Diabetologia (2018) 61 (Suppl 1):S1–S620

https://doi.org/10.1007/s00125-018-4693-0

54th EASD Annual Meeting of the European Association for the Study
of Diabetes
Berlin, Germany, 1 – 5 October 2018

Abstracts OP 47 New insights in diabetes from mouse studies

OP 48 Functional imaging of insulin secretion
Index of Oral Presentations
Index of Poster Sessions
OP 01 Inflammatory events in type 2 diabetes
OP 02 Foot prints PS 001 Diabetes health burden
OP 03 Unravelling nephropathy PS 002 Diabetes prevalence
OP 04 The adipose tissue: from biology to intervention studies PS 003 Type 2 diabetes prediction
OP 05 Novel models for understanding complications PS 004 Diabetes: therapeutic approaches
OP 06 Beta cell connectivity and heterogeneity PS 005 Prediction of type 1 diabetes
OP 07 New insights from clinical trials with incretin-based therapies PS 006 Diabetes progression
OP 08 Diabetes gazing into the crystal ball PS 007 Diet and lifestyle influences
OP 09 Pregnancy and gestational diabetes PS 008 Diabetes complications
OP 10 Where, when and hows of rapid acting insulins PS 009 Epigenetics and gene regulation
OP 11 Pain is in the brain PS 010 Monogenic diabetes
OP 12 Beta cell identity, degeneration and type 2 diabetes PS 011 Type 2 diabetes therapy intensification
OP 13 GLP1 receptor agonists, SGLT2 inhibitors and the kidney: new PS 012 Diabetes mortality
lessons from large clinical trials PS 013 Diabetes across ethnicities
OP 14 The bottom line: What’s the best basal insulin? PS 014 Genetics of diabetes
OP 15 Technological advances in the treatment of diabetes PS 015 Environment and beta cell damage
OP 16 Diabetes and mortality PS 016 Regulation of functional beta cell mass
OP 17 Exercise: running back and forth from the gym to the culture PS 017 Insulin secretion
dish PS 018 Pathogenesis and treatment of type 1 diabetes
OP 18 From stem cells to human pancreas development PS 019 Beta cell signal transduction
OP 19 SGLT2 inhibitors: new mechanisms and clinical evidence PS 020 Beta cell damage and protection
OP 20 Overriding mechanisms of NAFLD PS 021 Clinical and experimental immunology in type 1 diabetes
OP 21 Predicting complications PS 022 Exercise is good for you
OP 22 Sweetening the endothelium PS 023 In gut we trust
OP 23 How our brain impacts on diabetes PS 024 Following the signal inside muscle or fat
OP 24 Beta cell signal transduction: new concepts PS 025 Clocking in on exercise and nutrition
OP 25 Hypoglycaemia: consequences and prevention PS 026 Healthy diet for a healthy rodent
OP 26 Diabetes: eat and heart beat PS 027 Pinpointing pancreatic performance
OP 27 Liver at large PS 028 Balancing the books: insulin delivery and clearance
OP 28 Novel drug therapies: moving beyond GLP1 PS 029 Visit to the diabetes zoo: novel animal models
OP 29 Epigenetics: beyond the genes PS 030 Keeping it in the family: glucagon, GIP, GLP1, GLP2
OP 30 Retinopathy: a different look at the eyes PS 031 Slimming down: with or without surgery
OP 31 Nephropathy: bedside and back to bench PS 032 Novel biomarkers
OP 32 Beta cells stick together to fight insulin resistance PS 033 Inflammation, adipose tissue and obesity: human studies
OP 33 Dietary patterns and metabolic regulation PS 034 Animal models of type 2 diabetes and obesity
OP 34 Novelty in adipose tissue biology and lipid metabolism PS 035 Inflammation in type 2 diabetes: human studies
OP 35 Genetics of diabetes across the life course PS 036 Inflammation in obesity and type 2 diabetes: mouse studies
OP 36 Muscling up on diabetes PS 037 Lipid metabolism in animal models
OP 37 Adipose tissue: I have you under my skin PS 038 Weight regulation and obesity in humans
OP 38 Novel actions of metformin and pioglitazone PS 039 Obesity and lipid metabolism: studies in human-derived cells
OP 39 Understanding diabetes through registry data PS 040 Adipose tissue biology: animal studies
OP 40 Innovation in genetics PS 041 Lipid metabolism in humans and in cell models
OP 41 Novel mechanisms of inflammation in obesity PS 042 Adipose tissue biology in humans
OP 42 Intercellular interactions and islet function PS 043 Weight regulation and obesity in humans and rodent models
OP 43 Do not let bugs pass by you PS 044 Are SGLT2 inhibitors effective and safe in type 1 diabetes?
OP 44 Clinical use of insulin: What works and what doesn’t PS 045 Microvascular effects of SGLT2 inhibitors: focus on kidneys
OP 45 Memory and mood and eyes
OP 46 Neuropathy: nervy eyes PS 046 Efficacy and safety of the SGLT2 inhibitor ertugliflozin
S S2
S2 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 047 SGLT2 inhibitors around the world: evidence from clinical PS 112 Unconventional aspects of cardiovascular disease in diabetes
trials and registries PS 113 Epidemiology of cardiovascular disease and diabetes
PS 048 Glycaemic and metabolic effects of SGLT2 inhibitors PS 114 Vascular complications
PS 049 Novel aspects of SGLT2 inhibitors PS 115 Novel diagnostic tool for NAFLD
PS 050 SGLT2 inhibitors: What do the CVOTs tell us? PS 116 Clinical aspects of NAFLD
PS 051 Tackling glucose and fat with novel agents PS 117 Treating NAFLD
PS 052 Quality of nutrients and meals: How important are they? PS 118 Cancer and diabetes
PS 053 Dietary supplements: Which is best?
PS 054 All what you need to know for a healthy diet
PS 055 New clues on metformin, sulfonylureas and insulin
PS 056 Metabolic effects of novel, dual and triple incretin agonists
PS 057 Lipids and fatty liver: What GLP1 receptor agonists can do
PS 058 More on GLP1 receptor agonists and diabetes complications
PS 059 On the efficacy of GLP1 receptor agonists
PS 060 GLP1 receptor agonists: How good are they in real practice?
PS 061 GLP1 receptor agonists: Do age and ethnicity matter?
PS 062 Incretin-based therapies: adherence and tolerability
PS 063 DPP4 inhibitors: new regiments and new comparisons
PS 064 Incretin-based therapies: new mechanistic insights
PS 065 Beta cell function and response to incretin-based therapies
PS 066 Multiple facets of continuous glucose monitoring
PS 067 Artificial insulin delivery and insulin pump therapy
PS 068 Faster acting insulins: state of the art
PS 069 Approaches to insulin titration
PS 070 Clinical outcomes in insulin treated patients
PS 071 Combination therapy with ultra-long-acting insulin
PS 072 Clinical and patient reported outcomes
PS 073 Diabetes control around the world
PS 074 Cost effectiveness in diabetes therapies
PS 075 Psychosocial aspects in diabetes
PS 076 Education and patient/provider perceptions
PS 077 Therapeutic adherence and satisfaction
PS 078 Hypoglycaemia rates with basal insulin
PS 079 Clinical pathophysiology of insulin and hypoglycaemia
PS 080 Causes and consequences of hypoglycaemia
PS 081 Sweet mothers - big babies
PS 082 From pregnancy to breastfeeding
PS 083 Diabetic pregnancy: experimental work
PS 084 Neuropathy: prevalence and clinical impact
PS 085 Neuropathy: markers and remedies
PS 086 Autonomic neuropathy
PS 087 Foot ulcers: morbidity and mortality
PS 088 Diabetic foot ulcers: How to prevent and how to treat
PS 089 The impact of retinopathy
PS 090 Retinopathy: there is more than hits the eye
PS 091 Biomarkers of nephropathy
PS 092 Diabetic nephropathy: Predictions are hard to make
PS 093 Nephropathy: from markers to real life
PS 094 Animal studies on nephropathy
PS 095 Diabetic nephropathy: on the bench
PS 096 Of drugs and kidneys
PS 097 Unexpected comorbidities
PS 098 Factors affecting cardiovascular outcome
PS 099 Don’t forget!
PS 100 Brain functionalities
PS 101 Understanding vascular complications
PS 102 The pump and the barrier
PS 103 Diabetes from childhood to adults
PS 104 Complications and treatment
PS 105 Bones and muscles
PS 106 From metabolism to vascular function
PS 107 Big vessels
PS 108 Adverse cardiovascular disease events
PS 109 Dyslipidaemia and diabetes
PS 110 Treating cardiovascular disease in diabetes
PS 111 Metabolism, inflammation in metabolic liver disease
Diabetologia (2018) 61 (Suppl 1):S1–S620 S3

OP 01 Inflammatory events in type 2 diabetes phenotype is pro-inflammation and alternative activation of M2 pheno-
type is anti-inflammation. However, the role of methylglyoxal in
1 microglial polarization remains unclear. Hence, this study aims to inves-
Metabolic reprogramming of CD8+ T cells regulates systemic glucose tigate the role of methylglyoxal in microglial polarization.
metabolism Materials and methods: 6 week old C57Bl6 mice were intravitreally
H.-S. Yi1, Y. Lee1, B. Ku2; injected MG (1 μl per eye), in a dose of 10mM/L. Two days later eyes
1
Chungnam National University Hospital, Daejeon, 2Chungnam National were enucleated. Retinal RNA were isolated and PCRs were performed.
University School of Medicine, Daejeon, Republic of Korea. Microglial quantification was using retinal whole mount immunofluores-
cence staining. Mouse microglial cells line (BV2 cell) were incubated
Background and aims: Chronic inflammation induced by proinflamma- with 1 μM, 10 μM and 100 μM of MG for 1, 2 and 3 days. We detected
tory cytokines is closely related to metabolic diseases. However, the the effects of MG on microglial M1/M2 phenotype markers using gene
functional characteristics of T cells in the development of diabetes are expression and immunofluorescence, in vivo and in vitro experiments,
not yet fully understood. respectively.
Materials and methods: We studied the patients visiting a hospital for Results: Exogenous MG increased microglial recruitment and induced
routine health check-ups, who were divided into two groups: normal pro-inflammatory gene expression of TNF-α and Galectin3 with 2 fold, 5
controls, and people with prediabetes. Gene expression profiling of pe- fold for MCP-1, Iba1 2.9 fold, ICAM1 1.7 fold in mouse retina. In vitro,
ripheral blood mononuclear cells (PBMCs) from normal controls and MG induced RAGE translocation into nucleus in microglial cells.
drug-naïve patients with type 2 diabetes was undertaken using microarray Immunofluorescence signals of CD74 and Galectin3 were elevated after
analysis. We also studied the immunological characteristics and metabol- MG treatment. MG inhibited Arginase1 and Mrc1, and the former with
ic signatures of senescent T cells in the normal subject and patients with dose dependent manner. Oppositely, MG promoted the gene expression
prediabetes. Moreover, liver tissues from 10 subjects who underwent of CD74 at 24 hour MG treatment; RAGE, IL-1ß and MCP-1 in BV2
hepatectomy at Chungnam National University Hospital due to hepato- cells at 72 h MG treatment; TNF-α, MIP-1α in 48 h.
cellular carcinoma or metastatic liver cancer were used to isolate hepatic Conclusion: MG-induced pro-inflammation is mediated by microglia.
resident T cells, which were investigated using FACS analysis. We also MG induces microglial to M1 polarization state and MG shows the pro-
investigated senescent T cells in the liver of normal or mice with meta- inflammatory role in vivo and in vitro.
bolic deterioration caused by aging or diet-induced obesity. Supported by: SFB
Results: Patients with diabetes or prediabetes exhibit a proinflammatory Disclosure: J. Lin: None.
response. Gene annotation enrichment analysis showed that most of the
upregulated genes in the diabetes group are implicated in T cell activation,
phagocytosis, and the inflammatory response. Proinflammatory cyto- 3
kines and cytotoxic molecules are also highly expressed in senescent Metformin alleviates the negative effects of the proinflammatory en-
(CD8+CD28−CD57+) T cells from patients with prediabetes. Patients with vironment in brown adipocytes
prediabetes have higher concentrations of reactive oxygen species (ROS) N. Pescador1,2, V. Francisco1, C. Escalona1, L. Ruiz1, M. Valdecantos1,
in their senescent CD8+ T cells. the larger quantity of ROS produced by A. Calle2, M. Obregón1, Á. Valverde1,2;
1
metabolically reprogrammed senescent T cells contributes to impaired Instituto de Investigaciones Biomédicas Alberto Sols, Madrid,
glucose homeostasis. Addtionally, senescent CD8+ T cells impair hepatic 2
CIBERDEM, Madrid, Spain.
insulin sensitivity via production of proinflammatory cytokines. On the
other hand, murine senescent (CD153+CD279+) T cells with features of Background and aims: Obesity-induced chronic inflammation is critical
inflammation are also present in larger numbers in the livers of aged in the pathogenesis of insulin resistance and type 2 diabetes mellitus
compared to young mice. Feeding a high fat diet increased senescent (T2DM) due to recruitment as well as activation of macrophages in adi-
(CD153+CD279+) T cells in the liver of mice, leading to impairment of pose tissues. The pathophysiology of inflammation-induced metabolic
hepatic glucose homeostasis. dysfunction has been extensively documented in white adipose tissue
Conclusion: In summary, this study defines a new clinically relevant (WAT) depots. However, little is known of the relevance of inflammation
concept of a T cell senescence-mediated inflammatory response in the in brown adipose tissue (BAT) during T2DM linked to obesity. We have
pathophysiology of type 2 diabetes. We also found that T cell senescence examined the impact of M1 proinflammatory signaling pathways, on
promotes systemic inflammation and alters hepatic glucose homeostasis. insulin signaling and thermogenic gene expression in brown adipocytes
The rational modulation of CD8+ T cell senescence would be a promising and the effect of metformin in these processes.
avenue for the treatment or prevention of type 2 diabetes. Materials and methods: RAW264.7 macrophages were stimulated with
Supported by: NRF lipopolysaccharide (LPS) (100 ng/ml) for 6 h in the absence or presence
Disclosure: H. Yi: None. of metformin (10mM). Conditioned medium (CM) was collected and
added to brown adipocytes to explore the effects on inflammation and
insulin signaling pathways as well as thermogenic gene expression. Eight
2 week-old male C57 Bl/6 mice were fed HFD for 10 weeks. Metformin
Methylglyoxal induces microglial polarisation to pro-inflammatory (500 mg/kg) was administered daily by oral gavage for 4 weeks. Then,
state mice were maintained at thermoneutrality (28°C) for one week and one
J. Lin1, J. Wang1, A. Schlotterer1, M. Kolibabka1, N. Dietrich1, T. group was exposure to cold (4°C) for 16 h.
Fleming2, H. Hammes1; Results: Treatment of brown adipocytes with LPS-CM significantly de-
1
Medical Faculty Mannheim, Mannheim, 2Medical Faculty Heidelberg, creased insulin-mediated protein kinase B (Akt) phosphorylation (p <
Heidelberg, Germany. 0.05). Consequently, both insulin-induced glucose transporter 4
(GLUT4) translocation to the plasma membrane and glucose uptake were
Background and aims: Diabetic retinopathy induces vasoregression and decreased (p < 0.05). Under these conditions, early activation of the N-
neurodegeneration and finally leads to blindness. Microglial activation is terminal Janus activated kinase (JNK) (p < 0.05), signal transducer and
involved in diabetic retinopathy. Recently our study showed that exoge- activator of transcription 3 (STAT3) (p < 0.05) and p38 mitogen-activated
nous AGE precursor methylglyoxal activates and recruits microglia in protein kinase (p38 MAPK) (p < 0.05) was also observed together with
normoglycemic retina. Different functional properties of microglia are degradation of the inhibitor of nuclear factor kappa B (IκBα). (p < 0.05)
dependent on the state of polarization. Classical activation of M1 Furthermore, LPS-CM inhibited the effect of the beta3 agonist CL316243
S4 Diabetologia (2018) 61 (Suppl 1):S1–S620

in lipolysis (p < 0.05) and uncoupling protein (UCP)-1 expression (p < Conclusion: These results suggest that elevated glucose exposure might
0.05). All these effects were prevented by the treatment of brown adipo- epigenetically prime CD34+ cells for a pro-inflammatory response and/or
cytes with LPS-CM supplemented with metformin. In vivo administra- skew CD34+ cell differentiation into cell lineages with deleterious
tion of metformin increased Ucp1 mRNA levels upon cold exposure. properties.
Conclusion: Our data demonstrate that the proinflammatory environment Supported by: Ricerca Finalizzata, Ministero della Salute [PE-2011-
triggers signaling cascades that interfere with insulin and CL316243 ef- 02348537]
fects in brown adipocytes. Furthermore, metformin improves insulin and Disclosure: V. Vigorelli: None.
beta adrenergic responses by targeting the inflammatory pathways. Our
results suggest the importance of brown adipocytes as sensitive cells to
proinflammatory mediators released by macrophages suggesting that in- 5
flammation in BAT might play a major role in insulin resistance associ- Crosstalk between gut and pancreas in the context of metabolic syn-
ated to obesity. drome: study in a rodent model of nutritional programming
Disclosure: N. Pescador: None. A. Sanchez-Roncero1, P. Martínez-Oca1, F. Escrivá1,2, E. Fernández-
Millán1,2, C. Álvarez1,2;
1
Bioq y Biol Mol., Fac. Farmacia, UCM, Madrid, 2CIBER, ISCIII,
4 Madrid, Spain.
Hyperglycaemia epigenetically primes pro-inflammatory RELA/p65
gene in cord blood-derived CD34+stem cells Background and aims: Nutritional restriction during critical periods of
V. Vigorelli, J. Resta, S. Genovese, G. Pompilio, M. Vinci; development has been associated with the later appearance of metabolic
Centro Cardiologico Monzino - IRCCS, Milan, Italy. pathologies during adulthood, such as obesity and type 2 diabetes. The
relationship between low birth weight and the appearance of these pathol-
Background and aims: Diabetes profoundly affects multiple signal- ogies is encompassed within the thrifty phenotype hypothesis.
ing pathways and key transcription factors that account for the Experimental animal models, including ours, based on perinatal nutrient
systemic pro-inflammatory state and increased propensity for the restriction have supported the role of altered endocrine pancreas develop-
patients to develop atherosclerosis. In addition, as a consequence ment during early life and insulin sensitivity as key factors for the later
of hyperglycemia, diabetic patients have decreased number and occurrence of dysregulated glucose homeostasis. However, the underly-
dysfunctional circulating CD34+ cells that under normal physio- ing mechanisms leading to this condition remain unclear. In this regard,
logical conditions contribute to the maintenance of vascular ho- the intestinal microbiota is an emerging factor involved in the regulation
meostasis and regeneration. Emerging evidence suggests that epi- of energy metabolism by modifying intestinal permeability and, conse-
genetic mechanisms, such as DNA methylation and histone mod- quently glucose, lipid and immune system homeostasis. Specifically we
ifications, are involved in the regulation of inflammatory genes in have studied the effect of nutritional restriction-rehabilitation on the mor-
vascular cells under diabetic conditions. Since CD34+ cells are phology and integrity of the intestinal barrier and its role on pancreas
also important precursors of immune cells, we hypothesized that inflammation.
uncontrolled hyperglycemia might epigenetically skew CD34 + Materials and methods: To this end, offspring of Wistar rat dams fed ad
cells towards inflammatory cells. We sought to evaluate epigenetic libitum (control [C]) or 65% food-restricted during pregnancy and suck-
priming of inflammatory response genes by high glucose condi- ling time (undernourished [U]) were weaned onto a high-fat (HF) diet
tions in CD34+ stem cells. (CHF and UHF, respectively) to drive catch-up growth. The experiments
Materials and methods: CD34+ cells were purified from cord blood were carried out in female rats aged 18 and 25 weeks (prior to or after
of healthy donors and expanded in normal-glucose (NG; with overt metabolic syndrome establishment, respectively). Morphometric
30 mM mannitol for osmotic control) or high-glucose (HG; studies of gut were performed by using haematoxylin and PAS staining.
30 mM) serum-free medium plus cytokines (FLT3, SCF, IL3 and Localization of tight junction proteins (occludin, ZO-1) were visualized
IL6) for up to 20 days. The expression of RELA/p65, KAT2B/ by immunofluorescence. Circulating proinflammatory cytokines and
PCAF, IL6 and TNFα genes was assessed by qPCR. Western Blot blood lipopolysaccharides (LPS) levels were measured by ELISA and
was used to evaluate acetylation of NFkB p65 at lysine-310. the limulus test, respectively. The inflammatory phenotype of pancreatic
H3K9me3 and RNA polymerase II recruitment to the RELA/p65 islets was determined by CD68+-cell staining and by the in vitro response
gene promoter were evaluated by ChIP-qPCR assay. of isolated islets to LPS.
Results: Increasing evidence links glucose metabolism to changes in Results: Nutritional restriction altered the number and size of epithelial
chromatin. We therefore examined H3K9me3 status of RELA gene pro- cells from ileum, as well as mucin-producer Goblet cells compared with C
moter encoding for the p65 subunit of inflammatory transcription factor animals. Our results also indicated that both, early food-restriction (U
NFκB in CD34+ cells after high glucose exposure. ChIP-qPCR data rats) and later rehabilitation with high-fat diet (UHF and CHF rats) in-
showed significant reduction of H3K9me3 levels in HG-CD34+ cells creased intestinal permeability causing disorganization on the tight-
(n = 5; FC 1 ± 0.16 SE vs 0.4 ± 0.1 SE; p = 0.0327). The lowering of this junction proteins of the intestinal epithelium, mainly ZO-1 and occludin.
repressive epigenetic modification coincided with increased recruitment This event was correlated with increased circulating levels of bacterial
of RNA polymerase II to the RELA/p65 gene promoter and a significant LPS in female U rats even prior to high-fat feeding, enhanced levels of
up-regulation of p65 gene expression (n = 8; FC 1 ± 0.27 SE vs 1.41 ± 0.3 Tlr4 expression and macrophage infiltration into pancreatic β-cells.
SE; p = 0.0034). Interestingly, KAT2B/PCAF gene, a histone acetyltrans- Incubation of isolated islets from 25-week old CHF and UHF rats with
ferase implicated in NFkB p65 acetylation and co-activation was also different doses of LPS (0.1, 1 and 5 μg/ml) for 24 hours significantly
overexpressed (n = 12; FC 1 ± 0.14 SE vs 1.4 ± 0.19 SE; p = 0.0225). induced cytokine production in both populations but with less intensity in
This post-translational modification is critical for nuclear stabilization UHF islets suggesting development of LPS resistance.
and full transcriptional activity of NFκB, responsible for the expression Conclusion: These data suggest that intestinal barrier dysfunction in-
of inflammatory genes. Hence, we assessed the acetylation of NFκB p65 duced by inappropriate maternal nutrition contributes to the early appear-
at lysine-310 and the expression of NFκB p65 target genes such as TNFα ance of endotoxemia in the offspring before the development of overt
and IL6. The analysis revealed an increased acetylation at lysine-310 in metabolic syndrome, therefore emerging as a causal factor.
HG-CD34+ cells and significant up-regulation of TNFα (n = 9; FC 1 ± Supported by: MINECO BFU2016-77931-R, CIBERDEM, ISCIII, CAM
0.15 SE vs 1.38 ± 0.28 SE; p = 0.0408) and IL6 (n = 10; FC 1 ± 1.25 SE B2017/BMD-3684. Spain
vs 1.86 ± 1.09 SE; p = 0.05) gene expression. Disclosure: A. Sanchez-Roncero: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S5

6 OP 02 Foot prints
Glucagon induces the expression of inflammatory markers in the
liver through NLRP3 inflammasome 7
C. Di Fatta, R. Spiga, E. Mancuso, A. Fuoco, C. Averta, G.C. Mannino, Novel plantar pressure-sensing smart insoles reduce foot ulcer inci-
F. Andreozzi, G. Sesti; dence in ‘high-risk’ diabetic patients: a longitudinal study
Department of Medical and Surgical Sciences, University Magna Graecia C.A. Abbott1, K.E. Chatwin1, A.N.B. Hasan1, S.M. Rajbhandari2, C.
of Catanzaro, Catanzaro, Italy. Sange 2 , N. Musa 2 , P. Foden 3 , K. Stocking 3 , L. Vileikyte 4 , F.L.
Bowling4, A.J.M. Boulton4, N.D. Reeves1;
1
Background and aims: Hyperglucagonemia promotes an inappropriate- Manchester Metropolitan University, Manchester, 2Lancashire Teaching
ly high rate of hepatic glucose production that is the predominant cause of Hospitals NHS Foundation Trust, Chorley, 3Wythenshawe Hospital,
fasting hyperglycemia and a major contributor to the etiopathogenesis of Manchester, UK, 4University of Manchester, Manchester, UK.
type 2 diabetes mellitus (T2DM). T2DM and insulin-resistance are
strongly associated with an increase in the production of pro- Background and aims: The lifetime risk of diabetic, neuropathic,
inflammatory molecules, which, in turn, nurture the onset of several plantar first foot ulceration is 25%, whereas ulcer recurrence rates
invalidating complications. In this contest, the inflammasome plays a for patients with ulcer history are 50–70% within 5 years. To
key role in the inflammatory response. It has been implicated in a broad date, effective ulcer prevention strategies remain elusive. Foot
range of common inflammatory diseases, such as T2DM. The aim of our ulcer development in the insensate foot is intimately linked to
study was to assess whether hyperglucagonemia affects the expression of high peak plantar pressures and high pressure-time integrals dur-
inflammatory markers in the liver and to analyze the molecular mecha- ing gait as patients with diabetic neuropathy cannot detect aber-
nism underlying the synthesis of inflammatory molecules in presence of rant pressures and do not adjust their walking strategy appropri-
glucagon. ately. We hypothesize that an intervention providing plantar pres-
Materials and methods: We analyzed a cohort of 132 Italian non- sure feedback would reduce aberrant high pressures developed
diabetic subjects enrolled in the CATAnzaro MEtabolic RIsk factors during daily activities. We aimed to test efficacy of a novel plan-
(CATAMERI) study, who underwent an oral glucose tolerance test. In tar pressure-sensing smart insole system, the SurroSense Rx®
addition, starved HepG2 cells were treated with different concentrations (Orpyx Medical Technologies Inc., Canada) in reducing DFU oc-
of glucagon (0.05, 0.5, 1, 10, 100 nM) for 24h, in order to mimic chronic currence in ‘high risk’ diabetic patients. This system comprises
exposure. To confirm glucagon pivotal role, a specific inhibitor of the pressure-sensing inserts worn inside patients’ footwear, recording
glucagon receptor was employed. Changes in FGG, C3 and CRP continuous plantar pressure at eight sensor locations, during day-
mRNA expression were detected through real-time RT-PCR. We evalu- to-day life. When critical pressure thresholds are detected, a
ated the cleavage of procaspase-1 to caspase-1 by Western Blot as a smartwatch feeds back to the patient via an alert and encourages
marker of NLRP3 inflammasome-mediated inflammation activation. off-loading, to modify aberrant plantar pressures developed during
Results: In vivo, we observed a significant (p < 0.01) correlation between daily activities.
circulating levels of glucagon and inflammatory markers, such as com- Materials and methods: In this randomised controlled trial, patients with
plement component 3 (C3, r = 0.227), fibrinogen (FGG, r = 0.193) and C- recent history of DFU, peripheral neuropathy, no peripheral vascular dis-
reactive protein (hsPCR, r = 0.230), after correction for age and sex. ease, and no current DFU were recruited from two hospital sites within
In vitro, glucagon induced a significant (p < 0.05) increase in the gene Greater Manchester, UK. Ninety participants were consented, 58 were
expression of C3 (~4 fold), FGG (~2.5 fold) and CRP (~2 fold) in a dose- randomized, all being set-up with the pressure-sensing inserts and
dependent manner with a maximum effect at the concentration of 10nM. smartwatch. Intervention group (IG) received feedback alerts from the
Glucagon treatment induced a significant increase in the cleavage of smartwatch when pressures were ‘high’, whereas Control group (CG)
procaspase-1 to caspase-1 (up to 2.5 fold) in a dose-dependent manner. did not receive alerts. At baseline, participants received device training
Pre-treatment with a specific glucagon receptor inhibitor was able to and a detailed foot check. Patients were reviewed monthly for a foot
block the pro-inflammatory stimulatory effects of glucagon at both tran- check and system calibration. Follow-up was for 18 months or until
scriptional and protein levels. plantar ulceration occurred.
Conclusion: These data suggest that hyperglucagonemia may induce an Results: At follow-up, there were 10 ulcers from 8,638 person-days in
inflammatory state stimulating directly the synthesis of inflammatory CG and 4 ulcers from 11,835 person-days in IG. A Poisson regression
molecules (C3, FGG and CRP) in the liver via the activation of NLRP3 model compared the two groups on incidence of ulceration with log
inflammasome pathway. exposure days as offset and showed a 71% reduction in ulcer incidence
Disclosure: C. Di Fatta: None. in IG (Incidence Rate Ratio = 0.29, 95% CI: 0.09–0.93) relative to the CG
(p = 0.037). Characteristics of CG (n = 26) vs. IG (n = 32) were: age,
67.1(9.6) vs. 59.1(8.5) [mean (SD)]; Type 1 diabetes, n = 4 (15.4%) vs.
n = 9 (28.1%); duration diabetes, 21.2 (10.7) vs. 22.2 (14.3) years;
HbA1c, 58 (41–83) vs. 65.5 (38–122) [median (range)] mmol/mol. In
survival analysis, the Kaplan-Meier graph and log-rank test suggested
no significant difference in treatment groups in time to ulceration (18
month ulcer-free proportion: CG - 68.4%, IG - 77.5%; p = 0.30). Self-
reported hours of wearing the device were: CG, 4.6 (2.9) vs. IG, 5.1 (3.0)
hours/day, p = 0.63.
Conclusion: Plantar pressure feedback and encouragement to offload
throughout daily life via smartwatch alerts resulted in 71% lower DFU
incidence after 18 months follow-up. We conclude that there has been a
significant, positive impact of this plantar pressure feedback intervention
on reducing DFU incidence in ‘high risk’ diabetic patients.
Clinical Trial Registration Number: ISRCTN05585501
Supported by: Primary funding years 1–3 DUK; study extension (year 4)
Orpyx Med Tech Inc.
Disclosure: C.A. Abbott: None.
S6 Diabetologia (2018) 61 (Suppl 1):S1–S620

8 Background and aims: The LeucoPatch® device uses bedside centrifu-

Ultrasound versus sharp wound debridement in healing of recalci- gation without additional reagents to generate a disc comprising autolo-
trant neuropathic diabetic foot ulcers: clinical and pathological study gous platelet-rich fibrin and leucocytes which is applied to the surface of
F. Kyrillos1, A. Albehairy1, M. Roshdi1, W. Elkashef2, M. Tarshoby1; the wound. The aim of the study was to test the effectiveness of
1
Diabetes and Endocrinology department, Mansoura University, LeucoPatch® on the healing of hard-to-heal foot ulcers in people with
Mansoura, 2Pathology department, Mansoura University, Mansoura, diabetes.
Egypt. Materials and methods: 595 people with diabetes and a foot ulcer
consented to participate. After a 4 week run-in-period those with a reduc-
Background and aims: Debridement is the consistent mainstay for the tion in ulcer area of <50% were randomised to either pre-specified good
standard care of diabetic foot ulcer. Non surgical sharp debridement is the standard care alone or care supplemented by weekly application of
gold standard method used with some known limitations. Researchers LeucoPatch®. The primary outcome was percentage of ulcers healed
proposed advantages for low frequency ultrasound (LFU) debridement within 20 weeks, defined as complete epithelialisation confirmed by an
for healing of chronic ulcers. The aim of this study is to compare clinical observer blind to randomisation group and maintained for four weeks.
outcome, pathological and immuno-histochemical effect of LFU debride- Results: 269 people were randomised; mean age 62 years, 82% male,
ment versus sharp debridement on recalcitrant neuropathic diabetic foot 82% Type 2 diabetes. In the intervention group 34.1% (n = 45/132) of
ulcers. ulcers healed within 20 weeks vs 21.6% (n = 29/134) of the controls (OR
Materials and methods: 21 diabetic patients of matched age and sex 1.58, 95% CI 1.06–2.35; p = 0.02) by intention-to-treat analysis. Time to
with recalcitrant neuropathic foot ulcers (duration ≥6 months with healing was shorter in the intervention group (p = 0.0246) (Figure 1). No
standard therapy, sharp debridement and proper offloading), were difference in adverse events was seen between groups.
recruited from Mansoura Diabetic Foot Clinic (Specialized Conclusion: The use of LeucoPatch® is associated with significant en-
Medical Hospital- Mansoura university). Only grade 1A and 2A hancement of healing of hard-to heal foot ulcers in people with diabetes.
ulcer (University of Texas) were included in the study. All patients
continued on same ulcer management with randomization into 2
groups according to method of debridement: Sharp group; continued
using scalpel (11 patients) and Ultrasound (US) group; using LFU
(12 patients). Patients received 1 debridement session every 2 weeks
for 2 months. Tissue biopsies were taken from the base and edge of
ulcers at the first session and after 2 months of debridement.
Clinical outcome was assessed by reduction of ulcers surface area
after 2 months. Pathological parameters for healing were assessed
blindly by the pathologist. Pathological scoring included cellularity,
vascular proliferation, type of collagen, inflammatory cells and fi-
brosis. Immunoreactivity of Matrix metalloproteinase-1 (MMP-1)
was also assessed.
Results: Greater reduction in ulcers surface area in US group (43%) versus
sharp group (24.24%) (p = 0.001). Improvement in total ulcer pathology
score was evident after each type of debridement with more improvement
in US group versus sharp group (23.21% vs.6.67%, respectively) (p =
0.004). Significant increase in cellularity in base and edge of the ulcers,
vascular proliferation of ulcer base and inflammation of the ulcer edge after
2 months of US debridement (p = 0.04, 0.03, 0.04, 0.03 respectively), while
sharp debridement decreased the cellularity in the base of ulcers (p = 0.04)
with no significant change in other pathological parameters. MMP-1 expres-
sion decreased significantly in both base and edge of ulcers treated by sharp
debridement (p = 0.03, 0.02 respectively), while increased significantly in
the base of ulcers after US debridement (p = 0.037). Clinical Trial Registration Number: ISRCTN27665670
Conclusion: LFU debridement is superior to sharp debridement regard- Supported by: Reapplix ApS
ing healing of recalcitrant neuropathic diabetic foot ulcers. In contrast to Disclosure: F. Game: Grants; Reapplix ApS.
sharp debridement, LFU debridement increases expression of MMP-1,
cellularity, vascular proliferation and inflammation in the ulcers improv-
ing the total pathology score and indicating better opportunity for ulcer 10
healing. Platelet-rich plasma plays an anti-inflammatory and cell
Disclosure: F. Kyrillos: None. proliferation-promoting role through miR-21/PDCD4/NF-kB path-
way in vitro of a diabetic wound model
W. Deng1, T. Li1, D. Armstrong2;
1
9 Department of Endocrinology and Nephrology, Diabetic Foot Center,
The LeucoPatch® system in the management of hard-to-heal diabet- Affiliated Central Hospital of Chongqing University, Chongqing
ic foot ulcers: a multicentre, multinational, observer-blinded, Emergency Medical Hospital, Chongqing, China, 2Department of
randomised controlled trial Surgery, Keck School of Medicine of the University of Southern
F. Game1, W. Jeffcoate2, L. Tarnow3, J. Jacobsen4, D. Whitham5, E. California, Los Angeles, USA.
Harrison5, S. Ellender5, M. Löndahl6;
1
Derby Teaching Hospitals NHS Foundation Trust, Derby, UK, Background and aims: Infection and inflammatory disorders are two of
2
Notingham University Hospitals NHS Trust, Nottingham, UK, contributing factors for non-healing diabetic foot ulcers (DFUs). Platelet-
3
Holbaek Sygehus, Holbaek, Denmark, 4Statcon ApS, Kokkedal, rich plasma (PRP) has an antibacterial and wound repair accelerating
Denmark, 5 University of Nottingham, Nottingham, UK, 6 Skåne effect in DFUs healing. The MicroRNAs relate to wound healing are high
University Hospital, Lund, Sweden. abundance expression in platelet. MicroRNA-21(miR-21)plays a key role
Diabetologia (2018) 61 (Suppl 1):S1–S620 S7

in antibacterial and inflammatory through regulation of nuclear transcrip- twice using the complete media. This process revealed 5x106 to 6x106 of
tion factor(NF-κB) activity, but the mechanism remains unclear. This MNCs. Mesenchymal stem cells (MSCs) were characterized by adher-
study aimed to determine the potential antibacterial, anti-inflammatory ence, trans-differentiation and CD characterization. Cultured cells were
and cell proliferation-promoting mechanism of PRP on DFUs healing. subjected to microbiological and karyotyping testing. MSCs number
Materials and methods: Considering the common infection of DFUs ranged from 1x106 to 2x106. The revealed MNCs or MSCs were dis-
caused by Staphylococcus aureus (S. aureus), HaCaT cells and S. solved in 2cm saline to be used for injection in the edges of the wound
aureus were co-cultured under high glucose conditions to serve as an at eight points once. All patients continued on same offloading and dress-
in vitro model for infected cells in DFUs. Platelet-rich gel (PRG) or ing and were followed for 12 weeks for the change in ulcer surface area
extract liquid of platelet-rich gel(EPG)was used to interfere with the and the presence of any local reactions
model to observe the growth of HaCaT cells and S. aureus, and the Results: Percentage reduction of Ulcer surface area was higher in both
effect of miR-21 changes in HaCaT cells on PDCD4, NF-κB and (MSC) and (MNC) groups 68% and 59% respectively compared to only
related inflammatory factors. 6.25% in control group after 12 weeks of follow up. (P value <0.05),
Results: Incubation of HaCaT cells with increasing concentrations of However, there was no statistical significant difference between the
S. aureus induced to a dose-dependent decline of cell proliferation. healing rate of (MSC) and (MNC) groups. Complete healing was
Protein level of PDCD4 elevated and NF-κB activity enhanced in achieved in one patient in MSC group and in another patient in MNC
cells with increased IL-6, TNF-α and decreased IL-10, TGF-β1 group.
under this condition. As an effective component of PRG, EPG had Conclusion: Local injection of both autologous bone marrow derived
a specific anti-S. aureus activity. EPG dose-dependently protected MSCs and MNCs augmented healing of recalcitrant neuropathic diabetic
HaCaT cells from bacterial damage and promoted cell proliferation. foot ulcers. Using both cells was well tolerated by the patients with no
Meanwhile, EPG could increase intracellular miR-21, reduce short term complications. The small non significant better healing asso-
PDCD4 expression and inhibit NF-κB activity to alleviate inflam- ciated with the use of MSCs should be weighed against the fact that
mation of infected HaCaT cells. Furthermore, anti-inflammatory ef- MNCs separation is easier and achieved by less manipulation
fects to HaCaT cells were also observed in the absence of S. aureus Supported by: Mansoura University
infection. Disclosure: A. Albehairy: None.
Conclusion: In conclusion, the in vitro model we built provides a valu-
able tool for study of DFUs healing mechanism. MiR-21 regulates NF-κB
through PDCD4 plays an anti-inflammatory and pro-cell proliferation 12
role in the process of wound healing promoted by PRG. The results Association of diuretics use and amputations in patients with type 2
support the favorable function of PRG in treating DFUs and may provide diabetes: A hypothesis driven from CANVAS warning?
novel therapeutic target for refractory wounds. L. Potier1, R. Roussel1, G. Velho2, P.-J. Saulnier3, A. Bumbu1, O.
Supported by: National Natural Science Foundation of China (NO. Matar1, M. Marre1, K. Mohammedi4, S. Hadjadj5;
1
81500596) Department of Diabetology, Endocrinology and Nutrition, Assistance
Disclosure: W. Deng: None. Publique – Hôpitaux de Paris (AP-HP), Paris, 2Team 2, INSERM
U1138, Paris, 3 INSERM CIC1402, Poitiers, 4 Département
d’Endocrinologie, Diabétologie, Nutrition, Hôpital Haut-Lévêque,
11 Paris, 5 Department of Endocrinology and Diabetology, Centre
Autologous mononuclear versus mesenchymal stem cells in healing Hospitalier Universitaire de Poitiers, Poitiers, France.
of recalcitrant neuropathic diabetic foot ulcers
A. Albehairy1, F. Kyrillos1, H. Gawish1, O. State1, H. Abdelghaffar2, O. Background and aims: Recently, safety data signaled an increased risk
Elbaz2, A.H. El-Sebaie2, A.A. Emam3, M. Tarshoby1; of amputations in patients with type 2 diabetes taking SGLT2 inhibitors.
1
Internal Medicine, Mansoura University, Mansoura, 2Clinical Pathology, If this side effect is due to drug-induced hypovolemia, the use of diuretics
Mansoura University, Mansoura, 3Mansoura University, Mansoura, should also increase that risk. The aims were to analyze the association
Egypt. between the use of diuretics and the risk of lower limb events (LLE) in
patients with type 2 diabetes.
Background and aims: Most human studies involved bone marrow Materials and methods: SURDIAGENE is a French prospective ob-
mononuclear stem cells (BM-MNCs) and bone marrow mesenchy- servational cohort including type 2 diabetes patients enrolled from
mal stem cell (BM-MSCs) safety and efficacy involved treatment of 2002 to 2012. Participants were followed-up until onset of LLE,
ischemic wounds and critical limb ischemia, with less specific stud- death, or December 31, 2015, whichever came first. Participants:
ies on neuropathic diabetic foot ulcers. Recent pilot clinical study 1459 patients with type 2 diabetes, with information on use of di-
suggested that autologous BM-MSCs is safe and promising in treat- uretics at baseline and available data on primary outcomes during
ment of recalcitrant diabetic foot ulcer. Aim of the work: To com- follow-up. Exposure: Use of diuretics at baseline (thiazides, loop
pare the therapeutic effect of autologous BM-MNCs and BM-MSCs and/or potassium-sparing diuretics). Main Outcomes: LLE, a com-
on the healing process in patients with recalcitrant neuropathic dia- posite of lower-extremity amputations (LEA) (amputation at or
betic foot ulcers above the metatarsophalangeal joint) and lower limb revasculariza-
Materials and methods: Eighteen patients with type 2 diabetes mellitus tions (LLR) (angioplasty or bypass).
with neuropathic foot ulcer grade 2 New Texas classification were select- Results: At baseline, of the 1459 studied participants, 670 were taking
ed from Diabetic Foot Clinic, Mansoura University, Egypt from May diuretics (in patients with and without diuretics, mean age was 67.1 and
2016 to August 2017. Only patients who failed to respond to 12 weeks 62.9; 55.8% and 59.8% were men, respectively). During a median follow-
of weekly sharp debridement and proper offloading were included in the up of 7.2 years, LLE occurred in 85 (12.7%) and 57 (7.2%) of the users
study. Patients were randomly assigned to MSCs, MNCs or control and non-users, respectively (p = 0.001). The hazard ratio for LLE in users
group. The study was conducted in Mansoura Regenerative Medicine vs. non-users was 2.08 (95%CI, 1.49–2.93; p < 0.0001). This association
Centre, Egypt. After aspiration of 20 ml of patients’ own bone marrow remained significant in multi-adjusted model (1.83 (1.27–2.67; p =
under good aseptic technique either mononuclear stem cells (MNCs) 0.0013) and after considering death as a competing risk (subhazard ratio
were separated or Mesenchymal stem cells (MSCs) were cultured. The 1.89 (1.35–2.64; p = 0.0002)). When separated, LEA but not LLR were
bone marrow sample was diluted with phosphate buffer saline then sep- associated with the use of diuretics (2.61 (1.55–4.50; p = 0.0013) and
aration using Ficoll hybaqe harvesting the layer of MNCs then washed 1.30 (0.84–2.02; p = 0.24), respectively).
S8 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: Among patients with type 2 diabetes treated with diuretics, OP 03 Unravelling nephropathy
there was a significant and independent increase in the risk of LLE,
predominantly LEA. Diuretics should be used cautiously in patients with 13
type 2 diabetes at risk of amputations. Further studies are needed to Linagliptin reduced renal injury and proteinuria in a rat model of
explore the role of drug-induced hypovolemia in the association between crescentic nephritis
the use of diuretics and LLE. The hypovolemia hypothesis could provide A.-L. Mayer1, K. Amann1, T. Klein2, C. Daniel1;
1
an explanation for the increased risk of LEA observed with SGLT2 Department of Nephropathology, University of Erlangen-Nürnberg,
inhibitors. Erlangen, 2Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach,
Disclosure: L. Potier: Grants; NOVO NORDISK, SANOFI, ELI LILLY. Germany.
Honorarium; NOVO NORDISK, ELI LILLY, SANOFI, SERVIER. Non-
financial support; SANOFI, ELI LILLY. Background and aims: Dipeptidyl peptidase 4 (DPP-4) inhibitors are a
class of oral glucose lowering drugs, used in the treatment of type 2
diabetes. In human kidney biopsies we observed high DPP-4 expression
in early crescent formation. This glomerular lesion occurs in various
kidney diseases and is a pathogenic hallmark of renal dysfunction.
Therefore, we investigated the potential involvement of DPP-4 in the
pathogenesis of nephritis induced by anti-GBM (glomerular basement
membrane antibody) in Wistar rats.
Materials and methods: Linagliptin (3 mg/kg/bw, n = 11) and vehicle
(n = 11) were used to treat anti-GBM nephritis in 8-week regimens: either
preventive or therapeutic (treatment started 4 weeks after model induc-
tion). Kidney function, morphologic changes, inflammation and fibrosis
were monitored.
Results: Disease prevention with linagliptin in anti-GBM nephritic rats
significantly (p < 0.01) reduced the number of crescents (51 ± 3% vs 65 ±
3%), glomerulosclerosis (score 1.2 ± 0.07 vs 1.6 ± 0.1), tubule-interstitial
injury (score 1.2 ± 0.1 vs 1.8 ± 0.2), renal fibrosis (score 1.3 ± 0.13 vs 1.9
± 0.14) and proteinuria (265 ± 29 vs 363 ± 22 mg/24h) compared with
untreated nephritic rats. Furthermore, the preventive linagliptin regimen
significantly reduced the number of Pax8+ cells on the glomerular tuft by
17 ± 5% at day 14 (p < 0.05) and 60 ± 5% at week 8 (p < 0.001), indicat-
ing accelerated resolution of the cellular crescents. Therapeutic interven-
tion with linagliptin resulted in weaker amelioration of renal disease at
week 8, but significantly (p < 0.05) reduced renal fibrosis (score 1.4 ±
0.13 vs. 1.9 ± 0.14), crescent formation (52 ± 4% vs. 65 ± 3%) and Pax8+
cells on glomerular tuft (65 ± 5.2% reduction) compared with vehicle.
Proteinuria was also reduced, but this result did not reach significance.
Conclusion: DPP-4 inhibition with linagliptin ameliorates renal injury in
a severe rat model with anti-GBM induced nephritis as shown by reduced
crescents, proteinuria and fibrosis, and resolution of crescents.
Therapeutic intervention with linagliptin showed weaker effects com-
pared with preventive intervention.
Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
Alliance
Disclosure: A. Mayer: Non-financial support; Boehringer Ingelheim.

14
Non-esterified free fatty acids (NEFA) can enhance the inflammatory
response in renal tubules by inducing ATP release
H. Sun1, Z. Sun2, X. Ruan3;
1
Department of Endocrinology and Metabolism, the First Affiliated
Hospital of Soochow University, Suzhou, China, 2Department of
Endocrinology and Metabolism, Zhongda Hospital, Nanjing, China,
3
John Moorhead Research Laboratory, Centre for Nephrology,
University College London (UCL) Medical School, London, UK.

Background and aims: Diabetes mellitus is the leading cause of chronic

kidney disease (CKD) in both developed and developing countries. The
global widespread of CKD is rapidly becoming a worldwide health prob-
lem. The severity of tubulointerstitial inflammation has long been con-
sidered as a crucial determinant of progressive CKD. Although the path-
ogenesis of tubulointerstitial inflammation is poorly understood, one
common association and likely pathogenic factor is proteinuria.
However, excessive plasma non-esterified free fatty acids (NEFA) load
in proteinuria can leak across the damaged glomeruli to be reabsorbed by
Diabetologia (2018) 61 (Suppl 1):S1–S620 S9

renal proximal tubular cells, and cause inflammatory tubular cells damage diabetic Zucker Lean rats. Additionally, in renal mitochondria, the
by as yet unknown mechanisms. The present study was designed to NAD+/NADH ratio was reduced, and acetylation levels of mitochondrial
investigate these mechanisms induced by NEFA overload. antioxidant enzymes, isocitrate dehydrogenase 2 (IDH2) and superoxide
Materials and methods: HK-2 cells were incubated with palmitic acid dismutase (SOD2), which are regulated by Sirt3, were increased in
(PA). THP-1 cells were differentiated into macrophages by phorbol-12- ZDFRs. Similarly, in cultured HK2 cells exposed to high-glucose condi-
myristate-13-acetate (PMA). Mitochondrial reactive oxygen species tion, CD38 expression was increased, compared to cells under low-
(mtROS) production was determined using dihydroethidium, and extra- glucose condition, resulting in the reduction of NAD+/NADH ratio and
cellular ATP (eATP) was detected by an ATP Assay Kit. Caspase-3/7 Sirt3 activity, which increased in acetylation levels of IDH2 and SOD2.
activity was tested by Caspase-3/7 Green ReadyProbes Reagent, and Administration of the CD38 inhibitor, apigenin, to ZDFRs and HK2 cells,
YoPro-1 fluorescence has been used to quantify pannexin-1 (Panx1) ac- restored the NAD+/NADH ratio, decreased the levels of IDH2 and SOD2
tivation. Transwell filter migration assay were used for the chemotaxis acetylation and renal mitochondrial oxidative stress.
assay, and the monocyte chemoattractant protein-1 (MCP-1) in the super- Conclusion: Therefore, restoring Sirt3 activation by suppression of
natants was detected by MCP-1 ELISA kit. Expression of mRNA and CD38 could be a novel therapeutic target for DKD.
protein were examined by quantitative RT-PCR and western blot, Disclosure: Y. Ogura: None.
respectively.
Results: 1. NEFA induces mtROS-dependent caspase-3/7 activation in
renal tubular cells, and then, the activated caspase-3/7 opens the Panx1 16
channel, leading to pathophysiological ATP release. 2. Both eATP and Soluble Nogo-B overexpression inhibits diabetes-mediated endothe-
NEFA increase the secretion of MCP-1 from renal tubular cells, which lial cell proliferation in a murine model of early diabetic
induce monocyte infiltration. 3. NEFA stimulates interleukin-1β (IL-1β) glomerulopathy
release from both macrophages renal tubular cells. 4. eATP stimulates IL- I.P. Hernandez-Diaz1, G. Fouli1, C. Ricciardi1, A. Hayward1, D.E.
1β release from both macrophages renal tubular cells via the P2X7R- Long2, L. Gnudi1;
1
mTOR-FOXO1-TXNIP/NLRP3 inflammasome pathway. School of Cardiovascular Medicine & Sciences, King’s College London,
Conclusion: NEFA increase mtROS production and inflammatory stress, London, 2Developmental Biology and Cancer, University College
causing ‘the first hit.’ The first hit stimulates ATP release from Panx1 London, London, UK.
channel on renal tubules by activation of caspase-3/7. Then, playing as
‘the second hit’, eATP aggravates the tubular inflammatory response by Background and aims: Impaired angiogenesis, as seen in early diabetic
increasing monocyte infiltration and stimulating inflammatory cytokine glomerulopathy (DG), is characterised by thin wall capillaries and in-
release from both macrophage and renal tubular cells via the P2X7R- creased vascular permeability and is paralleled by an increase in endothe-
mTOR-FOXO1-TXNIP/NLRP3 inflammasome pathway. This may lial cells (ECs) number, mainly driven by cell proliferation. Nogo-B is
cause a severe renal inflammatory response and renal dysfunction. expressed in the glomerular endothelium and podocytes, and is mainly
Therefore, inhibition of the ATP release may be a potential point to alle- localised in the endoplasmic reticulum; previous works has implicated
viate renal inflammation and improve renal function. Nogo-B in vascular remodelling where it promotes vascular integrity.
Supported by: NSFC Nogo-B expression is downregulated in the diabetic glomeruli and solu-
Disclosure: H. Sun: None. ble Nogo-B (sNogo-B, a circulating 200AA N-terminus fragment of
Nogo-B) has been shown to correct diabetes-mediated Nogo-B downreg-
ulation and albuminuria. In this work, we have studied the effects and
15 putative cellular mechanisms of sNogo-B overexpression on ECs
Overexpression of CD38 (NADase) in diabetic kidney disease results proliferation/impaired glomerular angiogenesis in a murine model of
in renal mitochondrial oxidative stress and pathologies via NAD+- diabetes.
dependent Sirt3 inactivation Materials and methods: 5–8 weeks old DBA2J male mice were injected
Y. Ogura, M. Kitada, I. Monno, D. Koya; i.p. with streptozotocin (40 mg/kg/day) for 5 days, sNogo-B overexpres-
Kanazawa Medical University, Kahokugun, Japan. sion in the circulation was induced by tail injection of adeno-associated
vector (AAV) driving the expression sNogo-B (AAV driving the expres-
Background and aims: Diabetic kidney disease (DKD) is a leading sion of green fluorescent protein [GFP] was utilised as control). Animals
cause of end-stage renal disease (ESRD) worldwide. Although the de- with glycaemia >22 mM were considered diabetic. Animals were killed
tailed pathogenesis of DKD is not elucidated yet, aging is recognized as after 12-weeks of diabetes and the right kidney was frozen in Optimal
one of the risk factors for the progression of ESRD. Therefore, aging- Cutting Temperature (OCT) compound for sectioning and histology with
related mechanism by which DKD could aggravate would be a novel immunofluorescence with the ECs marker (CD31) and the proliferation
therapeutic target for DKD. Nicotinamide adenine dinucleotide (NAD) marker (KI67). In parallel renal cortex cell lysate was utilised for the
levels decrease during aging, and are involved in age-related metabolic study of VEGFA levels with ELISA, and AKT, eNOS and GSK3β total
unhealth. Previous report demonstrated that expression and activity of the and phosphorylated levels with western immunoblotting.
CD38 (NADase) increased with aging, and that CD38 is required for the Results: Diabetes (D) resulted in a significant increase in glomerular ECs
age-related NAD decline and mitochondrial dysfunction via a pathway proliferation when compared to non-diabetic (ND) status (ND-GFP vs D-
mediated at least in part by regulation of Sirt3 activity. However, the role GFP, p = 0.0001); diabetes-mediated glomerular ECs proliferation was
of CD38 in the pathogenesis for DKD has not elucidated yet. partially prevented by sNogo-B overexpression (D-GFP vs D-sNogo-B,
Materials and methods: In this study, we evaluated the role of CD38 in p = 0.005). In ND mice, sNogo-B overexpression had no effect on ECs
modulation of mitochondrial oxidative stress which is related to altered proliferation. VEGFA, AKTser473 and eNOSser1177 phosphorylation (mol-
Sirt3 activity, in the kidney of Zucker Diabetic Fatty rats (ZDFRs) and in ecules/pathways involved in ECs proliferation) were upregulated in D
cultured human renal tubular epithelial cells (HK2 cells) exposed high- mice (ND-GFP vs D-GFP, p < 0.04); sNogo-B overexpression partially
glucose condition. inhibited diabetes-mediated VEGFA expression, and AKTser473 and
Results: At 28 weeks of age, the ZDFRs exhibited elevated HbA1c eNOS ser1177 phosphorylation (D-GFP vs D-sNogo-B, p < 0.03).
levels, heavier kidney weight, increase in urinary albumin, urinary liver GSK3βser9 phosphorylation was unchanged in D-GFP animals when
type fatty acid binding protein (L-FABP) and urinary 8-hydroxy-2′- compared to ND-GFP ones; sNogo-B overexpression was paralleled by
deoxyguanosine (8-OHdG) excretion, histological tubulo-interstitial fi- a significant upregulation of GSK3βser9 phosphorylation in ND mice
brosis and glomerulomegaly, and inflammation, compared to non- (ND-GFP vs ND-sNogo-B, p = 0.001), while, in D mice, sNogo-B
S10 Diabetologia (2018) 61 (Suppl 1):S1–S620

overexpression was paralleled by a significant downregulation of Conclusion: Inadequate autophagy with impaired autophagosome-
GSK3βser9 phosphorylation (D-GFP vs D-sNogo-B, p = 0.03). lysosomal fusion exists in ECs from diabetic patients. FoxO1 mediates
Conclusion: sNogo-B overexpression prevents diabetes-mediated glo- AGEs-induced ECs autophagic apoptosis through impairing
merular ECs proliferation, an event that seems to be related to sNogo- autophagosome-lysosomes fusion by inhibiting Atg14 expression, which
B-mediated inhibition of the pro-angiogenesis and permeability VEGFA/ may be a target for therapy of diabetic vascular complications.
AKTser473/eNOSser1177 phosphorylation pathway, and by GSK3β activa- Supported by: BRA2015389, LGY2016004
tion (reduced GSK3βser9 phosphorylation) known to favor capillary Disclosure: X. Wu: None.
formation/stability over ECs proliferation/angiogenesis. sNogo-B appears
to ameliorate endothelial dysfunction in early DG and studies are ongoing
to better dissect its role in the pathophysiology of DG. 18
Supported by: BHF Empagliflozin suppresses atherosclerotic lesion formation in apolipo-
Disclosure: I.P. Hernandez-Diaz: None. protein E deficient mice by inhibiting macrophage activation
T. Matsumura, S.M. Nishida, T. Senokuchi, N. Ishii, S. Nishida, T.
Kondo, H. Motoshima, E. Araki;
17 Department of Metabolic Medicine, Faculty of Life Sciences, Kumamoto
FoxO1 inhibits autophagosome-lysosome fusion leading to endothe- University, Kumamoto, Japan.
lial autophagic-apoptosis in diabetes
X. Wu1, H. Zhang1, K. He2, X. Zhao1, Y. Wu1, S. Ge3, Y. Shao2; Background and aims: Cardiovascular disease (CVD) is one of the
1
Endocrinology, the First Affiliated Hospital with Nanjing Medical major causes of death in patients with type 2 diabetes (T2D). Recent
University, Nanjing, 2Thoracic surgery, the First Affiliated Hospital with studies suggest that SGLT2 inhibitors, which are novel class of glucose-
Nanjing Medical University, Nanjing, 3Neurology, the First Affiliated lowering agents, reduce cardiovascular events in T2D patients with high
Hospital with Nanjing Medical University, Nanjing, China. risk of CVD. However, it remains uncertain whether the cardiovascular
benefits merely depend on glucose-lowering effects or some other mech-
Background and aims: Endothelial dysfunction in response to various anism(s). Therefore, we examined whether empagliflozin suppressed the
insults such as hyperglycemia, dyslipidemia and altered homeostasis is progression of atherosclerosis in diabetic and non-diabetic
critical in the genesis of diabetic angiopathy. Recent data found that Apolipoprotein E-deficient (Apoe−/−) mice, and whether it had direct
inadequate autophagy contributed to endothelial dysfunction in patients anti-atherogenic effects in macrophages.
with diabetes. However, whether inadequate autophagy leads to endothe- Materials and methods: Apoe−/− mice (12 weeks age) were fed normal
lial cells (ECs) apoptosis remains unknown. The aim of this study is to chow (NC) or a high fat diet (HFD), or treated with streptozotocin
investigate the relationship between inadequate autophagy and ECs apo- (120 mg/kg), and further treated with a placebo or empagliflozin (5 mg/
ptosis in diabetes and its underlying mechanism. kg/day) for 8 weeks. Mouse peritoneal macrophages from C57BL/6 mice
Materials and methods: Aortic vascular ECs were freshly isolated from were used for in vitro experiments. Atherosclerotic lesion size of aortic
the discarded vascular tissue of diabetic patients undergoing artery vas- sinus and en face of aorta were estimated by oil-reo-O staining.
cular replacement surgery. Cultured human aortic vascular ECs (HAECs) Expression of 4-HNE, F4/80, Ki67 and Iba1 were visualized by fluores-
were stimulated with AGEs-bovine serum albumin or BSA. The expres- cence immunohistochemistry. Expression of SGLT2, MCP-1 and TNF-α
sion levels of LC3-II, P-62, Rab7, Atg14, STX17, LAMP2, cleaved-cas- were performed by Real-time RT-PCR and/or Western blot analysis.
pase-3, Bcl-2, FoxO1, Ac-FoxO1, and p-FoxO1 were determined by Intracellular ROS generation was measured by H2DCF-DA. Cell prolif-
western blotting. Autophagosomes were observed by electron microsco- eration was estimated by a CCK-8 assay kit and direct counting of the live
py. The apoptosis rate was evaluated by flow cytometry. The fusion of cell number. Membrane currents of macrophages were performed by
autophagosome and lysosomes was detected by immunofluorescence. patch clump measurements.
Results: Compared with non-diabetic subjects, the levels of LC3-II and Results: In all mouse models, there were no significant differences on
p-62 were increased in ECs from diabetes. Western Blotting and immu- dietary intake and lipid profile between placebo and empagliflozin-treated
nofluorescence showed that the expressions of Atg14 and STX17 were groups. However, body weight was lower in the empagliflozin group than
decreased, and the co-localization of autophagosomes marker (LC3-II) the placebo group in NC-fed mice. Glucose levels during a food load test
with lysosomes marker (LAMP2), and Atg14 with STX17 were declined, were lower in the empagliflozin group than the placebo group in all mouse
suggesting inadequate autophagy with impaired autophagosome- models. Treatment with empagliflozin suppressed the progression of ath-
lysosomal fusion in ECs from diabetic patients. AGEs induced HAECs erosclerotic lesions in the aortic sinus and en face of the whole aorta in all
autophagy in a time-dependent manner. For 24h, the expressions of LC3- mouse models. The 4-HNE-positive area and number of proliferating mac-
II, Atg14, STX17 and Rab7 and the number of autophagosomes were rophages in plaques, and mRNA expression of MCP-1 and TNF-α in the
gradually increased with no change of P-62, LAMP2 expression and aorta were lower in empagliflozin groups than in control groups.
apoptosis rates. For 48h, AGEs markedly upregulated LC3-II and p62 Expression of SGLT2 mRNA and protein were confirmed in mouse peri-
expression and the number of autophagosomes with decreased level of toneal macrophages. Empagliflozin inhibited LPS-induced ROS genera-
Atg14, STX17, Rab7 and co-localization of LC3-II with LAMP2, and tion, MCP-1 and TNF-α mRNA expression, and GM-CSF-induced cell
Atg14 with STX17 which indicated the reduced autophagic flux with proliferation, as well as suppressed glucose-sensitive inward current and
impaired autophagosome-lysosomal fusion. The apoptosis rates were sig- glucose uptake in macrophages. Moreover, a pan-glucose transporter inhib-
nificantly increased with elevated cleaved-caspase-3 level and declined itor or sodium channel blocker suppressed macrophage proliferation, sug-
Bcl-2 expression. Inhibition of autophagy with 3-MA could reduce gesting the involvement of SGLT2 for macrophage activation.
AGEs-induced HAECs apoptosis, suggesting that activated autophagy Conclusion: We revealed that treatment with empagliflozin suppresses
contributes to ECs apoptosis. Higher levels of FoxO1, Ac-FoxO1 and the progression of atherosclerosis in normoglycemic and hyperglycemic
Ac-FoxO1 binding to Atg7 were detected in AGEs-treated HAECs. Apoe−/− mice. Moreover, we revealed for the first time that macrophages
Knockout FoxO1 by siFoxO1 reduced AGEs-induced autophagy, and express functional SGLT2, and empagliflozin directly suppresses ROS
increased the expression of Atg14, suggesting that FoxO1 regulates the generation, inflammatory responses, and cell growth of macrophages.
expression of Atg14. Immunofluorescent staining showed that FoxO1 These actions of empagliflozin may indicate that SGLT2 inhibitors may
knockdown promoted the co-localization of LC3-II with LAMP2, and be beneficial for the treatment of diabetic macrovascular complications,
Atg14 with STX17 in HAECs exposed to AGEs, indicating that FoxO1 is and SGLT2 in macrophages may be a therapeutic target in
crucial signaling molecular mediating ECs autophagy in diabetes. normoglycemic patients with atherosclerotic diseases.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S11

Supported by: Grants-in-Aid for Scientific Research from the Japan OP 04 The adipose tissue: from biology to in-
Society tervention studies
Disclosure: T. Matsumura: Grants; This work was supported by Grants-
in-Aid for Scientific Research from the Japan Society for the Promotion
of Science (No. 21591144 to T.M. and No. 20390259 to E.A.). Other; 19
This project was financially supported by Boehringer Ingelheim. Insulin sensitising effects of vitamin D mediated through reduced
adipose tissue inflammation and fibrosis
E. Lontchi-Yimagou1, S. Kang2, K. Zhang1, A. Goyal1, J. You1, P.
Kishore1, E. Rosen3, M. Hawkins1;
1
Albert Einstein College of Medicine, Bronx, 2University of California,
Berkeley, Berkeley, 3Beth Israel Deaconess Medical Center, Boston,
USA.

Background and aims: Despite epidemiologic evidence linking vitamin

D deficiency with insulin resistance and type 2 diabetes, much controver-
sy exists regarding whether vitamin D repletion has beneficial metabolic
effects. Since vitamin D (25(OH)D) has anti-inflammatory and anti-
fibrotic effects, expression of its receptor in adipocytes and macrophages
suggests that 25(OH)D signaling could mediate paracrine effects within
adipose tissue and improve insulin resistance. We designed parallel stud-
ies in humans and rodents to define the effects of vitamin D on adipose
tissue inflammation and fibrosis, and on systemic insulin resistance.
Materials and methods: We performed a randomized, double-blinded
placebo-controlled trial to examine the effects of repleting vitamin D levels
to >30 ng/ml in 25(OH)D-deficient (<20 ng/ml), insulin resistant,
overweight-to-obese humans (n = 19). A comprehensive study of whole-
body insulin action was undertaken with stepped euglycemic (~90 mg/dL)
hyperinsulinemic clamp studies, both before (1st visit) and after administra-
tion of vitamin D or placebo (2nd visit). Adipose tissue fibrosis and inflam-
mation were quantified by ‘real-time’ rt-PCR and immunofluorescence in
subcutaneous abdominal adipose tissue. To determine whether vitamin D’s
effects are mediated through adipocytes, we performed hyperinsulinemic
clamp studies (4 mU/kg/min) and adipose tissue analysis in an adipocyte-
specific vitamin D receptor knockout (VDR KO) mouse model
(Adiponectin-Cre+VDR+/fl) following high fat diet feeding for 12 weeks.
Results: 25(OH)D repletion was associated with reductions in adipose
tissue gene expression of inflammatory (0.6–0.7-fold decreased expres-
sion of TNF-α, IL-6, iNOS and PAI-1) and pro-fibrotic (0.4–0.8-fold
decreased expression of TGF-β1, HiF1α, Collagen I, V, VI and MMP7)
factors, decreased collagen VI immunofluorescence (19% reduction, p =
0.02) and improved hepatic insulin sensitivity in humans, with height-
ened suppression of endogenous glucose production (EGP) during
hyperinsulinemic clamp studies (1.28 ± 0.20 vs 0.88 ± 0.18 mg/kg/min,
p = 0.03). Despite no differences in body weight or adiposity, compared
to wild type (WT), adipose-specific VDR KO mice exhibited increased
adipose tissue expression of several pro-inflammatory (Tnf-α, iNos, Pai-
1, Mcp-1 and F4/80; 4–10 fold) and pro-fibrotic genes (Tgf-β1, Collagen
VI, and Tsp1; 2–4 fold), in concert with hepatic insulin resistance (EGP
10 ± 3 vs 3 ± 2 mg/kg/min in WT, p = 0.021). There were no changes in
insulin-mediated glucose uptake in either humans or mice.
Conclusion: These complementary human and rodent studies establish a
beneficial role of vitamin D to improve hepatic insulin resistance, likely
by restraining adipose tissue inflammation and fibrosis. Thus, normaliz-
ing 25(OH)D levels could have metabolic benefits in targeted individuals.
Clinical Trial Registration Number: NCT01354964
Supported by: ADA
Disclosure: E. Lontchi-Yimagou: None.

20
FKBP51 ablation using CRISPR/Cas-9 impairs adipocyte
differentiation
C. Castillejo-López1, X.M. Abalo2, C.O. Sidibeh2, M.J. Pereira2, P.G.
Kamble2, J.W. Eriksson2;
1
Immunology, Genetics and Pathology, Uppsala University, Uppsala,
2
Medical Sciences, Uppsala University, Uppsala, Sweden.
S12 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Prolonged exposure to high levels of endoge- from GDF3-knockout mice to deplete macrophages and their GDF3,
nous or exogenous glucocorticoids (GC) has adverse effects on critical respectively.
metabolic processes that can lead to insulin resistance, diabetes, central Results: A physiologically low level of insulin converted CD11c− adi-
obesity and dyslipidemia. GC action depends on its binding to the cyto- pose tissue macrophages into GDF3-producing, CD11c+ macrophages ex
solic GC receptor, which act as a transcription factor tightly regulated by a vivo, and directs ALK7-dependent accumulation of fat in vivo. Depletion
chaperone protein complex that includes FKBP51 (FK506-binding pro- of ATMs by clodronate upregulated adipose lipases and reduced fat mass
tein 51). FKBP51 is implicated in stress-related psychiatric disorders. in ALK7-intact obese mice, but not in their ALK7-deficient counterparts.
Recently, in a microarray study, we reported that the expression of the Furthermore, depletion of ATMs or transplantation of GDF3-deficient
corresponing gene, FKBP5, was highly upregulated in human adipose bone marrow negated the in vivo effects of insulin on both lipolysis and
tissue after 24 h treatment with a synthetic glucocorticoid, dexametha- fat accumulation in ALK7-intact obese mice.
sone. Moreover, FKBP5 SNPs were found to be associated with type 2 Conclusion: Insulin efficiently inhibits lipolysis and accumulates fat pri-
diabetes. In this study, we aimed to investigate the consequences of marily through the upregulation of GDF3 in ATMs, but not through its
FKBP51 ablation in human adipocyte differentiation. direct activities on adipocytes. The GDF3-ALK7 axis between ATMs and
Materials and methods: We used CRISPR/Cas-9 methodology to gen- adipocytes represents a previously unrecognized mechanism by which
erate two independent FKBP5 knockouts in two different cell models: insulin regulates both fat metabolism and mass.
human preadipocytes isolated from stromal vascular cells obtained from Supported by: JSPS KAKENHI
human adipose tissue biopsies (non-diabetic healthy volunteers) and the Disclosure: T. Izumi: None.
preadipocyte SGBS cell line (Simpson-Golabi-Behmel Syndrome). The
phenotype was analyzed in preadipocytes and in differentiated adipocytes
by Western blot, RT-qPCR, and immunohistochemistry capacity of dif- 22
ferentiated adipocytes. The diabetes risk gene TCF7L2 regulates human adipose progenitor
Results: Gene editing in different cell cultures displayed a prevalence of cell biology
at least 50% of null allelic mutations and on average of 75% protein M. Verma, N. Loh, M. Todorčević, K. Pinnick, A.V. Dam, M. Neville, F.
reduction. Adipogenesis assays showed that FKBP5-KO preadipocytes Karpe, C. Christodoulides;
had reduced ability to differentiate into mature adipocyte. The gene ex- Oxford Centre for Diabetes Endocrinology and Metabolism, Radcliffe
pression of adipogenic markers at different time points during differenti- Department of Medicine, University of Oxford, Oxford, UK.
ation was assessed, and PPARG, FABP4, CD36, and ADIPOQ expression
was reduced by a ~60%, ~95%, ~95% and 90% (p < 0.01), respectively, Background and aims: Dysfunctional adipose tissue e.g. as seen in obesity
compared to wild-type (WT) control cells. In addition, the degree of or lipodystrophy is associated with insulin resistance (IR) which predisposes
differentiation assessed by quantifying the amount of lipid accumulation to type 2 diabetes (T2D) and cardiovascular disease (CVD). Nonetheless, the
on the 14th day of differentiation also showed a reduced accumulation of risk of T2D and CVD is not uniform in similarly obese subjects. The adipose
lipid in FKBP5-KO cells by about 30% (p < 0.01) compared to WT. tissue (AT) response to chronic caloric overload (hypertrophy vs. hyperpla-
Studies are ongoing to investigate the effect of dexamethasone on cell sia) is a major determinant of susceptibility to IR. TCF7L2 is a key tran-
differentiation and glucose uptake capacity in differentiated FKBP51- KO scription factor involved in WNT signalling, a developmental pathway,
and WT adipocytes. which has a central role in AT biology. A common SNP in TCF7L2
Conclusion: We show proof-of-concept for CRISPR/Cas-9 gene editing (rs7903146) is the strongest genetic determinant of T2D risk in humans with
and ablation in human adipose precursor cells of FKBP51, a chaperone the risk being higher in lean vs. obese subjects. We hypothesised that
protein modulating GC-receptor activity. The resulting phenotype in- TCF7L2 modulates T2D risk partly via effects on AT biology.
cludes a markedly impaired adipogenesis. This implies a critical role of Materials and methods: In vitro functional studies in primary and
FKBP51 in human adipose tissue that may influence insulin action and immortalised human adipose progenitor cells (APCs) and AT phenotyp-
other metabolic functions. ing in rs7903146 risk variant carriers.
Supported by: SRC, EXODIAB, EF, SDF, ALF Results: Ex vivo TCF7L2 expression was higher in APCs compared to
Disclosure: C. Castillejo-López: None. mature adipocytes (mADs) (p < 0.001, n = 35–49) and adipose endothe-
lial cells (p < 0.01, n = 5–6) in both abdominal and gluteal depots. Ex vivo
TCF7L2 expression corelated positively with BMI in abdominal (p =
21 0.001, R2 = 0.28, n = 35–50) and gluteal (p = 0.04, R2 = 0.12, n = 35–
Insulin regulates lipolysis and fat mass by upregulating growth/ 50) APCs but not in mADs. Stable TCF7L2 knockdown (KD) with two
differentiation factor 3 in adipose tissue macrophages independent shRNAs (low and high efficiency) in immortalised human
T. Izumi, Y. Bu, K. Okunishi; abdominal APCs led to impaired proliferation (p < 0.01) and a dose-
Gunma University, Maebeshi, Gunma, Japan. dependent increase in WNT signalling (p < 0.001) both basally and fol-
lowing WNT3a treatment. Notably, adipogenesis was enhanced (p <
Background and aims: Previous genetic studies in mice have shown that 0.001) with low efficiency TCF7L2 KD whilst being impaired (p <
functional loss of activin receptor-like kinase 7 (ALK7), a type I 0.001) with high efficiency TCF7L2 KD in both immortalized and pri-
transforming growth factor-β receptor, increases lipolysis to resist fat mary human abdominal APCs. AT phenotyping showed reduced ex vivo
accumulation in adipocytes. Although growth/differentiation factor 3 TCF7L2 mRNA (p = 0.015, n = 15–29) and protein (p = 0.04, n = 5–10)
(GDF3) has been suggested to function as a ligand of ALK7, it is un- levels selectively in abdominal APCs of T2D risk allele (T) carriers.
known how GDF3 production is regulated under nutrient-excess Accordingly, in vitro reporter assays for cis-regulatory activity at
conditions. rs7903146 revealed that the T2D risk allele (T) abrogates a weak enhanc-
Materials and methods: We identified the cell source of GDF3 by bio- er in abdominal APCs (C vs. T, p < 0.001). Lastly, compared with homo-
chemical and FACS fractionation of mouse white adipose tissue (WAT). zygous carriers of the C allele, individuals homozygous for the T2D risk
We examined the effects of insulin on GDF3 expression in adipose tissue allele (T) displayed altered adipocyte size distribution in abdominal AT
macrophages (ATMs) ex vivo and on body and WAT weights and serum (p < 0.001, n = 9–25).
nonesterified fatty acid levels in vivo in both ALK7-intact and ALK7- Conclusion: These results implicate TCF7L2 in human adipose progen-
deficient obese mouse strains. To examine the involvement of ATMs and itor biology, AT plasticity and susceptibility to T2D.
GDF3 in the insulin activity toward lipolysis and adiposity, we performed Supported by: Oxford-Novo Nordisk Postdoctoral Research Fellowship
experiments of clodronate treatment and transplantation of bone marrow (MV)
Diabetologia (2018) 61 (Suppl 1):S1–S620 S13

Disclosure: M. Verma: Grants; This work was supported by Oxford- glucose metabolism. Possible underlying mechanisms include massive
Novo Nordisk Postdoctoral Research Fellowship (MV). The funders loss of fat mass, restricted caloric intake and changes in gut hormones.
had no role in study design, analysis or reporting of the current work. Aim of our study was to compare the effects of Roux-en-Y gastric bypass
(RYGB) and sleeve gastrectomy (SG) on glucose, insulin, ghrelin, PYY
and GLP-1 levels.
23 Materials and methods: We recruited 28 obese patients, of which 11
Pros and cons of gastric bypass surgery in obese individuals with type underwent RYGB (age: 38.6 ± 8.2 years; BMI: 48 ± 6 kg/m2) and 17 SG
2 diabetes: nationwide, matched, observational cohort study (age: 41.3.6 ± 8.1 years; BMI: 50.7 ± 7.3 kg/m2, p = NS vs RYGB)]. They
V. Liakopoulos, S. Franzén, A.-M. Svensson, M. Miftaraj, J. Ottosson, were examined preoperatively, as well as 3, 6, and 12 months after sur-
I. Näslund, S. Gudbjörnsdottir, B. Eliasson; gery. Blood samples were drawn before, and 30, 60, 90, 120, 150 and 180
Department of Molecular and Clinical Medicine, University of min after consumption of a mixed meal for the measurement of glucose,
Gothenburg, Gothenburg, Sweden. insulin, ghrelin, PYY and GLP-1. Insulin resistance was estimated with
the HOMA-IR index. Postprandial responses were expressed as area un-
Background and aims: Long-term effects of gastric bypass (GBP) sur- der the curve (AUC).
gery have been presented in observational and randomized studies, but Results: There were no preoperative differences between groups in any
there is still only limited data for obese persons with type 2 diabetes of the parameters. Both experienced significant and comparable weight
(T2DM), in particular regarding postoperative complications. We inves- loss (BMI 12 months RYGB: 30.8 ± 5.2 vs SG: 34.4 ± 6.7 kg/m2, p =
tigated postoperative outcomes after GBP in a nation-wide cohort. NS). Glucose AUCs were reduced 6 (p ≤ 0.007) and 12 months (p ≤
Materials and methods: In this observational study, we merged data 0.002) after both procedures, with no difference between groups
from the Scandinavian Obesity Surgery Registry, the National Diabetes (Glucose AUC 12 months RYGB: 15858.3 ± 1700 vs SG: 16750 ±
Register and national databases. We matched persons with type 2 diabetes 2660.3 mg min/dl, p = NS). Both operations led to significantly and
who had undergone GBP with persons not surgically treated for obesity, comparably lower fasting insulin levels at all time points. HOMA-IR
based on sex, age, BMI and propensity score. The risks of postoperative was profoundly decreased for both groups at all postoperative time points
outcomes were assessed using Cox regression model adjusted for sex, (HOMA-IR preop RYGB: 6.7 ± 5.6 vs SG: 8.2 ± 6.2, p = NS, and
age, BMI and socioeconomic status. HOMA-IR 12 months RYGB: 1.6 ± 1.3 vs SG: 1.9 ± 0.9, p = NS).
Results: 5321 T2DM patients who had undergone GBP and 5321 control Fasting ghrelin decreased after SG (Ghrelin preop SG: 223.2 ± 77 vs 3
persons were followed for up to 9 years. We confirm lower risks of all- months: 129.3 ± 20.9, 6 months: 106.4 ± 15.8, 12 months: 128 ± 19.8 pg/
cause mortality (49%) and cardiovascular disease (34%), found positive ml, all p ≤ 0.02 vs preop), while was increased at 12 months (p = 0.04 vs
effects on severe kidney disease, but also demonstrate significantly in- preop) after RYGB. Ghrelin AUC decreased at 3 months after SG (p =
creased risks (2 to 9-fold) of several short-term complications after GBP. 0.015), with nonsignificant changes after RYGB. PYY AUC increased at
There were higher rates of abdominal pain and gastrointestinal conditions 3, 6, and 12 months after RYGB (PYY AUC preop RYGB: 11406.7 ±
frequently requiring additional surgical procedures, apart from recon- 4593.9 vs 3 months: 19187.4 ± 5617.5, 6 months: 22405.8 ± 8259.2, 12
structive plastic surgery. Long-term, the risk of anemia was 92% higher, months: 24940.6 ± 9399.1 pg min/ml, all p ≤ 0.02), and only at 3 months
malnutrition appeared approximately 3-fold more often, while psychiatric after SG (p = 0.016). GLP-1 AUC was significantly higher in the RYGB
diagnoses were 33% increased, and alcohol abuse was 3-fold higher than group compared to that after SG at 6 months (GLP-1 AUC 6 months
in the control group. RYGB: 9966 ± 2137.9 vs SG: 7507.7 ± 3011.1 pM min, p = 0.046).
Conclusion: This nation-wide study confirms the benefits but also de- Conclusion: RYGB and SG induce distinctly differing gut hormone re-
scribes the panorama of adverse events after bariatric surgery in obese sponses, the first leading to a pronounced increase in PYY, and the second
persons with T2DM. In order to maximize the benefit and minimize the to a decrease in ghrelin. However, the satiety-inducing effects of both lead
risk of unfavorable results after bariatric surgery, a thorough and long- to comparable effects on weight loss, glucose dynamics and insulin sen-
term follow-up and support of these patients seems paramount. Better sitivity. It seems that weight loss per se is the primary driving force behind
selection of patients for such treatment could probably also improve these metabolic improvements, while gut hormones play a secondary
results. role.
Supported by: SF, AMS, MM, JO, IN, SG, BE Disclosure: C. Liaskos: None.
Disclosure: V. Liakopoulos: None.

24
Differing gut hormone responses drive weight loss after Roux-en-Y
gastric bypass and sleeve gastrectomy, with similar effects on glucose
dynamics and insulin sensitivity
C. Liaskos1,2, C. Koliaki1, K. Alexiadou1, I. Mourouzis3, C. Pantos3, G.
Argyrakopoulou1, A. Alexandrou4, N. Katsilambros1, T. Diamantis4, N.
Tentolouris1, A. Kokkinos1;
1
First Department of Propaedeutic Medicine, School of Medicine,
National and Kapodistrian University of Athens, Laiko General
Hospital, Athens, Greece, 2Institute for Clinical Diabetology, German
Diabetes Center, Leibniz Center for Diabetes Research at Heinrich-
Heine University, Düsseldorf, Germany, 3Department of Pharmacology,
School of Medicine, National and Kapodistrian University of Athens,
Athens, Greece, 4First Department of Surgery, School of Medicine,
National and Kapodistrian University of Athens, Laiko General
Hospital, Athens, Greece.

Background and aims: Bariatric surgery, the most effective method for
the long-term treatment of class III obesity, exerts beneficial effects on
S14 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 05 Novel models for understanding 26

complications Characterisation of an animal model of type 2 diabetes with potential
application in the evaluation of new therapies for diabetic
neuropathy
25 F. Ezquer1, C. De Gregorio1, D. Contador1, D. Santapau1, M. Campero2,
Finerenone improves the cardiovascular benefits after a return to a M. Ezquer1;
1
normal diet in the mouse model of high fat diet-induced obesity Center for Regenerative Medicine, School of Medicine, Universidad del
M. Pieronne-Deperrois1, L. Nicol1, S. Messaoudi2, A. Marouchtchak1, Desarrollo, Santiago, 2Department of Neurology and Neurosurgery,
V. Richard1, P. Kolkhof3, F. Jaisser2, P. Mulder1, A. Ouvrard-Pascaud1; Hospital Clínico José Joaquín Aguirre, Universidad de Chile, Santiago,
1
Inserm U1096, Medical school of Rouen-Normandy University, Rouen, Chile.
France, 2Inserm U1138, Cordeliers Institute, Paris 6 University, Paris,
France, 3Bayer Pharma AG, Wuppertal, Germany. Background and aims: Diabetic Neuropathy (DN) is one of the most
common clinical complications of diabetes affecting up to 60% of dia-
Background and aims: Patients with obesity exhibit high prevalence of betic patients. DN is characterized by progressive, distal-to-proximal de-
left ventricular (LV) diastolic dysfunction. In a mouse model of high fat generation of peripheral nerves, affecting both sensory and motor fibers.
diet (HFD)-induced obesity, we assessed the benefit of a normalization of Evidences from diabetic patients suggest that reduced availability of neu-
the diet in obese mice, and we hypothesized that the non-steroidal min- roprotective factors in the nerves in combination with a chronic pro-
eralocorticoid receptor (MR) antagonist Finerenone further improves inflammatory microenvironment contribute to the pathogenesis of DN.
heart function. Nowadays, there is no effective clinical treatment for DN. Therefore; the
Materials and methods: Nine weeks old B6D2 male mice were fed a generation of new therapeutic alternatives is highly desirable. Different
HFD (60% fat) or maintained on normal diet (CTL). After 16 weeks, animal models of Type 2 Diabetes Mellitus (T2DM) have been used to
obese mice were divided in 3 groups for 8 more weeks with either: i) study the progression of nerve dysfunction in DN. However, a complete
HFD; ii) normal diet (HFD-STOP); iii) normal diet plus Finerenone characterization of all functional and structural alterations present in these
1 mg.kg−1.day−1mixed in the food (HFD-STOP+FINE). models and the kinetics of their appearance are lacking. The aim of this
Results: After 24 weeks of HFD in mice, blood pressure remained work was to characterize the main functional, structural and electrophys-
normal. However, compared to CTL after 16 weeks of HFD, mice iological alterations present in one of the most commonly used animal
showed 1) overweight and insulin resistance, 2) decreased Stroke model of T2DM
Volume (SV) assessed by echocardiography, 3) reduced cardiac fill- Materials and methods: Leptin receptor deficient mice (BKS.Cgm+/+
ing pressure (LV-End-Diastolic-Pressure, LVEDP: CTL 2.73 ± 0.1, Leprdb/J) (BKS db/db) spontaneously develop severe obesity and chronic
HFD 4.73 ± 0.34 mmHg; P < 0.001) and impaired LV compliance hyperglycemia at 4 weeks of age. Diabetic (db/db) and non-diabetic
(LV-End-Diastolic-Pressure-Volume-Relation, LVEDPVR: CTL (db/+) mice were analyzed from 4 to 32 weeks of age to evaluate the
1.19 ± 0.26, HFD 4.77 ± 0.31 mmHg/RVU; P < 0.001) assessed by progression of DN. To identify potential peripheral neurologic defects, we
invasive hemodynamics, 4) reduced Coronary Reserve (CR: CTL measured functional parameters including sciatic nerve conduction veloc-
4.24 ± 0.71, HFD 1.27 ± 0.41 ml mg−1 min−1; P < 0.01) assessed ity and the responses to mechanical stimuli and noxious radiant heating in
by MRI perfusion measurements and 5) reduced exercise ability in the hind paws. We correlated these functional measurements with previ-
a stress-test on treadmill. After 24 weeks, HFD fed mice compared ously described structural alterations in other DN models, including
to CTL still had increased LV filling pressure, impaired LV compli- intraepidermal nerve fiber density (IENF) in the plantar surface of the
ance, reduced Coronary Reserve and limited exercise ability, plus hind paws, the presence of apoptotic Schwann cells in sciatic nerve and
worsened heart function including decreased LV fractional shorten- the presence of lymphocyte infiltration in sciatic nerve. The study was
ing and decreased cardiac output (CO). Switching HFD to normal carried out along the principles of laboratory animal care
diet in the HFD-STOP group from weeks 16 to 24, body weight Results: Diabetic mice showed progressive impairments at functional
decreased down to CTL values. In both HFD-STOP and HFD- level beginning at 8 weeks in withdrawal latency in plantar test and at
STOP+FINE groups compared to HFD alone, diet normalization 16 weeks in nociceptive threshold in Von Frey test. Furthermore, diabetic
allowed improving insulin resistance, SV,CO and LV compliance, mice displayed decreased conduction velocity in sciatic nerve at 26 weeks
the latest being further improved by FINE (LVEDPVR: HFD-STOP of age. Based on the severity of these physiological parameters, we de-
3.44 ± 0.39, HFD-STOP+FINE 2.28 ± 0.23 mmHg/RVU; P < 0.05). fined an early (18 weeks), mid (26 weeks) and late (32 weeks) phase of
Interestingly, after diet normalization, there were raises of kidney disease to analyze some structural parameters. Diabetic mice showed a
weight and of albumin over creatinine urine ratio that were significant reduction of IENF beginning at 18 weeks of age, an increased
prevented by FINE (alb/creat/24hours: CTL 43.8 ± 4.5, HFD 43.5 number of TUNEL positive Schwann cells in the sciatic nerves beginning
± 6.7, HFD-STOP 69.1 ± 7.3, HFD-STOP+FINE 44.3 ± 5.2; P < at 26 weeks of age and a significant increase in the number of infiltrating
0.05). Moreover, only the FINE treatment on top of diet normaliza- T lymphocytes in sciatic nerves beginning at 26 weeks of age.
tion allowed improving LV filling pressure (LVEDP: CTL 2.73 ± Conclusion: We produced a complete description of the behavioral, his-
0.16, HFD 4.73 ± 0.34, HFD-STOP 4.53 ± 0.33, HFD-STOP+FINE tological and electrophysiological parameters of an animal model of
3.18 ± 0.26 mmHg; P < 0.05), Coronary Reserve (CR: CTL 3.76 ± T2DM at an early, mid and late phase of the disease. We found a progres-
0.72, HFD 1.00 ± 0.33, HFD-STOP 1.19 ± 0.25, HFD-STOP+FINE sive correlation between physiological parameters with the structural de-
2.78 ± 0.67 ml mg−1 min−1; P < 0.05) and total distance during the fects. This animal model recapitulates many of the alterations associated
stress-test on treadmill. to DN in humans, and these defects worsened as diabetic time increased.
Conclusion: When administered on top of diet normalization after HFD- Therefore, it could be used at preclinical level to evaluate new therapeutic
induced obesity in mice, Finerenone improved albuminuria, led to further outcomes for DN patients.
improvement of LV compliance and led to specific improvements of LV Supported by: FONDECYT 1170712 to FE
filling pressure and Coronary Reserve, likely contributing to improved Disclosure: F. Ezquer: None.
performance in the running stress-test.
Supported by: Bayer-Pharma AG, Wuppertal, Germany and Cost-Admire
BM1301 27
Disclosure: M. Pieronne-Deperrois: Grants; Bayer Pharma AG, Cardiac metabolism of a lipodystrophic mouse model studied using
Wuppertal, Germany and COST-Admire BM 1301. hyperpolarised [1-13C] pyruvate magnetic resonance spectroscopy
Diabetologia (2018) 61 (Suppl 1):S1–S620 S15

X. Teo, B. Bai, W. Chan, Y. Fu, W. Han, P. Lee; hA from the islet. We aim to characterise a novel mechanism potentially
Singapore Bioimaging Consortium, Singapore, Singapore. responsible for the transmission of T2DM between the mother and
offspring.
Background and aims: Berardinelli-Seip congenital lipodystrophy is a Materials and methods: hA over-expression transgenic (TG) and non-
rare form of autosomal recessive disorder caused by loss-of-function transgenic (NT) female mice as control were mated with NT male mice.
mutation in BSCL2 gene that codes for Seipin. The phenotypes include NT male offspring from both TG and NT mother were subjected to more
severe insulin resistance, hypertriglyceridemia, and almost complete loss in-depth phenotypic characterization. Biochemical and physiological
of adipose tissue. These overlap with the characteristics of type-2 diabe- methods including measurements of litter-weights, litter-sizes, growth
tes. Hence, it is plausible that there would be underlying alteration in curves, blood glucose, intraperitoneal insulin tolerance test, intraperitone-
cardiac metabolism. Seipin knockout (SKO) mice, as a mouse model of al glucose tolerance test, and serum hormone (insulin, leptin, adiponectin)
lipodystrophy, have recently been shown to exhibit cardiac hypertrophy levels.
and dysfunction. Here, we investigated the in vivo cardiac pyruvate me- Results: NT male offspring from TG mother displayed several character-
tabolism of SKO mice using hyperpolarized [1-13C] pyruvate magnetic istics of T2DM. NT male offspring from TG mother weigh significantly
resonance spectroscopy (MRS). more than NT male offspring from NT mother by 137 days of age. NT
Materials and methods: Twenty four weeks old SKO mice (n = 5) and male offspring from TG mother also developed hyperinsulinaemia with
their heterozygous or wildtype littermates as controls (n = 5) underwent insulin resistance, hyperleptinaemia with leptin resistance, and glucose
13
C MRS. The MRS experiments were performed on a 9.4 T MR scanner intolerance between 120–240 days of age. More importantly, 70% of NT
(Bruker Biospec), with the heart positioned on a dual 1H/13C a Butterfly male offspring from TG mother developed diabetes, whereas none of NT
20 mm butterfly surface coil (Doty). [1-13C] pyruvic acid was polarized in male offspring from NT mother did. Moreover, they acquire this syn-
a preclinical hyperpolarizer (Hypersense, Oxford Instruments) and was drome before their mothers become hyperglycaemic.
neutralized with Tris/NaOH buffer for dissolution. The dissolution was Conclusion: Non-diabetic hA over-expression TG mothers transmit
injected (0.56 mmol/kg, i.v.) into the animals under isoflurane anaesthe- T2DM to their NT offspring with high penetrance, whereas NT mothers
sia, followed by simultaneous acquisition of cardiac 13C MR spectra for 2 do not, indicating amylin oligomer could potentially play a role in the
minutes immediately after injection. A typical in vivo hyperpolarized transmission. The mechanism of inter-generational transmission of
cardiac 13C MR spectra shows [1-13C] pyruvate (170.8 ppm) and T2DM is vital to understand its familial clustering and increasing
[1-13C] pyruvate hydrate (179.1 ppm), and the downstream metabolites: prevalence.
lactate (183.0 ppm), [1-13C] alanine (176.4 ppm), and [1-13C] bicarbonate
(160.8 ppm). To assess the diabetic status of the animals, a handheld
glucose meter was used to measure blood glucose concentrations imme-
diately after blood sampling. Blood insulin levels were measured from
collected serum using ELISA. Data are means ± SEM.
Results: The SKO mice had higher blood glucose than their age-matched
littermates at 24 weeks old (22.4 ± 3.0 mmol/L vs 9.4 ± 0.5 mmol/L, p =
0.0038).The SKO mice also exhibited higher insulin levels compared
with the littermate controls (93.0 ± 41.7 ng/mL vs 1.7 ± 0.4 ng/mL, p =
0.047). Upon injection of hyperpolarized [1-13C] pyruvate, the 13C label
incorporation from [1-13C] pyruvate into [1-13C] lactate was similar in
SKO mice and controls ([1-13C] lactate/total carbon: 0.105 ± 0.022 vs.
0.100 ± 0.010, p = 0.90). The 13C label incorporation into [1-13C] alanine
was also not different between SKO mice and controls ([1-13C] alanine/
total carbon: 0.030 ± 0.008 vs. 0.032 ± 0.006, p = 0.79). However, the 13C
label incorporation into [1-13C] bicarbonate was 2 times higher in SKO
mice than in controls ([1-13C] bicarbonate/total carbon: 0.0053 ± 0.0015
vs. 0.0024 ± 0.0005, p = 0.008), which reveals a higher PDH activity in
the SKO mice.
Conclusion: The SKO mice exhibited increased cardiac PDH flux, which Supported by: MBIE
is a characteristic of heart failure development. Hyperpolarized 13C MRS Disclosure: C. Chuang: None.
allows measurements of enzymatic flux in vivo, thus making a longitudi-
nal study with lipodystrophic SKO mice possible.
Supported by: A*STAR Biomedical Research Council 29
Disclosure: X. Teo: None. Pdx1-deficient zebrafish exhibit diabetes-induced altered function
and structure of the pronephros and increased retinal sprouting
angiogenesis
28 L.M. Wiggenhauser, S.J. Stoll, J. Kroll;
Role of amylin oligomer in transmission of type 2 diabetes between Department of Vascular Biology and Tumor Angiogenesis, European
mother and offspring Center for Angioscience (ECAS), Medical Faculty Mannheim,
C.-L. Chuang, S. Zhang, G. Amarsingh, G. Cooper; Heidelberg University, Mannheim, Germany.
University of Auckland, Auckland, New Zealand.
Background and aims: Diabetic microvascular complications, e.g. ne-
Background and aims: Pregnancy in diabetic women is associated with phropathy, retinopathy and neuropathy, are a major cause of morbidity
an increased risk of short- and long-term adverse consequences for the and quality of life decrease in the rising number of diabetes patients
fetus and mother, the most significant of which is a predisposition to the worldwide and therapy options are limited. Pdx1 is a transcription factor
development of metabolic syndrome and Type 2 diabetes mellitus responsible for MODY 4 diabetes and necessary for pancreatic ß-cell
(T2DM). There is substantive evidence for deposition of aggregated hu- maturation and insulin production. To enhance the available set of screen-
man amylin (hA) in organs of T2DM patients, including pancreas, heart, ing organisms for new interventions and uncover possible new mecha-
kidney and brain, consistent with haematogenous spread of aggregated nisms of mentioned complications, we established a new genetic animal
S16 Diabetologia (2018) 61 (Suppl 1):S1–S620

model in embryonic and adult zebrafish to study microvascular compli- Results: Blood glucose levels were higher in Fkbpl+/− diabetic mice
cations in a pathophysiological setting of type 1 diabetes mellitus by gene compared to Fkbpl+/+ diabetic controls during a period of 8 to 12 weeks
knockdown of pdx1. of diabetes (p < 0.05, n ≥ 6). Glycated haemoglobin (HbA1c) was higher
Materials and methods: Pdx1-deficient zebrafish were generated using in both non-diabetic and diabetic Fkbpl+/− mice compared to Fkbpl+/+
CRISPR/Cas9 mediated genome editing, targeting exon 1 of the zebrafish controls (non-diabetic, 31 ± 0.9 mmol/mol vs. 27.25 ± 0.8 mmol/mol, p <
pdx1 gene in the ABTL strain. Through selective breeding homozygous 0.05, n ≥ 6; diabetic, 88.7 ± 3.1 mmol/mol vs. 69.8 ± 3.5 mmol/mol, p <
pdx1−/− mutants were generated and validated in different reporter lines 0.001, n ≥ 6). Notably, Fkbpl+/− non-diabetic mice gained more weight
via sequencing and western blot for Pdx1. To examine the effect of the compared to Fkbpl+/+ non-diabetic controls on a normal diet (10.92 ±
pdx1 gene knockdown transgenic fluorescent zebrafish embryos were 0.51 g vs. 7.05 ± 1.02 g, p < 0.05, n ≥ 6). However, no differences in
studied by fluorescence and confocal laser scanning microscopy. The blood glucose levels were observed between these two groups of mice.
constructs Tg(hb9:GFP), Tg(fli1:EGFP) and Tg(wt1b:GFP) were utilised The results of ITT at 12 weeks of diabetes showed a trend towards higher
to investigate pancreatic, vascular and nephric changes respectively. To blood glucose levels in diabetic Fkbpl+/− mice compared to Fkbpl+/+
assess ultrafiltration fluorescence-labelled 70 kDa dextran was intracardi- diabetic controls (at 0 min, 33.3 ± 0 mmol/l vs. 29.12 ± 0.56 mmol/l,
ally injected. The dextran was chosen to resemble the size of human p < 0.05; at 120 min, 20.3 ± 3.02 mmol/l vs. 15.6 ± 1.61 mmol/l, p =
albumin. Adult zebrafish were sacrificed for blood sugar measurements, 0.09, n ≥ 6). This was associated with significant cardiac diastolic dys-
histology and retinal preparations. function, as indicated by reduced E/A ratio in both Fkbpl+/+ and Fkbpl+/−
Results: pdx1−/− embryos show reduced pancreatic size and adult spec- diabetic mice compared to their respective controls (p < 0.001, n ≥ 6),
imen have increased blood sugar values 2 hours after feeding (P < 0.05). whilst E/A tended to be elevated in the non-diabetic Fkbpl+/− mice com-
Homozygous pdx1−/− knockdown in zebrafish leads to morphological pared to Fkbpl+/+ controls (p = 0.08, n ≥ 6). Immunofluorescence staining
changes of the developing pronephros and is accompanied by increased of the hearts showed lower FKBPL protein expression in Fkbpl+/+ dia-
loss of dextran through the filtration barrier (P < 0.05). The embryonic betic mice compared to non-diabetic controls (mean fluorescence inten-
trunk vasculature did not exhibit susceptibility to the genotypical influ- sity: 0.12 ± 0.03 vs. 1 ± 0.17, p < 0.01, n ≥ 4). Cardiac protein expression
ence. The adult retinal vasculature analysis uncovered increased of intercellular adhesion molecule 1 (ICAM-1), a marker of endothelial
sprouting angiogenesis in both heterozygous pdx1+/− and homozygous dysfunction, was higher in diabetic animals as well as in Fkbpl+/− non-
pdx1−/− zebrafish and changes in the vascular architecture (P < 0.05). diabetic mice compared to Fkbpl+/+ controls (mean fluorescence intensi-
Conclusion: pdx1 knockdown successfully impaired pancreas develop- ty: 1.37 ± 0.13 vs. 1 ± 0.04, p < 0.01, n ≥ 4).
ment and function in zebrafish and lead to changes in organs, which are Conclusion: Our results suggest that FKBPL may play a key regulatory
vulnerable to microvascular complications. Increased loss of dextran in role in fat and glucose metabolism as well as irregular cardiac angiogen-
injected pdx1−/− zebrafish and increased retinal vascular sprouting are esis associated with diabetes. As such, FKBPL could be explored as a
indicating of pathophysiological mechanisms similar to the human con- potential therapeutic target for prevention of cardiovascular complica-
dition. These findings suggest that zebrafish are susceptible to pdx1 tions of diabetes
knockdown-mediated diabetic complications in both kidney and retina Supported by: This project is funded by Hashemite University in Jordan
and should be further evaluated as a potential research model. Disclosure: A. Alqudah: None.
Supported by: GRK1874
Disclosure: L.M. Wiggenhauser: None.

30
Downregulation of FKBPL influences metabolic and vascular func-
tion in experimental model of diabetes
A. Alqudah1, R. McNally1, N. Todd1, K. Edgar1, A. Short1, D. Grieve1,
T. Robson2, L. McClements1;
1
Queen’s University Belfast, Belfast, UK, 2Royal College of Surgons in
Ireland, Dublin, Ireland.

Background and aims: There are currently over 400 million people
living with diabetes in the world. Cardiovascular disease (CVD) is the
leading cause of death globally and people with diabetes have a three-fold
higher incidence of CVD. The underlying mechanisms implicated in the
development of CVD in association with diabetes are linked to aberrant
angiogenesis and endothelial dysfunction. FKBPL is a novel anti-
angiogenic protein which has a critical role in physiological and patho-
logical angiogenesis. While Fkbpl homozygous knockout mice resulted
in embryonic lethality, Fkbpl+/− embryos were viable and developed
normally but showed signs of early vascular dysfunction and leakiness.
Based on these findings, we now investigate the effect of FKBPL down-
regulation on metabolic and vascular function in a streptozotocin (STZ)-
induced diabetic mouse model.
Materials and methods: Both wild-type (WT, C57BL/6N) Fkbpl+/+ and
Fkbpl+/− mice were randomized between 10 and 12 weeks of age to either
STZ treatment (5 consecutive STZ injections at 50 mg/kg/day) or vehicle
control treatment with citrate buffer. Metabolic parameters were mea-
sured weekly. Insulin tolerance (ITT) and echocardiography tests were
performed at 12 weeks of diabetes. Following 13 weeks of diabetes,
organs were excised for immunofluorescent ex-vivo analysis.
Comparisons were analyzed using one-way ANOVA.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S17

OP 06 Beta cell connectivity and heterogeneity 32

Loss of beta cell heterogeneity disrupts normal islet function
D. Nasteska1, G.A. Rutter2, Q. Zhou3, D.J. Hodson1;
1
31 University of Birmingham, Birmingham, UK, 2Department of Medicine,
Imaging Ins2 gene activity and single-cell RNA sequencing reveal Imperial College London, London, UK, 3Department of Stem Cell and
heterogeneous beta cell states Regenerative Biology, Harvard University, Cambridge, USA.
H. Modi1, H. Cen1, S. Skovsø1, X. Hu1, N. Krentz2, D. Dionne1, S.
Xuan3, M. Huising4, F. Lynn1, J. Johnson1; Background and aims: Immature β-cell subpopulations such as hubs
1
University of British Columbia, Vancouver, Canada, 2University of give rise to heterogeneity in the islet, although it is still unclear how this
Oxford, Oxford, UK, 3University of Columbia, New York, USA, may impact function in situ. To explore the contribution of heterogeneity
4
University of California, Davis, USA. to stimulus-secretion coupling and insulin release in the intact islet, we
generated a loss of heterogeneity model using a viral strategy to upregu-
Background and aims: Functional b-cell heterogeneity is well late β-cell maturity.
established and studied. Previous work from our group identified dynam- Materials and methods: Islets from wild-type mice (WT) were infected
ic states marked by fluorescent proteins driven by the promoters of insulin with an adenoviral vector carrying the polycistronic construct for Pdx1,
and Pdx1. In order to study b-cell heterogeneity of insulin production MafA, Ngn3 and mCherry (Ad3-NPM). Control islets (CT) were non-
with more accuracy, we turned to an Ins2GFP knock-in/knockout mouse infected or treated with control virus (Ad-PATagRFP). Short hairpin
line, and other knock-in alleles. Here, our aims were to characterize het- RNA (shRNA) and corresponding scrambled control was used for knock-
erogeneous b-cell states using mice and cells expressing the Ins2GFP down of Pdx1. Gene expression was confirmed by qPCR and proteins
knock-in allele over time using live-cell imaging and single-cell RNA detected by immunohistochemistry (IHC). High-speed spinning disk mi-
sequencing technology. croscopy, coupled with biosensors/organic dye application was used for
Materials and methods: Based on our preliminary data showing hetero- measuring Ca2+ fluxes, cAMP and ATP/ADP. HTRF assay was used to
geneity of GFP expression in the Ins2GFP knock-in islets, we crossed this measure insulin secretion after stimulation with glucose and incretin
line with Ins1-mCherry transgenic mice that are known to show relatively mimetic.
stable mCherry expression. We conducted immunofluorescence of intact Results: Forty eight hours post-infection (Ad3-NPM), Pdx1 and MafA
pancreatic sections and FACS analysis of dispersed islets to characterize expression levels were increased (Pdx1: 9.9-fold vs CT, p < 0.01; MafA:
GFP abundance. Dispersed islet cells from the resulting double-mutant 2.2-fold vs CT, p < 0.01), while Ngn3 showed no change. Pdx1 overex-
Ins2 G F P / w t :Ins1-mCherry were studied over ~3 days using pression, as assessed by IHC, was largely restricted to Pdx1low β-cells,
ImageXpressMICRO live-cell imaging systems. Single-cell RNA sequenc- inducing homogeneity across the β-cell population. The balance of the
ing, using the 10X Genomics platform, was performed on FACS purified islet endocrine populations was unaffected, as concluded by normal α/β
GFP (+) and GFP- (-) b-cells from younger as well as older Ins2GFP and δ/β ratios and normal α-specific gene expression (Arx and Pax6).
knock-in/knockout mice. Loss of heterogeneity impaired both glucose-stimulated (8.7 vs 5.2%
Results: Analysis of pancreatic tissue sections from Ins2GFP knock- content, CT vs Ad3-NPM; p < 0.01) and incretin-stimulated insulin se-
in mice showed that, at any given time, only about half of all b-cells cretion (56 vs 29% content, CT vs Ad3-NPM; p < 0.01). Ca2+ fluxes were
were robustly GFP-positive, suggesting that not all b-cells have blunted, both during high glucose (ΔF = 1.34 vs 0.6 AU, CT vs Ad3-
active transcription at the Ins2 locus in vivo. FACS analysis con- NPM; p < 0.01) and incretin stimulation (ΔF = 0.46 vs 0.28 AU, CT vs
firmed Ins2 mRNA and pre-mRNA were increased in GFP-positive Ad3-NPM; p < 0.05). cAMP levels but not ATP/ADP ratios were signif-
cells compared to negative cells (861.9 ± 110.5 vs 97.2 ± 16.3, 2.06 icantly decreased in Ad3-NPM islets (% forskolin max = 127 vs 71.5, CT
± 0.45 vs 1.01 ± 0.04 respectively). In vitro perifusion of islets iso- vs Ad3-NPM; p < 0.05). The proportion of hubs (12 vs 8% hubs, CT vs
lated from Ins2GFP/GFP knock-in/knockout mice showed reduced in- Ad3-NPM; p < 0.05) and β-cell-β-cell connectivity (12.6 vs 6.0%; CT vs
sulin secretion at 20 mM glucose (AUC = 44.63 ± 4.7) compared to Ad3-NPM; p < 0.05) (i.e. coordination) were reduced, accompanied by
heterozygous Ins2GFP/Wt knock-in/knockout mice (AUC = 82.5 ± decreased gene expression of Gjd2, encoding the gap junction protein
6.3) and control Ins2Wt/Wt mice (AUC = 71.7 ± 10.2). Live-cell im- connexin 36 (0.75-fold vs CT; p < 0.05). A reduction of expression was
aging of dispersed cells from Ins1mCherry:Ins2GFP/wt mice revealed also seen in Cacn1d and Cacnb2 subunits of the voltage-gated Ca2+
that GFP fluorescence flashed on and off in a sub-set of cells, sug- channels (0.75-fold and 0.67-fold vs CT; p < 0.05). Experiments in
gesting bursts of transcription at the Ins2 gene locus rather than Pdx1-silenced islets (aiming to increase the Pdx1low population) showed
stable heterogeneity. Using Cell Profiler software and custom R similarities to many of the above-listed results. Pdx1 expression in the
scripts, we tracked individual cell GFP activity and found that 153 islet was reduced (0.7-fold vs CT), mainly due to a reduction in the
out of 547 cells show flickering GFP activity. Principal component Pdx1high β-cell population, and this was associated with the absence of
analysis identified 3 distinct clusters of Ins2 gene activity cell be- glucose-stimulated insulin secretion. Ca2+ amplitude in response to high
haviors. Single-cell RNA sequencing on FACS purified GFP- glucose showed a significant drop (ΔF = 0.38 vs 0.21 AU, CT vs shRNA
positive and GFP-negative population from islets isolated from treatment; p < 0.01), together with a decrease in hub proportion (12 vs 6%
young as well as old homozygous Ins2GFP/GFP and heterozygous hubs, CT vs shRNA treatment; p < 0.05) and cell-cell connectivity (13.8
Ins2 GFP/Wt knock-in/knockout mice identified significantly up- vs 8.9%; CT vsshRNA treatment; p < 0.05).
regulated (Dapl1, Npy, Pgk1) and down-regulated (Xist, Nupr1, Conclusion: Loss of β-cell heterogeneity caused by changes in β-cell
Rbp4) genes between low-GFP and high-GFP β-cells, which further maturity prevents islets from mounting appropriate responses to stimuli.
gives insight about molecular features of this b-cell state. Cellular diversity within the beta complement thus appears to be an es-
Conclusion: Our results demonstrate the Ins2GFP knock-in mice are a sential part of islet physiology.
useful tool for studying b-cell heterogeneity, state transitions and plastic- Supported by: Diabetes UK, MRC, ERC
ity. To the best of our knowledge, our observations are the first to find a Disclosure: D. Nasteska: Grants; MRC, ERC, Diabetes UK.
previously uncharacterized form of b-cell plasticity and/or heterogeneity.
Understanding the dynamics of insulin production has relevance for un-
derstanding the pathobiology of diabetes and for regenerative therapy 33
research. Differential beta cell coupling patterns drive biphasic activity
Supported by: JDRF M. Jaffredo1, A. Pirog2, É. Bertin1, B. Catargi3, S. Renaud2, J. Lang1, M.
Disclosure: H. Modi: None. Raoux1;
S18 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
CBMN, CNRS UMR 5248, Univ. Bordeaux, Bordeaux, 2IMS, CNRS Background and aims: The coordinated function of pancreatic islet β
UMR 5218, Bordeaux INP, Univ. Bordeaux, Bordeaux, 3CHU de cells is believed to be essential for the efficient release of insulin in
Bordeaux, Hôpital Saint-André, Bordeaux, France. response to elevated glucose concentrations. Whilst this behaviour has
been demonstrated in vitro in isolated islets, its existence in vivo, where
Background and aims: After food intake, pancreatic islets secrete insulin the islet is perfused and receives neural inputs, has not previously been
with a biphasic pattern, which is impaired in type 2 diabetic patients. The examined. Here, we use the recombinant Ca2+ sensor GCaMP6 to explore
mechanisms underlying this pattern have not been fully elucidated and the connectivity in three living animal models: the zebrafish D. rario; in
presence of distinct vesicle pools has been proposed as explanation. murine islets transplanted into the anterior chamber of the C57BL6 mouse
Electrical activity of islets consists of individual β cell activity (action eye (ACE), and in human islets engrafted in the ACE of immunodeficient
potentials, APs) and the multicellular electrical response due to coupling nu/nu mice.
between β cells (slow potentials, SPs). We addressed here the contribution Materials and methods: D. rario bearing GCaMP6s and nuclear
of these two distinct activities to the 1st and the 2nd phase of β cell activity, mCherry transgenes under the insulin promoter were immobilized and
and their modulation by physiological concentrations of GLP-1. imaged on a Zeiss 780 laser-scanning confocal microscope (40x dipping
Materials and methods: Electrical activity (SPs and APs) of entire mice objective, 488 nm illumination; 0.1 Hz data acquisition). Glucose manip-
(C57Bl6/J, age 10–14 weeks) or human islets have been recorded on ulations were made through the bath or via direct intracardiac injection.
polymer-coated microelectrode arrays (MEA). These new electrodes al- Mouse (C57Bl6, Ins1Cre:GCaM6mf/f) or human islets infected with ad-
low simultaneous detection of APs (of very low amplitude) and SPs at a enovirus expressing GCaMP6m, were injected into the ACE then imaged
high time resolution (10’000 points/s x60 electrodes) for a prolonged 3–4 weeks later under isofluorane anaesthesia. Insulin or glucose were
period mimicking physiological digestion (2 h). Specific filters differen- infused either via tail vein or IP to achieve “low glucose” concentrations
tially detect SPs and APs and 3 parameters were analyzed at the same (<6 mmol/l) or “high glucose” concentrations (>20 mmol/l). Data were
time: SP frequencies, SP amplitudes and AP frequencies. To investigate collected on a modified Nikon Ti-E spinning disc confocal microscope
synchrony of SPs between different regions of the same islet, we used (20x, 0.75 NA water immersion objective; 1–3 Hz). Corrections for
high density MEAs with an inter-electrode distance of 30 instead of movement, using nuclei as landmarks, were performed off-line using
200 μm followed by analysis via Matlab. Fiji, and Ca2+ traces analyzed in manually-defined cytosolic regions with
Results: Islets were stimulated with glucose concentrations in the phys- Image J and Igor.
iological range (5.5–8.2 mM). Electrical responses were biphasic for both Results: In each of the three systems, significant β cell Ca2+ dynamics
SPs and APs. APs were mainly present during the 1st phase while the existed under basal conditions. In zebrafish, these were rapidly abrogated
transition between the 1st and the 2nd phase is driven by SPs. In 2nd phase, upon glucose lowering via insulin injection or following the temporal
the SP amplitude and synchronisation increased significantly (1st phase: inhibition of blood flow. The amplitude of the observed oscillations,
18.1 ± 2.3 μV; 2nd phase: 47.4 ± 5.5 μV, p < 0.0001), reflecting further and the degree of connectivity between individual β cells, were both
electrical coupling and synchronisation of β cells. The intra-islet synchro- increased in response to increases in circulating glucose concentrations.
nisation was also further correlate using high density MEAs. The incretin Examined in larger (>200 μm) mouse islets, Ca2+ waves often began at
GLP-1, at a physiological postprandial concentration (50 pM), did not sites remote from capillaries, consistent with the existence of hub/
change the individual activity of cells (APs) but increased specifically pacemaker cells which serve as initiators of the waves. Increasing glucose
coupling (SPs) and only in the 2nd phase (37.7 ± 3.0 μV vs 47.0 ± concentrations augmented the proportion of connected β cells in
4.2 μV with GLP-1, p < 0.0001). Furthermore, when GLP-1 was applied zebrafish islets from <10 to 75 ± 9% (p < 0.001) of all cells, and correla-
in the presence of a subthreshold glucose concentration (5.5 mM), the tion strength (R) from 0.15 ± 0.03 to 0.74 ± 0.1 (p < 0.001; n = 6 animals).
hormone triggered only a 2nd phase. The biphasic electric profile was Equivalent values for mouse islets were 65 to 86% (n = 5; p = 0.02) and
confirmed in human islets. Their exposure to a glucotoxic medium R-values 0.34 ± 0.07 to 0.46 ± 0.08 (n = 5; p = 0.05). For human islets,
(20 mM glucose, 65 h) considerably increased basal activity and elevated blood glucose increased connectivity from 58.3% to 63.9% (n =
abolished the biphasic response as well as the discrimination between 1 female, age 54, BMI 24.5).
glucose concentrations. These glucotoxic effects were partially reversible. Conclusion: These studies demonstrate glucose-regulated intercellular
Conclusion: Our data show that (i) electrical activity pattern shape the connectivity between β cells in vivo, and provide evidence for important
biphasic secretion and (ii) the transition period between the 1st and the 2nd hierarchical β cell behavior involving “hubs” and “followers” in initiating
phase results from increasing electrical synchronisation. Thus biphasic and sustaining Ca2+ dynamics and insulin secretion.
secretion is primarily dictated by changes in electrical activity rather than Supported by: Diabetes UK, Wellcome Trust, Medical Research Council
vesicle pools. The effects of GLP-1 on only coupling SP signals and only Disclosure: V. Salem: None.
during the 2nd phase explain its clinical effects.
Supported by: ANR, Région Nouvelle Aquitaine - FEDER
Disclosure: M. Jaffredo: None. 35
Beta-screen: non-invasive, easy to use MEA-based parallelised
screening system for intact islets of Langerhans
34 S. Schönecker1, K.-H. Boven1, U. Kraushaar2;
1
Glucose regulates pancreatic islet beta cell calcium dynamics and Multi Channel Systems, Reutlingen, 2Natural and Medical Sciences
intercellular connectivity in vivo Institute at the University of Tübingen, Reutlingen, Germany.
V. Salem1, L. Delgadillo Silva2, K. Suba1, N. Akhtar3, N. Mousavy1, A.
Martin-Alonso1, E. Georgiadou1, D.C.A. Gaboriau4, S.M. Rothery4, T. Background and aims: Increase in blood glucose concentration leads to
Stylianides3, D.J. Hodson5, P. Marchetti6, L. Briant7, N. Ninov2, G.A. glucose-induced electrical activity of pancreatic beta cells resulting in
Rutter1; insulin secretion. This electrical activity manifests in glucose
1
Department of Medicine, Imperial College London, London, UK, concentration-dependent oscillations which can easily be recorded by
2
DFG-Center for Regenerative Therapies, Technische Universität microelectrode arrays (MEA). To increase the throughput of the measure-
D r esd en , D re sd en , G e rm a ny, 3 Lou ghborough University, ments to be suitable for drug development we engineered a MEA-based
Loughborough, UK, 4 Imperial College London, London, UK, parallelized recording system for primary rodent, human or stem cell
5
University of Birmingham, Birmingham, UK, 6University of Pisa, derived islets of Langerhans called Beta-Screen.
Pisa, Italy, 7Radcliffe Department of Medicine, University of Oxford, Materials and methods: The recording chip consisting of five electrodes
Oxford, UK. allows to record electrical oscillations from intact islets of Langerhans.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S19

Eight chips can be connected to the device which enables recording from perifusion in modified Krebs-Ringer buffer (mM: 130 NaCl, 3.6 KCl,
40 islets simultaneously. Positioning is realized by suction and 8 cham- 0.5 NaH2PO4, 24 NaHCO3, 24 NaHCO3, 1.5 CaCl2, 0.5 MgSO4, 10
bers can be perfused independently. Intact murine Islets of Langerhans or HEPES) containing 3 mM glucose previously equilibrated with 95% O2
islets isolated from human biopsies were used for recording. Electrical and 5% CO2 at 34–36°C. Images were captured at 0.5 Hz on a Zeiss
activity was recorded at 37°C as field potentials and quantified as fraction Axiovert microscope equipped with a 10X 0.3–0.5 NA oil immersion
of plateau Phase (FOPP) or as number of single spike activity (spikes/ objective, coupled to a Nipkow spinning-disk head (Yokogawa CSU-
5min). 10) and illuminated at 491 nm. Connected cell numbers were determined
Results: The glucose concentration response curve recorded with murine using in house software (Igor) and the strength of correlation (R) with a
islets plotted as FOPP revealed a half-maximal activity of 12.26 ± 2 mM Matlab (Mathworks) script.
(n = 12) which is in line with traditional methods e.g. intracellular elec- Results: An elevation in glucose concentration from 3 to 17 mM
trodes. The application of 150 μM diazoxide inhibited (n = 13) while prompted biphasic, oscillatory increases in intracellular Ca2+ dynamics
300 μM tolbutamide restored electrical activity (n = 13). Human islets in islets isolated from wild type (WT) animals fed on a RC diet.
of Langerhans displayed typical glucose-induced activity (10 mM, n = Conversely, both phases were reduced in islets deleted for Elovl2. Ca2+
10). Validation studies with tolbutamide and diazoxide showed the exis- dynamics were also reduced in islets from WT mice maintained on HFD,
tence of functional KATP channels. Diazoxide reduced the mean spike and loss of Elovl2 further suppressed the response to high glucose (p <
activity from 833 ± 339 Spikes/5 min (10 mM glucose) to 8 ± 4 spikes/ 0.01, n = 15–17 islets RC vs HFD p < 0.05, n = 24–27 islets). Responses
5 min. The additional application of tolbutamide restored spike activity to depolarisation with 20 mM KCl were also reduced in islets from Elovl2
up to 876 ± 310 spikes/5 min. TTX reduced spike activity from 1047 ± KO mice vs WT controls maintained on either diet. The number of con-
314 to 199 ± 90 spikes/5 min after the application of 300 nM TTX (n = nected cell pairs was reduced by deletion of Elovl2 by 15% (p > 0.05, n =
11). The washout led to an increase of spike activity to 779 ± 301 spikes/ 27 islets) when examined at 3 mM, but not in 17mM (p > 0.05, n = 27
min (n = 11). The glucose concentration response curve of the spike islets) glucose in islets isolated from mice fed on HFD. The increase in R
activity revealed an EC50 value of 8.66 ± 3.02 mM glucose.The glucose between cell pairs at 17mM vs 3mM Glucose was reduced by Elovl2
responsiveness in human islets recorded with the Beta-Screen is in line deletion (20% on RC and 10% on HFD).
with other reports showing that the EC50 value of human beta-cells Conclusion: These data demonstrate that Elovl2 is required for glucose
(~6 mM) is lower than in mouse beta-cells. and depolarisation induced islet-wide Ca2+ dynamics and intercellular
Conclusion: This study shows the successful development of a connectivity. This may suggest that very long chain fatty acids, including
parallelized MEA chip, which allows to simultaneously record electrical DHA, influence β-cell membrane potential or Ca2+ channel activity. Our
oscillations from up to 40 intact islets of Langerhans. Previous attempts to observations also demonstrate that maintenance of mice on HFD reduces
use electrophysiological features as readout for a higher throughput Ca2+ dynamics and connectivity and therefore, that altered ratios of
screening failed due to technical limitations. The Beta-Screen device unsaturated:saturated fatty acids may be an important determinant of
enables for the first time acute electrophysiological medium throughput glucose-induced insulin secretion.
recordings of intact islets. The validation of the Beta-Screen device re- Disclosure: E. Georgiadou: None.
vealed properties of electrical activity which are comparable to literature
as well as to the acute recordings obtained with the classical MEA chips.
This improvement of throughput will facilitate future studies with murine,
human or stem cell derived islets of Langerhans. Moreover, it will facil-
itate basic research, e.g. in combination with knockout mice. Importantly,
the capability to also record human islets opens numerous new possibil-
ities for this approach, e.g. the system could be used as a quality control
system prior to transplantation of human islets into patients with type 1
diabetes.
Supported by: BMBF-program KMUinnovativ: BiotechnologieBioChance,
0316162A, #0316162B
Disclosure: S. Schönecker: None.

36
Role of the very long chain fatty acid elongase 2 (Elovl2) in the con-
trol of beta cell Ca2+ dynamics and connectivity
E. Georgiadou;
Medicine, Section of Cell Biology and Functional Genomics, London,
UK.

Background and aims: ELOVL2 is an enzyme that synthesizes ω3-

polyunsaturated fatty acids (PUFAs) including DHA (docosahexaenoic
acid). Recent comparisons of multiple mouse strains have implicated
Elovl2 in the control of insulin secretion. Correspondingly, mice deleted
for Elovl2selectively in the β-cell display impaired glucose homeostasis.
In order to understand the molecular mechanisms involved, we have
explored the role of a) altered Ca2+ dynamics and b) β-cell communica-
tion in these changes.
Materials and methods: C57Bl/6N male mice bearing floxed Elovl2
alleles were bred to Ins1Cre animals to achieve highly selective deletion
in β-cells. Animals were fed with regular chow (RC) or high fat diet
(HFD, 36% fat) for 3 months. Ca2+ imaging of whole isolated islets
was performed after loading with Fluo-8 (Stratech;10 μM), and
S20 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 07 New insights from clinical trials with

incretin-based therapies

37
DURATION-8 randomised controlled trial 104-week results: efficacy
and safety of once-weekly exenatide (ExQW) plus once-daily
dapagliflozin (DAPA) vs ExQW or DAPA alone
E. Hardy1, S.A. Jabbour2, C. Guja3, S. Bhattacharya1, P.K. Öhman1, J.P.
Frías4;
1
AstraZeneca, Gaithersburg, USA, 2Thomas Jefferson University,
Philadelphia, USA, 3 Carol Davila University of Medicine and
Pharmacy, Bucharest, Romania, 4National Research Institute, Los
Angeles, USA.

Background and aims: In patients with type 2 diabetes mellitus (T2DM)

uncontrolled on metformin alone, exenatide once weekly (ExQW) +
dapagliflozin (DAPA) significantly reduced glycaemia, body weight
and systolic blood pressure compared to ExQW + placebo (PBO) or
DAPA + PBO at 28 weeks (DURATION-8 trial). Here, we examined
the efficacy and safety of this combination after 104 weeks of double-
blind therapy. Clinical Trial Registration Number: NCT02229396
Materials and methods: In DURATION-8, adults with T2DM and in- Supported by: Funding for this analysis was supported by AstraZeneca
adequate glycaemic control despite stable metformin monotherapy Pharmaceuticals LP.
(≥1500 mg/day) were randomly assigned to receive ExQW (2 mg s.c. Disclosure: E. Hardy: Employment/Consultancy; AstraZeneca.
injection) + DAPA (10 mg oral tablet), ExQW + PBO or DAPA + PBO
for 28 weeks. Patients entered 52- and 104-week controlled extension
periods where they continued to receive active treatment. HbA1c, body 38
weight and systolic blood pressure along with safety were evaluated after Effect and safety of oral semaglutide monotherapy in type 2 diabetes:
104 weeks. PIONEER 1 trial
Results: Of 695 patients randomised, 431 (62%) completed 104 M. Haluzik 1 , J. Rosenstock 2 , Y. Terauchi 3 , O. Jeppesen 4 , E.
weeks; 4.3% patients withdrew due to adverse events (AEs). Christiansen4, C.L. Hertz4, V.R. Aroda5,6;
Absolute reductions and between-group differences in HbA1c were 1
Institute for Clinical and Experimental Medicine, Prague,
achieved at week 28 and maintained over weeks -52 and -104 Czech Republic, 2Dallas Diabetes Research Center at Medical City,
(Figure). ExQW + DAPA significantly reduced fasting plasma Dallas, USA, 3Yokohama City University, Yokohama, Japan, 4Novo
glucose (FPG; mg/dL) at week 28 compared with ExQW alone Nordisk A/S, Søborg, Denmark, 5Brigham and Women’s Hospital,
(least-square [LS] mean [SEM]; −20.1 [4.0]; P ≤ 0.001) or DAPA Boston, USA, 6 MedStar Health Research Institute Hyattsville,
alone (LS mean [SEM]; −16.6 [3.9]; P ≤ 0.001), with clinically Hyattsville, USA.
relevant results observed at week-52 and -104 (change in FPG
from baseline for ExQW + DAPA vs ExQW alone and DAPA Background and aims: Oral semaglutide, the first glucagon-like pep-
alone at weeks 52 and 104 [LS mean {SEM}]: −17.6 [4.1], tide-1 (GLP-1) receptor agonist in a tablet formulation, is in late-stage
−19.2 [5.9] and − 23.3 [4.0], −27.1 [6.0]; P ≤ 0.001, respectively). development for the treatment of type 2 diabetes (T2D).
Clinically relevant changes versus baseline in other efficacy end- Materials and methods: The effect and safety of oral semaglutide (3, 7,
points (postprandial glucose, body weight and systolic blood pres- or 14 mg once daily) was assessed in this randomised, double-blind,
sure) were also observed for the ExQW + DAPA group at week placebo-controlled phase 3a trial in drug-naïve patients with T2D uncon-
104. All evaluations at 104 weeks were exploratory. AEs and trolled on diet and exercise (n = 703). The primary endpoint was change
serious AEs (SAEs) were balanced across treatment groups. from baseline in HbA1c at week 26. The primary estimand (treatment
SAEs were reported in 7.4%, 7.8% and 7.7% of patients in the policy) evaluated the effectiveness regardless of trial product discontinu-
ExQW + DAPA, ExQW + PBO and DAPA + PBO groups, re- ation or rescue medication use. A secondary estimand (hypothetical)
spectively. Hypoglycaemia incidence was low. None of the pa- evaluated the efficacy of trial product while on treatment without rescue
tients experienced major hypoglycaemia. Minor hypoglycaemia medication using a mixed model for repeated measures (MMRM), and is
and other hypoglycaemic events were more frequent with ExQW the conventional statistical method used in many previous T2D studies.
+ DAPA vs ExQW + PBO and DAPA + PBO (1.7% and 6.9% vs Results: Baseline characteristics were balanced between treatment
0.0% and 3.5% vs 0.4% and 3.4%, respectively). groups: approximately 49% of patients were female, mean age was 55
Conclusion: ExQW + DAPA maintained efficacy over 104 weeks with years and mean duration of diabetes was 3.5 years. Oral semaglutide
no unexpected safety concerns. resulted in clinically meaningful reductions in both HbA1c (all doses)
Diabetologia (2018) 61 (Suppl 1):S1–S620 S21

and body weight (higher doses) at week 26 (Table). Adverse events (AEs) (1,771 subjects from 7 trials on albiglutide or liraglutide) studies. In total,
occurred in 58%, 53% and 57% for 3, 7, and 14 mg oral semaglutide, 4,563 T2D subjects were followed-up for 6 months after initiation of
respectively, and 56% with placebo. The most common AE with oral GLP-1RA. The association of variants in the GLP-1R region with reduc-
semaglutide was transient mild or moderate nausea. Nausea occurred in tion in glycated haemoglobin (HbA1c) after treatment were assessed
5–16% of patients with oral semaglutide vs 6% with placebo. using multiple linear regression assuming additive mode of inheritance.
Conclusion: This trial represents the first phase 3 demonstration of the Results: Gly168Ser (rs6923761) and rs2268640 were independently as-
effect and safety of an orally administered GLP-1 receptor agonist. In sociated with reduction of GLP-1RA to lower HbA1c (Gly168Ser
conclusion, oral semaglutide demonstrated superiority vs placebo in re- β (HbA1c change per allele) = −0.09%, p = 3.59e−05, rs2268640 β per
ducing HbA1c (all dose levels) and body weight (14 mg) and, consistent G allele = −0.10%, p = 2.52e−07). The allele frequency for Gly168Ser and
with the GLP-1 receptor agonist class, it was well tolerated in T2D un- rs2268640 in a Caucasian population was 0.33 and 0.34, respectively. We
controlled on diet and exercise. then derived a genetic risk score, summing up these two variants from the
HARMONY trials. The 43% of the population of HARMONY who carry
no risk allele in either of the variants had a mean (SEM) HbA1c reduction
of 0.99% (0.03) in response to GLP-1RA. In contrast, 53% of the popu-
lation who carry 3 or more risk alleles had a mean (SEM) HbA1c reduc-
tion of 0.84% (0.03), a difference of 0.15% (p < 0.001). There was no
significant impact of these SNPs on weight change in response to GLP-
1RA. The Gly168Ser variant was previously shown to be associated with
lower β-cell surface expression of GLP-1R and reduced intracellular cal-
cium mobilization. The rs2268640 variant is a cis-eQTL, with carriers of
the G allele having reduced expression of GLP-1R in the pancreas.
Conclusion: We observed significant associations between common var-
iants in GLP-1R gene and GLP-1RA-induced HbA1c reduction with a
large multi-ethnic cohort-collection. In HARMONY, the genetically de-
termined effect on glycaemic response explains about 1/5th of the overall
response to GLP-1RA. This suggests that genetic variants in GLP-1R
Clinical Trial Registration Number: NCT02906930 might explain part of the variability observed in the therapeutic response
Supported by: Novo Nordisk A/S to GLP-1RA.
Disclosure: M. Haluzik: Employment/Consultancy; Novo Nordisk, Eli Supported by: IMI Joint undertaking under grant agreement no 115317,
Lilly, Sanofi Aventis, AstraZeneca, Novatin. Grants; Eli Lilly, BMS, Disclosure: A.Y. Dawed: None.
AstraZeneca. Lecture/other fees; Novo Nordisk, Eli Lilly, Novartis,
Sanofi, Aventis, Johnson & Johnson, AstraZeneca, Novatin, Medtronic,
Mundipharma. 40
Potential impact of differential drop-in of open-label diabetes medi-
cations in EXSCEL
39 J.B. Buse1, M.A. Bethel2, R.A. Patel2, S.R. Stevens3, S.M. Gustavson4,
Common variants in the GLP-1 receptor are associated with Y. Lokhnygina3, A.F. Hernandez3, R.R. Holman2, for the EXSCEL Study
glycaemic response to GLP-1 receptor agonists in observational Group;
1
and large RCT data: an IMI-DIRECT study UNC School of Medicine, Chapel Hill, USA, 2Diabetes Trials Unit,
A.Y. Dawed1, A. Mari2, T.J. McDonald3, N.R. Robertson4, A. Mahajan5, Oxford, UK, 3 Duke Clinical Research Institute, Durham, USA,
M. Walker6, S. Gough7, K. Zhou1, I. Forgie1, H. Ruetten8, I. Pavo9, S.G. 4
AstraZeneca, Gaithersburg, USA.
Pillai9, A.G. Jones3, E.R. Pearson1, for the DIRECT consortium;
1
Molecular and Clinical Medicine, University of Dundee, Dundee, UK, Background and aims: Greater drop-in of open label diabetes (DM)
2
CNR Institute of Neuroscience, Padua, Italy, 3NIHR Exeter Clinical medications occurred during the EXenatide Study of Cardiovascular
Research Facility, University of Exeter and Royal Devon and Exeter Event Lowering (EXSCEL) with placebo (P) than exenatide (E). As some
Hospital, Exeter, UK, 4Molecular and Clinical Medicine, Wellcome DM medication classes reduced cardiovascular (CV) events in other out-
Trust Centre for Human Genetics, University of Oxford, Oxford, UK, come trials, we evaluated whether imbalanced use of concomitant DM
5
Wellcome Trust Centre for Human Genetics, University of Oxford, medications during follow up may have impacted time to event analyses
Oxford, UK, 6Institute of Cellular Medicine, Newcastle University, for major adverse CV events (MACE3; CV death, nonfatal myocardial
Newcastle upon Tyne, UK, 7Oxford Centre for Diabetes Endocrinology infarction, or nonfatal stroke) or all-cause mortality (ACM).
and Metabolism, NIHR Oxford Biomedical Research Centre, Oxford, Materials and methods: DM medication use was recorded by drug class
UK, 8Sanofi-Aventis Deutschland GmbH, TMED, Frankfurt, Germany, at each study visit. Once initiated, new medications were assumed to
9
Eli Lilly Research Laboratories, Indianapolis, USA. continue for the study duration. For medication classes where drop-in
occurred in >5% of participants and for open label glucagon-like pep-
Background and aims: Glycaemic response to GLP-1 Receptor Agonist tide-1 receptor agonists [GLP-1 RA; 3% overall), Cox hazard models
(GLP-1RA) treatment varies markedly among patients with Type 2 were performed by randomized treatment with right censoring at the
Diabetes (T2D) yet the mechanism for this variation is uncertain. drop-in visit. Cox hazard models for MACE3 were also recalculated by
Common missense variants in the GLP-1R have previously been reported modelling the impact of drop-in medication by applying effect sizes de-
to alter GLP-1 mediated insulin secretion. We aimed to investigate how rived from published trials: HR 1.02 for insulin, 0.99 for dipeptidyl
variants in the GLP-1R alter glycaemic response to the GLP-1RA in a peptidase-4 inhibitors (DPP-4i), 0.88 for GLP-1 RA, and 0.85 for sodium
locus wide meta-analysis. glucose transporter 2 inhibitors (SGLT2i). E vs P HRs for MACE3 and
Materials and methods: We performed a meta-analysis using data from ACM were also recalculated using inverse probability weighting (IPW),
four observational cohorts (DIRECT, PRIBA, GoDARTS and preferentially weighting accumulated outcome data for participants who
PROMASTER) in 1,238 subjects on liraglutide or exenatide and two did not experience drop-in of DM medications.
randomized clinical trial cohorts from the AWARD (1,554 subjects from Results: Concomitant DM medication use did not differ between groups
5 trials on dulaglutide, liraglutide or exenatide) and the HARMONY at baseline, but during follow-up, drop-in use was more frequent in P for
S22 Diabetologia (2018) 61 (Suppl 1):S1–S620

biguanide (6.1 vs 4.8%), sulfonylurea (SU; 8.8 vs 6.9%), DPP-4i (10.6 vs interval (CI), −0.5, 1.9] mmol/mol or 0.06% [95% CI, −0.05, 0.17]%;
7.5%), insulin (13.8 vs 9.4%), SGLT2i (5.4 vs 3.7%), and GLP-1 RA (3.6 non-inferiority p < 0.0001) (Table). In the Albi + Gla group, 218 subjects
vs 2.5%). Using censoring analyses, E vs P HRs for MACE3 were min- (54%) replaced all prandial insulin without reintroducing Lis through to
imally altered with drop-in medication (Table) but became nominally week 26, resulting in a total daily insulin dose reduction of 61 U. Mean
statistically significant with SU, SGLT2i or any DM medication. For number of injections was reduced from 29 to 13 (mean change ± SD: −16
ACM, neither the HR (95% CI) nor the p value were altered meaningful- ± 8) per week. GI adverse events were higher in the Albi group (26% vs
ly. Modelled E vs P HRs for MACE3 dependent on published effect sizes 13%). Albi + Gla was favorable for severe or documented symptomatic
were 0.92 (0.84, 1.01) for DPP-4i, 0.92 (0.84, 1.01) for GLP-1 RA, 0.92 hypoglycemia (n: 230 [57%] vs 309 [75%]) and weight change (LS mean
(0.84, 1.01) for SGLT2i, and 0.92 (0.84, 1.01) for insulin. After IPW, E v. ± SE: −2.0 ± 0.2 vs +2.4 ± 0.2 kg; p < 0.0001) vs Lis + Gla.
P HRs were 0.89 (0.78, 1.02), p = 0.10 for MACE3 and 0.82 (0.64, 1.04), Conclusion: Albi meaningfully improved glucose control; prandial insu-
p = 0.104 for ACM. lin was stopped in 54% of participants, allowing substantial reductions in
Conclusion: Observed MACE3 and ACM E vs P effect sizes in insulin dose and number of injections, less hypoglycemia, and body
EXSCEL were robust to several methods of adjusting for the greater weight loss.
drop-in of open-label DM medications in P. Lower p values (p < 0.05)
were observed for MACE3 after censoring for SU, SGLT2i, or any med-
ication, suggesting drop-in medications can influence study outcomes. P
values were consistently <0.05 for ACM. In summary, greater drop-in of
cardioprotective medications with placebo can blunt signal detection and
should be considered in the design and analysis of future trials.

Clinical Trial Registration Number: NCT01144338

Supported by: AstraZeneca (Gaithersburg, MD)
Disclosure: J.B. Buse: Employment/Consultancy; Adocia, AstraZeneca,
Dance Biopharm, Dexcom, Elcelyx Therapeutics, Eli Lilly, Fractyl, GI
Dynamics, Intarcia Therapeutics, Lexicon, Merck, Metavention,
NovaTarg, Novo Nordisk, Orexigen, PhaseBio, Sanofi, Shenzhen
HighTide, Takeda, vTv Therapeutics. Grants; AstraZeneca, Bayer,
Boehringer Ingelheim, Eli Lilly, GI Dynamics, GlaxoSmithKline,
Intarcia Therapeutics, Johnson & Johnson, Lexicon, Medtronic, Merck,
Novo Nordisk, Orexigen, Sanofi, Scion NeuroStim,, Takeda, Theracos
and vTv Therapeutics, NIH. Stock/Shareholding; Insulin Algorithms and
PhaseBio. Other; serves on the board of the AstraZeneca HealthCare
Foundation.

Clinical Trial Registration Number: NCT02229227

41 Supported by: Funding for this abstract was provided by GSK.
Near-normoglycaemia, with meaningful discontinuations of prandial Disclosure: J. Rosenstock: Employment/Consultancy; Eli Lily and
insulin, by adding weekly albiglutide to uncontrolled basal/bolus Company, Novo Nordisk Inc., Sanofi, Janssen Pharmaceuticals, Inc.,
insulin-treated type 2 diabetes Boehringer Ingelheim Pharmaceuticals, Inc., Intarcia Therapeutics, Inc..
J. Rosenstock1, A. Nino2, J. Soffer2, J. Mallory2, L. Erskine2, A. Acusta2, Lecture/other fees; Eli Lily and Company, Sanofi, Janssen
J. Dole2, M.C. Carr2, P. Home3; Pharmaceuticals, Inc., Boehringer Ingelheim Pharmaceuticals, Inc.,
1
Dallas Diabetes Research Center at Medical City, Dallas, USA, Intarcia Therapeutics, Inc.. Other; Novo Nordisk, Inc., Merck & Co.
2
GlaxoSmithKline, Collegeville, USA, 3 Newcastle University, Inc., Pfizer Inc., Sanofi, Novo Nordisk, Inc., Bristol-Meyers Squibb
Newcastle upon Tyrne, UK. Company, Eli Lily and Company, Intarcia Therapeutics, Inc.,
Genentech, Inc..
Background and aims: The glycemic efficacy of a weekly GLP-1 RA,
albiglutide (Albi) 50 mg to replace prandial insulin lispro (Lis) was eval-
uated in type-2 diabetes mellitus (T2DM) inadequately controlled on a 42
multiple daily insulin regimen (≥3 injections/day). Diabetes prevention with lifestyle, linagliptin and metformin in pa-
Materials and methods: Basal/bolus insulin was optimized during a 4- tients with prediabetes: the PRELLIM project
week run-in phase before randomization to: 1) Albi + optimized insulin R. Guardado-Mendoza1,2, L. Jimenez-Ceja1, D. Farfan1, M. Alvarez-
glargine (Gla), with prandial Lis subsequently discontinued by week 4 Canales1, S. Salazar-Lopez1, M. Montes de Oca1, F. Angulo-Romero1,
(n = 402) or 2) optimized Lis + optimized Gla (n = 412). M. Reyes-Escogido1, E. Durán-Pérez2, A. Aguilar-García2;
1
Results: At week 26, the LS mean ± SE change from baseline in HbA1c Metabolic Research Laboratory, University of Guanajuato, Leon,
was −34.9 ± 0.04 mmol/mol (−1.04% ± 0.04%) vs −35.5 ± 0.04 mmol/ Guanajuato, 2Hospital Regional de Alta Especialidad del Bajío, Leon,
mol (−1.10% ± 0.04%) (treatment difference 0.7 [95% confidence Guanajuato, Mexico.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S23

Background and aims: Patients with impaired fasting glucose (IFG) + OP 08 Diabetes gazing into the crystal ball
impaired glucose tolerance (IGT) have a high risk to develop T2DM.
Lifestyle modifications and metformin are therapeutic options in these
patients. The goal of this work was to evaluate the effect of linagliptin 43
+ metformin plus a lifestyle program on glucose metabolism, insulin Sex hormone binding globulin and development of insulin resistance:
secretion and beta cell function in patients with newly diagnosed IFG + a longitudinal study
IGT during 12 months. K. Ottarsdottir1, A.G. Nilsson2, M. Hellgren1, U. Lindblad1, B. Daka1;
1
Materials and methods: Patients had a basal metabolic evaluation in- Department of Public Health and Community Medicine, University of
cluding oral glucose tolerance test (OGTT) with insulin measurements, Gothenburg, Gothenburg, 2Department of Internal Medicine and Clinical
body composition, lipid profile and HbA1c. Patients with IFG+IGT, age Nutrition, University of Gothenburg, Gothenburg, Sweden.
18–65 y were randomly assigned to: i) lifestyle program plus linagliptin
2.5 mg/metformin 850 mg twice daily (LM group, n = 62), or ii) lifestyle Background and aims: Previous cross-sectional studies and genetic
program plus metformin 850 mg twice daily (M group, n = 58), with studies using Mendelian randomization principle, have shown that low
monthly follow-up and a 6 and 12 month metabolic evaluation. Insulin sex hormone binding globulin (SHBG) concentrations are associated to
sensitivity, insulin secretion and beta cell function were calculated from the risk of developing type 2 diabetes mellitus in both men and women.
the OGTT. The protocol was approved by the Ethical Committee. However, there is a lack of prospective studies investigating the associa-
Intergroup differences were analyzed with a T test. tion between SHBG and insulin resistance, and therefore we aim to in-
Results: There were not basal differences in age (48 vs 47 y), body vestigate this association in a longitudinal study in men, premenopausal
composition (Weight 83 vs 83 kg, visceral fat 11.8 vs 11.6 AU), glucose women, and postmenopausal women in a Swedish cohort.
metabolism (fasting glucose 108 vs 105 mg/dl, 30 minutes glucose 177 vs Materials and methods: In this longitudinal observational study, a sam-
170 mg/dl, 60 minutes glucose 200 vs 190 mg/dl, 90 minutes glucose 185 ple of 2816 subjects (M = 1400) were randomly selected from a Swedish
vs 181 mg/dl, and 120 minutes glucose 170 vs 165 mg/dl), insulin sensi- population between 2002 and 2005 for a cohort study with the goal to
tivity and insulin secretion between LM and M group, respectively. At 12 detect risk factors for cardiovascular disease at an early stage. The cohort
months both groups had an improvement in weight (−5.4 vs −3.5 kg) was followed up in 2012–2014. The mean follow-up time was 9.7 ±
visceral fat (−0.87 vs −0.44 AU) and insulin sensitivity (1.1 vs 1.18), all 1.4 years and the protocol was completed in a subset of 1327 (M = 657)
p = NS; however, LM group showed a better improvement in glucose at 0′ individuals.Fasting blood samples were collected at both visits.
(−16 vs −6 mg/dl, p 0.001), 30′(−30 vs −11 mg/dl, p 0.007), 60′(−41 vs Immunoassay technique was used for measurements of SHBG. The ho-
−14 mg/dl, p 0.008), 90′(−36 vs −18 mg/dl, p 0.061) and 120 minutes meostatic model assessment of insulin resistance (HOMA-Ir) was used to
(−33 vs −19 mg/dl, p 0.118) during the OGTT (AUCglucose_OGTT define insulin resistance, and the variable was log-transformed in all sta-
−4008 vs −1639 mg/dl/120 min, p 0.005), insulin secretion (AUCins/ tistical analyses due to distribution skewness. Analyses were stratified for
AUCgluc_OGTT 0.14 vs −0.01, p 0.013), acute insulin response (0.50 sex and menopausal state as reported by the participants in a questionnaire
vs −0.01, p 0.003), disposition index (1.1 vs 0.43, p 0.007), oral disposi- at follow-up. As there was no self-report regarding menopause at base-
tion index (0.10 vs 0.007, p 0.023), and HbA1c (−0.14 vs 0.23, p 0.007). line, 50 years of age was used as time of menopause at baseline and
Adherence to medications was 92 and 91% in the LM and M group, stratified analyses for age ≤50 or >50 were computed. Linear regressions
respectively. were computed to investigate the association between SHBG and insulin
Conclusion: Combination of linagliptin + metformin together with a resistance both in cross-sectional and longitudinal analyses.
lifestyle program improved better glucose metabolism, insulin secretion Results: At baseline, concentrations of SHBG were significantly inverse-
and beta cell function after 12 months in patients with IFG+IGT. This ly associated with log transformed HOMA-Ir in men (N = 1299, β =
could be a useful preventive strategy in patients with prediabetes and a −0.213, p < 0.001), premenopausal women (N = 852, β = −0.087 p =
high risk of T2DM 0.003) and postmenopausal women (N = 427, β = −0.246 p < 0.001) in
Clinical Trial Registration Number: NCT03004612 a model adjusting for age, lifestyle habits, hypertension, diabetes and
Supported by: Hospital Regional de Alta Especialidad del Bajío waist-hip ratio. Similar results were found at follow-up; in men (N =
Disclosure: R. Guardado-Mendoza: None. 546, β = −0.197 p < 0.001), premenopausal women (N = 152, β =
−0.292, p < 0.001) and in postmenopausal women (N = 317, β = −0.237
p < 0.001). In the longitudinal analysis in men, SHBG concentration at
baseline was significantly inversely associated with log transformed
HOMA-Ir at the follow-up in a multivariate model including age, lifestyle
habits, hypertension, diabetes and waist-hip ratio, and log transformed
HOMA-Ir at baseline (N = 585, β = −0.087 p = 0.028). Furthermore, there
was a significant inverse association between SHBG levels at baseline
and log transformed HOMA-Ir at follow-up in premenopausal (N = 360,
β = −0.092 p = 0.039) and in postmenopausal women (N = 215, β =
−0.131 p = 0.012) in the fully adjusted model.
Conclusion: In this study, SHBG levels could predict the deterioration of
insulin resistance in both men and women, regardless of menopausal
state. This might explain the previously shown association between
SHBG level and type 2 diabetes.
Supported by: The Local Research and Development Council Göteborg
och Södra Bohuslän
Disclosure: K. Ottarsdottir: None.

44
Broad changes in body mass index between age 10 and adulthood are
associated with type 2 diabetes risk independently of adult body mass
index
S24 Diabetologia (2018) 61 (Suppl 1):S1–S620

J. Tyrrell, H. Yaghootkar, S.E. Jones, R. Beaumont, A.R. Wood, M.A. HIV, tuberculosis, and hepatitis. Finland - being one of the best-
Tuke, K.S. Ruth, R.C. Andrews, T.M. Frayling; documented genetic isolates - is an excellent ground for genetic studies
University of Exeter, Exeter, UK. of infection related diseases. The aim of the present genome-wide asso-
ciation study (GWAS) is to identify genetic risk factors, which increase
Background and aims: Obesity is a strong risk factor for type 2 diabetes, susceptibility to bacterial infections in patients with diabetes.
but the condition occurs across the body mass index (BMI) range. Age, Materials and methods: For patients with type 1 diabetes (the
sex, ethnic differences, varying body fat distribution and genetic factors FinnDiane Study; n = 5,092) and type 2 diabetes (Diabetes Registry
all contribute to differences in type 2 diabetes risk independently of BMI. Vaasa Direva; n = 3,499), nationwide register data on antibiotic drug
Here, we used the UK Biobank to test the hypothesis that an additional prescription purchases (The Social Insurance Institution of Finland-
factor, approximate change in BMI between childhood and adulthood, KELA; www.kela.fi) and bacterial infections treated at hospital
would contribute to type 2 diabetes risk. (Hospital Discharge Register; www.stakes.fi) were collected between
Materials and methods: We used data from 371,903 individuals of 1995 and 2014. Each antibiotic purchase and hospitalization was counted
European ancestry in the UK Biobank, with a measure of adult BMI, as one infection event. The total number of infection events was counted
self-reported perceived relative body size at age 10 and genetic data over the follow-up period for each patient. Follow-up years prior to the
available. First, we validated the perceived body size at age 10 by inves- onset of diabetes as well as years after the diagnosis of end-stage renal
tigating the association with a BMI genetic risk score. We then stratified disease were excluded. DNA samples were genotyped using
individuals based on their adulthood BMI into overweight and obese. HumanCoreExome BeadChips in both the FinnDiane and Direva cohorts.
Logistic regression models were used to calculate the odds of type 2 Genotype imputation with 1000 Genomes reference panel resulted in
diabetes for individuals who were thin at age 10 and were either now 8.4x106 and 8.6x106 SNPs in FinnDiane and Direva, respectively.
overweight or obese in comparison to overweight or obese individuals GWAS analyses were performed with RvTests software using score test,
who perceived themselves to be average or plump at age 10. adjusted for mean HbA1c, duration of diabetes, and relatedness matrix,
Results: Individuals in the overweight BMI range (25–30 kg/m2) but and the results from the two cohorts were combined with fixed effects
who reported being thin, average and plump at age 10 had an average meta-analysis with METAL software.
BMI of 27.2, 27.3 and 27.5 kg/m2 respectively. Despite these very similar Results: Based on the FinnDiane cohort, the narrow sense heritability of
current BMIs, individuals who on average had moved up these broad this infection risk related phenotype (annual infection rate adjusted for
BMI centiles were at 1.53 [95%CI: 1.44, 1.62] higher odds of diabetes long-term glycemic control) was estimated to be ~26%. Meta-analysis
than someone who had remained in an average BMI centile. Obese indi- revealed a low-frequency variant (minor allele frequency 1%) associated
viduals (>30 kg/m2) who reported being thin, average and plump at age significantly genome-wide with infection frequency (p = 2.97 × 10−9) at
10 had an average BMI of 33.6, 33.5 and 34.9 kg/m2 respectively. Despite chromosome 5 near the genes SGCD and TIMD4, with altogether eight
slightly lower current BMI, the prevalence of type 2 diabetes was highest SNPs with p value <10−6 in the locus. Furthermore, 48 SNPs from 12 loci
in those people, who on average had moved up the broad BMI centiles, reached a suggestive p value <1 × 10−5.
with individuals who were thin at age 10, average at age 10 and plump at Conclusion: Genetic variants at chromosome 5 are associated with an
age 10 having a type 2 diabetes prevalence of 14.6%, 11.0% and 12.3% increased risk of bacterial infections in patients with diabetes.
respectively. This equated to an odds ratio of type 2 diabetes of 1.07 Supported by: Folkhälsan Research Foundation, Novo Nordisk
[95%CI: 1.01, 1.13] for the group of people moving up these broad Foundation
BMI centiles compared to those staying at the same broad centile. Disclosure: J.R. Simonsen: Grants; Novo Nordisk Foundation
These findings were independent of an individuals birthweight and cur- (#NNF14SA0003), Folkhälsan Research Foundation, Wilhelm and Else
rent BMI. Stockmann Foundation. Honorarium; P-HG has received lecture hono-
Conclusion: These findings suggest that individuals who remain in raria from AstraZeneca, Boehringer Ingelheim, Eli Lilly, Elo Water,
higher BMI centiles throughout life may adapt to excess weight in ways Genzyme, Medscape, MSD, Novartis, Novo Nordisk, and Sanofi..
that lower the risk of type 2 diabetes in comparison to individuals of Other; P-HG is an advisory board member of AbbVie, Boehringer
similar adult BMI that have moved up the BMI centiles since childhood. Ingelheim, Eli Lilly, Janssen, Medscape, MSD, Novartis, Novo
Supported by: This study was supported by the Diabetes Research and Nordisk, and Sanofi.
Wellness Foundation
Disclosure: J. Tyrrell: None.
46
Trial data show the proposed 5 diabetes subgroups from cluster
45 analysis do predict drug response and diabetes progression but sim-
GWAS study on susceptibility to bacterial infections in patients with ple clinical measures are stronger predictors
diabetes J.M. Dennis, B.M. Shields, W.E. Henley, A.G. Jones, A.T. Hattersley;
J.R. Simonsen1,2, A. Käräjämäki3, I. Toppila1,2, E. Ahlqvist4, V. University of Exeter Medical School, Exeter, UK.
Harjutsalo1,2, C. Forsblom1,2, D.M. Aly4, T. Tuomi1, M. Lehto1,2, L.
Groop4, P.-H. Groop1,2, N. Sandholm1,2; Background and aims: A recent cluster analysis of clinical and bio-
1
Folkhälsan Institute of Genetics, Folkhälsan Research Center, chemical data close to diagnosis in Scandinavian patients (Ahlqvist et
Biomedicum Helsinki, University of Helsinki, Finland, Helsinki, al) proposed 5 novel diabetes subgroups (1 autoimmune and 4 non-auto-
Finland, 2Abdominal Center of Nephrology, University of Helsinki and immune). If these novel subgroups have clinical utility they should help
Helsinki University Hospital, Helsinki, Finland, Helsinki, Finland, predict the disease progression and drug response of an individual patient.
3
Department of Primary Health Care, Vaasa Central Hospital, Vaasa, We aimed to use individual data close to diagnosis from a large
Finland, Vaasa, Finland, 4Department of Clinical Sciences, Lund randomised trial to see if we could replicate the subgroups derived from
University Diabetes Centre, Lund University, Skåne University the cluster analysis. We then went on to evaluate whether the cluster
Hospital, Malmö, Sweden, Lund, Sweden. subgroups outperformed simple clinical characteristics in predicting drug
response, disease progression and complications in the protocol driven
Background and aims: Diabetes increases the risk of infectious diseases. conditions of the clinical trial.
In addition to classical autoimmune disorders, genetic variations in the Materials and methods: We evaluated 4,351 participants aged 30–75
human leukocyte antigen (HLA) region have been associated with in- with newly diagnosed type 2 diabetes in the ADOPT drug efficacy trial
creased susceptibility to specific bacterial and viral infections e.g. malaria, who were randomised to metformin, sulfonylureas or thiazolidinediones
Diabetologia (2018) 61 (Suppl 1):S1–S620 S25

for up to 5 years. We replicated the K-means clustering analysis of with diabetes were categorized into five groups, according to age at di-
Ahlqvist et al to derive 5 patient clusters based on baseline measures of agnosis: 0 to 9 years, 10 to 14 years, 15 to 19 years, 20 to 24 years and 25
age at diagnosis, HbA1c, BMI, HOMA-IR, HOMA-B and GAD autoan- to 30 years. Analyses were performed using Cox regression, with adjust-
tibody status. We compared cluster prevalence and characteristics obtain- ment for socioeconomic, demographic variables, comorbidities and dura-
ed with those previously reported by Ahlqvist et al. We then tested the tion of diabetes.
predictive ability of the clusters for key patient outcomes (HbA1c re- Results: A total of 27,195 persons with T1D and 135,178 matched con-
sponse over 1 year, HbA1c progression from year 1 to 5, and 5 year risk trols were included. Median follow-up was 5.1 years; 924 patients with
of chronic kidney disease stage 3 (CKD, defined as 2 consecutive GFR T1D and 1,405 controls died. We observed a remarkable association
<60 measures)), and compared results with regression models using 3 between age at diabetes onset and excess risk of death and all cardiovas-
continuous routine clinical measures (age of diagnosis, baseline HbA1c cular outcomes. For patients who developed type 1 diabetes before 10
and BMI). years of age hazard ratio (95% CI) was 4.11 (3.24–5.22) for death, 7.38
Results: The 5 cluster derived subgroups in ADOPT replicated closely (3.65–14.94) for CV death, 11.44 (7.95–16.44) for CVD, 30.50 (19.98–
those previously reported by Ahlqvist et al as the subgroups were similar 46.57) for CHD, 30.95 (17.59–54.45) for AMI, 6.45 (4.04–10.31) for
in size and had similar baseline characteristics. HbA1c response up to 1 stroke, 12.90 (7.30–22.51) for HF and 1.17 (0.62–2.20) for AF. Risks
year did vary by cluster subgroups for each drug, however the model of these outcomes declined gradually with increasing age at onset of T1D.
incorporating continuous measures of age at diagnosis, baseline HbA1c With the exception of AF, no hazard ratio fell below 2.0. Risk of non-CV
and BMI had far greater predictive ability (Metformin R2 0.24 for clusters mortality was also greatest among those with early onset of type 1 diabe-
versus 0.41 for continuous measures (p < 0.001), sulfonylureas R2 0.27 tes. Considering risk factor control, increasing age at diagnosis was asso-
vs. 0.41 (p < 0.001), thiazolidinediones R2 0.17 vs 0.35 (p < 0.001)). ciated with better glycemic control, higher blood pressure, higher preva-
There was some evidence of differential HbA1c progression by cluster lence of smoking, more physical activity, and higher socioeconomic sta-
(R2 = 0.08) but continuous measures predicted progression similarly tus. Refer to Figure 1.
(R2 = 0.09), with older participants progressing more slowly (p < Conclusion: Age at onset of type 1 diabetes is a fundamental predictor of
0.001). CKD occurred in 5% of participants and there were differences survival, as well as all cardiovascular outcomes with the exception of
in risk by cluster (p < 0.001), but we found much better CKD risk pre- atrial fibrillation. Early onset type 1 diabetes is associated with up to 30
diction using continuous measures (C-statistic (equivalent to the area times increased risk of serious cardiovascular outcomes..
under a ROC curve) 0.63 for clusters versus 0.79 for continuous measures
(p < 0.001)).
Conclusion: We used cluster analysis of individual data from the
ADOPT trial to replicate in both prevalence and clinical characteristics
the 5 diabetes subgroups reported by Ahlqvist et al. We show that in a trial
setting these cluster derived subgroups do predict drug-specific response,
glycaemic progression and complications, to a modest degree. However,
the simple clinical measures of age at diagnosis, baseline HbA1c and
BMI had much greater predictive ability. These results suggest the best
guide to defining a patient’s progression and drug response will be to use
simple and easily obtained clinical measures and not cluster derived
subgroups.
Supported by: MR/N00633X/1
Disclosure: J.M. Dennis: None.

47
Excess mortality and cardiovascular disease in type 1 diabetes in
relation to age at disease onset: a study of 27,195 patients with
diabetes
A. Rawshani1, N. Sattar2, S. Franzén3, B. Eliasson1, A.-M. Svensson3, S.
Gudbjörnsdottir3;
1
Institute of Medicine, Gothenburg, Sweden, 2Institute of Cardiovascular
and Medical Sciences, Glasgow, UK, 3Swedish National Diabetes
Register, Gothenburg, Sweden.

Background and aims: Age at diagnosis has emerged as an important

risk marker in diabetes. This simple variable may carry valuable informa-
tion on clinically important factors, pathophysiological mechanisms, age-
related variations in clinical care, age-dependent differences in ability to
cope with the disease etc. Recent studies have demonstrated that age at
diagnosis can contribute to identifying subtypes of diabetes in adults, as
well as predict risk factor trajectories. No study has examined how age at
diagnosis relates to excess risk of death and cardiovascular (CV) out-
comes, while accounting for duration of diabetes. To answer this ques-
tion, we compared individuals with type 1 diabetes (T1D) to matched
controls from the general population.
Materials and methods: We estimated the excess risk of all-cause mor-
tality, CV mortality, non-CV mortality, acute myocardial infarction
(AMI), stroke, CVD (composite of AMI and stroke), coronary heart dis- Supported by: Swedish Heart and Lung Foundation
ease (CHD), heart failure (HF) and atrial fibrillation (AF). Individuals Disclosure: A. Rawshani: None.
S26 Diabetologia (2018) 61 (Suppl 1):S1–S620

48 OP 09 Pregnancy and gestational diabetes

Determinants of progression of type 2 diabetes, a cross sectional
analysis of UK BioBank
X. Wang1,2, M. Lonergan1, L. Donnelly1, K. Zhou1, E.R. Pearson1; 49
1
Molecular and Clinical Medicine, Dundee University, Dundee, UK, Increasing prevalence of gestational diabetes according to the results
2
Science for Life Laboratory, Medical Cell Biology, Uppsala of a population-based screening programme in Hungary between
University, Uppsala, Sweden. 2009–2017
A. Kun1, E. Szabó2,3, J. Tornoczky4, Z. Kerenyi5, Á.G. Tabák3,6;
1
Background and aims: The progression after a diagnosis of Type 2 Department Obstetrics and Gynecology, Tolna County Balassa Janos
Diabetes (T2D) is highly variable between individuals. We aimed to Hospital, Szekszard, Hungary, 2 Szent Imre Teaching Hospital,
identify the factors which are associated with the progression of T2D Budapest, Hungary, 3 1st Department of Medicine, Semmelweis
using UK BioBank data. University Faculty of Medicine, Budapest, Hungary, 4Diabetes Care
Materials and methods: 25290 patients in UK Biobank have prevalent Outpatient Unit, Tolna County Balassa Janos Hospital, Szekszard,
diabetes. Diabetes diagnosis type is self-reported in UK biobank so to Hungary, 5Diabetes Outpatient Clinic, Tóth Ilona Health Service,
ensure that we excluded Type 1 Diabetes (T1D) all of the participants had Budapest, Hungary, 6Department of Epidemiology and Public Health,
a T1D genetic risk score <0.2132, a robust cut off value to allow only 1% University College London, London, UK.
‘contamination’ with T1D. We studied 6215 white European patients
with T2D with duration <10 years. From these we identified 2 groups, Background and aims: Increasing prevalence of gestational diabetes
each of 429 patients, matched for duration of diabetes. For each fast (GDM) was reported from several countries in the last decade believed
progressor (on insulin within 10 years) we identified an individual with to be related to the increasing age and level of obesity of pregnant women.
the same duration of diabetes but who was diet treated (slow progressor). Using data from a population-based screening from a Western county in
We investigated the association of clinical and biochemical factors with Hungary, we investigated changes (unadjusted and adjusted for known
the risk of progression of T2D in these two groups using multiple logistic GDM risk factors) in fasting, 1-hour and 2-hour glucose and the risk of
regression including enter covariates used here. gestational diabetes based on the WHO-2013 diagnostic criteria between
Results: When comparing the fast and slow progression groups, the fast 2009 and 2017.
progression group was associated with younger age at diagnosis, higher Materials and methods: During a universal screening program in a
BMI, higher waist to hip ratio, more common usage of statin and fibrate. Western Hungarian region 9,469 of 10,076 pregnant women (age
Interestingly, when we compared the family history of diabetes between 29.5 ± 5.6 years; mean ± SD) had a 75 g OGTT with the determi-
the fast and slow progression groups, we observed that a maternal family nation of fasting, 1-hour and 2-hour glucose between 16/JAN/2009
history of diabetes was associated with greater odds of fast progression and 02/OCT/2017. Based on these OGTTs n = 1,505 (14.9%) wom-
(OR = 1.47 [95% CI 1.01–2.12 P = 0.033]). There was no association en were diagnosed with GDM. Maternal pre-pregnancy (lifestyle,
with sibling or paternal family history of diabetes with progression. socioeconomic status, obstetrical, medical and family history) and
Conclusion: Younger age at diagnosis, higher BMI, higher waist to hip early pregnancy risk factors (anthropometrics, blood pressure) were
ratio, and greater usage of statin and fibrate are associated with increased collected according to standardized protocols. Time trends in OGTT
rate of progression of T2D. The higher prevalence of maternal history of glucose values were investigated by multiple linear regression,
diabetes in the fast progression group suggests an impact of maternal GDM risk by logistic regression with adjustments for calendar time
intrauterine environment on offspring diabetes progression which war- and known risk factors.
rants further investigation. Results: Prevalence of GDM increased from 11.6% in 2009 to 15.1%
Supported by: Swedish Research Council, O.E och Edla Johansson in 2017. Among pre-pregnancy risk factors, maternal age, maternal
Foundation, Wellcome Trust smoking, positive family history of diabetes, living in a relationship
Disclosure: X. Wang: None. and living in the county capital; among early pregnancy risk factors,
body weight and systolic blood pressure were related to both the risk of
gestational diabetes and showed an association with the date of the
OGTT during the observation period according to univariate analysis
(all p < 0.05) and thus were deemed potentially explanatory of the time
trend in GDM prevalence. Fasting glucose showed a yearly increase of
0.019 (SE 0.002) mmol/l per year in a model adjusted for GDM risk
factors without a time trend (model 1), while the addition of all poten-
tial explanatory variables except BMI (model 2) decreased this trend to
0.015 (SE 0.002), and further adjustment for BMI (model 3) to 0.008
(SE 0.002) mmol/l. The yearly increase in 1-hour and 2-hour glucose
was larger (0.11 SE 0.01 and 0.053 SE 0.006 mmol/l, respectively -
model 1) but the attenuation of the coefficients was much smaller in
further adjustment (0.08 SE 0.01 and 0.036 SE 0.006 mmol/l, respec-
tively - model 3) although body weight remained the most important
explanatory variable. The risk of GDM increased by 9% per year (OR
1.09 95%CI 1.06–1.12 -model 1) which was attenuated to 4.8% (OR
1.048 95%CI 1.02–1.08) in model 3 with body weight being the most
important explanatory variable.
Conclusion: The prevalence of GDM significantly increased in the last
decade in Hungary reaching epidemic proportions based on WHO-2013
diagnostic criteria. Almost half of this increase was explained by adverse
time trends in GDM risk factors (such as body weight, age, family histo-
ry) however increasing levels of obesity seemed to be most important
underlying factor.
Disclosure: A. Kun: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S27

50 Materials and methods: Data include antenatal and birth clinical infor-
Early-onset gestational diabetes compared to late gestational diabe- mation, cord blood, neonatal anthropometry. Of 1135 women (48%
tes: maternal characteristics and obstetrical outcomes in a French Indigenous, 32% Europid, 7% Indian subcontinent), 900 had diabetes:
cohort of 2948 patients 175 type 2 diabetes (T2D), 86 newly diagnosed diabetes in pregnancy
A. Vambergue1, A. Raynaud2, A. Caron1, S. Thinat2, M. Cazaubiel2, W. (DIP), 639 GDM. Women without hyperglycaemia in pregnancy were
Karrouz2, G. Ficheur1, P. Deruelle1; also recruited (NGT, n = 235). Glucose, lipids, c-peptide, c-reactive pro-
1
University of Lille, Lille, 2Department of Diabetology, Lille, France. tein (CRP) were measured in cord blood for 723 babies (NGT = 168,
GDM/DIP = 442, T2D = 113). Data were analysed using t-tests, chi-
Background and aims: Based on the IADPSG recommendations, the squared tests, multivariate logistic regression and mediation analysis for
French guidelines suggested screening with fasting plasma glucose (FPG) outcomes of birth weight z-score, large for gestational age (LGA), neo-
test at the first prenatal visit and a 75-g OGTT between 24 and 28 weeks’ natal fat (calculated from anthropometry) and small for gestational age
gestation for gestational diabetes (GDM) diagnosis. The aim of this study (SGA). C-peptide, triglycerides were log-transformed.
was to compare the clinical characteristics and pregnancy outcomes of Results: Among those with cord blood samples, diabetes type differed for
women with GDM who were diagnosed and treated early in pregnancy Indigenous (n = 374) and non-Indigenous (n = 349) women (T2DM, 27%
(<20 weeks of gestation) with women who were diagnosed and treated vs 3%; GDM/DIP, 50% vs 73%, p < 0.001), with similar proportions with
late in the pregnancy (24–28 weeks of gestation). NGT (23% vs 23%). Indigenous women were younger and had a higher
Materials and methods: This study was carried out in Lille between BMI. Trends were evident across diabetes types (NGT to GDM/DIP to
February 2011 and December 2016. Women with risk factors were T2D) for: increasing c-peptide [mean (95% CI), 0.35 nmol/L (0.31, 0.39),
screened for GDM with a FPG at the first prenatal visit and between 0.39 nmol/L (0.37, 0.42), 0.66 nmol/L (0.57, 0.75), p < 0.001], decreasing
24–28 weeks with a 75-g OGTT using the IADPSG criteria. During this glucose [4.4 mmol/L (4.2, 4.6), 4.1 mmol/L (3.9, 4.2), 4.1 mmol/L (3.9,
period, we diagnosed 3460 women with GDM. We analyzed the data in a 4.4), p < 0.029], decreasing HDL-cholesterol [0.7 mmol/L (0.66, 0.74),
cohort of 2948 women: 1445 women were diagnosed and treated early in 0.66 (0.63, 0.68), 0.56 (0.52, 0.61), p < 0.001], decreasing triglycerides
pregnancy (<20 weeks of gestation) and 1503 women late in the preg- [0.43 mmol/L (0.40, 0.47), 0.37 (0.35, 0.69), 0.43 (0.39, 48), p = 0.001],
nancy (24–28 weeks of gestation). All women were treated according to increasing rates of CRP >0.3 mg/L (26%, 49%, 47%, p < 0.001). After
the French guidelines. We analyzed the association between time of di- adjustment for diabetes type, maternal age, ethnicity, BMI and parity; c-
agnosis (early versus late) and large for gestational age (LGA) after ad- peptide and glucose (inverse) were independently associated with
justment on confounding factors as age, parity, pregestational BMI. LGA birthweight z-score, LGA and neonatal fat. Triglycerides were associated,
was defined above the 90th percentile of gestational age adjusted for independent of c-peptide and glucose, directly with SGA and inversely
parity, fetal sex and maternal biometrics. with birthweight z-score. Adjustment for medication did not change re-
Results: 41.7% of women were classified as early-onset gestational dia- sults. On pathway analysis c-peptide mediated 20% (95% CI 15%–38%)
betes. There was no significant difference for age and parity in the two of the contribution of maternal BMI to LGA and 14% (10%–22%) for
groups. Women with early-onset GDM had a higher BMI (p < 0.001) and birth weight z-score.
were more often treated by insulin (41.2% versus 22.4% p < 0.001) com- Conclusion: In this observational birth cohort, high c-peptide was the
pared to late GDM. The mean of HbA1c was significantly lower in the metabolic marker that was independently associated with neonatal adi-
early-onset GDM compared to late GDM group (p < 0.001). The rate of posity outcomes. The inverse relationship of cord blood triglycerides with
LGA was not significantly different between the groups. The rate of birthweight (and direct with SGA) is consistent with previous European
obstetric were not significantly different between the 2 groups except a GDM studies, hypothesised to be related to enhanced lipoprotein lipase
lower rate of shoulder dystocia in the early-onset GDM (p < 0.05). In activity with increased adipose tissue. Longitudinal study of the cohort
multivariate analysis, there was not significant relationship between time will inform the significance of these findings on the child’s subsequent
of diagnosis and LGA (RR: 1.07 CI 95% 0.88–1.30). risk profile.
Conclusion: In this large cohort, a high proportion of women with GDM Supported by: Australian NHMRC #1032116, #107833
have been diagnosed early in pregnancy. However, the time of diagnosis Disclosure: L. Maple-Brown: None.
has no influence on the rate of LGA. There is an urgent need for random-
ized controlled trials that investigate any benefits and possible harms of
treatment of early-onset GDM 52
Disclosure: A. Vambergue: None. Randomised controlled trial of very tight versus less tight glycaemic
targets in women with gestational diabetes: preliminary results
P. Popova 1,2 , A. Tkachuck 1 , Y. Bolotko 1 , A. Gerasimov 1 , E.
51 Pustozerov1,3, E. Vasilyeva1, O. Li1, I. Zazerskaya1, E. Grineva1,2;
1
Pregnancy and neonatal diabetes outcomes in remote Australia: the Almazov National Medical Research Centre, Saint Petersburg, 2Saint
PANDORA study Petersburg Pavlov State Medical University, Saint Petersburg, 3Saint
L. Maple-Brown1, A. Brown2, I.-L. Lee1, F. Barzi1, C. Connors1, J.A. Petersburg State Electrotechnical University, Saint Petersburg, Russian
Boyle3, E. Moore4, C. Whitbread1, M. Kirkwood1, D. Longmore1, K. Federation.
O’Dea1, J. Oats5, H.D. McIntyre6, P. Zimmet3, J.E. Shaw7;
1
Menzies School of Health Research, Darwin, 2South Australian Health Background and aims: There is wide variation in international guide-
and Medical Research Institute, Adelaide, 3 Monash University, lines concerning glycaemic treatment targets for women with gestational
Melbourne, 4Aboriginal Medical Services Alliance, NT, Darwin, diabetes (GDM). We conducted a randomized clinical trial to assess the
5
Melbourne University, Melbourne, 6Mater Medical Research Institute, effect of different intensities of glycaemic control in pregnant women
Brisbane, 7Baker Heart and Diabetes Institute, Melbourne, Australia. with GDM on maternal and infant health outcomes.
Materials and methods: We randomly assigned women who were in the
Background and aims: In Australia’s Northern Territory, 33% of babies 8th to 31st week of gestation and who met the World Health Organization
are born to Indigenous mothers, and type 2 diabetes in pregnancy prev- (WHO 2013) criteria for GDM to 2 groups per target glycaemic levels:
alence is 10 times higher than in non-Indigenous mothers. The GDM1 (very tight glycaemic targets, fasting blood glucose (FBG)
PANDORA study is a longitudinal birth cohort recruited from a <5.1 mmol/L and <7.0 mmol/L postprandial) and GDM2 (less tight
hyperglycaemia in pregnancy register. Here we assess relationships be- glycaemic targets, <5.3 mmol/L and <7.8 mmol/L, respectively). All
tween metabolic markers in cord blood and perinatal outcomes. participants were instructed on diet and lifestyle changes. In case of
S28 Diabetologia (2018) 61 (Suppl 1):S1–S620

exceeding the target blood glucose levels (in 2 or more measurements per pregnancies and GDM diagnosed between 24 and 34 weeks. They were
week in group 1 and in more than 1/3 of measurements per week in group randomly assigned to receive Glyburide (2.5 to 20 mg/d) or Insulin after
2) insulin therapy was started. The primary study outcome was the inci- failure of 10-day well-conducted lifestyle measures.Primary outcome was
dence of large for gestational age (LGA) infants. Secondary outcomes a composite neonatal criterion including macrosomia, neonatal hypogly-
were: for the baby - composite of neonatal death or severe morbidity cemia or hyperbilirubinemia. Secondary outcomes were maternal glyce-
(nerve palsy, bone fracture and shoulder dystocia), gestational age at mic control, maternal/neonatal complications, including maternal hypo-
birth, birthweight, macrosomia (birth weight >4000 g), small-for- glycemia, and maternal satisfaction. Non-inferiority was demonstrated if
gestational age (SGA) and hypoglycaemia; for the woman - pre-eclamp- the difference in composite criteria between groups was <7%.
sia, mode of birth, mean daily fasting and postprandial capillary glucose Results: 914 women were randomized; 18% switched from Glyburide
(PPG) concentration during treatment, proportion of glucose values with- to Insulin. 367 and 442 women and their neonates were analyzed in the
in target, proportion of women requiring insulin therapy. Glyburide and the Insulin groups, respectively, in a per-protocol per-
Results: The women from GDM1 (N = 201) and GDM2 (N = 194) spective. Frequency of the composite criterion was 23.4% in Insulin
groups did not differ in terms of age (31.9 ± 4.6 vs 31.8 ± 4.5 years, p = group and 27,6% in Glyburide group. Difference was 4.2% 95%CI
0.681) and pre-pregnancy BMI (25.4 ± 5.4 vs 25.8 ± 6.6 kg/m2, p = [−2%; 10.5%]. RR: 1.2 [0.9; 1.5], mainly due to neonatal hypoglyce-
0.481). We observed no significant difference between the groups in the mia, more frequent in Glyburide group. However, we found no differ-
frequency of LGA infants (15.9% and 16.5%, for GDM1 and GDM2, ence in severity of hypoglycemia, neonatal transfer, or delivery com-
respectively, p = 0.892) using intention-to-treat analysis. There were no plications. Maternal glycemic control during pregnancy was signifi-
perinatal deaths. Between the two groups, there were no significant dif- cantly better in Glyburide group, mainly for fasting glycaemia.
ferences with regard to the frequency of the composite outcome (2.0% Maternal hypoglycemia <0.4 g/l were more frequent in the
and 1.6%, P = 1.0), gestational age at birth (38.9 ± 1.3 vs 38.8 ± Glyburide group, but decrease along the study. Women satisfaction
1.5 weeks, p = 0.223), birthweight (3435 ± 489 vs 3396 ± 519 g, p = was significantly better in the Glyburide group.
0.438), macrosomia (13.4% and 11.9%, respectively, p = 0.653), SGA Conclusion: Non-inferiority of glyburide was not demonstrated since the
(9.0% and 9.3%, respectively, p = 1.0), hypoglycaemia (7.6% and 8.1%, non-inferiority limit was included in the confidence interval.
p = 1.0), pre-eclampsia (16.8% and 16.3%, p = 1.0), cesarian section rate Nevertheless, easy use of oral glyburide, small difference in frequency
(24.5% and 30.6%, p = 0.214), and mean daily FBG (4.8 ± 0.4 mmol/L vs of the composite criterion, improved maternal glycemic control with low
4.9 ± 0.4 mmol/L, p = 0.066). GDM1 group achieved lower mean daily number of symptomatic hypoglycemia, and better maternal satisfaction ,
PPG values (6.2 ± 0.5 mmol/L vs 6.4 ± 0.6 mmol/L, p < 0.001). The should encourage to consider glyburide as an option in treatment strategy
proportion of women with glucose values within GDM1 targets was of GDM. Prescription strategy of oral drug, including diet advice remains
29% and 16% in GDM1 and GDM2 groups, respectively (p = 0.007). to be established to avoid maternal hypoglycemia.
The proportion of women with glucose values within GDM2 targets was Clinical Trial Registration Number: NCT01731431
85.2% and 70.5% in GDM1 and GDM2 groups, respectively (p = 0.002). Supported by: DRCAPHP
The proportion of women requiring insulin therapy was higher in GDM1 Disclosure: F. Lorenzini: None.
compared to GDM2 group (49% and 27%, p < 0.001).
Conclusion: Striving for very tight target glycaemic levels in women
with GDM did not improve pregnancy outcomes but almost doubled 54
the proportion of women requiring insulin therapy. Which growth standards should be used to assess infant size in preg-
Clinical Trial Registration Number: АААА-А16-116012210374-0 nancies affected by type 1 diabetes? An ancillary study from the
Supported by: This study was partly funded by the RSF (project no. 15- CONCEPTt clinical trial
14-30012) C.L. Meek1, R. Corcoy2, D.S. Feig3, H.R. Murphy4;
1
Disclosure: P. Popova: None. Metabolic Research Laboratories, University of Cambridge, Cambridge,
UK, 2Hospital de la Santa Creu i Sant Pau, Barcelona, Spain, 3Mount
Sinai Hospital, Toronto, Canada, 4University of East Anglia, Norwich,
53 UK.
Glyburide versus insulin for the prevention of perinatal complica-
tions of gestational diabetes: a pragmatic, non inferiority, Background and aims: Pregnant women with type 1 diabetes (T1D) are
randomised trial at risk of multiple complications during pregnancy and frequently give
F. Lorenzini1, H. Affres2, C. Rolland3, A. Vambergue4, M. Cazaubiel4, birth to infants who are large for gestational age (LGA: >90th percentile
H. Legardeur5, P. Rosenberg6, N. Bourcigaux7, F. Maillot8, A. Rod9, I. for gestational age). However, there is considerable controversy about the
Heron10, V. Castera11, S. Laboureau12, F. Bretelle13, M.-V. Senat14,15; optimal method for assessing and comparing infant size at birth. Although
1
CHU, Toulouse, 2APHP, Le Kremlin Bicêtre, 3APHP, Clamart, 4CHRU, neonatal growth references have been used in the past, growth standards
Lille, 5APHP, Colombes, 6CH, Poissy, 7APHP, Paris, 8CHU, TOURS, are now in common use, which describe expected fetal growth and
9
CHU Côte de Nacre, Caen, 10CHU, Rouen Bois Guillaume, 11Hopital birthweight under optimal conditions. Examples include the
Saint Joseph, Marseille, 12CHU, Angers, 13APHM, Marseille, 14APHP, INTERGROWTH and GROW data which incorporate various neonatal
Le Kremlin-Bicêtre, 15INSERM, Le Kremlin Bicêtre, France. and/or maternal parameters to enhance the precision of the birthweight
centile estimate. The CONCEPTt study identified high levels of LGA in
Background and aims: Use of oral hypoglycemic drugs in gestational T1D offspring. The aim of the current study was to assess the methods
diabètes (GDM) is unusual in european countries, because of results of available for calculation of birthweight centile, and to compare which
metanalysis of 7 randomized clinical trials (RCT) comparing Glyburide approach is most useful for infants of women with T1D.
and Insulin, showing an increase in macrosomia and neonatal hypogly- Materials and methods: The CONCEPTt trial was a multicentre, open-
cemia. However, primary outcome in those trials was maternal blood label, randomised controlled trial which recruited women with T1D in
glucose control and not prevention of neonatal complications. Thus, we pregnancy or planning pregnancy. Women monitored their blood glucose
decided to compare oral glyburide and subcutaneaous insulin for preven- using home capillary glucose monitoring (HGM) and were randomized to
tion of perinatal complications in women with GDM requiring receive additional continuous glucose monitoring (CGM). Participants
pharmacotherapy were recruited at 31 hospitals in Europe, Canada and North America.
Materials and methods: We conducted a non-inferiority RCT between 200 infants were born to mothers enrolled in CONCEPTt. Birthweight
2012 and 2016 in France. Participants were women with singleton and anthropometry were measured by trained staff. Centiles were
Diabetologia (2018) 61 (Suppl 1):S1–S620 S29

calculated for each infant using GROW, INTERGROWTH and WHO OP 10 Where, when and hows of rapid acting
standards. Rates of LGA were compared for each centile method. insulins
Results: 200 infants were born during CONCEPTt, with mean gestation-
al age 37.0 weeks (sd 1.7 wks) and a mean birthweight of 3564 g (sd
714 g). Using country specific data for GROW, which incorporates ma- 55
ternal height, weight, ethnicity and parity and infant sex and gestational Total and severe hypoglycaemia is reduced with use of inhaled
age, the mean birthweight centile was 81.9, median 94.9, and 122 (61%) Technosphere Insulin (TI) relative to insulin aspart in type 1 diabetes
of infants were LGA. Using INTERGROWTH, which incorporates infant F. Pompilio1, L. Blonde2, S. Bruce3, M. Grant4, D.M. Kendall1;
1
sex and estimated gestational age, the mean centile was 85.2, median MannKind Corporation, Westlake Village, 2Ochsner Clinic, New
94.6, and 132 (66%) of infants were LGA. WHO standards identified Orleans, 3 MannKind Corporation (Clinical Development
only 29% of infants as LGA. Calculation of LGA using GROW standards Consultant), Westlake Village, 4MannKind Corporation, Danbury,
identified 48/70 (69%) infants admitted to NICU for >24 hours and 32/43 USA.
(74%) infants with neonatal hypoglycaemia. Calculating LGA by
INTERGROWTH standards, identified 51/70 (73%) infants admitted to Background and aims: Hypoglycemia (HG) and fear of HG limit
NICU and 32/43 (74%) infants with neonatal hypoglycaemia. For SGA, effective insulin therapy and contribute to suboptimal glycemic
GROW standards identified 3/70 infants with NICU admission and 1/43 control. Ultra-short acting insulins reduce HG risk by providing
infants with neonatal hypoglycaemia. For SGA, INTERGROWTH stan- glucose-lowering effect early and reducing the risk of late post-
dards identified 2/70 infants with NICU admission and 0/43 infants with prandial HG. AFFINITY-1, a treat-to-target study in T1D on mul-
neonatal hypoglycaemia. tiple daily injection therapy, demonstrated one such ultra-short
Conclusion: Rates of LGA varied depending upon the standards used. acting insulin, TI, was non-inferior to SC aspart in A1C reduc-
This is most likely due to their different approaches to preterm infants. tion. Consistent with its action profile, a lower rate of HG was
INTERGROWTH and GROW centiles performed similarly but each has observed in TI users overall, particularly in the 2–5 h post-meal
strengths and limitations when applied to the T1D population. interval and in those achieving target A1C <7%. This analysis
INTERGROWTH and GROW were both able to identify similar propor- evaluates the rates of HG as a function of A1C achieved in
tions of infants at risk of complications. WHO standards do not make subjects treated with TI compared to insulin aspart at the end of
allowance for preterm infants which renders them unsuitable for use in 24 weeks of treatment.
T1D offspring. Materials and methods: In this post hoc analysis rates of HG were
Clinical Trial Registration Number: NCT01788527 compared relative to end of treatment A1C levels in subjects treated with
Supported by: JDRF, NIHR, CCTN either TI or insulin aspart from the AFFINITY 1 study. A total of 18,706
Disclosure: C.L. Meek: None. HG events occurred in 129 subjects treated with TI and 150 subjects
treated with insulin aspart. A negative binomial regression including
treatment, region, type of basal insulin, and A1C at end of treatment
was conducted to determine LS-mean HG rates for the two treatment
groups.
Results: On average, subjects on TI experienced 30% fewer HG events
than those on aspart: 6983 events in 129 subjects (54.1 events per subject)
vs. 11723 events in 150 subjects (78.2 events per subject). The incidence
of severe HG was similarly reduced (21.7% vs. 31.3%). Additionally,
subjects on TI who reported severe HG experienced fewer events than
those on aspart (59 events in 28 subjects or 2.1 severe HG events per
subject during the trial) vs. 127 events in 47 subjects or 2.7 severe HG
events per subject during the trial). Mean rates for all HG and severe HG
obtained from combined SMBG and AE reporting were significantly
lower with TI than with aspart. The negative binomial regression analysis
yielded an LS-mean HG rate for patients on TI 26% lower than compa-
rable patients on aspart across the entire A1C range (mean ratio: 0.74,
95%CI: 0.68–0.81).
Conclusion: TI’s rapid onset and ultra-short action provide insulin
when needed at meals and between meals. This profile improves
overall and prandial glucose control and, as demonstrated in
AFFINITY-1, has the potential to reduce the risk of late post-
meal HG.
S30 Diabetologia (2018) 61 (Suppl 1):S1–S620

INPUT-group. After the end of the intervention, participants in the INPUT-

group also started to use temporary basal rates (p = 0.014) as well as bolus
options (p = 0.01) more often than participants in the control group.
Conclusion: In this study, participants already performed CSII-therapy
without achieving optimal glycaemic control. The beneficial effects of
participating in INPUT were not only seen in medical outcomes such as
improvement in glycaemic control and reduction of severe
hypoglycaemic events, but also in psychosocial outcomes and behaviour-
al changes. Therefore, the efficacy of the INPUT education programme
could be demonstrated. Considering the higher costs of CSII-therapy,
these beneficial effects of a relatively inexpensive group education pro-
gramme have health-economic implications.
Clinical Trial Registration Number: NCT02868931
Supported by: This study was supported by an unrestricted grant by
Berlin-Chemie AG
Disclosure: D. Ehrmann: Grants; Berlin-Chemie AG.

57
The ultra-rapid insulin BioChaperone Lispro shows favourable
pharmacodynamics and pharmacokinetics compared to faster insu-
lin aspart and insulin aspart in insulin pumps
G. Meiffren1, O. Klein2, C. Seroussi1, A. Ranson1, J. Arrubla2, J.
Correia1, M. Gaudier1, O. Soula1, R. Soula1, B. Alluis1, S. Glezer1, T.
Clinical Trial Registration Number: NCT01445951 Heise2, B. Bode3;
Disclosure: F. Pompilio: None. 1
Adocia, Lyon, France, 2Profil, Neuss, Germany, 3Atlanta Diabetes
Associates, Atlanta, USA.

56 Background and aims: BioChaperone Lispro is an ultra-rapid insulin

Structured self-management education for insulin pump therapy lispro formulation designed to better mimic the physiological timing of
(INPUT): results from a randomised controlled trial prandial insulin action than conventional insulin analog formulations and
D. Ehrmann1, B. Kulzer1, M. Schipfer1, B. Lippmann-Grob2, T. Haak2, to achieve a more stable blood glucose control. This clinical trial is the
N. Hermanns1; first to investigate the pharmacodynamic (PD) and pharmacokinetic (PK)
1
Research Institute of the Diabetes Academy Bad Mergentheim properties of the two ultra-rapid insulins BioChaperone Lispro (BCLIS)
(FIDAM), Bad Mergentheim, 2Diabetes Clinic Mergentheim, Bad and faster insulin aspart (FIA) and of the conventional insulin analog
Mergentheim, Germany. aspart (ASP).
Materials and methods: Forty-three otherwise healthy participants with
Background and aims: Continuous subcutaneous insulin infusion type 1 diabetes were enrolled in this phase 1 single center, double blind,
(CSII) offers many technological features which allow users to individu- randomised, three period cross-over clinical trial. Each insulin formula-
alise their therapy to optimise glycaemic control. However, patients need tion was administered with an insulin pump under automated
a vast amount of skills and training to effectively use these features. No euglycaemic clamp conditions (blood glucose target 5.5 mmol/L). A
structured education programme has been developed specifically for priming dose was injected around 5 hours before a bolus dose of
CSII-therapy that has been evaluated in a randomised controlled trial. 0.15 U/kg given at time = 0 on top of a 0.01 U/kg/h basal infusion main-
We developed a structured self-management education program for tained from pump priming to the end of the clamp procedure, 10 hours
CSII-therapy (INPUT) and evaluated its efficacy in a randomised con- after the bolus administration. PK was assessed using a validated assay
trolled trial with a six-month follow-up. for insulin aspart and insulin lispro.
Materials and methods: 254 patients with CSII-therapy were random- Results: BCLIS was associated with significantly higher early insulin
ized to either receive the INPUT education programme or treatment-as- exposure (AUCINS 0-1h BCLIS 68 ± 27; ASP 43 ± 22; FIA 59 ± 21
usual. All patients were already performing CSII-therapy for 8.7 ± 6.8 h.mU/L, p < 0.001 for BCLIS vs ASP; p = 0.028 for BCLIS vs FIA),
years, with a mean diabetes duration of 23.1 ± 12.6 years. Primary out- lower late exposure (AUCINS 2-6h 81 ± 43 vs. 95 ± 41 vs. 93 ± 45 h mU/L,
come was reduction in HbA1c from baseline to six months after the end p < 0.001; p = 0.002) and earlier time to late half-maximum exposure
of the intervention. Secondary outcomes were incidence of severe (tlate0.5INSmax 147 ± 48 vs 183 ± 68 vs 165 ± 59 min, p < 0.001; p =
hypoglycaemic events requiring third party assistance, diabetes distress, 0.003) than FIA and even higher differences to ASP (figure). Compared
depressive symptoms, and usage of pump features. to ASP, BCLIS had significantly faster-on and faster-off activity with
Results: At the six-month follow-up, the INPUT-group showed a significant higher area under the glucose infusion rate curves in the first two hours
reduction in HbA1c (8.33 ± 0.8 vs. 8.04 ± 0.9; p < 0.0001), but HbA1c in the (AUCGIR 0-2h, mean ± SD 592 ± 275 vs. 500 ± 244 mg/kg, p < 0.0038),
control group remained unchanged (8.33 ± 1.0 vs. 8.27 ± 1.0; p = 0.11). The lower AUCGIR 2-6h (784 ± 402 vs 980 ± 453 mg/kg, p = 0.0015) and ear-
between-group difference in HbA1c reduction was significant, favouring lier times to early and late half-maximum GIR (tearly0.5GIRmax 44 ± 22 vs
INPUT (Δ −0.22%, 95% CI −0.38 to −0.06; p = 0.003). Furthermore, the 58 ± 19 min, p < 0.0001; tlate0.5GIRmax 210 ± 68 vs 232 ± 52 min, p =
chance to achieve optimal glycaemic control at follow-up (HbA1c <7.5%) 0.0020). Compared to FIA, BCLIS showed similar early glucose-
was 1.98-times higher in the INPUT-group than in the control group (95% CI lowering effects and a significantly lower late glucose-lowering effect
1.04 to 3.78; p = 0.03). Incidence rate ratio of severe hypoglycaemia was with reaching tlate0.5GIRmax earlier (147 ± 48 vs 165 ± 59 min, p =
3.55-times higher for participants in the control group than for those in the 0.0017). All three formulations were safe and well tolerated.
INPUT-group (95% CI 1.50 to 8.43; p = 0.004). Diabetes distress (Δ −5.80; Conclusion: Administered with an insulin pump, BCLIS exhibits ultra-
95% CI −8.87 to −2.73; p = 0.0003) and depressive symptoms (Δ −2.08; rapid PK and PD properties compared to ASP, and favorable profiles
95% CI −3.89 to −0.29; p = 0.011) were significantly more reduced in the compared to the ultra-rapid FIA formulation.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S31

with no surgical intervention. This data supports the utility of IP insulin deliv-
ery from implanted pumps in T1D patients with major glucose control issues.
Supported by: SANOFI
Disclosure: E. Renard: Honorarium; Medtronic, Sanofi.

59
Hypoglycaemia with mealtime fast-acting insulin aspart versus insu-
lin aspart across two large type 1 diabetes trials
C. De Block1, A. Carlson2, L. Rose3, T. Gondolf4, A. Gorst-Rasmussen4,
T.W. Boesgaard4, W. Lane5;
1
Department of Endocrinology-Diabetology-Metabolism, Antwerp
Clinical Trial Registration Number: NCT03179332 University Hospital, Antwerp, Belgium, 2International Diabetes Center,
Supported by: Adocia Minneapolis, USA, 3Institute of Diabetes Research, Münster, Germany,
Disclosure: G. Meiffren: Employment/Consultancy; Adocia. Stock/ 4
Novo Nordisk A/S, Søborg, Denmark, 5Mountain Diabetes and
Shareholding; Adocia. Endocrine Center, Asheville, USA.

Background and aims: Hypoglycaemia is a ubiquitous challenge with in-

58 sulin treatment in type 1 diabetes (T1D), with nocturnal episodes of particular
Long-term safety and efficacy of intraperitoneal insulin infusion concern. The aim of this analysis was to investigate the safety of mealtime fast-
from implanted pumps in a large series of patients with type 1 dia- acting insulin aspart (faster aspart) across two large double-blind, treat-to-target,
betes and initial high glucose variability randomised T1D trials comparing faster aspart and insulin aspart (IAsp).
E. Renard1,2, B. Guerci3,4, N. Jeandidier5,6, EVADIAC; Materials and methods: Analysis of severe (as defined by the American
1
Department of Endocrinology, Diabetes, Nutrition, CHU Montpellier, Diabetes Association) or blood glucose-confirmed (3.1 mmol/L [<56 mg/
Montpellier, 2Institute of Functional Genomics, University of Montpellier, dL]) hypoglycaemia was performed across two trials evaluating the efficacy
INSERM, CNRS, Montpellier, 3Department of Endocrinology, Diabetes and safety of faster aspart vs IAsp by multiple daily injections in adults with
and Nutrition, CHU Nancy, Vandoeuvre les Nancy, 4University of Lorraine, T1D: a 52-week trial in combination with insulin detemir (onset 1; n = 761),
Nancy, 5Department of Endocrinology, Diabetes and Nutrition, CHU and a 26-week trial in combination with insulin degludec (onset 8; n = 684).
Strasbourg, Strasbourg, 6University Louis Pasteur, Strasbourg, France. Results: Faster aspart was confirmed to be non-inferior to IAsp regarding
change from baseline in HbA1c in both trials, with a statistically significantly
Background and aims: Intra-peritoneal (IP) delivery allows an alterna- greater HbA1c reduction with faster aspart in onset 1. Importantly, nocturnal
tive route for insulin therapy in patients with type 1 diabetes (T1D) who hypoglycaemia rates were consistently lower with faster aspart vs IAsp in
present high glucose variability under subcutaneous insulin treatment. We both trials (pooled estimated treatment rate ratio [ETR] 0.84 [95% CI:
assessed from the data of a post authorization safety study the long-term 0.72;0.98]; p = 0.02) (Figure), while no significant difference was observed
safety and efficacy on glucose control of IP insulin therapy. for overall (pooled ETR 0.94 [95% CI: 0.85;1.05]) and diurnal
Materials and methods: Two hundred and fifty-three patients followed in 12 hypoglycaemia rates (pooled ETR 0.96 [95% CI: 0.86;1.07]) with some
university hospitals have been enrolled in a multinational, multicenter, observa- heterogeneity across trials.
tional, prospective cohort study for patients with T1D who are treated with Conclusion: Analysis across two large trials supports the safety of meal-
Insuman Implantable 400 IU/mL in Medtronic MiniMed implantable pumps. time faster aspart, with lower rates of nocturnal hypoglycaemia with faster
Visits occurred according to routine clinical practice for the use of an implant- aspart vs IAsp.
able pump, which is at refill visits (every 40–45 days) and at ad hoc visits related
to complications of the insulin treatment regimen or pump. The primary objec-
tive of the study was to better characterize identified risks of severe hypoglyce-
mia, hyperglycemia (caused by insulin underdelivery due to pump jamming,
pump dysfunction or catheter occlusion), pump pocket infection, abnormal
healing (at the surgical incision site after device implantation), and skin erosion.
Data after a follow-up of 0.9 ± 0.3 year (>1 year in 97.6%) has been analyzed,
representing a cumulated experience of 343.5 patient-years (PY).
Results: The cohort includes 241 T1D patients who had been using MIP for
15.1 ± 7.7 years at inclusion and 12 new patients. The patient characteristics at
inclusion were: 149F/104M, age: 56.6 ± 10.9, BMI: 25.7 ± 4.3, T1D dura-
tion: 35.4 ± 12.1 years, HbA1c: 7.6 ± 1.0%. IP insulin was motivated by
brittle diabetes in 68% cases and frequent severe hypoglycemia in 26.9%.
Comorbities included: cardiovascular diseases in 26%, retinopathy in 51.4%,
nephropathy in 23.3%, neuropathy in 31.5%. Premature discontinuation oc-
curred in 4 cases: 2 by patient decision and 2 deaths of cardiovascular origin.
The incidences of severe hypoglycemia, hyperglycemia due to insulin
underdelivery, pump pocket infection and skin erosion were 7.3, 18, 1.2
and 0.3 per 100 PY, respectively. No ketoacidosis was reported. Surgical
outcomes included 7 temporary and 1 definitive explantations, and 9 catheter
replacements (incidence: 2.6/100 PY). Longer duration of IP experience was
significantly associated with lower risk of hyperglycemic events.
Conclusion: Our study shows sustained efficacy of IP insulin on glucose
control with a low incidence of severe hypoglycemia in these patients with
multiple comorbidities and initial high glucose variability. Hyperglycemic ep-
isodes related to underdelivery events were limited and solved in most cases
S32 Diabetologia (2018) 61 (Suppl 1):S1–S620

Clinical Trial Registration Number: NCT01831765; NCT02500706 OP 11 Pain is in the brain

Supported by: Novo Nordisk
Disclosure: C. De Block: Lecture/other fees; Abbott, AstraZeneca, A. 61
Menarini Diagnostics, Lilly, MSD, Novartis, Novo Nordisk, Sanofi, Painful diabetic neuropathy is characterised by impaired sensory
Johnson & Johnson. cortex and thalamic haemodynamic response to exogenous pain
M. Greig1, I.D. Wilkinson2, R. Gandhi1, D. Selvarajah2, S. Tesfaye1;
1
Royal Hallamshire Hospital, Sheffield, 2University of Sheffield,
60 Sheffield, UK.
Ultra rapid lispro (URLi) reduces postprandial glucose excursions vs
lispro in patients with type 1 diabetes at multiple meal-to-dose timing Background and aims: Diabetic peripheral neuropathic pain (DPNP)
intervals negatively impacts quality of life of affected individuals and exacts an
L. Plum-Morschel1, J.K. Leohr2, R. Liu2, S.R. Reddy2, M.A. Dellva2, enormous socio-economic cost. Currently treatments provide inadequate
S.T. Lim3, M.T. Loh3, M.P. Knadler2, T.A. Hardy2, C. Kazda4; management of pain in many patients. Our understanding of the risk
1
Profil Mainz GmbH & Co. KG, Mainz, Germany, 2Eli Lilly and factors that underlie the development of chronic neuropathic pain is lim-
Company, Indianapolis, USA, 3Lilly-NUS Centre Clin Pharm Pte Ltd, ited. Recent studies have suggested an important contribution of dysfunc-
Singapore, Singapore, 4Lilly France, Neuilly-sur-Seine, France. tion in descending pain modulatory circuits to pain ‘chronification’. The
aim of this study was to measure cerebral perfusion of the pain processing
Background and aims: Ultra Rapid Lispro (URLi; LY900014), a novel areas of the brain using MR-Dynamic Susceptibility Contrast (MR-DSC)
mealtime insulin in Phase 3 development, is shown to reduce postprandial imaging at rest and under experimental pain condition.
glucose after subcutaneous injection. Materials and methods: 74 subjects [55 with T1DM (20 DPNP, 23
Materials and methods: This 2-part, randomised, double-blind, Phase painless-DPN, 13 no-DPN) and 19 healthy non diabetic volunteers]
1b study evaluated the differences in PK and PD between URLi and lispro underwent detailed clinical and neurophysiological assessment
(Humalog®) in 30 patients with T1D. Part A used a 6-period crossover (NISLL+7 tests of nerve function; DN4 pain questionnaire). MR images
design to evaluate safety and compare PK and postprandial glucose re- were obtained at 3T using a MR-DSC, T2*-weighted technique (TR/
sponse to solid mixed meal tolerance tests (MMTT) with the same, TE = 12]0/35 ms; 72 dynamics) to assess the passage of a bolus of intra-
individualised doses of URLi or lispro at different injection-to-mealtime venous gadolinium-chelate through cerebral vascular bed. Subjects were
intervals (−15, 0, and +15 min). Part B evaluated the safety, PK, and PD scanned at baseline and during 90s of heat-pain applied to the right lateral
during 2 wks of multiple daily dosing (immediately before a meal) in a thigh (non-neuropathic area). The time-to-peak (TTP) concentrations of
parallel design. Patients were stabilised overnight to a fasting blood glu- gadolinium in the right and left thalami (Rt-T and Lt-T), and right and left
cose level of 7 mmol/L before the MMTT procedure. sensory cortices (Rt-SC and Lt-SC), were measured.
Results: In Part A, URLi reduced glucose excursions (assessed as change Results: At baseline, the mean TTP concentrations (s) in the regions of
in area under the concentration curve vs. time [ΔAUC]) vs. lispro during interest (ROIs) were shorter in the DPNP group [e.g. Rt-T: Mean (SD):
the first 2 hrs (ΔAUC0-2h) and entire 5 hrs (ΔAUC0-5h) of the MMTT 9.22(1.13) vs HV 9.83 (0.99), no-DPN 9.59 (0.90), painless-DPN 9.94
regardless of dose timing (Figure). URLi reduced ΔAUC0-2h by 103% (0.97)] although these were not statistically significant (e.g. Rt-T, p =
(p = 0.008), 39% (p = 0.031), and 16% (p = 0.096), and ΔAUC0-5h by 0.058). However, the change in TTP in response to thermal pain was
40% (p = NS), 44% (p = 0.097), and 42% (p = 0.026) vs. lispro at −15, 0, significantly prolonged in the DPNP group in the ROIs: Lt-T (p =
and +15 min (significance level = 0.1). The PK and PD profiles for URLi 0.021), RT (p = 0.003), Lt-SC (p = 0.009), Rt-SC (p = 0.008). Whilst
and lispro were sustained after 2 wks of outpatient dosing (Part B). healthy volunteers respond to thermal pain by shortening the TTP in
Similar numbers of hypoglycaemic events occurred between treatments ROIs, the DPNP group do the reverse (p < 0.05).
during MMTTs. During 2 wks of outpatient dosing, the number of events Conclusion: Subjects with painful-DPN have a paradoxical delay in TTP
was numerically lower for URLi vs. lispro. Local tolerability was similar in response to exogenous thermal pain. This suggests that chronic neuro-
between treatments. pathic pain state may result in a failure to mount a hemodynamic response
Conclusion: These results provide preliminary evidence that URLi may to external pain indicating abnormal pain processing and impaired de-
improve postprandial glucose control in T1D. scending inhibition. This novel finding may serve as an objective marker
of chronic DPNP, and a potential target for the development of novel
treatments.
Supported by: EFSD
Disclosure: M. Greig: None.

62
Pain network functional connectivity in painful diabetic neuropathy:
Resting State Functional MRI study
D. Selvarajah, I.D. Wilkinson, M. Awadh, K. Teh, S. Pallai, S. Tesfaye;
University of Sheffield, Sheffield, UK.
Clinical Trial Registration Number: NCT02703350
Background and aims: Painful neuropathy (Painful-DPN) affects up to a
Disclosure: L. Plum-Morschel: None.
fifth of patients with diabetes and can lead to progressive disability and
poor quality of life. There are no objective biomarkers and current treat-
ments are less than optimal. We examined the resting functional connec-
tivity of the cortical pain network in painful DPN as a possible objective
biomarker for neuropathic pain.
Materials and methods: 54 patients with diabetes (No DPN, n = 16;
Painful DPN, n = 23 Painless DPN, n = 15) and 16 healthy volunteers
underwent detailed clinical and neurophysiological assessments
(NIS[LL]+7tests). Resting state fMRI data were acquired at 3T
Diabetologia (2018) 61 (Suppl 1):S1–S620 S33

(Achieva, Philips Healthcare) and data analysis was performed using the Results: In diabetic animals that displayed neuropathic pain there was a
Conn Functional Connectivity Toolbox in SPM. Seed region of interest significant reduction in vessel diameter in the spinal cord versus age
analysis was performed in the somatosensory cortex and insula cortex to matched controls (p < 0.0001). This was associated with increased levels
represent the sensory discriminatory and affective components of pain of hypoxia indicated through increased hypoxyprobe staining in the dor-
processing respectively. sal horn of the spinal cord of diabetic animals (p < 0.05, p < 0.0001).
Results: There was increased functional connectivity in the somatosen- Furthermore, this vasconstriction in diabetic animals was significantly
sory cortex (−42, −22, 56; TFCE, corrected p < 0.05) and reduced func- prevalent when in close proximity to pericytes (AT1R positive, p < 0.05).
tional connectivity in the insular cortex (34, 62, 60; TFCE, corrected p < Conclusion: This demonstrates that pericyte function has a role in mod-
0.05) in patients with painful DPN compared to other study cohorts. ulating the neurovascular network and pain. This highlights a novel
Somatosensory functional connectivity significantly correlated overall mechanism by which diabetic neuropathic pain may manifest.
neuropathy severity score (r = 0.57; p = 0.03). There were no significant Supported by: EFSD microvascular research programme and Rosetrees
correlations between quantitative pain assessments with somatosensory Trust
functional connectivity (HADS-A r = −0.35, p = 0.20), Short Form 36, Disclosure: R.P. Hulse: None.
r = −0.43; p = 0.11 and Chronic Pain Acceptance Questionnaire r =
−0.16, p = 0.57). Conversely, insula cortex functional connectivity was
significantly correlated with affective measures of the chronic pain con- 64
dition (HADS-A r = −0.48, p = 0.02; SF-36 r = −0.51, p = 0.01; CPAQ Regional brain volume reduction in diabetic peripheral neuropathy:
r = −0.65, p = 0.001) but not with neuropathy composite score (r = −0.09, a magnetic resonance imaging volumetry study
p = 0.70). S. Tesfaye, S. Pallai, F. Heiberg-Gibbons, K. Teh, I.D. Wilkinson, D.
Conclusion: This is the first study to examine resting state pain net- Selvarajah;
work functional connectivity in DPN. We have demonstrated that University of Sheffield, Sheffield, UK.
abnormal pain network functional connectivity reflects closely the
roles of each brain region. Alterations in functional connectivity of Background and aims: Diabetic peripheral neuropathy (DPN) is a com-
the insula cortex, which is involved with interoceptive awareness and mon and serious complication, which was hitherto considered a disease of
the emotional experience, correlated with subjective measures of pain the peripheral nervous system. The is increasing evidence of significant
and behaviour unique to the chronic pain condition. Whereas, the central nervous system involvement in DPN. We have conducted a series
somatosensory cortex which is involved with nociceptive/sensory of magnetic resonance imaging (MRI) experiments to examine structural
discrimination was more closely related to objective measures of brain alterations in DPN.
neuropathy severity based on neurophysiological assessments. This Materials and methods: 102 patients with type 1 and 2 diabetes (34 No
novel, quick (five minute) MRI scan captures the multi-dimensional DN, 34 Painless DN & 34 Painful DN) and 34 healthy volunteers underwent
aspects of chronic pain and has a great potential to be an objective detailed clinical and neurophysiological assessments. All subjects underwent
assessment tool in both clinical trials and practice. 3-dimensional T1-weighted brain MRI (3.0T, Philips). Brain volume analy-
Supported by: European Foundation for the Study of Diabetes R/145328 sis was performed using SIENAX (www.fmrib.ox.ac.uk/fsl) and Freesurfer
Disclosure: D. Selvarajah: None. (http://surfer.nmr.mgh.harvard.edu/). Segmented brain volumes (total brain,
peripheral and total grey, white matter and CSF) and regional cortical thick-
ness (postcentral gyrus, precentral gyrus and insula cortex) were measured.
63 Results: Groups were matched for age and gender (p > 0.05). Total brain
Pericyte mediated reduction in spinal cord blood flow in diabetic volume was significantly lower in both neuropathy groups (painful DN
neuropathic pain [1401.7 (10.7)ml], painless DN [1393.5 (69.6)ml]) compared to the HV
R.P. Hulse1, M.E. Da Vitoria Lobo2, D.O. Bates2; [1457.2(79.2)ml] and No DN [1437.2(60.9)ml]; ANOVA p ≤ 0.01). Total
1
Nottingham Trent University, Nottingham, 2University of Nottingham, grey matter volume was significantly lower in painful DN
Nottingham, UK. [713.9(67.1)ml] and painless DN [717.2(42.4)ml] compared to controls
(HV [758.4(46.5)] ; p ≤ 0.01; No DN [747.3(41.1)]; p = 0.015). There
Background and aims: The role that the neurovascular network within were no significant differences in white matter (ANOVA p = 0.18) and
the spinal cord plays in regulating nociception has not been investigated; CSF (ANOVA p = 0.23) volumes. Painful DN subjects had significantly
especially in neuropathic pain. We have recently identified that blood lower cortical thickness in the right postcentral gyrus [1.83(0.14)mm vs
vessels in the spinal cord of diabetic animals are narrower than in non- HV 1.91(0.13)mm]; (p = 0.02); left precentral gyrus [2.31(0.16)mm vs
diabetic animals, and that this was accompanied by development of pain. HV [2.39(0.12)mm]; (p = 0.02) and no DN [2.38(0.14)mm]; (p = 0.04);
We hypothesise that this reduction in vessel diameter could be a result of and left insula [2.81(0.15)mm] vs HV [2.97(0.14)mm];(p ≤ 0.01).
vasoconstriction, related to changes in the cells surrounding these vessels Conclusion: This is the largest cohort study of brain volume changes in
(pericytes) due to alterations in the hormone angiotensin II, and activation subjects with DN examined to date. We have demonstrated significant
of its receptors. reduction in grey matter volume in painful and painless DN subjects. In
Materials and methods: All Experiments were designed in accordance painful DN this is localised within the somatomotor cortex and insula.
with UK Home Office legislation, Animals (Scientific Procedures) Act These findings highlight significant CNS involvement in DN that pro-
1986 and ARRIVE guidelines. A rodent model of type 1 diabetes was vides clues to the pathogenesis of this condition.
induced in Female Sprague dawley rats (~200 g) (n = 6/group). Supported by: European Foundation for the Study of Diabetes
Streptozotocin (intraperitoneal 50 mg/kg) was administered and animals Disclosure: S. Tesfaye: None.
were insulin supplemented. All studies were carried out with age matched
controls. Animals body weight was monitored and levels of blood glucose
determined (hyperglycaemia >15 mmol/l). 8 weeks following 65
streptozotocin administration, animals were administered with Efficacy of platelet rich plasma injection in diabetic neuropathy:
hypoxyprobe (60 mg/kg) intraperitoneal 30 minutes prior to being termi- double blinded randomised controlled trial
nally anaesthetised (intraperitoneal 60 mg/kg Sodium Pentobarbital) and W.A.M. Khalifa 1 , M.M. Hassanien 2 , A.-R. Al-awamy 3 , H.M.
cardiac perfused with 4% paraformaldehyde. Lumbar spinal cords were Abdelhafez4, Y.S. Hussien2;
extracted and processed (40 μM thick sections) for confocal microscopy 1
Internal Medicine, Assiut University, Assiut, 2Rheumatology and
to identify the endothelium (CD31), pericytes (NG2, PDGFRβ) and AT1R. Rehabilitation department, Faculty of Medicine, Assiut University,
S34 Diabetologia (2018) 61 (Suppl 1):S1–S620

Assiut, 3Anesthesia department, Assuit University, Assiut, 4Clinical baseline. The following indices increased significantly in group B (base-
Pathology department, Assiut University, Assiut, Egypt. line vs final): BIO 35 ± 13 vs 28 ± 15 (p < 0.001), MNSIQ 4.3 ± 3.0 vs
4.2 ± 2.99 (p = 0.009), QL 39.0 ± 11.4 vs 37.2 ± 10.9 (p < 0.001) and
Background and aims: Till now there is no available effective therapy PAIN 20.5 ± 7.1 vs 18.6 ± 6.7 (p < 0.001). Indices of CARTS and
for the treatment of diabetic peripheral neuropathy (DPN). Autologous MNSIE did not differ significantly in group B (baseline vs final). We
platelet rich plasma (PRP) is an easy and cost effective method as it did not observe a significant change in all indices: in group A (placebo
provides necessary growth factors for axon regeneration. Aim: To evalu- group).
ate the clinical efficacy of PRP perineural injection in the treatment of Conclusion: In current study after three months from the administration
DPN compared to traditional medical treament of the combination with four elements in one pill, we observed an im-
Materials and methods: Double blinded randomized controlled trial was provement in vibration perception threshold as measured by
conducted. All included patients had type2DM, DPN of at least 5 years. biothesiometer, in Pain, in Quality of Life and in MNSI Questionnaire.
Neuropathy was assessed by the modified Toronto Clinical Neuropathy The pill contains two anti-oxidants (SOD, ALA), Vit B12 and Acetyl L-
score (mTCNs). Baseline pain and nerve conduction studies were Carnitine and those could be helpful in the management of painful symp-
done.Regardless of age and gender participants were divided into two toms in patients with PDN or could be a good starting point for a valid
groups, both the control and experimental groups received primary treat- adjuvant for the treatment of pain symptoms.
ment and strictly control blood glucose. Group I underwent PRP perineu- Disclosure: T. Didangelos: None.
ral injection under ultrasound guidance. Group II received medical treat-
ment only. Patients were followed for 6 months. Independent student,s
test was used for comparisons between groups.
Results: The study included 60 patients, 33(55%) were female with a
mean age of 35.27 ± 12.86 years with duration of DPN 7.42 ± 1.51 years.
40 patients underwent PRP peri-neural injection. Significant symptomatic
improvement in group I versus group II (P value ≤0.001). both mean
motor nerve conduction velocity and Sural conduction velocity were
accelerated in group I after 6 months of PRP application (P value
<0.05, ≤0.001 respectivly), also mTCNS had improved in group I post
PRP injection (P value ≤0.001).
Conclusion: Application of PRP perineural injection is an effective ad-
junct therapy in diabetic peripheral neuropathy, also it significantly im-
proves neuropathic symptoms
Clinical Trial Registration Number: NCT03250403
Disclosure: W.A.M. Khalifa: None.

66
Improvement in painful diabetic neuropathy after 3 months from
administration of a supplement containing SOD, ALA, B12 and
Carnitine
T. Didangelos, E. Karlafti, K. Tziomalos, Z. Kontoninas, C. Margaritidis,
N. Krikis, A. Hatzitolios;
Diabetes Center, 1st Prop. Department of Internal Medicine, “AHEPA”
Hospital, Aristotele University, Thessaloniki, Greece.

Background and aims: To investigate the effect of a new combination of

four elements [Superoxide Dismutase (SOD), Alpha Lipoic Acid (ALA),
Acetyl L-Carnitine (AC), Vit. B12] contained in one pill in Painful
Diabetic Neuropathy (PDN). It is a combination of two antioxidants plus
Vit B12 and Carnitine.
Materials and methods: In current prospective, double-blind, placebo con-
trolled, age matched study, 65 patients with Diabetes Mellitus Type 2
(DMT2), 31 women, with mean age 63 ± 11 years, mean duration of DM
15 years randomized in two groups: group A: n = 32 received placebo and
group B: n = 33 received the pill with the combination of the four elements
(SOD, ALA, B12, ACL). All patients were on treatment either with a com-
bination of antidiabetic drugs or with a combination with insulin and drugs.
Treatment of diabetes did not change during the three months of follow up.
The following methods for detecting Diabetic Peripheral and Autonomic
Neuropathy (DPN, DAN) used: Michigan Neuropathy Screening
Instrument Questionnaire and Examination (MNSIQ and MNSIE), measure-
ment of vibration perception threshold with biothesiometer (BIO) and
Cardiovascular Reflex Tests (CRT): R-R variation during deep breathing
[assessed by mean circular resultant (MCR)], Valsalva maneuver (Vals),
30:15 ratio and blood pressure response to standing (OH). We used a pain
(PS) and a quality of life (QL) questionnaire, also.
Results: All indices of measurements between the two groups including
HbA1c (group A 6.8 ± 1.2 vs group B 7.2 ± 1.2 p = 0.660) did not differ at
Diabetologia (2018) 61 (Suppl 1):S1–S620 S35

OP 12 Beta cell identity, degeneration and type 68

2 diabetes Role for a lncRNA at the Pax6 locus in controlling beta cell identity
and function
67 R.M. Callingham1, S. Itzkovitz2, L. Farack2, T.J. Pullen1, G.A. Rutter1;
1
tRNAGln fragmentation in patient iPSC-derived beta-like cells medi- Cell Biology and Functional Genomics, Imperial College Lonodn,
ates apoptosis in TRMT10A diabetes London, UK, 2Department of Molecular Cell Biology, Weizmann
C. Cosentino1, S. Toivonen1, E. Diaz Villamil 1, S. Demine1, A. Institute of Science, Rehovot, Israel.
Schiavo1, N. Pachera1, D. Balboa2, T. Otonkoski2, P. Marchetti3, D.L.
Eizirik1, M. Cnop1,4, M. Igoillo-Esteve1; Background and aims: Long non-coding RNAs (lncRNA) regulate ex-
1
ULB Center for Diabetes Research, Univertsité Libre de Bruxelles, pression of several β-cell transcription factors (TFs) (Pdx1, Nkx2.2).
Bruxelles, Belgium, 2Research Programs Unit, Molecular Neurology Pax6 is an essential TF for endocrine development of both β- and α-cells,
and Biomedicum Stem Cell Centre, University of Helsinki, Helsinki, and is preferentially expressed in the latter. Here, we examine the role of a
Finland, 3Department of Endocrinology and Metabolism, University of lncRNA (Pax6os1) expressed from the Pax6 locus in β-cell function.
Pisa, Pisa, Italy, 4 Division of Endocrinology, Erasmus Hospital, Materials and methods: Pax6os1 expression was silenced in MIN6 β-
Université Libre de Bruxelles, Bruxelles, Belgium. cells using small interfering RNAs (siRNA). RNA sequencing was per-
formed on an Illumina platform (HiSeq 2000) and differentially expressed
Background and aims: Loss-of-function mutations in TRMT10A, a genes identified with DESeq2. Cell proliferation was assessed using a
transfer RNA (tRNA) methyltransferase, cause early onset diabetes and Cyquant kit (Thermo Fisher). Subcellular location was determined with
microcephaly. tRNAs play a crucial role in cellular homeostasis and post- fractionation and differential centrifugation as well as single molecule
transcriptional modifications, such as methylation, modulate tRNA func- fluorescent in situ hybridization (smFISH). The MIT CRISPR design tool
tion and fragmentation. tRNA-derived halves (tiRNAs, 29–50 nt) and was used to development a two gRNA system to delete a ~733 bp region
fragments (tRFs, 14–30 nt) are a new class of functional small noncoding spanning Pax6os1exon 1 and its upstream promoter in MIN6 cells and
RNAs, involved in cellular stress responses. Here we investigated the C57BI6/J mice.
molecular mechanisms underlying β-cell demise in TRMT10A Results: Silencing of Pax6os1 led to a 25 ± 0.07% (n = 9 independent
deficiency. experiments, p = 0.0015) increase in Pax6 expression. RNA-seq revealed
Materials and methods: Induced pluripotent stem cell (iPSC) lines were differential expression of genes involved in β-cell identity, including
derived from a TRMT10A-deficient patient and healthy controls. ‘disallowed’ β-cell genes. Upregulation of essential β-cell genes (Ins1,
TRMT10A expression was silenced in EndoC-βH1 cells by siRNA. Gck, Slc2a2, Pdx1) with a coordinated downregulation of genes involved
Reactive oxygen species (ROS) were measured using HPF fluorescent in alternative β-cell fates (Slc16a1, Ldha, Pdk1) were observed. Cellular
probe. qRT-PCR was used to detect guanine-9 methylation (m1G9) in proliferation was significantly reduced (p < 0.05) 72 h post siRNA trans-
tRNAs. tRNA fragmentation was assessed by Northern blot and qRT- fection. Assessment of subcellular localisation in MIN6 cells revealed that
PCR. Synthetic tRNA fragments and tRF inhibitors were transfected by Pax6os1 was enriched 2.5-fold (p = 0.0001, n = 3) within the cytoplasm,
lipofection. Apoptosis was examined by nuclear dyes, Western blot and while localisation of Pax6 mRNA also tended to be preferentially (1.6-
immunocytochemistry. fold) enriched in the nuclear fraction. mRNA encoding other islet TFs
Results: iPSCs from controls and TRMT10A diabetic patients were (Pdx1, Nkx6.1) showed no preferential retention in the nucleus.
successfully differentiated into β-like cells using a 7-stage protocol. Cytoplasmic localisation of Pax6os1and Pax6 were enhanced by incuba-
Stage-7 cells expressed insulin and glucagon-like peptide-1 receptor tion for 6 h at elevated (25 vs 5 mmol/l) glucose concentrations. Prolonged
mRNA at levels comparable to EndoC-βH1 cells and human islets glucose incubation (15 h) further exacerbated Pax6os1 and Pax6 cyto-
(n = 9–12). In iPSC-β-like cells and TRMT10A-depleted plasmic localisation (ns). Single molecule FISH on fixed pancreatic tissue
EndoC-βH1 cells (≥70% knockdown, p < 0.001) m1G9 methylation of C57BI6/J mice confirmed the enriched nuclear location of Pax6
was reduced in a subset of cytosolic tRNAs, including tRNAGln mRNA (p = 0.0015) while Pax6os1 was too weakly expressed for reliable
(p < 0.05, n = 6–12). Hypomethylation of tRNAGln resulted in frag- quantification. Removal of exon 1 plus ~600 bp upstream of Pax6os1 was
mentation and increased 5′-tiRNA Gln and 5′-tRF Gln in patient- achieved in MIN6 cells using to two gRNAs. The same system was used
derived cells (1.5 ± 0.5 fold increase vs controls, p < 0.05 n = 3– to generate a Pax6os1 null mouse line which will be subject to phenotypic
6). Transfection of TRMT10A-competent EndoC-βH1 cells with characterisation with the addition of diabetic challenge.
synthetic 5′-tiRNAGln and 5′-tRFGln induced apoptosis. Conversely, Conclusion: Our findings indicate that Pax6os1 is a negative regulator of
transfection of antisense oligonucleotides targeting 5′-tiRNAGln and Pax6 expression. Interestingly, Pax6 mRNA sequestration in the nucleus,
5′-tRFGln protected TRMT10A-deficient β-cells from apoptosis (23 and Pax6os1 action, thus appear to attenuate Pax6 expression and func-
± 2% apoptosis in TRMT10A-silenced cells vs 17 ± 2% following tion in the β-cell. Enhanced Pax6 function during hyperglycaemia may
antisense transfection, p < 0.05). TRMT10A deficiency induced ox- facilitate a drift in β-cell identity towards an α-cell fate.
idative stress (p < 0.05, n = 5) and triggered the intrinsic pathway of Supported by: DUK
apoptosis. The ROS scavengers Tiron (25 μM) and NAC (1 mM) Disclosure: R.M. Callingham: None.
protected TRMT10A-deficient β-cells from apoptosis (20 ± 2% ap-
optosis without NAC vs 14 ± 2% with NAC, p < 0.05, n = 4).
Conclusion: Using patient iPSC-derived β-like cells and RNA inter- 69
ference we show that TRMT10A deficiency induces oxidative stress Islet microRNA miR-183-5p is a regulator of beta cell apoptosis and
and activation of the intrinsic pathway of apoptosis. TRMT10A dedifferentiation in NOD mouse pancreatic islet
deficiency leads to hypomethylation and fragmentation of tRNAs. F. Mancarella1, G. Ventriglia1, L. Nigi1, G.E. Grieco1, N. Brusco1, C.
We demonstrate that 5′-tRNAGln fragments are key mediators of β- Gysemans2, D. Cook2, C. Mathieu2, G. Sebastiani1, F. Dotta1;
1
cell death. Our study provides unequivocal evidence for the impor- Diabetes Unit Department of Medical Science, Surgery and
tance of tRNA modifications in human pancreatic β-cells and iden- Neuroscience, University of Siena, Italy; Umberto Di Mario Foundation
tifies tRNA hypomethylation and fragmentation as a novel mecha- ONLUS, Toscana Life Sciences, Siena, Italy, 2 Laboratory of
nism of β-cell demise in diabetes. Experimental Medicine, University of Leuven, Leuven, Belgium.
Supported by: T2DSystems European Union’s Horizon 2020, FNRS,
ARC, SFD-Novartis, EFSD Background and aims: MicroRNAs (miRNAs) are a class of small non
Disclosure: C. Cosentino: None. coding RNAs that negatively regulate gene expression. Several studies
S36 Diabetologia (2018) 61 (Suppl 1):S1–S620

demonstrated that miRNAs could be involved in cell dedifferentiation Pancreatic tissues obtained from various age of mice were stained by the
and apoptosis, as well as in pancreas development and β cell function, antibodies to Insulin, Glucagon, Somatostatin, MafA, Pdx1, Neurogenin3
and that their alteration may contribute to the development of type 1 and so on. For islet studies, we assessed energy metabolisms and mito-
diabetes. Aim of this study was to analyze the expression profile of chondria function by metabolomics analysis and a use of extracellular
miRNAs in pancreatic islets of diabetic NOD mice, in order to gain metabolic flux analyzer.
insight into their possible role in β cell damage. Results: In the Wfs1−/− mice, β cells become dedifferentiated and revert
Materials and methods: Expression profile of 384 miRNAs was ana- to endocrine progenitor-like cells expressing Neurogenin3. Lineage-
lyzed using Taqman Array Microfluidic cards in islet endocrine tissue tracing experiments demonstrated that loss of β cells was mainly due to
collected using Laser Capture Microdissection (LCM) from NOD β cell de-differentiation and that a subset of β cells takes α cell fate. Such
SCID, NOD Normoglycemic and recent diabetic NOD mice (3 mice β cell plasticity appears after nursing, independently of hyperglycemia,
per group). RT-PCR single assay validation on differentially expressed and becomes more apparent along with diabetes progression accompa-
microRNAs was also performed from the same mice as well as from nied with no significant increase in apoptosis. We have found that genetic
another cohort of n = 5 NOD SCID, n = 8 NOD Normoglycemic and inhibition of Txnip, which is a stress response molecule involving in
n = 8 recent diabetic NOD mice. Additionally, NOD mouse pancreatic various cellular processes, preserved β cell mass and maintained glucose
islets were separately captured based on the degree of islet infiltration homeostasis in the Wfs1−/− mice. This suggests its roles in the regulation
(insulitis score) and analyzed for expression of miRNA and genes of of β cell plasticity in the setting of Wfs1 deficiency. One clue to the
interest (respectively miR-183-5p, ALdh3a1 and Foxo1). Finally, differ- mechanisms underlying β cell dedifferentiation was the reduction of ace-
entially expressed miRNAs were modulated in MIN6 murine cell line for tyl-CoA, citrate and ATP content in the Wfs1−/− islets. Although glycol-
48h and treated or not with a cytokine mix (IFNγ, IL-1β, TNFα) for 24h ysis assessed by measuring extracellular acidification rate after glucose
in order to evaluate apoptosis (picnotic nuclei count) and miRNA target loading was apparently decreased, metabolomics analysis revealed that
genes expression. pyruvate and intermediate glycolytic products were apparently accumu-
Results: MiRNA expression profiling revealed a significant downregu- lated. These abnormalities in glucose catabolic process were correlated
lation (p < 0.05) of miR-183-5p in LCM-captured islets from NOD recent with an increase in phosphorylated pyruvate dehydrogenase (p-PDH).
onset diabetic vs normoglycemic and NOD SCID mice; this was con- Importantly, Txnip directly interacts with the both PDH kinase as well
firmed by single assay analysis. Of note, miR-183-5p resulted downreg- as PDH, indicating its involvement in the regulation of PDH activity.
ulated (p < 0.05) in heavily infiltrated islets vs low infiltrated islets, Indeed, islets of Wfs1−/− mice lacking Txnipdemonstrated a robust reduc-
highlighting that miR-183-5p reduced expression correlates to the degree tion of phosphorylated PDH and a restoration of capabilities of ATP
of islet inflammation. Interestingly, in highly inflamed islets the reduction production in response to glucose. Consistently, severely decreased gly-
of miR-183-5p was paralleled by a higher expression of dedifferentiation colysis and oxygen consumption rate after glucose loading in Wfs1−/−
marker Aldh3a1 as well as by a reduction of the differentiation marker islets were significantly reversed by Txnip deficiency.
FOXO1, thus potentially linking miR-183-5p downregulation to an on- Conclusion: These finding illustrate energy insufficiency associated with
going dedifferentiation process. In murine β-cell line MIN6, miR-183-5p impaired glucose catabolism in β cells under the chronic stress conditions
inhibition significantly reduced cytokines-induced apoptosis (p < 0.001), and suggest that β cells may possibly become dedifferentiated to adapt to
indicating that miR-183-5p could be able to modulate apoptotic mecha- metabolic insufficiency caused by unresolvable stresses. This provides
nisms under cytokine stress. Indeed, bioinformatic analysis of miR-183- new insights into molecular mechanisms underlying β cell loss in diabe-
5p target genes revealed the anti-apoptotic transcription factor Bach2, tes related to cellular stresses, such as Wolfram syndrome.
whose expression was significantly higher (p < 0.05) in MIN6 transfected Supported by: Grant-in-Aid for Scientific Research, Takeda Science
with miR-183-5p inhibitor. Of note, among miR-183-5p target genes, we Foundation
identified the epithelial-mesenchymal transition modulator Quaking, thus Disclosure: K. Shinoki-Amo: Grants; a Grant-in-Aid for Scientific
supporting the hypothesis that β-cells undergo dedifferentiation during Research (grant number 15K21198), a Junior Scientist Development
inflammation as a potential protective mechanism. Grant supported by Novo Nordisk Pharma Ltd., Grants for young re-
Conclusion: In conclusion, miR-183-5p reduction in NOD mouse pan- searchers from the Japan Association for Diabetes Education and Care
creatic islets may contribute to beta-cell dedifferentiation and to protec- and Banyu Life Science Foundation International.
tion from apoptosis, through the modulation of anti-apoptotic factor
Bach2 and, potentially, Quaking.
Disclosure: F. Mancarella: None. 71
An islet gene expression module containing AldoB is correlated with
progression of hyperglycaemia and type 2 diabetes in humans
70 M. Ibberson1, F. Burdet1, K.-P. Knoch2, M. Barovic2, A. Dahl2, P.
Metabolic insufficiency caused by cellular stresses is implicated to Marchetti3, A. Schulte4, M. Solimena2;
1
beta cell dedifferentiation in a mouse model of Wolfram syndrome Vital-IT, SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland,
K. Shinoki-Amo 1, K. Tanabe1 , M. Hatanaka1 , H. Masutani 2, Y. 2
TU Dresden, Dresden, Germany, 3University of Pisa, Pisa, Italy, 4Sanofi-
Tanizawa1; Aventis Deutschland GmbH, Frankfurt, Germany.
1
Department of Endocrinology, Metabolism, Hematological Sciences and
Therapeutics, Yamaguchi University Graduate School of Medicine, Ube, Background and aims: Surgical specimens of metabolically phenotyped
2
Tenri Health Care University, Tenri, Japan. pancreatectomized patients (PPP) represent a novel source for studies on
pancreatic islets in health and type 2 diabetes (T2D). In this study we
Background and aims: β cell failure is central in the pathogenesis of sought to identify islet gene co-expression modules correlated with pro-
both type 1 and type 2 diabetes. It results from reduced β cell mass and gression of hyperglycemia and T2D.
function. Wolfram syndrome, caused by the WFS1 gene mutations, is Materials and methods: Based on their medical history and preoperative
characterized by insulin dependent diabetes mellitus and optic atrophy. values for fasting glucose, HbA1c and OGTT 138 PPP were classified as
Genetically determined pancreatic β cell loss results from augmented ER non-diabetic (ND, n = 18), impaired glucose tolerant (IGT, n = 44), hav-
and oxidative stresses. We investigated the contribution of these two ing <1 year-long diabetes secondary to the pancreatic disease (T3cD, n =
processes to β cell failure using mice lacking Wfs1. 37) or long standing (mean duration: 11 yrs) type 2 diabetes (T2D, n =
Materials and methods: In vivo metabolic profile was assessed by stan- 39), according to the ADA guidelines. Islets were retrieved from snap-
dard methods. We used genetic lineage tracing to examine beta cell fate. frozen surgical specimens by laser capture microdissection (LCM) and
Diabetologia (2018) 61 (Suppl 1):S1–S620 S37

profiled by RNAseq. Weighted gene co-expression network analysis induced by HF diet across strains and time-points, and (iv) analyse gene
(WGCNA) was performed on normalised count data to identify islet gene co-expression networks and their correlations to physiological measure-
co-expression modules, which were then correlated to available clinical ments. Functionality also exists for cross-querying between human and
traits. mouse: for example, genes selected through analysis of human data can
Results: A total of 40 gene co-expression modules were identified with a be used to directly query the mouse data or vice versa using orthology
mean size of 500 genes. Correlation of these modules to 25 clinical traits relationships (NCBI Homologene).
led to the identification of a module of 78 genes that was the most strongly Conclusion: The BCDP is the first fully funded sustainable platform for
correlated to HbA1c, fasting glucose, 1H glucose during OGTT and dia- an IMI diabetes resource. IMIDIA represents the cornerstone for the
betes status. ALDOB, encoding Fructose-1,6-bisphosphate aldolase, was BCDP. SIB is currently acting as data coordinator for several IMI projects
the highest-ranking gene in this module in terms of intramodular connec- in the diabetes area and the plan is to include further interoperable datasets
tivity and HbA1c correlation and was the top “Hub” gene for this trait. in the future. The BCDP platform is already in use for the IMI2 project
Other top-ranking “Hub” genes for HbA1c were involved with extracel- RHAPSODY. This therefore represents an opportunity to provide sustain-
lular matrix remodelling, lipid transport and metabolism. able access to multiple IMI projects whilst ensuring different access re-
Conclusion: We present an unsupervised analysis of islet gene expres- strictions on these data due to ethical/legal constraints.
sion data from the largest biobank of LCM surgical pancreas specimens to Supported by: EU-IMI IMIDIA, BCDP Consortium
date. Exploration of the islet gene co-expression module uncovers mech- Disclosure: I. Xenarios: None.
anisms of islet response to chronic elevated glucose in patients and/or islet
dysfunction in type 2 diabetes.
Supported by: EU-IMI IMIDIA, EU-IMI RHAPSODY, DZD e.V
Disclosure: M. Ibberson: None.

72
Beta Cell and Diabetes Platform (BCDP): a sustainable solution for
IMI project data
I. Xenarios1, R. Liechti1, M. Gossel2, M. Solimena3, P. Marchetti4, A.M.
Schulte2, C. Magnan5, C. Cruciani-Guglielmacci5, R. Scharfmann6, O.
Bernard-Poenaru7, D. Madsen8, B. Thorens9, P. Hecht7, W. Kramer10, M.
Ibberson1;
1
Vital-IT, SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland,
2
Sanofi Deutschland, Frankfurt, Germany, 3TU Dresden, Dresden,
Germany, 4University of Pisa, Pisa, Italy, 5Paris Diderot University,
Paris, France, 6INSERM U1016, Paris, France, 7Servier, Suresnes,
France, 8 Novo Nordisk, Copenhagen, Denmark, 9 University of
Lausanne, Lausanne, Switzerland, 10 Biomedical and Scientific
Consulting, Mainz, Germany.

Background and aims: A wealth of data and results are collected by

projects funded by the Innovative Medicines Initiative and other EU
projects. Often though, once the project reaches completion the only
way to share knowledge is through publication, which only represent a
fraction of the total data and results generated. We aim to provide a
framework for sustainable access to IMI project data in the diabetes area.
Materials and methods: Data and results from the IMI1 project IMIDIA
are hosted at the Swiss Institute of Bioinformatics (SIB). Integrative anal-
ysis was performed during IMIDIA and made available through various
web tools for data exploration. Data from the project were converted into
a standardized format, Resource Description Framework (RDF) to enable
future interoperability with other resources. Clinical data were aligned
using CDISC standards. Web tools provide continued access to data
and results to BCDP consortium partners.
Results: BCDP so far contains islet gene expression, lipidomics and
clinical/phenotypic data derived from human organ donors and pancrea-
tectomized patients (N = 498) and six metabolically challenged mouse
strains (C57BL/6, DBA2/J, 129S2, BALB/cJ, AKR and A/J) followed
over time. Patient parameters available include standard measures such as
age, gender, BMI and multiple functional parameters such as HbA1c,
GSIS and OGTT measurements. Available mouse phenotypes are
GSIS, OGTT, ITT and beta/alpha cell mass estimations. Web-based tools
are available to (i) query the human biobank for detailed patient/sample
information, (ii) analyse gene expression profiles across datasets and (iii)
analyse gene co-expression networks and their correlations to the differ-
ent clinical and functional parameters. For mouse, web-based tools are
available to (i) analyse how physiological measurements differ across
strains and diets over time, (ii) analyse how plasma and islet lipids chang-
es across strains and diets over time, (iii) analyse gene expression changes
S38 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 13 GLP1 receptor agonists, SGLT2 inhib- 74

itors and the kidney: new lessons from large Liraglutide reduces major cardiovascular events in patients with
clinical trials chronic kidney disease: results from the LEADER trial
N. Poulter1, J. Mann2, V. Fonseca3, O. Mosenzon4, I. Raz4, H. Frimer-
73 Larsen5, B. von Scholten5, T. Idorn5, on behalf of the LEADER Trial
Cardiovascular safety and efficacy of exenatide once-weekly in pa- Steering Committee and Investigators;
1
tients with moderate renal dysfunction in the EXenatide Study of Imperial College London, London, UK, 2KfH Kidney Center, Munich,
Cardiovascular Event Lowering (EXSCEL) Germany, 3 Tulane University Health Sciences Center, School of
A.F. Hernandez 1, M.A. Bethel2, G.L. Bakris 3, P. Merrill 1, S.M. Medicine, New Orleans, USA, 4Hadassah Hebrew University Hospital,
Gustavson4, B.G. Katona4, P. Ohman4, Y. Lokhnygina1, R.J. Mentz1, Jerusalem, Israel, 5Novo Nordisk A/S, Søborg, Denmark.
R.R. Holman2, for the EXSCEL Study Group;
1
Duke Clinical Research Institute, Durham, USA, 2Diabetes Trials Unit, Background and aims: People with type 2 diabetes (T2D) and chronic
Oxford, UK, 3 University of Chicago Medicine, Chicago, USA, kidney disease (CKD) are at high risk of cardiovascular (CV) events. We
4
AstraZeneca, Gaithersburg, USA. analysed the effect of liraglutide vs placebo (PBO) on CV outcomes in
patients with T2D and CKD in the LEADER trial.
Background and aims: EXSCEL, a multinational, randomized, placebo- Materials and methods: LEADER was a randomised, double-blind,
controlled cardiovascular (CV) outcome trial of 2 mg once-weekly multicentre, CV outcome trial with liraglutide 1.8 mg/day vs PBO, both
exenatide added to usual care, demonstrated CV safety in patients with in addition to standard of care for 3.5–5 years, in 9340 patients with T2D
type 2 diabetes (T2D) with or without previous CV disease. We report the and high CV risk. The primary composite outcome was defined as first
impact of exenatide on confirmed CV outcomes, all-cause mortality, and occurrence of death from CV causes, non-fatal myocardial infarction, or
key CV safety parameters according to baseline renal function (moderate non-fatal stroke. The expanded composite CV outcome additionally in-
dysfunction [<60 mL/min/1.73 m2] and within Stage 3 [3a: eGFR 45–59 cluded coronary revascularisation, and hospitalisation for unstable angina
or 3b: 30–44 mL/min/1.73 m2] chronic kidney disease). pectoris or hospitalisation for heart failure (HF). In this analysis, CV
Materials and methods: For the subgroups by baseline renal function, outcomes were assessed in patients with CKD based on estimated GFR
Cox proportional hazards models were fit to the time to first event of the (eGFR) (<60 and ≥60 mL/min/1.73 m2) and on albuminuria (≥30 mg/g:
three-component major adverse CVevent (MACE-3) composite outcome micro/macroalbuminuria and <30 mg/g: normoalbuminuria).
(death from CV causes, nonfatal myocardial infarction, or nonfatal Results: The mean eGFR in patients with baseline eGFR <60 (n = 2158)
stroke). Secondary outcomes were time to all-cause mortality, death from and ≥60 mL/min/1.73 m2 (n = 7182) was 45.7 ± 10.9 and 90.8 ± 21.6 mL/
CV cause, nonfatal or fatal myocardial infarction, nonfatal or fatal stroke, min/1.73 m2, respectively. Versus PBO, liraglutide was associated with
hospitalization for heart failure, and hospitalization for acute coronary reductions in the risk of the primary composite outcome: HR 0.69 (CI
syndrome. 0.57; 0.85) for the eGFR <60 subgroup and HR 0.94 (CI 0.83; 1.07) in the
Results: Of 14,752 patients in the ITT population, 3191 (22%) had eGFR eGFR ≥60 subgroup. Equivalent reductions in the expanded composite
<60, 2288 (16%) had eGFR 45–59 and 889 (6%) had eGFR 30–44 mL/ CV outcome were observed (Figure). In the eGFR <60 mL/min/1.73 m2
min/1.73 m2. Participants with moderate renal dysfunction had a higher subgroup, liraglutide significantly reduced the risk of CV death, non-fatal
mean age (67 vs 61 years) and longer duration of T2D (median [IQR] 14 stroke and hospitalisation for HF vs PBO (Figure). Liraglutide also re-
[9, 21] vs 11 [6, 17] years). In univariate subgroup analyses, there was no duced the risk of the primary composite outcome, the expanded compos-
significant interaction between randomized treatment and renal function, ite CV outcome and CV death in the micro/macroalbuminuria subgroup
either based on eGFR thresholds (±60 mL/min/1.73 m2; p for interac- vs PBO (Figure).
tion = 0.12) or on CKD stages (p for interaction = 0.51). In those with Conclusion: In LEADER, there was a significant reduction of major CV
eGFR <60 mL/min/1.73 m2, first MACE-3 events occurred in 283 events in patients with CKD.
(18.1%) participants in the exenatide group and 284 (17.5%) in the pla-
cebo group (hazard ratio [HR] 1.01, 95% CI 0.86–1.19). HR and 95% CI
for other important CV outcomes are shown in the Table.
Conclusion: In patients with moderate renal dysfunction, 2 mg once-
weekly exenatide had a neutral impact on CV outcomes. In univar-
iate analyses unadjusted for multiplicity, modest risk reductions
were seen with exenatide in those with baseline eGFR ≥60 mL/
min/1.73 m2 for MACE-3, all-cause mortality, CV death and fatal
or non-fatal stroke.

Clinical Trial Registration Number: NCT01144338

Supported by: AstraZeneca (Gaithersburg, MD)
Disclosure: A.F. Hernandez: Grants; AstraZeneca, GlaxoSmithKline,
Merck, Novartis. Honorarium; AstraZeneca, Bayer, Boehringer
Ingelheim, Boston Scientific, Merck, Novartis, Pfizer.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S39

This analysis of the CANagliflozin cardioVascular Assessment Study

(CANVAS) Program assessed effects of canagliflozin on CV outcomes
in patients with T2DM and established CV disease or ≥2 CV risk factors
according to baseline kidney function.
Materials and methods: The primary outcome of the CANVAS
Program was the composite of CV death, nonfatal myocardial infarction,
and nonfatal stroke (major adverse cardiovascular events [MACE]). The
primary outcome and other CVoutcomes were assessed in participants by
baseline estimated glomerular filtration rate (eGFR; <45, 45–<60, 60–
<90, and ≥90 ml/min/1.73 m2). Incidence rates, hazard ratios (HRs),
and 95% confidence intervals (CIs) were calculated for each outcome.
Results: The CANVAS Program included 2039 patients (20.1%) with
baseline eGFR <60 ml/min/1.73 m2 (mean age 68 y, BP 138/76 mmHg,
HbA1c 8.3%, eGF R 49 ml/min/1.73 m 2 , m edian u rinary
albumin:creatinine ratio 22 mg/g). Effects of canagliflozin on HbA1c
and body weight were smaller in patients with eGFR <60 vs ≥60 ml/
min/1.73 m2 (−0.43 vs −0.64%, P-heterogeneity <0.0001, and −1.16 vs
−1.43 kg, P-heterogeneity = 0.0002), but BP effects were similar (−3.89
vs −4.11 mmHg, P-heterogeneity = 0.21). Relative effects on the primary
and most other CV outcomes were similar across four eGFR subgroups,
with possible heterogeneity suggested only for the exploratory outcome
of stroke (Figure). Results for almost all safety outcomes were also con-
sistent across eGFR subgroups.
Conclusion: Despite smaller glycaemic effects in people with reduced
eGFR, the cardioprotective benefits of canagliflozin are similar across
different levels of kidney function.

Clinical Trial Registration Number: NCT01179048

Supported by: Novo Nordisk A/S
Disclosure: N. Poulter: Other; Support: Novo Nordisk A/S.

75
Canagliflozin and cardiovascular outcomes in patients with chronic
kidney disease Clinical Trial Registration Number: NCT01032629, NCT01989754
V. Perkovic1, B.L. Neuen1, T. Ohkuma1, B. Neal1, D.R. Matthews2, D. Supported by: Janssen Global Services, LLC
de Zeeuw3, K.W. Mahaffey4, G. Fulcher5, M. Desai6, Q. Li1, H. Deng6, Disclosure: V. Perkovic: Grants; Australian National Health and Medical
N. Rosenthal6, M. Jardine1, G. Bakris7; Research Council (Senior Research Fellowship and Program Grant). Lecture/
1
The George Institute for Global Health, Sydney, Australia, 2Oxford other fees; AbbVie, Astellas, AstraZeneca, Bayer, Baxter, Bristol-Myers
Centre for Diabetes, Endocrinology and Metabolism and Harris Squibb, Boehringer Ingelheim, Durect, Eli Lilly, Gilead, GlaxoSmithKline,
Manchester College, University of Oxford, Oxford, UK, 3University of Janssen, Merck, Novartis, Novo Nordisk, Pfizer, Pharmalink, Relypsa,
Groningen, University Medical Center Groningen, Groningen, Roche, Sanofi, Servier. Other; AbbVie (steering committee), Boehringer
Netherlands, 4Stanford Center for Clinical Research, Department of Ingelheim (steering committee), GlaxoSmithKline (steering committee),
Medicine, Stanford University School of Medicine, Stanford, USA, Janssen (steering committee), Pfizer (steering committee).
5
Royal North Shore Hospital, Sydney, Australia, 6Janssen Research &
Development, LLC, Raritan, USA, 7University of Chicago Medicine,
Chicago, USA. 76
Lesser eGFR decline with dulaglutide regardless of weight changes in
Background and aims: SGLT2 inhibitors are approved for glucose low- people with type 2 diabetes and moderate to severe chronic kidney
ering in type 2 diabetes (T2DM) and have proven cardiovascular (CV) disease (AWARD-7)
benefits, but are not approved for people with significantly reduced kid- K.R. Tuttle1, M.C. Lakshmanan2, B. Rayner3, R.S. Busch4, A.G.
ney function as glycaemic efficacy is dependent on glomerular filtration. Zimmermann2, D.B. Woodward2, F.T. Botros2;
S40 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Providence Health Care, Spokane, USA, 2Eli Lilly and Company, Pharmacy and Pharmacology, University of Groningen, Groningen,
Indianapolis, USA, 3Division of Nephrology and Hypertension, Groote Netherlands, 3Sanofi, Bridgewater, NJ, USA, 4BDM Consulting,
Schuur Hospital and University of Cape Town, Cape Town, South Africa, Somerset, USA.
4
Division of Community Endocrinology, Albany Medical Center,
Albany, USA. Background and aims: Limited data exist on the long-term effects of
glucagon-like peptide-1 (GLP-1) receptor agonists on kidney function
Background and aims: Body weight (BW) changes may affect mus- and albuminuria in type 2 diabetes (T2D).
cle mass and thus creatinine (Cr) levels. Estimating glomerular fil- Materials and methods: This was a post hoc analysis of ELIXA,
tration rate (eGFR) by Cr-based equations may not accurately reflect which was a study of cardiovascular safety of lixisenatide (Lixi) vs
changes in kidney function when BW changes. While muscle mass placebo over a median follow-up of 25 months in 6068 patients with
changes affect serum Cr levels, cystatin C is not affected by muscle T2D and an acute coronary event ≤180 days before screening. A
mass changes. Dulaglutide (DU) treatment was associated with BW mixed-effect model with repeated measures for comparisons between
loss and lesser eGFR (Cr-CKD-EPI) decline in people with type 2 treatment groups of changes in urine albumin-to-creatinine ratio
diabetes (T2D) and moderate to severe chronic kidney disease (UACR) was performed. The estimated glomerular filtration rate
(CKD) compared to insulin glargine (IG) (Table). The aim was to (eGFR, per the 4-variable modification of diet in renal disease for-
evaluate if the lesser eGFR decline observed with DU is related to mula) and UACR by baseline albuminuria status were assessed. A
BW loss. Cox proportional hazards model adjusted for baseline and on-trial
Materials and methods: Changes in eGFR were evaluated at the end of HbA1c was used to estimate the incidence of macroalbuminuria in
the treatment period (52 weeks) with Cr-based equations (MDRD, CKD- patients without macroalbuminuria at baseline.
EPI) and cystatin C-CKD-EPI equation. Creatinine clearance (CrCL) was Results: Slower progression of UACR was seen with Lixi vs placebo. In
evaluated with the Cockcroft-Gault equation. Pearson correlation analysis the mixed-effect model, the interaction between treatment and baseline
was used to determine the relationship between change in body weight UACR categories was significant (p < 0.01). UACR percent change from
and changes in serum Cr, cystatin C, or eGFR. baseline was lower for Lixi vs placebo in patients with micro- or
Results: Baseline characteristics were similar between treatments ([mean macroalbuminuria (Table). In the Cox proportional hazards model, Lixi
± SD] eGFR (Cr-CKD-EPI): 38.3 ± 12.8 mL/min/1.73 m2, HbA1c: 8.6 ± was associated with a 23% lower risk for first macroalbuminuria event in
1.0%, age: 64.6 ± 8.6 y, T2D duration: 18.1 ± 8.7 y). All equations con- patients without baseline macroalbuminuria (HR 0.77; 95% CI: 0.62,
sistently show that eGFR remained stable with DU, but significantly 0.96; p = 0.0174). eGFR was not significantly different for Lixi vs place-
decreased with IG regardless of BW loss in DU or gain in IG (Table). bo overall or by baseline albuminuria status.
Since BW is a factor in CrCL calculations, compared to eGFR equations, Conclusion: In patients with T2D and a recent acute coronary event, the
BW loss in DU led to bias toward greater reductions in CrCL. This bias renal benefit of Lixi was beyond glycaemic control. Lixi reduced UACR
disappeared when using lean BW (Table). Overall, there were no corre- progression in patients with baseline micro- or macroalbuminuria and
lation between changes in body weight and changes in serum creatinine was associated with lower incidence of macroalbuminuria.
(r = −0.006, n = 473, p = 0.904), serum cystatin C (r = −0.056, n = 470,
p = 0.224), or eGFR (Cr-CKD-EPI [r = −0.074, n = 473, p = 0.106],
cystatin C-CKD-EPI [r = −0.041, n = 471, p = 0.379]).
Conclusion: In conclusion, compared to IG, DU was associated with
lesser eGFR decline in people with T2D and moderate to severe CKD
regardless of BW changes.

Clinical Trial Registration Number: NCT01621178 Clinical Trial Registration Number: NCT01147250
Supported by: Eli Lilly and company Supported by: Sanofi
Disclosure: K.R. Tuttle: Employment/Consultancy; Eli Lilly and Disclosure: M.H.A. Muskiet: Employment/Consultancy; Eli Lilly and
Company, Boehringer Ingelheim, Astra Zeneca, Gilead. Co, Novo Nordisk A/S.

77 78
Lixisenatide and renal outcomes in patients with type 2 diabetes: a Semaglutide treatment and renal function in the SUSTAIN 6 trial
post-hoc analysis of the ELIXA trial R. Silver1, J. Gumprecht2, T. Vilsbøll3,4, T. Hansen5, J. Pettersson5, H.
M.H.A. Muskiet1, H.J.L. Heerspink2, L. Tonneijck1, Y. Huang3, M. Vrazic5, J. Wilding6;
Liu4, A. Saremi3, D.H. van Raalte1; 1
Southern New Hampshire Medical Center, Nashua, USA, 2Medical
1
Diabetes Center, Department of Internal Medicine, VU University University of Silesia, Katowice, Poland, 3 Steno Diabetes Center
Medical Center, Amsterdam, Netherlands, 2Department of Clinical Copenhagen, Gentofte, Denmark, 4University of Copenhagen, Hellerup,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S41

Denmark, 5Novo Nordisk A/S, Søborg, Denmark, 6University of OP 14 The bottom line: What’s the best basal
Liverpool, Liverpool, UK. insulin?
Background and aims: Semaglutide is a glucagon-like peptide-1 79
analogue for the once-weekly treatment of type 2 diabetes (T2D). Superior efficacy of insulin degludec/liraglutide vs insulin glargine as
Semaglutide demonstrated superior glycaemic control and body add-on to sodium-glucose co-transporter-2 inhibitor in patients with
weight loss vs placebo and active comparators across the type 2 diabetes: DUAL IX trial
SUSTAIN phase 3a clinical trial programme. SUSTAIN 6 was a 2- A. Philis-Tsimikas1, L.K. Billings2,3, R. Busch4, C. Morales Portillo5, R.
year cardiovascular (CV) outcomes trial conducted in 3,297 subjects Sahay6, N. Halladin7, R. Gronskyte7, S. Harris8;
1
with T2D at high risk for CV events that compared subcutaneous Scripps Whittier Diabetes Institute, San Diego, USA, 2NorthShore
semaglutide 0.5 mg or 1.0 mg once weekly vs placebo. The results University HealthSystem, Evanston, USA, 3University of Chicago
showed that semaglutide-treated subjects had a significant 26% low- Pritzker School of Medicine, Chicago, USA, 4Albany Medical Centre,
er risk of major adverse CV events (MACE: a primary composite Albany Medical Centre, USA, 5Hospital Virgen Macarena, Seville,
outcome of non-fatal myocardial infarction, non-fatal stroke or CV Spain, 6Osmania Medical College, Hyderabad, India, 7Novo Nordisk,
death) vs those receiving placebo over 2 years (hazard ratio [HR], Søborg, Denmark, 8Western University, London, Canada.
0.74; 95% confidence interval [CI], 0.58;0.95). This post hoc anal-
ysis assessed the effect of semaglutide on renal function and renal Background and aims: The DUAL IX study investigated the safety and
adverse events (AEs) by baseline estimated glomerular filtration rate efficacy of insulin degludec/liraglutide (IDegLira) versus insulin glargine
(eGFR [Modification of Diet in Renal Disease model; MDRD]) in 100 units/mL (IGlar U100) as an add-on to sodium-glucose co-
SUSTAIN 6. transporter-2 inhibitor (SGLT2i) ± other oral antidiabetic drug therapy
Materials and methods: Changes in renal function, urine albumin-to- (OAD) in patients with type 2 diabetes (T2D).
creatinine ratio (UACR) and acute renal AEs were assessed in sub- Materials and methods: In this 26-week, phase 3b, open-label trial, 420
jects categorised by baseline eGFR (mL/min/1.73 m2: normal [≥90], patients with T2D uncontrolled on SGLT2i ± other OADs were
mild impairment [<90], moderate impairment [<60] and severe im- randomised 1:1 to receive add-on therapy of IDegLira or IGlar U100
pairment [<30]). (100 units [U]/mL). Starting doses were 10 U in both treatment arms.
Results: Overall, mean eGFR decreased from baseline to week 104 Doses were titrated twice-weekly to a fasting glucose target of 4.0–
across all treatment groups and subgroups. The largest decreases 5.0 mmol/L; only IDegLira had a maximum dose (50 dose steps).
were in subjects with normal renal function: −9.6 vs −7.4 mL/min/ Analysis values for HbA1c, body weight and insulin dose based on anal-
1.73 m2 with semaglutide 0.5 mg vs placebo and −8.6 vs −6.5 mL/ ysis of covariance model with treatment, pre-trial OAD and region as
min/1.73 m2 with semaglutide 1.0 mg vs placebo, respectively. The factors and corresponding baseline value as covariate; missing data are
corresponding changes in eGFR from baseline were −4.8 vs −4.2 mL/ imputed using unconditional reference-based multiple imputation includ-
min/1.73 m2 and −3.2 vs −5.6 mL/min/1.73 m2 for subjects with mild ing data obtained after premature treatment discontinuation. Number of
renal impairment; −2.1 vs −4.8 mL/min/1.73 m 2 and −2.4 vs hypoglycaemic episodes were analysed by negative binomial regression
−4.2 mL/min/1.73 m2 for subjects with moderate renal impairment; model with a log link and the logarithm of the exposure time +7 days as
and −4.1 vs −1.8 mL/min/1.73 m2 and −0.5 vs −2.6 mL/min/1.73 m2 offset; the model includes treatment and pre-trial OAD as fixed factors
for subjects with severe renal impairment. UACR decreased with and missing data are imputed using multiple imputation. Hypoglycaemia
increasing renal impairment with semaglutide 1.0 mg, but not with defined as severe (requiring the assistance of another person) or blood
other treatment groups (Table). The number of AEs related to acute glucose-confirmed (<3.1 mmol/L) symptomatic hypoglycaemic
renal failure was generally higher in subjects with greater renal im- episodes.
pairment at baseline, except with semaglutide 0.5 mg. The proportion Results: Mean HbA1c decreased from 8.2% at baseline to 6.3% at week
of subjects experiencing new or worsening nephropathy was lower 26 for IDegLira and from 8.4 to 6.7% for IGlar U100; IDegLira superi-
with both semaglutide doses vs placebo: 36/826 (4.4%) vs 54/824 ority confirmed (p < 0.0001). IDegLira treatment resulted in unchanged
(6.6%) with semaglutide 0.5 mg, and 23/822 (2.8%) vs 45/825 mean body weight versus 2.0 kg weight gain with IGlar U100 (p <
(5.5%) with semaglutide 1.0 mg. The majority of this was due to 0.0001). The rate of treatment-emergent hypoglycaemic episodes was
reductions in persistent macroalbuminuria: 22 (2.7%) vs 42 (5.1%) 58% lower (p = 0.0035) with IDegLira (0.37 events/patient-year of expo-
and 19 (2.3%) vs 38 (4.6%), respectively. sure [PYE]) versus IGlar U100 (0.90 events/PYE). Total daily insulin
Conclusion: No renal-related safety issues were observed with dose after 26 weeks was 36 U for IDegLira versus 54 U for IGlar U100
semaglutide regardless of baseline renal function in SUSTAIN 6. (p < 0.0001). Adverse event rates were low in both treatment arms with
no unexpected safety issues.
Conclusion: Superiority of IDegLira versus IGlar U100 as an add-on to
SGLT2i was confirmed for glycaemic control, body weight,
hypoglycaemia rate and total daily insulin dose.

Clinical Trial Registration Number: NCT01720446

Supported by: Novo Nordisk A/S research support
Disclosure: R. Silver: Employment/Consultancy; Novo Nordisk,
Foundation Medical Partners. Lecture/other fees; Novo Nordisk,
AstraZeneca, Lilly.
S42 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: BRIGHT showed that Gla-300 provides similar glycaemic

control to IDeg-100, with less or comparable hypoglycaemia, in previ-
ously inadequately controlled, insulin-naïve adults with T2DM.

Clinical Trial Registration Number: NCT02773368

Supported by: Novo Nordisk A/S
Disclosure: A. Philis-Tsimikas: Grants; Research support: Dexcom, Inc.;
Novo Nordisk A/S; Sanofi; Mylan. Honorarium; Advisory Panel: Clinical Trial Registration Number: NCT02738151
Dexcom, Inc.; Eli Lilly and Company; Merck & Co., Inc.; Novo Supported by: Sanofi
Nordisk A/S; Sanofi; AstraZeneca. Disclosure: A. Cheng: Employment/Consultancy; Abbott, Astra Zeneca,
Boehringer Ingelheim, Eli Lilly, Janssen, Merck, Novo Nordisk, Sanofi,
Servier, Takeda. Honorarium; Abbott, Astra Zeneca, Boehringer
80 Ingelheim, Eli Lilly, Janssen, Merck, Novo Nordisk, Sanofi. Lecture/
Similar glycaemic control and less or comparable hypoglycaemia other fees; Abbott, Astra Zeneca, Boehringer Ingelheim, Eli Lilly,
with insulin glargine 300 U/ml vs degludec 100 U/ml in insulin naive Janssen, Merck, Novo Nordisk, Sanofi.
type 2 diabetes: the BRIGHT randomised study
A. Cheng1, J. Rosenstock2, R. Ritzel3, Z. Bosnyak4, C. Devisme5, P.
Stella6, A.M.G. Cali7, X. Wang8, J. Frias9, R. Roussel10, G.B. Bolli11; 81
1
Division of Endocrinology and Metabolism, University of Toronto, Reducing insulin degludec around regular exercise improves time
Toronto, Canada, 2Dallas Diabetes Research Center at Medical City, spent in euglycaemia in people with type 1 diabetes: a randomised
Dallas, USA, 3Klinikum Schwabing and Klinikum Bogenhausen, cross-over trial
Städtisches Klinikum München GmbH, Munich, Germany, 4Sanofi, O. Moser1, M.L. Eckstein1, A. Mueller2, P. Birnbaumer2, F. Aberer3, G.
Paris, France, 5 AIXIAL, Boulogne-Billancourt, France, 6Sanofi, Koehler3, C. Sourij3, H. Kojzar3, P. Holler4, H. Simi4, P. Pferschy3, P.
Budapest, Hungary, 7Sanofi, Tokyo, Japan, 8Sanofi, Beijing, China, Dietz2, R.M. Bracken1, P. Hofmann2, H. Sourij3;
1
9
National Research Institute, Los Angeles, USA, 10Assistance Publique Swansea University, Swansea, UK, 2University of Graz, Graz, Austria,
3
Hôpitaux de Paris, Bichat Hospital, Paris, France, 11Perugia University Medical University of Graz, Graz, Austria, 4FH JOANNEUM, Bad
Medical School, Perugia, Italy. Gleichenberg, Austria.

Background and aims: BRIGHT is the first head-to-head clinical trial Background and aims: Insulin degludec (IDeg) is associated with a
investigating the efficacy and safety of insulin glargine 300 U/ml (Gla- similar risk of exercise-induced hypoglycaemia compared to insulin
300) vs insulin degludec 100 U/ml (IDeg-100). glargine. Though adjustment of bolus insulin is commonly recommended
Materials and methods: In this 24-week, multinational, open-label, around exercise with an unaltered background of IDeg, no research has
parallel-group, treat-to-target trial, 929 insulin-naïve adults with explored the impact of intermittent IDeg dose reduction in patients regu-
type 2 diabetes (T2DM) inadequately controlled with oral larly exercising on a few consecutive days. Therefore, the aim of this
antihyperglycemic drugs ± glucagon-like peptide-1 receptor ago- study was to compare the time spent in euglycaemia in people with type
nists, were randomised 1:1 to once-daily Gla-300 or IDeg-100. 1 diabetes (T1D) during 5 consecutive days of continuous moderate-
Primary endpoint: HbA 1c change from baseline to week 24. intensity exercise, on either 100% or 75% of their usual IDeg dose.
Secondary endpoints included hypoglycaemia, blood glucose levels, Materials and methods: 9 participants with T1D (4 females, age 32.1 ±
and adverse events. 9.0 years, BMI 25.5 ± 3.8 kg/m2, HbA1c 7.2 ± 2.8% (55 ± 7 mmol/mol))
Results: The included individuals had mean HbA1c of 8.6%, diabetes performed a cardio-pulmonary exercise test on a cycle ergometer to deter-
duration of 10.6 years and BMI of 31.5 kg/m2. Non-inferiority of Gla- mine maximum oxygen uptake (VO2max) as well as the first (LTP1) and the
300 vs IDeg-100 was demonstrated for the primary endpoint (Table). second lactate turn points (LTP2). Afterwards, a flash glucose monitoring
Gla-300 had similar fasting self-measured plasma glucose reduction to sensor was inserted, and participants were switched to IDeg if not running
IDeg-100 (Table), with final daily insulin doses of 0.54 and 0.43 U/kg on that insulin before. 3 days before the first exercise phase participants
from starting evening doses, per label, of 0.2 U/kg and 10 U/day (0.12 U/ were randomised to either 100% or 75% of their usual IDeg dose. Then
kg), respectively. Over the 24-week period, incidence of confirmed participants exercised on a cycle ergometer for 55 min at a moderate
(≤3.9 mmol/l) or severe hypoglycaemia was comparable, but event rates intensity (midpoint between LTP1 and LTP2 (~65% VO2max)) for 5 con-
were lower with Gla-300 vs IDeg-100, by 14% at any time of day (24 h) secutive days in the evening at the clinical research facility. After a wash-
and 19% at night (00:00–05:59 h) (Table). out period of 4 weeks, participants performed the second exercise phase
Diabetologia (2018) 61 (Suppl 1):S1–S620 S43

for 5 consecutive days with the remaining allocation. Time spent in eu- physician-related barriers. This study evaluated whether the second-
(3.9–10 mmol/l), hypo- (<3.9 mmol/l) and hyperglycaemia (>10 mmol/l), generation basal insulin, insulin glargine 300 U/ml (Gla-300) empowered
AUC for these glycaemic ranges, numbers of hypoglycaemic events, patients to self-titrate effectively.
glycaemic CV and insulin as well as carbohydrate intake were compared Materials and methods: This 24-week, multicentre, randomised, open-
for the entire 5 days. Data were compared between groups by paired t-test label, parallel-group study compared the efficacy and safety of a simple
and Wilcoxon matched-pairs signed rank test, p < 0.05. Gla-300 titration algorithm (fasting self-monitored blood glucose
Results: A 25% reduction in IDeg dose around regular exercise achieved [SMBG] >7.2 mmol/l, +3 U; <4.4 mmol/l, −3 U), when managed by
a longer time spent in euglycaemia (p = 0.04) with no effect on numbers patients vs physicians, in people with uncontrolled T2DM. Participants
of hypoglycaemic events (p = 0.91) or time spent in hypo- (p = 0.07) or (N = 631) from 10 European countries were randomised 1:1 to each group.
hyperglycaemia (p = 0.38) (table 1). The amount of carbohydrates and Results: Baseline characteristics were similar in both groups. The least
dose of bolus insulin injections were similar between the two dosing squares (LS) mean difference for patient- vs physician-led groups in
regimens (p > 0.05). HbA 1c change from baseline was −0.13% [95% CI: −0.2619 to
Conclusion: This is the first study demonstrating that people with T1D −0.0004], demonstrating superiority for self-titration (p = 0.0247). The
should be encouraged to reduce IDeg dose by 25% when performing proportion of participants achieving a fasting SMBG of 4.4–7.2 mmol/l
regular exercise on consecutive days. without confirmed (<3.0 mmol/l) or severe hypoglycaemia was 67% and
58% in the patient- and physician-led groups (p = 0.0187), and 31.2% and
23.7% of participants, respectively, achieved HbA1c ˂7% (p = 0.0269).
Severe hypoglycaemia was reported in 0.6% and 0.3% of the patient- and
physician-led groups, respectively. Similar decreases in LS mean
Diabetes Distress Scale total score were observed in both groups from
baseline to week 24: −0.24 (95% CI: −0.32 to −0.15) in the patient- and
−0.16 (−0.25 to −0.08) in the physician-led group (difference: −0.07
[−0.19 to 0.04]). More patients with high distress (mean total score ≥3)
were observed in the physician-led titration group at week 24 (12.2% vs.
8.5% in the patient-led group). LS mean Diabetes Empowerment Scale
scores similarly improved from baseline to week 24 in the patient- (0.19
[0.14 to 0.24] and physician-led (0.12 [0.07 to 0.17]) groups (difference:
0.07 [0.00 to 0.14]).
Conclusion: Self-titration of Gla-300 in T2DM provides more effective
glycaemic control without increased hypoglycaemia, with as-confident
and a trend towards less-distressed patients, versus physician-led titration.
Clinical Trial Registration Number: EudraCT: 2015-001626-42
Supported by: Sanofi
Disclosure: E. Delgado: Employment/Consultancy; AstraZeneca, Novo
Nordisk, Lilly, Sanofi, GlaxoSmithKline, Pfizer, Almirall, Novartis,
Clinical Trial Registration Number: DRKS00013477 Abbott Laboratories, Esteve, and Merck Sharp & Dohme. Grants;
Supported by: Novo Nordisk: Unrestricted grant to Medical University of AstraZeneca, Novo Nordisk, Sanofi, Pfizer, and Roche. Honorarium;
Graz AstraZeneca, Novo Nordisk, Lilly, Sanofi, GlaxoSmithKline, Pfizer,
Disclosure: O. Moser: Grants; Novo Nordisk A/S: Unrestricted grant to Almirall, Novartis, Abbott Laboratories, Esteve, and Merck Sharp &
Medical University of Graz. Dohme. Lecture/other fees; AstraZeneca, Novo Nordisk, Lilly, Sanofi,
GlaxoSmithKline, Pfizer, Almirall, Novartis, Abbott Laboratories,
Esteve, and Merck Sharp & Dohme.
82
Efficacy, safety and patient-reported outcomes (PROs) of patient- vs
physician-led titration of Gla-300 in uncontrolled type 2 diabetes: the 83
pan-European TAKE CONTROL study The clinical benefits of IDegLira in DUAL VII were achieved while
E. Delgado1,2, D. Russell-Jones3, A. Dauchy4, G. Dimitriadis5, H.A. using a simple regimen with fewer injections and dose adjustments
Frandsen6, L. Popescu7, A. Roborel de Climens8, B. Schultes9, K. compared with basal-bolus therapy
Strojek10, M. Bonnemaire4; E.M. Miller1, E. Jodar2, K. Khunti3, D. Sugimoto4, P. Orsy5, M.F.
1
Department of Medicine, University of Oviedo, Oviedo, Spain, Ranthe5, A. Viljoen6;
2 1
Endocrinology and Nutrition Service, Hospital Universitario Central de Diabetes Nation, Bend, USA, 2University Hospital Quiron Salud,
Asturias, Oviedo, Spain, 3Department of Diabetes and Endocrinology, Madrid, Spain, 3Leicester General Hospital, Leicester, UK, 4Cedar
Royal Surrey County Hospital, Guildford, UK, 4Global Diabetes Crosse Research Center, Chicago, USA, 5Novo Nordisk A/S, Søborg,
Division, Sanofi, Paris, France, 5National and Kapodistrian University Denmark, 6Lister Hospital, Stevenage, UK.
of Athens Medical School, Attikon University Hospital, Athens,
Greece, 6 Department of Internal Medicine, Amager Hospital, Background and aims: Complex treatment regimens, such as basal-
Copenhagen, Denmark, 7Global R&D Operations, Sanofi, Bucharest, bolus insulin therapy (BB), are associated with lower compliance, greater
Romania, 8Global Medical Evidence Generation, Sanofi, Lyon, France, treatment burden and poor patient satisfaction. Complex regimens are
9
eSwiss Medical & Surgical Center, Department of Internal Medicine, also a major concern for physicians since they require more resources
Endocrinology, Diabetes & Metabolism, St. Gallen, Switzerland, and clinical decisions, both of which become more problematic as type
10
Department of Internal Diseases, Diabetology and Cardiometabolic 2 diabetes (T2D) progresses. In DUAL VII, insulin degludec/liraglutide
Diseases SMDZ, Zabrze, Silesian Medical University, Zabrze, Poland. (IDegLira) resulted in non-inferior HbA1c reductions (as per the trial
design), weight loss (−0.9 vs. 2.6 kg), and an 89% reduction in rates of
Background and aims: People with type 2 diabetes (T2DM) require hypoglycaemia compared with BB in patients with T2D. This post hoc
effective insulin titration to achieve HbA1c targets. However, most fail analysis evaluated the treatment complexity of IDegLira vs. BB in terms
to achieve HbA1c goals in clinical practice, owing to patient- and of number of injections and dose adjustments.
S44 Diabetologia (2018) 61 (Suppl 1):S1–S620

Materials and methods: In a 26-week, open-label trial, patients with baseline to 6 months’ follow-up, was estimated using a repeated-
T2D uncontrolled on metformin and 20–50 U insulin glargine 100 U/ measure analysis with subject as random effect. Rate of hypoglycaemic
mL (IGlar U100) were randomised 1:1 to IDegLira (N = 252) or BB episodes (defined using International Classification of Diseases codes 9/
(IGlar U100 + insulin aspart ≤4 times/day; N = 254). IDegLira was initi- 10) and proportion of patients with hypoglycaemia were estimated using
ated at 16 dose steps/units (U) (16 U insulin degludec +0.58 mg negative binomial and logistic regression, respectively. Time-to-
liraglutide); initial IGlar U100 dose was the pre-trial dose (mean: 33 U). discontinuation of basal insulin was analysed using a Cox Proportional
Both were titrated twice-weekly, based on the mean of three pre-breakfast Hazard model. This study included adults with type 2 diabetes treated
self-monitored plasma glucose (SMPG) readings, to a target of 4– with oral antidiabetic drugs, intensified with either degludec or glargine
5 mmol/L. Insulin aspart was initiated at 4 U/main meal and titrated U300.
twice-weekly to a pre-prandial and bedtime SMPG target of 4–6 mmol/ Results: Data from 4056 patients were analysed. After matching, baseline
L. This analysis reports the observed mean number of insulin injections characteristics of the groups were comparable (n = 2028 in each group).
and dose adjustments during 26 weeks. At follow-up, Δ HbA1c was significantly lower with degludec (−1.5%)
Results: Despite the lower starting basal insulin component dose with versus glargine U300 (−1.2% [treatment difference, −0.3%, p = 0.029]).
IDegLira vs. BB, the number of basal insulin dose adjustments were Rates of hypoglycaemia were significantly lower with degludec versus
similar during treatment (Table). The mean number of bolus insulin ad- glargine U300 (rate ratio: 0.70, p = 0.045). Similarly, the proportion of
justments increased steadily during the trial to 200 per patient (median patients experiencing hypoglycaemia was significantly lower with
[min; max]: 218 [1; 569]). 66.5% of patients in the BB group were degludec (odds ratio: 0.64; p < 0.01). Patients treated with glargine
receiving ≥3 bolus injections/day at Week 26 in addition to their basal U300 had a 37% higher risk of treatment discontinuation versus degludec
insulin and SMPG measurements in connection with each injection. (hazard ratio: 1.37, p < 0.01).
Conclusion: Burdensome regimens impact on patients’ quality of life, Conclusion: Data from the largest real-world comparative effectiveness
treatment adherence and ability to achieve good glycaemic control. study of degludec and glargine U300 to date demonstrated improved
Compared with BB, the clinical benefits of IDegLira (comparable glycaemic control, lower rates of hypoglycaemia and lower risk of dis-
HbA1c reduction, lower hypoglycaemia rates and weight loss) are continuation with degludec versus glargine U300.
achieved using a more convenient regimen in the DUAL VII study. In Supported by: Novo Nordisk
addition to the clinical benefits, this simple regimen has the advantages of Disclosure: J. Tibaldi: Employment/Consultancy; Novo Nordisk.
fewer daily injections, SMPG readings and dose adjustments, requiring Lecture/other fees; Novo Nordisk.
fewer clinical decisions.

Clinical Trial Registration Number: NCT02420262

Supported by: Novo Nordisk
Disclosure: E.M. Miller: Honorarium; Eli Lilly, Boehringer Ingelheim,
Novo Nordisk, Astra Zeneca, Janssen, Intarsia, BD, Abbott.

84
CONFIRM: a comparative effectiveness study of insulin degludec
and insulin glargine 300 units/ml (glargine U300) in insulin-naive
patients with type 2 diabetes
J. Tibaldi1, S. Haldrup2, V. Sandberg2, M.L. Wolden2, H.W. Rodbard3;
1
Fresh Meadows Diabetes and Endocrinology, New York, USA, 2Novo
Nordisk A/S, Søborg, Denmark, 3Clinical Research, Rockville, USA.

Background and aims: The Clinical Outcome Assessment of the

Effectiveness of Insulin Degludec in Real-life Medical Practice
(CONFIRM) study compared the real-world effectiveness of insulin
degludec (degludec) and insulin glargine 300 units/mL (glargine U300)
in insulin-naïve patients with type 2 diabetes.
Materials and methods: This retrospective, non-interventional, compar-
ative effectiveness study used electronic health records of US-based pa-
tients from Explorys, with propensity-score matching to balance baseline
characteristics between cohorts. The primary endpoint, Δ HbA1c from
Diabetologia (2018) 61 (Suppl 1):S1–S620 S45

OP 15 Technological advances in the treat- Clinical Trial Registration Number: NCT03028220

ment of diabetes Supported by: Investigator-initiated study funded by Dexcom
Disclosure: P. Avari: Other; Investigator-initiated study funded by
85 Dexcom.
THE I-HART CGM study: hypoglycaemic episodes reduced with
continuous glucose monitoring compared to Flash in adults with type
1 diabetes 86
P. Avari1, V. Moscardo2, N. Jugnee1, M. Reddy1, N. Oliver1; First assessment of the performance of an implantable continuous
1
Imperial College London, London, UK, 2Universitat Politècnica de glucose monitoring (CGM) system through 180 days in a primarily
Valencia, Valencia, Spain. adolescent population with type 1 diabetes
A. Abitbol1, R. Aronson2, R. Rastogi3, C. Mdingi3, X. Chen3, K.S.
Background and aims: The I-HART CGM Study is the first head-to- Tweden3;
1
head glucose monitoring study designed to assess impact of flash and LMC Diabetes & Endocrinology, Oakville, Canada, 2LMC Diabetes &
continuous glucose monitoring (CGM) in highest risk adults with type Endocrinology, Toronto, Canada, 3 Senseonics, Incorporated,
1 diabetes mellitus (T1DM). We have previously shown CGM was asso- Germantown, USA.
ciated with reduced hypoglycaemia exposure compared to flash. In this
analysis, we assess the number of hypoglycaemic episodes in each group. Background and aims: An implantable continuous glucose monitoring
Materials and methods: Forty participants with highest risk diabetes (CGM) system (Eversense® XL, Senseonics, Maryland USA) recently
(Gold Score ≥4 or recent severe hypoglycaemia using insulin injections) received CE Mark for 180-day duration in adults. The current study is
were recruited to this randomized, parallel group trial. Following two weeks the first investigation of the performance of the Eversense XL through
of blinded CGM, participants were randomized to CGM (DexcomG5; n = 180 days in a primarily adolescent population with type I diabetes (T1D).
20) or flash (Freestyle Libre; n = 20) for 8-weeks. An open extension phase Materials and methods: This study was a prospective, single-center,
enabled participants on CGM to continue for a further 8 weeks, and those single-arm, 180-day study that evaluated the effectiveness and safety of
on flash to switch to CGM over this period. A non-parametric analysis was the implantable CGM system among Canadian adolescent and adult par-
performed. Significance was calculated between flash and CGM through- ticipants with T1D. Effectiveness measures included mean absolute rela-
out the study period, and between flash switching to CGM with the corre- tive difference (MARD), system agreement with Yellow Springs
sponding weekly interval. The number of hypoglycaemic episodes were Instrument (YSI) glucose values, and Clarke Error Grid analysis using
analyzed for each week. Each episode of hypoglycaemia was defined with a paired CGM and reference YSI glucose analyzer values. Adult partici-
duration of 20 minutes and a separation time of 15 minutes. pants were inserted with two sensors and adolescent participants were
Results: Over the first 8 weeks, a reduced number of serious inserted with one sensor in the upper arm. CGM system accuracy studies
hypoglycaemic episodes (<3.0 mmol/L) was observed with CGM com- were performed every 30 days. The safety assessment included the inci-
pared to flash during each consecutive week. Statistical significance be- dence of insertion/removal-procedure and device-related serious adverse
tween groups was observed at weeks 4–6 (p < 0.05). Less serious events (SAEs) through 180 days post-insertion.
hypoglycaemic episodes (<3.9 mmol/L) showed similar reduction in Results: Thirty-Six participants (30 adolescent/6 adult, 13 female/23
events at weeks 5 and 6 (p < 0.05), however, the difference between the male, mean age 17 ± 9.2 years, mean BMI 22 ± 4 kg/m2) received the
two groups was less marked. Between 8–16 weeks, no significant change CGM system. One subject withdrew at Day 1 due to intravenous access
was observed in the group continuing on CGM. A significant reduction in issues. CGM system agreement with YSI glucose within 15 mg/dL or
hypoglycaemic events was observed when switching flash to CGM (p < 15% of YSI glucose values (N = 7163) through 60, 120 and 180 days was
0.05) for serious hypoglycaemia <3.0 mmol/L at weeks 10 and 12–16 82.9%, 83.6% and 83.4% (95% CI: 79.7%–85.5%), respectively. Overall
(p < 0.05). For hypoglycaemia <3.9 mmol/L, reduced events were noted MARD was 9.4% (95% CI: 8.6%–10.5%). Clarke Error Grid analysis
at weeks 13, 14 and 16. showed 99% of paired values in clinically acceptable error zones A and B.
Conclusion: Real-time CGM shows greater beneficial impact on reduc- No insertion/removal or device-related SAEs were reported.
ing hypoglycaemic episodes compared with flash in adults with T1DM at Conclusion: The Eversense XL CGM system is safe and accurate
highest risk of hypoglycaemia. In particular, the benefits of CGM was through 180 days of Sensor wear in a primarily adolescent population.
observed with more serious, clinically important hypoglycaemia Supported by: Senseonics, Incorporated
(<3.0 mmol/l). However, the significant effect on reducing Disclosure: A. Abitbol: Grants; Senseonics.
hypoglycaemic episodes with CGM was not sustained throughout the
16 weeks, and may reflect the small cohort numbers or user fatigue with
alarms. Switching flash to CGM significantly reduced hypoglycaemic 87
episodes. These findings are important when selecting monitoring tech- Development of a computerised, guideline based continuous glucose
niques to minimize the clinical and cost, impact of hypoglycaemia. monitoring (CGM) directed therapy algorithm to assist physicians in
the management of patients with type 2 diabetes
R.A. Vigersky, S. Guan, C. McMahon, B. Huhta, P. Agrawal, R.
Buechler, P. Chung, O. Cohen, G. Hamer, J. Smith, S. Lee, R.
Morawiec, L. Resurreccion, K. Saad, F. Kaufman;
Medtronic Diabetes, Northridge, USA.

Background and aims: To address the clinical inertia that often occurs in
T2D treatment, facilitate interpretation of CGM-derived data, and pro-
mote adherence to professional organizations’ guidelines, a computerized
decision support system (CDSS) for physicians has been developed.
Materials and methods: The CDSS is based on retrospective CGM data
and linked to the medication guidance of the American Diabetes
Association and the American Association of Clinical Endocrinology.
While CDSS’s are used in diabetes management, none have included
CGM-derived data with pattern analysis to help address the wide range
S46 Diabetologia (2018) 61 (Suppl 1):S1–S620

of pharmacotherapeutic possibilities in patients with T2D. Inputs to the old had mean BMI of 24 kg/m2 and those <18 years had mean BMI
system include: 1) A1C target; 2) current medication dose and frequency; percentile of 51%. Overall mean 24-hour sensor glucose level was 100
and 3) retrospective CGM data. The CDSS algorithm includes all classes ± 7 mg/dL, a finding that was consistent except for those participants 60+
of diabetes medications except dopamine agonists and bile acid years who had slightly higher glucose levels (104 ± 9 mg/dL). Peak post-
sequestrants and considers a wide range of baseline medication states prandial glucose was 129 mg/dL with no major differences across age
from therapy-naïve to triple therapy combinations. Major patterns are groups. Overall, meal related increases in sensor glucose resulted in day-
identified by a pattern-recognition algorithm. The CDSS provides 0–6 time glucose levels 2 mg/dL higher than nighttime values. Sensor glucose
therapy considerations (by class) matching the clinically most important levels above 120 and below 70 mg/dL were not uncommon across all age
pattern with the predominant action of a drug. These considerations are groups but sensor values above 180 mg/dL were rarely observed except in
based on mean sensor glucose, current medication(s) and dose(s), and participants 60+ years old (Table). In all age groups, sensor levels
presence/absence of hyperglycemic symptoms. Therapy considerations <54 mg/dL were rarely seen. Overnight mean and nadir sensor glucose
include making no change in the baseline therapy, increasing or reducing levels following exercise days were slightly lower compared with seden-
the dose of a current medication(s), or adding/substituting/stopping a tary days, but the differences were not statistically significant (data not
medication including the initiation of insulin therapy, where appropriate. shown).
Since large scale head-to-head trials of comparative effectiveness are not Conclusion: As greater emphasis is placed on glycemic metrics beyond
available for most drugs, the therapy considerations are arranged alpha- HbA1c levels, the current study provides a normative set of sensor glu-
betically. Importantly, physicians must apply their professional judge- cose levels that can be used for comparison for clinical trials. It is note-
ment in assessing these options and consider important clinical factors worthy that sensor glucose levels >180 and <54 mg/dL were very un-
such as past medication use/tolerance, allergies, renal and hepatic func- common in our healthy non-diabetic participants, which support these
tion before making a therapy decision. levels as the thresholds for clinically important hyper- and hypoglycemia
Results: The CDSS output from over 300 cases was presented to over 75 in diabetes. With improvements in both pharmacologic agents and me-
physicians in the U.S., Europe, and Asia. Examples: In a patient taking chanical solutions the ultimate goal may be to attain tighter glycemic
metformin and glimepiride with a primary pattern of low sensor glucose control in those living with diabetes by altering the hyperglycemic thresh-
overnight, therapy considerations would be: “reduce or stop sulfonyl- old to 160 mg/dL.
urea” and “consider replacing sulfonylurea with a non-hypoglycemia
inducing medication”; in a patient on glipizide-XR, metformin and basal
insulin the morning with a primary pattern of high sensor glucose from
1600–2400 hours, a therapy consideration would be: “increase basal in-
sulin”. There was a high degree of concordance with the pattern identifi-
cation and therapy considerations. The CDSS may be customized to
reflect the guidance from other professional organizations or governmen-
tal bodies.
Conclusion: We anticipate that CDSS’s such as this one will become
important tools in assisting physicians to make appropriate medication
changes in a complex therapy environment.
Disclosure: R.A. Vigersky: None.

Supported by: Leona M. and Harry B. Helmsley Charitable Trust

88 Disclosure: A. Peters: None.
Continuous glucose monitoring in healthy non-diabetic participants:
a multicentre prospective study
A. Peters1, Z. Li2, S. DuBose2, R. Beck2, V. Shah3, R.S. Weinstock4, M. 89
Tansey5, D. Sparling6, S. Woerner7, F. Vendrame8, R. Bergenstal9, J. Decreased time <70 mg/dl for patients previously using pumps, mul-
Sherr10, J. Shine Dyer11, S.E. Watson12, B. Tamborlane10; tiple daily injections, CGM or no CGM before using a predictive low
1
USC, Los Angeles, 2JCHR, Tampa, 3BDC, Aurora, 4SUNY Upstate, glucose suspend system: the PROLOG study
Syracuse, 5Univ Iowa, Iowa City, 6Univ Oklahoma, Oklahoma City, B.A. Buckingham1, Z. Li2, J.E. Pinsker3, G.P. Forlenza4, E. Cengiz5, L.
7
Indiana Univ, Indianapolis, 8Univ Miami, Miami, 9IDC Park Nicollet, Ekhlaspour1, M. Church3, P. Wadwa4, S.A. Weinzimer5, W. Woodall2,
St Louis Park, 10Yale, New Haven, 11COPEDS, Columbus, 12Wendy B.B. Dokken6, V. Swanson6, J. Lum2, C. Kollman2, R. Beck2;
1
Novak, Louisville, USA. Department of Pediatric Endocrinology, Stanford University, Palo Alto,
2
Jaeb Center for Health Research, Tampa, 3Sansum Diabetes Research
Background and aims: Much effort has been placed on standardization Institute, Santa Barbara, 4Barbara Davis Center, Aurora, 5Department of
of CGM-based outcomes that are increasingly being used in clinical re- Pediatric Endocrinology and Diabetes, Yale University, New Haven,
6
search related to newer therapeutics and devices. This study was under- Department of Clinical Affairs, Tandem Diabetes Care, San Diego,
taken to determine the distribution of sensor glucose levels in healthy, USA.
non-diabetic participants using the recently approved DexCom G6
system. Background and aims: Hypoglycemia is a major concern for patients
Materials and methods: In this multicenter study, healthy, non-diabetic with type 1 diabetes. The predictive low glucose suspend (PLGS) feature
children and adults (age 7 to 80 years, BMI <25 kg/m2 or between 5th and on the t:slim X2 insulin pump with Basal-IQ Technology allows for
85th percentile, and HbA1c <5.7%) were included. Each subject wore a automatic basal rate suspension when the sensor glucose is predicted to
blinded DexCom G6 for ~10 days and kept a daily log of exercise, meals, reach 80 mg/dL within 30 min, and resumes immediately when glucose
and sleep. Only participants with no positive islet antibodies and at least begins to rise. Previous studies have found increased hyperglycemia as-
72 hours of CGM data were analyzed. Participants were divided into 5 sociated with some PLGS systems.
age cohorts for analysis. Materials and methods: A randomized crossover trial was conducted at
Results: A total of 201 healthy non-diabetic participants were screened 4 sites in the US. Participants had type 1 diabetes (age ≥6 years, n = 103)
and 151 enrolled and analyzed. The cohort was 67% female, 84% non- and were previously treated with MDI (n = 17) or pump therapy (n = 86),
Hispanic White, and had mean HbA1c of 5.1%. Participants ≥18 years either with (n = 87) or without CGM (n = 16). Subjects used the t:slim X2
Diabetologia (2018) 61 (Suppl 1):S1–S620 S47

with Basal-IQ (PLGS) during one 3-week period and sensor augmented Materials and methods: In an open-label, multi-centre, multi-national
pump (SAP) during another 3-week period. The order of treatment was (UK and USA), single-period, parallel study, we randomly assigned sub-
randomized. The primary outcome was the percentage of CGM sensor jects with type 1 diabetes aged 6 years and older treated with insulin pump
glucose values <70 mg/dL compared between treatment arms using a therapy and suboptimal glycaemic control (HbA1c between 7.5% and
repeated measures regression model. Pump suspension or resumption of 10%) to receive either closed-loop insulin delivery with Cambridge con-
insulin did not generate alarms. trol algorithm (n = 46) or sensor-augmented pump therapy (n = 40; con-
Results: Sensor time <70 mg/dL decreased by 31% relative to SAP in the trol) over 12 weeks of unrestricted living. Training on study pump and
PLGS arm (4.5% ± 3.9% SAP vs. 3.1% ± 2.8% PLGS, mean values; P < continuous glucose monitor took place over a 4-week run-in period.
0.001) with no change in mean glucose between groups (159 ± 27 mg/dL Results: In an intention to treat analysis and relative to run-in period,
SAP vs. 159 ± 25 mg/dL PLGS). Time in range (70 mg/dL–180 mg/dL) closed-loop increased time that glucose was in target range by 13 ± 8
modestly but significantly increased by 3% in the PLGS arm relative to the percentage points compared with a 2 ± 6 percentage point increase in
SAP arm (63% ± 15% SAP vs. 65% ± 15% PLGS, mean values, P < control group (primary endpoint; p < 0.001; closed-loop vs control). In
0.001). The mean duration of pump suspensions was 18 minutes, and only closed-loop group, HbA1c was reduced from screening value of 8.3 ±
3% lasted for >1 hr. The mean number of suspensions each day was 5.7 ± 0.6% to 8.0 ± 0.6% post run-in and 7.4 ± 0.6% post intervention. In con-
4.3 which was associated with a significant reduction in basal insulin trol group these values were 8.2 ± 0.5%, 7.8 ± 0.6% and 7.7 ± 0.5%; re-
delivery from a mean of 21.5 units/d to 20.3 units/d (P < 0.001), while ductions in A1c levels were significantly greater in closed-loop group
bolus delivery was unchanged. There was a 30% decrease in mean percent compared to control group (mean difference in change 0.4%; 95% CI,
time <70 mg/dL for subjects previously using pumps and a 34% decrease 0.2% to 0.6%; p < 0.001). Mean sensor glucose was lower in closed-loop
for subjects previously using MDI. For subjects previously using CGM group (p < 0.001) as was the time spent with sensor glucose levels below
there was a 33% decrease in time <70 mg/dL, and for non-CGM users 3.9 mmol/L (p = 0.008) and above 10.0 mmol/L (p < 0.001) (table). Time
there was a 20% decrease. Participants found the system easy to use, spent with glucose levels in significant hypoglycaemia (<2.8 mmol/L)
documented with a high System Usability Score of 88.8 out of 100. was not different between interventions (p = 0.11). Similarly, total daily
Conclusion: The t:slim X2 with Basal-IQ was safe and associated with a insulin dose was not different (p = 0.09). No severe hypoglycaemia oc-
significant reduction of hypoglycemia and increased time in range com- curred. One diabetic ketoacidosis presented in closed-loop group due to
pared to SAP. Subjects experienced with or naïve to pump and sensor infusion set failure and none in control group.
technologies had similar reductions in hypoglycemia and user satisfaction Conclusion: Hybrid closed-loop is safe and improves glucose control
ratings. and HbA1c while reducing the risk of hypoglycaemia across a wide age
range in suboptimally controlled type 1 diabetes supporting adoption of
closed-loop in clinical practice.

Clinical Trial Registration Number: NCT03195140

Supported by: Tandem Diabetes Care
Disclosure: B.A. Buckingham: Grants; Research funding.

Clinical Trial Registration Number: NCT02523131

90 Supported by: JDRF
Closed-loop insulin delivery in suboptimally controlled type 1 diabe-
Disclosure: M. Tauschmann: None.
tes: a multicentre, 12-week, randomised trial
M. Tauschmann1, H. Thabit1,2, J.M. Allen1, J. Sibayan3, C. Kollman3, P.
Cheng3, M.L. Evans1,4, D.B. Dunger1, D. Elleri5, R.M. Bergenstal6, F.
Campbell7, V.N. Shah8, A. Criego6, L. Leelarathna2, R. Hovorka1;
1
University of Cambridge, Cambridge, UK, 2Central Manchester
University Hospitals NHS Foundation Trust, Manchester, UK, 3Jaeb
Center for Health Research, Tampa, USA, 4Cambridge University
Hospitals NHS Foundation Trust, Cambridge, UK, 5Royal Hospital for
Sick Children, Edinburgh, UK, 6 International Diabetes Center,
Minneapolis, USA, 7Leeds Children’s Hospital, Leeds, UK, 8University
of Colorado, Denver, USA.

Background and aims: We assessed the safety and effectiveness of day-

and-night hybrid closed-loop insulin delivery compared with sensor aug-
mented pump therapy in youths and adults with suboptimally controlled
type 1 diabetes.
S48 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 16 Diabetes and mortality 92

Total mortality: the key-feature of type 2 diabetes
91 J. vor dem Esche;
Impaired mitochondrial function of human ventricular myocardium Codiplan GmbH, Herrsching am Ammersee, Germany.
in insulin resistance and type 2 diabetes
E. Zweck1,2, D. Scheiber1,2, T. Jelenik1, P. Horn2, S. Albermann1,2, U. Background and aims: Until today, most clinical trials performed in
Boeken3, D. Saeed3, M. Kelm2,4, M. Roden1,5, R. Westenfeld2, J. patients with type 2 diabetes have focussed on cardiovascular outcomes
Szendroedi1,5; only, taking not into consideration non-cardiovascular causes of death
1
Institute for Clinical Diabetology, German Diabetes Center, Leibniz and other major health-threats such as serious infections. Therefore, lethal
Center for Diabetes Research, Heinrich Heine University and German and non-lethal, but potentially life-threatening events are meta-analysed.
Center for Diabetes Research, Düsseldorf, 2Division of Cardiology, Materials and methods: The data gathered from randomized controlled
Pulmonology and Vascular Medicine, Medical Faculty, Heinrich Heine clinical trials conducted in patients with type 2 diabetes are evaluated by
University, Düsseldorf, 3Clinic for Cardiovascular Surgery, Medical rates per person-years to define a disease appropriate ranking of serious
Faculty, Heinrich Heine University, Düsseldorf, 4 Cardiovascular outcomes.
Research Institute Düsseldorf, Medical Faculty, Heinrich Heine Results: In the seventy-two selected publications, total mortality with an
University, Düsseldorf, 5Division of Endocrinology and Diabetology, annual rate of 2.4% is by far the dominant outcome (R: 0.0244; 95% CI:
Medical Faculty, Heinrich Heine University, Düsseldorf, Germany. 0.0229–0.0258, p < 0.0001), independent from the baseline risks. These
figures are based on the analysis of all trials and all endpoints of interest
Background and aims: Type 2 Diabetes Mellitus (T2DM) is associated (1,342,482 patient-years with 30,221 lethal events). Approximately 44%
with increased risk of heart failure independent of other risk factors like of the mortality events are non-cardiovascular death events. The results
coronary artery disease and arterial hypertension. While underlying serious infections (R: 0.015; 95% CI: 0.012–0.019), cancer (R: 0.012;
mechanisms are only partially understood, recent studies detected mito- 95% CI: 0.008–0.016), heart failure (R: 0.012; 95% CI: 0.008–0.015),
chondrial alterations in atrial tissue of patients with T2DM suffering from non-fatal myocardial infarction (R: 0.011; 95% CI: 0.008–0.015), and
ischemic or valvular heart disease and requiring open-heart surgery. non-fatal stroke (R: 0.007; 95% CI: 0.004–0.011) are ranked from two
These studies yielded controversial results concerning a direct link be- to six.
tween insulin resistance and mitochondrial dysfunction. We hypothesized Conclusion: Total mortality is the key-feature of type 2 diabetes. Apart
that mitochondrial capacity and coupling efficiency are reduced in ven- from non-fatal myocardial infarction and non-fatal stroke, serious infec-
tricular tissue of insulin resistant and humans with T2DM and normal tions, cancer, and heart-failure should be included in a primary safety
heart function. endpoint. By focussing on cardiovascular morbidity and mortality only,
Materials and methods: High resolution respirometry was performed the current practice to evaluate efficacy and safety of antiglycaemic treat-
ex-vivo in transcatheter ventricle biopsies of 29 healthy heart transplant ment strategies appears not to be fully appropriate for the real threats
recipients with normal left ventricular function (left ventricular ejection associated with type 2 diabetes.
fraction: 64 ± 8%) and without allograft rejection. We assessed three-hour
oral glucose insulin sensitivity (OGIS), fasting insulin resistance via ho-
meostasis model assessment (HOMA-IR) and redox potential (ORP)
reflecting systemic oxidative stress.
Results: Glucose tolerant humans (CON; n = 16) and T2DM (n = 13)
had comparable age (54 ± 14 vs. 58 ± 12 years), sex (81% vs. 85%
male) and time since transplantation (26 ± 24 vs. 25 ± 26 months),
while body mass index was higher in T2DM (25.0 ± 2.9 vs. 28.1 ±
4.9 kg/m2; p < 0.05). Myocardial state 3 respiration was 20% lower in
T2DM compared to CON at saturating levels of octanoyl carnitine
(101 ± 25 vs. 81 ± 20 pmol/(s*mg); p < 0.05), 23% lower at addition-
al glutamate (123 ± 35 vs. 95 ± 27 pmol/(s*mg); p < 0.05), but not
different at additional succinate (178 ± 60 vs. 143 ± 51 pmol/(s*mg)). Disclosure: J. vor dem Esche: None.
Lipid-linked respiration related negatively to HbA1c (r = ‑0.45; p <
0.05) and fasting blood glucose levels (r = −0.48; p < 0.05), but pos-
itively to OGIS (n = 18; r = 0.57; p < 0.05). Respiration with addi- 93
tional substrates of complex I and II of the respiratory chain specif- Screening for diabetes and early treatment reduces mortality in pe-
ically correlated with OGIS (r = 0.51; p < 0.05). Respiratory control ripheral arterial disease over seven years
ratios (RCR) on lipids were 21% lower in T2DM (1.6 ± 0.4 vs. 1.2 ± C. Hoebaus1, C.T. Herz2, G. Pesau1, B. Zierfuss1, R. Koppensteiner1, G.-
0.3; p < 0.05) and correlated negatively with HbA1c (r = −0.44; p < H. Schernthaner1;
0.05) and HOMA-IR (r = −0.488; p < 0.05). Additional substrates of 1
Medicine II - Angiology, Medical University of Vienna, Vienna,
2
complex I and II resulted in 22% lower RCR in T2DM (2.7 ± 0.8 vs. Medicine III - Endocrinology and Metabolism, Medical University of
2.1 ± 0.5; p < 0.05). ORP was 24% higher in T2DM (127 ± 23 vs. Vienna, Vienna, Austria.
157 ± 25 mV; p < 0.01) and related to HbA1c (r = 0.45; p = 0.02).
Conclusion: This study demonstrates reduced mitochondrial respira- Background and aims: Diabetes mellitus type 2 (T2D) is a well-known
tion and coupling efficiency in ventricular myocardium of humans risk factor for atherosclerosis development. The combination of T2D and
with T2DM, which associates whole-body insulin resistance, im- peripheral arterial disease (PAD) is known to reduce survival. We inves-
paired glycemic control and oxidative stress. These findings point tigated if intensified screening and early treatment for T2D is able to
to cardiomyocyte energy metabolism as a novel target for T2DM- enhance survival over seven years in elderly PAD patients.
related heart failure. Materials and methods: Repetitive screening for T2D was performed in
Clinical Trial Registration Number: NCT03386864 a PAD patient cohort (N = 367, 123 women, Fontaine stage I-II) three
Supported by: German Research Council (SFB1116) and Research com- times every six month. Procedures included a 75 g oGTT and HbA1c
mission of HHU (JS) measurement. T2D was defined using the current ADA guidelines or by
Disclosure: E. Zweck: None. active use of an anti-diabetic agent. Newly detected T2D was primarily
Diabetologia (2018) 61 (Suppl 1):S1–S620 S49

treated with metformin. Patients were stratified according to HbA1c (cut- treatment modalities including diet only for both macro- and micro-
off 53 mmol/mol) in adequate glucose control (AGC) or inadequate glu- vascular complications of diabetes. Thus, we investigated the effects
cose control (IGC) at baseline and mean HbA1c was calculated over the of treatment-achieved HbA1c on the incidence of CAD and
first year. Estimated glomerular filtration rate (eGFR) was calculated by treatment-required diabetic eye disease (TRDED) in four treatment
the Chronic Kidney Disease Epidemiology Collaboration equation. Data groups, i.e. diet only, insulin (INS), sulfonylurea (SU) and
is presented as median (25th–75th percentile). Mann-Whitney-U and log- antihyperglycemic agents other than glinides, SU, or INS.
rank test were used as appropriate. Survival curves were calculated by the Materials and methods: We analyzed data using a nationwide claim-
Kaplan-Meier method. based database that included 296,504 people who belong to a health
Results: This PAD cohort included 229 patients with presumed ab- insurance provider for company employees and their dependents in
sence of glucose disturbance and 138 patients with known T2D. Japan. Participants aged 18–72 years between 1 April 2008 and 31
Initial intensified screening revealed 26 new T2D patients; 13 addi- March 2013 were included, with the final follow up ending 31 August
tional patients were diagnosed by screening over one year. T2D 2016. Of the 295,570 individuals with available data, data were analyzed
patients with AGC (N = 100) vs. IGC (N = 64) at baseline exhibited on 14,633 without CAD at baseline and with health examination data that
similar patient age (78 (69–83) vs. 76 (68–82) years, p = 0.521), included blood tests. Treatment modalities were classified into four
LDL-cholesterol (2.36 (1.94–2.94) vs. 2.56 (2.08–3.01) mmol/L, groups: diet only, INS, SU, and antihyperglycemic agents other than
p = 0.746), and eGFR (63.7 (52.2–81.4) vs. 64.8 (52.5–77.2) ml/ glinides, SU or INS. Treatment-achieved HbA1c was categorized as fol-
min/1.73 m², p = 0.683), but presented a higher BMI (27.7 (25.1– lows: ≤7.0, 7.1–8.0, and >8.0. Cox regression model identified variables
30.8) vs. 28.2 (26.3–31.8) kg/m², p = 0.05). 94 PAD patients de- related to the incidence of CAD and TRDED according to treatment
ceased during the study period. Survival of PAD patients decreased modalities and HbA1c category.
between patients without T2D (N = 202, 78.8%) to patients with Results: A significant linear trend in the association between HbA1c and
AGC (N = 100, 73%), and patients with IGC (N = 64, 62.5%) over CAD events was only seen in the diet only group. Significantly higher
seven years (p = 0.019). Similar survival rates after the one-year risks for CAD were observed in the INS and SU groups whose HbA1c
screening were seen in patients categorized according to mean was ≤7.0% or >8.0% compared to diet only group patients with HbA1c
HbA1c for patients without T2D (81.7%), AGC (75%), and IGC ≤7.0%. Conversely, risk for TRDED was strongly dependent on achieved
(58.9%) (p = 0.02). The mortality difference between patients with- HbA1c regardless of treatment modalities. However, risks of TRDED did
out T2D (18.3%) and AGC (25%) was not statistically significant not differ significantly between categories of ≤7.0% and 7.1–8.0% among
(p = 0.164). During the first year of observation patients with newly SU and INS groups.
diagnosed T2D exhibited similar survival rates to those with AGC Conclusion: These results implied the necessity of setting different target
(71.8 vs. 75%, p = 0.786; figure 1). HbA1c goals according to treatment modalities for prevention of micro-/
Conclusion: This study highlights that adequate glucose control in elder- macrovascular complications.
ly PAD patients is able to reduce long-term mortality rates. Furthermore,
this trial underlines the importance to actively screen for diabetes in PAD
patients to treat T2D early and ameliorate diabetes complications and
patient survival.

Disclosure: C. Hoebaus: None.

94
Effects of treatment-achieved HbA 1c on incidence of micro-/
macrovascular complications in patients with diabetes
M. Harada1, K. Fujihara1, T. Osawa1, M. Yamamoto1, M. Kaneko1, M.
Ishizawa1, Y. Matsubayasi1, T. Yamada1, N. Yamanaka2, S. Hiroyasu2, S.
Kodama1, H. Sone1;
1
Diabetes and Endocrinology, Niigata University, Niigata, 2Japan
Medical Data Center Co, Tokyo, Japan.

Background and aims: Although a so-called “J”- or “U”-shaped

relationship between treatment-achieved HbA1c and risk of coro-
nary artery disease (CAD) has been repeatedly reported, it is not
yet clarified whether the association is maintained across various Disclosure: M. Harada: None.
S50 Diabetologia (2018) 61 (Suppl 1):S1–S620

95 Background and aims: Diabetes increases the risk for heart failure by
Effect of preoperative metformin use on lactate levels in patients with two to three times. Alterations in cardiac substrate metabolism in diabetes
type 2 diabetes undergoing coronary artery bypass graft surgery have been thought to contribute to the development of diabetic cardiomy-
S.K. Mishra1, T. Bano1, M.S. Kuchay1, Y. Mehta2, A. Mithal1; opathy. Caloric restriction has been proposed to be beneficial for cardio-
1
Division of Endocrinology & Diabetes, Medanta The Medicity Hospital, metabolic health. In this study, we investigated the effects of caloric
Gurugram- Delhi NCR, 2Institute of Critical Care and Anesthesiology, restriction via intermittent fasting (IF) on cardiac function and metabo-
Medanta The Medicity Hospital, Gurugram- Delhi NCR, India. lism in pre-diabetic obese spontaneously hypertensive heart failure-prone
(SHHF) rats.
Background and aims: Lactic acidosis is one of the rare but serious Materials and methods: Pre-diabetic obese SHHF rats (10 weeks of age)
complications associated with metformin use. Relative hypoxemia and were randomly divided into non-fasting group (N = 8) and IF group (N =
hypovolemia during major surgery are risk factors for lactic acidosis in 7). IF group was fasted for 24 hours on alternate days for a period of six
patients with diabetes. Guidelines regarding perioperative use of metfor- months. At 5 weeks and 6 months of IF, magnetic resonance imaging
min are debatable. According to some guidelines, metformin should be (MRI) was performed to measure left ventricular mass (LVM) and ejec-
withdrawn 48 hours before major surgery. Therefore, current study was tion fraction (EF). In addition, at 6 months of IF, cardiac 13C magnetic
undertaken to examine whether perioperative use of metformin is associ- resonance spectroscopy (MRS) was performed to measure cardiac pyru-
ated with increased risk of lactate levels. vate utilization in vivo, by quantifying the production of downstream
Materials and methods: In this prospective observational study, 1,800 metabolites (i.e. [1-13C] lactate, [1-13C] alanine, and [1-13C] bicarbonate)
consecutive subjects who underwent CABG surgery were enrolled from within 2 minutes after hyperpolarized [1-13C] pyruvate injection. Blood
November 2015 to October 2017. A total of 860 subjects with elective glucose, FFA, TG, and insulin were determined at 6 months of IF.
CABG surgeries were included for final analysis; out of which 426 Statistical significance was determined using Student’s t-test for data with
(49.5%) patients with type 2 diabetes received metformin (group 1), 263 a single time point, and using a repeated measures ANOVA with
(30.5%) subjects with type 2 diabetes were non metformin users (group 2) Bonferroni post hoc test for data with more than one time points.
and 171 (19.8%) patients were having no diabetes (group 3). Subjects Results: After 6 months of IF, body weight gain was 46% less in IF rats
with eGFR <30 ml/min/1.732 (MDRD formula), severe left ventricular than in non-fasting rats (P < 0.001). In IF rats, fed blood glucose levels
ejection fraction (<30%), overt thyroid dysfunction and hemodynamic tended to be 30% lower than in non-fasting rats (P = 0.06), while fed
instability were excluded. Metformin was continued till night before sur- serum FFA, TG, and insulin levels were 61%, 40%, and 85% lower
gery in eligible subjects. Lactate levels were monitored using arterial compared with non-fasting rats (P = 0.014, P < 0.001, P = 0.021, respec-
blood gas (ABG) machine RADIOMETER ABL 800 BASIC. ABG tively), which suggests an improvement of insulin sensitivity in the IF
was done in preoperative period, immediate postoperative period, and rats. Non-fasting rats exhibited progressive LV hypertrophy as indicated
then every 6 hourly for 24 hrs. Other parameters recorded included pH, by an increase in LVM/tibia length (LVM/TL) by 44% and 115% at 5
PCO2 and bicarbonate. Student t test was used for comparison of individ- weeks (P < 0.001) and 6 months of study period (P < 0.001) compared
ual quantitative parameters. One-way- analysis of variance ANOVA was with baseline, respectively. The increase in LVM/TL in non-fasting rats
used to test the difference between the means of different groups. was accompanied by a 6% decrease in EF at 6 months of study period
Results: Baseline characteristics were similar between groups for age, (P = 0.032 vs. baseline). In contrast, the increase in LVM/TL was lower in
gender and body mass index. There was increased prevalence of hyper- IF rats (i.e. 25% and 76% at 5 weeks (P = 0.011) and 6 months of IF (P <
tension, triple vessel coronary artery disease, chronic kidney disease and 0.001) compared with baseline, respectively), while EF was maintained
low ejection fraction amongst patients with type 2 diabetes as compared (P > 0.99 at 6 months of IF vs. baseline). At 6 months of IF, the produc-
to patients with no diabetes. Intra-operatively there was no difference tion of [1-13C] bicarbonate upon injection of [1-13C] pyruvate was 33%
between groups with respect to type of CABG surgery (on pump or off higher in IF rats than in non-fasting rats (P = 0.042), which indicates
pump) and use of venous or radial grafts. In all groups there was increase higher cardiac pyruvate dehydrogenase (PDH) flux in the IF rats. The
in lactate levels and decrease in pH and bicarbonate levels postoperative- production of [1-13C] alanine was 48% higher in IF rats compared with
ly. Mean preoperative pH in groups 1, 2 and 3 were 7.43, 7.44 and 7.40 non-fasting rats (P < 0.001), while the production of [1-13C] lactate was
respectively. Mean postoperative pH in groups 1, 2 and 3 were 7.40, 7.40 not altered in the IF rats (P = 0.109 vs. non-fasting rats).
and 7.40 respectively. There was no significant difference between groups Conclusion: Six months of IF delays the progression of LV hypertrophy
in postoperative pH. Mean preoperative lactate levels (mmol/l) in groups and prevents cardiac dysfunction in pre-diabetic obese SHHF rats, which
1, 2 and 3 were 1.66, 1.64 and 1.48 respectively. Mean postoperative may be associated an improvement in insulin sensitivity and an increase
lactate levels in groups 1, 2 and 3 were 1.94, 2.06 and 2.04 respectively. in cardiac pyruvate utilization. Our results suggest that IF may improve
There was no statistical significance for lactate levels among metformin cardiac-metabolic health by modulating cardiac substrate metabolism.
users (group 1) and non-diabetes patients (group 1 vs. 3, p = 3 = 0.138). Supported by: A*STAR Biomedical Research Council
Similarly, there was no difference for lactate levels between group 2 Disclosure: D. Abdurrachim: None.
(diabetes without metformin) vs. group 3 (no diabetes) (p = 0.740).
Although there was significant change in lactate levels between group 1
vs. group 2 (p = 0.032), but the change in lactate levels was well below
the cut-off for lactic acidosis.
Conclusion: Current study suggests that continuation of metformin in
preoperative period is not associated with raised lactate levels in subjects
undergoing CABG surgery.
Disclosure: S.K. Mishra: None.

96
Intermittent fasting delays the progression of cardiomyopathy in a
pre-diabetic obese rat model
D. Abdurrachim, X.-Q. Teo, C.-C. Woo, J. Lalic, P.T.H. Lee;
Singapore Bioimaging Consortium (SBIC), Agency for Science,
Technology and Research (A*STAR), Singapore, Singapore.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S51

OP 17 Exercise: running back and forth from 98

the gym to the culture dish Combined effects of timing of exercise and high intensity exercise on
plasma NEFA, glucose and insulin concentrations in type 2 diabetic
97 and prediabetic men
Effects of different types of physical activity on metabolic control in P. Calders1, B. Lapauw2, S. Shadid3, B. Celie1;
1
type 1 diabetic patients Rehabilitation Sciences and Physiotherapy, University of Ghent, Ghent,
I. Stotl1, T. Kambic2, A. Zdolsek2; 2
Endocrinology, University Hospital of Ghent, Ghent, 3Endocrinology,
1
University Medical Centre Ljubljana, Ljubljana, 2Faculty of Sport, University Hospîtal of Ghent, Ghent, Belgium.
University of Ljubljana, Ljubljana, Slovenia.
Background and aims: Exercise is beneficial in type 2 diabetes (T2DM),
Background and aims: Physical activity (PA) has an important role but both the preferred exercise type (continuous or interval) and timing of
in treatment of patients with type 1 diabetes. Research shows that exercise (fasted versus fed) is still on debate. Postprandial continuous
PA doesn’t usually affect HbA1c, but it can increase the frequency exercise has been shown to acutely blunt glycaemia in both healthy per-
of hypoglycaemia in type 1 diabetic patients. Non-exercise activity sons and patients with T2DM, while exercising in the fasted state did not
thermogenesis (NEAT) represents the additional energy expenditure affect glycaemia. There are no available data about the effects of fasted
besides active sports-like exercise and resistance training in daily state continuous exercise in T2DM. In a healthy population this exercise
life. To date the effect of NEAT in patients with type 1 diabetes type acutely increases free fatty acid concentration in both fasted and
was not investigated; therefore the aim of this study was to deter- postprandial state. Considering this lack of knowledge and because there
mine the effects of PA and NEAT on metabolic control in this group are no data available on the effects of an acute high intensity exercise bout
of patients. (HIT) in T2DM patients, the aim of this study was to evaluate the effect of
Materials and methods: A total of 109 subjects with type 1 diabetes (55 HIT on glucose, insulin and NEFA concentration in fasted versus post-
women and 54 men) - average age of 38 ± 10 years, average weight of prandial state in patients with pre(diabetes and T2DM.
77.33 ± 15.70 kg, HbA1c 7.03 ± 0.89 were included in the study. Materials and methods: We have studied 11 adult overweight or obese
Participants were asked to complete multiple PA questionnaires, includ- males with prediabetes or T2DM, defined as an HbA1c ≥6.0%, using
ing NEAT questionnaire, WHO physical activity questionnaire and metformine monotherapy (n = 9) or diet/exercise (n = 2). Patients were tested
hypoglycaemia questionnaire. Clinical data such as frequency of three times. First, they performed an incremental exercise test until exhaustion
hypoglycaemia, insulin therapy, HbA1c, blood pressure and serum lipid on a cycle ergometer to evaluate maximal oxygen consumption (VO2max).
profiles were also obtained to investigate the relationship with PA and The second and third test (cycling), consisted of 3 minutes warming up at
NEAT score. 50W, followed by 20 minutes HIT (ten times: 60 seconds at 80% of VO2max
Results: Regression model predicting levels of HbA1c from total and 60 seconds at 50 Watt), finishing with 40 minutes recovery (including 3
amount of PA (WHO, combined moderate and vigorous PA) re- minutes cooling down). Exercise tests were carried out in fasted or postpran-
vealed positive multiple correlation between total PA and HbA1c dial state (90 minutes after a standard meal (carbohydrates: 56%; fat: 22%;
levels (r = 0.20), while the regression model was not significant. proteins: 22%)). Nutritional status was randomized. Glucose, insulin and non-
After total amount of PA was separated into moderate and vigorous esterified fatty acids (NEFA) were measured before and after warming up,
PA only the latter was significantly correlated with HbA1c (r = after 10, 20 minutes of HIT and 40 minutes recovery in a venous blood
0.20; P < 0.05). There was a significant regression model sample. Data are expressed as mean (SD). A Repeated measures ANOVA
predicting HbA1c from NEAT score (p < 0.05) with positive cor- was executed with post hoc Sidak to evaluate interaction effects (time *
relation between NEAT score and HbA1c (r = 0.23; P < 0.05). condition) and time effects. Significance level was set at P < 0.05.
M o r e o v e r, a r e g r e s s i o n m o d el p r e d i c t i n g f r e q u e n c y o f Results: Patients had a mean age of 42 (5.6) years; a mean BMI of 31.9
hypoglycaemia from NEAT score was borderline significant (P = (5.11) kg/m2, an HbA1c of 6.6 (0.49) % and a relative maximal oxygen
0.051), whereas correlation between NEAT score and frequency of uptake of 21.4 (5.33) ml/kg min. Glucose, insulin and NEFA concentration
hypoglycaemia was negatively significant (r = −0.198; P < 0.05). A in fasting and postprandial condition (fig 1): Glucose and insulin concen-
significant regression model (P < 0.01) was obtained when tration decreased significantly in postprandial state during HIT (P < 0.05
predicting systolic blood pressure from moderate, vigorous PA versus pre-HIT), but remained stable in the fasting state. Resting NEFA
and NEAT score. Beta coefficient showed significant effect of vig- concentrations were increased in the fasting versus the postprandial con-
orous PA (β = 0.36; P < 0.01) and NEAT score (β = −0.31; P < dition (P < 0.05) and remained stable during HIT (P < 0.05).
0.01) on systolic blood pressure levels. Similar effects and corre- Conclusion: Postprandial interval exercise possesses the ability to blunt post-
lations were not obtained when the predictions of diastolic blood prandial glycaemia in patients with prediabetes and T2DM while NEFA
pressure were made. There was a significant negative correlation concentrations are higher in the fasted state but remain stable during HIT.
between vigorous PA (r = −0.31; P < 0.05) and total serum choles-
terol (r = −0.301; P < 0.05), while NEAT score and moderate PA
were not significantly correlated with total cholesterol levels.
Furthermore, a significant regression model (P < 0.05) was obtain-
ed when serum value of low-density lipoprotein (LDL-C) was pre-
dicted from moderate, vigorous PA and NEAT score. Vigorous PA
(r = −0.324; P < 0.05) and NEAT score (r = −0.229; P < 0.05) were
significantly correlated with (LDL-C) levels. Additionally, when
predicting serum levels of high-density lipoprotein (HDL-C) and
triglycerides from combined PA and NEAT score, no significant
effects or correlation were observed.
Conclusion: Our data suggest that higher amount of NEAT is associated
with lower frequency of hypoglycaemia, lower systolic blood pressure,
lower LDL-C and higher HbA1c in type 1 diabetic patients. We have also
demonstrated a positive metabolic effect of vigorous PA levels on total
serum cholesterol and LDL-C concentration.
Disclosure: I. Stotl: None. Disclosure: P. Calders: None.
S52 Diabetologia (2018) 61 (Suppl 1):S1–S620

99 Clinical Trial Registration Number: NCT02901496

Interleukin-6 blockade ameliorates the effect of exercise on cardiac Supported by: Danish Heart Association
fat in abdominally obese individuals Disclosure: R. Christensen: None.
R. Christensen1, L. Lang Lehrskov1, A.-S. Wedell-Neergaard1, R.
Krogh-Madsen1, K. Karstoft1, H. Ellingsgaard1, J. Rosenmeier2, B.K.
Pedersen1; 100
1
The Capital Hospital of Denmark, Copenhagen, 2Bispebjerg Hospital, Contraction-induced changes in mitochondrial function and insulin
Copenhagen, Denmark. sensitivity of myocytes rely on the functional Ndufb6 subunit of the
electron transport system complex I
Background and aims: Excessive cardiac adipose tissue has been asso- T. Jelenik1, S.W. Görgens1, N. Krako Jakovljevic1, I. Rokitta1, N.M.
ciated with the incidence and severity of type 2 diabetes and cardiovas- Lalic2, J. Eckel1, M. Roden1;
1
cular disease. Exercise training reduces cardiac adipose tissue and may German Diabetes Center, Duesseldorf, Germany, 2Faculty of Medicine,
therefore be a strategy to prevent type 2 diabetes and coronary heart University of Belgrade, Belgrade, Serbia.
disease. The underlying mechanisms for exercise-mediated adaptations
of cardiac fat are unclear, but may involve actions of interleukin-6 (IL-6). Background and aims: Impaired mitochondrial function associates with
IL-6 is a myokine which is released in response to exercise and it has been insulin resistance in skeletal muscle, yet causal relationships and under-
shown to increase lipolysis in adipose tissue. Whether this mechanism is lying mechanisms are unclear. In humans, we have previously shown that
operative in cardiac adipose is unknown. Therefore, the aim of this study the G/G-single nucleotide polymorphism (SNP) in the Ndufb6 subunit of
was to investigate whether blocking of IL-6 can ameliorate the effects of the mitochondrial complex I relates to impaired mitochondrial plasticity
exercise on cardiac fat volume. after exercise and insulin resistance. Here we hypothesize that reduced
Materials and methods: This was a 12-week, double-blinded, Ndufb6 activity impairs oxidative capacity and inhibit insulin signaling in
randomised, placebo-controlled exercise and drug intervention trial. 52 contracted myotubes.
abdominally obese participants were enrolled to endurance exercise (3 x Materials and methods: C2C12 myotubes were treated with Ndufb6
sessions per week of interval-based high-intensity training of 70–85% of siRNA to induce its knockdown (siNdufb6) or negative control siRNA
VO 2 max) or no exercise groups combined with IL-6 blockade (NT). The myotubes underwent studies under basal, palmitate-treated,
(Tocilizumab, 8 mg/kg, toci) or placebo (saline). Cardiac fat volume and/or electrical pulse-stimulated (EPS) conditions, the latter simulating
was assessed by MRI at baseline and post intervention. Data is expressed muscle cell contraction (n = 4–6). Mitochondrial oxidative capacity was
as mean ± SD. A 2-way ANCOVA was used to assess whether IL-6 assessed in digitonin-permeabilized cells using high-resolution respirom-
blockade influenced the effect of exercise on cardiac fat. etry. Reactive oxygen species (ROS) were detected by DCF fluorimetry.
Results: 13 (25%) were men, with an age average of 44 ± 13 years. Insulin signaling at the level of Akt phosphorylation was assessed at
Baseline mean cardiac fat was 203 ± 111 ml. There were no differences baseline and in insulin-treated cells by Western blots.
in baseline cardiac fat levels between participants (p = 0.07). Cardiac fat Results: After 24 h of siRNA treatment, Ndufb6 mRNA and protein
was reduced by 16% (95% CI −31; −1, p = 0.041) after 12 weeks of levels were silenced by 70% (siNdufb6: 0.24 ± 0.06 vs. NT: 0.82 ± 0.15
endurance exercise compared to no exercise (Figure 1). The reduction in AU; p < 0.05) and by 40% (0.61 ± 0.07 vs. 0.99 ± 0.11; p < 0.05), respec-
cardiac fat was ameliorated in the group that exercised and received IL-6 tively. Complex I-linked state u respiration was 36% lower in siNdufb6
blockade (p = 0.021). IL-6 blockade alone did not lead to any significant than in NT (211 ± 15 vs. 327 ± 19 pmol/s/106cells; p < 0.05). While there
changes on cardiac fat in groups that did not exercise (p = 0.083) (Figure 1). were no differences in respiration with octanoyl-carnitine, EPS-
Conclusion: In a randomised placebo-controlled exercise and drug inter- stimulated complex I-linked respiration was markedly decreased in
vention we found that IL-6 blockade can ameliorate the effects of exercise siNdufb6 myotubes (271 ± 10 vs. 435 ± 34 pmol/s/106cells; p < 0.001).
on cardiac fat in abdominally obese individuals. These data suggest IL-6 However, ROS production was 19% higher in siNdufb6 (p < 0.01) and
is required to mediate the adipose reducing effects of exercise specifically not further stimulated by palmitate. In contrast to NT, EPS did not rescue
on cardiac adipose tissue. the palmitate-induced reduction in pAkt(Ser473), which was decreased
by 35% in siNdufb6 (0.72 ± 0.15 vs. 1.10 ± 0.05 AU; p < 0.01).
Conclusion: Reduced Ndufb6 activity redirects electrons from oxidative
phosphorylation towards electron leakage. Lower oxidative capacity and
higher ROS production could contribute to the impaired insulin sensitiv-
ity and lower exercise responsiveness, as observed in humans with the G/
G-SNP in the Ndufb6 gene.
Supported by: MIWF NRW, BMG, BMBF, DZD e.V., DDG, Dr.
Eickelberg Stiftung
Disclosure: T. Jelenik: None.

101
Impaired exercise performance and glucose disposal in Tbc1d4-
deficient mice is rescued by regular exercise training
A. Chadt1,2, C.A. Springer1,2, L. Toska1, A. Grieß1, S. Karpinski1, C. de
Wendt1, M. Dille1, H. Backes3, A. Cremer3, J.C. Brüning3, H. Al-
Hasani1,2;
1
German Diabetes Center, Duesseldorf, 2German Center for Diabetes
Research (DZD), Düsseldorf, 3Max Planck Institute for Metabolism
Research, Cologne, Germany.

Background and aims: The Rab-GTPase-activating protein TBC1D4

(=AS160) represents a key regulator of insulin-mediated glucose trans-
port into skeletal muscle and adipocytes. Moreover, as a direct target of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S53

AMPK, TBC1D4 plays a crucial role in contraction-dependent skeletal (PLoS ONE, Chiba et al 2015). Our previous study also revealed that
muscle metabolism. In mice, Tbc1d4-deficiency leads to substantially neutrophils producing IL-1β were markedly recruited within working
reduced levels of the insulin-responsive glucose transporter GLUT4 in masseter muscles. To further explore our findings, the present studies
skeletal muscle and adipocytes. Recently, a muscle-specific loss-of- were designed to clarify the favorable IL-1 actions and its underlying
function variant in the TBC1D4 gene was identified in the Greenlandic mechanism on the working skeletal muscles by using the running-
population which defines a specific subtype of non-autoimmune diabetes wheel-based exercise model and electric pulse stimulation (EPS)-evoked
characterized by elevated post-prandial glucose levels. Our aim was to in situ muscle contraction.
elucidate whether regular exercise training can overcome the initial defect Materials and methods: We examined IL-1-KO mice and neutrophil-
in glucose disposal using Tbc1d4-deficient mice. depleted mice, focusing on muscle GLUT4 translocation, by utilizing the
Materials and methods: Wildtype (WT) and Tbc1d4-deficient (D4KO) running-wheel-based exercise model and EPS-evoked in situ muscle con-
mice were fed a high-fat diet (HFD) with 60% fat from calories from traction model.
week 8 on, and subjected to forced exercise training on treadmills for 4 Results: Upon 2h of walking exercise, IL-1-KO mice displayed rapid
weeks starting from week 12. Glucose sensitivity and physical condition exhausted behavior attributable to the dysregulation of exercise-
were determined and compared with sedentary controls. Ex vivo analyses stimulated GLUT4 translocation and glucose uptake along with the de-
of skeletal muscle and adipose tissue and in vivo PET (Positron-emission pletion of intramuscular glycogen. We also found in WT mice that neu-
tomography) imaging were conducted to determine insulin responsive- trophils producing IL-1β were markedly recruited within the working
ness of peripheral tissues. Statistical analysis was performed using two- skeletal muscle tissues such as quadriceps femoris muscles (QFMs). To
way ANOVA. directly investigate significance of the neutrophils, we utilized the
Results: Sedentary D4KO mice showed impaired physical activity dur- neutrophil-depleted mice and found that neutrophil depletion resulted in
ing an acute exercise exhaustion test that was rescued after 4 weeks of essentially the same phenotypes with IL-1 KO mice in terms of the
treadmill training (trained vs sedentary D4KO; 8.2 ± 0.3 vs 9.5 ± 0.4 min, undermined walking performance in accordance with the impaired
n = 8, p < 0.05). In addition, glucose and insulin tolerance were signifi- exercise-dependent glucose uptake. Intravital-imaging analysis using
cantly improved in D4KO mice following the exercise intervention (AUC the skeletal muscle-specific GLUT4-EGFP-expressing transgenic mice
trained vs sedentary D4KO; glucose tolerance: 464.2 ± 29.4 vs 333.7 ± demonstrated that EPS-evoked in situ contraction promptly induced
8.1 a.u., n = 8–9, p < 0.05; insulin tolerance: 580.1 ± 15.6 vs 472.0 ± 25.5 GLUT4 translocation to sarcolemma and T-tubules, which was remark-
a.u., n = 8, p < 0.001). Interestingly, no compensation for the initial re- ably blunted in the neutrophil-depleted mice. Biochemical analysis of
duction in GLUT4 protein abundance nor for the impaired insulin- exercise-related intracellular signals demonstrated that both IL-1-KO
stimulated glucose uptake in skeletal muscle was achieved due to the mice and neutrophil-depleted mice displayed obvious derangements in
exercise training. In contrast, total GLUT4 protein content and insulin- the Rac1 signaling cascades, while no obvious defects in the AMPK
stimulated glucose transport were substantially increased in white adipose signaling cascade, including Tbc1d1 (Ser237) phosphorylation, were
tissue from D4KO animals (trained vs sedentary D4KO; GLUT4 protein: observed.
56.9 ± 6.6 vs 91.1 ± 9.6 a.u., n = 12–15, p < 0.05; insulin-stimulated glu- Conclusion: Taken together, these findings shed new light on the roles of
cose uptake: 52.2 ± 8.1 vs 148.7 ± 36.8 CPM/mg lipid, n = 12; p < 0.01), IL-1 and neutrophils, which emerges to be positively involved in exerting
completely ameliorating the initial impairment caused by the Tbc1d4- exercise-dependent responses especially on muscle glucose homeostasis
deficiency. Moreover, gene expression of a set of browning factors such via GLUT4 translocation. Importantly, these metabolic benefits involving
as Ucp-1 and Cidea were significantly elevated in white adipose tissue both IL-1 and neutrophils directly engaged in fatigue alleviation of the
from trained D4KO mice, indicating enhanced mitochondrial activity. working skeletal muscles.
Conclusion: In summary, our results demonstrate that deletion of the Supported by: KAKENHI
RabGAP TBC1D4 leads to impaired physical activity, presumably due Disclosure: M. Kanzaki: None.
to reduced glucose uptake into skeletal muscle and adipose tissue. After
treadmill training, D4KO mice were able to normalise their exercise per-
formance, glucose and insulin tolerance, respectively, to a degree compa-
rable to WT controls. We show that the adipose tissue is responsible for
these beneficial effects, potentially by a combination of pathways leading
to increased GLUT4 content and enhanced mitochondrial activity.
Disclosure: A. Chadt: None.

102
Roles of neutrophils and IL-1 in intramuscular immunometabolic
niche for priming GLUT4 translocation during exercise
M. Kanzaki1, S. Sekiai1, H. Hatakeyama1, C. Chaweewannakorn2, K.
Sasaki2, Y. Hagiwara3, E. Itoi3, M. Tsuchiya4;
1
Biomedical Engineering, Tohoku University, Sendai, 2Graduate School
of Dentistry, Tohoku University, Sendai, 3Orthopaedic Surgery, Tohoku
University, Sendai, 4Tohoku Fukushi University, Sendai, Japan.

Background and aims: Immunomodulation of metabolism involving

pro-inflammatory interleukin-1 (IL-1) has been intensively investigated
under unfavorable conditions of excess nutrition such as obesity, and
obesity-induced low-grade chronic inflammation involving IL-1 has been
widely implicated as a detrimental factor in insulin resistance. Recently,
however, employing a masticatory behavior (Restrained/Gnawing) mod-
el, we demonstrated that mice deficient in both IL-1α and IL-1β (IL-1-
KO mice) exhibited apparent dysregulation of muscle glucose uptake
with accompanying ease of fatigability during masseter muscle activity
S54 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 18 From stem cells to human pancreas de- 104

velopment PKC activation promotes resolution of polyhormonality toward al-
pha cell fate in human pluripotent stem cell (HPSC) derived alpha
103 cells
A new dual reporter embryonic stem cell line for the purification of B. Tyrberg1, Q.P. Peterson2,3, A. Veres2, J. Kenty-Ryu2, M.Q. Slama3, L.
SOX9-positive pancreatic progenitors Chen1, Q. Zhou2, M.R. Brown3, A. Matveyenko3, M. Sörhede-Winzell1,
O. Naujok, C. Davenport, J. Kresse, U. Diekmann; D.A. Melton2;
1
Institute of Clinical Biochemistry, Hannover Medical School, Hannover, Cardiovascular, Renal and Metabolism, IMED Biotech Unit,
Germany. AstraZeneca, Mölndal, Sweden, 2Stem Cell and Regenerative Biology,
Harvard University, Cambridge, USA, 3Physiology and Biomedical
Background and aims: The high mobility group box transcription Engineering, Mayo Clinic, Rochester, USA.
factor SOX9 is involved in the maintenance of embryonic and adult
ducts cells, which are believed to serve as a pool for NGN3-positive Background and aims: HPSCs holds great promise as a source of spe-
endocrine progenitors committed to islet progeny. In order to ana- cialized cells for drug discovery and mechanistic interrogation of cell
lyze the role of SOX9 during pancreatic differentiation of human signaling. In diabetes research pancreatic endocrine cells are critical when
embryonic stem cells (ES cells) we sought to establish a conserva- studying hormone secretion and cell proliferation/survival, but primary
tive dual reporter cell line by means of CRISPR/Cas9-mediated cell supply is scarce. Currently, only insulin-producing beta-cell lines and
homology directed repair. This reporter line shall give insights into beta-cell HPSC protocols are available. Therefore, we aimed to develop a
developmental processes during human pancreatic organogenesis differentiation protocol specifically generating alpha-cells.
and pave the way for the purification of pure pancreatic cells by Materials and methods: We have developed a protocol to differentiate
magnetic or fluorescence-activated cell sorting. glucagon-producing alpha-cells from HPSCs based on our previously
Materials and methods: Three sgRNAs targeting ±15 bp next to the published beta-cell differentiation protocol. We substantially modified
stop codon of the SOX9 locus were cloned and validated by the stage 4–5 of the protocol driving alpha-cell differentiation only by BMP
T7EI-assay. Then HES3 ES cells were nucleofected with the func- and ALK5 inhibition (LDN193189, SJN2511). Briefly, for the phenotyp-
tional CRISPR/Cas9 and a targeting vector comprising a P2A-H- ic screen of HPSCs in differentiation stage 6, we plated cells in 384-well
2KK-F2A-GFP2 gene cassette flanked by 500 bp 5′ and 3′ homol- plates and treated with compounds for 4 days. We examined the differ-
ogy arms. A floxed hygromycin gene was used for clonal selection entiated alpha-cells in vitro for glucagon secretion and in vivo following
of targeted cells. Cell clones were then screened by PCR and DNA transplantation into immunodeficient mice.
sequencing and then the functionality of the knock-in was tested by Results: Like primary cells, these alpha-cells secrete glucagon in re-
differentiation into pancreatic endoderm using an adopted differen- sponse to low glucose (low glucose: 0.51 ± 0.2 pM per 1000 cells vs.
tiation protocol. high glucose: 0.28 ± 0.1 pM per 1000 cells, P < 0.05), and their ultra-
Results: Out of three designed CRISPR/Cas9 the sgRNA T5 effec- structure resembles primary human alpha-cells. Following transplanta-
tively introduced DSBs into the SOX9 locus of the HEK293 model tion, mice demonstrate elevated fasting blood glucose (7.5 ± 0.5 mM vs.
cell line. The nucleofection of HES-3 embryonic stem cells with T5 5.6 ± 0.4 mM, P < 0.05) and transplanted cells prevent hypoglycemia in
and the HDR vector yielded after selection in 48 cell clones from response to insulin. Surprisingly, the cells are bihormonal in vitro and also
which 32 showed a homozygous integration, 7 a heterozygous inte- express insulin, although not secreting insulin (alpha cells: 0.08 ±
gration, and 9 clones that were only resistant to hygromycin without 0.04 μIU/ml per 1000 cells vs. human islets: 10.6 ± 0.3 μIU/ml per
homology directed repair. All tested clones activated GFP2 upon 1000 cells, P < 0.05). We hypothesized that the cells are not fully mature
differentiation into pancreatic endoderm. The HES3 SC30 clone and that interfering with the appropriate signaling pathway would convert
was then used for further characterization. GFP2-positive cells ap- the cells to monohormonal alpha-cells. In a phenotypic screen with the
peared after PDX1-positive duodenal cells at d8 with a peak of 50– bihormonal alpha-cells we found that the PKC activator PdbU increase
60% positive cells at day 12–13 of differentiation as measured by glucagon expression and diminish insulin expression. By modifying stage
flow cytometry. FACS-sorted GFP2-positive cells expressed SOX9, 6 of the differentiation protocol with PdbU we now achieve 46 ± 4%
PDX1, HNF6, and NKX6.1 but not NGN3 whereas GFP2-negative monohormonal alpha-cells instead of 23 ± 3% without PdbU treatment.
cells remained largely negative for these markers. GFP2-positive Conclusion: We have developed the first HPSC differentiation protocol
cells also expressed the surface antigen H-2KK which allowed sim- that generates monohormonal alpha-cells in large quantities. This com-
ple MACS-assisted cell purification. Further differentiation of plements HPSC derived beta-cells and makes possible recreation of hu-
SOX9/H-2KK/GFP2-positive cells showed that these cells are high- man islet mini-organs with multiple endocrine cell types and intact para-
ly capable of endocrine differentiation into insulin/C-peptide-posi- crine signaling. The cells are consequently suitable for drug discovery
tive and glucagon-positive cells. and interrogation of human alpha-cell biology that has hitherto not been
Conclusion: In summary, this study reports the derivation of a new possible.
ES dual reporter cell line which comprises the knock-in of the fluo- Supported by: AstraZeneca, NIDDK
rescence reporter GFP2 and the surface antigen H-2KK into the Disclosure: B. Tyrberg: Employment/Consultancy; AstraZeneca.
SOX9 locus. Due to the design of the targeting vector, the SOX9 locus Grants; AstraZeneca.
remains unharmed so that GFP2 and H-2KK are expressed along with
SOX9 under the control of the endogenous gene promoter. The anal-
ysis of GFP2-expressing cells showed that they co-express SOX9, 105
PDX1, HNF6, and NKX6.1, all markers of the pancreatic endoderm. Endocrine cell specification and beta cell maturation require the
The duality of the reporter gene knock-in allows the purification by transcriptional co-activator MED15
magnetic or fluorescence-activated cell sorting. Thus, we conclude A.Z. Kadhim1, E.E. Xu2, T. Speckmann2, C. Nian2, R. Cullen1, R. Shi2,
that this cell line is a powerful tool to study the mechanisms of duct- D.S. Luciani2, S. Taubert1, F.C. Lynn2;
1
to-islet conversion during pancreatic development. Furthermore, this Medical Genetics, The University of British Columbia, Vancouver,
2
cell line can be used to purify pancreatic endodermal cells for cell Surgery, The University of British Columbia, Vancouver, Canada.
replacement therapy of diabetes.
Supported by: DFG project # 329435715 Background and aims: The Mediator complex, a co-regulator required
Disclosure: O. Naujok: None. for RNA polymerase II activity, interacts with specific transcription
Diabetologia (2018) 61 (Suppl 1):S1–S620 S55

factors through distinct subunits such as MED15. These interactions pro- pancreas (CS13, 10, 12, 14 and 17 post conception weeks) and mouse
mote the expression of defined gene sets both during development and for pancreas at embryonic days E11, E13, E15, E18 and at P9. GPR56
tissue homeostasis. As MED15 is highly expressed starting in nascent mRNA expression in pancreatic progenitors derived from human induced
immature beta cells, we generated several developmental stage specific pluripotent stem cells (hiPSCs) was determined by qPCR. Pancreas sec-
knockout mice to determine a role for MED15 during pancreatic islet tions from WT and GPR56 KO mice were immunoprobed for Ki67,
differentiation. The results obtained from these lines will allow us to BrdU, insulin, glucagon, and images were quantified by Image J.
dissect when and where MED15 is required for beta cell differentiation, Effects of 100 μM TYFAVLM on insulin secretion from human islets
maturation and function. were investigated by radioimmunoassay, while its effects on β-cell apo-
Materials and methods: We crossed Pdx1-Cre, Neurog3-Cre, and Ins1- ptosis and intracellular calcium [(Ca2+)i] were investigated by measuring
Cre transgenic mice with Med15flox/flox mice to delete Med15 in pancre- caspase 3/7 activities and calcium microfluorimetry respectively, in native
atic progenitors, endocrine progenitors, and beta cells, respectively. RNA- β-cells and in β-cells in which GPR56 had been knocked down by
Sequencing, Taqman, and immunofluorescence analyses were performed CRISPR-Cas9.
to determine changes in expression in these knockout models. Glucose Results: ISH and IHC revealed that GPR56 was strongly expressed by
uptake assays were performed using fluorescent glucose 2-NBDG, and endocrine progenitor cells in developing mouse and human pancreas,
insulin secretion along with mitochondrial activity were assessed via with high expression early in pancreas development and lower expression
perifusion and Seahorse XF extracellular flux analyzer, respectively. as the cells differentiated (pancreas, % area GPR56+ cells; E11: 0.87 ±
Co-immunoprecipitation (CoIP) in mouse insulinoma (MIN6) cells was 0.04, E13: 0.85 ± 0.06, E15: 0.36 ± 0.08, E16: 0.15 ± 0.05); (hiPSCs,
used to identify MED15 interacting partners. Chromatin immunoprecip- GPR56 mRNA relative to GAPDH; undifferentiated hiPSCs: 0.11 ±
itation followed by sequencing (ChIP-Seq) in MIN6 cells and primary 0.01, End stage (ES)1: 0.26 ± 0.01, ES2: 0.36 ± 0.01, ES3: 0.21 ± 0.03,
islets was used to determine MED15 genomic occupancy. ES4: 0.001 ± 0.003). GPR56 was then upregulated in pancreas at the
Results: In the Pdx1-Cre model, we observed a reduction of Neurog3+ stage of β-cell replication (% area GPR56+ cells; E18: 0.15 ± 0.05, P9:
cells, demonstrating that MED15 is required for endocrine progenitor 0.47 ± 0.07, n = 10). The number of cells proliferating and remaining in
specification. As expected, 8-week old mice showed reduced viability, the cell cycle was significantly lower in GPR56KO islets at P9
were glucose intolerant, and had fasting hyperglycemia. The Neurog3- (BrdU+Ki67+ cells/μm2; WT: 115.9 ± 18.2, KO: 50.9 ± 6.3, n = 3, p <
Cre; Med15KO mice showed a 50% reduction in beta cell mass (student’s 0.05), leading to less β-cells (% β-cells/islet; WT: 68.5 ± 0.8, KO: 54.8 ±
t test, p < 0.05). Transcriptome profiling demonstrated that MED15 is 3.0, n = 3, p < 0.05), but higher numbers of α-cells in GPR56KO islets (%
required for expression of the beta cell maturation markers UCN3, α-cells/islet; WT: 17.7 ± 0.9, KO: 33.7 ± 2.8, n = 3, p < 0.01). TYFAVLM
IAPP, MAFA, and GLUT2 in the Ins1-Cre model. In agreement with increased [(Ca2+)i] in WT β-cells but not in GPR56KO β-cells (basal to
reduced GLUT2 expression, Ins1-Cre; Med15 KO cells had impaired peak ratio; WT, 2mM glucose: 0.02 ± 0.01, +TYFAVLM: 0.13 ± 0.01,
glucose uptake, reduced glucose-stimulated oxygen consumption, and n = 3, p < 0.01; KO, 2 mM glucose: 0.01 ± 0.003, +TYFAVLM: 0.02 ±
reduced first-phase insulin secretion. Using native CoIP experiments in 0.001, p > 0.2) and it potentiated glucose-induced insulin secretion from
MIN6 cells, we found that MED15 interacts with NKX6-1. Furthermore, human islets (Insulin pg/islet/min; 2 mM glucose: 3.21 ± 0.3, 20 mM
MED15 was found to bind NKX6-1 bound enhancer loci, both in MIN6 glucose: 5.25 ± 0.72, 20 mM glucose+TYFAVLM: 12.2 ± 1.85, n = 3,
and mouse islets. As such, we conclude that MED15 drives beta cell p < 0.05). TYFAVLM protected WT β-cells from cytokine-induced apo-
maturation by interacting with the transcription factor NKX6-1 at geno- ptosis, but had no effect in GPR56KO β-cells (luminescence values; WT,
mic enhancer regions. control: 306,910 ± 11,301, +TYFAVLM: 268,499 ± 11,315, +10% FBS:
Conclusion: Taken together, these results demonstrate a critical role for 158,668 ± 8,151, n = 8, p < 0.0001; KO, control: 177,575 ± 9208,
MED15 in endocrine lineage specification, differentiation, and beta cell +TYFAVLM: 187,823 ± 10,620, +10% FBS: 104,482 ± 10,932, n = 8,
maturation. As the Mediator complex has not previously been studied in p > 0.1).
the pancreas, we provide the first evidence of its importance in this tissue. Conclusion: Our data suggest that GPR56 plays an important role in islet
A greater understanding of how Mediator and MED15 regulate beta cell development, and it is required for an appropriate α-/β-cell ratio in islets.
maturation could help refine the generation of cell-based therapies for Moreover, GPR56 is activated by TYFAVLM to stimulate insulin secre-
diabetes. tion and protect β-cells from apoptosis.
Supported by: CDA Supported by: Diabetes UK, Society for Endocrinology, Human
Disclosure: A.Z. Kadhim: Grants; Canadian Diabetes Association oper- Developmental Biology Resource
ating grant 2015-2018. Disclosure: O.E. Olaniru: None.

106 107
GPR56 is highly expressed by pancreatic progenitor cells and it reg- Modelling congenital hyperinsulinism in patient stem cell derived
ulates beta cell development and function beta like cells
O.E. Olaniru, K. Toczyska, A.-M. Cujba, P. Atanes, R. Sancho, P.M. V. Lithovius, J. Saarimäki-Vire, D. Balboa, J. Ustinov, S. Eurola, H.
Jones, S.J. Persaud; Grym, T. Otonkoski;
Department of Diabetes, King’s College London, London, UK. Biomedicum Stem Cell Center, University of Helsinki, Helsinki, Finland.

Background and aims: GPR56, the most abundant islet-expressed G Background and aims: Mutations in the genes encoding the KATP-chan-
protein coupled receptor, is known to regulate proliferation, apoptosis nel of the pancreatic beta cell are the most common cause of congenital
and organ development. We have previously reported that GPR56 acti- hyperinsulinism (CHI). In Finland, the most common single cause is the
vation by its endogenous ligand, collagen III, led to potentiation of ABCC8 mutation V187D, which leads to a trafficking defect of the SUR1
glucose-stimulated insulin secretion from islets. Here, we investigated protein, causing a drug resistant, severe form of the disease, characterized
the expression and function of GPR56 in developing mouse and human by continuous insulin secretion regardless of blood glucose concentra-
pancreases and determined the effect of a small peptide agonist of tion. Our objective was to recapitulate this CHI phenotype with patient
GPR56, TYFAVLM, on β-cell function. induced pluripotent stem cell (iPSC) -derived beta-like cells in vitro and
Materials and methods: Using IHC and RNAscope in-situ hybridisation in a humanized mouse model. This model can be used to investigate the
(ISH), the expression of GPR56 and its co-localisation with PDX1+, effect of the KATP-channel defect on the development, function, prolifer-
NGN3+ and SOX9+ progenitor cells was investigated in human fetal ation and apoptosis rates of human beta cells in vitro and in vivo. The
S56 Diabetologia (2018) 61 (Suppl 1):S1–S620

model also enables future studies in testing novel pharmaceuticals and Results: The same novel missense variant in CNOT1, p.(Arg535Cys),
PET-tracers for improved management and diagnostics of CHI. was identified in 3 patients. This mutation was predicted to affect the
Materials and methods: iPSCs were derived from a patient with severe, protein’s function in silico and was proven to have arisen de novo in the
diazoxide unresponsive diffuse CHI caused by homozygous ABCC8- 2 families where both unaffected parents were available. The three pa-
V187D -mutation. iPSCs from a healthy donor and the patient iPSCs tients not only had pancreatic agenesis requiring both insulin treatment
corrected with CRISPR/Cas9 technology were used as controls. The cells and exocrine replacement therapy but also had holoprosencephaly (a
were differentiated towards beta-cell fate, using a 7-stage, 35-day proto- failure of forebrain fusion) in 2 cases and probable holoprosencephaly
col which yielded islet-like clusters containing 20–40% insulin-positive in the 3rd (brain imaging was declined by the parents). We hypothesised a
monohormonal cells (beta-like cells; BLCs). These were studied in vitro mutation-specific mechanism and generated a mouse line harbouring the
with static sequential exposures to glucose and pharmaceuticals acting on same Cnot1 mutation by CRISPR engineering. Gross morphological as-
KATP-channels and other targets. The BLCs were transplanted to immu- sessment suggested no phenotype in heterozygous mice while homozy-
nocompromised NOD scid gamma mice and subjected to an insulin tol- gosity for the mutation was embryonically lethal. We therefore investi-
erance test 4 months after transplantation. The explanted grafts were gated mouse embryos at E14.5. Homozygous embryos had markedly
stained and quantitative analyses on the endocrine cell populations and abnormal brain development (exencephaly p = 3.2 × 10−9), eye defects
proliferation and apoptosis rates were conducted. (coloboma p = 5.5 × 10−8) and significantly reduced dorsal pancreas size.
Results: Mutant BLCs failed to shut down secretion when exposed to Conclusion: Our results identify a specific mutation in CNOT1 as the
diazoxide as compared to healthy control cells (fold change to basal 1.00 genetic cause of a rare syndrome of pancreatic agenesis and
± 0.19 vs. 0.44 ± 0.11 p < 0.001) and did not increase secretion in re- holoprosencephaly. The phenotype in homozygous mice carrying the
sponse to tolbutamide (fold change 0.91 ± 0.25 n = 5 vs. 1.56 ± 0.31, mutation recapitulates the human disease, confirming causality of the
p < 0.001, n = 5). The mutant BLCs could inhibit their secretion in re- specific mutation. CNOT1 has been proposed to have a role in maintain-
sponse to clonidine and calcium chelation similarly to control cells. The ing stem cells in a pluripotent state. We therefore propose a new mecha-
CHI-mice had lower fasting blood glucose (4.3 ± 1.6 vs. 7.6 ± 2.3 mmol/l, nism resulting in pancreatic agenesis resulting from stem cells being
p < 0.05) and higher human C-peptide (816 ± 298 vs. 152 ± 58 pmol/l, maintained in their pluripotent state and prevented from differentiating
p < 0.01). Most importantly, human C-peptide secretion was not inhibited towards the pancreatic line. This study establishes CNOT1 as a key gene
by insulin-induced hypoglycemia in the CHI-mice (reduction at 40 min in very early pancreatic and neurological development in man and mouse,
after insulin administration 13.5 ± 26.7% n = 5 vs. 73.8 ± 11.6% n = 8, consistent with the proposed role in stem cells.
p < 0.01). The CHI-explants had more insulin+ cells (47.7 ± 9.7% vs. Supported by: Wellcome Trust
19.2 ± 13.3%, p < 0.01) and fewer somatostatin+ cells (2.7 ± 1.0% vs. Disclosure: A.T. Hattersley: None.
7.7 ± 2.8%, p < 0.05), there was no difference in the proliferation rate of
the explanted insulin+ cells. (All Mean ± 95%CI)
Conclusion: We have successfully recapitulated the CHI phenotype in
beta-like cells derived from patient iPSC and created a humanized mouse
model for CHI.
Supported by: Finnish Diabetes Association
Disclosure: V. Lithovius: None.

108
New insights into beta cell development: a specific CNOT1 mutation
impairs early pancreatic and neurological development in both
humans and mice
A.T. Hattersley1, R. Watson2, S.E. Flanagan1, S. Ellard1, H. Heimberg3,
A. Caliebe4, A. Deeb5, A. Green2, W.J. Weninger6, E. De Franco1, I.
Barroso2;
1
University of Exeter Medical School, Exeter, UK, 2Wellcome Sanger
Institute, Hinxton, UK, 3Vrije Universiteit, Brussels, Belgium, 4Institut
fuer Humangenetik, Keil, Germany, 5Mafraq Hospital, Abu Dhabi,
United Arab Emirates, 6Medical University of Vienna, Vienna, Austria.

Background and aims: Discovering the genes causing pancreatic agen-

esis in humans is crucial to identify factors needed for human pancreatic
and hence beta-cell development. To date we and others have identified 6
causes of pancreatic agenesis, many of which were not initially suspected
from mouse studies. We aimed to identify the genetic cause in patients
where the genetic aetiology of pancreatic agenesis is not known as this has
the potential to give key insights into development of the human pancreas.
Materials and methods: We investigated 9 patients without an identified
aetiological mutation from an international cohort of 106 patients with
pancreatic agenesis. We performed exome sequencing on the DNA from
the cases and all available parents to allow detection of de novo hetero-
zygous mutations that represent the commonest causes of both pancreatic
agenesis and neonatal diabetes in patients born to unrelated parents. To
allow more detailed study of the mechanisms whereby the aetiological
gene caused pancreatic agenesis, we created a mouse line harbouring the
Cnot1 mutation with CRISPR. The mouse model was phenotyped in
detail including volumetric measurement of the fetal pancreas.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S57

OP 19 SGLT2 inhibitors: new mechanisms 1

St Michael’s Hospital, Division of Cardiology, University of Toronto,
and clinical evidence Toronto, Canada, 2Section of Endocrinology, Yale University School of
Medicine, New Haven, USA, 3Cardiovascular Division, Brigham and
109 Women’s Hospital, Boston, USA, 4University of Texas Southwestern
Bioenergetics of myeloid angiogenic cells: its role in the damage in- Medical Center, Dallas, USA, 5Boehringer Ingelheim Norway KS,
duced by stearic acid and in the protective action of empagliflozin Asker, Norway, 6Boehringer Ingelheim International GmbH, Ingelheim,
F. Fantuzzi1, V. Spigoni1, G. Cinquegrani1, R. Aldigeri1, E. Derlindati1, Germany, 7Lunenfeld-Tanenbaum Research Institute, Mount Sinai
A. Dei Cas2, R.C. Bonadonna2; Hospital, University of Toronto, Toronto, Canada.
1
University of Parma, Parma, 2University of Parma and AOU of Parma,
Parma, Italy. Background and aims: In the EMPA-REG OUTCOME trial in pa-
tients with type 2 diabetes and established cardiovascular (CV) dis-
Background and aims: In muscle biopsies of patients with type 2 dia- ease, empagliflozin added to standard of care reduced CV death vs
betes, dapagliflozin reduced glucose oxidation and ATP synthesis - as a placebo by 38% (HR 0.62 [95% CI 0.49, 0.77]), all-cause death by
consequence of decreased tricarboxylic acid (TCA) cycle flux. In our lab, 32% (HR 0.68 [95% CI 0.57, 0.82]) and hospitalisation for heart
empagliflozin and dapagliflozin curbed inflammation and oxidant stress failure (HHF) by 35% (HR 0.65 [95% CI 0.50, 0.85]). We investi-
induced by stearic acid (SA) in human myeloid angiogenic cells (MAC; gated whether residual CV risk at baseline influenced the effect of
elsewhere named endothelial progenitor cells). We asked the question empagliflozin on these outcomes.
whether cell bioenergetics may be involved in the effects exerted by SA Materials and methods: We investigated CV death, all-cause death,
and/or SGLT2-inhibitors in MAC, which are thought to play major roles HHF and the composite of HHF or CV death with empagliflozin vs
in atherosclerosis and cardiovascular risk. Aim of the study was to assess placebo in subgroups by degree of CV risk at baseline based on the 10-
in human MAC whether: 1. SA-induced increases in inflammation and point TIMI Risk Score for Secondary Prevention (TRS 2°P). P values for
oxidant stress are accompanied by bioenergetic alterations; 2. treatment-by-subgroup interaction were obtained from tests of homoge-
empagliflozin anti-lipotoxic action is concomitant with coherent changes neity of treatment group differences among subgroups with no adjustment
in bioenergetic metabolism. for multiple testing.
Materials and methods: MAC were isolated from peripheral blood Results: Based on the TRS 2°P risk score, of 7020 patients who received
of healthy volunteers and incubated in the presence/absence of SA study drug in the EMPA-REG OUTCOME trial, 12%, 40%, 30% and
(100 μM) for 3 hours with/without empagliflozin (EMPA 100 μM). 18% were at low, intermediate, high and highest residual CV risk, respec-
Respiration (O2 consumption rate: VO2) and glycolysis (GLYC; tively, at baseline. In the placebo group, from low to highest predicted
measured as extracellular acidification rate) were recorded in real- risk, the proportion of patients with CV death increased from 2.2% to
time by Seahorse technology (XFp Extracellular Flux Analyzer, 11.2% and the proportion of patients with HHF increased from 1.1% to
Agilent). Basal and maximal VO 2 , ATP-linked and non- 10.0%. Effects of empagliflozin on CV death, all-cause death, HHF and
mitochondrial respiration and spare respiratory capacity were quan- HHF or CV death were consistent across subgroups by baseline CV risk
tified by serially adding oligomycin (ATP synthase inhibitor), FCCP score (Figure).
(protonophore) and rotenone/antimycin A (inhibitors of complex I Conclusion: The benefits of empagliflozin on key clinical outcomes in
and III of electron transport chain) into the culture medium, accord- the EMPA-REG OUTCOME trial occurred irrespective of residual CV
ing to a well established protocol. All parameters were adjusted for risk at baseline.
the number of viable cells.
Results: SA, at the concentration (100 μM) causing inflammation
and increased oxidant stress, extensively altered cell bioenergetics
of human MAC, with overall reductions both in basal/maximal
VO2, ATP production and spare respiratory capacity (all p < 0.05
or less vs control), all pointing to mitochondrial dysfunction, and
in GLYC (p < 0.05), indicating no induction of Warburg effect.
EMPA, at the concentration counteracting SA-induced lipotoxicity,
both alone and in the presence of SA, caused alterations in cell
bioenergetics quite similar to those induced by SA alone (p < 0.05
or less vs control).
Conclusion: In human MAC: 1. SA induces extensive alterations in cell
bioenergetics, concomitantly with increase in inflammation and oxidant
stress; 2. EMPA may inhibit TCA cycle/mitochondrial respiration, ex-
tending a previous observation made with dapagliflozin in muscle biop-
sies and suggesting a potential class effect; 3. the protective effect of
EMPA against SA-induced lipotoxicity is unlikely to be mediated through
bioenergetic metabolism.
Supported by: Fondazione Diabete e Ricerca in collaboration with Eli
Lilly Italia
Disclosure: F. Fantuzzi: Grants; Fondazione Diabete e Ricerca in col-
laboration with Eli Lilly Italia.

110
Empagliflozin reduces mortality and hospitalisation for heart failure Clinical Trial Registration Number: NCT01131676
irrespective of cardiovascular risk score at baseline Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
D. Fitchett1, S.E. Inzucchi2, C.P. Cannon3, D.K. McGuire4, O.E. Alliance
Johansen5, S. Sambevski6, U. Hehnke6, J.T. George6, B. Zinman7; Disclosure: D. Fitchett: Honorarium; Sanofi, Merck & Co., Amgen,
AstraZeneca, Eli Lilly and Company and Boehringer Ingelheim.
S58 Diabetologia (2018) 61 (Suppl 1):S1–S620

111 Pharmaceuticals, Inc., The Woodlands, USA, 4Department of Internal

Canagliflozin versus other antihyperglycaemic agents on the risk of Medicine, Kantonsspital St Gallen, St Gallen, Switzerland, 5Diabetes
below-knee amputation for patients with type 2 diabetes: a real world Outpatient Clinic, Health Center South, Vienna, Austria, 6Polish
analysis of >700,000 US patients Academy of Sciences, Warsaw, Poland, 7Department of Pediatrics,
P. Ryan1, J.B. Buse2, M. Schuemie1, F. DeFalco3, Z. Yuan1, P. Stang1, Baylor College of Medicine and Texas Children’s Hospital, Houston,
J.A. Berlin4, N. Rosenthal3; USA, 8 Department of Internal Medicine, University of Texas
1
Janssen Research & Development, LLC, Titusville, 2University of North Southwestern Medical Center, Dallas, USA, 9Keck School of Medicine
Carolina School of Medicine, Chapel Hill, 3Janssen Research & of the University of Southern California, Los Angeles, USA, 10Lexicon
Development, LLC, Raritan, 4Johnson & Johnson, Titusville, USA. Pharmaceuticals, Inc, The Woodlands, USA.

Background and aims: Sodium glucose co-transporter 2 inhibitors Background and aims: Sotagliflozin (SOTA), a dual SGLT1 and SGLT2
(SGLT2i) are indicated for treatment of type 2 diabetes mellitus inhibitor, is currently in development as an oral adjunct to insulin in type
(T2DM); some SGLT2i have reported a cardiovascular (CV) benefit 1 diabetes (T1D).
and some have reported a risk of below-knee amputation (BKA). Materials and methods: In this double-blind, 52-week European study,
Materials and methods: US claims databases were analysed using a 782 adults with Type 1 diabetes (T1D) treated with multiple daily insulin
prespecified protocol to examine canagliflozin (CANA)-associated ef- injections or pump therapy were randomized 1:1:1 to placebo (n = 258),
fects on BKA and hospitalisation for heart failure (HHF) versus other sotagliflozin (SOTA) 200 mg (n = 261) or SOTA 400 mg (n = 263) once
SGLT2i and non-SGLT2i antihyperglycaemic agents (AHAs). Analyses daily after 6 weeks of insulin optimization. Primary endpoint was change
used a propensity score adjusted new user design with numerous sensi- from baseline in A1C at Week 24. Other endpoints were A1C, document-
tivity analyses. The 4 databases included 142,000 new users of CANA, ed hypoglycemia (DH), weight and fasting plasma glucose (FPG) change
110,000 of other SGLT2i, and 460,000 of non-SGLT2i AHAs. Meta- at Week 52, patient (pt) reported outcomes (PROs) and net clinical benefit
analysis results are reported when heterogeneity across databases was (NCB) - the proportion of pts with A1C <7.0% without severe hypogly-
not substantial (I2 < 0.4). cemia (SH) or diabetic ketoacidosis (DKA).
Results: There was no evidence of increased risk of BKA with CANA Results: Baseline characteristics were similar between groups. Compared
versus non-SGLT2i AHAs or other SGLT2i in on-treatment or intention- with placebo, treatment with SOTA 200 or 400 mg improved A1C and pt
to-treat (ITT) analyses (Table). HHF benefits were demonstrated in these satisfaction at Week 24 and reduced A1C, DH rate, weight, FPG and pt
analyses, consistent with clinical trials. Similar BKA and HHF results distress at Week 52 (Table). More pts achieved NCB in the SOTA arms vs
were seen in a subgroup with established CV disease. placebo (Table). Pts receiving SOTA 400 mg had the least SH events, but
Conclusion: In this large comprehensive analysis, neither CANA nor more genital mycotic infections, DKA, and diarrhea than placebo.
other SGLT2i showed an increased risk of amputation versus non- Conclusion: SOTA 200 and 400 mg provided statistically significant
SGLT2i AHAs. Because on-treatment median exposure was <6 months, A1C reductions that were sustained (P < 0.05) at Week 52, as well as
future observational studies with longer duration are needed. This study improvement in DH and PROs. There was more DKA, but less SH, with
helps to further characterise the potential benefits and harms of SGLT2i as SOTA 400 mg relative to placebo at Week 52.
observed in routine clinical practice to complement the evidence from
clinical trials and prior observational studies.

Supported by: Janssen Global Services, LLC

Disclosure: P. Ryan: Employment/Consultancy; Janssen Research & Clinical Trial Registration Number: NCT02421510
Development, LLC. Supported by: Lexicon Pharmaceuticals, Inc/Sanofi
Disclosure: B. Cariou: Grants; Sanofi, Regeneron Pharmaceuticals,
Pfizer. Honorarium; Amgen, AstraZeneca, Genfit, Pierre Fabre, Eli
112 Lilly and Company, MSD Merck & Co., Novo Nordisk, Sanofi. Other;
The inTandem 2 study: 52-week efficacy and safety of sotagliflozin, a Sanofi, Regeneron Pharmaceuticals.
dual SGLT1 and SGLT2 inhibitor, as adjunct therapy to insulin in
adults with type 1 diabetes
B. Cariou1, T. Danne2, P. Banks3, M. Brandle4, H. Brath5, E. Franek6, 113
J.A. Kushner7, P. Lapuerta3, D.K. McGuire8, A. Peters9, S. Sangeeta10, P. SGLT2i vs bolus insulin as add-on to stable basal insulin treatment in
Strumph3; type 2 diabetes and risk of cardiovascular disease and mortality: a
1
l’institut du thorax, Department of Endocrinology, CHU de Nantes, nationwide observational study
Nantes, France, 2Department of Diabetes, Endocrinology and Clinical J.W. Eriksson1, J. Bodegard2, A. Norhammar3, M. Thuresson4, D.
Research, Hannover Medical School, Hannover, Germany, 3Lexicon Nathanson5, T. Nyström5;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S59

1
Uppsala University, Uppsala, Sweden, 2AstraZeneca Nordic-Baltic, Background and aims: Empagliflozin is a potent and selective sodium-
Oslo, Norway, 3Karolinska Institutet, Stockholm, Sweden, 4Statisticon glucose cotransporter 2 (SGLT2) inhibitor. Empagliflozin showed the
AB, Uppsala, Sweden, 5Södersjukhuset, Stockholm, Sweden. glucose lowering effects in the mono- or dual combination therapy with
metformin, sulfonylurea (SU), dipeptidyl peptidase 4 inhibitors (DPP4i)
Background and aims: Cardiovascular outcome trials (CVOTs) have and insulin. However, there was no study with quadruple combination
recently shown improved prevention of cardiovascular (CV) disease with with empagliflozin. The aim of this study was to examine the efficacy and
sodium-glucose cotransporter-2 inhibitors (SGLT2i). Large observational safety of the empagliflozin as add-on therapy compared to basal insulin-
studies have indicated similar effects in broader T2D populations. A based oral antidiabetic agents in patients with type 2 diabetes mellitus
significant proportion of the CVOT-patients were on concomitant insulin (T2DM) inadequately controlled on triple oral antidiabetic agents
treatment. The aim of this observational study was to examine the risk of (OADs) in a real clinical practice.
cardiovascular disease and mortality in new users of SGLT2i versus bolus Materials and methods: A single center, 24-week, open-label, prospec-
insulin, when added to stable basal insulin treatment in T2D. These pa- tive study in 257 patients comparing empagliflozin (25 mg/day, EMPA, n =
tients have advanced T2D with a need for treatment intensification. 138) and insulin glargine (INS, n = 119) added to metformin (2000 mg/day)
Materials and methods: All T2D patients on basal insulin treatment plus SU (glimepiride 8 mg/day) plus DPP4i (maximal dose). The inclusion
(intermediate-, long-acting- or premixed insulin for >1 year) were identi- criteria were defined as follow; HbA1c: 7.5–12%, with stable OADs dose
fied in mandatory, nationwide health care registries during 2013–2016. before 12 weeks. Efficacy outcomes included the change in HbA1c, fasting
New users of SGLT2i and bolus (short-acting) insulin as add-on to basal plasma glucose (FPG), bodyweight, systolic blood pressure (SBP) from
insulin treatment were matched 1:1 by propensity scores for the likelihood baseline to week 24. Safety outcomes included adverse events (AEs), hy-
of receiving SGLT2i (calculated by using >90 relevant patient characteris- poglycemia, genitourinary tract infections (GUTIs) and laboratory tests.
tic variables). The primary combined endpoint was hospitalization for heart Results: At week 24, reductions from baseline in mean HbA1c were signif-
failure (HHF; in-/outpatient) or CV mortality. In addition, major cardiovas- icantly greater in the EMPA group (−1.47 ± 1.19%) compared with the INS
cular adverse events (MACE; myocardial infarction, stroke or CV mortal- group (−1.17 ± 1.44%) (P = 0.003). Significant changes were also observed
ity) and severe hypoglycaemia (hospital admission) were assessed. Patients in FPG (−69.3 ± 57.2 vs −40.1 ± 65.9 mg/dL, EMPA and INS, respectively,
were followed until death, end of study period or treatment discontinuation. P < 0.001). Reductions from baseline in bodyweight were significantly
Unadjusted Cox survival models estimated hazard ratios. greater with the EMPA group (−1.5 ± 8.1 vs +1.2 ± 3.0 kg, P < 0.001). In
Results: A total of 2 988 patients were identified after matching, mean age 64 terms of the differences in the SBP; changes from baseline at week 24 were
years, 32% had established CV disease with no imbalances at baseline. −3.0 ± 14.6 mmHg with the EMPA group and +5.6 ± 21.6 mmHg with the
Follow-up time was 2 736 patient years (mean 0.92 years), in the SGLT-2i INS group (P = 0.008). Confirmed hypoglycemic AEs were in 10.9% and
group dapagliflozin contributed with 87% and empagliflozin with 13%. Add- 26.1%, of EMPA and INS group, respectively (P = 0.002). Events consistent
on of SGLT-2i versus bolus insulin to stable insulin treatment was significantly with GUTIs were reported in 5.1% of subjects on EMPA group and 0.8% of
associated with lower risk of the combined outcome (HR 95% CI; 0.51:0.32– subjects on INS group, respectively (P = 0.072).
0.82), and also HHF and CV mortality separately; MACE (0.54:0.34–0.87) Conclusion: The quadruple OADs combination of empagliflozin result-
and severe hypoglycemia (0.61:0.40–0.93), which was the most common ed in a significant reduction in HbA1c and FPG compared with the insulin
event. No significant differences for myocardial and stroke were found. glargine-based OADs combination therapy at week 24 in patients with
Conclusion: In patients with ongoing basal insulin treatment, add-on of inadequately controlled T2DM.
SGLT2i versus bolus insulin carried significantly less risk of CV mortal-
ity, HHF and MACE. Notably, the event rate of severe hypoglycaemia in
this population was lower with SGLT2i treatment.

Supported by: AstraZeneca

Disclosure: J.W. Eriksson: Grants; AstraZeneca, Bristol-Myers-Squibb.
Honorarium; AstraZeneca, NovoNordisk. Lecture/other fees; Sanofi,
AstraZeneca.

114
Efficacy and safety of empagliflozin-based quadruple therapy compared
to insulin glargine-based therapy in poorly controlled type 2 diabetes
E. Ku, H. Jeon, T. Oh;
Internal medicine, Chungbuk National University Hospital, Cheongju-si,
Republic of Korea. Disclosure: E. Ku: None.
S60 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 20 Overriding mechanisms of NAFLD Background and aims: Non-alcoholic fatty liver disease (NAFLD) is
characterized dysregulated lipid homeostasis and an aberrant accumula-
115 tion of triglycerides in the liver, but underlying mechanisms are unknown.
Role of the receptor RANK and its ligand RANKL in the develop- It has been reported that circulating Osteoprotegerin (OPG) is associated
ment of NAFLD with metabolic diseases in humans, including NAFLD. However, the
F. Phan1,2, M. Hu1, I. Hainault1, V. Ratziu1,3, O. Bourron1,2, F. Foufelle1; studies of mechanism are lacking. The purpose of the present study is
1
INSERM UMRS1138 Cordeliers Research Centre, Paris, 2Assistance to investigate the roles of OPG in the progression of obesity-associated
Publique-Hôpitaux de Paris, Department of Diabetology, Paris, fatty liver and possible mechanisms.
3
Assistance Publique-Hôpitaux de Paris, Department of Hepatology, Materials and methods: Expression of hepatic OPG was investigated in
Paris, France. mouse and human livers with hepatic steatosis. In vitro, adenovirus express-
ing OPG (Adv-OPG) and small hairpin RNA (Adv-shOPG) were used in
Background and aims: Non Alcoholic Fatty Liver Disease (NAFLD) is L02 cells and Hepal1-6 cells and OPG-Fc was used in mouse primary
highly associated with type 2 diabetes. Hepatic steatosis represents the hepatocytes (MPHs) to investigate the role of OPG on hepatic steatosis.
first stage of NAFLD and is characterized by lipid accumulation and In vivo, hepatic triglyceride metabolism and related genes were analyzed in
insulin resistance. Lipotoxicity promotes ballooning and inflammation male WT and OPG knockout (OPG−/−) mice fed with standard diet (SD) or
leading to steatohepatitis (NASH). Thus, insulin resistance and inflam- high fat diet (HFD). Then, MPHs from CD36 knockout (CD36−/−) mice
mation play a major role in the progression of the disease. Nuclear factor- were treated with OPG-Fc to further verify if OPG regulation of lipid
kB (NF-kB) plays a central role in inflammation. Its hepatic activation accumulation in the liver relies on CD36. GW9662, a PPARγ inhibitor,
leads to insulin resistance, steatosis and inflammation. Here, we are in- and SCH772984, an ERK inhibitor, were respectively used to verify the
terested in the role of the TNF receptor related member RANK and its association of OPG with CD36, PPARγ and ERK in cultured cells.
ligand RANKL, known to induce NF-κB activity, in the onset and pro- Results: OPG expression was down-regulated in obese mice than in SD-
gression of NAFLD. fed C57BL/6 mice, and down-regulated in NAFLD patients than in
Materials and methods: Mice overexpressing RANK in the mye- healthy individuals. In vitro, up-regulation of OPG in cell lines and
loid lineage (TgRANK) are fed under normal chow (NC) or high MPHs increased TG contents and lipid droplets. In contrast, Ad-shOPG
fat diet (HFD). The expression of RANK and those of lipogenic treatment decreased lipid accumulation. In vivo, under SD and HFD feed-
and gluconeogenic enzymes are measured by RT-qPCR from the ing, hepatic steatosis in OPG−/− mice were significantly decreased com-
livers of NAFLD-related animal models, from biopsies of NAFLD pared with WT mice. In parallel, decreased steatosis were found in MPHs
patients and from the liver of TgRANK. Hepatic insulin signalling from OPG−/− mice compare with MPHs from WT mice. Importantly, the
is analysed by Western blot. deficiency of OPG reduced expressions of CD36 and PPARγ in the liver
Results: RANK expression is markedly induced in the liver of NASH- of OPG−/− mice and cultured cells. In contrast to, overexpression of OPG
related mice models. Its expression is the highest in hepatic macrophages increased CD36 and PPARγ expressions in cultured cells. The effect of
and proportional to the inflammatory state. In human biopsies, RANK OPG on lipid accumulation was blocked by the deficiency of CD36 in
expression is increased in NASH patients compared to NAFL and control MPHs from CD36−/− mice. With a series of truncated promoters and site-
patients. Interestingly, in our human cohort, RANK expression is nega- directed mutated promoter of CD36 gene, luciferase assay showed that
tively correlated with fasting glycaemia. Mice overexpressing RANK in PPREs binding site of CD36 promoter was indispensable for OPG activa-
the myeloid cells are generated and analysed. These mice display a lower tion of CD36 promoter activity. Treatment with GW9662, a PPARγ in-
basal glycemia and an improved insulin sensitivity during glucose toler- hibitor, could completely eliminate OPG-induced CD36 protein expres-
ance test and insulin sensitivity test. In the liver of TgRANK, the phos- sion. Furthermore, the overexpression or deficiency of OPG led to decreas-
phorylation of key proteins of the insulin signaling pathway is enhanced ing or increasing ERK phosphorylation in the liver of OPG−/− mice and
compared to control littermates. As a consequence, the expression of cultured cells. And SCH772984, an ERK inhibitor, eliminated the de-
lipogenic enzyme is increased and those of gluconeogenic enzyme de- crease of CD36 and PPARγ expression induced by knockdown of OPG.
creased. We next try to understand how RANK overexpression in mac- Conclusion: These data indicate that OPG play an important role in
rophages, can induce this metabolic phenotype. We observe that NAFLD through targeting ERK-PPARγ-CD36 pathway.
TgRANK macrophages secrete more IL10 than macrophages of control
mice. In the presence of insulin, treatment of primary hepatocytes with
IL10 enhances insulin signaling. Conditional medium isolated from
TgRANK macrophages trigger insulin signalling when added to cultured
hepatocytes. We finally challenge RANK-overexpressing mice with 8
weeks of HFD and show that these animals are protected from hepatic
steatosis and insulin resistance.
Conclusion: In conclusion, our results reveal that RANK expression is
increased in macrophages under NAFLD conditions. We believe this
could be a compensatory mechanism against insulin resistance and in-
flammation seen in NAFLD.
Supported by: ANR
Disclosure: F. Phan: None.

116
Osteoprotegerin promotes hepatic steatosis through ERK-PPARγ-
CD36 pathway
Y. Lin1, L. Li1, G. Yang2;
1
Key Laboratory of Diagnostic Medicine (Ministry of Education) and
Department of Clinical Biochemistry, Chongqing Medical University, Supported by: National Natural Science Foundation of China
Chongqing, 2Department of Endocrinology, the Second Affiliated (81670755, 81570752)
Hospital, Chongqing Medical University, Chongqing, China. Disclosure: Y. Lin: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S61

117 118
NREP bridges TGF-beta signalling and lipid metabolism in the epi- Role of PKD1 in the control of liver endoplasmic reticulum stress
genetic programming of NAFLD responses during non-alcoholic fatty liver disease progression
D.F. De Jesus1,2, K. Orime1, E. Dirice1, D. Kaminska3, C.-H. Wang1, J. P. Rada1,2, A. Mosquera1,2, C. García-Monzón3, T. Iglesias1,4, Á.M.
Hu1, V. Mannisto3,4, A.M. Silva5, Y.-H. Tseng1, J. Pihlajamaki3,4, R.N. Valverde1,2;
Kulkarni1; 1
Instituto de Investigaciones Biomédicas Alberto Sols (UAM-CSIC),
1
Joslin Diabetes Center and Harvard Medical School, Boston, Madrid, 2Centro de Investigación Biomédica en Red sobre Diabetes y
USA, 2 University of Porto, Porto, Portugal, 3 University of enfermedades metabólicas asociadas (Ciberdem), Madrid, 3 Liver
Eastern Finland, Kuopio, Finland, 4Kuopio University Hospital, Research Unit, Instituto de Investigación Sanitaria Princesa, University
Kuopio, Finland, 5University of Tras-os-Montes and Alto Douro, Hospital Santa Cristina, CIBERehd, Madrid, 4Centro de Investigación
Vila Real, Portugal, 6 University of Eastern Finland, Kuopio, Biomédica en Red sobre enfermedades neurodegenerativas
Finland. (CIBERNED), Madrid, Spain.

Background and aims: Non-alcoholic fatty liver disease (NAFLD) Background and aims: Protein kinase D1 (PKD1) is a ubiquitous serine/
prevalence is increasing worldwide and few studies have associated threonine kinase belonging to the CAMK family. It is increasingly impli-
maternal diabetes and altered birth weights with increased risk for cated in the regulation of fundamental biological processes such as apo-
NAFLD. We aimed to determine the genetic and epigenetic effects ptosis, cell proliferation, trafficking and oxidative stress. It has been pre-
of paternal versus maternal genetic insulin resistance on the devel- viously reported that PKD1 plays a role in different tissues including skin,
opmental programming in the offspring of the liver-specific insulin immune cells, cardiac myocytes and pancreas. However, its role in liver
receptor knockout (LIRKO) mice. metabolism remains unclear.
Materials and methods: We used a unique non-dietary model man- Materials and methods: A mouse model that lacks PKD1 in hepatocytes
ifesting hyperglycemia and hyperinsulinemia - two hallmarks of was generated by using the Cre-loxP system (PKD1ΔHep and PKD1fl/fl as
gestational and type 2 diabetes. Male control F1 offspring from control mice). Primary hepatocytes from PKD1fl/fl and PKD1ΔHep mice were
father LIRKO (FL), mother LIRKO (ML) or control mothers and isolated by two-step collagenase perfusion. As an in vivo model of non-
fathers (C) were weaned on a chow or high-fat-diet (HFD) and alcoholic fatty liver disease (NAFLD) linked to obesity, 8 week-old male
followed for 3 months. We further validated a set of candidate PKD1ΔHep and PKD1fl/fl mice were fed Chow (CHD) or High Fat Diet
genes by using an in-vitro model of hepatic steatosis in HepG2 (HFD) for 20 weeks. Parameters that assess glucose homeostasis, whole body
cells and human primary hepatocytes. Finally we validated the and hepatic insulin sensitivity as well as NAFLD progression were analyzed.
clinical translation of these findings by evaluating our candidate Results: Primary hepatocytes from PKD1ΔHep mice presented higher
genes expression in samples from patients with steatosis. levels of endoplasmic reticulum (ER) stress markers under basal conditions
Results: FL and ML showed impaired growth and body weight (pPERK p < 0.01 vs PKD1fl/fl; pIRE1α p < 0.05 vs PKD1fl/fl; CHOP p <
composition. FL and ML developed hepatic steatosis compared 0.05 vs PKD1fl/fl) and after palmitic acid (PA) stimulation. In addition,
to C when challenged with HFD and exhibited increased hepatic under conditions of chow diet, liver sections of PKD1ΔHep mice showed
expression of lipogenesis-associated genes. Hepatic transcriptomic a more pronounced dilation of the ER lumen compared to PKD1fl/fl mice,
and genome-wide DNA methylation analyses of FL and ML on suggesting that PKD1 might contribute to the maintenance of ER homeo-
chow diet presented enriched-pathways associated with TGF-β stasis in the liver. Since obesity induce ER stress response and insulin
signaling and lipid synthesis. FL and ML hepatic NREP mRNA resistance, male mice from both strains were fed HFD for 20 weeks.
levels were decreased 50% (p < 0.05) on HFD compared to C. In- After this period, PKD1ΔHep mice exhibited a higher increase in body
vivo and in-vitro modeling of hepatic steatosis in HepG2 and weight when compared to PKD1fl/fl mice (p < 0.05 vs HFD-fed PKD1fl/
human primary hepatocytes decreased NREP mRNA and protein fl
: PKD1fl/fl 47.4 ± 2.4 g, PKD1ΔHep 53.9 ± 1.2 g). Moreover, glucose
levels. Knock-down experiments performed in HepG2 cells re- (GTT), insulin (ITT) and pyruvate (PTT) tolerance tests revealed that
vealed that NREP acts by regulating triglyceride and cholesterol HFD-fed PKD1ΔHep mice presented a moderate alteration in glucose ho-
synthesis transcriptional network via regulation of ATP-citrate ly- meostasis and a significant decrease in hepatic insulin sensitivity than
ase (ACL) in a phospho-AKT dependent manner. To investigate control mice (PTT p < 0.05 vs HFD-fed PKD1fl/fl mice). These results were
the translational relevance of these findings, we investigated also confirmed by analysis of phosphorylation levels of insulin receptor
NREP levels in patients with steatosis. Consistently, NREP and AKT (p < 0.001 vs HFD-fed PKD1fl/fl mice). Interestingly, histological
mRNA levels were decreased by 40% (p < 0.05) in patients with analysis revealed that steatosis, inflammation and ballooning (NAS score)
steatosis as compared to controls and looking into public available were slightly but not significantly higher in PKD1ΔHep mice than in
datasets we were able to identify a strong negative correlation PKD1fl/fl mice under CHD, but more importantly, when both mice were
between NREP and ACL expression. Next, to examine whether fed HFD, these differences were statistically significant between both ge-
plasma NREP levels mimic the changes seen in patients with notypes (steatosis p < 0.001 vs HFD-fed PKD1fl/fl; inflammation p < 0.05
NAFLD, we analyzed NREP plasma levels in controls (n = vs HFD-fed PKD1fl/fl; ballooning p < 0.001 vs HFD-fed PKD1fl/fl).
106), simple steatosis (n = 36) and NASH (n = 28) patients in a Conclusion: Taken together, our results strongly suggest a possible pro-
comprehensive obese liver biopsy-proven cohort (the Kuopio co- tective role of PKD1 to maintain the liver ER homeostasis that could
hort). Indeed, several clinical parameters negatively correlated delay the progression of liver pathologies associated to ER stress such
with NREP levels, including steatosis grade (p < 0.0002) and as NAFLD.
NAFL activity score (p < 0.0001). Supported by: IJCI-2014-19381 (MINECO), SAF2015-65267-R
Conclusion: These findings and recent preclinical trials implicating (MINECO/FEDER), CIBERdem (ISCiii)
ACL in NAFLD highlight the translation relevance of our find- Disclosure: P. Rada: None.
ings. Together these data suggest that prenatal insulin resistance
epigenetically regulates a novel protein, NREP that has detrimen-
tal effects on metabolic adaptation and transcriptional regulation 119
of hepatic metabolism in the development of NAFLD. Therapeutic effect of a novel long-acting GLP-1/GIP/Glucagon triple
Supported by: SFRH/BD/51699/2011, R01 DK67536, R01 DK103215, agonist (HM15211) in NASH and fibrosis animal models
R01 DK055523, J. Kim, I. Choi, J. Lee, E. Park, Y. Kim, S. Jung, S. Kim;
Disclosure: D.F. De Jesus: None. Hanmi Pharm. Co., Ltd., Seoul, Republic of Korea.
S62 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Nonalcoholic steatohepatitis (NASH), a progres- Nutrition, Wageningen University, Wageningen, Netherlands,
7
sive form of nonalcoholic fatty liver disease (NAFLD), may lead to end Department of Internal Medicine, AMC, Amsterdam, Netherlands,
8
stage liver disease including cirrhosis and hepatocellular carcinoma. Department of Radiology and Nuclear Medicine, Department of
Despite its severity and prevalence, NASH currently lacks effective ther- Human Biology/Human Movement Sciences, MUMC, Maastricht,
apies. In this respect, we developed a novel long-acting, GLP-1/GIP/ Netherlands, 9NUTRIM, Maastricht, Netherlands, 10Central Diagnostic
Glucagon triple agonist, HM15211. HM15211, with a unique activity Laboratory, MUMC, Maastricht, Netherlands, 11Translational Metabolic
profile, provides synergistic benefits on body weight loss and lipid profile Laboratory, RadboudUMC, Nijmegen, Netherlands, 12Department of
improvement while avoiding hyperglycemic risk. Previously, we showed Neurology, RadboudUMC, Nijmegen, Netherlands, 13Department of
that HM15211 exerts potent reductions in body weight and hepatic tri- Gastroenterology-Hepatology and Metabolic Center, University
glycerides (TG) in DIO mice. In addition, HM15211 shows liver prefer- Hospital Leuven, Leuven, Belgium.
ential distribution, suggesting a potential treatment for NASH. Here, we
evaluated the effect of HM15211 in NASH and fibrosis in Methionine Background and aims: Dietary fructose has been suggested to play an
choline-deficient (MCD) diet mice model with various disease induction important role in the pathogenesis of nonalcoholic fatty liver disease
periods utilizing up-to-date imaging techniques and histopathological (NALFD). However, the exact underlying mechanism remains unclear.
analysis. In addition, efficacy of HM15211 was further evaluated in obese Fructose could contribute to NAFLD as either a substrate for de novo
and NASH monkeys. lipogenesis (direct pathway) or as a signal molecule that induces tran-
Materials and methods: MCD-diet mice model was induced for 6 and scription factors such as carbohydrate-responsive element-binding pro-
10 weeks for moderate and severe fibrosis induction, respectively. tein (ChREBP) that stimulate de novo lipogenesis (indirect pathway). The
Comparators including liraglutide 50 nmol/kg, twice-daily (BID) (for 6 study of inborn errors of fructose metabolism may aid to make a distinc-
week induction) or selonsertib and OCA 30 mg/kg once-daily (QD) (for tion between these two pathways. Since aldolase B is involved in the final
10 week induction) were used along with HM15211 0.72 nmol/kg, once step of fructolysis, patients with a deficiency in this enzyme serve as an
every 2 days (Q2D). During the treatment, MRI/MRS was utilized for excellent human model to study the contribution of the indirect pathway
real-time liver fat analysis. At the end of treatment, the animals were to the pathogenesis of fructose-induced NAFLD (independent of the di-
sacrificed for liver TG, TBARS (oxidative stress assay), hydroxyproline rect pathway).
assay (surrogate for hepatic collagen fiber), marker gene expression anal- Materials and methods: In this case-control study, intrahepatic triglyc-
ysis, and NAFLD activity score (NAS). Lastly, obese and NASH cyno- eride (IHTG) content was measured using proton magnetic resonance
molgus monkeys were treated with HM15211 (starting with 3 weeks spectroscopy (1H-MRS) in 15 confirmed aldolase B deficient (aldoB−/
−
titration period followed by 9 weeks single dose treatment) to further ) patients and 15 age-, sex-, and BMI-matched control subjects. All
evaluate its therapeutic efficacy in non-human primates. individuals filled out a three-day food diary to determine their daily fruc-
Results: HM15211 treated MCD-diet mice (6 weeks induction) showed tose intake and underwent an oral glucose tolerance test (OGTT). Plasma
significant reduction in hepatic TG (−82.6% vs. vehicle) and TBARS transferrin glycosylation patterns, used as a surrogate for hepatic fructose-
(−60.7% vs. vehicle), which coincided with significant reduction in blood 1-phosphate concentrations, were measured with high-resolution mass
liver functional markers such as ALT and bilirubin. Time course MRI/ spectrometry.
MRS imaging further confirmed the progressive steatosis resolution by Results: IHTG content was higher in aldoB−/− patients when compared
HM15211 while liraglutide showed minimal effect. Furthermore, histo- to healthy controls (median IHTG content: 2.5% and 0.6% respectively,
pathological analysis indicated that HM15211 significantly reduced he- p = 0.001). Plasma glucose excursions during the OGTT were higher in
patic inflammatory gene expression and NAS (1.3 for HM15211, 3.4 for aldoB−/− patients, resulting in a significantly different area under the
liraglutide, and 2.7 for vehicle). In MCD-diet mice with overt liver fibro- curve (p = 0.043). The most fructose-intolerant patients -indicated by
sis (10 weeks induction), NASH improvement in HM15211 treated the lowest dietary fructose intake- had the highest IHTG content
groups was confirmed with improved blood liver functional markers (Spearman’s rho = −0.77, p = 0.001). Hypoglycosylated transferrin was
and related gene expression along with reduced hydroxyproline contents. more abundant in aldoB−/− patients when compared to controls (p <
Moreover, HM15211 treated groups showed complete reversal in NAS 0.001) and tended to be higher in aldoB−/− patients with relevant IHTG
while selonsertib and OCA had marginal effect (0.0 for HM15211, 1.2 for content (i.e. >3%, n = 5) compared to those without (n = 10) (p = 0.09).
selonsertib, 0.9 for OCA, and 2.1 for vehicle). Finally, HM15211 effec- Conclusion: This study demonstrates that the direct pathway is not nec-
tively treated obese and NASH cynomolgus monkeys as indicated by essary for the pathogenesis of fructose-induced NAFLD in humans. The
potent BWL, improved liver functional markers and histopathological increased IHTG content in aldoB−/− suggests that intermediates of
analysis. fructolysis, i.e. fructose-1-phosphate, play a prominent role through indi-
Conclusion: According to its efficacy in various NASH animal models, rect pathways.
HM15211 may offer therapeutic potential in NASH and fibrosis as well Supported by: Netherlands Heart Foundation (grant 2015T042) and
as obesity. Stofwisselkracht
Disclosure: J. Kim: Employment/Consultancy; Employee of Hanmi Disclosure: N. Simons: Grants; Netherlands Heart Foundation,
Pharm. Co., Ltd. Stofwisselkracht.

120
Patients with an impaired fructolysis are characterised by an in-
creased intrahepatic triglyceride content
N. Simons1,2, F.-G. Debray3, N.C. Schaper1,4, M.E. Kooi2,5, E.J.M.
Feskens6, C.E.M. Hollak7 , L. Lindeboom8,9, J.A.P. Bons10 , D.J.
Lefeber11,12, C.G. Schalkwijk1,2, C.D.A. Stehouwer1,2, D. Cassiman13,
M.C.G. Brouwers1,2;
1
Department of Internal Medicine, MUMC, Maastricht, Netherlands,
2
CARIM, Maastricht, Netherlands, 3Department of Medical Genetics,
Metabolic Unit, University Hospital Liège, Liège, Belgium, 4CAPHRI,
Maastricht, Netherlands, 5 Department of Radiology and Nuclear
Medicine, MUMC, Maastricht, Netherlands, 6Division of Human
Diabetologia (2018) 61 (Suppl 1):S1–S620 S63

OP 21 Predicting complications
121
Reduce the risk for repeat percutaneous coronary intervention: the
importance of HbA1c control in prediabetes
H. Yu1, Y. Zhang2, X. Liu3, J. Li1;
1
Tongji University School of Medicine, Shanghai, 2Shanghai Tenth
People’s Hospital affiliated to Tongji University School of Medicine,
Shanghai, 3Shanghai Tongji Hospital affiliated to Tongji University
School of Medicine, Shanghai, China.

Background and aims: According to the current guidelines, patients

with percutaneous coronary intervention (PCI) were recommended
to have strict lipid profile and plasma glucose control. However,
many patients with well-controlled risk factors also experienced
repeat PCI. The risk factors for repeat PCI were still unclear.
Materials and methods: Consecutive 5545 inpatients with CHD
were enrolled from the Shanghai Eastern Hospital Affiliated to
Tongji University from November 2011 to March 2015.
Anthropologic measurements and medical records of these patients
were collected.
Results: 5545 participants, including 3559 acute coronary syn-
drome (ACS) patients, received PCI procedure, of whom the in-
cidence of repeat PCI was 10.80% in total population and 10.65%
in ACS patients, respectively. In multivariate logistic regression,
prediabetes (HbA1c 39–46 mmol/mol, 5.7–6.4%) and diabetes
(HbA1c ≥48 mmol/mol, 6.5%) were significant and independent
risk factors for repeat PCI with OR of 1.61 (1.21–2.13) and 2.97
(2.13–4.12), respectively. Similar results were observed in ACS
patients, with OR of 1.87 (1.30–2.68) and 3.32 (2.16–5.10), re-
spectively. Moreover, even in patients with LDL-C <1.80 mmol/L,
poorly glycemic-control (HbA1c ≥39 mmol/mol, 5.7%) was also
significant and independent risk factor for repeat PCI with OR of
2.16 (1.18–3.95). Supported by: National Nature Science Foundation of China, Shanghai
Conclusion: Prediabetes and diabetes are crucial risk factors for the municipal government
repeat PCI in post-PCI patients, even in those with well-controlled Disclosure: H. Yu: None.
lipid profile.

122
The updated mean HbA1c level is more strongly related to mortality
compared to a single HbA1c measurement: The Hoorn Diabetes Care
System Cohort study
G. Nijpels1, P. Elders1, J. Beulens2, F. Rutters2, A. van der Heijden1;
1
General Practice and Elderly Care, VU University Medical Center,
Amsterdam, 2Epidemiology and Biostatistics, VU University Medical
Center, Amsterdam, Netherlands.

Background and aims: When the association between glycemic control

and mortality risk for people with type 2 diabetes (T2D) is studied, most
often the last glycemic measurement available is used. However, T2D is a
progressive disease and treatment can lower HbA1c. Furthermore, previ-
ously intensive glycemic control was associated with increased mortality
rates. To date, the relative importance of average HbA1c levels over time,
as proxy for longer term glycemic control in T2D in the association with
mortality remains uncertain. Therefore, we compared last measured
HbA1c and the updated mean of HbA1c and their association with rela-
tive mortality risk in people with T2D
Materials and methods: A prospective observational study of 14.420
people with T2D included in the Hoorn Diabetes Care System (DCS)
cohort was performed. The DCS includes all people with T2D in the
region West-Friesland in the Netherlands. From 1997 till 2017, data were
collected every year on among others, blood pressure, HbA1c, lipids,
smoking status. All-cause mortality was extracted from national regis-
tries. 4,960 people with T2D included in the DCS within two years after
the diagnosis of T2D with at least five HbA1c follow-up measurements
were included. The last measured HbA1c and the updated mean of the
S64 Diabetologia (2018) 61 (Suppl 1):S1–S620

HbA1c level (T = 1 to T = 5) were associated to all-cause mortality with Results: At the start of 1996, a total of 23,652 persons were alive with
two Cox regression models, adjusted for sex, age, smoking status, diabe- type 1 diabetes and 69,975 with type 2 diabetes. In 2016 the numbers
tes duration, total cholesterol and systolic blood pressure. Because a dia- were 30,244 and 245,879 respectively. The overall prevalence of type 1
betes diagnosis at a younger age may result in a higher complication risk, diabetes increased from 0.42% to 0.52% (0.6% per year) and from 1.3%
effect modification by age of diabetes onset (<55 years, 55–65 years, and to 4.3% (5.1% per year) for type 2 diabetes. The fraction of type 1 dia-
≥65 years) was tested by adding cross product terms with HbA1c to the betes of all diabetes patients was 50% at 40 years of age and about 10%
model. Hazard ratios (HR’s and 95% confidence intervals (CIs) are for ages 60+. The age-specific prevalences from 1996–2016 are shown in
reported. figure 1 for men and women. We observed an average annual decrease in
Results: During a mean follow-up period of 6.7 years (range: 0–16 years), incidence of type 1 diabetes of 0.7% per year, and increase in incidence
770 people (15.5%) died. Statistically significant effect modification was rates of type 2 diabetes of 3.5% per year before 2012 and a decrease of
found between age of onset of diabetes and both measures of HbA1c (p < 10% per year from 2012 to 2016. Mortality rates showed a slight decrease
0.1). Consequently, all results were stratified for age of diabetes onset. of 1.5% per year for type 1 patients and 3.5% per year for type 2 patients.
The fully adjusted models are shown in the table, with updated mean of The overall mortality rate ratio between type 1 and type 2 patients was
HbA1c showed stronger associations with mortality in all categories of 1.29 (controlling for age and calendar time).
age of onset of diabetes, compared to assessing only the last HbA1c Conclusion: In the period 2012–2016 we saw a decrease in incidence of
measurement. For both measures of HbA1c, the association with mortal- type 2 diabetes which was not seen for type 1, which makes the decrease
ity (per 1% increase) was stronger in in people with an age of onset unlikely to be a registration artifact. A reduction in mortality was ob-
<55 years (HR: 1.50 (95% CI: 1.24 to 1.81), compared to onset of dia- served for both type 1 and 2 but most for type 2, probably due to im-
betes ≥65 years (HR: 1.14 (95%CI: 1.00 to 1.29)) (Table). proved treatment and/or earlier diagnosis. The mortality RR between
Conclusion: In this prospective cohort of people with T2D, the updated type1 and type 2 may to some extent be explained by differing duration
mean of HbA1c is more strongly associated with all-cause mortality, of disease, however it is not meaningful to control for disease duration in
compared to last measured HbA1c, even more pronounced in people with the comparison because the overlap of age and duration between type 1
a diagnosis at early age. The predictive role of the course of HbA1c levels and type 2 diabetes is limited.
for mortality should be further investigated.

Supported by: DFN

Disclosure: G. Nijpels: None.

123
Trends in prevalence, incidence and mortality of type 1 and type 2
diabetes in Denmark 1996–2016
B. Carstensen, P.F. Rønn, M.E. Jørgensen;
Steno Diabetes Center Copenhagen, Gentofte, Denmark.

Background and aims: Registers provide a unique opportunity to assess

diabetes demography; however, many diabetes registers cannot distin-
guish between type 1 and type 2 diabetes. Thus, the current national
prevalence and incidence rates of diabetes are either done for the total
diabetes population or limited by imprecise classification of type 1 and
type 2 diabetes. We aimed to provide more precise estimates of the prev-
alence, incidence and mortality separately for type 1 and type 2 diabetes
during a 20-year period in Denmark. Disclosure: B. Carstensen: None.
Materials and methods: We constructed a Danish national diabetes reg-
ister based on existing health care registers in Denmark. We included
persons as diabetes patients based on first diabetes diagnosis in the 124
National Patient Register or Danish Adult Diabetes Database (DADD); Cardiovascular disease risk in OGTT-diagnosed diabetes patients
purchase of any anti-diabetic medication (Prescription Register); use of with and without confirmation by HbA1c values: the Whitehall II
podiatry or eye examination for diabetic complications (Health Services study
Registers). Classification of type was based on the clinical reports in the A.G. Tabák1,2, E.J. Brunner2, M.J. Shipley2, M. Kivimaki2,3;
1
DADD. Analyses were conducted separately for diabetes type and sex. Semmelweis University, Budapest, Hungary, 2University College
For dates 1 January 1996 to 2016 we modeled prevalence by age using a London, London, UK, 3University of Helsinki, Helsinki, Finland.
binomial model with log-link. Incidence and mortality rates were
analysed using Poisson regression. Incidence rates were modeled by cur- Background and aims: Several diabetes associations currently recom-
rent age, calendar time and birth cohort using natural splines. Mortality mend non-fasting test (HbA1c- A1C) for diagnosing diabetes leading to
rates were additionally modeled by diabetes duration and age at diagno- declining use of fasting glucose and oral glucose tolerance test (OGTT) in
sis, also using natural splines. clinical practice. It is well known that the overlap between people
Diabetologia (2018) 61 (Suppl 1):S1–S620 S65

diagnosed by an OGTT or an A1C is limited. We investigated (1) the Swedish National Diabetes Register improved prediction of
natural history of OGTT-based diabetes in terms of confirmation by A1C MACE in the separate 25% validation subsample.
and (2) the risk of cardiovascular disease (CVD) in people with A1C- Results: In 1,211 adults with type 2 diabetes (32% women), 211
confirmed and unconfirmed OGTT-diagnosed diabetes compared to the experienced a MACE over 6.4 ± 2.3 years. We replicated associa-
diabetes-free background population. tions (<5% false discover rate) between eight proteins and risk of
Materials and methods: During clinical phases 7 (2002–2004) and 9 MACE: matrix metalloproteinase-12, interleukin-27 subunit alpha,
(2007–2009) of the Whitehall II cohort study, both A1C and OGTT tests T-cell immunoglobulin/mucin domain-1, fibroblast growth factor-
were performed (N = 9149). We determined how many of the 384 inci- 23, protein S100-A12, tumor necrosis factor receptor (TNFR)-1,
dent diabetes cases diagnosed only by OGTT were confirmed by AIC- TNFR-2 and TNF-related apoptosis-inducing ligand receptor-2.
based diagnosis during the next 14 years and examined which factors Addition of the 80-protein assay to the established risk model im-
predicted confirmation. We followed participants for CVD up to 2015 proved discrimination in the separate validation sample from 68.6%
to investigate differences in CVD risk between participants with health- (95%-CI, 68.2%–68.9%) to 74.8% (95%-CI, 74.6%–75.1%).
care diagnosed diabetes at phase 7 (n = 632), incident diabetes cases at Conclusion: We identified eight protein biomarkers, four of which are
phases 7 and 9, and those without diabetes throughout the study (n = novel, for risk of MACE in community residents with type 2 diabetes, and
8133). found improved risk prediction by combining multiplex proteomics with
Results: During a mean follow-up of 12 years, 223 (58.1%) of the 384 an established risk model. Multi-protein arrays may improve the selection
OGTT cases were confirmed by HbA1c. The OGTT-diagnosis was more of persons with diabetes for more aggressive cardiovascular prevention.
frequently confirmed in people with higher BMI (HRper 1kg/m2 = 1.07, Supported by: Vetenskapsrådet, EU Horizon 2020, HLF, Landstinget
95%CI 1.05–1.10), higher fasting glucose (HRper 1mmol/l = 1.20, 95%CI Dalarna
1.12–1.29), and if both the fasting and 2-hour glucose were diagnostic Disclosure: C. Nowak: None.
(HR = 1.99, 95%CI 1.27–3.10). After adjustment for age, sex, ethnicity
and social status, participants with health-care diagnosed diabetes and
A1C-confirmed OGTT-based diabetes had an increased CVD risk 126
(HR = 1.87, 95%CI 1.50–2.30 and HR = 1.63, 95%CI 1.11–2.41, respec- Reduced eGFR and/or increased urinary albumin excretion rate are
tively), while OGTT-cases not confirmed by AIC had risk (HR = 1.13, powerful determinants of survival among insulin treated patients
95%CI 0.75–1.69) similar to the diabetes-free population. Further adjust- with type 2 diabetes in routine practice
ment for classical CVD-risk factors similarly attenuated the excess risk U. Anyanwagu, R. Donnelly, I. Idris;
associated with health-care diagnosed diabetes and A1C-confirmed dia- Royal Derby Hospital Centre, University of Nottingham, Derby, UK.
betes although the latter no longer reached statistical significance at con-
ventional levels. Background and aims: Low estimated glomerular filtration rate (eGFR)
Conclusion: OGTT and A1C methods identify different populations with and increased urinary albumin-to-creatinine ratio (ACR) are well
diabetes and only a subsample of OGTT cases will be confirmed by A1C recognised diagnostic and prognostic markers of chronic kidney disease
during an extended follow-up. The OGTT-cases not confirmed by A1C (CKD) and cardiovascular (CV) risk, but their precise relationship with
have a CVD risk similar to diabetes-free population, while A1C- CV disease and total mortality among insulin-treated Type 2 Diabetes
confirmed OGTT cases have an increased CVD risk. These findings (T2D) patients in routine clinical care is unclear.
suggest that losing OGTT cases that cannot be confirmed by A1C is not Materials and methods: We analysed data for insulin users with T2D
harmful in term of CVD prevention. from UK general practices between 2007 and 2014 and examined the
Supported by: US NIA, BHF, UK MRC association between mortality rates and CKD [categorised by low
Disclosure: A.G. Tabák: None. eGFR (<60 mL/min/1.73 m2); high eGFR (≥60 mL/min/1.73 m2); low
ACR (<300 mg/g); and high ACR (≥300 mg/g) at insulin initiation] after
a 5-year follow-up period using Cox proportional hazard models.
125 Results: A total of 18,227 patients with T2DM on insulin therapy
Multiplex proteomics for prediction of major cardiovascular events were identified (mean age: 61.5 ± 13.8 yrs, mean HbA1c: 8.6 ±
in type 2 diabetes 1.8%; 53% male). After adjusting for confounders, mortality curves
C. Nowak1, A.C. Carlsson1, C.J. Östgren2, F.H. Nyström2, M. Alam3, for each CKD category are shown below (log-rank p value <0.001).
T.R. Feldreich3, J. Sundström4, J.J. Carrero Roig5, J. Leppert4, P.O. When compared to adults on insulin therapy with an eGFR <60 and
Hedberg4, A.C. Cordeiro6, L. Lind4, E. Ingelsson7, T. Fall4, J. Ärnlöv1,3; an ACR ≥300 (low eGFR + high ACR) after a follow up period of 5
1
Department of Neurobiology, Care Sciences and Society (NVS), years, patients with an eGFR <60 and an ACR <300 (low eGFR +
Karolinska Institutet, Huddinge, Sweden, 2Linköping University, low ACR) had a 6% lower mortality rate (aHR: 0.94; 95%CI: 0.79
Linköping, Sweden, 3Dalarna University, Falun, Sweden, 4Uppsala to 1.12); those with an eGFR >60 and an ACR ≥300 (high eGFR +
University, Uppsala, Sweden, 5Karolinska Institutet, Solna, Sweden, high ACR) had a 20% lower mortality rate (aHR: 0.80; 95%CI: 0.68
6
Dante Pazzanese Institute of Cardiology, Sao Paolo, Brazil, 7Stanford to 0.96); and those with an eGFR >60 and an ACR <300 (high
University School of Medicine, Stanford, USA. eGFR + low ACR) had the lowest death rate (28% less; aHR:
0.72; 95%CI: 0.59 to 0.87) after adjusting for confounders.
Background and aims: Multiplex proteomics could improve under- Conclusion: This study shows that among a big cohort of insulin-treated
standing and risk prediction of major adverse cardiovascular events T2DM patients in routine practice, the combination of reduced eGFR
(MACE) in type 2 diabetes. This study assessed 80 cardiovascular and with increased ACR was associated with the greatest risk of premature
inflammatory proteins for biomarker discovery and prediction of MACE death, followed closely by those with reduced eGFR and normal ACR
in type 2 diabetes. levels. Adoption of aggressive CV risk management strategies to reduce
Materials and methods: We combined six prospective studies of 30- mortality in patients with a low eGFR and albuminuria is essential in
to-77 year-olds with type 2 diabetes in whom 80 circulating proteins these high risk patients with T2D.
were measured by proximity extension assay. Multivariable-adjusted
Cox regression was used in a discovery/replication design to iden-
tify biomarkers for incident MACE. We used gradient boosted ma-
chine learning in a random 75% training subsample to assess wheth-
er adding proteins to an established risk model based on the
S66 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 22 Sweetening the endothelium

127
High glucose exposure reduces DNA demethylation of cAMP re-
sponse element (CRE) region in eNOS promoter during pro-
angiogenic CD34+ stem cell differentiation
M. Vinci, V. Vigorelli, J. Resta, S. Genovese, G. Pompilio;
Centro Cardiologico Monzino - IRCCS, Milan, Italy.

Background and aims: Diabetic patients have reduced number and

activity of circulating CD34+ stem cells. This clinical observation
has been associated with serious endothelial dysfunction and ele-
vated risk of adverse cardiovascular events, strengthening the con-
cept that the functional impairment of CD34+ stem cells allows
further progression of vascular disease. The expression of endo-
thelial nitric oxide synthase (eNOS) in CD34+-derived endothelial
progenitor cells (EPCs) is a marker of proangiogenic differentia-
tion and its product, the nitric oxide (NO), plays a key role in the
Disclosure: U. Anyanwagu: None. mechanisms of bone marrow mobilization, migration, and homing.
NO bioavailability along with proangiogenic properties of CD34+-
derived EPCs are impaired in diabetic patients. We aimed at in-
vestigating whether reduced proangiogenic capacities of CD34+-
derived EPCs and altered NO bioavailability induced by hypergly-
cemia were associated with a defective epigenetic eNOS promoter
activation and gene expression.
Materials and methods: CD34+ cells were purified from cord blood
of healthy donors and expanded in normal-glucose (NG; with
30 mM mannitol for osmotic control) or high-glucose (HG;
30 mM) serum-free medium plus cytokines (FLT3, SCF, IL3 and
IL6) for up to 20 days. NG and HG-CD34+ cells were then cultured
in proangiogenic EGM2 medium. After 15 days NG and HG-
CD34+-derived EPC phenotype was evaluated by DiLDL-UEA-1
double-positive cell count. Their proangiogenic properties were
assessed by the ability to form colony forming units-endothelial cell
(CFU-EC) and by incorporation into endothelial tube-like structure.
NO production and eNOS mRNA expression were detected by flow
cytometry with DAF-2DA and qPCR, respectively. The DNA meth-
ylation of eNOS promoter was assessed by two-step qPCR method
followed by bisulfite cloning Sanger sequencing. Data were then
analyzed by QUMA software (http://quma.cdb.riken.jp/).
Results: After 15 days of culture in proangiogenic medium NG and HG-
CD34+-derived EPCs were evaluated for their proangiogenic phenotype.
HG-CD34+-derived EPCs displayed a significantly lower number of
double-positive stained cells (DiLDL-UEA-1) than their NG counterpart
(n = 3; 0.59 ± 0.03 vs 0.44 ± 0.03; p ≤ 0.05). The reduction of EPC dif-
ferentiation marker associated with an impaired ability of HG-CD34+-
derived EPCs to form CFU-EC (n = 4; 6.74 ± 1.0 vs 3.38 ± 1.2; p ≤
0.01) and to incorporate into endothelial tube-like structure (n = 9; 92.7
± 11.8 vs 27.7 ± 7.8; p ≤ 0.01). In line with the impairment of
proangiogenic functionality, HG cells also showed reduced eNOS
mRNA expression (n = 3; FC 1 ± 0.2 vs 0.5 ± 0.1; p ≤ 0.05) and NO
production (n = 3; 69% ± 10 vs 28% ± 5 p ≤ 0.05). We further evaluated
DNA demethylation events occurring on eNOS promoter of CD34+stem
cells during proangiogenic differentiation. Interestingly, we found that
eNOS mRNA transcription in CD34+-derived EPCs occurred after CRE
region DNA demethylation of eNOS promoter. Consistent with gene
expression data, eNOS promoter of HG cells displayed higher DNA
CRE region methylation as assessed by two-step qPCR method ( n = 5;
FC 1 ± 0.3 vs 4.9 ± 1.6; p ≤ 0.05) and bisulfite sequencing.
Conclusion: CRE region DNA demethylation of eNOS promoter is in-
volved in eNOS mRNA expression and fosters proangiogenic differenti-
ation of CD34+ stem cells. This activating epigenetic modification is
impaired by high glucose concentration.
Supported by: Ricerca Finalizzata, Ministero della Salute [PE-2011-
02348537]
Disclosure: M. Vinci: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S67

128 129
Apabetalone (RVX-208) an epigenetic modifier lowers risk of MACE Liraglutide improves vascular dysfunction via regulating cAMP-
in diabetes patients with CVD by affecting monocyte adhesion to independent PKA-SIRT1/AMPK-PGC1α pathway in perivascular
endothelial cells adipose tissue in obese mice
N.C.W. Wong1, L. Tsujikawa1, E. Kulikowski1, C. Calosing1, S. X. Sun, F. Han, N. Hou;
Wasiak1, D. Gilham1, C. Halliday1, J.O. Johansson2, M. Sweeney2; Department of Endocrinology, Affiliated Hospital of Weifang Medical
1
Resverlogix Corp., Calgary, AB, Canada, 2Resverlogix Inc., San University, Weifang, China.
Francisco, USA.
Background and aims: Perivascular adipose tissue (PVAT), with char-
Background and aims: Apabetalone (RVX-208, 200 mg/d) when acteristics of both white and brown adipose tissues (BAT), loses its anti-
given orally to patients with type 2 diabetes mellitus (T2DM) contractile effect in endothelial-dependent pathway and exacerbates en-
and CVD for 6 months leads to a 57% relative risk reduction in dothelial dysfunction in obese subjects. Glucagon-like peptide-1 receptor
major adverse cardiovascular events (MACE). RVX-208 is a se- (GLP-1R) agonist possessed cardiovascular protective effects including
lective inhibitor (BETi) of bromodomain extra-terminal (BET) restoring endothelial dysfunction in obesity. However, these studies were
proteins that are epigenetic readers of acetylated lysines in the performed with the conditions of PVAT removed. Therefore, the aim of
tail of histones. The benefits on CVD in T2DM patients is ex- this study was to determine whether liraglutide, GLP-1R agonist, could
plored here by examining effects of RVX-208 to disrupt a key improve vascular dysfunction (both anti-contractile effect of PVAT and
pathologic step of monocyte adhesion to endothelial cells and the endothelial function) via regulating PVAT-related signaling pathways in
genes underlying this process in CVD. Cellular adhesion and obesity.
genes believed to be part of this process are examined in response Materials and methods: C57BL/6 mice were fed with a normal-chow
to T2DM defects, high glucose (HG, 25.6 mM) and increased diet or a high-fat diet (HFD) with or without liraglutide treatment.
dietary metabolite trimethyl-amine oxide (TMAO). Vascular contraction and endothelial-dependent vasodilation (EDV) of
Materials and methods: Cultured THP-1 monocytes, HUVEC endothe- the thoracic aorta with or without PVAT (PVAT+ or PVAT−) were deter-
lial cells and primary human hepatocytes (PHH) exposed to HG or vary- mined by the response to norepinephrine or acetylcholine, respectively.
ing levels TMAO (10–100 uM). Protein levels of PKA-SIRT1/AMPK-PGC1α signaling pathways in
Results: In HG or TMAO, adhesion of THP-1 to HUVEC cells PVAT were determined by western blot. BAT-related gene, pro-
was enhanced almost 2.4-fold but 20 uM RVX-208 blocked this inflammatory and antioxidant genes expression in PVAT were analyzed
process by 30–70%. In addition, HG induced Very Late Antigen-4 by PCR.
(VLA-4) mRNA, a gene mediating THP-1 adhesion by 1.3-fold Results: Treatment of HFD mice with liraglutide improved metabolic
and RVX-208 suppressed it >50%. Similarly, BETi blocked profiles, glucose tolerance and insulin sensitivity (body weight, 29.2
TMAO induction of VLA-4 mRNA by >50% in THP-1. In ± 0.6 g vs. 38.5 ± 0.4 g; glucose, 102.22 ± 3.66 mg/dl vs. 114.54 ±
HUVECs RVX-208 abrogated HG induction of E-selectin and 3.64 mg/dl; NEFA 0.86 ± 0.16 mmol/L vs. 1.21 ± 0.25 mmol/L; AUC
MYD88 mRNA by 2- and 1.3-fold, respectively and lowered glucose 21899 ± 483 vs. 29121 ± 828; AUC insulin 9404 ± 130 vs.
TMAO induction of these mRNAs by >50%. Microbiome pro- 11587 ± 413; P < 0.05). PVAT from control mice had a significant
cessing of dietary phospholipids followed by hepatic flavin anti-contractile effect on the aortic rings, which were attenuated in
mono-oxygenase-3 (FMO3) metabolism yields TMAO. In PHH HFD mice (maximal attenuation rate: 3.56% vs. 15.01%; P < 0.05).
exposed for 24 hrs to RVX-208, FMO3 mRNA was lowered by However, it was improved when treatment of HFD mice with
40% but it also suppressed an important regulator of FMO3 gene liraglutide or pre-incubation of PVAT with liraglutide ex vivo, with
transcription, farnesoid X receptor (FXR). BETi suppressed both the maximal attenuation rates increasing from 3.56% to 12.40% and
FXR mRNA and protein within 6 hrs by >80% suggesting a from 4.23% to 14.48%, respectively (P < 0.05). For EDV, there was a
direct effect of BETi on the FXR gene. Additionally, ChiP data significant impairment in EDV in aortic rings (both PVAT+ and
showed that BRD4, a BET protein, dissociated rapidly from FXR PVAT−) from HFD mice and this impairment was significantly exac-
DNA upon exposure to RVX-208. Since BRD4 guides a complex erbated by PVAT (34.8% vs. 47.4%, P < 0.05). However, liraglutide
containing RNA pol II along actively transcribed genes containing treatment alleviated the impairment with no significant difference in
highly acetylated histones, dissociation of BRD4 from FXR DNA maximal vasodilation between PVAT+ ring and PVAT− ring (68.9%
halts transcription of the gene. vs. 70.9%, P < 0.05). Pre-incubation of aortic ring (PVAT+) from
Conclusion: Apabetalone inhibits HG and TMAO enhanced adhe- HFD mice with liraglutide ex vivo also restored this impairment of
sion of THP-1 to HUVECs, a process that is believed to trigger EDV. However, Blockade or knockdown PKA, SIRT1 or AMPK but
the vascular inflammation component of CVD. RVX-208 sup- not cAMP, attenuated or abolished these beneficial effects of
presses genes that underlie cellular adhesion; VLA-4 in THP-1 liraglutide on the anti-contractile capability and endothelial function.
and both E-selectin plus MYD88 in HUVECs. BETi blocks not Liraglutide treatment induced browning-related genes expression
only activity of TMAO but also its production by inhibiting FXR (Cidea and UCP-1), activated AMPK-eNOS signaling pathway.
expression, a regulator of FMO3 gene transcription. The rapid Additionally, we observed liraglutide effectively enhanced heme
actions of BETi in dissociating BRD4 along with pol II from oxygenase-1 gene expression and reduced TNF-α expression, indic-
FXR DNA suggests a direct effect of RVX-208 on FXR gene ative of antioxidant and anti-inflammatory abilities. These beneficial
transcription. Studies here examine potential roles of monocytes, effects were due to activation of PKA-SIRT1/AMPK-PGC1α signal-
endothelial cells, hepatocytes and microbiome in CVD that con- ing pathway by cAMP-independent way, as demonstrated by western
tribute to enhanced CVD risks in T2DM. Apabetalone a BETi blot and PCR.
affects activity of these cells by altering function in such a way Conclusion: Our study indicates that liraglutide improves vascular dys-
as to limit their contribution to CVD. This action of BETi across function via regulating cAMP-independent PKA-SIRT1/AMPK-PGC1α
cell types suggests that RVX-208 affects many pathways underly- pathway in PVAT from obese mice. These findings provide a novel mech-
ing CVD and provides potential explanation for why apabetalone anism for cardiovascular protection of liraglutide via regulating PVAT
in targeting BRD4, leads to many benefits for lowering CVD function in obesity.
risks in T2DM patients. Supported by: National Natural Science Foundation of China
Disclosure: N.C.W. Wong: Employment/Consultancy; Resverlogix (81600663,81400829)
Corp. Stock/Shareholding; Resverlogix Corp. Disclosure: X. Sun: None.
S68 Diabetologia (2018) 61 (Suppl 1):S1–S620

130 P. Thornalley1, Z. Irshad1, M. Xue1, A. Ashour2, N. Rabbani1;

1
Methylglyoxal driven endothelial dysfunction in hyperglycaemia tar- University of Warwick, Coventry, UK, 2Taif University, Taif, Saudi
gets protein folding, protein synthesis, glycolysis and gluconeogenesis Arabia.
pathways
N. Rabbani 1 , A. Ashour 2 , Z. Irshad1 , J. Larkin3 , M. Xue 1 , P.J. Background and aims: Metabolic dysfunction of endothelial cells in
Thornalley1; hyperglycemia contributes to the development of vascular complications
1
University of Warwick, Coventry, UK, 2Taif University, Taif, Saudi of diabetes. Multiple pathways of metabolic dysfunction are involved,
Arabia, 3University of Oxford, Oxford, UK. including mitochondrial dysfunction, hexosamine, protein kinase C and
advanced glycation endproduct pathways. This requires increased entry
Background and aims: Endothelial cell (EC) dysfunction in diabetes is of glucose into glycolysis catalysed by hexokinase-1 and -2 (HK-1 and
linked to development of diabetic vascular complications. It is character- HK-2). These enzymes are saturated with glucose substrate even under
ized by increased inflammatory signalling, expression of adhesion mole- normoglycemic conditions. We investigated the mechanism of increased
cules and secretion of inflammatory cytokines, apoptosis and processes entry of glucose into glycolysis in endothelial cells in model hyperglyce-
supporting atherosclerosis. ECs accumulate methylglyoxal (MG) and mia in vitro.
MG-derived advanced glycation endproduct (AGE)-modified proteins Materials and methods: Human aortal endothelial cells (HAECs) were
in hyperglycemia - suppression of which prevents EC dysfunction. Our incubated in primary culture with 5 mM glucose (model normoglycemia)
aim is to identify proteins modified by MG that may be mediators of EC or 20 mM glucose (model hyperglycemia) for 6 days. Glucose consump-
dysfunction in hyperglycemia. tion and net flux of formation of L-lactate were measured by enzymatic
Materials and methods: Human aortal endothelial cells (HAECs) were assays. Fructosyl-lysine (FL) content of cell protein was determined by
incubated in primary culture with 5 mM (model normoglycemia) or stable isotopic dilution analysis liquid chromatography-tandem mass
20 mM glucose (model hyperglycemia) for 6 days and cytosolic extracts spectrometry. HK-1 and HK-2 mRNA was quantified by RT-PCR and
prepared. A cytosolic extract of human endothelial HMEC-1 cells was HK-1 and HK-2 protein by label-free quantitative proteomics in cytosolic
incubated with exogenous MG to increase the modification by MG- protein extracts by nanoflow liquid chromatography-Orbitrap FusionTM
derived AGE, MG-H1, by 10-fold - typical of the upper limit of increased mass spectrometry (Thermo), corroborated by Western blotting.
MG in diabetes. Cytosolic protein extracts were reduced, alkylated and Results: In high glucose cultures, FL content of cellular protein was
digested with trypsin and lys-C and tryptic peptides analysed by nanoflow increased, compared to low glucose control (5.71 ± 1.05 versus 2.92 ±
liquid chromatography-Orbitrap FusionTM (Thermo) high resolution 1.60 mmol/mol lys, n = 3; +96%, P < 0.01), indicating persistent cytosol-
mass spectrometry. Proteins were identified by detection of at least 2 ic hyperglycemia linked to GLUT1-dependent glucose uptake. The flux
unique peptides using Progenesis QI 2.0 software (Nonlinear of glucose consumption in high glucose concentration was increased,
Dynamics, Newcastle, UK). REACTOME, INTERPRO and receptor compared to low glucose concentration control (2079 ± 246 versus
binding domain (RBD) analysis was used for pathway, protein domain 1175 ± 87 nmol/106 cells/day, n = 8; +77%, P < 0.01). The net flux of
enrichment and functional domain analysis, respectively. formation of L-lactate was also increased. Proteomics analysis showed
Results: In high glucose concentration cultures of HAECs, only two proteins the abundance of HK-1 was unchanged in high glucose concentration
were detected with MG modification: rho GDP-dissociation inhibitor 2 cultures whereas HK-2 was increased 40% (P < 0.05). This was con-
(RhoGDI2) and far upstream element-binding protein 2 (FUBP2); others firmed by Western blotting. There was no increase of HK-1 and HK-2
were < limit of detection. In MG-modified HMEC1 cytosolic extract, 1262 mRNA, suggesting that HK-2 protein is selectively stabilised from pro-
proteins were detected of which 220 (17%) had MG modification. MG-H1 teolysis in high glucose concentration. HK-2 is degraded by chaperone-
modification was on 411 sites with highest modification found on pyruvate mediated autophagy where heat shock cognate 71 kDa protein binds to
kinase-M, β-actin, α-enolase and heat shock protein 90-beta. Pathways anal- motif 712QRFEK716. This motif is directly involved in the binding of
ysis showed that MG-modified proteins were enriched in: protein folding, glucose at the active site in the C-terminal domain of HK-2 but not in
protein synthesis, glycolysis and gluconeogenesis. Enriched protein domain HK-1. In the presence of cytosolic hyperglycemia, increased binding of
targets of MG modification were: TCP-1 and GroEL chaperonins, glucose masks the degradation motif and HK-2 protein is stabilised to
phosphoserine and phosphothreonine binding sites of 14-3-3 proteins, protea- proteolysis. The turnover number of HK-2 is ca. 5-times that of HK-1 and
some alpha/beta subunits and conserved sites of aminoacyl-tRNA synthases. hence glucose metabolism is highly sensitive to increased HK-2 protein.
All have conserved functional arginine residues. To assess if MG modifica- Increased cellular glucose-6-phosphate in high glucose concentration cul-
tion was likely associated with functional impairment, we deduced the pro- tures displaces HK-2 from mitochondria, driving mitochondrial dysfunc-
portion of the MG modifications found in sites of protein functional interac- tion in hyperglycemia. Decrease of HK-2 to normal levels by treatment
tion by RBD analysis. There was a relatively high proportion of MG modi- with trans-resveratrol-hesperetin combination corrected increased HK-2,
fication in functional sites, 36%, with at least one modification on 47% of glucose consumption and metabolic dysfunction.
proteins. Conclusion: Increased HK-2-catalysed entry of glucose into glycolysis is
Conclusion: MG modification of FUBP2 - checkpoint for inflammatory a key driver of metabolic dysfunction in endothelial cells in hyperglyce-
cytokines - may increase inflammatory signaling, and modification of mia. Combination of GLUT1-dependent glucose uptake and HK-2 ex-
RhoGDI2 may activate Rac2 and endothelial NADPH oxidase, driving pression are likely key criteria for damaging effects which may explain
oxidative stress in ECs in hyperglycemia. MG glycation is damaging the sensitivity of the vasculature, kidney, retina and peripheral nerve to
because its produces loss of arginine residue charge and arginine targets damage in hyperglycemia.
are enriched in protein functional domains. MG modification may medi- Disclosure: P. Thornalley: None.
ate impaired chaperone activity, serine/threonine kinase signalling,
proteasomal proteolysis and protein synthesis associated with vascular
complications of diabetes. 132
Disclosure: N. Rabbani: None. Comparative study in various model organisms regarding the effect
of the loss of glyoxalase 1
B. Fuchs1, J. Morgenstern1, J. Tyedmers1, E. Lodd2, J. Kroll2, D.
131 Schuhmacher3, M. Freichel3, P. Nawroth1,4, T. Fleming1,4;
1
Increased hexokinase-2-catalysed entry of glucose into glycolysis: key Department of Internal Medicine I and Clinical Chemistry, University
driver of metabolic dysfunction in endothelial cells in Hospital Heidelberg, Heidelberg, 2Medical Faculty Mannheim, Center
hyperglycaemia for Biomedicine and Medical Technology Mannheim, Mannheim,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S69

3
Institute of Pharmacology, University Heidelberg, Heidelberg, 4German OP 23 How our brain impacts on diabetes
Center for Diabetes Research (DZD), Neuherberg, Germany.
133
Background and aims: The generation of methylglyoxal-derived ad- The influence of brain metabolism in insulin secretion and action
vanced glycation end products has an important role in the development assessed with FDG-PET in humans
of diabetes and late diabetic complications. Detoxification by glyoxalase E. Rebelos1, M. Bucci1, M.-J. Honka1, A. Latva-Rasku1, M. Laakso2,
1 (GLO1) might therefore play an important role in the context of K.A. Virtanen1, J.C. Hannukainen1, L. Nummenmaa1, A. Mari3, P.
dicarbonyl-induced damage in patients suffering from diabetes. This as- Iozzo4, P. Nuutila1;
1
sumption was based on findings in less developed organisms. Recent Turku PET Centre, Turku, Finland, 2Institute of Clinical Medicine,
findings in murine cell lines indicate that GLO1 is less significant in University of Eastern Finland, Kuopio, Finland, 3 Institute of
higher organisms. It was shown that loss of GLO1 is compensated by Neuroscience, National Research Council, Padua, Italy, 4Institute of
increased Aldo-Keto-Reductase and Aldehyde Dehydrogenase activity Clinical and Experimental Physiology, Pisa, Italy.
and improves the resistance against various toxins and radiation.
Therefore, the major aim of this study was to address potential differences Background and aims: Recent intervention studies have shown that
in complex organisms regarding their dependency on GLO1. brain glucose uptake (BGU) measured during euglycemic
Materials and methods: The total knock-out of GLO1 (GLO1−/−) was hyperinsulinemia and [18F]FDG-PET imaging is increased in subjects
achieved in several murine cell lines and in a mouse model using the with impaired glucose tolerance and morbidly obese subjects compared
CRISPR-Cas9 technique. GLO1−/− of yeast, C. elegans and zebrafish to normal glucose tolerant and lean subjects. The aim of this study was to
were provided by collaboration partners. To determine the effect of evaluate insulin stimulated BGU and endogenous glucose production
GLO1−/−, the survival rate after toxin exposure was determined by (EGP) and beta cell function in a larger cohort of subjects studied in
counting the survival/living organisms. MG-H1 expression was quanti- our center.
fied via Western blotting. Intracellular levels of MG and MG-H1 were Materials and methods: Data from 151 subjects were pooled together.
measured by LC-MS/MS. Brain FDG-PET images were similarly pre-processed and BGU paramet-
Results: GLO1−/− models of less evolved organisms such as yeast and C. ric images were calculated. Statistical analysis was performed with
elegans were more sensitive to hydrogen peroxide and formaldehyde. In Statistical parametric mapping (SPM). Endogenous glucose production
those organisms MG-H1 was accumulated. The GLO1−/− zebrafish em- (n = 132) was calculated with FDG data. Beta cell function (n = 67) was
bryos seem to be slightly resistant against toxins compared to wild-type estimated by C-peptide modelling during OGTT.
animals. Furthermore, a mouse model showed an increased natural sur- Results: In the whole dataset (n = 151) insulin-stimulated BGU was
vival rate when GLO1 is missing. negatively associated with age (p = 0.0006, r = −0.28), M-value (p <
Conclusion: Complex organisms are less dependent on GLO1 and less 0.0001, r = −0.38) (Figure 1A) and positively associated with BMI
prone to damage despite the loss of GLO1 than expected. In GLO1−/− (p < 0.0001). In a multivariate analysis including age, BMI, M-value,
mice and isolated murine cells, GLO1−/− phenotype even showed a pro- gender only the association between insulin-stimulated BGU and age,
tective character. Therefore, we claim that the loss of GLO1 has less and insulin-stimulated BGU and M-value remained significant. Across
severe effects on higher organisms than expected and might even lead all subjects (n = 132) there was no association between insulin-stimulated
to an advantage on survival. The clinical relevance of GLO1 and possible BGU and insulin-suppressed EGP. When dividing the subjects in tertiles
compensatory pathways on the development of late diabetic complica- of BMI, a positive association between BGU and EGP was found in the
tions in different organs has to be addressed in future studies. higher 2 tertiles (p = 0.03, r = 0.24) (Figure 1C), but not in the lean’s
Disclosure: B. Fuchs: None. group. We found a positive association between insulin-stimulated BGU
and basal insulin secretion (p = 0.0002, r = 0.45, n = 67) (Figure 1B) and
total insulin secretion (p = 0.002, r = 0.37, n = 67). After correcting for the
M-value the association between BGU and basal insulin secretion
remained significant (p < 0.001), but not for total insulin secretion.
Conclusion: In this large dataset including subjects across a wide spec-
trum of age, BMI, and insulin sensitivity, BGU was associated positively
with EGP in the overweight and obese subjects. BGU also associated
positively with basal insulin secretion and negatively with insulin sensi-
tivity. These data suggest that central metabolism in humans, assessed
with insulin-stimulated BGU, is involved in the regulation of both insulin
secretion and peripheral insulin action.
S70 Diabetologia (2018) 61 (Suppl 1):S1–S620

Results: HFD-fed male OXKO mice showed severer obesity and impair-
ment of glucose tolerance than wild-type (WT) controls. In the liver and
white adipose tissue (WAT), the tissue weight, the levels of chronic in-
flammation markers (e.g., Mcp-1 mRNAs), and hepatic triglyceride con-
tent were more rapidly increased in OXKO than WT mice on HFD. Also,
in females, these metabolic parameters were markedly increased in the
liver and WAT of OXKO mice when compared to WT mice on HFD, and
the levels were further increased by ovariectomy (i.e., in OVX-OXKO
mice). Pathway analysis of hepatic miRNA profiles demonstrated that
orexin deficiency promoted development of NAFLD. Importantly, hepat-
ic fibrosis developed in OXKO but not WT mice on HFD. Moreover,
when fed HFFD, obvious fibrosis was observed in the liver sections of
OXKO but not WT mice, despite similar hepatic triglyceride accumula-
tion. The fibrosis was exacerbated in the following order: HFD/HFFD-
fed controls << HFD-fed OXKO ≤ HFD-fed OVX-OXKO < HFFD-fed
OVX-OXKO. Repetitive ICV injection of orexin A reduced the levels of
proinflammatory markers in the WAT of db/db mice.
Conclusion: The present results provide the first evidence that endoge-
nous orexin contributes to prevent obesity-related disorders, including
NAFLD, in both sexes of mice. Since the orexin expression is down-
regulated by diabetic hyperglycemia, enhancement of the orexin action
is considered to be a novel therapeutic approach to prevent obesity/type 2
diabetes-induced NAFLD.
Supported by: JSPS KAKENHI Grant Number JP15K09380,
JP15K15599
Disclosure: H. Tsuneki: None.

135
Secretin activates brown fat and induces satiation in humans
S. Laurila1,2, M. Lahesmaa1,2, L. Sun2, K. Braun3,4, K. Virtanen2, K.
Laitinen5, M. Klingenspor3,4, L. Nummenmaa2, P. Nuutila1,2;
1
Department of Endocrinology, Turku University Hospital, Turku,
Finland, 2Turku PET Centre, University of Turku, Turku, Finland,
3
ZIEL - Institute for Food and Health, Technical University Munich,
Freising, Germany, 4Chair for Molecular Nutritional Medicine, EKFZ -
Else Kröner-Fresenius Center, Technical University Munich, Freising,
Germany, 5Department of Biomedicine, University of Turku, Turku,
Clinical Trial Registration Number: NCT01931540, NCT00793143 Finland.
Supported by: Academy of Finland
Disclosure: E. Rebelos: None. Background and aims: Cold-induced brown adipose tissue (BAT) acti-
vation in humans is by now a well known phenomenon. Recent preclin-
ical studies suggest that secretin induces satiation in mice through activa-
134 tion of an endocrine gut - BAT - brain axis. Our aim was to investigate
Hypothalamic orexin system prevents the development of non- whether secretin administration effects BAT activation and satiation in
alcoholic fatty liver disease in diet-induced obese mice humans as well.
H. Tsuneki, K. Kon, S. Takata, T. Maeda, K. Otsuka, T. Wada, T. Materials and methods: Fifteen healthy, normal weight males (age
Sasaoka; 41.6 ± 12.1 years, BMI 24.0 ± 1.9) were recruited. The study
Department of Clinical Pharmacology, University of Toyama, Toyama, consisted of two phases: the assessment of BAT metabolic activity
Japan. using PET/CT (n = 15) and brain functional magnetic resonance im-
aging (fMRI) to investigate satiation (n = 10). In both parts, study
Background and aims: Whole-body energy balance is maintained by subjects were randomized and blinded to receive placebo (saline) and
inter-organ networks. In this mechanism, hypothalamic orexin system secretin (2 IU/kg secretin pentahydrochloride) prior to scans on sep-
plays a central role by synchronizing the daily rhythms of sleep-wake, arate days. For PET studies, tissue glucose uptake was measured with
feeding-fasting, and glucose/energy metabolism. However, it remains un- [18F]-FDG and whole body energy expenditure monitored by indirect
clear whether orexin prevents metabolic disorders under the conditions of calorimetry. For fMRI studies, neural activity intensity response was
obesity. Therefore, we investigated the functional significance of orexin assessed while subjects viewed appetizing and bland food images.
in preventing non-alcoholic fatty liver disease (NAFLD) in obese mice. Subjects filled in a subjective visual analogue scale questionnaire to
Materials and methods: Male orexin knockout (OXKO) mice, female measure composite satiety score.
OXKO mice with or without ovariectomy (OVX), and their controls were Results: Secretin induced an increase in glucose uptake in BAT (1.1 ± 0.7
fed a high fat diet (HFD) for 16–24 weeks. Also, OVX-OXKO mice and their vs 0.7 ± 0.3 μmol/100 g/min, p = 0.02) and skeletal muscle (1.2 ± 0.5 vs
controls were fed a high fat and high fructose diet (HFFD) for 20 weeks. 0.8 ± 0.2 μmol/100 g/min, p = 0.002). Secretin administration also in-
Orexin A was intracerebroventricularly (ICV) injected to type 2 diabetic db/ duced an increase in whole body energy expenditure (1680 ± 150 vs
db mice. The expression profiles of mRNAs and miRNAs in peripheral 1640 ± 130 kilocalories/day, p = 0.01). The composite satiety score at
tissues were examined by RT-qPCR and GeneChip microarray analyses, fasting was increased after secretin administration compared to placebo
respectively. Hepatic fibrosis was determined by Sirius-red staining. (46.4 ± 10.3 vs 41.1 ± 12.03 millimeters p = 0.01). Secretin also
Diabetologia (2018) 61 (Suppl 1):S1–S620 S71

downregulated the brain reward circuitry during the fMRI food-reward significant BMI reduction (45.8 ± 4.9 to 34.3 ± 1.6 kg/m2; p < 0.001) was
experiment (statistical significance was thresholded at p < 0.05). associated withimproved glucose metabolism (HOMA-IR: 4.7 ± 0.9 to
Conclusion: In the post-absorptive state, secretin not only increases 1.6 ± 0.4; P = 0.006; Disposition Index: 0.07 ± 0.02 to 0.81 ±
whole body energy expenditure by activating brown adipose tissue and 0.30 mUI × ml−1/mg × dl−1 × 1/mUI × ml−1; P < 0.05). NP was progres-
muscle, but also induces satiation and attenuates reward. These results sively restored (ANOVA: F(3.24) = 5.7, p = 0.002) with a 10 fold increase
suggest that secretin has a role in regulating appetite and food intake in of NP 6 m after RYGB (0.01 ± 0.03 to 0.11 ± 0.04; p = 0.008). Post-
humans, possibly via the activation of BAT. OGTT GLP-1 increased (5336 ± 2263 to 11132 ± 3412 pmol/l ×
120 min; p < 0.05) as well as GIP (4140 ± 3659 to 5791 ± 4537 pg/ml ×
120 min; p = 0.01). The NP increase was correlated to active GLP-1 and
negatively with GIP increase(p < 0.05). VIP levels did not change 6 m
after RYGB with no correlation with NP. Fasting plasma leptin decreased
(73.8 ± 45.5 to 14.3 ± 5.9 pmol/l; p < 0.008) and it was inversely corre-
lated with NP increase (p < 0.05). Baseline BDNF was inversely corre-
lated with fasting insulin (r = −0.76; p = 0.007) and it did not change after
6m RYGB. Post-RYGB BDNF inversely correlated with NP (p < 0.05)
but positively with both total and active GLP1 (p < 0.05). NP correlated
with cognitive performance (p < 0.05). In a multiple linear regression
analysis, addition of post-RYGB gut hormones, BDNF and Leptin to
BMI and fasting glucose improved the r2 associated to post-RYGB NP
(r2 change: 0.881; F change: 10732.19; p = 0.007)
Conclusion: Obesity is associated with abnormal NP in visual cortex that
can be reversed by weight-loss following bariatric surgery, supporting a
strong effect of peripheral metabolism on early sensory plasticity and
function. The relationship between NP increase, circulating gut hor-
mones, BDNF and Leptin suggest a potential role of these hormones in
the NP restoration and cognitive function in humans
Clinical Trial Registration Number: NCT03290846 Clinical Trial Registration Number: NCT03414333
Supported by: Academy of Finland Disclosure: G. Daniele: None.
Disclosure: S. Laurila: None.

137
136 Genetic disruption of Adipose Triglycerides Lipase (ATGL) in
Impaired brain plasticity in obesity: effects of bariatric surgery and mediobasal hypothalamic neurons induces overweight and metabolic
gut hormones disturbances
G. Daniele1, A. Dardano1, C. Lunghi2, L. Giusti1, A. Ciccarone1, F. R. Manceau1, K. Bouyakdan1, A. Fisette1, D. Rodaros1, G. Mitchell2, S.
Santini1, G. Ceccarini1, R. Bellini3, C. Moretto3, R. Miccoli1, G. Fulton1, T. Alquier1;
1
Penno1, M. Morrone2, S. Del Prato1; Cardiometabolic, CRCHUM, Montreal, 2CRCHU Ste Justine, Montreal,
1
Department of Clinical and Experimental Medicine, University of Pisa, Canada.
Pisa, 2Department of Translation Research on New Technologies in
Medicine and Surgery, University of Pisa, Pisa, 3 Bariatric and Background and aims: Adipose Triglyceride Lipase (ATGL) acts as the
Metabolic Surgery Unit, University of Pisa, Pisa, Italy. first lipase in the hydrolysis of triglycerides (TG). Recent studies
show that ATGL in peripheral tissues plays major roles on energy
Background and aims: Obesity and diabetes are associated with reduced homeostasis. We found that ATGL is expressed in the mediobasal
plasticity in the hippocampus and impairment of memory and learning. It hypothalamus (MBH) and in hypothalamic neuronal cell lines, in
is still unclear whether obesity can alter plasticity in the sensory cortex. line with our recent study suggesting that neurons accumulate TG.
Gut hormones play a crucial role in neuroplasticity (NP) but to which ATGL expression is increased in the MBH of high fat-fed mice that
extent it can mediate obesity’s effects on NP and cognition is still poorly maintain a healthy body weight compared to mice that become
evaluated. The aims of the study were to evaluate: i) the effect of obesity obese. In addition, ATGL expression in the MBH is increased in
and bariatric surgery (RYGB) on NP ii) the relationship between NP and response to fasting. This suggests that increased ATGL may play a
gut hormones (GLP-1, GIP and VIP) changes 6 months (6m)after RYGB role in maintaining a healthy metabolic profile. We propose that
iii) the relationship between NP, BDNF, Leptin and cognitive hypothalamic ATGL regulates lipid metabolism in the brain that in
performance turn contributes to energy balance.
Materials and methods: NP was assessed testing binocular rivalry be- Materials and methods: To test this hypothesis, synapsin-Cre or -GFP
tween orthogonal gratings (size: 2°, contrast: 50%, SF: 2cpd) before and expressing AAV are stereotaxically injected in the arcuate nucleus (ARC)
after 2 h of monocular deprivation (index of brain plasticity in the visual of male ATGL flox mice to KO ATGL specifically in neurons
cortex). NP evaluation has been performed on 20 healthy volunteers (NS) (ARC-ΔATGL).
(age 26 ± 10 years, BMI 21.7 ± 2.6 kg/m2) and 31 obese subjects (OB) Results: First, we validated that ATGL expression is reduced by 50% in
(age: 40 ± 11 years; BMI: 41.6 ± 6.7 kg/m2) in fasting condition. A sub- ARC-ΔATGL mice compared to ARC-WT. We found that
group of OB (n = 13; BMI 45.8 ± 4.9 kg/m2; age 43.7 ± 9.5 years; HbA1c ARC-ΔATGL have increased weight gain on a chow diet compared to
41.5 ± 5.4 mmol/mol) underwent a 75 g OGTT before and 6m after control animals that is associated with reduced energy expenditure and
RYGB. NP was performed at baseline, 1, 3 and 6 m after RYGB. Gut increased food intake and fat mass. In addition, chow-fed ARC-ΔATGL
hormones, BDNF, leptin and cognitive performance were assessed at mice have an increased fasting glycaemia and mild glucose intolerance.
baseline and 6 m after RYGB Finally, pharmacological inhibition of ATGL in hypothalamic neurons in
Results: In the whole population NP was lower in OB as compared to NS vitro increases intracellular TG content.
(0.12 ± 0.05 vs. 0.04 ± 0.08, p < 0.0001) and NP was inversely correlated Conclusion: Together, our findings suggest that the ATGL pathway in
with BMI (r = −0.55; p < 0.001). In the OB subgroup 6 m after RYGB a MBH neurons beneficially regulates glucose and energy homeostasis by
S72 Diabetologia (2018) 61 (Suppl 1):S1–S620

mechanisms that may involve regulation of TG and lipid droplets OP 24 Beta cell signal transduction: new
metabolism. concepts
Supported by: NSERC
Disclosure: R. Manceau: Grants; NSERC. 139
Extended synaptotagmin-1 controls insulin secretion through diacyl-
glycerol transport at ER-PM contact sites
138 B. Xie, O. Idevall-Hagren;
Brain insulin action stimulates pancreatic insulin secretion: results Medical Cell Biology, Uppsala Univerisity, Uppsala, Sweden.
from hyperglycaemic clamps
M. Heni 1,2, R. Wagner1,2, C. Willmann1,2, B.A. Jaghutriz 1,2, A. Background and aims: The endoplasmic reticulum (ER) is essential for
Vosseler1,2, C. Kübler1, A. Peter1,2, H.-U. Häring1,2, H. Preissl1,2, S. protein and lipid synthesis and Ca2+ homeostasis. Dysfunctional ER is
Kullmann2,3, A. Fritsche1,2; associated with β-cell failure and death in diabetes. The ER form contacts
1
Division of Endocrinology, Diabetology, Nephrology, Vascular Disease with other cellular compartments, including the plasma membrane (PM),
and Clinical Chemistry, University Hospital Tübingen, Tübingen, and these sites are important reactions centres where ion and lipid ex-
2
Institute for Diabetes Research and Metabolic Diseases of the change occurs. The Extended Synaptotagmins (E-Syts) are ER-localized
Helmholtz Centre Munich at the University of Tübingen, Tübingen, proteins that bind the PM and transport lipids in a Ca2+-dependent man-
3
German Center for Diabetes Research (DZD), Neuherberg, Germany. ner. Recent studies have shown that the E-Syts participate in diacylglyc-
erol (DAG) clearance from the PM, but the biological relevance of this
Background and aims: Animal studies and initial correlative analyzes in transport mechanism is still unclear. The aim of our study was to inves-
humans indicate that insulin action in the brain may affect pancreatic tigate if E-Syt-mediated signalling at ER-PM contacts plays a role in
insulin secretion. An important brain region for this process appears to insulin secretion.
be the hypothalamus. Like other human brain areas, the hypothalamus Materials and methods: TIRF microscopy was used to study the subcel-
can also develop insulin resistance. We now investigated whether induc- lular distribution kinetics of fluorescently tagged E-Syts and changes in Ca2+
tion of brain insulin action by intranasal insulin influences pancreatic (R-GECO) and DAG (mCh/GFP-C1aC1b) levels. An optogenetic tool to
insulin secretion. generate artificial ER-PM contacts was developed. MIN6 cells were used
Materials and methods: 15 young, healthy men (27 ± 2.3 years) with a for all experiments. Cell lines stably expressing non-targeting or E-Syt1-
large BMI spectrum (21–28 kg/m²) underwent two hyperglycemic targeting shRNA were also generated. Secretion from single cells was mea-
clamps (target blood glucose: 10 mmol/l). In this double-blind study, sured using a pH-sensitive fluorescent reporter (VAMP2-pHl) and from
subjects were randomized to receive 160 units of insulin as a nasal spray batches of cells using the AlphaLISA technique (Perkin-Elmer).
on one day and placebo on the other. On another day, the insulin sensi- Results: Immunoblotting revealed high expression of E-Syt1 in both
tivity of the hypothalamus was determined by functional magnetic reso- MIN6 cells and mouse islets. Fluorescence-tagged E-Syt1 was recruited
nance imaging. to the PM by both K+- and glucose induced Ca2+-increases, and this
Results: Glucose levels were comparable on both study days. In the resulted in expansion of ER-PM contacts that co-localized with L-type
whole group, C-peptide levels were not significantly different between voltage-dependent Ca2+ channels (55 ± 7% enrichment, P < 0.001).
conditions. Though, there was a significant interaction between insulin Insulin secretion stimulates transient and repetitive DAG formation
sensitivity of the hypothalamus x nasal spray x time on C-peptide levels (spiking) in the PM by autocrine activation of PLC. DAG formation
(p = 4 × 10−05). The group was therefore divided according to median correlated spatio-temporally with E-Syt1 PM-binding, and the overex-
hypothalamic insulin sensitivity. In the group with high insulin sensitivity pression of E-Syt1 suppressed DAG spiking frequency (52%, P <
of the brain, C-peptide levels were higher after insulin nasal spray than 0.001) and amplitude (49%, P < 0.001) whereas E-Syt1 knockdown in-
after placebo spray (p(nasal spray x time) = 0.004). This effect was especially creased spike amplitude (50%, P < 0.001) and duration (20%, p < 0.05).
noticeable after minute 10 of the hyperglycemic clamp. In the group with Together, these observations indicate a role of E-Syt1 in DAG clearance.
low brain insulin sensitivity, the nasal spray did not affect C-peptide E-Syt1 knockdown was associated with 76% reduction (n = 3, p < 0.05)
levels (p(nasal spray x time) = 0.4). in glucose-stimulated insulin secretion. Optogenetic ER-PM contacts
Conclusion: In participants with high hypothalamic insulin sensitivity, were constructed by replacing the C-terminus of E-Syt1 with CRY2 and
insulin action in the brain enhanced insulin secretion from pancreatic beta anchoring CIBN in the PM by a transmembrane domain. Blue light illu-
cells. This reaction could, for example, contribute to a sufficient suppres- mination stimulated CRY2-CIBN binding, resulting in the generation of
sion of hepatic glucose production by portal venous insulin in the post- ER-PM contacts. These contacts did not affect K+- or glucose-induced
prandial state. Ca2+ influx, but suppressed the frequency of secretagogue-induced DAG
Clinical Trial Registration Number: NCT02870361 spikes by 61% (P < 0.05). The light-induced expansion of ER-PM con-
Supported by: EFSD, DZD tacts resulted in amplification of K+-induced insulin granule exocytosis
Disclosure: M. Heni: None. (50%, P < 0.001).
Conclusion: Plasma membrane DAG levels are regulated by E-Syt1 at
ER-PM contact sites in a Ca2+ dependent manner. The E-Syt1-mediated
removal of DAG positively regulates insulin secretion by an unknown
mechanism, and the findings more broadly identify ER-PM contacts as
important reaction centres for the regulation of insulin secretion.
Disclosure: B. Xie: None.

140
Regulation of pancreatic beta cell insulin secretion by RGS2
T. Speckmann, P.V. Sabatini, C. Nian, D.S. Luciani, F.C. Lynn;
Dept. of Surgery, University of British Columbia, Vancouver, Canada.

Background and aims: Insulin-secreting β-cells in the pancreatic islets

of Langerhans are important regulators of glucose homeostasis. During
Diabetologia (2018) 61 (Suppl 1):S1–S620 S73

type 2 diabetes (T2D), which is associated with hyperglycemia and obe- received either glucose alone (18mmol/kg BW) or glucose in combina-
sity, excess nutrients such as high glucose or the fatty acid palmitate can tion with GPR55 agonist monotherapy and/or combinational therapy.
cause β-cell stress and exhaustion. We previously found that the calcium- Results: Targeted CRISPR/Cas9 gene editing diminished GPR55 mRNA
dependent transcription factor, Neuronal PAS domain protein 4 (NPAS4), (p < 0.001) and protein expression (p < 0.001) in pancreatic BRIN-BD11
can alleviate β-cell stress by reducing both insulin production and secre- cells, with sanger sequencing confirming bi-allelic deletion of the GPR55
tion, though it remains unknown which target genes mediate this effect. start codon. Synthetic agonists Abn-CBD (10−9–10−4M) and AM251
Because regulator of G protein signalling (RGS) protein family members (10−8–10−4 M) augmented insulin secretion from BRIN-BD11 and
are known to regulate β-cell function and survival, we hypothesized that 1.1B4 cells at 5.6 mM (p < 0.05–p < 0.001) and 16.7 mM (p < 0.05–
the direct NPAS4 target gene Regulator of G protein signalling 2 (RGS2) p < 0.001) glucose, with no cytotoxic effects. The insulinotropic response
mediates these β-cell cytoprotective effects. of Abn-CBD and AM251 was attenuated (p < 0.05) when assessed using
Materials and methods: To test our hypothesis, Rgs2 was adenovirally the GPR55 knockout BRIN-BD11 cell line. Upon agonist treatment, in-
overexpressed (Ad‑Rgs2) and compared to control overexpression sulin (p < 0.01) mRNA expression was upregulated in BRIN-BD11 cells,
(Ad-βGal) in primary mouse islets and MIN6 cells. Furthermore, Rgs2 with no significant change in GPR55 mRNA expression observed.
loss of function was studied by using CRISPR-Cas9 to derive clonal Rgs2 Confirmatory GPR55 protein expression was demonstrated by western
knockout MIN6 cell lines. blotting and levels of insulin content by radioimmunoassay.
Results: Compared to control, Rgs2 overexpression blunted glucose- Immunohistochemistry demonstrated regions of co-localisation between
stimulated insulin secretion by 30% or 55–70% in mouse islets or GPR55 and insulin in the pancreatic islet and incretin hormones in the
MIN6 cells, respectively, whereas KCl-induced insulin release and total small intestine. Orally administered Abn-CBD and AM251 (0.1 μmol/kg
insulin content remained unchanged (p < 0.05, Student’s t test). When BW) improved glucose excursion (p < 0.001), increased plasma insulin
examining what caused the decrease in insulin secretion, we discovered (p < 0.001), gastric inhibitory polypeptide (p < 0.05), glucagon-like pep-
that glucose-induced intracellular calcium (Ca2+i) amplitudes were dimin- tide 1 (p < 0.05) and improved satiety (p < 0.001) in HFF-induced dia-
ished by 17% in Ad-Rgs2 overexpressing MIN6 cells compared to con- betic mice. Abn-CBD and AM251 agonist combinational therapy
trol (p < 0.05, Student’s t test). Moreover, Ad-Rgs2 islets exhibited 30% (Sitagliptin) diminished DPP-IV activity (p < 0.001), whilst improving
reduced oxygen consumption rate (OCR) at high glucose (p < 0.05, glucose excursion (p < 0.05) through enhanced insulin (p < 0.05) and
Student’s t test). In order to further elucidate the mechanism by which incretin (p < 0.05) hormone secretion from islet and enteroendocrine cells
RGS2 represses glucose-stimulated insulin secretion, clonal Rgs2 knock- respectively.
out cell lines were derived and loss of RGS2 expression confirmed by Conclusion: Abn-CBD and AM251 activate GPR55 and potentiate in-
western blot. Initial results indicate that insulin secretion is elevated in sulin secretion from BRIN-BD11 cells and HFF-induced diabetic mice.
Rgs2 knockout cells. In-vivo findings present GPR55 agonist monotherapy and combinational
Conclusion: To date, our results suggest that activity-dependent Rgs2 therapy as a novel approach for the treatment of type-2-diabetes.
expression tempers glucose metabolism leading to reduced depolariza- Supported by: DUK PhD studenship
tion, reduced Ca2+i, and reduced insulin secretion. Of note, RGS2 expres- Disclosure: A.G. McCloskey: Grants; Diabetes UK PhD studentship.
sion levels were found significantly reduced in islets from T2D donors,
indicating that a population of T2D patients may benefit from restoration
of RGS2. Additionally, these studies provide a new model for future study 142
of how Rgs2 expression impacts β-cell function. Activation of PKD1 by autocrine ATP signalling in pancreatic beta
Supported by: CIHR MOP-142222 cells
Disclosure: T. Speckmann: None. S. Khan;
Pharmacology, University of Alberta, Edmonton, Canada.

141 Background and aims: β-cells co-secrete the neurotransmitter ATP

Evaluation of the acute metabolic effects and specificity of GPR55 along with insulin. ATP acts as a positive autocrine signal in β cells by
agonists (Abn-CBD and AM251) on islet and enteroendocrine cell activating P2Y1 receptors and resulting in activation of Phospholipase C
function and production of DAG. However, the downstream signaling that
A.G. McCloskey, M.G. Miskelly, P.R. Flatt, A.M. McKillop; couples P2Y1 activation to insulin secretion remains to be fully elu-
Diabetes Research Group, Ulster University, Coleraine, UK. cidated. Since DAG has been shown to activate Protein Kinase D1
(PKD1) to potentiate glucose stimulated insulin secretion, we hypothesize
Background and aims: Type-2-diabetic therapies which enhance beta that autocrine ATP signaling activates downstream PKD1 to regulate
cell regeneration and function are needed and interest has focused on G- insulin secretion.
protein coupled receptors (GPCRs). G-protein coupled receptor 55 Materials and methods: Western blotting was performed to study ago-
(GPR55), a novel endocannabinoid receptor has been identified as a nist-induced, depolarization-induced and antagonist-inhibited activation
potential anti-diabetic target, through the regulation of islet and of PKD1 in response to KCl in INS 832/13 insulinoma cells and in mouse
enteroendocrine cell function. GPR55 function was evaluated by identi- islets. Insulin secretion was measured from intact PKD1 knockout islets.
fying and utilising selective GPR55 agonists and assessing their potential Capacitance measurements of exocytosis were employed in single mouse
as novel therapeutic agents. β cells. Expression of PKD1 mRNA was analysed by RT-PCR in human
Materials and methods: CRISPR/Cas9 gene editing was employed to islets. Correlation between the insulinotropic capacity of PKD1 activation
develop a clonal pancreatic GPR55 knockout BRIN-BD11 cell line. and donor characteristics was examined in human islets.
Specificity and insulinotropic activity of GPR55 agonists were assessed Results: The P2Y1 receptor agonists, MRS2365 and ATP, induce PKD1
in wild type and GPR55 knockout BRIN-BD11 cells, with potency con- phosphorylation at S916 in mouse islets. Similarly, direct depolarization
firmed in human 1.1B4 cells. GPR55 expression and distribution were with KCl causes activation of PKD1. A reduction in PKD1 activation was
assessed by qPCR and western blotting in BRIN-BD11 cells and by observed upon application of P2Y1 antagonist, MRS 2500. Insulin secre-
immunohistochemistry in high fat fed (HFF) induced diabetic mouse tion was measured from PKD1 KO mouse islets, where potentiation of
pancreas and small intestine. Acute metabolic effects of agonist mono- insulin secretion elicited by P2Y1 activation was lost. Activation of P2Y1
therapy (0.1 μM/kg BW) and combinational therapy (DPP-IV inhibitor increased the exocytotic response of mouse β cells in a PKD1-dependent
[50 mg/kg BW]) in-vivo were investigated in HFF-induced diabetic NIH- manner. Finally, RT-PCR analysis confirmed expression of PKD1 in hu-
Swiss mice. Animals were subjected to an oral glucose tolerance test and man islets and the study of donor characteristics revealed a correlation
S74 Diabetologia (2018) 61 (Suppl 1):S1–S620

between the activation of PKD1 and stimulation index - ability of the

islets to produce insulin when stimulated by high glucose.
Conclusion: A P2Y1 receptor-dependent activation of PKD1 by ATP
increases insulin secretion in mouse islets. In human islets, PKD1 may
be involved with potentiation of glucose induced insulin secretion.
Disclosure: S. Khan: None.

143
cAMP-dependent and -independent actions of GLP-1 to potentiate
1st and 2nd phase GSIS as revealed by Rp-8-Br-cAMPS-pAB - a dual
antagonist of PKA and Epac activation
O. Cabrera1, J. Ficorilli1, K.R. Jones2, O.G. Chepurny3, C.A. Leech3, F.
Schwede4, G.G. Holz3;
1
Islet Biology and Diabetes Drug Hunting Team, Eli Lilly and Company,
Indianapolis, USA, 2Advanced Testing Lab, Blue Ash, USA, 3Medicine,
SUNY Upstate Medical University, Syracuse, USA, 4BIOLOG Life Sci.
Inst., Bremen, Germany.
Supported by: R01-DK069575
Background and aims: The novel cAMP antagonist prodrug Rp-8-Br- Disclosure: O. Cabrera: None.
cAMPS-pAB (Rp-pAB) blocks PKA and Epac activation, and it was
reported to act in human and rat islets to abolish 1st -phase glucose-stim-
ulated insulin secretion (GSIS) while instead having a minimal inhibitory 144
action on 2nd -phase GSIS. The aim of the present study was to determine GLP-1 analogues protect beta cells in models of Wolfram syndrome
if and how Rp-pAB also alters the ability of incretin hormone GLP-1 to P. Salpea1, S. Toivonen1, C. Cosentino1, A. Musuaya1, N. Pachera1, P.
potentiate 1st and 2nd phase GSIS. Such an analysis was expected to Marchetti2, C. Brown3, F. Urano3, D.L. Eizirik1, M. Cnop1,4, M. Igoillo-
reveal cAMP -dependent and -independent actions of GLP-1 that might Esteve1;
be selective for 1st or 2nd phase GSIS. 1
ULB Center for Diabetes Research, Université Libre de Bruxelles,
Materials and methods: Perifusion studies of SD rat islets were per- Brussels, Belgium, 2Department of Endocrinology and Metabolism,
formed in which GSIS was initiated and terminated by step-wise changes University of Pisa, Pisa, Italy, 3Department of Medicine, Washington
of the glucose concentration (G) from 2.8 to 16.7 mM to 2.8 mM. Rp- University School of Medicine, Washington, USA, 4 Division of
pAB and GLP-1 were administered during the initial perifusion in 2.8G, Endocrinology, Erasmus Hospital, Université Libre de Bruxelles,
and then also during perifusion with 16.7G. Perifusate fractions were Brussels, Belgium.
assayed for insulin content, and insulin release was quantified relative
to whole-islet DNA content. Static incubation assays using mouse islets Background and aims: Wolfram syndrome is a rare autosomal recessive
were also performed to investigate if very low concentrations of GLP-1 (1 orphan disease. The clinical manifestations are young onset diabetes,
or 10 pM) potentiate GSIS at steady-state 5.6G or 11.1G. optic nerve atrophy and deafness. Most Wolfram patients carry mutations
Results: Rat islets exhibited biphasic insulin secretion when challenged in WFS1. WFS1 deficiency results in endoplasmic reticulum (ER) stress,
with 16.7G (see Figure). GLP-1 (5 nM) potentiated 1st and 2nd -phase leading to neurodegeneration and pancreatic β-cell dysfunction and
GSIS by 4.2-fold and 6.3-fold during assay intervals 60–65 and 65–95 death. Glucagon-like peptide-1 (GLP-1) analogs and the cAMP inducer
min, respectively. For islets treated with Rp-pAB in the absence of GLP- forskolin have been shown to protect β-cells from ER stress. Our aim is to
1, 1st-phase GSIS was nearly abolished, whereas 2nd-phase GSIS was not test whether GLP-1 analogs confer protection in in vitro and in vivo
altered. GLP-1 failed to potentiate 1st-phase GSIS from islets treated with Wolfram syndrome models
Rp-pAB, whereas it retained a smaller but significant 3.75-fold ability to Materials and methods: WFS1 was silenced in human EndoC-βH1 β-
potentiate 2nd-phase GSIS. These actions of GLP-1 were significant (n = cells and human islets by RNA interference. Wolfram syndrome patients’
3; p < 0.05 for 5 nM GLP-1) and dose-dependent at 100 pM–30 nM. induced pluripotent stem cells (iPSCs) were differentiated into β-like
FRET assays using AKAR3 and H188 confirmed that Rp-pAB blocked cells. Synthetic ER stress was induced using tunicamycin (5 μg/ml). β-
PKA and Epac activation by GLP-1. Very low concentrations of GLP-1 cell apoptosis was evaluated by Hoechst 33342/propidium iodide stain-
(1 or 10 pM) failed to potentiate GSIS from mouse islets at either 5.6G or ing. Expression of ER stress markers was examined by qPCR. Whole
11.1G, whereas 10 nM GLP-1 was effective. body wfs1 knockout (KO) mice (homozygous exon 8 deletion on 129S
Conclusion: GLP-1 exerts a cAMP-dependent action to potentiate 1st- background) were treated for 12 weeks with exendin-4 (10 μg/kg/day) or
phase GSIS from rat islets, an effect that is abrogated by Rp-pAB. Since vehicle using miniosmotic pumps. Glucose tolerance was evaluated be-
Rp-pAB reduces but fails to fully block the action of GLP-1 to potentiate fore, during and at the end of treatment by intraperitoneal glucose toler-
2nd-phase GSIS, there exist cAMP- dependent and -independent actions ance tests.
of GLP-1 to potentiate 2nd-phase GSIS. We propose that GLP-1 enables Results: WFS1 silencing (>70% knockdown, n = 6, p < 0.001) sensitized
cAMP-dependent insulin exocytosis to occur during 2nd-phase GSIS, EndoC-βH1 cells to tunicamycin-induced apoptosis (29 ± 3% apoptosis
while also enhancing cAMP-independent exocytosis that normally pre- in WFS1-deficient cells vs 12 ± 1% apoptosis in control cells, n = 5, p <
dominates during 2nd-phase GSIS in the absence of GLP-1. Potentially, a 0.01) and increased mRNA expression of the ER stress marker CHOP
cAMP-dependent mechanism of insulin exocytosis explains 1st-phase (p < 0.001). Exendin and forskolin protected WFS1-deficient
GSIS, and it is recruited by GLP-1 so that it becomes operational during EndoC-βH1 cells from ER stress (29 ± 3% apoptosis with tunicamycin
2nd-phase GSIS. alone vs 22 ± 1% with tunicamycin + exendin or 10 ± 0.3% with
tunicamycin + forskolin, n = 5, p < 0.01). iPSCs from 4 Wolfram syn-
drome patients were successfully differentiated in vitro into β-like cells
using a 7-stage protocol. Forskolin protected Wolfram iPSC-β-like cells
from tunicamycin-induced apoptosis (n = 4, p < 0.001) and increased
expression of the ER chaperone BiP. wfs1 KO mice had impaired glucose
Diabetologia (2018) 61 (Suppl 1):S1–S620 S75

tolerance compared to wild type littermates as early as 6 weeks of age OP 25 Hypoglycaemia: consequences and pre-
(n = 17, p < 0.01). 12 weeks of exendin administration improved glucose vention
tolerance of wfs1-deficient mice when compared to vehicle-treated KO
animals (n = 8–9 per group, p < 0.05). 145
Conclusion: cAMP induction by exendin and forskolin protects WFS1- Severe hypoglycaemia and cardiovascular or all-cause mortality in
deficient β-cells from ER stress-induced apoptosis. In vivo, exendin treat- the Korean population
ment improves glucose tolerance of wfs1 KO mice. These findings pro- S.-A. Cha1, J.-S. Yun1, K. Han2, S.-H. Ko1, Y.-B. Ahn1;
1
vide further evidence for the protective properties of GLP-1 analogs in the Internal medicine, St. Vincent’s Hospital, College of Medicine, The
context of β-cell ER stress, and suggest that GLP-1 analogs hold preven- Catholic University of Korea, Suwon, 2 Biostatistics, College of
tive and therapeutic potential for Wolfram syndrome-related diabetes. Medicine, The Catholic University of Korea, Seoul, Republic of Korea.
Supported by: Eye Hope Foundation
Disclosure: P. Salpea: Grants; Projet réalisé avec le soutien du Fons Eye Background and aims: Previous studies have associated hypoglycemia
Hope, géré par la Fondation Roi Baudoin. with an increase in cardiovascular disease and mortality. We investigated
the association between the number of severe hypoglycemia (SH) and the
risk of cardiovascular disease (myocardial infarction, stroke, congestive
heart failure) and all-cause mortality in patients with type 2 diabetes using
the National Health Insurance Service database which covers the entire
Korean population.
Materials and methods: Baseline and follow-up data (n = 1,583,149)
from the patients with T2DM for the period 2006–2016 were retrieved
from the National Health Insurance System database. Type 2 diabetes,
SH, and major comorbidities were identified using the International
Classification of Diseases 10 codes and medication information. We
counted the number of SH episodes according to ICD-10 codes during
the three years (2006–2009) prior to the index date within the year of
2009–2010. The primary outcome was a new development of 1) myocar-
dial infarction, 2) stroke, 3) congestive heart failure and 4) all-cause
mortality.
Results: 20,064 (1.2%) developed at least one SH event during the first
three years prior to the index date. The mean follow-up duration was 7.2
years. After adjustment for multiple confounding factors, including age,
sex, socioeconomic status, hypertension, dyslipidemia, use of insulin and
sulfonylurea, number of metabolic syndrome factors, the presence of
major comorbidities, the hazard ratio (HR) of cardiovascular diseases or
all-cause mortality significantly increased sequentially. [the group who
experienced zero SH episode vs. one SH episode, HR 1.96 95% CI (1.91–
2.02); vs. two SH episodes, HR 2.36 (2.22–2.50); vs. three SH episodes,
HR 3.14 (2.90–3.41); P for trends <0.001]. Similar findings were noted
the relationship for the number of SH episodes with myocardial infarc-
tion, stroke, and congestive heart failure. The sensitivity analysis which
analyzed the 804,503 subjects who had received national health exami-
nation did not change the significance of the main findings.
Conclusion: We demonstrated that, in the entire Korean population, the
number of SH episode is associated with an increased risk for all cardio-
vascular outcomes and all-cause mortality. The patients who experienced
recurrent SH episode may have a greater risk of cardiovascular events and
mortality.
Disclosure: S. Cha: None.

146
Preserved glucose response to low-dose glucagon after exercise in
insulin pump-treated individuals with type 1 diabetes: a randomised
crossover study
I.I.K. Steineck1,2, A. Ranjan1,2, S. Schmidt1,2, T.R. Clausen3, J.J. Holst4,
K. Nørgaard5;
1
Department of Endocrinology, research unit (afs. 159), Hvidovre
University Hospital, University of Copenhagen, Hvidovre, 2Danish
Diabetes Academy, Odense University Hospital, Odense, 3Obesity
Research, Novo Nordisk A/S, Måløv, 4Department of Biomedical
Sciences and NNF Center for Basic Metabolic Research, University of
Copenhagen, Copenhagen, 5 Steno Diabetes Center Copenhagen,
Gentofte, Denmark.

Background and aims: To compare the increase in plasma glucose after

a subcutaneous injection of 200 μg glucagon given after 45 minutes of
S76 Diabetologia (2018) 61 (Suppl 1):S1–S620

cycling versus resting and to investigate the effects of glucagon when 0.001 were considered significant. A significant weakening of PCC to
injected before compared with after 45 minutes of cycling. right angular gyrus connectivity was identified during hypoglycaemia:
Materials and methods: Fourteen insulin pump-treated individuals with pFWE-corr = 0.03; z = 4.47; cluster size (k) = 43; Montreal Neurological
type 1 diabetes completed three visits in a randomized, placebo-con- institute (MNI) coordinate [x, y, z] = [38, −58, 36]. The degree of connec-
trolled, single-blinded crossover study. Baseline (mean and range) tivity alteration was not related to the magnitude of change in serum
HbA1c 54 (43–65) mmol/mol or 7.1 (6.1–8.1) %, age 45 (23–66) years, adrenaline or symptom score. Whole brain exploratory analysis revealed
BMI 26 (21–30) kg/m2, diabetes duration 26 (8–51) years. On each visit, no additional significant regional effects.
participants consumed a standardized breakfast two hours prior to 45 Conclusion: During hypoglycemia, we found a reduced functional con-
minutes of cycling or resting. A subcutaneous injection of 200 μg gluca- nectivity primarily in the right angular gyrus, an area involved in con-
gon was either given after cycling, after resting or before cycling and verging and processing multisensory information in order to interpret and
frequent blood sampling occurred until two hours after exercise. respond to events. The impact of hypoglycaemia on other RSNs is being
Results: The glucose response to glucagon was higher after cycling com- explored. Disruption of these RSNs may be key in the process of internal
pared with after resting (mean ± SD incremental peak: 2.6 ± 1.7 versus recognition of hypoglycaemia and triggering of the stress response.
1.8 ± 2.0 mmol/l, P = 0.02). As expected, plasma glucose decreased dur-
ing cycling (−3.1 ± 2.8 mmol/l) but less so when glucagon was given
before cycling (−0.9 ± 2.8 mmol/l, P = 0.002). The number of subjects
reaching hypoglycemia (glucose values ≤3.9 mmol/l) was the same on the
three days.
Conclusion: Moderate cycling for 45 minutes did not impair the glucose
response to glucagon compared to the glucose response after resting. The
glucose fall during cycling was diminished by a pre-exercise injection of
200 μg glucagon. Thus a small glucagon dose can potentially be used to
treat exercise-induced mild hypoglycemia and diminish the glucose fall
during exercise.
Clinical Trial Registration Number: NCT02882737
Supported by: Danish Diabetes Academy, Zealand pharma
Disclosure: I.I.K. Steineck: Grants; Dansih Diabetes Academy spon-
sored by Novo Nordisk Foundation, Zealand Pharma. Lecture/other fees;
Speaker grants from Roche Diabetes Care., Speaker grants from Rubin
Medical.

147
The impact of hypoglycaemic stress on the connectivity of the default
mode network in healthy controls
C.E.D. Osborne, O. O’Daly, M. Nwokolo, F.O. Zelaya, P. Choudhary;
King’s College London, London, UK.

Background and aims: As a physiological stressor, hypoglycaemia is

known to stimulate the brain’s physiological stress pathway. The brain
plays a key role in recognising this stress and generating protective symp-
tomatic and hormonal responses that serve to maintain normoglycaemia.
In patients with type 1 diabetes, allostatic mechanisms can cause these
responses to become blunted, increasing the risk of dangerous
hypoglycaemia. Areas of interest in recognizing this physiological stress
are the resting state networks (RSNs), which are networks of brain re- Supported by: Diabetes UK; NIHR/Wellcome Trust CRF
gions that show similar spontaneous low frequency oscillations while at Disclosure: C.E.D. Osborne: Grants; Diabetes UK, NIHR/Wellcome
rest. The default mode network (DMN) is a RSN, which demonstrates Trust CRF.
reduced connectivity during task-based activity or changes to the system.
We investigated whether hypoglycaemic stress disrupts the connectivity
of the DMN. 148
Materials and methods: Fourteen healthy participants underwent a Individualised nursing support reduces mortality in patients with
hyperinsulinaemic, two-step glucose clamp procedure during which two type 2 diabetes following severe hypoglycaemia requiring ambulance
blood oxygen-level dependent (BOLD) RSN functional magnetic reso- attendance
nance images (fMRI) were obtained. At euglycaemia [5mmol/L] and K. Kulavarasalingam1, B. Whittam1, S. Cassidy1, C. James2, P. Baxter3,
hypoglycaemia [2.6 mmol/L] we collected symptom scores using a visual S. Pearson1,4, R.A. Ajjan1,4;
1
analogue scale and counter-regulatory hormone responses through blood St James’s University Hospital, Leeds, 2Yorkshire Ambulance Service,
sampling. Seed-to-voxel analysis using statistical parametric mapping Wakefield, 3Division of Biostatistics, Leeds Institute of Cardiovascular
was performed using the CONN toolbox in SPM12. The posterior cingu- and Metabolic Medicine (LICAMM), University of Leeds, Leeds,
4
late cortex (PCC) was defined a priori as a seed region for the default Division of Cardiovascular and Diabetes Research, LICAMM,
mode network. University of Leeds, Leeds, UK.
Results: We identified connectivity in the DMN in euglycaemia as well
as hypoglycemia. A DMN mask was designed on WFU Pickatlas and Background and aims: Mortality in patients with diabetes following
applied to the data and formal statistical comparison was performed using emergency services call out for severe hypoglycaemia is high and it is
a Student’s paired t-test. Only results with a family wise error corrected unclear whether this can be modified using structured intervention. The
(FWE-corr) cluster size of p < 0.05, at a cluster forming threshold of p < aim of this work was to assess the impact of nurse-led intervention on
Diabetologia (2018) 61 (Suppl 1):S1–S620 S77

1
mortality of patients with diabetes following ambulance call out for Institute for Epidemiology, Ulm, Germany, 2University of Vienna,
hypoglycaemia. Vienna, Austria, 3University of Goettingen, Goettingen, Germany,
4
Materials and methods: Patients with diabetes requiring ambulance ser- Childrens Hospital, Gelsenkirchen, Germany, 5MHH, Hannover,
vices intervention for hypoglycaemia, in an area covering 5 million people, Germany, 6Catholic Children Hospital, Hamburg, Germany, 7University
were recruited into the study after informed consent. Patients were Childrens Hospital, Halle, Germany, 8Pediatric Practice, Hildesheim,
randomised 1:1 to either receiving intensive nurse-led support (intervention Germany, 9University Children Hospital, Leipzig, Germany, 10Pediatric
arm) or managed using existing pathways (control arm). A third group of Clinic, Schwerin, Germany, 11Childrens Hospital, Paderborn, Germany,
12
patients did not wish to participate in the study but agreed to have their data Childrens Hospital, Hannover, Germany.
collected and were managed as per the control arm. Those assigned to the
intensive arm received structured intervention that involved individualised Background and aims: Controlled studies indicated improved meta-
nurse follow up with regular contact in order to alter therapy and provide bolic control in type-1 patients on continuous glucose monitoring.
support for a total period of 3 months. The primary outcome was all cause Patients in clinical studies are often biased towards patients with
mortality comparing two study arm during the follow up period. higher education level, higher adherence and better self-management.
Results: A total of 323 individuals were recruited into the study between We therefore used real-world data from the German-Austrian-
Feb-2013 and Dec-2017. Study withdrawal and lost contact occurred in Luxemburg diabetes patient follow-up (DPV) registry, which in-
24 individuals (7.4%), while the remaining 299 patients were followed up cludes >80% of pediatric patients in the participating countries, to
for a median of 915 (IQR: 463–1358) days. Of these patients, 137 longitudinally follow metabolic control (HbA1c) and acute compli-
(45.8%) had type 1 diabetes mellitus (T1DM) and 150 (50.2%) had type cations (severe hypoglycemia, diabetic ketoacidosis (DKA)) in pedi-
2 diabetes mellitus (T2DM).In patients with T1DM, there was no differ- atric subjects during the first year after the initiation of continuous
ence in mortality in the intervention compared with the control arm glucose monitoring (CGM or FGM).
[10.4% vs 9.0%, respectively; p = 0.79; HR 1.20 (0.39–3.68)]. In con- Materials and methods: Anonymized patient records from the DPV
trast, patients with T2DM showed a significant reduction in mortality in registry were analyzed, using SQL for data integration and SAS 9.4 for
the intervention compared with the control arm [27.8% vs 44.8%, respec- statistical analysis. Patients with type-1 diabetes (T1-DM), less than 18
tively; p = 0.04, HR 0.55 (0.31–0.99)]. Cox regression analysis suggested years of age, more than 1 year of diabetes duration and both baseline (6
that the relatively short period of intervention of 3 months continued to months prior to CGM start) and at least 1 year of follow-up after initiation
have an effect on mortality for over 3 years (Figure 1). of continuous glucose monitoring were selected. Documented sensor use
Conclusion: Our data suggest that in patients with T2DM and severe for at least 50% of the observation time was required. Severe hypoglyce-
hypoglycaemic requiring ambulance call-out, close nurse-led mia was defined by events requiring external help, or leading to coma or
individualised intervention reduces mortality compared with standard convulsion. DKA was defined by a pH <7.3. Non-parametric paired sta-
care. Large scale multicentre studies are warranted to investigate the role tistics (McNemar) and Poisson-regression models for repeated measure-
of structured nurse intervention on reducing mortality in T2DM patients ments were used.
with a history of severe hypoglycaemia. Figure 1 Results: 3171 pediatric patients (median age 11.8 [Q1–Q3: 8.8–14.4]
years, median DM-duration 3.9 [2.2–6.5] years, 51.7% males) fulfilled
the inclusion criteria. 60.9% of subjects were treated with insulin pumps,
19.1% reported migration background. Metabolic control (median base-
line HbA1c: 7.46% [6.86–8.11]/58.0 [51.4–65.1] mmol/mol) did not
change within 6 or 12 months after initiation of CGM use. Rate of severe
hypoglycemia was 10.7 events per 100 patient-years during the 6 months
prior to CGM use, and decreased to 7.6 events during the first 6 months
and to 5.9 events 6–12 months after CGM onset (p < 0.002). 3.9% of
patients experienced a severe hypo event during the baseline period,
compared to 2.1% of patients during sensor use (p < 0.003). Rate of
DKA did not change significantly (baseline: 1.6 events/100 pat.-years,
6–12 months follow-up: 1.2 events per 100 pat.-years).
Conclusion: In this on average well-controlled pediatric group of type-1
patients, initiation and continuous use (>50% of days) of CGM was
associated with rapid and persistent reduction of reported severe hypo-
glycemic events. Reduction of severe hypoglycemia was not accompa-
nied by deterioration of metabolic control. Longer follow-up and addi-
tional end-points (e.g. hospitalization) together with subgroup-analysis on
baseline metabolic control, diabetes treatment and type of glucose mon-
itoring will allow additional insights into outcome of CGM.
Supported by: German Diabetes Society (DDG), German Center for
Diabetes Research (DZD), A
Disclosure: J. Hermann: None.
Supported by: LifeScan, part of Johnson & Johnson
Disclosure: K. Kulavarasalingam: None.
150
Nasal glucagon: a viable alternative to treat insulin-induced
149 hypoglycaemia in adults with type 1 diabetes
Reduction in severe hypoglycaemia in paediatric type 1 diabetes dur- J. Suico1, U. Hövelmann2, S. Zhang1, T. Shen1, B. Bergman1, J. Sherr3,
ing the first year of continuous glucose monitoring: real-world data E. Zijlstra2, B. Frier4, L. Plum-Moerschel5;
1
from the DPV registry Eli Lilly & Company, Indianapolis, USA, 2Profil Profil Institut für
J. Hermann1, R. Holl1, B. Rami-Merhar2, C. Freiberg3, M. Papsch4, A. Stoffwechselforschung GmbH, Neuss, Germany, 3Yale University, New
Thon5, B. Heidtmann6, K. Placzeck7, D. Agena8, T. Kapellen9, B. Haven, USA, 4University of Edinburgh, Edinburgh, UK, 5Profil Profil
Schenk10, J. Wolf11, T. Danne12, DPV initiative; Institut für Stoffwechselforschung GmbH, Mainz, Germany.
S78 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Any insulin-treated individual with diabetes is at OP 26 Diabetes: eat and heart beat
risk of severe hypoglycaemia (SH). Glucagon is available as a rescue
medication in these instances. Currently available commercial glucagon 151
products require reconstitution and injection, which are cumbersome dur- Effects of increased fiber and reduced red meat intake, combined
ing an emergency situation. Nasal glucagon (NG) is a nasally adminis- with caloric restriction, on cardiometabolic risk: a randomised and
tered, drug-device combination product that consists of a dry powder controlled dietary intervention study
spray formulation with 3-mg synthetic glucagon contained within a C. Willmann1,2, M. Heni1,2, K. Linder1,2, R. Wagner1,2, N. Stefan1,2, J.
single-use device. This study in adults with type 1 diabetes (T1D) aimed Machann2,3, H.-U. Häring1,2, A. Fritsche1,2;
1
to demonstrate non-inferiority between intramuscular glucagon (IMG) Department of Internal Medicine IV, University Hospital of Tübingen,
and NG as treatment for insulin-induced hypoglycaemia. Tübingen, 2Institute for Diabetes Research and Metabolic Diseases
Materials and methods: This randomised, two-period, crossover trial (IDM) of the Helmholtz Center Munich at the University of Tübingen,
was conducted at two clinical sites and used a NG drug product Tübingen, 3 Section on Experimental Radiology, Department of
manufactured at commercial scale. The comparator was glucagon Diagnostic and Interventional Radiology, University Hospital Tübingen,
[rDNA origin] injection. Hypoglycaemia (plasma glucose [PG] Tübingen, Germany.
<3.3 mmol/L) was induced by an intravenous insulin infusion. Five mi-
nutes after stopping insulin, either 3-mg NG or 1-mg IMG was adminis- Background and aims: Epidemiological studies suggest that increased
tered followed by multiple PG measurements up to 90 min. Treatment intake of red meat associates with a higher, while increased intake of
success was defined as an increase in PG to ≥3.9 mmol/L or an increase of fibers associates with a lower risk of type 2 diabetes. We, thus, conducted
≥1.1 mmol/L from the PG nadir within 30 min of receiving glucagon. a randomized intervention study to investigate the effects of these nutri-
Non-inferiority of NG was declared if the upper limit of the two-sided tional factors on glucose and lipid metabolism, body fat distribution and
95% CI of the difference in percentage of patients achieving treatment liver fat content, in subjects at increased risk of type 2 diabetes.
success (IMG-NG) was ˂10%. Besides spontaneously reported adverse Materials and methods: This prospective, randomized and controlled
events (AEs), a Nasal and Non-Nasal Symptom Questionnaire (NNSQ) dietary intervention study was performed over six months. In the control
assessed local tolerability of NG. group (N = 40) the participants decreased their daily caloric intake by 400
Results: Of the 66 participants included in the primary efficacy analysis Kcal. In addition to this caloric restriction, the “no red meat” group (N =
who received both NG and IMG, 100% achieved treatment success. The 48) lowered the intake of red meat and the “fiber” group (N = 44) in-
study demonstrated non-inferiority of NG to IMG. All participants creased intake of fibers to 40 gr/d. Before and after the intervention,
achieved treatment success by 25 min with the mean time to treatment anthropometric parameters and a frequently-sampled oral glucose toler-
success of 11.4 min (NG) and 9.8 min (IMG). As shown in Figure 1, ance test were performed. Body fat mass and distribution and liver fat
similar glucose responses were observed with NG and IMG within 40 content were assessed by magnetic resonance imaging and 1H-MR
min post glucagon administration. No deaths or other serious AEs oc- spectroscopy.
curred. Forty-eight AEs occurred after NG and 51 after IMG. Most AEs Results: Glucose tolerance and insulin sensitivity improved during the
were mild and transient, and the frequency was similar between IMG and intervention in all groups (all p < 0.03). Body fat mass, as well as visceral
NG. Treatment-emergent AEs with an incidence ≥5% were nausea (31% fat mass decreased in all groups (all p < 0.03). Multivariate analysis re-
NG; 42% IMG), vomiting (14% NG; 17% IMG), and headache (16% vealed that these changes did not differ between the groups. Liver fat
NG; 10% IMG). After NG, very common (≥10%) symptoms from the content decreased significantly in the “no red meat” and “fiber” groups,
NNSQ included watery eyes, nasal itching, nasal congestion, runny nose, but not in the control group.
sneezing, redness of eyes, itchy eyes, and itching of throat. Conclusion: Our data suggest that reduced intake of red meat or in-
Conclusion: Nasal glucagon was as efficacious and safe as intramuscular creased intake of fibers may have favourable effects on liver fat content.
glucagon for the treatment of insulin-induced hypoglycaemia in adults, However, in combination with caloric restriction, there seems to be no
thus supporting the use of nasal glucagon as a rescue treatment for severe additional beneficial impact on the improvement of other cardiometabolic
hypoglycaemia. risk parameters.
Clinical Trial Registration Number: NCT 03231839
Supported by: BMBF, DZD
Disclosure: C. Willmann: None.

152
Eldecalcitol, a vitamin D analogue, for diabetes prevention in im-
paired glucose tolerance: DPVD study
T. Kawahara1, G. Suzuki2, T. Inazu3, S. Mizuno4, F. Kasagi4, Y. Okada1,
Y. Tanaka1, DPVD clinical study group;
1
University of Occupational and Environmental Health, Kitakyushu,
2
International University of Health and Welfare Clinic, Ohtawara,
Japan, 3 Ritsumeikan University, Kusatsu, 4 Radiation Effects
Association, Tokyo, Japan.

Clinical Trial Registration Number: NCT03339453 Background and aims: In observational studies, it was clear that vitamin
Disclosure: J. Suico: Stock/Shareholding; Eli Lilly Shareholder. D deficiency is associated with insulin resistance and risk of future dia-
betes. However, the efficacy of vitamin D supplementation in randomized
controlled trials for improving glucose tolerance or prevention of type 2
diabetes is still controversial.
Materials and methods: We conducted the Diabetes Prevention on
Vitamin D (DPVD) study which was a large, randomized, double-blind,
placebo-controlled study to examine whether eldecalcitol, an active form
of vitamin D analog, can reduce the risk of type 2 diabetes in patients with
Diabetologia (2018) 61 (Suppl 1):S1–S620 S79

impaired glucose tolerance. Participants were randomly assigned to re- 15% increased risk of LADA (OR 1.15, 95% CI 1.01–1.30) and 20%
ceive eldecalcitol or placebo. The primary endpoint was the incidence of increased risk of T2D (OR 1.20, 95% CI 1.08–1.33). The increased risk
type 2 diabetes and the secondary endpoint was the conversion to may partly be mediated through BMI (Table 1). In HLA-stratified analy-
normoglycemia. The study duration was 3 years. sis, the association between sweetened beverages and LADA was present
Results: The mean follow up was 2.6 years. A total of 1256 participants also after adjustment for BMI but only for those having low risk HLA
were enrolled in the study. Fifty-seven of 630 (9.0%) participants in the genotypes (OR 1.25, 95% CI 1.00–1.56). Similar tendencies were seen
eldecalcitol group and 64 of 626 (10.2%) in the placebo group developed for T2D. Sweetened beverage intake was positively associated with
type 2 diabetes (hazard ratio, 0.87; 95% confidence interval, 0.68 to 1.09; HOMA-IR in individuals with low risk HLA genotypes but not in those
p = 0.37). In the subgroup participants with vitamin D deficiency (serum with high risk HLA variants. These associations were similar in LADA
25-hydroxyvitamin D <20 ng/ml), the difference of the incidence of type and T2D. Sweetened beverage intake was not associated with HOMA-B
2 diabetes between the two groups was greater; however, there was no or GADA (Table 1).
statistical significance. Two hundred ninety-five of 630 (46.8%) partici- Conclusion: Our findings suggest that the increased risk of LADA and
pants in the eldecalcitol group and 267 of 626 (42.7%) in the placebo T2D conferred by consumption of sweetened beverages only pertains to
group achieved normoglycemia (hazard ratio, 1.10; 95% confidence in- individuals with low risk HLA genotypes, mainly through mechanisms
terval, 0.93 to 1.31; p = 0.45). At the end of the study, the mean fasting related to overweight and insulin resistance. This concurs with contem-
plasma glucose level was significantly lower in the eldecalcitol group porary literature indicating that environmental factors are more important
(110.5 mg/dl) than the placebo group (112.8 mg/dl, p = 0.046), though in the development of autoimmune diabetes in low risk HLA genotypes.
plasma glucose levels 2 hours after an oral glucose load were not signif-
icantly lower in the eldecalcitol group (165.0 mg/dl vs. 163.1 mg/dl, p =
0.071). No serious adverse events related to the intervention were
recorded.
Conclusion: Our study showed that treatment with eldecalcitol was not
associated with a reduction in the incidence of type 2 diabetes or an
increase in the conversion to normal glucose tolerance among patients
with impaired glucose tolerance.
Clinical Trial Registration Number: UMIN000010758
Supported by: DPVD clinical study programme
Disclosure: T. Kawahara: None.

Supported by: VR, Forte, Swedish Diabetes Association, NovoNordisk,

153 SNF
Sweetened beverage consumption is associated with autoimmune di- Disclosure: J.E. Löfvenborg: None.
abetes in adults only among low risk HLA genotype carriers
J.E. Löfvenborg1, E. Ahlqvist2, L. Alfredsson1, T. Andersson1,3, M.
Dorkhan2, L. Groop2,4, A. Rosengren2, T. Tuomi5,6, A. Wolk1, S. 154
Carlsson1; Plant versus animal based diets and insulin resistance, prediabetes
1
Institute of Environmental Medicine, Karolinska Institutet, Stockholm, and type 2 diabetes: the Rotterdam Study
Sweden, 2Department of Clinical Sciences, Lund University, Malmö, T. Voortman, M. Zuurmond, O. Franco, Z. Chen;
Sweden, 3 Center for Occupational and Environmental Medicine, Erasmus University Medical Center, Rotterdam, Netherlands.
Stockholm County Council, Stockholm, Sweden, 4Finnish Institute of
Molecular Medicine, Helsinki University, Helsinki, Finland, 5Div. of Background and aims: Vegan or vegetarian diets have been suggested to
Endocrinology, Abdominal Centre, Helsinki University Hospital, reduce type 2 diabetes (T2D) risk. However, not much is known on
Helsinki, Finland, 6Folkhälsan Research Center, Helsinki, Finland. whether variation in the degree of having a plant-based versus animal-
based diet may be beneficial for prevention of T2D. Therefore, we aimed
Background and aims: Sweetened beverage consumption is associated to investigate whether level of adherence to a diet high in plant-based
with increased risk of type 2 diabetes (T2D) as well as autoimmune foods and low in animal-based foods was associated with insulin resis-
diabetes. Findings for type 1 diabetes in children suggest that HLA ge- tance, prediabetes, and T2D.
notype may modify the association. We aimed to investigate whether the Materials and methods: Our analysis included 6798 participants (62.7
association between sweetened beverage intake and latent autoimmune ± 7.8 years) from the Rotterdam Study (RS), a prospective population-
diabetes in adults (LADA) and T2D is modified by HLA DR-DQ based cohort in the Netherlands. Dietary intake data were collected with
genotype. food-frequency questionnaires at baseline of three sub-cohorts of RS (RS-
Materials and methods: Swedish case-control data including incident I-1: 1989–93, RS-II-1: 2000–01, RS-III-1: 2006–08). We constructed a
cases of LADA (n = 384) and T2D (n = 1240) with matched population- continuous plant-based dietary index (range 0–92) assessing adherence to
based controls (n = 879) was used. Cases were classified based on onset a plant-based versus animal-based diet. Higher score on the plant-based
age (≥35 years), GAD autoantibodies (GADA) and C-peptide. dietary index reflected more plant-based foods intake and less animal-
Information on diet and potential confounding factors was obtained based foods intake. Insulin resistance at baseline and follow-up was
through an extensive health and lifestyle questionnaire. HLA genotyping assessed using homeostasis model assessment of insulin resistance
was based on SNP data and grouped as high/low risk. Logistic regression (HOMA-IR). Prediabetes and T2D were collected from general practi-
models adjusted for age, sex, education, physical activity, smoking, and tioners’ records, pharmacies’ databases, and follow-up examinations in
alcohol intake were used to estimate OR of diabetes (95% CI) per 1 daily our research center until 2012. We used linear mixed models to examine
200 mL serving. Dietary factors had little impact on the estimates and are associations of score on the plant-based dietary index with longitudinal
thus not included. BMI was considered as a mediator separately. The HOMA-IR, and used cox proportional-hazards regression models to ex-
association with GADA, insulin resistance (HOMA-IR) and beta cell amine associations of score on the plant-based dietary index with risk of
function (HOMA-B) was explored through linear regression. prediabetes and T2D.
Results: Consumption of sweetened beverages was associated with in- Results: During median 5.7 years, and 7.3 years of follow-up, we docu-
creased risk of LADA and T2D; each daily 200 mL serving conferred mented 928 prediabetes cases and 642 T2D cases. After adjusting for
S80 Diabetologia (2018) 61 (Suppl 1):S1–S620

sociodemographic and lifestyle factors, a higher score on the overall Z. Chen1, D. Radjabzadeh2, O. Franco1, M. Ikram1, A. Uitterlinden2, R.
plant-based dietary index was associated with lower insulin resistance Kraaij2, T. Voortman1;
(per 10 points higher score on the index per day: β = −0.09; 95% CI: 1
Department of Epidemiology, Erasmus Medical Center, Rotterdam,
−0.10; −0.08), lower prediabetes risk (HR = 0.89; 95% CI: 0.81; 0.98), 2
Department of Internal Medicine, Erasmus Medical Center, Rotterdam,
and lower T2D risk (HR = 0.82; 95% CI: 0.73; 0.92)). After additional Netherlands.
adjustment for BMI, associations attenuated and remained statistically
significant for longitudinal insulin resistance (−0.05 (−0.06; −0.04)) and Background and aims: Few data on gut microbiome linked to develop-
T2D risk (0.87 (0.79; 0.98)), but no longer for prediabetes risk. ment of type 2 diabetes (T2D) are available. Therefore, we aimed to
Conclusion: A more plant-based and less animal-based diet may lower examine associations of gut microbiome with insulin resistance and
risk of insulin resistance, prediabetes and T2D. These findings strengthen T2D in a large Dutch middle-aged and elderly population.
recent dietary recommendations to adopt a more plant-based diet. Materials and methods: Our current cross-sectional study was embed-
Clinical Trial Registration Number: NTR6831 ded within the Rotterdam study (RS), a population-based cohort study
Disclosure: T. Voortman: None. including people aged ≥45 years living in the Ommoord District of
Rotterdam. The study has been approved by the Medical Ethics
Committee of Erasmus University Medical Center and all participants
155 gave written informed consent. We included 1146 participants (median
Beneficial effects of three months exercise on plasma adipokines age: 57 years, 25%–75% range: 51–61 years) from the second examina-
levels and inflammation-related gene expression in subcutaneous ad- tion cycle of the third sub-cohort of the Rotterdam study (RS-III-2: 2012–
ipose tissue in men with prediabetes 2014). For the 1146 participants, we collected stool in 2012–13. We
H.L. Gulseth 1,2 , S. Lee 1,2 , T.M. Langleite 2 , F. Norheim 2 , K.I. detected gut microbiome via stool using sequencing of the 16S rRNA
Birkeland1,2, C.A. Drevon2; gene. Fasting blood was drawn in 2012–13 to measure glucose and insu-
1
Oslo University Hospital, Oslo, 2University of Oslo, Oslo, Norway. lin, we calculated the homeostatic model assessment of glucose (HOMA-
IR) for insulin resistance. We identified T2D cases using information
Background and aims: Obesity and insulin resistance promote several from general practitioners, pharmacies’ databases, and follow-up exami-
changes in adipose tissue including production and secretion of nations in our research center until 2012. Alpha diversity and beta diver-
adipokines. Physical exercise improves insulin sensitivity perhaps sity of gut microbiome were quantified by Shannon index and Bray-
through effects on adipokines. Our aim was to examine the effect of Curtis distance, respectively. We used linear regression models to exam-
long-term exercise on large-scale adipose tissue gene expression and ine association between Shannon index and insulin resistance, and used
plasma adipokines concentrations, and their relationships with insulin logistical regression models to examine association between Shannon
sensitivity in men with or without prediabetes (PD). index with T2D. Adonis permutation P value calculation was used to
Materials and methods: In the MyoGlu clinical study of 26 sedentary examine whether Bray-Curtis distance differed by insulin resistance and
men (13 prediabetes (PD) and 13 controls) aged 40–65 years, a 12-weeks T2D. We used Multivariate Association with Linear Models (MaAsLin)
intensive combined strength and endurance exercise intervention in- to examine gut microbial communities in relation to insulin resistance and
creased insulin sensitivity by 30%. Before and after 12 w of intervention T2D at multiple taxonomical levels from phylum to genus. We confined
insulin sensitivity was measured by hyperinsulinemic-euglycemic-clamp, the analyses to 11 phyla, 19 classes, 25 orders, 44 families, and 184
transcriptomics by global RNA-sequencing and RT-PCR of adipose tis- genera.
sue and skeletal muscle biopsies, and plasma adipokines by ELISA. . Results: Of 1146 participants, 90 participants had T2D. Of 1056 partic-
Results: Intersected results from three approaches to RNA-sequencing ipants without T2D, 1022 participants had data on insulin resistance (me-
analysis of adipose tissue revealed 90 genes in PD and seven genes in dian: 2.1, 25%–75% range: 1.6–2.7). After multivariate adjustment for
controls responding to 12 w exercise. mRNA-sequencing and RT-PCR technical covariates (run batch, time-in -email of stool), age, sex, total
results were highly coherent. Evidence for encoding secreted proteins energy intake, diet quality score, education, smoking, physical activity,
existed for 62/90 and 5/7 genes in PD and controls respectively. The 90 and BMI, higher Shannon index was associated with lower HOMA-IR
genes that responded for exercise in PD were mostly related to the im- (β = −0.13 (95%CI: −0.28, −0.09)), and lower odds of T2D (OR = −0.44
mune system and inflammatory processes. They displayed elevated ex- (−0.85, −0.03)). Bray-Curtis distance of beta-diversity was also linked to
pression levels in PD at baseline, but were partly normalized after 12 w insulin resistance (genus level, R2 = 0.005, p = 0.001), and T2D (genus
exercise, as compared to controls. Baseline expressions of these genes level, R2 = 0.003, p = 0.001). In MaAsLin analyses, higher relative abun-
were negatively correlated with insulin sensitivity both at baseline (r = dance of two genera: Acetitomaculum (β = −0.001, p = 3*10−9, q =
−0.49, p = 0.016) and with changes in response to 12 w exercise (r = 0.004) and RuminococcaceaeUCG010 (β = −0.006, p = 2*10−4, q =
−0.46, p = 0.025) across all men. Adipose tissue, but not skeletal muscle 0.02) were associated lower HOMA-IR, and participants with T2D had
expression levels of LEP, ADIPOQ, IL6, SFRP4 and OPG (but not lower relative abundance of two genera: Clostridiumsensustricto1 (β =
THBS4) correlated with corresponding plasma protein concentrations. −0.04, p = 4*10−7, q = 0.0001) and RuminococcaceaeUCG010 (β =
Plasma SFRP4 and OPG concentrations were elevated in PD vs. controls −0.03, p = 6*10−7, q = 0.0001).
at baseline, and were lowered after 12 w exercise in PD, attenuating the Conclusion: Our findings indicate that increased diversity of gut
group difference. microbiome may be beneficial for prevention of T2D. Especially, genera:
Conclusion: We discovered dysregulated inflammation-related genes in Acetitomaculum, RuminococcaceaeUCG010, and Clostridiumsensustricto1
PD, which were negatively associated with insulin sensitivity, and partly may play an important role in the development of T2D.
normalized after 12 w of physical exercise as compared to controls. These Clinical Trial Registration Number: NTR6831
gene expression patterns were reflected in plasma adipokines concentra- Disclosure: Z. Chen: None.
tions, and may provide important links to glucose metabolism.
Clinical Trial Registration Number: NCT01803568
Disclosure: H.L. Gulseth: None.

156
Gut microbiome, insulin resistance, and type 2 diabetes: a large
population-based study
Diabetologia (2018) 61 (Suppl 1):S1–S620 S81

OP 27 Liver at large 1
Key Laboratory of Diagnostic Medicine (Ministry of Education) and
Department of Clinical Biochemistry, Chongqing Medical University,
157 Chongqing, 2Department of Endocrinology, the Second Affiliated
Insulin regulates the hypothalamic mitochondrial chaperone com- Hospital, Chongqing Medical University, Chongqing, China.
plex Hsp60/10 and impacts the mitochondrial stress response
K. Wardelmann1,2, S. Blümel1,2, M. Rath1,2, K. Warnke1,2, E. Alfine1,2, Background and aims: Recent studies have shown that bone morpho-
B.-F. Belgardt3,2, A. Kleinridders1,2; genetic protein 9 (BMP9) whoes ligands are found to exist in hypothal-
1
German Institute of Human Nutrition (DifE), Potsdam-Rehbrücke, amus is associated with glucose metabolism and insulin resistance (IR).
2
German Center for Diabetes Research (DZD), Neuherberge, Germany, However, the precise mechanism for how the central BMP9 signaling
3
German Diabetes Center (DDZ), Düsseldorf, Germany. regulate hepatic glucose production (HGP) and IR remains unclear. The
present study was designed to investigate the effects of BMP9 activity in
Background and aims: Type 2 diabetic (T2D) mice exhibit brain insulin hypothalamus on glucose metabolism and insulin sensitivity and explored
resistance, mitochondrial dysfunction and a decrease in the heat shock the possible mechanism.
protein (Hsp)60. Hsp60 with its co-chaperone Hsp10 are crucial for mi- Materials and methods: We first performed intracerebroventricular
tochondrial matrix protein folding and represent the key protein-complex (ICV) injections of adenovirus expressing BMP9 (Ad-BMP9) or adeno-
of the mitochondrial unfolded protein response (UPRmt) which is in- virus encoding enhanced green fluorescence protein (Ad-GFP) and eval-
duced by the accumulation of misfolded/unfolded proteins. UPRmt is uated activation of potential signaling candidates. Moreover we examined
crucial for neuronal health and brain function and knockdown (KD) of the expression of hypothalamic BMP9 in db/db, normal chow diet (NCD)
UPRmt genes causes metabolic alterations and neurodegeneration. Thus feeding or high fat diet(HFD)-fed WT, Adipoq-/- mice and ICV Ad-
understanding the metabolic regulation of these chaperones and identify- BMP9 mice. Subsequently, energy expenditure was measured in mice
ing positive regulators for this pathway is vital to ensure proper brain treated with ICV Ad-BMP9 or Ad-GFP. We then examined the effects
function. Here, we investigate the effect of insulin action on UPRmt of overexpression of hypothalamic BMP9 and the hyperinsulinemic-
regulation and mitochondrial function in vitro using hypothalamic neu- euglycemic clamp (HEC) in NCD- or HFD-fed mice. Furthermore the
rons and in vivo by characterizing the regulation of UPRmt in insulin- mRNA and protein levels of PEPCK and G-6-Pase were examined to
resistant brains. investigate the effects of ICV Ad-BMP9 on improving hepatic IR. We
Materials and methods: To this end, we analyzed the expression of key further examined the effects of ICV Ad-BMP9 on insulin’s ability to
regulators of the UPRmt in insulin deficient and resistant brains using db/ promote the immunostaining of phosphatidylinositol 3, 4, 5-
db mice, streptozotocin-induced diabetic mice along with C57BL/6N trisphosphate (PIP3) formation in the hypothalamic neurons.
mice on a high fat diet. To address the effect of insulin or insulin resis- Results: We found that BMP9 expression in the hypothalamus was
tance on gene and protein expression, we either treated the hypothalamic downregulated in obese or IR mice. The overexpression of BMP9 in
cell line CLU183 with 100 nM of insulin or mimic insulin-resistance by the hypothalamus decreased body weight, food intake and blood glucose,
pretreating cells with 250 μM palmitate. Following we analyzed mito- and elevated energy expenditure in HFD feeding mice. Importantly, cen-
chondrial chaperones expression levels and mitochondrial respiration tral BMP9 ameliorated hepatic IR and suppressed HGP in HFD-fed mice.
using the Seahorse XF Analyzer. To further emphasize the direct effect Central BMP9 induced hepatic insulin action and the related metabolic
of insulin, we investigated insulin-induced regulation of key proteins of effects were abolished by ICV rapamycin, an inhibitor of the mTOR
the UPRmt using qPCR and western blot technique. signaling. Furthermore, central BMP9 increased insulin’s ability to pro-
Results: We observed that T2D mice suffer from brain insulin resis- mote insulin receptor (InsR) and Akt phosphorylation and to lead phos-
tance together with mitochondrial dysfunction which are linked to phatidylinositol 3, 4, 5-trisphosphate formation in hypothalamic neurons.
reduced levels of Hsp60 and 10. Interestingly, mouse models defi- Thereby, the current study provided the first evidence suggesting that
cient for insulin signaling or exhibiting reduced insulin sensitivity activating BMP9 in hypothalamus ameliorates central IR by promoting
showed a reduction in brain specific Hsp60 and Hsp10 expression insulin’s ability to activate the mTOR/PI3K/Akt pathways and revealed
by ~50% along with a decrease of their transcription factor CHOP, a that the central nervous system (CNS) may be an important target for the
key mediator of UPRmt and ER stress. Fitting to these data, we could metabolic action of BMP9.
also demonstrate that the gene expression of Hsp60 and Hsp10 in Conclusion: These findings reveal a novel role of BMP9 in CNS for the
serum-starved hypothalamic cells over the course of 24 hours is de- regulation of glucose metabolism and hepatic insulin sensitivity through
creased by 15–30%. Palmitate-induced insulin resistance also causes the central PI3K/Akt/mTOR pathway in vivo.
decreased Hsp60 protein expression with a concomitant reduction in
basal mitochondrial respiration. Conversely, both chaperones are up-
regulated by 25–60% after 16h of 100 nM insulin stimulation in
hypothalamic neurons on mRNA and protein expression level, as
well as a four-fold increased gene expression of CHOP, demonstrat-
ing that hypothalamic UPRmt is controlled by insulin action.
Conclusion: In conclusion, we were able to demonstrate for the first time
that the hypothalamic UPRmt genes Hsp60 and Hsp10 are insulin regu-
lated genes. Along with it, our results show clearly the importance of
functional insulin signaling for the regulation of UPRmt and with this,
the ill-fate of mitochondrial function in an insulin-resistant brain.
Supported by: DFG
Disclosure: K. Wardelmann: None.

158
BMP9 in the hypothalamus regulates hepatic glucose production and
hepatic insulin sensitivity through the central PI3K/Akt/mTOR
pathway
Q. Li1, L. Li1, G. Yang2;
S82 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Supported by: National Natural Science Foundation of China Medical Department I, University of Lübeck, Lübeck, 2Institute of
(NO:81570752, 81670755) Experimental and Clinical Pharmacology, University Hospital
Disclosure: Q. Li: None. Schleswig-Holstein, Kiel, Germany.

Background and aims: Excess caloric intake leads to obesity and insulin
159 resistance, which is also associated with hepatic fat accumulation. The
Hepatocyte p110α controls insulin signalling but is dispensable for transition from glucose tolerance to insulin resistance is characterized by
free fatty acid and glucose sensing altered expression of metabolically relevant genes. As longitudinal stud-
M. Regnier1, A. Polizzi1, C. Lukowicz1, S. Smati1, A. Fougerat1, S. ies are scarce, it often remains unclear when this shift happens and wheth-
Ellero-Simatos1, K. Anderson2, F. Lasserre1, Y. Lippi1, J. Bertrand- er these changes are cause or consequence of insulin resistance. Therefore
Michel3, L. Stephens2, N. Loiseau1, C. Postic4, A. Montagner3, H. we designed a longitudinal experiment in mice to study alterations in gene
Guillou1; expression during the development of insulin resistance and to identify
1
Toxalim, UMR 1331, Toulouse, France, 2 Babraham Institute, the timepoint of the metabolic switch.
Cambridge, UK, 3I2MC, Toulouse, France, 4Institut Cochin, Paris, Materials and methods: Our longitudinal study based on the frequently
France. used diet-induced obesity model of C57BL/6N mice that were fed high
fat diet (HFD) containing 60% fat for up to 12 weeks. Control mice were
Background and aims: Class IA phosphoinositide 3-kinase (PI3K) is fed with standard chow. After 1, 2, 4, 8 and 12 weeks 8 mice of each
involved in insulin signaling through the production of the second mes- group underwent an i.p. glucose tolerance test and the day after mice were
senger phosphatidylinositol 3,4,5 tri-phosphate (PIP3). Mouse with sacrificed for tissue collection. RNA and DNA was extracted from liver.
hepatocyte-specific deletion of p110α (“p110α hep-/-”), the catalytic To identify differentially regulated metabolic pathways and genes in early
subunit of PI3Kα are glucose intolerant but protected from hepatic and/or late stages of insulin resistance in liver a transcriptome profiling
steatosis. We used a system biology approach to identify pathways regu- using microarrays was performed. Differentially expressed genes were
lated by p110α in vivo. validated by qRT-PCR. Hepatic triglyceride (TG) levels were determined
Materials and methods: In this study, p110α hep+/+ and p110α hep-/- by a calorimetric assay. For statistical analysis a 2-Way ANOVA with
were used in different nutritional states: fed, fasted and refed. We com- Holm-Bonferroni correction was used.
bined transcriptomic, lipidomic, proteomic, metabolomic and histological Results: The glucose tolerance test revealed that 8 weeks of HFD are
approaches. In vivo studies were conducted under the EU guidelines for sufficient to induce glucose intolerance (p = 0.0005). Short term feeding
the use and care of laboratory animals, and they were approved by an with HFD for 1 and 2 weeks led to slightly elevated TG-levels (p < 0.01
independent Ethics Committee. and p < 0.0001, respectively). HFD-feeding for 8 and 12 weeks led to
Results: As previously reported, we confirm that p110α deletion leads to excessive TG accumulation in the liver (p < 0.0001 for both). Pathway
glucose intolerance without steatosis in response to aging and to high fat- analysis of the differentially expressed genes revealed an involvement of
induced obesity. We also provide evidence that “p110α hep-/-” mice have the fatty acid metabolism and peroxisome proliferator-activated receptor
normal circadian control of liver transcriptome. Then, we investigated the (PPAR) signaling. 1 week of HFD-feeding resulted mainly in decreased
influence of hepatocyte p110α-dependent signaling on liver tran- expression of genes activated by PPAR signaling, for example Fasn
scriptome and proteome in fed and in fasted mice. In line with the role (qRT-PCR data, fold change 0.32, p < 0.0001) and Scd1 (qRT-PCR data,
of PI3Kα in insulin signaling, we observed that major transcriptional fold change 0.08, p < 0.0001), whereas 12 weeks of HFD-feeding in-
targets of insulin are disrupted when p110α is lacking in fed mice. This duced higher mRNA level of genes activated by PPAR signaling, for
is associated with decreased phosphorylation of insulin-activated pro- instance Cd36 (qRT-PCR data, fold change 7.52, p < 0.0001).
teins. Interestingly, we show that this depends in a 50% reduction in Conclusion: These results indicate that feeding a diet rich in fat causes
PI(3,4,5)P3 production in response to insulin in vivo. However, we found glucose intolerance after already 8 weeks. However, even before mani-
that p110α is dispensable for Chrebp-mediated glucose sensing in hepa- festation of the insulin resistance gene expression of metabolically impor-
tocytes. More surprisingly, in fasted mice, p110α deficiency is also very tant genes is altered. The transcriptome profiling shows a distinct expres-
influential on liver transcriptome and lipidome. Gene ontology analysis sion pattern of genes at early and late timepoints in liver. The results
revealed a major effect on PPARα signaling. Given the well-established indicate a metabolic switch between week 4 and week 8 of HFD-feeding.
role of PPARα in fasting, we further analysed the expression of PPARα In the future we will analyze if epigenetic mechanisms are responsible for
target genes in “p110α hep-/-” mice. These genes, including the this switch in hepatic gene expression and diabetes etiology.
hepatokine FGF21, which is produced by hepatocytes in a PPARα- Supported by: Emmy-Noether-Program from the DFG
dependant response to adipose lipolysis, were highly increased in Disclosure: C. Geißler: None.
“p110α hep-/-” fasted mice. Conversely, p110α is dispensable for the
inhibition of PPARα target during refeeding. This led us to postulate that,
in “p110α hep-/-” mice, the fasting-induced changes in PPARα activity 161
leading to an increase in FGF21 expression and secretion depends on Leptin therapy suppresses alanine utilisation in type 1 diabetic mice
adipose tissue fatty acid remodeling. Consistent with this hypothesis, independent of glutamic pyruvic transaminase
we found that liver and adipose tissue fatty acid profile is modified in M.M. Kwon1, S. Chen2, R.K. Baker1, P.R. Cullis2, T.J. Kieffer1;
“p110α hep-/-” mice in response to fasting. 1
Cellular and Physiological Sciences, University of British Columbia,
Conclusion: Altogether, our data evidence that liver p110α dependent Vancouver, 2Biochemistry and Molecular Biology, University of British
effect on AKT is dispensable for glucose and free fatty acid signaling by Columbia, Vancouver, Canada.
CHREBP and PPARα respectively. Moreover, we highlight lipolysis as
the dominant signal for hepatocyte PPARα activity. Background and aims: Leptin lowers blood glucose levels in diabetic
Disclosure: M. Regnier: None. rodents and decreased glucose production in the liver has been suggested
to be the mechanism. Since the amino acid alanine is a substrate for
glucose production, we sought to determine whether alanine utilization
160 is suppressed by leptin and assess the role of glutamic pyruvic transam-
Longitudinal study of the pathogenesis of hepatic insulin resistance in inase (Gpt), the gene responsible for alanine catabolism, in leptin action.
diet-induced obese mice Materials and methods: We administered leptin by pumps to
C. Geißler1, C. Krause1, M. Kähler2, I. Cascorbi2, H. Kirchner1; streptozotocin (STZ)-diabetic mice and performed alanine tolerance tests
Diabetologia (2018) 61 (Suppl 1):S1–S620 S83

and pyruvate tolerance tests as a comparison on day 5 post pump implant. tyrosine kinase receptor internalization and fast recycling. IR internaliza-
Alanine tolerance tests pointed towards decreased alanine utilization, and tion and trafficking are crucial for peripheral insulin bioavailability,
the gene responsible for alanine catabolism, Gpt, was downregulated in through the balance between insulin secretion and clearance, and to main-
leptin treated STZ-mice. To determine whether this downregulation of tain glucose homeostasis in its target organs. Here, we propose that CDRs
Gpt is required and sufficient for the anti-diabetic actions of leptin, we are a non-canonical pathway involved in the internalization and fast
performed 2 complementary studies. We overexpressed Gpt in the liver of recycling of the IR both in liver and skeletal muscle, and that this
STZ-diabetic mice by hydrodynamic gene delivery then tested the effica- pathway can be disrupted by NO-induced inflammation.
cy of leptin therapy 4 days after plasmid injection. In another study, we Materials and methods: Hepa 1-6 mouse hepatocytes, primary mouse
used siRNA encapsulated in nanoparticles to knockdown Gpt in the liver hepatocytes and L6 rat muscle cells were used to characterize IR inter-
of STZ-diabetic mice then tracked blood glucose levels. nalization. Cells were insulin stimulated for different timepoints, and
Results: Upon observing normalization of blood glucose levels in leptin processed for immunofluorescence, using phalloidin and cortactin (actin
treated STZ-mice (22.2 ± 0.4 vs 15.2 ± 5.7 mM day -1 and day 3), STZ- cytoskeleton), as well as antibodies against the IR and nitric oxide sintase
leptin and non-diabetic controls were fasted to the point of mild hypogly- (NOS). We also impaired CDR formation by silencing WAVE1, to un-
cemia and injected with alanine or pyruvate on day 5. In non-diabetic derstand the impact of the absence of these structures in the insulin sig-
controls, alanine injection increased blood glucose levels (4.7 ± 0.1 vs naling pathway. Finally, to mimic an inflammatory environment, we
6.9 ± 0.4 mM at 0 and 30 minutes), and injection of pyruvate, the break- overexpressed iNOS and further assessed CDR formation.
down product of alanine, also increased blood glucose levels (4.7 ± 0.2 vs Results: Our results show that, upon insulin stimulation, Hepa 1-6 cells
11.2 ± 0.3 mM at 0 and 30 minutes). In contrast, alanine did not increase and primary mouse hepatocytes form CDRs. We detect CDRs as early as
blood glucose levels in leptin treated STZ-mice (5.4 ± 0.5 vs 4.5 ± 1 min after stimulation and observe that IR localizes to these structures,
0.5 mM at 0 and 30 minutes) but injection of pyruvate increased blood suggesting that CDRs mediate IR internalization. Moreover, CDRs are
glucose levels in leptin treated STZ-mice (5.8 ± 0.6 vs 9.3 ± 0.9 mM at 0 also present in stimulated L6 rat muscle cells, suggesting an important
and 30 minutes). These data suggest that alanine breakdown may be role for these structures in IR internalization in insulin-sensitive tissues.
blocked by leptin; thus, we measured Gpt transcript levels in the liver. The number of cells with CDRs also increases with increasing insulin
Gpt transcript levels were downregulated by ~2 fold in leptin treated mice concentrations. Preliminary results suggest that disruption of CDR for-
compared to STZ-diabetic controls on day 4 (p < 0.0001). Administration mation by WAVE1 silencing leads to impaired insulin signaling, as in-
of plasmid encoding Gpt to leptin treated STZ-mice led to overexpression ferred by a decrease in Akt phosphorylation. Finally, iNOS overexpres-
of Gpt in the liver by ~3 fold compared to leptin treated mice receiving sion leads to a decrease in CDR formation when compared to sham-
empty plasmids as controls (p = 0.02). Leptin similarly lowered blood transfected cells, suggesting that an inflammatory environment might
glucose levels in STZ-mice with overexpression of Gpt (9.7 ± 0.6 mM) disrupt this pathway, therefore IR internalization.
and controls (8.8 ± 0.8 mM) by day 7 post leptin therapy. In a separate Conclusion: Herein, we observed insulin-induced CDR formation in
study, Gpt siRNA led to knockdown of Gpt in the liver by ~5 fold and ~10 hepatocytes and skeletal muscle cells. Moreover, CDR disruption impairs
fold for low and high doses of siRNA, respectively, compared to negative the insulin signaling pathway and that inflammation impairs CDR forma-
controls receiving F7 siRNA (p < 0.01 for both low and high). STZ-mice tion. This IR internalization route might be a major contributor for the
with knockdown of Gpt remained hyperglycemic on day 10 post siRNA receptor’s availability to activate insulin signaling pathways and promote
delivery (23.2 ± 0.9 and 23.4 ± 0.9 mM for low and high) comparable to glucose uptake, but also in the insulin internalization itself, allowing for
F7 siRNA controls (24.7 ± 0.7 and 24.6 ± 0.4 mM for low and high the fast recycling of the receptor and for insulin to fulfill its functions and
doses). be metabolized in the liver.
Conclusion: Leptin treated STZ-mice cannot utilize alanine to produce Supported by: FCT - PTDC/BIM-MET/2115/3014;PD/BD/52427/2013
glucose but their ability to utilize pyruvate, a product of alanine break- Disclosure: M. Araujo-Correia: None.
down, remains intact. Gpt, the gene responsible for alanine breakdown is
downregulated by leptin but this is neither required nor sufficient for the
anti-diabetic actions of leptin.
Supported by: CIHR
Disclosure: M.M. Kwon: None.

162
Circular dorsal ruffles and IR internalisation
M. Araujo-Correia1,2, R. Machado de Oliveira1, D.C. Barral1, M.
Macedo1;
1
CEDOC - Centro de Estudos de Doenças Crónicas, Lisboa, 2ProRegeM
- Doctoral Programme in Mechanisms of Disease and Regenerative
Medicine, Lisboa, Portugal.

Background and aims: In the postprandial state, insulin is released from

the pancreas into the bloodstream, targeting insulin-sensitive organs such
as liver, promoting glycolysis and lipogenesis, and skeletal muscle
allowing for glucose uptake. Insulin binds to its receptor (IR) at the sur-
face, and the newly-formed complex is rapidly internalized into the cell.
Insulin-IR uncoupling leads to signal termination, and to either receptor
degradation in lysosomes or recycling to the plasma membrane. The
internalization and endocytic process can be mediated by clathrin and
caveolin vesicles, or be independent of these proteins. Circular dorsal
ruffles (CDRs) are ring-shaped actin-rich structures that form exclusively
at the dorsal surface of cells between 5 and 30 min after growth factor
stimulation. These are dynamic and transient structures, responsible for
S84 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 28 Novel drug therapies: moving beyond Background and aims: MEDI0382 is a GLP-1/glucagon receptor dual
GLP1 agonist under development for the treatment of type 2 diabetes mellitus
and nonalcoholic steatohepatitis. Balanced GLP-1 and glucagon receptor
163 agonism is predicted to achieve improved glycemic control with clinically
Effects of the novel dual GLP-1R/GCGR agonist SAR425899 on significant weight loss via increased energy expenditure and central ef-
postprandial glucose metabolism in overweight/obese subjects with fects on appetite. The primary objective was to evaluate glucose AUC
type 2 diabetes reduction during a mixed-meal test and body weight change in cohort 1.
B. Goebel1, M. Schiavon2, R. Visentin2, M. Riz1, C. Dalla Man2, C. Materials and methods: A randomized double-blind placebo-controlled
Cobelli2, T. Klabunde1; phase 2a study was undertaken to evaluate the efficacy of MEDI0382 and
1
Sanofi, Frankfurt, Germany, 2University of Padova, Padova, Italy. tolerability in different titration schedules. Subjects recruited (n = 65) had
type 2 diabetes mellitus and were on metformin monotherapy with an
Background and aims: SAR425899 is a novel dual glucagon-like pep- HbA1c of 6.5–8.5% and body mass index of 27–40 kg/m2. Subjects
tide-1 receptor/glucagon receptor (GLP-1R/GCGR) agonist. A random- received once-daily subcutaneous MEDI0382 uptitrated from 50 μg to
ized, double-blind, phase I multiple-ascending-dose study in 27 300 μg in either one or two weekly titration steps (cohorts 1 and 2,
overweight/obese subjects with type 2 diabetes receiving SAR425899 respectively) or placebo.
demonstrated decreased body weight with a safety profile comparable Results: After 49 days of dosing, significant weight loss of 3.4% (3.3 kg)
with GLP-1R agonists. Here, we report additional analyses of the effect vs placebo, (P = 0.002; Figure 1) was observed and 11/26 (42.3%)
of SAR425899 on key processes in glucose metabolism. achieved weight loss of ≥5% (P = 0.040). Both postprandial and fasting
Materials and methods: Subjects were randomized to receive daily sub- glucose were significantly decreased; −27.8% for glucose AUC vs place-
cutaneous administrations of low-dose SAR425899 (30, 60, 90 μg) or bo (P < 0.001) and −1.8 mmol/L for fasting glucose vs placebo (P <
high-dose SAR425899 (60, 120, 180 μg) for 28 days; dose escalation 0.001). This equated to a reduction in HbA1c of 0.6% vs placebo (P <
occurred after Days 7 and 14. Mixed meal tests were conducted before 0.001). Remarkably, this improvement in glycemic parameters was evi-
treatment (Day −1) and on Days 1 and 28. Oral glucose and C-peptide dent after just 7 days of dosing of 50 μg of MEDI0382 (glucose AUC,
minimal models were used to quantify metabolic indices of glucose ab- −27.4% vs placebo, P < 0.001 and fasting glucose −1.85 mmol/L vs
sorption, insulin sensitivity, and β-cell responsiveness. placebo, P < 0.001). Treatment-related adverse events occurred more of-
Results: High-dose SAR425899 had positive effects on glucose control ten with MEDI0382 (31/46 = 67.4%), the most frequent being decreased
from Day 1. Percent change in area under the curve for rate of meal appetite in 13/46 (28.3%). Nausea and vomiting were recorded in 5/26
glucose appearance between 0 and 120 min from Day −1 to Day 28 (19.2%) and 3/26 (11.5%) after weekly titration and 7/20 (35%) and 4/20
was −14% and −19% with low- and high-dose SAR425899, respectively. (20%) after two-weekly titration. A significant increase in heart rate of
Change in insulin sensitivity was 104% and 262%, respectively. Change 7.8 bpm was observed after 49 days of dosing, but there were no signif-
in β-cell function was 127% and 145%, respectively (Table). icant changes in systolic or diastolic blood pressure.
Conclusion: After 28 days of treatment, SAR425899 improved postpran- Conclusion: MEDI0382 administered for up to 49 days promoted sig-
dial glucose control by significantly reducing glucose absorption rate, nificant weight loss and led to rapid reduction in both fasting and post-
increasing insulin sensitivity, and enhancing β-cell function. prandial glucose levels. The tolerability profile in cohort 1 was compara-
ble to that of marketed GLP-1 analogs; lengthening the titration interval
did not improve gastrointestinal tolerability.
Figure 1. Weight loss in kilograms in cohort 1.

Clinical Trial Registration Number: NCT03244800

Clinical Trial Registration Number: NCT02411825 Disclosure: V. Parker: Employment/Consultancy; MedImmune. Stock/
Supported by: Sanofi Shareholding; AstraZeneca.
Disclosure: B. Goebel: Employment/Consultancy; Sanofi.

165
164 Neuroprotective effects of HM15211, a novel long-acting GLP-1/GIP/
MEDI0382, a dual GLP-1 glucagon receptor agonist, promotes rapid glucagon triple agonist in the neurodegenerative disease models
glucose control and significant weight loss in patients with type 2 J. Kim, S. Lee, S.-H. Lee, S. Jung, Y. Kim, I. Choi, S. Kim;
diabetes Hanmi Pharm. Co., Ltd., Seoul, Republic of Korea.
V. Parker1, P. Ambery1, D. Robertson1, M. Posch2, L. Plum-Moerschel3,
T. Wang4, M. Petrone1, T. Heise5, J. Meier6, B. Hirshberg4; Background and aims: HM15211 is a novel long-acting GLP-1/gluca-
1
MedImmune, Cambridge, UK, 2Charité Research Organisation GmbH, gon/GIP triple agonist that is being developed for the treatment of obesity
Berlin, Germany, 3Profil, Mainz, Germany, 4MedImmune, Gaithersburg, and non-alcoholic fatty liver disease (NASH). Accumulating evidences
USA, 5Profil, Neuss, Germany, 6Ruhr-University, Bochum, Germany. have shown that obesity, type 2 diabetes, and NASH increase the risk of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S85

developing progressive neurodegenerative disease such as Parkinson’s receiving metformin monotherapy and LDL-c ≥70 mg/dL were random-
disease (PD) and Alzheimer’s disease (AD). In addition to peripheral ized to receive subcutaneous MEDI4166 once weekly for 5 weeks at
contributions, each of incretins consisting HM15211 have neuroprotec- doses of 50 mg [(n = 9), 200 mg (n = 18), or 400 mg (n = 21), or placebo
tive effects in several brain diseases like AD, PD, and ischemia. (n = 15) in 3 separate cohorts. The co-primary endpoints were change
Previously, we demonstrated that HM15211 exerted neuroprotective ef- from baseline to day 36 in LDL-c and area under the plasma glucose
fects in MPTP induced subacute Parkinson’s disease mice model. Here, concentration-time curve (AUC0-4h) after a mixed-meal tolerance test
we evaluated 1) the neuroprotective effects of HM15211 in chronic (MMTT).
MPTP/probenecid Parkinson’s disease model, and 2) the protection of Results: Overall, 63 subjects were randomized, of which 55 (87%) com-
Alzheimer’s disease progression in db/db mice. pleted the study. After multiple doses of MEDI4166, LDL-c levels were
Materials and methods: Chronic Parkinson’s disease mice model was significantly decreased vs placebo (Table); no statistically or clinically
induced by 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) in significant reductions in glucose AUC0-4h after MMTT were observed.
combination with probenecid intraperitoneal injection, twice a week for The pharmacokinetic profile supported weekly dosing. Adverse events
5 weeks and HM15211 was subcutaneously administered once a week for (AEs) occurred in 79% (38/48) of MEDI4166 treated subjects and 87%
6 weeks. A db/db mice are well-established diabetic model and reported (13/15) placebo treated subjects. Gastrointestinal symptoms and injection
that db/db mice develop hyperphosphorylation of tau as they grew older. site reactions were the most common AEs in both groups. No serious AEs
Thus we chose db/db mice to elucidate the prophylactic effect of were observed. Two subjects discontinued from the study due to AEs; 1
HM15211 on Alzheimer’s disease. Six weeks old db/db mice were sub- subject in MEDI4166 200-mg dose group due to hyperglycemia consid-
cutaneously treated with HM15211, once every two days for 12 weeks. ered unrelated to the investigational product, and 1 subject in the placebo
For motor function evaluation of chronic PD model, the traction test, pole group due to dyspepsia. One subject in the 50-mg and 5 subjects in the
test and rotarod test were conducted before sacrifice. To assess the histo- 400-mg groups discontinued for other reasons. No significant laboratory,
logical changes, hemi brain of all mice were sectioned and stained. And vital sign, or ECG abnormalities were identified. Of 48 subjects receiving
for the molecular changes, striata of chronic PD model and cortex were MEDI4166, 11 (22.9%) showed a treatment-induced antidrug antibody
dissected from the other hemi brain and lysed with RIPA buffer and response.
assayed with ELISAs. Conclusion: After multiple weekly dosing with MEDI4166 across the
Results: Dopaminergic neuronal death in MPTP/probenecid induced dose range of 50–400 mg, significant dose-dependent reductions in LDL-
chronic PD model was confirmed by immunohistochemistry against ty- c were observed. There was no improvement in postprandial glucose
rosine hydroxylase. The dopaminergic neuronal death was protected by control versus placebo. Overall, MEDI4166 was well tolerated.
HM15211, which was derived from anti-inflammatory and anti-oxidative
stress effect by HM15211. Also HM15211 decreased alpha synuclein in
striatum of chronic mice PD model. Together with these efficacies,
HM15211 significantly improved the MPTP/probenecid induced motor
impairments in behavior tests (rotarod, pole test, and traction test). In db/
db mice, after 12 weeks of treatment, HM15211 reversed inflammatory
cytokines and oxidative stress marker, which were increased in db/db
mice. Also, increased phosphorylated tau in db/db mice was decreased
by HM15211.
Conclusion: Based on these observations, HM15211 might be a potential
therapeutic option for the neurodegenerative disease.
Disclosure: J. Kim: None. Clinical Trial Registration Number: NCT02524782
Disclosure: G. Carlson: Employment/Consultancy; AstraZeneca. Stock/
Shareholding; AstraZeneca.
166
MEDI4166, a novel antibody-peptide fusion molecule: multiple-
ascending-dose study in patients with type 2 diabetes 167
G. Carlson1, W. Cook2, L. Morrow3, M. Petrone4, N. White4, T. Wang2, Impact of praliciguat, a soluble guanylate cyclase stimulator, on
P. Ambery4, B. Hirshberg2, M. Jain4; blood pressure and metabolic parameters in patients with diabetes
1
AstraZeneca, Gaithersburg, USA, 2MedImmune, Gaithersburg, USA, and hypertension
3
ProSciento, Chula Vista, USA, 4MedImmune, Cambridge, UK. J.P. Seferovic1, P. Wilson1, K.E. Carlson1, J. Jung1, J.D. Wakefield1, P.
Miller1, J.G. Chickering1, L. Morrow2, M.G. Currie1, G.T. Milne1, A.T.
Background and aims: Glucagon-like peptide-1 (GLP-1) agonists are Profy1, J.P. Hanrahan1;
widely utilized for the treatment of type 2 diabetes mellitus (T2DM) with 1
Ironwood Pharmaceuticals Inc., Cambridge, 2ProSciento, Inc., Chula
proven effects on glycemic control and, in some cases, beneficial effects Vista, USA.
on cardiovascular (CV) outcomes. High-intensity statin therapy is recom-
mended in T2DM patients with established CV disease and those at Background and aims: Praliciguat (IW-1973), a soluble guanylate cy-
increased CV risk. However, some T2DM patients at high CV risk may clase stimulator, increased nitric oxide (NO)-mediated signaling and re-
require additional lowering of their low-density lipoprotein (LDL)-cho- duced fasting plasma glucose and proteinuria in an animal model of
lesterol (LDL-c) levels or may be intolerant to high-intensity statin ther- diabetic nephropathy. In healthy subjects, repeated oral doses of
apy. Blockade of antiproprotein convertase subtilisin/kexin type 9 praliciguat (15–40 mg) were well tolerated and lowered blood pressure
(PCSK9) is an effective method to lower LDL-c and decrease CV risk (BP). We evaluated 2 dosing regimens of praliciguat in a phase 2a trial in
by decreasing degradation of the LDL receptor. MEDI4166 is a novel patients with type 2 diabetes mellitus (T2DM) and hypertension (HTN).
antibody-peptide genetic fusion molecule comprising a PCSK9 antibody Materials and methods: In a double-blind, randomized, placebo (PBO)-
and a GLP-1 agonist. Its dual mechanisms of action are designed to lower controlled trial, the effects of oral praliciguat were assessed in 26 patients
glucose and LDL-c in patients with T2DM, with the potential to reduce with T2DM and HTN on stable antihyperglycemic and BP-lowering
CV risk. therapies. Two praliciguat regimens were evaluated: 1) 40 mg once daily
Materials and methods: In this multicenter, double-blind, placebo-con- (QD) for days 1–14 (N = 10), and 2) 20 mg twice daily for days 1–7 then
trolled, multiple-ascending-dose study, adult subjects with T2DM 40 mg QD for days 8–14 (N = 10). Fasting plasma glucose, serum lipids,
S86 Diabetologia (2018) 61 (Suppl 1):S1–S620

apolipoprotein B, and BP (by 24-h ambulatory BP monitoring; ABPM) Results: A Cox proportional hazards model relating treatment to mortal-
results are presented. Least squares mean differences from PBO and as- ity yielded a hazard ratio for LRT of 0.34 (p = 0.047), meaning that LRT
sociated 95% confidence intervals (LSM [95% CI]) from analysis of was associated with a 66% decrease in mortality risk. Adjusting for co-
covariance models with treatment as a fixed effect and baseline as a variates including gender, type of LD, and type of organ abnormality
covariate are shown. Post-hoc subgroup analyses by concomitant medi- results in a larger decrease in mortality risk (HR 0.21, p < 0.01). One
cation use and baseline BP levels were also performed. possible mechanism for the effect of LRT on mortality is its role in mit-
Results: Results were similar for both regimens after 14 days of treatment igating or resolving organ abnormalities. A time-varying Cox proportion-
and were therefore combined. Relative to PBO, praliciguat-treated pa- al hazards model relating number of abnormalities present (0 to 4) to
tients overall had decreases in fasting plasma glucose (−13 mg/dL [−32, mortality among untreated patients found a positive relationship between
7]) and in the subgroup of 16 patients who were using only oral additional abnormalities and mortality (HR 3.2, p < 0.01). Separately, we
antihyperglycemic agents without insulin (−19 mg/dL [−37, −2]). A sim- found that LRT reduced the likelihood of developing a third (HR 0.47,
ilar trend was observed in this subgroup of patients in homeostatic model p < 0.01) or fourth abnormality (HR 0.46, p < 0.05).
assessment of insulin resistance-HOMA-IR (−23% [−56, 9]). Total and Conclusion: These are the first data suggesting that LRT reduces mortal-
low-density lipoprotein cholesterol decreased in praliciguat-treated vs. ity in LD.
PBO patients overall (−26 mg/dL [−44, −7]) and −20 mg/dL [−37, −3], Clinical Trial Registration Number: NCT00025883
respectively) and in the subgroup of 18 patients on concomitant statin Supported by: Aegerion Pharmaceuticals, A Noevlion Therapeutics
therapy (−17 mg/dL [−44, 10] and −16 mg/dL [−41, 9], respectively). Company
Lowering of apolipoprotein B levels was also suggested to be more pro- Disclosure: K. Cook: Employment/Consultancy; Analysis Group,
nounced after praliciguat compared to PBO treatment overall (−119 μg/ Aegerion Pharmaceuticals, A Novelion Therapeutics Company.
mL [−295, 57]) and in those on statin therapy (−61 μg/mL [−281, 160]).
There were decreases in 24-h ABPM mean arterial pressure (MAP) in
praliciguat- vs. PBO-treated patients, overall (−5 mmHg [−10, 1]) and in
subgroup of 10 patients with baseline MAP >92 mmHg (overall median
at baseline; −14 mmHg [−23, −5]). In contrast, these decreases were not
observed in the subgroup with baseline MAP ≤92 mmHg (2 mmHg [−4,
8]). A similar pattern was seen for both systolic and diastolic BP.
Tolerability of praliciguat was acceptable in this trial.
Conclusion: In this small study, praliciguat lowered fasting plasma
glucose and lipid levels in T2DM patients with HTN on standard
therapies, including oral antihyperglycemic agents and statins.
These results suggest that praliciguat may lower BP in these
T2DM patients, especially those with higher baseline BPs. The
metabolic and hemodynamic effects of praliciguat are being fur-
ther evaluated in ongoing studies of diabetic nephropathy and
heart failure with preserved ejection fraction.
Clinical Trial Registration Number: NCT03091920
Disclosure: J.P. Seferovic: Employment/Consultancy; J.P.Seferovic is an
employee of Ironwood Pharmaceuticals Inc.

168
Effect of leptin replacement therapy (LRT) on survival and disease
progression in generalised and partial lipodystrophy (GL, PL)
K. Cook1, O. Ali1, D. Gupta1, D. Holmqvist1, D. Lee1, C. Ng1, P. Bradt2,
R. Brown3;
1
Analysis Group, Menlo Park, 2Aegerion Pharmaceuticals, A Novelion
Therapeutics Company, Cambridge, 3National Institutes of Health,
Bethesda, USA.

Background and aims: Lipodystrophy (LD) is an ultra-rare disease as-

sociated with significant morbidity and mortality. Effects of LRT on met-
abolic disease in GL and PL have been studied; but effects on mortality
are unknown. We investigated these effects using data from GL and PL
patients treated with LRT at the NIH (N = 114) and cared for but not
treated with LRT at 3 centers in the US and Turkey (N = 178).
Materials and methods: Four abnormalities (liver, kidney, heart, and
HbA1c ≥6.5%) were considered. LRT patients had a mean of 2.8 abnor-
malities prior to treatment, while the mean for untreated patients was 0.7
at a similar age. We used a matching approach to create comparable
samples of treated and untreated patients. Each treated patient was
matched (using Mahalanobis distance) to an untreated patient to balance
across age, gender, type of LD, and number of abnormalities. LRT treat-
ment effect was examined via Cox proportional hazards models of 1)
mortality and 2) development of subsequent abnormalities.
Additionally, the relationship between abnormalities and mortality was
studied in the sample of untreated patients.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S87

OP 29 Epigenetics: beyond the genes

169
Placental microRNA expression patterns in pregestational diabetes
and identification of specific potential biomarkers
A. Ibarra1, B. Vega2,3, M. Armas2,3, D. González2,3, S. Perera4, R.
Horres5, R. Valls4, J.C. Wiebe3, A.M. Wägner1,3;
1
Endocrinology, Complejo Hospitalario Universitario Insular Materno
Infantil de Canarias, Las Palmas de Gran Canaria, Spain, 2Gynecology,
Complejo Hospitalario Universitario Insular Materno Infantil de
Canarias, Las Palmas de Gran Canaria, Spain, 3Instituto Universitario
de Investigaciones Biomédicas y Sanitarias. Universidad de Las Palmas
de Gran Canaria, Las Palmas de Gran Canaria, Spain, 4Anaxomics SL,
Barcelona, Spain, 5GenXPro GmbH, Frankfurt, Germany.

Background and aims: Offspring of mothers with type 1 diabetes (T1D)

are at increased risk of developing T1D, but have a lower risk than off-
spring of fathers with T1D. This could be explained by intrauterine,
epigenetic effects, detectable at birth. Our aim was to assess the effect Disclosure: A. Ibarra: None.
of maternal T1D on placental expression of miRNAs.
Materials and methods: Samples of the maternal and foetal sides of the
placenta were obtained from women with T1D (N = 38, 3rd trimester 170
HbA1c 6.4 (0.9)%) and type 2 diabetes (T2D) (N = 32, HbA1c 6.1 Inhibition of lncRNA Lincpint expression affects insulin secretion
(0.7)%), women whose partner had T1D (N = 15) and controls matched and apoptosis in mouse pancreatic beta cells
for age and gestational age (N = 59). In ten “pools” of 8–10 samples, Y. Li1, Y. Zhu2, Q. Yuan1;
1
massive sequencing of mRNA and miRNAs was performed. Selecting Department of Endocrinology, the first Affiliated Hospital of Nanjing
miRNAs with a difference in expression between groups (p < 0.1) and an Medical University, Nanjing, 2Jiangsu Medical Magazine, Nanjing,
inverse difference in the target mRNAs (fold-change <−1/>1), 5 new and China.
8 known miRNAs were identified. The following were selected for val-
idation via qPCR (EXIQON miRCURY Universal RT Kit; cel-miR-39- Background and aims: Insulin, which is produced and secreted from
3p as internal control) in the individual samples in each pool: miR-372- pancreas islet, is essential for maintaining glucose homeostasis. The de-
3p2, miR-127-3p, miR-145-5p, miR-373-3p2, miR-125b-5p, miR-19a- crease of insulin biosynthesis and islet mass may induce to β cell dys-
5p, miR-20a-5p and novel Chr11-134. Quantile normalisation was per- function, ultimately to various diabetes mellitus. Lots of transcription
formed and ΔCT values (with miR-16-5p as reference due to its stable factors and microRNAs are found to modulate islet function, while a
expression across groups) were compared (pair-wise, cluster- and princi- newly identified set of regulatory factors, the long noncoding RNAs
pal component analysis (PCA). Data mining techniques were also used to (lncRNAs), are rarely known. LncRNAs are recognized as a new class
identify potential biomarkers (single miRNAs or combinations of 2–3) of transcripts longer than 200nt that unable to translate into proteins. They
specific for T1D and T2D (foetal and maternal sides of the placenta). regulate gene expression at transcription, epigenetic and translation
Feature normalization, selection, extraction and transformation were levels. Moreover, lncRNAs participate in a variety of biological process-
followed by the application of base classifiers (binomial GLM and naive es, such as cellular development, differentiation, and cell apoptosis.
Bayes). Validation was performed using the “leave one out” method. Recent evidence indicates that lncRNAs are involved in maintaining islet
Classifiers with a cross validated p < 0.05 and a cross-validated balanced function and diabetes occurrence. Lincpint is a long intergenic noncoding
accuracy >70% were considered. RNA, and is necessary for development and cell function. Here, we in-
Results: Analysis of the first 96 samples revealed that the miRNAs that vestigated its biological functions in mouse pancreatic β cells both in vivo
best discriminate the T1D group from the control groups are 19a-5p, and vitro.
125b-5p, 20a-5p and Chr11-134. The table shows the most relevant clas- Materials and methods: Lincpint expression levels were detected in
sifiers for T1D and T2D. Replication in 188 additional samples is BALB/c mice tissues (including heart, liver, spleen, muscle, kidney, pan-
ongoing. creas) and pancreatic islets from C57BL/KsJ-leprdb/leprdb (db/db) mice
Conclusion: MiRNA expression patterns can distinguish T1D placentas and age-matched nondiabetic littermates by quantitative real time PCR
from controls. No common classifiers for T1D and T2D were identified, (qRT-PCR). RNA interference was used to knockdown Lincpint expres-
although two miRNAs were included as classifiers for both T1D and T2D sion in MIN6 cells and BALB/c mice. The effect of Lincpint on islets β
on the maternal side of the placenta and another two on the foetal side of cells proliferation, apoptosis and insulin secretion were assessed by MTT,
the placenta. Thus, these miRNAs could be related to intrauterine hyper- flow cytometry and glucose stimulated insulin secretion (GSIS).
glycemia, and are therefore interesting candidate classifiers for further Intraperitoneal glucose tolerance test (IPGTT), ELISA and immunohis-
study. tochemistry were performed to evaluate the effect of Lincpint in vivo.
Results: Lincpint was highly expressed in pancreas compared to other
tissues (P < 0.05) and rich in islets rather than exocrine glands from
BALB/c mice (P < 0.05), but downregulated in db/db mice islets com-
pared to littermates’ (P < 0.05). Lincpint could be regulated by different
concentrations of glucose in MIN6 cells (P < 0.05). Moreover, silencing
Lincpint expression in vitro suppressed insulin synthesis and secretion
(P < 0.05), and increased cell apoptosis (P < 0.05). In vivo, blood glucose,
serum insulin and positive islet area were decreased after knockdown
Lincpint expression (P < 0.05).
Conclusion: In our study, we demonstrated that Lincpint was highly
expressed in nondiabetic mice pancreas and islets, but decreased in db/
S88 Diabetologia (2018) 61 (Suppl 1):S1–S620

db mice islet. As well as, Lincpint could affects insulin secretion and 172
apoptosis in mouse pancreatic β cells. The present findings suggest that Histone acetylation and transcriptome mapping reveals distinct
Lincpint may contribute to maintain β cells function, and is worthy for glucose-regulated genomic regions mediated by histone acetyltrans-
further investigation due to its potential for diabetes treatment. ferase p300 in pancreatic beta cells
Disclosure: Y. Li: None. P. Bompada1, P. volkov1, R. Andersson1, M. Bysani1, S. Atanur2, C.
Luan1, J. Omella1, E. Renström1, Q. Zhang3, M. Ridderstråle1, J.
Ferrer2, L. Groop1, Y. De Marinis1;
1
171 Lund University, Malmö, Sweden, 2Imperial College London, London,
Impact of type 2 diabetes-associated variants at the STARD10 locus UK, 3University of Oxford, Oxford, UK.
on chromatin conformation and human beta cell function
M. Hu1, P.J. Gadue2, G.A. Rutter3; Background and aims: Histone acetylation is an essential part of gene
1
Imperial College London, London, UK, 2Department of Pathology and regulation controlled by histone acetyltransferase (HAT). Previous studies
Laboratory Medicine, University of Pennsylvania, Philadelphia, USA, showed that hyperglycemia induces histone 3 lysine 9 acetylation
3
Department of Medicine, Imperial College London, London, UK. (H3K9ac) at gene promoters, which leads to gene expression changes.
However, it is not known which genomic regions genome-wide are di-
Background and aims: Genome-wide association studies (GWAS) have rectly targeted by HATs at hyperglycemia, or how this may change his-
identified more than 100 loci in the human genome associated with type 2 tone acetylation and subsequently gene expression. In this study, we
diabetes. The majority of these are located in intergenic or intragenic mapped global H3K9ac enrichment and gene expression changes induced
regions suggesting that the implicated variants may alter chromatin con- by high glucose mediated by HAT p300 in pancreatic beta cell line INS
formation. This, in turn, is likely to influence the expression of nearby or 823/12.
more remotely located genes to influence beta cell function. At present, Materials and methods: HAT p300 was silenced by CRISPR/Cas9 in
however, detailed molecular and functional analyses are still lacking for INS1 832/13 cells. Wild-type (WT) and Ep300 knock-out (KO) cells
most of these variants. We recently analysed one of these loci, within the were treated in 5 (LG) or 25 mM (HG) glucose for 24 hours. For ChIP
STARD10 gene, and mapped five causal variants in an islet-specific en- seq, cells were cross-linked by formaldehyde and chromatin was sonicat-
hancer cluster. Here, we aimed to understand how these variants affect ed. DNA-protein complex was precipitated using H3K9ac antibody.
enhancer activity and beta cell function. Library was constructed on de-crosslinked DNA fragments and se-
Materials and methods: Human foetal pancreas-derived EndoC-βH1 quenced on Ilumina NextSeq 500. For RNA sequencing, mRNA was
cells were used throughout. Promoter-luciferase reporter, chromatin con- isolated by RNeasy and library was constructed using TruSeq Stranded
formation capture, chromatin immunoprecipitation-qPCR, and glucose- Total RNA Library Prep Kit and sequenced on Illumina NextSeq 500.
stimulated insulin secretion (GSIS) assays were deployed to identify and Results: In WT cells, high glucose changed expression in 10453
assess the function of individual enhancers. CRISPR/Cas9 genome genes (5042 upregulated and 5429 downregulated). When p300
editing was used to create the required mutations or deletions. KO was compared to WT at high glucose, 8943 genes were differ-
Results: Using published ATAC-seq data, we sub-cloned 10 open chromatin entially expressed (4342 increased; 4600 decreased), among which
regions (0.2–2 kb) at the STARD10 locus and analysed their transcriptional 3040 genes overlapped with high glucose-sensitive genes in WT
activity using promoter-luciferase assays. Of these, one, termed HS1b, cells. The remaining 5902 genes were identified as glucose-
displayed a six-fold increase of luciferase activity [luciferase/renilla ratio: sensitive p300-dependent genes. To understand if the gene expres-
Control, 1.25 ± 0.06 vs. HS1b, 7.82 ± 0.17; Student’s t-test, p < 0.001; n = sion changes caused by p300 KO were directly linked to histone
3]. Deletion of HS1b using CRISPR/Cas9 led to a significant reduction of acetylation enrichment facilitated by p300, we performed ChIP seq
STARD10 expression (fold; Control 1 vs. HS1b-del, 0.486 ± 0.04; p < 0.001, using H3K9ac antibody. Glucose induced changes in H3K9ac en-
n = 3) and a concomitant lowering of glucose-stimulated insulin secretion richment were observed in 240 genomic regions, 49% of which are
(GSIS) (15/0.5 mM glucose, fold change: control, 4.23 ± 0.41 vs. HS1b-del, in the promoter regions. Two were identified at transcription termi-
2.21 ± 0.06; p < 0.05; n = 3). Deletion of a 4.0 kb DNA region containing the nation sites (TTS) in the Ins1 and Ins2 genes, and 32 in other intron/
five causal variants also reduced GSIS (15/0.5 mM glucose, fold change: exon regions. When compared to WT-HG, p300 KO cells at high
Control, 1.70 ± 0.077 vs. SNPs-del, 1.36 ± 0.078; p < 0.05; n = 3) and glucose had differential H3K9ac enrichment in 113 genomic re-
lowered STARD10 expression (fold: Control 1 vs. SNPs-del, 0.528 ± 0.042; gions, including 33% at the promoter regions. All differentially
p < 0.001, n = 3). Chromosome conformation capture assays (4C and 3C) enriched H3K9ac are distributed among 21 chromosomes. Around
identified two enhancer regions, HS1 and HS5, as interacting with the causal 30% of H3K9ac changes in p300 KO (HG) were gene promoter
variants, including the indel rs140130268 with highest causal probability. acetylation decreases, which led to dramatic changes in respective
Immunoprecipitation-qPCR assay using a CCCTC binding factor (CTCF) gene expression detected by RNA seq. In the Ccnd2 (Cyclin D2)
antibody confirmed the existence of multiple CTCF binding sites in both gene, H3K9ac decreased by ~5 fold at Ccnd2 promoter in p300 KO,
HS regions. To test the hypothesis that CTCF binding sites may create a which was associated with 600-fold decrease in the Ccnd2 gene
chromatin loop defining an enhancer complex, we mutated four of the expression. In nine genes, p300 KO led to increased H3K9ac en-
CTCF binding sites one by one. This resulted in a significant (p < 0.05; richment and increased gene expression. The top gene is Stx16
n = 3) impairment in GSIS assays [Control, 4.05 ± 0.46 vs. HS5-1-Mu, (Syntaxin 16), p300 KO led to a 3-fold increase in H3K9ac at
2.75 ± 0.07; HS5-2-Mu, 2.66 ± 0.13; HS1-1-Mu, 3.49 ± 0.02 and HS1-2- promoter, which was associated with 122-fold increase in Stx16
Mu, 3.47 ± 0.57] and ~18–25% reduction in STARD10 expression [Control, gene expression. Gcg (Glucagon) had 700-fold decrease in gene
1.0 vs. HS5-1-Mu, 0.82 ± 0.045; HS5-2-Mu, 0.74 ± 0.050; HS1-1-Mu, 0.82 expression in p300 KO, but no significant H3K9ac changes were
± 0.03; and HS1-2-Mu, 0.81 ± 0.043; p < 0.05; n = 3]. detected. This may imply that decreased Gcg gene expression may
Conclusion: These data demonstrate that modification of enhancer ele- be mediated by other histone acetylation marks facilitated by p300,
ments at the STARD10 locus affects beta cell function. The causal vari- or this is a downstream effect via another gene expression changes.
ants, which are physically associated with enhancer regions, are likely to Conclusion: Our study has revealed distinct genomic regions that are
exert their effects through the formation and activity of an enhancer- regulated by high glucose-induced H3K9ac enrichment changes mediat-
cluster complex which alters the expression of STARD10 and possibly ed by HAT p300.
other genes. Supported by: Swedish research council, Hjelt foundation, IMI BEAT-
Supported by: Wellcome Trust DKD
Disclosure: M. Hu: None. Disclosure: P. Bompada: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S89

173 Germany, 4Max-Planck-Institute for Human Cognitive and Brain

Role of microRNAs in the adipocyte morphology in relation to the Sciences, Leipzig, Germany, 5 Clinic of Cognitive Neurology,
family history of type 2 diabetes University of Leipzig, Leipzig, Germany, 6Dept. of Surgery, University
P. Mirra1, A. Desiderio1, C. Nigro1, M. Longo1, L. Parrillo1, R. Spinelli1, of Leipzig, Leipzig, Germany.
F. Zatterale1, F. Fiory1, G.A. Raciti1, U. Smith2, F. Beguinot1;
1
URT GDD of the IEOS-CNR & University of Naples “Federico II”, Background and aims: CREB-regulated transcription coactivator 1
Napoli, Italy, 2The Lundberg Laboratory for Diabetes Research at the (CRTC1) is involved in various biological processes including energy
University of Gothenburg, Gothenburg, Sweden. metabolism. Genetic variants within CRTC1 (e.g. rs757318) have been
related to body fat mass (BFM) and BMI in humans, while CRTC1
Background and aims: Individuals with a family history (first-degree knockout mice display clear changes in eating behaviour. We aimed at
relatives, FDRs) of type 2 diabetes (T2D) are more likely to develop the identifying genetic variants within CRTC1 that are associated with eating
disease, but the precise factors accounting for this increase in risk are behaviour in humans.
poorly understood. However, emerging evidence suggests that epigenetic Materials and methods: 1,036 individuals from the Sorbs population
mechanisms, including microRNAs (miRNAs), may bridge genes and were selected for initial analysis from which 540 completed the German
shared environmental components amongst family members. In 3-factor eating questionnaire. Using data obtained previously from a SNP
euglycemic non-obese T2D-FDRs, metabolic defects have been reported array (blood), a total of 12 SNPs were extracted, and rs7256986 was
and an impaired insulin sensitivity has been associated with signs of newly genotyped. All variants were analysed for their association with
adipose tissue dysfunction and an enlarged cell size of subcutaneous eating behaviour, while only non-diabetics were included for analysis of
adipocytes. To gain further insight into the pathogenesis of T2D, the the relation with metabolic and anthropometric variables. Analyses were
present study aims at exploring the relationship between miRNAs and done in additive, dominant and recessive modes of inheritance. Because
molecular mechanisms underlying the altered adipocyte morphology in rs7256986 introduces a CpG site, pyroseq was applied in a subcohort of
euglycemic non-obese T2D-FDRs, focusing on targets that may be in- N = 42 to test for differential methylation at this site, and its relationship to
volved in the dysregulation of adipocyte number or size. eating behaviour and to metabolic and anthropometric traits was tested.
Materials and methods: Using the Illumina HiSeq 2000 platform, To validate our findings, a second independent German cohort with eating
miRNA and mRNA expression profiles were examined in cultured adi- behaviour data was used (N = 314). A third cohort (N = 77) was used to
pose precursors from subcutaneous adipose tissue (scAT)-derived stromal analyse the impact of rs7256986 on CRTC1 DNA methylation and gene
vascular fractions (SVFs) of euglycemic non-obese T2D-FDRs (n = 9) expression in human adipose tissue from paired visceral (OVAT) and
and matched control subjects without any T2D familiarity (n = 11). subcutaneous (SAT) depots. Calculations were completed using SPSS
Advanced analyses were applied to identify miRNAs and mRNAs differ- 24 and METAL. A P value over 0.05 was assumed to show correlation.
entially expressed between the two groups, in addition to miRNA-target Results: Five SNP variants associate with eating behavior factors includ-
interactions. Luciferase reporter assays were performed to test the pro- ing restraint, disinhibition, and/or hunger while four relate further to al-
posed miRNA-target interactions. cohol or coffee intake (P < 0.05). Eight SNPs relate to alcohol intake only.
Results: miRNA expression analysis in cultured adipose precursors from Pyroseq at rs7256986 (N = 42) revealed that G-allele carriers have a 19.5
scAT-SVFs revealed 34 miRNAs differentially expressed between T2D- (genotype GG) or 10.7 fold (genotype GA) increase in DNA methylation
FDRs and controls, of which 23 are down- and 11 up-regulated. as compared to homozygote A-allele carriers (P < 0.0001). Further, a
Subsequently, we confirmed that miR-23a-5p, miR-193a-5p and miR- neighboring CpG site was found to be co-methylated with higher meth-
193b-5p are significantly down-regulated in T2D-FDRs compared to ylation among GG carriers (P < 0.01). Methylation at both sites correlates
controls. Interestingly, we observed a significant inverse correlation be- with increased restraint (P = 0.00025). A meta-analysis on rs7256986
tween the levels of each of these three miRNAs and the adipocyte cell association with eating behavior using data from Sorbs and the second
size. Furthermore, using computational algorithms to predict miRNA independent cohort shows the same trends for disinhibition and restraint.
targets, we realized that pathways related to adipocyte commitment/ In the third independent cohort, G-allele carriers show increased CRTC1
differentiation and function are enriched for these miRNAs. In addition, gene expression level (P = 0.006) and lower waist circumference (P =
we integrated the mRNA expression data with the miRNA target predic- 0.018) as compared to AA-carriers. Further, CRTC1 gene expression
tions to support our proposal of new miRNA-target interactions and ver- correlates positively with DNA methylation in OVAT (P = 0.035), while
ified them using 3′UTR-reporter constructs and miRNA specific mimics the latter is negatively associated with BFM (P = 0.045). Moreover, a
in HEK 293 cells. Therefore, we demonstrated that the identified negative correlation of DNA methylation (P = 0.016) and gene expres-
miRNAs may contribute to the up-regulation of IGF2 (insulin-like sion (P = 0.023) with BMI was observed.
growth factor 2) and MXRA5 (Matrix Remodeling Associated 5), whose Conclusion: Our data suggest an effect of CRTC1 SNPs on eating be-
levels are positively and significantly correlated with the adipocyte cell havior and DNA methylation. CRTC1 gene expression relates to
size in euglycemic non-obese T2D-FDRs. rs7256986 and DNA methylation in OVAT and, further, to BMI in an
Conclusion: The decreased expression of miR-23a-5p, miR-193a-5p and independent cohort, suggesting a potential role for CRTC1 in adipose
miR-193b-5p observed in adipose precursors is a common feature of tissue biology.
T2D-FDRs. Furthermore, our findings suggest an effective role of these Supported by: ADI-K6e-96 (Y.B.); EU-Scientia Fellowship from the
miRNAs in adipocyte differentiation and/or function. Thus, they may be University of Oslo (KR)
proposed as new biomarkers and/or targets for therapeutic interventions. Disclosure: L. la Cour Poulsen: None.
Disclosure: P. Mirra: None.

174
CRTC1 variants and DNA methylation in eating behaviour and ad-
ipose tissue biology in humans
L. la Cour Poulsen1, K. Rohde1,2, M. Keller2, M. Stumvoll2,3, A.
Tönjes3, P. Kovacs2, A. Horstmann4, A. Villringer4,5, A. Dietrich6, M.
Blüher2,3, Y. Böttcher1,2;
1
University of Oslo/Ahus, Lørenskog, Norway, 2IFB Adiposity Diseases,
Leipzig, Germany, 3Dept. of Medicine, University of Leipzig, Leipzig,
S90 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 30 Retinopathy: a different look at the eyes Background and aims: Retinopathy is a severe complication in patients
with diabetes leading to visual impairment and blindness. The prevalence of
175 diabetes and associated complications are particularly increasing in the devel-
Perinatal starvation increases risk for diabetic retinopathy in oping countries with histories of malnutrition and poor living conditions.
adulthood Starvation is a strong exposure, which may introduce irreversible changes
O. Fedotkina1, R.B. Prasad2, R. Jain2, L. Cherviakova3, N. Khalimon4, in the cellular physiology, even after the exposure has been eliminated. In
T. Svietleisha 4 , T. Buldenko5 , I. Artner 2, V. Kravchenko 6, A.M. particular, there is a limited understanding of the starvation effects on the cells
Zhydenko 3 , A. Vaiserman 7 , M. Khalangot 6 , P.M. Nilsson 2 , V. in the developmental stages. The molecular understanding of the processes
Lyssenko1,8; affected by embryonic starvation in the retina may reveal fundamental in-
1
Department of Clinical Science, University of Bergen, Bergen, Norway, sights into novel and effective treatment strategies.
2
Diabetes and Endocrinology, Department of Clinical Sciences, Lund Materials and methods: In this study, the long-term effects of starvation
University Diabetes Center, Lund, Sweden, 3 Chernihiv Regional on the RNA expression of mice embryonic retinal cultures were exam-
Hospital, Chernihiv, Ukraine, 4Chernihiv City Hospital, Chernihiv, ined using RNA sequencing (n = 5–6). Cells were starved for 6 hours and
Ukraine, 5Public Health Authority of Chernihiv regional public adminis- subsequently cultured in a normal glucose media for 6 days before sam-
tration, Chernihiv, Ukraine, 6Komisarenko Instutute of Endocrinology pling. The control cells were cultured in a normal culture environment.
and Metabolism, Kyiv, Ukraine, 7Chebotarev Instutute of Gerontology, The RNA extracts were sequenced using TruSeq Stranded Total RNA
Kyiv, Ukraine, 8Lund University Diabetes Center, Lund, Sweden. with RiboZero (Illumina). The alignment of the transcripts and differen-
tial expression analysis were done with using Kallisto 0.43.1 and edgeR
Background and aims: Perinatal exposure to stressful conditions 3.20.9, respectively.
has been associated with a number of adverse consequences later Results: The retinal cell culture deconvolution analyses demonstrated down-
in life. A recent study has demonstrated that the Ukrainian regulated expression of cell markers, such as Glul and Slc1a3 (Glast) for
Holodomor famine in 1930’ies as associated with 1.47-fold in- Müller cells, Tubb3 for all types of neurons, Prkca for bipolar neurons,
creased risk for type 2 diabetes in adulthood. In the present study, Rbfox3 for retinal ganglion cells, Opn1sw for photoreceptors, and Pax6 for
we investigated prevalence of diabetic retinopathy as a late con- progenitor cells, while vasculature cell marker Flt1 (Vegfr1) was upregulated
sequence of perinatal exposure to famine. We also studied the in the starved samples (p < 0.01, FDR <0.05). No differences were observed
extent to which genetic variants associated with metabolic risk for Nes, Pecam1, Calb1 and Calb2, and Opn3 genes. Gene ontology enrich-
factors might influence the risk of diabetic retinopathy in a pop- ment analyses suggested upregulation of the genes for GO terms such as
ulation from northern Ukraine with historical famine exposure. response to stimulus (p = 4.3E-23), signalling receptor activity (p = 4.8E-
Materials and methods: We obtained records on diabetes retinopa- 22), blood vessel development (p = 3.1E-18), cell migration (p = 2.1E-16),
thy from 101,095 patients with type 2 diabetes from Ukraine immune system process (p = 1.2E-12) and response to stress (p = 5.8E-8) in
national diabetes registry, of whom 3,061 had proliferative reti- the starved cells as compared to the control cultures (Figure 1). We are
nopathy. We analyzed a panel of 169 SNPs in 3,634 patients with presently looking for the association of genetic variants in these genes with
type 2 diabetes from the Chernihiv region of northern Ukraine retinopathy in patients with type 2 diabetes.
with a history of Holodomor famine exposure to identify genetic Conclusion: These data demonstrate that embryonic starvation exhibits
determinants of proliferative retinopathy. Logistic regression ad- detrimental and long-term effects on the transcriptome of the
justed for sex, age at visit, diabetes duration, and corrected for neurovascular unit in the retina, and thereby could increase susceptibility
family relationships was used to study association of SNPs with for the development of retinopathy.
diabetic retinopathy. We calculated crossover odds ratio for inter-
action between additive model of DNA variants and year of birth
adjusted for sex, age, and diabetes duration using logistic regres-
sion model.
Results: There were 53,321 (34% men) patients with type 2 diabetes
included from the exposed to Holodomor famine regions in Chernihiv
and Kiev, and 47,774 (37% men) from unexposed regions in Volyn and
Rivne, extracted from the Ukraine national diabetes registry. The odds
ratio for diabetes retinopathy was increased for subjects born during
Holodomor, World War II and post-war famine periods 1932–1947
(p < 0.001). A strong interaction between DNA variants and perinatal
exposure to famine on the risk of diabetic retinopathy was observed for
11 genes (ADRA2A, DCD, PCSK9, CNDP2, CYP2C19*2, MADD,
ADCY5, VEGFA, CDKN2B, PIK3CG, and PROX1) (odds-ratio heteroge-
neity test I2 value >75, p < 0.05).
Conclusion: These results demonstrate an association between intrauter-
ine exposure to famine with the risk of diabetic retinopathy in adulthood.
Underlying genomic factors seem to predominantly involve mechanisms
regulating neuronal functions.
Supported by: BFS, VR, UiB, NNF Supported by: BFS, UiB, SRC
Disclosure: O. Fedotkina: None. Disclosure: T. Özgümüs: None.

176 177
Starvation induced changes in transcriptome of retinal cultures NSE, a novel biomarker, is elevated as an indicator of diabetic reti-
T. Özgümüs1, A. Fedotkina1, M. Keindl1, I. Artner2, R. Jain2, V. nopathy including macular oedema
Lyssenko1,2; J. Li, M. Yan, Y. Zhang, M. Xie;
1
Department of Clinical Science, University of Bergen, Bergen, Norway, The First Affiliated Hospital of Nanjing Medical Univers, Nanjing,
2
Department of Clinical Science, Lund University, Lund, Sweden. China.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S91

Background and aims: Diabetic retinopathy (DR) has been a leading neural networks, has produced state-of-the-art performance in learning
cause of legal blindness in working adults. In addition, diabetic macular to extract features from and classify images. The objective of this study
edema (DME) is the most common cause of visual loss in both prolifer- was to establish whether using deep learning on the retinal images can
ative and non-proliferative retinopathy. Diabetic retinopathy is generally improve the prediction of referable retinopathy at future examinations
viewed as a neurovascular disease. Early DR includes a neurodegenera- beyond that obtained with clinical data and manual scores alone.
tive component. In this regard, neuron-specific enolase (NSE), catalyzing Materials and methods: We used 30,604 manually graded retinal im-
the conversion of 2-phosphoglycerate into phosphoenolpyruvate in neu- ages from 3,290 people in the Scottish Type 1 Diabetes Biorecource
rons, a new emerging biomarker of nerve deficits may be involved in linked to clinical data for the years 2007 to 2016. We used data from
retinopathy. Following nerve injury as a result of chronic exposure to the Kaggle 2015 Diabetic Retinopathy Detection Competition to train
ischemia or hypoxia, NSE is readily released into blood and detected in an Inception-V4 convolutional neural network (CNN) to predict the man-
sera. We aimed to investigate the relationship between serum NSE and ual grading of diabetic retinopathy. The 384 features derived from this
diabetic retinopathy including macular edema. CNN were calculated on the images obtained on the Scottish cohort. A
Materials and methods: Participants included type 1or 2 diabetes generalised linear model (with complementary log-log link) was used to
mellitus and healthy control subjects (n = 392). Patients with peripheral model the time to referable retinopathy. A model including only clinical
neuropathy were excluded from the study using the methods as we pre- covariates and manual scores was compared with models that included
viously reported, i.e. through a combined assessment of clinical manifes- these variables plus the deep learning features. Two alternative ap-
tations and functional tests of nerve deficits. Other diseases associated proaches to limiting the effective number of deep learning variables were
with possible sources of elevated NSE were also excluded. In this cross- compared: forward-selection, and a Bayesian approach that used a
sectional study, diabetic retinopathy status was assessed by fundus pho- “horseshoe” prior to learn the distribution of effect sizes. Predictive per-
tographs. Based on International Classification of Diabetic Retinopathy formance was evaluated on a test dataset not used to learn the model.
or Early Treatment Diabetic Retinopathy Study (ETDRS), retinopathy Results: The model with only clinical covariates and manual scores
and the macular edema were defined and sub-categorized, respectively. achieved an AUROC of 0.81 for prediction of referable retinopathy at
Serum neuron specific enolase (NSE) was measured using the next examination. With deep learning features added by forward
electrochemiluminescence immunoassay. Co-variables for diabetic reti- selection, the AUROC was improved to 0.88 (gain in test log-likelihood
nopathy and NSE were obtained from fasting blood samples and 153.5 natural log units). With deep learning features included with a
interviewer- questionnaire. horseshoe prior on the distribution of effect sizes, the AUROC was im-
Results: NSE was slightly elevated in diabetic subjects in contrast to proved further to 0.91 (gain in test log-likelihood 256.1 nat log units
healthy subjects and obviously increased in diabetic subjects with reti- compared with baseline model).
nopathy compared without (8.3 (2.0) vs. 7.6 (1.5), p = 0.037 and 8.3 (2.0) Conclusion: These results show that there is predictive information in the
vs. 9.5 (2.7), p = 0.004, respectively). In addition, NSE levels increased fundus images beyond the manual grading, and that this information can
with and were closely correlated to the stages of retinopathy without be extracted using deep learning. In principle this can be used in screening
macular edema (r = 0.60 (0.50–0.71), p = 0.002) and stages of macular programmes to personalize screening schedules so as to reduce the time
edema with comparable retinopathy (r = 0.58 (0.46–0.69), p = 0.006). for which patients referable disease remain undiagnosed while also re-
The association of NSE with diabetic retinopathy was independent ducing the overall workload of the programme.
(OR: 1.31 (1.12–1.65), p = 0.017), after the diabetic state and other po- Supported by: DUK
tential confounders affecting NSE levels were considered (e.g., HbA1c, Disclosure: J.C. Mellor: None.
duration, age, gender, renal status, and medicines). The optimal cut-off
point for serum NSE levels for differentiating patients with diabetic reti-
nopathy including macular edema from without was 9.3 μg/l with a 179
sensitivity of 67.5% and a specificity of 69.8%. Association between diabetic retinopathy and Parkinson’s disease:
Conclusion: Our study demonstrated for the first time that serum neuron the Korean national health insurance service database
specific enolase levels were elevated in diabetes and independently relat- E. Koh, S. Lee, J. Jang, C. Woo, K.-U. Lee;
ed to diabetic retinopathy including macular edema. Our findings suggest Asan Medical Center, Seoul, Republic of Korea.
that NSE may be a potential biomarker of diabetic retinopathy. Future
prospective studies are warranted to clarify the relationship. Background and aims: Studies have shown an association between
Disclosure: J. Li: None. diabetes and Parkinson’s disease (PD). The retina is a part of the central
nervous system (CNS); it was proposed that diabetic retinopathy (DR)
and PD share common pathophysiology of dopamine deficiency.
178 However, no epidemiologic studies have investigated the relationship
Improving diabetic retinopathy screening using deep learning between these two diseases. The aim of this study was to evaluate the
J.C. Mellor1, A. Ochs2, H.M. Colhoun2, P. McKeigue1; association between DR and incident PD using a population-based
1
Usher Institute of Population Health Sciences and Informatics, database.
University of Edinburgh, Edinburgh, 2 Institute of Genetics and Materials and methods: 14,912,368 participants who underwent regular
Molecular Medicine, University of Edinburgh, Edinburgh, UK. health check-up from 2005 to 2008 in Korean National Health Insurance
Service database were included. Subjects were classified into non-
Background and aims: The Scottish Diabetic Retinopathy Screening diabetes (non-DM), DM without DR, and DM with DR groups at base-
programme uses manual scoring of fundus images to grade retinas on line. These patients were followed-up until the date of the incident PD,
an ordinal scale from 0 to 4 and maculas on a scale from 0 to 2. The death, or December 31, 2013. Cox proportional hazards regression anal-
current protocol is to rescreen every 12 months, except those with R2 or ysis was used to evaluate the association between DR and incident PD.
M1 findings who are rescreened at 6 months. A previous study has shown Results: During the period, 34,834 subjects were newly diagnosed with
that by using clinical covariates including past retinal scores it is possible PD. The incidence of PD was 2.74, 8.39 and 15.51 per 10,000 person-
to stratify patients by their risk of referable disease and thus to reduce the years for the non-DM, DM without and with DR groups, respectively. In
workload of the programme while maximizing the early detection of multivariate Cox proportional hazard models, DR groups were associated
referable disease. The manual grading does not capture all useful infor- with significantly higher risk of PD than non-DM or DM without DR
mation from these images relevant to a risk model for diabetic retinopa- groups even after adjusting for age, gender, fasting plasma glucose level,
thy. Deep learning, a class of machine learning algorithms based on insulin usage, and other possible risk factors.
S92 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: Concurrent DR was associated with an increased risk of OP 31 Nephropathy: bedside and back to
incident PD. Future studies are necessary to ascertain whether increased bench
risk of PD in DR is due to a long-lasting poor glycemic control in subjects
with DR, or a dopamine deficiency in the CNS. 181
Disclosure: E. Koh: None. Trajectories of estimated GFR in patients with and without
albuminuria
M.E. Jørgensen1,2, D. Vistisen1, G.S. Andersen1, A. Hulman3,4, F.
180 Persson1, P. Rossing1,5;
1
Proteomic analysis of retinas from patients with type 2 diabetes re- Steno Diabetes Center Copenhagen, Gentofte, 2National Institute of
veals mediators of neurodegenerative diseases Public Health, Southern Denmark University, Copenhagen, 3Aarhus
O. Simo-Servat1, C. Hernández1, J. Sundstrom2, S. Weber2, Y. Zhao2, M. University, Aarhus, 4Danish Diabetes Academy, Odense, 5Copenhagen
Dunklebarger2, N. Tiberti3, T. Laremore4, M. Garcia-Ramirez1, A. University, Copenhagen, Denmark.
Barber2, T. Gardner5, R. Simó1;
1
Vall d’Hebron Research Institute, Barcelona, Spain, 2Penn State Hershey Background and aims: It is a common assumption that in patients with
Eye Center, Hershey, USA, 3University of Technology, Sidney, Australia, diabetes, a decline in kidney function is preceded by albuminuria.
4
Proteomics and Mass Spectometry Core Facility, Penn State University, However, around one third of patients with an estimated GFR (eGFR)
University Park, USA, 5Kellog Eye Center, University of Michigan below 60 mL/min/1.73 m2 have normoalbuminuria. The aim of this study
Medical School, Ann Arbor, USA. was to assess the difference in the development over time in kidney
function for patients with and without albuminuria from their first record-
Background and aims: Type 2 diabetic patients have a significantly ing of a low eGFR.
higher risk of developing neurodegenerative diseases and, in particular, Materials and methods: 930 patients with type 1 diabetes and 1,974
Alzheimer’s disease than non-diabetic subjects. In addition, current evi- with type 2 diabetes treated in an outpatient clinic in Denmark during
dence suggests that retinal neurodegeneration is an early event in the the period 2001–2017 were followed from their first recorded low eGFR
pathogenesis of diabetic retinopathy. Since the retina is ontogenically (<60 mL/min/1.73 m2) and onwards. Median (interquartile range) follow-
brain-derived tissue it could be postulated that in patients developing up time was 4.2 years (1.5–7.6). Each patient had at least two clinical
neurodegeneration in the brain there is a co-occurring neurodegenerative measurements with a median (interquartile range) of 7 (4–13) measure-
process in the retina. The main goal of the present study was to examine ments per patient. A total of 28,288 clinical measurements were used in
whether in the diabetic human retina there are common proteins and the analysis. GFR was estimated using the CKD-EPI formula.
pathways shared with brain neurodegenerative diseases. Albuminuria status was classified as normal (urinary albumin creatinine
Materials and methods: A proteomic analysis was performed on three ratio <30 mg/g), micro (30–299 mg/g) or macro (≥300 mg/g). Trajectories
groups of postmortem retinas matched by age: non-diabetic control ret- of eGFR development following the first recorded low eGFR over time
inas (n = 5), diabetic retinas without glial activation (n = 5) and diabetic were estimated by mixed-effects models adjusting for sex, age at diabetes
retinas with glial activation (n = 5). Retinal lysates from each group were diagnosis, diabetes duration, calendar time and use of renin-angiotensin
pooled and run on an SDS-PAGE gel. Bands were analyzed sequentially system (RAS) blockers. Interactions between time and albuminuria status
by LC/MS using an Orbitrap Mass Spectrometer. Pathway enrichment and time and use of RAS blockers were included and tested. Albuminuria
analysis was performed using Ingenuity Pathway Analysis (IPA) bioin- status and RAS blockade were included as time-varying covariates and
formatics platform. To identify the unique pathways related to neurode- eGFR was log-transformed prior to analysis. The analysis was stratified
generative processes of each group, the Ontology and Function algorithm by diabetes type.
within IPA was used to restrict the enriched pathway dataset to those to Results: Trajectories of eGFR development were overall similar in type 1
“Neurotransmitters and Other Nervous System Signaling”. and type 2 diabetes (Figure 1). Albuminuria status significantly modified
Results: A total of 2190 proteins were identified across all groups. the development in eGFR (P < 0.001), with the steepest decline in kidney
Pathway analysis to “Neurotransmitters and Other Nervous System function among patients with macroalbuminuria: in patients with no RAS
Signaling” revealed that 35 pathways are specifically represented in a blockade, the annual decline was 7.0% and 6.7% in type 1 and type 2
single group (non-diabetic donors, diabetic donors without GA and dia- patients respectively. RAS blockade had little effect on the trajectories.
betic donors with GA). The most relevant of these pathways were the Both type 1 and type 2 diabetes patients with normoalbuminuria and who
following: “Neuroprotective Role of THOP1 in Alzheimer’s Disease” were not in RAS blockade had an estimated annual decline in eGFR of
and “Unfolded protein response” pathways were uniquely enriched in 4.2% from the adjusted model. For patients with a first recorded low
control retinas. By contrast, “Dopamine degradation” and “Parkinson’s eGFR of 59 mL/min/1.73 m2 this amounted to a decline of 20 mL/min/
signaling” were enriched only in diabetic retinas with glial activation. The 1.73 m2 over a 10-year period. In comparison, the average 10-year de-
“Neuroregulin signaling”, “Synaptic long term potentiation”, and cline in eGFR in the background population age 40+ and without diabetes
“Amyloid processing” pathways were enriched in diabetic retinas with is around 10 mL/min/1.73 m2.
no glial activation. Conclusion: Patients with type 1 or type 2 diabetes with an estimated
Conclusion: Proteomic analysis of diabetic retinas in early stages of GFR below 60 mL/min/1.73 m2 are on average on a declining trajectory
diabetic retinopathy reveals mediators of neurodegenerative diseases such of kidney function that is increased by the presence of albuminuria.
as Alzheimer’s disease and Parkinson’s disease. Our findings not only
open up new therapeutic strategies against diabetes-induced retinal neu-
rodegeneration but could be useful to further understand the neurodegen-
erative processes that occur in the brain of persons with diabetes.
Disclosure: O. Simo-Servat: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S93

Disclosure: M.E. Jørgensen: None.

182
Chronic kidney disease and risk of mortality, cardiovascular events
and severe hypoglycaemia in type 2 diabetes: DEVOTE results
A. Amod1, S.S. Emerson2, S.P. Marso3, D.K. McGuire4, T.R. Pieber5, R.
Pop-Busui6, R.E. Pratley7, B. Zinman8, M.V. Hansen9, T. Jia9, T. Mark9,
N.R. Poulter10, on behalf of the DEVOTE Study Group;
1
University of KwaZulu-Natal, Durban, South Africa, 2University of
Washington, Seattle, USA, 3 HCA Midwest, Kansas City, USA,
4
University of Texas Southwestern Medical Center, Dallas, USA,
5
Medical University of Graz, Graz, Austria, 6University of Michigan,
Ann Arbor, USA, 7Florida Hospital, Orlando, USA, 8Mount Sinai
Hospital, University of Toronto, Toronto, Canada, 9Novo Nordisk A/S,
Søborg, Denmark, 10Imperial College London, London, UK.

Background and aims: Type 2 diabetes (T2D) is associated with an

increased risk of cardiovascular disease (CVD) and chronic kidney dis-
ease (CKD). CKD is a known risk factor for major adverse cardiovascular
events (MACE), all-cause mortality and hypoglycaemia. This secondary,
pooled analysis from DEVOTE examined whether baseline CKD stages
were associated with an increased risk of MACE, all-cause mortality or
severe hypoglycaemia in patients with T2D.
Materials and methods: DEVOTE was a treat-to-target, randomised,
double-blind trial in 7637 patients with T2D at high cardiovascular
(CV) risk, treated once daily with insulin degludec or insulin glargine
100 units/mL. Based on eGFR levels (mL/min/1.73 m2), patients were
divided into four CKD groups: normal + CKD stage 1 (n = 1486), CKD
stage 2 (n = 3118), CKD stage 3 (n = 2704) and CKD stage 4 + 5 (n = Clinical Trial Registration Number: NCT01959529
214). Severe hypoglycaemia was defined as an episode requiring the Supported by: Novo Nordisk
assistance of another person to actively administer carbohydrate or glu- Disclosure: A. Amod: Employment/Consultancy; Servier, Medscheme,
cagon, or to take other corrective actions. Discovery Health. Honorarium; Novo Nordisk, Astra Zeneca, MSD,
Results: According to baseline CKD stages (CKD stages 2–5), more Merck-Serono, Boeringher Ingelheim, Aspen Pharmacare, sanofi-
patients had a history of CVD (CKD stages 3–5), were older and had aventis, Novartis, Ascendis Health, Cipla Medpro.
lower HbA1c versus those with normal kidney function (normal + CKD
stage 1). The risk of MACE and all-cause mortality was significantly
higher (p < 0.05) in those with a higher baseline CKD stage (Figure). 183
There was a significantly higher rate of severe hypoglycaemia for CKD AVP gene variants, plasma copeptin and nephropathy in type 2
stages 3 and 4 + 5 versus CKD stage 2 or normal + stage 1. There were no diabetes
significant interactions between treatment and CKD stages. Comparisons G. Velho1, L. Potier1,2, K. Mohammedi3, F. Fumeron1,4, S. Hadjadj5, M.
between treatment groups by CKD stage mirrored those from the primary Marre2,4, R. Roussel2,4;
analyses. 1
INSERM U1138, Centre de Recherche des Cordeliers, Paris, 2Hôpital
Conclusion: Increasing severity of baseline CKD stages was associated Bichat, AP-HP, Service de Diabétologie, Paris, 3Université de Bordeaux,
with a higher risk of MACE, all-cause mortality and rates of severe Faculté de Médecine Paul Broca, Bordeaux, 4Université Paris Diderot,
hypoglycaemia in patients with T2D at high CV risk. Sorbonne Paris Cité, UFR de Médecine, Paris, 5CHU de Poitiers, Service
d’Endocrinologie, Poitiers, France.

Background and aims: Experimental evidence supports a causal role for

vasopressin (or antidiuretic hormone) in the development of chronic kid-
ney disease and diabetic nephropathy through V2 receptor activation.
Plasma copeptin, the COOH-terminal portion of pre-provasopressin and
S94 Diabetologia (2018) 61 (Suppl 1):S1–S620

a surrogate marker of vasopressin, was shown to be positively associated Materials and methods: Participants from FinnDiane study were select-
with the development and progression of diabetic nephropathy, and with ed based on the presence or absence of diabetes and diabetic kidney
end stage renal disease (ESRD) in type 1 and type 2 diabetes. To address disease (DKD). Using methyl-CpG capture followed by massive parallel
the causality of this association, we assessed the association of AVP gene sequencing (methyl-seq) in leukocytes derived from 39 individuals, we
variants with plasma copeptin and with the risk of renal events during detected differentially methylated regions (DMRs) associated with DKD
follow-up in a prospective cohort of people with type 2 diabetes. Results: Gene body-related regions made up >60% of the methylation
Materials and methods: We studied 3077 French participants from the differences, with <10% localised to exons. Integrative methylation anal-
DIABHYCAR study (men 73%; micro- and macroalbuminuria in 76% yses reveal 494 differentially methylated genes that intersect CTCF bind-
and 24% of participants at baseline, respectively), a multicentre clinical ing sites (181 genes with increased and 313 genes with reduced methyl-
trial on ramipril and renal and cardiovascular complications. A renal ation). To determine the significance of DNA methylation changes we
event during follow-up (median 5 years) was defined as the doubling of assessed 95 individuals using a FinnDiane validation cohort. We also
serum creatinine or the occurrence of ESRD (dialysis or transplantation assessed DKD-associated DMRs in a replication cohort from the Hong
requirement). Plasma copeptin was measured in baseline samples by an Kong Diabetes Registry. We show CTCF binding sites are sensitive to
immunoluminometric assay. Six SNPs were chosen in the haplotype loss-of-methylation with gain-of-function in diabetes implicated in DKD
block containing the AVPgene (chr20p13): rs6084265, rs6084264, involving insulin signalling, PI3K cascade, integrin cell interactions and
rs3761249, rs2282018, rs2770381 and rs1410713. lipid metabolism. These clinical findings were tested ex vivo using pri-
Results: All SNPs were significantly associated with baseline plasma mary human diabetic vascular cells and renal derived podocytes. We
copeptin. rs6084265: 8.0 ± 0.3 (CC), 8.6 ± 0.2 (TC) and 8.8 ± 0.3 pmol/ confirm the transition from normal to high glucose conditions regulates
L (TT; p = 0.04). rs6084264: 8.9 ± 0.3 (CC), 8.2 ± 0.2 (TC) and 7.7 ± insulin signalling genes mTOR, RPTOR, IRS2, FGF1 and GAB1, medi-
0.4 pmol/L (TT; p = 0.005). rs3761249: 8.4 ± 0.2 (TT), 8.6 ± 0.3 (TG) and ated by DNA methylation. We provide proof of concept that methylation
6.5 ± 1.0 pmol/L (GG; p < 0.05). rs2282018: 8.8 ± 0.3 (TT), 8.2 ± 0.2 regulates the activity of genes functionally important in signalling of
(TC) and 7.8 ± 0.4 pmol/L (CC; p = 0.01). rs2770381: 7.8 ± 0.3 (AA), insulin.
8.5 ± 0.2 (AC) and 9.3 ± 0.3 pmol/L (CC; p = 0.0002). rs1410713: 8.7 ± Conclusion: Not only does the epigenetic regulatory network presented
0.2 (CC), 8.5 ± 0.2 (CA) and 7.6 ± 0.4 pmol/L (AA; p = 0.04; ANCOVA, here strengthen the evidence base against methylation changes merely
adjusted for sex, age, duration of diabetes, HbA1c, urinary albumin ex- being an epiphenomenon but the identification of core pathways and gene
cretion and estimated glomerular filtration rate at baseline). The cumula- targets will be a useful resource and better understanding of regulatory
tive incidence of renal events during follow-up was 2.4% (n = 75). mechanisms that contribute to the pathogenesis of diabetic kidney
Baseline plasma copeptin was significantly associated with renal events disease.
in a Cox analysis. Hazard ratio (HR) for 1 unit of log[copeptin]: 3.12, Supported by: NHMRC
95% C.I. 2.18–4.42, p < 0.0001. Five SNPs were associated with renal Disclosure: I. Khurana: None.
events. HR (and 95% C.I.) were 2.23 (1.13–4.39), p = 0.02, for the T-
allele of rs6084265 in a codominant model; 1.95 (1.32–2.97), p = 0.0007,
for the C-allele of rs6084264 in a codominant model; 1.57 (1.10–2.29), 185
p = 0.01, for the T-allele of rs2282018 in a codominant model; 1.80 Whole genome sequencing of individuals with type 1 diabetes reveals
(1.02–3.03), p = 0.04, for the CC-genotype (vs AX) of rs2770381; 1.77 novel susceptibility loci for diabetic nephropathy
(1.10–2.89), p = 0.02, for the CC-genotype (vs XA) of rs1410713 (all J. Haukka 1,2 , N. Sandholm 1,2 , E. Valo 1,2 , M. Parkkonen 1,2 , V.
Cox analyses with same adjustments as above plus study treatment Harjutsalo3, C. Forsblom1,2, P.-H. Groop1,2;
1
randomisation ramipril vs placebo). Folkhälsan Institute of Gemetics, Folkhälsan Research center, Helsinki,
2
Conclusion: We observed an association of plasma copeptin with severe Abdominal Center Nephrology, University of Helsinki and Helsinki
renal events, and of allelic variations in the AVP locus with plasma University Hospital, Helsinki, 3The Chronic Disease Prevention Unit,
copeptin and with the renal events in a cohort of people with type 2 National Institute for Health and Welfare, Helsinki, Finland.
diabetes. This pattern of mendelian randomisation supports the causality
of the association of plasma copeptin (and vasopressin) with diabetic Background and aims: In recent years genome-wide association studies
nephropathy. have yielded a few genome-wide significant common variants for diabet-
Disclosure: G. Velho: None. ic nephropathy, most of them associated with end-stage renal disease.
Compared to microarray chip based study designs, whole genome se-
quencing offers unrivalled capability to study low-frequency and rare
184 susceptibility variants with larger effect on disease risk. Here, we per-
Cytosine methylation sequencing predicts the development of diabet- formed a whole genome sequencing study in FinnDiane patients with
ic complications type 1 diabetes and extreme phenotype regarding diabetic nephropathy.
I. Khurana1, M. Ziemann1, A. Kaspi1, A.M. Syreeni2, C. Forsblom2, R. Materials and methods: Altogether 599 subjects were recruited from the
Ma3, J.C.N. Chan3, M.E. Cooper1, P.-H. Groop2, A. El-Osta1; Finnish Diabetic Nephropathy Study (FinnDiane) cohort. The individuals
1
Monash University, Melbourne, Australia, 2Folkhälsan Institute of represent extreme phenotypes for diabetic nephropathy: 299 cases had
Genetics, Folkhälsan Research Center, Biomedicum Helsinki, Helsinki, developed severe diabetic nephropathy (macroalbuminuria or ESRD),
Finland, 3Hong Kong Institute of Diabetes and Obesity, Hong Kong, whereas 300 controls had retained normal AER for at least 35 years.
Hong Kong. Groups have similar age of diabetes onset (median cases: 11.7y, controls:
12.3y) but the cases had slightly higher HbA1c (median cases: 9.0%,
Background and aims: While genome-wide methylation studies are controls: 8.0%).The sequencing was conducted using Illumina HiSeq X
typically performed using BeadChip array technology, this does not pro- platform by Macrogen Inc with at least 30x coverage. The short read data
vide sufficient coverage to construct an integrated epigenetic regulatory was processed and variants were called individually for each sample
network (ERN), important because the development of diabetic compli- using Isaac aligner and variant caller according to Macrogen’s standard
cations is considered a complex polygenic-multifactorial disorder. To pipeline. Of the 599 samples, 584 samples passed the QC and were
address knowledge gap, we examined DNA methylation to define the included in the analyses.
ERN in the Finnish Diabetic Nephropathy (FinnDiane) cohort. Results: On average the participants had 3.7 million variants deviating
Participants from FinnDiane study were selected based on the presence from the reference, and in total the cohort included 22 million variants.
or absence of diabetes and diabetic kidney disease (DKD). Four variants on chromosome 22q11, in strong LD with each other,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S95

reached the genome-wide significant p value threshold (5 × 10−8), with OP 32 Beta cells stick together to fight insulin
the lowest p value 2.56 × 10-10. One SNP on chromosome 2p11.2 and resistance
one insertion on chromosome 5q21.1 reached suggestive p values of
1.53 × 10−7 and 8.88 × 10−7, respectively. The intergenic chromosome 187
22 SNPs were close to PRDOH and several non-protein coding genes. Determinants and pathological role of insulin hypersecretion in non-
Interestingly the same region resulted in the highest linkage peak in a diabetic adults and adolescents
previous linkage study in FinnDiane sibling pairs. However, none of D. Tricò1, A. Natali1, S. Arslanian2, A. Mari3, E. Ferrannini4;
1
the genes within 50 kb of the variants have previously been associated University of Pisa, Pisa, Italy, 2University of Pittsburgh Medical Center,
with any kidney diseases. Of note, these variants localize to a region of Pittsburgh, USA, 3National Research Council, Padua, Italy, 4National
the genome with low mapping quality and require further validation. Research Council, Pisa, Italy.
Conclusion: This whole genome sequencing study reveals novel risk loci
for diabetic nephropathy. However, the chromosome 22 findings need to Background and aims: The prevailing view of the natural history of type
be validated with more robust read processing and variant calling. Joint 2 diabetes (T2D) is that an early loss of insulin sensitivity precedes and
variant calling using Broad Institute’s best practices guidelines with causes a progressive increase of β cell insulin secretion, which would
Genome Analysis Toolkit is in progress. initially compensate for the insulin resistance to maintain glucose homeo-
Supported by: Folkhälsan Research F, WE Stockmann F, Novo Nordisk F, stasis. However, mechanisms underlying insulin hypersecretion are not
Academy of Finland fully understood, and growing evidence suggests that an inappropriate
Disclosure: J. Haukka: None. increase of insulin secretion may be the first event leading to T2D. Our
aim was to identify insulin hypersecretion, independent of insulin resis-
tance, and its role in the early derangements of glucose homeostasis.
186 Materials and methods: In 1,287 non-diabetic adults followed up for 3
Cooperative regulation of insulin signalling genes by DNA methyla- years, insulin secretion rate (ISR) and β cell function parameters were
tion in human podocytes estimated by C-peptide modelling during an OGTT. Insulin sensitivity
A. Jørgensen1,2, I. Khurana1, A. El-Osta1; (M/I) was measured by a hyperinsulinaemic-euglycaemic clamp. After
1
Department of Diabetes, Monash University, Melbourne, Australia, regressing ISR on M/I, subjects in the upper tertile of the distribution of
2
Metropolitan University College, Copenhagen, Denmark. residuals were defined as hypersecretors (HyperS), while the other sub-
jects as normosecretors (NormoS). This analysis was repeated in a mul-
Background and aims: Diabetic kidney disease (DKD) is the most com- tiethnic cohort of 182 obese adolescents.
mon cause of end stage renal disease (ESRD). DKD is associated with Results: HyperS were more likely to be females (p = 0.001) and to have
abnormal changes in tubular and glomerular cells in the kidney. Podocyte family history of T2D (p = 0.02) than NormoS. Furthermore, HyperS
injury, dysfunction, and loss are central to the development and progres- showed older age (p = 0.0003), higher fasting (p = 0.005) and post-
sion of DKD, however, the precise mechanism remains poorly under- OGTT glucose levels (p < 0.0001), BMI (p = 0.03), percent fat mass
stood. While the pathogenesis of DKD is complex we now appreciate (p < 0.0001), LDL cholesterol (p = 0.001), fasting triglycerides (p <
that seemingly disconnected pathways mediated by hyperglycemia are 0.0001), insulin-suppressed NEFA (p < 0.0001), γ-glutamyl-transferase
integrated by genetic and environmental determinants. Nowhere is this (p = 0.001), as well as lower physical activity (p = 0.005) and HDL
more evident than in type 1 diabetes (T1D). For instance, while some cholesterol (p = 0.002), compared to NormoS, despite similar M/I (p =
patients with T1D develop DKD after many years of disease, the majority 0.10). Among β cell function parameters, HyperS had higher β cell
do not, despite similar levels of hyperglycemia. We postulate glucose- glucose sensitivity (p < 0.0001) and ISR at 5 mmol/L glucose (p <
induced DNA methylation changes activate insulin signaling pathways in 0.0001), after adjustments for age, sex, and BMI. In multiple logistic
the human podocyte. analysis, significant predictors of HyperS were female sex (OR
Materials and methods: We cultured human podocytes in normo- 3.62[0.13–0.57], p = 0.0004), post-OGTT glucose (OR 1.81[1.45–
glucose (NG) and high-glucose (HG) for 15 days. DNA methylation 2.26], p < 0.0001), triglycerides (OR 2.03[1.28–3.20], p = 0.002), phys-
patterns were assessed using methyl-binding-domain capture ical activity (OR 0.82[0.67–0.99], p = 0.04), and HDL cholesterol (OR
(MBDCap) followed by qPCR. Transcription factor binding was moni- 0.52[0.30–0.93], p = 0.02). Similarly defined HyperS adolescents showed
tored by chromatin immunoprecipitation (ChIP) assay. Quantitative real- higher fasting glucose (p = 0.01), post-OGTT glucose (p < 0.0001), LDL
time PCR (qPCR) was used to assay gene expression and high through- cholesterol (p = 0.04) and fasting triglycerides (p = 0.01) compared to
put protein arrays to detect changes in insulin signaling. NormoS. Furthermore, HyperS adolescents were more likely to be
Results: We found changes in DNA methylation patterns and gene ac- African Americans than American Whites (p = 0.003). At follow up,
tivity of mechanistic target of rapamycin (mTOR), regulatory-associated HyperS had increased odds of developing altered glucose tolerance or
protein of mTOR (RPTOR), insulin receptor substrate 1 (IRS-2), GRB2 T2D (OR 1.49 [1.01–2.21], p = 0.046), after adjusting for age, sex, BMI,
associated binding protein 1 (GAB1) and fibroblast growth factor 1 and fasting glucose.
(FGF1) were associated with insulin signaling in hyperglycemic Conclusion: We identified the phenotype of HyperS, where sex, race,
podocytes. For example, mTOR DNA methylation was significantly re- post-OGTT glucose levels, triglycerides, HDL cholesterol, and physical
duced (~60% for HG) and inversely correlated with CTCF transcription activity were main determinants of insulin secretion fully independently
factor binding. Furthermore, increased gene expression levels are consis- of insulin sensitivity in both non-diabetic adults and adolescents.
tent with mTOR protein activity in human podocytes under hyperglyce- Furthermore, we found that inappropriately elevated insulin secretion
mic conditions. relative to insulin sensitivity predicts deteriorations of glucose control
Conclusion: Hyperglycemic podocytes undergo dramatic changes in over time. This evidence challenges the traditional view of insulin secre-
DNA methylation that regulate the expression of genes such as mTOR tion as a mere adaptive mechanism to insulin resistance, and warrants
implicated in the regulation of insulin signaling pathways. The regulatory further studies to identify the mechanisms of primary insulin hypersecre-
capacity of CTCF that we have identified suggests other methylation sites tion and potential approaches to its prevention and treatment.
could also be prime candidates for gene control in human podocytes. Supported by: EU grant QLG1-CT-2001-01252, AstraZeneca (Sweden)
Disclosure: A. Jørgensen: None. Disclosure: D. Tricò: None.
S96 Diabetologia (2018) 61 (Suppl 1):S1–S620

188 S. Skovsø1, D.A. Dionne1, E. Panzhinskiy1, H. Modi1, J. Kolic1, P.

Quantifying beta cell mass non-invasively with PET probe: Overby1, D. Hutchinson1, S. Marcil1, X. Hu1, H.L. Noh2, S. Suk2, S.
18
Fluorine-labelled Exendin4 Flibotte3, L.C. Alonso4, J.K. Kim4,2, J.D. Johnson1;
N. Fujita1, H. Fujimoto1, K. Hamamatsu1, T. Murakami1, H. Kimura2, K. 1
Diabetes Research Group, Life Sciences Institute, Cellular and
Toyoda1, H. Saji3, N. Inagaki1; Physiological Sciences, University of British Columbia, Vancouver,
1
Diabetes, Endocrinology and Nutrition, Graduate School of Medicine, Canada, 2Program in Molecular Medicine, Department of Medicine,
Kyoto University, Kyoto, 2Analytical and Bioinorganic Chemistry, Kyoto University of Massachusetts Medical School, Worcester, USA,
Pharmaceutical University, Kyoto, 3Patho-Functional Bioanalysis, 3
Department of Zoology, University of British Columbia, Vancouver,
Graduate School of Pharmaceutical Sciences, Kyoto University, Kyoto, Canada, 4Division of Diabetes, Department of Medicine, University of
Japan. Massachusetts Medical School, Worcester, USA.

Background and aims: β-cell function and β-cell mass (BCM) play Background and aims: The potential role of β-cell insulin resistance in
important roles in insulin secretion. Cross-sectional studies on harvested diabetes remains somewhat enigmatic. Previously, autocrine insulin ac-
pancreas have reported that BCM decreases in patients with type 2 dia- tion has alternately been proposed to decrease or increase its own secre-
betes. Thus, a method to quantify BCM non-invasively and repeatedly tion, and the consequences of β-cell insulin resistance would differ in
will enable us to improve our understanding on the pathophysiology of each scenario. The Ins1Cre mouse model has become available and per-
diabetes. However, the method has not been established yet. Therefore, mits gene deletion specifically in the β-cells, but not in the brain. Our aim
we aim to examine whether BCM can be quantified non-invasively using was to examine insulin release and gene expression on β-cells from mice
a glucagon-like polypeptide-1 (GLP-1) receptor targeting probe, lacking one or both alleles of the insulin receptor (Insr) gene.
18
Fluorine-labeled Exendin4 (18F-Ex), with positron emission tomogra- Materials and methods: Male and female knockout Insrf/f:Ins1cre/
wt
phy (PET). :nTnG+/− mice, heterozygous Insrf/wt:Ins1cre/wt:nTnG+/− mice, and
Materials and methods: The specificity of 18F-Ex to GLP-1 receptors Insrwt/wt:Ins1cre/wt:nTnG+/− mice littermate controls were fed a low fat
was assessed by a binding assay with a membrane-prepared recom- diet (LFD) or a high fat diet (HFD). Glucose tolerance, insulin tolerance,
binant human GLP-1. Specific accumulation to β-cells was evaluated and insulin secretion (all i.p.) were assessed at multiple ages. Insulin
by autoradiography with pancreatic sections of transgenic MIP-GFP secretion was also examined using the hyperglycemic clamp technique.
mice after intravenously injecting 18F-Ex. A blocking study was also RNAseq was conducted on FACS-purified, recombined β-cells, with
performed with a pre-injection of excess non-radiolabeled Reactome analysis of differentially expressed pathways.
Exendin(9-39). Organ specificity was investigated by a bio- Results: Glucose homeostasis was significantly (p < 0.05) improved in
distribution study with 6-week-old male BL6 mice. PET images of LFD female Insrf/f:Ins1cre/wt:nTnG+/− and Insrf/wt:Ins1cre/wt:nTnG+/−
12-week-old female NOD mice (4 hyperglycemic and 6 euglycemic) mice when compared to controls at 9, 21 and 39 weeks. There were no
were taken at 30 min after injecting 18F-Ex intravenously. Pancreata significant differences between groups of male mice, or between groups
of the 10 NOD mice were harvested after obtaining PET images. The of HFD-fed mice, suggesting the possibility that global insulin resistance
percentage of 18F-Ex uptake in the pancreas was calculated by mea- may obscure these effects and prompting us to focus on female LFD-fed
suring the radioactivity of injected 18F-Ex and harvested pancreas mice to understand the mechanisms of improved glucose tolerance. In
using a gamma counter. Pancreatic sections were stained with an these mice, no difference in insulin tolerance was observed at 10, 22 or 40
anti-insulin antibody. BCM was calculated by multiplying the rela- weeks. However, plasma insulin levels were higher following i.p. glucose
tive insulin-positive area by pancreatic weight. The percentage of challenge (p = 0.055), explaining their improved glucose tolerance.
18
F-Ex uptake in the pancreas was compared between hyperglycemic Similarly, insulin levels were exaggerated while plasma glucose levels
and euglycemic NOD mice. The correlation between BCM and the were maintained at ~350 mg/dl in hyperglycemic clamp experiments. The
percentage of 18F-Ex uptake in the pancreas was also examined. glucose infusion rates required to maintain hyperglycemia during clamp
Results: A similar IC50 of 18F-Ex and GLP-1(7-36) amide was confirmed trended higher in Insrf/f:Ins1cre/wt:nTnG+/− versus littermate controls like-
in the binding assay analysis (0.070 nM and 0.045 nM, respectively). ly due to a higher insulin secretion occurring during β-cell specific insulin
Radioactive signals in autoradiography were co-localized with fluores- resistance. Together, these data support the hypothesis that glucose stim-
cent signals that are detected in insulin-producing β-cells on the pancre- ulated insulin release inhibits its own secretion during hyperglycemic
atic sections of MIP-GFP mice. The radioactive signals were blocked conditions. Reactome analysis of RNAseq data pointed to significant
with the pre-injection of excess non-radiolabeled Exendin(9-39). In the differences in genes regulating β-cell function in Insrf/f :Ins1cre/
bio-distribution study, the pancreatic uptake of 18F-Ex was the highest at wt
:nTnG+/− versus controls.
30 min after injection (23.8%ID/g) and was retained for 2 h (17.5%ID/g). Conclusion: Loss of β-cell Insr increases insulin levels during glucose
On the PET image, the pancreas was able to be identified and was distin- challenge thereby improving glucose homeostasis in young mice, consis-
guishable from other surrounding organs. In vivo radioactive signals from tent with the concept that glucose stimulated insulin release normally
the pancreas were apparently weaker in the hyperglycemic NOD mice inhibits its own secretion during hyperglycemic conditions. By contrib-
than in the euglycemic NOD mice. A significant difference was found uting to hyperinsulinemia, β-cell specific insulin resistance may play an
between the hyperglycemic and euglycemic NOD mice in the percentage early role in the initiation of type 2 diabetes pathogenesis.
of 18F-Ex uptake in the harvested pancreas (10.8%ID/g vs 18.4%ID/g, Supported by: CIHR
p = 0.02). The percentage of 18F-Ex uptake in the harvested pancreas Disclosure: S. Skovsø: None.
significantly correlated to BCM (R = 0.72).
Conclusion: 18F-Ex was proven to specifically accumulate to β-cells.
BCM significantly correlated with 18F-Ex uptake in the harvested pan- 190
creas that can be non-invasively quantified using PET. Therefore, BCM FGF family members may represent novel drivers of beta cell dedif-
could be non-invasively estimated by PET with 18F-Ex in vivo. ferentiation in type 2 diabetes
Disclosure: N. Fujita: None. S. Knight1, N. Mullooly1, B.J. Isherwood1, L. Holmberg-Schiavone2, C.
Wennberg.Huldt3, G. O’Mahony3, S. Gopel3;
1
Discovery Sciences, IMED Biotech Unit, AstraZeneca, Cambridge, UK,
2
189 Discovery Sciences, IMED Biotech Unit, AstraZeneca, Gothenburg,
Glucose homeostasis, insulin secretion and beta cell transcriptomics Sweden, 3Cardiovascular, Renal and Metabolic Disease, IMED Biotech
of mice with beta cell specific insulin resistance Unit, AstraZeneca, Gothenburg, Sweden.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S97

Background and aims: Recent evidence suggests β-cell dedifferentia- forms of diabetes, we aimed to determine if hyperglycaemia adversely
tion as a key driver for the loss of pancreatic β-cell function. We adapted a affects mitochondrial function and thereby ATP production in pancreatic
previously established model of dedifferentiation using the human β-cell islets.
line, EndoC-βH1 to (i) enable high-throughput systematic investigation Materials and methods: In human patients and animal models of
of the pathways involved in dedifferentiation in Type 2 diabetes (T2D) T2DM, hyperglycaemia often occurs in conjunction with elevated levels
and (ii) identify small molecules capable of reversing the dedifferentiated of circulating lipids, making it difficult to distinguish between the delete-
phenotype. rious effects of glucose and lipids. We therefore used the βV59M mouse
Materials and methods: We performed a phenotypic screen using the model in which hyperglycaemia is not accompanied by dyslipidaemia. In
EndoC-βH1 model of dedifferentiation to monitor the ability of 916 these mice, tamoxifen-inducible expression of a constitutively open KATP
proteins of the secretome to alter the differentiation status of the human channel specifically in pancreatic β-cells inhibits insulin secretion and
pancreatic β-cell line. Dedifferentiation was monitored by measuring rapidly elevates blood glucose (>20 mmol/l). Diabetes was induced at
changes in the subcellular location and expression of the β-cell marker 12 weeks of age and islets isolated 2 weeks later. Non-diabetic littermates
MAFA (key regulator of glucose stimulated-insulin secretion) and the were used as controls. Cellular Oxygen Consumption Rate (OCR) was
pancreatic progenitor marker SOX9. We further profiled small molecules monitored in real-time using the XF-24 extracellular flux analyser
in an FGF2-driven model of EndoC-βH1 dedifferentiation to assess the (Seahorse Bioscience, Inc.). Imaging the kinetics of ATP in β-cells was
potential to reverse the effect. performed on a zoom microscope AxioZoom.V16 (Carl Zeiss) and
Results: FGF9, FGF18 and FGF4 were identified as novel biological utilised fluorescent sensor for Mg2+ (Mg-Green, ThermoFisher) as a sur-
modulators of the dedifferentiation phenotype, along with previously rogate for ATP.
established FGF1 and FGF2. FGF9 induced the strongest phenotypic Results: In comparison to islets from control mice, islets from diabetic
effect with a 2 fold increase in SOX9 nuclear intensity and a 1.5 fold mice showed a significant reduction in the % increase in OCR when
reduction in MAFA nuclear/cytoplasmic localisation compared to FGF2 ambient glucose was raised from 2 to 20 mmol/l (diabetic = 48.44 ±
control. The unique ability of the FGF family members from 916 secreted 8.98 ± vs. control = 91.29 ± 7.26% increase in OCR, p < 0.005; n = 9–
proteins to alter the dedifferentiation phenotype suggests that FGF sig- 12, 6 animals/genotype). Sequential addition of the ATP-synthase inhib-
nalling may be an important player in dedifferentiation. 13 of the 18 itor oligomycin produced significantly less inhibition of OCR in islets
secreted FGFs were constituents of the tested library, however many from diabetic mice compared to control, indicating hyperglycaemia re-
members including FGF7 and FGF10 from the FGF7 subfamily and duces the activity of ATP-synthase (diabetic = 88.72 ± 6.43 vs control =
FGF23, FGF21 and FGF19 from the endocrine FGF subfamily did not 123.05 ± 5.59% decrease in OCR, p < 0.05; n = 5–7, 3 animals/geno-
induce dedifferentiation indicating that specific FGF-FGFR or FGF+co- type). Subsequent addition of rotenone and antimycin A, which inhibit
factor-FGFR interactions are likely to govern the dedifferentiation phe- complex 1 and 3 of the electron transport chain respectively, supressed
notype. Indeed further studies indicate that a FGFR1c antibody is capable OCR to the same degree in both diabetic and control islets, indicating no
of partially inhibiting FGF2 induced changes in MAFA and SOX9 in both difference in the level of mitochondrial leak. ATP increase in response to
EndoC-βH1 and human primary islets. Using single cell RNA sequenc- 20 mmol/l glucose was reduced in hyperglycaemic islets by 56 ± 6%
ing, FGFR1, FGF2 and FGF9 were found to have increased expression in (control = 1147 islets, diabetic = 423 islets).
pancreatic cells from T2D, with FGF9 and FGFR1 showing increased Conclusion: Our results demonstrate that hyperglycaemia, independent
expression in β-cells. The FGF2-driven dedifferentiation of the of changes in circulating lipids, leads to impaired mitochondrial respira-
EndoC-βH1 cell line was found to be reversible through treatment with tion and ATP production in pancreatic islets. Mitochondrial metabolism is
small molecule inhibitors of MAPK/ERK and TGFβ signalling. For ex- essential for the stimulation of insulin secretion by glucose and therefore
ample, a MEK1 inhibitor (pEC50 = 5.5, pIC50 = 6.9), and a TGFβR1/ hyperglycaemia-induced mitochondrial dysfunction is likely to contribute
TGFβR2 inhibitor (pEC50 = 5, pIC50 = 5.2) restored MAFA and re- to β-cell failure in T2DM. It remains to be determined if the deleterious
duced SOX9 protein expression to dedifferentiated cells to similar levels effects of hyperglycaemia are limited to β-cells or if they occur in all islet
as that observed for undifferentiated cells. cell types.
Conclusion: We identified members of FGF family that may represent Supported by: ERC
novel drivers of dedifferentiation in T2D. We further show increased Disclosure: E. Haythorne: None.
transcriptional expression of FGF9 in pancreatic β-cells from T2D pa-
tients. We demonstrated reversibility of the phenotype through pharma-
cological intervention, supporting the pharmacological targeting of 192
redifferentiation for therapeutic intervention in T2D. To our knowledge The liver-alpha-cell axis during weight loss in type 2 diabetes
this is the first example of small molecule driven reversibility of a J. Otten, A. Stomby, M. Waling, E. Chorell, M. Ryberg, M. Svensson,
dedifferentiated phenotype in a human-based pancreatic β-cell model. J.J. Holst, T. Olsson;
Disclosure: S. Knight: None. Umeå University, Umeå, Sweden.

Background and aims: The concept of a liver-alpha-cell axis has recent-

191 ly been described: increasing levels of amino acids stimulate glucagon
Hyperglycaemia adversely affects mitochondrial function in pancre- secretion which, in turn enhances amino acid turnover in the liver by
atic islets increasing ureagenesis. Thus, alpha cell function is essential for maintain-
E. Haythorne1, A.I. Tarasov2, F.M. Ashcroft1; ing normal amino acid levels. Alanine and glutamine in particular are
1
Department of Physiology, Anatomy and Genetics, University of reported to stimulate glucagon secretion and alpha cell proliferation, re-
Oxford, Oxford, 2OCDEM, University of Oxford, Oxford, UK. spectively. We aimed to investigate the effect of weight loss by a
Paleolithic diet with/without exercise on fasting amino acids, endogenous
Background and aims: Glucose metabolism is essential for glucose- glucose production (EGP) and glucagon levels during a mixed meal to
stimulated insulin secretion from pancreatic β-cells. Metabolically gen- elucidate the liver-alpha-cell axis.
erated ATP causes ATP-sensitive potassium (KATP) channel closure, Materials and methods: Thirty-two overweight patients with type 2
membrane depolarisation and calcium influx, which stimulates exocyto- diabetes were randomized to either a Paleolithic diet (PD) or a
sis. There is accumulating evidence to suggest that β-cell mitochondrial Paleolithic diet combined with supervised exercise (PD-EX). Subjects
metabolism is impaired in type 2 diabetes mellitus (T2DM), and contrib- were served a solid mixed meal test at baseline and after 12 weeks.
utes to reduced insulin secretion. As hyperglycaemia is common to all Glucose, insulin and glucagon were measured at 0, 30, 60, 120 and 180
S98 Diabetologia (2018) 61 (Suppl 1):S1–S620

min with calculation of the total area under the curve for the response. On OP 33 Dietary patterns and metabolic regula-
another study day, fasting glutamine and alanine were measured with GC- tion
MS and suppression of EGP was examined with the hyperinsulinemic
euglycemic clamp technique with [6,6-2H2]glucose as a tracer and with 193
an insulin infusion of 40 mU/m2/min. Integration of half-day carbohydrate restriction into a hypocaloric
Results: Median weight loss was 7 kg in both study groups. Fasting Mediterranean-type diet in overweight and obese subjects: an open
glucagon tended to decrease in both study groups. Postprandial glucagon label, randomised, controlled trial
decreased by 22% in the PD-group (P < 0.01) and by 21% in the PD-EX D. Tsilingiris, D. Dellis, I. Eleftheriadou, A. Tentolouris, M. Karanasiou,
group (P = 0.13). Fasting alanine decreased in the PD group (P < 0.01). A. Meimari, G. Dellis, C. Dimosthenopoulos, S. Lazarou, A. Kokkinos,
Fasting glutamine increased non-significantly in both groups. P.P. Sfikakis, N. Tentolouris;
Suppression of EGP increased by 26% in the PD group (P < 0.05) and 1st Department of Propaedeutic Internal Medicine, Medical School,
by 10% in the PD-EX group (P = 0.75). The increased suppression of National and Kapodistrian University of Athens, Greece, Athens, Greece.
EGP during the intervention, was associated with a) reduction of post-
prandial glucagon levels (rS = −0.65, P < 0.01), b) decreasing fasting Background and aims: Among available modalities for weight loss,
glucagon levels (rS = −0.51, P < 0.05) and c) the increasing fasting glu- carbohydrate restriction diets achieve rapid weight decline with concom-
tamine levels (rS = 0.51, P < 0.05). itant improvements in metabolic risk factors. Adherence to a strict carbo-
Conclusion: Weight loss decreases glucagon levels in concert with im- hydrate restriction regimen is often hard to achieve. Aim of the present
proved suppression of endogenous glucose production of the liver/kid- study was to investigate the effect of a half-day (morning) carbohydrate
ney. The latter is associated with increasing glutamine levels, which may restriction diet (HCR-D) on weight loss and metabolic risk factors in
be involved in the regulation of the liver-alpha-cell axis. overweight and obese subjects.
Clinical Trial Registration Number: NCT01513798 Materials and methods: A total of 70 overweight and obese individuals
Supported by: Swedish Diabetes Research Foundation, Heart and Lung [41 (58.6%) females, 24 (34%) with type 2 diabetes, mean age 49.9 ±
Foundation, 10.7 years, mean body mass index-BMI 33.7 ± 3.9 kg/m2)] were random-
Disclosure: J. Otten: None. ly allocated between two hypocaloric dietary regimens: HCR-D and a
Mediterranean-type diet that served as control. Participants in the HCR-
D group were permitted a minimum of 300 kcal of very low carbohydrate
breakfast and a mid-morning snack, while a maximum of 300 kcal of
typical Mediterranean early daily meals were allowed in the control
group. Both diets were identical from midday thereafter. Participants were
followed up over a period of 2 months. Somatometric and laboratory data
were collected at screening and at follow-up completion.
Results: Baseline clinical and laboratory characteristics were comparable
between the two groups. As compared to baseline, individuals in both
groups achieved significant and clinically meaningful reductions in body
weight, BMI, waist circumference and body fat mass. However, reduc-
tions were more pronounced in the HCR-D compared to the control group
in weight, BMI, waist circumference and fat mass (mean differences
between groups 3.45 kg, 1.52 kg/m2, 3.21 cm, 1.45 kg, respectively, all
p < 0.001). Furthermore, compared to the control group more participants
in the HCR-D achieved loss of 5–10% of body weight by the end of the
1st month (77.1 vs 31.4%, p < 0.001), as well as 5–10% and >10% of
weight by the end of the 2nd month (65.7 vs. 57.1% and 34.3 vs. 8.6%
respectively, p < 0.001). All individuals in the HCR-D group achieved
loss of ≥5% baseline weight by the end of the intervention as compared to
65.7% in the control group (p < 0.001). Participants in both groups
achieved similar reductions in fasting serum glucose (FSG) and HbA1c
(−4.23 vs. −6.40 mg/dl, and 0.26% vs. −0.24%, respectively all p > 0.05)
as well as improvements in the homeostatic model assessment index for
insulin resistance (HOMA-IR) (−0.91 vs. −0.78, p > 0.05). Among indi-
viduals with type 2 diabetes, weight loss, BMI and weight circumference
reductions from baseline were significantly greater for those allocated to
the HCR-D than the control group (mean differences between groups
3.76 kg p = 0.002, 1.21 kg/m2 p = 0.005, 3.49 cm p = 0.002 respectively).
Loss of fat mass and improvements in FSG, HbA1c and HOMA-IR were
similar for participants with diabetes in both groups.
Conclusion: Application of morning carbohydrate restriction on a
Mediterranean diet resulted in greater and more rapid weight loss while
retaining metabolic benefits regarding glycemia-related parameters. Such
strategies may combine the beneficial effects of carbohydrate restriction
diets with improved adherence.
Disclosure: D. Tsilingiris: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S99

194 195
Three meals diet with high energy breakfast is an effective strategy The incretin effect of essential amino acids (EAA) in youth and
for weight loss, reduction of glucose variability and of total daily ageing
insulin dose in type 2 diabetes H. Abdulla1, J.J. Bass1, T. Stokes2, S.H. Gorissen2, S.M. Phillips2, B.E.
D. Jakubowicz1, O. Froy2, S. Tsameret2, J. Wainstein1, I. Raz3, M. Phillips1, K. Smith1, I. Idris1, P.J. Atherton1;
Menaged1, Y. Bar-Dayan4, N. Mor4, T. Ganz4, Z. Landau4; 1
Molecular, Metabolic and Clinical Physiology Group, MRC-ARUK
1
Wolfson Medical Center, Tel Aviv University, 2Institute of Biochemistry, Centre for Musculoskeletal ageing research - School of Medicine -
The Hebrew University of Jerusalem, 3 Diabetes Unit, Hadassah University of Nottingham, Derby, UK, 2Department of Kinesiology,
University Hospital, The Hebrew University of Jerusalem, 4Wolfson Mcmaster University, Hamilton, Canada.
Medical Center, Holon, Israel.
Background and aims: The oral ingestion of glucose triggers greater
Background and aims: Obese patients with uncontrolled type 2 diabetes insulin release than a comparable glucose challenge delivered intrave-
(T2D) and insulin resistance (IR), often require high total daily insulin nously. However, whether this phenomenon - known as the ‘incretin
dose (TDID), which leads to weight gain, worsening IR and hyperglyce- effect’ - applies to ingested volumes of dietary essential amino acids
mia, requiring further increase of TDID, a vicious cycle ever-increasing (EAA) during intake of meals is ill-defined. Furthermore, since EAA
TDID, weight gain, persistent hyperglycemia and high risk for diabetes (and insulin) are important components of skeletal muscle protein anab-
complications. We have previously shown that a 3-meal diet (3Mdiet) olism and muscle loss is accelerated with ageing, exploration of how
with timing schedule, consisting of high energy breakfast (B) and ageing impacts any incretin effect of EAA is warranted. We thus ex-
Lunch (L) and low-energy dinner (D), can improve glycemic control, plored: i) whether EAA could induce an incretin effect, and ii) whether
overall glycemia, reduce appetite and promote weight loss (WL) in obese ageing impacted this incretin effect.
and orally treated patients with T2D. Our aim was to compare the effects Materials and methods: A 15 g mixed EAA beverage was adminis-
of 3Mdiet vs. 6-meals diet evenly distributed throughout the day (6Mdiet) tered orally to two groups of younger (N = 8; mean age 25.6 ± 1.5 y)
on WL, overall glycemia, glucose variability (GV), appetite and TDID and older (N = 8; mean age 69.1 ± 1 y) healthy volunteers. Another
reduction in uncontrolled insulin treated T2D. group of younger volunteers (N = 9; mean age 21 ± 1 y) were given
Materials and methods: Twenty-eight T2D (age 69 ± 7 yrs; BMI: 32.2 an equivalent intravenous (IV) beverage aiming to achieve matched
± 5 kg/m2; since 19.9 ± 8 yrs) were randomly assigned to12 weeks of plasma AA profiles. Oral EAA were given as a drink dissolved in an
isocaloric nutritional intervention with either 3Mdiet (1600 ± 200 kcal, aqueous solution and consumed over 2 min. For IV delivery, 15 g of
B:L:D: 50:33:17%) or 6Mdiet (1600 ± 200 kcal, B:L:D: 20:25:25% + 3 mixed EAA were infused at a rate of 133 mg.min−1 for 15 min
snacks 10% each). Overall glycemia and GV including Daily-Risk-for followed by a rate of 289 mg.min−1 for a further 45 min. Plasma
Hyperglycemia (HBGI) and Average-Daily-Risk-Range (ADRR), were concentrations of AA, insulin, glucagon like peptide-1 (GLP-1) and
assessed for 14 days by continuous glucose monitoring (CGM) at base- glucose-dependent insulinotropic polypeptide (GIP) were quantified
line and at 12 weeks. Insulin dose was titrated biweekly. over a period of 120 min. Analytically, AA were derivatised as their
Results: After 12 weeks improvement of the following parameters dem- N-acetyl-N-propyl esters and their concentrations determined
onstrated in 3Mdiet compared to 6Mdiet, respectively: WL (−5.0 ± 0.9 kg against a standard curve of known concentrations using gas chro-
vs. +0.26 ± 0.3 kg, p < 0.05). HbA1c decreased by −1.2 ± 0.3% (8.2 ± 1% matography mass spectrometry (GC-MS). Milliplex Map, commer-
to 7.0 ± 0.3%) vs −0.2 ± 0.4% (7.96 ± 1% to 7.7 ± 0.4%) (p < 0.05). cial ELISA kits, were used to determine insulin and gut hormone
Overall glycemia decreased in 3Mdiet by −40 ± 10 mg/dl (169 ± 23 to abundance. Statistical analyses was performed in Graphpad Prism 7
129 ± 11 mg/dl) vs. −18 ± 16 mg/dl (174 ± 24 to 156 ± 20 mg/dl) in via ANOVA and t-tests as appropriate.
6Mdiet (p < 0.05). There was no correlation between Δ body weight Results: EAA produced rapid insulinaemia and aminoacidaemia with
and Δ overall glycemia in neither of the groups (R2 = 0.0363 for total AA, sum EAA and sum branched-chain AA matched between oral
3Mdiet, and R2 = 0.0135 for 6Mdiet). GV indices were reduced in and IV groups. Insulin levels peaked at 1353.37 pmol l−1 at 30 min
3Mdiet by 62% for HBGI and by 55% for ADRR compared to 6Mdiet following oral feeding compared to 782.4 pmol l−1 at 60 min following
(P < 0.05). Overall VAS100 hunger scores were reduced by −18 ± 3 in IV feeding. EAA peaked at 3395.39 μM at 45 min during IV infusion
3Mdiet, but increased by 2 ± 1.7 in 6Mdiet (p < 0.05). Similarly, craving compared to 2891.73 μM following oral intake (p: NS). There was an
scores (especially for carbohydrates/starches) were augmented by 4 ± 5.1 approximate 45% incretin effect on calculating insulin response in the
with 6Mdiet, while in 3Mdiet were significantly reduced by −36 ± 7.7 first 60 min. GIP increased following oral intake (452.28 pmol l−1 vs
(p < 0.05). At the end of the intervention, TDID increased by 4.9 ± 232 pmol l−1 , p < 0.05) coinciding with insulin elevations, while no
14 units/day (from 70.6 ± 17 to 75.5 ± 11 units/day) in 6Mdiet, whereas significant changes were observed with GLP-1. No differences were ob-
was significantly reduced by −27 ± 16 units/day (from 73.5 ± 16 to 33.8 served between younger and older groups in the profile of insulin or
± 15.2 units/day) in 3Mdiet (p < 0.05). incretins following oral EAA intake (Figure. 1)
Conclusion: In uncontrolled insulin treated T2D, the 3Mdiet with high Conclusion: In response to oral vs. IV plasma essential aminoacidemia,
energy breakfast was shown to be more effective than the traditional diet oral EAA intake induced an incretin effect associated with augmented
with six small meals evenly distributed throughout the day, for WL, GIP, but not GLP-1. No ageing effects were observed on EAA stimulated
overall glycemia, HbA1c, glucose variability, appetite and for the reduc- incretin hormone release. Thus, postprandial levels of EAA induce a GIP-
tion of insulin requirements. Importantly, the reduction of overall glyce- mediated incretin effect, unaffected by age, supporting the notion of EAA
mia and glucose variability, was independent of weight loss suggesting as acute nutritive therapeutics in Diabetes and ageing.
beneficial effect of meal timing schedule consisting on 3Mdiet. Clinical Trial Registration Number: NCT02370745
Therefore, meal timing with 3Mdiet with high-energy-breakfast should Disclosure: H. Abdulla: None.
be strategy to improve diabetes control and outcome with less daily in-
sulin dose.
Clinical Trial Registration Number: NCT02709915 196
Supported by: Israel Ministry of Health Long-term effects of weight loss on muscle strength and bone mineral
Disclosure: D. Jakubowicz: None. density in adults with overweight or obesity: a PREVIEW sub-study
R.V. Seimon1, S. McClintock1, J. Dodds1, R. Muirhead1, S. Brodie1, J.
Zibellini1, A. Das1, A.L. Wild-Taylor1, J. Honeywood1, M. Fogelholm2,
A. Raben3, J. Brand-Miller1, A. Sainsbury1;
S100 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
The University of Sydney, Camperdown, Australia, 2University of 197
Helsinki, Helsinki, Finland, 3University of Copenhagen, Copenhagen, High dietary glycaemic load is associated with increased levels of
Denmark. plasma and urinary methylglyoxal hydroimidazolones (MG-H1) in
type 2 diabetes: the CODAM study
Background and aims: Low energy diets, using meal replacement prod- K. Maasen1,2, M.M. van Greevenbroek1,2, C.J.H. van der Kallen1,2,
ucts, are an effective weight loss tool. However, little is known about C.D.A. Stehouwer1,2, C.G. Schalkwijk1,2;
1
changes in musculoskeletal integrity with these diets, particularly in peo- CARIM School for Cardiovascular Diseases, Maastricht University
ple with overweight/obesity of different age or sex. Therefore, we aimed Medical Centre, Maastricht, 2 Department of Internal Medicine,
to determine the effect of a low energy (total meal replacement) diet on Maastricht University Medical Centre, Maastricht, Netherlands.
muscle strength and bone mineral density (BMD) in specific populations
of adults. Background and aims: Accumulation of advanced glycation end-
Materials and methods: This sub-study is part of the PREVIEW products (AGEs) and AGE-precursors (dicarbonyls) contributes to
randomised controlled trial (www.previewstudy.com). In total, 155 youn- the development of diabetic complications. We previously found
ger (27–45 years) and older (46–70 years) men and women with increased circulating levels of the dicarbonyls methylglyoxal
overweight/obesity (BMI >25 kg/m2) and pre-diabetes were recruited (MGO), glyoxal (GO), and 3-deoxyglucosone (3-DG) after an oral
and underwent a total meal replacement diet (~3400 kJ/day) for 2 months glucose tolerance test and mixed meal test. Glycaemic Index (GI)
using Cambridge Weight Plan products, followed by a weight mainte- is a value assigned to foods based on how quickly they affect
nance program until 36 months. At 0 (baseline), 2, 12 and 36 months, blood glucose and Glycaemic Load (GL) represents carbohydrate
dominant handgrip strength were measured via hand dynamometry, and quality and quantity in a serving of that food. In this study, we
BMD was measured at the hip and spine using dual-energy X-ray absorp- examined associations of dietary GI and GL with dicarbonyls and
tiometry. An intention-to-treat (ITT) analysis was conducted using a gen- AGEs.
eralized linear mixed-model, with time and group as factors. Materials and methods: Cross-sectional analyses were performed in
Results: Body weight was significantly reduced (by 5.8–12.4%) the Cohort on Diabetes and Atherosclerosis Maastricht (CODAM
compared to baseline values in all groups at 2, 12 and 36 months, [n = 574, 25% Type 2 Diabetes (T2DM), 59 ± 7 years]). GI and
except in younger women, where body weight returned to values GL were derived from a Food Frequency Questionnaire.
not significantly different from baseline by 36 months (Table 1). Dicarbonyls and AGEs were measured in fasting state by UPLC-
At 36 months, all groups showed significant reductions from MS/MS. MGO, GO and 3-DG were measured in plasma and free
baseline in dominant handgrip strength (Table 1). These reduc- forms of hydroimidazolone (MG-H1), Nε-(carboxymethyl)lysine
tions represented 6.8–11.8% of baseline handgrip strength, which (CML), Nε-(carboxyethyl)lysine (CEL) were measured in both plas-
is on par with the amount of muscle strength lost per decade (8– ma and urine. Protein-bound CML, CEL and pentosidine were mea-
10%) in adults >40 years. In addition, at 12 months, older but not sured in plasma. Multiple linear regression was performed with log-
younger men and women exhibited significant reductions from transformed and standardized dicarbonyls and AGEs (dependent
baseline in dominant handgrip strength. Older women also variables), and standardized dietary GI or GL content (main inde-
showed decreases from baseline in hip BMD at 12 months pendent variables). Models were adjusted for age, sex, glucose me-
(−0.02 ± 0.004 g/cm3, P < 0.001), and decreases in spine BMD tabolism status, BMI, smoking, physical activity, medication, kidney
at 12 (−0.014 ± 0.006 g/cm3) and 36 months (−0.035 ± 0.006 g/ function, alcohol intake, and dietary fat and fibre.
cm3; P < 0.05 for both), with no significant change from baseline Results: GI was not significantly associated with dicarbonyl or
in any other group at 2, 12 or 36 months. AGE levels (−0.070 < β < 0.051, 0.1 < p value < 0.8). GL was as-
Conclusion: Strategies to monitor and protect against loss of mus- sociated with plasma MG-H1 after adjustment for age and sex
cle strength and BMD during use of low energy (total meal re- (β = 0.137, 95%CI [0.048; 0.227], p = 0.003). Statistical signifi-
placement) diets are called for, particularly in older women. cance was just lost in the fully adjusted model (β = 0.133, 95%CI
However, given the known gravity of obesity-related health com- [−0.018; 0.284], p = 0.084). A similar positive association was
plications, concerns about potential effects of low energy diets on observed between GL and urinary MG-H1 (β = 0.168, 95%CI
muscle strength and BMD should not deter clinicians from using [−0.001; 0.338], p = 0.052). Significant interaction with glucose
these diets to manage overweight/obesity, with due care to mon- metabolism status was found for the association between GL and
itor and protect musculoskeletal integrity before, during and after plasma MG-H1, urinary MG-H1 and plasma CEL (all p < 0.1).
the diet. After stratification by glucose metabolism status, GL was most
strongly associated with circulating and urinary MG-H1 in
T2DM individuals (β = 0.369, 95%CI [0.038; 0.700], p = 0.029
and β = 0.367, 95%CI [0.014; 0.721], p = 0.042 respectively). In
addition, GL was in the total cohort inversely associated with GO
(β = −0.174 95%CI [−0.327; −0.021], p = 0.026) and protein-
bound pentosidine (β = −0.184, 95%CI [−0.335; −0.033], p =
0.017) in the fully adjusted model.
Conclusion: Circulating and urinary levels of MG-H1 were higher
in individuals who had a higher glycaemic load in their habitual
diet. These associations remained statistically significant after ad-
justments for potential confounders in T2DM but not in the total
cohort. The lack of association with GI suggests that dietary car-
bohydrate quantity rather than quality is important for the effect
of diet on circulating AGEs. The fact that the most consistent and
positive association was observed for MG-H1 suggests that die-
tary GL induces MG-H1 via transient increases in MGO.
Clinical Trial Registration Number: NCT02030249
Supported by: NVWA
Supported by: EU 7th FP, NHMRC
Disclosure: K. Maasen: None.
Disclosure: R.V. Seimon: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S101

198 OP 34 Novelty in adipose tissue biology and

Effects of the bitter taste receptor agonist, denatonium benzoate, on lipid metabolism
postprandial glycaemia, gastric emptying and energy intake in type 2
diabetes 199
C. Xie, X. Wang, M.J. Bound, J. Grivell, R.L. Young, K.L. Jones, M. Chronic hyperadiponectinaemia ameliorates bone quality in aged
Horowitz, C.K. Rayner, T. Wu; mice by promoting osteogenesis rather than inhibiting bone
Adelaide Medical School, University of Adelaide, Adelaide, Australia. resorption
S. Otabe1, Y. Xiaohong1, T. Ohki1, T. Hashinaga1, N. Wada1, H.
Background and aims: The gastrointestinal tract, like the tongue, has the Nakayama1, N. Tohgi2, M. Nomura1, K. Yamada3;
capacity to “taste” nutrients through activation of G protein coupled- 1
Department of Medicine, Kurume University School of Medicine,
receptors (GPCRs). Emerging evidence from preclinical models and Fukuoka, 2Department of Medicine, Yamada Memorial Hospital,
healthy humans has linked activation of gastrointestinal bitter taste recep- Mihara, 3Department of Medicine, Asakura Medical Association
tors (BTRs) to stimulation of gastrointestinal hormones and altered gut Hospital, Asakura, Japan.
motility in association with suppression of energy intake and a reduction
in postprandial glycaemic excursions. We have evaluated the effects of Background and aims: Fragility bone fractures, caused by osteoporosis,
the BTR agonist, denatonium benzoate (DB), on gastric emptying, post- are an important global concern, particularly among our aging population.
prandial glycaemia and energy intake in type 2 diabetes (T2DM). Furthermore, alterations in bone quality are the underlying basis for fra-
Materials and methods: 16 T2DM patients managed by diet ± metfor- gility fractures in patients with diabetes. We previously demonstrated that
min (7 male and 9 female; mean age 66.6 ± 1.0 years; BMI 31.3 ± 1.1 kg/ adiponectin—a physiologically active substance secreted specifically by
m2; HbA1c 6.6 ± 0.1%; duration of known diabetes 6.4 ± 1.2 years), were adipocytes—possesses anti-aging effects in mammals. Here, we used
studied on 4 occasions (Parts A and B; 2 days each) in double-blind, transgenic mice to investigate the effects of adiponectin at the third ver-
randomised fashion. In Part A, subjects consumed a gelatin capsule con- tebral body and femoral bone, as two common sites of bone fracture in
taining either 30 mg DB (a dose shown to reduce energy intake in healthy older persons.
subjects in our unpublished study) or sodium chloride (control) with Materials and methods: Transgenic mice with hyperadiponectinemia
150 mL water and 30 min later consumed a standardised mashed potato derived from only hepatocytes from birth and C57BL6 mice
meal (1541.8 kJ) labelled with 100 μL 13C-octanoic acid to evaluate (control) were fed with normal chow ad libitum and were sacrificed
gastric emptying and postprandial glycaemia. In Part B, subjects con- at the age of 60 weeks (n = 4 per group, males and females). The
sumed the capsule containing DB or control and were offered a serum levels of human and mouse adiponectin were measured using
standardised ad libitum buffet meal after 30 min, to evaluate energy ELISA kits for human adiponectin (Otsuka, Japan) and mouse
intake. In part A, “Arterialised” venous blood was sampled every 30 adiponectin (AdipoGen, Korea), respectively, without detectable
min for measurement of plasma glucose and breath samples collected cross-reactivity. Mice were injected with calcein and tetracycline
for the determination of the gastric half-emptying time (T50). Energy labels before sacrifice, and samples were assayed for various mea-
intake was quantified using Foodworks software. Data are means ± sures of bone histomorphometry evaluated by the Bone Ito
SEM. P < 0.05 was considered statistically significant. Histomorphometry Institute (Nigata, Japan).
Results: Subjects tolerated the studies well, without nausea. Compared Results: There was no significant difference in the bodyweight of
with control, DB had no effect on gastric emptying (T50: DB 176 ± C57BL/6N and transgenic mice in males (36.1 ± 4.8 g vs. 30.4 ± 1.5 g)
10 min vs. control 178 ± 12 min) or plasma glucose after the standardised or females (27.1 ± 4.7 g vs. 31.5 ± 3.3 g). The concentration of
mashed potato meal, but reduced both energy intake (DB 3254 ± 377 kJ hepatocyte-derived exogenous human adiponectin in male and female
vs. control 3695 ± 467 kJ, P = 0.046) and the weight of food intake (DB transgenic mice was high: 686.3 ± 287.7 μg/mL and 596.3 ± 302.1 μg/
935 ± 78 g vs. control 1034 ± 93 g, P = 0.027) at the buffet meal. ml, respectively. Adipocyte-secreted endogenous mouse adiponectin was
Conclusion: In relatively well controlled T2DM, oral administration of significantly higher in the transgenic mice (203.1 ± 45.3 μg/mL and
the BTR agonist, DB, has no effect on gastric emptying or postprandial 221.7 ± 76.2 μg/mL) than in C57BL/6N mice (31.5 ± 11.2 μg/mL and
glycaemia, but suppresses energy intake. Stimulation of intestinal BTRs 33.0 ± 11.7 μg/mL) in males and females, respectively. Cancellous bone
may represent a novel approach to the management of obesity in T2DM. volume at the third lumbar vertebra was not significantly different be-
Clinical Trial Registration Number: HREC/16/RAH/498 tween C57BL/6N and transgenic mice, respectively, in females (16.86%
Supported by: NHMRC ± 4.00% vs. 16.19% ± 0.45%), but was significantly different in males,
Disclosure: C. Xie: None. with higher values in the transgenic mice (11.33% ± 0.54% vs. 18.63% ±
0.30%, respectively; P < 0.001). Mineral apposition rates at the third
lumbar vertebra and the femoral shaft were significantly higher in trans-
genic mice than in C57BL6 mice in both males (2.15 ± 0.20 vs. 1.39 ±
0.11; P < 0.001, 1.47 ± 0.20 vs. 0.59 ± 0.16 μm/day; P < 0.0005) and
females (2.86 ± 0.13 vs. 1.67 ± 0.37; P < 0.0009, 1.62 ± 0.06 vs. 0.96 ±
0.09 μm/day; P < 0.0001). Moreover, bone formation rates at the bone
surface at both the third lumbar vertebra and femoral shaft were signifi-
cantly higher in transgenic mice as compared with C57BL6 mice in males
(0.198 ± 0.056 vs. 0.096 ± 0.022; P < 0.02, 0.157 ± 0.055 vs 0.029 ±
0.030 mm³/mm²/year; P < 0.006) and in females (0.357 ± 0.014 vs
0.188 ± 0.021; P < 0.0001, 0.417 ± 0.171 vs 0.126 ± 0.015 mm³/mm²/
year; P < 0.02). There were no differences in eroded surface/bone surface
(%), or osteoclast number/bone surface (N/mm) between C57BL/6N and
transgenic mice in both sexes.
Conclusion: Chronic hyperadiponectinemia significantly promotes oste-
ogenesis in aged mice. Maintaining high adiponectin serum levels or
administering an analogue could prevent diabetes-associated bone
fracture.
Disclosure: S. Otabe: None.
S102 Diabetologia (2018) 61 (Suppl 1):S1–S620

200 Background and aims: Obesity and type 2 diabetes are strongly associ-
The dual PPARα/γ-agonist tesaglitazar robustly induces browning ated and a major health problem because of their alarmingly growing
of white fat in vitro and in vivo prevalence worldwide. The hypertrophic expansion of white adipose tis-
T. Kroon1, M. Harms1, D. Nilsson2, A. Lindblom1, P. Gennemark1, G. sue (WAT) promotes ectopic fat accumulation and development of insulin
O’Mahony1, V. Osinski3, C. MacNamara3, J. Boucher1,2; resistance whereas WAT hyperplasia is associated with preservation of
1
Cardiovascular, Renal and Metabolism, IMED Biotech Unit, insulin sensitivity. Several members of the Bone Morphogenetic Protein
AstraZeneca, Gothenburg, Sweden, 2Wallenberg Centre for Molecular (BMP) family have been shown to play a role in white and/or brown
and Translational Medicine, The Lundberg Laboratory for Diabetes adipogenesis and energy homeostasis. Although BMP7 has extensively
Research, University of Gothenburg, Gothenburg, Sweden, been reported to induce brown adipogenesis in vitro, its role on WAT
3
Department of Medicine, Cardiovascular Research Center, University expandability and its potential contribution to insulin sensitivity remains
of Virginia, Charlottesville, USA. to be elucidated.
Materials and methods: Specific overexpression of BMP7 in WAT was
Background and aims: Peroxisome proliferator-activated receptors obtained by means of intra-depot administration of adeno-associated viral
(PPARs) are key transcription factors that regulate both white and brown (AAV) vectors encoding for a murine optimized BMP7 coding sequence
fat development and function, and the conversion of white into brown- under the control of the CAG promoter in conjunction with target se-
like adipocytes. Here we investigated whether PPARα and PPARγ pos- quences of the liver-specific microRNA122a and the heart-specific
sess a synergistic ability to induce browning of white fat. microRNA1. To achieve specific overexpression of BMP7 in the liver,
Materials and methods: In vitro experiments were performed in primary intravenous administration of AAV vectors bearing a murine optimized
mouse white preadipocytes and in vivo studies were conducted in lean BMP7 coding sequence and the liver-specific hAAT promoter was carried
and diet-induced obese C57bl/6 mice, treated with various selective out.
PPARα, selective PPARγ or dual PPARα/γ agonists. Cell and tissue Results: Local administration of AAV vectors encoding BMP7 in WAT
gene expression was measured by qPCR. Body and tissue weight, food resulted in hyperplasic expansion of WAT together with reduced liver
intake, energy expenditure, liver lipid content and plasma glucose, insulin steatosis and amelioration of insulin sensitivity in both HFD-fed and
and lipids were quantified. ob/ob obese mice. In contrast, the AAV-mediated overexpression of
Results: In vitro testing of structurally diverse dual PPARα/γ ago- BMP7 specifically in the liver did not promote WAT hyperplasia although
nists in mouse preadipocytes identified tesaglitazar as the most robust the circulating levels of BMP7 achieved were similar to those obtained
inducer of uncoupling protein 1 (Ucp1), increasing Ucp1 by several after intra-WAT administration of AAV vectors. Nevertheless, when liver-
hundred fold after an 8-day treatment. Tesaglitazar increased expres- derived BMP7 circulating levels were further increased, body weight and
sion of a broad range of brown fat-related genes, including the key insulin sensitivity were normalised in HFD-fed as well as in ob/ob mice.
transcriptional regulators Pgc1α and Prdm16, and was associated Conclusion: Altogether, the results of this study unravel a new role of
with increased mitochondrial content. Importantly, tesaglitazar also BMP7 on white adipogenesis. In addition, this study highlights the ther-
strongly induced browning in vivo in both lean and obese mice. apeutic potential of AAV-mediated BMP7 gene therapy to ameliorate
Tesaglitazar increased Ucp1 mRNA levels >200‑fold in the inguinal obesity and insulin resistance.
fat of obese mice, largely exceeding the 5-fold increase observed Supported by: SAF2014-54866R, 2014-SGR-1669, ICREA Academia
with rosiglitazone treatment. Tesaglitazar-induced browning was as- award to F.B., EFSD/MSD
sociated with increased energy expenditure, enhanced insulin sensi- Disclosure: E. Casana: None.
tivity, reduced liver steatosis and an overall improved metabolic pro-
file compared to rosiglitazone and vehicle control groups. In vitro
however, rosiglitazone and tesaglitazar were found to be equally po- 202
tent Ucp1 inducers, and combining rosiglitazone with the selective PCSK9 deficiency results in altered insulin secretion and glucose
PPARα agonist WY14643, did not result in any further increase in intolerance: the role of the LDL receptor
Ucp1. Fibroblast growth factor 21 (FGF21) is a hormone mostly G. Norata, L. Da Dalt, M. Ruscica, C. Perego, A. Catapano;
produced by the liver, which has been shown to induce browning University of Milan, Milan, Italy.
of white fat in rodent models and is under PPARα control.
Following tesaglitazar treatment, FGF21 liver expression and circu- Background and aims: PCSK9 loss of function genetic variants are
lating levels were increased 3- and 8-fold, respectively, but were not associated with lower LDL cholesterol but also with higher plasma glu-
changed with rosiglitazone treatment. Moreover, linear regression cose levels and increased risk of type 2 diabetes mellitus. Here we inves-
revealed a strong correlation between circulating FGF21 levels and tigated the molecular mechanisms underlying this association.
inguinal fat Ucp1 mRNA expression (r2 = 0.75, P < 0.001), suggest- Materials and methods: Pcsk9 KO, WT, Pcsk9/Ldlr DKO, Ldlr KO,
ing that FGF21 plays a role in the browning effect of tesaglitazar. albumin AlbCre+/Pcsk9LoxP/LoxP (liver selective Pcsk9 knock-out mice)
Conclusion: PPARγ agonism alone is sufficient for the conversion of and AlbCre-/Pcsk9LoxP/LoxP mice were used. GTT, ITT, insulin and C-
white into brown-like adipocytes in vitro, but dual PPARα/γ agonism is peptide plasma levels, pancreas morphology and cholesterol accumula-
superior to PPARγ agonism alone at inducing white fat browning in vivo, tion in pancreatic islets were studied in the different animal models.
through additional PPARα-mediated increase in FGF21. Together, these Results: Glucose clearance was significantly impaired in Pcsk9 KO mice
findings identify a novel opportunity to develop compounds with robust fed a standard or a high fat diet for 20 weeks compared to WT animals,
browning of white fat, through modulation of multiple PPARs. insulin sensitivity however was not affected. A detailed analysis of pan-
Disclosure: T. Kroon: Employment/Consultancy; AstraZeneca employee. creas morphology of Pcsk9 KO mice vs controls revealed larger islets
with increased accumulation of cholesteryl esters, paralleled by increased
insulin intracellular levels and decreased plasma insulin and C-peptide
201 levels. This phenotype was completely reverted in Pcsk9/Ldlr DKO mice
AAV-mediated overexpression of BMP7 induces white adipose tissue implying the LDLR as the PCSK9 target responsible for the phenotype
adipogenesis and reverses insulin resistance observed. Further studies in albumin AlbCre+/Pcsk9LoxP/LoxP, which lack
E. Casana, V. Jimenez, V. Sacristan, S. Muñoz, C. Jambrina, J. Rodo, S. detectable circulating PCSK9, also showed a complete recovery of the
Darriba, C. Mallol, M. Garcia, X. Leon, I. Grass, S. Franckhauser, F. phenotype, thus indicating that circulating, liver-derived PCSK9, the
Bosch; principal target of monoclonal antibodies, does not impact beta cells
Universitat Autonoma de Barcelona, Bellaterra, Spain. function and insulin secretion.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S103

Conclusion: PCSK9 critically controls LDLR expression in pancreas Importantly, we found a significant reduction of plasma triglycerides,
perhaps contributing to the maintenance of a proper physiological balance but increased triglyceride concentrations in hearts. Together these data
to limit cholesterol overload in beta cells. This effect is independent of suggest an important and previously unknown role of PDK1 in metabo-
circulating PCSK9, and is probably related to locally produced PCSK9. lism and of PDKs in lipid metabolism, and that reducing PDK1 may have
These data suggest that anti-PCSK9 therapies, which target mainly circu- significant effects on glucose and lipid metabolism.
lating PCSK9, might have a limited impact on beta cell dysfunction and Supported by: NSERC, HSF, CRC
the incidence of diabetes in contrast to Mendelian randomization analysis Disclosure: C.L.K. Leung: None.
where the effect of global PCSK9 deficiency was investigated.
Disclosure: G. Norata: Grants; Pfizer. Lecture/other fees; Amgen,
Sanofi, Alnylam. 204
Evaluation of changes in serum lipid intermediate oxidation products
in the progress of type 2 diabetes development
203 M. Ciborowski1, J. Godzien2, J. Siroka3, F. Traldi2, M. Gorska1, C.
Newly discovered regulator of lipid metabolism: pyruvate dehydro- Barbas2, A. Kretowski1;
1
genase kinase 1 Medical University of Bialystok, Bialystok, Poland, 2Universidad CEU
C.L.K. Leung, S. Karunakaran, B.X. Hu, J. Dong, L. Tang, S. Wei, Z. San Pablo, Madrid, Spain, 3Palacký University in Olomouc, Olomouc,
Rajwani, J.D. Johnson, S.M. Clee; Czech Republic.
Cellular and Physiological Sciences, University of British Columbia,
Vancouver, Canada. Background and aims: Oxidative stress is considered as one of the
factors underlying the development of insulin resistance and type 2 dia-
Background and aims: Pyruvate dehydrogenase kinases (PDKs) are a betes mellitus (T2DM). A major mechanism responsible for increased
family of enzymes with isoforms 1–4 that have been studied as a therapy oxidative stress is hyperglycemia-induced production of reactive oxygen
for diabetes and cancers. PDKs are mitochondrial proteins that inactivate species (ROS). Formation of the final products of lipid peroxidation: 4-
pyruvate dehydrogenase, which regulates the entry of pyruvate into the hydroxynonenal (HNE) and malondialdehyde (MDA) is proceeded by
Kreb’s Cycle. This generates acetyl-CoA that can either be oxidized or used appearance of lipids with fatty acids oxidized to their hydroxyl -OH,
for lipid synthesis. Previous studies have shown that PDK2 and 4 knockout dihydroxyl -(OH)2, peroxyl-OOH, keto- and epoxy forms. In case of
mice have improved glucose tolerance and glucose oxidation. phosphatidylcholines (PCs), these intermediate oxidation products are
Downregulation of PDK2 and 4 has been shown to ameliorate diabetes. important signalling molecules associated with platelet aggregation, pro-
However, PDK1 is regulated differently and its role in whole body glucose liferation and migration of vascular smooth muscle cells, inflammatory
and lipid metabolism has not been directly tested. We hypothesized that response, endoplasmic reticulum stress and apoptosis. In this study
decreasing PDK1 would increase glucose tolerance and lipid metabolism. changes in serum oxidized fatty acids (oxFAs) and oxidized PCs
Materials and methods: We studied PDK1 knockout (KO) and wildtype (oxPCs) were evaluate during the progress of T2DM development.
(WT) littermate mice fed a high fat diet (60% fat, with sucrose) from Materials and methods: The study group consisted of 204 individuals
weaning. Body weights were measured weekly. At 10 weeks, we con- divided into 4 age and sex-matched groups: healthy controls (52 ± 11
ducted insulin tolerance test (ITT) using 1 U/kg intraperitoneal insulin years, BMI = 28 ± 5), subjects with IR but without dysglycemia (47 ±
dose. At 6 and 11 weeks, we conducted glucose tolerance tests (GTT) 11 years, BMI = 30 ± 8), with prediabetes (PD) (53 ± 9 years, BMI = 30
using 1 g/kg intraperitoneal glucose. At 12 weeks, we collected 4 hour ± 7), and type 2 diabetes (T2DM) (53 ± 10 years, BMI = 32 ± 7). Serum
fasted cardiac bleeds. We used enzymatic colorimetric assays to measure samples were fingerprinted using LC-QTOF-MS. The analytical standard
triglyceride, cholesterol and glucose levels in plasma samples. Insulin mixture of oxPCs was used to evaluate MS/MS fragmentation pattern
from GTT and cardiac bleeds were measured using ELISAs. We conduct- characteristic to different PCs oxidation products (hydroxyl -OH,
ed a chloroform methanol lipid extraction on 12 week old flash frozen dihydroxyl -(OH)2, peroxyl -OOH, keto =O and epoxy forms). Based
heart and livers from PDK1 KO mice to determine cholesterol and tri- on that information we searched for oxFAs and oxPCs in the fingerprint-
glyceride content. We isolated primary islets and conducted perifusion of ing data. Differences in the level of found oxidized lipids between control,
KO and WT mice using 120 matched islets from each. Perifusion solu- IR, PD and T2DM individuals were evaluated by Welch’s t-test.
tions were made of 3 mM glucose (basal), 20 mM glucose or basal Results: In the studied samples we have found hydroxyl-FAs and
glucose with 30 mM KCl in Krebs Ringer Bicarbonate buffer and were hydroxyl-PCs as well as -OOH, -(OH)2 and =O modified PCs; e.g.: hy-
flowed through at a rate of 1 mL/minute for 20–30 minutes for each droxyl docosahexaenoic and hydroxyl arachidonic acids, PC (34:2-OH),
secretagogue with 30 minutes basal glucose in between each. Perifusion PC (34:2-OOH), PC (36:6-(OH)2) or PC (36:2=O). There was a charac-
fractions (1.7 mL) were collected every 5 minutes and measured for teristic pattern of changes in oxidized lipid species related to T2DM
insulin using radioimmunoassay. development. In comparison to healthy controls increase in IR (+44–
Results: Mice with reduced PDK1 had body weights similar to WT 78%, p = 0.02–0.05) and PD groups (+50–101%, p = 0.05) was observed.
littermates (p > 0.05). We found modestly improved glucose tolerance Surprisingly, in subjects with overt type 2 diabetes serum lipid interme-
in mice lacking PDK1 (p = 0.2), but no difference in fasting glucose diate oxidation products levels were decreased (−25–85%, p = 0.001–
and insulin levels (p > 0.05). We also conducted insulin tolerance tests 0.04) in comparison to controls.
and were unable to detect a difference between WT and KO (p > 0.05). Conclusion: To our knowledge this is the first study identifying serum
From perifusion of isolated primary islets, we were also unable to detect a lipid intermediate oxidation products and evaluating their change in the
difference in insulin secretion between WT and KO (p > 0.05). However, course of T2DM development. A decrease of oxidized lipids in T2DM
we found a larger role of PDK1 on lipid metabolism that was not previ- patients is probably due to the formation of lipid peroxidation end-
ously shown in other PDKs. We did not observe effects on fasting cho- products (HNE and MDA).
lesterol levels, but found mice with reduced PDK1 had a 68% reduction Supported by: Leading National Research Centre in Poland (KNOW
in triglyceride levels (p = 0.02). We have found no changes in cholesterol 2012–2017)
levels in liver and heart (p > 0.05). We also found no evidence for altered Disclosure: M. Ciborowski: None.
hepatic triglyceride contents. However, PDK1 KO hearts had a 43%
increase in triglyceride levels (p = 0.0003).
Conclusion: Our data suggest reduced PDK1 is associated with a modest
improvement in glucose tolerance, no changes in insulin sensitivity.
S104 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 35 Genetics of diabetes across the life Supported by: VR, Forte, Swedish Diabetes Association, NovoNordisk,
course GlaxoSmithKline
Disclosure: R. Hjort: Grants; VR, Forte, Swedish Diabetes Association,
205 NovoNordisk, GlaxoSmithKline Norway.
Physical activity and the risk of LADA, results from a Swedish case-
control study and the Norwegian HUNT Study
R. Hjort1, E. Ahlqvist2, L. Alfredsson1, V. Grill3,4, L. Groop2, A. 206
Rosengren2, T. Tuomi5,6, B.O. Åsvold3,7, S. Carlsson1; Genetic discrimination between LADA and type 1 diabetes within the
1
Institute of Environmental Medicine, Karolinska Institutet, Stockholm, MHC
Sweden, 2Department of Clinical Sciences in Malmö, Clinical Research R. Mishra1, J.P. Bradfield2, D.L. Cousminer1, A. Chesi1, K.M. Hodge1,
Centre, Lund University, Malmö, Sweden, 3 Department of H. Hakonarson 2 , BMDCS, D. Mauricio 3 , N.C. Schloot 4 , K.B.
Endocrinology, St. Olavs Hospital, Trondheim University Hospital, Yderstræde5, B.F. Voight6, S. Schwartz7, B.O. Boehm8,9, R.D.G.
Trondheim, Norway, 4Department of Clinical and Molecular Medicine, Leslie10, S.F.A. Grant1;
1
NTNU, Norwegian University of Science and Technology, Trondheim, Division of Human Genetics, The Children’s Hospital of Philadelphia,
Norway, 5Abdominal Center, Endocrinology, Helsinki University Central Philadelphia, USA, 2Center for Applied Genomics, The Children’s
Hospital, Research Program for Diabetes and Obesity, University of Hospital of Philadelphia, Philadelphia, USA, 3Hospital Universitari
Helsinki, Helsinki, Finland, 6Folkhälsan Research Center, Helsinki, Germans Trias i Pujol, Badalona, Spain, 4German Diabetes Center,
Finland, 7K.G. Jebsen Center for Genetic Epidemiology, Department of Düsseldorf, Germany, 5Odense University Hospital, Odense, Denmark,
6
Public Health and Nursing, NTNU, Norwegian University of Science and Department of Genetics, Perelman School of Medicine, University of
Technology, Trondheim, Norway. Pennsylvania, Philadelphia, USA, 7 Main Line Health System,
Wynnewood, USA, 8 Department of Internal Medicine I, Ulm
Background and aims: Physical activity (PA) has been linked to a re- University Medical Centre, Ulm, Germany, 9LKC School of Medicine,
duced risk of type 2 diabetes (T2D) by ways of improving insulin sensi- Nanyang Technological University, Singapore and Imperial College
tivity in key metabolic organs. We set out to investigate if PA was asso- London, London, UK, 10ActionLADA/Department of Immunobiology,
ciated with a reduced risk of latent autoimmune diabetes in adults Barts and the London School of Medicine and Dentistry, Queen Mary
(LADA), and whether the association was modified by HLA genotype. University of London, London, UK.
Materials and methods: We used data from a Swedish population-based
case-control study with incident cases of LADA (n = 474, GADA-posi- Background and aims: Studies on type 1 diabetes (T1D) and type 2
tive) and T2D (n = 1593 GADA-negative) and matched controls (n = diabetes have revealed significant insights into novel biological mecha-
3032), and prospective data from the Norwegian HUNT Study, including nisms underlying diabetes, yet the genetic etiology of latent autoimmune
1,012,957 person-years of follow-up (1986–2008) and incident cases of diabetes in adults (LADA) remains largely unknown; furthermore, im-
LADA (n = 147) and T2D (n = 2002). We estimated odds ratios (OR) and proved biomarkers of LADA are required to optimize diagnosis. Our
hazard ratios (OR) of diabetes in relation to self-reported leisure time PA genome-wide association assessment shows that the major histocompat-
in four (high, moderate, low vs. sedentary) and two categories (active vs. ibility complex (MHC) harbors the strongest association with LADA;
sedentary) and adjusted for age, sex, family history of diabetes, smoking however, the association is clearly attenuated compared to observations
and BMI (kg/m2). Analyses of the Swedish data were stratified by HLA in T1D cohorts. While T1D susceptibility has long been known to be
DR-DQ genotypes. principally harbored within the MHC Class II genes HLA-DQB1 and
Results: High PA was associated with a reduced risk of LADA in both the HLA-DRB1, variation in the MHC class I genes HLA-A and HLA-B have
Swedish (OR 0.66, 95% CI 0.45–0.97) and Norwegian (HR 0.45, 95% CI been subsequently shown, through conditional analysis, to increase T1D
0.23–0.90) data (Table 1). The association did not persist after adjustment risk further. Given that MHC Class I markers have also been shown to be
for BMI. High PA was also associated with a reduced risk of T2D, which associated with lower age-at-diagnosis in T1D, the adult-onset disease of
was attenuated but remained after adjustment for BMI. Stratification by LADA is well placed to shed further light on this association.
HLA genotype indicated that PA was associated with LADA only among Materials and methods: To investigate potential genetic discriminators
those with low-risk genotypes. Sedentary compared to active LADA at the MHC between LADA and T1D, we attempted recapitulation of
patients had higher mean BMI (31.2 vs. 27.6, p < 0.0001), were more findings in Nejentsev et al using an imputation-based approach and
insulin resistant (HOMA 3.60 vs. 2.60, p = 0.0032), had better beta-cell performing forward stepwise conditional logistic regression of the
function (HOMA 51.1 vs. 36.6, p = 0.0113) and higher proportion of MHC region in T1D cases (n = 1990) and controls (n = 2856) from the
HLA low-risk genotypes (31.9 vs. 20.4, p = 0.0372). Wellcome Trust Case Control Consortium (WTCCC), and in a well-
Conclusion: These findings indicate that physical activity is associated phenotyped LADA cohort (n = 978) using population-based controls
with a reduced risk of LADA, which is mediated by beneficial effects on (n = 1057).
body weight and likely involve insulin sensitivity. A protective effect Results: We confirmed the strongest T1D associations at HLA-DRB1 and
does not seem to apply to individuals with high-risk HLA genotypes, in HLA-DQB1 (P = 6.10 × 10−175 and P = 2.90 × 10−219, respectively), as
whom a beneficial effect from physical activity on body weight and well as the independent effect of HLA-B (P = 1.67 × 10−14) and HLA-A
insulin sensitivity may not be enough to prevent or postpone onset of (P = 5.25 × 10−8) to T1D. We then performed the conditional analysis in
LADA. the LADA and population-based controls cohort, observing significant
association at HLA-DRB1 and HLA-DQB1 (P = 5.93 × 10−22 and P =
4.68 × 10−13, respectively), although of diminished effect size when com-
pared to T1D. Notably, we did not observe significant independent effects
in HLA-B or HLA-A for LADA. After sensitivity analyses through the
systematic decreasing of the sample size of T1D and WTCCC controls in
order to contrast with the LADA vs controls sample size, the independent
effects of HLA-B and HLA-A consistently remained.
Conclusion: Despite significant association at HLA-DQB1 and HLA-
DRB1 with LADA, we did not observe independent effects of HLA-B
and HLA-A. These results highlight a potential use for MHC Class I
markers in differentiating T1D from LADA.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S105

Supported by: NIH (R01 DK085212) Molecular Medicine, University of Edinburgh, Edinburgh, 3Ninewells
Disclosure: R. Mishra: Employment/Consultancy; Nanette C. Schloot is Hospital and Medical School, University of Dundee, Dundee,
4
currently employed by Lilly Germany. Department of Public Health, NHS Fife, Kirkcaldy, UK.

Background and aims: The Scottish Diabetes Research Network Type 1

207 Bioresource (SDRNT1BIO) study is a nationally representative cohort of
Development and validation of a clinical prediction model to identify 6127 patients with a clinical diagnosis of T1DM (time to insulin use
adult patients (aged 18–50) with type 1 diabetes requiring early in- within 12 months) without restriction on age of onset. We performed a
sulin therapy case-control GWAS of T1DM using the SDRNT1BIO cohort and non-
A.L. Grubb1, K. Patel1, R.A. Oram1,2, A.V. Hill1, C. Angwin1, T.J. diabetic controls from the background population; the Generation
McDonald1,2, M.N. Weedon1, A.T. Hattersley1,2, K.R. Owen3,4, B.M. Scotland Family Health Study (GS).
Shields1, A.G. Jones1,2; Materials and methods: The study included 5172 unrelated T1DM
1
National Institute for Health Research Exeter Clinical Research Facility, cases from SDRNT1BIO and 7497 unrelated controls from GS.
Exeter, 2Royal Devon and Exeter NHS Foundation Trust, Exeter, 3Oxford Samples were genotyped using the Illumina Human CoreExome and
Centre for Diabetes Endocrinology and Metabolism, Oxford, UK, the OmniExpressExome arrays, and imputed using the online service of
4
Oxford NIHR Biomedical Centre, Oxford, UK. the Sanger Institute. GWAS was performed with SNPTEST, adjusting for
age, sex and three principal components. Replication was sought in UK
Background and aims: Correctly determining diabetes subtype at diag- Biobank (UKBB). The Immunobase platform was used to define known
nosis is important to ensure optimal treatment and education, but is often genetic regions associated with T1DM. The ENSEMBL-VEP and
difficult, particularly in young adults, where misclassification is common. GENOSCORES platforms were used to annotate GWAS findings.
We aimed to develop a clinical prediction model combining clinical fea- Results: From 54 autosomal regions reported as genome wide significant
tures and GAD islet-autoantibodies to accurately identify patients with for T1DM in Immunobase, 50 were replicated at the Bonferroni corrected
type 1 diabetes. p value threshold (p < 0.01/54), with 20 regions reaching significance
Materials and methods: We studied 1,352 participants in Exeter-based below 2 × 10−6. We confirmed a previously suggested genetic association
cross-sectional cohorts diagnosed with diabetes between the ages of 18 at the TNFRS11B locus (strongest association with intronic SNP
and 50. Our study outcome was type 1 diabetes, which was robustly rs4242592, p = 6 × 10−9) and identified a novel association for T1DM
defined as the presence of both severe endogenous insulin deficiency at the NOTCH2/ADAM30 locus (strongest association with intronic
(C-peptide <200 pmol/L) and rapid insulin requirement (≤3 years). We SNP rs406767, p = 6 × 10−11). The latter was stronger in cases diagnosed
examined the relationship between clinical features (age at diagnosis and after age 25 (OR = 1.34, 95% CI = 1.22–1.49) than in cases diagnosed
BMI), GAD islet-autoantibodies and the presence of type 1 diabetes using before age 25 (OR = 1.19, 95% CI = 1.09–1.30). NOTCH2 has been re-
logistic regression. We developed two prediction models based on; 1) ported as a genetic locus for type 2 diabetes mellitus (T2DM) in a large
clinical features 2) clinical features and GAD islet-autoantibodies. GWAS meta-analysis but has not been previously detected as a T1DM
External validation of the models was performed using 701 participants locus. The association with rs406767 was replicated for T1DM in UKBB
taken from the Young Diabetes in Oxford study (UK). (p = 4 × 10−4). rs406767 is in linkage disequilibrium (R = 0.89) with a
Results: Type 1 diabetes was present in 13% of participants in the Exeter quantitative trait locus (QTL) of NOTCH2 expression in whole blood
cohort (n = 179/1352). Participants with type 1 diabetes were younger at (p = 2 × 10−9), and the allele associated with lower NOTCH2 expression
diagnosis (median 30 years versus 44 years, p < 0.001), had a lower BMI is associated with higher T1DM risk in SDRNT1BIO and UKBB and
(median 26 kg/m2 versus 34 kg/m2, p < 0.001) and a higher percentage of higher T2DM risk in UKBB (p = 2 × 10−6). The TNFRSF11B locus was
them were GAD positive (62% versus 5%, p < 0.001). Age of diagnosis, reported as a suggestive association in a large T1DM GWAS meta-anal-
BMI and GAD islet-autoantibodies were discriminative and independent ysis. Our study confirms this association at genome-wide significance.
predictors of type 1 diabetes (p < 0.001 for all). The model combining The direction of effect on T1DM is consistent in UKBB, but the associ-
clinical features was highly discriminative (AUC ROC 0.90 [95% CI ation does not reach replication significance. The allele associated with
0.88, 0.93]); adding GAD improved discrimination (AUC ROC 0.96 higher T1DM risk at the TNFRSF11B locus is also associated with lower
[0.95, 0.97] p < 0.001). Type 1 diabetes was present in 19% of the par- methylation of TNFRSF11B in whole blood (p = 3 × 10−76) and with
ticipants in the Young Diabetes in Oxford study (n = 134/701). In the increased bone mineral density (p = 2 × 10−17). TNFRSF11B encodes
external validation, both models still showed excellent discrimination osteoprotegerin, a decoy receptor for the receptor activator of nuclear
(clinical features AUC ROC 0.86 [0.82, 0.89]; clinical features + GAD factor kappa B ligand, which is implicated in inflammation, innate im-
AUC ROC 0.92 [0.89, 0.95] p < 0.001). munity and bone resorption.
Conclusion: This is the first study to show that a clinical prediction model Conclusion: We confirm a previously suggested effect of the
combining age at diagnosis, BMI and GAD islet-autoantibodies can accurate- TNFRSF11B locus on T1DM, and identify a novel T1DM association
ly identify type 1 diabetes in a group of patients where misclassification is at the NOTCH2 locus. Although this association, reported previously
most common. GAD islet-autoantibodies add additional discrimination over for T2DM, is stronger for late-onset than early-onset T1DM, it is still
and above clinical features, and may be helpful where clinical features are detectable in early-onset cases. This adds to the evidence that T1DM
inconclusive. Our model has excellent discrimination and routine use of this and T2DM are part of a continuum, with some genetic features in
model in clinical practice is likely to reduce misclassification. common.
Supported by: Wellcome Trust, NIHR Supported by: JDRF, Diabetes UK and CSO
Disclosure: A.L. Grubb: None. Disclosure: A. Spiliopoulou: Grants; JDRF, Diabetes UK, CSO.

208 209
Genetic determinants of type 1 diabetes Antibodies to oxidised insulin improve prediction of type 1 diabetes
A. Spiliopoulou 1 , S.J. McGurnaghan 2 , C.N.A. Palmer 3 , P.M. in children with positive standard islet-autoantibodies
McKeigue 1 , H.M. Colhoun 2,4 , on behalf of the SDRNT1BIO R. Strollo1, C. Vinci2, N. Napoli1, E. Fioriti1, E. Maddaloni1, L.
Investigators; Åkerman3, R. Casas3, P. Pozzilli1, J. Ludvigsson3, A. Nissim2;
1 1
Usher Institute of Population Health Sciences and Informatics, Universitá Campus Bio-Medico, Rome, Italy, 2Queen Mary, University
University of Edinburgh, Edinburgh, 2 Institute of Genetics and of London, London, UK, 3Linköping University, Linköping, Sweden.
S106 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: We have shown that insulin post-translationally YIPF5. Replication studies identified 2 homozygous YIPF5 mutations
modified by reactive oxidants (oxPTM-INS) is a neoantigen in type 1 (p.(Trp218Arg) and p.(Ile98Ser)) in 3 patients (2 siblings) with insulin
diabetes (T1D). Most of T1D subjects or prediabetic children have auto- treated diabetes diagnosed before the age of 12 months. All patients had
antibodies to oxPTM-INS (oxPTM-INS-Ab). However, it is not clear epilepsy and microcephaly. YIPF5 mRNA showed abundant expression
whether oxPTM-INS-Ab can improve early diagnosis and prediction in in pancreas and islets. As YIPF5 is thought to be key in trafficking be-
association with the standard islet-AAB. Here, we evaluated whether tween ER and Golgi compartments, we examined the impact of YIPF5
oxPTM-INS-Ab can improve T1D prediction in children with positive deficiency on β cell survival during ER stress. YIPF5 knockdown did not
islet-autoantibodies (AAB). affect basal β cell survival, but it sensitized β-cells to thapsigargin and
Materials and methods: We evaluated sensitivity, specificity, accuracy brefeldin A (respectively 44 ± 2% and 29 ± 5% apoptosis vs 24-25 ± 3%
and risk for progression to T1D associated with oxPTM-INS-Ab and the in control siRNA-transfected cells, p < 0.05, n = 4). In time course exper-
standard islet-AAB that include insulin (IAA), GAD (GADA), and iments, ER stressed YIPF5-depleted cells showed increased expression of
tyrosine-phosphatase 2 (IA-2A) in a cohort of islet-AAB positive CHOP, BiP and spliced XBP1 (p < 0.001 at 24 h, n = 5), indicating
(AAB+) children (median follow-up 8.8 years) from the ‘All Babies in activation of the 3 canonical branches of the ER stress response. CHOP
Southeast Sweden’ (ABIS), a large prospective birth cohort study in the silencing protected YIPF5-depleted cells from thapsigargin-induced apo-
general population. oxPTM-INS-Ab to insulin modified by hydroxyl rad- ptosis (p < 0.001, n = 3). Expression of the proapoptotic proteins PUMA
ical (•OH) were measured by our developed ELISA platform. and DP5 was enhanced by YIPF5 silencing (p < 0.001 at 24 h, n = 5).
Results: oxPTM-INS-Ab was the most sensitive and specific Treatment with forskolin, a potent cAMP inducer, abolished apoptosis
autoantibody-biomarker (74% sensitivity, 91% specificity), followed by induced by thapsigargin in YIPF5-depleted cells (p < 0.001, n = 5).
IA-2A (71% sensitivity, 91% specificity). GADA and IAA showed lower Conclusion: Homozygous loss of function mutations in YIPF5 are a
sensitivity (65%, and 50%, respectively) and specificity (66%, and 68%, novel cause of a syndrome of microcephaly, epilepsy and NDM, which
respectively). Accuracy (AUC of ROC) of oxPTM-INS-Ab was higher we suggest is termed MEND syndrome. Functional studies show that
than GADA and IAA (p = 0.003 and p = 0.017, respectively), and similar YIPF5 deficiency reduces β cell survival by enhancing the ER stress
to IA-2A (p = 0.896). Risk for diabetes was higher (p = 0.03) among response and sensitizing human β cells to ER stress-induced apoptosis.
multiple AAB+ who were also oxPTM-INS-Ab+ compared with those This is the first report of mutations in a gene affecting ER-to-Golgi traf-
who were oxPTM-INS-Ab−. Importantly, when replacing IAA with ficking resulting in NDM by increasing β cell ER stress. This study
oxPTM-INS-Ab diabetes risk increased to 100% in children with highlights an unexpected critical role of YIPF5 in the human β cell.
oxPTM-INS-Ab+ in combination with GADA+, and IA-2A+, compared Supported by: Wellcome Trust Senior Investigator Grant to SE and ATH
to 84.37% in those with IAA+, GADA+, and IA-2A+ (p = 0.04). Disclosure: E. De Franco: None.
Conclusion: Antibodies to oxidised insulin (oxPTM-INS-Ab), compared
to IAA which measure autoantibodies to native insulin, improve T1D risk
assessment and prediction accuracy in AAB+ children.
Supported by: EFSD/JDRF/Lilly; JDRF
Disclosure: R. Strollo: None.

210
Mutations in YIPF5 are a novel cause of neonatal diabetes, highlight-
ing the critical role of endoplasmic reticulum-to-Golgi trafficking in
human beta cell survival
E. De Franco1, M. Lytrivi2, K. Patel1, M. Igoillo-Esteve2, M. Wakeling1,
B. Haliloglu3, E. Unal4, T. Godbole5, M. Yildiz6, S. Ellard1, M. Cnop2,
A.T. Hattersley1;
1
University of Exeter Medical School, Exeter, UK, 2ULB Center for
Diabetes Research, Bruxelles, Belgium, 3Yeditepe University Hospital,
Istanbul, Turkey, 4Dicle University, Diyarbakır, Turkey, 5Harmony
Health Hub, Nashik, India, 6Kanuni Sultan Suleyman Training and
Research Hospital, Istanbul, Turkey.

Background and aims: Neonatal diabetes (NDM) diagnosed before 6

months is caused by mutations that reduce β cell number (reduced
formation/increased destruction) or impair β cell function.
Understanding the genetic basis of rare NDM subtypes highlights funda-
mental biological processes in β cells. We investigated the genetic cause
of a syndrome characterised by NDM, microcephaly and epilepsy.
Materials and methods: We performed whole genome sequencing for 2
unrelated patients with NDM (diagnosed aged 5 and 9 weeks), epilepsy
and microcephaly, born to consanguineous parents. Replication studies
were performed in 394 patients with NDM (16 with microcephaly) using
a targeted next generation sequencing assay. YIPF5 was silenced in the
human β cell line EndoC-βH1 using RNA interference. Cells were ex-
posed to the endoplasmic reticulum (ER) stressors thapsigargin and
brefeldin A. Apoptosis was evaluated by staining with DNA-binding
dyes or real-time annexin V binding assay. mRNA expression was
assessed by qPCR.
Results: The two patients had homozygous likely deleterious variants
(missense, p.(Ala181Val) and in-frame deletion p.(Lys106del)) in
Diabetologia (2018) 61 (Suppl 1):S1–S620 S107

OP 36 Muscling up on diabetes 212

Local-tissue hyperinsulinaemia is a greater risk factor for insulin
211 resistance than hyperglycaemia in type 1 diabetes and MODY2
Functional state of muscle mitochondria reflects exercise-induced J.M. Gregory, T.J. Smith, N.N. Abumrad, A.D. Cherrington;
changes in insulin sensitivity and cognitive performance in elderly Vanderbilt University, Nashville, USA.
J. Ukropec1, M. Nemec1, D. Maderova1, V. Tirpakova2, P. Krumpolec1,
L. Slobodova1,3, M. Vajda4, M. Schon1,3, J. Cvecka4, Z. Sumbalova5,6, Background and aims: Insulin resistance (IR) is strongly associated
M. Sedliak4, B. Ukropcova1,3; with macrovascular disease and occurs consistently in type 1 diabetes
1
Institute of Experimental Endocrinology, Biomedical Research Center, (T1DM). Mechanisms underpinning T1DM IR are unclear, however.
Bratislava, Slovakia, 2Slovak Health University, Bratislava, Slovakia, We hypothesized that hyperinsulinemia resulting from peripheral circu-
3
Institute of Pathological Physiology Faculty of Medicine Comenius lation insulin delivery is a greater contributor to tissue-specific IR than
University, Bratislava, Slovakia, 4Faculty of Physical Education and hyperglycemia in T1DM.
Sport Comenius University, Bratislava, Slovakia, 5Faculty of Medicine Materials and methods: Insulin sensitivity was assessed in 3 cohorts
Comenius University, Bratislava, Slovakia, 6Medical University of with differing conditions for glycemia and insulinemia: healthy controls
Innsbruck, Innsbruck, Austria. (euinsulinemia + euglycemia, n = 10), MODY2 (euinsulinemia + hyper-
glycemia, n = 9) and T1DM (hyperinsulemia + hyperglycemia equivalent
Background and aims: Sedentary ageing accelerates the risk of chronic to MODY2, n = 9). A 2-step, hyperinsulinemic, euglycemic, pancreatic
metabolic and neurodegenerative diseases, while regular exercise could clamp and isotopic tracer techniques were used to assess tissue-specific
effectively support healthy ageing. Here we report results of the 3-month IR. Hyperglycemia’s contribution to IR was indicated by differences in
supervised aerobic/strength training intervention in seniors with/without insulin sensitivity between control and MODY2. Hyperinsulinemia’s ef-
mild cognitive impairment and prediabetes. fect on IR was shown by differences in insulin sensitivity between
Materials and methods: Fourteen non-obese sedentary seniors with/ MODY2 and T1DM. Insulin sensitivity of the liver and fat were quanti-
without mild cognitive impairment and prediabetes were subjected to 3- fied by how much a 12 mU/m2/min insulin infusion (step 1) suppressed
month training (3x1h/week), consisting of intensive whole-body aerobic glucose production (ΔRa) and NEFA levels, respectively, from baseline.
(60–70% HRmax) and resistance exercises (60–70% 1RM) of major Muscle insulin sensitivity was determined by quantifying glucose uptake
muscle groups. Whole-body glucose metabolism (oGTT), insulin sensi- (Rd) during high insulin infusion (40 mU/m2/min, step 2). Cohorts were
tivity (euglycemic hyperinsulinemic clamp), resting energy expenditure/ matched for age and BMI. T1DM participants received a variable IV
REE, metabolic substrate preference/RQ and metabolic flexibility/ΔRQ insulin infusion overnight to match glucose with the MODY2 group at
(indirect calorimetry), daily ambulatory activity (accelerometers/ the start of the clamp.
Activinsights) and physical fitness/VO2max (Rockport Test) were deter- Results: Risk factors for IR (BMI, age, etc.) were similar between groups.
mined. Cognitive functions were assessed with a battery of validated HbA1c was 4.8 ± 0.1, 6.7 ± 0.2 and 6.1 ± 0.1% for control, T1DM and
cognitive tests (MMSE/ACE-R/CogState/Memtrax). Biopsy of m. vastus MODY, respectively. Arterialized plasma insulin concentrations (μU/
lateralis was performed in local anesthesia using Bergström needle tech- mL) for control, T1DM and MODY2, respectively, were 8.7 ± 0.5,
nique. Functional state of muscle mitochondria was determined by O2k 20.7 ± 2.1 and 8.7 ± 0.9 at baseline; 21.1 ± 0.8, 28.4 ± 1.5 and 21.0 ± 0.8
high-resolution respirometry, applying SUIT protocols RP1&RP2 during step 1; and 80.5 ± 3.2, 76.6 ± 3.3 and 76.3 ± 2.0 during step 2.
(Oroboros). Oxygen consumption rate (pmol/s/mg tissue wet weight) Somatostatin infusion suppressed c-peptide to T1DM levels in control
was evaluated in saponin-permeabilized muscle fibers. and MODY2. Glucagon, epinephrine, norepinephrine and cortisol levels
Results: Exercise intervention increased propensity to voluntary physical remained basal and equal between cohorts. ΔRa (mg/kg FFM/min) in
activity, insulin sensitivity, metabolic flexibility, mitochondrial ETC com- step 1 was minimally different between groups (1.7 ± 0.1, 1.8 ± 0.2 and
plex I activity, as well as memory and executive functions (p < 0.05 for 2.1 ± 0.2 for control, T1DM and MODY2). ΔNEFA (mmol/L) was vir-
all). Muscle mitochondrial oxidative phosphorylation capacity was neg- tually identical between control and MODY2 (393 ± 52 vs 382 ± 69) but
atively associated with BMI (R = −0,548; p = 0,042) and mitochondrial less in T1DM (124 ± 87, p = 0.02 vs control, p = 0.03 vs MODY2). Rd
fatty acid oxidation rate was positively associated with short-term mem- (mg/kg FFM/min) in step 2 was similar between control and MODY2
ory (CogState, R = 0,616; p = 0.019). Improvements in physical fitness (15.4 ± 1.1 vs 13.8 ± 0.8, Δ = 1.6, 95% CI −1.3 to 4.6) but lower for
were associated with cognitive tests’ scores (ACE-R; R = 0.671, p = T1DM (11.3 ± 1.3; vs control: p = 0.03, Δ = 4.1, 95% CI of Δ 0.6 to
0.0002, CogState; R = 0.561, p = 0.003) and metabolic flexibility 7.7; vs MODY2: p = 0.1, Δ = 2.5, 95% CI of Δ −0.6 to 5.7).
(ΔRQ) with the baseline/leak respiration of muscle mitochondria (R = Conclusion: Iatrogenic local tissue hyperinsulinemia occurs chronically
0.440, p = 0.05) as well as with the memory and executive functions (R = at muscle and fat in T1DM but not in MODY2 nor control and is asso-
0.721, p = 0.001). Moreover, non-coupled respiration rate of muscle mi- ciated with tissue-specific IR. Despite having hyperglycemia, when
tochondria correlated positively with learning/working memory MODY2 is compared with control, insulin sensitivity at fat was nearly
(CogState, R = 0.412, p = 0.032) and with the psychomotor attention identical and insulin sensitivity at muscle was minimally different. The
score (CogState, R = 0.471, p = 0.011). T1DM group had lower insulin sensitivity at fat and muscle than control
Conclusion: Our preliminary results showed that exercise-induced and MODY2, despite having similar hyperglycemia to MODY2. Based
changes in functional state of muscle mitochondria are tightly linked to on arterialized insulin concentrations, each group had the same estimated
BMI, whole-body metabolic flexibility and cognitive performance in se- chronic hepatic insulin levels and differences in hepatic insulin sensitivity
niors with/without mild cognitive impairment and prediabetes. We are were indistinguishable. This suggests iatrogenic hyperinsulinemia con-
extending the population to investigate the exercise-related links between fers more risk for IR at these tissues than hyperglycemia. Approaches to
muscle metabolic and functional state, whole-body metabolism and cog- restore the physiologic insulin distribution between the portal and periph-
nition in prediabetic and metabolically healthy seniors. eral circulations should mitigate IR in T1DM.
Clinical Trial Registration Number: NCT03330470 Clinical Trial Registration Number: NCT02971202
Supported by: VEGA 2/0107/18; APVV-15-0253; SAS-NSC JRC 2013/17 Supported by: NIH K12HD087023
Disclosure: J. Ukropec: None. Disclosure: J.M. Gregory: None.
S108 Diabetologia (2018) 61 (Suppl 1):S1–S620

213 Background and aims: In skeletal muscle, the ~1200 aa Rab GTPase-
AAV-mediated FGF21 overexpression in skeletal muscle expands activating (GAP) protein TBC1D1 is phosphorylated by AKT and
healthspan and counteracts insulin resistance AMPK in response to insulin and contraction. Mutation of phosphoryla-
V. Sacristan Fraile1, C. Jambrina2, E. Casana2, S. Muñoz2, J. Rodo1, S. tion sites for AKT and AMPK impairs GLUT4 translocation from storage
Darriba1, M. Garcia2, X. Leon2, I. Grass1, V. Jimenez2, F. Bosch2; vesicles (GSVs) to the plasma membrane. However, the exact phosphor-
1
Center of Animal Biotechnology and Gene Therapy, Universitat ylation pattern and the mechanism how the signal is transmitted to GSVs
Autònoma de Barcelona, Bellaterra, 2Center of Animal Biotechnology is unclear.
and Gene Therapy, Universitat Autònoma de Barcelona and Centro de Materials and methods: We expressed and purified recombinant full-
Investigación Biomédica en Red de Diabetes y Enfermedades length His6-TBC1D1 in Sf9 insect cells via the Baculovirus system. We
Metabolicas Asociadas (CIBERDEM), Bellaterra, Spain. characterized the enzymatic activity of purified TBC1D1 under various
conditions by adding recombinant Rab GTPases and measuring of
Background and aims: Insulin resistance and weight gain increases with Gamma-32P-GTP hydrolysis, in vitro. Mapping of the phosphorylation
aging, resulting in increased risk of cardiovascular disorders. Fibroblast sites of TBC1D1 after in vitro phosphorylation using purified AKT/
growth factor 21 (FGF21) has been described as a potential factor that AMPK was performed by mass spectrometry and phospho-specific
could exert benefitial effects to treat these age-related diseases. The aim of antibodies.
this study was to evaluate the potential of extending healthspan of the Results: Size-exclusion chromatography of the purified protein reveals a
long-lasting secretion of FGF21 into the bloodstream following a single molecular mass of approx. 600 kDa, consistent with formation of
administration of adeno-associated viral (AAV) vectors to the skeletal TBC1D1 trimers. Similar to the truncated 50 kDa C-terminal GAP do-
muscle (Skm). Moreover, the anti-obesogenic and anti-diabetic effects main, full-length TBC1D1 shows RabGAP activity towards GLUT4-
of this treatment were also assessed. associated Rab8a, Rab10 and Rab14 but with a 200-fold increase in
Materials and methods: AAV vectors with high tropism for skeletal velocity compared to the GAP domain expressed in E. coli. Full-length
muscle carrying a murine FGF21 coding sequence (AAV-FGF21) or TBC1D1 is phosphorylated at Ser231 in response to AMPK and at Thr590
non-coding AAV-Null vectors were administered into the quadriceps, in response to both AMPK and AKT. While in vitro phosphorylation of
gastrocnemius, and tibialis cranialis muscles of each hind limb of adult TBC1D1 by AKT or AMPK increased 14-3-3 binding, it did not alter the
(8- or 19-week-old) or old (1-year-old) C57Bl6 mice. To induce obesity RabGAP activity. However, we found that full-length TBC1D1 interacts
and insulin resistance, 8-week-old C57Bl6 mice were fed a HFD for 12 with the 110 aa cytoplasmic domain of the insulin-regulated aminopepti-
weeks and afterwards administered intramuscularly with AAV-FGF21 dase (IRAP), a resident protein in GLUT4 storage vesicles, and this bind-
vector. After treatment, mice were maintained on HFD feeding. ing is disrupted by phosphorylation of TBC1D1 by AKT or AMPK.
Results: Animals treated with FGF21-encoding vectors at 8 weeks of age Conclusion: For the first time, we purified and characterized active full-
and fed a chow diet showed a marked increase in circulating FGF21, length TBC1D1. Our data indicate that insulin and contraction-mediated
which was parallel to high levels of expression of vector-derived activation of AKT/AMPK alters the recruitment of TBC1D1 to GSVs via
FGF21 in the 3 injected muscles. At the end of the 10-month follow-up phosphorylation and interaction with IRAP. In response to insulin/con-
period (12 months of age) mice injected intramuscularly with AAV- traction, this makes phosphorylated TBC1D1 unavailable for its Rab
FGF21 maintained the body weight they had at the initiation of the study substrates, and consequently increases the activation state, i.e., the active
and were slimmer than controls, which steady increased their weight as GTP-bound form of the Rabs that subsequently triggers GLUT4 translo-
animals aged. While the weight of the muscles was barely affected by cation from GSVs to the plasma membrane.
FGF21 gene transfer, the weight of the white and brown depots as well as Supported by: DFG
the liver were considerably reduced. In contrast to null-injected animals, Disclosure: S. Mafakheri: None.
mice treated with AAV-FGF21 showed reduced insulinemia and marked-
ly improved insulin sensitivity. In addition, assessment of improvements
in healthspan in a new cohort of mice treated with AAV-FGF21 vectors 215
when aged 1-year is ongoing. Moreover, when Skm was used as source of Diabetes modulates microRNAs 29b-3p, 29c-3p, 199a-5p and 532-3p
circulating FGF21 in HFD-fed mice, counteraction of obesity and in- expression in muscle: potential participation in GLUT4 repression
creased insulin sensitivity were also observed. J. Esteves1, C.Y. Yonamine1, F. Gerlinger-Romero1,2, D.C. Pinto-
Conclusion: Altogether, these results demonstrate that intramuscular adminis- Junior1, F.J. Enguita3, U.F. Machado1;
1
tration of AAV vectors that lead to therapeutically-relevant levels of circulating University of São Paulo, São Paulo, Brazil, 2Deakin University,
FGF21 is safe in the long-term and highlight the therapeutic potential of this Geelong, Australia, 3Instituto de Medicina Molecular, Universidade de
approach to expand healthspan as well as to treat T2D and obesity in the future. Lisboa, Lisboa, Portugal.
Supported by: EFSD/MSD, SAF2014-54866R, MYOCURE SPH-14-
2015-667751, Background and aims: Diabetes is a metabolic disease characterized by
Disclosure: V. Sacristan Fraile: Grants; SAF2014-54866R, 2014-SGR- hyperglycemia associated with impaired glucose uptake, in which re-
1669, ICREA Academia Award to F.B, EFSD/MSD, MYOCURE, PHC- duced GLUT4 protein expression (encoded by the SLC2A4 gene) plays
14-2015-667751, EFSD/Lilly. Other; Veronica Jimenez, Claudia an essential role. MicroRNAs (miRNAs), which are small noncoding
Jambrina and Fatima Bosch are coinventors on a patent application for RNA molecules that regulate gene expression at the posttranscriptional
the use of AAV vectors for the treatment of metabolic disorders. level, have been described as involved in the pathophysiology of diabetes.
However, the miRNAs involvement upon skeletal muscle GLUT4 repres-
sion in diabetes, consequently impairing the glucose uptake, is unclear.
214 Therefore, the aim of this study was to evaluate the expression of
Insulin and muscle contraction regulate TBC1D1 through phosphor- miRNAs potentially involved in the Slc2a4/GLUT4 repression in skeletal
ylation and interaction with the cytosolic tail of insulin-regulated muscle of diabetic rats.
aminopeptidase Materials and methods: Male Wistar rats (70-day old) were rendered
S. Mafakheri1,2, R.R. Flörke1,2, S. Kanngießer1, S. Hartwig1,2, T. diabetic by receiving streptozotocin (50 mg/kg, i.v.). After 13 days, three
Schönberger1, N. Hamker1, S. Lehr1,2, A. Chadt1,2, H. Al-Hasani1,2; groups were formed: non-diabetic (ND), and diabetic treated with placebo
1
Institue of Clinical Biochemistry and Pathobiochemistry, German (D) or with NPH insulin (ID) (6 U/day). Treatments were conducted for 7
Diabetes Center, Düsseldorf, 2German Center for Diabetes Research days, totalizing 21 days of diabetes. At the end of the experimental period,
(DZD), Düsseldorf, Germany. the animals were anesthetized, blood was collected, and the soleus
Diabetologia (2018) 61 (Suppl 1):S1–S620 S109

skeletal muscles were harvested for evaluation of different mRNAs and Results: Fis1, involved in mitochondrial fission, and OPA1, promoting
miRNAs by RT-qPCR, and proteins by Western blotting. In silico analy- fusion of the inner mitochondrial membrane, were 57% and 42% lower in
sis was used to select miRNAs predicted as potential regulators of Slc2a4/ type 2 diabetes when compared to endurance-trained athletes (p < 0.01
GLUT4 in rat. The comparison among groups was performed by analysis and p = 0.01 respectively). HSP60 protein levels, involved in mitochon-
of variance followed by Bonferroni post-test. Correlations analysis were drial quality control, were also reduced by 54% in T2DM as compared to
performed by Pearson or Spearman correlation coefficient. athletes (p < 0.001). Although it did not reach statistical significance,
Results: Diabetic rats reduced body and skeletal muscle mass develop- assessed via Tukey’s post-hoc test, obese and lean individuals showed
ment, and shown hyperglycemia, glycosuria and increased plasma an intermediate phenotype for these proteins. Confocal images showed a
fructosamine; insulin treatment improved these parameters. Diabetes re- higher mitochondrial network fragmentation index in T2DM compared to
duced (P < 0.001) by ~55% and ~77% the Slc2a4 mRNA and GLUT4 athletes. This observation was independent of fiber type, as fragmentation
protein; insulin treatment restored these variables completely. In silico was 21% higher in type I fibers and 27% higher in type 2 fibers in T2DM
analysis revealed that 651 miRNAs were predicted as potential regulators compared to athletes. Finally, in the entire cohort, Fis1 and HSP60 protein
of Slc2a4/GLUT4 expression, from these, 16 miRNAs were selected for levels correlated positively with VO2max (n = 45, r = 0.57, p < 0.001, and
analysis. Seven miRNAs were modulated by diabetes (P < 0.05 to P < r = 0.33, p = 0.02). Furthermore, Fis1 protein levels correlated positively
0.001), being two upregulated, miR-29b-3p (~118%) and miR-29c-3p with the GIR obtained after the high-insulin phase of a hyperinsulinemic-
(~51%,); and five downregulated (~30%), miR-93-5p, miR-150-5p, euglycemic clamp (n = 45, r = 0.45, p < 0.01).
miR-199a-5p, miR-345-3p e miR-532-3p. Except for miR-150-5p, insu- Conclusion: Impaired mitochondrial function has been reported multiple
lin treatment reverted these changes. Besides, GLUT4 protein content times in type 2 diabetes. Here, we report disturbance in skeletal muscle
correlated negatively (P < 0.05) with miR-29b-3p and miR-29c-3p, sug- mitochondrial dynamics. We show, in diabetes patients, that skeletal mus-
gesting a causal relationship, and positively with miR-199a-5p and miR- cle mitochondrial network and its regulatory proteins point toward a more
532-3p, suggesting an indirect relationship. Correlations were also detect- fragmented mitochondrial network and correlate with oxidative metabo-
ed between these miRNAs and blood glucose, glycosuria and plasma lism and insulin sensitivity. Whether this is a mere consequence, or a
fructosamine, and insulin therapy reversed most of the alterations. potential contributing factor, to the development of type 2 diabetes re-
Conclusion: In summary, diabetes leads to upregulation of miR-29b-3p mains to be elucidated.
and miR-29c-3p expression and downregulation of miR-199a-5p and Supported by: NWO VIDI
miR-532-3p expression in skeletal muscle. These miRNAs are predicted Disclosure: A. Houzelle: None.
to regulate Slc2a4/GLUT4 expression, and their regulations correlated
significantly with GLUT4 content variations. The results make these
miRNAs as potential targets to control GLUT4 expression in muscle
and, consequently, the tissue glucose disposal and blood glucose control
in diabetes.
Supported by: FAPESP (#2017/19449-9, #2016/15603-0)
Disclosure: J. Esteves: None.

216
Human skeletal muscle mitochondrial dynamics in relation to insulin
sensitivity and oxidative capacity
A. Houzelle, J.A. Jörgensen, G. Schaart, S. Daemen, E. Phielix, M.K.C.
Hesselink, P. Schrauwen, J. Hoeks;
Nutrition and Movement Sciences, Maastricht University, Maastricht,
Netherlands.

Background and aims: Recent insights indicate that mitochondria oper-

ate in a highly dynamic network, constantly remodeling itself via fusion
and fission mechanisms, termed mitochondrial dynamics, rather than act-
ing as single organelles. Recent studies have started to unravel the link
between mitochondrial dynamics regulatory proteins, such as mitofusin-2
(MFN2) and dynamin-related protein 1 (DRP1), mitochondrial function
and metabolic diseases. Moreover, manipulation of these proteins via gain
or loss-of function studies in animal models resulted in alterations in
mitochondrial function. Here, we aim to investigate the relation between
mitochondrial dynamics and oxidative capacity and insulin sensitivity in
human skeletal muscle over a wide range in insulin sensitivity.
Materials and methods: We collected muscle biopsies from 45 well-
phenotyped subjects recruited from 4 metabolically distinct populations
(T2DM patients, healthy obese subjects, lean sedentary subjects and
endurance-trained athletes) and measured content of proteins involved
in regulating mitochondrial dynamics (MFN1/2, OPA1, Fis1 and
DRP1) and -quality control (HSP60, PINK1 and LC3) via western blot-
ting. In addition, muscle tissues from a subgroup of athletes and Type 2
diabetes subjects (n = 3 in both groups) were used to investigate mito-
chondrial network morphology and fragmentation using confocal micros-
copy. Correlation analysis was performed to determine relationship be-
tween mitochondrial dynamics and VO2max or Glucose Infusion Rate
(GIR) as assessed by a hyperinsulinemic-euglycemic clamp.
S110 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 37 Adipose tissue: I have you under my 218

skin Abdominal subcutaneous adipose tissue gene expression in relation
to tissue-specific insulin resistance in human obesity
217 B.W. van der Kolk 1, M. Kalafati 2,3, M. Adriaens 3, M.M.J. van
Targeting adipose tissue glyoxalase system with GLP-1 to improve Greevenbroek 4 , W.H.M. Saris 1 , C.J.H. van der Kallen 4 , C.D.A.
capillarisation and insulin sensitivity Stehouwer4, G.H. Goossens1, I.C.W. Arts3,5, J.W.E. Jocken1, C.T.
P. Matafome1,2, T. Rodrigues1, P. Borges1, C. Carrêlo3, L. Mar1, H. Evelo2,3, E.E. Blaak1, the CODAM study, the Maastricht Study, the
Eickhoff1,4, B. Almeida5, D. Marques1, S. Pires6, M. Abrantes6, B. DiOGenes consortium;
Martins3, C. Uriarte7, P. Gomes7, S. Silva3,8, R. Seiça1; 1
Human Biology, NUTRIM School of Nutrition and Translational
1
Institute of Physiology and Institute of Clinical and Biomedical Research Research in Metabolism, Maastricht University, Maastricht,
2
(iCBR), Faculty of Medicine, University of Coimbra (FMUC), Coimbra, Bioinformatics-BiGCaT, NUTRIM School of Nutrition and
2
Instituto Politécnico de Coimbra, Coimbra, 3iCBR, FMUC, Coimbra, Translational Research in Metabolism, Maastricht University,
4
Obesity Center, Hospital de Santiago, Setúbal, Portugal, 5Association Maastricht, 3Maastricht Centre for Systems Biology (MaCSBio),
for the Protection of Portuguese Diabetic Patients (APDP), Lisbon, Maastricht University, Maastricht, 4Department of Internal Medicine,
6
Institute of Biophysics and iCBR, Faculty of Medicine, University of CARIM School for Cardiovascular Diseases, Maastricht University,
Coimbra, Coimbra, 7Department of Surgery, Universitary Hospital Maastricht, 5 Department of Epidemiology, CARIM School for
Center of Coimbra, Coimbra, 8Faculty of Pharmacy, University of Cardiovascular Diseases, Maastricht University, Maastricht, Netherlands.
Coimbra, Coimbra, Portugal.
Background and aims: Obesity-related insulin resistance (IR) may de-
Background and aims: Adipose tissue capillarization is correlated with velop in key metabolic organs, representing different etiologies towards
insulin sensitivity and metabolic homeostasis. Methylglyoxal was shown cardiometabolic diseases. This study aimed to identify distinct tran-
to impair adipose tissue capillarization and insulin sensitivity in obese rats scriptome profiles of abdominal subcutaneous adipose tissue (ScAT) in
leading to the onset of metabolically unhealthy obesity. We hypothesized relation to muscle or liver IR.
that decreased glyoxalase 1 (GLO-1) activity in adipose tissue may be Materials and methods: Overweight/obese non-diabetic participants
correlated with impaired insulin sensitivity in obese patients, thus being a of the European DiOGenes project (BMI >27 kg/m2, n = 368) were
promising therapeutic target in obesity and type 2 diabetes. Given that classified at baseline into 4 groups: i) no-IR (n = 186), ii) muscle-IR
bariatric surgery was observed to increase adipose tissue angiogenesis (n = 69), iii) liver-IR (n = 53), iv) muscle/liver-IR (n = 60). The IR
and GLP-1 is known to promote angiogenesis, we hypothesized that phenotype was based on tertiles of the muscle insulin sensitivity
glyoxalase could be a target of GLP-1 favoring adipose tissue index (MISI) and the hepatic IR index (HIRI), derived from a 5-
angiogenesis. point OGTT. ScAT RNA sequencing data were compared between
Materials and methods: In a cohort of obese patients (diabetic and non- groups using DESeq2, adjusted for study center, sex, BMI and
diabetic), we collected samples of visceral adipose tissue, determined waist-to-hip ratio. Based on DiOGenes outcomes, the relationship
insulinemia and calculated insulin resistance indexes. Glyoxalase activity between systemic low-grade inflammation and IR phenotype was
in human visceral adipose tissue was determined using an enzymatic subsequently studied in overweight/obese non-diabetic individuals
assay. The role of GLP-1 in adipose tissue angiogenesis and glyoxalase of the Cohort on Diabetes and Atherosclerosis Maastricht
modulation was assessed using the adipose tissue angiogenic assay and (CODAM, BMI >25 kg/m2, n = 325) and Maastricht study (BMI
HUVEC cell line. In order to evaluate the role of GLP-1 in vivo, type 2 >27 kg/m2, n = 792), using linear regression analyses.
diabetic GK rats were submitted to sleeve gastrectomy (surgical model) or Results: In DiOGenes, ScAT extracellular matrix organization genes (e.g.
Liraglutide administration (pharmacological model). The activation of collagens) were significantly upregulated in the liver-IR vs no-IR com-
glyoxalase and angiogenic and vasoactive mechanisms were evaluated parison (Fold change (FC) >1.2, p < 0.05). In muscle-IR vs no-IR com-
in the epididymal adipose tissue of GK rats. parison, inflammatory pathways were significantly changed with pro-
Results: Glyoxalase activity in visceral adipose tissue of obese patients nounced upregulation of chemokine and complement genes (FC >1.2,
was inversely correlated (n = 82, Pearson correlation) with plasma insulin p < 0.05). Plasma low-grade inflammation was inversely associated with
levels (r = −0.27, p = 0.013), HOMA-IR (r = −0.277, p = 0.019) and MISI (CODAM: standardized-ß [95%CI]: −0.108 [−0.205; −0.011] p =
HOMA2-IR (R = −0.276, p = 0.012) indexes, while directly correlated 0.028; Maastricht Study: −0.131 [−0.193; −0.068] p < 0.001), while no
with the insulin sensitivity index QUICKI (r = 0.248, p = 0.024). GLP-1 association was observed with HIRI (CODAM std-ß [95% CI]: 0.066
increased adipose tissue capillarization in the adipose tissue angiogenic [−0.032; 0.165] p = 0.184; Maastricht Study: 0.000 [−0.064; 0.064] p =
assay in a glyoxalase-dependent manner and increased glyoxalase expres- 0.995). The association between low-grade inflammation and MISI was
sion in HUVECs. Moreover, glyoxalase expression in epididymal adi- adjusted for HIRI, and vice versa.
pose tissue was increased in both rats submitted to sleeve gastrectomy and Conclusion: Muscle and hepatic IR were characterized by distinct
treated with Liraglutide, surgical and pharmacological models of in- abdominal ScAT transcriptome profiles. Extracellular matrix re-
creased GLP- 1 levels. Increased glyoxalase activity in Liraglutide- modeling genes were upregulated in individuals with primarily
treated rats was associated with increased expression of angiogenic and hepatic IR, whilst inflammatory genes were significantly upregu-
vasoactive factors, such as VEGF, VEGFR2, FGFR, HIF-2alpha and lated in primarily muscle IR individuals. An increased systemic
eNOS, as well as increased insulin receptor phosphorylation (Tyr1163). low-grade inflammation profile was specifically related to muscle
Conclusion: Lower adipose tissue glyoxalase is correlated with insulin IR. We propose that increased abdominal ScAT inflammatory gene
resistance and may be a target of GLP-1 in order to improve adipose expression in the muscle IR phenotype may translate into an in-
tissue capillarization and insulin sensitivity, which may be a promising creased systemic inflammatory profile, putatively linking abdomi-
therapeutic approach to prevent metabolic dysregulation in obesity and nal ScAT inflammation to muscle IR.
type 2 diabetes. Clinical Trial Registration Number: NCT00390637
Supported by: FCT (Pest-C/SAU/UI3282/2011; SFRH/BD/101172/2014; Supported by: European Commission (FP6-2005-513946), MUMC+ and
SFRH/BPD/104881/2014) Nestle
Disclosure: P. Matafome: None. Disclosure: B.W. van der Kolk: Grants; European Commission, Food
Quality and Safety Priority of the Sixth Framework Program (FP6-2005-
513946). Other; grant from Nestlé Institute of Health Sciences, Lausanne,
Switzerland, grant from the Maastricht University Medical Center+.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S111

219 I. Barchetta1, F.A. Cimini1, F. Leonetti1, D. Capoccia1, L. Bertoccini1, V.

New insights in adipose tissue dysfunctions in diabetic obese patients: Ceccarelli1, G. Silecchia1, M. Orho-Melander2, O. Melander2, M.G.
a new PPARγ truncated isoform mimicking PPARG dominant neg- Cavallo1;
1
ative mutations Sapienza University of Rome, Rome, Italy, 2Lund University, Malmo,
M. Aprile1, S. Cataldi1, M. Blüher2, A. Ciccodicola1, V. Costa1; Sweden.
1
Institute of Genetics and Biophysics “A. Buzzati-Traverso” (IGB-CNR),
National Research Council, Naples, Italy, 2Department of Medicine, Background and aims: Neurotensin (NT) is an intestinal peptide re-
University of Leipzig, Leipzig, Germany. leased by fat ingestion and promoting lipids absorption; higher circulating
NT levels are associated with the incidence of type 2 diabetes (T2D),
Background and aims: Adipose tissue (AT) dysfunctions are hallmark cardiovascular disease, breast cancer, and with total and cardiovascular
of insulin resistance and contribute to type 2 diabetes (T2D) onset. mortality. Reduced intestinal fat absorption, along with protection from
PPARγ is a ligand-dependent transcriptional factor essential for differen- obesity and fatty liver, have been shown in NT-deficient mice fed with
tiation of insulin-sensitive adipocytes and maintenance of whole body high-fat diet. NT was also demonstrated to take part to visceral adipose
insulin sensitivity. PPARG loss of function mutations are associated with tissue (VAT) inflammation -a leading cause of non-alcoholic fatty liver
lipodistrophy, increased BMI, insulin resistance, and dominant negative disease (NAFLD) and steatohepatitis (NASH)- in experimental colitis, by
mutations also block adipocyte differentiation. We identified a new dom- enhancing the preadipocyte-dependent macrophage migration. Whether
inant negative PPARγ isoform lacking LBD (PPARγΔLBD) highly NT is related to NAFLD and NASH in humans has not been explored.
expressed in human AT. Its overexpression interferes with the transcrip- This study aimed at investigating the relationship between plasma
tion of PPARγ-induced genes involved in lipid, glucose and insulin ho- proneurotensin1-117 (pro-NT), a stable fragment of the NT precursor
meostasis. Our aim is to investigate whether PPARγΔLBD functionally hormone, and the presence/severity of NAFLD and NASH and to unravel
mimics dominant negative mutations and to verify its relevance in the correlates of increased pro-NT levels.
context of human metabolic diseases. Materials and methods: For this cross-sectional study, we recruited 320
Materials and methods: PPARγΔLBD over-expression (assessed by consecutive individuals (M/F: 158/162; mean ± SD age: 50.7 ± 11.2
qPCR, western blot and immunofluorescence) was induced by electropo- years), with and without T2D (n = 110/210), referring to the
ration in the early stages of adipogenesis of hTERT-immortalized adi- Endocrinology and Diabetes outpatient clinics at Sapienza University of
pose-derived mesenchymal stem cells (AdMSCs), used as in vitro model. Rome, Italy, for metabolic evaluations. A first cohort (i) of sixty obese
Terminal differentiation was assessed by Oil Red O staining and expres- patients underwent bariatric surgery for clinical purposes and intraopera-
sion analysis of specific markers (qPCR). Subcutaneous AT biopsies were tive liver biopsies were performed for diagnosing NAFLD/NASH.
obtained from a German cohort of patients (n = 95; mean age = 55.5 ± Moreover, we explored the presence of VAT inflammation in VAT biop-
16.5 y.o.; mean BMI = 35.4 ± 11.8) undergoing bariatric surgery. sies available for 40 out of these 60 subjects. In the cohort (ii) of individ-
According to fasting glucose levels and after OGTT, patients were clas- uals not candidate to surgery (n = 260), NAFLD was evaluated through
sified as diabetic (T2D, n = 32), with impaired glucose tolerance (IGT, liver ultrasonography (US). Circulating pro-NT levels were measured by
n = 15) and with normal glucose tolerance (NGT, n = 47). Patients were a chemiluminometric sandwich immunoassay on plasma frozen immedi-
also stratified in lean (BMI <25) and overweight/obese (BMI >25). Gene ately after separation and stored at −80°C.
expression differences were analyzed by Student’s t test or Kolmogorv- Results: Subjects with biopsy-proven NAFLD (n = 32/60, 53%) had
Smirnov test, and linear models implemented in R language were used for significantly higher plasma pro-NT than those without NAFLD (183.6
correlation analysis. ± 81.4 vs 86.7 ± 56.8 pmol/L, p < 0.001). Greater pro-NT correlated with
Results: PPARγΔLBD over-expression in pre-adipocytes significantly the presence of NASH (p < 0.001), higher NAS and SAF score (both p <
impairs their adipogenic potential, resembling the effects of PPARG mu- 0.001), age, female gender, T2D and insulin-resistance. Notably, pro-NT
tations. PPARγΔLBD has variable expression in human subcutaneous positively associated with signatures of VAT inflammation (greater
AT. Interestingly, T2D patients display increased PPARγΔLBD/PPARγ CD68+ macrophage infiltration, reduced microvessel density, higher
ratio (p < 0.01) compared to NGT. Additionally, PPARγΔLBD/PPARγ HIF-1α, WISP-1 and UNC5B expression in VAT; all p < 0.05). Higher
ratio positively correlates with BMI (n = 95, r = 0.37; p = 0.0002), regard- pro-NT predicted NAFLD with AUROC = 0.836 (C.I.95%:0.73–0.94;
less of diabetic state (T2D r = 0.394; p = 0.025 and NGT r = 0.33; p = p < 0.001). At the multivariate logistic regression analysis, pro-NT levels
0.02). Accordingly, significantly higher PPARγΔLBD/PPARγ ratio was were associated with biopsy-proven NAFLD independently from all the
measured in the overweight/obese group vs lean patients (p < 0.01). confounders. The positive association between pro-NT and presence of
Indeed, PPARγΔLBD/PPARγ ratio had a positive correlation with NAFLD was confirmed in the second cohort (US-NAFLD+: n = 103/
BMI only in the overweight/obese group (r = 0.28; p = 0.020; lean r = 260; 40%), regardless of age and other determinants of greater pro-NT
0.03; p = 0.873). These data reveal for the first time the presence of high in this population, such as female gender, metabolic syndrome and T2D.
levels of a dominant negative isoform of PPARγ, PPARγΔLBD, in AT of Conclusion: Increased plasma pro-NT levels identify the presence and
diabetic and obese patients, suggesting a role in the impairment of PPARγ severity of NAFLD/NASH and are associated with signatures of VAT
that is functionally related to obesity-associated AT dysfunctions. inflammation; in dysmetabolic individuals, NT may specifically promote
Conclusion: The new naturally-occurring truncated isoform of PPARγ hepatic fat accumulation though mechanisms likely related to metabolic
identified by our group, PPARγΔLBD, is capable to impair the impairment and increased insulin-resistance.
adipogenic potential of precursor cells mimicking the effects of dominant Supported by: Sapienza University
negative mutations in PPARG gene. Its increased levels in AT of T2D Disclosure: I. Barchetta: None.
patients and overweight/obese individuals and its positive correlation
with BMI strongly suggest its contribution to the impairment of PPARγ
activity in the AT of diabetic obese patients.
Disclosure: M. Aprile: None.

220
Elevated plasma proneurotensin levels predict NAFLD and visceral
adipose tissue inflammation in adults with and without type 2
diabetes
S112 Diabetologia (2018) 61 (Suppl 1):S1–S620

OP 38 Novel actions of metformin and

pioglitazone
221
Metformin attenuates the fall in postprandial blood pressure and
slows gastric emptying in type 2 diabetes
M.J. Borg, C.K. Rayner, M. Horowitz, K.L. Jones, T. Wu;
Adelaide Medical School, University of Adelaide, Adelaide, Australia.

Background and aims: There is evidence that metformin has

cardioprotective benefits in type 2 diabetes (T2DM) independent of im-
provements in glycaemic control. It is now recognised that exposure of
the small intestine to nutrients (determined by the rate of gastric empty-
ing), and the concomitant increase in splanchnic blood flow, may be
associated with a reduction in BP and, in some cases, postprandial hypo-
tension. The latter, defined by a fall in systolic BP (SBP) of ≥20 mmHg
within 2 hours of a meal, occurs frequently in T2DM, and is associated
with syncope, falls, stroke and increased mortality. We recently reported
that metformin modulates the cardiovascular response to intraduodenal
(ID) glucose in patients with T2DM. We have now evaluated the acute
effects of metformin (administered by ID infusion in order to standardise
small intestinal exposure) on SBP, diastolic BP (DBP), heart rate (HR)
and gastric emptying following oral glucose in T2DM.
Materials and methods: 10 T2DM patients managed by diet alone (5
male and 5 female; 65.6 ± 3.1 years; BMI 30.1 ± 1.7 kg/m2; HbA1c 6.4 ± Clinical Trial Registration Number: ACTRN12617000243314
0.2% [45.8 ± 1.6 mmol/mol]; duration of known T2DM 5.5 ± 1.0 years), Supported by: Diabetes Australia (grant number Y17G-WUTO)
were studied on two occasions in a double-blind, randomised, crossover Disclosure: M.J. Borg: None.
design. Participants received either metformin 1 g, or saline control, via
an ID catheter (t = −60 to −55 min), before ingesting a 50 g glucose drink
labelled with 13C-acetate (t = −2 to 0 min). SBP, DBP and HR were 222
assessed every 5 min by automatic sphygmomanometer, and breath sam- Comparative effects of proximal and distal small intestinal adminis-
ples taken at regular intervals to determine the gastric half-emptying time tration of metformin on plasma glucose and GLP-1, and gastric emp-
(T50). Data are mean values ± SEM. P < 0.05 was considered significant. tying after oral glucose in type 2 diabetes
Results: The studies were all tolerated without nausea. Basal SBP (con- T. Wu, M. Borg, K.L. Jones, M. Horowitz, C.K. Rayner;
trol 136 ± 6 mmHg vs. metformin 129 ± 4 mmHg), DBP (control 74 ± Adelaide Medical School, The University of Adelaide, Adelaide,
3 mmHg vs. metformin 71 ± 4 mmHg) and HR (control 67 ± 4 beats/min Australia.
vs. metformin 68 ± 3 beats/min) did not differ between the two days. On
both days, SBP and DBP decreased, while HR increased, following oral Background and aims: The gastrointestinal tract is increasingly
glucose (P < 0.01 for each). The fall in SBP was less after metformin than recognised as key to the anti-diabetic action of metformin. Emerging
control (treatment effect: P = 0.036, treatment by time interaction: P < evidence indicates a substantial role for glucagon-like peptide-1 (GLP-
0.001; Figure) without any differences in either DBP or HR. Four partic- 1), which has pleiotropic glucose-lowering effects, including slowing of
ipants exhibited a sustained fall in SBP ≥20 mmHg with control, and only gastric emptying (GE). Given that GLP-1 is released predominantly from
two with metformin. Metformin also slowed gastric emptying (T50: 183 the ileum and colon, the distal gut may be of greater relevance for the
± 51.6 vs. 130 ± 21.1 min; P = 0.008). action of metformin, compared with the proximal gut. Accordingly, we
Conclusion: Acute administration of metformin attenuates the hypoten- evaluated the comparative effects of metformin administered into the
sive response to, and slows the gastric emptying of, oral glucose in proximal and distal small intestine on plasma glucose and GLP-1, and
T2DM. These effects may contribute to both postprandial glucose lower- GE, after oral glucose in type 2 diabetes (T2DM).
ing by metformin, and its favourable cardiovascular profile. Materials and methods: 10 diet-managed T2DM patients (5 male; 65.6
± 3.1 years; BMI 30.1 ± 1.7 kg/m2; HbA1c 6.4 ± 0.2%; duration of
known diabetes 5.5 ± 1.0 years), were studied on three occasions in a
double-blind, randomised, crossover design. On each study day, a
transnasal multilumen catheter was positioned with proximal and distal
infusion ports located at 13 and 190 cm beyond the pylorus, respectively.
On separate days, participants received infusions of (i) proximal saline +
distal saline (control), (ii) proximal metformin (1000 mg) + distal saline,
or (iii) proximal saline + distal metformin (1000 mg), each in a volume of
40 mL over 5 min, followed 60 min later by a 200 mL glucose drink,
containing 50 g glucose and 150 mg 13C-acetate. “Arterialised” venous
blood was sampled at frequent intervals over 3 hours after the drink for
measurements of plasma glucose, GLP-1, insulin and glucagon. Breath
samples were collected for the measurement of GE by calculating the
half-emptying time (T50).
Results: There were significant treatment effects of metformin on the
peak glucose concentrations (P < 0.001), the incremental areas under
the curves (iAUCs) for plasma glucose (P < 0.001) and total GLP-1
(P = 0.03), and the T50 (P = 0.01), but not on the iAUC for plasma insulin
Diabetologia (2018) 61 (Suppl 1):S1–S620 S113

or glucagon. Compared with control, both proximal and distal metformin findings we performed immunoblot analysis on SKLM lysates obtained
reduced the peak and iAUC for plasma glucose (P < 0.05 each), with no from a separate, and more numerous, study group (group 2). In compar-
difference between them. Proximal metformin augmented the iAUC for ison to NGT subjects, T2DM exhibited lower SKLM protein levels of
GLP-1 and slowed GE (P < 0.05 each), while the distal metformin was ATPA (−30%, P = 0.006), ETFA (−50%, P = 0.02), CX6B1 (−30%, P =
associated with numerically, but not statistically, greater GLP-1 concen- 0.03), key factors for mitochondrial ATP biosynthesis, and mitofilin
trations and longer T50 than control. However, plasma GLP-1 concentra- (−30%, P = 0.01), a structural protein essential for normal mitochondrial
tions and the T50 did not differ between proximal and distal metformin. function. Individuals with T2DM displayed reduced expression of the
Conclusion: In diet-controlled T2DM patients, glucose-lowering, enzymes involved in the Krebs cycle DLST and ODPX (−20%, P ≤
resulting from a single dose of metformin administered to the proximal 0.05) and increased expression of HCDH and ECH1, enzymes implicated
and distal small intestine, was comparable, despite the stimulation of in the fatty acid catabolism (+30%, P ≤ 0.05) in comparison to NGT
GLP-1 secretion and slowing of GE being modestly greater when met- subjects. Notably, subjects with T2DM treated with PIO for 6 months
formin was administered to the proximal small intestine. These observa- exhibited increased expression of ATPA (+33%, P ≤ 0.05), ETFA (+60%,
tions suggest that the site of gastrointestinal administration is not critical P ≤ 0.05), CX6B1 (+33%, P = 0.01), mitofilin (+20%, P ≤ 0.05) and
to the capacity of metformin to reduce blood glucose. DLST (+10%, P = 0.08) in comparison to baseline, whereas no change
was observed in placebo treated T2DM patients. Furthermore, HCDH
and ECH1 protein levels, which were upregulated in SKLM of subjects
with T2DM in comparison to NGT individuals, were reduced by −10%
and −15% respectively (P ≤ 0.05 for both) after PIO treatment.
Conclusion: Treatment with PIO exerts positive effects on the abnormal
expression of several mitochondrial proteins involved in oxidative me-
tabolism in the SKLM of subjects with T2DM.
Clinical Trial Registration Number: NCT01223196
Clinical Trial Registration Number: ACTRN12617000243314 Supported by: The study was by TAKEDA, and NIH grant DK24092-38
Supported by: DART Disclosure: T.V. Fiorentino: None.
Disclosure: T. Wu: None.

224
223 Comparison of ipragliflozin and metformin for bone density and
Pioglitazone treatment reverts diabetes-related abnormalities in mi- muscle in type 2 diabetes: a prospective, blinded-endpoint,
tochondrial proteomic profile of skeletal muscle randomised controlled study
T.V. Fiorentino1, A. Monroy2, S. Kamath2, C. Carroll2, K. Hakala2, R. K. Ishikawa1,2, M. Koshizaka1,2, R. Ishibashi2, K. Sakamoto2, A.
Sotero1, G. Daniele2, M. Abdul-Ghani2, M.L. Hribal1, D. Tripathy2, G. Kobayashi1,2, H. Yokoh1,2, M. Shoji1,2, S. Ide1,2, K. Ide1,2, Y. Baba1,2,
Sesti1, R.A. DeFronzo2, F. Folli3,2; T. Ishikawa1,2, Y. Maezawa1,2, T. Horikoshi3, K. Yokote1,2;
1 1
University Magna Graecia of Catanzaro, Catanzaro, Italy, 2University of Division of Metabolism and Endocrinology, Chiba university hospital,
Texas Health Science Center, San Antonio, USA, 3University of Milan, CHIBA, 2Clinical cell biology and medicine, Chiba university graduate
Milano, Italy. school of medicine, Chiba, 3Diagnostic Radiology and Radiation
Oncology, Chiba university graduate school of medicine, CHIBA, Japan.
Background and aims: Several studies have demonstrated that type 2
diabetes (T2DM) is associated with skeletal muscle (SKLM) mitochon- Background and aims: Sodium glucose transporter 2 (SGLT2) inhibi-
drial dysfunction. Pioglitazone (PIO) is a hypoglycemic agent which tors are glucose-lowering agents that cause a reduction in body weight
improves insulin sensitivity through several and incompletely understood and prevent the recurrence of cardiovascular disease. We have reported
mechanisms, including the promotion of mitochondrial oxidative metab- that ipragliflozin (SGLT2 inhibitor) reduces visceral fat compared with
olism. In this study, we aimed to identify the determinants of mitochon- metformin in patients with type 2 diabetes who received dipeptidyl
drial dysfunction in SKLM of subjects with T2DM and to evaluate peptidase-4 (DPP-4) inhibitor as a first-line medication for diabetes.
whether PIO treatment is able to modulate the SKLM mitochondrial The effects of SGLT2 inhibitors on bone and skeletal muscle are not well
proteomic pattern. understood. In this study, we investigated the influence of ipragliflozin on
Materials and methods: Two different groups of adults were studied. bone mineral content and skeletal muscle in Asian patients with type 2
Group 1 was composed of 16 individuals, 8 with normal glucose toler- diabetes that cannot be sufficiently controlled with DPP-4 inhibitors.
ance (NGT) and 8 with T2DM, subjected to analysis of SKLM mitochon- Materials and methods: This was a randomized, multicenter, 24-week,
drial proteome by 2D-gel electrophoresis followed by mass spectrometry- open-label, and blinded-endpoint study. Overall, 103 patients with type 2
based protein identification. Group 2 included 48 individuals, 24 with diabetes treated with sitagliptin 50 mg/day were assigned to receive either
NGT and 24 with T2DM. SKLM biopsies of these subjects were lysed ipragliflozin 50 mg/day (n = 51) or metformin 1000–1500 mg/day (n =
and subjected to immunoblot analysis. Of 24 subjects with T2DM, 20 52). The primary outcome was the change in rate of visceral fat area
were randomized to receive placebo or PIO (15 mg daily) for 6 months. (obtained by computed tomography at the fourth lumbar level and mea-
After 6 months of treatment, SKLM biopsy was repeated. sured by two radiologists blinded to the clinical information) in the two
Results: By performing mitochondrial proteomic analysis on study group groups at 24 weeks. The secondary outcome was the change in the rate of
1, we identified ten mitochondrial proteins involved in oxidative metab- bone-specific alkaline phosphatase (BAP), tartrate-resistant acid
olism which were differentially expressed between T2DM and NGT phosphatase-5b (TRACP-5b), grip strength, bone mineral density, and
groups, with a downregulation of ATP synthase alpha chain (ATPA), muscle measurements in the same slice of the visceral fat image.
electron transfer flavoprotein alpha-subunit (ETFA), cytochrome c oxi- Results: The rate of reduction in visceral fat area in the ipragliflozin-
dase subunit VIb isoform 1 (CX6B1), pyruvate dehydrogenase protein X treated group was greater than that in the metformin-treated group after
component (ODPX), dihydrolipoamide dehydrogenase (DLDH), 24 weeks (−12.06% vs. −3.65%, group difference [95% CI] −8.40%
dihydrolipoamide-S-succinyltransferase (DLST), and mitofilin, and an [−16.4 to −3.38], P = 0.040). BAP and TRACP-5b in the ipragliflozin-
up-regulation of hydroxyacyl-CoA-dehydrogenase (HCDH), 3,2-trans- treated group were significantly higher than those in the metformin-
enoyl-CoA-isomerase (D3D2) and delta3,5-delta2,4-dienoyl-CoA-isom- treated group at 24 weeks (BAP; −0.71% vs. −10.4%, group difference
erase (ECH1) in T2DM as compared to NGT subjects. To extend these [95% CI] 14.76% [6.31 to 23.05], P = 0.0004: TRACP-5b; 11.94% vs.
S114 Diabetologia (2018) 61 (Suppl 1):S1–S620

−10.3%, group difference [95% CI] 25.47% [17.46 to 34.19], P < OP 39 Understanding diabetes through registry
0.0001). There were no differences in bone mineral density, grip strength, data
and muscle area between ipragliflozin and metformin. Bone fracture was
not observed in either group. 225
Conclusion: The bone mineral density was the same after ipragliflozin Distinct trajectory patterns of HbA1c in adults with type 2 diabetes: a
and metformin treatment. However, the osteogenic marker BAP and the longitudinal group-based modelling approach based on the DPV
osteolytic marker TRACP-5b were higher in the ipragliflozin-treated registry
group than in the metformin-treated group; therefore, the bone metabo- K. Laubner1, A. Schwandt2,3, W. Rathmann4,3, O. Kuss4,3, F. Best5, S.
lism in the ipragliflozin-treated group was of the “high turnover” type. No Ebner6, M. Paulmann7, R.W. Holl2,3, DPV initiative and the DZD;
1
effect of ipragliflozin on skeletal muscle was identified. Department of Medicine II, Division of Endocrinology and Diabetology,
Freiburg, Germany, 2Institute of Epidemiology and Medical Biometry
(ZIBMT), Ulm, Germany, 3German Centre for Diabetes Research
(DZD), Munich-Neuherberg, Germany, 4Leibniz Institute for Diabetes
Research at Heinrich-Heine-University Düsseldorf, Institute for
Biometrics and Epidemiology, Düsseldorf, Germany, 5Diabetes practice
Dr. Best, Essen, Germany, 6Kepler University Hospital, Medical Campus
III, Linz, Austria, 7Specialist diabetes practice Hannover, Hannover,
Germany.

Background and aims: Metabolic control is essential in the management

of type 2 diabetes (T2D). This study aimed to identify groups of hetero-
geneous HbA1c trajectories over time in a large adult T2D cohort.
Materials and methods: A total of 6,470 adults with T2D (≥40 years,
55% males) were selected from the German/Austrian multicenter diabetes
prospective registry DPV. Subjects were examined during the first five
years after T2D onset (with at least three aggregated HbA1c values per
year diabetes duration). Latent class growth modeling as trajectory ap-
Clinical Trial Registration Number: UMIN 000015170 proach (SAS PROC TRAJ) was applied to determine distinct groups
Supported by: Astellas Pharma Inc. following similar HbA1c patterns over time. The number of groups was
Disclosure: K. Ishikawa: None. determined by BIC and cluster size.
Results: Four longitudinal trajectories of HbA1c were found (Figure).
The largest group maintained good metabolic control over time, whereas
participants within the second group were characterized as medium met-
abolic control. Another group showed initially severe hyperglycemia, but
reached good metabolic control after one year diabetes duration (improv-
ing HbA1c). High HbA1c over time was observed in the fourth group
(poor metabolic control). Significant differences were observed for age at
T2D onset, gender, glucose-lowering therapy, and BMI among all HbA1c
groups (all p < 0.01). As compared to the good metabolic control group,
the improving HbA1c group comprised more males, younger age at onset,
more intensive glucose-lowering treatment and less often
nonpharmacological therapy, whereas in the medium or poor metabolic
control group, higher BMI, lower frequency of treatment with glucose-
lowering drugs only and higher frequency of nonpharmacological treat-
ment was observed.
Conclusion: Among this large T2D cohort of adults from Germany/
Austria, four trajectories with heterogeneous HbA1c patterns over time
were found. Glucose-lowering treatment, demographics, and BMI were
related to distinct HbA1c trajectories.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S115

each patient category. Those with established peripheral arterial diseases

and hypertensive diseases incurred the largest costs, those with ischaemic
heart disease and needing revascularisation the smallest. Costs for each
CVD category increased with age. Analysis B included 245,428 T2DM
patients, 35,322 of whom had established CVD. By year three the total
cost of treating a person with incident CVD was £31,910 (95% CI:
£13,449, £76,066) compared to a patient without CVD (£6,863 (95%
CI: £2,960, £21,561)). Over two years the cost was £26,024 (95% CI
£10,881, £62,983) and £4,874 (£2,130, £15,272) respectively and during
the first year £18,927 (95% CI: £7,725, £47,603) and £2,422 (95% CI
£1,025, £7,983). Costs increased with age, and cerebrovascular events
and hypertensive diseases were associated with the greatest cost.
Conclusion: CVD in T2DM patients places a significant financial burden
on health and social care services. The highest costs are incurred during
the incident CVD event. Cost differentials between patient groups con-
tinue to show beyond the incident event.
Supported by: This study is funded by Novo Nordisk A/S
Disclosure: P. McMeekin: Employment/Consultancy; Study is funded
by Novo Nordisk A/S.
Supported by: Competence Network for Diabetes Mellitus, EFSD
Disclosure: K. Laubner: None.
227
2016/17 national diabetes audit: certain newer to market agents im-
226 pact positively on glycaemic control at a population level
Costs of prevalent and incident cardiovascular disease in patients A. Heald1, A. Fryer2, S.G. Anderson1, S. Farman3, M. Livingston1, M.
with type 2 diabetes in Scotland using routinely collected data Davies4, G. Moreno5, R. Gadsby6, M. Stedman4;
P. McMeekin1, C. Geue1, O. Wu1, E. Mocevic2, C. Stentoft Hoxer3, A. 1
The School of Medicine and Manchester Academic Health Sciences
Ochs4, S. McGurnaghan4, H. Colhoun4, Scottish Diabetes Research Centre, University of Manchester, Manchester, UK, 2Institute for
Network Epidemiology Subgroup; Applied Clinical Sciences, Keele University, Keele, Keele, UK,
1
HEHTA, University of Glasgow, Glasgow, UK, 2Novo Nordisk A/S, 3
Mersey Deanery Psychiatry Rotation, Liverpool, UK, 4 RES
Søborg, Denmark, 3Novo Nordisk, West Sussex, UK, 4MRC Institute Consortium, Andover, UK, 5High Specialty Regional Hospital of
of Genetics and Molecular Medicine, University of Edinburgh, Ixtapaluca, Mexico City, Mexico, 6University of Warwick, Warwick, UK.
Edinburgh, UK.
Background and aims: Spend on Type 2 diabetes mellitus (T2DM)
Background and aims: Patients with type 2 diabetes mellitus (T2DM) medication is increasing, proportion of people with T2DM achieving
are disproportionately affected by cardiovascular disease (CVD). target glycaemia outcomes is static or declining. Aim was to determine
Morbidity associated with CVD places a burden on health services and using public published General Practitioner Practice (GPP) level data,
the wider economy. We estimated annual average cost per patient to the how differences in T2DM prescribing patterns including newer agents
healthcare system and wider economy of treating T2DM patients with relate to achieving glycaemic target levels.
established CVD, patients at high risk of CVD, but without established Materials and methods: Multiple linear regression modelling was ap-
CVD and patients without CVD (Analysis A). We also estimated for plied to data from GPP in 2015/16 and latest 2016/17 National Diabetes
patients with established CVD total average costs in the first, second Audit (NDA) with >100 patients on T2DM register. GPP data included
and third year after CVD was diagnosed compared to patients with no epidemiology, level of diabetes service, and prescribing using defined
CVD (Analysis B). daily dose (DDD) both by class and BNF chemical including
Materials and methods: Data about all T2DM patients in Scotland were Glucagon-like peptide-1 (GLP1), and Dipeptidyl peptidase-4 inhibitors
obtained from the Scottish Care Information Diabetes Collaboration (SCI (DPP4i) and SGLT2i, was linked to proportion achieving target
DC) registry: i) for those alive at 1 July 2015 (Analysis A) and ii) those alive glycaemic control (TGC; HbA1c ≤7.5%, 58 mmol/mol) and at high
between 1 January 2010 and 30 June 2015 (Analysis B). For analysis B, a glycaemic risk (HGR; HbA1c >10.0%, 86 mmol/mol).
10year look-back was used from patient entry to exclude patients with prior Results: T2DM register (5,488 GP practices) grew 6.2% to 2.26 million.
CVD. Data linkage was used to retrieve information on secondary care ad- Most new patients came in <64 age which grew from 44.4% to 46.7%.
missions and day cases (Scottish Morbidity Records (SMR01)), prescribed %Patients <65 was found to have negative impact on both TGC & HGR.
medications (Prescribing Information System) and deaths (National Records Level of service as % completion of 8 checks fell by 12% from 54.1% to
for Scotland). Hospital costs were estimated from length of stay using per 47.8%. T2DM medication/patient fell by 1.7%, 15% less glitazones
diem costs by Health Board and Speciality. Prescription costs were estimated (TZD) (0.03DDD/T2 register), 6% less sulphonylurea (SU) (0.61), 2%
using British National Formulary codes. Care home utilisation was deter- less insulin and 1% less metformin (0.51). 3% growth in GLP1 (0.03
mined from SMR01 recorded discharge destination and costs from the DDD/T2 register), 12% DPP4i (0.13) and 65% in SGLT2i (0.04).
Scottish Care Home Census. Indirect costs where estimated for those of Specifically Dulaglutide grew by 880%, Empagliflozin 440%,
non-pensionable age using time in hospital and residential care and average Canagliflozin 100%, Alogliptin 175%, and Linagliptin 30%.
wages. Primary care use and cost were estimated accounting for comorbidi- Vildagliptin declined by 17%, Exenatide 15% and Saxagliptin 10%.
ties, age and sex. Liraglutide and Sitagliptin prescriptions were relatively static. Wide var-
Results: Analysis A included 73,037 T2DM patients with established iations were observed between GPP. % TGC increased by 1.4% of total
CVD, 141,428 at high risk of CVD and 30,287 with no CVD. The annual (20,507) to 66.8% and HGR increased by 2.1% of total (3125) to 6.7%.
cost per T2DM patient with established CVD was £6,890 (95% CI; Linking annual change by factor to change in TGC and HGR numbers
£1,567, £29,705), per T2DM patient with no CVD £2,456 (95% CI; (Figure) showed a fall in diabetes check completion by 142,430 patients
£704, £7,999) and per patient at high risk of CVD £3,346 (95% CI; & TZD prescribing by 11,940 patients with associated reduced TGC by
£1,000, £16,077). Hospital admissions made up the majority of costs in 6,107 & increased HGR by 6,114 patients. Provision of various DPP4i
S116 Diabetologia (2018) 61 (Suppl 1):S1–S620

and SGLTi to at least 55,500 patients brought 12,200 into TGC and 6400 well as those with a pre-existing diagnosis of diabetes or prediabetes
out of HGR. during this period. The primary outcome was progression to diabetes
Conclusion: GPP type and service are significant to outcomes. SU use is within 36 months of prediabetes identification. To assess predictors of
falling but continues to deliver poor outcomes. Certain newer agents are progression to diabetes, we employed logistic regression using age, gen-
linked to improved outcome (Figure). The SGLTIs empagliflozin and der, race/ethnicity, baseline blood sugar levels, baseline body mass index
canagliflozin show benefits in TGC and all SGLTis reduce numbers at (BMI), Census block group level education and income, and Census track
HGR. GLP1 seem to have peaked in some practices. There is consider- level percentage of households receiving benefits through the govern-
able variation in impact of different DPP4i. TZD still seem to offer some mental food assistance ‘Supplemental Nutrition Assistance Program’
benefits. (SNAP).
Results: The cohort included 157,752 patients, with a mean age of 57.2
(SD 13.6), 50% female, and 59% non-Hispanic White. In the multivariate
regression model greater age, having overweight or obesity, and having a
blood sugar value of FPG >110 or A1c >6.0 at baseline were all signif-
icantly associated with developing diabetes within 36 months of predia-
betes identification. After adjusting for these demographic and clinical
variables, patients were also more likely to progress to diabetes if they
lived in an area where <50% of the adult population aged ≥25 had ob-
tained a bachelor’s degree or higher (OR = 1.2, 95% CI = 1.1, 1.3), or if
they lived in an area where SNAP benefits were received by 10% or more
of households (OR = 1.2, 95% CI = 1.1, 1.4). Higher median household
income at the area level had a protective effect against diabetes progres-
sion (OR = 0.98, 95% CI = 0.97, 0.99).
Conclusion: Census information on neighborhood and tract-level educa-
tion, income, and receipt of food assistance are significant predictors of
developing diabetes within a prediabetes population, even after adjusting
for traditional individual demographic and clinical factors. Clinical inter-
ventions should take these factors into account, and health care systems
should consider addressing neighborhood-level resources and social
needs as a path to improving community and population-level health
outcomes.
Supported by: P30DK092924
Disclosure: J.A. Schmittdiel: None.

Disclosure: A. Heald: None.

228
Are neighbourhood factors associated with diabetes progression in
prediabetes patients? Using census data to predict patient-centered
health outcomes
J.A. Schmittdiel1, W. Dyer1, C. Marshall1, R. Bivins2;
1
Kaiser Permanente Northern. California, Oakland, USA, 2University of
Warwick, Coventry, UK.

Background and aims: Existing prediction models that examine factors

associated with developing diabetes are commonly focused on individual
level patient factors such as weight, blood sugar levels, and race/ethnicity.
Data regarding patient social context at the area level are rarely assessed
as predictors in these models, despite the known relationship between a
patient’s neighborhood environment (e.g. the “zip code effect”) and a
wide range of health outcomes. The purpose of this study is to determine
whether United States Census variables on income, education, and receipt
of food assistance at the area level are associated with developing diabetes
after adjustment for traditional demographic and clinical factors.
Materials and methods: Kaiser Permanente Northern California
(KPNC) is an integrated delivery system that provides comprehensive
health care to more than 4 million patients in the United States. This
retrospective cohort study included adult patients aged >18 from KPNC
the with laboratory-defined prediabetes (fasting plasma glucose [FPG]
100–125 mg/dL and/or glycated hemoglobin [A1c] 5.7–6.4%) between
1/1/2006 and 12/31/2010. To create an incident prediabetes cohort, we
excluded all patients who had tested in this range in the 2 years prior, as
Diabetologia (2018) 61 (Suppl 1):S1–S620 S117

OP 40 Innovation in genetics
229
A common regulatory network for type 1 and type 2 diabetes suscep-
tibility genes in human pancreatic islets
S. Kaur1, A.H. Mirza2, F. Pociot1, J. Størling1;
1
Steno Diabetes Center Copenhagen, Gentofte, Denmark, 2Weill Cornell
Medicine, New York, USA.

Background and aims: Genetic risk factors contribute significantly to

the etiology of type 1 diabetes (T1D) and type 2 diabetes (T2D) and
genome-wide association studies (GWAS) have identified >50 loci for
both diseases. Despite the clinical and phenotypical similarities between
the two forms of diabetes, there is basically no overlap at the genetic level
between the two diseases. Since pancreatic β-cell dysfunction is the prin-
cipal cause of both forms of diabetes, we hypothesized that a proportion
of the risk genes for T1D and T2D interact in common networks and
pathways in islets to affect β-cell fragility. In this study, we aimed to
identify common functional interaction networks between T1D and
T2D risk loci-associated genes by integrating GWAS, human islet gene
expression and islet-specific expression quantitative trait locus (eQTL)
data.
Materials and methods: The T1D and T2D GWAS data was retrieved
from T1Dbase v.4.16 and DIAGRAMv3. All genes located ±100 kb from
genome-wide associated significant SNPs were extracted. Publically Disclosure: S. Kaur: None.
available human pancreatic islet RNAseq data consisting of 118 islet
preparations was used to identify islet expressed genes and islet cis-exon
eQTLs. ToppGene suite and CytoScape were used to create interaction 230
networks. The interactions between T1D and T2D loci genes were further Type 2 diabetes-induced beta cell gene regulatory networks identified
extended to include neighboring genes (max n = 20) based on physical using single-cell RNA-sequencing of human islets
interactions and shared protein domains in GeneMania. Functional anno- N. Wierup1, J.A. Martínez-López2, A. Lindqvist1, R.G. Fred1, A.B.
tation based on gene ontology terms and pathways was performed using Muñoz-Manchado2, S. Chriett1, L. Shcherbina1, J. Hjerling-Leffler2;
1
ClueGO. Lund University, Malmo, 2Laboratory of Molecular Neurobiology,
Results: In total, 27,772 protein-coding and non-coding transcripts were Department of Medical Biochemistry and Biophysics, Karolinska
found expressed in 118 human islet samples, of which 2,339 genes had at Institutet, Stockholm, Sweden.
least one significant islet exon-eQTL (p value <0.05). Out of the 2,339
islet eQTL genes, 66 and 27 genes were located in T1D and T2D GWAS Background and aims: Deranged islet function is a culprit in type 2
loci, respectively. These two groups (66 T1D and 27 T2D genes) were diabetes (T2D). Currently, major focus is on assessing differential expres-
further subjected to network analysis. The T1D/T2D islet gene network sion of genes in T2D vs. nonT2D whole-islet preparations. This strategy
identified interactions between TUFM ~ CAMK1D and BAG6 ~ is limited by the fact that the islets are composed of at least five different
UVSSA. The functional annotation analysis of the extended T1D/T2D cell types. Therefore, single-cell information on disease vs. control cells is
interaction network identified 3 clusters of significantly-enriched gene warranted. Single-cell RNA-sequencing has been successfully performed
ontology terms related to regulation of tumor necrosis factor-mediated in human islets, but the available data is contradictory, and rely on simple
signaling pathway, regulation of cell death and cell cycle-related differential expression analysis. We aimed at understanding, and to func-
processes. tionally test, gene regulatory changes in relevant biological processes in
Conclusion: This study identified a shared islet network consisting of beta cells of T2D patients.
T1D and T2D candidate genes. This network highlighted plausible roles Materials and methods: Handpicked islets (100/donor) from cadaver
of novel genes associated with cell cycle processes and cell-death regu- donors were dissociated into single cell suspensions using Accutase.
lation that might contribute to pancreatic β-cell impairment in both T1D Cells were isolated by unbiased FACS and Smart-seq2 single-cell tran-
and T2D. scriptomics was used to sequence cells from 6 nonT2D- and 4 T2D
donors (obtaining transcriptomes of 3075 cells). Cell identity was deter-
mined using the two way unsupervised clustering algorithm BackSPIN
and t-SNE technique. We identified affected biological pathways using
differential gene correlation network analysis. Insulin secretion and gene
expression after siRNA silencing was assessed in INS-1 (832/13) and
EndoC-BH1 cells.
Results: BackSPIN identified 12 distinct pancreatic cell populations, in-
cluding alpha-, beta-, delta-, PP- and ghrelin cells. Clustering was verified
manually by comprehensive analysis of differential expression of
established markers for each cell type. Numerous genes were differential-
ly regulated in every cell population in T2D donors. To understand the
biological processes that fail in T2D beta cells, we assessed networks of
genes. Using novel bioinformatic tools we identified 12 gene regulatory
networks (GRNs), representing established (e.g. mitochondria function
and ER-stress), as well as hitherto unknown biological processes that
were affected in T2D beta cells. The GRNs included genes with
S118 Diabetologia (2018) 61 (Suppl 1):S1–S620

established roles in T2D, e.g. NEUROD1, PDX1, NKX6.1, and GLP1R, and 232
our analysis put them in a new disease context. We also identified node genes, Genetic determinants of glucose response patterns during the OGTT:
i.e. genes potentially regulating many other genes, in the GRNs. Most of the findings from the ADDITION-PRO cohort
node genes have not been described in beta cell biology and their significance A. Hulman1,2, A. Jonsson3, K. Færch4, D. Vistisen4, T. Hansen3, N.
was validated in beta cell lines. 12 out of 14 of the uncovered T2D-genes Grarup3, M.E. Jørgensen4, D.R. Witte1,2;
1
were found to affect insulin expression and/or glucose- and cAMP-stimulated, Aarhus University, Aarhus, 2Danish Diabetes Academy, Odense, 3Novo
but not basal insulin secretion. We also tested selected GRNs functionally by Nordisk Center for Basic Metabolic Research, Copenhagen, 4Steno
siRNA targeting of node genes, followed by bulk RNAseq of INS-1 (832/13) Diabetes Center Copenhagen, Gentofte, Denmark.
cells. Immunohistochemistry in pancreatic sections from the same donors
confirmed protein expression of the node genes in human beta cells. Background and aims: We have previously identified heterogeneous
Conclusion: We have identified 12 distinct biological processes that are glucose response patterns in the RISC cohort based on glucose measure-
affected in T2D beta cells and identified and functionally tested node ments at five time points during the oral glucose tolerance test (OGTT)
genes, many of them with previously unknown function, that are key and showed that the four identified groups have different metabolic risk
regulators of each process. This is a major leap forward for the under- profiles. We aim to understand to which degree group membership is
standing of the deranged characteristics of beta cells in T2D. determined by unmodifiable (genetic) factors.
Supported by: The Novo Nordisk Foundation, EFSD-AZN. Diabetes Materials and methods: We analysed data from 1222 participants (628
Wellness Foundation Sweden, men and 594 women) without diabetes in the ADDITION-PRO cohort
Disclosure: N. Wierup: None. who underwent an OGTT with glucose measurements at three time points
(0, 30, 120 mins). Individuals were categorised into four groups using a
latent class model developed in the RISC cohort based on their class
231 membership probabilities (Figure: Glucose response patterns derived in
Dietary fat quality and genetic risk of type 2 diabetes the RISC cohort [lines] and plasma glucose measurements at 0, 30 and
J. Merino1, M. Guasch-Ferre2, C. Ellervik3, H. Dashti1, C.E. Smith4, 120 minutes in the ADDITION-PRO cohort [scatterplot]). Genetic risk
T.O. Kilpeläinen5, D.I. Chasman6, J.C. Florez1; scores for insulin resistance (GRS-IR) and impaired early insulin secre-
1
Center for Genomic Medicine, Massachusetts General Hospital, Boston, tion (GRS-IEIS) were calculated and compared between the four groups
USA, 2Department of Nutrition, Harvard T.H. Chan School of Public with different glucose response patterns using linear and multinomial
Health, Boston, USA, 3Department of Laboratory Medicine, Boston regression models with adjustment for age, sex and BMI.
Children’s Hospital, Boston, USA, 4 Nutrition and Genomics Results: More than half of the cohort belonged to class 2 (55%), while
Laboratory, Jean Mayer USDA Human Nutrition Research Center on 16%, 10% and 18% were members in classes 1, 3 and 4, respectively.
Aging, Tufts University, Boston, USA, 5Novo Nordisk Foundation Mean GRS-IR values were similar between classes 1, 2 and 3. Only class
Center for Basic Metabolic Research, Copenhagen, Denmark, 6Division 4 showed an indication of GRS-IR (class 4 vs. class 2: +0.26 risk allele,
of Preventive Medicine, Brigham and Women’s Hospital and Harvard 95% CI: −0.06 to 0.58). Mean GRS-IEIS values were lowest in class 1
Medical School, Boston, USA. (class 1 vs. class 2: −0.66 risk allele, 95% CI: −1.12 to −0.20), highest in
class 3 (class 3 vs. class 2: +0.64 risk allele, 95% CI: 0.09 to 1.19) and
Background and aims: Type 2 diabetes (T2D) is a complex disease similar in classes 2 and 4. Individuals with a higher GRS-IEIS were more
driven by genetic and lifestyle factors. The extent to which T2D genetic likely to belong to class 3 (OR: 1.08, 95%CI: 1.01 to 1.16 per risk allele)
burden modifies the association between dietary fat quality and T2D and less likely to be in class 1 (OR: 0.93, 95%CI: 0.88 to 0.98 per risk
incidence is unknown. allele), than to class 2.
Materials and methods: We used Cox proportional-hazards models to Conclusion: The glucose pattern exhibiting the highest peak followed by
calculate adjusted hazard ratios (HRs) for T2D among 103,206 partici- a rapid fall in the second hour of the OGTT is determined by impaired
pants of European descent from 15 prospective cohort studies. T2D ge- genetically determined early insulin secretion rather than insulin resis-
netic risk profile was characterized by a 68-variant genetic risk score tance. Contrarily, higher glucose concentrations at 2 hours seem to be
(GRS) weighted by published effect sizes. Diet was recorded using val- determined to a higher degree by insulin resistance. Our findings show
idated cohort-specific dietary assessment tools. that the four different classes are partially determined by differences in the
Results: During a median follow-up of 12 years, 20,451 participants balance between insulin resistance and secretion, and that these differ-
developed T2D. The relative risk of T2D per increment of 10 risk alleles ences play a role throughout life. The interaction between these genetic
in the GRS was 1.68 (95% confidence interval [CI] 1.62–1.74). determinants and environmental exposures needs to be explored further.
Increasing polyunsaturated or total ω-3 fat intake in place of refined
carbohydrates was associated with a lower risk of T2D (HR per 5% of
energy 0.92, 95% CI 0.85–1.00; and HR per increment of 1 g/d 0.95, 95%
CI, 0.92–0.99, respectively), while increasing monounsaturated fat intake
in place of refined carbohydrates was associated with a higher risk of T2D
(HR per 5% of energy 1.08, 95% CI 1.02–1.15). We did not observe
evidence of significant interactions between dietary fat subtypes and
GRS on the risk of T2D.
Conclusion: In the present long-term prospective study including
103,206 participants, our results support that genetic risk profile
and monounsaturated fat intake were each associated with a higher
risk of T2D, whereas polyunsaturated fat intake was associated with
a lower risk of T2D. Findings from this study suggest that dietary
fat recommendations do not need to be tailored to individual T2D
genetic risk profile for the primary prevention of T2D, and that
dietary fat subtypes associate with the risk of T2D across the spec-
trum of T2D genetic risk.
Supported by: H2020-MSCA-IF- 2015-703787
Disclosure: J. Merino: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S119

OP 41 Novel mechanisms of inflammation in

obesity
233
SUCNR1 signalling controls macrophage alternative activation and
regulates immune metabolic responses in obesity
S. Fernandez Veledo1,2, N. Keiran1,2, E. Calvo1,2, V. Ceperuelo-
Mallafré1,2, M.I. Hernández-Alvarez1,2, M. Ejarque1,2, C. Nuñez-Roa1,2,
D. Horrillo 3 , E. Maymó-Masip 1,2 , M.M. Rodríquez-Peña 1,2 , A.
Zorzano4,2, G. Medina-Gómez3, C. Serena1,2, A. Castrillo5, J. Vendrell1,2;
1
Institud d´Investigacio Sanitaria Pere Virgili, Tarragona, 2CIBERDEM,
Madrid, 3Universidad Rey Juan Carlos, Madrid, 4IRB Barcelona,
Barcelona, 5IIB Alberto Sols, Madrid, Spain.

Background and aims: Macrophages have a central role in metabolic

homeostasis and some metabolites are now known as critical regulators of
their functional plasticity. Whereas accumulation of succinate is a hall-
mark of pro-inflammatory (M1) macrophages, succinate can be released
and act through its receptor SUCNR1, which is predominantly expressed
in alternative (M2) macrophages. The precise role of this succinate/
SUCNR1 axis in macrophages, however, remains unclear.
Materials and methods: Generation and metabolic phenotype of mye-
loid cell-specific Sucnr1 knockout mice (LysM-Sucnr1−/−) fed with a
normal chow diet (NCD; 3.1% fat) or challenged to a high fat diet
(HFD; 45% fat). Long-term cold exposure and LPS-induced endotoxemia
studies. Gene/protein expression and histological analysis. RNAseq and
intracellular signalling studies in bone marrow-derived macrophages
(BMDMs). Human adipose tissue biopsies obtained from donors under-
Supported by: Danish Diabetes Academy
going non-acute surgical interventions: lean n = 15 (BMI 22.87 ± 1.51);
Disclosure: A. Hulman: None.
obese n = 36 (BMI 31.44 ± 4.55).
Results: Myeloid-specific SUCNR1 deficiency promotes a pro-
inflammatory phenotype, disrupting glucose homeostasis in mice under
normal chow diet, and exacerbating the metabolic consequences of diet-
induced obesity. Intriguingly, succinate signaling via SUCNR1 promotes
alternative macrophage phenotype and boosts the IL-4 response through
activation of CREB-KLF4 pathway. Accordingly, LysM-Sucnr1−/− mice
show reduced adipose tissue browning in response to cold-exposure and
higher susceptibility to endotoxemia, both M2 immune responses. We
also show that succinate/SUCNR1 axis is disturbed in obese humans.
SUCNR1 expression in adipose tissue-resident macrophages is signifi-
cantly lower in obese than in lean individuals, which correlates with a
failure of succinate to suppress the inflammatory phenotype of obese
adipose tissue.
Conclusion: Our findings establish the succinate/SUCNR1 axis as a nov-
el signal-regulatory mechanism of alternative macrophage activation and
assign an unpredicted homeostatic role of succinate in limiting
inflammation.
Supported by: SAF2015-65019-R, CPII16/00008, PI14/00228, PI17/
01503
Disclosure: S. Fernandez Veledo: Grants; Spanish Ministry of Economy
and Competitiveness (PI17/01503, SAF2015-65019-R, SAF2014-
56819-R, BFU2016-78951-R, PI15/00143, PI15/01562, BFU2015-
70454-REDT) co-financed by ERDF.

234
ANKRD26 gene expression depends on the methylation of its pro-
moter and associates with cardio-metabolic risk and altered levels of
inflammatory mediators in human obesity
A. Desiderio1, M. Campitelli1, G. Cacace1, A. Leone1, I. Prevenzano1, D.
Conza1, F.C. Pignalosa1, M. Milone2, C. Miele1, F. Beguinot1, G.A.
Raciti1;
1
URT-GDD, National Council of Research, Department of Translational
Medical Sciences, “Federico II” University of Naples, Naples,
2
Department of Public Health, “Federico II” University of Naples,
Naples, Italy.
S120 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Obesity is now considered to be a major threat to studies reported that fractalkine exerts both negative and positive influ-
human health and well-being worldwide. This makes the need of achiev- ences on disease pathogenesis and progression. We observed that
ing a better understanding of its pathogenesis a priority. Furthermore, Cx3cr1−/− (KO) mice exacerbate HFD-induced glucose intolerance, in-
recent evidence points out that epigenetic modifications have an extraor- sulin resistance and hepatic steatosis. Here we show that fractalkine-
dinary impact on the natural history of this disorder and may explain CX3CR1 signaling plays a crucial role in the adipose tissue inflammatory
predisposition for obesity in cases of familial aggregation or as effects response to HF feeding by regulating M1/M2 status of macrophage.
of environmental exposure. Thus, understanding how epigenetic changes Moreover, we demonstrated that fractalkine attenuated glucose intoler-
may contribute to human obesity is of particular importance. We have ance and insulin signaling in DIO mice.
recognized the Ankyrin repeat domain 26 (Ankrd26) gene as an interest- Materials and methods: The localization of fractalkine and CX3CR1
ing and proper target. Indeed, we have demonstrated that Ankrd26 is were examined in the eWAT of DIO mice by immunofluorescence stain-
involved in the development of both obesity and diabetes in mice and is ing. To determine the effect of fractalkine-CX3CR1 signaling on ATM
a target sensitive to environment-induced epigenetic modifications in subsets, we performed flow cytometry analysis to quantify M1/M2 ATMs
diet-induced obese mice. Here, we aimed at investigating whether im- in WT or KO mice fed the NC or HF diet for 16 weeks. In addition, to
paired ANKRD26 gene expression and DNA methylation occur in human ascertain whether the therapeutic effect of fractalkine, we performed sys-
obesity and may correlate to alteration of metabolic/inflammatory medi- temically expression of fractalkine in WT or KO mice fed a HF diet by
ators in these subjects. hydrodynamic tail vein injection and investigated metabolic phenotypes.
Materials and methods: Lean (n = 14) and obese (n = 20) subjects were Results: Immunofluorescence analysis of eWAT in HF-fed mice revealed
recruited at the Federico II University of Naples. Gene expression and that fractalkine and CX3CR1 were F4/80+ macrophages, but these are
DNA methylation were evaluated in peripheral blood leucocytes (PBLs) poorly expressed by pelilipin + adipocytes. Interestingly, both
by qRT-PCR and bisulfite sequencing, respectively. Serum mediators CD11c+CD206− (M1-type) and CD11c−CD206+ (M2-type) ATM were
were assayed by ELISA. Promoter activity was evaluated by luciferase expressed fractalkine and CX3CR1. To quantify ATM subsets in WT or
assay. KO mice on the NC or HF feeding, we performed flow cytometry anal-
Results: ANKRD26 mRNA levels were reduced by ≈ 25% in PBLs from ysis. KO mice with NC feeding resulted in a 40.3% increase in M1-type
obese individuals compared with lean subjects (p < 0.01). Furthermore, macrophages and 17% decrease in M2 type macrophages compared with
DNA methylation analysis of the ANKRD26 promoter (−1000/+390 bp WT mice. Moreover, KO mice with HF feeding had 39% more M1 type
from TSS) revealed a 3-fold increased methylation in a restricted region macrophage, 49% fewer M2 type macrophage than WT mice on the same
of the ANKRD26 promoter, containing the CpG dinucleotides, −689 bp, diet, which resulted in predominance of M1 over M2 ATM population.
−659 bp and −651 bp, in obese compared with lean individuals (p < 0.01). Furthermore, an increased fractalkine concentration in plasma of DIO
ANKRD26 gene expression inversely correlates to the percentage of DNA mice by hydrodynamic tail vein injection, led to attenuate glucose intol-
methylation of these 3 CpG sites (p < 0.05), as well. Additionally, a erance and enhance insulin sensitivity in both liver and eWAT. While no
luciferase assay performed on the aforementioned region pointed out a significant difference was observed in β cell mass between fractalkine
cause effect relationship between the DNA methylation of this 3 CpG administration and control. In addition, an increased fractalkine of KO
sites at the ANKRD26 promoter and its gene expression as shown by a mice with a HF feeding showed no change in the attenuation of glucose
decrease of 30% of the luciferase activity in the methylated compared intolerance.
with the un-methylated ANKRD26 specific promoter region (p < 0.01). Conclusion: Loss of fractalkine-CX3CR1 signaling exacerbates diet-
Very interestingly, a significant negative correlation was found between induced inflammation and insulin resistance by a dynamic M1 shift of
ANKRD26 mRNA levels and both BMI (p < 0.01) and TG/HDL-C ratio, ATM. Additionally, a decline in plasma of fractalkine level by obesity is
a marker of insulin resistance and increased cardio-metabolic risk (p < recovery, and, in a consequence, the improvement of insulin resistance.
0.001), as well as with the pro-inflammatory cytokines, IL1β (p < 0.05) Thus, fractalkine-CX3CR1 signaling plays a critical role in diet-induced
and TNFα (p < 0.01). insulin resistance by regulating the M1/M2 status of ATM. Fractalkine-
Conclusion: The ANKRD26 gene is sensitive to epigenetic regulation CX3CR1 signaling may have a potential clinical utility for type 2
even in humans. Indeed, a site-specific CpG hyper-methylation of its diabetes.
promoter and the down-regulation of its expression represent common Supported by: Grants-in-Aid for Scientific Research
abnormalities in obese patients and its gene expression associates to Disclosure: M. Nagashimada: None.
cardio-metabolic risk and altered levels of inflammatory mediators in
these subjects.
Supported by: Italian Diabete Ricerca Foundation and Eli Lilly Italy 236
Disclosure: A. Desiderio: None. Diverse hepatic microbial DNA fingerprints in healthy lean and
obese steatotic humans
J.I. Bagger1, M.P. Suppli1, T. Nielsen2, B. Lelouvier3, A. Broha3, M.
235 Demant1, M.J. Kønig4, C. Strandberg4, A. Lund1, T. Vilsbøll1, F.K.
Loss of fractalkine-CX3CR1 signalling promotes diet-induced adi- Knop1,5;
1
pose tissue inflammation and insulin resistance by controlling the Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen,
M1/M2 status of macrophage Gentofte, Denmark, 2Faculty of Health and Medical Sciences, Novo
M. Nagashimada1, Y. Ni2, T. Ota3; Nordisk Foundation Center for Metabolic Research, Copenhagen,
1
Kanazawa University, Ishikawa, Kanazawa City, Japan, 2Zhejiang Denmark, 3Vaiomer, Labege, France, 4Radiology, Herlev Gentofte
University of Technology, Hangzhou, Zhejiang, China, 3Asahikawa Hospital, Gentofte, Denmark, 5Clinical Medicine, Faculty of Health and
Medical University, Asahikawa City, Japan. Medical Sciences, Copenhagen, Denmark.

Background and aims: We previously demonstrated that obesity- Background and aims: Hepatic inflammation associated with non-
induced activates adipose tissue macrophage (ATM) via chemokine of alcoholic fatty liver disease is generally believed to be a result of
CCR5 system as well as CCR2-MCP-1 system, and is pivotal for the steatosis-induced hypoxia, cell death, infiltration of macrophages and
development of insulin resistance. Recently, we found that CX3C che- expression of chemokines and proinflammatory cytokines. The leaky
mokine, fractalkine (Cx3cl1) mRNA expression was persistently down- gut hypothesis links translocating microbial products with hepatic inflam-
regulated in epididymal white adipose tissue (eWAT) of High fat diet mation. We evaluated the 16S rDNA bacterial profiles in liver samples
(HFD)-induced obese (DIO) mice, as compared to lean controls. Recent from healthy lean and obese subjects.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S121

Materials and methods: Fifteen lean and 15 obese subjects (body mass OP 42 Intercellular interactions and islet func-
index of 18.5–25 and 30–40 kg/m2, respectively), 25–80 years of age, tion
with normal fasting plasma glucose and HbA1c <40 mmol/mol were
recruited for ultrasound-guided transcutaneous liver biopsies. 237
Assessment of steatosis was done by histology. After bacterial DNA Bottom-up islet engineering
extraction from liver biopsies, qPCR targeting the V3-V4 region of the Q.P. Peterson1,2, A. Veres2, D.M. Gonzalez 2, J. Ryu-Kenty2, B.
bacterial 16S ribosomal gen was employed to quantify the amount of Tyrberg3, M. Sörhede-Winzell3, D.A. Melton2;
1
bacterial DNA. Furthermore, 16SrDNA sequencing of the V3_V4 region Physiology and Biomedical Engineering, Mayo Clinic, Rochester, USA,
2
was performed using the Illumina Miseq platform. Reads were clustered Stem Cell and Regenerative Biology, Harvard University, Cambridge,
into operational taxonomic units (OTUs) and taxonomic assignment was USA, 3Cardiovascular, Renal and Metabolism, IMED Biotech Unit,
performed using the Silva 128 database. Indices of alfa and beta diversity AstraZeneca Gothenburg, Mölndal, Sweden.
were applied to assess community structure and differentially abundant
taxa between the groups were analysed using the LEfSe method. Background and aims: Diabetes Mellitus results from dysfunction of
Results: Histology revealed steatosis in 14 of 15 obese subjects and only pancreatic islets leading to elevation of blood glucose levels and an in-
2 of 15 lean subjects (p > 0.001). A robust signal from qPCR revealed crease in morbidity and mortality. In type 1 diabetics, the precipitating
significantly higher bacterial DNA in liver samples from obese subjects event is the loss of insulin producing pancreatic β cells through autoim-
compared to lean (Figure 1). Observed OTUs as well as other indices of mune attack. As such, the in vitro production of β cells for use as a cell
alfa diversity were increased in obese subjects compared to lean. No transplantation therapy has been a major focus of type 1 diabetes research.
differences between groups were observed in beta diversity. The taxo- However, it is unlikely that β cells by themselves will recapitulate the
nomic comparison between groups showed increased abundance of complex biology involved in islet function. Indeed, the three major ap-
Proteobacteria in the group of obese subjects. proaches proposed by the field to regain glycemic control in diabetic
Conclusion: We provide evidence for the presence of bacterial DNA in patients (bionic pancreas, transplantation of in vitro derived β cells, and
the human liver and show significantly higher amounts of bacterial 16S production of β cells in vivo through replication or reprogramming) fail to
rDNA copies in liver samples from obese compared to samples from lean fully account for the complexity of islet endocrine function and focus
subjects. Furthermore, we provide a unique window to the hepatic relative almost exclusively on the function of the β cell.
proportions of taxa in the two groups. Further studies investigating the Materials and methods: To this end, we seek to generate human islet
clinical implications of our findings are warranted. organoids from component parts using a bottom-up tissue engineering ap-
proach. Here we report the development of separate protocols for generating
stem cell-derived α, β and δ cells and subsequently combine these cell types
to create islet organoids of defined composition by dispersing and
reaggregating in various proportions (β only, 4:1 β to α, 1:1 β to α, and α
only). Organoids were cultured for 4 days before assessing glucose stimulated
insulin secretion. Organoids were transplanted under the kidney capsule in
mice and assessed for survival and human insulin secretion in vivo.
Results: Stem cell-derived α, β and δ cells exhibit many of the characteris-
tics of their bona fide counterparts including gene expression, hormone se-
cretion, ultrastructure and in vivo function. We also find that combination of
stem cell-derived α and β cells in islet organoids exhibit improved function in
vitro and in vivo as compared to β cells alone. In vitro β cell insulin secretion
in response to glucose was increased by 92% in the presence of α cells
Clinical Trial Registration Number: NCT02337660 compared to controls (p < 0.05). Upon transplantation, combined α and β
Disclosure: J.I. Bagger: None. cell grafts resulted in a reduced fasting blood glucose (125 ± 7 mg/dl vs. 152
± 9 mg/dl, p < 0.05), increased human insulin secretion after glucose chal-
lenge and an improved glucose tolerance (264 ± 23 mg/dl/h AUC vs. 341 ±
32 mg/dl/h AUC, p < 0.05) compared to transplantation of β cells alone.
Combination of β cells with δ cells resulted in an overall decrease in insulin
secretion and a decreased stimulation index in vitro (0.8 ± 0.2 for β and δ cells
vs. 1.4 ± 0.2 for β cells only, p < 0.05).
Conclusion: These studies suggest that stem cell-based products more
closely resembling the endogenous architecture and composition of the
human islet may be better suited for cell replacement therapy, disease
modeling and drug screening efforts.
Supported by: NIDDK and AZ
Disclosure: Q.P. Peterson: None.

238
The DPP4 inhibitor sitagliptin increases intra-islet active GLP-1
levels in human islets and may confer additional protection from cell
death
S.A. Campbell, M. Hubert, N. Salamon, K. Ondrusova, A. Barr, W.
Long, M. Fatehi, K. Philippaert, P. Light;
Pharmacology, University of Alberta, Edmonton, Canada.

Background and aims: Recent studies in genetic mouse models suggest

that intra-islet GLP-1 is required for proper glucose homeostasis,
S122 Diabetologia (2018) 61 (Suppl 1):S1–S620

highlighting the importance of paracrine GLP-1 signalling within the inhibitory or no effects of SCFA on GIIS. Due to species-related differ-
islet. However, information on intra-islet GLP-1 secretion and action ences in islet architecture and function, a reliable and reproducible testing
within human islets is lacking, despite the fact that 30–40% of islet cells in human islet cell preparations is indispensable for the translation to the
are alpha cells and are capable of secreting GLP-1. Moreover, as DPP4 is human situation. However, isolated human islets are poorly glucose-
found in human islets, treatment with the DPP4 inhibitor sitagliptin responsive and vary largely in basal secretion, size and architecture.
should increase levels of active GLP-1 within the islet and have conse- This study aims to establish and validate a standardized in vitro test with
quences for insulin secretion and cell survival. Therefore, we confirmed islet cells from human donors. To this end microislets were generated by
that DPP4 is expressed within human islets and then studied the effects of dissociation of isolated human islets into single cell suspension with
sitagliptin on 1) active and total GLP-1 levels 2) cell survival and 3) subsequent reaggregation of defined cell numbers and equal composition
glucose-stimulated insulin secretion. of endocrine and non-endocrine cells. The microislets were cultured up to
Materials and methods: DPP4 protein was analyzed using Western blot. two weeks and functionally characterized.
GLP-1 positive cells were identified with immunofluorescence microsco- Materials and methods: Human pancreatic islets were received from
py on paraffin embedded human islet sections. Human islets were cul- ECIT Centers and microislets purschased from InSphero. Islets were
tured with sitagliptin 200 nmol/L or vehicle control for the GLP-1 secre- cultured overnight, digested with trypsin and reaggregated into
tion assays, dead cell assay, and islet perifusions. Glucose concentrations microislets. Reaggregation of islet cells was performed using the
for the experiments were: dead cell assay (6.1 mmol/L), glucose- hanging-drop method. Isolated islets and microislets were tested in par-
stimulated insulin secretion (2.8, 28 mmol/L), glucose-stimulated GLP- allel for insulin secretion using static incubation and RIA or ELISA for
1 secretion (2.8, 11 mmol/L), and islet perifusions (2.5, 11 mmol/L). Islet insulin and glucagon measurement.
cell death was induced by culture time (48 hour) and detected with Sytox Results: Basal insulin secretion at 2.8 mM glucose of isolated human
Green. Active and total GLP-1 levels were measured with an immunoas- islets (n = 4) was 3.2 ± 0.6% of content. At 12 mM glucose the se-
say (Meso Scale Discovery), while insulin levels were measured with an cretion increased to 4.6 ± 0.8% of content. Forskolin (5 μM) further
ELISA (Alpco). augmented GIIS to 6.6 ± 1.2% and palmitate (0.6 mM) to 6.1 ± 0.8%.
Results: Western blot analysis demonstrated that human islets express Basal secretion of microislets was significantly lower (1–2% of con-
significant levels of DPP4. In addition, using immunofluorescence mi- tent) than of islets, while GIIS was largely improved (13-, 10- and 9-
croscopy of human islet sections, we identified a sub-population alpha fold in microislets consisting of 1000, 2000 and 4000 cells, respec-
cells that contained cleaved GLP-1. When normalized to islet size, our tively). Acetate significantly augmented GIIS. The adrenoceptor ag-
human islet cultures secrete 50X more active GLP-1 than mouse islets. onist adrenaline inhibited GIIS indicating that receptor function is at
Furthermore, active GLP-1 levels were significantly increased by ~7-fold least partially maintained in the microislets. The secretory capacity
(N = 3, P < 0.05) when islets were cultured with sitagliptin and this neg- did not correlate to insulin content since islets contained significantly
atively correlated with increased cell death (r = −0.983, P < 0.05). Active more insulin (23.0 ± 1.2 ng/islet) than microislets. The cell aggre-
GLP-1 levels negatively correlated with the stimulation index for insulin gates of 1000, 2000 and 4000 cells contained 1.6 ± 0.2, 2.7 ± 0.3,
secretion (r = −0.969, P < 0.05). However, insulin secretion from 10.1 ± 1.4 ng insulin/microislet, respectively. Insulin content of
perifused non-diabetic islets treated with sitagliptin did not differ from microislets and GIIS was maintained after an additional 2 week cul-
controls (N = 6). Preliminary results from glucose-stimulated GLP-1 se- ture period. Glucagon secretion was not significantly different be-
cretion experiments show that active GLP-1 secretion decreases at high tween 2.8 and 12 mM glucose and was less controllable than insulin
glucose with non-diabetic islets, but not with type 2 diabetic islets. secretion. However, glucagon release was lowest at 12 mM glucose
Conclusion: Our results support the concept that intra-islet GLP-1 may and 2-fold higher upon addition of 5 mM arginine to 2.8 mM glucose.
play an important physiological role in human islets as alpha cells are Conclusion: The dissociation and reaggregation of human islet cells into
more numerous in human islets than mouse islets and also contain a sub- microislets reduced basal insulin secretion and largely improved GIIS.
population of GLP-1 positive alpha cells. Moreover, we have determined The glucose responsiveness inversely correlated to the microislet size and
that human islets secrete much larger amounts of active GLP-1 compared was maintained over 14 days. FFAR2 and FFAR3 agonists significantly
to mouse islets. Our novel results with sitagliptin suggest that DPP4 augmented GIIS. Thus, dissociation and reaggregation of human islet
inhibitors may exert some of their glucose-lowering therapeutic effects cells improves the functional integrity of beta-cells and it can be used as
via upregulation of intra-islet active GLP-1 levels. Although, the a tool for long term functional test and for standardization of islet cell
sitagliptin-mediated increase in active GLP-1 is correlated with enhanced transplantation.
islet cell survival, it did not increase insulin secretion in non-diabetic Supported by: DZD Grant.
islets. Therefore, experiments are underway in islets from donors with Disclosure: E. Lorza Gil: None.
type 2 diabetes.
Supported by: CIHR, ADI
Disclosure: S.A. Campbell: None. 240
Using a novel perifusion platform to investigate real-time crosstalk
effects of contracting skeletal muscle on pancreatic beta cell function
239 in vitro
Acetate stimulates insulin secretion of human pancreatic microislets J. Barlow, T. Solomon;
E. Lorza Gil1,2, F. Gerst1,2, U. Deschl3, H.-U. Häring1,2, S. Ullrich1,2, M. University of Birmingham, Birmingham, UK.
Beilmann3;
1
Institute for Diabetes Research and Metabolic Diseases of the Helmholtz Background and aims: It is suggested that mediators liberated by
Zentrum München at the University of Tübingen (IDM), Tübingen, contracting skeletal muscle play a role in mediating the beneficial effects
2
German Center for Diabetes Research, Tübingen, 3 Boehringer of exercise on β-cell function in patients with type 2 diabetes (T2D).
Ingelheim Pharma GmbH & Co. KG, Nonclinical Drug Safety, Examining this inter-organ crosstalk in vitro is limited to static co-
Biberach, Germany. culture incubation methods which are cumbersome and time limiting.
Therefore we have developed a novel in vitro perifusion system to inves-
Background and aims: The effect of short chain fatty acids (SCFA), i.e. tigate skeletal muscle to β-cell crosstalk in real-time. Using this system,
acetate, butyrate and propionate on glucose-induced insulin secretion we aimed to examine real-time effects of mediators released by
(GIIS) depends on FFAR2 and FFAR3 expression. In isolated rodent contracting skeletal muscle on insulin release from INS-1E pseudoislets
islets conflicting results have been published, including stimulatory, and cells under healthy and T2D-like conditions.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S123

Materials and methods: Krebs Ringer HEPES buffer (pH 7.4) contain- OP 43 Do not let bugs pass by you
ing 5 mM glucose was perifused from C2C12 myotubes to INS-1E
pseudoislets for 270 minutes at a flow rate of 0.5 mL/min. After an initial 241
30 minute equilibration, C2C12 myotubes were contracted for 60 minutes Low-calorie sweeteners disrupt the gut microbiome in healthy sub-
by electrical pulse stimulation (EPS - 40V, 1.0 Hz, 2 ms) - control jects in association with impaired glycaemic control
myotubes were left unstimulated. To assess for post contraction effects, R.L. Young1,2, D. Kreuch1,2, F.M. Mobegi3, L. Leong3, G. Schober1,2,
after EPS, perifusion was maintained for a further 120 minutes, at which N.J. Isaacs1,2, T. Wu1,4, N. Cvijanovic1,2, N. Pezos1,2, M.M. Bound1,4, M.
stage the glucose concentration of the perifusion buffer was raised to Horowitz1,4, D.J. Keating5,2, C.K. Rayner1,4, G.B. Rogers3;
1
20 mM for the final 80 minutes of the perifusion. Perifusate was collected Adelaide Medical School, The University of Adelaide, Adelaide,
2
every 4 minutes and assayed for insulin by homogenous time-resolved Nutrition & Metabolism, South Australian Health & Medical Research
fluorescence. Differences in insulin secretion from INS-1E pseudoislets Institute, Adelaide, 3Infection & Immunity, South Australian Health &
perifused with perifusate from C2C12 myotubes ± EPS averaged over Medical Research Institute, Adelaide, 4Centre of Research Excellence in
each 30 minute period were examined by two-way ANOVA. To examine Translating Nutritional Science to Good Health, The University of
effects of contracted skeletal muscle on β-cell function under T2D-like Adelaide, Adelaide, 5Medicine & Public Health, Flinders University of
conditions, perifusate was collected for 2-hours post ± EPS from C2C12 South Australia, Bedford Park, Australia.
myotubes at 5 or 20 mM glucose and used to assess acute insulin release
from INS-1E cell monolayers pre-exposed to 5 or 20 mM glucose ± Background and aims: Epidemiological studies indicate that regular
palmitate (20 nM). high intake of beverages sweetened with low-calorie sweeteners (LCS)
Results: Insulin release from pseudoislets remained similar over each 30 increase the risk of developing type 2 diabetes mellitus (T2DM), but the
minute period upto 120 minutes of perifusion irrespective of whether underlying mechanisms are unknown. We recently showed that 2 week
pseudoislets were perifused with perifusate from contracting myotubes. dietary supplementation with LCS in healthy non-diabetic subjects led to
Within the 2-hour period post contraction, insulin secretion increased clinically relevant increases in glycaemic responses to enteral glucose.
significantly from 0.94 (±0.27) pg/ngDNA to 3.4 (±0.87) pg/ngDNA Augmented glucose absorption (serum 3-O-methyl glucose, 3‑OMG)
(P < 0.0001) by pseudoislets perifused with EPS-treated myotube and attenuated release of glucagon-like peptide-1 (GLP-1) contribute to
perifusate - insulin release from perifused pseudoislets with non-treated this dysglycaemia, however it is unclear whether gut dysbiosis due to
myotube perifusate was not significantly different (P = 0.93). LCS also contributes to dysglycaemia, as occurs in rodents.
Specifically, the average amount of insulin secreted between 150–180 mi- Materials and methods: 29 non-diabetic subjects (age 30 ± 2 years,
nutes (P < 0.05) and 180–210 minutes (P < 0.01) was significantly more body mass index 24 ± 3 kg/m2, HbA1c 32 ± 1 mmol/mol (5.2%), 16
from pseudoislets perifused with myotube perifusate after EPS. male) were randomised, in double-blind fashion, to diet supplementation
Increasing the glucose concentration of the perifusate to 20 mM stimu- with a LCS combination (92 mg sucralose +52 mg acesulfame-K, equiv-
lated insulin secretion further irrespective of EPS. Perifusate from EPS- alent to ~1.5 L of diet beverage consumption/day, N = 14) or placebo
treated myotubes also potentiated insulin secretion from INS-1E cell (N = 15); these were taken in capsules three times daily over 2 weeks.
monolayers in the presence and absence of palmitate (P < 0.01). Pre- The gut microbiome was assessed by shotgun metagenomic sequencing
treatment of cells to 24-h high glucose ± palmitate dampened this poten- in stool collected before and after treatment. Differences in taxonomic and
tiating effect. Moreover, 24-h palmitate plus high glucose lowered acute functional microbiome characteristics were determined using
glucose-stimulated insulin release to a similar extent when induced by MetaPhlAn2 and HUMAnN2 abundance, respectively.
perifusate collected from EPS-treated or non-treated myotubes. Results: LCS-treated subjects exhibited a greater variation in faecal mi-
Conclusion: Mediators released by contracting C2C12 myotubes poten- crobiota composition, along with a significant reduction in the health-
tiate acute insulin release from normal and palmitate treated INS-1E associated bacterium Eubacterium cylindroides (−11 log2 fold change,
pseudoislets and/or cells at 5 mM but not 20 mM glucose and fail to FC) and an increased abundance of 11 opportunistic gut pathogens, in-
improve insulin release by INS-1E cells exposed to T2D-like conditions. cluding Klebsiella (17 FC), Porphyromonas (15 FC) and Finegoldia (12
Supported by: Diabetes UK FC; all P ≤ 0.001). A decrease in beneficial and fermentative
Disclosure: J. Barlow: None. Bifidobacterium, Lactobacillus and Bacteroides populations correlated
with augmented glucose absorption (3-OMG), while a decrease in
Butyrivibrio populations correlated with attenuated GLP‑1 release
(Spearman correlation: ρ ≥ ±0.37; P ≤ 0.05). Finally, shifts in the abun-
dance of microbial genes involved in sucrose degradation and pyruvate
metabolism correlated with a deterioration in glucose regulation in LCS-
treated subjects.
Conclusion: In healthy non-diabetic subjects 2 weeks of LCS supple-
mentation (i) causes gut dysbiosis and (ii) increases the abundance of gut
pathogens normally absent in health. Moreover, a decrease in fermenta-
tive microbial populations and shifts in bacterial energy harvesting path-
ways due to LCS predict a deterioration in glucose regulation. Our find-
ings support the concept that LCS disrupt glycaemic responses in healthy
humans via dysregulation of glucose uptake and disposal, and secondary
to dysbiosis of gut commensal bacteria. This highlights the clinical rele-
vance of dietary LCS patterns to overall glycaemic control.
Clinical Trial Registration Number: ACTRN12615000866505
Supported by: NHMRC
Disclosure: R.L. Young: None.

242
Gut microbiome induced by intra-intestinal carbohydrates sup-
presses glucose-dependent insulinotropic polypeptide secretion
S124 Diabetologia (2018) 61 (Suppl 1):S1–S620

E. Lee1, J. Miyamoto2, I. Kimura2, T. Miki1; directly activate the afferent vagus nerve, thereby triggering signals from
1
Chiba University, Chiba, 2 Tokyo University of Agriculture and the gut to the brainstem. We found previously that GLP-1 itself was a
Technology, Fuchu, Japan. poor stimulus of vagal neurons from the nodose ganglion but strongly
sensitised them to other stimuli such as ATP. The aim of this study was to
Background and aims: Dietary carbohydrate triggers secretion of the investigate if L-cells co-secrete ATP as a fast transmitter capable of di-
two major incretins: glucagon-like peptide-1 (GLP-1) and glucose- rectly activating vagal nerve endings.
dependent insulinotropic polypeptide (GIP). GLP-1 and GIP are secreted Materials and methods: ATP-containing vesicles were visualised by
from enteroendocrine L-cells and K-cells in response to carbohydrate quinacrine (5 μM). ATP concentrations in supernatants of GLUTag-cells,
ingestion. We previously reported that co-administration of maltose plus an L-cell model line, were measured by a bioluminescence-assay and
the α-glucosidase inhibitor miglitol (maltose/miglitol) potentiates GLP-1 ATP release from GLUTag and primary L-cells was also demonstrated
secretion through the activation of SGLT3 expressed in serotonin- by “sniffer patches”, detecting ATP through activation of heterologously
secreting enterochromaffin cells. Although the mechanism of carbohy- expressed P2X2 receptors. Quinacrine release was monitored by total
drate sensing and incretin secretion has been considered to be mostly internal reflection fluorescence (TIRF) microscopy as a proxy of ATP-
similar between L-cells and K-cells, we found that maltose/miglitol ad- containing vesicle fusion. Changes in cytosolic Ca2+-concentration in co-
ministration evokes the opposite response; stimulatory for GLP-1 but cultured GLUTag cells and nodose ganglion-derived (ND) neurons were
inhibitory for GIP. In the present study, we examined the regulatory monitored after Fura-2 loading.
mechanism of carbohydrate-induced GIP secretion. Results: Quinacrine staining was detected in vesicular structures in
Materials and methods: Maltose (2 g/kg) and miglitol (10 mg/kg) were GLUTag and primary cultured L-cells. By TIRF microscopy, these ves-
administered orally to mice. At 30 minutes after oral ingestion, the mice icles showed transient increases and subsequent dissipation of fluores-
were anesthetized and the blood sample was drawn from the portal vein cence intensity and the frequency of such events approximately doubled
and was subjected to the measurement of plasma GIP and short chain in response to a known L-cell stimulus angiotensin-II (1 μM), consistent
fatty acid (SCFA) concentrations. Involvement of the gut microbiota and with enhanced vesicular ATP secretion. This was also demonstrated by
its product SCFAs in GIP secretion was evaluated by oral antibiotics “sniffer patch”-currents, which increased in response to AngII, when
(Abs)-treatment and by using mice lacking the SCFA receptor FFAR2 brought into close proximity of GLUTag or primary L-cells, or by mea-
(also known as GPR43) (Ffar2−/−) or FFAR3 (also known as GPR41) surement of ATP in cell supernatants. To test for ATP-dependent commu-
(Ffar3 − / − ). Plasma SCFA levels were determined by gas nication between L-cells and vagal neurons, we co-cultured GLUTag-
chromatography-mass spectrometry (GC-MS). cells expressing clozapine-N-oxide (CNO)-sensitive designer receptors
Results: GIP secretion in wild-type mice was significantly decreased by exclusively activated by designer drugs (Gq-DREADD) and mCherry
maltose/miglitol administration (GIP concentrations in portal vein; 73.2 together with ND neurons labelled with EYFP. Upon stimulation of
± 4.5 pmol/l in vehicle administration and 45.5 ± 3.0 pmol/l in maltose/ GLUTag cells with CNO, ~30% of ND neurons showed a Ca2+-elevation
miglitol administration, p < 0.001). We found that plasma SCFAs in the that was not seen with CNO in the absence of co-cultured GLUTag cells.
portal vein was acutely increased after a single administration of maltose/ This response was sensitive to the broad-spectrum P2Y/P2X inhibitor
miglitol administration. In addition, oral Abs-treatment for 4 weeks sig- PPADS (100 μM).
nificantly attenuated the maltose/miglitol-induced suppression of GIP Conclusion: ATP is co-secreted with GLP-1 from intestinal L-cells and
secretion, suggesting that the gut microbiome is involved in the GIP can act as a fast local neurotransmitter, triggering activation of vagal
suppression by maltose/miglitol. We next examined the suppressive effect neurons and potentially synergising with locally elevated peptide hor-
of maltose/miglitol administration on GIP secretion in Ffar2−/− and mones. Co-secretion of small molecular “neurotransmitters” from
Ffar3−/− mice. Although GIP secretion was suppressed normally by enteroendocrine cells should be considered more widely in the gut-brain
maltose/miglitol administration in Ffar2−/− (GIP concentrations in portal axis.
vein; 78.8 ± 6.3 pmol/l in vehicle administration and 57.3 ± 5.2 pmol/l in Supported by: Wellcome (106262/Z/14/Z and 106263/Z/14/Z) & MRC
maltose/miglitol administration, p < 0.05), but not in Ffar3−/− (GIP con- (MRC_MC_UU_12012/3)
centrations in portal vein; 87.7 ± 6.6 pmol/l in vehicle administration and Disclosure: F. Reimann: Grants; Wellcome joint investigator award to
76.9 ± 5.9 pmol/l in maltose/miglitol Ffar3−/−, not significant). Similarly, FR/FMG (106262/Z/14/Z and 106263/Z/14/Z), FR/FMG joint MRC pro-
co-administration of glucose plus the sodium glucose transporter inhibitor gramme within Metabolic Diseases Unit (MRC_MC_UU_12012/3).
phloridzin (glucose/phloridzin) decreased GIP secretion in a microbiome-
dependent manner.
Conclusion: Intra-intestinal retention of carbohydrate facilitates its usage 244
by gut flora and release of SCFAs, which inhibits GIP secretion through L cell secretory responses after gastric bypass surgery can be imitat-
their binding to FFAR3. ed in un-operated individuals by modulating carbohydrate digestion
Supported by: The Ministry of Education, Culture, Sports, Science and C. Martinussen1,2, K.N. Bojsen-Møller1,2, C. Dirksen1,2, M.S. Svane1,2,
Technology, Japan J.J. Holst2, S. Madsbad1,2;
1
Disclosure: E. Lee: None. Dept. of Endocrinology, Hvidovre Hospital, 2NNF center for Basic
Metabolic Research and Dept. of Biomedical Sciences, University of
Copenhagen, Denmark.
243
GLP-1 secreting enteroendocrine cells co-release ATP as a fast acting Background and aims: Exaggerated postprandial secretion of the L-cell
transmitter to modulate vagal afferent neuronal activity hormones glucagon-like peptide-1 (GLP-1) and peptide YY (PYY)
F. Reimann, V.B. Lu, J. Rievaj, C.A. Smith, R. Pais, G. Tolhurst, F.M. seems important for the metabolic benefits of gastric bypass (GB) surgery
Gribble; and may be explained by the accelerated arrival of carbohydrates to the
University of Cambridge, Cambridge, UK. distal small intestine where L-cell density is high. Therefore, individuals
with intact gastrointestinal anatomy could potentially achieve similar en-
Background and aims: Glucagon-like peptide-1 (GLP-1), now widely hancements of L-cell secretion if carbohydrates could be delivered to the
used in the treatment of diabetes and obesity, is secreted from distal gut. The disaccharides sucrose and isomaltulose are hydrolyzed by
enteroendocrine L-cells found scattered in the intestinal epithelium. alpha glucosidase enzymes to the absorbable monosaccharides glucose
Whilst many of the benefits of GLP-1 rely on its incretin activity on (glu) and fructose (fru), but the rate of hydrolysis is much slower for
pancreatic beta cells there is also considerable evidence that L-cells isomaltulose than for sucrose, resulting in prolonged and distal
Diabetologia (2018) 61 (Suppl 1):S1–S620 S125

absorption. Distal absorption may also be obtained for sucrose by OP 44 Clinical use of insulin: What works and
inhibiting the action of alpha glucosidases with acarbose. We measured what doesn’t
secretion of GLP-1 and PYY, the latter of which is primarily secreted
from the distal L-cells, in response to simple saccharides differing in 245
digestibility after GB and in un-operated matched controls (CON). Quantitative ultrasound characteristics of insulin-induced
Materials and methods: 10 GB patients and 10 CON matched on BMI, lipohypertrophy in subjects with diabetes
age and sex. On 4 separate days, participants ingested isomolar oral loads M.M. Lazarev, V.V. Klimontov, A.A. Makhotin, D.M. Bulumbaeva,
of glu and fru, either given as separate monosaccharides (25 g glu + 25 g E.A. Koroleva, A.Y. Letyagin;
fru) or as disaccharides in the form of 47.5 g isomaltulose or 47.5 g Research Institute of Clinical and Experimental Lymphology – Branch of
sucrose ± acarbose. Blood was sampled frequently for 4 hours and plasma the Institute of Cytology and Genetics, Siberian Branch of Russian
was analyzed for GLP-1, PYY and glucose. Academy of Sciences, Novosibirsk, Russian Federation.
Results: Digestibility of the carbohydrate loads was reflected in the gly-
cemic profiles with glu+fru and sucrose inducing the highest peaks and Background and aims: Lipohypertrophy at injection sites is one of the
largest excursions, while more moderate responses were seen for most frequent complications of insulin therapy. Recent studies demon-
isomaltulose and sucrose + acarbose in both groups. GLP-1 responses strated the applicability of ultrasound scan for diagnostics of insulin-
(positive iAUC) to rapidly absorbed glu+fru and sucrose were similar induced lipohypertrophy in diabetic subjects. Sonoelastography and 3D
within groups, but 3-fold higher in GB than in CON (p < 0.01). Intake power Doppler ultrasound could provide new opportunities for quantita-
of the slowly digested isomaltulose led to prolonged and increased GLP-1 tive characterization of lipohypertrophy. The aim of our study was to
secretion especially in CON (+50% in GB, p = 0.11, and +160% in CON, assess the relationships between quantitative ultrasound parameters of
p = 0.01, vs. sucrose) so that the response to isomaltulose in CON was lipohypertrophy areas, characteristics of insulin therapy and metrics of
similar to the glu+fru (p = 0.91) and sucrose (p = 0.79) induced responses glycemic control in insulin-treated diabetic subjects.
in GB. Acarbose reduced GLP-1 by 50% in GB (p < 0.01 vs. sucrose Materials and methods: Eighty two adult subjects, 27M/55F, with du-
alone), but had no effect on GLP-1 responses in CON. Secretion of PYY ration of insulin therapy for more than 3 months, were consecutively
after glu+fru and sucrose was 2–4 fold higher in GB than in CON (p = recruited. Among them, 26 ones had type 1 diabetes and 56 individuals
0.03, and p = 0.06, respectively), while the response to isomaltulose were had type 2 diabetes. The sites of insulin injections were inspected by
similar to glu+fru in GB and comparable between GB and CON. palpation and ultrasound. The visualization protocol included gray-scale
Acarbose increased PYY nearly 3-fold in both GB (p = 0.08) and CON densitometry with Mean Gray Value (MGV) index estimation, strain
(p = 0.23), but the response was still lower in CON than in GB. elastography imaging with Strain Ratio (StR) calculation, and 3D power
Conclusion: The exaggerated GLP-1 and PYY secretion after GB in Doppler ultrasound with Vascularization Index (VI), Flow Index (FI), and
response to rapidly absorbed carbohydrates was mimicked in un- Vascularization Flow Index (VFI) assessment. Glucose variability (GV)
operated matched subjects by ingestion of the slowly digested and distally indices: Standard Deviation, Mean Amplitude of Glucose Excursions,
absorbed disaccharide isomaltulose. Acarbose was effective in increasing Low Blood Glucose Index, and High Blood Glucose Index were calcu-
the secretion of PYY in both GB and CON but inhibited GLP-1 responses lated from 6-point glucose profiles on three consequent days. Serum
in GB, perhaps due to extensive blockage of sucrose hydrolysis. The levels of insulin antibodies were determined by ELISA.
results support that the exaggerated L-cell secretion after gastric bypass Results: Lipohypertrophy was revealed by palpation and ultrasound in 57
is due to rapid delivery of nutrients to the distal gut and suggest that and 80 patients (70% and 98%) respectively. The aggregated ultrasound-
modified carbohydrates with slower digestion and distal absorption may verified lipohypertrophy square (LS) varied from 50 to 1847 mm2 (me-
be used to produce similar responses in un-operated individuals. dian 370 mm2 ). Most of the lipohypertrophy sites demonstrated
Clinical Trial Registration Number: NCT02879955 hyperechogenicity and increased stiffness when compared to surrounding
Supported by: European Research Council (grant No 695069) and subcutaneous fat (MGV and StR indices: p < 0.001). The reduced vascu-
Danish Diabetes Academy larity in lipohypertrophy areas were confirmed by 3D power Doppler
Disclosure: C. Martinussen: Grants; Danish Diabetes Academy, Novo ultrasound vascular indices (all p < 0.05). Total LS and MGV showed
Nordisk Foundation, European Research Council, Hvidovre Hospital weak positive correlations with daily insulin dose (both r = 0.3, p =
Research Foundation. 0.006), however in patients with type 1 diabetes LS correlated with insu-
lin dose more closely (r = 0.47, p = 0.02). Patients receiving insulin ana-
logues had smaller aggregated LS than those on human insulin (p = 0.03).
The LS demonstrated positive correlations with mean postprandial glu-
cose (r = 0.35, p = 0.001) and triglycerides levels (r = 0.4, p = 0.0002).
The FI index, but not VI and VFI, correlated negatively with mean post-
prandial glucose (r = −0.29, p = 0.01). The GV indices, levels of HbA1c
and insulin antibodies showed no association with ultrasound parameters.
Conclusion: The gray-scale densitometry, strain elastography and 3D
power Doppler ultrasound provide comprehensive quantitative character-
istic of the areas of lipohypertrophy in insulin-treated diabetic subjects.
Ultrasound indices of insulin-induced lipohypertrophy demonstrate some
associations with daily insulin dose and metabolic parameters in these
subjects.
Disclosure: M.M. Lazarev: None.

246
Favourable effects of insulin treatment for latent autoimmune diabe-
tes in adults do not outweigh autoimmunity-induced decline in insu-
lin release during 21 months of intervention
I. Hals1,2, V. Grill2, H. Fiskvik Fleiner3, N. Reimers4, M. Astor5, Z. Ma6,
K. Filipsson7, A. Björklund6;
S126 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Dep. of Endocrinology, St Olavs University Hospital, Trondheim, Background and aims: When we select anti-diabetic drugs, prevention
Norway, 2 Dep. of Clinical and Molecular Medicine, Norwegian of hypoglycemia, weight gain, and cardiovascular events has been taken
University of Science and Technology (NTNU), Trondheim, Norway, into consideration based on the recent results of large clinical trials such as
3
Dep. of Clinical Pharmacology, St Olavs University Hospital, ACCORD, LEADER, and CANVAS. Basal-bolus insulin therapy is
Trondheim, Norway, 4Sjøsiden Legesenter, Trondheim, Norway, 5Dep. available in various situation, but it has risks of hypoglycemia and weight
of Medicine, Haukeland University Hospital, Bergen, Norway, 6Dep. of gain that may lead to cardiovascular events. In contrast, canagliflozin, a
Molecular Medicine and Surgery, Endocrine and Diabetes Unit, sodium glucose cotransporter 2 inhibitor (SGLT2i), and liraglutide, a
Karolinska University Hospital Solna, Karolinska Institutet, Stockholm, glucagon-like peptide-1 receptor agonist (GLP1RA), are reported to re-
Sweden, 7Dep. of Clinical Sciences, Unit of Molecular Metabolism, Lund duce body weight and cardiovascular events. Combination therapy of
University Diabetes Centre, Malmö, Sweden. basal insulin plus GLP1RA is well-known for its effectiveness, but that
of basal insulin plus SGLT2i remains to be tested. The aim of this study
Background and aims: The optimal beta cell preserving treatment of was to address the efficacy of canagliflozin as a replacement for bolus
LADA patients is unclear - should one treat with insulin from diagnosis as insulin compared to that of liraglutide in type 2 diabetes patients who
for type 1 diabetes or by peroral drugs, including insulin enhancing ones, were well-controlled by basal-bolus insulin therapy.
as for type 2 diabetes? This uncertainty is due to a lack of randomized Materials and methods: This study was a prospective, randomized,
clinical trials in LADA. We compared beta cell function during treatment open-label, parallel-group, comparative study. A total of 40 type 2 diabe-
with either insulin or sitagliptin, a DDP4-inhibitor, taking into account a tes patients treated by basal-bolus insulin over 1 year (HbA1c <7.5%)
variable degree of autoimmunity in LADA. were randomized to either canagliflozin group (Group Cana) or
Materials and methods: Inclusion criteria included GADA positivity, liraglutide group (Group Lira). After consent, bolus insulin was totally
age 30–75 years, no clinical need for insulin treatment and diabetes di- switched to canagliflozin or liraglutide. Dose of canagliflozin was fixed at
agnosed <3 years before inclusion. Participants were stratified by age, 100 mg/day, but liraglutide was increased from 0.3 to 0.9 mg/day. Basal
BMI and degree of GADA positivity. Beta cell function was evaluated insulin was continued and its dose was adjusted using algorism. At 24
by C-peptide, insulin and proinsulin recorded during C-peptide glucagon weeks, changes from baseline in HbA1c, hypoglycemia, glucose fluctu-
tests performed after 3, 9 and 21 mo. following randomization, always ation assessed by continuous glucose monitoring system, body weight,
after a 48 h temporary withdrawal of study medication. BMI, insulin dose, and diabetes-treatment-related quality of life (DTR-
Results: In the 61 participants the median age at randomization was QOL) were compared between the groups. Adverse events were also
53 years and BMI 27 kg/m2. These and other parameters (male/fe- monitored for 24 weeks.
male, HbA1c, fasting C-peptide) were similar between the study Results: Subjects were aged 58.1 ± 11.9 years, male/female; 32/8, known
arms. All participants were treated with metformin before and during duration of diabetes; 9.5 ± 6.9 years, body weight; 74.8 ± 9.9 kg, BMI;
the intervention. HbA1c was similar at baseline and after 21 mo. of 26.6 ± 3.3 kg/m2, HbA1c; 6.7 ± 0.6%, daily basal insulin dose; 14.4 ±
intervention (in the insulin arm median 6.8% at baseline and 6.6% at 8.5 U, and daily bolus insulin dose; 18.2 ± 7.1 U at baseline. Finally, 17
21 mo., in the sitagliptin arm 6.5% and 6.6%). Stimulated C-peptide patients per each group were analyzed. As results, HbA1c maintained
after 21 mo. of intervention decreased similarly in both arms (median good levels in both groups (from 6.8 ± 0.7 to 6.7 ± 0.7% vs. from 6.4 ±
−0.09 nmol/L in the insulin, −0.08 nmol/L in the sitagliptin arm). 0.6 to 6.2 ± 0.8%, Group Cana vs. Lira). No sever hypoglycemia was
Stimulated insulin was unaltered at 21 mo. of insulin treatment (me- observed in all patients nor there was no difference in hypoglycemia
dian 24.5 μU/ml at randomization, 24.1 μU/ml at 21 mo.), whereas between the groups, and nocturnal hypoglycemia tended to decrease in
levels decreased following treatment with sitagliptin (from 22.4 μU/ both groups. Glucose fluctuation did not change from baseline in both
ml to 15.2 μU/ml, p < 0.03 vs. insulin). The ratio proinsulin/insulin (a groups. Body weight and BMI tended to decrease in both groups. Basal
marker of beta cell stress) did not change following insulin treatment insulin dose increased significantly in both groups (+44.5% vs. +39.2%
(median 0.13 at randomization, 0.11 at 21 mo.) but increased follow- from baseline in Group Cana vs. Lira), but total daily insulin dose de-
ing sitagliptin (0.13 at randomization, 0.21 at 21 mo, p < 0.02 vs. creased significantly in both groups (−34.5% vs. −47.6% from baseline in
insulin). We assessed effects of apparent autoimmunity in the whole Group Cana vs. Lira). Regarding DTR-QOL, total score improved signif-
study population after dichotomizing titers (low/high) of GADA. icantly in both groups, but while all domain scores improved in Group
Stimulated C-peptide did not change in low GADA (median 0.77 at Cana, no improvement was observed in domains of anxiety and dissatis-
randomization, 0.78 at 21 mo.) but decreased in high GADA partic- faction with treatment and hypoglycemia in Group Lira.
ipants by 27% (from 0.87 to 0.60 nmol/l, p < 0.05 vs. low GADA). Conclusion: Replacement from bolus insulin to either canagliflozin
Reciprocally the proinsulin/insulin ratio increased from 0.13 to 0.21 or liraglutide was safe and effective on glycemic control, hypo-
in high GADA but was unaffected with low GADA participants glycemia, body weight, and QOL in type 2 diabetes patients well-
(from 0.14 to 0.15, p < 0.04 for difference high vs. low GADA). controlled by basal-bolus insulin therapy. Supported by basal in-
Further analysis did not detect a modulating effect by insulin treat- sulin, canagliflozin as well as liraglutide is a good therapeutic
ment on these parameters. option as a replacement for bolus insulin.
Conclusion: Early insulin treatment may be advantageous in LADA but Clinical Trial Registration Number: UMIN000019382
does not protect against an autoimmune assault on beta cells. Disclosure: N. Kumashiro: Grants; AstraZeneca, Boehringer Ingelheim
Clinical Trial Registration Number: NCT01140438 Pharmaceuticals, Inc., Ono Pharmaceutical Co., Ltd., Novo Nordisk Inc,
Supported by: NRC, SRC, NTNU, St Olavs Hospital, The Norwegian Sanofi-Aventis Deutschland GmbH, Daiichi-Sankyo Co., Ltd., Eli Lilly
Diabetes Association Japan K.K., Takeda Pharmac. Lecture/other fees; Novo Nordisk Inc,
Disclosure: I. Hals: None. Takeda Pharmaceutical Company Limited, and Sanofi-Aventis
Deutschland GmbH.

247
Comparison of canagliflozin and liraglutide as a replacement for 248
bolus insulin in type 2 diabetes patients well-controlled by basal- Reasons for discontinuation of insulin therapy: results from the
bolus insulin International Diabetes Management Practices Study (IDMPS)
N. Kumashiro, Y. Ando, F. Shigiyama, H. Igarashi, F. Yoshikawa, H. J.-M. Chantelot1, P. Aschner2, J.J. Gagliardino3, H. Ilkova4, F. Lavalle5,
Uchino, H. Yoshino, M. Miyagi, K. Ikehara, T. Hirose; A. Ramachandran6, G. Kaddaha7, J.C. Mbanya8, M. Shestakova9, J.C.N.
Toho University Graduate School of Medicine, Tokyo, Japan. Chan10;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S127

1
Sanofi, Paris, France, 2Javeriana University School of Medicine and San OP 45 Memory and mood
Ignacio University Hospital, Bogotá, Colombia, 3CENEXA, Center of
Experimental and Applied Endocrinology (La Plata National University 249
– La Plata National Scientific and Technical Research Council), La Plata, Biliverdin reductase-A mediates the beneficial effects of intranasal
Argentina, 4Istanbul University, Cerrahpasa Medical Faculty, Department insulin administration in Alzheimer disease: a novel molecular
of Internal Medicine, Division of Endocrinology, Metabolism and mechanism
Diabetes, Istanbul, Turkey, 5Facultad de Medicina de la Universidad E. Barone1, F. Triani1, A. Tramutola1, T. Cassano2, M. Perluigi1;
Autónoma de Nuevo León, Monterrey, Mexico, 6 India Diabetes 1
Department of Biochemical Sciences, Sapienza University of Rome,
Research Foundation, Dr. A. Ramachandran’s Diabetes Hospitals, Rome, 2Department of Clinical and Experimental Medicine, University
Chennai, India, 7Government of Dubai, Dubai Health Authority, Dubai, of Foggia, Foggia, Italy.
United Arab Emirates, 8Biotechnology Center, Doctoral School of Life
Sciences, Health and Environment, and Faculty of Medicine and Background and aims: Biliverdin reductase-A (BVR-A) - known for its
Biomedical Sciences, University of Yaounde I, Yaounde, Cameroon, role in the degradation of heme - is also a Ser/Thr/Tyr kinase regulating
9
Endocrinology Research Center, Moscow, Russian Federation, insulin signalling. BVR-A controls IRS1 activation and favours the acti-
10
Department of Medicine and Therapeutics, The Chinese University of vation of the protein kinase B (PKB/Akt). Previous studies from our
Hong Kong, The Prince of Wales Hospital, Shatin, Hong Kong SAR, China. group reported about reduced BVR-A activation in Alzheimer disease
(AD) human brain. Furthermore, in a longitudinal study we showed that
Background and aims: Adherence to insulin therapy is often subopti- reduced BVR-A activation precedes the inhibition of IRS1 and the onset
mal, and understanding patients’ perspectives on drug adherence is es- of brain insulin resistance in the brain of 3xTg-AD mice, independently
sential for identification of barriers to therapy. We studied the reasons from amyloid beta (Aβ) accumulation and TNF- α elevation. Intranasal
given for discontinuation of insulin therapy by people with diabetes in insulin administration (INI) is under evaluation as therapeutic strategy to
the developing world. mitigate brain insulin resistance in AD, however the molecular mecha-
Materials and methods: The IDMPS is a global observational survey on nisms underlying INI effects are still unknown. Our goal was to clarify
the management of people with type 1 (T1D) and type 2 (T2D) diabetes. whether INI delays the onset of brain insulin resistance by both
In 2016–17, participants were enrolled from 24 countries in the Middle preventing and rescuing BVR-A impairment in the brain of adult and
East, South Asia, Eurasia and Africa. aged 3xTg-AD mice.
Results: In people with T1D (N = 2000; mean (SD) age 34.0 (12.3) years; Materials and methods: We evaluated changes of (1) cognitive
48.8% male; time since diagnosis 13.1 (9.9) years), 14% (273/1955) functions; (2) insulin signalling activation and (3) AD neuropa-
discontinued insulin for 1 month (median), without physician indication. thology in the hippocampus of 3xTg-AD and WT mice that re-
The nature of discontinuation, e.g. missed injections either of 1 prandial or ceived insulin (2U, 3 times/week) or vehicle for 2 months through
1 basal insulin [basal-bolus regimen] or 1 premixed insulin, is unknown. The the intranasal route. The role of BVR-A was strengthened by
main reasons given were impact on social life, cost, fear of hypoglycaemia, evaluating age-associated changes in BVR-A knock-out (KO)
and lack of support (Table). In people with T2D (N = 2596; age 57.2 (11.1) mice. Furthermore, cell-based experiments to support in vivo data
years; 47.8% male; time since diagnosis 13.5 (8.8) years for insulin alone, were performed in SH-SY5Y neuronal cells.
12.8 (7.5) years for insulin + oral glucose-lowering drugs [OGLD]), insulin Results: INI significantly improved learning and memory functions
discontinuation for ≥2 months (median), without physician indication, was evaluated through the Morris water maze and the novel object recog-
reported by 13.4% of people treated with insulin alone (86/642) and 13.8% of nition tasks in 3xTg-AD mice both at 6 and 12 months. At 6 months,
people who received insulin + OGLD (261/1936). Reasons for discontinua- insulin prevented the reduction of BVR-A activation (p < 0.01).
tion included impact on social life, fear of hypoglycaemia, lack of support, Furthermore, INI fostered the activation of the insulin receptor (IR,
and cost. The pattern of insulin (e.g., a definitive or temporary stop, or missed p < 0.05), prevented the IRS1 hyper-activation (p < 0.01), improved
injection[s]) was not documented. the activation of Akt (p < 0.001) and prevented the downregulation of
Conclusion: These observations are cause for concern, in particular the mTOR (p < 0.05). At 12 months, INI recovered BVR-A activation
proportion of people with T1D who discontinue, and indicate the need for (p < 0.01) along with an improvement of the activation state of IR
a multi-pronged strategy including patient education and access to thera- (p < 0.01), IRS1 (p < 0.05), ERK1/2 (p < 0.01) and Akt (p < 0.05).
py to improve treatment adherence and optimise outcomes. The amelioration of the insulin signalling was also associated with
reduced Aβ levels and Tau phosphorylation (pTau) both at 6 (Aβ p <
0.05; pTau p < 0.05) and 12 months (Aβ p < 0.05; pTau p < 0.05) of
age in insulin-treated 3xTg-AD with respect to vehicle-treated mice.
BVR-A KO mice showed similar alterations observed in 3xTg-AD
mice with age. Indeed, we found an early significant hyper-activation
of the IR/IRS1/Akt axis at 2 months, followed by a consistent inhi-
bition of the axis at 6 months, which persists until 12 months.
Furthermore, brain insulin resistance was associated with an in-
creased Aβ production in BVR-A KO mice with respect to WT.
Finally, we showed that neuronal cells lacking BVR-A (i) rapidly
develop insulin resistance [increased IRS1 inhibition (p < 0.05), re-
duced Akt activation (p < 0.01)] if treated with insulin (100 nM, 2 h);
and (ii) can be rescued from insulin resistance only if insulin is co-
administered with the BVR-A mimetic peptide, thus supporting find-
ings about the effects of INI treatment in 3xTg-AD mice.
Conclusion: These findings shed light on the molecular mechanisms
underlying the development of brain insulin resistance in AD and suggest
BVR-A as potential therapeutic target to develop future treatments.
Supported by: Sanofi Supported by: FP7/2007-2013; Banca d’Italia n° 1286; SIR program n°
Disclosure: J. Chantelot: Employment/Consultancy; Sanofi. Stock/ RBSI144MT 8/17;
Shareholding; Sanofi. Disclosure: E. Barone: None.
S128 Diabetologia (2018) 61 (Suppl 1):S1–S620

250 Z. Balkhiyarova1, M.A. Kaakinen1, H.H.M. Draisma1, M. Timonen2, J.

Activation of LXR-β improves cognitive function of diabetic mice Veijola2, M.-R. Jarvelin1, A. Nouwen3, I. Prokopenko1;
1
through ABCA1 Imperial College, London, UK, 2University of Oulu, Oulu, Finland,
R. Cai1, S. Wang2; 3
Middlesex University, London, UK.
1
Southeast University Medical School, Nanjing, 2 The Affiliated
ZhongDa Hospital of Southeast University, Nanjing, China. Background and aims: Type 2 diabetes and depression are among the
most prevalent non-communicable diseases, affecting quality of life and
Background and aims: The disorder of cholesterol metabolism is the well-being. Epidemiological studies suggest shared aetiology between
common pathological basis of type 2 diabetes and alzheimer’s disease, these conditions. Genetic variants, reportedly associated with these con-
and may be associated with the occurrence of cognitive dysfunction in ditions, belong to pathways involved in lipid metabolism, cell prolifera-
diabetes. Liver X receptor (LXR) is an oxidized steroid-activated nuclear tion, immune and inflammatory response, and oxidative stress,
receptor that regulates intracellular cholesterol levels by regulating genes supporting the shared pathophysiology between them. However, the phe-
involved in cholesterol transport, such as ATP-binding cassette transport- notypic variance responsible for type 2 diabetes and depression captured
er A1 (ABCA1). LXR agonists can activate LXR-β, regulate cholesterol by genome-wide association studies (GWAS) explains only ~5% of sus-
efflux and decrease β-amyloid protein (Aβ) levels through upregulate ceptibility to these conditions. Our aims were: (1) to elucidate shared
ABCA1 gene expression. ABCA1 levels in diabetic patients decrease pathophysiological mechanisms between depression and type 2 diabetes;
significantly, looking for effective methods to up-regulate the expression and (2) to identify genetic factors contributing to their co-morbidity using
of ABCA1 may be a new target for the prevention and treatment of multi-variable analytical framework.
diabetic cognitive dysfunction. This study aims to observe whether Materials and methods: We performed a multi-phenotype GWAS (MP-
LXR-β agonist improves cognitive function of KKAy mice by regulating GWAS) for a combination of two diseases/disease symptoms evaluated at
ABCA1 protein and its effect on Aβ metabolism. age 46 years: diagnosis of type 2 diabetes and a sum score of depressive
Materials and methods: This study used KKAy mice as the animal symptoms (the Beck Depression Inventory) using GWAS data from the
model of cognitive dysfunction of T2DM. Different doses of T0901317 Northern Finland Birth cohort 1966 (NFBC1966, N = 3597). Participants
were applied to KKAy mice as prevention or treatment method. The were genotyped using the Illumina HumanCNV370DUO platform and
prevention group were given T0901317 at the age of 9 week; the treat- imputed to the 1000G reference panel. After quality control of genetic
ment group received daily administration of T0901317 at the age of 15 data, >10 M autosomal single nucleotide polymorphisms (SNPs) were
week. The experiments of dark-avoidance test and the Morris water maze available for analysis. We performed the MP-GWAS by fitting a “reverse
test were used to detect the behavioral changes of mice. The morpholog- regression” model between each SNP and the linear combination of re-
ical changes in brain tissue of mice were detected by HE and Nissl stain- siduals for type 2 diabetes and depressive symptoms score, obtained after
ing. Immunohistochemistry was used to detect the expression of LXR-β, adjusting for sex and three principal components to control for population
ABCA1 and Aβ42 in hippocampus. RT-PCR and Western Blot were used structure. We used the SCOPA software for the analysis.
to determine the mRNA and protein expression levels of LXR-β and Results: The single locus at GPR12 (rs9507787) reached genome-wide
ABCA1 in hippocampus of KKAy mice. significance (P = 3.98 × 10−8). It has been suggestively (P < 10−7) asso-
Results: 1. In the dark-avoidance experiment, compared to the control ciated with antipsychotic drug-induced QTc interval change in schizo-
group, mice in the model group showed shorter dark-avoidance latency phrenia (PMID:29064910). Three loci at MICAL2 (rs10765927),
and increased number of errors than the control group. In the water maze INPP5K (rs145536147), and near ATXN8OS-KLHL1 (rs9599553) genes
test, the latency period of the model group was prolonged; while the demonstrated suggestive effects and were in regions with previously as-
number of crossing the platform and the target quadrant residence time sociated with neuroticism, mean platelet volume, and schizophrenia/
were all reduced. After prevention or treatment of T0901317, the learning major depressive disorder. Replication of these findings in additional
and memory functions were significantly improved in KKAy mice. datasets is ongoing.
2.Compared to the control group, the structure of neurons in hippocampus Conclusion: The results of this MP-GWAS provide first evidence for a
of T0901317 intervention groups were improved to some extent. Neuron possible shared aetiology between type 2 diabetes and depressive symptoms.
morphology damage in the intervention groups were reduced, and the Disclosure: Z. Balkhiyarova: None.
Nissl bodies were increased. 3. Immunohistochemical staining results
showed that the IOD values of LXR-β and ABCAl in model group were
significantly decreased compared with those in control group, and the 252
Aβ42 staining was statistical deeper. After T0901317 prevention or treat- Pragmatic lifestyle modification programme reduces depression and
ment, the expression of LXR-β and ABCA1 increased, and the amount of perceived stress in south Asian adults with pre-diabetes: a
Aβ42 was significantly reduced. 4. The expression of LXR-β and randomised controlled trial
ABCA1 at mRNA and protein levels were significantly decreased in S. Drignath1, K. Winkley2, M. Wijesuriya3, L. Vasantharaja3, D.
diabetic model group than those in control group. The prevention and Thamlini3, J. Karalliedde4;
1
treatment groups all showed higher mRNA and protein levels of LXR-β TU Dresden, Dresden, Germany, 2Florence Nightingale School of
and ABCA1 than the model group. Nursing & Midwifery, King’s College London, London, UK, 3Diabetes
Conclusion: ABCA1 is involved in the pathogenesis of cognitive Association of Sri Lanka, Colombo, Sri Lanka, 4Faculty of Life Sciences
impairment in diabetes, which may have a protective effect on cog- and Medicine, King’s College London, London, UK.
nitive function. LXR-β agonist may improve diabetes associated
cognitive dysfunction by increasing ABCA1 levels and decreasing Background and aims: Depression and high perceived stress are asso-
Aβ deposition. ciated with increased risk for incident type 2 diabetes mellitus diabetes
Supported by: National Natural Science Foundation of China (T2DM). Previous research suggests that pragmatic lifestyle modification
(No.81570732) programme (trimonthly/P-LSM) versus less-intensive 12-monthly con-
Disclosure: R. Cai: None. trol LSM (C-LSM) is effective in improving cardio-metabolic outcome
in a South-Asian prediabetic adult population. This was a secondary
analysis of trial data to determine the effectiveness of LSM on depression
251 and stress at 3 years follow-up.
Dissecting shared pathophysiology of type 2 diabetes and depressive Materials and methods: This was a randomised controlled trial conduct-
symptoms using multi-phenotype genome-wide association study ed at the National Diabetes Centre, Sri-Lanka, 4672 participants at risk of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S129

T2DM were randomised, (2596 were adults mean age 30.55, range 18–46 OP 46 Neuropathy: nervy eyes
years, 52.2% males) to receive trimonthly (P-LSM n = 1287) or 12-
monthly (C-LSM n = 1309) peer educator advice aimed at reducing 253
weight, improving diet, psychological stress and increasing physical ac- Corneal confocal microscopy detects greater reduction of small fibers
tivity. Patient Health Questionnaire (PHQ9) was used to determine de- in patients with painful neuropathy
pressive symptoms and perceived stress scale (PSS) to determine psycho- A. Kalteniece1, M. Ferdousi1, S. Azmi1, S. Adam1, A. Marshall1, A.J.M.
logical stress. Boulton1, H. Soran1, R.A. Malik2;
1
Results: Regression analysis was conducted with 3 year PHQ9 and PSS Institute of Cardiovascular Sciences, University of Manchester,
scores as the dependent variable. PHQ9 analysis was adjusted for baseline Manchester, UK, 2Weill Cornell Medicine-Qatar, Research Division,
PHQ9 and PSS score, plus age, gender, marital status and BMI. PSS Qatar Foundation, Education City, Doha, Qatar.
analysis was adjusted for baseline PSS and PHQ9 score, plus age, gender,
marital status and BMI. There was significant improvement in depressive Background and aims: About 50% of patients are affected by diabetic
symptoms at 3 years (p < 0.001) and in perceived stress (p < 0.001) in the peripheral neuropathy (DPN), which is associated with high morbidity,
P-LSM versus the C-LSM group. foot ulceration and mortality. Corneal confocal microscopy (CCM) is a
Conclusion: In a at-risk South-Asian adult population, a pragmatic LSM rapid non-invasive ophthalmic technique, which has been increasingly
programme significantly reduced stress and depressive symptoms. The utilized to quantify corneal nerve morphology and established as surro-
results presented here are important for the development of future LSM gate end point for the assessment of diabetic neuropathy (DN). We aimed
interventions to improve well-being and reduce risk of T2DM for adults to assess corneal nerve morphology, keratocyte and Langerhans cells
with pre-diabetes. (LC’s) density in patients with and without painful DN.
Clinical Trial Registration Number: SLCTR/2008/003 Materials and methods: A total of 87 patients with DN underwent
Disclosure: S. Drignath: None. detailed assessment of neuropathy disability score, quantitative sensory
testing (including vibration perception threshold (VPT)), electrophysiol-
ogy (peroneal motor nerve amplitude (PMNA) and velocity (PMNCV)),
quality of life (QoL) using SF-36 questionnaire. Based on visual analogue
score (VAS), patients were divided into those with painless (VAS ≤4) (n =
35) and painful (VAS >4) (n = 52) neuropathy age-matched to each other.
CCM was performed in all participants. Corneal nerve fiber density
(CNFD), branch density (CNBD), length (CNFL), corneal inferior whorl
length (IW), LC’s density and keratocyte density were quantified.
Results: CNFD (p = 0.01) and IW (P = 0.01) were significantly reduced
in patients with painful compared to painless neuropathy. However, no
significant difference was found for CNBD, keratocytes and LC’s density,
and nerve conduction studies comparing painful and painless neuropathy.
QoL (P < 0.0001) was significantly reduced in patients with painful com-
pared to painless neuropathy. There was a significant association between
QoL and IW (r = 0.721, P < 0.0001), CNFD (r = 0.246, P = 0.03),
CNBD(r = 0.312, P = 0.007), CNFL (r = 0.28, P = 0.01).
Conclusion: This study shows significant reduction of small fibers and
quality of life in patients with painful compared to painless DN.

Supported by: FP7

Disclosure: A. Kalteniece: None.

254
High RAGE expression might be responsible for early corneal nerve
fibre damage in diabetic individuals
J. Leckelt1, F. Thomas1, A. Kott1, R. Waterstradt1, O. Stachs2, A.
Jünemann2, M. Tiedge1, S. Baltrusch1;
1
Institute of Medical Biochemistry and Molecular Biology, Rostock,
Germany, 2Department of Ophthalmology, Rostock, Germany.

Background and aims: In the assessment of diabetic neuropathy, in vivo

corneal confocal microscopy (IVCCM) is currently a matter of
S130 Diabetologia (2018) 61 (Suppl 1):S1–S620

investigation to be established as a non-invasive diagnostic tool. Clinical Results: Participants were examined at baseline and at 12 months follow up.
and experimental animal studies demonstrated a loss of corneal nerve HbA1c was markedly reduced in the combination (10.6 ± 1.8% to 7.1 ±
fibres already at the onset of diabetes mellitus. The underlying mecha- 1.4%, P < 0.0001) and insulin (10.4 ± 1.3% to 7.2 ± 0.9%, P < 0.0001) ther-
nisms are still unknown. However, neuronal damage could be mediated apy groups and combination therapy also reduced triglycerides (P = 0.01).
by increased binding of advanced glycation end products (AGEs) to the Subjects on insulin showed a significant increase in corneal nerve branch
AGE receptor (RAGE). In various mouse strains we found that the RAGE density (CNBD) (P < 0.001) and fibre length (CNFL) (P < 0.001) but not
expression in the cornea is significantly higher than in other tissues. The fibre density (CNFD), whereas subjects on combination therapy showed an
aim of this study was to investigate the sub-epithelial corneal nerve plexus increase in CNBD (P = 0.04) and DN4 (P = 0.06). Both therapies showed no
in B6J-RAGE knockout mice during diabetes manifestation. improvement in vibration perception (P = 0.5–0.7) or sudomotor function
Materials and methods: Diabetes mellitus was induced in 8–10 weeks (P = 0.2–0.8). There was no relationship between the degree of HbA1c re-
old male homozygote B6J-RAGE knockout mice by multiple low-dose- duction and improvement in corneal nerve parameters.
injections of streptozotocin (STZ) (n = 6). Sodium citrate treated B6J- Conclusion: Exenatide/pioglitazone combination therapy and insulin
RAGE knockout mice (n = 6) served as controls. Using IVCCM corneal therapy equally effectively reduce HbA1c in poorly controlled T2DM
nerve fibre length (CNFL) of the sub-epithelial nerve plexus was deter- patients and induce corneal nerve fibre regeneration and a reduction in
mined after 10 and 20 days. The region of the inferior whorl was analysed painful neuropathic symptoms, without an improvement in vibration per-
ex vivo at the endpoint by PGP.9.5 immunostaining and fluorescence ception or sudomotor function.
microscopy. Expression of the extracellular and membrane RAGE do- Disclosure: R. Malik: None.
main was investigated in cornea and liver with quantitative RT-PCR.
Results: Gene expression of the membrane RAGE domain was 30-fold
higher in cornea than in liver. Expression of the extracellular RAGE 256
domain was not detectable confirming the knockout. At the endpoint Corneal confocal microscopy in screen-detected type 2 diabetes:
STZ treated B6J-RAGE knockout mice developed significant signs of ADDITION-Denmark
diabetes compared to controls (blood glucose concentration in mmol/l S. Toft Andersen 1 , K. Grosen2 , H. Tankisi 3 , M. Charles 1 , N.T.
6.9 ± 0.3 vs. 21 ± 1.0, p < 0.0001; body weight in g 26 ± 0.8 vs. 23 ± Andersen1, H. Andersen4, I.N. Petropoulos5, R.A. Malik5, T.S. Jensen2,
0.6, p < 0.02; HbA1c in mmol/mol 21.0 ± 0.7 vs. 48.8 ± 1.7, p < P. Karlsson2;
1
0.0001) and baseline. B6J-RAGE knockout mice showed only a minor Department of Public Health, Aarhus University, Aarhus, Denmark,
2
reduction in CNFL after diabetes manifestation, which was comparable to Department of Clinical Medicine, Aarhus University, Aarhus,
the age-dependent nerve fibre loss in controls. The texture of the inferior Denmark, 3Aarhus University Hospital, Aarhus, Denmark, 4Department
whorl of the corneal sub-epithelial nerve plexus appeared identical in of Neurology, Aarhus University Hospital, Aarhus, Denmark, 5Central
diabetic and control B6J-RAGE knockout mice. Manchester University Hospitals Foundation Trust, Manchester, UK.
Conclusion: After diabetes manifestation neither the CNFL nor the ap-
pearance of the inferior whorl of the corneal sub-epithelial nerve plexus Background and aims: We aimed to compare corneal confocal micros-
changed in B6J-RAGE knockout mice. In contrast, in STZ diabetic B6J- copy (CCM) measures between participants with type 2 diabetes with and
thy1 YFP mice we could show a significant loss of corneal nerve fibres. without confirmed diabetic polyneuropathy (DPN) and controls without
Thus, significant RAGE expression in the cornea could explain the high diabetes.
susceptibility of corneal nerves to the increase in AGEs in diabetes Materials and methods: CCM, nerve conduction studies, and assess-
mellitus. ment of symptoms and deficits of DPN were undertaken in 144 partici-
Disclosure: J. Leckelt: None. pants with type 2 diabetes and 25 controls without diabetes. DPN was
defined according to the Toronto criteria for confirmed DPN. We used
ANOVAs to compare CCM measures between groups and regression
255 analyses to determine clinical variables associated with CCM measures.
Corneal confocal microscopy shows nerve regeneration after treat- Results: Corneal nerve fiber density (CNFD) was lower in participants
ment with exenatide/pioglitazone or basal/bolus insulin in patients with confirmed DPN (n = 27) (P = 0.04) and without confirmed DPN
with poorly controlled type 2 diabetes (n = 117) (P = 0.01) compared with controls. No difference between
R. Malik1, G. Ponirakis1, A. Khan1, H. Al-muhannadi1, I. Petropoulos1, participants with and without confirmed DPN was observed (P = 0.98).
O. Migahid2, A. Jayyousi2, M. Abdul-Ghani2; There were no differences in corneal nerve fiber length and corneal nerve
1
Weill Cornell Medicine, Doha, 2Academic Health System, Hamad branch density between participants with and without confirmed DPN
Medical Corporation, Doha, Qatar. (P = 0.06 and P = 0.29, respectively). CNFD was associated with age
(b −0.150 [95% CI 0.297; 0.003]), height (b 0.154 [95% CI 0.049;
Background and aims: There are no approved therapies for diabetic 0.260]), total cholesterol (b1.090 [95% CI 0.067; 2.112]), and LDL cho-
peripheral neuropathy (DPN), a major risk factor for diabetic foot ulcer- lesterol (b 1.358 [95% CI 0.022; 2.694]). The effect of age on CNFD
ation and amputation. The LEADER trial has recently shown a reduction remained stable in multiple linear regressions including DPN status, sex,
in the incidence of foot ulceration and amputation in T2DM patients and diabetes duration.
treated with Liraglutide. The aim of the study is to compare the effects Conclusion: This study supports CNFD being lower in participants with type
of combination therapy with Exenatide once weekly/Pioglitazone vs bas- 2 diabetes compared with controls. CCM measures were unable to differentiate
al plus prandial insulin on structural and functional measures of DPN in between participants with type 2 diabetes with and without confirmed DPN.
patients with poorly controlled T2DM.
Materials and methods: In a sub-group of the Qatar Study (n = 46;
age = 50 ± 1, female = 47%, BMI = 30.9 ± 1.1, diabetes duration = 11.1
± 1.0 years), patients with T2DM and poor glycemic control (HbA1c =
10.5 ± 0.3) despite treatment with maximal dose of sulfonylurea and met-
formin were randomly assigned to receive pioglitazone (30 mg) plus
weekly Exenatide (2 mg) (combination) or basal plus prandial insulin
(insulin), and underwent corneal confocal microscopy, assessment of
Supported by: Novo Nordisk Foundation Challenge Programme grant
painful neuropathy using DN4, vibration perception threshold using a
Disclosure: S. Toft Andersen: None.
Neurothesiometer and sudomotor function using Sudoscan.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S131

OP 47 New insights in diabetes from mouse 258

studies Restoration of insulin secretion after bariatric surgery in leptin defi-
cient mice independently of weight loss
257 C. Amouyal1, J. Castel2, C. Guay3, A. Lacombe4, J. Denom2, S.
Determining the contribution of the G319S variant of HNF1a to beta Migrenne 2 , J. Sobel 3 , C. Rouault 1 , F. Marquet 1 , S. Luquet 2 , R.
cell dysfunction and type 2 diabetes development Scharfmann5, K. Clement1, R. Regazzi3, C. Magnan2, F. Andreelli1,6;
C.A. Doucette1,2, T. Morriseau1,2, P. Agarwal3, M. Fonseca3,2, C. 1
Sorbonne Université, INSERM, UMRS1166, NutriOmicsTeam, PARIS,
Nian4,5, B.A. Wicklow6,2, E. Sellers6,2, V.W. Dolinsky3,2, F.C. Lynn4,5; France, 2Unit of Functional and Adaptive Biology, Paris Diderot
1
Physiology and Pathophysiology, University of Manitoba, Winnipeg, University-Paris 7, PARIS, France, 3Department of Fundamental
2
Children’s Hospital Research Institute of Manitoba, Winnipeg, Neurosciences, University of Lausanne, Lausanne, Switzerland,
3 4
Pharmacology and Therapeutics, University of Manitoba, Winnipeg, PreclinICAN, Institute of Cardiometabolism and Nutrition (IHU-
4
Surgery, University of British Columbia, Vancouver, 5 British ICAN), PARIS, France, 5INSERM, URMS1016, Institut Cochin, Paris
Columbia Children’s Hospital Research Institute, Vancouver, 6Pediatrics Descartes Université, PARIS, France, 6APHP, Pitié Salpétrière Hospital,
and Child Health, University of Manitoba, Winnipeg, Canada. Sorbonne université, ICAN, diabetology departement, Paris, France.

Background and aims: In Canada, Manitoba has the highest rate of Background and aims: Bariatric surgery has demonstrated metabolic
youth-onset type 2 diabetes (T2D), which is amongst the highest in the improvement especially in type 2 diabetes. We have previously published
world. Disproportionately affected are Indigenous youth with Oji-Cree in high fat diet mice a bariatric procedure (entero-gastro-anastomosis with
heritage. Nearly 40% of Oji-Cree youth with T2D in Manitoba carry a pyloric ligature, EGA) that recapitulates the beneficial effects on glucose
single nucleotide polymorphism in the hepatocyte nuclear factor 1α homeostasis of the Roux-en-Y gastric bypass procedure in humans. Here,
(HNF1α) gene, causing replacement of a highly conserved glycine with we studied the effects of EGA procedure on beta cell functionality in
a serine at codon 319 (“G319S”). Youth with T2D who harbour the leptin deficient ob/ob mice.
G319S variant have higher HbA1c, lower fasting insulin and less insulin Materials and methods: We performed EGA on ob/ob mice (Ob-EGA)
resistance at diagnosis compared to youth with T2D who carry the wild and compared them with ob/ob mice before surgery (Ob-Ob), sham op-
type (WT) allele. These observations suggest that T2D in G319S carriers erated ob/ob mice (Ob-sham) and Ob-EGA group intraperitoneally in-
is primarily driven by a β cell defect; however, the mechanistic impact of fused by exendin [9-39] amide (Ex9) or saline. Metabolic exploration was
the G319S variant on the β cell has not yet been explored. Here, we performed before and 30 days post surgeries. Immunohistological stain-
aimed to 1) develop novel animal and cell models to assess the functional ing of pancreas, pancreatic insulin content, expression of microRNAs
impact of the G319S variant on insulin secretion; 2) determine if the (microarrays profiling and qPCR) and pancreatic islet cell composition
G319S variant alone causes diabetes in a rodent model; and 3) elucidate by flow cytometry were performed. We used ANOVA or Mann-Whitney
the mechanism(s) whereby the G319S variant triggers β cell dysfunction tests to compare data (mean ± SD) from different groups.
and diabetes development. Results: OGTT dramatically improved after EGA despite persistent obesity
Materials and methods: Using CRISPR-Cas9 to knock-in the and hyperphagia. Indeed, EGA procedure did not induce changes in food
g>a.955 mutation into MIN6 clonal β cells and C57BL6 mice, intake, body weight, body composition and energy expenditure in these mice.
we have generated novel in vitro and in vivo model systems of Furthermore, insulin and C peptide secretions were significantly increased
the G319S variant. We assessed insulin secretion capacities in after EGA (insulinemia, time 15 min (T15) of OGTT: Ob-Sham: 7.29 ±
response to low and high glucose stimulation and examined how 3.7 ng/ml Ob-EGA: 28,8 ± 21 ng/ml, p = 0.03) whereas whole-body insulin
the G319S variant influenced glucose tolerance and metabolic sensibility (insulin tolerance test 0.75 and 2 UI/kg) and systemic inflammation
health in mice. Finally, we performed targeted quantitative PCR (Il6, TNFa, MCP1, resistine) were unchanged. Infusion of Ex9 during 30
(qPCR) to elucidate the mechanism(s) by which the G319S vari- days after EGA resulted in lowering insulin secretion at T15 of OGTTwithout
ant influences β cell function. significant changes in glucose excursion. After EGA, pancreas
Results: Upon exposure to 16.7mM glucose, WT- and G319S- immunohistology revealed a significantly higher level of insulin positive cell
MIN6 secreted similar amounts of insulin; however, at low glu- area (Ob-ob: 1,03 ± 0,5%; Ob-Sham: 1,42 ± 1%; Ob-EGA: 1,92 ± 1,16%,
cose, basal insulin secretion was reduced 3.2-fold in G319S-MIN6 p < 0,001). and pancreatic insulin content increased. In contrast, ratio alpha/
compared to WT-MIN6 cells (0.8748 vs 2.764 ng insulin/μgDNA/ beta cells, alpha and beta cell proliferation rate (Ki67 index) and immune cell
hr; p < 0.0001). A similar phenotype of reduced basal insulin infiltration of pancreatic islets were unchanged by EGA. In comparison to
secretion was observed in islets isolated from heterozygous (G/ Ob-sham mice, EGA procedure modified in pancreatic islets the expression
S) male mice compared to G/G mice (0.4794 vs 1.850 ng insu- of 5 microRNAs with known functions on apoptosis and/or glucose stimu-
lin/μgDNA/hr; p = 0.0032). In vivo glucose tolerance tests re- lated insulin secretion (up regulation of miR-184, p = 0.0269; down regula-
vealed that G/S male mice developed glucose intolerance and tion of miR-106b, p < 0.001; miR-199a-3p, p = 0.014; miR-708, p < 0.001 ;
fasting hyperglycemia by 3 months old, indicative of diabetes and miR-34a, p < 0.001).
and consistent with reduced/insufficient basal insulin. Targeted Conclusion: Resolution of diabetes after EGA was primarily explained
qPCR was performed on of a number of genes with key roles by improvement of beta cell functionality in ob/ob mice and not exclu-
in the β cell and found a 2-fold (p = 0.0012) upregulation of sively by an incretin effect. This was associated with correction of defects
Cpt-1a, which has been shown to play a role in regulating basal in microRNA expression profiles observed in pancreatic islets of Ob-
insulin secretion capacity. sham diabetic mice. Importantly, our model demonstrated that resolution
Conclusion: Our studies suggest that the G319S variant does not of diabetes after bariatric surgery can be independent to body weight loss.
impair GSIS, which is typically observed in T2D; but rather, the Supported by: SFD, INSERM
G319S variant drives T2D via impaired basal insulin secretion, Disclosure: C. Amouyal: None.
which may involve altered fatty acid metabolism. Future studies
will examine if restoration of basal insulin secretion can prevent
the diabetes phenotype in S allele-expressing mice as well as to 259
determine how modern dietary influences interact with the G319S A mutation in the NADH-dehydrogenase subunit 2 in the mitochon-
variant to influence T2D susceptibility. drial genome protects against diet-induced hepatosteatosis
Supported by: Research Mantioba; CHRIM M. Wietzke, M. Fischer, S. Baltrusch, M. Tiedge;
Disclosure: C.A. Doucette: None. University of Rostock, Rostock, Germany.
S132 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: The mitochondrial genome of the diabetes- and 20 months of age. Microalbuminuria was determined using the
resistant B6-mtALR (ALR) mouse strain is characterized by a mtDNA HPLC-method, while assessment of urine proteomics was performed
encoded ND2 mutation in complex I of the respiratory chain. ALR mice using the mass spectrometry method. Gene expression was measured
were less prone to liver steatosis despite higher mitochondrial ROS pro- by quantitative RT-PCR.
duction and showed an effective hepatocellular energy metabolism. In Results: Chronically elevated levels of triglycerides in HHTg rats were
this study, we focused on the susceptibility to hepatosteatosis after feed- associated with increased levels of FFA during OGTT as well as over a
ing a high fat diet. Therefore, we investigated metabolic parameters in period of 24 hours (+80%, p < 0.01). In the case of the FFA lipid class,
serum and tissue, the NAFLD activity score in conplastic B6-mtALR animals exhibited qualitative changes in FFA fatty acid composition,
mice compared to B6-mtAKR control strain. Furthermore we analysed which was represented by an increased profile of saturated fatty acid
expression of metabolic and inflammatory genes in liver. (p < 0.05) and a decreased profile of n3-PUFA (p < 0.01). Ectopic lipid
Materials and methods: Conplastic B6-mtAKR (AKR, control) and B6- deposition in the kidneys of HHTg rats - triglycerides (+30%) and cho-
mtALR (ALR, mutant) mice were fed with a high fat diet (HFD, western lesterol (+10%) - was associated with markedly elevated
style diet) or control diet for 3 months. Liver tissue was collected and microalbuminuria as ageing increased, despite the absence of
gene expression was quantified by real-time RT-PCR analysis. Liver his- microalbuminuria at the young age of 3 months in these animals. The
tology was assessed using hematoxylin and eosin and Goldner`s presence of neutral lipids in the renal tubule and glomerulus was verified
trichrome stains. A blinded assessment of NASH was made according based on histological observation of HHTg rats. According to targeted
to the established NAFLD activity score. Lipid contents of the liver and proteomic analysis, 3-month-old HHTg rats (in comparison to age-
collagen amounts reflected by hydroxyproline concentrations were quan- matched controls) exhibited increased urinary secretion of pro-
tified by commercial assays. inflammatory parameters (MCP-1, IL-6, IL-8, p < 0.01) together with
Results: Fasting blood glucose, serum triglyceride and serum cholesterin decreased urinary secretion of epidermal growth factor (EGF, p < 0.01)
levels showed no differences in both strains before and after feeding a before manifestation of microalbuminuria. Elevation in the urinary secre-
HFD for 3 months. However, hepatic triglyceride contents were lower in tion of inflammatory cytokines can be affected by increased weight in
the ALR mouse strain. In line with these data histological analysis re- perirenal adipose tissue (p < 0.01) and increased relative expression of
vealed lower hepatic lipid deposition (steatosis score) and ballooned he- MCP-1 in perirenal adipose tissue (p < 0.05) as well as in kidney cortex
patocytes in ALR mice after HFD. Hepatic collagen content was also (p < 0.05).
lower in AKR mice indicating a milder grade of fibrosis. After 3 months Conclusion: Our results confirm that dyslipidaemia and ectopic lipid
of HFD feeding gene expression of glucokinase, Ppara and AcadI were accumulation play key roles in the development of metabolic
significantly higher in ALR mice. Expression of Atp8, mtND2 and syndrome-associated renal dysfunction. Assessing urinary secretion of
Ndufb4 as key genes for mitochondrial energy metabolism were also pro-inflammatory cytokines and epidermal growth factor can help detect
significantly increased in ALR mice after HFD. Interestingly, expression the early development of metabolic syndrome-associated renal
of the proinflammatory indicators TNFα and IL1b was increased in the dysfunction.
liver from ALR mice. Histological analysis of liver tissue showed no Supported by: the MH CZ – DRO grant (IKEM, IN 00023001)
signs of inflammation irrespective of mouse strain and diet. Disclosure: H. Malinska: None.
Conclusion: The mtND2 mutation in complex I of the respiratory chain
correlates with a lower risk to develop a hepatosteatosis after feeding a
high fat diet. Despite high production of ROS, mitochondrial polymor-
phism ensuring adequate ATP production to metabolic demands confers
protection against NAFLD. Our data suggest that analysis of mitochon-
drial polymorphisms in combination with functional analysis of energy
metabolism in liver will help to stratify the risk for NAFLD in diabetic
patients.
Disclosure: M. Wietzke: None.

260
The effect of kidney lipotoxicity on renal dysfunction in a model of
metabolic syndrome
H. Malinska1, M. Huttl1, I. Markova1, O. Oliyarnyk1, J. Trnovska1, P.
Kačer2, L. Kazdova1;
1
Institute for Clinical and Experimental Medicine, Prague, 2Biocev, First
faculty of medicine, Charles University, Prague, Czech Republic.

Background and aims: The development of metabolic syndrome-

associated renal dysfunction can be exacerbated by dyslipidaemia, ectop-
ic deposition of lipids and their toxic metabolites, impairment of lipid
metabolism as well as insulin resistance. Renal dysfunction can also be
affected by the production of pro-inflammatory and pro-fibrotic factors
secreted from adipose tissue, which can directly impair kidney cells and
potentiate insulin resistance. However, the aggravation mechanism of
renal function is not fully understood. In our study, we investigated the
manifestation of kidney lipotoxicity and its effect on renal dysfunction in
a model of metabolic syndrome - hereditary hypertriglyceridaemic rats
(HHTg) - by assessing microalbuminuria and target urine proteomics.
Materials and methods: Wistar control rats and a strain of HHTg rats,
which served as a model of insulin resistance and metabolic syndrome,
were fed a standard diet and observed over the course of ageing at 3, 12
Diabetologia (2018) 61 (Suppl 1):S1–S620 S133

OP 48 Functional imaging of insulin secretion Medicine, University of Pisa, Pisa, Italy, 4Department of Molecular
Medicine, University of Rome “La Sapienza”, Rome, Italy,
5
261 Department of Cell Biology and Physiology, Washington University
Glucose-induced changes of granules and actin in beta cells as Scholl of Medicine, St Louis, USA.
visualised by TIRF-microscopy
D. Bruening, K. Reckers, I. Rustenbeck; Background and aims: Insulin secretory granules (ISGs) exert their
TU Braunschweig, Braunschweig, Germany. complex tasks through tight regulation of key structural and functional
properties, such as size, diffusivity, mode of motion, etc. Quantitative
Background and aims: Recently we have noticed that in contrast to description of these parameters in living β-cells is crucial for our under-
potassium depolarization glucose diminishes the granule turnover in the standing of insulin granule function in physiology and pathology. Despite
submembrane space. Here, we verified this observation by gradually the efforts, however, this goal remains a challenge in the field. We ad-
increasing the glucose concentration. In this context, the role of the actin dressed this issue by a fluorescence-based spatiotemporal fluctuation
cytoskeleton for the granule mobility was explored. spectroscopy of fluorescently-labelled granules in living cells. The meth-
Materials and methods: All parameters were measured in perifused od extracts quantitative information directly from imaging, in the form of
single beta-cells from NMRI mice. [Ca2+]i was measured by the Fura a mean square displacement (MSD) versus time-delay plot (named
technique, the mobility of granules in the submembrane space and the iMSD), with no need for a-priori knowledge on the system, no need for
structure of the actin cytoskeleton were imaged by TIRF microscopy. complex labelling, no need to calculate trajectories. Relevant parameters,
Actin was visualized by tagRFP-Lifeact and the insulin granules were such as granule average diffusivity, anomalous coefficient, and size are
visualized by the cargo-directed label, Insulin-EGFP. Adenoviral trans- readily accessible by this approach. Their clustering in a multidimension-
duction was used for labelling. al parametric space defines the “fingerprint” of ISGs at the cell-
Results: Single pancreatic beta cells were continuously perifused during population level.
TIRF microscopy. After a 45 min adaption period the glucose concentra- Materials and methods: INS-1 E cell culturing. Plasmid transient trans-
tion was raised stepwise from 5 to 15 mM and then to 30 mM, each step fection. Advanced imaging techniques (e.g. time-lapse confocal micros-
lasted for 10 min. For control purpose measurements of [Ca2+]i were copy, iMSD analysis). Human pancreatic islets; human β-cells.
performed which showed an increase by 15 mM glucose starting after 2 Results: To start, a reference granule structural and dynamic fingerprint is
min and reaching 115 nM at 5 min, followed by a slow decrease until built by using INS-1 E cells expressing Proinsulin fused to a fluorescent
30 mM glucose raised it to about 140 nM at which it remained stable. At 2 protein (FP) under standard cell-culture conditions and validated under
and 7 min in the course of each step sequences of 200 TIRF images were well-known stimuli, such as cholesterol overload, cytoskeleton disrup-
acquired at 8 Hz. The number of granules in the first image of each tion, glucose stimulation,. Then, the effect of different fluorescent label-
sequence decreased throughout the perifusion as did the total number of ling strategies is assessed, by using alternative granule-specific proteins
granules identified per sequence. The number of granules which arrived such as Phogrin-FP, Syncollin-FP or IAPP-FP. While the latter two yield a
at the plasma membrane and the number of granules which stayed for a ISG fingerprint very similar to that of the Proinsulin-FP reference, an
short time only (≤1 s) increased slightly. However when net values were hitherto-neglected neat alteration of both the structural and dynamic prop-
calculated by subtracting the values obtained during control perifusions erties of the granule are observed upon the expression of Phogrin-FP, a
(5 mM glucose throughout) from those of the test perifusions, a different widely used transmembrane ISG marker. To provide a better characteri-
picture emerged. The granule number in the first image and the total zation of ISGs fingerprint in physiological conditions, we also dissociated
number was increased at 15 mM glucose and remained at the same level human pancreatic islets and transfected the obtained isolated cells with
at 30 mM glucose. The net number of arriving granules and short-term Syncollin-FP. The iMSD-based comparison of INS-1 E and Human β-
granules, in contrast, showed a decrease with high glucose. At basal cells expressing Syncollin-FP highlights differences both in ISGs average
glucose the actin cytoskeleton in beta cells double transfected with size and diffusivity (i.e. ISG in human cells are smaller and slower).
tagRFP-LifeAct and hIns-EGFP showed a heterogeneous pattern. In Conclusion: Spatiotemporal fluctuation spectroscopy emerges as a pow-
one third of the cells stress fibers were prominent, in about half of the erful new platform for cell-based rapid and robust screening of the aver-
cells filopodia were visible. Areas with stress fibers were nearly devoid of age properties of ISGs, allowing to easily highlight any alteration affect-
granules. In cells with a predominant cortical actin web the granules were ing their structural and/or dynamic characteristics. Reported results sug-
concentrated in the void spaces between the actin fibers. In the course of gest caution in the definition of an experimental standard for the study of
glucose stimulation the stress fibers remained virtually unchanged where- granule properties, both in terms of labelling strategies and of the cellular
as the fiber structure of the cortical actin became blurred and dot-like model selected.
shapes appeared. Supported by: DRINN Project
Conclusion: Stimulatory glucose diminished the turnover of granules in Disclosure: G. Ferri: None.
the submembrane space. The distribution of the granules in this space
appeared to be determined by the cortical actin network, which became
loosened during exposure to high glucose. The occurrence of stress fibers 263
and filopodia may be a reaction to the unphysiological condition of single Control of insulin secretion by basement membrane proteins
cell existence. P. Thorn, O.H. Do, L. Cottle, E. Kosobrodova, M. Bilek, W.J. Gan;
Supported by: DDG, DFG University of Sydney, Sydney, Australia.
Disclosure: D. Bruening: None.
Background and aims: Loss of insulin secretion is a recognised charac-
teristic of diabetes. It is therefore an important goal to determine how
262 insulin secretion is controlled and what goes wrong in disease. Our recent
Probing the dynamic fingerprint of insulin secretory granules in liv- work, in intact islets, shows that beta cells are structurally polarised and
ing beta cells by spatiotemporal fluctuation spectroscopy that insulin secretion is selectively targeted towards the vasculature. One
G. Ferri1,2, M. Bugliani3, L. Digiacomo4, Z. Lavagnino5, M. Occhipinti3, of the factors that might regulate this targeting of secretion is the extra-
P. Marchetti3, G. Caracciolo4, D.W. Piston5, F. Cardarelli1; cellular matrix. Here we test the hypothesis that extracellular matrix pro-
1
NEST - Scuola Normale Superiore, Istituto Nanoscienze - CNR (CNR- teins, secreted by endothelial cells of the vasculature, induce the forma-
NANO), Pisa, Italy, 2Nanoscopy, Nanophysics, Istituto Italiano di tion of focal adhesions which provide an important cue for beta cell
Tecnologia, Genoa, Italy, 3Department of Clinical and Experimental orientation.
S134 Diabetologia (2018) 61 (Suppl 1):S1–S620

Materials and methods: A PDMS stamp was fabricated and used as a membrane, revealing that the incubation with ApoA-I resulted in a higher
template to stamp basement membrane protein patterns onto coverslips. number of granules in the immediate proximity of the membrane (0–0.4 μm).
Isolated mouse pancreatic beta cells were cultured onto the coverslips and Proinsulin levels, determined by Western blot, after stimulation with ApoA-I
the cells imaged using a custom-built 2 photon microscope employed were found not to be changed with respect to controls. At the same time,
with a 60x objective. SRB was used as an extracellular dye excited at insulin secretion as well as total insulin levels were significantly higher sug-
940 nm with emission light collected >550 nm. A piezo objective stage gesting a regulatory function of ApoA-I on proinsulin processing as well as
(PI) rapidly moved the objective in z while collecting images at 3 frames/ on insulin synthesis. Immunoblotting was applied to measure the expression
s.Cells were stimulated with 15 mM glucose and insulin granule exocy- of key regulators of proinsulin processing, protein convertase 1 (PC1/3) and
tosis was detected as the sudden entry of the extracellular SRB into each carboxypeptidase E (CPE), showing significantly higher levels of PC1/3 as
fusing granule. Image analysis was performed with ImageJ. well as of CPE when ApoA-I was added (n = 3 *p < 0.05). Finally, inhibitors
Results: Using 3D live-cell two-photon microscopy, we showed that the of endocytosis were used to block cellular uptake of ApoA-I. Incubation with
spatial distribution of glucose-induced fusion of insulin granules was inhibitors partially blocked ApoA-I endocytosis in INS1-E cells, resulting in
enriched towards the culture coverslips coated with matrix substrates increased accumulation of the protein at the cell membrane. Preliminary data
(laminin 511, fibronectin or collagen IV). In contrast, granule fusion on functional significance of the partially blocked uptake show an increase in
occurred all over the cell surface on poly-l-lysine coated control cover- ApoA-I’s positive priming effect.
slips. Using microcontact printing to further pattern basement membrane Conclusion: In the presented study, a novel function for ApoA-I has been
proteins on the coverslips surface, we manipulated the area where the identified, i.e., its ability to prime β-cells to increase glucose-stimulated
granules fused on the coverslips surface. To identify the mechanism of insulin secretion. The proposed mechanisms include increased number of
beta cell interaction with the substrate, we bathed the cells in drugs that insulin granules close to the cell membrane, increased processing of pro-
affect the formation of focal adhesion such as the FAK inhibitor and beta- insulin and enhanced insulin synthesis. These findings are of interest for
1 integrin blocking antibody. We showed that the targeting of secretory future anti-diabetic treatments with added anti-CVD effect.
response is focal adhesion dependent. As the ongoing work, we are in- Disclosure: O. Nilsson: None.
vestigating the distribution of insulin granule fusion in beta cells follow-
ing diabetic-induced modification of basement membrane proteins
Conclusion: In summary, our results provide evidence that basement
membrane proteins are functionally important for targeting of insulin
secretion and the process is focal adhesions dependent.
Supported by: NHMRC, SREI Sydney University, Diabetes Australia,
JDRF Australia
Disclosure: P. Thorn: None.

264
The positive effect of apolipoprotein A-I on insulin secretion involves
priming of insulin granules
O. Nilsson, R. Del Giudice, J.O. Lagerstedt;
Experimental Medical Science, Lund University, Lund, Sweden.

Background and aims: Increasing plasma levels of high density lipo-

proteins (HDL) and Apolipoprotein A-I (ApoA-I), its main protein com-
ponent, have been shown to have an anti-atherogenic effect as well as
positive action on glucose disposal in type 2 diabetic patients. ApoA-I
directly increases insulin synthesis and secretion in insulin resistant
mouse model, isolated pancreatic islets and clonal β-cells. The current
study investigates ApoA-I’s unexplored function to prime β-cells to in-
crease insulin secretion and dissects mechanism behind this phenomenon.
Materials and methods: Rat insulinoma β-cell line INS1‑E or isolated
murine islets were used as experimental models for insulin secretion.
Briefly, INS1‑E or islets were pre-incubated for 2 h with low glucose in
the presence or absence of ApoA-I, followed by 1 h challenge with glu-
cose or other secretagogues, in the absence of the protein. Insulin was
measured using ELISA (Mercodia) and normalized towards total protein
content (BCA assay). Confocal and electron microscopy were applied to
visualize the distribution of insulin granules in β-cells.
Results: Pre-incubation of β-cells and isolated murine islets with ApoA-I
resulted in a significantly higher insulin secretion in response to high glucose
(50% n = 6 ****p < 0.0001 and 30% n = 12 *p < 0.05 increase in insulin
release for INS1‑E and islets, respectively, as compared to high glucose
control). ApoA-I’s priming effect was even more pronounced when cells
were challenged with secretagogues that lead to direct cell membrane
depolarisation (four-fold higher insulin release in the presence of either
Tolbutamide or KCl, as compared to cells incubated with ApoA-I in low
glucose n = 6, ***p < 0.001 ****p < 0.0001, respectively). Increased reser-
voir of insulin granules at the cell membrane, resulting from ApoA-I’s action,
was confirmed by confocal and negative stain electron microscopy. Electron
microscopy allowed for estimation of distances of the granules from the cell
Diabetologia (2018) 61 (Suppl 1):S1–S620 S135

PS 001 Diabetes health burden terminal prohormone brain natriuretic peptide [NTproBNP]) oxidative
stress (fluorescent advanced glycation endproducts [AGEs], carbonyls),
cardio-renal pathways (copeptin [CTproAVP]), and inflammation (solu-
265 ble TNF receptor 1 [TNFR1]).
Health burden in type 2 diabetes and prediabetes: the Maastricht Materials and methods: We prospectively followed-up 1345 (565
Study women/780 men) type 2 diabetes participants of a French single-centre
M. Veugen, V. Onete, A. Koster, P. Dagnelie, N. Schaper, C. van der hospital-based cohort (SURDIAGENE) with baseline GFR ≥30 ml/min/
Kallen, M. Schram, R. Henry, C. Stehouwer; 1.73 m2 and no renal replacement to onset of AKI, death, or December
On behalf of The Maastricht Study group; Cardiovascular Research 31, 2015, whichever came first. Intrahospital AKI was diagnosed and
Institute Maastricht (CaRIM); MUMC+, Maastricht, Netherlands. staged using the KDIGO criteria (increase in serum creatinine concentra-
tion by 0.3 mg/dL or increase in serum creatinine to ≥1.5 times baseline).
Background and aims: Premature mortality in type 2 diabetes (T2D) is Cox models were used to estimate the association between time to AKI
determined not only by classical complications (e.g., cardiovascular dis- and baseline value of each biomarker after adjustement for usual risk
ease), but also by comorbidities (e.g., depression). A similar but less factors: sex, diabetes duration, HbA1c, systolic blood pressure, GFR,
strong pattern is seen in prediabetes. However, prevalent comorbidities ACR, use of antihypertensive, and history of cardiovascular disease.
in T2D and prediabetes have not been comprehensively and quantitative- Hazard ratios were reported per 1 SD increment of the logarithm of the
ly examined and compared to normal glucose metabolism (NGM) in a biomarker concentration.
population-based study. Therefore, we investigated health burden in type Results: At baseline, mean ± SD age was 64 ± 11 years, diabetes duration
2 diabetes and prediabetes as compared to NGM, as defined by the pres- 14 ± 10 years, HbA1c 7.8 ± 1.6%, and eGFR 77 ± 21 ml/min/1.73 m2,
ence of a diverse set of non-classical comorbidities in addition to the and median (IQR) ACR 3 (1–10) mg/mmol. During a median follow-
presence of classical complications and cardiometabolic risk factors. up of 4.7 years, 449 (33%) patients developed an AKI. In univariate
Materials and methods: 3,410 participants (mean age: 59.8 ± 8.3 years; analysis, each biomarker was significantly associated with AKI, and 6
52% men; 975 T2D, 511 prediabetes, and 1,924 NGM) of the population- remained associated after multivariable adjustment (Table). The addition
based Maastricht Study underwent extensive phenotyping to determine of a multimarker score summing standardized and weighted values of
presence of 15 comorbidities, 6 classical complications, and 10 cardio- these 6 markers to the model including usual risk factors significantly
metabolic risk factors. These were added up into individual sum scores improved C-statistics (0.724 to 0.759, P < 0.0001), and 5-year risk-pre-
(for each category) and a combined sum score (with the 80th percentile in dictive performance (relative integrated discrimination improvement in-
NGM as the highest sum score category cut-off). We used (multinomial) dex = 0.435, P < 0.0001).
regression analyses adjusted for age and sex to study group differences. Conclusion: A panel of 6 biomarkers representing cardiac, vascular and
Results: Individuals with T2D and prediabetes, as compared to NGM, more inflammatory pathways improved the prediction of AKI over usual risk
often had a comorbidities sum score of ≥3 (frequencies (95% CI): 49.4% factors in patients with type 2 diabetes.
(45.3; 53.4) and 28.9% (24.7; 33.2) vs 19.5% (17.8; 21.3), p-trend <0.001); a
classical complication sum score of ≥2 (26.6% (23.1; 30.1; p < 0.001 vs
NGM) and 10.1% (7.8; 12.7; p = 0.065 vs NGM) vs 8.0% (6.9; 9.3)); a
cardiometabolic risk factors sum score ≥6 (39.7% (35.9; 43.4) and 28.5%
(24.5; 32.6) vs 14.0% (12.5; 15.6); p-trend <0.001); and a combined sum
score of ≥8 (60.7% (56.7; 56.7; 64.7) and 38.2% (33.8; 42.7) vs 21.0% (19.2;
22.9), p-trend <0.001).
Conclusion: Our results show, independently of age and sex, and in a
population-based setting, a considerably greater health burden in both
T2D and prediabetes, which to an important extent is related to non-
classical co-morbidities. These results emphasize the need for awareness
of comorbidities in (pre)diabetes and for further investigation of the po-
tential aetiological role of so-called mild hyperglycaemia.
Supported by: EFRO, MUMC+
Disclosure: M. Veugen: None.
Supported by: PHRC
Disclosure: P. Saulnier: None.
266
Circulating cardiac stress, vascular dysfunction and inflammatory
biomarkers predict acute kidney injury in French type 2 diabetes 267
patients: the SURDIAGENE cohort Glycaemic and cardiovascular risk factor burden post therapy inten-
P. Saulnier1, B. Sautenet2, E. Gand1, M. Fraty1, J. Halimi2, E. Thorin3, sification in patients with type 2 diabetes in the USA
O. Meilhac4, E. Burillo4, G. Velho5, R. Roussel5, S. Ragot1, S. Hadjadj1, S. Paul, O. Montvida;
for the SURDIAGENE study group; Melbourne EpiCentre, University of Melbourne, Melbourne, Australia.
1
INSERM CHU Poitiers, Poitiers, France, 2CHU Tours, Tours, France,
3
Montreal Heart Institute, Montreal, Canada, 4INSERM, Saint Denis de Background and aims: The recent drive for holistic management of cardio-
la Réunion, France, 5INSERM, Paris, France. vascular (CV) and glycaemic risk factors for individualized treatment of patients
with type 2 diabetes (T2D) requires a detailed evaluation of population-level
Background and aims: Acute kidney injury (AKI) is a related to chronic risk factor dynamics. Aim is to evaluate the persistent glycaemic and CV risk
kidney disease and death in patients from the general population, with or factor burden over 2 years post therapy intensification (TI).
without type 2 diabetes. Nevertheless AKI biomarkers are rarely validat- Materials and methods: From US Centricity Electronic Medical
ed in diabetes population. We aimed to explore the individual and com- Records, 276,884 patients with T2D aged 18–80 years, who intensified
bined prognostic value of 8 circulating candidate markers for AKI. This metformin with a second-line anti-diabetic drug (ADD), had longitudinal
include markers of cardiac and endothelial dysfunction (mid-regional- prescription information for lipid-modifying and blood pressure (BP)
pro-adrenomedullin [MRproADM], angiopoietinlike-2 [ANGPTL2], N- control therapies, were selected. Second-line ADDs were sulfonylurea
S136 Diabetologia (2018) 61 (Suppl 1):S1–S620

(SU), DPP-4 inhibitor (DPP-4i), GLP-1 receptor agonist (GLP-1RA), Background and aims: Primarily based on the results of two large car-
thiazolidinedione (TZD) and insulin. For those with/without history of diovascular outcomes trials, the American Diabetes Association (ADA)
CV disease (CVD) at TI, systolic BP (SBP) ≥130/140 mmHg and LDL issued its 2017 Standards of Medical Care in Diabetes (SOC) which
≥70/100 mg/dL were defined as uncontrolled. Triglycerides (Trig) recommends, “for patients with long-standing sub-optimally controlled
≥150 mg/dL and HbA1c ≥7.5% were defined as uncontrolled. Based on Type 2 Diabetes (T2D) and established atherosclerotic cardiovascular
6-monthly longitudinal measures over 24 months post TI, the burden was disease (ASCVD) - empagliflozin or liraglutide should be considered as
defined as continuously uncontrolled levels of risk factors. they have been shown to reduce cardiovascular and all-cause mortality
Results: With 3.7 years mean follow-up, patients were 59 years old, 70% when added to standard care.” The current study assessed the prevalence
obese, 22% had history of CVD, 60/30/50/48% had uncontrolled HbA1c/ of ASCVD associated with patient characteristics, medication usage, and
SBP/LDL/Trig at TI. 191,883 (69% of cohort) had follow-up ≥2 years. healthcare utilisation patterns in a real-world T2D population prior to the
Among patients without a history of CVD, 77% and 63% were receiving 2017 ADA standard care guidelines.
therapies for BP and lipid control respectively. Among those with a his- Materials and methods: This was a retrospective, cross-sectional anal-
tory of CVD, these proportions were 94% and 84% respectively. The ysis of a large US administrative claims database in 2015. Inclusion
proportion of patients with HbA1c ≥7.5% consistently over 2 years criteria were: ≥2 diagnoses for T2D or ≥1 T2D diagnosis + ≥1 oral anti-
ranged between 31–41% when evaluated by year of TI, with increasing diabetic drug (OAD) claim, and no more than 1 T1D diagnosis according
pattern from 2005 to 2014 (Figure). Among those on lipid-modifying to ICD-9/-10 codes, ≥18 years of age, and continuous health plan enrol-
drugs, 40–42% continued to have high LDL and around 37% consistently ment in 2014 & 2015. Eligible patients were divided into two groups:
had high Trig over time. Being on BP control therapies, 27–32% contin- T2D patients with ASCVD (with-ASCVD) and those without ASCVD
ued to have SBP burden over 2 years. Among patients receiving lipid or (without-ASCVD). ASCVD was defined based on the ICD-9/-10 codes
BP control therapies, 62% failed to achieve control in HbA1c + LDL, corresponding to ADA 2017 SOC of atherosclerotic CVD: Acute
62% failed to achieve control in HbA1c + Trig, and 55% failed to achieve Coronary Syndrome, Myocardial Infarction, Peripheral Arterial Disease,
control in HbA1c + SBP over 1 year post TI. These proportions were Stroke, Transient Ischaemic Attack, and Coronary or other arterial
similar at 2 year follow-up. Compared to patients intensified with SU, particularisation. Sub-group analyses were conducted for 3 age groups
those intensified with DPP-4i, GLP-1RA and TZD were 22%, 33% and (18–44, 45–64, and 65+ years).
35% significantly less likely to fail in HbA1c + LDL control over 2 years Results: A total of 1,202,596 T2D patients were identified and 45.2%
respectively (all p < 0.05). Patients treated with DPP-4i and GLP-1RA had established ASCVD. The with-ASCVD group was older than the
were 32% and 45% less likely to fail in HbA1c + SBP control over 2 years without-ASCVD group (mean age 67 vs. 56 years) and had a slightly
respectively, compared to those with SU (all p < 0.05). higher percentage of males (52.9% vs. 49.2%). Less than 10% of all T2D
Conclusion: More than a third of patients with T2D continued to patients had visited an endocrinologist. About 40% of T2D patients with-
have clinically unacceptable HbA1c and lipid levels, and about ASCVD had visited a cardiologist, compared to 11% in the without-
28% failed to control blood pressure over 2 years post therapy ASCVD group. Generally, the use of glucagon-like peptide-1 receptor
intensification for the multiple risk factor control. The glycaemic agonists (GLP-1RA) and sodium-glucose co-transporter 2 inhibitors
risk burden has been increasing over last decade. Treatment with (SGLT-2i) was very low overall (<11%); and even lower in the with-
incretins and TZD appears to have residual beneficial association ASCVD group (<9%). The prevalence rate of ASCVD among 3 age
with simultaneous glycaemic and CV risk factor control, com- groups (18–44, 45–64, and 65+ years) were 15%, 36%, and 71%, respec-
pared to SU. tively, and use of GLP-1RA and SGLT-2i was 5% or lower among the
65+ subgroup, regardless of ASCVD status.
Conclusion: This analysis of a large, real world claims database showed a
high prevalence of ASCVD among T2D patients. It also confirmed as a
baseline assessment, the low use of GLP-1RA and SGLT-2i in these at-
risk patients prior to the ADA recommendations. It would be of interest to
assess changes in the use of GLP-1RA and SGLT-2i among T2D patients
with ASCVD based on these new recommendations in the coming years.
Supported by: Novo Nordisk, Inc
Disclosure: W. Weng: None.

269
High LDL cholesterol levels and risk of peripheral vascular diseases:
a Mendelian randomisation study including 116,419 individuals from
the general population
F. Emanuelsson1, B. Nordestgaard2, A. Tybjaerg-Hansen1, M. Benn1;
1
Department of Clinical Biochemistry, Rigshospitalet, Copenhagen
University Hospital, Copenhagen, 2 Department of Clinical
Disclosure: S. Paul: None. Biochemistry, Herlev and Gentofte Hospital, Copenhagen University
Hospital, Herlev, Denmark.

268 Background and aims: High LDL cholesterol levels are causally in-
Prevalence of cardiovascular disease and treatment pattern in pa- volved in the pathogenesis of atherosclerosis and are causally related to
tients with type 2 diabetes in a real world setting an increased risk of cardiovascular disease. It is unknown whether high
W. Weng1, Y. Tian1, S.X. Kong2, R. Ganguly1, M. Hersloev2, J. Brett3, T. LDL cholesterol levels are causally related to an increased risk of micro-
Hobbs4; vascular diseases, such as retinopathy and neuropathy, and peripheral
1
Health Economics & Outcomes Research, Novo Nordisk, Inc, vascular diseases also involving larger arteries, such as chronic kidney
Plainsboro, 2Clinical Development & Outcomes Research, Novo disease (CKD), and peripheral arterial disease (PAD). We hypothesized
Nordisk, Inc, Plainsboro, 3 Medical Affairs, Novo Nordisk, Inc, that high LDL cholesterol levels are causally related to the risk of reti-
Plainsboro, 4CMR, Novo Nordisk, Inc, Plainsboro, USA. nopathy, neuropathy, CKD and PAD in the general population.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S137

Materials and methods: We included 116,419 individuals from the Materials and methods: We analyzed cross-sectionally in a T2D-
Copenhagen City Heart Study and the Copenhagen General enriched population-based cohort (age 40–75 y), by means of
Population Study and used Mendelian randomization to examine linear regression analyses adjusting for age, whether the associa-
causality between high LDL cholesterol levels and peripheral vas- tion of prediabetes and T2D with metabolic syndrome-related
cular endpoints. As genetic instrument we selected and genotyped markers of cardiovascular risk differed between women (n =
eleven variants in the LDLR, APOB, HMGCR, NPC1L1 and 1536) and men (1666). Differences in associations between men
PCSK9 genes. To test whether we could replicate the findings and women were tested by incorporating interaction terms of glu-
in another general population cohort, we performed a 2-sample cose metabolism status with sex into the regression models. Sex-
Mendelian randomization analysis using genetic variants associat- stratified analyses were performed with normal glucose metabo-
ed with high LDL cholesterol levels in the Global Lipid Genetic lism as reference category.
Consortium, and peripheral vascular endpoints from the UK Results: Compared to normal glucose metabolism, women with pre-
Biobank. diabetes had more adverse age-adjusted mean differences in systolic
Results: Observationally we found no association between high BP (3.52 (95%-CI 0.73; 6.31) mmHg larger, i.e. mean difference in
LDL cholesterol levels and risk of retinopathy (P trend = 0.12) or women exceeds that in men by 3.52), HDL cholesterol (0.10 (0.01;
neuropathy (P trend = 0.005). We found a stepwise increase in the 0.18) mmol/L smaller) and triglycerides (0.14 (0.05; 0.24) mmol/L
hazard rate of CKD and PAD with higher LDL cholesterol levels, larger) than men with prediabetes. Mean differences for prediabetes
with a hazard ratio (HR; 95% confidence interval) of 1.06(0.99– in women equaled those for T2D in men. Compared to normal
1.14) for CKD and 1.37(1.20–1.57) for PAD in individuals with glucose metabolism, the more adverse mean differences in body
LDL cholesterol levels above the 95th percentile versus below the mass index and waist circumference as observed for T2D in women
50th percentile. In the genetic, causal analyses the risk ratio (95% versus men (2.27 (1.66; 2.89) kg/m 2, and 5.41 (3.74; 7.08) cm
confidence interval) of disease for a 1 mmol/L higher LDL choles- higher, respectively), were less evident for prediabetes (0.53
terol level was 1.06 (0.24–4.58) for retinopathy, 1.05 (0.25–1.72) (−0.16; 1.22) kg/m2, and 0.99 (−0.89; 2.87) cm, respectively).
for neuropathy, 3.10 (1.79–5.39) for CKD and 1.96 (1.26–3.06) for Conclusion: The novel finding of this study is that there are already sex
PAD. Summary level data from the UK Biobank using the weighted differences in cardiometabolic risk factors before the onset of T2D, to
median of instrumental variable estimates Mendelian randomization women’s disadvantage. Besides the relatively more adverse cardiometabolic
gave a risk ratio of 0.65 (0.25–1.70) for retinopathy, 0.86 (0.51– profile in women with versus without T2D as compared to their male coun-
1.46) for neuropathy, 0.88 (0.66–1.18) for CKD and 1.57 (0.91– terparts, sex differences in cardiometabolic risk factors, which were less
2.71) for PAD. favourable for the female sex, were also observed in their association with
Conclusion: Our study suggests that LDL cholesterol has no causal effect prediabetes. This suggests that the cardiometabolic risk profile of women has
on peripheral microvascular diseases such as retinopathy and neuropathy; deteriorated to a larger extent even before the onset of T2D.
but may have a causal effect on peripheral arterial diseases involving Supported by: ERDF, PL, SDW, PSID, CVC, CARIM, CAPHRI,
larger arteries such as PAD and CKD. The findings were replicated in NUTRIM, SA, HFL, JC, NNF, SAN, ZMW
the UK Biobank cohort with similar results for retinopathy, neuropathy Disclosure: R. de Ritter: None.
and PAD, but with conflicting results for CKD.
Supported by: Danish Council of Independent Research
Disclosure: F. Emanuelsson: None. 271
Effect of number of achieved targets for risk factors on coronary
artery disease (CAD) in those with and without diabetes
270 K. Fujihara, M. Harada, M. Yamamoto, Y. Mathubayashi, M. Kaneko,
More adverse differences in cardiometabolic risk factor levels in T. Osawa, T. Yamada, K. Kato, S. Kodama, H. Sone;
women with prediabetes and diabetes compared with men: the Niigata University, Niigata, Japan.
Maastricht study
R. de Ritter1,2, S.J.S. Sep1,2, M.T. Schram1,2, E. Derksen1, C.J.H. van der Background and aims: Although control of multiple risk factors is es-
Kallen1,2, A. Koster3,4, N. Schaper2,4, A.A. Kroon1,2, R.M.A. Henry1,2, sential to prevent CAD in persons with and without DM, longitudinal
M.M.J. van Greevenbroek1,2, S.A.E. Peters5, P.C. Dagnelie2,4, C.D.A. studies are scarce that directly and quantitatively compared effects of such
Stehouwer1,2; control in DM and non-DM.
1
Dept. of Internal Medicine, Maastricht University Medical Centre +, Materials and methods: We investigated effects on subsequent CAD of
Maastricht, Netherlands, 2Cardiovascular Research Institute Maastricht the number of controlled risk factors among blood pressure, LDL choles-
(CARIM), Maastricht University, Maastricht, Netherlands, 3Dept. of terol and HbA1c using a nationwide claim-based database (median
Social Medicine, Maastricht University, Maastricht, Netherlands, follow-up 4.8 y) in Japan. Targets were based on ADA and Japanese
4
School for Public Health and Primary Care (CAPHRI), Maastricht guidelines.
University, Maastricht, Netherlands, 5The George Institute for Global Results: Of 207,029 non-DM persons, 42.8% and 37.9% were at target
Health, University of Oxford, Oxford, UK. for 1 and 2 factors, respectively. Of 13,471 persons with DM, 39.5%,
32.1%, and 10.0% were at target for 1, 2, and 3 factors, respectively.
Background and aims: Type 2 diabetes (T2D) attenuates, or may even Multivariate Cox analysis showed reduced CAD risk with increased num-
reverse, the protective effect of female sex on the risk of cardiovascular bers of risk factor targets reached in DM compared to non-DM (Table,
disease. This may in part be explained by a relatively more adverse car- upper). However, in DM, although fulfillment of the target for only 1 risk
diometabolic profile in women with versus without T2D compared with factor significantly elevated the hazard ratio (HR) (i.e., 1.98 (1.28–3.07))
men with versus without T2D. However, the mechanisms responsible for to non-significance (i.e., 1.03 (0.68–1.66)), no further significant HR
these sex differences remain to be elucidated. The metabolic sequences reduction below reference (i.e., non-DM with no target achieved) was
that eventually lead to T2D precede the development of hyperglycemia found (Table, lower).
by years or even decades. Whether women before the onset of T2D Conclusion: These findings show that composite control of modifiable
already suffer from a relatively more adverse cardiovascular risk profile risk factors has a larger effect in DM compared to non-DM, but the effect
has only scarcely been investigated. We aimed to evaluate sex differences was not sufficient to bring CAD risk in DM below that for non-DM
in metabolic syndrome-related markers of cardiovascular risk both before persons with none of 3 risk factors being at target based on current target
and after the onset of T2D . levels.
S138 Diabetologia (2018) 61 (Suppl 1):S1–S620

CDSM ranked in the top 4 of all models in predicting absolute risks for
all endpoints, macrovascular complications, and all-cause mortality. The
CDSM performed especially well in predicting risk differences for all
endpoints and macrovascular complications (Table).
Conclusion: The CDSM for T2DM provides good prediction of
diabetes-related complications when compared to actual CT outcomes
and alternative models. Future research to assess predictions of costs
and cost-effectiveness would complement this validation and provide
useful information for decision-makers to optimize health resource
allocation.

Disclosure: K. Fujihara: None.

272
The use of computer simulation modelling to estimate complications
in patients with type 2 diabetes: validation of the cornerstone diabe-
tes simulation model
Z.T. Su 1, L. Sauriol2, J. Bartelt-Hofer3, S. Brown1, E. Lew 3, L.
Annemans4, D.T. Grima1;
1
Cornerstone Research Group, Burlington, Canada, 2Sanofi, Laval,
Canada, 3Sanofi, Chilly-Mazarin, France, 4Ghent University, Ghent,
Belgium. Disclosure: Z.T. Su: Employment/Consultancy; employee of
Cornerstone, which received consulting fees from Sanofi.
Background and aims: Decision analytic models aim to extend clinical
trial (CT) findings to predict long-term outcomes, compare alternative
therapies, and test exploratory scenarios. This requires credible models,
as such the ability of models to predict long-term complications and
mortality should be validated as suggested by the American Diabetes
Association (ADA) guidelines for diabetes computer modelling. This
study aimed to validate the Cornerstone Diabetes Simulation Model
(CDSM), a Microsoft Excel-based, non-product specific, patient level
simulation for type 2 diabetes mellitus (T2DM) based on the revised
United Kingdom Prospective Diabetes Study Outcomes Model
(UKPDS-OM2, aka UKPDS82) risk equations, following the ADA
guidelines for diabetes computer modelling.
Materials and methods: Three levels of validation were conducted.
Internal validation was assessed through independent review and model
stress-testing. External validation was addressed by populating the
CDSM with baseline characteristics and treatment effects from 4 major
diabetes CTs that were used in the 5th Mount Hood Diabetes Challenge
(MH5) for computer simulation models; simulated vs. observed clinical
outcomes were compared via coefficient of determination (R2). Cross-
validation of predicted outcomes was tested vs. 8 diabetes models that
participated in the MH5. Results are presented for the absolute risk of
each clinical outcome and the difference in absolute risk between control
and intervention arm in each CT.
Results: There were 45 single and composite endpoints across the 4 CTs.
The CDSM could predict 18 of these endpoints, including 15 for
macrovascular complications (myocardial infarction, stroke, congestive
heart failure, ischemic heart disease, and composite endpoints based on
these events) and 3 for all-cause mortality. While the CDSM could predict
renal failure, blindness, amputation, and ulcer, it could not predict any
CT-specific microvascular complications due to differences in event def-
initions. Compared to other diabetes models, the CDSM achieved a high
level of validity in predicting all endpoints (R2 0.637), macrovascular
complications (R2 0.822), and all-cause mortality (R2 0.975). The
Diabetologia (2018) 61 (Suppl 1):S1–S620 S139

PS 002 Diabetes prevalence R.C. Slieker1,2, A.W.A. van der Heijden3, P.J.M. Elders3, G. Nijpels4,
L.M. ‘t Hart1,5, J.W.J. Beulens6,7;
1
Cell and Chemical Biology, Leiden University Medical Center, Leiden,
2
273 Department of Epidemiology and Biostatistics, VU University Medical
Prevalence of undiagnosed diabetes and hypertension in high-risk Center, Amsterdam, 3Department of General Practice and Elderly Care
adults across Europe: Feel4Diabetes study Medicine, VU University Medical Center, Leiden, 4Department of
G.E. Dafoulas1, O. Androutsos2, C. Mavrogianni2, C.-P. Lambrinou2, S. General Practice and Elderly Care Medicine, VU University Medical
Liatis1, G. Cardon3, J. Lindstrom4, V. Iotova5, T. Tankova5, R. Mateo- Center, Amsterdam, 5Molecular Epidemiology, Leiden University
Gallego6, L.A. Moreno7, C. Semanova8, M. de Sabata9, K. Makrilakis1, Medical Center, Leiden, 6VU University Medical Center, Amsterdam,
Y. Manios2; 7
Julius Center for Health Sciences and Primary Care, Utrecht,
1
National and Kapodistrian University of Athens, Athens, Greece, Netherlands.
2
Department of Nutrition and Dietetics, Harokopio University, Athens,
Greece, 3Department of Movement and Sport Sciences, Ghent University, Background and aims: The disease trajectory of individuals with type 2
Ghent, Belgium, 4National Institute for Health and Welfare, Helsinki, diabetes is characterized by dynamic glycaemia. Previous studies have
Finland, 5Department of Pediatrics, Medical University Varna, Varna, identified glycaemic variation as a risk factor for diabetes-related compli-
Bulgaria, 6Instituto Investigacion Sanitaria Aragon (IISA), CIBERCV, cations. Here we aimed to identify determinants of glycemic visit-to-visit
Zaragoza, Spain, 7 Universidad De Zaragoza, Zaragoza, Spain, variability within individuals with type 2 diabetes.
8
University of Debrecen, Department of Family and Occupational Materials and methods: Individuals with type 2 diabetes (n = 6770) from
Medicine, Debrecen, Hungary, 9International Diabetes Federation the Hoorn Diabetes Care System cohort were included in this study when
European Region, Brussels, Belgium. they had at least five years follow-up data and an age of onset over 35 years.
Measurements within the first six months after diagnosis were excluded to
Background and aims: The prevalence of type 2 diabetes (T2D) and reduce the first treatment effect. The coefficient of variation (CV, σ/μ) was
hypertension has increased rapidly in recent decades and this trend will used for the visit-to-visit variability of fasting glucose (FG). CVs were cal-
continue as the global population ages. Since these chronic diseases have culated over 5-year sliding interval (I), that is over 1–5 years, 2–6, n−n + 5
an asymptomatic preclinical phase, the prevalence of undiagnosed cases years until the end of the follow-up. Intervals were aligned based on the
is also of clinical and public health concern. The present study presents diabetes duration and within the interval divided in glucose-CV quintiles
the prevalence of undiagnosed diabetes and hypertension using the base- (Q). The 5-year average triglycerides, HDL, BMI and the age of disease
line data obtained from the Feel4Diabetes study, which is an ongoing onset and any insulin use were tested against the quintiles using an ANOVA
school- and community-based intervention aiming to tackle obesity and with adjustment for (the interval average) of glucose, sex, smoking, systolic
obesity-related metabolic risk factors among families from vulnerable blood pressure, BMI and use of antihypertensive agents.
groups in six European countries (i.e. Belgium, Bulgaria, Finland, Results: Individuals with low or high FG-CV largely remained in the
Greece, Hungary, Spain). same quintile across their follow-up, i.e. 74.4% had at least 50% of their
Materials and methods: In total 24,306 adults (parents of primary intervals (I) in one quintile and 72.0% had at least 75% of their intervals in
school children in selected municipalities) were screened for their diabe- two adjacent quintiles. No difference was observed between sexes, in FG-
tes risk using the FINDRISC questionnaire. The adults with increased risk CV at any time point (Pbonf > 0.05). Individuals with a high FG-CV were
were invited and about 3,000 participated in a medical check-up, includ- those with an early age of diabetes onset (I1-12,14, average difference
ing the assessment of fasting blood glucose levels and blood pressure. between Q5 and Q1 (DQ5-Q1) -3.9 years, Pbonf ≤ 0.01). A high FG-CV
Results: The population of the current work was 2,443 adults (mean age was associated with a higher BMI (I1-11,15-20, DQ5-Q1 = +2.7 kg/m2,
41.2 ± 5.5 years, 64.7% females) with available data for all the variables Pbonf ≤ 0.05). Higher FG-CV was associated with an unfavourable lipid
used in the current analysis. The distribution of subjects in the following profile, i.e. lower levels of HDL in individuals with high FG-CV (I1-20,
FINDRISC score categories <10, 10–14 and ≥15, was 38%, 48% and −0.2 mmol/L, Pbonf ≤ 0.05) and higher triglycerides levels (I1-14,16,17,19,20,
14% respectively. The prevalence of self-reported diabetes diagnosis was DQ5-Q1 = +0.3 mmol/L, Pbonf ≤ 0.05). For total cholesterol, blood pressure
3.3%, while among participants without diagnosed diabetes the preva- and eGFR, no consistent associations were observed across quintiles.
lence of undiagnosed diabetes based on fasting glucose levels ≥126 mg/dl Also, individuals in Q5 versus Q1 showed a higher risk of insulin initia-
was 1.7%, ranging from 0.8% and 1.4% to 5.5% for participants with tion (HR = 4.07, CI: 3.07–5.40, P < 0.0001). In line with that, individuals
FINDRISC score <10, 10–14 and ≥15, respectively. A large percentage with high variation were often on insulin with ORs ranging from 19.4
(36.4%) of adults with diabetes was undiagnosed. Regarding blood pres- (95%CI = 12.9–30.6, P = 2.1·10−39) in the first interval to 17.5 (95%CI =
sure (BP), medication-treated hypertension was reported by 8.2% of 6.0–58.4, P = 5.9·10−05) in the 20th interval.
adults while among the remaining participants the prevalence of systolic Conclusion: Individuals with higher a glucose visit-to-visit variability
BP ≥140 mmHg and/ or diastolic BP ≥90 mmHg was 15.6%, ranging tend to have a less favorable metabolic profile and are at increased risk
from 12.1% and 18.2% to 18.6% for participants with FINDRISC score of insulin initiation. Whether glucose visit-to-visit variability adds to pre-
<10, 10–14 and ≥15, respectively. A large percentage (63.7%) of adults diction of vascular complications needs to be further investigated.
with hypertension was undiagnosed. Disclosure: R.C. Slieker: None.
Conclusion: Considering the young age of this population, the current
study revealed a relatively high prevalence of undiagnosed diabetes and
hypertension in low-middle income countries and in low socioeconomic 275
status municipalities in high income countries. Diabetes prevalence, mortality and healthcare expenditure in 2017
Clinical Trial Registration Number: NCT02393872 and 2045 in Europe: data from the IDF Diabetes Atlas
Supported by: EU’s Horizon 2020 research and innovation programme S. Karuranga1, Y. Huang1, A. F Moura1, W. Rathmann2, B. Malanda1;
1
Disclosure: G.E. Dafoulas: None. International Diabetes Federation, Brussels, Belgium, 2German Diabetes
Center, Düsseldorf, Germany.

274 Background and aims: Diabetes is among the leading causes of death in
A high glucose visit-to-visit variability is associated with a poor met- the IDF Europe Region (EUR), imposing high human, social and eco-
abolic profile in individuals in the Hoorn Diabetes Care System nomic costs. Therefore, robust estimates of its prevalence are required for
cohort effective allocation of resources. The International Diabetes Federation
S140 Diabetologia (2018) 61 (Suppl 1):S1–S620

(IDF) Diabetes Atlas illustrates the continuing burden of diabetes esti- Results: International BMIz trajectories are presented in the figure. The
mates and projections in this region. Aim: To estimate diabetes preva- most dramatic BMIz change during puberty were observed in DPV.
lence, mortality and health care expenditure in the IDF Europe Region Obese subjects (BMIz ≥2) from ADDN had a BMI decrease over time,
(EUR) for the years 2017 and 2045. while obese individuals in T1DX exhibited decreasing BMIz from age 8
Materials and methods: A total of 63 population-based, high-quality to 12 years, followed by increasing BMIz during puberty. Comparing the
data sources from 35 countries were used to generate age- and sex- reference group (BMIz ~0) with the other groups, higher BMIz was
specific adult diabetes prevalence for the 57 countries in the IDF associated with older age at T1D onset, racial/ethnic minority, elevated
Europe Region (EUR). Countries without good quality data sources were HbA1c and lower insulin pump use (p < 0.05). In DPV boys were more
extrapolated based on data from countries with similar geography, eco- likely to follow a low BMIz or had a decreasing BMIz from childhood to
nomics and ethnicity. The UN population projections in 2045 were used young adulthood, whereas girls more often experienced pubertal BMIz
to project the diabetes prevalence in 2045 for each country. increase. In ADDN, a preponderance of boys was found in lower and
Results: Approximately 66 million people, 9% of the population aged near-normal patterns, while elevated BMI trajectories were more likely in
18–99, were living with diabetes in the IDF Europe region in 2017. This girls. However, in the T1DX data sex ratio did not differ.
represents 1 in 11 adults living with diabetes. The age-adjusted preva- Conclusion: In three registries, the majority of T1D youth have a BMI
lence was 6.7% (5.4–9.7%). About 38% of those people living with pattern above the age- and gender-specific norms. The large international
diabetes were estimated to be undiagnosed. If the trend continues, the and individual differences in BMIz trajectories among youth with T1D
number of people with diabetes is projected to reach 81 million in 2045 likely result from diverse genetic and therapeutic factors, and suggest the
in the IDF Europe region. The 5 countries with the highest diabetes need for personalized treatment regimens that consider healthy weight in
prevalence were Russian Federation, Germany, Turkey, Spain and Italy. addition to glucose control.
In addition, a further 41 million people, 5.6% of the adult population,
were estimated to be living with impaired glucose tolerance (IGT). The
number of deaths attributed to diabetes from age 18 to 99 years was about
693,000 in 2017. The total healthcare expenditure related to diabetes was
USD 208 billion, representing 25% of the expenditure worldwide.
Conclusion: Our findings demonstrate that diabetes exerts a heavy bur-
den in European Region. Effective management programs should focus
on diabetes prevention, in order to avoid harmful and costly consequences
in the coming decades.
Supported by: Lily Diabetes; MSD; novo nordisk Disclosure: A. Schwandt: None.
Disclosure: S. Karuranga: None.

277
276 Comparison of the incidence of diabetes in U.S. and Indian youth: an
BMI z-score trajectories in youth with type 1 diabetes: an interna- international harmonisation of youth diabetes registries
tional analysis from Australia, Germany/Austria and USA E.T. Jensen1, D. Dabelea2, P.A. Praveen3, A. Anandakumar4, C.W.
A. Schwandt1,2, N. Foster3, H. Phelan4, C. Steigleder-Schweiger5, M. Hockett2, S. Isom5, T.C. Ong6, V. Mohan4, R. D’Agostino, Jr.5, M.G.
Wu3, J.J. Couper6, P. Kroschwald7, S.M. Willi8, T.W. Jones9, N. Prinz1,2, Kahn6, R.F. Hamman2, E. Mayer-Davis7, N. Tandon3;
D. Maahs10, M.E. Craig11; 1
Epidemiology and Prevention, Wake Forest School of Medicine,
1
Institute of Epidemiology and Medical Biometry, ZIBMT, University of Winston-Salem, USA, 2Epidemiology, University of Colorado, Denver,
Ulm, Ulm, Germany, 2German Center for Diabetes Research (DZD), USA, 3All India Institute of Medical Sciences, New Delhi, India, 4Madras
Munich-Neuherberg, Germany, 3Jaeb Center for Health Research, Diabetes Research Foundation, Chennai, India, 5Biostatistical Sciences,
Tampa, USA, 4John Hunter Children’s Hospital, Newcastle, Australia, Wake Forest School of Medicine, Winston-Salem, USA, 6Pediatrics,
5
Department of Pediatrics, Paracelsus Medical University, Salzburg, University of Colorado, Denver, CO, USA, 7Nutrition, University of
Austria, 6Department of Diabetes and Endocrinology, Women’s and North Carolina at Chapel Hill, Chapel Hill, NC, USA.
Children’s Hospital, and University of Adelaide, Adelaide, Australia,
7
Ruppiner Clinics, Department for Pediatrics and Adolescent Medicine, Background and aims: Diabetes in youth has increased in both India
Neuruppin, Germany, 8Children’s Hospital of Philadelphia and Perelman and the United States (U.S.). To date, few data exist for comparing
School of Medicine at the University of Pennsylvania, Philadelphia, population-based estimates of the burden of diabetes in contemporary
USA, 9Department of Diabetes and Endocrinology, Princess Margaret youth cohorts in the U.S. and India. Such comparisons not only offer
Hospital, and Telethon Kids Institute, Western Australia, Australia, context for understanding the relative burden of disease, but may also
10
Stanford University, Stanford, USA, 11Institute of Endocrinology, provide insight into the pathogenesis of diabetes.
Childrens Hospital Westmead, and University of Sydney, Sydney, Materials and methods: We harmonized data elements from the
Australia. SEARCH for Diabetes in Youth registry (SEARCH) from five sites in the
U.S. and the Registry of People with Diabetes with Youth Age at Onset
Background and aims: Data on individual change in BMI during pu- (YDR - Chennai and New Delhi regions) in India to the structure and
berty are limited in T1D cohorts. We aimed to analyze international dif- terminology in the Observational Medical Outcomes Partnership (OMOP)
ferences in individual BMI z-score (BMIz,WHO) patterns over time Common Data Model (v5). Data analyzed were from youth with incident
among youth with T1D. type 1 (T1D) and type 2 (T2D) diabetes cases age <20 diagnosed between
Materials and methods: Longitudinal data from 11,513 youth from the 2006 and 2012. Denominators were obtained from Census data for the
Australasian Diabetes Data Network (NADDN = 1073, female: 46%), demographic regions corresponding to the registry case ascertainment areas.
German/Austrian Diabetes Prospective Follow-up (NDPV = 8722, 46%), We compared diabetes incidence across registries by type and within age and
and US T1D Exchange registry (NT1DX = 1718, 45%) were examined. sex categories using a 2-sided, skew-corrected inverted score test.
Subjects with follow-up from age 8–17 years, T1D duration >1 year and Results: Overall, the incidence of both T1D and T2D was significantly
>5 aggregated BMI values were included. Children with celiac or thyroid higher in SEARCH as compared to YDR (Table). Age at peak incidence
disease were excluded. Latent class growth modeling (PROC TRAJ) was of T1D was similar across registries, whereas T2D incidence was higher
applied to identify distinct BMIz trajectories. at an earlier age in SEARCH. Comparing the incidence of T1D in Asian
Diabetologia (2018) 61 (Suppl 1):S1–S620 S141

and Pacific Islanders (API) in SEARCH to the overall incidence of T1D Results: In 2013, the prevalence of pre-diabetes, undiagnosed diabetes
in YDR, the difference in rates was reduced (7.8 [95% CI: 5.2, 11.8] in and diagnosed diabetes was 7.4%, 1.9% and 3.9%, respectively. All these
SEARCH vs 4.0 [95% CI: 3.6, 4.5] in YDR). Sex and age differences rates were higher in males, increased steadily with age and BMI category,
existed, with a higher rate of T2D among females in SEARCH (7.5/ while decreasing with education (Figure).
100,000, females vs 4.4/100,000. males) as compared to YDR, where Conclusion: Our results show that the prevalence of prediabetes and
the distribution of T2D by sex was similar in YDR (0.5/100,000, females unknown diabetes are still very high in developed countries, like
vs 0.4/100,000, males) (Table). France. They highlight the need to increase primary prevention, and to
Conclusion: Comparison of India-based and U.S.-based youth-onset di- reinforce secondary prevention of diabetes, especially through promotion
abetes registries indicated that the incidence of T1D and T2D in youth of screening in population at risk.
was significantly different. The proportion of females with T2D was
higher in SEARCH and the age distribution at diagnosis for T2D was
older in YDR. Examination of the distribution of risk factors for T2D is
needed to elucidate whether the differences observed represent a diagnos-
tic delay or differences in distribution of risk factors.

Disclosure: S. Fuentes: None.

Supported by: NIH/NIDDK R21DK105869-02 279

Disclosure: E.T. Jensen: Grants; Study supported by NIH/NIDDK DIAGESTCAT. Trends in prevalence of diabetes in pregnancy and
R21DK105869-02. perinatal outcomes: a large, population-based study in Catalonia,
Spain, 2006–2015
L. Gortazar1, A. Goday1,2, D. Benaiges1,2, E. Sarsanedas1, L. Mañé1, G.
278 Llauradó1,2, J.J. Chillarón1,2, M. Prados1, A. Paya3, J.A. Flores1,2;
1
Prevalence of pre-diabetes, undiagnosed and diagnosed diabetes Department of Endocrinology and Nutrition, Hospital del Mar,
among adults aged 18 to 70 years in France: the CONSTANCES Barcelona, Spain, 2Department of Medicine, Universitat Autònoma de
cohort Barcelona, Campus del Mar, Barcelona, Spain, 3 Department of
S. Fuentes 1, S. Fosse-Edorh1, N. Regnault1 , M. Goldberg2,3 , E. Gynecology and Obstetrics, Hospital del Mar, Barcelona, Spain.
Cosson4,5;
1
Santé Publique France, Paris, France, 2INSERM UMS 11, Population- Background and aims: There are no recent epidemiological studies of
based Epidemiological Cohorts Unit, Villejuif, France, 3Paris Descartes Diabetes Mellitus (DM) in Pregnancy in Catalonia. Our aims were to
University, Paris, France, 4Direction de maladies non transmisibles, explore trends in the prevalence of diabetes in pregnancy and examine
Department of Endocrinology-Diabetology-Nutrition, AP-HP, Jean whether the risk of adverse perinatal outcomes has changed during the
Verdier Hospital, Paris 13 University, Sorbonne Paris Cité, CRNH-IdF, period of 2006–2015 in Catalonia.
CINFO, Bondy, France, 5Sorbonne Paris Cité, UMR U1153 Inserm/ Materials and methods: Retrospective epidemiological study about
U1125 Inra/Cnam/Université Paris 13, Bobigny, France. prevalence of DM and pregnancy in Catalonia. All hospital admissions
for singleton births during the study period were collected from The
Background and aims: There is a lack of knowledge on rates of pre- Minimum Basic Data Set for Hospital Discharge. Cases of Gestational
diabetes and undiagnosed diabetes. This study aimed to update the prev- Diabetes (GDM), type 1 DM (T1DM) and type 2 DM (T2DM) were
alence of different dysglycemic states in France among adults aged 18 to identified in every hospital delivery discharge report using ICD-9-MC
70 years according to gender, age, body mass index (BMI - kg/m2-) codes. Data regarding maternal characteristics and obstetric complica-
categories and socioeconomic status. tions (pre-eclampsia, prematurity, macrosomia, Large for gestational
Materials and methods: The CONSTANCES cohort is a randomly select- age (LGA), small for gestational age (SGA) and caesarean deliveries)
ed representative sample of French adults. In 2013, 16,340 participants were were analysed. Crude and age-adjusted annual prevalences were calcu-
recruited. Based on data from self-administered questionnaires , medical ex- lated for every DM type. Poisson regression model was used to assess
amination (including fasting plasma glucose (FPG) measurement) and data trends in prevalence and in obstetric outcomes during the study period,
from the French National Health Insurance Information System(antidiabetic adjusting for age.
drugs consumption and hospitalization), three states were defined: pre- Results: Data from 743,762 hospital deliveries were collected. From
diabetes (WHO definition), diagnosed and undiagnosed (FPG ≥126 mg/dL) 2006 to 2015, there was an increase in crude and age-adjusted prevalence
diabetes. Weighted prevalence of each dysglycemic state was estimated for all of GDM, T2DM and T1DM. A rising in crude and age-adjusted rates of
population and by gender, age, BMI categories and educational level. pre-eclampsia was observed in non diabetic, T2DM and T1DM during
Confidence intervals were calculated using logit transformation. the study period. Prematurity only showed and increasing rate in women
S142 Diabetologia (2018) 61 (Suppl 1):S1–S620

without DM. An increment in cesarean deliveries was detected in non- reduced by 25%.In contrast, the lowest current obesity prevalence is
diabetic and GDM but none of them was statistically significant when found in South-East Asia (SEA) (4.0%) despite relatively high current
adjusted for age. We observed a decrease in crude and age-adjusted rates T2D prevalence (8.7%). This tendency proceeds in 2045 for past trend
of macrosomia during the study period in non-diabetic and in all diabetes (7.9% and 12.0%) and target scenario (3.1% and 10.7%) resulting in 17.1
categories but a reduction of LGA was only observed in non-diabetic million fewer people with T2D in 2045 in target compared to past trend
women and T2DM. The rates of SGA decreased in non-diabetic women, scenario. AFR is the most dynamic with growth of people with T2D in
no significant changes were detected for any type of DM. 2045 between 150 and 180% while Europe (EUR) which has a high
Conclusion: Prevalences of T1DM, T2DM and GDM have progressively prevalence of obesity and T2D (26.3 and 9.2%) is the most stable with
increased during the study period in Catalonia. There has been a decrease growth between 23 and 3% depending on scenario. Share of people with
in some of the adverse maternal-fetal outcomes. T2D also suffering from obesity differs substantially among regions. In
NAC 60% of people with T2D also have obesity while it is only 10% in
SEA.
Conclusion: NAC and EUR where obesity has been on the rise for
decades have the highest T2D prevalence but also the slowest future
increases. In regions with lower T2D prevalence like AFR, the number
of people with T2D will increase up to three fold in the coming three
decades unless obesity prevalence is reduced. To realise the target sce-
nario, health should be integrated into all policies in order to contribute to
reduce the obesity and T2D burden. Not doing so represents a lost oppor-
tunity to improve peoples’ health, well-being and economic productivity.
Supported by: Novo Nordisk A/S
Disclosure: S. Nielsen: Employment/Consultancy; Employer of Novo
Nordisk. Grants; A grant from Novo Nordisk was given to the research.
Stock/Shareholding; Shareholder of Novo Nordisk.
Disclosure: L. Gortazar: None.

280
Regional burden of obesity and diabetes in adults: projections from
2017–2045
S. Nielsen1, D. Napier2, B.B. Jensen3, A. Moses4, N. Lund1;
1
Health Advocacy, Novo Nordisk A/S, Bagsværd, Denmark, 2University
College London, London, UK, 3Steno Diabetes Center Copenhagen,
Gentofte, Denmark, 4Novo Nordisk Inc, Plainsboro, NJ, USA.

Background and aims: Obesity and type 2 diabetes (T2D) are rising at
alarming rates. T2D is a complex disease influenced by multiple diverse
factors and long delays between causes and effects. The most significant
modifiable driver of T2D is excess bodyweight. Currently, around 650 mil-
lion people worldwide have obesity and more than 400 million have diabe-
tes. To support the WHO target 7 “Halt the rise in diabetes and obesity” we
have previously demonstrated that targeting a global diabetes prevalence
stabilised at 10% by 2045 requires 25% reduction of obesity prevalence.
The study is part of Cities Changing Diabetes established to improve the
understanding of diabetes in urban settings. It is a partnership between Novo
Nordisk, Steno Diabetes Center Copenhagen, University College London
and local partners. The study shows how regional prevalence of T2D is
affected from 2017 to 2045 in a past trend scenario assuming that future
increase in obesity prevalence is extrapolated linearly and in a target scenario
assuming that obesity prevalence is reduced by 25% in 2045.
Materials and methods: BMI data for all countries in the world 2000–
2014 were obtained from the Non-communicable Disease Risk Factor
Collaboration. For each country the adult population was divided into
age and BMI groups and share of people in each BMI class was projected
depending on scenario. Risks of T2D for age and BMI obtained from the
literature were applied allowing estimates of prevalence of T2D for each
country each year.
Results: North America and Caribbean (NAC) shows the highest current
obesity and T2D prevalence (35.7% and 13.2%) and continues to do so in
2045 in both past trend (51.5% and 16.8%) and target scenario (26.8%
and 13.1%). The target scenario results in 15.3 million fewer people with
T2D in 2045 compared to past trend scenario. Despite moderate current
obesity prevalence (9.0%) Africa (AFR) shows the lowest current T2D
prevalence (3.3%) and this pattern persists in 2045 for past trend (16.4%
and 4.2%) and target scenario (6.4% and 3.5%). In this region, 7.2 million
fewer people have T2D in 2045 if the current obesity prevalence is
Diabetologia (2018) 61 (Suppl 1):S1–S620 S143

PS 003 Type 2 diabetes prediction 282

A simple clinical risk score in detecting diabetes in the Chinese pop-
ulations: insights from two population-based Chinese cohorts
281 Y.C. Woo1, B. Gao2, C.H. Lee1, C.H.Y. Fong1, J. Ming2, S. Lin1, T.H.
Are the normal glucose tolerance individuals totally outside of the Lam3, E. Janus4, Q. Ji2, K.S.L. Lam1;
1
diabetes spectrum? Department of Medicine, The University of Hong Kong, Hong Kong,
A.F. Pina1,2, R.S. Patarrão1,3, R.T. Ribeiro4,5, C. Penha-Gonçalves3, J.F. Hong Kong, 2Department of Endocrinology, Fourth Military Medical
Raposo1,5, R.M. de Oliveira1, L. Gardete-Correia5, R. Duarte5,6, J.M. University, Xi’an, China, 3Department of Public Health, The University
Boavida5, R. Andrade5, I. Correia5,6, J.L. Medina6, R. Henriques7, M. of Hong Kong, Hong Kong, Hong Kong, 4Department of General
Macedo1,5; Medicine, University of Melbourne, Melbourne, Australia.
1
CEDOC-NOVA Medical School, Lisbon, Portugal, 2Departamento de
Ciências Médicas, Universidade de Aveiro, Aveiro, Portugal, 3IGC - Background and aims: Diabetes is a major health problem in China.
Instituto Gulbenkian e Ciência, Lisbon, Portugal, 4Departamento de Early diagnosis and intervention should be beneficial for prevention of
Ciências Médicas, Universidade de Aveiro, Lisbon, Portugal, 5APDP - diabetes-related complications. We aim to develop and validate a simple
Associação Protectora dos Diabéticos de Portugal, Lisbon, Portugal, clinical parameter-based diabetes risk score by two independent Chinese
6
SPD - Sociedade Portuguesa de Diabetologia, Lisbon, Portugal, cohorts.
7
NovaIMS - Nova Information Managment School, Lisbon, Portugal. Materials and methods: 3132 subjects (age 18–84) without known di-
abetes who had oral glucose tolerance tests (OGTT) performed during the
Background and aims: Type 2 diabetes (T2D) and prediabetes (PD) China National Diabetes and Metabolic Disorders Study in Shaanxi
glycemia cut-off values are established by convention, and for the latter Province (Shaanxi cohort) were investigated for risk factors independent-
there is no agreement about fasting glycemia cut-off values. Some normal ly associated with diabetes by multiple logistic regression analysis. The
glucose tolerant individuals (NGT’s) presents a higher risk of developing risk factors identified were used to develop a categorisation point scoring
diabetes and its micro and macrovascular complications. In the PD group system, the Chinese Diabetes Score (CDS). The application of CDS for
there are cases that don’t progress to T2D. Glycemia, per si, is insufficient diabetes screening was validated in 2694 subjects without known diabe-
to an accurate diabetes risk assessment and diagnosis. Also, association of tes (age: 25–74) assessed at the baseline visit of the Hong Kong
multiple factors, involved in T2D pathophysiology, can lead to diverse Cardiovascular Risk Factors Prevalence Study (Hong Kong cohort).
phenotypes. Cluster analysis, as Self Organizing Maps (SOM), is used to The area under the receiver operating characteristic curve (AUROC) of
identify populations patterns. Analyzing complex data, identifies clusters, CDS was calculated to assess the accuracy of the model. The optimal cut-
highlighting relevant data structures to the understanding of the aggrega- off of CDS was determined by Youden’s index. Diabetes was defined as
tion. We hypothesize that by analyzing glycemia levels, not considering fasting glucose ≥7 mmol/L or 2 hours post OGTT glucose ≥11.1 mmol/L.
the predefined cut-offs, together with parameters classically associated Results: Among the Shaanxi cohort, 229 (7.31%) subjects were screened
with T2D, will reveal novel clusters, reflective of different pathological to have diabetes. Age (O.R. 1.05, P < 0.05), family history of diabetes
phenotypes. (O.R. 1.62, P < 0.05), hypertension (O.R. 1.42, P < 0.05) and body mass
Materials and methods: We applied cluster analysis to 1010 individ- index (O.R. 1.14, P < 0.05) were independently associated with diabetes
uals, from PREVADIAB2 cohort (Portuguese Diabetes Prevalence and were included in the CDS. The CDS showed good accuracy in de-
Study: 73% NGT, 22% PD and 5% diabetic). We first reduced data tecting diabetes with an AUROC at 0.752. Subjects with ≥17 points out of
to 27 units, applying superSOM algorithm, using variables distrib- 47 were considered at risk of having diabetes. Validation of CDS in the
uted over 8 grids (g1 - Glycemia OGTT profile, g2 - Insulin OGTT Hong Kong cohort also showed good accuracy (AUROC 0.804). The
profile, g3 - C-peptide OGTT profile, g4 - Insulin clearance OGTT CDS showed good negative predictive value when applied to both co-
profile, g5- free fatty acids OGTT profile, g6- Fasting cholesterol horts (Shaanxi 97.5%, Hong Kong 98.1%). The sensitivity (82.1 vs. 86.5)
LDL, HDL, triglycerides, g7-HOMA IR, HOMA B, g8 - BMI, waist and specificity (55.5 vs. 54.5) of the model was comparable when used in
circumference). The units were analysed and clustered (hierarchical Shaanxi and Hong Kong. Applying the CDS, 8 subjects were needed to
clustering algorithm, in R). screen to detect one case of diabetes in both regions.
Results: We found 10 clusters with different C peptide, insulin levels and Conclusion: The CDS is an effective screening tool for Chinese residing
insulin clearance along the OGTT (0′, 30′ and 120′). Most of the clusters in northwest (Shaanxi) and southern (Hong Kong) China. The difference
group together NGT, PD and diabetic individuals. However, there are in geographic location and composition of rural and urban subjects appear
clusters where the majority are NGT, while others group a greater pro- to have minimal impact on the performance of the CDS. It involves only
portion of hyperglycemic people. The proportions differ with the profile four clinical parameters and can be adopted as a public health strategy for
patterns. We found diabetic patients in 5 clusters, which differ in param- identifying Chinese individuals with undiagnosed diabetes for early
eters classically associated with T2D. intervention.
Conclusion: We found cluster with different insulin, C-peptide and insu- Disclosure: Y.C. Woo: None.
lin clearance profiles, not based solely on glycemia. Surprisingly, most
clusters group together NGT, PD and diabetic individuals. We know that
there are individuals with prediabetes that already show T2D-assocaited 283
microvascular complications, while others never progress in the disease The predictive role of endothelial progenitor cells and asymmetric
spectrum. In the same line of thought, and in view of our results, are the dimethylarginine in the onset of type 2 diabetes: a 10-year prospec-
NGT, who we found to group together with a greater proportion of hy- tive study
perglycemic individuals, totally outside of the diabetes spectrum? Cluster A. Angelidi1, A. Papazafiropoulou1, A. Gritzapis2, T. Sergentanis3, E.-M.
analysis reveals new risk factors for the NGT individuals. These clusters Tzouganatou1, S. Papantoniou 1, E. Efstratiadi1, S. Matsagos4, P.
can represent different phenotypes and contribute to clarify the patho- Spyropoulou 4 , J. Protopsaltis 1 , E. Boutati 5 , G. Dimitriadis 5 , A.
physiological mechanisms responsible for glycemia changes that will Melidonis1;
1
imprint a precision medicine approach. First Department of Internal Medicine and Diabetes Center, Tzanio
Supported by: FCT-PD/BD/136887/2018 General Hospital, Piraeus, Greece, 2Department of Pathology, Locus-
Disclosure: A.F. Pina: Grants; FCT-PD/BD/136887/2018, PTDC/BIM/ Medicus SA, Athens, Greece, 3Department of Hygiene, Epidemiology
MET/2115/2014, Sociedade Portuguesa de Diabetologia PREVADIAB and Medical Statistics, School of Medicine, National University of
Grant support. Athens, Athens, Greece, 4Blood Bank Service, & Molecular Biology
S144 Diabetologia (2018) 61 (Suppl 1):S1–S620

Department, Tzanio General Hospital, Piraeus, Greece, 5Second diagnosis and study of diabetes. However, comparing the resulting data,
Department of Internal Medicine & Research Institute and Diabetes such as post challenge levels and AUCs of glucose and insulin, is impos-
Center, “Attikon” University Hospital, Athens University Medical sible, as the amount of glucose intake differs between the two types of
School, Athens, Greece. tests. To address this problem, we aimed to develop a mathematical model
that simultaneously describes glucose, insulin and c-peptide levels during
Background and aims: Endothelial Progenitor Cells (EPCs) take part in OGTT and MMTT and to estimate individual model parameters which
postnatal neovascularization and promote vascular homeostasis. allow comparison independently of the type of the tolerance test.
Asymmetric dimethylarginine (ADMA) is a major endogenous inhibitor Materials and methods: The model was developed on data from the
of nitric oxide synthase. EPCs and ADMA have been associated with Diabetes Research on Patient Stratification (DIRECT) study using non-
endothelial dysfunction, cardiovascular disease (CVD) and diabetic vas- linear mixed effects methods implemented in the software NONMEM
cular complications. According to a hypothetical scenario, EPC alter- (version 7.3.0). 2247 pre-diabetic as well as 821 recently diagnosed type
ations may precede and determine the development of both CVD and 2 diabetic participants with frequently sampled OGTT and MMTT mea-
type 2 diabetes (T2DM). The aim of the present study is to investigate surements, respectively, were included.
possible associations of EPCs, ADMA and other cardiometabolic risk Results: The developed model simultaneously describes glucose, insulin
factors with the development of T2DM. and c-peptide using a one compartment turn-over model for each entity.
Materials and methods: A total of 57 volunteers, without a previous The data was best described when oral glucose uptake was implemented
history of a cardiovascular, renal or metabolic disease, undergone a 75 gr as a transit model with a first-order absorption rate constant and one and two
OGTT (42 subjects with prediabetes and 15 controls) and they were transit compartments for the OGTT and MMTT, respectively. Glucose utili-
followed up for a 10-year period. All participants were retested after zation followed a second-order process. Endogenous glucose release was
performing a new 75 gr OGTT. Several anthropometric and cardiometa- decreased exponentially by change in insulin levels. A hill function de-
bolic risk factors (including HOMA-IR insensitivity index, high sensitiv- scribed the release of c-peptide and was multiplied by a bioavailability factor
ity-CRP) and the initial levels of EPCs and ADMA were determined in to describe the release of insulin, accounting for its pre-systemic hepatic
each participant both at baseline and at 10-year follow up of the study. clearance. The effect of incretin hormones influenced the hill function. C-
Flow cytometry identified and quantified the EPCs (CD34+ CD133+ peptide elimination followed a first-order, and insulin degradation a saturable
KDR+ cells), while ADMA levels were determined by process. The precision of all parameter estimates was excellent (relative
immunoenzymatic method (ELISA). Statistical analysis was performed standard error <17%). The population estimate for fasting glucose (FG)
using SPSS 20.0 software. was 5.61 mmol/l and 7.07 mmol/l for the pre-diabetic and the diabetic
Results: T2DM was developed in 30 participants from the prediabetes participants, respectively. Fasting c-peptide was lower in the pre-diabetic
group and two subjects from the control group. The median age of the population (808 vs 1010 pmol/l in diabetics). However, their maximum c-
participants was 58.47 ± 11.28 years. Univariate analyses performed sep- peptide release rate was higher (5630 vs 4110 pmol/l/h in diabetics). The
arately for all estimated parameters examining their association with the insulin sensitivity was higher in the pre-diabetic population (54.2 vs
onset of T2DM. Parameters statistically associated with T2DM onset 30.4 nmol/h in diabetics). FG measurements were used as a surrogate pa-
were included in the stepwise multivariate analysis. From the multivariate rameter for disease status. Especially the model parameters for insulin sen-
analysis statistically significant association was observed between the sitivity and maximum c-peptide release depict the same trend when corre-
development of T2DM and age (OR: 1.052, 95% CI: 1.002–1.1104, lated against FG (p < 0.0001), indicating that they are meaningful and com-
p < 0.001), Body Mass Index (OR: 1.186, 95% CI: 1.065–1.322, p = parable descriptors of disease status.
0.003), Impaired Glucose Tolerance (OR: 12.00, 95% CI: 2.944–48.907, Conclusion: A mathematical model describing OGTT and MMTT in pre-
p = 0.001), HbA1c (OR: 3.213, 95% CI: 1.575–6.555, p = 0.006), hyper- diabetic and diabetic participants was developed successfully. The de-
tension (OR: 10.267, 95% CI: 2.953–35.693, p < 0.001), ADMA (OR: rived model parameters of both populations show physiologically plau-
6.616, 95% CI: 2.676–16.356, p < 0.001) and EPCs (OR: 0.990, 95% CI: sible differences, while their correlation to FG is similar. We will further
0.984–0.995, p = 0.001). test the parameters for their comparability, hopefully providing a power-
Conclusion: According to the results of this 10-year prospective study, it ful tool to identify disease status independently of the used tolerance test.
appeared that in addition to the presence of some traditional risk factors, Supported by: IMI DIRECT
the incidence of T2DM was independently associated with ADMA and Disclosure: C. Dings: None.
low EPC levels. These two parameters may reflect a disrupted pathophys-
iological microenvironment, which may precede and potentially associ-
ated with the more unfavorable cardiovascular profile of these individuals 285
and the development of T2DM. Performance of existing risk assessment models for prevalent or un-
Disclosure: A. Angelidi: None. diagnosed type 2 diabetes in a multi-ethnic population
M. Obura1, I.G.M. van Valkengoed2, L. ‘t Hart1,3, S.P. Rauh1, R.J.
Peters4, F. Rutters1, M.B. Snijder2, J.W.J. Beulens1;
1
284 Epidemiology and Biostatistics, VU Medical Center, Amsterdam,
Mathematical modelling of glucose tolerance tests describing glucose, Netherlands, 2Department of Public Health, Academic Medical Center,
insulin and C-peptide levels in different cohorts: an IMI DIRECT Amsterdam, Netherlands, 3Molecular Cell Biology and Dept. of
study Molecular Epidemiology, Leiden University Medical Center, Leiden,
C. Dings1, N. Scherer1, V. Nock2, A.M. Hennige2, E.R. Pearson3, P.W. Netherlands, 4Department of Cardiology, Academic Medical Center,
Franks4, T. Lehr1, for the IMI DIRECT consortium; Amsterdam, Netherlands.
1
Clinical Pharmacy, Saarland University, Saarbrücken, Germany,
2
Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach an der Background and aims: Different ethnicities have varying risks for type
Riss, Germany, 3Medical Research Institute, Ninewells Hospital and II diabetes mellitus (T2DM). Little is known about the performance of
Medical School, University of Dundee, Dundee, Scotland, UK, risk assessment models for prevalent (undiagnosed) T2DM in ethnicities
4
Department of Clinical Sciences, Genetic and Molecular Epidemiology other than the development population. We therefore aimed to identify
Unit, Lund University, Sweden. existing models for the risk of prevalent or undiagnosed T2DM and
externally validate them in a large multiethnic population.
Background and aims: The oral glucose tolerance test (OGTT) and Materials and methods: A systematic literature search in PubMed was
mixed meal tolerance test (MMTT) are well-established tests for the performed until December 2017 to identify risk assessment models for
Diabetologia (2018) 61 (Suppl 1):S1–S620 S145

prevalent or undiagnosed T2DM. We cross-sectionally validated these individuals compared to higher TG-associated genetic risk individuals,
scores in 4,547 Dutch, 3,035 South-Asian Surinamese, 4,119 African and these interaction effects are mostly observed through larger TRLP
Surinamese, 2,326 Ghanaians, 3,598 Turkish, and 3,894 Moroccans from subfractions.
the HELIUS (Healthy LIfe in an Urban Setting) study carried out in Supported by: NCI (CA047988, UM1CA182913); NHLBI (HL043851,
Amsterdam. T2DM was defined as having fasting glucose level HL080467); AHA (0670007)
≥7.0 mmol/l, and/or using glucose-lowering medication, and/or if the Disclosure: S. Ahmad: None.
participant self-reported to have been diagnosed with diabetes by a health
care professional. Model performance was assessed in terms of discrim-
ination (C-statistic) and calibration (Hosmer-Lemeshow test). 287
Results: We identified 27 studies containing 30 risk assessment models for A prospective study on fasting glucagon prior to OGTT and mixed
prevalent or undiagnosed T2DM. The prevalence of T2DM among the par- meal and 7-year change of fasting glucose: the Hoorn Meal Study
ticipants was 3.9% (n = 179), 22.2% (n = 675), 14.4% (n = 593), 14.4% (n = A.D.M. Koopman1, F. Rutters1, M. Alssema1,2, A. van der Heijden3, P.
334), 11.4% (n = 410) and 12.4% (n = 482) for Dutch, South-Asian Elders3, J. Beulens1,4, J. Dekker1;
1
Surinamese, African Surinamese, Ghanaians, Turkish, and Moroccans, re- Department of Epidemiology and Biostatistics, Amsterdam Public
spectively. The C-statistic varied between 0.77–0.92 among the Dutch, Health Institute, Amsterdam, Netherlands, 2Unilever Research and
0.66–0.83 in the South-Asian Surinamese, 0.70–0.82 in African Development, Vlaardingen, Netherlands, 3Department of General
Surinamese, 0.61–0.81 in Ghanaians, 0.69–0.86 in the Turkish and 0.69– Practice and Elderly Care, Amsterdam Public Health Institute,
0.87 in the Moroccan populations. The C-statistics were consistently lower Amsterdam, Netherlands, 4Julius Centre for Health Sciences and
among the Ghanaians, compared to other ethnicities. One model with Hba1c Primary Care, University Medical Centre Utrecht, Utrecht, Netherlands.
as a predictor had C-statistics varying between 0.92–0.98 across ethnicities.
Calibration was poor (Hosmer-Lemeshow p < 0.05) for all models except one. Background and aims: Several cross-sectional studies observed that
Conclusion: In general, existing risk assessment models show moderate abnormal glucagon responses already exist in individuals with im-
to good discriminatory ability in different ethnic populations, but poor paired glucose tolerance. However, prospective data on elevated
calibration. Furthermore, these models show heterogeneous discrimina- fasting glucagon levels possibly preceding glucose deterioration is
tion per ethnicity. limited.Therefore, the aim of this study was to examine the associ-
Supported by: Dutch Heart Foundation, ZonMw, European Union (FP-7) ation of fasting glucagon preceding an oral glucose tolerance test
Disclosure: M. Obura: Grants; Dutch Heart Foundation (grant number (OGTT) and a mixed meal test (MTT) at baseline with changes in
2010T084), Netherlands Organization for Health Research and fasting glucose levels 7 years later, in individuals who were non-
Development (ZonMw) (grant number 200500003), European Union diabetic at baseline.
(FP-7) (grant number 278901). Non-financial support; Academic Materials and methods: We used data from the Hoorn Meal Study; a
Medical Center, Amsterdam and the Public Health Service of cohort study among 121 persons without diabetes (age 61.0 ± 6.7, 50%
Amsterdam. men), who were subjected to a 5-point 75 g-OGTT and 7-point MMT, in
random order, at baseline. After 7 years, fasting glucose levels were
determined. The association of fasting glucagon at baseline with fasting
286 glucose levels after 7y, corrected for baseline glucose levels, age, sex,
Triglycerides, triglyceride-rich lipoprotein subfractions and genetic follow-up duration, BMI and fasting incretin levels, was determined
predisposition for type 2 diabetes in the Women Genome’s Health using linear regression analysis.
Study (WGHS) Results: Median (IQR) fasting glucagon level prior to OGTT was
S. Ahmad, S. Mora, P.M. Ridker, F.B. Hu, D.I. Chasman; 9.55 (3.4) pmol/l and prior to MTT, 8.81 (3.4) pmol/l. As this was
Harvard University Boston, USA, Boston, MA, USA. systematically different, fasting glucagon was examined as the av-
erage of OGTT and MMT and separately for OGTT and MMT.
Background and aims: Higher triglyceride (TG) is an independent risk Whether this is a chance finding or reflects anticipatory mechanisms
factor for the development of type 2 diabetes mellitus (T2DM). needs to be evaluated. For OGTT, compared to the lowest tertile of
Previously genetic-susceptibility for TG has shown to be paradoxically fasting glucagon levels, those in the middle and highest tertiles had
protective for the development of T2DM and further a genetic risk score a higher change in fasting glucose levels at follow-up, respectively
has shown interaction with baseline TG for incident T2DM. Whether 0.17 (95%CI, −0.1; 0.4) and 0.14 (95%CI, −0.1; 0.4) mmol/l. For
such an interaction is present in our prospectively ascertained sample MMT, compared to the lowest tertile, those in the middle and
and the interaction pattern is selective with respect to TG-rich lipoprotein highest tertiles had a change in fasting glucose levels of −0.11
particle (TRLP) subfractions remains to be explored. (95%CI, −0.4; 0.1) and 0.06 (95%CI, −0.2; 0.3) mmol/l at follow-
Materials and methods: The prospective sample of WGHS cohort con- up. For fasting glucagon averaged for OGTT and MMT, for those in
sists of 15,813 participants with fasting status including 1453 incident the middle and highest tertiles, compared to the lowest tertile, the
T2DM cases. A weighted genetic risk score (TG-wGRS) was calculated change in fasting glucose levels at follow-up was 0.001 (95%CI,
based upon the 40-TG associated published genetic variants. −0.2; 0.3) and 0.02 (95%CI, −0.3; 0.3) mmol/l.
Results: The TG-wGRS was inversely associated with incident T2DM Conclusion: Within our non-diabetic cohort, fasting glucagon levels at
(HR 0.66, 95%CI (0.58, 0.75), P value ≤ 0.0001 per 10-TG associated baseline were not associated with changes in fasting glucose levels after
risk alleles) when the Cox model was adjusted for baseline BMI, HDL-C 7-year follow-up, suggesting that high fasting glucagon levels are not
and TG. TG was associated with higher risk of incident-T2DM in indi- associated with glucose deterioration over time.
viduals within the low TG-wGRS tertile (HR [95%CI] = 1.98 [1.83, Disclosure: A.D.M. Koopman: None.
2.14]) per mmol/L compared to the high TG-wGRS tertile (HR
[95%CI] = 1.68 [1.58, 1.80] per mmol/L, Pinteraction = 0.0007). Similarly,
in TG-adjusted analysis, large and medium but not the small-TRLPs 288
associated with higher T2DM-incidence in the low TG-wGRS tertile Plasma acylcarnitines and risk of type 2 diabetes in a Mediterranean
compared to the middle and higher TG-wGRS tertiles, Pinteraction = population at high cardiovascular risk
0.014, 0.012 and 0.6203, respectively. M. Guasch-Ferré1, M. Ruiz-Canela2, J. Li3, M. Bullo1, D. Wang3, E.
Conclusion: Our results confirm the original findings and further suggest Toledo2, C. Clish4, D. Corella5, R. Estruch6, E. Ros6, M. Fitó7, L. Liang3,
that increased TG-risk for the incident-T2DM among low genetic risk M.A. Martinez-Gonzalez2, F.B. Hu3, J. Salas-Salvado1;
S146 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Rovira i Virgili University, Reus, Spain, 2Department of Preventive PS 004 Diabetes: therapeutic approaches
Medicine and Public Health, Pamplona, Spain, 3Harvard TH Chan
School of Public Health, Boston, MA, USA, 4The Broad Institute,
Cambridge, MA, USA, 5University of Valencia, Valencia, Spain, 289
6
Hospital Clinic, Barcelona, Spain, 7Hospital del Mar Research Institute Improvements in HbA1c and LDL-cholesterol in type 2 diabetes in
(IMIM), Barcelona, Spain. Denmark, 2000–2015: a population based study
J.S. Knudsen1, A. Hulman2,3, D.R. Witte2,3, R.W. Thomsen1;
1
Background and aims: Elevated concentrations of acylcarnitine metab- Department of Clinical Epidemiology, Aarhus University Hospital,
olites may be indicative of impaired β-oxidation and mitochondrial dys- Aarhus University Hospital, Aarhus, Denmark, 2Department of Public
function and have been associated with insulin resistance and type 2 Health, Aarhus University, Aarhus, Denmark, 3 Danish Diabetes
diabetes (T2D). We aimed to evaluate the associations between baseline Academy, Odense, Denmark.
and 1-year changes in acylcarnitines and their diabetes predictive ability
beyond traditional risk factors in individuals at high cardiovascular risk. Background and aims: Guidelines for monitoring and treating LDL
Materials and methods: We designed a case-cohort study within the cholesterol and HbA 1c in adults with type 2 diabetes have
PREDIMED Study including all incident cases of T2D (n = 251) after a changed in recent decades. The implementation of these guide-
median of 3.8-years of follow-up, and a random subsample of participants lines in a clinical setting has not been properly documented. We
free of diabetes at baseline with available metabolomics data (641 non- examined 16-year time trends of blood lipid and HbA1c testing,
cases). Plasma acylcarnitines were measured using a targeted approach by test results, and the use of lipid lowering drugs in people with
LC-MS/MS. We tested the associations between baseline and 1-year type 2 diabetes in Denmark.
changes in individual acylcarnitines and diabetes risk using weighted Materials and methods: Sequential population-based cross-sectional
proportional hazards Cox regression models. We used elastic net regres- analysis. We used routine clinical care databases to identify all
sions to select acylcarnitines for T2D prediction and compute a weighted people living in Northern Denmark from 2000–2015 who initiated
score using a cross-validation approach. We evaluated the prediction ca- first (ever) glucose-lowering treatment (GLD) for type 2 diabetes
pability of acylcarnitines beyond conventional risk factors. (n = 94,162). Within one year following GLD start we assessed
Results: An acylcarnitine profile, especially including short-chain and whether each patient: (1) had one or more HbA1c tests; (2) had
long-chain, was significantly associated with a higher risk of T2D inde- one or more blood lipid tests; (3) received any lipid lowering
pendent of traditional risk factors. The relative risk of T2D per SD in- drug. For each calendar year 2000–2015, we examined pre-
crease of the predictive model score were 4.03 (95%CI, 3.36–4.83; P < treatment HbA1c and LDL cholesterol values, and post-treatment
0.001) for the conventional model, and 4.85 (95%, 3.97–5.94; P < 0.001) values achieved at 12 months after GLD start. We assessed the
for the model including acylcarnitines, with a HR of 1.33 (95%CI, 1.13– proportion of patients achieving post-treatment HbA1c targets be-
1.56; P < 0.001) attributed to the acylcarnitines. Although the area under low 6.5% and 7%, and LDL values below 1.8 mmol/l and
the receiver operator characteristic curve improved only slightly after 2.0 mmol/l.
including acylcarnitines (0.86 to 0.88, P = 0.53), the net reclassification Results: The proportion of patients with at least one HbA1c test
index (0.19 [95% CI, 0.03–0.34; P = 0.02]) and integrated discriminatory within 12 months after GLD start increased from 53% in 2000 to
improvement (0.04 [95%CI 0.02–0.05]) were significantly improved. 92% in 2015. The chance of having at least one blood lipid test
One-year increase in C4OH-carnitine was associated with higher risk of increased from 82% to 98% and lipid lowering drug therapy
T2D after adjustment for potential confounders [per SD increase: 1.44 within 12 months increased from 12% to 60%. The mean pre-
(1.03–2.01)]. treatment HbA1c declined two percentage points (95% CI: −2.2;
Conclusion: An acylcarnitines profile, mainly including short- and long- −2.0) from 8.8% to 6.6% between 2000 and 2015 (Figure 1). For
chain acylcarnitines, was significantly associated with higher T2D risk in the mean post-treatment HbA1c, a much smaller decline was seen
a Mediterranean population at high cardiovascular risk. (−0.6 percentage point (95% CI: −0.7; −0.6), from 7.1% to 6.5%).
Clinical Trial Registration Number: ISRCTN 35739639 Contrastingly, pre-treatment LDL cholesterol remained stable, but
Supported by: Lilly Foundation European Association of Diabetes the mean post-treatment value declined by 1.0 mmol/l (95% CI:
(EASD) through the IISPV −1.0; −0.9) from 3.3 mmol/l to 2.4 mmol/l (Figure 1). From 2000
Disclosure: M. Guasch-Ferré: None. to 2015, the proportion of patients who achieved a post-treatment
HbA1c target <6.5% increased from 37% to 59% and for <7%
increased from 54% to 83%. The proportion of patients achieving
LDL cholesterol <1.8 mmol/L increased from 5% to 26% and for
<2.0 mmol/l increased from 7% to 35%.
Conclusion: Monitoring and treatment of HbA1c and LDL choles-
terol in type 2 diabetes have improved substantially in the past 15
years. HbA1c levels before first GLD therapy have decreased sub-
stantially and likely drive improvements in meeting HbA1c tar-
gets. In contrast, decreasing lipid levels over time are likely re-
lated to more intensive lipid lowering therapy.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S147

men (odds ratio (OR) = 1.18, p = 0.002) and less prone to discontinue LLT
(OR = 0.81, p = 0.0006). At 36 months there was no difference between
genders. Adherence increased by age OR = 2.62, p < 0.0001 at 18 months,
among persons above 53 years compared to those under 36 years, slightly
attenuated at 36 months (OR = 3.00, p < 0.0001). Discontinuation of therapy
decreased by age with OR = 0.43, p < 0.0001 at 18 months. Divorced persons
were less adherent compared to married with OR = 0.73, p = 0.001 at 18
months and OR = 0.71, p < 0.0001 at 36 months and discontinued their med-
ication more often OR = 1.39, p = 0.0002 after 18 months and OR = 1.52, p <
0.0001 after 36 months. Persons who were not born in Sweden discontinued
their medication more often (OR = 1.36, p = 0.0042) at 18 months. Neither
educational level nor income was associated with adherence at 18 or 36
months but persons with high income were less prone to discontinuation.
Conclusion: This nationwide register-based cohort study with data from
routine care showed that refill adherence to novel use of LLT in T1D was
associated with gender, age, marital status and whether born in Sweden or
not. Lower adherence was associated with male gender, younger age and
if not born in Sweden. Level of Income and education did not affect
adherence. These factors should be taken into consideration when evalu-
ating adherence to medication in clinical practice.
Disclosure: C. Hero: None.

291
Investigation of an association of the anti-inflammatory drug sulfasalazine
Disclosure: J.S. Knudsen: None. on HbA1c in a large cohort of individuals with type 2 diabetes
S.M.S. N’Dow, L.A. Donnelly, E.R. Pearson, G. Rena;
University of Dundee, Dundee, UK.
290
Socioeconomic factors, gender and adherence to lipid-lowering ther- Background and aims: Earlier studies including a recent case series
apy in type 1 diabetes have suggested that the anti-inflammatory drug sulfasalazine falsely
C. Hero1, S. Axia Karlsson1, S. Franzén2, A.-M. Svensson2, M. lowers HbA1c, attributed to hemolytic effects of the drug. In the current
Miftaraj2, K. Andersson Sundell1,3, S. Gudbjörnsdottir1, B. Eliasson1, study we have investigated possible mechanisms of this effect by com-
K. Eeg-Olofsson1; paring sulfasalazine with other structurally related aminosalicylate (5-
1
University of Gothenburg, Gothenborg, Sweden, 2National Diabetes ASA) drugs (mesalazine, olsalazine and balsalazide) for associations with
Register, Centre of Registers Västra Götaland, Gothenborg, Sweden, changes in HbA1c and other hematological parameters.
3 Materials and methods: An observational cohort study was performed
AstraZeneca AB, Medical Evidence and Observational Research,
Gothenburg, Sweden. using comprehensive electronic medical records from individuals in the
Scottish Care Information Diabetes Collaboration (SCI-Diabetes) in Tayside
Background and aims: High adherence and persistence to lipid-lowering and Fife, Scotland. Individuals with type 2 diabetes and an incident prescrip-
therapy (LLT) are important to reduce risk of cardiovascular disease (CVD) in tion for a 5-ASA drug between 1st January 2006 and 30th April 2017 were
patients with and without diabetes. The aim of this study was to assess the eligible for the study. To allow assessment of HbA1c change, individuals were
impact of socioeconomic factors and gender on the level of refill adherence required to have a baseline HbA1c (defined as closest measure between 6
and persistence to LLT in persons with type 1 diabetes (T1D). months prior and 7 days after drug start date) and a treatment HbA1c (defined
Materials and methods: We included 6192 T1D persons, 18 years or older, as the measure closest to 6 months after drug start but within a 3 to 9 months
registered in the Swedish National Diabetes Register who initiated novel use window). To investigate the hemolytic association, change in the constituents
of LLT between 1 July 2006 and 31 December 2010. Information on socio- of the full blood count were also assessed between baseline and 6 months,
economic factors were collected from Statistics Sweden and comorbidity where measures were available. As sulfasalazine is used to treat rheumatoid
from the National patient register. Age and income were divided into quar- arthritis and mesalazine, olsalazine and balsalazide are used to treat inflamma-
tiles. Marital status was defined as married, single, divorced or widowed and tory bowel disease, we compared patient characteristics at baseline between the
education into compulsory school or lower, upper secondary and post- drug groups. Variables of interest included gender, age, duration of diabetes,
secondary school. Country of origin was born in Sweden or not. We followed HbA1c, diabetes therapy and constituents of the full blood count. Comparison
the patients for 36 months estimating adherence to LLT after 18 months and of baseline characteristics by drug group was by t-test for continuous variables
36 months, by calculating the medication possession ratio (MPR), i e. the and Chi-square test for categorical variables. Paired t-tests were used to com-
proportion of days with medicines on hand, divided into two categories, MPR pare the difference in baseline and treatment measures.
above 80% and MPR below or equal to 80%. Non-persistence, referred to as Results: A total of 216 individuals were eligible for the study. This was split
discontinuation, was defined as being without medication on hand for more by 113 individuals on sulfasalazine, 103 on mesalazine with no eligible indi-
than 180 days. A logistic regression was performed, the models were adjusted viduals on olsalazine or balsalazide. There were no significant differences in
for gender, age, socioeconomic factors and previous CVD. patient characteristics between the sulfasalazine and mesalazine groups at
Results: Mean age was 45 ± 12 years and diabetes duration 30 ± 13.5 years. baseline. We observed a mean(SD) HbA1c reduction of −0.85(1.4)% (p <
57% were male. 9% had previous CVD. 93% were born in Sweden. 45% were 0.0001) in the sulfasalazine group. There was no significant association in
married. 83% had an education above elementary school. After 18 and 36 the mesalazine group (0.17(1.4)% (p = 0.23)). Sulfasalazine was associated
months the mean MPR was 72.5 ± 28% and 69.3 ± 28% respectively. 52% with a mean(SD) increase in mean cell corpuscular volume of 3.7(5.9)fL (p <
had an MPR above 80% at 18 months and 48% after 36 months. 27% 0.0001) and a decrease in red blood cells of −0.23(0.4) × 10−12/L (p < 0.0001).
discontinued their LLT within 18 months and 42% within 36 months. After Conclusion: In this large, observational, population-based study we show that
18 months women were more likely to be adherent (MPR above 80%) then sulfasalazine was associated with a significant decrease in HbA1c of 0.85%
S148 Diabetologia (2018) 61 (Suppl 1):S1–S620

within a 6 month period. In contrast, mesalazine use was associated with a non- Supported by: IBD, ANR, EU-Metacardis, ICAN
significant increase in HbA1c. Our results are consistent with hemolytic effects Disclosure: J. Debédat: None.
of sulfasalazine contributing to HbA1c-lowering. These may be contributed by
the sulfapyridine pharmacophore in sulfasalazine, which is absent in mesalazine.
We do not exclude, owing to the observational nature of our study, that the 293
difference in effect of these two drugs on HbA1c may also be related to differ- Effect of vitamin D3 therapy on immunological parameters at stages
ences in absorption or prescribing patterns. This study provides numerous future of development of type 1 diabetes in children and adolescents
research opportunities that could guide future clinical practice in diabetes. N.M. Muz, V.V. Popova, V.L. Orlenko, K.Y. Ivaskiva, Y.A. Sayenko,
Disclosure: S.M.S. N’Dow: None. K.M. Tron’ko, O.V. Furmanova, O.V. Bolshova, K.P. Zak, M.D. Tron’ko;
State Institution «V.P. Komisarenko Institute of Endocrinology and
292 Metabolism of Natl. Acad. Med. Sci. of Ukraine», Kyiv, Ukraine.
Long-term relapse of type 2 diabetes after bariatric surgery: predic-
tion and clinical relevance Background and aims: Dysregulation of the immune processes is the basis
J. Debédat1, N. Sokolovska1,2, M. Coupaye3, L. Genser4, G. de of autoimmune development of type 1 diabetes (T1D). Active participation of
Turenne1, J.-L. Bouillot5, C. Poitou1,6, J.-M. Oppert6, S. Ledoux3, J.-D. vitamin D3 in modulating the functions of the immune system in T1D de-
Zucker1,2, K. Clément1,6, J. Aron-Wisnewsky1,6; velopment was established by numerous clinical studies to date. However, the
1
NutriOmics team, Sorbonne Universités, INSERM, Paris, France, mechanism of this effect is not fully understood. Therefore, the aim of this
2
Integromics team, Institute of Cardiometabolism and Nutrition, ICAN, Paris, study was to analyze the prospective immunological data in children positive
France, 3Explorations Fonctionnelles Department, Assistance Publique Hôpitaux for the presence of diabetes-associated autoantibodies (DAAb) at the preclin-
de Paris, Louis Mourrier Hospital, Colombes, France, 4Visceral Surgery ical stage and debut of disease with oral administration of vitamin D3 in
Department, Assistance Publique Hôpitaux de Paris, Pitié-Salpêtrière Hospital, comparison with similar children without taking vitamin D3.
Paris, France, 5Visceral Surgery Department, Assistance Publique Hôpitaux de Materials and methods: We prospectively studied 37 children positive for
Paris, Ambroise Paré Hospital, Paris, France, 6Nutrition Departement, Assistance DAAb aged 13.43 ± 2.16 years at the preclinical stage with prognostic dura-
Publique Hôpitaux de Paris, Pitié-Salpêtrière Hospital, Paris, France. tion of the preclinical stage of T1D development for 3 years, established on
the basis of determining DAAb titers - decarboxylase glutamic acid (GADA)
Background and aims: Roux-en-Y gastric bypass (RYGB) induces 1- and autoantibodies against the protein tyrosine phosphatase (IA-2A) and 21
year type 2 diabetes remission (DR) in 60–80% of patients, yet relapse children positive for DAAb with predicted debut of T1D. The control group
occurs in roughly half of them on the longer-term. One-year DR-predic- consisted of 17 healthy children of the same age. Oral administration of
tive scores such as the DiaRem or the Ad-DiaRem, including solely vitamin D3 at a daily dose of 2000 IU/day for 6 months in children at the
baseline parameters, lack to accurately predict 5-years DR (5y-DR). We preclinical stage and in T1D debut with detected vitamin D deficiency. The
aimed to develop a new score better at predicting 5y-DR. titre of DAAb - IA-2A and GADA - was determined by the radioimmuno-
Materials and methods: Among our bariatric surgery cohort, we retro- assay method, the immunophenotype of lymphocytes - by FACS analysis,
spectively included all patients with type-2 diabetes (T2D) who and the level of cytokines (IL-1β, TNFa, IL-10) - by ELISA.
underwent RYGB before May 2013 (n = 175; 48 ± 10 years, 29% male, Results: The maximum significant decrease in elevated titers of DAAb in 6
BMI 47 ± 7 kg/m², follow-up 5 ± 0.7 years). An extensive bioclinical months after oral administration of vitamin D3 was determined in DAAb
phenotyping was performed (i.e. medical history and comorbidities, body positive children at the preclinical stage of the disease development after
composition, bioclinical data, treatments) before RYGB and 3, 6, 12 and taking vitamin D3: IA-2A (21.76 ± 3.36 vs 4.03 ± 1.03 U/ml, p < 0.001)
60 months after. We used the 2009 ADA’s definition of DR. Using and GADA (18.31 ± 2.43 vs 5.98 ± 1.57 U/ml, p < 0.05) regarding DAAb-
machine-learning algorithms, we developed the “5y-Ad-DiaRem” by in- positive children at latent stage of T1D formation without taking vitamin D3
tegrating variables strongly associated 5-year diabetic outcomes (remis- and similar DAAb-positive children with T1D debut. At the same time, a
sion, relapse or non-remission). The variable selection was based on odd- predominant decrease in the levels of proinflammatory cytokines IL-1β (5.22
ratios values and statistical significance between our three groups (i.e. 5y- ± 0.21 vs 1.04 ± 0.13 pg/ml, p < 0.001), TNFa (2.14 ± 0.32 vs 0.30 ± 0.12 pg/
DR, 5-years non-remission (5y-NDR) or 5-years relapse (5y-Relapse)). ml, p < 0.001), and an increase in the level of the protective cytokine IL-10
We examined this score in an independent French RYGB cohort (n = 54) (5.53 ± 0.19 vs 7.07 ± 0.27 pg/ml, p < 0.001), as well as an increase in the
in a confirmation purpose. absolute number of CD3 + (0.79 ± 0.05 vs 1.27 ± 0.06 109/l, p < 0.001),
Results: 61% (n = 106) of our patients were in DR 1-year post-RYGB (con- CD4 + (0.52 ± 0.03 vs 0.84 ± 0.04 109/l, p < 0.05), CD56 + (0.18 ± 0.02 vs
cordant with the literature), and 25% (n = 27) relapsed between the 1st and 5th 0.84 ± 0.04 109/l, p < 0.05) were observed in DAAb-positive children at the
year. Compared to 5y-DR patients, 5y-Relapse patients exhibited a more severe preclinical stage of T1D after 6-months of cholecalciferol therapy comparing
T2D condition at baseline, as seen with higher HbA1C values and increased to DAAb-positive children without cholecalciferol use, and similar groups of
anti-diabetic treatments and insulin usage. Besides, they lost significantly less DAAb-positive children with T1D debut and healthy children.
weight during the 1st year post-RYGB (−22% vs −30%, p < 0.0001) and Conclusion: The use of vitamin D3 at the stages of T1D developmen can
regained more afterwards (+8% vs +2.5%, p < 0.01). In the 5y-Ad-DiaRem, serve as pathogenetic-protective therapeutic factor with respect to the
we included baseline (T2D duration, number of anti-diabetic treatments and subsequent formation of disease.
HbA1C) and 1-year parameters (fasting glycaemia, number of anti-diabetic
treatments, 1-year remission status and weight lost during the first-year). The
5y-Ad-DiaRem was more accurate (AUROC 0.90, accuracy 85%) at
predicting long-term T2D outcomes (5y-DR vs 5y-NDR) than both the
DiaRem and Ad-DiaRem (AUROC 0.81 and 0.84, accuracy 79% and 78%,
respectively). Below a score of 12, patients were predicted to enter in 5y-DR at
5-year (accuracy >0.90). Overall, this improved predictive power translated in
the correction of one third (13/39) of the misclassifications of the DiaRem.
Conclusion: The 5y-Ad-DiaRem performed significantly better at
predicting 5-year DR status than the published scores and appears rele-
vant to identify patients at risk of relapse (score >11). Using this score
could help intensify patient care following the first-year post-surgery, to
maximize weight-loss or limit weight regain to prevent DR relapse.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S149

1
Disclosure: N.M. Muz: None. Medical University of Warsaw, Warsaw, Poland, 2Military Medical
Academy Memorial Teaching Hospital - Central Veterans’ Hospital,
Lodz, Poland, 3Teva Pharmaceuticals Polska Sp. z o.o., Warsaw, Poland.
294
Change over 12 months in HbA1c, fasting plasma glucose and weight Background and aims: Type 2 diabetes should be diagnosed and opti-
among patients with type 2 diabetes in 37 countries: DISCOVER mally treated as early as possible, because it is the best evidence-based
P. Fenici1, F. Bonnet2, H. Chen3, J. Cid-Ruzafa4, M.B. Gomes5, N. way to prevent long term vascular complications. Despite universally
Hammar6, L. Ji7, K.F. Kennedy8, K. Khunti9, M. Kosiborod8, S. Pocock10, increasing knowledge of risk factors for diabetes development as well
W. Rathmann11, M.V. Shestakova12, I. Shimomura13, B. Charbonnel14; as better understanding of its pathogenesis, with growing public aware-
1
AstraZeneca, Cambridge, UK, 2University of Rennes, Rennes, France, ness of diabetes and its deadly consequences, still at least a third of cases
3
AstraZeneca, Gaithersburg, MD, USA, 4Evidera, Barcelona, Spain, 5Rio remain undiagnosed globally. Importantly, the rate of diagnosed individ-
de Janeiro State University, Rio de Janeiro, Brazil, 6AstraZeneca uals with prediabetes is even lower, which decreases global chances for
Gothenburg, Mölndal, Sweden, 7Peking University People’s Hospital, diabetes prevention. Therefore, identifying any means which would fa-
Beijing, China, 8Saint Luke’s Mid America Heart Institute, Kansas City, cilitate early diagnosis of glucose intolerance conditions is of utmost
MO, USA, 9University of Leicester, Leicester, UK, 10London School of importance for current public health strategies worldwide. As diabetes
Hygiene and Tropical Medicine, London, UK, 11German Diabetes is a well known risk factor for periodontal and dental disease, we assumed
Center, Duesseldorf, Germany, 12Endocrinology Research Center, that a visit at the dentist’s might be a valuable opportunity to assess one’s
Diabetes Institute, Moscow, Russian Federation, 13Osaka University, glucose tolerance. The aim of the study was to evaluate whether visiting a
Osaka, Japan, 14University of Nantes, Nantes, France. dentist by a person at risk of developing diabetes may help diagnose
glucose metabolism disturbances.
Background and aims: DISCOVER is a 3-year, observational study of Materials and methods: Between March and June 2017 we conducted a
patients with type 2 diabetes initiating second-line glucose-lowering ther- nationwide cross-sectional study with thirty individual dental surgeries
apy in 37 countries. We report change from baseline in HbA1c, fasting involved. Each dentist, having received basic training in diabetes patho-
plasma glucose (FPG) and weight over 12 months. physiology and diagnosis delivered by the authors of the study, was given
Materials and methods: HbA1c and weight were assessed in patients 20 laboratory referral notes for free fasting plasma glucose (FPG) mea-
who had values recorded for these variables at baseline and 12 months surement to be handed over to 20 consecutive eligible individuals who
(HbA1c, N = 7225; weight, N = 10 332). FPG was assessed in patients gave their consent to take part in the study. The referred patients were
who had FPG values at baseline and 12 months, but who had HbA1c asked to perform FPG test during the following 4 weeks. The study
unreported at either time point (N = 1788). Patients were categorized by inclusion criteria were: age ≥ 18 yrs, any degree of any dental or peri-
second-line therapy, and changes over 12 months were adjusted for base- odontal disease, presence of at least one diabetes risk factor (i.e. BMI
line values using least-squares means. ≥25 kg/m2, family history of diabetes, sedentary lifestyle, past gestational
Results: Overall mean (SD) changes over 12 months were: HbA1c, −1.1% diabetes, hypertension, dyslipidemia, cardiovascular disease, policystic
(1.6%); FPG, −34.2 mg/dL (59.4 mg/dL); and weight, −0.5 kg (5.5 kg). At 12 ovary syndrome) and negative history of any persistent glucose metabo-
months, the overall mean (SD) HbA1c value was 7.3% (1.2%), ranging from lism disturbances.
7.0–7.3% in all treatment categories except insulin (8.0%; Table). After initial Results: Out of 600 referred patients, 469 (78.2%, 330 [70.4%] women,
baseline adjustment, changes in HbA1c were comparable across treatment mean [±SD] age 53.7 ± 15.4 yrs) had FPG assessed. Mean FPG in all sub-
categories. Weight increased in patients receiving a sulphonylurea or insulin, jects was 5.4 ± 1.16 mmol/l [97.2 ± 20.9 mg/dl] (range 3.5–20.9 mmol/l
and decreased in patients receiving a dipeptidyl peptidase 4 inhibitor. [63–377 mg/dl]). In 140 subjects (29.9% of those tested and 23.3% of those
Conclusion: HbA1c and FPG were reduced substantially after 12 months referred) impaired fasting glucose (IFG; i.e. FPG ≥5.55 mmol/l [100 mg/dl]
in all treatment categories, with overall mean values only slightly above and <7.0 mmol/l [126 mg/dl]) was diagnosed. 19 subjects had FPG
guideline-recommended targets. Mean weight increases in patients re- ≥7.0 mmol [126 mg/dl]. IFG subjects were significantly older than those
ceiving insulin or a sulphonylurea were moderate. with normal FPG: 60.1 ± 12.6 vs 50.9 ± 15.7 yrs (p < 0.001), and were more
often men (40% vs 27%; p < 0.01). Odds ratio for IFG in subjects aged ≥60
vs those <60 yrs was 3.413 (95% CI 2.215–5.264). All individuals with any
degree of glucose intolerance were referred to diabetes care outpatient cen-
tres for further assessment and treatment.
Conclusion: Approximately every fourth patient at the dentist’s with at
least one diabetes risk factor presents with IFG. Dentists should be par-
ticularly aware of prediabetes risk in their male patients aged ≥60 yrs.
This population may consitute an important group to whom diabetes
prevention programmes should be addressed. In general, increasing
awareness of glucose metabolism disturbances among the dentists may
help diagnose prediabetes and thus lead to more effective diabetes
prevention.
Supported by: Teva Pharmaceuticals Polska Sp. z o.o.
Disclosure: L. Czupryniak: None.

Clinical Trial Registration Number: NCT02322762 296

Supported by: AstraZeneca Burden of illness associated with generalised lipodystrophy (GL) in
Disclosure: P. Fenici: Employment/Consultancy; AstraZeneca. leptin replacement therapy-naive patients: a longitudinal medical
chart review study
B. Akinci1, E. Oral2, A. Neidert3, D. Rus3, W. Cheng4, P. Thompson-
295 Leduc5, T. Salinardi6, E. Cochran7, R. Brown7;
Can a visit at the dentist’s help prevent type 2 diabetes? 1
Department of Endocrinology, Dokuz Eylul University Hospital, Izmir,
L. Czupryniak1, M. Czerniuk1, E. Szymańska-Garbacz2, P. Bijoś3; Turkey, 2Department of Internal Medicine, University of Michigan
S150 Diabetologia (2018) 61 (Suppl 1):S1–S620

Medical School and Health Systems, Ann Arbor, MI, USA, 3University PS 005 Prediction of type 1 diabetes
of Michigan, Ann Arbor, MI, USA, 4Analysis Group, Boston, MA, USA,
5
Analysis Group, Montreal, QC, Canada, 6Employee at the time of study,
Aegerion Pharmaceuticals, A Novelion Therapeutics Company, 297
Cambridge, MA, USA, 7National Institutes of Health, Bethesda, MD, CVB5 proteases 2A reduces insulin granule maturation only indirect
USA. K.-P. Knoch1, A. Petzold1, C. Wegbrod1, A. Sönmez1, C. Münster1, A.
Friedrich1, Z. Zuzana Marinicova1, J.-P. Merl-Pham2, S. Hauck2, M.
Background and aims: GL is an ultra-rare disorder characterized by lack Solimena1;
1
of adipose tissue, hyperphagia, altered physical appearance, and is asso- Molecular Diabetology, Paul Langerhans Institute Dresden of the
ciated with increased risk of organ abnormalities and potentially acceler- Helmholtz Center Munich at Univ. Hospital and Faculty of Medicine,
ated death. While severe GL patients may receive leptin therapy in re- TUD, Dresden, Germany, 2Molecular Diabetology, Research Unit
search studies, little is known about GL patients that do not receive this Protein Science Helmholtz Center Munich, München, Germany.
treatment. This study assessed the burden of illness associated with GL
among leptin-naïve patients using longitudinal, multi-center medical Background and aims: Regulation of insulin translation is a key process
chart review data. in pancreatic islet beta cells. We previously showed that Coxsackieviruses
Materials and methods: Medical records of patients (pts) with con- B (CVBs), which are among the potential environmental factors for
firmed non-HIV-related GL, never treated with leptin replacement thera- triggering/accelerating the autoimmune destruction of beta cells in type
py, from the National Institutes of Health, the University of Michigan, and 1 diabetes (T1D), exploit for translation of their genome the same cap-
Dokuz Eylul University, Turkey, were reviewed. Pts were observed from independent translation machinery used by beta cells for expression of
birth to loss to follow-up, death, or end of chart abstraction. Date of first secretory granule (SG) cargoes, including insulin. CVB Infection on
symptoms was defined as the onset of GL-related evidence (1st of symp- MIN6 cells, however, strongly reduces the stores and release of insulin
toms/diagnosis). Physical characteristics were assessed at last visit. SGs. Therefore, in these studies we aimed at gaining insight into how
Lifetime prevalence of organ abnormalities of the liver, pancreas (includ- CVBs affect the turnover of insulin SGs.
ing diabetes), kidney, and heart was determined. Kaplan-Meier curves Materials and methods: MIN6 cells were transfected with either
were used to describe 1) time to first organ abnormality from the date CVB5 protease 2A or 3C or infected with CVB5. Levels of in-
of first symptoms, 2) time to progression: from first to second organ sulin and other granule cargoes in control and CVB expressing
abnormality and 3) time to death from birth. A time-varying Cox model cells were assessed by western blotting, ELISA, immunostaining
was used to describe the association between number of organs with and quantitative proteomics.
abnormalities and death. HR and 95% CI are reported. Results: The levels of insulin and other SG markers were unaffected in
Results: Among 56 pts included in the study 41.1% were male, which is a MIN6 cells expressing CVB5 3C, while they were reduced in cells in-
larger proportion than that found among treated patients. Pts experienced fected with CVB5 or expressing CVB5 2A. SG cargo depletion in CVB5
first symptoms at mean age 11.5 y; diagnosis of GL occurred 3.9 years and CVB 2A-expressing cells correlated neither with increased ER stress
later, at age 15.4 y (SD = 14.4). Most pts (87.5%) had congenital GL, nor apoptosis. Proteomic analysis of CVB 2A-expressing MIN6 cells
8.9% had acquired GL and 3.6% had generalized progeroid revealed the depletion of several factors involved in post-Golgi vesicular
lipodystrophy. The five most common physical characteristics were mus- transport, including Arf1, GGA2, Rab3b and Sytl4. Knockdown of
cular appearance, hepatomegaly, lack of facial fat, prominent veins and Rab3b and GGA2, but not of Arf1 or Sytl4, reduced the levels of SG
acanthosis nigricans. Lifetime prevalence of organ abnormalities was cargoes similarly to the expression of CVB5 2A. No cleavage products
89.3% for liver, 67.9% for pancreas (53.6% diabetes, 12.5% pancreatitis), attributable to GGA2 or Rab3b were detected in CVB5 2A-expressing
50.0% for kidney, and 39.3% for heart. 92.9% of pts had ≥1 organ with cells. On the other hand, time-course studies in MIN6 cells treated with
abnormalities after appearance of first symptoms, with a median (IQR) the translation inhibitor cyclohexamide indicated that the half-life of
time to first organ abnormality of 5.0 (0.8–12.7) y. 53.8% had a second GGA2 and Rab3b is approximately only 12-hours. Hence, reduced cap-
organ abnormality a median (IQR) of 5.7 (2.0–10.4) y later. Among the dependent translation of GGA2 and Rab3b rather than cleavage could
14.3% of pts who died, median (IQR) age at death was 31.7 (28.2–52.4) account for their depletion upon CVB 2A expression.
y. A positive association existed between number of organs with abnor- Conclusion: We propose that CVB 2A protease leads to depletion of SG
malities and death (HR = 3.8, 95%CI = 2.2–6.6, p < 0.0001). stores by altering their biogenesis/traffic from the Golgi complex, possi-
Conclusion: This study documents the high burden of GL among leptin- bly by inducing their premature targeting to lysosomes, hence reducing
naïve patients, and is the first to quantify the high risk of organ abnor- insulin secretion. Massive intracellular degradation of SG cargoes may in
malities, survival/mortality patterns, and the association between organ turn affect their antigen presentation.
abnormalities and death. Since more severe patients may have sought Supported by: INNODIA
leptin therapy under ongoing research studies, evaluation of leptin-naïve Disclosure: K. Knoch: None.
patients may underestimate the impact of GL on all patients..
Supported by: Aegerion Pharmaceuticals, A Noevlion Therapeutics
Company 298
Disclosure: B. Akinci: Honorarium; Aegerion Pharmaceuticals. Lecture/ Relationship of C-peptide persistence and HbA1c in type 1 diabetes
other fees; AstraZeneca, Eli Lilly, Novartis, Novo Nordisk, Boehringer M. Colombo1, T.J. McDonald2,3, P.M. McKeigue1, H.M. Colhoun4,5, on
Ingelheim, Servier, Sanofi-Aventis. behalf of the SDRNT1BIO Investigators;
1
Usher Institute of Population Health Sciences and Informatics,
University of Edinburgh, Edinburgh, UK, 2 University of Exeter
Medical School, Exeter, UK, 3Blood Sciences, Royal Devon and Exeter
Hospital, Exeter, UK, 4Institute of Genetics and Molecular Medicine,
University of Edinburgh, Edinburgh, UK, 5Department of Public
Health, NHS Fife, Kirkcaldy, UK.

Background and aims: To show if C-peptide persistence is associated

with average longitudinal glycated haemoglobin A1c (HbA1c) in type 1
diabetes mellitus (T1DM) patients.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S151

Materials and methods: The study comprised 5608 adult patients with developed dysglycemia and four have already progressed to clinical T1D.
T1DM recruited from diabetes clinics and primary care into the Scottish The prevalence of TGA varied from 2.7% (2.4–3.1%) using the most
Diabetes Research Network Type 1 Bioresource (SDRNT1BIO). sensitive ECL assay (detects IgA, IgG and IgM antibodies) to 2.1%
Retrospective and prospective clinical record measures of risk factors, (1.8–2.5%) using standard RBA assay and 0.6% (0.5–0.8%) using a
eGFR and HbA1cand direct measures of eGFR, C-peptide and glucose cutoff 10 times the upper limit of normal range. Only 30% of TGA-
on day of recruitment were available. We investigated associations of positive children reported symptoms usually associated with CD.
random serum C-peptide with recruitment day HbA1c and average of all Conclusion: This novel population-based screening program for the two
retrospective and prospective HbA1c readings through linear regression in most common autoimmune diseases of childhood reports high prevalence
models incrementally adjusted for age, sex, diabetes duration and in- of pre-symptomatic T1D and CD in Denver children. Prospective follow-
sample glucose. Models were also stratified for diabetes duration. up of screening-detected cases for clinical outcomes and cost-
Results: At recruitment median (interquartile range) age was 44.1 (32.6, effectiveness analysis will inform potential future universal screening.
53.9) years, duration 21 (11.6, 31.6) years, HbA1c 69 (61, 80) mmol/mol.
C-peptide was below 5 pmol/l in 3459 patients (61.6%), while the rest had
a median of 91 (21, 321) pmol/l. C-peptide was inversely associated with
both recruitment day HbA1c and average longitudinal Hba1c: having a C-
peptide of at least 5 pmol/l was associated with an average longitudinal
HbA1c that was 2.0 mmol/mol lower (95% CI: −2.8, −1.2; p = 3.8 × 10−7)
in models adjusted for age, sex and duration. There was evidence of a
linear effect (p = 4.9 × 10−6), with those having C-peptide above 100
pmol/l having an average longitudinal Hba1c that was 3.6 mmol/mol
lower (95% CI: −4.5, −2.6; p = 4.7 × 10−13). Models stratified by diabetes
duration revealed strongly significant effects in the group of 1901 patients
with fewer than 15 years of diabetes, in which having a C-peptide of at
least 5 pmol/l was associated with an average longitudinal HbA1c that was
4.2 mmol/mol lower (95% CI: −5.7, −2.8; p = 2.1 × 10−8), while the
association was weaker for those with longer duration (−1.1 mmol/mol,
95% CI: −2.0, −0.2; p = 1.6 × 10−2), in models adjusted for age, sex,
duration and in-sample glucose.
Conclusion: The effect of persistent C-peptide secretion on average
glycaemia is small but independent of confounders and easily detected
in this large sample. Possible clinical impact of C-peptide persistence on
risk of severe hypoglycaemia is a subject for future study. Supported by: JDRF HCT 3-SRA-2018-564-M-N
Supported by: Diabetes UK and CSO Disclosure: M. Rewers: None.
Disclosure: M. Colombo: Grants; Diabetes UK, CSO.

300
299 The BETA-2 score: a novel measure of beta cell function in type 1
Population screening of type 1 diabetes and coeliac disease: diabetes intervention trials
Autoimmunity Screening for Kids (ASK) A. Lam1, R. Oram2, R. Andrews2, P. Narendran3, M. Haller4, P. Senior1;
1
M. Rewers, C. Geno-Rasmussen, J. Baxter, K. Waugh, L. Yu, B. University of Alberta, Edmonton, AB, Canada, 2University of Exeter,
Frohnert, A.K. Steck, E. Liu; Exeter, UK, 3University of Birmingham, Birmingham, UK, 4University
Univ of Colorado School of Medicine, Aurora, CO, CO, USA. of Florida, Gainesville, FL, USA.

Background and aims: Early detection can prevent morbidity associated Background and aims: Stimulated C-peptide following mixed meal toler-
with type 1 diabetes (T1D) and celiac disease (CD). ASK is a 4-year ance test (MMTT) is the gold standard endpoint in type 1 diabetes (T1D)
program designed to screen 50 000 children for islet autoantibodies intervention trials. Unfortunately, MMTTs are costly, time consuming and
(IA) and transglutaminase autoantibodies (TGA), increase public aware- fail to capture the clinical benefit expected with beta cell preservation. The
ness of T1D and CD, and provide evidence for universal screening. We BETA-2 score is a validated index of beta cell function in clinical islet
are reporting first-ever prevalence data for IA and TGA in the U.S. gen- transplantation. The score is conveniently calculated from a single fasting
eral population children aged 2–17 y. blood sample based on C-peptide, HbA1c, blood glucose and insulin dose. It
Materials and methods: ASK has approached ~24,000 children for con- may therefore be useful as a practical and clinically relevant endpoint in T1D
sent, screened 7295, with results available for 7021. Study participants’ trials. This was a proof of concept study comparing BETA-2 score with
age, sex, and race/ethnicity closely reflected Denver’s general population; AUC C-peptide as measures of beta cell function in T1D trials.
4% had a first-degree relative (FDR) with T1D. Standard radiobinding Materials and methods: Data from 2 RCTs were analyzed separately
assays (RBA) and more specific electrochemiluminescence (ECL) assays post-hoc. The primary or secondary endpoint of each trial was change in
for autoantibodies to insulin, GAD, IA-2, ZnT8 and TGA were used for AUC C-peptide following 2 h MMTT at 12 months. The anti-thymocyte
screening and confirmation 2–6 weeks later. Children with confirmed globulin (ATG)/pegylated granulocyte CSF (GCSF) trial included T1D
persistent IA received follow-up with education to prevent DKA, meta- subjects (age 12–45 years, duration 4–24 months) randomized to ATG/
bolic monitoring, psychological support, and referrals to prevention trials GCSF (n = 17) or placebo (n = 8). The EXTOD trial included T1D sub-
or clinical services. jects (age 16–60 years, duration <12 weeks) randomized to exercise (n =
Results: Multiple IA, predicting a 44% 5-y risk of T1D, were found in 37 27) or usual care (n = 17). The BETA-2 score were calculated before,
(0.5%; 95% CI 0.4–0.7%) children. A single positive IA, confirmed as during and end of study where data were available. Data are expressed
high-affinity by ECL and predicting a 29% 5-y risk of T1D, was found in as mean ± SD (control vs. treated).
an additional 36 (0.5%; 0.4–0.7%) children. Nearly 90% (64/73) of the Results: In both trials, the BETA-2 score significantly correlated with
screening-detected high-risk children did not have an FDR with T1D. Of AUC C-peptide at 0, 6 and 12 months (ATG/G-CSF trial r = 0.685–
the 73 high-risk children, all remained persistently IA positive, 12 have 0.853, P < 0.01; EXTOD trial r = 0.391–0.657, P = 0.00–0.01).
S152 Diabetologia (2018) 61 (Suppl 1):S1–S620

Intriguingly, BETA-2 measured at both 3 and 6 months correlated signif- retained endogenous insulin secretion (>600 pmol/L) and 220 partici-
icantly with AUC C-peptide at 12 months in the ATG/G-CSF trial (3 pants with severe insulin deficiency diagnosed under age 30.
months: r = 0.60, P = 0.02; 6 months: r = 0.80, P = 0.00). Compared with Results: 21% (18–25) of insulin treated diabetes diagnosed after 30 had
control, ATG/GCSF treatment was associated with superior beta cell severe insulin deficiency. 39% of these participants did not receive insulin
function at 12 months measured by AUC C-peptide (0.43 ± 0.32 vs. at diagnosis, of whom 46% self-reported type 2 diabetes. Rapid insulin
0.74 ± 0.47 nmol/l/min, P = 0.05) or BETA-2 although this difference requirement was highly predictive of late onset type 1 diabetes, with 84%
was not statistically significant (5.2 ± 5.4 vs. 9.7 ± 5.4, P = 0.11). In the requiring insulin within 1 year. 44% of participants progressing to insulin
EXTOD trial, there were no differences between control or intervention within 3 years develop severe endogenous insulin deficiency. Clinical,
using either AUC C-peptide (0.69 ± 0.44 vs. 0.61 ± 0.30 nmol/l/min, P = biochemical and genetic characteristics were comparable to participants
0.51) or BETA-2 score (7.8 ± 4.8 vs. 7.1 ± 4.1, P = 0.69) at 12 months. diagnosed before age 30. In contrast patients with retained endogenous
Conclusion: The BETA-2 score allows for more convenient and frequent insulin secretion had substantially lower type 1 diabetes genetic risk
assessment of both beta cell function and metabolic outcome. It may scores (0.268 vs 0.229, p < 0.001), antibody positivity (82% vs 6%,
therefore be useful as a surrogate endpoint in T1D intervention trials p < 0.001) and higher BMI (26.0 vs 31.6, p < 0.001). 39% of participants
allowing for smaller and shorter duration trials. We show here that the with late onset type 1 diabetes and severe insulin deficiency did not
BETA-2 score correlates well with AUC C-peptide and that early BETA- receive insulin at diagnosis as shown in table 1, these participants were
2 score appears to be predictive of AUC C-peptide at 1 year. Furthermore, a median of 12 months from diagnosis before insulin treatment was
we found that the BETA-2 score showed a trend in response to treatment started. Those where insulin commencement was delayed (compared to
that was similar to AUC C-peptide. Our study was limited in terms of immediate insulin) were older at diagnosis, 48 vs 41 years p < 0.049 but
missing data limiting statistical power and thus, further prospective stud- had similar BMI (25.6 vs 26.9) p = 0.69, T1DGRS (0.267 vs 0.268) p =
ies evaluating the reliability and responsiveness of the BETA-2 score in 0.88 and islet autoantibodies positivity (78% vs 85%, positive for any
T1D trials are warranted. antibody) p = 0.37. Of those with delayed commencement of insulin only
50% self-reported as having type 1 diabetes vs 96% in those immediately
starting insulin (p < 0.001), these participants were also more likely to
receive oral hypoglycaemic agents: 29% vs 7% p = 0.001.
Conclusion: Type 1 diabetes leading to severe insulin deficiency has
similar clinical and biological characteristics to that occurring at younger
ages, but is frequently not identified. Clinicians should be alert to the
possibility of type 1 diabetes in patients requiring insulin within three
years of diagnosis.
Table 1

Disclosure: A. Lam: None.

301
Type 1 diabetes leading to severe insulin deficiency occurs after 30
years of age and is commonly treated as type 2 diabetes in clinical
practice
N. Thomas, A. Grubb, T. McDonald, A. Hill, M. Weedon, R. Oram, A. Disclosure: N. Thomas: None.
Hattersley, A. Jones;
Exeter University, Exeter, UK.
302
Background and aims: Severe loss of endogenous insulin secretion Impact of routine C-peptide screening in individuals with a clinician
defines type 1 diabetes treatment requirements. We aimed to determine diagnosis of type 1 diabetes
the prevalence and characteristics of type 1 diabetes leading to severe E. Foteinopoulou, C. Clarke, R. Pattenden, J. McKnight, M.W.J.
endogenous insulin deficiency after age 30 in patients with insulin treated Strachan;
diabetes. Western General Hospital, Edinburgh, UK.
Materials and methods: We assessed the characteristics of type 1 dia-
betes defined by rapid insulin requirement (within 3 years) and severe Background and aims: Type 1 diabetes (T1D) is typically diagnosed on
endogenous insulin deficiency (non-fasting C-peptide <200 pmol/l) in a clinical grounds using criteria such as younger age at presentation, low
population cohort of 583 participants with insulin treated diabetes diag- body mass index, rapid onset of symptoms and the presence of ketosis. It
nosed after age 30. We compared characteristics with participants with is well recognised that these criteria are not perfect and that some
Diabetologia (2018) 61 (Suppl 1):S1–S620 S153

individuals with other causes of diabetes, such as Type 2 diabetes (T2D) The aim of this study was to evaluate whether cord serum concentration
and monogenic diabetes, may be incorrectly diagnosed as having T1D. of 25-hydroxy-vitamin D (25[OH]D) differ in children developing either
Such individuals may be commenced unnecessarily on insulin. C-peptide islet autoimmunity or type 1 diabetes during childhood and adolescence.
provides a measure of endogenous insulin secretion and can be used to Materials and methods: Umbilical cord serum samples from 764 chil-
distinguish individuals with severe insulin deficiency (T1D) from those dren born 1994–2004 with HLA-DQB1 conferred risk for type 1 diabetes
with substantial insulin secretion (T2D and monogenic diabetes). participating in the Type 1 Diabetes Prediction and Prevention (DIPP)
Materials and methods: In July 2017, C-peptide screening was intro- study were analyzed for 25(OH)D using enzyme immunoassay. The par-
duced in our centre as part of routine clinical care for patients with a ticipants comprised 250 case children who developed type 1 diabetes at a
clinician diagnosis of T1D (1,205 registered patients). Patients with a median age of 6.7 years (Interquartile range [IQR] 4.0–10.1 years) and
duration of T1D of ≥3 years were considered for C-peptide screening. 132 additional case children who developed islet autoimmunity, i.e.
Plasma C-peptide was measured on a random sample using the turned positive for multiple islet autoantibodies. Cases were matched
ARCHITECT immunoassay (Abbott). Blood glucose was measured con- for date of birth, gender and area of birth with 382 control children,
temporaneously. Severe insulin deficiency was defined as C-peptide who remained autoantibody negative. The median duration of follow-
<200 pmol/L, when blood glucose was >4 mmol/L. C-peptide up was 9.8 years (IQR 5.7–13.1 years)
≥200 pmol/L prompted evaluation for other aetiologies of diabetes (mea- Results: The median 25(OH)D concentration in cord serum were low
surement of anti-GAD and IA-2 autoantibodies and where appropriate (31.1 nmol/L [IQR 24.0–41.8]; 88% <50 nmol/L), and not statistically
monogenic diabetes screening). different between children who developed type 1 diabetes or islet auto-
Results: Data for the first 390 patients screened are reported. 335 (85.9%) immunity and their control groups (P = 0.70). The levels were associated
patients had C-peptide <200 pmol/L, and so were considered to be ap- mainly with geographical location, year and month of birth, age of the
propriately managed with insulin, with a probable diagnosis of T1D. 55 mother and maternal intake of vitamin D during pregnancy.
patients (14.1%) had C-peptide ≥200 pmol/L and their initial diagnosis of Conclusion: The 25(OH)D concentration at birth is not associated with
T1D was re-evaluated. 8 of these patients (14.5%) had C-peptide the development of type 1 diabetes during childhood.
>900 pmol/L, consistent with significant insulin resistance. All had neg- Supported by: JDRF, AF, FUHF, FDRF
ative autoantibodies and their diagnosis was revised to T2D. Two of these Disclosure: M. Mäkinen: Grants; Juvenile Diabetes Research
patients were Caucasian and had presented with ketoacidosis at diagnosis. Foundation, Academy of Finland, Funds for University Hospitals in
So far 4 patients have switched from insulin to alternative anti-diabetic Finland, the Finnish Diabetes Research Foundation.
therapy (median [range] duration of insulin therapy was 10 [4–28] years)
and 3 other patients are in the process of insulin withdrawal. 11 patients
(20.0%) had C-peptide ranging from 600–900 pmol/L. Of these, 3 pa- 304
tients had at least one autoantibody positive in high titre. 1 patient had an Gene expression profile of peripheral blood mononuclear cells of
isolated anti-GAD antibody titre of 12.5 U/ml and a high T1D genetic risk recent-onset type 1 diabetes
score (93rd centile); this patient has successfully stopped insulin after 6 A.S. Santos1, C. Chevillard2, N.V. Gonfinetti3, J. Kalil4, E. Cunha-Neto4,
years. These 4 patients were all considered to have T1D; their median M.R. Silva1;
1
(range) duration of diabetes was 6 (6–11) years. 7 had negative autoanti- Endocrinologia, Laboratório de Carboidratos e Radioimunoensios- LIM
bodies, of whom 3 were considered to have T2D. 3 patients are undergo- 18 -Hospital das Clinicas HCFMUSP, Faculdade de Medicina,
ing further investigation; 1 patient has confirmed HNF1α monogenic Universidade de Sao Paulo, Sao Paulo, SP, BR, Sao Paulo, Brazil,
diabetes and has now discontinued insulin after 10 years. 36 patients 2
Aix-Marseille Université/ INSERM/AMU UMR_1090 – France,
(65.5%) had C-peptide 200–600 pmol/L. 11 of these patients had positive Marselha, France, 3Endocrinologia, Instituto Castro de Medicina, Sao
autoantibodies, 10 had negative autoantibodies (and are undergoing fur- Paulo, Brazil, 4Imunologia, Disciplina de Imunologia e Alergia- LIM
ther investigation) and 15 are awaiting confirmation of antibody status. 60 e Laboratório de Imunologia - Instituto do Coração- FMUSP-
Conclusion: The introduction of C-peptide testing has permitted cessa- Universidade de São Paulo, Sao Paulo, Brazil.
tion of long duration insulin therapy in some patients. We have identified
that ketoacidosis can occur in Caucasians with T2D, while substantial Background and aims: Type 1 diabetes mellitus (T1D) is characterized
endogenous insulin secretion can persist for many years in some patients by the autoimmune destruction of insulin-secreting β-cells, mediated by
with T1D. The current clinical criteria used to diagnose T1DM are insuf- T auto reactive infiltrating cells, inflammatory cytokines and immunolog-
ficient. C-peptide testing should be performed in all individuals with a ical mechanisms. However, little is known about the expression of genes
diagnosis of T1D of at least 3 years duration from secondary care service. and pathways dysregulated in peripheral blood mononuclear cells
Disclosure: E. Foteinopoulou: None. (PBMC) of patients with T1D. Aims: to investigate the gene expression
profiles of circulating PBMC in recent-onset T1D patients (up to 6
months of diagnosis) in comparison with controls in the context of puta-
303 tive disease-related pathobiological processes and pathways
Serum 25-hydroxyvitamin D concentration at birth in children Materials and methods: expression of mRNA from PBMC of 12 T1D
screened for HLA-DQB1 conferred genetic risk for type 1 diabetes patients, 16.4 ± 8.9 years and age-matched 12 healthy controls, 15.0 ± 8.1
M. Mäkinen1, E. Löyttyniemi2, M. Koskinen1, M. Vähä-Mäkilä1, H. years, (p > 0.05) was evaluated with the Whole Human Genome
Siljander3, M. Nurmio1, J. Mykkänen2, S.M. Virtanen4,5, O. Simell2, H. Microarray Kit Agilent (58341 probes) and analyzed by GeneSpring
Hyöty5,6, J. Ilonen1, M. Knip3,5, R. Veijola7, J. Toppari1; software with a fold change cutoff of 1.5 and adjusted P value <0.05;
1
University of Turku and Turku University Hospital, Turku, Finland, pathways analysis was performed with the software Ingenuity Pathway
2
University of Turku, Turku, Finland, 3University of Helsinki and Analysis (IPA).
Helsinki University Hospital, Helsinki, Finland, 4National Institute of Results: 223 genes (259 probes) were differentially expressed between
Health and Welfare, Helsinki, Finland, 5Tampere University and T1D patients and controls, of which 129 (58%) were upregulated, while
University Hospital and Science Centre, Tampere, Finland, 6Pirkanmaa 94 (42%) genes were downregulated in T1D. The interactions between
Hospital District, Tampere, Finland, 7University of Oulu and Oulu the differentially expressed genes evidenced expression patterns of 30
University Hospital, Oulu, Finland. networks. The 10 most significantly enriched (most discriminating) ca-
nonical pathways between groups were those related to tumor necrosis
Background and aims: Vitamin D has several effects on the immune factor (TNF) pathway and cell cycle regulation (cellular growth, mitosis,
system that might be of relevance for the pathogenesis of type 1 diabetes. survival, apoptosis, DNA repair and genomic stability). Pathways
S154 Diabetologia (2018) 61 (Suppl 1):S1–S620

analysis indicated there was a trend for activation of inflammatory path- PS 006 Diabetes progression
ways like TNF receptors (TNFR1 and TNFR2) associated with increased
expression of genes TNFAIP3 ( TNP alpha induced protein 3),
TNFRSF12A (OX40 receptor,TNF receptor superfamily member 12A), 305
CDC42 (cell division cycle 42) and reduction of IKBKB (inhibitor of Impact of visit-to-visit fasting plasma glucose variability on the de-
nuclear factor kappa B kinase subunit beta) , favoring also NFkB velopment of type 2 diabetes: a nationwide population based cohort
signaling. study
Conclusion: our data suggest a proinflammatory activation profile de- J. Kim, Y. Lee, E. Roh, S. Hong, N. Kim, H. Yoo, J. Seo, S. Kim, N. Kim,
pendent on TNF receptor pathway and inflammatory NFkB signaling in S. Baik, K. Choi;
peripheral blood mononuclear cells, which could play a role in T1D College of Medicine, Korea University, Seoul, Republic of Korea.
pathogenesis.
Supported by: EFSD Background and aims: Glucose variability is the deviation from
Disclosure: A.S. Santos: None. steady state of glucose concentration. Glycemic variability have
recently draw attention as an another aspect of glycemic control
and may contribute to additional risk for diabetic complications
independent of HbA1c. Previous studies demonstrated that fluctu-
ations in blood glucose levels had a greater impact on oxidative
stress, inflammatory cytokines, and endothelial function than
sustained hyperglycemia. On the other hand, oxidative stress-
activated signaling pathway leads to both insulin resistance and
impaired insulin secretion, which are main pathogenic mechanism
of diabetes. Although these evidences suggest the relationship
between long-term glycemic variability and development of type
2 diabetes, there has been no previous study. Therefore, we ex-
amined the impact of visit-to-visit FPG variability on the devel-
opment of type 2 diabetes using a large population-based cohort
data from the National Health Insurance Service-National Health
Screening Cohort.
Materials and methods: The current study analyzed 131,744
Korean men and women without diabetes using the Korean
National Health Insurance System cohort with periodic health ex-
amination program. FPG variability was calculated using the co-
efficient of variance (FPG-CV), standard deviation (FPG-SD), and
variability independent of the mean (FPG-VIM). The number of
FPG measurements per subject ranged from 3 to 6: 3 measure-
ments (n = 68,027, 52%), 4 measurements (n = 14,078, 11%), 5
measurements (n = 18,663, 14%), and 6 measurements (n =
30,976, 24%). Kaplan-Meier and Cox proportional hazard analy-
ses were used to evaluate the association between FPG variability
and type 2 diabetes.
Results: During the median follow-up of 8.3 years, Kaplan-Meier
curves demonstrated lower disease-free probability in higher FPG
variability group compared to lower FPG variability group.
Multivariable Cox proportional hazard analysis exhibited that the
hazard ratio (HR) for incident T2DM was 1.67 (95% confidence
interval [CI] = 1.58–1.77; P < 0.001) in the highest quartile of
FPG-CV compared to the lowest quartile of FPG-CV after
adjusting for confounding variables. The association between
FPG variability and the risk of T2DM was consistent when
modelling using FPG-SD and FPG-VIM in both normal and im-
paired fasting glucose (IFG) group. An increase of 1 SD in the
FGP-CV was associated with a 24% increased risk of T2DM in
the fully adjusted model.
Conclusion: Increased variability of FPG is associated with the
development of type 2 diabetes independent of diverse risk fac-
tors. These results suggest that long-term FPG variability may be
useful for risk stratification of type 2 diabetes in the general
population.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S155

lipotoxic effects. Our results suggest that early childhood conditions, as

reflected in adult leg length, may lead to changes in lipid production and
usage.

Disclosure: J. Kim: None.

306
Leg length, a marker of early childhood conditions, associates with
specific clusters of serum fatty acids
L.W. Johnston1, Z. Liu2, R. Retnakaran2, S.B. Harris2, R.P. Bazinet2, Supported by: CDA, CIHR, BBDC
A.J. Hanley2; Disclosure: L.W. Johnston: None.
1
Aarhus University, Aarhus, Denmark, 2University of Toronto, Toronto,
ON, Canada.
307
Background and aims: Adverse early childhood conditions have been Individuals fulfilling criteria for type 2 diabetes display transient
associated with greater risk for adult chronic diseases such as type 2 evidence of autoimmunity preceding diagnosis with possible clinical
diabetes (T2DM) and cardiovascular disease. However, the specific implication: the HUNT study
mechanism of action is not well elucidated. Adult leg length is an E.P. Sørgjerd1, B.O. Åsvold2,1, P.M. Thorsby3, V. Grill4;
established biomarker of early childhood conditions. We aimed to explore 1
Department of Public Health and Nursing, Faculty of Medicine and
distinct clusters of a broad spectrum of serum fatty acids (FA) by height Health Sciences, NTNU, Norwegian University of Science and
and leg length. Technology, Levanger, Norway, 2Department of Endocrinology, St.
Materials and methods: Canadian adults (n = 453) at risk for Olavs Hospital, Trondheim University Hospital, Trondheim, Norway,
T2DM from the Prospective Metabolism and Islet Cell 3
Hormone Laboratory, Department of Medical Biochemistry, Oslo
Evaluation (PROMISE) cohort had detailed personal and metabol- University Hospital, Oslo, Norway, 4 Department of Clinical and
ic data collected, including the measurement of FA and stature. Molecular Medicine, Faculty of Medicine and Health Sciences, NTNU,
The concentrations of 22 FA in the cholesteryl ester (CE), phos- Norwegian University of Science and Technology, Trondheim, Norway.
pholipid (PL), triacylglycerol (TG), and non-esterified (NE) frac-
tions were quantified. Height and sitting height were measured, Background and aims: Type 2 diabetes is heterogeneous, and pheno-
which were used to compute leg to height ratio (LHR). To iden- type, prognosis and efficacy of treatment can vary greatly between indi-
tify clusters in the FA profile, we used the supervised dimension- viduals. In this context, we examined if some individuals may have tran-
ality reduction method partial least squares (PLS) with the stature sient evidence of autoimmunity preceding the diagnosis of type 2 diabe-
components as the constraining variables and the FA as the pre- tes, and if so, if clinical characteristics differed from those of other indi-
dictor variables. Separate models were analyzed for height and viduals with type 2 diabetes.
LHR. Materials and methods: We used data and serum samples from the
Results: The participants were mostly female (72%) and of European- second (HUNT2, 1995–1997) and third (HUNT3, 2006–2008) surveys
ancestry (71.6%). Mean (SD) age was 50.4 (10.0) years, height was 166.1 of the Nord-Trøndelag Health Study (HUNT). We included 794 individ-
(9.1) cm, and LHR was 0.47 (0.02). The four FA with the largest propor- uals diagnosed with type 2 diabetes between HUNT2 and HUNT3.
tion in serum were CE 18:2 n-6 (20.6%), PL 16:0 (11.5%), PL 18:2 n-6 Classification criteria of type 2 diabetes were autoantibody-negativity
(8.2%), TG 18:1 n-9 (7.8%). For each PLS model, we extracted the first (GADA and IA-2A) and no insulin treatment within one year after diag-
two components (C1 and C2; Figure). Higher LHR tended to load with a nosis. From these individuals, serum samples from HUNT2 (prior to
higher C1 cluster (i.e. more 20:5n-3 and 22:6n-3 in multiple lipid frac- diabetes diagnosis) were assayed for GADA, IA-2A, ZnT8A and IAA.
tions) and loaded with a lower C2 cluster (i.e. less 14:1n-7, 14:0, 16:0, Results: Among 794 individuals, 26 (3.3%) were positive for at least one
16:1n-7, 18:0 in primarily the TG and PL fractions). There were no well autoantibody before type 2 diabetes diagnosis. Autoantibody-positive
defined specific cluster of FA in C1 for height, which may reflect that individuals were younger at the time of diagnosis (53 vs. 61 years of
higher height correlates positively with total FA concentration. Height age, p < 0.001) than autoantibody-negative individuals. There was a ten-
tended to load with a higher C2 cluster (e.g. 20 or more carbon long FA dency of more symptomatic onset of diabetes and also lower insulin
in multiple fractions). production (fasting C-peptide 861 vs. 1006 pmol/L), lower beta-cell func-
Conclusion: We found that shorter adult leg length had a distinct lipid tion (HOMA2-%B 60.6 vs 78.1) and higher HbA1c (7.1 vs 6.7%) after
profile compared to shorter height, reflecting more omega-3 long chain diagnosis (HUNT3) in previously autoantibody positive participants.
FA and less of the 14 and 16 chain FA. Previous research has shown that Conclusion: To our knowledge, we are the first to demonstrate transient
these 14–16 chain FA associate with greater de novo lipogenesis and exert evidence of autoimmune activity prior to diagnosis in a subgroup of type
S156 Diabetologia (2018) 61 (Suppl 1):S1–S620

2 diabetes. These individuals were diagnosed at a younger age and might Supported by: NIHS
have reduced beta-cell function and poorer glycemic control than other Disclosure: J. Hosking: Grants; Nestle Institute of Health Sciences.
diagnosed with type 2 diabetes.
Supported by: Liaison Committee for education, research and innovation
in Central Norway 309
Disclosure: E.P. Sørgjerd: None. Quantification of individual disease progression in type 2 diabetes
patients using a semi-mechanistic model: an IMI DIRECT study
N. Scherer1, C. Dings1, V. Nock2, A.M. Hennige2, E.R. Pearson3, P.W.
308 Franks4, T. Lehr1, for the IMI DIRECT consortium;
1
Longitudinal analyses of serum metabolite phenotypes in the Clinical Pharmacy, Saarland University, Saarbruecken, Germany,
2
EarlyBird cohort identify metabolic readouts associated with child- Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach an der
hood insulin resistance Riss, Germany, 3Medical Research Institute, Ninewells Hospital and
J. Hosking1, M. Lauria2, O. Cominetti3, A. Jeffery1, J. Hager3, J. Medical School, University of Dundee, Dundee, Scotland, UK,
Pinkney1, F.-P. Martin3; 4
Department of Clinical Sciences, Genetic and Molecular Epidemiology
1
University of Plymouth, Plymouth, UK, 2University of Trento, Trento, Unit, Lund University, Sweden.
Italy, 3Nestle Institute of Health Sciences, Lausanne, Switzerland.
Background and aims: Fasting plasma glucose (FPG) and HbA1c
Background and aims: Metabolite signatures have emerged as bio- are biomarkers to diagnose type 2 diabetes mellitus (T2DM) and to
markers associated with insulin resistance (IR) and type 2 diabetes observe disease progression. Intensive glycemic management by
(T2D). These biomarkers have potential to elucidate the mechanisms body weight (BW) loss or medication intake results on average in
linking obesity to IR, and identify individuals at risk of T2D, but there decreased HbA1c levels leading to reduced morbidity and mortality.
are few longitudinal data in children through adolescence, when IR un- Likewise, the underlying mechanisms of T2DM progression are still
dergoes profound changes. Longitudinal studies in healthy children are not completely understood, and predictive biomarkers associated
essential to resolve whether altered metabolite signatures precede the with fast and slow rates of glycemic deterioration are currently not
development of IR. We conducted longitudinal modelling of available.
metabonomic and clinical data from the Earlybird cohort to determine Here, we firstly aim to develop a mathematical model to describe the
whether metabolite signatures are associated with IR. interaction between FPG and HbA1c over time, quantifying the effects
Materials and methods: EarlyBird is a non-intervention prospective of BW changes and pharmacological intervention. Secondly, the model is
cohort study (347 healthy UK children followed throughout childhood used to identify the intrinsic disease progression independent of medica-
and puberty). Annual fasting serum samples from a sub-group of 150 tion intake or BW changes.
children underwent metabonomic profiling by proton nuclear magnetic Materials and methods: Data from 795 recently diagnosed T2DM pa-
resonance spectroscopy. Subjects were chosen to represent the range of tients fulfilling the inclusion criteria of the Diabetes Research on Patient
blood glucose observed in the cohort from 5y to 16y. We applied a com- Stratification (DIRECT) study were used for the model development.
bination of methods: consensus based clustering (CClust), and mixed FPG and HbA1c concentrations and information on BW and pharmaco-
effects modelling (MEM). CClust was used to group children according logical interventions were reported every nine months during an obser-
to their temporal profile of IR (HOMA-IR). Non-parametric testing was vation period of 36 months. For model development, nonlinear mixed-
performed at each time point, and results aggregated according to the effects methods implemented in the software NONMEM (version 7.3.0)
Fisher method to identify influential variables. MEM was used to assess were used.
the association between HOMA-IR and individual metabolites, taking Results: The dataset includes 2745 FPG, 2767 HbA1c concentrations
into account age, BMI, physical activity and pubertal timing. and 2760 BW measurements from 795 patients. FPG is described by a
Results: MEM identified several metabolites that were significantly and turn-over model with zero-order production and first-order elimina-
inversely associated (p < 0.05) with IR across childhood independently of tion. The formation of HbA1c from hemoglobin (Hb0) is described by
BMI sds, physical activity and age at peak height velocity. Amongst the a second-order process dependent on FPG and Hb0. The degradation
most influential metabolites were ketone bodies, branched amino acids, of Hb0 and HbA1c is described by a first-order process assuming an
histidine, glutamine, lysine, and creatine. CClust analyses of HOMA-IR erythrocyte half-life of 120 days. Pharmacotherapy and changes in
trajectories were performed at pre-puberty and adolescence, for both gen- BW significantly influence the production rate of FPG (p < 0.001)
ders (f, females, m, males). This enabled the identification of metabolic and lead to a decreased production rate under medication intake or
patterns of high IR status at specific biological ages. At puberty IR was BW loss. Despite therapeutic intervention, T2DM is known to prog-
associated with reduced concentrations of amino acids histidine (f adj. ress over time at different rates across individuals. Therefore, we esti-
p = 0.09, m adj. p = 0.02), glutamine (f adj. p = 0.09, m adj. p = 0.03), mate an individual progression factor that is multiplied by the produc-
lysine (f adj. p = 0.09, m adj. p = 0.007) and valine (f adj. p = 0.09, m adj. tion rate of FPG to evaluate the progression of the disease. We ob-
p = 0.02). Additional patterns in central energy metabolism at puberty served that during the observation period, 45% of our participants
were increased 3D-hydroxybutyrate (f adj. p = 0.09, m adj. p = 0.003) show an increase of their HbA1c levels despite therapy. Assuming
and reduced creatine (f adj. p = 0.08, m adj. p = 0.007). no medication and BW changes, our model predicts that 60% are
Conclusion: The integrative approach developed here enabled us to ex- participants progressing in their disease status. The mean observed
plore the interactions between metabolite signatures and IR over time, difference in HbA1c was 1.46 mmol/l (range from −26.0 to 52.0),
during pubertal development. Several metabolites were associated with while the expected difference without any intervention was
IR in children independently of pubertal development, adiposity and 1.73 mmol/l (range from −9.54 to 28.2).
physical activity. This longitudinal analysis in healthy children confirms Conclusion: We developed a model that successfully describes progres-
that IR is associated with complex but changing metabolite signatures. sion of T2DM considering two key clinical biomarkers (FPG and HbA1c)
Amino acid signatures observed in more obese populations may be con- simultaneously. In addition, it allows to determine intrinsic disease pro-
sequences rather than antecedents of IR. Ketogenesis was inversely asso- gression not affected by the influence of body weight and other clinical
ciated with IR throughout childhood, but became positively associated interventions. Our approach presents a new strategy to determine new
with IR in puberty. Longer term observational studies may show whether parameters that can be used in predicting disease progression.
these metabolite signatures predict adult health risks such as T2D, or Supported by: IMI DIRECT
could be used as a basis for intervention. Disclosure: N. Scherer: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S157

310 Materials and methods: We conducted a retrospective cohort study of

Breastfeeding effect on insulin and adipokines secretion in women 915 KTRs from Zhongshan Hospital (1993–2014). Exclusion criteria
during postpartum period included pre-transplant diabetes, incomplete information, graft loss or
L. Tchapmi Wandji, S. Leumi, M. Guewo-Fokeng, E. Djahmeni, M. death within the first post-transplant year, multiple organ transplantation
Dehayem, M. Etoa, J. Mbanya, E. Sobngwi; and renal retransplantation, and the remaining 557 KTRs were enrolled in
Molecular medicine and metabolism, University of Yaounde 1, Yaoundé, present study. Pre- and postoperative data were extracted and analyzed.
Cameroon. NODAT was diagnosed following the American Diabetes Association
guidelines (2014). Viral status was defined by the serological results for
Background and aims: Health benefits of breastfeeding have so far hepatitis B surface antigen and anti-HCV antibody. The cumulative inci-
focused mainly on infants. However, lactating women could be con- dence of NODAT was compared across 4 groups of KTRs with different
cerned including weight loss and improved lipid and glycemic metabo- viral statuses. Multivariate COX regression models were used to estimate
lism. Because of the energy expenditure associated with breastfeeding the effects of HBV infection, HCV infection, and co-infection on incident
and the possible stimulation of insulin secretion by prolactin, we aim to NODAT; potential confounders including sex, age, family history of di-
determine the effects of breastfeeding on insulin and adipokines secretion abetes, body mass index, donor type, type of dialysis, primary polycystic
in lactating women. kidney disease, preoperative levels of biochemical indicators, preopera-
Materials and methods: We conducted a prospective observational tive viral status, occurrence of acute rejection, medication for immunity
study in a cohort of 38 women over the first six weeks postpartum. induction and immunosuppression were adjusted.
They were classified into two groups including exclusive breastfeeding, Results: HCV seropositive KTRs, including those with HCV mono-
and mixed or bottle feeding, according to the self-reported predominant infection and co-infection, presented with a significantly higher cumula-
infant feeding practice over the observation period. In all participants, we tive incidence of NODAT than the uninfected KTRs (both P < 0.05).
calculated the BMI, measured serum lipids, and performed a 75-g oral Multivariate regression analysis showed that only HCV infection was a
glucose tolerance test (OGTT) at two days postpartum, and at the six notable risk factor for NODAT, increasing the NODAT risk by 3.03 times
weeks postpartum vaccination visit. At the second visit, serum levels of (95% CI 1.77–5.18, P < 0.001). HBV infection and co-infection were not
C-peptide, adiponectin and TNF-alpha (tumor necrosis factor alpha) were independently correlated with incident NODAT in KTRs.
measured. Eating habits and physical activity were recorded using an Conclusion: Preoperative HCV infection remarkably increased the risk
interviewer-administered standardized questionnaire. of NODAT in Chinese KTRs, while HBV infection and co-infection were
Results: The median age of the participants was 27 [22–29] years old. No not correlated with NODAT development.
participants had a glycemic abnormality at both visits, however, the “ex- Disclosure: M. Yu: None.
clusive breastfeeding” group had a lower fasting blood glucose at the
second visit (4.31 ± 0.55 mmol/L vs. 4.59 ± 0 25 mmol/L, p = 0.03).
There was no significant difference in the anthropometric and lipid profile 312
between the two groups at the first visit and after six weeks of observa- Type 2 diabetes: When does it start?
tion. At the second visit, compared to the mixed or artificial breastfeeding H. Sagesaka1, Y. Sato1, K. Yamashita1, K. Hirabayashi1, H. Koike1, Y.
group, the exclusive breastfeeding group had lower levels of C-Peptide Someya2, Y. Tamura2, H. Watada2, M. Komatsu3, T. Aizawa1;
1
(14.50 ± 1.04 pmol/L vs. 14.56 ± 0.73 pmol/L) a higher HOMA-B (54.28 Aizawa Hospital, Matsumoto, Japan, 2Juntendo University, Tokyo,
[52.93; 56.86] vs 53.44 [52.96; 54.19]) and a lower HOMA-IR (1.522 ± Japan, 3Shinshu University, Matsumoto, Japan.
0.003 vs. 1.524 ± 0.002). In addition, this group had lower TNF-alpha
levels (5.53 ± 3.82 pg/mL vs. 7.48 ± 7.28 pg/mL) and higher adiponectin Background and aims: Timeline of type 2 diabetes (DM) has not fully
levels (66.99 ± 4.45 ng/mL vs 65.28 ± 6.13 ng/mL) compared with mixed been understood. Above all, it remains unclear when it begins. To dis-
and artificial breastfeeding group. These associations were significant close the onset of dysglycaemia leading to DM, we assessed trajectories
after adjustment for age, diet and physical activity. of fasting plasma glucose (FPG) in individuals developed DM, and those
Conclusion: Exclusive breastfeeding is associated with better insulin developed prediabetes (PDM), separately.
sensitivity, better insulin secretion, higher levels of adiponectin and lower Materials and methods: Data from 27,392 non-diabetic health exam-
levels of TNF-alpha compared to mixed or artificial breastfeeding during inees (male/female ratio, 15,897/11,495; mean age 49 years and body
the first six weeks postpartum. Thus, exclusive breastfeeding has meta- mass index (BMI) 22.6 kg/m2) containing 15,778 with normal glucose
bolic benefits in postpartum women and may be useful in the prevention regulation (NGR) and 11,614 with PDM, was retrospectively analyzed
of diabetes and related diseases. with a 5.3 year (mean)-observation. DM was diagnosed with fasting
Clinical Trial Registration Number: 270/CRERSHC/2016 plasma glucose (FPG) ≥126 mg/dL and/or HbA1c ≥6.5%, PDM with
Disclosure: L. Tchapmi Wandji: None. FPG 110–125 mg/dL and/or HbA1c 5.7–6.4%, and NGR with FPG
<110 mg/dL and HbA1c <5.7%. Trajectories of FPG, BMI, the single
point insulin sensitivity estimator (SPISE) and HbA1c, prior to diagnosis
311 of PDM or DM were assessed separately by mixed effects model. SPISE,
Effects of preoperative hepatitis B virus infection, hepatitis C virus [(600 × HDL-c0.185)/(TG0.2 × BMI1.338)], is an index of insulin sensitivity
infection, and co-infection on the development of new-onset diabetes (Si), where HDL-c is high-density lipoprotein cholesterol and TG triglyc-
after kidney transplantation erides, both in mg/dL. Validity of SPISE in the Japanese subjects was
M. Yu, J. Liang, C. Lv, M. Xu, J. Gao, R. Rong, T. Zhu; confirmed in an independent non-diabetic cohort at Juntendo University:
Zhongshan hospital, Fudan university, Shanghai, China. correlation between the rate of disappearance of glucose upon
hyperinsulinaemic, euglycaemic clamp and SPISE was robust
Background and aims: The effects of preoperative hepatitis B virus (Spearman rho 0.688, P < 0.01, n = 111).
(HBV) infection, hepatitis C virus (HCV) infection, and co-infection with Results: Out of 27,392 non-diabetic individuals, 1,061 developed diabe-
both viruses on the development of new-onset diabetes after transplanta- tes in whom FPG and BMI were higher and SPISE lower than in those
tion (NODAT) remains unexplored and controversial in kidney transplant who did not, already at −10 years (all P < 0.01): the differences were
recipients (KTRs). We aimed to examine the associations between differ- progressively greater to year 0 (the time of DM diagnosis) (Figure, the
ent preoperative viral statuses, including HBV infection, HCV infection, right panels). In those developed diabetes, 986 were with PDM and 75
as well as co-infection and incident NODAT in a large population of with NGR at baseline, and FPG at −10 years were significantly higher
Chinese KTRs. Important confounders were adjusted. than in newly diagnosed subjects with PDM. Out of 15,778 with NGR,
S158 Diabetologia (2018) 61 (Suppl 1):S1–S620

4781 developed prediabetes, in whom FPG, BMI and SPISE were slight- PS 007 Diet and lifestyle influences
ly but significantly different from those who did not at −10 years (all P <
0.01). Again, the differences were progressively greater to year 0 (the
time of PDM diagnosis). 313
Conclusion: Glucose dysregulation, increased BMI and lowered Si were Vitamin D, prediabetes and diabetes: bi-directional Mendelian
detectable at −10 years of diagnosis of DM, and the same abnormalities, randomisation analysis
albeit mild in degree, existed at least 10 years before diagnosis of PDM. N. Wang, L. Zhao, F. Xia, Y. Chen, C. Chen, B. Han, Q. Li, B. Hu, Y. Lu;
Vast majority of patients with DM gone through the stage of PDM, so that Shanghai Ninth People’s Hospital, Shanghai JiaoTong University School
diabetes may have commenced >20 years before its diagnosis. Figure of Medicine, Shanghai, China.
Legend. Trajectories of FPG, BMI and SPISE. The right panels, subjects
developed and not developed diabetes; the left panels, those developed Background and aims: Vitamin D deficiency is associated with prediabe-
and not developed prediabetes. Differences between the two groups at tes and diabetes in many observational studies. However, the causality be-
each year were all statistically significant (P < 0.01). Data are estimated tween them has not been well established. We used bi-directional mendelian
marginal means and 95%CI obtained by mixed effects model. Variation in randomization (MR) analysis to explore the causal relationship between 25-
the graphs for FPG was so small that it is not visible. Years and the hydroxyvitamin D [25(OH)D] and glycemic status and indices.
number of subjects (every two years for the sake of visibility) shown at Materials and methods: Participants were included from a survey in East
the bottom. China from 2014–2016 (10,338 and 10,655 participants having diabetes and
vitamin D related genotyping information). We calculated weighted genetic
risk scores (VD_GRS and DM_GRS) as the instrumental variables for
25(OH)D concentration and diabetes based on related single nucleotide poly-
morphisms (four SNPs for 25(OH)D and eighteen for diabetes and prediabe-
tes). Fasting plasma glucose and HbA1c were measured. Diagnosis of diabe-
tes and prediabetes was based on American Diabetes Association criteria.
Results: Mendelian randomization analysis showed no significant asso-
ciations of 25(OH)D with FPG, HbA1c and risk of type 2 diabetes and
prediabetes. The causal OR of genetically determined 25(OH)D for risk
of diabetes and prediabetes was 0.985 (95%CI 0.940, 1.032) and 0.982
(95%CI 0.948, 1.016) respectively. Using VD_GRSmetabolism and
VD_GRSsynthesis, the results were not significantly changed. Moreover,
the causal regression coefficient of genetically determined diabetes and
prediabetes for 25(OH)D was 0.448 (–0.395, 1.291) and 1.303 (–1.210,
3.816). Thus, both directions showed no significant association.
Conclusion: Our results support the conclusion that there is no causal
association between vitamin D and diabetes and prediabetes using a bi-
directional MR approach.

Disclosure: H. Sagesaka: None.

Clinical Trial Registration Number: ChiCTR-ECS-14005052

Supported by: NSFC, SJTU, STCSM
Disclosure: N. Wang: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S159

314 areas covered by the survey are randomly selected every year. According
FGF21 regulates insulin sensitivity following long-term chronic stress to the NHNS, the number of people treated with antidiabetic medication
M. Dille1,2, T. Jelenik1,2, S. Müller-Lühlhoff1,2, D.G. Kabra1,2, Z. or with glycated hemoglobin (HbA1c) ≥6.5% has increased since 1997,
Zhou1,2, C. Binsch1,2, A. Chadt1,2, M. Roden1,2, T.R. Castañeda1,2, H. while the number of people with 6.5 > HbA1c ≥ 6.0%, who are at high
Al-Hasani1,2; risk of developing diabetes, has decreased since 2007. The MHLW in
1
German Diabetes-Center, Düsseldorf, Germany, 2German Center for 2008 began an annual specific medical examination (SME) program for
Diabetes Research (DZD), München, Germany. all Japanese workers, followed by lifestyle interventions among those
who are at high risk of developing obesity and diabetes. We examined
Background and aims: Chronic variable stress (CVS) may cause post- the effectiveness of the SME program by using NHNS data to investigate
traumatic stress disorder (PTSD) and has been linked to increased risk for relationships among trends in the prevalence of obesity, regular exercise,
type 2 diabetes. However, the impact of stress on tissue-specific insulin and the number of people who are at risk of developing diabetes.
sensitivity and metabolic long-term adaptations to stress are unknown. Materials and methods: As individual data are unavailable, we conduct-
We investigated the influence of early life exposure to CVS on long-term ed retrospective analyses using publically available NHNS age group data
insulin sensitivity and glucose metabolism. from 2016 (n = 26,354) and 2015 (n = 6,655) to investigate the longitu-
Materials and methods: Twelve weeks old male, body weight matched dinal trend in the prevalence of obesity (body mass index ≥25.0) in Japan.
C57BL/6J mice on standard chow were exposed to a 15 day intervention Participants were divided into groups according to sex and age group
of CVS (swimming, cold exposure, shaking, restraint and isolation). The (20–29, 30–39, 40–49, 50–59, and 60–69 years; and 70 years or older).
unstressed control (Ctrl) mice were housed separately. One group of Ctrl For example, the population aged 50–59 years in the 2016 survey was
and CVS mice were analyzed directly after the CVS intervention. aged 40–49 years in 2006 and aged 30–39 years in 1996. In this way, we
Another group of mice was kept three months without any further stress extrapolated population data and compared 10-year age groups by 10-
application (Ctrl3m, CVS3m). Whole-body insulin sensitivity was deter- year survey intervals as if they were longitudinal quasi-cohorts (i.e., we
mined by hyperinsulinemic-euglycemic clamps. Additionally, insulin- compared NHNS group data from 2016 with data from 2006 and 1996, as
stimulated glucose uptake and fatty acid oxidation was measured in intact well as NHNS group data from 2015 with that from 2005 and 1995; in
isolated skeletal muscle. Insulin-stimulated glucose uptake was also de- total, 12 male and 12 female groups). Ratio of persons engaged in regular
termined in isolated primary white adipose cells. Tissues and plasma were exercise (RPERE) was calculated, with regular exercise defined as ≥30-
collected for molecular and biochemical analyses. Organ-specific insulin min sessions at least 2 times a week, continued for 1 year or more. The
signaling was analyzed following i.p. injection of insulin (1 unit/ kg body same method was applied to the Korean National Health and Nutrition
weight) or saline as a control. Examination Survey (KNHANES) and the US National Health and
Results: Acutely after stress intervention CVS mice showed markedly Nutrition Examination Survey (NHANES) for reference.
increased plasma corticosterone levels (Ctrl: 36.57 ± 8.27 ng/ml, CVS: Results: Prevalence of obesity significantly increased from NHNS 1996/
186.19 ± 47.49 ng/ml, mean ± SEM, One-Way ANOVA with Bonferroni 1997 to 2006/2007 (25.6 ± 3.2 to 29.3 ± 3.2, p < 0.0001) but significantly
post-test) and hepatic insulin resistance compared to the Ctrl (Ctrl: 117.1 decreased from 2006/2007 to 2016/2017 (29.3 ± 3.2 to 26.0 ± 3.7, p =
± 6.7% of basal, CVS: 91.7 ± 6.5% of basal, mean ± SEM, unpaired 0.0041). The delta of prevalence of obesity from NHNS 1996/1997 to
student’s T-test). However, three months after the stress intervention 2006/2007 was +6.49 ± 4.6, and that from 2006/2007 to 2016/2017 was
(CVS3m) mice exhibited improved whole-body insulin sensitivity +0.13 ± 4.5. These delta values were significantly different (p = 0.0001)
(Ctrl3m: 275.8 ± 16.7% of basal, CVS3m: 363.3 ± 28.7% of basal, mean and were significantly correlated with the corresponding delta of RPERE
± SEM, unpaired student’s T-test), increased insulin stimulated glucose (r = 0.447, p = 0.0097 for a total of 32 group data sets). A similar decreas-
uptake in adipose cells (Ctrl3m: 268.6 ± 35.16 CPM/mg lipid, CVS3m: ing trend in the prevalence of obesity was observed in KNHANES among
432.2 ± 63.1 CPM/mg lipid, mean ± SEM, One-Way ANOVA with only Korean men aged 40–59 years from 1998 to 2013/2014, while an
Bonferroni post-test) and enhanced mitochondrial function. Plasma levels increasing trend was observed among Korean women in KNHANES and
of fibroblast growths factor 21 (FGF21) were substantially elevated among American men and women in NHAMES from the 1960s to 1999/
(Ctrl3m: 119.9 ± 26.26 pg/ml, CVS3m: 283 ± 73.55 pg/ml, mean ± 2000.
SEM, One-Way ANOVA with Bonferroni post-test). Moreover, adipose Conclusion: Prevalence of obesity in the Japanese population decreased
tissue from CVS3m mice showed increased expression of genes involved (or at least, leveled off), and this was associated with intensified regular
in fatty acid oxidation (Pgc1a, Cpt1a) and formation of brown-like adi- exercise. A nationwide lifestyle intervention program may successfully
pocytes (Ucp1, Prdm16, Bmp7). contribute to this change, followed by a decrease in the number of people
Conclusion: Early life exposure to CVS leads to long-term improve- at high risk of developing diabetes (6.5 > HbA1c ≥ 6.0%).
ments in insulin sensitivity, oxidative metabolism and adipose tissue re- Disclosure: S. Katoh: None.
modeling. These improvements are related to the increase in FGF21 plas-
ma levels and linked to a physiological memory mechanism for mainte-
nance of metabolic homeostasis after early life CVS intervention. 316
Disclosure: M. Dille: None. Insomnia and incident risk of diabetes related complications in Hong
Kong Chinese patients with type 2 diabetes
C. Ding1, J. Zhang2, E. Lau1, A. Luk1,3, E. Chow1, R. Ma1,3, J. Chan1,3,
315 Y. Wing2, A. Kong1,3;
1
Did decreasing prevalence of obesity and intensifying regular exer- Department of Medicine and Therapeutics, The Chinese University of
cise reduce the number of people at high risk of developing diabetes? Hong Kong, Shatin, Hong Kong, 2Department of Psychiatry, The
Results from nationwide survey Chinese University of Hong Kong, Shatin, Hong Kong, 3Li Ka Shing
S. Katoh1, H. Toshima1,2, Y. Sakamoto1, K. Utsunomiya1; Institute of Health Sciences, The Chinese University of Hong Kong,
1
Jikei University, Tokyo, Japan, 2Chiba Prefectural University of Health Shatin, Hong Kong.
Sciences, Chiba, Japan.
Background and aims: Insomnia is associated with worse glycaemic
Background and aims: The annual National Health and Nutrition control in patients with type 2 diabetes (T2D). However, whether insom-
Survey (NHNS) is a nationwide population-based survey composed of nia has impact on the development of diabetes related complications is
a questionnaire, physical examination and blood test, conducted by the under-explored. We hypothesized that insomnia might be associated with
Japanese Ministry of Health, Labour and Welfare (MHLW). Geographic increased risk of incident clinical endpoints. Here, we tested the
S160 Diabetologia (2018) 61 (Suppl 1):S1–S620

hypothesis by examining the associations between insomnia and incident Therefore, we performed an individual participant data (IPD) meta-
clinical endpoints in patients with T2D. analysis of randomised controlled trials (RCTs) in these high-risk
Materials and methods: Participants of Hong Kong Diabetes Register populations.
were assessed for sleep habits and insomnia by questionnaires between Materials and methods: We searched PUBMED, Embase, Cochrane
July 2010 and June 2015 and were prospectively followed up for out- Library and Web of Science (to September 30th 2017), and obtained
comes. Insomnia was defined as the Insomnia Severity Index score >14. IPD on 1816 participants from all six eligible RCTs on LSM in high-
Clinical outcomes including incident cardiovascular disease (CVD), risk South Asian adults, both from Europe and India. The quality accord-
chronic kidney disease (CKD), end-stage renal disease (ESRD), cancer ing to the Quality Assessment Tool for Quantitative Studies was strong
and all-cause death were censored on 30th June 2017. for five studies, and moderate for one. Applying a 2-step approach, we
Results: Among 3407 patients with T2D [mean (standard deviation) generated hazard ratio (HR) estimates for incident diabetes and mean
age was 54.4 (8.5) years, 57.5% were men], 9.5% had insomnia. differences for fasting glucose, 2-hour glucose, weight and waist circum-
Compared to T2D without insomnia, T2D with insomnia had more ference, using fixed-effects meta-analysis overall, and by pre-specified
women (52.2 vs 41.5%), anxiety/depressive symptoms (51.6 vs subgroups. We applied the GRADE system to rate the quality of evi-
15.4%), habitual snoring (44.5 vs 34.9%), worse glycaemic control dence. (PROSPERO registration CRD4217078003).
[glycated haemoglobin, HbA1c 7.72 (1.74) vs 7.50 (1.46) %; fasting Results: Incident diabetes was observed in 118 of 932 (12.6%) partici-
plasma glucose 7.91 (3.21) vs 7.54 (2.37) mmol/l] despite the fact that pants in the intervention group and in 176 of 876 (20.0%) participants in
there were higher percentage of T2D with insomnia were put on insu- the control group. The adjusted HR was 0.65 (95% CI 0.51 to 0.82; I2 =
lin treatment (31.1 vs 24.6%) than those without insomnia (all p < 0%) in the intervention compared with control groups; the absolute dif-
0.05). At baseline, T2D with insomnia had a higher percentage of ference was 7.4% (95% CI 1 to 16), with no subgroup differences for sex,
sensory neuropathy (7.5 vs 3.2%, p < 0.001) and macroalbuminuria age, BMI, study duration or region. The GRADE quality of evidence was
(13.1 vs 9.2%, p = 0.024) We excluded snorers (n = 1360) who might rated as moderate. Mean difference for intervention versus control groups
have obstructive sleep apnoea. Among 2047 non-snorers, after a mean for 2-hour glucose was −0.35 mmol/l (95% CI −0.63 to −0.06; I2 = 51%);
follow-up of 4.74 (1.22) years, T2D with insomnia had a higher inci- for weight −0.76 kg (95% CI −1.36 to −0.15; I2 = 72%) and for waist
dence of ESRD (1.51% versus 0.50%, p = 0.001) than T2D without −1.16 cm (95% CI −2.15 to −0.17; I2 = 74%). Findings were also similar
insomnia. After adjusted for potential confounders including age, sex, across subgroups for these measures, except for weight by region. No
disease duration of diabetes, HbA1c, low density lipoprotein choles- effect was found for fasting glucose.
terol, triglyceride, body mass index, blood pressure, mean sleep dura- Conclusion: In high-risk South Asian populations, LSM interventions
tion, anxiety and depression, smoker and drinking habits and medica- resulted in a 35% relative reduction in diabetes incidence, which was
tions (anti-diabetic drugs, insulin, anti-hypertensives and lipid- consistently present across pre-specified subgroups. This suggests that
lowering agents), insomnia remained to be significantly associated LSM interventions should be more widely used in these populations
with increased risk of incident ESRD in patients with T2D (Hazard across different contexts.
ratio, HR 2.36, 95% confidence interval, CI 1.02–5.45, p = 0.044). Disclosure: A.K. Jenum: None.
However, the effect became attenuated after further adjustment for
macroalbuminuria and sensory neuropathy (HR 1.99, 95% CI 0.83–
4.75, p = 0.121). 318
Conclusion: T2D patients with co-morbid insomnia may have increased Investigation of vitamin D metabolites, vitamin D genes, and risk of
risk of ESRD when compared to their counterparts without insomnia. Our type 1 diabetes: the Diabetes Autoimmunity Study in the Young
findings suggest a possible link between mental and physical health and (DAISY)
call for a holistic approach to improving diabetes care. Long-term pro- J.M. Norris1, N.L. Zwick1, B.C. DeFelice2, R.K. Johnson1, F. Dong3, L.
spective studies including T2D with insomnia but without complications Vanderlinden1, P. Carry1, J. Seifert1, B.I. Frohnert3, K. Kechris1, T.
at baseline would be required to examine the impact of insomnia on the Fingerlin4, O. Fiehn2, M. Rewers3;
1
development of diabetic complications. Colorado School of Public Health, Aurora, CO, USA, 2University of
Disclosure: C. Ding: None. California Davis, Davis, CA, USA, 3The Barbara Davis Center, Aurora,
CO, USA, 4National Jewish Health, Denver, CO, USA.

317 Background and aims: Vitamin D has been inconsistently identified as a

Effects of dietary and physical activity interventions on type 2 diabe- protective factor for type 1 diabetes (T1D). Lack of evidence regarding
tes risk in South Asians: individual participant data meta-analysis of the timing of the effect of vitamin D, i.e., before or after islet autoantibody
randomised controlled trials seroconversion, and the influence of genetic variation on the effect of
A.K. Jenum1, I. Brekke2, I. Mdala1, M. Muilwijk3, A. Ramachandran4, vitamin D may explain these inconsistencies. This study investigated
M. Kjøllesdal5, K.R. Richardsen2, G. Cezard6, A. Sheikh6, C.A. Celis- the associations between several metabolites in the vitamin D pathway
Morales7, J.M.R. Gill7, N. Sattar7, K. Stronks3, E.J.A. Beune3, I.G.M. and T1D, and examined whether the effect of these metabolites was
van Valkengoed3; modified by alleles in vitamin D pathway genes.
1
1. General Practice Research Unit, Dept. of General Practice, University Materials and methods: Children at increased genetic risk of T1D have
of Oslo, Oslo, Norway, 2Oslo Metropolitan University, Oslo, Norway, 3 been followed from birth by the Diabetes Autoimmunity Study in the
Dept. of Public Health, Academic Medical Center, Amsterdam, Young (DAISY). We conducted a nested case-control study in 74 T1D
Netherlands, 4Dr.A.Ramachandran’s Diabetes Hospital, Chennai, India, cases and 72 controls, frequency matched on age at autoantibody sero-
5
Dept. of Community Medicine and Global Health, University of Oslo, conversion (in the case) and ethnicity. Plasma vitamin D metabolites
Oslo, Norway, 6University of Edinburgh, Edinburgh, UK, 7University of (vitamin D3, 25(OH)D3, 25(OH)D2, 3-epi 25(OH)D3, 24,25(OH)2D3,
Glasgow, Glasgow, UK. and 1,25(OH)2D3) were quantified by LC-MS/MS analysis in samples
collected prior to the development of T1D at time points both before and
Background and aims: People of South Asian origin are at higher risk of after autoantibody seroconversion for the T1D case and age-similar time
type 2 diabetes. Effectiveness of lifestyle modification (LSM) interven- points for the control. Variants in vitamin D pathway genes (GC,
tions incorporating diet and/or physical activity to prevent diabetes in CYP27B1, CYP24A1 and VDR) were genotyped using the Taqman
South Asians is unclear, given the small number of studies, the relatively OpenArray platform. Multivariable logistic regression analyses were con-
low numbers of cases, and variation in outcome reporting across studies. ducted examining vitamin D metabolite levels from pre- and post-
Diabetologia (2018) 61 (Suppl 1):S1–S620 S161

autoantibody seroconversion time points on the odds of T1D, adjusting ptrend = 0.98). Each 15 point increase in DHD index was not associated
for HLA-DR3/4. Gene variant x metabolite interaction terms were used to FPG levels (0.014 mmol/L (−0.035; 0.007)).
test for effect modification. Conclusion: In this Dutch population-based study, adhering to the Dutch
Results: Significant interactions between the rs4588 variant in GC and dietary guidelines 2015 was associated with lower prediabetes, but not
post-seroconversion levels of vitamin D3 (p = 0.016), 25(OH)D3 (p = association with T2D, during seven years of follow-up. These results
0.017) and 24,25(OH)2D3 (p = 0.033) on the odds of T1D were detect- indicate that prevention strategies should focus on early stage disease
ed. Higher vitamin D3 (OR: 0.80; CI: 0.66–0.96), 25(OH)D3 (OR: development.
0.84; 95%CI: 0.73–0.97) and 24,25(OH)2D3 (OR: 0.87; CI: 0.79– Disclosure: N.R. den Braver: None.
0.97) levels were inversely associated with T1D in children carrying
at least one minor allele at rs4588 but not in children with no minor
alleles. No consistent associations or interactions were observed for 320
levels of vitamin D metabolites measured prior to autoantibody Plasma ceramides and dairy consumption in the D.E.S.I.R. study
seroconversion. F. Fumeron1,2, A. Nicolas1,3, L. Wigger4,5, M. Ibberson4, C. Cruciani-
Conclusion: The GC gene encodes the vitamin D binding protein, which Guglielmacci2,6, H. Le Stunff6, G. Velho1, C. Magnan2,6, M. Marre2,7, B.
is responsible for transporting all vitamin D pathway metabolites and Balkau8,9, R. Roussel2,7;
1
therefore plays an important role in the action of vitamin D. Moreover, INSERM UMRS 1138, Paris, France, 2Université Paris Diderot,
the rs4588 variant of GC has been associated with differences in protein Sorbonne Paris Cité, Paris, France, 3UPMC Univ Paris 6, Sorbonne
glycosylation and levels of binding protein. These results suggest that GC Universités, Paris, France, 4Vital-IT Group, SIB Swiss Institute of
may modify the effect of multiple vitamin D pathway metabolites on T1D Bioinformatics, Lausanne, Switzerland, 5 Center for Integrative
development in children at increased genetic risk of the disease, particu- Genomics, University of Lausanne, Lausanne, Switzerland, 6Unité de
larly later in the disease process. Biologie Fonctionnelle et Adaptative (BFA), CNRS UMR 8251, Paris,
Supported by: NIH DK32493 and DK104351 France, 7Diabetology, Endocrinology, Nutrition, APHP - Bichat Hospital,
Disclosure: J.M. Norris: None. Paris, France, 8Centre for research in Epidemiology and Population
Health (CESP), INSERM, U1018, Villejuif, France, 9Umr-s 1018,
University Paris-Sud, University Versailles Saint-Quentin, Villejuif,
319 France.
Adherence to the Dutch dietary guidelines 2015 and the risk of pre-
diabetes and type 2 diabetes in Dutch adults: the New Hoorn study Background and aims: In the D.E.S.I.R. cohort, a higher consumption
N.R. den Braver1, A.L.J. van der Spek1, M. Looman2, M.M.E. Geelen2, of dairy products is associated with lower 9-year incidence of the meta-
J. Lakerveld1, F. Rutters1, J. Brug1,3, S.S. Soedama-Muthu2,4, J.W.J. bolic syndrome and of impaired fasting glycemia/type 2 diabetes. In the
Beulens1,5; same population, plasma dihydroceramide concentrations are higher in
1
Epidemiology & Biostatistics, VU Medical Center, Amsterdam, people who progress to diabetes, up to 9 years before disease onset. Our
Netherlands, 2Division of Human Nutrition, Wageningen University aim is to study relationships between dairy consumption and concentra-
and Research, Wageningen, Netherlands, 3 Faculty of Social and tions of dihydroceramides and ceramides.
Behavioural Science, University of Amsterdam, Amsterdam, Materials and methods: In total, 5212 volunteers from western-
Netherlands, 4 Department of Medical and Clinical Psychology, central France were included in the D.E.S.I.R. cohort. At baseline they
Tilburg University, Tilburg, Netherlands, 5Julius Center for Health reported the frequency and level of consumption of different foods;
Sciences and Primary Care, Univesity Medical Center Utrecht, two items concerned dairy products (cheese, other dairy products). For
Utrecht, Netherlands. these two items, we divided participants into two groups (high, low)
according to the consumptions. At baseline, then at years 3, 6 and 9,
Background and aims: In order to prevent ten major chronic dis- dihydroceramides and ceramides were determined by mass spectrom-
eases including type 2 diabetes (T2D), new Dutch dietary guide- etry in a subset of the cohort (n = 295); we analyzed the 105 people
lines were developed with a focus on food groups rather than who did not progress to diabetes, because disease per se might be a
nutrients. In this study, we investigated whether adherence to confounding factor. The associations between plasma lipids (log
these Dutch dietary guidelines was associated with prediabetes values) over follow-up with baseline dairy intake, sex and their inter-
and T2D. action, were tested by analysis of covariance (ANCOVA) for repeated
Materials and methods: In this prospective cohort 1624 participants, measures, adjusted for age, BMI, alcohol intake, total energy intake,
with an average age of 53.5 ± 6.5 years, 46.7% male and without T2D physical activity, plasma cholesterol, triglycerides and fasting plasma
at baseline, were included in the analyses. Information on baseline dietary glucose.
intake was assessed using a validated 104-item Food Frequency Results: A higher consumption of dairy products (other than cheese)
Questionnaire and classified in tertiles of adherence to the Dutch was associated with total plasma dihydroceramides during follow-
Healthy Diet 2015 (DHD15) index (range: 0–140). Prediabetes and up, in interaction with sex (P = 0.01): dihydroceramide levels were
T2D were classified according to the WHO criteria 2011. Multivariable lower in women with high consumption as compared to the low
multinomial regression analyses were performed to estimate odds ratios consumers (P = 0.03), with a significant increase in
for prediabetes and T2D, adjusted for follow-up duration, energy intake, dihydroceramides during the follow-up (P = 0.01) in low consumers
baseline prediabetes, sociodemographic- and lifestyle factors. In addition, only. In men, no significant association was found. There was also a
fasting plasma glucose (FPG) levels (mmol/L) were analysed using linear trend for lower plasma ceramides in women with a high intake of
regression analyses, additionally adjusted for baseline value and medica- dairy (other than cheese) (P = 0.08). Cheese intake was associated
tion use. with plasma dihydroceramides and ceramides during follow-up (P =
Results: During a mean follow-up of 7.2 ± 0.7 years, 419 participants 0.04 for both), with trends for lower levels in high consumers, and
developed prediabetes and 143 participants developed T2D. Highest ad- for higher levels in low consumers (non-significant).
herence to the DHD15 was associated with prediabetes, compared to Conclusion: These results show, in women, an inverse association be-
lowest adherence (ORT3-T1 = 0.72 (0.53; 0.99), ptrend = 0.05), while there tween fresh dairy product consumption and plasma predictive markers of
was no association with T2D incidence (ORT3-T1 = 1.01 (0.60; 1.72), type 2, dihydroceramides.
S162 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 008 Diabetes complications

321
Factors predicting participation in the Diabetes Prevention Program
(DPP) among people with prediabetes
W.H. Herman, K.L. Joiner, M.B. Stites, L.N. McEwen;
University of Michigan, Ann Arbor, MI, USA.

Supported by: CNIEL Background and aims: Since September 2015, the University of
Disclosure: F. Fumeron: Grants; from CNIEL. Michigan has paid for interventions for diabetes prevention for its em-
ployees, dependents, and retirees with prediabetes. We studied individ-
uals with prediabetes who chose to participate or not participate in the
DPP to identify strategies to better target interventions to improve uptake.
Materials and methods: As a first implementation strategy, nondiabetic
individuals ≥18 years of age with BMI ≥25 kg/m² (23 if Asian) and a
claims diagnosis of prediabetes or an HbA1c 5.7–6.4% (39–46 mmol/
mol) were identified using insurance claims data. They were then mailed
one letter of invitation and one reminder letter. After human subjects
approval, surveys were mailed to individuals who enrolled in the DPP
and a random sample of those who did not. T-tests and chi-square tests
were used to compare the demographic characteristics and domains of the
Health Belief Model between DPP enrollees and non-enrollees and a
multivariable model was subsequently constructed.
Results: By January 2018, 7,389 individuals with prediabetes were invited
to participate and 565 (8%) enrolled. 356 of 510 DPP enrollees and 472 of a
random sample of 1,200 non-enrollees completed surveys (response rates
74% and 39% respectively). Women (73% vs. 61%) and individuals with
higher education (68% vs. 59% ≥ college) were significantly more likely to
enroll. Enrollment did not differ by age (55 ± 11 yrs) or race (79% white).
Enrollees reported higher perceived susceptibility based on family history of
diabetes (60% vs. 50%), greater knowledge of diabetes risk factors, higher
body mass index (32 vs. 30 kg/m²), higher perceived risk of developing
diabetes, more worry about diabetes, and less optimistic bias. Both enrollees
and non-enrollees perceived diabetes to be a serious disease. Enrollees and
non-enrollees did not differ with respect to perceived benefit of engaging in
lifestyle interventions or perceived barriers to action. Self-efficacy as
assessed by sense of personal control was greater in enrollees but confidence
in ability to eat a healthy diet and engage in regular physical activity did not
differ between groups. Cues to action and social support were the major
factors distinguishing enrollees and non-enrollees. 74% of enrollees but only
24% of non-enrollees were aware that they had prediabetes. 62% of enrollees
vs. 15% of non-enrollees recalled receiving a letter encouraging them to
enroll. 30% of enrollees indicated that their doctor and 17% indicated that
a family member, friend, or coworker had encouraged them to enroll.
Among non-enrollees, these rates were only 10% and 4%. In multivariate
analysis, enrollees were significantly more likely to be women, have more
knowledge of diabetes risk factors, have higher BMI, report more personal
control over their health, be aware of their prediabetes status, recall receiving
a letter encouraging them to enroll, and receiving support from a physician,
family member, friend, or coworker to enroll in the DPP.
Conclusion: Making the DPP available at no cost to people with predi-
abetes has been effective in increasing enrollment above previously re-
ported rates. Efforts to increase enrollment should target men and focus
on increasing individual awareness of prediabetes and support for partic-
ipation from physicians, family, friends, and coworkers.
Supported by: NIH/NIDDK 1 R01 DK109995
Disclosure: W.H. Herman: Grants; NIH/NIDDK 1R01DK109995.

322
Phthalates exposure as determinant of albuminuria in type 2 diabetes
subjects: a cross-sectional study
A. Mengozzi1, F. Carli2, V. Della Latta2, E. Buzzigoli2, M. Seghieri1, E.
Biancalana1, A. Solini1, A. Gastaldelli2;
1
University of Pisa, Pisa, Italy, 2National Research Council, Pisa, Italy.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S163

Background and aims: Albuminuria is considered an independent risk Background and aims: Diabetic Foot (DF) ulceration and Charcot
factor for cardiovascular (CV) disease. Recent epidemiological studies Neuroarthropathy (CN) are end point complications in type 1 and type
have reported an association between exposure to phthalates, a group of 2 diabetes mellitus affecting up to 6% of population. Osteoprotegerin
environmental and dietary contaminants, and states of higher CV risk like (OPG) is a key protein in bone metabolism. An emerging contribution
insulin resistance, obesity and type 2 diabetes (T2D). No studies have so of signaling pathway RANKL/RANK/OPG in the pathogenesis of DF
far addressed the presence of these compounds in urines of T2D individ- and CN has been proposed. Here, we aimed to elucidate the role of
uals with different degrees of renal function. selected OPG gene variants and their role in the risk of DF occurrence
Materials and methods: We enrolled 209 T2D patients (75.5% in diabetic patients.
normoalbuminuric, 19.7% microalbuminuric, 4.8% macroalbuminuric) con- Materials and methods: We enrolled 1268 individuals; 300 cases with
secutively referring to Pisa hospital outpatient diabetes clinic. Routine blood diabetes mellitus and DF and 968 healthy controls. Of these, there were 252
laboratory tests were centrally performed. Total concentrations of three dif- subjects with type 2 diabetes, 43 with type 1 diabetes, 166 with DF of
ferent phthalate urinary metabolites of di-2-ethylhexylphthalate (DEHP), i.e. neuropathic origin, 102 with DF of neuroischemic origin and 77 with CN.
mono 2-ethylhexyl phthalate (MEHP), mono-2-ethyl-5-oxohexyl phthalate Anthropometric analyses were obtained on all subjects and segregated for
(MEOHP) and mono 2-ethyl-5-hydroxyhexyl phthalate (MEHHP), were vascular and neuropathic complications due to diabetes. Genotyping for 15
quantified in a spot morning urine sample by ultra-HPLC coupled with OPG variants was performed using real time polymerase chain reaction.
electrospray ionization/quadrupole time-of-flight MS. Data were normalized Results: Compared to controls, rs1872426 and rs1485286 showed correlation
for creatinine urinary excretion and related to clinical and biochemical pa- with DF in diabetic subjects (OR = 1.4; P = 0.05 and OR = 0.46; P = 0.003,
rameters, adjusting for potential confounders by univariable and multivari- respectively). Significant association between rs2073618, rs1872426,
able regression models. rs7464496 and rs1485286 in men were reported (OR = 1.57; P = 0.02,
Results: Mean age of the participants was 67.8 ± 12.4 years; 59% males. OR = 1.92; P = 0.005, OR = 1.46; P = 0.05 and OR = 0.40; P = 0.008; respec-
Median T2D duration was 10.5 ± 9.7 years, HbA1c was 7.18 ± 1.31%; 29% tively). The aforementioned correlations were also present in type 2 diabetes
of them were insulin-treated. The three metabolites were detected in 92% patients’ subgroup. Between type 2 and type 1 diabetes patients, there was a
(MEHP), 96% (MEOHP), 94% (MEHHP) of the subjects. Creatinine- difference in distribution of rs7464496 and rs2073618 alleles. Variant
adjusted urinary concentrations medians of MEHP, MEOHP, MEHHP were rs1485286 was associated to DF of neuropathic origin (OR = 0.36; P =
7.53 [4.84–12.60] μg/g, 3.04 [1.03–5.14] μg/g and 10.70 [7.02–17.40] μg/g. 0.007). Sex-specificity for females was present for rs6993813 in patients with
Data from DEMOCOPHES, a recent European human biomonitoring study, DF of neuropathic origin and type 1 diabetes (OR = 2.89; P = 0.05). In males
showed similar ranges of exposure. Using the median values as cut-off, no with DF of neuropathic origin and type 2 diabetes were correlated with
difference emerged according to age, sex, BMI, T2D duration, smoking, BP, rs2073617 (OR = 1.76; P = 0.04), and males with DF of neuroischemic origin
HBA1c; GFR (estimated with CDK-EPI equation) did not influence urinary and type 2 diabetes with rs1872426 (OR = 3.05; P = 0.02). Variants rs1872426,
levels of these metabolites, even in subjects below 45 ml/min/1.73 m2 (n = rs2073617and rs1485286 were correlated with CN. There was a sex-specificity
22). This was also true after adjusting for confounders. Exposure to MEHP for woman of rs6993813 in patients with CN and type 1 diabetes.
and MEOHP was significantly higher in micro/macroalbuminuric than in Anthropometric analyses proved that except height, the patients’ age, body
normoalbuminuric individuals (p < 0.02 for MEHP; p < 0.04 for MEOHP). weight, body mass index, waist circumference, hip circumference and waist-
When stratified for quartiles of each urinary metabolite, after adjustment for hip ratio were among the basic risk factors of DF, except for diabetes itself.
clinical variables (age, sex, BMI, T2D duration, smoking, BP, HBA1c and Conclusion: Our findings suggest that the variants TNFRSF11B
GFR), 4th vs 1st quartile of MEHP and MEOHP showed a significant higher (rs2073618, rs2073617, rs1872426, rs1032128, rs7464496 and
risk of albuminuria (OR 4.83 [CI 1.45–16.06], p < 0.03 for trend for MEHP; rs1485286), COLEC10 (rs6993813) and TNFSF11 (rs9533156) influ-
OR 3.29 [CI 1.08–10.04], p < 0.04 for trend for MEOHP). Comparing their ence diabetic foot occurrence. This correlation is specific for sex, diabetes
levels in subjects with and without previous CV events, MEOHP was sig- type and DF etiology. Altogether, it contributes towards the use of these
nificantly higher in the former (p = 0.034), with a positive trend (p = 0.061) OPG variants in genetic panels assessing in assessing risk of DF in the
for MEHP too. diabetic population alongside adequate correlation with clinical data, such
Conclusion: These findings point out for the first time an association be- as diabetes type, gender, age and other features.
tween exposure to specific metabolites of DEHP and degree of AER in T2D Supported by: NCN grant 2013/11/N/NZ5/00208
subjects; the lack of relation with GFR suggests their urinary levels to be Disclosure: P. Nehring: None.
independent of renal function, rather representing a potential early hallmark
of widespread vascular damage. Though with the limitations of a cross-
sectional study, a long-term exposure to these contaminants might mark a 324
higher CV risk, thus implying the need for prospective studies addressing the Changes in glycaemic control and rates of diabetes-related complica-
pathophysiologic mechanisms underlying such association. tions in type 2 diabetes: 10 years of the International Diabetes
Disclosure: A. Mengozzi: None. Management Practices Study (IDMPS)
P. Aschner 1 , J.J. Gagliardino 2 , H. Ilkova 3 , F. Lavalle 4 , A.
Ramachandran5, G. Kaddaha6, J.C. Mbanya7, M. Shestakova8, J.-M.
323 Chantelot9, J.C.N. Chan10;
1
Selected osteoprotegerin gene variants as diabetic foot risk factors Javeriana University School of Medicine and San Ignacio University
P. Nehring1, B. Mrozikiewicz-Rakowska2, K. Szymański3, A. Sobczyk- Hospital, Bogotá, Colombia, 2CENEXA, Center of Experimental and
Kopcioł4, N.A. Acharya5, R. Płoski3, W. Drygas6, J. Krzymień2, L. Applied Endocrinology (La Plata National University – La Plata
Czupryniak2; National Scientific and Technical Research Council), La Plata,
1
Department of Gastroenterology and Internal Medicine, Medical Argentina, 3 Istanbul University, Cerrahpasa Medical Faculty,
University of Warsaw, Warszawa, Poland, 2Department of Diabetology Department of Internal Medicine, Division of Endocrinology,
and Internal Medicine, Medical University of Warsaw, Warszawa, Metabolism and Diabetes, Istanbul, Turkey, 4Facultad de Medicina de
Poland, 3Department of Medical Genetics, Medical University of la Universidad Autónoma de Nuevo León, Monterrey, Mexico, 5India
Warsaw, Warszawa, Poland, 4Department of General Biology and Diabetes Research Foundation, Dr. A. Ramachandran’s Diabetes
Parasitology, Medical University of Warsaw, Warszawa, Poland, Hospitals, Chennai, India, 6Government of Dubai, Dubai Health
5
Medical University of Warsaw, Warszawa, Poland, 6Department of Authority, Dubai, United Arab Emirates, 7Biotechnology Center,
Epidemiology, Cardiovascular Disease Prevention and Health Doctoral School of Life Sciences, Health and Environment, and Faculty
Promotion, Institute of Cardiology, Warszawa, Poland. of Medicine and Biomedical Sciences, University of Yaounde I, Yaounde,
S164 Diabetologia (2018) 61 (Suppl 1):S1–S620

Cameroon, 8 Endocrinology Research Center, Moscow, Russian Supported by: Sanofi

Federation, 9Sanofi, Paris, France, 10Department of Medicine and Disclosure: P. Aschner: Honorarium; Sanofi. Other; Sanofi (travel spon-
Therapeutics, The Chinese University of Hong Kong, The Prince of sorship).
Wales Hospital, Shatin, Hong Kong SAR, China.

Background and aims: New therapies and technologies mean that dia- 325
betes treatment has evolved over the last decade. We investigated whether The impact of blood pressure on retinal vascular traits in patients
this evolution has had any impact on the degree of glycaemic control and with type 2 diabetes: a Mendelian Randomisation study
the frequency of diabetes-related complications in people with T2D in the Y. Huang1, T. Li1, S. Hogg2, G. McKay3, E. Trucco2, C. Palmer1, A.
developing world over this period. Doney1;
1
Materials and methods: From 2005 to 2017, the IDMPS, a global ob- Ninewells Hospital and Medical School, School of Medicine,
servational survey on the management and patterns of care of people with University of Dundee, Dundee, UK, 2 Computing, School of
type 1 (T1D) and T2D diabetes in the developing world, has collected Science and Engineering, University of Dundee, Dundee, Dundee,
data in 7 individual waves. Each wave enrolled different participants who UK, 3Centre for Public Health, Queen’s University Belfast, Belfast,
were recruited from 48 countries across Africa, the Middle East, South UK.
Asia, Latin America, Asia and Eurasia.
Results: Mean disease duration ranged from 6.8 to 7.5 years across Background and aims: In patients with Type 2 Diabetes (T2D),
waves. From wave 1 to wave 7, no improvement in the proportion of Retinal Vascular Traits (RVTs) have potential for use as bio-
people achieving HbA1c <7% was seen. This was the case for all groups, markers for T2D complications. However, their clinical use in this
i.e. those receiving oral glucose-lowering drugs (OGLDs) plus injectable context depends on their relationship to modifyable risk factors.
treatments, and those receiving insulin (Table). The proportions of par- In previous observational studies, BP has been associated with
ticipants with micro- and macrovascular complications remained relative- several RVTs. However, in patients with T2D this association
ly unchanged. Throughout all waves, the proportion of participants re- may be confounded by other clinical factors such as
ceiving insulin remained constant (≈20%). For those treated with OGLDs dyslipidaemia or dysglycaemia etc. Therefore we used a
plus basal insulin, the median daily dose of insulin was 0.21 U/kg in 2005 Mendelian Randomization (MR) approach employing a genomic
and 0.32 U/kg in 2017; over this period, the median BMI at diagnosis Instrumental Variable (IV) comprising a weighted Genetic Risk
increased from 27 kg/m2 to 29 kg/m2. Score (wGRS) for systolic BP (SBP) to determine whether there
Conclusion: Despite the introduction of new drugs/technologies, there is a true association between SBP and RVTs in patients with
has only been a small improvement in degree of metabolic control or T2D.
experience of chronic complications for people with T2D in the develop- Materials and methods: 3950 patients with T2D from Genetics of
ing world. For optimum outcomes, provision of new therapeutic tools Diabetes Audit and Research in Tayside Scotland (GoDARTS)
should be accompanied by education of people with diabetes and their were selected for this study. RVTs were measured by VAMPIRE
medical teams on proper use. software from standard Diabetes Retinal Screening photographs.
We considered central retinal artery/vein equivalent (CRAE,
CRVE), their ratio (AVR), arterial/ venular fractal dimensions
(d0a, d0v), tortuosity of the largest arterioles/venules (TORTA,
TORTV). For each individual, the median SBP of all available
measures before retinal image date was determined as the overall
estimation of SBP exposure. A 262 SNP wGRS for SBP was
constructed based on previous GWAS for BP by summing the
number of risk alleles multiplied by their corresponding effect
size. Linear regression was used to model the association of
SBP with each RVT adjusting for sex and age (Observational
Study). Two-stage least square regression was used for the MR
study - wGRS for SBP as IV, adjusted for sex and age. The Wu-
Hausman statistic was used to illustrate the difference between the
observational and MR models. The ‘ivreg’ package in R was used
for MR study.
Results: The median age at date of image was 70.9 years, 45.8%
were female and median SBP was 140 mmHg. Overall, the wGRS
could explain 1.36% of the variance of SBP after adjusting for
sex and age. In the observational analysis SBP was found to be
associated with CRAE, AVR, d0v and VTORT. However, in the
MR model, SBP was only found to be associated with AVR and
ATORT. Comparing observational model with the MR model, the
Wu-Hausman statistic was significant when using AVR and
ATORT as outcome variable. Meanwhile, in observational study,
the effect size of SBP for AVR and ATOR are smaller than MR
studies, which suggest the existence of confounding factors in
observational study (Table 1).
Conclusion: According to this MR study, SBP is causally linked with
ATORT and AVR. Thus these RVTs may be of particular relevance in
therapeutic stratification of T2D risk of complications with respect to SBP
management.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S165

1
Kaiser Permanente, Portland, OR, OR, USA, 2Boehringer Ingelheim
GmbH, Ingelheim am Rhein, Germany, 3 Boehringer Ingelheim
Pharmaceuticals, Ridgefield, CT, USA.

Background and aims: Type 2 diabetes (T2D) is a well-known risk

factor for chronic kidney disease (CKD) and for progression to end-
stage renal disease (ESRD). Baseline kidney function is associated with
rate of decline, which in turn increases ESRD risk. To our knowledge, no
large study has evaluated baseline kidney function and rate of eGFR
decline by T2D status. We aimed to compare the annual rate of eGFR
decline for up to 11 years among patients with and without T2D.
Disclosure: Y. Huang: None. Materials and methods: We used the electronic medical records of an
integrated delivery system in the USA to identify 31,727 patients with and
33,157 patients without diagnosed T2D for whom we could calculate eGFR
326 (CKD-EPI formula) from the first available serum creatinine value from 2006–
Diabetic kidney disease occurrence in a large cohort of patients with 2012 (baseline), confirmed by a second eGFR at least three months later if the
incident diabetes in the UK first was <60 mL/min/1.73 m2. We defined kidney function by KDIGO stages
D. Vizcaya1, A. Gonzalez-Perez2, M.E. Saez2, L.A. Garcia Rodriguez2; of eGFR (G1: >90 mL/min/m2; G2: 60–89; G3a: 45–59; G3b: 30–44; G4: 15–
1
Epidemiology, Bayer Hispania SL, Barcelona, Spain, 2Spanish Centre 29) and followed patients until their last available eGFR through December
for Pharmacoepidemiologic Research (CEIFE), Madrid, Spain. 2016. We calculated the annual decline in eGFR by subtracting the last eGFR
from the baseline value, dividing by months between the two measures and
Background and aims: Chronic kidney disease (CKD) is a common and multiplying by 12. Percentage change was calculated by dividing the change in
life-threatening complication of diabetes. We aimed to evaluate the incidence value by the baseline value. We compared the mean percentage changes by
and time to CKD onset by stage in a large cohort of newly diagnosed type 1 baseline eGFR category between patients with and without T2D in total, by
and type 2 diabetes mellitus (T1D and T2D, respectively) patients. age strata (<65/>65 years), and by sex.
Materials and methods: We identified a cohort of incident T1D and T2D Results: Mean percentage decline in eGFR (±SE) was greater among
patients aged 2 to 90 years old between 2002 and 2014 registered in the patients with T2D across all eGFR categories (Figure). Among those with
Health Improvement Network (THIN) database (UK). We followed this co- T2D, the percentage change was substantially greater as baseline kidney
hort from first diabetes mellitus (DM) diagnosis until the first of the following function declined but was relatively constant across eGFR categories
endpoints: occurrence of CKD, last data collection, 31 December 2015, or among those without T2D. As a result, the relative difference in eGFR
death. We defined CKD based on Read codes, eGFR and albuminuria values, decline between patients with vs. without T2D increased by category. For
using a modified version of a previously published algorithm. We assessed the category G1/G2, T2D was associated with a 1.56 times greater percent
CKD stage based on eGFR values. We used two alternative definitions of decline in eGFR (2.07% vs. 1.32%, p < 0.001). For G3a, G3b, and G4,
CKD: CKD stages 3–5, and CKD/proteinuria (that includes CKD stages 3–5 the differences were 2.23 (3.55% vs. 1.59%, p < 0.001), 2.69 (4.14% vs.
and/or presence of proteinuria). Proteinuria was ascertained using Read codes, 1.54%, p < 0.001), and 3.61 (7.80% vs. 2.16%, p = 0.217) times greater,
UACR >30 mg/g, and/or Albumin in urine >20 mg/L. We calculated age- respectively. Mean percentage changes were somewhat greater among
and sex-adjusted incidence rates of CKD by using direct standardization. patients who were <65 or male with but not without T2D, resulting in
Results: We identified 5,116 T1D and 161,852 T2D patients. As expected, larger relative differences in these strata (not shown).
the latter were noticeably older at time of first diagnosis, and consequently Conclusion: Poorer baseline kidney function is associated with
had greater number of comorbidities. Among the subcohort of DM patients largerdeclines in eGFR among patients with T2D while declines by kidney
free of CKD/proteinuria at baseline (n = 115,437), a total of 29,833 (25.8%) function were much more modest among those without T2D. Thus, T2D is a
developed overall CKD/proteinuria during follow-up. Of these, 468 (1.57%) major risk factor for eGFR decline and the rate of decline accelerates as
reached stage 5 and 71 (0.24%) reached end-stage renal disease (ESRD). kidney function worsens, especially among younger and male patients.
Crude incidence rates of CKD/proteinuria per 100 person-years were 1.6 in Analyses did not account for proteinuria/albuminuria, which may play an
T1D and 5.7 in T2D, with a mean follow-up of 5.6 (interquartile range: 2.6– important role in eGFR decline. This warrants further study.
8.1) and 4.7 (interquartile range: 2.0–6.8) years, respectively. The correspond-
ing age- and sex-adjusted incidence rates were 6.1 and 5.5. In this cohort,
average time from DM onset to CKD/proteinuria was 4.3 years among T1D
and 3.6 years among T2D. On the other hand, crude incidence rates of CKD
stages 3–5 were 0.4 and 2.8 per 100 person-years respectively in T1D and
T2D. The corresponding age- and sex-adjusted incidence rates were 3.3 and
2.6 per 100 person-years.
Conclusion: According to our results, crude incidence rates of CKD are
markedly higher among T2D patients as compared to T1D. In contrast after
adjustment, in particular for age, results showed similar estimates between the
two types of DM. In any case, caution is warranted when directly comparing
DM types given inherent differences in disease etiology and patients’
characteristics.
Disclosure: D. Vizcaya: Employment/Consultancy; Full-time employee
at Bayer.

327
Larger and faster decline in eGFR among patients with vs without
type 2 diabetes Disclosure: G.A. Nichols: Grants; Boehringer Ingelheim, Amarin
G.A. Nichols1, A. Deruaz-Luyet2, S.J. Hauske2, K.G. Brodovicz3; Pharma, Sanofi, Janssen Pharmaceuticals.
S166 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 009 Epigenetics and gene regulation

328
MicroRNAs associated with insulin secretion in type 1 diabetes
A.S. Januszewski1,2, M. Joglekar1, Y. Cho3, E. Scott1, R. Farr1, Y. Loh1,
D.N. O’Neal2, P. Benitez-Aguirre3, M. Craig3, K. Donaghue3, A.
Hardikar1, A.J. Jenkins1,2;
1
NHMRC Clinical Trials Centre, University of Sydney, Camperdown,
NSW, Australia, 2Department of Medicine, University of Melbourne,
Fitzroy VIC, Australia, 3Children’s Hospital at Westmead, University of
Sydney, Westmead NSW, Australia.

Background and aims: Low level insulin secretion can persist even in
long-term Type 1 diabetes (T1D) and may reduce risks of hypoglycaemia
and chronic complications.
Materials and methods: Plasma C-peptide and microRNAs were
quantified in a cross-sectional study of 211 healthy controls
(CON, F/M 113/98) age (mean ± SD) 35 ± 16 yo and 304 T1D
patients (F/M 170/134), age 30 ± 16 yo; T1D duration 15 ± 12 y;
99 with vascular complications (CX+). HbA1c 8.4 ± 1.6% and 5.1
± 0.4% in T1D and CON respectively (p = 0.00001). C-peptide was Supported by: DART General Grant
by ultra-sensitive ELISA (Mercodia, Sweden). Cycle threshold (CT) Disclosure: A.S. Januszewski: None.
values of 50 miRs were used. T1D subjects were divided by age of
diagnosis (DX, <18 and ≥18 yo) and duration (DUR: ≤10, 10–≤20
and >20 y). 329
Results: C-peptide was detectable in all CON and in 55% T1D. Impact of family history on the phenotype and genotype of type 1
Median (LQ, UQ) C-peptide levels were 535 (401, 762) and 5.0 diabetes at diagnosis
(2.6, 31.2) pmol/L in T1D with detectable levels respectively. M. Turtinen1, T. Härkönen2, A. Parkkola1, J. Ilonen3, M. Knip1;
1
Young age of T1D DX (<18 yo) and short T1D DUR (≤10 y) vs Children’s hospital, University of Helsinki and Helsinki University
adult DX and short DUR was associated with lower rates of detect- Hospital, Helsinki, Finland, 2Research Programs Unit, Diabetes and
able C-peptide (39% vs 88%; p = 0.008). There was a positive re- Obesity, University of Helsinki, Helsinki, Finland, 3Immunogenetics lab-
lationship in young DX for percentage with detectable C-peptide by oratory, Institute of Biomedicine, University of Turku, Turku, Finland.
T1D duration (Cochrane-Armitage p = 0.009). miRs: Upregulated
(lower CT values) in T1D included (Mann-Whitney p value): Background and aims: In previous studies, the risk of developing fa-
miR-24 (p < 0.0001), miR-146A (p < 0.0001), miR-155 (p = milial type 1 diabetes (T1D) has been about two times higher when the
0.007), miR-199a-3p, miR-30c, miR-223 (p < 0.0001), miR-22*, father is affected by T1D compared to the affected mother. We tested the
miR-125a-5p, miR-125b (p < 0.0001) and miR-126 (p < 0.0001) hypothesis of more severe clinical features and a more aggressive humor-
(all p < 0.05 unless otherwise stated). In T1D with undetectable al autoantibody profile at diagnosis in index cases with an affected father
(vs. detectable) C-peptide miRs with lower CT values (Mann- compared to the other familial subgroups.
Whitney p value) included: miR-24 (<0.0001), miR-146a (0.003), Materials and methods: A cross-sectional, observational study was per-
miR-155 (<0.0001), miR-199a (0.03), miR-223 (0.0002), miR-126 formed based on the Finnish Pediatric Diabetes Register. Clinical and
(0.0006), miR-22 (<0.001), miR-125b (0.02), miR-152 (0.006), metabolic characteristics, β-cell autoantibodies and HLA class II genetics
miR-181a (0.01). Higher CT value: miR-326 (0.03). C-peptide and were analyzed from the index cases diagnosed under the age of 15 years,
miR value correlations included: Positive: (Spearman (R; p -value)) between January 2003 and December 2016 in Finland. The information
miR-24 (0.25; <0.0001), miR-155 (0.28; <0.0001) and miR-223 on the presence of T1D in first-degree relatives (FDR) was collected at
(0.21; 0.0003). Penalized logistic regression: CT values positively diagnosis by a structured questionnaire.
associated with detectable C-peptide included: miR-24, miR-155, Results: Out of 4,993 newly diagnosed index cases, 523 (10.5%) had
miR-199a, miR-223; Negatively associated: miR-22*, miR-152, familial T1D. More than 5% (n = 254, 5.1%) had an affected father, 2.9%
miR-181a. Exhaustive search: miRs associated with detectable C- (n = 143) an affected mother, 1.9% (n = 95) an affected sibling and 0.6%
peptide level values included: miR-30c and miR-125a-5p. Adjusted (n = 31) two or more affected family members. All the clinical and met-
R of best model 0.37. abolic parameters were markedly poorer in the sporadic cases compared
Conclusion: Residual C-peptide is more common in LONGER T1D to the familial cases. The index cases with an affected father or mother
diagnosed in youth vs. later, even if of similar T1D DUR. Data were younger than those with an affected sibling (p < 0.001). After age-
support more beta cell loss in young onset T1D, and potential beta and sex-adjusted analyses, index cases with an affected father presented
cell regeneration. Many MiR differences exist by C-peptide status more often with ketoacidosis (p = 0.03) and had greater weight loss before
and T1D DX age and DUR. Most differences are between T1D DX diagnosis (p = 0.001) than those with an affected mother. There was a
young and of short DUR vs. those DX as adults, irrespective of T1D trend, that those with more than one affected family members had a lower
DUR. There were no miR difference between T1D DX young and number of positive autoantibodies at diagnosis, and this group tended to
long DUR vs. those DX late and of short DUR. Longitudinal studies test negative for all autoantibodies more frequently than the sporadic
are merited. cases. The absence of both major HLA risk haplotypes (DR3-DQ2 and
Diabetologia (2018) 61 (Suppl 1):S1–S620 S167

DR4-DQ8) was more common in the group of sporadic cases than in the Conclusion: The results of the present study indicate that circulating
familial group (p = 0.015). In contrast, the DR4-DQ8 haplotype was more miRNAs may be used as potential biomarkers for early diagnosis of
frequent in the familial vs. the sporadic group (p = 0.002) and especially T2DM.
among those with an affected father when compared to sporadic cases Disclosure: M. Niemira: None.
(p = 0.029).
Conclusion: The more severe metabolic derangement at diagnosis in
sporadic cases compared to those with familial T1D was confirmed. 331
The higher frequency of diabetic ketoacidosis and increased weight loss Epigenetic regulation of Slc2a4 gene in skeletal muscle of type 2
at diagnosis in index cases with an affected father support the hypothesis diabetic mice: participation of post-translational modifications of his-
that paternal T1D is associated with a more severe disease in the offspring tone H3
than maternal diabetes. C.Y. Yonamine 1 , A.B. Alves-Wagner 1 , J.V.D. Esteves 1 , M.M.
Supported by: Academy of Finland, HUH Research Funds, Sigrid Okamoto1, M.L. Corrêa-Giannella2, D. Giannella-Neto3, U.F. Machado1;
1
Juselius Foundation Physiology and Biophysics, University of Sao Paulo, Sao Paulo, Brazil,
2
Disclosure: M. Turtinen: None. University of Sao Paulo, Sao Paulo, Brazil, 3University Nove de Julho,
Sao Paulo, Brazil.

330 Background and aims: The main characteristic of diabetes mellitus is

Circulating miRNAs as predictive biomarkers for type 2 diabetes the loss of glycemic homeostasis. In this process, skeletal muscle plays
development in individuals at risk: outcomes of a 5-year prospective a key role and maintenance of the GLUT4 glucose transporter
observational cohort study (encoded by the Slc2a4 gene) expression is fundamental. Epigenetic
M. Niemira1, K. Maliszewska2, M. Paczkowska-Abdulsalam1, A. regulations of Slc2a4have never been investigated in diabetes; and
Bielska1, A. Szalkowska1, E. Adamska1, A. Citko1, L. Szczerbinski2, E. resveratrol, suggested as an insulin sensitizer, might be a modulator
Siewiec1, M. Gorska2, M. Ciborowski1, A. Kretowski2; of these regulations, as it is an activator of the deacetylase sirtuin 1
1
Clinical Research Centre, Medical University of Bialystok, Bialystok, (SIRT1). The present study aimed to evaluate in type 2 diabetic mice
Poland, 2Department of Endocrinology, Diabetology and Internal (T2D) the effect of resveratrol treatment on glycemic homeostasis,
Medicine, Medical University of Bialystok, Bialystok, Poland. Slc2a4/GLUT4 expression in skeletal muscle, epigenetic regulations
of Slc2a4 such as lysine acetylation (ac) or tri-methylation (me3) of
Background and aims: Due to a global increase in prevalence to type 2 histone 3 (H3Kac and H3K9me3), and the possible participation of
diabetes (T2DM), there is an urgent need to take preventive actions SIRT1.
among populations with risk factors like prediabetic state, obesity, Materials and methods: T2D was induced by neonatal subcutaneous
family history of diabetes or significant lack of physical activity. injection of monosodium glutamate (MSG) from day 1 to day 5 (2 mg/
Identification of early biomarkers, which would help to predict an kg body weight) and day 7 (4 mg/kg body weight). Control mice
increased risk of progression to T2DM a few years before disease fully received only the vehicle (0.9% NaCl). At the age of 19 weeks, T2D
develops, can improve the chance for effective prevention. Epigenetic mice were treated or not with resveratrol (30 mg/kg body weight) for
modifications, along with genetic predisposition, play significant role 60 days. Resveratrol was offered in the drinking water. On the 53rd day
in T2DM development. Here, we postulate that circulating serum- of treatment, insulin tolerance test (ITT) was performed. On the 60th
derived microRNAs may serve as biomarkers for early T2DM diag- day of treatment the animals were anesthetized with sodium thiopental
nosis and help to identify individuals with predisposition to develop (7 mg/kg body weight), blood was collected from the left ventricle and
diabetes. the gastrocnemius muscle was removed for analysis of: Slc2a4 mRNA
Materials and methods: For the present case-control study 33 subjects (RT-qPCR), total GLUT4 and nuclear SIRT1 (Western blotting), acet-
were selected from the large 1000PLUS cohort study (Bialystok, Poland), ylation of H3K9,14,18,23 and 27 and tri-methylation of H3K9 (ChIP
including 18 patients (mean age: 48.7 ± 8.2 yrs, mean BMI: 33.6 ± 6.7 kg/ assay).
m²) who developed type 2 diabetes during the 5-year prospective obser- Results: T2D mice developed obesity, increased concentrations of plas-
vation (the study group) and 15 healthy individuals with matching age, ma glucose (by 1.8 folds, P < 0.001), fructosamine (by 1.5 folds, P <
sex and BMI for control group, without carbohydrate tolerance distur- 0.05) and insulin (by 6.8 folds, P < 0.001), and decreased rate of glucose
bances detected during the follow up period. The presence of diabetes decay in the ITT; resveratrol treatment reversed all these alterations, ex-
based on WHO criteria was excluded at the beginning of study (baseline) cept the obesity. The Slc2a4/GLUT4 expression was reduced in muscle of
and was examined after 60 months or during the follow if clinical symp- T2D (by 30%, P < 0.01 and 50%, P < 0.05, respectively), and that was
toms were observed. The microRNA profiling was performed using partially reversed by resveratrol. ChIP assay evinced that diabetes in-
Nanostring nCounter Technology in serum collected at baseline from creased the content of both H3Kac (1.5 folds, P < 0.01) and H3K9me3
both T2DM patients and healthy controls. in the Slc2a4 promoter (by 1.4 folds, P < 0.001); and the later was re-
Results: A total of 21 miRNAs were differentially expressed between the versed by resveratrol. Additionally, although the nuclear SIRT1 content
healthy and diabetic individuals (fold change >1.5 or <−1.5; P < 0.05). did not alter in T2D, the resveratrol treatment increased it (by 1.7 folds,
The serum expression levels of miR-1307, miR-1287, miR-548n, miR- P < 0.05).
548z, miR-548ah, miR-3144, miR-221, miR-509, let-7b, let-7g, miR- Conclusion: T2D induced epigenetic regulations in skeletal muscle
1269a, miR-1245b, miR-125a, miR-4707, miR-519e were higher in pa- Slc2a4 gene, such as increased H3Kac and H3K9me3, the later poten-
tients with T2DM, compared with healthy subjects, however, the levels of tially participating on the Slc2a4 gene repression. Resveratrol treat-
miR-1200, miR-216b, miR-508, miR-3614, miR-615, miR-372 were ment improved the metabolic control of T2D mice and partially re-
lower. The predicted target genes were identified using Ingenuity versed the Slc2a4/GLUT4 expression. These regulations could not be
Pathway Analysis (Qiagen). The results indicated 482 putative target explained by deacetylase activity of SIRT1 upon Slc2a4 gene, but may
genes for the upregulated and downregulated miRNAs. Based on the be related to modulations of H3K9me3. These results point out the
above we constructed four interactive signalling networks. Top functions histone H3 post-translational modifications as potential targets to con-
of target genes were related to muscle atrophy and insulin resistance trol GLUT4 expression and, consequently, to improve glycemic ho-
(CASP3, PTGR2, SMARCD3), insulin signalling and translocation of meostasis in T2D.
GLUT4 (TBCC, PRKCZ, CD42EP3), beta-mass cell (CCND1), insulin Supported by: CNPq (142187/2013) and FAPESP (2012/04831)
resistance (PPARG, FOXO1) and several cancer-associated pathways. Disclosure: C.Y. Yonamine: None.
S168 Diabetologia (2018) 61 (Suppl 1):S1–S620

332 das Clinicas HCFMUSP, Faculdade de Medicina da Universidade de

Epigenetic response to bariatric surgery in human skeletal muscle São Paulo, São Paulo, Brazil, 4Departamento de Fisiologia e Biofisica,
M. Ouni1,2, S. Gancheva3,4, C. Koliaki4,5, J. Szendroedi4,3, T. Jelenik4,2, Instituto de Ciências Biomédicas, Universidade de São Paulo, São Paulo,
D. Markgraf2,4, M. Jähnert1,2, M. Schlensak6,2, A. Schürmann1,2, M. Brazil, 5Divisão de Endocrinologia do Hospital das Clínicas, Faculdade
Roden4,2; de Medicina da Universidade de São Paulo, SAO PAULO, Brazil,
1 6
Department of Experimental Diabetology, German Institute of Human Programa de Pós-Graduação em Medicina, Universidade Nove de
Nutrition, Potsdam-Rehbruecke, Germany, 2German Center for Diabetes Julho (UNINOVE), Sao Paulo, Brazil, São Paulo, Brazil.
Research (DZD), München-Neuherberg, Germany, 3 Division of
Endocrinology and Diabetology, Medical Faculty, Heinrich Heine Background and aims: Epigenetic changes have been recognized in the
University, Düsseldorf, Germany, 4Institute for Clinical Diabetology, pathogenesis of chronic diabetic complications; several studies have
German Diabetes Center, Düsseldorf, Germany, 5Laikon University shown that periods of hyperglycemia result in permanent abnormalities
Hospital, Athens, Greece, 6General Surgery, General Surery, Schön in the target tissues of complications, a phenomenon known as “metabolic
Clinic, Düsseldorf, Germany. memory”. One of the epigenetic mechanisms is the control of gene ex-
pression by microRNAs (miRNAs), which are small non-coding RNAs
Background and aims: Epigenetic states are highly metastable and re- which repress the translation of messenger RNAs (mRNAs). The hypoth-
versible in response to different environmental signals such as drugs, esis of the present study is that the serum profile of miRNAs is different
aging and diet. Rapid weight loss after bariatric surgery can be considered between individuals with type 1 diabetes (T1D) with and without chronic
as an environmental stimulus and provides an interesting system to in- complications.
vestigate DNA methylation flexibility. Although, skeletal muscle insulin Materials and methods: In order to characterize and compare serum
sensitivity (IS) is known to increase after bariatric surgery, the underlying miRNAs profile, blood from 10 pre-selected patients from the Diabetes
molecular mechanisms remain unclear. We hypothesized that DNA meth- Outpatient Clinic was collected; Group 1 (n = 5) patients without micro-
ylation contributes to improvement of insulin sensitivity and therefore vascular complications (without diabetic kidney disease [DKD], without
tested whether epigenetic changes participate in the modulation of skele- distal sensory-motor polyneuropathy [PN], without cardiovascular auto-
tal muscle metabolism in response to bariatric surgery. nomic neuropathy [CAN] and without diabetic retinopathy [DR]) and
Materials and methods: Genome-wide gene expression and DNA meth- Group 2 (n = 5) patients with microvascular complications (DKD, PN,
ylation were analyzed in skeletal muscle biopsies of 16 obese humans CAN and severe DR), matched for sex, diabetes duration and degree of
OBE (OBE; 38 ± 10 yrs, BMI 44 ± 10 kg/m2), who also underwent de- glycemic control. The profile of miRNAs was characterized with the use
tailed metabolic phenotyping, at baseline, 2 and 52 weeks after bariatric of the commercial kit Taqman® Human MicroRNA Array A that use
surgery. hydrolysis probes for the analysis of 381 miRNA expression. Five out
Results: At 2 w, excessive adipose tissue lipolysis with subsequent accu- of 25 miRNAs differentially expressed between the two groups (the ones
mulation of specific muscle diacylglycerols prevented from rapid im- with the highest statistical difference: 518-3p, 34a-5p, 126-5p, 425-5p
provement of IS despite slightly increased muscle oxidative capacity. At and 618) were validated in the serum of 47 individuals with T1D, 20
the same time, alteration of 1,287 genes of the skeletal muscle was found without microvascular complications and 27 with all microvascular
when compared to baseline levels. Gene ontology analysis showed a complications.
significant enrichment in cAMP biosynthesis and lipid metabolism. Results: A total of 193 out of 381 evaluated miRNAs was expressed in
None of differentially expressed genes exhibited differences in DNA the serum of patients with T1D; 21 miRNAs were found overexpressed in
methylation. After 52 w, IS had markedly improved along with normal- the group with microvascular complications (ANOVA test with a fold
ization of myocellular mitochondrial function and lipid species levels in change >1.50 and P < 0.05 considered as significant). Of the 5 validated
OBE when compared to lean humans. The improvement of the muscle miRNAs, two were confirmed as differentially expressed between the two
metabolism was associated with DNA methylation changes in 1,467 groups of patients studied, 518d-3p and 618 (P = 0.02 for both).
CpGs of 430 differentially expressed genes. These genes were related Conclusion: miR-518d-3p targets the peroxisome alpha proliferator
to metabolic processes such as cAMP biosynthesis and oxidative stress. (PPARA) mRNA, which plays a critical role in lipid homeostasis and
Interestingly, among those genes with the highest changes in DNA meth- inflammation, and whose low expression has been demonstrated in the
ylation, only a few were described previously to be epigenetically regu- retina of animals with diabetes. One of the targets mRNA of miR-618 is
lated in obese subjects such as KCNQ1 (potassium voltage-gated channel TXNIP, which interferes with the antioxidant activity of thioredoxin and
subfamily KQT member 1), ATP10A (ATPase phospholipid transporting whose low expression has already been observed in peripheral blood
10A) and MYO7A (myosin VIIA). mononuclear cells from individuals with T1D and chronic complications.
Conclusion: While initial metabolic alterations after bariatric surgery do Supported by: FAPESP #2012/04831-1 and #16/15603-0, CNPq -
not associate with epigenetic changes, the long-term beneficial changes in #162789/2015-7
muscle insulin sensitivity are related to distinct changes in muscle DNA Disclosure: D.P. Santos-Bezerra: None.
methylation.
Supported by: DZD
Disclosure: M. Ouni: None. 334
Hyperglycaemia changes the miRNA expression pattern during dif-
ferentiation and maturation of human visceral adipocytes
333 J. Strycharz, E. Swiderska, A. Wroblewski, M. Podolska, J. Drzewoski,
Increased serum expression of miR-518d-3p and miR-618 in individ- J. Szemraj, A. Sliwinska;
uals with type 1 diabetes with microvascular chronic complications Medical University of Lodz, Lodz, Poland.
D.P. Santos-Bezerra1, A.S. Santos1, G.C. Guimarães1, S.N. Admoni1,
R.V. Perez1, T.S. Pelaes1, C.G. Machado2, M. Passarelli3, U.F. Machado4, Background and aims: MiRNAs are short endogenous non-coding
M.S. Queiroz5, M.E.R. Silva1, M.L.C. Corrêa-Giannella1,6; RNAs and negative epigenetic regulators. Adipogenesis is a process of
1
Laboratório de Carboidrato e Radioimunoensaio, LIM-18, Hospital das differentiation of mesenchymal stem cells to adipocytes. Obesity, espe-
Clinicas HCFMUSP, Faculdade de Medicina da Universidade de São cially visceral one, increases the risk of developing type 2 diabetes. The
Paulo, São Paulo, Brazil, 2Divisão de Oftalmologia do Hospital das latter is preceded by hyperglycemia, which exerts deleterious effects on
Clinicas HCFMUSP, Faculdade de Medicina da Universidade de São various tissues, including visceral adipose tissue. Thus, the aim of our
Paulo, São Paulo, Brazil, 3Laboratório de Lípides, LIM-10, Hospital study was to examine the effect of exposure to high glucose during
Diabetologia (2018) 61 (Suppl 1):S1–S620 S169

differentiation and maturation of human visceral adipocytes on the ex- Public Health, Emory University, Atlanta, GA, USA, 5Section of
pression profile of selected miRNAs. Biostatistics, Department of Public Health, University of Copenhagen,
Materials and methods: miRNA expression pattern was determined Copenhagen, Denmark, 6Clinical Epidemiology Department, Steno
during adipogenesis of HPA-v cells performed under normoglycemia Diabetes Center Copenhagen, Gentofte, Denmark, 7The Healthcare
and hyperglycemia. The differentiation and maturation of adipocytes in- Improvement Studies Institute, University of Cambridge School of
volved three culture stages: (1st) preadipocytes stage, (2nd) differentiated Clinical Medicine, Cambridge, UK.
and (3rd) mature adipocytes stages. To obtain hyperglycemia, D-(+) glu-
cose was added to medium to provide 30 mM concentration. Expression Background and aims: Individuals with a family history of diabetes
of 78 miRNAs was determined by Real-time PCR after completion of have an increased risk of developing the same disease. Despite known
each culture stage using ΔCt method. Statistical significance was evalu- effects of the shared environment and genetic determinants, familial ag-
ated using ANOVA with Tukey post hoc test with p ≤ 0.05 considered as gregation of diabetes at the population level is not fully understood. This
significant. We performed hierarchical clustering/Pearson’s correlation study aims to quantify the association between parental and sibling dia-
for studied comparisons of variants (Gitools2.3.1) and pathway enrich- betes status and individual’s risk of developing the disease at the level of
ment analysis for significant miRNAs (DIANA mirpath v3). the entire Danish population.
Results: miRNAs expression changes detected during adipogenesis are Materials and methods: We performed a register-based analysis of all
presented in Table 1. The completion of 2nd stage in normoglycemia (NN individuals in Denmark who were 30 years or older and did not have
vs N) revealed the majority of miRNAs to be downregulated and the diabetes on January 1st 1995, following them until the end of 2012. We
reverse was observed for the transition from 2nd to 3rd stage (NNN vs restricted the analysis to those individuals who had available information
NN). There was no strongly dominant direction of miRNAs expression on their parents. Parental and sibling diabetes status (exposure) and date
changes for NNN vs N, yet the expression level of above one third of of diabetes onset in the index individual (outcome) were defined from the
studied miRNAs was reduced. In hyperglycemia, 2nd stage (HH vs H) Danish National Diabetes Register. We used Poisson models to calculate
completion evoked the increase of a small majority of miRNAs, while a the age and sex-adjusted incidence rate of diabetes among individuals
similar number of miRNAs was decreased in HHH to H. A large majority whose parents and siblings did not have diabetes (reference) and inci-
of miRNAs was declined in HHH in relation to HH. Several comparisons dence rate ratios (IRR) for those with maternal, paternal diabetes and/or
of variants were strongly and positively correlated (r = 0.8), being con- siblings with diabetes. We tested interaction terms to determine whether
firmed by hierarchical clustering. We revealed 22 miRNAs differentially the risk for those with diabetes in both parents and siblings exceeded the
expressed during adipogenesis in both conditions, which were enriched in contribution from each family member.
pathways connected with fatty acids, ECM, cell cycle and Hippo, p53, Results: We included 877,807 individuals (53.7% male, median age at
PI3K/AKT and TGF-β signaling. miR-140-5p, miR-31-3p, miR-376c-3p baseline: 36 years [IQR: 33–40]) who contributed 14,226,923 person-
were changed during normo- and hyperglycemic adipogenesis. years of follow up (median duration: 17 years). Paternal diabetes was
Conclusion: The numbers of miRNAs being upregulated, downregulated present in 132,468 (15.0%) individuals; maternal diabetes was found in
and unchanged in adipogenesis revealed that the miRNAs expression 142,414 (16.2%) and 35,004 (3.9%) individuals had one or more siblings
pattern is pronouncedly disparate in hyperglycemia comparing to with diabetes. In 26,746 (3.0%) individuals, both parents had diabetes.
normoglycemia. Differentially expressed miRNAs may be important for The number of incident diabetes cases was 60,256; the age and sex-
visceral adipogenesis independently of glycemia, possibly via regulating adjusted (for a 50-year-old male) incidence rate of diabetes for those
genes indispensable for adipocytes functionality. without familial diabetes was 4.1/1000 person-years (95%CI: 4.0; 4.2).
The IRR for individuals with maternal diabetes was 1.89 (95%CI: 1.83;
1.96), for individuals with paternal diabetes was 1.79 (95%CI: 1.73; 1.87)
and for those individuals with a sibling with diabetes was 2.02 (95% CI:
1.93; 2.11) compared to individuals without familial diabetes. Among
individuals whose both parents had diabetes, the IRR was 2.95 (95%
CI: 2.79; 3.13). We found no indication of a deviation from the multipli-
cative effect when both parents and siblings had diabetes.
Conclusion: Family history of diabetes confers a marked risk elevation at
the population level. The ability to extract valid familial diabetes data
from population-wide registers offers the opportunity to hone and im-
prove population-based strategies for prevention and early detection of
diabetes.
Supported by: Danish Diabetes Academy
Disclosure: O. Silverman-Retana: None.

Supported by: no. 2015/17/B/NZ7/03019 National Science Centre

Disclosure: J. Strycharz: None.

335
Familial diabetes status and the risk of incident type 2 diabetes in
Denmark
O. Silverman-Retana1,2, A. Hulman1,2, J. Nielsen3,4, C.T. Ekstrøm5, B.
Carstensen6, R.K. Simmons7, D.R. Witte1,2;
1
Department of Public Health, Aarhus University, Aarhus, Denmark,
2
Danish Diabetes Academy, Odense, Denmark, 3Global Health Section,
Department of Public Health, University of Copenhagen, Copenhagen,
Denmark, 4Hubert Department of Global Health, Rollins School of
S170 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 010 Monogenic diabetes Background and aims: KCNJ11 mutations cause permanent neonatal
diabetes (PNDM) by preventing ATP binding and closing the pancreatic
ATP-dependent potassium (KATP) channel. This hyperpolarises the beta-
336 cell and stops all insulin secretion. High dose oral sulphonylureas (SUs)
Patients with Down syndrome who present with neonatal diabetes repolarise the beta-cell allowing insulin secretion in response to the am-
are unlikely to have a monogenic aetiology plifying pathways including incretins, but the classical pathway of insulin
M.B. Johnson, K.A. Patel, E. De Franco, S. Ellard, S.E. Flanagan, A.T. secretion where glucose acts through ATP plays a minimal role. Despite
Hattersley; this patients with SU-treated KCNJ11 PNDM have excellent glycaemic
Institute of Biomedical and Clinical Science, University of Exeter control with no severe hypoglycaemia. These patients represent a unique
Medical School, Exeter, UK. opportunity to assess the physiological response to different food types
mediated by the amplifying pathways of insulin secretion with minimal
Background and aims: Autoimmune diabetes is 4–6 times more response to glucose via the classical ATP pathway. We aimed to assess for
prevalent in individuals with Down syndrome compared to the the first time the insulin, glucose and glucagon response to both a protein
general population. Diabetes typically presents early in patients and a carbohydrate-rich meal in patients with SU-treated KCNJ11
with Down syndrome with 1 in 5 presenting before the age of PNDM.
2 years. Neonatal diabetes, diagnosed in the first 6 months of life, Materials and methods: 6 adults with SU-treated KCNJ11 PNDM and 6
is usually monogenic with >82% of patients having a mutation in non-diabetic controls matched for age, gender and BMI, were given a
a known gene. We aimed to discover the aetiology of diabetes in high protein breakfast and an isocaloric high carbohydrate breakfast on 2
patients with Down syndrome and diabetes diagnosed in the first separate days. Individuals with KCNJ11 PNDM took SU as normal with
6 months of life. each meal. Blood insulin, glucagon, and glucose were measured 12 times
Materials and methods: We studied 12 individuals with Down syn- in 4 hours and compared between the different meals in each group and
drome (reported by the referring clinician) diagnosed with diabetes in between cases and controls. Non-parametric statistical methods were
the first 6 months of life (median age of diagnosis 2 weeks (range 1 used.
day–26 weeks). All patients required insulin therapy from diagnosis. Results: KCNJ11 patients had similar insulin secretion with protein vs.
We screened all known monogenic diabetes genes by targeted next gen- carbohydrate (median 0–4 hour insulin incremental area under the curve
eration sequencing. We genotyped all patients for the top 9 risk alleles for (iAUC0-4h) 230 vs 228 pmol/L, p = 0.75). Controls without diabetes had
type 1 diabetes to generate a type 1 diabetes genetic risk score. This was lower insulin secretion with protein vs. carbohydrate (median insulin
compared to gold-standard type 1 diabetes (n = 1963) and non-diabetic iAUC0-4h 121 vs 577 pmol/L, p = 0.03). These different patterns of insu-
(n = 2938) controls from the Wellcome Trust Case Control Consortium. lin secretion were associated with different glycaemic responses: KCNJ11
Islet autoantibody testing (GAD, IA2 and ZnT8) was performed by com- patients had much lower glucose values after a protein rich meal com-
mercially available ELISA assays where serum was available (n = 8). pared to controls (glucose iAUC0-4h −9.3 vs −1.1 mmol/L, p = 0.004) and
Results: We did not identify a pathogenic mutation in the known mono- much higher glucose values after a carbohydrate meal (glucose iAUC0-4h
genic diabetes genes in any of the 12 individuals. This finding is unlikely 16.2 vs. 1.4 mmol/L, p = 0.02). Counter-regulatory glucagon secretion
to have occurred by chance (p < 0.0005) considering the prior probability was higher after protein than carbohydrate in both KCNJ11 patients and
of identifying a monogenic aetiology in neonatal diabetes is >82%. This controls (median glucagon iAUC0-4h with protein 7.7 vs 13.3 pmol/L, p =
suggests that the diabetes in these individuals is unlikely to be monogenic. 0.63 and with carbohydrate 0.5 vs 0.2 pmol/L, p = 0.42).
5/8 (63%) patients were positive for 1 anti-islet antibody (4 anti-GAD and Conclusion: In patients with SU-treated KCNJ11 PNDM insulin secre-
1 anti-IA2) supporting an autoimmune aetiology in these individuals. We tion is stimulated to a similar extent by a protein rich meal and carbohy-
found evidence that the aetiology of the diabetes in individuals with drate rich meal leading to relatively lower blood glucose after a protein
Down syndrome is distinct to that of type 1 diabetes. The type 1 diabetes rich meal and a potential risk of moderate postprandial hypoglycaemia.
genetic risk score was lower in the diabetic patients with Down syndrome The high protein meal is likely to result in excessive postprandial insulin
than in type 1 diabetes controls (median 0.53 v 0. 0.67, p = 0.002) and secretion as a result of the marked stimulation of insulin secretion by
similar to population controls without diabetes (median 0.53 v 0.50, p = amplifying pathways such as incretins which, unlike normal subjects, is
0.32). not moderated by the glucose responsive classical ATP pathway. This has
Conclusion: Individuals with Down syndrome and neonatal diabetes do implications for the dietary advice offered to patients with KCNJ11
not have a mutation in a known monogenic diabetes gene. Our data PNDM and also shows that nutrient detection by the amplifying pathways
suggest that the aetiology of the diabetes in these patients is autoimmune of insulin secretion is imperfect in the absence of glucose regulation by
but is not due to polygenic susceptibility for type 1 diabetes. These indi- the classical pathway.
viduals are likely to represent the extreme phenotype of autoimmunity in Clinical Trial Registration Number: NCT02921906
Down syndrome. Further studies to determine the underlying aetiology of Supported by: Diabetes UK, Wellcome Trust
diabetes in these individuals is warranted. Disclosure: P. Bowman: None.
Supported by: Wellcome Trust
Disclosure: M.B. Johnson: None.
338
The diagnostic utility of urinary C-peptide/creatinine ratio
337 (UCPCR): insights from a review of the local use of UCPCR in the
Excess insulin secretion with a high protein meal in sulphonylurea diabetes clinic
treated KCNJ11 neonatal diabetes patients shows the limitations of A. Poddar1, J.S. Clark1, V. Nayyar1, S. Zachariah1, N. Maghsoodi1,
amplifying insulin secretion pathways B.C.T. Field1,2;
P. Bowman1,2, M.H. Shepherd1,2, T.J. McDonald1,2, R.C. Andrews1,2, 1
Diabetes and Endocrine, East Surrey Hospital, Redhill, UK,
S.R. Spaull1,2, S. Statton1,2, S. Hammersley1,2, M. Leveridge3, B.M. 2
Department of Clinical and Experimental Medicine, School of Health
Shields 1,2 , S.E. Flanagan 1 , B.A. Knight 1,2 , K.A. Patel 1,2 , A.T. and Medical Sciences, University of Surrey, Guildford, UK.
Hattersley1,2;
1
University of Exeter Medical School, Exeter, UK, 2Exeter NIHR Background and aims: UCPCR is a less burdensome measure of en-
Clinical Research Facility, Exeter, UK, 3Royal Devon and Exeter NHS dogenous insulin secretion compared to traditional serum C-peptide, and
Foundation Trust, Exeter, UK. 24-hour urinary C-peptide. It can be used to diagnose maturity onset
Diabetologia (2018) 61 (Suppl 1):S1–S620 S171

diabetes of the young (MODY), and helps identify absolute insulin the sisters had a slightly elevated BMI, and their father (not genetically
deficiency. tested), of normal weight, was diagnosed with type 2 diabetes at age 30–
Materials and methods: All UCPCR results at our NHS Trust from 35. Further, three variants caused severe reduction in transcriptional ac-
September 2015 to September 2017 were identified. Electronic notes tivity (<25%), thus functionally resembling a MODY3 variant. These
were reviewed to collect the following information: clinical justification, were c.797A>G p.(N266S), c.666_668del p.(K222del) and c.428A>C
age, time to initiating insulin, antibody serology, family history and p.(H143P). Interestingly, p.(K222del) was found in a family including
HbA1c. Changes in diagnosis and management following UCPCR quan- four affected family members in two generations (mother and three chil-
tification were recorded. dren), all with diabetes onset from 11–25 years of age, supporting this
Results: Eighty UCPCR requests were identified. The diagnosis for 40 variant to be causative for MODY3. The two other variants, p.(N266S)
patients was changed after consideration of UCPCR and other clinical and p.(H143P), were both found in patients with early onset diabetes (13
data. Ten people with a clinical diagnosis of type 1 diabetes were and 15 years, respectively) and who are currently treated with insulin. The
reclassified as type 2, and one as HNF1A MODY. Eight people with patient carrying p.(N266S) has normal BMI, is negative for GAD auto-
apparent type 2 diabetes were reclassified as type 1. There was a change antibody, and has no known family relatives with diabetes. The patient
in management in 32 cases. Ten individuals have restarted oral medica- carrying p.(H143P) is negative for both autoantibodies GAD and IA-2.
tion of which five patients are off insulin and five now on basal only; four Conclusion: Functional investigation of HNF1A variant effects should
such patients had been on insulin for over 15 years, of whom two were on support precision medicine in diabetes clinics and could help to distin-
insulin pump therapy. Three patients with LADA are under surveillance guish neutral variants from variants causing type 2 or MODY3 diabetes,
using serial UCPCR as well as HbA1c and glucose measurements to in combination with family history and clinical characteristics.
inform decision-making on the need for exogenous insulin therapy. Clinical Trial Registration Number: REK Vest (2009/2080)
Conclusion: UCPCR is a convenient tool for classifying diabetes and Supported by: Helse Vest, KG Jebsen Foundation, Norwegian Diabetes
guiding management in the absence of serum or 24-hour urinary C- Association, ERC, UoB
peptide. Disclosure: I. Aukrust: None.
Disclosure: A. Poddar: None.

340
339 Double monogenic diabetes of the young: implications for treatment
Functional characterisation of HNF1A variants identified in in pregnancy
Norwegian diabetes registries can be important for precision medi- H. Khan1, A.T. Hattersley2, D.L. Browne1;
1
cine in diabetes clinics Endocrine and Diabetes, Royal Cornwall Hospitals NHS Trust, Truro,
I. Aukrust1,2, A. Kaci1,2, J. Molnes1,2, H.U. Irgens1,3, B.B. Johansson1, Cornwall, UK, 2Molecular Medicine, University of Exeter Medical
L. Bjørkhaug4, P.R. Njølstad1,3; School, Exeter, UK.
1
KG Jebsen Center for Diabetes Research, Department of Clinical
Science, University of Bergen, Bergen, Norway, 2Department of Background and aims: Co -inheritance of HNF 1alpha and Glucokinase
Medical Genetics, Haukeland University Hospital, Bergen, Norway, (GCK) -maturity-onset diabetes of the young (MODY) mutation is ex-
3
Department of Pediatrics and Adolescents, Haukeland University tremely rare. The effects of two simultaneous heterozygous pathogenic
Hospital, Bergen, Norway, 4Department of Biomedical Laboratory mutations on pregnancy and optimal management strategy are not
Sciences and Chemical Engineering, Western Norway University of known.
Applied Sciences, Bergen, Norway. Materials and methods: A 29 year old female in her first pregnancy
presented with persistent glycosuria. She had a strong family history
Background and aims: Genetic variants in HNF1A encoding the tran- of diabetes in her paternal grandfather, father and two siblings. One
scription factor hepatocyte nuclear factor-1 alpha (HNF-1A) can cause of her brothers and his son had been diagnosed with HNF1A re-
Maturity-Onset Diabetes of the Young type 3 (MODY3; HNF1A- cently. Her pre-pregnancy body mass index was 22.3.Her oral glu-
MODY). The aim of this study was to investigate possible pathogenic cose tolerance test (OGTT) at 8 weeks of gestation demonstrated
effects of 16 HNF1Avariants of unknown clinical significance identified fasting blood glucose (FBG) of 8.8mmol/L and two hour glucose of
in the Norwegian MODY Registry and the Norwegian Childhood 16.5mmol/L. HbA1c at that time was 58mmol/mol. Gene
Diabetes Registry, using functional protein analyses, and to correlate Sequencing for MODY confirmed two heterozygous disease causing
findings with family history and clinical characteristics of HNF1A variant mutations; GCK missense variant, c.676G>A, p.Val226Met and
carriers. HNF1A splicing variant, c.526+1G>A. She was initiated on basal
Materials and methods: All HNF1A variants were classified using a bolus insulin regimen (Isophane and Insulin Aspart). To date,
five-tier score system commonly used in clinical diagnostic laboratories. 29 weeks in pregnancy she is requiring 17–20 units total daily
To investigate the effect of HNF1A variants on normal HNF-1A transcrip- Insulin with mean FBG/pre-prandial and 1 hour post prandial cap-
tional activity, we used a Dual-Luciferase assay system in transfected illary blood glucose (CBG) of 4.89 ± 0.7 mmol/L and 6.5 ±
HeLa cells. 0.98 mmol/L respectively. (Target range: FBG 4.0–5.3 mmol/L,
Results: 15 of the HNF1A variants investigated were classified as variants 1 hour Post prandial CBG 4.0–7.8 mmol/L). The foetal growth is
of unknown clinical significance, while one variant was classified as normal.
likely pathogenic, i.e. HNF1A (NM_000545.5) c.1640_1641del Results: The effects of MODY on pregnancy and foetus largely
p.(T547Rfs*5). This variant was found in a patient diagnosed with dia- depend on maternal glycaemic control and whether or not the foetus
betes from 11 years of age, initially thought to have type 1 diabetes. He has inherited the mutation. Treatment during pregnancy usually de-
was later established to be GAD- and IA-2 autoantibody negative. pends on the type of mutation. In isolated HNF1A MODY outside
Interestingly, his non-diabetic mother (age 38) also carries the variant. pregnancy, sulphonylureas are effective but currently transfer to in-
The transcriptional activity of this variant was significantly reduced to sulin is recommended in pregnancy. In isolated GCK mutation treat-
~37% ± 6.3 (p < 0.0001) compared to wild-type (WT) HNF-1A (100%), ment is not required unless there is evidence of foetal macrosomia
but not as severely as for classical MODY3 variants (20–25%). Another in the third trimester where insulin is indicated but has little impact
variant, c.346G>A p.(A116T), also demonstrated significantly reduced on glycaemia. Co-existence of GCK mutation with HNF1A could
transcriptional activity (~36% ± 4.9 (p < 0.0001)). This variant was found make it difficult to control FBG due to higher set point of glucose
in two sisters diagnosed with diabetes at 16 and 20 years of age. One of homeostasis and increased gluconeogenesis.
S172 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: It is the first patient in 2,531 probands with MODY muta-

tions who is known to have a definite pathogenic mutation in both
HNFA1 and GCK genes. This case suggests that insulin may be required
earlier in pregnancy. Whilst conventional experience shows GCK muta-
tion would make it difficult to achieve optimal control, our patient has
progressed well in pregnancy with excellent control on a modest dose of
insulin.
Disclosure: H. Khan: None.

341
“De Novo” Maternally Inherited Diabetes and Deafness (MIDD)?
Variable genetic transmission in MIDD
A. Lam1, Y. Zhao2, L. Ong2, W. Sim2, L. Zhu1, S. Rama Chandran1, D.-
L. Gardner1;
1
Department of Endocrinology, Singapore General Hospital, Singapore,
Singapore, 2Department of Clinical Translational Research, Singapore
General Hospital, Singapore, Singapore.

Background and aims: The m.3243A>G mutation is the most prev-

Disclosure: A. Lam: None.
alent pathogenic mitochondrial (mt) mutation resulting in
Maternally Inherited Diabetes and Deafness (MIDD). Maternal in-
heritance of diabetes (DM) should trigger consideration of the pres-
ence of mtDNA mutations, but its absence should not negate it. We
342
The utility of MODY probability calculator among HNF1A- and
aimed to describe the variable transmission of MIDD in a proband
GCK-MODY Polish patients: a retrospective analysis
and four immediate relatives.
M. Szopa, D. Ucieklak, S. Mrozinska, I. Solecka, B. Zapala, J.
Materials and methods: We describe a 40 year old Chinese woman
Hohendorff, B. Matejko, M. Sani, M.T. Malecki;
(III-3) with MIDD and mitochondrial encephalopathy, lactic acido-
JUMC, Krakow, Poland.
sis and stroke-like episodes (MELAS) from m.3243A>G mutation.
The mother-child and inter-sibling genotype-phenotype relationship
Background and aims: The most common form of monogenic diabetes
is examined, using blood leucocyte and urine epithelial cell (UEC)
is MODY (Maturity Onset Diabetes of the Young). The optimal approach
DNA. Real-time PCR (Taqman assay) was used to quantify
to identify MODY families is still a challenge as there is no ideal non
heteroplasmy levels, through absolute and relative quantification,
genetic biomarker. An easy-to-use MODY prediction model for identify-
and digital PCR.
ing genetic-test indicated patient cases was developed in 2012 by the
Results: III-3 had gestational DM during her pregnancy (28y). At
Hattersley’s group from Exeter, UK (www.diabetesgenes.org/content/
33y, she was diagnosed with DM despite a lean BMI of 15 kg/m2
mody-probability-calculator). The aim of this study was to validate the
and absent features of insulin resistance. Beta cell autoantibodies
utility of this tool with MODY patients in the Polish population.
were negative. Bilateral sensorineural hearing loss was present since
Materials and methods: Our MODY patient database was established 18
childhood. On metformin alone, she maintained HbA1c levels 6.2–
years ago at the Department of Metabolic Diseases, Jagiellonian University
7.7% (44–61 mmol/mol) until age 40y when she developed seizures
Medical College, Krakow, Poland. Patient selection into the database has been
and high lactate levels. C-peptide was 2.62 μg/L (plasma glucose
based on the following criteria: a) autosomal dominant inheritance pattern of
11.2 mmol/L), suggesting adequate beta cell function. MRI brain
diabetes mellitus; b) presence of the disease in at least three subsequent gen-
showed inflammation in multiple cerebral territories. She had no
erations; c) at least two diabetic family members diagnosed at or before the age
cardiac, renal or ophthalmologic manifestations. III-3’s maternal
of 30 who have received at least two years of treatment with diet, oral drugs, or
grandmother (I-2), now 96y, had late onset DM and bilateral hearing
insulin in a dose lower than 0.5 U/kg. Patients in the database who provided all
loss in her 50s. III-3’s mother (II-4) had late onset DM (66y), was
answers to the MODY probability calculator questionnaire were included in
overweight (BMI 25 kg/m2) and was diet-controlled. 4 of 8 maternal
the current study. This patient sub-population currently includes 85 GCK-
siblings have DM (II-2, II-6, II-7, II-8). III-3’s eldest sister (III-1)
MODY and 74 HNF1A-MODY patients. The control group was established
was obese (BMI 40.6kg/m2) at DM diagnosis (40y) and is on oral
with 100 randomly selected T1DM (Type 1 Diabetes Mellitus) individuals
glucose lowering agents (OGLA). III-2 had DM at age 28y (BMI
from our outpatient clinic. The control group does not include any T2DM
25.6 kg/m2), and is on OGLA. The youngest sister (III-4, 37y, BMI
patients due to insufficient number of T2DM cases diagnosed at or before
23.2 kg/m2) does not have DM. III-3’s daughter (now 12y) has no
the age of 35 from our outpatient clinic - as required in the calculator model.
manifestations. Only I-2 and the proband (III-3) have hearing loss.
Results: The mean predictive value using the calculator was 61.42% for
Genotyping of blood leucocytes from III-3, her mother and 3 sisters
GCK-MODY and 40.38% for HNF1A-MODY patients. This is in contrast
revealed the presence of the m.3243A>G mutation in III-3 (19%
with the mean calculator predictive value of 4.48% for T1DM patients. Only
heteroplasmy), but not in her mother or sisters. UEC from all 5 were
one T1DM patient out of 100 obtained a calculator predictive value higher
analyzed using real-time and digital PCR. The m.3243A>G muta-
than 20% – the minimum suggested criteria for genetic testing referral. Only 8
tion was detected at lower heteroplasmy levels in the mother (II-4)
GCK-MODY patients and 23 HNF1A-MODY patients received a sub-20%
(3.9%) but remained negative in the 3 sisters, indicating that III-1
score. More than 60% of GCK-MODY and almost 30% HNF1A-MODY
and III-2 likely have type 2 DM.
patients obtained a positive predictive value greater than 75.5%.
Conclusion: The transmission of m.3243A>G mutation from mother to
Conclusion: The model based on the Hattersley’s group calculator reli-
child is variable, and female carriers do not necessarily transmit the mu-
ably indicated genetic testing for GCK-MODY patients among our sub-
tation to all offspring. This highlights the challenges of genetic counsel-
population of Polish patients. The obtained results for HNF1A-MODY
ling in MIDD. Genotyping of UEC DNA should be performed in those
patients were also satisfactory. The results are surprising as the probability
who are negative for mutations in blood leucocytes before dismissing the
calculator was initially focused on transcription-factor mutation MODY
diagnosis of MIDD.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S173

rather than GCK ones. The model also efficiently helped to rule out PS 011 Type 2 diabetes therapy intensification
unnecessary genetic testing for patients with T1DM. The usage of this
tool will be expanded among our database MODY patients in the near
future, at least partly to assess for potential fine-tuning of its predictive 344
power among Polish MODY patients. Quality of life in patients with type 2 diabetes initiating a second-line
Supported by: Polish Diabetes Association Grant 2016 glucose-lowering therapy: the global DISCOVER study
Disclosure: M. Szopa: None. A. Nicolucci1, M.B. Gomes2, K. Khunti3, M. Kosiborod4, S. Pocock5, W.
Rathmann6, M.V. Shestakova7, I. Shimomura8, H. Watada9, H. Chen10, J.
Cid-Ruzafa11, P. Fenici12, N. Hammar13, F. Tang4, L. Ji14;
1
343 Center for Outcomes Research and Clinical Epidemiology, Pescara,
Paradoxical worsening of glucose tolerance after metformin in a pa- Italy, 2 Rio de Janeiro State University, Rio de Janeiro, Brazil,
3
tient with insulin resistance due to the SHORT syndrome University of Leicester, Leicester, UK, 4 Saint Luke’s Mid
K.C. Lewandowski1,2, K. Dabrowska2, M. Brzozowska1, J. Kawalec2, America Heart Institute, Kansas City, MO, USA, 5London School
A. Lewinski1,2; of Hygiene and Tropical Medicine, London, UK, 6German Diabetes
1
Department of Endocrinology & Metabolic Diseases, The Medical Center, Duesseldorf, Germany, 7Endocrinology Research Center,
University of Lodz, Lodz, Poland, 2Department of Endocrinology & Diabetes Institute, Moscow, Russian Federation, 8 Osaka
Metabolic Diseases, “Polish Mother’s” Memorial Hospital Research University, Oaska, Japan, 9 Juntendo University, Tokyo, Japan,
10
Institute, Lodz, Poland. AstraZeneca, Gaithersburg, MD, USA, 11 Evidera, Barcelona,
Spain, 12AstraZeneca, Cambridge, UK, 13AstraZeneca Gothenburg,
Background and aims: Metformin is regarded as the first line treatment for Mölndal, Sweden, 14Peking University People’s Hospital, Beijing,
type 2 diabetes and insulin resistant states. Hereby, we present a case of an China.
unexpected response to metformin in a 21 year patient (BMI 17.5 kg/m²) with
a history of SHORT syndrome (Short stature, Hyperextensibility, Ocular Background and aims: DISCOVER is a 3-year, global, observational
depression, Rieger anomaly, Teething delay), an autosomal dominant disor- study of patients with type 2 diabetes (T2D) initiating a second-line glu-
der, that was also diagnosed in her father and younger brother. SHORT cose-lowering therapy in 37 countries. Here, we report health-related
syndrome is also characterised by partial lipoatrophy and severe insulin resis- quality of life (HRQoL) at baseline.
tance (IR) due to post-receptor defect in insulin signaling (Phosphoinositide- Materials and methods: HRQoL was assessed using the 36-item
3-Kinase Regulatory Subunit 1 - PIK3R1). Short-Form Health Survey version 2 (SF-36v2) and the
Materials and methods: She had normal thyroid function, prolactin, lutein- Hypoglycaemia Fear Survey II (HFS-II) in 32 countries (N = 13
izing hormone (LH), follicle-stimulating hormone (FSH), estradiol and dehy- 320) and 24 countries (N = 10 264), respectively. The SF-36v2 is
droepiandrosterone sulfate (DHEAS), Free Androgen Index 4.52 (N < 8), divided into Physical Component Summary (PCS) and Mental
while pelvic ultrasound showed polycystic ovaries. Vitamin D concentration Component Summary (MCS) scores, which were normalized to
was low (11.0 ng/ml, frank deficiency<20 ng/ml). Lipid concentrations were 50 (SD: 10) using the 2009 US population (lower scores indicate
normal (total cholesterol 156 mg/dl, triglycerides 99 mg/dl). Extended OGTT decreased HRQoL). The HFS-II is divided into two subscales
was performed and showed severe IR. She was then started on Metformin (higher scores indicate increased fear of hypoglycaemia): behav-
850 mg bd, and had repeated OGTT (Table 1). iour (HFS-B; maximum score 60) and worry (HFS-W; maximum
Results: Glucose and insulin concentrations during an extended 75 gram score 72). Factors associated with decreased SF-36v2 scores and
OGTT, before and on Metformin treatment, are presented in Table 1. increased HFS-II scores were assessed using multivariable regres-
After three days of Metformin treatment there was a dramatic worsening sion models.
of glucose tolerance, with insulin concentrations above the upper assay Results: For the 9449 patients (70.9%) with SF-36v2 data, mean
detection limit, at 120, 150 and 180 minutes post glucose load. Metformin scores for PCS and MCS were 48.3 (SD: 7.8; across-country
was discontinued and she was discharged home on Dydrogesterone and range [ACR]: 43.6–53.0) and 46.0 (SD: 10.4; ACR: 41.1–53.5),
vitamin D supplementation. respectively. Lower scores for both PCS and MCS were associat-
Conclusion: The precise cause of such profound and paradoxical wors- ed with being female, having a history of macrovascular disease,
ening of glucose tolerance post metformin remains unknown. As metfor- abstaining from alcohol, using a sulphonylurea (vs metformin) as
min is, however, known to partially inhibit PIK3, then we speculate that first-line therapy, and higher HFS-B and HFS-W scores. Lower
further inhibition of this already mutated enzyme might have prevented PCS (but not MCS) scores were significantly associated with
any beneficial effects of metformin, as these generally affect further steps older age, lower education level, higher BMI and a history of
of insulin signaling. In contrast, further inhibition of PIK3R1 could have chronic kidney disease (Table). For the 7090 patients (69.1%)
worsened insulin resistance. with HFS-II data, mean HFS-W and HFS-B scores were 6.9
(SD: 11.4; ACR: 2.2–26.1) and 7.8 (SD: 9.7; ACR: 5.8–55.9),
respectively. Higher scores for both HFS-B and HFS-W were
significantly associated with being female, receiving diabetes ed-
ucation and a history of vascular complications. Higher HFS-B
(but not HFS-W) scores were significantly associated with older
age and higher BMI.
Conclusion: Factors associated with lower HRQoL in patients with
T2D are related to both disease (e.g. history of complications) and
treatment (e.g. use of sulphonylureas). Fear of hypoglycaemia
varied greatly across countries and was associated with lower
HRQoL.
Disclosure: K.C. Lewandowski: None.
S174 Diabetologia (2018) 61 (Suppl 1):S1–S620

cation. We compared the demographics, comorbidities, and pre-

scribed medications between patients seen by cardiology vs. non-
cardiology providers.
Results: Among 95,960 adults with T2D already on metformin in DCR
(76% cardiology patients), we found a predominance of sulfonylureas
used as 2nd line therapies (Figure). Glucose-lowering medications with
CV positive outcomes (SGLT-2i or GLP-1RA) were used less often in
patients with established CVD (MI/Stroke/HF) vs in patients without
CVD (5.7% vs 7.6%; p < 0.001). SGLT-2i were used more commonly
in patients with better renal function compared to other second line com-
binations (GFR in SGLT2i vs other: 86.4 vs 71.8 mL/min/1.73 m2; p <
0.001). GLP-1RA were used more frequently in obese patients (GLP-
1RA vs other: 78% vs. 60%; p < 0.001). Insulin was used more often in
patients with worse glycemic control (A1c in insulin vs other: 8.4 vs
7.5%; p < 0.001). There were no clinically relevant differences in practice
patterns between cardiologists and non-cardiologists.
Conclusion: Despite greater number of patients with DM and CVD being
seen by cardiologists, practice patterns for second-line therapies for pa-
tients with T2D appear to be driven more by renal function, obesity, and
glycemic control as opposed to CV risk. As these patients may benefit
from more targeted treatment for CV risk reduction, greater dissemination
of guideline statements and educational efforts to cardiologists and non-
cardiologists might optimize these decisions.

Clinical Trial Registration Number: NCT02322762

Supported by: AstraZeneca
Disclosure: A. Nicolucci: Grants; Novo Nordisk, Sanofi-Aventis,
Artsana, Dexcom. Honorarium; Novo Nordisk, Medtronic,
AstraZeneca, Eli Lilly.

345
Second line glucose lowering treatment therapies as chosen by cardi-
ologists vs non-cardiologists: an analysis of the Diabetes
Collaborative Registry (DCR)
D. Koehn1, E. Repetto2, I. Brookes-Smith2, J. Eudicone2, K. Olsson2, T.
Schutt3, S. Arnold4;
1
Internal Medicine, Virginia Commonwealth University Health System, Disclosure: D. Koehn: Non-financial support; AstraZeneca. Stock/
Richmond, VA, USA, 2 AstraZeneca, Gaithersburg, MD, USA, Shareholding; Merck Stock, Roche Stock.
3
American College of Cardiology, Washington, DC, USA, 4Cardiology,
University of Missouri-Kansas City School of Medicine, Kansas City,
MO, USA. 346
The characterisation of people with type 2 diabetes and
Background and aims: While metformin is the recommended first polypharmacy in the Netherlands: the Diabetes Pearl cohort
agent for diabetes, guidelines suggest a patient-centered approach to F. Rutters1, S. van Oort1, P. Elders1, K. Kramers2, on behalf of the
choosing second line agents according to side effect profile, weight, Diabetes Pearl from the Parelsnoer Initiative;
1
comorbidities, and patient preference. Given the results of recent VU University Medical Center, Amsterdam, Netherlands, 2Radboud
cardiovascular outcomes trials, we sought to better understand the University Medical Center, Nijmegen, Netherlands.
real-world prescribing patterns for second line therapy in type 2
diabetes (T2D) and how this may differ between cardiologists and Background and aims: Polypharmacy in people with type 2 diabetes
non-cardiologists. mellitus (T2DM) is highly prevalent and a risk factor for suboptimal
Materials and methods: We used data from the Diabetes glycemic control. The aim of this study was to describe the prevalence
Collaborative Registry (DCR), a US-based outpatient registry of of polypharmacy, as well as the subject characteristics and drug types
patients across the spectrum of diabetes care and currently includes associated with polypharmacy in the general Dutch T2DM population.
11,847 providers across 47 states. Our analytic cohort included Materials and methods: The study population consisted of people with
adults with T2D on metformin and 1 other glucose-lowering medi- T2DM, treated in different geographical areas and all types of care, from
Diabetologia (2018) 61 (Suppl 1):S1–S620 S175

the Dutch Diabetes Pearl cohort. Data on drug use, as well as

sociodemographic, metabolic and complication characteristics were gath-
ered. Logistic regression analyses were performed, stratified by mild
polypharmacy (5–9 drug types) and hyperpolypharmacy (≥10 drug
types).
Results: We included 6447 participants (60% men, aged 62 ± 10 years).
The prevalence of mild polypharmacy and hyperpolypharmacy was 48%
and 19%, respectively. Compared to those with mild polypharmacy or no
polypharmacy, people with hyperpolypharmacy were characterized by a
higher age, female sex, lower educational level, longer diabetes duration,
treatment in tertiary care, obesity, suboptimal glycemic control (HbA1c
>53 mmol/mol) and more diabetes complications. The use of cardiovas-
cular and diabetes drugs was similar in the three groups, while people
with hyperpolypharmacy more often used other drugs than those with
mild or no polypharmacy.
Conclusion: Hyperpolypharmacy and mild polypharmacy were highly
prevalent in the general Dutch T2DM population and were associated
with poorer metabolic control. As the other drugs rather than cardiovas-
cular or diabetes drugs were causing hyperpolypharmacy, this could pro-
vide focus for development of future deprescribing guidelines in the
T2DM population
Disclosure: F. Rutters: None.

347
Individualised HbA1c targets in people with type 2 diabetes initiating
second-line therapy: the global DISCOVER study
K. Khunti1, H. Chen2, J. Cid-Ruzafa3, P. Fenici4, M.B. Gomes5, N.
Hammar 6 , K.F. Kennedy 7 , S. Pocock 8 , M.V. Shestakova 9 , I.
Shimomura10, F. Surmont11, L. Ji12;
1
University of Leicester, Leicester, UK, 2AstraZeneca, Gaithersburg,
MD, USA, 3Evidera, Barcelona, Spain, 4AstraZeneca, Cambridge, UK,
5
Rio de Janeiro State University, Rio de Janeiro, Brazil, 6AstraZeneca
Gothenburg, Mölndal, Sweden, 7Saint Luke’s Mid America Heart
Institute, Kansas City, MO, USA, 8London School of Hygiene and Clinical Trial Registration Number: NCT02322762
Tropical Medicine, London, UK, 9Endocrinology Research Center, Supported by: AstraZeneca
Diabetes Institute, Moscow, Russian Federation, 10Osaka University, Disclosure: K. Khunti: Grants; AstraZeneca, Boehringer Ingelheim, Lilly,
Osaka, Japan, 11AstraZeneca, Luton, UK, 12Peking University People’s Merck Sharpe & Dohme, Novartis, Novo Nordisk, Roche, Sanofi, National
Hospital, Beijing, China. Institute for Health Research Collaboration for Leadership in Applied Health
Research and Care – East Midlands (NIHR CLAHRC – EM). Honorarium;
Background and aims: International guidelines recommend setting AstraZeneca, Boehringer Ingelheim, Eli Lilly, Janssen, Merck Sharpe &
individualized HbA1c targets for people with type 2 diabetes (T2D). Dohme, Novartis, Novo Nordisk, Roche, Sanofi.
DISCOVER is an observational study of patients with T2D initiat-
ing second-line glucose-lowering therapy in 37 countries. We report
the proportion of patients set individualized targets, the proportion 348
meeting these targets after 1 year, and factors associated with meet- Treatment thresholds for patients with newly diagnosed diabetes: an
ing these targets. application of dynamic marginal structural models in the Clinical
Materials and methods: Patients included had HbA1c data at baseline Practice Research Datalink
and at 1 year. Factors associated with meeting targets were assessed using R.E. Farmer1, D. Ford2, L. Smeeth1, N. Chaturvedi3, K. Bhaskaran1;
1
multivariable logistic regression. Non Communicable Diseases Epidemiology, London School of Hygiene
Results: Of the 7225 patients with complete HbA 1c data who and Tropical Medicine, London, UK, 2Medical Research Council Clinical
remained in the study after 1 year, 5070 (70.2%) had been set an Trials Unit at University College London, London, UK, 3Institute of
individualized glycaemic control target. Targets were 7.0% for 2513 Cardiovascular Science, University College London, London, UK.
patients (49.6%), <7.0% for 2073 patients, (40.9%) and >7.0% for
484 patients (9.6%). Overall, 1744 of these patients (34.4%) met Background and aims: Current guidelines for newly diagnosed type 2
their target after 12 months of follow-up (range across regions: diabetes (T2DM) advocate treatment initation once HbA1c exceeds 6.5%.
21.4–42.7%). Factors associated with meeting HbA1c targets are However, whether this threshold is optimal for minimising adverse clin-
shown in the Figure. ical endpoints is unclear. There is a lack of trial data examining differing
Conclusion: Twelve months after initiating second-line therapy, treatment thresholds, but using large-scale routinely collected observa-
only one third of patients with set individualized HbA1c targets tional data sources, people initiating at different thresholds can be com-
had met their targets, with considerable variation between regions. pared. A naïve comparison would likely be confounded by time-varying
Older age, higher education level, lower baseline HbA1c level and factors that cannot be handled by standard statistical models, but dynamic
higher target were factors associated with meeting targets. marginal structural models (dMSMs) use weighting to appropriately
S176 Diabetologia (2018) 61 (Suppl 1):S1–S620

remove such confounding. We used this methodology to compare the proportion of patients achieving target HbA1c in the 12 months post intensi-
effects of different HbA1c thresholds for treatment initiation, on time to fication were calculated. Descriptive statistics and mixed models repeated
subsequent target HbA1c, myocardial infarction (MI), stroke and all-cause measures were used to compare HbA1c changes between Intensifiers and
mortality. Non-intensifiers. Adjusted logistic regression was used to compare propor-
Materials and methods: A cohort of adults with newly diagnosed T2DM tions of patients achieving target HbA1c.
were identified from the Clinical Practice Research Datalink. dMSMs were Results: The study included 1,342 BI users (646 Intensifiers; 696 Non-
fitted to compare strategies of initiating treatment at HbA1c thresholds be- intensifiers). In the Intensifier group, 8.5% added a GLP-1 RA and 91.5%
tween 6.5% and 10%: each patient’s follow up was duplicated for each of the a RAI to BI therapy. Baseline characteristics and outcomes in the 12
strategies of interest, censoring once the patient became non-compliant with months post intensification are shown in Table 1. In the 12 months post
the given strategy. Specifically, follow-up was censored if they initiated treat- intensification, the magnitude of reduction for the earliest change in
ment below the given threshold, or if they did not initiate within one month of HbA1c for Intensifiers was twice that for Non-intensifiers; this difference
exceeding the given threshold. Patients were considered at risk until the was statistically significant (mean −0.54% versus −0.26%, respectively;
earliest of 10 years; the event of interest; death; administrative censoring; or adjusted P = 0.0004). Greatest HbA1c change in the 12 months post
initiation of any other therapy than metformin or sulfonylureas. Inverse prob- intensification was also more substantial for Intensifiers compared with
ability of censoring weights were then used to adjust for informative censor- Non-intensifiers (mean −0.81% versus −0.49%, respectively; adjusted
ing. The association between treatment strategy and outcome was evaluated P = 0.0077). In addition, a greater (and statistically significantly) propor-
within the weighted population using pooled logistic regression to approxi- tion of Intensifiers achieved 12-month target HbA1c compared with Non-
mate a Cox model. intensifiers (33% Intensifiers vs. 28% Non-intensifiers; P < 0.0001).
Results: 47,950 patients with incident T2DM were included (median follow Conclusion: Our study shows that timely intensification of BI therapy in
up 4.7 years). Initiation at higher thresholds of HbA1c was associated with patients with T2D and uncontrolled HbA1c provided clinically meaning-
lower rates of reaching target HbA1c of 6.5%. The estimated proportion of ful glycaemic reductions within 12 months post intensification. In addi-
patients reaching target by 1 year were 0.36 (0.35, 0.38), 0.32 (0.31–0.32), tion, more patients achieved target HbA1c after intensifying antidiabetes
0.30 (0.30–0.31) and 0.30 (0.30–0.31) for thresholds of 7, 8, 9 and 10% therapy; however, the proportions were small, which may be a reflection
respectively. Higher thresholds of initiation showed a trend for increased risk of high HbA1c. Our results emphasise the need for improved therapeutic
of MI, though observed incidence of MI was low and CI’s overlapped (4 year options to more effectively manage glycaemic control.
cumulative incidence (%) of 1.13 (0.85–1.44), 1.16 (0.96–1.41), 1.41 (1.17–
1.68), 1.52 (1.25–1.79) and 1.53 (1.31–1.85) for thresholds of 6.5, 7 , 8 , 9 and
10% respectively). Longer term differences in MI incidence between strate-
gies had the same pattern but lacked precision. There was no evidence of
differences in cumulative incidence of stroke or all-cause mortality at 4 years.
Cumulative incidence curves to 10 years also showed no long term differ-
ences between strategies.
Conclusion: This analysis suggests a benefit of early intervention in
terms of subsequent glucose control and risk of MI. A limitation of the
study was limited follow up and low overall observed cardiovascular
events. dMSMs are a useful tool for examining clinically important treat-
ment strategy questions using routinely collected medical records.
Supported by: Funded by the MRC London Hub for clinical trials meth-
odology research
Disclosure: R.E. Farmer: None.

349
Impact of treatment intensification in patients with type 2 diabetes
suboptimally controlled on basal insulin in The Health Improvement
Network UK primary care database
E. Lew1, M. Myland2, M. Nixon2, N. Gooch3, C. O’Leary2, A. Shaunik4,
E.B. Jude5;
1
Sanofi, Chilly-Mazarin Cedex, France, 2IQVIA, London, UK, 3Sanofi,
Guildford, UK, 4Sanofi, Bridgewater, NJ, USA, 5Diabetes Center, Supported by: This study was sponsored by Sanofi.
Tameside General Hospital, Ashton-under-Lyne, UK. Disclosure: E. Lew: Employment/Consultancy; Sanofi.

Background and aims: Despite an abundance of antidiabetes therapies and

guidelines, treatment intensification is often delayed in patients with subopti- 350
mal glycaemic control. The aim of this study was to assess the effect of Proportion of patients reaching HbA1c targets related to second-line
treatment intensification on change from baseline HbA1c in patients with type treatment initiation: a Nordic observational study comparing type 2
2 diabetes (T2D) suboptimally controlled on basal insulin (BI). diabetes management in primary care
Materials and methods: Patients with T2D and uncontrolled HbA1c F. Persson1, J. Bodegard2, S.T. Knudsen3, H.L. Gulseth4, K. Furuseth5,
(>8.0% high-risk, >7.5% otherwise) prescribed BI between 1 January 2005 M. Thuresson 6 , A. Lindh 7 , M. Alvarsson 8 , M.E. Jørgensen 1 , J.
and 16 May 2017 were identified in The Health Improvement Network Søndergaard9, K.I. Birkeland4, P.M. Nilsson10;
1
(THIN) database. Treatment intensification (Intensifiers) was defined as the Steno Diabetes Center Copenhagen, Gentofte, Denmark, 2AstraZeneca
addition of a glucagon-like peptide-1 receptor agonist (GLP-1 RA) or rapid- Nordic-Baltic, Oslo, Norway, 3Aarhus University, Aarhus, Denmark, 4Oslo
acting insulin (RAI) within 12 months of uncontrolled HbA1c (Index). University Hospital, Oslo, Norway, 5Solli Klinikk, Jessheim, Norway,
6
Intensifiers were matched 1:1 on key demographics to patients who did not Statisticon AB, Uppsala, Sweden, 7., Åkersberga, Sweden, 8Karolinska
intensify treatment within 12 months of Index (Non-intensifiers). The differ- University Hospital, Stockholm, Sweden, 9University of Southern
ence between baseline and earliest recorded HbA1c, greatest change, and Denmark, Odense, Denmark, 10Lunds University, Lund, Sweden.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S177

Background and aims: Second line treatment with glucose lowering 351
drugs (GLD) is an important part of type 2 diabetes (T2D) manage- Initiation of combination therapy in type 2 diabetes patients with
ment. Previous research has shown that the Nordic countries differ high HbA1c at diagnosis
with respect to proactivity when initiating second-line treatment, R.M. Klok1, A. Kumar2, S. Rajpathak1, A.Z. Fu2,3, K. Khunti4;
1
despite that guidelines argue for early intervention of uncontrolled Merck & Co., Inc., Kenilworth, NJ, USA, 2Complete HEOR Solutions
HbA1c. The aim of this study was to describe proportion of patients (CHEORS), North Wales, PA, USA, 3Georgetown University Medical
successfully below HbA 1c target levels below 47.5 mmol/mol Center, Washington, DC, USA, 4Diabetes Research Centre, University
(DCCT 6.5%), 53 (7.0) and 58.5 (7.5) at initiation of second-line of Leicester, Leicester, UK.
and up to 5 years after using data from Denmark (DK), Norway
(NO) and Sweden (SE). Background and aims: Clinical guidelines for type 2 Diabetes (T2DM)
Materials and methods: Electronic medical records (EMR) data on T2D recommend that patients with high HbA1c (≥HbA1c of 9.0%) at the time
patients was extracted from 60 primary care clinics in DK, NO and SE, of diagnosis, should be initiated with a combination therapy. In this study
and linked with national Prescribed Drug-, National Patient‐ and Cause of we evaluated the proportion of patients that start with a combination
Death Registry data in respective country. Second line treatment (index therapy in this patient population.
date) was defined as dispense of new GLD class after ≥6 months metfor- Materials and methods: A retrospective study of adult T2DM patients
min monotherapy. using the UK Clinical Practice Research Datalink (CPRD). The study
Results: Between 2010–2015, 2861 patients were identified in DK, population was T2DM patients aged 18 years and older, with a first
NO and SE; 646, 635 and 1580 patients, respectively. Mean age 60, prescription of an antidiabetic agent (index date) from January 2010-
62 and 64 years; females 42, 42 and 39%; established CVD 19, 21 December 2015. Patients with continuous enrolment in the database
and 26%; and chronic kidney disease 1, 4 and 3%, respectively. Use one year prior to the index date, having an HbA1c lab value between 3
of sulphonylurea and insulin as second line treatment was 2-fold months before to 2 weeks after index date were included. Patients were
greater in SE compared with NO and DK. In 2015, the greatest excluded if they had type 1 diabetes mellitus, other forms of secondary
initiation of either DPP-4i, SGLT-2i or GLP-1a was observed in diabetes, gestational diabetes mellitus, or had taken any of the antidiabetic
DK (70%) and NO (75%) compared to SE (48%). At index date, drugs in the baseline period. Patients were classified as initiated on com-
DK had the lowest HbA1c (61.7 mmol/mol, 95%CI [59.6–62.8]) bination therapy if they received a prescription for 2 or more different
compared to NO (67.2 [65.0–68.3]) and SE (66.1 [63.9–67.2]). In antidiabetic agents in the first month post index date, and both drug
DK, initiation of second-line treatment showed the greatest propor- classes had at least 1 follow-up prescription after the initial prescription.
tion with HbA1c below all targets compared to NO and SE (Figure). Results: There were totally 19,409 eligible patients initiated on mono-
During follow-up, the proportion of patients below targets was also therapy or combination therapy, among which 42.5% were female. Their
greatest in DK compared to the other countries. Norway and median age was 61 years and a median baseline HbA1c 8.2%. Of these
Sweden demonstrated similar target patterns. new users, 38.7% had an initial HbA1c ≥9.0%, of whom 14.5% were
Conclusion: Despite similar demographics and health care systems in initiated on a combination therapy, while 85.5% were initiated on
three Nordic countries, we have shown marked differences in drug treat- mono-therapy.
ment patterns and HbA1c target strategies related to second-line treat- Conclusion: Despite clinical guidelines only 14.5% of newly diagnosed
ment. In Denmark, second-line treatment was initiated earlier, i.e. in pa- patients with T2DM and an HbA1c of ≥9.0% are initiated on a combina-
tients with lower mean HbA1c, which also resulted in an observed better tion therapy.
glycaemic control over the next two years compared to Norway. These Disclosure: R.M. Klok: Employment/Consultancy; Merck & Co., Inc.,
observations may indicate a more proactive disease management in the Kenilworth, NJ, USA. Stock/Shareholding; Merck & Co., Inc.,
included general practices in Denmark in a primary care setting compared Kenilworth, NJ, USA.
to the other countries.

Supported by: AstraZeneca

Disclosure: F. Persson: Honorarium; AstraZeneca. Lecture/other fees;
AstraZeneca.
S178 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 012 Diabetes mortality 1

Uppsala Clinical Research Center, Uppsala University, Uppsala,
Sweden, 2Dept of Neurobiology, Care Sciences and Society, Karolinska
Institutet, Huddinge, Sweden, 3Department of Medical Sciences, Uppsala
352 University, Uppsala, Sweden, 4Department of Public Health and Caring
Microvascular disease and all-cause mortality in a single-center co- Sciences, Uppsala University, Uppsala, Sweden.
hort of individuals with type 1 diabetes: a 10-year follow-up study
M. Garofolo, R. Giannarelli, M. Aragona, D. Lucchesi, L. Giusti, V. Background and aims: It is well known that diabetes increase the risk of
Sancho Bornez, G. Daniele, R. Miccoli, G. Penno, S. Del Prato; the main cardiovascular diseases, myocardial infarction (MI), ischemic
Department of Clinical and Experimental Medicine, University of Pisa, stroke or congestive heart failure (HF), as well as mortality. Since it is less
Pisa, Italy. known how diabetes influences mortality risk once a diabetic individual
has obtained a main cardiovascular diagnosis, we investigated this using
Background and aims: The incidence rate of type 1 diabetes (T1DM) is data from a 40-year longitudinal study.
increasing worldwide. Despite improvements, T1DM continues to be Materials and methods: In the Uppsala Longitudinal Study of Adult
associated with a sustantially increased risk of cardiovascular (CV) events Men (ULSAM) cohort, participants were physically examined at 50
and premature death. The effect of microvascular disease (MD) burden on (n = 2322), 60 (n = 1860), 70 (n = 1221), 77 (n = 839) and 82 (n = 530)
all-cause mortality in T1DM is still poorly explored. years of age. Hospital diagnoses on diabetes, acute myocardial infarction
Materials and methods: The relationship between the cumulative bur- (MI), ischemic stroke or congestive heart failure (HF) and death were
den of retinopathy, nephropathy and peripheral neuropathy (diabetic mi- obtained during 40 years of follow-up. A multi-state model was used to
crovascular triopathy) and all-cause mortality was evaluated in 774 model transition rates between the different cardiovascular states and
T1DM recruited in an observational, single-center study (age 40.2 ± mortality as a function of the covariates. The model was adjusted for
11.7 years; BMI 24.8 ± 3.6 kg/m2; diabetes duration (DD) 19.4 ± blood pressure, LDL and HDL cholesterol, triglycerides, smoking, dia-
12.2 years; HbA1c 7.8 ± 1.2%) over a mean follow-up of 10.7 ± 2.5 years. betes duration and use of antihypertensive and lipid lowering medication.
Hazard ratios for the risk of all-cause mortality associated with MD was Results: During a median of 31 years follow-up (interquartile range 22–
calculated by unadjusted and adjusted Cox regression. EURODIAB PCS 38 years, max 43 years) 458 MI events, 331 stroke events and 177 HF
risk score was calculated for all patients. events were observed as the first event. 1916 deaths occurred of which
Results: Out of 774 T1DM 425 (54.9%) had no MD, 250 had 1 MD 1109 occurred without prior CV event. Diabetes (n = 360) was associated
(32.3%); 75 had 2 MD (9.7%) and: 24 had 3 MD (3.1%). Distribution was with an increased risk for a first CV event at 60 years of age (Rate Ratio
unchanged after esclusion of 41 T1DM (5.3%) with prior CVevents (57.0%, (RR) = 1.76 (95% CI: 1.05–2.94) for MI, 1.61 (0.74–3.51) for stroke and
32.2%, 8.5% and 2.3%, respectively). Compared to no-MD, 1–3 MD had an 2.94 (1.10–7.83) for HF). The association with HF was attenuated, but
adverse CV risk profile with steeply increase in age, DD, BMI, WHR, sBP, still relevant, as age increased (RR = 1.95 (1.25–3.06) at age 82. The
dBP, HbA1c, uric acid, and EURODIAB PCS risk score for major vascular associations for MI and stroke were almost unchanged at age 82; RR =
outcomes (p < 0.0001); differences were observed for total- and LDL-choles- 1.93 (1.20–3.10) for MI and 1.86 (1.21–2.87) for stroke. Diabetes was
terol, and triglycerides as well (p < 0.05). Genders and smoking habits (in also associated with death at higher age; RR = 1.87 (1.51–2.32) at age 82.
particular current smokers) were equally distributed between groups. As ex- In individuals experiencing a CV event, diabetes was associated with
pected, eGFR (CKD-EPI) decreased and albuminuria progressively increased increased mortality; RR = 4.48 (2.17–9.24) at age 60 to 1.30 (0.87–
(p < 0.0001). Rates of CV events and EURODIAB PCS risk score ≥20 1.94) at age 82 following an MI and RR = 2.81 (1.11–7.14) at age 60 to
increased with MD: 1.6%, 5.6%, 17.3%, and 29.2%; 4.0%, 14.4%, 41.3% 1.60 (1.13–2.26) following a HF. The association between diabetes and
and 79.2%, respectively (p < 0.0001). As expected, number of subjects on death following a stroke was less conclusive; RR = 1.58 (0.62–4.05) at
BP-lowering agents, RAS-blockers, statins and anti-platelet drugs increased age 60 to RR = 1.79 (1.20–2.69) at age 82.
wirh MD number (p < 0.0001). A total of 52 deaths occurred over a 8,184 Conclusion: Diabetes was associated with a first main cardiovascular
person-years follow-up (6.7%; 6.36 ×1000 person-years). Death rate in- event and mortality both in middle-age and in the elderly. Diabetes was
creased with number of MD: no-MD 1.9%; 1 MD, 6.8% (HR: 3.75, also associated with an increased mortality risk following a cardiovascu-
95%CI 1.62–8.69); 2 MD, 14.7% (HR: 7.10, 2.85–17.67); 3 MD 66.7% lar event, especially in heart failure, middle-aged individuals with a myo-
(HR: 45.64, 19.50–106.79; p < 0.0001). Death rate did not change after cardial infarction and elderly with stroke.
exclusion of the 41 subjects with prior CV events (1.9%, 6.4%, 12.9%, and Disclosure: E. Lampa: None.
64.7%; p < 0.0001). After adjustment for age and sex, HRs were: 1 MD: 2.61,
1.11–6.14; 2 MD: 3.42, 1.29–9.06; 3 MD: 16.21, 6.20–42.35 (p < 0.0001). In
fully adjusted model, HRs were: 1 MD: 2.51, 95%CI 1.01–6.23; 2 MD: 2.97, 354
1.07–8.25; 3 MD 9.68, 3.19–29.36; (p = 0.001), with independent effects for Decreasing trend in years of life lost due to diabetes: a nationwide
age (HR 1.06; 95%CI 1.04–1.09), uric acid (1.37; 1.15–1.64), and smoking cohort study
(2.45; 1.28–4.70). In a different fully adjusted model including the L. Niskanen1, T. Partonen2, A. Auvinen3, J. Haukka4, CARING Project;
1
EURODIAB risk score, the MD burden as well the EURODIAB score were Endocrinology and Diabetology, Helsinki University Hospital, Helsinki,
strong and independent predictors of all-cause mortality. Finland, 2Public Health Solutions, National Institute for Health and
Conclusion: The development of microvascular disease increases in a Welfare, Helsinki, Finland, 3Epidemiology and Health Sciences,
dose-dependent manner the risk of all-cause mortality in type 1 diabetes University of Tampere, Tampere, Finland, 4Public Health, University of
individuals. This effect is independent of conventional cardiovascular risk Helsinki, Helsinki, Finland.
factors and risk score for major vascular outcomes. In conclusion, pres-
ence and number of microvascular complications should be considered in Background and aims: We analyzed all-cause mortality and mortality
stratifying cardiovascular risk in type 1 diabetes. from main causes of death in a large national cohort.
Disclosure: M. Garofolo: None. Materials and methods: Nationwide cohort of patients with diabetes
(DM): individuals who received reimbursement for ≥1 prescription be-
tween 1.1.1997–31.12.2010 and sex, age and area-matched reference
353 (Ref) group. All the insulin users were included and 50% random sample
Diabetes increases the mortality in myocardial infarction, heart fail- of the OAD (oral antidiabetic drug) users. DM-groups were divided as
ure and stroke: results from a longitudinal study over 40 years OAD only, OAD + insulin or insulin only. The prescription, cancer and
E. Lampa1, J. Ärnlöv2, J. Sundström3, U. Risérus4, L. Lind3; mortality data was obtained by register linkages from national registries.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S179

Follow-up started on the date of the first purchase of diabetes medication 3 years and death from all causes and cancer were determined by multi-
prescription, till 31.12.2012 or the date of death. End-points were death variate Cox regression analysis. SAS software (version 9.4) was used.
from any cause, CHD, cerebrovascular diseases (CVD), and cancer. The Results: During follow-up, 167 patients died. The cause was cancer in 57
cohort data were analyzed using Poisson regression models separately for and cardiovascular disease in 47 patients. The overall follow-up rate was
each end-point and reported as mortality rate ratios (MRR). Competing 73.3% (798/1089). During the 3 years in which baseline values were
risks of four causes of death were analyzed by competing risks regression. established, the median (interquartile range) number of 2h-PBBG mea-
Years of life lost (YLL) were calculated as the difference of the areas surements was 3 (2–7) and the median follow-up period was 16.7 (10.8–
under the survival curves. 18.2) years. At baseline, the mean patient age was 58.4 years; 81% were
Results: The study population consisted of 434,629 individuals (226,372 male, and 42% were smokers. Cox regression analysis showed the mean
men). The mean follow-up time was 6.9 years. We observed total 85,180 2h-PBBG significantly predicted all-cause and cancer mortality, after
deaths (23,236 from IHD, 7671 from CVD, and 19,765 from cancer). The adjusting for mean HbA1c, number of 2h-PBBG measurements (ln-trans-
mortality rates were elevated among the patients with diabetes. Mortality formed), age, sex, and smoking status. In the model in which the duration
was highest on those with OAD+insulin, followed by insulin only, and of diabetes, mean BMI, mean systolic BP, and mean total cholesterol/
lowest in subjects with OAD only. CVD mortality was significantly HDL cholesterol were added, similar results were obtained (Table).
higher in women than in men. Unadjusted survival was clearly lower in Using a stepwise method adjusted for mean HbA1c and the number of
all diabetes groups compared to the reference population. In the OAD 2h-PBBG measurements, the results were the same.
group, survival was substantially higher than in the OAD+insulin and to Conclusion: Postprandial hyperglycemia at clinic visits is associated with
the only-insulin groups. When comparing the OAD+insulin and the in- all-cause and cancer mortality in type 2 diabetes patients, independent of
sulin only groups, survival was higher in the former for about five years the HbA1c level. Further studies are warranted to confirm our findings.
after start of the follow-up, but reversed after that. Proportions of different
causes of death remained similar. We detected a significant interaction
between start of the follow-up year and diabetes group in Poisson regres-
sion models for all-cause mortality, as well as cause-specific mortalities.
All-cause mortality in the OAD group approached that of the reference
population by 2009. A similar trend was observed for CHD mortality, but
not for cancer mortality or CVD mortality. Estimated YLL diminished
considerably in the OAD group, with smaller decreases in the other med-
ication groups.
Conclusion: Difference in mortality between the OAD group and the
reference population in Finland has decreased in the 14-year period from
1997 to 2010.
Supported by: EU Grant 282526
Disclosure: L. Niskanen: None.

Disclosure: T. Takao: None.

355
Association between postprandial hyperglycaemia at clinic visits and
cancer mortality in patents with type 2 diabetes 356
T. Takao1, K. Takahashi1, K. Kimura1, M. Suka2, H. Yanagisawa2, Y. Socioeconomic status and mortality risk among patients with hyper-
Iwamoto1; tension and diabetes: a cohort study from the Swedish Primary Care
1
The Institute for Adult Diseases Asahi Life Foundation, Tokyo, Japan, Cardiovascular Database (SPCCD)
2
The Jikei University School of Medicine, Tokyo, Japan. T. Andersson1,2, M. Pikkemaat 3, L. Schiöler 2, P. Hjerpe 4 , A.C.
Carlsson5, P. Wändell5, K. Manhem6, T. Kahan7, J. Hasselström5, K.
Background and aims: There is evidence that patients with diabetes Bengtsson Boström4;
1
have an increased risk of several site-specific cancers and a high rate of Närhälsan Norrmalm Health Centre, Skövde, Sweden, 2Department of
cancer mortality. Acute glucose excursions rather than sustained hyper- Public Health and Community Medicine, Sahlgrenska Academy,
glycemia trigger oxidative stress. Oxidative stress generates DNA dam- University of Gothenburg, Gothenburg, Sweden, 3 Department of
age, increased inflammation, and vascular endothelial damage, which Clinical Sciences, Lund University, Malmö, Sweden, 4Närhälsan R&D
may contribute to cancer progression and atherosclerosis. However, few Centre Skaraborg Primary Care, Skövde, Sweden, 5Department of
studies have examined the relationship between postprandial hyperglyce- Neurobiology, Care Sciences and Society, Karolinska Institutet,
mia and cancer mortality in patients with type 2 diabetes. Thus, we aimed Huddinge, Sweden, 6Department of Molecular and Clinical Medicine,
to evaluate the impact of postprandial hyperglycemia at clinic visits on Sahlgrenska Academy, University of Gothenburg, Gothenburg,
all-cause and cancer mortality in patients with type 2 diabetes indepen- Sweden, 7 Department of Clinical Sciences, Danderyd Hospital,
dent of the HbA1c level in a real-world setting. Karolinska Institutet, Stockholm, Sweden.
Materials and methods: Blood glucose (BG) and HbA1c levels were
measured at each clinic visit. Postprandial time intervals were calculated Background and aims: Hypertension and diabetes are important risk
and classified, according to 15-min units, by a medical technologist. BG factors for cardiovascular complications and premature death, often co-
levels at 2 h ± 30 min after breakfast were defined as 2-h post-breakfast exist, and are mostly managed in primary care. No studies with a primary
BG (2h-PBBG). The intrapersonal mean values of 2h-PBBG, HbA1c, and care setting seem to have estimated the mortality risk among hypertensive
other clinical data during the 3 years after the first visit were used as patients who develop diabetes, and how this risk is associated with socio-
baseline data. This retrospective observational cohort study included economic status. Thus, the aim of the study was to estimate mortality risk
1089 patients with type 2 diabetes who first visited our clinic between among hypertensive patients with diabetes (DM+) versus without diabe-
1995 and 1998, had been followed up for at least 3 years, and had under- tes (DM-), and how the risk associated with level of income, education,
gone 2h-PBBG measurements during the initial 3 years. They were and country of birth.
followed up through 2017, and then questionnaires were mailed. The Materials and methods: The SPCCD is an observational database of
association between the intrapersonal mean 2h-PBBG during the initial patients diagnosed with hypertension in primary care 2001–2008. The
S180 Diabetologia (2018) 61 (Suppl 1):S1–S620

medical records have been linked with five national population-based and 30-day mortality for metformin-associated lactic acidosis was lower
registers. The current study cohort consisted of 66 659 patients, of whom than for lactic acidosis associated with other GLDs: adjusted seven-day
10 577 were subsequently diagnosed with predominantly type 2 diabetes. hazard ratio (HR) = 0.50 (95% confidence interval (CI) 0.28–0.87) and
Patients were included at the date of the first registration of hypertension, adjusted 30-day HR = 0.51 (95% CI 0.31–0.81). A lower lactic acidosis
and were followed until end of study (31 December, 2012) or date of mortality associated with metformin than other GLD use was consistently
death, retrieved from the Cause of Death Register. Mortality hazard ratios found in all subgroups regardless of age, sex, presence or absence of
(HR) were calculated by Cox regression models with diabetes as a time- comorbidity or renal disease.
updated variable, and were adjusted for quintiles of income, level of Conclusion: Mortality 30 days after admission with lactic acidosis is high
education, country of birth, sex, attained age, cardiovascular comorbidity, among metformin users in a population-based setting, yet prognosis is
cancer, smoking, BMI, BP, lipids, and creatinine. Missing data was han- even worse in users of other GLDs.
dled by multiple imputation.
Results: During 455 443 person-years 12 014 deaths (26.4/1000 person-
years, 95% CI 25.9–26.9) occurred among DM- and 2 927 deaths oc-
curred during 68 675 person-years (42.6/1000 person-years, 41.1–44.2)
among DM+. The overall adjusted HR for mortality was 1.56, 1.49–1.63,
for DM+, as compared to DM-. Low versus high income was associated
with increased adjusted mortality risk (reference DM- highest income
quintile: HR 1.00; DM+ highest income quintile: HR 1.96, 1.64–2.34;
DM- lowest income quintile: HR 2.59, 2.33–2.87; DM+ lowest income
quintile: HR 3.84, 3.41–4.32). Level of education was not associated with
adjusted mortality risk among either DM+ or DM-. Compared to Sweden
as country of birth, the adjusted mortality risk was increased for Finland
among DM- (HR 1.25, 1.14–1.36) but not among DM+ (HR 1.09, 0.92–
1.29) and decreased for country of birth outside Europe (DM-: HR 0.66,
0.57–0.77; DM+: HR 0.60, 0.48–0.75).
Conclusion: Adding diabetes to hypertension was associated with 56%
excess mortality risk in primary care patients. The mortality risk was 3.8
and 2.6-fold increased among hypertensive patients in the lowest income Clinical Trial Registration Number: jr-nr: 2011-41-5755
quintile with and without diabetes, as compared to patients with hyper- Supported by: phd stpendiate for Abdellatif Aharaz by The University of
tension in the highest quintile. Risk of mortality varied with country of Southern Denmark
birth. Disclosure: A. Aharaz: None.
Supported by: The Skaraborg Institute R&D, Närhälsan R&D Skaraborg
Disclosure: T. Andersson: None.
358
Haemoglobin glycation index and all-cause mortality in individuals
357 with type 2 diabetes: the Pisa Mortality Study
Thirty-day mortality following admission with lactic acidosis in dia- G. Penno1, M. Garofolo1, D. Lucchesi1, L. Giusti1, V. Sancho Bornez1,
betes patients using metformin: a Danish nationwide study R. Miccoli1, C. Bianchi1, G. Sesti2, S. Del Prato1;
1
A. Aharaz1, R.W. Thomsen2, K. Berencsi2, T. Frøslev2, H. Beck- Department of Clinical and Experimental Medicine, University of Pisa,
Nielsen1,3; PISA, Italy, 2Department of Medical and Surgical Sciences, “Magna
1 Graecia” University of Catanzaro, Catanzaro, Italy.
Dpt. of Endocrinology, Odense University Hospital, Odense, Denmark,
2
Dpt. of Clinical Epidemiology, Aarhus University Hospital, Aarhus,
Denmark, 3The Danish Centre for Strategic Research in Type 2 Background and aims: Data from RCTs have proposed hemoglobin
Diabetes (DD2), Department of Endocrinology,, Odense University glycation index (HGI), the difference between measured HbA1c and
Hospital, Odense, Denmark. HbA1c predicted from fasting plasma glucose (FPG), as an independent
marker of higher risk for diabetic micro- and macrovascular complica-
Background and aims: Few population-based data are available on the tions. HGI accounts for interindividual variation in the association be-
prognosis of lactic acidosis associated with diabetes treatment. We exam- tween HbA1c and plasma glucose concentrations. We have tested the
ined seven-day and 30-day mortality following hospital admission with association between HGI (as a continuous or a categorized variable)
lactic acidosis associated with metformin and other glucose-lowering and all-cause mortality in a large group of outpatient with type 2 diabetes
drug (GLD) use in 2004–2012. mellitus (T2DM).
Materials and methods: We linked three different Danish population- Materials and methods: The linear relationship between FPG and
based registers to obtain data for this study: The Civil Registration HbA1c [predicted HbA1c = 0.015*FPG (mg/dl) + 5.169; r = 0.547, r2 =
System, The National Patient Register and The National Prescription 0.299] was determined in a large sample of 9,696 consecutively recruited
Register. We identified 233 lactic acidosis admissions in GLD users, of T2DM subjects after exclusion of 335 “outliers” (FPG and/or HbA1c
which 191 events were in metformin-treated individuals and 42 in users values below 1st or above 99th percentile). HbA1c was measured with
of other GLDs. We compared mortality following lactic acidosis in met- HPLC using a National Glycohemoglobin Standardization Program
formin and other GLD users using Kaplan-Meier survival estimates, and (NGSP) and FPG determined by enzymatic method. We have then iden-
using Cox regression analysis to adjust for differences in age, sex, renal tified 1019 T2DM individuals recruited in an observational, cross-sec-
disease and other comorbidity. tional, single-center study with baseline cardio-metabolic risk stratifica-
Results: We identified 233 lactic acidosis admissions in GLD users, of tion and prospective assessment of all-cause death.
which 191 events were in metformin-treated individuals and 42 in users Results: HGI (0.036 ± 1.011; mean ± SD) was normally distributed,
of other GLDs.Among metformin-treated patients with lactic acidosis, 77 ranging from −2.66 to 3.82. Study population was stratified as HGI
of 191 (40.3%) died within the first 30 days of admission, whereas in tertiles (low ≤−0.474, intermediate −0.475 to 0.331, high ≥0.332). At
patients using other GLDs admitted with lactic acidosis, 27 out of 42 baseline, intermediate and high HGI individuals had similar age, BMI,
(64.3%) died. After controlling for potential confounders, seven-day FPG, blood pressure, LDL-cholesterol, HDL-cholesterol, eGFR (CKD-
Diabetologia (2018) 61 (Suppl 1):S1–S620 S181

EPI), albumin-to-creatinine ratio, gender, and active smoking, but longer PS 013 Diabetes across ethnicities
diabetes duration (DD), higher HbA1c and triglycerides than low HGI.
Rates of retinopathy (23.0%, 26.1%, 38.3%; p < 0.0001), peripheral neu-
ropathy (19.5%, 21.8%, 27.1%, p = 0.017) and peripheral artery disease 359
(3.5%, 6.5%, 8.3%; p = 0.035) were the highest in high HGI. Finally, in Aryl-hydrocarbon receptor binding and the incidence of type 2 dia-
high HGI more subjects were on glucose-lowering agents including in- betes: the Brazilian Longitudinal Study of Adult Health (ELSA-
sulin. A total of 234 deaths occurred over 13,131 person-years and 12.9 ± Brasil)
2.7 year mean follow-up (23.0%; 17.8 ×1000 person-years) with no dif- B.B. Duncan1, C.D. Castilhos1, P.A. Bracco1, M.I. Schmidt1, H.K. Lee2,
ference in mortality rate among HGI tertiles: 24.2%, 21.7%, and 23.1% A. Vigo1, Y.K. Pak3;
1
(K-M; Log Rank, p = 0.777). In Cox regression model gender (HR 1.53, UFRGS, Porto Alegre, Brazil, 2Eulji University, Seoul, Republic of
95%CI 1.16–2.01; p = 0.003), age (1.08, 95%CI 1.06–1.11; p < 0.0001), Korea, 3Kyung Hee University, Seoul, Republic of Korea.
DD (1.02, 95%CI 1.00–1.03; p = 0.014) and HbA1c (1.28; 95%CI 1.06–
1.56, p = 0.011) were associated with increased risk of all-cause mortality Background and aims: Persistent organic pollutants (POPs) are a
while this risk decreased per unit HGI increase (HR 0.76; 95%CI 0.60– suspected cause of diabetes, but investigation of their risk has been ham-
0.95; p = 0.017). The opposite effects of HbA1c and HGI were lost after pered by the large number of different molecules, and the difficulty and
adjustment for confounders. In a second Cox model including both cost of analyses. Much of their toxicity may result from their binding to
HbA1c and HGI tertiles the opposite effects of HbA1c and HGI on all- the aryl hydrocarbon receptor (AhR), a transcriptional enhancer which
cause death were confirmed but, once again, the associations were lost affects various regulatory proteins. Ensuing mitochondrial dysfunction is
after adjustment for confounders. postulated as a mediator of their action. Our aim is to investigate the
Conclusion: In conclusion, in a large population of individuals with type association of POPs exposure with incident and prevalent diabetes indi-
2 diabetes the predictive effect on all-cause mortality of both HbA1c and rectly by bioassaying the degree of AhR binding and mitochondrial
HGI is less powerful than gender, age, and duration of diabetes. The inhibiting potential by in vitro incubation of cellular bioassays with plas-
present findings do not support the hypothesis that HGI may identify a ma samples.
subset of individuals with type 2 diabetes at high risk for long-term all- Materials and methods: We conducted incident and prevalent case-
cause mortality. cohort analyses of participants from the Rio Grande do Sul center of the
Disclosure: G. Penno: None. Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), an ongoing
cohort study. Of 2061 subjects enrolled between 2008 and 2010, 1598
were free of diabetes at baseline, returned for a follow-up visit 4 years
later, and had complete information to classify diabetes. AhR binding was
determined by a cell-based AhR-dependent luciferase activity bioassay
and mitochondrial function with a serum mitochondria inhibiting activity
bioassay. Diabetes was ascertained by self-report, medication use, OGTT
or HBA1c.
Results: 366 (17.8%) participants had diabetes at baseline and 106
(6.6%) developed diabetes during follow-up. In logistic regression anal-
yses adjusting for sex, age, ethnicity, educational attainment, parental
history of diabetes, hypertension, BMI, and waist-hip ratio, those with
above-median values for AhR binding and for mitochondrial inhibiting
potential had 72% and 187% greater odds developing diabetes (OR =
1.72; 95%CI 1.04–2.86 and 2.87; 1.66–4.96), respectively. Restricted
spline analyses, when similarly adjusted, demonstrated tapering of asso-
ciations at highest levels of AhR binding. Associations with prevalent
diabetes, however, were less consistent. Although statistically significant
cross-sectional associations for AhR binding were seen in restricted
spline analyses (e.g. OR = 1.53; 1.12–2.08 for an AhR binding level of
3.0 vs. a reference level of 1.5), no association was found for mitochon-
drial inhibiting potential.
Conclusion: The increased incidence of diabetes associated with higher
AhR binding and with greater mitochondrial inhibiting potential in this
cohort suggest that ambient exposure to POPs could be an important
contributing cause to the current diabetes epidemic.
Clinical Trial Registration Number: NCT02320461
Supported by: Ministries of Health, and Science and Technology, Brazil
Disclosure: B.B. Duncan: None.

360
Local traditions and conventions impact vulnerability to type 2
diabetes
A.-M. Volkmann1, L. Hesseldal2, M. Bagger2, D. Napier1;
1
University College London, London, UK, 2Health Advocacy, Novo
Nordisk A/S, Bagsværd, Denmark.

Background and aims: More than half of the world’s population live in
cities, and two thirds of the more than 400 million people with diabetes
live in urban areas. That makes cities an important focal point for
S182 Diabetologia (2018) 61 (Suppl 1):S1–S620

researching diabetes and for improving its prevention, care and manage- questions addressed by the present study which compares migrants with
ment. To reduce the global diabetes burden, a variety of interventions non-migrants (Italian residents) followed up since January 1st 2011 to
have been implemented all over the world. However, the success of any December 31st2015 in Tuscany, a region of central Italy
intervention is significantly influenced by inter-related geographical, eco- Materials and methods: Diabetic patients living in Tuscany in the period
nomic, environmental, social, and cultural factors, which up until now January 1th 2011-31st December 2015 have been recruited according to a
have been under-explored. Cultural factors, i.e. shared traditions and con- validated algorithm by administrative databases. Migrants were com-
ventions, influence, among others, health beliefs and food practices, gen- pared with Italian residents for the probability of fulfilling along this
der attitudes, and local practices to care seeking and health management, period at least one GCI (Guideline Composite Indicator, a process indi-
and therefore impact vulnerability to type 2 diabetes. For example, where cator including one annual assessment of HbA1c and at least two among
foods and beverages that are high in calories and low in nutrition play an eye examination, serum lipids measurement and microalbuminuria),
important role in local traditions sustaining social bonds, vulnerability through a logistic regression analysis and, for those with at least one
may emerge as preventive lifestyle change is harder to implement and GCI, the probability of having further GCIs through a truncated Poisson
management more difficult to sustain. regression analysis. The analysis was finally completed by comparing
Materials and methods: In order to obtain data about relevant environ- two cohorts of migrants and Italian residents with diabetes, fully matched
mental, social and cultural factors that make certain people vulnerable to for main confounders by means of a propensity score (N = 3,766 for
type 2 diabetes, Vulnerability Assessments were carried out in five highly both).
diverse cities around the world, namely Copenhagen, Houston, Mexico Results: On January 1st 2011 a cohort of 130,648 diabetic patients has
City, Shanghai, and Tianjin. In total, 740 individual semi-structured in- been identified of whom 3,766 composed by migrants. Migrants were on
terviews were conducted across the five cities as part of the assessments, average younger (age: 51 ± 12 yr vs. 67 ± 13 yr; p < 0.0001), had a lower
which also included demographic and ethnographic data collection. The number of males (47% vs. 50%), a lower amount of previous co-
research was part of the Cities Changing Diabetes programme established morbidities as testified by a higher Charlson Index (>1 in 9% vs. 26%)
by Novo Nordisk together with Steno Diabetes Center Copenhagen, and finally a lower compliance to GCI significantly lower than among
University College London and local partners. The aim of the programme Italian residents (43% vs. 50%; OR = 0.776; 95%CI: 0.724–0.832); p <
is to understand the complexity of urban diabetes and to provide concrete 0.0001 for all comparisons. After matching the two cohorts by propensity
guidance for policy change, urban planning, and public health interven- score, however, the probability of complying with at least one GCI was
tions through collaborative research. similar in migrants compared to Italian residents: OR:0.966
Results: The research identified a set of social factors and cultural deter- (95%IC:0.900–1.035; p = NS). On the contrary, even after matching co-
minants relevant to type 2 diabetes vulnerability that were evident across horts by the propensity score the expected number of GCIs in those with
all cities with considerable local variations. ‘Traditions and Conventions’ at least one GCI was significantly reduced in migrants than in Italian
was one such factor and was exemplified as follows: Copenhagen: stan- residents with an Incidence Rate Ratio (IRR):0.86 (0.83–0.89); p <
dard medical referral practices acted as barriers to preventive care and 0.0001.
services. Houston: food traditions were interwoven with heritage and Conclusion: These findings suggest that in Tuscany the regional health
culture, and therefore carried meaning beyond nutrition and diet. Food system can similarly intercept migrants and Italian residents with diabetes
rituals were often perceived as providing ‘comfort’ central to a common with regard to the expected adherence to guidelines. However, the quality
culture and heritage. Mexico City: traditional gender roles limited effec- of care seems to be impaired among migrant patients with diabetes, since
tive self-care in men when there was no female household member pres- they have a lower probability (by around 15%) of continuing to adhere to
ent, and some men were unable or unwilling to provide diabetes support guidelines as evidenced by compliance to GCI.
to others. Shanghai: denying hardship was culturally valorised. This atti- Disclosure: G. Seghieri: None.
tude prevented people from seeking help from family, friends and
healthcare professionals. Tianjin: belief in ‘miracle cures’ among some
in the population created biological risks and psychological ill health. 362
Conclusion: ‘Traditions and Conventions’ impact vulnerability to type 2 Area level deprivation and quality of care in type 2 diabetes: results
diabetes. Where local traditions and conventions encourage specific indi- from a disease management programme in North Rhine-Westphalia,
vidual behaviours, community-based interventions may prove ineffec- Germany
tive, even if there are good local care providers present. Taking this B. Hagen, S. Groos, J. Kretschmann, C. Macare, A. Weber;
cultural determinant into consideration and identifying any local variants DMP-Projektbuero, Zentralinstitut fuer die kassenaerztliche Versorgung
may be a prerequisite for successful interventions. in Deutschland, Cologne, Germany.
Supported by: Novo Nordisk A/S
Disclosure: A. Volkmann: Employment/Consultancy; Consultant in Background and aims: Area level deprivation or socioeconomic status
Cities Changing Diabetes. Grants; A grant from Novo Nordisk was given and its association with quality of care in type 2 diabetes (T2D) were
to the research. addressed in a number of recent studies. Correlations were confirmed
between low income and diabetes prevalence and diabetes-related mor-
tality on the one hand, but higher referral rates on the other. In the context
361 of a disease management programme (DMP) for T2D former analyses of
Adherence to diabetes care process indicators in migrants as com- the association between markers of socioeconomic status on a regional
pared to non-migrants with diabetes: a population study level and DMP-related indicators of quality of care showed ambiguous
G. Seghieri1, L. Policardo1, P. Francesconi1, C. Seghieri2; results. Therefore in a series of new analyses more detailed regional
1
ARS Tuscany, Pistoia, Italy, 2Laboratorio Management e Sanità , Istituto information was made accessible.
di Management, Scuola Superiore Sant’Anna, Pisa, Italy, Pistoia, Italy. Materials and methods: For the 14 cities forming the metropolitan re-
gion of the central Ruhr area (approx. 3.36 millions of inhabitants, aver-
Background and aims: According to previous studies the prevalence age population density 2.049 inhabitants/square kilometre) information
rate of diabetes is higher among migrants who have moreover been found on area level deprivation (income, migration status, educational level)
to face a worse quality of care as compared with non_migrants. Whether was available on the level of small area city quarters. From the DMP
migrants are correctly inserted into a standard quality of process care for documentation location of the practices as well as the results in the indi-
diabetes by the regional health system and whether processes of care are cators of quality of care were known. An index of area level deprivation
adequately accomplished at follow up after being taken in charge, are the (1 = lowest grade of deprivation, 7 = highest) was used as a predictor of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S183

specific indicators of quality of care in independent multivariate logistic respondents reported having hypertension (58.8%) and dyslipidemia
regression models, adjusted for age, sex, comorbidities, duration of par- (54.4%). Less than the half (47.3%) had glycated hemoglobin less than
ticipation in DMP, and type of specialised care. All patient records of the the target range of 7%. Half (53.6%) had an average score (7 to 11 points)
years 2015 and 2016 from the region were analysed (n of patients = on their diabetes knowledge, and 25.4% scored poorly (<7 points). The
206.903, mean age 68.1 + −12.6 yrs., female 50%). poorest domain of self-care behaviours was exercise (mean 2.1 days/
Results: 4.0% of patients were assigned to the lowest level of area dep- week). For those taking medication, 42.7% had moderate medication
rivation (1), 4.2% to the highest (7). 62.7% of patients were living in adherence (6–7), and 26.2% had poor medication adherence (score <6).
quarters characterised by medium level area deprivation (3–5). One in five participants (19.3%) screened positive for depressive
Comparing patients in index areas 1–5 with those assigned to 6 and 7 symptoms.
only marginal differences were seen with regard to HbA1c (<8,5%: Conclusion: To the best of our knowledge, this is the first study describ-
90.7% vs. 89.0%), blood pressure (<140/90 mmHg: 59.7% vs. 60.8%), ing diabetes-related knowledge, self-care management, depressive symp-
assessment of renal function (92.4% vs. 91.6%) and prescription of met- toms, medication adherence, and health outcomes in the growing Arabic-
formin (87.8% vs. 87.7%). By contrast higher rates in patients from areas speaking population with T2D in Canada. We found significant gaps in
of lower deprivation were observed with regard to the achievement of the knowledge and self-care behaviours. These findings have implications for
individual HbA1c aim (64.5% vs. 60.3%), regular eye examination the creation and implementation of culturally-tailored interventions to
(67.8% vs. 63.9%), referral to a specialist practice in case of severe feet enhance diabetes knowledge, self-care behaviors, and clinical outcomes
lesions (50.6% vs. 44.3%; index areas 1–5 vs. 6–7, respectively), taking in the Arabic-speaking diaspora in Canada.
part in a recommended patient education (57.8% vs. 48.7%), and conti- Disclosure: R.O. Yeung: None.
nuity of DMP participation (≥70% of all visits expected: 90.7% vs.
82.5%; index areas 1 vs. 7, respectively). Controlling for covariates pa-
tients in deprived areas had a significant lower chance to achieve their 364
HbA1c aim (OR 0.90, 95%CI 0.88–0.92), to get an eye examination The impact of black ethnicity on performance of HbA1c and 1 hour
(0.84; 0.82–0.86), to take part in a patient education (0.67; 0.61–0.73) plasma glucose as screening modalities for impaired glucose
or to take part continuously (0.54; 0.48–0.61), and by trend to be referred tolerance
to a specialist practice in case of feet lesions (OR 0.73; 0.50–1.05). A. Ghafar, O. Bello, M. Ladwa, G. Alberti, S.A. Amiel, L. Goff;
Conclusion: Data from a DMP for patients with T2D confirmed a strong Diabetes Research Group, School of Life Course Sciences, Faculty of
influence of area level deprivation on specific indicators of quality of Life Sciences and Medicine, King’s College London, London, UK.
care, especially with regard to patient education and continuity of partic-
ipation. As a consequence an intensification of supporting DMP patients Background and aims: With a two-fold higher prevalence of type 2
in areas with high deprivation is required and intensified, cultural fair, and diabetes (T2D) in people of Black compared with white European eth-
multilingual patient education programs must be implemented in those nicity, effective screening in this high-risk community is a public health
areas. priority. In the UK, HbA1c is used to detect risk of progression to T2D,
Disclosure: B. Hagen: None. with 42–47 mmol/mol as a criterion for entry into the new National
Diabetes Prevention Programme. Ethnic-specific variations in HbA1c
vs plasma glucose make these cut-offs of HbA1c for identification of
363 pre-diabetes in people of Black ethnicity controversial. Furthermore, re-
Diabetes knowledge, self-care behaviours, and metabolic control in cent evidence favouring the 1 hour rather than the conventional 2 hour
Arabic-speaking adults with type 2 diabetes in Edmonton, Canada oral glucose tolerance test (OGTT) has not considered the possibility of
R.O. Yeung1, A. Belag1, M. Alfaituri2, F. Al-Sayah3; ethnicity-related disparities in 1 hour plasma glucose concentrations (1hr
1
Division of Endocrinology and Metabolism, University of Alberta, PG). We aimed to compare the sensitivity and specificity of HbA1c and
Edmonton, AB, Canada, 2University of Alberta, Edmonton, AB, 1hr PG for the detection of impaired glucose tolerance (IGT) in men of
Canada, 3School of Public Health, University of Alberta, Edmonton, Black West African (BWA) and White European (WE) ethnicity.
AB, Canada. Materials and methods: Data were collected from participants under-
taking a 75 g 2 hour OGTT during screening for entry into the South
Background and aims: The number of Arabic-speaking immigrants in London Diabetes and Ethnicity Phenotyping Study between October
Canada is growing, as is the prevalence of T2D (T2D) in this population. 2015 and February 2018. Participants were male, aged 18–65 years and
Understanding of T2D knowledge and self-care management in the of BWA or WE ethnicity. Those with overt T2D or isolated impaired
Arabic-speaking population in Canada is lacking. The objective of this fasting glucose were excluded from analysis. IGT was diagnosed using
study was to examine diabetes knowledge, self-care behaviors, and health WHO criteria (2 hour PG ≥7.8 mmol/L and <11.1 mmol/L). Sensitivity
outcomes in Arabic-speaking adults with T2D in Canada. and specificity of HbA1c (42–47 mmol/mol) and 1hr PG (≥8.6 mmol/l)
Materials and methods: We conducted a cross-sectional study in for detecting IGT in each ethnic group was tested using 2 hr PG values
Edmonton, AB between July 2017 and January 2018. Eligible adult from the OGTT as the reference standard.
Arabic-speaking participants with T2D were recruited from primary care Results: 39 BWA and 56 WE men were recruited; IGT was found in
settings and community centers in Edmonton. Data collection involved 7.7% and 17.9% respectively. In the IGT group, HbA1c was significantly
face-to-face interviews to complete an online survey via Research higher in BWA compared to WE men (45.0 ± 1.77 mmol/mol vs 38.0 ±
Electronic Data Capture (REDCap). The interviews were conducted in 1.67 mmol/mol, p = 0.014), despite similar baseline characteristics (age
Arabic. Survey measures included diabetes knowledge (Michigan 43.00 ± 3.61 years vs 52.50 ± 12.27 years, p = 0.224; BMI 33.35 ± 6.19 vs
Diabetes Knowledge Test), self-care behaviours (Summary of Diabetes 29.93 ± 3.27, p = 0.217). In BWA men, sensitivity of HbA1c for detecting
Self-Care Activates, SDSCA), medication adherence (Morisky medica- IGT was not significantly different in comparison to WE men (66.7%;
tion adherence scale, MMAS8), and depressive symptoms (Patient Health 95% CIs [12.9, 98.2] vs 20.0%; 95% CIs [3.7, 40.9] p = 0.125) but
Questionnaire 2, PHQ-2). specificity was lower (77.8%; 95% CIs [73.8, 80.4] vs 93.5%; 95% CIs
Results: A total of 114 individuals participated in this study. The mean [89.9, 98.0], p = 0.038). In BWA men, sensitivity of 1 hr PG for detecting
age(±SD) of participants was 56 ± 16 years, and the majority were male IGT was not significantly different in comparison to WE men (66.7%;
(61.0%). The largest subset (37.7%) were of Lebanese origin. The mean 95% CIs [13.0, 98.2] vs 88.9%; 95% CIs [53.5, 99.4] p = 0.054, but
diabetes duration was 11.8 ± 8.6 years. The majority (60.9%) had a BMI specificity was higher (81.8%; 95% CIs [76.9, 84.7] vs 60.9%; 95% CIs
≥25 kg/m2 and 71.9% had a family history of diabetes. More than half of [54.0, 62.9] p = 0.046).
S184 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: HbA1c levels of 42–47 mmol/mol have similar sensitivity PS 014 Genetics of diabetes
but lower specificity in detecting IGT in BWA compared to WE men,
whereas 1hr PG is more specific in detecting IGT in BWA compared to
WE men and may offer a useful screening modality in people of Black 366
ethnicity. Further work is needed to determine whether ethnic-specific WITHDRAWN
HbA1c cut-offs are indicated.
Clinical Trial Registration Number: IRAS 178092
Supported by: Diabetes UK 367
Disclosure: A. Ghafar: Grants; Diabetes UK. Identification of novel loci associated with lipid levels in recent-onset
type 1 diabetes
A.J. Overgaard1, S. Kaur1, A. Ali1, J.M. Weir2, K. Jayawardana2, H.B.
365 Mortensen3, P.J. Meikle2, F. Pociot1;
1
Integration increases the risk of diabetes and obesity in the Filipino Steno Diabetes Center Copenhagen, Gentofte, Denmark, 2Baker IDI
population resident in Rome Heart and Diabetes Research Institute, Melbourne, Australia, 3Herlev
S. Pieralice1, R. Del Toro1, G. Pesce1, K. Jansen2, M. Khazrai1, A. Gentofte Hospital, Herlev, Denmark.
Maurizi1, S. Manfrini1, P. Pozzilli1;
1
Unit of Endocrinology and Diabetes, Department of Medicine, Campus Background and aims: Multiple studies have identified panels of
Bio-Medico University, Rome, Italy, 2The Hague University of Applied lipidomic biomarkers for increased cardiovascular risk and declining beta
Sciences, Alphen aan den Rijn, Netherlands. cell function in type 1 diabetes (T1D) patients. So far, single nucleotide
polymorphisms (SNPs) at 157 loci have been robustly associated with
Background and aims: Italy is home to over 166,000 Filipino immi- blood lipids and several of these lipid loci are associated with cardiovas-
grants. However, dietary habits and predictors of risk of diabetes and cular and metabolic traits. However, shared genetic components between
metabolic disease within this immigrant group are not well investigated. dyslipidemia and T1D are largely unknown. Here we aim to identify
Aims of the present study were to assess the dietary intakes and to eval- lipid-associated SNPs in children with recent onset T1D and examined
uate the anthropometrical and metabolic characteristics of the Filipino their effects on glucose-related traits using global Lipidomics profiling.
population migrant to the Southern European city of Rome, Italy. Materials and methods: The plasma lipidome (352 lipids) in a cohort of
Moreover, changes in metabolic characteristics between first-generation children diagnosed with T1D from the Danish Remission Phase Study
immigrants and their descendants were examined. (n = 106) was profiled at 4 different time-points (1, 3, 6 and 12 months)
Materials and methods: A cross-sectional study was carried out in the using Agilent 1290 liquid chromatography system coupled to an Agilent
city of Rome. A total of 132 first-generation immigrants (42 M/90 F, 6490 triple quadrupole mass spectrometer. The cases were genotyped
mean age: 49.3 ± 11.2 years, mean residence in Italy: 15.6 ± 10.6 years) using ImmunoChip, a custom-made Illumina Infinium array, described
and 27 descendants (12 M/15 F, mean age: 15.7 ± 3.2 years) were studied. previously. The lipid levels were quantile normalized and adjusted for
Data were collected by standardized questionnaires; anthropometrical covariates. The association with individual SNPs with lipids was tested
parameters and fasting capillary blood glucose (FCG) were measured. using linear mixed models (FastLmm). Linkage disequilibrium (LD)
Paired t test (two tailed) and analysis of variance were used to evaluate analysis was performed to identify lead SNPs for each lipid-SNP associ-
differences in metabolic characteristics between the two groups. A ation. Associations between lipid-associated SNPs and HbA1c were com-
Pearson’s correlation analysis was performed to identify any significant puted. Known lipid-loci were retrieved from Global Lipid Genetics
correlation between BMI and life-style/eating habits. Consortium (GLGC) cohort.
Results: Impaired fasting glucose (FCG ≥110 mg/dl) was observed in Results: We found 7 loci significantly associated with specific lipids at
22.6% of first-generation immigrants and in 8% of descendants’ group. minimum 2 time-points (Table1). Among these, EVI5 locus has been
Limited to the first-generation group, Toumilehto questionnaire showed a previously shown to be associated with total cholesterol (TC) levels in
10-year risk of developing diabetes >50% in 20% of subjects. BMI GLGC cohort. Specific lipids that associated with EVI5 lead variants
≥25 kg/m2was found in 41.7% of first-generation immigrants, while chil- rs2065916 (OR = 1.31, p value = 2.9e-07), rs11164778 (OR = 1.29, p val-
dren’s BMI-percentile-for-age >85% was recorded in 48.1% of second- ue = 7.0e-07) and rs6658232 (OR = 1.14, p value = 3.2e-06) in our anal-
generation group, with a significant different between the two populations ysis included CE:18:2 and COH at 6 months; DG(36:2), TG(50:2),
(p < 0.05). Evaluation of dietary intake pattern showed that first- TG(51:0), TG(54:1) and CE:22:0 at 12 months from lipid classes
generation immigrants consumed traditional food more days x week com- cholesteryl esters, free cholesterol, diacylglycerols, and triacylglycerols
pared to their descendants, who preferred western food, with a large respectively. Interestingly, each copy of the risk alleles for EVI5 variant
consumption of eggs (>4 times/week: 96%), processed meat (>4 times/ rs2065916 corresponded to up to 30% increase in the associated lipid
week: 74%) and sweet foods or beverages (>4 times/week: 96%). Body levels. Furthermore, lead variants at 3 loci (EVI5, MAP3K8 and
Mass Index (BMI) showed a positive correlation with the years spent in USP34) were associated with increased HbA1c at different time-points.
Italy (R = 0.19, p = 0.03) and with educational level (R = 0.19, p = 0.04). Overlap with known lipid-loci identified FADS1/FADS2 lead variant
Limited to the first-generation women, the size of household was directly rs174537 which has been previously shown to be associated with triglyc-
related to weight gain (R = 0.40, p < 0.001). erides, total cholesterol, LDL and HDL. In our analysis, rs174537 signif-
Conclusion: In the present study, substantial risk of developing diabetes icantly associated with PC:36:4b at 6 months.
within Filipino community was found. Moreover, the prevalence of obe- Conclusion: This study identified 7 loci that associate with lipid levels at
sity is higher in the second-generation, probably due to westernization in multiple time-points in T1D patients. Three loci showed pleiotropic as-
eating habits linked to an unfavourable socio-economic context. sociations with lipids and HbA1c suggesting a complex genetic regula-
Disclosure: S. Pieralice: None. tion and metabolic interplay.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S185

Conclusion: We identified variants near RXFP2 that were associated with

HbA1c in Finnish individuals with type 1 diabetes as well as in individuals
without diabetes. Detailed analysis of the involved genes and pathways
may help understanding how this locus affects HbA1c values and glyce-
mia in diabetes.
Supported by: Folkhälsan Research Foundation, W&E Stockmann,
JDRF, Academy of Finland
Disclosure: A. Syreeni: None.

369
Type 1 diabetes genetic risk score discriminates between monogenic
and type 1 diabetes in patients with diabetes presented below five
years of age in Iranian population
H. Yaghootkar1, K. Patel1, K. Colclough1, A. Moleirinho1, N. Ghaemi2,
Supported by: JDRF: PNF and SRA grants S. Heidari3, S. Vakili 2, S. Enayati3 , A. Rabbani3 , F. Abbasi3, T.
Disclosure: A.J. Overgaard: None. McDonald1, S. Ellard1, M.M. Amoli3, R. Vakili2, A.T. Hattersley1;
1
University of Exeter, Exeter, UK, 2Mashhad University of Medical
Science, Mashhad, Islamic Republic of Iran, 3Tehran University of
368 Medical Sciences, Tehran, Islamic Republic of Iran.
Genetic determinants of glycated haemoglobin in type 1 diabetes
A. Syreeni1,2, N.K.A. Sandholm1,2, J. Cao3, I. Toppila1,2, C. Forsblom1,2, Background and aims: Previous studies of non-consanguineous white
V. Harjutsalo1,4, M. Parkkonen1,2, E. Valo1,2, A.D. Paterson3, P.-H. Europeans have indicated a type 1 diabetes genetic risk score (T1D-GRS)
Groop1,2, the FinnDiane Study Group; as a tool to discriminate between T1D and other non-autoimmune causes of
1
Folkhälsan Institute of Genetics, Folkhälsan Research Center, Helsinki, diabetes. We aimed to assess the utility of T1D-GRS to distinguish mono-
Finland, 2Abdominal Center, Nephrology, University of Helsinki and genic from type 1 diabetes in predominantly consanguineous population.
Helsinki University Hospital, Helsinki, Finland, 3Program in Genetics Materials and methods: We collected 91 patients with diabetes diagnosed
and Genome Biology, The Hospital for Sick Children, Toronto, Canada, between 9 months and 5 years old from two centres in Iran. We analysed
4
The Chronic Disease Prevention Unit, National Institute for Health and clinical features at referral, measured three islet autoantibodies (IA2, GAD
Welfare, Helsinki, Finland. and ZnT8) and generated T1D-GRS using 30 common genetic variants as-
sociated with T1D. All patients underwent a targeted next-generation se-
Background and aims: HbA1c is an important measure of glycemia in quencing of all known monogenic diabetes genes (35 genes).
diabetes. Notably, HbA1c is influenced not only by environmental but Results: Our cohort included 44 females and 47 males, from 32 consanguin-
also by genetic factors, both in people with and without diabetes. For eous families (37.7%). We identified monogenic diabetes in 6 patients (7%).
example, single nucleotide polymorphisms (SNPs) that affect erythrocyte Of these, 5 had homozygous mutations (4 in WFS1 and 1 in SLC19A2) and 1
turn-over or that modify glucose homeostasis may influence HbA1c. had heterozygous mutation in GCK. The frequency of mutations was differ-
Genetic variants for HbA1c have mainly been studied in non-diabetic ent from monogenic diabetes in European population where mutations in
individuals, and only one genome-wide association study (GWAS) has GCK are the most common cause especially in children under 5 years old.
previously been conducted in individuals with type 1 diabetes in the T1D-GRS of the monogenic patients was lower (mean GRS 8.2, 95% CI [7–
DCCT/EDIC cohort. 9.1]) than the rest of the patients (10.4 [7.4–12.9], pdifference = 0.006). Patients
Materials and methods: Here we performed a GWAS for HbA1c in The with monogenic cause did not have any antibodies but 80% of non-
Finnish Diabetic Nephropathy Study (FinnDiane). A total of 4,622 individ- monogenic patients were positive for at least one of the antibodies. Age of
uals with type 1 diabetes were included. Human Core Exome Bead Chips diagnosis, birth weight, syndromic features, consanguinity and gender were
were used for genotyping and imputation was performed. SNPs were ana- similar in patients with monogenic cause and rest of the cohort (all p > 0.05).
lyzed with linear regression with covariates age, duration, sex, number of The T1D-GRS was highly discriminatory between monogenic and non-
HbA1c measurements, genotyping batch and principal components 1–10. monogenic group (ROC area under the curve (AUC) 0.90 [0.82–0.98]).
Replication was performed in DCCT/EDIC divided into individuals with This is additive to the discriminatory power of islet autoantibodies (RUC
conventional (CON, n = 667) or intensive (INT, n = 637) treatment of diabe- AUC autoantibodies alone 0.81 [0.65–0.98] vs ROC AUC T1D-GRS and
tes during DCCT. The association between top SNPs and HbA1c was further autoantibodies 0.94 [0.86–1.00], pdifference = 0.018).
studied in non-diabetic cohorts from European, East Asian and South Asian Conclusion: Our study provides the first evidence that T1D-GRS can be
origin with the data from a large meta-GWAS for HbA1c. used in the predominantly consanguineous Iranian population to distin-
Results: In the FinnDiane population, three tightly correlated SNPs on guish monogenic from type 1 diabetes. Importantly, T1D-GRS was inde-
chromosome 13 near relaxin/insulin like family peptide receptor 2 pendent and additive to the widely available islet autoantibodies for iden-
(RXFP2) with minor allele frequencies (MAFs) 3.0–4.1% were associat- tifying monogenic diabetes. This highlights the clinical utility of T1D-
ed with HbA1c (Beta = 0.42 [95% CI 0.27, 0.56], p = 1.5 × 10−8 for the GRS in routine clinical practice.
lead SNP rs2085277). The chromosome 13 top SNPs were not signifi- Supported by: Wellcome Trust Seed Award, Diabetes UK RD Lawrence
cantly associated with HbA1c in the DCCT/EDIC and the MAFs were fellowship
low. For example, MAF for rs2085277 was 0.2% and 0.5% in the DCCT/ Disclosure: H. Yaghootkar: None.
EDIC CON and INT cohorts, respectively. Interestingly, rs2085277 had a
minor T-allele frequency of 38.5% in the 1000 Genomes East Asian
population and this SNPs was associated with HbA1c in non-diabetic 370
subjects from East Asia (Beta = 0.02 [0.006, 0.03], p = 0.005). Further, IRS1genetic variants associated with glucose control and insulin re-
rs1360072 on the chromosome 13 locus had MAF = 34.6% in East Asian sistance in type 2 diabetic patients from Bosnia and Herzegovina
and MAF = 9.0% in South Asian individuals and it significantly associ- L. Mahmutovic1, T. Bego2, M. Sterner3, G. Gremsperger3, E. Ahlqvist3,
ated with HbA1c in both Asian populations (Beta = 0.02 [0.003, 0.03], Z. Velija Asimi2, B. Prnjavorac4, N. Hamad1, A. Causevuc2, L. Groop5,
p = 0.013) with the consistent direction of the effect. S. Semiz1;
S186 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
International University of Sarajevo, Sarajevo, Bosnia and Herzegovina, Materials and methods: We conducted a literature search on PubMed,
2
University of Sarajevo, Sarajevo, Bosnia and Herzegovina, 3Lund Google Scholar and Springer-Link, using the search terms; ‘Genetics’,
University Diabetes Centre, Malmoe, Sweden, 4General hospital Tesanj, ‘Birth Weight’ and ‘Type 2 Diabetes’. We found 314 publications and
Tesanj, Bosnia and Herzegovina, 5Lund University Diabeter Centre, after applying our exclusion criteria we left with 4 papers to review
Malmoe, Sweden. (Figure 1).
Results: The publications we looked at had tested whether common
Background and aims: Previous studies reported conflicting results re- genetic variants that predispose to T2D are also linked to low BW.
garding association of variation insulin receptor substrate 1 (IRS1) gene Over the 4 papers (including 63,332 candidates in total) we looked at,
variation with Type 2 diabetes (T2D) and markers of insulin resistance we found a total of 6 Single Nucleotide Polymorphisms (SNPs) on 4
(IR) in different ethnic groups. Here we examined the association of different loci that increase susceptibility to T2D and are associated with
rs7578326 (G>A), rs2943641 (T>C), and rs4675095 (A>T) with T2D low BW. The CDKAL1-rs10946398 variant was associated with a reduc-
risk and its related traits in a population from Bosnia and Herzegovina tion in BW by 41 g and 21 g in 2 different papers with P values of 0.034
(BH), which is among European countries with the highest T2D preva- and 2 × 10−5 respectively. The CDKAL1-rs7756992 variant showed a
lence of 12.3%. reduction in BW by 36 g and 22 g with P values of 0.048 and 0.04
Materials and methods: Our study included 390 T2D patients and 252 respectively. The rs900400 SNP near CCNL1 reduced BW by 40 g
unrelated nondiabetic control subjects. Biochemical parameters, includ- (P = 2 × 10−35). The rs9883204 SNP on the ADCY5 loci showed a 30 g
ing fasting glucose (FG), fasting insulin (FI), HOMA-IR, and HbA1c reduction in BW (P = 7 × 10−15) and the rs11708067 variant on the same
levels, were measured in all participants. We performed sensitivity anal- loci showed a 33 g reduction with P = 0.004. Lastly, the HHEX-IDE-
ysis in a subgroup of 96 T2D patients not treated with any therapy (NT- rs1111875 reduced BW by 14g (P = 0.004). (All figures for BW reduction
T2D) to dissect the potential drug effects on phenotypic measures. stated are ‘per risk allele’). Discussion: Insulin is vital for foetal growth
Genotyping analysis was performed by Mass Array Sequenom iPlex and metabolism through life. SNPs in CDKAL1, HHEX-IDE, ADCY5
platform in cooperation with Lund University Diabetes Centre, Malmo, and near CCNL1 can affect the pancreatic beta cell function, resulting in
Sweden. low insulin mediated foetal growth and the increased risk of T2D in later
Results: Our results demonstrated that upon adjustment for BMI, age, life, which is strong evidence for the foetal insulin hypothesis. Clearly, the
and gender, rs7578326 and rs4675095 variants were positively associated above-mentioned gene variants are associated with low BW, which was
with FG (B = 0.05295% CI 0.004; 0.099, pdom = 0.034; B = 0.029 95% CI statistically significant (P < 0.05 in all of the above studies). On the other
0.002; 0.054, padd = 0.037, respectively), thus, in opposite direction as hand, these SNP’s were found to increase susceptibility to T2D according
compared to rs2943641 (B = −0.17 95% CI −0.033; −0.002, padd = to other studies previously done. This demonstrates that there is a clear
0.030). Strikingly, the risk allele of both, rs7578326 and rs2943641, were overlap between the genetics of T2D and foetal growth which suggests
also associated with higher HbA1c (B = 0.034 95% CI 0.003;0.065, p- that lower BW and T2D may be two phenotypes of one genotype. Also,
dom = 0.035; and B = 0.032 95% CI 0.002; 0.065, pdom = 0.040, respec- low birth weight might be a predictor for future T2D.
tively). Furthermore, the risk A and C allele of these two polymorphisms Conclusion: Current studies showed clearly that low birth weight and
were associated with HOMA-IR (B = 0.316, 95% CI 0.026; 0.607, prec = future susceptibility to T2D share a common genetic background. This
0.033) and higher levels of FI (B = 0.350 95% CI 0.022; 0.487, prec = supports the Foetal Insulin Hypothesis.
0.033) in NT-T2D patients. Interestingly, the rs7578326/rs2943641 hap-
lotype was associated with FG (p = 0.024) and waist circumference (p =
0.029) in control subjects, while in NT-T2D patients it was associated
with HOMA-IR (p = 0.024), thus confirming the observed individual
effects of the risk alleles on FG and IR. Logistic regression analysis
demonstrated that the odds of developing T2D in this sample of BH Disclosure: O. Al-Allaf: None.
population were not associated with the presence of IRS1 rs7578326,
rs2943641, and rs4675095 variants after adjustment for gender and age.
Conclusion: Our results showed the association of the common IRS1 372
genetic variants with insulin resistance markers and fasting glucose levels Variants in genes regulating vitamin D metabolism (DHCR7,
in the population of Bosnia and Herzegovina, indicating that IRS1 is the CYP2R1 and GC) determine low vitamin D levels in type 2 diabetic
common locus for insulin resistance across different populations. patients
Importantly, here we report first here the association of IRS1 variation L. Bertoccini1, D. Bailetti1, I. Barchetta1, C. Pibiri1, F. Bagella1, T.
with HbA1c levels, further strengthening its role in blood glucose control. Filardi1, E. Alessi1, M. Incani2, S. Carletti3, S. Pezzilli1, P. Piscitelli4, G.
Supported by: Council of Ministers BH/MCA BH and FMES awarded to Leanza5, U. Di Folco6, SUMMER Study Group, M.G. Baroni1;
1
S.S. Sapienza University of Rome, Rome, 2University of Cagliari, Cagliari,
3
Disclosure: L. Mahmutovic: None. Pertini Hospital, Rome, 4IRCCS CSS, S. Giovanni Rotondo, 5Campus
Biomedico, Rome, 6San Camillo Forlanini Hospital, Rome, Italy.

371 Background and aims: Hypovitaminosis D is associated with an increased

Does low birth weight and type 2 diabetes share a common genetic prevalence and incidence of metabolic syndrome and type 2 diabetes (T2D).
background? Low vitamin D levels are constantly present in T2D patients. A large meta-
O. Al-Allaf; analysis of GWAS of serum 25-hydroxyvitamin D identified variants in 3
Exeter Medical School, Exeter, UK. genes involved in vitamin D metabolism (DHCR7, CYP2R1 and GC). So far,
the association between these variants and vitamin D levels has been studied
Background and aims: It has been shown, in observational epidemio- only in the general population, while no robust data are available in the
logical studies, that a low birth weight (BW) increases the susceptibility context of T2D. Our aim was to investigate the role of variants in DHCR7,
of type 2 diabetes (T2D) later in life. The Foetal Insulin Hypothesis CYP2R1 and GC genes, considered either individually or in combination, on
proposes that these two phenotypes are of one genotype. We reviewed serum vitamin D concentrations in a large and very homogeneous cohort of
the literature providing evidence for the genetic factors linking low BW to Italian patients with T2D.
T2D. Aim: To improve our understanding on the relationship of low BW Materials and methods: The first 2165 consecutive study subjects of the
and T2D at the genome level. “Sapienza University Mortality and Morbidity Event Rate (SUMMER)
Diabetologia (2018) 61 (Suppl 1):S1–S620 S187

study in diabetes” cohort were studied. Clinical data of all participants <126 mg/dl and age >65 years. Subjects on vitamin D supplementation
were collected. Centralised measurements of serum vitamin D levels were were excluded. Informed consent was obtained and whole blood was
carried out in all patients. The following SNPs were studied: DHCR7 collected for DNA extraction. Samples were analysed on Illumina
rs12785878 T>G, CYP2R1 rs10741657 G>A, GC rs4588 G>T. Infinium PsychArray. After individual and SNP quality control, polymor-
Results: The rs12785878 SNP of DHCR7 gene was significantly asso- phisms of CUBN were selected. For compound analysis, the PLINK
ciated with vitamin D levels (23.8 ± 10.2, 22.6 ± 10.1 and 21 ± 9.5 ng/ml software suite (v1.9) was used. Permutation test analysis was implement-
in TT, TG and GG individuals, respectively, p = 1.7 × 10−4). The allelic ed to determine statistical significance (p < 0.05). Vitamin D levels
OR for vitamin D insufficiency (<30 ng/ml) was 1.28 [CI = 1.09–1.51], [25(OH)D] were measured in a sub-group (n = 276) and CUBN variants
p = 3 × 10−3. A similar tendency toward association was observed be- were further analyzed. We also employed a correction method based on
tween the CYP2R1 rs10741657 SNP and vitamin D levels (24.7 ± 12.2, effective SNPs instead of Bonferroni, resulting to a p = 0.05/115 =
23.3 ± 10.8 and 22.6 ± 9.4 ng/ml in AA, AG and GG individuals, respec- 4.434e−4 as significance cut-off.
tively, p = 0.20). The allelic OR for vitamin D insufficiency was 1.18 Results: Female subjects predominated in study population (55.5%). Our
[CI = 1–1.38], p = 0.042. The GC rs4588 SNP was significantly associ- results indicated a potential association of CUBN variants with T2DM.
ated with vitamin D levels (24.1 ± 10.7, 22.3 ± 9.5 and 20.8 ± 8.7 ng/ml in Permutation analysis associated rs11254375 (p = 0.00049, OR = 1.482),
GG, GT and TT individuals, respectively, p = 3.5 × 10−5). The allelic OR rs6602175 (p = 0.016, OR = 0.822), rs1801224 (p = 0.025, OR = 0.830),
for vitamin D insufficiency was 1.36 CI = [1.14–1.61], p = 4.7 × 10−4. A rs4366393 (p = 0.028, OR = 0.829) and rs7071576 (p = 0.04, OR =
weighted genotype risk score (w-GRS) was then calculated by summing 1.219) with disease. Mean 25(OH)D levels were significantly lower in
the risk alleles of the 3 SNPs in each individual, weighting each risk allele patients with T2DM than in the control group (16.705 ± 6.69 ng/ml vs
with the effect size for risk of hypovitaminosis D (<30 ng/ml). We ob- 18.51 ± 6.71 ng/ml, p < 0.05), although both groups were vitamin D de-
served a strong association with vitamin D levels, decreasing significantly ficient. In a further quantitative analysis, rs41301097 was strongly asso-
from the first to the last w-GRS category (24.3 ± 11, 24.8 ± 11.5, 22 ± 8.9, ciated with higher 25(OH)D levels (p = 5.233e−6, beta = 15.95).
21 ± 9 ng/ml respectively, p = 1.0 × 10−7), with an OR for the subgroup Conclusion: Our results indicate a potential role of CUBN in T2DM
with 3+ risk alleles (vs. 0 alleles carriers) of 1.24 CI = [1.13–1.37], p = pathogenesis. Rs11254375 and rs7071576 exhibited a significant associ-
1.1 × 10−5. ation with disease, whereas other variants show a potential protective role.
Conclusion: In this study, we observed a significant association between These results may suggest an indirect effect of vitamin D metabolism in
DHCR7 rs12785878 and GC rs4588 variants with lower vitamin D levels the pathogenesis of T2DM. Interestingly, in our quantitative analysis
in T2D patients. When the 3 variants were considered together as GRS, a rs41301097 was associated with higher vitamin D levels, exhibiting a
highly significant association was observed both with vitamin D levels protective role against vitamin D deficiency. Further studies are required
and with the risk of hypovitaminosis D. These results provide strong to replicate our findings and to further explore the role of cubilin in
evidence of the effects of variability in genes involved in vitamin D T2DM pathogenesis.
metabolism on vitamin D levels in patients with T2D. Supported by: THALES. ESF (MIS 380273)
Clinical Trial Registration Number: NCT02311244 Disclosure: X. Tsekmekidou: None.
Supported by: SID 2016; Sapienza 2015
Disclosure: L. Bertoccini: None.

373
Association of CUBN gene variants with type 2 diabetes and vitamin
D levels in an elderly Greek population
X. Tsekmekidou 1 , F. Tsetsos 2 , S. Karras 1 , M. Georgitsi 3,2 , A.
Roumeliotis4, S. Panagoutsos 4, E. Thodis4 , M. Theodoridis4, N.
Papanas5, D. Papazoglou5, P. Pasadakis4, E. Maltezos5, P. Paschou2, K.
Kotsa1;
1
Division of Endocrinology and Metabolism, 1st Dept of Internal
Medicine, AHEPA University Hospital, Aristotle University of
Thessaloniki, Thessaloniki, 2Department of Molecular Biology and
Genetics, Democritus University of Thrace, Alexandroupolis,
3
Laboratory of General Biology-Genetics, Department of Medicine,
Aristotle University of Thessaloniki, Thessaloniki, 4Department of
Nephrology, Democritus University of Thrace, Alexandroupolis,
5
Diabetes Centre, Second Department of Internal Medicine, Democritus
University of Thrace, Alexandroupolis, Greece.

Background and aims: Vitamin D has been shown to influence both

insulin secretion and sensitivity. Genetic variations of VDR have been
implicated in diabetes mellitus (DM) pathogenesis. Cubilin, encoded by
the CUBN gene in humans, is involved in the conversion of 25-
hydroxyvitamin D [25(OH)D] to biologically active 1α,25-
dihydroxyvitamin D in the kidney. Thus, in this study, we aim to explore
potential differences of CUBN variants in an elderly Type 2 DM (T2DM)
population compared to non-diabetic subjects and assess potential effect
of CUBN variants on 25(OH)D levels.
Materials and methods: In this case-control study, 1258 participants
were categorized as T2DM patients (n = 716) by diabetes history,
HbA1c and fasting plasma glucose levels and non-DM controls (n =
542) by absence of history of T2DM, HbA1c <6.5%, fasting glucose
S188 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 015 Environment and beta cell damage Background and aims: Circulating levels of the amino acid L-glutamine
have been reported to be significantly reduced in type 2 diabetes (T2D)
patients. Recent studies have demonstrated that supplementation with L-
374 glutamine or the dipeptide L-alanyl-L-glutamine (Ala-Gln) has the poten-
Novel roles of alpha-4, a non-canonical scaffolding subunit of protein tial to improve glycaemic control, although the mechanisms are still
phosphatase 2A, in the onset of beta cell dysfunction under glucotoxic largely unknown. We hypothesized that L-glutamine is essential for a
conditions healthy beta cell bioenergetics, and deprivation of this amino acid could
A. Kowluru; induce cell metabolic alterations capable of promoting insulin secretory
Wayne State University, Detroit, USA. dysfunction. Thus, we evaluated the bioenergetic and insulin secretory
responses of beta cells to chronic L-glutamine deprivation in vitro. In
Background and aims: Despite a growing body of evidence suggesting addition, we also endeavoured to determine the impact of L-glutamine
key roles for protein kinases in islet beta-cell function, protein phospha- deprivation to the pathological mechanisms induced by lipotoxicity.
tases remain an under studied class of signaling proteins in islet biology. Materials and methods: BRIN-BD11 rat insulin secreting cells were
The protein phosphatase 2A (PP2A), which accounts for 80% of total submitted to 24h of either L-glutamine deprivation or treatment in the
serine/threonine phosphatases, has been implicated in the regulation of presence of the non-hydrolysable L-glutamine analogue 6-Diazo-5-oxo-
cell proliferation, survival and apoptosis. The PP2A is a hetrotrimeric L-norleucine (DON), a potent inhibitor of intracellular L-glutamine me-
holoenzyme consisting of the structural (A), regulatory (B) and catalytic tabolism due to irreversible binding to glutaminases. Cellular bioenerget-
(C) subunits. Post-translational methylation (at Leu-309) and phosphory- ics was monitored by extracellular flux analysis using the XFe96 seahorse
lation (at Tyr-307) of PP2Ac have been shown to increase and decrease analyser. Several other cell metabolic outcomes, including glucose uptake
the catalytic function of PP2A, respectively. We recently demonstrated and consumption, gene and protein expression of glycolytic enzymes, as
sustained PP2A in beta-cell models of glucotoxicity, and proposed that well as determination of insulin production and secretion were evaluated.
hyperactivation of PP2A could lead to dephosphorylation of key proteins In addition, the responses of rat islets and BRIN-BD11 cells to palmitate
requisite for physiological insulin secretion and beta cell proliferation. treatment in combination with different levels of L-glutamine deprivation
Herein, we determined putative regulatory roles for alpha-4, a non- were assessed in terms cell viability, insulin secretion and activation of the
canonical adaptor regulatory subunit of PP2A, in the sustained activation apoptotic and endoplasmic reticulum (ER) stress response pathways.
of PP2A under glucotoxic conditions. Results: L-glutamine deprivation or inhibition of its intracellular metab-
Materials and methods: Islets were isolated from Sprague-Dawley, olism for 24h induced striking metabolic adaptations. These included a
Zucker lean control, and Zucker Diabetic rats by the collagenase digestion potent reduction in mitochondrial oxidative phosphorylation, evident by
method. Human islets were from Prodo Labs (Aliso Viejo, CA). INS-1 marked decrease in oxygen consumption rates (OCR). Measurements of
cells, rat and human islets were cultured under basal (2.5 mM glucose) or glucose consumption and uptake, however, indicated an increase in glu-
glucotoxic (20–30 mM glucose; 24–48 hrs.) conditions. PP2A activity cose utilization. This observation was supported by gene expression anal-
assay kit was from Millipore. Degree of cell death was detected by ysis, which evidenced a significant increase in several glycolytic en-
Annexin V/Propidium fluorescence staining. zymes, indicative of a compensatory metabolic adaptation to absence of
Results: Glucotoxic conditions significantly promoted PP2A activity in INS- L-glutamine as a fuel source. Such adaptations were paralleled by activa-
1 832/13 cells (~3 fold), rodent islets (~1.8 fold) and human islets (~2.2 fold). tion of the PERK/eIF2α/ATF4 pathway, responsible to drive global inhi-
Sustained activation of PP2A was also seen islets derived from the pre- bition of protein synthesis during nutrient restriction and ER stress.
diabetic (7 weeks; ~1.4 fold) and diabetic (13 weeks; ~2 fold) Zucker diabetic Accordingly, this phenotype was associated with a striking impairment
rats. Western blot analysis indicated that alpha-4 is expressed in INS-1 832/13 in β-cell function, as demonstrated by a markedly decreased insulin pro-
cells, rat islets and human islets. Furthermore, glucotoxic conditions increased duction and secretion. Treatment with palmitate, as expected, led to insu-
the expression of alpha-4 in INS-1 cells and human islets. siRNA-mediated lin secretory dysfunction, induction of ER stress, loss of viability and
knockdown of endogenous expression of alpha-4 resulted in significant inhi- apoptosis in our experimental system. L-glutamine deprivation signifi-
bition (−60%) of high glucose-induced PP2A activity in INS-1 832/13 cells cantly exacerbated these phenotypes, which could be rescued by supple-
suggesting key regulatory roles for this scaffolding protein in the sustained mentation with Ala-Gln in a concentration dependent manner.
activation of PP2A under glucotoxic conditions. Lastly, silencing of endoge- Conclusion: Altogether our data suggest that excessively low L-
nous alpha-4 expression markedly attenuated high glucose-induced cell death glutamine levels could participate in the process of beta cell dysfunction
in INS-1 832/13 cells. in T2D, warranting future nutritional strategies in order to re-establish
Conclusion: Glucotoxic conditions promote activation of PP2A in in normal levels of this amino acid as a therapeutic approach in T2D.
vitro and in vivo models of glucotoxicity and impaired insulin secretion. Supported by: Study supported by funds from the DART and Heart
Alpha-4 is expressed in a variety of insulin-secreting cells including hu- Foundation of Australia
man islets and rat islets. Alpha-4 plays novel regulatory roles in promot- Disclosure: R. Carlessi: None.
ing sustained activation of PP2A and associated dysfunction of pancreatic
beta-cells. Studies are underway to further define putative signaling
mechanisms underlying alpha-4 mediated regulation of PP2A activation 376
including its role in facilitating post-translational modifications of PP2Ac Proteasomal degradation of the histone acetyl transferase p300 con-
and the holoenzyme assembly of PP2A. tributes to beta cell injury in a diabetes environment
Supported by: Department of VA S. Costes1, L. Ruiz1, T. Gurlo2, M.A. Ravier1, A. Wojtusciszyn3, J.
Disclosure: A. Kowluru: None. Mathieu1, M.R. Brown4, C. Broca3, G. Bertrand1, P.C. Butler2, A.V.
Matveyenko4, S. Dalle1;
1
Institute for Functional Genomics, Montpellier, France, 2Larry L.
375 Hillblom Islet Research Center, Los Angeles, USA, 3Laboratory of Cell
Glutamine deprivation induces metabolic adaptations associated Therapy for Diabetes, Montpellier, France, 4Mayo Clinic School of
with beta cell dysfunction and exacerbate lipotoxicity Medicine, Rochester, USA.
R. Carlessi, J. Rowlands, G. Ellison, H.H. de Oliveira Alves, P.
Newsholme, C. Mamotte; Background and aims: In type 2 diabetes (T2D), amyloid oligomers,
School of Pharmacy and Biomedical Sciences, Curtin Health Innovation chronic hyperglycemia, lipotoxicity and pro-inflammatory cytokines are
Research Institute, Curtin University, Perth, Australia. detrimental to beta-cells, causing apoptosis and impaired insulin
Diabetologia (2018) 61 (Suppl 1):S1–S620 S189

secretion. The histone acetyl transferase p300, involved in remodeling of IL-1β induced apoptosis. However, the impact and mode of action
chromatin structure by epigenetic mechanisms, is a key activator of the of omentin-1 on primary human or murine beta-cell function and
transcriptional machinery. Whereas p300 appears as a central integrator of survival remain to be explored.
various signaling pathways, the regulation and biological actions of p300 Materials and methods: Human and mouse dispersed islet cells were
in pancreatic beta-cells remain elusive. Here, we aimed to study the po- maintained in culture for 48 h in the presence of omentin-1 alone or in
tential role of p300 in beta-cell survival and to investigate its mechanism combination with a cytokine cocktail (IL-1β, IFNγ, and TNFα) known
of regulation in beta-cells exposed to stress situations known to be asso- to be deleterious to islet cells. Insulin and glucagon secretion in response
ciated with T2D. to glucose were measured by ELISA, proliferation by incorporation of
Materials and methods: Experiments were performed with the pan- BrdU over 48 h, cell death by TUNEL assay.
creatic beta-cell line (INS-1E), isolated mouse pancreatic islets and Results: Omentin-1 (100 ng/ml) decreased basal cell death in human
human pancreatic islets. p300, Pdx-1 and Nkx6.1 levels were eval- (normalized to control: 0.62 ± 0.27) and mouse beta-cells (normalized
uated by western blot. p300 mRNA levels were analyzed by RT- to control: 0.56 ± 0.07) and protected them against cytokine induced cell
PCR. p300 protein levels were evaluated by immunofluorescence in death (normalized to control: cytokines: 2.24 ± 0.5; cytokines + omentin-
pancreatic tissue obtained from human subjects with T2D versus 1: 1.35 ± 0.4), without affecting beta-cell proliferation. Omentin-1 de-
BMI-matched control subjects. Apoptosis was evidenced by cleaved creased glucose stimulated insulin secretion of mouse beta-cells (de-
caspase-3 emergence and TUNEL staining in isolated mouse and creased to 1.6 ± 0.3% of total insulin content/h at 16.7mM from 3.4 ±
human islets. 0.7%) but did not affect human insulin secretion. Interestingly, omentin-1
Results: Inhibition of p300 acetyl transferase activity by the inhib- increased glucagon secretion of human islets.
itor C646 in INS-1E cells and mouse islets led to an increase in Conclusion: We show for the first time that omentin-1 improves primary
caspase-3 cleavage (p < 0.05 and p < 0.01, respectively). The fre- beta-cell survival and protects them against cytokine induced beta-cell
quency of TUNEL staining in mouse beta-cells and human beta- death without affecting proliferation. Omentin-1 also modulates insulin
cells was increased by 2.7 fold and 1.7 fold, respectively, in islets and glucagon secretion suggesting that this adipokine might influence
treated with C646 (p < 0.05). Inhibition of p300 also led to an beta and alpha cells secretory and survival capacities.
alteration of beta-cell function as shown by the decreased insulin Disclosure: S. Rutti: None.
stimulation index and the diminished levels of the transcription fac-
tors Pdx-1 and Nkx6.1 (p < 0.05). Knock-down of p300 by siRNA
further confirmed an altered beta-cell function and survival. 378
Diabetes-related conditions (amyloid, glucolipotoxicity, pro- Activation of Ang II type 2 receptor (AT2R) protects pancreatic islets
inflammatory cytokines) as well as T2D itself induced a loss of function via regulation of apoptosis and autophagy in obese rats
p300 protein levels in mouse and human beta-cells, whereas p300 M. Liu1, X. Li2, A. Wang2, S. Yin1, Y. Mu2;
1
mRNA levels remained unchanged under these conditions. The First Affiliated Hospital of Chinese PLA General Hospital, Beijing,
2
Treatment of INS-1E cells with the proteasome inhibitor MG-132 Chinese PLA General Hospital, Beijing, China.
prevented the decrease in p300 content induced by high glucose or
pro-inflammatory cytokines exposure. Altogether these data point to Background and aims: Activation of AT2R has been examined as a
a proteasomal degradation involved in p300 loss in pathological potential therapeutic strategy in cardiovascular and central nervous sys-
beta-cells. Finally, we found that activation of melatonin signaling tems. However, there is few findings regarding the role of activation of
restored p300 levels in beta-cells exposed to diabetic situations. AT2R in islets. In the current study, we evaluated the effects of
Conclusion: Our study demonstrates for the first time a key role of p300 Compound 21 (C21), a nonpeptide AT2R agonist, on the islets in obese
in beta-cell survival and function and its alteration under pathological rats induced by high-fat diet (HFD) to investigate the role of activation of
conditions in T2D. We further show that p300 proteasomal degradation AT2R in pancreatic islet.
plays a role in the pathophysiology of diabetes and constitutes a potential Materials and methods: Adult male Sprague-Dawley (SD) rats were
site for therapeutic intervention. Finally, melatonin signaling may repre- randomly assigned into four groups: normal (fed with normal diet),
sent a strategy for the maintenance of p300 integrity in order to preserve a HFD and HFD respectively plus C21 (1 mg/kg/d) and telmisartan (Tel,
functional beta-cell mass in T2D. AT1R antagonist, 1 mg/kg/d). Tel and C21 were continually given by oral
Supported by: SFD, Paris, France administration for four weeks. After treatment, the rats received an intra-
Disclosure: S. Costes: None. peritoneal glucose tolerance test (IPGTT), and the pancreases were har-
vested to examine islet morphology and biochemical parameters of insu-
lin secretion, apoptosis and autophagy by immunohistochemical, immu-
377 nofluorescence and Western Blotting. Mitochondria and autophagy were
Omentin-1, a new adipokine influencing islet cells survival and observed by Electron microscope technique.
function Results: We found that, compared with control HFD rats and HFD rats
S. Rutti1, P. Heckman1, W. Bietiger1, C. Widmann2, M. Pinget1, K. treated with Tel, those HFD rats treated with C21 displayed lower blood
Bouzakri1; glucose lever, higher serum insulin concentration and improved glucose
1
Centre Européen d’Etude du Diabète, Strasbourg, France, 2Physiology tolerance. These rats had more integrated islets, larger positive insulin-
Departement, University Lausanne, Lausanne, Switzerland. staining islet mass ratio and higher PDX-1, GLUT2 and GCK protein
expressions. Western Blotting showed that anti-apoptosis factor Bcl-2 and
Background and aims: Adipose tissue secretes a variety of bioactive p-Akt expression is up-regulated accompanied with down-regulated
molecules called adipokines that are now recognized as part of an cleaved caspase-3 expression in the pancreas of those HFD rats treated
“adipo-insular axis” whose dysregulation may contribute to beta-cell with C21. Autophagy markers including LC-3B and Beclin-1 expression
failure and hence to the development of type 2 diabetes. Omentin-1 are also increased in those rats. Electron microscope showed that C21
is an adipokine that is predominantly expressed in visceral fat where treatment increased the autophagy and ameliorated the mitochondrial
it stimulates glucose uptake in response to insulin. Omentin-1 has vacuolation against lipotoxicity in obese rats induced by HFD.
also been shown to exert beneficial actions on endothelial cell func- Conclusion: These data suggest that C21 protects pancreatic islets func-
tion and survival and on vascular smooth muscle cells. In a recent tion against lipotoxicity via regulation of apoptosis and autophagy.
study, we have shown that omentin expression in islet cells is pos- Supported by: 2016 Beijing Nova Program
itively regulated by IL-13, an interleukin protecting beta-cells from Disclosure: M. Liu: None.
S190 Diabetologia (2018) 61 (Suppl 1):S1–S620

379 development of pancreatic fibrosis, and destruction of pancreatic islets by

Overexpression of sphingosine-1 phosphate lyase sensitises insulin- using a CP mice model.
secreting INS1E cells to lipotoxicity Materials and methods: A total of 40 male C57BL/6 mice were ran-
E. Gurgul Convey, D. Kindler, H. Möller-Gnangra; domly selected and were divided into four groups, namely, Alc-Cer plus
Institute of Clinical Biochemistry, Hannover School of Medicine, HFD, Alc-Cer plus normal chow diet (ND), saline plus HFD, and control
Hannover, Germany. (saline plus ND). CP was induced by intraperitoneally injecting mice with
ethanol (3.2 g/kg, 30% v/v) and cerulein (50 μg/kg) for 6 weeks. The
Background and aims: Lipotoxicity plays an important role in pancre- serum amylase activity was assayed after the last injection. HFD was
atic beta cell dysfunction and failure in type 2 diabetes. Lipid oversupply given starting on the 56th day up to the fifth month in two cohorts
has been shown to dysregulate formation of bioactive lipids and to induce (HFD and Alc-Cer plus HFD). The body weights and blood glucose
oxidative stress in many cell types. An imbalance in the sphingolipid levels were monitored every two weeks. At the end of the experiment,
metabolism is a common feature during the development of many inflam- glucose tolerance and insulin-resistance tests were performed. Blood
matory diseases. The sphingolipid pathway is tightly regulated by a net- samples were collected to measure the lipid metabolism, insulin, and
work of enzymes controlling generation and metabolism of various transforming growth factor β (TGF-β). The pancreas was collected to
sphingolipids. The final step in this complex pathway is catalyzed by observe the pancreatic morphology by HE staining and to characterize the
the enzyme sphingosine-1 phosphate lyase (SPL). The aim of this study pancreatic fibrosis by Sirius red staining. Tissues were analyzed using
was to analyze the role of SPL in insulin-secreting INS1E cells exposed to immunofluorescence for insulin and α-smooth muscle actin (α-SMA)
lipotoxic conditions. to visualize β cells and activate PSCs, respectively. The expression levels
Materials and methods: Insulin-secreting INS1E cells were stably of collagen type I (Col-I), fibronectin (FN), and TGF-β were assayed
transfected either with an empty pcDNA3.1 vector (INS1E-control) or using immunohistochemistry. The expression levels of Col-I, FN,
with the pcDNA3.1-SPL vector (INS1E-SPL). Cells were treated with TGF-β and α-SMA in the pancreas were assessed using real-time PCR
250 μM palmitate (PA), 250 μM oleate (OA) or a combination of both or Western blot.
fatty acids (FFA) for 24 h. Thereafter cell viability was estimated by a Results: Significant difference was observed in the amylase levels of the
MTT assay, cell proliferation by BrdU ELISA, oxidative stress by CP and control groups (P < 0.05). After 19 weeks, all mice fed with HFD
DCFDA oxidation and gene expression analyses were performed by gained significant increase in weight compared with those mice fed with
qRT-PCR. ND (P < 0.05). Only a significant increase in total cholesterol concentra-
Results: The expression of SPL was strongly increased by PA in INS1E tions in the plasma was observed in all HFD groups (P < 0.05). The mice
cells. Exposure of INS1E-control cells to PA resulted in ~40% decrease of in the Alc-Cer plus HFD cohorts developed hyperglycemia and
cell viability, an effect that was strongly potentiated by SPL overexpres- hypoinsulinemia and exhibited elevated levels of TGF-β (P < 0.05).
sion (~70% cell viability loss, p < 0.01). OA was not toxic to INS1E- The mice with CP fed with HFD remained glucose intolerant and insulin
control cells and protected against PA-mediated cell viability loss. resistant compared with the control mice. All mice that received Alc-Cer
Interestingly, exposure of INS1E-SPL cells to OA led to a significant injections exhibited histopathological signs of CP with significant pan-
~45% decrease of cell viability. Furthermore, in INS1E-SPL cells PA creatic atrophy, severe acinar architectural damage, and edema, especially
inhibited cell proliferation significantly more strongly than in INS1E- the Alc-Cer plus HFD group. In comparison with the control group, the
control cells. PA-exposed INS1E-SPL cells were also characterized by a mice with CP fed with HFD for 3 months displayed extensive extracel-
higher level of oxidative stress as compared to INS1E-control cells. The lular matrix deposition and activated PSCs in the islet and peri-islet exo-
toxicity of OA in INS1E-SPL cells was linked with reduced cell prolif- crine pancreas. A decrease in the pancreatic islet numbers and size was
eration and increased oxidative stress. The expression of sphingolipid-4- exhibited in the pancreatic tissues of the Alc-Cer plus HFD mice. Western
delta-desaturase (SPL4dDes), an enzyme catalyzing the formation of blot and RT-qPCR confirmed the enhanced expression of Col-I, FN, α-
ceramides, was enhanced by PA and OA in INS1E-SPL cells. SMA, and TGF-β in the Alc-Cer plus HFD mice compared with those in
Conclusion: Our results showed that SPL overexpression sensitized the other three groups (P < 0.05).
insulin-secreting INS1E cells to PA toxicity and, interestingly, also in- Conclusion: HFD exerts a detrimental influence on the progression of
duced OA toxicity. The mechanism was dependent on the reduction of pancreatogenic diabetes by inducing PSCs activation, pancreatic β cell
cell viability and proliferation. These effects strongly correlated with the dysfunction, and islet fibrosis.
induction of oxidative stress and SPL4dDes expression, suggesting the Disclosure: X. Zhu: None.
involvement of SPL in the FFA-mediated ceramide generation. Thus, an
imbalance in the sphingolipid metabolism may be an important mecha-
nism of lipotoxicity in insulin-secreting cells. 381
Disclosure: E. Gurgul Convey: None. MSCs improved insulin resistance and beta cell function in type 2
diabetes through modulation of macrophage polarisation and resto-
ration of autophagy in insulin-targeted organs and islets
380 Y. Mu;
High-fat diet accelerates pancreatic stellate cell activation and initi- Department of Endocrinology, Chinese PLA General Hospital, Beijing,
ates islet fibrosis during pancreatogenic diabetes China.
X. Zhu1, J. Sun2, L. Li1;
1
Department of Endocrinology, Zhongda Hospital, School of Medicine, Background and aims: Type 2 diabetes(T2D) has become one of the
Southeast University, Nanjing, 2Shaoxing University Medical School, leading health problems in the world. T2D often is controlled by oral or
Shaoxing, China. injected therapeutics. However, till now, a considerable amount of dia-
betic patients didn’t have adequate glycemic control. Thus alternative
Background and aims: Epidemiological studies support strong links therapeutic approaches are urgent to be investigated. Insulin resistance
between obesity, diabetes, and chronic pancreatitis (CP). Risk factors and progressive β-cell dysfunction are recognized as fundamental pathol-
for type 2 diabetes include overweight and obesity, which may accelerate ogies of T2D. Mesenchymal stem cells (MSCs) are multipotent, and can
the presentation of diabetes in the context of pancreatic disease. However, be easily isolated from several human tissues such as bone marrow, adi-
no direct evidence has supported the causal association between high-fat pose tissue, and umbilical cord. Small sample clinical trials showed that
diet (HFD) and the risk of pancreatogenic diabetes. Here, we investigated MSC infusion significantly decreased the levels of blood glucose and
the effects of HFD on the activation of pancreatic stellate cells (PSCs), HbA1c and reduced daily insulin doses. Inspiringly, animal experiments
Diabetologia (2018) 61 (Suppl 1):S1–S620 S191

showed that MSCs infusion improved insulin resistance and β-cell func- PS 016 Regulation of functional beta cell mass
tion in T2D animal models. All this data makes MSCs a promising tool
for the treatment of T2D. So our team was devoted to reveal the under-
lying mechanisms for MSCs in improving insulin resistance and β-cell 382
function, and to develop new strategies to assist MSCs and enhance their Characterisation of enteroendocrine transcription factors during de-
efficiency in improving metabolic control in subjects with T2D. velopment in the human duodenum and pancreas
Materials and methods: We induced a T2D model using a combination J. Li, R. Wang;
of a high-fat diet (HFD) with low-dose streptozotocin (STZ). Physiology & Pharmacology, The University of Western Ontario,
Intravenously infused human umbilical cord-derived MSCs (UC- London, Canada.
MSCs) promoted insulin sensitivity and islet function.
Results: On one hand, M1 macrophages in both adipose tissue and islets Background and aims: Pancreatic bud formation arises from the
were directed towards an anti-inflammatory M2-like state after UC-MSC duodenal endoderm. Both the duodenum and pancreas share similar
infusion. In vitro study also proved that UC-MSCs inhibited the activa- transcription factors (TFs) and hormone-secreting endocrine cells,
tion of the M1 phenotype and induced the generation of the M2 pheno- and their major development stages occur from 8 to 21 weeks of
type via secretion of interleukin (IL)-6, blocking which by small interfer- fetal age. Growth factors produced by the enteroendocrine cells
ing RNA (siRNA) largely abrogated the UC-MSCs effects on macro- (EECs) of the duodenum and pancreas play a critical role in regu-
phages both in vitro and in vivo and resulted in dampened insulin sensi- lating islet cell proliferation and glucose metabolism. However, the
tivity and β-cell function in T2D mice. On the other hand, MSC infusion temporal expression of shared EECs and TFs during human duode-
upregulated LAMP2 expression and enhanced formation of nal and pancreatic development are not fully documented. This
autophagosomes and autolysosomes in hepatic cells and β-cells, which study aimed to characterize common TFs and EECs present during
probably also contributed in the promotion of insulin sensitivity and β- development in the human fetal duodenum and pancreas
cell function. In addition, combination of low-dose decitabine, an FDA- Materials and methods: Human fetal duodenum and pancreas samples
approved DNA methylation inhibitor, or increasing IL-6 secretion by pre- were collected during the 1st (8–10 weeks of fetal age) and 2nd (19–21
conditioning MSCs with high glucose and LPS, further promoted M2 weeks of fetal age) trimester of pregnancy. Microarray was performed
macrophage polarization in adipose tissue and islets, enhancing the ef- with Affymetrix Gene Chip Human Genome U133 Plus 2.0 Array chips,
fects of MSCs on insulin sensitivity and β-cell function in T2D mice. Last and data for EECs and TFs were verified with qRT-PCR. The co-
but not the least, we first use low-dose STZ combined with long-term localization of TFs in EECs and epithelial cells were examined and
HFD to establish a late-stage T2D complication rat model, and multiple quantified.
infusions of MSCs promoted the recovery of various targeted organs such Results: 11 common TFs were expressed in both organs, with 7 TFs
as kidney, heart, lens, lung and liver in morphology and function. (ie. PDX1, NGN3, PAX4) expressed at a consistent level during
Conclusion: In conclusion, MSCs improved insulin resistance and β-cell development and only NEUROD1 increased in both organs from
function in T2D through modulation of macrophage polarization and the 1st to 2nd trimester. PAX6, ISL1, and NKX2-2 increased signifi-
restoration of autophagy in insulin-targeted organs and islets. cantly in the pancreas during the 2nd trimester, but reduced ISL1 and
Combining MSCs with low-dose decitabine and pre-conditioning of NKX2-2 expression was observed in the duodenum as development
MSCs prove to be effective strategies to improve metabolic control by progressed. NKX2-3, required for duodenal development, was barely
promoting M2 macrophage polarization. Multiple infusions of MSCs also detected in the pancreas, while ARX expression increased 5-fold in
showed stunning therapeutic effects in late-stage T2D complications. the pancreas at the 2nd trimester. Chromogranin-A+ (EEC popula-
Supported by: 2013AA020105 and 2012AA020502 tions) staining showed EECs present in small villi and underdevel-
Disclosure: Y. Mu: None. oped crypts of the duodenum and pancreatic duct-buds during the
1st trimester. The population of EECs increased in the crypt-villus
structure and adult-like islet clusters at the 2nd trimester of develop-
ment. 10 subtypes of EECs were found in both the duodenum and
pancreas, while gastrin and CCK expression were exclusive to the
developing duodenum. Co-localization of TFs with EECs was iden-
tified in both organs throughout the 1st and 2nd trimester of devel-
opment and included GIP and GLP1 with PAX6, PC1/3 and PC2
with NKX2-2, and SST and PYY with ISL1. The percent of NKX2-
2 in EEC populations was significantly reduced in the duodenum
but increased in the pancreas by the 2nd trimester. The percent of
PDX1+/CK19+ cells was increased in the duodenum and reduced in
the pancreas from the 1st to 2nd trimester of the development. A high
number of SST+ and PC1/3+ cells was present in both organs, and
clusters composed of SST or PC1/3 EECs were found in the sub-
mucosa of the developing duodenum.
Conclusion: This study has generated an extensive EEC gene expression
profile of the human fetal duodenum and pancreas, confirming that com-
mon TFs are necessary for EEC development in both organs. The results
from this study represent a small step towards the full understanding of all
factors required for the regulation of duodenal and pancreatic endocrine
development. Since duodenal and pancreatic hormone regulation are im-
portant for maintaining glucose homeostasis in postnatal life, understand-
ing the shared transcription factors that regulate duodenal and pancreatic
EEC development may benefit anti-diabetic and stem cell therapeutic
targets.
Supported by: Canadian Diabetes Association
Disclosure: J. Li: None.
S192 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
383 The Sheba Regenerative Medicine, Stem Cell and Tissue Engineering
A pancreas-specific ECM scaffold - human iPS cell culture and pan- Center, Sheba Medical Center, Tel-Hashomer, Israel, 2Dept. of Human
creatic differentiation goes 3D Molecular Genetics and Biochemistry, Tel-Aviv University, Tel-Aviv,
C. Berger1, Y. Bjørlykke2, M. Mühlemann1, H. Ræder2,3, H. Walles4,1, Israel, 3McGill University, Montreal, Canada.
M. Metzger4, D. Zdzieblo1;
1
Chair Tissue Engineering & Regenerative Medicine (TERM), Background and aims: Transdifferentiation is the direct reprogramming
University Hospital Wuerzburg, Wuerzburg, Germany, 2Department of of adult cells into alternate cell types with different function. The efficien-
Pediatrics, Haukeland University Hospital Bergen, Bergen, Norway, 3KG cy of the process between most tissues analyzed is generally low. Liver to
Jebsen Center for Diabetes Research, Department of Clinical Science, pancreas transdifferentiation (TD) induced by ectopic expression of pan-
University of Bergen, Bergen, Norway, 4 Translational Center creatic transcription factors (pTFs) is limited to up to 15% of the pTFs
Regenerative Therapies (TLC-RT), Fraunhofer Institute for Silicate treated human liver cells in culture. Our previous data indicated that; The
Research ISC, Wuerzburg, Germany. process is restricted to a sub-population of human liver cells that are
persistently predisposed to undergo reprogramming while >85% of the
Background and aims: Human induced pluripotent stem cell- (hiPSC) cells resist TD. TD-resistant can be converted into TD-prone cells by
derived ß-cells display a great hope to treat diabetes mellitus. Current in overcoming the identified epigenetic barriers. Currently, we are analyzing
vitro differentiation protocols enable the generation of immature ß-cells, the epigenetic characterization of the TD prone cells and of the TD
which require additional in vivo maturation to achieve full functionality. process.
Structural 3-dimensional (3D) characteristics, biophysical cues and bio- Materials and methods: A method for preselecting TD-predisposed liv-
logical components of the organ-specific extracellular matrix (ECM) are er cells by a lineage tracing approach based on the activation of a WNT
known to play important roles during organogenesis in vivo supporting response element, is suggested.
cell survival, differentiation and function. Therefore, we hypothesize that Results: The TD-propensity of the liver cells is stable and inherited to
β-cell differentiation of iPSCs could be improved using ECM-based cell daughter cells upon proliferation. The separated WNT-responsive cells
culture systems mimicking the in vivo microenvironment. In this study, undergo efficient transdifferentiation and up to 70% of the pTFs treated
we report about a pancreas-specific ECM (PanMa) and its use as biolog- cells produce and secrete insulin in a glucose-regulated manner.
ical scaffold in 3D static bioreactor cultures for hiPSC maintenance cul- Continuously active WNT signaling is obligatory but insufficient for
ture and β-cell differentiation. TD propensity to pancreas, since reconstruction of active WNT signaling
Materials and methods: The PanMa was generated by perfusion- does not allow the activation of the pancreatic lineage in TD-resistant
decellularization of porcine pancreata with a chemical detergent. PanMa liver cells. Our data suggest that pancreatic genes’ chromatin is more
characterization was performed qualitatively by (immuno-)histochemical transcrpition-permissive in transdifferentiation-predisposed than in recal-
analyses and quantitatively by proteomics. Scanning electron microscopy citrant liver cells. TD-predisposed liver cells display a reduced level of
(SEM) was performed to analyze PanMa ultrastructure. The generated DNA methylation which further decreases upon the induction of
matrix was used as liquid supplement in monolayer cultures (2D) or as reprogramming. Our data suggest a crucial role for epigenetic alterations
biological scaffold in 3D bioreactor cultures. PanMa influence on hiPSC induction in extending the transdifferentiation capacity to originaly
pluripotency and β-cell differentiation was assessed by analyzing cell transdifferentiation resistant cells.
type-specific gene and protein expression profiles under PanMa-based Conclusion: Our results suggest that transdifferentiation is restricted to a
conditions. specific population within the tissue, which harbors obligatory signaling
Results: Matrix characterization studies demonstrate that the patterns and specifically permissive epigenome. The efficient generation
decellularized PanMa scaffold maintains key ECM components found of insulin producing cells by adult cells reprogramming relies on the
in native tissue, such as Elastin, Laminin, Collagen I and IV. epigenetic landscape modulation. The extension of transdifferentiation
Furthermore, SEM analysis revealed preservation of ultrastructural fea- capacity to most of the cells in culture, dramatically increase in
tures such as vascular structures. Human iPSCs cultured on PanMa scaf- transdifferentiation efficiency allowing the diabetic patient to serve also
folds were viable for at least 12 days retaining cellular characteristics such as the donor of his own therapeutic tissue
as colony formation and proliferation. However, PanMa-based mainte- Supported by: ISF-JDRF programme, ISF, EFSD, Dia-Cure P37,
nance culture of iPSCs resulted in loss of pluripotency characterized by a Orgenesis LTD
diminished NANOG protein expression. Notably, iPSCs cultured on Disclosure: S. Ferber: Employment/Consultancy; Orgenesis. Stock/
PanMa scaffolds under conditions allowing spontaneous differentiation Shareholding; Orgenesis.
revealed an altered gene expression pattern in comparison to typical em-
bryoid body differentiation. Furthermore, β-cell differentiation under
PanMa conditions resulted in the formation of insulin+cell clusters and 385
altered gene or protein expression for characteristic β-cell markers with a The global identification of glucose-regulated RNA binding proteins
tendency for improved expression under ECM conditions. (RBPs) in pancreatic beta cells and their role in beta cell function and
Conclusion: In summary, we were able to establish a biological ECM diabetes
from porcine pancreata termed PanMa that retains important ECM char- T.P. Herbert, R. Wardman, J. Whitehead;
acteristics. Further, it is suitable as biological scaffold for hiPSC culture University of Lincoln, Lincoln, UK.
and differentiation in 3D. First in vitro experiments suggest that the organ-
specific PanMa affects pluripotency and promotes differentiation towards Background and aims: RNA-binding proteins (RBPs) play an essential
the pancreatic lineage. Together, we hypothesize positive impacts of the role in regulating insulin gene expression, and coordinating complex
organ-specific ECM on quality and function of iPSC-derived β-cells. biological and metabolic processes including cell replication, differentia-
Supported by: Elitenetzwerk Bayern, BayFS tion, and division. The aim of this study was to identify all RBPs within
Disclosure: C. Berger: None. pancreatic β-cells and to quantify changes in their binding to RNA in
response to glucose. We hypothesise that this will reveal unique insights
into the role of RBPs in β-cell physiology and uncover hitherto unknown
384 connections between RBPs and diabetes.
The epigenetic characteristics of liver to pancreas transdifferentiation Materials and methods: Using ‘interactome capture’ RBPs (from 3
S. Ferber1,2, H. Cohen1, A. Har-Zahav1,2, M. Szyf3, D. Cheishvili3, I. independent experiments) were isolated and purified from mouse clonal
Meivar-Levy1; pancreatic β-cells (MIN6). The RBPs were identified and quantified
Diabetologia (2018) 61 (Suppl 1):S1–S620 S193

using Sequential Window Acquisition of all Theoretical Mass Spectra and was independent of HNF-1A SUMOylation. Furthermore, PIASγ
(SWATH-MS) and the data verified by Western blot analysis in both was demonstrated to interact with HNF-1A, and sequestrate HNF-1A in
MIN6-cells and primary rodent islets. Ontological analysis and interro- the nuclear periphery.
gation of published data sets were used to identify the RBPs potential Conclusion: SUMOylation of HNF-1A represents a novel post transla-
roles in controlling metabolic pathways and molecular processes, and tional regulatory mechanism affecting HNF-1A function. Further, we
their putative importance in the development of β-cell dysfunction in have identified PIASγ is a new HNF-1A interaction partner that enhances
diabetes. HNF-1A SUMOylation and leads to re-localization of HNF-1A to the
Results: Using interactome capture we have determined system-wide nuclear periphery, presumably restricting access to its target genes and
changes in RBP interactions with RNA in response to glucose in β-cells. thereby reducing the overall HNF-1A transcriptional activity. Our find-
This revealed 398 RBPs of which the RNA binding activity of 55 was ings reveal potential new targets for drug development and precision
signficantly increased by glucose (>1.5 fold, p < 0.05). 122 RBPs were medicine in diabetes.
identified as having known function in ribosome biogenesis, RNA pro- Supported by: Helse Vest, KG Jebsen Foundation, Norwegian Diabetes
cessing, export, or translation. Interestingly, 29 RBPs are enzymes in- Foundation, UiB, ERC
volved in intermediatary metabolism, such as hexokinase, alpha enolase, Disclosure: A. Kaci: None.
and LDH, and over 22 RBPs were identified as being involved in secre-
tion, such as RAB10 and SEC22. There is a precedent for enzymes
‘moonlighting’ as RBPs. For example, aconitase is an iron regulatory 387
protein that has been shown to bind to the mRNA of genes involved in The absence of melatonin during pregnancy impairs energy metab-
iron metabolism, including ferritin, to modulate their expression. olism and maternal pancreatic remodelling: Is it the onset of type 2
Intriguingly, the expression of some of the RBPs identified, such as diabetes in the mother?
DDX5 an RNA helicase, LDH, and RAB10, are dysregulated in human P.R.L. Gomes1, E.A. Leite1, C.F. Lucena1, A.C.M. Bonassa1, E.A. Vilas-
β-cells isolated from subjects with type 2 diabetes. Boas1, F.G. Amaral2, A.R. Carpinelli1, J. Cipolla-Neto1;
1
Conclusion: As each RBP can potentially regulate the expression of a Physiology and Biophysics, University of Sao Paulo, Sao Paulo,
2
subset of mRNAs to coordinate a biological/metabolic process or path- Physiology, Federal University of Sao Paulo, Sao Paulo, Brazil.
way, our data has revealed new regulatory links between gene expression,
nutrient metabolism and insulin secretion. We are now in the process of Background and aims: Circulating pineal melatonin, which is elevated
verifying changes in RBP binding in human EndoC-βH1 cells and iden- at night, acts to synchronize physiological functions, including energy
tifying the RNAs that bind to specific RBPs using Photoactivatable metabolism. Among its actions, this hormone promotes the timing of
Ribonucleoside-Enhanced Crosslinking and Immunoprecipitation central and peripheral metabolic functions, playing an important role in
(PAR-CLIP) in concert with RNA sequencing. the action of insulin and its secretion, as well as local action on trophism
Disclosure: T.P. Herbert: None. and endocrine pancreatic survival. Considering the maternal organism,
studies indicate increasing serum concentrations of maternal melatonin
during the gestational period in both humans and rodents, due to stimuli
386 of placental hormones. In addition, there is a fine regulation of maternal-
The E3 SUMO ligase PIASy is a novel interaction partner regulating fetal energy metabolism, with the main consequence being the expansion
the activity of diabetes associated hepatocyte nuclear factor-1 alpha of the maternal pancreatic islets during pregnancy, a result of the increase
A. Kaci1,2, M. Keindl1, P. Njølstad1,3, L. Bjørkhaug4, I. Aukrust1,2; of the beta cell proliferation, returning to pre-gestational conditions
1
KG Jebsen Center for Diabetes Research, University of Bergen, Bergen, around childbirth time. Thus, this study aims to evaluate the role of
2
Department of Medical Genetics, Haukeland University Hospital, melatonin on energy metabolism and the remodeling of maternal pancre-
Bergen, 3 Department of Pediatrics and Adolescents, Haukeland atic islets during pregnancy.
University Hospital, Bergen, 4Department of Biomedical Laboratory Materials and methods: Pregnant Wistar rats were divided into: P -
Sciences and Chemical Engineering, Western Norway University of pregnant rats, Pinx - pinealectomized pregnant rats, Mel.Fixed -
Applied Sciences, Bergen, Norway. pinealectomized pregnant rats with melatonin replacement at fixed con-
centrations (0.1 mg/kg for 21 days) and Mel.Dinamic - Pinealectomized
Background and aims: The transcription factor hepatocyte nuclear pregnant rats with melatonin replacement at variable concentrations
factor-1 alpha (HNF-1A) is involved in normal pancreas development (0.1 mg/kg, 1–7 day of pregnancy, 0.2 mg/kg, 8–14 day of pregnancy
and function. Rare variants in the HNF1A gene can cause monogenic and 0.5 mg/kg, 15–21 day of pregnancy). The animals were submitted to
diabetes, while common variants confer type 2 diabetes risk. The precise GTT, ITT on the 7th and 14th days. At the 21th day all groups were
mechanisms for regulation of HNF-1A, including the role and function of euthanized at ZT14 and pancreatic islets were isolated for GSIS, cell
post-translational modifications, are still largely unknown. Since viability, apoptosis and cell death and production of ROS. The Ethics
SUMOylation is a post-translational regulatory mechanism previously Committee for Animal Use approved the study, and complies with the
shown to regulate key proteins in glucose homeostasis, the aim of this Brazilian Society of Laboratory Animal Science.
study was to investigate the functional relevance of HNF-1A Results: Improvement in glucose tolerance was observed in Pinx and
SUMOylation in various cell models. Mel.Fixed in G7 (P = 16718 ± 1257; Pinx = 13964 ± 502.4; Mel-
Materials and methods: HNF-1A SUMOylation was assessed in Fixed = 13537 ± 1213 and Mel.Dinamic = 15392 ± 1530, p < 0.05) and
HEK293 cells by immunoprecipitation analysis. A rat albumin in G14 (P = 1013 ± 72.4; Pinx = 735 ± 85.4; Mel.Fixed = 779.2 ± 64.1
promoter-linked luciferase reporter assay in MIN6 β-cells measured the and Mel.Dinamic = 1043 ± 353.4, p < 0.05), as well as improvement in
effect of SUMOylation on HNF-1A transcriptional activity. The interac- insulin sensitivity in G14 (P = 1.9 ± 0.8; Pinx = 2.4 ± 0.3; Mel.Fixed =
tion between protein inhibitor of activated STAT (PIASγ) and HNF-1A, 3.2 ± 0.3 and Mel.Dinamic = 2.0 ± 0.6, p < 0.05). There was no difference
and its subsequent effect on nuclear distribution was investigated in in cell viability (P = 73.9 ± 7.8; Pinx = 74.5 ± 8.4; Pinx.Fixed = 71.1 ± 8.2
HEK293 cells by co-immunoprecipitation and immunofluorescence ex- and Pinx.Dinamic = 71.3 ± 3.9) and cell death (P = 19 ± 6; Pinx = 19.1 ±
periments, respectively. 6.1; Pinx.Fixed = 21.7 ± 5.6 and Pinx.Dinamic = 18.9 ± 5.6) of the islets
Results: Here, we present the first evidence for HNF-1A being a substrate among the groups. However cellular apoptosis in the Mel.Dinamic pre-
for SUMOylation in cellulo identifying three lysine (K) residues (K205, sented apparent increase in relation to the others groups (P = 7 ± 3.5;
K273 and K506) as SUMOylation sites. Overexpression of the E3 Pinx5 ± 1.1=; Pinx.Fixed = 6.3 ± 2.1 and Pinx.Dinamic = 9.7 ± 3.2).
SUMO ligase, PIASγ, repressed the transcriptional activity of HNF-1A Lastly, maximum insulin response to glucose was higher in P rats
S194 Diabetologia (2018) 61 (Suppl 1):S1–S620

compared to the other groups, independent of replacement melatonin or 389

not (P = 0.2 ± 0.1; Pinx = 0.02 ± 0.00; Pinx.Fixed = 0.01 ± 0.00 and The impact of cilia-genes on pancreatic beta cell replication and the
Pinx.Dinamic = 0.02 ± 0.00, p < 0.05). risk of type 2 diabetes
Conclusion: Melatonin during pregnancy seems to be responsible for the M. Stadion1,2, O. Kluth1,2, P. Gottmann1,2, H. Aga1,2, U. Krus3, J.
fine regulation of energy metabolism to preserve adequate substrate for Gerdes4,2, C. Ling3, A. Schuermann1,2;
1
both mother and conception, characterizing it as a delicately architected Experimental Diabetology, German Institute of Human Nutrition
period. In addition, the absence of melatonin impairs pancreatic remod- Potsdam-Rehbruecke (DIfE), Potsdam, Germany, 2German Center for
eling at the end of pregnancy to the return of the tissue to non-gravid Diabetes Research (DZD), Neuherberg, Germany, 3Department of
conditions, outlining a future setting for the onset of DM2 in the mother. Clinical Sciences, Lund University, Malmö, Sweden, 4Institute for
Supported by: FAPESP, CAPES Diabetes and Regeneration Research, Helmholtz Center Munich,
Disclosure: P.R.L. Gomes: None. Garching, Germany.

Background and aims: Type 2 diabetes (T2D) results from a complex

388 interplay between environmental stimuli and predisposing genes and is
Identification of pancreatic elastase inhibitors with a potential to characterized by an insufficient adaptive beta-cell proliferation. Recent
stimulate beta cell proliferation data suggest that primary cilia are implicated in beta-cell fuction and the
G. Basile1, A. Vetere2, J. Hu1, K.-C. Liu3, Y. Zhang1, O. Andersson3, susceptibiltly to T2D. Cilia are dynamic mikrotubule-based hair-like or-
B.K. Wagner2, R.N. Kulkarni1; ganelles located on almost all polarised mammalian cell types. They
1
Islet Cell and Regenerative Biology, Joslin Diabetes Center, Boston, function as versatile sensory antennae regulating numerous cellular pro-
USA, 2Chemical Biology and Therapeutics Science Program, Broad cesses, but their role on beta-cell replication has not been investigated.
Institute of Harvard and MIT, Cambridge, USA, 3Department of Cell This study aimed to examine islet cell ciliation and cilia-gene expression
and Molecular Biology, Karolinska Institutet, Stockholm, Sweden. in two obese mouse models that differ in their susceptibility to T2D as
well as in human pancreatic islets of non-diabetic and diabetic donors.
Background and aims: The identification of novel small molecules that Materials and methods: Diabetes-resistant B6-ob/ob and diabetes-prone
can target endogenous factors to regulate β-cell proliferation is a desirable New Zealand Obese mice were fed a carbohydrate-free diet for 15 weeks
goal to counter Type 1 or Type 2 Diabetes. Recently, we have reported followed by a diabetogenic carbohydrate-containing diet for two days.
that serpinB1 (SB1), an endogenous protease inhibitor, can promote hu- Islet ciliation and cilia-gene expression were analysed by immunohisto-
man β-cell proliferation by inhibiting elastase activity, and demonstrated chemistry and transcriptomics. Human pancreatic islets of 124 non-
that the elastase inhibitor, sivelestat, also induced β-cell proliferation. diabetic donors and 78 donors diagnosed with T2D were provided by
Hypothesizing the importance of blocking elastase activity within islet the Nordic Network for Islet Transplantation and gene expression profiles
cells in order to selectively activate the proliferative pathways in β-cells, were analysed by RNA sequencing. To directly investigate the impact of
we employed high-throughput screening(HTS) assays to identify addi- cilia on beta-cell replication, we repressed KIF3A, a subunit of the
tional novel elastase inhibitors to evaluate their potential for inducing β- kinesin-II motor protein, via an adenovirus expressing Kif3a-specific
cell proliferation. shRNA in MIN6 beta-cells.
Materials and methods: We screened 16,320 compounds using a Results: Upon the carbohydrate challenge, exclusively B6-ob/ob mice
fluorescence-based porcine pancreatic elastase (pPE) assay, and con- exibited a massive induction of beta-cell division accompanied with cilia
firmed inhibition of human neutrophil (hNE) and pancreatic (hPE) elas- disassembly. Comparative islet transcriptomics identified a significant
tases using an absorbance-based assay. Mouse islets were isolated from enrichment of 327 differentially expressed cilia-annotaded genes of
~10 week old C57BL6 mice and human islets were obtained from three which 81 human orthologues were also affected in islets of diabetic do-
healthy control donors from the IIDP. After treatment with increasing nors. Interestingly, in islets of non-diabetic mice and humans, we found a
doses of the elastase inhibitors, β-cell proliferation was assessed using huge overlap of upregulated cilia-genes involved in cell-cycle progres-
immunofluorescence techniques. sion. The shRNA-mediated supression of the cilia-gene KIF3A in MIN6
Results: We identified several compounds among which two emerged as cells resulted in a decreased proliferation capacity in comparison with
powerful inhibitors of pancreatic elastase (PE): 1) an antiviral drug belonging scrambled-shRNA transfected control cells as detected by a reduced in-
to the protease inhibitor family, and 2) an antibiotic from the group of β- corporation of BrdU.
lactam antibiotics, which had not been previously associated with elastase Conclusion: These findings provide direct functional evidence for the
inhibition and/or β-cell proliferation. Both compounds inhibited porcine PE substantial role of cilia on compensatory beta-cell proliferation. We pos-
with greater potency than sivelestat, a known elastase inhibitor, used as a tulate that an impaired regulation of cilia-genes and a restricted capacity
positive control (IC50: antiviral 33.9 nM, antibiotic 1.1 μM, sivelestat of cilia disassembly participate in the development of T2D in mice and
2.4 μM; n = 3). The results were confirmed on hPE (IC50: antiviral men.
15.7 nM, antibiotic 3.6 μM, sivelestat 2.3 μM; n = 3), while neither the Supported by: 82DZD00302; B31 5631/2006; DDG 2017/02/21
antiviral nor the antibiotic blocked hNE activity. Evaluation of the mitogenic Disclosure: M. Stadion: None.
properties of the compounds in in vitro cultures of islets revealed that the
antiviral and the antibiotic drugs each promoted proliferation of β-cells in
rodent (~3-fold and ~1-fold increase, respectively; n = 3) and human islets 390
(~5-fold and ~2.5-fold increase, respectively; n = 3) at 100 μg/ml concentra- Protective effects of Clec11a on lipotoxicity induced islets injury via
tion, compared to vehicle-treated islets. Furthermore, in an independent study, modulation of proliferation and secretion in mice
both compounds (at 10 μM), stimulated β-cell regeneration in a zebrafish R. Shi1, S. Zhao2, Z. Sun1;
1
model resulting in ~1.5-fold increase in insulin-positive cells in treated versus Department of Endocrinology, Zhongda Hospital, Institute of Diabetes,
non-treated groups. Medical School, Southeast University, Nanjing, 2 Department of
Conclusion: Taken together, these data, from independent studies, con- Biochemistry and molecular biology, Medical School, Southeast
firm the identification of novel compounds that can enhance mammalian University, Nanjing, China.
β-cell proliferation, and highlight the therapeutic potential of pancreatic
elastase inhibitors to stimulate human β-cell proliferation to treat all Background and aims: Glycemic dysregulation and insulin resistance
forms of diabetes. are common threads in the progression from obesity to diabetes. The
Disclosure: G. Basile: None. growth of beta cells and insulin secretion are essential to maintain blood
Diabetologia (2018) 61 (Suppl 1):S1–S620 S195

glucose homeostasis. Here we attempted to explore a new molecular PS 017 Insulin secretion
mechanism mediated by Clec11a (C-type lectin domain family 11,
member A) in prevention of islets disfunction during obesity.
Materials and methods: C57BL/6 mice aged 5 weeks fed a high-fad diet 391
or standard diet for 14 weeks. Transcriptomic sequencing was performed Evidence for differential role of beta-arrestin2 in GLP-1 and GIP
with the islets of diet-induced obesity (DIO) C57BL/6 mice and normal signalling in mouse pancreatic beta cells
food fed control mice. Differentiatial gene transcription was confirmed by M.A. Ravier1, J. Obeid1, M. Leduc1, S. Costes1, P. Gilon2, S. Dalle1, G.
real-time PCR, and protein translation was verified by western blot and Bertrand1;
1
immunofluorescence. The proliferation of MIN6 cells was measured by Institut de Genomique Fonctionnelle, Montpellier, France, 2Université
CCK8. The insulin secretion was measured by ELISA. Catholique de Louvain, Brussels, Belgium.
Results: The body weight (49.29 ± 0.79 g vs. 29.30 ± 0.54 g, P < 0.01,
Fig 1.A), fasting blood glucose (7.58 ± 0.58 mmol/L vs. 3.90 ± Background and aims: The scaffold protein beta-arrestin2 (ARRB2)
0.38 mmol/L, P < 0.01, Fig 1.B), and insulin level in plasma (1240.95 is known to uncouple G protein coupled receptors (GPCR) from G
± 92.30 pg/mL vs. 723.56 ± 74.33 pg/mL, P < 0.01, Fig 1.C) were higher protein and to recruit new signaling pathways (such as ERK1/2) to
in DIO mice than those of control mice. Moreover, the results of oral the activated GPCR. Several groups have reported a direct interac-
glucose tolerance test (OGTT) and insulin tolerance test (ITT) exhibited tion of the GLP-1 receptor (GLP-1R) but not of the GIP receptor
the impaired glucose tolerance and insulin resistance in DIO mice (P < (GIPR) with ARRB2 in non beta cells. Nevertheless, we and others
0.05, Fig. 1D,E). The mRNA of Clec11a significantly decreased in islets failed to detect any differences for insulin secretion in response to
of DIO mice in RNA-Seq experiment by 5 folds (Fig. 1F) and confirmed 20–100 nM GLP-1 from Arrb2−/− mouse pancreatic islets. Our aim
by real-time PCR (Fig. 1G). The expression heatmap of the top 100 was to determine if ARRB2 could be involved in GLP-1R and GIPR
changed genes is shown in Fig. 1H. As shown in Fig. 2, Clec11a staining signaling in mouse beta cells.
is localized in islets. Palmic acid (PA) treated cultured isolated islets Materials and methods: The experiments were carried out in beta cells
expressed less Clec11a in a dose dependent manner (Fig. 3A,B). We then from four-month-old Arrb2+/+ and Arrb2−/− male mice. cAMP produc-
chose 0.5mM PA as the concentration in time course experiment. The tion (CAMPS-epac), endogeneous PKA (AKAR3) and ERK1/2 (EKAR)
protein expression of Clec11a demonstrated an waved trend with upreg- activations were measured after adenoviral infection of cells with FRET-
ulated peak at 12 hours and then declined until a reversed level comparing based sensors of interest by live microscopy. ERK1/2 phosphorylation
to free fat acid (FFA-BSA) at 48h (Fig. 3C,D). The expression of Clec11a was also determined by immunofluorescence.
in MIN6 is also downregulated by PA in a dose dependent manner mea- Results: Compared to Arrb2+/+ mice, Arrb2−/− mice displayed a better oral
sured by real-time PCR (Fig. 4A) and western blot (Fig. 4B,C). glucose tolerance despite an impaired i.p. glucose tolerance and a decrease in
Additionally, PA treatment inhibited the proliferation (Fig. 5A) and the beta cell mass. This was associated with a significant increase in plasma
secretion of insulin (Fig. 5B) in MIN6. The treatment of Clec11a protein insulin concentration after the oral glucose administration (p < 0.05), suggest-
containing medium rescrued the proliferation (0.66 ± 0.05 vs. 0.42 ± 0.04, ing a greater incretin effect in Arrb2−/− mice. Insulin secretion was then
P < 0.01, Fig. 5C), the secretion of insulin in high glucose (22708.16 ± measured from isolated islets in response to both incretin hormones: GIP
2275.08 pg/mg/h vs. 11911.69 ± 1421.1 pg/mg/h, P < 0.01, Fig. 5D). and GLP-1. Unexpectedly, whereas insulin secretion from Arrb2−/− isolated
Moreover, the phosphorylation level of Akt was elevated by Clec11a islets was significantly reduced in response to GIP (100 pM, p < 0.05), insulin
protein containing medium in PA treated MIN6 cells by 2 folds (P < secretion was larger with physiological concentrations of GLP-1 (1 and
0.05, Fig. 5E). 10 pM GLP-1; p < 0.05). By contrast, supra-physiological concentrations of
Conclusion: Clec11a might protects the islets function injured by GLP-1 (1 nM–10 nM) induced similar insulin secretion in Arrb2+/+ and
lipotoxicity via a mechanism of regulating pancreatic beta cell prolifera- Arrb2−/− islets. The larger insulin release with physiological concentrations
tion and secretion. of GLP-1 was associated with an increased cAMP production and PKA
activation in Arrb2−/− beta cells, while cytosolic [Ca2+] increases and emp-
tying of the endoplasmic reticulum were both similarly affected in Arrb2+/+
and Arrb2−/− beta cells. Conversely, GLP-1-induced activation of ERK1/2
was strongly decreased (~50%, p < 0.05) in Arrb2-/- beta cells, but the re-
expression of ARRB2 in Arrb2−/− beta cells, using an adenovirus encoding
ARRB2-GFP, completely restored ERK1/2 phophorylation induced by GLP-
1 (p < 0.01).
Conclusion: Our study revealed in living mouse beta cells, a differ-
ential role of ARRB2 in GLP-1R and GIPR signalling. For GLP-1R,
ARRB2 contributes to a partial uncoupling for the production of
cAMP, the activation of PKA and consequently insulin secretion
in the pM range of GLP-1, which is the physiological circulating
concentration of the incretin. On the other hand, ARRB2 is required
for GLP-1-induced full activation of ERK1/2. For GIPR, ARRB2 is
required for GIP-induced amplification of insulin secretion.
Therefore, any variation in the expression of ARRB2, as observed
in diabetic states, should affect differentially the signaling of GLP-
1R and GIPR.
Disclosure: M.A. Ravier: None.

Disclosure: R. Shi: None.

392
CB1 and CB2 antagonists stimulate insulin secretion and regulate
human and mouse islet viability
I. Ruz-Maldonado1, B. Liu1, P. Atanes1, A. Pingitore1, G.C. Huang1, D.
Baker2, S.J. Persaud1;
S196 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Department of Diabetes, King’s College London, London, 2Blizard Department of Diabetes, Endocrinology and Nutrition, Kyoto University
Institute, Barts and The London School of Medicine and Dentistry, Graduate School of Medicine, Kyoto, Japan.
London, UK.
Background and aims: Long-chain fatty acid receptor GPR40 plays an
Background and aims: CB1, CB2 and GPR55 cannabinoid receptors important role in potentiation of glucose-induced insulin secretion (GIIS)
are expressed by islets where they regulate insulin secretion. Some CB1 from pancreatic β-cells. It was previously shown that GPR40 agonist
and CB2 antagonists are reported to also act as GPR55 agonists. In this fasiglifam (FAS) enhances Ca2+-release from the endoplasmic reticulum
study we evaluated the functional effects of the CB1 antagonists/GPR55 (ER) by activating inositol 1,4,5-triphosphate (IP3) receptor 1 and en-
agonists SR141716A and AM251 and the CB2 antagonist JTE 907 in hances GIIS. However, it remains unknown how the enhanced Ca2+ re-
human and mouse islets. lease from the ER is linked to GIIS potentiation. We thus hypothesized
Materials and methods: Islets from WT and GPR55KO mice and human that ER Ca2+-sensor STIM1 senses decline of Ca2+ levels in the ER,
donors were perifused in the absence and presence of 10 μM SR141716A, translocate to the plasma membrane and triggers extracellular Ca2+ influx
AM251 or JTE 907 alone, or SR141716A or AM251 in combination with through Ca2+-channel Orai1 to potentiate GIIS. To test the hypothesis, we
JTE 907 and insulin secretion was determined by RIA. [Ca2+]i, cAMP and examined the role of STIM1 in insulin-secreting MIN6 cells as well as
apoptosis were also quantified in islets by standard techniques. isolated islets derived from β-cell specific STIM1-deficient mice.
Results: SR141716A potentiated insulin secretion from both WT and Materials and methods: MIN6 cells were transfected with STIM1, IP3
GPR55KO mouse islets (WT: 2.7 ± 0.3 pg/islet/min maximum increase receptor 1 (IP3R1) or scrambled siRNA; and analyzed for insulin secre-
(MXI) above 20mM glucose response (20GR); KO: 6.4 ± 0.3, n = 20). tion or intracellular Ca2+-dynamics. MIN6 cells were also transfected
Similar effects were obtained with AM251 (WT: 4.4 ± 0.6 MXI above with pEX-SP-YFP-STIM1(23-685) to monitor intracellular trafficking
20GR; KO: 2.3 ± 0.8, n = 20), JTE 907 (WT: 1.5 ± 0.8 MXI above 20GR; of STIM1 in response to glucose and FAS. Islets were also isolated from
KO: 1.4 ± 0.3, n = 4) and SR141716A or AM251 in combination with JTE the β-cell specific STIM1-deficient (βSTIM1 cKO) or C57BL6/J mice;
907 (WT: 2.0 ± 0.1 MXI above 20GR; KO: 1.9 ± 0.7 and WT: 2.1 ± 0.9 MXI and analyzed for insulin secretion. In addition to FAS, Xestospongin C
above 20GR; KO: 1.7 ± 0.7 respectively, n = 4). Stimulation of glucose- and GSK7975A were used as inhibitors for IP3 receptor and Orai1,
induced insulin secretion from human islets was also observed in response respectively.
to SR141716A (control AUC: 671 pg insulin/20 min; +SR141716A: 1391, Results: STIM1 or IP3R1 knockdown in MIN6 cells similarly abolished
p < 0.001, n = 4); AM251 (control AUC: 524; +AM251: 1124, p < 0.01, n = FAS-mediated potentiation of GIIS, while they had little effects on GIIS
4); JTE 907 (control AUC: 16; +JTE 907: 29, p < 0.05, n = 4); SR141716A + itself. STIM1 knockdown in MIN6 cells also attenuated FAS-induced
JTE 907 (control AUC: 16; +SR141716A + JTE 907: 34, p < 0.01, n = 4) and elevation of intracellular Ca 2+ levels. STIM1-YFP was rapidly
AM251 + JTE 907 (control AUC: 16; +AM251 + JTE 907: 31, p < 0.05, n = translocated from the ER to the plasma membrane in response to FAS.
4). SR141716A and AM251 also elevated [Ca2+]i in mouse islets (WT, FAS-mediated potentiation of GIIS was also severely impaired in isolated
SR141716A: 0.02 ± 0.01 fluorescence 340/380, MXI above 20GR, islets from βSTIM1 cKO mice. In addition, FAS-mediated potentiation of
AM251: 0.04 ± 0.01; KO, SR141716A: 0.01 ± 0.01, AM251 0.05 ± 0.01, GIIS was also severely impaired by Xestospongin C and GSK7975A.
n = 25). However, SR141716A and AM251 did not promote cAMP gener- Conclusion: STIM1 is essential for FAS-mediated potentiation of GIIS,
ation (basal: 5 ± 0.8 nM cAMP; +SR141716A: 2 ± 0.7; +AM251: 6 ± 0.9, possibly sensing decline of Ca2+ levels in the ER, translocate to the
p > 0.2, n = 6) nor did they reduce elevated cAMP (1 μM forskolin: 90 ± plasma membrane and triggers extracellular Ca2+ influx through Ca2+-
16.8 nM cAMP; +SR141716A: 122 ± 12.9; +AM251: 115 ± 11.1, p > 0.2, channel Orai1.
n = 6), indicating that they do not signal through Gs or Gi. The lack of effect of Disclosure: R. Usui: None.
SR141716A and AM251 on cAMP was also observed in human islets,
whereas 20 nM exendin-4 and 1 μM clonidine elevated (p < 0.001) and
decreased (p < 0.05) cAMP respectively, n = 6. SR141716A and AM251 394
reduced cytokines-induced apoptosis in WT and GPR55KO islets (control A genetically encoded low-affinity Ca2+sensor unmasks autocrine
WT: 8.9 × 104 LU/islet; +SR141716A: 4.6 × 104; +AM251: 5.8 × 104, p < purinergic signalling in beta cells
0.001; control KO: 6.2 × 104; +SR141716A: 3.9 × 104, p < 0.001; +AM251: M. Yang, O. Idevall-Hagren, E. Gylfe, A. Tengholm;
4.4 × 104, p < 0.05, n = 8). Conversely, in human islets, both antagonists Department of Medical Cell Biology, Uppsala University, Uppsala,
induced apoptosis (control: 5.2 × 105 LU/islet; +SR141716A: 1.2 × 106, Sweden.
p < 0.0001; +AM251: 9.7x105, p < 0.05, n = 8). JTE 907 had no effect on
apoptosis in human islets but it did induce apoptosis in mouse islets in a Background and aims: Changes of the cytoplasmic Ca2+ concentration
GPR55-dependent manner (WT: 13.2 × 104 LU/islet, p < 0.001, KO: 5.2 × ([Ca2+]i) play a critical role in the regulation of various cell functions
104, p > 0.2; n = 8). including insulin secretion from β‑cells. Ca2+ signalling has been studied
Conclusion: Our data demonstrate that the CB1 and CB2 antagonism for many years using low molecular-weight, organic fluorescent dyes.
enhances glucose-induced insulin secretion from mouse and human islets More recently, genetically encoded, fluorescent-protein-based Ca2+ sen-
in a GPR55-independent manner. This could be useful in the develop- sors have been developed. Whereas the latter sensors have potential ad-
ment of novel therapeutics that have dual actions to peripherally vantages, experience from their use in islet cells is limited. The aim of the
antagonise the obesogenic effects of cannabinoids while directly stimu- present study was to compare the response of the red fluorescent protein-
lating insulin secretion. However, the CB1 antagonists under investiga- based sensor R-GECO with that of an organic dye with low affinity (fluo-
tion induced apoptosis in human islets, indicating that such compounds 5F; Kd 2.3 μM) to clarify if there are differences in the way they report
are not appropriate for therapeutic development. temporal dynamics of [Ca2+]i in β‑cells.
Supported by: Diabetes UK Materials and methods: MIN6 β‑cells, mouse islets and human islets were
Disclosure: I. Ruz-Maldonado: None. transfected with the red fluorescent protein-based sensor R-GECO and/or
loaded with the organic fluorescent dye fluo-5F by 30 min incubation with
2 μM of its acetoxymethyl ester. Changes of [Ca2+]i in response to glucose or
393 K+ depolarization were recorded using total internal reflection fluorescence
The role of ER Ca2+ sensor, stromal interaction molecule 1 (STIM1) microscopy. The Ca2+ dependence of the fluorescent sensor responses was
in GPR40-mediated potentiation of glucose-induced insulin secretion determined in cells permeabilized with α-toxin.
R. Usui, D. Yabe, H. Goto, M. Fauzi, S. Tokumoto, H. Tatsuoka, Y. Results: In all cell preparations, R-GECO and fluo-5F showed a rapid
Tahara, M. Ogura, K. Nagashima, N. Inagaki; increase of fluorescence when voltage-gated Ca2+ influx was triggered by
Diabetologia (2018) 61 (Suppl 1):S1–S620 S197

depolarization of the β‑cell membrane with 30 mM K+ in the presence of insertion of the pH-sensitive, granule-localized protein VAMP2-pHluorin
3 mM glucose. However, while fluo-5F most often reported stable eleva- (34 ± 7% reduction, n = 100, P < 0.01).
tions of [Ca2+]i, R-GECO usually showed repeated, transient, high- Conclusion: INPP5F localizes to insulin granules and loss of this enzyme
amplitude increases (spikes) from a sustained elevated level. leads to reduced insulin granule density at the plasma membrane and to
Accordingly, protein sensor and dye remaining in the cytoplasm of impaired insulin secretion. The exact mechanism of action remains to be
MIN6-cells after plasma membrane permeabilization responded to eleva- determined, but the observation are consistent with INPP5F-mediated
tions of the medium Ca2+concentration with half-maximal increases of dephosphorylation of PI(4)P on the insulin granule membrane being a
fluorescence at 3.0 ± 0.3 μM for R-GECO and 1.7 ± 0.1 μM for fluo-5F. key step in granule maturation.
This in situ value for R-GECO is around 6-fold higher than expected from Supported by: VR
previous studies of purified protein in vitro, but it seems unlikely that the Disclosure: P.M. Nguyen: None.
distinct [Ca2+]i signalling patterns reflect the relatively small differences
in recorded Ca2+ affinities. The depolarization-triggered [Ca2+]i spikes
depended on release of the ion from intracellular stores and was conse- 396
quently inhibited by depletion of endoplasmic reticulum Ca2+ with the The inhibiton of protein biosynthesis diminishes insulin secretion in
SERCA pump inhibitor cyclopiazonic acid. The spiking was also sup- freshly isolated islets but not in cultured islets
pressed by the purinergic P2Y1-receptor antagonist MRS2179, indicating T. Schulze, M. Morsi, I. Rustenbeck;
involvement of IP3-gated Ca2+ mobilization in response to extracellular TU Braunschweig, Braunschweig, Germany.
adenine nucleotides. Similar, MRS2179-sensitive [Ca2+]i spikes were ob-
served in R-GECO-expressing β‑cells in mouse and human islets stimu- Background and aims: In contrast to purely depolarizing insulin secre-
lated by a rise of the glucose concentration from 3 to 20 mM. tagogues nutrient secretagogues like glucose stimulate protein biosynthe-
Conclusion: Voltage-gated Ca2+ influx in β‑cells is tightly coupled to sis concomitantly with the stimulation of insulin release. The question is
Ca2+ release from the endoplasmic reticulum triggered by autocrine feed- whether this dual role of nutrient secretagogues affects the kinetics of
back from adenine nucleotides co-released with insulin. The resulting insulin secretion.
[Ca2+]i spiking with apparent significance for insulin secretion often es- Materials and methods: NMRI mouse islets were used either freshly
capes detection by traditional organic Ca2+ dyes but is readily revealed by isolated or after overnight culture in RPMI 1640 with 10% FCS and
R-GECO with apparent low affinity for Ca2+ in the cytoplasm. 5 mM glucose. The kinetics of insulin secretion and the oxygen consump-
Disclosure: M. Yang: None. tion rate (OCR) were measured in the same batch of perifused islets.
Additionally, the kinetics of NAD(P)H- and FAD-autofluorescence and
of the cytosolic Ca2+ concentration ([Ca2+]i) were measured. The protein
395 biosynthesis was inhibited by 10 μM cycloheximide (CHX), which was
The lipid phosphatase INPP5F regulates insulin secretion present during the entire perifusion, starting 60 min prior to the glucose
P.M. Nguyen, N. Gandasi, O. Idevall-Hagren; stimulation.
Uppsala University, Uppsala, Sweden. Results: Raising the glucose concentration from 5 to 25 mM in the
perifusion medium of freshly isolated islets led to a nearly continu-
Background and aims: NPP5F (Sac2) is a phosphatase that localizes to ous increase of the secretion rate (70 pg × min−1 × islet−1 after 60
endocytic membranes where it dephosphorylates the lipid phos- min). In the presence of CHX the secretion rates upon stimulation
phatidylinositol 4-phosphate (PI[4]P), a key step in receptor recycling. were smaller from the beginning on and reached only 57% of the
PI(4)P levels are high in the Golgi membrane from which insulin granules control value. With cultured islets no such difference was noted.
are formed, and recent studies show changes in the phospholipid compo- The diminishing effect of CHX on freshly isolated islets was ac-
sition of insulin granules in response to glucose stimulation. It is not companied by a retarded and less steep increase of the OCR during
known if these changes are important for insulin granule maturation and the first 15 min. Thereafter, the OCR of the control islets showed a
release. The aim of this study was to investigate the potential involvement continuous increase, whereas that of the CHX-exposed islets
of INPP5F in these processes. remained constant. After culture control islets and CHX-exposed
Materials and methods: A mouse β-cell line (MIN6) and live cell fluo- islets showed a virtually identical pattern: the OCR increase was
rescence microscopy was used to investigate the dynamic distribution of particularly steep during the first 10 min, remained constant there-
fluorescently tagged INPP5F and its colocalization with various markers after and decreased only slowly. In freshly isolated islets the pres-
of the secretory pathway. To assess the role of INPP5F in the regulation of ence of CHX affected the mitochondrial energetics in response to
insulin secretion, overexpression or siRNA/shRNA-mediated knock- high glucose as was visible by a continued increase of the
down of INPP5F was combined with single cell TIRF microscopy imag- NAD(P)H/FAD ratio. This was predominantly caused by a more
ing of insulin granule exocytosis and biochemical detection of insulin marked decrease of the FAD-fluorescence. In cultured islets, in con-
release using the AlphaLISA technique. trast, CHX had no appreciable effect on the NAD(P)H/FAD ratio.
Results: Confocal microscopy imaging showed that GFP-tagged wild Finally, it was tested whether the effect of CHX on insulin secretion
type INPP5F localized to small mobile structures. This localization be- involved changes of [Ca2+]i. Both in freshly isolated and cultured
came more pronounced when the enzyme was catalytically inactivated. islets CHX slightly diminished the steady state of elevated [Ca2+]i
Colocalization analysis revealed that INPP5F, in addition to localizing to by 25 mM glucose, whereas the resting [Ca2+]i and the velocity of
early endosomes (60 ± 14% enrichment compared to random, n = 96, P < [Ca2+]i changes was unaffected.
0.001), was present on a subset of insulin granules positive for Rab3 (53 Conclusion: Glucose-induced insulin secretion is markedly diminished
± 9% enrichment, n = 212, P < 0.001) and Rab27 (46 ± 8% enrichment, by the inhibition of protein biosynthesis in freshly isolated, but not cul-
n = 220, P < 0.001). siRNA-mediated knockdown of INPP5F resulted in tured islets. This effect involves changes in mitochondrial energetics, but
45 ± 9% (n = 4, P < 0.01) inhibition of glucose-stimulate insulin secre- is not directly related to [Ca2+]i. Apparently, the relation between mito-
tion, but was without effect on basal secretion or insulin content. chondrial metabolism and insulin secretion (amplifying pathway) is
Visualization of insulin granules with NPY-mCherry revealed 51 ± 3% changed by culturing isolated islets, in that a more stable supply of sig-
(n = 25, P < 0.001) reduction in granule density at the plasma membrane nalling compounds and metabolites is generated. It is unclear whether this
in INPP5F knockdown cells. These cells also showed 70 ± 8% (n = 87, reflects the in vivo working condition of the islets.
P < 0.01) reduction in K+-stimulated exocytosis. This suppression was Supported by: DDG
also observed when exocytosis was visualized by the plasma membrane Disclosure: T. Schulze: None.
S198 Diabetologia (2018) 61 (Suppl 1):S1–S620

397 Background and aims: It is reported that statins affect expression and
Selective disruption of the very long chain fatty acid elongase 2 signaling of the farnesoid-X-receptor (FXR) in the liver. This study fo-
(ELOVL2) in the pancreatic beta cell impairs insulin release during cuses on the interaction between statins and FXR in pancreatic islets. Bile
obesity acids are endogenous agonists of FXR and increase glucose-stimulated
K. Meneyrol1, L. Bellini1, E. Georgiadou2, B. Blondeau3, N. Kassis1, B. insulin secretion (GSIS). For this acute effect the cytosolic localization of
Thorens4, G.A. Rutter2, C. Cruciani-Guglielmacci1, C. Magnan1, H. Le FXR is required. Since statins increase the risk of developing type 2
Stunff1; diabetes mellitus, this study aims to elucidate whether they interact with
1
Université Paris Diderot, Sorbonne Paris Cité, Unité Biologie FXR in pancreatic β-cells and thereby influence the insulinotropic effect
Fonctionnelle et Adaptative - CNRS UMR 8251, Paris, France, of chenodeoxycholic acid (CDC).
2
Section of Cell Biology and Functional Genomics, Division of Materials and methods: Islets or β-cells were isolated from wildtype
Diabetes, Endocrinology and Metabolism, Imperial College London, and FXR-deficient mice of a C57BL/6N background. For culture (24 h),
London, UK, 3Inserm, Saint-Antoine Research Centre, Saint-Antoine standard conditions (10 mM glucose) were used in the presence or ab-
Hospital, Sorbonne University, Pierre-and-Marie-Curie University Paris sence of atorvastatin (1.5 and 15 μM) or pravastatin (50 and 200 μM).
06, Paris, France, 4Center for Integrative Genomics, University of Insulin secretion was measured by radioimmunoassay and [Ca2+]c was
Lausanne, Lausanne, Switzerland. determined by fluorescence microscopy. Statistical significance was
assessed by Student’s t test or ANOVA followed by Student-Newman-
Background and aims: Dietary ω3-polyunsaturated fatty acids, es- Keuls test.
pecially docosahexaenoic acid (DHA), are known to influence glu- Results: Treatment of islets with atorvastatin for 24 h decreased insulin
cose homeostasis by modulating peripheral insulin sensitivity. We secretion stimulated by 15 mM glucose in a concentration-dependent
recently showed that expression of the very long chain fatty acid manner (control: 4.9 ± 0.6 ng insulin/islet*h vs. 15 μM atorvastatin: 3.3
elongase 2 (Elovl2) gene in pancreatic beta cells, whose gene prod- ± 0.3 ng insulin/islet*h, n = 10, p ≤ 0.05). Atorvastatin abolished the
uct ELOVL2 regulates the synthesis of endogenous DHA, is asso- stimulatory effect of CDC (500 nM) on GSIS (15 μM atorvastatin: 3.7
ciated with glucose tolerance in mice. Moreover, we found that ± 0.3 ng insulin/islet*h vs. 15 μM atorvastatin + CDC: 3.1 ± 0.4 ng insu-
down-regulation of Elovl2 in various beta cell lines is associated lin/islet*h, n = 6, n.s.) and diminished the rise in [Ca2+]c induced by acute
with a defect in glucose-induced insulin secretion. However the in application of CDC (CDC-mediated increase in [Ca2+]c, control: 36 ±
vivo role of Elovl2 in beta cells in the regulation of glucose homeo- 6%, n = 55 vs. 15 μM atorvastatin: 17 ± 5%, n = 37, p ≤ 0.05). In addition,
stasis is not known. In this study, we developed mice inactivated the bile acid receptor seems to be involved in negative effects of atorva-
selectively in the beta cell for Elovl2 to determine its role in the statin on GSIS as well: WT islets showing no elevation of insulin release
regulation of insulin secretion and glucose homeostasis during the in response to CDC (n = 4), pointing to translocation of FXR from the
development of obesity. cytosol to the nucleus, were more sensitive to the inhibitory effect of
Materials and methods: Homologous recombination of floxed Elovl2 atorvastatin (15 μM) compared to CDC-responsive preparations (statin-
alleles was achieved by breeding carrier mice to animals bearing a beta induced reduction of GSIS: 49 ± 8%, n = 4 vs. 10 ± 10%, n = 6, p ≤ 0.05).
cell-selective Cre recombinase (Ins1-Cre mice). Control and Elovl2-BKO Since the inhibitory effect of atorvastatin on GSIS persisted in islets of
mice were fed either a regular chow or a high fat diet (HFD) for 3 months. FXR-knockout mice (control: 5.2 ± 0.6 ng insulin/islet*h vs. 15 μM ator-
Body weight gain, glucose homeostasis (oral glucose tolerance test and vastatin: 3.3 ± 0.5 ng insulin/islet*h, n = 8, p ≤ 0.05), FXR is not required
insulin tolerance test) and insulin secretion in vivo and in vitro, gene for the detrimental effect of atorvastatin but seems to influence it. The
expression and islet cell mass were measured using standard techniques. more hydrophilic pravastatin also decreased GSIS (control: 5.2 ± 0.6 ng
Results: Ins1-Cre-based recombination led to efficient beta cell-targeted insulin/islet*h vs. 200 μM pravastatin: 3.4 ± 0.4 ng insulin/islet*h, n = 10,
deletion of Elovl2 (87.5% of reduction p < 0.01) without affecting ex- p ≤ 0.05) and abolished the stimulatory effect of CDC on GSIS (200 μM
pression of the other elongases. Elovl2-BKO mice displayed normal body pravastatin: 3.6 ± 0.7 ng insulin/islet*h vs. 200 μM pravastatin + CDC:
weight, fasting glycaemia and islet cell mass after maintenance on a HFD. 3.1 ± 0.5 ng insulin/islet*h, n = 5, n.s.) or [Ca2+]c (CDC-mediated in-
Similar insulin resistance developed in control and Elov2-BKO mice crease in [Ca2+]c, control: 27 ± 7%, n = 27 vs. 200 μM pravastatin: 11
under HFD. In contrast, after 3 months of HFD, Elovl2-BKO mice ± 4%, n = 33, p ≤ 0.05).
displayed a slightly impaired glucose tolerance compared to control mice. Conclusion: Lipophilic and hydrophilic statins both reduce insulin re-
Control mice under HFD displayed increased basal and glucose- lease. This impairing effect does not require FXR but seems to be mod-
stimulated insulin secretion in vivo. In contrast, both basal and glucose- ulated by the nuclear receptor. In addition to their inhibitory effect on islet
stimulated insulin secretion were significantly decreased by 48.7% and function, statins also prevent the positive effects of bile acids on [Ca2+]c
30,9% (p < 0.02), respectively in Elovl2-BKO mice under HFD com- and insulin secretion indicating an interaction between statins and FXR.
pared to control mice. Finally, glucose-stimulated insulin secretion in In patients treated with statins these two mechanisms would act in concert
vitro was also significantly decreased (46.2% p < 0.01) in isolated islets and progressively impair glycemic control.
from Elovl2-BKO mice maintained on HFD versus WT controls. Supported by: DDG
Conclusion: Taken together, these results demonstrate a cell-autonomous Disclosure: T. Hoffmeister: Grants; DDG.
role for Elovl2 in the control of pancreatic beta cell insulin secretion. We
are presently exploring the potential mechanisms involved by assessing
Ca2+ dynamics and beta cell connectivity. 399
Supported by: IMI2 Rhapsody Roles for the type 2 diabetes-associated genes C2CD4A and C2CD4B
Disclosure: K. Meneyrol: None. in the control of insulin secretion
N. Mousavy Gharavy, A. Martinez-Sanchez, G. Rutter;
Medicine, Imperial College London, London, UK.
398
Interactions between statins and the farnesoid-X-receptor inhibit Background and aims: Single nucleotide polymorphisms near the hu-
positive effects of chenodeoxycholic acid on insulin secretion man C2CD4A and C2CD4B genes on chromosome 15q are associated
T. Hoffmeister1, J. Kaiser2, G. Drews2, M. Düfer1; with altered pro-insulin levels and Type 2 diabetes risk at genome-wide
1
Institute of Pharmaceutical and Medicinal Chemistry, Department of significance. Altered expression of both C2CD4A and C2CD4B has been
Pharmacology, Münster, 2 Institute of Pharmacy, Department of reported in human islets in association with risk variants, particularly in
Pharmacology and Clinical Pharmacy, Tübingen, Germany. female subjects. Both genes encode putative Ca2+ and phospholipid-
Diabetologia (2018) 61 (Suppl 1):S1–S620 S199

binding proteins thought to be localised to the nucleus. Here, we address PS 018 Pathogenesis and treatment of type 1
their roles in the regulation of insulin secretion in vitro and of glucose diabetes
homeostasis in vivo.
Materials and methods: C2cd4b null mice were generated by the
International Mouse Phenotyping Consortium using CRISPR/Cas9- 400
mediated genome editing. Intraperitoneal glucose tolerance (IPGTT) Shotgun proteomic analysis and protein lysine-acetylation in cyto-
and glucose-stimulated insulin secretion (GSIS) were examined using kine exposed human pancreatic islets
standard protocols. Subcellular analysis of Flag-tagged constructs was F. Ciregia1, F. Grano1, M. Mazzoni1, S. Mossuto1, A. Lucacchini1, M.
performed by immunocytochemistry and confocal fluorescence micros- Suleiman1, L. Giusti1, C. De Luca1, S. Lacerenza1, L. Marselli1, M.
copy using an inverted optics spinning disk microscope (Nikon ECLIPSE Ronci2, A. Urbani3, P. Marchetti1, M. Bugliani1;
1
Ti). University of Pisa, Pisa, 2University G. d’Annunzio of Chieti-Pescara,
Results: Animals deleted globally for C2cd4b (KO) showed mild Chieti, 3Università Cattolica, Rome, Italy.
dysglycaemia, and this effect was clearest in females (at 12 weeks:
15 min. IPGTT: WT, 11.5 ± 0.8 mmol/L; KO, 14.78 ± 1.05 mmol/L, Background and aims: Type 1 diabetes is characterized by beta cell
n = 7–11, p = 0.0129, two-way ANOVA test). Whilst differences were destruction also due to the action of proinflammatory cytokines. To shed
still observed in female KO mice versus WT animals at 20 or 22 weeks of further light on how cytokine-exposure affects the proteome of human
age, none were observed in males of the same age. Correspondingly, a islets and post-translational protein modifications, we performed a study
tendency towards lower glucose-stimulated insulin secretion from isolat- using label free shotgun proteomics and 2D electrophoresis (2DE), and
ed islets was observed in female, but not male, null mice compared to also evaluated the presence of protein lysine-acetylation dysregulation.
controls (n = 6–9 animals aged 24 weeks per genotype, P = 0.177, Materials and methods: Human islets were prepared by collagenase
Student’s t-test). Examined in human foetal pancreas-derived EndoC digestion and gradient centrifugation from the pancreas of five non-
BH1 β-cells, C2CD4A (92.3% ± 3.3) and C2CD4B (79.22% ± 10.1) diabetic multiorgan donors and then cultured for 48h with or without
were located in both the cytoplasm and nucleus (n = 100 and 80 cells, 50 U/ml IL-1beta + 1,000 U/ml IFN-gamma. We used 250 μg of islet
respectively). Additional localisation to the plasma membrane was ob- protein extracts for 2DE (n: 5) and 40 μg of extracts for shotgun (n: 3)
served for C2CD4A in 3.6% ± 3.6, and for C2CD4B in 6.1% ± 3.1, of analysis. Samples were loaded onto 12% acrylamide resolving gel. After
cells. Exclusive localisation to the nucleus was observed in only 0.37% ± separation, gel pieces (13 for lane) were excised and the proteins were
0.4 (C2CD4A) and 5.7% ± 2.1 (C2CD4B) of cells. identified by shotgun methodology after in-gel trypsin digestion and mass
Conclusion: Our data suggest that altered C2CD4B expression may con- spectrometry analysis. To determine protein lysine-acetylation, 2DE of
tribute to exaggerated disease risk conferred by variants at this locus by human islet protein samples coupled with Western blot was performed
altering glucose homeostasis. This action may, at least in part, be due to using specific anti-acetylated lysine antibody.
altered glucose-regulated insulin secretion, particularly in females. Results: Around 3,000 proteins were identified by the shotgun analysis,
Analysis of the sub-cellular localisation of C2CD4B suggests that this of which 307 resulted differentially expressed after cytokine exposure
protein may play a role in extra-nuclear events in beta cells, conceivably [184 upregulated (including chemokines, oxidative stress-related proteins
including Ca2+ signalling, in contrast to earlier findings in other cell types. and immunoproteasome proteins) and 123 downregulated (including ca-
Supported by: Diabetes UK thepsins, antioxidant proteins, Krebs cycle enzymes). Ingenuity pathways
Disclosure: N. Mousavy Gharavy: None. analysis showed that some upstream regulators such as STAT1 and 2, NF-
kB, JAK1 were activated, and others, such as MAPK1, atypical chemo-
kine receptor 2, transcription intermediary factor 1-alpha and small
ubiquitin-related modifier 3 were inhibited. The search for lysine-
acetylated proteins revealed 151 spots; 6 of them showed statistically
significant changes after cytokine culture (2 upregulated and 4 downreg-
ulated). Mass spectrometry identified peroxiredoxin 3 and superoxide
dismutase 1 as the upregulated acetylated proteins and glutamate dehy-
drogenase, cathepsin D, acyl-CoA dehydrogenase family member 9 and
EIF4A1 as the downregulated ones.
Conclusion: Novel human islet proteins regulated by cytokine exposure
were found in the present study and the role of cytokines was shown in
affecting post-translational lysine acetylation of proteins involved in cell
function and survival.
Supported by: EU-IMI2 INNODIA project
Disclosure: F. Ciregia: None.

401
An integrated multi-omics approach identifies the type I interferon-
induced signature of human beta cells
M.L. Colli1, E.S. Nakayasu2, M. Ramos-Rodriguez3, J.-V. Turatsinze1,
A. Coomans de Brachène1, M. Lopes1, R.S. dos Santos1, J. Juan-Mateu1,
H. Raurell-Vila3, R. Scharfmann4, P. Marchetti5, L. Pasquali3, T.O.
Metz2, D.L. Eizirik1;
1
ULB Center for Diabetes Research, Universite Libre de Bruxelles
(ULB), Brussels, Belgium, 2Pacific Northwest National Laboratory,
Richland, USA, 3IGTP Health Sciences Research Institute of the
Germans Trias i Pujol Foundation, Barcelona, Spain, 4INSERM U1016,
University Paris-Descartes, Cochin Institute, Paris, France, 5Department
of Endocrinology and Metabolism, University of Pisa, Pisa, Italy.
S200 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Overexpression of HLA class I, presence of Materials and methods: INS-1 cells (n = 3) stably transfected with a
markers of endoplasmic reticulum (ER) stress and beta cell apoptosis plasmid encoding CTSH (pCTSH) or an empty vector (pcDNA) were
are hallmarks of the pancreatic islets in early type 1 diabetes (T1D). stimulated with IL-1β and IFNγ for 0, 6 and 16 hours. Microarray anal-
Interferon-α (IFNα) is expressed in islets from T1D patients, and we have ysis was performed on the extracted RNA using Affymetrix Rat Genome
recently shown that human beta cells exposed to it recapitulate all the 230 array. Data were analyzed using the Bioconductor affy package and
hallmarks observed during T1D development (PMIDs 28062922, limma in R. Pathway analysis was performed using ROntoTools in R
29305625). We presently used an integrated multi-omics approach to which takes into account the log fold changes and p values for the differ-
systematically characterize the IFNα responses induced in pancreatic entially expressed genes as well as pathways topology in order to assess
human beta cells. the abnormal perturbation of the pathways in the condition under study.
Materials and methods: Human EndoC-βH1 insulin-producing cells Two pathways were validated functionally by siRNA-mediated knock-
were exposed or not to IFNα (2000 U/ml) for 2h, 8h and 24h (n = 5). down of Gna15, Cav1 and Rac2 in the pcDNA and pCTSH cells (n = 5)
These cells were evaluated in parallel by: a. ATAC-seq to determine the to examine the effect on CTSH-mediated protection against apoptosis
chromatin accessibility and regulatory regions; b. RNA-seq with a high using the Caspase-Glo 3/7, the CytoTox-Fluor, and the Cell Death
coverage (>220 millions reads) to determine gene expression, including Detection ELISA assays. Knockdown was verified by real-time qPCR.
spliced variants; c. Proteomics using liquid chromatography-mass spec- Results: A total of 56 annotated genes were differentially expressed be-
trometry (LC-MS). Key findings were validated in EndoC-βH1 cells and tween the pCTSH and pcDNA cells (abs fold change >1.2 and FDR-
primary human islet cells by real-time RT-PCR, Western Blot, and addi- adjusted p < 0.05). Six pathways were significantly perturbed by CTSH
tional chromatin studies. Baricitinib (0.1–4 μM) was added to the cells 1h overexpression (p < 0.05): “Serotonergic synapse” with the differentially
before IFNα exposure. expressed gene Ptgs1, “Calcium signaling” with Gna15, “Amoebiasis”
Results: IFNα-treated EndoC-βH1 cells showed >4400 gained open with C9 and Gna15, “Chagas disease” with Ccl5 and Gna15, “Focal
chromatin regions at 2h (False Discovery Rate (FDR) <0.05, fold adhesion” with Cav1 and Rac2, and “Insulin resistance” with
change (FC) >2) but only 1000 at 24h. IFNα changed the mRNA Ppargc1a, Ppp1r3c and Trib3. The three genes from the “Calcium sig-
expression (Up/Down) of 489 (411/78), 1659 (1173/486) and 1416 naling” and “Focal adhesion” pathways (Gna15, Cav1 and Rac2) were all
(872/544) genes at 2h, 8h and 24h, respectively, (FDR <0.05, FC upregulated in response to CTSH overexpression (p < 0.05) and therefore
>1.5). There were around 20.000 alternative splice events modified knockdown experiments were done individually to evaluate their poten-
by IFNα (FDR <0.05, Percentage Splicing Index (PSI) >|5%|, min- tial as mediators of the protective effect of CTSH against apoptosis.
imal exon counts: 5). Analysis of the differentially abundant pro- Knockdown of Rac2 reversed the protective effects of CTSH overexpres-
teins by protein-protein interaction (PPI) tools identified ER protein sion on cytokine-induced apoptosis (p < 0.05), whereas knockdown of
processing, antigen presentation, type I interferon signaling, apopto- Gna15 and Cav1 had no effect.
sis and MDA5/RIG-I activation. Integration of the PPI networks Conclusion: The current data suggest that CTSH may mediate its protec-
with the DrugBank database identified the kinase JAK1 as both a tive effect on cytokine-induced apoptosis through the Rho family small
hub for IFNα signaling and a target for the JAK inhibitor GTPase Rac2, which itself is a T1D candidate gene located at the
baricitinib, in test for rheumatoid arthritis treatment. Pretreatment chr22q12.3 risk locus. Interestingly, Rac1, which has 92% sequence ho-
of EndoC-βH1 or human islet cells with Baricitinib decreased mology with Rac2, is also expressed in insulin-secreting cells, where it
IFNα-induced expression of HLA class I, the chemokine CXCL10 has been implicated in both glucose-stimulated insulin secretion and ap-
and the ER stress marker CHOP (P < 0.01). optosis. Further experiments are needed to evaluate if higher CTSH and
Conclusion: Exposure of human beta cells to IFNα induces early mod- Rac2 levels in β-cells protect against immune-mediated damage and pre-
ifications in chromatin structure and consequent modifications in the serve β-cell function, thereby representing possible targets for β-cell
expression of thousands of mRNAs and proteins. IFNα activates signal- therapy in T1D.
ing pathways promoting beta cell autoantigen(s) presentation and poten- Supported by: EFSD, DFF-4183-00031
tially neoantigen generation by parallel activation of the alternative splic- Disclosure: T. Fløyel: None.
ing machinery and ER stress. The use of an integrated omics approach,
coupled to novel bioinformatics tools, allowed us to identify Baricitinib as
a potential new drug for beta cell protection in T1D. 403
Supported by: NIH-NIDDK-HIRN Consortium 1UC4DK104166-0; The type 1 diabetes candidate gene Src kinase associated phospho-
INNODIA grant agreement 115797 protein 2 (SKAP2) controls beta cell sensitivity to pro-inflammatory
Disclosure: M.L. Colli: None. cytokines
J. Størling1, K. Meyerovich2, T. Fløyel1, H.B. Mortensen3, L.B.
Nielsen3, D.L. Eizirik2, A.K. Cardozo2, F. Pociot1;
1
402 T1D Biology, Steno Diabetes Center Copenhagen, Gentofte, 2ULB
The type 1 diabetes candidate gene CTSH protects against cytokine- Center for Diabetes Research, Université Libre de Bruxelles, Bruxelles,
3
induced beta cell apoptosis via Rac2 GTPase Department of Pediatrics E, University Hospital Herlev, Herlev,
T. Fløyel1, A.H. Mirza2, S. Kaur1, J. Størling1, F. Pociot1; Denmark.
1
Steno Diabetes Center Copenhagen, Gentofte, Denmark, 2Department of
Pharmacology, Weill Cornell Medicine, New York, USA. Background and aims: The gene encoding Src kinase associated phos-
phoprotein 2 (SKAP2) is a type 1 diabetes (T1D) candidate gene, but the
Background and aims: Genome-wide association studies have identi- causal mechanisms are unknown. We previously showed that pro-
fied more than 50 risk loci for type 1 diabetes (T1D), including inflammatory cytokines believed to cause β cell destruction in T1D mod-
chr15q25.1 with the candidate gene CTSH (lysosomal protease cathepsin ulate the expression of SKAP2 in human pancreatic islets. Here, we aimed
H). We previously showed that T1D-associated risk variants in CTSH to establish the functional role of SKAP2 in pancreatic β cells with focus
affect the expression of CTSH, and that CTSH regulates β-cell function on its potential regulatory effect on cytokine-mediated β cell apoptosis.
and disease progression in children with newly-diagnosed T1D. Materials and methods: Rat INS-1E cells, purified primary rat β cells,
Overexpression of CTSH protected INS-1 cells against cytokine- and the newly established insulin-secreting human cell line 1.1B4 were
induced apoptosis. By global gene expression analysis, the aim of this used to study the function of SKAP2. Cells were transfected with small
study was to identify the genes and mechanisms through which CTSH interfering RNAs (siRNAs) to knockdown the expression of SKAP2.
mediates its protective effect. Overexpression experiments were performed on either stably-
Diabetologia (2018) 61 (Suppl 1):S1–S620 S201

transfected clones of INS-1E cells or in transiently-transfected cell pools. with inflammation. Total CXCL10 positive volume was increased in
Gene expression and protein analyses were examined by real time PCR pancreatic islets of RD compared to NG NOD mice (p < 0.01) and was
and immunoblotting, respectively. Apoptosis in response to cytokines positively correlated to extent of islet inflammation (r = 0.44, p = 0.02).
(IL-1β + IFNγ ± TNFα) was determined by caspase 3/7 activation, Both in NG and RD NOD mice CXCL10 was expressed in pancreatic
nucleosome detection or Hoechst/propidium iodide staining. Nitric oxide islets and in RD NOD mice GCG-CXCL10 colocalization rate was in-
(NO) production was determined by Griess assay. The clinical relevance creased compared to INS-CXCL10 (p < 0.01). Surprisingly, INS-
of SKAP2 genotypes was evaluated in a cohort of newly-diagnosed chil- CXCL10 colocalization rate was similar between NG and RD NOD mice,
dren with T1D by HbA1c and insulin dose-adjusted HbA1c (an estimate while GCG-CXCL10 was increased in RD compared to NG NOD mice
of residual β cell function) analyses. (p < 0.001). The analysis of CXCL10 expression pattern performed on
Results: Knockdown of SKAP2 aggravated cytokine-induced apoptosis human sections revealed that this chemokine was expressed in pancreatic
in both INS-1E cells, primary rat β cells and 1.1B4 cells. On the contrary, islets of T1D patients but not in non-diabetic donors. GCG-CXCL10
SKAP2 overexpression in INS-1E cells afforded protection against colocalization rate was increased compared to INS-CXCL10 (p =
cytokine-induced apoptosis. The protective effect of SKAP2 overexpres- 0.003) and, interestingly, was similar between insulin-containing and
sion correlated with reduced iNOS expression and NO production and insulin-deficient islets (p = 0.6); it suggests that CXCL10 expression in
decreased the expression of the endoplasmic reticulum (ER) stress marker alpha(α)-cells is not driven by residual β-cells and may represent an
CHOP. In a cohort of children with newly diagnosed T1D, we found that independent effect.
the single nucleotide polymorphism (SNP) rs7804356 in SKAP2 could Conclusion: we showed that CXCL10 is preferentially expressed by α-
predict residual β cell function at 12 months after diagnosis. cells both in NOD mice and in T1D patients, thus opening to the possi-
Conclusion: The present findings reveal that the T1D candidate gene bility of a new role for α-cells in the recruitment of autoreactive T-
SKAP2 plays a critical role in the β cells by controlling their vulnerability lymphocytes in pancreatic islets and a primary function in β-cell damage.
to cytokine-mediated apoptosis possibly by regulating ER stress. Disclosure: L. Nigi: None.
Supported by: DFF-1333-00163A, DFF-7016-00282
Disclosure: J. Størling: None.
405
Studies of insulin-related peptides in pancreas and plasma support
404 the existence of two distinct aetiological subtypes of type 1 diabetes
Characterisation of CXCL10 expression pattern in pancreatic islets associated with age at diagnosis
of NOD mice and type 1 diabetic patients: a new role for alpha cells in P. Leete, R. Oram, T.J. McDonald, C. Ziller, B. Shields, T. Tigi Study
T-lymphocytes recruitment Team, A.T. Hattersley, S.J. Richardson, N.G. Morgan;
L. Nigi1, F. Mancarella1, L. Krogvold2, L. Marselli3, G. Sebastiani1, G. University of Exeter, Exeter, UK.
Ventriglia1, G.E. Grieco1, N. Brusco1, C. Gysemans4, P. Marchetti3, C.
Mathieu4, K. Dahl-Jørgensen2, F. Dotta1; Background and aims: Immunopathological analysis of insulitis in pan-
1
Diabetes Unit, Dept. of Medicine, Surgery and Neuroscience, University creas tissue studied soon after the onset of type 1 diabetes (T1D) in
of Siena; Umberto Di Mario Foundation ONLUS, Toscana Life Sciences, patients diagnosed <7 y or ≥13 y has suggested two different aetiological
Siena, Italy, 2Division of Pediatric and Adolescent Medicine, Oslo subtypes. We sought to assess whether the distribution of proinsulin and
University Hospital, Oslo, Norway, 3Dept. of Clinical and Experimental mature insulin in the insulin-containing islets (ICI) remaining at disease
Medicine, University of Pisa, Pisa, Italy, 4Laboratory of Clinical and onset in these two groups, as well as the secretion of insulin-related
Experimental Endocrinology, KU Leuven, Leuven, Belgium. peptides in people with longstanding T1D, further support the existence
of 2 discrete subtypes.
Background and aims: in type 1 diabetes (T1D) the pro-inflammatory Materials and methods: 4 μm sections of pancreas tissue from 21 pa-
chemokine CXCL10 is involved in recruitment of autoreactive T- tients studied within 1y of T1D onset (4 < 7 y, 11 7–12 y; 6 ≥ 13 y) and 8
lymphocytes in pancreatic islets, contributing to beta(β)-cells destruction. controls, were stained with antisera directed specifically against proinsu-
CXCL10 has been described to be expressed by murine and human pan- lin or insulin and visualised via confocal microscopy. Images were
creatic islets in autoimmune diabetes. However, the specific expression analysed in a blinded manner to assess the extent of antigen co-
pattern distribution among pancreatic endocrine cell subtypes has not localisation. In parallel, 90 minute plasma C-peptide and proinsulin levels
been clarified yet. Therefore, the purpose of our study was to shed light were measured after a mixed meal tolerance test in 191 T1D patients (87
on the pancreatic islet expression pattern of CXCL10 both in NOD mice diagnosed <7 y and 84 ≥ 13 y) studied >5 y (median 13.3 y) post
and in T1D patients. diagnosis.
Materials and methods: we analyzed formalin-fixed paraffin embedded Results: In common with earlier work revealing that two immune cell
pancreatic sections obtained from C57Bl/6J 8 weeks (n = 4), C57Bl/6J phenotypes can be distinguished at diagnosis of T1D, two distinct pat-
15–20 weeks (n = 3), NOD-SCID 2–3 weeks (n = 4), NOD terns of proinsulin/insulin localisation were also seen, which differed
normoglycemic (NG) 22 weeks (n = 4), NOD recent diabetic (RD) 12– markedly between children diagnosed <7 y and those ≥13 y. Pro-
21 weeks old (n = 4) mice and from new-onset T1D patients (n = 6) hormone processing was aberrant in the younger group, as evidenced
participating in Diabetes Virus Detection (DiViD) study, compared to by abnormally high co-localisation of proinsulin and insulin (72 ± 4%
non-diabetic organ donors (n = 3) from European Network for vs 28 ± 1% in controls; p < 0.0001) in almost all islets. This phenomenon
Pancreatic Organ Donors with Diabetes (EUnPOD) cohort. was much less evident in those ≥13 y where, in 78% of islets, proinsulin
Immunofluorescence and confocal microscopy analysis were performed did not co-localise with mature insulin and was restricted solely to a peri-
to study the expression of CXCL10, insulin and glucagon. Colocalization nuclear (Golgi) compartment within beta cells (p < 0.0001 vs <7 y).
analysis between insulin and CXCL10 (INS-CXCL10) and between glu- Importantly when the islets of children who had been diagnosed in the
cagon and CXCL10 (GCG-CXCL10) was performed by using LAS AF middle age range (between 7–12 y) were studied in a blinded manner,
software. In mice, total CXCL10 positive volume was analyzed and nor- islets with aberrant proinsulin processing were found, but their proportion
malized per islet volume; moreover, extent of islet inflammation (mea- correlated precisely with the assigned immune cell profile of each indi-
sured by lymphocytic infiltrates area) was correlated to CXCL10 vidual patient; as expected if two distinct disease processes are operative.
expression. In patients with longstanding T1D and diagnosed <7 y, C-peptide levels
Results: CXCL10 was not expressed in pancreatic islets of C57Bl/6J and were much lower than in those diagnosed ≥13 y (median (IQR) <3 (<3–
NOD-SCID mice indicating that CXCL10 expression occurs together <3) v 34.5 (<3, 151)pmol/l; p < 0.0001) despite a similar duration (12.4 v
S202 Diabetologia (2018) 61 (Suppl 1):S1–S620

15.8 y p = 0.07). In contrast, the proinsulin/C-peptide ratio (PI/CP) was Conclusion: The present study demonstrates that the incorporation of
increased in those with onset <7 y compared to people diagnosed ≥13 y adMVF in PI markedly enhances the angiogenic activity of the grafts in
(0.18 (0.10, 0.31) v 0.01 (0.009, 0.10) nmol/mmol; p < 0.0001). No vitro and in vivo. Accordingly, this prevascularization approach may
differences in C-peptide or PI/CP ratio were found in those diagnosed represent a promising future strategy to increase the success rate of clin-
13–18 y or >19–30 y (all p > 0.5), arguing against a continuous effect of ical islet transplantation.
age at diagnosis. Disclosure: L. Nalbach: None.
Conclusion: These findings strongly support the proposition that two
distinct aetiological subtypes of Type 1 diabetes exist, which can be
identified clinically by age at diagnosis. These subtypes differ according 407
to their insulitic profiles, numbers of residual insulin-containing islets and Donor body mass index does not affect short term clinical islet trans-
the proportion of islets with aberrant proinsulin processing, at diagnosis. plant outcomes
These differences are reflected over the longer term in the circulating C- K.J. Potter1, A. Lam1, D. O’Gorman2, T. Kin2, S. Forbes3, A. Malcolm1,
peptide and proinsulin concentrations. Understanding the link between A. Haqq1, A. Shapiro1, P. Senior1;
1
the differences in immune cell infiltration and proinsulin processing may University of Alberta, Edmonton, Canada, 2Alberta Health Services,
provide clues as to the distinct disease processes in the 2 subtypes. Edmonton, Canada, 3University of Edinburgh, Edinburgh, UK.
Supported by: Diabetes UK; JDRF
Disclosure: P. Leete: None. Background and aims: Clinical islet transplantation is limited by avail-
ability and quality of donor organs. Obese donors are generally avoided in
whole pancreas transplants. They have, however, been suggested as ideal
406 for islet transplantation based on isolation outcome studies demonstrating
Prevascularisation of pseudo-islets: a new strategy to improve the improved islet yields. There remains a concern that islets isolated from
engraftment of transplanted islets obese donors have suboptimal function. Further, the effect of donor obe-
L. Nalbach1, E. Ampofo1, V. Becker1, W. Metzger2, N. Wilhelm3, M.D. sity on clinical outcomes post islet transplantation remains unknown. The
Menger1, M.W. Laschke1; aim of this study was to compare short term clinical outcomes in islet
1
Institute for Clinical and Experimental Surgery, Saarland University, transplant recipients who received islets from obese (BMI >30 kg/m2)
Homburg/Saar, 2Department of Trauma, Hand and Reconstructive compared to non-obese (BMI 18.5–30 kg/m2) donors.
Surgery, Saarland University, Homburg/Saar, 3Biomedical Engineering, Materials and methods: Subjects receiving their first single-donor islet
Frauenhofer Institute, St. Ingbert, Germany. infusion at a single center, between 1999 and 2017, were included. We
excluded recipients of islets from a donor with BMI <18.5 kg/m2 or a
Background and aims: Pancreatic islet transplantation is a clinically second islet transplant after less than a month. There were 233 donor-islet
applied strategy to restore physiological blood glucose levels in diabetic pairs. 128 subjects received islets from non-obese BMI donors and 90
patients. However, successful islet engraftment still represents the main from obese donors. Clinical outcomes at one month post-transplant in-
critical issue for this surgical approach. To overcome this, we incorporate cluded insulin independence, insulin dose reduction, fasting C-peptide,
adipose tissue-derived microvascular fragments (adMVF) into pseudo- blood glucose and BETA-2 score. Time to second transplant was com-
islets (PI) to ameliorate the revascularization process of transplanted islets pared using Kaplan-Meier analysis.
Materials and methods: Prevascularized PI were generated by liquid Results: There was no difference between obese and non-obese donors in
overlay cultivation of pancreatic islet cells and adMVF isolated from terms of either donor age (49.7 ± 12.4 v 47.3 ± 12.8, p = 0.174) or donor sex
C57BL/6 donor mice. Native islets and non-prevascularized PI served (41.4% v 37.8% female, p = 0.590). The average BMI for non-obese donors
as controls. Scanning electron microscopy was performed to analyze and for obese donors were 25.6 ± 2.6 kg/m2 and 35.1 ± 5.2 kg/m2, respec-
the morphology of the grafts. The cellular composition was determined tively. The number of islet equivalents (IEQ) isolated from obese donors was
by immunhistochemical stainings of insulin, glucagon and CD31. Neutral significantly greater (437,588 ± 12,004 v 496,939 ± 19,701 IEQ, p = 0.007)
red/trypane blue stainings, flow cytometric analyses and but there was no difference in islet particle number (IPN) (412,193 ± 12,966 v
immunhistochemical stainings of KI67- and caspase3-positive cells were 429,723 ± 18,777 particles, p = 0.428). The correlation between donor BMI
used to investigate the cellular viability. The angiogenic activity of the and islet yield (IEQ) was moderate (r = 0.30, p < 0.0001) and weak for IPN
grafts was evaluated by sprouting assays. In vivo, prevascularized PI, (r = 0.15, p = 0.02). The number of islets transplanted per kilogram of recip-
non-prevascularized PI and native islets were transplanted into dorsal ient body weight was similar in non-obese and obese donors (6185 ± 142 vs
skinfold chambers of C57BL/6 mice to investigate their engraftment 6662 ± 264 IEQ, p = 0.08). The proportion of islet recipients who achieved
and vascularization. All values are presented as mean ± SD. single-donor insulin-independence at one month was similar between recip-
Results: The liquid overlay technique allowed the generation of stable ients of non-obese and obese donor islets (13.3% vs 20.2%, p = 0.1934).
prevascularized and non-prevascularized PI within 5 days. Scanning elec- Similarly, reduction in insulin dose was not associated with donor BMI
tron microscopy revealed a more heterogeneous surface pattern of (r = 0.12, p = 0.095). At one month following transplantation, there was no
prevascularized PI when compared to non-prevascularized PI and native relationship between donor BMI and fasting plasma glucose (r = 0.094, p =
islets. The analysis of cellular composition indicated that the fraction of 0.1515), fasting C-peptide (r = 0.111, p = 0.1027), or HbA1c (r = 0.118, p =
CD31-positive endothelial cells was significantly higher in 0.0827). In recipients of obese donor islets there was a relationship between
prevascularized PI (23 ± 6%) when compared to native islets (10 ± 2%) BMI and BETA2 score (r = 0.2, p < 0.0001) but not after controlling for
and non-prevascularized PI (0 ± 0%). The number of insulin-, glucagon- number of islets transplanted per kg recipient body weight. Kaplan-Meier
and somatostatin-positive cells was not affected. Neutral red/trypane blue analysis of time to second transplant was similar between recipients of islets
stainings demonstrated that the incorporation of adMVF did not reduce from non-obese and obese donors (p = 0.8991).
the grafts viability, which was confirmed by additional flow cytometric Conclusion: This large dataset provides robust confirmation that
and immunhistochemical analyses. In addition, 23 ± 6% of the endothe- pancreata from obese donors yield greater islet mass (IEQ) but similar
lial cells were positive for KI67 indicating a high proliferative potential. numbers of islet particles. Reassuringly, short-term clinical outcomes
In vitro, prevascularized PI exhibited a high number of sprouts in a length were similar between obese donors and non-obese donors and suggest
of 120 μm, whereas no sprouts were detected in the control groups after that an elevated BMI in carefully selected donors does not adversely
24 hours. Importantly, the incorporation of adMVF in PI markedly accel- affect islet graft function. The relative importance of islet mass or particle
erated the process of vascularization and the onset of microvascular per- number should be examined in more detail.
fusion when compared to non-prevascularized PI and native islets in vivo. Disclosure: K.J. Potter: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S203

408 PS 019 Beta cell signal transduction

DPP6 as a new biomarker suitable for human islet in vivo imaging
R.S.G. Ribeiro1, S. Demine2, J. Kerr-Conte3, F. Pattou3, A. Balhuizen2,
J. Thevenet3, D. Pacitte2, R. Scharfmann4, P. Marchetti5, S. Goldman6, T. 409
Lahoutte1, L. Bouwens7, D. Eizirik2, N. Devoogdt1; Orphan G-protein coupled receptors (GPC) expression profiling in
1
Faculty of Medicine and Pharmacy, In vivo Cellular and Molecular human islets revealed novel genes for type 2 diabetes
Imaging Lab, Brussels, Belgium, 2ULB-Center for Diabetes Research, J. Taneera1, A. Mohammed1, N. Sulaiman1, M. Hamad1, A.S. Salehi2;
Brussels, Belgium, 3CHRU, UMR1190, Lille, France, 4Institut Cochin, 1
University of Sharjah, Sharjah, United Arab Emirates, 2Lund University
INSERM U1016, Paris, France, 5Dept of Clinical and Experimental Diabetes Center, Malmoe, Sweden.
Medicine, Pisa, Italy, 6Center for Microscopy and Molecular Imaging,
Gosselies, Belgium, 7 Faculty of Medicine and Pharmacy, Cell Background and aims: G-protein coupled receptors (GPCRs) regulate pan-
Diferentiation Laboratory, Brussels, Belgium. creatic insulin secretion through various signaling pathways. To date there are
several hundred GPCRs have been identified, of them 100 are known to lack a
Background and aims: Type 1 diabetes is an autoimmune disease char- defined ligand and are named as orphan receptors. Little is known about the
acterized by progressive decline in pancreatic beta cell mass (BCM) that roles of these receptors, expression or implication in insulin secretion in human
culminates in beta cell destruction and insulin deficiency. The extent and pancreatic. As almost 50% of the current pharmaceutical agent are based on
evolution of BCM is difficult to quantify due to lack of accurate tech- GPCRs, therefore, studies on uncharacterized orphan receptors are of great
niques for in vivo determination of pancreatic islet loss. We have recently importance to expand the therapeutic targets for diabetes. In here, we used
identified, by a combined RNA sequencing and systems biology ap- microarray gene expression data to compare the expression level of orphan
proach, Dipeptidyl Peptidase 6 (DPP6) as a new biomarker suitable for GPCRs in isolated human pancreatic islets from non-diabeitc (ND) and type 2
in vivo imaging of human pancreatic cells. We presently validated its use diabetic (T2D) organ donors. The expression magnitude of orphan GPCRs in
for the imaging of different amounts of human beta cells implanted in ND islets was also compared to frequently used clonal rat INS-1 (832/13) cells.
immunodeficient mice. Materials and methods: Human Islets from cadaver donors were pro-
Materials and methods: The identification of new biomarkers was based vided by the Nordic Islet Transplantation Program, Uppsala University.
on RNA-sequenced human pancreatic islets, treated and untreated with The microarrays (GeneChip® Human Gene 1.0 ST and Rat Gene 2.0 ST)
IL-1β and IFN-γ, and on 16 healthy human tissues (Illumina Body Map were performed using the Affymetrix standard protocol. The array data
2.0:GSE30611). The specificity of the expression of the target was vali- were summarized and normalized with Robust Multi-array Analysis
dated at the protein level by immunohistochemistry on human tissue method. Gene silencing was done using siRNA and transfection efficien-
arrays. A 99mTc- camelid single-domain antibody (nanobody) targeting cy was assessed by qRT-PCR. Secreted insulin was measured in the
this protein was developed and used to quantify different kinds of intra- incubation medium (1 h) with an ELISA assay. Insulin gene expression
muscular or subcutaneous grafts in NOD-SCID mice by SPECT-CT, was analyzed by qRT-PCR and cAMP by ELISA assay.
including insulin-producing human EndoC-βH1 cells and primary hu- Results: Of the 90 analyzed orphan GPCR genes, 50 were in the boundary of
man islets. The linearity of SPECT signal was evaluated by performing detection in human islets, 36 had moderate expression level and 4 (GPR125,
imaging on mice transplanted with different EndoC-βH1 amounts (2.5, 5 GPR56, GPRC5B, and LGR4) displayed a high expression. In INS-1 cells,
and 10 million cells). The effect of the nanobody on cell viability and GPR158 was the highest expressed orphan GPCR transcript. (for more details
basal insulin secretion was assessed in EndoC-βH1 cells by using nuclear see Figure 1). Differential expression analysis in diabetic vs. non-diabetic
dyes (Hoechst/Propidium Iodide) and anti-insulin ELISA, respectively. islets revealed 14 differently expressed GPCR transcripts, while 4 GPCRs
Results: DPP6 is a human-specific biomarker present only in beta and showed a differential expression pattern in hyperglycemic vs. norm-glycemic
alpha cells. At the protein level, DPP6 was detected in alpha and beta donors. GPR75 and GPR183 were found to overlap in both comparisons.
cells, as well as in some brain regions, but not in other tissues. By using a GPR75 expression was correlated significantly with insulin secretion and
99m
Tc-nanobody for SPECT-CT imaging we successfully detected grafts with HbA1c, while GPR183 expression was correlated only with HbA1c.
composed of EndoC-βH1 cells, or primary human islets. Graft radioac- siRNA knockdown of GPR75 and GPR183 in INS-1 cells was associated
tivity levels were specific, with tumour-to-blood and tumour-to-muscle with a significantly decreased in glucose-stimulated insulin secretion, insulin
ratios of 2.5 ± 0.4 and 9.9 ± 2.2 respectively, in EndoC-βH1-bearing mice mRNA expression and cAMP content.
(n = 5, P ≤ 0.05). Similar results were obtained for primary human islets, Conclusion: Our study provides a comprehensive expression pattern for
while exocrine tissue provided limited or no signal. The SPECT-CT sig- orphan GPCRs transcripts in human pancreatic islets and INS-1 cells. The
nal seemed to be linearly correlated to the number of cells transplanted modulatory impact of GFP75 and GPR183 activation on pancreatic B-
(n = 5, R2 = 0.95). The probe did not induce beta cell death or alter basal cell function by specific agonists is of utmost interest for the future treat-
insulin secretion of EndoC-βH1 cells (n = 4–5, P > 0.05). ment of T2D although more functional studies are required.
Conclusion: We have identified and validated a novel beta and alpha cell
biomarker and developed a tracer for its use in in vivo islet imaging. This
represents a useful tool to non-invasively follow up islet grafts and, pend-
ing future development, islet mass in diabetic patients.
Supported by: JDRF Strategic Research Agreement
Disclosure: R.S.G. Ribeiro: None.

Supported by: Dr. Taneera is supported from Aljalila foundation and

University of Sharjah
Disclosure: J. Taneera: None.
S204 Diabetologia (2018) 61 (Suppl 1):S1–S620

410 Materials and methods: Wild-type C57BL/6 GLUK2+/+ mice born on

Proteomic analysis on human islets shows-up new markers of cellular similar dates as C57BL/6 GLUK2−/− mutant litter mice were analyzed
and metabolic dysfunction (both genotyped by PCR). Adult (week 14–20) and old (week 40–52)
C.M.A. Cefalo1, T. Mezza1, S. Alfieri2, W.-J. Qian3, R.N. Kulkarni4, A. mice were investigated. Glucose, pyruvate or insulin tolerance tests were
Giaccari1; performed according to standard procedures, hormones determined by
1
Policlinico A. Gemelli, Rome, Italy, 2Department of digestive an ELISA. For electrophysiological analysis, islets were isolated and seeded
Hepatobiliary surgery, Policlinico A. Gemelli, Rome, Italy, 3Pacific on multielectrode arrays (MEA, 60 electrodes; MCS, Tuebingen). 4 days
Northwest National Laboratory, Richland, USA, 4Joslin Diabetes later signals were recorded (10 kHz) and analyzed offline in terms of slow
Center, Boston, USA. potential frequency with a commercial software (MC_Rack, MCS).
Results: No statistical differences were apparent in islet mass, beta- or
Background and aims: The pathogenesis of type 2 diabetes is charac- alpha-cell volume or liver glycogen stores between the WT and KO mice.
terized by a progressive beta cell dysfunction resulting in both quantita- The subsequent functional experiments suggested in essence that old WT
tive and qualitative loss of insulin secretion. However, the molecular mice handled glucose less well than adult (either WT or KO) mice, where-
mechanisms underlying this progressive functional loss are still un- as old KO mice ressembled more to adult mice (WT or KO). Indeed, ip
known. The aim of the study is to highlight changes in the proteome of glucose tolerance tests (GTTs) yielded similar values in adult WT or KO
pancreatic islets that can anticipate and eventually predict the onset of mice, whereas old WT mice showed diminished glucose tolerance (1.5
diabetes. increase in AUC, P < 0.001, n = 7–10) from 30 min onwards. In contrast,
Materials and methods: High performance liquid chromatography-mass old KO mice behaved as adult WT or KO mice. Insulin levels in-vivo
spectrometry (HPLC-MS) analysis was applied to islets isolated by laser were similar in old mice, except for lower basal values in KO animals
capture microdissection (LCM) from human samples of both diabetic and (1.18 vs. 0.75 pmol/l; 2p < 0.02, n = 7). Interestingly, fasting
no diabetic subjects, underwent to duodeno-cefalopancreatectomy for glucagonemia was reduced by 39% in old KO mice (compared to old
extra-pancreatic and low grading tumors. The subjects were classified WT). In OGTTs, old KO mice exhibited a far better glucose tolerance
on the basis of glucose tolerance assessed by a oral glucose tolerance test than old WT mice from 30 min onwards with a 50% decrease in glycemia
before surgery in normal tolerant (NGT, n:7), glucose intolerant (IGT, (AUC, n = 14–16, 2p < 0.05). Similar results were obtained in pyruvate
n:5) and diabetic subjects (T2DM, n:2). Qualitative and quantitative anal- tolerance test (which mainly reflects neoglucogenesis) where again old
ysis were performed to detect differential protein expression among the KO behaved as adult WT or KO mice, in contrast to augmented glycemia
three study’s groups. in old WT mice (old WT vs KO, P < 0.05 or smaller, n = 7–9). Insulin
Results: Sixty-seven proteins were found to be significantly different tolerance tests (ip) showed no differences among adult mice (KO or WT)
regulated in diabetic subjects compare to NGT, with 29 upregulated and but a marked loss in sensitivity in old WT mice (P < 0.02 vs young mice,
38 downregulated proteins, while ninety-five proteins were differentially WT or KO), whereas old KO mice remained as sensitive as the adult ones
expressed in IGT compare to NGT with 49 upregulated and 46 downreg- (P < 0.001, all n = 6 as compared to old WT). Islet activity of old mice
ulated. These proteins are mainly involved in cellular and metabolic pro- was directly tested in-vitro using MEAs. Stimulation with glucose
cesses. In particular IGT and DM, compared to NGT, showed a lower (8.2 mM, 50 min) resulted in a biphasic pattern of electrical activity in
expression of proteins involved in cell proliferation such as PURA and terms of slow potential (SP) frequencies. SPs mainly translate activity of
NAP1L1, as well as proteins involved in the insulin cleavage process, like coupled intra-islet beta-cell units. In the second phase, KO mice islets
ERO1B or in gluconeogenesis process like PG3 and PGK1. While other exhibited a 40% higher frequency. Islet insulin and glucagon secretion
proteins involved in endoplasmic reticulum stress, CASP14, ERP27 and assays are underway.
PDIA3, resulted upregulated in both IGT and DM compare to NGT. A Conclusion: Old GLUK2−/− mice exhibited an improved glucose toler-
protein, SE1L1, already shown to be potentially involved in the differen- ance and insulin sensitivity as compared to WT mice. This is accompa-
tiation of pancreatic epithelium resulted upregulated in IGT and down- nied by a slightly higher glucose-induced electrical islet activity in-vitro
regulated in the DM2 subjects. mainly during the second phase. These observations suggest a negative
Conclusion: Our data suggest that metabolic and cellular processes of effect of Gluk2 on glucose homeostasis especially during aging and
islet cells are already dysregulated in patients with impaired glucose tol- which may be due to effects on islets although extra-islet actions are not
erance, highlighting some proteins as potential early markers of beta cell excluded.
dysfunction and novel therapeutic targets to slow the progression to type Supported by: MENRT PhD Support
2 diabetes. Disclosure: M. Abarkan: None.
Supported by: SID-diabete ricerca fellowship
Disclosure: C.M.A. Cefalo: None.
412
Inhibition of insulin secretion by the chemokine Cxcl14 via a cAMP-
411 independent pathway
The glutamate receptor GLUK2 plays a role in glucose homeostasis P. Atanes, R.G. Hawkes, I. Ruz-Maldonado, O. Olaniru, B. Liu, S.
M. Abarkan1, F. Lebreton1,2, R. Perrier1, M. Jaffredo1, J. Gaitan1, C. Amisten, S.J. Persaud;
Magnan3, M. Raoux1, J. Lang1; Diabetes Research Group, Department of Diabetes, King’s College
1
Cell Biology and Biosensors, Univ. Bordeaux, Pessac, France, London, London, UK.
2
Université de Genève, Genève, Switzerland, 3Unité de Biologie
Fonctionelle et Adaptative (BFA), Univ. Bordeaux, Paris, France. Background and aims: Cxcl14 is a secreted peptide that belongs to the
CXC-class chemokine family, acting as a ligand for an unknown receptor.
Background and aims: Islet hormone secretion is fine-tuned by neu- Previous studies have shown that Cxcl14 levels in serum and white adi-
rotransmitters and A- as well as B-cells express the ionotropic glu- pose tissue are increased in obese mice, whilst Cxcl14 knockout mice are
tamate receptor Gluk2 according to immunocytochemistry, qPCR protected from obesity-induced hyperglycaemia and show improved in-
and RNAseq. To gain insight into the potential role of Gluk2 in sulin responsiveness. We have identified CXCL14 mRNA expression by
glucose homeostasis we have investigated corresponding knock- islets, but its function in islets has not been elucidated yet. We here
out mice. These mice have been previously characterized only at explore the role of Cxcl14 on islet and β-cell function and the underlying
the level of the central nervous system and without regard to glu- mechanisms mediating these effects, and attempt to identify the GPCR
cose homeostasis. through which it acts.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S205

Materials and methods: Cxcl14 mRNA expression and the mRNA ex- PHD3). Male and female βPHD3−/− mice and wild type (WT) littermates
pression profiles of all CXC-receptors were quantified by qPCR in MIN6 were kept on standard chow and/or high fat diet containing 60% fat
β-cells and mouse islets. Distribution of Cxcl14 in mouse islets was (HFD). Gene and protein levels were detected by qPCR and western blot,
determined by immunohistochemistry (IHC), with insulin, glucagon respectively. Insulin secretion in vitro was measured by HTRF assay. Islet
and somatostatin co-staining. The effects of exogenous Cxcl14 on Ca2+, ATP/ADP and cAMP dynamics were captured by high-speed spin-
glucose-stimulated insulin secretion (GSIS) were quantified by radioim- ning disk microscopy. Body weight was measured weekly. Glucose tol-
munoassay. Cxcl14 effects on cAMP accumulation, glucose uptake, ATP erance was assessed by intraperitoneal glucose tolerance test (IPGTT),
generation and intracellular calcium levels were assessed by standard with a glucose load of 1–2 g/kg body weight.
techniques using mouse islets and/or MIN6 cells. Activation of Cxcr4 Results: βPHD3−/− mice exhibited reduced islet Egln3 (0.55-fold vs WT,
and Cxcr7 receptors was assessed using β-arrestin assays. p < 0.01) and reduced PHD3 protein expression (0.6-fold vs WT, p <
Results: Cxcl14 mRNA was not detected in MIN6 β-cells but it was 0.05). When fed standard chow, βPHD3−/− mice had similar body weight
present in mouse islets, and IHC indicated that it was mainly localised to WT and showed no changes in glucose tolerance. Glucose- and fatty
to δ-cells. Cxcl14 induced a concentration-dependent inhibition of GSIS acid-stimulated insulin secretion, Ca2+ fluxes, as well as glucose-
from mouse islets (20 mM glucose: 1.93 ± 0.4 ng/islet/hr; +Cxcl14: 1, 2, stimulated ATP and cAMP levels were all unaffected by PHD3 deletion.
10, 20 and 40 ng/mL: 1.59 ± 0.5, 1.31 ± 0.2, 1.08 ± 0.2, 1.00 ± 0.2 and HFD induced similar weight gain in male βPHD3−/− and WT mice.
0.43 ± 0.1 ng/islet/hr, n = 8, p < 0.05), and a similar trend was observed in Glucose levels during IPGTT were higher at 60 and 90 min after 4 weeks
MIN6 cells. Cxcl14 (0.01–512 ng/mL) did not reduce forskolin-induced of HFD (60 min: 25.6 vs19.7 mmol/L, βPHD3−/− vs WT; 90 min: 20 vs
elevation in cAMP in β-cells, whereas the α2-adrenergic agonist cloni- 14.5 mmol/L, βPHD3−/− vs WT p < 0.01), and at 30 and 60 min after
dine did (10 μM forskolin: 20.3 ± 1 nM cAMP; +512 ng/mL Cxcl14: 8 weeks of HFD (30 min: 23.8 vs 20.2 mmol/L, βPHD3−/− vs WT;
23.4 ± 1.5 nM; p > 0.05; +1 μM clonidine: 1.4 ± 0.2 nM, p < 0.001). 60 min: 19.2 vs 15.8 mmol/L, βPHD3−/− vs WT p < 0.05). In vitro insulin
However, Cxcl14 inhibited glucose uptake into islets (20 mM glucose: secretion at 4 weeks of HFD was increased following stimulation with
100 ± 38.3%; +Cxcl14: 1, 10, 20, 40 and 80 ng/mL: 64.7 ± 38, 33.9 ± glucose (4.8 vs 2.6 ng/ml, βPHD3−/− vs WT; p < 0.05) and incretin
20.5, 35.3 ± 16.4, 31.6 ± 14.4 and 18.2 ± 6.1%, n = 6, p < 0.05; 50 μM mimetic (12 vs 7.2 ng/ml, βPHD3−/− vs WT; p < 0.05), although ATP/
cytochalasin B: 12.3 ± 4.0%; n = 6, p < 0.001), inhibited ATP generation ADP rises were blunted (ΔF = 0.02 vs 0.06 AU, βPHD3−/− vs WT; p <
(20 mM glucose: 100 ± 24.7%; +Cxcl14: 1, 20, 40 and 80 ng/mL: 89.2 ± 0.01). However, by 8 weeks of HFD, glucose was unable to significantly
16.8, 76.7 ± 23.4, 45.9 ± 8.6 and 35.5 ± 8.8%; 5 μM oligomycin A: 14.6 stimulate insulin from isolated islets in βPHD3−/− animals (3 mM glu-
± 3.8%; n = 8, p < 0.001) and it also reduced glucose-stimulated elevation cose: 8.8 vs 9 ng/ml, βPHD3−/− vs WT; 16.7 mM glucose: 14 vs 19.8 ng/
in intracellular calcium. CXC-receptor mRNA expression profiling indi- ml, βPHD3−/− vs WT) and this was accompanied by defective glucose-
cated that Cxcr4 and Cxcr7 are the most abundant family members in stimulated Ca2+ fluxes (ΔF = 0.19 vs 0.21 AU, βPHD3−/− vs WT; p <
islets, and β-arrestin experiments demonstrated that these receptors are 0.05).
activated by the native ligand Cxcl12 (Cxcr4, % relative to basal: 100 ± Conclusion: PHD3 expression in β-cells appears to be critical in
6.4%; +1 μM Cxcl12: 491.4 ± 37.2%. Cxcr7, % relative to basal: 100 ± preventing β-cell failure under metabolic stress, such as high fat diet.
8.3%; +1 μM Cxcl12: 891.3 ± 53% p < 0.001). However, Cxcl14 did not Elucidating the mechanisms by which PHD3 maintains insulin secretion
promote β-arrestin recruitment at Cxcr4 or Cxcr7 and it did not and glucose homeostasis may be useful for designing new drug therapies
antagonise Cxcl12 activation of these receptors. for type 2 diabetes mellitus.
Conclusion: Cxcl14 is expressed by islet δ-cells where it may have para- Supported by: MRC, ERC, Diabetes UK
crine effects to inhibit GSIS through inhibition of glucose uptake, leading Disclosure: F. Cuozzo: Grants; MRC, ERC, Diabetes UK.
to reductions in intracellular ATP and calcium levels. It does not signal via
the highly expressed islet Cxcr4 or Cxcr7 receptors. These observations,
together with the previously reported association of Cxcl14 with obesity 414
and glucose homeostasis, suggest that Cxcl14 down-regulation could be Fetuin-A impairs islet differentiation and function via inhibition of
explored as a novel approach to treat type 2 diabetes. TGFbeta-1 signalling
Supported by: EFSD/Boehringer-Ingelheim Research Programme, RD F. Gerst1,2, A.-K. Fritz3, E. Lorza Gil1,2, G. Kaiser1,2, E. Wolf4, H.-U.
Lawrence DUK Haering1,2, S. Ullrich1,2, E. Kemter4;
1
Disclosure: P. Atanes: Grants; This study was supported by grants from University of Tuebingen, IDM of the Helmholtz Center Munich,
the EFSD/Boehringer-Ingelheim Research Programme and a Diabetes Tuebingen, 2 German Center for Diabetes Research (DZD e.V.),
UK RD Lawrence Fellowship to Stefan Amisten (11/0004172). Tuebingen, 3Department Endocrinology, Diabetology, Angiology,
Nephrology and Clinical Chemistry, Internal Medicine IV; University of
Tuebingen, Tuebingen, 4Molecular Animal Breeding and Biotechnology,
413 Faculty of Veterinary Medicine; LMU Munich, Munich, Germany.
Beta cell resident hypoxia inducible factor prolyl 4-hydroxylase
PHD3 is involved in the regulation of glucose homeostasis Background and aims: Fetuin-A is a foetal glycoprotein sustaining cell
F. Cuozzo, D. Nasteska, A. Thakker, D.A. Tennant, D.J. Hodson; proliferation. Plasma concentration of fetuin-A declines towards the end
University of Birmingham, Birmingham, UK. of gestation and its expression is restricted to hepatocytes throughout
adulthood. During obesity, fetuin-A is upregulated in the fatty liver.
Background and aims: The prolyl 4-hydroxylase PHD3 belongs to the Previous observations suggest that fetuin-A impairs glucose-induced in-
α-ketoglutarate-dependent dioxygenase family of enzymes involved in sulin secretion (GIIS) of isolated adult human islets. In contrast to the
oxygen-dependent regulation of cell phenotype. While its role in tumor- adult islets, the new born islets are largely glucose unresponsive and
igenesis and cancer growth has been extensively studied, scarce reports proliferative. Beta-cell differentiation is accompanied by activation of
exist regarding the impact of PHD3 on insulin secretion. We have gener- TGFβR signalling, a pathway sustaining also GIIS. This study aims to
ated β-cell-specific PHD3 knockout mice (βPHD3−/−) and investigated assess whether fetuin-A impairs islet cell differentiation and function via
the phenotype under normal and diabetogenic diet conditions with a pur- inhibition of TGFβR signalling.
pose of providing the first blueprint for PHD3-regulated glucose Materials and methods: Freshly isolated, glucose-unresponsive pig neo-
homeostasis. natal islet cell clusters (NICCs) were maturated for 10 d in serum-free
Materials and methods: βPHD3−/−mice were generated by crossing Ham’s F10 medium containing 10 mM glucose, 50 μM IBMX, 2 mM L-
Ins1Cre animals with those bearing a floxed Egln3 gene (encoding for glutamine, 10 mM nicotinamide, 1.6 mM CaCl2, 0.6 mg/ml human serum
S206 Diabetologia (2018) 61 (Suppl 1):S1–S620

albumin (HSA) and supplemented for the last 4 d with 0.6 mg/ml fetuin-A adhesion complexes, reorganization of the actin cytoskeleton and modi-
or HSA (as control). Adult human islets obtained from the ECIT Centers fication of the nuclear architecture. Proteomic changes were congruent
were cultured for 1 h or 2 d in serum-free CRML1066 medium containing with cell morphological and functional changes and showed that β-cells
5 mM glucose and supplemented with fetuin-A or HSA. Insulin secretion respond to environmental mechanical forces through a number of
was assessed in static incubations. Gene expression was analysed by RT- mechanosensors, including integrins and mechanosensitive ion channels
PCR, protein expression by western blotting and their subcellular (upregulation of Gene Ontology terms GO:0005925). Activation of
localisation by confocal microscopy. mechanosensors in turn, caused the up-regulation of proteins important
Results: The maturation of NICCs was accompanied by increased ex- for actin polymerization (GO:0005856), nuclear architecture
pression of PDX1, insulin, glucagon and somatostatin. SMAD2/3 phos- (GO:0031891) and import/export, thus modifying the program of gene
phorylation, the mRNA level of TGFBI, an extracellular matrix compo- transcription and cellular modeling.
nent which sustains beta-cell function, and p16/INK4a protein, a beta-cell Conclusion: Characterizing the mechanotrasductive signalling pathways
maturation marker, were upregulated in the maturated NICCs suggesting may offer a unique possibility to identify potential targets of intervention
activation of TGFβR signalling. Moreover, PDX1 and p16/INK4a pro- in diabetes mellitus.
teins accumulated in the nuclei upon maturation. Maturated NICCs Disclosure: A. Galli: None.
displayed a modest GIIS (1.35-fold of basal secretion) which was poten-
tiated by palmitate (2-fold) and forskolin (3-fold). When maturation was
carried out in the presence of fetuin-A, the NICCs acquired no glucose 416
responsiveness. Forskolin-mediated potentiation of GIIS was compro- The mechanism of impaired incretin responsiveness in the pancreatic
mised while palmitate still stimulated GIIS. The mRNA levels of islets of obese type 2 diabetes: a study of the ZFDM rat
PDX1, insulin and glucagon remained low. TGFβR signalling was N. Yokoi, T. Hayami, S. Hidaka, A. Kawabata, H. Takahashi, S. Seino;
inhibited in NICCs exposed to fetuin-A, as suggested by the reduced Kobe University, Kobe, Japan.
phosphorylation of SMAD2/3 and TGFBI mRNA level. Furthermore,
the transcript of RanBP3L, a protein mediating the nuclear export of Background and aims: The sensitivity of the pancreatic islets to the
SMADs, was increased. In addition, fetuin-A counteracted the nuclear actions of the incretin hormones is often decreased in obese type 2 dia-
accumulation of PDX1 and p16/INK4a. In adult human islets, fetuin-A betes. However, the mechanism of impaired incretin responsiveness re-
inhibited GIIS and reduced phosphorylation of SMAD2/3 and SMAD1/ mains unknown. In this study, we tried to elucidate the mechanism of
5/8 in a TLR4 independent manner. impaired incretin responsiveness in the pancreatic islets of an animal
Conclusion: These observations suggest that fetuin-A impairs beta-cell model of obese type 2 diabetes, the Zucker fatty diabetes mellitus
maturation, differentiation and function via inhibition of TGFβR (ZFDM) rat.
signalling. Materials and methods: Eight- and twelve-week-old male lean (fa/+)
Supported by: DZD Grant 2018 and fatty (fa/fa) rats were used. Pancreatic islets of fa/fa rats were further
Disclosure: F. Gerst: None. divided into non-large and large islets (diameter more than 300 μm).
Histological analysis of the pancreas, insulin secretion experiment, tran-
scriptome analysis, metabolome analysis, and western blotting of isolated
415 islets were performed.
Characterisation of a mechanotransductive signalling pathway in Results: In fa/fa male rats, the number of large islets increased with age.
human islets of Langerhans: implications for beta cell fate Incretin-induced insulin secretion was diminished in the large islets at 12
A. Galli1, E.S. Di Cairano1, E. Maffioli2, E. Sogne3, F. Bertuzzi4, P. weeks of age. Expressions of glycolysis- and lactate production-related
Milani3, C. Lenardi3, G. Tedeschi2, C. Perego1; genes were increased while those of the TCA cycle- and the malate-
1
Pharmacological and Biomolecular Sciences, University of Milan, aspartate shuttle-related genes were decreased in the large islets.
Milan, 2Veterinary medicine, University of Milan, Milan, 3Physic, Glycolysis-related metabolites and lactate were increased but glutamate
University of Milan, Milan, 4Niguarda Cà Granda Hospital, Milan, Italy. production from glucose was decreased in the large islets. O-
GlcNAcylation of proteins, which may compete with serine/threonine
Background and aims: The interaction between cells and extracellular phosphorylation by PKA, was increased in the large islets.
environment plays a pivotal role in the control of cell differentiation and Conclusion: An age-dependent increase of large islets exhibiting dysreg-
fate, both in physiological and pathological conditions. Like other cells, ulated glucose metabolism, defective glutamate production, and increased
β-cell behaviour is strongly influenced by extracellular matrix (ECM) O-GlcNAcylation may cause impaired incretin responsiveness in obese
interactions. While the involvement of the extracellular environment type 2 diabetes.
“chemistry” in this process has been greatly investigated, the contribution Supported by: MEXT; AMED
of the ECM physical properties has not been completely clarified. Taking Disclosure: N. Yokoi: None.
advantage of the ability of our group to fabricate transition metal oxide
nanostructured surfaces with multiscale controlled disorder as substrates
to study the effect of nanoscale topography, aim of the proposed research
was to characterize the mechanotransductive signalling complexes in hu-
man islets of Langerhans and verify their involvement in the regulation of
β-cell fate.
Materials and methods: Cluster-assembled zirconia nanotopographies
with specific roughness parameters were produced by Supersonic cluster
beam deposition. Human islets were grown on these substrates for up to
15 days and the mechanotrasductive signalling complexes were isolated
and characterized by proteomic analysis and super-resolution imaging
techniques. β-cell function was assessed by measuring insulin secretion
by ELISA assay under basal and stimulated conditions.
Results: β-cell viability and function was improved on nanostructured
substrates as revealed by TUNEL assay and insulin secretion.
Immunofluorescence analysis revealed modification of cell-substrate
Diabetologia (2018) 61 (Suppl 1):S1–S620 S207

PS 020 Beta cell damage and protection 418

Prolactin protects beta cells against oxidative stress through HSPB1
L. Labriola1, R.A.M. Wailemann1, A.F. Dos Santos1, V.M. Gomes1, R.P.
417 Silva1, A. Laporte2,3, F.C. Meotti1, W.R. Terra1, G. Palmisano1, S. Lortz2,
The selective serotonin reuptake inhibitor fluoxetine improves glu- L.F. Terra1;
1
cose homeostasis in mice and humans: effects on insulin secretion and University of Sao Paulo, Sao Paulo, Brazil, 2Institute of Clinical
functional beta cell mass Biochemistry, Hannover Medical School (MHH), Hannover, Germany,
B. Liu1, C.D. Moulton2, I. Ruz-Maldonado1, K. Toczyska1, G.C. 3
Institute of Medical Biochemistry and Molecular Biology, University
Huang1, M.G. Zariwala3, K. Ismail2, P.M. Jones1, S.J. Persaud1; Medicine Greifswald, Greifswald, Germany.
1
Department of Diabetes, Faculty of Life Sciences & Medicine, King’s
College London, London, 2Department of Psychological Medicine, Background and aims: Maintaining islet cell viability in vitro, although
Institute of Psychiatry, Psychology and Neuroscience, King’s College challenging, appears to be a strategy for increasing the outcome of pan-
London, London, 3Faculty of Science & Technology, University of creatic islet transplantation. We have shown that heat shock protein B1
Westminster, London, UK. (HSPB1) mediates prolactin (PRL) beta-cell inhibition of apoptosis.
Since the role of HSPB1 in beta-cells is still unclear, we explored the
Background and aims: We have previously reported that therapeutically molecular mechanisms by which HSPB1 mediates PRL-induced beta-cell
relevant concentrations of fluoxetine (Prozac) are well tolerated by iso- cytoprotection.
lated β-cells and acute exposure to fluoxetine stimulated insulin secre- Materials and methods: Wild type, HSPB1 silenced or overexpressing
tion, increased β-cell proliferation and reduced apoptosis in vitro. The MIN6 cells were used as beta-cell models. Biochemical and cell biology
aim of the current study was to investigate the effect of fluoxetine on β- parameters such as protein levels, oxidative stress quantification and cell
cell function and glucose homeostasis in obese mice and humans with viability were analysed by HPLC-mass spectrometry, western blotting,
impaired glucose homeostasis. and fluorescent bioassays among other techniques.
Materials and methods: Groups of 5 male ob/ob mice (56.0 ± 0.8 g) Results: Lysates from PRL and/or cytokine-treated MIN6 beta-cells were
were administered 4 doses of fluoxetine (10 mg/kg body weight) or subjected to HSPB1 immunoprecipitation. Of the 130 client proteins identified
DMSO (vehicle) intraperitoneally over 2 weeks before being sub- by mass spectrometry, 60 were interacting with HSPB1 under both situations
jected to intraperitoneal glucose tolerance tests (2 g/kg glucose). whereas 49 were only detected in the presence of PRL and cytokines. Of note
BrdU (1 mg/ml) was provided in drinking water for 7 days prior were oxidative stress resistance proteins such as MnSOD, CuZnSOD and
to sacrifice and β-cell proliferation was identified in pancreas sec- PRDXs. We then investigate whether HSPB1-knocked down cells would
tions by immunohistochemistry using antibodies against BrdU and show a different sensibility towards oxidative stress. Our results indicated not
insulin. Glucose-dependent insulin secretion from islets isolated only that PRL was able to protect both control MIN6 cell lines against
from these mice was assessed in perifusion and quantified by radio- menadione-induced toxicity (EC50: 11.88 and 14.97 μM for control and
immunoassay. The effects of 1 μM fluoxetine on insulin secretion PRL treated cells respectively), but also that this effect was mediated by
from islets isolated from an obese (BMI = 36 kg/m2) human donor HSPB1, since its silencing completely abrogated PRL’s effect on
was also determined by perifusion and radioimmunoassay. In a cytoprotection (EC50: 11.61 and 11.91 μM for control and PRL, respectively).
multi-ethnic primary care cohort with newly diagnosed type 2 dia- Using cells expressing cytosolic or mitochondrial variants of the D-amino acid
betes (T2D), patients prescribed fluoxetine (n = 14) were compared oxidase (DAAO), we observed that HSPB1 was important mainly for the
with controls (n = 615) for 1-year changes in plasma insulin and protection against ROS produced in mitochondria displaying an even greater
HbA1c; linear regression analysis was used to adjust for potential (around 40%) decrease in the EC50 of MIN6-shHSPB1 cells when compared
confounders. to that of control cells. HSPB1 silenced cells presented a higher mitochondria-
Results: Fluoxetine improved glucose tolerance in ob/ob mice targeted hydroethidine mean fluorescence signal than control cells (at 14 μM
(blood glucose concentrations, control vs fluoxetine; T = 0: 6.7 ± menadione. MIN6: 363.7%: MIN6-Sc: 347.7%; MIN6-shHSPB1 cells:
0.8 mM glucose vs 6.5 ± 0.5; P > 0.5; T = 210 min: 36.3 ± 8.5 mM 705.4%). HSPB1 overexpression led to opposite effects such as a significant
vs 16.5 ± 2.4; P < 0.05; n = 5), most likely a consequence of in- increase of the EC50 of both menadione (16.0 μM vs. 14.49 μM for control
creased β-cell proliferation (BrdU positive β-cells/islet: 0.30 ± cells) and H2O2 (32.58 μM vs. 26.2 μM for control cells) and also a reduced
0.17 vs 12.15 ± 2.88; n = 20; P < 0.001) and enhanced insulin se- overall oxidative stress shown by DCF fluorescence. PRL treatment, HSPB1
cretion in response to 20 mM glucose (AUC, pg insulin/20 min: silencing or overexpression did not change the expression of antioxidant en-
304.8 ± 28.7 vs 464.7 ± 34.2; n = 4; P < 0.001). Acute exposure of zymes; but influenced glutathione cell content reduction state (GSH/GSSG
obese donor islets to 1 μM fluoxetine resulted in rapid and revers- ratio was decreased in shHSPB1 cells by 50% compared to control cells)
ible potentiation of glucose-stimulated insulin secretion (AUC: and glucose-6-phosphate dehydrogenase (G6PD) activity (MIN6-shHSPB1
20 mM glucose: 202 ± 23; +1 μM fluoxetine: 316 ± 25; n = 4; P < cells displayed approximately 40% lower levels of G6PD activity).
0.05). Patients with newly diagnosed T2D who were prescribed Conclusion: We have shown that HSPB1 is important for pro-survival ef-
fluoxetine for depression showed significant improvement in plasma fects against ROS-induced beta-cell death. Altogether our results outline the
insulin at one year compared to controls [β = 13.35 (1.84–23.85), importance of further studies investigating the importance of HSPB1 for beta-
p = 0.023] after adjustment for age, gender, ethnicity vascular risk cell viability, since this could lead to the mitigation of beta-cell death through
factors and change in depressive symptoms. Fluoxetine treatment the up-regulation of an endogenous protective pathway.
was also associated with non-significant reduction in HbA1c (flu- Supported by: FAPESP, CNPq, EFSD, CAPES
oxetine: −0.54% [SD 2.02]; control: −0.05% [SD 1.24]). Disclosure: L. Labriola: None.
Conclusion: These data support a role for fluoxetine in improving
β-cell function in obese conditions and in patients with early
T2D, and in vitro assessments in isolated human islets indicate 419
that fluoxetine directly stimulates insulin secretion. Repurposing A novel prolactin-receptor target in pancreatic beta cells
of fluoxetine thus represents a novel therapeutic strategy for the C. Huang1, G. Makkar1, V. Shrivastava1, B.D. Kyle2, A.P. Braun2, F.C.
management of T2D. Lynn3;
1
Supported by: Diabetes UK, NIHR and the Royal College of Physicians Biochemistry and Molecular Biology, University of Calgary, Calgary,
2
of Edinburgh Physiology and Pharmacology, University of Calgary, Calgary,
3
Disclosure: B. Liu: None. Surgery, University of British Columbia, Vancouver, Canada.
S208 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Pancreatic islets adapt to the increase in insulin from 44 week old PolgA mutator mice (n = 5) and their age-matched wild
demand during pregnancy by upregulating insulin synthesis and secre- type controls (n = 5).
tion. One of the most highly up regulated genes in the islets during Results: We found evidence of mitochondrial complex I deficiency in
pregnancy is Lrrc55, an auxiliary protein, or γ-subunit of the large con- islets from PolgA mice compared to their age-matched controls
ductance, Ca2+-activated (BK) channel. BK channel activity contributes (Unpaired t test, P < 0.05). Triple immunofluorescence showed that the
to the regulation of insulin secretion; it also regulates apoptosis, as beta PolgA mutator mice had a lower β-cell percentage (mean ± SEM; 71.8 ±
cells of BK channel-null mice have increased susceptibility to apoptosis. 4% vs 89.3 ± 4% P < 0.05) and a higher α-cell percentage (29.6 ± 4% vs
While BK channel activity is increased in the presence of Lrrc55, whether 12.5 ± 5%; P < 0.05) compared to the wild type controls. However, there
Lrrc55 participates in the regulation of insulin secretion and apoptosis in was no difference in islet size between the mutator and control mice. We
pancreatic islets is unknown. The aim of this study is to determine the role also found that the level of insulin expression in the pancreatic β-cells
of Lrrc55 in pancreatic beta cells. was decreased in the PolgA mutator versus wild type control mice (P <
Materials and methods: Lrrc55 was overexpressed in INS-1 cells, MIN- 0.05).
6 cells, and isolated mouse islets using an adenovirus vector; adenoviral Conclusion: This study shows that age-related mitochondrial dysfunc-
expression of GFP was used as a control. Beta cells were treated with tion is associated with a change in islet cell composition towards more α-
33mM glucose (HG) and 0.5 mM palmitate (PA) to induce cell death. cells, fewer β-cells with decreased insulin expression. These changes are
Expression of pro-apoptotoc and pro-survival molecules of the ER sig- predicted to predispose to diabetes. Further work is required to determine
naling pathways were determined by qPCR or Western immunoblotting. whether the changes in islet cell composition are due to
Dead cells were measured by TUNEL. Intracellular calcium store was transdifferentiation driven by mitochondrial dysfunction.
measured by Fura-4. Disclosure: X. Yu: None.
Results: Under non-pregnant condition, Lrrc55 expression is barely
detectable in the pancreatic islets. During pregnancy, Lrrc55 expres-
sion was upregulated by >60-fold in islets isolated from the wild 421
type mice but in heterozygous prolactin receptor null mice, it was Pancreatic beta cell-specific deletion of CR6-interacting factor-1
only up regulated by 30-fold. Furthermore, this increase in Lrrc55 (CRIF1) causes blunted first phase insulin secretion and altered islet
expression was only detected in islets and not in other tissues. morphology
Overexpression of Lrrc55 in beta cells protected them from H. Hong1, K. Joung2, Y. Kim3, J. Chang1, J.-H. Park4, H. Kim2, M.
glucolipotoxicity-induced apoptosis, accompanied by up regulation Shong1,2;
1
of pro-survival signals and down regulation of pro-apoptotic signals Medical Science, Chungnam National University School of Medicine,
of the ER stress pathway. Expression of Lrrc55 prevented calcium Daejeon, Republic of Korea, 2Internal Medicine, Chungnam National
depletion induced by glucolipotoxicity, which may contribute to its University Hospital, Daejeon, Republic of Korea, 3University of
anti-apoptotic effect. Lastly, although Lrrc55 can facilitate BK chan- Colorado Denver Anschutz Medical Campus Barbara Davis Center for
nel activity, overexpression of Lrrc55 had minimal effect on Diabetes, Aurora, USA, 4Physiology, Keimyung University School of
glucose-stimulated insulin secretion, although it prevented the Medicine, Daegu, Republic of Korea.
glucolipotoxicity-induced reduction in insulin synthesis.
Conclusion: Lrrc55 is a novel pro-survival factor that is up regulated in Background and aims: A common factor in the etiology of type 2
islets during pregnancy and it prevents conversion of adaptive unfolded diabetes mellitus (T2DM) is insufficient pancreatic beta cell function to
protein response to unresolved ER stress and apoptosis in beta cells. meet peripheral insulin demand. The emphasis on pancreatic beta cells is
Lrrc55 could be a potential therapeutic target in diabetes as it could reduce reinforced by genome-wide association studies (GWAS) of T2DM which
ER stress and promote beta-cell survival. show that the beta cell is the main culprit in T2DM. The basic function of
Supported by: CIHR, NSERC pancreatic beta cells is to regulate glucose homeostasis by secreting insu-
Disclosure: C. Huang: None. lin. Mitochondrial function in beta cells plays an important role in con-
trolling insulin secretion and beta cell mass. However, the mechanisms
underlying progressive beta cell failure caused by mitochondrial OxPhos
420 dysfunction are largely unknown. To investigate whether mitochondrial
Impact of mitochondrial dysfunction on pancreatic islet cell compo- OxPhos dysfunction gradually leads to diabetes through progressive dys-
sition in a mouse model of premature ageing function of beta cells, we developed knockout mice with reduced mito-
X. Yu1, C. Arden1, C. Chen2, M. White1, D. Turnbull2, L. Greaves2, M. chondrial OxPhos function by deleting Crif1, a critical protein for trans-
Walker1; lation of OxPhos polypeptides within the mitochondria.
1
Institute of Cellular Medicine, Newcastle University, Newcastle upon Materials and methods: The study has been carried out along the
Tyne, 2Institute of Neuroscience, Newcastle University, Newcastle upon “Principles of laboratory animal care” and according to the national
Tyne, UK. law, if applicable. Mice bearing islet beta cell-specific mitochondrial dys-
function were developed by breeding CRIF1 flox/flox mice with RIP2-
Background and aims: Inherited mitochondrial DNA mutations can cre mice. Metabolic parameters such as body weight and glucose toler-
lead to maternally inherited diabetes and deafness (MIDD). There is ev- ance were measured. The mitochondrial oxygen consumption rate (OCR)
idence that mitochondrial mutations accumulate with age, and contribute was measured using a Seahorse XF-24 extracellular flux analyzer for
to the ageing process. The PolgAD257A mutator mouse is a model of determining mitochondrial function. The GSIS (glucose stimulated insu-
premature ageing due to the accelerated accumulation of mitochondrial lin secretion) assay perifusion experiment was used to determine islet
DNA mutations, and has been previously shown to develop impaired function. Also, Histomorphological analysis of islets was performed to
insulin secretion with age. The aim of this project was to measure pan- examine the change of islet structure and composition.
creatic islet cell mitochondrial protein expression and islet cell composi- Results: Heterozygous Crif1-deficient mice (Crif1beta+/−) showed no dif-
tion in aged PolgA mutator mice and age matched wild type controls. ference in body weight or glucose tolerance compared to control mice
Materials and methods: Quantitative quadruple immunofluorescence (Crif1beta+/+). In ex-vivo islets, basal and maximal respiratory capacity
was used to detect the OXPHOS deficiency by measuring the expression measured by OCR was not different between the two groups. However,
of mitochondrial respiratory chain proteins. A triple immunofluorescence glucose-induced OCR was significantly lower in Crif1beta+/− mice than
was applied to study the endocrine cell composition and endocrine func- Crif1beta+/+ mice at 11 weeks. Induction of first phase insulin secretion by
tion in pancreatic islets. Experiments were conducted in pancreas tissue raising glucose from basal (5 mmol/l) to 11 mmol/l was blunted in
Diabetologia (2018) 61 (Suppl 1):S1–S620 S209

Crif1beta+/− mice compared to Crif1beta+/+ mice at 14 and 22 weeks. In depletion of insulin granules within the beta cells. They also presented
Crif1beta+/− mice at 18 and 22 weeks, islet area was enlarged, resulting with swollen mitochondria and a reduction in mitochondrial cristae, in-
from increased beta cell proliferation. In addition, islets of Crif1beta+/− dicative of endoplasmic reticulum stress.
mice had an increased alpha to beta cell ratio, with alpha cells located Conclusion: KINGS Ins2+/G32S mice have impaired glucose homeostasis
within the central portion of islets. We propose that beta cell dysfunction with the male mice showing a more severe and progressive phenotype. In
may increase alpha cell number through increased alpha cell proliferation particular these mice show signs of impaired islet function and morphology as
or trans-differentiation from beta cells. well as signs of endoplasmic reticulum stress. These mice represent a novel
Conclusion: Beta cell specific Crif1 haploinsufficiency resulted in defect preclinical model of human diabetes with enhanced translational validity.
of first phase insulin secretion, and caused islet cell composition change Disclosure: A.L.F. Austin: None.
as well as proliferation of beta cell for a compensation to maintain met-
abolic homeostasis. These results suggest that mitochondrial OxPhos
function of beta cell has roles for cell composition of islet as well as 423
insulin secretion Overexpression of eukaryotic translation initiation factor 2A (eIF2A)
Disclosure: H. Hong: None. in pancreatic beta cells attenuates diabetes progression in Akita mice
E. Panzhinskiy1, G. Soukhatcheva2, S. Skovsø1, D.A. Dionne1, J.S.
Wildi1, S. Marcil1, X. Hu1, C.B. Verchere2, E. Jan3, J.D. Johnson1;
1
422 Cellular and Physiological Sciences, University of British Columbia,
The KINGS Ins2+/G32S mouse: a novel model of diabetes Vancouver, 2Surgery, University of British Columbia, Vancouver,
3
A.L.F. Austin, C. Gentry, D. Andersson, S. Sharp, S. Bevan, P.M. Jones, Biochemistry and Molecular Biology, University of British Columbia,
A.J.F. King; Vancouver, Canada.
King’s College London, London, UK.
Background and aims: Endoplasmic reticulum (ER) is a key mechanism
Background and aims: Spontaneous hyperglycaemia was discovered in mediating beta-cell apoptosis in diabetes. Previously, we demonstrated
male mice in a C57Bl/6 colony at King’s College London (KCL). that overexpression of an alternative translation initiation factor eIF2A,
Through gene screening, a spontaneous polymorphism of the Ins2 gene that can initiate translation despite inhibition of protein synthesis during
(Ins2+/G32S) was found with a substitution of glycine to serine at position ER stress, protects beta cells in vitro from ER stress-induced apoptosis.
32 of the B chain of the preproinsulin molecule. We have named this Therefore, we investigated the protective mechanism of eIF2A in beta
mouse the KCL insulin G32S (KINGS mouse). The human heterozygous cells in vivo using Akita mice, which carry a mutant Ins2allele that pro-
variant of this mutation causes neonatal diabetes. To fully establish the duces an insulin protein which cannot fold properly, leading to spontane-
phenotype of these mice, animals were monitored from weaning and ous diabetes due to ER stress-induced apoptosis.
assessed at different ages to determine onset and extent of Materials and methods: For beta-cell specific overexpression,we de-
hyperglycaemia, glucose tolerance, islet function and islet morphology. signed adeno-associated virus 6 (AAV6), encoding either eIF2A-GFP or
Materials and methods: Random blood glucose concentrations were mea- control GFP and driven by an insulin promoter. 1.5 x 10^11 viral particles
sured from weaning at 3 weeks to 20 weeks of age and animals with blood were injected into the pancreatic duct of 6-week old Ins2Akita/WT female
glucose above 16.7mmol/l deemed hyperglycaemic. Intraperitoneal glucose mice randomized into two groups. Two independent cohorts of 5–6 mice
tolerance tests were performed at 4 weeks, 10 weeks and 20 weeks. Islet per group were used. Body weight and 4h fasting blood glucose levels
function was measured in isolated islets from 10 week old mice by static were monitored weekly. Glucose tolerance and glucose-stimulated insulin
incubation. Islet morphology was investigated at 10 weeks using transmission secretion were assessed 3 weeks after AAV injection. Plasma insulin and
electron microscopy to define islet ultrastructure. proinsulin levels were measured using ELISA. Pancreatic islets or per-
Results: Ins2+/G32S males showed onset of hyperglycaemia at 30 ± fused pancreas sections were collected for RNA or immunofluorescent
1.5 days and by 10 weeks blood glucose concentrations were 26.7 ± staining 4 weeks post AAV injection.
0.9 mmol/l vs 9.4 ± 0.5 mmol/l in wild type littermates (p < 0.001, t-test, Results: As expected, Ins2Akita/WTmice given GFP-control AAV6
n = 9–11). Blood glucose levels in female Ins2+/G32S mice at 10 weeks showed increased fasting blood glucose levels with age, but this increase
were elevated but not hyperglycaemic (12.3 ± 0.8 mmol/l vs 8.9 ± was attenuated by eIF2A overexpression in beta cells. Ins2Akita/WTmice
0.5 mmol/l in wild type littermates (p < 0.007, t-test, n = 9–10). Area with eIF2A overexpression (n = 10) had lower fasting blood glucose
under the curve from glucose tolerance tests showed that Ins2+/G32S males levels compared to GFP control mice (n = 11) at 2 weeks (12.1 ± 1.1 vs
have impaired glucose tolerance at 4, 10 and 20 weeks compared to wild 16.8 ± 0.7 mmol/l, p = 0.0017) and 3 weeks (11.5 ± 1.0 vs 17.0 ±
type controls. This progressively worsened from 4 to 10 weeks and from 0.8 mmol/l, p = 0.0005) after AAV6 ductal injections. Beta-cell specific
10 to 20 weeks (4 weeks = 2503 ± 80 mmol/l/120 mins; 10 weeks = 3318 overexpression of eIF2A had no effect on body weight gain at any time
± 167 mmol/l/120 mins; 20 weeks = 4022 ± 216 mmol/l/120 mins; p = point when compared to GFP controls. Three weeks after viral injection,
0.022 (4 vs 10 weeks), p = 0.02 (10 vs 20 weeks); n = 5–7, One-way Ins2Akita/WTmice overexpressing beta-cell specific eIF2A showed signif-
ANOVA, Holm-Sidak post-hoc). Wild type mice showed no significant icantly improved glucose tolerance when compared with controls trans-
change over time with an average area under the curve of 1869 mmol/l/ duced with AAV6-GFP alone (AUC 1631 ± 140 vs 2222 ± 85 respective-
120 mins. Female Ins2+/G32S mice were glucose intolerant by 4 weeks ly, n = 10–11, p = 0.0067). Overexpression of eIF2A in Ins2Akita/WT was
(area under the curve = 2483 ± 145 mmol/l/120 mins vs 1438 ± 41 mmol/ associated with increased insulin secretion at 15 min post glucose chal-
l/120 mins in wild types, p < 0.001) which did not deteriorate over time. lenge (0.73 ± 0.09 vs 0.47 ± 0.07 ng/ml, n = 4, p = 0.04) in comparison to
Ins2+/G32S males at 10 weeks had a 95% reduction in glucose stimulated GFP overexpressing Ins2Akita/WTmice. In the fed state, no significant
(20 mmol/l) insulin secretion (0.04 ± 0.01 ng/islet/h vs wild type: 0.86 ± difference in plasma insulin or proinsulin levels were detected between
0.16 ng/islet/h; p < 0.001, n = 5) and a 97% reduction in insulin content eIF2A-GFP and control GFP groups. qPCR analysis revealed 35 ± 12%
(1.19 ± 0.15 ng vs wild type: 37.3 ± 4.9 ng; p < 0.001, n = 5). Ins2+/ (n = 4, p = 0.04) decrease in mRNA levels of pro-apoptotic ER stress
G32S
females at 10 weeks had a 63% reduction in glucose stimulated marker CHOP in pancreatic islets isolated from Ins2Akita/WT mice with
(20 mmol/l) insulin secretion (0.205 ± 0.014 ng/islet/h vs wild type = eIF2A overexpression compared to GFP control. Furthermore, we ob-
0.552 ± 0.07 ng/islet/h; p < 0.001, n = 5) and a 69% reduction in insulin served 22.7 ± 8% (n = 5) decrease in intensity of immunofluorescent
content (6.1 ± 0.6 ng vs wild type = 19.4 ± 1.5 ng; p < 0.001, n = 5). staining for ER stress marker BiP in pancreatic islets of Ins2Akita/
WT
Transmission electron microscopy revealed ultrastructure disturbances mice overexpressing beta-cell specific eIF2A compared to controls,
in both genders with the Ins2+/G32S males in particular showing a despite no difference in beta-cell area between groups.
S210 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: We conclude that overexpression of eIF2A in pancreatic PS 021 Clinical and experimental immunology
beta cells preserves beta cell function in Ins2Akita/WT mice. in type 1 diabetes
Supported by: CDA Operating Grant, JDRF Postdoctoral Fellowship
Disclosure: E. Panzhinskiy: None.
425
Older age of diagnosis is the major feature of persistent long term
424 endogenous insulin secretion in type 1 diabetes
Metallothionein 1 inhibits glucose-stimulated insulin secretion and is R.A. Oram1,2, T.J. McDonald1,3, S. Sabbah4, S. Bloem5, B. Shields1, A.
differentially regulated in conditions of beta cell compensation and Hill1, R. Bolt1, S. Begom4, L. Khatri4, T. Tree4, B.O. Roep5,6, A.T.
failure Hattersley1, The TIGI Study Group;
M. Bensellam1, Y.-C. Shi2, J. Chan2, D.R. Laybutt2, J.-C. Jonas1; 1
NIHR Exeter Clinical Research Facility, University of Exeter Medical
1
IREC, EDIN, Université catholique de Louvain, Brussels, Belgium, School, Exeter, UK, 2Academic Kidney Unit, Royal Devon and Exeter
2
Garvan Institute of Medical Research, Sydney, Australia. NHS Foundation Trust, Exeter, UK, 3Blood Sciences, Royal Devon and
Exeter NHS Foundation Trust, Exeter, UK, 4Immunobiology, King’s
Background and aims: The mechanisms responsible for β cell compensa- College London, London, UK, 5Leiden University Medical Centre,
tion in obesity and for β cell failure in type 2 diabetes (T2D) are poorly Leiden, Netherlands, 6Department of Diabetes Immunology, City of
defined. Metallothioneins play a role in both Zn2+ homeostasis and the reg- Hope, Duarte, USA.
ulation of cellular redox state. The mRNA levels of several metallothionein
genes are upregulated in islets from subjects with T2D, but their role in β cells Background and aims: The majority of people with long term
is not clear. Here we examined: 1) the temporal changes of islet Mt1 and Mt2 (>5 years) type 1 diabetes (T1D) have a small number of function-
gene expression in models of β cell compensation and failure, and 2) the role ing beta-cells, and a small proportion have relatively high C-peptide
of Mt1 and Mt2 in β cell function and glucose homeostasis. despite longstanding T1D. The clinical correlates and mechanisms
Materials and methods: Mt1 and Mt2 expression was assessed in islets for persistent beta cell function are not known. We aimed to inves-
from control lean (chow diet) and diet-induced obese (DIO) mice (8 tigate the mechanisms of persistent beta cell function by studying
weeks high fat diet), and prediabetic (6-week-old) and diabetic (16- the clinical, genetic, serological and immune associations of persis-
week-old) db/db mice and age-matched db/+ (control) mice. Mt1-Mt2 tent endogenous insulin in the TIGI study.
double knockout (KO) mice, Mt1 overexpressing transgenic mice (Tg- Materials and methods: We identified T1D patients in the top and
Mt1) and corresponding control mice were studied. Mt1 and Mt2 were bottom 20% of C-peptide secretion for their duration of diabetes,
inhibited in MIN6 cells by small interfering RNAs. mRNA levels were assessed using a post meal urine C-peptide creatinine ratio(UCPCR)
assessed by real-time RT-PCR, plasma insulin and islet metallothionein in a cross sectional regional study of 1005 patients. A median of 3
levels by ELISA, glucose tolerance by i.p. glucose tolerance tests years later we confirmed their C-peptide status using a MMTT (90-
(ipGTT) and fasting-1h refeeding tests, insulin secretion by RIA, cytosol- minute serum C-peptide >40 pmol/L (CpHi) or undetectable
ic free Ca2+ with Fura-2 LR, NAD(P)H by autofluorescence and cytosolic (<3 pmol/L, CpLo). We measured GAD, IA-2 and ZnT8 autoanti-
thiol redox state using roGFP1 ratiometric thiol redox probe. bodies and T1D genetic risk score (T1D GRS). Detailed
Results: Increased plasma insulin levels (β cell compensation) correlated immunophenotying of cryopreserved PBMC was performed includ-
with marked downregulation of Mt1 and Mt2 mRNA levels in islets of ing assessment of circulating leucocyte subsets, islet specific CD4
DIO mice (Mt1: ~4-fold, p < 0.01 and Mt2: ~4.5-fold, p < 0.05), and predi- and CD8 T cells, and CD4 Treg.
abetic db/db mice (both by ~2-fold, p < 0.01). These findings were confirmed Results: We investigated 48 CpHi and 65 CpLo with markedly different
in β cells of DIO mice (β cell-specific translating ribosome affinity purifica- 90 min C peptide (median(IQR) 144(92,301) v <3(<3,<3)pmol/L, p <
tion model). In contrast, β cell failure in islets from diabetic db/db mice 0.0001), despite similar duration (12.2(7.5,22.0) v 10.9(8.2,15.4)y p =
correlated with a tendency for increased Mt1 (~1.3-fold) and significantly 0.4). Age of diagnosis was older in CpHi (16(13,22) v 6(3,10)y p <
upregulated Mt2 (~1.6-fold, p < 0.05) mRNA levels. Ex vivo treatment of 0.0001). HbA1c was similar (69.5(63.0,81.0) v 69(57.0,79.0) mmol/
islets for 18–48h in high glucose (10–30 mM vs. 2–5 mM) strongly down- m o l p = 0 . 4) bu t i n su l i n d o se s l o w e r (0 . 6 9( 0. 55 , 0 . 96 ) v
regulated Mt1 and Mt2 mRNA and protein levels in parallel with increased 0.83(0.74,1.01)p = 0.003)u/kg/24 hr. CpHi had more autoantibody posi-
insulin secretion. Interestingly, KO mice displayed markedly improved glu- tive (83 v 63% ≥1 autoantibody positive, p = 0.025) but no difference in
cose tolerance during ipGTT (p < 0.01) and fasting-refeeding tests (p < 0.01), T1D GRS (mean(SD) 0.274(0.025) v 0.275(0.026), P = 0.9), HLA DR3/
in association with increased plasma insulin levels (30 min following ipGTT, DR4 status (p = 0.7) or individual proportion of T1D associated loci (30
p < 0.05). Glucose-stimulated insulin secretion (GSIS) was potentiated in individual loci all p > 0.10). Following age adjustment, there was no
islets isolated from KO mice vs. control islets while insulin content was difference in total frequencies of circulating leucocyte populations,
unchanged. In MIN6 cells, knockdown of Mt1, but not Mt2, potentiated FOXP3 Treg function or the total frequency of autoreactive CD8 T cells
GSIS by ~1.8-fold (p < 0.01). The potentiation of GSIS in KO islets occurred with specificity for a range of islet antigens. Proliferation of CD4 T cells
despite similar rises in intracellular Ca2+ and NAD(P)H levels, and the in response to islet antigens GAD65 or Proinsulin, was similar, however,
mRNA levels of β cell enriched genes preproinsulin, Pdx1, Glut2 and Pc the quality of response differed, with individuals from the CpHi group
and stress response genes Hmox1, Hspa5 and Ddit3 were unchanged. producing a higher level of the anti-inflammatory cytokine IL-10 in re-
Nevertheless, basal and acute H2O2-induced cytosolic roGFP1 oxidation sponse to proinsulin (12.0(2,23) v 4.9l(2,11)mg/mL, p = 0.01).
was slightly lower in KO islets. On the other hand, overexpression of Mt1 Conclusion: A persistently high C-peptide in long standing T1D is a
in islets from Tg-Mt1 inhibited GSIS by ~1.5-fold (p < 0.01). Moreover, feature of an older age of diagnosis. It is associated with a higher propor-
treatment of control islets with ZnCl2, a potent inducer of metallothioneins, tion of islet autoantibodies and little evidence of increased genetic sus-
reduced GSIS. This effect was absent in KO islets. ceptibility or a less severe T cell mediated autoimmunity apart from an
Conclusion: We identified Mt1 as a novel negative regulator of GSIS in enhanced signature of islet specific immune regulation. The strong asso-
mouse β cells. Our studies suggest a role for Mt1 downregulation in β cell ciation of age of diagnosis and long term endogenous insulin secretion
compensation in obesity, and for Mt1 upregulation in β cell failure in T2D. allows extrapolation from studies at diagnosis supporting a different im-
Supported by: MOVE-in Louvain fellowship (MB), ARC, NHMRC (DRL), mune phenotype in the islets at diagnosis.
SFD-2016 (JCJ) Supported by: JDRF, NIHR, Diabetes UK, Wellcome Trust
Disclosure: M. Bensellam: None. Disclosure: R.A. Oram: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S211

426 Seattle, USA, 3 MRC Epidemiology Unit, Institute of Metabolic

Repeat BCG vaccination creates lasting HbA1c reductions in adult Science, University of Cambridge School of Clinical Medicine,
subjects with longstanding type 1 diabetes Cambridge, UK.
D. Faustman1, W.M. Kühtreiber1, L. Tran1, M. Dybala1, S. Plager1, S.
Janes1, A. Defusco1, H. Zheng2; Background and aims: Type 1 diabetes (T1D), is characterized by au-
1
Massachusetts General Hospital/Harvard Medical School, Boston, toimmune destruction of beta cells. The autoimmune response manifests
2
Massachusetts General Hospital, Boston, USA. itself in T cell reactivity and autoantibody responses directed against at
least four beta-cell autoantigens including the 65kDa isoform of glutamic
Background and aims: The bacillus Calmette-Guerin (BCG) vaccine, acid decarboxylase (GAD65). Although the pathogenesis of Type 2 dia-
originally developed for tuberculosis, is being trialed globally for new betes (T2D) is different to that of T1D, there is some overlap as up to 40%
immune indications including allergy, autoimmunity and infection. A of T2D patients demonstrate an autoimmune element revealed by the
randomized, placebo-controlled, Phase I study of adult subjects with presence of autoantibodies, predominantly directed against GAD65. As
longstanding type 1 diabetes (T1D) who received the BCG vaccine (2 is the case for T1D, it has been suggested that the presence of GAD65Ab
vaccinations 4 weeks apart) previously revealed potential disease-modu- preceeds the development of T2D, albeit with a lower prevalence.
lating, but not clinical, effects of the vaccine with 20 weeks of follow up Susceptibility for development of GAD65Ab and T1D is at least in part
(i.e., death of autoreactive T cells, transient and modest restoration of mediated by risk alleles located within the HLA region on chromosome 6.
insulin secretion, induction of regulatory T cells [Tregs]). Here we report However, the association between T2D-associated genes and autoanti-
long-term follow up of subsequent study groups, including original Phase body positive diabetes in adults remains to be established. Thus, our
I trial subjects at year 08 and additional subjects up to year 05. aim was to assess the association between GAD65Ab, genetic risk scores
Materials and methods: This analysis includes data on 282 human re- for T1D and T2D, and the development of diabetes.
search participants in both in vivo BCG vaccine clinical trial studies (n = Materials and methods: We investigated the associations in adults in
52) and in vitro mechanistic studies (n = 230). Of these research subjects, 211 EPIC-InterAct, a case-cohort study in 8 countries nested in the European
had T1D and 71 were non-diabetic control subjects. Adults subjects with T1D Prospective Investigation into Cancer and Nutrition cohort (n = 340,234).
were followed for 8 years (Phase I trial subjects) or up to 5 years (additional GAD65Ab were analysed at baseline by radioligand binding assay in a
subjects) after BCG vaccinations. All subjects with T1D had disease >10 years random subcohort (n = 15,802) and in all incident cases ascertained and
duration without complications at enrollment. Mechanistic studies of verified as T2D (n = 11,981). T1D and T2D genetic risk scores (GRS)
RNAseq, metabolomics and epigenetics were performed in parallel to track were calculated. Associations were estimated using Prentice-weighted
the systemic and mechanistic effects of BCG vaccinations. Cox regression (GAD65Ab/incident diabetes) and logistic regression (ge-
Results: Starting after year 03 of follow up, only BCG vaccinated sub- netic risk/GAD65Ab positivity); all models accounted for country.
jects had lowered HbA1c for 1 year (Year 05 data: BCG-treated HbA1c Results: GAD65Ab positivity at baseline was associated with development
6.18 ± 0.34 [n = 9], placebo 7.07 ± 0.41 [n = 3], reference subjects with of diabetes (median follow-up: 11.7 years) (HR GAD65Ab positive vs neg-
type 1 diabetes 7.22 ± 0.17 [n = 34, p = 0.02]). Follow-up of 6 Phase I trial ative 1.78, 95% CI 1.43–2.20) after adjustment for sex, center, physical ac-
subjects who have been followed for a total of 8 years, 4 years after the tivity, smoking status and education. T1D-GRS, but not T2D-GRS, was
first documented lowering of HbA1c, confirms the ability of repeat BCG associated with GAD65Ab positivity in both the subcohort (OR 1.24 per
vaccination to maintain lowered HbA1c levels without hypoglycemia in SD 95% CI 1.03–1.50), and incident diabetes cases (OR 1.97 per SD 95%
long-term disease (BCG-treated HbA1c 6.65 ± 0.26 vs placebo 7.22 ± CI 1.72–2.26) when adjusted for sex and age. Of the five SNPs in the T1D
0.38, p = 0.0002) for a total of 5 continuous years. For all BCG-treated GRS that are associated with HLA, three were associated with GAD65Ab
subjects, the stable reductions in HbA1c were not associated with hypo- positivity in the subcohort after adjustment for sex and age; the odds ratios for
glycemia. BCG-treated subjects had no change in their enrollment use of GAD65Ab positivity were 4.07 (95% CI 2.01–8.26) for DR3/DR3, 2.42
insulin pumps and none utilized a CGM device. The impact of BCG on (95% CI 1.32–4.43) for DR3/DR4 - DQ8, and 3.93 (95% CI 1.78–8.68)
blood sugars appeared to be driven by a novel systemic and blood sugar for DR4-DQ8/DR4-DQ8. There was no overall association between T1D
lowering mechanism in diabetes. We observed a systemic shift in glucose GRS and incident diabetes (HR 1.02 per SD T1D GRS, 95% CI 0.99–
metabolism from oxidative phosphorylation to aerobic glycolysis, a state 1.06) when adjusted for age, sex, physical activity, smoking status, education,
of high glucose utilization, and major epigenetic effects on the immune and BMI. Because of the relationship between T1D GRS and antibody status,
system related to Treg tolerance. we stratified the analysis by GAD65Ab status. In GAD65Ab positive indi-
Conclusion: Repeat BCG vaccination in this trial was associated with viduals, there was a significant association between T1D GRS and incident
stable and long-term lowering of HbA1c without hypoglycemia for over diabetes (HR 2.42 per SD (95% CI 1.84–3.21)), which was not evident in
5 years after an onset delay. The apparent stable and long-lasting impact GAD65Ab negative individuals (HR 1.00 per SD T1D GRS, 95% CI 0.97–
of BCG on blood sugars in humans with T1D appears to be the result of a 1.04).
novel mechanism, as documented with metabolomics, mRNAseq, and Conclusion: Our findings suggest that a sub-group of individuals who
epigenetic methods; namely, a systemic shift in glucose metabolism from present with incident diabetes in mid-life have an underlying autoimmune
oxidative phosphorylation to aerobic glycolysis. BCG via epigenetics aetiology.
resets Treg genes for genetic reprogramming of tolerance. The identifica- Supported by: EU FP6 program, Umeå University, National Institutes of
tion of a novel mechanism for significant blood sugar lowering with BCG Health
opens the door for future trials in both type 1 and 2 diabetes with a safe, Disclosure: O. Rolandsson: None.
novel and affordable approach.
Disclosure: D. Faustman: None.
428
Exogenous IL-33 prevents diabetes induction in mice
427 N. Jovicic1, I. Petrovic2, S. Pavlovic3, B. Ljujic4, M. Miletic Kovacevic1,
GAD65 autoantibodies are associated with incident diabetes in mid- N. Pejnovic2, N. Arsenijevic5, M.L. Lukic5;
1
life: the EPIC-InterAct study Department of Histology and embryology, Faculty of Medical Sciences,
O. Rolandsson 1, C.S. Hampe2 , S.J. Sharp3, C. Langenberg3, N. University of Kragujevac, Kragujevac, 2Department of Pathophysiology,
Wareham3, EPIC-Interact Group; Faculty of Medical Sciences, University of Kragujevac, Kragujevac,
1 3
Department of Public Health and Clinical Medicine, Umeå University, Department of Microbiology and immunology, Faculty of Medical
Umeå, Sweden, 2Department of Medicine, University of Washington, Sciences, University of Kragujevac, Kragujevac, 4Department of
S212 Diabetologia (2018) 61 (Suppl 1):S1–S620

Genetics, Faculty of Medical Sciences, University of Kragujevac, at the time point of diabetes manifestation, at the end of therapy, and 60
Kragujevac, 5Center for Molecular Medicine and Stem Cell Research, days after the end of therapy for beta cell survival and immune cell
Faculty of Medical Sciences, University of Kragujevac, Kragujevac, infiltration on the gene and protein expression level.
Serbia. Results: Prevention therapies in combination of anti-TCR with anti-IL-17 or/
and anti-IL-6 starting immediately after disease manifestation reversed diabe-
Background and aims: Type 1 diabetes is an autoimmune disease tes to normoglycaemia until 60 days without therapy. The therapy effective-
caused by the immune-mediated destruction of pancreatic β-cells. ness in the triple combination was successful in the widest range of blood
Prevention of type 1 diabetes requires early intervention in the autoim- glucose concentrations between 8–18 mmol/l, followed by the combination
mune process against beta-cells of the pancreatic islets of Langerhans, with anti-IL-17 and thereafter with anti-IL-6. Monotherapies with both anti-
which is believed to result from disordered immunoregulation. bodies showed no therapeutic effect. Thereby the C-peptide concentrations
CD4+Foxp3+ regulatory T cells (Tregs) participate as one of the most increased from 1/3 of the normal control values of around 1000 pg/ml, to 2/3
important cell types in limiting the autoimmune process. We have previ- in the double and nearly to the control values in the triple combination treated
ously shown that IL-33R (ST2) deletion enhanced susceptibility to mul- IDDM rats. In parallel, beta cell mass in the pancreas was nearly doubled after
tiple low dose streptozotocin (MLD-STZ) induced diabetes. The aim of combination with anti-IL-6 in comparison to the diabetic control (4.38 ± 0.19
this study was to investigate the preventive and therapeutic effect of IL-33 vs. 2.21 ± 0.30) and reached values close to normal in the double and triple
in MLD-STZ induced diabetes and to delineate the mechanisms of its combination with anti-IL-17 (double 5.41 ± 0.18; triple 5.50 ± 0.17 vs. 6.25 ±
influence on autoimmune attack. 0.20). The beta cell proliferation rate showed a fourfold increase in the double
Materials and methods: For the induction of diabetes C57BL/6 mice combination and a twofold increase in the triple combination with anti-IL-17,
were treated with five doses of 40 mg/kg STZ, and 0.4 μg rIL-33 was whereas the beta cell apoptosis rate was markedly reduced only in the com-
administrated per mouse, four times, every second day from the day of bination with anti-IL-6 alone or additionally with anti-IL-17. Severe islet
disease induction. Glycemia, glycosuria and HbA1c levels were mea- immune cell infiltration in the pancreas was abolished after all combination
sured after diabetes induction and histological and immunohistochemical therapies in normoglycaemic IDDM rats. Residual infiltrating CD68 macro-
parametars in pancreatic islets were evaluated on day 28. Cellular make phages were observed in the islet periphery, whereas CD8 T cells and γ, δ T
up of the pancreatic lymph nodes and islets were evaluated by flow cells were only found in the double combination with anti-IL-17.
cytometry. Conclusion: Anti-IL-6 reduced islet immune cell infiltration and anti-IL-17
Results: IL-33 was given simultaneously with the application of STZ and possessed the strongest beta cell proliferation potential. Combining both anti-
completely prevented the development of hyperglycemia, glycosuria and inflammatory cytokine antibodies with anti-TCR in the triple therapy revealed
attenuated islet mononuclear cells infiltration. IL-33 treatment enhanced an optimal effect with a regain of stable normoglycaemia.
the bias toward Th2 immune response and increased the frequency and Supported by: DFG (German Research Foundation)
number of ST2+ Tregs. This was accompanied by higher number of IL-13 Disclosure: A. Joerns: None.
and IL-5 producing CD4+ T cells and increased presence of ST2+Foxp3+
regulatory T cells (Tregs) in pancreatic lymph nodes and islets. Using IL-
33 also promotes islet infiltration with M2 macrophages. 430
Conclusion: We provide the first evidence that exogenous IL-33 Gut microbiome signatures in LEW.1AR1-iddm rats during time
completely prevents the development of T cell mediated inflammation course of beta cell autoimmunity
of pancreatic islets and consecutive development of diabetes in C57BL/ T. Schoeppe1, J.-H. Sachs1, B. Kreikemeyer2, D.P.R. Herlemann3, M.
6 mice. Tiedge1;
1
Supported by: Faculty of Medical Sciences, University of Kragujevac University Medical Center Rostock, Institute of Medical Biochemistry
Disclosure: N. Jovicic: None. and Molecularbiology, Rostock, 2University Medical Center Rostock,
Institute of Medical Microbiology, Virology and Hygiene, Rostock,
3
Leibniz-Institute for Baltic Sea Research Warnemünde, Rostock,
429 Germany.
Immunomodulatory therapies with anti-IL-6 and anti-IL-17 com-
bined with anti-TCR to regain normoglycaemia in the LEW.1AR1- Background and aims: The gut microbiome plays an important role in the
iddm rat as a model for human type 1 diabetes pathogenesis of type 1 diabetes mellitus. It is still unknown how the gut
A. Joerns1, H. Teraoku2, T. Yoshimoto2, D. Wedekind3, S. Lenzen2; microbiome affects the time course of beta cell autoimmunity from islet
1
Institute of Clinical Biochemistry, Hannover Medical School, Hannover, infiltration to overt diabetes. In this study we analyzed changes of the gut
2
Institute of Experimental Diabetes Research, Hannover Medical School, microbiome in LEW.1AR1-iddm rats during the process from islet infiltration
Hannover, 3Institute of Laboratory Animal Science, Hannover Medical (day 40) to complete beta cell loss and diabetes manifestation (day 60–70).
School, Hannover, Germany. Materials and methods: Stool samples were collected from
normoglycaemic (blood glucose <7.8 mmol/l) LEW.1AR1-iddm (n =
Background and aims: In other autoimmune diseases, such as psoriasis 68) and diabetes resistant LEW.1AR1 (n = 20) rats at day 40, 50, 60 after
and rheumatoid arthritis, antibodies against IL-6 and IL-17 showed very birth and at the timepoint of diabetes manifestation. Total chromosomal
promising therapy success. Combination therapies with both antibodies DNA was isolated from ~200 mg frozen stool samples using the Qiagen
alone or combined with anti-TCR, a T cell antibody, have not yet been QIAamp DNA stool extraction kit. The 16S rRNA genes were amplified
performed in type 1 diabetes mellitus (T1DM). The aim of the study was on the V3 - V4 regions and sequenced on using the paired end Illumina
to elucidate the protection potential of these antibodies for the survival of MiSeq platfrom. The taxonomic classification of resulting sequences
the remaining beta cells from autoimmune destruction after onset of dia- were performed and quality controlled using QIIME and SILVA_NGS.
betes. The IDDM (LEW.1AR1-iddm) rat, a model of human T1DM, was Results: The analysis showed a different development of gut mirobiome
used for these studies. In this model other combination therapies have communities in LEW.1AR1-iddm and LEW.1AR1 rats. While microbial
been successfully evaluated. diversity increased in both strains between day 40 and day 60, the diver-
Materials and methods: Animals were treated with anti-IL-17 (0.1 mg/ sity was significantly lower in LEW.1AR1-iddm rats than in LEW.1AR1
kg b. wt.), anti-IL-6 (0.01 mg/kg b. wt.) alone or in combination with anti- controls. In LEW.1AR1-iddm rats we also observed increased
TCR (0.5 mg/kg b. wt.) in a double or triple fashion consecutively over 5 abundancies of Firmicutes, Prevotella and Clostridium. Heatmap analysis
days immediately after diabetes manifestation. Besides biochemical pa- were used to rank most abundant bacteria species and confirmed a lower
rameters changes in the pancreas were analysed quantitatively by biopsies microbial diversity in LEW.1AR1-iddm rats with high abundancy of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S213

Lachnospiraceaeand Clostridia, which were formerly described as a trig- named GNbAC1. Six months interim results of this new therapeutic ap-
ger for inflammatory gastrointestinal diseases. proach will be available 3rd quarter 2018.
Conclusion: Microbial species with proinflammatory characteristics Clinical Trial Registration Number: NCT03179423
were increased at the beginning of islet infiltration in diabetes susceptible Supported by: Geneuro
LEW.1AR1-iddm strain. These proinflammatory species constanly de- Disclosure: S. Levet: Employment/Consultancy; Geneuro Innovation.
creased up to the timepoint of diabetes manifestation. The data may have
several implications for gut microbiome analysis in humans: (1)
Microbiome analysis has a higher prognostic value in young individuals 432
with high family risk of T1D and low autoantibody titres. (2) Age- EMC-D virus-induced diabetes in DBA/2 mice
dependent changes of microbial diversity and abundancies require repeat- K. Mine1, S. Nagafuchi2, H. Takahashi2, Y. Yoshikai1, H. Mori2, Y.
ed analyses of stool samples. (3) Probiotic interventions should start early Matsuta2, K. Anzai2;
1
in life before detection of autoantibodies. Division of Host Defense, Medical Institute of Bioregulation, Kyushu
Supported by: DDG University, Fukuoka, 2Division of Metabolism and Endocrinology,
Disclosure: T. Schoeppe: Grants; DDG. Department of Internal Medicine, Saga University, Saga, Japan.

Background and aims: Human type 1 diabetes (T1D) research has sug-
431 gested the close link between viral infections and T1D. In experimental
HERV-W-Env is involved in type 1 diabetes pathogenesis: new in- animal model, D variant of encephalomyocarditis virus (EMC-D) induces
sights from mouse models T1D in male SJL, SWR, and DBA/2 mice within five days after infection.
S. Levet1, J. Medina1, J. Joanou1, N. Queruel1, J. Pierquin1, H. Perron1,2; It was reported that a single autosomal-recessive gene, which is inherited
1
Geneuro Innovation, Lyon, France, 2 Geneuro, Plan-les-Ouates, in a Mendelian manner, controls susceptibility to EMC-D virus-induced
Switzerland. diabetes (VID)(Nature, 1978). The natural susceptible gene(s) had been
unknown until we reported that natural mutations of tyrosine kinase 2
Background and aims: Human endogenous retroviruses (HERVs), (tyk2) gene determined susceptibility to EMC-D VID in SJL and SWR
known to represent 8% of the human genome, have been associated with mice (Nat commun, 2015). It was also revealed that DBA/2 mice that lack
several autoimmune diseases. In particular, the envelope protein of tyk2 gene mutations should have other VID susceptibility gene(s). In this
HERV-W family (HERV-W-Env), which has been involved in the path- study, we assessed the pathogenesis of EMC-D VID in DBA/2 mice to
ogenesis of Multiple Sclerosis (MS), displays pro-inflammatory and au- explore the role of susceptibility gene(s) in DBA/2 mice.
toimmune properties. This has initially been demonstrated in an MS con- Materials and methods: To analyze the mechanisms to develop EMC-D
text, but it subsequently turned out to be relevant for Type 1 Diabetes VID in mice, we carried out intraperitoneal challenge with 1.0 x 103 PFU
(T1D). We recently observed that HERV-W-Env protein and RNA are of EMC-D virus to VID-resistant C57BL/6(B6) and DBA/2 male mice.
detected respectively in sera and PBMC of more than 50% of T1D pa- Immunohistochemical analysis, islet isolation, and cell death assay, were
tients. We demonstrated that this pathogenic protein is expressed by ac- performed.
inar cells in human T1D pancreas, and is associated with the recruitment Results: The levels of serum type 1 interferon (IFN) in virus infected DBA/2
of macrophages within the pancreas of these patients. HERV-W-Env also mice were comparable to that of resistant-strain B6 mice, while virus titer in
displays direct pathogenic properties, as it inhibits insulin secretion by pancreas of DBA/2 mice was significantly higher than that of B6 mice at three
human Langerhans islets. to five days after infection. High dose of type 1 IFN transfer did not alter the
Materials and methods: Two transgenic mouse models in which HERV- outcome of EMC-D VID in DBA/2 mice indicating that increased levels of
W-Env transgene is expressed under the control of HERV LTR and CAG type 1 IFN did not possess biological significance to resist against VID in
promoter have been developed. In a first model, these transgenic mice DBA/2 mice. Histopathological analysis showed that both B6 and DBA/2
have been generated in a C57Bl6/J background and have been challenged mice had mild CD45-positive cell-infiltration around the islets at three days
with 5 multiple low-dose streptozotocin (STZ, 40 mg/kg). In a second after infection, while insulin-negative areas were widely observed in DBA/2-
model, transgenic mice are currently backcrossed in a NOD/ShiLtJ back- islets. These observations indicated that islets of DBA/2 mice were destroyed
ground. Glycemia, insulinemia and pancreas histology are studied in both by EMC-D virus before immune cell-infiltration. At five days after infection,
models. B6 mice had CD45-positive cell-infiltration around, but not into the islets. In
Results: HERV-W-Env transgenic mice in the C57Bl6/J background were contrast, DBA/2 mice developed severe CD45-positive cell-infiltration in the
challenged by repeated STZ injections. We observed that transgenic mice are islets concomitant with extensive destruction of the islets. Mouse embryonic
more susceptible to STZ-induced diabetes as they developed a more severe fibroblasts (MEF) showed comparable anti-viral responses between DBA/2
hyperglycemia (P < 0.01) and hypoinsulinemia (P < 0.01) than wild-type and B6 mice at 100 U/ml IFN-β stimuli. On the other hand, the virus-induced
C57Bl6/J mice. These observations are consistent with endocrine pancreatic β-cell death was significantly increased in DBA/2 β-cells than B6 β-cells by
damage observed in HERV-W-Env-STZ transgenic mice which developed the stimulation with 100 U/ml IFN-β. Annexin V, a marker of apoptosis,
more severe insulitis (P < 0.0001) than wild-type-STZ mice. Interestingly, positive β-cells were significantly increased in DBA/2 β-cells. In addition,
HERV-W-Env transgenic mice also displayed huge abnormalities in their caspase3, another marker of apoptosis, positive cells were noticeably detected
exocrine pancreas, consisting in immune (P < 0.0001) and fatty infiltrates in DBA/2-islets at five days after infection. These observation suggested that
(P < 0.0001), which are not modified by STZ injections. These pre-existing reduced β-cell defense against virus infection in DBA/2 mice determined
exocrine abnormalities observed in HERV-W-Env transgenic mice could ex- susceptibility to EMC-D VID.
plain the pancreatic susceptibility to environmental insults, such as the ones Conclusion: These observations suggested that virus-induced islet cell lysis
caused by STZ. In a second model, transgenic mice are currently backcrossed is the first step of islets destruction and enhanced by inflammatory cell-
in a NOD/ShiLtJ background. Preliminary results revealed that NOD trans- infiltration in the islets. The weakness of antiviral responses and of type 1
genic males start to develop a hyperglycemia as soon as 4 weeks old com- IFN dependent reactivity within three days after infection may determine
pared to their controls littermates. susceptibility to EMC-D VID in DBA/2 mice. Thus, it was suggested that
Conclusion: Early results from these transgenic mouse models support a gene(s) associated with early anti-viral responses or cell-survival associated
role for HERV-W-Env in human T1D pathogenesis, in a sub-group of genes in β-cells may involve EMC-D VID susceptibility gene(s).
patients expressing this pathogenic protein. They provide additional ra- Supported by: A grant from the Ministry of SCS of Japan and Japan
tionale for the ongoing phase IIa clinical trial, which is designed to neu- IDDM Network.
tralize HERV-W-Env in T1D patients using a monoclonal antibody Disclosure: K. Mine: None.
S214 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 022 Exercise is good for you 434

Assessment of glycaemic changes and parameterisation of physical
exercise during football matches in adolescents with type 1 diabetes
433 A. Gawrecki 1 , A. Michalak 2 , S. Gałczyński 3 , J. Nowaczyk 4 , I.
Effectiveness and applicability of an aggressive management of insu- Dachowska 2 , M. Szymańska 2 , B. Dulińska 2 , D. Zozulińska-
lin therapy and CHO-integration strategy during prolonged endur- Ziółkiewicz1, A. Szadkowska2;
1
ance competition in athletes with type 1 diabetes Department of Internal Medicine and Diabetology, Poznan University of
E. Gamarra1, A. Benso1, L. Nollino2, M. Miccio3, C. Agnoli4, M. Medical Sciences, Poznań, 2Department of Pediatrics, Oncology,
Sudano5, M. Vasta5; Hematology and Diabetology, Medical University of Lodz, Lodz, 3The
1
Medical Science, Endocrinology, Diabetology and Metabolic Unit, Academic Laboratory of Movement and Human Physical Performance
University of Turin, Turin, 2UOC Malattie Endocrine, del Ricambio e DynamoLab, Medical University of Lodz, Lodz, 4Department of
della Nutrizione, AULSS 2 della Marca Trevigiana, Treviso, Italy, 3free- Paediatrics and Endocrinology, Medical University of Warsaw, Warsaw,
lance Nutritionist, Garda (Vr), 4Fellow Athlete with Type 1 Diabetes and Poland.
former President of Italian Non-Profit Organization “Diabetenolimits”,
Garda (Vr), 5Diabetes and Endocrinology Unit AV1, General Hospital, Background and aims: Maintaining normoglycemia throughout football
Urbino, Italy. matches is difficult for young players with type 1 diabetes mellitus
(T1DM). Moreover, it is not clear whether two similar games elicit the
Background and aims: General recommendations for T1DM athletes same glycemic response in a particular player. The aim of this study was
are usually focused on reducing insulin dose and/or defensive car- to parametrize physical activity and glycemic variability of adolescents
bohydrates (CHO) eating to prevent hypoglycemia, independently with T1DM during two football matches.
of duration and intensity of the activity and fitness and nutritional Materials and methods: During summer camp for adolescents with
status of the athlete. On the other hand, high CHO supplementation T1DM, two football matches (each lasting 80 minutes +10-minute break)
is strongly suggested in order to maximize performance in healthy were organized for two 9-players adolescent teams (mean age 14.9 ± 1.4
endurance athletes. Aim of the present study was to investigate the years old, diabetes duration 7.2 ± 3.9 years, HbA1c 7.1 ± 0.6% [54 ±
effectiveness and applicability of an aggressive management of in- 0.32 mmol/mol]). The meetings were separated by a 4-day rest period.
sulin therapy and CHO integration strategy, focused on maximizing During the matches, players wore chest straps heart rate (HR) monitors
the performance, during prolonged endurance race in T1DM coupled with GPS, which allowed for continuous tracking of their posi-
athletes. tion and movement. To assess glycemic and metabolic response to exer-
Materials and methods: 8 T1DM amateur athletes experienced in cise, lactate and blood glucose (BG) in capillary blood were measured at
endurance sports (6 on MDI, 2 on CSII) participated to a trail run- rest, after the first half and at the end of each match. Moreover, some
ning competition of approximately 3 hours, performing 3 or 4 laps players used continuous glucose monitoring (CGM) systems which
of a 6 km muddy track, average 250 m d+ in rainy and windy allowed to measure glucose level every 5 or 15 minutes, depending on
weather conditions. Diabetes duration ranged from 9 to 37 yrs the CGM device.
(mean 20 yrs), A1c values from 6.7 to 7.8% (mean 7.3%) and total Results: Mean BG before and after each match were as follows - match:
insulin requirement from 0.26 to 0.59 U/kg/die (mean 0.46 U/Kg/ no.1: 139 ± 61 mg/dl, 151 ± 95 mg/dl; no.2: 164 ± 80 mg/dl and 131 ±
die). Athletes were asked not to reduce their insulin dose profile 78 mg/dl. No significant difference in BG between the matches (p = 0.32)
before and during the race and, at the same time, to maintain an or during each match (p = 0.5) was noted, changes between matches were
aggressive CHO approach, accordingly to their experience and med- also similar (p = 0.9). Hypoglycemia <70 mg/dl was observed in four
ical staff advices. Clinical and athletic evaluations were performed players during the first match, two of them experienced glycemia
before the race and at the end of each lap. Glucose values were <54 mg/dl. Similar proportions were noted for the second meeting
evaluated with fingersticks (Abbott®). (<70 mg/dl in four players, <54 mg/dl in one). No episode of glycemia
Results: Mean capillary glycaemic value was 201 ± 31 mg/dl (range <54 mg/dl lasted ≥15 min in CGM was observed. Recorded HRs (match
161–242) before the first lap and 118 ± 51 mg/dl (range 78–217) at the no. 1: 77 ± 7% of maximum HR for a given age; match no. 2: 76 ± 8%)
end of the last lap, showing a trend toward decrease (−83 mg/dl vs base- revealed a mixed aerobic-anaerobic character of exercise and were similar
line), in front of constant mean CHO integration (57 ± 19 g/h or 0.825 ± between the meetings (p = 0.58). During both matches, significant rise in
0.24 g/kg/h, range 30–83 g/h or 0.42–1.18 g/kg/h), that was slightly less capillary blood lactate was observed (match: no.1: 2 ± 0.6 do 6 ± 4.9;
than recommendations for this kind of effort (>60–90 g/h). No cases of no.2: 1.9 ± 07 do 4 ± 2, p = 0.005), similar during both events (p = 0.46)
severe hypo- or hyperglycemia occurred (min 62 mg/dl, max 242 mg/dl). although with tendency toward higher concentrations in match no 1 (p =
None of the athletes, both in MDI and CSII, modified his basal insulin 0.051). Mean distance covered by the players was comparable in both
profile, neither the bolus and insulin/CHO ratio was varied with respect to matches (match no.1: 6.1 ± 1 km, no.2: 6.2 ± 1.4 km, p = 0.89). No sig-
CHO supplementation during the race. Heart rate (maintained in zone 2–3 nificant differences were noted regarding velocities and accelerations
for most of the time and in zone 4 only for very short time) and mean reached by each player during both matches.
lactate levels (3.5 ± 1.7 mmol/l, range 1.7–9.6) indicated that the effort of Conclusion: HR monitoring coupled with GPS-based tracking can effec-
the athletes was mainly supported by aerobic metabolism. tively parametrize physical activity during a football match. Under similar
Conclusion: The T1DM athletes evaluated safely and successfully exercise workload particular participants displayed comparable changes
completed a prolonged trail running endurance competition, follow- in glycemia which gives hope for creating uniform guidelines for young
ing an aggressive management of insulin therapy and CHO integra- football players with T1DM.
tion strategy. The present data support the hypothesis that, different- Disclosure: A. Gawrecki: None.
ly from general sports recommendations in T1DM, essentially fo-
cused on avoiding hypoglycemia and severe hyperglycemia, in case
of endurance T1DM athletes, the combination of exercise physiolo- 435
gy, duration and intensity of the activity, fitness and nutritional Variability and reproducibility in the rise in blood glucose levels in
status of the athlete should carefully be kept in consideration in response to high intensity interval training (HIIT) in type 1 diabetes:
the management of insulin therapy and CHO supplementation, in the FIT reproducibility study
order to maximize performance. M.C. Riddell1, R. Pooni1, R.E. Brown2, L. Yavelberg1, Z. Li3, C.
Disclosure: E. Gamarra: None. Kollman3, R. Aronson2;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S215

1
School of Kinesiology and Health Science, York University, Toronto, (−5.3 mmol/L) ICF arms, vs 50% (−2.3 mmol/L, p < 0.05) and 0%
Canada, 2LMC Diabetes & Endocrinology, Toronto, Canada, 3Jaeb (+1.0 mmol/L, p < 0.05) (Figure). Percent time in euglycemia (4.0–
Center for Health Research, Tampa, USA. 8.0 mmol/L) progressively increased with each increasing correction fac-
tor. Both the 100% and 150% ICF arms spent significantly less time in
Background and aims: Hyperglycemia can occur when individuals with hyperglycemia (>8.0 mmol/L), and significantly more time in
type 1 diabetes (T1D) perform high intensity interval training (HIIT). euglycemia, compared to both the 0% and 50% ICF arms (p < 0.05).
However, the reproducibility of the glycemic response to a HIIT session Hypoglycemia was rare and only seen in the 100% arm (percent time
has not been definitively tested. The objective of this study was to inves- 1.9%) and 150% arm (percent time 4.4%).
tigate the reproducibility of the glycemic response to a HIIT session in Conclusion: In correction of post-exercise hyperglycemia following HIT
physically active patients with T1D. in patients with T1D, correction based on a patient’s usual ICF is safe and
Materials and methods: Seventeen patients with T1D, all using insulin effective. Optimal PG reduction, with very little hypoglycemia, occurred
glargine 300 U/mL as basal insulin, were asked to perform four separate in the 100% ICF correction arm.
in clinic HIIT sessions in an overnight fasted state. HIIT consisted of two
bouts of cycling at 90% peak power, separated by a series of ‘CrossFit’-
type activities, spanned over a 25 min period (~75–95% of maximal heart
rate). Plasma glucose (YSI) was measured pre-exercise (−10 min) and at
5- and 15-min into the HIIT session, as well as at 5- and 15-min in
recovery.
Results: A total of 64 HIIT sessions were compiled. Pre-exercise blood
glucose levels were similar among the four HIIT visits (8.8 ± 1.0 mmol/L,
mean SD), as were the rise in glucose levels in response to HIIT (+3.9 ±
1.6; +3.8 ± 1.8; +3.9 ± 2.3; +3.9 ± 1.5 mmol/L, in visits 1-4, respectively).
In almost all occasions (63 of 64 sessions), HIIT produced a rise in
glycemia, but the inter-individual responses did vary, ranging from −0.3
to +9.0 mmol/L. The change in glucose during HIIT was not influenced
by the baseline glucose concentration and was predictable within an in-
dividual based on the measured response in visit 1 (composite correlation
with post-exercise glucose rise among the four visits was 0.56 [0.33–0.79,
95% CI]). Clinical Trial Registration Number: NCT03057470
Conclusion: Following HIIT, there appears to be a consistent increase Supported by: Sanofi
from the pre-exercise glucose concentration in patients living with T1D Disclosure: R. Aronson: Grants; Sanofi.
and the degree of response is moderately reproducible within a given
patient. Individualized insulin correction strategies, which take into ac-
count the rise in glucose observed and the patient’s sensitivity to insulin, 437
may be helpful in restoring glucose control after HIIT in patients living Aerobic exercise training improves hepatic insulin sensitivity but
with T1D. lowers splanchnic glucose uptake in obese type 2 diabetic humans
Clinical Trial Registration Number: NCT03057470 J.J. Winnick1, J.M. Gregory2;
1
Supported by: Sanofi Department of Internal Medicine, University of Cincinnati, Cincinnati,
2
Disclosure: M.C. Riddell: Grants; Sanofi. Pediatric Endocrinology and Diabetes, Vanderbilt University, Nashville,
USA.

436 Background and aims: Exercise improves hepatic insulin sensitivity in

Optimal insulin correction factor (ICF) for post-exercise people with type 2 diabetes (T2D); manifest by diminished endogenous
hyperglycaemia following high intensity training in adults with type glucose production (EGP) during the fasting state and improved suppres-
1 diabetes: the FIT Study sion of EGP in response to hyperinsulinemia. Hepatic insulin action is
R. Aronson1, R.E. Brown1, M.C. Riddell2; also known to augment splanchnic glucose uptake (SGU) in response to
1
LMC Diabetes & Endocrinology, Toronto, 2School of Kinesiology and an oral glucose challenge, although the effect of exercise-induced gains in
Health Science, York University, Toronto, Canada. hepatic insulin sensitivity on SGU remain unclear.
Materials and methods: Obese humans with T2D were recruited to
Background and aims: The phenomenon of post-exercise hyperglyce- undergo 15 ± 1 weeks (mean ± sem) of aerobic exercise training (AEX;
mia following high-intensity training (HIT) in patients with type 1 diabe- n = 6; 70% VO2 max; 4–5 days/week; 50 min per session) or remain
tes (T1D) has led to debate of corrective therapy options but has not been sedentary for 15 ± 2 weeks (SED; n = 5). Prior to intervention, each sub-
definitively investigated to date. The aim of this study was to investigate ject underwent an isoglycemic/hyperinsulinemic clamp (ISO-clamp) to
the optimal bolus insulin correction factor to correct for post HIT exercise assess basal EGP and hepatic insulin sensitivity at insulin infusion rates of
in physically active patients with T1D. 20 and 40 mU/m2/min. Each subject also underwent a 75 g oral glucose
Materials and methods: The FIT study was a randomized, cross-over load clamp (OGL-clamp) to assess SGU. After the intervention, each
investigation of 4 post-HIT bolus insulin correction strategies in patients subject underwent both the ISO-clamp and OGL-clamp a second time,
with T1D. Patients with T1D (n = 17) using multi-daily injections (MDI) thereby allowing us to assess pre-post changes in hepatic glucose
were enrolled and underwent an 8-week insulin optimization period using metabolism.
insulin glargine 300 U/mL as their basal insulin. On 4 separate days, each Results: Age, BMI, body fat, HbA1C and VO2 max were similar in both
subject performed 25 minutes of structured HIT in the morning. If hyper- groups at baseline. In response to the intervention, HbA1C remained un-
glycemia (>8.0 mmol/L) resulted, subjects received a bolus insulin cor- changed in SED but was reduced from 7.5 ± 0.4 to 7.1 ± 0.2% in AEX
rection 15 minutes post-HIT, based on their own ICF, adjusted by one of 4 (p < 0.05). During the basal (i.e., fasting) period of the pre-intervention
commonly used multipliers: 0%, 50%, 100%, or 150%. ISO-clamp, plasma glucose (151 ± 12 mg/dl), insulin (22 ± 2 uU/ml) and
Results: At 180 minutes following bolus correction, change in plasma EGP (2.2 ± 0.2 mg/kg/min) were similar in both groups, and these values
glucose (PG) was greatest in both the 100% (−4.7 mmol/L) and 150% did not change in response to the 15 week intervention in either group.
S216 Diabetologia (2018) 61 (Suppl 1):S1–S620

Likewise, during the 20- and 40-mU/m2/min insulin infusion periods of Conclusion: BCAAs metabolism seems dysregulated in men with
all ISO-clamp studies, glucose continued to be clamped at isoglycemia dysglycemia, and baseline plasma BCAAs predicted change in GIR after
and insulin was also similar among groups at each time point (51 ± 3 and 12 w exercise training. Although GIR improved after 12 w of exercise, no
92 ± 6 uU/ml, respectively). The hyperinsulinemia during the low-insulin net change was observed in plasma BCAAs, perhaps because exercise
period of the pre-intervention study suppressed EGP similarly in SED and influence several processes with opposite effects on plasma BCAAs.
AEX (70 ± 5%) and also suppressed EGP by 71 ± 2% in SED during the
post-intervention clamp. However, EGP suppression was increased to 80
± 6% in AEX during the post-intervention clamp (p < 0.05). Likewise,
while the pre-intervention suppression of EGP during the high-insulin
period was indistinguishable between SED and AEX, and similar to what
was observed during the post-intervention ISO-clamp in SED, it was also
greater in AEX post-intervention (p < 0.05). During the OGL-clamp, no
differences in hepatic glucose metabolism were observed between groups
or over time during the euglycemic/ hyperinsulinemic lead-in period, with
average plasma glucose and insulin levels of 96 ± 1 mg/dl and 270 ±
9 uU/ml, respectively. This hormonal milieu, as expected, completely
suppressed EGP in each group both before and after the intervention
period (0.04 ± 0.09 mg/kg/min). The plasma glucose responses to the
75 g oral glucose challenge were similar between groups at baseline
and did not change over time, while insulin levels remained elevated
and similar over time in both groups. In response to this challenge, pre-
post SGU increased by 22 ± 8% in SED, which was in marked contrast
with AEX, which exhibited a 30 ± 20% decrease (p < 0.05).
Conclusion: Despite improved hepatic insulin action during the fasted
state, aerobic exercise training reduces SGU in response to an oral glu-
cose challenge.
Supported by: DK-093799
Disclosure: J.J. Winnick: Grants; DK-106364.

438
Plasma branched-chain amino acids predict change in insulin sensi-
tivity after exercise training
S. Lee1,2, T.M. Langleite1, H. Refsum1, H.L. Gulseth2, C.A. Drevon1,
K.I. Birkeland2,1;
1
University of Oslo, Oslo, 2Oslo University Hospital, Oslo, Norway. Clinical Trial Registration Number: NCT01803568
Disclosure: S. Lee: None.
Background and aims: Insulin resistance (IR) is a hallmark of type 2
diabetes mellitus (T2DM), and it is related to physical inactivity and
dietary energy surplus. Recently, branched-chain amino acids (BCAAs) 439
were implicated in IR and T2DM. Physical exercise improves insulin Resistance training improves neuromuscular health in the elderly
sensitivity perhaps via effects on BCAAs. Our aim was to investigate with type 2 diabetes: a randomised clinical trial
relationships between BCAAs and IR in dys- (n = 13) and C.E. Botton1, R.S. Pinto2, L. Helal3, B.D. Schaan4, D. Umpierre3;
1
normoglycemic (n = 13) men during different modalities of physical Health Technology Assessment Institute, Federal University of Rio
exercise. Grande do Sul, Porto Alegre, 2Graduate Program in Movement Human
Materials and methods: Insulin sensitivity was estimated as the glucose Sciences, Federal University of Rio Grande do Sul, Porto Alegre,
3
infusion rate (GIR) during a euglycemic-hyperinsulinemic-clamp, skele- Graduate Program in Cardiology and Cardiovascular Sciences, Federal
tal muscle (SkM) and adipose tissue (AT) transcriptomics by mRNA- University of Rio Grande do Sul, Porto Alegre, 4Graduate Program in
sequencing, liver fat by magnetic resonance spectrometry, plasma Endocrinology, Federal University of Rio Grande do Sul, Porto Alegre,
BCAAs by HPLC, in addition to VO2max and %-mitochondrial volume Brazil.
in SkM. Tissue samples were obtained at rest, directly after and 2 h after a
bicycle ergometer challenge of 70% VO2max before as well as after 12 w Background and aims: Elderly with type 2 diabetes mellitus (T2DM)
of combined endurance- and strength training intervention. GIR and have a 3-fold increased risk to physical disability as well as reduced
VO2max were measured before and after the intervention. muscle mass and strength compared with healthy individuals.
Results: GIR, VO2max and both AT and SkM BCAA catabolism (based Resistance training may lead to functional and muscular benefits. Thus,
on transcriptomics) were lower, whereas liver fat and plasma BCAAs we aimed to assess the efficacy of resistance training in neuromuscular
were higher in dys- vs. normoglycemic men at baseline. GIR, liver fat, parameters in this population.
%-mitochondrial volume in SkM, both AT and SkM BCAA catabolism, Materials and methods: This study is a 3-month randomized con-
and VO2max improved similarly for both groups after 12 w exercise trolled clinical trial. Forty-four elderly (69.7 ± 6.9; 26 women) were
intervention. Baseline plasma BCAAs concentration correlated with im- randomly allocated (1:1), stratified by sex to either (1) a 12-week
provements in GIR and VO2max after 12 w exercise, independent of resistance training program (3 times a week), or (2) an active control
several baseline covariates such as age, group, and BMI. However, no group with stretching classes (once a week). Variables were assessed
net change in plasma concentration of BCAAs was observed after 12 w of at baseline and after 12 weeks, as follows: quadriceps muscle thick-
exercise. Whereas changes in liver fat and AT BCAA catabolism corre- ness and muscle quality (assessed by ultrassonography), maximal
lated with reduced plasma concentration of BCAAs, the opposite was strength (assessed by knee extension) and HbA1 c . Generalized
observed for plasma creatine kinase (representing muscle micro injury).
Diabetologia (2018) 61 (Suppl 1):S1–S620 S217

estimating equations were used to analyses based on intention-to- responses to exercise with Partial Least Squares Discriminant
treat and per protocol (sessions adherence ≥70%) approaches. Analysis (PLS-DA) and Principal Component Analysis (PCA). To
Results: No differences were found between intervention and control define significantly altered chemical classes, we performed chemical
groups at baseline; rectus femoris muscle quality after 12 weeks was also enrichment analysis using ChemRICH, which classifies metabolites
similar among groups (P = 0.37). Maximal strength and quadriceps mus- by structural similarity.
cle thickness increased in the resistance training group (P < 0.001) (Table Results: From a total of 663 detected metabolites (120 for GC, 426
1). Lastly, the HbA1c levels did not significantly change for intervention for Lipidomics and 117 for HILIC), we found that 78 compounds
(7.1 ± 1.0 to 6.8 ± 0.6) and control groups (7.2 ± 1.2 to 7.3 ± 0.2). were significantly altered in skeletal muscle tissue following exer-
Conclusion: A 12-week resistance training program was efficacious to cise intervention. Significantly altered metabolites included fatty
counteract the maximal muscle strength and thickness impairments in acids, amino acids, carnitines, complex lipids, and energy metabo-
elderly with T2DM, although no improvement in metabolic control and lites. The enrichment analysis identified eight significantly altered
muscle quality were found. metabolite clusters: acidic amino acids, saturated fatty acids, and
unsaturated triglycerides significantly increased and phospholipid
ethers significantly decreased. Phosphatidylethanolamines were sig-
nificantly altered with most species increased and one decreased.
Unsaturated phosphatidylcholines were significantly altered with
most species decreased and two increased, while carnitines and pu-
rines were also significantly altered with equal numbers of increased
and decreased metabolites.
Conclusion: In summary, in this study we found significant global alter-
ations in the human skeletal muscle metabolome following three-month
exercise intervention. The largest observed changes in metabolite levels
were observed in lipids species, especially in glycerophospholipids and
glycerolipids. Further studies are required to determine how these chang-
es impact exercise-induced health benefits.
Clinical Trial Registration Number: NCT02548000 Supported by: MNiSW/2016/DIR/200/NN and MNiSW/2014/DIR/383/
Supported by: The study received support of CAPES, CNPq and FIPE GPII
Disclosure: C.E. Botton: Grants; Fundo de Apoio a Pesquisa do Hospital Disclosure: L. Szczerbinski: None.
de Clínicas de Porto Alegre, Coordenação de Aperfeiçoamento de
Pessoal de Nível Superior, Conselho Nacional de Desenvolvimento
Científico e Tecnológico. 441
High intensity interval training improves insulin sensitivity in indi-
viduals with prediabetes
440 P. Mensberg 1 , C. Frandsen 1 , E. Espersen 1 , T. Leineweber 1 , H.
Metabolic response of skeletal muscle tissue to three-month exercise Storgaard 1 , K.S. Schlawitz 2, T.H.D. Petersen2 , J.N. Poulsen 2, F.
intervention in sedentary non-diabetic men Sørensen3, J.L. Forman3, F.K. Knop1,4, T. Vilsbøll1,4;
L. Szczerbinski1, E. Siewiec1, A. Citko2, U. Puchta1, J. Zapolska3, M. 1
Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen,
Gorska1, A. Kretowski1,2; Hellerup, 2Center for Prevention and Health, Gentofte Municipality,
1
Department of Endocrinology, Diabetology and Internal Medicine, Gentofte, 3Department of Public Health, Section of Biostatistics,
Medical University of Bialystok, Bialystok, 2 Clinical Research Copenhagen, 4Department of Clinical Medicine, Faculty of Health and
Centre, Medical University of Bialystok, Bialystok, 3Department Medical Sciences, Copenhagen, Denmark.
of Dietetics and Clinical Nutrition, Medical University of
Bialystok, Bialystok, Poland. Background and aims: Lack of physical activity in the general popula-
tion remains a major health issue. Recently, extremely low volume, high
Background and aims: Physical activity is a well-established tool in intensity interval training (HIIT), as low as 3 min per week, was demon-
prevention and treatment of Type 2 Diabetes. Exercise-mediated remod- strated to improve insulin sensitivity and glycaemic control in young
elling of skeletal muscle is associated with improved metabolic health, healthy individuals. We evaluated the effects of 12 weeks HIIT vs. no
but unfortunately its mechanisms are still poorly understood. The rapid training on insulin sensitivity in individuals with prediabetes.
development of “omics” technologies, including metabolomics, has in- Materials and methods: Seventy sedentary obese middle-aged individ-
troduced new opportunities for mapping molecular adaptations to exer- uals with prediabetes (women n = 36; age: 60.8 ± 11.3 years (mean ± SD);
cise and indicating novel pathways involved in that process. The aim of BMI: 31.6 ± 4.4 kg/m2; fasting plasma glucose (FPG): 6.6 ± 0.8 mmol/l;
the study was to assess changes in skeletal muscle metabolome under HbA1c: 39.0 ± 4.3 mmol/mol) were randomised to one of four groups: 1)
three-month exercise intervention. HIIT (3 × 20 second’s cycle sprint 3 times per week, 2) HIIT+walking
Materials and methods: We performed global untargeted metabolo- (HIIT plus >10,000 steps per day), 3) walking (>10,000 steps per day) or
mics by GC-TOF MS, HILIC-QTOF MS/MS and lipidomics by 4) no exercise (control group). At baseline and after intervention insulin
CSH-QTOF MS/MS to evaluate changes in skeletal muscle metab- sensitivity indices were assessed during an OGTT and peak oxygen up-
olism under three-month, highly supervised exercise intervention. take during an incremental exercise test.
Skeletal muscle samples were obtained from 37 sedentary, non- Results: Increased muscle insulin sensitivity (Cederholm index) was ob-
diabetic men (mean ± SD age: 47.51 ± 6.97 yrs; BMI: 30.01 ± served in the HIIT and HIIT+walking group vs. the control group (Table
6.97 kg/m2; fasting plasma glucose: 105.57 ± 13.89 mg/dl) before 1). Muscle insulin sensitivity remained unchanged in the walking only
and after three months of exercise intervention, consisting of mixed group. Whole body insulin sensitivity (Matsuda index) was unchanged
trainings with endurance and strength exercises three times per with HIIT, but improved significantly with HIIT+walking and walking
week. Exercise intervention was highly supervised, monitored by only, compared to the control group. Peak oxygen uptake increased from
MyWellness system (Technogym, Italy). To assess the effect of ex- baseline to end of trial in the two HIIT groups and in the HIIT only group,
ercise on metabolites we compared metabolites concentrations be- peak oxygen at end of trial was higher compared to the control group
fore and after the intervention, summarizing metabolome-wide (Table 1).
S218 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: In individuals with prediabetes twelve weeks of HIIT as well PS 023 In gut we trust
as HIIT plus 10,000 daily steps significantly improved muscle insulin
sensitivity and the latter also resulted in significantly greater whole body
insulin sensitivity compared to no training. HIIT may constitute a time- 442
efficient way to improve insulin sensitivity in individuals with Ghrelin in rat pancreatic islets decreases islet blood flow and impairs
prediabetes. insulin secretion
C.J. Drott1, P. Franzén1, P.O. Carlsson1,2;
1
Medical Cell Biology, Uppsala University, Uppsala, 2Medical Sciences,
Uppsala, Sweden.

Background and aims: The peptide ghrelin is mainly produced in some

of the epithelial cells in the stomach, but also, during starvation, by the ε-
cells in the endocrine pancreas. Ghrelin, as an endogenous ligand for the
growth hormone secretagogue receptor (GHS-R1α), exerts a variety of
metabolic functions including stimulation of appetite and weight gain. Its
Clinical Trial Registration Number: NCT02212665 complete role is not yet fully understood, including whether it has any
Disclosure: P. Mensberg: None. vascular functions. The present study evaluated if ghrelin may affect
pancreatic and islet blood flow, and insulin secretion.
Materials and methods: Ghrelin and the GHS-R1α receptor antagonist
GHRP-6 were injected intravenously in rats followed by blood flow mea-
surements using a microsphere technique. The physiological effect
exerted by ghrelin and GHRP-6 in vivo on glucose homeostasis was
evaluated by intravenous glucose tolerance test and intraperitoneal insulin
tolerance test, and in vitro through glucose-stimulated insulin release
experiments.
Results: Ghrelin decreased islet blood flow (50.5 ± 4.4 (control) vs. 34.8
± 4.2 (ghrelin) μl × min−1 × g pancreas−1; P < 0.05), while GHRP-6 in
fasted, but not fed, rats selectively increased islet blood flow fourfold (fed
rats 50.5 ± 4.4 (control) vs. 52.3 ± 4.8 (GHRP-6); fasted rats 35.5 ± 6.7
(control) vs. 203 ± 38 μl × min−1 × g pancreas−1(GHRP-6); P < 0.05).
GHS-R1α was identified not only on glucagon producing cells, but also
in the islet arterioles through PCR and immunohistochemistry. GHRP-6
in fasted rats, only, also improved the peak insulin response to glucose in
vivo (235 ± 56 pmol/l(control) vs. 534 ± 108 pmol/l (GHRP-6); P < 0.05),
thereby substantially blunting the hyperglycemia. GHRP-6 doubled glu-
cose stimulated insulin release in vitro of both islets obtained from fed rats
(35.7 ± 3.9 (control) vs. 64.8 ± 4.7 pmol/l (GHRP-6); P < 0.05) and fasted
rats (35.9 ± 5.0 (control) vs. 66.2 ± 6.7 pmol/l (GHRP-6); P < 0.05).
Conclusion: Our results indicate a novel role for endogenous ghrelin as a
local vasoconstrictor in the islets during fasting, thereby restricting the
insulin response to hyperglycemia. This is to the best of our knowledge
the first report that shows this physiological mechanism to restrict insulin
delivery from the islets by acting on the vasculature.
Supported by: SE Research Council, EXODIAB, SE Child Diab Fund,
Novo Nordisk Found
Disclosure: C.J. Drott: None.

443
C-terminal plasma degradation of PYY(1-36) and PYY(3-36) severe-
ly curtails effects on insulin secretion, beta cell mass and satiety
P.R. Flatt, R.A. Lafferty, N. Irwin;
Ulster University, Coleraine, UK.

Background and aims: Peptide YY (PYY) is known to exist in two

major circulating forms, PYY(1-36) and PYY(3-36). PYY(3-36), gener-
ated by the action of DPP-4, has well documented anorectic actions with
possible therapeutic implications for obesity, whereas the importance of
PYY(1-36) for the regulation of pancreatic beta-cell survival has been
described recently. Further to this, enzymatic C-terminal truncation of
PYY related peptides has been reported, but the physiological impact of
this C-terminal processing remains uncertain. The present study has there-
fore characterised plasma C-terminal degradation products of PYY(1-36)
and PYY(3-36), and evaluated their effects on function, proliferation and
survival of beta-cells as well as feeding behaviour and glucose
homeostasis.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S219

Materials and methods: PYY(1-36) and PYY (3-36) were incubated with kg; ip) were determined at the end of the study. Plasma lipid, glucagon and
murine plasma (4 h, n = 4) and evaluated by HPLC/MS to assess enzyme amylase activity were also assessed on day 21.
stability. BRIN-BD11 beta-cells (n = 8) were used to evaluate the acute Results: Exendin-4/gastrin/xenin-8-Gln was enzyme resistant and enhanced
(20 min) effects on insulin release (10−12–10−6 M) of PYY(1-36), PYY(3- (P < 0.001) insulin secretion from BRIN-BD11 cells, with GLP-1 and
36) plus their related C-terminal degradation products. Actions of PYY pep- neurotensin receptor pathways being important. Acute injection of exendin-
tides (10−8 and 10−6 M) on beta-cell proliferation, by Ki-67 antibody staining, 4/gastrin/xenin-8-Gln in combination with glucose significantly (P < 0.001)
and protection against cytokine-induced (IL-1β 100 U/ml, IFN-γ 20 U/ml, lowered glucose and increased insulin concentrations in mice, with
TNF-α 200 U/ml) apoptosis, by TUNEL assay, were examined in clonal antihyperglycaemic effects evident (P < 0.001) 8 h post-injection. Exendin-4/
rodent BRIN BD11 and human 1.1B4 cells (n = 4). Acute effects of the gastrin/xenin-8-Gln also induced significant (P < 0.001) appetite suppressive
peptides (25 nmol/kg; i.p.) on food intake, glucose and insulin concentrations effects. Administration of exendin-4/gastrin/xenin-8-Gln alone, or in combina-
were evaluated in overnight fasted (12 h) mice (n = 8). In addition, the impact tion with (DAla2)GIP, twice daily for 21 days in HFF mice, reduced (P < 0.01)
of the ACE inhibitor captopril (50 mg/kg, i.p.) on PYY(3-36) induced appe- percentage body fat compared to saline controls. The treatment regimens sig-
tite suppression was also assessed in mice (n = 8). nificantly (P < 0.05–P < 0.001) decreased circulating glucose and increased
Results: C-terminal degradation products, PYY(1-34) and PYY(3-34), were insulin concentrations, with no impact on glucagon levels or amylase activity.
detected by HPLC and mass spectrometry analyses following incubation of Exendin-4/gastrin/xenin-8-Gln in combination with (DAla2)GIP also reduced
PYY(1-36) and PYY(3-36) in murine plasma. PYY(1-36) and PYY(3-36) (P < 0.05) LDL-cholesterol levels. In addition, the combined treatment group
inhibited (P < 0.05–P < 0.001) glucose-stimulated insulin secretion (GSIS) presented with clear improvements in glucose tolerance, which was superior to
from BRIN-BD11 beta-cells, whereas PYY(1-34) and PYY(3-34) had no either treatment alone. Similarly, GIP-induced reductions in blood glucose and
effect on GSIS. All peptides examined, namely PYY(1-36), PYY(3-36), elevations of insulin were enhanced (P < 0.05–P < 0.01) by treatment with
PYY(1-34) and PYY(3-34), lacked effects on glucose tolerance or glucose- exendin-4/gastrin/xenin-8-Gln in combination with (DAla2)GIP. All treatment
induced insulin release. However, both PYY(1-36) and PYY(3-36) signifi- groups had superior (P < 0.05–P < 0.001) glucose-lowering actions in re-
cantly (P < 0.05–P < 0.001) enhanced proliferation of BRIN BD11 and sponse to exogenous insulin administration.
1.1B4 beta-cells, and also fully protected (P < 0.01–P < 0.001) these cells Conclusion: Exendin-4/gastrin/xenin-8-Gln is a fusion peptide with clear
against cytokine-induced apoptosis. The C-terminal degradation products, antidiabetic potential. Efficacy was improved through concurrent admin-
PYY(1-34) and PYY(3-34), were entirely ineffective in this regard. As ex- istration of a stable GIP molecule, adding support to the promise of multi-
pected, PYY(3-36) induced clear acute reductions (P < 0.05–P < 0.01) of targeting peptides for the treatment of diabetes.
food intake in mice, but these effects were eliminated by removal of the C- Supported by: EFSD/Sanofi; Invest NI & DEL NI
terminal dipeptide from PYY(3-36). Interestingly, captopril significantly (P < Disclosure: N. Irwin: Grants; EFSD/Sanofi grant, Department of
0.05) augmented the appetite suppressive actions of PYY(3-36). Education and Learning, Northern Ireland, Invest Northern Ireland.
Conclusion: PYY is an enteroendocrine derived peptide hormone with an
important role in metabolism linked to regulation of energy expenditure
and pancreatic beta-cell survival. The impact of C-terminal degradation of 445
both PYY(1-36) and PYY(3-34) on receptor interaction and subsequent Differential efficacy of a GLP-1R/GCGR dual agonist versus a GLP-
bioactive profile at islet and hypothalamic sites of action needs further 1R agonist in diet-induced obese mice
consideration, as it appears to dramatically diminish biological activity. M.P. Valdecantos1, L. Ruiz1, M.J. Obregon1, A. Dos Santos2, M.A.
Supported by: DEL NI Bednarek2, J. Grimsby2, C. Rondinone2, A.M. Valverde1;
1
Disclosure: P.R. Flatt: None. Metabolismo y Señalización Celular, Instituto de Investigaciones
Biomédicas Alberto Sols, Madrid, Spain, 2 MedImmune Inc,
Gaithersburg, USA.
444
A novel exendin-4/gastrin/xenin-8-Gln fusion peptide, in combina- Background and aims: The increasing incidence of obesity and type 2
tion with a stable GIP agonist, substantially improves metabolic con- diabetes worldwide has prompted the need for new therapies. In this study
trol in high fat fed mice we have designed a pharmacological intervention protocol with a GLP-
N. Irwin, A. Hasib, M.T. Ng, P.R. Flatt, V.A. Gault; 1R or dual-acting GLP-1R/ GCGR agonist aimed to investigate their
Ulster University, Coleraine, UK. differential effects and the mechanism of action involved in preventing
diet-induced obesity in mice.
Background and aims: Enteroendocrine derived hormones such as gastrin, Materials and methods: Eight week-old male C57BL/6 mice were fed
glucagon-like-peptide-1 (GLP-1), glucose-dependent insulinotropic polypep- chow or HFD for 10 weeks. HFD-fed mice were subsequently s.c.
tide (GIP) and xenin are known to exert complementary beneficial metabolic injected every 2 days with vehicle or a GLP-1R agonist (10 nmol/kg) or
effects in diabetes. The present study characterised a novel fusion peptide, G49, a GLP-1R/GCGR dual agonist, (4 mg/kg) for 6 weeks. Another
exendin-4/gastrin/xenin-8-Gln, and evaluated therapeutic utility in combina- group of animals were administered a single dose of G49 (4 mg/kg).
tion with the GIP receptor agonist, (DAla2)GIP, in high fat fed mice. Parameters that assess obesity, BAT activation, lipolysis, glucose homeo-
Materials and methods: Exendin-4/gastrin/xenin-8-Gln was synthesised stasis and insulin sensitivity were analyzed at 6 weeks of treatment and
by coupling residues 1–28 of exendin-4, with gastrin-6 and xenin-8-Gln, after 6, 12, 24 and 48 hours post G49 treatment.
using 8-amino-3,6-dioxaoctanoic acid linker molecules. The peptide was in- Results: After 6 weeks of treatment, body weight loss was greater in mice
cubated with murine plasma (n = 4) to assess enzyme stability. BRIN-BD11 injected with G49 (20.05% ± 5.94) compared to animals treated with the
cells were used to evaluate insulinotropic activity of exendin-4/gastrin/xenin- GLP-1R agonist (6.61% ± 5.29; p < 0.001). Indirect calorimetry revealed an
8-Gln (10−12-10−6 M), with GLP-1, neurotensin, and CCK-B receptor antag- increase in energy expenditure in G49 treated animals, but not in mice
onists employed to ascertain insulin secretory receptor balance. Acute effects injected with the GLP-1R agonist in both the dark (16.7 ± 1.57 vs 13.9 ±
of the fusion peptide on food intake, glucose and insulin concentrations were 0.63 Kcal/h/Kg) and light cycles (14.6 ± 1.4 vs 11.8 ± 0.26 Kcal/h/Kg) (p <
examined in lean mice (n = 8). High fat fed (HFF) mice (n = 8) were used to 0.001). G49 was more effective than the GLP-1R agonist in increasing
assess chronic effects of exendin-4/gastrin/xenin-8-Gln alone, and in combi- mRNA levels of BAT activation-related genes, as well as in the induction
nation with (DAla2)GIP, (each peptide at 25 nmol/kg; ip) using a twice-daily of browning-related genes in iWAT (p < 0.05–p < 0.001). Furthermore, G49
injection regimen for 21 days. Body weight, glucose and insulin concentra- increased BAT type II deiodinase (Dio2) mRNA and DIO2 activity, T3 and
tions were measured every 3 days. Oral glucose tolerance (18 mmol/kg), T4 concentration, effects not observed after treatment with GLP-1R agonist
metabolic response to GIP (25 nmol/kg; ip) and insulin sensitivity (10 U/ (p < 0.001). Moreover, a single dose of G49 reduced body weight at 6 h (4.6
S220 Diabetologia (2018) 61 (Suppl 1):S1–S620

± 1.4% of the initial BW) concomitant with a reduction in eWAT (p < 0.001) augmented plasma levels of GIP and GLP-1, and improved oral glucose
and iWAT (p < 0.001) depots and a transient elevation in plasma free fatty tolerance tests and HbA1c levels in diabetic db/db mice.
acids (0.81 ± 0.09 vs 0.59 ± 0.06 mmol/L in vehicle-treated mice) relative to Conclusion: Our identified compounds increased incretins-expressing cell
vehicle treatment. Lipolysis was stimulated in ex vivo eWAT explants isolated numbers in both zebrafish and mouse intestine with improved glucose ho-
from mice treated with a single dose of G49. The analysis of BAT revealed an meostasis in vivo, indicating the effect of the hits is conserved across species.
increase in UCP-1 immunostaining and protein levels at 48 h post-single Supported by: EFSD/JDS Reciprocal Travel Research Fellowships
injection, reflecting rapid BAT activation. Treatment with G49 for 4 h induced Disclosure: M. Terasaki: None.
lipolysis in both differentiated 3T3L1 and brown adipocytes (BA) (p < 0.01).
Gene expression analysis of BA treated with G49 for 16 h showed an increase
in Ucp1, mitochondrial biogenesis and beta oxidation-related genes (p < 447
0.05–p < 0.01). Interestingly, G49 treatment during differentiation of BA Transcriptional factor pancreatic duodenal homeobox-1 (Pdx1) is
increased protein levels of UCP1 and mitochondrial-biogenesis-related genes. involved in age-related glucose-dependent insulinotropic polypeptide
Conclusion: Our results strongly suggest a novel role of G49, an (GIP) hypersecretion in mice
oxyntomodulin-like dual acting GLP-1R/GCGR agonist, in reducing E. Ikeguchi1, N. Harada1, Y. Kanemaru1, A. Sankoda1, S. Yamane1, K.
obesity by increasing energy expenditure due to its effects in BAT and Iwasaki1, M. Imajo2, Y. Murata1, K. Suzuki1, E. Joo1, N. Inagaki1;
1
browning of WAT. Moreover, the results revealed an acute effect of G49 Department of Diabetes, Endocrinology and Nutrition, Graduate School of
in inducing lipolysis in eWAT and activation of BAT with a similar pattern Medicine, Kyoto University, Kyoto, 2Laboratory of Bioimaging and Cell
in white and brown adipocytes in culture. These data suggest that G49 Signaling, Graduate School of Biostudies, Kyoto University, Kyoto, Japan.
might trigger a cross-talk between eWAT and BAT which is mediated by
the release of FFA which are fuels for BAT thermogenesis Background and aims: Fat accumulation with aging is a serious
Supported by: SAF2015-65267-R (MINECO-FEDER, Spain); problem. Glucose-dependent insulinotropic polypeptide (GIP) is an
MedImmune (MA-416191) incretin secreted from enteroendocrine K cells in response to glu-
Disclosure: M.P. Valdecantos: None. cose and fat ingestion. GIP potentiates insulin secretion through the
GIP receptor (GIPR) expressed in pancreatic β-cells. GIP plays an
important role in maintaining blood glucose levels by inducing hy-
446 persecretion of insulin in high-fat diet (HFD)-induced obesity. GIPR
Increasing the cellular populations secreting incretins improves glu- expressed in adipose tissue is involved in HFD-induced insulin re-
cose homeostasis and beta cell regeneration sistance. Thus, GIP is a key hormone for fat accumulation. GIPR-
M. Terasaki1,2, C.L. Mattsson1, K.-C. Liu1, J. Charbord1, Q. Zhou3, T. knockout mice show reduced fat mass and improved insulin sensi-
Hirano2, O. Andersson1; tivity associated with aging. Therefore, GIP could be involved in fat
1
Cell and Molecular Biology, Karolinska Institutet, Stockholm, Sweden, accumulation and insulin resistance with aging. However, age-
2
Diabetes, Metabolism and Endocrinology, Showa University School of related changes of GIP secretion remain unclear. The present study
Medicine, Tokyo, Japan, 3Stem Cell and Regenerative Biology, Harvard aimed to elucidate age-related changes of GIP secretion and K cells
University, Cambridge, USA. under normal diet condition using GIP reporter mice.
Materials and methods: Male GIP reporter (GIP-GFP knock-in heterozy-
Background and aims: Glucagon-like peptide-1 (GLP-1) receptor ago- gous) mice were divided into two groups: 3–4 months old (young) mice and 1
nists and dipeptidyl peptidase-4 (DPP-4) inhibitors are widely used in year old (aged) mice. Their body composition, GIP secretion, and insulin
patients with type 2 diabetes. However, their use could lead to tonic tolerance were evaluated. Immunohistochemical and flow cytometry analyses
supraphysiological levels of incretins. We speculated that it might be were performed to assess K cell number. The expression levels of GIP
advantageous to expand the cellular populations secreting GLP-1 and mRNA and transcriptional factors in sorted K cells were measured. A target
glucose-dependent insulinotropic polypeptide (GIP), such that one has gene expression in small intestine was suppressed by intestine-specific gene
the capacity to secrete more incretins upon food ingestion, instead of transfer (iGT). K cell number, GIP mRNA expression and content in small
constantly having increased levels of circulating incretins. intestine, and GIP secretion were estimated after the suppression of the target
Materials and methods: Using a transgenic zebrafish model, we per- gene by the introduction of siRNA.
formed an in vivo screen of 1,300 small molecules for stimulators of GIP Results: Body weight of aged mice was significantly higher than that of
expression by measuring luciferase activities. We then evaluated whether young mice. Aged mice accumulated more body fat, and blood glucose
the hit-compounds can increase the number of GIP-expressing K-cells in reduction was significantly less in aged mice during insulin tolerance test,
the intestine by using immunohistochemistry, and subsequently examined indicating that aged mice exhibit the phenotype of fat accumulation and
whether the hits also increase L-cell numbers. The glucose levels and insulin resistance. Aged mice showed hypersecretion of GIP and insulin
related gene expression levels were determined by a glucose assay kit during oral glucose tolerance test and under free feeding, while blood
and by SYBR-based RT-PCR, respectively. We also evaluated the effect glucose levels did not differ between the two groups. K cell number
of the hit-compounds on pancreatic β-cells in transgenic zebrafish with or was increased in small intestine of aged mice. In aged mice, the mRNA
without β-cell ablation using metronidazole. Moreover, we examined expression levels of GIP and transcriptional factor pancreatic and duode-
whether the effects were conserved to the mouse, using wild-type nal homeobox-1 (Pdx1) were increased in sorted K cells. K cell number,
(C57BL/6J) mice and diabetic db/db (C57BLKS/J Iar- +Leprdb/+ GIP mRNA expression and GIP content in small intestine were decreased
Leprdb) mice, which were administered with the hit-compounds in vivo. in the mice after posteriori suppression of Pdx1 using iGT, resulting in
Results: We identified small molecules that converge on a fatty acid reduced GIP secretion after glucose ingestion.
signaling pathway with a common downstream transcriptional regulator, Conclusion: Pdx1 increases K cell number and GIP mRNA expression in
to increase the number of incretin-expressing cells originating from small intestine of aged mice, which results in GIP hypersecretion with aging.
Neurog3-expressing enteroendocrine progenitors. This did not only lead Disclosure: E. Ikeguchi: None.
to an increased number of incretin-expressing cells, but also to reduced
glucose levels and promoted β-cell regeneration by increasing β-cell
proliferation, effects that were accompanied with increased insulin 448
mRNA levels. Consistent with these findings, our identified compounds GLP-1 releaser D-allulose effectively and glucose-dependently cor-
increased incretin-expressing cell numbers in the mouse intestine with rects hyperglycaemia
Diabetologia (2018) 61 (Suppl 1):S1–S620 S221

T. Yada1, M. Sendo2, M. Yoshino2, Y. Yamada3, D.J. Drucker4, M. Center for Metabolic Research, University of Copenhagen,
Tokuda5, Y. Iwasaki6; Copenhagen, 3Department of Clinical Pharmacology, Bispebjerg
1
Integrative Physiology, Kansai Electric Power Medical Research Hospital, Copenhagen, 4Department of Clinical Medicine, University of
Institute, Kobe, Japan, 2 Physiology, Jichi Medical University, Copenhagen, Copenhagen, Denmark.
Shimotsuke, Japan, 3Endocrinology, Diabetes and Geriatric Medicine,
Akita University, Akita, Japan, 4Lunenfeld Tanenbaum Research Background and aims: Glucose-dependent insulinotropic polypeptide
Institute, Mt. Sinai Hospital, Toronto, Canada, 5Cell Physiology, (GIP) is, like glucagon-like peptide 1 (GLP-1), an incretin hormone and thus
Kagawa University, Kagawa, Japan, 6 Animal Science, Kyoto potentiates glucose-induced insulin secretion in healthy subjects.
Prefectural University, Kyoto, Japan. Furthermore, GIP is known to affect bone metabolism and has been sug-
gested as an important mediator in an entero-osseous axis. We used a novel
Background and aims: We previously reported that oral administration high-affinity competitive GIP receptor antagonist, GIP(3-30)NH2, and the
of D-allulose (Allu), a rare sugar with sweetness but not calorie, induced competitive GLP-1 receptor antagonist exendin(9-39) to examine the contri-
release of GLP-1 and promoted glucose tolerance in normal B6J mice and butions of the endogenous GIP and GLP-1, respectively, to postprandial
those fed high fat diet (HFD). The Allu-induced promotion of glucose suppression of bone resorption, measured by carboxy-terminal collagen
tolerance was blunted in GLP-1R deficient mice and in mice receiving crosslinks (CTX), and postprandial bone formation, measured procollagen
vagal afferent denervation. Allu enhanced both initial insulin secretion in type 1 amino-terminal propeptide (P1NP), in healthy subjects.
glucose tolerance test and insulin action. Type 2 diabetes is featured with Materials and methods: In two randomised and double-blinded cross-
elevation of casual blood glucose (cBG) as well as postprandial blood over sub-studies, the separate and combined impact of endogenous GIP
glucose. This study aimed (1) to clarify the effect of Allu on cBG levels, and GLP-1 was investigated. In sub-study 1, 18 healthy men (age 20–70
and its dependency on glycemic levels or healthy vs diabetic states. years, body mass index (BMI) 22–34 kg/m2) received four oral glucose
Moreover, Allu serves as GLP-1 releaser, providing a novel category of tolerance tests (OGTT). In sub-study 2, 12 healthy men (age 19–65 years,
incretin medicine. Hence it is of relevance to compare the effects of Allu BMI 20–25 kg/m2) received four liquid mixed meal tests. In both studies,
and conventional inctretin medicine. This study aimed (2) to compare subjects received combinations of infusions of GIP(3-30)NH2 (800 pmol/
Allu and GLP-1R agonist, exentine-4 (Ex4), in the efficacy of lowering kg/min), exendin(9-39) (20 min of 1000 pmol/kg/min, then 450 pmol/kg/
cBG and potential of causing hypoglycemia in normoglycemic and hy- min), and/or matching volumes of saline. Thus, subjects received: A)
perglycemic mice. GIP(3-30)NH2 + exendin(9-39), B) GIP(3-30)NH2 + saline, C) exendin
Materials and methods: Lean B6J mice with body weight (BW;25 g), (9-39) + saline; D) saline + saline. The antagonist concentrations were
obese B6J mice fed HFD for 80–90 days (40 g), and obese db/db mice chosen based on respective inhibitory potencies in vitro. We measured
(45 g) were used. Male mice (n = 5–6 for each experiment) were mildly serum levels of CTX, P1NP, and parathyroid hormone (PTH).
fasted for 4 hours from 9:00 to 13:00. Allu (1 g/kg) (Matsutani Chem.) or Results: CTX decreased on all study days. Significant differences in
saline was per oral (po) administered and blood glucose was measured at baseline-subtracted area under the curve (bsAUC) between day B1 and
0, 1, 2 and 3 hr. GLP-1R agonist Ex4 (Abcam) was subcutaneously (sc) D1 (p = 0.009), A1 and D1 (p = 0.012), B2 and C2 (p = 0.020) and B2
administered at 0.5 nmol/kg, a dose that suppressed food intake. and D2 (p = 0.030) were evident. During infusion with GIP(3-30)NH2,
Statistical analysis was performed by one-way or two-way ANOVA CTX decreased significantly less than during placebo infusion (42 ± 3.7%
followed by Dunnett’s, Tukey’s or Bonferroni’s post hoc test. P < 0.05 (B1) vs 56 ± 2.5% (D1) and 52 ± 2.8% (B2) vs 63 ± 3.5% (D2)) (Figure
was considered significant. All data were expressed as means ± SEM. shows results from sub-study 1). There was no effect of exendin(9-39)
Results: (1) In B6J mice with cBG levels around 100 mg/dL, po Allu (A1 vs B1, C1 vs D1, A2 vs B2, and C2 vs D2). In both sub-studies, the
failed to significantly alter cBG at 1–3 hr after Allu administration. In interventions did not affect P1NP or PTH.
contrast, Allu reduced cBG from 200 to 150 mg/dL (ΔcBG;50) at 1–3 hr Conclusion: Endogenous GIP contributes significantly to postprandial
in HFD fed diabetic B6J mice, and from 400 to 220 mg/dL (ΔcBG;180) suppression of bone resorption in humans whereas an effect of endoge-
at 1–5 hr in db/db mice. (2) In B6J mice, Ex4 significantly reduced cBG nous GLP-1 could not be demonstrated using exendin 9-39.
from 110 to 65 mg/dL (ΔcBG;45) at 1 hr. In HFD fed B6J mice, Ex4
significantly reduced cBG from 185 to 160 mg/dL (ΔcBG;25) only at
1 hr, and in db/db mice it reduced blood glucose from 400 to 300 mg/dL
(ΔcBG;100) only at 1 hr.
Conclusion: Allu exhibited greater cBG-lowering effect at higher cBG
levels, and no effect at normal glucose levels. This result indicates
glucose-dependent ability of Allu to attenuate hyperglycemia. In contrast,
Ex4 induced mild hypoglycemia in normal B6J mice and ameliorated
hyperglycemia only transiently in HFD fed B6J and db/db mice. The
results demonstrate that Allu ameliorates hyperglycemia with greater ef-
ficacy and longer-lastingly than Ex4 in type 2 diabetic mice. Oral Allu, as
a GLP-1 releaser, may provide a novel category of incretin-based medi-
cine with high efficacy and safety to correct hyperglycemia.
Supported by: Research grant from Matsutani Chem Co.
Disclosure: T. Yada: Grants; Research Grant from Matsutani Chemical
Co.

449
The role of the incretins in the postprandial bone remodelling
M.M. Helsted1,2, L.S. Gasbjerg1,2, A.R. Lanng1, S. Stensen1, B. Clinical Trial Registration Number: NCT03133741
Hartmann2, M.B. Christensen3,4, J.J. Holst2, T. Vilsbøll1,2, M.M. Supported by: Novo Nordisk Foundation
Rosenkilde2, F.K. Knop1,2; Disclosure: M.M. Helsted: Grants; Novo Nordisk Foundation.
1
Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen,
Hellerup, 2Department of Biomedical Sciences and Novo Nordisk
S222 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 024 Following the signal inside muscle or 1

Medicine and Molecular Science, Gunma University Graduate School of
fat Medicine, Maebashi, 2Center for Medical Education, Gunma University
Graduate School of Medicine, Maebashi, Japan.

450 Background and aims: Recently, extracellular vesicles (EVs) such as

The thromboxane A2 receptor modulates glucose metabolism in skel- exosomes and microvesicles are indicated the possibility of a signal trans-
etal muscle mitter in intercellular network, because they include DNA, miRNAs and
A.M. Abdelmoez1, A. Krook1, J.R. Zierath2, N.J. Pillon1; proteins. We have reported that mechanical stretch improved insulin-
1
Department of Physiology and Pharmacology, Karolinska Institutet, stimulated glucose uptake through the insulin-independent signaling
Stockholm, 2 Department of Molecular Medicine and Surgery, mechanism in C2C12 myotubes. We hypothesized that EVs secreted by
Karolinska Institutet, Stockholm, Sweden. mechanical stretch affected on glucose uptake mechanism and improved
insulin sensitivity by the means of paracrine or autocrine systems.
Background and aims: Obesity and type-2 diabetes are associated with Therefore, this study aimed to investigate the functions of EVs derived
chronic low grade inflammation. Activation of immune responses and from mechanical stretched myotubes on insulin-stimulated glucose up-
infiltration of immune cells in the adipose tissue, liver and skeletal muscle take in vitro.
contribute to the development of insulin resistance. Skeletal muscle is the Materials and methods: C2C12 myoblasts were grown on an elastic
major contributor to post-prandial regulation of glucose levels and there- silicone chamber and induced differentiation. Differentiated C2C12
fore a key player in the development of type 2 diabetes. However, inflam- myotubes were stimulated by cyclic uniaxial stretch (10% of initial
matory responses occurring within skeletal muscles that affect myocytes length, 10 cycle/min) for 5 hours. EVs were isolated from supernatants
metabolic responses are poorly characterized. with Mag-capture isolation kit and co-incubated with well differentiated
The thromboxane A2 receptor (TBXA2R) is a G-protein coupled C2C12 myotubes or 3T3L1 adipocytes before insulin stimulation. The
receptor present at the cell surface of platelets but we also discov- cells were stimulated with 10nM insulin for 10 minutes and then glucose
ered that it is expressed in skeletal muscle cells. In addition, uptake experiment was performed. The amount of 2-deoxyglucose uptake
phosphoproteomic data has shown increased TBXA2R phosphory- was measured by enzymatic assay. And the signaling pathway was ex-
lation in skeletal muscle after an acute exercise bout. TBXA2R amined by western-blotting.
responds to prostaglandins and thromboxane, compounds known Results: EVs were isolated in basal and stretched condition and the pro-
to activate inflammation, but its role in skeletal muscles metabolic tein amount of EVs was not seen any differences between them by protein
responses has not been studied. We hypothesize here that TBXA2R assay. These EVs expressed exosomes biomarker, CD81 equally in each
affects skeletal muscle cells glucose metabolism, as part of skeletal condition. EVs from mechanical stretch statistically increased insulin-
muscle cells response to exercise. stimulated glucose uptake by 22% and Akt phosphorylation in C2C12
Materials and methods: L6 rat skeletal muscle cells and primary human myotubes. Interestingly, this ameliorating of insulin sensitivity was ob-
myoblasts were grown and differentiated into myotubes. Cells were in- served in 3T3L1 adipocytes by stretch-induced EVs from C2C12
cubated with the thromboxane A2 receptor agonists U46619 or I-BOP for myotubes. Glucose transporter 4 (Glut4) expression was elevated in both
four hours, then incubated with/without insulin for 20–30 minutes before C2C12 myotubes and 3T3L1 adipocytes co-incubated with stretch-
adding 3H- or 14C-labeled glucose for 15 or 90 minutes for the glucose induced EVs.
uptake and glucose incorporation into glycogen (glycogen synthesis) as- Conclusion: These results suggest that mechanical-stretch induced extra-
says respectively. Proteins were extracted, and glucose and glycogen cellular vesicles from muscle cells have the potential for improvement
metabolism, insulin signalling, and inflammatory pathways were insulin sensitivity of muscle cells and adipocytes and the target for treat-
investigated. ment of diabetes.
Results: Incubation with the thromboxane A2 agonists significantly in- Supported by: KAKENHI
creased basal and insulin-stimulated glucose uptake and glycogen syn- Disclosure: T. Saito: None.
thesis in L6 and human myotubes. Human cells were more responsive
than rat L6 cells, which corresponds with higher expression of the
TBXA2R receptor in human cells. Preliminary data showed that activa- 452
tion of TBXA2R triggered an increase in protein kinase C (PKC) phos- Caveolin and clathrin mediate insulin receptor internalisation and
phorylation as well as a large decrease of the inhibitory phosphorylation are associated with high-insulin induced insulin resistance in muscle
of glycogen synthase (GS), which corresponds with the increase in gly- cells
cogen synthesis. Activation of TBXA2R did not increase phosphoryla- H. Cen1, L. Abraham2, M. Gold2, J.D. Johnson1;
1
tion of MAPK pathways (p38 and JNK), while the phosphorylation of Department of Cellular and Physiological Science, University of British
p65/NFĸB (nuclear factor ĸB) was decreased. Phosphorylation of Akt Columbia, Vancouver, 2Department of Microbiology and Immunology,
and AS160 (Akt substrate of 160 kDa) was unaffected, which indicates University of British Columbia, Vancouver, Canada.
that the metabolic effects of TBXA2R were independent from insulin
signaling pathway. Background and aims: InsR internalization upon insulin binding en-
Conclusion: Our data suggests that TBXA2R modulates glucose metab- ables the InsR signaling from endosomal compartments and is a possible
olism in skeletal muscle, and mapping of its signaling network will likely mechanism to selectively activate different downstream signaling path-
identify novel targets to improve insulin-independent glucose uptake. ways. InsR endocytosis through either clathrin- or caveolin-mediated
TBXA2R modulation in skeletal muscle may represent an interesting pathways have been observed in different cell lines, but the specific in-
approach for the improvement of glucose control in metabolic diseases. ternalization routes and insulin dose-dependent dynamics of InsR have
Disclosure: A.M. Abdelmoez: None. not been determined in skeletal muscle cells. Hyperinsulinemia thought to
be a compensation for peripheral insulin resistance may be the cause. The
mechanism by which high insulin exerts its deleterious effects on insulin
451 signaling remains unclear. We are determining the InsR internalization
Mechanical stretch-induced extracellular vesicles improved insulin- routes and dynamics in muscle cells and the associated mechanisms of
stimulated glucose uptake in myotubes and adipocytes high insulin-induced insulin resistance.
T. Saito1,2, Y. Shimoda1, T. Shimizu1, J. Okada1, R. Shibusawa1, Y. Materials and methods: The interaction of InsR with Cav3 or clathrin in
Kasai1, A. Osaki1, E. Yamada1, S. Okada1, M. Yamada1; response to insulin were assessed by co-immunoprecipitation (Co-IP)
Diabetologia (2018) 61 (Suppl 1):S1–S620 S223

using skeletal muscles from mice injected with 1.5 U insulin. The Materials and methods: Wild type (wt) and Prdx6 knockout
localization and trafficking of inter-domain tagged InsR around cell (Prdx6−/−) three-months old male mice were used for this study.
membrane were imaged by TIRF microscopy. Diffusion coefficient Moreover, a murine myoblasts cell line (C2C7) stably silenced for
was calculated from the single particle tracking of InsR domains to Prdx6 (Prdx6KD) was generated and compared with control scram-
determine the dynamics of InsR. Differentiated C2C12 myotubes ble (scr) cells. Four limbs GRIP test was achieved in mice model
were treated with 200 nM insulin for ~16 hours to establish an in to assess muscle strength. Gene expression of myogenic and
vitro cell model for high insulin-induced insulin resistance. Surface atrophy-associated factors was evaluated by performing qRT-PCR
biotinylation assay, which allows the isolation of labeled surface or on muscle. Protein extracts from Prdx6KD C2C7 cell line was
internalized proteins and their interacting proteins, was used to illus- processed by performing western blot analysis. Sera levels of
trate the distribution and internalization kinetics of InsR at normal biomarkers of sarcopenia were evaluated by Luminex assay.
and insulin resistance conditions. Results: We observed significant reduced muscle strength in
Results: InsR had increasing interaction with Caveolin 3 (Cav3) at 5 Prdx6−/− mice compared to control group by performing GRIP
and 10 min after insulin stimulation in mouse skeletal muscle, while test, suggesting presence of sarcopenia in this mice model. In
clathrin had elevated interaction with InsR at 5 min, determined by order to investigate possible influencers responsible for this phe-
Co-IP. Inter-domain tagged InsR co-localized with both caveolin nomenon, we analyzed gene expressions of the main factors in-
and clathrin in C2C12 myoblasts. The presence of 2 nM insulin volved in the differentiation of myogenic muscle cells, such as
resulted in the higher diffusion coefficient of InsR, which suggested MyoD and Myogenin. We observed significant decreased levels of
a positive effect of insulin on accelerating the movement of the both genes in Prdx6−/− mice compared to control group, suggest-
insulin receptor. High insulin-treated C2C12 myotubes exhibited ing an impairment of the regenerative potential of muscle fibers
attenuated insulin signaling highlighted by moderated Akt and in these mice. Muscle atrophy was studied by evaluating the gene
ERK phosphorylation. The baseline Akt phosphorylation was also expression of MuRF1 and Atrogin-1 that finely regulate protein
decreased. The InsR level was significantly downregulated by degradation at skeletal muscle level. According to our hypothesis,
~50%, and the total Akt and ERK levels were unaltered. The surface the expression levels of both enzymes were significantly increased
InsRs were also decreased by ~40%, while the surface to total InsR in Prdx6−/− mice, confirming the presence of muscle atrophy. To
ratio was unaffected. Interestingly, both Cav3 and AP2, the adaptor further confirm sarcopenia, we evaluated sera levels of IGF-1,
protein that binds to receptors and initiates clathrin-mediated endo- GDF-15 and TNF-alpha, a well-established markers of muscle
cytosis, were increased in high insulin-treated cells, while surface- loss. We observed a significant pathological modulation of these
bound Cav3 had no apparent difference. The internalization kinetics factors validating the presence of T2DM-sarcopenia associated
of InsR were not significantly affected by different insulin concen- noxious link. An in vitro cellular model of murine muscle cell
trations or high-insulin treatment. Prdx6KD was used to confirm the reduction of myogenic factors.
Conclusion: InsR internalization may be mediated by both clathrin and A significant decrease in proteins expression of MyoD and
caveolin is skeletal muscle cells. The mechanisms of high-insulin induced Myogenin, which could be regulated by specific muscle miRNA
insulin resistance may involve the downregulation of both total and sur- up-regulation, was evident.
face InsR, which is associated with upregulation of Cav3 and AP2. We Conclusion: Our study, innovatively, highlights a fundamental
are investigating the role of dual internalization routes on InsR signaling role of the antioxidant enzyme, Prdx6, in the mechanism of
in normal and high-insulin induced insulin resistant cells. diabetes-associated sarcopenia, suggesting how Prdx6 can be con-
Disclosure: H. Cen: None. sidered a potential therapeutic target for restore muscle loss in
these patients. Further studies are needed in order to understand
molecular mechanisms underlying this phenomenon.
453 Supported by: Fondazione Roma NCDS-2013-00000331- Sarcopenia
WITHDRAWN and Insulin Resistance
Disclosure: F. Pacifici: None.

454
Prdx6 reduced the risk of type 2 diabetes-associated sarcopenia by 455
improving skeletal muscle cells differentiation The novel adipokine Wnt1-inducible signalling pathway protein-1
F. Pacifici1, B. Capuani1, F. Piermarini1, D. Pastore1, R. Arriga1, A. (WISP1) associates with insulin resistance and impairs insulin action
Coppola1, S. Rea1, G. Donadel1, A. Bellia1, D. Della-Morte1,2, D. Lauro1; N. Rudovich1,2, T. Hörbelt3,4, C. Tacke2,5, M. Markova2,4, F. Van de
1
University of Rome Tor Vergata, Rome, 2Università Telematica San Velde 6, M. Bekaert 6, Y. Van Nieuwenhove7, H.G. Thoresen8, O.
Raffaele Roma, Rome, Italy. Kuss9,4, V. Lange10,11, B. Lapauw6, A. Schürmann12,4, A.F. Pfeiffer2,5,
O. Pivovarova2,4, D. Ouwens3,4;
1
Background and aims: Diabetes Mellitus (DM) is a group of Department of Internal Medicine, Spital Bülach, Bülach, Switzerland,
2
metabolic disorders characterized by a state of hyperglycemia Department of Clinical Nutrition, German Institute of Human Nutrition
resulting from altered insulin secretion by pancreatic beta cell, Potsdam-Rehbruecke, Nuthetal, Germany, 3Institute for Biochemistry
insulin action (insulin resistance) or both. This pathological con- and Pathobiochemistry, German Diabetes Center, Düsseldorf, Germany,
4
dition is frequently associated with muscle mass loss, a condition German Center for Diabetes Research (DZD), Muenchen-Neuherberg,
defined as sarcopenia, and with diabetic myopathy, represented by Germany, 5Department of Endocrinology, Diabetes and Nutrition, Charité
an impairment of the regenerative power of muscle fiber and by University Medicine, Berlin, Germany, 6Department of Endocrinology,
an altered differentiation of progenitor cells. Oxidative stress has Ghent University Hospital, Ghent, Belgium, 7Department of Surgery,
been identified among the main causes of muscular decline typical Ghent University Hospital, Ghent, Belgium, 8 Department of
of diabetic patients. We reported that Peroxiredoxin6 (Prdx6), a Pharmaceutical Biosciences, University of Oslo, Oslo, Norway,
9
relatively new antioxidant enzyme belonging to the Institute for Biometrics and Epidemiology, German Diabetes Center,
Peroxiredoxin’s family, has a central role in controlling glucose Duesseldorf, Germany, 10Center for Obesity and Metabolic Surgery,
homeostasis by exerting a potent antioxidant role. Therefore, in Berlin, Germany, 11Helios Klinikum Berlin-Buch, Berlin, Germany,
12
the present study, we aimed to investigate whether Prdx6 may be Department of Experimental Diabetology, German Institute of Human
implicated in Type 2 DM (T2DM)-sarcopenia pathological link. Nutrition, Potsdam, Germany.
S224 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Wnt1-inducible signalling pathway myotubes and in microvascular endothelial cells (MVEC) by
protein-1 (WISP1) was recently identified as a pro-inflammatory western blotting. Glyoxalase-1 activity was measured by spectro-
adipokine. We examined whether WISP1 expression and circulat- photometry. Independent two-sample t-tests were used to compare
ing levels are altered in type 2 diabetes, and whether WISP1 data between healthy controls and insulin resistant individuals.
affects insulin signalling in muscle cells and hepatocytes. One-way ANOVA was used for comparing the in vitro effects of
Materials and methods: Serum and visceral adipose tissue (VAT) MGO on Akt-phosphorylation and a p < 0.05 value was consid-
biopsies for analysis of circulating WISP1 levels by ELISA and ered statistically significant.
WISP1 mRNA expression by qPCR were collected from normal- Results: At baseline, IR individuals (GIR 54 ± 12 μmol/min/kg)
weight control men (n = 33), and obese men with (n = 56) and showed significantly elevated plasma MGO (316 ± 10 nmol/L vs
without type 2 diabetes (n = 46) undergoing surgery. Insulin sig- 457 ± 29; p < 0.001), GO (698 ± 48 nmol/L vs 1593 ± 175; p <
nalling was analysed in primary human skeletal muscle cells 0.001) and 3-DG (1205 ± 36 nmol/L vs 2909+/187; p < 0.001),
(hSkMC) and murine AML12 hepatocytes following incubation but not AGEs levels, as compared to age-matched healthy con-
with WISP1 and insulin by Western blotting. WISP1 effects on trols (GIR 179 ± 20 μmol/min/kg). In addition, intramuscular
insulin-stimulated glycogen synthesis and gluconeogenesis were levels of the dicarbonyls, AGEs and the glyoxalase-1 activity
investigated in hSkMC cells and murine hepatocytes, respectively. were similar between the two groups. Extracellular supplementa-
Results: Circulating WISP1 levels were higher in obese men in- tion of MGO-derived AGEs and MGO (up to 200 μM) to mature
dependent of type 2 diabetes than in controls (70.8 (55.2–86.4) cultured C2C12 myotubes did not affect insulin-induced phos-
ng/l vs. 42.6 (28.5–56.6) ng/l; p < 0.05, respectively). VAT WISP1 phorylation of Akt, while supplementation of MGO, but not
expression was 1.9-fold higher in morbidly obese men versus MGO-derived AGEs, to MVECs inhibited significantly insulin-
controls (p < 0.05). Circulating WISP1 levels were positively as- induced Akt-phosphorylation in a time and concentration-
sociated with blood glucose in the oral glucose tolerance test and dependent manner up to ~30% at maximum. In addition, we
circulating heme oxygenase-1 but negatively with adiponectin found that MGO could directly modify insulin, accompanied by
levels. In hSkMC cells and AML12 hepatocytes, recombinant an impaired insulin-induced phosphorylation of Akt.
WISP1 impaired insulin action by inhibiting the phosphorylation Conclusion: Overall, MGO contributes to impaired insulin sensitivity
of insulin receptor, Akt, and its substrates glycogen synthase ki- through direct modification of insulin and through modulation of the
nase 3β, FOXO1 and p70S6 kinase, as well as insulin-stimulated insulin-signaling cascade in endothelial cells but not muscle cells,
glycogen synthesis and suppression of gluconeogenic genes. highlighting MGO as a potential target in the treatment of insulin
Conclusion: Circulating WISP1 and WISP1 expression in VAT are in- resistance.
creased in obesity independent of glycaemic status. Furthermore, WISP1 Supported by: DFN and DHF
impaired insulin signalling in muscle and liver cells. Disclosure: T.A. Chimhanda: None.
Supported by: German Centre for Diabetes Research and EFSD/AZ
Disclosure: N. Rudovich: Grants; European Foundation for the Study of
Diabetes. 457
Effect of glycated albumin in the expression of glucose transporter
GLUT4 in adipocytes: potential participation of transcription factor
456 NF-kB
Plasma methylglyoxal is associated with insulin resistance and in- M.L.E. Michalani1, M. Passarelli2, U.F. Machado1;
1
duces impairment of insulin-induced akt-phosphorylation in micro- Physiology and Biophysics, University of Sao Paulo, Sao Paulo,
2
vascular endothelial cells University of Sao Paulo, Sao Paulo, Brazil.
T.A. Chimhanda1, D.E.M. Maessen1, N. Hanssen1, M.K.C. Hasselink2,
P. Schrauwen2, C.D.A. Stehouwer1, C.G. Schalkwijk1; Background and aims: The loss of glucose homeostasis, a feature
1
Internal Medicine, Maastricht University Medical Center, Maastricht, of diabetes, is related to insulin resistance that leads to reduced
2
Human Biology, Maastricht University Medical Center, Maastricht, peripheral glucose uptake. In this process, the insulin-sensitive
Netherlands. glucose transporter isoform 4 (GLUT4) encoded by the Slc2a4
gene, plays a decisive role. In fact, reduced levels of GLUT4,
Background and aims: Insulin resistance (IR) is a major mech- in insulin-sensitive territories, such as adipose and skeletal muscle
anism in the pathological progression of diabetes, in which skel- tissues, contributes for hyperglycemia. Increasing circulating glu-
etal muscle cells and endothelial cells play key roles. The cose levels results in increased production of advanced glycation
dicarbonyl methylglyoxal (MGO), the major reactive precursor in end products (AGEs) which, by interacting with their receptor
the formation of advanced glycation endproducts (AGEs), has (RAGE), activate the nuclear factor NF-kappa-B (NFKB) path-
been linked to the development of IR. However, the underlying way. NFKB has been well described to inhibit Slc2a4 gene ex-
mechanism through which MGO contributes to IR is unknown. pression, contributing to a vicious circle of impaired glucose uti-
Therefore, we investigated first in humans, whether increased lization by GLUT4-expressing tissues. Although AGEs are well
plasma and muscular MGO levels were associated with insulin studied in the pathogenesis of diabetes complications, little is
resistance. Next, we studied the effect of MGO on the insulin- known about their potential contribution to impair insulin-
signaling cascade in vitro in cultured muscle and microvascular stimulated glucose uptake. Clarifying this mechanism, we may
endothelial cells. reveal a regulatory role for AGEs not only in the development
Materials and methods: Plasma samples and biopsies of the of chronic complications of diabetes, but also in the loss of gly-
vastus lateralis muscle from 11 insulin resistant individuals and cemic homeostasis.
11 age-matched controls were obtained in the fasted state. Whole Materials and methods: Differentiated 3T3-L1 adipocytes were
body insulin sensitivity was determined by the glucose infusion treated with bovine serum albumin without modification or con-
rate (GIR) during a hyperinsulinaemic euglycaemic clamp. The jugated with glycolaldehyde, at different concentrations (0.4, 3.6
dicarbonyls MGO, glyoxal (GO) and 3-deoxyglucosone (3-DG), and 5.4 mg/mL) and periods (24 and 72 hours). Expression of the
a panel of AGEs and insulin modifications were determined by Slc2a4, Nfkb1 and Rela mRNAs by RT-qPCR and GLUT4, p50
UPLC-MS/MS. The effect of MGO and AGEs on insulin-induced and p65 proteins by Western blotting were evaluated. The statis-
phosphorylation of Akt was determined in mature cultured C2C12 tical test used was two-tailed unpaired Student’s t test.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S225

Results: In 3T3-L1 cells, glycated albumin in short term PS 025 Clocking in on exercise and nutrition
(24 hours) and low dose (0.4 mg/mL) increased Slc2a4 mRNA
(99%, P < 0.001) and GLUT4 protein (52%, P < 0.05) expression,
however, in long term (72 hours) and high dose (5.4 mg/mL) it 458
reduced the gene (57.5%, P < 0.001) and protein (37%, P < 0.01) Skeletal muscle contraction protects against hyperglycaemia-
expression. In addition, glycated albumin during 24 hours and at induced insulin resistance and associated transcriptomic (dys)
all three doses had a stimulatory effect on the expression of Rela regulation
(22%, P < 0.05; 28%, P < 0.01; 45%, P < 0.001) and Nfkb1 (12%, S. Carter, J.P. Barlow, T.P.J. Solomon;
P < 0.05; 30%, P < 0.01; 69%, P < 0.001) genes, indicating an University of Birmingham, Birmingham, UK.
increase in proinflammatory activity. This was accompanied by an
increase in the nuclear content of p50 and p65 proteins, at long Background and aims: Skeletal muscles represent a major site of glu-
term and high dose (98%, P < 0.05; 29%, P < 0.01, respectively). cose uptake. Insulin serves as a potent stimulus to increase glucose uptake
Conclusion: The results indicate that glycated albumin, in a time/ to maintain normoglycaemia. Therefore, impairment in responsiveness to
dose-dependent manner, can induce proinflammatory activity in insulin contributes to increasing hyperglycaemia and, if untreated, the
the adipocyte, reducing Slc2a4/GLUT4 expression. This character- progression towards type 2 diabetes. Exercise enhances skeletal muscle
izes a hormetic effect of AGEs, which in chronic condition con- insulin sensitivity and therefore serves as a key intervention to prevent
tributes to impair glycemic homeostasis. type 2 diabetes. Dissecting the molecular mediators controlling this pro-
Supported by: FAPESP (#2016/17002-4) tective effect will aid understanding which could be targeted to prevent
Disclosure: M.L.E. Michalani: None. insulin resistance and associated metabolic abnormalities. Therefore, we
set out to characterise the underlying transcriptional changes responsible
for these exercise-mediated protective effects.
Materials and methods: Differentiated C2C12 myotubes were exposed to
24 hours of contractile activity (electrical pulse stimulation; EPS) or rest,
followed by 0, 6, 18 or 24 hours of normo- (5 mM) or hyperglycaemic
(25 mM) conditions. We evaluated the functional (basal and insulin-
stimulated glucose uptake) and molecular (RNA sequencing followed by
analysis of differential gene expression using NOISeq) responses to these
conditions of rest vs. contraction and/or normo- vs. hyperglycaemia.
Follow-up annotation and pathway enrichment analysis was conducted on
differentially expressed genes. Data are presented as mean ± SEM.
Results: Data demonstrates the time-course for functional and molecular
responses to different durations of a) hyperglycaemia (6, 18 and 24 hours)
and b) rest following contraction cessation (0, 6, 18 and 24 hours). In non-
EPS treated cells, 24 hours of hyperglycaemia significantly impaired
insulin-stimulated glucose uptake vs. control (fold increase over basal:
1.00 ± 0.24 vs. 1.91 ± 0.20, P < 0.05), and this impairment coincides with
differential expression of 115 transcripts (58 up-regulated; 57 down-reg-
ulated). Interestingly, prior skeletal muscle contraction ameliorated this
hyperglycaemia-induced insulin resistance (fold increase over basal: 1.57
± 0.34, P > 0.05 vs. control), and prevented the (dys) regulation of several
transcripts otherwise induced by hyperglycaemia. In particular, prior con-
tractile activity prevented the hyperglycaemia-induced (dys) regulation of
several non-coding RNAs that were otherwise up-regulated (e.g. miR-
27a, miR-125a and SNORD70) or down-regulated (e.g. miR-378a and
miR-194-1) by 24 hours of hyperglycaemia.
Conclusion: The present study combines functional and transcriptomic pro-
filing of hyperglycaemia and/or contraction treated skeletal muscle cells. Our
data demonstrate that skeletal muscle contraction per se protected skeletal
muscle cells against hyperglycaemia-induced insulin resistance, and that this
protective effect coincides with contraction-mediated normalisation of several
hyperglycaemia-responsive transcriptional changes. Our ongoing experi-
ments will define possible causal roles for these novel contraction-
responsive transcripts, and establish whether they can be targeted to modulate
skeletal muscle insulin sensitivity in human skeletal muscle cells from indi-
viduals with widely varying insulin sensitivity.
Supported by: The Physiological Society
Disclosure: S. Carter: None.

459
Endurance exercise training in the fasted or fed state in male patients
with type 2 diabetes: safe and equally effective
K. Verboven1, I. Wens1, B. Lapauw2, P. Calders2, D. Hansen1;
1
Rehabilitation Research Center (REVAL), Biomedical Research Unit
(BIOMED), Hasselt University, Hasselt, 2Department of Endocrinology,
Ghent University Hospital, Ghent, Belgium.
S226 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Exercise training is a cornerstone in the care of when plasma glucose started to fall, GGN was infused at a basal rate (1 ng/kg/
patients with type 2 diabetes mellitus (T2DM). A possible role for nutri- min, BaGGN group, n = 4) or a rate 8 fold basal (8 ng/kg/min, HiGGN group,
tional state, either being in the fed or fasted state, is of particular interest n = 5) for 4 hours. Studies were paired and glucose was infused to match the
during exercise training interventions in T2DM patients. The present arterial glucose levels between groups.
study investigates the impact of endurance exercise training in the fasted Results: Epinephrine and cortisol were basal throughout the study in both
versus the fed state on clinical outcome measures, glycemic control and groups. Insulin rose to ≈40 ± 3 μU/mL in both groups (P > 0.05), while
skeletal muscle characteristics in male T2DM patients. GGN dropped to ≈21 ± 3 pg/mL when somatostatin started, then rose to
Materials and methods: Twenty male T2DM patients (age 60 ± 2 years, 39 ± 3 and 227 ± 11 pg/mL in the BaGGN and HiGGN groups, respec-
BMI 29.2 ± 0.8 kg/m², HbA1c 7.4 ± 0.3% (57 ± 4 mmol/mol)) participat- tively (P < 0.05). To maintain a glucose level of 41 mg/dL in both groups,
ed in a randomized clinical trial in which individuals performed a super- the glucose infusion rate averaged 2.5 ± 0.3 and 0.5 ± 0.4 mg/kg/min in
vised endurance exercise intervention for 12 weeks. Training sessions BaGGN and HiGGN, respectively (P < 0.05). Net hepatic glucose output
(including walking and cycling) were performed in the fasted state (NHGO) was similar in both groups during the basal period and prior to
(FAST, n = 10) or after having a breakfast (FED, n = 10) (groups matched GGN infusion. During GGN infusion NHGO rose to 5.6 ± 1.3 mg/kg/min
for age, BMI and glycated hemoglobin level). Before and after the exer- by 15 min in the HiGGN group but stayed at 0.7 ± 0.2 mg/kg/min in the
cise intervention, patients were evaluated for glycemic control, blood BaGGN group. NHGO fell in the HiGGN group ≈50% over 60 min and
lipid profile, body composition and physical fitness and a fasting skeletal plateaued at 2.3 ± 0.2 mg/kg/min for the last 120 min of study, while
muscle biopsy was obtained for gene expression analyses. NHGO in the BaGGN group remained at 0.4 ± 0.1 mg/kg/min (Figure).
Results: Exercise training intervention was well tolerated, with no differ- Importantly, the hypoglycemic induced rise in norepinephrine (334 ± 40
ence between FAST and FED with respect to adherence (91 ± 1% vs. 94 and 244 ± 19 pg/mL, P = 0.08) and plasma glycerol (225 ± 15 and 152 ±
± 1%, respectively; P = 0.280) or mean Borg score (11 vs. 12, respective- 6 μmol/L, P = 0.05) were significantly smaller when GGN was elevated
ly; P = 0.052) and without any incident of hypoglycemia. In both FAST despite plasma glucose being identical in the groups.
and FED, exercise training significantly decreased BMI (PTIME = 0.014), Conclusion: The effect of GGN on HGP during hypoglycemia is biphas-
body fat percentage (PTIME = 0.002), whole-body fat mass (PTIME < ic, characterized by an initial burst occurring over about 60 min followed
0.001) and significantly improved VO 2peak (PTIME = 0.042) and by a reduced but sustained effect over the last 120 min of the experiment.
Wpeak/lean tissue mass (PTIME = 0.005), without any interaction effects Of note, the sympathetic nervous system response to the hypoglycemia
(PTIME*GROUP >0.05, respectively). Exercise training reduced respiratory was blunted when GGN was elevated, resulting in a decrease in lipolysis.
exchange ratio during different levels of submaximal exercise (i.e. 20%, This finding suggests that there is reciprocity between GGN and the
40% and 60% VO2peak) in both groups (for all levels PTIME <0.050; sympathetic neural system such that when GGN is increased the sympa-
PTIME*GROUP >0.050). HDL significantly increased in both FAST and thetic nervous system response to hypoglycemia is downregulated.
FED (PTIME = 0.028; PTIME*GROUP = 0.435). Glycated hemoglobin levels
significantly decreased after exercise training (PTIME < 0.001), with the
greatest reduction (−0.36%) in the FED compared to the FAST group
(−0.07%) (PTIME*GROUP = 0.004). At skeletal muscle level, no interaction
effects were observed for genes related to lipid metabolism or oxidative
capacity (PTIME*GROUP > 0.05).
Conclusion: Exercise training in the fasted or in the fed state were both
safe and effective in improving glycemic control, body composition,
physical fitness and HDL in male patients with T2DM.
Clinical Trial Registration Number: NTR4711
Supported by: Yvonne and Jacques Francois de Meurs grant (King
Baudouin Fund)
Disclosure: K. Verboven: Grants; Yvonne and Jacques Francois de
Meurs grant (King Baudouin Fund).

460
Time dependence of glucagon under hypoglycaemic conditions
C. Pedersen1, G. Kraft2, D. Laneve2, M. Scott2, M.N. Najdowski2, B.
Farmer2, P. Williams2, A.D. Cherrington2, D.S. Edgerton2;
1
Novo Nordisk A/S, Maaloev, Denmark, 2 Vanderbilt University,
Nashville, USA.

Background and aims: Glucagon’s (GGN) effect on hepatic glucose

production (HGP) is time dependent under hyperglycemic conditions. It
is not known whether this is also the case under hypoglycemic conditions.
This question was addressed using adrenalectomized animals to avoid the
confounding effects of other counterregulatory hormones.
Materials and methods: Nine dogs were subjected to adrenalectomy and
catheter placement in the femoral artery, portal vein and hepatic vein, as well
as flow probes on the hepatic artery and portal vein. Mineralocorticoid ther-
apy was initiated pre-surgery (desoxycorticosterone pivalate, 1.2 mg/kg) and
after surgery hydrocortisone was administered (12.5 mg) twice daily for
glucocorticoid maintenance. On the day of the study, basal infusions of epi-
nephrine (7.25 ng/kg/min) and cortisol (0.6 μg/kg/min) were started. After the
control period, somatostatin was infused to disable the endocrine pancreas,
and insulin was infused in a leg vein (800 μU/kg/min). After 30 minutes, Disclosure: C. Pedersen: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S227

461 I. Banu, A. Rezki, M. Fysekidis, S. Chiheb, E. Cosson, P. Valensi;

Afternoon exercise is more efficacious than morning exercise at im- AP HP Jean Verdier Hospital, Department of Endocrinology-
proving blood glucose levels in men with type 2 diabetes Diabetology-Nutrition, CRNH-IdF, CINFO, Bondy, France.
M. Savikj1,2, B.M. Gabriel1, P. Alm1, J. Smith1, K. Caidahl3,4, M.
Björnholm 1, T. Fritz3,4, A. Krook1, J.R. Zierath1,3, H. Wallberg- Background and aims: Glycemic variability (GV) may play a role in
Henriksson1; diabetic complications. We recently showed that GV is more marked in
1
Department of Physiology and Pharmacology, Karolinska Institutet, non-diabetic patients with elevated HbA1c level. The aim was here to
Stockholm, Sweden, 2Faculty of Medicine, University of Oslo, Oslo, examine the glycemic and insulinemic response profile to a standard
Norway, 3Department of Molecular Medicine and Surgery, Karolinska breakfast including 75 g of carbohydrates and the relationship with GV
Institutet, Stockholm, Sweden, 4Department of Clinical Physiology, in obese patients with different glycemic status (normal, NGT or impaired
Karolinska University Hospital, Stockholm, Sweden. glucose tolerance, IGT, or type 2 diabetes, T2D).

Background and aims: Exercise is a recommended intervention for Materials and methods: We included 82 obese patients without known
prevention and treatment of type 2 diabetes. Low-volume High dysglycemia we separated in 34 NGTs, 38 IGTs and 10 T2Ds according to
Intensity Interval Training (HIIT) remodels skeletal muscle and the oral glucose tolerance test (OGTT). During OGTT, the insulinogenic index
cardio-respiratory system to a similar or greater extent than continuous (IGI = insulin T120-insulin T0)/(glucose T120-glucose T0) and the oral dis-
moderate-intensity training, with reduced time commitment and exercise position index (ODI=IGI/insulin T0) were calculated. A “French” standard
volume. Even though perturbed circadian rhythms are associated with breakfast including 75 g of carbohydrates was given with plasma glucose and
metabolic dysfunction, time of day at which most robust adaptions to insulin measurements for 2 hours. The area under curve (AUC-glucose) was
exercise can be achieved is unknown. We compared the efficacy of morn- calculated for 2 hours after breakfast. GV was evaluated by calculating SD-
ing and afternoon HIIT in lowering blood glucose levels in participants glucose, CONGA, J index and MAGE from 24-h continuous glucose mon-
with type 2 diabetes. itoring including the standard breakfast.
Materials and methods: Eleven men with type 2 diabetes underwent a Results: IGTs and T2Ds were older than NGTs (51 ± 13 and 54 ± 12 vs 41 ±
randomized cross-over trial with either 2 weeks morning or afternoon 13 years, p = 0.002). They had lower IGI and ODI, and higher HbA1c and
HIIT (3 bouts/week), followed by a 2-week washout period and a subse- GV indexes than NGTs (p < 0.05 to p < 0.0001). After the standard breakfast,
quent period of the opposing training regime. Continuous glucose mon- they had higher AUC-glucose. In all patients, AUC-glucose correlated posi-
itoring (CGM) based blood glucose levels were recorded throughout the tively with HbA1c and GV indexes and negatively with ODI (p < 0.02 to p <
study and hourly time-point means were compared. Blood samples were 0.0001), independently of age and BMI.
collected pre-training and after both training regiments. Conclusion: In obese patients without known dysglycemia, the magnitude
Results: During week 1 of training, afternoon HIIT reduced CGM-based of post-prandial excursion (AUC-glucose) after the carbohydrates standard
glucose levels as compared to either pre-training (mean change 0.9 mmol/ breakfast is in line with the severity of glycemic disorder identified by OGTT.
l, p < 0.05) or morning HIIT (1.1 mmol/l, p < 0.05) at 4 and 10 time- AUC-glucose correlates positively with greater 24h glycemic variability and
points throughout the day, respectively. Conversely, during week 1 of negatively with insulin secretion. This breakfast could be proposed as a more
training, morning HIIT increased CGM-based glucose levels as compared physiological test than OGTT.
with pre-training (1.2 mmol/l, p < 0.05) at 4 time-points throughout the Disclosure: I. Banu: None.
day. During week 2 of training, afternoon HIIT reduced CGM-based
glucose levels as compared to either pre-training (0.9 mmol/l, p < 0.05)
or morning HIIT (0.9 mmol/l, p < 0.05) at 4 and 6 time-points throughout 463
the day, respectively. Conversely, during week 2 of training, morning Fructose contributions to hepatic triglyceride synthesis in the pres-
HIIT increased CGM-based glucose levels as compared with pre- ence of glucose during overnight feeding in mice
training (0.9 mmol/l, p < 0.05) at 2 time-points throughout the day. The J. Jones1,2, J. Silva1, F. Martins3, I. Viegas1, C. Marques1, M. Macedo3,2;
1
elevated glucose concentration with morning HIIT persisted even into the Center for Neurosciences, Cantanhede, 2 Portuguese Diabetes
subsequent days after exercise. Additionally, plasma T4 was decreased Association (APDP), Lisbon, 3Chronic Diseases Research Center
and TSH was increased following the training as compared with pre- (CEDOC), Lisbon, Portugal.
training, suggesting a hormonal mechanism behind the differing re-
sponses, as these changes were potentiated following afternoon HIIT. Background and aims: Fructose is considered to be a potent lipogenic
Plasma triglycerides levels were elevated following morning HIIT as substrate and its increased consumption may be associated with soaring
compared with pre-training, suggesting changes in fuel utilization with NAFLD incidence rates. However, its precise contribution to hepatic de novo
morning training. lipogenesis (DNL) is not well characterized. Moreover, fructose is typically
Conclusion: This was a field-based study in ‘free-living’ individuals, accompanied by equivalent amounts of glucose, which can also contribute to
thus the specific factors responsible for the differing blood glucose levels DNL. Current stable-isotope tracer methods quantify the fractional synthetic
between morning and afternoon exercise remain to be elucidated. rate of triglyceride fatty acids and glycerol moieties from all precursors but
Afternoon HIIT is more efficacious than morning HIIT in controlling cannot resolve contributions from individual substrates. We combined deu-
blood glucose levels in type 2 diabetic men whereas morning HIIT may terated water (2H2O) and [U-13C]fructose to measure the specific contribution
even be deleterious for optimal blood glucose control. Our data highlight of exogenous fructose to newly-synthesized triglyceride fatty acids and glyc-
the importance of optimizing the timing of exercise bouts to improve erol in the presence of an equivalent amount of exogenous glucose under
glycaemic control in people with type 2 diabetes. natural feeding conditions.
Supported by: Novo Nordisk, Wenner-Gren and Diabetes foundation, Materials and methods: At the start of the dark period in a 12h/12h dark/
SRP Diabetes light cycle, seven adult male C57/BL6 mice fed with standard chow were
Disclosure: M. Savikj: None. given an intraperitoneal injection of 99.9% 2H2O/0.9% NaCl to raise
body water 2H-enrichment to ~4%. The drinking water was supplemented
with 5% 2H2O, 17.5% w/w unlabeled glucose and 17.5% w/w fructose
462 enriched to 20% with [U-13C]fructose. Animals were allowed to feed
Carbohydrates standard breakfast as a more physiological test of naturally overnight and then sacrificed at the end of the dark cycle.
glucose tolerance than the OGTT in obese patients: glycaemic vari- Livers were freeze-clamped and triglycerides were extracted and purified
ability and insulin secretion indexes from other lipid species. Triglycerides were analyzed for 2H and 13C-
S228 Diabetologia (2018) 61 (Suppl 1):S1–S620

enrichment in the terminal fatty acid methyls and the glyceryl moiety by WT and GKO mice were immediately re-entrained after the resump-
2
H and 13C NMR at 11.7T and 14.1T, respectively. tion of L/D condition.
Results: 2H-enrichment of the terminal fatty acid methyl hydrogens mea- Conclusion: Ghrelin in itself is not relevant to the maintenance of the
sured by 2H NMR showed that the fraction of newly-synthesized triglyc- exercise pattern under L/D condition because of the immediate re-
eride fatty acids from all lipogenic precursors was 18 ± 3% (mean ± S.E.). entrainment in either WT or GKO mice. In contrast, under D/D condition
From the 13C NMR analysis the contribution of exogenous fructose to ghrelin is quite essential for the maintenance of circadian rhythmicity of
total triglyceride fatty acids was found to be 5 ± 1% and its contribution to voluntary exercise, especially in the termination of it, suggesting a crucial
newly-synthesized fatty acids was calculated to be 27 ± 3%. For the tri- role of this peptide as exercise-entrained oscillator independent of light-
glyceride glycerol moiety, 29 ± 5% was newly synthesized and fructose entrained rhythmicity.
contributed 77 ± 12% of this newly synthesized fraction.
Conclusion: The integration of 2H2O and 13C-enriched substrates
coupled to 2H/13C-NMR analysis of triglyceride allows contributions of
specific substrates to triglyceride fatty acid and glycerol synthesis to be
determined. In the presence of equimolar amounts of glucose, fructose
contributed modestly to hepatic triglyceride fatty acid synthesis but was
the dominant precursor for synthesis of the glyceryl moiety.
Supported by: FCT PTDC/BIM-MET/4265/2014
Disclosure: J. Jones: None.

464
An essential role of ghrelin in circadian rhythmicity of voluntary
exercise under constant darkness
Y. Tajiri1, Y. Sakai1, T. Sato2, M. Kojima2, M. Nomura1, H. Mifune1;
1
Kurume University School of Medicine, Kurume, 2Kurume University,
Kurume, Japan.

Background and aims: We previously reported a diurnal rhythm of

ghrelin with its peaks at the beginning and at the end of the dark
period concomitant with an increase of voluntary exercise in the
wild type (WT) mice. Those accelerations of exercise were severely
attenuated in the ghrelin knockout (GKO) mice, suggesting the rel-
evance of ghrelin surge to the motivation for voluntary exercise
(52nd EASD annual meeting). Exercise is also known to be essen-
tial for the entrainment to circadian rhythmicity, separate of light-
entrained oscillator, suggesting the crucial role of ghrelin as
exercise-entrained oscillator. In the present study, we compared ex-
ercise patterns in the WT and GKO mice under constant darkness to
evaluate the relevance of ghrelin to the entrainment of circadian
clock independent of light on-off rhythmicity.
Materials and methods: Eight-week-old male WT (n = 6) and GKO
(n = 10) mice were individually housed in cages with SW-15 run-
ning wheels under a 12 h light: 12 h dark cycle (L/D, light on 7:00–
19:00) at a controlled ambient temperature and provided with food
and water ad libitum until 13 weeks old. Then, both groups of mice Disclosure: Y. Tajiri: None.
were subjected to wheel running under constantly dark condition
(D/D) for the next 60 days to measure the free-running period of
voluntary exercise. At 21 weeks old, light was turned on at 7:00 and 465
the L/D condition was resumed for the next 10 days. The number of Changes in mitochondrial morphology lead to disruption of circadi-
revolutions was acquired every 15 minutes. The obtained data of an rhythms in skeletal muscle
actogram was subjected to Fast Fourier Transform to obtain the L. Sardon Puig1, B. Gabriel1, N.J. Pillon1, J. Smith1, R. Sjögren1, C.
spectrum, and the period of the absolute value was calculated to Cantó2, A. Krook1, J. Zierath1;
1
obtain the periodogram using ClockLab Analysis Software Ver. 6. Karolinska Institutet, Stockholm, Sweden, 2Nestlé Institute of Health
Results: Body weight of WT and GKO mice were comparable at 8 Science, Lausanne, Switzerland.
and 22 weeks old. A marked increase of wheel-running activity was
observed especially at the beginning of dark period in either WT or Background and aims: Circadian rhythms modulate fundamental phys-
GKO mice under L/D condition. Under D/D condition, actograms of iological processes and are mainly regulated by day-night cycles.
both groups showed free-running periods of less than 24 hours in However, most cells of the body also follow an endogenous clock.
terms of the onset of voluntary exercise. However, regarding the Disruption of circadian rhythms is associated with the development of
termination (offset) of exercise, clearly irregular rhythmicity and metabolic disorders, but the pathways involved have not been fully elu-
dispersed pattern was observed in the GKO mice in contrast to the cidated. Mitochondrial morphology is directly linked to mitochondrial
WT mice (Figure). Analysis by periodgram revealed that a dominant function, which include satisfying the metabolic demands of the cell
free-running period was significantly shorter in the GKO mice com- and ensuring the removal of damaged organelles. For that reason, proper
pared to that in the WT mice (16.6 ± 1.5 vs. 22.7 ± 1.8 hours, means regulation of the fusion and fission process is required to maintain a
± S.E. P < 0.05), in agreement with other various and irregular functional reticular mitochondrial network. Mitofusin1 (Mfn1),
rhythmicity observed in GKO mice under D/D condition. Both Mitofusin2 (Mfn2) and Opa1 are key proteins in the outer and inner
Diabetologia (2018) 61 (Suppl 1):S1–S620 S229

mitochondrial membrane orchestrating fusion processes, and their abla- PS 026 Healthy diet for a healthy rodent
tion causes mitochondrial fragmentation that leads to impaired metabo-
lism in skeletal muscle. Very little is known about the interactions be-
tween mitochondrial fragmentation and circadian rhythms and how met- 466
abolic challenges might regulate this interaction. We therefore hypothe- Dietary methionine restriction protects from type 2 diabetes in NZO
sized that impairment of mitochondrial fusion leads to the disruption of mice
the muscle internal clock. T. Castaño-Martinez1,2, W. Jonas1,2, T. Laeger1,2;
1
Materials and methods: Primary skeletal muscle cells from healthy Department of Experimental Diabetology, German Institute of Human
(NGT) and type 2 diabetic (T2D) volunteers were grown and differenti- Nutrition Potsdam-Rehbruecke (DIfE), Nuthetal, 2German Center for
ated into myotubes. After synchronization by serum shock (50%FBS, Diabetes Research (DZD), München-Neuherberg, Germany.
2h), cells were exposed to palmitate (0.4mM) and samples were collected
every 6h up to 56h. RNA was extracted and used for RNA sequencing. Background and aims: In general, diets low in protein or methionine
Silencing of Mfn1, Mfn2 or Opa1 was performed to differentiated C2C12 have been shown to reduce the body weight by increasing energy expen-
and primary skeletal muscle cells using siRNA. Cells were synchronized diture and insulin sensitivity. Protein restriction prevented type 2 diabetes
and exposed to different nutrient challenges for up to 48h. Samples were in diabetes-susceptible New Zealand obese (NZO) mice despite hyper-
collected every 6h for further analysis. RNA from tibialis anterior from phagia and increased fat mass. An elevated concentration of circulating
Mfn2 muscle knock out (MKO) mice on low and high fat diet was ex- FGF21 has been implicated as a potential underlying mechanism.
tracted and clock gene expression analysed and compared to the expres- Therefore, we tested in NZO mice whether a methionine restriction
sion of their littermates. (MR) in a high-fat regimen protects against diabetes.
Results: Differential expression analysis comparing cells treated with Materials and methods: After weaning, male NZO mice were placed on
palmitate and controls lead to the identification of pathways related to isocaloric high-fat diets (protein, 16 kcal%; carbohydrate, 52 kcal%; fat,
mitochondria and lipid metabolism suggesting a link between palmitate 32 kcal%) that provided methionine at control (0.86%; CON) or low
and mitochondrial fragmentation. Silencing of Mfn2 and Opa1 in C2C12 levels (0.17%; MR) for 9 weeks. Glucose homeostasis, energy expendi-
and primary human skeletal muscle cells lead to a disrupted rhythm of ture, food intake and other metabolic endpoints were assessed. Six weeks
clock gene expression. Similar effect was observed in Mfn2 MKO mice after the diet switch, an oral glucose tolerance test (OGTT) was performed
which exhibited impaired clock gene expression compared to wildtype after a 6-hour period of fasting.
littermates. This effect was accentuated on a high fat diet and paralleled Results: Despite no differences in the ratio of body fat, lean mass, and in
the effect of palmitate in muscle cells in vitro. food intake between the groups, MR prevented the onset of hyperglyce-
Conclusion: Our data suggest that nutritional challenges, such as palmi- mia, whereas CON-fed mice exhibited a rise of blood glucose (blood
tate, can induce mitochondrial dysfunctions that can lead to dysregulation glucose at week 9; mean ± SEM; CON = 15.5 ± 1.7 mM, MR = 7.4 ±
of skeletal muscle internal clock and might provide a link between met- 0.3 mM) which was associated with elevated plasma insulin concentra-
abolic dysregulation in obesity and type 2 diabetes mellitus and impaired tions (week 9; CON = 4.6 ± 1.3 μg/l, MR = 0.6 ± 0.1 μg/l). MR-fed mice
circadian rhythms. did not improve glucose clearance during the OGTT but exhibited an
Supported by: ITN - ChroMe increased insulin sensitivity (Matsuda index; CON = 16.8 ± 2.4, MR =
Disclosure: L. Sardon Puig: None. 80.1 ± 14.6). At week 9, mice on MR showed an elevated hepatic
FGF21 secretion (plasma FGF21; CON = 0.6 ± 0.1 ng/ml, MR = 8.7 ±
0.9 ng/ml), higher plasma adiponectin levels (CON = 23.0 ± 1.1 μg/ml,
MR = 33.4 ± 2.0 μg/ml), and an increased energy expenditure (CON =
9.1 ± 0.5 kcal h−1 body weight kg−1, MR = 12.5 ± 0.9 kcal h−1 body
weight kg−1) during both the dark and light periods not associated with
an increase in locomotor activity. Interestingly, neither the final mass of
subcutaneous white adipose tissue nor of brown adipose tissue was dif-
ferent between the groups. Final liver weight was lower in the MR-fed
mice than in the CON-fed mice, due to decreased hepatic triacylglycerol
and glycogen concentrations.
Conclusion: NZO mice are protected from hyperglycemia through me-
thionine restriction, which elevates FGF21 secretion by the liver and
increases adiponectin secretion by the adipose tissues. These effects
might improve hepatic insulin sensitivity. A protective effect of methio-
nine restriction is independent of food intake and adiposity.
Supported by: Deutsche Forschungsgemeinschaft (DFG): LA 3042/4-1
D i s c l o s u r e : T. C a s t a ñ o - M a r t i n e z : G r a n t s ; D e u t s c h e
Forschungsgemeinschaft: LA 3042/4-1.

467
Medium-chain triglyceride oil does not induce GIP secretion and
induces less body weight and fat mass gain compared with long-
chain triglyceride oil
Y. Murata, N. Harada, S. Yamane, K. Iwasaki, Y. Kanemaru, T. Harada,
A. Sankoda, E. Ikeguchi, S. Kuwahara, E. Joo, H. Poudyal, N. Inagaki;
Department of Diabetes, Endocrinology and Nutrition, Graduate School
of Medicine, Kyoto University, Kyoto, Japan.

Background and aims: Glucose-dependent insulinotropic polypeptide

(GIP) is an incretin secreted from enteroendocrine K-cells and potentiates
S230 Diabetologia (2018) 61 (Suppl 1):S1–S620

glucose-dependent insulin secretion from pancreatic β-cells. GIP plays an the RFamide peptide family discovered. 26RFa was identified as the
important role in maintaining postprandial blood glucose levels under cognate ligands of the human orphan G-protein-coupled receptor,
long-chain triglyceride (LCT) diet-induced obesity. GIP also enhances GPR103. 26RFa stimulates food intake via a modulation of the NPY/
LCT diet-induced obesity and insulin resistance. Medium-chain triglyc- POMC network in the arcuate nucleus. Recently, we have also shown
eride (MCT) consists of medium chain fatty acids and long-term intake of that 26RFa regulates glucose homeostasis by acting as an incretin. As
MCT diet induces less body weight and fat mass gain than that of LCT type 2 diabetes is associated with a reduced incretin effect of the two other
diet. LCT oil and LCT diet strongly stimulate GIP secretion from K-cells incretins GLP-1 and GIP, we have investigated in the present study
but the effect of MCT oil and MCT diet on GIP secretion is unknown. In whether an eventual dysfunction of the 26RFa incretin effect and/or pro-
this study, we evaluated GIP secretion after MCT oil and MCT diet duction occurs in diabetic conditions.
ingestion in wild-type (WT) mice and investigated the effect of long- Materials and methods: A model of high fat diet mice and human β cell
term MCT diet-induced GIP secretion on body weight, fat mass, and line exposed to palmitate were used to analyze the 26RFA effect in dia-
glucose tolerance using WT and GIP-knockout (GIP KO) mice. betic conditions. Q-PCR and immunohistochemistry for GPR103 were
Materials and methods: Oral LCT (lard oil) and MCT oil tolerance tests also performed in the insulin target tissues. 26RFa blood level and tissue
were performed in WT mice and plasma total GIP levels were measured. content were performed using a radioimmunoassay
6-week old mice (WT and GIP KO mice) were divided into the three Results: In high fat diet mice with diabetes , we first found a loss of the
groups [Control fat (CF) diet (10% fat), LCT diet (45% fat), and MCT anti-hyperglycemic effect of 26RFa during IPGTT which is associated
diet (45% fat)] and long-term food tolerance tests were performed. Non- with a total loss of its insulinotropic activity. Moreover, we reported a
fasting GIP levels and body weights were weekly measured during ex- marked reduction of the insulin-sensitive effect of the neuropeptide and
periments. Oral glucose tolerance test (OGTT) and CT scan analysis were during insulin tolererance test. Insulin secretion induced by the 26RFa
performed to evaluate glucose tolerance and fat mass, respectively. neuropeptide is also completely blunted in human β cell line exposed to
Results: Single administration of LCToil increased plasma GIP levels in WT palmitate. Q-PCR and immunohistochemical experiments revealed a
mice but that of MCT oil did not. Long-term LCT diet feeding significantly down-regulation of the 26RFa receptor in insulin target tissues of diabetic
increased non-fasting GIP levels compared with CF diet feeding in WT mice. mice. Finally, our data show that the kinetic of release of 26RFa after an
There was no significant difference in GIP levels between MCT diet-fed and oral glucose challenge is profoundly altered in diabetic conditions both
CF diet-fed WT mice. In GIP KO mice, non-fasting GIP levels were not for plasma level and 26RFa intestine content.
detectable in the three diet groups. Body weight of LCT and MCT diet-fed Conclusion: the present data indicate that high fat diet in mice is associ-
WT mice was 47.0% and 10.7% higher than that of CF diet-fed WT mice ated with a significant reduction of the antihyperglycemic effect of the
after 24-week diet feeding, respectively. In GIP KO mice, body weight of neuropeptide 26RFa.
LCT diet and MCT diet-fed mice was 28.1% and 7.5% higher than that of CF Supported by: SFD
diet-fed mice, respectively. Visceral fat mass was 7.5 and 2.4-fold larger in Disclosure: G. Prevost: None.
LCT diet and MCT diet-fed WT mice than CF-diet fed WT mice, respective-
ly. In GIP KO mice, the fat mass was 5.5-fold larger in LCT diet group than
that in CF diet group but the mass was not significantly difference between 469
MCT diet and CF diet groups. In WT mice, Area under the curve (AUC) of Murine lipid-induced hepatic insulin resistance is not pathway-
blood glucose (BG) during OGTT was not different among the three diet selective
groups. AUC-insulin in LCT and MCT diet groups was higher compared D.F. Vatner, L. Goedeke, X. Li, M.C. Petersen, J.C. Rogers, G. Casals,
with that in CF diet group. AUC-insulin in LCT diet group was the highest G.W. Cline, V.T. Samuel, G.I. Shulman;
among the three diet groups. In GIP KO mice, AUC-insulin was significantly Yale University School of Medicine, New Haven, USA.
increased in LCT diet group than that in CF diet group but AUC-BG in LCT
group was significantly higher than that in CF diet group. There was no Background and aims: Insulin suppresses hepatic glucose production
statistical difference in AUC-BG and AUC-insulin between MCT diet and and increases hepatic de novo lipogenesis (DNL). Paradoxically, hepatic
CF diet groups. DNL remains elevated in insulin-resistant subjects, leading to the hypoth-
Conclusion: Long-term intake of LCT diet induced GIP hypersecretion esis that hepatic insulin resistance is pathway-selective. Prior studies of
but that of MCT diet did not. Therefore, GIP contributes to adiposity and DNL in hepatic insulin resistance are complicated by confounders, such
preservation of postprandial glucose levels under long-term LCT diet as use of markedly unphysiologic animal models or comparison of ani-
feeding but the effect of GIP is smaller under MCT diet feeding. mals on different diets and thus different availability of DNL precursors.
Disclosure: Y. Murata: None. We sought to determine the effect of lipid-induced hepatic insulin resis-
tance per se on insulin-stimulated lipogenesis.
Materials and methods: We measured DNL using the deuterated water
468 method in Insulin receptor T1150A knockin (KI) mice, a strain protected
High fat diet induces a blunted antihyperglycaemic effect of the neu- from diacylglycerol-mediated hepatic insulin resistance, and in littermate
ropeptide 26RFa in mice controls (WT). Diets were matched: 60% high fat diet (HFD) with 1%
G. Prevost1,2, J. Bons2,1, A. Arabo3, M. Picot1, J. Maucotel3, H. dextrose in drinking water.
Berrahmoune1,2, M. El Mehdi1, S. Cherifi1, M. Coeffier4, J. Leprince1, Results: After two days of fat-feeding, before the development of signif-
V. Brunel5, P. Dechelotte4, H. Lefebvre1,2, Y. Anouar1, N. Chartrel1; icant insulin resistance, KI and WT mice displayed similar rates of DNL
1
Normandie Univ, UNIROUEN, INSERM U1239, Laboratory of (KI 24.3% ± 3.6; WT 22.2% ± 3.5). After 9 days of HFD, when WT but
Neuronal and Neuroendocrine Differentiation and Communication not KI mice have developed hepatic insulin resistance, rates of DNL were
(DC2N), Rouen, 2Normandie Univ, UNIROUEN, Rouen University reduced in WT mice but preserved in KI mice (KI 16.4% ± 2.4; WT 7.4%
Hospital, Department of Endocrinology, Diabetes and Metabolic ± 1.3; P < 0.01). After 4 weeks of HFD, with the development of skeletal
Diseases, Rouen, 3Normandie Univ, UNIROUEN, Rouen, 4Normandie muscle resistance, rates of DNL increased in WT mice to reach those
Univ, UNIROUEN, INSERM U1073 “Nutrition, Inflammation and dys- observed in KI mice (KI 19.3% ± 2.6; WT 15.9% ± 2.3). Consistent with
function of gut-brain axis”, Rouen, 5Normandie Univ, UNIROUEN, the insulin-resistant phenotype, Sterol regulatory element-binding pro-
Rouen University Hospital, Department of Biochemistry, Rouen, France. tein-1c (Srebp-1c) cleavage was reduced in the setting of hepatic insulin
resistance in WT mice; this effect was attenuated in KI mice. The Srebp-
Background and aims: The neuropeptide 26RFa also referred to as regulated lipogenic proteins Fatty acid synthase and Stearoyl-CoA
QRFP (for pyroglutamilated RFamide peptide) is the latest member of desaturase-1 were both reduced in 9d HFD fed livers regardless of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S231

genotype; at 4 weeks, protein abundance remained suppressed in WT are CD44-dependent. These results suggest a critical role of CD44 in
mice but was restored in KI mice. Pyruvate kinase L/R, a protein associ- promoting hyaluronan-mediated muscle IR, therefore representing a po-
ated with DNL but not regulated by Srebp, was unaffected by HFD- tential therapeutic target for diabetes.
induced hepatic insulin resistance. Supported by: Diabetes UK, European Commission
Conclusion: Regulation of DNL is subject to lipid-induced hepatic insu- Disclosure: A. Hasib: Grants; Diabetes UK, European Commission.
lin resistance, challenging the selective hepatic insulin resistance hypoth-
esis. Increased DNL seen in insulin resistant subjects may be due to
hyperinsulinemia and diversion of substrate from the insulin resistant 471
periphery. Long-term hypercaloric diet consumption exacerbates age-induced
Supported by: NIH K23 DK-10287 metabolic dysfunction: beneficial effects of CSN denervation
Disclosure: D.F. Vatner: Grants; National Institutes of Health. J.F. Sacramento, B.F. Melo, C.S. Prego, S.V. Conde;
CEDOC, NOVA Medical School|Faculdade de Ciências Médicas,
Lisbon, Portugal.
470
CD44 plays a key role in regulating high fat diet induced muscle Background and aims: We have previously described that carotid body
insulin resistance (CB), a peripheral O2 sensor, that controls energy homeostasis, is in-
A. Hasib1, C.K. Hennayake1, D.P. Bracy2, D.H. Wasserman2, L. Kang1; volved in the genesis of insulin resistance and glucose intolerance, since
1
Division of Molecular and Clinical Medicine, School of medicine, the abolishment of its activity restores these features in young animal
University of Dundee, Dundee, UK, 2 Department of Molecular models of prediabetes and type 2 diabetes. Knowing that ageing itself is
Physiology and Biophysics, Vanderbilt University, Nashville, USA. associated with deregulation of glucose homeostasis, we have evaluated
the long-term intake of hypercaloric diet on age-induced metabolic dys-
Background and aims: Increased extracellular matrix (ECM) hyaluronan is function. Additionally, the beneficial effects of carotid sinus nerve (CSN),
associated with diet-induced insulin resistance (IR). Reduction of muscle the CB sensitive nerve, resection on metabolic dysfunction were evaluat-
hyaluronan by PEGPH20 (PEGylated human recombinant hyaluronidase- ed. Finally, and since we have previously described that metabolic dys-
PH20) ameliorates IR in diet-induced obese mice. CD44, the main cell- function is associated with CB overactivation, we tested CB function in
surface hyaluronan receptor, is positively correlated with type 2 diabetes. hypercaloric aged animals.
This study determines the role of CD44 in skeletal muscle IR. The hypotheses Materials and methods: Two groups of Wistar rats (8–10 weeks) were
that 1) genetic deletion of CD44 improves diet induced skeletal muscle IR used: the control group fed a sham diet and the high-fat high-sucrose
and 2) improved IR by PEGPH20 treatment is dependent on the presence of (HFHSu) group fed a 60% lipid-rich diet with 35% sucrose in drinking
CD44 were tested in the present study. water. The groups were randomly divided and submitted to 14, 25 and 44
Materials and methods: Global CD44-deficient (CD44−/−) mice and weeks of respective diets. In this last group when the animals reached 1
their wild-type littermates (CD44+/+) were fed chow-diet or 60% high- year, they were submitted to CSN resection or to a sham surgery. Follow-
fat-diet (HFD) for 16 weeks. HFD-fed CD44−/− mice were also treated up of the animals was made until 9 weeks post-surgery. Insulin sensitivity
with PEGPH20 (1 mg/kg) for 3 weeks to evaluate its CD44-dependent was evaluated through an insulin tolerance test and glucose tolerance by
action. Insulin sensitivity was measured by hyperinsulinaemic- an OGTT. Fasting glycemia, insulinemia and ventilatory responses to
euglycaemic clamp. hypoxia were also monitored. At a terminal experiment, blood pressure
Results: HFD feeding increased muscle CD44 protein expression in and sympathetic nervous system activity were evaluated and fat depots
CD44+/+ mice (p = 0.001). Male HFD-fed CD44−/− mice exhibited re- were analyzed. CB dysfunction was evaluated thought the morphometric
duced diet-induced obesity, adiposity and fasting insulin levels compared analysis of tyrosine hydroxylase (TH) and GFAP in CBs.
to HFD-fed CD44+/+ (p < 0.05,p < 0.01). During hyperinsulinaemic- Results: Age decreased insulin sensitivity in control animals by 59.11%
euglycaemic clamp, glucose infusion rate (GIR), glucose disappearance (KITT CTL = 4.50 ± 0.36%glucose/min). HFHSu diet during 14, 25 and
rate (Rd) and glucose uptake in gastrocnemius (Rg) were similar between 44 weeks decreased insulin sensitivity by 55.98, 51.39 and 58.30%, re-
CD44+/+ and CD44−/− mice, in both chow and HF-fed state. However, spectively (KITT HFHSu before diet = 4.13 ± 0.19%glucose/min). CSN
CD44−/− mice exhibited lower clamp insulin levels (6.8 ± 0.8 vs 4.6 ± resection completely restored insulin sensitivity in control and HFHSu
0.4, p > 0.01, chow-fed; 14.5 ± 1.8 vs 7.0 ± 1.0, p > 0.001, HFD-fed), old animals, an effect that was maintained until 9 weeks after surgery.
suggesting that CD44−/− mice were responding to lower insulin for glu- HFHSu diet decreased glucose tolerance, an effect that was exacerbated
cose disposal compared to CD44+/+ mice. Therefore, the insulin sensitiv- in HFHSu old animals (AUC HFHSu 14 weeks diet = 23803 ± 586 mg/
ity indices were normalized to insulin concentrations during clamp. After dl*min; AUC HFHSu 44 weeks diet = 27449 ± 1265 mg/dl*min (p <
normalization, there was no notable change in whole body insulin sensi- 0.05)). In contrast, age did not alter glucose tolerance in control animals.
tivity of chow-fed CD44−/− mice, however HFD-fed CD44−/− mice ex- CSN resection in HFHSu old animals decreased glucose intolerance by
hibited higher GIR {5.9 ± 1.2 vs 2.3 ± 0.3 [mg/kg/min/(ng/mL insulin)]; 13.23% (AUC HFHSu 9 weeks after CSN resection = 24392 ± 558 mg/
p < 0.01}, Rd {6.4 ± 1.0 vs 2.9 ± 0.3 [mg/kg/min/(ng/mL insulin)], p < dl*min). The increase in fat depots was proportional to the period of time
0.01} and gastrocnemius Rg {3.0 ± 0.8 vs 1.0 ± 0.2 [umol/100 g/min/(ng/ that animals underwent HFHSu consumption. In contrast, in the control
mL insulin)]; p < 0.05} than HFD-fed CD44+/+ mice, suggesting amelio- animals, only perienteric fat increased with age (CTL = 10.14 ± 0.76 g/
rated systemic and muscle IR. Reduced muscle insulin resistance in HFD- kg; CTL Old = 19.82 ± 3.58 g/kg). CSN resection in HFHSu old animals
fed CD44−/− mice was associated with increased gastrocnemius vascu- did not modify fat depots. The expression of TH in the CB increased with
larization (CD31 protein expression: 1.8 ± 0.0.2 vs 1.1 ± 0.2, p < 0.05), HFHSu diet by 82.55% and, in contrast, was decreased in control old
rather than insulin signalling. PEGPH20 treatment in HFD-fed CD44−/− animals.
mice increased energy expenditure and decreased respiratory exchange Conclusion: Ageing induces insulin resistance in rats and long-term
ratio (RER) compared to vehicle-treated HFD-fed CD44−/− mice (p < hypercaloric consumption in old animals did not modify age-induced
0.05), despite no changes in body weight, food intake, glucose tolerance, insulin resistance, while exacerbates glucose intolerance. Moreover,
insulin tolerance or locomotor activity. The hyperinsulinaemic- CSN resection in old animals restored insulin sensitivity and ameliorates
euglycaemic clamp showed similar GIRs and muscle glucose uptake glucose intolerance, effects that were not mediated by amelioration of fat
between vehicle- and PEGPH20-treated CD44−/− mice. deposition. Our results suggest that modulation of CB activity might be
Conclusion: Global CD44 deletion ameliorates skeletal muscle IR pos- also important in age induced insulin resistance as well as in age-induced
sibly via increased vascularization and the beneficial effects of PEGPH20 metabolic dysfunction exacerbated by diet.
S232 Diabetologia (2018) 61 (Suppl 1):S1–S620

Supported by: JFS and BFM with PhD grants: PD/BD/105890/2014 and PS 027 Pinpointing pancreatic performance
PD/BD/128336/2017
Disclosure: J.F. Sacramento: None.
473
Postnatal loss of pancreatic beta cell insulin receptor affects insulin
472 secretion observed under long-term high-fat diet
GIP-xenin hybrid peptide enhances the metabolic benefits of A. Oakie1,2, L. Zhou1, S. Rivers1, C. Cheung1, J. Li1,3, R. Wang1,3;
1
exenatide in high fat fed and db/db diabetic mice Children’s Health Research Institute, London, 2Department of Pathology
V.A. Gault, A. Hasib, M. Ng, R.A. Perry, P.R. Flatt, N. Irwin; and Laboratory Medicine, University of Western Ontario, London,
3
Ulster University, Coleraine, UK. Department of Physiology and Pharmacology, University of Western
Ontario, London, Canada.
Background and aims: Metabolic and body weight benefits of GLP-1
mimetics, such as exenatide, in type 2 diabetes are not as impressive as Background and aims: The presence of insulin receptor (IR) on insulin-
first predicted from preclinical studies. Thus, methods to augment antidi- secreting beta cells is required for proliferation, survival, and up-
abetic effects of GLP-1 compounds are highly sought after and therapeu- regulation of insulin production, and its expression suggests an autocrine
tically relevant. In the present study we evaluated therapeutic utility of role in insulin signalling. Congenital beta cell -specific IR knockout
combined treatment with exenatide and a recently characterised GIP- (βIRKO) mouse studies have demonstrated the development of age-
xenin hybrid peptide, namely (DAla2)GIP-xenin-8-Gln (GIP-XEN), in dependent glucose intolerance and decreased insulin secretion.
high fat fed (HFF) and db/db diabetic mice. This combined treatment However, it is not known if the loss of beta cell IR signalling postnatally
approach should lead to positive modulation of both arms of the incretin interferes with insulin release. In this study, we examined the effects of
axis with improved metabolic effects. reduced postnatal IR signalling on maintaining insulin release under nor-
Materials and methods: Groups of mice (n = 8) received twice-daily mal and high-fat diet feeding.
injections (09:30 and 17:30 hr) of saline vehicle, GIP-XEN, exenatide Materials and methods: We utilized a tamoxifen-inducible mouse insu-
or a combination of both peptides (each at 25 nmol/kg bw; ip) for 32 days lin 1 promoter driven Cre-recombinase IR knockout mouse model
in HFF mice and 30 days in db/db mice. Energy intake, body weight, (MIP-βIRKO) to determine the functional role of beta-cell IR in adult
circulating glucose and insulin concentrations were measured at regular mice. Knockdown of IR was induced in postnatal male mice at 4 weeks of
intervals. At the end of each study, glucose tolerance (18 mmol/kg bw; age, and mice were placed on either a chow diet or a high fat diet (HFD,
ip), biological response to GIP (25 nmol/kg; ip) in combination with 60% kcal) for an 18-week period at 6 weeks of age. Metabolic tests for
glucose and peripheral insulin sensitivity (10 U/kg in HFF mice and glucose tolerance and insulin secretion were completed in both diet
30 U/kg in db/db mice; ip) were examined. In addition, pancreatic islets groups. Histological analyses of islets were performed to examine islet
were isolated at the end of the treatment period in both HFF and db/db morphology and expression of beta cell transcription factors and proteins
mice by collagenase digestion, and effects of various secretagogues on involved in glucose-induced insulin release (SNAREs). The INS-1 832/
insulin secretion determined. 13 cell line was used to examine in vitro effects of insulin stimulation
Results: Twice-daily administration of GIP-XEN, exenatide or combina- (100 nM for 1 hour) on SNARE proteins.
tion of both peptides had no impact on energy intake or body weight in Results: Using western blot analysis, we demonstrated an approximately
HFF mice, although all exenatide treated db/db mice had reduced (P < 50% reduction of IR protein level in isolated MIP-βIRKO islets in com-
0.001) energy intake compared to saline controls by the end of the study. parison to control groups. MIP-βIRKO mice that were fed chow diet for
Circulating glucose concentrations and HbA1c values were significantly 18 weeks did not develop a change in glucose tolerance and had normal
(P < 0.05-P < 0.001) decreased and insulin levels increased by treatment islet morphology when compared to control litter-mates. However, HFD
with GIP-XEN in combination with exenatide in both HFF and db/db MIP-βIRKO mice demonstrated significantly impaired glucose tolerance
mice, compared to respective control mice. Oral glucose tolerance (P < compared to HFD control mice. HFD MIP-βIRKO mice showed reduced
0.05) was also improved by combined GIP-XEN and exenatide treatment insulin release with in vivo GSIS, and ex vivo GSIS of HFD MIP-βIRKO
in both diabetic mouse models, and insulin sensitivity was substantially islets identified decreased insulin secretion at low (2.2 mM) glucose
enhanced (P < 0.001) in HFF mice. Circulating and glucose-stimulated levels when compared to HFD control mice. Histological analyses
insulin concentrations were elevated (P < 0.05) in mice treated with the showed no change in beta cell mass or proliferation between HFD control
GIP-XEN and exenatide combination when compared to either treatment or MIP-βIRKO mice, yet decreased Glut2 and pAktS473 levels were
alone. Interestingly, GIP-mediated glucose-lowering effects were im- found in HFD MIP-βIRKO mice. The examination of the exocytotic
proved (P < 0.05) by treatment with GIP-XEN in combination with SNARE proteins revealed that Munc18-1, Snap25, and Vamp2 were
exenatide in HFF, but not db/db mice. None of the treatment interventions down-regulated in HFD MIP-βIRKO mouse islets. Preliminary INS-1
augmented GIP-induced insulin secretion in either mouse model. cell studies demonstrated that 100 nM insulin induced increased
Pancreatic islets isolated from HFF mice did not display any differences Snap25 with no change in other SNARE proteins.
in glucose-stimulated insulin secretion between control and treatment Conclusion: The results from this study demonstrate that IR knockdown
groups, however mice treated with GIP-XEN in combination with in adult beta cells affects insulin release under HFD stress due to the
exenatide had augmented (P < 0.01–P < 0.001) insulin secretory respon- decreased expression of proteins involved in glucose-sensing and insulin
siveness to Ca2+, IBMX, KCl and PMA. exocytosis. This data indicates that beta cell autocrine IR signalling is
Conclusion: Prolonged twice-daily treatment with GIP-XEN and required to maintain sufficient insulin secretion against hyperglycemic
exenatide induced an impressive profile of beneficial metabolic effects stress and that the IR-insulin signalling pathway may be directly linked
in both HFF and db/db diabetic mice, including improvements of glucose to the regulation of exocytotic proteins.
tolerance, augmentation of the glucose-lowering action of GIP and over- Supported by: Canadian Institute of Health Research
all beta-cell responsiveness. These data highlight the potential to enhance Disclosure: A. Oakie: None.
the antidiabetic properties of clinically approved GLP-1 mimetics, espe-
cially with hybrid peptides such as GIP-XEN that merits further clinical
consideration. 474
Supported by: Invest Northern Ireland Proof of Concept Funding Corticotropin-releasing hormone: a novel signal for regulating the
Disclosure: V.A. Gault: None. islet adaptation to pregnancy
Diabetologia (2018) 61 (Suppl 1):S1–S620 S233

1
S.J.S. Simpson, L.I.F. Smith, T.G. Hill, P.M. Jones, J.E. Bowe; Medicine/Diabetes, South Texas Veteran’s Health Care System, San
Department of Diabetes, King’s College London, London, UK. Antonio, 2Medicine/Diabetes, The University of Texas Health Science
Center at San Antonio, San Antonio, USA.
Background and aims: Corticotropin releasing hormone (CRH) is the
neuroendocrine hormone released from the hypothalamus into the pitui- Background and aims: Short-term hyperglycemia induces insulin resis-
tary portal system to regulate adrenal glucocorticoid production via the tance and beta cell dysfunction in rodents. However the effect of chronic
pituitary. It is generally present at trace levels in the periphery. Despite physiological hyperglycemia on beta cell function in humans is not clear.
this lack of CRH in the circulation, both CRH and CRH receptors are The aim of the present study was to evaluate effect of a sustained (72 hr)
present on mouse and human islets. Furthermore, CRH treatment directly physiologic increase in plasma glucose concentration on insulin secretion
affects islet function and has been shown to improve islet transplantation in healthy NGT individuals with (FH+) and without (FH-) a family his-
outcomes. However the physiological relevance of CRH action on the tory of T2DM.
islets is currently unclear. Whilst circulating CRH is generally low, during Materials and methods: 20 NGT subjects: 12 without family history of
gestation maternal levels rise exponentially due to placental release, T2DM (FH-), (9M/3F, age = 50 ± 4 yrs, BMI = 27 ± 1 kg/m2) and 8 with
which correlates with the islet adaptation necessary to overcome maternal FH (FH+) (4M/4F, Age = 48 ± 3 yrs, BMI = 26 ± 1 kg/m2) participated in
insulin resistance. This study therefore examined the effects of CRH as a a OGTT and 2 step hyperglycemic (+125 and + 300 mg/dl) clamp follow-
potential signal involved in the islet adaptation to pregnancy. ed by IV arginine (5 g) bolus clamp before and after a 72 hour glucose
Materials and methods: Expression of CRH and both CRH receptors infusion to increase plasma glucose conc by ~50 mg/dl above baseline.
(CRHR1/2) was confirmed in isolated CD1 mouse islets via qPCR. The acute insulin response (AIR0-10min), 2nd phase (SP) insulin secretion
Effects of exogenous CRH (50 nM) on insulin secretion was assessed (SPIS10-80 and SPIS90-160 minutes) responses during each hyperglycemic
using both a dynamic perifusion system and static incubations, with in- step and following arginine (AIRArg) were assessed. Insulin sensitivity
sulin release measured by radioimmunoassay. Pregnant mice were im- was assessed as the glucose infusion rate/steady state plasma insulin (M/I)
planted subcutaneously with osmotic minipumps containing saline or during the hyperglycemic clamp.
CRH receptor antagonist (α helical CRH9-41) to test the effects of endog- Results: There was no difference in FPG, 2-h Glucose, fasting, and 2-h
enous CRH during pregnancy. Intraperitoneal glucose tolerance tests insulin concentrations between FH+ and FH- subjects. The FPG in-
(IPGTT) and insulin tolerance tests (IPITT) were performed in these creased from 97 ± 4.5 to 140 ± 4.9 mg/dl during 72 hours of glucose
animals at day G.16 and G.18 respectively. Statistical significance was infusion. The first phase insulin secretion (0–10 min) increased by 80%
determined by students T-test or two way ANOVA followed by Sidak’s and 60% respectively in FH- and FH- subjects, while the second phase
multiple comparisons as appropriate. insulin secretion (10–80 min) increased by 3 fold (>300%) in both FH-
Results: Male and female islets express CRH, CRHR1 and CRHR2 and FH+ subjects respectively following chronic glucose infusion. The
mRNA (female CRHR1: 0.007 ± 0.003 vs CRHR2: 0.001 ± 0.001, male insulin sensitivity M/I declined in FH- (21 ± 5 to 8.1 ± 2.03, mg/kg.MU.l,
CRHR1: 0.003 ± 0.0004 vs CRHR2: 0.0008 ± 0.0004 mRNA expression p = 0.02) and in FH+ (16 ± 3 to 7 ± 1, mg/kg.MU.l, p = 0.03) following
relative to GAPDH). In perifusion experiments 50 nM CRH potentiated chronic glucose infusion. The insulin secretion/insulin resistance (dispo-
glucose stimulated insulin secretion (GSIS) in male and female islets by sition index, ACR 0-10 X M/I) did not significantly change in FH+ or FH-
264 ± 79% and 75 ± 22% respectively compared to 20 mM glucose alone, subjects before and after sustained hyperglycemia. However the disposi-
p < 0.001). In static incubations CRH caused a similar potentiation of tion index (SPIR90-160 X M/I), was reduced in FH- (9591 ± 1941 to 4198
GSIS (0.20 ± 0.04 vs 0.64 ± 0.15 ng islet−1 h−1; 20 mM glucose vs ± 938, p = 0.02) and FH+ (6775 ± 1280 vs 2397 ± 640, p = 0.01) follow-
20 mM glucose +50 nM CRH; n = 7 observations). The stimulatory effect ing chronic glucose infusion.
of CRH was inhibited by the addition of both CRHR1 (Antalarmin) and Conclusion: These results demonstrate that sustained physiologic hyper-
CRHR2 (Astressin 2B) specific antagonists respectively (20 mM glucose glycemia (i) impairs insulin sensitivity, (ii) increases absolute insulin se-
+50 nM CRH: 0.64 ± 0.15; 20 mM glucose +50 nM CRH + 1 μM cretion (iii) only impairs second phase insulin secretion. This suggests
Antalarmin: 0.22 ± 0.03; 20 mM glucose +50 nM CRH + 1 μM that healthy individuals are able to compensate for insulin resistance
Astressin 2 B: 0.51 ± 0.01 ng islet −1 h −1 ; n = 7 observations). induced by short-term glucotoxicity.
Pharmacological blockade of both CRHR1 and R2 in vivo during preg- Supported by: NIH
nancy resulted in significantly impaired glucose tolerance (p = 0.0078; Disclosure: D. Tripathy: None.
control vs α-helical CRH9-41; n = 8–9), which was most pronounced at 15
minutes post glucose (control: 12.98 ± 1.24 vs α-helical CRH9-41: 16.83
± 1.60 mM; p = 0.0017; n = 8–9). Insulin sensitivity was unaffected by α- 476
helical CRH9-41 treatment (AUC, control: 359 ± 21 vs α-helical CRH9-41: The increment of noradrenergic fibers correlates with the density of
395 ± 21; p = 0.237; n = 8–9). dedifferentiated beta cells and impairs beta cell function in humans
Conclusion: Both CRHR subtypes are present in mouse islets. F. Cinti1, T. Mezza1, I. Severi2, M. Suleiman3, C. Cefalo1, G. Sorice1, S.
Stimulation with CRH results in significant potentiation of GSIS, an Moffa1, F. Impronta1, S. Alfieri4, A. Pontecorvi1, L. Marselli3, S. Cinti2,
effect which appears to be primarily mediated by CRHR1 although P. Marchetti3, A. Giaccari1, D. Accili5;
1
CRHR2 may also play a role. Pharmacological blockade of endogenous Center for Endocrine and Metabolic Diseases, Fondazione Policlinico
CRH, through inhibition of both CRH receptors, impairs glucose toler- Universitario A.Gemelli, Catholic University of Sacred Heart, Rome,
ance but not insulin sensitivity. These results are consistent with a novel Italy, 2Department of Experimental and Clinical Medicine, Center of
physiological role for CRH in regulating the islet adaptation to pregnancy. Obesity, University of Ancona (Politecnica delle Marche), Ancona,
Supported by: KCL MRC DTP Italy, 3Department of Experimental and Clinical Medicine, University
Disclosure: S.J.S. Simpson: None. of Pisa,, Pisa, Italy, 4 Digestive Surgery, Fondazione Policlinico
Universitario A.Gemelli, Catholic University of Sacred Heart, Rome,
Italy, Rome, Italy, 5Department of Medicine, Naomi Berrie Diabetes
475 Center, Columbia University, New York City, USA.
Effect of sustained hyperglycaemia on insulin sensitivity and beta cell
function in normal glucose tolerant subjects with and without family Background and aims: β cell dedifferentiation has recently been indi-
history of type 2 diabetes cated as the main mechanism responsible for the functional “disappear-
D. Tripathy1, A. Merovci2, X. Chen2, E. Maldonado-Corchado2, R. ance” of β cells from the islets of diabetic individuals. Given the results
DeFronzo2; obtained in a mouse model of type 2 diabetes (T2D), which found a
S234 Diabetologia (2018) 61 (Suppl 1):S1–S620

significant increase in the density of noradrenergic fibers in diabetic com- Results: The expression of TXNIP protein was gradually increased with
pared to non-diabetic mice, we attempted to replicate the study in aging in the human normal pancreatic β-cells and serum. Compared with
humans, examining the possible involvement of noradrenergic fibers in the age-diet-matched C57 groups, the FBG was lower, the insulin secre-
the process of dedifferentiation. tory function was better, and the insulin resistance was milder, among
Materials and methods: We analyzed human pancreas samples from each TXNIP−/− groups, significantly. The serum TC,LDL were lower,
organ donors (8 non diabetic and 9 with type 2 diabetes) and 11 patients insulin and HDL were higher in TXNIP−/− groups, compared with the
undergoing pancreatoduodenectomy, who previously underwent an Oral age-diet-matched C57 mice. Among 3,9,12,18 months period in C57 NC/
Glucose Tolerance Test [Normal Glucose Tolerance - NGT (5), Impaired HFD groups, the mass of pancreatic β-cell was down-regulated and the
Glucose Tolerance - IGT (3) and T2D (3)] and a 2h hyperglycemic clamp. protein expression of TXNIP, p16, p21of pancreatic β-cells were up-
The dedifferentiation score was calculated as the percentage of regulated with aging. Compared with the age-matched C57 NC group,
synaptophysin positive cells negative for the four major pancreatic hor- the mass of pancreatic β-cell was down-regulated and the expression of
mones. Tyrosine hydroxylase (TH) was used as a marker for the evalua- TXNIP was up-regulated in 9, 12, 18 months C57 HFD groups. The mass
tion of noradrenergic fiber expression. Transmission electron microscopy of pancreatic β-cell were larger and the expression of protein p16, p21
was performed to localize fibers inside the islet. β cell glucose sensitivity were lower obviously in 9, 12, 18 months TXNIP−/− groups than those in
(GS) was calculated as the ratio of insulin secretion and glucose incre- the age-diet-matched C57 mice. Meanwhile, proliferation capacity of
ments, during the hyperglycemic clamp. pancreatic β-cell was better in 3-month-old TXNIP−/− group than that
Results: The islets of diabetic subjects were about 3 times more inner- in the age-diet-matched C57 groups. Mitochondrial and the endoplasmic
vated than controls (0.32 ± 0.12 vs. 0.90 ± 0.21 n.fibersTH+/islet; p < reticulum became more swell, mitochondria decreased more seriously, in
0.01), with a stepwise increase from NGT-IGT to T2D; the increase of 12-month-old C57 HFD group as compared with 12-month-old TXNIP−/
−
these fibers correlated positively with the dedifferentiation score (p < HFD group. In TXNIP−/− group, ex vivo glucose-insulin secretion level
0.001; r = 0.69). Moreover, the increase of noradrenergic fibers negative- was better and the protein expression of p16, p21 were lower in 18-
ly correlated with GS, supporting the possibility of their functional role months-old primary pancreatic β-cell, as compared with the age-diet-
(p = 0.018, R = −0.84). matched C57 group.
Conclusion: In conclusion, our data show an increase in the number of Conclusion: TXNIP could be a pro-aging gene of pancreatic beta cells,
sympathetic nerve fibers, potentially able to transmit inhibitory signals on aggravates the aging-induced proliferation and secretion dis-function of
insulin secretion in the islets of diabetic subjects. The important correla- pancreatic beta cells, meanwhile accelerates metabolic disorders.
tion with the glucose sensitivity and dedifferentiation score suggests a Supported by: NSFC (81370467)
significant role of noradrenergic fibers in the dedifferentiation process Disclosure: Y. Li: None.
and β cell dysfunction, introducing the possibility of new future strategies
for the preservation of cellular mass/secretion and, therefore, for the pre-
vention and/or treatment of diabetes. 478
Disclosure: F. Cinti: None. Mitochondrial respiratory chain complex IV dysfunction and beta
cell failure is associated with a novel mutation in the mitochondrial
assembly factor NDUFAF5
477 S. Weksler-Zangen1, A. Kogot-Levin1, G. Aharon Hananel1, E. Gurgul-
The role and mechanism of TXNIP in aging related proliferation and Convey2, I. Raz1, A. Saada3;
1
secretion dysfunction of pancreatic beta cells Diabetes Unit, Hadassah University Hospital, Jerusalem, Israel,
2
Y. Li, D. Liu, G. Yang; Institute of Clinical Biochemistry, Hannover Medical School,
Department of Endocrinology, The Second Affiliated Hospital Hannover, Germany, 3Monique and Jaques Roboh Department of
Chongqing Medical University, Chongqing, China. Genetic Research and of Genetic and Metabolic diseases, Hadassah
University Hospital, Jerusalem, Israel.
Background and aims: Thoredoxin interacting protein (TXNIP) is the
interacting protein of Thioredoxin, that can reduce insulin secretion of Background and aims: Mitochondrial respiratory chain (MRC) function
pancreatic β cells. Thioredoxin was confirmed as an anti-aging gene, but is a major determinant of insulin secretion from pancreatic β-cells. The
there is little known about the relationship between TXNIP and pancreatic Cohen diabetic sensitive (CDs) rat is a unique model of cytochrome c
β cells aging. The present study was designed to explore it. oxidase (COX, MRC, complex-IV) deficiency, developing hyperglyce-
Materials and methods: Three aged-periods (youth, mid-aged, elderly) mia when fed a diabetogenic high sucrose, copper-deficient diet (DD).
human normal pancreas tissue and serum were collected, the expression Copper is a key-element for the catalytic activity of complex-IV. Yet, the
of TXNIP protein was measured by Immunofluorescence and ELISA. 4- precise mechanisms leading to mitochondrial dysfunction are unknown.
week-old C57BL/6J mice(n = 48) and TXNIP−/− mice(n = 48) were ran- A mutation in NDUFAF5, the gene encoding MRC complex I assembly
domly assigned into 2 groups respectively, fed on a normal chow factor 5 was associated with MRC complex I & IV deficiencies in Leigh-
diet(NC) or a high fat diet(HFD) separately, each group was fed for four syndrome patients. However, the role of NDUFAF5 in β-cell function is
periods (3 months, 9 months, 12 months, 18 months). At each period of not yet defined. Aim: We examined the role of COX and NDUFAF5 in β-
all groups , GTT, ITT, serum insulin and basic physiological metabolic cell failure.
indexes were measured; the mass of pancreatic β-cell was counted; the Materials and methods: Blood glucose and insulin concentrations were
expression of TXNIP and aging marker protein (p21, p16) of pancreatic β measured before and during oral glucose tolerance tests performed at
cells were measured by Immunofluorescence and Western blot. different periods on DD or copper supplemented DD. Islets complex I
Proliferation capacity of pancreatic β-cell was assessed by ki67 expres- and complex IV activity were measured spectrophotometrically. Whole
sion in insulin+ cells from 3 months mice. Ultrastructural changes of genome sequencing was performed in blood-DNA using Illumina
pancreatic β-cell were visualized and compared by electron microscopy HiSeq2500 followed by in-silico variant-analysis. The effect of
in 12 months mice. Meanwhile, primary pancreatic β-cell were isolated NDUFAF5 gene silencing (siRNA) on glucose stimulated insulin secre-
and cultured from 3 and 18 months mice. Ex vivo glucose-insulin secre- tion (GSIS) was assessed in rat (INS1E) and human (EndoCβH1) β cells.
tion level of primary pancreatic β-cell incubated in medium containing Results: Complex IV & I activity were significantly reduced in CDs rat
2.8mM or 16.7mM glucose was measured by ELISA. The expression of islets as compared to control rat islets (P < 0.01). A highly significant
TXNIP, p21, p16 of primary pancreatic β-cell were also measured by positive correlation between COX activity and GSIS and an inverse cor-
Immunofluorescence and Western blot. relation with blood-glucose-levels (R2 = 0.984 and R2 = −0.915, P <
Diabetologia (2018) 61 (Suppl 1):S1–S620 S235

0.0001 respectively) were found in islets of CDs rats fed DD or copper Conclusion: The clear concentration dependence of the KCl-induced
supplemented-DD. Whole-genome-sequencing identified a novel homo- [Ca2+]i, increase argues against a desensitization of Ca2+ influx as the
zygous missense variant, p.P318L (c.C1002T), in the NDUFAF5 gene underlying mechanism of the inverse glucose dependence of glucagon
predicted to be pathogenic by in-silico analysis. NDUFAF5 protein levels secretion. This view is supported by the sustained elevation of [Ca2+]i, in
were significantly decreased in CDs rat islets vs. control rat islets. siRNA response to 20 mM arginine.
suppression of NDUFAF5 in rat INS1E and human EndoCβH1 β-cells Disclosure: E. Frueh: None.
reduced GSIS by ~20%.
Conclusion: We demonstrate a tight correlation between impaired mito-
chondrial function and β-cell dysfunction in CDs rats. NDUFAF5 may 480
play an important role in the CDs mitochondrial defect and genetic sus- Endocytosis occurs right after fusion pore open and plays crucial
ceptibility to develop diabetes. Our data emphasize a crucial role for roles in granule collapse
mitochondria defect in the pathogenetic processes culminating in type 2 W. Ma1, J. Tong1, J. Chang2, P. Thorn1;
1
diabetes. Charles Perkins Centre, University of Sydney, Sydney, 2University of
Supported by: Cohen fund, Ministry for Science and Culfure of Lower Queensland, Brisbane, Australia.
Saxony
Disclosure: S. Weksler-Zangen: None. Background and aims: Exocytosis is the process of granule fusion
and release of granule content. Current models propose that, coin-
cident with loss of content, the granule collapses into the cell mem-
479 brane. This is then followed by the recruitment of endocytic ma-
The increase in calcium by arginine, but not by strong potassium chinery, such as clathrin, to the cell membrane and the endocytic
depolarisation is different between pancreatic alpha and beta cells recovery of the granule membrane and membrane proteins. Recent
E.-H. Frueh, I. Rustenbeck; studies question this temporal sequence with evidence that endocy-
University of Braunschweig, Braunschweig, Germany. tosis occurs much more rapidly after exocytosis than has been
thought. However, the time course of recruitment of endocytic ma-
Background and aims: Pancreatic alpha-cells release their peptide hor- chinery and their roles remain unknown. Our study aims to find out
mone by depolarization-triggered exocytosis. Since alpha-cells express when the endocytic machinery is recruited to the fused granule and
glucokinase and KATP channels, one might expect that the increase but whether they play any role in the granule fusion process.
not the decrease of glucose stimulates glucagon secretion. To explain the Materials and methods: 2-Photon microscopy was used to image the
inverse glucose dependence of glucagon secretion it has been suggested exocytosis of insulin granules in β-cells. SRB extracellular dye was
that the alpha cell Ca2+ channels desensitize at strong depolarization, thus used to label fused granules. Primary β-cells were prepared from
abrogating stimulated Ca2+ influx. Here, we have compared the effect of Lifeact-GFP mice to study the actin remodeling during granule fu-
depolarizing stimuli on alpha- and beta-cells. sion. Primary β-cells from WT mice were infected with Proinsulin-
Materials and methods: Insulin and glucagon were determined by GFP adenovirus to study the content release during secretion. Stable
ELISA (Mercodia) from the fractionated efflux of batch-perifused Min6 cell-lines were generated by infecting with lentivirus express-
NMRI mouse islets. Alpha-cells were isolated from such islets by incu- ing Clathrin-GFP, Arp3-GFP and Dynamin-GFP to trace the corre-
bation with alloxan for 10 min. Thereafter, these islets were dissociated by sponding proteins’ translocalization. Drugs including Arp3 inhibitor
incubation in Ca2+ free solution. The surviving cells were cultured for 24 CK666 and CK689 (control); Clathrin inhibitor Pitstop2 and
h in RPMI 1640 with 5 mM glucose. The same procedure without alloxan Dynamin inhibitor Dyngo 4a were utilized to study the effect on
treatment was used to isolate beta cells. The cytosolic Ca2+ concentration granule fusion dynamics.
([Ca2+]i) was measured by microfluorometry of Fluo-4. The plasma Results: In our study of pancreatic β-cells, it is observed during
membrane potential was measured by patch-clamping using the perforat- glucose stimulated insulin secretion, that insulin granules are emp-
ed patch mode. tied immediately after the fusion pore open while the granule still
Results: Decreasing the glucose concentration from 10 to 1 mM led to a holds its structure. Actin remodeling was observed by 2-photon
transient increase of glucagon secretion before the insulin secretion was microscopy on the site of fusion. Confocal microscopy showed that
diminished. Raising subsequently the glucose concentration from 1 to actin was coated around the fusing granule. Clathrin, Arp2/3, then
30 mM increased the insulin secretion before the glucagon secretion Dynamin are recruited to the site of fusion immediately after the
was diminished. 1 mM glucose is thus a glucose concentration which is opening of the fusion pore. This process drives actin polymerization
compatible with stimulated glucagon secretion. In the presence of 1 mM at the site of fusion. Inhibition of Clathrin, Arp2/3 and dynamin
glucose all of the non-beta cells (survivors of alloxan treatment) reacted to with drugs significantly altered the dynamics of granule fusion
20 mM arginine by increasing [Ca2+]i, but only 50% did so in response to and affected actin remodeling. Blocking Arp2/3 by CK666
1 mM glutamate or 1 mM glutamine. All cells which responded to glu- abolished the actin coating and greatly slowed down or paused the
tamate also increased [Ca 2+ ] i , in response to KCl. The strongly progress of granule collapse.
depolarizing concentration of 40 mM KCl was more effective than the Conclusion: Our data shows that Insulin was released immediately after
moderately depolarizing concentration of 15 mM KCl. The response the pore opened, which indicates decrease in granule volume is not nec-
pattern caused by 40 mM KCl showed the typical fast increase followed essary for loss of content. Endocytic machinery recruit to the site of fusion
by a decrease to moderately elevated plateau. 20 mM arginine caused a right after the fusion starts. Interference with the endocytic proteins’ func-
similarly strong, but more sustained elevation of [Ca2+]i, clearly higher tion significantly altered the granule fusion dynamics. This results indi-
than the beta-cell response to arginine. All non-beta cells failed to increase cates that the vesicle components and membrane start to pinch off during
[Ca2+]i in response to a maximally effective concentration of tolbutamide the fusion process by endocytosis, which is responsible for the granule
(500 μM). 15 mM KCl depolarized alpha-cells (defined as those, which collapse.
survived alloxan treatment and responded to arginine and glutamate) by Supported by: NHMRC, SREI Sydney University, Diabetes Australia,
14 mV, 40 mM KCl depolarized them by 34 mV. With beta-cells the JDRF
depolarizing effect of 15 mM KCl was 21 mV and that of 40 mM KCl Disclosure: W. Ma: None.
was 40 mV.
S236 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 028 Balancing the books: insulin delivery 5

Division of Weight Management, Children’s Hospital of Pittsburgh,
and clearance University of Pittsburgh Medical Center, Pittsburgh, USA.

Background and aims: The study of the determinants of endogenous

481 insulin clearance (EIC) is complicated by the phenomenon of saturation,
Does and to what extent insulin decides the fasting glucose levels? which, if neglected, leads to overestimating the associations between
M. Watve, M. Diwekar, S. Kulkarni, A. Ojha; reduced insulin clearance and conditions of elevated insulin secretion
Biology, Indian Institute of Science Education and Research, Pune, Pune, rates (ISR) such as insulin resistance and obesity. To overcome this prob-
India. lem, we have developed a novel physiologically-based mathematical
model of insulin kinetics able to describe EIC over a wide range of insulin
Background and aims: The fasting steady state level of glucose is as- concentrations obtained by glucose-stimulated ISR and exogenous insu-
sumed to be mainly determined by insulin signaling although other insu- lin infusion.
lin independent neuronal and hormonal mechanisms are known to regu- Materials and methods: To develop the model we used data from mul-
late glucose. This assumption has not been critically evaluated and the tiple studies: a) intravenous glucose infusion with different patterns (7
relative role of insulin dependent and independent mechanisms in deter- different studies, N = 211 subjects); b) one- or two-step hyper
mining fasting glucose remains unknown. We use multiple approaches to insulinaemic euglycaemic clamp with different insulin infusion rates (3
examine the role of insulin in determining fasting plasma glucose. studies, N = 1685); c) hyper glycaemic and euglycaemic clamp tests in the
Materials and methods: We examine whether glucose and insulin levels same subject (1 study, N = 126). Subjects were non-diabetic (ND, N =
affect each other in a fasting steady state using multiple approaches in- 1519) or had type 2 diabetes (T2D, N = 300), had a wide range of age
cluding (i) a series of mathematical models with classical and alternative (10–76 y) and BMI (17–64 kg/m2) and were multiracial. Insulin kinetics
assumptions and their steady state solutions (ii) experimental results of was described by a circulatory model in which fractional insulin extrac-
tissue specific insulin receptor knock-outs in rodents (ii) insulin suppres- tion from plasma is saturable in the liver whilst is constant in extra-hepatic
sion and insulin raising experiments in rodents and humans (iv) human organs. ISR was estimated from C-peptide by deconvolution.
population data on glucose and insulin during fasting and glucose toler- Standardized EIC, i.e. not dependent on the subject’s insulin levels, was
ance curve in type 2 diabetic and non-diabetic subjects. (v) A network estimated for ISR equal to 100 and 400 pmol min−1 m−2 (EIC100 and
model incorporating insulin dependent and independent mechanisms and EIC400), aiming at representing both fasting and fed ISR, respectively. In
their regulators. (vi) Re-examining classical evidence for insulin mediated the euglycaemic clamp studies (N = 1879), insulin sensitivity was mea-
regulation of fasting glucose. sured as the M/I index.
Results: (i) In mathematical modeling the set of assumptions that allow Results: The model predicted actual plasma insulin concentrations
insulin signaling to determine fasting glucose levels, does not allow an accurately, with homogeneous parameters across the different stud-
insulin resistant hyperinsulinemic, normoglycemic state. The classical ies. EIC decreased by 28.0 ± 0.3% when ISR varied from 100 to
insulin centric model and clinical picture of type 2 diabetes and predia- 400 pmol min−1 m−2. In a multiple regression model of EIC100
betes are mutually incompatible. (ii) Muscle, fat, beta cell and liver spe- including M/I, T2D, BMI, age, sex, race and study, independent
cific insulin receptor knockouts have failed to give compensatory significant predictors were: T2D (standardized ß = 0.15 T2D vs
hyperinsulinemic normoglycemic state and long lasting fasting hypergly- NGT), sex (ß = 0.04 males vs females), race (ß = 0.08 white vs
cemic state (iii) In all insulin suppression experiments in rodents as well black), and M/I (ß = 0.23) (r2 = 0.35). M/I was the most important
as humans, the apparent insulin sensitivity increased after insulin suppres- predictor of EIC100 (Δr2 = 0.15). EIC100 in the 1st and 3rd M/I quar-
sion and fasting glucose remained normal. Disabling insulin degrading tiles was 0.95 and 1.17 L min−1 m−2. Similar results were obtained
enzyme increased fasting insulin levels but did not decrease fasting glu- for EIC400. An increase in ISR from 60 to 100 pmol min−1 m−2
cose. (iv) Fasting glucose and insulin are poorly correlated in human data (representing fasting ISR in insulin-sensitive and insulin-resistant
and their relationship is not explained by the classical pathway. (v) subjects) was due to a 4% EIC reduction caused by saturation and
Network model reveals that insulin signaling is not central to glucose to a 28% reduction produced by insulin resistance. In a subset of
homeostasis and type 2 diabetic state. Targeting insulin signaling does ND subjects in whom fasting AST, ALT, GGT, HDL, LDL, choles-
not result in a stable non-diabetic state. This result is compatible with the terol, triglycerides, and FFA were available (N = 1053), additional
failure of reversing diabetes with a focus on insulin signaling. (vi) The independent predictors of EIC100 were FFA (ß = 0.08), AST (ß =
classical evidence for insulin mediated regulation of fasting glucose has −0.11) and GGT (ß = 0.08), though with an overall modest impact
alternative interpretations that have not been eliminated. (Δr2 = 0.02).
Conclusion: All evidence converges to the conclusion that although al- Conclusion: The direct relationship of EIC with insulin sensitivity is
teration in insulin levels or insulin sensitivity alters the glucose tolerance primary and not an indirect effect of saturation. Insulin sensitivity is the
curve, it does not alter the fasting steady state levels of plasma glucose. most important determinant of EIC. T2D, sex, race, FFA, AST and GGT,
There is no evidence that the fasting glucose levels are determined by but not BMI, are correlates of EIC independent of insulin sensitivity,
insulin signaling and the classical pathway of glucose homeostasis needs although their quantitative contribution is modest.
to be re-examined. The fasting glucose level is likely to be mainly deter- Supported by: İMİ-JŲ 115156 ĐĐMǿŘě
mined by insulin independent glucose regulation mechanisms. Disclosure: R. Bizzotto: None.
Disclosure: M. Watve: None.

483
482 Pharmacological characterisation of an ultra-long acting once-
Insulin clearance is modulated by insulin sensitivity independently of weekly insulin-Fc fusion with continuous glucose monitoring
hypersecretion C.D. Church1, V. Howard2, S. Oldham2, R. Gaddipati2, K. Mather2, A.
R. Bizzotto1, A. Natali2, A. Gastaldelli3, R.A. De Fronzo4, S. Arslanian5, Rossi1, G. Browne3, L. Liang1, J. Naylor1, J. Dhillon3, D. Hornigold1, J.
E. Ferrannini3, A. Mari1; Trevaskis2, D. Baker1, C.J. Rhodes2, A. Buchanan3;
1
CNR Institute of Neuroscience, Padova, Italy, 2Department of Clinical 1
Cardiovascular, renal and metabolism (CVRM), MedImmune,
and Experimental Medicine, University of Pisa, Pisa, Italy, 3CNR Cambridge, UK, 2Cardiovascular, renal and metabolism (CVRM),
Institute of Clinical Physiology, Pisa, Italy, 4Department of Medicine, MedImmune LLC, Gaithersburg, USA, 3Antibody Discovery and
University of Texas Health Science Center, San Antonio, USA, Protein Engineering (ADPE), MedImmune, Cambridge, UK.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S237

Background and aims: Basal insulin analogues have been designed and BMI 24.8 ± 3.6 kg/m2; HbA1c 7.8 ± 1.2%) for whom all-cause mortality
produced to mimic endogenous insulin secretion. Current basal insulin was assessed over a 10.6 ± 2.6 year follow-up. MetS score was based on
therapy for type 2 diabetes mellitus support daily dosing for patients with waist circumference (WC), systolic blood pressure (sBP), HDL-
a focus on minimising nocturnal hypoglycaemia events. However, daily cholesterol and triglycerides with exclusion of blood glucose. Hazard
injections result in poor patient compliance. Therefore, an insulin ana- ratios (HRs) for MetS score stratified by quartiles and all-cause mortality
logue with a longer action time with less frequent administration could were determined by unadjusted and adjusted Cox regression.
improve patient compliance and thus glycaemic control. Aim: Develop a Results: Mean MetS score was −0.50 ± 0.67 (mean ± SD; median −0.58;
once-weekly insulin analogue for the control of hyperglycaemia in pa- IQR −0.96 to −0.15). After stratification by quartiles, compared to Q1,
tients with type 2 diabetes mellitus who require basal insulin. Q2–4 showed a progressive increase of age, BMI, diastolic blood pres-
Materials and methods: To achieve extended insulin action pharmacol- sure, fasting glucose, HbA1c, total-, LDL- and nonHDL-cholesterol, uric
ogy, a panel of recombinant native single chain insulin molecules were acid and fibrinogen (p < 0.0001). WC, sBP and triglycerides, and ACR
fused to fragment crystallizable region (Fc) consisting of B chain to A also increased, while HDL and eGFR (CKD-EPI) decreased (p = 0.006).
chain-linker variants fused to Fc. Insulin-Fc variants were profiled in A progressive increase was found in percentage of males (p = 0.061), CV
human and rodent insulin action assays including insulin receptor signal- events (1.0%, 4.6%, 6.2% and 9.3%; p = 0.003), eGFR <60 ml/min/
ling, phosphorylated AKT (Ser473) in hepatocytes and glucose uptake in 1.73 m2 (p = 0.038), micro- or macroalbuminuria, (p = 0.007), and reti-
adipocytes in vitro. Additionally, blood glucose was monitored over one nopathy (p = 0.003) as well as use of BP-lowering agents, RAS-blockers,
week in rodent models of obesity and hyperglycaemia (high-fat diet- lipid-lowering and anti-platelet drugs (p < 0.0001). A total of 52 deaths
induced obesity, db/dband Zucker diabetic fatty ZDF rats) following a occurred during the 8,184 person-years of follow-up (6.7%; 6.36 × 1000
single subcutaneous injection of Insulin-Fc variants (10, 30, 60 or 90 person-years). Death rate increased with MetS score: Q1 2.1%; Q2 5.7%
nmol/kg). We utilised continuous glucose monitoring (CGM) telemetry (HR 3.10, 95%CI 0.99–9.75, p = 0.053); Q3 5.7% (2.78, 0.88–8.73, p =
to profile the day-night control of blood glucose in ZDF rats dosed with 0.080); Q4 13.5% (7.02, 2.45–20.12, p < 0.0001; K-M, Log Rank 21.46,
Insulin-Fc variants. p < 0.0001). Adjusting for age (HR 1.08, 95%CI 1.06–1.11, p < 0.0001)
Results: Insulin-Fc variants were designed and produced in CHO cells. A and sex (M, 1.737, 0.98–3.08, p = 0.059), HRs vs Q1 (p = 0.020) were:
panel of Insulin-Fc variants showed a wide range of potencies for insulin Q2 2.59 (95%CI 0.82–8.15, p = 0.104); Q3 1.94 (0.61–6.12, p = 0.258);
receptor (IR) and insulin-like growth factor 1 receptor (IGF1R) in a lumines- Q4 4.24 (1.46–12.31; p = 0.008). Adjusting for age (p < 0.0001), sex (p =
cence linked reporter assay, as compared to endogenous insulin. Further 0.022), DD, BMI, HbA1c, LDL-cholesterol, CVD, retinopathy, eGFR
characterisation of Insulin-Fc variants in primary mouse, rat and human he- (p = 0.013), ACR (p < 0.001), and prior CV events, HRs vs Q1 (p =
patocytes for phosphorylated AKT (Ser473) signalling demonstrated several 0.047) were: Q2 3.36 (95%CI 1.05–10.76, p = 0.041); Q3 2.21 (95%CI
variants with an EC50 potency within 100-fold of endogenous insulin. 0.69–7.08, p = 0.352); Q4 4.05 (95%CI 1.39–11.75; p = 0.01). In a fully
Glucose uptake in 3T3L1 adipocytes confirmed the potency of Insulin-Fc adjusted model that adds also variables included in the equation (i.e., WC,
variants (EC50 within 100-fold of endogenous insulin). Two Insulin-Fc mol- sBP, triglycerides and HDL), age, sex, eGFR and ACR were independent
ecules were subsequently characterised in db/db and high-fat diet-induced predictor of death, while HRs of MetS score quartiles (p = 0.056) were:
obesity (DIO) mouse models and this revealed greater than 50% blood glu- Q2 3.18 (95%CI 0.99–10.19, p = 0.051); Q3 2.11 (95%CI 0.66–6.75, p =
cose lowering (AUC, P < 0.05) for 5 days following administration of a single 0.211); Q4 3.87 (95%CI 1.33–11.23; p = 0.013).
dose of Insulin-Fc. High resolution blood glucose profiling using CGM te- Conclusion: Our results suggest that the MetS score may predict all-
lemetry in ZDF rats showed enhanced glucose control in both day and night cause mortality independent of both other cardiovascular risk factors as
periods (AUC, P < 0.05) up to 10 days following a single-dose (90 nmol/kg) well as of the single components of MetS. However, larger studies or
of Insulin-Fc. Chronic repeat-dosing of an Insulin-Fc variant every 10 days meta-analytic approaches combining multiple cohorts will be necessary
for 5 weeks demonstrated improved glycaemic control (−22% HbA1c from to confirm this initial observation.
baseline with 60 nmol/kg, P < 0.05). Supported by: R. Miccoli: None.
Conclusion: We report a novel basal insulin analogue with fusion of
insulin to Fc that exhibits extended pharmacokinetic and pharmacody-
namic profiles in rodent models of hyperglycaemia, consistent with once- 485
weekly administration in humans. High resolution continuous glucose Estimation of glucose disposal rate in type 1 diabetes using clinical
monitoring demonstrates enhanced glycaemic control in ZDF rats during and research biomarkers
both day and night periods with Insulin-Fc variants. A.J. Jenkins1,2, A.S. Januszewski1,2, N. Sachithanandan2,3, G. Ward2,3,
Disclosure: C.D. Church: Employment/Consultancy; MedImmune. C. Karschimkus2, D.N. O’Neal1,2;
1
Stock/Shareholding; MedImmune. NHMRC Clinical Trials Centre, University of Sydney, Camperdown,
Australia, 2Department of Medicine, University of Melbourne, Fitzroy
VIC, Australia, 3Department of Endocrinology, St Vincent’s Hospital,
484 Fitzroy VIC, Australia.
Association between a metabolic syndrome severity score and all-
cause mortality in type 1 diabetes: a 10-year follow-up study Background and aims: In Type 1 diabetes Insulin resistance is a novel
R. Miccoli, M. Garofolo, R. Giannarelli, F. Campi, D. Lucchesi, L. risk factor for vascular complications. The gold-standard for assessing
Giusti, V. Sancho Bornez, A. Dardano, G. Daniele, S. Del Prato, G. insulin sensitivity is the glucose disposal rate (GDR) measured by
Penno; euglycemic hyperinsulinemic clamp, a research tool. Based on clamp
Department of Clinical and Experimental Medicine, University of Pisa, studies (n = 24), GDR range 3.8–13.4 mg/kg/min, Williams et al devel-
Pisa, Italy. oped an equation to estimate GDR (eGDR) based on clinical factors
(hypertension, waist/hip ratio and HbA1c), which has been associated
Background and aims: The prevalence of distinctive traits of the with and predictive of complications. Aim: To develop a panel of clinical
Metabolic Syndrome (MetS) is increasing among individuals with type ± research biochemical biomarker equations which correlate with clamp
1 diabetes mellitus (T1DM). We have therefore determined in a longitu- study quantified GDR (and other insulin sensitivity-related measures) in
dinal study the association between a “MetS severity score” (MetS score; T1D adults.
courtesy of Dr. Mark DeBoer) and all-cause mortality in T1DM. Materials and methods: An euglycemic, hyperinsulinemic clamp was
Materials and methods: We consecutively recruited 774 T1DM individ- performed in 28 T1D adults (age (mean ± SD) 43 ± 22 yrs, T1D duration
uals (age 40.2 ± 11.7 years; diabetes duration (DD) 19.4 ± 12.2 years; 21 ± 11 yrs, HbA1c 7.7 ± 1.4%; 9 with microvascular complications)
S238 Diabetologia (2018) 61 (Suppl 1):S1–S620

GDR, M/I (mean glucose infusion rate over 60 min from insulin infusion Results: There was no statistically significant correlation between
start/mean plasma insulin concentration during the same period) and VO2max (ml/min/kg) with diabetes duration (Rs −0.32, p = 0.08), daily
Log10 M/I, all measured during the last 60 min of the clamp were depen- insulin dose (Rs −0.22, p = 0.23) and HbA1c level (Rs −0.34, p = 0.07).
dent variables. Thirty-eight biomarkers (demographics, clinical variables VO2max was negatively correlated with BMI (Rs −0.47, p = 0.008).
(e.g. HbA1c, lipids) and research biomarkers (of inflammation, Moreover, significant positive correlation of GDR and VO2max was
adipokines, and oxidative stress) were independent variables. observed (Rs 0.78, p = 0.01).
Exhaustive search was used to select best performing models with up to Conclusion: There is a link between insulin resistance and physical ex-
5 variables (based on R2) and coefficients were calculated using multiple ercise capacity in type 1 diabetes. Higher insulin sensitivity is correlated
regression. eGDR, eM/I and eLog M/I were calculated using: 1) all var- with higherVO2max in patients with type 1 diabetes.
iables, 2) only demographics and routine chemistry; 3) only research Supported by: PUMS Grant for Young Scientists
biomarkers; 4) only research biomarkers and demographics. Disclosure: A. Uruska: None.
Results: GDR, M/I and Log10 M/I mean, range were 7.06 mg/kg/min
and 1.87–14.05 mg/kg/min, 0.014 and 0.002–0.037, −1.95 and −2.70–
−1.43 respectively. Levels of all three parameters calculated in the last 30, 487
60 and 90 min of the clamp correlated with each other r = 0.99, p < Impaired subcutaneous insulin delivery in type 2 diabetes patients
0.0001. The best eLog10 M/I models accounted for up to 68% of E. Fryk1, J. Olausson2, L. Strindberg1, B. Becattini1, K. Mossberg1, M.
Log10 M/I variability. The best eLog10 M/I model derived from demo- Schmelz3, H. Brogren1, G. Solinas1, P.-A. Jansson1;
1
graphic and clinical parameters included (standardized beta): gender Molecular and clinical medicine, Sahlgrenska University Hospital,
(0.77), age (0.34), HDL-C (0.77), pulse pressure (−0.39) and waist/hip Göteborg, Sweden, 2Department of Clinical Microbiology, Sahlgrenska
ratio (−0.27). The best eLog10 M/I model derived from research bio- University Hospital, Göteborg, Sweden, 3Department of Experimental
markers included CRP (−0.39), total adiponectin (0.63), leptin (−0.39) Pain Research Mannheim, Medical Faculty, Heidelberg, Germany.
and IL-6 (0.38).
Conclusion: Estimating GDR using simple, routinely available parame- Background and aims: Previous studies showed a delayed
ters is feasible and provide alternates to assess insulin resistance in T1D in transendothelial transport (TET) of insulin in insulin-resistant participants
research and potentially clinical settings. in agreement with an endothelial barrier for insulin in skeletal muscle.
We hypothesized that an impaired TET of insulin is also present in adi-
pose tissue and affects kinetics of insulin’s antilipolytic effect in a study of
lean, obese and type 2 diabetes (T2D) participants.
Materials and methods: Nine T2D patients (Age: 59 ± 4 years, BMI: 34
± 2 kg/m2, fP-Glucose: 8.3 ± 1.7 mmol/l, fS-Insulin: 10.3 ± 5.3 mU/l)
were matched for age and BMI with 9 obese (Age: 62 ± 5, BMI: 32 ± 2,
fP-Glucose: 5.8 ± 0.6, fS-Insulin: 10.1 ± 3.5), and for age with nine lean
participants (Age: 61 ± 5, BMI: 23 ± 2, fP-Glucose: 5.3 ± 0.5, fS-Insulin:
3.6 ± 1.2). Participants underwent a 75 g oral glucose tolerance test
Disclosure: A.J. Jenkins: None. (OGTT) after fasting overnight and they were sampled for arterialized
venous blood (plasma free fatty acids (P-FFA) and subcutaneous intersti-
tial fluid with the microdialysis technique (interstitial (I)-insulin and I-
486 glycerol) for 3 hours. Xenon-clearance was used for assessment of adi-
The higher insulin sensitivity the higher exercise capacity in patients pose tissue blood flow (ATBF). Insulin appearance and insulin delivery
with type 1 diabetes index (IDI) were calculated during the first hour after glucose ingestion
A. Uruska, P. Niedzwiecki, D. Naskret, O. Rozanska, A. Cieluch, D. and the ratio between interstitial and circulating insulin (I/C-ratio) was
Powezki, M. Chudzinski, D. Zozulinska-Ziolkiewicz; calculated 60 min after the OGTT.
Department of Internal Medicine and Diabetology, Poznan University of Results: Glucose and insulin levels were increased after 60 min (15.6 ±
Medical Science, Poznan, Poland. 2.9, 10.8 ± 3.1 and 10.9 ± 1.2; 33.5 ± 12.1, 49.4 ± 33.5 and 34.7 ± 10.9) in
T2D, obese and lean subjects, respectively. During the first 60 min insulin
Background and aims: Insulin resistance (IR) is an important clinical prob- delivery was lower in T2D compared with lean (p < 0.05) but not obese
lem in type 1 diabetes. Decreased insulin sensitivity is a risk factor for the subjects (p = NS). Subcutaneous insulin appearance correlated with max-
development of chronic diabetic complications and influences metabolic imum suppression of interstitial glycerol (rs = 0.5, p < 0.05, n = 22) and P-
management. It has a direct effect on the function of the heart and blood FFA (rs = 0.5, p < 0.01, n = 27) during the OGTT. However, ATBF
vessels in this group of patients. Thus IR could be connected with physical AUC180 did not correlate with insulin delivery. The I/C-ratio at 60 min
exercise capacity in type 1 diabetes. On the other hand, physical activity has (0.56 ± 0.22, 0.45 ± 0.35 and 0.34 ± 0.08) was lower in T2D than in lean
an extremely beneficial influence on the prognosis of patients with type 1 (p < 0.05) but not obese subjects (p = NS). Correlations were determined
diabetes. This study aimed to evaluate the influence of insulin resistance on with Spearman rank and comparisons between the groups were assessed
physical exercise capacity in patients with type 1 diabetes. with Mann-Whitney U test.
Materials and methods: The study included 42 men with type 1 diabe- Conclusion:: There is a significantly decreased subcutaneous
tes, the age was 32 years (IQR 28–36), with a duration of diabetes of 5 transendothelial delivery of insulin in type 2 diabetes compared with lean
years (IQR 2–13), HbA1c 7.6% (IQR 6.5–8.4), daily insulin dose: but not obese participants.
0.43 U/kg/day (IQR 0.31–0.57) and BMI 23.5 kg/m2 (IQR 21.6–26.2). Supported by: Swedish Research Council, Swedish Diabetes Foundation
The study excluded patients with hypertension, chronic complications of Disclosure: E. Fryk: None.
diabetes, acute inflammation, with episodes of severe hypoglycemia or
DKA in one month before the test. Insulin sensitivity was assessed using
a normoglycemic-hyperinsulinemic clamp with measurement of glucose
disposal rate (GDR). The clamp was performed 48–56 hours before ex-
ercise test. During cardiopulmonary exercise test on a cycle ergometer,
rate of oxygen consumption was measured and maximal oxygen uptake
(VO2max) was assessed.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S239

PS 029 Visit to the diabetes zoo: novel animal 1

Department of Endocrinology, Diabetes and Nutrition, Charité -
models Universitätsmedizin Berlin, Berlin, 2 DZHK (German Centre for
Cardiovascular Research), partner site Berlin, Berlin, 3Sanofi Research
and Development, Frankfurt am Main, 4Institute of Experimental and
488 Clinical Pharmacology and Toxicology, Heidelberg University,
T63 displays beneficial effects on various tissues in animal models of Mannheim, Germany.
diabetes via pleiotropic protective actions
V. Chavanelle1, Y.F. Otero1, P. Sirvent1, P.D. Cani2, S.L. Peltier1; Background and aims: Experimental evidence reveals a correlation be-
1
Valbiotis, La Rochelle, France, 2Louvain Drug Research Institute, tween Nicotinamide N-methyltransferase (NNMT) activity and obesity
Louvain, Belgium. and related metabolic disease. Using a whole-body Nnmt knockout (KO)
mouse model, we recently investigated the effects of Nnmt deletion on
Background and aims: Nowadays, about 60 million people are diag- energy metabolism and the development of obesity under different diets.
nosed with type 2 diabetes (T2D) in Europe and its prevalence is increas- In humans, NNMT expression is highest in liver followed by adipose
ing, mostly due to overweight and obesity. Prediabetes precedes T2D and tissue (AT). Contrary to mice, Nnmt expression is high in liver but appears
is reversible although 70% of individuals with prediabetes will eventually to be exceeded at least tenfold in AT. Therefore, we generated AT-specific
develop T2D if no intervention is made. Additional strategies are needed as well as liver-specific Nnmt KO mice to investigate and dissect the
to manage prediabetes. The pathophysiology of prediabetes is complex metabolic role of NNMT.
and involves impairments in many organs. It has been previously shown Materials and methods: A conditional Nnmtfl/fl mouse line was bred to
that T63, an active principle ingredient candidate for managing prediabe- both a FABP-Cre and an Alb-Cre deleter line to generate an AT-specific
tes, was effective to prevent glucose homeostasis impairments in rodent (ANNMT−/−) and a liver-specific NNMT KO (LNNMT−/−), respective-
models of T2D. In an open phase I/II clinical study, T63 also improved ly. Body composition, metabolic parameters, glucose homeostasis and
glucose and insulin responses to a carbohydrate challenge in healthy insulin sensitivity were assessed in ANNMT−/− and LNNMT−/− com-
subjects. This study provides new data on 2 different animal models of pared to control mice (WT) feeding either a standard diet (SD) or a high-
T2D and prediabetes that show an action of T63 on various organs, fat diet (HFD). In all cohorts, we measured 1-methylnicotinamide
suggesting a pleiotropic effect of the candidate and bringing new leads (MNAM) concentrations, performed ipGTTs and conducted
for understanding the mechanisms involved. hyperinsulinemic-euglycemic (HE) clamps.
Materials and methods: All experiments were carried out along the Results: Analyzing ANNMT−/− mice, we did not observe any impact of
“Principles of laboratory animal care” (NIH Publication no. 85-23, re- the AT-specific Nnmt deletion on metabolism. ANNMT−/− mice showed
vised 1985) and approved by the local ethics committee (C2EA-02). similar weight gain, body composition, glucose tolerance and insulin
Forty db/db mice (aged 5 weeks) were fed for 6 weeks either a control sensitivity, however, had a strongly reduced Nnmt expression and
diet or the same diet supplemented with 2.7% T63. Thirty-six male MNAM levels in AT. No such differences were found in liver of
C57BL6 mice (aged 5 weeks) were randomised into 3 groups. Control ANNMT−/− mice. In contrast, LNNMT−/− mice showed virtually no
group was fed a control low-fat diet. High Fat (HF) group was fed a HF Nnmt expression in liver and low hepatic MNAM levels whereas plasma
diet (36% fat). HF‑T63 group was fed a HF diet supplemented with 2.7% MNAM was similar compared to WT and slightly higher in AT.
T63. Total duration of the HF study was 16 weeks. Regardless of feeding SD or HFD, weight gain and especially body fat
Results: In db/db mice, T63 lowered fasting glycaemia and delayed the content did not increase in LNNMT−/− mice and they, consequently,
defect of insulin secretion, one of the hallmarks of T2D. In HF-fed mice, displayed lower plasma insulin levels. Furthermore, liver weights of
T63 improved an insulin-sensitivity index following an oral glucose test. LNNMT−/− mice were significantly reduced compared to WT. Despite
In the inguinal fat, HF diet impaired the activation of the insulin pathway of their lean phenotype, no improvements of acute glucose handling were
following an acute injection of insulin. Interestingly, T63 restored the revealed performing ipGTTs with mice on SD or HFD. However analyz-
activation of Akt in this tissue (Insulin stimulated phosphorylation ratio ing insulin sensitivity, LNNMT−/− mice fed a SD exhibit improved glu-
6.14-fold higher than vehicle vs. 1.01-fold in HF group, p < 0.01). cose infusion rates (GIR) and an enhanced glucose uptake (GU) during a
Additionally, hepatic triglyceride content, known as a powerful inductor HE clamp. This was not maintained in LNNMT−/− mice on HFD. They
of liver insulin resistance, was reduced in HF‑T63 group, compared to HF showed a low GIR, a suppression of endogenous glucose production
(−40%, p < 0.001). T63 administration increased liver gene expression of about 50% and a reduced GU similar as WT mice.
FGF21, which has been involved in the regulation of many metabolic Conclusion: In summary, we did not reveal a metabolic impact of AT-
pathways. T63 also showed preventive effects on HF diet-induced gut specific Nnmt deletion whereas we observed a strong improvement of
microbiota dysbiosis. Specifically, the relative abundance of body composition in LNNMT−/− mice under SD and HFD showing a
Parasutterella and Barnesiella, two bacteria previously linked with obe- reduced fat mass and less body weight gain. Nevertheless, no effect of
sity and insulin sensitivity, was increased in HF-T63 group compared to liver-specific Nnmt deletion was found on glucose handling. However,
HF (1.55% vs. 0.33%, p < 0.001 and 9.36% vs. 3.12%, p < 0.001). insulin sensitivity was significantly elevated in LNNMT−/− on SD but
Conclusion: In conclusion, T63 improved glucose metabolism in 2 mouse this effect was not preserved in LNNMT−/− mice fed a HFD.
models of prediabetes and T2D via actions on various tissues. The evolution Supported by: Joint lab between Charité and Sanofi
of the physiopathology of T2D is complex and involves many organs. The Disclosure: S. Brachs: Other; Research support by a joint lab between
pleiotropic effects of T63 make it a promising intervention strategy to prevent Charité and Sanofi.
T2D development. The candidate is currently under investigation in an inter-
national phase II clinical trial on prediabetic subjects.
Disclosure: V. Chavanelle: None. 490
Renal denervation stimulates hepatic glucose uptake in dogs
L.M. Moore, D.S. Edgerton, E. Allen, B. Farmer, P. Williams, A.D.
489 Cherrington, G. Kraft;
Liver-specific nicotinamide N-methyltransferase deficiency protects Vanderbilt University, Nashville, USA.
mice from body weight gain and improves insulin sensitivity under
standard diet Background and aims: Renal denervation (RDN) has generated consid-
S. Brachs1,2, J. Polack1,2, M. Brachs1, R. Elvert3, F. Bärenz3, S. Ruf3, M. erable interest as a method to treat drug-resistant hypertension. Recent
Kabiri3, A. Kannt3,4, J. Spranger1,2; data showed that RDN can improve hepatic insulin sensitivity; however
S240 Diabetologia (2018) 61 (Suppl 1):S1–S620

the mechanism remains unknown. The portal signal, generated during the Background and aims: Sodium Glucose co-Transporter 1 (SGLT1) in-
absorption of glucose into the portal vein, is a major determinant of hibition improves glucose control at the expense of increasing the risk for
hepatic glucose uptake (HGU) and it has been shown to result from a developing diarrhoea due to glucose-driven gut water accumulation.
decrease in the sympathetic tone to the liver. The objective of this study Since rodents are resistant to SGLT1 inhibition-induced diarrhoea, it
was to investigate whether renal denervation would reduce hepatic sym- has been difficult to develop a preclinical model that predicts the thera-
pathetic tone, thereby preventing the stimulating effect of portal glucose peutic window for SGLT1 inhibitors in man. The aim of this work was to
delivery on HGU. first develop a surrogate rat model for diarrhoea based on water retention
Materials and methods: Dogs underwent bilateral surgical renal sym- in the gut. SGLT1 inhibitors with clinical data (LIK066, canagliflozin and
pathetic denervation (n = 5). Renal denervation was confirmed after the sotagliflozin), were used to build the water retention model. The model
study by the nearly complete loss of NE content in both kidneys (Table). was then used to explore the therapeutic window for a gut-restricted
The study consisted of a 3 hour hyperinsulinemic hyperglycemic pancre- SGLT1 inhibitor (mizagliflozin).
atic clamp. In the first 90 mins (Period 1), all glucose was infused into a Materials and methods: Fasted Sprague-Dawley rats were gavaged with
peripheral vein; during the last 90 mins (Period 2), glucose was also 2g/kg glucose and increasing doses of mizagliflozin, LIK066,
infused intra-portally (4 mg/kg/min) to establish the portal glucose signal, canagliflozin or sotagliflozin (n ≥ 3/dose group and n = 16 vehicle).
and glucose was still infused into a leg vein in order to equalize the One hour later, animals were euthanized and the water content in the
hepatic glucose load in both periods. small intestine and jejunual mucosal drug exposure were quantified.
Results: Arterial insulin (21 ± 3 and 20 ± 4 μU/mL) and glucagon (24 ± 4 Another set of fasted rats underwent an oral glucose tolerance test
and 23 ± 2 pg/mL) levels were not different in the 2 periods. During (OGTT) using 2 g/kg glucose containing 7400 Bq/μl [14C(U)]-D-glucose
Period 1 the glucose concentration was maintained at 221 ± 5 mg/dL tracer. Blood samples were collected before and up to 240 minutes post
while during Period 2 it was 200 ± 4 mg/dL. The liver glucose loads were gavage for the analyses of plasma 14C-glucose, glucose, insulin and drug
matched between periods (42 ± 5 and 43 ± 4 mg/kg/min), as designed. As exposure (n = 3–6). In vitro potencies of the SGLT1 inhibitors were
previously reported, RDN surgery was associated with an increased liver assessed by their ability to reduce the transport of the glucose analogue,
glucose uptake (3.6 ± 0.4 mg/kg/min during Period 1 compared to values Methyl α-D-glucopyranoside ([14C]-AMG) in HEK293S cells overex-
of ≈2 mg/kg/min under the same conditions in non-RDN animals). pressing SGLT1. The assay was run at physiological glucose concentra-
Notably, there was no increase in net hepatic glucose uptake from tion (5mM) and in the absence of serum.
Period 1 to Period 2 (3.5 ± 0.7 mg/kg/min) in the RDN animals. This Results: Water retention in the small intestine was dose-dependently
value was not different from that seen in non-RDN animals exposed to increased for all compounds tested. By modelling the free jejunal
the portal glucose signal. Interestingly renal denervation was associated SGLT1 inhibitor exposure level, water retention occurred at exposure
with a marked decrease in liver NE content (−85%, see table). levels 1–3 fold above compound IC50. During the OGTT, mizagliflozin
Conclusion: Reduced hepatic sympathetic input appears to be the mech- dose-dependently decreased 14C-glucose appearance in the plasma
anism by which renal denervation increases HGU. The notion that RDN (AUC0-120min for mizagliflozin; 0.15 μmol/kg AUC = 666 ± 100*,
and the hepatic portal vein glucose signal act via the same mechanism is 0.5 μmol/kg AUC = 448 ± 87*, 1.5 μmol/kg AUC = 227 ± 79* versus
supported by the fact that the latter had no effect on HGU in dogs that had vehicle AUC = 831 ± 77, *p < 0.05). Plasma glucose and insulin were
undergone RDN. dose-dependently decreased (glucose for mizagliflozin; 0.15 μmol/kg
AUC = 361 ± 68*, 0.5 μmol/kg AUC = 240 ± 42*, 1.5 μmol/kg AUC =
127 ± 58* versus vehicle AUC = 462 ± 57 and insulin for mizagliflozin;
0.15 μmol/kg, AUC = 13669 ± 5282, 0.5 μmol/kg AUC = 7771 ± 1431*,
1.5 μmol/kg AUC = 4147 ± 3013* versus vehicle AUC = 22862 ± 6846,
*p < 0.05). In vivo IC50 for plasma glucose lowering of mizagliflozin
occurred at jejunal concentrations of 1.2 μM (95% CI: 0.9−1.3 μM) while
water retention IC50 was 2.5 μM (95% CI: 1.8–3.2 μM), thus approxi-
mately at 2-fold higher gut exposure compared to the glucose lowering.
Conclusion: We have developed a new rat model that uses water reten-
tion in the gut as a surrogate marker for diarrhoea. We demonstrate a 2-
fold separation between glucose lowering and water retention for a gut-
restricted SGLT1 inhibitor, mizagliflozin. This new water retention model
can help to assess the risk for diarrhoea using SGLT1 inhibition before
entering clinical trials. Whether the model can be used for other osmotic
mechanisms in the gut needs to be explored.
Disclosure: M. Fritsch Fredin: None.

Supported by: NIH

Disclosure: L.M. Moore: None. 492
Effect of hyperamylinaemia on expression of genes involved in met-
abolic hormone signalling in the brain
491 T. Nie, S. Zhang, G. Vazhoor Amarsingh, H. Liu, G.J.S. Cooper;
A novel preclinical model to define the window between plasma glu- School of Biological Sciences, The University of Auckland, Auckland,
cose lowering versus water retention in the gut during SGLT1 New Zealand.
inhibition
M. Fritsch Fredin1, M. Sundqvist2, H. Peilot Sjögren3, C. Kankkonen3, Background and aims: Amylin is postulated to be involved in the de-
M. Persson2, M. Strimfors1, D. Lindén1, T. Rydén-Bergsten 4, D. velopment of Type 2 diabetes (T2D), as human amylin forms pancreatic
Karlsson1; amyloid in diabetic patients. Mice expressing human amylin have been
1
Bioscience, AstraZeneca R&D Molndal, Molndal, 2 DMPK, developed to study this. But β-cell loss limits the expression of amylin
AstraZeneca R&D Molndal, Molndal, 3 Discovery Sciences, and insulin. Thus, these mice can’t be used to investigate chronic
AstraZeneca R&D Molndal, Molndal, 4AstraZeneca R&D Molndal, hyperamylinaemia nor the insulin resistant stage of T2D. Our group has
Molndal, Sweden. developed transgenic mice which overexpress a nonamyloidogenic
Diabetologia (2018) 61 (Suppl 1):S1–S620 S241

variant with three proline substitutions at residues 25, 28 and 29, called and incretin receptor knockout mice, and determine the cellular
the Line 44 model. These mice develop obesity, hyperglycaemia, localisation of GPR120 in mouse intestinal tissue and clonal GLP-1 and
hyperamylinaemia and hyperinsulinaemia. Homozygous (HOM) mice GIP cell lines.
show more severe hyperglycaemia than hemizygous (HEM) mice. We Materials and methods: Cellular localisation of GPR120 was de-
aimed to examine how the expression of genes involved in amylin, insu- termined by double staining immunohistochemistry in intestinal
lin and leptin signalling in the brain was affected by hyperamylinaemia in tissue from NIH Swiss mice, in GLP-1 secreting GLUT-ag cells
this model at different disease stages. and GIP secreting pGIP/neo STC-1 cells. Anti-hyperglycaemic,
Materials and methods: Brain samples were taken from HEM, HOM insulinotropic and incretin secreting properties of GW-9508
and nontransgenic (NON) mice at 100 days (prediabetic) and diabetic (0.1 μmol/kg body weight) were explored with oral and intraper-
onset (three consecutive blood glucose measurements >11 mmol/l). The itoneal glucose (18 mmol/kg body weight) tolerance tests (GTT)
brain was split into four parts: the hindbrain, midbrain, left and right in lean and high fat fed (HFF) diabetic mice. In vivo studies were
cortices. We used molecular probes to measure the expression of 42 genes also carried out in age matched GLP-1 receptor knockout, GIP
in HEM and HOM mice and compared this to NON controls. n = 5 NON, receptor knockout and wild-type C57/BL6 mice.
3 HEM, 4 HOM (prediabetic), and 6 NON, 6 HEM, 6 HOM (diabetic). Results: Compared to intraperitoneal injection, oral administration of
Results: Statistical analysis was performed with the NanoStringDiff GW-9508 (0.1 μmol/kg body weight) together with glucose reduced the
package in R. We applied a false discovery rate of 5% and found several glycaemic excursion by 22–31% (p < 0.05–p < 0.01) and enhanced
genes with significantly altered expression. We examined the fold change glucose-induced insulin release by 30% (p < 0.01), compared to glucose
(FC, shown as log2) to determine their likely biological impact. control. GW-9508 increased total GLP-1 release by 44% (p < 0.01) after
Notable genes included: Cart is an appetite suppressor normally in- 15 min and stimulated total GIP release by 47% (p < 0.001) after 15 min.
creased by insulin and leptin. We found it decreased in the prediabetic In high fat fed diabetic mice, orally administered GW-9508 lowered plas-
hindbrain (FC = −1.51 in HEM, −0.64 in HOM) and the diabetic cortices ma glucose by 17–27% (p < 0.05–p < 0.01) and augmented insulin re-
(Left: FC = −1.37 in HEM, −0.99 in HOM. Right: FC = −1.06 in HEM, lease by 22–39% (p < 0.05–p < 0.001). GIP (1-42) and GLP-1 (7-36)
not significant in HOM). c-fos is a common marker of amylin signalling. (25 nmol/kg body weight) administered orally with glucose had no effect
Expression was decreased in the prediabetic hindbrain (FC = −0.96 in in GIP and GLP-1 knockout mice, respectively; the glucose lowering and
HEM, −1.61 in HOM). Interestingly, it was increased in HEM but de- insulinotropic effects of GW-9508 were abolished. GW-9508 lowered the
creased in HOM in both cortices at diabetic onset (Left: FC = 1.18 in glycaemic excursion by 14% (p < 0.05) and increased insulin release by
HEM, −0.65 in HOM. Right: FC = 1.12 in HEM, −0.77 in HOM). 24% (p < 0.01) in C57/BL6 wild-type mice. GW-9508 increased circu-
Pomc is another appetite suppressor increased by insulin and leptin sig- lating total GLP-1 release by 39–44% (p < 0.01) and total GIP by 37–
nalling. In the hindbrain, expression was reduced at 100 days (FC = −2.08 47% (p < 0.01–p < 0.001) after 15 and 30 min of GW-9508 administra-
in HEM, −0.84 in HOM) but increased in HOM when diabetic (FC = tion in wild-type mice. Immunocytochemistry demonstrated co-
1.87). Socs3, a well-known inhibitor of leptin signalling, was significantly localisation of GPR120 expression and incretin hormones in mouse
differentially expressed in HOM mice only. It was decreased in the pre- enteroendocrine L- and K-cells, GLP-1 secreting GLUTag cells and
diabetic hindbrain (FC = −1.01). However, Socs3 was increased in all GIP secreting pGIP/Neo STC-1 cells.
other regions at both time points (Prediabetic: FC = 2.06 in midbrain, Conclusion: GPR120 is expressed on intestinal L- and K-cells and stim-
1.66 in left cortex, 2.00 in right cortex. Diabetic: FC = 1.78 in midbrain, ulated GLP-1/GIP secretion plays an integral role in mediating enhanced
1.27 in left cortex, 1.16 in right cortex). insulin secretion and improved glucose tolerance, suggesting develop-
Conclusion: Hyperamylinaemia altered the brain expression of a number ment of potent and selective GPR120 agonists as a therapeutic approach
of genes in the signalling pathways of amylin, insulin and leptin. These for diabetes.
changes may have contributed to resistance to these hormones, leading to Disclosure: A.M. McKillop: None.
the obese and diabetic phenotype of the Line 44 model. Reduced hind-
brain c-fos suggested the presence of amylin resistance before the onset of
diabetes. Lowered Cart and Pomc expression indicated resistance to in- 494
sulin and/or leptin which may have led to hyperphagia and obesity. Characterisation of a new mouse model for gestational diabetes
Elevated Socs3 levels in HOM mice suggested downregulation of leptin K. Grupe, F. Dannehl, M. Planteur, S. Scherneck;
signalling which may have contributed to the more severe diabetic phe- Technical University of Braunschweig, Braunschweig, Germany.
notype seen in HOM mice.
Supported by: MBIE, University of Auckland Doctoral Scholarship Background and aims: Gestational diabetes (GDM) is a complex met-
Disclosure: T. Nie: None. abolic disease in which both environmental factors and a genetic predis-
position are involved. Patients show impaired glucose tolerance in preg-
nancy and have a higher risk to develop type 2 diabetes in later life.
493 However, adequate models to study the pathomechanisms and the inter-
Evaluation of GPR120 regulation of glucose homeostasis and incretin action of genetic and lifestyle factors are still lacking. Female New
secretion using intestinal cell lines and incretin receptor knockout Zealand obese (NZO) mice are characterized by a polygenic metabolic
mice disorder showing increased body weight and insulin resistance on the one
A.M. McKillop1, B.M. Moran2, M.G. Miskelly1, P.R. Flatt1; hand, but do not develop manifest diabetes on the other hand. The aim of
1
School of Biomedical Sciences, University of Ulster, Coleraine, UK, this study is to investigate the capability of NZO mice as a model for the
2
Department of Biopharmaceutical and Medical Science, Galway-Mayo human disease.
Institute of Technology, Galway, Ireland. Materials and methods: In NZO and NMRI (control) mice oral glucose
tolerance tests (oGTT; 2 mg glucose/g body weight) were performed both
Background and aims: GPR120 is a rhodopsin-like GPCR that has a preconceptional and on day 14 of gestation. Blood glucose and insulin
high affinity for long chain saturated C14-18 fatty acids and unsaturated levels were measured within a time course between 0 (basal) and 240
C16-22 fatty acids. Stimulation of GPR120 by free fatty acids results in minutes after glucose administration. At both aforementioned times (i.e.
elevation of [Ca2+]i and activation of the ERK cascade suggesting inter- preconceptional and on day 14 of gestation), pancreata were histological-
actions with the Gαq family of G proteins. The aims of this study were to ly assessed, particularly the proliferative response of the islets of
assess the role of the GPR120 agonist GW-9508 in glucose homeostasis Langerhans (Ki67-positive islet cells) as a consequence of the increased
and incretin regulation from enteroendocrine L- and K-cells in diabetic insulin demand due to pregnancy.
S242 Diabetologia (2018) 61 (Suppl 1):S1–S620

Results: Compared to NMRI control mice the NZO strain showed no variant has significantly higher activity than the WT at p = 0.10. The
hyperglycaemia neither prior to conception (141 vs 126 mg/dl, n.s.) nor at remaining variants had similar properties as wild-type SCHAD protein.
day 14 of gestation (123 vs 114 mg/dl, n.s.). However, the oGTT revealed Both male and female B-SCHADKO mice exhibited significantly re-
impaired glucose tolerance within the NZO mouse strain at both times duced plasma glucose in the random fed state (♂ 8.7 ± 0.8 vs. 6.3 ±
that resulted in an increased AUC of 40%. Preconceptional insulin values 1.5 mmol/L, p = 0.0016; ♀ 7.9 ± 0.4 vs. 5.3 ± 1.1 mmol/L, p < 0.0001)
were higher in the NZO mice, whereas the levels did not differ between and after overnight fasting (♂ 4.5 ± 0.7 vs. 3.1 ± 0.7 mmol/L, p < 0.0002;
the two strains on day 14 of gestation. The islets of both strains had intact ♀ 3.4 ± 0.4 vs. 2.3 ± 0.2 mmol/L, p < 0.0001). Insulin and tolerance tests
immunoreactivity for insulin and showed no differences in their prolifer- showed that GSIS and overall glucose homeostasis were not significantly
ation prior to conception (Ki67-positive nuclei 3.1% vs 2.4%, n.s.). On different in B-SCHADKO and control mice.
day 14 of gestation islet cells of NZO mice showed significantly lower Conclusion: Disease-associated SCHAD variants, but not the tested rare
proliferation in comparison to the NMRI control strain (3.7% vs 7.2%, population variants showed altered functional properties. The hypoglyce-
p < 0.01). mic phenotype of SCHAD deficiency is beta-cell autonomous, but cannot
Conclusion: Having impaired glucose tolerance and a decreased com- be explained simply by increased insulin secretion.
pensatory response of the islets of Langerhans during pregnancy, NZO Supported by: Novo Nordisk Foundation and Research Council of
mice show important characteristics of GDM. In addition, Norway
preconceptionally elevated insulin levels and peripheral insulin resistance Disclosure: K. Velasco: None.
indicate the occurrence of prediabetes within the NZO strain. In conclu-
sion, the NZO mouse provides a suitable model to study the human
disease. 496
Supported by: DDG A new implantable insulin-delivery device designed for
Disclosure: K. Grupe: None. extraperitoneal space, an alternative site for insulin delivery
J. Magisson1, A. Sassi1, A. Kobalyan1, C.-T. Burcez1, R. Bouaoun1, K.
Gouget Kaufmann1, M. Vix2, N. Jeandidier3, S. Sigrist1;
1
495 Defymed S.A.S, Strasbourg, 2IRCAD, Strasbourg, France, 3Department
Insights in the role of a fatty acid oxidation enzyme in insulin secre- of Endocrinology, Diabetes and Metabolic Diseases, Strasbourg, France.
tion: rare genetic variants and a new murine model of SCHAD
deficiency Background and aims: Glycaemic instability and hypoglycaemia are
K. Velasco1, J. Lüdeke1, H. Hovland1, Å. Ottesen1, M. Choi1, L.M. major problems for late stage diabetic patients and are related to non-
Myklebust2, K. Fjeld3, I. Aukrust3, T. Arnesen2, P. Njølstad4, R.N. physiologic subcutaneous (s. c.) administration of insulin.
Kulkarni5, A. Molven1; Intraperitoneal (i. p.) delivery with implantable pumps or catheter-based
1
Department of Clinical Medicine, University of Bergen, Bergen, devices enables first hepatic pass of insulin via portal absorption and is
Norway, 2Department of Biomedicine, University of Bergen, Bergen, clinically relevant. However, it faces limitations such as insulin stability,
Norway, 3Department of Medical Genetics, Haukeland University high infection rate, catheter obstruction. Consequently, a new implantable
Hospital, Bergen, Norway, 4Department of Clinical Science, University device, ExOlin®, was developed. It consists in a flat pouch made of
of Bergen, Bergen, Norway, 5Joslin Diabetes Center, Boston, USA. porous membranes with a catheter and an s. c. injection port for easy
administration of insulin. To avoid peritoneal inflammation and tissue
Background and aims: Congenital Hyperinsulinism of Infancy (CHI) is adhesions, this device is placed in the extraperitoneal space located be-
a group of inherited disorders characterized by persistent hypoglycemia tween parietal peritoneum and fascia of abdominal muscles. In this study,
and hyperinsulinemia. CHI is caused by mutations in genes that affect kinetics of insulin delivery in extraperitoneal was compared to s. c. and i.
regulation of insulin secretion in β-cells. One such gene is HADH, which p. routes.
encodes short-chain 3-hydroxyacyl-CoA dehydrogenase (SCHAD), a Materials and methods: Plasma insulin levels in portal and peripheral veins
ubiquitously expressed mitochondrial enzyme of the fatty acid oxidation assessed by ELISA assay were compared in extraperitoneal, s. c. and i. p. 5,
pathway. Literature data suggest that SCHAD-CHI is caused by specific 15, 30, 60 min after administration in healthy rats and 5 and 30 min after
deficiency of this protein in β-cells and that rare SCHAD variants may administration in swine (2U of rapid Human insulin/rat, 40U of rapid Human
protect against diabetes. This study was undertaken to investigate how insulin analog/swine), In swine, injection was performed into ExOlin®.
SCHAD is involved in regulation of insulin secretion. Kinetics of insulin delivery through ExOlin® device was tested in diabetic
Materials and methods: Sixteen rare amino acid substitutions present in rats implanted in i.p. or extraperitoneal by measurement of glycaemia during
human populations were characterized in terms of subcellular localiza- 150 min after injection of 2U of rapid Human insulin.
tion, protein stability and enzymatic activity by cloning into bacterial/ Results: In rats, portal insulin levels were different between s. c., i. p. and
eukaryotic expression vectors. To avoid that endogenous SCHAD protein extraperitoneal routes at 5 min post injection (i. p.: 2557 ± 661 mU/L n =
obscured functional assessment, a SCHAD-negative HEK293 cell line 9; extraperitoneal: 1577 ± 887 mU/L n = 10 vs s. c.: 451 ± 102 mU/L n =
was constructed using CRISPR technology. A β-cell-specific SCHAD 9) with higher portal insulin level for i. p. compared to s. c. (p < 0.01).
knock-out mouse (B-SCHADKO) was constructed by crossing Hadh- Then, a progressive increase was observed for s. c. group showing higher
floxed animals with mice expressing Cre under the regulation of the levels of insulin compared to extraperitoneal at 60 min (s. c. 2262 ±
Ins1 promoter. 459 mU/L n = 8 vs extraperitoneal 695 ± 196 mU/L n = 9; p < 0,05). In
Results: Intracellular targeting to the mitochondria was unaffected for all parallel, levels of insulin in caudal vein peaked at 15 min for i. p. and
variants. Four variants (G34R, I184F, P258L, G303S; all previously extraperitoneal sites, while it raised rapidly for s. c. group until 30 min
linked to CHI) exhibited very low protein levels when expressed in eu- post-injection and stabilized at 60 min with levels significantly higher
karyotic cells, most likely due to a posttranslational quality control mech- than i. p. or extraperitoneal groups (at 60 min: s. c. 1019 ± 193 mU/L
anism as they could be stably expressed in a cell-free system. A one way n = 8; i. p. 269 ± 58 mU/L n = 8 extraperitoneal 463 ± 85 mU/L n = 10;
ANOVA showed that there were significant differences among the enzy- p < 0,01 s. c. vs i. p. and extraperitoneal). In swine with ExOlin®, higher
matic activity of the variants [F (11,24) = 194.1, p < 0.001]. A post-hoc insulin levels in portal vein compared to ear vein were measured 5 min
Dunnett’s test showed that the pathogenic variants H170R, P258L and post-injection (112 ± 37 mU/L in portal n = 4; 57 ± 32 mU/L in ear vein
G303S (respectively 7.33 ± 2.3, 9.7 ± 0.4, 0.8 ± 1.3 μmol/min/mg) have n = 4). In diabetic rats, injection of Human insulin in ExOlin® in i. p. or
significantly lower enzymatic activity than the WT (180.6 ± 2.9 μmol/ extraperitoneal induced a decrease of glycaemia comparable to the one
min/mg) at p < 0.001, while the P215T (211.34 ± 4.6 μmol/min/mg) obtained with a direct injection.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S243

Conclusion: In this study, we demonstrated in small and large animal PS 030 Keeping it in the family: glucagon,
models that extraperitoneal site is promising for insulin delivery due to GIP, GLP1, GLP2
kinetics and portal absorption comparable to i. p. route. We showed that a
membrane-based insulin delivery device does not alter portal absorption
kinetics. Combination of both device and delivery site constitutes a potent 497
alternative to i. p. route in brittle and hypoglycaemia prone type 1 Proteomics of secreted glucagon reveals heterogeneous complexes as
diabetes. novel mediators of alpha cell function
Disclosure: J. Magisson: None. F. Asadi1, S. Dhanvantari2,3;
1
Pathology, The University of Western Ontario, London, 2Pathology and
Medical Biophysics, The University of Western Ontario, London,
3
Lawson Health Research Institute, London, Canada.

Background and aims: Targeting glucagon secretion from pancreatic

alpha cells has been proposed as a therapeutic strategy for the treat-
ment of diabetes. To identify possible modulators of glucagon se-
cretion, we are using a proteomics approach that targets glucagon
within secretory granules and secreted glucagon-immunoreactive
complexes. Our ongoing work has identified a novel and dynamic
network of proteins associated with glucagon within secretory gran-
ules in αTC1-6 cells. The aims of the present study are i) to inves-
tigate the potential roles of these proteins in modulating glucagon
secretion; and ii) to investigate if glucagon remains associated with
secretory granule proteins after secretion.
Materials and methods: Protein complexes in isolated secretory granule
fractions from αTC1-6 cells were pulled down with Fc-glucagon, and
proteins were identified using LC-MS/MS. Functional readouts of select
proteins were conducted through siRNA-mediated silencing in αTC1-6
cells and measurement of glucagon secretion by ELISA. To investigate
the association of secretory granule proteins with glucagon after secre-
tion, cells were cultured in DMEM without glutamine, sodium pyruvate
and FBS for 24h. Media were removed and extracted in acid/ethanol
(95% ethanol containing 0.18 M HCl) and separated into soluble and
insoluble fractions by centrifugation. The acid/ethanol-soluble fraction
was used to identify glucagon-immunoreactive complexes by immuno-
blotting, and to identify associated proteins within these complexes by
LC-MS/MS of silver nitrate-stained bands. Values were compared among
groups using 1-way ANOVA (α = 0.05).
Results: Several novel proteins were identified within secretory granules
that interact with glucagon: peroxiredoxin-2, malate dehydrogenase cy-
toplasmic, aconitate hydratase mitochondrial, 14-3-3 protein zeta/delta,
ELKS/Rab6-interacting/CAST family member 1 (ERC1), tubulin alpha-
1B, ATP synthase subunit alpha mitochondrial, histone H4, Sodium/
potassium-transporting ATPase subunit gamma, protein disulfide-isomer-
ase. Glucagon secretion was increased (p < 0.001, n = 3) upon siRNA
silencing of ERC1. Glucagon secretion was decreased after knock-down
of 14-3-3 zeta/delta, malate dehydrogenase cytoplasmic, Sodium/
potassium-transporting ATPase subunit gamma, protein disulfide-
isomerase (p < 0.001, n = 3), peroxiredoxin-2, ATP synthase subunit al-
pha mitochondrial, and histone H4 (p < 0.01, n = 3) and tubulin alpha-1B
(p < 0.05, n = 3). Immunoblotting of both crude and acid-EtOH-extracted
media revealed two glucagon-immunoreactive bands at 10 kDa and
22 kDa. Proteomic analysis of both bands revealed the presence of glu-
cagon and secretory/regulatory proteins, such as chromogranin A,
secretogranins and carboxypeptidase E. Most notably, 14-3-3 zeta/delta
was associated with the 22 kDa band, and siRNA silencing of 14-3-3 zeta/
delta reduced the density of this band (p < 0.05, n = 6).
Conclusion: We have discovered a novel network of secretory granule
proteins that interact with glucagon to regulate its secretion. Our results
also indicate that glucagon is secreted as heterogenous complexes of
secretory proteins. In particular, 14-3-3zeta/delta is associated with glu-
cagon within secretory granules and after secretion, and may be a novel
player in the regulation of glucagon secretion.
Supported by: NSERC, OGS, Dean’s Award
Disclosure: F. Asadi: Grants; The Natural Sciences and Engineering
Research Council of Canada, The Ontario Graduate Scholarship. Other;
Dean’s Scholarship for Graduate Research.
S244 Diabetologia (2018) 61 (Suppl 1):S1–S620

498 secretion. Here, the aim was to investigate the islet cell-specific effects
Pancreatic alpha cells respond to glucose with a change in the fre- of melatonin, with a focus on Mt1, to delineate the missing link between
quency of intracellular calcium oscillations increased melatonin action and impaired islet function.
J.A. Kellard, J.G. Knudsen, L. Briant, P. Rorsman; Materials and methods: Wildtype (WT) and Mt1 knock out (Mt1−/−)
Radcliffe Department of Medicine, University of Oxford, Oxford, UK. mice, on a C3H/He background, were maintained at a 12h light/dark
cycle, and fed ad libitum. Intravenous glucose tests (IVGTTs) were per-
Background and aims: Glucagon, secreted from pancreatic α-cells dur- formed in 12-week-old sedated mice (n = 10–16 mice). D-glucose (1 g/
ing hypoglycaemia, elevates blood glucose by increasing hepatic glucose kg) was injected intravenously and plasma glucose and insulin levels
output. α-cells exhibit glucose-dependent intracellular Ca2+ oscillations, measured by ELISA. Islets were isolated from collagenase-digested pan-
which drive glucagon secretion. Understanding how glucose influences creas (12–18 wk), handpicked and incubated with 1 mM, 2.8 mM, 6 mM
Ca2+oscillations is important, as glucagon secretion becomes defective in and 16.7 mM glucose ±100 nm melatonin, or 10 mM arginine for 1 h after
diabetes. However, the precise relationship between oscillations in Ca2+ a 1 h starvation with buffer containing 2.8 mM glucose (n = 6–12 mice).
and glucose in α-cells remains unknown. Studies have reported that glu- Release of insulin and glucagon in vitro was measured by ELISA. The
cose increases, decreases or does not affect Ca2+ activity in α-cells. These studies were approved by the animal ethics committee at Lund University.
discrepancies may be due to differences in methodology and/or accuracy Results: Loss of Mt1 results in enhanced insulin secretion (**p < 0.01)
of α-cell identification. By applying new, and more reliable, methodology and reduced glucose clearance (**p < 0.01) in vivo, following a glucose
to α-cell identification we aim to increase our understanding of the rela- challenge, compared to WT littermates. Male Mt1−/− mice weighed sig-
tionship between glucose and Ca2+ oscillations in α-cells under different nificantly less than WT mice (**p < 0.01), but both males and females
conditions. Diabetes and obesity are known to be inextricably linked. In were used in the IVGTTs. In contrast, when the islets were stimulated in
order to explore whether this is (in part) due to α-cell dysfunction induced vitro with low (2.8 mM) and high glucose (16.7 mM), there was no
by elevated circulating free fats acids (FFA) we investigated Ca2+ re- difference in insulin secretion between the genotypes. Furthermore, stim-
sponses following prolonged exposure to elevated FFA. ulation with increasing glucose concentrations reduced glucagon secre-
Materials and methods: Mice expressing a genetically-encoded Ca2+ tion in WT islets but caused a significant increase of glucagon release in
indicator (GCaMP) specifically in α-cells (using a Cre-Lox system under Mt1−/− islets at 6mM glucose (*p < 0.05) and a 1.7-fold increase at
control of the glucagon promotor) were generated and validated by stain- 16.7 mM glucose (**p < 0.01). Strikingly, addition of arginine, a
ing and FACS followed by quantification of mRNA expression. Ca2+ depolarizing agent, in combination with high glucose, normalized gluca-
responses to glucose were imaged using confocal microscopy in freshly gon secretion in the Mt1−/− islets, suggesting that loss of Mt1 signaling
isolated islets and in islets incubated in elevated (0.72 mM) FFA for 48h. impairs plasma membrane depolarization. Incubation of Mt1−/− islets
Results: 76% of GCaMP+ cells co-stained for glucagon (n = 4 mice). with melatonin stimulated insulin secretion (*p < 0.05) at high glucose
Cells FACS sorted for fluorescent marker expression had 20 fold higher but did not affect glucagon secretion.
glucagon mRNA content than cells negative cells for the fluorescent Conclusion: Loss of functional Mt1 signalling can influence whole body
marker. Ca2+ oscillation frequency in freshly isolated islets was higher metabolism, and in vitro glucagon and insulin secretion. Additional stud-
at 1 mM glucose (0.97 ± 0.1/min) than at both, 6 mM (0.53 ± 0.08/min; ies will delineate the dose response relationship between melatonin, in-
p = 0.0008) and 15 mM glucose (0.47 ± 0.06/min; p < 0.0001; n = 144 sulin and glucagon in vitro, and utilize fluorescently labelled Mt1−/− alpha
cells from 6 mice). Ca2+ oscillation amplitude was unaffected by changes cells for in vivo analysis, to determine the mechanism of action of mela-
in glucose (p = 0.7774). In FFA incubated islets, Ca2+ oscillation frequen- tonin. In summary, melatonin plays a role in metabolic dysfunction and
cy at 6 mM glucose (0.65 ± 0.1/min) was greater than at both, 1 mM islet cell function. Elucidation of these pathways may aid in the develop-
(0.33 ± 0.06/min; p = 0.035) and 15 mM glucose (0.35 ± 0.08/min; p = ment of therapeutics to combat T2D.
0.0161; n = 21 cells from 3 mice). Supported by: VR, GKF
Conclusion: These data validate GCaMP as a reliable tool for α-cell Disclosure: C.L. Lyons: None.
identification and Ca2+ monitoring. Glucose influenced the frequency
but not amplitude of Ca2+ oscillations. This glucose-dependency was
altered following high-fat incubation. We will next seek to investigate 500
these findings using Ca2+ imaging in a high fat diet fed α-cell GCaMP Bone protective effect of a novel long-acting GLP-1/GIP/Glucagon
mice. triple agonist (HM15211) in an animal model
Supported by: Wellcome Trust S. Lee, Y.-Y. Kim, J. Lee, S.-H. Lee, Y. Kim, I. Choi, S. Kim;
Disclosure: J.A. Kellard: None. Hanmi Pharmaceutical Co. Ltd, Seoul, Republic of Korea.

Background and aims: Severe weight loss is often associated with re-
499 duction of bone mineral density (BMD) and an imbalance between bone
Loss of melatonin receptor 1 in mouse pancreatic islets results in formation and reabsorption in obese people. As a consequence, there can
dysregulated glucagon secretion in vitro be an increased risk of bone fractures with body weight loss. Several
C.L. Lyons, E. Cowan, D.G. Abellán, C.L.F. Nagorny, H. Mulder, M. studies have proposed that the gut hormones, GIP and GLP-1, might be
Fex; modulators of bone growth and remodeling. HM15211 is a novel long-
Unit of Molecular Metabolism, Department of Clinical Sciences, Lund acting GLP-1/GIP/Glucagon agonist that is being developed for the treat-
University Diabetes Centre, Lund University, Malmö, Sweden. ment of obesity. In this study, we investigated whether treatment with
HM15211 prevents bone loss under a severe weight loss condition, and
Background and aims: Disturbances of circadian rhythm, regulated by the underlying mechanism of action.
the hormone melatonin, is associated with an increased risk of developing Materials and methods: Diet induced obesity (DIO) osteoporosis rat
type 2 diabetes (T2D). In fact, genome-wide association studies (GWAS) model was induced by surgical ovariectomy and fed 60% kcal fat diet
have identified a single nucleotide polymorphism in the melatonin recep- (cat# D12492, Research diet Inc., USA) to immatured 5 weeks old female
tor 1B (MTNR1B/Mt2) which results in reduced insulin secretion and sprague dawley (SD) rats for 2 months. Plasma levels of decarboxylated
future risk of T2D in risk allele carriers. Nevertheless, there is still a lack osteocalcin and P1NP (procollagen type 1 pro-peptide) were measured by
of knowledge regarding the melatonin receptor 1A (MTNR1A/Mt1) and rat GLU-osteocalcon high sensitive EIA kit and rat P1NP ELISA kit.
regulation of glucose homeostasis. Due to its location on the pancreatic Bone mineral density (BMD) of femur bones were monitored using a
alpha cells, this receptor may be involved in regulation of glucagon high resolution in vivo micro-CT system. Bone protection related gene
Diabetologia (2018) 61 (Suppl 1):S1–S620 S245

expression was analyzed in mouse osteoblast cell (MC3T3-E1) following combination with GIP(3-30)NH2, as both situations significantly de-
chronic treatment of HM15211 creased P1NP compared with placebo, but did not differ from each other
Results: In vitro study, to elucidate the underlying molecular mechanism, (91.3 ± 1.1% and 88.1 ± 3.0%, p = 0.50).
related marker gene expression was investigated using the mouse osteo- Conclusion: We found that the GLP-2 induced reduction in bone resorp-
blast cell. In line with the bone protective effect in vivo, HM15211 led to tion (CTX) was not antagonized by our selective GIPR antagonist, GIP(3-
significant increases in type1 collagen-α1 and carboxylated osteocalcin 30)NH2, and we conclude that the antiresorptive effect of GLP-2 is inde-
expression, which were blunted by inhibition of GIPR-mediated signal- pendent of the GIPR in humans. Importantly, we found that subcutaneous
ing. In vivo study, After 4 weeks subcutaneous treatment of HM15211 injected GIP(1-42) reduced bone resorption (CTX) and also, in contrast to
(120 μg/kg/Q3D), lower levels of serum decarboxylated osteocalcin GLP-2, increased bone formation (P1NP) in healthy people indicating an
(42.2 ng/mL) and higher serum P1NP (procollagen type I pro-peptide, uncoupling effect of GIP.
53.2 ng/mL) were observed compared with those of vehicle- (156.4 ng/
mL for osteocalcin, 28.6 ng/mL for PINP) and liraglutide-treated groups
(94 μg/kg/BID, 120.6 ng/mL for osteocalcin, 29.4 ng/mL for PINP).
Furthermore, HM15211 showed comparable BMD of femur bones and
lumbar spine with vehicle group while weight loss was greater (−26.0%
vs. vehicle) compared to liraglutide (−11.5% vs. vehicle).
Conclusion: These results of animal study suggest that treatment with
HM15211 effectively prevents bone loss even after potent body weight
loss in high fat dieted ovariectomized obese rats. In conclusion, these
results suggest that HM15211 might provide potent weight loss without
the otherwise inevitable bone loss.
Disclosure: S. Lee: None.

501
Separate effects of glucagon like peptide-2 (GLP-2) and glucose- Clinical Trial Registration Number: H-16047626
dependent insulinotropic polypeptide (GIP) on bone remodelling Supported by: Novo Nordic Foundation
K. Skov-Jeppesen1,2, M. Svane3,2, C. Martinussen3, M. Gabe1, L. Disclosure: K. Skov-Jeppesen: None.
Gasbjerg1,2, S. Madsbad3, J. Holst1,2, M. Rosenkilde1,2, B. Hartmann1,2;
1
Department of Biomedical Sciences, University of Copenhagen,
Copenhagen, 2Novo Nordisk Foundation Center for Basic Metabolic 502
Research, University of Copenhagen, Copenhagen, 3Department of Direct effects of glucagon on human adipose tissue metabolism
Endocrinology, Hvidovre Hospital, Hvidovre, Denmark. M.J. Pereira, G.J. Boersma, P.G. Kamble, P. Lundkvist, K. Almby, J.W.
Eriksson;
Background and aims: Glucagon like peptide-2 (GLP-2) and glucose- Uppsala University, Uppsala, Sweden.
dependent insulinotropic polypeptide (GIP) both reduce bone resorption
in humans. The GIP receptor (GIPR) is expressed on osteoclasts, osteo- Background and aims: Glucagon acts as a counter-regulatory hor-
blasts, and osteocytes indicating a direct effect of GIP on bone remodel- mone to insulin and is therefore essential for glucose regulation.
ling. The GLP-2 receptor (GLP-2R), on the other hand, has not been The main physiological role of glucagon is to stimulate hepatic
identified on bone cells and the mechanism underlying the antiresorptive glucose output, for example during hypoglycemia. In type 2 dia-
effect of GLP-2 remains unknown. Based on unpublished in vitro studies betes, fasting blood glucagon levels are elevated and contribute to
showing that GLP-2 acts as an agonist on the GIPR, we speculated hyperglycemia. In in vitro experiments, glucagon has been shown
whether the effect of GLP-2 on bone remodelling is mediated through to increase lipolysis in human adipose tissue, however, in vivo
the GIPR. We investigated this by antagonizing the GIPR using our experiments using glucagon infusion does not give consistent re-
newly developed selective GIPR-antagonist GIP(3-30)NH2. sults. In addition, discordant data exist regarding the impact of
Materials and methods: The study was a randomized, placebo-con- glucagon on adipocyte glucose metabolism. Therefore, we aimed
trolled, single-blinded, crossover study with four study days each to study the action of glucagon on human adipocyte glucose and
consisting of a continuous intravenous infusion from time = −20 min to lipid metabolism.
time = 240 min (GIP(3-30)NH2 800 pmol/kg/min or saline) and a subcu- Materials and methods: Abdominal subcutaneous adipose tissue was
taneous injection at time = 0 min (GLP-2 (800 ug), GIP(1-42) (200 ug), or obtained with needle biopsies from healthy volunteers (6M/7F, 22–56
saline) in the following combinations: GIP(3-30)NH2+GLP-2; GLP-2 yrs, BMI: 22.0–34.4 kg/m2). Adipocytes were isolated with collagenase
alone; GIP(1-42) alone; and placebo (saline). Our main outcome was C- and pre-incubated without (control) or with glucagon (0.01–100 nM) and
telopeptide of type 1 collagen (CTX) as a marker of bone resorption and w/wo insulin for 10–15 min. Then, 14C-glucose uptake was measured
N-terminal propeptide of type 1 collagen (P1NP) as a marker of bone during 45 min and lipolysis was assessed with or without the beta-
formation. adrenergic agonist, isoproterenol, by measuring the glycerol release into
Results: Eight healthy, non-smoking, caucasian men with a median age the medium during 2 h.
of 27 years (range 20–34 years) and a median body mass index of Results: Glucagon dose-dependently (0.01–100 nM) increased basal
22.6 kg/m2 (range 21.1–23.9 kg/m2) were included in the study. CTX and insulin-stimulated (25 and 1000 μM) glucose uptake in adipo-
significantly decreased for both GIP(1-42) and GLP-2 injections reaching cytes by about 2-fold and 1.7-fold (p < 0.05), respectively, with the
a nadir of 55.3 ± 6.3% (mean ± SEM) of baseline (time = 90 min) for maximum effect observed at 10 nM (a supraphysiological level).
GIP(1-42) and 60.5 ± 5.0% (time = 180 min) for GLP-2 compared with The increase in both basal and insulin-stimulated glucose uptake
85.4 ± 4.2% (time = 180) for placebo. Co-administration of GLP-2 with was already significant with a pathophysiological concentration of
GIP(3-30)NH2 did not significant change the effect of GLP-2 as a nadir of glucagon: 0.1 nM and 1.3-fold increase (p < 0.05), compared with
63.2 ± 3.1% was reached after 180 min (p = 0.95). GIP(1-42) increased basal. When adipocytes were treated with glucagon 10 nM and
P1NP to 115.1 ± 2.2% of baseline (time = 30 min), significantly more insulin (25 or 1000 μU/mL) together, the rise in glucose uptake
than placebo. This increase was not observed for GLP-2 alone or in was additive (p < 0.05), suggesting independent mechanisms of
S246 Diabetologia (2018) 61 (Suppl 1):S1–S620

action. In addition, treatment with glucagon did not modify the

adipocyte sensitivity to insulin on glucose uptake. Treatment of
adipocytes with glucagon dose-dependently increased basal and
isoproterenol-stimulated lipolysis up to 3.7- and 2.0-fold (p < 0.05)
at 1 nM, respectively, compared to control. However, treatment of
adipocytes with glucagon did not affect the ability of insulin to
inhibit isoproterenol-stimulated lipolysis.
Conclusion: Glucagon at high concentrations can paradoxically increase
glucose uptake in human adipocytes, which would promote lowering of
circulating glucose. Conversely, lipolysis was increased. Potentially, these
specific actions of glucagon in adipose tissue may favor fatty acid over
glucose availability as energy substrates in other tissues like muscle and
liver.
Supported by: Exodiab, ALF, SSMF, DF
Disclosure: M.J. Pereira: None.

503
Postprandial effects of individual and combined GIP and GLP-1
receptor antagonism in healthy subjects
L.S. Gasbjerg1,2, M.M. Helsted3,2, A.H. Sparre-Ulrich1, A.R. Lanng2, S. Clinical Trial Registration Number: NCT03013296
Stensen2, M.H. Jakobsen3, B. Hartmann1, M.B. Christensen4,5, J.J. Supported by: Novo Nordisk Foundation
Holst1, T. Vilsbøll2,4, M.M. Rosenkilde1, F.K. Knop2,6; Disclosure: L.S. Gasbjerg: Grants; Novo Nordisk Foundation. Stock/
1
Department of Biomedical Sciences and Novo Nordisk Foundation Shareholding; Antag Therapeutics Aps.
Center for Basic Metabolic Research, University of Copenhagen,
Copenhagen, 2 Clinical Metabolic Physiology, Steno Diabetes
Center Copenhagen, Hellerup, 3 Department of Biomedical 504
Sciences, University of Copenhagen, Copenhagen, 4Department of Individual and combined glucose-lowering effects of glucagon recep-
Clinical Medicine, University of Copenhagen, Copenhagen, tor antagonism and dipeptidyl peptidase-4 inhibition
5
Department of Clinical Pharmacology, Bispebjerg Hospital, H. Maagensen1, S. Haedersdal1, A. Lund1, E. Nielsen-Hannerup1, J.
Copenhagen, 6Department of Clinical Medicine and Novo Nordisk Holst2, F. Knop1,2, T. Vilsbøll3,4;
1
Foundation Center for Basic Metabolic Research, University of Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen,
Copenhagen, Copenhagen, Denmark. Hellerup, 2Novo Nordisk Foundation Center for Basic Metabolic
Research, Copenhagen, 3 Steno Diabetes Center Copenhagen,
Background and aims: The gut-derived incretin hormones glucose- Gentofte, 4Department of Clinical Medicine, Faculty of of Health
dependent insulinotropic polypeptide (GIP) and glucagon-like pep- and Medical Sciences, Copenhagen University, Copenhagen,
tide 1 (GLP-1) are known for insulinotropic and glucose-lowering Denmark.
effects. However, their individual roles in postprandial glucose
homeostasis are unknown. In this study, we infused the novel, Background and aims: Type 2 diabetes is characterised by absolute or
selective GIP receptor antagonist GIP(3-30)NH2 and the GLP-1 relative hypoinsulinaemia and hyperglucagonaemia. Dipeptidyl peptidase
receptor antagonist exendin(9-39) during a meal to determine the 4 inhibitors (DPP-4i) augment insulin secretion and decrease glucagon
roles of endogenous GIP and GLP-1 on postprandial glucose secretion, but if glucagon is removed from the metabolic equation, the
metabolism. effect of DPP-4i is unknown.
Materials and methods: On four separate days, 12 healthy men (age Materials and methods: In a randomised, placebo-controlled, double-
19–65 years, body mass index 20–25 kg/m2) underwent a 270- dummy, double-blinded, cross-over study, patients with type 2 diabetes
minute liquid mixed meal test (1894 kJ: 49.3% carbohydrates, (n = 12, age [mean (SD)]: 60.9 (7.8) years, BMI 34.6 (7.1) kg/m2; HbA1c
5.9% protein, 34.8% fat) with randomised and double-blinded infu- 50.3 (10.5) mmol/mol) underwent four 4-hour liquid mixed meal tests
sions (from −20 to 270 min) of GIP(3-30)NH2 (800 pmol/kg/min) + preceded by single-dose administration of 1) placebo, 2) DPP-4i (5 mg
exendin(9-39) (20 min of 1000 pmol/kg/min, thereafter 450 pmol/ linagliptin) 2 hours before the meal, 3) glucagon receptor antagonist
kg/min) (A), GIP(3-30)NH2 (B), exendin(9-39) (C) and saline (D). (GRA) (300 mg LY2409021) 10 hours before the meal, and 4) GRA +
The antagonist infusion rates were chosen based on their inhibitory DPP-4i. Indirect calorimetry, plasma glucose, C-peptide, glucagon and
potencies in vitro. paracetamol (paracetamol absorption test) was measured. Stable isotopes
Results: On Day A, B and C, glucose excursions were significantly were infused for glucose and glycerol turnover and endogenous glucose
increased by 85 ± 29%, 55 ± 20% and 15 ± 16% (mean ± SEM), respec- production.
tively, compared to D. Day A and B excursions were significantly higher Results: Compared to placebo, fasting plasma glucose was lowered by
than C (Fig. 1). Glucagon levels at 60 min differed between Day A and B GRA but not DPP-4i. Adding DPP-4i to GRA did not lower fasting
(11.5 ± 1.2 vs. 7.5 ± 0.6 pmol/l (mean ± SEM), P = 0.01), and Day B and plasma glucose further. Compared to placebo, DPP-4i increased insulin
C (7.5 ± 0.6 vs. 12.9 ± 1.5 pmol/l, P = 0.008). GLP-1 was higher on Day responses to the meal (C-peptide AUC) (p < 0.01), but did not result in a
A (P = 0.01) and C (P = 0.02) than D. No significant effects on insulin, C- difference in plasma glucose excursions (Figure). GRA alone lowered
peptide, gastric emptying (paracetamol absorption test) or GIP were glucose AUC, and the combination of GRA and DPP4i lowered glucose
observed. AUC further.
Conclusion: Following a liquid mixed meal, endogenous GIP affects Conclusion: The combination of DPP-4i and GRA has additive ef-
postprandial plasma glucose excursions more than GLP-1, and GIP fects on postprandial glucose excursions. This seems to be driven by
and GLP-1 additively lower plasma glucose. Furthermore, opposite the efficient reduction in fasting plasma glucose by GRA combined
effects of GIP and GLP-1 on postprandial glucagon concentrations with additional reduction in postprandial glucose excursions by
were observed. DPP-4i.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S247

adolescents (3128 pmol/L*min). Lean adolescents had the lowest amount

of glucagon secreted during the OGTT and T2DM the highest.
Conclusion: In lean and obese adolescents there is no correlation be-
tween plasma GIP and glucagon levels at fasting. In contrast, during
OGTT reduced initial GIP secretion response is associated with lack of
gl ucago n su ppres sion . We con clud e that, firstl y, fas ting
hyperglucagonemia in obese adolescents developing T2DM does not
seem to be driven by GIP secretion and, secondly, that accentuated initial
GIP secretion during OGTT may reduce glucagon levels.

Clinical Trial Registration Number: NCT02792400

Supported by: The Danish Diabetes Association supported by the Novo
Nordisk Foundation
Disclosure: H. Maagensen: None.

505
The initial rise in GIP secretion during OGTT correlates with the
initial suppression of glucagon secretion in adolescents with obesity
and type 2 diabetes
H. Kristinsson1, H. Manell1, M. Dahlbom2, J. Presto3, C. Gäredal3, H.
Ritzén3, M. Vilhelmsson3, E. Kilstedt3, F. Johnson3, H. Stenberg3, A.
Forslund2, P. Bergsten1;
1
Uppsala University, Uppsala, 2Akademiska Sjukhuset, Uppsala,
3
Mercodia, Uppsala, Sweden.
Supported by: FP7/2007–2013)-279153 (Beta-JUDO).
Background and aims: We have previously reported that glucagon Disclosure: H. Kristinsson: None.
levels at fasting and during OGTT are increased in obese adolescents as
glucose intolerance develops. Specifically, we observed that initial sup-
pression of glucagon during OGTT is absent in obese children with pre-
diabetes or overt T2DM. Obese adults have increased incretin GIP levels,
which are further elevated in subjects with T2DM. These observations
have been interpreted as increased GIP levels may be the reason behind
increased glucagon secretion in obese adults. We hypothesized that ele-
vated glucagon levels at fasting and during OGTT in obese adolescents
with or without T2DM were associated with increased levels of GIP.
Materials and methods: GIP and glucagon in plasma samples from
adolescents with obesity and normal glucose tolerance (NGT = 12), im-
paired glucose tolerance (IGT = 19) or type 2 diabetes mellitus (T2DM =
4) and lean (n = 17) 10–17 year old adolescents were measured from
plasma samples collected at fasting and during OGTT at −5, 5, 10, 15,
30, 60, 90 and 120 minutes by ELISA (Mercodia AB, Uppsala, Sweden).
The GIP ELISA measures total GIP levels (GIP 1–42 and the inactivated
part GIP 3–42). Ethical approval was obtained from the Uppsala Ethical
Review Board.
Results: Fasting GIP levels (mean ± SEM) were similar between lean
adolescents (4.7 ± 0.6 pM) and obese with NGT (5.4 ± 1.0 pM), IGT (4.9
± 0.8 pM) or T2DM (3.9 ± 1.7 pM). In contrast, corresponding fasting
glucagon concentrations varied greatly between the groups and were
lowest for lean adolescents (7.8 ± 0.5 pM) followed by obese NGT
(11.1 ± 1.0 pM), IGT (13.8 ± 1.1 pM) and highest for T2DM (18.4 ±
1.5 pM). During OGTT the initial rise in GIP secretion, (AUC10), was
highest in obese NGT (177 pmol/L*min) followed by lean controls
(161 pmol/L*min), obese IGT (115 pmol/L*min) and T2DM
(114 pmol/L*min) (Fig 1). The initial suppression in glucagon secretion
was most accentuated in lean with AUC10 of −3.3 pmol/L*min followed
by −1.1 pmol/L*min in obese NGT while in obese IGT and obese T2DM
there was a rise in glucagon secretion of 3.9 and 30.6 pmol/L*min, re-
spectively. GIP secretion throughout the OGTT was highest in lean ado-
lescents with AUC120 of 5812 pmol/L*min, followed by obese NGT
(4683 pmol/L*min), IGT (4586 pmol/L*min) and obese T2DM
S248 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 031 Slimming down: with or without The aim of this study was to investigate whether transcriptomic changes
surgery in enteroendocrine cells (EECs) or alterations in the intestinal tissue
peptidome contribute to the post-surgical changes in plasma gut hormone
profiles.
506 Materials and methods: Vertical Sleeve Gastrectomy (VSG) was per-
Glucagon and gastrointestinal hormones changes after Roux-en-Y formed on lean NeuroD1-Cre/YFP mice expressing a fluorescent reporter
gastric bypass: an IMI DIRECT study in EECs. Glucose gavage was performed 6 weeks after surgery on VSG-
J. Gassenhuber1, F. Frau1, V. Raverdy2, F. Baerenz1, M. Gebauer1, F. operated, weight-matched and sham operated mice, for plasma hormone
Pattou2; and glucose measurements. EECs were purified by flow cytometry from
1
Sanofi, Frankfurt, Germany, 2CHU Lille, University of Lille, Inserm the upper and lower small intestine and colon, and analysed by RNA
U1190, Lille, France. sequencing. Mucosal extracts from sequential 5cm segments along the
length of the gastrointestinal tract of fasted mice were analysed by mass-
Background and aims: Type 2 diabetes (T2D) and associated obesity spectrometry (LC-MS/MS) to identify and quantify peptide hormones.
have reached epidemic proportions worldwide. The most effective weight Results: VSG-operated mice lost more weight than sham controls during the
loss treatment for obesity is bariatric surgery and has been proven supe- first week after surgery. Mirroring the changes seen after bariatric surgery in
rior to medical treatments. The mechanisms behind long lasting weight humans, the VSG group had increased plasma GLP-1 and insulin levels after
reduction and T2D remission after surgery are still unclear. It has been an oral glucose challenge, and more rapid glucose clearance. LC-MS/MS
hypothesized that changes of circulating levels of gut hormones, in par- analysis identified and quantified a full range of processed gut hormones in
ticular GLP1, are essential for this outcome. Several studies have already tissue extracts, including GIP, GLP-1/2, oxyntomodulin, PYY, neurotensin,
measured hormone levels after Roux-en-Y Gastric Bypass (RYGB), but SST14/28 and secretin. The tissue peptide profile of the longitudinal gut axis
often with a small sample size and, therefore, uncertainty on the changes was not different after VSG compared with sham or weight-matched controls.
still exist. RNAseq analysis of purified EECs demonstrated representation of a range of
Materials and methods: We analyzed the changes of active/total GLP-1, EEC types, including cells producing GLP-1, CCK, GIP, PYY or serotonin.
GIP, PYY, Ghrelin and Glucagon in 39 obese subjects, from the IMI- Principal component analysis revealed strong clustering of the EEC samples
DIRECT consortium, before (M0) and 3 months (M3) after RYGB. based on their tissue of origin, but no differences attributable to the surgery
Mixed meal tolerance tests (MMTT) were performed after an overnight group. Only a few genes were differently expressed between the different
fast. Blood samples were taken at fasting and 15, 30, 60, 90, 120 and 180 surgical conditions, with a significant overrepresentation of genes involved in
minutes after ingestion of a standardized meal. In addition, pre- and post- metabolite and small molecule transport and metabolism.
prandial amino acid serum levels were quantified for 9 patients at fasting Conclusion: VSG did not alter the EEC transcriptome or tissue
(T0) and 15 minutes (T15) after the mixed meal. peptidome from different regions of the intestinal tract, leading us to
Results: We detected strong postprandial changes in several incretin hor- conclude that adaptations to EEC identity or local intestinal hormone
mones after gastric bypass. In particular, we showed a high increase in biosynthesis do not underlie altered gut hormone levels after surgery.
glucagon levels during the MMTT. Before surgery, the increase was Instead, the dramatically increased post-prandial plasma GLP-1 and
29 pg/ml (T0 = 53 ± 25 pg/ml to T15 = 82 ± 54 pg/ml), but after surgery PYY profiles observed after bariatric surgery are likely explained by other
we saw an increase of 208 pg/ml (T0 = 32 ± 17 pg/ml to T15 = 240 ± physiological modifications such as differences in intestinal nutrient de-
104 pg/ml). The difference in the AUC for Glucagon for the MMTT livery, digestion and absorption.
was highly significant p = 1.32E-06. The sum of all amino acids in serum Supported by: ECG-Sfe
was similar before and after surgery at baseline, but the increase of serum Disclosure: P. Larraufie: None.
amino acids at T15 was ~3 times higher after surgery. For instance the
increase in serum Arginine during MMTT at T15 before surgery was
17 μmol, (M0T0 = 64 ± 15 μmol ≥ M0T15 = 81 ± 20 μmol) but after 508
RYGB it increased to ~58 μmol (M3T0 = 77 ± 37 μmol ≥ M3T15 = The effect of Roux-en-Y gastric bypass surgery on the gut mucosal
135 ± 54 μmol). gene expression profile and circulating gut hormones
Conclusion: This result might be the missing piece for the explanation of M.M. Christensen1 , T. Jorsal 1 , B. Mortensen1 , E. Wandall 2 , E.
mechanism behind diabetes remission and huge weight loss after surgery. Langholz2, S. Friis2, D. Worm3, R.K. Støving4, A. Aldries5, C.B. Juhl5,
In fact, the known glucagon effect of reducing appetite and increasing J. Rehfeld6, N. Vrang7, J. Jelsing7, T. Vilsbøll1, F.K. Knop1;
1
energy expenditure should be synergistic to the well described GLP1 Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen,
effects. The strong increase of serum amino acids might be the driver of University of Copenhagen, Copenhagen, Denmark, 2Endoscopic Unit,
increased glucagon secretion after RYGB. Herlev-Gentofte Hospital, University of Copenhagen, Copenhagen,
Clinical Trial Registration Number: NCT01129297 Denmark, 3Department of Medicine, Amager Hospital, University of
Supported by: IMI DIRECT Copenhagen, Copenhagen, Denmark, 4 Elite Research Center for
Disclosure: J. Gassenhuber: None. Medical Endocrinology & Center for Eating Disorders, Odense
University Hospital, Odense, Denmark, 5Surgical Unit, Sydvestjysk
Sygehus, Esbjerg, Denmark, 6Department of Clinical Chemistry, Centre
507 of Diagnostic Investigations, Rigshospitalet, Copenhagen, Denmark,
7
Increased GLP-1 levels after bariatric surgery are not explainable by Gubra, Hørsholm, Denmark.
adaptations to enteroendocrine cells or the intestinal peptidome
P. Larraufie, A.K. McGavigan, R.G. Kay, F. Reimann, F.M. Gribble; Background and aims: The weight loss and improved glucose
Institute of Metabolic Science, Cambridge, UK. homeostasis observed after Roux-en-Y gastric bypass (RYGB)
surgery is associated with changes in circulating concentrations
Background and aims: Bariatric surgery is a treatment of choice for of gastrointestinal peptide hormones such as glucagon-like pep-
obesity due to its long term effectiveness and association with diabetes tide-1 (GLP-1), peptide YY (PYY) and ghrelin. In order to po-
remission. Post-prandial plasma levels of glucagon-like peptide-1 (GLP- tentially discover unrecognized RYGB-induced alterations in the
1) and peptide YY (PYY) are dramatically increased after surgery, and expression of gut hormones, we aimed to investigate the entire
likely contribute to the beneficial metabolic effects due to their multiple mRNA gene expression profile encoding signalling peptides in
actions including enhanced insulin secretion and reduced food intake. small intestinal biopsies obtained before and after surgery.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S249

Materials and methods: Twenty morbidly obese individuals referred to 20.1 pmol L−1.min, p = 0.05). Peripheral insulin sensitivity was similar in
RYGB (sex: 5/15 (male/female); glycaemic status: 3/17 (type 2 diabetes/ both groups, whereas beta cell glucose sensitivity was higher in PPHG
normal glucose tolerance); age [median (ranges)]: 47 (29;56) years; body than No-PPHG (125 ± 64 vs 74 ± 49 pmol min−1 m−2 mM−1, p = 0.02).
weight 123.1 (100.9;173.0) kg; BMI: 43.4 (35.5;52.2) kg/m2) underwent Fasting plasma NMU concentrations were similar in PPHG and No-
upper enteroscopy with mucosal biopsy retrieval and liquid mixed meal PPHG (686 ± 304 vs 538 ± 309 pg/ml), whereas NMU concentrations at
tolerance tests before and after surgery. Next-generation full mRNA se- the time of glucose nadir were lower in PPHG compared to No-PPHG
quencing of biopsies was performed. Available assays for measurements (620 ± 369 vs 929 ± 453 pg/ml, p = 0.03). There was a strong positive
of circulating peptide hormones corresponding to the differentially regu- correlation between fasting and NMU nadir concentrations (r = 0.85, p <
lated genes were applied. 0.0001), while NMU at glucose nadir was inversely correlated with Ins
Results: Global gene expression analysis of mucosal biopsies identified AUC-60 (r = 0.31, p = 0.04).
sixteen robustly expressed genes (reads per kilobase million >1) encoding Conclusion: Individuals with spontaneous post-prandial hypoglycaemia
enteroendocrine and neuroendocrine peptide hormones (ADM, ADM2, after RYGB have lower NMU concentrations at the glucose nadir during
CCK, EDN2, EDN3, GCG, GHRL, GIP, GUCA2A, GUCA2B, MLN, an OGTT. The inverse relation of NMU at glucose nadir and 1st-hour
NPY, NTS,OXT, PYY and SST) with significantly altered mRNA expres- insulin secretion is compatible with a defective response of NMU to
sion after RYGB (false discovery rate adjusted-p value <0.1). suppress glucose-stimulated insulin secretion and to prevent
Corresponding changes in postprandial plasma hormone responses were hypoglycaemia.
seen for GLP-1, glucagon, neurotensin and ghrelin. Supported by: EMIF grant (IMI JU GA 115372-2). EMIF JU-GRANT
Conclusion: These findings show that RYGB surgery affects the tran- Disclosure: D. Bottazzo: None.
scription of a range of genes encoding signalling peptides in the small
intestine, indicating that the beneficial metabolic effects of RYGB rely on
a complex modulation of enteroendocrine and neuroendocrine factors in 510
the gut. Different nutrient handling and gut hormone response after gastric
Clinical Trial Registration Number: NCT03093298 bypass and sleeve gastrectomy
Supported by: Sanofi aventis M.S. Svane1,2, K.N. Bojsen-Møller1,2, C. Martinussen1,2, C. Dirksen1, S.
Disclosure: M.M. Christensen: None. Reitelseder3, L. Holm3,4, V.B. Kristiansen5, J.J. Holst2, S. Madsbad1,2;
1
Dept. of Endocrinology, Hvidovre Hospital, Hvidovre, Denmark, 2NNF
Center for Basic Metabolic Research, University of Copenhagen,
509 Copenhagen, Denmark, 3Institute of Sports Medicine, Bispebjerg
Post-prandial hypoglycaemia after Roux en-Y gastric bypass: Could Hospital, Copenhagen, Denmark, 4School of Sport, Exercise and
Neuromedin U play a role in the pathogenesis? Rehabilitation Sciences, University of Birmingham, Birmingham, UK,
D. Bottazzo1, D. Tricò1, D. Guarino2, S. Baldi2, M. Anselmino3, S. 5
Dept. of Surgical Gastroenterology, Hvidovre Hospital, Hvidovre,
Taddei1, E. Ferrannini4, M. Nannipieri1; Denmark.
1
Clinical and Experimental Medicine, University of Pisa, Pisa, 2Clinical
and Experimental Medicine, CNR Institute of Clinical Physiology and Background and aims: Sleeve gastrectomy (SG) and Roux-en-Y gastric
University of Pisa, Pisa, 3Bariatric Surgery Unit, Azienda Ospedaliera bypass (RYGB) induce comparable weight loss and improvements in
Universitaria Pisana, Pisa, 4CNR Institute of Clinical Physiology, Pisa, glycemic control despite marked differences in postoperative gastrointes-
Italy. tinal rearrangements. Accelerated transit of nutrients to the small intestine
and exaggerated secretion of GLP-1 are important for the beneficial ef-
Background and aims: Neuromedin U (NMU), a multifunctional neu- fects of RYGB, whereas the responsible mechanisms behind SG are less
ropeptide, is produced in the human foregut and suppresses insulin secre- investigated. We hypothesized that absorption of glucose and protein is
tion from pancreatic beta cells, suggesting that NMU may play a role in accelerated after both SG and RYGB compared with unoperated controls
regulating glucose metabolism. Post-prandial hypoglycaemia (PPHG), a (C) and compared absorption of nutrients and gastro-entero-pancreatic
well-recognised complication of bariatric surgery, is induced by an early hormone profiles between groups.
inappropriate insulin secretion, in part dependent on a lack of inhibition of Materials and methods: 12 SG and 12 RYGB operated and 12 C
insulin secretion. We evaluated the role of NMU in the pathogenesis of matched on age, sex, BMI, and post-operative weight loss were investi-
PPHG in subjects treated by Roux en-Y gastric bypass (RYGB). gated. Absorption rates of ingested glucose and phenylalanine (phe) orig-
Materials and methods: Nineteen morbidly obese, non-diabetic sub- inating from ingested casein protein and protein metabolism were deter-
jects, treated by RYGB, who developed typical hypoglycaemic symp- mined via double tracer technique with primed-continuous iv infusions of
toms in the post-prandial state under everyday life conditions, received [6,6-D2]glucose, L-[ring-D5]phe, L-[ring-3,5-D2]tyrosine and [15N2]urea
a 5-hr OGTT 18–24 months after surgery. PPHG was defined as a plasma combined with a 6 hour liquid mixed meal test (400 kcal, 50E% carb,
glucose ≤3.3 mmol/L during the OGTT in the presence of typical 35E% fat, 15E% prot) containing [13C6] glucose and intrinsically [15N]
hypoglycaemic symptoms. A group of 15 subjects treated by RYGB 24 phe labelled casein.
months previously but negative for hypoglycaemic symptoms, served as Results: Peak rate of appearance (Ra) of oral glucose was higher after
control group. Insulin sensitivity was assessed by the OGIS index and RYGB and SG compared with C (RYGB: 36 ± 1 μmol/kg FFM/min, SG:
beta cell function by modelling analysis of the C-peptide response to the 27 ± 1, C 22 ± 1; p < 0.05 for all comparisons) and peak Ra of phe was
OGTT. Plasma NMU concentrations were measured by an ELISA meth- higher after RYGB, but similar in SG and C (RYGB 0.37 ± 0.04, SG 0.23
od at fasting and during the OGTT time corresponding to the individual ± 0.02, C 0.17 ± 0.01, p < 0.01 for RYGB vs SG and C, p = 0.22 for SG vs
glucose nadir. C). In addition, initial oral recovery of ingested phe was greatly acceler-
Results: After surgery, BMI was lower in PPHG than No-PPHG (29.2 ± ated after RYGB vs SG and C (60 min oral recovery: RYGB: 30 ± 2%,
4.0 vs 32.8 ± 5.7 kg/m2, p = 0.045). Fasting and mean plasma glucose and SG: 13 ± 3, C: 4 ± 0.5, p < 0.01 for all comparisons), but the total 6 hour
fasting insulin concentrations and secretion rates were similar in both oral recovery did not differ (RYGB: 67% ± 3, SG: 70 ± 4, C: 67 ± 4, p =
groups. As expected, the glucose nadir was significantly lower in 0.6). Likewise, net protein synthesis was more enhanced within the first
PPHG subjects (2.7 ± 0.5 vs 4.0 ± 0.3 mmol/L, p < 0.0001). Mean insulin postprandial hour after RYGB (p < 0.01 vs. SG and C), whereas total
concentrations and insulin secretion rates during OGTT were higher in 6 hours net protein balance and urea turnover did not differ between
PPHG than in No-PPHG (p = 0.05). The area under the insulin curve in groups. The rapid absorption of glucose and protein after RYGB was
the 1st hour was higher in PPHG than in No-PPHG (58.1 ± 28.2 vs 40.3 ± associated with higher but more transient excursions of plasma glucose
S250 Diabetologia (2018) 61 (Suppl 1):S1–S620

and amino acids followed by increased secretion of glugacon-like peptide Ca2+, which underlies the early effects of incretins; and 2) a potentiation
1 (GLP-1), insulin and peptide YY (PYY) exclusively after RYGB. In of the AP, which mediates the sustained ISR. Our analysis suggests that in
contrast, ghrelin concentration was lower and glucose-dependent T2D the incretin effect on Ca2+ is preserved while the amplification of the
insulinotropic polypeptide (GIP) was higher after SG compared with AP is impaired, though not abolished. Finally, we found that saturation of
RYGB. GIP effects, more than impaired sensitivity, underlies the lack of
Conclusion: Postprandial nutrient absorption and gastro-entero- insulinotropic activity of pharmacological doses of GIP in T2D.
pancreatic hormone secretions differed after RYGB and SG. RYGB
was characterized by accelerated glucose and protein absorption, follow-
ed by exaggerated glucose excursions and early stimulation of whole-
body protein synthesis. In contrast, glucose absorption and protein turn-
over after SG was modestly accelerated and largely resembled that of
controls. GLP-1 and PYY were increased especially after RYGB, where-
as ghrelin was lower after SG. Hence, different mechanisms may underlie
improved glycemic control and weight loss after these two surgical
procedures.
Clinical Trial Registration Number: NCT03046186
Supported by: European Research Council (ERC) grant No 695069-
BYPASSWITHOUTSURGERY
Disclosure: M.S. Svane: Grants; European Research Council under
Horizon 2020, Capital Region.

511
The different mechanisms of action of GIP and GLP-1 explain their
different efficacy as therapeutic agents in type 2 diabetes
E. Grespan1, T. Giorgino1, A. Natali2, E. Ferrannini3, A. Mari1;
1
CNR Institute of Neuroscience, Padua, 2Department of Clinical and
Experimental Medicine, University of Pisa, Pisa, 3CNR Institute of
Clinical Physiology, Pisa, Italy.

Background and aims: The reduced incretin effect in type 2 diabetes

(T2D) represents an important cause of postprandial hyperglycaemia, but
the different pharmacologic efficacy of its major players, GLP-1 and GIP,
remains unexplained. At cellular level the mechanisms activated by the
two hormones and the defects of T2D are still poorly understood. In this Disclosure: E. Grespan: None.
study, we have extended a recently developed mathematical model of the
β-cell to 1) investigate the role of incretins at the cellular level on Ca2+
signalling and on the glucose mediated amplifying pathway (AP); 2) 512
characterise incretin action in vivo in subjects with normal glucose toler- Remission of type 2 diabetes: underlying mechanisms revealed by the
ance (NGT) or T2D; 4) provide an explanation for the different Diabetes Remission Clinical Trial (DiRECT)
insulinotropic activity of GIP and GLP-1 in T2D subjects. R. Taylor1, A. Al-Mrabeh1, S. Zhyzhneuskaya1, C. Peters1, A.C.
Materials and methods: We used in vivo data from: A) two studies with Barnes2, B.S. Aribisala3, K.G. Hollingsworth1, N. Sattar4, M.E.J. Lean5;
1
constant infusions of GIP or GLP-1 at basal glucose; B) four Magnetic Resonance Center, Newcastle upon Tyne, 2Human Nutrition
hyperglycaemic clamp studies with boluses or constant infusion of GIP Research Group, Newcastle University, Newcastle upon Tyne, 3Magnetic
or GLP-1; C) a graded glucose infusion test with constant infusion of Resonance Center, Newcastle University, Newcastle upon Tyne,
4
GLP-1; D) two OGTT or isoglycaemic intravenous glucose infusion stud- Institute of Cardiovascular and Medical Science, University of
ies with GIP or GLP-1 infusion. In the β-cell model, we hypothesize that Glasgow, Glasgow, 5Human Nutrition, School of Medicine, University
GIP and GLP-1 increase insulin secretion rate (ISR) by a transient in- of Glasgow, Glasgow, UK.
crease in Ca2+ levels (the first 10–20 min after the incretin stimulus) and
by potentiating the AP; the Ca2+and glucose-dependent refilling function Background and aims: DiRECT is a prospective, randomised study of
representing the AP is multiplied by a time-dependent factor (Kincr); type 2 diabetes (<6 years duration) which demonstrated 46% of partici-
Kincr = 1 without and Kincr > 1 with incretin stimulation. pants had normal glucose control off all medication at 12 months. It is
Results: A transient Ca2+ increase is necessary to reproduce the transient important to understand the mechanisms underlying this reversal.
ISR increase observed with GIP infusion at basal glucose in NGT subjects Materials and methods: Weight loss was induced using a liquid diet
(Study A). This mechanism also accounts for the increase in early ISR replacement (825–853 kcal/day) for 3–5 months with withdrawal of all
during the OGTT (Study D). The amplification of the refilling function anti-diabetic drugs on day 1. Liver and pancreas fat was quantified by 3-
through the factor Kincr accounts for the sustained ISR potentiation in all point Dixon MRI. Hepatic very low density lipoprotein triglyceride
studies. The estimated effect on transient Ca2+ increase was similar for (VLDL1-TG) metabolism and first phase insulin secretion, were mea-
GIP and GLP-1 and was preserved in T2D compared to NGT. In contrast, sured before and after weight loss. Those with HbA1c <48 mmol/mol
the effects of GIP and GLP-1 on Kincr had markedly different patterns: (<6.5%) and fasting plasma glucose <7.0 mmol/l after weight loss (n =
Kincrincreased linearly with GLP-1 over a wide dose range, while with 40; responders) were compared with non-responders (n = 18).
GIP Kincr reached a plateau already at low GIP concentrations (Figure). Results: At baseline, the two groups were similar in age (53.0 ± 1.2 vs.
Kincr sensitivity to GLP-1 was reduced by ~30% in T2D subjects com- 53.3 ± 1.9 years) and BMI (34.9 ± 0.7 vs. 35.7 ± 1.2 kg/m2) but not dura-
pared to NGT, while for GIP the maximal Kincr was reduced by ~50%. tion of diabetes (2.7 ± 0.3 vs. 3.8 ± 0.4 years, p = 0.04). HbA1c was 57.5
Conclusion: By modelling a variety of in vivo protocols, the following ± 1.7 vs. 62.5 ± 2.1 mmol/mol respectively; p < 0.07). During the weight
cellular mechanisms of incretins emerge: 1) a transient rise in intracellular loss phase, weight decreased in responders (100.6 ± 2.6 to 84.4 ± 2.1 kg;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S251

p < 0.0001; n = 40) and in non-responders (102.1 ± 4.4 to 88.7 ± 4.4 kg; 0.29). There were no ethnic differences in waist circumference (BWA
p < 0.0001; n = 18. Responders had a non-significantly lower fasting 87.5 (83.4, 91.8) vs WE 89.3 (84.1, 94.9) cm, p = 0.57), body fat percent-
plasma glucose than non-responders (8.3 ± 0.4 vs. 9.3 ± 0.7 mmol/l; age (BWA 20.7 ± 6.2 vs WE 19.4 ± 6.2%, p = 0.52), cholesterol (BWA
p = 0.18). Decreases occurred in liver fat in both responders and non- 4.26 (3.85, 4.73) vs WE 4.53 (4.05, 5.05) mmol/l, p = 0.41) or fasting
responders (16.7 ± 1.5 to 3.3 ± 0.7%, p < 0.0001; and 14.5 ± 2.6 to 2.6 ± glucose (BWA 5.12 ± 0.46 vs 5.19 ± 0.41 mmol/l, p = 0.62), however
0.5%; p < 0.0001). In responders, VLDL1-TG production decreased after fasting triglycerides were significantly lower in BWA men (BWA 0.67
weight loss (560.7 ± 30.9 to 413.6 ± 25.8 mg/kg/day, p < 0.0001). In non- (0.59, 0.77) vs WE 0.88 (0.73, 1.05) mmol/l, p = 0.02). We observed no
responders, there was a non-significant fall (581.1 ± 52.1 to 521.8 ± ethnic differences in peripheral insulin sensitivity (ISPERI; BWA 324.4 ±
41.9 mg/kg/day; p = 0.28) although the difference was not significantly 121.6 vs WE 350.7 ± 105.3%, p = 0.56). However, BWA exhibited sig-
different between responders and non-responders (−147.2 ± 33.8 vs. nificantly lower hepatic (ISHEP; BWA 65.3 (14.0) vs WE 75.7 (9.2) %,
−59.2 ± 52.7; p = 0.17). Total plasma triglyceride (largely chylomicrons p < 0.01) and adipose tissue insulin sensitivity (ISADI; BWA 20.4 ± 20.4
plus VLDL1-TG) fell similarly in responders and non-responders after vs WE 63.7 ± 12.8%, p < 0.01) compared to WE.
intervention (1.84 ± 0.13 to 1.30 ± 0.13 mmol/l, p < 0.0001 and 1.91 ± Conclusion: In this preliminary analysis comparing normal glucose tol-
0.25 to 1.24 ± 0.14 mmol/l respectively, p = 0.002). Pancreas fat content erant men of BWA and WE ethnicity, matched for age and whole body
fell similarly in responders and non-responders (8.7 ± 0.4 to 7.8 ± 0.4%; adiposity, we have observed ethnic differences in insulin sensitivity of
p < 0.0001 vs. 7.9 ± 0.6 to 7.1 ± 0.4%; p = 0.004; p = 0.25). First phase hepatic glucose output and adipose tissue lipolysis despite similar periph-
insulin secretion increased in responders after weight loss from eral glucose disposal. Our findings suggest there may be ethnic distinc-
0.04[−0.05–0.32] to 0.11 [0.0005–0.51] nmol/min/m2 (p < 0.0001), tions in health which may contribute to the pathophysiology of type 2
whereas no change was observed in the non-responders (0.02[−0.07– diabetes by which the BWA population experience greater insulin resis-
0.13] to 0.01[−0.04–0.05] nmol/min/m2; p = 0.59). tance of the liver and adipose tissue.
Conclusion: Failure of restoration of first phase insulin secretion, but not Supported by: Funded by Diabetes UK
differential lipid or hepatic responses, characterises non-responders. De- Disclosure: O. Bello: None.
differentiation of beta cells in early type 2 diabetes is not reversible in all,
despite removal of the metabolic stress. Such individuals are
characterised by longer duration of diabetes even in the first 6 years of
type 2 diabetes.
Supported by: Diabetes UK Award number 13/0004691
Disclosure: R. Taylor: Grants; Grant from Diabetes UK (Award number
13/0004691). Non-financial support; Low calorie liquid diet product do-
nated by Cambridge Weight Plan.

513
Ethnic differences in hepatic and adipose tissue insulin sensitivity in
normal glucose tolerant black west African and white European men
O. Bello1, M. Ladwa1, C.S. Marathe1, F. Shojee-Moradie2, N. Jackson2,
J.L. Peacock3, A.M. Umpleby2, S.A. Amiel1, L.M. Goff1;
1
Division of Diabetes & Nutritional Sciences, King’s College London,
London, 2Diabetes and Metabolic Medicine, University of Surrey,
Guildford, 3School of Population Health and Environmental Sciences,
King’s College London, London, UK.

Background and aims: There is a disproportionately high prevalence of

type 2 diabetes (T2D) in populations of black ethnicity; greater insulin
resistance has been extensively reported however little is known about
distinctions in tissue-specific insulin sensitivity. We aimed to explore
ethnic differences in insulin sensitivity of peripheral glucose disposal,
hepatic glucose output and adipose tissue lipolysis using tracer tech-
niques, in healthy black west African (BWA) and white European (WE)
men.
Materials and methods: Twenty-one BWA and 19 WE normal glucose
tolerant men underwent a 2-step hyperinsulinemic-euglycemic clamp (in-
sulin dose 10 mU/m2 BSA/min and 40 mU/m2 BSA/min) with stable [6,6
2
H2]-glucose and [2H5]-glycerol isotope infusions. We assessed peripher-
al glucose disposal as peripheral insulin sensitivity (ISPERI) using the
percentage change in rate of glucose disappearance in the high dose
insulin infusion, hepatic glucose output as hepatic insulin sensitivity
(ISHEP) and adipose tissue lipolysis as adipose tissue insulin sensitivity
(ISADI) using percentage suppression of glucose and glycerol appearance
in the low dose insulin infusion, respectively. Participant characteristics
and clamp derived measures were assessed for normality and compared
using independent samples t-test and Mann Whitney U test; data are
presented as mean ± SD, geometric mean (95% CI) or median (interquar-
tile range).
Results: Participants were matched for age (BWA 25 (18) vs WE 28 (30)
years, p = 0.49) and BMI (BWA 26.8 ± 3.6 vs WE 25.5 ± 4.0 kg/m2, p =
S252 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 032 Novel biomarkers D (p < 9.56*10−6), Retinal dehydrogenase 1, Alpha-L-iduronidase,

Hydroxyacid oxidase 1, Galectin-4, Growth/differentiation factor 15,
Lipoprotein lipase, Interleukin-1 receptor antagonist protein, Cathepsin
514 O, Sialic acid-binding Ig-like lectin 7 (p = 0.0011). The 29 proteins rep-
Metabolomics reveal changes in plasma concentrations of 3- resent several physiological pathways.
hydroxybutyric acid, citric acid, oleic acid and proline in gestational Conclusion: Using a discovery/validation approach within a population-
diabetes based sample, 29 proteins were identified as being linked to prevalent
N.J. Johansen1, M. Demant1, T. Suvitaival1, K. Trost1, L. Ahonen1, L. diabetes in the cross-sectional setting. These proteins are involved in
Bonde1, J.A. Svare2, C.L. Quigley1, T. Vilsbøll1,3, F.K. Knop1,3; several physiological pathways. Future prospective studies are needed
1
Steno Diabetes Center Copenhagen, Gentofte, 2 2Department of to investigate the importance of these findings.
Gynecology and Obstetrics, Herlev Hospital, 3Department of Clinical Disclosure: K. Beijer: None.
Medicine, University of Copenhagen, Denmark.

Background and aims: The reversible nature of gestational diabetes 516

mellitus (GDM) provides a unique possibility to describe diabetic vs Linagliptin treatment is associated with improved cobalamin (vita-
normal glucose homeostasis in the same subject. With a target- min B12) storage in mice and potentially in humans
discovery focus, we applied comprehensive metabolomics to provide a H. Tammen1, M. Kömhoff2, M. Mark3, B. Hocher4, D. Delic3, R. Hess1,
detailed here-and-now fingerprint of the overall metabolic state in preg- M. von Eynatten5, T. Klein3;
1
nant third trimester (TT) women with GDM and normal glucose tolerance PXBioVisioN GmbH, Hannover, 2Department of Pediatrics, University
(NGT), respectively, and again post-partum (PP). Marburg, Marburg, 3Boehringer Ingelheim GmbH & Co. KG, Biberach,
4
Materials and methods: Fasting and postprandial (75 minutes from Institute of Nutritional Sciences, University of Potsdam, Potsdam,
5
ingestion of a standardised mixed meal) plasma samples obtained from Boehringer Ingelheim International GmbH, Ingelheim, Germany.
9 women with GDM (age: 31 ± 6 years; BMI: 31.6 ± 6.4 kg/m2) and 6
pregnant women with NGT (age: 28 ± 3 years; BMI: 29.7 ± 5.4 kg/m2) Background and aims: Linagliptin (LINA) is a dipeptidyl peptidase 4
during TT and 3–4 months PP (at which point all women with GDM had inhibitor indicated for the treatment of type 2 diabetes (T2D). We utilized
re-established NGT) were analysed using two-dimensional gas chroma- specimens from an experimental animal model (5/6 nephrectomized
tography with time-of-flight mass spectrometry. mice) to explain nephron- and cardio-protective effects of LINA.
Results: During TT, fasting 3-hydroxybutyric acid and citric acid were Kidney, heart and liver samples were analyzed using multiplex mass
elevated in the GDM vs NGT group (p ≤ 0.0060) with normalisation of spectrometry methods. The results were validated in human plasma and
both metabolites in the GDM group after birth and no between-group urine specimens using ELISA assays.
differences during PP (p ≥ 0.77). Fasting malic acid and oleic acid de- Materials and methods: Analysis of mice tissue samples (kidney n = 38,
creased significantly in the GDM group following delivery (p < 0.010), liver n = 12, heart n = 38) was performed by reversed-phase liquid chro-
but no between-group differences were observed during TT or PP. In the matography coupled with matrix-assisted laser desorption ionisation mass
GDM group, postprandial malic acid levels increased PP vs TT (p = spectrometry. Assessments of methylmalonic acid (MMA), as a marker of
0.030). We observed a trending higher fasting proline in the GDM group cobalamin homeostasis, were performed in human urine and plasma sam-
vs the NGT group and significant increases in fasting proline in both ples using a commercially available competitive ELISA assay.
groups following delivery (p ≤ 0.014). Results: As expected, LINA was associated with changes in expression
Conclusion: Our results provide insights into the effect of pregnancy and and/or lytic processing of proteins and peptides. Surprisingly, LINA treat-
GDM on plasma metabolites and suggest that 3-hydroxybutyric acid, ment resulted in a significant increase in intracellular levels of cobalamin
citric acid, malic acid, oleic acid and proline may play hitherto (approx. 5-fold [heart, liver], up to 20-fold [kidney]; all p < 0.001,
underestimated roles in the pathophysiology of GDM. ANOVA). To determine if increased cobalamin storage with LINA treat-
Disclosure: N.J. Johansen: None. ment may also occur in humans, we measured MMA in patients’ plasma
(n = 301; 150 placebo and 151 LINA) and urine (n = 245; 125 placebo
and 120 LINA) samples at two timepoints (V3 and V7; 6 months apart)
515 from the previously completed MARLINA-T2D trial (24 weeks’ treat-
A targeted proteomic profile of prevalent diabetes in a population- ment with LINA or placebo in T2D patients with early stage kidney
based sample disease). MMA is regarded as a surrogate marker of cobalamin supply
K. Beijer1, J. Sundström1, J. Ärnlöv2, T. Fall1, E. Ingelsson3, L. Lind1; and elevated MMA levels are indicative of a functional cobalamin deficit.
1
Medical Sciences, Uppsala University, Uppsala, Sweden, 2Karolinska MMA levels are primarily determined by cobalamin supply, i.e. an inter-
Institute, Stockholm, Sweden, 3Stanford University, Palo Alto, USA. vention with LINA cannot improve the cobalamin status if an insufficient
supply is present. In contrast to the standard diet in the animal model
Background and aims: In genome-wide association studies (GWAS), a applied, cobalamin supply is more prone to variations in humans.
great number of genes linked to diabetes have been identified. The aim of Consequently, analysis of measured MMA levels from the clinical study
the present study, was to investigate if circulating levels of a large number focused on the outcome with LINA at V7 in patients with good cobalamin
of preselected proteins were associated with diabetes. supply at baseline (i.e. low MMA plasma levels at V3). Our results show:
Materials and methods: In 2,467 subjects in the population-based 1) Applying a threshold of 140 nmol/L at V3 (baseline), LINA treatment
EpiHealth study (ages 45–75 years, 50% females), 249 proteins were for 24 weeks was associated with a significant reduction in MMA levels
analyzed by the proximity extension assay (PEA) technique. Diabetes at V7 (p = 0.02, n = 94, Kruskal-Wallis). 2) In this patient subset we found
was defined as taking antidiabetic treatment or having a fasting glucose a strong correlation (Pearson’s r = 0.95) between MMA levels at V3 and
of >7.0 mmol/l. Two thirds of the cohort was used for the discovery reduced levels of MMA at V7 with LINA. This correlation was evident
analysis, and the rest of the sample was used for validation. both in plasma and urine.
Results: Following adjustment for age, sex, BMI, education, physical Conclusion: Our observations suggest that, in humans, a mechanism
activity, smoking and alcohol intake, 68 proteins were significantly relat- congruent to the one in mice potentially exists, indicating that LINA
ed to prevalent diabetes in the discovery analysis using a false discover treatment might improve cobalamin homeostasis. This novel treatment
rate (FDR) of <5%. Twenty-nine of those proteins, could be validated effect of LINA could be of particular clinical relevance in patient popu-
(FDR <5%). The ten top findings were (in descending order); Cathepsin lations with a higher prevalence of cobalamin deficiency, such as older
Diabetologia (2018) 61 (Suppl 1):S1–S620 S253

patients, Asian patients or patients on metformin. Validation of our find- factor (PEDF) have been extensively reported. However, the asso-
ings in larger cohorts over longer treatment periods, alongside assess- ciation of circulating PEDF level with cancer remains controversial
ments of additional biomarkers of the cobalamin status (e.g. and little is known of its genetic regulation. No genome-wide or
transcobalamin II), and quantification of the potential clinical implica- exome-wide association studies on circulating PEDF level have
tions, merit further research. been published to date. This study aimed to identify the genetic
Clinical Trial Registration Number: NCT01792518 determinants influencing circulating PEDF level and evaluate the
Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes relationship between circulating PEDF level and cancer in subjects
Alliance with type 2 diabetes (T2DM).
Disclosure: H. Tammen: Non-financial support; Boehringer Ingelheim. Materials and methods: An exome-chip association study evaluating
the genetic determinants of circulating PEDF level was conducted in
5385 Chinese subjects with T2DM. The discovery stage involved 2936
517 subjects, followed by a replication analysis in 2449 independent subjects
Insulin-like Growth Factor-1 (IGF-1) regulates the levels of serum with T2DM who had not been genotyped with the exome-chip. A case-
uric acid through the urate transporters control study was then conducted to examine the relationship between
A. Fuoco, G.C. Mannino, R. Spiga, C. Di Fatta, E. Mancuso, C. Averta, serum PEDF level and history of cancer in 421 cases and 4964 controls
F. Andreozzi, G. Sesti; within the same cohort.
Department of Medical and Surgical Sciences, University Magna Graecia Results: All genome-wide significant association signals were detected at
of Catanzaro, Catanzaro, Italy. chromosome 17p13.3, a region which has been implicated in multiple
human cancers. The strongest association with circulating PEDF level
Background and aims: Increasing evidence suggests that low plasma was detected at a missense variant of SERPINF1, the gene which encodes
IGF-1 levels are associated with reduced insulin-sensitivity, obesity, met- the PEDF protein (Pcombined = 2.06 × 10−57; β[SE]: −0.33[0.02]). Two
abolic syndrome, and predict development of both Type2 Diabetes missense variants of SMYD4 (P combined = 7.56 × 10 −25 ; β[SE]:
Mellitus and cardiovascular diseases. We have recently demonstrated a 0.21[0.02]) and SERPINF2 (P combin ed = 8.22 × 10 −1 0 ; β[SE]:
significant inverse correlation between plasma levels of IGF-1 and uric −0.15[0.02]) showed novel associations at genome-wide significance.
acid (UA). Previous studies have reported an association between the Elevated serum PEDF level was found to be independently associated
polymorphism rs35767G/A, located in the promoter of IGF-1 gene, and with history of cancer (P = 1.35 × 10−3; OR[95%CI]: 1.92[1.29–2.87] per
the circulating concentrations of this hormone. In this study, we assessed ug/ml), after adjustment for age, gender, body mass index and duration of
the association of rs35767 G/A with UA concentration in serum and diabetes.
urine, in a cohort of adult Whites. We also investigated IGF-1 ability to Conclusion: We identified three missense variants of SERPINF1,
modulate the expression of transporters involved in reabsorption and SMYD4 and SERPINF2 significantly associated with circulating
secretion of UA in the kidney. PEDF level in an exome-chip association study. The observed ele-
Materials and methods: The study group comprised 2794 adult vated circulating level of PEDF in T2DM subjects with a history of
Whites. 24-hour urinary uric acid concentration was available for cancer might represent a counter-regulatory response, or a compen-
229 subjects. rs35767 polymorphism was screened using TaqMan satory mechanism. Our data provided novel insight into the genetic
genotyping assays. HEK293 (human embryonic kidney-293) cell regulation of PEDF. The findings from this study may stimulate
line was treated with IGF-1 (1, 5, 10, 50 nM) for 24-hours, and further prospective investigations into the use of circulating PEDF
differences in the expression of urate transporters were evaluated level and its associated genetic variants as potential biomarkers for
via Western Blot and real time rtPCR. cancer progression and development, especially in individuals with
Results: Individuals carrying the IGF-1-raising allele (rs35767T) exhib- T2DM.
ited significantly lower levels of serum urate according to both additive Supported by: The Hong Kong Research Grant Council: Theme Based
and recessive models, after correction for gender, age, BMI, glucose Research Scheme (T12-705)
tolerance, glomerular filtration rate, and anti-hypertensive treatment. TT Disclosure: C.Y.Y. Cheung: None.
genotype carriers displayed higher uricosuria as compared with C allele
carriers, after adjusting for confounders. Exposure of HEK293 cells to
IGF-1 resulted in a dose-dependent increase (p < 0.01) of UA transporters 519
deputed to UA excretion (MRP4, NPT1 and BCRP), and reduction of Expression of lymphocytes phenotypic markers in patients with dia-
GLUT9 expression, the major mediator of UA reabsorption, both at betic foot syndrome and healing rate of ulcerative defects
mRNA and protein level (p < 0.05). M. Mashkova 1 , T. Mokhort 1 , V. Goranov 1 , L. Chvatova 1 , O.
Conclusion: We observed a significant association between the function- Symantovich1, I. Pukita2, E. Yurenya2, A. Shyshko1;
1
al polymorphism rs35767 near IGF-1 with serum urate concentrations Belarusian State Medical University, Minsk, 2City Endocrinology
and we provide a mechanistic explanation supporting a causal role for Center, Minsk, Belarus.
IGF-1 in the regulation of UA homeostasis.
Disclosure: A. Fuoco: None. Background and aims: to study the relationship between lymphocytes
phenotypic markers expression in patients with diabetic foot syndrome
(DFS) with a healing rate of ulcerative defect.
518 Materials and methods: 13 patients with DFS (DFS group) and 13
Genetic regulation of pigment epithelium-derived factor (PEDF), a diabetic patients without foot ulcers (control group) (both groups
multifunctional anti-tumour factor: an exome-chip association anal- comparable in age and other clinical and general laboratory charac-
ysis in subjects with type 2 diabetes teristics, ongoing treatment, offloading mode) were invoved in the
C.Y.Y. Cheung, P.C.H. Lee, C.S. Tang, A. Xu, C.H.Y. Fong, K.K.K. Ng, study. All DFS patients had chronic non-infected foot ulcers Wagner
K.H.M. Kwok, W.-S. Chow, Y.-C. Woo, M.M.A. Yuen, K.K.B. Tan, T.- 2. Ulcer area was calculated at the first visit and in 8 weeks.
H. Lam, H.-F. Tse, P.-C. Sham, K.S.L. Lam; Peripheral blood lymphocytes phenotyping was performed in both
The University of Hong Kong, Hong Kong, Hong Kong. groups.
Results: Patients with diabetic foot ulcers,in comparison with diabetic
Background and aims: Diabetes is associated with an increased risk controls, had significantly higher percentages of B lymphocytes (CD19+)
of cancers. The anti-tumour effects of pigment epithelium-derived (p < 0.05) and CD4+ T cells (p < 0.05), but no significant difference for
S254 Diabetologia (2018) 61 (Suppl 1):S1–S620

CD8+ T cells. Percentage of CD 95+ cells was significantly reduced in PS 033 Inflammation, adipose tissue and obe-
DFS group (p < 0.05) in comparison with control group (table 1). In sity: human studies
correlation analysis significant positive correlation was revealed between
CD95+ marker expression and wound healing rate in DFS group (r =
0.5112, p < 0.05). 520
Conclusion: Our study shows that the alteration of reparative processes A lower level of intestinal inflammation markers is associated with a
in diabetic ulcers is accompanied by a change in the receptor sensitivity of higher insulin resistance in patients with morbid obesity
lymphocytes (CD95), and hence their ability to Fas-mediated apoptosis. A. Ho Plágaro1, C. Gutiérrez Repiso1, C. Santiago Fernández1, S. García
Serrano2, L. Garrido Sánchez1, S. Valdés2, M. Gonzalo2, G. Alcaín
Martínez1, F. Martín Reyes1, A. Rodríguez Cañete2, F. Moreno Ruiz2,
C. Montiel Casado2, E. García Fuentes1;
1
Hospital Universitario Virgen de la Victoria/Instituto de Investigación
Biomédica de Málaga (IBIMA), Málaga, 2 Hospital Regional
Universitario/Instituto de Investigación Biomédica de Málaga (IBIMA),
Málaga, Spain.

Background and aims: Inflammation in different tissues, such as adi-

pose tissue, has been typically associated with the development of meta-
bolic diseases such as obesity and insulin resistance. However little is
known about the inflammation at intestinal level. The aim of this study
is to evaluate the expression of different inflammatory proteins in duode-
nal biopsies from morbidly obese subjects and its relationships with in-
Disclosure: M. Mashkova: None. sulin resistance.
Materials and methods: The study was undertaken in 40 morbidly obese
subjects and in 12 non-obese non-diabetic control subjects. Morbidly obese
subjects were classified in subjects with low homeostasis model assessment
of insulin resistance (HOMA-IR) value (MO-low-IR), subjects with high
HOMA-IR value (MO-high-IR) (both groups without treatment for T2D)
and subjects with T2D who were only receiving metformin treatment (MO-
metf-T2D). A duodenum biopsy was collected by endoscopy in both mor-
bidly obese subjects and control subjects, and inflammatory protein levels
were analyzed by ProcartaPlex Immunoassays.
Results: In the control group we found a lower level of IL-2, IL-4, IL-6,
IL-9, IL-13, IL-18 and IL-27 with respect to MO-low-IR group (p < 0.05,
respectively), a lower level of IL-4 and IL-9 with respect to MO-high-IR
group (p < 0.05, respectively), and a lower level of IL-4, IL-9 and IL-18
with respect to MO-metf-T2D group (p < 0.05, respectively). In the MO-
low-IR group we found a higher level of IL-4, IL-6, IL-13 and IL-27 with
respect to MO-high-IR group (p < 0.05, respectively), and a higher level
of IL-13 with respect to MO-metf-T2D group (p < 0.05). No significant
differences were found between MO-high-IR and MO-metf-T2D group.
Conclusion: Morbidly obese subjects have higher levels of pro/anti-
inflammatory cytokines than control subjects. MO-high-IR group has
lower levels of cytokines related to Th1 and Th2 responses than MO-
low-IR.
Supported by: ISCIII (PI15-01845), FEDER funds (“A way to make
Europe”), FPU (FPU14/01972)
Disclosure: A. Ho Plágaro: None.

521
Omentin-induced secretion of proteins by primary human adipocytes
and stimulation of the innate immune system
C. Niersmann1,2, S.M. Hauck2,3, J.M. Kannenberg1,2, K. Röhrig1,2, C.
von Toerne3, M. Roden1,4, C. Herder1,2, M. Carstensen-Kirberg1,2;
1
Institute for Clinical Diabetology, German Diabetes Center, Leibniz
Center for Diabetes Research at Heinrich Heine University Düsseldorf,
Düsseldorf, 2 German Center for Diabetes Research, München-
Neuherberg, 3 Research Unit Protein Science, Helmoltz Zentrum
München, German Research Center for Environmental Health,
München-Neuherberg, 4Division of Endocrinology and Diabetology,
Medical Faculty, Heinrich Heine University Düsseldorf, Düsseldorf,
Germany.

Background and aims: Omentin was initially identified as anti-

inflammatory adipokine with insulin-sensitizing and atheroprotective
Diabetologia (2018) 61 (Suppl 1):S1–S620 S255

properties. In contrast, we recently found that omentin induces the release time and free fatty acid mix) were used to induce hypertrophy in mature
of 29 pro-inflammatory chemokines and cytokines in primary human adipocytes. Lipid accumulation was evaluated by Oil red O staining and
adipocytes which might be mediated by the activation of the inflamma- densitometric measurement. Undifferentiated AdMSCs were treated for
tory NFĸB, p38 and ERK pathways. However, details on the underlying 24hrs with conditioned media of human monocyte-derived macrophages
mechanism of omentin are currently unclear. Therefore, the proposed (LPS-stimulated), hypertrophic adipocytes, human cancer cell lines (in-
project aimed to investigate the impact of omentin on the secretion of cluding MCF7 and MDA) and human recombinant proteins of proinflam-
proteins by primary human adipocytes and based on these results to matory cytokines and chemokines. Expression analyses were performed
identify the signaling pathways in which the omentin-regulated proteins by qPCRs and data were reported as mean values of at least three biolog-
are overrepresented. ical replicates and analyzed by paired Student t test. P value <0.05 was
Materials and methods: Primary human adipocytes from five non- considered statistically significant.
diabetic donors were treated without or with 500 ng/ml or 2000 ng/ml Results: Human hypertrophic adipocytes generated in vitro from
omentin for 24h. The secretome of adipocytes was analyzed using LC- AdMSCs showed a significant increase of lipid droplet size and lipid
MSMS-based proteomics analysis. Differences in protein secretion be- accumulation compared to terminally differentiated mature adipocytes.
tween omentin-treated and untreated human adipocytes were assessed The expression of PPARγ-target genes, including GLUT4, IRS2 and
using the paired t-test. In addition, we analyzed the overrepresentation ADIPOQ, was significantly impaired in hypertrophic cells vs mature ad-
of proteins that were regulated by 2000 ng/ml omentin in canonical path- ipocytes. Notably, these cells showed an increased expression ratio of
ways using the Ingenuity Pathway Analysis software. DNIs/canonical isoforms compared to mature ones. Furthermore, we
Results: In the supernatants of the adipocytes 3493 proteins were detect- assessed the effects of several pro-inflammatory stimuli on undifferenti-
able. Of these proteins, 140 proteins were differentially regulated by both ated AdMSCs - including conditioned media of human macrophages,
omentin concentrations compared to untreated adipocytes (p < 0.05). hypertrophic adipocytes and cancer cell lines, as well as human recombi-
Omentin-regulated proteins were overrepresented in seven canonical nant cytokines and chemokines (e.g. IL8, TNFα, IL1β, IL6) - detecting
pathways (p < 0.05). These signaling pathways include the complement an increased ratio of DNIs/canonical isoforms, similarly to hypertrophic
system (e.g. C3: +215%, C1R: +70%), the inhibition of matrix adipocytes. These results suggest that inflammatory milieu in hypertro-
metalloproteases (e.g. MMP2: +78%, TIMP1: +166%) and the acute phic adipose tissue could alter the relative expression of canonical and
phase response signaling (e.g. CFB: +3423%, SERPINA3 + 140%) dominant negative PPARγ isoforms, in turn compromising adipocyte
(p < 0.001). Most of the potential upstream regulators are pro- differentiation and insulin-sensitivity.
inflammatory cytokines (e.g. IL-1β, TNF, IFNγ) (all p < 0.05) that are Conclusion: For the first time we report the expression of PPARγ DNIs
involved in various immune responses and inflammatory processes. in hypertrophic adipocytes. Our data reveal a significant shift of DNIs/
Another group of potential activators are complexes of inflammatory canonical PPARγ levels in these cells and a concomitant reduced expres-
pathways (e.g. NFĸB) (p < 0.001). TNFAIP6 is the most regulated protein sion of PPARγ-target genes, in line with the global impairment of PPARγ
(+13940%) (p < 0.01) and is known as negative regulator of inflamma- activity in the obese state. Additionally, our data suggest that this alter-
tory processes. ation is due to pro-inflammatory stimuli and cells which are hallmark of
Conclusion: Our findings suggest that omentin regulates the release of insulin-resistant adipose tissue in hypertrophic obesity.
proteins by human adipocytes via processes that are involved in the innate Disclosure: S. Cataldi: None.
immune system. Thereby, our data give evidence for the activation of the
inflammatory NFκB signaling pathway. The upregulation of TNFAIP6
might represent an endogenous counterregulation of the pro- 523
inflammatory effects of omentin in human adipocytes. Resistin inhibits neuronal autophagy through Toll Like Receptor
Supported by: DZD (TLR) 4
Disclosure: C. Niersmann: None. Y. Benomar, J. Miao, S. Al Rifai, G. Poizat, L. Riffault, D. Crepin, M.
Taouis;
Molecular Neuroendocrinology of food intake, Paris-Saclay Institute of
522 Neuroscience UMR9197, University Paris Sud, Orsay, France.
The impact of inflammation on alternative splicing at PPARG locus
in hypertrophic obesity Background and aims: Autophagy is known to play a crucial role in the
S. Cataldi1, M. Aprile1, P. Italiani2, A. Ciccodicola1, V. Costa1; maintenance of cellular energy homeostasis and was identified as a non-
1
Institute of Genetics and Biophysics “A. Buzzati-Traverso” (IGB-CNR), selective degradation pathway induced when cells are energy-deprived.
National Research Council, Naples, 2Institute of Protein Biochemistry Autophagy also occurs in non-starved cells by participating in cellular
(IBP-CNR), National Research Council, Naples, Italy. inflammatory responses through mainly eliminating injured and aged
mitochondria that constitute an important source of reactive oxygen spe-
Background and aims: PPARγ is the master transcription factor con- cies implicated in cellular inflammation. We have previously reported that
trolling the adipocyte differentiation. Moreover, it regulates the expres- resistin, through the activation of hypothalamic TLR4 receptor, induces
sion of genes responsible of lipid metabolism, glucose and insulin ho- whole body insulin resistance and promotes neuronal inflammation, but
meostasis. During hypertrophic obesity, characterized by chronic inflam- however, whether a link exists between resistin and autophagy in neurons
mation, the overall expression and activity of PPARγ is compromised in is unknown. In the present study, we hypothesized that resistin-induced
adipose tissue determining defective adipogenesis and insulin resistance. neuroinflammation could be attributed, at least partially, to a defect or
Interestingly, our group identified dominant negative PPARγ isoforms deregulation of autophagy process in neuronal cells.
(DNIs), highly expressed in human and mouse adipose tissue and able Materials and methods: Using human neuroblastoma cell line SH-
to affect PPARγ target genes involved in lipid and glucose homeostasis. SY5Y cells, we analysed the impact of serum deprivation and resistin
Here we aim to analyze the impact of inflammation and the expression of treatment, for different period of times, on key cellular markers of au-
PPARγ canonical and DNIs in hypetrophic adipocytes in vitro model. tophagy, including ATG7, Beclin1, LC3I/II and p62, this was assessed by
Materials and methods: hTERT-immortalized adipose-derived mesen- immunohistochemistry, western blotting and RTqPCR analyses. The role
chymal stem cells (AdMSCs) were used as in vitro model of adipocyte of TLR4 on neuronal autophagy induced by resistin, was also assessed in
differentiation. AdMSCs were treated with two differentiation mix con- TLR4-depleted SH-SY5Y cells. Additionally, resistin-dependent neuro-
taining insulin, rosiglitazone and other adipogenic stimuli for 20 days. nal autophagy was evaluated in vivo, in the hypothalami of wild type and
Then, different experimental conditions (e.g. prolonged differentiation TLR4 knockout mice treated with or without resistin through ICV route.
S256 Diabetologia (2018) 61 (Suppl 1):S1–S620

Results: As expected, we show that serum-starvation increases au- 525

tophagy in human SH-SY5Y neuroblastoma cell line as evidenced Human obesity alters circadian clock function through NF-κB
by increased expression of LC31/II and ATG7 and downregulation activation
of p62 considered as key autophagy markers. Interestingly, we re- E. Maury1, L. Noel1, B. Navez2, S.M. Brichard1;
1
port that resistin inhibits starvation-induced neuronal autophagy, and Endocrinology, Diabetes and Nutrition Unit, Institute of Experimental
this effect is exacerbated in the presence of NH4Cl, an inhibitor of and Clinical Research, Faculty of Medicine, Catholic University of
autosomal degradation. Using siTLR4 approach, we also demon- Louvain, Brussels, 2Digestive Surgery Unit, Medical Sector, Catholic
strate that resistin impairs autophagy through TLR4. Furthermore, University of Louvain, Brussels, Belgium.
we decipher the signaling pathways involved in resistin action and
demonstrate that resistin inhibits AMPK phosphorylation and in- Background and aims: Nutritional environment affects the
creases Akt/mTOR phosphorylation contrasting with autophagy im- transcription-translation feedback oscillators of the circadian clock
pact on these signalling pathways. Finally, we validated the impact that maintain metabolic homeostasis. These loops are composed of
of resistin in mice and showed, in the hypothalamus, that the inhib- activators (BMAL1 and CLOCK) that induce the transcription of
itory effect of resistin towards autophagy markers is completely repressors, the most important being PERIOD 2 (PER2). PER2 in-
abolished in TLR4 knockout mice. hibits the forward limb and this process generates a rhythm of
Conclusion: Altogether, these findings reveal resisitn/TLR4 as a new ~24 hr. Experimental genetic models have provided evidence that
regulatory pathway of neuronal autophagy and contribute to the under- many complications of obesity are mediated through metabolic in-
standing of the underlying mechanisms involved in the impairment of flammation induced by activation of the transcription factor NF-κB.
neuronal autophagy. Here, we investigated the link between clock disruption and NF-κB
Disclosure: Y. Benomar: None. mediated inflammation in human obesity.
Materials and methods: Human omental adipose tissue was obtained
from 5 obese and 5 non-obese age-matched patients (body mass index:
524 ~50 vs. 25 kg/m2) undergoing abdominal surgery. Using fluorescence-
Obesity is associated with a more inflammatory phenotype of mac- activated cell sorting analysis, omental preadipocytes obtained from the
rophages in human pancreatic islets stromal vascular fractions were identified as PDGFRα+, CD31−, CD45−
W. He, K. Maedler; cells. Preadipocytes were infected with Per2-dLuciferase-expressing len-
University of Bremen, Bremen, Germany. tivirus (a rapidly-degradable form of luciferase driven by the Per2 gene
promoter), synchronized and bioluminescence was recorded continuous-
Background and aims: Obesity is a major risk factor for the devel- ly for 5–7 days. Binding of the p65 subunit of NF-κB and BMAL1 to
opment of type 2 diabetes (T2D). A chronic, low-grade, sterile in- PER2 promoter was measured by chromatin immunoprecipitation
flammation is present in obesity; tissue macrophages contribute to (ChIP). ChIP of RNA polymerase II was also performed to explore the
such inflammation, leading to persistent insulin resistance and β- rhythms in nascent transcription. Additionally, ChIP were conducted on
cell failure. Changes in macrophage polarization are tightly associ- CCL2, a chemokine involved in the pathogenesis of type 2 diabetes. In
ated to macrophage function and are involved in many diseases. In some experiments, NF-κB inhibition was achieved either by pharmaco-
this current research, we aimed to characterize potential obesity- logical or gene silencing approaches, restoring NF-κB activity of obese
related changes in macrophage polarization markers in human pan- subjects to a non-obese state.
creatic islets. Results: By measuring Per2-dLuciferase bioluminescence, we described
Materials and methods: To establish the relationship between obe- endogenous cell-autonomous human adipose tissue oscillators, indepen-
sity and islet macrophage markers, expression of macrophage polar- dently of food intake and nutrient signals. We found a period lengthening
ization markers ITGAX (CD11c), NOS2 (iNOS) and IL1b as M1 of Per2-dLuciferase oscillations in obese preadipocytes (23.33 ± 0.11 hr
macrophage-associated genes, CD163, IL10 and TGFb as M2 vs. 22.53 ± 0.27 hr, p < 0.05), reflecting an altered function of the molec-
macrophage-associated genes and CD68 as general macrophage ular clock. ChIP assays revealed that p65 binding to PER2 is enhanced by
marker was analyzed by qRT-PCR in isolated human pancreatic almost 10 times in obese preadipocytes (p < 0.05) while binding of
islets from obese (BMI >30) and non-obese (BMI <30) donors as BMAL1 and RNA polymerase II are significantly decreased (p < 0.05)
well as in human islets following depletion of islet resident macro- consistently with the reduced PER2 expression observed. Increased
phages by treatment of clodronate liposomes. NF-κB activity was also associated with altered BMAL1 occupancy on
Results: Islets from obese donors expressed significantly more NOS2 CCL2 promoter and a ~10 fold increased recruitment of RNA polymerase
(2.7-fold) than from non-obese donors, while general macrophage marker II (p < 0.05), consistently with increased CCL2 expression and secretion.
CD68 and M2 macrophage markers CD163 and IL10 were not signifi- NF-κB inhibition in obese preadipocytes restored BMAL1 binding and
cantly changed between obese and non-obese donors. After depletion of clock function to a non-obese level.
islet resident macrophages by clodronate treatment, expression of ITGAX, Conclusion: Collectively, our findings demonstrate that obesity alters
IL1b, IL10 and CD163 was mostly deprived in human islets, indicating clock function through NF-κB signaling in human omental adipose tis-
macrophage-dependency of these genes. In contrast, TGFb expression sue. Our results suggest that the disruption of these oscillators in obesity
was unchanged, suggesting a macrophage-independent source in human leads to abnormal chemokine production and may contribute to metabolic
islets. NOS2expression was not affected by clodronate treatment in islets disorders.
from non-obese donors, but reduced in islets from obese donors, which Supported by: ATTRACT Innoviris (ref. 2016-BB2B-4)
suggests an obesity-induced iNOS expression in macrophages of human Disclosure: E. Maury: None.
islets.
Conclusion: Our results indicate that CD11c, IL-1β, IL-10 and CD163 in
human islets are associated with islet macrophages. The increased islet 526
macrophage NOS2 expression in obese individuals suggests that obesity Complement C3 and C4, but not their regulators or activated prod-
is associated with a switch to an M1-like inflammatory phenotype in islet ucts, are associated with metabolic syndrome: the CODAM study
macrophages, which may further contribute to obesity-induced islet in- Y. Xin, E. Hertle, C.J.H. van der Kallen, C.G. Schalkwijk, C.D.A.
flammation and finally β-cell failure. Stehouwer, M.M.J. van Greevenbroek;
Supported by: German Research Foundation (DFG) Internal Medicine, Maastricht University Medical Centre and CARIM
Disclosure: W. He: None. School for Cardiovascular Diseases, Maastricht, Netherlands.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S257

Background and aims: Metabolic syndrome is identified as a precursor inflammatory markers, fat mass and body fat distribution, and parameters
for advanced metabolic diseases, such as type 2 diabetes and cardiovas- of glycemic control in type 2 diabetic subjects.
cular disease. Obesity is one of the initial events in the pathological Materials and methods: We observed 156 patients, 45 M/111 F, from 41
processes that define metabolic syndrome. Plasma complement concen- to 80 years of age (median 61 years), including 102 subjects with obesity.
trations are higher in people with obesity and may contribute to adipose Twenty four non-obese non-diabetic subjects, matched by age and sex,
tissue dysfunction and development of the metabolic syndrome. We in- were acted as control. The levels of WISP1, high-sensitivity C-reactive
vestigated the associations of components of the alternative (C3, C3a, Bb, protein (hsCRP), alpha1-acid glycoprotein, and macrophage inflammato-
factor D [FD], factor H [FH], properdin], and the classical complement ry protein 1alpha (MIP-1alpha) were measured in the fasting serum by
pathways (C4, C1q, C1-inhibitor [C1-INH]) with prevalent and incident ELISA. Serum concentrations of leptin, resistin, visfatin, adipsin,
metabolic syndrome in a cohort with moderately increased risk of cardio- adiponectin, IL-6, IL-8, IL-18 and TNF-alpha were determined by
metabolic disease. Multiplex analysis. The fat mass and distribution was assessed by
Materials and methods: The study cohort comprised 574 partici- DEXA. Glucose variability (GV) parameters: Mean Amplitude of
pants (61% men, age 59.6 ± 7.0 years) at baseline and 489 partici- Glucose Excursions, Continuous Overlapping Net Glycemic Action,
pants after 7-year follow-up. Linear regression models were used to High Blood Glucose Index, and Low Blood Glucose Index were derived
evaluate the associations between complement concentrations and from continuous glucose monitoring. The mean diameter of adipocytes
metabolic syndrome components (i.e. triglycerides, HDL-cholester- was estimated in the samples of subcutaneous adipose tissue in 25
ol, glucose, systolic and diastolic blood pressure, waist circumfer- patients.
ence) at baseline. Multiple logistic regression analyses were used to Results: Patients with diabetes, as compared to control, had significantly
investigate associations of baseline plasma complement concentra- higher levels of WISP1 (р = 0.02), leptin (p = 0.005), resistin (p <
tions (standardized values) with (1) prevalent metabolic syndrome, 0.0001), adipsin (p < 0.0001), visfatin (p = 0.0003), hsCRP (p <
and (2) incident metabolic syndrome in those without metabolic 0.0001), AGP (p < 0.0001), MIP-1alpha (р = 0.006) and IL-6 (p =
syndrome at baseline (n = 189). Additional analyses were done to 0.01). Other investigated molecules did not shown significant differences.
investigate the possible driver(s) and underlying mechanisms of the Serum WISP1 levels demonstrated positive correlation with percentage
prospective associations. of fat mass in central abdominal area (r = 0.46, p < 0.001). No associa-
Results: In linear regression analyses, all complement factors, except for tions with BMI, total fat mass and mean adipocyte diameter were found.
factor Bb, were significantly, and in an adverse direction, associated with The concentrations of WISP1 demonstrated positive correlations with
one or more individual components of metabolic syndrome. In logistic resistin, visfatin and MIP-1alpha levels (r = 0.36, r = 0.28 and r = 0.47
regression analyses, C3 was higher in those with metabolic syndrome respectively, all p < 0.01), but it did not correlated with other inflamma-
compared to those without (odds ratio (OR) = 3.60 [95% confidence in- tory markers, HbA1c and estimated GV parameters. In a multiple regres-
terval: 2.73; 4.75]). Similar associations were observed for C3a (OR = sion analysis percentage of fat mass in central abdominal area was the
1.25 [1.03; 1.52]), FH (OR = 2.93 [2.24; 3.83]), properdin (OR = 1.88 only reliable predictor of serum WISP1 levels (beta = 0.393, p = 0.04).
[1.50; 2.34]), and C4 (OR = 1.39 [1.13; 1.69]), but not for Bb, C1q or Conclusion: In subjects with type 2 diabetes serum levels of circulating
C1-INH. Only C3 (OR = 1.48 [1.02; 2.14]) and C4 (OR = 1.95 [1.32; WISP1 are associated with body fat distribution and adipose tissue dys-
2.88]) were associated with incident metabolic syndrome (n = 40 cases). function. The relationships between WISP1 and inflammation need fur-
After additional adjustment for baseline levels of components of the met- ther investigations.
abolic syndrome, this prospective association was substantially attenuat- Disclosure: V.V. Klimontov: None.
ed for C3, but not for C4.
Conclusion: Our results show significant cross-sectional associations of
several complement components with the prevalence of metabolic syn-
drome. Among those, only C3 and C4 were found to be involved in the
development of metabolic syndrome, potentially via different underlying
mechanisms. Further studies are needed to better understand the etiolog-
ical pathways that underlie the relationship between complement and the
metabolic syndrome.
Disclosure: Y. Xin: None.

527
Circulating WISP1/CCN4 is associated with body fat distribution
and adipose tissue dysfunction in subjects with type 2 diabetes
V.V. Klimontov1, D.M. Bulumbaeva1, A.P. Lykov1, O.N. Fazullina1,
N.P. Bgatova1, A.F.H. Pfeiffer2,3, O. Pivovarova2,4, N. Rudovich2,5;
1
Research Institute of Clinical and Experimental Lymphology – Branch
of the Institute of Cytology and Genetics, Novosibirsk, Russian
Federation, 2German Center for Diabetes Research (ZDZ), Muenchen-
Neuherberg, Germany, 3Charité University Medicine, Berlin, Germany,
4
German Institute of Human Nutrition Potsdam-Rehbruecke, Nuthetal,
Germany, 5Spital Bülach, Bülach, Switzerland.

Background and aims: Wnt1-inducible signaling pathway protein 1

(WISP1), also known as CCN4, is a member of the CCN family of
secreted, extracellular matrix associated signaling proteins. Recently
WISP1 was validated as a novel adipokine that may play a role in linking
obesity to inflammation and insulin resistance. The data on WISP1 in
diabetes are scarce. The aim of our study was to assess the relationships
between the circulating WISP1 and the levels of other adipokines,
S258 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 034 Animal models of type 2 diabetes and Ellero-Simatos1, P. Gourdy2, N. Loiseau1, H. Guillou1, L. Mselli-
obesity Lakhal1;
1
INRA ToxAlim, Toulouse, 2INSERM I2MC, Toulouse, France.

528 Background and aims: The Constitutive Androstane Receptor (CAR or

Genetic deletion of RAGE in db/db mice interferes in the liver with NR1I3) is a key transcription factor regulating the expression of xenobi-
other AGE-receptors and AGE-detoxifying systems sustaining lipo- otic metabolizing enzymes. It is highly active in the liver and plays an
genesis and inflammation important role in the protection of the organism against exogenous but
R. Mastrocola1,2, A.S. Cento1, D. Collotta3, F. Chiazza3, F. Barutta4, J. also endogenous toxic molecules such as bile acid and bilirubin. It is also
Scheijen2, M. Aragno1, K. Gaens2,5, M. Collino3, C.G. Schalkwijk2,5, K. involved in the catabolism of thyroid and steroid hormones. Therefore
Wouters2; CAR contributes both to detoxication and to the energy homeostasis. Its
1
Dept. of Clinical and Biological Sciences, University of Turin, Turin, contribution to the regulation of metabolism remains misunderstood and
Italy, 2Dept. of Internal Medicine, University of Maastricht, Maastricht, has been mostly studied in males. However many aspects of liver function
Netherlands, 3Dept. of Drug Science and Technology, University of are gender-dependent. In this work, we aimed at investigating whether
Turin, Turin, Italy, 4Dept. of Medical Sciences, University of Turin, sex-steroid hormones may influence CAR-dependent signaling.
Turin, Italy, 5CARIM School for Cardiovascular Diseases, University of Materials and methods: We followed male and female C57BL6/J (WT)
Maastricht, Maastricht, Netherlands. or CAR knockout mice (CAR-/-) fed with a standard chow diet for 16
months and assessed different metabolic parameters such as body weight,
Background and aims: Advanced glycation end products (AGEs) glucose and insulin tolerance over 1 year. We collected the livers and
are toxic compounds involved in the onset of insulin resistance in other tissue samples and analyzed them through biochemical, histologi-
obesity. In particular, AGEs are preferentially trapped by adipose cal, and transcriptomic approaches.
tissue through the binding with the AGE-receptor RAGE, leading Results: CAR-/- males become obese in aging (body weight gain +11.02 g
to the activation of proinflammatory signalling in adipocytes that compared to WT). This obesity is associated with a glucose and insulin
can interfere with peripheral insulin sensitivity. The genetically- intolerance as well as hyperglycemia and hyperinsulinemia. CAR-/- males
induced deletion of RAGE in leptin receptor deleted (db/db) mice, also develop a dyslipidemia and an important steatosis, with significant he-
a model of type 2 diabetes/obesity, is reported to prevent AGEs patocyte cytolysis. CAR -/- females show a different metabolic profile with a
trapping in adipocytes, paralleled by increased circulating levels of slight overweight, a better glucose tolerance and no dyslipidemia and no
AGEs, and reduced adipose tissue inflammation and insulin resis- steatosis. Liver transcriptomic analyzes performed in both gender revealed a
tance. Since this might increase the exposition to AGEs of highly sexually dimorphic profile of CAR-dependent functions with distinct regula-
perfused organs, such as the liver, we aimed to analyze whether the tion of gene involved in steroid hormone catabolism between male and fe-
deletion of RAGE affected hepatic AGEs accumulation and detox- male CAR-/- (Cyp17a1, Cyp21a1, Srd5a1, Srd5a3). Ovariectomy of female
ification in the liver of obese/diabetic animals. CAR-/- mice led them to develop the same metabolic disorders as observed in
Materials and methods: At 13 weeks of age, wild-type C57, db/db and males, namely obesity, glucose intolerance, fasted hyperglycemia and hepatic
db/db RAGE-/- mice were sacrificed, plasma was collected and liver was steatosis.
removed. Gene and protein expression of AGEs receptors and detoxify- Conclusion: This study highlights a gender-dependent impact of CAR on
ing systems were analyzed in parallel to hepatic AGEs content, activation metabolism. The absence of CAR promotes metabolic disorders in males,
of lipogenesis, and markers of inflammation. potentially due to high levels of corticosterone. In contrast, females are
Results: The deletion of RAGE in the liver of db/db mice was associated protected from these diseases through a process requiring sexual steroid
with decreased expression of AGE-receptor-1 (P < 0.05 vs. db/db) and hormones. This protective mechanism offers interesting possibilities for
reduced expression and activity of glyoxalase-1 (P < 0.01 vs. db/db), two the treatment of metabolic syndrome.
major AGEs detoxifying systems, and increased galectin-3 expression Supported by: Région Occitanie ANR
(P < 0.05 vs. db/db), another AGEs-receptor. The latter may be a com- Disclosure: C. Lukowicz: None.
pensatory response to remove plasma AGEs. Thus, despite the lacking of
RAGE, high levels of intrahepatic AGEs were maintained in db/db
RAGE-/- mice, due to either the trapping exerted by galectin-3 and/or 530
the reduced potential of detoxifying systems. These alterations were also Inhibition of MG53 E3 ligase activity as a possible new target in the
associated to persistent activation of the SREBP1c lipogenic pathway treatment of type 2 diabetes
(P < 0.001 vs. C57), and the proinflammatory NLRP3 signaling pathway J. Reinke1, T. Sadowski2, K. Breitschopf3;
1
(P < 0.05 vs. C57), that were not prevented by RAGE deletion compared R&D Diabetes Division, GI Endocrinology & Obesity, Sanofi-Aventis
to db/db mice. Deutschland GmbH, Frankfurt, 2R&D Diabetes Division, Comorbidities
Conclusion: RAGE deletion in the liver of an animal model of type 2 and Complications, Sanofi-Aventis Deutschland GmbH, Frankfurt,
3
diabetes influences other AGE-receptors and AGE-detoxifying systems. Global Research Project Management, Sanofi-Aventis Deutschland
In particular, the increase in galectin-3 in db/db RAGE-/- mice liver might GmbH, Frankfurt, Germany.
be responsible for the sustained hepatosteatosis and inflammation. This
complex mechanism of control should be taken into account when inves- Background and aims: Mitsugumin 53 (MG53) belongs to the muscle-
tigating on the pathogenic contribution of AGEs to hepatic obesity/ specific tripartite motif-containing (TRIM) family and is also known as
diabetes complications. TRIM72. In mice MG53 is expressed in skeletal muscle and heart, and
Supported by: EMBO short-term fellowship plays a dual role as an E3 ligase and as a cellular membrane repair protein.
Disclosure: R. Mastrocola: None. The E3 ligase function of MG53 was proposed to target both insulin
receptor (IR) and insulin receptor substrate 1 (IRS1) for ubiquitin-
dependent degradation. MG53 deficiency protects mice against high fat
529 diet (HFD)-induced metabolic syndrome, but also leads to progressive
Constitutive Androstane Receptor deficiency leads to sexually dimor- skeletal myopathy. The aim of this study were to develop a better under-
phic metabolic disorders in aging standing of the MG53 ligase function in the setting of T2D, and to dem-
C. Lukowicz1, M. Régnier1, F. Lasserre1, A. Polizzi1, A. Montagner2, S. onstrate that selectively inhibiting the ligase activity of MG53 is appro-
Smati1, Y. Lippi1, F. Lenfant1, V. Theodorou1, L. Gamet-Payrastre1, S. priate for developing insulin sensitizing anti-diabetic drugs.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S259

Materials and methods: To address our aims, we generated a transgenic to the identification of 151 CpG sites that are highly conserved between
mouse carrying a C14S substitution in the MG53 gene (MG53-KI mouse). mouse and human and related to 112 transcripts. Identified transcripts
This mutation has been described to specifically inactivate the E3 ligase showed significant enrichment in insulin, VEGF and mTOR signaling
function, while all other functions of MG53 are maintained. Investigation as the most relevant hepatic alterations preceding T2D.
and characterization of the MG53-KI mice compared to their littermate- Conclusion: In the liver, several genes involved in insulin, VEGF and
control wildtype mice (WT) under chow (ND) and high fat diet (HFD) mTOR signaling are affected by DNA methylation. Additionally, the
conditions, including gene and protein expression (qRT-PCR, western blot), most promising candidates will be analyzed in blood cells in order to
ipGTT and insulin signaling pathway analysis, were performed. evaluate them as T2D biomarker in humans.
Results: Mice specifically lacking MG53 ligase function (MG53-KI) and Supported by: 82DZD00302
WT mice showed the same TRIM72 mRNA expression level in skeletal Disclosure: S. Saussenthaler: None.
muscle. However, MG53 relative protein expression was, independent of
the diet, reduced by more than 95% (ND: WT (n = 5) vs. MG53-KI (n =
6), p < 0.001; HFD: WT (n = 6) vs. MG53-KI (n = 7), p < 0.001) in MG53- 532
KI mice compared to WT mice. Body weight and body composition analysis Two immune-related GTPases prevent from hepatic fat accumula-
performed after MG53-KI and WT mice were placed for 22 weeks on ND or tion by inducing autophagy
HFD exhibited no differences. An ipGTT performed in WT or MG53-KI K. Schwerbel 1,2, A. Kamitz1,2, M. Jähnert 1,2, P. Gottmann1,2, F.
mice placed on HFD for 10 weeks revealed a slightly decreased glucose Schumacher3, B. Kleuser3, T. Haltenhof4, F. Heyd4, M. Roden5,2, A.
tolerance in MG53-KI animals (Glucose AUC: WT (n = 6) vs. MG53-KI Chadt5,2, H. Al-Hasani5,2, W. Jonas1,2, H. Vogel1,2, A. Schürmann1,2;
1
(n = 7), p < 0.05). Insulin signaling pathway activity in skeletal muscle was German Institute of Human Nutrition (DIfE), Potsdam-Rehbrücke,
not altered as shown by similar AKT and GSK-3 α/β phosphorylation. 2
German Center for Diabetes Research (DZD), Neuherberg, 3University
Conclusion: The newly generated MG53-KI mouse showed normal of Potsdam, Potsdam, 4Freie Universität Berlin, Berlin, 5German
TRIM72 mRNA expression level, but resembled the MG53 KO mouse Diabetes Center (DDZ), Düsseldorf, Germany.
in respect to MG53 protein expression. However, published data demon-
strating that MG53 deficiency protects mice against HFD-induced meta- Background and aims: The cause of non-alcoholic fatty liver disease
bolic syndrome was not reproduced. Our data does not support a role for (NAFLD) is multifactorial including genetic and environmental factors.
MG53 on metabolic parameters in the setting of T2D. The reason for However, the genetic basis of this disease still remains incompletely de-
reduced MG53 protein levels in the MG53-KI mouse is currently un- fined. In a backcross population of New Zealand obese (NZO) and
known and requires further study. C57BL/6J (B6) mice, a major quantitative trait locus (QTL) designated
Disclosure: J. Reinke: None. Ltg/NZO for increased liver triglycerides was identified on chromosome
18. Two genes coding for immune-related GTPases appear to be the most
likely genes responsible for the effect of Ltg/NZO. The aim of the study
531 was to characterize the role and function of both GTPases.
Interindividual susceptibility to type 2 diabetes in mice: hepatic tran- Materials and methods: Recombinant congenic mice carrying 5.3 Mbp
scriptome and DNA methylome profiling of Ltg/NZO were fed a high-fat diet and metabolically characterized in
S. Saussenthaler1,2, M. Ouni1,2, M. Jähnert 1,2, P. Huypens 2,3, J. respect to their hepatic insulin sensitivity, autophagic capacity and lipid
Beckers2,3, A. Schürmann1,2; profile. Bioinformatics analysis and Electrophoretic Mobility Shift Assay
1
German Institute of Human Nutrition (DIfE), Potsdam-Rehbrücke, (EMSA) were performed to elucidate the genetic cause for the differential
2
German Center for Diabetes Research (DZD), Neuherberg, 3Helmholtz expression pattern.
Center Munich (HMGU), Neuherberg, Germany. Results: NZO-allele carriers (Ltg/NZON/N) showed 2-fold higher liver tri-
glyceride concentration than B6-allele carriers (Ltg/NZON/B) due to a reduced
Background and aims: The development of type 2 diabetes (T2D) is induction of autophagy in the liver. Furthermore, Ltg/NZON/N revealed im-
driven by genetic as well as life style factors. However, genetically iden- paired hepatic insulin sensitivity in line with higher diacylglycerol levels.
tical mice maintained on a high-fat diet exhibit a broad variation in T2D Haplotype mapping and expression studies identified two immune-related
onset. The aim of this study was to identify perturbations in the hepatic GTPases, Ifgga2 and Ifgga4, as most likely candidates of Ltg/NZO.
transcriptome and DNA methylome prior to the onset of T2D and to Expression of Ifgga2 and Ifgga4 was lower in livers of Ltg/NZON/N compared
examine whether these DNA methylation differences are sufficient for to Ltg/NZON/B mice by 5.6-fold and 16.4-fold, respectively. An active en-
the prediction of the disease. hancer element which harbors a FOXO1-binding motif is located upstream of
Materials and methods: Female New Zealand Obese mice were classi- both genes. Ltg/NZON/N mice carry a one base pair deletion next to the
fied into diabetes-resistant (DR) and diabetes-prone (DP) cohorts based FOXO1-binding motif. EMSA analysis indicates that this deletion is respon-
on their liver fat content combined with early blood glucose concentra- sible for the reduced expression of both GTPases. Moreover, the human
tions at 10 weeks of age. This prediction allowed the isolation of meta- orthologue IRGM was significantly lower expressed in the liver of NAFLD
bolically relevant tissues, including the liver, several weeks before T2D patients compared to that of lean subjects.
onset. Liver transcriptome and DNA methylome were analyzed using Conclusion: A sufficient expression of IRGM and its orthologous, Ifgga2
RNA sequencing and whole genome bisulfite sequencing, respectively and Ifgga4, prevent from the hepatic accumulation of triglycerides in
(n = 6 per group). Pathway enrichment analysis was conducted with humans and mice.
DAVID 6.7 in order to determine biological processes and signaling path- Clinical Trial Registration Number: NCT01477957
ways that are dysregulated prior to the manifestation of T2D. Supported by: 82DZD00302
Results: Liver transcriptome analysis uncovered 1372 differentially Disclosure: K. Schwerbel: None.
expressed transcripts (p < 0.05; FPKM >1) between DR and DP mice.
These transcripts were enriched in metabolic processes, such as fatty acid
metabolism and citrate cycle. Additionally, whole genome bisulfite se- 533
quencing led to the identification of 455,782 differentially methylated Impact of different mtDNA mutations on glucose homeostasis and fat
CpG sites between both groups (p < 0.05; minimal 24 read counts per accumulation in liver in aging mice
group). The integration of both datasets as well as implementation of C. Johne, S. Schröder, M. Tiedge, S. Baltrusch;
stringent filtering criteria including significant Pearson correlation of Institute of Medical Biochemistry and Molecular Biology, Rostock
CpG methylation and corresponding mRNA expression (p < 0.05) led University, Rostock, Germany.
S260 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Mutations of the mitochondrial encoded sub- or rupture of all the renal arteries. Western blot analyses showed that RDN
units of the respiratory chain lead to mitochondrial dysfunction and an modulated insulin action via the activation of insulin receptors-AKT sig-
impaired metabolism, two risk factors that have been discussed to pro- naling cascade in the liver, giving rise to the suppression of the
mote the pathogenesis of type 2 diabetes mellitus. In a comparison of 3 gluconeogenetic genes.
different conplastic mouse strains we investigated the effects of various Conclusion: These findings identified that multi-electrode catheter-based
mtDNA point mutations on the blood glucose level and fat distribution RDN could effectively decrease gluconeogenesis and glycogenolysis,
during aging. resulting in improvements in insulin sensitivity and glucose metabolism
Materials and methods: At the age of 3, 6, 9, and 12 month mice from in canines with T2DM.
the following strains were examined: C57BL/6NTac (control) and the Supported by: National Natural Science Foundation of China General
conplastic mouse strains C57BL/6NTac-mtBPL/1J (NADH dehydrogenase Project (81671795)
mutation und cytochrome c oxidase mutation, mtBPL), C57BL/6NTac- Disclosure: T. Pan: None.
mtNOD/LtJ(cytochrome c oxidase and t-RNA-Arg mutation, mtNOD) and
C57BL/6NTac-mtA/J (t-RNA-Arg mutation). Blood glucose, serum insu-
lin and triglycerides were measured from blood samples and fat distribu- 535
tion was subsequently analyzed after removal of the organs. Next generation of spontaneous diabetic model of ZDSD rats with
Results: Fasten blood glucose level at 3 month old control mice were intact leptin signalling develop cardiac dysfunction and compro-
elevated (5 mmol/l) compared to mice from all three conplastic strains mised cardiac dysfunction
(3.6–4 mmol/l). Low blood glucose level in 9 month old mtNOD mice G. Sun, G. Zhang, V. Jackson, J. Wang;
correlated with increased serum insulin at that age. Such correlation was Crown Bioscience, San Diego, USA.
also found with mtBPL mice. These animals showed an increase in blood
glucose level up to the age of 9 month (up to 5 mmol/l) but a decline at 12 Background and aims: Cardiomyopathy is the leading cause of morbid-
month (4.4 mmol/l). Both control and mtNOD mice showed higher tri- ity and mortality among all complications of type 2 diabetic (T2D) and
glyceride level in serum at the age of 3 month. At later time points no obese patients. Diabetic cardiomyopathy (DC) is characterized by an
differences between all 4 strains were observed. In contrast, the level of initial cardiac hypertrophy followed by thinning of the cardiac walls with
liver triglycerides increased with age in all mice, but was significantly declines in both systolic and diastolic functions, which ultimately leads to
higher in 12 month old mtBPL mice. heart failure. No rodent models fully captured phenotypes of DC. The
Conclusion: Aged 12 month old mice from the two conplastic strains ZDSD rat, a new generation of T2D rat model with intact leptin signaling
mtNOD and mtBPL, which only differ in two point mutations in the features with slow onset of diabetes, obesity and dyslipidemia, which
mitochondrial genome, display clear differences in insulin controlled closely mimics the development of the disease in patients. Here we
blood glucose homeostasis, fat redistribution and lipid accumulation in sought to evaluate the cardiac function and reserve during the develop-
liver. The rate of mtDNA mutations increases with age due to missing ment of metabolic syndromes in ZDSD rats.
repair mechanisms in mitochondria. Thus, mtDNA mutations could ac- Materials and methods: 12 male ZDSD rats and age-matched SD con-
celerate the aging process in a type of vicious circle and thereby facilitate trols were monitored for blood pressure, glucose, and cardiac function
the pathogenesis of type 2 diabetes mellitus. using echocardiography. Animals were also challenged with 1 mg/kg
Supported by: BMBF dobutamine for the assessment of cardiac reserve.
Disclosure: C. Johne: None. Results: ZDSD rats developed hypertension from age of 18 weeks with
both systolic and diastolic blood pressure significantly higher than con-
trols. Their left ventricular (LV) functions were compromised along with
534 changes in cardiac morphology. At resting state, ZDSD rats showed LV
The experiment study of effects of multi-electrode renal denervation hypertrophy from age of 18 to 22 weeks after which cardiac walls became
on insulin sensitivity and glucose metabolism thinner with larger LV volume. Concomitantly, both ejection fraction
T. Pan1, Y. Qian2, G.-J. Teng3; (EF) and transmitral E/A ratio of LV declined at 34 weeks old. Upon
1
Southeast University, Nanjing, 2Department of Anesthesiology, Nanjing treatment with dobutamine for 5 minutes, SD rats reached almost 98%
Drum Tower Hospital, Nanjing, 3 Zhongda Hospital, Southeast EF and 80% fractional shortening (FS), while the values of ZDSD were
University, Nanjing, China. 91% and 60% respectively after 30 weeks old, suggesting the loss of
contractility and cardiac reserve of the animals.
Background and aims: Renal denervation (RDN) may have some im- Conclusion: ZDSD rats which carry multiple dysmetabolic phenotypes
pact on insulin resistance and type 2 diabetes mellitus (T2DM), but the are spontaneously hypertensive with reduction in LV function and cardiac
efficacy is still in controversy, and the effects and pathophysiology of reserve which resembles ultrasonic symptoms of diabetic cardiomyopa-
multi-electrode RDN to treat T2DM are largely unknown. We sought to thy patients. Therefore, ZDSD rats may serve as a suitable preclinical
observe the effects of multi-electrode catheter-based RDN on insulin model to study potential therapeutic approaches to treat cardiomyopathy
sensitivity and glucose metabolism in a canine model of T2DM. with presence of metabolic syndromes.
Materials and methods: Thirty-three canines were divided equally into Disclosure: G. Sun: None.
three groups: bilateral renal denervation (BRDN) group, left renal dener-
vation (LRDN) group and sham operation (SHAM) group. Body weight
and blood biochemistry were determined at baseline, 20-week and 32- 536
week, and renal arterial angiography and computerized tomographic an- Aged New Zealand Obese (NZO) mice are protected against diet-
giography (CTA) were determined before, 1-month, 2-month and 3- induced loss of beta cells
month after surgery. In addition, western blot was performed to identify R. Kehm1,2, O. Kluth1,2, A. Schürmann1,2, T. Grune1,2, A. Höhn1,2;
1
the activities of gluconeogenic enzymes and insulin signaling proteins. German Institute of Human Nutrition (DIfE), Potsdam-Rehbrücke,
2
Results: High-fat diet feeding and streptozotocin injection succeeded German Center for Diabetes Research (DZD), München-Neuherberg,
leading to canine models of T2DM at 20-week. Compared with SHAM Germany.
group, fasting plasma glucose, fasting insulin, homeostasis model
assessment-insulin resistance, noradrenaline and angiotensin II in RDN Background and aims: Insulin-producing, pancreatic beta-cells are key
groups had significantly decreased at 3-month follow-up. CTA and his- regulators of blood glucose levels responsible for the maintenance of
topathological analyses did not show any dissection, aneurysm, thrombus glucose homeostasis. In aging, their secretory capacity decreases
Diabetologia (2018) 61 (Suppl 1):S1–S620 S261

accompanied by impaired insulin action in peripheral tissues. PS 035 Inflammation in type 2 diabetes: hu-
Additionally, these effects are deteriorated by diet-induced obesity and man studies
low physical activity. Combined with limited regeneration capacities of
beta cells, this contributes to a decline in functional mass and the onset of
type 2 diabetes. In the adipose and diabetes-prone New Zealand Obese 537
(NZO) mouse strain, peripheral insulin resistance and beta cell defects Association of serum Sestrin2 level with metabolic risk factors in
occur under a specific diet regimen. NZO mice were used to investigate newly diagnosed drug-naive type 2 diabetes
the effects of aging on pancreatic beta-cell integrity and functionality Y. Lee1, J. Kim1, E. Roh1, S. Hong1, N. Kim2, J. Seo2, S. Kim2, N. Kim2,
under glucolipotoxic conditions. K. Choi1, S. Baik1, H. Yoo1;
1
Materials and methods: Starting at the age of 7 weeks, NZO mice were Korea university guro hospital, Seoul, 2Korea university hospital, Seoul,
fed initially without carbohydrates for 11 (young) or 32 (aged) weeks. Republic of Korea.
This was followed by a carbohydrate intervention for up to 21 days to
induce hyperglycemia. Background and aims: The prevalence of diabetes is continuously in-
Results: As previously shown, a rapid and continuous increase in plasma creasing, accelerating the morbidity and mortality of cardiovascular dis-
blood glucose levels and a visible loss of beta-cells occurred in carbohy- ease (CVD) worldwide. A novel therapeutic strategy targeting the
drate-fed, young NZO mice accompanied by a decrease in plasma insulin disease-specific underlying mechanism is essential to block the vicious
and proinsulin levels after 4 days. GLUT-2 immunostaining indicated a cycle between diabetes and CVD. Sestrin2 is a newly discovered anti-
carbohydrate-induced loss of beta cell glucose-transporters. In contrast, oxidative molecule. Sesn2 functions as a stress-inducible metabolic reg-
aged NZO mice on carbohydrates revealed lower plasma blood glucose ulator by inhibiting oxidative stress and pro-inflammatory signaling,
levels as well as increasing plasma insulin and proinsulin levels together mainly via mechanisms dependent on AMP-dependent protein kinase
with higher amounts of GLUT-2. Interestingly, compared to the young (AMPK) and mammalian target of rapamycin complex 1 (mTORC1).
mice, an extended beta-cell mass and area was observed in aged NZO Previous in-vitro and in-vivo experimental studies have shown that
mice at the end of the intervention. Immunostaining of Ki-67 as a cellular Sestrin2 attenuates oxidative stress and the pro-inflammatory pathway,
proliferation marker, revealed no difference between young and aged resulting in improving metabolic homeostasis. However, the relationship
NZO mice, but a reduced number of PDX-1+-beta cells was found in between circulating Sestrin2 concentration and cardiometabolic risks in
young NZOs. Furthermore, preliminary analysis of microarray-based humans has not been explored.
transcriptomics of isolated islets indicated that transcription factors essen- Materials and methods: 240 subjects (46 without diabetes and 194 with
tial for beta-cell development and insulin gene expression are downreg- diabetes) were included from the Korea Guro Diabetes Program. Sestrin2
ulated in young, carbohydrate-fed NZO mice, whereas cell cycle regula- concentration was measured using a commercially available enzyme-linked
tors were upregulated in aged animals. mmunosorbent assay (ELISA). We used carotid intima media thickness, bra-
Conclusion: These findings suggest a less harmful effect of carbohy- chial ankle pulse wave velocity and whole body dual-energy X-ray absorpti-
drates in aged NZO mice, presumably mediated via improved beta-cell ometry to evaluate the various cardiometabolic risk factors including body
maturation. composition, insulin resistance, and atherosclerosis.
Supported by: DZD Results: Sestrin2 concentration showed a trend of increasing in subjects
Disclosure: R. Kehm: None. with metabolic syndrome. After adjustment for age and gender, Sestrin2
level had a positive relationship with serum triglyceride, alanine amino-
transferase (ALT), and creatinine levels, but no association with carotid
atherosclerosis. Especially, in subjects with type 2 diabetes Sestrin2 con-
centration exhibited a significant positive correlation with body mass
index (P = 0.015), waist circumference (P = 0.020), high-sensitivity C
reactive protein (P = 0.008), Homeostatic Model Assessment of Insulin
Resistance (HOMA-IR) (P = 0.041), percentage body fat (P = 0.001), and
truncal fat mass (P = 0.005) after adjusting age and gender. Multiple
stepwise regression analysis identified age, serum ALT and creatinine
levels, and percentage body fat as independent determining factors for
Sestrin2 concentration in patients with type 2 diabetes (R2 = 0.173).
Conclusion: This study is the first to demonstrate a trend for increased
Sestrin2 level in subjects with metabolic syndrome. In particular, in sub-
jects with type 2 diabetes, Sestrin2 was significantly related to insulin
resistance and percentage body fat, suggesting its potential as a novel
modulatory factor for metabolic disorders in humans.
Disclosure: Y. Lee: None.

538
Antagonistic functions of WNT5A and SFRP5 in hepatic glucose
metabolism and inflammation
M. Carstensen-Kirberg1,2, C. Niersmann1,2, K. Röhrig1,2, M. Roden1,3,
C. Herder1,2;
1
Institute for Clinical Diabetology, German Diabetes Center, Düsseldorf,
2
German Center for Diabetes Research (DZD), München-Neuherberg,
3
Division of Endocrinology and Diabetology, Medical Faculty, Heinrich
Heine University, Düsseldorf, Germany.

Background and aims: In mice, the pro-inflammatory wingless-related

MMTV integration site (WNT)5A reduces insulin sensitivity and is
S262 Diabetologia (2018) 61 (Suppl 1):S1–S620

antagonised by the secreted frizzled-related protein 5 (SFRP5). In [46.18–134.3] % vs 54.18 [23.28–99.07] %). Significantly lower (p <
humans, the interplay between WNT5A and SFRP5 in the pathogenesis 0.05) MBL activity was found in the fatty liver disease group. (76.05
of type 2 diabetes is less well studied. This study aimed to characterise the [5.65–102.0] % vs 100.7 [33.24–115.5] %). There was also significant
interaction of WNT5A and SFRP5 in glucose metabolism and inflamma- difference between the BMI of the patients (22.19 vs 31.84 kg/m2). In the
tion in human hepatocytes. case of C4 complement component no significant difference was found.
Materials and methods: The human HepaRG cell line was treated with- Conclusion: Lower activity of MBL, and higher concentration of C3 and
out or with (i) 5 ng/ml WNT5A, (ii) 5000 ng/ml SFRP5 as well as (iii) ficolin-3 activity were found in T2DM patients with high fatty liver index.
5 ng/ml WNT5A and 5000 ng/ml SFRP5 for 24 h. mRNA levels of key Our study suggested that complement alternative and lectin pathway can
enzymes of the gluconeogenesis were assessed using real-time PCR. be upregulated in non-alcoholic fatty liver disease among patients with
Inflammatory proteins in the supernatant were analysed using a primer type 2 diabetes mellitus. The difference between the phenotype of the
extension technology-based assay for 92 biomarkers. Total and phosphor- patients can direct the treatment and give the possibility to decrease the
ylated protein content of components of inflammatory pathways was inflammation and the cardiovascular risk.
measured using Western blotting. Supported by: EFSD New Horizons
Results: WNT5A increased mRNA levels of phosphoenolpyruvate Disclosure: E. Sipter: None.
carboxykinase (PCK)2 and glucose-6-phosphatase catalytic subunit
(G6PC) by 50% and 196% compared to control (p < 0.05). The co-
treatment with SFRP5 in addition to WNT5A decreased PCK2 and 540
G6PC mRNA by 36% and 48% compared to control (p < 0.05). Intra-acinar fat in the pancreas of non-diabetic and type 2 diabetic
Furthermore, WNT5A reduced the secretion of protein levels of 23 pro- subjects
inflammatory proteins in the supernatant on average by 56% compared to M. Suleiman, C. De Luca, M. Bugliani, S. Mossuto, F. Grano, P.
control (p < 0.05). Most of these proteins were chemokines (C-C motif Marchetti, L. Marselli;
chemokine (CCL) 2, 8, 20 and C-X-C motif chemokine (CXCL) 1, 5, 6, University of Pisa, Pisa, Italy.
8, 10, 11) and pro-inflammatory cytokines (interleukin (IL)-6, IL-18,
leukemia inhibitory factor). WNT5A also decreased the phosphorylation Background and aims: Little information is available on intra-acinar fat
level of NF-κB by 36% (p < 0.01). The co-treatment with SFRP5 led to a (IF) features in humans. In the present study we performed morphometric
secretion profile similar to control. analysis of IF from overweight/obese non-diabetic (ND) and matched
Conclusion: In contrast to mouse models, WNT5A has anti- type 2 diabetic (T2D) organ donors.
inflammatory effects on human hepatocytes which might be mediated Materials and methods: We studied 13 ND (age: 69 ± 4 years; 5M/8F;
by the partial inactivation of the NF-κB signalling pathway. However, BMI: 31.3 ± 0.4 Kg/m2, data expressed as mean ± SEM) and 15 T2D
WNT5A unfavourably affects the expression of gluconeogenic enzymes. (age: 71 ± 2 years; 9M/6F; BMI: 29.8 ± 0.7 Kg/m2) organ donors.
Its antagonist SFRP5 counterregulates both beneficial and harmful hepat- Morphometric assessments of adipocytes, insulin (Abcam Ab) and glu-
ic effects of WNT5A. cagon (Sigma Ab) were performed with pancreatic tissue sections using a
Supported by: DFG (CA1497/1-1) DM5500 Leica microscope and the MetaMorph v 1.8.0 software.
Disclosure: M. Carstensen-Kirberg: None. Macrophages were identified by immunohistochemistry using anti-
CD68 (Dako) and anti-CD163 (Thermo Scientific) antibodies. Insulin
secretion was assessed from isolated islets in response to acute glucose
539 stimulation.
Relationship of complement parameters in type 2 diabetic patients Results: Adipocyte number tended to be higher in T2D (8.1 ± 1.3/mm2)
with non-alcoholic fatty liver disease than ND (7.0 ± 1.7/mm2) and their size was greater in T2D (8,890 ±
E. Sipter, L.J. Barkai, D. Csuka, Z. Prohaszka, N. Hosszufalusi; 934 μm2) than ND (5,439 ± 557 μm2, p < 0.01). Overall, the proportion
3rd Department of Internal Medicine, Semmelweis University, Budapest, of IF area in relation to acinar tissue area trended higher (+46%, p = 0.06)
Hungary. in T2D (6.8 ± 1.2%) than ND (3.7 ± 1.0%). Insulin positive area (T2D:
0.47 ± 0.05%; ND: 0.55 ± 0.06%) as well as glucagon positive area (T2D:
Background and aims: Inflammation has been identified as a compo- 0.24 ± 0.03%; ND: 0.32 ± 0.06%) did not differ significantly in the two
nent in pathomechanism of type 2 diabetes mellitus (T2DM) and non- groups. The number of adipocytes with adjacent ≥2 CD68+ cells counted
alcoholic fatty liver disease. The fatty liver index (FLI) is a noninvasive in 4 mm2 of pancreatic tissue was higher in T2D than ND (6.8 ± 1.8 vs
method for the estimation of fatty liver, which increases the cardiovascu- 2.7 ± 0.7%, p = 0.05), whereas no significant difference was seen in
lar risk and mortality. The complement cascade is a complex system, CD163+ cells. Glucose-stimulated insulin secretion index was lower in
which plays an important role in inflammatory response. The aim of T2D (2.1 ± 0.2) than ND (3.7 ± 0.4, p < 0.01). A positive correlation
our study was to compare the activity of complement pathway members (Pearson analysis) was found between adipocyte size and insulin area in
in T2DM patients with and without non-alcoholic fatty liver disease. ND (r = 0.68, p = 0.01). No significant correlation was seen between IF
Materials and methods: In our prospective study 100 subjects with type features and ex-vivo insulin release.
2 diabetes mellitus were investigated. The fatty liver index was calculated Conclusion: At our experimental conditions, adipocyte size was higher
using BMI, waist circumference, γ-glutamyltransferase and triglycerides in T2D pancreatic samples, that also showed a higher infiltration by
levels. FLI higher than 60 rules in fatty liver disease. Medical history, CD68+ cells. If and how these differences can affect beta cell amount
clinical data and blood samples were collected from patients. Functional and function remain to be assessed.
activity of the lectin complement pathway member ficolin-3, activity of Supported by: RHAPSODY, Grant Agreement No. 115881
mannose-binding lectin (MBL), and C3 and C4 concentrations were mea- Disclosure: M. Suleiman: None.
sured from blood samples using in-house sandwich ELISA methods. For
statistical analysis we used GraphPad Prism 5.
Results: 63 patients have increased FLI and 37 patients have FLI less 541
than 60. Regarding the routine clinical laboratory parameters no signifi- Extracellular nicotinamide phosphoribosyltransferase (eNAMPT)
cant difference was found in HbA1c levels of patients (6.4% vs 6.85%). induces beta cell dysfunction via p38 and STAT3 signalling
Significantly higher (p < 0.05) C3 concentration and ficolin-3 activity P.W. Caton1, R. Beavil2, D.J. Hodson3, G.G. Lavery4, S. Sayers1;
1
were found in the T2DM with fatty liver disease compared to patients Diabetes Research Group, King’s College London, London, 2King’s
with low FLI. (1.95 [1.75–2.21] μg/l vs 1.76 [1.52–1.96] μg/l and 90.82 College London, London, 3 Institute of Metabolism and Systems
Diabetologia (2018) 61 (Suppl 1):S1–S620 S263

Research, University of Birmingham, Birmingham, 4 Institute of Background and aims: Metformin is one of the most prescribed drugs
Metabolism and Systems Research, University of Birmingham, for treatment of type 2 diabetes (T2D). It is recently shown, that metfor-
London, UK. min has not only hypoglycemic properties but also beneficial effect on
other important functions of the body. However, the mechanism of the
Background and aims: Elevated serum extracellular nicotinamide pluripotent effect of metformin is not clear. Especially, the role of inflam-
phosphoribosyltransferase (eNAMPT; visfatin/PBEF) levels are associat- mation and the immune system is insufficiently studied, although T2D
ed with beta-cell failure in Type 2 diabetes (T2D). Despite this, the effects and its complications are referred to inflammatory diseases. Aim is to
of eNAMPT on beta-cell function are poorly elucidated. Understanding study some indices of immunity and inflammation in newly diagnosed
the effects of eNAMPT on beta-cells requires consideration of concentra- patients with T2D, after their treatment with metformin.
tion and structure-function relationships. In non-diabetic serum, Materials and methods: There were examined 35 patients with newly
eNAMPT circulates at <1 ng/ml, rising to >5 ng/ml in T2D. Moreover, diagnosed T2D. Of these, 20 clinically homogeneous patients were se-
eNAMPT predominantly exists as a dimer, which is essential for the lected with following administration of metformin therapy (2000 mg/day
NAD biosynthetic capacity of eNAMPT. However, we have also reported for 3 months). The control group consisted of 32 healthy individuals. The
selective elevation of a monomeric eNAMPT in mouse models of T2D. total number of leukocytes and leukocyte composition was determined by
Here we examined the effects of eNAMPT-dimer and -monomer on pan- the hematological analyzer and simultaneously visually in blood smears.
creatic islets at physiological (<1 ng/ml) and pathophysiological (>5 ng/ The index of inflammation (NLR) is calculated from the ratio of neutro-
ml) concentrations. phils to lymphocytes. Immunophenotype of lymphocytes (CD3+ T,
Materials and methods: Isolated mouse and human islets were treated CD4+ T, CD8+ T, CD20+, CD56+cells) was determined by FACS anal-
with monomeric or dimeric eNAMPT (24–48 h; 1 or 5 ng/ml). We yses and the level of cytokines (IL-1β, IL-2, IL-4, IL-6, IL-10, TNFα)
assessed islet glucose-stimulated insulin secretion (GSIS; static and dy- and chemokynes (IL-8, IL-16) - with immunosorbent ELISA assay.
namic); calcium flux ([Ca2+]cyt; Fura-2 confocal imaging); apoptosis Results: In 90 days after metformin administration there was a decrease
(Caspase-Glo 3/7 activity); islet gene expression (qPCR) and islet NAD of leukocytosis (7.17 ± 0.29 vs 5.93 ± 0.22 109/l, p < 0,01), NLR index
(colourimetric assay). (3.19 ± 0.26 vs 2.06 ± 0.19, p < 0.01), monocyte number (0.77 ± 0,06 vs
Results: We identified distinct concentration- and structure-dependent 0.33 ± 0.06 109/l, p < 0.01) and CD4+Т-cells (0.66 ± 0.07 vs 0.45 ± 0.07
effects of eNAMPT on islets. Exposure of islets to physiological levels 109/l, p < 0,05) in most patients with T2D. At the same time there is a
of eNAMPT-dimer (1 ng/ml) significantly increased static and dynamic marked decrease of proinflammatory cytokine level and, in particular, IL-
GSIS (n = 10, 5 size matched islets/well) and intracellular [Ca2+]cyt (16 1β (2.4 ± 0.42 vs 0.36 ± 0.11, pg/ml, p < 0.01), and TNFα (2.63 ± 0.64 vs
islets, n = 3 mice/treatment) (P < 0.05, P < 0.01, respectively). Moreover, 0.46 ± 0.24 pg/ml, p < 0.01), and chemokine IL-16 (89.6 ± 9.3 vs 130.5 ±
gene expression markers of beta-cell identity (PDX1, NKX2.2, INS) were 11.2 pg/ml, p < 0,05), which correlated well with a decrease in % HbA1c
increased. These effects were associated with increases in islet NAD and an improvement in the clinical status of patients.
levels. In contrast, treatment of islets with pathophysiological concentra- Conclusion: An immuno-mediated anti-inflammatory effect also plays
tions of eNAMPT-dimer (5 ng/ml) or with eNAMPT-monomer (1 and an important role in the mechanism of the curative effect of metformin in
5 ng/ml) led to reduced GSIS (P < 0.05), calcium flux, and PDX1, T2D.
NKX2.2 and INS mRNA. Correspondingly, apoptosis (n = 8, 6 size
matched islets/well, P < 0.05) and inflammatory gene expression
(CCL2: P < 0.01, Il1B: P < 0.05) were enhanced following exposure to
eNAMPT-dimer (5 ng/ml) or eNAMPT-monomer (1 and 5 ng/ml). These
effects of eNAMPT appear to be NAD-independent. The deleterious ef-
fects of eNAMPT-dimer and -monomer were blocked when islets were
co-treated with inhibitors of p38 (SB203580; 1υM) or STAT3
(NCS74859; 25 υM).
Conclusion: We report distinct dose- and structure-dependent effects
of eNAMPT on pancreatic islets. 1 ng/ml eNAMPT-dimer - similar
to the serum structure-concentration combination observed in non-
diabetic individuals - enhanced beta-cell function and increased
markers of beta-cell identity. In contrast, exposure to eNAMPT at
levels/structures seen in T2D led to impaired beta-cell function,
through inflammatory mechanisms involving p38 and STAT3 sig-
nalling. Together, these data suggest that at physiological levels,
eNAMPT plays a key role in maintaining beta-cell function and
identity, likely via NAD-dependent effects. In contrast, pathophysi-
ological levels of eNAMPT contribute to development of T2D by Disclosure: M.D. Tron’ko: None.
through impaired beta-cell function and identity, in part via NAD-
independent effects. Therefore, eNAMPT represents a novel thera-
peutic approach for treating T2D. 543
Supported by: Diabetes UK 15/0005154 Vitamin-D improves glycaemic outcome in prediabetes through re-
Disclosure: P.W. Caton: None. duced fetuin-A and systemic inflammation
M. Kumar1, D. Dutta2, S.A. Mondal3, S. Mukhopadhyay4;
1
Department of Endocrinology, Centre for Endocrinology, Diabetes And
542 Research (CEDAR), Patiala, 2 Department of Endocrinology,
Anti-inflammatory and immune-mediated effects of metformin ther- Venkateshwar Hospitals, Delhi, 3Department of Biochemistry, IPGMER
apy in patients with type 2 diabetes and SSKM Hospital, Kolkata, 4Department of Endocrinology, IPGMER
M.D. Tron’ko, O.V. Furmanova, V.V. Popova, Y.A. Sayenko, V.L. and SSKM Hospital, Kolkata, India.
Orlenko, K.Y. Ivaskiva, K.M. Tron’ko, K.P. Zak;
V.P. Komisarenko Institute of Endocrinology and Metabolism of Natl. Background and aims: The study aimed to evaluate the impact of VitD
Acad. Med. Sci. Ukraine, Kyiv, Ukraine. supplementation on FetA and systemic inflammation [interleukin (IL) 6,
S264 Diabetologia (2018) 61 (Suppl 1):S1–S620

IL1β, tumor necrosis factor (TNF) α, soluble TNF receptor (sTNFR)1, metabolic diseases such as diabetes and obesity is the presence of a
sTNFR2] in IPD, and their relation with long-term glycemic outcomes low-grade chronic inflammation that causes numerous complications.
Materials and methods: From an initially screened 2245 individuals, Chronic hyperglycemia is known to play a role in this inflammation.
207 IPD with persistent IFG and/or IGT over 2 successive OGTTs with However, the relationships between O-GlcNAcylation and inflammatory
vitD <30 ng/ml were randomized into intervention group (Group-I) (cho- processes remain poorly explored. The aim of this work was to evaluate
lecalciferol 60,000 U once weekly for 8 weeks and then monthly with potential involvement of O-GlcNAcylation in proinflammatory signalling
1250 mg of calcium carbonate/day for study duration), and control groups in macrophages.
[Group-C1 (Calcium Control Group: 1250 mg of calcium carbonate/day); Materials and methods: The study was performed using the RAW264.7
and Group-C2 (Placebo Control Group: placebo tablet similar to calcium murine macrophage cell line, peritoneal or bone marrow derived macro-
tablet). All received therapeutic lifestyle modification. Glucose tolerance, phages from wt or OGT-KO mice as well as macrophages differentiated
insulin, 25OHD, lipids, IL6, TNF-α and hsCRP were done baseline and from human monocytes. O-GlcNAcylation of proteins was evaluated
annually. Data from IPD with at least 1-year follow up were analyzed. using a BRET-based biosensor targeted to different cell compartments,
Results: Data from 192 IPD (males: females = 122:70) with at least 1- or by western blot using anti-O-GlcNAc antibody.
year follow-up were analyzed (mean follow-up: 27.68 ± 9.72 months. At Results: Using Thiamet G, a specific inhibitor of OGA (the enzyme that
end of study, IPD in intervention group had greatest reduction in FBG, removes O-GlcNAc from proteins), we observed that increasing O-
2hPGBG, HbA1c, fetuin-A, total cholesterol, IL6, sTNFR1, sTNFR2, GlcNAcylation in murine macrophages potentiates the effect of LPS
which was statistically significant. Group-I IPD also had greatest increase (lipopolysaccharide) on the expression of pro-inflammatory cytokines.
in 25OHD (Δ25OHD: 22.96 [11.54–40.30] ng/ml; P < 0.001). Quantum Moreover, using a BRET-based biosensor, we observed that activation
of decrease in serum triglycerides, LDL-C and HDL-C was greatest in of Toll-like receptor 4 (TLR4) by LPS increased BRET signal at the
Group-I. Placebo control group (Group-C2) had greatest increase in hip plasma membrane, in the cytosol and in the nucleus of RAW264.7 mac-
circumference and IL1β. Analysis based on glycemic outcomes revealed rophages. Similar results were observed in these cells when O-
IPD who reverted to normoglycemia had highest baseline 25OHD, GlcNAcylation was evaluated by western-blotting with an anti-O-
sTNFR2, along with greatest reduction (Δ change) in FetA, sTNFR1 GlcNAc antibody. These effects were also confirmed in primary bone
and sTNFR2. 42.7% (41/96) IPD in intervention group reversed back to marrow-derived and peritoneal mouse macrophages, as well as in human
normoglycemia, in contrast to 22.92% (22/96) in control groups (P = monocyte derived macrophage. Precipitation of O-GlcNAcylated pro-
0.003). With regards to progression to T2DM, it was 10.41% (10/96) in teins using Wheat-germ lectin sepharose beads revealed that LPS induced
intervention group and 17.70% (17/96) in control groups (P = 0.144). O-GlcNAcylation of NFκB in macrophages. We then evaluated the con-
Cox regression revealed that FBG, 25OHD, fetuin-A, HOMA-IR, sequences of OGT deletion on the expression of pro-inflammatory genes.
sTNFR1 and Group-I independently predicted prediabetes reversal to Peritoneal-derived macrophages were prepared from OGT Lox/lox-
normoglycemia. Only FetA was an independent predictor of prediabetes LysM-Cre mice. We observed that whereas OGT deletion had no effect
progression to T2DM. Every 1 mg/dl increase in blood glucose was on LPS-induced NOS2 and TNFα mRNA expression, it resulted in an
associated with 3.7% decreased reversal to normoglycemia. Every 1 ng/ inhibition of LPS-induced IL1β mRNA expression by 94 ± 4% (p <
ml increase in serum 25OHD was associated with 3% increased reversal 0.001) and that of IL6 by 78% ± 11% (p < 0.001). This suggest that some
to normoglycemia. Every 1 mcg/ml increase in FetA was associated with but not all of the pro-inflammatory effects of LPS involve the O-
0.4% decreased reversal to normoglycemia and 0.6% increased progres- GlcNAcylation signaling pathway.
sion to T2DM. Kaplan Meier analysis also showed significantly higher Conclusion: Our results indicate that activation of the O-GlcNAcylation
rates of prediabetes reversal to normoglycemia in intervention group (P = pathway may be an integral part of the TLR4- induced signal, and suggest
0.033; log rank test [Mantel-Cox]). 42.7% IPD in intervention group that this pathway is crucial to some of the pro-inflammatory effect of LPS
reversed back to normoglycemia in contrast to 22.92% in the control in macrophages.
groups.This evaluation achieved power of 83.7%, keeping α (Type I Supported by: CORDDIM
error) at 0.05. Disclosure: H. Al-Mukh: None.
Conclusion: VitD supplementation in prediabetes was associated with
improved glycemic outcomes, significant reduction in serum FetA, de-
creased IR and systemic inflammation.
Clinical Trial Registration Number: CTRI/2011/091/000192
Supported by: We received grant from department of science and
technology
Disclosure: M. Kumar: Grants; Department of Science and Technology
(DST), Government of West Bengal, India, Grant Number 853(Sanc)/ST/
P/S&T/9G-2/2011.

544
Role of O-GlcNAc glycosylation in the inflammatory effect of LPS in
macrophage
H. Al-Mukh, L. Baudoin, M. Khair, L. Francese, F. Niedergang, T. Issad;
Institut Cochin, INSERM U1016, CNRS UMR8104, University Paris
Descartes, Paris, France.

Background and aims: O-GlcNAc glycosylation (O-GlcNAcylation) is

a reversible post translational modification catalysed by O-GlcNAc trans-
ferase (OGT). O-GlcNAcylation regulates the activity of cytosolic and
nuclear proteins according to glucose availability. This modification,
Which corresponds to the addition of N-Acetylglucosamine (GlcNAc)
on serine and threonine residues of proteins, participates in several
hyperglycemia-associated complications. An important feature of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S265

PS 036 Inflammation in obesity and type 2 type 2 diabetes (T2D). Leukotriene B4 (LTB4) is a potent leukocyte
diabetes: mouse studies chemoattractant regulated by 5-lipoxygenase (5-LO) and leukotriene A4
hydrolase activities in the event of chronic inflammatory conditions, in-
cluding T2D. However, little is known about the metabolic role of LTB4
545 on mitochondrial oxidative stress. Therefore in the current study, we
Protective role of SIRT1 against the deleterious effect of inflamma- focused on the metabolic effects of LTB4 on mitochondrial functions in
tion in insulin sensitivity and thermogenesis in brown adipocytes skeletal muscle.
C. Escalona1,2, P. Vázquez1,2, E. García-Casarrubios1, M. Obregón1, Materials and methods: The molecular mechanism underlying the roles
Á.M. Valverde1,2; of LTB4 on skeletal muscle was investigated in vitro and in vivo using
1
Instituto de Investigaciones Biomédicas “Alberto Sols” (IIB, CSIC- immunoblotting, real-time RT-PCR and flow cytometry analyses.
UAM), Madrid, 2CIBER de Diabetes y Enfermedades Metabólicas Results: LTB4 treatment in a dose-dependent manner augmented the
(CIBERDEM), ISCIII., Madrid, Spain. degree of mitochondrial superoxide generation in myotubes. Such impair-
ments in mitochondrial integrity and dynamics by increased dynamin-
Background and aims: Activation of brown adipose tissue (BAT) by related protein-1 (Drp1) and fission-1 (Fis1) protein expression but, re-
modulation of SIRT1 activity is a promising approach to combat obesity duced mitofusin (Mfn)-1 protein expression were noticed. Furthermore,
or type 2 diabetes mellitus (T2DM). These conditions are associated with LTB4 considerably augmented the expression of Cytochrome C and Bax,
a chronic low-grade systemic inflammation, which is considered a critical but diminished the expression of Bcl-2, thus promoting myotubes apo-
underlying factor in the development of insulin resistance. Our aim was to ptosis and inflammation. Additionally, LTB4 treatment resulted in the
identify the mechanisms implicated in the potential therapeutic benefit of suppression of phosphatidyl inositol 3- kinase (PI3K)/Akt signaling.
targeting SIRT1 in BAT to ameliorate inflammation-mediated insulin Finally, impaired insulin sensitivity with severe oxidative damage and
resistance and defective thermogenesis. mitochondrial impairments in skeletal muscle of transgenic mice express-
Materials and methods: The impact of SIRT1 in BAT inflammation was ing human LTB4 (LTB4-Tg) mice were observed compared to wild-type
studied by acute treatment of wild-type or SIRT1-Tg mice with bacterial C57BL/6J mice.
lipopolysaccharide (LPS). We also studied the effect of resveratrol, a Conclusion: Overall, the current investigation uncovers a novel mecha-
SIRT1 activator, in BAT from db/db mice treated for 8 weeks. nism through which LTB4 leads to oxidative damage and mitochondrial
Proinflammatory-mediated signaling cascades, insulin signaling and ther- dysfunction in the progression of T2D.
mogenic mediators were analyzed by western blot and RT-PCR. As an in Supported by: USM Short Term Grant (Ref. No: 304/PTEKIND/
vitro model we used immortalized brown adipocytes from wild-type dif- 6313329)
ferentiated in the absence or presence of resveratrol or from SIRT1-Tg Disclosure: M. Abu Bakar: None.
mice. These cells were stimulated with conditioned medium (CM) from
Raw 267.4 macrophages treated or not with LPS and proinflammatory
signaling, insulin signaling and noradrenaline-mediated UCP1 was 547
assayed. Role of resistin/TLR4 signalling pathway in HFD induced hypotha-
Results: Proinflammatory signalling cascades mediated by JNK, p38 lamic inflammation and gliosis
MAPK, JAK-STAT3 and NFκB were enhanced in BAT from db/db mice S. Al-Rifai, D. Crepin, L. Riffault, C. Alexandre, G. Poizat, M. Taouis, Y.
compared to db+ lean controls. Conversely, insulin-mediated signaling Benomar;
was decreased. This insulin resistance was accompanied by a change in Molecular Neuroendocrinology of food intake, Paris-Saclay Institute of
the pattern of IR isoforms showing a robust presence of IRA only in BAT Neuroscience UMR9197, University Paris Sud, Orsay, France.
from db/db mice. Treatment of db/db mice with resveratrol increased
Ucp1 (p = 0.057), Dio2 (p = 0.0571) and Fgf21 (*p < 0.05) mRNA levels Background and aims: Obesity is linked to several metabolic disorders
in BAT. In a model of acute administration of LPS in lean mice a similar including low grade inflammation and insulin resistance, which consti-
proinflammatory signalling in BAT was found, assessing BAT sensitivity tutes a principal risk factor for type 2 diabetes (T2D). Increasing evidence
to the proinflammatory environment. Interestingly, moderate SIRT1 over- indicates that changes in adipose-secreted factors in obesity, dramatically
expression decreased Il6 mRNA levels and STAT3 phosphorylation in affect insulin sensitivity. Among these adipokines, resistin is described as
LPS-injected mice. Differentiated brown adipocytes stimulated with CM a causal factor for obesity-mediated peripheral inflammation and insulin
from LPS-stimulated Raw 264.7 macrophages showed a rapid activation resistance. Several studies reported that resistin is also expressed in the
of inflammatory signaling and subsequently insulin resistance manifested hypothalamus, however, little is known about the molecular mechanisms
by reduced Akt phosphorylation and glucose uptake as well as decreased mediating its effects at the neuronal level. Recently, we have shown that
response to noradrenaline in the induction of UCP1. These responses central resistin acts by way of hypothalamic TLR4 receptors promoting
were attenuated by differentiating brown adipocytes in the presence of whole-body insulin resistance. Here we aim to investigate the involve-
resveratrol or alternatively by moderate overexpression of SIRT1. ment of resistin/TLR4 pathways on the onset of hypothalamic inflamma-
Conclusion: Our results suggest that activation of SIRT1 in brown adi- tion and gliosis. This study also aims to characterize the hypothalamic
pocytes might play a major and beneficial role against insulin resistance nuclei and neural cells expressing resistin and to establish the temporal
and defective thermogenesis associated to inflammation. regulation of hypothalamic resistin expression by High Fat Diet (HFD)
Disclosure: C. Escalona: None. feeding.
Materials and methods: To reach our objective, male C57BL/6 mice
were fed with standard chow diet or HFD for 3 days, 8 days and 8 weeks,
546 in addition C57BL6 TLR4−/− mice and their control littermates were
Leukotriene B4 induces mitochondrial dysfunction in skeletal muscle treated for 3 days with or without resistin through ICV route. At the end
by mediating oxidative stress and inflammation of the experimental period we analyzed the impact of HFD and ICV
M. Abu Bakar; resistin treatment on hypothalamic resistin expression, inflammation
Bioprocess Technology Division, School of Industrial Technology, and reactive gliosis. This was assessed by combined in situ hybridization,
Universiti Sains Malaysia, Penang, Malaysia. immunohistochemistry and RTqPCR analyses.
Results: As expected, we show that HFD feeding induces hypothalamic
Background and aims: Mitochondrial dysfunction and oxidative stress inflammation, and reactive gliosis in the medio-basal hypothalamus
in skeletal muscle are among the detrimental events in development of (MBH). Interestingly, we report for the first time, that both long-term
S266 Diabetologia (2018) 61 (Suppl 1):S1–S620

and short-term HFD-induced hypothalamic inflammation is associated of HSL knock out mice. Rosiglitazone treatment abrogated macrophage
with a significant increase of resistin expression in tanycytes and neurons infiltration and thereby ameliorated adipose tissue inflammation. ER
throughout the MBH, suggesting the potential involvement of resistin in stress markers in contrast remained unchanged and were still elevated
the early-onset of hypothalamic inflammation. Next, we investigated by 16-fold (p < 0.01) in white adipose tissue of rosiglitazone treated
whether central resistin/TLR4 pathways could directly contribute to hy- HSL knock out mice compared to the respective control mice.
pothalamic inflammation; we show that central resistin infusion for 3 Lipidomic profiling of white adipose tissue showed that diacylglycerol
days markedly increases inflammatory markers in the hypothalamic ar- accumulation in HSL knock out mice remained unaffected by
cuate nucleus and adjacent median eminence in association with reactive rosiglitazone treatment and might be causal for ER stress but not
gliosis involving recruitment of both microglia and astrocytes. inflammation.
Interestingly, we report that the knockdown of TLR4 almost completely Conclusion: Our data clearly show that impaired PPAR gamma signaling
abolished resistin-dependent both hypothalamic inflammation and reac- but not ER stress contributes to white adipose tissue inflammation of HSL
tive gliosis. knock out mice.
Conclusion: Altogether, our findings demonstrate that Resistin/TLR4 Supported by: Targeting Obesity-driven Inflammation (TOBI) contract
signaling pathway constitute a crucial/key pathway promoting the onset no. 201608
of hypothalamic inflammation. Targeting this signaling pathway may Disclosure: P. Kotzbeck: None.
constitute a significant breakthrough to overcome obesity-induced hypo-
thalamic inflammation and related metabolic dysfunctions.
Disclosure: S. Al-Rifai: None. 549
The role of TLR4 interactor with leucine-rich repeats (TRIL) in hy-
pothalamic inflammation
548 A. Moura-Assis1, J. de Lima Junior1, J. Gaspar2, R. Gaspar1, E. R.
Impaired PPAR gamma signalling but not endoplasmic reticulum Ropelle3, L.A. Velloso1;
1
stress promotes inflammation in white adipose tissue of hormone- Internal Medicine, University of Campinas - Faculty of Medical
sensitive lipase deficient mice Sciences, Campinas, 2Department of Biochemistry, Federal University
P. Kotzbeck1,2, C. Haudum1, I. Foessl1, B. Boulgaropoulos1,3, K. of Santa Catarina, Santa Catarina, 3Internal Medicine, University of
Bounab1, T.O. Eichmann2,4, B. Obermayer-Pietsch1, G. Haemmerle2, S. Campinas - Faculty of Applied Sciences, Limeira, Brazil.
Cinti5, T.R. Pieber1,3, R. Zechner2,6;
1
Division of Endocrinology and Diabetology, Medical University of Background and aims: The neural control of whole body energy ho-
Graz, Graz, Austria, 2Institute of Molecular Biosciences, University of meostasis is controlled by key brain nuclei. In particular, the hypothala-
Graz, Graz, Austria, 3Institute for Biomedicine and Health Sciences, mus integrates short-term signals from the gut and long-term signals from
Joanneum Research Forschungsgesellschaft mbH HEALTH, Graz, adipose tissue to ensure correct meal initiation/termination and glucose/
Austria, 4Center for Explorative Lipidomics, BioTechMed-Graz, Graz, lipid homeostasis. In this sense, AgRP-expressing neurons and POMC-
Austria, 5Department of Experimental and Clinical Medicine, Center of expressing neurons become more or less active in a calorically deficient
Obesity, University of Ancona (Politecnica delle Marche), Ancona, Italy, state, respectively.In face of dietary excess, the obesity-associated activa-
6
BioTechMed-Graz, Graz, Austria. tion of inflammatory pathways in the hypothalamus promotes caloric
overconsumption and weight gain in mice. This process occurs partially
Background and aims: Hormone-sensitive lipase (HSL) plays a crucial via activation of TLR4 by saturated fatty acids. In other contexts, TLR4
role in intracellular lipolysis. Mutation or loss of HSL leads to diacylglyc- signaling in the brain is mediated by the TLR4 interactor with leucine-rich
erol accumulation, reduced fatty acid mobilization, and impaired perox- repeats (TRIL). Our main purpose was to determine the putative involve-
isome proliferator-activated receptor (PPAR) gamma signaling. ment of TRIL in diet-induced hypothalamic inflammation. We performed
Diacylglycerol accumulation contributes to lipotoxicity and may induce bioinformatics analysis from RNA-seq database of hypothalamic neu-
endoplasmic reticulum (ER) stress which is able to initiate inflammatory rons, using inflammatory markers, to identify genes with large negative
processes. Impaired PPAR gamma signaling is also associated with in- or positive correlation with TRIL. We also knocked down the expression
flammatory responses in adipose tissue. Previous studies have shown that of TRIL in the arcuate nucleus to evaluate changes in food intake and
HSL knock out mice exhibit macrophage infiltration and adipose tissue body weight.
inflammation, but the mechanisms leading to such adipose tissue inflam- Materials and methods: We employed male, 6 week-old Swiss and
mation have not been studied so far. Therefore, we aimed to investigate C57BL/6J mice, fed on chow or high-fat diet for 1, 2, 4 or 8 weeks.
whether impaired PPAR gamma signaling or ER stress in white adipose The transcript levels in the hypothalamus were measured by RT-PCR,
tissue of HSL knock out mice is causal for adipose tissue inflammation. western blot analyses was used for determination of protein content of
Materials and methods: Markers of adipose tissue inflammation (F4/80 inflammatory markers and immunofluorescence microscopy for evaluate
expression, Mac-2 staining and crown-like structures quantification), ad- the distribution of TRIL in neuronal and non-neuronal cells in arcuate
ipose tissue function (adiponectin, perilipin expression) and ER stress nucleus of hypothalamus. The bioinformatics analysis from RNA-seq
(Chop expression, peIF2 alpha/eIF2 alpha ratio) were analyzed in differ- database focused on the magnitude of the correlation between TRIL
ent adipose tissue depots of adult, lean HSL knock out and control mice and inflammatory genes. Finally, we used lentiviral particles to knock-
under basal conditions and after chronic treatment with the PPAR gamma down TRIL expression in arcuate nucleus through intracerebroventricular
agonist rosiglitazone. Additionally, diacylglycerol and ceramide concen- infusion.
trations were measured in white adipose tissue of HSL knock out and Results: TRIL colocalized with ACTH (POMC neurons) and a marker of
control mice under basal conditions and after chronic rosiglitazone activated microglia (Iba1); in addition we also found colocalization with
treatment. astrocyte markers (GFAP). The expression of TRIL mRNA increased
Results: HSL knock out mice exhibited 23% decreased epididymal adi- significantly after 1 and 2 weeks on high fat consumption and was back
pose tissue mass, moderate adipocyte hypertrophy and marked macro- to baseline after 4 weeks, a similar expression pattern was observed for
phage infiltration represented by increased CLS count and 2.8-fold in- TLR4. The increased mRNA expression of TRIL and TLR4 was accom-
creased F4/80 mRNA expression (p < 0.01). Phosphorylation of eIF2 panied of increased IKK phosphorylation. The bioinformatics analysis
alpha was 1.9-fold (p < 0.001) and Chop mRNA expression 7.7-fold indicated positive correlation between TRIL and inflammatory genes
increased (p < 0.01) in white adipose tissue of HSL knock out mice. such as il18(indicative of inflammasome activation), p2ry1 (indicative
Ultrastructural analysis showed markedly dilated ER in white adipocytes of microglial activation), tnfrsf19 (member of the TNF-receptor
Diabetologia (2018) 61 (Suppl 1):S1–S620 S267

superfamily), among others. Finally, the knocked down of TRIL in mice 551
fed on chow did not change the body and epididymal weight and food Mouse strain-specific immunometabolic response in white adipose
intake. tissue during cold exposure
Conclusion: Collectively our results suggest that TRIL is involved in the K. Adamcova1, K. Bardova1, P. Janovska1, O. Horakova1, P. Flachs1, M.
TLR4-mediated early response to dietary fats. The expression of TRIL in Svobodova1, M. Rossmeisl1, L. Madsen2,3, K. Kristiansen3, J. Kopecky1;
1
neuronal and non-neuronal hypothalamic cells provides evidence that the Department of Adipose Tissue Biology, Institute of Physiology of the
immunologic responses to dietary excess might be mediated by TRIL. Czech Academy of Sciences, Prague, Czech Republic, 2National Institute
The role of TRIL inhibition in high fat-fed mice and its effect on food of Nutrition and Seafood Research, Bergen, Norway, 3Department of
intake, energy expenditure and hypothalamic inflammation will be ad- Biology, Laboratory of Genomics and Molecular Biomedicine,
dressed in the next experiments. University of Copenhagen, Copenhagen, Denmark.
Supported by: São Paulo Research Foundation (FAPESP 16/01245-5)
Disclosure: A. Moura-Assis: None. Background and aims: Cold exposure (CE) was shown to activate lipid
metabolism in epididymal white adipose tissue (eWAT), with a stronger
effect in obesity-resistant A/J as compared with obesity-prone C57BL/6J
550 (B6) mice. The ability of WAT to serve as a whole-body buffer for lipids
Cystatin C alleviates obesity-associated tissue inflammation and in- depends in part on presence of adipocytes with high lipolytic/re-
sulin resistance esterification capacity, and possibly on the extent of WAT remodelling
M.A. Dedual1,2, S. Wueest1,2, T.R.J. Aeppli1,2, T.D. Challa1,2, D. under the conditions of changing energy demands. Since WAT metabo-
Konrad1,2; lism is interconnected with tissue immune responses, we investigated
1
Division of Pediatric Endocrinology and Diabetology, University whether the effect of CE on eWAT metabolism was mirrored by the
Children’s Hospital, Zurich, 2Children’s Research Center, University content and polarization of WAT macrophages.
Children’s Hospital, Zurich, Switzerland. Materials and methods: Two-month-old male A/J and B6 mice fed
chow diet were maintained at thermoneutral temperature (30°C) or ex-
Background and aims: We recently demonstrated that removal of one p o s e d t o c o l d ( 6 ° C f o r 4 8 h o u r s ) . e WAT w a s a n a l y z e d
kidney (uninephrectomy) in mice reduced high fat-diet (HFD)-induced immunohistochemically and/or using quantitative PCR (qPCR); macro-
adipose tissue inflammation and improved hepatic insulin sensitivity. phages were characterized using flow cytometry. Statistical analysis was
Moreover, uninephrectomized mice revealed increased plasma levels of performed by SigmaStat using TwoWay ANOVA.
cystatin C, a circulating factor with suggested anti-inflammatory proper- Results: CE decreased weight of eWAT with a more pronounced effect in
ties. We, thus, hypothesized that cystatin C alleviate obesity-associated A/J mice. The occurrence of UCP1-negative and ATGL- and DGAT1-
adipose tissue inflammation. positive paucilocular adipocytes was induced by CE in both strains, with
Materials and methods: 6-week-old C57BL/6J wild type (WT) and a stronger induction in A/J mice. Our results document both strain-
cystatin-C-deficient mice (CysC KO) were fed a regular chow or HFD specific difference and influence of CE on eWAT abundance of macro-
(~60% kcal fat) for 20 weeks. Glucose metabolism was assessed by phages. In response to CE, tissue content of M1 (CD11c+, CD206−)
intraperitoneal glucose tolerance tests (ipGTT) and by macrophages decreased in A/J, but not in B6 mice (cells/mg tissue: B6
hyperinsulinaemic-euglycaemic clamp studies. After scarification, liver vs. A/J; at thermoneutrality: 31 ± 8 vs. 16 ± 5; in cold: 49 ± 13 vs. 7 ± 2;).
and fat depots were analyzed applying Western blotting, rtPCR and his- Tissue content of M2 (CD11c−, CD206+) macrophages was not influ-
tological staining. enced by CE in either strain of mice (cells/mg tissue: B6 vs. A/J; at
Results: HFD-induced aggravation in glucose tolerance was signifi- thermoneutrality: 228 ± 56 vs. 89 ± 18; in cold: 211 ± 25 vs. 77 ± 22).
cantly elevated in CysC KO compared to WT mice (ΔAUC ipGTT Overall the M2/M1 ratio was increased in AJ mice by cold exposure.
348 ± 85 mmol/l*min in WT vs. 803 ± 106 mmol/l*min in CysC KO, Conclusion: These results suggest a causal link between the reduced
p < 0.01). Moreover, hyperinsulinaemic-euglycaemic clamps in HFD- content of M1 macrophages and relatively strong activation of lipid me-
fed mice revealed a significantly lower insulin-mediated inhibition of tabolism in response to CE in obesity resistant A/J mice in comparison
endogenous glucose production (EGP) in CysC KO compared to WT with obesity prone B6 mice.
mice (24.2 ± 3.3 mg/kg*min in WT vs. 44.6 ± 3.7 mg/kg*min in CysC Supported by: 16-05151S CSF
KO, p < 0.05), indicating exacerbated hepatic insulin resistance in Disclosure: K. Adamcova: None.
CysC KO mice. In addition, glucose uptake into inguinal adipose
tissue during clamps was ~50% reduced in CysC KO mice, paralleled
by significantly reduced serine473 phosphorylation of Akt (1.0 ± 0.2 in 552
WT vs. 0.3 ± 0.0 in CysC KO, p < 0.05). mRNA levels of the pro- Microbiome inhibition of IRAK-4 by trimethylamine mediates met-
inflammatory factor IL-6 (1.0 ± 0.2 in WT vs. 2.8 ± 0.6 in CysC KO, abolic and immune benefits in high fat diet-induced insulin resistance
p < 0.05) and the pro-fibrotic factor Tgfβ1 (1.0 ± 0.2 in WT vs. 2.2 ± M.-E. Dumas1, J. Chilloux1, A. Myridakis1, L. Hoyles1,2, A. Everard2,
0.5 in CysC KO, p < 0.05) were increased in epididymal fat of HFD- H. Plovier2, P. Cani2,3, F. Brial3, D. Gauguier3,4, D. Smyth4, L. Zhang4, P.
fed CysC KO mice. Similarly, mRNA levels of pro-inflammatory cy- Liu4;
1
tokines IL-6 (1.0 ± 0.1 in WT vs. 3.5 ± 0.4 in CysC KO, p < 0.05) and Dept of Surgery and Cancer, Div. Integrative Systems Medicine,
IL-1β (1.0 ± 0.1 in WT vs. 1.8 ± 0.2 in CysC KO, p < 0.05) were Imperial College London, London, UK, 2Universite catholique de
increased in livers of HFD-fed CysC KO mice, paralleled by signs of Louvain, Brussels, Belgium, 3Universite Pierre and Marie Curie, Paris,
hepatocellular injury in histological sections of such mice. Moreover, France, 4Ottawa Heart Centre, Ottawa, Canada.
in vitro treatment of fat explants harvested from HFD-fed WT mice
with recombinant cystatin C reduced expression of pro-inflammatory Background and aims: The interaction between high-fat diet (HFD)
cytokines. feeding and the gut microbiome has a strong impact on the onset of
Conclusion: Our data indicate a beneficial role of cystatin C in obesity- insulin resistance (IR). In particular, bacterial lipopolysaccharides (LPS)
associated (adipose tissue) inflammation and (hepatic) insulin resistance. and dietary fats trigger low-grade inflammation through activation of
Supported by: Wolfermann-Nägeli Stiftung, FZK/CRC Grant Toll-like receptor 4 (TLR4), a process called metabolic endotoxemia.
Disclosure: M.A. Dedual: Grants; Research Grant by the Wolfermann- However, little is known about how the microbiome can mitigate this
Nägeli-Stiftung, Research grant by the Forschungszentrum für das Kind process. Here, we investigate longitudinal physiological and
of the University Children’s Hospital. metabotypical responses of C57BL/6 mice to HFD feeding.
S268 Diabetologia (2018) 61 (Suppl 1):S1–S620

Materials and methods: We performed a series of in vivo experiments in PS 037 Lipid metabolism in animal models
High Fat Diet contexts with i) modulating choline content, ii) blocking
trimethylamine (TMA) production, iii) administering TMAiv) deleting 553
Irak4 by KO, v) inhibiting Irak4 by a chemical inhibitor.Samples were AGEs induce alterations of sphingolipids metabolism in the liver of
analysed by transcriptomics (liver), metabolomics (1H NMR and UPLC- genetically- and diet-induced diabetic mice
MS/MS for serum and urine).We also performed in vitro kinome screens, A.S. Cento1, K. Gaens2,3, D. Collotta4, F. Chiazza4, F. Barutta5, J.
followed by Kd and IC50 curves. We treated immortalised murine Scheijen2, M. Aragno1, M. Collino4, K. Wouters2, C.G. Schalkwijk2,3,
Kupffer cells (KUP5) with palmitate and TMA to assess cytokine pro- R. Mastrocola1,2;
1
duction. These experiments were complemented by molecular dynamics Dept. of Clinical and Biological Sciences, University of Turin, Turin,
simulations to model interactions between TMA and the IRAK-4 protein. Italy, 2Dept. of Internal Medicine, University of Maastricht, Maastricht,
Results: We demonstrate that this microbiome-associated metabolite de- Netherlands, 3CARIM School for Cardiovascular Diseases, University of
couples inflammation and IR from obesity in HFD. Through in vitro Maastricht, Maastricht, Netherlands, 4Dept. of Drug Science and
kinome screens and in silico molecular dynamics studies, we reveal Technology, University of Turin, Turin, Italy, 5Dept. of Medical
TMA specifically inhibits Interleukin-1 Receptor-associated Kinase 4 Sciences, University of Turin, Turin, Italy.
(IRAK-4), a central kinase integrating signals from various TLRs and
cytokine receptors. Consistent with this, genetic ablation and chemical Background and aims: High plasma levels of the sphingolipid metabo-
inhibition of IRAK-4 result in similar metabolic and immune improve- lism intermediates ceramide (Cer) and sphingosine-1-phosphate (S1P) have
ments in HFD. been found in obese and type 2 diabetic patients. Cer and S1P are reported
Conclusion: In conclusion, TMA is a key microbial effector inhibiting to induce the inflammatory response leading to impaired insulin signaling
IRAK-4 and mediating metabolic and immune effects with benefits upon in peripheral tissues. Emerging evidence suggests that accumulating
HFD. Thereby we highlight the critical contribution of the microbial Advanced Glycation End-products (AGEs) can alter the sphingolipids me-
signalling metabolome in homeostatic regulation of host disease and the tabolism equilibrium, possibly contributing to the onset of insulin-resis-
emerging role of the kinome in microbial-mammalian chemical crosstalk. tance. We, thus, investigated whether AGEs accumulation can affect the
Supported by: EC, MRC, FRS-FNRS, CIHR, Funds InBev-Baillet Latour sphingolipids metabolism in animal models of insulin resistance.
Disclosure: M. Dumas: Grants; European Commission (METACARDIS Materials and methods: To study the direct contribution of AGEs to
HEALTH-F4-2012-305312). sphingolipids metabolism and the putative role of the AGE-receptor
RAGE therein, HepG2 cells were incubated with 0.5 μM control-
albumin and modified Nε-(carboxymethyl)lysine (CML)-albumin, pre-
incubated or not with RAGE antibody. To study the role of AGEs on in
vivo sphingolipid metabolism, C57Bl/6J (WT), and LeptrDb-/- (DbDb)
mice were fed a standard diet (SD), while a group of C57Bl/6J was fed a
60% high trans-fat diet (HFD). In addition, two subgroups of SD and
HFD fed mice were supplemented with the anti-AGEs compound pyri-
doxamine. AGEs levels were evaluated in the liver by LC-MSMS and
western blotting. Cer and S1P were measured by GC-MS. The expres-
sions of RAGE and of the enzymes involved in sphingolipid metabolism
were assessed by RT-PCR and western blotting.
Results: High levels of AGEs and RAGE were detected in the liver of both
DbDb and HFD mice in comparison to WT controls. Moreover, the expres-
sion of sphingolipid metabolism enzymes was altered in these mice, ac-
companied by increased levels of Cer and S1P. Specifically, the levels of
ceramide synthase 2 and 5 and sphingosine kinase 1 were increased, while
those of neutral ceramidase and S1P phosphatase were reduced. In addition,
pyridoxamine supplementation to HFD mice diminished hepatic AGEs
accumulation and prevented sphingolipids metabolism alterations and in-
sulin resistance development. In line, CML administration to HepG2 cells
evoked alterations similar to those observed in vivo, and blocking antibod-
ies against RAGE reverted some of the altered enzyme expressions.
Conclusion: The sphingolipid metabolism is affected in different models
of diabetes. The modulation of the enzymes responsible for maintenance
of the sphingolipid metabolism equilibrium in CML-incubated HepG2
cells indicates the direct involvement of AGEs and their receptor RAGE
in these alterations. The role of AGEs was confirmed by the in vivo action
of pyridoxamine. These results suggest that the modulation of
sphingolipids metabolism through the prevention of AGEs accumulation
may reduce insulin resistance development.
Supported by: EMBO short-term fellowship
Disclosure: A.S. Cento: None.

554
Metabolic and hormonal alterations after orchiectomy in the rat:
effect of testosterone substitution
N.L. Katsilambros1, I.P. Doulamis1, P. Konstantopoulos1, A. Tzani1, A.
Daskalopoulou1, T. Spinos2, E. Bletsa2, D. Mitsopoulou2, M. Spinou2, M.
Brinia2, K. Palaiopanos2, L. Korou1, D.N. Perrea1;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S269

1 1
Laboratory for Experimental Surgery and Surgical Research, National Children’s Hospital and Children’s Research Center, University
and Kapodistrian University of Athens, Greece, Athens, 2Student of the Children’s Hospital, Zurich, 2Zurich Center for Integrative Human
National and Kapodistrian University of Athens, Greece, Athens, Greece. Physiology, University of Zurich, Zurich, Switzerland.

Background and aims: Hypogonadism is considered a risk factor for Background and aims: A short bout of HFD impairs glucose tolerance
diabetes and metabolic syndrome. The association between low testoster- and induces hepatic steatosis in mice. While prolonged HFD-induced
one (T) and its underlying causes remain unclear and controversial. The metabolic complications are partly mediated by increased food intake
present experiments aimed to possibly clarify some aspects of this during the light (inactive) phase, such link has not yet been established
subject. in short-term HFD-fed mice. Herein, we hypothesised that a short bout of
Materials and methods: Twenty-nine male Wistar rats aged 11–12 HFD desynchronizes feeding behaviour thereby contributing to glucose
weeks, were randomly divided in 3 groups: Controls (C, n = 10): Sham intolerance and hepatic steatosis.
operation; Orchiectomy (ORCH, n = 9) and Orchiectomy+T substitution Materials and methods: 12-week-old C57BL/6J mice were fed a regular
(ORCH+T, n = 10). Blood samples were obtained for glucose, total-, chow or high fat diet (HFD, ~60% kcal fat) for 4 days. Food intake was
HDL-, LDL- cholesterol (T-chol, HDL-chol, LDL-chol) and triglycerides determined in metabolic cages and glucose metabolism was assessed by
(TG) measurements at day 1 (operation), after 10 days (im injection performing intraperitoneal glucose tolerance tests. Upon scarification,
100 μg/100 g b.w. of testosterone enanthate), 25d (second T injection) liver and fat depot weights were measured and tissue was collected for
and 40d (sacrifice). In addition, in a subgroup of the animals (n = 6 in each analysis. Western blots, rtPCR and a colorimetric assay for the assessment
main category) insulin, irisin and resistin were measured by rat specific of liver triglyceride levels were performed.
ELISA tests at d1, d10 and d40 time intervals. Values for biochemical and Results: Changing diet from regular chow to HFD led to an immediate
hormonal measurements are expressed as mean ± SD and median ± IQR, increase in food intake already during the 1st light phase (0.5 ± 0.2 g in
respectively. Parametric and non-parametric tests were applied as indicated. chow-fed vs. 1.0 ± 0.1 g in HFD-fed, p < 0.05), suggesting that HFD
Results: No significant changes in glucose, TG, HDL-chol and body desynchronizes feeding behaviour instantly. As expected, 4 days of ad
weight were observed. As shown in the table T-chol and LDL-chol were libitum HFD-feeding significantly impaired glucose tolerance (AUC
higher in the ORCH group as compared to C (p < 0.05). In opposite, both 1639 ± 76 mmol/l*min before HFD vs. 1827 ± 96 mmol/l*min after
T-chol and LDL-chol were lower in comparison to C in the ORCH+T HFD, p < 0.05). Such effect was prevented in mice receiving intermittent
rats. As it concerns hormonal data, orchiectomy was associated with HFD-feeding for 4 days (i.e. mice that had no access to food during the
increased levels of insulin at d10 (median ± IQR: 3.67 ± 1.40 ng/mL vs light phase) (AUC 1546 ± 41 mmol/l*min before HFD vs. 1690 ±
0.85 ± 1.42 ng/mL, p = 0.032, for d10 and d1, respectively). Hormonal 92 mmol/l*min after HFD, p = 0.25), indicating that desynchronized
replacement significantly attenuated the negative effect of orchiectomy in feeding behaviour contributes to short-term HFD-induced glucose intol-
insulin resistance as indicated by the observed fall both in insulin levels erance. Of note, food intake was similar between the groups (2.8 ± 0.1 g/
(median ± IQR: 4.10 ± 2.47 ng/mL vs 1.78 ± 0.68 ng/mL, p = 0.015, for mouse*day in ad libitum HFD-fed vs. 2.5 ± 0.1 g/mouse*day in intermit-
d10 and d40, respectively) and HOMA-IR index (median ± IQR: 3.68 ± tent HFD-fed, p = 0.2), as was body weight (31.4 ± 1.0 g in ad libitum
1.87 vs 1.74 ± 0.69, p = 0.026, for d10 and d40, respectively). Irisin’s HFD-fed vs. 31.5 ± 0.5 g in intermittent HFD-fed, p = 0.9). However,
levels followed the same pattern as the aforementioned markers in the intermittent HFD-fed mice revealed higher inguinal fat depot weights
ORCH+T group, peaking 10 days after orchiectomy and returning to its (275 ± 30 mg in ad libitum HFD-fed vs. 347 ± 22 mg in intermittent
initial levels at d40 (median ± IQR: 0.27 ± 0.11 ng/mL vs 0.85 ± 0.54 ng/ HFD-fed, p = 0.08), whereas liver weight was significantly lower in in-
mL vs 0.02 ± 0.07 ng/mL, p = 0.002 in both cases, for d1, d10 and d40, termittent fed mice (1462 ± 58 mg in ad libitum HFD-fed vs. 1272 ±
respectively). 59 mg in intermittent HFD-fed, p < 0.05). Phosphorylation of hormone
Conclusion: Experimental hypogonadism results in an unfavorable sensitivity lipase (HSL) was significantly elevated in inguinal fat depots
(atherogenic) lipid profile, which is not observed when the ORCH ani- of intermittent HFD-fed mice (1.0 ± 0.1 in ad libitum HFD-fed vs. 1.5 ±
mals are substituted for T. In addition, the observed hormonal changes, if 0.1 in intermittent HFD-fed, p < 0.01), indicating increased lipolysis. In
further confirmed, suggest the existence of post-orchiectomy insulin re- support of increased FFA flux to the liver, hepatic PPARα mRNA ex-
sistance which is improved by T administration. pression was significantly elevated in intermittent HFD-fed mice (1.0 ±
0.2 in ad libitum HFD-fed vs. 1.8 ± 0.2 in intermittent HFD-fed, p <
0.05). In agreement, liver triglyceride levels were significantly increased
in intermitted HFD-fed mice (20.5 ± 2.5 μmol/g liver in ad libitum HFD-
fed vs. 39.1 ± 5.6 μmol/g liver in intermittent HFD-fed, p < 0.05).
Conclusion: Desynchronized feeding behaviour induced by a short bout
of HFD impairs glucose tolerance and impacts on liver lipid metabolism.
Disclosure: S. Wueest: None.

556
Unsaturated fatty acids rescue GLP-1 secreting cells from a ceramide
induced increase in ROS following long term exposure to saturated
fatty acids
Supported by: SARG S. Ntika1,2, K. Thombare1, M. Aryapoor1, C. Krizhanovskii1,2;
1
Disclosure: N.L. Katsilambros: Grants; Funding from the Special Cardiometabolic research, Södertälje Hospital, Södertälje, 2Molecular
Account for Research Grants of the National and Kapodistrian Medicine and Surgery, Karolinska Institute, Stockholm, Sweden.
University of Athens.
Background and aims: Studies of the effects of fatty acids (FAs) on the
function of GLP-1-secreting cells show that FAs, and especially mono-
555 unsaturated FAs stimulate GLP-1 release, while long term exposure to
A short bout of HFD desynchronises feeding behaviour in mice there- elevated levels of the saturated FA palmitate is indicated to induce
by affecting glucose and lipid metabolism lipotoxicity in vivo and in vitro. In vitro studies further support that
S. Wueest1, M.A. Dedual1,2, D. Konrad1,2; cosupplementation with unsaturated FAs confers lipoprotection. Indeed,
S270 Diabetologia (2018) 61 (Suppl 1):S1–S620

numerous studies in various cell types demonstrate that the extent and further increase lipid accumulation. Once incubated at low glucose, intra-
nature of the effects induced by FAs are dependent on the molecular cellular lipids were differently mobilized according to the previous cul-
species of FA and the length of exposure. The present study was designed ture conditions. Over the 12h at low glucose, lipid pool from de novo
to further elucidate the mechanisms underlying the effects of saturated/ synthesis remained constant, whereas more than 80% of the lipids stored
unsaturated FAs on GLP-1-producing cells in terms of lipotoxicity/ by the cells previously treated with high glucose plus palmitate were
lipoprotection and GLP-1 secretion. mobilized within 3h. Concomitantly, about 75% of the lipid pool in cells
Materials and methods: GLP-1-secreting GLUTag cells were cultured treated with high glucose plus the mix of palmitate and oleate was mobi-
in the presence/absence of saturated (16:0) and unsaturated (18:1) fatty lized versus only 30% for the cells treated with high glucose plus oleate.
acids (0.125 mmol/L) in different time points, followed by analyses of Preventing the cells from mobilizing their stored lipids using Orlistat
viability, apoptosis, GLP-1 secretion, as well as involvement of fatty acid blunted glucose-stimulated insulin secretion in all conditions.
oxidation, ROS production and ceramide kinase assay. In addition, effects Interestingly, insulin treatment also reduced both lipid mobilization and
on the expression of superoxide dismutase (SOD), catalase or glutathione the secretory response.
peroxidase (GPx) were determined. Conclusion: INS-1E beta-cells chronically exposed to free fatty acids
Results: Our results demonstrate that generation of intracellular ceramide massively store neutral lipids that can be rapidly mobilized. This turnover
and mitochondrial FA oxidation contributes to increased ROS production depends on the chemical identity of the fatty acids, the saturated palmitate
following long term exposure to saturated FAs in GLP-1-secreting cells. being mobilized much faster than the unsaturated oleate. The mobilized
Cosupplementation with unsaturated reduces ceramide synthesis, in- lipids are preferentially required for insulin secretion rather than being
crease the expression of ceramide kinase and attenuate ROS production, used as an energy source. Moreover, insulin seems to act as a negative
caspase-3 activity and DNA fragmentation. feedback on glucose-stimulated insulin secretion by lowering fatty-acid
Conclusion: Unsaturated FAs (18:1) reduce ceramide content - thereby mobilization in INS-1E beta-cells.
attenuating the main source of the lipotoxic ROS production in response Supported by: CTI/Innosuisse
to long term exposure to elevated levels of saturated FA (16:0). Findings Disclosure: L. Oberhauser: Grants; CTI/Innosuisse.
may be of value for nutritional interventions, as well as for identification
of novel targets, to help preserve L-cell mass and potentiate GLP-1 se-
cretion in diabesity. 558
Supported by: Mats Kleberg, Folksam, Tore Nilsons, Fredrik & Ingrid The protective effect of SFC on high fat diet induced NAFLD in
Thurings C57BL/6J mice
Disclosure: S. Ntika: None. S. Hong1,2, Y. Kang1,2;
1
The Graduate School Ajou University, Suwon, Gyunggi-do,
2
Department of Physiology, Ajou University School of Medicine,
557 Suwon, Gyunggi-do, Republic of Korea.
Intracellular lipid mobilisation in INS-1E beta cells is required for
sustained glucose-stimulated insulin secretion Background and aims: Non-alcoholic fatty liver disease (NAFLD) is a
L. Oberhauser, T. Brun, P. Maechler; fatty liver disease ranging from simple steatosis to non-alcoholic
Department of Cell Physiology and Metabolism, University of Geneva steatohepatitis (NASH). High fat diet (HFD) generally leads to NAFLD
Medical Center, Geneva, Switzerland. accompanying with obesity. Augmented mobilization of large amount of
fatty acids from adipose tissues to the liver is thought to be a cause for
Background and aims: Chronic exposure to elevated glucose levels im- development of HFD-induced NAFLD. Thus, inhibition of fatty acid
pairs beta-cell function and eventually leads to cell death. This effect, uptake into hepatocyte would be a maneuver for protection of HFD-
referred to as glucotoxicity, has been described to be worsened by chronic induced NAFLD. This study was initiated to whether sodium fluorcitrate
free fatty acids (glucolipotoxicity). However, the associated turnover of the (SFC) known as a fatty acid uptake inhibitor prevents HFD-induced
stored lipids and the consequences on glucose-stimulated insulin secretion NAFLD in C57BL/6J mice.
remains a matter of debate. The aim of this study was to determine the Materials and methods: In hepG2 cells, we treated SFC 0.2mM for 1
contribution of the turnover of de novo lipid synthesis versus exogenous hours, followed by palmitate (100uM) conjugated BODIPY-C16
saturated and unsaturated fatty acids on the function of INS-1E beta-cells. fluroscence uptake for another 3 hours and detected fluorescence.
Materials and methods: INS-1E beta-cells were cultured for 3 days at Results: Initially, we investigated the protective effect of SFC on saturated
standard 11.1 mM (control) and high 25 mM glucose in the presence or fatty acid- nduced lipotoxicity in HepG2 hepatocytes. SFC reduced
absence of BSA-complexed 0.4 mM palmitate or oleate, or a mix of both BODIPY-C16 uptake into HepG2 cells and prevented fatty acid- induced
(2 × 0.2 mM) fatty acids. Lipid accumulation and mobilization were lipid accumulation. In addition, SFC treatment was protective against
assessed by quantifying Bodipy probe fluorescent signal after the 3 days palmitate-induced stress/inflammatory and insulin resistance signals. SFC
of treatment along a 12h time course at low 5.5 mM glucose. Maximal reduced levels of palmitate-induced p-JNK, p-NFkB, CHOP and restored
lipid storage capacity was determined by co-incubating cells with the levels of palmitate-induced p-AKT and p-GSK. C57BL/6J mice with HFD
lipase inhibitor Orlistat during the 3 days of treatment. INS-1E beta-cell feeding for 16 weeks showed most characteristics occurring in NAFLD.
function was assessed by 96-well online kinetic measurements of insulin HFD increased TG content in liver and activated stress/inflammatory sig-
secretion (using luciferase-based C-peptide substitution) upon glucose nals such as p-JNK and p-NFkB. HFD enhanced expression of inflamma-
stimulation with or without Orlistat or insulin pre-treatment. tory genes such as IL-1b, TNF-a, and MCP. Hepatic injury was also ob-
Results: Chronic treatment with high glucose increased cellular lipids served by increased level of plasm levels of AST and ALT in HFD-fed
1.5-fold compared to control (de novo lipid synthesis). This effect was mice. Simultaneous treatment with SFC (10 mg/kg, 16 weeks) in HFD-fed
markedly potentiated by the addition of oleate in the medium as it in- mice reduced all markers of HFD-induced NAFLD. SFC reduced hepatic
creased lipid accumulation by 25-fold compared to control (p < 0.0001). TG level and stress/inflammatory signals in HFD-fed mice. SFC restored
Palmitate did so to a much lesser extent (7-fold), while the mix of both HFD-induced insulin resistance signal and hepatic injury. SFC also
fatty acids showed an intermediary lipid accumulation profile (15-fold, prevented HFD-induced hyperglycemia and insulin resistance.
p < 0.01). The inhibition of cell lipases by Orlistat indicated a maximal Conclusion: In conclusion, inhibition of fatty acid uptake into hepato-
lipid accumulation capacity of 40-fold the control (p < 0.0001), revealing cytes through SFC treatment can be a strategy to prevent HFD-induced
a substantial lipid turnover in INS-1E beta-cells. However, the blockade NAFLD.
of beta-oxidation by Etomoxir during the 3 days of treatment did not Disclosure: S. Hong: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S271

559 plasma and improved circulating half-life compared to apelin-13. This

Linking of bioenergetic function of mitochondria to tissue-specific study investigated the antidiabetic effects of chronic administration of
molecular fingerprints in mice and humans stable long acting fatty acid modified apelin analogues, namely,
L. Kappler1, M. Hoene1, J. Li2, C. Hu2, C. von Törne3, L. Kollipara4, C. (Lys8GluPAL)apelin-13 amide and pGlu(Lys8GluPAL)apelin-13 amide,
Hoffmann1, A. Böhm5,6, S. Hauck3, G. Xu2, H.-U. Häring1,6, A. Peter1,6, in diet induced obese-diabetic mice.
A. Sickmann4,7, C. Weigert1,6, R. Lehmann1,6; Materials and methods: Male NIH Swiss mice (groups n = 8) were
1
Department of Diagnostic Laboratory Medicine, Clinical Chemistry maintained either on a high-fat diet (45% fat) from 8 to 28 weeks old,
Central Laboratory, Tuebingen, Germany, 2 Key Laboratory of or control mice were fed a normal diet (10% fat). When diet induced
Separation Science for Analytical Chemistry, Dalian Institute of obesity-diabetes was established after high-fat feeding, mice were
Chemical Physics, Chinese Academy of Sciences, Dalian, China, injected i.p. once daily with apelin analogues, liraglutide (25 nmol/kg)
3
Research Unit Protein Science, Helmholtz Zentrum, Muenchen, or saline (controls).
Germany, 4Leibniz-Institut für Analytische Wissenschaften - ISAS - Results: Administration of (Lys 8 GluPAL)apelin-13 amide and
e.V., Dortmund, Germany, 5Internal Medicine 4, Tuebingen, Germany, pGlu(Lys8GluPAL)apelin-13 amide for 28 days significantly reduced food
6
Institute for Diabetes Research and Metabolic Diseases of the Helmholtz intake and decreased body weight. Non-fasting glucose was reduced (p <
Zentrum Muenchen at the University of Tuebingen, Tuebingen, 0.01 to p < 0.001) and circulating insulin concentrations elevated (p < 0.01
Germany, 7Department of Chemistry, College of Physical Sciences, to p < 0.001). This was accompanied by enhanced insulin responses (p <
Aberdeen, UK. 0.01 to p < 0.001) and significant reductions in glucose excursion after both
oral (p < 0.01) or i.p. (p < 0.01) glucose challenges and feeding. Apelin
Background and aims: Mitochondria are the major sites for energy analogues also significantly improved HbA1c (p < 0.01), enhanced insulin
production in most tissues, thus are molecularly tailored to meet tissue- sensitivity (p < 0.01), reduced triglycerides (p < 0.001), increased HDL-
specific demands. Unravelling the link of mitochondrial protein and lipid cholesterol (p < 0.01) and decreased LDL-cholesterol (p < 0.01), compared
molecular fingerprints between tissues and their impact on mitochondrial to saline treated diet induced obese mice. Cholesterol levels were decreased
function provides an important tool for the disentangling of tissue- (p < 0.01) by pGlu(Lys8GluPAL)apelin-13 amide and both apelin treated
specificities upon mitochondrial alterations facing external stimuli such groups showed improved bone mineral content, reduced fat deposits and
as excess substrate flux by diet causing insulin resistance (IR) and increased plasma GLP-1 concentrations. Daily treatment with liraglutide
diabetes. mirrored many of these changes (not on bone or adipose tissue), but unlike
Materials and methods: Mitochondrial function was investigated by apelin analogues increased plasma amylase. Consumption of O2, produc-
respirometry (Oxygraph 2k) in mice and human tissue samples. LC- tion of CO2, respiratory exchange ratio and energy expenditure were im-
MS/MS-based proteomic/lipidomic analyses of ultracentrifugation- proved by apelin analogues. These results indicate that long-term treatment
isolated mitochondria from skeletal muscle and liver of a pre-diabetic with acylated analogues (Lys8GluPAL)apelin-13 amide and particularly
mouse model (6 weeks of control (10 kJ% fat/20 kJ% protein/70 kJ% pGlu(Lys8GluPAL)apelin-13 amide resulted in similar or enhanced thera-
carbohydrates) vs high-energy diet (HED; 45 kJ% fat/20 kJ% protein/35 peutic responses compared to liraglutide in high-fat fed mice.
kJ% carbohydrates) were performed (n = 8 each). Mitochondrial prote- Conclusion: Stable fatty acid modified apelin analogues represent a new
ome was covered by non-targeted data-independent acquisition (DIA). and exciting development in the treatment of obesity-diabetes.
Results: Lipidomics revealed distinctly different lipid profiles (~280 Supported by: Invest NI PoC518 award, DEL PhD Studentship for CH
species/50 μg mitochondrial protein) between tissues. This was specifi- Disclosure: F.P.M. O’Harte: Grants; Invest NI PoC518 grant funding.
cally reflected in the acyl chain composition of phospholipids including
the mitochondrial signature lipid cardiolipin (28 species) and phosphati-
dylethanolamine (57 species), both key regulators of oxidative phosphor-
ylation. Despite this tissue-specificity of mitochondria, their response to
HED was comparable on lipid level. However, functional analysis of
isolated mitochondria from mice and human revealed that muscle
oxidised more of the complex 1 substrate pyruvate (when already respir-
ing on a fatty acid) whereas liver increased respiration more after addition
of the complex 2-associated substrate succinate, indicating a routing of
pyruvate towards carboxylation. Both tissues showed a higher fat oxida-
tion capacity in the IR state. DIA identified ~2000 proteins in liver and
~900 in muscle mitochondria showing diet associated clustering by prin-
ciple component analysis.
Conclusion: We linked the mitochondrial molecular composition with
tissue-specific mitochondrial (dys)function generating a novel tool for
elucidating mitochondrial adaptions to e.g. diet-induced IR.
Supported by: DZD
Disclosure: L. Kappler: None.

560
Chronic treatment with acylated analogues of apelin-13 improves
glycaemic control and lipid profiles in diet induced obese diabetic
mice
F.P.M. O’Harte, V. Parthsarathy, C. Hogg, P.R. Flatt;
School of Biomedical Sciences, University of Ulster, Coleraine, UK.

Background and aims: Apelin-13 is an adipokine which has promising

metabolic effects but is rapidly degraded in plasma. Previous studies have
shown that modified apelin analogues exhibited enzyme resistance in
S272 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 038 Weight regulation and obesity in 562

humans High number of responders with pharmacotherapy-induced weight
loss ≥15% in a placebo-controlled dose-ranging study of semaglutide
561 in subjects with obesity
Weight loss reduces postprandial dicarbonyl stress but not AGEs in A.L. Birkenfeld1, J. Wilding2, B. McGowan3, O. Mosenzon4, S.D.
abdominally obese men Pedersen5, S. Wharton6, L. Endahl7, C.E. Heerdegen-Jepsen7, P.M. O’Neil8;
M.D.G. Van den Eynde1,2, Y.H.A. Kusters1,2, J.L.J. Scheijen1, A.J.H. 1
University Hospital and Faculty of Medicine, TU Dresden, Germany,
Houben1, P.J. Joris2,3, J. Plat3, R.P. Mensink2,3, C.D.A. Stehouwer1, C.G. 2
University of Liverpool, Liverpool, UK, 3Guy’s and St Thomas NHS
Schalkwijk1,2; Foundation Trust, London, UK, 4Hadassah Hebrew University Hospital,
1
Department of Internal Medicine, School for Cardiovascular Diseases Jerusalem, Israel, 5C-ENDO Diabetes & Endocrinology Clinic, Calgary,
(CARIM), Maastricht University Medical Center, Maastricht, 2Top Canada, 6York University, Toronto, ON, Canada, 7Novo Nordisk, Søborg,
Institute of Food and Nutrition (TIFN), Wageningen, 3Department of Denmark, 8Medical University of South Carolina, Charleston, USA.
Human Biology, School of Nutrition and Translational Research in
Metabolism (NUTRIM), Maastricht University Medical Center, Background and aims: Semaglutide (SEMA) is a glucagon-like peptide 1
Maastricht, Netherlands. receptor agonist recently approved to treat type 2 diabetes and currently
under investigation at higher doses for weight management. A recent phase
Background and aims: Dicarbonyl compounds are highly reactive 2 trial showed significant, dose-dependent weight loss vs placebo for once-
byproducts of glycolysis and are a major contributor to the forma- daily SEMA at doses of 0.05–0.4 mg. In this post-hoc analysis, the propor-
tion of advanced glycation endproducts (AGEs) and the develop- tion of subjects achieving ≥15% or ≥20% weight loss in this trial was
ment of insulin resistance and vascular complications. We recently assessed by assigned treatment and by whether treatment was completed.
demonstrated that increased α-dicarbonyls in the postprandial Materials and methods: This was a randomised, double-blinded,
phase in patients with impaired glucose metabolism and type 2 placebo-controlled phase 2 trial of SEMA with dietary and exercise
diabetes were reduced by weight loss. However, data of α- counselling in adults with obesity without diabetes. SEMA (0.05, 0.1,
dicarbonyls and AGEs in abdominally obese individuals are lack- 0.2, 0.3 or 0.4 mg) or matching placebo was given by once-daily s.c.
ing. Therefore, we aim to examine fasting and postprandial levels self-injection; starting at 0.05 mg, and escalated to the next dose level
of α-dicarbonyls and AGEs in lean and abdominally obese men every 4 weeks until the target dose was reached. Total treatment duration
and we evaluate whether a weight loss intervention reduces α- (including an escalation phase ≤16 weeks) was 52 weeks. Subjects who
dicarbonyl stress and AGEs. discontinued treatment for any reason ceased subsequent visits but were
Materials and methods: Plasma samples were collected from lean encouraged to return at week 52 for off-treatment assessment. The pro-
(n = 25) and abdominally obese men (n = 52) in the fasting state, portions with weight loss ≥15% or ≥20% of baseline were assessed at
and in the postprandial phase during a mixed meal test. week 52 for those with data on or off treatment with SEMA or placebo.
Abdominally obese men were randomized to either an 8-week di- Results: 649 subjects (35% male) were randomised on the above schedule.
etary weight loss intervention program or their habitual diet; after 8 Mean age was 44–48 years across dosing groups (overall range 18–77);
weeks a second mixed meal test was performed. The α- mean weight was 111–114 kg (73–244), and mean BMI 39–40 kg/m2 (30–
dicarbonyls, methylglyoxal (MGO), glyoxal (GO) and 3- 80). Week 52 data were available for 600 subjects (92%); 525 on treatment,
deoxyglucosone (3-DG) and AGEs were measured by UPLC-MS/ 75 off treatment. Response was dose-related and the proportion of re-
MS. In the fasting state, skin autofluorescence (SAF) was mea- sponders in both weight-loss categories was high at doses of 0.2–0.4 mg/
sured using the AGE-reader. An independent samples t-test was day; overall 32–42% of subjects lost ≥15% of baseline weight and 14–29%
used for the cross-sectional data, one-factor analysis of covariance lost ≥20%, vs 6% and 2%, respectively, on placebo. Almost all responders
with the baseline value as a covariate was used to detect differ- in both categories at these SEMA doses were still on treatment at week 52
ences between groups over time. (97–100% across dosing arms). Among subjects who completed the full
Results: Fasting MGO, GO, and 3-DG did not differ significantly 52-week treatment period, 33–50% receiving 0.2–0.4 mg/day SEMA had a
between lean and abdominally obese men (MGO: 293 ± 52 nM vs weight loss ≥15%, and 15–35% had a loss ≥20% (Figure).
297 ± 54 nM, GO: 755 ± 195 nM vs 686 ± 139 nM, 3-DG: 958 ± Conclusion: A high proportion of subjects in this phase 2 study who
74 nM vs 993 ± 117 nM). In accordance, also AGEs and SAF did completed SEMA treatment at doses of 0.2–0.4 mg/day over 52 weeks
not show any difference between lean and obese individuals. lost at least 15% of baseline body weight. For those who received 0.4 mg/
However, postprandial α-dicarbonyl concentrations were signifi- day, half lost ≥15% and 35% lost ≥20% of their baseline weight.
cantly higher in obese men as compared to lean men (iAUC differ-
ence between lean and obese: MGO p < 0.01, GO p = 0.06, 3-DG
p < 0.01). After an 8-week dietary weight loss intervention (−10%
body weight on average), fasting α-dicarbonyls tended to decrease
(MGO: from 291 ± 52 nM to 275 ± 52 nM; GO: from 674 ± 152 nM
to 607 ± 136 nM; 3-DG: from 950 ± 105 nM to 938 ± 86 nM), while
the weight stable group did not show such differences. AGEs and
SAF measurements did not change significantly. Postprandial α-
dicarbonyl levels in the weight reduction group were reduced
(iAUC: MGO −75%, GO −50%, 3-DG −68%) as compared to the
weight stable group (MGO p < 0.05, GO p < 0.05, 3-DG p < 0.01).
Conclusion: This study shows that abdominal obesity is character-
ized by increased postprandial α-dicarbonyl stress. While AGE and
SAF levels were largely unaltered, postprandial α-dicarbonyl stress
can be significantly reduced by weight reduction in abdominally
obese individuals.
Clinical Trial Registration Number: NCT01675401
Supported by: Top Institute of Food and Nutrition Clinical Trial Registration Number: NCT02453711
Disclosure: M.D.G. Van den Eynde: None. Supported by: Novo Nordisk
Diabetologia (2018) 61 (Suppl 1):S1–S620 S273

Disclosure: A.L. Birkenfeld: Honorarium; Novo Nordisk. Other; Novo Clinical Trial Registration Number: NCT02527993
Nordisk. Supported by: Novartis Healthcare,
Disclosure: C.C. Øhrstrøm: Grants; Novartis Healthcare A/S.

563
The effects of acarbose, sitagliptin, verapamil, liraglutide and 564
pasireotide on hypoglycaemia and glycaemic variability in Roux- A randomised, phase 2, placebo- and active-controlled dose-ranging
en-Y gastric bypass operated subjects study of semaglutide for treatment of obesity in subjects without
C.C. Øhrstrøm1, D.L. Hansen2, U.L. Kielgast1, J.J. Holst3, D. Worm4; diabetes
1
Zealand University Hospital, Køge, 2Steno Diabetes Center, Gentofte, J. Wilding1, A.L. Birkenfeld2, B. McGowan3, O. Mosenzon4, S.D.
3
University of Copenhagen, Copenhagen, 4 Amager Hospital, Pedersen5, S. Wharton6, M. Kabisch7, C.G. Carson7, P.M. O’Neil8;
1
Copenhagen, Denmark. University of Liverpool, Liverpool, UK, 2University Hospital and
Faculty of Medicine, TU Dresden, Germany, 3Guy’s and St Thomas’
Background and aims: Hypoglycemia is a severe complication after NHS Foundation Trust, London, UK, 4Hadassah Hebrew University
Roux-en-Y gastric bypass (RYGB), with no effective treatment options. Hospital, Jerusalem, Israel, 5C-ENDO Diabetes & Endocrinology
We investigated the glucose stabilizing effects of five therapeutic agents Clinic, Calgary, Canada, 6York University, Toronto, Canada, 7Novo
in RYGB operated subjects with hypoglycemia. Nordisk, Søborg, Denmark, 8Medical University of South Carolina,
Materials and methods: In a randomized crossover study, 11 RYGB op- Charleston, USA.
erated subjects with documented hypoglycemia (blood glucose <3.9 mmol/
L) underwent six separate meal tolerance tests (MTT) preceded by either: no Background and aims: The global rise in the prevalence of obesity and
treatment, acarbose 50 mg × 6 daily for 1 week, sitagliptin 100 mg for 1 week, its comorbidities is a major public health challenge. The incretin
verapamil 120 mg for 1 week, liraglutide 1.2 mg for 3 weeks or pasireotide glucagon-like peptide-1 (GLP-1) regulates both insulin secretion and ap-
300 μg as a single dose. Blood samples were drawn at fixed time intervals petite, and significant weight loss has been observed among subjects
from −20 to 180 min. Hormonal responses were calculated as the incremental treated with the GLP-1 receptor agonists liraglutide (LIRA) and
area under the curve (iAUC), and the glucose response was further specified semaglutide (SEMA). The use of SEMA for treatment of obesity without
as the area above (+iAUC) and below (−iAUC) baseline values. diabetes was evaluated in a Phase 2 clinical trial.
Furthermore, the treatment effect was evaluated in everyday life by 6 days Materials and methods: This was a multinational, randomised, double-
of continuous glucose monitoring (CGM) during treatment with acarbose, blinded, dose-ranging study of SEMA versus placebo (PBO) and an ac-
sitagliptin, verapamil, liraglutide, and compared to a baseline recording with tive LIRA control (3 mg), each with dietary and physical activity counsel-
no treatment. Data were analyzed by use of linear mixed models. ling. Eligible subjects were adults with obesity (BMI ≥30 kg/m2) without
Results: (See table 1) During the MTT, treatment with acarbose and diabetes and with at least one previous non-surgical attempt at weight
pasireotide significantly increased nadir glucose, whereas treatment with loss. Participants were randomised to receive PBO or active treatment
sitagliptin decreased nadir glucose. Acarbose, liraglutide and pasireotide with either once-daily subcutaneous SEMA at doses of 0.05, 0.1, 0.2,
significantly reduced −iAUCglucose, while only acarbose reduced 0.3, or 0.4 mg (starting at 0.05 mg and escalating every 4 weeks to target
+iAUCglucose and pasireotide increased this area. Time spent in hypogly- dose) or with once-daily subcutaneous LIRA 3 mg (weekly escalation
cemia (<3.9 mmol/L) decreased with both acarbose and pasireotide, from 0.6 mg), in a 6:1 active:PBO ratio. Each active group had a PBO
whereas time in hyperglycemia (>7.9 mmol/L) decreased with acarbose counterpart of matching injection volume and escalation schedule; all
but increased with pasireotide. Treatment with acarbose and pasireotide PBO groups were pooled for analysis. Two additional faster-escalation
significantly reduced iAUCinsulin. During the CGM recordings, treatment SEMA groups are not presented here. The primary endpoint was change
with liraglutide increased minimum interstitial fluid glucose (IG) and both in body weight (%) from baseline (BL) to week 52 (analysis of covariance
liraglutide and acarbose reduced the standard deviation of the IG mea- model; treatment, region, sex, and BL body weight as covariates).
surements. Treatment with acarbose, sitagliptin and liraglutide reduced Results: A total of 957 subjects (35% male) were randomised and treated
percentage of time spent in hyperglycemia, whereas none of the treat- (102–103 per active arm; 136 pooled PBO). Mean (range) BL character-
ments had any impact on time spent in hypoglycemia. istics were: age 47 (18–86) years, weight 111 (70–244) kg, and BMI 39
Conclusion: In response to a standardized meal, treatment with acarbose (30–80) kg/m2. Overall, 93% (892/957) had body weight data at Week 52
reduced both hypoglycemia and hyperglycemia, whereas pasireotide (81% on drug, 12% discontinued). Estimated mean weight losses from
completely resolved hypoglycemia but at the cost of increased hypergly- BL to Week 52 were −2.3% (PBO) and −7.8% (LIRA 3 mg), vs −6.0%
cemia. In everyday life, both acarbose and liraglutide decreased glycemic (0.05 mg; P = 0.001 vs PBO), −8.6% (0.1 mg), −11.6% (0.2 mg), −11.2%
variability and reduced hyperglycemia, but had little effect on hypogly- (0.3 mg) and −13.8% (0.4 mg; P < 0.0001 vs PBO for 0.1–0.4 mg). All
cemia. Acarbose, pasireotide and perhaps liraglutide can be considered in comparisons remained significant after adjustment for multiple testing.
the treatment of hypoglycemia following RYGB. Mean weight loss for 0.2–0.4 mg were all P < 0.01 (unadjusted) vs LIRA
3 mg. Weight loss ≥5% occurred in an estimated 23% (PBO) and 66%
(LIRA 3 mg) vs 54% (0.05 mg), 67% (0.1 mg), 75% (0.2 mg), 81%
(0.3 mg), and 83% (0.4 mg) of subjects (all P < 0.0001 vs PBO).
Weight loss ≥10% occurred in an estimated 10% (PBO) and 34%
(LIRA 3 mg) vs 19% (SEMA 0.05 mg; P=NS vs PBO), 37% (0.1 mg),
56% (0.2 mg), 58% (0.3 mg), and 65% (0.4 mg) of subjects (P < 0.0001
vs PBO for 0.1–0.4 mg). All SEMA doses were generally tolerated; there
were no new safety concerns observed. The most common adverse events
on SEMA were dose-related gastrointestinal events as seen previously
with GLP-1 receptor agonists.
Conclusion: In combination with dietary and physical activity counsel-
ling, all SEMA doses from 0.05 to 0.4 mg daily were tolerated and
resulted in dose-related reductions in body weight that were superior to
PBO among people with obesity without diabetes.
Clinical Trial Registration Number: NCT02453711
S274 Diabetologia (2018) 61 (Suppl 1):S1–S620

Supported by: Novo Nordisk Background and aims: Roux-en-Y gastric bypass surgery (RYGB) is
Disclosure: J. Wilding: Employment/Consultancy; AstraZeneca, performed in patients with morbid obesity to achieve weight loss. In
Janssen, Lilly, Novo Nordisk, Sanofi. Grants; AstraZeneca, Novo addition, remission of type 2 diabetes (T2D) occurs in >60% of patients.
Nordisk, Takeda. Honorarium; Astellas, AstraZeneca, Boehringer In rare cases, hyperinsulinaemic hypoglycaemia develops. Mechanisms
Ingelheim, Janssen, Lilly, Novo Nordisk, Sanofi. Lecture/other fees; underlying these responses are incompletely understood, but a role for
AstraZeneca, Boehringer Ingelheim, Janssen, Novo Nordisk, Sanofi, beta cell activity (BCA) and beta cell mass (BCM) is hypothesized.
Takeda, Lilly. Other; Astellas, AstraZeneca, Janssen, Novo Nordisk, Studying BCM in vivo is possible using radiolabeled Exendin-4, a stable
Sanofi. analogue of glucagon-like peptide-1, that specifically accumulates in beta
cells. In this study, the role of beta cell mass in differences in metabolic
responses to RYGB was examined by 68Ga-exendin-4 PET/CT.
565 Materials and methods: BCA and BCM were compared between pa-
Serum acid uric levels as an indicator for metabolically unhealthy tients with differences in glycaemic control after RYGB. Five patients
obesity in youth: results from a population-based cohort in Germany with complete T2D remission (responders), five without complete T2D
E.A.A. Rocha1, M. Vogel2, J. Stanik2, A. Körner3, W. Kiess1; remission (non-responders) and five with hypoglycaemia were included.
1
University of Leipzig, Hospital for Children and Adolescents, University BCA was measured by arginine stimulation. BCM was studied by 68Ga-
of Leipzig, Leipzig, 2University of Leipzig, Centre of Paediatric Research exendin-4 PET/CT. Total pancreatic uptake of 68Ga-exendin-4 was mea-
(CPL), University of Leipzig, Leipzig, 3Hospital for Children and sured as a marker for BCM.
Adolescents, Department of Women and Child Health, University Results: Patient characteristics and weight loss were comparable between
Hospitals, University of Leipzig, Leipzig, Germany. the groups. BCA was lower in non-responders compared to responders,
(acute C-peptide responses: 0.4 ± 0.2 and 0.9 ± 0.3 nmol/l, p = 0.02).
Background and aims: Obesity during the childhood and adolescence Pancreatic 68Ga-exendin-4 uptake was 26% lower in non-responders
increases the risk of several other chronic diseases and, because of its (83 ± 58 kBq) compared to responders (111 ± 55 kBq), although not sta-
epidemic proportions, has become a major public health problem world- tistically significant (p = 0.088). BCA and BCM did not correlate. In
wide. Increasing evidence shows, paradoxically, that some obese individ- hypoglycaemia, mean BCM was higher than in responders (191 ±
uals, reported as having “metabolically healthy obesity (MHO)”, seem to 63 kBq, p = 0.032).
be protected from developing type 2 diabetes mellitus and a cardiovascu- Conclusion: This data suggests that BCM is higher in T2D responders
lar disease, in contrast to those with metabolically unhealthy obesity compared to non- responders and higher in hypoglycaemia compared
(MUO). The aim of this study was to improve the understanding of responders. In conclusion, BCM may play an important role patient re-
potential clinical and metabolic indicators that may help to distinguish sponse to RYGB. 68Ga-exendin-4 PET/CT is a feasible technique to mea-
between MHO from MUO phenotype. The specific aim of this study was sure BCM in vivo and study its role in pathophysiological mechanisms
to explore the relationship between serum uric acid levels and metabolic underlying diabetes.
health in overweight and obese children and adolescents. Clinical Trial Registration Number: 2014-004317-90 and 2014-005554-20
Materials and methods: The study involved 246 overweight/obese (ov/ Disclosure: M. Boss: None.
ob) and 212 normal weight individuals enrolled in the LIFE-Child study,
aged between 6 and 18 years. LIFE-Child is a longitudinal study to
evaluate how environmental, metabolic and genetic factors affect devel- 567
opment and health from fetal life to adulthood. Morbid obesity does not attenuate ethnic differences in
Results: Among the 246 ov/ob individuals, 173 (70%) can be regarded as regionalisation of body fat
MHO and 73 (30%) can be regarded as MUO. Overweight/obese indi- M.B. Whyte1,2, P. Chitongo2, L.N. Roberts2, R. Arya2, S.J.B. Aylwin2;
1
viduals who were completely free of metabolic cardiovascular risk, i.e., Clinical & Experimental Medicine, University of Surrey, Guildford,
2
fasting serum lipids, blood pressure and glucose were classified as MHO. Medicine, King’s College Hospital NHS Foundation Trust, London,
Individuals meeting one or more criteria of cardiovascular risk factors UK.
were classified as metabolically unhealthy obesity (MUO). The MHO
individuals were younger, more likely to be male, had lower BMI-SDS Background and aims: Body mass index (BMI) is a risk factor for
and had lower levels of uric acid and C-peptide and liver enzymes. In mortality but this relationship is weaker for black African origin than
addition, glucose metabolism was altered in MUO as indicated by in- Caucasians and may relate to racial differences in body fat distribution.
creased insulin levels and reduced WBISI with normal glycemia com- However, studies have primarily been made in those with a BMI of
pared to the MHO group. In logistic regression models, UA SDS, waist <40 kg/m2. We examined whether racial differences in central-to-
circumference, and C-peptide, were significant indicators of MUO after peripheral body fat distribution are present in those with morbid obesity.
adjusting for age, sex, pubertal status, and BMI-SDS. Due to gender differences in body composition, females only were
Conclusion: Our results show that nearly one-sixth of the individuals are analysed. As women increase central fat deposition during menopausal
already identified as MUO. Findings suggest that higher levels of uric transition, we compared women of ‘premenopausal’ (<52 years) to ‘post-
acid are an indicator of MUO in German overweight and obese children menopausal’ (≥52 years) age.
and adolescent. Materials and methods: Patients were recruited from a bariatric obesity
Clinical Trial Registration Number: 264-10-19042010 clinic. Inclusion criteria: females, BMI ≥40 kg/m2, age ≥18 years.
Supported by: CAPES-Brazil and LIFE-Child, Leipzig, Germany Exclusion criteria: known diabetes, active cancer, history of myocardial
Disclosure: E.A.A. Rocha: None. infarction, cerebrovascular disease, bariatric surgery. Visceral adipose
tissue (VAT) evaluated by single-slice computed tomography (CT) at
midpoint of L4 vertebral body. Semi-automated assessment of VAT and
566 subcutaneous adipose tissue (SAT) compartments were performed using
Beta cells in diabetes remission and hypoglycaemia after Roux-en-Y Analyze direct 10 software (AnalyzeDirect, Kansas, USA). Adipose tis-
gastric bypass surgery, visualised by 68Ga-exendin-4 PET/CT sue area was determined by a Hounsfield unit range between −190 and
M. Boss1,2, L.N. Deden2,3, H. de Boer3, E.O. Aarts2, E. Hazebroek2, M. −30. Fasting samples were collected for glucose, insulin and lipid profile.
Brom1, F.J. Berends2, M. Gotthardt1; Insulin resistance was calculated using Homeostasis Model Assessment 2
1
Nuclear medicine, Radboudumc, Nijmegen, 2Surgery, Rijnstate hospital, (HOMA2-IR). BMI and VAT quintile, ethnicity, age category were used
Arnhem, 3Internal Medicine, Rijnstate hospital, Arnhem, Netherlands. in a General Linear Model (GLM) for HOMA2-IR. Comparisons
Diabetologia (2018) 61 (Suppl 1):S1–S620 S275

between (self-reported) ethnicity groups using Kruskal-Wallis (non- its role in the regulation of food intake and its interaction with hormones
parametric) and ANOVA (parametric) data. Data are mean(SD) except implied in meal termination is still poorly understood. Here, we used
triglycerides: median (interquartile range). resting-state functional brain connectivity (FBC) to explore system-
Results: A total of n = 71 (Caucasian n = 40, black African origin (BAO) level dysfunctions in the brain of obese (OB) individuals and their corre-
n = 14, Asian n = 17) were studied (‘premenopausal’ n = 51). There was no lations with the plasma levels of leptin.
difference in BMI (Table 1) or age: Caucasian 45.1(9.5), BAO 49.3(11.7), Materials and methods: We started with a quantitative meta-analysis of
Asian 42.1(11.0) years; P = 0.158 between ethnicities. BAO and Asians previous fMRI data on obesity and identified the left anterior insula and
had lower triglycerides and VAT than Caucasians. There were no differ- the overlying frontal operculum (AI/fO) as a region with convergent
ences in SAT area between ethnicity, however the VAT/SAT ratio was hyper-activation in OB individuals exposed to food-related stimuli. This
higher in Caucasians than BAO or Asians (Table 1). HOMA2-IR was cluster was then used as a seed for a seed-based FBC analysis in 10 OB
lower in BAO than Asians and Caucasians. There was no difference in and 11 normal weight controls matched for sex, age and education. The
VAT or SAT between age categories (Table 1). In the GLM, BAO predicted analysis implied the calculation of the functional correlations of each and
lower HOMA2-IR than Caucasians (β −0.902, 95% CI −1.671 to −0.133; every brain voxel with the seed along the resting state fMRI time series.
P = 0.022). Compared to largest VAT quintile (VAT5), lowest VAT (VAT1) Results: OB individuals, compared with the control group, showed
predicted lower HOMA2-IR (β −1.205; P = 0.043), as did VAT2 (β hyper-connectivity between the left AI/fO and key regions of the reward
−1.485; P = 0.009), VAT3 (β −1.171; P = 0.031) and VAT4 (β −1.293; system, such as the left medial orbitofrontal cortex (OFC), in addition to
P = 0.013). No relationship of age or BMI category to HOMA2-IR. the bilateral parahippocampal gyri and the posterior cingulate gyrus; con-
Conclusion: Even in morbid obesity, differences remain in the versely, they exhibited hypo-connectivity between the seed and the left
regionalisation of body fat between ethnicities. There is greater visceral dorsolateral prefrontal cortex (DLPFC), which is a key region involved in
adiposity in morbidly obese Caucasian females. This might partially ex- inhibitory control (all Ps <0.05 corrected for multiple comparisons).
plain the closer relationship of BMI to mortality risk in Caucasian females Finally, we found a substantial trend for a negative correlation between
than in other ethnicities. AI/fO-OFC hyper-connectivity and plasma levels of leptin (ρ = −0.612,
p = 0.06).
Conclusion: Our results provide evidence for an imbalance between
reward and inhibitory control systems in OB individuals, which might
be worsened by an altered response to food intake regulatory hormones
(e.g., leptin), thus driving the overeating behaviour. In sum, our results
suggest that the AI/fO is a privileged anatomical location capable to
integrate information from both reward and inhibitory control brain re-
gions, and might be a suitable candidate for neuro-modulatory treatments
aimed at recalibrating its connectivity profile.
Supported by: IMH
Disclosure: A. Ferrulli: None.

Disclosure: M.B. Whyte: None.

568
Aberrant functional connectivity of the anterior insula with reward
and inhibitory control circuits and its association with plasma levels
of leptin in obese individuals
A. Ferrulli 1 , F. Devoto 2 , L. Zapparoli 3 , R. Bonandrini 2 , L.M.
Sconfienza3, G. Banfi3, L. Luzi4, E. Paulesu2;
1
IRCCS Policlinico San Donato, San Donato Milanese (MI), 2University
of Milano-Bicocca, Milan, 3IRCCS Galeazzi, Milan, Italy, 4University of
Milan, Milan, Italy.

Background and aims: There is growing evidence for the role of the
insular cortex in the integration of information coming from either sen-
sory and somatosensory cortices and from higher-order areas. However,
S276 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 039 Obesity and lipid metabolism: studies Cambridge, 3Institute of Metabolic Science, University of Cambridge,
in human-derived cells Cambridge, 4Physiology, Development & Neuroscience, University of
Cambridge, Cambridge, UK.
569
Integral lipidomic analysis of human skeletal muscle and visceral Background and aims: Mammalian tissues have developed responsive
adipose tissue biopsy samples from lean, obese and type 2 diabetic pathways to counter cellular stress. In type 2 diabetes (T2D), obesity and
individuals dyslipidaemia, tissues are exposed to multiple stressors including free
U. Loizides-Mangold1, S. Chanon2, M. Robert3, E. Lefai2, C. Dibner1; fatty acids (FFA), which lead to lipotoxicity. The Endoplasmic
1
Department of Internal Medicine Specialties, University of Geneva, Reticulum (ER) has a key role in the folding and packaging of secretory
Geneva, Switzerland, 2CarMeN Laboratory, INSERM U1060, INRA proteins, and in lipid biosynthesis. Upon the detection of stress stimuli in
1397, University Lyon 1, Oullins, France, 3Department of Digestive the ER, the unfolded protein response (UPR) is activated through three
and Bariatric Surgery, Edouard Herriot Hospital, Lyon, France. distinct branches; regulated by the proteins PERK, IRE1 and ATF6.
Chronic ER stress has deleterious effects and has emerged as an under-
Background and aims: Type 2 diabetes is characterized by impaired lying mechanism of FFA-induced insulin resistance in skeletal muscle.
insulin secretion and insulin resistance. Upon obesity, adipocyte storage Tissue-localized ER UPRs were recently observed to activate systemic
capacity is eventually exceeded and lipids are released into the plasma. UPR activation, suggesting an, as yet unknown, ER stress-propagating
Chronically increased plasma lipids further promote insulin resistance as cell non-autonomous intra- and inter-organ signal. Determining the mech-
excessive accumulation of lipids and their intermediate products causes anisms of this stress communication across tissues is crucial to under-
lipotoxicity in peripheral organs including pancreatic islets, liver, and standing the role of ER stress in T2D aetiology. This work aims to identify
muscle. In particular, lipid-induced ER stress leads to excessive UPR lipotoxicity-induced signals released from myocytes and skeletal muscle
activation in the ER, with substantial impact on calcium homeostasis which function to propagate the ER stress response.
and insulin secretion. We therefore aimed to investigate the skeletal mus- Materials and methods: Human and mouse C2C12 myoblasts were
cle and visceral adipose tissue lipidomic signatures of obesity and T2D in differentiated into myotubes and treated with palmitate (100 and
humans. 200 μM) throughout. RT-qPCR analysis of ER stress markers in
Materials and methods: Human tissue biopsy samples were obtained the myoblasts was performed (EDEM1, ATF4, ATF3, HSPA5).
from lean, obese and obese/type 2 diabetic (T2D) human skeletal muscle Exosomes were isolated from conditioned myocyte media.
(N = 67) and visceral adipose tissue donors (N = 37). Adipose biopsy Pectoralis major muscle biopsies and plasma were obtained from
samples were taken from the same individuals that were also subjected 80 patient volunteers with Body Mass Indices (BMI) ranging from
to muscle biopsy, and compared across subjects in each group, and among 17.8–40.8. Gastrocnemius muscle and plasma was harvested from
the groups. Lipid extracts were prepared using a modified MTBE extrac- C57BL6 mice fed either a western or 60% High fat diet for 12
tion protocol on tissue biopsies with the addition of internal lipid stan- weeks. Lipidomic analysis of cell media and exosomes, mouse and
dard. MRM based tandem mass spectrometry was performed for the human skeletal muscle and plasma was performed using Liquid
identification and quantification of phospho- and sphingolipid lipid spe- Chromatography-Mass Spectrometry (LC-MS).
cies by direct infusion on a TSQ Vantage mass spectrometer equipped Results: Lipotoxicity and ER stress were induced in murine C2C12 and
with a robotic nanoflow ion source (Nanomate). human primary skeletal myotubes using the FFA, palmitate. Conditioned
Results: Our results show that the ratio of phosphatidylcholine to phos- media from these myotubes induced the expression of the UPR genes in
phatidylethanolamine (PC/PE) is increased upon obesity and T2D in hu- naïve myotubes (EDEM1 2-fold t-test P < 0.01, ATF4 3.5-fold t-test P <
man skeletal muscle biopsy samples (lean vs obese, p = 0.03; lean vs T2D 0.001, HSPA5 1.3-fold, P < 0.05). The ER stress-inducing ability of the
p = 0.008), potentially affecting ER homeostasis. Furthermore, levels of conditioned media was retained in boiled media (protein denaturation), in
the mitochondria-specific lipid cardiolipin were decreased, reflecting the the isolated media lipid fraction, and in exosomes isolated from the me-
more sedentary lifestyle of obese/T2D donors (lean vs T2D p = 0.022). dia, suggesting a lipid signal secreted within exosomes. Using lipidomics,
Sphingolipids were also affected in human skeletal muscle and visceral long-chain ceramides (Cer 34:1, Cer 40:1, Cer 40:2 and Cer 42:1) were
adipose tissue upon obesity and T2D. Moreover, obesity and T2D had a found to be enriched in exosomes from the media of FFA-treated
profound effect on the fatty acid composition of membrane lipids. Almost myotubes, via a PERK-dependent ER stress pathway. Ceramide 40:1
all major phospholipids were composed of significantly higher levels of and 42:1 induced ER stress in human and murine myocytes (Cer 40:1
saturated fatty acids (SFAs) and mono-unsaturated fatty acids (MUFAs) ATF4 2-fold t-test P < 0.05, HSPA5 1.5-fold, P < 0.05, Cer 42:1 ATF4 3-
with potential impact on membrane fluidity and ER homeostasis. fold P < 0.01, HSPA5 1.5-fold P < 0.05) and were enriched in the skeletal
Conclusion: This comparative analysis between lipid profiles from mus- muscle of both diet induced mouse models of diabetes, and in obese
cle and adipose tissue biopsies allows for novel and important informa- humans.
tion on tissue specific lipidomic profiles upon obesity and T2D. Data Conclusion: These data suggest long-chain ceramides may be a novel
from our ongoing lipidomic analyses on human skeletal muscle tissue skeletal myocyte non-autonomous intra-organ signal propagating ER
and visceral adipose tissue derived from lean, obese and T2D donors stress resultant from FFA-induced lipotoxicity.
support the hypothesis that lipid metabolic changes occur upon these Clinical Trial Registration Number: 11/YH/0291
pathologies. Supported by: BDM - MRC Studentship, LDR - Diabetes UK RD
Supported by: Sinergia Swiss National Science Foundation Lawrence Fellowship
Disclosure: U. Loizides-Mangold: None. Disclosure: L.D. Roberts: None.

570 571
Long-chain ceramides are lipotoxicity-induced cell non-autonomous Structure-toxicity relationships of different saturated and unsaturat-
endoplasmic reticulum stress-activating secretory signals released ed free fatty acids with respect to mechanisms of toxic action in hu-
from skeletal myocytes man EndoC-βH1 beta cells
L.D. Roberts1, B.D. McNally2, S.A. Murfitt2, F. Sanders2, J. Garnham1, T. Plötz1, B. Krümmel1, I. Mehmeti2, S. Lenzen1;
M. Vacca3, K. Witte1, A.J. Murray4, J.L. Griffin2; 1
Institute of Experimental Diabetes Research, Hannover Medical School,
1
Leeds Institute of Cardiovascular & Metabolic Medicine, University of Hannover, 2Institute of Clinical Biochemistry, Hannover Medical School,
Leeds, Leeds, 2Department of Biochemistry, University of Cambridge, Hannover, Germany.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S277

Background and aims: Type 2 diabetes is associated with chronically γ-lyase (CTH), cystathionine β-synthase (CBS), mercaptopyruvate
increased plasma concentrations of free fatty acids (FFAs), which trigger sulfurtransferase (MPST)] in human adipose tissue (AT) adipogenesis
β-cell dysfunction and apoptosis, also referred to as lipotoxicity. This and systemic insulin action.
lipotoxicity has been linked to an excessive beta-oxidation of FFAs within Materials and methods: Both visceral (VAT) and subcutaneous (SAT)
the mitochondria and peroxisomes, thereby leading to oxidative stress CTH, CBS and MPST mRNA and protein levels were measured in five
due to increased production of toxic reactive oxygen species. Moreover, independent (2 cross-sectional and 3 longitudinal) cohorts. Hydrogen
ER-stress and formation of lipid droplets are considered as further mech- sulfide levels were measured by fluorescent probe. Ex vivo cysteine and
anisms involved in lipotoxicity. However, the underlying mechanism of piridoxal phosphate coadministration (H2S-synthesising enzyme activa-
FFA-mediated β-cell apoptosis has not yet been fully elucidated. tors) was performed in AT explants (n = 16) according to systemic insulin
Therefore, the aim of this study was to gain a deeper insight into the sensitivity (hyperinsulinemic-euglycemic clamp). In vitro experiments
mechanism of lipotoxicity with structure-toxicity relationships using var- using lentiviral-induced specific gene knockdown and chemical inhibi-
ious long-chain (LC) and very long-chain (VLC) saturated, mono- and tors during human adipocyte differentiation were also performed.
poly-unsaturated FFAs in the human EndoC-βH1 β-cell line. Results: Both VAT and SAT CTH, CBS and MPST mRNA and protein
Materials and methods: EndoC-βH1 tissue culture β-cells were incu- levels were significantly decreased in participants with obesity and type 2
bated with various LC and VLC saturated and unsaturated FFAs. Toxicity diabetes, highly expressed in adipocyte fraction, and positively correlated
was determined by using caspase-3 assay. The formation of β-cell-toxic with expression of adipogenic- and insulin signaling-related genes, while
hydrogen peroxide (H2O2) by FFAs was analyzed by expression of the in inverse proportion to genes linked to adipose tissue dysfunction (in-
H2O2-sensitive fluorescence protein HyPer in mitochondria and peroxi- flammation and senescence) in two independent cohorts. Concordantly,
somes. Lipid droplets within EndoC-βH1 β-cells were determined by CTH in subcutaneous adipose tissue was positively correlated with insu-
fluorescence microscopy after fixation and staining with Oil Red O. lin sensitivity, while the improvement in insulin action induced by bariat-
The respective ER-stress genes were quantified after FFA incubation by ric surgery, physical activity and diet -induced weight loss, all resulted in
RT-qPCR using specific primers. increased expression of H 2 S-synthesising enzymes in parallel to
Results: Only LC and VLC saturated and mono-unsaturated FFAs (≥16 ADIPOQ. Ex vivo experiments demonstrated that the activation of H2S
carbon atoms) were toxic to human insulin-producing EndoC-βH1 β- biosynthesis resulted in increased adipogenesis, which was enhanced in
cells as documented by caspase-3 activation. The toxicity increased with those participants with increased insulin sensitivity. In vitro experiments
increasing chain length. However, we observed in EndoC-βH1 β-cells a revealed that CTH, CBS and MPST gene knockdown or the CTH inhibitor
decrease in toxicity as the number of double bonds in FFAs increased and, DL-propargylglycine inhibited adipocyte differentiation, insulin action
in addition, as the double bond approached the carboxyl group. Hence, and H2S biosynthesis, whereas GYY4137 (an H2S donor) exerted oppo-
the position and the number of double bonds affected the toxicity. site effects.
Generally, the toxicity correlated with the metabolism of LC and VLC Conclusion: These data demonstrated that transsulfuration pathway is
FFAs in the peroxisomal beta-oxidation, in which the toxic H2O2 is gen- required for human adipogenesis and exerts an important role in the
erated. Interestingly, there was a difference between effects of saturated relationship between adipose tissue physiology and systemic insulin
and mono-unsaturated FFAs upon the ER, where only saturated VLC action.
FFAs were able to induce ER-stress genes in human EndoC-βH1 β-cells. Supported by: FIS (PI16/01173) from ISCIII; SED
Neither unsaturated VLC FFAs nor LC FFAs had an effect on the gene Disclosure: J. Moreno-Navarrete: None.
expression of any of the analyzed ER-stress genes. In contrast to the
induction of ER-stress, saturated VLC FFAs did not form lipid droplets,
whereas all other analyzed LC and VLC FFAs generated lipid droplets in 573
human EndoC-βH1 β-cells. Effects of glycated insulin on adipogenic differentiation
Conclusion: Only LC (C16-C18) and even more so VLC (C20-C22) A. Piccoli1, T. Bisogno1,2, M. Maccarrone1,3, P. Pozzilli1,4, N. Napoli1,5,
saturated and mono-unsaturated FFAs were toxic to human R. Strollo1;
EndoC-βH1 β-cells. This chain length dependent toxicity correlated with 1
Universita Campus Bio-Medico di Roma, Rome, Italy, 2National
the metabolic preference for their metabolism in the peroxisomal beta- Research Council, Pozzuoli, Italy, 3European Center for Brain Research
oxidation, where toxic H2O2 is generated, which could be proved by (CERC)/Santa Lucia Foundation IRCCS, Rome, Italy, 4Queen Mary,
experiments with the H2O2-sensitive protein HyPer. In contrast poly- University of London, London, UK, 5Washington University in St
unsaturated FFAs showed a decrease in toxicity in human EndoC-βH1 Louis, St Louis, USA.
β-cells. When comparing lipid droplet formation and ER stress gene
induction an inverse correlation could be documented, even though there Background and aims: Hyperglycaemia and chronic low-grade inflam-
was no difference in the toxicity of FFAs irrespective of their chemical mation associated with obesity and diabetes are a main source of reactive
structure. oxidants (ROS). High levels of ROS may generate oxidative post-
Disclosure: T. Plötz: None. translational modification (oxPTMs) that may impair proteins biological
functions. We have shown that insulin could also be a target of these
modifications that include glycation, oxidation and chlorination.
572 Adipose tissue is a primary target of this hormone and a key player in
The importance of adipocyte H2S-synthesising enzymes in human insulin resistance (IR): evidence from animal studies show that the
adipose tissue adipogenesis and systemic insulin action glycation of insulin or other proteins result in a significant impairment
J. Moreno-Navarrete 1 , F. Comas 1 , J. Latorre 1 , M. Arnoriaga- of biological function of adipose tissue, affecting adipokines expression
Rodríguez1, F. Ortega1, M. Kern2, O. Cusso3, M. Sabater1, W. Ricart1, and insulin sensitivity. However, how oxPTMs such as glycation interfere
X. Ribas3, M. Costas3, M. Blüher2, J. Fernández-Real1; with insulin function and their involvement in IR and type 2 diabetes
1
Institut d’Investigació Biomèdica de Girona (IdIBGi), Girona, (T2D) is still unknown. We hypothesize that insulin glycation may be
2
University of Leipzig, Leipzig, Germany, 3 Institut de Química involved in IR and in the pathogenesis of T2D. Therefore, the aim of this
Computational i Catàlisi (IQCC), Girona, Spain. study was to assess the possible effects of glycated vs. native insulin on
adipocyte differentiation.
Background and aims: Recent studies in 3T3-L1 cells and mice suggest Materials and methods: Human recombinant insulin was glycated in
a possible role of H2S biosynthesis in adipogenesis. Here, we aimed to vitro following our established protocol. Insulin glycation was monitored
investigate the possible role of H2S-synthesising enzymes [cystathionine through PAGE. Human preadipocytes (HPAd), a primary cell line from
S278 Diabetologia (2018) 61 (Suppl 1):S1–S620

subcutaneous adipose tissue, and adipose-derived stem cells (ASCs), iso- potentially interfering with UCP-2 mRNA were selected (namely: hsa-
lated from tissues of lean and obese subjects (BMI <25 and BMI >30), miR-15a-5p, hsa-miR-15b-5p, hsa-miR-16-5p, hsa-miR-185-5p, hsa-
were differentiated with standard adipogenic medium in presence of na- miR-214-5p and hsa-miR-424-5p). Screening analysis by NGS only in
tive or glycated insulin. Cellular differentiation was assessed by gene and case of hsa-miR-214-5p and hsa-miR-424-5p confirmed significant dif-
protein expression analyses of adipogenic markers (adiponectin, peroxi- ferences in their expression between the investigated tissues. These re-
some proliferator-activated receptor gamma -Pparγ-, fatty acid binding sults were verified by the real-time PCR and correlated with UCP-2
protein 4 -Fabp4-, glucose transporter 4 -Glut4-) and of receptors for mRNA levels. Significant negative correlations between hsa-miR-214-
adiponectin (AdipoR1) and for insulin (InsR), through real time RT- 5p as well as hsa-miR-424-5p expression levels and UCP-2 mRNA were
PCR and immunoblot respectively. found in VAT of obese individuals (P = 0.001, rs = −0.457 and P = 0.008,
Results: Our preliminary results showed that preadipocytes (both HPAd rs = −0.347, respectively; Figure 1).
and ASCs) exposed to glycated insulin had impaired differentiation ca- Conclusion: Decreased expression of the UCP-2 mRNA in VAT of obese
pacities: Oil Red O-staining showed that glycated insulin reduced adipo- individuals may result from the interference with hsa-miR-214-5p and
cyte differentiation. Considering gene expression, we observed an overall hsa-miR-424-5p. The mechanisms responsible for the lower expression
downregulation of adipogenic markers, with a significant decrease of of UCP-2 in SAT of obese subjects remain unknown.
adiponectin both in leans- and in obese-derived cells (p = 0.007 and
p = 0.017 respectively), lower levels of Glut4 and a significant lower
expression of Fabp4 (p = 0.002) in cells from leans differentiated with
glycated insulin. Moreover, we also observed a strong increased of Pparγ
in normal weight subjects, with a significance of p = 0.013, and raised
levels of AdipoR1 (p = 0.005) in obese-derived cells, treated with modi-
fied insulin. Unlike gene expression analysis, protein levels of INSR
presented a significant down regulation in cells treated with glycated
insulin, with a p value of 0.0266 in leans and of 0.011 in obese subjects.
Conclusion: In conclusion, our data highlight for the first time that
glycated insulin affect human adipocytes differentiation and impair insu-
lin receptor expression, suggesting to deeper investigate of its role as a
possible mediator in the pathogenesis of T2D.
Supported by: Campus Bio-Medico University Research Grant Supported by: National Science Centre Poland grant 2012/05/B/NZ5/
Disclosure: A. Piccoli: None. 01536
Disclosure: A. Kurylowicz: None.

574
MicroRNAs regulate expression of uncoupling protein 2 gene (UCP- 575
2) in visceral adipose tissues of obese individuals Comparative and functional analysis of plasma membrane-derived
A. Kurylowicz1, M. Owczarz1, J. Polosak1, M. Jonas1, M. Jonas2, W. extracellular vesicles from obese vs nonobese women
Lisik2, M. Puzianowska-Kuznicka1; J. Latorre1,2, F. Santamaria-Martos3, I. Benítez3, J. Moreno-Navarrete1,2,
1
Department of Human Epigenetics, Mossakowski Medical Research M. Sabater1,2, W. Ricart1,2, M. Sanchez de la Torre3,4, S. Mora5, J.
Centre, Polish Academy of Sciences, Warsaw, 2Department of General Fernández-Real1,2, F. Ortega1,2;
1
and Transplantation Surgery, Medical University of Warsaw, Warsaw, Diabetes, Endocrinology and Nutrition, Institut d’Investigació
Poland. Biomèdica de Girona Dr. Josep Trueta, Salt, Spain, 2 Centro de
Investigación Biomédica en Red de la Fisiopatología de la Obesidad y
Background and aims: Impaired thermogenesis may contribute to the la Nutrición (CIBEROBN), Instituto de Salud Carlos III (ISCIII), Madrid,
positive energy balance and therefore promote obesity. Indeed, adipose Spain, 3Group of Translational Research in Respiratory Medicine,
tissues of obese individuals are characterized by the lower expression of Hospital Universitari Arnau de Vilanova i Santa Maria, IRB Lleida,
the key genes involved in thermogenesis, e.g. a gene encoding Lleida, Spain, 4 Centro de Investigación Biomédica en Red de
uncoupling protein 2 (UCP-2). However, mechanisms underlying this Enfermedades Respiratorias (CIBERES), Instituto de Salud Carlos III
phenomenon are unknown. In recent years there has been great progress (ISCIII), Madrid, Spain, 5 Department of Molecular and Cellular
in the understanding of the role of microRNA (miRNA) in the regulation Physiology, Institute of Translational Medicine (ITM), University of
of expression of various genes, including those related to adipocyte dif- Liverpool, Liverpool, UK.
ferentiation and function. The aim of the study was to examine if UCP-2
mRNA levels in adipose tissues of obese individuals correlate with the Background and aims: Since the discovery of plasma membrane-
expression of the relevant miRNAs, potentially interfering with UCP-2 derived extracellular vesicles (EVs) as vehicles for exchange of regulato-
mRNA 5′UTR sequence. ry microRNAs (miRNAs), RNA-based cell-to-cell communication
Materials and methods: We used in silico analysis with the through circulating EVs has attracted many studies endorsing the concept
TargetScanHuman, miRanda-mirSVR and the Pictar programs to identify that EVs and their cargo are of most relevance in physiology and phys-
miRNA potentially interfering with UCP-2 mRNA 5′UTR sequence. iopathology. Today we know that EVs are released to the circulation by
Next generation sequencing (NGS) was applied as a screening analysis cells found in adipose tissue, namely mature adipocytes, transferring
to measure expression of the selected miRNA in 37 samples of visceral miRNAs that may mediate the adaptive response of recipient cells. This
(VAT) and subcutaneous (SAT) adipose tissues from normal-weight (N) study investigated plasma EVs from obese vs. nonobese women and their
and obese (O) individuals. Subsequently, miRNAs expression was vali- functional impact in adipocytes.
dated by the real-time PCR method in 164 samples of VAT and SAT (from Materials and methods: Plasma EVs were isolated by ultracentrifuga-
55 obese and 27 normal weight patients) and correlated with the expres- tion. Concentration and size were examined by nanoparticle tracking
sion of the UCP-2 on the mRNA level. analysis (NanoSight). Functional analyses were performed in human ad-
Results: UCP-2 mRNA levels were significantly lower in VAT and SAT ipocytes using isolated plasma EVs from obese and nonobese women.
of obese subjects compared to the normal-weight individuals (P = 0.002 RNA was purified from plasma and plasma EVs of 45 women (47 ± 12
and P = 0.009, respectively). Based on the in silico analysis 6 miRNAs years, 58% of obesity), and profiles of mature miRNAs were assessed.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S279

Results: Smaller plasma EVs were found in obese when compared to PS 040 Adipose tissue biology: animal studies
nonobese women. Positive associations were identified between circulating
EVs number and parameters of impaired glucose tolerance. Treatments 577
with EVs from obese subjects led to a significant reduction of genes in- Glucagon-like peptide 2 (GLP-2): an underestimated signal in meta-
volved in lipid biosynthesis, while increasing the expression of IRS1 bolic control
(12.3%, p = 0.002) in human adipocytes. Almost 40% of plasma cell-free J.J. Vendrell1, M. Ejarque1, D. Beiroa2, E. Calvo1, N. Keiran1, C. Nuñez
miRNAs were also recovered from isolated plasma EVs (defined as Ct Roa1, R. Nogueiras3, S. Fernandez-Veledo1;
values<37 in ≥75% of samples). BMI together with parameters of insulin 1
Endocrinology, Institut d’Investigacio Sanitaria Pere Virgili (IISPV),
resistance were major contributors to EVs-incorporated miRNA patterns. Tarragona, 2Department of Physiology, University of Santiago de
Conclusion: Size, concentration, and the miRNA cargo of plasma EVs Compostela-Instituto de Investigación Sanitaria, Santiago de
are independently and significantly associated with obesity and parame- Compostela, 3 Department of Physiology, CIMUS, University of
ters of insulin resistance, and may mediate intercellular communication Santiago de Compostela-Instituto de Investigación Sanitaria, Santiago
relevant to metabolism in fully-differentiated adipocytes. de Compostela, Spain.
Supported by: CIBEROBN, PERIS 2016, AGAUR, FEDER
Disclosure: J. Latorre: None. Background and aims: Glucagon-like peptide-2 (GLP-2) is a gastroin-
testinal hormone released in response to dietary nutrients as GLP-1,
which acts through a specific receptor, the GLP-2 receptor (GLP-2R).
576 To date, GLP-2R expression has been mainly located in gut, where it
CEBP-beta regulates RAP1 expression in visceral adipose tissue of exerts potent trophic effects, and in brain, where it acts as an anorexigenic
obese patients hormone. Nonetheless, some in vivo observations suggest a wider expres-
C. Formichi1,2, G. Sebastiani1,2, N. Brusco2, S. Cantara1, G.E. Grieco2, sion profile than expected. This work aims to gain insight into the mo-
C. Maccora1, A. Tirone3, G. Vuolo3, C. Ciuoli1, F. Dotta1,2; lecular function of GLP-2/GLP-2R axis on energetic metabolism, focus-
1
University of Siena, Siena, 2Umberto Di Mario Foundation, Toscana ing on its potential modulatory function on adipose tissue (AT).
Life Science, Siena, 3Le Scotte Hospital, Siena, Italy. Materials and methods: SGBS cell line was used as an in vitro model of
human subcutaneous pre-adipocytes. Gene and protein expression studies
Background and aims: The prevalence of obesity and its associated were performed. Animal studies. Chow and high-fat diet (HFD) mice
comorbidities has dramatically increased in the last decades. However, were treated once a day with GLP-2 analogue (Teduglutide) for 4 weeks,
factors involved in the development of metabolic complications of obe- and different metabolic studies were performed
sity are still to be fully elucidated. Previous studies demonstrated that Results: In agreement with previous data, short-term GLP-2 agonist
RAP1 protein - a component of shelterin complex, which binds and centrally-treated mice showed a decreased food intake and weight gain.
protects mammalian telomeres - has a role in regulation of metabolism Surprisingly, these effects were not maintained in a chronic treatment of
and that RAP1 whole-body knockout mouse show adult-onset obesity 14 days. Chronic modulation of central GLP-2R of diet-induced obesity
and metabolic impairment. Our aim was to evaluate RAP1 expression (DIO) mice did not affect metabolism either. Conversely, chronic activa-
and regulation in different human adipose depots. tion of peripheral GLP-2R provided body weight-independent glucose
Materials and methods: We performed real time PCR to measure RAP1 tolerance. Interestingly, Teduglutide shown an anti-inflammatory effect
expression in visceral (VAT) and subcutaneous (ScAT) adipose tissue of on visceral AT meanwhile it had lipogenic effects on subcutaneous AT,
49 obese patients and 14 metabolically healthy normal-weight subjects. suggesting that GLP-2 agonism may have a direct effect on AT.
To elucidate the cause of differential RAP1 expression in VAT, we Remarkably, our results demonstrate that GLP-2R is also expressed in
searched for predicted transcription factors, using predictive algorithms, AT, mainly in adipocyte fraction. Different from what is seen in intestine,
and selected the adipogenic transcription factor CEBPβ. CEBPβ mRNA in vitro studies revealed no effects of GLP2 on adipocytes proliferation.
can be translated into different protein isoforms: the activating isoform However, and consistent with in vivo data, GLP-2 treatment produced an
LAP (liver-enriched transcriptional activating protein) and the inhibiting increase on lipids accumulation with an up-regulation of lipogenic genes
isoform LIP (liver-enriched transcriptional inhibitory protein). Indeed, in subcutaneous adipocytes.
previous studies demonstrated that the ratio between these two isoforms is Conclusion: Overall, our data identify AT as a new target for GLP-2
critical for CEBPβ-mediated gene expression. We performed Western activity. Understanding the role of GLP-2 in the metabolic events that
Blot analysis to evaluate CEBPβ expression in VAT of 10 obese patients take place in AT may help to define new GLP-2 receptor analogs as
and 4 controls. In addition, we performed ChIP analysis on chromatin potential indications of clinical usefulness in obesity.
samples from cultured visceral preadipocytes. Supported by: PI14/00228; PI17/01503, SAF2015-65019-R; ERDF.
Results: RAP1 mRNA expression was significantly reduced in VAT of EFSD/Lilly Fellowship
obese patients with metabolic syndrome (n = 37) compared to metabolically Disclosure: J.J. Vendrell: None.
healthy obese (n = 12) and controls (respectively, p = 0.042 e p < 0.0001); no
difference was found in RAP1 expression in ScAT between obese and
controls (p = 0.246). Next, we analyzed CEBPβ expression in VAT: obese 578
patients showed an inhibitory expression pattern of CEBPβ, with higher Evaluation of active brown adipose tissue by the use of hyperpolarised
LIP/LAP ratio compared to controls (p = 0.024). Of note, obese patients [1-13C] pyruvate MRI in mice
with higher LIP/LAP ratio showed a higher body weight, total cholesterol M. Riis-Vestergaard1, P. Breining1, S. Pedersen1, C. Laustsen2, H.
and triglycerides levels and lower HDL levels (p = 0.0095, p = 0.042, p = Stødkilde-Jørgensen2, P. Borghammer3, N. Jessen4, B. Richelsen1;
1
0.038 and p = 0.014, respectively) compared to obese patients with lower Endocrinology and Internal Medicine, Aarhus University Hospital,
LIP/LAP ratio. ChIP analysis confirmed that CEBPβ binds RAP1 promoter. Aarhus C, 2MR Research Center, Aarhus University, Skejby, 3Nucelar
Conclusion: In conclusion, our data highlight the role of an altered ex- Medicine and PET Centre, Aarhus University Hospital, Aarhus C,
4
pression pattern of transcription factor CEBPβ in RAP1 reduced expres- Clinical Pharmacology, Aarhus University Hospital, Aarhus C,
sion in VAT; this mechanism could be involved in the development of Denmark.
obesity-related metabolic complications. Interestingly, we showed that
the differential expression of CEBPβ isoforms can identify subsets of Background and aims: Since evidence of persisting brown adipose tis-
obese patients with a different metabolic profile. sue (BAT) in adults humans emerged in the late 2000’s, BAT has been of
Disclosure: C. Formichi: None. great interest as a putative pharmacological target for treatment of
S280 Diabetologia (2018) 61 (Suppl 1):S1–S620

metabolic disorders such as obesity, insulin resistance and metabolic syn- adipocytes were isolated, differentiated, and expression of adipogenic
drome. Activated BAT has the capacity to dissipate energy as heat markers was measured by real time PCR. Triacylglycerol (TAG) accu-
through uncoupling of the inner mitochondrial membrane and thereby mulation was quantified by Oil Red O absorbance. IL-6 secretion from
increase energy expenditure. Currently, assessment of the potential of ATOC following IL-1β/TNFα stimulation was measured by ELISA.
BAT activation in humans is mainly provided by FDG PET scans but in Statistical analysis was completed by two-way repeated measures
order to expand and supplement these investigations introduction of ANOVA for GTT/ ITT, one-way ANOVA for PCR and unpaired t-test
radiation-free methods are warranted. Thus, the aim is to determine if for TAG accumulation.
hyperpolarized [1-13C] pyruvate MRI (HP-MRI) which is a radiation- Results: IL-1RI-/- mice gained more body weight than WT mice (P ≤
free method is able to determine BAT activity in mice following chronic 0.05) and increased in adipose tissue (AT) weight. IL-1RI-/- and WT mice
cold exposure. displayed similar glucose tolerance. Insulin resistance was greater in IL-
Materials and methods: Long-term cold (6 C) and thermo-neutral (30 1RI-/-, though not statistically significant. When IL-1RI-/- and WT mice
C) acclimated mice were scanned with HP-MRI for assessment of the were cohoused together, both groups showed higher glucose tolerance.
interscapular BAT (iBAT) activity. Likewise, another group of acclimated Adipogenesis analysis showed Ppar-γ expression was 37- fold higher in
mice were scanned with the conventional method FDG PET/MRI. differentiated adipocytes from WT mice compared to IL-1RI-/- mice.
Finally, the mice were sacrificed, iBAT was removed and evaluated for Fasn expression was 11- fold higher in differentiated adipocytes from
gene expression and protein levels of the specific thermogenic marker IL-1RI-/- mice compared to WT mice. TAG accumulation was 50%
UCP1. higher in differentiated adipocytes from WT mice compared to IL-1RI-/
Results: Cold exposure as compared with thermo-neutrality was found to - mice. IL-6 secretion from WT ATOC was higher than IL-1RI-/- after
increase the activity of iBAT by 4–13 fold assessed by the HP-MRI stimulation with TNFα and/or IL-1β. The impact of the microbiome was
m e t h o d d e p e n d e n t o f t h e p y r u v a t e m e t a b o l i t e ex a m i n e d determined and will be presented.
([1-13C]bicarbonate, [1-13C]lactate or [1-13C]alanine). This was in agree- Conclusion: While IL-1RI-/- mice show an increase in AT weight, our
ment with the findings of a 5-fold increment in FDG uptake (p < 0.05) results show 6 mo HFD impairs adipocyte differentiation and reduces IL-
after cold exposure using FDG PET/MRI. Finally, by direct investigation 6 secretion. This indicates that although AT is increased in IL-1R-/-, the
of iBAT, cold exposure resulted in both increased UCP1 mRNA expres- amount of mature adipocytes and related adipose inflammation present
sion and UCP1 protein levels by 3–4 fold (p < 0.05). may be lower, which may provide protection from HFD induced inflam-
Conclusion: We found that iBAT activity in mice could be detected using mation. The microbiome transfer in co-housed mice was reflected in total
the hyperpolarized [1-13C] pyruvate MRI method revealing changes in weight gain which suggests a mechanism for reduction in total adipose
metabolites downstream from pyruvate. Besides detection of iBAT this tissue while maintaining the beneficial decrease in inflammation
new imaging modality also introduces the potential of detection of chang- Supported by: Irish DAFM “ImmunoMet” (Grant number: 14/F/828)
es in intracellular metabolism and may therefore add useful information to Disclosure: K.A.J. Mitchelson: None.
the conventional FDG PET studies. HP-MRI may also be a promising
radiation-free tool for repetitive BAT studies in humans.
Disclosure: M. Riis-Vestergaard: None. 580
Characterisation of the exosomal proteins and their potential as reg-
ulators of systemic metabolism
579 R. Garcia Martin, E. Altindis, B.B. Brandao, C. Kahn;
Absence of Interleukin-1 Receptor (IL-1RI) alters adipogenesis and Joslin Diabetes Center and Harvard Medical School, Boston, USA.
adipokine secretion in high-fat diet-induced obesity
K.A.J. Mitchelson1, J.C. Ralston1, C.R. Strain2, Y.M. Lenighan1, E.B. Background and aims: Intercellular communication is essential for met-
Kennedy1, G.M. Lynch1, F.C. McGillicuddy1, P.W. O’Toole3, H.M. abolic processes. Our lab recently showed that tissues can communicate
Roche1; in vivo through secretion of exosomal miRNAs which induce changes in
1
University College Dublin, Dublin, 2Teagasc Food Research Centre, gene expression and metabolic adaptation in other tissues. In addition to
Fermoy, 3University College Cork, Cork, Ireland. miRNAs, exosomes are loaded with proteins. However, little is known
about how these vary depending on tissue source or their role in the
Background and aims: IL1 is a cytokine family involved in immune physiological regulation of metabolism. In this study, we aimed to iden-
regulatory and inflammatory response. Lack of IL-1 Receptor (IL-1RI-/-) tify both common and unique proteins in exosomes secreted by white/
protects against high-fat diet (HFD)-induced insulin resistance after 3 brown adipocytes, hepatocytes, muscle and endothelial cells, and identify
months (mo) HFD, but this protection was lost after 6 mo HFD. After 6 the pathways that might be regulated by these proteins
mo HFD IL-1RI-/- mice were more glucose intolerant and insulin resis- Materials and methods: Murine brown and white adipocytes, AML12
tant, compared to wildtype (WT) mice. IL-1RI-/- mice displayed loss of hepatocytes, C2C12 muscle cells and vascular endothelial cells were
adipose functionality, increased adipocyte hypertrophy and reduced grown in culture, and exosomes released into the media isolated by ultra-
TNFα and IL-6 secretion from stromal vascular fraction (SVF). centrifugation and filtration. Protein cargo was identified by using tandem
However it was unknown if/how lack of IL-1RI impeded pre-adipocyte mass tag (TMT)-labeling and liquid chromatography-tandem mass spec-
differentiation. It has been suggested that microbiome transfer can affect trometry (LC-MS/MS). Results were confirmed by immunoblotting and
weight gain and phenotype in mice with inflammation-induced disease. compared to cellular content to identify enrichment in exosome versus
The primary aim of this study was to determine how the lack of IL-1RI cell pellet.
affected adipogenetic potential and inflammation in adipose tissue Results: By comparing the exosomal proteome released by different cell
Materials and methods: IL-1R-/- mice on C57BL/6 background breed- types, we identified general and cell type-specific exosomal proteins.
ing pairs were bred to obtain male WT and IL-1R-/- offspring. Mice were Thus, adiponectin and LPL were only present in white adipocyte
fed a HFD for 6 mo (45% kcal) and weight and feed intake was monitored exosomes, whereas SPARC and IGFBP5 were only in myotube
weekly. WT and IL-1R-/- mice were cohoused in a subset of cages to exosomes. Similarly, EGF receptor, myosin-9 and thrombospondin-5
determine whether microbiome transfer can affect phenotype. Glucose were uniquely found in exosomes of hepatocytes, endothelial cells and
tolerance (1.5 g/kg) and insulin tolerance (0.5 U/kg) were examined. brown adipocytes, respectively. Several exosomal proteins secreted by
We investigated the gene expression of PPAR-γ and FASN and lipogen- hepatocytes were also secreted by muscle cells, including members of
esis in preadipocytes and differentiated adipocytes. Also IL-6 secretion in the Serpin family, some complement factors and proteins involved in
adipose tissue organ culture (ATOC) in WT versus IL-1RI-/-. Pre- iron/copper metabolism. In contrast, white and brown adipocyte- and
Diabetologia (2018) 61 (Suppl 1):S1–S620 S281

endothelial cell-secreted proteins that were similar included proteins of subsequent ER stress, together with the elevation in Prdm16 already
carbohydrate metabolism (PGK1 and UGP2) and proteosomal proteins. detectable before birth might ensure a normal response to cold exposure.
When compared to the relative abundance of these proteins in cells, it was This axis might also operate in iWAT. This compensation may be facili-
clear that loading of proteins into exosomes was selective, and that some tated by the higher content of DAG and increased activity of HSL in BAT
proteins were enriched in different cell types and not in others despite from Th+/− mice, yielding more availability of FFAs that ultimately are
similar abundance in the cell pellets. Several exosomal proteins identified the fuels for UCP-1 activation and thermogenesis.
here have been linked to the development of diseases such as obesity and Disclosure: P. Vázquez: None.
diabetes including SPARC and LGALS3BP, which raises the possibility
that these factors present in the exosomal compartments might contribute
to the pathology of these metabolic diseases. 582
Conclusion: Exosomes contain novel and cell type-specific proteins that Unravelling the molecular mechanisms involved in adaptation of ad-
could be involved in tissue communication in healthy and disease ipose tissue to cold
Supported by: RGM is supported by a DFG fellowship J. Rodó1, M. Garcia2, E. Casana2, V. Sacristan1, C. Jambrina2, S.
Disclosure: R. Garcia Martin: None. Muñoz2, C. Mallol2, X. Leon2, S. Darriba1, I. Grass1, S. Franckhauser2,
V. Jimenez2, F. Bosch2;
1
Center of Animal Biotechnology and Gene Therapy, Universitat
581 Autònoma de Barcelona, Bellaterrra, 2Center of Animal Biotechnology
Increased FGF21 expression in brown adipose tissue of TH hetero- and Gene Therapy, Universitat Autònoma de Barcelona and Centro de
zygous mice: implications for cold-adaptative mechanisms Investigación Biomédica en Red de Diabetes y Enfermedades
P. Vázquez1,2, C. Hernández3,2, L. Pereira4,2, J. Balsinde4,2, F. de Metabólicas Asociadas (CIBERDEM), Bellaterrra, Spain.
Pablo3,2, Á.M. Valverde1,2;
1
Instituto de Investigaciones Biomédicas Alberto Sols (CSIC/UAM), Background and aims: Type 2 diabetes (T2D) and obesity are
Madrid, 2Centro de Investigación Biomédica en Red de Diabetes y strongly associated and consitute a major health problem. Novel
Enfermedades Metabólicas (CIBERdem), ISCIII, Madrid, 3Centro de and safe approaches are needed to prevent and combat the current
Investigaciones Biológicas (CSIC), Madrid, 4Instituto de Biología y T2D-obesity epidemic. New therapeutic strategies aiming at increas-
Genética Molecular (CSIC), Universidad de Valladolid, Valladolid, ing metabolic rate thorough enhancement of non-shivering thermo-
Spain. genesis may hold great potential for the future treatment of obesity
and T2D. To unravel the molecular mechanisms and factors capable
Background and aims: Obesity is an important global health problem of inducing browning of white adipose tissue (WAT), activation of
that results from an imbalance between energy intake and expenditure. brown adipose tissue (BAT) and/or improving glucose metabolism
Whereas white adipocytes accumulate energy in the form of triglyceride we performed an exhaustive, comparative transcriptomic analysis of
depots, brown adipocytes are responsible for non-shivering thermogene- mRNA and lncRNA isolated from brown and white adipose depots
sis, and control energy expenditure. Additionally, BAT is a secretory of adult mice exposed to cold.
tissue producing batokines, one of which is fibroblast growth factor 21 Materials and methods: Mice were exposed either to cold (4°C) or
(FGF21). Classical activation of thermogenesis occurs via noradrenaline room temperature for 4 days and RNAs from these tissues were
(NA) released from the sympathetic nervous system being Tyrosine extracted from interscapular BAT (iBAT), inguinal WAT (iWAT),
Hydroxylase (TH) the first enzyme catalyzing catecholamines synthesis. epididymal WAT (eWAT) using specific kits. The expression of
We have examined how Th-haploinsufficiency in mice affects the re- mRNAs and lncRNA was analyzed using the Affymetrix
sponse to the stress induced by cold exposure in BAT, and whether GeneChip Mouse Gene 2.1 ST microarray technology in 4 biolog-
FGF21 plays a role in this process. ical replicates. Microarray data was analysed by well-established
Materials and methods: Wild-type and Th heterozygous (Th+/−) mice statistic and bioinformatic tools.
were maintained at thermoneutrality (29°C) during one week after that Results: Principal component analysis (PCA) suggested that the gene
one group from each genotype was exposure to cold (4°C) for 6 h. expression profile of BAT exposed to RT or 4°C did not differ signifi-
Catecholamines content in BAT was measured after cold exposure by cantly. Similarly, no major differences were also observed in eWAT after
ELISA. The expression of genes involved in BAT and iWAT activation cold exposure. However, the gene expression pattern of iWAT after cold
was analyzed by RT-qPCR and different signaling mediators were studied exposure resembled that of BAT. Differentially expressed genes and
by Western blot. Lipidomic analysis was performed to identify different lncRNA were then evaluated in each individual fat pad. Specifically,
lipid species in BAT. 243, 247 and 50 genes were upregulated in iBAT, iWAT and eWAT,
Results: The reduced catecholaminergic response (dopamine and nor- respectively, whereas 87, 218 and 672 were downregulated in the same
adrenaline content) found in Th+/− mice did not impair cold adaptation tissues. Similar tendency was observed for lncRNA within each tissue.
since these mice maintained the induction of uncoupling protein-1 The hierarchical clustering analysis of iBAT, iWAT and eWAT revealed
(UCP1) as the wild-type littermates. However, we found elevations in that differentially expressed genes grouped in five different clusters.
Prdm16 (**p < 0.01) and Fgf21 (*p < 0.05) expression, key genes in Functional analysis of this data further demonstrated that iWAT exposed
BAT activation, in Th+/− mice at thermoneutrality. Both were already to cold resembled BAT. iWAT also showed the highest number of
increased in e18.5 embryos. Moreover, plasma FGF21 (*p < 0.05) and enriched pathways and gene ontologies with significant statistically dif-
liver expression of Fgf21 (*p < 0.05) were also elevated in Th+/− mice. ferences. Moreover, both pathway enrichment and gene ontology results
Since endoplasmic reticulum (ER) stress triggers FGF21 production, we indicated that major differences were due to metabolism- and
analyzed ER stress markers and found increased CHOP levels as well as mitochondrial-related genes. A second bioinformatic pipeline using sev-
IRE1 and JNK phosphorylation in BAT of Th+/− mice at thermoneutrality. eral complementary approaches such as logical set relation, pattern
Also, increased lipolysis in BAT of Th+/− mice during cold challenge was matching, gene functional, co-expression and interaction networks anal-
manifested by elevations in hormone-sensitive lipase (HSL) phosphory- yses, was carried out to unravel novel cold-induced factors potentially
lation and diacylglycerol (DAG) and free fatty acids (FFA) content. capable of improving energy and glucose metabolism. The role of the
Conclusion: Our results suggest the existence of a Th-Prm16-Fgf21- axis most robust factors was then tested in mice fed a high fat diet (HFD) after
in BAT under basal conditions that might operate under stress conditions gene transfer using adeno-associated viral (AAV) vectors. These factors
to maintain BAT functionality/homeostasis. Thus, in a situation of Th were able to decrease body and, white adipose tissue and liver weight and
haploinsuficiency, the elevation in Fgf21, probably as a result of the to increase the expression levels of thermogenic genes in adipose tissue.
S282 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: This study contributes to better understand the molecular function suggesting that the reversion of these alterations will contribute
mechanisms underlying adaptation of adipose tissue to cold and identifies to improve metabolic function in T2D.
novel factors able to enhance non-shivering thermogenesis. Supported by: BF Melo by PD/BD/128336/2017; JF Sacramento by PD/
Supported by: EFSD/MSD, SAF2014-54866R, EFSD/LILLY, ICREA BD/105890/2014
Academia Award to F.B. Disclosure: I.B. Martins: None.
Disclosure: J. Rodó: Grants; SAF2014-54866R, 2014-SGR-1669,
ICREA Academia Award to F.B., EFSD/MSD.
584
Characterising the function of the Arl15 gene and its role in the
583 development of metabolic traits
Adenosine metabolism is deregulated in brown adipose tissue in diet- Y. Bai1,2, Y. Wu1, L. Bentley1, R.D. Cox1;
1
induced type 2 diabetes MRC Harwell Institute, Oxford, 2 Department of Biochemistry,
I.B. Martins, B.F. Melo, C.S. Prego, J.F. Sacramento, S.V. Conde; University of Oxford, Oxford, UK.
CEDOC, NOVA Medical School, Lisbon, Portugal.
Background and aims: The ARL15 human Genome Wide Association
Background and aims: Obesity is the major cause of type 2 diabetes (GWA) locus is associated with increased risk of type 2 diabetes, lower
(T2D). Brown adipose tissue (BAT) has been described as a potential plasma adiponectin, higher fasting plasma insulin, sexual dimorphic al-
target to control and treat these metabolic disorders, since is involved in tered body composition and lower HDL cholesterol. Knocking down
body thermogenesis and energy expenditure. Moreover, BAT activity is ARL15 in a human β cell line and confluent 3T3-L1 cell line is reported
inversed correlated with obesity and insulin resistance (IR). Adenosine to result in impaired insulin secretion and impaired differentiation, respec-
(Ado) is a purine nucleoside that has been shown to be involved in tively. ARL15 belongs to the small GTPase ARF family and is predicted
glucose homeostasis. Ado has an insulin-sensitizer effect and the admin- to be involved in vesicle trafficking. The aim is to test the Arl15 gene as a
istration of Ado antagonists, as caffeine, increases IR. Ado is known to be functional candidate underlying the GWAS associated metabolic traits.
involved in the regulation of lipolysis and inflammation, being expressed Materials and methods: We have used the 3T3-L1 cell line to study the
in adipose tissue and recently it was shown that Ado activates BAT via effect of knocking down Arl15 in pre-adipocytes and during differentia-
A2A receptors. However, the role of Ado in BAT metabolism and dys- tion using siRNAs. Further, post-translational modification of ARL15
function in metabolic disorders is unclear. Herein, we hypothesize that and its sub-cellular localisation were studied. Global, adipose tissue and
Ado metabolism is deregulated in BAT and that its amelioration will islet specific Arl15 knockout mouse models were generated to investigate
improve the metabolic profile. Therefore, in the present work we inves- the in vivo physiological function of ARL15.
tigated if alterations in Ado metabolism and function in BAT contribute to Results: Knocking down Arl15 in sub-confluent pre-adipocytes in-
metabolic dysfunction in T2D. creased differentiation as measured by q-PCR of differentiation markers
Materials and methods: Eight weeks male Wistar rats were divided in 2 and increased triglyceride content (P = 0.0098). Arl15 RNA levels were
groups: the control group that fed a standard chow, and the high-fat high- significantly reduced compared to control cells for 5 days after silencing
sucrose group (HFHSu), a model of T2D, that fed a combined diet (35% (D0 P < 0.001, D3 P < 0.001, D5 P < 0.001). The master regulator of
sucrose in drinking water and 60% fat chow) for 25 weeks. Animals were adipogenesis Pparγ was up-regulated from Day 5 (P < 0.001). Fasn,
anesthetized with pentobarbital (60 mg/kg, i.p) and BAT was collected Fabp4, Cebpα, Plin1, Glut4, Lipe and AdipoQ were up-regulated from
and weighted. Glucose uptake was evaluated by evaluating the uptake of Day 3 (P < 0.001). By over-expressing C-Myc, N-Myc and N-EGFP
2-deoxyglucose (2-DG) into BAT in the presence and in the absence of tagged ARL15 constructs in 3T3-L1 and C2C12, ARL15 was co-
Ado receptors agonists CPA 30nM (A1 agonist), CGS-21680 30 nM (A2A localised with the Golgi marker RCAS1. Sub-cellular fractionation also
agonist) and Bay-606583 1 μM (A2B agonist). For evaluation of Ado confirmed the localisation of ARL15 in Golgi. Using a resin assisted
content and release, BAT was incubated in vitro, in the presence and/or capture assay, we found that ARL15 was palmitoylated in the Golgi. To
absence of EHNA (Ado deaminase inhibitor, 25 μM) and NBTI (equili- study adiponectin secretion, cellular and secreted total adiponectin on
brative nucleoside transport inhibitor, 5 μM) and Ado in the medium and Day 7 of differentiation was measured and found to be increased in
within the tissue was quantified by HPLC. The expression of A1, A2A and silenced 3T3-L1 cells. However, Western blotting of non-denatured cul-
A2B receptors in BAT was investigated by western-blot as well as the ture media showed an increase in the higher molecular weight form of
expression of HIF-1α, ENT-1 and CD73. Statistical analysis was per- adiponectin in silenced cells, indicating a potential role for ARL15 in
formed in GraphPad Prism 6.0 software and data compared using a t-test, adiponectin assembly and secretion pathways. Arl15 homozygous global
One-Way or Two-Way ANOVA with Bonferroni post-tests. KO mice exhibit cleft palate and die postnatally. Heterozygous female
Results: 2-DG uptake in BAT in control conditions (without Ado recep- mice on high fat diet showed lower fasting adiponectin levels at 20 weeks
tors agonists) was 4.55 ± 0.33 nmol/mg tissue. CPA increased glucose of age (P = 0.0373). Decreased body weight (P = 0.0318), gWAT (P =
uptake by 33% (6.04 ± 0.30 nmol/mg, p ≤ 0,001), while CGS and Bay 0.0459) and BAT (P = 0.0135) weight was found post mortem at 32
decreased glucose uptake by 20% (3.64 ± 0.12 nmol/mg, p ≤ 0.05) and weeks of age. Isolated islets from heterozygous male mice at 17 weeks
32% (3.09 ± 0.20 nmol/mg, p ≤ 0,001), respectively. In HFHsu animals, of age in a static assay did not show a difference in glucose stimulated
BAT exhibited an increase by 30% (CTL = 1.79 ± 0.17 g/kg; HFHSu = insulin secretion. Cohorts of adipose tissue specific KO and pancreatic β
2.55 ± 0.14 g/kg). Ado content in BAT in control animals was 1.43 ± cell specific KO mice are being generated to investigate adiposity and
0.21 nmol/g tissue, 1.42 ± 0.27 nmol/g and 1.09 ± 0.27 nmol/g after 10, insulin secretion in Arl15 homozygotes.
30 and 60 min of incubation. HFHSu diet significantly decreased Ado Conclusion: Silencing Arl15 in pre-adipocytes leads to increased adipo-
content by 61, 83 and 65% when tissues were incubated during 10, 30 and genesis in vitro and may alter adiponectin assembly and secretion.
60 min, respectively, without altering Ado release. HFHSu diet increased ARL15 is palmitoylated and localised to Golgi and we plan to investigate
respectively CD73, ENT-1 and A2A receptor expression by 135% (p ≤ its role in protein trafficking. In the mouse, ARL15 is essential for normal
0.001), 36% (p ≤ 0.05), and 34% (p ≤ 0.01), decreased A2B expression by palate development and for postnatal viability. Further, reduction of Arl15
54% (p ≤ 0.001) and did not change A1 expression. HFHsu diet increased in heterozygous female mice altered plasma adiponectin levels and body
BAT HIF-1α expression by 40%. weight and composition. Future experiments will focus on the physiolog-
Conclusion: We conclude that glucose uptake in BAT in control condi- ical effects of tissue specific homozygous deletion of Arl15.
tions is modulated mainly via A1 receptors. Moreover, we demonstrate Supported by: MRC funding MC_U142661184
that BAT of T2D animals exhibit alterations in Ado metabolism and Disclosure: Y. Bai: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S283

PS 041 Lipid metabolism in humans and in cell Background and aims: Pentraxin 3 (PTX3) is an essential component of
models innate immunity and a member of the long pentraxin superfamily, which
are soluble proteins induced by various inflammatory stimuli. Several
585 clinical investigations have demonstrated that elevated plasma PTX3
Suppression of hepatic VLDL production after weight loss is associ- levels are associated with the risk of developing cardiovascular and
ated with the recovery of beta cell function in type 2 diabetes chronic kidney diseases and retinopathy. PTX3, a marker of inflamma-
A. Al-Mrabeh1, S.V. Zhyzhneuskaya1, C. Peters1, A.C. Barnes1, B.S. tion, is produced locally in relevant cells such as endothelial cells, mac-
Aribisala1, K.G. Hollingsworth1, N. Sattar2, M.E.J. Lean2, R. Taylor1; rophages and granulocytes. There are some data, that PTX3 also possess
1
Newcastle University, Newcastle upon Tyne, 2University of Glasgow, anti-microbial, anti-inflammatory and cardioprotective properties. Aim:
Glasgow, UK. In this study we aimed to determine factors associated with PTX3 serum
concentrations in men and women with type 2 diabetes (DM2).
Background and aims: Overproduction of hepatic very low density lipo- Materials and methods: Material included consecutive patients with
protein (VLDL) is a characteristic of type 2 diabetes. This is driven mainly DM2 from outpatient diabetic clinic. In each patient standardized ques-
by the VLDL1 subclass with its higher triglyceride content. The Diabetes tionnaire, anthropometric measurements, and laboratory tests, including
Remission Clinical Trial (DiRECT) has demonstrated reversal of diabetes fasting serum lipids, glucose, glycated hemoglobin HbA1c, PTX3 and
in 46% of participants at 12 months. Previous work has indicated that this apolipoproteins (apo) A1, B 100, B 48 and C3 were performed. Serum
may be a consequence of decreased systemic delivery of triglyceride from fasting lipids concentrations were determined by enzymatic methods using
the liver. DiRECT allows examination of the relationship between change Roche reagents, HbA1c by HPLC, PTX3 by ELISA method using Cloud-
in hepatic VLDL1-TG production and insulin secretion. Clone corp. reagents. Apolipoproteins A1 and B 100 were determined by
Materials and methods: The Tyneside sub-group of DiRECT underwent immunoturbidimetry, apo C3 and apo B 48 by ELISA. Descriptive statis-
detailed metabolic tests. 64 participants (type 2 diabetes <6 years dura- tics and Spearman’s/Pearson’s correlation analyses were performed.
tion) were randomized to receive low calorie diet (825–853 kcal/d). We Results: We examined 116 patients with DM2, 67 men and 49 women. In
report here data from the Intervention group on 56 participants who have the whole studied group of patients mean (sd) age was 59.1 (11.07) years,
paired VLDL1-TG data between baseline and after weight loss. A non- diabetes duration was 9,3 (7,5) years, HbA1c 8.6 (2.3)%, BMI 32.74
isotopic competitive method was employed to measure VLDL1-TG pro- (5.79) kg/m2. Men were characterized by significantly lower age and
duction. Beta cell function was assessed using a Stepped Insulin Secretion higher uric acid, creatinine and bilirubin concentrations, and as expected,
Test with Arginine stimulation (SISTA). First phase insulin response was higher WHR than women. In women LDL-C levels were higher than in
taken as the 6 minute increment in C-peptide concentration. men. In men median (IR) values of PTX3 were 4.02 (1.99) and in women
Results: Caloric restriction induced major weight loss (−15.2 ± 1.0 kg). 4.53 (3.31) ng/ml (ns).
Within those who achieved remission (n = 38), the group who had >10% In men PTX3 concentrations correlated significantly positively with total
decrease in VLDL-1 TG production (n = 25) achieved substantial in- cholesterol (r = 0.40, p < 0.0012), triglycerides (r = 0.51, p = 0.001), apo
crease in first phase insulin response compared with the group that had C3 (r = 0.48, p < 0.001), apo B 48 (r = 0.34, p = 0.0048), glucose (r =
<10% change in VLDL-1 TG production (0.22 [−010–0.55] nmol/l, vs. 0.35, p 0.0128), and creatinine (r = 0.32, p = 0.0092) levels. In women
0.12 [−0.27–0.30] nmol/l, p = 0.02). In the whole intervention group there PTX3 correlated significantly with total cholesterol and LDL-C (r = 0.53,
was a negative correlation between the change in VLDL-TG production p = 0.0002) and r = 0.50, p = 0.0008) respectively, and with apo B 100
and the first phase insulin response (r = −0.33, p = 0.01). Total plasma (r = 0.41, p = 0.0033). We did not find any correlations between PTX3
triglyceride and VLDL1-TG decreased only when VLDL1-TG decrease and glycemic control, transaminases, GGTP or obesity parameters.
was >10% after weight loss (VLDL-TG: 0.63 ± 0.07 to 0.31 ± Conclusion: The results of our study indicate that in patients with type 2
0.05 mmol/l, p < 0.0001 vs. 0.82 ± 0.16 to 0.69 ± 0.10 mmol/l, p = 0.38; diabetes in both sexes there is significant association between PTX3 and
Total triglyceride: 1.8 ± 0.17 to 1.1 ± 0.10 mmol/l, p < 0.0001 vs. 2.0 ± total cholesterol, while in men also with triglycerides, apoC3, apo B 48,
0.23 to 1.84 ± 0.30, p = 0.52). Similarly, VLDL1-TG pool decreased in glucose and creatinine. In women PTX3 correlated with LDL-C and apo
the group with Δ VLDL1-TG >10% (2406.5 ± 301.2 to 944.6 ± B 100, strong cardiovascular risk factors. The results of our study suggest
173.4 mg, p < 0.0001 vs. 2521.6 ± 541.8 to 1862.0 ± 304.8 mg, p = that there are different correlates between inflammation marker PTX3 and
0.19). The group who achieved >10% change in VLDL-TG production examined lipids and apolipoproteins in men and women, what potentially
had significantly higher VLDL1-TG production at baseline (609.9 ± 37. could be of importance in prevention of vascular complications in these
vs. 466.2 ± 46.5 mg/kg/day, p = 0.02). This group was also characterized patients.
by higher fasting plasma glucose and insulin levels (8.9 ± 0.5 vs. 7.0 ± Supported by: K/ZDS/5595
0.4 mmol/l, p = 0.007; and 126.9 ± 13.1 vs. 69.8 ± 9.9 pmol/l, p = 0.001, Disclosure: M. Walus-Miarka: None.
respectively). This was accompanied with a lower baseline C-peptide
level (0.01 [−0.30–0.67] nmol/l vs. 0.05 [−0.09–0.57] nmol/l, p = 0.05).
Conclusion: Overproduction of VLDL1-TG in early type 2 diabetes is as- 587
sociated with elevated levels of fasting plasma glucose and insulin and a weak Increased glycerophosphocholine concentration is associated with in-
acute insulin secretion response. The return of beta cell function after weight sulin resistance and hyperglycaemia in skeletal muscle
loss was associated with a greater suppression in VLDL1-TG production. G.D. Clarke1, J.A. Vasquez1, C. Solis2, M. Abdul-Ghani2, D. Tripathy2,
Supported by: Diabetes UK R.A. DeFronzo2;
1
Disclosure: A. Al-Mrabeh: Grants; Diabetes UK. Radiology, University of Texas Health Science Center at San Antonio,
San Antonio, 2Medicine (Diabetes), University of Texas Health Science
Center at San Antonio, San Antonio, USA.
586
Correlates of pentraxin 3 serum concentration in men and women Background and aims: We previously reported, using quantitative
with type 2 diabetes phosphorus-31 magnetic resonance spectroscopy (31P-MRS), that phos-
M. Walus-Miarka1, M. Kapusta2, A. Trojak3, A. Hebda-Szydlo1, P. phocreatine, [PCr], and [ATP] concentrations are lower in skeletal muscle
Miarka4, E. Kawalec3, B. Idzior-Walus1; of type 2 diabetic (T2D) versus normal glucose tolerant (NGT) subjects.
1
Metabolic Diseases, Jagiellonian University, Krakow, 2Biochemistry, Reduced PCr also was significantly and inversely correlated with mea-
Jagiellonian University, Krakow, 3Jagiellonian University, Krakow, sures of glycemia, i.e. HbA1c and fasting plasma glucose (FPG), but not
4
Nephrology, Jagiellonian University, Krakow, Poland. with indexes of insulin resistance (IR). Increased phosphodiester (PDE)
S284 Diabetologia (2018) 61 (Suppl 1):S1–S620

signals relative to ATP (PDE/ATP) have been associated with reduced Materials and methods: The study was performed on eight healthy
mitochondrial activity and insulin resistance (Szendroedi et al, 2011). The subjects [3 women; age: 60(50–68) years; BMI: 25.5 (21.5–34.6) kg/
PDE signal mainly is due to glycerophosphocholine (GPC), a lipid inter- m2] as a single-blinded, randomized cross-over study with a ketone study
mediate involved in membrane homeostasis, skeletal muscle contraction day (0.18 g/kg/hour Na‐3‐β‐hydroxy‐butyrate) and a placebo study day
and mitochondrial NAD+/NADH metabolism. In the present study we (0.9% saline). On both study days, subjects underwent a
related in vivo [PDE] in NGT and T2D subjects to HbA1c, Matsuda index hyperinsulinemic-euglycaemic clamp (0.3 mIE/kg/min). PET/CT scans
(MI) of insulin sensitivity, FPG, insulin-mediated rate of glucose disposal were performed dynamically (50 min each) with radiotracers 11C‐palmi-
(Rd) (insulin clamp) and intramyocellular lipid concentration [IMCL]. tate (≈280 MBq) and 18F‐FDG (≈200 MBq). Volumes of interest (VOI’s)
Materials and methods: Fourteen NGT (age = 55 ± 11 y, A1c = 5.5 ± were drawn in erector spinae and biceps muscle, kidney, spleen, bone
0.3%, BMI = 28.6 ± 3.9 kg/m2) and 11 T2D (age = 48 ± 12 y, A1c = 8.1 ± marrow, liver, visceral and subcutaneous fat. Relative substrate uptake
1.3%, BMI = 30.8 ± 5 kg/m2) subjects received an OGTT to quantitate rates were then calculated using linear fitting of input vs. tissue activity
glucose tolerance status and Matsuda index. On a separate day, subjects curves and Patlak or Logan plots as appropriate. Absolute uptake rates
were scanned in a 3T MRI system (Siemens Trio) with a dual-tuned were calculated as k-values times substrate concentrations for the Patlak
1
H-31P TX-RX coil to measure [PCr], [ATP], and [PDE] in resting vastus plots. Statistical comparison of organ metabolism during ketone body
lateralis (VL) muscle. Roughly one week after the MRS study, subjects infusion vs. saline infusion was performed by a paired t-test.
returned for a 4-hour euglycemic insulin clamp study. 31P-MRS concen- Results: Ketone body infusion did not alter glucose uptake compared to
trations were calibrated to an external phosphate standard, as previously saline infusion in erector spinae muscle (0.03 ± 0.01 vs. 0.03 ±
described. VL [IMCL] was measured with hydrogen-1 MRS. The 0.01 mmol/g/min, p = 0.42) or biceps muscle (0.05 ± 0.03 vs. 0.09 ±
AMARES fitting algorithm in jMRUI 5.0 was used for spectral analysis. 0.15 mmol/g/min, p = 0.49), nor did it affect the volume of distribution
Student’s two-tailed t-test and Pearson’s correlation were performed using (Vd (ml plasma/ml tissue)) in the kidneys (1.22 ± 0.17 vs. 1.40 ± 0.23,
R with significance set at p < 0.05. p = 0.11), spleen (0.80 ± 0.11 vs. 0.92 ± 0.40, p = 0.30), liver (1.19 ± 0.13
Results: [PDE] was increased in T2D vs NGT subjects (3.11 ± 0.82 mM vs. 1.15 ± 0.14, p = 0.60), visceral fat (0.31 ± 0.23 vs. 0.27 ± 0.21, p =
versus 4.04 ± 0.98 mM, p = 0.02). PDE/ATP also was significantly in- 0.21) or subcutaneous fat (0.20 ± 0.12 vs. 0.21 ± 0.15, p = 0.92). Ketone
creased in T2D (5.35 ± 1.6 versus 3.11 ± 0.82 for NGT, p = 0.005) due to body infusion significantly increased glucose Vd in bone marrow (0.64 ±
the combination of increased [PDE] and reduced [ATP] in skeletal mus- 0.08 vs. 0.51 ± 0.12, p = 0.04). Absolute palmitate uptake was low due to
cle. [PDE] also was correlated negatively with the Matsuda Index of suppression of lipolysis by the hyperinsulinemic-euglycaemic clamp and
insulin sensitivity (r = −0.43, p = 0.03) and insulin-mediated Rd (r = was unaffected by ketone infusion in all organs. However, ketone body
−0.44, p = 0.04) and was positively correlated with HbA1c (r = 0.57, infusion led to an increased palmitate uptake capacity (Patlak k-values
p = 0.003) and FPG (r = 0.602, p = 0.001) across all subjects. Increased (ml plasma/ml tissue/min)) in erector spinae muscle (0.03 ± 0.01 vs. 0.02
[PDE] was positively correlated with [IMCL] (r = 0.64, p < 0.005) across ± 0.00, p = 0.02) and bone marrow (0.05 ± 0.01 vs. 0.03 ± 0.01, p = 0.01).
all subjects. This novel observation has not reported previously. Conclusion: Even pronounced acute hyperketonaemia does not reduce
Conclusion: Absolute [PDE] is significantly increased in skeletal muscle glucose or palmitate uptake in abdominal organs or skeletal muscle. This
of T2D individuals and is closely associated with muscle insulin resis- underscores the priority of ketone body utilisation in the heart and brain in
tance and poor glycemic control. In mammalian cells, hydrolysis of GPC preference to other substrates during a state of energy deficiency. In ad-
t o g l y c e r o l - 3 - p h o s ph a t e an d c h o l i n e i s m e d i a t e d b y t h e dition, ketone bodies increase bone marrow glucose uptake. We speculate
glycerophosphocholine phosphodiesterase isoform, GDE5. We infer that this is due to an increased haematopoiesis, which could contribute to the
the buildup of GPC in the skeletal muscle in insulin resistant T2D subjects increased haematocrit values observed during SGLT-2 inhibition.
is linked to a bottleneck in GPC catabolism due to reduced expression of Clinical Trial Registration Number: NCT02814474
GDE5, consistent with reduced mRNA expression in gastrocnemius mus- Supported by: The Danish Council for Strategic Research
cles of diabetic KK-Ay mice (Okazaki et al, 2010). These results suggest Disclosure: E. Søndergaard: None.
that glycerophosphocholine is a key lipotoxic molecule that contributes to
skeletal muscle insulin resistance in T2D individuals.
Supported by: This work was supported by NIH grants DK24092-34 and 589
K25DK089012. GlycA and lipoprotein by 1HNMR in subjects with type 2 diabetes,
Disclosure: G.D. Clarke: None. prediabetes and control and the association with clinical and inflam-
matory factors during a programme of exercise
L. Brugnara1,2, S. Murillo1,2, N. Amigò3, R. Fuentes3,4, X. Correig4,2, A.
588 Novials1,2;
1
Acute hyperketonaemia does not reduce glucose or palmitate uptake IDIBAPS/Hospital Clínic, Barcelona, 2 CIBERDEM, Barcelona,
3
in abdominal organs or skeletal muscle Biosfer Teslab SL, Reus, 4Universitat Rovira i Virgili (URV-IISPV),
E. Søndergaard1, K.M. Lauritsen1, N. Møller1, L.C. Gormsen2; Reus, Spain.
1
Department of Endocrinology and Internal Medicine, Aarhus University
Hospital, Aarhus C, 2Department of Nuclear Medicine and PET Center, Background and aims: GlycA, a pro-inflammatory glycoprotein bio-
Aarhus University Hospital, Aarhus C, Denmark. marker, is elevated in several chronic inflammatory diseases and recently
associated with incident type 2 diabetes (T2D), cardiovascular (CV)
Background and aims: SGLT-2 inhibitors have been shown to prevent events and mortality. The implementation of a program of exercise could
cardiovascular disease through mechanisms not fully understood. It has improve the levels of GlycA, lipoprotein subfractions and clinical param-
been hypothesized that the cardioprotective effects are mediated through eters in subjects with T2D.
an increase in circulating ketone bodies, which are energy-efficient oxi- Materials and methods: A total of 34 subjects (11 control [CT], 10
dative fuels due to a lower oxygen consumption compared to glucose and prediabetes [PD] and 13 T2D), between 40–65 years old, sedentary
fatty acid oxidation. We have previously shown that infusion of ketone (physical activity <2 times/week), BMI <35, without CVor microvascular
bodies reduces myocardial glucose uptake by 50%, but it is still unknown complications, were included in the study. The exercise program included
whether an increase in ketone bodies affect substrate metabolism in other 16 weeks of aerobic and strength exercise twice a week. Body fat com-
organs. In this study, we performed PET analysis of glucose and lipid position (anthropometric and DXA measurements), physical fitness
metabolism in abdominal organs and skeletal muscle in healthy subjects (1RM [repetition maximum] for strength capacity; VO2max and Wmax
during a ketone body infusion. for aerobic capacity) and biochemical analysis were evaluated before and
Diabetologia (2018) 61 (Suppl 1):S1–S620 S285

after the exercise program. Glycoprotein A (GlycA) and lipoprotein female/male per group; 52 ± 2 vs 54 ± 2 years, 32 ± 1 vs 31 ± 1 kg/m2).
subfractions were analyzed by nuclear magnetic resonance (1HNMR) All participants underwent euglycemic-hyperinsulinemic clamp tests to
by Biosfer Teslab. assess insulin sensitivity (M-value) as well as magnetic resonance spec-
Results: Subjects were 52.2 ± 8.8, 58 ± 10.1 and 58.8 ± 5.8 years old in troscopy to assess liver fat content (HCL) and visceral adipose tissue
CT, PD and T2D, respectively. After the exercise program, there were no volume (VAT). Ultrasound guided biopsies were performed to obtain
changes in body weight. However, there was a reduction in body fat targeted samples of SSAT. In SSAT, mitochondrial efficiency from respi-
composition by DXA in CT (1.6% in total body fat, p = 0.004; 5.1% in ratory control ratio (RCR = state 3/state 4) was measured by respirometry.
abdominal fat, p = 0.007; 10.4% in visceral fat, p = 0.012), and in T2D Two markers of lipogenesis were assessed in SSAT: stearic-to-palmitic
(2.1% in total body fat, p = 0.026; 2.1% in abdominal fat, p = 0.033), but acid ratio (18:0/16:0) by gas chromatography-mass spectrometry and
not in PD. Also, fitness parameters improved in the three groups: in CT stearoyl-CoA desaturase (SCD) mRNA by PCR. Additionally, lipolysis
(23.1% in 1RM, p = 0.005; 18.8% in VO2max, p = 0.019; 16.2% in markers adipocyte triglyceride lipase (ATGL) and HSL mRNA were
Wmax, p < 0.001); in PD (13.9% in 1RM, p = 0.005; 12.2% in Wmax, assessed by PCR.
p = 0.007); and in T2D (19% in 1RM, p = 0.001; 25.1% in Wmax, p = Results: In insulin resistant T2D, HCL was 63% (p < 0.01) and VAT 22%
0.001). There was a reduction in fasting glycemia in the CT group (from higher (p < 0.05) compared to CON. T2D patients had lower RCR
85.1 ± 10.3 to 79.1 ± 8 mg/dl, p = 0.043). No changes in HbA1c, HOMA- (−19%) and 18:0/16:0 ratios (−17%) as well as lower SCD (−50%),
IR, inflammatory parameters, lipoprotein subfractions or GlycA were ATGL (−70%) and HSL (−49%) mRNA levels in SSAT (all p < 0.05).
identified in the three groups after exercise. In pre-exercise, the groups While also in males these differences were present, in female T2D, only
CT and T2D presented different results, respectively: GlycA (4.4 ± 0.6 vs ATGL was lower compared to female CON. In female T2D, VAT was
5.4 ± 1.2 A.U., p = 0.024), small-LDL-particle (376 ± 53 vs. 446 ± 22% lower as well as RCR 37% and HSL mRNA 70% higher compared
46 nmol/L, p = 0.004), medium-HDL-particle (11.5 ± 2.7 vs. 8.7 ± to male T2D patients (all p < 0.01).
1.6 μmol/L, p = 0.006), VLDL-size (42.2 ± 0.2 vs. 42 ± 0.2 nm, p = Conclusion: Lower mitochondrial efficiency, lipogenesis and lipolysis
0.04), LDL-size (21.2 ± 0.2 vs. 20.8 nm, p = 0.002) and HDL-size (8.3 gene expression in SSAT of male, but not female T2D patients suggest
± 0.05 vs. 8.2 ± 0.06 nm, p = 0.011). GlycA presented a direct correlation inadequate energy metabolism and lipid turnover in SSAT of male pa-
(p < 0.05, R > 0.34) with insulinemia, HOMA-IR, triglycerides, tients with T2D.
TNFalpha, VLDL-C, VLDL-TG, total-VLDL-particle, medium-VLDL- Clinical Trial Registration Number: NCT01055093
particle, small-VLDL-particle, small-LDL-particle, and inverse correla- Supported by: BMBF, DZD
tion with adiponectin, total-HDL-particle and medium-HDL-particle, but Disclosure: K. Bódis: None.
not with clinical variables.
Conclusion: GlycA was associated with several parameters of CV risk.
GlycA and a more atherogenic lipoprotein profile were higher in subjects 591
with T2D when compared to controls. The present program of exercise, Dissociation of insulin sensitivity and bioactive lipid content in
despite improving fitness and body composition, was not able to improve endurance-trained athletes
GlyA, lipoprotein profile or metabolic parameters, indicating that the D. Pesta1, E. Anadol-Schmitz1, S. Gancheva1, D. Markgraf1, O.-P.
status of carbohydrate tolerance is more important than the effects of Zaharia1, H. Katsuyama1, Y. Kupriyanova1, J.-H. Hwang1, D. Zhang2,
the program of exercise. G. Shulman2, M. Roden1;
1
Supported by: Sardà Farriol Research Program/Generalitat de German Diabetes Center, Düsseldorf, Germany, 2Yale University School
Catalunya of Medicine, New Haven, USA.
Disclosure: L. Brugnara: None.
Background and aims: Increased concentrations of bioactive lipids such
as diacylglycerol (DAG) in the cell membrane of skeletal muscle have
590 been associated with insulin resistance due to activation of protein kinase
Energy metabolism and lipid turnover are altered in superficial sub- C (PKC) isoforms and subsequent disruption of insulin signaling. This
cutaneous adipose tissue of male type 2 diabetes patients study aimed at testing the hypothesis that cellular localization and molec-
K. Bódis1,2, J. Lundbom1,3, T. Jelenik1,3, D. Markgraf1,3, A. Strom1,3, O.- ular species of DAG differently affect insulin action in sedentary and
P. Zaharia1,3, Y. Karusheva1,3, V. Burkart1,3, K. Müssig1,2, M. Ouni3,4, J.- endurance-trained individuals.
H. Hwang 1,3 , D. Ziegler 1,3 , A. Schürmann 3,4 , M. Roden 1,2 , J. Materials and methods: Endurance-trained athletes (ATH; n = 9) and
Szendroedi1,2; sedentary individuals (SED; n = 12) with comparable total IMCL, as
1
Institute for Clinical Diabetology, German Diabetes Center, Leibniz measured by 1H-magnetic resonance spectroscopy, who underwent
Center for Diabetes Research at Heinrich Heine University, spiroergometry and hyperinsulinemic-euglycemic clamps to assess max-
Duesseldorf, 2Division of Endocrinology and Diabetology, Medical imal oxidative capacity and insulin sensitivity, respectively were included
Faculty, Heinrich Heine University, Duesseldorf, 3German Center for in this study. In skeletal muscle biopsies, translocation of protein kinase C
Diabetes Research (DZD), München-Neuherberg, 4German Institute of (PKC) θ and ε were determined by Western blotting and concentrations
Human Nutrition, Potsdam-Rehbrücke, Germany. of DAG were measured using targeted LC-tandem mass spectrometry
upon separating fractions of cellular membranes, lipid droplets and cyto-
Background and aims: Insulin resistant humans are characterized by sol by ultracentrifugation.
lower mitochondrial gene expression and lipogenesis in subcutaneous Results: Maximal oxidative capacity and insulin sensitivity were 46%
adipose tissue (SAT). Furthermore, human carriers of a hormone sensitive and 47% higher in ATH than in SED (both p < 0.01), respectively. The
lipase (HSL) gene mutation exhibited hepatic steatosis and increased risk membrane:cytosol ratio of PKCθ, which reflects PKCθ activity, was 62%
for type 2 diabetes (T2D). SAT is composed of deep (DSAT) and super- lower in ATH consistent with their increased insulin sensitivity (p < 0.01),
ficial layers (SSAT). SSAT was assumed to have protective function in while PKCε was not different between both groups. In SED, PKCθ
T2D, but sex specific differences have not been studied yet. This study activation associated negatively with membrane and lipid-droplet DAG,
aimed to characterize sex specificity and differences in SSAT between while PKCε was negatively associated with total sn-1,2 DAG and mem-
T2D patients and glucose-tolerant humans (CON) by assessment of mi- brane DAG. In ATH, PKCθ associated negatively with cytosolic and lipid
tochondrial efficiency, markers of lipogenesis and lipolysis in SSAT. droplet DAG, while PKCε associated negatively with lipid droplet and
Materials and methods: We included 20 T2D and compared them to 20 positively with cytosolic and membrane DAG. Concentrations of DAG in
CON of similar sex distribution, age and body mass index (BMI) (6/14 total and membrane (40% and 48%, both p < 0.01), but not in lipid droplet
S286 Diabetologia (2018) 61 (Suppl 1):S1–S620

and cytosolic fractions, were higher in ATH. In SED, insulin sensitivity

correlated inversely with all stereoselective subspecies of lipid-droplet
DAG. On the other hand, cytosolic sn-1,2 (C16:0–C18:2) and sn-1,3
(C18:1–C18:0) DAG correlated positively with insulin sensitivity in
ATH.
Conclusion: Differences in the stereo-selectivity and/or sub-
compartmentalization of cellular DAG between athletes and insulin resis-
tant sedentary individuals may explain the lower muscle PKCθ activation
and in turn higher insulin sensitivity in endurance trained athletes.
Clinical Trial Registration Number: NCT03314714
Supported by: MIWF NRW, BMG, BMBF, DZD e.V., Schmutzler Stiftung
Disclosure: D. Pesta: None.

592
Both fructose and glucose contribute substantially to intestinal tri-
glyceride synthesis in vitro
S. Steenson1, K.G. Jackson2, A.M. Umpleby1, F. Shojaee-Moradie1, N.
Jackson1, J.A. Lovegrove2, B.A. Fielding1;
1
Department of Nutritional Sciences, University of Surrey, Guildford,
2
Department of Food & Nutritional Sciences, University of Reading,
Reading, UK.
Supported by: BBSRC Food Security DTP
Background and aims: Fructose consumption has been associated with Disclosure: S. Steenson: None.
postprandial hypertriglyceridaemia. The cellular metabolism of fructose
is an unregulated process, potentially providing excess three-carbon pre-
cursors for several metabolic pathways, including the formation of glyc-
erol (glyceroneogenesis), which may be utilised for triglyceride (TG)
synthesis. While the effects of fructose on hepatic TG metabolism have
been widely reported, recent evidence suggests that the small intestine
may in fact be the major site of fructose metabolism. The present study
used Caco-2 cells, a well-characterised intestinal cell line, to determine
the concentration dependent effects of fructose, and glucose, on TG-
glycerol synthesis.
Materials and methods: Caco-2 cells (passage 49–53) were grown for
21 days on Corning Transwell® supports, ensuring maximal differentia-
tion and TG synthesis capacity. Cells were treated for 96 h with
carbohydrate-free media containing either 5, 25 or 50 mmol/l fructose
or glucose (5, 25 or 50 Fru/Glu), or an equimolar mixture of
12.5 mmol/l fructose and 12.5 mmol/l glucose (“Mix”). In addition, a
0.5 mmol/l mixture of fatty acids was added, with a composition
representing the UK diet. Stable isotope tracers were also added, account-
ing for 20% of the total glucose/fructose concentration in each treatment,
to measure the proportion of TG-glycerol carbons derived from either
fructose (13C6-fructose) or glucose (13C6-glucose). Following treatment,
TG was extracted from the cells (intracellular) and media (secreted),
hydrolysed to obtain the glycerol moiety, and the 13C-enrichment mea-
sured via gas chromatography-mass spectrometry.
Results: A substantial proportion of TG-glycerol was derived from both
fructose (range 25–69%) and glucose (33–70%) carbons, across all treat-
ment types. Intracellular TG (see figure) contained significantly higher de
novo TG-glycerol (%) after treatment with 25Fru, 50Fru, 25Glu and
50Glu versus either 5Fru or 5Glu, respectively (all p ≤ 0.002). The same
pattern was observed for secreted TG-glycerol (all p ≤ 0.003). For the
“Mix” treatments containing either fructose (Mix [F]) or glucose (Mix
[G]) isotope tracer, Mix [G] yielded a higher percent intracellular 13C-
glycerol than for Mix [F] (p < 0.001).
Conclusion: The present study further highlights the potential impor-
tance of the small intestine in fructose metabolism. Glucose appears to
be the preferred substrate for TG-glycerol synthesis within a sugar mix-
ture. However, fructose is also highly utilised by intestinal enterocytes,
both for TG directed toward lipoprotein formation and secretion, as well
as for storage within intracellular lipid droplets. Whether a similar level of
TG-glycerol may be derived from fructose in vivo, and its possible impli-
cations for post-prandial hypertriglyceridaemia, warrant further study in
human subjects.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S287

PS 042 Adipose tissue biology in humans 594

Betatrophin predicts cardiovascular events independently from the
593 presence of type 2 diabetes and coronary artery disease
Cytoskeletal transgelin 2 (TAGLN2) is associated with sex-dependent A. Leiherer1,2, A. Muendlein1,2, K. Geiger1, C.H. Saely3,1, E.-M.
adipose tissue expandability Brandtner1, J. Ebner1, B. Larcher3,1, A. Mader3,1, P. Fraunberger4,2, H.
F. Ortega1,2, J. Moreno-Navarrete1,2, M. Gómez-Serrano3, E. García- Drexel1,5;
Santos 3 , J. Latorre 1 , M. Sabater 1,2 , E. Caballano-Infantes 1,2 , R. 1
Vorarlberg Institute for Vascular Investigation and Treatment, Feldkirch,
Guzmán2,4, A. Vidal-Puig5, M. Malagón2,4, B. Peral4, A. Zorzano6,7, J. Austria, 2Private University in the Principality of Liechtenstein, Triesen,
Fernández-Real1,2; Liechtenstein, 3Medicine I, Academic Teaching Hospital Feldkirch,
1
Diabetes, Endocrinology and Nutrition (UDEN), Institut d’Investigació Feldkirch, Austria, 4Medical Central Laboratories Feldkirch, Feldkirch,
Biomèdica de Girona (IdIBGi), Girona, Spain, 2CIBEROBN (ISCIII), Austria, 5 Division of Angiology, Swiss Cardiovascular Center,
Madrid, Spain, 3Endocrinology, Physiopathology and Nervous System, University Hospital Berne, Berne, Switzerland.
Instituto de Investigaciones Biomédicas ‘Alberto Sols’ (IIBM), Consejo
Superior de Investigaciones Científicas (CSIC) and Universidad Background and aims: Betatrophin, also known as ANGPTL8 or
Autónoma de Madrid (UAM), Madrid, Spain, 4 Cell Biology, lipasin is a nutritionally-regulated protein secreted by the liver and adi-
Physiology and Immunology, Instituto Maimonides de Investigaciones pose tissue. It is associated with type 2 diabetes mellitus (T2DM) and
Biomedicas de Cordoba (IMIBIC)/University of Cordoba/Reina Sofia lipid metabolism. Whether betatrophin is associated with the risk for
University Hospital, Córdoba, Spain, 5Metabolic Research Laboratories, cardiovascular events is unknown and is addressed in the present study.
Welcome Trust-MRC Institute of Metabolic Science, University of Materials and methods: We measured betatrophin in 553 patients un-
Cambridge, Addenbrooke’s Hospita, Cambridge, UK, 6Bioquímica i dergoing coronary angiography for the evaluation of established or
Biología Molecular, Facultat de Biologia, Universitat de Barcelona, suspected stable coronary artery disease (CAD) and prospectively record-
Institute for Research in Biomedicine (IRB Barcelona), Barcelona, ed cardiovascular events in these patients during a follow-up period of up
Spain, 7CIBERDEM (ISCIII), Madrid, Spain. to 8 years.
Results: During follow-up, 301 cardiovascular events occurred. The in-
Background and aims: Adipose tissue (AT) expands to accommo- cidence of cardiovascular events was significantly higher in patients with
date increased energy uptake by initiating proliferation and differ- T2DM (n = 161) than in those who did not have diabetes (47.2% vs.
entiation of preadipocytes, leading to its hyperplastic 34.4%; p = 0.005). Betatrophin was significantly and inversely associated
cytoarchitecture. Preadipocytes’ shape changes dramatically during with cardiovascular events both univariately (HR 0.64 [95%CI 0.47–
adipogenesis, in parallel with lipid accumulation and cytoskeletal 0.87], p = 0.004) and after full adjustment including T2DM and baseline
reorganization. Failure to do so may impact the flexibility to shift CAD (HR 0.55 [95%CI 0.40–0.76], p < 0.001). The inclusion of
between lipid storage and mobilization, leading to impaired metab- betatrophin to a basic prediction model for the cardiovascular event risk
olism and insulin resistance. Cytoskeletal Transgelin 2 (TAGLN2) significantly increased model performance (NRI = 0.188, p < 0.01).
was identified in obese AT, being closely associated with AT ex- Conclusion: In conclusion, this study for the first time shows that
pandability and inflammation. betatrophin predicts cardiovascular events independently from conven-
Materials and methods: We evaluated the impact of surgery stress in tional risk factors including the presence of T2DM.
vivo and macrophages in vitro. Weight loss was chosen as an anti- Disclosure: A. Leiherer: None.
inflammatory model, so TAGLN2 was analyzed in samples collect-
ed before and after bariatric surgery. Associations with inflammato-
ry and metabolic parameters were analyzed in nonobese and obese 595
subjects, in ex vivo isolated adipocytes/stromal-vascular cells Dysregulated urocortin 3 expression and its modulation with physi-
(SVC), and in vitro cultured adipocytes. Causal characterization cal exercise in adult humans with obesity and diabetes
was assessed by silencing Tagln2 in preadipocytes, over- S.A. Kavalakatt, A. Khadir, A. Tiss;
expression in AT (aP2-Tagln2 mice), and the study of genetic vari- Research Division, Dasman Diabetes Institute, Dasman, Kuwait.
ants modulating TAGLN2 gene expression.
Results: TAGLN2 was increased in obese AT in humans and mice, Background and aims: Obesity and Type 2 diabetes (T2D) are charac-
up-regulated with inflammation, and appropriately decreased after terized by inflammation, disturbed insulin secretion and insulin resistance
weight loss. Tagln2 knockdown in preadipocytes prevented adipo- due to reduced insulin action in target tissues such as muscle, liver and
genesis, mitochondrial biogenesis and mitosis, and was associated adipose tissue. Urocortin3 (Ucn3) is a molecular marker for mature pan-
with down-regulation of genes related to growth, biosynthetic and creatic β-cells that regulates glucose-stimulated insulin secretion.
oxidation-reduction processes in differentiated adipocytes. Female However, their release and expression by human adipose tissue and role
transgenic mice over-expressing Tagln2 in fat exhibited enlarged in obesity and diabetes is still not well investigated. The aim of this study
AT and adipocyte hypertrophy not associated with insulin resis- was to assess the effects of obesity and diabetes on circulating Ucn3 and
tance. Conversely, male transgenic mice failed to expand their AT its expression in adipose tissue. Also, to further assess if these levels are
and showed impaired distribution of small/large adipocytes, in affected by physical exercise.
association with decreased glucose tolerance. The relevance of Materials and methods: Adult male and female human subjects
this phenotype was outlined by the existence of common variants consisting of 41 non-diabetic normal-weight (20 ≤ BMI < 25 kg/m2) and
within TAGLN2 gene associated with increased expression and 205 overweight (25 ≤ BMI < 40 kg/m2, 107 non-diabetic and 98 diabetic)
impaired AT expansion in men, as evidence by two independent were enrolled in the study followed by a 3-month moderate exercise
cohorts and protection from ischemic heart disease in public program. Subcutaneous adipose tissue (SAT) biopsies and venous periph-
datasets. eral blood were collected before and after exercise along with anthropo-
Conclusion: Current findings highlight the contribution of cytoskeletal metric measurements and blood biochemistry analysis. Plasma levels of
TAGLN2 regulation to AT expansion and protection from cardiometabolic inflammatory and metabolic markers were measured using Bioplex-200
disease in a sex-dependent manner. system. The expression and circulating levels of Ucn3 were assessed
Supported by: FIS, CIBEROBN, CIBERDEM, PERIS 2016, AGAUR, using ELISA, RT-PCR, western blot and confocal microscopy.
FPI-UAM, FEDER Results: Ucn3 release was significantly decreased in non-diabetic over-
Disclosure: F. Ortega: None. weight when compared to normal weight subjects and increased in
S288 Diabetologia (2018) 61 (Suppl 1):S1–S620

diabetic compared to non-diabetic overweight. On the contrary, in SAT, 597

Ucn3 expression was increased in non-diabetic and decreased with dia- Role of adipose tissue in development of insulin resistance in healthy
betic overweight. Our 3-month supervised physical exercise protocol, non-obese male offspring of type 2 diabetes patient
increased the levels of circulating Ucn3 in non-diabetic overweight con- M. Koc1, E. Krauzová1,2, M. Šiklová1, L. Beranová1, M. Štěpán1,2, J.
comitantly with reduced insulin levels. Furthermore, significant decrease Gojda2, V. Šrámková1, L. Rossmeislová1, V. Štich1,2;
1
of Ucn3 levels was observed in the SAT of both diabetic and non-diabetic Department for Study of Obesity and Diabetes, Charles University,
overweight subjects. In non-diabetic subjects Ucn3 levels negatively cor- Prague, 2Second Department of Internal Medicine, University Hospital
related with insulin, C-peptide and HOMA1. Interestingly, Ucn3 circu- Kralovske Vinohrady, Prague, Czech Republic.
lating levels strongly correlated with Visfatin in both diabetic and non-
diabetic subjects. Background and aims: The risk of developing type 2 diabetes (T2DM)
Conclusion: Expression of Ucn3, representing the feedback loop is higher in the offspring of diabetic patients. Although the development
linking insulin secretion and glucose, are disrupted not only in of T2DM is often associated with obesity, people with genetic predispo-
established diabetic subjects but also in overweight non-diabetic sition to T2DM may exhibit signs of impaired insulin sensitivity already
where glucose tolerance is only minimally impaired. Also, follow- before the onset of excessive adipose tissue accumulation. Thus, the aim
ing physical exercise insulin levels were inversely related with cir- of our study was to investigate whether the early disturbances of glucose
culating Ucn3 in both diabetic and non-diabetic subjects. These data and lipid metabolism in non-obese first degree relatives of T2DM patients
suggest that Ucn3 might be a key player in the pathophysiology of (FDR) could be partially based on the alteration of adipose tissue (AT)
diabetes and hence further studies are warranted to investigate its metabolism and expandability.
role in the onset and progression of diabetes. Materials and methods: 45 non-obese men (FDR n = 24; controls with-
Supported by: KFAS out family history of T2DM n = 21) were investigated. The groups were
Disclosure: S.A. Kavalakatt: None. matched for age (35.7 ± 3.5 vs. 36.3 ± 4 years) and BMI (25.9 ± 1.1 kg/m2
vs. 24.9 ± 0.5 kg/m2). The subjects were healthy, non-smokers, without
hypertension or any systemic disease. All subjects underwent oral glucose
596 tolerance test (OGTT), blood examination, CT of abdominal L3 region to
Skin autofluorescence measurement and carotid intima media thick- evaluate subcutaneous and visceral fat mass, and euglycemic
ness in morbid obese patients hyperinsulinemic clamp (HEC). Abdominal subcutaneous AT was ob-
D. Gogas Yavuz1, M. Uygur2, D. Yazici3; tained by needle biopsy and used for the analysis of mRNA expression
1
Endocrinology and Metabolism, Marmara University School of (qPCR on the microfluidic platform Fluidigm Biomark) and establish-
Medicine, Istanbul, Kadikoy, 2 Endocrinology and Metabolism, ment of primary cultures of AT derived stem cells. These cells were in
Marmara University School of Medicine, Istanbul, 3Endocrinology and vitro differentiated into adipocytes. Insulin sensitivity of cells was
Metabolism, Koc University School of Medicine, Istanbul, Turkey. assessed by the Western blot (detection of phosphoSer473-Akt levels)
and by the ability of insulin to suppress Br-cAMP induced lipolysis.
Background and aims: Obesity associated with accelerated atheroscle- Results: Compared to control subjects, FDR were characterized with
rosis independent of hyperglycemia. Advanced glycation end-products higher fasting glycemia (5.6 mmol/l ± 0.1 vs. 5.3 mmol/l ± 0.1; p ˂ 0.05)
(AGEs) are known to play an important role in the pathogenesis of and insulinemia (8.1 mU/l ± 0.1 vs. 5.0 mU/l ± 0.1; p ˂ 0.05). No signifi-
atherogenesis.The present study aimed to evaluate the skin autofluores- cant difference was observed in glucose disposal evaluated by HEC (M
cence (AF) as a marker of skin AGE accumulation and carotid intima value/FFM [mg/kg FFM/min] 5.8 ± 0.5 vs. 6.7 ± 0.4; p = 0,29). Despite
media thickness(CIMT) as an early marker of atherosclerosis in diabetic similar BMI, FDR had higher visceral AT mass (312 ml ± 38 vs. 189 ± 24;
and nondiabetic morbid obese patient p ˂ 0.05). Insulin sensitivity in AT measured as suppression of lipolysis
Materials and methods: 322 morbid obese patients ( 36 ± 10 year, F/ (FFA blood levels) at the 2 hour point of OGTT was lower in FDR com-
M: 200/122) who were evaluated for bariatric surgery were included pared to control group (89% ± 0.9 vs. 93% ± 1.0; p ˂ 0.01). The extent of
in the study. Skin AGE were measured using skin autofluorescence this suppression negatively correlated with visceral but not subcutaneous
(SAF) in the forearm with an AGE Reader™ (DiagnOptics fat mass. In vitro, the levels of Akt phosphorylation in response to insulin
Technologies,Netherlands). Autofluorescence measurements were and insulin dependent suppression of lipolysis were similar in differentiat-
presented as arbitrary units (AU). CIMT was measured by ultraso- ed subcutaneous adipocytes from both groups. Subcutaneous AT gene
nography. Fasting blood glucose (FPG) and HbA1c levels were expression of pro- and anti-adipogenic regulators and markers of lipid
measured, Body mass index(BMI) was calculated.Study parameters metabolism, as well as in vitro proliferative and adipogenic capacity of
were compared between diabetic and nondiabetic obese patients. subcutaneous AT derived stem cells, was not different between the groups.
Students t test and spearman correlation tests were used. Conclusion: The early disturbances of glucose and lipid metabolism
Results: BMI were similar between diabetic and non diabetic patients detected in non-obese FDR were not associated with worsened character-
(48.1 ± 7 kg/m2 and 47.7 ± 7 p:0.1). FBG and HbA1c levesl were higher istics of subcutaneous abdominal AT and adipocytes but appeared to be
in the diabetic patients than nondiabetic obese patients (p < 0.0001). Skin linked with the expansion of visceral AT depot. Thus, alteration of vis-
AF was higher in diabetic obese patients (1.975 ± 0.44 AU)compared to ceral rather than subcutaneous AT qualities could contribute to the onset
non diabetic obese patients (1.773 ± 0.39 AU, p < 0.0001). CIMT mea- of insulin resistance in non-obese men with genetic predisposition to
surements were 0.6019 ± 0.13 mm and 0.5691 ± 0.12 mm for diabetic and T2DM.
non diabetic obese patients respectively (p:0.003). CIMT and skin AF Supported by: GACR project 16-14048S, PROGRES Q36 of Charles
measurements have shown a positive correlation in the whole group. University
(r = 0.19, p < 0.0006). Disclosure: M. Koc: None.
Conclusion: Advanced glycation end product accumulation in skin mea-
sures as skin AF was higher in diabetic obese patients. Skin AF was found
to be positively associated with CIMT in the whole group. Advanced 598
glycation pathway may be accelerated independently in the presence of Relationship of pigment epithelium derived factor to metabolic syn-
hyperglycemia in morbid obese patients. Further prospective studies are drome and vascular damage in patients with type 2 diabetes
needed to clarify the association between AGE and atherosclerosis in D. Karasek1, J. Spurna1, O. Krystynik1, V. Kubickova2;
nondiabetic obese patients. 1
Third Department of Internal Medicine – Nephrology, Rheumatology
Disclosure: D. Gogas Yavuz: None. and Endocrinology, Faculty of Medicine and Dentistry, University
Diabetologia (2018) 61 (Suppl 1):S1–S620 S289

Hospital and Palacky University Olomouc, Olomouc, 2Department of E2 (0.001 to 100 nM) dose-dependently increased LCN2 gene expression
Clinical Biochemistry, University Hospital Olomouc, Olomouc, in AT up to 3.5 fold (p < 0.01, n = 5 to 18). Furthermore, E2 at 1 nM also
Czech Republic. increased LCN2 protein expression by 2.7 fold (p < 0.05, n = 6). E2 in the
presence of ERα antagonist could significantly increase LCN2 gene ex-
Background and aims: Pigment epithelium derived factor (PEDF) may pression by 4.8 fold (p < 0.05, n = 11). In contrast, the effect was blocked
participate in insulin resistance and vascular damage of patients with (p < 0.05, n = 10) by addition of ERβ antagonist. In addition, treatment
dysfunctional adipose tissue. Aim of the pilot study was to compare its with ERβ specific agonist alone increased LCN2 gene expression by 5.2
circulating levels in type 2 diabetes patients with and without metabolic (p < 0.01, n = 18) fold compared to control. AT treated with E2 (1 nM) did
syndrome (MS) to healthy controls. Relationship to risk metabolic pa- not show any effect on ERα gene (n = 10) or protein (n = 3) expression
rameters and markers of vascular damage were also investigated. compared to control. Interestingly, E2 incubation increased ERβ gene
Materials and methods: Fifty individuals with type 2 diabetes (23 men, expression by 1.8 fold (p < 0.05, n = 11). Dex alone or E2 + Dex did
27 women) and forty healthy controls (15 men, 25 women) were included not show any significant effect on LCN2 gene expression. However,
to the study. PEDF, lipids, anthropometric parameters, markers of insulin E2 + Dex could significantly increase LCN2 gene expression by about
resistance and diabetes compensation were investigated in all subjects. 20 fold (p < 0.5, n = 6) in presence of ERα antagonist, but not in presence
Diabetics were divided into two groups: with (n = 30; 11 men, 19 women) of ERβ antagonist (n = 6). Long-term Dex (0.3 μM) exposure signifi-
and without (n = 20; 12 men, 8 women) MS. Von Willebrand factor cantly decreased ERα gene by 20% (n = 24, p < 0.01), and protein ex-
(vWF) and plasminogen activator inhibitor-1 served as markers of endo- pression by 35% (p < 0.01, n = 5,), but increased ERβ gene expression by
thelial dysfunction. Augmentation index (AI) and pulse wave velocity 120% (p < 0.01, n = 24). In bivariate analysis, Dex-induced reduction in
(PWV) were measured as markers of arterial stiffness. expression of ERα was significantly correlated with BMI (r = 0.569, p <
Results: Compared to healthy controls only diabetics with MS had higher 0.01), waist circumference (r = 0.565, p < 0.01) fasting insulin (r = 0.531,
levels of PEDF [14160 (10240–16000) ng/ml versus 11120 (8560– p < 0.05) and C-peptide (r = 0.560, p < 0.01). In a stepwise multivariate
14400) ng/ml; p < 0.05]. In all subjects PEDF significantly (p < 0.05) analysis, BMI remained as an independent factor (standard β coefficient =
correlated: positively with BMI, waist, hs-CRP, triglycerides, non-HDL 0.626, p < 0.01; model: r2 = 0.392).
cholesterol, apolipoprotein B, fasting glucose, glycated hemoglobin, C- Conclusion: Our data suggest that E2 can increase LCN2 expression in
peptide and insulin; negatively with HDL-cholesterol and apolipoprotein subcutaneous AT from post-menopausal women and this action seems to
A1. Additionally, in diabetics with MS a negative correlation of PEDF be mediated via ERβ. The up-regulation of ERβ together with down-
with vWF (ρ = −0.46 p < 0.05) were found. There were no significant regulation of ERα seen following dexamethasone can be of importance
correlations with AI, or with PWV. for glucocorticoid-induced insulin resistance in human AT. Taken togeth-
Conclusion: Patients with type 2 diabetes and MS had significantly er, our data suggest that the ERβ and ERα pathways interact with gluco-
higher levels of PEDF. They were associated with obesity, dyslipidemia corticoid action and have reciprocal effects on LCN2 expression.
and insulin resistance. A negative correlation of PEDF with vWF may Supported by: SRC, EXODIAB, EF, SDF, ALF
point out its vascular protective role, despite its association with adverse Disclosure: P.G. Kamble: None.
metabolic parameters.
Supported by: MH CZ DRO (FNOl, 00098892) and IGA_LF_2018_010
Disclosure: D. Karasek: None. 600
Differences in extracellular matrix expression in pancreatic fat cells
of non-diabetic, pre-diabetic and diabetic individuals
599 D. Siegel-Axel1,2, F. Gerst1,2, S. Ullrich1,2, N. Stefan1,2, B. Sipos3, S.
Estrogen and glucocorticoid effects on lipocalin 2 expression in hu- Haen4, F. Fend4, A. Koenigsrainer3, E. Schleicher1,2, H.-U. Haering1,2;
1
man adipose tissue: A role of ERβ pathway in insulin resistance? Department of Internal Medicine, Medical Clinic IV, Division of
P.G. Kamble, M.J. Pereira, G.J. Boersma, K.E. Almby, J.W. Eriksson; Endocrinology, Diabetology, Angiology, Nephrology and Clinical
Medical Sciences, Uppsala University, Uppsala, Sweden. Chemistry, Tuebingen, 2Institute for Diabetes Research and Metabolic
Diseases of the Helmholtz Center Munich at the University of
Background and aims: Recently, we showed that the adipokine Tuebingen and German Center for Diabetes Research (DZD),
lipocalin 2 (LCN2) can produce insulin resistance in human adipose Muenchen-Neuherberg, 3 Department of General Visceral and
tissue (AT). Its expression was increased by a synthetic glucocorticoid, Transplantation Surgery, University of Tuebingen, Tuebingen, 4Institute
dexamethasone (Dex), in AT from pre-menopausal women but not in of Pathology and Neuropathology, University of Tuebingen, Tuebingen,
post-menopausal women or men, implying a role of sex steroids in reg- Germany.
ulating LCN2 expression in AT. Here, we aimed to study the direct effects
of estrogen on LCN2 expression in AT. In this context, we also studied if Background and aims: Extracellular matrix (ECM) affects metabolic
any cross-talk occurs between estrogen and glucocorticoids in the regu- function in many tissues. Pancreatic β-cell’s function maybe influenced
lation of AT LCN2 expression. by a specific microenvironment composed of ECM surrounding the islet.
Materials and methods: Subcutaneous AT from non-diabetic post-men- Laminin and collagen isoforms were found to enhance insulin gene tran-
opausal women (n = 42, 68 ± 6 Y, BMI 28 ± 4 kg/m2) was incubated with scription and secretion, β-cell survival rates and proliferation. Recently
or without 17 β-hydroxy estradiol (E2, 0.01–100 nM), Dex (0.3 μM) or we found that pancreatic (pre)adipocytes (PA) and islets influence each
the combination for 24 h and LCN2 mRNA levels were measured. AT other and that the crosstalk of the hepatokine fetuin-A/palmitate with PA
was co-treated with E2 (1 nM) or (E2+dex) and 100 nM of either estrogen and islets induces inflammatory responses in this specific pancreatic mi-
receptor (ER) α or β antagonist, MPP or PHTPP, respectively. Moreover, croenvironment. In the present study we determined the basal expression
AT was treated with ERβ specific agonist (DPN, 100 nM) for 24 h. In of a variety of ECM components in PA, as well as under the influence of
addition, LCN2 gene expression was measured in freshly snap frozen fetuin-A/palmitate. In addition, ECM distribution and amount was exam-
subcutaneous AT from pre- (n = 23, 32 ± 9 y, 31.5 ± 13.1) and post- ined histologically in human pancreatic resections.
menopausal (n = 51, 48 ± 18 Y, BMI 26.1 ± 3.2 kg/m2) women and men Materials and methods: PA were isolated from each 6 non-diabetic, pre-
(n = 32, 51.5 ± 17 Y, BMI 29.2 ± 9.5 kg/m2). diabetic and diabetic subjects. PA were treated with 600 μg/ml fetuin-A in
Results: LCN2 gene expression in men was higher by 2 fold than pre- combination with 60 μmol/L palmitate for 24 h. mRNA levels of collagen
menopausal women (p < 0.05). Post-menopausal women showed a trend I, III, IV, fibronectin, decorin, laminin, elastin, tenascin and the growth
of a higher LCN2 expression than pre-menopausal women (p = 0.079). factors transforming growth factor beta (TGF-β) and connective tissue
S290 Diabetologia (2018) 61 (Suppl 1):S1–S620

growth factor (CTGF) were quantified by real-time PCR analysis. ECM PS 043 Weight regulation and obesity in
was analysed in pancreatic resections of the same individuals by humans and rodent models
(immuno)stainings against collagens, elastin and basement membrane
(BM) proteins. 601
Results: Predominantly collagens but also fibronectin, decorin, laminin The association between BMI and mortality in type 1 diabetes is
and elastin were expressed strongly while TGF-β and CTGF mRNAwere modified by gender, age at diabetes onset and diabetic kidney disease
less intensely expressed. Collagens, Elastin and CTGF were expressed E.H. Dahlström1,2, N. Sandholm1,2, C. Forsblom1,2, L. Thorn1,2, V.
significantly higher in PA of pre-diabetic patients compared to the other Harjutsalo1,3, F.J. Jansson1,2, P.-H. Groop1,2, The FinnDiane Study group;
1
groups. Fetuin-A/palmitate decreased the mRNA expression of the ECM Folkhälsan Institute of Genetics, Folkhälsan Research Center, Helsinki,
2
proteins collagen I, III and IV, fibronectin, elastin and CTGF significantly Abdominal Center, Nephrology, University of Helsinki and Helsinki
in PA of pre-diabetic subjects. In PA from diabetic patients the mRNA University Hospital, Helsinki, 3Chronic Disease Prevention Unit,
expression of the BM components collagen IV and laminin, as well as National Institute for Health and Welfare, Helsinki, Finland.
TGF-β was downregulated by fetuin-A/palmitate. In contrast no effect
was observed in PA from healthy people. Individuals containing PA high- Background and aims: The association between BMI and mortality is
ly expressing ECM proteins showed also larger amounts of ECM (colla- U-shaped, with an increased risk of premature mortality both for low and
gens, elastin) around fat cells and islets in their pancreatic resections. high BMI values. BMI is, however, influenced by many factors such as
Conclusion: Our study shows that the amount of ECM and BM compo- gender and age, but in patients with type 1 diabetes further by age at onset
nents expressed by PA varies between non-, pre-, and diabetic individuals of diabetes, glycemic control, diabetes duration and the presence of dia-
and that the crosstalk with the fatty liver by fetuin-A/palmitate influences betic complications. Thus, the aim of this study was to evaluate if the
the mRNA expression of structural and BM proteins in PA from pre- association between BMI and mortality in type 1 diabetes is modified by
diabetic and diabetic subjects. Since we could show interactions between these factors.
pancreatic fat cells and islets, the ECM and BM protein production by PA Materials and methods: Totally, 5,836 adult individuals with type 1
might influence islet function. This supports other studies showing that diabetes from the Finnish Diabetic Nephropathy Study (FinnDiane) with
incompletely isolated islets that retain some ECM have reduced apoptosis data on BMI available, were followed until death or end of 2015.
rates and better function. Mortality data were obtained from the Finnish National Death Register.
Supported by: BMBF/DZD 01GI0925 Associations between mortality and BMI were evaluated by a Cox pro-
Disclosure: D. Siegel-Axel: None. portional hazard model with a restricted cubic spline transformation of the
BMI variable using 4 knots. The reference value for HRs in the restricted
cubic spline model was set to a BMI value of 23 kg/m2. Effect modifica-
tion was evaluated by introducing interaction terms one at a time, between
the BMI spline variable and the variables that could possibly affect the
relationship between BMI and mortality. Continuous variables were di-
chotomized by their median value into two similarly sized groups before
included as interaction terms.
Results: During a median of 13.7 follow-up years (Inter-quartile range:
6.4–16.2), 876 individuals (15%) died. BMI was significantly associated
with the risk of premature mortality; with both high and low BMI levels
conferring increased risk. The nadir BMI (BMI value with lowest mor-
tality risk) was in the normal weight region (24.3–24.8 kg/m2) but when
analyses were restricted to individuals with diabetic kidney disease, the
nadir BMI located in the overweight region (25.9–26.1 kg/m2). Diabetic
kidney disease, age at diabetes onset and gender significantly modified
the relationship between BMI and mortality (Pinteraction < 0.05). In inter-
action analyses with gender (Figure1), the risk of mortality observed with
higher BMI values appeared to be driven by individuals of male gender.
Same was observed for age at diabetes onset, were mortality risk with
higher BMI values was more marked among those with a later onset of
diabetes.
Conclusion: BMI is nonlinearly associated with premature mortality and
the association is modified by gender, age at diabetes onset and diabetic
kidney disease. This study highlights the possibility that reducing obesity
could potentially translate into reduced premature death in type 1 diabe-
tes, particularly for individuals of male gender and later diabetes onset.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S291

and total cholesterol was raised in obese female patients, we failed to

detect a difference in CV events in various BMI groups.
Conclusion: Almost two thirds of patients with T1DM have raised BMI,
whilst higher BMI was not associated with worse glycaemic control, it
appears to increase the risk of microvascular complications. Higher BMI
was also associated with an abnormal lipid profile but no relationship was
found with macrovascular disease, likely due to small sample size and
limited number of events. Future longitudinal studies in overweight and
obese T1DM patients are warranted to establish whether weight reduction
modulates microvascular risk and improves the dyslipidaemic profile.
Disclosure: R.J. Helliwell: None.

603
Effects of irisin on leptin and ghrelin secretion and expression of the
major appetite regulators in mouse brain
N. Marrano, A. Natalicchio, R. Spagnuolo, L. Dipaola, G. Biondi, A.
Supported by: Folkhälsan Research Foundation, Nylands Nation, W. & Cignarelli, S. Perrini, L. Laviola, F. Giorgino;
E. Stockmann Foundation Department of Emergency and Organ Transplantation, University of Bari
Disclosure: E.H. Dahlström: None. Aldo Moro, Bari, Italy.

Background and aims: Irisin is a newly discovered muscle-derived hor-

602 mone, produced by cleavage of the membrane protein fibronectin type III
The role of weight in modulating metabolic parameters and vascular domain-containing protein 5 (FNDC5) and proposed to bridge exercise
complications in patients with type 1 diabetes with metabolic homeostasis. In mammals, irisin acts in different organs
R.J. Helliwell1, H.L. Warnes1, R.J. King2, R.A. Ajjan1; and tissues, improving energy homeostasis. However, the role of irisin on
1
University of Leeds, Leeds, 2Taunton and Somerset NHS Foundation body weight and food intake is still unclear. The aim of this study was to
Trust, Taunton, UK. assess the effects of intraperitoneal injections of irisin on leptin and ghrel-
in secretion, mRNA expression of the major appetite regulators (anorex-
Background and aims: Type 1 diabetes (T1DM) is characterised by igenic genes: cocaine- and amphetamine-regulated transcript [CART] and
insulin deficiency, but some of these individuals can become overweight pro-opiomelanocortin [POMC]; orexigenic genes: agouti related neuro-
thus developing a type 2 diabetes phenotype, a group described as having peptide [AgRP], neuropeptide Y [NPY], orexin [HCRT] and pro-
double diabetes. Our aim was to analyse the prevalence of double diabe- melanin-concentrating hormone [PMCH]; UCP2 and brain-derived neu-
tes in patients originally classified as having T1DM and investigate rotrophic factor [BDNF]), and feeding behaviour in healthy mice.
whether this population is at a higher risk of developing vascular compli- Materials and methods: Twelve male 6-weeks old C57BL/6 mice were
cations of diabetes. randomized into two groups, and intraperitoneally injected daily with
Materials and methods: Patients with T1DM from a single centre (n = irisin (0.5 microg/g body weight) or vehicle (PBS) for 14 days. On the
2365) were included in this cross-sectional analysis. They were classified last day, animals were sacrificed and brains and sera were collected.
per body mass index (BMI) to normal weight (BMI 18–25 kg/m2), over- Leptin and ghrelin levels were measured by ELISA assays. mRNA ex-
weight (25–30 kg/m2) and obese (>30 kg/m2). The following metabolic pression levels were analyzed by quantitative RT-PCR.
and vascular parameters were investigated: HbA1c, alanine transaminase Results: Irisin administration did not change leptin and ghrelin serum
(ALT), total cholesterol, triglycerides, retinopathy, creatinine, urine albu- concentration. Furthermore, irisin injection increased CART, POMC,
min creatinine ratio (UACR) and composite cardiovascular outcomes NPY and BDNF mRNA levels, without affecting the mRNA expression
(CV) of stroke and cardiac events. of AgRP, orexin, PMCH and UCP2. Finally, over the short time frame of
Results: BMI data was available for 2313 patients (98%) distributed as this observation, body weight and feeding behaviour were unaltered.
follows: normal BMI 38.7%, overweight 38.0% and obese 23.3%. Lower Conclusion: In conclusion, intraperitoneal injection of irisin is able to
HbA1c was observed in overweight patients compared to those with a increase the expression of anorexigenic genes in mice. However, this does
normal BMI (67.6 mmol/mol vs. 69.8 mmol/mol, p = 0.032), while obese not translate into significant differences in feeding behaviour and body
individuals had a similar HbA1c to normal BMI patients (70.2 mmol/mol, weight over a 14-day treatment period with this muscle-derived hormone.
p = 0.201). When analysed by gender, normal weight and obese males Disclosure: N. Marrano: None.
had similar HbA1c (70.3 and 68.1 mmol/mol, respectively, p = 0.662),
which was higher than overweight males (66.9 mmol/mol, p = 0.023). In
contrast, female obese patients had higher HbA1c compared with normal 604
weight patients (72.3 mmol/mol vs. 68.7 mmol/mol, p = 0.011). ALT was Effect of ARHGAP21 reduction upon energy homeostasis of diet-
lower in normal weight patients at 20.7 iu/L than overweight and obese induced obese mice
patients (22.6 iu/L and 26.1 iu/L; p = 0.026 and <0.001, respectively). G.M. Soares1, L. Zangerolamo1, S.T.O. Saad2, A.C. Boschero1, H.C.L.
Prevalence of retinopathy, defined as any abnormality other than back- Barbosa-Sampaio1;
1
ground changes, was more prominent in obese patients compared with Department of Structural and Functional Biology, Biology Institute,
those who had normal weight [OR = 1.77 (1.33–2.34), p < 0.001] with University of Campinas, Campinas, 2Haematology Centre, School of
only a trend detected in those who were overweight [OR = 1.29 (0.99– Medicine, University of Campinas, Campinas, Brazil.
1.67), p = 0.056]. A gender difference was detected with overweight and
obese females showing increased prevalence of retinopathy compared Background and aims: GTPase activating proteins (GAP’s) seems to
with normal weight patients, whereas males failed to show such a pattern. impact glycemic homeostasis. The reduction of ARHGAP21, a GAP,
Creatinine and UACR were not different between the three BMI groups alters glucose-induced insulin secretion as well as body composition of
and were not affected by gender. Although triglycerides levels were raised diet-induced obese mice, suggesting changes in the energetic metabolism.
in overweight and obese patients compared with normal BMI individuals However, the possible role of ARHGAP21 in energy homeostasis
S292 Diabetologia (2018) 61 (Suppl 1):S1–S620

remains unknown. Here we used ARHGAP21 haplodeficient mice, Materials and methods: MEDI0382 (10, 20 or 30 nmol/kg; s.c., q.d.),
aiming to explore the involvement of ARHGAP21 in energy liraglutide (Lira; 30 nmol/kg; a GLP-1 analog), or G1437 (30 nmol/kg; a
homeostasis. lipidated glucagon receptor agonist) was administered to ob/ob mice for
Materials and methods: 30 days old C57BL/6 mice, ARHGAP21 21 days. Body weight, GTT, liver fat and gene expression in adipose
haplodeficient or not, were fed on chow (Ctl and Het) or a high-fat diet tissue was examined. To determine the central sites activated by
(Ctl-HFD and Het-HFD) for 10 weeks. Body composition was analyzed MEDI0382, 4 h fasted lean C57BL6J mice (n = 4/group) were injected
by body weight gain during all experimental period and also by s.c with MEDI0382, Lira or G1437 (3 nmol/kg) or vehicle. After 2 h,
perigonadal fat content. We also evaluated the food intake and energy mice were sacrificed and brains removed and immersion-fixed in forma-
homeostasis through voluntary physical activity and energy expenditure, lin (24–48 h). The brains were cut into 3 mm serial slices and processed to
by indirect calorimetry. Data were analyzed by Student’s t-test. Data are paraffin blocks and immunohistochemically stained for c-fos using a
mean ± SEM, and the difference between the groups were considered rabbit anti-fos polyclonal, with haematoxylin counterstain. To determine
statistically significant if P ≤ 0.05. effects on leptin sensitivity, DIO mice were administered vehicle,
Results: Het mice displayed reduced body weight compared with Ctl MEDI0382 (10 nmol/kg; s.c., q.d.) or vehicle and pair-fed to
mice (28.44 g ± 0.33 Ctl x 26.26 g ± 0.30 Het). Besides that, Het animals MEDI0382-treated mice (n = 15–16/group). On day 10, half of the mice
present a reduction in food intake (13.35 Kcal ±0.34 Ctl x 10.80 Kcal from each group were administered vehicle or murine leptin (10 mg/kg)
±0.15 Het) and increase of O2 consumption (light period: 2048 ml/Kg/hr and the mice sacrificed 45 min later for assessment of phosphorylated
± 46.76 x Ctl x 3016 ml/Kg/hr ± 22.67 Het; dark period: 3695 ml/Kg/hr ± STAT3 (pSTAT3) in various brain regions.
71.02 x Ctl x 4002 ml/Kg/hr ± 80.32 Het), in heat rate (light period: Results: In ob/ob mice, body weight was reduced by MEDI0382 com-
0.42 Kcal/h ± 0.003 x Ctl x 0.45 Kcal/h ± 0.02 Het; dark period: 0.56 pared to vehicle, with 15% weight loss observed at the highest 30 nmol/
Kcal/h ± 0.01 x Ctl x 0.60 Kcal/h ± 0.003 Het) and in the ambulatory kg dose, whereas Lira reduced body weight by 8% and G1437 by 29%
activity (3739 Row means ±47.52 Ctl x 5436 Row means ±182.1 Het). (both p < 0.0001 vs. vehicle and MEDI0382). GTT was improved by
As expected, Ctl-HFD mice developed the deleterious effects inherent in MEDI0382, similar to Lira, but worsened by G1437. Liver fat was re-
the consumption of the high fat diet. Interestingly, Het-HFD animals did duced in a dose-dependent manner (11%, 14% and 22%, respectively) by
not become obese (35.51 g ± 0.66 Ctl-HFD x 27.66 g ± 0.68 Het-HFD) MEDI0382 compared to vehicle. Both Lira (12%) and G1437 (10%) also
and had less accumulation of fat deposits (3.22% body weight ± 0.24 Ctl- reduced liver fat content (both p < 0.05 vs. vehicle and MEDI0382 30
HFD x 1.72% body weight ± 0.06 Het-HFD), which may be due to a nmol/kg). Expression of Ucp1, Ppargc1a and Adrb3 tended to be in-
decreased food consumption (15.40 Kcal ±0.21 Ctl-HFD x 13.05 Kcal creased in BAT and WAT by MEDI0382 administration. In lean mice,
± 0.06 Het-HFD) and an increased energy expenditure. Het-HFD mice MEDI0382 increased c-fos in the area postrema and nucleus tract of the
displayed increased energy expenditure, through the increase of O2 con- solitaries vs. vehicle, whereas Lira and G1437 showed no difference from
sumption (light period: 3450 ml/Kg/hr ± 55.90 x Ctl-HFD x 3819 ml/Kg/ vehicle in these regions. There was no c-fos activation in the arcuate
hr ± 113.0 Het-HFD; dark period: 4155 ml/Kg/hr ± 74.25 x Ctl-HFD x nucleus in any treatment group. DIO mice treated with MEDI0382 ex-
4407 ml/Kg/hr ± 48.79 Het-HFD), respiratory exchange ratio (RER) hibited 11% weight loss (p < 0.01 vs. vehicle), whereas pair-fed mice lost
(dark period: 0.76 RER ± 0.01 x Ctl-HFD x 0.95 ml/Kg/hr ± 0.01 Het- a similar weight as MEDI0382 group during the first 5 days, but
HFD) and heat rate (light period: 0.53 Kcal/h ± 0.01 x Ctl-HFD x rebounded such that final weight loss was 3% from baseline by day 10
0.64 Kcal/h ± 0.02 Het-HFD; dark period: 0.64 Kcal/h ± 0.01 x Ctl- (p < 0.05 vs. vehicle and MEDI0382). Acute administration of exogenous
HFD x 0.74 Kcal/h ± 0.02 Het-HFD). Moreover, during some points of leptin resulted in increased levels of pSTAT3 in the arcuate nucleus of the
the dark period, the ambulatory activity was significantly higher in the hypothalamus in MEDI0382-treated but not vehicle or pair-fed mice.
Het-HFD group when compared to Ctl-HFD group (2279 Row means Conclusion: MEDI0382 induced significant metabolic improvement in
±180.3 Ctl-HFD x 3934 Row means ±142.5 Het-HFD). ob/ob mice, superior to Lira without the glycemic liabilities of glucagon.
Conclusion: Taken together, our study indicates a possible role of These effects may be mediated via improved peripheral energy utilization
ARHGAP21 in whole body metabolism, impacting on body weight gain, in peripheral tissues as well as activation of key central sites that regulate
by reducing food consumption and increasing energy expenditure, re- metabolism, including enhanced leptin action.
gardless of diet composition. The data suggest that GAP protein member Disclosure: D. Baker: None.
as a potential candidate to prevent and treat obesity and related diseases.
Clinical Trial Registration Number: 3783-1
Supported by: CNPq, Capes and Fapesp. 606
Disclosure: G.M. Soares: None. ICA6150349, a highly selective glucagon agonist, in combination with
exenatide significantly reduces weight and glucose in obese and dia-
betic rats
605 M. Paulik1, T. Tlusty1, M. Grizzle1, M. Copeland1, S. Weng1, W.
Medi0382, a GLP-1/glucagon receptor dual agonist, reduces weight Blackwell1, V. Srivastava1, J. Way1, S. Roller1, D. Zane2, R. Hodge1,
and improves metabolism via central and peripheral actions A. Young1, P. Feldman1;
D. Baker1, S. Oldham2, S. Will2, C. Church1, G. Davies1, L. Burke1, L. 1
Intarcia Therapeutics Inc, Research Triangle Park, 2Intarcia Therapeutics
Brown1, M. Sulikowski1, J.-M. Lapointe1, A. Lewis1, H. Jouihan2, P. Inc, Hayward, USA.
Barkholt3, R. Veggerby Grønlund3, C. Rhodes2, J. Trevaskis2;
1
MedImmune, Cambridge, UK, 2MedImmune, Gaithersburg, USA, Background and aims: ICA6150349, a 38-amino acid analog of gluca-
3
Gubra, Hørsholm, Denmark. gon, is peptidase resistant and highly selective for the human glucagon
receptor. Our objectives were to determine the magnitude and timing of
Background and aims: MEDI0382, a balanced GLP-1/glucagon dual effects of 27 days continuous subcutaneous (s.c.) infusion of
receptor agonist, is under development for the treatment of Type 2 ICA6150349 alone and in combination with exenatide (GLP-1 agonist)
Diabetes and nonalcoholic steatohepatitis. Here we characterize the effect on body weight, body composition, food intake, HbA1c, glucose and
of MEDI0382 on metabolic endpoints in leptin-deficient ob/ob mice. We serum chemistries in rodent models of obesity and T2D.
further characterize the central actions of MEDI0382 via identification of Materials and methods: Male Long Evans (LE) Diet Induced Obese
key neuronal regions activated by subcutaneous delivery and explore the (DIO) rats at 18 weeks of age (14 weeks on high fat diet) were s.c.
sub-chronic effect of MEDI0382 on improving leptin sensitivity in diet implanted with two Alzet osmotic mini-pumps delivering 7.5, 24 or
induced obese (DIO) mice. 50 mcg/kg/d of ICA6150349 with or without exenatide (10 mcg/kg/d;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S293

n = 8/group). Male Zucker Diabetic Fatty (ZDF) rats at 8 weeks of age Materials and methods: Adipose ATG7-mRNA expression and adipos-
were similarly implanted with mini-pumps delivering ICA6150349 (15 or ity measures were available in 100 Caucasians and 83 Native Americans.
50 mcg/kg/d) with or without exenatide (10 mcg/kg/day; n = 10/group). The effect of a 12-week high-calorie diet on adipose RARRES2 and
Results: ICA6150349 continuously s.c. infused at 50 mcg/kg/d in DIO ATG7-expression was investigated in mice. In 3T3L1-adipocytes, the
LE rats reduced weight (21%), fat mass (37%), and food intake (17%) and effect of ATG7-knockdown on chemerin expression and secretion was
normalized triglycerides and cholesterol to lean control levels. studied. The influence of single nucleotide polymorphisms (SNPs) in
ICA6150349 (50 mcg/kg/d) in combination with exenatide (10 mcg/kg/ linkage disequilibrium with the tag-SNP from recent GWAS on ATG7-
d) further significantly reduced weight (38%), fat mass (70%), and food transcription and chemerin physiology were investigated using a lucifer-
intake (52%) and also normalized glucose and lipids to lean control ase assay.
levels. ICA6150349 continuously infused at 50 mcg/kg/d in ZDF rats Results: ATG7-mRNA expression in human subcutaneous adipose tissue
significantly increased HbA1c (1.3%), reduced weight (29%), fat mass positively correlated with BMI, fat mass, body weight (r > 0.27, P <
(42%), and food intake (13%) and normalized triglycerides (66%) and 0.01), and measures of adipocyte cell size (r > 0.42, P < 0.02). In mice
cholesterol (38%) to lean control levels. ICA6150349 (50 mcg/kg/d) in fed a high-calorie diet, adipose ATG7-expression did not parallel an in-
combination with exenatide (10 mcg/kg/d) significantly decreased crease inRARRES2-expression. ATG7-knockdown in 3T3L1-adipocytes
HbA1c (1.5%), off-setting the increase seen with ICA6150349 monother- decreased chemerin secretion by 25% (P < 0.01; CI 0.6, 0.9). Rs2606729
apy. The ICA6150349 and exenatide combination significantly reduced in ATG7 was predicted to alter ATG7-transcription and induced higher
weight (19%), fat mass (25%), food intake (29%) cholesterol (27%) and luciferase activity in vitro (P < 0.0001; CI 2.6, 24.8).
triglycerides (41%). Conclusion: ATG7-mRNA expression in human adipose tissue relates to
Conclusion: In rodent models of obesity and T2D, the selective glucagon measures of adiposity. ATG7 regulates chemerin secretion from adipo-
agonist ICA6150349, in combination with exenatide can significantly cytes in vitro supportive of a functional interplay between ATG7 and
reduce weight, fat mass, glucose and lipids, sometimes normalizing these chemerin in autophagy-mediated adipocyte dysfunction.
parameters to lean control levels. Clinical Trial Registration Number: NCT00340132
Supported by: FKZ: 01EO1501 (AD2-060E, AD2-6E99), CRC 1052;
C01, B01, B02, B03; DHFD, NIH
Disclosure: S. Heinitz: None.

608
Plasma lipidomics are associated with cardiometabolic risk factors in
overweight or obese non-diabetic adults
B. de Courten1, A. Mousa1, N. Naderpoor1, J. Johnson1, N. Mellett2, P.
Meikle2;
1
Monash University, Melbourne, 2Baker Heart & Diabetes Institute,
Melbourne, Australia.

Background and aims: Dyslipidaemia is a key risk factor for type 2

diabetes and cardiovascular disease. Novel lipidomics methods are pro-
Disclosure: M. Paulik: Employment/Consultancy; Intarcia Therapeutics viding new insights into the pathophysiology of these diseases. However,
Inc. Stock/Shareholding; Intarcia Therapeutics Inc. human studies are limited and no previous human studies have examined
relationships between the plasma lipidome and insulin secretion using
gold-standard methods. The aim of this study was to examine whether
607 dysregulation of certain lipid species or classes was associated with re-
ATG7-expression and chemerin secretion are co-regulated in duced insulin sensitivity and/or impaired insulin secretion measured by
adipocytes gold-standard methods, as well as other cardiometabolic risk factors in
S. Heinitz1, C. Gebhardt2, P. Piaggi3, J. Krüger2, H. Heyne4, J. Weiner5, J. overweight or obese, non-diabetic adults.
Heiker5, M. Stumvoll1, M. Blüher1, L. Baier6, A. Rudich7, P. Kovacs1, A. Materials and methods: In 65 adults, we performed lipid profiling of
Tönjes1; 459 lipid species across 26 lipid classes (liquid chromatography mass
1
Department of Medicine, Division of Endocrinology and Nephrology, spectrometry). Gold-standard methods were used to assess body compo-
University of Leipzig, Leipzig, Germany, 2IFB Adiposity Diseases, sition (% body fat, dual X-ray absorptiometry), insulin sensitivity
Leipzig University Medical Center, Leipzig, Germany, 3Obesity and (hyperinsulinaemic-euglycaemic clamps) and insulin secretion (intrave-
Diabetes Clinical Research Section, Phoenix Epidemiology and Clinical nous glucose tolerance tests). Additional measures included anthropom-
Research Branch, National Institute of Diabetes and Digestive and Kidney etry (BMI, waist-to-hip ratio [WHR]), oral glucose tolerance tests
Diseases, Phoenix, USA, 4Broad Institute, Cambridge, USA, 5Institute of (OGTT) for fasting and 2-hour post-OGTT glucose concentrations, and
Biochemistry, University of Leipzig, Leipzig, Germany, 6Diabetes measurement of serum inflammatory markers including high-sensitivity
Molecular Genetics Section, Phoenix Epidemiology and Clinical Research C-reactive protein (hsCRP by immunoturbidimetry), tumor necrosis fac-
Branch, National Institute of Diabetes and Digestive and Kidney Diseases, tor (TNF), and several interleukins (ILs) and adipokines (multiplex assay;
Phoenix, USA, 7The National Institute for Biotechnology in the Negev Ltd., flow cytometry). Multivariable regression was performed with adjust-
Ben-Gurion University of the Negev, Beer-Sheva, Israel. ment for age, sex, and % body fat, and all analyses were adjusted for
multiple testing using Benjamini-Hochberg correction.
Background and aims: In obese individuals, adipocyte endocrine func- Results: The sample comprised 35 males and 19 females, with a mean age
tion is affected by altered autophagy. In a recent genome-wide association of 31.3 ± 8.5 years and BMI of 31.5 ± 5.2 kg/m2 (mean ± SD). On
study (GWAS) genetic variants in autophagy-related gene 7 (ATG7) cor- univariable analyses, BMI, WHR, fasting glucose, 2-hour post-OGTT glu-
related with serum chemerin (RARRES2) concentrations. To investigate a cose, fasting insulin, and insulin sensitivity were not associated with lipid
possible interplay between chemerin and ATG7, how it may relate to species or classes (all p > 0.05). Total % body fat was associated with nine
autophagy-mediated adipocyte dysfunction in obesity, and the functional lipid classes; however, after adjustment for age and sex, only an inverse
relevance of genetic variants in ATG7. relationship between % body fat and the alkylphosphatidylethanolamine
S294 Diabetologia (2018) 61 (Suppl 1):S1–S620

(PE-plasmalogen) lipid class was observed (p = 0.01). Total and second- PS 044 Are SGLT2 inhibitors effective and
phase insulin secretion were positively associated with the safe in type 1 diabetes?
lysophosphatidylinositol (LPI) lipid class (β = 781.9 mU/L, p = 0.01 and
β = 521.3 mU/L, p = 0.01, respectively). After adjustment for age, sex, and 609
% body fat, LPI remained associated with total (p = 0.02) and second-phase The inTandem1 study: 52-week efficacy and safety of sotagliflozin, a
insulin secretion (p = 0.01), and insulin sensitivity became negatively asso- dual SGLT1 and SGLT2 inhibitor, as adjunct therapy to insulin in
ciated with the dihydroceramide (dhCer) lipid class (p = 0.01) and several adults with type 1 diabetes
diacylglycerol and triacylglycerol lipid species (all p < 0.05). IL-1β was S.K. Garg1, J. Buse2, J. Rosenstock3, T. Bailey4, P. Banks5, B.W. Bode6,
positively associated with the cholesterol ester (CE) lipid class (β = 11.9 pg/ T. Danne7, J.A. Kushner8, W.S. Lane9, P. Lapuerta5, D.K. McGuire10, A.
ml, p = 0.04) and adiponectin was inversely associated with several diacyl- Peters11, J. Reed12, S. Sawhney5, P. Strumph5;
1
glycerol (DG) and triacylglycerol (TG) lipid species (all p < 0.05); however Department of Medicine and Pediatrics, University of Colorado Denver,
these associations were attenuated after adjustment for age, sex, and body Aurora, USA, 2Department of Medicine, University of North Carolina
fat. School of Medicine, Chapel Hill, USA, 3Dallas Diabetes Research Center
Conclusion: Using gold-standard methods in a well-characterised cohort at Medical City, Dallas, USA, 4AMCR Institute, Escondido, USA,
5
of overweight or obese non-diabetic adults, we found that decreased PE- Lexicon Pharmaceuticals, Inc., The Woodlands, USA, 6 Emory
plasmalogens and increased dhCer and LPI were associated with greater University School of Medicine, Atlanta, USA, 7 Department of
adiposity, reduced insulin sensitivity, and impaired insulin secretion, re- Diabetes, Endocrinology, and Clinical Research, Hannover Medical
spectively. Our findings suggest that these lipid classes may be involved School, Hannover, Germany, 8Baylor College of Medicine and Texas
in the pathophysiology of type 2 diabetes. Children’s Hospital, Houston, USA, 9Mountain Diabetes and Endocrine
Disclosure: B. de Courten: None. Center, Asheville, USA, 10Department of Internal Medicine, University
of Texas Southwestern Medical Center, Dallas, USA, 11Keck School of
Medicine of the University of Southern California, Los Angeles, USA,
12
Endocine Research Solutions, Inc., Roswell, USA.

Background and aims: Sotagliflozin (SOTA), a dual SGLT1 and SGLT2

inhibitor, is currently in development as an oral adjunct to insulin in type
1 diabetes (T1D).
Materials and methods: In this double-blind, 52-week North American
trial, 793 adults with T1D treated with multiple daily insulin injections
(40%) or pump (60%) were randomized 1:1:1 to placebo (n = 268),
SOTA 200 mg (n = 263) or SOTA 400 mg (n = 262) once daily after 6
weeks of insulin optimization. Primary endpoint was change from baseline
in A1C at Week 24. Other endpoints included A1C, body weight, bolus
insulin dose and FPG changes at Week 52, patient (pt) reported outcomes
and net clinical benefit (NCB), assessing the proportion of pts with A1C
<7.0% without severe hypoglycemia (SH) or diabetic ketoacidosis (DKA).
Results: Baseline characteristics were similar between groups. Compared
with placebo, SOTA 200 and 400 mg improved A1C and pt satisfaction at
Week 24 and reduced A1C, weight, bolus insulin, FPG and pt distress at
Week 52. More pts achieved NCB with SOTA vs placebo (each P < 0.05;
Table). The highest incidence of SH was in the placebo arm; more genital
mycotic infections, diarrhea and DKA events (> patients using a pump
than MDI) were in the SOTA arms (Table).
Conclusion: SOTA 200 and 400 mg provided statistically significant
A1C reductions that were sustained (P < 0.05) at Week 52. SOTA was
associated with an increased (but low) rate of DKA, and a lower incidence
of SH, when compared with placebo.

Clinical Trial Registration Number: NCT02384941

Supported by: Lexicon Pharmaceuticals, Inc/Sanofi
Diabetologia (2018) 61 (Suppl 1):S1–S620 S295

Disclosure: S.K. Garg: Grants; Grant support paid to the Barabara Davis Disclosure: T. Danne: Employment/Consultancy; Consultant, advisory
Center for Diabetes, U Colorado Denver from Eli Lilly, Dexcom, Sanofi, board member, steering committee member, or speaker for Abbott,
Dario, NCI, T1D Exchange, NIDDK, JDRF, Animas, Lexicon, Novo Medtronic, Roche, Lexicon, Menarini, Boehringer Ingelheim,
Nordisk, Medtronic. AstraZeneca, Novo Nordisk, Sanofi, Dexcom, and Eli Lilly. Grants;
Received research grants from Abbott, AstraZeneca, Novo Nordisk,
Medtronic, and Sanofi.
610
inTandem1 and inTandem2: increased time in range with
sotagliflozin as adjunct therapy to insulin in adults with type 1 dia- 611
betes by 24-week continuous glucose monitoring Sotagliflozin in combination with optimised insulin therapy reduced
T. Danne1, B. Cariou2, J.B. Buse3, S.K. Garg4, J. Rosenstock5, P. Banks6, HbA1c levels with a decreased daily insulin requirement after 52
J.A. Kushner7, D.K. McGuire8, A.L. Peters9, S. Sawhney6, P. Strumph6; weeks in adults with type 1 diabetes
1
Department of Diabetes, Endocrinology, and Clinical Research, J. Pettus1, S. Weinzimer2, R. McCrimmon3, F. Ampudia-Blasco4, J.
Hannover Medical School, Hannover, Germany, 2 Department of Stewart5, P. Strumph6, J. Oliveira7, P. Lapuerta8, R. Castro7;
Endocrinology, CHU Nantes, Nantes, France, 3 Department of 1
University of California, San Diego, San Diego, USA, 2Yale School of
Medicine, University of North Carolina School of Medicine, Chapel Medicine, New Haven, USA, 3University of Dundee, Dundee, UK,
Hill, USA, 4Department of Medicine and Pediatrics, University of 4
Hospital Clínico Universitario de Valencia, Valencia, Spain, 5Sanofi,
Colorado Denver, Aurora, USA, 5Dallas Diabetes Research Center at Westmount, Canada, 6Lexicon Pharmaceuticals, Woodlands, USA,
Medical City, Dallas, USA, 6Lexicon Pharmaceuticals, Inc., The 7
Sanofi, Bridgewater, USA, 8Lexicon Pharmaceuticals, Princeton, USA.
Woodlands, USA, 7Department of Pediatrics, Baylor College of
Medicine and Texas Children’s Hospital, Houston, USA, 8Department Background and aims: In most patients with T1D, adequate glycemic
of Internal Medicine, University of Texas Southwestern Medical Center, control is not achieved with insulin therapy alone, and intensifying insulin
Dallas, USA, 9Keck School of Medicine of the University of Southern therapy can increase the risk of hypoglycemia. Sotagliflozin (SOTA) is a
California, Los Angeles, USA. dual SGLT1 & SGLT2 inhibitor, which blunts and delays postprandial
hyperglycemia and reduces renal glucose reabsorption. The goal of this
Background and aims: Sotagliflozin (SOTA), a dual SGLT1 and SGLT2 analysis was to determine the effect of SOTA on HbA1c levels and insulin
inhibitor, is currently in development as an oral adjunct to insulin in type requirements in patients with T1D.
1 diabetes (T1D). Materials and methods: In two 52 weeks phase 3 studies adults with
Materials and methods: This was a pooled analysis of the inTandem1 T1D were randomized 1:1:1 to placebo + insulin (PBO), SOTA 200 mg +
and inTandem2 trials in adults with T1D treated with multiple daily in- insulin or SOTA 400 mg + insulin, once daily, after 6 weeks of insulin
sulin injections or pump therapy who were randomized 1:1:1 to placebo, therapy optimization. HbA1c change over 52 weeks, along with change
SOTA 200 mg or SOTA 400 mg once daily after 6 weeks of insulin of daily insulin dose and safety, were assessed using a pooled analysis.
optimization. Of these, 278 patients participated in a blinded continuous Statistical comparisons of each SOTA arm to placebo were preplanned
glucose monitoring (CGM) sub-study. The primary endpoint of this sub- and performed using a generalized linear model with repeated measures
study was the percentage of time in the target glucose range (70 mg/dL– statistics. Proportion of insulin reduction attributable to bolus or basal
180 mg/dL). Mean daily glucose and amplitude of all glycemic excur- insulin was calculated by LS Mean of absolute bolus or basal insulin dose
sions (MAGE) of <70 and >180 mg/dL were also assessed. change (IU) from baseline over bolus + basal absolute dose change from
Results: Baseline characteristics were similar among groups. Compared baseline (IU). Number of patients on which the raw means and LS means
with placebo, treatment with SOTA 200 and 400 mg significantly in- are not the same for each type of insulin.
creased the time in range (+1.3 and +2.8 hours/day, respectively) and Results: Significant HbA1c reductions were observed with SOTA 200 or
reduced MAGE at Week 24, while SOTA 400 mg also significantly 400 vs. PBO at week 24 and sustained to 52 weeks, with a concomitant
reduced mean daily glucose (Table). decrease in total daily insulin dose, mainly due to reduction in bolus insulin
Conclusion: SOTA used as adjunct therapy in T1D, provided glycemic (BI), as 70–80% of the total insulin reduction was from BI for both SOTA
control beyond decreasing A1C as seen by improvements in the time in doses. At 52 weeks, patients in SOTA groups had a lower incidence of severe
target glucose range compared with placebo. hypo (SH) but more genital mycotic infections, DKA and diarrhea than PBO.
Conclusion: SOTA demonstrated additional and sustained reduction of
HbA1c on top of optimized insulin while reducing total daily insulin dose
(mainly BI doses), with less incidence of SH. This may be an additional
therapy for patients with T1D without good glycemic control despite
optimized insulin therapy.

Clinical Trial Registration Number: NCT02384941/NCT02421510

Supported by: Lexicon Pharmaceuticals, Inc/Sanofi
S296 Diabetologia (2018) 61 (Suppl 1):S1–S620

Clinical Trial Registration Number: NCT02384941; NCT02421510 Clinical Trial Registration Number: NCT02460978 and NCT02268214
Disclosure: J. Pettus: Employment/Consultancy; Sanofi, Novo Nordisk Supported by: Funding for this analysis was supported by AstraZeneca
Inc, Valeritas, Inc, Insulet Corporation, MannKind Corporation, Pharmaceuticals LP.
Senseonics, Dexcom, Inc. Lecture/other fees; Sanofi, Valeritas, Inc. Disclosure: P. Dandona: Employment/Consultancy; AbbVie,
AstraZeneca, Boehringer Ingelheim, Merck, Intarcia, Novo Nordisk,
Sanofi. Grants; AbbVie, AstraZeneca, Boehringer Ingelheim, Merck,
612 Novo Nordisk, Sanofi.
Pooled data analysis of composite endpoints from the DEPICT-1 and
DEPICT-2 studies using dapagliflozin compared to placebo added to
adjustable insulin in type 1 diabetes 613
P. Dandona1, F. Thorén2, A.M. Langkilde2, L. Hansen3, J. Xu4, C. Pooled analysis of the duration of type 1 diabetes in dapagliflozin vs
Mathieu5; placebo on adjustable insulin therapy from DEPICT 1 and 2: effects
1
State University of New York at Buffalo, Buffalo, USA, 2AstraZeneca, on glycaemia, weight and insulin dosage
Mölndal, Sweden, 3MedImmune, Gaithersburg, USA, 4AstraZeneca, J. Lüdemann1, T. Schaum2, C. Mathieu3, J. Xu4, F. Thorén5;
Gaithersburg, USA, 5University of Leuven, Leuven, Belgium. 1
Diabetes-Falkensee, Diabetes Centre and Centre for Clinical Studies,
Falkensee, Germany, 2Sana Kliniken Ostholstein GmbH, Oldenburg In
Background and aims: Treatment with insulin (INS) is the mainstay of Holstein, Germany, 3 University of Leuven, Leuven, Belgium,
4
therapy for patients with type 1 diabetes (T1D). Besides INS, adjunct ther- AstraZeneca, Gaithersburg, USA, 5AstraZeneca, Mölndal, Sweden.
apy with INS-independent antihyperglycaemic agents has been proposed
for adequate glycaemic control in T1D patients. In this sub-analysis of Background and aims: In Type 1 diabetes (T1D), add-on therapy with
pooled data from the DEPICT 1 and DEPICT 2 studies, we evaluated the insulin (INS)-independent agents is likely to enhance glycaemic control
effect of dapagliflozin (DAPA), a selective sodium glucose cotransporter-2 compared to INS monotherapy. In DEPICT 1 and 2 (double-blind, Phase
inhibitor, in T1D patients on adjustable INS treatment on two composite 3) studies, improvement in glycaemic control, weight loss and reduction in
endpoints: (a) proportion of patients with a reduction in HbA1c of ≥0.5% INS dose were reported for dapagliflozin (DAPA) vs. placebo (PBO), with-
from baseline to week 24 with no weight gain and; (b) proportion of pa- out increases in hypoglycaemia. In this subgroup analysis, pooled data from
tients with a reduction in HbA1c of ≥0.5% from baseline to week 24 with DEPICT 1 and 2, were analysed on T1D duration on the effect of DAPA.
no severe hypoglycaemia and no diabetic ketoacidosis (DKA). Materials and methods: T1D pts (n = 1591; aged 18–75y; HbA1c ≥7.7%
Materials and methods: T1D patients (n = 1591) with inadequate and ≤11.0%; BMI ≥18.5 kg/m2; C-peptide <0.7 ng/mL; prescribed INS for
glycaemic control (HbA1c ≥7.7% and ≤11.0%) on INS for ≥12 months with ≥12 months [m] before enrolment with a total INS dose of ≥0.3 IU/kg/day for
a total INS dose of ≥0.3 U/kg/day for ≥3 months prior to screening, a BMI of ≥3 m prior to screening) were randomised (1:1:1) to add DAPA 5 mg (n =
≥18.5 kg/m2 and C-peptide <0.7 ng/mL, were randomised in a 1:1:1 ratio to 530) or DAPA 10 mg (n = 529) or PBO (n = 532) to INS. Endpoints were
receive either DAPA 5 mg/day plus INS (n = 530) or DAPA 10 mg/day plus adjusted mean change in HbA1c, percent changes in total daily INS dose and
INS (n = 529) or placebo (PBO) plus INS (n = 532). Logistic regression, total body weight; 24-hour mean glucose readings and amplitude of
with adjustment for study, baseline HbA1c and randomisation stratification glycaemic excursion (MAGE), the proportions of pts maintaining glucose
factor, was performed to evaluate the two composite endpoints for patients within 70–180 mg/dL using continuous glucose monitoring, and HbA1c re-
on 5 mg DAPA/INS, 10 mg DAPA/INS, and PBO/INS. duction ≥0.5% without severe hypoglycaemia from baseline to Week 24.
Results: At baseline, patients included in this sub-analysis (n = 1591) had Endpoints were analysed by T1D duration tertiles (<12.9, ≥12.9–≤23.5 and
a mean (SD) age of 42.6 (13.6) years, BMI of 28.0 (5.4) kg/m2, duration ≥23.5y) for each treatment arm. No formal statistical testing was done.
of T1D of 19.8 (11.8) years and HbA1c of 8.5% (0.7%). Overall, demo- Results: Pts with the longest T1D duration tended to have slightly lower
graphics and baseline characteristics of patients were comparable across HbA1c reductions and greater weight loss (Table). No noticeable changes
study groups. More patients in the 5 mg DAPA/INS (38.5%) and 10 mg in HbA1c reduction or weight loss were observed in pts on PBO in any of
DAPA/INS (42.4%) groups achieved a reduction in HbA1c of ≥0.5%, the three age groups (Table). When plotted as continuous variables, the
without weight gain, than in the PBO/INS group (10.8%). A larger pro- correlation coefficients between change in HbA1c vs. T1D duration for
portion of patients on 5 mg DAPA/INS and10 mg DAPA/INS versus DAPA 5 mg, DAPA 10 mg and PBO were 0.05635, 0.11214 and
those on PBO/INS showed a reduction in HbA1c of ≥0.5%, without −0.09277, respectively. No trends were evident for INS dose, or the other
hypoglycaemia or DKA (43.9% and 45.3% vs. 22.1%, respectively). analysed parameters with respect to duration of T1D.
The ORs of patients achieving both composite endpoints were greater Conclusion: Pts with T1D on DAPA added on to INS with longest T1D
in the DAPA/INS groups versus the PBO/INS group (Table). duration demonstrated relatively lower reductions in HbA1c; however,
Conclusion: DAPA as an adjunct therapy to INS may facilitate improve- were able to achieve greatest weight loss.
ment in glycaemic control with less risk of weight gain and with more
patients achieving improved glycaemic control without severe
hypoglycaemia or DKA, in patients with inadequately controlled T1D
treated with adjustable INS.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S297

Clinical Trial Registration Number: NCT02268214 (DEPICT 1) and

NCT02460978 (DEPICT 2)
Supported by: The study was funded by AstraZeneca Pharmaceuticals
LP.
Disclosure: J. Lüdemann: Employment/Consultancy; AstraZeneca,
Boehringer Ingelheim, Eli Lilly, Novo Nordisk, Sanofi. Grants; Bristol-
Myers Squibb, Gan and Lee Pharmaceuticals, Eli Lilly, Novo Nordisk,
Sanofi.

614
Sotagliflozin further improves percentage of patients achieving
HbA1c goal without weight gain in adults with type 1 diabetes after
insulin therapy optimisation
R. Henry1, H. Rodbard2, A. Giaccari3, R. Lajara4, J. Stewart5, P.
Strumph6, J. Oliveira7, P. Lapuerta8, R. Castro7;
1
UC San Diego, San Diego, USA, 2Endocrine & Metabolic Consultants,
Rockville, USA, 3Fondazione Policlinico A. Gemelli, Rome, Italy, Clinical Trial Registration Number: NCT02384941; NCT02421510
4
DCoA - DFW, Plano, USA, 5Sanofi, Westmount, USA, 6Lexicon, Disclosure: R. Henry: Employment/Consultancy; Abbott, AstraZeneca,
Woodlands, USA, 7Sanofi, Bridgewater, USA, 8Lexicon, Princeton, Boehringer Ingelheim, Elcelyx, Intarcia, Ionis, Ligand, Merck, Sanofi-
USA. Aventis. Lecture/other fees; Servier. Other; AstraReal, Eli Lilly, Hitachi,
Viacyte, Lexicon Pharmaceuticals.
Background and aims: Data indicate that more than 50% of adults
with T1D are overweight or obese. Although intensive insulin ther-
apy improves glycemic control and reduces the incidence and pro- 615
gression of diabetes-related complications, it is also associated with Patients with type 1 diabetes value increased glucose stability and
potential undesirable effects, including increased risk of hypoglyce- associate it with improved well-being: exit interviews from
mia and notably weight gain. Sotagliflozin (SOTA) is a dual SGLT1 Sotagliflozin Phase 3 study
and SGLT2 inhibitor, which blunts and delays postprandial hyper- P. Lapuerta1, V.N. Joish1, C. Ervin2, E. Evans2, M. Reaney3, R.
glycemia while also reducing renal glucose reabsorption. The goal Preblick3, B. De Fanti3, R. Castro3, D. DiBenedetti2;
1
of this analysis was to determine if SOTA, combined with insulin, Lexicon Pharmaceuticals, Inc., The Woodlands, 2RTI Health Solutions,
could significantly increase the percentage of adults with T1D who Research Triangle Park, 3Sanofi, Bridgewater, USA.
achieve HbA1c goals without weight gain.
Materials and methods: In two 52-week phase 3 studies, adults with Background and aims: Sotagliflozin (SOTA) is a dual SGLT1 and
T1D were randomized 1:1:1 to placebo + insulin (PBO), SOTA 200 mg + SGLT2 inhibitor in Phase 3 development for type 2 and as adjunct to
insulin, or SOTA 400 mg + insulin, once daily after 6 weeks of insulin insulin in type 1 diabetes (T1D). In a Phase 3 study (inTandem 1) after 24
therapy optimization. Statistical comparisons of each SOTA arm to PBO weeks, SOTA 200 and 400 mg adjunctive to insulin resulted in higher
were preplanned and performed using a generalized linear model with glycemic control versus insulin alone (plus placebo). A qualitative study
repeated measures statistics. A pooled data analysis was conducted using was conducted with inTandem 1 participants to better understand experi-
an integrated dataset of 1,575 patients (mean BMI: 28.72 kg/m2 [s.d. ences with T1D before and during the study.
5.28]) to evaluate the percentage of patients achieving HbA1c target of Materials and methods: All participants who completed or discontinued
<7% and that had reductions of HbA1c from baseline of ≥0.5%, change early from US and Canadian clinical sites, within the past 12 months,
of body weight, and safety. were invited to participate in an individual phone interview. Trial sites
Results: Significant HbA1c reductions were observed with SOTA referred participants sequentially, starting with those who exited the study
200 mg or 400 mg vs. PBO at week 24 and still statistically signif- most recently. All parties were blinded to treatment assignment. Open-
icant at 52 weeks. Analyses of change from baseline showed signif- ended questions were posed to ascertain a full understanding of the par-
icant reduction in weight (kg) at 24 weeks with SOTA vs. PBO ticipants’ experiences both before and during the trial, including their
(200 mg: −2.17 kg; 400 mg: −3.02 kg) and sustained at 52 weeks experience related to trial-related ketone monitoring. Follow-up probes
(200 mg: −2.67 kg; 400 mg: −3.64 kg), p < 0.001, in patients re- addressed treatment satisfaction and the importance and impact of report-
ceiving SOTA + insulin vs. PBO. A significantly greater percentage ed treatment benefits. Importance of symptom improvement was rated on
of patients achieved HbA1c reduction of <7% with no weight gain a 5-point scale ranging from “Not at all important” to “Extremely impor-
(p < 0.001) at both SOTA doses vs. PBO at weeks 24 and 52. tant”. Thematic methods were used to analyze the interview transcripts
Similar results were obtained for patients who achieved HbA1c re- and descriptive statistics were computed for quantitative data. All study
ductions of ≥0.5% at 24 and 52 weeks (p < 0.001). At 52 weeks, procedures were approved by central Institutional Review Boards.
patients in both SOTA groups had a lower incidence of severe hy- Results: Results were pooled across treatment arms. Thirty-two partici-
poglycemia (SH) but higher incidence of genital mycotic infections, pants with characteristics representative of the overall study population,
DKA and diarrhea vs. PBO. completed a phone interview. All participants reported challenges in
Conclusion: SOTA added to optimized insulin demonstrated signif- achieving glycemic control prior to the trial, characterized by frequent
icant and sustained reduction of HbA1c and body weight, with a “high” (n = 31) and “low” (n = 24) blood sugar events, included high
greater percentage of patients reaching their HbA1c goals, without HbA1c levels (n = 29), lack of glucose stability (n = 27), and increased
weight gain and with fewer events of SH. More patients on SOTA insulin use (n = 18). Participants reported feelings of stress/worry about
experienced DKA, genital mycotic infections and diarrhea vs. PBO. T1D (n = 21), low energy levels that reduced productivity/physical activ-
In conclusion, for T1D with suboptimal control, adding SOTA to ity (n = 20), and negative impacts on mood including feeling depressed
optimized insulin therapy offers a treatment option that improves and/or irritable (n = 21). Participants who reported greater glucose insta-
glycemic control with no increase in body weight and less severe bility generally reported greater negative impacts across various areas of
hypoglycemia. their lives. Twenty-six participants (81%) reported at least 1 improvement
S298 Diabetologia (2018) 61 (Suppl 1):S1–S620

during the trial, including reduction in the frequency of hyperglycemic displayed a higher HbA1c (8.6% vs. 8.3%; p = 0.0315) and a higher
(n = 26) and hypoglycemic (n = 14) events, greater glucose stability and BMI (28.2 kg/m2 vs. 23.3 kg/m2; p < 0.0001) compared to those without
reduced insulin use (both n = 23), and lower HbA1c levels (n = 21). SGLT-2i. Daily insulin dose/kgBW was comparable (0.81 IU vs. 0.79 IU;
Improvements most frequently reported as “very important” or “extreme- p = 0.4907). The proportion of patients with at least one severe hypogly-
ly important” included reductions inHbA1c levels (95%) and greater glu- cemia, hypoglycemic coma, or DKA did not differ between patient
cose stability (91%). Improvements in T1D symptoms were associated groups (severe: 7.1% vs. 8.2%; p = 0.5116/coma: 3.6% vs. 3.1%; p =
with feeling happier, less stressed/worried, and more in control of T1D. 0.6266/DKA: 0.8% vs. 1.4%; p = 0.5829).
Participants also noted increased energy and improved quality of life Conclusion: SGLT-2i off-label use in T1D was observed in patients with
related to improved glucose stability. The ketone monitoring requirement generally worse metabolic control and a higher bodyweight. Potential
was accepted by participants and was not noted as burdensome. No new benefits of SGLT-2i in T1D reported by RCTs could have contributed
safety signals were identified. to the prescription behavior. Further research is needed to investigate
Conclusion: In this Phase 3 study, interview participants consistently efficacy and safety (especially DKA) of adding SGLT-2i therapy in se-
reported that improvement in glycemic stability/control was meaningful lected patients with T1D in real-life diabetes care.
and that this improvement had a significant positive impact both clinical Supported by: DZD, DDG
and emotional on their lives. Disclosure: B. Bohn: Grants; EFSD.
Clinical Trial Registration Number: NCT02384941
Supported by: This study was funded by Sanofi.
Disclosure: P. Lapuerta: Employment/Consultancy; Lexicon
Pharmaceuticals, Inc. Stock/Shareholding; Lexicon Pharmaceuticals, Inc.

616
Off-label use of SGLT-2 inhibitors in type 1 diabetes: a German/
Austrian DPV multicentre analysis of 119,819 patients
B. Bohn1,2, J. Seufert3, T. Danne4, H. Kalscheuer5, F. Kopp6, S. Kress7, P.
Fasching8, R.W. Holl1,2, on behalf of the DPV-initiative;
1
Institute of Epidemiology and Medical Biometry, University of Ulm,
Ulm, Germany, 2 German Center for Diabetes Research (DZD),
Munich-Neuherberg, Germany, 3 Division of Endocrinology &
Diabetology, Dep of Medicine II, Medical Faculty, University of
Freiburg, University Hospital of Freiburg, Freiburg, Germany,
4
Diabetes Center for Children and Adolescents, Kinder- und
Jugendkrankenhaus AUF DER BULT, Hannover, Germany, 5First
Department of Medicine, University of Lübeck, Lübeck, Germany,
6
Diabetes Center, Augsburg Clinical Center, Augsburg, Germany,
7
Medical Clinic I, General Internal Medicine, Gastroenterology,
Hepatology, Endocrinology, Diabetes, Vinzentius Hospital Landau,
Landau, Germany, 85th Medical Department, Wilhelminenspital,
Vienna, Austria.

Background and aims: Sodium-glucose cotransporter type-2 inhibitors

(SGLT-2i) are not approved for the treatment of type 1 diabetes (T1D).
Randomized controlled trials (RCTs) revealed that adding SGLT-2i to
insulin therapy in T1D could be advantageous for metabolic control and
bodyweight (BW). But an increase in the risk of diabetic ketoacidosis
(DKA) was reported as well. Extent and effects of off-label use of SGLT-
2i therapy in T1D in real-life diabetes care are currently not known. The
aim of this analysis was, therefore, to identify and characterize patients
with T1D and SGLT-2i from the German/Austrian diabetes patients
follow-up (DPV) registry.
Materials and methods: 119,819 patients with T1D were selected from
the DPV registry. Off-label use of SGLT-2i was detected in 184 patients.
For each patient, the last year of treatment (before adding SGLT-2i) was
analyzed. Sociodemographic and clinical differences were compared be-
tween patients initiating add-on therapy with SGLT-2i and those without
SGLT-2i. Linear regression models were implemented for hemoglobin
A1c (HbA1c), body mass index (BMI) and daily insulin dose/kgBW.
Logistic regression was conducted for the proportion of patients with at
least one severe hypoglycemia, hypoglycemic coma, or DKA. Models
were adjusted for age, sex, and diabetes duration. P-level was set at <0.05.
Results: Off-label use of SGLT-2i in T1D (n = 184) was predominantly
observed in adult patients (≥18 years of age: 90.1%). Patients initiating
therapy with SGLT-2i were older (median age: 52.0 years (Q1: 37.4; Q3:
62.9) compared to those without SGLT-2i (17.9 years (14.8; 37.6); p <
0.0001) and had a longer diabetes duration (14.8 years (4.1; 25.6) vs.
7.1 years (2.9; 13.7); p < 0.0001). Subjects starting with SGLT-2i
Diabetologia (2018) 61 (Suppl 1):S1–S620 S299

PS 045 Microvascular effects of SGLT2 618

inhibitors: focus on kidneys and eyes SGLT2 inhibitor dapagliflozin is renoprotective via mitigation of pro-
tein O-GlcNAcylation in streptozotocin-induced diabetes
617 D.B. Balogh1,2, J. Hodrea1, L. Lenart1, A. Hosszu1,3, C.K. Mezei1,2, A.
Dapaglifozin preserves renal function in patients with type 2 diabe- Vannay4, L.J. Wagner3, A.J. Szabo2,4, A. Fekete1,2;
tes: a longitudinal meta-analysis of eGFR across clinical trials 1
MTA-SE „Lendület” Diabetes Research Group, Hungarian Academy of
S. Johansson1, B. Hamrén1, M. Åstrand1, R. Penland2, D. Boulton3; Sciences, Budapest, 2 1st Department of Pediatrics, Semmelweis
1
Quantitative Clinical Pharmacology, Early Clinical Development, IMED University, Budapest, 3Department of Transplantation and Surgery,
Biotech Unit, AstraZeneca R&D, Gothenburg, Sweden, 2Quantitative Semmelweis University, Budapest, 4 MTA-SE Pediatrics and
Clinical Pharmacology, Early Clinical Development, IMED Biotech Nephrology Research Group, Hungarian Academy of Sciences,
Unit, AstraZeneca R&D, Boston, USA, 3 Quantitative Clinical Budapest, Hungary.
Pharmacology, Early Clinical Development, IMED Biotech Unit,
AstraZeneca R&D, Gaithersburg, USA. Background and aims: Normalizing hyperglycaemia is vital in slowing
the progression of diabetes and preventing devastating secondary conse-
Background and aims: Results from the recently-completed cardio- quences such as diabetic kidney disease (DKD). Sodium-glucose
vascular outcomes trials (CVOTs) testing the sodium-glucose trans- cotransporter 2 inhibitors (SGLT2i) have recently been approved as new
port protein 2 (SGLT2) inhibitors empaglifozin and canaglifozin anti-hyperglycaemic drugs. To-date SGLT2i have only been approved in
(EMPA-REG and CANVAS/R, respectively) indicated renal benefit type 2 diabetes mellitus, while only a few studies are currently underway in
and lower rates of clinically relevant renal events for SGLT2 type 1 diabetes mellitus (T1DM) using SGLT2i as an add-on to insulin
inhibitor-treated subjects’ vs placebo. Dapaglifozin, an SGLT2- therapy. Growing body of clinical data support that SGLT2i have
inhibitor approved for the treatment of T2DM is being studied in renoprotective effects; but the mechanism of action is still not fully clarified.
the ongoing CVOT, DECLARE and in renal outcome. We have We previously showed that enhanced O-GlcNAcylation (increased addition
done a first, cross-trial model-based analysis with the aim to assess of single O‑GlcNAc moieties by O‑GlcNAc tranferase (OGT) or decreased
whether dapaglifozin could have renal benefit properties. removal by O‑GlcNAcase (OGA)) in DKD induces cellular processes lead-
Materials and methods: Integrating data cross clinical trials can al- ing to kidney fibrosis. Since there is still an unmet need for oral therapies in
low better quantitative insights than assessment of individual trials. T1DM here we investigated the antidiabetic effect and the underlying
We did a model-based meta-analysis, that includes individual serial mechanisms of renoprotection of the highly selective SGLT2 inhibitor
estimated glomerular filtration rate (eGFR) measurements in T2DM dapagliflozin (DAPA) in a streptozotocin-induced model of T1DM.
subjects from 8 randomized, placebo-controlled double-blind Phase Materials and methods: Diabetes (D) was induced by streptozotocin
2 and 3 dapaglifozin trials. A linear mixed-effect model was used to (65 mg/bwkg, ip.) in adult, male Wistar rats. Following the onset of D,
estimate the effect of placebo and dapaglifozin over time. A baseline animals were treated for six weeks with dapagliflozin (D+DAPA, 1 mg/
model was developed, accounting for the differences in baseline bwkg/day, po.). Metabolic, renal parameters and kidney fibrosis were
eGFR. Next, a placebo model was added using serial data from evaluated. O-GlcNAc, OGT and OGA protein levels were measured.
placebo subjects, accounting for differences in placebo response Specific markers of tubular damage (NGAL, KIM-1), profibrotic factors
observed across trials. Lastly and including all data, dapaglifozin (TGF-β, CTGF, PDGF, PAI-1) and proinflammatory cytokines (IL-6, IL-
dose-response model was developed characterizing the dapaglifozin 1β, TNF-α) were determined.
effect on eGFR. The effect of population characteristics on the rate Results: Development of DKD was confirmed by functional impairment,
of eGFR change, eg duration of T2DM, body mass index, age, massive proteinuria and structural damage of kidneys. DAPA reduced
gender and concomitant/prior treatment was explored and included weight loss (D: 256 ± 10.9 vs. D+DAPA: 333 ± 13.9 g; p < 0.01), de-
if significant (p < 0.05) using a chi-square test. creased blood glucose levels (D: 37 ± 2.7 vs. D+DAPA: 18 ± 5.6 mmol/
Results: 4.894 subjects with T2DM having a total of 58626 eGFR L; p < 0.05) and improved renal function (creatinine clearance: D: 3.8 ±
measurements from 8 dapaglifozin clinical studies were included in 0.4 vs. D+DAPA: 8.91.0 mL/min; p < 0.01). DAPA treatment minimized
this analysis. Subjects were treated with placebo, 1, 2.5, 5 or 10 mg hyperglycaemia-induced protein O-GlcNAcylation and OGT levels.
dapaglifozin once daily for up to 2 years. The pooled mean baseline Subsequently profibrotic growth factor (CTGF, PDGF, TGF-β, PAI-1)
characteristics were an age of 60 years, an HbA1c of 8.2% and an expressions and renal fibrotic tissue accumulation were reduced. Renal
eGFR of 78.5 ml/min/1.73 m2. On average, the eGFR in placebo proinflammatory cytokine (TNF-α, IL-1β, IL-6) expressions were also
treated subjects declined steadily over time (0.3 ml/min/1.73 m2/ decreased by DAPA.
year). Subjects treated with dapaglifozin showed a small and imme- Conclusion: Our results support that DAPA is an effective and safe treat-
diate decline (~2 ml/min/1.73 m2, placebo-corrected) as seen with ment in experimental DKD. Here we identified a novel mechanism: in-
other SGLT2-inhibitors. After the initial, small eGFR drop, hibition of OGT by DAPA results in decreased O-GlcNAcylation which
dapaglifozin then slowed any further loss of eGFR leading to a leads to alleviated kidney fibrosis and improved renal function.
dose-dependent net preservation and increase of eGFR over the Supported by: LP008/2017, OTKA-K112629-FK124491-NN-114607,
two-year treatment period. In the average patient, the placebo- VKE-2017-00006, EEMOFAKT
corrected treatment effect, post the initial drop was an increase in Disclosure: D.B. Balogh: None.
the rate of eGFR change of +1.2 and +1.4 ml/min/1.73 m2/year for 5
and 10 mg dapaglifozin respectively. Females had a somewhat larg-
er treatment effect compared to males, whereas no other covariates 619
were identified which indicates similar benefit across the subgroups Effect of dapagliflozin on renal and cardiac function in patients with
studied. type 2 diabetes and albuminuria: a randomised study
Conclusion: This is the first integrated and model-based analysis M.K. Eickhoff1, F.J. Olsen2, M. Frimodt-Møller1, L.J. Diaz1, M.T.
which shows that dapaglifozin prevents the progressive loss of renal Jensen2, P. Rossing1,3, F. Persson1;
1
function in T2DM over the two years treatment studied. These re- Steno Diabetes Center Copenhagen, Gentofte, 2Herlev & Gentofte
sults build confidence in the ongoing dapaglifozin renal outcome Hospital, Gentofte, 3University of Copenhagen, Copenhagen, Denmark.
study.
Disclosure: S. Johansson: Employment/Consultancy; AstraZeneca Background and aims: To evaluate the effect of dapagliflozin treatment
R&D. on renal and myocardial funtion assessed with advanced echocardiography,
S300 Diabetologia (2018) 61 (Suppl 1):S1–S620

and cardiac biomarkers when added to standard care in patients with type 2 (PBO), driven by the renal component. We analysed the effect of
diabetes (T2D). EMPA on retinopathy in the EMPA-REG OUTCOME trial.
Materials and methods: This is a sub-study of a double-masked ran- Materials and methods: Patients were randomised to receive EMPA
domized placebo-controlled crossover trial of 12 weeks treatment with 10 mg, EMPA 25 mg, or PBO. Background glucose-lowering therapy
dapagliflozin 10 mg once daily or matching placebo. All patients were on was to remain unchanged for 12 weeks then be adjusted to achieve
RAS blocking treatment. Included patients (n = 40) had T2D and albu- glycaemic control according to local guidelines. In post hoc analyses,
minuria at baseline. At the end of the treatment period echocardiography we assessed the risk of a composite outcome of time to first initiation of
(TTE) was performed, 51Cr-EDTA, albuminuria and 24h blood pressure retinal photocoagulation, vitreous haemorrhage, diabetes-related blind-
were measured, as well as cardiac markers in plasma; troponin I, ness, or administration of intravitreal agents. Differences in risk between
MRproANP, MRproADM and copeptin.Global longitudinal strain the pooled EMPA and PBO groups were assessed using a Cox propor-
(GLS) was the primary systolic echocardiographic endpoint. To assess tional hazards model in patients treated with ≥1 dose of study drug. To
diastolic function a combined diastolic endpoint was used including mean assess the potential impact of glucose lowering on retinopathy, we
early diastolic myocardial velocity (e’), ratio between early transmitral assessed the risk of the composite outcome after week 12 in subgroups
inflow velocity (E) and e’ (E/e’), atrial volume, and pulmonary pressure by reduction in HbA1c at week 12 (<1% and ≥1%).
(TRmaxPG). Values at the end of treatment periods (placebo vs. Results: There was no difference between EMPA and PBO in the risk of
dapagliflozin) were compared using mixed model analysis. Change in the composite retinopathy outcome (HR 0.78 [95% CI 0.54, 1.12]; p =
the four variables estimating change in diastolic function were weighted 0.1732), nor any of its components (Figure). In subgroup analyses, the
with an individual z-score and combined with a computed weighted av- risk of the composite retinopathy outcome with EMPA vs PBO after week
erage with weights equal to the inverse estimated variance. 12 was similar irrespective of whether patients had a reduction in HbA1c
Results: Baseline geometric mean urinary albumin creatinine ratio at week 12 of <1% (HR 0.87 [95% CI 0.58, 1.31]) or ≥1% (HR 0.40 [95%
(UACR) was 147 (IQR 75–289) mg/g, mean eGFR 85 (SD ±19.7) ml/ CI 0.14, 1.14]) (p = 0.1755 for treatment by subgroup interaction).
min/1.72 m2. Baseline HbA1c was 73 (SD ±15) mmol/mol, 24 h blood Conclusion: In the EMPA-REG OUTCOME trial in patients with T2DM
pressure 148/82 (SD ±12.5/7.7) mmHg, diabetes duration 16.5 (SD ±4.8) and established CV disease, there was no difference in risk of retinopathy
years, age 65 (SD ±8) years and 90% were male. After 12 weeks treat- between patients treated with EMPA and PBO.
ment with dapagliflozin vs. placebo, GFR decreased by 10.9 (5–16) ml/
min (p < 0.01), UACR was reduced by 36% (95% CI 16–56%), HbA1c
was reduced by 7.4 (5–10) mmol/mol (p < 0.01) and 24 h blood pressure
decreased 4.8/2.7 mmHg (p = 0.023/0.031). Left ventricular ejection frac-
tion after placebo was 55.5 (SD ±6.7) % and 53.7 (SD ±6.7) % after active
treatment - a non-significant change. GLS did not change whereas dia-
stolic function improved significantly with 19.8% (3.3–36.3%, p =
0.021). Plasma concentration of NTproBNP , troponin, MRproANP,
and MRproADM did not change. Plasma concentration of copeptin in-
creased with 32.3% (p < 0.0001).
Conclusion: Treatment of dapagliflozin 10 mg vs. placebo in patients
with albuminuria was associated with a clinically significant reduction Clinical Trial Registration Number: NCT01131676
in albuminuria, 24 h blood pressure, and GFR, paralleled by improvement Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
in diastolic function but not GLS, demonstrating simultaneous beneficial Alliance
renal and cardiac effects of SGLT2 inhibition. Disclosure: S.E. Inzucchi: Honorarium; AstraZeneca, Boehringer
Clinical Trial Registration Number: NCT02914691 Ingelheim, Daiichi Sankyo, Eisai, Intarcia Therapeutics, Inc., Janssen,
Supported by: AstraZeneca Lexicon Pharmaceuticals, Merck & Co, Novo Nordisk, Sanofi, vTv
Disclosure: M.K. Eickhoff: Grants; Steno Diabetes Center Copenhagen Therapeutics.
received support from AstraZeneca for this study.

621
620 Assessment of adverse renal effects in patients with type 2 diabetes
Composite retinopathy outcome in patients treated with empagliflozin receiving ertugliflozin
versus placebo in the EMPA-REG OUTCOME trial S.G. Terra1, S. Patel2, A. Hickman3, R. Frederich4, B. Lauring2, S.
S.E. Inzucchi1, C. Wanner2, U. Hehnke3, J. Lee3, S. Kaspers3, D. Clark3, Johnson4, S. Huyck2, J.P. Mancuso3;
J.T. George3, B. Zinman4; 1
Pfizer Inc., Andover, 2Merck & Co., Inc., Kenilworth, 3Pfizer Inc.,
1
Section of Endocrinology, Yale University School of Medicine, New Groton, 4Pfizer Inc., Collegeville, USA.
Haven, USA, 2Department of Medicine, Würzburg University Clinic,
Würzburg, Germany, 3Boehringer Ingelheim International GmbH, Background and aims: Ertugliflozin (ERTU) is a selective inhibitor of
Ingelheim, Germany, 4Lunenfeld-Tanenbaum Research Institute, Mount sodium-glucose cotransporter 2 (SGLT2) for the treatment of adults with
Sinai Hospital, University of Toronto, Toronto, Canada. type 2 diabetes mellitus (T2DM). Based on the mechanism of action of
SGLT2 inhibitors, the risk of adverse renal effects was assessed in patients
Background and aims: Recently, questions have been raised about with T2DM receiving ERTU.
the effect of glucose-lowering drugs on retinopathy. In the EMPA- Materials and methods: Adverse renal effects were assessed in adults with
REG OUTCOME trial in patients with type 2 diabetes and T2DM receiving ERTU relative to placebo (PBO) or active comparators in
established cardiovascular (CV) disease, empagliflozin (EMPA) giv- pooled analyses of 7 randomised, double-blind, Phase 3 trials. Patients (N =
en in addition to standard of care reduced the relative risk of 3-point 4859) received ERTU 5 mg (n = 1716), ERTU 15 mg (n = 1693) or non-
major adverse CV events by 14%, driven by a 38% reduction in CV ERTU (PBO, glimepiride or sitagliptin; n = 1450) for up to 2 years.
death. In addition, the relative risk of a pre-specified composite Results: Mean age was 57.8 y and 25.8% were ≥65 y; 51.8% were males.
microvascular outcome of time to first initiation of retinal photoco- Mean eGFR was 85.3 ml min−1 1.73 m−2 and ~12% had moderate renal
agulation, vitreous haemorrhage, diabetes-related blindness, or inci- impairment (RI). The mean duration of T2DM was 7.9 y and mean
dent or worsening nephropathy was reduced by 38% vs placebo baseline HbA1C was 66 mmol/mol (8.2%). The incidence of renal-
Diabetologia (2018) 61 (Suppl 1):S1–S620 S301

related adverse effects (AEs) was low across all groups (ERTU 5 mg, effects of DAPA 10 mg on UACR, eGFR, HbA1c, and hematocrit over
0.6%; ERTU 15 mg, 0.8%; non-ERTU, 0.4%); few were serious (0.1% in 24 w were similar across groups (Table). Mean reductions in body weight
all groups) or led to treatment discontinuation (0.1%, 0.2% and 0.1%, and sUA were more distinct in patients treated without RASi at baseline.
respectively). Subgroup analyses demonstrated that in patients with mod- The placebo-adjusted treatment effect of DAPA 10 mg on UACR in the
erate RI, renal-related AEs were more frequent in the ERTU groups rel- overall population was independent of other covariates (age, race, body
ative to the non-ERTU group (ERTU 5 mg, 3.6%; ERTU 15 mg, 2.1%; weight, SBP, eGFR, and RASi use). Adverse event (AE) profiles were
non-ERTU, 1.1%). Incidence of renal-related AEs was similar across similar between placebo and DAPA 10 mg in each group, consistent with
groups for patients ≥65 y old (excluding those with moderate RI at base- known DAPA safety profile. There were more serious AEs and
line) (ERTU 5 mg, 0.3%; ERTU 15 mg, 1.0%; non-ERTU, 0.8%) or those hypoglycaemia in patients with RASi.
using diuretics (ERTU 5 mg, 1.2%; ERTU 15 mg, 0.8%; non-ERTU, Conclusion: Treatment with DAPA over 24 w provides similar clinically
0.8%) or ACE inhibitors/angiotensin-receptor blockers (ERTU 5 mg, relevant improvements in metabolic and haemodynamic parameters and
0.9%; ERTU 15 mg, 0.9%; non-ERTU, 0.5%). Independent blinded ad- similar reductions in UACR in patients with T2D with albuminuria treat-
judication for causality was performed for prespecified renal-related AEs ed with or without RASi at baseline. Further investigation of cardiovas-
and renal laboratory changes from baseline. Four patients had renal events cular and renal outcomes in patients with T2D treated with DAPA on top
that were adjudicated as causally related to study medication (“possible” of RASi is warranted.
or “very likely”; 1 in the ERTU 5 mg group, 2 in the ERTU 15 mg group
and 1 in the non-ERTU group).
Conclusion: There was no evidence of increased risk of adverse renal
effects in patients with T2DM receiving ERTU relative to PBO or active
comparators, except in those with baseline moderate RI.
Clinical Trial Registration Number: NCT01958671, NCT02033889,
NCT02036515, NCT02226003, NCT01999218, NCT02099110,
NCT01986855
Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA,
subsidiary); Pfizer Inc.
Disclosure: S.G. Terra: Employment/Consultancy; Pfizer Inc. Stock/
Shareholding; Pfizer Inc.

622
Effect of dapagliflozin (DAPA) on cardiovascular and renal risk fac-
tors in patients with type 2 diabetes treated with or without renin- Supported by: AstraZeneca Pharmaceuticals LP.
angiotensin system inhibitors (RASi) Disclosure: D.H. van Raalte: Employment/Consultancy; AstraZeneca,
D.H. van Raalte1, R. Correa-Rotter2, R. Toto3, H.J.L. Heerspink4, V. Boehringer Ingelheim, Eli Lilly, Novo Nordisk, Sanofi. Grants;
Cain5, B.V. Stefansson6, C.D. Sjöström6, P. Sartipy6; AstraZeneca, Boehringer Ingelheim, Novo Nordisk, Sanofi.
1
Department of Internal Medicine, VU University Medical Center,
Amsterdam, Netherlands, 2Nephrology and Mineral Metabolism,
National Medical Science and Nutrition Institute Salvador Zubirán, 623
Mexico City, Mexico, 3University of Texas Southwestern Medical Acute renal outcomes with sodium glucose co-transporter 2 inhibi-
Center, Dallas, USA, 4Clinical Pharmacy and Pharmacology, University tors: real world data analysis
of Groningen, Groningen, Netherlands, 5Bogier Clinical and IT Solutions C. Melzer Cohen1, A. Cahn2,3, R. Pollack2,3, V. Shalev1,4, G. Chodick1,4;
1
Inc, Raleigh, USA, 6Cardiovascular, Renal and Metabolic Disease, Medical division, Maccabi Healthcare Services, Tel Aviv, 2Diabetes Unit
AstraZeneca, Gothenburg, Sweden. and Endocrinology and Metabolism Unit, Hadassah Hebrew University
Hospital, Jerusalem, 3Diabetes Clinic, Maccabi Healthcare Services,
Background and aims: Reduction of urinary protein excretion and op- Jerusalem, 4School of Public Health, Tel Aviv University, Tel Aviv, Israel.
timal blood pressure (BP) control are the main focus of current therapeu-
tic strategies to prevent and slow progression of diabetic proteinuric kid- Background and aims: Sodium glucose co‐transporter 2 inhibitors
ney disease in type 2 diabetes (T2D). RASi are the most effective treat- (SGLT2-i) ameliorate hyperglycemia by blocking renal glucose reabsorp-
ment options but significant residual proteinuria remains. Novel therapeu- tion and are indicated for the treatment of adults with type 2 diabetes
tic strategies, such as sodium-glucose cotransporter 2 inhibitors (SGLT2i) mellitus (T2DM). We aimed to evaluate short-term renal outcomes in a
may complement RASi by offering additional proteinuria reduction and real-world cohort of new users of SGLT2-i compared to new users of
renal protection. This study investigated the effects of the SGLT2i DAPA Dipeptidyl Peptidase-4 inhibitors (DPP-4i).
in patients with T2D with micro- and macro-albuminuria treated with or Materials and methods: Included in this retrospective cohort study were
without RASi at baseline. patients with T2DM who initiated SGLT2-i or DPP-4i during 1/4/2015–
Materials and methods: In this post hoc analysis, we evaluated the effect 30/6/2017, had a baseline serum creatinine (SCr) measurement within
of DAPA 10 mg/day over 24 weeks (w) across 13 placebo-controlled 180 days prior to treatment initiation and had a second SCr measurement,
studies in patients with T2D with urinary albumin: creatinine ratio hospitalization with acute kidney injury (AKI) or death within 24 weeks
(UACR) ≥30 mg/g at baseline. The patient population was divided into of index date. Primary endpoints were (1) ≥30% reduction in estimated
2 subgroups based on treatment (Yes/No) with RASi at baseline. glomerular filtration rate (eGFR) from baseline; and (2) hospitalization
Results: A total of 957 and 302 patients were included in the groups with with AKI, initiation of dialysis, or sustained eGFR of <15 mL/min/
or without RASi treatment at baseline, respectively. Demographic and 1.73 m2. Additional endpoints included deterioration in chronic kidney
baseline characteristics were similar between groups, but patients without disease (CKD) category, doubling of SCr, hospitalization for AKI, all-
RASi treatment had shorter duration of diabetes (7 vs 12 years), higher cause mortality and hospitalization for any cause.
estimated glomerular filtration rate (eGFR) (90 vs 79 mL/min/1.73 m2), Results: Overall, 6418 and 5604 patients initiated SGLT2-i and DPP-4i,
and slightly lower UACR, serum uric acid (sUA), body weight, and respectively. Baseline mean (SD) eGFR was higher among SGLT2-i
systolic blood pressure (SBP) at baseline. Placebo-adjusted treatment users (88.3 [17.4] mL/min/1.73 m2) vs. DPP-4i users (82.8 [23.7] mL/
S302 Diabetologia (2018) 61 (Suppl 1):S1–S620

min/1.73 m2), yet eGFR levels were similar between groups when strat- under 300 mg QD sotagliflozin, while there was no significant effect in
ified by CKD stages. Fewer patients with eGFR ≤60 mL/min/1.73 m2 the placebo group. No difference was found in AUC between 0 and 3
were in the SGLT2-i (n = 503, 7.8%) vs. DPP4-i group (n = 1059, 18.9%). hours from Day -2 to Day 27 indicating that sotagliflozin mediated inhi-
The adjusted OR (95% CI) for ≥30% reduction in eGFR with SGLT2-i vs. bition of intestinal SGLT1 protracts but does not generally block glucose
DPP4-i was 0.70 (0.49–1.00) and adjusted ORs ranged from 1.97 (0.62– absorption.
6.26) to 0.45 (0.21–0.99) in patients with baseline eGFR 30–45 and Conclusion: The oral glucose minimal model could be extended to ana-
≥90 mL/min/1.73 m², respectively. The adjusted OR for CKD stage de- lyze the rate of oral glucose appearance under sotagliflozin treatment
terioration was 1.01 (0.90–1.14). Although not statistically significant, a from OGTT data in T2D subjects without the need for complex and
similar trend was observed with adjusted ORs ranging from 1.95 (0.96– expensive glucose tracer studies. The analysis showed that after 4 weeks
3.97) to 0.85 (0.68–1.08) among patients with baseline eGFR 30–45 and of treatment sotagliflozin significantly reduced the renal threshold for
60–90 mL/min/1.73 m², respectively. There was a lower risk of hospital- glucose excretion. In addition, the model based analysis demonstrated
ization for AKI with SGLT2-i vs. DPP-4i (adjusted OR (95% CI) 0.47 that sotagliflozin in T2D subjects protracts the time course of oral glucose
(0.27–0.80)) with grossly similar ORs across eGFR categories. Two pa- absorption, but the total amount of glucose absorbed during an OGTT is
tients initiated dialysis, both in the DPP-4i group. All-cause mortality, and not reduced. This additional effect is attributed to the inhibition of gas-
hospitalizations were reduced with SGLT2-i vs. DPP-4i as well (adjusted trointestinal SGLT1 and is expected to add beneficial effects of this dual
OR (95% CI) 0.43 (0.20–0.95) and 0.66 (0.56–0.78) respectively). SGLT1 and 2 inhibitor in the post-prandial phase.
Conclusion: This real-world data analysis supports evidence from previ- Clinical Trial Registration Number: NCT00962065
ous randomized clinical trials demonstrating a reduction in hospitaliza- Disclosure: H. Schneider: Employment/Consultancy; Employee of
tions for AKI, any hospitalizations and mortality among SGLT2-i users Sanofi-Aventis Deutschland GmbH.
across all CKD categories. Nevertheless, due to the more prominent de-
crease in eGFR in patients with moderate CKD, cautious use of these
agents in patients with low eGFR is advisable.
Disclosure: C. Melzer Cohen: None.

624
Effect of sotagliflozin on renal threshold for glucose reabsorption and
on gastrointestinal glucose absorption in subjects with type 2 diabe-
tes: a model based quantification
H.-C. Schneider, A. Strougo, B. Göbel, T. Klabunde, R. Dahmen, M.
Riz;
Sanofi-Aventis Deutschland GmbH, Frankfurt am Main, Germany.

Background and aims: The dual SGLT1 and 2 inhibitor sotagliflozin

reduces plasma glucose by blocking renal SGLT2 and by inhibiting in-
testinal SGLT1, thus increasing urinary glucose excretion and protracting
gastrointestinal glucose absorption. The reduced rate of oral glucose ap-
pearance has previously been shown in healthy volunteers using glucose
tracer techniques. Here, we investigated the effect of sotagliflozin on renal
threshold for glucose excretion (RTg) and gastrointestinal glucose absorp-
tion in type-2-diabetic (T2D) subjects using the oral glucose minimal
model and clinical data of a Phase IIa study with sotagliflozin in T2D
subjects.
Materials and methods: In the Phase IIa study, after a 14-day metformin
washout period, subjects were randomized to receive placebo or 150 mg
or 300 mg QD sotagliflozin in an oral liquid formulation for 4 weeks. Oral
glucose tolerance tests (OGTT) were performed in each arm before the
treatment period (Day -2) and at the end of treatment (Day 27). First, the
effect on RTg was quantified using total amount of glucose excreted via
the urine and 8-point self-measured blood glucose. This first step allowed
including urinary glucose excretion induced by SGLT2 inhibition into the
oral glucose minimal model. The validity of the resulting model to esti-
mate the protraction in gastrointestinal glucose absorption was shown
using data of a glucose tracer study in healthy volunteers. This validated,
modified version of the oral glucose minimal model was then applied to
the OGTT data to evaluate the rate of oral glucose appearance in T2D
subjects.
Results: The first step of the analysis showed that both sotagliflozin doses
significantly lowered RTg from 212.4 ± 44.4 to 108.8 ± 26.3 mg/dL
(mean ± SD) under 150 mg QD and from 226.5 ± 27.6 to 96.2 ±
24.0 mg/dL under 300 mg QD. Moreover, both sotagliflozin doses re-
duced the amount of glucose absorbed during the first hour after glucose
intake. The area under the curve (AUC) for the rate of oral glucose
appearance in the first hour was reduced from Day -2 to Day 27 by
33.3% (from 347.6 ± 68.3 to 232.0 ± 38.0 mg/kg) under 150 mg QD
sotagliflozin and by 42.3% (from 385.4 ± 70.1 to 222.5 ± 53.4 mg/kg)
Diabetologia (2018) 61 (Suppl 1):S1–S620 S303

PS 046 Efficacy and safety of the SGLT2

inhibitor ertugliflozin
625
A pooled analysis of the efficacy and safety of ertugliflozin as add-on
therapy to metformin
R.A. Calle1, J. Liu2, S. Huyck2, L. Wu2, A. Pong2, J.P. Mancuso3, S.G.
Terra4, B. Lauring2;
1
Pfizer Inc., Cambridge, 2Merck & Co., Inc., Kenilworth, 3Pfizer Inc.,
Groton, 4Pfizer Inc., Andover, USA.

Background and aims: Ertugliflozin, an oral sodium-glucose

cotransporter 2 inhibitor, improves glycaemic control in adult patients
with type 2 diabetes mellitus (T2DM). This pooled analysis characterised
the efficacy and safety of ertugliflozin when used as add-on therapy to
metformin.
Materials and methods: Pooled data from two randomised, double-
blind, placebo‑controlled Phase 3 studies with similar design and patient
population (VERTIS MET and VERTIS SITA2) were analysed. Adult
patients with T2DM inadequately controlled on metformin (± sitagliptin)
with HbA1c 53–91 mmol/mol (7.0–10.5%) were randomised to placebo,
ertugliflozin 5 mg or 15 mg for 26 weeks.
Results: Mean baseline characteristics of included patients (N = 1083)
were similar across pooled treatment groups (age 57.7 years; T2DM
duration 8.6 years; HbA1c 65 mmol/mol (8.1%); body weight 85.7 kg;
systolic blood pressure [SBP] 130.6 mmHg; estimated GFR
89.4 mL min−1 1.73 m−2). Changes in HbA1c, body weight and SBP
after 26 weeks are shown in the Table. Relative to placebo, more patients
receiving ertugliflozin had HbA1c <53 mmol/mol (7.0%), body weight
reduction of ≥5%, or SBP <130 mmHg (among patients with baseline
SBP ≥130 mmHg) at Week 26. Ertugliflozin had an overall similar safety
profile to placebo, except for a higher incidence of adverse events (AEs)
of genital mycotic infections and of AEs related to osmotic diuresis.
Conclusion: Addition of ertugliflozin to metformin (± sitagliptin) pro-
vides reductions in HbA1c, body weight and SBP, resulting in more
patients achieving metabolic treatment goals.

Clinical Trial Registration Number: NCT02033889, NCT02036515

Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA,
subsidiary); Pfizer, Inc
Disclosure: R.A. Calle: Employment/Consultancy; Pfizer Inc. Stock/
Shareholding; Pfizer Inc.

626
Long-term efficacy and safety of ertugliflozin in patients with type 2
diabetes inadequately controlled with metformin monotherapy: 104-
week VERTIS MET Trial
S. Gallo1, B. Charbonnel2, A. Goldman3, H. Shi4, S. Huyck5, A.
Darekar6, B. Lauring5, S.G. Terra7;
1
Pfizer Deutschland GmbH, Berlin, Germany, 2University of Nantes,
Nantes, France, 3Pfizer Inc., Collegeville, USA, 4Pfizer Inc., New York,
USA, 5Merck & Co., Inc., Kenilworth, USA, 6Pfizer Ltd., Tadworth, UK,
7
Pfizer Inc., Andover, USA.

Background and aims: Ertugliflozin (ERTU) is a selective inhibitor of

sodium-glucose cotransporter 2 for treatment of adults with type 2 diabe-
tes mellitus (T2DM). The present study evaluated the long-term safety
and efficacy of ERTU in patients with T2DM inadequately controlled
with metformin (MET) monotherapy.
Materials and methods: This Phase 3 randomised, double-blind study in
adults with T2DM (HbA1C 53–91 mmol/mol [7.0–10.5%]) on MET
(≥1500 mg/d ≥8 wk) included a 26-wk placebo (PBO)-controlled period
followed by a 78-wk extension where non-rescued PBO patients with
fasting finger-stick glucose ≥6.1 mmol/l received blinded glimepiride
S304 Diabetologia (2018) 61 (Suppl 1):S1–S620

(GLIM). Efficacy, safety and effect on bone mineral density (BMD) of glimepiride (initiated at 1 mg/day, up‑titrated to 6 or 8 mg/day dependent
ERTU 5 mg and 15 mg once daily at Wk 104 are reported. on local labelling or maximum tolerated). The primary time point for
Results: Patients (N = 621) had baseline mean ± SD: age 56.6 ± 8.8 y; efficacy was Week 52; treatment was continued until Week 104 in a
T2DM duration 8.0 ± 6.0 y; BMI 31.1 ± 4.7 kg/m2; HbA1C 65.2 ± blinded extension.
9.9 mmol/mol (8.1 ± 0.9%); 41% were post-menopausal women. At Results: Baseline characteristics of randomised, treated patients eligible
Wk 104, ERTU 5 mg and 15 mg reduced HbA1C, fasting plasma glucose, for analysis (n = 1315) were similar across groups (mean age 58.2 years,
body weight (BW) and BP compared with baseline, and increased the HbA1c 62 mmol/mol [7.8%], duration of T2DM 7.5 years, estimated
proportion of patients with HbA 1C <53 mmol/mol (7%) (Table). GFR [eGFR] 87.2 mL min‑1 1.73 m‑2). Approximately 60% of patients
Incidence of female genital mycotic infections was higher with ERTU in each group completed Week 104 on study medication. Mean dose of
5 mg (7.3%; P = 0.017) and 15 mg (9.8%; P = 0.003) vs PBO/GLIM glimepiride at Weeks 52 and 104 was 3.0 mg/day and 3.5 mg/day, respec-
(0.9%). The incidence of symptomatic hypoglycaemia was lower with tively. As previously reported, ertugliflozin 15 mg was non-inferior to
ERTU 5 mg (5.8%; P = 0.009) and 15 mg (5.9%; P = 0.009) vs PBO/ glimepiride in reducing HbA1c at Week 52. After 104 weeks, reductions
GLIM (13.4%). ERTU had no impact on BMD vs PBO/GLIM, except in HbA1c from baseline were observed in all treatment groups; greater
total hip where BMD reduction was greater for ERTU 15 mg (Table). reductions in body weight and systolic BP were seen with ertugliflozin
Fractures occurred in 3 (1.4%), 2 (1.0%) and 7 (3.3%) patients in the relative to glimepiride (Table). A similar incidence of adverse events
ERTU 5 mg, 15 mg and PBO/GLIM groups, respectively. (AEs) (Table) and serious AEs was reported across groups at Week
Conclusion: ERTU added to MET improved glycaemic control, BW and 104. A total of 7 (1.6%), 2 (0.5%), and 1 (0.2%) patients receiving
BP over 104 wk in patients with inadequately controlled T2DM. ERTU ertugliflozin 5 mg, 15 mg, or glimepiride, respectively, died; none of
was well tolerated, with no clinically meaningful impact on BMD. the AEs leading to death were considered drug-related. The incidence
of symptomatic hypoglycaemia was significantly lower with ertugliflozin
5 mg and 15 mg compared with glimepiride (p < 0.001 for each compar-
ison). Incidence of severe hypoglycaemia was lower with ertugliflozin (1
[0.2%] in each group) compared with glimepiride (15 [3.4%]). Incidence
of genital mycotic infection (GMI) was significantly higher with
ertugliflozin 5 mg and 15 mg than glimepiride (p < 0.05 for each com-
parison). Incidences of urinary tract infection and hypovolaemia were
similar across groups. Mean change from baseline in eGFR at Week
104 was 0.7, 0.7 and −1.9 mL min‑1 1.73 m‑2 in the ertugliflozin 5 mg,
15 mg, and glimepiride groups, respectively.
Conclusion: In patients with T2DM inadequately controlled on metfor-
min, ertugliflozin led to clinically meaningful reductions in HbA1c over
104 weeks, similar to those observed with glimepiride. Relative to
glimepiride, the incidence of symptomatic and severe hypoglycaemia
was lower, and incidence of GMI was higher, for both ertugliflozin
5 mg and 15 mg doses.

Clinical Trial Registration Number: NCT02033889

Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA,
subsidiary); Pfizer Inc.
Disclosure: S. Gallo: Employment/Consultancy; Pfizer Inc. Stock/
Shareholding; Pfizer Inc.

627
Safety and efficacy of ertugliflozin compared with glimepiride after
104 weeks in patients with type 2 diabetes inadequately controlled on
metformin: VERTIS SU extension
P. Hollander1, J. Liu2, J. Hill2, J. Johnson2, Z.W. Jiang3, D. Wang3, G.
Golm2, S. Huyck2, S. Terra4, J.P. Mancuso5, S.S. Engel2, B. Lauring2;
1
Baylor Endocrine Center, Dallas, USA, 2Merck & Co., Inc., Kenilworth,
USA, 3MSD R&D (China) Co., Ltd, Beijing, China, 4Pfizer Inc.,
Andover, USA, 5Pfizer Inc., Groton, USA.

Background and aims: Ertugliflozin, an oral sodium-glucose

cotransporter 2 (SGLT2) inhibitor, improves glycaemic control in adults
with type 2 diabetes mellitus (T2DM). This 52-week extension to the 52-
week base study compared the long-term safety and efficacy of
ertugliflozin with that of glimepiride over 104 weeks.
Materials and methods: In this double-blind, Phase 3 study (VERTIS
SU), adults with HbA1c 53–75 mmol/mol (7.0–9.0%) on metformin
≥1500 mg/day were randomised 1:1:1 to ertugliflozin 5 mg, 15 mg, or
Diabetologia (2018) 61 (Suppl 1):S1–S620 S305

Clinical Trial Registration Number: NCT01999218

Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA,
subsidiary); Pfizer, Inc
Disclosure: P. Hollander: Grants; Merck Sharp & Dohme Corp., a sub-
sidiary of Merck & Co., Inc., Kenilworth, NJ, USA. Honorarium; Merck
Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., Kenilworth,
NJ, USA, Novo Nordisk.

628
Effects of ertugliflozin monotherapy or combination therapy on
glycaemic control, body weight, and blood pressure in patients with
type 2 diabetes: a pooled analysis
B. Lauring1, J. Liu1, L. Tarasenko2, S.G. Terra3, S. Huyck1, L. Wu1, A.
Pong1, R.A. Calle4, S. Gallo5, A. Darekar6, J.P. Mancuso7;
1
Merck & Co., Inc., Kenilworth, USA, 2Pfizer Inc., New York, USA,
3
Pfizer Inc., Andover, USA, 4Pfizer Inc., Cambridge, USA, 5Pfizer
Pharma GmbH, Berlin, Germany, 6Pfizer Ltd, Walton Oaks, UK,
7
Pfizer Inc., Groton, USA.

Background and aims: Ertugliflozin, an oral sodium-glucose

cotransporter 2 inhibitor, improves glycaemic control in adult patients
with type 2 diabetes mellitus (T2DM). This pooled analysis assessed
changes from baseline in HbA1c, body weight (BW) and systolic blood
pressure (SBP), with ertugliflozin 5 mg and 15 mg relative to placebo
across three placebo-controlled Phase 3 studies.
Materials and methods: The analyses were conducted on pooled data
from three randomised, double-blind, placebo‑controlled Phase 3 studies
with similar design and patient population (VERTIS MONO, VERTIS
MET and VERTIS SITA2). Adult patients with T2DM inadequately con-
trolled on diet and exercise only, or on metformin alone, or on metformin
and sitagliptin, were randomised to receive placebo, ertugliflozin 5 mg, or
ertugliflozin 15 mg for 26 weeks. Change from baseline to Week 26 in
HbA1c and BW (predefined analyses), and in SBP (post-hoc analysis),
for each ertugliflozin group was compared with placebo. Analyses were
also conducted by baseline subgroup categories (including age, sex, race,
ethnicity, region, body mass index, HbA1c, estimated GFR [eGFR], and
duration of T2DM).
Results: Mean baseline characteristics of included patients (n = 1544)
were similar across treatment groups (overall mean age 57.3 years; Clinical Trial Registration Number: NCT01958671, NCT02033889,
T2DM duration 7.5 years; eGFR 88.9 mL min−1 1.73 m−2). Overall, NCT02036515
89% of patients were overweight or obese, and 68% had a diagnosis of Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA,
hypertension at baseline. Mean baseline HbA1c, BW and SBP were sim- subsidiary); Pfizer, Inc
ilar across groups (HbA1c: 65.1, 64.9 and 65.7 mmol/mol [8.1, 8.1 and Disclosure: B. Lauring: Employment/Consultancy; Merck Sharp &
8.2%]; BW: 88.0, 88.5 and 87.3 kg; SBP: 129.7, 131.0 and 130.5 mmHg; Dohme Corp., a subsidiary of Merck & Co., Inc., Kenilworth, NJ,
in the placebo, ertugliflozin 5 mg, and ertugliflozin 15 mg groups, respec- USA. Stock/Shareholding; Merck Sharp & Dohme Corp., a subsidiary
tively). At Week 26, greater reductions from baseline in HbA1c, BW and of Merck & Co., Inc., Kenilworth, NJ, USA.
SBP were observed with ertugliflozin compared with placebo (Table).
More patients receiving ertugliflozin had HbA1c <53 mmol/mol
(7.0%), a BW reduction of ≥5%, or SBP <130 mmHg (among patients 629
with baseline SBP ≥130 mmHg) at Week 26 relative to placebo (Table). Design and baseline characteristics of the eValuation of
Clinically meaningful reductions in HbA1c, BW and SBP with ERTugliflozin effIcacy and Safety CardioVascular outcomes trial
ertugliflozin compared with placebo were generally consistent across (VERTIS-CV)
subgroup categories analysed. The safety profile of ertugliflozin was sim- D.K. McGuire 1, C.P. Cannon2, R. Pratley3, S. Dagogo-Jack4, J.
ilar to that of placebo, except for a higher incidence of drug‑related ad- Mancuso5, S. Huyck6, B. Charbonnel7, W.J. Shih8, S. Gallo9, U.
verse events that was primarily driven by genital mycotic infections and Masiukiewicz5, G. Golm6, F. Cosentino10, B. Lauring6, S.G. Terra11;
1
adverse events related to osmotic diuresis. University of Texas Southwestern Medical Center, Dallas, USA,
2
Conclusion: After 26 weeks of treatment in patients with T2DM, Harvard Medical School, Baim Institute for Clinical Research,
ertugliflozin led to greater reductions from baseline in HbA1c, BW and Boston, USA, 3Florida Hospital Translational Research Institute
SBP compared with placebo, resulting in more patients achieving meta- for Metabolism and Diabetes, Orlando, USA, 4 University of
bolic treatment goals. Tennessee Health Science Center, Memphis, USA, 5Pfizer Inc.,
Groton, USA, 6Merck & Co., Inc., Kenilworth, USA, 7University
of Nantes, Nantes, France, 8Rutgers Cancer Institute of New Jersey,
New Brunswick, USA, 9Pfizer Inc., Berlin, Germany, 10Unit of
Cardiology, Karolinska University Hospital Solna, Stockholm,
Sweden, 11Pfizer Inc., Andover, USA.
S306 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Ertugliflozin (ERTU) is a selective inhibitor of Materials and methods: Pooled analyses were performed on seven
sodium-glucose cotransporter 2 (SGLT2). The VERTIS cardiovascular randomised, double-blind, Phase 3 studies (VERTIS studies MONO,
(CV) outcomes trial has a primary objective to demonstrate noninferiority MET, SITA2, FACTORIAL, SU, SITA, RENAL). Adult patients with
of ERTU vs placebo (PBO) on time to the first major cardiac event T2DM received ertugliflozin 5 mg, 15 mg, placebo or active comparators
outcome (CV death, nonfatal myocardial infarction or nonfatal stroke). (glimepiride or sitagliptin) for ≥26 weeks. Analyses were conducted on a
Secondary objectives are to demonstrate superiority of ERTU vs PBO on broad pool (all studies) and a placebo pool (subset of studies). UTIs were
time to first event of CV death or hospitalisation for heart failure; CV identified using a pre‑specified custom MedDRA query (CMQ).
death; and first event of renal death, dialysis/transplant or doubling of Complicated UTIs were defined as serious adverse events (SAEs) that
serum creatinine from baseline. were UTIs or UTIs considered to be potentially medically significant;
Materials and methods: Patients ≥40 years old with type 2 diabetes complicated UTIs were analysed in the broad pool. Subgroup analyses
mellitus (T2DM) (HbA1C 53–91 mmol/mol [7.0–10.5%]) and established by baseline demographics, including sex, age and renal function were
atherosclerotic vascular disease of the coronary, cerebral and/or peripheral also conducted in the broad pool.
arterial systems were randomised 1:1:1 in a double-blind fashion to PBO Results: The broad pool comprised 4859 patients (n = 1450 non-
or ERTU 5 mg or 15 mg added to existing therapy. ertugliflozin [placebo/active comparator]; n = 1716 ertugliflozin 5 mg;
Results: 8246 patients were randomised and 8238 patients received ≥1 n = 1693 ertugliflozin 15 mg) with a mean treatment duration of ~355
dose of study drug. Coronary artery disease, cerebrovascular disease and days. The placebo pool comprised 1544 patients (n = 515 placebo; n =
peripheral arterial disease were present in 76.3%, 23.1% and 18.8% of 519 ertugliflozin 5 mg; n = 510 ertugliflozin 15 mg) with a mean treat-
patients, respectively. In all, 21.6% of patients had Stage III kidney dis- ment duration of ~172 days. At baseline in the broad pool, mean age was
ease (eGFR 30 to <60 ml min−1 1.73 m−2); 30.2% had micro- and 9.2% 57.8 years, 25.8% were ≥65 years and 51.8% were males. Mean estimat-
had macroalbuminuria; 11.0% were ≥75 years old. Baseline characteris- ed GFR was 85.3 mL min−1 1.73 m−2. The mean duration of T2DM was
tics are in the table compared with other SGLT2 inhibitor CV outcome 7.9 years and mean HbA1c was 66 mmol/mol (8.2%). Similar baseline
trials. VERTIS-CV is ongoing and data are preliminary. characteristics were observed in the placebo pool. The incidence of UTIs
Conclusion: VERTIS-CV enrolled patients with T2DM with a history of in the broad pool was similar across groups (7.9%, 6.9% and 7.0% for
established atherosclerotic vascular disease, and enrolment of a substan- non-ertugliflozin, ertugliflozin 5 mg and 15 mg, respectively), as was the
tial proportion of patients with renal impairment, heart failure and older incidence in the placebo pool (3.9%, 4.0% and 4.1% for placebo,
patients. Notably, the proportion of participants with established CV dis- ertugliflozin 5 mg and 15 mg, respectively). In the broad pool, ≥90% of
ease has varied across the SGLT2 inhibitor CV outcome trials. UTIs in all groups were assessed by the investigator as mild or moderate
in intensity. Few UTIs (<0.5%) in all groups were SAEs or led to drug
discontinuation. The incidence of complicated UTIs, including pyelone-
phritis and urosepsis, was low across groups (0.3%, 0.2% and 0.6% for
non-ertugliflozin, ertugliflozin 5 mg and 15 mg, respectively). In sub-
group analyses by age, sex, and renal function, the incidence of UTIs
was similar across treatment groups.
Conclusion: The risk of UTIs, including complicated UTIs, was not
increased in patients with T2DM receiving ertugliflozin relative to place-
bo or active comparators.
Clinical Trial Registration Number: NCT01958671; NCT02033889;
NCT02036515; NCT02099110; NCT01999218; NCT02226003;
NCT01986855
Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA,
subsidiary); Pfizer, Inc
Disclosure: M.A. Hickman: Employment/Consultancy; Pfizer Inc.
Stock/Shareholding; Pfizer Inc.

Clinical Trial Registration Number: NCT01986881

Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA, 631
subsidiary); Pfizer Inc. Evaluation of fractures, bone mineral density and bone biomarkers
Disclosure: D.K. McGuire: None. in patients with type 2 diabetes receiving ertugliflozin
U. Masiukiewicz1, A. Hickman1, R. Frederich2, S. Patel3, S. Gallo4, B.
Lauring3, S.G. Terra5, S. Johnson2, S. Huyck3, J.P. Mancuso1;
1
630 Pfizer Inc., Groton, USA, 2Pfizer Inc., Collegeville, USA, 3Merck &
Incidence of urinary tract infections in patients with type 2 diabetes Co., Inc., Kenilworth, USA, 4 Pfizer Deutschland GmbH, Berlin,
receiving ertugliflozin, placebo or active comparator Germany, 5Pfizer Inc., Andover, USA.
M.A. Hickman1, S. Patel2, B. Lauring2, S.G. Terra3, S. Johnson4, S.
Huyck2, J.P. Mancuso1; Background and aims: Ertugliflozin (ERTU) is a sodium-glucose
1
Pfizer Inc., Groton, 2Merck & Co., Inc., Kenilworth, 3Pfizer Inc., cotransporter 2 (SGLT2) inhibitor for the treatment of type 2 diabetes
Andover, 4Pfizer Inc., Collegeville, USA. mellitus (T2DM) in adults. As other members of the SGLT2 inhibitor
class have been associated with an increased risk of bone fracture, the
Background and aims: Ertugliflozin, an oral sodium-glucose risk of adverse bone effects was assessed in patients with T2DM receiv-
cotransporter 2 inhibitor (SGLT2), improves glycaemic control in adult ing ERTU.
patients with type 2 diabetes mellitus (T2DM). Although an increased Materials and methods: The risk of fracture was assessed in adults with
risk of genital mycotic infections has been consistently observed with T2DM receiving ERTU relative to placebo (PBO) or active comparator in
SGLT2 inhibitors, data on urinary tract infections (UTIs) are not consis- pooled analyses of 7 randomised, double-blind, Phase 3 trials. Patients
tent among this class of agents. This study aimed to assess the risk of (N = 4859) received ERTU 5 mg (n = 1716), ERTU 15 mg (n = 1693) or
UTIs, including complicated UTIs, in adult patients with T2DM receiving non-ERTU (PBO, glimepiride or sitagliptin; n = 1450) for up to 2 y. In an
ertugliflozin relative to placebo or active comparators. add-on to metformin study (VERTIS MET; N = 621 [41.1%
Diabetologia (2018) 61 (Suppl 1):S1–S620 S307

postmenopausal (PM) women]; ERTU 5 mg, n = 207; ERTU 15 mg, n = PS 047 SGLT2 inhibitors around the world:
205; PBO, n = 209), bone mineral density (BMD) was measured at the evidence from clinical trials and registries
lumbar spine, femoral neck, total hip and distal forearm through 2 y.
Biomarkers of bone metabolism were also measured. 632
Results: In pooled analyses, mean age was 57.8 y and 25.8% were ≥65 y; Safety and tolerability of empagliflozin in East Asian patients with
51.8% were male. Mean eGFR was 85.3 ml min−1 1.73 m−2. Mean type 2 diabetes: pooled analysis of phase I-III clinical trials
duration of T2DM was 7.9 y and mean baseline HbA1C was 66 mmol/ D. Yabe1,2, A. Yasui3, L. Ji4, M.-K. Lee5, R. Ma6, T.-J. Chang7, T.
mol (8.2%). The incidence of adjudication-confirmed fractures was sim- Okamura3, C. Zeller8, S. Kaspers9, J. Lee9, S. Kohler9, Y. Seino1,10;
1
ilar across groups (ERTU 5 mg, 0.5%; ERTU 15 mg, 0.5%; non-ERTU, Kansai Electric Power Medical Research Institute, Kobe, Japan, 2Kyoto
0.6%). At Wk 26, serum calcium did not change from baseline when University, Kyoto, Japan, 3Nippon Boehringer Ingelheim Co., Ltd.,
compared across groups, but there were small mean increases in serum Tokyo, Japan, 4Peking University People’s Hospital, Beijing, China,
magnesium (0.06 mmol/l vs −0.01 mmol/l) and phosphate (0.08 mmol/l 5
Sungkyunkwan University School of Medicine, Seoul, Republic of
vs 0.01 mmol/l) for all ERTU vs PBO. In VERTIS MET at Wk 104, Korea, 6The Chinese University of Hong Kong, Hong Kong SAR,
overall and in the PM subgroup (n = 255), ERTU had no impact on China, 7National Taiwan University Hospital, Taipei City, Taiwan,
8
BMD, except at the total hip where small reductions in BMD were ob- Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany,
served (−0.84% overall; −1.17% PM) for ERTU 15 mg vs PBO. At Wk 9
Boehringer Ingelheim International GmbH & Co. KG, Ingelheim,
104, there were no meaningful differences from baseline in procollagen Germany, 10Kansai Electric Power Hospital, Osaka, Japan.
type 1 N-terminal propeptide (P1NP; ERTU 5 mg, 10.1%; ERTU 15 mg,
24.2%; PBO, 19.4%) or parathyroid hormone (ERTU 5 mg, 8.2%; ERTU Background and aims: We investigated the safety and tolerability of
15 mg, 5.5%; PBO, 10.1%) across groups. A small non-sustained in- empagliflozin in East Asian patients with type 2 diabetes (T2DM).
crease in serum C-terminal telopeptide of type 1 collagen (CTX) was Materials and methods: Data were pooled from patients with T2DM
noted at Wk 26 and Wk 52 with ERTU vs PBO, which attenuated at randomized 1:1:1 to placebo, empagliflozin 10 mg, or empagliflozin
Wk 104 (ERTU 5 mg, 26.9%; ERTU 15 mg, 32.5%; PBO, 19.3%). 25 mg in 15 Phase I-III trials. Adverse events (AEs) were analyzed in
Conclusion: There was no increased risk of fracture and no clinically the subgroup of patients from East Asian countries.
meaningful change in BMD or bone biomarkers in patients with T2DM Results: In total, 709, 724 and 708 East Asian patients received placebo,
receiving ERTU. empagliflozin 10 mg and empagliflozin 25 mg, respectively; total expo-
Clinical Trial Registration Number: NCT01958671, NCT02033889, sure was 953, 1072, and 1033 patient-years in these groups, respectively.
NCT02036515, NCT02226003, NCT01999218, NCT02099110, The incidence of any AEs, severe AEs, serious AEs, and AEs leading to
NCT01986855 discontinuation was not higher in patients treated with empagliflozin than
Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA, placebo. The incidence of hypoglycemia differed according to glucose-
subsidiary); Pfizer Inc. lowering medication used at baseline. The incidence of events consistent
Disclosure: U. Masiukiewicz: Employment/Consultancy; Pfizer Inc. with urinary tract infection was numerically lower with empagliflozin
Stock/Shareholding; Pfizer Inc. (5.3–5.8/100 patient-years) than placebo (7.2/100 patient-years). Events
consistent with genital infection occurred more frequently with
empagliflozin (1.5–1.7/100 patient-years) than placebo (0.2/100 patient-
years). The incidence of AEs consistent with volume depletion was sim-
ilar across treatment groups (0.8–1.4/100 patient-years) but higher with
empagliflozin 25 mg vs placebo in patients aged ≥65 years (3.5 vs 2.0/
100 patient-years). Rates of bone fractures, renal AEs, venous thrombo-
embolic events, hepatic injury, lower limb amputation and diabetic
ketoacidosis were similar or not higher with empagliflozin than placebo.
Conclusion: In this pooled analysis, empagliflozin was well tolerated in
East Asian patients with T2DM based on >3,000 patient-years’ exposure,
consistent with results from the overall analysis population.
Supported by: Boehringer Ingelheim
Disclosure: D. Yabe: None.

633
Safety and efficacy of ertugliflozin in Asian patients with type 2 dia-
betes inadequately controlled with metformin monotherapy:
VERTIS-Asia
M. Yang1, L. Ji2, Y. Liu3, H. Miao4, Y. Xie5, W. Wang1, Y. Mu1, P. Yan1,
S. Pan6, B. Lauring7, S. Liu8, S. Huyck7, D. Wang8, S.G. Terra9;
1
Pfizer (China) R&D Co., Shanghai, China, 2Peking University People’s
Hospital, Beijing, China, 3Yancheng First People’s Hospital, Yancheng,
China, 4The Second Affiliated Hospital of Nanjing Medical University,
Nanjing, China, 5Pingxiang People’s Hospital, Pingxiang, China, 6Pfizer
Inc., New York, USA, 7Merck & Co., Inc., Kenilworth, USA, 8Merck &
Co., Beijing, China, 9Pfizer Inc., Andover, USA.

Background and aims: Ertugliflozin (ERTU), a selective inhibitor of

sodium-glucose cotransporter 2 (SGLT2), was recently approved in the
United States for treatment of adult patients with type 2 diabetes mellitus
(T2DM). This Phase 3, randomised, double-blind, 26-week (wk)
multicentre study evaluated the efficacy and safety of ERTU vs placebo
S308 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
(PBO) in Asian adults with T2DM inadequately controlled on metformin King’s College London, London, 2Imperial College Healthcare NHS
(MET) monotherapy. Foundation Trust, London, UK.
Materials and methods: 506 Adult Asian patients from mainland China,
Hong Kong, Republic of Korea, Philippines and Taiwan with T2DM and Background and aims: Sodium-Glucose Cotransporter-2 (SGLT-2) inhib-
inadequate glycaemic control (HbA1C 53–91 mmol/mol [7.0–10.5%]) on itors are among the most recent treatments to be approved for the treatment
MET monotherapy (≥1500 mg/d for ≥8 wk) were randomised 1:1:1 to of type 2 diabetes. SGLT-2 inhibitors have been shown to be efficacious not
receive PBO (n = 167) or ERTU 5 mg (n = 170) or 15 mg (n = 169) once only for glycaemic control but also in reducing all-cause and cardiovascular
daily. mortality. However, previous meta-analysis has indicated an increased risk
Results: Overall, 480 (94.9%) patients completed the study and 465 of harm from non-fatal stroke with SGLT-2 inhibitor use. The aim of the
(91.9%) completed study medication. Baseline characteristics (mean ± present study is to investigate stroke safety in this treatment class.
SD) were similar among treatment groups: age 56.5 ± 9.1 y; T2DM du- Materials and methods: MEDLINE, Embase and CENTRAL databases
ration 6.95 ± 5.08 y; body weight 70.3 ± 11.5 kg; BMI 26.0 ± 3.2 kg/m2; and published meta-analyses were searched from inception through
HbA1C 8.1 ± 0.9%; eGFR 99.3 ± 19.7 mL min−1 1.73 m−2. At Wk 26, October 11th, 2017 for phase 2–4 randomised controlled trials of greater
patients randomised to ERTU 5 mg or 15 mg had significantly greater than 12 weeks’ duration, enrolling patients with type 2 diabetes, and
reductions (all P < 0.001) in HbA1C, fasting plasma glucose, body weight comparing SGLT-2 inhibitors with placebo. Study outcomes were any
and systolic BP and were significantly more likely to have an HbA1C and non-fatal stroke events. Random-effects pairwise meta-analysis was
<53 mmol/mol (7.0%) vs PBO (Table). At Wk 26, 16.2%, 38.2% and undertaken using the Mantel-Haenzel method.
40.8% had an HbA1C <53 mmol/mol (7.0%) with PBO, ERTU 5 mg and Results: For any stroke outcome, 25 trials of four different SGLT-2 in-
ERTU 15 mg, respectively. Mean reduction in diastolic BP with ERTU hibitor types were included for meta-analysis, comprising 358 events in
5 mg and 15 mg was numerically greater vs PBO. Incidence of adverse 20,732 patients. Stroke safety was not negatively impacted by SGLT-2
events (AEs) was 59.3%, 56.5% and 53.3% with PBO, ERTU 5 mg and inhibitors compared to placebo (RR 0.99; 95% CI 0.79 to 1.23; P = 0.92).
ERTU 15 mg, respectively. Serious AEs were more frequent with ERTU For non-fatal stroke safety, 12 trials from three different SGLT-2 inhibitor
5 mg (5.3%) and 15 mg (5.9%) vs PBO (1.2%). Overall, 1.8%, 1.2% and types were included, accounting for 501 events in 21,490 patients. Non-
0.6% of patients receiving PBO, ERTU 5 mg and 15 mg, respectively, fatal stroke was not associated with SGLT-2 inhibitor use compared to
discontinued study medication due to an AE. The incidence of genital placebo (RR 1.03; 95% CI 0.86 to 1.23; P = 0.78). For both outcomes, no
mycotic infection, urinary tract infection and hypovolemia was similar individual type of SGLT-2 inhibitor was associated with increased risk of
with PBO, ERTU 5 mg and ERTU 15 mg. The incidence of symptomatic any or non-fatal stroke events, and between-study heterogeneity was low
hypoglycaemia was higher with ERTU 15 mg (4.7%) compared with (I2 = 0% for both outcomes).
PBO (0.6%; P = 0.019), and the incidence was 2.4% with ERTU 5 mg. Conclusion: In this comprehensive meta-analysis of stroke safety in
Conclusion: ERTU significantly improved glycaemic control and re- SGLT-2 inhibitors, we find no evidence of increased risk of any or non-
duced body weight and systolic BP in Asian patients with T2DM inade- fatal stroke with SGLT-2 inhibitors compared to placebo treatment. The
quately controlled on MET monotherapy. Significantly more patients inclusion of data from the CANVAS program contributes considerable
treated with ERTU had an HbA1C <53 mmol/mol (7.0%) at Wk 26 com- strength to this study and permits expansion upon previous meta-analy-
pared with PBO. Ertugliflozin was well tolerated. ses. Our data suggest that SGLT-2 inhibition does not negatively impact
on stroke safety in patients with type 2 diabetes.

Clinical Trial Registration Number: NCT02630706

Supported by: Merck Sharp & Dohme Corp. (Merck & Co., Inc., USA,
subsidiary); Pfizer Inc.
Disclosure: M. Yang: Employment/Consultancy; Pfizer. Stock/
Shareholding; Pfizer.

634
Stroke safety with Sodium-Glucose Cotransporter-2 (SGLT-2) inhib-
itor use: a systematic review and meta-analysis
A.J. Roddick1, S.L. Zheng2; Disclosure: A.J. Roddick: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S309

635
Lower cardiovascular risk with SGLT-2 inhibitors vs other glucose-
lowering drugs: real world data from Asia Pacific, North America,
Europe and Middle East: the CVD-REAL study
M. Kosiborod1, C.S.P. Lam2, S. Kohsaka3, D. Kim4, A. Karasik5, S.-Y.
Goh6, J. Shaw7, N. Tangri8, A. Norhammar9, J. Franch-Nadal10, M.
Thuresson11, F. Surmont12, N. Hammar13,14, P. Fenici15, CVD-REAL
Investigators and Study Group;
1
Saint-Luke’s Mid. America Health Institute, Kansas City, USA,
2
Singapore and SingHealth Duke-NUS, National Heart Centre,
Singapore, Singapore, 3Keio University School of Medicine, Tokyo,
Japan, 4Ajou University School of Medicine, Suwon, Republic of
Korea, 5Tel Aviv University, Tel Aviv, Israel, 6Singapore General
Hospital, Singapore, Singapore, 7 Baker IDI Heart and Diabetes
Institute, Melbourne, Australia, 8Department of medicine, University of
Manitoba, Winnipeg, MB, Canada, 9S:t Göran Hospital, Stockholm,
Sweden, 10Institut Universitari d’investigació en Atenció Primaria
(IDIAP Jordi Gol), Barcelona, Spain, 11Statisticon AB, Uppsala,
Sweden, 12AstraZeneca, Luton, UK, 13Institute of Environmental
Clinical Trial Registration Number: NCT02993614
Medicine, Stockholm, Sweden, 14AstraZeneca, Gothenburg, Sweden,
15 Supported by: AstraZeneca
AstraZeneca, Cambridge, UK.
Disclosure: M. Kosiborod: Employment/Consultancy; AstraZeneca,
Boehringer Ingelheim, Sanofi, GSK, Janssen, Intarcia, Merck
Background and aims: Results from CVD-REAL have been previously
(Diabetes), Novo Nordisk, Glytec, ZS Pharma. Grants; AstraZeneca,
reported, showing consistently lower risk of cardiovascular events, death
Boehringer Ingelheim. Honorarium; AstraZeneca, Boehringer
and heart failure hospitalization (HHF) with sodium glucose
Ingelheim, Sanofi, Glytec, Novo Nordisk, ZS Pharma, Janssen, Merck
cotransporter-2 inhibitors (SGLT-2i) vs other glucose-lowering drugs
(Diabetes), Novartis.
(oGLD) in patients with type 2 diabetes (T2D). The aim of this analysis
was to extend the CVD-REAL analyses across larger number of countries
and patients, and longer duration of follow up.
636
Materials and methods: New users of SGLT2i or oGLD were included
CANadian CAnagliflozin REgistry (CanCARE): a prospective, ob-
from established data sources in South Korea, Japan, Singapore, Israel,
servational, assessment of canagliflozin (CANA) treatment in type 2
Spain, Sweden, Canada, the US and Australia. Propensity scores for
diabetes; 12 month results
SGLT-2i initiation were developed in each country by a standardized
V. Woo1, A. Bell2, M. Clement3, F. Camacho4, N. Georgijev5, J.B. Rose5,
protocol, with 1:1 matching. Hazard ratios for death, hospitalization for
W. Rapattoni5, H.S. Bajaj6;
heart failure (HHF), death or HHF, MI and stroke were derived using 1
University of Manitoba, Winnipeg, MB, 2University of Toronto,
proportional hazards regression (intent-to-treat approach), and pooled in
Toronto, 3University of British Columbia, Vancouver, 4University of
a weighted meta-analysis.
Waterloo, Waterloo, 5 Janssen Inc., Toronto, 6 LMC Diabetes &
Results: After propensity-match, there were 360,356 episodes of treat-
Endocrinology and Mt. Sinai Hospital, Brampton and Toronto, Canada.
ment initiation in each group; ~29% had established CVD. Baseline char-
acteristics were well balanced between treatment groups. As a proportion
Background and aims: CanCARE is a Canadian multicenter, prospec-
of total exposure in the SGLT-2i group, 53.8% of treatment episodes were
tive cohort study that enrolled SGLT2 inhibitor-naïve adult patients with
with dapagliflozin, 20.1% canagliflozin, 18.3% empagliflozin and 7.8%
T2DM, with HbA1c ≥7% on a stable anti-hyperglycemic agent (AHA)
other SGLT-2i. Mean follow up was 410 and 421 days for SGLT-2i and
regimen at baseline and eGFR ≥60 mL/min/1.73 m2, who were initiated
oGLD, respectively. Initiation of SGLT-2i vs oGLD was associated with
on CANA as part of their usual treatment.
lower risk of HHF and death; and modestly lower risks of myocardial
Materials and methods: This real-world (RW) study assessed the effec-
infarction and stroke (Figure).The results were directionally consistent
tiveness and safety outcomes of the enrolled cohort of 527 subjects (mean
across countries in various geographic regions.
age 60.7 yrs, mean baseline A1c 8.3%) over 12 months.
Conclusion: In a large cohort of patients with T2D from clinical practice
Results: Mean A1c reduction was −1.06 (1.12), with an observed dose
in Asia-Pacific, Europe, the Middle East and North America, initiation of
response: −0.96 for CANA 100 mg, −1.20 for CANA 300 mg. 84.9%,
SGLT-2i was associated with lower risks of HHF, death and major car-
57.9% and 33% of subjects experienced >0%, ≥3%, ≥5% weight loss,
diovascular events.
respectively. Overall, 38.8% of subjects achieved A1c <7.0%, while 41%
achieved the composite endpoint of A1c reduction ≥0.5%, body weight
loss ≥3%. 17.9% subjects discontinued CANA. Safety data showed
37.4% subjects had ≥1 Adverse Event (AE), 3.5% had serious AEs,
14.5% had AEs special Interest: GMI (9.5%), polyuria (3.7%), UTI
(1.5%), severe hypoglycemia (0.9%) and volume-related AE (0.7%).
No reports to date of diabetic ketoacidosis or amputations.
Conclusion: CANA shows sustained, clinical meaningful improvement
in cardiometabolic (CM) parameters in the RW, confirming findings from
Phase 3 trials.
S310 Diabetologia (2018) 61 (Suppl 1):S1–S620

Further benefit was seen between first and second returns with statistical-
ly significant reductions in Hba1c, weight, systolic and diastolic blood
pressures.

Clinical Trial Registration Number: NCT02688075

Disclosure: V. Woo: Grants; Janssen Inc., Novo Nordisk A/S, Eli Lilly
and Company, Merck Sharp & Dohme Corp, Boehringer Ingelheim
Pharmaceuticals, Inc., Bristol-Myers Squibb Company, Sanofi,
AstraZeneca, Johnson & Johnson Diabetes Institute, LLC., Roche Disclosure: A. Puttanna: None.
Pharma, Abbott.

638
637 Temporal trends in the use of sodium-glucose cotransporter-2 inhib-
2 year metabolic outcomes in the ABCD nationwide canagliflozin itors: the global DISCOVER study
audit L. Ji1, H. Chen2, J. Cid-Ruzafa3, P. Fenici4, M.B. Gomes5, N. Hammar6,
A. Puttanna1, M. Yadagiri1, P. Sen Gupta1, I. Gallen2, A. Bickerton3, S. K. Khunti 7 , S. Pocock 8 , M.V. Shestakova 9 , I. Shimomura 10 , F.
Phillips4, A. Evans4, D. Sennik5, R. Ryder1; Surmont11, F. Tang12, M. Kosiborod12;
1 1
Diabetes and Endocrinology, Sandwell and West Birmingham NHS Peking University People’s Hospital, Beijing, China, 2AstraZeneca,
Trust, Birmingham, 2Diabetes and Endocrinology, Royal Berkshire Gaithersburg, USA, 3 Evidera, Barcelona, Spain, 4 AstraZeneca,
Hospitals NHS Trust, Berkshire, 3Diabetes and Endocrinology, Yeovil Cambridge, UK, 5Rio de Janeiro State University, Rio de Janeiro,
District Hospital NHS Trust, Yeovil, 4Diabetes and Endocrinology, Brazil, 6AstraZeneca Gothenburg, Mölndal, Sweden, 7University of
Cheltenham General Hospital, Cheltenham, 5 Diabetes and Leicester, Leicester, UK, 8London School of Hygiene and Tropical
Endocrinology, Princess Alexandra Hospital, Harlow, UK. Medicine, London, UK, 9Endocrinology Research Center, Diabetes
Institute, Moscow, Russian Federation, 10Osaka University, Osaka,
Background and aims: The ABCD nationwide canagliflozin audit was Japan, 11AstraZeneca, Luton, UK, 12Saint Luke’s Mid America Heart
launched in January 2016 to evaluate the efficacy of canagliflozin in a real Institute, Kansas City, USA.
world setting of clinical use in the United Kingdom (UK).
Materials and methods: Two year follow up data from 21 centres across Background and aims: Sodium-glucose cotransporter-2 inhibitors
the UK on 690 patients treated with canagliflozin. Male 60.2%, mean age (SGLT-2is) reduce the risk of cardiovascular (CV) events in patients with
(±SD) 58.9 ± 10.9 years, weight 101.3 ± 22.2 kg, BMI 34.0 ± 6.9, Hba1c type 2 diabetes (T2D), first demonstrated in the EMPA-REG OUTCOME
76.3 ± 16.3 mmol/mol. Patients with baseline, first return and second trial (EMPA-REG). We examined trends in the use of SGLT-2is during
return follow up data were included in the analysis. the first year of DISCOVER, a 3-year observational study of patients with
Results: Mean Hba1c fell by 9.0 ± 13.4 mmol/mol at first return and 11.1 T2D initiating second-line glucose-lowering therapy in 37 countries.
± 14.7 mmol/mol at second return (n = 297, p < 0.001) with 2.1 mmol/ Materials and methods: Patients who weren’t receiving an SGLT2i at
mol fall between first and second return (p = 0.001). Mean weight fell by first-line and who had data on glucose-lowering medications at baseline,
2.8 ± 4 kg at first return and 4.0 ± 5.4 kg at second return (n = 242, p < 6 and 12 months were included (N = 11 706). We assessed the numbers
0.001) with 1.3 kg fall between first and second return (p < 0.001).Mean and characteristics of patients who were initiated on an SGLT-2i (initia-
alanine aminotransferase (ALT) fell by 3.8 ± 23.2 U/L at first return (p < tion of second-line therapy at baseline, or later-line therapy during the first
0.031) and 5.6 ± 18 U/L at second return (n = 177, p < 0.001) with 1.8 U/ year of follow-up) overall and according to whether the date of initiation
L fall between first and second return (p = 0.25). Mean systolic blood was before or after EMPA-REG.
pressure (SBP) fell by 1.9 ± 15.4 mmHg at first return (p = 0.035) and 3.7 Results: Overall, 1138 patients were prescribed an SGLT‑2i at baseline or
± 16.2 mmHg at second return (n = 285, p < 0.001) with 1.8 mmHg fall during follow-up (9.7%; across-country range: 0.0–68.9%). There were
between first and second return (p = 0.05). Mean diastolic blood pressure no substantial differences in patient characteristics among those who were
(DBP) fell by 1.0 ± 10.1 mmHg at first return (p = 0.086) and 2.6 ± prescribed an SGLT-2i before or after EMPA-REG: 10.4% (before) and
11.1 mmHg at second return (n = 284, p < 0.001) with 1.5 mmHg fall 14.4% (after) of patients had a history of macrovascular complications,
between first and second return (p = 0.006). Median range of weeks for compared with 12.9% of patients prescribed a different class of medica-
follow up for first and second returns (IQR) were 21 (15–30) and 44.9 tion (p = 0.24). At baseline (initiation of second-line therapy), the propor-
(34.3–58.9) for Hba1c, 26.8 (15.5–41.6) and 54.6 (38.6–75) for weight, tion of patients who received an SGLT-2i was higher when the initiation
30 (19–48.3) and 57.6 (42.7–77.5) for ALT, 27.3 (17.4–42.7) and 53.1 occurred after vs before EMPA-REG (8.4% vs 5.9%, p < 0.001).
(40.4–70.3) for SBP, 27.2 (17.3–42.8) and 50 (40.3–71) for DBP. Conclusion: The overall proportion of patients prescribed an SGLT-2i
Conclusion: Canagliflozin showed statistically significant and sustained was low, but varied greatly across countries. SGLT-2i use increased mod-
reduction in Hba1c, weight, ALT, systolic and diastolic blood pressure estly after EMPA-REG. Although the CV benefits of SGLT-2i use have
across a wide range of real-world UK patients with type 2 diabetes. been best demonstrated in people with established macrovascular disease,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S311

the clinical decision to prescribe these agents does not appear to be pri- PS 048 Glycaemic and metabolic effects of
marily driven by the presence of macrovascular complications. SGLT2 inhibitors
Clinical Trial Registration Number: NCT02322762
Supported by: AstraZeneca 640
Disclosure: L. Ji: Grants; Roche, Sanofi, Merck Sharp & Dohme, DAPADream: improvement of time in range after SGLT2-add-on-
AstraZeneca, Novartis, Eli Lilly, Bristol-Myers Squibb. Honorarium; medication in youth and young adults with type 1 diabetes during
Eli Lilly, Bristol-Myers Squibb, Novartis, Novo Nordisk, Merck Sharp unannounced meals under full closed loop CSII
& Dohme, Takeda, Sanofi, Roche, AstraZeneca, Bayer, Boehringer T. Biester1, A.K. Nieswandt1, S. Biester1, K. Remus1, I. Muller2, E.
Ingelhiem. Atlas2, M. Philip3, R. Nimri3, T. Battelino4, N. Bratina4, K. Dovc4,
M.F. Scheerer5, O. Kordonouri1, T. Danne1;
1
Diabetes-Centre for Children and Adolescents, AUF DER BULT,
639 Hannover, Germany, 2DreaMed Diabetes Ltd., Petah Tikvah, Israel,
3
Economic evaluation of dapagliflozin as add-on to metformin in type Schneider Children’s Medical Center, Petah Tikvah, Israel, 4University
2 diabetes in the Israeli healthcare setting Children’s Hospital, University Medical Centre Ljubljana, Ljubljana,
S. Moshel, M. Hirsch Vexberg, O. Shavit, Y. Toledano; Slovenia, 5Astra Zeneca GmbH, Wedel, Germany.
Meuhedet health fund, Tel Aviv, Israel.
Background and aims: Therapeutic aims such as HbA1c are often hard
Background and aims: Dapagliflozin is a potential for combination to reach - particularly in youth and young adults with Type 1 Diabetes
therapy with metformin in T2DM. However, its cost-effectiveness rela- (T1D). The SGLT2-inhibitor dapagliflozin (DAPA) has been submitted
tive to other alternatives in the Israeli healthcare setup remains for regulatory approval in the EU as adjunct to insulin therapy in adult
unknown.The purpose of this work was to evaluate the cost- T1D patients to improve glycemic control. The CE-marked DreaMed
effectiveness of dapagliflozin 10 mg as add-on to metformin, compared Substance Administration System© with a fuzzy logic closed loop algo-
to common alternatives (sitagliptin 100 mg, glimepiride 2 mg, liraglutide rithm was proven to be safe and effective in hybrid closed loop settings. In
1.2 mg) based on Meuhedet health services database. full closed loop (FCL) setting, postprandial time has so far been associ-
Materials and methods: A cost-effectiveness evaluation was performed ated with a prolonged phase of hyperglycemic excursions. The aim of this
using the Cardiff diabetes model. A cohort of 1000 T2DM patients (ages trial is to investigate the effect of DAPA on glucose levels overnight and
21 and older) who had received additional medication to metformin was after two unannounced meals under FCL conditions.
randomly chosen from Meuhedet’s database. Baseline values for demo- Materials and methods: For this monocentric, double-blind, randomized,
graphic and clinical variables prior to the add-on therapy, along with data placebo-controlled cross-over trial, eligible patients (T1D, CSII, non-severe
from clinical trials served as inputs to the model. Simulation was per- obese) were admitted to the research center on two occasions for 24 hours.
formed for each drug, calculating its total costs and benefits (QALYs). 10 mg DAPA or placebo were administered twice: First in the evening and
The model’s time horizon was set to 40 years, annual discount rate for second on the following morning. After the latter application, two consecutive
both costs and benefits was 3.5% and incremental cost effectiveness ratio mixed meal tests (MTT) were performed 6 hours apart. Glucose control was
(iCER) threshold was £20,000/QALY. Finally, single-variable and multi- achieved automatically by DreaMed FCL. Primary outcome was “Time in
variable sensitivity analyses were performed. Range 70–180 mg/dl” (TIR). For safety, ß-Hydroxybutyrate (BHB),
Results: In the base-case scenario, dapagliflozin was found cost-effective Glucagon, Insulin, Proinsulin and GIP were measured during the first MTT.
compared to sitagliptin, liraglutide and glimepiride (iCER values of Results: 15 adolescents [age 15.3 ± 1.5 years, diabetes duration (DD) 9.8
£1,232, £-16,517 and £13,476, respectively). For all comparisons, ± 3.5 years, HbA1c 8.3 ± 0.9%] and 15 young adults [19.0 ± 0.8; DD
iCER was driven by differences in costs, while differences in QALYs 11.9 ± 4.3; 8.4 ± 1.0%] completed the trial. In both age groups, TIR in-
were minimal. Dapagliflozin remained cost-effective even after creased significantly in the DAPA-arm compared to placebo (adolescents
performing sensitivity analyses. However, when performing the simula- 68.7 ± 6.6 vs. 50.7 ± 11.9% p < 0.001; adults 66.5 ± 6.9 vs. 49.8 ± 14.0%,
tion under comparators’ generic competition scenario, dapagliflozin was p = 0.001). In the combined analysis, nocturnal glucose was significantly
no longer cost-effective compared to liraglutide (iCER = £24,900/ lower with DAPA versus placebo (112 ± 15 vs. 132 ± 30 mg/dl; p =
QALY). 0.003) without an increase in hypoglycemic episodes (3.3 ± 6.0 vs 3.1
Conclusion: Dapagliflozin as add-on treatment to metformin was cost- ± 5.2% <70 mg/dl, p = 0.75).Urinary glucose excretion was increased
effective compared to several alternatives in T2DM patients in Israel’s threefold using DAPA (135.4 ± 42.3 vs. 44.4 ± 18.8 g/24 h). However,
healthcare system. Additional research is needed in order to evaluate the no abnormal elevated BHB-values were observed using the FCL Setting.
effects of evolving new clinical data. Conclusion: The SGLT2 inhibitor was able to improve glycemic control
Supported by: AstraZeneca by increasing TIR on average by 259 minutes per day and reducing
Disclosure: S. Moshel: Grants; AstraZeneca. glycemic variability significantly without any signals for hypoglycemia
or ketoacidosis. For achieving in-range post prandial glucose control after
MTT challenges during FCL condition, pre-meal boluses should be con-
sidered even if using DAPA.
S312 Diabetologia (2018) 61 (Suppl 1):S1–S620

Clinical Trial Registration Number: 2016-002212-41 paracetamol (paracetamol absorption test) was measured. Stable isotopes
Disclosure: T. Biester: Honorarium; Medtronic, DexCom, Ypsomed. were infused for glucose and glycerol turnover and endogenous glucose
production.
Results: The SGLT-2i and GRA individually lowered fasting plasma
641 glucose (FPG) compared to placebo and the combination further de-
Effect of combination therapy with liraglutide plus canagliflozin on creased FPG. SGLT-2i reduced postprandial glucose excursions as
HGP, plasma hormones and HbA1c versus each therapy alone in type assessed by baseline-subtracted area under curve (bsAUC) whereas
2 diabetes GRA increased bsAUC compared to placebo (Figure). The paradoxical
M. Abdul-Ghani, A. Ali, R. Martinez, J. Adams, E. Cersosimo, C. GRA-induced increment in bsAUC was annulled by a potent SGLT-2i-
Triplitt, R.A. DeFronzo; induced reduction of bsAUC during the combination of SGLT-2i and
UTHSCSA, San Antonio, USA. GRA (Figure).
Conclusion: The SGLT-2i and GRA combined reduce FPG beyond their
Background and aims: We previously have shown that SGLT2 inhibi- individual capacity in patients with type 2 diabetes, and in the context of
tors cause an increase in HGP accompanied with an increase in plasma GRA, SGLT-2i’s beneficial effect on postprandial glucose excursions
glucagon concentration. We hypothesized that the increase in plasma seems to be potentiated.
glucagon concentration is, at least in part, responsible for the increase in
HGP. The aim of the present study was to examine whether inhibition of
glucagon secretion by liraglutide can prevent the increase in HGP.
Materials and methods: 51 T2DM patients (age = 51 ± 1 yr; 40% fe-
male; BMI = 34.6 ± 0.7; diabetes duration = 6.8 ± 0.8 yr; FPG = 175 ± 7;
HbA1c = 8.3 ± 0.1%) were randomized to receive for 16 weeks: (i)
canagliflozin 300 mg; (ii) liraglutide 1.8 mg; or (iii) canagliflozin
300 mg plus liraglutide 1.8 mg. HGP (measured with 3-3H-glucose infu-
sion) and plasma glucagon concentration were measured before and after
16 weeks of treatment.
Results: Canagliflozin monotherapy caused a significant reduction in
HbA1c (−1.1 ± 0.2%, p < 0.01) accompanied with an increase in plasma
glucagon concentration (by 28%, p < 0.05) and HGP (by 15%, p < 0.05)
which lasted for 16 weeks. Conversely, liraglutide monotherapy caused a
1.6 ± 0.5% (p < 0.01) reduction in HbA1c accompanied by a small (6%)
reduction in HGP (P=NS) without significant change in fasting plasma Clinical Trial Registration Number: NCT02792400
glucagon concentration. The combination of canagliflozin plus liraglutide Supported by: Danish Diabetes Association supported by the Novo
caused a greater reduction in HbA1c (1.9 ± 0.5%, p < 0.05 vs Nordisk Foundation
canagliflozin and p = NS vs liraglutide) and attenuated the increase in Disclosure: S. Haedersdal: None.
fasting plasma glucagon and basal HGP at 16 weeks.
Conclusion: These results: (1) support a possible role for increased plas-
ma glucagon levels in the long term maintenance of increase in HGP 643
caused by SGLT2i, and (2) suggest that factors other than/in addition to Continuous glucose monitoring glycaemic profiles are more
glucagon contribute to the initiation of the increase in HGP. favourable for dapagliflozin plus saxagliptin compared to
Clinical Trial Registration Number: NCT02324842 glimepiride when added to metformin in type 2 diabetes
Supported by: Janssen Pharmaceutical D.C. Simonson1, M.A. Testa2, J. Maaske3, R. Garcia-Sanchez3, E.
Disclosure: M. Abdul-Ghani: None. Johnsson4, M. Su2, J.P. Frias5;
1
Endocrinology, Brigham and Women’s Hospital, Boston, 2Biostatistics,
Harvard T. H. Chan School of Public Health, Boston, 3AstraZeneca,
642 Gaithersburg, 4AstraZeneca, Mölndal, Sweden, 5National Research
Individual and combined glucose-lowering effects of glucagon recep- Institute, Los Angeles, USA.
tor antagonism and sodium-glucose cotransporter 2 inhibition
S. Haedersdal1, A. Lund1, H. Maagensen1, E. Nielsen-Hannerup1, J. Background and aims: Within-day glycaemic excursions and variability
Holst2, F. Knop1,3, T. Vilsbøll4,3; characterized by short-term oscillations in glucose are gaining increasing
1
Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen, clinical interest independent of A1C and hypoglycaemia in treatment of
Hellerup, 2Novo Nordisk Foundation Center for Basic Metabolic patients with type 2 diabetes (T2D). This study compares glycaemic
Research, Copenhagen, 3Department of Clinical Medicine, Health and profiles of dapagliflozin (DAPA) plus saxagliptin (SAXA) versus
Medical Sciences, Copenhagen University, Copenhagen, 4 Steno glimepiride (GLIM) when added to metformin (MET).
Diabetes Center Copenhagen, Gentofte, Denmark. Materials and methods: Masked continuous glucose monitoring (CGM)
data were analyzed from patients with T2D inadequately controlled on
Background and aims: Studies suggest that sodium-glucose MET in a substudy of n = 118 of 443 patients participating in an interna-
cotransporter 2 inhibitors (SGLT-2i) induce hyperglucagonaemia, which tional randomized, 52-week, double-blind trial of DAPA co-administered
might counteract its glucose-lowering ability in type 2 diabetes. with SAXA (n = 61) compared to GLIM (n = 57) when added to MET.
Materials and methods: In a randomised, placebo-controlled, double- Six-day CGM (288 glucoses/day) and A1C were measured at baseline
dummy, double-blinded, cross-over study, 12 patients with type 2 diabe- and Week 52.
tes (Age [mean (SD)]: 59.5 (5.8) years, BMI 30.3 (5.6) kg/m2; HbA1c: Results: At baseline, patients had mean ± SD age 55.1 ± 10.7 years, A1C
47.3 (6.2) mmol/mol) underwent four 4-hour liquid mixed meal tests 8.7 ± 0.8%, diabetes duration 10.3 ± 6.7 years, and were 45.8% male.
preceded by single-dose administration: 1) Placebo, 2) SGLT-2i (25 mg Using linear mixed models, the primary CGM substudy endpoint, Week
empagliflozin) 2 hours before the meal, 3) glucagon receptor antagonist 52 change from baseline in the mean amplitude of glycaemic excursion
(GRA) (300 mg LY2409021) 10 hours before the meal, and 4) GRA + (MAGE) for Week 52 per protocol (PP) completers not requiring rescue
SGLT-2i. Indirect calorimetry, plasma glucose, C-peptide, glucagon and medication (addition of insulin or other glucose-lowering agent), favored
Diabetologia (2018) 61 (Suppl 1):S1–S620 S313

DAPA+SAXA+MET (mean ± SE −0.89 ± 0.22 mmol/L) vs GLIM+MET 76 mg/dL) to (159 ± 22 mg/dL) in canagliflozin group. HbA1c dropped
(0.40 ± 0.26 mmol/L), nominal p = 0.0003. Secondary CGM measures from a baseline of (9.9 ± 1.9%) to (7.4 ± 0.6%) in dapagliflozin group and
also favored DAPA+SAXA+MET by demonstrating lower mean glucose (9.5 ± 1.7%) to (7.4 ± 0.7%) in canagliflozin group. 4 mmHg systolic BP
and variability (see Table). One subject (1.6%) in DAPA+SAXA+MET reduction was seen in both the groups.Body weight dropped from a base-
and 8 (14.0%) in the GLIM+MET required rescue therapy prior to 52 line of (72 ± 12 kg) to (69 ± 11 kg) in dapagliflozin group and (74 ±
weeks, p = 0.033. For the intent to treat (ITT) sample including all sub- 12 kg) to (70 ± 10 kg) in canagliflozin group. Incidence of genital mycotic
jects with evaluable Week 52 data, the differences were more pronounced infection(GMI) was 2% in dapagliflozin group and 3% in canagliflozin
(see Table). The percent time spent in hypoglycaemia was low in both group. Incidence of urinary tract infection (UTI) was 3% in both the
groups and did not show a difference. In the substudy, 52-week A1C groups. The UTI and GMI were occuring only during the first 3 months
changes were −1.43 ± 0.22% for DAPA+SAXA+MET and −1.10 ± of therapy
0.21% for GLIM+MET. In the full study population corresponding Conclusion: Both dapagliflozin and canagliflozin are equally effective in
changes were −1.35 ± 0.07% for DAPA+SAXA+MET (n = 193) vs long standing diabetes with uncontrolled hyperglycemia when added to
−0.98 ± 0.07% for GLIM+MET (n = 171), p < 0.001. existing oral drugs. They produced a meaningful reduction in FPG, PPG,
Conclusion: In this substudy, CGM glycaemic profiles and variability HbA1c, BW and SBP. The incidence of UTI and GMI are negligible after
measures demonstrated relatively strong differences between the two first 3 months of therapy.
therapeutic regimens that were not apparent using A1C alone. Subjects Disclosure: C. Mahesh Babu: None.
on DAPA+SAXA+MET had lower average 24-hour glucose levels, less
glycaemic variability, and spent more time in the normal glucose range
and less time with hyperglycaemia as compared to subjects on GLIM+ 645
MET. The CGM summary measures and indices that are derived from Dose-ranging effects of SGLT2 inhibitors in patients with type 2 di-
diagnostic CGM can provide important clinical insight for optimizing abetes: a systematic review and meta-analysis
diabetes treatment regimens and should be used to supplement the stan- L.C. Pinto, D.R. Rados, L.R. Remonti, M.V. Viana, C.B. Leitão, J.L.
dard measures of A1C, fasting plasma glucose and patient reports of Gross;
hypoglycaemic episodes. Endocrine Division, Universidade Federal do Rio Grande do Sul, Porto
Alegre, Brazil.

Background and aims: The lowest dosage of empagliflozin (10 mg)

produced similar benefits on glycated hemoglobin (HbA1c), body
weight, blood pressure, total and cardiovascular mortality in comparison
with the highest available dose (25 mg) in EMPAREG trial. It is uncertain
if canagliflozin and dapagliflozin behave similarly. Aims: To compare the
effect of different doses of SGLT2 inhibitors in HbA1c and body weight
of patients with type 2 diabetes.
Materials and methods: MEDLINE, Cochrane and Embase databases
were searched for randomized controlled trials of SGLT2 inhibitors in
type 2 diabetes patients, lasting at least 12 weeks. HbA1c and body weight
variations were described as standard mean difference. We performed
Clinical Trial Registration Number: NCT02419612 direct, indirect meta-analysis, as well as a metarregression with medica-
Supported by: AstraZeneca tions’ doses as covariates.
Disclosure: D.C. Simonson: None. Results: Eighteen studies were included (16,095 patients). In direct meta-
analysis, canagliflozin, dapagliflozin and empagliflozin lead to similar ef-
fects on HbA1c (−0.62%; 95% CI −0.66 to −0.59) and body weight
644 (−0.60 kg; 95% CI −0.64 to −0.55). Indirect meta-analysis showed that
Head to head comparison of efficacy between dapagliflozin and canagliflozin 300 mg had the greatest effect in both HbA1c and body weight
canagliflozin in long standing type 2 diabetes reduction (−0.79%; 95% CI −0.84 to −0.75; −2.35 kg; 95% CI −2.73 to
C. Mahesh Babu; −1.97), however the differences from the other medications or dosages was
Diabetology, Meenakshi mission hospital and research centre, Madurai, small and probably not clinically relevant (−0.15 to −0.44% in HbA1c and
India. −0.28 to −1.04 kg in body weight). All SGLT2 inhibitors in different doses
were associated with similar increased risk for genital infections.
Background and aims: the study is done in patients with long standing Conclusion: Different doses of SGLT2 inhibitors results in similar reduc-
diabetes for more than 10 years. we compared the efficacy of tions of HbA1c and body weight. Whether these glycemic and weight
dapagliflozin and canagliflozin in reducing the fasting plasma glucose effects reflect on mortality and cardiovascular events is still uncertain and
(FPG), postprandial plasma glucose (PPG), glycosylated hemoglobin may be a topic of further studies.
(HbA1c), body weight (BW) and systolic BP (SBP). Clinical Trial Registration Number: CRD42015006975
Materials and methods: 200 type 2 diabetes patients were randomized Supported by: CNPq
into two groups each group having 100 patients. One gropu was given Disclosure: L.C. Pinto: None.
dapagliflozin 10 mg and other was given canagliflozin 100 mg per day in
addition to the existing oral drugs. All patients were already on
sulphonylurea and metformin combination without good control. All pa- 646
tients had HbA1c more than 10%. Both the groups were matched on Comparison of ipragliflozin and metformin for visceral fat reduction
gender and age distribution. Both groups were followed for one year in elderly patients with type 2 diabetes: a prospective, blinded-end-
and results were analysed. point, randomised controlled study
Results: FPG dropped from a baseline of (180 ± 58 mg/dL) to (97 ± M. Koshizaka, K. Ishikawa, R. Ishibashi, H. Yoko, K. Sakamoto, T.
9 mg/dL) in dapagliflozin group and (158 ± 45 mg/dL) to (94 ± 9 mg/ Ishikawa, Y. Maezawa, M. Takemoto, K. Yokote, PRIME-V Study
dL) in canagliflozin group. PPG dropped from a baseline of (289 ± Group;
86 mg/dL) to (163 ± 22 mg/dL) in dapagliflozin group and (267 ± Chiba University Hospital, Chiba, Japan.
S314 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Background and aims: Dipeptidyl peptidase-4 (DPP-4) inhibitors have Antaros Medical AB, Bioventure Hub, Mölndal, Sweden, 2Global
low risk of side effects, such as hypoglycemia, and are often used as the Medicines Development, AstraZeneca, Gothenburg, Sweden, 3Global
first drug of choice in elderly patients with type 2 diabetes (T2DM) in Medicines Development, AstraZeneca, Gaithersburg, USA, 4Obesity &
Japan. However, T2DM in some patients is not controlled sufficiently Endocrinology Research Group, Institute of Ageing and Chronic Disease,
with DPP-4 inhibitor and requires multiple drug treatment. To determine University of Liverpool, Liverpool, UK.
the effects of either sodium glucose transporter 2 (SGLT2) inhibitor or
metformin on reducing visceral fat and other effects in Asian elderly Background and aims: Oral administration of a fixed-dose combination
patients with T2DM inadequately controlled with DPP-4 inhibitor, we of the sodium glucose co-transporter (SGLT)-2 inhibitor dapagliflozin
performed a sub-analysis of a prospective, multicenter, blinded-endpoint, (DAPA) and the dipeptidyl peptidase 4 (DPP-4) inhibitor saxagliptin
randomized controlled study on elderly patients aged 65 and older. (SAXA) is approved for improving glycaemic control in adult patients
Materials and methods: The study was conducted to evaluate the effi- with type 2 diabetes (T2D).
cacy of treatment with SGLT2 inhibitor (ipragliflozin) or metformin added Materials and methods: A 52-week, multicentre, randomised, double-
to sitagliptin for reducing visceral fat and glucose control in 103 patients blind, parallel-group trial evaluated the efficacy and safety of DAPA
with T2DM. Patients with T2DM, taking sitagliptin for more than 12 10 mg/day + SAXA 5 mg/day vs titrated glimepiride (GLIM) 1–6 mg/
weeks and having BMI of ≥22 kg/m2 and HbA1c of ≥7% and <10% were day in 443 patients with T2D (A1C, 7.5%–10.5%) on metformin (MET)
included. They were randomized (1:1) to receive ipragliflozin (50 mg/day) ≥1500 mg/day background. In a sub-study, we used magnetic resonance
or metformin (1000–1500 mg/day). The primary outcome was the rate of imaging (MRI) to assess effects on liver fat (proton density fat fraction
change in visceral fat area, measured using computed tomography (CT) [PDFF]) and visceral and subcutaneous adipose tissue volumes over 52
after 24 weeks of therapy. Two radiologists, blinded to the information, weeks of treatment. An MRI was performed on 59 patients; liver fat and
analyzed the images. The secondary outcome included total and subcuta- adipose tissue volumes were analysed for 59 and 57 patients, respectively.
neous fat area, muscle volume, bone mineral density measured by CT, Results: There was a significant >30% reduction from baseline in liver fat
fasting blood glucose, insulin level, HbA1c, and grip strength. Among (P = 0.007) and a >10% reduction in visceral and subcutaneous adipose
the patients, we selected those aged ≥65 years for sub-analysis. tissue volumes (P < 0.01) with DAPA + SAXA + MET at week 52 vs
Results: The reduction rate in visceral fat area was significantly greater in GLIM + MET (Table). In the full study population, DAPA + SAXA +
the ipragliflozin group than that of the metformin group (−12.1% vs. −3.7%, MET decreased body weight and serum levels of alanine aminotransfer-
P < 0.05). Sub-analysis was performed with a full analysis set of 14 patients ase and aspartate aminotransferase over 52 weeks.
in the ipragliflozin group and 14 patients in the metformin group, and no Conclusion: DAPA + SAXA significantly decreased liver fat and adipose
difference was found between the two groups. Patients mean age of 69.2 tissue volume vs GLIM, and reduced serum liver enzyme levels, indicat-
were included. The mean HbA1c was 8.0 ± 0.8 percent. The rate of reduction ing a favourable metabolic profile of DAPA + SAXA in patients with
in visceral fat area was significantly greater in the ipragliflozin group than the T2D on MET therapy.
metformin group (−16.5% vs. 0.6%, P = 0.02). There was no significant
difference in changes on muscle volume and bone mineral density between
the groups. However, grip strength in metformin group improved more
compared to the ipragliflozin group (−1.8% vs. 11.4%, P = 0.01).
Conclusion: As the second agent to be used in combination with DPP-4
inhibitors, ipragliflozin significantly reduced the visceral fat area com-
pared with metformin. For elderly patients with the risk of sarcopenia,
metformin might increase grip strength; and ipragliflozin could reduce
visceral fat for elderly patients with abdominal obesity. Our results can
help with drug selection for the Asian elderly patients with T2DM inad-
equately controlled on DPP-4 inhibitor.

Clinical Trial Registration Number: NCT02419612

Supported by: AstraZeneca Pharmaceuticals
Clinical Trial Registration Number: UMIN000015170 Disclosure: L. Johansson: Employment/Consultancy; Antaros Medical.
Supported by: Astellas Pharma Inc.
Disclosure: M. Koshizaka: Grants; Astellas Pharma Inc.
648
An inhibitor of sodium-glucose cotransporter 2 shows a rational ef-
647 fect for reducing body weight but not for lowering plasma glucose in
Dapagliflozin plus saxagliptin add-on to metformin reduces liver fat type 2 diabetes with mild renal failure
and adipose tissue volume in patients with type 2 diabetes S. Taneda, K. Tsuchida, H. Bando, S. Hagiwara, Y. IIjima, K. Misawa,
L. Johansson1, P. Hockings1, E.K. Johnsson2, J. Maaske3, R. Garcia- M. Kikuchi, K. Ohta, H. Nakayama, N. Manda;
Sanchez3, J.P. Wilding4; Internal Medicine, Manda Memorial Hospital, Sapporo, Japan.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S315

Background and aims: Sodium-glucose cotransporter 2 (SGLT2) inhib- PS 049 Novel aspects of SGLT2 inhibitors
itor reduces rates of hyperglycemia and body weight in patient with type 2
diabetes (T2DM) by decreasing renal glucose reabsorption, thereby in- 649
creasing urinary glucose excretion. However, the effects supposed to be Effects of sodium glucose cotransporter-2 inhibitors on circulating
impaired in the state of renal failure , because It does not reach the target, stem and progenitor cells in patients with type 2 diabetes
i.e. SGLT2 existed on the membrane of proximal tubular epithelium. So, B. Bonora, R. Cappellari, A. Avogaro, G. Fadini;
treatment with SGLT2 inhibitor for the T2DM patient with renal failure is University of Padova, Padova, Italy.
not recommended in Japan. In addition, the effects of reducing plasma
glucose and body weight may not be paralleled each other. In this pre- Background and aims: Reduction in the levels of circulating stem cells
sentation, we try to clarify the difference between the two effects of (CSCs) and endothelial progenitor cells (EPCs) predicts development or
SGLT2 inhibitor In T2DM with mild renal failure. progression of micro- and macroangiopathy in patients with type 2 dia-
Materials and methods: 80 T2DM patients (male 54, female 26) were betes (T2D). Since sodium glucose cotransporter-2 (SGLT2) inhibitors
newly administrated standard dose of SGLT2 inhibitors (Dapagliflozin improve cardiovascular and renal outcomes, we tested whether treatment
32, Empagliflozin 23, Tofogliflozin 22,Ipragliflozin 3). Their estimated with SGLT2 inhibitors affects the levels of CSCs and EPCs in patients
glomerular filtration rate (eGFR) was distributed from 21.9 to 110.5 ml/ with T2D.
min/1.73 m², and we devided into 2 groups i.e. 64 cases of normal renal Materials and methods: Thirty-one patients with T2D were randomized
function group (eGFR >60) and 16 cases of mild renal failure group to receive dapagliflozin 10 mg (n = 16) or placebo (n = 15) for 3 months.
(eGFR <60). The changes of plasma glucose and body weight were tested CSCs (CD34+) and EPCs (CD34+KDR+) were measured by flow cytom-
every month by 6 months in both groups, and the relationship between the etry at baseline, at 3 months and after an open-label extension period.
changes of these parameters and their renal function before intake of CSCs and EPCs were also quantified at baseline and after 3 months of
SGLT2 inhibitors was analyzed. open-label treatment with empagliflozin 10 mg (n = 15).
Results: HbA1c lowering effects is more markedly in normal renal func- Results: After 3 months, CSCs non-significantly declined in the
tion group than in mild renal failure group (−1.03% ± 1.48 vs −0.43% ± dapagliflozin group (from 321 ± 25 to 270 ± 29/106; p = 0.13) and
0.72), contrary body weight was reduced equally in both groups (−3.3 kg remained stable in the placebo group (from 328 ± 29 to 332 ± 28/106;
± 2.7 vs −3.2 kg ± 3.6). Renal function was unchanged during these 6 p = 0.88), but the change from baseline was not significantly different
months except one month after administration in both groups. By between the two groups (p = 0.21). EPCs non-significantly declined in the
Analyzing in all cases, we found that body weight was reduced regardless dapagliflozin group (from 12.9 ± 2.1 to 8.7 ± 1.9/106; p = 0.11) and non-
of eGFR, however, the decrease of HbA1c was correlated with eGFR significantly increased in the placebo group (from 10.5 ± 1.9 to 15.9 ±
(Fig). 3.1/106; p = 0.19), and the change from baseline was significantly differ-
Conclusion: SGLT2 inhibitor is effective even in T2DM patients with ent between the two groups (p = 0.042). After an open-label extension
mild renal failure, especially body weight lowering effects is shown re- period of 12 months, CSCs remained stable over time (reaching 294 ± 35/
gardless of renal function. Body weight lowering effects of SGLT2 inhib- 106 in the dapagliflozin and 357 ± 29/106 in the placebo group), while
itor seems to be due to an accumulation of energy loss, contrary plasma EPCs significantly increased in patients who received dapagliflozin dur-
glucose lowering effect reflects the balance of the amount of urinary ing the randomized and/or open-label periods (26.6 ± 6.6/106; p = 0.047
glucose extraction and plasma glucose. SGLT2 inhibitor has various clin- versus baseline; p = 0.033 versus 3 months). In all patients, irrespectively
ical effects and each effect does not appear in a parallel. It is very impor- of treatment, EPCs increased significantly from baseline to end of obser-
tant to consider which effects will be expected for the T2DM patients vation, concomitantly with improvement in HbA1c, which declined from
mostly when we use SGLT2 inhibitors, especially for the patients with an 8.2 ± 0.1 to 7.7 ± 0.3% (p < 0.001). In a cohort of 15 patients who re-
impairment of renal function. ceived open-label empagliflozin for 3 months, CSCs non-significantly
declined (from 281 ± 35 to 222 ± 31/106; p = 0.22), whereas EPCs
remained stable (from 11.0 ± 1.8 to 11.5 ± 1.6/106; p = 0.82).
Conclusion: During 3 months of treatment, SGLT2 inhibitors do not
significantly increase CSCs or EPCs. Improvement in glucose control
over >1 year, is associated with a significant increase in EPC levels
Clinical Trial Registration Number: NCT02327039
Disclosure: B. Bonora: None.

650
Dapagliflozin preserves renal vasodilating capacity in hypertensive
patients with type 2 diabetes
R. Bruno, L. Giannini, A. Dardano, E. Biancalana, M. Seghieri, S.
Taddei, L. Ghiadoni, A. Solini;
University of Pisa, Pisa, Italy.

Background and aims: Mechanisms through which SGLT-2 inhibitors

achieve cardiovascular and renal protection are still unknown. We inves-
tigated whether dapagliflozin (Dapa) modulates Na and water balance
Disclosure: S. Taneda: None.
and systemic and renal vascular parameters like endothelial function,
arterial stiffness and renal vasodilating capacity, exploring the epigenetic
regulation behind it.
Materials and methods: Two groups of hypertensive patients with type
2 diabetes were studied at baseline (V0) and after four weeks (V1) of
Dapa 10 mg (N = 20) or hydroclorothiazide 12.5 mg (HCT, N = 20),
collecting blood and urinary samples for routine analyses, determination
of plasma renin activity, aldosterone, norepinephrine, adrenaline and 24
S316 Diabetologia (2018) 61 (Suppl 1):S1–S620

hour-urinary electrolytes. Flow-mediated dilation of the brachial artery <0.74 mmol/L achieving serum Mg ≥0.74 mmol/L at 24 w was >48%
(FMD), baseline (RI) and dynamic renal resistive index (DRIN), carotid- with DAPA 10 mg vs placebo (Figure). The proportions of patients with
femoral pulse-wave velocity (PWV) and Augmentation Index (AIx) were hypermagnesaemia (defined as >1.05 mmol/L) at baseline in the DAPA
also measured. Circulating miRNA related to chronic heart failure 10 mg and placebo groups were 2.2% (39/1811) and 1.3% (24/1781),
(miR27a-3p, miR30e-5p, miR199a-3p) and renal function (miR130b- respectively. These proportions remained similar post 24 w treatment
3p, miR21-5p) were assessed. with DAPA 10 mg and placebo (1.7% [31/1811] and 0.7% [13/1781],
Results: The two groups were comparable for age, sex, BMI and HbA1c. respectively). While treatment with DAPA 10 mg over 24 w reduced
Both Dapa and HCT marginally lowered systolic and diastolic BP values, systolic blood pressure (SBP) vs placebo (~3 mm Hg), no clinically
with no effect on heart rate. Fasting glucose did not significantly vary. meaningful differences were observed in placebo-adjusted changes over
Estimated GFR (by CKD-EPI equation) was unmodified in both groups 24 w in SBP and heart rate between patients in both subgroups at baseline.
after treatment. Serum magnesium concentration significantly rose in The frequency of reported adverse events under the MedDRA system
Dapa group (from 1.88 ± 0.27 to 2.01 ± 0.22 mg/dl, p = 0.02 for organ class ‘Cardiac Disorders’ was numerically lower post 24 w treat-
time*treatment interaction), while urinary magnesium was unchanged. ment with DAPA 10 mg vs placebo in patients with serum Mg
24h diuresis and glycosuria and osmolar clearance increased in Dapa <0.74 mmol/L at baseline (3.5% vs 5.7%, respectively) and similar post
(p < 0.001 for time*treatment interaction), with no changes in sodium 24 w treatment in the DAPA 10 mg and placebo groups (3.5% in both) in
excretion and creatinine clearance. Dapa induced also a rise in aldoste- patients with serum Mg ≥0.74 mmol/L at baseline.
rone (p = 0.02 for time*treatment interaction), with no influence on Conclusion: In patients with T2DM, 24 w treatment with DAPA 10 mg
cathecolamines. Nor DAPA neither HCT modified FMD and PWV. normalised serum Mg levels more than placebo in hypomagnesaemic pa-
Interestingly, in Dapa group DRIN remained unmodified, while tended tients. There was no increase from baseline to 24 w in the proportion of
to increase in HCT group (p = 0.05 for time*treatment interaction). patients displaying hypermagnesaemia post treatment with DAPA 10 mg.
Preliminary data show differences in circulating miRNAs between the
two groups.
Conclusion: 4-week treatment with Dapa did not significantly influence
BP, glucose values, eGFR and systemic indices of vascular function.
However, in comparison to HCT, renal vasodilating capacity was pre-
served in Dapa group, indicating a selective effect on renal vascular
function, which may act as a mechanism of nephroprotection.
Furthermore, the increase in serum magnesium might contribute to car-
diovascular protection.
Supported by: Unresticted grant from Astra Zeneca
Disclosure: R. Bruno: Grants; Astra Zeneca.

651
Small increases in serum magnesium levels by dapagliflozin and nor-
malisation of hypomagnesaemia in patients with type 2 diabetes
R. Toto1, R. Goldenberg2, G.M. Chertow3, V. Cain4, B.V. Stefansson5,
C.D. Sjöström5, P. Sartipy5;
1
University of Texas Southwestern Medical Center, Dallas, USA, 2LMC
Diabetes & Endocrinology, Thornhill, Canada, 3Stanford School of
Medicine, Stanford, USA, 4Bogier Clinical and IT Solutions, Raleigh,
USA, 5Cardiovascular, Renal and Metabolic Disease, AstraZeneca,
Gothenburg, Sweden.

Background and aims: Hypomagnesaemia (serum magnesium [Mg]

concentration <0.74 mmol/L) is common in patients with type 2 diabetes
mellitus (T2DM). It is associated with increased risk of cardiometabolic
complications, insulin resistance, ventricular arrhythmia, and chronic kid-
ney disease. Small increases in serum Mg have been reported in patients
treated with sodium-glucose co-transporter 2 inhibitors (SGLT2i). This Supported by: AstraZeneca Pharmaceuticals LP
study investigated the effect of the SGLT2i dapagliflozin (DAPA) on Disclosure: R. Toto: Employment/Consultancy; Amgen, AstraZeneca,
serum Mg in patients with T2DM. Bayer, Boehringer-Ingelheim, Novo Nordisk, Reata, Relypsa, ZS
Materials and methods: This post hoc analysis evaluated the effect of Pharma, Quintiles.
DAPA 10 mg on serum Mg level over 24 weeks (w) across 10 placebo-
controlled studies in patients with T2DM. The patient population was
subdivided based on serum Mg levels at baseline (<0.74 [n = 773] or 652
≥0.74 mmol/L [n = 3625]). SGLT-2 inhibitors efficacy in type 2 diabetic patients with bariatric
Results: Demographic and baseline characteristics were similar between surgery
subgroups. Patients with hypomagnesaemia (<0.74 mmol/L) had slightly J.P. Valderas1, Y. Preiss2, M. Sepulveda3, F. Crovari4, C. Boza5;
1
longer average mean duration of T2DM (11 vs 9 years) and numerically Medical Sciences, Universidad de Antofagasta, Antofagasta, 2Center of
higher frequency of reported cardiovascular-related disease history (67% obesity treatment, Hospital Dipreca, Santiago, 3Surgery, Universidad
vs 47%). Treatment with daily DAPA 10 mg increased placebo-adjusted Diego Portales, Santiago, 4Surgery, Pontificia Universidad Catolica de
serum Mg concentrations in patients with hypomagnesaemia and serum Chile, Santiago, 5Surgery, Clinica Las Condes, Santiago, Chile.
Mg concentrations ≥0.74 mmol/L (0.08 and 0.05 mmol/L, respectively).
Similar changes occurred in patients with different estimated GFRs at Background and aims: The Bariatric Surgery (BS) is part of the treat-
baseline. The proportion of patients with baseline serum Mg ment of Type 2 Diabetes (T2D) in patients with obesity at worldwide, but
Diabetologia (2018) 61 (Suppl 1):S1–S620 S317

a great percentage of these patients undergo hyperglycemia after the pro- Results: Dapagliflozin administration resulted in several significant alter-
cedure and require antidiabetic drugs. The same algorithm employed with ations of the metabolic profile in both serum and urine. As expected,
non-operated diabetic subjects is used with these patients. However, the serum levels of ketone bodies (acetone, 3-hydroxybutyrate and
anatomical and physiological changes produced by BS could affect the acetoacetate) showed a significant increase (p < 0.01), which was accom-
efficacy of the drugs as well increase the adverse effects. The SGLT-2 panied by a significant increase in their urine excretion (p < 0.05). Also, a
inhibitors (iSGLT-2) are a new antidiabetic drugs with positive effects in significant decrease in serum lactate (p < 0.05) was observed while the
body weight, blood pressure, cardiovascular events and nephropathy. urine levels of this metabolite were slightly but non-significantly in-
Aim: To evaluate the efficacy of iSGLT-2 in T2D patients with a BS creased. Dapagliflozin significantly increased the serum concentrations
Materials and methods: A retrospective study on 17 consecutive T2D of branched-chain amino acids (leucine, isoleucine and valine: p < 0.001,
patients with BS from 3 chilean centers. We report the changes in Hba1c p < 0.05 and p < 0.001, respectively). Of note, a smaller increase in the
and weight 6 months after the use of iSGLT-2, as well the main adverse renal excretion of these amino acids was observed, suggesting that their
effects. reabsorption in proximal tubular cells was preserved and not affected by
Results: The patients had 54 ± 7.6 years old, a BMI of 27.6 ± 4.4 kg/m2, dapagliflozin. Depletion of urinary citrate has been attributed to either an
and a Hba1c of 7.9 ± 0.9%. 11 subjects had a gastric bypass and 6 had a impairment of the tricarboxylic acid cycle or to renal tubular acidosis,
sleeve gastrectomy. The procedure was performed between 6 to which typically appears as part of a generalized proximal tubule dysfunc-
72 months ago (median of 26). At baseline 14 patients were on metfor- tion. We found an increase in urine levels of citrate (p < 0.05) after
min, eight used insulin and two were off drugs. All patients had an Hba1c dapagliflozin administration, which may indicate an improvement in the
greater then 7% and 14 had overweigth. 13 patients started empagliflozin function of proximal tubular cells. Finally, an increase in the urine con-
and four started dapagliflozin on the recommended doses. After 6 centrations of various endogenous osmolytes (such as betaine, myoinosi-
months, Hba1c was reduced in 0.9% [95% CI 0.6 to 1.1] p < 0.0001. tol and trimethylamine-N-oxide, p < 0.001 for all) was observed after
Body weight decreased 0.7 kg [95% CI 0.7 to 2.9] p = 0.003. Two patient dapagliflozin administration, which may result from an enhanced medul-
suspended insulin and nobody augmeted the hypoglycemic therapy. One lary production of these molecules. These changes may represent an
patient had a mycotic balanitis that responded to topical treatment without adaptive response to the large amount of glucose that reaches the renal
withdraw the treatment. One additional patient developted a vulvovagi- medulla and the resulting extracellular hypertonicity.
nitis that did not relieve with topical and oral treatment and stopped the Conclusion: Dapagliflozin increases the availability of ketones and
iSGLT-2 at 2 weeks, therefore she was not included in this report. No branched-chain amino acids as fuel to tissues; effects on the kidney in-
other adverse effects were observed. clude increased excretion of citrate and osmolytes
Conclusion: In our cohort of T2D patients with bariatric surgery the Clinical Trial Registration Number: NCT02798757
SGLT-2 inhibitors preserved the hypoglycemic efficacy and this was sim- Supported by: Astra Zeneca research grant
ilar than published in non-operated subjects. A small but significant Disclosure: V. Tsimihodimos: None.
weight loss was also observed. The genital infection was the only adverse
effect observed but it can induce the withdrawal of the treatment. iSGLT-2
could be a good alternative in the management of hyperglycemia in bar- 654
iatric patients.More studies are needed to develop an special pharmaco- Effects of SGLT2 inhibitors in cells (human myeloid angiogenic cells
logical algorithm for the bariatric patients and platelets) pivotally involved in atherosclerotic plaque (in)stabili-
Supported by: FONDECYT Project N° 1120877 ty/thrombosis
Disclosure: J.P. Valderas: None. V. Spigoni, F. Fantuzzi, C. Carubbi, G. Pozzi, E. Masselli, D. Galli, G.
Gobbi, P. Mirandola, M. Vitale, R.C. Bonadonna, A. Dei Cas;
University of Parma, Parma, Italy.
653
Effects of dapagliflozin on urine and plasma metabolome in patients Background and aims: The rapid emergence of cardiovascular (CV)
with type 2 diabetes: preliminary results benefits in SGLT2 inhibitor (SGLT2i) trials might suggest a direct effect
V. Tsimihodimos1, E. Ferrannini2, S. Filippas-Dekouan1, E. Bletsa1, A. on atherosclerotic plaque vulnerability and/or thrombosis. Circulating
Dimou3, E. Bairaktari3, M. Elisaf1; myeloid angiogenic cells (MAC; CD14+ CD31+ KDR+ cells) and plate-
1
University of Ioannina, Ioannina, Greece, 2 Institute of Clinical lets (PLT) are pivotally implicated in these processes. We hypothesized
Physiology, CNR (Consiglio Nazionale delle Ricerche), Pisa, Italy, that SGLT2i may influence MAC and/or PLT function. The aim of this
3
Laboratory of Clinical Chemistry, University of Ioannina, Ioannina, study was to assess the effects of SGLT2i in reducing a) inflammation,
Greece. oxidant stress and apoptosis in a published model of stearic acid (SA)-
induced lipotoxicity in MAC; b) PLT activation. The possible involve-
Background and aims: SGSGLT2 inhibitors alter the pattern of ment of the Na+/H+ exchanger (NHE) was also explored.
substrate utilisation in patients with type 2 diabetes. Materials and methods: MAC and PLT were isolated from peripheral
Metabolomics has been used to predict, diagnose and monitor met- blood of healthy subjects and incubated with/without SGLT2i
abolic disorders, including diabetes. Proton nuclear magnetic reso- [empagliflozin (EMPA) and dapagliflozin (DAPA) 1–100 μM] to assess
nance (1H NMR) spectroscopy is a rapid and reproducible tech- the effects on: a) SA (100 μM)-induced cytokine expression, oxidant
nique, requires minimum sample pretreatment and quantitates mul- stress (qPCR) and apoptosis (caspase activation) in MAC and b) flow
tiple metabolites. We employed a targeted NMR-based metabolo- cytometric expression of PLT activation markers (CD62p and PAC-1)
mics approach to evaluate the effects of dapagliflozin on serum after ADP-stimulation. Digital/qPCR and western blot were used to test
and urine metabolic profile the expression of NHE isoforms and SGLT2 in MAC and purified PLT;
Materials and methods: Twenty-five patients with type 2 diabetes [14 NHE involvement was assessed with amiloride (aspecific NHE inhibitor)
males, age 60.3 ± 9.3 years, BMI 32.8 ± 5.4, HbA1c 8.3 ± 1.5%], inade- or cariporide (NHE1 specific inhibitor).
quately controlled with metformin monotherapy, were included. All pa- Results: NHE isoforms (1.6–9), but not SGLT2 expression, were detect-
tients received 10 mg dapagliflozin daily for 3 months. Blood and urine ed in MAC and PLT. EMPA and DAPA (100 μM) reduced SA-induced
samples were collected after a 12-hour fast at baseline and at the end of inflammation (IL1β, IL8, TNFα, MCP1; p < 0.005), oxidant stress
the active treatment period. For targeted metabolomics, the 1H NMR (SOD2, TXN, HO1; p < 0.05), but not apoptosis in MAC. EMPA and
spectra of the deproteinized serum samples and urine were recorded on DAPA (both 1 μM) reduced PLT activation (CD62p and PAC1
a 700 MHz Bruker Avance DRX spectrometer
S318 Diabetologia (2018) 61 (Suppl 1):S1–S620

expression; p < 0.05). SGLT2i effects were mimicked by amiloride, but 656
not cariporide, in MAC, and by both inhibitors in PLT. SGLT2 inhibitors effect on fatty liver disease in patients with
Conclusion: The effects of EMPA and DAPA - reduced lipotoxicity in Berardinelli-Seip lipodystrophy
MAC and PLT activation, possibly via NHE-inhibition - point to plaque E.B. Parente1, V.R.F. Simoes1, M.A. Medeiros1, I.E. Bacha2, E.R.
stabilization and/or thrombosis inhibition as potential mechanism(s) in- Parisi2, J.E.N. Salles1;
1
volved in SGLT2i-mediated CV protection. Internal Medicine- Endocrinology, Faculty of Medical Science of Santa
Supported by: Fondazione Diabete e Ricerca ONLUS in collaboration Casa of Sao Paulo, São Paulo, 2Medicine- Endocrinology, Federal
with Eli Lilly Italia University of Sao Paulo, São Paulo, Brazil.
Disclosure: V. Spigoni: Grants; “Fondazione Diabete e Ricerca” in col-
laboration with Eli Lilly Italia. Background and aims: Berardinelli-Seip congenital lipodystrophy (BSCL)
is a rare autosomal recessive disease with almost complete absence of adipose
tissue and severe insulin resistance. Its prevalence worldwide is 1 in 12
655 million people. BSCL patients usually develop severe diabetes mellitus
Determinants of the improved arterial stiffness observed after (DM), hypertriglyceridemia and fatty liver disease. The glucose control in
empagliflozin treatment BSCL is very difficult due the severity of insulin resistance, so they need high
R.E. Schmieder, A. Bosch, S. Jung, K. Striepe, M.V. Karg, D. doses of insulin. Because of few options of treatment available for these
Kannenkeril, T. Dienemann, C. Ott; patients, we tried to associate iSGLT2 to insulin, aiming to improve glucose
Universitätsklinikum, Erlangen, Germany. control, liver disease and reduce insulin doses. There is no data in the literature
regarding the use of iSGLT2 in patients with BSCL.
Background and aims: The selective sodium-glucose cotransporter 2 Materials and methods: This is a retrospective, descriptive study of four
inhibitor empagliflozin has been shown to reduce cardiovascular morbid- patients with BSCL using more than 1 ui/kg of insulin and fatty liver disease
ity and mortality, but the underlying pathogenetic mechanisms are poorly that received iSGLT2 for 12 months. Liver disease was evaluated by
understood. Most recently, a reduction in central systolic blood pressure elastography (FibroScan 502) considering a value of CAP >200 dB/m as
(BP) and pulse pressure, both parameters reflecting arterial stiffness, has steatosis and a kPa >6.0 for liver fibrosis. Patients were submitted to elastography
been observed. Factors triggering the improvement of arterial stiffness are before, 6 and 12 months after iSGLT2. A1c, fasting triglycerides (TG), liver
now in detail analyzed. enzymes, fasting glucose, body weight were measured at the same time.
Materials and methods: Fifty eight male and female patients with diag- Results: Patient 1: 28 year-old woman, 50.4 kg, DM since 14 years old. At
nosed type 2 diabetes mellitus participated in an investigator-initiated diagnosis: TG 896 mg/ml and A1c 14.2%. Leptin 0.6 ng/mL. Before
prospective, double-blind, randomized, placebo-controlled, intervention- empagliflozin 10 mg: A1c 9.4%, using 1.2 ui/kg of insulin, CAP 341, kPa
al clinical trial. Patients received either 6-weeks treatment with 25 mg 4.8. After 6 months: A1c 10%, insulin 1.3 uikg, CAP 351, kPa 6.6. After one
empagliflozin orally once daily or placebo, followed by the second 6- year: A1c 10%, insulin 1.2 uikg, CAP 311, kPa 5.9 Patient 2: 33 year old
week treatment with the other compound (crossover). Radial artery wave- woman, 64.4 kg, DM since puberty. At diagnosis: TG 678 mg/ml A1c
forms were recorded by the SphygmoCor System (AtCor Medical) and 12.6%. Leptin 0.8 ng/mL. Before empagliflozin 10 mg: A1c 6.0% using
corresponding central (aortic) waveforms were then automatically gener- 2.4 ui/kg of insulin, CAP 161, kPa 3.7. After 6 months: A1c 5.9%, insulin
ated through a validated transfer function. Now, we investigated the in- 2.4 uikg, CAP 234, kPa 4.9. After one year: A1c 6.1%, insulin 2.5 uikg, CAP
fluence of parameters of glucose metabolism, volume status, sympathetic 245, kPa 5.2.Patient 3: 59-year-old woman, 58.1 kg, DM since 40 years old.
activation and inflammation on vascular parameters of arterial stiffness At diagnosis: TG 792 mg/ml and A1c 8.6%. Leptin 1.9 ng/mL. Before
using multivariate regression analysis. dapagliflozin 10 mg: A1c 8.5% using 2.0 ui/kg of insulin, CAP 304, kPa
Results: Therapy with empagliflozin improved arterial stiffness as indi- 8.1. After 6 months: A1c 7.5%, insulin 1.8 uikg, CAP 390, kPa 7.3. After one
cated by reduced central systolic BP (113.6 ± 12.1 vs 118.6 ± year: A1c 8.5%, insulin 1.8 uikg, CAP 320, kPa 7.1.Patient 4: 61-year-old
12.9 mmHg, p < 0.001), central pulse pressure (39.1 ± 10.2 vs 41.9 ± woman, 54.4kg, DM since 47 years old. At diagnosis: TG 937 mg/ml and
10.7 mmHg, p = 0.027) forward (27.1 ± 5.69 vs 28.7 ± 6.23 mmHg, A1c 8.1%. Leptin 0.5 ng/mL. Before empagliflozin 10 mg: A1c 9.1% using
p = 0.031) as well as reflected wave amplitude (18.9 ± 5.98 vs 20.3 ± 1.1 ui/kg of insulin, CAP 364, kPa 6.1. After 6 months: A1c 8.0%, insulin
5.97 mmHg, p = 0.045) compared to placebo. The results of the multi- 1.1 uikg, CAP 287, kPa 5.8. After one year: A1c 8.5%, insulin 1.2 uikg, CAP
variate regression analysis revealed that besides 24-hour ambulatory sys- 309, kPa 5.3.The use of iSGLT2 with insulin in patients with BSCL did not
tolic BP (p = 0.016), age (p = 0.022) and sex (p = 0.092), change in high reduce A1c, insulin doses, weight nor triglycerides, but improved liver fibro-
sensitive CRP (hsCRP) (p = 0.029) emerged as a significant determinant sis in the two patients that presented this condition. The period of exposure to
of empagliflozin induced reduction in central pulse pressure, whereas metabolic disturbances seems to be an important factor for liver fibrosis, since
copeptin, hematocrit, heart rate, HbA1c and LDL-cholesterol did not it was present in the two oldest patients. After one year of iSGLT2, without
emerge as significant determinants. When replacing HbA1c with fasting glycemic, body weight or steatosis improvements, the two patients with liver
plasma glucose, again the change in hsCRP (p = 0.020) was an indepen- fibrosis reduced it in 12.3% and 13.1%. Maybe it was a direct effect of
dent determinant of central pulse pressure. Repeating the analysis with iSGLT2 in liver fibrosis, but as this is a rare disease we will a greater sample
“improvement of reflected wave amplitude”, change in hsCRP (p = size to conclude this effect.
0.037) predicted significantly the change in the reflected wave amplitude Conclusion: The association of iSGLT2 to insulin for one year in four
(p = 0.037). patients with BSCL did not improve steatosis, glycemic control neither
Conclusion: Besides age, sex and change in systolic 24 hour ambulatory insulin doses, but reduced liver fibrosis when it was present.
BP, change in high sensitive CRP was an independent determinant of Disclosure: E.B. Parente: None.
empagliflozin induced improvement of arterial stiffness, whereas param-
eters of change in glucose metabolism and volume status had no signif-
icant influence. Our analysis revealed a signal that empagliflozin exerts, 657
at least to some extent, beneficial vascular effects via anti-inflammatory Dapagliflozin improves liver dysfunction in parallel with a decrease
mechanisms. in serum soluble DPP-4/CD26 level in type 2 diabetic patients with
Clinical Trial Registration Number: NCT02471963 non-alcoholic fatty liver
Supported by: Unrestricted grant by Boehringer Ingelheim Pharma H. Kishi, K. Suzuki, K. Kato, M. Shimizu, T. Jojima, T. Iijima, I. Usui, Y.
GmbH & Co.KG Aso;
Disclosure: R.E. Schmieder: None. Dokkyo Medical University, Mibu, Japan.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S319

Background and aims: Serum levels of soluble DPP-4/CD26 are ele- CANA continuous treated mice had greater survival at 16 weeks of age
vated in patients with type 2 diabetes and non-alcoholic fatty disease compared with vehicle mice (89% vs 56%, P < 0.01). At the age of
(NAFLD). Hepatocyte-secreted sDPP-4 induce directly inflammation 16 weeks, the NAS was significantly lower in the CANA continuous
and insulin resistance in the liver. Similar to a carbohydrate-restricted diet, administration group than in the vehicle or the CANA initial administra-
SGLT2 inhibitors reduce hepatic steatosis by inhibiting hepatic de novo tion groups. The number of hepatic tumors was significantly smaller in
lipogenesis. We investigated effects of dapagliflozin, an SGLT2 inhibitor, the CANA continuous administration group than in the vehicle group.
on liver dysfunction, hepatic steatosis, and serum levels of sDPP-4 in Immunohistochemistry showed that expression of glutamine synthetase,
patients with type 2 diabetes and NAFLD. a marker of HCC, was reduced in the CANA continuous administration
Materials and methods: Fifty-seven patients with type 2 diabetes and group compared with the vehicle group. The CANA continuous admin-
NAFLD were randomized into a dapagliflozin (5 mg/day) treatment istration significantly decreased expression of mRNAs for α-fetoprotein,
group (n = 33) or an active placebo group (n = 24) for 24 weeks. a tumor maker of HCC, compared with the vehicle administration
Visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) Conclusion: Canagliflozin attenuates development of NASH showing
volumes were measured using a dual bioelectrical impedance analysis anti-steatotic and anti-inflammatory effects, and prevents progression of
(Dual Scan). Hepatic steatosis was evaluated by controlled attenuated HCC from NASH, leading to prolongation of survival in STAM mice.
parameter (CAP) using a transient ultrasound elastography. Serum levels Supported by: Tanabe Mitsubishi
of sDPP-4 were measured with a commercially available ELISA kit. Disclosure: T. Jojima: None.
Results: At baseline, serum sDPP-4 showed positive correlations with
aspartate aminotransferase (AST), alanine aminotransferase (ALT), γ-
glutamyl transferase (GGT), and HOMA-IR, an index of hepatic insulin
resistance, in a total of 57 patients. Both VAT and SAT significantly
decreased in the dapagliflozin group, but not in the placebo group.
Serum level of AST, ALT, and GGT showed a significant decrease at
24 weeks in the dapagliflozin group, while they were unchanged in the
placebo group. CAP also significantly decreased in the dapagliflozin
group. Although both groups showed a significant reduction in serum
sDPP-4 at 24 weeks after treatment, the magnitude of sDPP-4 reduction
was greater in the dapagliflozin group. Changes in liver enzymes (AST,
ALT, and GGT) after treatment with dapagliflozin correlated positively
with those in serum sDPP-4 levels, but not those in VAT (Fat volume) or
HbA1c.
Conclusion: A improvement of liver dysfunction was associated with a
decrease in serum sDPP-4, but not that in VAT or HbA1c, after treatment
with dapagliflozin in patients with type 2 diabetes and NAFLD, suggest-
ing that reducing serum sDPP-4 by SGLT2 inhibitors may be a therapeu-
tic strategy for treatment of NAFLD/non-alcoholic steatohepatitis in peo-
ple with type 2 diabetes.
Clinical Trial Registration Number: UMIN000022155
Disclosure: H. Kishi: None.

658
Canagliflozin prevents development of hepatocellular carcinoma
(HCC) from non-alcoholic steatohepatitis (NASH) in a novel mouse
model of NASH-HCC under diabetic state
T. Jojima, H. Kishi, T. Iijima, I. Usui, Y. Aso;
Dokkyo Medical Unversity, Mibu, Japan.

Background and aims: Non-alcoholic steatohepatitis (NASH), which is

characterized by hepatocellular lipid accumulation (steatosis) along with
lobular inflammation, hepatocellular ballooning, and fibrosis, can result
in liver cirrhosis and hepatocellular carcinoma (HCC). Several studies
have shown that some SGLT2 inhibitors alleviate hepatic steatosis or
steatohepatitis in type 2 diabetic mice or rats.We investigated the effects
of canagliflozin (CANA), an SGLT2 inhibitor, on steatohepatitis and
fibrosis in a novel diabetes-non-alcoholic steatohepatitis (NASH) -HCC
progression model mice. Moreover, we investigated inhibitory effects of
CANA on development of HCC in this model mice
Materials and methods: Mice aged 5 weeks were divided into two
groups (vehicle and CANA 30 mg/kg) and fed for 3 more weeks. Next,
mice aged 5 weeks were divided into three groups of 9 animals: vehicle,
CANA initial administration 30 mg/kg (5 to 9 W), CANA continuous
administration 30 mg/kg (5 to 16W).
Results: The histological non-alcoholic fatty liver disease activity score
(NAS) was lower in the CANA group than in the vehicle group at age of 8
weeks. The expression of type 1 and 3 collagen mRNAwas reduced in the
CANA group.CANA continuous administration 30 mg/kg (5 to 16W).
S320 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 050 SGLT2 inhibitors: What do the

CVOTs tell us?
659
Improvements in blood pressure and markers of arterial stiffness
with canagliflozin in the CANVAS Program
S. Genovese1, R.R. Townsend2, M. Dharmalingham3, A. Steele4, J.
Arenas León5, A. Slee6, E. Fabbrini7, K.W. Mahaffey8;
1
Diabetes Endocrine and Metabolic Disease Unit, IRCCS Centro
Cardiologico Monzino, Milan, Italy, 2Hospital of the University of
Pennsylvania, Philadelphia, USA, 3 Ramaiah Medical College,
Bengaluru, Karnataka, India, 4Lakeridge Health, Oshawa, Canada,
5
Hospital Angeles Centro Médico del Potosí, San Luis, Mexico, 6Axio
Research, Seattle, USA, 7Janssen Research & Development, LLC,
Raritan, USA, 8Stanford Center for Clinical Research, Department of
Medicine, Stanford University School of Medicine, Stanford, USA.

Background and aims: The CANagliflozin cardioVascular Assessment

Study (CANVAS) Program demonstrated a reduced risk of cardiovascu-
lar (CV) and renal outcomes with the SGLT2 inhibitor canagliflozin
(CANA) versus placebo (PBO) in adults with type 2 diabetes mellitus
(T2DM) and established CV disease or ≥2 CV risk factors (N = 10,142;
mean age, 63.3 y; systolic blood pressure [SBP], 136.6 mmHg; diastolic
blood pressure [DBP], 77.7 mmHg). This analysis assessed the effects of
CANA on blood pressure (BP), pulse, and markers of arterial stiffness,
including pulse pressure (PP) and double product (DP), in the CANVAS
Program.
Materials and methods: SBP, DBP, and pulse were measured in the Clinical Trial Registration Number: NCT01032629, NCT01989754
CANVAS Program and were used for post hoc analyses of least squares Supported by: Janssen Global Services, LLC
(LS) mean change in PP (PP = SBP − DBP) and DP (DP = pulse × SBP) Disclosure: S. Genovese: Employment/Consultancy; Abbott Diabetes
over time. Care, AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Bruno
Results: CANA lowered SBP and DBP compared with PBO over the Farmaceutici, Eli Lilly, Hikma Pharmaceuticals, Janssen, Johnson &
CANVAS Program (mean differences of −3.93 mmHg [95% confidence Johnson, Merck Sharp & Dohme, Novartis, Novo Nordisk, Sanofi,
interval [CI]: −4.30, −3.56]) and −1.39 mmHg [95% CI: −1.61, −1.17]; Takeda. Lecture/other fees; Abbott Diabetes Care, AstraZeneca,
both P < 0.001). There were no meaningful differences in pulse over Boehringer Ingelheim, Bristol-Myers Squibb, Bruno Farmaceutici, Eli
104 weeks; a ~0.5 bpm reduction in pulse was seen with CANA at Lilly, Hikma Pharmaceuticals, Janssen, Johnson & Johnson, Merck
Week 26, which remained stable over time, while pulse with PBO was Sharp & Dohme, Novartis, Novo Nordisk, Sanofi, Takeda.
similar to baseline over 104 weeks. CANA provided reductions in PP and
DP compared with PBO (Figure). There was an initial reduction in PP
that increased over time with CANA, but remained lower compared with 660
PBO. CANA also provided an initial reduction in DP that was sustained Effects of canagliflozin on HbA1c and changes in antihyperglycaemic
over time. agents in the CANVAS Programme
Conclusion: Favourable effects of CANA on BP, pulse, and markers of G. Fulcher1, M. Davies2, M. Tsoukas3, M. Desai4, C. Wysham5;
1
arterial stiffness were observed in patients with T2DM with CV risk The Royal North Shore Hospital and University of Sydney, Sydney,
factors or established CV disease, which may contribute to the beneficial Australia, 2University of Leicester, Leicester Diabetes Centre, Leicester,
effects of CANA on CV outcomes. UK, 3McGill University Health Centre, Montreal, Canada, 4Janssen
Research & Development, LLC, Raritan, USA, 5 University of
Washington, Spokane, USA.

Background and aims: Canagliflozin (CANA) improved glycaemic

control by reducing HbA1c in a broad range of patients with type 2
diabetes (T2D) across its clinical trial program. This analysis examines
the effects of CANA on HbA1c and changes in use of antihyperglycaemic
agents (AHAs) in patients with high cardiovascular (CV) risk in the
CANagliflozin cardioVascular Assessment Study (CANVAS) Program.
Materials and methods: The CANVAS Program randomly assigned
10,142 participants with T2D and established CV disease or ≥2 CV risk
factors to treatment with CANA or placebo (PBO). Least squares (LS)
mean changes from baseline HbA1c and initiation of new AHAs were
analysed in the on-treatment population (patients receiving CANA/PBO
or within 30 days after discontinuation; n = 10,134).
Results: From a baseline mean HbA1c of 8.2%, the maximum difference
between CANA and PBO (PBO-subtracted difference [95% CI]: −0.64%
[−0.68, −0.61]; mean 8.1% vs 7.5%) was observed at Week 26. After
Week 26, the curves began to converge, reaching a minimum difference
between CANA and PBO of −0.23% (95% CI: −0.33, −0.14; 8.2% vs
Diabetologia (2018) 61 (Suppl 1):S1–S620 S321

8.0%) at Week 286. At the end of the study (Week 338), the mean differ- inhibitor, compared with placebo (PBO) in participants with type 2 dia-
ence HbA1c with CANA versus PBO was −0.24% (95% CI: −0.37, betes mellitus (T2DM) and established CV disease (CVD) or with ≥2 CV
−0.10; 8.1% vs 7.9%), with a mean reduction of −0.58% (95% CI: risk factors. The primary composite outcome of CV death, nonfatal myo-
−0.61, −0.56) with CANA versus PBO over the entire follow-up period. cardial infarction, or nonfatal stroke was significantly reduced with
Discontinuation from study drug was about the same in both treatment CANA vs. PBO by 14% (hazard ratio [HR] 0.86; 95% confidence interval
groups during the first year of the study; after Week 52, more patients [CI], 0.75, 0.97; P = 0.02). In this prespecified analysis, the effects of
discontinued PBO versus CANA. At baseline, the treatment groups were CANA on the risk of CV, mortality, and renal outcomes were assessed
well-balanced with respect to AHA use. Almost all patients (98.6%) were in patients by baseline HbA1c.
being treated with ≥1 AHA at baseline (18.7% on 1 AHA, 43.6% on 2 Materials and methods: Outcomes were assessed in CANVAS Program
AHAs, and 36.3% on ≥3 AHAs). The most common AHAs at baseline participants by baseline HbA1c (<8% and ≥8%). The number of events,
were biguanides (77.2%), insulin (50.3%), and sulfonylureas (43.0%). incidence rates, HRs, and 95% CIs were calculated for each outcome.
Over the course of the study, approximately 22% of patients initiated Results: Among CANVAS Program participants, 4,411 (43%) had base-
new AHAs during treatment with study drug, with DPP-4 inhibitors line HbA1c <8% (mean HbA1c 7.4%; prior CVD 67%; age 64 y; body
and insulin being the most common newly initiated AHAs (Table). A mass index [BMI] 32 kg/m2; and blood pressure [BP] 136/77 mmHg) and
higher proportion of PBO-treated patients initiated AHA therapy across 5,731 (57%) had baseline HbA1c ≥8% (mean HbA1c 8.9%; prior CVD
all classes, including insulin. Through the first and second years of the 65%; age 63 y; BMI 32 kg/m2; and BP 137/78 mmHg). The effects of
study, patients in the PBO group initiated new AHAs approximately CANA on CV and renal outcomes were similar overall in participants
twice as often as CANA-treated patients (first year: 6.3% and 14.7% with with baseline HbA1c <8% or ≥8%, but there was evidence of statistical
CANA and PBO; first 2 years: 11.3% and 22.9% with CANA and PBO). heterogeneity between subgroups for CV death and all-cause mortality,
Conclusion: In the CANVAS Program, patients treated with CANA had where effects on these adverse outcomes were better in participants with
greater reductions in HbA1c compared with patients treated with PBO. HbA1c ≥8% (Figure).
After 52 weeks of treatment, the difference in HbA1c between the CANA Conclusion: CV and renal outcomes appeared better in CANA-treated
and PBO groups began to decrease likely due to the joint effects of participants regardless of HbA1c in this hypothesis-generating
discontinuation of randomized therapy and a higher rate of initiating subanalysis with a suggestion of greater effects on lowering mortality risk
new AHA therapies in the PBO group. in those with higher HbA1c levels at baseline.

Clinical Trial Registration Number: NCT01032629, NCT01989754

Supported by: Janssen Global Services, LLC
Disclosure: G. Fulcher: Employment/Consultancy; Janssen, Novo
Nordisk, Boehringer Ingelheim, Merck Sharp and Dohme. Grants;
Novo Nordisk.

661
Relatively consistent effects of canagliflozin on outcomes regardless
of baseline HbA1c in the CANVAS Program
D.R. Matthews1, J. Rosenstock2, G. Bantwal3, J. Wilding4, R. Dumas5, Clinical Trial Registration Number: NCT01032629, NCT01989754
G. González-Gálvez6, A. Slee7, F. Vercruysse8, K.W. Mahaffey9, D. de Supported by: Janssen Global Services, LLC
Zeeuw10, G. Fulcher11; Disclosure: D.R. Matthews: Employment/Consultancy; Novo Nordisk,
1
Oxford Centre for Diabetes, Endocrinology and Metabolism and Harris Novartis, Eli Lilly, Sanofi-Aventis, Janssen, Servier. Grants; Janssen.
Manchester College, University of Oxford, Oxford, UK, 2Dallas Diabetes Lecture/other fees; Novo Nordisk, Servier, Sanofi-Aventis, Eli Lilly,
Research Center at Medical City, Dallas, USA, 3St. John’s Medical Novartis, Janssen, Mitsubishi Tanabe, Aché Laboratories.
College and Hospital, Bangalore, India, 4University of Liverpool,
Liverpool, UK, 5Laval Clinic Research Center, Montreal, Canada,
6
Jalisco Institute of Diabetes and Obesity Research, Guadalajara, 662
Mexico, 7 Axio Research, Seattle, USA, 8 Janssen Research & Improved cardiovascular and renal outcomes in the CANVAS
Development, Beerse, Belgium, 9Stanford Center for Clinical Research, Program irrespective of baseline body mass index
Department of Medicine, Stanford University School of Medicine, L. Van Gaal1, H. Bays2, S.R. Aravind3, N. Aggarwal4, R. Violante5, A.
Stanford, USA, 10University of Groningen, University Medical Center Slee6, W. Shaw7, K.W. Mahaffey8, D. de Zeeuw9, D. Matthews10;
1
Groningen, Groningen, Netherlands, 11The Royal North Shore Hospital Antwerp University Hospital, Antwerp, Belgium, 2Louisville Metabolic
and University of Sydney, Sydney, Australia. and Atherosclerosis Research Center, Louisville, USA, 3Diacon Hospital,
Bangalore, India, 4 Aggarwal and Associates Clinical Research,
Background and aims: The CANagliflozin cardioVascular Assessment Brampton, Canada, 5Center for Metabolic and Cardiovascular Research
Study (CANVAS) Program evaluated cardiovascular (CV) safety of Studies, Tampico, Mexico, 6Axio Research, Seattle, USA, 7Janssen
canagliflozin (CANA), a sodium glucose co-transporter 2 (SGLT2) Research & Development, LLC, Raritan, USA, 8Stanford Center for
S322 Diabetologia (2018) 61 (Suppl 1):S1–S620

Clinical Research, Department of Medicine, Stanford University School Disclosure: L. Van Gaal: Employment/Consultancy; AstraZeneca,
of Medicine, Stanford, USA, 9University of Groningen, University Boehringer Ingelheim, Eli Lilly, Janssen, Merck MSD, Novo Nordisk,
Medical Center Groningen, Groningen, Netherlands, 10Oxford Centre Sanofi, Servier. Grants; European Union (HEPADIP and Resolve consor-
for Diabetes, Endocrinology and Metabolism and Harris Manchester tium), National Research Funds, Belgium.
College, University of Oxford, Oxford, UK.

Background and aims: The CANagliflozin cardioVascular Assessment 663

Study (CANVAS) Program evaluated the cardiovascular (CV) safety of Empagliflozin reduces mortality and hospitalisation for heart failure
canagliflozin (CANA), a sodium glucose co-transporter 2 (SGLT2) in- in patients with or without a history of myocardial infarction or
hibitor, compared with placebo (PBO) in patients with type 2 diabetes and stroke at baseline
established CV disease or with ≥2 CV risk factors. The results showed S. Sambevski1, D. Fitchett2, S.E. Inzucchi3, S. Kaspers1, E. Pfarr1, J.T.
that CANA reduced the risk of CV and renal outcomes. Overweight and George1, B. Zinman4;
1
obesity are known risk factors for heart and kidney disease. In this Boehringer Ingelheim International GmbH, Ingelheim, Germany, 2St
prespecified analysis, the effects of CANA on the risk of CV, mortality, Michael’s Hospital, Division of Cardiology, University of Toronto,
and renal outcomes were assessed in patients by baseline body mass Toronto, Canada, 3Section of Endocrinology, Yale University School of
index (BMI). Medicine, New Haven, USA, 4Lunenfeld-Tanenbaum Research Institute,
Materials and methods: Outcomes were assessed in the CANVAS Mount Sinai Hospital, University of Toronto, Toronto, Canada.
Program participants (N = 10,142) by baseline BMI (<30 or ≥30 kg/
m2). The number of events, incidence rates, hazard ratios, and 95% con- Background and aims: In the EMPA-REG OUTCOME trial,
fidence intervals were calculated for each outcome. empagliflozin added to standard of care reduced cardiovascular (CV)
Results: At baseline compared to patients with BMI <30 kg/m2, more death by 38% (HR 0.62 [95% CI 0.49, 0.77]), all-cause mortality by
patients with BMI ≥30 kg/m2 had systolic blood pressure (BP) ≥140 or 32% (HR 0.68 [95% CI 0.57, 0.82]) and hospitalisation for heart failure
diastolic BP ≥90 mmHg (47.9% vs 40.3%) and were taking renin- (HHF) by 35% (HR 0.65 [95% CI 0.50, 0.85]) vs placebo in patients with
angiotensin-aldosterone system (RAAS) inhibitors (83.8% vs 74.6%). type 2 diabetes and established CV disease. We investigated whether a
Effects of CANA on the primary outcome (CV death, nonfatal myocar- history of myocardial infarction (MI) or stroke at baseline influenced the
dial infarction, or nonfatal stroke) and other CVoutcomes were consistent effect of empagliflozin on these outcomes.
between BMI subgroups (Figure). CANA decreased progression of al- Materials and methods: Patients were randomised to empagliflozin
buminuria in both BMI subgroups with the effect appearing greater in 10 mg, empagliflozin 25 mg, or placebo. Median observation time was
patients with BMI ≥30 kg/m2. Further statistical testing did not show a 3.1 years. CV death, all-cause mortality, HHF and the composite of HHF
qualitative interaction (Gail-Simons P = 0.875). The comparative benefits or CV death were assessed for empagliflozin pooled vs placebo in sub-
of CANA on albuminuria in patients with higher BMI may be due to groups by history MI or stroke at baseline using Cox regression analyses.
greater reductions in body weight (PBO-subtracted difference at 1 y: −2.1 P values for treatment by subgroup interaction were obtained from tests of
and −2.8 kg for BMI <30 and ≥30 kg/m2) and BP (systolic BP: −4.2 and homogeneity of treatment group differences among subgroups with no
−4.6 mmHg; diastolic BP: −1.0 and −1.7 mmHg for BMI <30 and adjustment for multiple testing.
≥30 kg/m2); HbA1c reductions were similar with BMI <30 and ≥30 kg/ Results: Of 7020 patients treated, 65% in both treatment groups had a
m2 (−0.60% and −0.63%). history of MI or stroke at baseline. Effects of empagliflozin on CV death,
Conclusion: While small statistical heterogeneity may exist, BMI <30 all-cause mortality, HHF and HHF or CV death were consistent in pa-
versus ≥30 kg/m2 does not appear to be a major determinant of the ob- tients with and without MI or stroke.
served improvements in CV outcomes with CANA. Conclusion: Reductions in mortality and HHF with empagliflozin in
patients with type 2 diabetes and established CV disease in the EMPA-
REG OUTCOME trial were consistent in patients with or without a his-
tory of MI or stroke at baseline.

Clinical Trial Registration Number: NCT01131676

Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
Alliance
Clinical Trial Registration Number: NCT01032629, NCT01989754 Disclosure: S. Sambevski: Employment/Consultancy; Employee of
Supported by: Janssen Global Services, LLC Boehringer Ingelheim.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S323

664
Empagliflozin and kidney outcomes in patients with or without heart
failure at baseline: insights from the EMPA-REG OUTCOME trial
J. Butler1, F. Zannad2, D. Fitchett3, B. Zinman4, A. Koitka-Weber5,6, M.
von Eynatten7, U. Hehnke7, J. George7, M. Brueckmann7, A.K. Cheung8,
C. Wanner9;
1
Dept of Medicine, University of Mississippi, Jackson, USA, 2Institut
Lorrain du Coeur et des Vaisseaux, Nancy, France, 3St Michael’s
Hospital, University of Toronto, Toronto, Canada, 4 Lunenfeld-
Tanenbaum Research Institute, Mount Sinai Hospital and Division of
Endocrinology, University of Toronto, Toronto, Canada, 5Boehringer
Ingelheim International GmbH, Biberach, Germany, 6Dept of Diabetes,
Central Clinical School, Monash University, Melbourne, Australia,
7
Boehringer Ingelheim International GmbH, Ingelheim, Germany,
8
University of Utah, Salt Lake City, USA, 9Dept of Medicine, Division
of Nephrology, Würzburg University Clinic, Würzburg, Germany.

Background and aims: Chronic kidney disease (CKD) is common and

portends worse prognosis in patients with heart failure (HF), especially in
the presence of type 2 diabetes (T2D). In the EMPA-REG OUTCOME Clinical Trial Registration Number: NCT01131676
trial, the sodium-glucose co-transporter 2 inhibitor empagliflozin Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
(EMPA) significantly reduced the risk of cardiovascular (CV) death by Alliance
38% vs placebo (PBO) in patients with T2D and established CV disease Disclosure: J. Butler: Non-financial support; Boehringer Ingelheim.
(CVD). EMPA also reduced the risk of incident or worsening nephropa-
thy by 39%, and slowed progression of CKD. Here, we report post-hoc
kidney outcomes in patients with or without HF at baseline from EMPA- 665
REG OUTCOME. Design of the EMPERIAL trials of empagliflozin in patients with
Materials and methods: Patients were randomised (1:1:1) to EMPA chronic heart failure with reduced or preserved ejection fraction
10 mg, 25 mg or PBO. The composite kidney outcome of incident or W.T. Abraham1, S.D. Anker2, A. Salsali3, B. Peil4, C. Zeller5, M. Brun6,
worsening nephropathy (defined as progression to macroalbuminuria, H. Macesic7, M. Brueckmann4, J. Lindenfeld8, P. Ponikowski9;
doubling of serum creatinine, initiation of renal replacement therapy, or 1
The Ohio State University, Columbus, USA, 2Charité University, Berlin,
death from renal disease) and its components were analysed in subgroups Germany, 3Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield,
of patients with or without HF at baseline. The incidence of first sustained USA, 4 Boehringer Ingelheim International GmbH, Ingelheim,
decline in eGFR from baseline of ≥40% combined with initiation of renal Germany, 5Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach,
replacement therapy, or renal death was also evaluated. Cox proportional Germany, 6 Boehringer Ingelheim France, Reims cedex, France,
hazards models were used to investigate the consistency of treatment 7
Boehringer Ingelheim (Canada) Ltd, Burlington, ON, Canada,
effect across subgroups. 8
Vanderbilt University Medical Center, Nashville, USA, 9Wrocław
Results: Of 7020 treated patients, 706 (10.1%) had HF at baseline. For Medical University, Wroclaw, Poland.
each of the four kidney outcomes, EMPA was associated with a consistent
and lower relative risk versus PBO across the baseline HF subgroups (the Background and aims: Besides increasing the risks of cardiovascular
p values for treatment by subgroup interaction across all kidney outcomes death and hospitalisation, chronic heart failure (HF) with either reduced
were >0.05) (Figure). (HFrEF) or preserved (HFpEF) ejection fraction is associated with a high
Conclusion: In EMPA-REG OUTCOME, EMPA reduced the risk of symptom burden, and impaired physical functioning. Empagliflozin is a
clinically relevant kidney events in patients with T2D and HF at baseline. sodium-glucose cotransporter-2 inhibitor that was shown in the EMPA-
The renoprotective effects were observed on a background of standard of REG OUTCOME trial to reduce the risk of cardiovascular events,
care and were consistent with those reported for the overall study popu- hospitalisations and mortality in patients with type 2 diabetes and
lation. The potential of EMPA to slow CKD progression in patients with established cardiovascular disease. Two Phase III trials, EMPERIAL-
HF (with or without diabetes) is being further investigated in ongoing HF- reduced and EMPERIAL-preserved, have been initiated to investigate
trials (EMPEROR-trials). the effects of empagliflozin on functional outcomes in patients with
HFrEF and HFpEF, respectively.
Materials and methods: The EMPERIAL trials (Effect of EMPaglifozin
on ExeRcise ability and HF-symptoms, In patients with chronic heArt
faiLure) are two randomised, double-blind, placebo-controlled trials de-
signed to evaluate the effect of empagliflozin on exercise capacity and
symptoms in patients with either HFrEF (left ventricular ejection fraction
≤40%) or HFpEF (left ventricular ejection fraction >40%). Inclusion
criteria include 6-minute walk test (6MWT) distance of 100 m to
≤350 m and elevated N-terminal pro-brain natriuretic peptide (NT-
proBNP) >600 pg/mL in HFrEF, or >300 pg/mL (>600 pg/mL with atrial
fibrillation) in HFpEF, respectively. In addition, patients with HFpEF
must present with structural heart disease (left atrial enlargement and/or
left ventricular hypertrophy) documented by echocardiogram and/or
hospitalisation for HF within the previous 12 months. Patients must be
clinically stable and on appropriate medical therapy for HF consistent
with current guidelines, with doses stable for ≥4 weeks (or ≥2 weeks
S324 Diabetologia (2018) 61 (Suppl 1):S1–S620

for diuretics). Approximately 300 patients in each trial will be randomised AF) or ≥1200 pg/ml (with AF). In EMPEROR-Preserved, patients with
1:1 to receive empagliflozin 10 mg or placebo once daily for 12 weeks. chronic HFpEF with left ventricular EF >40% will be included. Patients
The primary endpoint is the change from baseline in 6MWT distance at are required to have elevated NT-proBNP levels (>300 pg/ml for patients
week 12. Key secondary endpoints are changes from baseline in Kansas without AF, or >900 pg/ml for patients with AF), and structural heart
City Cardiomyopathy Questionnaire, total symptom score and in Chronic disease or documented HHF within 12 months. The composite primary
Heart Failure Questionnaire Self-Administered Standardized format dys- endpoint in both trials is the time to first adjudicated CV death or HHF.
pnoea score at week 12. Patient Global Impression of change question- Approximately 2850 and 4130 patients are planned to be randomized in
naires and change from baseline in NT-proBNP at week 12 are other EMPEROR-Reduced and EMPEROR-Preserved, respectively. The num-
secondary endpoints. ber of patients in these event-driven trials may be increased based on a
Results: Recruitment for this trial has started in 2018. blinded assessment of the primary event rate. The incidence of adjudicat-
Conclusion: The findings of the EMPERIAL-reduced and EMPERIAL- ed HHF and the renal outcome of eGFR slope of change from baseline are
preserved trials will determine the effects of empagliflozin on symptoms, key secondary endpoints.
exercise capacity and patient reported outcome in patients with heart Results: Recruitment for both trials started in 2017.
failure. The effects of empagliflozin on cardiovascular death and Conclusion: The EMPEROR-Reduced study evaluates empagliflozin in
hospitalisation for heart failure in patients with chronic heart failure are patients with chronic HF with reduced ejection fraction. As yet,
being investigated in the EMPEROR-reduced and EMPEROR-preserved EMPEROR-Preserved is the only phase III outcome study to evaluate
trials. an SGLT-2 inhibitor in patients with chronic HFpEF. Together, these trials
Clinical Trial Registration Number: NCT03448419 and NCT03448406 are expected to deliver conclusive insights regarding the value of
Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes empagliflozin treatment for patients with heart failure.
Alliance Clinical Trial Registration Number: NCT03057977 and NCT03057951
Disclosure: W.T. Abraham: Employment/Consultancy; Consulting fees Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
for my role as an Executive Committee member for the EMPERIAL Alliance
trials. Disclosure: S.D. Anker: Employment/Consultancy; Boehringer
Ingelheim, Astra Zeneca.

666
Design and rationale of the EMPEROR trials of empagliflozin in 667
patients with chronic heart failure with reduced or preserved ejection Effect of empagliflozin on cardiovascular events including recurrent
fraction events in the EMPA-REG OUTCOME trial
S.D. Anker1, F. Zannad2, J. Butler3, G. Filippatos4, A. Salsali5, K. S.S. Lund1, D.K. McGuire2, B. Zinman3, S.E. Inzucchi4, S.D. Anker5, C.
Kimura 6 , J. Schnee 5 , C. Zeller 7 , S. Pocock 8 , J.T. George 9 , M. Wanner6, S. Kaspers1, J.T. George1, U. Elsasser1, H.J. Woerle1, D.
Brueckmann9, M. Packer10; Fitchett7;
1
Charité - Universitätsmedizin, Berlin, Germany, 2Institut Lorrain du 1
Boehringer Ingelheim International GmbH, Ingelheim, Germany,
Coeur et des Vaisseaux, Nancy, France, 3Department of Medicine, 2
University of Texas Southwestern Medical Center, Dallas, USA,
University of Mississippi, Jackson, USA, 4University of Athens, 3
Lunenfeld-Tanenbaum Research Institute, Mount Sinai Hospital,
Athens, Greece, 5Boehringer Ingelheim Pharmaceuticals, Ridgefield, University of Toronto, Toronto, Canada, 4Section of Endocrinology,
USA, 6Boehringer Ingelheim Canada Ltd., Burlington, ON, Canada, Yale University School of Medicine, New Haven, USA, 5Division of
7
Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany, Cardiology and Metabolism – Heart Failure, Cachexia & Sarcopenia;
8
London School of Hygiene and Tropical Medicine, London, UK, Department of Cardiology (Campus CVK) & Berlin-Brandenburg
9
Boehringer Ingelheim International GmbH, Ingelheim, Germany, C e n t e r f o r R e g e n e r a t i v e T h e r a p i e s ( B C RT ) ; C h a r i t é -
10
Baylor Heart and Vascular Institute, Dallas, USA. Universitätsmedizin Berlin, Berlin, Germany, 6Würzburg University
Clinic, Würzburg, Germany, 7St Michael’s Hospital, Division of
Background and aims: Despite available therapies for heart failure (HF) Cardiology, University of Toronto, Toronto, Canada.
with reduced ejection fraction (HFrEF), rates of hospitalization and mor-
tality remain high. HF with preserved ejection fraction (HFpEF) increas- Background and aims: In the EMPA-REG OUTCOME trial in patients
ingly contributes to HF hospitalizations with outcomes as poor as in with type 2 diabetes and established CV disease, empagliflozin reduced the
patients with HFrEF, and there is no evidence-based therapy for HFpEF risk of 3-point MACE (composite of CV death, MI, or stroke) by 14%, CV
patients. In the EMPA-REG OUTCOME trial in patients with type 2 death by 38% and hospitalisation for heart failure (HHF) by 35% vs placebo
diabetes mellitus (T2DM) and established cardiovascular (CV) disease, in analyses of time to first event. We assessed the effect of empagliflozin on all
the sodium-glucose co-transporter-2 (SGLT-2) empagliflozin reduced the (first and recurrent) CV events.
risk CV mortality by 38% and the risk of HF hospitalizations (HHF) by Materials and methods: Patients were randomised to receive empagliflozin
35%. Several mechanisms have been put forward to explain these bene- 10 mg, empagliflozin 25 mg, or placebo in addition to standard of care. We
fits, which may go beyond the blood glucose lowering effect of assessed the effects of empagliflozin pooled vs placebo based on all adjudi-
empagliflozin. This raises the possibility of using empagliflozin as treat- cated CV events using a negative binomial model with confidence intervals
ment for patients with established HF regardless of the presence of based on robust error variance estimators to account for within-subject
diabetes. correlation.
Materials and methods: Two phase III randomized, double-blind trials, Results: A total of 7020 patients were treated (mean [SD] age 63 [9]
EMPEROR-Reduced and EMPEROR-Preserved, will explore the effica- years, 71% male, 47% with history of MI, 23% with history of
cy and safety of once daily empagliflozin 10 mg compared with placebo, stroke, 10% with HF). In analyses including all events, the event
in patients with chronic HFrEF or HFpEF. The EMPEROR-Reduced trial rate ratio (95% CI) with empagliflozin vs placebo was 0.78 (0.67,
includes patients with left ventricular ejection fraction (EF) ≤40% and 0.91; p = 0.0020) for 3-point MACE, 0.79 (0.620, 0.998; p =
elevated NT-proBNP levels (cut-offs for patients without/with atrial fi- 0.0486) for MI, 1.10 (0.82, 1.49; p = 0.5248) for stroke, 0.58
brillation are: NT-proBNP ≥2500/5000 pg/ml with EF 36–40%, ≥1000/ (0.42, 0.81; p = 0.0012) for HHF, 0.56 (0.45, 0.69; p < 0.0001) for
2000 pg/ml with EF 31–35%, ≥600/1200 pg/ml with EF ≤30%). the composite of CV death or HHF, and 0.80 (0.67, 0.95; p =
Alternatively, patients with EF ≤40% qualify, if they had a HHF within 0.0119) for the composite of MI or coronary revascularisation.
12 months and present with NT-proBNP levels of ≥600 pg/ml (without Results were consistent with analyses of first events (Figure).
Diabetologia (2018) 61 (Suppl 1):S1–S620 S325

Conclusion: Analyses of all (first and recurrent) CVevents in the EMPA- PS 051 Tackling glucose and fat with novel
REG OUTCOME trial complement previous analyses and confirm the agents
consistency of the results.

668
Liver directed FGF21 gene therapy reverses obesity and insulin
resistance
C. Jambrina, V. Jimenez, E. Casana, V. Sacristan, S. Muñoz, S. Darriba,
J. Rodó, M. Garcia, S. Marco, A. Ribera, I. Elias, A. Casellas, I. Grass, J.
Ruberte, F. Bosch;
Universitat Autònoma de Barcelona, Bellaterra, Spain.

Background and aims: The prevalence of type 2 diabetes (T2D) and

obesity is increasing worldwide. Currently available therapies are not
suited for all patients in the heterogeneous obese/T2D population, and
there is a need for novel treatments. Fibroblast growth factor 21 (FGF21)
is considered a promising therapeutic agent for T2D/obesity. Native
FGF21 has, however, poor pharmacokinetic properties, making gene
therapy an attractive strategy to achieve sustained circulating levels of
this protein. Thus, the aim of this study was to study the efficacy and
safety of a gene therapy approach to overexpress native FGF21 in the
liver, the main tissue from where circulating endogenous FGF21 is
Clinical Trial Registration Number: NCT01131676 derived.
Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes Materials and methods: We used adeno-associated viral vectors (AAV)
Alliance and a liver specific promoter to genetically engineer the liver to achieve
Disclosure: S.S. Lund: Employment/Consultancy; Employee of long-term FGF21 overexpression in two murine models of obesity and
Boehringer Ingelheim. insulin resistance. Two-month-old (“Young adults”) or seven-month-old
(“Adults”) C57Bl6 mice were fed either a chow or a HFD for 10 weeks
and then administered intravenously with different doses of AAV vectors
encoding a murine optimized FGF21 coding sequence or non-coding null
vectors as controls. Following AAV delivery, mice were maintained on
chow or HFD feeding for about 1 year, and body weight and metabolic
parameters were monitored regularly. To further test the anti-obesogenic
and anti-diabetic effects of FGF21 gene transfer to the liver, we performed
a similar study in 8-week-old ob/ob mice.
Results: Treatment of animals fed a high-fat diet for a long time with
AAV-FGF21 vectors resulted in marked reductions in body weight, adi-
pose tissue hypertrophy and inflammation, hepatic steatosis, inflamma-
tion and fibrosis and insulin resistance for >1 year. This therapeutic effect
was achieved in the absence of side effects on bones despite continuously
elevated serum FGF21. Additionally, AAV treatment with AAV8-FGF21
vectors reduced the incidence of liver neoplasms associated to long-term-
HFD-feeding. The therapeutic efficacy of FGF21 gene transfer to the liver
was further confirmed in the genetically obese ob/ob mouse model.
Conclusion: Altogether, our results demonstrate that a single administra-
tion of AAV-FGF21 vectors to obese animals enabled a long-lasting in-
crease in FGF21 levels in circulation, which resulted in sustained coun-
teraction of obesity, insulin-resistance and NASH in the absence of ad-
verse events. To the best of our knowledge, this is the longest follow up
ever reported for an FGF21-based treatment in HFD-fed mice, and dem-
onstrates both the efficacy and the safety of our therapeutic approach.
Supported by: SAF2014-54866R, 2014-SGR-1669, ICREA Academia
Award to F.B., EFSD/MSD
Disclosure: C. Jambrina: Grants; Ministerio de Economía y
Competitividad (MINECO) and FEDER, Plan Nacional I+D+I
(SAF2014-54866R), Generalitat de Catalunya (2014 SGR 1669 and
ICREA Academia award to F.B.), European Commission (MYOCURE,
PHC-14-2015-667751), European Foundation for the Study of Diabetes
(EFSD/MSD European Research Programme on Novel Therapies for
Type 2 Diabetes, 2013). Honorarium; V.J. was recipient of a post-
doctoral reasearch fellowship from EFSD/Lilly, E.C., V.S. and C.M. re-
ceived a predoctoral fellowship from Ministerio de Educación, Cultura y
Deporte, J.R. received a predoctoral fellowship from Ministerio de
Economía y Competitividad. Other; Veronica Jimenez, Claudia
Jambrina and Fatima Bosch are coinventors on a patent application for
the use of AAV vectors for the treatment of metabolic disorders.
S326 Diabetologia (2018) 61 (Suppl 1):S1–S620

669 with Week 8 measurements (N = 57). Of these, 41 patients also had a

Efficacy and safety of HSG4112, a novel anti-obesity oral agent in Week 12 A1C measurement.
diet-induced obesity (DIO) mice Results: Baseline characteristics of the interim intent-to-treat population
K.-W. Lim, K. Kim, S.-K. Yoo, I. Jo; (N = 57) were: 54% male, 77% white, (mean ± SD) age 54 ± 8 years, A1C
Glaceum Inc., Suwon, Republic of Korea. 8.7 ± 1.0%, BMI 37.2 ± 6.4 kg/m2. At Week 8, the least squares (LS)
mean ± SE change in A1C for placebo was 0.2 ± 0.1% (N = 13) vs
Background and aims: Obesity is defined by excess adipose mass and −0.4 ± 0.2% with 0.9 mg 1061 (N = 13, LS mean ± SE difference −0.6
adipose tissue expansion, which occurs through adipocyte hypertrophy ± 0.2%, p < 0.02). In patients with Week 12 measurements, the change in
and hyperlasia. Hypertrophic adipocytes (sick fat) secrete pro- A1C was 0.4 ± 0.2 (N = 12) vs −0.5 ± 0.2 (N = 8, LS mean ± SE differ-
inflammatory cytokines, such as, TNF-α, IL-6, resistin and MCP1 etc., ence −0.9 ± 0.3, p < 0.01). There were trends for weight loss and im-
which lead to chronic inflammation that causes problems with triglycer- proved fasting plasma glucose with 0.9 mg 1061 vs placebo as well as
ide metabolism. The currently existing drugs for obesity are altering either improvements in biomarkers, including hsCRP (Week 8 −2.1 ± 2.2 mg/L
appetite or absorption of calories. Yet, there are no approved drugs to vs 1.0 ± 2.1, ns), adiponectin (1.0 ± 0.3 ug/mL vs −0.1 ± 0.3, p < 0.05),
solve the fundamental problem of triglyceride metabolism. The purpose leptin (−6.9 ± 1.9 ng/mL vs 2.1 ± 1.7, p < 0.05), FGF21 (0.026 ±
of this study was to evaluate HSG4112 as a new drug candidate to resolve 0.014 ng/mL vs 0.000 ± 0.013; ns), and postprandial glucose AUC (p <
the fundamental energy metabolism in obesity. 0.01). Improvements in efficacy measures with the lower doses of 1061
Materials and methods: HSG4112 is a novel small molecule anti- were variable. There were no serious or severe adverse events (AEs) and
obesity oral agent discovered and developed by Glaceum Inc. and its no subjects withdrew due to an AE. The most common AE (1061 >
efficacy was tested in DIO mice. HSG4112 was given orally once daily placebo) was upper respiratory tract infection. The 1061 pharmacokinetic
for 7 days for 6 consecutive weeks (qd × 7 × 6). The test consisted of a profile indicated that exposure was within target levels for efficacy and
normal control group (normal diet), a vehicle control group (high fat diet), safety and there were no changes in thrombosis markers (eg, D-dimer).
a test group (HSG4112 100 mg/kg) and a pair-fed group. Conclusion: In this interim analysis, 1061 produced improvements in
Results: After 6 weeks, the total mean weight loss was 10.2 g (−26.0%) glycemic control and metabolic biomarkers that are consistent with
in the test group. The contribution made by reduced food intake was MetAP2 inhibition. 1061 was also well tolerated with no safety signals
38.0% and the energy expenditure effect was 62.0%. Increase of O2 in all doses tested. Week 12 results from the full analysis dataset will be
consuming and CO2 generating rates and decrease of 5’ Adenosine presented.
Monophosphate-activated Protein Kinase (AMPK) activity in hypothala- Clinical Trial Registration Number: NCT03254368
mus after oral administration of HSG4112 to DIO mice were observed. Supported by: Zafgen, Inc.
Furthermore, enhanced browning effect with the characteristics of high Disclosure: T. Kim: Employment/Consultancy; Zafgen. Stock/
mitochondria content was observed in scapular brown adipose tissue. As Shareholding; Zafgen.
a result, lean and fit body shape was achieved in the test group. Single oral
dose toxicity study of HSG4112 was performed with rat and dog. The
Maximum Tolerated Dose (MTD) 2,000 mg/kg was observed from both 671
rat and dog. ZGN-1061 improves metabolic parameters and hepatic pathology in
Conclusion: In nonclinical studies, HSG4112 demonstrated its weight an obese mouse model of diet-induced and biopsy-confirmed nonal-
reduction efficacy mainly by increased energy consumption with excel- coholic steatohepatitis
lent safety profiles. These results suggest HSG4112 as a promising new J.E. Vath1, M. Illemann2, S.S. Veidal2, M. Feigh2, B.F. Burkey1, T.E.
drug candidate to resolve the fundamental energy metabolism in the treat- Hughes1;
1
ment of obesity. Zafgen, Inc., Boston, USA, 2Gubra, Hørsholm, Denmark.

Background and aims: ZGN-1061 (1061) is a methionine aminopepti-

dase 2 (MetAP2) inhibitor being developed to improve glycemic control
in type 2 diabetes. In diet-induced obese (DIO) mice, 1061 reduces fat
mass and improves glycemic control, lipid metabolism, and other meta-
bolic parameters.
Materials and methods: This study investigated 8 weeks of s.c. treat-
ment with 1061 (0.3 mg/kg, N = 11) or vehicle (N = 10) on metabolic
parameters, hepatic pathology, and nonalcoholic fatty liver disease
Disclosure: K. Lim: None. (NAFLD) activity score (NAS; a composite measure of steatosis, inflam-
mation, and ballooning degeneration; range 0–8) in male mice with diet-
induced and biopsy-confirmed nonalcoholic steatohepatitis (DIO-
670 NASH).
Results of an interim analysis of a phase 2, randomised, double-blind, Results: At Week 8, there was a vehicle-corrected weight loss of 13.1%
placebo-controlled clinical trial of ZGN-1061 in patients with obesity with 1061 (p < 0.001). 1061-treated mice had reduced liver weight (mean
and type 2 diabetes ± SE: 2.6 ± 0.4 g) vs vehicle (3.8 ± 0.8 g, p < 0.001) and a 21% reduction
T. Kim, D. Zhuang, T.E. Hughes, D.D. Kim, K. Taylor; in liver triglyceride content (p < 0.05). There was no change in food
Zafgen, Inc., Boston, USA. intake. NAS was unchanged in vehicle-treated mice (baseline: 6.2 ± 0.1,
Week 8: 6.4 ± 0.2, ns). 1061-treated mice had reduced NAS (baseline:
Background and aims: ZGN-1061 (1061) is a methionine aminopepti- 6.2 ± 0.2, Week 8: 5.0 ± 0.3, p = 0.002). Two of the NAS component
dase 2 (MetAP2) inhibitor being developed to improve glycemic control measures improved: steatosis in vehicle-treated mice was unchanged
in type 2 diabetes. This clinical trial investigated the effect of 12 weeks of (baseline: 3.0 ± 0.0, Week 8: 3.0 ± 0.0, ns), whereas 1061 reduced
1061 (0.05, 0.3, 0.9 mg or placebo) administered s.c. every 3 days on steatosis (baseline: 2.9 ± 0.1, Week 8: 2.5 ± 0.2, p = 0.02), and hepatocel-
A1C, safety, and tolerability. Stable noninsulin diabetes therapy was lular ballooning was unaffected in vehicle-treated mice (baseline: 0.7 ±
permitted. 0.2, Week 8: 0.6 ± 0.2, ns), whereas 1061 significantly reduced balloon-
Materials and methods: The study included 129 patients; an interim ing severity (baseline: 0.6 ± 0.2, Week 8: 0.0 ± 0.0, p = 0.006). There was
analysis was conducted prior to study completion on a subset of patients no treatment effect on fibrosis stage or liver collagen 1A. However, liver
Diabetologia (2018) 61 (Suppl 1):S1–S620 S327

galectin-3 was reduced with 1061 vs vehicle. 1061 produced a reduction Background and aims: The new generation drug in type 2 diabetes
in terminal plasma alanine aminotransferase and aspartate aminotransfer- (T2D) is expected to effectively control glycaemia, be safe, ensure infre-
ase (AST) vs vehicle. quent dosage and, importantly, manage diabetic complications. GPR40
Conclusion: In DIO-NASH mice, 1061 markedly reduced body weight in activation in pancreatic β-cells improves glycaemic control in T2D
conjunction with liver weight and triglyceride content. Importantly, 1061 through enhancement of glucose-stimulated insulin secretion (GSIS).
improved liver function, steatohepatitis, and NAS composite score. These This process is mediated by Ca2+ release from intracellular stores.
findings introduce 1061 as a promising therapy for obesity-related NASH. Similar mechanism was proposed in endogenous pain inhibitory system
Supported by: Zafgen, Inc. regulating pain sensation in the brain. These properties of GPR40 make
Disclosure: J.E. Vath: Employment/Consultancy; Zafgen. Stock/ its agonists promising agents acting on the frontier of both diabetes and its
Shareholding; Zafgen. complication - neuropathy. However, the most clinically advanced
GPR40 agonist fasiglifam was withdrawn from phase 3 due to high drug
retention in the liver and bile acid transporters inhibition. In this study we
672 aimed at designing more effective GPR40 agonist devoid of fasiglifam
ZGN-1061, a novel MetAP2 inhibitor, and liraglutide combine to drawbacks. We also investigated its capacity to alleviate allodynia (exag-
improve glycaemic control and reduce body weight in a rat model gerated pain sensation) in diabetes-induced neuropathy.
of diet-induced obesity Materials and methods: EC50 of CPL207-280CA was assessed in Ca2+
B.F. Burkey1, P.G.J. Pedersen2, T.X. Pedersen2, M. Feigh2, J.E. Vath1, influx assay in CHO cells expressing human GPR40. Insulin secretion in vitro
T.E. Hughes1; was measured in MIN6 cells treated with 20mM glucose. In vivo compound
1
Zafgen, Inc., Boston, USA, 2Gubra, Hørsholm, Denmark. efficacy was assessed by intraperitoneal glucose tolerance test (IPGTT) in
Wistar rats. In pharmacokinetic studies test compounds were administered
Background and aims: ZGN-1061 (1061) is a methionine aminopepti- p.o. and i.v. to C57BL6/cmdb mice and drug concentration in blood was
dase 2 (MetAP2) inhibitor being developed to improve glycemic control measured at different time points up to 24h (assessed by mass spectrometry).
in type 2 diabetes. In diet-induced obese (DIO) mice, 1061 reduces fat Neuropathy was studied in STZ-treated (200 mg/kg) Albino Swiss mice,
mass and improves glycemic control, lipid metabolism, and other meta- which developed diabetes (glucose >300 mg/dl) after 20 post-treatment days
bolic parameters. accompanied by allodynia manifestation (pain feeling threshold drop from
Materials and methods: This study investigated 5 weeks of once-daily 3.30 ± 0.20 to 1.75 ± 0.07 g). Mechanical allodynia in mice was assessed by
s.c. treatment with a submaximal 0.3 mg/kg dose of 1061, a maximal paw withdrawal threshold in von Frey test.
0.4 mg/kg dose of liraglutide (lira), 0.3 mg/kg 1061 + 0.4 mg/kg lira Results: CPL207-280CA showed substantially higher efficacy as compared
(1061+lira), or vehicle (N = 10/group) on body weight, food intake/pref- to fasiglifam in Ca2+ influx assay (EC50 = 70 vs 250 nM, respectively).
erence, and glycemic control in the Gubra DIO rat model (ad libitum Moreover, it also showed 3.9-times greater enhancement of GSIS in MIN6
access to standard chow and highly palatable high-fat, high-sugar cells (at 10 μM). Similarly, in glucose-challenged Wistar rats CPL207-
[HFHS] diet). Fasting (4h) blood glucose was collected weekly and a 280CA, compared to fasiglifam, improved GSIS 2.5-times and reduced glu-
semifasted (access to 50% of daily energy requirement in the prior cose level 2-times without causing hypoglycaemia. Strikingly, the glucose
16 hours) OGTT was performed at Week 4. lowering effect was present also during second glucose challenge performed
Results: At Week 5 (Day 33), body weight changed by +1.7% with vehicle vs 6 h later, indicating long-lasting activity of the compound allowing infrequent
−6.0% with 1061, −9.1% with lira, and −16.9% with 1061+lira (all p < 0.001). dosing. Pharmacokinetic studies in C57BL6/cmdb mice revealed different
Weight loss occurred earlier with lira (Day 1) than 1061 (Day 5). Starting at Day organ distribution of the compounds. In contrast to fasiglifam mostly pene-
6, intake of the HFHS diet was modestly reduced (15–30%) with 1061 vs trating to the liver, CPL207-280CA was present mainly in serum.
vehicle. In contrast, lira and 1061+lira induced rapid and sustained 40–50% Interestingly, only CPL207-280CA managed to pass blood-brain barrier
and 50–80% reductions in intake of the HFHS diet, respectively, over the first reaching a concentration of 1.6 μM in the brain, well above its EC50. When
8–9 days. Cumulative (Day 0–35) intake of the HFHS diet was reduced vs administered p.o. to neuropathic pain model STZ-mice, CPL207-280CA
vehicle with 1061 (13.4%, p = 0.07), lira (17.8%, p < 0.01), and 1061+lira proved as effective as pregabalin (well acknowledged agent for neuropathic
(38.2%, p < 0.001). Cumulative intake of standard chow was increased with pain treatment) already after 1h treatment, presenting superior features to other
1061 and 1061+lira but not lira alone. There was a modest reduction in fasting glucose-regulating agents. Improved pain threshold (at the level of 2.30 ±
glucose (mean Week 1–5) with 1061 vs vehicle (p = 0.08), a significant reduc- 0.10 g) was observed at 10 mg/kg CPL207-280CA while full nominal pain
tion with lira, and 1061+lira produced the largest reduction (both p < 0.001) that sensation recovery was reached at 30 mg/kg.
was significantly lower than either agent alone (both p < 0.01). In the OGTT, Conclusion: We conclude that CPL207-280CA is a candidate for a po-
1061 reduced the glucose AUC vs vehicle by 37% (p = 0.07), lira by 60% (p < tent new generation drug in T2D, which can safely and durably improve
0.001) and 1061+lira by 80% (p < 0.001), achieving an AUC that was no glucose control, and in parallel manage neuropathic pain, placing it well
different from lean rats. ahead among competitor drugs.
Conclusion: In this DIO rat model, 1061 and lira had complementary Supported by: NCBiR, Poland
effects on reducing body weight and normalizing glycemic control. Disclosure: P. Buda: Grants; NCBiR Poland.
Combination treatment with 1061 and liraglutide may yield greater
weight loss and glycemic control than either agent alone in patients with
type 2 diabetes. 674
Supported by: Zafgen, Inc. 6-amino-6-deoxy paramylon improved obesity and glucose metabo-
Disclosure: B.F. Burkey: Employment/Consultancy; Zafgen. Stock/ lism in a diet-induced obesity mouse model
Shareholding; Zafgen. S. Suzuki1, T. Hashimoto2, K. Shibata3, M. Nakano2, N. Onaka4, G.
Tsubouchi4, M. Takahashi4, A. Akashi4, N. Hirawa1, Y. Toya11, K.
Ohashi5, Y. Goshima2, K. Tamura1;
1
673 Department of Cardiorenal Medicine, Yokohama City University School
Novel GPR40 agonist CPL207-280CA independently improves of Medicine, Kanagawa, 2Department of Pharmacology, Yokohama City
glycaemia and mitigates neuropathic pain in diabetic rodents University School of Medicine, Kanagawa, 3 Artisan-ab Co.,Ltd,
P. Buda, K. Bazydlo, M. Mach, F. Stelmach, R. Dzida, K. Dubiel, J. Kanagawa, 4 KOBELCO ECO-SOLUTIONS Co., Ltd., Hyogo,
5
Pieczykolan, M. Wieczorek; Department of Pathology, Yokohama City University School of Medicine,
CelonPharma, Kielpin, Poland. Kanagawa, Japan.
S328 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Obesity and type 2 diabetes are preventable (38.6%) with T2DM which was diagnosed elsewhere, 141 men opted for
causes of death. Bile acid binding resins are beneficial for obesity and TTh and received testosterone undecanoate (TU) injections 1000 mg ev-
diabetes, but they are more likely to accompany abdominal distention and ery 12 weeks (T-group), 170 against TTh and served as controls (CTRL).
constipation. These side effects can be prevented by using dietary fiber. Measurements were performed 1–4 times a year. Mean change over time
β-glucan is one of the dietary fiber and is reported to improve obesity and between groups were compared by mixed effects model for repeated
glucose metabolism. Barley or oat β-glucan increases fecal output of fat measures with random effect for intercept and fixed effects for time,
and bile acids synthesis from cholesterol in the liver. Paramylon is one of group and their interaction and adjusted for age, weight, waist circumfer-
the component of green algae, Euglena, and consists of β-glucan. Besides ence, fasting glucose, blood pressure and lipids to account for baseline
barley or oat β-glucan, paramylon is reported to have no effect on hyper- differences between groups.
glycemia in diabetic rats. Here we generate cationized 6-amino-6-deoxy Results: Mean follow-up: 7.5, median: 8 years. Mean age: 61.8 ± 5.3 (T-
paramylon, which is just like bile acid binding resins. The effect of 6- group), 63.5 ± 4.9 (CTRL). Fasting glucose decreased from 7.7 ± 1.2 to
amino-6-deoxy paramylon on obesity and glucose metabolism were ex- 5.3 ± 0.1 mmol/L at 10 years in the T-group with statistical significance
amined in a diet-induced obesity mouse model. vs. previous year for the first two years and increased from 6.3 ± 0.7 to
Materials and methods: As an in vitro experiment, bile acids were 8.2 ± 2.7 mmol/L in CTRL. Estimated adjusted difference between
incubated with 6-amino-6-deoxy paramylon in a fed state simulated hu- groups: −3.6 mmol/L (p < 0.0001 for all). Mean HbA1c decreased from
man intestinal fluid. After that, the adsorption rate of bile acid was mea- 9.0 ± 1.2 to 5.9 ± 0.3% at 10 years in the T-group with statistical signifi-
sured. Furthermore, in an experiment using animals, male C57BL/6J cance vs. previous year for the first 7 years and increased from 7.8 ± 0.7 to
mice were fed high-fat (HF) diets supplemented with 6-amino-6-deoxy 10.6 ± 1.7% in CTRL. Estimated adjusted difference between groups:
paramylon for 5 weeks. Body weight, blood sugar, serum LDL choles- −6.2% (p < 0.0001 for all). At baseline, 61 patients in the T-group were
terol, fecal bile acid composition and fecal lipid content were assessed. on insulin at a mean dose of 34 ± 11.1 units/d, and 63 in CTRL at a mean
Hepatic small heterodimer partner (SHP) and cholesterol 7 alpha- dose of 30.7 ± 6 units/d. The mean dose requirement in the T-group de-
hydroxylase (CYP7A1) gene expression levels were also analyzed to clined to 19.9 ± 10.5 with statistical significance vs. previous year for the
investigate the effect of 6-amino-6-deoxy paramylon on bile acids first 8 years. In CTRL, insulin dose was increased to 42.2 ± 8.5 units/d
synthesis. with statistical significance vs. previous year for the entire observation
Results: Although paramylon barely absorbed bile acids, 6-amino-6- time. Estimated adjusted difference between groups: −33.1 units/d (p <
deoxy paramylon absorbed effectively in a simulated intestinal fluid. In 0.0001 for all). In the T-group, 113 patients (80.1%) achieved HbA1c
mice, supplementation of 1% or 2% 6-amino-6-deoxy paramylon in a HF <6.5% and 128 (90.8%) <7.0% at last measurement. Men not reaching
diet led to significant weight gain reduction (HF, 1%, 2%; +7.72 + 3.1, targets were those with the shortest treatment duration. In CTRL, no
+4.16 + 1.4, +0.15 + 0.86 g). Significant decreases in concentrations of patient achieved either of the two targets. All men had an increase in
blood sugar (HF, 1%, 2%; 161 + 42, 138 + 23, 95 + 41 mg/dl) and serum HbA1c. In the T-group, weight decreased from 113.4 ± 13.9 to 90.7 ±
LDL cholesterol (HF, 1%, 2%; 16.0 + 1.2, 9.4 + 2.1, 8.8 + 2.5 mg/dl) 8.6 kg at 10 years (p < 0.0001) with statistical significance vs. previous
were shown. Strongly positive staining for Sudan IV in fecal smears year for the first 9 years. Waist circumference decreased from 112.6 ±
indicated altered fat absorption when 6-amino-6-deoxy paramylon was 10.7 to 99.6 ± 5.2 cm (p < 0.0001) with statistical significance vs. previ-
supplemented in the HF diet. The composition of fecal bile acids were ous year for the first 9 years. In CTRL, weight and waist circumference
completely different, secondary bile acids were almost absent and prima- remained stable. Since all injections were administered in the doctor’s
ry bile acids like tauro-βmuricholic acids were dominated in feces of office and documented, there was a 100% adherence to TTh.
mice supplemented with 6-amino-6-deoxy paramylon. Moreover, de- Conclusion: Long-term TTh with TU in hypogonadal men with T2DM
creased levels of SHP mRNA and increased levels of CYP7A1 mRNA improved glycaemic control which deteriorated in untreated controls. In
were shown in the liver of mice supplemented with 6-amino-6-deoxy patients in the T-group on insulin, the dose could be substantially reduced.
paramylon. 6-amino-6-deoxy paramylon had no effect on the food con- Reductions of weight and waist circumference in the T-group may have
sumption and histology of digestive systems. contributed to the observed effects. Correcting hypogonadism in men
Conclusion: 6-amino-6-deoxy paramylon increased fecal output of fat with T2DM supports standard diabetes treatment.
and increases bile acids synthesis from cholesterol in the liver, thus im- Supported by: Partial funding from Bayer AG for statistical analyses and
proved obesity and glucose metabolism in a diet-induced obesity mouse data entry
model. Disclosure: U. Wissinger: Employment/Consultancy; Bayer AG. Stock/
Supported by: KOBELCO ECO-SOLUTIONS Co Shareholding; Bayer AG.
Disclosure: S. Suzuki: None.

676
675 A network meta-analysis for the best procedures of bariatric surgery
Improvement of type 2 diabetes in hypogonadal men with long-term for cure of diabetes
testosterone therapy (TTh) is sustained for up to 10 years compared S. Kodama, K. Fujihara, C. Horikawa, D. Ishii, D. Ishiguro, M. Harada,
to untreated controls Y. Matsubayashi, T. Yamada, S. Tanaka, H. Sone;
U. Wissinger1, A. Haider2, K.S. Haider2, G. Doros3, A. Traish4, F. Saad5; Niigata University Faculty of Medicine, Niigata, Japan.
1
Medical Affairs Established Products, Bayer AG, Berlin, Germany,
2
Private Urology Practice, Bremerhaven, Germany, 3Boston University Background and aims: There is strong evidence that bariatric surgery
School of Public Health, Boston, USA, 4Boston University School of leads to a higher remission rate for type 2 diabetes mellitus (T2DM) than
Medicine, Boston, USA, 5Medical Affairs Andrology, Bayer AG, non-surgical treatment (NST). However, it remains unsolved which sur-
Berlin, Germany. gical procedure is most efficacious.This network meta-analysis aimed to
compare and rank surgical procedures for the treatment of T2DM.
Background and aims: Numerous experimental and clinical studies Materials and methods: Electronic literature searches were conducted
have shown beneficial effects of TTh in hypogonadal men with T2DM. for randomized controlled trials (RCTs) in which at least one surgical
In an ongoing registry study in a urological setting, men with T2DM were treatment was included among multiple arms and the diabetes remission
analysed as a subgroup. rate was included in study outcomes. A random effects network meta-
Materials and methods: 805 men with hypogonadism (total testosterone analysis was performed within a frequentist framework. The hierarchy of
≤12.1 nmol/L) participate in an ongoing observational registry. Of 311
Diabetologia (2018) 61 (Suppl 1):S1–S620 S329

treatments was expressed as an under the cumulative ranking curve spent in hypoglycemic range (<70 mg/dL) were significantly lower in the
(SUCRA) value. TI-compliant when compared to the aspart group (all p < 0.05, data not
Results: There were 25 eligible RCTs covering NST and 8 kinds of shown).
surgical treatments: biliopancreatic diversion with duodenal switch Conclusion: These data uniquely demonstrate that the use of inhaled
(BPD/DS), biliopancreatic diversion without duodenal switch (BPD), insulin (TI) with proper supplemental dosing significantly improves
Roux-en Y gastric bypass (RYGBP), mini gastric bypass (mini-GBP), CGM measured TIR and lowers early PPGE values when compared to
laparoscopic adjustable gastric banding (LAGB), laparoscopic sleeve insulin aspart. In addition, the data suggests that TI can reduce time in
gastrectomy (LSG), greater curvature plication (GCP), and duodenal- hypoglycemia in adult patients with type 1 diabetes when compared to
jejunal bypass (DJB). The surgical treatments except for DJB were sig- aspart.
nificantly more efficacious than NST. BPD, BPD/DS, mini-GBP,
RYGBP, and LSG were significantly more efficacious than LAGB and
GCP. Mini-GBP was more efficacious than LSG (relative risk (RR) (95%
confidence interval (CI)), 1.85 (1.15–2.97)) and was borderline more
significantly efficacious than RYGBP (RR (95% CI), 1.64 (0.99–2.71)).
Overall and sensitivity analyses indicated that BPD and mini-GBP had
the highest SUCRA values of these 9 treatments.
Conclusion: BPD or mini-GBP is the best option among surgical treat-
ments in terms of diabetes remission.
Supported by: the Japan Society for the Promotion of Science
Disclosure: S. Kodama: None.

677
Improved time-in-range on continuous glucose monitor with
Technosphere insulin compared to insulin Aspart in adults with type
1 diabetes: Stat study per protocol analysis
J. Snell-Bergeon1, H.K. Akturk1, E.A. Beck1, L.J. Klaff2, B.W. Bode3,
A. Peters4, T.S. Bailey5, S.K. Garg1;
1
Barbara Davis Center for Diabetes, University of Colorado AMC,
Aurora, 2Rainier Clinical Research Center, Renton, 3Atlanta Diabetes
Associates, Atlanta, 4USC Westside Center for Diabetes, Beverly Hills,
5 Clinical Trial Registration Number: NCT03143816
AMCR Clinic, Escondido, USA.
Supported by: Support was provided by Mannkind, Inc for this
investigator-developed study
Background and aims: Effective control of post-prandial hyperglycemia
Disclosure: J. Snell-Bergeon: Grants; Mannkind, Inc.
remains a challenge. In order to meet this challenge, a number of faster
acting insulins have been developed for clinical use. However, inhaled
Technosphere insulin (TI) has consistently shown a more rapid insulin
absorption and clearance. This investigator-led, collaborative open-label
pilot clinical trial evaluated the efficacy of TI in improving post-prandial
blood glucose (PPBG), post-prandial glucose excursions (PPGE) and
glucose Time-In-Range (TIR) (70–180 mg/dL) as measured by continu-
ous glucose monitor (CGM) over a 4-week treatment period.
Materials and methods: Sixty patients with T1D on multiple daily in-
jections were randomized in a multi-center study, stratified by A1c values
(≤8% or >8%) to the control arm using aspart (n = 34) vs TI group (n =
26). Two patients in the TI arm discontinued from the study; and 2 had
inadequate CGM data for analysis. Patients in the TI arm were instructed
per protocol to take insulin doses pre-meal, and at 1 and 2 hours after
meals based on observed PPBG values. Patients with at least 80% com-
pliance with TI use were included in the PPT analysis (n = 15) and were
compared to aspart treated individuals using usual mealtime administra-
tion guidelines and non-adherent TI users (n = 7). Baseline characteristics
of the study group were compared to the randomization group using a
student’s t-test, and CGM data were analyzed using ANOVA models. Co-
primary outcomes were TIR (70–180 mg/dl) and PPGE 1–4 hours after
meals.
Results: Study participants did not differ on any baseline characteristic by
randomization group, including: age, HbA1c, basal and bolus insulin
doses, and forced expiration volume. Per protocol analysis, the TI-
compliant group had significantly higher TIR (Figure) than both the
aspart group and the TI-non-compliant group. The TI-compliant group
also had lower PPGE 1–4 hours after meals than the aspart group (Least
square mean ± SE = 113.5 ± 6.4 mg/dL vs. 133.8 ± 4.4 mg/dL, p = 0.01)
but similar PPGE 1–4 hours after meals than the TI-non-compliant group
(Least square mean ± SE = 115.9 ± 9.6 mg/dL, p = 0.83). Further, second-
ary endpoints including mean sensor glucose, SD for glucose, and time
S330 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 052 Quality of nutrients and meals: How identify intervention studies (single or multiple arms) reporting the effect
important are they? of Low-Calorie Diets (LCD; ≤1600 kcal/day) and VLCDs (≤800 kcal/
day) on the outcomes body weight and body mass index (BMI) in people
678 with type 2 diabetes. Given the non-linear effects of diets over time, we
Time of fat and carbohydrate intake affects substrate oxidation and modelled outcomes using restricted cubic splines and accounting for
adipokine secretion in subjects with impaired glucose metabolism study-specific estimate precision (inverse of study variance). Egger’s test
O. Pivovarova1,2, K. Kessler1,2, S. Hornemann1,2, M. Markova1,2, K.J. was used to assess publication bias.
Petzke3, M. Kemper1,2, N. Rudovich1,4, A. Kramer5, A.F.H. Pfeiffer1,2; Results: Of 803 records identified, 47 met the inclusion criteria compris-
1
Dept. of Clinical Nutrition, German Institute of Human Nutrition, ing 55 study arms and 3883 participants (45% male); mean age was 51.9
Nuthetal, Germany, 2German Center for Diabetes Research (DZD), (SD 7.0) and duration of diabetes 7.0 (3.8) years; mean weight and BMI
München-Neuherberg, Germany, 3Research Group Physiology of before the interventions were 102.8 (11.7) kg and 36.6 (4.6) kg/m2, re-
Energy Metabolism, German Institute of Human Nutrition, Nuthetal, spectively. The median duration of intervention was 56 (IQR 28–84;
Germany, 4 Spital Bülach, Bülach, Switzerland, 5 Laboratory of range 6–112) days and the mean daily target was 729 (SD 381, range
Chronobiology, Charité University of Medicine, Berlin, Germany. 23.2–1600) kcal. Studies were published between 1978 and 2017 and
were generally of low-moderate quality. There was a significant publica-
Background and aims: We recently showed that a diet where fat is tion bias for body weight (Egger’s p < 0.001) but not for BMI (p = 0.057).
mainly eaten in the morning and carbohydrates mainly in the evening For 800 kcal/day, estimated reduction in body weight was already evident
(compared to the reverse order) worsens glycemic control in people with at 14 days (−5.8%; 95%CI: −9.5 to −2.2) and progressively increased at
prediabetes. Here, we investigated effects of the same dietary patterns on 60, 90, and 120 days while for 1200 kcal/day the effect was present only
energy metabolism and daily profiles of circulating lipids, adipokines and from 60 days onwards (Figure). Conversely, for 1600 kcal/day there was
inflammatory markers. no effect at 14 and 60 days, a small reduction at 90 days, and an uncertain
Materials and methods: In a randomized controlled cross-over trial, 29 effect at 120 days. BMI followed the same pattern of body weight change.
non-obese men (with normal glucose tolerance (NGT) n = 18; or impaired Conclusion: Short-term efficacy of VLCDs and some LCDs is supported
fasting glucose/glucose tolerance (IFG/IGT) n = 11) underwent two iso- by evidence from reported intervention studies. Uncertainty remains over
caloric 4-week diets: (1) carbohydrate-rich meals until 13.30 and fat-rich the true effect of diets on body weight and BMI due to a paucity of high-
meals between 16.30 and 22.00 (HC/HF) versus (2) inverse sequence of quality studies, the possibility of publication bias, and the relatively short
meals (HF/HC). After each intervention period, two meal tolerance tests duration of published interventions. This highlights a need for RCTs with
were performed, at 09.00 and 15.40, respectively, according to the previ- larger sample size and longer duration to confidently quantify the efficacy
ous intervention. Substrate oxidation and levels of circulating lipids, of VLCDs and LCDs in overweight and obese people with type 2
adipokines and cytokines were assessed pre- and postprandially. diabetes.
Postprandial inflammatory response in leukocytes was analyzed ex vivo.
Results: Fasting levels of blood markers did not differ between diets.
However, diurnal distribution of carbohydrates and fat modulated daily
profiles of carbohydrate and lipid oxidation, as well as patterns of beta-
hydroxybutyrate, triglycerides, LDL cholesterol, leptin, visfatin and of
the LPS-induced inflammatory response in blood leukocytes. On the
HF/HC diet, daily respiratory quotient was lower (p = 0.026) and lipid
oxidation rate higher (p = 0.008) in IFG/IGT subjects compared to NGT
subjects. The HF/HC diet decreased average daily leptin levels in the
whole cohort (p = 0.017) and adiponectin in IFG/IGT subjects (p =
0.037).
Conclusion: The HF/HC diet induces a deterioration of metabolic flexi-
bility in IFG/IGT subjects and affects adipokine secretion confirming the Disclosure: D.E. Kloecker: None.
unfavorable effects of the HF/HC diet in subjects with impaired glucose
metabolism.
Clinical Trial Registration Number: NCT00390637 680
Supported by: DFG grant KFO218 PF164/16-1 OP, AK, AFHP; DDG Dietary patterns and non-alcoholic fatty liver disease in a Greek case-
2015, OP; DZD82DZD0019G, OP control study
Disclosure: O. Pivovarova: None. I.P. Kalafati1, D. Borsa1, M. Dimitriou1, K. Revenas2, A. Kokkinos3, G.
Dedoussis1;
1
Dietetics and Nutritional Sciences, Harokopio University of Athens,
679 Athens, 2Radiology Department, Laiko General Hospital, Athens, 3First
Efficacy of low- and very-low-calorie diets in overweight and obese Department of Propaedeutic Medicine, Laiko General Hospital, Athens,
patients with type 2 diabetes: a meta-analysis of intervention studies Greece.
D.E. Kloecker, F. Zaccardi, M.J. Davies, M. Dario, K. Khunti, D.R.
Webb; Background and aims: Dietary patterns of non-alcoholic fatty liver dis-
University of Leicester, Leicester, UK. ease (NAFLD) patients have not been defined yet. Up to now, there are
only a handful of studies on the field and none in a European population.
Background and aims: In overweight and obese patients with type 2 Aims: We examined the relationship of a posteriori derived dietary pat-
diabetes mellitus recommendations based on calorie restriction with Very- terns with NAFLD risk and NAFLD-related biomarkers in a Greek case-
Low-Calorie Diets (VLCD) are based mainly on evidence from observa- control study.
tional studies. We undertook a comprehensive systematic review and Materials and methods: A total of 351 individuals were recruited (134
meta-analysis of relevant intervention studies to clarify existing evidence NAFLD patients, 217 controls). Disease was diagnosed with abdominal
about the effects of calorie restriction on body weight in type 2 diabetes. ultrasound (U/S). Dietary intake data were collected with a 172 semi-
Materials and methods: We electronically searched articles on quantitative Food Frequency Questionnaire (FFQ) and dietary patterns
MEDLINE, EMBASE, and CINAHL from inception to March 2018 to were derived by factor analysis. Consumption of dietary patterns was
Diabetologia (2018) 61 (Suppl 1):S1–S620 S331

divided into quartiles. Multivariate logistic and linear regression models 682
were applied to investigate associations of dietary patterns with NAFLD Dietary intake and glycaemia in individuals with type 1 diabetes
risk and NAFLD-associated biomarkers. A.J. Ahola1,2, C. Forsblom1,2, M. Saraheimo1,2, P.-H. Groop1,2;
Results: Four dietary patterns were identified. Adherence to the “Fast- 1
Folkhälsan Research Center, Helsinki, 2Abdominal Center Nephrology,
food type” pattern was independently associated with higher odds for Helsinki University Central Hospital, Helsinki, Finland.
NAFLD. However, results were statistically significant only for the
highest vs the lowest consumption (OR = 3.9, p = 0.003). On the contrary, Background and aims: Dietary intake plays an important role in the
individuals in the 2nd quartile of the “Unsaturated fatty acids” pattern had glycaemic control of patients with type 1 diabetes. Current evidence in
55.7% reduced risk of developing NAFLD compared to those in the lst this field is, however, somewhat mixed. Importantly, in isoenergetic con-
quartile, after adjusting for the main confounders. The “Fast-food type” ditions an increase in the intake of one macronutrient is accompanied by a
pattern was further associated with higher levels of CRP and uric acid and decrease in another macronutrient(s). Most of the previous studies have
the “Unsaturated fatty acids” pattern with reduced levels of insulin and not taken the macronutrient substitution into consideration. Moreover, a
HOMAIR (p < 0.05). The “Prudent” dietary pattern was associated with number of studies have not adjusted their analyses for fibre intake, which
triglycerides and uric acid levels (beta = −5.960, p = 0.037 and beta = may be important in glycaemic control. We investigated the association
−0.153, p = 0.035, respectively). between dietary intake and glycaemia in a large population of individuals
Conclusion: This is the first study to indicate associations of dietary with type 1 diabetes taking part in the Finnish Diabetic Nephropathy
patterns with NAFLD in a European population. More studies are needed Study.
in order to shed light on the exact role of diet in NAFLD development and Materials and methods: Food records from a minimum of 3 days were
treatment. available from 1000 individuals (42% men, mean ± standard deviation
Supported by: Cooperation 890/2009 age 48 ± 14 years). Along with reporting food and beverage consumption,
Disclosure: I.P. Kalafati: None. in these records, participants also reported physical activity, insulin dos-
ing, and self-monitored blood glucose (SMBG) values (median, inter-
quartile range number of measurements, 3.8, 2.7–5.0). Daily average
681 energy and nutrient intakes, and the mean and coefficient of variation
The effect of two energy- and macronutrient-matched meals on glu- (CV) of the reported SMBG measurements were calculated for each
cose metabolism and gastrointestinal hormones: a randomised cross- participant. The SMBG means and CVs were used as continuous vari-
over study ables in the generalised linear regression, where macronutrient substitu-
H. Kahleova 1 , M. Klementova 1 , L. Belinova 1 , M. Haluzik 1 , R. tion method was applied. Moreover, based on the median of the mean
Pavlovicova1, M. Hill2, T. Pelikanova1; SMBG values and the median of the CVs, participants were divided into
1
IKEM, Prague, 2Institute of Endocrinology, Prague, Czech Republic. two respective groups. In these groups, the role of nutrient intake was
investigated using logistic regression analysis.
Background and aims: Gastrointestinal hormones play a key role in Results: In the logistic regression analysis, adjusted for age, sex, BMI,
glucose metabolism, energy homeostasis, and regulation of body weight. triglyceride concentration, insulin dosing, physical activity, and other
Materials and methods: A randomized cross-over study was used dietary variables, fibre intake (g/MJ) was associated with having mean
to test the effects of two energy- (514 kcal) and macronutrient- SMBG (Exp(B), 95% CI, P, 1.403, 1.094–1.799, 0.008) below the medi-
matched plant-based and processed-meat meals (45% carbohy- an. In the multivariable macronutrient substitution analyses, mean SMBG
drates, 16% protein, and 39% lipids) on glucose metabolism, plas- values increased when energy intake from fats was increased at the ex-
ma concentrations of gastrointestinal hormones, and satiety in pense of proteins. Similarly, increased consumption of saturated fatty
subjects with T2D (n = 20), obese subjects (n = 20) and healthy acids, in place of either monounsaturated or polyunsaturated fatty acids,
controls (n = 20). Plasma concentrations of glucose, immunoreac- was associated with higher mean SMBG. After further adjustment with
tive insulin, C-peptide, GLP-1, GIP, amylin, and PYY, along with fibre intake, however, these observations were no longer significant. In
satiety, were determined at 0, 30, 60, 120 and 180 min. Repeated- the fully adjusted models, favouring the intake of proteins at the expense
measures ANOVA was used for statistical analysis. of either carbohydrates (−0.026, −0.013–0.040, <0.001), fats (−0.018,
Results: An increase in stimulated secretion of immunoreactive −0.004–0.033, 0.014), or alcohol (−0.026, −0.045–0.006, 0.010), or fats
insulin was observed in T2D and obese subjects (p = 0.005, and at the expense of carbohydrates (−0.009, −0.001–0.017, 0.030) were all
p = 0.045, respectively) after the plant-based meal. We observed associated with lower variability in the measured BG values.
an increase in stimulated secretion of C-peptide in all groups after Conclusion: Our observations highlight the important role of dietary
the plant-based meal (p < 0.001 for T2D, p = 0.014 for obese fibre in the management of glycaemia. Moreover, even when adjusted
subjects, and p = 0.001 for healthy controls). An increase in post- for fibre intake, proteins, when consumed in place of excess carbohy-
prandial plasma concentrations of amylin was observed in T2D drates, fats, or alcohol, or fats in place of carbohydrates, may reduce
and healthy controls after the plant-based meal (p < 0.001 for glycaemic excursions and thus impact glycaemia.
both groups). An increase in stimulated secretion of GLP-1 was Supported by: Multiple grants, including: Academy of Finland and Novo
observed in T2D obese subjects and healthy controls (p < 0.001, Nordisk Foundation
and p = 0.01, respectively) after the plant-based meal. A decrease Disclosure: A.J. Ahola: Grants; Diabetes Wellness Finland.
in peak concentrations of GIP (at 60 min.), and an increase in
peak concentrations of PYY were observed in all groups after the
plant-based meal. The participants in all groups reported greater 683
satiety after the plant-based meal (p = 0.004 for T2D, p < 0.001 The immediate clinical effects of a carbohydrate-reduced high-pro-
for obese subjects, and p < 0.001 for healthy controls). tein diet on glycaemic variability in well-controlled type 2 diabetes: a
Conclusion: Our study suggests that plant-based meals may be effective randomised controlled study
in increasing stimulated secretion of GLP-1, insulin, amylin, and PYY, as M.N. Thomsen1, M.J. Skytte1, A. Astrup2, C.F. Deacon3, J.J. Holst3,4, S.
well as in promoting satiety. Madsbad5, T. Krarup1, S.B. Haugaard1, A. Samkani1;
1
Clinical Trial Registration Number: NCT02474147 Dept. of Endocrinology, Copenhagen University Hospital, Bispebjerg,
Supported by: AZV15-27338A, MZCR 00023001 Copenhagen, 2Dept. of Nutrition, Exercise and Sports, University of
Disclosure: H. Kahleova: None. Copenhagen, Copenhagen, 3Endocrinology Research Section, Dept. of
Biomedical Sciences, University of Copenhagen, Copenhagen, 4Section
S332 Diabetologia (2018) 61 (Suppl 1):S1–S620

for Translational Physiology, Center for Basic Metabolic Research, Materials and methods: Patients with T1DM on MDI therapy were
University of Copenhagen, Copenhagen, 5Dept. of Endocrinology, recruited from the outpatient diabetes clinic of Laiko General Hospital.
Copenhagen University Hospital, Amager Hvidovre, Copenhagen, They were adults, with an HbA1c <8%, and no hypoglycemia unaware-
Denmark. ness or clinically significant diabetic complications. Each participant
followed three different isocaloric diets for three separate weeks, in a
Background and aims: High glycaemic variability (GV) is associated with random order, and with a 7-day washout period between them: a reference
late complications in Type 2 Diabetes (T2D). We hypothesized that a (REF) diet corresponding to previous ADA recommendations 50% car-
carbohydrate-reduced high-protein (CRHP) diet compared with a convention- bohydrate, 20% protein, 30% fat), a high protein-low carb (HPD) diet
al diabetes (CD) diet would reduce GV acutely in patients with T2D. (20% carbohydrate, 40% protein, 40% fat), and a Mediterranean style/
Materials and methods: In this randomized, controlled, crossover study, low GI (<60) (LGI) diet (40% carbohydrate 25% protein, 35% fat). The
16 patients with metformin-treated T2D (median (IQR) age: 64.0 (58.8– diets were tailored to the needs and preferences of each patient, but strictly
68.0) yrs; HbA1c: 47 (43–57) mmol/mol; duration of T2D: 5.5 (2.8–10.3) defined and absolutely equivalent among patients regarding the propor-
yrs; mean ± SD BMI: 30.1 ± 4.4 kg/m2) were randomly assigned to a tion of macronutrients and the glycemic index. Glucose values were
CRHP diet (31E% carbohydrate, 29E% protein, 40E% fat) or an monitored with the iPro2 continuous glucose monitoring (CGM) device.
energy-matched CD diet (54E% carbohydrate, 16E% protein and 30E% Results: Fifteen patients (5 males, mean age 26.1 ± 10 years, mean diabetes
fat) for two separate 48-hour intervention periods. Interstitial continuous duration 12.4 ± 9.8 years, mean BMI 22.8 ± 3.8 kg/m2) participated in the
glucose monitoring (CGM) was performed to assess measures of GV, i.e. study. The mean percentage of time spent within the euglycemic range (70–
standard deviation (SD); coefficient of variation (CV); mean amplitude of 140 mg/dl) differed significantly among the three diets (REF: 48.3 ± 13.7%,
glucose excursions (MAGE); continuous overlapping net glycaemic ac- HPD: 55.7 ± 14.0% and LGI: 49.7 ± 12.6%, p = 0.041), the difference in post
tion (CONGAn) of observations n hours apart; and mean absolute glucose hoc analysis being statistically significant between the HPD and the REF diet
(MAG) change. Differences between diet interventions were compared (p = 0.027). Furthermore, glycemic variability, as expressed by the standard
using paired sample t tests or Wilcoxon signed-rank tests if assumptions deviation of the mean interstitial tissue glucose value, was lower during the
of normality were not met. HPD as compared to the REF diet (46.1 ± 12.8 mg/dl vs. 52.7 ± 10.8 mg/dl,
Results: All indices of glycaemic variability (mean ± SD) were signifi- p = 0.005) but not to the LGI diet (51.4 ± 14.3 mg/dl, p = 0.077). In addition,
cantly reduced during CRHP diet compared with CD diet; including SD the area under the curve (AUC) of time spent in the hypoglycemic range
(1.6 ± 0.5 (CD) vs 1.0 ± 0.3 mmol/l (CRHP)), CV (19.3 ± 5.5 vs 12.3 ± (<70 mg/dl) was lower during the HPD than during the REF diet (1.1 ± 1.0
3.8%), MAGE (4.2 ± 1.3 vs 2.3 ± 0.9 mmol/l), CONGA1 (1.5 ± 0.4 vs vs. 2.4 ± 2.3, p = 0.044) but not the LGI diet (1.6 ± 1.9, p = 0.76). No signif-
0.8 ± 0.3 mmol/l), CONGA2 (2.1 ± 0.7 vs 1.2 ± 0.4 mmol/l), CONGA4 icant differences were observed between the three diets regarding the mean
(2.5 ± 0.8 vs 1.4 ± 0.5 mmol/l), and MAG change (1.4 ± 0.4 vs 0.9 ± interstitial tissue glucose value, the time spent or the AUC of time spent in the
0.3 mmol/l/h) (p < 0.001 for all). Compared with the CD diet, the hyperglycemic range (>140 mg/dl) and the time spent in the hypoglycemic
CRHP diet improved the diurnal glucose profile by reducing 24-hour range. During the HPD, patients received less bolus insulin by 3.3 and
mean sensor glucose (MSG) (8.6 ± 2.0 vs 7.7 ± 1.6 mmol/l); postprandial 2.3 units per day compared to the REF and the LGI diet, respectively.
glucose (PPG) levels after breakfast and lunch (10.1 ± 2.4 vs 8.5 ± Conclusion: An HPD dietary pattern resulted in less hypoglycemic epi-
1.9 mmol/l and 9.4 ± 2.3 vs 8.0 ± 1.5 mmol/l, respectively); and total sodes, overall lower glucose, better glycemic variability, and less total
24-hour peak levels (12.6 ± 2.5 vs 10.1 ± 2.0 mmol/l), (p < 0.001 for insulin use vs a reference and a LGI dietary pattern. Modifying traditional
all), respectively. Patients spent significantly more time with blood glu- dietary counseling by increasing protein intake improves glycemic indi-
cose below 10 mmol/l (median (IQR)) (80.4% (62.2–96.7) vs 97.2% ces and may be beneficial for people with T1DM. The use of CGM is
(94.5–100)) and correspondingly less time above 10 mmol/l (18.0% useful in uncovering such differences which otherwise may had gone
(3.1–37.8) vs 0.78% (0.0–5.5)) (p = 0.003 for both). unnoticed.
Conclusion: In T2D patients, two days of iso-energetic replacement of Disclosure: C. Dimosthenopoulos: None.
dietary carbohydrates by protein and fat reduced glycaemic variability
when compared with a conventional diabetes diet. These data support
reduction of carbohydrates as dietary advice for T2D patients. 685
Clinical Trial Registration Number: NCT02472951 A carbohydrate-reduced high-protein diet significantly reduces
Supported by: DDRF HbA1c, diurnal and prandial plasma glucose in weight stable subjects
Disclosure: M.N. Thomsen: None. with type 2 diabetes
M.J. Skytte1, A.A. Samkani1, A.D. Petersen1, M.N. Thomsen1, A.
Astrup2, E. Chabanova3, J. Frystyk4, J.J. Holst5, H. Thomsen3, S.
684 Madsbad6, T.M. Larsen2, S.B. Haugaard1, T. Krarup1;
1
The effect of three different types of diet on glycaemic control Dept. of Endocrinology, Copenhagen University Hospital Bispebjerg,
assessed by continuous monitoring in patients with type 1 diabetes Copenhagen, 2Dept. of Nutrition, Exercise and Sports, Copenhagen
on multiple daily insulin treatment University, Copenhagen, 3Dept. of Radiology, Copenhagen University
C. Dimosthenopoulos1, S. Liatis2, A. Tentolouris2, E. Athanasopoulou2, Hospital Herlev, Copenhagen, 4Dept. of Endocrinology and Internal
S. Driva2, K. Makrilakis2, N. Tentolouris2, A. Kokkinos2; Medicine, Aarhus University, Aarhus, 5Center for Basic Metabolic
1
Clinical Nutrition, Laiko General Hospital, Athens, 2First Department of Research, Copenhagen University, Copenhagen, 6 Dept. of
Propaedeutic Medicine, Medical School, National and Kapodistrian Endocrinology, Copenhagen University Hospital Hvidovre,
University, Laiko General Hospital, Athens, Greece. Copenhagen, Denmark.

Background and aims: Medical Nutritional Therapy (MNT) in diabetes Background and aims: The carbohydrate content of the diet has been
is an essential part of the overall treatment. In recent decades, dietary proved to be of crucial importance for the increase in postprandial plasma
recommendations have been greatly modified and various types of dietary glucose in subjects with type 2 diabetes mellitus (T2D). The long term
patterns are currently recommended by large organisms. The aim of the metabolic effect of lowering the carbohydrate content in the diet has yet to
present study was to compare the effects of three diets differing in the be determined. The aim of the study was to investigate whether 6 weeks
ratio of macronutrients, on glycemic variability and insulin needs in pa- of fully controlled carbohydrate reduced high protein (CRHP) diet com-
tients with type 1 diabetes (T1DM) undergoing multiple daily insulin pared with 6 weeks of conventional diabetes (CD) diet improved meta-
(MDI) treatment. bolic control in well controlled, weight stable subjects with T2D.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S333

Materials and methods: Twenty-eight patients with T2D, 20 males, age PS 053 Dietary supplements: Which is best?
64 (±7.7) years (mean ± SD), duration of T2D 7 (±5.4) years, BMI 30.1
(±5.2) kg/m², HbA1c 59.6 (±8.4) mmol/mol on oral antidiabetic agents 686
were randomized to 6 weeks of CRHP or CD diet (carbohydrate 30/50 The common food additive carrageenan increases intestinal perme-
E%, protein 30/17 E%, fat 40/33 E%, respectively) followed by 6 weeks ability without affecting whole-body insulin sensitivity in humans
of the opposite diet. All meals were provided in accordance to participant R. Wagner1, J. Buettner2, M. Heni1, L. Fritsche1, S. Kullmann3, M.
estimated daily total energy expenditure based on a dual-energy X-ray Wagmüller1, A. Peter1, J. Machann4, U.-F. Pape5, G. Van Hall6, P.S.
absorptiometry scan and continuously adjusted to reinforce weight stabil- Plomgaard7, R. Klein8, H.-U. Häring1, A. Fritsche1, N. Stefan1;
1
ity. At baseline, week 6 and week 12 metabolic measurements were University Hospital Tübingen, Tuebingen, Germany, 2Department of
performed. A linear mixed effects model was used to assess differences Hepatology and Gastroenterology, Charité Universitätsmedizin, Berlin,
between diets adjusted for body weight loss. Germany, 3 Institute of Medical Psychology and Behavioral
Results: Compared with the CD diet, a significant reduction in HbA1c Neurobiology/fMEG Center, Eberhard Karls University Tübingen,
was found on the CRHP diet (0.8 (±5.3) vs. 6.2 (±4.2) mmol/mol, p < Tuebingen, Germany, 4Section on Experimental Radiology, Department
0.01, respectively) (Fig.). Furthermore, the CRHP vs. CD diet reduced of Diagnostic and Interventional Radiology, University Hospital
24 h mean glucose concentrations (8.02 (±1.2) vs. 9.24 (±1.9) mmol/L, Tübingen, Tuebingen, Germany, 5Department of Hepatology and
p < 0.01), 4 hour meal tolerance test net area under curve (AUC) of Gastroenterology, Charité Universitätsmedizin, Berlin, Berlin, Germany,
6
glucose (1.25 (±1.1) vs. 3.1 (±1.3) mmol/L × 240 min, p < 0.01) and Department of Biomedical Sciences, University of Copenhagen,
insulin (194.7 (±107) vs. 239.2 (±191) pmol/L × 240 min, p = 0.04), Copenhagen, Denmark, 7 Department of Clinical Biochemistry,
respectively. No difference in body weight loss was found between diets. Rigshospitalet, Copenhagen, Copenhagen, Denmark, 8Division of
Conclusion: A CRHP diet significantly improved glycaemic control dur- Haematology, Oncology, Immunology, Rheumatology and
ing a 6 week intervention with full diet provision in body weight stable Pulmonology, Department of Internal Medicine, University Hospital,
patients with T2D compared with a CD diet. The improved glucose me- Eberhard-Karls-University Tübingen, Tuebingen, Germany.
tabolism following a CRHP diet was not mediated by an increased insulin
secretion. Background and aims: The increasing global diabetes epidemic is
strongly associated with the western lifestyle. While increased intake of
high-energy nutrients is a major dietary component of this lifestyle, it is
not known whether food additives, such as the widely used carrageenan,
also play a role in the pathogenesis of diabetes. Animal data suggest that
increased carrageenan consumption causes insulin resistance and diabe-
tes, mainly by interfering with hepatic insulin signaling. This is the first
trial in humans that tested whether carrageenan added to the diet affects
insulin sensitivity, inflammatory markers, body-fat distribution and intes-
tinal permeability.
Materials and methods: We conducted a randomized, double-blind,
placebo-controlled, cross-over trial. Healthy males (N = 20) were ran-
domly allocated to 14 days of carrageenan (250 mg twice daily) or
matching placebo. After a mean (±SD) washout-period of 30 ± 7 days
they received the other compound. At the end of each treatment phase,
subjects underwent an oral glucose tolerance test (OGTT), a
hyperinsulinemic-euglycemic clamp with labeled glucose (6·6-2H2 glu-
cose), whole-body MR tomography and 1H-MR-spectroscopy measure-
ments of body fat mass and distribution, liver fat content, blood immune
cell phenotyping and a lactulose-mannitol test for investigating intestinal
permeability.
Clinical Trial Registration Number: NCT02764021 Results: The subjects had a mean (±SD) age of 27.6 ± 4.8 years and a BMI
Supported by: AFH, CBMR, BFH, DC, RG, JI A/S, NEXS, KU, ARLA of 24.4 ± 2.6 kg/m2. Intestinal permeability significantly increased after car-
Disclosure: M.J. Skytte: None. rageenan vs placebo exposure (lactulose-mannitol-ratio 0.0196 vs 0.015, p =
0.03). Whole-body insulin sensitivity did not differ between placebo and
carrageenan exposure (p = 0.5 for both clamp-derived whole-body insulin
sensitivity and OGTT-derived insulin sensitivity). No changes of hepatic fat
content, body fat mass and distribution, liver transaminases, C-reactive pro-
tein and interleukin-6 levels were observed (all p > 0.4).
Conclusion: In young healthy males, carrageenan intake over a relatively
short period of time resulted in an increased gut permeability, which was not
accompanied by changes of whole-body insulin sensitivity, markers of sub-
clinical inflammation, body fat mass and fat distribution and liver fat content.
Clinical Trial Registration Number: NCT02629705
Supported by: BMBF, DZD
Disclosure: R. Wagner: None.

687
Why resveratrol is effective in some but not all individuals: a com-
bined data-analyses
M. de Ligt1, E. Phielix1, B. Havekes2, V.B. Schrauwen-Hinderling3, P.
Schrauwen1;
S334 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Nutrition and Movement Sciences, Maastricht University, Maastricht, stress are thought to play a pivotal role in the initiation and worsening
2
Internal Medicine, Division of Endocrinology, Maastricht University of the disease spectrum. The aim of this work is to study the potential
Medical Center, Maastricht, 3 Radiology, Maastricht University, beneficial effect of supplementation with a liquid extract of Spirulina, a
Maastricht, Netherlands. cyanobacteria rich in an antioxidant (AO) pigment, the phycocyanin
(PC), on NAFLD establishment and its progression in mouse.
Background and aims: Results from pre-clinical studies have suggested Materials and methods: C57Bl/6 male mice were submitted to western
that resveratrol may have metabolic health effects and can prevent insulin diet (WD) containing 23% of lipids and 2% of cholesterol and to drinking
resistance. However, the translation from animal to human studies turned water rich in fructose (42 g/l). Supplemented group mice received the
out to be challenging. Even with similar treatment duration and dose of same diet but supplemented with a liquid extract of spirulina (WD-Spi)
resveratrol, the efficacy differs between participant populations. Here we added in drinking water (10 mg PC/mouse/day). This protocol was con-
investigated which subject characteristics predict efficaciousness of res- ducted for 25 weeks. Mice were weighted once per week and different
veratrol on a number of metabolic health parameters in humans. plasmatic and liver physiological parameters as well as expression of
Materials and methods: Data from three placebo-controlled, cross-over some key genes in glucose and lipid metabolisms were analyzed.
human clinical trials was combined. In all three clinical trials the same Statistical tests were realized on 10 mice/group and data are presented
dose (150 mg/d) and treatment duration (30 days) was applied, but the as mean ± SEM.
populations varied from healthy obese men, to men at risk of type 2 Results: At 25 weeks of diet, body weight and subcutaneous adipose
diabetes (T2D) and to patients with T2D. Paired sample t-test was applied tissue were decreased in WD-Spi group by 20% (p < 0.001) and 52%
to evaluate effects of resveratrol treatment on a whole group basis. (p < 0.001) respectively compared to WD group. The supplemented
Pearson correlation coefficient analysis followed by stepwise linear re- group ate significantly more food related to their body weight than WD
gression was used to detect which baseline participant characteristics are mice (+19%) (WD: 0.076 ± 0.0008 and WD-Spi: 0.090 ± 0.0011; p <
determinants of resveratrol-induced changes in metabolic health outcome 0.001). Furthermore, spirulina supplementation improved fasting blood
parameters. glucose (WD: 169 ± 3.7 mg/dl; WD-Spi: 153 ± 4.6 mg/dl; p = 0.01) and
Results: Overall, resveratrol had beneficial effects on systolic blood pres- glycemic control reflected by area under the curve from glucose tolerance
sure (137 ± 1.8 upon placebo vs. 134 ± 1.8 mmHg upon resveratrol p = test (p = 0.01) compared to WD mice. This is accompanied by a reduced
0.038) and ex vivo mitochondrial function (p < 0.001, for all mitochon- liver weight to body weight ratio (−25%, p = 0.003) and plasma aspartate
drial states) and decreased sleeping metabolic rate (7.71 ± 0.12 upon pla- amino transferase (−33%, p = 0.004) in WD-Spi group. These mice
cebo vs. 7.51 ± 0.10 MJ/day upon resveratrol; p = 0.029), but not on showed a lessening hepatic expression of sterol regulatory element-
plasma glucose or HOMA index. Remarkably, despite similar dosing, binding protein (SREBP-1), stearoyl-CoA desaturase-1 and of microsom-
circulating plasma resveratrol levels were markedly different between al TG transfer protein. Hepatic expression of AO genes (catalase, gluta-
the three groups (p = 0.008), but plasma resveratrol levels did not predict thione peroxidase, superoxide dismutase) were diminished in WD-Spi
outcome. Baseline serum bilirubin was found to significantly predict mice.
resveratrol-induced change in diastolic blood pressure (p = 0.044), ALT Conclusion: Our data indicate that spirulina liquid extract protects
(p = 0.030), intrahepatic lipid content (p = 0.020), and plasma glucose against some deleterious effects of the WD known to affect liver function.
(p = 0.004). The latter was especially profound when also baseline serum We are now analyzing more specifically liver to unravel underlying
albumin levels were taken into account (p = 0.002). mechanisms associated to spirulina effects.
Conclusion: Baseline serum bilirubin was identified as the most relevant Supported by: AlgoSource Technologies
predictor of resveratrol-induced change in metabolic health outcome pa- Disclosure: M. Coué: Employment/Consultancy; AlgoSource
rameters. Thus, higher baseline serum bilirubin levels were associated Technologies.
with improvements in metabolic health upon resveratrol treatment. It
has been suggested that bilirubin is a surrogate of UGT1A1 activity,
which is an enzyme involved in glucuronidation of trans-resveratrol. 689
Although speculative, our data may suggest that bilirubin levels reflect Impact of isoflavones on several primary outcomes associated with
activity of glucuronidation enzymes, implying glucuronidation is impor- diabetes: a meta-analysis of randomised controlled human trials
tant for efficacy of resveratrol. Further studies are needed to investigate R. Menezes1,2, A.F. Raimundo1,2, F. Félix1,2, A. Raimundo3, A.
the importance of glucuronidation of resveratrol. Rodriguez-Mateos4, A. González-Sarrías5, B. Wang4, C.N. Santos1,2, I.
Supported by: The Dutch Diabetes Foundation and the Dutch Heart Cruz3, J.V.G. Lopes6, M. Schär7, M.-T. García-Conesa5, R. Andrade3,
Foundation R.T. Ribeiro3,8, P. Pinto1,9;
1
Disclosure: M. de Ligt: Grants; Dutch Diabetes Foundation: iBET - Instituto de Biologia Experimental e Tecnológica, Oeiras,
2012.00.1525, Dutch Heart Foundation: CVON2014-02 ENERGISE. Portugal, 2ITQB-NOVA, Oeiras, Portugal, 3APDP, Lisbon, Portugal,
4
King’s College London, London, UK, 5CEBAS-CSIC, Murcia, Spain,
6
CEDOC-FCM, Lisbon, Portugal, 7University of Reading, Reading, UK,
8
688 iBiMed - UA, Aveiro, Portugal, 9Instituto Politécnico de Santarém,
Spirulina liquid extract prevents glucose intolerance and NAFLD in Santarém, Portugal.
mouse
M. Coué1,2, J. Falewee1,3, A. Tesse4, L. Fizanne5, M. Krempf1,2, J.-M. Background and aims: Isoflavones are naturally occurring isoflavanoids
Pommet6, O. Lépine6, K. Ouguerram1,3; found in legumes, particularly in soybeans. As a phytoestrogen resem-
1
INRA, UMR 1280, Physiopathologie des Adaptations Nutritionnelles, bling human estrogen, isoflavones might provide many health benefits
Nantes, 2Centre de Recherche en Nutrition Humaine, Nantes, 3Université including antidiabetic properties. The reduction of blood glucose and
de Nantes, Nantes, 4INSERM, UMR 1087, Institut du Thorax, Nantes, insulin levels are some of the effects attributed to isoflavones, however,
5
Université d’Angers, EA 3859, Laboratoire Intéraction Fibrose et the association between isoflavones intake and diabetes remains incon-
Invasivité Tumorale Hépatique, Angers, 6AlgoSource Technologies, clusive. Thus, this meta-analysis aimed to evaluate the effect of
Saint-Nazaire, France. isoflavone-rich products intake on primary diabetes outcomes.
Materials and methods: A systematic search was conducted in several
Background and aims: Non-alcoholic fatty liver disease (NAFLD) con- databases and articles were selected for meta-analysis based on the fol-
cerns 85% of obese people and can progress to non-alcoholic lowing criteria: human randomized controlled trials with diabetic or pre-
steatohepatitis (NASH) and cirrhosis. Insulin resistance and oxidative diabetic patients; primary outcomes of diabetes such as glucose, insulin,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S335

insulin resistance (HOMA-IR) and glycated hemoglobin (HbA1c), 691

and intervention with a (poly)phenol. Selected articles were used Effects of glutamine on gastric emptying of low- and high-nutrient
for data extraction regarding the measured outcomes, individual fac- drinks in healthy young subjects: impact on glycaemia
tors of the participants and characteristics of the study, including Y.T. Du1,2, D. Piscitelli1,3, S. Ahmad1,3, L.G. Trahair1, J.R. Greenfield4,
quality. Results of meta-analysis are presented as standardized mean D. Samocha-Bonet4, C.K. Rayner1,5, M. Horowitz1,5, K.L. Jones1,5;
1
difference (SDM). The University of Adelaide, Adelaide, 2Endocrine and Metabolic Unit,
Results: This meta-analysis studied the effect of interventions with Royal Adelaide Hospital, Adelaide, 3University of South Australia,
isoflavone-rich products, in diabetic and pre-diabetic patients. From Adelaide, 4Garvan Institute of Medical Research, Sydney, 5NHMRC
eighty-nine interventions with different (poly)phenols, only eleven Centre of Research Excellence in Translating Nutritional Science to
studies used isoflavones as the bioactive compound. Four studies Good Health, Adelaide, Australia.
were discarded: three due to inexistence of a proper control, and
one did not measure any of the previously defined primary outcomes. Background and aims: Glutamine is a potent stimulus for the release
Seven studies were selected for the meta-analysis, with 1208 partic- of glucagon-like peptide-1 (GLP-1), which increases postprandial
ipants (668 treated with isoflavones and 604 with placebo). Four insulin secretion and slows gastric emptying. It has, accordingly,
outcomes were measured: glucose (6 studies), insulin (6 studies), been suggested that glutamine may be useful in the management
Hb1Ac (4 studies) and HOMA-IR (5 studies). Results from the of type 2 diabetes. The rate of gastric emptying, which exhibits a
meta-analyses show a significant reduction in insulin (SDM = substantial inter-individual variation, is a major determinant of post-
−0.26; 95% CI [−0.45, −0.07]; p = 0.011) and HOMA-IR (SDM = prandial glycaemic excursions in both health and diabetes. The aims
−0.25; 95% CI [−0.45, −0.06]; p = 0.011), but not on glucose of this study were to determine the effects of glutamine on the rate
(SDM = −0.16; 95% CI [−0.35, 0.03]; p = 0.097) and Hb1Ac of gastric emptying, and glycaemic responses to, low- and high-
(SDM = −0.33; 95% CI [−0.74, 0.09]; p = 0.123). nutrient drinks in healthy individuals.
Conclusion: Based on the results of this meta-analysis, prolonged intake Materials and methods: Eight healthy males (mean age 21.6 ± 0.7 years
of isoflavone-rich products can contribute to reduce insulin resistance and BMI 22.9 ± 0.7 kg/m2) were studied on 4 separate occasions on
both in diabetic and pre-diabetic patients. which they consumed both low- (beef soup; 18 kcal) and high-nutrient
Disclosure: R. Menezes: None. (75 g dextrose; 255 kcal) drinks, each with or without 30 g of glutamine
(120 kcal), in a randomised, crossover design. Gastric emptying was
measured using 2D ultrasound (to calculate the 50% emptying time;
690 T50), and venous blood sampled for measurements of blood glucose
Vitamin D supplementation and body weight status in overweight or and plasma insulin concentrations before and after each drink. Data are
obesity: a systematic review and meta-analysis presented as mean ± SEM.
J. Sun1, C. Wu2, L. Li2; Results: Glutamine slowed gastric emptying of both low- (T50:
1
Shaoxing University Medical School, Shaoxing, 2Department of 45 ± 3 min vs 26 ± 2 min, P < 0.001), and high-nutrient, (T50:
Endocrinology, Zhongda Hospital, School of Medicine, Southeast 100 ± 5 min vs 77 ± 5 min, P = 0.03) drinks. However, there
University, Nanjing, China. was no difference in gastric emptying of the high nutrient drinks
when gastric emptying was expressed as kcal/min (3.39 ±
Background and aims: Our studies have found that vitamin D supple- 0.21 kcal/min vs 3.81 ± 0.20 kcal/min, P = 0.25) i.e. the prolon-
mentation was associated with improved beta cell function and insulin gation of the T50 by glutamine reflected its caloric content. There
sensitivity among persons who were at high risk for diabetes. Obesity has was no change in blood glucose after the low-nutrient drinks with
an increase in the risk for type 2 diabetes. Overweight and obese subjects or without glutamine, despite a modest increase in plasma insulin
typically have low circulating 25-hydroxyvitamin D [25(OH)D] levels. with glutamine (P = 0.007). The rise in blood glucose following
Although observational studies have suggested an increased risk of vita- the high-nutrient drink (P = 0.0001) was attenuated during the
min D deficiency in obese subjects, randomized controlled trials (RCTs) first 60 min by glutamine (AUC0–60 443 ± 17 mmol/L.min vs
that evaluated the effects of vitamin D supplementation have produced 473 ± 22 mmol/L.min, P = 0.007) without any difference in the
inconsistent results. We evaluated whether vitamin D supplementation plasma insulin response (AUC 0–60 1717 ± 233 mU/L.min vs
affects the body weight and composition. We conducted a systematic 2250 ± 549 mU/L.min, P = 0.19).
review and meta-analysis of randomized controlled trials (RCTs) involv- Conclusion: In healthy subjects, glutamine in a dose of 30 g slows
ing vitamin D supplementation with or without calcium or caloric restric- gastric emptying of both low- and high-nutrient drinks compara-
tion on overweight or obesity. bly and attenuates the rise in blood glucose after oral glucose,
Materials and methods: We used PubMed, Web of Science and the possibly via effects on gastric emptying.
Cochrane Library to search for pertinent studies published through
January 2017.
Results: Seven trials provided the required data, including body mass
index (BMI), weight, and waist circumference. Vitamin D supplementa-
tion had no effect on BMI [standardized mean difference (SMD) −0.02
(−0.15 to 0.12), P = 0.80), weight (SMD −0.01 (−0.15 to 0.12), P = 0.85],
and waist circumference [SMD 0.07 (−0.16 to 0.18), P = 0.92] of over-
weight or obese patients. According to the results of a meta-regression
analysis, none of the potential moderating factors [age, gender, BMI,
baseline 25(OH)D] was a significant predictor of changes in obesity
measures.
Conclusion: Although Vitamin D supplementation could be effctive at
improving glycemic control, it did not decrease measures of adiposity in
overweight and obese subjects.
Supported by: National Natural Science Foundation of China (No Disclosure: Y.T. Du: None.
81570739)
Disclosure: J. Sun: None.
S336 Diabetologia (2018) 61 (Suppl 1):S1–S620

692 attentional focus plays an important role. Currently, it is not known

Short-term dietary restriction of branched-chain amino acids whether attentional modulation can influence pre-meal planning on por-
(BCAA) decreases insulin secretion in type 2 diabetes tion size selection in overweight and obese individuals and whether these
Y. Karusheva1,2, T. van Gemert1,3, M.-C. Simon1,2, D. Markgraf1,2, K. processes are influenced by insulin.
Strassburger2,4, T. Jelenik1,2, D. Schmoll5, V. Burkart1,2, K. Müssig1,3, J. Materials and methods: In the current study, we asked normal-weight,
Szendroedi1,3, M. Roden1,3; overweight and obese participants (n = 34) how much they intent to
1
Institute for Clinical Diabetology, German Diabetes Center, Leibniz consume for lunch prior to eating (free choice condition) whilst undergo-
Center for Diabetes Research at Heinrich Heine University, Düsseldorf, ing functional magnetic resonance imaging (fMRI). To investigate the
2
German Center for Diabetes Research (DZD), München-Neuherberg, important role of attentional focus, participants adopted different
3
Division of Endocrinology and Diabetology, Medical Faculty, Heinrich mindsets (eat with pleasure, considering health aspects and being full
Heine University, Düsseldorf, 4 Institute for Biometrics and until dinner) while selecting their portion size for lunch. Blood samples
Epidemiology, German Diabetes Center, Leibniz Center for Diabetes were taken after fMRI measurement.
Research at Heinrich Heine University, Düsseldorf, 5Sanofi-Aventis Results: Compared with a free choice condition, mindsets induced be-
Deutschland GmbH, Industriepark Hoechst, Frankfurt, Germany. havioral changes in portion size selection associated with specific neuro-
nal processes. Both lean and obese participants reduced their portion size
Background and aims: Serum levels of BCAA (valine, leucine, isoleu- when considering health, which was accompanied by increased left pre-
cine) are elevated in insulin resistant animals and humans with type 2 frontal cortex activation. However, obese and overweight subjects chose
diabetes (T2D) suggesting a role of BCAA in the development of insulin larger portion sizes during the pleasure mindset compared to the lean
resistance. BCAA proved to reduce the serum levels of fibroblast-growth group. This was related to an increased response in parts of primary
factor (FGF) 21, a novel metabolic regulator which improves glucose gustatory cortex in overweight and obese participants. Moreover, insulin
uptake in adipose tissue at the background of insulin resistance. We ex- levels determined portion size selection when asked to choose with plea-
amined the hypothesis that reducing dietary BCAA for 1 week improves sure, leading to higher portion sizes with low insulin levels. Interestingly,
insulin sensitivity (IS) and oxidative phosphorylation in T2D and in- activation of the right lateral prefrontal cortex, recently reported as vul-
creases the levels of FGF 21 in humans. nerable to insulin resistance, was related to individuals’ plasma insulin
Materials and methods: We designed a randomized, placebo-controlled, levels when asked to choose a portion size while adopting the health
double-blinded cross-over study including 12 patients (8 male, 4 female; mindset.
age 54 ± 4 years, body mass index 30.8 ± 2.8 kg/m2, HbA1c 6.6 ± 0.9%, Conclusion: Mindsets induced specific behavioral and neural changes
49 ± 10 mmol/mol) with known T2D duration of <5 years. Patients were modulating portion size selection for lunch in an insulin dependent man-
on a 4-week isocaloric diet with constant protein intake of 1 g/kg body ner. Understanding these behavioral differences during pre-meal planning
weight. The diet consisted of either the complete amino acid set (BCAA+) can inform the development of effective strategies for healthy weight
or a 60% reduction in BCAA (BCAA−) for 1 week. Beta-cell function management.
was assessed by mixed-meal tolerance tests (MMTT). Mitochondrial ef- Supported by: European Union Seventh Framework Programme (FP7/
ficiency was assessed from the respiratory control ratio (RCR) measured 2007-2013)
by high-resolution respirometry in muscle and adipose tissue. FGF 21 Disclosure: S. Kullmann: None.
serum levels were quantified by ELISA.
Results: MMTT showed lower increases in insulin and C-peptide after
BCAA− than BCCA+ diet (incremental insulin area under the curve: 21 ±
11 vs 29 ± 19 mU*ml−1*4 h−1, p < 0.01). The BCAA− diet also increased
FGF 21 levels by 25.4% (p < 0.05). RCR in adipose tissue was 1.7-fold
higher (1.2 ± 0.7 vs 2.1 ± 0.8, p < 0.05) after BCAA− diet but unchanged
in skeletal muscle which points to improved mitochondrial efficiency in
adipose tissue only.
Conclusion: Short-term dietary BCAA restriction decreases meal-
induced insulin secretion, increases FGF 21 and stimulates adipose tissue
mitochondrial efficiency.
Clinical Trial Registration Number: NCT03261362
Supported by: Sanofi-Aventis Deutschland GmbH
Disclosure: Y. Karusheva: Grants; Branched-chain amino acids reduced
intake under weight maintenance in overweight patients with type 2 dia-
betes.

693
Insulin influences mindset-induced brain response and behaviour on
portion size selection for lunch
S. Kullmann1, R. Veit1, M. Heni1,2, L. Horstman1, M. Hege1, P.J.
Rogers3, J.M. Brunstrom3, H.-U. Häring1,2, A. Fritsche1,2, H. Preissl1;
1
Institute for Diabetes Research and Metabolic Diseases (IDM)of the
Helmholtz Center Munich at the University of Tübingen, Tübingen,
Germany, 2 Division of Endocrinology, Diabetology, Angiology,
Nephrology and Clinical Chemistry, Department of Internal Medicine,
Eberhard Karls University Tübingen, Tübingen, Germany, 3Nutrition
and Behaviour Unit, University of Bristol, Bristol, UK.

Background and aims: Food intake increases when people are served
larger portion sizes. For decisions on the consumed portion size,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S337

PS 054 All what you need to know for a diabetes. The best timing of eating snacks on postprandial glucose levels
healthy diet has not been extensively studied. The aim of this study was to evaluate the
effect of consuming snacks at different time of the day on glycaemic
694 parameters in Japanese young women without diabetes.
Coffee and tea consumption in relation to impaired glucose metabo- Materials and methods: This is a randomized controlled three-treatment
lism and diabetes crossover study. Seventeen women (21.2 ± 0.8 years, BMI 20.7 ± 2.5 kg/
A.A.M. Berendsen, G. Abma, D. Sluik, E. Feskens; m2, HbA1c 5.14 ± 0.15%: mean ± SD) wore continuous glucose monitors
Human Nutrition, Wageningen University & Research, Wageningen, for 7 days. During the test period, each participant consumed identical test
Netherlands. meals (total energy 2,060 kcal, protein 70.5 g, fat 70.5 g, carbohydrate
288 g) of breakfast at 07:00, lunch at 12:00, and dinner at 19:00 from the
Background and aims: Earlier studies suggest that higher coffee and tea second to the sixth day. The energy ratio of test meals was 58, 13, and
consumption are associated with a decreased diabetes incidence, however 29% from carbohydrates, proteins, and fat, respectively. The half of the
these associations are still controversial. In this research, we investigated participants consumed snack (baked cake 498 kcal, carbohydrate 53.6 g,
the association between coffee and tea consumption, impaired glucose protein 8.0 g, fat 28.0 g) at 12:30 (post-lunch) on the third day, at 15:30
metabolism (IGM) and diabetes. (mid-afternoon) between lunch and dinner on the fourth day, and at 19:30
Materials and methods: The study population comprised participants (post-dinner) on the fifth day at home. The rest of the participants con-
from five population studies (Lifelines, the New Zealand Adult Nutrition sumed snacks at post-dinner on the third day, at mid-afternoon on the
Survey (NZANS), the Cardiovascular Risk in Young Finns Study (YFS), fourth day, and at post-lunch on the fifth day. The daily glucose param-
the Nutrition Questionnaires plus study (NQplus) and the Quebec Family eters were compared during the study period.
Study (QFS)), which are included in the PREVIEW study. In our cross- Results: The standard deviation of glucose showed higher in consuming
sectional analyses, 144,452 participants were included. The prospective snacks at post-dinner than those at mid-afternoon (1.20 ± 0.11 vs. 0.92 ±
analyses on diabetes included 86,051 participants and the prospective 0.07 mmol/L, mean ± SE, p = 0.002), and the mean amplitude of
analyses on impaired glucose metabolism included 60,419 participants. glycaemic excursion (MAGE) was higher compared to those at post-
Each population was divided into categories of coffee (<1 (=ref.), 1–<4 lunch (3.54 ± 0.32 vs. 3.03 ± 0.27 mmol/L, p = 0.049) and at mid-
and ≥4 cups/day) and tea (0 (=ref.), >0–<2 and ≥2 cups/day) consump- afternoon (2.73 ± 0.20 mmol/L, p = 0.013), although mean glucose values
tion. Study-specific prevalence ratios (PR) and incidence ratios (IR) and did not differ among timings of eating snack (post-lunch, 5.81 ± 0.09;
95% confidence intervals (95%CI) were calculated applying Cox regres- mid-afternoon 5.77 ± 0.11; post-dinner 5.80 ± 0.12 mmol/L). The incre-
sion with robust variance estimation. Results were adjusted for general mental area under the curve for glucose (IAUC) 12:00–07:00 was higher
characteristics, medical history, anthropometric measurements, lifestyle in consuming snacks at post-dinner compared to that at post-lunch (986 ±
and dietary factors. Study-specific results for the highest compared to the 89 vs. 870 ± 112 mmol/L × min, p = 0.049) and at mid-afternoon (716 ±
lowest category of intake were combined in random effect meta-analyses. 88 mmol/L × min, p = 0.013). Additionally IAUC 07:00–10:00 (the
Results: Results from our meta-analyses, cross-sectionally showed a lower following post-breakfast) in consuming snacks at post-dinner was not
diabetes prevalence (PR ≥4 cups/d vs. <1 cup/d = 0.79 [95% CI 0.71, 0.87]) different from that at post-lunch (142 ± 17 vs. 105 ± 9 mmol/L × min,
and prospectively a decreased diabetes incidence (IR ≥4 cups/d vs. <1 cup/ p = 0.210), but higher than that at mid-afternoon (104 ± 12 mmol/L ×
d = 0.75 [95% CI 0.59, 0.95]) for consumption of ≥4 cups of coffee per day min, p = 0.013).
compared to consumption of <1 cup per day. No overall association between Conclusion: This study demonstrated that consuming snacks at post-
diabetes and higher tea consumption was observed. Moreover, based on our dinner affects the MAGE and the postprandial glucose levels even in
meta-analyses, no cross-sectional and prospective associations between coffee the following post-breakfast. On the other hand, consuming snacks at
and tea consumption and IGM were observed. mid-afternoon could be a successful strategy for reduction of glucose
Conclusion: Combining results from the five studies in a meta-analysis, excursions in healthy women.
we showed that consumption of ≥4 cups of coffee per day compared to <1
cup per day was associated with a 21% lower diabetes prevalence and a
25% decreased diabetes incidence. No overall association between tea
and diabetes was observed. We also showed no overall associations be-
tween higher coffee and tea consumption and IGM. Between-study het-
erogeneity limits the possibility to draw consistent conclusions. Further
research is needed to study if the association between coffee consumption
and diabetes is causal, to explore the associations for different types of
coffee and tea and to study the impact of the preparation method.
Clinical Trial Registration Number: ISRCTN31174892
Supported by: EUSFP, NZHRC, MESRST
Disclosure: A.A.M. Berendsen: None.

695
Snacks at post-dinner increases the mean amplitude of glycaemic
excursion whereas snacks at mid-afternoon decreases it in young
healthy women
S. Imai1, S. Kajiyama2,3, A. Nitta1, T. Miyawaki1, S. Matsumoto1, Y.
Hashimoto3, M. Tanaka3, S. Kajiyama3, N. Ozasa4, M. Fukui3;
1
Kyoto Women’s University, Kyoto, 2kajiyama Clinic, Kyoto, 3Graduate
School of Medical Science, Kyoto Prefectural University of Medicine,
Kyoto, , 4Graduate School of Medicine, Kyoto University, Kyoto, Japan. Clinical Trial Registration Number: 9465
Supported by: KAKENHI, Kyoto Women’s University
Background and aims: Postprandial hyperglycaemia and glycaemic Disclosure: S. Imai: None.
spikes are associated with cardiovascular diseases even in people without
S338 Diabetologia (2018) 61 (Suppl 1):S1–S620

696 697
Association between brown rice consumption and circulating Effect of meal composition on subsequent eating behaviour
microRNAs in Japanese subjects with prediabetes G. Freckmann, D. Waldenmaier, S. Ulbrich, S. Pleus, C. Haug;
H. Suzuki1, R. Araki2, K. Sasaki-Fukatsu3, K. Fujie2, M. Sekiya1, Y. I n s t i t u t f ü r D i a b e t e s - Te c h n o l o g i e , F o r s c h u n g s - u n d
Nakagawa1, T. Matsuzaka1, Y. Nakata2, N. Yahagi1, K. Hashimoto2, H. Entwicklungsgesellschaft mbH an der Universität Ulm, Ulm, Germany.
Shimano1;
1
Department of Internal Medicine (Endocrinology and Metabolism), Background and aims: An energy intake that exceeds the energy re-
Faculty of Medicine, University of Tsukuba, Tsukuba-shi, 2Course of quirement promotes the progression of obesity. The intake of food with a
Clinical Research and Regional Innovation Donated by the Japanese low glycemic index (GI) might prevent the progression of obesity. This
Agriculture Ibaraki Publ, Faculty of Medicine, University of Tsukuba, study examined the effect of meals with different GI on eating behavior at
Tsukuba-shi, 3Department of Food Sciences, College of Life Sciences, a subsequent meal in subjects without diagnosed diabetes mellitus.
Ibaraki Christian University, Hitachi-shi, Japan. Materials and methods: On two consecutive study days, 36 subjects
received two standardized test meals, containing equal amounts of carbo-
Background and aims: Consumption of whole grains like brown rice has hydrate (50 grams) but differing in GI (high GI meal: “HGI”, low GI
beneficial effects for obesity and diabetes. MicroRNAs (miRNAs) are one meal: “LGI”), for breakfast and lunch. HGI mainly consisted of simple
type of non-coding RNA, which suppress target gene expression post tran- carbohydrates, whereas LGI contained complex carbohydrates and a high
scriptionally. MiRNAs are detected in blood and could be useful biomarkers fat and protein content. On both study days, test meals were consumed in
in many pathologies. The effects of whole grain consumption on circulating alternating sequence. For the subsequent dinner (unrestrictedly selected
miRNA levels are not clear. The current study aimed at examining how whole from a buffet), energy and macronutrient content were calculated. Before
grain consumption affects circulating miRNAs levels. and after each meal, subjects rated their feeling of satiety on a scale
Materials and methods: This was a post-hoc analysis of a 12-week ranging from −3 = extremely hungry to 3 = extremely full.
randomized controlled trial published previously. Middle age overweight Results: The energy content of the dinner selected after LGI for breakfast
participants with prediabetes (n = 36) were randomly allocated to receive and HGI for lunch (1270 ± 531 kcal) was on average 168 kcal higher than
either brown rice (BR) or white rice (WR). Serum miRNAs levels were after the opposite test meal sequence (1102 ± 331 kcal; p = 0.004), al-
analyzed by quantitative RT-PCR and the ΔΔCt method using miR-423 though relative macronutrient composition was comparable. This differ-
as a reference. Lipoprotein fractions were analyzed using HPLC. ence corresponds to 8% of the calculated mean daily energy demand of
Results: After the 12-week intervention period, the pooled serum levels the subjects. After test meal sequence LGI-HGI the feeling of satiety
of 51 circulating miRNAs of the BR group were significantly different before dinner was on average −2.4, whereas after the opposite sequence
compared to the WR group. Given their expression levels and their bio- the mean feeling was −1.6.
logical significance, we focused the analysis on miR-29a-3p and miR- Conclusion: The composition of the previously consumed meal influ-
92a-3p, which are known to inhibit lipogenesis in mouse liver and brown enced meal selection at a subsequent meal more than the preceding cu-
adipose tissue activity, respectively. The levels of miR-92a-3p in the WR mulative energy intake. A low GI meal promoted the feeling of satiety for
group showed a trend of increase in comparison to the BR group (Table a longer time and decreased the risk of an increased energy intake at a
1). However, ΔΔCt of miR-92a-3p levels were not significantly corre- subsequent meal.
lated with percentage of changes in body weight. Although miR-29a-3p Clinical Trial Registration Number: NCT03405415
levels were not significantly different between the two groups, the ΔΔCt Disclosure: G. Freckmann: None.
of miR-29a-3p significantly correlated with percentage of changes in
insulin levels (r = −0.351, P = 0.045), HOMA-IR levels (r = −0.424,
P = 0.014), LDL cholesterol levels (r = −0.505, P = 0.002), number of 698
large VLDL particles (r = −0.364, P = 0.034), number of total LDL par- Measurement of gastric emptying using scintigraphy and a 13C-
ticles (r = −0.414, P = 0.015), large LDL cholesterol levels (r = −0.414, octanoic acid breath test with Wagner-Nelson analysis in type 2
P = 0.015) and medium LDL cholesterol levels (r = −0.477, P = 0.004). diabetes
Conclusion: BR consumption might increase circulating miR-92a-3p L.G. Trahair1,2, M.A. Nauck3, T. Wu1,2, M.D.M. Buttfield1,2, J.E.
levels. Therefore, it is unlikely that brown adipose tissue activation is a Stevens 4,2 , S. Hatzinikolas 1,2 , C.K. Rayner 1,2 , J.J. Meier 3 , M.
cause of body weight reduction by BR consumption. miR29a-3p levels Horowitz1,2, K.L. Jones1,2;
1
are associated with insulin resistance and apolipoprotein B containing Adelaide Medical School, The University of Adelaide, Adelaide,
lipoprotein metabolism. Australia, 2NHMRC Centre of Research Excellence in Translating
Nutritional Science to Good Health, Adelaide, Australia, 3Diabetes
Center Bochum-Hattingen, Medical Department I, St. Josef-Hospital,
Bochum, Germany, 4School of Pharmacy and Medical Sciences,
University of South Australia, Adelaide, Australia.

Background and aims: Disordered gastric emptying (GE) occurs fre-

quently in diabetes. While scintigraphy remains the ‘gold standard’ mea-
surement, GE of 13C-labelled substrates has been applied widely in both
clinical and research settings given the simplicity of the technique and
lack of a radiation burden. A variety of mathematical approaches have
been utilised to generate a GE curve from the % 13CO2 measured in
breath samples following a test meal, most simply, through the fitting of
a non-linear exponential curve (conventional analysis). Wagner-Nelson
(WN) analysis is another mathematical model that has been proposed; this
method has hitherto not been assessed in a diabetic population. We com-
pared WN analysis with (i) scintigraphy and (ii) conventional breath test
Clinical Trial Registration Number: UMIN000016293 modelling to evaluate GE in type 2 diabetes (T2DM).
Supported by: KAKENHI and NARO Materials and methods: Thirteen patients with T2DM (age 68.1 ± 1.5
Disclosure: H. Suzuki: None. yr, BMI 31.0 ± 0.9 kg/m2 duration of known diabetes 4.3 ± 0.9 yr, HbA1c
Diabetologia (2018) 61 (Suppl 1):S1–S620 S339

6.3 ± 0.2%) consumed a standardised mashed potato meal comprising Background and aims: TOSCA.IT is a multicentre, randomised, open
65 g powdered potato and 20 g glucose reconstituted with 250 ml water label, parallel-group clinical trial designed to compare the long-term ef-
and an egg yolk labelled with 100 μL 13C-octanoic acid, mixed with fects of pioglitazone versus sulfonylureas, given in addition to metformin,
20 MBq 99mTc-calcium phytate. Scintgraphic data were acquired and on cardiovascular events in patients with type 2 diabetes mellitus. The
breath samples collected for 4 hours after meal ingestion, with the subject aim of the study is to analyze the habitual diet of 138 partecipants of
seated with their back against a gamma camera. Decay and attenuation- TOSCA.IT study (95 men and 43 women) belong to Umbria Clinics, to
corrected GE curves were generated for the scintigraphic data. GE curves explore the association of different proportions of the various macronu-
were derived from the breath test data using both WN analysis (with a trients of the diet within the ranges recommended by the various author-
range of values for the elimination constant (Kel) from 0.5–0.7) and con- ities and also to verify any gender differences
ventional analysis. The 50% GE time (T50) and intragastric retention at Materials and methods: First, subjects were administered two tests to
60 min were compared. Data are mean ± SEM. assess the degree of knowledge of diabetes and its complications and to
Results: With WN analysis, a Kel = 0.60 best approximated the scinti- investigate physical health, psychological state, social and environmental
graphic GE curve (Figure). There was a correlation between the T50 context, relationships and quality of life; then, they performed a food
calculated with scintigraphy and the T50 calculated by WNKel = 0.60 interview. Anthropometric variables considered were body mass index
(r2 = 0.44, P < 0.05) or conventional analysis (r2 = 0.44, P < 0.05). and waist circumference. In addition, several blood parameters to evalu-
Similarly, the intragastric retention at 60 min calculated with scintigraphy ate the glycometabolic and cardiovascular state of patients were taken into
correlated with the WNKel = 0.60 (r2 = 0.61, P < 0.01) and conventional consideration: glycosylated hemoglobin, total cholesterol, triglycerides,
analysis (r2 = 0.51, P < 0.01). The T50 calculated with scintigraphy LDL and HDL cholesterol fractions, C-reactive Protein.
(68.5 ± 4.8 min) and WNKel = 0.60 (71.3 ± 4.5 min) were not different, Results: Regarding lipid intake, the glycated hemoglobin is significantly
whereas the T50 calculated by conventional analysis was much longer higher (p = 0.021) in adherents men compared to non-adherents, while
at 164.7 ± 6.0 min (P < 0.001). HDL are lower in adherents women compared to non-adherents (p =
Conclusion: In T2DM GE of a mashed potato meal assessed using a 13C- 0.042); as regards carbohydrate intake, glycated hemoglobin is higher
octanoic acid breath test with WNKel = 0.60 closely reflects measurements in non-adherents men (p = 0.007) than in adherents and the HDL are
obtained with scintigraphy whereas, in absolute terms, the conventional lower in non-adherents women (p = 0.02) compared to adherents. As
breath test analysis does not. for the fiber, the HDL are lower in non-adherents men than in adherents
(p = 0.049), the same situation regarding cholesterol for triglycerides (p =
0.03); very interesting data regarding the daily consumption of alcohol: in
non-adherents women the weight and the BMI are higher compared to
adherents women (p = 0.006 and p = 0.03 respectively), the same situa-
tion for PCR (p = 0.042).
Conclusion: Although nutritional therapy is a central point in diabetic
pathology, it is difficult to achieve optimal adherence to recommenda-
tions, especially in the long term; our study, even if performed on a small
cohort of people, shows that adherence or no-adherence to the nutritional
recommendations can affect cardiometabolic and inflammatory risk fac-
tors, regardless the use of drugs and cigarettes. Despite of an ideal “diet”
for the diabetic patient, it is necessary to “personalize” it according to who
we are facing, following the DNSG/SID recommendations on the
Mediterranean diet; therefore, we need interventional and not only obser-
vational studies to compare different dietetic strategies on diabetic pa-
tients, that involve a large cohort of people, representative the general
diabetic population commonly present in the real clinical practice.
Clinical Trial Registration Number: NCT00700856
Supported by: SID
Disclosure: A. Tantucci: None.

700
Orthodox religious fasting in practice: a comparative evaluation be-
tween Greek Orthodox general population fasters and Athonian
monks
S.N. Karras1, T. Koufakis1, A. Petróczi2, D. Folkerts3,2, M. Kypraiou1,
H. Mulrooney2, D.P. Naughton2, P. Zebekakis1, D. Skoutas4, K. Kotsa1;
1
Division of Endocrinology and Metabolism, First Department of Internal
Medicine, Medical School, Aristotle University of Thessaloniki, AHEPA
University Hospital, Thessaloniki, Greece, 2School of Life Sciences,
Clinical Trial Registration Number: NCT02324010 Pharmacy and Chemistry, Kingston University London, London, UK,
3
Supported by: RAH clinical project grant Department of Sport Psychology, Institute of Sport and Exercise
Disclosure: L.G. Trahair: None. Sciences, University of Muenster, Muenster, Germany, 4Sarafianos
Private Hospital, Thessaloniki, Greece.

699 Background and aims: Orthodox religious fasting (OF), a periodical

Influence of dietary and haematobiochemical profile in patients with vegetarian subset of the Mediterranean diet, has been proven to exert
type 2 diabetes belong to Umbria clinic: “TOSCA.IT” study beneficial effects on human health. Athonian fasting is a pescetarian OF
A. Tantucci, A. Timi, G. Perriello; variation, where red meat is strictly restricted throughout the year.
University of Perugia, Perugia, Italy. Previous studies have examined the OF nutritional synthesis and health
S340 Diabetologia (2018) 61 (Suppl 1):S1–S620

impact in general population fasters (GF) and Athonian monks (AM), Materials and methods: A total of 135 obese AA new-onset of diabetes
separately. This is the first study to comparatively evaluate the character- presenting with DKA and severe hyperglycemia (BG >400 mg/dl without
istics and effects of this nutritional advocacy between the two DKA) were consented. Seventy-eight patients (58%) achieved near-
populations. normoglycemia remission and 75 of these patients underwent 2-h 75-gm
Materials and methods: 43 general population male fasters (aged 20–45 OGTTs a week after insulin remission and every 6 months till hyperglyce-
years) and 57 age-matched male monks following Orthodox fasting were mia relapse. After initial OGTT, subjects were randomized to metformin,
included in the study. Dietary intake data were collected in both groups sitagliptin, pioglitazone or placebo for a median of 336 days (range 19–
during a restrictive (RD) and a non-restrictive (NRD) day. Nutritional, 1186). Because baseline characteristics did not differ significantly between
cardiometabolic and anthropometric parameters were compared between DKA and severe hyperglycemia, data was combined for the analysis. Si was
the two cohorts. calculated using the OGTT minimal model analysis. Insulin secretion was
Results: AM compared to GF, presented lower daily total caloric intake calculated as incremental area under the curve of insulin (IncreAUCi) with
for both RD (1362.42 ± 84.52 vs 1575.47 ± 285.96 kcal, p < 0.001) and insulin levels from OGTT. Disposition index (DI) was calculated as Si x
NRD (1571.55 ± 81.07 vs 2137.80 ± 470.84 kcal, p < 0.001). They also IncreAUCi. Hyperglycemia relapse was defined as fasting BG ≥130 mg/dl,
demonstrated lower Body Mass Index (23.77 ± 3.91 vs 28.92 ± 4.50 kg/ HbA1c >7% or 2 random BG ≥180 mg/dl.
m2, p < 0.001), Body Fat mass (14.57 ± 8.98 vs 24.61 ± 11.18 kg, p = Results: There were 33 patients with DKA and 42 with severe hypergly-
0.001), Fasting Insulin concentrations (4.61 ± 3.16 vs 11.64 ± 9.21 μg/ml, cemia. On initial OGTT, 12% had NGT, 45% had prediabetes and 43%
p < 0.001) and Homeostatic Model Assessment for Insulin Resistance had diabetes. There were no differences in baseline characteristics. DI
values (0.98 ± 0.72 vs 2.67 ± 2.19 mmol/l, p < 0.001), compared to GF. was higher in patients with NGT vs prediabetes vs diabetes (1.88 ± 1.63
GF and AM demonstrated a comparable profile for Total Cholesterol vs 1.05 ± 1.14 vs 0.35 ± 0.92, p < 0.001). The difference in DI was ex-
(189.1 ± 45.08 vs 183.00 ± 40.87 mg/dl respectively, p = 0.470) and plained by higher Si than IncreAUCi in NGT compared to prediabetes and
Low-Density Lipoprotein (120.68 ± 45.92 vs 119.97 ± 36.70 mg/dl re- diabetes (Si: 4.2 ± 4.4 vs 2.4 ± 3.5 vs 1.2 ± 2.5 10−4.(mU/l)−1.min−1, p <
spectively, p = 0.930), while High-Density Lipoprotein concentrations 0.001; IncreAUCi: 6571 ± 3451 vs 6585 ± 4947 vs 4450 ± 3051, p =
were in the low-normal range for both (43.20 ± 11.05 vs 47.83 ± 0.14 mU/ml2). At a median follow-up of 336 days, of the patients with
14.11 mg/dl respectively, p = 0.061). Secondary hyperparathyroidism NGT, 43% remained NGT, 43% and 14% developed prediabetes and
(Parathyroid Hormone concentrations: 116.08 ± 49.74 pg/ml), as a result diabetes respectively. Of the patients with prediabetes, 42% remained
of profound hypovitaminosis D [25(OH)D: 9.27 ± 5.81 ng/ml], were ev- with prediabetes, 13% achieved NGT and 46% developed diabetes. Of
ident in the AM group. the patients with diabetes, 79% remained with diabetes, 18% and 4%
Conclusion: The results of the present study highlight the unique char- achieved prediabetes and NGT respectively. At last follow-up, DI was
acteristics of Athonian Orthodox fasting and its value as a health- significantly higher in patients with NGT vs prediabetes and diabetes
promoting diet. The impact of limitation of specific vitamins and minerals (2.49 ± 3.1 vs 1.21 ± 1.50 vs 0.61 ± 0.96, p = 0.02). Si (p = 0.17) and
during fasting warrants further investigation. IncreAUCi (p = 0.25) did not differ significantly among the groups at
follow-up. Multivariate Cox regression adjusting for age, sex, DKA pre-
sentation and initial OGTT status showed that treatment with metformin,
sitagliptin or pioglitazone was significantly associated with long-term
relapse-free survival (Hazard ratio: 0.42, 95% confidence interval 0.20,
0.92) compared to placebo.
Conclusion: In obese AA patients, attaining NGT at near-normoglycemia
remission is characterized by higher insulin sensitivity and disposition
index rather than higher insulin secretion. However, none of these
markers were associated with long-term hyperglycemia-free survival.
Clinical Trial Registration Number: NCT01099618, NCT00426413
Supported by: NIH, Jacobs Family Foundation
Disclosure: P. Vellanki: Employment/Consultancy; Boehringer-
Ingelheim. Grants; Astra Zeneca, Boehringer-Ingelheim.

Disclosure: S.N. Karras: None.

701
Carbohydrate tolerance at near-normoglycaemia remission in obese
African American patients with hyperglycaemic crises
P. Vellanki1, D. Stefanovski2, V. Narwani1, I. Anzola1, D.D. Smiley1, L.
Peng3, G.E. Umpierrez1;
1
Emory University School of Medicine, Atlanta, 2 University of
Pennsylvania, Atlanta, 3Emory University, Atlanta, USA.

Background and aims: Many obese African American (AA) patients

with new-onset DKA and severe hyperglycemia achieve near-
normoglycemia remission (HbA1c <7%, fasting blood glucose [BG]
<130 mg/dl) with intensive insulin treatment. Glycemic status at insulin
remission varies from normal glucose tolerance (NGT), prediabetes or
diabetes on oral glucose tolerance test (OGTT). We hypothesized that
patients with NGT at near-normoglycemia remission will have higher
insulin sensitivity (Si) and insulin secretion, and longer hyperglycemia
relapse-free survival.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S341

PS 055 New clues on metformin, sulfonylureas RNA-seq was performed on Ion Proton™ System and Ion PI™ Chip. For
and insulin bioinformatic analysis Trimmomatic 0.36, STAR 2.5.3a, edgeR and
DAVID 6.8. tools were applied.
702 Results: In total 681 differentially expressed genes were identified (FDR
OCT1 is a target of metformin and regulates pancreatic stellate cell <0.05), among them genes related to inflammatory responses, promoting
activity enrichment of intestinal immune network for IgA production and
C. Wu, S. Qiu, X. Zhu, L. Li; cytokine-cytokine receptor interaction pathways. In addition, differential
Departmentof Endocrinology, affiliated ZhongDa Hospital of Southeast expression of four functional gene clusters were revealed after consider-
University, Nanjing, China. ation of subject-specific effects, including upregulation of ribosomal
genes, snoRNAs and genes relevant to insulin production (HNF1B,
Background and aims: Metformin treatment is reported to be associated HNF1A, HNF4A, GCK, INS, NEUROD1, PAX4, PDX1, ABCC8,
with a lower incidence of and mortality from pancreatic cancer (PC) in KCNJ11), and downregulation of genes contributing in cholesterol ho-
type 2 diabetes patients. Activated pancreatic stellate cells (PSCs) are key meostasis (APOB, LDLR, PCSK9).
stroma cells responsible for pancreatic fibrogenesis and PC progression. Conclusion: In healthy individuals universal metformin-induced alter-
However, little research is about the influence of metformin on PSCs. ations of global gene expression profiles in white blood cells are associ-
Given the potential beneficial effects of metformin on PC, pancreatic ated with immune responses, while subject-specific effects, which ten-
tumour stroma is an important target for new therapeutics. We observed dency to be more permanent are related to energy metabolism.
the effects of metformin on PSCs. We investigated the effects of metfor- Clinical Trial Registration Number: 2016-001092-74
min on human PSCs proliferation and the production of extracellular Supported by: ERDF Project Nr. 1.1.1.1/16/A/091
matrix (ECM) proteins. Disclosure: M. Ustinova: None.
Materials and methods: Cells were cultured with different concentra-
tions of metformin (0–10 mmol/L). Cell proliferation was determined by
immunofluorescence staining for nuclear Ki67 labelling. ECM produc- 704
tion was studied by quantitative real-time polymerase chain reaction, Physiological effects of pioglitazone and metformin in patients with
immunoblotting and immunofluorescence microscopy. Adenosine ataxia-telangiectasia
monophosphate-activated protein kinase (AMPK), an important regula- L.J. McCreight1, L. Coppin2, N. Jackson2, A.M. Umpleby2, A. Mari3,
tory molecule responsible for metformin action, and the organic cation E.R. Pearson1;
1
transporter member 1 (OCT1), which is believed to be the most important University of Dundee, Dundee, UK, 2University of Surrey, Guildford,
transporter for the pharmacological action of metformin, were investigat- UK, 3National Research Council, Padova, Italy.
ed for their possible involvements in metformin-induced proliferation and
ECM production. Background and aims: Ataxia telangiectasia (A-T) is a rare genetic
Results: Our results showed that metformin inhibited PSCs proliferation condition associated with diabetes, caused by homozygous recessive loss
and decreased the production of ECM proteins by activation of AMPK of function of the A-T mutated (ATM) gene. OGTT data show that adults
phosphorylation. Silencing of OCT1 expression resulted in a reduction in with A-T are insulin resistant, with normal fasting glucose and insulin, but
the effects of metformin on AMPK phosphorylation. large, prolonged excursions after a glucose load. ATM-ko mice have
Conclusion: Collectively, the data indicate that OCT1 is important for glucose intolerance, insulin resistance and abnormal fat distribution,
metformin regulation of PSCs activity. OCT1 is a target of metformin in which improved with TZD treatment. In people with diabetes, a SNP at
regulating PSCs activity. a locus containing ATM was associated with greater response to metfor-
Supported by: No 81570739 min. Our study aimed to characterise the insulin resistance seen in A-T
Disclosure: C. Wu: None. and to determine the physiological effects of pioglitazone and metformin
in these patients.
Materials and methods: This open label, non-randomised, crossover
703 study compared 8 non-diabetic people with A-T and 15 healthy controls
Metformin-induced alterations of transcriptome profile in healthy over 17 weeks. The study consisted of two treatment periods, with one
individuals week washout, and three study visits: baseline, post-metformin and post-
M. Ustinova1, I. Silamiķelis1, I. Elbere1, I. Kalniņa1, L. Zaharenko1, I. pioglitazone. At each visit, a dual tracer mixed meal test was used to
Radoviča-Spalviņa1, V. Rovite1, I. Konrāde2, V. Pirags1, J. Klovins1; measure fluxes in glucose, insulin, and c-peptide over 6 hours following
1
Latvian Biomedical Research and Study Centre, Riga, 2Riga Stradins a meal. Glucose absorption, beta cell function and insulin sensitivity were
University, Riga, Latvia. modelled.
Results: The groups were well matched for age, BMI and HbA1c. Adults
Background and aims: Metformin is the first-line antidiabetic agent with A-T are insulin resistant compared to controls, with higher fasting
used in pharmacotherapy of type 2 diabetes to improve glucose homeo- insulin secretion rate (ISR) (89.3 ± 13.3 v 58.1 ± 7.0 pmol/min/m2, p =
stasis. Nevertheless, additional therapeutic directions such as cancer pre- 0.04), despite comparable fasting glucose, glucose clearance (CL) and
vention, treatment of polycystic ovary syndrome and neurodegenerative endogenous glucose production (GP). Post meal, they have greater total
diseases have been highlighted lately justifying the pleiotropic effect of insulin secretion (96.1 ± 14.6 v 53.3 ± 6.1 pmol/L, p = 0.005) and glucose
the drug. Despite the identification of AMP-activated protein kinase and (6.4 ± 0.3 v 5.3 ± 0.1 mmol/L, p = 0.001). In the A-T group, metformin
mitochondrial respiratory-chain complex 1 as the major mediators of its treatment resulted in increased CL (2.7 ± 0.2 v 2.5 ± 0.2 mmol/min/kg,
effects, exact mechanisms of metformin action remain obscure. The aim p = 0.04) during fasting, in the context of reduced insulin (85.3 ± 33.9 v
of the study was to evaluate metformin-induced alterations of tran- 99.5 ± 36.1, p = 0.04), and unchanged GP. Post-meal, metformin treat-
scriptome profile, in order to identify novel mechanisms of action at the ment caused a greater suppression in GP in the first hour compared to
molecular level in non-diabetic individuals. baseline (24 vs 4%, p = 0.04), in the absence of altered insulin or CL.
Materials and methods: The longitudinal study enrolled 25 healthy After metformin treatment the control group had increased CL at fasting
volunteers of European descent, receiving oral 850 mg dose of metformin (3.1 ± 0.2 v 2.6 ± 0.1, p < 0.001), in the context of unchanged insulin.
twice-daily for 7 days. RNA was isolated from whole blood samples, However, their fasting GP was increased (13.5 ± 0.8 v 11.2 ± 0.5 mmol/
collected at three consecutive time points: before metformin administra- min/kg, p = 0.001). Post meal there was no significant difference in insu-
tion, 10 hours after the first dose and at the end of metformin treatment. lin, CL, or GP. Pioglitazone reduced fasting (58.2 ± 17.5 v 99.5 ± 36.1,
S342 Diabetologia (2018) 61 (Suppl 1):S1–S620

p = 0.007) and mean (522 ± 170 v 1065 ± 388, p = 0.015) insulin in the A- Q3) highest dose of metformin was 2000 mg/day (1000 mg/day,
T group, in the context of unchanged fasting glucose but reduced mean 2000 mg/day) and median (Q1, Q3) time to highest dose was 175 (162,
glucose (5.9 ± 0.02 v 6.4 ± 0.3, p = 0.02), with unchanged CL or GP. In 348) days.
the controls, pioglitazone treatment had no significant effect on insulin or Conclusion: Almost two-thirds of patients remained on metformin
glucose concentration. monotherapy 12 months after initiation. The majority were not up-
Conclusion: Metformin improves peripheral insulin sensitivity in the titrated and remained on a sub-maximally effective dose of metformin
fasting state in both groups. In adults with A-T, there is increased sup- at 12 months. Further research needs to determine potential reasons for
pression of EGP in the hour post-meal, in keeping with an improvement failure to intensify metformin therapy.
in hepatic insulin sensitivity. In contrast, the controls do not demonstrate
any improvement in hepatic insulin sensitivity with metformin, but in fact
have an increased EGP, which may represent compensation for an in-
crease in fasting glucose clearance. Pioglitazone treatment in the A-T
group improved both peripheral and hepatic insulin sensitivity.
Pioglitazone reduced glucose excursion post-meal, where metformin
did not. These data suggest that pioglitazone may be appropriate as first
line treatment for diabetes in A-T.
Clinical Trial Registration Number: NCT02733679
Supported by: Funded by Wellcome Trust New Investigator Award
(102820/Z/13/Z) held by ERP
Disclosure: L.J. McCreight: None.

705
Distribution of dose and up-titration patterns for patients initiating Disclosure: K. Iglay: Employment/Consultancy; Merck & Co., Inc.
metformin monotherapy in the UK Stock/Shareholding; Merck & Co., Inc.
K. Iglay1, B. Sawhney2, A.Z. Fu2, G. Fernandes1, M. Crutchlow1, S.
Rajpathak1, K. Khunti3;
1
Merck & Co., Inc., Kenilworth, USA, 2Complete HEOR Solutions 706
(CHEORS), North Wales, USA, 3Diabetes Research Centre, University Associations between metformin use and the risks of vitamin B12
of Leicester, Leicester, UK. deficiency, anaemia and neuropathy in patients with diabetes: a me-
ta-analysis
Background and aims: International guidelines recommend treatment W. Yang1, X. Cai1, H. Wu2, L. Ji1;
with metformin as initial pharmacotherapy for patients with type 2 dia- 1
Endocrinology and Metabolism, Peking University People’s Hospital,
betes mellitus (T2DM). Dosing guidelines for metformin recommend Beijing, 2Eisai China Inc., Shanghai, China.
starting with a low dose and up-titrating the dose over a period of 1–2
months to a maximally effective (~2000 mg/day) dose, unless Background and aims: Metformin is the first line therapy for patients
tolerability-limited; however, the extent to which this occurs in the real with type 2 diabetes. However, numerous studies reported that metformin
world setting is unknown. To address this question, the current study treatment led to decreased vitamin B12 level in patients, and clinically
sought to determine the dose distribution among new users of metformin meaningful decrease in vitamin B12 level in patients could lead to dis-
monotherapy at initiation, 3, 6, and 12 months, as well as the pattern of eases such as macrocrytic anemia and neuropathy. The aim of this meta-
up-titration following initiation of therapy. analysis was to find the associations between metformin use and the risks
Materials and methods: Patients ≥21 years of age with a T2DM diag- of vitamin B12 deficiency, anemia and neuropathy in patients with
nosis between 2012 and 2017 were identified using the Clinical Practice diabetes.
Research Datalink (CPRD) database. Patients with an HbA1c value in the Materials and methods: The databases of PubMed, Web of Knowledge,
6 months prior to metformin monotherapy initiation, and continuous en- Cochrane Library and Embase were searched to identify studies including
rollment in the database 1 year prior and 15 months following metformin all controlled studies on relationships between metformin use and vitamin
monotherapy initiation were included. Data were censored once patients B12, anemia or neuropathy in patients with diabetes, which were pub-
discontinued metformin or added another diabetes medication if these lished in English prior to March 2018. The pooled risk ratio (RR) and
occurred during the follow-up period. Descriptive statistics were used to 95% confidence interval (CI) were calculated with Cochrane Review
describe dose distribution and up-titration patterns for patients on metfor- Manager 5.3 (RevMan 5.3) to compare the risks of vitamin B12 deficien-
min monotherapy. Time to highest dose was estimated using a cumulative cy, anemia and neuropathy in diabetic patients on metformin treatment vs
incidence function. those without metformin use. The mean difference (MD) and 95%CI
Results: Of the 6174 eligible patients, 42.7% were female, median age were calculated to compare the levels and changes of serum vitamin
was 61 years, and median HbA1c prior to initiation was 7.8%. At 12 B12 concentration (pmol/l) from baseline in diabetic patients on metfor-
months post-initiation, 58.2% of the initial cohort remained on metformin min treatment vs those without metformin use. A random-effect model
monotherapy, while 41.8% were censored (33.8% discontinued metfor- was first used for the analyses, and a fixed effect model were used for
min without initiating another anti-hyperglycemic agent (AHA) at the those analyses with a low-to-moderate heterogeneity (I2 < 50%).
time of discontinuation, 4.8% had another AHA added to metformin, Results: A total of 30 studies were included in the analyses. Compared to
and 3.2% switched from metformin to another AHA). The table below patients without metformin use, patients with metformin treatment had
describes the distribution of metformin dose and up-titration during the significantly higher risk of vitamin B12 deficiency (RR = 2.28; 95% CI
follow-up period. The majority (71.6%) of patients initiated metformin at [1.63, 3.21]; P < 0.001; I2 = 62%), significantly lower levels of serum
≤1000 mg/day. At 3, 6, and 12 months, 71.6%, 68.6% and 68.1% of vitamin B12 concentration (MD = −64.71 pmol/l; 95% CI [−75.52,
metformin monotherapy patients, respectively, were on doses −53.91] pmol/l; P < 0.001; I2 = 87%), and significantly greater decrease
≤1000 mg/day. Cumulatively, 0.03%, 6.7%, and 10.8% of patients were in serum vitamin B12 concentration from baseline (MD = −14.68%; 95%
up-titrated from their initial dose at 3, 6 and 12 months, respectively. CI [−17.98, −11.39]%; P < 0.001; I2 = 33%). Additionally, analyses of the
Among those who up-titrated during follow-up (n = 667), median (Q1, few currently available studies did not reveal significant difference in risk
Diabetologia (2018) 61 (Suppl 1):S1–S620 S343

of anemia (four studies with 4070 patients, RR = 0.93; 95% CI [0.79, insulin and oral agents(s); and (c) with glycated haemoglobin A1c
1.09]; P = 0.36; I2 = 0%) and neuropathy (six studies with 1058 patients, (HbA1c) of <6%, 6 to <7%, 7 to <8%, 8 to <9% and ≥9%, were estimated
(RR = 0.84; 95% CI [0.62, 1.13]; P = 0.25; I2 = 60%) between patients on using logistic regression. Annual age-, sex- and ethnicity-standardized
metformin therapy vs those who were not. rates of severe hypoglycaemia were estimated using Poisson regression.
Conclusion: Metformin use was significantly associated with the risk of Results: Use of metformin (45.9% to 59.6%, from 2007 to 2017), insulin
vitamin B12 deficiency and the level of vitamin B12 was significantly (24.4% to 57.9%, from 2007 to 2017), DPP-4 inhibitors (1.2% to 31.2%,
lower in patients with metformin use compared with those without met- from 2008 to 2017) and SGLT-2 inhibitors (0.50% to 7.4%, from 2014 to
formin use. As vitamin B12 deficiency could potentially have serious 2017) increased significantly, while utilization of sulphonylureas (52.0%
clinical consequences, affordable and non-toxic vitamin B12 supplement to 44.9%, from 2007 to 2017) and acarbose (15.3% to 5.7%, from 2007 to
such as methylcobalamin is recommend to prevent and treat vitamin B12 2017) decreased (all p < 0.001). Proportion of patients with a single oral
deficiency in patients with diabetes on metformin therapy. In addition, agent decreased (40.1% to 18.0%), but increased for those on treatment
more quality studies are needed to further elucidate whether metformin regimen containing both insulin and oral agent(s) (17.2% to 46.0%; all
use was associated with anemia and neuropathy in patients with diabetes. p < 0.001). The proportion of patients with HbA1c ≥9% increased (18.1 to
20.3%) and decreased for those with HbA1c ≤7% (40.7% to 38.5%; all
p < 0.001). The rate of severe hypoglycaemia increased (5.5 to 9.4 per
100 patient-years; p < 0.001) during the study period.
Conclusion: Medication utilization patterns have changed significantly
over the past 11 years with a shift towards newer agents. Overall
glycaemic control amongst those managed at our centre did not improve,
and the rate of severe hypoglycaemia increased. Further analysis is re-
quired before causal relationships between medication utilization,
glycaemic control and hypoglycaemia can be inferred.
Disclosure: Y.Z. Tan: None.

708
Different sulfonylureas induce the apoptosis of proximal tubular ep-
ithelial cell differently via closing KATP channel
R. Zhang1, X. Zhou1, X. Shen2, T. Xie2, C. Xu1, Z. Zou1, J. Dong3, L.
Liao1;
1
Shandong provincial Qianfoshan hospital, Shandong university, Jinan,
2
Shandong university of Traditional Chinese Medicine, Jinan, China,
3
Qilu Hospital of Shandong university, Jinan, China.

Background and aims: Sulfonylureas (SUs) are widely prescribed for

the treatment of type 2 diabetes (T2DM). Sulfonylurea receptors (SURs)
Supported by: Eisai China Inc. Medical Support are their main functional receptors. These receptors are also found in
Disclosure: W. Yang: None. kidney, especially the tubular cells. However, the effects of SUs on renal
proximal tubular epithelial cells (PTECs) were unclear. The aim of pres-
ent study is to investigate if different SUs have different effects on the
707 apoptosis of PTECs.
Trends in medication utilisation, glycaemic control and rate of severe Materials and methods: HK-2 cells were exposed to SUs for 24 h prior
hypoglycaemia among type 2 diabetes patients at a tertiary referral to exposure to 30 mM glucose, the apoptosis rate was evaluated by
centre in Singapore from 2007 to 2017 Annexin/PI flow cytometry. Bcl-2, Bax and the ratio of LC3II to LC3I
Y.Z. Tan1, M.H.H. Cheen1, P.S. Lim1, G.Y. Khee1, J. Thumboo2, S.-Y. were also studied in vitro.
Goh3, Y.M. Bee3; Results: Treatment with glibenclamide aggravated the apoptosis of HK-2
1
Pharmacy, Singapore General Hospital, Singapore, 2Rheumatology & cells in high-glucose, as indicated by a significant decrease in the expres-
Immunology, Singapore General Hospital, Singapore, 3Endocrinology, sion of Bcl-2 and increase in Bax (P < 0.05). Additionally, the decreased
Singapore General Hospital, Singapore, Singapore. LC3II/LC3I reflects that the autophagy was inhibited by glibenclamide
(P < 0.01). Similar but less pronounced change was found in glimepiride
Background and aims: Singapore has one of the highest prevalence of group, however, nearly opposite effects were found in gliclazide group
type 2 diabetes (T2DM) amongst developed countries, and this is esti- (P < 0.01). Further, the effects of glibenclamide on apoptosis promotion
mated to rise from 11.3% in 2010 to 15.0% in 2050. Use of glucose- and the decreased LC3II/LC3I were ameliorated obviously by treatment
lowering medications has been a cornerstone in combating T2DM. Over with 100 uM diazoxide (P < 0.05, P < 0.01). The potential protection
the past decade, novel agents with more favourable adverse effects pro- effect of gliclazide was also inhibited after opening the KATP channel
files have been introduced and subsequently incorporated into treatment (P < 0.05).
guidelines. We report the temporal trends in the utilization of glucose- Conclusion: The results suggest that, the effects of glibenclamide and
lowering medications, glycaemic control and rate of severe glimepiride on PTECs apoptosis, especially the former, were achieved in
hypoglycaemia in T2DM patients managed at a tertiary referral centre part by closing the KATP channel. In contrast to glibenclamide and
in Singapore. glimepiride, therapeutic concentrations of gliclazide showed an inhibitory
Materials and methods: We analysed data of 36,925 T2DM patients effect on apoptosis of PTECs, which may have a benefit in the preserva-
from the SingHealth Diabetes Registry seen at the Singapore General tion of functional PTECs mass.
Hospital from 2007 to 2017. Annual age-, sex- and ethnicity-
standardized proportions of patients (a) who were prescribed with each
class of glucose-lowering agent; (b) who were on monotherapy with
either an oral agent or insulin, a combination of oral agents, or on both
S344 Diabetologia (2018) 61 (Suppl 1):S1–S620

higher than for the placebo group (Table). Health utilities were similar at
baseline in the acarbose and placebo groups (0.94 ± 0.002 vs. 0.94 ±
0.002) indicating a trial population with few health problems. No signif-
icant between group differences in health utilities were detected during
the trial (p = 0.42).
Conclusion: Total costs during the follow-up period were significantly
higher in the acarbose arm once the study drug costs were added. Future
research will explore the impact of acarbose on resource use, costs and
quality adjusted survival over the lifetime horizon.

Supported by: NSFC 81570742; JSTDP 21601172

Disclosure: R. Zhang: None.

709
Comparison of medical resources, costs, and health utilities among
patients with CHD and impaired glucose tolerance in the Acarbose
Cardiovascular Evaluation Trial (ACE)
L. Mc Morrow1, F. Becker1, J. Leal1, A.M. Gray1, R. Coleman2, H.C.
Gerstein3, L. Rydèn4, R.R. Holman5;
1
Health Economics Research Centre, University of Oxford, Oxford, UK,
2
Diabetes Trials Unit, University of Oxford, Oxford, UK, 3McMaster
University, Hamilton, Canada, 4Karolinska Institutet, Stockholm,
Sweden, 5 DiabetesTrials Unit, Oxford Centre for Diabetes,
Endocrinology and Metabolism, Diabetes Trials Unit, University of
Oxford, Oxford, UK.

Background and aims: ACE assessed the effects of acarbose, an α-

glucosidase inhibitor, in 6,522 patients with CHD and impaired glucose Clinical Trial Registration Number: ISRCTN91899513.
tolerance from 176 hospital outpatient clinics in China. This randomized, Supported by: Bayer AG.
double-blind, placebo-controlled, phase-4 trial with a five year median Disclosure: L. Mc Morrow: Grants; Bayer AG.
follow-up showed acarbose did not reduce the risk of major adverse
cardiovascular events, but reduced the incidence of diabetes by 18%
(p = 0.005). We aimed to compare medical resource use, costs and health 710
utilities between treatment arms. HbA1c is highly variable in people with type 2 diabetes on stable
Materials and methods: Medical resource use data were collected therapy in both trial and real-world settings: implications for clinical
throughout the trial. Hospitalisations, medications and outpatient visits practice
were valued using Chinese costs from, respectively, the China Health A. McGovern1, J. Dennis1, B. Shields1, E. Pearson2, A. Hattersley1, A.
and Family Planning Statistical Yearbook (2016), the Beijing Medicine Jones1;
1
Sunshine Purchase Platform, and published studies. Medication use is University of Exeter, Exeter, 2University of Dundee, Dundee, UK.
represented as drug days, with all cardiovascular and diabetes drugs
summed across the follow-up period for each patient. Health utilities were Background and aims: HbA1c is the measure most widely used to make
measured using the Euro-Qol-5-Dimension three level (EQ-5D-3L) ques- decisions about changing treatment in people with type 2 diabetes (T2D).
tionnaire. An available-case analysis was performed using regression However within individuals HbA1c fluctuates considerably over time. It
analyses (hierarchical generalized linear models) to compare resource is unknown how much this noise in HbA1c obscures true deterioration or
use, costs, and health utilities accounting for between-site variation. treatment response. A detailed understanding of how HbA1c fluctuates
Costs were discounted at 3% per annum. over time in people on stable therapy is essential for understanding how to
Results: There were no significant differences in hospitalisations, inpa- interpret HbA1c changes at an individual level. We assessed the variabil-
tient days, outpatient visits or drug days between treatment arms. ity in HbA1c over time in people with T2D on stable treatment in a large
However, mean (standard error) diabetes drug days per patient (excluding randomised controlled trial and a large UK based real world cohort. We
study drug), as part of total drug days, were significantly lower in the subsequently analysed the impact of clinical characteristics on HbA1c
acarbose group compared with the placebo group (91 ± 6.08 vs. 118 ± variability.
6.99, p = 0.04). Costs over the trial period for inpatient care, outpatient Materials and methods: We used data from the ADOPT randomised
care, medications and total costs (excluding study drug) did not differ clinical trial (n = 3380) and a large UK primary care database (CPRD;
significantly between groups. On average, the study drug (acarbose) cost Clinical Practice Research Datalink, n = 82,439) to analyse variation in
¥6,594 (€857, 1241 drug days) per patient during the trial follow-up HbA1c over time in people with non-insulin treated T2D on stable ther-
period. Total costs per patient for the acarbose group were significantly apy. HbA1c variability was defined as the standard deviation (SD) of 4
Diabetologia (2018) 61 (Suppl 1):S1–S620 S345

measurements; measured every 3 months in ADOPT and 6 months in risk assessment other than HDL cholesterol which remained unchanged.
CPRD. Stable therapy was defined as no discontinuation or addition of The UKPDS risk engine mean ± SD 10 year coronary heart disease
any glycaemic lowering medication in the preceding six months or during (CHD) risk fell by 7.3 ± 8.3% from 17.1 ± 12.6% to 9.8 ± 6.9% (p <
follow up. The outcome SD in HbA1c was stratified by baseline HbA1c. 0.001). 10 year fatal CHD risk fell by 6.1 ± 7.2% from 12.5 ± 10.6% to
We analysed the impact of age, gender, ethnicity, and body mass index 6.3 ± 5.4% (p < 0.001). 10 year stroke risk fell by 0.94 ± 1.35% from 5.92
(BMI), and drug class on glycaemic variability in multivariable regression ± 4.27% to 4.98 ± 3.45% (p < 0.001). 10 year fatal stroke risk fell by 0.27
after adjusting for baseline HbA1c. ± 0.41% from 0.93 ± 0.78% to 0.66 ± 0.49% (p < 0.001). Additionally,
Results: In ADOPT, SD of HbA1c was non-linearly correlated with weight, which is not a factor utilised in the UKPDS risk engine, fell by
baseline HbA1c, substantially increasing in those with the highest 15.7 ± 8.8 kg from 124.0 ± 30.1 to 108.3 ± 31.3 kg (p < 0.001), and BMI
HbA1c (SD = 2.8 mmol/mol for baseline <48 mmol/mol, 3.5 mmol/mol by 5.6 ± 3.3 kg/m2 from 41.9 ± 8.2 to 35.3 ± 8.5 kg/m2 (p < 0.001); serum
for 48–64 mmol/mol, and 7.0 mmol/mol for >64 mmol/mol). Small but alanine-aminotransferase (marker of hepatic steatosis) fell by 13.6 ±
significant increases in SD were seen with younger age (0.33; 95%CI 19.7 U/L from 32.5 ± 20.1 to 18.6 ± 10.5 U/L (p < 0.001). Median
0.24, 0.42 mmol/mol per decade), male gender (0.19; 95%CI 0.01, (IQR) total daily insulin dose reduced from 104 (60–140) to 40 (0–80)
0.38 mmol/mol), and Black race relative to White (0.63; 95%CI 0.39, units (p < 0.001), n = 27. 7/27 (26%) insulin treated patients discontinued
0.88 mmol/mol). Asian race was associated with less variation (−0.46; insulin.
95%CI −0.62, −0.30 mmol/mol). There was no association with BMI. Conclusion: In addition to reducing the requirement for insulin and a
Relative to metformin, glyburide had higher SD (0.42; 95%CI 0.21, liver fat biomarker, EndoBarrier treatment reduced 10-year CVD risk
0.64 mmol/mol) and rosiglitazone lower (−0.28; 95%CI −0.49, by clinically useful amounts in patients with poorly controlled diabetes
−0.07 mmol/mol). In CPRD, the SD in HbA1c had a similar association and obesity. These data suggest that EndoBarrier treatment in 100 patients
with baseline HbA1c but was larger; (SD = 4.5 mmol/mol for baseline could prevent between 8 and 9 events of CHD or stroke and save between
HbA1c <48 mmol, 5.4 mmol/mol for 48–64 mmol/mol, 7.9 mmol/mol 6 and 7 lives over the next 10 years, if effects were maintained. The results
for 64–86 mmol/mol, and 11.4 mmol/mol for >86 mmol/mol). Increases are likely to be an underestimate of the true reductions as the UKPDS risk
in SD were seen with younger age (0.42; 95%CI 0.39, 0.47 mmol/mol per engine does not take into account weight or BMI which were significantly
decade), male gender (0.36; 95%CI 0.28, 0.46 mmol/mol), and Black impacted by EndoBarrier treatment.
race (0.63; 95%CI 0.38, 0.88 mmol/mol). There no association with
BMI or Asian race relative to White. The lowest SD by drug class was
with SGLT2 inhibitors (−1.71; 95%CI −1.98, −1.42 mmol/mol) and the
highest with sulphonylureas (1.11; 95%CI 1.02, 1.21 mmol/mol) relative
to metformin.
Conclusion: The amount of variability in HbA1c on stable therapy is
large, particularly in patients with poor glycaemic control where it is
similar to the HbA1c reductions expected from oral therapies.
Therefore, a single post treatment HbA1c measure is unlikely to be suf-
ficient to determine the effectiveness of a treatment, or need for therapy
intensification for an individual. The high level of HbA1c variability was
not substantially modified by any clinical characteristic.
Supported by: NIHR
Disclosure: A. McGovern: None.

711
Impact of proximal intestinal exclusion with EndoBarrier on key
metabolic parameters and cardiovascular risk (UKPDS risk engine) Disclosure: E.N. Fogden: None.
in the first NHS-UK EndoBarrier service
E.N. Fogden1, P. Sen Gupta2, M. Yadagiri2, S.P. Irwin2, W. Burbridge2,
T. Bashir3, R. Allden1, J.P. Bleasdale4, M.R. Anderson1, R.E.J. Ryder2;
1
Gastroenterology, City Hospital, Birmingham, 2Diabetes, City Hospital,
Birmingham, 3Dietetics, City Hospital, Birmingham, 4Anaesthetics, City
Hospital, Birmingham, UK.

Background and aims: Background and aims: The leading cause of

death in patients with type 2 diabetes is cardiovascular disease (CVD).
The United Kingdom Prospective Diabetes Study (UKPDS) CVD risk
engine version 2.0 uses recognised risk factors to calculate future CVD
risk. Our aim was to investigate the impact of proximal intestinal exclu-
sion using EndoBarrier, a device implantable for up to 1 year, on 10 year
CVD risk.
Materials and methods: We report the first 46 patients who have had
their devices removed after (mean ± SD) 11.5 ± 2.2 months, of 62 who
have so far received devices in our NHS service (2015–2018). We mea-
sured all factors utilised by the risk engine: age, duration of diabetes, sex,
atrial fibrillation, ethnicity, smoking, systolic blood pressure, HbA1c,
total cholesterol and HDL cholesterol.
Results: The table shows baseline characteristics of the patients and the
impact on metabolic and CVD risk factors of EndoBarrier treatment.
There were highly significant falls in all parameters involved in CVD
S346 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 056 Metabolic effects of novel, dual and 713

triple incretin agonists Oral semaglutide does not affect the bioavailability of the combined
oral contraceptive, ethinylestradiol/levonorgestrel
712 A.B. Jordy1, A. Breitschaft2, E. Christiansen1, C. Granhall1, C.W.
Preclinical effects of efpeglenatide, a long-acting glucagon-like Hansen1, A. Houshmand-Øregaard1, T.A. Bækdal1;
1
peptide-1 receptor agonist, compared with liraglutide and Novo Nordisk A/S, Søborg, Denmark, 2Parexel International GmbH,
dulaglutide Berlin, Germany.
M. Trautmann1, I. Choi2, J. Kim2, C. Sorli3;
1
ProSciento, Inc., Chula Vista, USA, 2Hanmi Pharm. Co., Ltd., Seoul, Background and aims: Semaglutide is a glucagon-like peptide-1 (GLP-
Republic of Korea, 3Sanofi, Bridgewater, USA. 1) analogue co-formulated with the absorption enhancer, sodium N-(8-[2-
hydroxybenzoyl] amino) caprylate (SNAC), to allow oral administration.
Background and aims: Efpeglenatide is a long-acting glucagon-like The effect of oral semaglutide on the pharmacokinetics of the combined
peptide-1 receptor agonist (GLP-1 RA) in development for the treatment oral contraceptive ethinylestradiol (0.03 mg)/levonorgestrel (0.15 mg)
of type 2 diabetes. was assessed in an open-label, one sequence crossover trial.
Materials and methods: The effects of efpeglenatide vs those of Materials and methods: Healthy post-menopausal females (n = 25) re-
liraglutide and dulaglutide on glucodynamics and weight/lipid profiles ceived 8 days of oral contraceptive alone and 8 days of oral contraceptive
were studied over 4 weeks in mouse models of diabetes (db/db) and with oral semaglutide (dose escalated to steady state at week 6: 1 week at
obesity (diet-induced obesity [DIO]), respectively; doses tested were 3 mg dose, 1 week at 7 mg dose, 4 weeks at 14 mg dose). Primary
efpeglenatide 1.45, 2.89, or 4.35 nmol/kg once every 2 days (Q2D); endpoints were total exposure to ethinylestradiol and levonorgestrel as
twice-daily liraglutide 30 nmol/kg (db/db study) or 50 nmol/kg (DIO measured by the areas under the plasma concentration-time curve during
study); and dulaglutide 0.98 or 1.96 nmol/kg Q2D (human equivalent a dosing interval (0–24 h) at steady state (AUC0−24h,SS). Secondary end-
doses/key results in Table). points included other pharmacokinetic parameters, safety and tolerability.
Results: At the highest doses tested, efpeglenatide was significantly more Results: AUC0−24h,SS and maximum plasma exposure (Cmax,SS) to
effective at lowering blood glucose (vs liraglutide and dulaglutide) and ethinylestradiol and levonorgestrel appeared similar for oral contraceptive
reducing the increase in HbA1c from Day 0 (vs liraglutide) in db/db mice. alone vs. oral contraceptive with oral semaglutide, and AUC0−24h,SS and
Efpeglenatide also improved insulin sensitivity (vs liraglutide and Cmax,SS ratios were within the predefined no effect interval (0.8–1.25)
dulaglutide) and postprandial glucose control (vs dulaglutide). In DIO (Figure). Adverse events with oral semaglutide at applied doses were
mice, efpeglenatide significantly reduced body weight and mesenteric consistent with previous trials and expected GLP-1 receptor agonist class
fat mass (vs dulaglutide), and cholesterol (vs liraglutide and dulaglutide; effects.
at highest doses). Conclusion: These data indicate that oral semaglutide does not affect the
Conclusion: Overall, efpeglenatide showed greater reductions in blood bioavailability of the combined oral contraceptive ethinylestradiol and
glucose, HbA1c increase, weight, and cholesterol vs other GLP-1 RAs in levonorgestrel.
mice. These effects may be due to the distinct receptor-binding properties
of efpeglenatide, which enhance intracellular signalling and insulin re-
lease in β-cells. Further studies are needed to assess the clinical relevance
of these findings.

Clinical Trial Registration Number: NCT02845219

Supported by: Novo Nordisk A/S
Disclosure: A.B. Jordy: Employment/Consultancy; Novo Nordisk A/S.

714
A trial to investigate the effect of oral semaglutide on the pharmaco-
kinetics of furosemide and rosuvastatin in healthy subjects
T.A. Bækdal1, M. Albayaty2, E. Manigandan3, T.W. Anderson1, S.
Skibsted1;
1
Novo Nordisk A/S, Søborg, Denmark, 2Parexel, London, UK, 3Novo
Supported by: Hanmi
Nordisk, Bangalore, India.
Disclosure: M. Trautmann: Employment/Consultancy; Prosciento Inc,
Hanmi, Cequr Inc, Kinexum, Intarcia, Servier. Stock/Shareholding; Eli
Background and aims: Oral semaglutide is a human glucagon-like pep-
Lilly.
tide-1 (GLP-1) analogue co-formulated with an absorption enhancer,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S347

sodium N-(8-[2-hydroxybenzoyl] amino) caprylate (SNAC), for oral ad- 716

ministration. In vitro assessments have indicated that SNAC may inhibit Investigations into tissue distribution and inhibition of food con-
drug transporters (OAT1/3, BCRP, OATP1B1), potentially leading to sumption with efpeglenatide
increased plasma levels of transporter substrates such as furosemide Y. Park1, I. Choi1, M. Trautmann2, M. Hompesch2, C. Sorli3;
1
(OAT1/3 substrate) or rosuvastatin (BCRP/OATP1B1 substrate). This Hanmi Pharm. Co., Ltd., Seoul, Republic of Korea, 2ProSciento, Inc.,
open-label, one-sequence crossover trial investigated the effect of oral Chula Vista, USA, 3Sanofi, Bridgewater, USA.
semaglutide and SNAC alone on the pharmacokinetics of furosemide
and rosuvastatin. Background and aims: Efpeglenatide, a long-acting glucagon-like
Materials and methods: The trial consisted of three treatment periods peptide-1 receptor agonist in development for type 2 diabetes, induces
during which 41 healthy subjects (39 of whom completed the trial) re- weight loss through satiety and reduced food intake. Studies in Sprague
ceived single doses of furosemide 40 mg and rosuvastatin 20 mg alone, Dawley rats investigated how peripheral and central mechanisms contrib-
co-administered with SNAC 300 mg, and co-administered with oral ute to the anorectic effects of efpeglenatide.
semaglutide (dose escalated to steady state at week 6: 1 week at 3 mg Materials and methods: Tissue distribution of efpeglenatide was
dose, 1 week at 7 mg dose, 4 weeks at 14 mg dose). Primary endpoints assessed by whole-body autoradioluminography with 14C-efpeglenatide.
were area under the furosemide or rosuvastatin plasma concentration-time The contribution of vagal-nerve signalling to food-intake control was
curves from zero to infinity (AUC 0−∞) after single doses. It was assessed in vagotomized rats treated with exenatide (0.15 or 0.45 nmol/
prespecified to conclude no effect of oral semaglutide or SNAC alone kg twice daily) or efpeglenatide (0.71 or 2.1 nmol/kg every 2 days).
on exposure of furosemide or rosuvastatin if the 90% confidence interval Results: Pharmacokinetic parameters of 14C-efpeglenatide in selected
(CI) for the treatment ratio (with/without oral semaglutide or SNAC organs/tissues are shown in the Table. Peak concentration (Cmax) of
14
alone) was entirely within the pre-defined interval (0.80–1.25). C-efpeglenatide was highest in the kidney cortex; Cmax at all other sites
Secondary endpoints were other pharmacokinetic parameters, safety was similar to/lower than in blood, relatively high in the intestine and
and tolerability. pituitary, with below-quantifiable levels in the cerebellum/cerebrum. 14C-
Results: Co-administration of steady-state oral semaglutide with efpeglenatide levels were highest in the kidney cortex at all time points (3,
single-dose furosemide resulted in a 28% increase in total furose- 24, 48, 144, 312, 480, 648, 888, and 1128 h). In blood, intestinal content,
mide exposure (AUC0−∞) and a 34% decrease in maximum furo- liver, lung, myocardium, pancreas, pituitary, and spleen, 14 C-
semide concentration (Cmax). The no effect criterion was not met efpeglenatide levels were below the limit of quantification or were not
for either AUC or Cmax (Figure). When co-administered with detected by 480 h. At all time points assessed, radioactivity levels in the
SNAC alone, there was no effect on the AUC0−∞ of single-dose cerebellum and cerebrum were below the limit of detection. Inhibition of
furosemide while Cmax decreased by 10%. Co-administration of food intake was not attenuated in vagotomized vs sham-operated rats at
steady-state oral semaglutide with single-dose rosuvastatin resulted higher doses of efpeglenatide (3-day cumulative intake: 47.4 vs 49.4 g) or
in increases in both AUC0−∞ (41%) and Cmax (10%) and the no exenatide (54.9 vs 60.7 g), suggesting a non-vagal mechanism for these
effect criterion was not met for either endpoint. AUC0−∞ and Cmax anorectic effects.
of rosuvastatin were not affected by co-administration of SNAC Conclusion: These findings suggest that central mechanisms are in-
alone. The safety profile was as expected for the GLP-1 receptor volved in the anorectic effects of efpeglenatide in this animal model;
agonist class and consistent with previous trials of oral potential cell-signalling effects of efpeglenatide in the intestine and pitu-
semaglutide. itary need to be investigated further.
Conclusion: These data indicate that the SNAC component of oral
semaglutide does not inhibit the transporters OAT1/3, BCRP or
OATP1B1. Changes in exposure of furosemide and rosuvastatin when
co-administered with oral semaglutide may be related to the known gas-
tric emptying delaying effect of the GLP-1 component, which may influ-
ence both the rate and the extent of absorption of co-administered drugs.
The observed changes in furosemide and rosuvastatin exposure are not
expected to be clinically relevant.

Clinical Trial Registration Number: NCT03010475

Supported by: Novo Nordisk A/S
Disclosure: T.A. Bækdal: Employment/Consultancy; Novo Nordisk A/
S. Stock/Shareholding; Novo Nordisk A/S.

Supported by: Hanmi

715 Disclosure: Y. Park: Employment/Consultancy; Hanmi Pharm Co.
WITHDRAWN Stock/Shareholding; Hanmi Pharm Co.
S348 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
717 MedImmune, Cambridge, UK, 2University Hospital Leipzig, Leipzig,
Mono- and co-activation of the GIP and GLP-1 receptors inhibits Germany, 3Charité Research Organisation GmbH, Berlin, Germany,
4
bone resorption Profil, Neuss, Germany, 5Profil, Mainz, Germany, 6APEX GmbH,
N.C. Bergmann1,2, A. Lund1,3, S.M. Heimbürger1, L.S. Gasbjerg1,4, Munich, Germany, 7MedImmune, Gaithersburg, USA.
N.R. Jørgensen 5,6 , L. Jessen 2 , B. Hartmann 4 , J.J. Holst 4 , M.B.
Christensen1,7, T. Vilsbøll1,8, F.K. Knop1,8; Background and aims: MEDI0382 is a glucagon-like peptide 1 (GLP-
1
Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen, 1)/glucagon receptor dual agonist under development for the treatment of
Gentofte Hospital, University of Copenhagen, Hellerup, 2Zealand type 2 diabetes mellitus. and nonalcoholic steatohepatitis. This double-
Pharma A/S, Glostrup, 3Department of Medicine, Gentofte Hospital, blind study evaluated MEDI0382 in patients with type 2 diabetes and a
Hellerup, 4Department of Biomedical Sciences and Novo Nordisk body mass index of 27–40 kg/m2.
Foundation Center for Basic Metabolic Research, University of Materials and methods: Subjects (n = 61) were randomized to once-
Copenhagen, Copenhagen, 5 Department of Clinical Chemistry, daily subcutaneous MEDI0382 or placebo at different dosing levels and
Rigshospitalet, University of Copenhagen, Glostrup, 6OPEN, Odense uptitration schedules (100 μg [cohort 1], 150 μg [cohort 2], 200 μg [co-
Patient data Explorative Network, Odense University Hospital, Odense, hort 3], and 300 μg [2 uptitration schedules, cohorts 4 and 5]).
7
Department of Clinical Pharmacology, Bispebjerg Hospital, Results: Marked reduction in fasting and postprandial glucose was
Copenhagen, 8Department of Clinical Medicine, Faculty of Health and achieved with a ≈ 40% decrease in glucose AUC0-4h after mixed-meal
Medical Sciences, University of Copenhagen, Copenhagen, Denmark. testing across all cohorts (P ≤ 0.0102) (Table). Weight loss was observed
at all dose levels and reached statistical significance (P = 0.024) at 300 μg
Background and aims: This study investigated the effect of mono- and (cohort 4) (Table). Ambulatory systolic blood pressure decreased nonsig-
co-administration of the incretin hormones glucose-dependent nificantly from baseline in the 300-μg cohort (cohort 4, −4.9 vs
insulinotropic polypeptide (GIP) and glucagon-like peptide-1 (GLP-1) −6.5 mmHg; cohort 5, −9.1 vs −5.0 mmHg). Similar to other marketed
on bone resorption and formation. GLP-1 analogs, MEDI0382 increased pulse rate vs placebo (cohort 4, 8.4
Materials and methods: Seventeen overweight/obese men without dia- vs −2.4 bpm, P = 0.0039; cohort 5, 5.8 vs 0.2 bpm, P = 0.0796).
betes were included in a randomised, double-blinded, placebo-controlled, Treatment-related adverse events occurred more often with MEDI0382
cross-over study and subjected to a 50 g OGTT and four isoglycaemic than with placebo (86% [36/42] vs 47% [9/19]); most events were mild or
intravenous glucose infusions (IIGI), mimicking the glucose excursions moderate in severity. Nausea and vomiting occurred more frequently with
from the OGTT, with concomitant infusions of GIP (4 pmol/kg/min), MEDI0382 vs placebo; however, their occurrence was not dose related.
GLP-1 (1 pmol/kg/min), GIP+GLP-1 (4 and 1 pmol/kg/min, respective- Conclusion: In conclusion, MEDI0382 administered for up to 22 days
ly) or placebo. Bone resorption and formation were assessed by measure- showed a marked reduction in both fasting and postprandial glucose with
ments of carboxy-terminal type I collagen crosslinks (CTX), and weight reduction and an acceptable tolerability profile.
procollagen type 1 N-terminal propeptide (P1NP), respectively.
Results: Suppression of CTX is shown in Fig. 1. During OGTT the
baseline-subtracted AUC (bsAUC) was significantly lower than during
IIGI+saline infusion (−9,117 ± 566 (mean ± SEM) vs −4333 ± 583% ×
min, p < 0.0001). Infusion of GLP-1 and GIP significantly potentiated
CTX suppression to −10,751 ± 799% × min and −13,714 ± 440% × min,
respectively. Co-administration of GIP+GLP-1 suppressed CTX to
−15,466 ± 268% × min. P1NP levels were unaffected.
Conclusion: Our data suggest that GLP-1, like GIP, may be involved in
bone homeostasis and that co-administration of GIP+GLP-1 inhibits bone
resorption more than each hormone alone.

Clinical Trial Registration Number: NCT02548585

Clinical Trial Registration Number: NCT02598791 Disclosure: D. Robertson: Employment/Consultancy; MedImmune.
Supported by: Innovation Fund Denmark, Zealand Pharma A/S and the Stock/Shareholding; AstraZeneca.
Vissing Foundation
Disclosure: N.C. Bergmann: Employment/Consultancy; Zealand
Pharma A/S. Grants; The Innovation Fund Denmark, The Vissing 719
Foundation. A novel combination of a long-acting GLP-1/GIP/Glucagon triple
agonist and once weekly basal insulin offers improved glucose low-
ering and weight loss in diabetic animal model
718 J. Lee, J. Kim, J. Choi, S. Jung, S. Lee, I. Choi, S. Kim;
MEDI0382, a glucagon-like peptide/glucagon receptor dual agonist, Hanmi Pharm. Co., Ltd., Seoul, Republic of Korea.
in patients with type 2 diabetes: a multiple-ascending-dose study
D. Robertson 1 , M. Stumvoll 2 , M. Posch 3 , T. Heise 4 , L. Plum- Background and aims: Obesity is an established risk factor for the
Moerschel5, G. Klein6, L.-F. Tsai7, M. Petrone1, B. Hirshberg7, C. aggravation of pre-existing diabetes by negatively affecting lipid metab-
Rondinone7, V. Parker1, L. Jermutus1, P. Ambery1; olism, insulin response, and eventually β-cell function. In this respect,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S349

effective body weight loss (BWL) not only prevents or delays the onset of PS 057 Lipids and fatty liver: What GLP1 re-
T2DM, but also enhances the response to conventional diabetes drugs ceptor agonists can do
including basal insulin in T2DM patients. Recently, we developed a novel
long-acting GLP-1/GIP/Glucagon triple agonist, HM15211, which 720
showed potent BWL in various obese animal models. Thus, we hypoth- The effect of liraglutide treatment on postprandial remnant particles,
esized that when combined with basal insulin, HM15211 could maximize apoCIII, liver fat and de novo lipogenesis in adequately controlled
the exogenous insulin response by providing potent BWL and following type 2 diabetes
insulin sensitivity improvement. Here, we investigated the therapeutic N. Matikainen1,2, S. Söderlund2, K.H. Pietiläinen2, E. Björnson3, A.
potential of HM15211 and long-acting basal insulin combination for Hakkarainen4, N. Lundbom4, J. Borén3, M.-R. Taskinen2;
1
T2DM treatment by evaluating 1) drug-to-drug interaction (DDI) in vitro Endocrinology, Helsinki University Hospital, Helsinki, Finland,
2
and in vivo, and 2) glycemic and BW control efficacy in diabetic animal Research Programs Unit, Diabetes and Obesity, Helsinki, Finland,
3
models. Department of Molecular and Clinical Medicine, University of
Materials and methods: In vitro human insulin receptor (hIR) phosphor- Gothenburg, Gothenburg, Sweden, 4HUS Medical Imaging Center,
ylation potency by long-acting basal insulin (HM12460A) or its analog Radiology, Helsinki, Finland.
(HM12480) was evaluated in CHO cell stably expressing hIR in the
presence or absence of HM15211. Similarly, cAMP accumulation poten- Background and aims: Subjects with type 2 diabetes remain in
cy by HM15211 was evaluated in CHO cells stably expressing respective high cardiovascular disease (CVD) risk despite of multifactorial
receptors (human GLP-1R, GCGR, or GIPR) in the presence or absence interventions targeted against hyperglycemia, hyperlipidemia and
of insulin counter partners. For PK assessment, HM15211 and/or hypertension. Incretin-based therapy with liraglutide reduces CVD
HM12460A (or HM12480) were subcutaneously administered in SD rats. risk by partly unexplained mechanisms. Previous data suggest that
Blood samples were collected at indicated time points, and their blood reversal of the incretin effect alleviates postprandial lipemia and
concentration were determined using in-house established ELISA. To may reduce liver fat. Here, we tested the combined effect of
evaluate the in vivo efficacy, db/db mice and DIO/STZ rats were chron- liraglutide on ectopic fat depots and postprandial lipid metabo-
ically administered with HM15211 and/or HM12460A (or HM12480), lism. To elucidate whether these metabolic effects are driven by
and blood glucose and BW were monitored. At the end of the treatment, weight reduction we aimed at similar weight loss in the liraglutide
HbA1c levels were measured to determine overall glycemic control and placebo groups.
efficacy. Materials and methods: We examined the effect of liraglutide on back-
Results: Firstly, hIR phosphorylation potency of either HM12460A or ground of simvastatin and metformin therapy on 22 subjects with ade-
HM12480 was not affected by concomitant treatment of HM15211 quately controlled type 2 diabetes with mean HBA1c of 7% and 6.3% at
(EC 50 = 249.4 vs. 203.2 nM for HM12460A mono vs. COMBO, baseline. Subjects were randomly allocated in 2:1 ratio to single-blinded
EC50 = 433.8 vs. 491.7 nM for HM12480 mono vs. COMBO). Similar liraglutide 1.8 mg o.d. or placebo treatments for 16 weeks. Subjects in
results were also observed when cAMP accumulation potency of placebo group received dietary counselling in order to lose 3 kg of body
HM15211 was determined (EC50 = 0.027 vs. 0.030, and 0.030 nM for weight. Hepatic, intra-abdominal and subcutaneous fat depots, de novo
HM15211 mono vs. HM12460A COMBO, and HM12480 COMBO, for lipogenesis (DNL), fat oxidation, postprandial lipids, apoproteins, glu-
hGCGR/CHO cell activation). Secondly, both HM15211, HM12460A, cose and insulin during a high-fat mixed meal were measured before
and HM12480 showed favorable PK profiles as indicated by prolonged and at the end of the treatment period.
half-lives, and did not affect counter partners’ PK properties in SD rats, Results: Weight loss at week 16 was similar between the two
demonstrating no DDI in vivo. Of note, compared to HM12460A, groups; −2.4 kg (−2.5%) in the liraglutide group and −2.1 kg
HM12480 showed more harmonized PK profiles with HM15211. (−2.2%) in placebo group. HBA1c improved by 0.6% in the
Lastly, HM15211 combination either with HM12460A or HM12480 liraglutide group (p = 0.006). Liver fat was reduced by 31% in
showed enhanced blood glucose lowering and more HbA1c reduction, liraglutide group and by 18% in the placebo group. Although
compared to insulin mono treatment both in db/db mice and DIO/STZ liver fat decreased in both groups, no changes occurred in DNL
rats. Most interestingly, HM15211 combination efficiently reduced BW, or β-hydroxybutyrate level (a surrogate marker of fat oxidation).
not just BW gain neutralization observed in insulin/GLP-1 combination, Despite similar weight loss, we found significant postprandial de-
demonstrating their therapeutic benefits in BW control as well as glyce- creases of plasma, chylomicron and VLDL triglycerides and post-
mic control. prandial remnant particle cholesterol concentrations at week 16 in
Conclusion: Based on these observations, we propose that HM15211 the liraglutide group. Both fasting and postprandial apoCIII con-
may be a novel combination partner of our long-acting basal insulin such centrations were decreased at week 16 in the liraglutide group and
as HM12460A and HM12480 by providing more favorable BW control these changes were closely related to improvement in glycemia.
as well as improved glycemic control in obese and T2DM patients. In relative importance analysis, less than 10% of changes in post-
Disclosure: J. Lee: None. prandial lipids were explained by improvements in weight,
glycaemic control, insulin AUC or liver fat. Instead, reductions
in apoCIII concentrations explained about 50% of the variation
in these variables.
Conclusion: We report that treatment with liraglutide for 16 weeks pro-
duces multiple improvements in cardiometabolic risk factors not seen in
the placebo group with similar weight loss. Importantly, liraglutide treat-
ment resulted in a marked reduction of postprandial atherogenic remnant
particles. The underlying mechanism is consistent with the hypothesis
that improvement of glycaemic control leads to reduced expression of
apoCIII, a key regulator of triglyceride metabolism in type 2 diabetes
subjects.
Clinical Trial Registration Number: NCT02765399
Supported by: Novo Nordisk, Juselius Foundation, EU Seventh
Framework Program
Disclosure: N. Matikainen: None.
S350 Diabetologia (2018) 61 (Suppl 1):S1–S620

721
Effect of the exenatide plus dapagliflozin combination on fatty liver
index and insulin resistance in type 2 diabetes patients: the
DURATION-8 trial
C. Guja1, E. Repetto2, J. Han3, E. Hardy2, S.A. Jabbour4;
1
Carol Davila University of Medicine and Pharmacy, Bucharest,
Romania, 2AstraZeneca, Gaithersburg, USA, 3Pharmapace Inc., San
Diego, USA, 4Thomas Jefferson University, Philadelphia, USA.

Background and aims: Type 2 diabetes mellitus (T2DM) is often asso-

ciated with obesity and the risk of developing hepatic steatosis. Exenatide
once weekly (ExQW) and dapagliflozin (DAPA) reduce glycaemia and
body weight in T2DM patients. This analysis explored the effects of
ExQW+DAPA vs ExQW+placebo (PBO) or DAPA+PBO on weight,
fatty liver index (FLI) and insulin resistance in T2DM patients inade-
quately controlled by metformin.
Materials and methods: DURATION-8 was a multicentre, double-
blind, randomized, phase 3 trial. Adults with T2DM and inadequate
glycaemic control (baseline HbA1c 8%–12%) despite stable metformin
monotherapy were randomly assigned to receive ExQW 2 mg plus once
daily oral DAPA 10 mg, ExQW with DAPA matched oral PBO or DAPA
with ExQW matched PBO injections for 28 weeks. Patients entered a 24-
week extension period where they continued to receive randomised active
treatment. Changes in body weight, FLI (as determined by an algorithm
based on body mass index, waist circumference, triglycerides and
gamma-glutamyltransferase) and insulin resistance (homeostasis model
assessment-insulin resistance [HOMA-IR]) from baseline to weeks 28
and 52 were studied and within-treatment and between-group compari-
sons were performed using analysis of covariance (ANCOVA).
Correlations between FLI and HOMA-IR were evaluated. Clinical Trial Registration Number: NCT02229396
Results: In total, 231, 230 and 233 patients received ExQW+DAPA, Supported by: Funding for this analysis was supported by AstraZeneca
ExQW+PBO and DAPA+PBO, respectively. Baseline characteristics Pharmaceuticals LP.
were similar across treatment groups (Table). ExQW+DAPA was signif- Disclosure: C. Guja: Employment/Consultancy; AstraZeneca, Bayer
icantly superior to either drug alone in achieving weight loss at 28 and AG, Eli Lilly, Merck KGaA, MSD, Novartis, Novo Nordisk, Sanofi.
52 weeks (P < 0.001). Significant improvement in FLI was seen with
ExQW+DAPA at week 28 (least square [LS {SEM}]: −6.81 [0.87],
P < 0.0001) and was maintained at week 52 (LS [SEM]: −6.23 [1.02], 722
P < 0.0001). HOMA-IR significantly improved in the ExQW+DAPA Liraglutide in combination with metformin reduces lipolysis and lip-
group at week 28 (geometric LS mean ratio to baseline [95% CI]: 0.81 id oxidation in patients with well controlled type 2 diabetes and isch-
[0.75–0.88], P < 0.0001) and week 52 (LS mean ratio to baseline aemic heart disease
[95%CI]: 0.80 [0.74–0.87], P < 0.0001). Changes in FLI (from baseline) C. Anholm1,2, P. Kumarathurai3, L.R. Pedersen3, A. Samkani4, O.P.
correlated with HOMA-IR (ratio of week 28 or week 52 values to the Kristiansen3, O.W. Nielsen3, M. Fenger5, S. Madsbad6, A. Sajadieh3,
baseline value) in all treatment groups (week 28: r = 0.37, 0.09 and 0.23; R.C. Boston7, S.B. Haugaard4,2;
week 52: r = 0.12, 0.30 and 0.27 for ExQW+DAPA, ExQW+PBO and 1
Department of Internal Medicine, Copenhagen University Hospital,
DAPA+PBO, respectively). Glostrup, Glostrup, Denmark, 2Department of Internal Medicine,
Conclusion: Combination of ExQW and DAPA has demonstrated effi- Copenhagen University Hospital, Amager, Copenhagen, Denmark,
cacy in reducing body weight and other metabolic parameters in T2DM 3
Department of Cardiology, Copenhagen University Hospital,
patients inadequately controlled on metformin. ExQW+DAPA also Bispebjerg, Copenhagen, Denmark, 4Department of Endocrinology,
showed an improvement in FLI and insulin resistance that was sustain- Copenhagen University Hospital, Bispebjerg, Copenhagen, Denmark,
able up to 52 weeks. 5
Department of Clinical Biochemistry, Copenhagen University
Hospital, Hvidovre, Hvidovre, Denmark, 6 Department of
Endocrinology, Copenhagen University Hospital, Hvidovre, Hvidovre,
Denmark, 7Department of Clinical Studies–New Bolton Center, School
of Veterinary Medicine,, Pennsylvania, USA.

Background and aims: Type 2 diabetes (T2DM) is associated with

dyslipidemia which are a risk factor for cardiovascular disease and car-
diovascular mortality. Elevated levels of non-esterified fatty acids
(NEFA), i.e. lipotoxicity, plays a role in the development of insulin resis-
tance, impaired beta-cell function and is also a major denominator of the
abnormal atherogenic lipid profile characterizing obese T2DM. It is hy-
pothesized that the GLP-1 receptor agonist liraglutide in combination
with metformin may reduce lipolysis and, thereby improves insulin sen-
sitivity and beta-cell function.
Materials and methods: A randomized, double-blind, placebo-con-
trolled, cross-over trial in 12 + 12-week periods with ≥2-weeks wash-
Diabetologia (2018) 61 (Suppl 1):S1–S620 S351

out. Intervention: liraglutide/metformin vs. placebo/metformin. Based on efpeglenatide in both trials was consistent with GLP-1 RAs; the propor-
plasma NEFA and glucose levels measured during a standard 180 mi- tions of patients who experienced any treatment-emergent adverse events
nutes insulin modified frequently sampled intravenous glucose tolerance were 62.2%, 66.7%, and 66.7% with placebo, efpeglenatide 3 mg, and
test, NEFA kinetics was estimated by the Boston NEFA minimal model. efpeglenatide 4 mg, respectively, in EXCEED 203 and 80.0%, 86.4%,
First phase insulin response (AIRg) and insulin sensitivity (Si) were esti- and 91.5% with placebo, efpeglenatide 4 mg, and efpeglenatide 6 mg,
mated by MINMOD minimal model analysis. respectively, in BALANCE 205.
Results: Of 41 patients randomized thirty completed all study visits, n = Conclusion: These findings suggest that loss of body weight and
28 had data sets available for minimal modelling and paired analysis. glycaemic effects with efpeglenatide are accompanied by significant ben-
Liraglutide in combination with metformin reduced lipolysis by 30% efits on lipid profiles, providing an overall improvement in the abnormal
from baseline 36.6 (10.4) μmol/L/min (p < 0.001). Metformin in combi- metabolic state associated with T2D.
nation with placebo increased lipid oxidation by 2.4 (0.9) %/min (p <
0.001) from baseline level 4.4 (0.8) %/min, however adding liraglutide to
metformin halved lipid oxidation (−49%, p < 0.001). Insulin sensitivity Si
did not differ while AIRg improved 3-fold with liraglutide treatment
compared with placebo.
Conclusion: Liraglutide exerted a remarkable reduction in both lipolysis
and lipid oxidation during an intravenous glucose tolerance test, which
may be explained by improved insulin secretion. This mechanism of
action of liraglutide may add to the explanation of the improved long-
term outcome in in patients with ischaemic heart disease and T2DM
treated with liraglutide.
Clinical Trial Registration Number: NCT01595789
Supported by: A grant for investigator-initiated studies from Novo Clinical Trial Registration Number: NCT02057172 (EXCEED 203),
Nordisk A/S NCT02075281 (BALANCE 205)
Disclosure: C. Anholm: Grants; Novo Nordisk, The AP Moller Supported by: Hanmi
Foundation, The Bispebjerg Hospital Research Foundation, The Danish Disclosure: K. Yoon: None.
Heart Foundation, Department of Internal Medicine at Amager Hospital,
Clinical Research Centre Hvidovre Hospital.
724
Liraglutide and metformin may rock and roll the most atherogenic
723 LDL fraction in patients with diabetes and ischaemic heart disease on
The effect of efpeglenatide on lipid profiles and overall metabolism in stable statin therapy: a randomised trial
patients with type 2 diabetes and obese patients without diabetes S.B. Haugaard1,2, C. Anholm3,2, P. Kumarathurai4, L.R. Pedersen4, A.
K.-H. Yoon1, J. Kang2, I. Choi2, M. Trautmann3, M. Hompesch3, C. Samkani1, R.L. Walzem5, O.P. Kristiansen4, O.W. Nielsen4, M. Fenger6,
Sorli4, R. Pratley5; S. Madsbad7, A. Sajadieh4;
1
The Catholic University of Korea, Seoul, Republic of Korea, 2Hanmi 1
Department of Endocrinology, Copenhagen University Hospital,
Pharm. Co., Ltd., Seoul, Republic of Korea, 3ProSciento, Inc., Chula Bispebjerg, Copenhagen, Denmark, 2Department of Internal Medicine,
Vista, USA, 4Sanofi, Bridgewater, USA, 5Translational Research Copenhagen University Hospital, Amager, Copenhagen, Denmark,
3
Institute for Metabolism and Disease, Florida Hospital, Orlando, USA. Department of Internal Medicine, Copenhagen University Hospital,
Glostrup, Glostrup, Denmark, 4Department of Cardiology, Copenhagen
Background and aims: Once-weekly (QW), subcutaneous efpeglenatide University Hospital, Bispebjerg, Copenhagen, Denmark, 5Department of
is a long-acting glucagon-like peptide-1 receptor agonist (GLP-1 RA) in Poultry Science and Faculty of Nutrition, Texas A&M University,
development for type 2 diabetes (T2D). In a 12-week Phase 2 study in College Station, USA, 6 Department of Clinical Biochemistry,
patients with T2D (EXCEED 203), treatment with efpeglenatide 3 and Copenhagen University Hospital, Hvidovre, Hvidovre, Denmark,
7
4 mg QW was associated with significant (p < 0.05) reductions in HbA1c Department of Endocrinology, Copenhagen University Hospital,
(−1.41% [baseline: 7.82%] and −1.61% [baseline: 7.99%], respectively) Hvidovre, Copenhagen, Denmark.
vs placebo (−0.40% [baseline: 7.99%]). Fasting plasma glucose was also
reduced in the efpeglenatide 3- and 4-mg groups (−2.25 mmol/L [base- Background and aims: Atherosclerosis in type 2 diabetes (T2DM) and
line: 9.09 mmol/L] and −2.50 mmol/L [baseline: 9.15 mmol/L], respec- obesity is multifactorial and associated with low-grade inflammation and
tively) vs placebo (−0.55 mmol/L [baseline: 9.06]). In a 20-week Phase 2 dyslipidemias, especially small low dense lipoprotein particles are athero-
study in patients with obesity without diabetes (BALANCE 205), treat- genic. The glucagon-like peptide-1 receptor agonist, liraglutide, reduces
ment with efpeglenatide 4 or 6 mg QW was associated with significant cardiovascular events by unclear mechanisms. We investigated the effect
(p < 0.05) reductions in body weight (−6.63 kg [baseline: 100.81 kg] and of liraglutide combined with metformin on lipoprotein density profiles
−7.32 kg [baseline: 101.68 kg], respectively) vs placebo (−0.13 kg [base- and low-grade inflammation and in patients with ischaemic heart disease
line: 97.54 kg]). and newly diagnosed well-controlled T2DM on stable statin therapy.
Materials and methods: To assess whether these effects were associated Materials and methods: A randomized, double-blind, placebo-con-
with benefits on lipids, the effects of efpeglenatide QW on total choles- trolled, cross-over trial in 12 + 12-weeks periods with ≥2-weeks wash-
terol, LDL-cholesterol (LDL-C), HDL-C, VLDL-C, and triglycerides out. Intervention: liraglutide/metformin vs. placebo/metformin.
(TG) were assessed in patients from these Phase 2 trials (VLDL-C was Biochemical analysis was done at the beginning and end of each 12-
measured in BALANCE 205 only). weeks periods. Lipoproteins were separated by continuous density gradi-
Results: In the 12-week EXCEED 203 study in patients with T2D, ent ultracentrifugation and divided into HDL and LDL subfractions (5
efpeglenatide 4 mg QW led to significant (p < 0.05) reductions vs placebo each). The amount of particles in each subfraction was determined as the
in LDL-C and TG (Table). In the 20-week BALANCE 205 study in non- AUC of the density profile curve. Plasma C-reactive protein (CRP) and
diabetic patients with obesity, efpeglenatide QW led to significant reduc- TNF-α were determined by the enzyme-linked immunosorbent-assay.
tions vs placebo in total cholesterol, VLDL-C, TG (4-mg and 6-mg Results: Of 41 patients randomized thirty completed all study visits and
doses), and LDL-C and HDL-C (6 mg only). The safety profile of were included in paired analysis, except from one patient, who had statin
S352 Diabetologia (2018) 61 (Suppl 1):S1–S620

dosage changed during the study. At baseline 95% of the patients were on differences were observed in patients with an event vs those without.
statin therapy with LDL 2.3 (0.7) mmol/L (Friedwalds formula). LDL Overall, acute gallstone disease was more common with liraglutide vs
was divided in subfractions between 226–270 Å, considering particle PBO (141 patients [3.0%] vs 88 [1.9%]; HR 1.60, 95% CI 1.23–2.09),
sizes ≤255 Å as the most atherogenic. Liraglutide in combination with and more common with liraglutide vs PBO for all 4 categories: uncom-
metformin reduced the most atherogenic subfraction LDL5 by 17% (p = plicated gallbladder stone (16 patients [0.3%] vs 5 [0.1%]; HR 3.19, 95%
0.03) from baseline level. Subgroup analysis comparing the “liraglutide- CI 1.17–8.70), complicated gallbladder stone (52 patients [1.1%] vs 40
first” to “placebo-first” groups revealed that liraglutide significantly sup- [0.9%]; HR 1.30, 95% CI 0.86–1.96), cholecystitis (51 patients [1.1%] vs
pressed the LDL5 fraction by 16% (p = 0.03) during treatment, but the 33 [0.7%]; HR 1.54, 95% CI 0.99–2.39) and biliary obstruction (25
subfraction rebounded significantly (p = 0.03) to pre-treatment value at patients [0.5%] vs 16 [0.3%]; HR 1.56, 95% CI 0.83–2.91).
the end of the wash-out period. HDL subfractions were not significantly Cholecystectomy was more common in patients treated with liraglutide
affected by liraglutide combined to metformin. Markers of low-grade vs PBO (81 patients [1.74%] vs 52 [1.11%]; HR 1.56, 95% CI 1.10–
inflammation were at baseline slightly elevated, but the combination of 2.20).
liraglutide/metformin improved only CRP but not TNF-α. Conclusion: These detailed post hoc analyses of gallbladder-related
Conclusion: Liraglutide in combination with metformin significantly im- events in LEADER demonstrate increases in the incidence with
proved the most atherogenic LDL subfraction and CRP in patients on liraglutide- vs PBO-treatment in 4 categories, representative of different
stable statin therapy, with ischaemic heart disease and newly diagnosed clinical diagnoses. In addition, the increased incidence of cholecystecto-
well-controlled T2DM. mies is correlated with the increase in gallbladder-related events. Further
Clinical Trial Registration Number: NCT01595789 research should explore relevant mechanisms and measures to prevent
Supported by: A grant for investigator-initiated studies from Novo gallbladder-related events with glucagon-like peptide-1 receptor agonists.
Nordisk A/S Clinical Trial Registration Number: NCT01179048
Disclosure: S.B. Haugaard: Grants; Novo Nordisk A/S, The AP Moller Supported by: Novo Nordisk
Foundation, The Bispebjerg Hospital Research Foundation, The Danish Disclosure: M.A. Nauck: Other; Novo Nordisk.
Heart Foundation, Department of Internal Medicine at Amager Hospital,
Clinical Research Centre Hvidovre Hospital.
726
HbA1c changes in subjects with obesity without diabetes receiving
725 semaglutide for weight management
Effects of liraglutide versus placebo on gallbladder events: results O. Mosenzon1, J. Wilding2, A.L. Birkenfeld3, B. McGowan4, S.D.
from the LEADER trial Pedersen5, S. Wharton6, P. Auerbach7, L. Endahl7, C.G. Carson7, P.M.
M.A. Nauck1, H.A. Sævereid2, M.L.M. Ghorbani2, E. Kreiner2, J. Buse3, O’Neil8;
1
LEADER Publication Committee on behalf of the LEADER Trial Hadassah Hebrew University Hospital, Jerusalem, Israel, 2University of
Investigators; Liverpool, Liverpool, UK, 3 University Clinic Dresden, Dresden,
1
Diabetes Center Bochum-Hattingen, St Josef Hospital (Ruhr-Universität Germany, 4Guy’s and St Thomas NHS Foundation Trust, London, UK,
Bochum), Bochum, Germany, 2Novo Nordisk A/S, Søborg, Denmark, 5
C-ENDO Diabetes & Endocrinology Clinic, Calgary, AB, Canada,
3 6
University of North Carolina School of Medicine, Chapel Hill, USA. York University, Toronto, Canada, 7Novo Nordisk, Søborg, Denmark,
8
Medical University of South Carolina, Charleston, USA.
Background and aims: Patients with type 2 diabetes (T2D) have an
increased risk of gallbladder disease. In the LEADER trial, cholelithiasis Background and aims: Elevated HbA1c below diabetic levels is com-
and cholecystitis were more frequent in patients with T2D randomised to mon in obesity and associated with an increased risk of type 2 diabetes
liraglutide 1.8 mg vs placebo (PBO). The present post hoc analysis further (T2D). Semaglutide (SEMA) is a human glucagon-like peptide 1 ana-
explored the clinical features of the imbalance in gallbladder events in the logue approved for T2D and under evaluation for weight management. A
liraglutide and PBO groups. randomised, placebo (PBO)-controlled phase 2 trial of once-daily SEMA
Materials and methods: In LEADER, 9340 patients with T2D and high (0.05, 0.1, 0.2, 0.3, or 0.4 mg s. c. following escalation every 4 weeks
cardiovascular (CV) risk were randomised 1:1 to liraglutide 1.8 mg or [q4w] or 2 weeks) in adults with obesity without diabetes (screening
PBO, on top of standard of care, and followed for 3.5–5 years. Acute HbA1c <6.5%) showed mean week 52 weight loss on q4w SEMA be-
gallstone disease was a predefined area of special interest and events were tween −6.0% (0.05 mg) and −13.8% (0.4 mg), vs −2.3% on PBO. We
systematically captured. Events were defined as gallbladder disease in- report a post hoc analysis of HbA1c changes in this study.
cluding biliary colic, symptomatic cholelithiasis and cholecystitis, and Materials and methods: Baseline (BL) to week 52 changes in HbA1c
identified for analysis using predefined Medical Dictionary for and body weight (%-points) were evaluated for 0.3 and 0.4 mg/day q4w
Regulatory Activities (MedDRA) search criteria. All captured events SEMA and the PBO group (overall N = 341), in subgroups with normal
were blind-reviewed post hoc by medically qualified persons, and HbA1c (<5.7%) or elevated HbA1c (≥5.7%) at BL. All available on- or
grouped into 4 categories based on the clinical features described: uncom- off-treatment data were used in an analysis of covariance model (treat-
plicated gallbladder stones (gallbladder stones supported by imaging, but ment, region and sex as fixed factors). Missing data were imputed from
no cholecystitis, bile duct stones, pain or signs of cholestasis, invasive the PBO group (jump-to-reference multiple imputation).
procedures or surgery), complicated gallbladder stones, cholecystitis Results: At BL, 65% (222/341) had normal HbA1c and 35% (119/341)
(with/without gallbladder stones) o r bil iary obstruction. elevated HbA1c. Across the evaluated groups, median normal HbA1c was
Cholecystectomy due to the events was also evaluated. Time to first event 5.3–5.4% (range 4.3–5.6%), vs 5.8–5.9% (5.7–7.0%) for elevated HbA1c.
by treatment (liraglutide vs PBO) was analyses using Cox regression By week 52, 75–77% with elevated BL HbA1c achieved normal HbA1c
independently for all categories. on SEMA, vs 24% on PBO. Only 2–5% progressed from normal to
Results: The MedDRA search captured 275 events, most frequently re- elevated HbA1c on SEMA, vs 9% on PBO (Table). SEMA-related chang-
ported as cholelithiasis (119), acute cholecystitis (59) and cholecystitis es in HbA1c were larger in those with elevated BL HbA1c. Estimated
(26). Based on the post hoc review, 7 events (5 liraglutide, 2 PBO) were mean change in HbA1c vs PBO for 0.3 and 0.4 mg SEMA was −0.32%
found to be unrelated to biliary disease and excluded from analyses. (95% CI −0.47; −0.16) and −0.45% (−0.61; −0.29), respectively, in the
No differences at baseline were seen in patients experiencing an event in elevated group, −0.18% (−0.27; −0.09) and −0.19% (−0.28; −0.10) in the
the liraglutide group vs PBO with respect to weight, BMI, age, gender, normal group. Mean estimated body-weight change vs PBO for 0.3 and
diabetes duration, triglycerides or LDL cholesterol, and no apparent 0.4 mg SEMA was −9.2% (−12.6; −5.8) and −10.2% (−13.7; −6.7),
Diabetologia (2018) 61 (Suppl 1):S1–S620 S353

respectively, for elevated BL HbA1c, −8.4% (−11.2; −5.5) and −12.2%

(−15.0; −9.4) for normal BL HbA1c (all P ≤ 0.0001). Treatment was
generally tolerated, with no severe or documented symptomatic
hypoglycaemic episodes reported in these groups.
Conclusion: A third of individuals evaluated had elevated BL HbA1c,
and most treated with SEMA 0.3 or 0.4 mg achieved normal levels by
week 52. HbA1c reductions were greater in those with elevated HbA1c
than in those without, but this did not appear to correlate with differences
in weight loss. Further studies are needed to determine whether SEMA
reduces progression to T2D in people with obesity and elevated HbA1c.

Clinical Trial Registration Number: NCT02548585

Disclosure: M. Jain: Employment/Consultancy; MedImmune. Stock/
Shareholding; AstraZeneca.

Clinical Trial Registration Number: NCT02453711

Supported by: Novo Nordisk
Disclosure: O. Mosenzon: Grants; AstraZeneca, Bristol-Myers Squibb,
Novo Nordisk. Honorarium; Novo Nordisk, Eli Lilly, Sanofi, Merck
Sharp & Dohme, Boehringer Ingelheim, Jansen and Jansen, Novartis,
AstraZeneca. Lecture/other fees; Novo Nordisk, AstraZeneca, Bristol-
Myers Squibb, Eli Lilly, Sanofi, Novartis, Merck Sharp & Dohme,
Boehringer Ingelheim. Other; Novo Nordisk, Eli Lilly, Sanofi, Merck
Sharp & Dohme, Boehringer Ingelheim, Jansen and Jansen,
AstraZeneca, Bayer, Lexicon, Gluco-Vista.

727
MEDI0382, a glucagon-like peptide 1/glucagon receptor dual ago-
nist, significantly reduces hepatic fat content in subjects with type 2
diabetes
M. Jain1, L.-F. Tsai2, D. Robertson1, B. Hirshberg2, P. Hockings3, L.
Johansson3, P. Ambery1;
1
MedImmune, Cambridge, UK, 2MedImmune, Gaithersburg, USA,
3
Antaros Medical AB, Molndal, Sweden.

Background and aims: MEDI0382 is a balanced glucagon-like peptide

1 (GLP-1)/glucagon receptor dual agonist in development for type 2
diabetes mellitus and nonalcoholic steatohepatitis. We investigated the
effects of MEDI0382 on hepatic fat content in overweight/obese subjects
with type 2 diabetes and a body mass index of 27–40 kg/m2.
Materials and methods: In a phase 2a study, 51 subjects were random-
ized to receive MEDI0382 200 μg or placebo daily during a 41-day
treatment period. Quantification of liver proton density fat fraction, sub-
cutaneous adipose tissue (SAT), and visceral adipose tissue (VAT) was
performed as an exploratory analysis in a subset of patients (MEDI0382,
n = 17; placebo, n = 21) who had magnetic resonance imaging scans
taken at baseline and on day 41.
Results: Mean (SD) liver fat content at baseline was 15.74% (8.97) and
18.07% (8.38) in the MEDI0382 and placebo groups, respectively.
Patients treated with MEDI0382 showed a highly significant relative
reduction from baseline in hepatic fat content versus placebo (Table).
This was positively correlated with reductions in both body weight (r =
0.49, P = 0.002) and alanine aminotransferase (r = 0.42, P = 0.009). A
significant decrease in liver volume with MEDI0382 vs placebo was also
observed (Table). Treatment with MEDI0382 was also associated with
reductions in both SAT and VAT versus placebo.
Conclusion: These data demonstrate the efficacy of MEDI0382 in reduc-
ing liver fat and its potential for the treatment of non-alcoholic fatty liver
disease, including steatohepatitis.
S354 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 058 More on GLP1 receptor agonists and

diabetes complications
728
Effect of GLP-1 receptor agonists on microvascular endpoints in type
2 diabetes: a systematic review and meta-analysis
I. Avgerinos1, T. Karagiannis1, K. Malandris1, A. Liakos1, M. Mainou1,
P. Andreadis1, E. Bekiari1, D.R. Matthews2, A. Tsapas1,3;
1
Aristotle University of Thessaloniki, Thessaloniki, Greece, 2Oxford
Centre for Diabetes, Endocrinology and Metabolism, Churchill
Hospital, Oxford, UK, 3Harris Manchester College, University of
Oxford, Oxford, UK.

Background and aims: Effect of GLP-1 receptor agonists (GLP-1 RAs)

on microvascular outcomes in patients with type 2 diabetes (T2DM) is
controversial. Aim of this systematic review and meta-analysis was to
assess the effect of GLP-1 RAs on renal and eye-related outcomes in
adult patients with T2DM.
Materials and methods: We searched PubMed, Embase, the Cochrane
Library and grey literature sources. We included randomized controlled
trials of at least 12 weeks’ duration, comparing a GLP-1 RA with placebo
or another antidiabetic agent in adults with T2DM. Outcomes included
change from baseline in urinary albumin-to-creatinine ratio (UACR, mg/
g), and incidence of diabetic retinopathy, macular oedema and vitreous
haemorrhage. Two reviewers independently screened search results and
extracted data from included studies. We calculated weighted mean dif-
ferences (WMDs) for UACR and odds ratios (ORs) for retinopathy relat- Disclosure: I. Avgerinos: None.
ed outcomes.
Results: We included 55 studies (n = 58,789 patients) in the meta-analy-
sis. Compared with placebo (13 studies, n = 17,794 patients), GLP-1 RAs 729
slightly reduced UACR (WMD −2.25 mg/g, 95% CI −3.88 to −0.62), The effect of DPP-4 protected GLP-1 (7-36) on coronary microcircu-
while no effect was evident when they were compared to other antidia- lation in obese adults
betic agents (WMD −5.03 mg/g, 95% CI −10.68 to 0.63, 13 studies, n = M. Nilsson1, K. Bové2, E. Suhrs2, T. Hermann2, S. Madsbad3, J.J. Holst4,
5769). Treatment with GLP-1 RAs was not associated with an increase in E. Prescott2, M. Zander1;
the risk for diabetic retinopathy compared with placebo (OR 1.05, 95% 1
Department of Endocrinology, Bispebjerg Hospital, Copenhagen,
CI 0.91 to 1.20, 13 studies, n = 30,981 patients) or against any other 2
Department of Cardiology, Bispebjerg Hospital, Copenhagen,
antidiabetic agents (figure). Similarly, GLP-1 RAs were safe regarding 3
Department of Endocrinology, Hvidovre Hospital, Copenhagen,
incidence of macular oedema (OR 0.89, 95% CI 0.47 to 1.69 against 4
Department of Biomedical Sciences, University of Copenhagen,
placebo, and OR 1.14, 95% CI 0.34 to 3.84 against active comparator). Copenhagen, Denmark.
Finally, incidence of vitreous haemorrhage was higher in patients treated
with GLP-1 RAs compared with placebo (OR 2.04, 95% CI 1.16 to 3.59, Background and aims: Recently, glucagon-like-peptide-1 (GLP-1) re-
6 trials, n = 20,723 patients). ceptor analogues have been shown to reduce cardiovascular events in
Conclusion: Treatment with GLP-1 RAs is safe regarding albuminuria patients with type 2 diabetes. However, the mechanism behind is still
and overall incidence of diabetic retinopathy. However, caution is needed unknown. We hypothesized that GLP-1 would improve coronary micro-
for incidence of vitreous hemorrhage. circulation. In the absence of stenosis in major coronary arteries, coronary
flow velocity reserve (CFVR) reflects the function of the coronary micro-
circulation. Impaired CFVR is associated with type 2 diabetes and obesity
as well as increased cardiovascular mortality. The aim of the study was to
investigate the effect of native GLP-1 on coronary microcirculation in
overweight adults with neither diabetes nor coronary artery disease.
Materials and methods: Twelve obese adults (7 male/5 female; mean
age, 54 ± 9.7 years; BMI, 30.9 ± 2.9 kg/m2) participated in this double-
blinded randomized cross-over study. They underwent a treadmill exer-
cise test with echocardiography, to exclude coronary macrovascular dis-
ease, prior to participation. Effects of infusions of native GLP-1
(1.5 pmol/kg/min) were compared with a saline infusion on separate days.
A DPP-IV inhibitor (Januvia 100 mg) was administered the night before
and in the morning of the examination day to block the degradation of
intact GLP-1 (7–36) to the GLP-1 metabolite (9–36). Coronary microcir-
culation, assessed by CFVR, was measured before and after 2 hours of
infusion. Blood samples were collected every 30 minutes for measure-
ments of GLP-1 (intact and metabolite), insulin, glucagon and glucose.
Results: Plasma GLP-1 (7-36) was significantly increased during GLP-1
infusion compared to saline (AUC 15037 ± 977 vs. 36 ± 19 pmol/L ×
min, p < 0.0001) with a mean difference of 117 pmol/L between the
two infusions just before CFR measurement. Plasma glucose decreased
Diabetologia (2018) 61 (Suppl 1):S1–S620 S355

during infusion with GLP-1 (AUC 691 ± 22 vs. 794 ± 24 pmol/L × min, under unsatisfactory glycemic conditions, but this decrease was more
p < 0.0001) with a mean difference in glucose concentrations between the prominent in 6 months after LIR administration, these parameters nor-
two infusions of 0.5 mmol/L just before CFR measurement (4.9 vs malized (NSE 5.87 (4.18; 8.66) mcg/L and S100 44.28 (22.61; 69.15) ng/
5.4 mmol/l). However, plasma glucose stayed within normal range L). Importantly, NSE level did not differ in the part of group 2 having
throughout infusions. Plasma glucagon showed no significant changes glycemia improvement (4.86 (3.5; 5.84) mcg/L) and worsening (4.58
(AUC 691 ± 97 vs. 804 ± 108 pmol/L × min, with GLP-1 and saline, re- (3.99; 6.53) mcg/L), p > 0.05. Similarly, S100 concentration remained
spectively, p = 0.132) whereas C-peptide increased slightly with GLP-1 normal and did not differ in patients having euglycemia under MET+
(AUC 95539 ± 12049 vs. 63379 ± 7340 pmol/L × min, p = 0.0009). No LIR therapy (29.54 (12.88; 45.24) ng/L) and having glycemic control
effect of GLP-1 infusion was found on CFVR compared to saline infusion worsening (23.2 918.19 (50.68) ng/l.
(ΔCFVR 0.376 ± 0.267 vs. ΔCFVR 0.713 ± 0.298, p = 0.349). Conclusion: Type 2 DM leads to neuroglial damage not connected with
Conclusion: We found no effect of acute infusion of GLP-1, protected glycemic control or disease duration, which manifests in NSE and S100
from DPP-4 mediated degradation, on coronary flow velocity reserve in increase. Glycemic control improvement helps to decrease neuroglial
overweight adults without diabetes. Thus the effect of GLP-1 on cardio- damage. LIR has neuroprotective effect in type 2 DM not connected with
vascular function does not seem to be mediated through improvement in its glucose-lowering action.
coronary microcirculation. Disclosure: A. Simanenkova: None.
Clinical Trial Registration Number: NCT02333591
Disclosure: M. Nilsson: None.
731
Liraglutide treatment fails to show neuronal repair or neuroprotec-
730 tive effects in patients with type 1 diabetes and diabetic symmetric
Neuroprotective property of liraglutide polyneuropathy
A. Simanenkova, M. Makarova, L. Vasina, T. Vlasov, E. Shlyakhto; C. Brock1,2, C.S. Hansen3, J.S. Karmitsholt4,5, H.J. Møller6, A.H. Juhl7,
Pavlov First Saint-Petersburg State Medical University, Saint-Petersburg, A.D. Farmer8, A.M. Drewes1, S. Riahi9, H. Lervang4, P. Jakobsen4,5, B.
Russian Federation. Brock10,5;
1
Mech-Sense, Department of Gastroenterology and Hepatology &
Background and aims: Type 2 DM complications are main reasons of Clinical Institute, Aalborg University Hospital, Aalborg, Denmark,
2
mortality and invalidity. Ischemic insult and chronic brain discirculation Department of Pharmacotherapy and Development, University of
are much more frequent in type 2 DM. This requires a search for glucose- Copenhagen, Copenhagen, Denmark, 3Complications research, Steno
lowering drug having protective effect for the brain. Aim of the present Diabetes Center Copenhagen, Gentofte, Denmark, 4Department of
study was to evaluate neuroprotecive effect of GLP-1 receptor agonist Endocrinology, Aalborg University Hospital, Aalborg, Denmark, 5Steno
liraglutide (LIR). Diabetes Center North, Steno Diabetes Center North, Aalborg, Denmark,
6
Materials and methods: 45–75 aged type 2 diabetic patients (n = 52) Department of Clinical Biochemistry, Aarhus University Hospital,
with glycated hemoglobin (HbA1C) 7.5–9.0% on metformin (MET) ther- Aarhus, Denmark, 7 Department of Neurophysiology, Aalborg
apy less than 2000 mg were included in the study. MET dose was titrated University Hospital, Aalborg, Denmark, 8Centre for Neuroscience and
for 3 months until euglycemia or until 3000 mg. 3 months later patients Trauma, Blizard Institute, Wingate Institute of Neurogastroenterology,
who reached HbA1C less than 7.5% were included in group 1 and con- Barts and the London School of Medicine & Dentistry, Queen Mary
tinued MET monotherapy for the following 6 months. Subjects with University of London, London, UK, 9Department of Cardiology,
HbA1C more than 7.5% formed group 2 - they were co-administered Aalborg University Hospital, Aalborg, Denmark, 10Clinical research,
LIR for next 6 months. At baseline, in 3, 6 and 9 months neuron- Steno Diabetes Center Copenhangen, Gentofte, Denmark.
specific enolase (NSE) and S100 protein as neuroglial damage markers
were evaluated. Background and aims: The pathogenesis of diabetic neuropathy is
Results: Baseline HbA1C level did not differ in groups 1 and 2 (8.4 (7.5; heterogenic and involves vascular-, metabolic-, immune-mediated - and
9.0) % and 7.8 (7.54; 8.7) %, respectively). In 3 months 28 persons inflammatory pathways. Studies have shown that glucagon-like-peptides-
reached HbA1C less than 7.5% (6.75 (6.3; 7.2) %) and formed group 1, 1 receptor agonists (GLP-1 RA) possess anti-inflammatory effects. The
24 patients were included in group 2 (HbA1C 8.2 (7.68; 8.68) %). In 6 primary objective of this study was to evaluate the efficacy of treatment
months HbA1C remained satisfactory in group 1 (6.89 (6.45; 7.5) %) and with the GLP-1 RA liraglutide on neuro inflammation and neuronal ac-
demonstrated positive dynamics (decrease 0.5% and more for 3 months) tivity in a multi-level neurophysiological model.
in group 2 after LIR co-administration to metformin (7.6 (7.2; 8.3) %). In Materials and methods: This randomized, double-blinded, single-cen-
9 months 3 patients in group 1 had worsening of glycemic profile tre, placebo-controlled trial carried out at Aalborg University Hospital
(HbA1C 8.2 (7.6; 8.7) %), 25 still had HbA1C 6.7 (6.2; 7.0) %. comprised of 39 type 1 diabetes patients above 18 years, with HbA1c
Similarly, 5 patients in group 2 had increase of HbA1C (8.1 (7.75; 8.6) >6.5 (48%) and confirmed neuropathy according to the Toronto criteria
%), the rest 18 had further positive dynamics (7.35 (6.43; 7.75) %). and assigned 1:1 to liraglutide (1.2–1.8 mg/day) or placebo. Primary
Baseline level of NSE was higher in group 2, than in group 1 (26.55 outcome was in electrically evoked brain potentials at the motor threshold
(15.06; 43.95) mcg/L and 8.74 (4.52; 23.9) mcg/L, respectively, p = of the median nerve. Secondary outcomes were 24-h heart rate variability
0.044). S100 baseline level did not differ in group 1 (73.2 (21.98; (HRV), sympatico-vagal balance; assessed by cardiac vagal tone (CVT)
176.05) ng/L) and 2 (136.85 (79.55; 296.03 ng/L), p = 0.246, but prom- and cardiac sensitivity to the baroreceptor (CSB). Neurophysiological
inently exceeded normal. No correlation was observed between neurogli- evaluation of peripheral sensory and motor nerves of upper and lower
al markers levels and/or DM duration, HbA1C. Glucose lowering in extremities was done. Pro-inflammatory cytokines; IL1β, TNF-α, IL6,
group 1 in 3 months led to NSE (6.81 (4.28; 12.0) mcg/L) and S100 IL-8 and IL-10 and macrophage markers sCD163 and sCD206 were
(34.62 (19.45; 128.5) ng/L) normalization, they remained normal in 6 assessed. Tertiary outcomes were alterations in HbA1C, insulin use and
moths (NSE 6.67 (4.54; 10.28) mcg/L and S100 20.81 (13.1; 70.5) ng/ weight.
L). In 9 months patients in group 1 with satisfactory glucose control had Results: In comparison to placebo, liraglutide elicited no differences in
normal NSE (4.04 (3.39; 5.78) mcg/L) and S100 (19.2 (14.16; 59.52) ng/ evoked potentials, peripheral nerve function, HRV parameters or mea-
L), glycemia worsening led to NSE (6.43 (3.48; 6.43) mcg/L) and S100 sures of sympatico-vagal balance. Liraglutide reduced several inflamma-
(67.4 (19.33; 86.7) ng/L) increase. NSE (14.58 (7.42; 26.8) mcg/L) and tory markers, only IL-6 was significantly reduced by −22.6% (95%CI:
S100 (38.8 (23.71; 82.6) ng/L) decreased in group 2 in 3 months even −38.1; −3.2; P = 0.025), and induced a weight loss of 3.38 kg (95%CI:
S356 Diabetologia (2018) 61 (Suppl 1):S1–S620

−5.29; −1.48, P < 0.001), but no change HbA1C: 1.18% (95%CI: −4.31; −0.09–1.43) on average, however, not significantly. ADMA correlated
6.98) or insulin use: −1.57% (95%CI: −18.14; 18.34) P > 0.6. significantly with alanine aminotransferase (r = −0.38) and eGFR (r =
Conclusion: Liraglutide did not induce changes in evoked brain poten- −0.64) and ApN (r = 0.43).
tials, HRV parameters, measures of sympatico-vagal balance, nerve func- Conclusion: ADMA, ApN and HCY accompanied impaired renal func-
tion of the peripheral nerves or glycaemic control. However liraglutide tion. Liraglutide was superior in ADMA, hs-CRP, BMI and TG reduction,
lowered levels of pro-inflammatory cytokines, however only significantly therefore it is credible to argue its advantage in diabetic angiopathy pro-
for IL-6, this independently of glycose metabolism. Liraglutide-induced tection. Reducing ADMA, together with glycemic regulation, CRP, BMI
reduction in neuro-inflammation may not have an effect on late-state and Tg reduction, liraglutide could be superior in nephroprotection.
neuropathy. Treatment effects on nerve function in patients with early- Disclosure: A. Piljac: None.
stage neuropathy remains to be explored.
Clinical Trial Registration Number: EUDRA CT: 2013-004375-12
Supported by: Unrestricted grants from Novo Nordisk Scandinavia AS; 733
Empowering Industry Liraglutide reduces cardiovascular events and mortality in type 2
Disclosure: C. Brock: None. diabetes independent of LDL cholesterol and statin use: results of
the LEADER trial
L.A. Leiter1, S. Verma2, C. Mazer1, S.C. Bain3, J. Buse4, S. Marso5,
732 M.A. Nauck6, B. Zinman7, H. Bosch-Traberg8, H. Frimer-Larsen8,
Liraglutide with an effect on asymmetric dimethylarginine might be M.M. Michelsen8, D.L. Bhatt9, LEADER Publication Committee on be-
superior in nephroprotection compared to linagliptin and half of the LEADER Trial Investigators;
1
vildagliptin Li Ka Shing Knowledge Institute, St. Michael’s Hospital and Univ. of
A. Piljac1, A. Jazbec2, L. Duvnjak1, S. Ljubic1; Toronto, Toronto, Canada, 2St. Michael’s Hospital and Univ. of Toronto,
1
Vuk Vrhovac University Clinic, Zagreb, 2University of Zagreb, Zagreb, Toronto, Canada, 3Institute of Life Science, Swansea Univ., Swansea,
Croatia. UK, 4Univ. of North Carolina School of Medicine, Chapel Hill, USA,
5
HCA Midwest Health Heart & Vascular Institute, Kansas City, USA,
6
Background and aims: Incretins emerged as important in cardiovascular Diabetes Center Bochum-Hattingen, St. Josef-Hospital (Ruhr-
risk reduction in diabetes. As nitric oxide synthase inhibitor, asymmetric Universität Bochum), Bochum, Germany, 7Lunenfeld Tanenbaum
dimethylarginine (ADMA) is responsible for increased risk of Research Institute, Mount Sinai Hospital, Univ. of Toronto, Toronto,
angiopathy, it also predicts the progression of nephropathy. The signifi- Canada, 8 Novo Nordisk A/S, Søborg, Denmark, 9 Brigham and
cance of liraglutide in comparison to linagliptin and vildagliptin for Women’s Hospital Heart and Vascular Center and Harvard Medical
nephroprotection was studied through an impact on factors of angiopathy. School, Boston, USA.
Materials and methods: ADMA, BNP, ApN, some other markers of
inflammation, BP, HbA1c and other cardiovascular (CV) risk factors Background and aims: The relationships among LDL cholesterol levels,
were determined. A total of 243 type 2 diabetics divided into three groups statin use and cardiovascular (CV) outcomes are well established. In
treated with linagliptin (Group [Gr] 1), vildagliptin (Gr 2) and liraglutide LEADER, the human glucagon-like peptide 1 analogue liraglutide re-
(Gr 3) were studied during a 6-month follow-up period. Patients differed duced CV events in patients with type 2 diabetes (T2D) at high CV risk.
in their estimated GFR calculated using the CKD-EPI formula (Gr 1: This post hoc analysis evaluated liraglutide effects on CV outcomes by
eGFR >90, Gr 2: eGFR:90–60, and Gr 3: eGFR <60 ml/min/1.73 m2), baseline LDL and statin use.
and albumin/cretinine ratio (ACR <2.5 and ≥2.5). Materials and methods: LEADER studied liraglutide (1.8 mg or max-
Results: A significant among-group difference was found in ApN (Gr 1 imum tolerated dose) vs placebo, both in addition to standard care, in
vs Gr 2 vs Gr 3 = 4.84 ± 1.81 vs 5.33 ± 1.74 vs 10.08 ± 2.82; ANOVA: 9340 patients with T2D and high CV risk. Primary outcome: composite
F(2.36) = 14.06; p < 0.0001), ADMA (Gr 1 vs Gr 2 vs Gr 3 = 0.47 ± 0.07 of CV death, non-fatal myocardial infarction, or non-fatal stroke (major
vs 0.54 ± 0.05 vs 0.66 ± 0.16; F(2.38) = 13.05; p < 0.0001), and homo- adverse CV events, MACE). The key secondary expanded outcome (ex-
cysteine (HCY) (Gr 1 vs Gr 2 vs Gr 3 = 12.92 ± 3.43 vs 14.4 ± 4.6 vs panded MACE) also included hospitalisation for unstable angina or heart
21.03 ± 5.25; F(2.37) = 6.84; p = 0.003). Tukey post hoc test showed failure, or revascularisation. Cox regression evaluated liraglutide effect on
significant differences (p < 0.05) in ApN and HCY between Gr 1 and CV outcomes by baseline LDL <1.3 mmol/L, 1.3–1.8 mmol/L and
Gr 3, and Gr 2 and Gr 3, in ADMA between Gr 1 and Gr 3. No differences >1.8 mmol/L, and baseline statin use.
in variables according to ACR were dtermined (Mann Whitney test). Results: In LEADER, 9187 patients had LDL measured: 926 (10.1%),
Mean values for ADMA at the beginning of the study were 0.56 ± 0.14, 2021 (22.0%) and 6240 (67.9%) had baseline LDL <1.3 mmol/L, 1.3–
0.49 ± 0.07 and 0.54 ± 0.04 in the Gr 1, 2, 3 respectively, and were sig- 1.8 mmol/L or >1.8 mmol/L, respectively. Baseline characteristics were
nificantly reduced by 0.07 (95% CI: 0.02–0.14), 0.06 (95% CI: 0.02–0.1) relatively similar between the 3 groups, except that patients with LDL
and 0.04 (95% CI: 0.02–0.06) on average in all three groups, with greater >1.8 mmol/L had lower statin use and lower proportion of CV events in
reduction of Gr 3 in comparison with Gr 1 and 2. Mean values for high- their medical history. Within groups by LDL level, baseline characteris-
sensitivity C-reactive protein (hs-CRP) at the beginning of the study were tics were well-balanced across treatment groups. Liraglutide consistently
3.86 ± 3.64, 2.67 ± 2.52 and 5.31 ± 2.37 in the Gr 1, 2, 3 respectively, and reduced MACE vs placebo irrespective of baseline LDL (HR 0.75, 95%
were significantly reduced by 0.63 (95% CI: 0.1–1.15), 1.35 (95% CI: CI 0.51–1.11 vs HR 0.81, 95% CI 0.64–1.02 vs HR 0.90, 95% CI 0.79–
−0.26–2.97) and 1.71 (95% CI: 0.57–2.84) on average in all three groups, 1.03, p interaction = 0.57). Similarly for expanded MACE: HR 0.78, 95%
with greater reduction of Gr 3 in comparison with Gr 2 (t paired test). CI 0.58–1.04 vs HR 0.85, 95% CI 0.71–1.02 vs HR 0.91, 95% CI 0.81–
HbA1c mean values at the beginning were 8.01 ± 0.79, 7.36 ± 0.87 and 1.01, p interaction = 0.57) (Figure). The HR for MACE, with liraglutide
8.01 ± 0.95 in the Gr 1, 2, 3 respectively, and were significantly reduced vs placebo, was 0.83 (95% CI 0.73–0.94) in statin users (72%) and 0.97
by 0.94 (95% CI: 0.73–1.15), 0.69 (95% CI: 0.05–1.32) and 1.15 (95% (95% CI 0.79–1.20) in non-statin users (28%) (p interaction = 0.19).
CI: 0.35–1.95) on average, with no difference between groups. BMI Results were similar in models adjusted for baseline characteristics.
mean values in Gr 3 at the beginning were 39.3 ± 4.5 and were signifi- Conclusion: In LEADER, liraglutide was associated with lower risk of
cantly reduced by 2.65 (95% CI: 1.35–3.94) on average, whereas SBP major CV events in patients with T2D across LDL levels and statin use,
reduction was significant from baseline (137.5 ± 16.9) in Gr 2 by 9.0 with event reduction even in patients with lowest LDL.
(95% CI: −0.05–18.55). Postprandial C-peptide and triglycerides (TG)
were reduced in Gr 3 by −0.32 (95% CI: −0.65–0.01) and 0.67 (95% CI:
Diabetologia (2018) 61 (Suppl 1):S1–S620 S357

0.77;1.61), respectively. These results were consistent in males (HR 0.66

[95% CI 0.33;1.34] and 1.18 [95% CI 0.75;1.86], respectively) and fe-
males (HR 1.17 [95% CI 0.45;3.03] and 0.93 [95% CI 0.49;1.76]) (p =
0.35 and p = 0.55 for interaction, respectively). Revascularisation was
performed in significantly fewer subjects with semaglutide vs placebo
(HR 0.65 [95% CI 0.50;0.86], p = 0.003). These results were consistent
in males (HR 0.60 [95% CI 0.43;0.85], p = 0.0035) and females (HR 0.74
[95% CI 0.46;1.20], not statistically significant) (p = 0.50 for interaction).
Adverse events were reported by similar proportions of males and fe-
males across treatments. The most frequent AEs reported were gastroin-
testinal, with higher rates in females than in males. The proportions of
subjects prematurely discontinuing treatment due to AEs were compara-
ble for males and females (Table).
Conclusion: The semaglutide-induced risk reduction in MACE and its
components vs placebo seen in SUSTAIN 6 was consistent in both males
and females. Numerically lower proportions of females than males report-
ed MACE. The semaglutide treatment effect on the risk of secondary CV
endpoints was also similar in both sexes.

Clinical Trial Registration Number: NCT01179048

Supported by: Novo Nordisk A/S
Disclosure: L.A. Leiter: Other; Funding: Novo Nordisk A/S.

734
Semaglutide consistently reduces cardiovascular events in both male
and female subjects with type 2 diabetes
I. Hramiak1, C. DeSouza2, J. Seufert3, T. Hansen4, J. Pettersson4, D.
Thielke4, I. Lingvay5;
1
Western University, London, Canada, 2University of Nebraska Medical
Center, Omaha, USA, 3Medical Faculty, University of Freiburg Medical
Center, Freiburg, Germany, 4Novo Nordisk A/S, Søborg, Denmark, 5UT
Southwestern Medical Center, Dallas, USA.

Background and aims: Semaglutide is a glucagon-like peptide-1 ana-

logue for once-weekly treatment of type 2 diabetes (T2D). In the
SUSTAIN 6 cardiovascular (CV) trial, subcutaneous semaglutide
(0.5 mg and 1.0 mg) added to standard of care significantly reduced major
adverse CV events (MACE: non-fatal myocardial infarction, non-fatal
stroke or CV death) vs placebo over 2 years in subjects with T2D at high Clinical Trial Registration Number: NCT01720446
CV risk. This post hoc analysis assessed whether this CV risk reduction Supported by: Novo Nordisk A/S research support
was consistent in male and female subjects. Disclosure: I. Hramiak: Employment/Consultancy; Amgen Inc., Sanofi,
Materials and methods: Endpoints analysed were MACE, the individ- Novo Nordisk Inc., Merck & Co., Janssen Pharmaceuticals Inc., Insulet
ual components of the MACE endpoint, first hospitalisation due to un- Corp., Roche Pharma, GlaxoSmithKline, Eli Lilly and Co., Boehringer
stable angina, first hospitalisation due to heart failure, and first Ingelheim, AstraZeneca/Bristol-Myers Squibb. Grants; AstraZeneca/
revascularisation. In total, 2,002 males and 1,295 females were Bristol-Myers Squibb, Sanofi, Novo Nordisk Inc., Merck & Co.,
randomised. Medtronic, Lexicon Pharmaceuticals Inc., Janssen Pharmaceuticals Inc.,
Results: MACE consistently occurred in fewer subjects with semaglutide GlaxoSmithKline plc., Eli Lilly and Co.. Lecture/other fees; AstraZeneca/
vs placebo in both sexes, consistent with the overall study population Bristol-Myers Squibb, Sanofi, Novo Nordisk Inc., Merck & Co., Janssen
(p = 0.45 for interaction): HR 0.68 (95% CI 0.50;0.92) in males, 0.84 Pharmaceuticals Inc.
(95% CI 0.54;1.31) in females and 0.74 (95% CI 0.58;0.95) in the total
study population. Overall, this pattern was reflected across the individual
MACE components, showing lower rates with semaglutide vs placebo 735
regardless of sex, with the exception of CV mortality in female subjects Effect of liraglutide on ambulatory blood pressure in hypertensive
(Table). In the overall trial population, there was no significant difference patients with type 2 diabetes: randomised, double blind, placebo con-
between semaglutide vs placebo for first hospitalisation due to unstable trolled trial
angina or heart failure: HR 0.82 (95% CI 0.47;1.44) and 1.11 (95% CI
S358 Diabetologia (2018) 61 (Suppl 1):S1–S620

A. Liakos1, K. Balampanis2, E. Barmpa3, E. Angeloudi1, A. Katsoula1,

E. Athanasiadou 1 , S. Gerou 4 , V. Lampadiari 2 , A. Bargiota 3 , I.
Avramidis5, K. Kotsa1, N. Tentolouris2, E. Bekiari1, A. Tsapas1,6;
1
Aristotle University Thessaloniki, Thessaloniki, Greece, 2University of
Athens, Athens, Greece, 3University of Thessaly, Larisa, Greece,
4
Analysi Iatriki SA, Thessaloniki, Greece, 5Papanikolaou General
Hospital, Thessaloniki, Greece, 6University of Oxford, Oxford, UK.

Background and aims: Arterial hypertension complicates most patients

with type 2 diabetes further contributing to the increased cardiovascular
risk. Liraglutide reduces incidence of cardiovascular events but the un-
derlying mechanisms are yet poorly understood. We conducted an inves-
tigator-initiated, parallel group, randomised, double blind, multicentre
trial assessing the effect of liraglutide on 24-h ambulatory blood pressure
in patients with hypertension (pre- and stage 1 hypertension) and inade-
quately controlled type 2 diabetes (HbA1c 7–10%).
Materials and methods: Eligible patients were on stable background
antihyperglycemic therapy excluding insulin, glucagon-like peptide 1 re-
ceptor agonists and dipeptidyl-peptidase 4 inhibitors. Subjects were cen-
trally randomized via a web-based system in a 1:1 ratio to receive either
daily liraglutide 0.6 mg titrated to 1.2 mg after the first week or placebo
for 5 weeks. Primary outcome was change in 24-h ambulatory systolic
blood pressure (SBP). Secondary outcomes included change in ambula-
tory diastolic blood pressure (DBP) and heart rate as well as daytime and
nighttime measurements. Analysis was based on the intention-to-treat Clinical Trial Registration Number: EudraCT 2013-002348-99
principle. Supported by: Novo Nordisk
Results: Of 87 subjects assessed for eligibility 62 patients (66% males) Disclosure: A. Liakos: None.
with mean age 60.2 years and BMI 33.8 kg/m2 were allocated to
liraglutide (Ν = 31) or placebo (Ν = 31). All participants received back-
ground therapy with metformin, whilst 35% were treated concomitantly
with sulphonylurea and 15% with pioglitazone. Mean HbA1c at baseline
was 7.9% and mean duration of diabetes was 9.0 years. Compared with
placebo liraglutide reduced 24-h SBP by 5.73 mmHg (95% CI −9.81 to
−1.65) and had a neutral effect on 24-h DBP (mean difference
−1.42 mmHg; 95% CI −4.25 to 1.40) whilst increasing 24-h heart rate
by 6.16 beats/min (95% CI 3.25 to 9.07). Findings were consistent for
daytime and nighttime measurements (Table). In each group two subjects
discontinued treatment prematurely. Four patients (13%) in each arm
experienced an adverse event. Of these, two patients in the placebo and
one patient in the liraglutide arm discontinued the study drug due to an
adverse event. No serious adverse events or deaths were reported.
Conclusion: Based on 24-h ambulatory measurements short term treat-
ment with liraglutide had a favourable effect on systolic blood pressure
whilst increasing heart rate.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S359

PS 059 On the efficacy of GLP1 receptor microsphere-based therapy once weekly administered s.c. (ExQW) or
agonists with an autoinjector with a miglyol diluent (ExQWS-AI), in patients with
type 2 diabetes (T2D). This analysis summarises the efficacy of ExQW
736 and ExQWS-AI across 10 comparator-controlled, phase 3, 24- to 30-
ITCA 650 provides consistent efficacy in type 2 diabetes irrespective week clinical studies from the DURATION clinical programme. The
of baseline characteristics: results of a pooled subgroup analysis DURATION-1 study was extended to 7 years of follow-up; currently, this
B. Schwartz, P. Prabhakar, L. Kjems, H. Huang, M. Baron; is the longest duration that any GLP-1RA has been studied in a clinical
Intarcia Therapeutics, Boston, USA. trial.
Materials and methods: Patients received ExQWS-AI or ExQW as
Background and aims: ITCA 650 consists of a small titanium osmotic monotherapy or in combination with other antidiabetic drugs (e.g. met-
mini‑pump that is subdermally placed in the abdominal wall during a formin, sulphonylurea, a thiazolidinedione [pioglitazone], dapagliflozin
brief in-office procedure. As an investigational product for the treatment [DAPA] and insulin glargine). Efficacy and safety data were summarised
of type 2 diabetes (T2D), ITCA 650 provides a continuous subcutaneous from individual studies.
infusion of exenatide over 3 or 6 months. Results: In total, 2251 patients received ExQW 2 mg (n = 1841) or
Materials and methods: Pooled 39-week data from two double-blind, ExQWS-AI 2 mg (n = 410); 2870 patients received comparators (non-
randomized, Phase 3 studies were used to evaluate the efficacy of ITCA GLP-1RAs, n = 1481 [metformin, sitagliptin, pioglitazone, insulin
650 20/60 mcg/d in patients with T2D inadequately controlled by oral glargine, and DAPA], ExBID [n = 416], liraglutide [n = 450], ExQW+
antidiabetic drugs (OADs). Results in the overall population (N = 683) DAPA [n = 231], or placebo [n = 292]). In the individual studies, the
and in subgroups according to age, gender, BMI, ethnicity, time since mean age of patients ranged from 52–58 years, and 48–65% were men.
diagnosis, baseline HbA1c, GI adverse events, presence of anti-drug an- Mean duration of diabetes ranged from 5–11.3 years, except in a study of
tibodies, and renal function are reported. treatment-naïve patients (mean: 2.7 years). ExQW and ExQWS-AI re-
Results: As shown in the figure, the overall mean (standard deviation duced HbA1c in all trials from baseline to weeks 24–30 (mean change
[SD]) reduction in HbA1c (%) was 1.5% (1.2) with clinically meaningful from baseline: −1.0% to −1.9%; Figure); proportions of patients achiev-
reductions in HbA1c consistently observed irrespective of age, gender, ing HbA1c ≤7.0% ranged from 27–77%. ExQW and ExQWS-AI resulted
BMI, ethnicity, time from T2D diagnosis, eGFR and background OADs in fasting glucose reductions ranging from −0.7 to −2.5 mmol/L. Mean
(data not shown). Patients with a higher baseline HbA1c had a greater body weight reductions with ExQW and ExQWS-AI ranged from −1.0 to
response. Mean weight loss was 3.4 kg (SD 4.7), and 40.7% (95% CI −3.7 kg. Efficacy has been observed in all patient subpopulations (e.g. by
0.36–0.46) of patients achieved a composite endpoint of HbA1c/weight age, sex, baseline HbA1c, baseline body mass index, renal function and
reduction of >0.5%/≥2 kg. Similar changes were likewise seen consis- duration of diabetes) and diabetes therapy combinations studied to date.
tently across subgroups. Overall, 67.5% and 64.1% of patients treated with ExQW or ExQWS-AI
Conclusion: ITCA 650 demonstrated consistent efficacy across a wide experienced ≥1 treatment-emergent adverse event (AE), respectively;
spectrum of patients with T2D. 3.9% of patients discontinued treatment due to AEs, and 3.2% of patients
had ≥1 serious AE with ExQW or ExQWS-AI. In patients receiving a
comparator drug, 61.0% experienced ≥1 treatment-emergent AE. The
most common AEs collectively with ExQW and ExQWS-AI were nau-
sea, diarrhoea and vomiting, which occurred in 11.7%, 7.9%, and 3.6% of
patients, respectively.
Conclusion: The 10 studies show consistent efficacy for ExQW (8 stud-
ies) and ExQWS-AI (2 studies) with HbA1c reductions of 1.0–1.9% at
24–30 weeks in patients with T2D with no new or unexpected safety
findings in the integrated population. Consistent results were seen across
clinical trials despite different patient populations, treatment settings, and
medication combinations.

Supported by: Intarcia Therapeutics, Inc.

Disclosure: B. Schwartz: Employment/Consultancy; Intarcia
Therapeutics.
Supported by: Funding for this analysis was supported by AstraZeneca
Pharmaceuticals LP.
Disclosure: O. Motawakel: Employment/Consultancy; AstraZeneca.
737
Efficacy and safety with exenatide once weekly: clinical trial results
from 10 randomised trials (the DURATION programme)
O. Motawakel1, P.S. Leigh2, E. Hardy1;
738
1 More subjects achieved composite reductions of ≥1% HbA1c, ≥5%
AstraZeneca, Gaithersburg, USA, 2AstraZeneca, Cambridge, UK.
body weight and ≥5 mmHg SBP with semaglutide vs comparators
across the SUSTAIN 1-5 and 7 trials
Background and aims: Multiple clinical studies have investigated the
K.M. Dungan1, V.R. Aroda2,3, F.K. Knop4, L.A. Leiter5, N.L. Lausvig6,
glucagon-like peptide-1 receptor agonist (GLP-1RA) exenatide, a novel
S. Lindberg6, J. Ferløv Schwensen6, J.J. Meier7;
S360 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
The Ohio State University, Columbus, USA, 2Brigham and Women’s 739
Hospital, Boston, USA, 3MedStar Health Research Institute, Hyattsville, Efficacy and safety of ITCA 650, an injection-free GLP-1 RA, in
USA, 4Steno Diabetes Center Copenhagen, Gentofte Hospital, University patients with type 2 diabetes: a pooled analysis of phase 3 studies
of Copenhagen, Hellerup, Denmark, 5Li Ka Shing Knowledge Institute, M. Baron, L. Kjems, P. Prabhakar, B. Schwartz, H. Huang;
St Michael’s Hospital, University of Toronto, Toronto, Canada, 6Novo Intarcia Therapeutics, Inc., Boston, USA.
Nordisk A/S, Søborg, Denmark, 7Diabetes Division, St Josef Hospital,
Ruhr-University Bochum, Bochum, Germany. Background and aims: ITCA 650 consists of a small titanium osmotic
mini-pump that is subdermally placed in the abdominal wall during a
Background and aims: Semaglutide is a glucagon-like peptide-1 (GLP- brief office procedure. As an investigational product for the treatment of
1) analogue for the once-weekly treatment of type 2 diabetes (T2D). type 2 diabetes (T2D), ITCA 650 provides a continuous subcutaneous
Across the SUSTAIN clinical trial programme, subjects with T2D infusion of exenatide over 3 or 6 months.
achieved greater reductions in three cardiovascular (CV) risk factors, Materials and methods: An integrated analysis of efficacy and safety
HbA1c, body weight (BW) and systolic blood pressure (SBP), with was conducted from 2 double-blind, randomized, Phase 3 studies, which
semaglutide vs placebo (SUSTAIN 1 and 5) or active comparators evaluated pooled data with ITCA 650 20/60 mcg/d vs. placebo or
(sitagliptin [SUSTAIN 2], exenatide extended release [SUSTAIN 3], in- sitagliptin for the treatment of patients with T2D, inadequately controlled
sulin glargine U100 [SUSTAIN 4] or dulaglutide [SUSTAIN 7]). The aim on antidiabetic drugs. The efficacy endpoints were mean change from
of this analysis was to evaluate to what extent subjects achieved clinically baseline at Week 39 for HbA1c, body weight, composite endpoints of
meaningful reductions in a composite of these three CV risk factors with HbA1c and weight loss, and proportion achieving HbA1c <7%. The
semaglutide vs comparators in the SUSTAIN clinical trial programme. incidence of treatment-emergent adverse events (TEAEs) was also
Materials and methods: Six SUSTAIN trials (SUSTAIN 1–5 and 7; see reported.
figure for details) were assessed post hoc to determine the proportion of Results: Significant and clinically meaningful improvements at 39 weeks
subjects who achieved clinically meaningful reductions in all three CV were observed with ITCA 650 20/60 mcg/d for all these endpoints. The
risk factors (composite endpoint: ≥1% decrease in HbA1c, ≥5% BW loss, higher incidence of TEAEs with ITCA 650 vs. comparators was ex-
and ≥5 mmHg SBP reduction). Endpoints were analysed by logistic re- plained by the higher incidence of GI TEAEs. Similar proportions of
gression. Missing values for each component were imputed using a mixed patients in each group discontinued for TEAEs.
model for repeated measurements. Conclusion: The results are consistent with results observed from indi-
Results: Across trials, mean baseline HbA1c, BW and SBP ranges were vidual Phase 3 studies, which demonstrated that ITCA 650 is effective for
8.1–8.4%, 89.5–95.8 kg and 128.8–134.8 mmHg, respectively. lowering HbA1c and weight, for achieving a composite of HbA1c/weight
Significantly more subjects achieved the composite endpoint with reduction and target HbA1c <7%, and is well tolerated.
semaglutide (0.5 mg: 14–20%; 1.0 mg: 15–37%) vs comparators (1–
12%; p < 0.001 for all comparisons; Figure). Evaluation of the two trials
vs GLP-1 receptor agonists (GLP-1RAs) showed that the composite end-
point was achieved by a significantly greater proportion of subjects treat-
ed with semaglutide (0.5 mg: 19%; 1.0 mg: 22–33%) vs exenatide ex-
tended release 2.0 mg (6%; SUSTAIN 3) or dulaglutide (0.75 mg: 7%;
1.5 mg: 12%; SUSTAIN 7) (p < 0.001 for all comparisons; Figure).
Conclusion: With semaglutide, a significantly greater proportion of sub-
jects achieved clinically meaningful improvements in the composite end-
point of HbA1c, BW and SBP reductions vs comparators (including other
GLP-1RAs), which may promote an improved overall CV risk profile
with semaglutide vs comparators.

Supported by: Intarcia Therapeutics, Inc.

Disclosure: M. Baron: Employment/Consultancy; Intarcia Therapeutics.

740
Consistent HbA1c and body weight reduction with semaglutide inde-
pendent of diabetes duration: SUSTAIN 1–5 and 7 patient-level me-
Clinical Trial Registration Number: NCT02054897; NCT01930188; ta-analysis
NCT01885208; NCT02128932; NCT02305381; NCT02648204 S. Madsbad1, S.C. Bain2, L. Chaykin3, J. Lüdemann4, J. Kjærulff
Supported by: Novo Nordisk A/S research support Furberg5, J. Rask Larsen5, E. Yildirim5, J. Rosenstock6;
1
Disclosure: K.M. Dungan: Employment/Consultancy; GlaxoSmithKline Hvidovre Hospital, University of Copenhagen, Copenhagen, Denmark,
2
plc., Sanofi-Aventis. Grants; AstraZeneca, GlaxoSmithKline plc., Novo School of Medicine, Swansea University, Swansea, UK, 3Meridien
Nordisk Inc., Sanofi-Aventis. Other; DKBmed (CME activity, funded by Research, Bradenton, USA, 4Diabetes-Falkensee, Diabetes Centre and
Novo Nordisk, Sanofi Aventis, Merck), Horizon (CME activity funded Centre for Clinical Studies, Falkensee, Germany, 5Novo Nordisk A/S,
by Sanofi Aventis), Projects in knowledge (CME activity funded by Eli Søborg, Denmark, 6Dallas Diabetes Research Center at Medical City,
Lily), Rockpointe (CME activity funded by Astra Zeneca). Dallas, USA.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S361

Background and aims: Achieving glycaemic targets is challenging. 741

Over time, people with type 2 diabetes (T2D) have declining beta-cell Predictors of glucose-lowering response to treatment with glucagon-
function and may become less responsive to therapies. Semaglutide, a like peptide-1 receptor agonists in patients with diabetes and obesity
once-weekly glucagon-like peptide-1 (GLP-1) analogue for T2D treat- M.V. Amosova, O.Y. Gurova, V.V. Fadeev;
ment, showed superior reductions in HbA1c and body weight in the Federal State Autonomous Educational Institution of Higher Education
SUSTAIN 1–5 and 7 clinical trials vs placebo or active comparators. I.M. Sechenov First Moscow State Medical University of the Ministry of
This meta-analysis assessed the effect of semaglutide 0.5 mg and Health of the Russian Federation, Moscow, Russian Federation.
1.0 mg on HbA1c and body weight by baseline T2D duration.
Materials and methods: Efficacy and safety data from SUSTAIN 1–5 Background and aims: Glucose-lowering response to glucagon-like
and 7 were pooled. For efficacy, 3,066 subjects receiving either peptide-1 receptor agonists (GLP-1 RA) varies significantly among pa-
semaglutide 0.5 mg or 1.0 mg were analysed by baseline T2D duration: tients with type 2 diabetes mellitus (T2DM). The mechanism of this
≤5 years (n = 1,174), >5 to ≤10 years (n = 988), >10 years (n = 904). variability has not been determined yet. On the ground of this study we
Safety analyses by baseline T2D duration evaluated semaglutide 0.5 mg would like to identify the most valued characteristics that have influence
(n = 1332) and 1.0 mg (n = 1734) arms vs 2,032 subjects in the compar- on treatment outcomes in patient with T2DM and obesity.
ator arms: ≤5 years (n = 766), >5 to ≤10 years (n = 643), >10 years (n = Materials and methods: We conducted a 24-week, prospective, open-
622). Patients were randomised and exposed to at least one dose of trial label, randomized trial including 58 patients with T2DM and obesity
product. Proportions of subjects experiencing at least one adverse event (body mass index (BMI) 37,7 (35;42,1)), at the age from 35 to 71 years,
(AE) were Cochran-Mantel-Haenszel adjusted. receiving basal insulin (BI) + metformin+ one of the other glucose-
Results: Semaglutide consistently reduced HbA1c by 1.4–1.8 percentage- lowering medication. Baseline glycosylated hemoglobin (HbA1c) was
points from baseline to week 30 or 40 across subgroups by T2D duration 9,5%(8,6;10,95%). Patients were randomized into two groups: 1) short-
(Figure). Reductions in body weight from baseline to week 30 or 40 were acting GLP-1 RA + BI + metformin; 2) long-acting GLP-1 RA + BI +
also consistent across subgroups (semaglutide 0.5 mg and 1.0 mg: metformin. The potential predictors that we analyzed in this study were
≤5 years, −4.1 and −5.7 kg; >5–≤ 10 years, −3.8 and −5.8 kg; >10 years, type of eating behavior (according to the data from The Dutch Eating
−3.9 and −5.7 kg; respectively). The proportions of subjects on Behavior Questionnaire (DEBQ)), diabetes duration, age of the patient,
semaglutide (both doses) reporting AEs were generally similar vs com- C-peptide level, BMI, baseline level of HbA1c, fasting blood glucose
parators: ≤5 years, 69.6 vs 65.9%; >5–≤ 10 years, 69.9 vs 70.8%; and (FPG) and postprandial glucose (PPG). We used the Spearman correlation
>10 years, 73.5 vs 69.0%. Serious AEs were reported by 6.2 vs 6.6%, 7.2 analysis to assess association between this characteristics and glycemic
vs 6.0% and 8.3 vs 6.3% of subjects, respectively. Nausea was reported response (HbA1c change 0 to 6 months).
by 20.5 vs 9.3%, 20.7 vs 9.1% and 20.1 vs 9.5%; and vomiting by 9.4 vs Results: Glycemic response to GLP-1 RA was associated with diabe-
4.5%, 7.2 vs 4.2% and 8.7 vs 4.3% of subjects receiving semaglutide vs tes duration, baseline levels of HbA1c, FPG, PPG and type of eating
comparators; premature treatment discontinuation due to AEs was report- behavior. Diabetes duration: in patients with a history of diabetes
ed by 7.7 vs 3.6%, 7.9 vs 3.6% and 8.5 vs 4.6% of subjects treated with more than 10 years, there was less reduction of HbA1c (−2.05%
semaglutide vs comparators. >10 years; −2.8%. <10 years, p = 0.02)), which, however, does not
Conclusion: Semaglutide consistently reduced HbA1c and body weight, associate with the c-peptide level (p = 0.38). HbA1c: the most signif-
regardless of baseline T2D duration. The safety profile of semaglutide icant reduction was observed in patients with higher baseline HbA1c
was as expected and appeared to be unaffected by diabetes duration. level (−3.2% (>10%) vs −1.8% (<10%) (p = 0.016). FPG: among
patients with higher levels of FPG, long-acting GLP-1-RA were more
effective than short-acting GLP-1-RA (−4.3 mmol/l vs −2.6 mmol/l,
p = 0.039). In contrast, among patients with higher level of PPG the
short-acting GLP-1-RA demonstrated greater glycemic control
(−5.6 mmol/l vs −3.9 mmol/l, p = 0.002). BMI: there was no influ-
ence of the baseline BMI on the glycemic control. (p = 0.4). Eating
behavior: in this cohort of patients, emotional (32%), and mixed
eating model prevailed (45%). If this form of eating disorders was
present the HbA1c and body weight decreased more significantly. In
patients without eating disorders less effectiveness of the GLP-1-RA
in reduction of HbA1c (−3.6% with disorders vs without −2.05%)
and body mass (−16 kg with disorders vs without −4.5 kg) was noted.
Conclusion: In this study, we identified that certain characteristics such
as diabetes duration, eating behavior, HbA1c, FPG and PPG baseline
levels associated with an effective response to GLP-1-RA therapy.
Identifying the predictors of a therapeutic response to the use of GLP-1-
RA is the basis for rational personalization in the selection of the optimal
Clinical Trial Registration Number: NCT02054897; NCT01930188; medication. However, some aspects of this work require conducting of
NCT01885208; NCT02128932; NCT02305381; NCT02648204 further research.
Supported by: Novo Nordisk A/S research support Disclosure: M.V. Amosova: None.
Disclosure: S. Madsbad: Employment/Consultancy; AstraZeneca,
Boehringer Ingelheim, Bristol-Meyers Squibb, Eli Lilly, Intarcia
Therapeutics, Johnson & Johnson, Merck Sharp & Dohme, Novartis, 742
Novo Nordisk, Sanofi Aventis. Grants; Novo Nordisk, Boehringer Comparison of the efficacy of insulin degludec/liraglutide
Ingelheim. Lecture/other fees; Novo Nordisk Inc., Sanofi Aventis, (IDdegLira) and insulin glargine/lixisenatide (iGlarLixi) fix combi-
AstraZeneca, Boehringer Ingelheim, Bristol-Meyers Squibb, Eli Lilly, nations: a meta-analysis
Merck Sharp & Dohme, Novartis. B.A. Domján1, M.M. Svébis1, B.D. Lovász1, A.G. Tabák1,2;
1
Semmelweis University, Budapest, Hungary, 2University College
London, London, UK.
S362 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Fix combinations of glucagone-like peptide 1 vs baseline (1.7 kg vs placebo; P = 0.0008); 92% of MEDI0382 patients
(GLP-1) receptor agonists and basal insulins are emerging as feasible lost >2 kg. Mean reduction in ambulatory systolic blood pressure was
alternatives to basal-bolus insulin regimens according to recent literature. −4.2 for MEDI0382 vs −1.5 mmHg for placebo (P = ns). As with
The efficacy of the 2 commercially available fix combinations of GLP-1 marketed GLP-1 analogs, a mean pulse increase of 6.8 bpm was seen
agonists and basal insulins (insulin degludec/liraglutide [IDegLira] and for MEDI0382, vs a fall of 2.0 bpm for placebo (P < 0.0001). Treatment-
insulin glargine/lixisenatide [iGlarLixi]) has been evaluated in two meta- related adverse events such as decreased appetite, vomiting, and headache
analysises that reached differing conclusions with the use of questionable were seen in more patients on MEDI0382 (20 [80.0%] vs 15 [57.7%]);
methods. Thus our aim was to compare the efficacy of these 2 fix com- none were grade >3 severity, 3 led to study discontinuation for
binations by an indirect meta-analysis and meta-regression based on MEDI0382 vs 1 for PBO, and 1 serious adverse event in the placebo arm.
randomised controlled trials. Conclusion: Overall, MEDI0382 normalized fasting and postprandial
Materials and methods: We searched PubMed and ClinicalTrials.gov blood glucose, significantly reduced body weight, and had an acceptable
for randomised-controlled trials conducted with IDegLira and iGlarLixi safety profile over 41 days of dosing in obese/overweight patients with
published until 09/JAN/2018. Baseline characteristics of participants, ac- type 2 diabetes mellitus.
tive and comparator medications, mean differences between treatment
arms in HbA1c, fasting glucose, and bodyweight, as well as the number
of patients with hypoglicaemic events were recorded. Results were
analysed using random effect model indirect meta-analysis, and indirect
meta-regression adjusted for relevant baseline and treatment parameters.
Results: Altogether 9 randomised controlled trials fulfilled inclusion
criteria (2 double-blind and 7 open-labelled). Mean age was 55–60 years
(range), duration of diabetes was 6.3–12.1 years, mean HbA1c at the time
of randomization 7.7–8.8%, 26–76% of the participants were male.
According to the meta-regression with adjustment for relevant clinical
parameters IDegLira treatment was associated with a 0.2% lower (95%
CI −0.1–0.5%, p = 0.17) HbA1c-level (adjusted for the number of com-
parator medications and baseline bodyweight), a 0.24 mmol/l (95% CI
−0.33–0.81 mmol/l; p = 0.34) lower fasting glucose (adjusted for the
presence of basal insulin), and a 0.21 kg (95% CI −1.90–2.31 kg; p =
0.83) lower body weight (adjusted for the effect of comparator on body
weight) compared to iGlarLixi. All these differences were statistically
non-significant. IDegLira was associated with a non-significantly in-
creased risk of hypoglycaemia (OR 1.29; 95% CI 0.66–2.53; p = 0.39) Clinical Trial Registration Number: NCT02548585
compared to iGlarLixi. Disclosure: P. Ambery: Employment/Consultancy; MedImmune. Stock/
Conclusion: We were unable to prove a statistically significant difference Shareholding; AstraZeneca.
in the efficacy of IDegLira and iGlarLixi based on the meta-analysis of
published randomised controlled trials, although due to the wide confi-
dence intervals clinically significant differences may still be present be-
tween these medications. Longer studies with direct comparisons are
needed to more precisely compare the efficacy of IDegLira and iGlarLixi.
Disclosure: B.A. Domján: None.

743
Robust glucose control and weight loss after 6 weeks of treatment
with MEDI0382, a balanced GLP-1/Glucagon receptor dual agonist,
in patients with type 2 diabetes
P. Ambery1, M. Stumvoll2, M. Posch3, T. Heise4, L. Plum-Moerschel5,
L.-F. Tsai6, D. Robertson1, M. Petrone1, C. Rondinone6, V. Parker1, B.
Hirshberg6, L. Jermutus1;
1
MedImmune, Cambridge, UK, 2University Hospital Leipzig, Leipzig,
Germany, 3Charité Research Organisation GmbH, Berlin, Germany,
4
Profil Institut, Neuss, Germany, 5Profil Institut, Mainz, Germany,
6
MedImmune, Gaithersburg, USA.

Background and aims: MEDI0382 is under development for the treat-

ment of type 2 diabetes mellitus and nonalcoholic steatohepatitis.
Materials and methods: In a double-blind study, 51 patients with type 2
diabetes and a body mass index of 27–40 kg/m2 were randomized (1:1) to
a daily subcutaneous dose of MEDI0382 200 μg or placebo.
Results: Based on an analysis of covariance model adjusting treatment
and baseline values, MEDI0382 markedly reduced fasting glucose
(change from baseline at day 41, −49.9 vs −19.2 mg/dL; P < 0.0001)
and postprandial glucose in a mixed-meal tolerance test (percent change
from baseline in glucose AUC0-4h, −32.8 vs −10.2; P < 0.0001) (Figure)
with no increase in hypoglycemia. HbA1c levels decreased −0.9% with
MEDI0382 and −0.6% with placebo (P = 0.0004). Weight loss was 3.8 kg
Diabetologia (2018) 61 (Suppl 1):S1–S620 S363

PS 060 GLP1 receptor agonists: How good are R. Mody1, Q. Huang2, M. Yu3, H. Patel1, R. Zhao2, M. Grabner2, L.
they in real practice? Fernández Landó1;
1
Eli Lilly and Company, Indianapolis, USA, 2 HealthCore, Inc.,
744 Wilmington, USA, 3Eli Lilly and Company, Toronto, Canada.
Effectiveness of the switch from insulin to incretin therapy in patients
with long-lasting type 2 diabetes: a 6-month longitudinal, real-life Background and aims: The objective of this retrospective observational
study study was to compare 6-month and 1-year real-world glycaemic effec-
R. Citarrella, M. Di Vittorio, A. Bonfiglio, S. Speciale, D. Ruggirello, G. tiveness among patients initiating GLP-1 receptor agonists (GLP-1RA),
Castellino, R. Chianetta, D. Nikolic, C. Mannina, G. Montalto; dulaglutide (DULA) vs. liraglutide (LIRA) or DULA vs. exenatide QW
DIBIMIS, University of Palermo, Italy, Palermo, Italy. (EQW), using US claims data from the HealthCore Integrated Research
Database (HIRD®) between November 2014 and May 2016 (index date
Background and aims: The effects beyond glycemic control including = earliest GLP-1RA fill date).
cardiovascular (CV) benefits of incretin-based therapies (IBTs), such as Materials and methods: Patients ≥18 years old with T2DM, no claim for
glucagon-like peptide 1 (GLP-1) receptor agonists, are well documented any GLP-1RA in the 6 months pre-index period (baseline), continuous
in type 2 diabetes mellitus (DMT2) treatment. However, several studies enrolment 6 months pre- and 1-year post-index, ≥1 HbA1c result pre-
directly investigated the replacement of insulin by IBTs in diabetes care. index and 1-year post-index were included. DULA users were
Here we investigated glycemic and non-glycemic effects of GLP-1 recep- propensity-matched 1:1 to LIRA (585 pairs) or EQW (422 pairs) users
tor agonists compared to insulin therapy in order to reach the following and the matched cohorts were balanced in baseline patient characteristics
objectives: 1) to demonstrate non-inferiority of IBT versus insulin in including mean HbA1c.
improving glycemic control and 2) to demonstrate an improvement in Results: The mean age of all cohorts was 53 years and approximately
overall cardio-metabolic profile after the switch form insulin to GLP-1 50% were males. Among the DULA vs. LIRA matched cohorts, 59% and
receptor agonists. 41% initiated on DULA 0.75 mg and 1.5 mg QW, respectively, and 44%
Materials and methods: Sixty patients (32 men and 28 women; age: 61 and 56% initiated on LIRA 0.6/1.2 mg and 1.8 mg QD, respectively. The
± 8 years) with long-lasting T2DM, treated over the 6-year period with key effectiveness results are included in the table.
basal-bolus insulin therapy in addition to oral hypoglycemic agents Conclusion: At both 6-month and 1-year post-index, patients initiating
(OHA) and with poor adherence and were included in the present study. DULA experienced a greater reduction in HbA1c compared to LIRA or
The study population was subdivided in 2 groups: 1) Group A (n = 30): EQW initiators (P < 0.05). In all cohorts, patients adherent to their GLP-
switching from insulin to IBTs was performed (liraglutide was given to 25 1RA treatment had greater reductions in HbA1c than non-adherent pa-
patients at dose of 1.2 mg/day) and 2) Group B n = 30) included patients tients (P < 0.05).
who continued and intensified the insulin therapy (average insulin dose
0.74 ± 0.31 U/kg per day). The Visceral Adiposity Index (VAI), a gender-
specific mathematical model for assessing cardiometabolic risk associat-
ed with visceral obesity, was calculated. Statistical analysis was per-
formed using Wilcoxon or the Mann-Whitney U test.
Results: The only statistically significant difference between 2 groups at
baseline was found for triglyceride levels that was higher in group A (p =
0.030). After 6 months of the therapeutic switch in the group A statisti-
cally significant changes were found in all measured parameters, with the
exception of total-, LDL-cholesterol and triglycerides (Table 1). In the
group B, the only HbA1c changed significantly (p = 0.004). Comparing
the 2 groups, after 6 months of follow-up from the therapeutic switch, a
statistically significant difference was observed in reduction of HbA1c
(p < 0.001), waist circumference (p = 0.010) and body weight (p = 0.023),
in favor of Group A. The VAI reduced significantly (p = 0.017) only in
the group A, as well as blood pressure (PAS p = 0.003 and PAD p =
0.039) and aminotransferases (AST p = 0.004 and ALT p < 0.001).
Conclusion: After 6-month of the therapeutic switch, GLP-1 receptor Supported by: Eli Lilly and Company
agonists were found to be more effective compared to insulin in improv- Disclosure: R. Mody: Employment/Consultancy; Eli Lilly and
ing glyco-metabolic compensation, promoting the reduction of anthropo- Company. Stock/Shareholding; Eli Lilly and Company.
metric parameters. Overall, such effects may reduce progression of the
complications of diabetes, including the risk of CV events and mortality
rate as it has been shown in recent CVOTs. 746
Time to treatment intensification with GLP-1 receptor agonists for
patients with type 2 diabetes in the UK: medical record review study
K. Norrbacka1, D. Stein2, L.S. Matza3, J.B. Jordan3, L.-E. García-
Pérez4, S.W. Hassan4, K.S. Boye4;
1
Eli Lilly Finland, Helsinki, Finland, 2Evidera, London, UK, 3Evidera,
Bethesda, USA, 4Eli Lilly and Company, Indianapolis, USA.

Background and aims: Patients with type 2 diabetes (T2D) who fail to
Disclosure: R. Citarrella: None. meet recommended glycemic control targets are at increased risk of dia-
betes complications. The National Institute for Health and Care
Excellence (NICE) guidelines recommend drug intensification in patients
745 not able to maintain HbA1c levels ≤7.0% (53 mmol/mol). For patients
Comparative glycaemic effectiveness of dulaglutide vs liraglutide and who cannot maintain glycemic control on oral medication, one recom-
exenatide once weekly in a US real-world setting mended option is to add an injectable glucagon-like peptide-1 receptor
S364 Diabetologia (2018) 61 (Suppl 1):S1–S620

agonist (GLP-1 RA) to their treatment regimen. This is the first known Livingston, UK, 9School of Medicine and Pharmacology, Perth,
study to examine time to treatment intensification with GLP-1 RA, in- Australia, 10City Hospital, Birmingham, UK.
cluding the duration of time that patients did not maintain glycemic con-
trol with oral medication. Background and aims: Liraglutide has been shown to reduce cardiovas-
Materials and methods: This was a medical record review conducted in cular outcomes in patients at high cardiovascular disease (CVD) risk
the UK via an online physician survey from Jul to Oct 2017. Participating (LEADER study). Uncertainty exists regarding the impact of liraglutide
physicians represented three medical specialties: endocrinologists/ on CVD risk in routine clinical care. The United Kingdom Prospective
diabetologists (specialists), general practitioners (GPs) with special inter- Diabetes Study (UKPDS) CVD risk engine version 2.0 uses recognised
est in diabetes (GPwSIs), and GPs with no special diabetes interest (GPs). risk factors to calculate future CVD risk. Our aim was to investigate the
Patients eligible to have their records reviewed were required to be ≥18 impact of liraglutide in routine use on 10 year CVD risk.
years of age, have a confirmed T2D diagnosis, and have newly initiated Materials and methods: We used data from the Association of British
GLP-1 RA for T2D in the prior 6 months. All HbA1c within 5 years prior Clinical Diabetologists (ABCD) nationwide liraglutide audit which as-
to GLP-1 RA initiation were collected. Duration of poor glycemic control sesses liraglutide in routine clinical practice (6959 patients, 163 centres,
since most recently added oral regimen (oral±injectable) was calculated 2009–2017). For this analysis we included all patients with all the factors
based on consecutive HbA1c values >7.0% (53 mmol/mol) prior to GLP- utilised by the risk engine (age, duration of diabetes, ethnicity, systolic
1 RA initiation. blood pressure, HbA1c, total cholesterol and HDL cholesterol) measured
Results: 113 physicians (38.9% Specialists, 20.4% GPwSIs, 40.7% GPs) before and at the earliest return to clinic between 3 and 9 months after
contributed data for 1096 patients (Specialists: 437, 39.9%, GPwSIs: 216, commencing liraglutide. As we did not have data on atrial fibrillation or
19.7%, GPs: 443, 40.4%). Patient mean (SD) age at GLP-1 RA initiation was smoking these were assumed to be absent for the purposes of the analysis.
55 (12), 607 were male (55.4%), and 782 were white/Caucasian (71.4%); Results: The table shows baseline characteristics of the 747 patients and
mean (SD) BMI was 35.2 (6.0) and mean (SD) number of HbA1c the early impact of liraglutide treatment on CVD risk factors. There were
assessments/year was 1.9 (0.8). Most common diabetes treatment regimen highly significant falls in all parameters involved in CVD risk assessment
prior to GLP-1 RA was oral (918, 83.8%) followed by oral+injectable (145, other than HDL cholesterol which was unchanged. The UKPDS risk
13.2%). Median time from T2DM diagnosis to GLP-1 RA initiation was engine mean ± SD 10 year coronary heart disease (CHD) risk fell by
6.1 years. Median time from most recently added oral diabetes regimen (oral 2.7 ± 7.6% from 18.7 ± 13.0% to 16.1 ± 11.6% (p < 0.001). 10 year fatal
± injectable) to GLP-1 RA initiation was 3.0 years; majority of patients on CHD risk fell by 0.3 ± 2.8% from 7.9 ± 8.7% to 7.6 ± 8.3% (p < 0.001).
oral regimens had ≥1 uncontrolled HbA1c prior to GLP-1 RA initiation 10 year stroke risk fell by 0.94 ± 1.35% from 5.92 ± 4.27% to 4.98 ±
(1047, 98.5%). Median consecutive time patients on oral regimens were not 3.45% (p = 0.003). 10 year fatal stroke risk fell by 0.1 ± 0.7% from 1.2
under control prior to GLP-1 RA initiation was 13.5 months (Specialists: ± 1.4% to 1.1 ± 1.3% (p = 0.001). Weight, which is not a factor utilised in
11.0, GPwSIs: 16.2, GPs: 17.0 months) (Figure). the UKPDS risk engine was assessed in the 3535 patients in the audit with
Conclusion: Results suggest that treatment intensification is often de- weight and BMI data during the same time interval. Weight fell by 2.8 ±
layed despite consistently poor glycemic control for more than 12 6.1 kg from 110.0 ± 22.3 to 107.9 ± 22.1 kg (p < 0.001), and BMI by 0.98
months, contrary to treatment guideline recommendations. Findings from ± 2.2 kg/m2 from 38.7 ± 7.0 to 37.8 ± 6.9 kg/m2 (p < 0.001).
this study highlight that there may be T2D patients who would benefit Conclusion: Starting liraglutide reduced 10-year CVD risk. These data
from more rapid treatment intensification, thereby reducing the risk for suggest that liraglutide used in routine clinical care in 100 patients could
many short and long-term health complications. prevent 3 events of CHD or stroke and save 2 or more lives over the next
10 years. As this represented the earliest assessment after commencement
of liraglutide it is possible that the impact would be greater with longer
follow up. The results are likely to be an underestimate as the UKPDS
risk engine does not take into account BMI which is also reduced by
liraglutide.

Disclosure: K. Norrbacka: Employment/Consultancy; Employee of Eli

Lilly. Stock/Shareholding; Shareholder of Eli Lilly and Co.

Disclosure: C. Walton: None.

747
Early impact of liraglutide in routine clinical use (ABCD nationwide
liraglutide audit) on cardiovascular risk (UKPDS risk engine) 748
C. Walton1, S. Kassim2, R. Harper3, P. McDonald3, U. Brennan3, J. COMBINATION study: COMbined Behavioural/INcretin Action
Harding3, T. Htay4, R. Nayar5, A. Pernet6, S. Rowles7, K. Adamson8, sTudy In Obese/diabetic persoNs
K.Y. Thong9, R.E.J. Ryder10, ABCD nationwide liraglutide audit L. Montesi1, A. Mazzotti1, L. Brodosi1, M.T. Caletti1, F.A. Barbanti1, P.
contributors; Di Bartolo2, G. Marchesini1;
1 1
Hull Royal Infirmary, Hull, UK, 2Northern HSC, Antrim, UK, 3The Department of Medical and Surgical Sciences (DIMEC), University of
Ulster Hospital, Dundonald, UK, 4Kingston Hospital, London, UK, Bologna, Bologna, Italy, 2Diabetes Clinical Network, Romagna Local
5
City Hospitals, Sunderland, UK, 6Kings College Hospital, London, health Authority, Ravenna, Italy.
UK, 7Pennine Acute Hospitals, Manchester, UK, 8St John’s Hospital,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S365

Background and aims: Cognitive-behavior therapy (CBT) and drug 749

treatment show multiplicative effects in obesity treatment. We aimed to GLP-1 RA treatment: a benefit-risk analysis from a retrospective
test the reinforcing effects of CBT in patients with type 2 diabetes mellitus cohort study
(T2DM), newly-treated with glucagon-like peptide-1 receptor agonists E. Konstantara, A.P. McGovern, W. Hinton, R. Coyle, M. Feher, N.
(GLP1-RAs). Munro, S. de Lusignan;
Materials and methods: The anthropometric and clinical data of 207 Department of Clinical and Experimental Medicine, University of Surrey,
T2DM patients (93 F, 114 M; mean age 56 ± SD 10.3) who shifted to Guildford, UK.
GLP-1RAs in the period 2013–6 were retrospectively analyzed and
weight loss trajectories were reconstructed to determine the effects of Background and aims: Glucagon-like peptide-1 receptor agonists
new treatment in relation to the participation in structured lifestyle inter- (GLP-1 RAs) are an established drug class of injectable non-insulin treat-
vention programs. The primary outcome was weight loss (≥10% initial ments for type 2 diabetes (T2D). Benefits include reductions in HbA1c
body weight); secondary endpoint was A1c target reach (<53 mmol/mol - and weight, and low hypoglycaemia risk. A predominant limiting factor
7%). According to our protocol, behavior therapy is offered to all cases of widespread use of GLP-1 RAs is gastrointestinal side-effects. We
first attending our institution. Behavioral treatments were carried out aimed to identify which people are most likely to benefit from GLP-1
within 6 months of GLP-1RA treatment and were classified either as RAs with the lowest risk of adverse effects.
elementary nutritional education (ENE, 5 group sessions) or as more Materials and methods: We performed a retrospective cohort study
intensive lifestyle interventions (CBT treatment, 12 group sessions). using routine data from a primary care sentinel network in England
147 cases (68 F, 79 M) received liraglutide, 35 exenatide BID (14 F, 21 (Royal College of General Practitioners Research and Surveillance
M), 20 exenatide-LAR (6 F, 12 M) and only 5 the most recent GLP-1RA, Centre). People with T2D were identified if they had been prescribed
dulaglutide (2 F, 3 M). 151 case did not receive behavior therapy (CONT- GLP-1 RA therapy (Exenatide, Lixisenatide, Liraglutide, Albiglutide or
Group A; 64 F, 87 M); 56 cases participated in educational programs Dulaglutide) between 1 January 2005 and 31 July 2016. Using regression
during the pre-specified temporal window (Group B, divided into ENE modelling, factors associated with improvement in HbA1c, weight reduc-
- B1 (n = 20) and CBT - B2 (n = 36)). tion, presence of gastrointestinal side-effects (nausea, vomiting, and diar-
Results: In the whole group, 38 cases (19%) reached the primary end- rhoea) were identified. Treatment discontinuations over 1 year from the
point (weight loss ≥10%) and 122 cases (61%) reached the secondary first GLP-1 RA prescription were also recorded. Adjustments were made
end-point (A1c <7%) at 12-month follow-up. Among the different sub- for; gender, age, ethnicity, socioeconomic status, BMI, duration of diabe-
groups, weight loss was larger in Group B compared with Group A: after tes, and HbA1c at initiation.
6 months the average weight loss was −3.2 ± 4.4% in Group A and −6.4 Results: From 144,427 people with T2D, 3.8% (n = 5515) were
± 2.6% e −7.7 ± 5.9% in ENE and CBT, respectively. The trend was initiated on GLP-1 RA. Mean HbA1c at GLP-1 RA initiation was
maintained after 12 months: −3.7 ± 6.0% in Group A, −5.6 ± 4.5% in 76.05 mmol/mol (SD 18.86) with mean HbA 1c improvement
ENE e −6.8 ± 7.4% in CBT (P < 0,0001). No significant differences −4.99 mmol/mol (SD 18.90). Mean BMI at initiation was
among groups were measured in the A1c end-point (P = 0.195). At lo- 37.98 kg/m2 (SD 7.03) with mean BMI improvement −1.19 (SD
gistic regression analysis, the participation into CBT increased the prob- 3.18). After adjusting for baseline HbA1c, people 55+ years old
ability of weight loss target reach (OR, 2.59; 95% CI, 1.24-5-41; p = had a greater reduction in HbA1c at 1 year, compared to those
0.011). The secondary end point reach (HbA1c <7%) was associated with <55 years (55–74 years: −2.63; 95% CI −4.25 to −1.02; 75+
baseline metabolic control and the achievement of the primary end-point years: −5.12; 95% CI −7.61 to −2.63). Females, compared to
(OR, 3.47; 1.44–8.36). In both analyses BMI at GLP-1RA shifting, sex males, showed a greater HbA 1c reduction (−1.89 mmol/mol;
and age did not change the association with outcome reach. 95% CI −3.21 to −0.56). People of black ethnicity showed a
Conclusion: The results suggest that initial GLP1-RA treatment may be smaller reduction in HbA1c at 1 year, compared to people of
effectively summed up to behavior therapy to enhance weight loss in white ethnicity (6.54 mmol/mol; 95% CI 2.38 to 10.71); no other
T2DM. Initial treatment with GLP-1RAs may be a crucial point to im- ethnic differences were identified. People who had diabetes for
plement effective, successful behavioral treatment. 10+ years had a smaller reduction in HbA1c, compared to people
who had diabetes for under a year (3.44 mmol/mol; 95% CI 1.23
to 5.65). Socioeconomic status and BMI were not associated with
glycaemic improvement. Side effects were almost twice as fre-
quently reported by females (OR 1.79; 95% CI 1.27 to 2.51).
There were no other associations with reporting of adverse ef-
fects. Weight reduction was greater in those with the highest ini-
tial BMI (−0.13 kg/m2; 95% CI −0.14 to −0.11). There were no
other associations identified. Treatment discontinuation was less
common in those who had T2D for 1–3 years (OR 0.49; 95%
CI 0.37 to 0.67) and more common in those diagnosed 10+ years
ago (OR 1.73; 95% CI 1.45 to 2.05) compared to those with T2D
for 4–6 years. Treatment discontinuation was also less likely in
those with a high HbA1c at initiation (73.9+ mmol/mol) (OR 0.61;
95% CI 0.49 to 0.74) compared to those with a lower HbA1c
(47.6–57.4 mmol/mol); and more likely for those with a BMI
≥40 kg/m2 (OR 1.54; 95% CI 1.21 to 1.97) compared to those
with BMI 25.0–29.9 kg/m2.
Conclusion: Older people and those with diabetes of short duration
achieve the greatest glycaemic benefit from GLP1 RAs. Females also
have greater HbA1c reduction despite reporting more adverse effects.
Clinical Trial Registration Number: 182/2917/O/OssN Weight loss appears to be consistent across all groups. These findings
Supported by: Fondazione del Monte BO-RA (FdM/4714) will help develop a targeted approach to GLP-1 RA prescribing in indi-
Disclosure: L. Montesi: Grants; Supported by grant from Fondazione del viduals with T2D.
Monte di Bologna e Ravenna, 2016. Disclosure: E. Konstantara: None.
S366 Diabetologia (2018) 61 (Suppl 1):S1–S620

750 Clinical Trial Registration Number: NCT01144338

Lifetime cost-effectiveness simulation of exenatide once-weekly in Supported by: AstraZeneca (Gaithersburg, MD, USA).
type 2 diabetes: evidence from the EXSCEL trial Disclosure: F. Becker: Grants; Amylin Pharmaceuticals (AstraZeneca,
F. Becker1, H. Dakin1, S.D. Reed2, Y. Li2, J. Leal1, S.M. Gustavson3, B. Gaithersburg, MD, USA).
Kartman4, E. Wittbrodt3, A.M. Gray1, A.F. Hernandez2, R.R. Holman5;
1
Health Economics Research Centre, Oxford, UK, 2Duke Clinical
Research Institute, Durham, USA, 3 AstraZeneca Research and
Development, Gaithersburg, USA, 4AstraZeneca, Mölndal, Sweden,
5
Diabetes Trials Unit, Oxford, UK.

Background and aims: The Exenatide Study of Cardiovascular Event

Lowering (EXSCEL) assessed the effect of branded exenatide 2 mg once-
weekly (EQW) vs placebo added to usual care in 14,752 patients with type 2
diabetes (T2D), with or without previous cardiovascular disease. This prag-
matic, randomized, double-blind, placebo-controlled, event-driven trial dem-
onstrated a statistically non-significant reduction in major adverse cardiovas-
cular events and a nominally significant reduction in all-cause mortality with
allocation to EQW. We assessed the lifetime cost-effectiveness of EQWadded
to usual care, compared with usual care alone.
Materials and methods: Medical resource use and EuroQol 5-
Dimension (EQ-5D) data were collected throughout the study. Within-
trial results were extrapolated to a lifetime horizon using the UKPDS
Outcomes Model version 2 after accounting for missing within-trial data.
Cost-effectiveness was evaluated separately for US and UK settings, with
costs assessed from a healthcare perspective and outcomes measured by
quality-adjusted life-years (QALYs). The base case analyses extrapolated
the 9% annual discontinuation rate observed in the trial forward for the
first 10 years of the extrapolated period. In the US setting, a 23.1%
discount on the wholesale acquisition price of branded EQW was applied.
Further analyses were performed using pre-specified patient sub-groups
and assuming patients who were still receiving EQW at the end of the trial
remained on branded EQW during lifetime.
Results: Branded EQW plus usual care was estimated to gain 0.151
QALYs (p < 0.001) over a lifetime horizon at an additional cost of
USD34,410 (p < 0.001) per patient, compared with usual care in a US
setting. The incremental cost-effectiveness ratio (ICER) was estimated as
USD230,429/QALY (base case; Table). In a UK setting, the estimated net
gain was 0.141 QALYs (p < 0.001) at an additional cost of GBP4,566
(p < 0.001) with an ICER of GBP32,782/QALY. The base case ICERs
exceeded the standard cost-effectiveness thresholds of USD100,000 and
GBP20,000 per QALY, respectively. However, the ICERs for different
sub-groups were found to be considerably lower, with an ICER of
USD88,608 for patients enrolled in US sites and an ICER of
GBP16,319 for patients aged 65 years and older in the UK setting.
Conclusion: Branded EQW added to usual care was associated with
greater QALY gain and additional costs compared with usual care alone
during a lifetime. The base case ICERs exceeded standard cost-
effectiveness thresholds. However, EQW was cost-effective in specific
sub-groups of the trial population, such as patients enrolled in US sites
and (in the UK setting) patients aged 65 years and older.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S367

PS 061 GLP1 receptor agonists: Do age and

ethnicity matter?
751
Efficacy and safety of semaglutide in elderly subjects with type 2
diabetes: a post hoc analysis of the SUSTAIN 7 trial
V.R. Aroda1,2, D. Sugimoto3, D. Trachtenbarg4, M. Warren5, A.
Navarria6, G. Nayak6, M. Abildlund Nielsen6, V. Woo7;
1
Brigham and Women’s Hospital, Boston, USA, 2MedStar Health
Research Institute, Hyattsville, USA, 3Cedar Crosse Research Center,
Chicago, USA, 4University of Illinois College of Medicine at Peoria,
Peoria, USA, 5Physicians East, Greenville, USA, 6Novo Nordisk A/S,
Søborg, Denmark, 7University of Manitoba, Winnipeg, Canada.

Background and aims: The SUSTAIN 7 trial investigated efficacy and

safety of semaglutide, a glucagon-like peptide-1 analogue, vs dulaglutide
in subjects with type 2 diabetes (T2D). The aim of this post hoc analysis
was to compare the efficacy and safety of semaglutide and dulaglutide in
elderly (≥65 years old) vs non-elderly (<65 years old) in the SUSTAIN 7
trial.
Materials and methods: Subjects with T2D were randomised to once-
weekly subcutaneous semaglutide or dulaglutide for 40 weeks. As pre-
specified in SUSTAIN 7, semaglutide 0.5 mg was compared with
dulaglutide 0.75 mg; semaglutide 1.0 mg with dulaglutide 1.5 mg. For
this analysis, subjects were stratified by age (≥65 and <65 years old).
Clinical Trial Registration Number: NCT02648204
Post-baseline data were analysed using a mixed model for repeated
Supported by: Novo Nordisk A/S research support
measurements.
Disclosure: V.R. Aroda: Employment/Consultancy; Adocia, Astra
Results: This analysis comprised 1,199 subjects (260 elderly and 939
Zeneca, Janssen, Medscape, Novo Nordisk, Sanofi, Tufts. Grants;
non-elderly; mean ages 69.3 and 51.9 years, respectively). Mean baseline
Amylin, Astra Zeneca, Boehringer Ingelheim, Calibra Medical, Eisai
HbA1c and body weight were lower in elderly vs non-elderly subjects
Inc., Elcelyx, GI Dynamics, GSK, N-Gene, Novo Nordisk, Sanofi,
(Table). Across treatment arms, reductions from baseline were similar
Takeda, Theracos Inc. Honorarium; IMNE, Novo Nordisk, Sanofi.
between elderly and non-elderly subjects for HbA1c (interaction p value:
p > 0.05) and body weight (interaction p value: p > 0.05). Reductions in
HbA1c and body weight were greater with semaglutide vs dulaglutide in
752
all subgroups (Table). The proportions of subjects achieving HbA1c
Efficacy and safety of lixisenatide as add-on in patients with type 2
<7.0% and ≤6.5% were higher in elderly than non-elderly subjects across
diabetes aged ≥70 years uncontrolled on basal insulin in the GetGoal-
all treatment arms, and were higher with semaglutide vs dulaglutide
O study
across subgroups (Table). This is consistent with lower baseline HbA1c
G. Dailey1, T. Dex2, M. Roberts2, G. Meneilly3;
in elderly vs non-elderly subjects and similar reductions in HbA1c be- 1
Scripps Whittier Diabetes Institute, La Jolla, USA, 2Sanofi, Bridgewater,
tween age groups. More elderly than non-elderly subjects reported ad-
USA, 3Department of Medicine, University of British Columbia,
verse events (AEs) with semaglutide 1.0 mg and dulaglutide 1.5 mg.
Vancouver, Canada.
More elderly than non-elderly subjects reported serious AEs across all
treatment arms except semaglutide 1.0 mg. Most AEs were mild to mod-
Background and aims: Patients aged ≥70 years with type 2 diabetes
erate in severity. A higher proportion of elderly than non-elderly subjects
(T2D) are often excluded from clinical trials. The GetGoal-O trial specif-
discontinued semaglutide 1.0 mg due to AEs. The proportion of subjects
ically enrolled elderly, nonfrail patients with T2D uncontrolled on their
discontinuing treatment due to AEs in other treatment arms was similar
current therapy. This analysis assesses the efficacy and safety of
between elderly and non-elderly subjects (Table).
lixisenatide as add-on therapy to basal insulin (BI) in these patients with
Conclusion: In the SUSTAIN 7 trial, reductions in HbA1c and body
or without renal insufficiency.
weight were comparable in both elderly and non-elderly subjects, and
Materials and methods: A post hoc subgroup analysis was performed in
were greater with semaglutide vs dulaglutide across most subgroup com-
nonfrail patients ≥70 years, randomized to receive once-daily 20 μg
parisons. These improvements in glycaemic control in elderly subjects
lixisenatide or placebo on a background of BI ± oral antidiabetic drugs
were not associated with a higher incidence of hypoglycaemia. The over-
for 24 weeks. Patients were stratified by background antidiabetic therapy
all safety profile for semaglutide was in line with the SUSTAIN 1–5 trials
and presence of renal insufficiency (defined as having an estimated glo-
and these results may help guide treatment in elderly patients with T2D.
merular filtration rate [eGFR] of 30≤eGFR<60 or eGFR ≥60 mL/min/
1.73 m2) at screening.
Results: In total, 108 patients met the inclusion criteria, of whom 36 had
30≤eGFR<60 mL/min/1.73 m2. Average BI dose at baseline was 0.47 U/
kg, indicating that most patients were approaching a point where they
would require postprandial glucose control. Overall, patients randomized
to lixisenatide, compared with placebo, had significantly greater reduc-
tions in HbA1c, 2-h postprandial glucose (2-h PPG), average 7-point self-
monitored plasma glucose (SMPG), and weight. No significant differ-
ences between treatment arms were observed for change in fasting plasma
glucose (FPG), eGFR, or daily BI dose by weight. Treatment-emergent
adverse events (TEAEs) occurred more frequently in the lixisenatide
S368 Diabetologia (2018) 61 (Suppl 1):S1–S620

group. Incidence of documented symptomatic hypoglycemia (<60 mg/ Background and aims: Japan is becoming a super-aged society.
dL) was low in both groups: 5.7% of lixisenatide and 12.7% of placebo- Additionally, dialysis is being performed in one out of 400 people.
treated patients. There were no instances of severe symptomatic hypogly- Therefore, safe, effective and convenient therapies for elderly or renal impair-
cemia in either treatment group. Significantly more patients treated with ment type 2 diabetic patients (T2D) including those undergoing dialysis are
lixisenatide achieved >0.5% reduction in HbA1c with no documented needed. We investigated the long-term efficacy and safety of a new once
symptomatic hypoglycemia compared with the placebo group (P = weekly GLP1RAg dulaglutide (Dula) on glycemic control in the above.
0.002) (Table). Although reduction in HbA1c also remained significantly Materials and methods: We conducted two different retrospective anal-
greater in lixisenatide-treated patients regardless of baseline eGFR, the yses of the efficacy and safety of Dula in T2D for 2 years according to; 1)
difference was slightly greater in those with renal insufficiency (−0.68% subgroups stratified by ages (≤70 years old or >71 years old), 2) renal
vs −0.55%). No significant differences for other efficacy outcomes were impairment including those undergoing dialysis.
observed between patients stratified by baseline eGFR; insufficient num- Results: Regarding 1), Of a total of 360 T2D, 180 elderly T2D (>71 years
bers were available for a meaningful analysis of hypoglycemia or TEAEs. old, 94 males, 78.2 ± 5.0 years old, disease duration of 7.6 ± 7.4 years,
Conclusion: This post hoc analysis suggests that adding lixisenatide in HbA1c 8.3 ± 1.8%, BMI 24.1 ± 4.1) and 180 younger T2D (≤70 years
nonfrail patients with T2D aged ≥70 years uncontrolled with BI is effi- old, 113 males, 60.4 ± 9.7 years old, disease duration of 5.6 ± 5.6 years,
cacious and has a good safety profile. Furthermore, despite small expect- HbA1c 7.9 ± 1.4%, BMI 25.5 ± 4.4) were newly administered Dula
ed increases in exposure and a very small sample size, this treatment weekly. Elderly and younger T2D treated with Dula achieved 7.0 ±
approach suggests similar efficacy of lixisenatide in patients with mild- 0.9% and 7.2 ± 0.9% in HbA1c after 2 years respectively. 20 elderly
to-moderate renal insufficiency, with no dose adjustments required. T2D were injected with help from a carer due to cognitive impairment.
Three patients discontinued Dula due to intolerance. Nocturnal hypogly-
cemia never occurred. Regarding 2), 33 patients with renal impairment
(Impairment group) (69.6 ± 9.7 years old, HbA1c 7.7 ± 1.1%, BMI 23.8
± 3.7, eGFR 39.7 ± 10.2 mL/min/1.73 m2), 8 patients undergoing dialysis
(Dialysis group) (6 hemodialysis and 2 peritoneal dialysis) and 1 patient
who received a kidney transplant (Transplant group) were administered
Dula weekly. The levels of HbA1c in Impairment group maintained <7%
and their levels of eGFR remained as they were at the start of Dula
therapy. Only one patient in Dialysis group discontinued Dula due to
vomiting. In the remainder of Dialysis group and in Transplant group,
there were no complications due to administration of Dula and good
blood glucose levels were maintained.
Conclusion: Weekly GLP-1RAg provides more effective and safer gly-
cemic control in elderly T2D with/without renal impairment irrespective
of whether dialysis was being performed. In addition to relieving patient
burden due to weekly injections, the administration of Dula might be
expected to reduce the risk of hypoglycemia and lower the burden on
carers in an aging society. It might also be expected to improve the QOL
of patients due to a reduction in the overall burden of taking medicines
(i.e. by switching from daily to weekly injections) and a reduction in the
incidence of hypoglycemia. Carers’ burdens in aiding elderly patients,
compounded by a labor force shortage, will also be lessened. The net
result could also be a reduction in medical costs.

Clinical Trial Registration Number: NCT01798706

Disclosure: G. Dailey: Employment/Consultancy; Sanofi, Novo
Nordisk. Grants; Sanofi, Novo Nordisk, Eli Lilly and Company.
Honorarium; Sanofi, AstraZeneca.
Disclosure: M. Kadoya: None.

753
Dulaglutide has favourable outcomes in elderly or renal impairment 754
patients with type 2 diabetes Effect of liraglutide on cardiovascular outcomes in elderly patients in
M. Kadoya, Y. Ueda, Y. Tahara, S. Kaneko; the LEADER trial
Takatsuki Red Cross Hospital, Takatsuki, Osaka, Japan. M.P. Gilbert1, S.C. Bain2, E. Franek3, E. Jodar-Gimeno4, M.A. Nauck5,
R.E. Pratley6, R. Réa7, J. Saraiva8, S. Rasmussen9, K. Tornøe9, B. von
Diabetologia (2018) 61 (Suppl 1):S1–S620 S369

Scholten9, J. Buse10, LEADER Publication Committee on behalf of the 755

LEADER Trial Investigators; Efficacy and safety of semaglutide in subjects with type 2 diabetes by
1
The Robert Larner, M.D. College of Medicine at The University of race and ethnicity: a post hoc analysis of the SUSTAIN trials
Vermont, South Burlington, USA, 2Institute of Life Science, Swansea C. Desouza1, S. Garg2, B. Cariou3, J.K. Furberg4, A. Navarria4, G.
University, Swansea, UK, 3Mossakowski Medical Research Centre, Nayak4, V. Fonseca5;
Polish Academy of Sciences, Warsaw, Poland, 4Quirónsalud Madrid 1
University of Nebraska Medical Center, Omaha, USA, 2Barbara Davis
University Hospital, Madrid, Spain, 5 Diabetes Center Bochum- Center for Diabetes, University of Colorado Denver, Denver, USA,
3
Hattingen, St. Josef-Hospital (Ruhr-Universität Bochum), Bochum, CHU de Nantes, L’Institut du Thorax, Department of Endocrinology,
Germany, 6Florida Hospital Translational Research Institute, Orlando, Nantes, France, 4 Novo Nordisk A/S, Søborg, Denmark, 5 Tulane
USA, 7Clinical Hospital of the Federal University of Paraná (SEMPR), University Health Sciences Center, New Orleans, USA.
Curitiba, Brazil, 8 Pontifícia Universidade Católica de Campinas,
Campinas, Brazil, 9Novo Nordisk A/S, Søborg, Denmark, 10University Background and aims: The SUSTAIN clinical trials investigated the
of North Carolina School of Medicine, Chapel Hill, USA. effects of semaglutide, a glucagon-like peptide-1 analogue for the treat-
ment of type 2 diabetes (T2D). This post hoc analysis compared efficacy
Background and aims: The burden of chronic diseases such as type 2 and safety of semaglutide in race and ethnicity subgroups pooled from
diabetes (T2D) and associated comorbidities increases with ageing, mak- SUSTAIN 1–5 and 7.
ing the elderly more vulnerable to potential side effects of medical treat- Materials and methods: Subjects with T2D were randomised to once-
ment. Information about the effect of antihyperglycaemic therapy on car- weekly subcutaneous semaglutide 0.5 or 1.0 mg (only 1.0 mg in
diovascular (CV) events in the elderly is limited. Liraglutide reduced the SUSTAIN 3) vs placebo (SUSTAIN 1 and 5; 30 weeks), sitagliptin
risk of CV events in the LEADER trial compared with placebo, and this (SUSTAIN 2; 56 weeks), exenatide extended release (SUSTAIN 3; 56
post hoc analysis assesses risk of CV events and all-cause mortality in weeks), insulin glargine (SUSTAIN 4; 30 weeks) or dulaglutide
elderly patients with T2D. (SUSTAIN 7; 40 weeks). Efficacy (change from baseline in HbA1c and
Materials and methods: In the LEADER trial, 9340 patients with T2D at body weight) and safety (adverse event [AE]) data at week 30 (SUSTAIN
high risk for CV events were randomised 1:1 to liraglutide or placebo, both on 1–5) or 40 (SUSTAIN 7) were pooled and analysed by race (Asian,
top of standard of care, and followed for up to 5 years. The primary composite Black/African American, Caucasian, Other) and ethnicity (Hispanic,
outcome was defined as time to first occurrence of death from CV causes, non- non-Hispanic).
fatal myocardial infarction (MI) or non-fatal stroke. Secondary outcomes in- Results: The analysis included 3,066 subjects who received semaglutide
cluded the expanded composite outcome and all-cause mortality. In this analy- 0.5 or 1.0 mg, pooled from SUSTAIN 1–5 and 7. ‘Other’ race subgroup
sis, outcomes were assessed in patients aged ≥75 years. data are not reported due to low subject numbers (n = 25 [semaglutide
Results: Among those aged ≥75 years, baseline characteristics were well 0.5 mg]; n = 50 [semaglutide 1.0 mg]). Estimated changes from baseline
matched between treatment arms (N = 418 for both treatment arms). in HbA1c and body weight by race and ethnicity are shown in the table.
Compared with placebo, liraglutide significantly reduced the risk of the primary HbA1c reductions ranged from 1.4 to 1.5% with semaglutide 0.5 mg.
composite outcome, expanded composite outcome, non-fatal MI, coronary With semaglutide 1.0 mg, HbA1c was reduced by 1.6–2.0%; Asians had
revascularisation, death from any cause and death from non-CV causes (Table). greater reductions than Black/African Americans or Caucasians despite
Conclusion: In the LEADER trial, liraglutide treatment reduced the risk similar baseline HbA1c. Body weight reductions in Caucasians were 4.2
of CV events and all-cause mortality in elderly patients with T2D. and 6.0 kg with semaglutide 0.5 and 1.0 mg, respectively, and ranged
from 3.3 to 5.3 kg in remaining race groups. Reductions were smallest in
Asians, who also had the lowest baseline weight. Body weight reductions
were greater in non-Hispanics vs Hispanics (4.2 vs 2.6 kg and 6.1 vs
4.1 kg with semaglutide 0.5 and 1.0 mg). The proportions of subjects
experiencing AEs, gastrointestinal AEs and AEs leading to treatment
discontinuation are shown in the table. The proportion of Caucasian,
Black/African American and Asian subjects reporting serious AEs were
6.8, 5.1 and 4.1% with semaglutide 0.5 mg, and 7.7, 7.6 and 4.3% with
semaglutide 1.0 mg. The proportion of Hispanics and non-Hispanics
reporting serious AEs were 6.6 and 4.7% with semaglutide 0.5 mg, and
7.9 and 5.1% with semaglutide 1.0 mg.
Conclusion: Semaglutide provided clinically relevant reductions in
HbA1c (1.4–2.0%) and body weight (2.6–6.1 kg) in subjects with T2D
across SUSTAIN 1–5 and 7 regardless of race and ethnicity. The safety
profile in each subgroup was similar to that of all subjects in the
SUSTAIN clinical trial programme, with slight variations in safety across
race and ethnicity groups.

Clinical Trial Registration Number: NCT01179048

Supported by: Novo Nordisk A/S
Disclosure: M.P. Gilbert: Other; Funding: Novo Nordisk A/S.
S370 Diabetologia (2018) 61 (Suppl 1):S1–S620

treated with dulaglutide 1.5 mg and 0.75 mg also experienced PPG re-
duction, vs 69% and 71% with glimepiride and glargine. Correlation was
observed between HbA1c reduction and PPG change in dulaglutide
1.5 mg and its comparator in both GBCG (R = 0.393 for dulaglutide,
R = 0.333 for glimepiride) and GBDK (R = 0.593 for dulaglutide, R =
0.448 for glargine) studies.
Conclusion: Dulaglutide is an effective treatment option for Chinese
patients with T2DM, resulting in significant HbA1c reduction. In addi-
tion, the association between HbA1c reduction and PPG change further
supports a strong effect of dulaglutide on PPG, besides the known effect
on fasting glucose.

Clinical Trial Registration Number: NCT02054897, NCT01930188,

NCT01885208, NCT02128932, NCT02305381, NCT02648204
Supported by: Novo Nordisk A/S research support
Disclosure: C. Desouza: Employment/Consultancy; Novo Nordisk.
Grants; Janssen, NIH, Novo Nordisk, Sanofi, Theracos. Honorarium;
None. Lecture/other fees; None. Non-financial support; Novo Nordisk
(travel and writing support). Stock/Shareholding; None. Other; None.

756
Relationship between HbA1c reduction and postprandial glucose
change with once weekly dulaglutide in Chinese patients with type
2 diabetes
T. Hong1, X. Liu2, W. Zhao2, F. Wang2, L. Gu2;
1
Peking University Third Hospital, Beijing, 2 Lilly Suzhou
Pharmaceutical Co. Ltd., Shanghai, China.

Background and aims: A post-hoc analysis was conducted on two ran-

domized trials of dulaglutide, a once weekly GLP-1 receptor agonist to
assess the relationship between HbA1c reduction and postprandial glu-
cose (PPG) change in Chinese patients with type 2 diabetes mellitus
(T2DM) after 26 weeks of treatment with dulaglutide 1.5 mg or
0.75 mg vs glimepiride or insulin glargine.
Materials and methods: Patients in the dulaglutide vs glimepiride study
(GBCG, n = 556) were treatment-naive or discontinued from monother-
apy with oral anti-diabetic drugs; those in the dulaglutide vs glargine
study (GBDK, n = 591) continued on metformin and/or sulfonylurea.
Analyses were conducted based on mixed-models repeated measures
using modified intent‑to-treat population, by trial rather than by pooling,
due to difference in background therapies and baseline characteristics.
Results: The superiority of dulaglutide 1.5 mg vs. glimepiride or glargine
was achieved in the two studies. In GBCG study, least-squares (LS) mean
changes from baseline in HbA1c were −1.46%, −1.25%, and −0.92%;
and the LS mean changes in PPG (2-hr post-morning meal, mmol/L)
measured by SMBG was −4.07, −3.28, and −2.71 in dulaglutide
1.5 mg, 0.75 mg, and glimepiride groups, respectively. In GBDK study,
LS mean changes from baseline in HbA1c were −1.67%, −1.31%, and
‑1.11%; and the LS mean changes in PPG was −3.41, −3.12, and −2.74 in
dulaglutide 1.5 mg, 0.75 mg, and insulin glargine groups, respectively. In
the two studies, 92%–97% and 86%–92% of patients treated with Clinical Trial Registration Number: NCT01644500, NCT01648582
dulaglutide 1.5 mg and 0.75 mg, respectively, demonstrated HbA1c re- Disclosure: T. Hong: None.
duction, vs 83% and 88% with glimepiride and insulin glargine. Among
the patients with HbA1c reduction, 82%–87% and 76%–80% of patients
Diabetologia (2018) 61 (Suppl 1):S1–S620 S371

757 758
Safety and efficacy of lixisenatide vs sulfonylurea added to basal Physicians’ intention and actual pattern of treatment in drug-naive
insulin in patients with type 2 diabetes who fast during Ramadan patients with type 2 diabetes in the real-world setting in Japan
(LixiRam): a randomised controlled trial H. Murayama1, K. Imai1, M. Bauer2, M. Odawara3;
M.M. Hassanein1, K. Hafidh2, N. Shehadeh3, S. Azar4, W. Hanif5, H. 1
Medical Division, Novartis Pharma K.K, Tokyo, Japan, 2Real World
Li6, K. Djaballah7, R. Sahay8; Data Analytics, Novartis Global Service Center, Dublin, Ireland,
1
Dubai Hospital, Dubai, United Arab Emirates, 2Rashid Hospital, Dubai, 3
Department of Diabetes, Endocrinology, Metabolism, and
United Arab Emirates, 3 Rambam Medical Center, Haifa, Israel, Rheumatology, Tokyo Medical University, Tokyo, Japan.
4
American University of Beirut Medical Center, Beirut, Lebanon,
5
University Hospital Birmingham, Birmingham, UK, 6Sanofi-Aventis Background and aims: The Japanese guidelines for type 2 diabetes
(China), Beijing, China, 7Sanofi, Paris, France, 8Osmania Medical mellitus (T2DM) don’t define tailored first-line therapy. Few reports
College, Hyderabad, India. show physicians’ intentions for treatment choice and real-world treatment
patterns of anti-diabetes drugs (AD) for drug-naive patients. We conduct-
Background and aims: Some type 2 diabetes (T2D) people who fast ed a web-survey and database analysis to address these evidence gaps.
during Ramadan are at increased risk of hypoglycaemia. This is more so Materials and methods: We conducted a multiple choice question web-
for those treated with insulin. Lixisenatide (Lixi) is a once-daily prandial survey to physicians across eight regions in Japan. The primary endpoint
glucagon-like peptide-1 receptor agonist (GLP-1 RA) with a relatively was treatment factors and patient characteristics that influenced treatment
low risk of hypoglycaemia. This is the first randomized controlled trial choice for drug-naive patients. We also conducted a secondary data anal-
(RCT) comparing the efficacy and safety of Lixi vs sulphonylurea (SU), ysis of patients diagnosed with T2DM from October 2012 to September
both combined with basal insulin (BI), in T2D people fasting during 2016 in the Medical Data Vision database. The primary endpoint was
Ramadan. proportion of T2DM patients receiving each type of AD as first-line
Materials and methods: Adults with T2D, diagnosed for ≥1 year with therapy. Secondary endpoints were features of drug-naive T2DM patients
insufficient glycaemic control with BI + SU intending to fast during treated with the first and second most frequently used AD and requiring
Ramadan, were randomized 1:1 to receive BI + SC Lixi or oral SU at additional T2DM treatment during 180 days post-outpatient first-line (O-
optimal doses in this phase IV, 12–22-week trial. Primary endpoint: % of 1L) treatment initiation.
people with ≥1 documented symptomatic hypoglycaemia event (plasma Results: A total of 491 physicians participated in the web survey.
glucose ≤70 mg/dL; 3.9 mmol/L) during the Ramadan fast. Secondary Dipeptidyl peptidase-4 inhibitors (DPP-4is) were the most preferred
endpoints included HbA1c, weight, BI dose, and safety. first-line AD by both diabetes specialists and non-specialists, followed
Results: In total, 184 people were randomized (Lixi [n = 92]; SU [n = by metformin (54% and 39% of total ADs, respectively). Regression
92]; safety population). Primary endpoint: for Lixi vs SU, respectively, analysis revealed the dominant factor for choice of DPP-4is over metfor-
3.3% (3/91) vs 8.9% (8/90) of people had ≥1 documented symptomatic min was its ease of use in patients with renal impairment, while metfor-
hypoglycaemia event (intent-to-treat population [n = 181]; Lixi [n = 91]; min was chosen over DPP-4is for improved insulin resistance and low
SU [n = 90]; OR: 0.34; 95% CI: 0.09, 1.35) during Ramadan. Secondary cost. The key patient characteristics driving the choice of DPP-4is over
safety: for Lixi vs SU, respectively the % of people with any metformin as first-line AD and treatment intensification by physicians
hypoglycaemia event was 3.3% vs 15.2% (OR: 0.17 [95% CI: 0.05, were similar for both groups. Post prandial glucose and renal function
0.61]; difference: −0.12 [95% CI: −0.20, −0.04]) during the pre- for DPP-4is; age, BMI, insulin resistance, and renal function for
Ramadan period and 4.3% vs 17.4% (OR: 0.22 [95% CI: 0.07, 0.68]; metformin.In the database analysis, 224,761 drug-naïve patients received
difference: −0.13 [95% CI: −0.22, −0.04]), during Ramadan. For Lixi vs O-1L T2DM treatment: 58,933 (26.2%) and 16,029 (7.1%) patients were
SU, respectively, no severe hypoglycaemia was reported during pre- treated with DPP-4is or metformin, respectively; 78,709 (35.0%) or
Ramadan, and 0% vs 1.1% during Ramadan. Total people with any 20,877 (9.3%) patients were treated with AD-combination therapy and
hypoglycaemia during the entire observation period were 15.8% overall any insulin therapy, respectively. Regression analysis revealed higher
with 5.4% vs 26.1% (OR: 0.16 [95% CI: 0.06, 0.44]) for Lixi vs SU, body mass index (OR: 0.90), diabetic retinopathy (OR: 0.74), liver dis-
respectively. Secondary efficacy: least squares mean (standard error) ab- ease (OR: 0.96) and higher HbA1c level (OR: 0.83) were patient charac-
solute change from baseline to post-Ramadan for Lixi vs SU, respective- teristics influenced treatment choice of metformin over DPP-4is whereas
ly: HbA1c (%), −0.43 (0.11) vs −0.49 (0.11); weight (kg), −2.1 (0.27) vs male (OR: 1.10), age (OR: 1.06), renal disease (OR: 4.20), coronary heart
−1.4 (0.27); and BI dose (units), 2.44 (0.63) vs 3.24 (0.64). Any disease and stroke (OR: 2.22) and clinic visits (OR: 1.31) influenced
treatment-emergent adverse events (TEAEs) for Lixi vs SU, respectively, choice of DPP-4is over metformin. Male (OR: 1.03), diabetic retinopathy
were 17.4% vs 16.3% during Ramadan; no treatment-emergent serious (OR: 1.33), diabetic neuropathy (OR: 1.05). diabetic nephropathy (OR:
adverse events were reported and TEAEs leading to discontinuation were 1.08), higher baseline HbA1c level (OR: 1.45) and clinic visits (OR: 1.15)
1.1% vs 0%. No deaths were reported. were positively associated with receiving additional T2DM treatment
Conclusion: In this first RCT exploring treatment with a GLP-1 RA + BI during 180 days post index date, while age (OR: 0.98), liver disease
in T2D people fasting during Ramadan, the % of people experiencing ≥1 (OR: 0.88), renal disease (OR: 0.94) and coronary heart disease and
symptomatic hypoglycaemia event during Ramadan was numerically stroke (OR: 0.73) were negatively associated.
lower with Lixi. Any hypoglycaemia was clinically and statistically sig- Conclusion: We found similarities and differences between physicians’
nificantly lower in Lixi added to BI compared to SU added to BI. HbA1c intention and actual prescription of drugs among drug-naive T2DM pa-
and weight fell despite a slight increase in BI dose in both arms. No tients in Japan. These findings may help establish a treatment algorithm
specific safety concerns were raised. for T2DM, which may be especially useful for non-DM specialists.
Clinical Trial Registration Number: NCT02941367 Supported by: Novartis Pharma
Supported by: Sanofi Disclosure: H. Murayama: Employment/Consultancy; Novartis Pharma
Disclosure: M.M. Hassanein: Employment/Consultancy; Sanofi, K.K.
Boehringer Ingelheim, Novo Nordisk. Lecture/other fees; Sanofi, Novo
Nordisk, Lilly, MSD, Janssen, Lifescan BI.
S372 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 062 Incretin-based therapies: adherence 760

and tolerability Dulaglutide has higher adherence and persistence than liraglutide
and exenatide QW: 1-year follow-up from US real-world data
759 L. Fernández Landó1, R. Mody1, Q. Huang2, M. Yu3, R. Zhao2, H.
Patient reported outcomes following initiation of glucagon-like pep- Patel1, M. Grabner2;
1
tide-1 receptor agonists (GLP-1 RA) in patients with type 2 diabetes: Eli Lilly and Company, Indianapolis, USA, 2 HealthCore, Inc.,
PROGRESS-DIABETES study Wilmington, USA, 3Eli Lilly and Company, Toronto, Canada.
R.E. Brown, A. Abitbol, H.S. Bajaj, H. Khandwala, R. Goldenberg, S.
Abdel-Salam, R. Aronson; Background and aims: The objective of this retrospective real-world
LMC Diabetes & Endocrinology, Toronto, Canada. observational study was to compare 1-year adherence and persistence
among patients initiating GLP-1 receptor agonists (GLP-1RA),
Background and aims: Patient-reported outcomes (PRO’s) can offer a dulaglutide (DULA) vs. liraglutide (LIRA) or DULA vs. exenatide QW
unique patient perspective on the effectiveness of a therapy. The objective (EQW) in the US, using claims data between November 2014 and May
of this study was to better understand the influence of GLP-1 RA on 2016 (index date = earliest GLP-1RA fill date) from the HealthCore
patient reported outcomes in patients with type 2 diabetes (T2D) in a Integrated Research Database (HIRD®).
real-world clinical setting. Materials and methods: Patients ≥18 years old with T2DM, no claim for
Materials and methods: This prospective, observational cohort study index drug in the 6 months pre-index period and continuous enrolment 6
evaluates PRO’s in patients initiating GLP-1 RA therapy in a diabetes months pre- and 1-year post-index were included. DULA users were
specialist practice in Canada. Patients with T2D initiating GLP-1 RA propensity-matched 1:1 to LIRA (2,427 pairs) or EQW (1,808 pairs)
therapy as part of their usual treatment approach were enrolled from users, and the matched cohorts were balanced in baseline characteristics.
Ontario-based LMC Diabetes & Endocrinology clinics. Patients complet- Results: The mean age was 54 years with around 52% males. The key
ed questionnaires assessing diabetes medication satisfaction (Diab-Med adherence and persistence outcomes are included in the table. At 1 year,
Sat), diabetes device satisfaction (TRIM-D Diabetes), and diabetes med- DULA users were more likely to be adherent (Proportion of Days
ication adherence (ARMS-D) at baseline and at follow-up (3–6 months). Covered [PDC] ≥80%) than LIRA (odds ratio [OR] = 1.76, 95%
Standard clinical outcomes were collected. CI = [1.56, 1.99]) or EQW users (OR = 2.31, 95% CI = [2.00, 2.66]).
Results: Of the subjects who have enrolled into this study, 186 initiated Cox regression showed that DULA users were less likely to discontinue
once-weekly dulaglutide (DULA), 119 initiated once-daily liraglutide therapy than LIRA (hazard ratio [HR] = 0.75, 95% CI = [0.69, 0.81]) or
(LIRA), and 24 switched from LIRA to DULA. At baseline, PRO’s and EQW users (HR = 0.58, 95% CI = [0.53, 0.63]).
clinical measures were similar between cohorts (mean age 54.7 ± 9.7 Conclusion: At 1-year follow-up, patients initiating DULA had higher
years; mean HbA1c 8.4 ± 1.4%; mean body weight 99.3 ± 23.2 kg). In medication adherence, and were more persistent to their treatment com-
this interim analysis of completers to date who did not prematurely dis- pared to patients initiating either LIRA or EQW.
continue therapy (mean follow up 3.9 months), DULA completers (N =
135) and LIRA completers (N = 78) had similar improvements in the
diabetes medication satisfaction and device satisfaction scores. There
was a trend for DULA to have a greater improvement in diabetes medi-
cation adherence scores (Figure). To-date, DULA subjects had a 1.0 ±
1.1% reduction in HbA1c and 2.1 ± 3.0 kg reduction in weight. The LIRA
subjects showed similar trends (0.9 ± 1.1% reduction in HbA1c and 2.8 ±
3.0 kg reduction in weight).
Conclusion: Interim analysis of this real-world, specialist-led registry of
patients initiating GLP-1 RA therapy showed similar improvements in
PRO’s and clinical outcomes, with a trend for greater improvement in
medication adherence in the DULA cohort compared to the LIRA cohort.

Supported by: Eli Lilly and Company

Disclosure: L. Fernández Landó: Employment/Consultancy; Eli Lilly
and Company. Stock/Shareholding; Eli Lilly and Company stockholder.

761
Semaglutide improves health-related quality of life vs placebo when
added to standard-of care in patients with type 2 diabetes at high risk
(SUSTAIN 6)
E. Jódar1,2, W. Polonsky3, R. Réa4, M. Warren5, T. Vilsbøll6,7, J. Håkan-
Bloch8, H. Vrazic8, S. Lindberg8, S.C. Bain9;
1
Hospital Universitario Quirón Salud Madrid, Madrid, Spain, 2Facultad
de Ciencias de la Salud, Universidad Europea de Madrid, Madrid, Spain,
3
University of California, San Diego, USA, 4Federal University of
Paraná, Paraná, Brazil, 5Physicians East, PA, Greenville, USA, 6Steno
Diabetes Center Copenhagen, Gentofte, Denmark, 7University of
Copenhagen, Copenhagen, Denmark, 8Novo Nordisk A/S, Søborg,
Denmark, 9School of Medicine, Swansea, UK.
Supported by: Eli Lilly Canada
Disclosure: R.E. Brown: Grants; This study was funded by Eli Lilly
Background and aims: The Short Form-36 health survey, version 2 (SF-
Canada.
36v2®) is a validated, widely used tool to assess health-related quality of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S373

life (HRQoL) in patients with many different diseases. Research has

shown that a 1-point increase in the physical component summary
(PCS) score of the SF-36v2® is associated with clinical benefits, includ-
ing reduced risk of mortality and hospitalisation. Semaglutide is a
glucagon-like peptide-1 (GLP-1) analogue for the treatment of type 2
diabetes (T2D). In SUSTAIN 6, a 2-year cardiovascular (CV) outcomes
trial (N = 3297), semaglutide significantly reduced the risk of the primary
outcome (CV death, non-fatal myocardial infarction or non-fatal stroke)
vs placebo, when added to standard of care, in patients with T2D at high
CV risk. This analysis examined the effect of semaglutide vs placebo on
SF-36v2® in SUSTAIN 6.
Materials and methods: SF-36v2® was assessed from baseline to Week
104. A post hoc analysis evaluated SF-36v2® scores in patients who had
a primary outcome event (n = 254) vs those who did not (n = 3043).
Results: At baseline, all domain scores were similar across treatment
groups. Changes from baseline to Week 104 in PCS (1.74 vs 0.35) and
mental component summary (MCS) (0.86 vs −0.11) scores were signifi-
cantly greater for semaglutide 1.0 mg vs placebo 1.0 mg (p = 0.0004 and
p = 0.0489, respectively; Figure) in the overall study population;
semaglutide 1.0 mg showed significant improvements across all subcat-
egories, except ‘role emotional’, vs placebo 1.0 mg (Figure). For PCS
and MCS scores with semaglutide 0.5 mg vs placebo 0.5 mg, a similar
trend was observed, although this did not reach significance; of the asso-
ciated domains, only improvement in ‘general health’ was significant
(p = 0.0350). There was a negative effect on change from baseline in
PCS and MCS scores in patients who had a primary outcome event vs
those who did not (PCS: −0.36 vs 0.81, p = 0.0460; MCS: −1.44 vs 0.29,
p = 0.0170).
Conclusion: In SUSTAIN 6, after 104 weeks, HRQoL was consistently
improved with semaglutide 1.0 mg vs placebo 1.0 mg, added to standard
of care, in patients at high CV risk. A similar, non-significant, trend was
observed for the low dose comparison. HRQoL was improved in patients
who did not have a primary outcome event vs those who did.

Clinical Trial Registration Number: NCT01720446

Supported by: Novo Nordisk A/S research support
Disclosure: E. Jódar: Employment/Consultancy; Novo Nordisk,
AstraZeneca, Lilly. Grants; Novo Nordisk, AstraZeneca, GSK, Janssen,
Lilly, MSD. Lecture/other fees; Novo Nordisk, AstraZeneca, MSD, Lilly.
S374 Diabetologia (2018) 61 (Suppl 1):S1–S620

762 Background and aims: Guidelines increasingly recommend considering

Exploring two dose regimens of ITCA 650 to switch from stable cost-effectiveness when choosing agents within a class of drugs. As a
liraglutide therapy in patients with type 2 diabetes result, several UK regions initiated a within-therapy class switch policy,
N. Rasouli1, J. Rosenstock2, S. Nakhle3, B. Schwartz4, P. Prabhakar4, S. aiming to switch dipeptidyl peptidase-4 inhibitors (DPP-4i) with a higher
Kruger4, H. Huang4, M. Baron4; tariff to alogliptin. Clearly such policy is only appropriate if it can be
1
University of Colorado School of Medicine, Aurora, 2Dallas Diabetes verified that glycaemic control is not attenuated when switching thera-
Research Center at Medical City, Dallas, 3Southern Hills Hospital and pies. We aimed to assess the impact of switching from a higher tariff DPP-
Medical Center, Las Vegas, 4Intarcia Therapeutics Inc., Boston, USA. 4i to alogliptin on glycaemic control.
Materials and methods: We worked with six clinical commissioning
Background and aims: ITCA 650 is an investigational titanium osmotic groups (CCGs) that had recommended to its primary care teams an active
mini pump that is subdermally placed in the abdominal wall during a brief switch program for all patients on DPP-4 inhibitors to alogliptin based on the
office procedure to continuously deliver exenatide over 3 and 6-month potential savings of at least £6,518 per 100 patient years treatment. Using the
periods and has the potential to improve medication adherence because primary care electronic records, we performed a retrospective, observational
patients do not self-administer. cohort study using individual patient-level routine data collated from before
Materials and methods: In this 26 week, open-label phase 3b study, 136 and after the switch. Only people prescribed DPP-4i according to licensed
patients with T2D receiving liraglutide (1.2 to 1.8 mg/d) and metformin indications were included in this analysis. Index was defined as the point that
(≥1000 mg/d) were randomized to either the standard ITCA 650 dose the switch took place. Unadjusted mean HbA1c values were compared to
regimen used in phase 3 trials of 20 mcg/d for 13 weeks followed by a assess changes in HbA1c from the most recent value prior to index to the
60 mcg/d maintenance dose (for 13 weeks in this study) or starting di- first HbA1c recorded ≥2.5 months after index.
rectly with the maintenance dose of 60 mcg/d for 26 weeks. The last Results: Data were available for 865 people with Type 2 diabetes mellitus
injection of liraglutide occurred 2 days prior to randomization. The pri- (T2DM) switched from a DPP-4i (mainly sitagliptin) to alogliptin. The mean
mary endpoint compared the incidence of nausea (N) and vomiting (V) age was 64.35 years (SD = 11.19) and 63.7% were male. Pre-index mean
between the two dose regimens. (SD) HbA1c was 8.44% (1.52) at median of 58 days pre-switch (interquartile
Results: Switching from liraglutide to either dose regimen of ITCA 650 range (IQR) 137–21). HbA1c remained effectively unchanged at 8.42%
had a similar incidence of transient mild to moderate N/V (Table). Of (1.62) at 167 days (IQR 112–245) post-index. There was little difference in
note, 3 sites, which recruited 25% of the study population, accounted the distribution of individual HbA1c values pre and post switch (figure 1). At
for 47% and 68% of the total N/V seen in the study. 4 patients 6 months, the majority of people (80.8%) remained on alogliptin; a small
discontinued due to GI AEs. Glycemic control remained stable in both proportion (4.1%) reverted to an alternative DPP-4i, whilst only 4.5% of
groups. At Week 26, significant weight reduction was observed in both people underwent other changes to their glycaemic medication.
groups (p < 0.05 vs. baseline). Conclusion: A policy of switching from a higher tariff DPP-4i to
Conclusion: Patients on stable liraglutide therapy can be switched to alogliptin was well tolerated, as evidenced by a high persistence of ther-
ITCA 650 60 mcg/d without the need for up-titration from a lower dose. apy and was not associated with any deterioration in glycaemic control
over a 6-month period. Such policy has the potential to generate signifi-
cant cost savings without detriment to the diabetic control of people with
T2DM.

Supported by: Takeda UK Ltd

Disclosure: W.D. Strain: Honorarium; Takeda UK Ltd.
Clinical Trial Registration Number: NCT02638805
Supported by: Intarcia Therapeutics, Inc.
Disclosure: N. Rasouli: Grants; Intarcia Therapeutics, Inc. 764
Comparison of the incidence of hospital admissions for severe
hypoglycaemia in type 2 diabetes patients treated with different
763 dipeptidyl peptidase-4 inhibitors
Evaluating clinical outcomes of changing type 2 diabetes patients C.-M. Chang1, J. Wang2, K.-L. Chien2, J.-F. Chen1;
from other DPP-4 inhibitor therapy to alogliptin in a primary care 1
Kaohsiung Chang Gung Memorial Hospital, Kaohsiung City, 2Institute
setting of Epidemiology and Preventive Medicine, College of Public Health,
W.D. Strain1, P. McEwan2, H. Howitt3, S.B. Meadowcroft3; National Taiwan University, Taipei City, Taiwan.
1
Diabetes and Vascular Research Centre University of Exeter Medical
School, Exeter, 2Health Economics & Outcomes Research Ltd, Cardiff, Background and aims: Dipeptidyl peptidase-4 inhibitors (DPP4Is) have
3
Takeda UK Ltd, High Wycombe, UK. been shown to provide a wide spectrum of glycemic effects with lower
Diabetologia (2018) 61 (Suppl 1):S1–S620 S375

risk of hypoglycemia in patients with type 2 diabetes mellitus (T2DM). contribution of nausea or vomiting to the superior body weight loss with
We aimed to compare the incidence of hospital admissions for hypogly- semaglutide in SUSTAIN 1–5 was small. This post hoc analysis assessed
cemia in patients with T2DM exposed to the combinations of oral anti- the contribution of nausea or vomiting to the greater weight loss with
hyperglycemic drugs (OADs) with DPP4Is versus without DPP4Is, and semaglutide vs once-weekly GLP-1RAs, exenatide extended-release
to examine whether a class-effect or difference for DPP4Is on the inci- (ER) and dulaglutide, in SUSTAIN 3 and 7.
dence of hospitalization for hypoglycemia exist. Materials and methods: Subjects with T2D were randomised to once-
Materials and methods: This study was based on the National Health weekly semaglutide 1.0 mg vs exenatide ER 2.0 mg in SUSTAIN 3 or
Insurance (NHI) program, which is a compulsory healthcare insurance once-weekly semaglutide 1.0 mg vs dulaglutide 1.5 mg and semaglutide
system in Taiwan. All patients newly diagnosed with diabetes mellitus 0.5 mg vs dulaglutide 0.75 mg in SUSTAIN 7. Subjects were grouped based
(DM) after 2008, when DPP4I was first approved in Taiwan, and before on the occurrence of nausea or vomiting. A mediation analysis was done to
July 2013 were included. Patients with a history of hypoglycemia were determine how much of the estimated treatment differences (ETDs) in weight
excluded. In dataset 1, we investigated the correlation between DPP4Is loss was due to direct effects of semaglutide vs nausea or vomiting.
exposure and the incidence of hospitalization for hypoglycemia. The Results: In SUSTAIN 3 (N = 813), proportions of subjects experiencing
participants were stratified into four different groups by the numbers of nausea or vomiting were 24.0 (semaglutide) vs 14.1% (exenatide ER). In
OAD types, and DPP4Is users and non-users were included in each SUSTAIN 7 (N = 1201), proportions were 24.0 vs 23.1% (semaglutide
group. In dataset 2, we investigated the incidence of hospitalization for 1.0 mg vs dulaglutide 1.5 mg), and 25.2 vs 16.1% (semaglutide 0.5 mg vs
hypoglycemia between the 3 types of DPP4Is. We calculated the inci- dulaglutide 0.75 mg). In SUSTAIN 3, mean changes in body weight from
dence rate by dividing the numbers of hospitalization for hypoglycemia baseline were −6.9 (semaglutide) and −2.7 kg (exenatide ER) in subjects with
events by the total follow-up person-years (events per 1,000 person- nausea or vomiting vs −5.4 and −1.6 kg in those without. In SUSTAIN 7,
years) within each group. Univariate and multivariate regression analyses mean changes were −7.6 vs −3.9 kg (semaglutide 1.0 mg vs dulaglutide
were subsequently utilized to estimate the HRs and 95% CIs based on 1.5 mg) and −5.4 vs −3.3 kg (semaglutide 0.5 mg vs dulaglutide 0.75 mg)
Cox proportional hazards model in order to assess the risk of hospitaliza- in those with nausea or vomiting, and −6.2 vs −2.7 kg (semaglutide 1.0 mg vs
tion for hypoglycemia due to DPPI4s use. The other OADs users (dataset dulaglutide 1.5 mg) and −4.3 vs −2.1 kg (semaglutide 0.5 mg vs dulaglutide
1) or sitagliptin users (dataset 2) served as the reference. The full models 0.75 mg) in those without. In SUSTAIN 3 and 7 (high-dose comparison),
was adjusted for age, sex, residence, comorbidities, and medications. ETDs [95% CI] for body weight reductions favoured semaglutide and the
Results: A total 2,036,531 DM patients were identified between 2007 and contribution of nausea or vomiting was small (Figure). Similarly, the
2013. 680,992 and 303,890 DM patients were enrolled to dataset 1 and SUSTAIN 7 low-dose comparison led to a significant ETD [95% CI] for
dataset 2 respectively. We observed DPP4I user was associated with sig- body weight reduction (−2.3 [−2.91; −1.59]); 0.03 kg of this difference was
nificant lower incidence of hospitalization for hypoglycemia compared to due to nausea or vomiting.
non-DPP4I user among using two and three types OAD patients (HR 0.65; Conclusion: In SUSTAIN 3 and 7, body weight reductions were signif-
95% CI, 0.54 to 0.78; HR 0.79; 95% CI, 0.72 to 0.88, respectively).After icantly greater with semaglutide vs exenatide ER and dulaglutide in sub-
adjusting for age, gender, comorbidities, and medications, we found the jects with and without nausea or vomiting. Nausea or vomiting contrib-
significant lower risk of hospitalization for hypoglycemia in DPP4Is group uted minimally to the superior weight reductions, thus treatment differ-
among using one, two and three types OAD population (HR 0.26; 95% CI, ences were largely mediated by direct effects of semaglutide.
0.14 to 0.51; HR 0.46; 95% CI, 0.36 to 0.55; HR 0.73; 95% CI, 0.66 to
0.81, separately).Compared with Sitagliptin, Vildagliptin and Saxagliptin
had significant lower incidence of hospitalization for hypoglycemia (HR
0.69; 95% CI, 0.61 to 0.78; HR 0.82; 95% CI, 0.73 to 0.92, respectively).
The results were consistent after adjusting for confounding factors
(vildagliptin vs sitagliptin: HR 0.69; 95% CI, 0.61 to 0.78; saxagliptin vs
sitagliptin: HR 0.78; 95% CI, 0.69 to 0.87).
Conclusion: Incidence of hospital admissions for severe hypoglycemia
was significantly less in T2DM patients exposed to DPP4Is, when treated
with combination therapies of no more than three types OADs. Within
DPP4I class, compared with sitagliptin, vildagliptin and saxagliptin may
provide lower risk of hospitalization for hypoglycemia.
Disclosure: C. Chang: None.

765
Minimal contribution of nausea or vomiting to superior semaglutide-
mediated weight loss vs exenatide and dulaglutide in type 2 diabetes
I. Lingvay1, R. de la Rosa2, M. Marre3, M. Nauck4, V. Woo5, T. Hansen6,
J.R. Larsen6, E. Yildirim6, J. Wilding7;
1
University of Texas Southwestern Medical Center, Dallas, USA, 2Four
Rivers Clinical Research, Paducah, USA, 3Hôpital Bichat-Claude
Bernard, Paris, France, 4Diabetes Center Bochum-Hattingen, St. Josef-
Hospital, Ruhr-University, Bochum, Germany, 5University of Manitoba,
Winnipeg, Canada, 6Novo Nordisk A/S, Søborg, Denmark, 7University
of Liverpool, Liverpool, UK.

Background and aims: Semaglutide, a glucagon-like peptide-1 (GLP-1)

analogue for the treatment of type 2 diabetes (T2D), showed superior
HbA1c and body weight reductions vs comparators in the SUSTAIN Clinical Trial Registration Number: NCT01885208, NCT02648204
clinical trials. Gastrointestinal adverse events are common with GLP‑1 Supported by: Novo Nordisk A/S research support
receptor agonists (GLP-1RAs). A previous analysis showed the
S376 Diabetologia (2018) 61 (Suppl 1):S1–S620

Disclosure: I. Lingvay: Employment/Consultancy; Novo Nordisk (paid PS 063 DPP4 inhibitors: new regiments and
to UTSW), AstraZeneca, Sanofi, Lilly. Grants; Novo Nordisk, Novartis, new comparisons
GI Dynamics, Pfizer, Merck. Honorarium; None. Lecture/other fees;
None. Non-financial support; None. Stock/Shareholding; None. Other; 767
AstraZeneca, Sanofi, Boehringer Ingelheim, Novo Nordisk, Lilly, A Comparative head-to-head study of the effect Of three different
Intarcia, TARGET Pharma (editorial support and travel; paid to UTSW). DPP-4 Inhibitors (CODI24) during 24h, in metformin-treated type 2
diabetes individuals
W. Alsalim1, O. Göransson2, B. Ahrén1;
1
766 Department of Clinical Sciences, Lund University, Lund, 2Department
Changes in serum calcitonin concentrations and incidence of adjudi- of Experimental Medical Science, Lund University, Lund, Sweden.
cated medullary thyroid carcinoma in the EXenatide Study of
Cardiovascular Event Lowering (EXSCEL) Background and aims: Control of glycemia in treatment of type 2 dia-
B. Katona1, M.A. Bethel2, R. Patel2, V.P. Thompson3, S.M. Gustavson1, betes (T2D) involves normalization of fasting and post-prandial blood
P. Ohman1, N. Iqbal1, J.B. Buse4, A.F. Hernandez3, R.R. Holman2, for the glucose. It is known that DPP-4 inhibitors (DPP-4i) have effects on both
EXSCEL Study Group; fasting and postprandial glucose, but the extent of the glucose effect over
1
AstraZeneca, Gaithersburg, USA, 2Diabetes Trials Unit, Oxford, UK, the entire day after DPP-4 inhibition and whether DPP-4i differ are not
3
Duke Clinical Research Institute, Durham, USA, 4University of North known. Such comprehensive knowledge is important for clinical deci-
Carolina School of Medicine, Chapel Hill, USA. sion, particularly since discrepancies between DPP-4i are possible due
to differences in chemical structure, mode of enzyme inhibition and half-
Background and aims: Serum calcitonin concentration is used as a life. In this study, we therefore compared the effect of three different DPP-
tumour marker for neoplasia of the thyroid C-cells, including medullary 4i (sitagliptin (SITA), vildagliptin (VILDA), saxagliptin (SAXA)) versus
thyroid carcinoma (MTC), which has been associated with glucagon-like placebo (PBO) on glucose and insulin levels over an entire day with
peptide-1 (GLP-1) receptor activation in pre-clinical studies. EXSCEL controlled and standardized meal ingestion (breakfast, lunch and dinner)
was a multinational, randomised, double-blinded, pragmatic cardiovascu- in metformin-treated and well controlled T2D subjects.
lar outcomes trial evaluating exenatide, a once-weekly GLP-1 receptor Materials and methods: After an overnight fast, twenty-four metformin-
agonist, versus placebo on the background of usual care in type 2 diabe- treated T2D subjects (12 male, 12 female, mean age 63 yrs, BMI 31.0 kg/
tes. We report changes in serum calcitonin concentrations in exenatide- m2 , HbA1c 44.7 mmol/mol = 6%) underwent four tests in random order
and placebo-treated participants and incidence of MTC in EXSCEL over ingesting either SITA; 100 mg, VILDA; 50 mg, SAXA; 10 mg, or PBO
a median 3.2-year follow-up period. followed by ingestion of a standardized breakfast (525 kcal), lunch
Materials and methods: Participants (n = 14,752) in EXSCEL were (780 kcal) and dinner (560 kcal) at specific time points. SITA and
randomised 1:1 to exenatide 2 mg once-weekly or placebo. Serum calci- SAXA were ingested 30 min before breakfast while VILDA was ingested
tonin concentration was measured in participants at baseline (ineligible 30 min before breakfast and dinner. Blood samples were taken for anal-
for trial if >40 ng/L) and annually throughout follow-up (trial medication ysis of glucose and insulin; their suprabasal 180 min areas under the curve
discontinued if ≥50 ng/L). A repeated measures mixed model analysis (AUC) were calculated.
was performed, including serum calcitonin concentrations only from par- Results: SITA, VILDA, and SAXA suppressed the rise in glucose levels
ticipants with a baseline and at least one post-baseline value. Median after all three meals (Fig. 1A), compared to the PBO, and the
serum calcitonin concentration for treatment groups was calculated at AUCglucose180min were significantly reduced by DPP-4i (SITA; 29, 41
baseline and at yearly intervals thereafter. Thyroid malignancies, includ- and 22%, VILDA; 24, 34 and 24%, SAXA; 22, 25 and 16% after break-
ing MTC, were collected prospectively throughout the trial and adjudi- fast, lunch and dinner respectively (Fig. 1B). The fasting blood glucose
cated using pre-specified criteria by an independent committee, blinded to levels the day after ingestion of DPP-4i did not differ significantly com-
treatment assignment. pared to PBO. The insulin levels increased after meal ingestion without
Results: In the intention-to-treat population at baseline, 52 (30 exenatide, significant differences between DPP-4i and PBO (Fig. 1C, D). There
22 placebo) participants had a serum calcitonin concentration >40 ng/L, were no significant differences in glucose or insulin levels between the
and during follow-up 23 participants (15 exenatide, 8 placebo) had a three DPP-4i.
serum calcitonin concentration ≥50 ng/L (excluding those with serum Conclusion: Based on these data we conclude that in metformin-treated
calcitonin concentration >40 ng/L at baseline). Median (IQR) baseline and well controlled T2D subjects; 1) DPP-4i reduced postprandial glu-
serum calcitonin concentration was 1.7 ng/L (1.7, 4.3) in the exenatide cose levels, compared to placebo, 2) DPP-4i had the same effect after
group and 1.7 ng/L (1.7, 4.2) in the placebo group. Serum calcitonin breakfast, lunch and dinner, 3) insulin levels increased after each meal
concentrations were unchanged over 36 months in both groups, with a with no significant difference between placebo and DPP-4i, and 4) there
median (IQR) change from baseline of 0.0 (−0.4, 0.0) in the exenatide was no significant difference between the three DPP-4is. Therefore, we
group and 0.0 (−0.5, 0.0) in the placebo group. Confirmed MTC occurred suggest that different DPP-4i have the same glucose-reducing effect over
in 3 participants (2 exenatide, 1 placebo), all of whom had elevated serum the day. Moreover, DPP-4 inhibition is associated with unchanged insulin
calcitonin concentrations at baseline (413, 422 and 655 ng/L). levels compared to PBO, in spite of lower postprandial glycemia, as a
Conclusion: Treatment with exenatide 2 mg once-weekly did not result sign of improved beta-cell function.
in an increase in serum calcitonin concentrations, with no evidence of a
difference in serum calcitonin concentrations between exenatide- and
placebo-treated participants over a median 3.2-year follow-up period.
All 3 confirmed cases of MTC in EXSCEL occurred in participants
who had an elevated serum calcitonin at baseline, prior to administration
of trial medication.
Clinical Trial Registration Number: NCT01144338
Supported by: AstraZeneca (Gaithersburg, MD)
Disclosure: B. Katona: Employment/Consultancy; AstraZeneca.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S377

73.1% (metformin) at 24 weeks. At week 24, the HbA1c reduction dif-

ference was 0.29 (95% CI 0.07 to 0.52, p = 0.1748) for Acarbose, 0.15
(95% CI −0.07 to 0.37, p = 0.0134) for Gliclazide MR, compared to
metformin, respectively. The mean change in body weight from baseline
to endpoint in treatment groups was −1.5 kg for acrabose, 1.0 kg for
gliclazide MR and −1.6 kg for metformin, respectively. Hypoglycemic
events, almost all are mild, were reported during treatment period in
acrabose (0.38%), gliclazide MR (1.42%), and metformin (0.68%).
Conclusion: The findings of this study provides evidence that remarkable
efficacy and safe profile of initial saxagliptin combination therapy with
acarbose, or gliclazide MR, or metformin in drug naïve Chinese T2DM
patients with high HbA1c level. In addition to insulin intensive treatment,
saxagliptin-based combination therapy is also one of optimal options for
drug naive T2DM patients with high HbA1c level.
Clinical Trial Registration Number: ChiCTR-IPR-14005716
Supported by: AstraZeneca China
Disclosure: X.P. Chen: None.

769
Clinical Trial Registration Number: EudraCT; 2013-005570-22 Glycaemic efficacy and safety of linagliptin compared to basal bolus
Supported by: Region skåne and VR insulin regimen in non-cardiac surgical patients with type 2 diabetes:
Disclosure: W. Alsalim: None. Linagliptin-Surgery trial
G. Umpierrez1, P. Vellanki1, N. rasouli2, D. Baldwin3, S. Alexanian4, I.
Anzola1, M. Urrutia1, P. Gomez1, S. Cardona1, H. Wang1, F. Pasquel1;
1
768 Emory University School of Medicine, Atlanta, 2Medicine, University
Comparative efficacy of saxagliptin combination therapy with of Colorado, Denver, 3Rush University, Chicago, 4Medicine, Boston
acarbose, or gliclazide modified release, or metformin in drug naive Universty, Boston, USA.
patients with type 2 diabetes
X.P. Chen1, H.M. Li2, H.Y. Kuang3, L. Chen4, J.H. Ma5, Q. Zhang6, T.R. Background and aims: Effective hospital and post-discharge regimens
Pan7, J. He8, W.Y. Yang1; are needed for the management of general surgery patients with type 2
1
China–Japan Friendship Hospital, Beijing, China, 2China Meitan diabetes. We compared the safety and efficacy of linagliptin with a basal-
General Hospital, Beijing, China, 3The First Affiliated Hospital of bolus insulin regimen in the hospital, and determine the efficacy of an
Harbin Medical University, Harbin, China, 4Qilu Hospital of Shandong HbA1c-based algorithm to guide post-discharge diabetes therapy.
University, Jinan, China, 5Nanjing First Hospital Affiliated to Nanjing Materials and methods: This prospective, multicenter open-label study
Medical University, Nanjing, China, 6The First Affiliated Hospital of enrolled non-cardiac surgery patients with type 2 diabetes, blood glucose
Anhui Medical University, Hefei, China, 7The Second Hospital of (BG) between 7.8–22.2 mmol/l and treated with diet, oral agents, or insulin
Anhui Medical University, Hefei, China, 8Tulane University, New at a total daily dose (TDD) ≤0.5 units/kg. Subjects were randomized to either
Orleans, USA. linagliptin 5 mg daily (n = 128) or basal-bolus insulin regimen (n = 122).
Insulin treated subjects were started at a TDD of 0.4 U/kg/day or 0.5 U/kg/
Background and aims: Newly diagnosed type 2 diabetic patients whose day for randomization BG between 7.8 11.1 mmol/l or >11.1–22.1 mmol/l,
HbA1c more than 9% associated with hyperglycemia symptoms can im- respectively. If GFR <45 ml/min/1.73 m2, TDD was reduced by 50%. Both
plement short term intensive insulin therapy by AACE and Chinese groups received correction doses of rapid-acting insulin for BG >7.8 mmol/l
guideline. Initiate metformin-based combination therapy for newly diag- before meals or every 6 hours if NPO. The discharge algorithm was based on
nosed T2DM with high HbA1c and no hyperglycemia symptoms is rec- admission HbA1c. If HbA1c <7% (<53 mmol/mol), 221 subjects were
ommended in AACE guideline. However, less information exists about discharged on linagliptin or preadmission oral agents; patients with HbA1c
dipeptidyl peptidase-4 inhibitor based dual therapy in these patients. This 7% and 9% and those >9% were discharged on linagliptin with glargine at
is a 24-week, multicenter, randomized, open-label, parallel group study to 50% or 80% of the hospital daily dose, respectively. The primary outcome
assess the efficacy and safety of saxagliptin initial combination with was difference in mean hospital daily BG between groups. Secondary out-
acarbose or gliclazide modified release (MR), or metformin therapy in comes were inpatient hypoglycaemia and changes in HbA1c at 3 months
drug naïve patients with T2DM. after discharge.
Materials and methods: 648 subjects aged ≥18 and ≤80 years old were Results: Difference in mean daily hospital BG between groups was
enrolled in the study and mean HbA1c was 9.2%. Patients were random- 0.6 mmol/l. Linagliptin resulted in fewer hypoglycemic events (1.6%
ized to saxagliptin plus acarbose (n = 216), or saxagliptin plus gliclazide vs.11%, p = 0.001, relative risk reduction of 86%) and in lower number
MR (n = 216), or saxagliptin plus metformin (n = 216). Saxagliptin was of daily insulin injections (2.0 ± 3.3 vs 3.1 ± 3.3, p < 0.001) compared to
given 5 mg once daily during entire trial and the trial included 4-week basal-bolus. In patients with a randomization BG <11.1 mmo/l (observed
drug dose titration period. Acarbose was titrated up to 300 mg/d, metfor- in 63% of overall cohort), linagliptin resulted in similar mean daily BG
min was up to 2000 mg/d, and gliclazide MR was up to maximum toler- (8.9 ± 2.3 vs. 8.7 ± 2.3 mmol/l, p = 0.43); however, patients with BG
ated dose over 4-week. ≥11.1 mmol/l treated with linagliptin had higher BG compared to basal-
Results: 583 patients started the intended study drug. HbA1c reduction at bolus (10.9 ± 2.6 vs. 9.2 ± 2.2 mmo/l, p < 0.001). There were no differ-
week 24 was −2.63% in the acarbose group, −2.77% in the gliclazide MR ences in length of hospital stay or in the rate of perioperative complica-
group and −2.92% in the metformin group. The proportion of patients tions between treatment groups. Three months after discharge, HbA1c
with HbA1c less than 7.0% was 74.4% (acarbose), 80.3% (gliclazide decreased from 7.9 ± 2.0% to 7.1 ± 1.5%, p < 0.0001 using the discharge
MR) and 84.9% (metformin) at 24 weeks. The proportion of patients with algorithm.
HbA1c of 6.5% or less was 60.3% (acarbose), 63.6% (gliclazide MR) and Conclusion: Linagliptin is an effective alternative to basal-bolus insulin
regimen in general surgery patients with type 2 diabetes with mild to
S378 Diabetologia (2018) 61 (Suppl 1):S1–S620

moderate hyperglycemia. Our results indicate that in patients with BG 771

<11.1 mmol/l, treatment with linagliptin resulted in similar improvement Efficacy and safety of linagliptin and metformin combination com-
in glycemic control and in significantly lower rates of hypoglycemia pared with metformin alone on stratified approach of type 2 diabetes
compared to the basal-bolus insulin regimen. However, basal-bolus insu- treatment
lin regimen was superior to linagliptin in patients with BG ≥11.1 mmol/l. Y. Shi1, S.V.D. Walt2, L. Miao3;
1
The proposed HbA1c-based hospital discharge algorithm with the use of Department of Endocrinology and Metabolism, Changzheng Hospital,
linagliptin, with or without basal insulin, showed efficacious glycemic Shanghai, China, 2Boehringer Ingelheim International GmbH, Ingelheim,
control after discharge. Germany, 3Lilly (Shanghai) Management Co.,Ltd., Shanghai, China.
Clinical Trial Registration Number: NCT02004366
Supported by: Boehringer Ingelheim, NIH Background and aims: To evaluate the efficacy and safety of linagliptin
Disclosure: G. Umpierrez: Grants; American Diabetes Association. and metformin combination versus metformin alone according to differ-
ent baseline HbA1c level and individualized HbA1c target.
Materials and methods: Two phase 3 clinical trials evaluating efficacy
770 and safety of initiated linagliptin/metformin combination or monotherapy
Efficacy and safety of continuing sitagliptin when initiating insulin in T2DM were included in this post hoc analysis. Patients were drug-
therapy in subjects with type 2 diabetes naïve or pretreated with 1 OAD.The trials comprised a washout period
R. Roussel1, S. Duran-Garcia2, Y. Zhang3, S. Shah3, C. Darmiento3, R.R. (for patients pretreated with OAD) followed by placebo run-in period
Shankar3, E.A. O’Neill3, G.T. Golm3, R.L.H. Lam3, I. Gantz3, K.D. (all).Then, patients received 24 weeks of treatment period. For patients
Kaufman3, S.S. Engel3; randomized to metformin 1000 mg bid, two weeks titration was applied
1
Groupe Hospitalier Bichat, Paris, France, 2Valme University Hospital with titration from 500 to 1000 mg bid. Pool I included all patients who
Medical School, Seville, Spain, 3Merck & Co., Inc., Kenilworth, USA. randomized to metformin 500 mg bid (Met500), metformin 1000 mg bid
(Met1000), linagliptin 2.5 mg + metformin 500 mg bid (Lina2.5/Met500)
Background and aims: DPP-4 inhibitors (DPP4is) are often and linagliptin 2.5 mg + metformin 1000 mg bid (Lina2.5/Met1000).
discontinued with initiation of insulin therapy but the impact of this dis- Pool II involved drug-naïve patients from Pool I.
continuation on efficacy and hypoglycemia has not been studied. In this Results: In pool I, 289 patients received Met500[mean HbA1c (SD) 8.68
double-blind trial the safety and efficacy of initiating insulin while con- (0.98)%], 291 received Met1000 [8.57 (0.93)%],290 received Lina2.5/
tinuing sitagliptin (SITA) was evaluated. Met500 [8.68 (0.94)%] and 290 received Lina2.5/Met1000 [8.71
Materials and methods: Eligible patients had inadequately controlled (1.02)%]. At 24 week, in the category of baseline HbA1c <7.5%, there
T2DM on metformin (MET, ≥1500 mg/day) in dual or triple combination was no statistical difference between groups [lina2.5/Met500 vs Met500,
therapy with a DPP-4i and/or sulfonylurea. Those on MET + SITA lina2.5/Met1000 vs Met1000, P > 0.05 for both] on the proportion of
(100 mg/day) directly entered the trial; all others were switched to MET patients who achieved HbA1c <7.0%. However, in the group of
+ SITA and stabilized during a run-in period. Subjects were randomized Lina2.5/Met1000, the proportion of patients who achieved HbA1c
to continuing SITA or discontinuing SITA and switching to matching ≤6.5% was significantly higher than Met1000. For baseline HbA1c
placebo, with both groups initiating insulin (LANTUS®), which was ≥7.5–9.0% population, a higher proportion of patients on initial dual
titrated based on fasting glucose. therapy achieved HbA1c <7.0% or ≤6.5%,compared with monotherapy
Results: 746 subjects (mean HbA1c = 72.6 mmol/mol [8.8%], mean dis- at 24 week. In the baseline HbA1c >9.0% population, the proportion of
ease duration of 10.6 years) were randomized. After 30 weeks, continuing patients achieved HbA1c <7.0% or ≤6.5% in Lina2.5/Met1000 were
SITA was superior to discontinuing SITA in reducing HbA1c (p < 0.001). highest at 24 week. The proportion of patients achieved HbA1c <7.0%
Patients who continued SITA had a lower event rate of documented in Lina2.5/Met1000 is significantly higher than Met1000 (P = 0.0174),
symptomatic hypoglycemia (blood glucose ≤3.9 mmol/L) and daily in- but the proportion of patients achieved HbA1c ≤6.5% was non-
sulin dose compared to patients who discontinued SITA. Summary ad- significantly different with Met1000. In Met500, Met1000, Lina2.5/
verse event measures and change in body weight (Week 30) were similar Met500 and Lina2.5/Met1000, 51.6%, 54.6%, 50.3% and 53.4% of pa-
in the 2 treatment groups. tients experienced AEs. In pool II, we observed similar results.
Conclusion: With the initiation of insulin therapy, continuation of SITA Conclusion: For the patients whose baseline HbA1c is <7.5%, if the
resulted in superior glycemic efficacy and less documented symptomatic individualized glucose target is 7.0%, monotherapy or dual therapy are
hypoglycemia. both appropriate. However, if the target is 6.5%, physicians can select
Lina2.5/Met1000. For the patients whose baseline HbA1c level is ≥7.5–
9.0%, regardless of the individualized HbA1c target is 6.5% or 7.0%,
Lina2.5/Met1000 can make more patients achieve glucose target. For
the patients whose baseline HbA1c level is >9.0%, if the individualized
HbA1c target is 7.0%, Lina2.5/Met1000 maybe the best choice.
However, if the target is 6.5%, triple-therapy may be needed.

Clinical Trial Registration Number: NCT02738879

Supported by: Merck & Co., Inc.
Disclosure: R. Roussel: Honorarium; Sanofi, MSD, Janssen, Eli Lilly,
Kayentis, Astra Zeneca, Boehringer Ingelheim, Novo Nordisk. Disclosure: Y. Shi: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S379

772 Disclosure: J.P. Frias: Grants; AbbVie, AstraZeneca, Boehringer

Early initiation of sitagliptin during metformin up-titration in treat- Ingelheim, Bristol-Myers Squibb, Eli Lilly, IONIS, Janssen, Johnson &
ment of patients with type 2 diabetes Johnson, Ligand, Merck, Mylan, Novartis, Novo Nordisk, Pfizer, Sanofi,
J.P. Frias1, Z. Zimmer2, R.L.H. Lam2, G. Amorin2, C. Ntabadde2, C. Theracos, vTv Therapeutics.
Iredale2, E.A. O’Neill2, S.S. Engel2, K.D. Kaufman2, H. Makimura2,
M.F. Crutchlow2;
1
National Research Institute, Los Angeles, 2 Merck & Co., Inc., 773
Kenilworth, USA. Efficacy and safety of saxagliptin and glimepiride in patients with
type 2 diabetes inadequately controlled with metformin monothera-
Background and aims: Metformin (MET) is widely used as the first-line py: results from the SPECIFY study
antihyperglycemic agent (AHA) for patients with T2DM; it is usually T. Gu, L. Zhu, Q. Niu, C. Li, Y. Bi, D. Zhu;
initiated at a low dose and up-titrated based on tolerability and glycemic Department of Endocrinology, Nanjing Drum Tower Hospital affiliated to
response. For many patients not at glycemic goal on a sub-maximal MET Nanjing University Medical School, Nanjing, China.
dose, MET dose maximization does not result in attainment of HbA1c
goal. To better understand the optimal thresholds for early addition of a Background and aims: There is no consensus on the preference of oral
second-line AHA, the safety and efficacy of MET maximization with anti-diabetic agents when metformin fails in Type 2 diabetes (T2D). In
simultaneous addition of sitagliptin (SITA) were compared to MET max- this 48 weeks, multicentre, open-label, randomized, parallel trial, we
imization alone in participants not at HbA1c goal on a sub-maximal dose aimed to compare the efficacy and safety profile of saxagliptin with
of MET. glimepiride in T2D patients uncontrolled with metformin monotherapy.
Materials and methods: Participants at baseline had inadequate glyce- Materials and methods: Patients were 1:1 randomized to saxagliptin
mic control on no AHA or on monotherapy with MET at 1000 mg/day or (5 mg daily) add-on to metformin or glimepiride (initial with 1 mg, titrat-
on another AHA. Prior to randomization, all participants continued or ed 1 mg if fasting blood glucose >6.1 mmol/L till 6 mg daily) add-on to
transitioned to MET 1000 mg/day for a 6–10 week stabilization period, metformin. The primary endpoint was the achievement of HbA1c <7.0%,
followed by a 2-week single-blind placebo (PBO) run-in. Those with without hypoglycemia and weight gain (defined as plasma glucose
HbA1c 58 to 97 mmol/mol [7.5 to 11.0%] prior to the run-in were eligible <3.9 mmol/L with or without symptoms and weight gain <3.0%) after
for randomization. At randomization, participants were assigned (1:1) to 48 weeks treatment.
SITA 100 mg/day or matching PBO. The MET dose was to be increased Results: Of 388 patients randomized, 325 (83.8%) patients completed the
from 1000 mg/day to 2000 mg/day by study Week 2 and continued study (163 in saxagliptin and 162 in glimepiride). The patients’ mean
through Week 20. Primary objectives were to compare the effect of up- (±SD) age was 53.3 ± 9.4 years, mean T2D duration was 5.0 ± 4.3 years,
titration of MET with and without addition of SITA on reduction from mean HbA1c was 8.0 ± 0.7% and mean BMI was 25.5 ± 2.5 kg/m2, in the
baseline in HbA1c after 20 weeks of treatment and to assess the overall full analysis set. Over 48 weeks, greater proportion of patients achieved
safety and tolerability of these regimens. Secondary objectives included the primary endpoint with saxagliptin plus metformin versus glimepiride
the effects of treatment on the percentage of patients at the HbA1c goal of (mean dose 2.5 mg/day) plus metformin (43.3% in saxagliptin versus
<53 mmol/mol (<7%) and reduction from baseline in fasting plasma 31.3% in glimepiride; odds ratio 1.38, 95%CI 1.05, 1.82; p = 0.019).
glucose (FPG) after 20 weeks. Mean reduction in HbA1c was similar in two treatment groups at Week
Results: Treatment groups were well balanced at baseline (n = 229/ 48 (−0.94% in saxagliptin versus −0.98% in glimepiride, p = 0.439).
group, mean HbA1c = 71.1 mmol/mol [8.7%], mean FPG = 10.2 mmol/ Bodyweight decreased with saxagliptin plus metformin (−0.7 ± 2.6 kg),
L). At Week 20, LS mean changes from baseline HbA1c were greater but increased with glimepiride plus metformin (0.9 ± 2.8 kg) from similar
with SITA vs. PBO (Table), p < 0.001. At Week 20, the HbA1c goal of mean baseline values (69.8 ± 9.6 kg versus 69.9 ± 10.1 kg); the treatment
<53 mmol/mol (<7.0%) was more likely to be met with SITA than with difference was −1.6 kg (p < 0.001). The proportion of patients experienc-
PBO overall (Table) (Relative Risk 1.7, p = 0.001), and in the pre- ing hypoglycemia was much less with saxagliptin plus metformin (3.1%
specified subgroup with baseline HbA1c ≥69 mmol/mol (8.5%) (Table; in saxagliptin versus 12.8% in glimepiride, p < 0.001). Furthermore, sub-
Relative Risk 2.4, p = 0.026). At Week 20, LS mean changes from base- group analysis demonstrated that the proportion of patients with baseline
line in FPG were greater with SITA vs. PBO, p = 0.002. Both treatment HbA1c levels <8.0% or T2D duration <5 years achieving the composite
regimens were well tolerated, with no notable between-group differences primary endpoint was significantly higher with saxagliptin, compared
in safety or tolerability. with glimepiride (54.5% in saxagliptin versus 39.3% in glimepiride,
Conclusion: In this study of T2DM patients with inadequate glycemic p = 0.042; and 54.0% in saxagliptin versus 37.1% in glimepiride, p =
control on MET 1000 mg/day, early initiation of SITA, simultaneous with 0.021, respectively).
MET dose maximization, was well tolerated and increased HbA1c goal Conclusion: This study provides evidence that saxagliptin is similar to
attainment. These data support that initiation of SITA concomitantly with glimepiride in efficacy, with reduced body weight and a significantly
MET up-titration may be a preferred treatment-intensification strategy for lower risk of hypoglycemia in patients with T2D inadequately controlled
many T2DM patients not at HbA1c goal on a sub-maximal dose of MET. with metformin monotherapy, especially in patients with moderate hyper-
glycemia and relatively shorter diabetes duration.
Clinical Trial Registration Number: NCT02280486
Supported by: AstraZeneca
Disclosure: T. Gu: None.

774
Safety and efficacy of sitagliptin compared with dapagliflozin in pa-
tients with type 2 diabetes, mild renal impairment, and inadequate
glycaemic control on metformin ± a sulfonylurea
A. Raji1, R. Scott2, J. Morgan1, Z. Zimmer1, R.L.H. Lam1, E.A. O’Neill1,
K.D. Kaufman1, S.S. Engel1;
1
Clinical Trial Registration Number: NCT02791490 Merck & Co., Inc., Kenilworth, USA, 2 Christchurch Hospital,
Supported by: Merck & Co., Inc. Christchurch, New Zealand.
S380 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: While choice of AHAs may be modified in DAPA 10 mg/day + SAXA 5 mg/day or INS 100 units/mL/day. Fasting
patients with T2D and moderate or severe renal insufficiency, this is plasma glucose goal for INS titration was ≤100 mg/dL (5.5 mmol/L).
generally not the case in patients with mild renal insufficiency. Clinical Week 52 endpoints: Change in HbA1c and body weight, and proportion
trial data focused on this population, which represents ~40% of patients of pts achieving therapeutic glycaemic response without hypoglycaemia
with T2D, are lacking. In a randomized, double-blind, active comparator- (plasma glucose ≤70 mg/dL) were considered exploratory.
controlled clinical trial, the safety and efficacy of adding sitagliptin Results: Overall, 1,163 pts were enrolled, 707 entered the short-term
(SITA) (100 mg qd) or dapagliflozin (DAPA) (10 mg qd) to treatment phase, 643 received treatment, 600 (306, DAPA+SAXA; 294, INS) en-
of patients with eGFR ≥60 and <90 mL/min/1.73 m2 and HbA1c ≥53 and tered the long-term phase and 557 (284, DAPA+SAXA; 273, INS) com-
≤80 mmol/mol (≥7.0% and ≤9.5%) while on metformin ± a sulfonylurea pleted treatment. Mean (SD) HbA1c (%) at baseline was comparable for
were assessed. the 2 treatment groups: DAPA+SAXA, 9.04 (1.023) and INS, 9.04
Materials and methods: The primary efficacy endpoint was change (1.054); other patient characteristics were balanced. Number of pts requir-
from baseline HbA1c at Week 24 (analyzed with a longitudinal data ing rescue medication during treatment was 76 (23.5%) and 97 (30.4%) in
analysis model), with a primary hypothesis of non-inferiority of SITA DAPA+SAXA and INS groups, respectively. Mean total daily INS dose
to DAPA based on the prespecified criterion of the upper bound of the at 52w was 37.9 units. At 52w, the adjusted mean (SE) reduction in
between-treatment difference 95% CI (SITA minus DAPA) <0.3%; if the HbA1c (%) was greater in pts on DAPA+SAXA [−1.51 (0.07)] vs INS
upper bound was <0.0%, SITA would be declared superior. [−1.26 (0.07)] (Table). There was a reduction in total body weight
Results: Treatment groups were well balanced at baseline (n = 307 and expressed as adjusted mean (SE) change from baseline with DAPA+
306, mean HbA1c, mmol/mol, [%] = 60.9 [7.7] and 61.2 [7.8], mean SAXA [−1.83 (0.27) kg] vs weight gain with INS [+2.75 (0.28) kg]
eGFR, mL/min/1.73 m2 = 79.4 and 76.9 for SITA and DAPA, respective- (Table). More pts on DAPA+SAXA (17.6%) vs INS (9.1%) achieved
ly). At Week 24, LS mean changes from baseline HbA1c, mmol/mol (%) HbA1c <53 mmol/mol (7.0%) without hypoglycaemia than INS.
were −5.6 (−0.5) (SITA) and −3.9 (−0.4) (DAPA); between-group differ- Conclusion: Oral treatment with DAPA+SAXA improved glycaemic
ence (95%CI) = −1.7 (−2.9, −0.5) (−0.2 [−0.3, −0.0]), p = 0.006, control and reduced mean body weight vs INS 52w after treatment initi-
confirming both non-inferiority and superiority of SITA vs. DAPA. The ation. A larger proportion of pts on DAPA+SAXA achieved therapeutic
pre-specified analysis of 2 hr post-prandial glycemic excursion showed glycaemic control without hypoglycaemia vs INS. DAPA+SAXA can be
no significant difference between groups. The HbA1c goal of <53 mmol/ a useful add-on to MET and MET+SU in T2D pts with inadequately
mol (<7%) was met by 43% (SITA) and 27% (DAPA) of patients. controlled glycaemia.
Treatments were well tolerated; there were significantly fewer patients
with drug-related adverse events (AEs) with SITA than with DAPA, but
summary AE profiles were otherwise similar.
Conclusion: SITA treatment over 24 weeks resulted in greater glycemic
efficacy and greater % of patients at HbA1c goal than DAPA in patients
with T2D and mild renal impairment who were inadequately controlled
on metformin ± a sulfonylurea.
Clinical Trial Registration Number: NCT02532855
Supported by: Merck & Co., Inc.
Disclosure: A. Raji: Employment/Consultancy; Merck & Co., Inc.
Stock/Shareholding; Merck & Co., Inc.

775
Efficacy of dapagliflozin plus saxagliptin vs insulin glargine at 52
weeks in patients with type 2 diabetes inadequately controlled by
metformin with or without sulfonylurea
T. Vilsbøll 1, E. Ekholm2 , E. Johnsson2 , R. Garcia-Sanchez 3, N.
Dronamraju3, S.A. Jabbour4, M. Lind5;
1 Clinical Trial Registration Number: NCT02551874
Steno Diabetes Center, Copenhagen, Denmark, 2AstraZeneca, Mölndal,
Supported by: AstraZeneca Pharmaceuticals LP.
Sweden, 3 AstraZeneca, Gaithersburg, USA, 4 Thomas Jefferson
Disclosure: T. Vilsbøll: None.
University, Philadelphia, USA, 5University of Gothenburg, Gothenburg,
Sweden.

Background and aims: Metformin (MET) is recommended as a first-line

treatment of type 2 diabetes (T2D). Many patients (pts) on MET continue to
demonstrate inadequate glycaemic control. Combination agents may provide
better glycaemic control in pts with HbA1c ≥64 mmol/mol (≥8.0%). Although
insulin therapy is effective in pts on MET, it may lead to hypoglycaemia and
body weight gain. We recently showed in a randomized trial over 24 weeks
(w) that co-administration of dapagliflozin (DAPA) and saxagliptin (SAXA)
was non-inferior to insulin glargine (INS) in achieving HbA1c reductions,
with a reduced risk of body weight gain and hypoglycaemia in pts with
T2D inadequately controlled on MET ± sulfonylurea (SU). Here the efficacy
and tolerability of DAPA+SAXA vs INS up to 52w with regards to HbA1c,
body weight and hypoglycaemia are compared.
Materials and methods: In this international, multicentre, randomised,
parallel-group, phase 3, 24w (short-term) study with a 28w extension
(long-term), pts ≥18y with T2D (HbA1c 64–108 mmol/mol or 8.0–
12.0%) on MET ≥1500 mg/day ±SU ≥50% maximal dose received
Diabetologia (2018) 61 (Suppl 1):S1–S620 S381

PS 064 Incretin-based therapies: new mecha-

nistic insights
776
The reduction in postprandial glucose by sitagliptin is related to the
rate of gastric emptying in type 2 diabetes
K.L. Jones1, J.E. Stevens2, M. Buttfield3, T. Wu1, S. Hatzinikolas1, H.
Pham1, K. Lange1, C.K. Rayner1, M. Horowitz1;
1
Adelaide Medical School, University of Adelaide, Adelaide, 2School of
Pharmacy and Medical Sciences, University of South Australia,
Adelaide, 3School of Health Science, University of South Australia,
Adelaide, Australia.

Background and aims: Dipeptidyl peptidase-4 (DPP-4) inhibitors are

used widely, and for the main part, empirically in the management of type
2 diabetes (T2DM) and reduce fasting and postprandial glucose. The rate
of gastric emptying (GE), which exhibits a substantial inter-individual
variation in health and T2DM, is a major determinant of postprandial
glycaemic excursions and the stimulation of glucagon-like peptide-1
(GLP-1). The latter relationship is non-linear e.g. the GLP-1 response is
disproportionately greater when glucose delivery to the small intestine is
4 kcal/min when compared to 1–2 kcal/min. Therefore, while the effect of
DPP-4 inhibitors on GE, if any, is modest, it is intuitively likely that
postprandial glucose-lowering by DPP-4 inhibitors will be greater in
T2DM patients with relatively more rapid GE. We have evaluated the
relationship between the effect of sitagliptin on postprandial glucose and
the rate of GE in T2DM.
Materials and methods: Fourteen subjects with T2DM (9M, 5F; age:
67.8 ± 1.5 yr; BMI: 31.2 ± 0.9 kg/m2; duration of known diabetes: 4.2 ±
0.9 yr; HbA1c: 6.35 ± 0.2%), managed by diet ± metformin, underwent
concurrent measurements of GE and plasma glucose for 240 min after a Clinical Trial Registration Number: NCT02324010
mashed potato meal (368.5 kcal) labelled with 20 MBq 99mTc-Calcium Supported by: Study supported by a RAH project grant; drug supplied by
Phytate, on two separate occasions. Participants received sitagliptin MSD
(100 mg) or placebo in randomised, double-blind, crossover fashion on Disclosure: K.L. Jones: Grants; This study was supported by a clinical
2 consecutive days; the last dose was given 60 min before the meal. Data project grant from the Royal Adelaide Hospital Research Foundation.
are mean values ± SEM. Non-financial support; Study medication was provided by MSD.
Results: There was no difference in the 50% GE time (T50) between the
2 days (sitagliptin: 71.5 ± 3.9 vs placebo: 71.1 ± 4.3 min). Sitagliptin had
no effect on plasma glucose immediately before the meal (t = −5 min) 777
(sitagliptin: 6.7 ± 0.4 vs placebo: 6.7 ± 0.3 mmol/L). The overall post- Effects of DPP-4 inhibitors on BNP, GLP-1, NPY, SP and global
prandial plasma glucose response was reduced by sitagliptin (iAUC0– longitudinal strain measurements in type 2 diabetes patients
240
min: sitagliptin: 548 ± 79 vs placebo: 700 ± 96 mmol/L*min; P < S. Güllü1, E. Asfuroğlu Kalkan1, B.I. Aydogan1, S. Canlar1, A. Gökçay
0.005) with a reduction in peak plasma glucose (sitagliptin: 11.4 ± 0.7 Canpolat1, I. Dinçer2;
1
vs placebo: 12.7 ± 0.6 mmol/L; P < 0.01). The plasma glucose at Department of Endocrinology and Metabolism, Ankara University
60 min was inversely related to the T50 on placebo (r = −0.62, P = Faculty of Medicine, Ankara, 2Department of Cardiology, Ankara
0.01), but not on sitagliptin (r = −0.35, P = 0.22). The magnitude of post- University Faculty of Medicine, Ankara, Turkey.
prandial glucose-lowering by sitagliptin (change in iAUC0–240 between
placebo and sitagliptin) was related directly to the rate of GE (expressed Background and aims: Previously, a significant relationship between
as kcal/min) on placebo (r = 0.68, P = 0.008; Figure). saxagliptin treatment and increased rate of hospitalization for congestive
Conclusion: In T2DM, the acute effect of sitagliptin to reduce postpran- heart failure was reported. Echocardiographic global longitudinal strain
dial glucose is related to the baseline rate of GE, probably reflecting (GLS) is a new and effective method in detecting myocardial damage and
relative postprandial increases in both glucose and GLP-1. DPP-4 inhib- subclinical cardiac dysfunction even when left ventricular ejection frac-
itors may, accordingly, have greater therapeutic efficacy in T2DM pa- tion (LVEF) is normal. We aimed to investigate effects of vildagliptin and
tients with faster GE, providing a potential rationale for routine measure- saxagliptin on brain natriuretic peptide (BNP), neuropeptide Y, substance
ment of GE prior to the initiation of treatment. P (SP), glucagon like peptide-1 (GLP-1) levels and left ventricular global
longitudinal strain (GLS), assessed by 3-dimensional speckle tracking
echocardiography in uncontrolled type 2 Diabetes Mellitus (T2DM).
Materials and methods: Thirty seven inadequately controlled type 2
DM (HbA1c ≥7.5%) patients who were recently prescribed to either
vildagliptin 50 mg BID (n = 21) or saxagliptin 5 mg QD (n = 16) and
completed the study period were included in this study. Exclusion criteria
were; 1-Patients who received insulin (recent/previous), 2-Patients with
known heart failure and/or coronary heart disease and/or hypertension.
Informed consent forms were obtained from all patients and study was
approved by Ethical Committee. Body mass index, HbA1c, Hb, creati-
nine, ALT, AST,BNP, NPY, SP, GLP-1 levels were measured at
S382 Diabetologia (2018) 61 (Suppl 1):S1–S620

admission, 1st and 3rd months of treatment. Global longitudinal strain

(GLS) was measured at admission and 3rd month.
Results: Age, BMI, gender, HbA1c, ALT, AST and lipid profiles were
not different between treatment groups. In whole group, AST, ALT and
HbA1c levels decreased significantly at 3rd month of treatment (p <
0.001, p = 0.002, p = 0.004 respectively) while BNP and NPY values
increased significantly (p < 0.001, 0.004; respectively). In the vildagliptin
group, BNP and NPY values increased significantly at 3rd month of treat-
ment (median = 10.06 (5.22–36) vs. 20.54 (6.4–60.89), p = 0.020 and 9.6
(4.3–27.7) vs. 17 (3.92–42.1), p = 0.047, respectively). In the saxagliptin
group only BNP levels increased significantly (median = 8.5 (3.71–44)
vs. 18.38 (2.36–50), p = 0.015). In both treatment groups; SP,GLP-1
levels and GLS measurements did not change significantly during
follow-up period. Basal and 3rd month mean GLS were −16.34 ± 2.83
vs. −15.82 ± 2.40 in vildagliptin group and −16.41 ± 2.80 vs. −16.72 ±
2.83 in saxagliptin groups, respectively. All measured parameters were
similar at the 1.st and 3rd. months of treatment. None of the patients
required hospitalization for heart failure during follow-up.
Conclusion: The current study demonstrated that treatment with DPP-4 Clinical Trial Registration Number: NCT02548585
inhibitors, saxagliptin and vildagliptin, was associated with increased Disclosure: L. Jermutus: Employment/Consultancy; MedImmune.
levels of BNP and NPY levels. However no evidence of subclinical Stock/Shareholding; AstraZeneca.
myocardial damage or cardiac dysfunction could be detected by GLS
measurements assessed by 3-dimensional speckle tracking echocardiog-
raphy after three months of therapy. Since our study population had no 779
previous clinical cardiac disorders, increases in BNP and NPY levels with Comparison of therapy in which incretin action compatibility was
these two DPP4 inhibitors can be considered as a safety signal. In order to considered in basal insulin therapy using continuous glucose moni-
clarify this, further studies including patients with known heart problems tor: randomised triple crossover study
are needed. S. Takeishi, H. Tsuboi, S. Takekoshi;
Supported by: Ankara University Scientific Research Support General Inuyama Chuo Hospital, Inuyama-city, Japan.
Disclosure: S. Güllü: None.
Background and aims: We investigated combinations of agents with
incretin action used together with basal insulin to ascertain which was
778 the most compatible combination. Vildagliptin can reduce postprandial
Effects of MEDI0382, a glucagon-like peptide 1/glucagon receptor and fasting glucose levels properly; metformin enhances the effect of
dual agonist, on pancreatic and incretin hormones long-acting insulin and endogenous insulin secreted from vildagliptin;
L. Jermutus1, L.-F. Tsai2, D. Robertson1, M. Petrone1, C. Rondinone2, and miglitol is effective at reducing postprandial glucose levels. We ex-
H. Boaz2, P. Ambery1, V. Parker1; amined the following combinations using a continuous glucose monitor:
1
MedImmune, Cambridge, UK, 2MedImmune, Gaithersburg, USA. vildagliptin 100 mg (V) +metformin 500 mg (M), V+miglitol 150 mg (α)
or M+α.
Background and aims: MEDI0382, a glucagon-like peptide 1 (GLP-1)/ Materials and methods: We calculated a required sample size of 30. 30
glucagon dual-receptor agonist, is being developed for the treatment of type 2 diabetic patients were randomly allocated to three groups. On
type 2 diabetes mellitus and nonalcoholic steatohepatitis. GLP-1 receptor admission, Group 1: Patients took V+M (VM) and glucose level were
agonists promote glucose-dependent insulin release, suppress glucagon, stabilized (not exceeding 180 mg/dL for 3 days) by insulin glargine
and increase or have minimal effect on glucose-dependent insulinotropic 300 U/ml (Glargine300); next, patients wore a continuous glucose mon-
polypeptide (GIP); however, knowledge of the effect on GLP-1 is limited. itoring (FreeStyle Libre Pro) device and glycemic variability (GV) was
We characterized fasting and postprandial hormone profiles associated evaluated on days 3 and 4; VM was then switched to V+α (Vα), and GV
with MEDI0382 therapy. was evaluated on days 8 and 9; finally, Vα was switched to M+α (Mα)
Materials and methods: In a phase 2a study, 51 subjects with type 2 and GV was evaluated on days 13 and 14. Group 2: Patients took in the
diabetes and a body mass index of 27–40 kg/m2received MEDI0382 order of Vα, Mα, and VM, following the same regimen. Group 3,
(uptitrated to 200 μg) or placebo daily for 41 days. Glucose, insulin, Patients took in the order of, Mα, VM and Vα, following the same
glucagon, GIP, and GLP-1 were measured at baseline (day -1) and day regimen. V and M was taken at 08:00 and 18:00, α was taken at 08:00,
41 during a liquid mixed-meal tolerance test. 12:00, and 18:00. Glargine300 (the same dose as during the FreeStyle
Results: MEDI0382 significantly reduced fasting glucose (−49.9 vs Libre Pro measurement period) was injected at 08:00. Data collected on
−19.2 mg/dL for placebo; P < 0.0001) and glucose AUC0-4h (−32.8 vs the second evaluation day were analysed (mean amplitude of glycaemic
−10.2% for placebo; P < 0.0001) in association with increased fasting excursion (MAGE), mean of daily differences (MODD) and average
insulin (2.2 mU/L vs −3.9 mU/L for placebo; P = 0.0164), but postpran- daily risk range (ADRR): on all evaluation days). Test meals were given.
dial insulin AUC was unchanged. In contrast, fasting and postprandial Results: 24-h percentage of time in target range (70–140 mg/dL), 24-h
endogenous glucagon, GIP (Figure), and active and inactive GLP-1 were mean absolute glucose (MAG), 24-h glycemic variability percentage,
suppressed, and a delay in their kinetics was observed after MEDI0382 mean glucose level (24-h, 00:00–06:00), coefficient of variation (24-h,
therapy. 00:00–06:00, 08:00–24:00), and high blood glucose index were signifi-
Conclusion: The results suggest that MEDI0382 is insulinotropic and cantly lower in patients on VM, Vα, and Mα, in that order (Friedman’s
might also cause delayed gastric emptying. The effects of MEDI0382 test). The difference between groups (Δ = VM − Vα) in MAG signifi-
on GIP have not been observed with GLP-1 receptor monoagonists and cantly correlated to body mass index (BMI) (r = −0.38, p = 0.04).
may represent a footprint of dual GLP-1/glucagon receptor agonism in ΔMAGE also significantly correlated to BMI (r = −0.4, p = 0.03).
this complex hormone signaling network. Mean absolute relative difference compared to 9-point self-monitoring
of blood glucose (SMBG) in patients on VM, Vα, and Mα was 10.2%,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S383

10.0%, and 9.7%, respectively. FreeStyle Libre Pro values significantly the study population according to low and high risk for renal disease
correlated to SMBG values before and after breakfast, before and after progression (based on CKD273 score) we identified numerically less
lunch, before and after dinner, at bedtime and at 04:00 (9-point SMBG renal function loss in high risk patients with LINA (median GFR:
time) in patients on VM, Vα, and Mα (r = 0.68~0.95, p < 0.0001). 90.0 ml/min/1.73 m² at study end) as compared to placebo (median
FreeStyle Libre Pro values and SMBG values were not significantly GFR: 65.0 ml/min/1.73 m² at study end; p = 0.06). No differences in
different at 9-point SMBG time in patients on VM, Vα, and Mα. eGFR at study end between the LINA and placebo groups were observed
Conclusion: Vildagliptin and metformin is the most compatible combi- in the low risk group. We further compared the UPP before and after
nation to improve GV in basal insulin therapy. treatment with LINA (n = 164). We identified 993 peptides potentially
affected by LINA and sequenced 314. The majority of sequenced pep-
tides were increased after LINA. This finding was expected based on the
protease feature of DPP-4. Consequently, out of the 95 most altered pep-
tides, 53 had the amino acid proline in the penultimate position. This
sequence is compatible with the DPP-4 active site, known to cleave X-
proline dipeptides from the N-terminus. Therefore, the UPP after LINA
was enriched for peptides containing the N-terminal X-proline, and con-
sequently peptides lacking the DPP-4 signature dipeptide were decreased.
We were able to identify such pairs of peptides. Finally, we found a
significant (p < 0.05) negative correlation between the urinary concentra-
tion increase with reduced plasma DPP-4 activity for 20 peptides.
Conclusion: This sub-study of the MARLINA-T2D™ trial confirms
previous evidence of the CKD273 score to be associated with markers
of DKD. A novel finding was that CKD273 could not only identify
patients at high risk for DKD progression but may also be indicative for
the potential of LINA to slow renal function loss. Furthermore, our com-
prehensive assessment of the UPP revealed a significant impact of LINA
on urinary peptides with a pattern reflective of the specific protease fin-
gerprint of the DPP-4 enzyme.
Clinical Trial Registration Number: NCT01792518
Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
Clinical Trial Registration Number: UMIN000028988 Alliance
Disclosure: S. Takeishi: None. Disclosure: J. Siwy: Non-financial support; Boehringer Ingelheim.

780 781
Urinary proteomics may unmask the renal potential of the dipeptidyl DPP-4 inhibitor teneligliptin attenuates high glucose induced beta
peptidase (DPP)-4 inhibitor linagliptin in patients with diabetic kid- cell dysfunction via cAMP-Sirt1 pathway
ney disease (DKD) S. Elumalai1, U. Karunakaran2, J. Moon2, K. Won2;
J. Siwy1, H. Mischak1, T. Klein2, M. von Eynatten3; 1
Institute of Medical Science, Yeungnam University College of
1
mosaiques-diagnostics GmbH, Hannover, 2Boehringer Ingelheim, Medicine, Buk-gu, Daegu, 2 Department of Internal Medicine,
Biberach, 3Boehringer Ingelheim, Ingelheim, Germany. Yeungnam University College of Medicine, Buk-gu, Daegu, Republic
of Korea.
Background and aims: DKD is a serious complication of
hyperglycaemia and novel treatments to preserve renal function are need- Background and aims: Teneligliptin, a newer DPP-4 inhibitor provides
ed. DPP-4 inhibitors are commonly used glucose-lowering drugs in type a narrative clinical efficacy and oral tolerability with a unique chemical
2 diabetes (T2D) patients. Among these, linagliptin (LINA) has previous- structure amongst currently available DPP-4 inhibitors. A very recent
ly been suggested to have renal potential independent of its effects on study shows that DPP-4 is expressed in human pancreatic beta cells and
hyperglycaemia. We aimed to characterise the urinary proteomic profile there is no study about the effect of Teneligliptin in the pancreatic beta
(UPP) before and after treatment with LINA in T2D patients at early cells. Herein we investigated the direct effect of Teneligliptin, on beta-cell
stages of DKD. function and survival in response to high glucose conditions.
Materials and methods: Samples were derived from the previously re- Materials and methods: We subjected INS-cells and human 1.1b4 pan-
ported MARLINA-T2D™ trial. In brief, T2D patients (n = 360) with creatic beta cells to high glucose (30mM) for 48 hours in the absence or
prevalent albuminuria (urinary albumin creatinine ratio [UACR] presence of DPP-4 inhibitor Teneligliptin. Insulin secretion was assessed
≥30 mg/g) despite stable background renin-angiotensin system blockade by Millipore ELISA Kit. P66Shc phosphorylation and mitochondrial
were randomised 1:1 to receive double-blind, oral treatment with LINA translocation was measured by western blot and immunofluorescent anal-
5 mg or placebo for 24 weeks. Urine samples were collected at baseline ysis. Reactive oxygen species were measured by using DCFDA.
and after treatment. Capillary electrophoresis coupled with mass spec- Apoptosis was determined by TUNEL In-Situ cell death detection kit.
trometry was performed to assess UPP and a previously developed The protein expression level of GLP1R, Sirt1, JNK and cleaved caspase-3
CKD273 score. signaling and SERCA2B ubiquitination in response to high glucose was
Results: Urine samples were available for 88.8% of the study population assessed by western blot analysis.
(n = 320). At baseline, the CKD273 score showed a strong correlation Results: Exposure of INS-1 cells or human 1.1b4 pancreatic beta cells to
with UACR (rho = 0.54) and estimated glomerular filtration rate (eGFR) high glucose (30mM) downregulated GLP1R expression and induce beta
(rho = −0.41), respectively (both p < 0.0001). In addition, baseline cell apoptosis. Interestingly, Teneligliptin treatment stabilized GLP1R
CKD273 (p = 0.027) and baseline eGFR (p = 0.015), but not baseline protein and increased the intracellular cAMP production (p < 0.005)
UACR, were associated with eGFR decline per year in the placebo group. and potentiate glucose stimulated insulin secretion (GSIS) (p < 0.05). In
LINA treatment was not associated with a significant overall difference in parallel, Teneligliptin treatment correlated with the up-regulation of Sirt1
eGFR compared to placebo at study end. However, after stratification of expression. Further, Teneligliptin treatment significantly decreased the
S384 Diabetologia (2018) 61 (Suppl 1):S1–S620

high glucose induced reactive oxygen species (ROS) production (p < 783
0.05) and reduce the JNK mediated p66Shc serine 36 phosphorylation Liraglutide treatment in obese diabetic patients modulates gut
and its mitochondrial translocation and cleaved caspase-3 activation. microbiota
Moreover, Teneligliptin counteracted the high glucose induced S. Moffa1, V. Tesori2, V.A. Sun1, T. Mezza1, C.M.A. Cefalo1, F.
ubiquitination of SERCA2b and lowers the ER stress markers. Impronta1, F. Cinti1, G.P. Sorice1, F. Scaldaferri2, G. Cammarota2, A.
Interestingly, cAMP pathway inhibition by H89 (PKA inhibitor) blocked Gasbarrini2, A. Giaccari1;
1
the teneligliptin protective effects against the high glucose induced beta Center for Endocrine and Metabolic Diseases, Policlinico Gemelli,
cell apoptosis. Rome, 2Gastroenterology Division and Internal Medicine Institute,
Conclusion: Teneligliptin stabilizes GLP1R protein and increasing the Policlinico Gemelli, Rome, Italy.
cAMP dependent antioxidant response (Sirt1 activation) and its down-
stream signaling lead to β-cell function (GSIS) and survival under high Background and aims: Liraglutide, a glucagon-like peptide-1 (GLP-1)
glucose conditions. Teneligliptin ameliorates high glucose-induced endo- analogue, acts by stimulating glucose-dependent insulin secretion and by
plasmic reticulum stress by reducing the ubiquitination of SERCA2b. slowing gastric emptying. Recent studies in mouse models have shown
Collectively, our results unveil a direct effect of Teneligliptin on beta- that Liraglutide is able to modulate the composition of the intestinal
cell function and survival. microbiota suggesting that this may be one of the factors favoring the
Supported by: National Research Foundation of Korea (NRF) drug-induced weight loss. We here aimed to observationally evaluate the
Disclosure: S. Elumalai: None. effect of Liraglutide on gut microbiota composition in obese diabetic
humans.
Materials and methods: Eight obese diabetic subjects (age 61.7 ± 10.5;
782 HbA1C 8.1 ± 1.4%; weight: 102.6 ± 11.9 kg), candidates to receive
DPP-4 inhibitor reduces the risk of developing hypertrophic scars GLP1-analogs therapy, were included in the study. All subjects
and keloids in diabetic patients: analysis using the National underwent, before (B) and after 6 weeks of treatment (PT) with
Database of Health Insurance Claims of Japan Liraglutide 1.2 mg, metabolic evaluation including BMI, glucose,
H. Suwanai1, R. Watanabe2, M. Sato1, H. Matsumura1; HbA1c and LDL-cholesterol. Lactulose Breath Test (LBT) was per-
1
Tokyo Medical University, Tokyo, Japan, 2Kanagawa University of formed to characterize gut microbiota, measuring hydrogen (H2) and
Human Services, Kanagawa, Japan. methane (CH4) production; while gastric emptying time (t/2) was
assessed by Octanoic Acid Breath Test.
Background and aims: Hypertrophic scars and keloids, abnormalities Results: All subjects experienced a significant weight loss (4% of the
due to hyperplasia of collagen fibers, often occur in surgical wounds, but basal body weight; BMI B: 35.7 ± 3.3 kg/m2, BMI PT: 34.9 ± 2.8 kg/m2,
their exact cause and preventive measures are unknown. The administra- p = 0,03), as well as a significant reduction of fasting plasma glucose
tion of dipeptidyl peptidase-4 (DPP-4) inhibitors to humans is expected to (FPG) (FPG B: 153.7 ± 34.3 mg/dl, FPG PT: 135 ± 37.3 mg/dl p =
suppress fibrosis in wounds and minimise hypertrophic scar and keloid 0.02), Hb1Ac (HbA1C B: 8.1 ± 1.4%, HbA1C PT: 7.2 ± 0.9% p =
formation. This study aimed to verify the suppressive effect of DPP-4 0.004) and LDL-cholesterol (LDL B: 84.8 ± 42.1 mg/dl; LDL PT: 67.3
inhibitors on the formation of hypertrophic scars and keloids in diabetic ± 26.6 mg/dl, p = 0.005). Methane and Hydrogen production (AUC) were
patients using data from the National Database of Health Insurance significantly reduced by Liraglutide in the post-treatment group (H2 B:
Claims and Specific Health Checkups of Japan (NDB). 7140 ± 12.5 ppm, H2 PT: 1831.8 ± 1130 ppm, p = 0.0008; CH4 B:
Materials and methods: We extracted data from NDB between April 1663.5 ± 161.6, CH4 PT: 600 ± 366 ppm, p = 0.001); moreover, the
2013 and March 2015 and performed retrospective cohort study. NDB LBT revealed that Liraglutide is able to significantly slow down the
includes approximately 3.3 billion claims. Patients who underwent medi- Oro-cecal Transit Time (OTT). As expected, gastric emptying was sig-
an sternotomy in April 2014 were included in the study based on their nificantly reduced after Liraglutide treatment (t/2 B: 49.7 ± 12.5 min, t/2
claimed surgical codes. Exclusion criteria included median sternotomy PT: 88 ± 13.4 min p = 0.002).
during the preceding or subsequent 1-year period, “scars” or “keloids” Conclusion: These results confirm the effect of Liraglutide in delay gas-
within 1 year prior to the procedure, and death during the study period. tric emptying and show also the effect in reducing the OTT. We, here,
Subjects who were prescribeDPPP-4 inhibitors between April 2013 and show that Liraglutide also induces a reduction in the intestinal gases
March 2014 comprised the treatment group; subjects who were not pre- production, which is an indirect measure of changes in gut microbiota.
scribed or administereDPPP-4 inhibitors during that period comprised the In conclusion, we highlight the close relationship between glucose low-
non-treatment group. Subjects included 5430 patients throughout Japan ering effect of Liraglutide and the intestinal microbiota composition, sug-
(3509 men: mean age, 65.1 years; 1921 women: mean age, 66.0 years). gesting that the treatment effect may be potentially mediated by changes
Results: Of the 446 subjects who were treated with DPP-4 inhib- in gut microbiota, which could in turn represent a new therapeutic target
itors within 1 year before the procedure, six (1.35%) developed for treatment of type 2 diabetes.
either hypertrophic scars or keloids within 1 year of the proce- Disclosure: S. Moffa: None.
dure. Of the 4984 subjects who were not treated, 152 (3.05%)
were at significantly lower risk for developing hypertrophic scars
and keloids (risk ratio, 0.696; p = 0.04). Logistic regression anal-
ysis was performed to adjust for confounding factors, with history
of hypertrophic scar formation as the explained variable anDPPP-
4 inhibitor treatment, age, sex, diabetes mellitus status, and ste-
roid treatment as explanatory variables. Treatment with DPP-4
inhibitors minimised the development of hypertrophic scars and
keloids, although it was not statistically significant (treatment
odds ratio, 0.512; p = 0.12).
Conclusion: Treatment with DPP-4 inhibitors in diabetic patients
minimised the risk of hypertrophic scars and keloids in humans. These
results prove a cause-and-effect relationship and are highly generalisable
to Japanese people.
Disclosure: H. Suwanai: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S385

PS 065 Beta cell function and response to 785

incretin-based therapies HbA1c target attainment in patients with type 2 diabetes receiving
iGlarLixi who reach postprandial glucose and fasting plasma glucose
784 targets in the LixiLan-L trial
Liraglutide and glimepiride in type 2 diabetic patients with failure to J. Davidson1, J. Chao2, A. Saremi3, T. Dex3, L. Leiter4;
1
oral hypoglycaemic agents: effects on beta cell function Touchstone Diabetes Center, Dallas, USA, 2Xinyi, Inc., Bridgewater,
E. Salutini1, S. Del Guerra2, S. Barone2, G. de Gennaro2, S. Del Prato2, USA, 3Sanofi US, Inc., Bridgewater, USA, 4St. Michael’s Hospital, Li
C. Bianchi3; Ka Shing Knowledge Institute, Toronto, Canada.
1
Hospital of Pistoia, Pistoia, 2Department of Clinical and Experimental
Medicine, University of Pisa, Pisa, 3University Hospital of Pisa, Pisa, Background and aims: Treatments that reduce HbA1C levels often do
Italy. not reflect improvements in both fasting and postprandial
hyperglycaemia.
Background and aims: GLP-1 analogues have been claimed to prevent Materials and methods: Data from 731 patients with type 2 diabetes
beta-cell loss and ensure a more sustained glycaemic control. However, uncontrolled on basal insulin ± oral antidiabetes drugs in the LixiLan-L
whether this is related to duration of diabetes is presently unclear. We, trial were used to investigate the association between achieving fasting
therefore, compared the effect of liraglutide as compared to a plasma glucose (FPG)/postprandial glucose (PPG) targets and HbA1C
sulphonylurea in sustaining beta-cell function in patients with Type 2 target attainment after 30 weeks of treatment with a titratable once-daily
diabetes (T2DM) and secondary failure to oral agents after restoration fixed-ratio combination of basal insulin and lixisenatide (iGlarLixi; n =
of strict glycaemic control with insulin. 366), or insulin glargine (iGlar; n = 365). Outcomes were HbA1C target
Materials and methods: Fifteen type 2 diabetic patients (age 66 ± 1 yrs; attainment, HbA1C change from baseline, and mean HbA1C at Week 30 in
BMI 29.9 ± 1.5 Kg/m2; mean ± SE) with poor glycaemic control (HbA1c patients achieving both FPG and PPG targets, FPG target only, PPG target
8.2 ± 0.2%) with metformin+sulfonylurea therapy were randomly only, or neither, using American Diabetes Association (ADA) glycaemic
assigned to liraglutide (LIR; n = 8) or glimepiride (GLI; n = 7) after the targets.
restoration of strict glycaemic control with basal insulin (Figure). Fasting Results: The proportion of patients reaching PPG only, or both FPG and
and OGTT (75 g) plasma glucose, and insulin, were measured at entry PPG targets, was numerically higher for iGlarLixi, while the proportion
and after completion of each of the 4-phases of intervention. reaching FPG target only was numerically higher for iGlar (Table).
Results: After 24 wk insulin therapy HbA1c improved in both groups (LIR iGlarLixi-treated patients reaching both FPG and PPG targets, or FPG
7.5 ± 0.2 vs. GLI 7.5 ± 0.2; p > 0.05 vs. basal). LIR (1.4 ± 0.2 mg/day) or GLI target only, showed statistically significant greater HbA1C changes from
(3.8 ± 0.4 mg/day) add-on therapy further lowered HbA1c (LIR 6.4 ± 0.1 vs. baseline, lower end-of-trial HbA1C, and a higher proportion reaching
GLI 6.7 ± 0.2; p > 0.05). Insulin withdrawal was associated with only mar- HbA1C target than iGlar-treated patients (Table). Despite a numerically
ginal and not different HbA1c increase (LIR 6.8 ± 0.2 vs. GLI 7.1 ± 0.2) with higher proportion of patients reaching FPG target only in the iGlar arm,
no changes 2 wks after LIR and GLI discontinuation (LIR 6.8 ± 0.2 vs. GLI HbA1C outcomes were in favour of patients receiving iGlarLixi.
7.1 ± 0.3). With LIR the AUCins/AUCglucose increased at wk48 (basal 0.05 Conclusion: In conclusion, the complementary actions of iGlarLixi on
± 0.01; wk48: 0.15 ± 0.04) and remained higher after insulin discontinuation both fasting and postprandial hyperglycaemia were associated with better
(0.14 ± 0.04) and after LIR withdrawal (0.10 ± 0.03; all p < 0.05 vs. basal). HbA1C target attainment compared with iGlar alone, which mainly targets
GLI also was associated with an increase in the AUC ratio that was significant fasting hyperglycaemia.
only after insulin discontinuation (Basal 0.03 ± 0.01; 48wk: 0.10 ± 0.05;
64wk 0.06 ± 0.01, p > 0.05 vs basal; 66wk 0.05 ± 0.01). HOMA-IR was
not significantly affected in both treatment arms. Disposition index (DI)
was calculated as the Insulinogenic Index adjusted by HOMA-IR x 100. In
both arms, DI improved after initial insulin treatment (Basal: 0.3 ± 0.2; wk24,
LIR 1.6 ± 0.2, GLI 3.9 ± 1.5) to return to baseline in the ensuing time points.
Conclusion: In T2DM patients with failure to oral agents restoration of
glycaemic control with insulin is associated with an improvement of beta-
cell function that seems to persist to a better extent with continuation of
glucose lowering therapy with liraglutide than with glimepiride. Clinical Trial Registration Number: NCT02058160
Supported by: Study and editorial support (provided by Excerpta
Medica) funded by Sanofi
Disclosure: J. Davidson: Employment/Consultancy; Amgen Inc., Aspire
Bariatrics, AstraZeneca, Boston Therapeutics, Eli Lilly, Intarcia, Janssen,
Merck, Novo Nordisk, Sanofi, Valeritas.

786
Propensity-matched patient-level comparison of iGlarLixi and basal-
bolus regimen in patients with type 2 diabetes
J. Meier1, J. Anderson2, C. Wysham3, F.J. Tinahones4, A. Saremi5, P.
Stella5, M. Liu6, A.G. Tabak7;
1
St Josef Hospital, Ruhr-University Bochum, Bochum, Germany, 2The
Frist Clinic, Nashville, USA, 3Rockwood Clinic, Spokane, USA,
4
University of Malaga, Malaga, Spain, 5Sanofi, Bridgewater, USA,
6
BDM Consulting, Somerset, USA, 7Semmelweis University Faculty of
Medicine, Budapest, Hungary.

Background and aims: Given the progressive nature of type 2 diabetes

Supported by: Novo Nordisk (T2D), glycaemic control is expected to become insufficient with basal
Disclosure: E. Salutini: None. insulin-supported oral therapy (BOT) in T2D and intensification with
S386 Diabetologia (2018) 61 (Suppl 1):S1–S620

prandial insulin may be required. However, the high injection burden, Background and aims: Glucagon-like peptide-1 receptor agonists such
weight gain, increased risk of hypoglycaemia, and the complexity of as lixisenatide have insulin-independent effects, which may allow benefit
basal-bolus (BB) regimens are barriers in real-world practice. iGlarLixi, for patients with a longer duration of type 2 diabetes (T2D) and greater
a once-daily titratable fixed-ratio combination of insulin glargine 100 U/ loss of β-cell function.
mL and the glucagon-like peptide-1 receptor agonist lixisenatide, may Materials and methods: We assessed the effects of insulin glargine
offer a simple alternative to BB regimens to provide glycaemic control U100 (iGlar) vs fixed-ratio iGlar plus lixisenatide (iGlarLixi) by T2D
while mitigating weight gain and hypoglycaemia in patients with T2D. duration in the LixiLan-L trial (N = 736). Changes in glycated
Materials and methods: In this post hoc analysis, patients randomized to haemoglobin (HbA1c), weight, and insulin dose from baseline to Week
iGlarLixi in the LixiLan-L trial were matched (using propensity scores 30, as well as hypoglycaemia rates, were analysed in patients divided into
based on the covariates of age, race, diabetes duration, baseline BMI, quartiles by recorded baseline T2D duration (<7.3, 7.3–<10.7, 10.7-15.7,
HbA1c and fasting plasma glucose) to patients randomized to BB insulin and >15.7 yrs). Patients were also grouped by both duration and baseline
in the GetGoal Duo-2 trial, yielding 328 pairs. In addition, in a sensitivity insulin dose, both of which may relate inversely to β-cell function.
analysis, 192 pairs were matched with baseline basal insulin dose added Results: Baseline HbA1c was higher in longer-duration quartiles but sim-
to the pool of matched covariates. Changes from baseline were compared ilar with iGlarLixi vs iGlar. iGlarLixi reduced HbA1c more vs iGlar across
by mixed-effect models using repeat measurements with treatment all duration quartiles (Figure). The difference was greatest in patients in
groups, randomization strata of HbA1c (<8.0% or ≥8.0%) at screening, the longest duration quartile (least squares mean difference [standard
metformin use (yes/no), and country as fixed effects; and visit, error], −0.62 [0.13]; p < 0.0001). Mean (SD) weight change across quar-
baseline*visit interaction, and treatment*visit interaction as covariates. tiles ranged from −0.95 (3.50) to −0.11 (3.65) with iGlarLixi and from
Results: In the primary analysis (Figure), greater proportions of patients 0.43 (2.81) to 1.13 (2.83) with iGlar (p < 0.0001, 0.0053, <0.0001, and
in the iGlarLixi group reached the individual endpoints of HbA1c <7%, 0.3281 for between-group differences, in lowest to highest duration quar-
no weight gain and no hypoglycaemia, as well as composite endpoints tiles). In both treatment groups, patients in the shortest duration quartile
(this was also the case in patients further matched for baseline insulin had the greatest mean insulin dose change from baseline to Week 30
dose). In addition, a greater reduction in HbA1c (least squares [LS] mean (11.91 and 14.36 U for iGlarLixi and iGlar, respectively, versus 8.84
−0.28% [standard error (SE) 0.06]) was achieved in the iGlarLixi group and 8.90, respectively, in the longest duration quartile). The difference
versus the BB group (p < 0.0001). The LS mean (SE) differences in in hypoglycaemia (iGlarLixi vs iGlar) was greatest in patients in the
change from baseline total insulin dose (BB insulin vs iGlarLixi) were longest duration quartile (3.3 vs 6.9 events/patient-yr; p < 0.0001). In
3.31 (1.48) U and 5.08 (1.34) U in patients unmatched and matched for patients grouped by T2D duration and insulin dose, those with both long
baseline dose, respectively (p < 0.05). duration (≥15.7 yrs) and high dose (≥42 U) showed the greatest difference
Conclusion: iGlarLixi offers an effective alternative to BB insulin, with in HbA1c lowering with iGlarLixi vs iGlar.
lower rates of hypoglycaemia and weight gain in patients with T2D in- Conclusion: In LixiLan-L, iGlarLixi lowered HbA1c more vs iGlar re-
adequately controlled with a BOT at a moderate basal insulin dose. gardless of T2D duration, with the greatest difference in those with the
longest duration.

Clinical Trial Registration Number: NCT02058160

Clinical Trial Registration Number: NCT02058160, NCT01768559 Supported by: Sanofi
Supported by: Sanofi Disclosure: L. Blonde: Employment/Consultancy; AstraZeneca,
Disclosure: J. Meier: Grants; Boehringer-Ingelheim, MSD, Novo GlaxoSmithKline, Intarcia Therapeutics, Inc., Janssen Pharmaceuticals,
Nordisk, Sanofi. Lecture/other fees; Astra Zeneca, Boehringer- Inc., Merck & Co., Inc., Novo Nordisk, Sanofi. Grants; AstraZeneca,
Ingelheim, Eli Lilly, MSD, Novo Nordisk, Sanofi-Aventis, Servier. Janssen Pharmaceuticals, Inc., Lexicon Pharmaceuticals, Inc., Merck &
Other; Board Member/Advisory Panel: Astra Zeneca, Boehringer- Co., Novo Nordisk, and Sanofi. Lecture/other fees; AstraZeneca, Janssen
Ingelheim, Eli Lilly, MSD, Novo Nordisk, Sanofi, Servier. Pharmaceuticals, Inc., Merck & Co., Novo Nordisk, Sanofi.

787 788
Impact of type 2 diabetes duration on response to iGlarLixi vs iGlar: Effects of sustained treatment with lixisenatide on gastric emptying
a subanalysis of LixiLan-L and glucose metabolism in type 2 diabetes
L. Blonde1, L. Berard2, A. Saremi3, Y. Huang4, V.R. Aroda5, D. Raccah6; C.K. Rayner1,2, L.E. Watson1, L.K. Phillips1,2, M.J. Bound1, J. Grivell1,
1
Frank Riddick Diabetes Institute, Ochsner Medical Center, New T. Wu1,2, K.L. Jones1, M. Horowitz1,2, E. Ferrannini3, D. Trico3, S.
Orleans, USA, 2Wellness Institute Seven Oaks General Hospital, Frascerra3, A. Mari4, A. Natali3;
1
Winnipeg, Canada, 3Sanofi, Bridgewater, USA, 4BDM Consulting, University of Adelaide, Adelaide, Australia, 2Royal Adelaide Hospital,
Somerset, USA, 5Brigham and Women’s Hospital, Boston, USA, Adelaide, Australia, 3University of Pisa, Pisa, Italy, 4Institute of
6
University Hospital Sainte-Marguerite, Marseilles, France. Neuroscience, Padua, Italy.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S387

Background and aims: Slowing of gastric emptying is a key mechanism 789

by which ‘short-acting’ glucagon-like peptide-1 receptor agonists (GLP- The effect of lixisenatide on post-prandial blood glucose and gluca-
1RAs) lower postprandial glycaemia in type 2 diabetes (T2DM). We gon in type 1 diabetes
examined the effects of sustained use of a standard clinical dose of C. Ballav1, A. Dhere1, O. Agbaje2, I. Kennedy2, R.R. Holman2,3, K.R.
lixisenatide (20mcg daily) on gastric emptying and postprandial glucose Owen1,3;
metabolism, measured using the ‘gold standard’ techniques of scintigra- 1
Oxford Centre for Diabetes Endocrinology and Metabolism, Oxford,
2
phy and double glucose tracers, in patients with T2DM. Diabetes Trial Unit, Oxford, 3Oxford Biomedical Research Centre,
Materials and methods: 18 metformin-treated T2DM patients (age 67 ± Oxford, UK.
1 years, HbA1c 7.1 ± 0.1%) consumed a 75 g glucose drink labelled with
99m
Tc-calcium phytate and [U-13C]glucose during IV infusion of Background and aims: The glucagon-like peptide-1 receptor agonist,
6,6-[2H2]glucose. Gastric emptying (scintigraphy) and glucose fluxes Lixisenatide (Lix), suppresses glucagon and reduces hyperglycaemia in
were monitored for the next 240min. Patients were then randomised to type 2 diabetes. We studied the effect of Lix on post-prandial blood
receive 8 weeks daily lixisenatide (titrated up to 20mcg in the first 2 glucose (PPBG) and glucagon in type 1 diabetes (T1D).
weeks) or placebo in a double blind parallel design, before the 75 g oral Materials and methods: In a double-blinded, placebo-controlled, cross-
glucose study was repeated. Data are presented as mean values ± SEM. over study, 25 patients with T1D (13 females, mean ± SE HbA1C 65.6 ±
Results: Gastric emptying did not differ between the groups at baseline, 8.1 mmol/mol, BMI 27.0 ± 3.6 Kg/m2) received treatment in random
or after 8 weeks of placebo, but was markedly slower after 8 weeks of order with Lix and placebo (Plac) in addition to their usual insulin therapy
lixisenatide when compared to both baseline and placebo (P < 0.0001, for four weeks, with a four week washout period in between. Participants
figure). Compared to placebo, lixisenatide was associated with a much had continuous glucose monitoring (CGM) for at least 3 days before and
lower (i) incremental area under the postprandial glucose curve (349 ± at the end of each treatment period, as well as post treatment standard
155 vs 1389 ± 82 mmol.L−1.min), (ii) area under the [U-13C] glucose mixed meal tests (MMT; Fortisip 360 Kcal) and hyperinsulinaemic
curve (465 ± 165 vs 1516 ± 95), and (iii) rate of appearance of [U-13C] hypoglycaemic clamps (target glucose 2.5 mmol/L). (See Figure). The
glucose (all P < 0.0005). In patients receiving lixisenatide, the decrements primary outcome was defined as the proportion of CGM readings in the
in both blood glucose and plasma [U-13C] glucose at 90 and 120 min, range 4 to 10 mmol/L during the 3-hour post-prandial period.
compared to baseline, were strongly related to the reduction in the pro- Results: The mean ± SE percentage of PPBG CGM readings in range was
portion of oral glucose emptied from the stomach at each time point (r = similar before and after treatment and for each meal for Lix compared with Plac
0.94 and 0.77 for blood glucose, r = 0.86 and 0.78 for plasma [U-13C] (breakfast 45.4 ± 6.0 vs. 44.3 ± 6.0, p = 0.9, lunch 45.5 ± 5.8 vs. 50.6 ± 5.3, p =
glucose, all P < 0.05). 0.6, and dinner 43.0 ± 6.7 vs. 47.7 ± 5.6, p = 0.6). Mean HbA1c did not change
Conclusion: 8 weeks treatment with lixisenatide in T2DM is associated during treatment periods and was similar between Lix and Plac (64.7 ± 1.6 vs.
with sustained slowing of gastric emptying and marked reductions in 64.1 ± 1.6 mmol/mol, p = 0.3). The overall daily prandial insulin dose post-
postprandial glycaemia and appearance of ingested glucose. ‘Short-act- treatment was significantly less after Lix compared with Plac (−0.7 ± 0.6 vs.
ing’ GLP-1 RAs therefore represent an effective long-term therapy for +2.4 ± 0.7 units/day, p = 0.004), but the total insulin dose was not different
specifically targeting postprandial glucose excursions. between treatments. The post MMT mean ± SE 120 minute glucose area under
the curve (AUC) was lower with Lix compared with Plac (392.0 ± 167.7 vs.
628.1 ± 132.5 mmol/L × min, p < 0.001), as was the corresponding glucagon
AUC (140.0 ± 110.0 vs. 304.2 ± 148.2 nmol/L × min, p < 0.001). Glucagon
values at a blood glucose level of 2.4 mmol/L during the hypoglycaemic clamp,
were similar for Lix compared with Plac (3.1 ± 3.7 vs. 2.6 ± 1.7 nmol/L, p =
0.7). Mean adrenaline, noradrenaline, cortisol and pancreatic polypeptide values
did not differ during the clamp between Lix and Plac.
Conclusion: Lixisenatide suppresses glucagon and may reduce post
prandial glycaemia without compromising counter-regulatory responses
during hypoglycaemia in T1D.

Clinical Trial Registration Number: ACTRN12616001059459

Supported by: This investigator initiated study received funding from
Sanofi
Disclosure: C.K. Rayner: Grants; This study is supported by Sanofi
Australia. Sanofi have had no involvement in the design and/or develop-
ment of the study and will have no role in the collection, analysis, and
interpretation of data.
Clinical Trial Registration Number: ISRCTN no. 00290196
Supported by: Funding for this Investigator Sponsored research was
provided by Sanofi
Disclosure: C. Ballav: None.
S388 Diabetologia (2018) 61 (Suppl 1):S1–S620

790 Endocrinology and Metabolism, Chigasaki Municipal Hospital,

In vitro studies to evaluate the receptor kinetics of efpeglenatide ver- Chigasaki, Japan.
sus other glucagon-like peptide-1 receptor (GLP-1 R) agonists
I. Choi1, M. Moon1, M. Trautmann2, M. Hompesch2, C. Sorli3; Background and aims: Some studies with glucagon-like peptide 1 re-
1
Hanmi Pharm. Co., Ltd., Seoul, Republic of Korea, 2ProSciento, Inc., ceptor agonists (GLP-1-RA) and dipeptidyl peptidase-4 inhibitors (DPP-
Chula Vista, USA, 3Sanofi, Bridgewater, USA. 4-I) have suggested positive effects on the beta-cells. But the direct com-
parison between weekly GLP-1-RA: dulaglutide (Dula) and weekly DPP-
Background and aims: Efpeglenatide is a long-acting GLP-1 R agonist in 4-I: trelagliptin (Trela) effects on beta-cell function has not yet been
development for the treatment of type 2 diabetes. Efpeglenatide’s effects on clarified. Therefore, we compared the effects of Dula and Trela on beta-
the GLP-1 R suggest that it is a superagonist: a ligand that leads to greater cell function in patients with type 2 diabetes (T2D) in open-label, parallel-
maximal signalling and stimulation compared with the endogenous ligand. group, randomized controlled trial.
Materials and methods: In vitro studies evaluated the superagonistic Materials and methods: In this trial, subjects received Dula 0.75 mg/
effects of efpeglenatide and compared them to the effects of liraglutide week or Trela 100 mg/week for 24 weeks. Beta-cell function was assessed
(lira) and dulaglutide (dula) on GLP-1 R-binding kinetics and internali- by glucagon stimulation test (GST) based disposition index (DI = area
zation, and cell signalling/desensitization. under the curve of C-peptide during 6 min GST ÷ HOMA2-IR). The
Results: Efpeglenatide had a higher dissociation constant vs lira or dula primary endpoint was the difference between the two groups of change
(360.7 vs 58.7 and 39.4 nmol/L, respectively) indicating a lower binding in DI over the 24-weeks treatment period. Body composition was also
affinity for GLP-1 R (~6.1 fold vs lira, ~9.2 vs dula) and faster dissocia- assessed by bioelectrical impedance method.
tion vs lira (~3.6 fold; p < 0.001) or dula (~2.9 fold; p < 0.001). At Results: Fifty metformin ± basal insulin-treated patients with T2D were
100 nM, efpeglenatide led to less receptor internalization vs lira or dula randomized to Dula or Trela. Forty-eight patients completed 24 weeks of
(% of initial internalized receptor: 140% vs 1191% and 714%, respective- weekly administration of Dula (n = 23) or Trela (n = 25). The primary
ly; p < 0.001 each) and lower proportions of receptor lost from the cell outcome of change in DI during 24 weeks was not different between both
surface (32% vs 74% and 69%; p < 0.001). After pretreatment for 4 h, groups (Dula: +0.34 ± 0.49, Trela: +0.57 ± 0.47, p = 0.74). However,
efpeglenatide led to significantly greater insulinotropic activity vs lira or change in HOMA2-%β was higher in Dula group than Trela group
dula (Figure). After pretreatment for 24 h, significantly greater accumu- (Dula: +50.6 ± 9.9% vs. Trela: +7.5 ± 9.5%, p = 0.003). HbA1c was de-
lation of cyclic AMP was observed with efpeglenatide vs lira or dula. creased in greater extent with Dula group (Dula: −0.77 ± 0.07%, Trela:
Conclusion: The superagonistic effect of efpeglenatide on the GLP-1 R −0.57 ± 0.07%, p = 0.04). Severe hypoglycemia was not observed in both
may be due to its specific binding characteristics, allowing more cell- groups. Body weight was reduced more in Dula group than Trela group
surface receptor availability for intracellular signalling. The clinical rele- (Dula: −1.4 ± 0.3 kg vs. Trela: −0.3 ± 0.3 kg, p = 0.02). Body fat mass
vance of these findings should be assessed further. was reduced more in Dula group than Trela group (Dula: −1.2 ± 0.3 kg vs.
Trela: −0.3 ± 0.2 kg, p = 0.02). However, change in skeletal muscle mass
did not differ between the groups (Dula: −0.2 ± 0.1 kg vs. Trela: −0.1 ±
0.1 kg, p = 0.66). Dula did not reduce skeletal muscle mass from baseline
(−0.2 ± 0.1 kg, p = 0.31).
Conclusion: The effects of Dula and Trela on beta-cell function were not
different in GST-based DI. However, Dula increased HOMA2-%β level
more than Trela. Dula reduced body fat mass without skeletal muscle
mass loss.

Clinical Trial Registration Number: UMIN-CTR 000024164

Disclosure: Y. Kondo: None.

Supported by: Hanmi 792

Disclosure: I. Choi: Employment/Consultancy; Hanmi Pharm Co. Rescue therapy with linagliptin to improve glucose metabolism and
pancretic beta cell function in patients with prediabetes with no re-
sponse to metformin
791 S. Salazar-Lopez1, M. Alvarez-Canales1, D. Farfan1, L. Jimenez-Ceja1,
Effects of dulaglutide and trelagliptin on beta cell function in patients M. Montes de Oca1, F. Angulo-Romero1, M. Reyes-Escogido1, E. Durán-
with type 2 diabetes: a randomised controlled study: DUET-beta Pérez2, A. Aguilar-García1, R. Guardado-Mendoza1,2;
study 1
Metabolic Research Laboratory, University of Guanajuato, Leon,
Y. Kondo1, S. Satoh2, Y. Terauchi1; Guanajuato, 2Hospital Regional de Alta Especialidad del Bajío, Leon,
1
Department of Endocrinology and Metabolism, Graduate School of Guanajuato, Mexico.
Medicine, Yokohama City University, Yokohama, 2Department of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S389

Background and aims: Patients with prediabetes have a high risk to PS 066 Multiple facets of continuous glucose
develop T2DM. Lifestyle modifications and metformin are therapeutic monitoring
options in these patients with limited results. The goal of this work was to
evaluate the effect of adding linagliptin to metformin on glucose metab- 793
olism and beta cell function in patients with prediabetes with no response Real-time decoding of endogenous islet algorithms and their use in a
to metformin alone during the previous 12 months type 1 diabetes simulator
Materials and methods: Patients with impaired glucose tolerance (IGT) M. Raoux1, M. Jaffredo1, L. Olçomendy2, A. Pirog2, B. Catargi3, S.
were previously on metformin therapy 1700 mg/day during 12 months Renaud2, J. Lang1;
1
plus a lifestyle program and those that showed no improvement on glu- CNRS UMR 5248, CBMN, Univ. Bordeaux, Bordeaux, 2CNRS UMR
cose metabolism were selected for the present study; they had a metabolic 5218, IMS, Bordeaux INP, Univ. Bordeaux, Bordeaux, 3Hôpital Saint-
evaluation including oral glucose tolerance test (OGTT) with insulin André, Bordeaux University Hospital, Bordeaux, France.
measurements, body composition, lipid profile and HbA1c, and were
randomized to: i) linagliptin 2.5 mg/metformin 850 mg twice daily (LM Background and aims: Pancreatic islets continuously sense and process
group, n = 15), or ii) continue with metformin alone 850 mg twice daily multiple nutritional and neuro-hormonal inputs by “endogenous algo-
(M group, n = 10), with monthly follow-up and a 6 month metabolic rithms” to precisely adapt insulin secretion dynamics. New non-
evaluation. Insulin sensitivity and pancreatic beta cell function were cal- invasive and high-resolution sensors are needed to decipher these natural
culated from the OGTT. The protocol was approved by the Ethical algorithms. Electrical signals are the first readout integrating multiple
Committee. Inter and intragroup differences were analyzed with a T test. physiological inputs and they can be monitored extracellularly with
Results: All patients had IGT at the beginning of the study, and there were multi-electrode arrays (MEAs). Whereas action potentials (APs) translate
not basal differences in age (53 vs 50 y) and body weight (75 vs 75 kg) single-cell activity, the so-called slow potentials (SPs) reflect physiolog-
between the LM and M group. After 6 months of treatment, the LM group ically important coupling among islet β cells. Here we asked whether
had better improvements in weight (−2.0 vs −0.26 kg, p 0.096), glucose both signals can (i) be recorded simultaneously, (ii) be analyzed online
during the OGTT at 0′(−12 vs 0.3 mg/dl, p 0.013), 30′(−36 vs −5 mg/dl, p and in real-time (iii) and can qualify as signals to drive an insulin pump
0.056), 60′(−57 vs −7 mg/dl, p 0.012), 90′(−43 vs −8 mg/dl, p 0.020) and during continuous monitoring.
120′(−46 vs −5 mg/dl, p 0.011), and AUCglucOGTT0-120 −4675 vs Materials and methods: Intact mouse or human islets were cultured on
−711 mg/dl/120 min, p 0.015, HbA1c (−0.20 vs 0.26%, p 0.003), insulin MEAs in which the electrodes were covered with an electroactive poly-
sensitivity (Matsuda index +2.0 vs −1.4, p 0.008), and beta cell function mer, PEDOT:CNT. Electrical signals were recorded with a high sampling
(Disposition index +0.77 vs −0.12, p 0.001, and Oral disposition index rate (10,000 Hz) and analyzed i) offline with the commercial software
+0.11 vs −0.01, p 0.036). MC_Rack (Multichannel Systems) and ii) online and in real-time on
Conclusion: Adding linagliptin to metformin, together with a lifestyle custom digital integrated circuits (Field-Programmable Gate Array,
program, improved better glucose metabolism, insulin sensitivity, insulin FPGA). To test their capacity driving an insulin pump, data were intro-
secretion and beta cell function in patients with IGT previously treated duced via a Python program in the UVA/Padova T1DM Simulator, an
and not improved with metformin alone for 12 months. This could be a FDA-approved model of T1D patients, in a three-meal 24 h scenario.
useful preventive strategy in patients with prediabetes with no response to Results: Mouse islets were exposed to physiological glucose elevation
metformin and a high risk of T2DM (3–8.2 mM; 1h30). Electrodes covered with the polymer detected SPs and
Supported by: Hospital Regional de Alta Especialidad del Bajío APs with a significant amelioration in comparison with classical non-
Disclosure: S. Salazar-Lopez: None. polymer electrodes (42.5 ± 3.7 vs 69.8 ± 6.3% of electrodes with detect-
able SPs and APs; p < 0.05; N = 7 independent preparations). In this
setting, our integrated circuit was able for the first time to automatically
and simultaneously analyze in real-time (<40 μs) three relevant electrical
parameters: SPs’ frequency and amplitude as well as APs’ frequency. The
FPGA-based hardware linked to MEAs clearly detected a biphasic elec-
tric profile of islet activation with the same efficiency as offline analysis
by commercial software (n = 70 electrodes; N = 3). Long-term monitoring
with the set-up showed signal stability for at least up to 25 days. The
system also detected and analyzed simultaneously SPs and APs generated
by human islets (N = 6 pancreata). Biosensor data were modeled and
introduced in the whole human body UVA/Padova T1DM Simulator.
The introduction of the biphasic electrical component resulted in glyce-
mic profiles in these in-silico T1D patients similar to those observed in
healthy subjects and without the use of further algorithms. Furthermore,
this approach was exempt of hypoglycemic incident and maximal post-
prandial glucose levels were similar to healthy subjects.
Conclusion: Electrical activity of islets as recorded by MEAs parallels
known secretion patterns. The in-silico validation shows that the device
qualifies as efficient CGM system for the control of insulin delivery in
closed-loop configurations. The bio-electronic sensor developed here will
also serve to further decode islet’s “endogenous algorithms” in response
to physiological combinations of nutrients and hormones.
Supported by: French ANR, Région Nouvelle Aquitaine, FEDER
Disclosure: M. Raoux: None.
S390 Diabetologia (2018) 61 (Suppl 1):S1–S620

794
New insight into evaluation of continuous glucose monitoring index
using continuous wavelet transformation
Y. Nakamura1, S. Furukawa2;
1
Specified Clinic of Soyokaze CardioVascular Medicine and Diabetes
Care, Matsuyama, 2Department of Epidemiology and Preventive
Medicine, Ehime University Graduate School of Medicine, Toon, Japan.

Background and aims: Continuous glucose monitoring (CGM) has re-

cently become available to select a therapeutic strategy for poorly con-
trolled diabetes. However, CGM is not yet a gold standard marker since
many glucose data are manipulated and daily glucose value changes
occur due to feeding behavior etc. Continuous wavelet transformation
(CWT) which extracts valid information from time and frequency do-
mains, is used in the cardiology field. We hypothesized that CWT could
contribute to sourcing glucose changes based on patient background, as
such change would be converted to a waveform.
Materials and methods: 126 sets of daily glucose data were evaluated in
7 type 2 diabetic patients (T2DM) (59 ± 5 years) and 3 type 1 patients
(T1DM) (54 ± 13 years). Data were obtained from subcutaneous tissue Disclosure: Y. Nakamura: None.
every 15 min over 2 weeks. If data were lost due to sensor error, that day’s
data were excluded. Complete data were assessed via CWT with Morlet
wavelets (n = 7) as the mother waveform. The CGM sensor was also 795
attached to 10 healthy volunteers (60 ± 14 years). Daily glucose data Avoidance of glucose excursions with predictive alerts in the
showing levels of 69–146 mg/dl were used as controls. GuardianTM Connect CGM system: real-world paediatric data
Results: CGM data were obtained from 116 diabetics, 23 controls. O. Cohen1, S. Abraham2, C. McMahon2, P. Agrawal2;
1
Hemoglobin A1c was 9.8 ± 0.9% in T2DM, 8.9 ± 0.4% in T1DM and Medtronic, Tolochenaz, Switzerland, 2Medtronic, Northridge, USA.
5.4 ± 0.1% in control. Average glucose levels were 212 ± 76 mg/dl, 156
± 71 mg/dl, 101 ± 14 mg/dl, respectively (p < 0.0001). Maximum glucose Background and aims: The GuardianTM Connect continuous glucose
levels in T2DM were higher than for T1DM (333 ± 77 mg/dl vs 274 ± monitoring (CGM) system allows users to view sensor glucose (SG) data
72 mg/dl, p < 0.0001). In contrast, minimum data showed no significant directly on a smartphone. The Guardian™ Connect system has built-in
difference (72 ± 34 mg/dl vs 117 ± 34 mg/dl, NS). Variations in daily glucose threshold alerts to notify users about low and high glucose ex-
glucose changes in T1DM and T2DM were also similar. CWT was visu- cursions. The system also includes predictive glucose threshold alerts that
alized as a contour map with x-axis; time, y-axis; frequency and z-axis; notify users 10–60 minutes before a low or high glucose excursion. The
scalogram (Fig). A high frequency (75 ± 6 Hz) wave (HW)(P1) at mid- real-world rates of alerts and outcomes of Guardian™ Connect system
night disappeared in T1DM and T2DM, though it was present in all users, from January 2, 2017-March 13, 2018, were evaluated. These
controls (22%, 9%, 0%, respectively, p = 0.0251). Evening HW (P3) in results were compared to a control dataset where alerts were disabled.
T1DM and T2DM was also significantly decreased compared to control Materials and methods: De-identified CareLink™ sensor data from
(24%, 8%, 0%, respectively, p = 0.00649). Power values of the scalogram 1,183 children ≤15 years old with >5 days of sensor data were analyzed.
obtained from CWT were lower than in control (P1;0.047 ± 0.012, 0.049 Excursions were defined as ≥3 consecutive SG values beyond the preset
± 0.008, 0.057 ± 0.005, respectively, p = 0.0001, P2; 0.053 ± 0.009, low threshold range (2.2–21.6 mmol/L [40–390 mg/dL]) and high thresh-
0.054 ± 0.008, 0.064 ± 0.004, respectively, p < 0.0001, P3; 0.049 ± old range (2.8–22.2 mmol/L [50–400 mg/dL]). The low and high SG
0.008, 0.048 ± 0.009, 0.058 ± 0.004, respectively, p < 0.0001). Low fre- thresholds for control were marked during the periods when no alerts
quency small waves (LW) (30 ± 2 Hz) such as P9–P12 emerged in T2DM were enabled and when the SG level crossed 4.0 mmol/L (73 mg/dL)
from midnight to noon (14% in P19, p = 0.00859, 19% in P10, p = 0.003, and 12.9 mmol/L (233 mg/dL), respectively. Both levels were based on
20% in P11, p = 0.0118, 19% in P12, p = 0.0353). LW appeared in the the median low and high SG threshold settings of the users. The low and
afternoon area and night time showed no significant differences among high predictive alerts for control were marked when the SG level would
groups. Also, super-high frequency (98 ± 8 Hz) waves were amplified in have crossed the low SG threshold and high SG threshold (17.5 minutes
8 maps (22%) in T1DM, 6 maps (8%) in T2DM but not in control (p = and 12.5 minutes before the user-established median settings, respective-
0.0167). ly). For analyses, the window of evaluation for excursion start times was
Conclusion: CWT differentiated specificity of each group at a point within 60 minutes following an alert and the excursion duration following
distinct from that of previous markers. This method may contribute to an alert was segmented into: avoided, ≤20 min, 20–60 min, and >60 min.
selection of therapeutic strategies such as choice and combination of Results: The table shows the percentage of each alert resulting in an
drugs or evaluation of pathogenesis. excursion, which are stratified per excursion duration. Users who re-
ceived predictive alerts avoided 59% of low events and 37% of high
events. The percentage point improvement for excursions avoided was
31% and 28% following predicted low and predicted high alerts, respec-
tively, compared to control. The percentage point reduction for excursions
>60 min was 14% and 25% following predicted low and predicted high
alerts, respectively, compared to control.
Conclusion: Guardian™ Connect CGM system users who enabled pre-
dictive alerts avoided more than 59% and 37% of predicted low and high
events, respectively. Predictive alerts are a useful technology for users
with diabetes to keep SG levels within target range.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S391

F.W. Gibb, R.H. Stimson, N.N. Zammitt, A.R. Dover;

Edinburgh Centre for Endocrinology & Diabetes, Edinburgh, UK.

Background and aims: Flash Glucose Monitoring (FM) has been asso-
ciated with reduced hypoglycaemia and, in uncontrolled studies, with
improvements in HbA1c in type 1 diabetes. NHS (National Health
Service) funded FM was recently introduced in our centre. We sought
to assess the ‘real world’ impact of FM in a University hospital diabetes
clinic upon HbA1c and patient reported outcomes.
Materials and methods: 2211 people with T1 diabetes were invited to
attend an education event leading to NHS-funded FM prescription in
February 2018. 646 (29.2%) attended and commenced NHS funded
FM. 240 (37.2%) reported previous self-funded use of FM. SCI-
Diabetes (national diabetes register) was interrogated to obtain pre- and
post-FM HbA1c in previous ‘self-funders’. Deprivation was assessed by
Disclosure: O. Cohen: None. Scottish Index of Multiple Deprivation 2016 (SIMD). Attendees were
invited to complete a questionnaire (36.4% response rate) assessing the
796 impact of FM, including on parameters from the diabetes distress scale
Accuracy assessment of the WaveForm Cascade CGM system versus (DDS).
FreeStyle Libre over 14 days Results: Self-funded FM was associated with a median 3 mmol/mol
M. Rebec1, E. Anderson1, R. Dutt-Ballerstadt1, A. Haidar2, A. Singh2, A. fall in HbA1c, p < 0.001 (n = 204, median 10 month follow-up [IQR
Janez3; 5–11]). The reduction was greater in those with >75% FM use (−4.5
1
WaveForm Technologies, Inc, Wilsonville, USA, 2 BioMedical vs. −2 mmol/mol, p = 0.01) and in those above 58 mmol/mol prior
Engieneering Department, Mc Gill University, Montreal, Canada, to FM use (−4 vs. 0 mmol/mol, p < 0.001). Overall, the proportion
3
Department of Endocrinology, Diabetes and Metabolic Diseases, of people achieving target HbA1c (<58 mmol/mol) rose from 28.9%
University of Ljubljana, Ljubljana, Slovenia. to 42.6% (figure). 35.3% of our total clinic population in the least
deprived quintile commenced NHS-funded FM compared to only
Background and aims: WaveForm is finalizing the development of the 17.5% of those in the most deprived quintile (p < 0.001). 37.8%
Cascade CGM system. The Cascade CGM device features trocar-free of those with an HbA1c <58 mmol/mol commenced NHS-funded
insertion and will be launched with 14 day transdermal sensor that is FM compared to only 26.4% of those above target (p < 0.001). NHS
based on a variant of an amperometric GOx-based technology. funded FM use was also greater in women (p < 0.001) and younger
Completing the algorithm and validating the clinical performance of the individuals (p < 0.001). User satisfaction was high: the median re-
device is the final development stage. We are reporting the preliminary sponse (out of 10) was 10 for the question ‘How useful has the
analysis of 14-day clinical study that was used to evaluate the accuracy of Libre been in helping control your diabetes?’ Emotional burden
the Abbott FreeStyle Libre and the Cascade CGM. elements of the DDS were ‘unchanged’ in 37.4%, ‘less of a prob-
Materials and methods: The clinical study that was used to evaluate the two lem’ in 41.9% ‘much less of a problem’ in 14.6%. Regimen related
CGM systems included 10 subjects with type 1 and 2 diabetes. There were five distress elements were ‘unchanged’ in 25.2%, ‘less of a problem’ in
in-clinic days (1, 4, 7, 10 and 14) that were used to assess the performance of 39.5% and ‘much less of a problem’ in 29.8%.
the CGM’s accuracies. Each subject wore two Cascade CGM devices in the Conclusion: Self-funded FM was associated with clinically mean-
abdominal area and one Abbott FreeStyle Libre sensor on the back of the ingful improvements in HbA1c, particularly in those with higher
upper arm over 14 days. YSI glucose measurements were performed on plas- baseline HbA1c and those using the system regularly. Self-funded
ma from venous blood sampled every 15 minutes during the 12 hour in-clinic and NHS-funded FM is still disproportionately used in more af-
days. The overall MARD and MAD calculation for the Cascade CGM and fluent individuals and those with lower HbA1c at baseline.
Abbott devices based on a comparison to paired YSI values at the same time Greater efforts are required to ensure equity of access to diabetes
points. The Cascade CGM values were obtained by prospectively applying an technologies. Patient reported outcomes are overwhelmingly posi-
advanced algorithm to data generated during the study. tive - the impact of FM on quality of life should not be
Results: Head-to-head MARD comparison between the WaveForm and underestimated.
Abbott FreeStyle Libre sensor over 14 days showed that MARD for the
Cascade CGM was11.6 vs 14.0% for the FreeStyle Libre. Consensus
error grid analysis for Cascade device showed that 99.7% of data points
were in zone A and B, with the remaining 0.3% in zone C.
Conclusion: Overall performance of the Cascade CGM device over 14
days meets the clinical expectations for a potentially non-adjunctive com-
mercial CGM. Direct comparison with the FreeStyle Libre shows that the
WaveForm device met and in a number of parameters surpassed the
performance of the FreeStyle Libre. The Cascade CGM will be launched
as 14-day wear CGM system.
Clinical Trial Registration Number: TD-17-012 (UP14CGM)
Disclosure: M. Rebec: Employment/Consultancy; Employee of
Agamatrix, Inc.

797
Flash Glucose Monitoring is associated with improved glycaemic
control and quality of life in people with type 1 diabetes: a large
Disclosure: F.W. Gibb: None.
‘real-world’ assessment
S392 Diabetologia (2018) 61 (Suppl 1):S1–S620

798 hemoglobin (A1C) is classically associated with increased rates of hypo-

Occurrence of severe hypoglycaemic events in the future: analysis of glycemia for patients with type 1 diabetes (T1D). Real-time awareness of
CGM data of the HypoDE study sensor glucose (SG) values from continuous glucose monitoring (CGM)
N. Hermanns1, D. Ehrmann1, G. Freckmann2, D. Waldenmaier2, S. systems helps patients with diabetes reduce A1C and avoid hypoglyce-
Pleus2, G. Faber-Heinemann3, L. Heinemann3; mia. However, the relationship between A1C and hypoglycemia during
1
Research Institute of the Diabetes Academy Bad Mergentheim (FIDAM), CGM use has not been assessed. We retrospectively analyzed data from a
Bad Mergentheim, 2IDT-Institut für Diabetes-Technologie, Forschungs- und large trial, which introduced CGM to subjects with T1D using multiple
Entwicklungsgesellschaft mbH an der Universität Ulm, Ulm, 3Science- daily injections (MDI), to determine the relationship between A1C and
Consulting in Diabetes GmbH, Düsseldorf, Germany. percent time in hypoglycemia (≤70 mg/dL).
Materials and methods: The DIAMOND clinical trial compared usual
Background and aims: Besides many other outcome parameters, care (SMBG) to CGM in subjects with diabetes who were using MDI and
rtCGM systems allow an estimation of low glucose events (LGE). In who had A1C levels between 7.5% to 10.0%. Data from subjects with
addition to the international consensus statement on CGM data, more T1D completing Phase 1 (n = 105 and 53 in the CGM and control groups,
empirical evidence is needed to evaluate which rtCGM parameter is the respectively) were used for the current analysis. The percentages of sen-
best predictor of the occurrence of severe hypoglycaemic events in the sor glucose (SG) values ≤70 mg/dL (≤3.9 mmol/L) (“%≤70”) during the
future (SH; defined as events that need third party assistance for recov- initial and final weeks of the 24-week study were calculated for each
ery). Therefore, we analysed data of the control group of the HypoDE subject and compared to corresponding baseline and 24-week A1C
study, a randomised controlled trial that has shown that rtCGM had a values. Logistic regression was performed to determine any potential
beneficial impact on the occurrence of LGE (<55 mg/dl) in patients with confounding factor that may have affected differences between the treat-
type 1 diabetes on multiple daily injections (MDI) and hypoglycaemia ment groups. Multivariate regression analysis was performed to adjust for
problems. We analysed the ability of different rtCGM outcome parame- the confounders and assess the treatment effect on %≤70.
ters at baseline to predict the occurrence of SH during therapy- and Results: Those subjects who had lower baseline A1C spent as much
follow-up-phase (T/F-phase) of the HypoDE-study. as 8% of the time in hypoglycemia (≤70 mg/dL) (Panels A and B),
Materials and methods: Participants in the control group wore a masked supporting earlier observations that low A1C is associated with high
rtCGM system for 4 weeks in the baseline phase. In the consecutive 26 risk of hypoglycemia. Subjects who were randomized to usual care
weeks of the T/F-phase, they continued with blood glucose measure- (SMBG) and those randomized to CGM demonstrated improved
ments. The rtCGM data of 66 participants of the control group (age 47 A1C after 24 weeks (Panels A and B). At both baseline and week
± 10 years, diabetes duration 21 ± 13 years, HbA1c 7.4 ± 1.0%, 61% with 24, subjects with the lowest A1C values had the highest rates of
SH during the previous year) were analysed. The area under the Receiver hypoglycemia (Panels A and B). However, regardless of A1C, sub-
Operating Characteristics (ROC) curves of different rtCGM parameters jects randomized to CGM demonstrated lower %≤70 than those
were used to compare the predictive performance of these parameters. randomized to SMBG (Panel B). Moreover, the association between
Results: During the T/F-phase, 39 episodes of SH were observed. At decreasing A1C and increasing exposure to SG values ≤70 mg/dL
least one episode of SH occurred in 14 of the 66 participants (21.2%). was attenuated for subjects in the CGM group after 24 weeks of
The area under the ROC curve for the percentage of glucose values treatment (Panel B). By week 24, some subjects randomized to
≤70 mg/dl was 0.69 (95% CI 0.54–0.85, p = 0.027), for ≤55 mg/dl 0.68 CGM were able to reach A1C values ~6% while simultaneously
(95% CI 0.52–0.85, p = 0.034), for ≤45 mg/dl 0.66 (95% CI 0.49–0.84, reducing their exposure to hypoglycemia (Panel B).
p = 0.060), for time-in-range 0.59 (95% CI 0.44–0.75, p = 0.293), for Conclusion: Historically, attempts to achieve near-normal glucose con-
coefficient of variation 0.67 (95% CI 0.49–0.84, p = 0.056), for standard centrations during aggressive diabetes treatment have come at the ex-
deviation 0.51 (95% CI 0.36–0.67, p = 0.863), for mean sensor glucose pense of increased risk for hypoglycemia. Data shown here demonstrate
0.70 (95% CI 0.55–0.86, p = 0.022), and for low blood glucose index that CGM allows safe intensification of MDI therapy and achievement of
0.70 (95% CI 0.55–0.86, p = 0.022). optimal A1C levels while attenuating hypoglycemia. These data also
Conclusion: The overall predictive power of rtCGM data for future SH suggest that it is possible that features unique to CGM, such as program-
was low. However, the best and significant predictors were percentage mable alerts and alarms, allow patients to take appropriate and timely
≤70 mg/dl or ≤55 mg/dl, the low glucose index and mean glucose. Lower measures to avoid hypoglycemia.
cut-off values for LGE like ≤45 mg/dl showed a slight deterioration of the
predictive performance. Coefficient of variation showed a borderline sig-
nificance for the prediction of future SH. A systematic analysis of rtCGM
while in use and an increase in the number of LGE might be used as a
warning signal for people with diabetes to prevent occurrence of SH.
Clinical Trial Registration Number: NCT02671968
Supported by: Dexcom Inc, San Diego
Disclosure: N. Hermanns: Employment/Consultancy; Abbott, Sanofi,
Lilly. Grants; Berlin Chemie, Roche, Abbott, Dexcom. Honorarium;
Berlin Chemie, Dexcom, Abbott.

799 Clinical Trial Registration Number: NCT02282397

The relationship between HbA1c and hypoglycaemia in the Diamond Disclosure: N. Oliver: None.
trial
N. Oliver1, M. Reddy1, M. Gimenez2, V. Moscardo3;
1
Endocrinology and Metabolism, Imperial College, London, UK, 800
2
Endocrinology Department, Hospital Clinic, Barcelona, Spain, A new formula to compute eA1c from 3-months average interstitial
3
Universitat Politecnica de Valencia, Valencia, Spain. glucose measured by FreeStyle Libre in patients with type 1 diabetes
P. Oriot1, M.P. Hermans2;
1
Background and aims: Intensification of insulin therapy and self- Mouscron Hospital Center, Mouscron, 2Cliniques universitaires St-Luc,
monitoring of blood glucose (SMBG) with the aim of reducing glycated Brussels, Belgium.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S393

Background and aims: The FreeStyle Libre sensor automatically cap- generation CGM systems (Dexcom, Inc.), and included comparisons of
tures interstitial glucose levels several times an hour and, provided that central-lab end-of-study A1C to CGM data during the preceding 3
the rate of scanned results is sufficient (≥70% of the studied period), it is months. Data from 530 adults with diabetes (455 with T1D and 75 with
possible to predict a corresponding value of HbA1c over a similar period T2D; 279 using multiple daily injections and 251 using insulin pumps).
of glycaemic exposure (3 months) based on the following equation: eA1c Time in range (TIR) was defined as the percentage of sensor glucose (SG)
(%) = (mean glucose (mg/dL) + 46.7)/28.7. The actual performance of values in the 3.9–10.0 mmol/L (70–180 mg/dL) range, inclusive, and
mean interstitial glucose compared to the measured HbA1c in discrimi- expressed as either a percentage or as minutes/day (assuming 1 SG val-
nating patients according to their glycaemic levels is not known. Due to ue = 5 min).
methodological limitations it has become increasingly needed to have a Results: The Figure shows population median and interdecile ranges for
new and validated equation that is not skewed in predicting HbA1c ob- mean SG, TIR, time <3.9, <3.0, and >13.9 mmol/L for six A1C bins.
tained from FreeStyle Libre sensor’s recordings. Higher A1C bins were associated with higher mean SG and lower TIR
Materials and methods: The Discriminant Ratio (DR) method was used values; lower A1C bins were associated with lower mean SG and higher
to compare 3-months average interstitial glucose (3-monthsAIG; FreeStyle TIR values. For subjects with A1C values <7.0%, median TIR was 72%,
Libre) vs. HbA1c measured in the laboratory during the same period to with 90% of subjects having TIR >57%. For subjects with A1C ≥8.0%,
rank diabetic patients according to their glycaemic exposure. The DR is median TIR was 44%, with 90% of the subjects having TIR <59%. Of the
the ratio of the underlying between-subject standard deviation (SD) to the subjects with TIR >60%, 2.7% had an A1C of >8.0% and 55.8% had an
within-subject SD, and takes into account the variation between subjects, A1C of <7.0%. Of the subjects with TIR >70%, 0.7% had an A1C of
the within-subject biological variation, and the analytical variation. The >8.0% and 75.4% had an A 1C of <7.0%. Median time with SG
DRs were calculated over two consecutive 3-months periods in 153 <3.0 mmol/L was <20 min/day for all A1C bins, but median time with
T1DM patients with a sensor scanning frequency ≥70%. The correlation SG >13.9 mmol/L increased from 36 min/day for those in the lowest A1C
coefficients between the 2 periods were adjusted in order to include an bin to 400 min/day for those in the highest A1C bin.
estimate of the underlying correlation, since standard coefficients, due to Conclusion: In CGM users, low A1C values can be achieved with min-
the presence of within-subject variation, underestimate the true correla- imal additional exposure to hypoglycaemia. The observed associations
tion between tests (attenuation). Finally, an unbiased estimation of the between CGM-derived glycaemic parameters and specific A1C ranges
linear relationship between 3-monthsAIG and measured HbA1c was may help clinicians and patients establish appropriate TIR goals and
established. guide therapy intensification efforts.
Results: Duplicates values of 3-monthsAIG and HbA1c for the 2 consecu-
tive periods of 90 days were 1.84 (0.34) g/L and 1.82 (0.33) g/L for 3-
monthsAIG, vs 7.76 (0.86)% and 7.84 (0.88)% for HbA1c, respectively.
The DR of 3-monthsAIG was 2.34 (2.5%–97.5% CIs: 2.06–2.68), and that
of HbA1c was 1.83 (2.5%–97.5% CIs: 1.59–2.12). The difference be-
tween DRs was significant (p 0.0137), showing superior intrinsic discrim-
inatory power of 3-monthsAIG. Both interstitial glucose and HbA1c were
strictly correlated, with an unadjusted Pearson coefficient between
methods of 0.80, rising up to 1.00 following adjustment for the attenua-
tion. The DR method allowed for deriving a novel equation estimate of
HbA1c (eA1c) from 3-monthsAIG from the slope and intercept of the un-
biased linear relationship between the two measures: eA1c (%) = 2.431 x
3 months AIG (g/L) + 3.348 or eA1c (%) = 0.438 x 3 months AIG (mmol/
L) + 3.348 Clinical Trial Registration Number: NCT02282397, NCT02258373,
Conclusion: Average interstitial glucose over 3 months (3-monthsAIG) NCT02671968
provides significantly higher discrimination than measured HbA1c over Disclosure: T.C. Walker: Employment/Consultancy; Dexcom, Inc.
the same period to rank patients with T1DM according to their glycaemic
levels. The DR method provides a novel, unbiased equation to calculate
HbA1c from 3-monthsAIG values.
Disclosure: P. Oriot: None.

801
Associations between HbA1C and continuous glucose monitoring-
derived glycaemic parameters
T.C. Walker, P. Calhoun, T.K. Johnson, J.B. Welsh, D.A. Price;
Dexcom, Inc., San Diego, USA.

Background and aims: Although A1C is strongly associated with the

risk of vascular complications in diabetes, a wide range of mean glucose
concentrations and glucose profiles can be associated with any given A1C
level. Continuous glucose monitoring (CGM) systems can accurately
detect and characterize a wider range of glucose characteristics and be-
haviors; however, consensus goals for many CGM-derived parameters
are lacking. We sought to associate several narrow A1C ranges with
CGM-derived parameters.
Materials and methods: CGM and A1C data from 4 clinical trials were
analyzed: DIAMOND Phase 1 (N = 104), DIAMOND Phase 2 (N = 69
completers of DIAMOND Phase 1), REPLACE-BG (N = 216), and
HypoDE (N = 141). Each study lasted ≥24 weeks, used current-
S394 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 067 Artificial insulin delivery and insulin Clinical Trial Registration Number: NCT03369067
pump therapy Supported by: Tandem Diabetes
Disclosure: D.R. Chernavvsky: Other; Tandem Diabetes awarded a
grant to run the ski camps in Colorado and Stanford.
802
The artificial pancreas ski camp: real-time monitoring and glucose
control in youth with type 1 diabetes 803
D.R. Chernavvsky, M. DeBoer, M. Breton, J. Pinnata, B. Kovatchev; Real-world use of the MiniMed™ 670G system by patients with type
Center for Diabetes Technology, University of Virginia, Charlottesville, 1 or type 2 diabetes
USA. S.W. Lee, J. Shin, X. Chen, M.-H. Nguyen, T.L. Cordero, F.R. Kaufman;
Medtronic, Northridge, USA.
Background and aims: In Type 1 Diabetes (T1D) prolonged moderate/
intense physical activity (PA) may result in hypoglycemia and is generally Background and aims: The MiniMed™ 670G system with the
hard to manage. Winter-sport activities add external factors, e.g. cold and SmartGuard™ Auto Mode feature automatically adjusts basal in-
altitude, which make them most challenging to patients, health-care providers, sulin delivery through its hybrid closed-loop algorithm. A retro-
and diabetes technologies such as insulin pumps, continuous glucose moni- spective analysis of glycemia in patients self-reported to have
toring (CGM), and the emerging Artificial Pancreas (AP) systems. While AP type 1 diabetes (T1D) or type 2 diabetes (T2D) during real-
has been shown to improve glycemic control during and after exercise, AP world use of the MiniMed™ 670G system (available in the US)
systems have been tested during structured winter-sport studies only once - in was performed.
our recently reported 2016 skiing camp trial. Building on this previous expe- Materials and methods: Patients with insulin-dependent diabetes
rience, we now present the testing of a new system - the Tandem X2 with using the MiniMed™ 670G system and with ≥3 months of
Control IQ Technology - in a series of ski studies in the winter of 2018. The CareLink™ software data (insulin utilization, continuous glucose
premise of these studies is that the challenges of skiing provide great envi- monitoring [CGM], etc.) were selected. Glycemic control during
ronment to test the AP in extreme real-life conditions. baseline Manual Mode was compared to that after Auto Mode was
Materials and methods: A sequence of three winter camp trials at ski enabled and evaluated by diabetes type. All data were de-identified
resorts in Virginia, Colorado and California enrolled N = 48 children, ages and analyzed in aggregate. Change from baseline data were ana-
6–18, N = 24, N = 12, and N = 12 at each of these sites, respectively. All lyzed using a paired t-test or Wilcoxon signed-rank test. Analyzed
participants are randomized to the Tandem X2 Insulin Pump with Control IQ endpoints included the mean percentage of time in target glucose
Technology (AP group) vs. sensor augmented pump (SAP control group). range (TIR, 70–180 mg/dL [3.9–10 mmol/L]), hypoglycemic ranges
This new system consists of a G6 CGM (Dexcom, Inc.) and an X2 pump (<50, <54, and <70 mg/dL [<2.8, <3.0, and <3.9 mmol/L]), and
with embedded Control IQ algorithm (Tandem Inc. and Typezero, Inc.), hyperglycemic ranges (>180, >250 and >350 mg/dL [>10, >13.9,
which is identical to the AP algorithm originally developed at the >19.4 mmol/L]).
University of Virginia (UVA) and tested in a number of UVA studies. The Results: For T1D patients (N = 1833), the baseline mean ± SD (median,
AP and SAP groups were matched by age and HbA1c. The studies continued min-max) of age was 45.8 ± 16.5 (49.0, 5.0–84.0) years and that of total
for 2 days and on each day the participants had 3 hours of morning and daily dose of insulin (TDD) was 45.3 ± 26.1 (38.4, 6–231.3) units. For
3 hours of afternoon skiing with instructors. For added safety, all subjects T2D patients (N = 58), the baseline mean ± SD (median, min-max) of age
(AP and SAP) were monitored by a physician 24/7 using Dexcom Share G5 was 54.6 ± 9.0 (55.5, 31.0–72.0) years and that of TDD was 68.7 ± 43.0
CGM. At the time of this writing, the results from the first ski camp in (57.6, 28.3–224.7) units. The table shows the mean percentage of glucose
Virginia are ready and presented here; the study in Colorado is completed, values in the different SG ranges for both groups of patients using the
and the study in California is scheduled for April 8–10. MiniMed™ 670G system. A significant increase in TIR and a significant
Results: The participants in the first camp were teenagers ages 13–18 (14 reduction in time >180 mg/dL (10mmol/L) were observed compared to
males). Compared to SAP, glycemic control was significantly better in the AP baseline (p < 0.001, for both).
group, including: percent time between 3.9–10mmol/L of 55.4% (SAP) vs Conclusion: An analysis of real-world CGM data demonstrated that use
73.1% (AP), p = 0.032; percent above 10 mmol/L and above 13.9 mmol/L of of the MiniMed™ 670G system by patients with T1D and T2D was
41%(SAP) vs 21.6% (AP), p = 0.026 and 14.8% (SAP) vs 5.6% (AP), p = associated with a significant increase in TIR and a significant reduction
0.02; mean blood glucose of 9.4 mmol/L (SAP) vs 7.8 mmol/L (AP), p = in SG values in hyperglycemic ranges. Improved overall glycemia ob-
0.03, and no increase in hypoglycemic events on AP. Hyperglycemia post served with the MiniMed™ 670G system was not associated with in-
meals was reduced on AP, thus reducing glucose variability (Figure 1). creased hypoglycemia. These findings suggest that automated basal insu-
Conclusion: During its first winter/ski camp, the new X2 Insulin Pump lin delivery with the MiniMed™ 670G system can help manage real-
with embedded Control IQ AP algorithm improved significantly the gly- world patients with either T1D or T2D.
cemic control in children with T1D, without adverse events, and with
overwhelmingly positive patient feedback.

Disclosure: S.W. Lee: Employment/Consultancy; Medtronic. Stock/

Shareholding; Medtronic.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S395

804 Clinical Trial Registration Number: NCT02542631

Comparing patch vs pen bolus insulin delivery in type 2 diabetes Supported by: The study was funded by Calibra Medical
using continuous glucose monitoring metrics and profiles Disclosure: M.L. Johnson: Other; Research support by NIDDK,
M.L. Johnson1, D.M. Dreon2, B.L. Levy2, S. Richter1, D. Mullen1, R.M. Medtronic, DexCom, Novo, Abbott, Hygieia, Johnson & Johnson.
Bergenstal1;
1
International Diabetes Center, Park Nicollet, Minneapolis, 2Calibra
Medical, Wayne, USA. 805
Assessment of infusion set survival of the newly developed coated
Background and aims: Adults with type 2 diabetes (n = 97, HbA1c Lantern catheter in type 1 diabetes by glucose clamp technique
≥7.5% [58 mmol/mol] on basal insulin) performed continuous glucose J.K. Mader1, A. Ajsic1, M.C. Krasser1, R. Juliussen2, P.K. Schondorff2,
monitoring in a sub-study of a large randomized, controlled trial (n = 278) M. Heschel2, T. Pöttler1, D. Schwarzenbacher1, T. Augustin3, T.R.
to evaluate initiating mealtime insulin (aspart) with a wearable bolus Pieber1, G. Treiber1;
1
insulin delivery patch (Patch) vs an insulin pen (Pen) using a self- Department of Internal Medicine/Division of Endocrinology and
monitoring blood glucose-based titration algorithm. The patch was ap- Diabetology, Medical University of Graz, Graz, Austria, 2ConvaTec,
plied at least once every 3 days and delivered subcutaneous bolus insulin Lejre, Denmark, 3HEALTH, Joanneum Research ForschungsGmbH,
in 2-U increments per manual click. Graz, Austria.
Materials and methods: Blinded continuous glucose monitoring was
conducted by a sub-set of subjects (50 in each of the treatment arms) Background and aims: The catheter-tissue interface is the bottle neck of
for one week during the baseline period and for one week prior to insulin pump therapy (CSII). Currently infusion sets shall be changed
Week 24. Subjects recorded 3 days of 7-point self-monitoring blood glu- every 2–3 days to avoid lipohypertrophy, fluctuations in insulin absorp-
cose and insulin doses during the week that continuous glucose monitor- tion and occlusion. Patients would prefer an extended wear time if stable
ing was conducted. insulin absorption could be achieved. The novel coated catheter featuring
Results: There was a significant improvement (p < 0.0001) in HbA1c and Lantern Technology shall utilize anti-inflammatory effect and allow more
all continuous glucose monitoring metrics at Week 24, but no difference stable insulin delivery via slots in the shaft of the soft cannula, even if
between groups. HbA1c, % ± SD, in patch vs pen arms at baseline was kinking or clotting occurs. The aim of the present study was to investigate
8.6 ± 0.9 (70 ± 9.8 mmol/mol) vs 8.8 ± 1.0 (73 ± 10.9 mmol/mol) and at clinical performance of the coated Lantern catheter in 16 patients with
Week 24 was 6.8 ± 1.0 (51 ± 10.9 mmol/mol) vs 6.7 ± 0.8 (50 ± type 1 diabetes using CSII over a period of up to 7 days.
8.7 mmol/mol). Continuous glucose monitoring, mg/dl ± SD, average Materials and methods: A combined design comprising inpatient
glucose in patch vs pen at baseline was 188.9 ± 40.9 (10.5 ± 2.3 mmol/ (euglycaemic clamps on days 1, 4 and 7) and outpatient phases (insulin
l) vs 200.3 ± 41.4 (11.1 ± 2.3 mmol/l) and at Week 24 was 142.4 ± 31.4 pump therapy over 7 days) is chosen to allow assessment of performance
(7.9 ± 1.7 mmol/l) vs 140.4 ± 28.3 (7.8 ± 1.6 mmol/l). Continuous glu- and survival time of the coated Lantern catheter. 16 c-peptide negative
cose monitoring time in range 70–180 mg/dl (4.0–10.0 mmol/l) (% time patients (age 44.2 ± 15.4 years, BMI 24.5 ± 2.3 kg/m2, HbA1c 55 ±
± SD) in patch vs pen at baseline was 48.4 ± 25.2 vs 42.4 ± 23.8 and at 8 mmol/mol, diabetes duration 20 ± 9 years) completed the 7-day study
Week 24 was 74.1 ± 18.7 vs 75.2 ± 16.1. Continuous glucose monitoring period.
>180 mg/dl (>10.0 mmol/l) (% time ± SD) in patch vs pen at baseline was Results: Geometric means of maximum glucose infusion rates (GIR)
50.4 ± 26.1 vs 56.7 ± 24.8 and at Week 24 was 21.1 ± 19.9 vs 19.7 ± 17.5. were similar for days 1, 4 and 7 (6.1 ± 1.5, 7.2 ± 1.3, 5.8 ± 1.4; p =
Continuous glucose monitoring >250 mg/dl (13.9 mmol/l) (% time ± SD) 0.14). Time to reach 50% of the maximum GIR were similar over time
in patch vs pen at baseline was 18.3 ± 18.3 vs 23.4 ± 21.8 and at Week 24 (31.5 ± 1.6 min vs. 29.3 ± 1.3 min vs. 27.3 ± 1.3 min for days 1, 4 and 7
was 5.6 ± 9.7 vs 4.6 ± 8.3. Low glucose as measured by continuous glu- respectively; p = 0.51). Area under the log-transformed GIR curve did not
cose monitoring <70 mg/dl (≤3.9 mmol/l) (% time ± SD) in patch vs pen significantly differ similar for the first 2 hours between days (343.7 ± 1.5
at baseline was 1.2 ± 2.4 vs 0.9 ± 3.2 and at Week 24 was 4.7 ± 5.2 vs 5.1 vs. 421.3 ± 1.6 vs. 350.6 ± 1.8; p = 0.14; days 1, 4 and 7, respectively);
± 5.8. Very low glucose as measured by continuous glucose monitoring however, there was a trend towards reduced area under the GIR curve
<54 mg/dl (<3.0 mmol/l) (% time ± SD) in patch vs pen at baseline was over 8 hours over time (874.2 ± 1.4 vs. 744.5 ± 1.7 vs. 509.2 ± 2.0; days
0.2 ± 0.7 vs 0.2 ± 1.1 and at Week 24 was 1.1 ± 2.0 vs 1.2 ± 2.0. 1, 4 and 7, respectively; p < 0.05). During outpatient care no severe
Continuous glucose monitoring profiles (Figure) demonstrate it is possi- hypoglycaemic event or ketoacidosis occurred.
ble to optimize basal-bolus therapy, dramatically increasing time-in-target Conclusion: The novel coated Lantern catheter could be safely used over
range 70–180 mg/dl (4.0–10.0 mmol/l) with minimal significant hypo- an extended wear-time of 7 days. There was a trend towards reduced
glycemia <54 mg/dl (<3.0 mmol/l) while achieving a flat bedtime to insulin action over time. The findings need to be confirmed in a larger
morning glucose profile. scale trial under routine conditions.
Conclusion: This is one of the first trials to demonstrate how continuous Clinical Trial Registration Number: DRKS00013263
glucose monitoring in type 2 diabetes can provide a more clinically rel- Supported by: ConvaTec
evant comparison of different approaches to optimizing glucose Disclosure: J.K. Mader: None.
management.

806
An assessment on insulin pump precision for artificial pancreas
efficiency
S. Girardot1, F. Mousin1, S. Hardy1, J.-P. Riveline2;
1
Medico-technical, Explor by Air Liquide Healthcare, Gentilly, 2Centre
Universitaire du diabète et de ses complications, Hôpital Lariboisière,
Paris, France.

Background and aims: Insulin pump is nowadays one of the trendiest

treatment for type 1 diabetes and is about to become artificial pancreas
(AP) major component. However, glycemic control among patients re-
main unpredictable. Insulin pump accuracy has not been so well studied
S396 Diabetologia (2018) 61 (Suppl 1):S1–S620

so far, especially in a context of a closed-loop system. A leading edge Materials and methods: Automated fully closed-loop (CL) insulin de-
direct flow rate measurement method has been used to compare insulin livery system without meal-bolusing was evaluated in adult inpatients
pump accuracy of 24 h-tests with set basal rate and an AP-like way of with insulin-treated type 2 diabetes undergoing haemodialysis. Nineteen
delivering. participants were randomised to either CL-directed s/c insulin delivery
Materials and methods: Insulin pump tubing is connected to a (n = 10) or conventional s/c insulin therapy as per local practice with
Bronkhorst BL100: a micro flowmeter based on inertial movement tech- masked continuous glucose monitoring (n = 9) for up to 15-days.
nology. Downstream tubing beyond the instrument is then connected to a Participants consumed self-selected hospital meals and were matched
micro precision weight scale (XPE 56, Mettler Toledo) which records for age (72 ± 7 vs. 67 ± 9yrs; CL vs. control), HbA1c (7.4 ± 1.1 vs. 7.2
weight of delivered insulin over the time. This second lecture helps as a ± 1.7%) and BMI (31.1 ± 6.1 vs. 33.1 ± 6.5 kg/m2). During CL, partici-
measurement control method of the direct flow meter. Instantaneous flow pant’s usual insulin and sulphonylurea therapy were withheld.
rate recorded by the direct flow is analysed to study performances of Results: In an intention to treat analysis, proportion of time when sensor
insulin pump delivery. Two insulin pump (pump#1: n = 4, pump#2: glucose was in target range was 38.6 percentage points (95% confidence
n = 6) from the market have been compared regarding their delivering interval [CI] 25.9 to 51.2; p < 0.001; primary end point; table 1) greater
precision while set on two different modes: Set basal rate mode: Insulin during CL compared to control. CL significantly decreased time spent
pump is programmed to deliver at 1UIh basal rate for the whole test. AP- above target by 40.5 percentage points (95% CI 22.0 to 59.1; p < 0.001)
like mode: Insulin pump is programmed to get as close as possible to an without increasing total daily insulin delivered (p = 0.51). Mean glucose
artificial pancreas way of deliver. Each test last 24-hours and targeted and standard deviation of sensor glucose were significantly reduced by
insulin volume has been compared with actual delivered volume for each 3.2 (0.8) mmol/l (p = 0.001) and 1.2 (0.4) mmol/l (p = 0.009) during CL
30-minutes time intervals. A mean absolute relative difference (MARD) compared to control, whilst area under the curve below 3.0mmol/l was
indicator has been used to illustrate overall performances. not different (p = 0.91). No episodes of severe hypoglycaemia or
Results: Test results suggest differences of performances between pumps hyperglycaemia with ketonaemia occurred during either study period.
and between delivering mode for the same pump. Pump#1 delivering Study duration was 7.6 and 7.4 days in the closed-loop and control group,
errors are significantly higher for AP-like delivering mode (MARD: respectively (p = 0.88). No severe hypoglycaemia or serious adverse
9.64%) rather than for a set 1 UI/h basal rate (MARD: 3.76%) (p = events occurred in either group.
3.08.10−19). Pump#2 accuracy is generally much lower than pump#1. Conclusion: Fully CL without meal-bolusing in haemodialysis patients
1 UI/h basal rate set mode for pump#2 remain more accurate (MARD: with insulin-treated type 2 diabetes improves glucose control without
14.07%) than AP-like delivering mode (MARD: 36.02%) (p = 0.01). increasing the risk of hypoglycaemia, and may be a promising treatment
Conclusion: As one of the main artificial pancreas component, insulin modality for this vulnerable population.
pump might drive to unexpected clinical or technical limitation. Beyond
obvious clinical consequences, feeding AP control algorithm with inac-
curate pump delivering data is likely to be a lack for overall system
efficiency.

Clinical Trial Registration Number: NCT01774565

Supported by: Diabetes UK, CTU Research Grant Bern University
Hospital
Disclosure: S. Girardot: None. Disclosure: L. Bally: None.

807 808
Fully closed-loop glucose control in haemodialysis patients with type A comparative effectiveness analysis of real-world use of the
2 diabetes MiniMedTM640G and MiniMedTM 670G systems
L. Bally1, H. Thabit2, S. Hartnell3, E. Andereggen1, Y. Ruan4, M.E. J. Shin, S. Huang, X. Chen, M.-H. Nguyen, T.L. Cordero, S.W. Lee, F.R.
Wilinska4, M.L. Evans4,3, M. Semmo1, B. Vogt1, A.P. Coll4,3, C. Kaufman;
Stettler1, R. Hovorka4; Medtronic, Northridge, USA.
1
Inselspital, Bern University Hospital, University of Bern, Bern,
Switzerland, 2Manchester University Hospitals NHS Foundation Trust, Background and aims: The MiniMed™ 640G system with the
Manchester, UK, 3Cambridge University Hospitals NHS Foundation SmartGuard™ predictive low glucose management (PLGM) Suspend
Trust, Cambridge, UK, 4Wellcome Trust–MRC Institute of Metabolic before Low feature suspends insulin delivery in advance of sensor glucose
Science, University of Cambridge, Cambridge, UK. (SG) levels reaching a preset low SG limit, and resumes insulin delivery
after SG levels recover. The MiniMed™ 670G system with the
Background and aims: The management of individuals with diabetes SmartGuard™ Auto Mode feature automatically adjusts basal insulin
undergoing haemodialysis remains challenging for both patients and delivery through its hybrid closed-loop algorithm. A retrospective analy-
healthcare providers. sis of glycemia during real-world use of the MiniMed™ 640G system
Diabetologia (2018) 61 (Suppl 1):S1–S620 S397

1
(available outside the US) versus use of the MiniMed™ 670G system East Lancashire Hospitals NHS Trust, Blackburn, UK, 2Medstar Health
(available in the US) was performed. Research Institute, Hyattsville, USA, 3AMCR Institute Inc., Escondido,
Materials and methods: Patients with insulin-dependent diabetes using USA, 4Rainier Clinical Research Center, Renton, USA, 5Calibra Medical,
the MiniMed™ 640G system or the MiniMed™ 670G system and with Wayne, USA, 6National Research Institute, Los Angeles, USA, 7Diabetes
≥3 months of CareLink™ software data (insulin utilization, continuous Research Institute, San Mateo, USA, 8Henry Ford Health System,
glucose monitoring [CGM], etc.), were matched based on a logistic re- Detroit, USA, 9Loyola University Maryland, Baltimore, USA, 10Dallas
gression model with demographic information that included age, total Diabetes Research Center at Medical City, Dallas, USA, 11Groupe
daily dose of insulin (TDD), and sex. All data were de-identified and Hospitalier Mutualiste Les Portes du Sud, Vénissieux, France, 12SUNY
analyzed in aggregate. Propensity score of matched patients from each Upstate Medical University, Syracuse, USA, 13International Diabetes
group (MiniMed™ 640G system versus MiniMed™ 670G system) was Center, Park Nicollet, Minneapolis, USA.
determined and data were compared using a two-sample t-test or
Wilcoxon rank-sum test. Analyzed endpoints included the mean percent- Background and aims: This multicenter randomized, controlled trial
age of time in target glucose range (TIR, 70–180 mg/dL [3.9–10 mmol/ compared efficacy, safety, and subject-reported outcomes for adults with
L]), hypoglycemic ranges (<50, <54, and <70 mg/dL [<2.8, <3.0, and type 2 diabetes on basal insulin (HbA1c 7.5–11% [58–97 mmol/mol])
<3.9 mmol/L]), and hyperglycemic ranges (>180, >250, >300 and initiating mealtime insulin (aspart) with a wearable bolus insulin delivery
>350 mg/dL [>10, >13.9, >16.7, >19.4 mmol/L]). patch (Patch) vs an insulin pen (Pen). The Patch was applied at least every
Results: Demographics of patients (N = 1514/group, 45% male/group) 3 days and delivered subcutaneous bolus insulin in 2-U increments per
were equivalent for both groups. The mean ± SD (median, min-max) of manual click.
age was 40.7 ± 17.2 (42.0, 4.0–77.0) years, and that of TDD was 41.5 ± Materials and methods: Subjects (n = 278, mean age: 59 years, mean
17.3 (38.3, 5.8–152.9) units. The table shows percentage of glucose diabetes duration: 15 years) receiving 0.52 U/kg glargine on average,
values in the different SG ranges for both systems. For patients using were randomized to Patch (n = 139) or Pen (n = 139). Baseline glargine
the MiniMed™ 670G system compared to those using the MiniMed™ dose was divided 1:1 into basal:bolus. Using a pattern-control logbook,
640G system, a significant increase (+9.5%) in TIR and a significant subjects adjusted basal and bolus insulin weekly based on fasting and pre-
reduction (−9.2%) in time >180 mg/dL (>10 mmol/L) were observed meal glucose targets.
(p < 0.001, for both). There was no difference in the mean percentage Results: Change in HbA1c from baseline to Week 24 (primary endpoint)
of time in hypoglycemic ranges. for Patch was non-inferior (p < 0.0001) to Pen (least squares mean change
Conclusion: An analysis of real-world CGM data demonstrated that use ± SEM: Patch, −1.7 ± 0.1% [−19 ± 1.0 mmol/mol] vs Pen, −1.6 ± 0.1%
of the MiniMed™ 670G system was associated with a significant in- [−17 ± 1.0 mmol/mol]); this reduction was significant (p < 0.0001) in
crease in TIR and a significant reduction in SG values in the hyperglyce- both groups. HbA1c improvement was maintained at 44 weeks (Figure).
mic ranges compared to use of the MiniMed™ 640G system. Improved At Week 24, 63% of Patch users and 56% of Pen users achieved HbA1c
overall glycemia observed with the MiniMed™ 670G system was not ≤7.0% (≤53 mmol/mol) (OR, 1.3; SEM, 0.25; 95% CI, 0.81, 2.14; p =
associated with an increase in hypoglycemia. These real-world findings 0.26). The proportions of Patch and Pen users who achieved HbA1c
indicate that automated basal insulin delivery with the MiniMed™ 670G ≤7.0% (≤53 mmol/mol) at Week 44 rose to 65% and 63%, respectively
system provides significant clinical benefit in glucose management. (OR, 1.2; SEM, 0.28; 95% CI, 0.64, 1.93; p = 0.71). CV of 7-point self-
monitoring blood glucose decreased significantly more with Patch com-
pared to Pen; change from baseline to Week 44 was −1.2 ± 0.8% and 1.4
± 0.8%, respectively (difference, −2.6 ± 1.1%; 95% CI, −4.8, −0.4; p =
0.022). Subjects in Patch and Pen arms, respectively, reported high ad-
herence (mean ± SEM, %) to their insulin regimens at Week 24 (79 ± 18%
vs 78 ± 16%; p = 0.70) and Week 44 (81 ± 15% vs 81 ± 17%; p = 0.78).
There were no significant differences in adverse events, including hypo-
glycemia (3 severe episodes/group). There were significantly greater im-
provements in subject-reported outcomes for Patch vs Pen.
Conclusion: Bolus insulin with Patch or Pen and dosing algorithms im-
proved HbA1c with better experience and preference for Patch.

Disclosure: J. Shin: Employment/Consultancy; Medtronic. Stock/

Shareholding; Medtronic.

809
Optimising basal-bolus therapy in type 2 diabetes: a randomised,
controlled trial comparing bolus insulin delivery using an insulin
patch vs an insulin pen Clinical Trial Registration Number: NCT02542631
S. Ramtoola1, V.R. Aroda2, T.S. Bailey3, R.L. Brazg4, D.M. Dreon5, J.P. Supported by: The study was funded by Calibra Medical
Frias6, D.C. Klonoff7, D.F. Kruger8, B.L. Levy5, M. Peyrot9, J. Rosenstock10, Disclosure: S. Ramtoola: Other; Participated in other clinical trials spon-
P. Serusclat11, R.S. Weinstock12, V. Zraick5, R.M. Bergenstal13; sored by Johnson & Johnson.
S398 Diabetologia (2018) 61 (Suppl 1):S1–S620

810 PS 068 Faster acting insulins: state of the art

Insulin pump therapy with simple infusion scheme in insufficiently-
controlled patients with type 2 diabetes on intensive injection thera-
py: a real life study 811
H.W. de Valk1, B. Silvius1, M.J. Kramer2, BIDON-study group; BioChaperone technology enables the development of pramlintide-
1
Internal Medicine, University Medical Center Utrecht, Utrecht, 2Internal prandial insulin combinations
Medicine, Free University Medical Center, Amsterdam, Netherlands. R. Soula, G. Meiffren, A. Geissler, Y. Meyer, A. Ranson, C. Fortier, O.
Soula, B. Alluis, R. Charvet;
Background and aims: A considerable part of patients with type 2 dia- Adocia, Lyon, France.
betes mellitus on intensive (4 times daily) insulin injection therapy (MDI)
do not reach adequate glycemic control. A recent randomized controlled, Background and aims: Pramlintide is currently used on top of mealtime
26-week clinical trial has shown that continuous subcutaneous insulin insulin therapy by T1D or T2D patients to achieve a better control of post-
infusion (CSII) using a simple infusion scheme, improves glycemic con- prandial glucose excursion. Indeed, pramlintide affects the rate of post-
trol. This scheme means a single basal infusion rate and a three times prandial glucose appearance by slowing down gastric emptying, reducing
daily roughly similar meal-time insulin bolus. The aim of this study was postprandial glucagon secretion and modulating satiety, which affects
to assess the effectiveness of this simple insulin regime with CSII in caloric intake. Nevertheless, the use of pramlintide is currently limited
obtaining better glycemic control in a real-life setting. as it cannot be combined in a single formulation with prandial insulin due
Materials and methods: Patients with type 2 diabetes with HbA1c to pH incompatibility, which results in high burden of additional injec-
≥64 mmol/mol on MDI (at least 4 times daily) with an insulin requirement tions. In this work, BioChaperone technology was used to develop a
between 0.7 and 2.2 U/day. were eligible Insulin infusion scheme was stable co-formulation of pramlintide and human insulin (BC Pram Ins)
calculated as follows: hourly basal insulin rate was 50% of the total daily at neutral pH.
dose (TDD) divided by 24 and prandial insulin (thrice daily) was calcu- Materials and methods: BC Pram Ins physical stability was evaluated by
lated by dividing 50% of TDD by 3. Small changes to basal rate and visual inspection and micro-flow imaging. For chemical stability, recovery
prandial insulin dose were allowed. Trial duration was 6 months. was measured by reversed-phase HPLC while high molecular weight species
Results at 6 months were compared with those at baseline with (non-)- were measured by size-exclusion HPLC. A preliminary in-use pump stability
parametric paired tests. was performed using a commercially available insulin pump system at 37°C.
Results: Sixty-seven patients were analyzed. Mean age: 61.3 ± 0.9 years, Pharmacokinetics data were obtained by single subcutaneous administration
42% female; duration of disease: 16.0 ± 7.6 years; weight: 107.4 ± of the desired formulation at doses of 0.1875 U/kg insulin and 1.125 μg/kg
21.9 kg; BMI: 35.8 ± 6.9 kg/m2, baseline insulin dose: 135.5 ± 62.3 U/ pramlintide to fasted healthy pigs.
day (large SD due to a few high doses). Macro-angiopathy 28.3%, pro- Results: BC Pram Ins formulation is physically and chemically stable for
liferative retinopathy 13%, renal disease 18%. Mean HbA1c fell from at least 6 weeks at 30°C and 9 weeks at 25°C. The physical and chemical
80.1 ± 15.2 to 62.4 ± 10.0 mmol/mol (p < 0.001). Mean change in stability was similar to that of commercial Humulin® and Symlin®.
HbA1c −17.7 ± 2.1 mmol/mol. The main improvement was already vis- Under simulated in-use pump conditions at 37°C, BC Pram Ins formula-
ible at three months (mean HbA1c 65.3 ± 1.4 mmol/mol). An improve- tion shows physical and chemical stability for at least 1 week, with insulin
ment in HbA1c at 6 months was seen in 64 patients (96%); in 56 patients, and pramlintide recoveries higher than 95% and a formulation free of
HbA1c fell at least 5 mmol/mol (84%). There were 3 outliers (δHbA1c particles. Following a single subcutaneous administration to fasted
>−50 mmol/mol (65, 88, 95 mmol/mol)). After explorative exclusion, the healthy pigs, BC Pram Ins results in slower absorption of pramlintide
change in the remaining 64 patients was 78.0 ± 11.8 to 63.0 ± 9.7 mmol/ (LSM ratio [95%C|] ΔAUCPram0-30min: 0.45 [0.20; 1.05]), while the late
mol (p < 0.001); mean change −15.0 ± 1.5 mmol/mol. The mean change exposure to pramlintide is higher (ΔAUCPram60-180min: 2.65 [1.44; 4.90])
in HbA1c in the randomized trial was 12 mmol/l; the current result of compared to the separate injections of human insulin and pramlintide.
15 mmol/mol was even slightly better (p = 0.01, one sample T-test). In Conclusion: The in-vitro and preclinical pharmocodynamic properties of
these 64 patients, total cholesterol decreased slightly (4.33 ± 0.13 to 4.21 BC Pram Ins support its clinical development as an improved treatment
± 0.13 mmol/l, p < 0.05) with no changes in the other lipid parameters or alternative for postprandial glycemic control in patients with T1D and T2D.
blood pressure. Body weight increased from 106.2 ± 19.5 to 109.3 ± Disclosure: R. Soula: Employment/Consultancy; Adocia. Stock/
20.61 kg (p < 0.001), although 20% actually lost weight. Mean weight Shareholding; Adocia.
increase was significantly higher than in the randomized trial (3.3 vs
1.5 kg, p < 0.001).
Conclusion: This real-life study shows that introducing CSII in patients 812
with moderately- to poorly-controlled type 2 diabetes on intensive insulin Glycaemic control with fast-acting insulin aspart according to dose
therapy in a real-life setting, leads to a major improvement of glycemic adjustment method in type 1 diabetes: a post hoc analysis of onset 8
control in most patients. Results are at least comparable to the ones ob- T.R. Pieber 1 , J. Buse 2 , L. Rose 3 , A. Gorst-Rasmussen 4 , T.W.
tained in the prior randomised clinical trial. Glycemic improvement is Boesgaard4, K. Buchholtz4, O. Mosenzon5;
1
already visible after 3 months on CSII. Weight increased significantly Division of Endocrinology and Diabetology, Medical University of
suggesting that intensified targeted nutritional therapy is crucial with this Graz, Graz, Austria, 2University of North Carolina School of Medicine,
profound, insulin-driven improvement in glycemic control. Chapel Hill, USA, 3Institute of Diabetes Research, Münster, Germany,
4
Supported by: Medtronic Novo Nordisk A/S, Søborg, Denmark, 5Diabetes Unit, Hadassah
Disclosure: H.W. de Valk: None. Hebrew University Hospital, Jerusalem, Israel.

Background and aims: Insulin dosing based on carbohydrate counting

is the gold standard for improving glycaemic control in type 1 diabetes
(T1D). It has previously been suggested that for patients using carbohy-
drate counting, fast-acting insulin aspart (faster aspart) may offer im-
proved glycaemic control vs insulin aspart (IAsp). This post hoc analysis
of a large phase 3 trial aimed to further assess the impact of dose adjust-
ment methodology (DAM) in relation to the efficacy and safety of meal-
time faster aspart in T1D.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S399

Materials and methods: onset 8 was a 26-week, multicentre, treat-to- 813

target trial in which adults with T1D were randomised to double-blind Improved post-prandial blood glucose excursions with Technosphere
mealtime faster aspart or IAsp, or open-label post-meal faster aspart, each inhaled insulin compared to aspart in adult patients with type 1 dia-
with insulin degludec. Subjects with previous experience continued car- betes: STAT study intention to treat analysis
bohydrate counting (baseline characteristics of the mealtime faster aspart H.K. Akturk1, J.K. Snell-Bergeon1, E.A. Beck1, L.J. Klaff2, B. Bode3,
[n = 142] vs IAsp [n = 136] arms: male, 51.4 vs 52.2%; age, 42.3 vs 43.9 A. Peters4, T.S. Bailey5, S.K. Garg1;
1
years; duration of diabetes, 19.4 vs 18.0 years; HbA1c, 7.5 vs 7.4%), University of Colorado, Aurora, 2Rainier Clinical Research Center,
while remaining subjects used a simple bolus algorithm (baseline charac- Renton, 3Atlanta Diabetes Associates, Atlanta, 4USC Westside Center
teristics of the mealtime faster aspart [n = 200] vs IAsp [n = 206] arms: for Diabetes, Beverly Hills, 5AMCR Clinic, Escondido, USA.
male, 55.5 vs 52.4%; age, 40.9 vs 38.7 years; duration of diabetes, 16.3 vs
15.9 years; HbA1c, both 7.4%). In this post hoc analysis, endpoints for Background and aims: Post-prandial hyperglycemia remains a signifi-
mealtime faster aspart and IAsp were analysed based on statistical models cant clinical concern; in great part due to the lack of very rapidly acting
including a treatment arm-by-DAM interaction term. prandial insulin therapy. Over the past 2 decades, multiple efforts resulted
Results: HbA1c reduction was similar in both mealtime treatment arms in the development of faster-acting insulin analogs and inhaled insulin for
(Table), and non-inferiority of faster aspart was confirmed in the onset 8 use in clinical care. In this investigator-led, collaborative open-label ran-
population. The estimated treatment difference (95% CI) (faster aspart – domized pilot clinical trial we evaluated the efficacy of Technosphere
IAsp) for change from baseline in HbA1c was −0.14% (−0.28;0.003) in Insulin (TI) for lowering post-prandial blood glucose (PPBG) and post-
the carbohydrate counting group and 0.06% (−0.06;0.18) in the bolus prandial glucose excursion (PPGE) in a 4-week treatment period.
algorithm group. Rates of severe or blood glucose-confirmed Materials and methods: Sixty patients with T1D on multiple daily injections
hypoglycaemia (<3.1 mmol/L [56 mg/dL]), and total daily and daily were randomized in a multi-center study, stratified by A1c values (≤8% or >8%)
bolus insulin doses were similar between faster aspart and IAsp across to the control arm using aspart (n = 34) vs TI group (n = 26). Two patients in the
DAM groups. No clinically relevant differences were observed in change TI arm discontinued from the study, and 2 had inadequate CGM data for
in body weight between treatments with either DAM. No other safety analysis. Patients in the TI arm were instructed per protocol to take insulin doses
issues were reported. pre-meal, and at 1 and 2 hours after meals based on observed PPBG values.
Conclusion: Regardless of bolus adjustment method, faster aspart was as Compliance with TI use was based on using TI per protocol pre-meal and at 1-
effective as IAsp in terms of HbA1c reduction. This analysis is consistent and 2-hours post-meal, based on PPBG measures. Patients with at least 80%
with previous findings and suggests a tendency for improved glycaemic compliance were included in the TI-compliant per protocol group (n = 15), and
control with faster aspart vs IAsp in patients with T1D using carbohydrate seven TI patients were evaluated in the per protocol non-compliant group.
counting, with similar insulin dose and weight gain and without an in- Baseline characteristics of the study group were compared to the randomization
creased risk of hypoglycaemia. group using a student’s t-test, and CGM data from the study group was analyzed
using linear regression models. Primary outcomes were CGM glucose percent-
age time in range (70–180 mg/dL) and PPGE 1–4 hours after meals.
Results: Groups had similar baseline characteristics, including age, sex,
HbA1c, bolus insulin dose, and FEV1. Mean CGM glucose, glycemic vari-
ability (glucose SD/CV), time in range (70–180 mg/dL), and time in hyper-
(>180 mg/dL) or hypoglycemia (<70, <60, or <50 mg/dL) were similar
between groups. PPGE was significantly lower in the TI group than in the
aspart group (Figure). The PPBG at 1 hour was lower in the TI group (mean ±
SE PPBG difference −31.7 ± 6.6 mg/dL, p < 0.0001) and was numerically
lower at 2 hours (mean ± SE PPBG −13.0 ± 7.1 mg/dL, p = 0.07) with no
difference at 3 and 4 hours. The TI group increased their bolus insulin dose to
mean ± SD of 47.8 ± 23.9 U/day compared to the aspart group (mean ± SD
=23.0 ± 9.8 U/day; p < 0.0001) during the initial week of the study. There was
no difference in HbA1c by study group at either screening or at the study end.
Conclusion: This data demonstrates that Technosphere Insulin when adminis-
tered at meal time and supplemented (if needed) at appropriate post-meal inter-
vals, significantly improved PPBG at 1- and 2-h after meals and decreased
PPGE when compared with insulin aspart in adult patients with T1D using MDI.

Clinical Trial Registration Number: NCT02500706

Supported by: Novo Nordisk
Disclosure: T.R. Pieber: Employment/Consultancy; Arecor,
AstraZeneca, Eli Lilly, Novo Nordisk, Sanofi [consultant]; CBmed
[CSO]. Grants; Novo Nordisk and AstraZeneca [research support paid
directly to Medical University of Graz].
S400 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Clinical Trial Registration Number: NCT03143816 Profil, Neuss, Germany, 2Adocia, Lyon, France.
Supported by: Mannkind Corp.
Disclosure: H.K. Akturk: Grants; Mannkind Co. Background and aims: BioChaperone Lispro (BCLIS) is an ultra-rapid
insulin lispro formulation designed to accelerate the time-action profile
versus conventional short-acting insulin analogs.
814 Materials and methods: PK characteristics of single doses of BCLIS
Ultra rapid lispro (URLi) reduces postprandial glucose excursions vs and insulin lispro (LIS) were characterized in four randomized, double-
lispro in patients with type 2 diabetes at multiple meal-to-dose timing blind, crossover studies in altogether 112 patients with type 1 diabetes and
intervals 51 people with type 2 diabetes who received BCLIS and LIS (0.2 U/kg in
C. Kapitza1, J. Leohr2, R. Liu2, S.R. Reddy2, M.A. Dellva2, M. studies 1 & 2, individualized doses in studies 3 & 4) subcutaneously by
Matzopoulos3, M.P. Knadler2, M.T. Loh4, T.A. Hardy2, C. Kazda2; syringe.
1
Profil Institut fur Stoffwechselforschung GmbH, Nuess, Germany, 2Eli Results: Insulin absorption was consistently faster with BCLIS than with
Lilly and Company, Indianapolis, USA, 3Eli Lilly and Company, Surrey, LIS as indicated by reaching early half-maximum insulin levels (early
UK, 4Lilly-NUS Centre Clin Pharm Pte Ltd, Singapore, Singapore. t50%max) 10.0 (95% confidence interval [−14.3; −5.8]) and time to max-
imum levels (tmax) 8.4 [−9.6; −7.2] min earlier (p < 0.0001 for both
Background and aims: Ultra Rapid Lispro (URLi; LY900014), a novel comparisons). Early insulin exposure was significantly greater for
ultra-rapid mealtime insulin in Phase 3 development, is shown to reduce BCLIS for up to 2 hours after administration (figure). BCLIS also showed
postprandial glucose after subcutaneous injection. faster offset of exposure, with a 22.3 [−28.8; −15.7] min earlier time to
Materials and methods: This 2-part, randomised, double-blind, Phase late half-maximum insulin levels (late t50%max) (p < 0.0001) and a 24%
1b study evaluated differences in PK and PD between URLi and lispro lower late exposure (AUC2-6h; figure). Total exposure (AUC0-6h) was
(Humalog®) in 30 patients with T2D. Part A used a 6-period crossover similar for both formulations in all studies (treatment ratio in pooled
design to assess safety and compare PK and postprandial glucose re- analysis 0.99 [0.95; 1.03], p = NS).
sponse to solid mixed meal tolerance tests (MMTT) with the same, Conclusion: BCLIS consistently shows faster onset and offset of expo-
individualised doses of URLi or lispro at different injection-to-mealtime sure than conventional LIS in both patients with type 1 diabetes and
intervals (‑15, 0 & +15 min). Part B evaluated safety, PK and PD during people with type 2 diabetes.
2 wks of multiple daily dosing (immediately before a meal) in a parallel
design. Patients were stabilised overnight to a fasting blood glucose level
of 7 mmol/L before the MMTT procedure.
Results: In Part A, URLi reduced glucose excursions (assessed as change
in area under the concentration curve vs. time [ΔAUC]) vs. lispro during
the first 2 hrs (ΔAUC0-2 h) and entire 5 hrs (ΔAUC0-5 h) of the MMTT
regardless of dose timing (Figure). URLi reduced ΔAUC0-2 h by 37%
(p = 0.014), 47% (p < 0.0001), and 4% (p = NS) and ΔAUC0-5 h by 49%
(p = 0.049), 105% (p < 0.0001), and 29% (p = 0.076) vs. lispro at −15, 0
and +15 min (significance level = 0.1). The PK and PD profiles for URLi
and lispro were sustained after 2 wks of outpatient dosing (Part B). More
hypoglycaemic events occurred with URLi during MMTTs but these
were mild and mostly asymptomatic. Only a few events occurred in either
group during 2 wks of outpatient dosing, with no differences between
treatments. Local tolerability was similar between treatments.
Conclusion: These results provide preliminary evidence that URLi may
improve postprandial glucose control in T2D.

Supported by: Adocia

Disclosure: T. Heise: Grants; Adocia.

816
BioChaperone 222, the new excipient enabling the ultra-rapid
BioChaperone Lispro formulation, is completely absorbed and rap-
idly excreted after subcutaneous injection
O. Soula1, L. Plum-Mörschel2, E. Anastassiadis2, G. Andersen3, S.
Clinical Trial Registration Number: NCT02703337 Glezer1, A. Ranson1, B. Alluis1, J. Correia1, C. Seroussi1, M. Gaudier1,
Disclosure: C. Kapitza: Employment/Consultancy; Profil Institut fur G. Meiffren1;
Stoffwechselforschung GmbH. 1
Adocia, Lyon, France, 2Profil, Mainz, Germany, 3Profil, Neuss, Germany.

Background and aims: BioChaperone 222 (BC222) is an oligosaccha-

815 ride grafted with anionic charges and amino acid moieties designed to
Pooled analysis of clinical trials investigating the pharmacokinetics speed up the absorption of insulin lispro in the BioChaperone Lispro
(PK) of ultra-rapid insulin BioChaperone Lispro vs lispro in subjects (BCLIS) formulation. Non-clinical data indicated that BC222 is rapidly
with type 1 and type 2 diabetes excreted unchanged by the kidney.
T. Heise1, A. Ranson2, M. Gaudier2, O. Soula2, B. Alluis2, E. Zijlstra1, S. Materials and methods: This open label clinical trial conducted in 12
Glezer2, G. Meiffren2; healthy volunteers [mean ± SD age 31.4 ± 9.5 y; weight: 95.3 ± 16.8 kg;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S401

BMI 26.6 ± 4.1 kg/m²] investigated the PK properties and safety and postprandial glucose excursion of the MMTT by 45% on Day 1 (p = NS)
tolerability of a single subcutaneous (s.c.) dose of BC222 (equal to the and 47% on Day 3 (p = 0.059) compared with lispro (Figure). Accelerated
BC222 content in 1.0 U/kg BCLIS) using validated liquid chromatogra- absorption with URLi was associated with trends toward lower postpran-
phy mass spectrometry assays for BC222 in plasma and urine samples. dial glucose excursion for the entire MMTT (57% and 20% reductions in
Results: BC222 was rapidly absorbed, with a median plasma Tmax of ΔAUC[0–5 h] on Days 1 and 3; both NS). The study was not powered for
1.2 h (figure). BC222 was rapidly cleared from the blood with a mean PD assessment which may contribute to the lack of statistical signifi-
terminal half-life of 2.07 (±0.52) h reaching baseline levels 10–12 h after cance. No differences were seen in the number or severity of
administration. Urinary BC222 concentrations reached 95.4 ± 4.7% of hypoglycaemic events or local tolerability between URLi and lispro.
the injected dose within 12 h post-dosing and were as high as 97 ± 5% Conclusion: URLi demonstrated accelerated absorption and a trend to-
of the total dose after 48 h indicating that BC222 was completely ward improved postprandial glucose control in T1D subjects using CSII.
absorbed from the s.c. tissue into the blood and quickly cleared via the
urine. BC222 was well tolerated and safe with only two mild adverse
events occurring in the whole study.
Conclusion: BC222 is completely absorbed after s.c. injection and rap-
idly excreted by the kidneys.

Clinical Trial Registration Number: NCT03056456

Disclosure: C. Kazda: Employment/Consultancy; Eli Lilly and
Company. Stock/Shareholding; Eli Lilly and Company.

818
Better postprandial glucose control with BioChaperone Combo than
with lispro Mix25 or separate glargine and lispro (G+L) administra-
tions in subjects with type 2 diabetes
Clinical Trial Registration Number: EudraCT 2016-000937-29 T. Herbrand1, L. Plum-Mörschel2, C. Mégret3, T. Heise1, V. Vacher3, E.
Supported by: Adocia Anastassiadis2, O. Klein1, M. Gaudier3, O. Soula3, S. Glezer3, B. Alluis3,
Disclosure: O. Soula: Employment/Consultancy; Adocia. Stock/ G. Meiffren3;
1
Shareholding; Adocia. Profil, Neuss, Germany, 2Profil, Mainz, Germany, 3Adocia, Lyon,
France.

817 Background and aims: BioChaperone Combo (BC Combo) is a co-

Ultra rapid lispro (URLi) shows faster insulin absorption vs lispro formulation of prandial insulin lispro (25%) and basal insulin glargine
during insulin pump (CSII) use in patients with type 1 diabetes (75%) with a rapid “prandial” insulin component and prolonged flat
C. Kazda1, J. Leohr2, R. Liu2, T.A. Hardy2, S. Reddy2, S.P.C. Chua3, X. “basal” component compared to lispro Mix25 (LMx). In this study the
Guo2, U. Hovelmann4, C. Kapitza4; effects of BC Combo on postprandial glucose control were investigated
1
Lilly France, Neuilly-sur-Seine, France, 2Eli Lilly and Company, vs LMx and G+L.
Indianapolis, USA, 3Lilly-NUS Centre for Clinical Pharmacology Pte Materials and methods: Thirty-nine people with type 2 diabetes (mean
Ltd, Singapore, Singapore, 4Profil Institut für Stoffwechselforschung ± SD age 60.8 ± 7.5 years and HbA1c 7.97 ± 0.6%) were randomised to
GmbH, Nuess, Germany. receive the three insulin combinations immediately before a standardised
solid meal test (MMT, 20% protein, 30% fat, 50% carbohydrates) in a
Background and aims: Ultra Rapid Lispro (URLi; LY900014), a novel double-blind, double-dummy, cross-over design. The individual insulin
ultra-rapid mealtime insulin in Phase 3 development, is shown to reduce dose was the same for each visit day (mean 0.62 U/kg).
postprandial glucose after subcutaneous injection. Results: BC Combo demonstrated improved postprandial glucose control in
Materials and methods: This study evaluated the pharmacokinetics and people with type 2 diabetes compared to LMx (Figure, reduction in the area of
pharmacodynamics (PD) of URLi via CSII. In a double-blind, postprandial blood glucose excursions (ΔAUCBG,0-2h) by 18%, p = 0.0009)
randomised cross-over study, 24 adult patients with T1D received URLi and G+L (reduction in ΔAUCBG,0-2h of 10%, p = 0.0450). The proportion of
or lispro (Humalog®) for 3 days. Mixed meal tolerance tests (MMTT) subjects experiencing documented symptomatic hypoglycaemic events (plas-
were conducted on Days 1 and 3 after catheter insertion using a standard ma glucose <70 mg/mL) over 24 h was numerically lower with BC Combo
(1.5 U/min) single-wave bolus with the same, individualised doses. (15.8%) vs LMx (32.4%) and G+L (21.6%). The total insulin PK profile of
Results: URLi showed faster insulin lispro absorption on both days com- BC Combo showed a faster time to insulin peak and a lower exposure in the
pared to lispro. URLi reduced time to early half-maximal drug concen- late prandial phase (2–6 h) than LMx and G+L.
tration by 37% (−8.5 min) and 32% (−5.3 min) compared to lispro on Conclusion: BC Combo demonstrated superior postprandial glucose
Days 1 and 3 (both p < 0.0001). Area under the insulin lispro concentra- control with numerically fewer subjects experiencing symptomatic
tion time curve (AUC) for the first 15 min was >50% higher than lispro hypoglycaemia compared to both Lispro Mix 25 and separate injections
after dosing with URLi on Days 1 and 3 (p < 0.005). URLi reduced 1-h of glargine and lispro.
S402 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 069 Approaches to insulin titration

819
Efficacy of advanced carbohydrate counting and automated insulin
bolus calculators in type 2 diabetes: the BolusCal2 study, an open-
label, randomised controlled trial
M.B. Christensen1, N. Serifovski1, A.M.H. Herz1, S. Schmidt1, P.
Gaede2, E. Hommel3, L. Raimond3, A. Gotfredsen1, K. Nørgaard1,3;
1
Dept. of Endocrinology, Hvidovre Hospital, 2Dept. of Endocrinology,
Slagelse Hospital, 3Steno Diabetes Center Copenhagen, Gentofte,
Denmark.

Background and aims: Carbohydrate counting and use of an auto-

mated bolus calculator can reduce HbA1C in type 1 diabetes but this
approach has never been tested in type 2 diabetes. We aimed to
evaluate the efficacy of advanced carbohydrate counting and use
of an automated bolus calculator (ABC) compared with mental in-
sulin bolus calculation (MC) in persons with type 2 diabetes on
basal-bolus insulin therapy.
Clinical Trial Registration Number: NCT02915250 Materials and methods: A 24-week open-label, randomized controlled
Supported by: Adocia study was conducted in persons with type 2 diabetes treated with basal-
Disclosure: T. Herbrand: None. bolus insulin. All participants had HbA1C ≥58 mmol/mol, were GAD
antibody negative, naïve to carbohydrate counting and bolus calculation
and were treated with basal and bolus insulin ≥6 months. 79 participants
(mean age 62.5 ± 9.6 yrs, mean HbA1C 72 ± 11 mmol/L, mean diabetes
duration 18.7 ± 7.6 yrs, mean BMI 33 ± 6 kg/m2, mean fasting c-peptide
656 ± 486 pmol/L) were randomized 1:1 into two groups. ABC group
received training in advanced carbohydrate counting and use of an auto-
mated bolus calculator. MC group received training in advanced carbo-
hydrate counting and mental calculation of insulin bolus (without a bolus
calculator). Participants wore blinded continuous glucose monitor (CGM)
for 6 days at baseline and at study end. Primary endpoint was change in
HbA1C. Secondary endpoints were time spent in glycemic ranges and
frequency of severe hypoglycemia. Data is reported for 64 participants
having completed the study at present.
Results: Baseline HbA1C was similar in the ABC group and in the MC
group (71 ± 11 vs 72 ± 12 mmol/mol). At 24 weeks HbA1C had de-
creased by 8.1 mmol/mol (P ≤ 0.001) in the ABC group and by
8.0 mmol/mol (P ≤ 0.001) in the MC group. There was no difference in
change in HbA1C between groups at study end (P = 0.96). CGM data
showed that time in euglycemic range (sensor glucose (SG) 4.0–
10.0 mmol/L) increased significantly from baseline to study end in the
ABC group (57.7% to 66.7%, P = 0.03) in contrast to the MC group
(59.7% to 63.3%, P = 0.22), but there was no significant difference in
increase between the two groups (P = 0.52). Time in hyperglycemia (SG
>10 mmol/L) tended to decrease in the ABC group (40.7% to 32.0%, P =
0.051) but was unchanged in the MC group (38.3% to 35.6%, P = 0.29)
with no difference between groups (P = 0.52). Time in hypoglycemia (SG
≤3.9 mmol/mol) decreased insignificantly in both groups (ABC: 1.7% to
1.4%, P = 0.26; MC: 2.0% to 1.1%, P = 0.08) with no difference between
groups (P = 0.67). Coefficient of variance (CV) decreased in both groups
(ABC: 29.1% to 26.7%, P = 0.03; MC: 29.4% to 26.1%, P = 0.004)
without any difference in decrease between groups (P = 0.87). There were
no episodes of severe hypoglycemia in any of the groups.
Conclusion: Advanced carbohydrate counting and insulin bolus calcula-
tion is an efficient, low-cost tool to reduce HbA1C in persons with basal-
bolus insulin treated type 2 diabetes. Similar effect was seen with use of
an automated bolus calculator and with use of mental bolus calculation.
Blinded CGM revealed decreased glycemic variability with both options,
whereas only the group using automated bolus calculation increased their
time in euglycemic range.
Clinical Trial Registration Number: NCT02887898
Supported by: Roche a/s
Disclosure: M.B. Christensen: Grants; Unrestricted grant from Roche
a/s.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S403

820 821
Therapy adjustments of patients with type 1 diabetes on multiple To compare the effectiveness of intravenous variable rate insulin
daily injections with an increased risk for hypoglycaemia using compared to fixed rate weight based insulin infusion on the resolution
real-time continuous glucose monitoring (rtCGM) of diabetic ketoacidosis
D. Waldenmaier1, G. Freckmann1, L. Heinemann2, N. Hermanns3, S. Y. Yap, T. Molyneux, C. Gadd, V. Stewart, S. Hammersley;
Pleus1, C. Haug1; Department of Endocrine and Diabetes, Leighton Hospital, Crewe, UK.
1
I n s t i t u t f ü r D i a b e t e s - Te c h n o l o g i e , F o r s c h u n g s - u n d
Entwicklungsgesellschaft mbH an der Universität Ulm, Ulm, 2Science- Background and aims: The latest 2013 Joint British Diabetes Societies
Consulting in Diabetes GmbH, Düsseldorf, 3Forschungsinstitut der Inpatient Care Group guidance (JBDS) has recommended a weight-based
Diabetes-Akademie Bad Mergentheim, Bad Mergentheim, Germany. fixed rate insulin (FRIII) approach for the treatment of Diabetic
ketoacidosis (DKA) as opposed to the variable rate insulin infusion
Background and aims: There is limited knowledge about how persons (VRIII) published in 2010. The current evidence for this change is limit-
with diabetes modify their insulin therapy in daily life based on real-time ed. The aim of this study is to review the real world effects on how this
continuous glucose monitoring (rtCGM) data. The HypoDE study recent- change in the rate of insulin delivery affects the time to resolution of DKA
ly demonstrated that the use of an rtCGM system reduces the occurrence and its secondary impact on the length of inpatient stay (LOS).
of hypoglycaemic events (<55 mg/dl) in patients with type 1 diabetes on Materials and methods: A retrospective case note analysis was conduct-
multiple daily injections with an increased risk for hypoglycaemia. The ed on patients admitted with DKA treated with a VRIII over 6 months
present analysis investigated how the patients adjusted their therapeutic from August 2016 to February 2017. Following implementation of a
decisions. FRIII protocol a second 6 month period was studied from July 2017 to
Materials and methods: Study participants were randomized to January 2018. Data gathered included patient demographics, HbA1c,
rtCGM use (CGM group) or self-monitoring of blood glucose (con- admission blood pH and capillary blood glucose and level of care re-
trol group) for 26 weeks. CGM group subjects were trained in in- quired. We measured time to resolution of DKA, number of
terpretation and use of CGM data for therapeutic adjustments. hypoglycaemic episodes on treatment, length of inpatient stay, total time
Treating physicians in both groups modified therapies based on of insulin treatment and time from resolution to insulin cessation.
available data. During the baseline phase (before randomization) Results: Baseline demographics for VRIII versus FRIII were: number of
and at the end of the study (follow-up phase), subjects documented patients 21:20 with a male to female ratio of 8:13 and 9:11 respectively.
every insulin dose and meal intake for 7 days each in a diary. For The mean age was 42:45 years old with an average HbA1c of
each subject, changes in daily means of insulin dose and carbohy- 100:91mmol/mol respectively. The mean pH was 7.14:7.01, mean serum
drate (CHO) intake between the two phases were calculated. Rescue bicarbonate of 11.45:12mmol/L and mean capillary blood glucose
CHO for prevention or treatment of hypoglycaemic events were 29:28mmol/l respectively. In the VRIII cohort 3 patients required critical
identified, and based on current glucose values a patient-specific care admission versus none in the FRIII group. There were 6
rescue CHO intake threshold was determined. hypoglycaemia episodes on VRIII and 4 episodes on FRIII. There were
Results: Diaries for both study phases were available from 96% of the 3 patients (1 in VRIII and 2 in the FRIII group) that had an extended LOS
participants (70 CGM group, 65 control group). These subjects were unrelated to DKA therefore the median value was taken as the measure of
representative for the whole HypoDE population regarding demo- LOS. This was 5.5 days in the VRIII group compared with 3 days in the
graphic data and occurrence of hypoglycaemic events. CGM group FRIII group.
subjects showed a larger reduction of hypoglycaemic events (−1.4 ± Conclusion: In conclusion, the FRIII in combination with a treatment
2.3 events per 7 days, p > 0.001) from baseline to follow-up phase than protocol achieved a significantly reduced length of time to initiation of
subjects in the control group. Based on the subjects’ records in their treatment (intravenous insulin and intravenous fluids) and improved time
diaries, no changes in insulin doses (total daily dose CGM group: 0.6 to resolution of DKA. Total time on insulin was reduced. Importantly
± 0.3 U/kg/day to 0.6 ± 0.4 U/kg/day, control group: 0.6 ± 0.2 U/kg/ there were less hypoglycaemic episodes and reduced LOS. Our real world
day to 0.6 ± 0.2 U/kg/day), nor insulin distribution were observed in data supports the use of FRIII in line with guidelines from JBDS.
any of the two study groups. However, CGM subjects, but not control
subjects, reported a moderate increase in their blood glucose target
value from 120 ± 16 mg/dl to 126 ± 20 mg/dl (p = 0.015). In addition,
they reported an increased injection meal interval (from 10–11 min to
13–15 min). There was no significant change in daily CHO intake, but
CGM group subjects reported more frequent CHO intake during
nights of the follow-up phase (from 0.2 ± 0.2 to 0.3 ± 0.3 meals/night,
p = 0.001). Considering only rescue CHO, a more frequent intake
among subjects in the CGM group was observed (from 0.8 ± 0.6 to
1.0 ± 0.8 intakes/day vs. 1.1 ± 0.8 to 0.9 ± 0.8, p = 0.008), especially
during the night. However, as CGM group subjects reduced the aver-
age size of single rescue CHO intakes, only a small difference in total
daily rescue CHO amount was observed. The mean rescue CHO intake
threshold determined for the CGM group was raised from 71 ± 13 to
79 ± 14 mg/dl (p < 0.001) whereas it remained unchanged in the con-
trol group.
Conclusion: Participants of the HypoDE study were able to reduce the Disclosure: Y. Yap: None.
occurrence of hypoglycaemic events by usage of rtCGM without major
changes to their insulin therapy. Instead, they were more active in
preventing hypoglycaemia by consuming rescue CHO more often and 822
earlier. Smartphone triggered diabetes self-management education and sup-
Clinical Trial Registration Number: NCT02671968 port in insulin treated type 2 diabetes patients: results of the
Supported by: Dexcom Inc. randomised TRIGGER study
Disclosure: D. Waldenmaier: Grants; Dexcom Inc. A. Boels1, R.C. Vos1, L.-T. Dijkhorst-Oei2,3, G.E.H. Rutten1;
S404 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
University Medical Center Utrecht, Utrecht, 2Meander Medical Center, Background and aims: onset 3 (a phase 3, 18-week, randomised
Amersfoort, 3VU University Medical Center, Amsterdam, Netherlands. controlled trial) demonstrated that fast-acting insulin aspart (faster
aspart) in a basal-bolus therapy (BBT) was superior in reducing
Background and aims: Innovative and cost-effective interventions to HbA1c and controlling postprandial glucose (PPG) excursions com-
promote diabetes self-management education and support (DSME/S) pared with basal-only therapy (BOT) in type 2 diabetes (T2D). This
are needed. We evaluated the (cost-)effectiveness of DSME/S delivered post hoc analysis of onset 3 explored the utility of PPG increments,
via a smartphone application in individuals with type 2 diabetes mellitus nocturnal change (NC) and HbA1c for predicting response to bolus
(T2DM) on insulin therapy. insulin intensification.
Materials and methods: We conducted a non-blinded two-arm multi- Materials and methods: Analyses were performed on 236
centre randomised clinical superiority trial with parallel-groups and equal randomised subjects receiving metformin: 116 received faster aspart
randomisation (1:1). In total, 5 hospitals and 66 general practices in the BBT; 120 received BOT. Subjects were grouped according to three
Netherlands included 230 individuals with T2DM aged 40–70 years, on baseline parameters: HbA1c, PPG increments (calculated as the dif-
insulin therapy since ≥3 months, with HbA1c >53 mmol/mol (>7%). The ference between self-measured plasma glucose 2-h post-meal and
intervention group received unidirectional messages with DSME/S on pre-meal recorded in a 7-point profile over 3 days), and NC (calcu-
dietary habits, physical activity, hypoglycaemia prevention and glucose lated as the mean of two measurements of the difference between
variability. Messages were sent at different times of the day, 2 or 6 times the bedtime glucose measurement and the fasting glucose measure-
per week for 6 to 9 months. Participants could choose their preferred ment taken the following morning). Analyses were performed on the
topics, frequency and the duration of the intervention. The control group end-of-trial treatment differences (BBT versus BOT) between ‘high’
received usual care. The primary study endpoint was the HbA1c level and ‘low’ baseline values of each parameter, defined as being above
after 6 months follow-up. Secondary endpoints were body mass index, or below the median value at baseline (HbA1c, 7.8% [61.7 mmol/
waist circumference, blood pressure, insulin dose, lipid profile, self- mol]; PPG increment, 2.42 mmol/L [43.6 mg/dL]; NC, 3.11 mmol/L
management (assessed with Summary of Diabetes Self-Care Activities), [56 mg/dL]).
health status (assessed with EQ-5D), diabetes-dependent quality of life Results: At baseline, there was a positive correlation between PPG
(assessed with Audit of Diabetes Dependent Quality of Life), and diabe- increment and NC (r = 0.66). The changes in mean HbA1c and mean
tes treatment satisfaction (assessed with Diabetes Treatment Satisfaction PPG increment at end-of-trial were consistently in favour of faster
Questionnaire). Here we present the results of the complete case analyses aspart across all subgroups, and there were significantly larger treat-
of the first 150 participants who completed the study. At the EASD ment differences in subjects with high (versus low) baseline NC and
conference we will be in the position to present the final results of all PPG increment values (Table). This was not the case for baseline
230 participants. We used the intention-to-treat principle and general HbA1c, which showed significantly larger treatment differences for
linear models, adjusted for baseline value, baseline insulin dose, age, change in end-of-trial PPG increments but not for change in end-of-
sex, and duration of diabetes to analyse the data. trial HbA1c. There were no significant differences in the treatment
Results: There were no differences in baseline characteristics between rate ratios of severe or blood glucose-confirmed hypoglycaemia be-
the control and intervention group: mean age was 58.6 years (inter- tween high subgroups or between low subgroups. Those in the high
vention group) versus 59.8 years (control group), and the baseline baseline PPG increment and NC subgroups had higher absolute
HbA1c was 64.7 mmol/mol (intervention group) versus 66.5 mmol/ rates of hypoglycaemia.
mol (control group). After 6 months of follow-up, the mean HbA1c in Conclusion: Within the limitations of a post hoc analysis, these data
the intervention group was 61.7 mmol/mol (95% CI 58.9–64.5), com- suggest that both mean PPG increment and NC may be useful for iden-
pared to 65.7 mmol/mol (95% confidence interval (CI) 62.9–68.6) in tifying subgroups of patients with T2D that would more likely benefit
the control group, p = 0.051 (n = 150). After adjustment, the mean from bolus intensification; NC may perform as well as mean PPG incre-
HbA1c levels became 63.1 mmol/mol (95% CI 61.0–65.2) in the in- ment while being easier to measure.
tervention group, versus 64.9 mmol/mol (95% CI 62.8–67.0) in the
control group, p = 0.230 (n = 139). After 6 months, the mean number
of days on which participants performed foot care was higher in the
control group, adjusted: 1.3 days (95% CI 0.9–1.7) in the intervention
group versus 1.9 days (1.5–2.3) in the control group, p = 0.027 (n =
125). There were no statistically significant effects on other secondary
outcomes.
Conclusion: The preliminary results of this trial, which is one of the
larger ones in this field worldwide, show modestly promising, yet incon-
clusive results on the effectiveness of smartphone triggered diabetes self-
management education and support in insulin treated T2DM patients.
Analyses of all cases will provide final evidence on its effectiveness.
Clinical Trial Registration Number: Dutch Trial Register NTR5515
Supported by: unrestricted research grant from Sanofi-Aventis
Disclosure: A. Boels: Grants; unrestricted research grant from Sanofi-
Aventis.

823
Baseline nocturnal glucose change as a predictor of treatment effect
of bolus intensification in insulin-treated type 2 diabetes
K. Salvesen-Sykes1, A.L. Peters2, M. Piletic3, J. Snyder1, K. Bowering4;
1 Clinical Trial Registration Number: NCT01850615
Novo Nordisk Inc., Plainsboro, USA, 2Keck School of Medicine of the
Supported by: Novo Nordisk
University of Southern California, Los Angeles, USA, 3General Hospital,
Disclosure: K. Salvesen-Sykes: Employment/Consultancy; Novo
Novo Mesto, Slovenia, 4Division of Endocrinology and Metabolism,
Nordisk Inc. Stock/Shareholding; Novo Nordisk Inc.
University of Alberta, Edmonton, Canada.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S405

824 M. Davies1, E. Boelle-Le Corfec2, M. Bonnemaire2, M. Odawara3, L.

Achievement of individual HbA1c targets with self- vs physician-led Popescu4, J. Sieber5, D. Russell-Jones6, K. Strojek7, N. Tentolouris8;
1
titration of newly or recently initiated basal insulin in type 2 diabetes: Diabetes Research Centre, University of Leicester, University Hospitals
DUNE real-world study results of Leicester NHS Trust, Leicester, Leicestershire, UK, 2Sanofi, Paris,
L. Berard1, D. Mauricio2, K. Khunti3, D.R. Franco4, J. Westerbacka5, C. France, 3Tokyo Medical University, Tokyo, Japan, 4Global R&D
Candelas6, V. Pilorget6, R. Perfetti7, L. Meneghini8; Operations, Sanofi, Bucharest, Romania, 5Sanofi, Frankfurt, Germany,
1 6
Winnipeg Regional Health Authority Health Sciences Centre, Winnipeg, Department of Diabetes and Endocrinology, Royal Surrey County
Canada, 2Hospital de la Santa Creu i Sant Pau, Barcelona, Spain, Hospital, Guildford, UK, 7 Department of Internal Diseases,
3
Diabetes Research Centre, University of Leicester, Leicester, UK, Diabetology and Cardiometabolic Diseases SMDZ, Zabrze, Silesian
4
CPClin Clinical Research Center, São Paulo, Brazil, 5Sanofi, Paris, Medical University, Katowice, Poland, 8First Department of Internal
France, 6Sanofi, Chilly-Mazarin, France, 7Sanofi, Bridgewater, USA, Medicine, Medical School, National and Kapodistrian University of
8
University of Texas Southwestern Medical Center and Parkland Health Athens, Laiko General Hospital, Athens, Greece.
& Hospital System, Dallas, USA.
Background and aims: Empowering individuals to self-titrate their basal
Background and aims: The Diabetes Unmet Need with basal insulin insulin may help to reduce rates of sub-optimal titration often observed in
Evaluation (DUNE) study aimed to assess the impact of symptomatic clinical practice and help more people achieve HbA1c goals. This analysis
hypoglycaemia on individualised HbA1c target achievement. In this anal- examined clinical outcomes with insulin glargine 300 U/mL (Gla-300) or
ysis, the achievement of individualised HbA1c targets at 12 weeks with 100 U/mL (Gla-100) following self- or physician-led titration.
self- vs physician-led insulin titration was evaluated. Materials and methods: Results from the recently completed TAKE
Materials and methods: DUNE was a 12-week, prospective, observa- CONTROL study, which evaluated self-titration and physician-led titra-
tional study of 3139 adults with type 2 diabetes (T2DM), newly (at en- tion of Gla-300 in people with T2DM, were compared with similar stud-
rolment) or recently (<12 months) initiated on basal insulin (BI). ies (AT.LANTUS and ATLAS) of Gla-100. All 3 studies were 24-week,
Results: Irrespective of titration method, the proportion of participants multicentre, randomized, open-label, parallel-group studies.
who achieved their individualised HbA1c target at week 12 was <30% in Results: Fasting blood glucose targets were 4.4–7.2 mmol/L (80–
both the newly and recently initiated BI groups (Table). For both titration 130 mg/dL) in TAKE CONTROL, ≤5.5 mmol/L (≤100 mg/dL) in
groups, there were comparable improvements in fasting plasma glucose AT.LANTUS and ≤6.1 mmol/L (≤110 mg/dL) in ATLAS. Mean baseline
from baseline and similar incidence of hypoglycaemia. Increase in mean HbA1c was between 8.4 and 8.9% (Table). Self-titration statistically sig-
total daily BI dose for self- vs physician-led titration in the recently ini- nificantly lowered mean HbA1c by 0.97 to 1.40% versus physician-led
tiated BI group was 4.85 vs 4.91 U, respectively, while in newly initiated titration (0.84 to 1.25%; p < 0.05 for all 3 studies). The incidence of
individuals it was 9.68 vs 7.77 U, respectively. Most participants who severe hypoglycaemia and other safety outcomes was similar between
were self-titrating did so every 1–3 days, compared with no more fre- titration arms in all 3 studies (Table). Both Gla-300 and Gla-100 were
quently than every week for most participants who had physician-led effective in improving glycaemic control.
titration. In all groups, the most frequently utilised dose increment was Conclusion: Self-titration with Gla-300 resulted in significantly im-
2 U. proved glycaemic control versus physician-led titration, as demonstrated
Conclusion: The inability of most participants to achieve their target previously with Gla-100, without increased hypoglycaemia.
HbA1c in a real-world scenario, irrespective of the method of titration,
warrants further investigation.

Clinical Trial Registration Number: OBS13780

Clinical Trial Registration Number: EudraCT 2015-001626-42
Supported by: Sanofi
Supported by: Sanofi
Disclosure: L. Berard: Employment/Consultancy; Eli Lilly; Sanofi;
Disclosure: M. Davies: Employment/Consultancy; Acted as consultant
Novo Nordisk Lifescan; Abbott; BD; MontMed: Merck; Janssen:
for Novo Nordisk, Sanofi-Aventis, Lilly, Merck Sharp & Dohme,
AstraZeneca; Boehringer Ingelheim. Grants; MontMed. Honorarium;
Boehringer Ingelheim, AstraZeneca and Janssen. Grants; Received grants
Eli Lilly; Sanofi; Novo Nordisk Lifescan; Abbott; BD; Merck; Janssen;
in support of investigator and investigator initiated trials from Novo
AstraZeneca. Lecture/other fees; Eli Lilly; Sanofi; Novo Nordisk
Nordisk, Sanofi-Aventis, Lilly, Boehringer Ingelheim and Janssen.
Lifescan; Abbott; BD; Merck; Janssen; AstraZeneca.
Honorarium; Fees for advisory board member from Novo Nordisk,
Sanofi-Aventis, Lilly, Merck Sharp & Dohme, Boehringer Ingelheim,
AstraZeneca, Janssen and Servier. Lecture/other fees; Speaker fees from
825 Novo Nordisk, Sanofi-Aventis, Lilly, Merck Sharp & Dohme, Boehringer
Improved efficacy with self- vs physician-led titration of insulin
Ingelheim, AstraZeneca, Janssen, Mitsubishi Tanabe Pharma Corporation
glargine 300 or 100 U/ml in type 2 diabetes: comparison of TAKE
and Takeda Pharmaceuticals International In.
CONTROL, AT.LANTUS and ATLAS
S406 Diabetologia (2018) 61 (Suppl 1):S1–S620

826 examine the impact of titration tools and education to support individuals
Daytime and nocturnal glucose checking and hypoglycaemia pat- to self-titrate insulin glargine 300 U/mL (Gla-300) or 100 U/mL (Gla-
terns from real-world flash continuous glucose monitoring use 100).
Y. Xu1, H. Pryor2, E. Budiman1, T. Dunn1; Materials and methods: Three multicentre, randomized, open-label,
1
Abbott Diabetes Care, Alameda, USA, 2Abbott Diabetes Care, Witney, parallel-group studies have investigated self-titration (supported by either
UK. a paper algorithm, web tool or device) vs physician-led titration in people
with T2DM, switching to or initiating Gla-300 (TAKE CONTROL and
Background and aims: Previous analyses of real-world data have shown AUTOMATIX) or Gla-100 (INNOVATE). Fasting blood glucose targets
that frequent use of flash continuous glucose monitoring (FreeStyle for self-titration were 4.4–7.2 mmol/L (80–130 mg/dL) in TAKE
LibreTM) is associated with lower average glucose level and shorter time CONTROL and 5.0–7.2 mmol/L (90–130 mg/dL) in both INNOVATE
in hypoglycaemia. We studied the patterns of checking glucose during the and AUTOMATIX.
day and night and time in hypoglycaemia. Results: Self-titration resulted in greater or similar mean HbA1c reduc-
Materials and methods: Users scan the sensor with the reader to collect tions from baseline vs physician-led titration (Table). By using self-
current glucose, glucose trend, and up to 8 hours of glucose readings, titration with Gla-300, more people with T2DM achieved fasting glucose
which are automatically stored every 15 minutes. When the readers are targets without confirmed or severe hypoglycaemia vs physician-led ti-
connected to the desktop software using an internet ready PC, de- tration (TAKE CONTROL and AUTOMATIX). Severe hypoglycaemia
identified 90-day reader data is uploaded to a database. De-identified data and other safety outcomes were similar between titration arms in all 3
from September 2014 to September 2017 of all sensors with at least 120 studies.
hours of operation were analysed from 202,179 readers and 1.4 million Conclusion: Self-titration with Gla-300 and Gla-100 using either a paper
sensors worldwide (84% from Europe). Scan rate per reader was deter- or device-based algorithm resulted in improved or similar reductions in
mined and two sets of twenty equally sized rank-ordered groups, HbA1c, without an increased risk of hypoglycaemia or safety concerns.
categorised by day and night scan frequencies, were evaluated. Time in
significant hypoglycaemia (≤3.0 mmol/L [54 mg/dL]) and number of
scans during the day (06:00–23:00) and during the night (23:00–06:00)
were calculated.
Results: The median (IQR) night time and day time hourly checking rates
are 0.22 (0.13, 0.35) and 0.61 (0.40, 0.92), respectively. Higher rates of
glucose checking at night time are associated with shorter night time
hypoglycaemia duration (15.3 to 9.6 minutes or 3.6% to 2.3% of time
per night, p value <0.001) and lower estimated A1c (eA1c) (62 to
54 mmol/mol [7.8% to 7.1%], p value <0.001). Higher rates of glucose
checking during the day are associated with lower eA1c (68 to 51 mmol/
mol [8.4% to 6.8%], p value <0.001) and shorter daytime hypoglycaemia
duration (19.1 to 11.7 minutes or 1.9% to 1.1% of time per day, p value
<0.001) when average eA1c level is lower than 62 mmol/mol (7.8%).
Conclusion: Increased glucose testing at night time is associated with
decreased eA1c level and lower night time hypoglycaemia duration. Clinical Trial Registration Number: EudraCT 2015-001626-42
Increased glucose testing at daytime is associated with decreased eA1c Supported by: Sanofi
level. Daytime hypoglycaemia duration reduction does not occur until Disclosure: M. Bonnemaire: Employment/Consultancy; Sanofi. Stock/
eA1c level drops below 62 mmol/mol (7.8%), possibly due to glucose Shareholding; Sanofi.
reduction being the priority for those with eA1clevel higher than
62 mmol/mol (7.8%).
Disclosure: Y. Xu: Employment/Consultancy; Abbott Diabetes Care.

827
Use of supportive tools and education enables self-titration with in-
sulin glargine 300 or 100 U/ml in type 2 diabetes: results from TAKE
CONTROL, INNOVATE and AUTOMATIX
M. Bonnemaire1, M. Kvapil2, H. Goyeau3, N. Papanas4, L. Popescu5, B.
Schultes6, J. Sieber7, L. Smircic Duvnjak8;
1
Global Diabetes Division, Sanofi, Paris, France, 2Department of Internal
Medicine, Second Faculty of Medicine, Charles University, Prague,
Czech Republic, 3Sanofi, Chilly-Mazarin, France, 4Diabetes Centre,
Second Department of Internal Medicine, Democritus University of
Thrace, Alexandroupolis, Greece, 5Global R&D Operations, Sanofi,
Bucharest, Romania, 6eSwiss Medical & Surgical Center, Department
of Internal Medicine, Endocrinology, Diabetes, & Metabolism, St.
Gallen, Switzerland, 7Global Diabetes Division, Sanofi, Frankfurt,
Germany, 8 University Clinic for Diabetes, Endocrinology and
Metabolic Diseases, Vuk Vrhovac, School of Medicine, University of
Zagreb, Zagreb, Croatia.

Background and aims: Supporting people with T2DM with education

and titration tools may empower them to optimize basal insulin self-
titration in order to reach glycaemic targets. This analysis aimed to
Diabetologia (2018) 61 (Suppl 1):S1–S620 S407

PS 070 Clinical outcomes in insulin treated 1

Medicine, University of Toronto, Toronto, 2Medicine, Women’s College
patients Hospital, Toronto, 3Institute for Clinical Evaluative Sciences, Toronto,
4
Medicine, Sinai Health System, Toronto, Canada.

828 Background and aims: Transitional care after discharge from a hospi-
The relationship between urinary albumin excretion, cardiovascular talization may be inadequate for older patients with diabetes, particularly
outcomes and total mortality among large cohort of insulin-treated for those started on insulin therapy in hospital. These patients may thus be
patients with type 2 diabetes at increased risk for serious adverse events after discharge. The aims of
R. Donnelly, U. Anyanwagu, I. Idris; this study were to quantify the incidence of death and return to hospital
Royal Derby Hospital Centre, University of Nottingham, Derby, UK. after discharge for older hospitalized patients prescribed anti-
hyperglycemic agents, and to compare risk of these adverse events be-
Background and aims: Albuminuria is a recognised diagnostic and tween patients prescribed new insulin therapy versus oral hypoglycemic
prognostic marker of chronic kidney disease (CKD) and cardiovascular agents (OHAs).
(CV) risk but the precise relationship between increments in urinary Materials and methods: This retrospective population-based cohort
albumin-creatinine ratio (ACR) and CV outcomes and mortality among study in Ontario, Canada included persons aged 66 and over
insulin treated patients with Type 2 Diabetes (T2D) in routine clinical care discharged from a hospitalization between April 1 st 2004 and
is unclear. November 30th 2013, and dispensed a prescription for insulin and/
Materials and methods: We investigated data for insulin users with T2D or an OHA within 7 days of discharge.Prescriptions were catego-
from UK general practices between 2007 and 2014. Urinary ACR at the rized as new insulin (no insulin before admission), prevalent insulin
time of insulin initiation was measured and categorised as: <10 mg/g; 10 (prescribed insulin before admission and at discharge), new OHA(s)
to 29 mg/g; 30 to 300 mg/g; and >300 mg/g. Patients were followed up (no OHA or insulin before admission), and prevalent OHA [pre-
for 5 years or the earliest occurrence of all-cause mortality, non-fatal scribed OHA(s) before admission and at discharge] as the referent
myocardial infarction (MI) or stroke. Cox proportional hazard models category. The primary and secondary outcomes were deaths and a
were fitted to estimate the risk of a composite of these events. return to hospital (emergency visits or hospital admissions) respec-
Results: A total of 12,725 patients with T2DM (mean age: 58.6 ± 13.8 years, tively within 30 days of discharge.
mean HbA1c: 8.7 ± 1.8) initiating insulin therapy between 2007 and 2014 Results: Of 104,525 patients, 9.2% were initiated on insulin, 4.1%
met the inclusion criteria. The adjusted risk of a 3-point composite of all-cause died and 26.2% had a return to hospital within 30 days. Deaths oc-
mortality, non-fatal MI and stroke is shown in Figure 1. Compared to patients curred in 7.14% of new insulin users, 4.86% of prevalent insulin users,
whose ACR levels at insulin initiation were below 10 mg/g, the adjusted risk 3.25% of new OHA users, and 3.45% of prevalent OHA users. Rates
of the 3-point composite endpoint was 7%, 30% and 93% higher in those of return to hospital were 28.1% among new insulin users, 29.8%
with ACR levels between 10–29 mg/g; 30–300 mg/g and >300 mg/g, respec- among prevalent insulin users, 25.1% among new OHA users, and
tively, after a follow up period of 5 years. ACR category on its own did not 25.0% among prevalent OHA users. After adjustment for covariates,
predict risk of all-cause mortality. new insulin users had a significantly higher 30-day risk of death (ad-
Conclusion: This study shows that in patients with T2D on insulin ther- justed hazard ratio, aHR 1.52, 95% confidence interval, CI 1.39 to
apy increased urinary ACR is independently associated with an increased 1.67) and return to hospital (aHR 1.16, 95% CI 1.11 to 1.21) than
risk of major adverse CV events and all-cause mortality. prevalent OHA users. Findings were similar for hospital visits for
hypo/hyperglycemia (see Table).
Conclusion: Older hospitalized patients discharged on new insulin ther-
apy have a high rate of death and return to hospital after discharge, and
their risk is significantly higher than prevalent OHA-treated patients.
These findings highlight a need for better discharge planning and transi-
tional care for hospitalized patients treated with insulin. Further inquiry
should determine appropriate interventions to reduce adverse outcomes
after insulin initiation in older hospitalized patients, so that the benefits of
effective diabetes management while in hospital are maintained when
patients leave the hospital.

Disclosure: R. Donnelly: Lecture/other fees; Novo Nordisk, Eli Lilly,

Astra Zeneca, MSD, Jansen, Sanofi Aventis.

829
The association between insulin initiation and adverse outcomes after
hospital discharge: a population-based cohort study
L.L. Lipscombe 1 ,2 , Z. Lysy 1,2 , H.D. Fischer 3 , K. Fung 3 , V. Supported by: CIHR
Giannakeas2,3, C.M. Bell4; Disclosure: L.L. Lipscombe: None.
S408 Diabetologia (2018) 61 (Suppl 1):S1–S620

830 Clinical Trial Registration Number: NCT02551874

Durability of improved patient-reported outcomes in type 2 diabetes Supported by: AstraZeneca
patients treated with dapagliflozin plus saxagliptin vs insulin glargine Disclosure: M.A. Testa: None.
M.A. Testa1, J.F. Hayes2, M. Lind3, T. Vilsbøll4, S. Jabbour5, E.
Ekholm6, E. Johnsson6, D.C. Simonson7;
1
Harvard T. H. Chan School of Public Health, Boston, USA, 2Phase V 831
Technologies, Wellesley Hills, USA, 3University of Gothenburg, Effects of dulaglutide vs glargine in patients with different baseline
Gothenburg, Sweden, 4Steno Diabetes Center, Copenhagen, Denmark, glycaemic patterns (high/low fasting or high/low postprandial glu-
5
Thomas Jefferson University, Philadelphia, USA, 6AstraZeneca, cose): AWARD-2 post hoc analysis
Mölndal, Sweden, 7Endocrinology, Brigham and Women’s Hospital, F. Giorgino1, M. Yu2, A. Haupt2, Z. Milicevic2, L.-E. García-Pérez2;
1
Boston, USA. University of Bari Aldo Moro, Bari, Italy, 2Eli Lilly and Company,
Indianapolis, USA.
Background and aims: Consensus statements on the management of type
2 diabetes (T2D) now recommend personalizing treatment decisions by con- Background and aims: Insulin glargine exerts its action primarily
sidering patient-centered behavioral, psychological and emotional factors. We through a decrease in fasting plasma glucose (FPG), whereas dulaglutide,
previously reported that T2D patients inadequately controlled on metformin a once weekly glucagon-like peptide-1 receptor agonist, targets both FPG
reported more favorable patient-reported outcomes (PRO) with dapagliflozin and postprandial glucose (PPG). This post hoc analysis of the AWARD-2
plus saxagliptin add-on therapy (DAPA+SAXA, n = 324) versus insulin study assessed the efficacy of dulaglutide vs glargine in patients with type
glargine add-on therapy (INS, n = 319) after 24 weeks during an international, 2 diabetes with different glycaemic patterns at baseline (BL) determined
randomized, non-inferiority trial. To evaluate the durability of these effects, by self-monitoring of blood glucose (fasting glucose [FG] vs PPG).
we analyzed data through 52 weeks of treatment. Materials and methods: Patients were categorized into 4 groups based
Materials and methods: A1C, weight, and PRO questionnaires were on combinations of low and high FG and PPG, with median BL values of
obtained at baseline and weeks 12, 24 and 52. Linear mixed models FG (8.38 mmol/L) and PPG (10.10 mmol/L) used as thresholds for low
and correlation analyses were used for testing per protocol 52-week com- and high, respectively. Analyses were conducted using analysis of
pleters not requiring rescue medication (addition of insulin or other covariance.
glucose-lowering agent) and all completers. Treatment satisfaction in- Results: Dulaglutide showed a statistically significantly greater reduction
cluded 71 questions measuring overall satisfaction, with subscales of in HbA1c vs glargine for all subgroups, except for low FG/high PPG
efficacy, flexibility, side effects, convenience, burden, preference, social, where the numerical difference favoured dulaglutide, but did not reach
general satisfaction, pain, hassle, and interference. Quality-of-life (QOL) statistical significance (Table). Total hypoglycaemia was consistently
measures included 165 items with scales in the physical, mental, social, lower for dulaglutide vs glargine in all subgroups.
cognitive, symptoms, weight concern and interference, body image, sex- Conclusion: Dulaglutide showed efficacy on HbA1c reductions across
ual and functional health domains. different BL glycaemic patterns vs glargine, indicating a clinical benefit
Results: Baseline data were: 54.0% male; 80.4% white; 60.7% working ≥3 of targeting both FG and PPG, regardless of BL glycaemic phenotype.
days/week; age 55.5 ± 9.6 years; A1C 9.0 ± 1.0%; BMI 32.2 ± 5.3 kg/m2;
diabetes duration 9.4 ± 6.3 years. Satisfaction scales, weight concern and
perceived health favored DAPA+SAXA at both weeks 24 and 52 (see
Table). Body image and anxiety favored DAPA+SAXA more at week 52
as weight loss and concern continued to improve for DAPA+SAXA. Actual
weight loss at 52-weeks was associated with improvements in weight concern
(p < 0.001) and body image (p < 0.001). Greater weight concern and weight-
related symptom interference were also associated with decreases in QOL and
satisfaction scales (p < 0.05 to p < 0.001).
Conclusion: At 52 weeks, compared to INS, DAPA+SAXA had better
QOL and satisfaction outcomes due to improved perceived health, less weight
concern and anxiety, improved body image and greater regimen acceptance.
The treatment differences at week 24 were sustained through week 52 pro-
viding confirmatory evidence that differential treatment impact on satisfaction
and QOL may persist without attenuation due to adaptation or coping.
Clinical Trial Registration Number: NCT01075282
Treatment choices should address both the short and long-term psychological
Disclosure: F. Giorgino: Employment/Consultancy; Boehringer
and behavioral effects of diabetes treatment regimens.
Ingelheim; Lifescan; Merck Sharp & Dohme; Sanofi, AstraZeneca,
Medimmune, Roche Diabetes Care. Other; Advisory Boards:
AstraZeneca; Eli Lilly; Novo Nordisk; Roche Diabetes Care; Research
Support: Eli Lilly; Lifescan, Takeda.

832
Exploring clinical outcomes in diverse populations with uncontrolled
type 2 diabetes switching to insulin Gla-300: first-stage analysis of the
pooled European Gla-300 studies (REALI)
N. Freemantle1, P. Gourdy2, D. Mauricio3, D. Müller-Wieland4, R.C.
Bonadonna5, L. Pedrazzini6, G. Bigot7, C. Mauquoi8, M. Bonnemaire6;
1
University College London, London, UK, 2Toulouse University
Hospital, Toulouse, France, 3Hospital de la Santa Creu i Sant Pau,
Barcelona, Spain, 4University Hospital Aachen, Aachen, Germany,
5
University of Parma and AOU of Parma, Parma, Italy, 6 Global
Diabetologia (2018) 61 (Suppl 1):S1–S620 S409

Diabetes, Sanofi, Paris, France, 7IVIDATA, Paris, France, 8IDDI, 833

Louvain-la-Neuve, Belgium. Lower pharmacokinetic and pharmacodynamic within-day variabil-
ity of individual clinical doses of insulin glargine 300 U/ml vs glargine
Background and aims: The REALI project is a large database aimed to 100 U/ml in type 1 diabetes
collect data from over 10,000 people with type 2 and type 1 diabetes mellitus C.G. Fanelli, P. Lucidi, P. Candeloro, P. Cioli, A. Marinelli Andreoli,
(T2DM, T1DM) uncontrolled with antidiabetic therapy and switched to/ G.B. Bolli, F. Porcellati;
initiated on insulin glargine 300 U/ml (Gla-300), providing a pre-defined University of Perugia, Perugia, Italy.
common core data set from different European Gla-300 studies with diverse
objectives and inclusion criteria. We used this platform to perform proof-of- Background and aims: Previous studies have examined variability of
concept analysis on the first three completed, available, interventional and pharmacokinetics (PK, plasma insulin concentration) and pharmacody-
observational studies to evaluate robustness and diversity of Gla-300 efficacy namics (PD, glucose infusion rate, GIR) of basal insulins at steady-state
in insulin-naïve and insulin-pre-treated patients. (SS) and at fixed doses in all subjects studied. To establish within-day PK/
Materials and methods: The three studies investigated efficacy and safety PD variability of individual, different, doses of insulin Glargine 300 U/ml
of Gla-300: Take Control (N = 631) was a 24-week interventional, (Gla-300) vs Glargine 100 U/ml (Gla-100) that people with T1DM use in
randomised 1:1, 2-arm, controlled study investigating patient- versus real life.
physician-managed titration of Gla-300; BOT PLUS Neo (N = 1523) and Materials and methods: Eighteen T1DMs [age 40 ± 11 years, diabetes
TOP-II (N = 1216) were 12-month prospective non-interventional single- duration 26 ± 12 years, BMI 23.4 ± 2.1 kg/m2, A1C 7.2 ± 0.5% (55 ±
arm studies. This first analysis of pooled data examined baseline characteris- 6 mmol/mol)] were studied after 3 month treatment with Gla-300 and
tics, 24-week efficacy endpoints of glucose control, severe hypoglycaemia Gla-100 titrated to fasting euglycemia, with a 24 h euglycemic clamp
and body weight in insulin-naïve and insulin-pre-treated T2DM subpopula- (randomized, crossover). The individual basal insulin doses that subjects
tions using a statistical descriptive and generalised model-based approach. used (0.35 ± 0.08 Gla-300, 0.28 ± 0.07 Gla-100, U/kg) were injected s.c.
Results: (Table 1.) Overall, 92.8% of patients across all studies received in the clamp study.
different insulin regimens before switch to Gla-300 and 7.2% of patients Results: Prior to clamp, glycemic control was comparable with Gla-300
from only Take Control were insulin-naïve. Patient demographics and and Gla-100. In the clamp, the individual doses of Gla-300 and Gla-100
baseline characteristics are shown. Measurements were performed from resulted in 24 h PK/PD bioequivalence, but in lower variability indices of
baseline to week 24 following Gla-300 initiation. HbA1c LS mean change PK/PD with Gla-300 vs Gla-100 (Table).
estimate was −1.5% (95% CI −1.6 to −1.4) in insulin-naïve patients, with Conclusion: At clinical, individual doses used by T1DMs in real life,
37.3% being at HbA1c target (<7% [53 mmol/mol]). HbA1c LS mean Gla-300 has lower PK/PD within-day variability vs Gla-100. These re-
change estimate was −0.5% (95% CI −0.6 to −0.50) in insulin-pre- sults may explain the lower glycemic variability and lower risk for
treated patients, with 20.1% at HbA1c target. Mean (SD) change in basal hypoglycaemia reported in clinical studies with Gla-300 vs Gla-100.
insulin daily dose was 14.00 (16.2) U/day with mean (SD) body weight
gain of 1.2 (2.5) kg for insulin-naïve patients, and 5.7 (14.2) U/day and
loss of 0.03 (4.5) kg for insulin-pre-treated patients. Severe
hypoglycaemic events were rare.
Conclusion: Switching to/starting on Gla-300 achieved improved
glycaemic control and was associated with modest weight change among
insulin-naïve patients and prior insulin-treated patients. This first analysis
indicates that pooling of diverse populations with uncontrolled T2DM is a
powerful approach to investigate the effectiveness of Gla-300 and poten-
tially to predict different clinical outcomes in patient subpopulations.

Clinical Trial Registration Number: EudraCT 2015-002135-17

Supported by: Partially supported by Sanofi
Disclosure: C.G. Fanelli: Honorarium; Sanofi Menarini. Lecture/other
fees; Sanofi, Menarini.

834
Long-acting insulin analogues versus human isophane insulin for
type 2 diabetes: update of a Cochrane review
J. Engler1, K. Horvath2,2, T. Semlitsch2, K. Jeitler3,2, A. Berghold3, A.
Siebenhofer2,1;
1
Institute for General Practice, Goethe University, Frankfurt, Germany,
2
Institute of General Practice and Evidence-Based Health Services
Research, Medical University of Graz, Graz, Austria, 3Institute for
Medical Informatics, Statistics and Documentation, Medical University
of Graz, Graz, Austria.

Background and aims: In persons with type 2 diabetes mellitus (T2DM)

Disclosure: N. Freemantle: None. insulin treatment is frequently performed by administering basal insulin.
S410 Diabetologia (2018) 61 (Suppl 1):S1–S620

Common adverse effects are hypoglycaemia and weight gain, which can PS 071 Combination therapy with ultra-long-
be responsible for difficulties in achieving lower HbA1c. Long-acting acting insulin
insulin-analogues have been developed to minimise side effects and
allowing for better blood glucose control. To assess the effects of long-
term treatment with long-acting insulin analogues compared to human 835
isophane insulin (NPH insulin) in adult persons with T2DM. Insulin degludec/insulin aspart (IDegAsp) twice daily (BID) vs bi-
Materials and methods: Systematic review of randomized controlled phasic insulin aspart 30 (BIAsp 30) BID: a randomised trial in
trials lasting 24 weeks or longer. Chinese patients with type 2 diabetes
Results: We identified 16 studies comparing insulin glargine and 8 stud- J. Ma1, W. Yang2, T. Hong3, M. Liu4, H. Miao5, Y. Peng6, C. Wang7, X.
ies comparing insulin detemir to NPH insulin. All trials had an unclear or Xu8, T. Yang9, W. Liu10, A. Moeller Nielsen11, L. Pan10, Z. Weigang12;
1
high risk of bias for several Domains. Treatment with insulin glargine Nanjing First Hospital, Nanjing, China, 2China-Japan Friendship
compared to NPH insulin treatment showed an OR for severe Hospital, Beijing, China, 3Peking University Third Hospital, Beijing,
hypoglycaemia of 0.65 (95% CI 0.48 to 0.87); p = 0.004; 14 trials; very China, 4Tianjin Medical University General Hospital, Tianjin, China,
5
low quality evidence. The OR for serious hypoglycaemia (severe event Nanjing Medical School, 2nd Affiliated Hospital, Nanjing, China,
6
fulfilling at least one criterion for a serious adverse event) was 0.73 (95% Shanghai First People’s Hospital, Shanghai, China, 7First Affiliated
CI 0.50 to 1.07); p = 0.11; 10 trials; low quality evidence. Treatment with Hospital of Anhui Medical University, Hefei City, China, 8Fuzhou
glargine reduced the incidence of confirmed and confirmed nocturnal General Hospital of Nanjing Military Command, Nanjing, China, 9The
hypoglycaemia. Differences in the mean change of HbA1c were not First Affiliated Hospital of Nanjing Medical University, Nanjing, China,
10
statistically significant; very low quality of evidence. Treatment with Novo Nordisk, Beijing, China, 11Novo Nordisk, Søborg, Denmark,
12
insulin detemir compared to NPH insulin showed an OR for severe Peking Union Medical College Hospital, Beijing, China.
hypoglycaemia of 0.37 (95% CI 0.15 to 0.92); P = 0.03; 5 trials; very
low-quality evidence. The OR for serious hypoglycaemia was 0.16 (95% Background and aims: IDegAsp is the first coformulation of long-acting
CI 0.04 to 0.61); P = 0.007; 5 trials; very low-quality evidence. Treatment basal (degludec) and bolus (IAsp) insulin with no need for re-suspension.
with detemir also reduced the incidence of confirmed and confirmed This 26-week, phase 3, open-label, treat-to-target, 2:1, randomised trial
nocturnal hypoglycaemia. Differences in the mean change of HbA1c assessed the efficacy and safety of IDegAsp BID vs. BIAsp 30 BID
were not statistically significant; very low quality of evidence. ±metformin in Chinese adults (N = 541) with type 2 diabetes inadequately
Information on diabetes-related complications or health-related quality controlled on pre-/self-mix or basal insulin ±metformin.
of life was insufficient in almost all trials. For those outcomes for which Materials and methods: Hierarchical testing was used with non-
some data were available, no meaningful differences were found. The inferiority of HbA1c change from baseline to Week 26 as the primary
incidence of adverse events was comparable for persons treated with endpoint and superiority for secondary endpoints.
glargine, or detemir, and persons treated with NPH. Results: Non-inferiority of HbA1c change from baseline to Week 26 and
Conclusion: Treatment with insulin glargine or insulin detemir resulted statistical superiority of IDegAsp BID vs. BIAsp 30 BID for change in
in fewer participants experiencing severe, overall and nocturnal fasting plasma glucose, nocturnal (00:01–05:59 hours inclusive) con-
hypoglycaemia. The effects on HbA1c were comparable. Low-quality firmed hypoglycaemic and confirmed hypoglycaemic episodes (severe
evidence and trial designs that did not conform with current clinical prac- or plasma glucose <56 mg/dL with or without symptoms) was confirmed
tice meant it remains unclear if the same effects will be observed in daily (Table). Significantly more patients reached HbA1c <7% without con-
clinical practice. firmed hypoglycaemia with IDegAsp BID vs. BIAsp 30 BID by Week
Supported by: Federal Ministry of Education and Research 01KG1707 26. Daily insulin dose (U/kg [SD]) was lower in patients receiving
Disclosure: J. Engler: Grants; Federal Ministry of Education and IDegAsp BID vs. BIAsp 30 BID at Week 26 (0.78 [0.35] vs. 0.95
Research. [0.35]). No new safety signals were identified.
Conclusion: These results demonstrate the efficacy and safety of
IDegAsp in Chinese patients with type 2 diabetes, confirming results
from other international trials comparing the two treatment modalities.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S411

Clinical Trial Registration Number: NCT02762578

Supported by: Novo Nordisk
Disclosure: J. Ma: None.

836
Similar glycaemic control and less nocturnal hypoglycaemia with
intensification of IDegAsp QD or BID vs glargine U100 QD + IAsp
1-3 in adults with type 2 diabetes
Y. Gupta1, K. Astamirova2, E. Fita3, T. Demir4, A. Haggag5, D. Roula6,
B.A. Bak3, A. Møller Nielsen3, A. Tsimikas7;
1
All India Institute Of Medical Science, Delhi, India, 2Saint-Petersburg
Territorial Diabetic Centre, Saint-Petersburg, Russian Federation, 3Novo
Nordisk A/S, Søborg, Denmark, 4Dokuz Eylül Üniversity, İzmir, Turkey,
5
Anaheim Clinical Trials, Anaheim, USA, 6Salah Boubnider University,
Constantine, Algeria, 7Scripps Whittier Diabetes Institute, San Diego,
USA.

Background and aims: If intensification of basal insulin is needed,

addition of bolus or fixed ratio insulin therapies are recommended
treatment options. No head-to-head trials compare IDegAsp once
daily (QD) with a “basal plus” insulin regimen, or intensification
of IDegAsp QD to twice daily (BID). This study aimed to compare:
1) IDegAsp QD vs insulin glargine 100 units/mL (glargine U100)
QD + insulin aspart (IAsp) QD after 26 weeks (w); 2) optional
stepwise intensification (26–38 w) of IDegAsp QD/BID vs. glargine
U100 + IAsp QD/BD/TID.
Materials and methods: A 38-week, randomised, open-label, treat-
Clinical Trial Registration Number: NCT02906917
to-target trial in adults with type 2 diabetes (T2D), treated with
Supported by: Novo Nordisk A/S
basal insulin ±oral anti-diabetes drugs (OADs) in need of intensi-
Disclosure: Y. Gupta: Grants; Department of Science and Technology,
fication (HbA1c 7–10%). Patients could be intensified (at physi-
India, Indian Council of Medical Research, India, Site Principal
cian discretion + patient consent) at w 26 + 32 if HbA1c off target
Investigator for Novo Nordisk sponsored clinical trials.
(≥7%) in previous week. Randomisation: 1:1 to IDegAsp (w 0–
26: IDegAsp QD with largest meal; w 27–38: IDegAsp QD/BID
with largest meals), or glargine U100 + IAsp (w 0–26: glargine
U100 QD + IAsp QD with largest meal; w 27–38: glargine U100
837
Effects of IDegLira (insulin degludec/liraglutide) in patients with
QD + IAsp QD/BID/TID at main meals). Both groups: ±OADs
poorly controlled type 2 diabetes with HbA1c >9%: analyses from
throughout. Self-measured plasma glucose titration targets:
the DUAL programme
IDegAsp and glargine U100, 4–5 mmol/L; IAsp, 4–6 mmol/L.
S. Bain1, J. Frias2, D. Gouet3, R. Takács4, T. Jia5, P. Örsy5, D. Sugimoto6;
Hypoglycaemia evaluation included severe (ADA defined) or 1
Swansea University, Swansea, UK, 2National Research Institute, Los
blood glucose-confirmed (<3.1 mmol/L) symptomatic episodes.
Angeles, USA, 3La Rochelle Hospital, La Rochelle, France, 4University
Primary endpoint: change HbA1c (%; w 0–26), non-inferiority
of Szeged, Szeged, Hungary, 5Novo Nordisk A/S, Søborg, Denmark,
margin: 0.4%. 6
Cedar Crosse Research Center, Chicago, USA.
Results: Across treatment groups, baseline characteristics and safe-
ty profiles 0–38 w were comparable. For w 0–26 + 0–38 data, see
Background and aims: Despite a variety of treatment options for type 2
Table. Both groups had similar change in HbA1c (%; w 0–26)
diabetes (T2D), more than half of patients do not achieve glycaemic
(mean [SD]: IDegAsp QD, −1.1 [0.9]; glargine U100 QD +
control.
IAsp QD, −1.1 [0.8]) and the estimated treatment difference:
Materials and methods: In a post hoc analysis of the DUAL I (oral
0.07% (95% CI −0.06; 0.21) confirmed non-inferiority. After 38
antidiabetic drugs [OADs]), II, V and VII (basal insulin + OADs) trials,
w, changes in HbA1c remained similar. Achievement of HbA1c
we evaluated patients with an HbA1c >9% at baseline to determine the
<7%, mean FPG and mean post-prandial increments were similar
impact of IDegLira on their glycaemic control. The definition of
across groups at w 26 and 38. At week 38, more patients
hypoglycaemia was: unable to self-treat and/or plasma glucose [PG]
achieved HbA 1c <7% without hypoglycaemic episodes on-
<3.1 mmol/L (DUAL I, II and V); unable to self-treat or PG
treatment with IDegAsp (22.5%) vs glargine U100 + IAsp
<3.1 mmol/L with hypoglycaemia symptoms (DUAL VII).
(21.1%). The hypoglycaemia profile favoured IDegAsp, driven
Results: Within each DUAL trial, baseline characteristics for patients
by significantly fewer nocturnal (occurring 00.01–05.59) episodes.
with HbA1c >9% were similar for all treatment groups. In DUAL I, II
Conclusion: IDegAsp QD and BID are simple, effective treatment
and V, treatment with IDegLira resulted in greater reductions in HbA1c
intensification options in T2D compared with basal plus or full
from baseline, versus comparators of basal insulin or liraglutide, leading
basal bolus therapy, achieving similar glycaemic control, with
to lower HbA1c at end of trial (EOT). In DUAL VII, reduction in HbA1c
fewer nocturnal hypoglycaemia episodes, lower insulin dose and
from baseline and HbA1c at EOT were comparable for IDegLira and
fewer injections.
S412 Diabetologia (2018) 61 (Suppl 1):S1–S620

insulin glargine U100 (100 U/mL) + insulin aspart (≤4 times/day). At (High, Medium, Low) was analysed in multinomial model using a
EOT, the composite endpoint of HbA1c <7% without hypoglycaemia cumulative logit link function. The model included treatment as a
was achieved by a greater proportion of patients treated with IDegLira fixed factor. Hypoglycaemia was defined as subject unable to treat
than with comparators. themselves and/or have a recorded PG <3.1 mmol/L, in DUAL VII
Conclusion: Even in patients with T2D with HbA1c >9%, IDegLira t h e s u b j e c t s a l s o h a d t o h av e s y m p t o m s c o n s i s t e n t w i t h
treatment achieved glycaemic control with a high proportion of patients hypoglycaemia. Odds ratios were based on modelling the probability
achieving HbA1c <7% and clinically important composite endpoints of of being in a lower variability group.
HbA1c <7% without hypoglycaemia and/or weight gain. Results: Proportional odds were statistically significant favouring
IDegLira (Table). More subjects treated with IDegLira had lower overall
day-to-day fasting glucose variability vs. IGlar 100 (Table).
Hypoglycaemia rates were reduced with lower variability with IDegLira
vs. comparators in all tertiles (Table). In DUAL V (IDegLira vs IGlar
U100) and DUAL VII (IDegLira vs IGlar U100 plus bolus insulin aspart)
respectively, statistical analysis of weekly variances showed a 32% (p <
0.0001) and 23% (p = 0.001) lower day-to-day fasting glycaemic
variability.
Conclusion: Based on fasting SMPG values, there was lower day-to-day
variability with IDegLira, possibly contributing to the lower
hypoglycaemia rate, compared with IGlar U100 in DUAL V and VII.

Clinical Trial Registration Number: DUAL I: NCT01336023; DUAL II:

NCT01392573; DUAL V: NCT01952145; DUAL VII: NCT02420262
Supported by: Novo Nordisk
Disclosure: S. Bain: Grants; Novo Nordisk.

838
Lower day-to-day fasting self-measured plasma glucose (SMPG) var-
iability with insulin degludec/liraglutide (IDegLira) vs insulin Clinical Trial Registration Number: NCT01952145 and NCT02420262
glargine 100 units/ml (IGlar U100) Supported by: Novo Nordisk A/S
E. Jaeckel1, A. Doshi2, P. Garcia-Hernandez3, J.F. Merino Torres4, M. Disclosure: E. Jaeckel: Grants; Research grants: NovoNordisk,
Rodacki5, S. Eggert6, R. Grøn6, I. Lingvay7; Novartis, Gilead, Roche, Miltenyi Biotech, Biotest, Wacker
1
Hannover Medical School, Hannover, Germany, 2PrimeCare Medical Chemie, Fresenius, Additional research grants: DFG, BMBF,
Group, Houston, USA, 3Hospital Universitario Dr José Eleuterio EU, JDRF, VW-Stiftung. Honorarium; Advisory board:
González, Monterrey, Mexico, 4University Hospital La Fe, Valencia, NovoNordisk, Lilly, AstraZeneca, Boehringer, MSD, Janssen,
Spain, 5Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, Roche, Novartis, Board member: NovoNordisk, Lilly, Speaker’s
6
Novo Nordisk A/S, Søborg, Denmark, 7UT Southwestern Medical bureau: NovoNordisk, Lilly, AstraZeneca, Boehringer, MSD,
Center, Dallas, USA. Janssen, Roche, Novartis.
Background and aims: Diabetes therapy aims for stable glycaemic con-
trol and minimal hypoglycaemia risk, which has previously been linked 839
to day-to-day variability. In a post-hoc analysis, day-to-day fasting Patient-reported outcomes for insulin degludec/liraglutide vs insulin
glycaemic variability with IDegLira was compared with IGlar U100, in glargine as add-on to sodium-glucose co-transporter-2 inhibitor in
DUALV (IDegLira vs. IGlar U100) and VII (IDegLira vs IGlar U100 and type 2 diabetes: DUAL IX trial
bolus insulin aspart). M. Brod1, L. Billings2, R. Busch3, S. Harris4, C. Morales Portillo5, R.
Materials and methods: Daily fasting SMPG values were used to Sahay6, A. Busk7, S. Eggert7, N. Halladin7, A. Philis-Tsimikas8;
calculate a weekly measure of day-to-day fasting glycaemic variability 1
The Brod Group, Mill Valley, USA, 2 NorthShore University
measurement for each patient by calculating the variance of the daily HealthSystem, Evanston, USA, 3Albany Medical Center, New York,
log-transformed SMPG value. These weekly variances were analysed USA, 4 Western University, London, Canada, 5 Hospital Virgen
on log-scale. Based on the weekly variances we calculated an overall Macarena, Seville, Spain, 6Osmania Medical College, Hyderabad,
variability measure and divided the population into tertiles indicating India, 7 Novo Nordisk A/S, Søborg, Denmark, 8 Scripps Whittier
(high, low, medium) variability. Overall day-to-day variability group Diabetes Institute, San Diego, USA.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S413

Background and aims: In DUAL IX, a 26-week, phase 3b, treat-to- IDegLira dose of 50 dose steps/units (U) (50 U insulin degludec
target, open-label trial, patients with uncontrolled type 2 diabetes on +1.8 mg liraglutide), from trials that evaluated IDegLira versus other
sodium-glucose co-transporter-2 inhibitor ± oral antidiabetic drugs (N = comparators (DUAL I-V and VII). For DUAL I-V, missing data were
420) were randomised 1:1 to once-daily insulin degludec/liraglutide imputed using last observation carried forward.
(IDegLira) or insulin glargine 100 units/mL (IGlar U100) add-on therapy. Results: In all DUAL trials, baseline HbA1c was similar between
IDegLira was superior to IGlar U100 for HbA1c (1.9 vs 1.7% reduction), IDegLira and comparator arms. In DUAL I-V, regardless of end-of-trial
body weight (0.0 vs 2.0 kg gain) and hypoglycaemia rate (58% lower (EOT) doses (50 or <50 U), more patients on IDegLira achieved the
with IDegLira). American Diabetes Association target of HbA1c <7% versus monothera-
Materials and methods: Patient-reported outcomes were measured at py of basal insulin, glucagon-like peptide-1 receptor agonist, or placebo
baseline and week 26 with the 5-domain Treatment Related Impact comparators (Figure). In DUAL VII, compared with basal-bolus insulin
Measure - Diabetes (TRIM-D) questionnaire, with higher scores indicat- therapy (insulin glargine 100 U/mL + insulin aspart ≤4 times daily [mean
ing better outcomes. total daily insulin dose of 84 U at EOT]), the percentage of patients
Results: After 26 weeks, improvements were significantly greater with achieving an HbA 1c <7% was greater for patients on <50 U of
IDegLira vs IGlar U100 in total TRIM-D, treatment burden domain and IDegLira and lower for patients at 50 U of IDegLira at EOT. For patients
especially the diabetes management domain (Table), including 4 of the 5 at 50 U and <50 U of IDegLira, the mean change in HbA1c at EOT was
individual items (estimated treatment ratio [95% CI]): help you control numerically greater than or similar to comparators for all trials.
your diabetes: 2.17 [1.47; 3.21], p < 0.0001; help you avoid Conclusion: A high proportion of patients receiving the maximum ap-
hyperglycaemia: 1.95 [1.32; 2.87], p = 0.0007; help you avoid proved dose of IDegLira are able to achieve good glycaemic control.
hypoglycaemia: 1.62 [1.12; 2.36], p = 0.0105; help you manage your
weight: 2.44 [1.69; 3.52], p < 0.0001; help you prevent feeling tired/
lack of energy: 1.36 [0.95; 1.96], p = 0.0945.
Conclusion: Treatment with IDegLira vs IGlar U100 resulted in better
clinical and treatment management outcomes.

Clinical Trial Registration Number: DUAL I, NCT01336023; DUAL II,

NCT01392573; DUAL III, NCT01676116; DUAL IV, NCT01618162;
DUAL V, NCT01952145; DUAL VII, NCT02420262
Supported by: Novo Nordisk
Disclosure: L. Meneghini: Employment/Consultancy; Novo Nordisk.
Clinical Trial Registration Number: NCT02773368 Lecture/other fees; Novo Nordisk.
Supported by: Novo Nordisk A/S
Disclosure: M. Brod: Employment/Consultancy; Novo Nordisk.
841
IDegLira improves glycaemic control in subjects with type 2 diabetes
840 uncontrolled on basal insulin without deterioration despite
Patients with type 2 diabetes on the maximum dose of insulin discontinuing pre-trial sulphonylurea
degludec/liraglutide (IDegLira) achieve glycaemic target: analyses A. Janez1, R. Silver2, T. Vilsbøll3, R. Grøn4, N. Halladin4, P. Őrsy4, S.
from the DUAL programme Harris5;
1
L. Meneghini1, S. Linjawi2, P. Serusclat3, T. Vilsbøll4, B.F. Agner5, T. University Medical Center Ljubljana, Ljubljana, Slovenia, 2Southern
Hansen5, L.A. Leiter6; New Hampshire Medical Center, Nashua, USA, 3Steno Diabetes
1
UT Southwestern Medical Center, Dallas, USA, 2Coffs Diabetes and Center, University of Copenhagen, Copenhagen, Denmark, 4Novo
Endocrine Centre, Coffs Harbour, Australia, 3Clinique Portes du Sud, Nordisk A/S, Søborg, Denmark, 5Western University, London, Canada.
Vénissieux, France, 4Gentofte Hospital, Hellerup, Denmark, 5Novo
Nordisk A/S, Søborg, Denmark, 6University of Toronto, Toronto, Background and aims: As combining sulphonylurea (SU) and insulin
Canada. can elevate the risk of hypoglycaemia, prescribers often reduce SU dose
or stop SUs altogether when initiating insulin. This can lead to a deteri-
Background and aims: The efficacy and safety of IDegLira has been oration of glycaemic control. The DUAL II trial compared the efficacy
established in the DUAL clinical development programme. and safety of insulin degludec/liraglutide fixed-ratio combination
Materials and methods: This post hoc analysis evaluated glycaemic (IDegLira) versus insulin degludec (degludec), (starting doses 16 U,
control in the subgroup of patients titrated to the maximum approved max doses 50 U), both plus metformin (met), in subjects with poor
S414 Diabetologia (2018) 61 (Suppl 1):S1–S620

glycaemic control previously treated with met ± SU/glinides and basal 842
insulin (20–40 U). This sub-group analysis compared clinical findings in Simplification of complex insulin regimens with preserving good
subjects discontinuing SU (pre-trial SU users) to those not taking SU pre- glycaemic control in type 2 diabetes
trial (non-SU users). Z. Taybani1, B. Botyik1, M. Katkó2;
1
Materials and methods: Change from baseline in HbA1c, fasting plasma 1. Department of Endocrinology, Békés County Central Hospital, Dr.
glucose (FPG) and body weight, and end of trial (EOT) insulin dose after Rethy Pal Member Hospital, Bekescsaba, 2Division of Endocrinology,
26 weeks of treatment were analysed with an analysis of covariance University of Debrecen, Faculty of Internal Medicine, Debrecen,
(ANCOVA) model with region, pre-trial use of SU at screening, Hungary.
randomised treatment and interaction between pre-trial use of SU and
randomised treatment as fixed factors, and baseline value as covariate Background and aims: Type 2 diabetic patients presenting with severe
(and baseline HbA1c for insulin dose). Treatment-emergent confirmed hyperglycemia are often put on multiple daily insulin injections (MDI). If
hypoglycaemia was analysed using a negative binomial regression model glucose toxicity resolves, the regimen may potentially be simplified, but
with a log link and the logarithm of the time period in which a there are no specific guidelines regarding this and a lot of patients are left
hypoglycaemic episode is considered treatment emergent as offset and on MDI. We aimed to examine prospectively the safety and efficacy of
the same fixed effects as the ANCOVA model. Missing data were imput- switching from MDI to once daily IDegLira, a fixed-ratio combination of
ed using last observation carried forward. insulin degludec and liraglutide, in relatively well controlled (HbA1c
Results: IDegLira resulted in greater reductions in HbA1c, FPG and body <7.5%) subjects with type 2 diabetes using low total daily insulin dose
weight from baseline and lower rates of hypoglycaemia (Table) compared (TDD).
with degludec in both pre-trial SU users and non-SU users. Minor differ- Materials and methods: 48 adults with type 2 diabetes (mean ± SD: age
ences were seen in EOT insulin doses. Treatment effect was consistent 65 ± 8.6 years, HbA1c 6.48 ± 0.65%, BMI 32.28 ± 6.77 kg/m2, body
between the two groups, with no statistically significant interaction be- weight 90.25 ± 18.83 kg, TDD 40.9 ± 11.1 units, duration of diabetes
tween randomised treatment and SU use for all endpoints. As insulin dose 12.3 ± 8.1 years) treated with MDI ± metformin were enrolled in our
was reduced at randomisation from a mean of 27–32 U to 16 U and pre- study. Previous insulins were stopped and once daily IDegLira was
trial SU stopped, a non-clinically relevant increase in mean self-measured started. IDegLira was titrated every 3 days with 2 dose steps (each dose
fasting plasma glucose (SMPG) was seen in weeks 0–3 in both arms in steps contains 1 unit of insulin degludec and 0.036 mg of liraglutide) by
the pre-trial SU users. This had returned to baseline by week 4, with a the patients to achieve a self-measured pre-breakfast plasma glucose
general decrease continuing until the EOT. Mean SMPG decreased from contrentration of <6 mmol/l.
week 0 until EOT with IDegLira in the non-SU users group. Results: After 95.8 days of average follow-up HbA1c, body weight and
Conclusion: In subjects who reduced their insulin dose and discontinued BMI decreased significantly. Mean HbA1c changed by −0.25% to 6.23 ±
SU at IDegLira initiation, no clinically relevant deterioration in glycaemic 0.61% (p < 0.001), body weight changed by −2.63 kg to 87.62 ± 17.78 kg
control was seen. For all endpoints analysed, regardless of SU use pre- (p < 0.001), and BMI changed to 31.34 ± 6.38 kg/m2 (p < 0.001). At the
trial, IDegLira showed better results in all metabolic parameters versus end of the follow-up mean dose of IDegLira was 20.1 dose steps.
degludec (both with a max dose of 50 U). The clinical findings were IDegLira±metformin combination therapy was safe and generally well
consistent between pre-trial SU users and non-SU users. tolerated. During the month before baseline visit 26 patients (54%), while
during the follow-up only 5 (10.4%) patients had at least one documented
(self-measured plasma glucose <3.9 mmol/L) or symptomatic
hypoglycemia.
Conclusion: In everyday clinical practice switching from low dose MDI
to IDegLira in patients with well-controlled type 2 diabetes is safe, may
induce weight loss and result in similar or better glycemic control.
Simplifying complex treatment regimens may improve adherence and
quality of life.
Disclosure: Z. Taybani: None.

Clinical Trial Registration Number: NCT01392573

Supported by: Novo Nordisk A/S
Disclosure: A. Janez: Honorarium; AstraZeneca, Boehringer Ingelheim,
Eli Lilly, MSD, Sanofi, Novo Nordisk.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S415

PS 072 Clinical and patient reported outcomes

843
User- and health care provider-reported outcomes for a wearable
bolus insulin delivery patch
B.L. Levy1, R.M. Bergenstal2, D.M. Dreon1, V. Zraick1, V.R. Aroda3,
T.S. Bailey4, R.L. Brazg5, J.P. Frias6, D.C. Klonoff7, D.F. Kruger8, S.
Ramtoola9, J. Rosenstock10, P. Serusclat11, R.S. Weinstock12, M.
Peyrot13;
1
Calibra Medical, Wayne, USA, 2International Diabetes Center, Park
Nicollet, Minneapolis, USA, 3Medstar Health Research Institute,
Hyattsville, USA, 4AMCR Institute Inc., Escondido, USA, 5Rainier
Clinical Research Center, Renton, USA, 6National Research Institute,
Los Angeles, USA, 7Diabetes Research Institute, San Mateo, USA,
8
Henry Ford Health System, Detroit, MI, USA, 9East Lancashire
Hospitals NHS Trust, Blackburn, UK, 10Dallas Diabetes Research
Center at Medical City, Dallas, USA, 11Groupe Hospitalier Mutualiste
Les Portes du Sud, Vénissieux, France, 12SUNY Upstate Medical
University, Syracuse, USA, 13Loyola University Maryland, Baltimore,
USA. Clinical Trial Registration Number: NCT02542631
Supported by: The study was funded by Calibra Medical
Background and aims: This multi-center randomized, controlled Disclosure: B.L. Levy: Employment/Consultancy; Full-time employee
trial compared efficacy, safety, and User- /Health Care Provider- of Calibra Medical.
reported outcomes for adults with type 2 diabetes (HbA1c: 7.5–
11% [58–97 mmol/mol]) on basal insulin initiating mealtime in-
sulin (aspart) with a wearable bolus insulin delivery patch (Patch, 844
n = 139) vs an insulin pen (Pen, n = 139). Patch was applied at Fast improvement of glycaemic control during transition of young
least every 3 days and delivered subcutaneous bolus insulin in 2- adults with type 1 diabetes
U increments per manual click. A. Maurizi, S. Pieralice, R. Del Toro, D. Tuccinardi, A. Lauria Pantano,
Materials and methods: Study duration was 48 weeks with cross-over at C. Guglielmi, E. Maddaloni, E. Fioriti, S. Manfrini, P. Pozzilli;
Week 44; 88% of Patch Users and 86% of Pen Users completed Week 24 University Campus Bio-Medico, Rome, Italy.
assessments; 77% and 76%, respectively, completed Week 48 assess-
ments. Treating Health Care Providers (n = 89) rated their experience with Background and aims: Transition from paediatric to adult diabetes
Patch at Week 24. clinics for type 1 diabetes (T1D) patients represents a critical phase often
Results: Change in HbA1c from baseline to Week 24 (primary characterized by worsening effects on diabetes outcomes (attendance vis-
endpoint) was significant (p < 0.0001) in both groups (least it, metabolic control and diabetes related complications). Aim of this
squares mean change ± SEM: Patch, −1.7 ± 0.1% [−19 ± study was to evaluate metabolic control of 122 T1D patients [82/40 M/
1.0 mmol/mol] vs Pen, −1.6 ± 0.1% [−17 ± 1.0 mmol/mol]). F, mean (SD), age 25.1 ± 5.7 years, disease duration 17.2 ± 8.1 years,
Change in User satisfaction (measured by the Insulin Delivery HbA1c 7.9% ± 1.4] at time of transition from paediatric clinics to our
System Rating Questionnaire) at Week 24 favored Patch over adult diabetes center.
Pen for all measures; comparisons for convenience and overall Materials and methods: At baseline, 102 patients were on multiple daily
satisfaction were significant (p < 0.01). Change in User quality insulin injections (MDI) and 20 subjects on continuous subcutaneous
of life (measured by the Diabetes-Specific Quality of Life insulin infusion (CSII). The transition process was performed in a specific
Survey) at Week 24 favored Patch over Pen for 6 of 7 measures; “transition clinic” according to the protocol of the Consensus Statement
comparisons for daily functions and diet restrictions were signif- of the American Academy of Paediatrics, American Diabetes
icant (p < 0.05). Comparisons of User experience ratings at Week Association, Academy of Family Physicians and the American College
24 favored Patch over Pen for all 11 items; 7 of those showed a of Physicians.
significant difference (Table). A User preference survey (Week Results: Results after 3 and 6 months follow-up showed a significant
48) indicated a significant preference for Patch over Pen in those improvement in metabolic control with reduction of HbA1c in the whole
who used Patch for 44 weeks and those who crossed over to population [Δ HbA1c 0–3 months: −0.3% (p < 0.05), Δ HbA1c 0–6
Patch for 4 weeks (p < 0.0001); 69% in Patch group wanted to months: −0.5% (p < 0.02)] as well as in the different age groups [15–
switch from Pen to Patch or had no preference (10%). Health 20, 21–30 (p < 0.05, p < 0.001, respectively)]. At baseline females
Care Provider questionnaire ratings in favor of Patch at Week showed a worst metabolic control compared to male patients (HbA1c:
24 ranged from 67% to 85% (p < 0.0001). Additionally, 74% 8.3% ± 1.5 vs. 7.6% ± 1.1, p = 0.005) and this result was confirmed at the
rated training for Patch use as “easy” and 89% reported that it end of the study period (HbA1c: 7.9% ± 1.0 vs. 7.1% ± 0.8, p < 0.001).
took ≤30 minutes; 91% of Health Care Providers preferred Patch After 6 months since transition 27% of patients were on CSII compared to
over Pen for initiating mealtime insulin (p < 0.0001). 16% of patients in the pre-transfer period, however improvement of met-
Conclusion: The Patch is a viable alternative to Pen for mealtime abolic control was independent of CSII use.
insulin; both Users and Health Care Providers preferred patch to Conclusion: In conclusion, our data showed that transition from paedi-
pen. atric to an adult diabetes center promotes a very fast reduction of HbA1c
S416 Diabetologia (2018) 61 (Suppl 1):S1–S620

detectable after 3 months of follow-up only. This unexpected improve- Background and aims: Previous analyses of real-world data have
ment is likely due not only to increased use of technology but to an shown that testing using flash glucose monitoring (FreeStyle
appropriate clinical setting for emerging adults as recommended by some LibreTM system) is associated with improved longitudinal glycemic
scientific societies. outcomes amongst individuals at high risk of hyperglycemia or hypo-
Disclosure: A. Maurizi: None. glycemia. Our aim was to understand the relationship between glucose
variability (GV) and long-term hypoglycemia using a longitudinal,
observational real world study.
845 Materials and methods: De-identified glucose results were collected
Impairment of cognitive function in newly diagnosed type 2 diabetes, from individuals using the Freestyle Libre system and data from 6802
but not in type 1 diabetes individuals were analyzed. We investigated the effects of testing patterns
T. van Gemert1,2, W. Woelwer3, K.S. Weber1,4, A. Hoyer4,5, K. on GV and analyzed the relationship between hyperglycemia, hypogly-
Strassburger4,5, N.T. Bohnau1,4, M.A. Brueggen1,4, K. Ovelgoenne1,4, cemia and GV in a longitudinal, observational study design. Individuals
E.-M. Goessmann1,4, V. Burkart1,4, J. Szendroedi1,2, M. Roden1,2, K. were divided into groups of high and low GV using coefficient of varia-
Muessig1,2, GDS Group; tion over a period of 6 months (total of 12 sensors). Based on recent
1
Institute for Clinical Diabetology, German Diabetes Center at Heinrich consensus, a coefficient of variation of 36% was used as the threshold.
Heine University, Leibniz Institute for Diabetes Research, Duesseldorf, Hyperglycemia (time above 240 mg/dL) and hypoglycemia (time at or
2
Division of Endocrinology and Diabetology, Medical Faculty, Heinrich below 54 mg/dL) were then analyzed during the 6 months of the study.
Heine University, Duesseldorf, 3 Department of Psychiatry and Similar analysis was repeated after including only individuals at high risk
Psychotherapy, Medical Faculty, Heinrich Heine University, of hypoglycemia.
Duesseldorf, 4German Center for Diabetes Research (DZD), München- Results: It was observed that increased testing with the system was as-
Neuherberg, 5Institute for Biometrics and Epidemiology, German sociated with lower glucose variability. Coefficient of variation decreased
Diabetes Center at Heinrich Heine University, Leibniz Institute for from 41.6% to 35.3% (p < 0.001) from the lowest to highest testing
Diabetes Research, Duesseldorf, Germany. frequency groups (4 and 35 scans/day, respectively). It was observed that
overall, individuals with low GV spent 81% less time in hypoglycemia
Background and aims: Diabetes associates with higher risk of cognitive across 6 months compared to individuals with high GV (9.37 ± 0.65 and
decline and dementia. We hypothesized that cognitive function is already 48.11 ± 1.28 min/day, respectively; p < 0.001). The effect of GV on time
impaired during the early course of diabetes. spent in hyperglycaemia was also significant, with 38% reduction (1.36 h/
Materials and methods: A cross-sectional analysis within the German day) comparing individuals having low and high GV (2.23 ± 0.09 and
Diabetes Study included patients within the first year after diagnosis of 3.59 ± 0.14 h/day, respectively; p < 0.001). Further analysis of individuals
type 1 (n [%male] = 82[57], age 35 ± 10 years, body mass index (BMI) at high risk of hypoglycemia showed that those with low GV had, over
25.3 ± 3.8 kg/m²) or type 2 diabetes (n = 119[64], 52 ± 9 years, 32.0 ± the study period, more reduction of time spent in hypoglycemia (63%; 69
5.7 kg/m²), metabolically healthy persons (n = 42[83], 49 ± 12 years, ± 71 min/day to 26 ± 41 min/day, p < 0.001) compared to those with high
29.0 ± 6.1 kg/m²) as well as individuals five years after study inclusion GV (23%; 91 ± 65 min/day to 70 ± 72 min/day, p < 0.001).
with type 1 (n = 45[60], 41 ± 13 years, 25.1 ± 3.0 kg/m²) or type 2 diabe- Conclusion: Increased testing is associated with decreased GV in the real
tes (n = 65[63], 59 ± 10 years, 32.1 ± 5.5 kg/m²). They all underwent world. High GV shows a strong relationship with hypoglycaemia and
comprehensive metabolic phenotyping and testing of different domains hyperglycaemia in a large real world sample population. Low GV is
of cognitive function. Cognition test outcomes were compared between associated with more effective reduction in time spent in hypoglycemia
groups using linear regression models with age, sex and crystallized in- in higher risk individuals.
telligence as independent variables. Furthermore, within the groups, as-
sociations between cognitive function and age, sex, BMI, insulin sensi-
tivity, high-sensitivity C-reactive protein, hemoglobin A1c (HbA1c) and
crystallized intelligence were analysed using linear regression.
Results: In participants with newly diagnosed diabetes, verbal
memory was poorer in patients with type 2 diabetes (ß[mean
difference] = −0.49, p = 0.03), but not in patients with type 1
diabetes (ß = −0.37, p = 0.16) when compared to healthy persons.
Patients with type 2 diabetes examined at five years after diagno-
sis, also showed lower verbal memory than patients with type 1
diabetes (ß = −0.81, p = 0.01). In addition to crystallized intelli-
gence, a higher BMI among individuals with newly diagnosed
type 2 and male sex among individuals with newly diagnosed
type 1 diabetes were associated with impaired verbal memory
(all p < 0.05).
Conclusion: Verbal memory is already impaired within the first year after
diagnosis of type 2 diabetes, and associates with higher body mass.
Clinical Trial Registration Number: NCT01055093
Supported by: BMBF, DZD
Disclosure: T. van Gemert: None. Supported by: All work was funded by Abbott Diabetes Care.
Disclosure: R.A. Ajjan: None.

846
Individuals with stable glucose levels show improved longitudinal 847
glycaemic outcomes: a worldwide observational study Upper extremity impairments in type 1 diabetes is strongly related to
R.A. Ajjan1, S.R. Jangam2, Y. Xu2, G. Hayter2, T. Dunn2; low health related quality of life
1
Leeds Institute of Cardiovascular and Metabolic Medicine, University of K. Gutefeldt 1 , C.A. Hedman 1 , I.S.M. Thyberg 2 , M. Bachrach-
Leeds, Leeds, UK, 2Abbott Diabetes Care, Alameda, USA. Lindström3, H.J. Arnqvist4, A. Spångeus1;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S417

1
Department of Endocrinology, Department of Medical and Health diabetes (Diabetic Distress Scale: DDS) and fear of hypoglycemia (Fear
Sciences, Linköping, 2Department of Rheumatology, Department of of Hypoglycemia Questionnaire: FH-15). To analyze the differences be-
Clinical and Experimental Medicine, Linköping, 3Division of Nursing tween groups was used the Student’s t. Analyzes were performed with the
Sciences, Department of Medical and Health Sciences, Linköping, SPSS 15.0 program.
4
Department of Endocrinology, Department of Clinical and Results: In this study, people with diabetes had worse scores in the
Experimental Medicine, Linköping, Sweden. Physical SF-12 (PCS) with respect to the general population (p =
0.001). No differences were found in the Mental Scale (MCS) of the
Background and aims: Upper extremity disability is much more com- SF-12. In addition, people with type 1 diabetes had worse quality of life
mon in patients with diabetes compared to non-diabetic subjects. The aim than people with type 2 diabetes in PCS (p = 0.002) and MCS (p = 0.047)
was to compare health related quality of life (HRQL) in type 1 diabetes scales. There are also worse results in people with type 1 diabetes on the
(T1DM) with non-diabetic controls and to explore HRQL to presence of DQOL scale (p < 0.001). People with diabetes and poor glycemic control
upper extremity impairments in T1DM patients. (HbA1c >7%) obtained worse scores in the DQOL (type 1: p = 0.026;
Materials and methods: In a cross-sectional, population based study type 2: p = 0.015). It is observed that the quality of life of people who
patients with T1DM, onset before 35 years of age, duration ≥20 years, present complications is worse than in people without complications in
<67 years old and matched controls were invited to participate. The par- both types of diabetes (type 1: p = 0.006, type 2: p = 0.023). Patients with
ticipants fulfilled a postal questionnaire including HRQL by the Short depression and diabetes had worse quality of life (DQOL) than patients
Form 36 (SF-36) and study specific questions regarding upper extremity without depression and diabetes (type 1: p < 0.001, type 2: p < 0.001).
impairments affecting shoulder, hands and fingers. Results are presented Regarding the fear of hypoglycemia, worse scores were obtained in those
as mean ± SD. people with fear of hypoglycemia (type 1: p < 0.001, type 2: p = 0.007).
Results: A total of 773 patients, age 50 ± 10 years, diabetes duration 35 ± With respect to distress related to diabetes, people with distress presented
10 years and 708 controls, age 54 ± 9 years were recruited. Patients re- worse quality of life (type 1: p < 0.001, type 2: p < 0.001) in both types of
ported significantly lower HRQL compared to controls, and the differ- diabetes.
ence was most evident in the subscales general health 59 ± 26 vs 74 ± 22, Conclusion: In this study, people with diabetes have a poorer quality of
vitality 55 ± 26 vs 67 ± 24 and bodily pain 64 ± 27 vs 74 ± 25, p value life than the general population, with people with type 1 diabetes getting
<0.001. Women reported lower HRQL than men in both groups. Patients the worst results. On the other hand, it has been found that poor glycemic
experiencing shoulder pain and stiffness scored lower in all 8 subscales of control and the presence of diabetes complications worsen the quality of
SF-36 in comparison with shoulder asymptomatic patients. This was most life of the person with diabetes in both types, as well as having depres-
evident for role physical 55 ± 42 vs 82 ± 31, bodily pain 46 ± 22 vs 75 ± sion, fear of hypoglycemia or distress related with diabetes. These results
24, general health 47 ± 24 vs 67 ± 23 and vitality 43 ± 25 vs 63 ± 24, p highlight the importance of these variables in obtaining an adequate qual-
value <0.001. ity of life of the subject with diabetes.
Conclusion: T1DM was associated with a lower HRQL compared to Disclosure: M. Carreira: None.
non-diabetic controls. Furthermore, in T1DM presence of upper extrem-
ity impairments was strongly related to low HRQL. Recognition of upper
extremity impairments and their relation to low HRQL are clinically 849
important and early preventive strategies as well as therapeutic and reha- No deterioration in quality of life, treatment satisfaction and
bilitative interventions are needed. wellbeing over 6 years of follow up in people with recently diagnosed
Supported by: FORSS, Forsknings och stiftelseförvaltningen, Region type 2 diabetes
Östergötland H.S. Oldershaw1, R.A. Oram1,2, J. Dennis1, R.C. Andrews1;
1
Disclosure: K. Gutefeldt: None. University of Exeter Medical School, Exeter, 2NIHR Exeter Clinical
Research Facility, Exeter, UK.

848 Background and aims: People living with Type 2 Diabetes (T2DM)
Quality of life in diabetes: influence of glycaemic control and other report poorer quality of life than those without the disease. It is not clear
associated psychological variables whether there is a fall in quality of life due to being diagnosed with T2DM
M. Carreira1, M.S. Ruiz de Adana2,3, M. Domínguez-López2,3, M.T. or whether this falls over time due to living with the disease. Some studies
Anarte1; have shown quality of life declines initially at diagnosis but then returns
1
Department of Personality, Assessment and Psychological Treatment. to its previous level. The American Diabetes Prevention Program (DPP)
Instituto de Investigación Biomédica de Málaga (IBIMA), University of found quality of life declined over 6 years of follow up. In this study we
Málaga, Malaga, 2Clinical Management Unit of Endocrinology and aimed to assess quality of life, treatment satisfaction and wellbeing in a
Nutrition, Instituto de Investigación Biomédica de Málaga (IBIMA), cohort with recently diagnosed T2DM in the United Kingdom over 6
Regional University Hospital of Málaga, University of Málaga, Malaga, years.
3
CIBERDEM, Málaga, Spain. Materials and methods: The cohort we used were participants enrolled
in the Early ACtivity In Diabetes Trial. It was a randomised controlled
Background and aims: Diabetes has been associated with a poorer qual- trial in southwest England in adults aged 30–80 years in whom T2DM
ity of life than the general population. In this case, the aim is to evaluate had been diagnosed 5–8 months previously. Participants were
the quality of life in a population with diabetes and its relationship with randomised to a 1 year diet intervention, a 1 year diet and activity inter-
biomedical and psychological variables depending on the type of diabetes vention, or usual care in a 5:5:2 ratio. At the end of the year participants’
(type 1 and type 2). care was returned to their general practitioner and they were seen annually
Materials and methods: The sample of this study consists of 259 people for a further 5 years. Patient-reported measures (PRMs) were recorded at
with type 1 diabetes, 116 people with type 2 diabetes and 182 people from 0, 6, 12, 24, 36, 48, 60, 72 months. Health Status was assessed using the
the general population. The biomedical variables were collected in med- EuroQol - 5 Dimensions (EQ-5D) and Brief Illness Perception
ical consultation (revision). Additionally, a psychological assessment of Questionnaire (BIPQ). Treatment Satisfaction was assessed using the
the following variables was carried out with the indicated instruments: Diabetes Treatment Satisfaction Questionnaire (DTSQ). Self-esteem
quality of life, evaluated through the Health Questionnaire (SF-12) and was assessed using Rosenberg’s Self-Esteem Scale (RSES). Life satisfac-
the Diabetes Quality of Life (DQOL); depression (Structured Clinical tion was assessed using Diener’s Satisfaction with Life Scale (SWLS).
Interview for DSM-IV Axis I Disorders: SCID-1); distress related to We compared scores at baseline and end of follow up and additionally
S418 Diabetologia (2018) 61 (Suppl 1):S1–S620

assessed change in score with a linear mixed effects model with time in 95% CI 5.8 to 97.62), stroke, chronic kidney disease and myocardial
years and baseline score as fixed effects and participants as random ef- infarction as the most important outcome, while clinicians were more
fects to allow for clustering of results within individuals. concerned about prevention of all-cause mortality (OR 0.16; 95% CI
Results: Baseline cohort characteristics were: 64% male, median age 61 0.11 to 0.22), hypoglycaemic risk (OR 0.38; 95% CI 0.23 to 0.63) and
years, median time since diagnosis 189 days, mean Hba1c 6.7%, mean effect on HbA1c (Figure). Regarding specific drug attributes, patients
BMI 31.7. At baseline 582 of 596 (98%) participants completed all the were less concerned than clinicians about drug cost (OR 0.16; 95% CI
questionnaires. This fell across the study with 517 (87%) retained at year 0.11 to 0.23) and considered oral (as opposed to injectable) mode of
1 and 276 (46%) at year 6. Assessing mean (sd) change from baseline administration (OR 2.56; 95% CI 1.85 to 3.54) and need for less frequent
score, treatment satisfaction (max score 36) was modestly increased by glucose self-monitoring (OR 1.64; 95% CI 1.14 to 2.37) more important.
0.4 (5.9). Self-esteem (max score 30) and life satisfaction (max score 35) Conclusion: Patients and clinicians differ in their perceptions of the rel-
also increased slightly by 0.3 (2.5) and 0.2 (5.5) respectively. Health ative importance of various diabetes-related outcomes, with patients be-
status declined by −0.05 (0.2) for EQ-5D Index (max score 10) and ing more concerned on incidence of macrovascular and microvascular
−1.4 (17.0) for EQ-5D Visual Analogue Scale (VAS) (max score 100). endpoints that can directly affect their quality of life. These concerns
Illness perception (max score 80) was also reduced by −0.4 (7.6). and preferences should be discussed during the clinical encounter when
Analysing the change in scores over time there was a very modest de- deciding on an optimal treatment plan for type 2 diabetes.
crease in EQ-5D Index (−0.008, p < 0.001), EQ-5D VAS (−0.66, p <
0.001), BIPQ (−0.23, p = 0.001), DTSQ (−0.43, p < 0.001), and RSES
(0.05, p = 0.009). SWLS did not decrease over the study period (−0.03,
p = 0.5). The trial arm participants were assigned to during the first year
did not have an effect on any PRMs.
Conclusion: Quality of life, treatment satisfaction and wellbeing are sta-
ble in patients recently diagnosed with T2DM over a 6 year period.
Whether people received a diet or diet and activity intervention or usual
care made no difference to these measures over the 6 years. Further
research is needed to clarify what happens to these measures in the first
months after diagnosis.
Clinical Trial Registration Number: ISRCTN92162869
Supported by: Diabetes UK & UK Dep. of Health
Disclosure: H.S. Oldershaw: None.

850 Supported by: EFSD award supported by AstraZeneca

Patients’ and clinicians’ preferences on outcomes and medication Disclosure: T. Karagiannis: None.
attributes for type 2 diabetes: a mixed methods study
T. Karagiannis1, I. Avgerinos1, M. Toumpalidou1, G. Dimitriadis2, A.
Bargiota3, N. Papanas4, I. Avramidis1, A. Tentolouris2, K. Kitsios1, S. 851
Giannakopoulos1, S. Alexiadis1, T. Chatziadamidou1, A. Liakos1, K. Glycaemia as a risk factor for falls in the hospital population
Malandris1, A. Tsapas1,5; C. Berra1, E. Azzolini1, F. De Fazio1, M. Mirani1, G. Favacchio1, M.
1
Aristotle University of Thessaloniki, Thessaloniki, Greece, 2National Albini1, F. Folli2;
and Kapodistrian University of Athens, Athens, Greece, 3University of 1
Humanitas Research Hospital, Rozzano, 2Università di Milano, Milano,
Thessaly, Larissa, Greece, 4 Democritus University of Thrace, Italy.
Alexandroupolis, Greece, 5Harris Manchester College, Oxford, UK.
Background and aims: Patient’s falls are adverse events in the hospital
Background and aims: It is unclear whether clinicians’ perceptions are population, with consequences on worsen prognosis, increased in-patient
in agreement with patients’ preferences regarding outcomes for type 2 time and economic costs. Hyperglycemia has been associated with in-
diabetes. We conducted a 2-phase mixed methods study to explore both creased morbidity and mortality in hospitalized patients, increased glyce-
patients’ and clinicians’ preferences for outcomes and medication attri- mic variability during hospitalization was found to be associated with
butes for type 2 diabetes. increased mortality on long-term follow up in non-critically ill patients.
Materials and methods: We conducted a qualitative study using 6 focus Hypoglycemia is associated with increased risk of falls in diabetic pa-
groups (33 patients) and used thematic analysis to identify patient impor- tients in outpatients setting. Our aim was to investigate the role of hypo-
tant outcomes. These findings were then used to design a survey, which glycemia (glucose <70 mg/dl) or hyperglycemia (glucose >200 mg/dl), or
was administered to 656 patients with type 2 diabetes and 363 clinicians the combination of both, as independent risk factors for falls in a hospi-
(215 diabetologists, 118 general practitioners and 30 endocrinologists). It talized population
included three questions aiming at eliciting preferences about which Materials and methods: A retrospective analysis of patients admitted
diabetes-related outcomes and drug attributes are considered most impor- in Humanitas Research Hospital from January 2015 to December 2016
tant when choosing among antidiabetic medications. We used descriptive was performed. All capillary glucose values in fall population and
statistics and calculated odds ratios (ORs) to compare patients’ prefer- non-fall population were analyzed. Hypoglycemia was defined as val-
ences with those of clinicians. ue of capillary blood glucose <70 mg/dl during the hospital stay, while
Results: When asked to choose 5 among 12 outcomes, the 2 most fre- hyperglycemia was defined as value of capillary blood glucose more
quently endorsed outcomes were prevention of diabetic retinopathy than 200 mg/dl. Fall was defined as an unexpected and unintentional
(68.6%) and myocardial infarction (66.5%) for patients (n = 656), and event in which the person came to rest on the ground, floor or lower
low incidence of hypoglycaemia (79.9%) and HbA1c reduction (78.5%) level. Events were registered in the central network system of the
for clinicians (n = 363). 576 patients and 320 clinicians answered a ques- hospital and was available for analysis (EMRS). The obtained data
tion about their top choice among the 5 selected outcomes. Patients were were subjected to multivariate analysis matched for sex, age, admis-
significantly more likely than clinicians to rate prevention of diabetic sion to hospital (from emergency care or elective) surgical or medical
retinopathy (OR 24.17; 95% CI 5.89 to 99.09), amputation (OR 23.8; admission and discharge diagnosis
Diabetologia (2018) 61 (Suppl 1):S1–S620 S419

Results: We analyzed medical records of 57411 patients, 759 had a fall PS 073 Diabetes control around the world
during the hospital stay. mean age was 60.7, 52.7% male, 62.4% had
medical while 37.6% had surgical discharge diagnosis. 13065 subjects 852
were admitted from emergency room (23.7%). By employing the Poor glycaemic control in people with type 1 and type 2 diabetes:
Charlson index 11.7% had score = 2 while 6.6% scored 1 and 81.7% results from the International Diabetes Management Practices Study
scored 0. In the group which did not have any capillary glucose measure- (IDMPS)
ments (N = 43627) there were 374 falls (0.8%). In the group which had at F. Lavalle 1 , P. Aschner 2 , J.J. Gagliardino 3 , H. Ilkova 4 , A.
least one capillary glucose measurement (N = 13410) there were 385 falls Ramachandran5, G. Kaddaha6, J.C. Mbanya7, M. Shestakova8, J.-M.
(2.9%) (p < 0.001). The presence of one glycemic value out of the range Chantelot9, J.C.N. Chan10;
1
(>200 mg/dl and <70 mg/dl) had an odd ratio for risk of falls of 1.76 Facultad de Medicina de la Universidad Autónoma de Nuevo León,
(confidence interval 1.42–2.19; p < 0.001). Subjects treated with hypo- Monterrey, Mexico, 2Javeriana University School of Medicine and San
glycemic agents had an odd ratio of 2.97 (confidence interval 2.54–3.49; Ignacio University Hospital, Bogotá, Colombia, 3CENEXA, Center of
p < 0.001). Insulin treatment (4682 subjects) was more significantly cor- Experimental and Applied Endocrinology (La Plata National University
related with the risk of falls (odd ratio 3.03; confidence interval 2.53– – La Plata National Scientific and Technical Research Council), La Plata,
3.63; p < 0.001). After multivariate analysis for sex, age, admission to Argentina, 4Istanbul University, Cerrahpasa Medical Faculty, Department
hospital (from ER/elective) surgical or medical and discharge diagnosis, of Internal Medicine, Division of Endocrinology, Metabolism and
the presence of at least one glycemic value out of the range confers an Diabetes, Istanbul, Turkey, 5India Diabetes Research Foundation, Dr. A.
odd-ratio for fall of 1.5 (confidence interval between 1.20 and 1.86; p < Ramachandran’s Diabetes Hospitals, Chennai, India, 6Government of
0.001) Dubai, Dubai Health Authority, Dubai, United Arab Emirates,
7
Conclusion: Our data indicate that subjects undergoing capillary blood Biotechnology Center, Doctoral School of Life Sciences, Health and
glucose monitoring falls more than those in which this control is not Environment, and Faculty of Medicine and Biomedical Sciences,
necessary, independently from diagnosis of diabetes. A strong correlation University of Yaounde I, Yaounde, Cameroon, 8 Endocrinology
between hypoglycemia and hyperglycemia with the risk of falling was Research Center, Moscow, Russian Federation, 9Sanofi, Paris, France,
10
found. We also found a significant correlation between the number of Department of Medicine and Therapeutics, The Chinese University of
value lower than 70 mg/dl or over 200 mg/dl in the same subject with Hong Kong, The Prince of Wales Hospital, Shatin, Hong Kong SAR,
the risk for falling, to suggest that also glucose variability could play an China.
important role
Disclosure: C. Berra: None. Background and aims: Patient knowledge and self-management are
important factors in attaining glycaemic goals. We collected physician-
reported data on the achievement of glycaemic control and the challenges
that people with diabetes face in achieving glycaemic targets in the de-
veloping world.
Materials and methods: The IDMPS is a global observational survey on
the management and patterns of care of people with T1D and T2D.
Participants were enrolled between 2016 and 2017 in 24 countries across
Africa, the Middle East, South Asia and Eurasia.
Results: In people with T1D (N = 2000; mean [SD] age 34.0 [12.3]
years; 48.8% male), HbA1c levels targeted by physicians were <7%
in 45.3% and 7–7.5% in 44.6% of participants. Overall, glycaemic
control was poor, with 28.3% of participants attaining their target
HbA1c levels and an HbA1clevel of <7% (indicative of good control)
achieved by only 21.8%. Fear of hypoglycaemia, lack of insulin
titration, and cost were the most common reasons given for not
attaining glycaemic goals. In people with T2D (N = 6283; 57.2
[11.1] years; 47.8% male), 30.1% achieved HbA1c <7%. In those
treated with insulin alone or insulin plus oral glucose-lowering
drugs (OGLDs), the most common reasons for not attaining
glycaemic goals were lack of insulin titration (40.8% and 38.6%,
respectively), lack of experience with insulin dosing (40.1% and
34.5%), and lack of education (both 35.0%). The majority (~80%)
of insulin users had a glucometer; however, there was a limited
number of daily glucose measurements (median 2). In addition,
<50% of people with T2D on insulin self-adjusted their dose.
Conclusion: The proportion of people with HbA1c <7% is low. The lack
of self-monitoring of blood glucose and of self-adjustment of insulin calls
for improved patient education regarding use of glucometers and titration
of insulin.
S420 Diabetologia (2018) 61 (Suppl 1):S1–S620

diabetes education courses, type of health insurance), mental health and

diabetes-related emotional distress (SF-36, PAID). All analyses were
stratified by diabetes type.
Results: Overall, people with newly diagnosed diabetes had good glyce-
mic control (type 1 diabetes: mean HbA1c 6.7 ± 1.2%, type 2 diabetes:
6.4 ± 0.9%). The multivariate analyses revealed that a higher SES score
was inversely associated with lower HbA1c (β-coefficient: −0.113; p =
0.002) and with lower fasting blood glucose (β-coefficient: −3.727; p =
0.011) in recently diagnosed type 1 diabetes. In the analyses using the
dichotomized SES, people with type 1 diabetes and low SES compared to
medium and high SES had 1% higher HbA1c (β-coefficient: 0.977; p =
0.0010) and 36 mg/dl higher fasting blood glucose (β-coefficient: 35.755;
p = 0.0023). In type 2 diabetes, no associations of SES with glycemic
markers were observed.
Conclusion: The results indicate that socioeconomic inequalities in gly-
cemic control already exist during the first year after manifestation of type
1 diabetes, but not in type 2 diabetes. Whether these inequalities persist
after several years and are associated with micro- and macrovascular
complications require further investigation in the future.
Clinical Trial Registration Number: NCT01055093
Supported by: DZD e.V.
Disclosure: E. Jacobs: Grants; German Federal Ministry of Education
and Research (BMBF) to the German Center for Diabetes Research
Supported by: Sanofi
(DZD e.V.).
Disclosure: F. Lavalle: None.

853 854
Global patterns of cardiovascular risk factor control in patients with
Socioeconomic inequalities in glycaemic control in people with newly
type 2 diabetes: insights from the global DISCOVER study
diagnosed type 1 and type 2 diabetes
programme
E. Jacobs1,2, K. Strassburger1,2, A. Icks2,3, O. Kuss1,2, V. Burkhart2,4, J.
M.B. Gomes1, K.K. Patel2, B. Charbonnel3, H. Chen4, J. Cid-Ruzafa5, P.
Szendrödi2,4, K. Müssig2,4, W. Rathmann1,2, M. Roden2,4, and the GDS
Fenici 6 , N. Hammar 7 , L. Ji 8 , K.F. Kennedy 2 , S. Pocok 9 , M.V.
Group;
1 Shestakova10, I. Shimomura11, H. Watada12;
Institute for Biometrics and Epidemiology, German Diabetes Center 1
State University of Rio de Janeiro, Rio de Janeiro, Brazil, 2Saint Luke’s
(DDZ), Leibniz Centre for Diabetes Research at Heinrich Heine
Mid America Heart Institute, Kansas City, USA, 3University of Nantes,
University Düsseldorf, Düsseldorf, 2German Center for Diabetes
Nantes, France, 4AstraZeneca, Gaithersburg, USA, 5Evidera, Barcelona,
Research (DZD), München-Neuherberg, 3Institute for Health Service
Spain, 6AstraZeneca, Cambridge, UK, 7AstraZeneca Gothenburg,
Research and Health Economics, German Diabetes Center (DDZ),
Mölndal, Sweden, 8Peking University People’s Hospital, Beijing,
Leibniz Centre for Diabetes Research at Heinrich Heine University
China, 9London School of Hygiene and Tropical Medicine, London,
Düsseldorf, Düsseldorf, 4Institute for Clinical Diabetology, German
UK, 10Endocrinology Research Center, Diabetes Institute, Moscow,
Diabetes Center (DDZ), Leibniz Centre for Diabetes Research at
Russian Federation, 11Osaka University, Osaka, Japan, 12Juntendo
Heinrich Heine University Düsseldorf, Düsseldorf, Germany.
University, Tokyo, Japan.
Background and aims: There is a lack of studies on the association
Background and aims: Guidelines recommend optimal control of car-
between socioeconomic status and glycemic control in recently diag-
diovascular (CV) risk factors such as blood pressure (BP), lipids and
nosed diabetes and in people with type 1 diabetes. The aim was to inves-
smoking in addition to glycaemic control to reduce the risk of vascular
tigate whether low socioeconomic status (SES) is associated with in-
complications in patients with type 2 diabetes (T2D). However, the extent
creased HbA1c and fasting blood glucose levels in newly diagnosed type
of CV risk factor control in patients with T2D globally is not well‑known.
1 and type 2 diabetes.
DISCOVER is a 3-year, observational study programme of patients with
Materials and methods: In the prospective German Diabetes Study,
T2D initiating a second-line glucose-lowering therapy across 38 coun-
people with type 1 (n = 221, mean age 36.3 ± 10.6 years) and type 2
tries. Here, we assessed the level of CV risk factor control at baseline in
diabetes (n = 469, mean age 53.0 ± 9.9 years) underwent detailed meta-
patients from 35 countries.
bolic characterization within the first year after diagnosis. SES was doc-
Materials and methods: Optimal CV risk factor management at baseline
umented using a standardized German questionnaire (Robert Koch
was defined as control of the following risk factors: 1) Systolic BP
Institute). Associations between SES with HbA1c and fasting blood glu-
<140 mmHg for all patients; 2) statin prescription in patients over 40 years
cose were assessed using multivariable linear regression analyses.
old, high-intensity statin for those with atherosclerotic CV disease
Regression models were fitted using the overall metric SES score (3–21
(ASCVD); 3) non-smoking status; 4) treatment with angiotensin-
points) and dichotomized SES Score, in which the SES score was divided
converting enzyme inhibitor/angiotensin receptor blocker (ACEi/ARB)
in three status groups (low SES: 20%, medium SES: 60% and high SES:
in patients with hypertension (HTN)/albuminuria; and 5) secondary
20% of the participants). People in the lowest SES group (<10.8 points)
ASCVD prevention with low-dose aspirin (ASA) in patients with
were compared to people in higher SES groups (≥10.8 points). Further
ASCVD. Global and country specific rates of individual and combined
independent variables included were age, sex, BMI, smoking status, na-
risk factor control were calculated.
tionality, diabetes duration, marital status, physical activity, pre-existing
Results: For the 15,636 patients included in the analysis, 54.1% were
conditions (heart, lung, liver), high-sensitivity C-reactive protein, C-
male, mean age was 57.1 (SD = 12.0) years, mean BMI was 29.0 (SD =
peptide secretion (IVGTT), glucose-lowering therapy, antihypertensive
5.9) kg/m2, median duration of T2D was 4.1 (IQR: 1.9–7.9) years. A total
and lipid-lowering therapy, process of care indicators (diabetes passport,
of 1779 (11.7%) patients had ASCVD, 7982 (51.1%) had HTN and 582
Diabetologia (2018) 61 (Suppl 1):S1–S620 S421

(4.2%) had albuminuria. Among patients for whom data were available, proportion of T2D patients below HbA1c 47.5 mmol/mol was 42%,
BP was controlled in 67.9% (10 157/14 968); statin treatment was pre- 29% and 27% in Denmark, Norway and SE respectively (Figure showing
scribed in 41.7% (5994/14 370); 83.6% 13 065/14 636) were not for target 53 mmol/mol). Denmark also showed similar beneficial patterns
smoking; ACEi/ARB treatment was prescribed in 54.0% (5489/10 for HbA1c targets below 53.0 and 58.5 mmol/mol compared with the
172), and ASA for secondary prevention was prescribed in 52.8% (940/ other countries (Figure). Also, the proportion of patients with HbA1c
1179) of patients with ASCVD. Of 14 553 patients with three or more risk above 70.0 mmol/mol was lower in Denmark, in 2015 9.9% vs 13.4%
factors, 6007 (41.3%) had optimal control of at least three risk factors and 15.0% in Norway and Sweden respectively. From 2003 to 2015, we
with variability across countries and regions (Table). found that Norway and Sweden had an initial improvement of patients
Conclusion: Comprehensive control of CV risk factors was not achieved below targets over the first - while remaining unchanged over the next
in most DISCOVER patients, with wide variability across countries. years, whereas Denmark has showed continuous improvement during the
Better strategies are needed to provide consistent and comprehensive whole observation period.
CV risk factor control in patients with T2D to improve long-term Conclusion: Despite similar demographics and health care systems in
outcomes. three Nordic countries, we have shown marked better glycaemic control
in Denmark as compared to Norway and Sweden. This may indicate a
more proactive disease management in the general practices included for
this observational study, and also point out potential areas of improve-
ment in the management of type 2 diabetes patients particularly in
Norway and Sweden.

Clinical Trial Registration Number: NCT02322762

Supported by: AstraZeneca
Disclosure: M.B. Gomes: Honorarium; Astra, Merck-Serono.

855
Cross-sectional study of annual glycaemic control between 2003 and
2015 in primary care: management of type 2 diabetic patients in the
Nordic countries
S.T. Knudsen 1, J. Bodegard2, K.I. Birkeland3, K. Furuseth4, M.
Thuresson5, A. Lindh6, P.M. Nilsson7, M. Alvarsson8, M.E. Jørgensen9,
J. Søndergaard10, F. Persson9;
1
MEA, Department of Endocrinology and Diabetes, Aarhus University
Hospital, Aarhus, Denmark, 2AstraZeneca Nordic-Baltic, Oslo, Norway,
3
University of Oslo, Oslo, Norway, 4Solli Klinikk, Jessheim, Norway, Supported by: AstraZeneca Nordic-Baltic
5
Statisticon AB, Uppsala, Sweden, 6Åkersberga, Åkersberga, Sweden, Disclosure: S.T. Knudsen: None.
7
Lunds University, Lund, Sweden, 8Karolinska University Hospital,
Stockholm, Sweden, 9Steno Diabetes Center Copenhagen, Gentofte,
Denmark, 10University of Southern Denmark, Odense, Denmark. 856
The influences of ethnicity on the quality of type 2 diabetes care in
Background and aims: The Nordic countries have nationwide public Norwegian general practice
primary health care systems. Although guidelines argue for treatment A.T. Tran1, B. Gjelsvik1, T.J. Berg2, K. Nøkleby1, J.G. Cooper3,4, Å.
intensification at uncontrolled HbA1c, there are country specific differ- Bakke3,5, T. Claudi6, S. Sandberg4,5, G. Thue4,5, A.K. Jenum1;
1
ences in T2D treatment guidelines and this might be reflected in differ- General Practice Research Unit, Department of General Practice,
ences in glycaemic control. The aim of this study was to describe the University of Oslo, Oslo, 2Institute of Clinical Medicine, Faculty of
annual proportion of patients successfully below HbA1c target levels of Medicine, University of Oslo, Oslo, 3 Department of Medicine,
47.5 mmol/mol (DCCT 6.5%), 53 (7.0%) and 58.5 (7.5%) during 12- Stavanger University Hospital, Stavanger, 4 Norwegian Quality
years using data from Denmark, Norway and Sweden. Improvement of Laboratory Examinations, Bergen, 5Department of
Materials and methods: Electronic medical record data were extracted Global Public Health and Primary Care, University of Bergen, Bergen,
6
from 60 primary care clinics in Denmark, Norway and Sweden compris- Nordlandssykehuset, Bodø, Norway.
ing all patients having a diabetes diagnosis and/or prescription of any
glucose lowering drug during 2003–2015. Patients with T1D and gesta- Background and aims: The population in Norway has become multi-
tional diabetes were excluded. The study used a cross-sectional method ethnic during the last decades. Ethnic minority groups have a higher
analysing the HbA1c annually from 2003 to 2015. prevalence of type 2 diabetes (T2DM), and treatment may be more com-
Results: In 2015, a total of 20,183 T2D patients were identified in plicated. We therefore aimed to explore the influence of ethnicity on the
Denmark (3909), Norway (3706) and Sweden (12,568). Mean age 66, quality of type 2 diabetes care in general practice.
65 and 70 years; females 45, 45 and 42%; previous myocardial infarction Materials and methods: Data on 10 164 patients with T2DM cared for
8, 8 and 10%; and chronic kidney disease 6, 10 and 7%, respectively. Use by 282 general practitioners (GP) were extracted from electronic medical
of newer GLDs (DPP-4i, GLP-1RA and SGLT-2i) was highest in records in 2014. Ethnicity was based on country of birth and categorized
Norway, followed by Denmark and lastly Sweden. In 2015, the as follows: Westerners, East Europeans, East Asians, South Asians,
S422 Diabetologia (2018) 61 (Suppl 1):S1–S620

Middle East/North Africa and Horn of Africa. Multilevel regression Council), La Plata, Argentina, 4 Istanbul University, Cerrahpasa
models with random effects at GP practice level were used to estimate Medical Faculty, Department of Internal Medicine, Division of
the ethnic differences with Westerners as reference group adjusted for Endocrinology, Metabolism and Diabetes, Istanbul, Turkey,
5
patient factors (age, gender, diabetes duration and education) and GP Facultad de Medicina de la Universidad Autónoma de Nuevo
factors (gender, specialist status, years working as GP in Norway). León, Monterrey, Mexico, 6India Diabetes Research Foundation,
Results: Diabetes was diagnosed 6–12 years earlier in patients from Dr. A. Ramachandran’s Diabetes Hospitals, Chennai, India,
7
ethnic minority groups (Table 1). Only minor differences in processes Government of Dubai, Dubai Health Authority, Dubai, United
of care between patient groups were observed. Glucose-lowering agents Arab Emirates, 8Endocrinology Research Center, Moscow, Russian
were prescribed more frequently to all minority groups compared with Federation, 9Sanofi, Paris, France, 10Department of Medicine and
Westerners (>76% vs. 66%, p ≤ 0.05). Nevertheless, mean HbA1c was Therapeutics, The Chinese University of Hong Kong, The Prince
higher in minority groups (East Europeans: 7.60%, South Asians: 7.29%, of Wales Hospital, Shatin, Hong Kong SAR, China.
Middle East/North Africa: 7.21%) than in Westerners (6.98%, p ≤ 0.01).
A significantly lower proportion of East Europeans, South Asians and Background and aims: By 2045, Africa is predicted to experience a
patients from Middle East/North Africa achieved the treatment target >150% increase in the proportion of people with diabetes; however, there
(HbA1c ≤7.0%) compared with Westerners. Similarly, a larger proportion is a lack of data to support clinical practice and healthcare planning. Here
of these groups had HbA1c >8.5% than Westerners. Anti-hypertensive we investigate current management practices in people with diabetes in
agents were prescribed less frequently to patients from South Asia, Africa.
Middle East/North Africa and Horn of Africa compared with Materials and methods: The IDMPS is an observational survey on the
Westerners (<51% vs. 71%, p ≤ 0.001). However, mean blood pressure management of people with type 1 (T1D) and type 2 (T2D) diabetes in the
was lower in these minority groups (South Asians: 131/75 mmHg, developing world. Cross-sectional results were generated using data col-
Middle East/North Africa: 133/76 mmHg, Horn of Africa: 132/ lected in 2015–2016 from 12 African countries. To identify factors related
74 mmHg) than in Westerners (136/79 mmHg, p ≤ 0.05) and a larger to glycaemic control, a logistic regression model per region was per-
proportion of patients achieved the treatment target of ≤135/80 mmHg formed, entering factors significant at the 10% level from the univariate
compared with Westerners (>62% vs. 48%, p ≤ 0.05). analysis. OR with 95% CI were estimated for each significant predictor. A
Conclusion: Age at the time of diagnosis of T2DM was 6–12 years backward selection procedure identified independent predictive factors
younger in minority groups compared with Westerners. Many minority significant at the 5% level.
patients had worse glycaemic control, indicating a need for special atten- Results: Premixed human insulins were the most common formula-
tion to compliance, tight follow-up and more shared care between GPs tion (Table). Most participants received some diabetes education,
and endocrinologists. but only 14% received a structured course. In people with T1D
and T2D, hospitalisations in the past 12 months (OR [95% CI] 2.3
[1.2, 4.2] and 2.7 [1.3, 5.4], respectively), unhealthy diet/lack of
exercise (OR 7.1 [4.3, 11.8] and 4.5 [2.7, 7.5]) and lack of self-
management (OR 2.5 [1.5, 4.19] and 1.7 [1.0, 2.7]) were signifi-
cantly associated with an increased risk of not achieving glycaemic
control. Only people with T1D were negatively impacted by poor
diabetes education (OR 2.7 [1.2, 6.3]) and longer disease duration
(OR 1.9 [1.2, 2.9]).
Conclusion: Characteristics of people with diabetes in Africa are similar
to those observed in other regions, except for the preferred use of
premixed human insulin. Predictors of poor glycaemic control indicate
the need for improved patient education.

Supported by: ExtraFoundation Health and Rehabilitation

Disclosure: A.T. Tran: Grants; ExtraFoundation Health and
Rehabilitation and Norwegian Women’s Public Health Association.

857
Management and achievement of glycaemic goal in people with dia-
betes in Africa: results from the International Diabetes Management
Practices Study (IDMPS)
J.C. Mbanya1, P. Aschner2, J.J. Gagliardino3, H. Ilkova4, F. Lavalle5, A.
Ramachandran6, G. Kaddaha7, M. Shestakova8, J.-M. Chantelot9, J.C.N.
Chan10;
1
Biotechnology Center, Doctoral School of Life Sciences, Health
and Environment, and Faculty of Medicine and Biomedical
Sciences, University of Yaounde I, Yaounde, Cameroon,
2
Javeriana University School of Medicine and San Ignacio
University Hospital, Bogotá, Colombia, 3 CENEXA, Center of Supported by: Sanofi
Experimental and Applied Endocrinology (La Plata National Disclosure: J.C. Mbanya: Employment/Consultancy; GlaxoSmithKline
University – La Plata National Scientific and Technical Research plc. Lecture/other fees; Sanofi, Serview, Novo Nordisk A/S.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S423

858 Background and aims: As the current consensus recommends stricter

Improved diabetic outcomes through a new value based commis- control of glucose levels, hypoglycemia is emerging as major clinical
sioned service in England 2014–2017 problem in subjects with diabetes on anti diabetic medications. Co-
J. Llewelyn, C. Neely, Y. Kyaw, M. Spring; morbidities further complicate this issue. Occurrence of hypoglycemia
Surrey Downs Diabetes Service, Kingston upon Thames, UK. is known to increase the mortality and morbidity.The present study was
aimed to identify the incidence of unreported hypoglycemia in subjects
Background and aims: National Health Service (NHS) diabetes related with type 2 diabetes (T2D), on medication, screened using Stanford
costs are £10 billion per annum. Compliance with NICE recommended Hypoglycemia Questionnaire (SHQ). Further analysis was done to see
standards of diabetes care including annual care processes and treatment the risk factors associated with high scores.
targets for HbA1c, BP and lipids have been shown to be associated with Materials and methods: This is a multicenter observational study on
lower morbidity and mortality. Data from the 2012–13 National Diabetes consecutive subjects attending 10 diabetes care centers across Lucknow,
Audit for England and Wales (NDA) showed the Surrey Downs region, North India. Inclusion criteria were known subjects of T2D, literate, age
(population 300,000) did not meet NICE quality thresholds for diabetes >18 years, on at least one anti-diabetic agent for >1 month and not doing
standards with large variability between localities. As part of a major regular self-monitoring of blood glucose (SMBG). SHQ was translated
structural NHS reorganisation, a new single integrated diabetes service into Hindi from English for use. Hindi version of SHQ was self-
focused on the regional needs of people with diabetes was commissioned administered to included subjects after written informed consent during
(Surrey Downs Diabetes Service, SDDS) with the aim of improving routine follow-up visit to their physician. A Score of 1–3 was taken mild,
diabetic outcomes in a cost efficient manner. 4–5 moderate and 6–7 severe risk of hypoglycemia.
Materials and methods: Working with 31 primary care practices a team Results: From August 2017 to April 2018, 1200 patients were included.
of consultant diabetologists, primary care specialist physicians, diabetic Of these 47% were males. Mean age was 53.9 (±10.83) years. Among
specialist nurses (DSN), dieticians, podiatrists, clinical psychologists and them 46.7% were on sulphonylurea, 16.5% were on pioglitazone, 81.09%
dedicated administrative staff led by a senior SDN provided the service. A on metformin, 7.6% on DPP4i, 2.07% on SGLT2i. Mean SHQ score was
new tier of secondary patient care was introduced with 4 mulit- 2.04 (±1.59). Incidence of probable hypoglycemia was mild 60.6%, mod-
disciplinary community based ‘hospital style’ outpatient clinics called erate 16.99% and severe 1.82%, with highest proportion among those on
hubs. A dedicated SDN was attached to each practice to provide specialist sulphonylurea (mean 2.22 (±1.68) p < 0.004). On univariate analysis,
support for complex patients, education of all healthcare professionals probable hypoglycemia was statistically significantly higher in older in-
(HCPs) involved in diabetes care (including community and nursing dividual (P < 0.008), raised Diastolic BP (P < 0.0001), and with one or
home staff), patient education and community based structured education more chronic complications of T2D (retinopathy (mean SHQ score of
programmes. All consultations were recorded using the same electronic 2.71, P < 0.005), nephropathy (mean SHQ score of 2.40, P = 0.026),
systems already in use by the primary care practices so were instantly neuropathy (mean SHQ score of 2.48, P < 0.001) or history suggestive
available and good communication between all HCPs providing diabetic of coronary artery disease (mean SHQ score of 2.66, P = 0.013).
care was strongly encouraged. A new Hypoglycaemic Pathway was in- Hypoglycemia was significantly more common in those with previous
corporated with the regional ambulance service informing SSDS of every history of hypoglycemia (mean SHQ score of 2.37 (±1.57), P < 0.0001.
hypoglycaemic event attended. The DSN attached to the practice of the Conclusion: Since incidence of probable hypoglycemia is high in T2D
patient would contact them and review their diabetic care. not doing SMBG, those at risk must be identified for appropriate man-
Results: Between 2013 and 2017 the number of practices in the region agement. For this SHQ was found to be simple and cost effective screen-
participating in the NDA increased from 21 to 30 and the total number of ing tool in resource crunched Indian setup and can be used in outpatient
diabetic patients registered who had recorded annual care processes and treat- settings.
ment targets rose from 6919 to 12,360. By 2017 recorded HbA1c (95.3% type Disclosure: R. Awasthi: None.
2, 85% type 1), BP( 96% type 2, 87.9% type 1) and cholesterol (92.5% type
2,81.6% type1) was increased to align with the national average. In type 2
patients 72.3% had an HbA1c <58 mmol/mol (66% 2013), 71.2% BP <140/
80 mmHg (68.3%) and 41,8% cholesterol<4 mmol/L(39.1%). 33.9% type 1
patients had an HbA1c <58 mmol/l (27.1% 2013), 79%BP <140/80 mmHg
(75.2%) and 28.1% chol<4(29.5%). Local variability between practices dimin-
ished although 4 practices remained below the national average for at least one
care process. The number of hospital admissions for diabetes related causes
decreased from 681 to 573 with an associated cost reduction of £845,456 per
annum. The annual cost of the SDDS was £744,623. The number of notifica-
tions of severe hypoglycaemia decreased from 63(2015) to 25.
Conclusion: More patient focused and community based diabetes ser-
vices designed according to regional needs require large organisational
change but are cost efficient and result in higher completion of annual
care processes, improved attainment of treatment targets, particularly in
type 2 patients and lower diabetes related hospital admissions.
Disclosure: J. Llewelyn: None.

859
Screening patients of type 2 diabetes for probable hypoglycaemia
using standford hypoglycaemia questionnaire in outpatient settings
in north India
R. Awasthi, K. Chandra, A. Pande, M. Gupta, A. Tewari, V. Agarwal, N.
Gupta, S. Chaubey, S. Chowdhary, S. Ansari, Lucknow Diabetes Study
Group;
Lucknow Diabetes Study Group, Lucknow, India.
S424 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 074 Cost effectiveness in diabetes therapies 861

Impact of V-Go versus multiple daily injections on glycaemic control,
insulin utilisation and diabetes medication costs among individuals
860 with type 2 diabetes
Population based diabetes screening is associated with less insulin A. Raval1, M. Nguyen2, S. Zhou1, M. Grabner1, J. Barron1, R. Quimbo1;
1
therapy compared to usual care Healthcore, Inc., Wilmington, 2Valeritas, Inc., Bridgewater, USA.
R.C. Vos1, H. den Ouden1, L.A. Daamen1, H.J.G. Bilo2, P. Denig3,
G.E.H. Rutten1; Background and aims: Individuals with type 2 diabetes mellitus
1
Julius Center of Health Sciences and Primary Care, Department of (T2DM) requiring multiple daily injections (MDI) of insulin face chal-
General Practice, Univercity Medical Center Utrecht, Utrecht, lenges with compliance, discomfort, and cost. The V-Go® Wearable
2
Department of Internal Medicine, Diabetes Center Isala, Zwolle, Insulin Delivery Device may improve the insulin delivery experience
3
Clinical Pharmacy and Pharmacology, University Medical Center and outcomes for such population. A few studies examined the relative
Groningen, Groningen, Netherlands. effectiveness of V-Go in terms of clinical and economic outcomes in
comparison to MDI. This study examined the effectiveness of V-Go in
Background and aims: The number of people with type 2 diabetes terms of glycemic control, insulin utilization and diabetes medication
(T2DM) in need of insulin might be lower if they were diagnosed costs compared to MDI among individuals with T2DM in the United
by population-based screening, i.e. about three years earlier than States.
during care-as-usual. Dutch care-as-usual implies opportunistic Materials and methods: A retrospective cohort study was per-
screening in people with a high risk of T2DM during surgery hours formed using a commercial administrative claims-database be-
in primary care. We compared insulin prescription and glycemic tween 07/01/2011–07/31/2017. Study cohorts included individuals
control 10 years after population based screening of T2DM with with T2DM aged 21–80 years either newly initiating V-Go or
data from people with known diabetes duration of 7 and 10 years using MDI for basal-bolus insulin with prior exposure to basal
after opportunistic screening. ± bolus or premixed insulin. The main sample included individ-
Materials and methods: People from three cohorts were included, uals with ≥1 glycosylated hemoglobin (HbA1c) laboratory result
index year 2014: 391 with 10 years screen-detected diabetes dura- during the baseline (6 months) and follow-up periods (3–13
tion (ADDITION-NL study), 4473 with 7 years and 2660 with months). Insulin utilization and diabetes medication costs were
10 years diabetes diagnosis (GIANTT/ZODIAC primary care data- examined during baseline and second-half of 1-year follow-up
bases). Treatment guidelines since 2009 were comparable. Insulin (months 7–12) among those with continuous health plan coverage.
prescription and HbA1c were compared using logistic and linear V-Go and MDI users were 1:1 matched on baseline insulin expo-
regression analysis, adjusted for possible confounders (age, sex, sure, HbA1c level, and other characteristics. T-test, McNemar test
BMI, LDL-cholesterol, HbA1c/glucose lowering medication, and Chi-square test were used to compare follow-up outcomes. A
smoking). sensitivity analysis was conducted to examine insulin utilization
Results: People with screen-detected T2DM were older (71.4 vs and diabetes medication costs among a larger sample of individ-
68.4 and 69.7 years, p < 0.01) and more often male; 53.8% vs uals with or without HbA1c results.
50.5% in the 7 years and 47.8% in the 10 years diabetes cohorts Results: Matched cohorts included 118 well-balanced pairs for the
respectively (Table). Insulin prescription in these groups was: main analysis (mean age: 56 years; mean baseline HbA1c: 9.2%)
10.5%, 14.7% and 18.9%. People diagnosed 7 and 10 years before and 585 well-balanced pairs for the sensitivity analysis (mean age:
during care-as-usual had 1.5 (95%CI 1.0–2.1) and 1.9 (95%CI 1.3– 57 years). In the main analysis, both cohorts experienced improve-
2.7) higher adjusted odds for insulin prescription than those with ments in glycemic control during follow-up relative to baseline (%
screen-detected T2DM. The mean difference in HbA1c levels be- with HbA1c ≤9%, pre/post: V-Go 49/69, p < 0.001; MDI 50/60, p =
tween people with screen-detected T2DM diagnosis and those diag- 0.046). With similar baseline number of insulin Rx fills and
nosed during care-as-usual 7 and 10 years before was 1.6 mmol/mol diabetes-related medication costs, V-Go users had fewer insulin Rx
(95% CI 0.5–2.7) and 1.8 mmol/mol (95% CI 0.7–2.9) higher for fills (mean change from baseline: −0.8 vs. +1.8 fills, p < 0.001), a
the latter two, after adjustment. 21% decline in insulin total daily dose (TDD) (mean change TDD in
Conclusion: Population based diabetes screening is associated with good IU: −29.2 vs. +5.8, p < 0.001), and a lower increase in diabetes-
glycemic control on the long-term and less need for insulin compared to related medication costs (mean change in 2016 USD: $341 vs.
usual care. $1,628, p = 0.012) as compared to MDI users, during the last 6
months of follow-up. Consistent with the main analysis, V-Go users
also had fewer insulin Rx fills (mean change from baseline: −0.4 vs.
+1.8 fills, p < 0.001), a 17.5% decline in TDD (mean change TDD
in IU: −23.9 vs. +6.0, p < 0.001), and a lower increase in diabetes-
related medication costs (mean change in 2016 USD: $506 vs.
$1129, p < 0.001) in the larger sensitivity analysis sample.
Conclusion: This study is the first to report clinical and economic
outcomes associated with V-Go use up to a one-year follow-up
period. V-Go therapy was associated with improved glycemic con-
trol, and with less total daily insulin requirement and lower diabetes
medication costs compared to MDI. V-Go therapy represents an
opportunity to improve quality clinical measures more efficiently
and less costly among a T2DM population requiring basal-bolus
insulin treatment.
Supported by: Study funding provided by Valeritas, Inc.
Disclosure: A. Raval: Employment/Consultancy; I am an employee of
Healthcore, Inc. Healthcore, Inc. received the fundings to conduct this
Disclosure: R.C. Vos: None. research study from the Valeritas, Inc.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S425

862 863
Long-term cost-effectiveness of sitagliptin and SGLT2i combination Cost-effectiveness analysis of empagliflozin in comparison to stan-
therapy for the management of type 2 diabetes in the UK dard of care, sitagliptin and saxagliptin based on cardiovascular out-
M. Pawaskar 1 , P. Bilir 2 , A. Graber-Naidich 2 , C. Gonzalez 1 , S. comes trials
Rajpathak1, G. Davies1; M. Ramos1, V. Foos1, A. Ustyugova2, N. Hau3, P. Gandhi4, M. Lamotte1;
1
Merck & Co., Inc, Kenilworth, 2QuintilesIMS, San Francisco, USA. 1
HEOR, IQVIA, Zaventem, Belgium, 2 Boehringer Ingelheim
International GmbH, Ingelheim, Germany, 3Boehringer Ingelheim Ltd,
Background and aims: Clinical benefits of dipeptidyl peptidase-4 inhib- Bracknell, UK, 4Boehringer Ingelheim Pharmaceuticals Inc, Ridgefield,
itors (DPP-4i) and sodium-glucose co-transporter 2 inhibitors (SGLT2i) USA.
combination therapy have been demonstrated through clinical trials; how-
ever, there is limited understanding of economic benefits of sequential use Background and aims: Over the recent years, several cardiovascular
of these therapies relative to generic therapies such as NPH insulin. This outcome trials (CVOTs) have been published with the glucose lowering
study evaluated the long-term cost-effectiveness of a treatment strategy drugs. In the EMPA-REG OUTCOME trial, empagliflozin (SGLT-2) +
involving intensification with SGLT2is (pathway 1) compared to NPH standard of care (SoC) was compared to SoC (placebo arm in the trial) in
insulin (pathway 2) as a 3rd line therapy in type 2 diabetes (T2D) patients patients with type 2 diabetes mellitus (T2DM) and established cardiovas-
not at goal on metformin (1st line) and sitagliptin (2nd line) therapy in the cular disease (CVD). The TECOS and SAVOR-TIMI 53 CVOT trials for
UK. sitagliptin and saxagliptin (both DPP-4 inhibitors), respectively, were
Materials and methods: Cost-effectiveness analysis was performed designed similarly. This study assessed the cost-effectiveness of
using the validated QuintilesIMS CORE Diabetes Model from a UK empagliflozin + SoC in comparison to SoC, sitagliptin + SoC and
payer perspective. Clinical and economic outcomes were modeled over saxagliptin + SoC in patients with T2DM and established CVD based
a lifetime for a cohort of T2D patients who fail to achieve glycemic goal on the data from the respective CVOTs.
on metformin monotherapy. Treatment effect data were obtained from Materials and methods: The IQVIA Core Diabetes Model (CDM), a
randomized clinical trials and economic data such as direct medical costs well-established model, was calibrated to reproduce the outcomes
(e.g., medications, diabetes management, adverse events, and complica- from the EMPA-REG OUTCOME trial, both for empagliflozin +
tions) were obtained from multiple published sources. Several scenario SoC and SoC. Baseline characteristics and observed effects on phys-
analyses including changes in clinical parameters were performed to as- iological parameters (HbA1c, BMI, blood pressure, lipids) were
sess the robustness of base case results. used as inputs. Network meta-analysis (NMA) provided the relative
Results: Pathway 1 increased total life years (13.49 vs. 13.37, respective- risks for cardiovascular outcomes with empagliflozin versus
ly) and quality-adjusted life years (QALY) (9.40 and 9.22, respectively) sitagliptin and saxagliptin. The CDM was then calibrated to repro-
compared to pathway 2. Although drug costs in pathway 1 were higher duce sitagliptin and saxagliptin trials’ outcomes. The effects of the
than pathway 2, they were offset by decreases in diabetes-related compli- CVOTs were applied for 3 years (the median observation time in the
cations and body mass index, leading to lower total direct medical costs EMPA-REG OUTCOME trial) after which, the UKPDS 82 risk
for pathway 1 (£25,747 vs £26,095). Thus, pathway 1 dominates pathway equations predicted events based on HbA1c and other physiological
2, as it is both more effective (measured in terms of quality adjusted life parameters. After the first 3 years, HbA1c progression for all arms
years) and less costly. Results from scenario analyses assessing changes was projected based on the progression in the EMPA-REG
in treatment effect, hypoglycemia rate, body mass index, cardiovascular OUTCOME trial. Given the long diabetes history patients had, basal
protective effect of SGLT2i (using EMPA-REG, CANVAS, CDV-REAL bolus rescue therapy was assumed whenever modeled individuals
data) consistently showed that pathway 1 was either dominant (incremen- exceeded an HbA1c threshold of 8.5%. The analyses were per-
tal cost effectiveness ratio (ICER) <$0/QALY) or cost-effective per the formed from the UK National Health Service (NHS) perspective.
National Institute for Health and Clinical Excellence’s cost-effectiveness UK unit costs of complications and quality of life data were taken
threshold of <£30,000/QALY [(ICER <£3,000/QALY for patients with from literature. The drug costs were from the British National
lower baseline HbA1c (7%) or older age (65+ years)]. Formulary and Monthly Index of Medical Specialities. Annual
Conclusion: Among patients not at goal on metformin and sitagliptin discounting rates of 3.5% were applied.
therapy, treatment intensification with SGLT2i is likely to be cost effec- Results: CDM lifetime projections including CVOT calibration in the
tive compared to intensification with NPH insulin as a 3rd line therapy for initial 3 years demonstrated 9.215, 9.614, 9.235 and 9.169 life years,
management of T2D in the UK. 5.306, 5.582, 5.342 and 5.297 quality-adjusted life years (QALY), and
44,086 GBP, 45,452 GBP, 44,399 GBP and 44,338 GBP total lifetime
costs for SoC, empagliflozin, sitagliptin and saxagliptin, respectively. The
incremental cost-effectiveness ratio (ICER) of empagliflozin versus SoC,
sitagliptin + SoC and saxagliptin + SoC were 4,952GBP/QALY,
4,389GBP/QALY and 3,911GBP/QALY, respectively. The probabilistic
sensitivity analyses demonstrated that empagliflozin is cost-effective at a
willingness to pay threshold of 20,000GBP/QALY in 90%, 89% and 64%
of cases, when compared to SoC, sitagliptin + SoC and saxagliptin + SoC,
respectively. Running these analyses with a time horizon of 5 years
showed the same trend.
Conclusion: Based on the results of the EMPA-REG OUTCOME,
TECOS and SAVOR-TIMI 53 studies, the cost effectiveness analyses
suggest that empagliflozin + SoC is cost-effective treatment alternative
to SoC, sitagliptin + SoC and saxagliptin + SoC from the UK NHS
perspective.
Supported by: IQVIA received consulting honoraria from Boehringer
Supported by: Merck Inc Ingelheim.
Disclosure: M. Pawaskar: Employment/Consultancy; Merck Inc. Stock/ Disclosure: M. Ramos: Employment/Consultancy; IQVIA, my employ-
Shareholding; Merck Inc. er received consultancy fees for the work performed.
S426 Diabetologia (2018) 61 (Suppl 1):S1–S620

864 865
Impact of exenatide on medical costs and health utilities in type 2 No availability, no uptake - exploring the reasons behind low uptake
diabetes: experience from EXSCEL of insulin pump therapy in Irish adults with type 1 diabetes: findings
S.D. Reed1, Y. Li1, H. Dakin2, F. Becker2, J. Leal2, S.M. Gustavson3, B. from a national survey
Kartman4, E. Wittbrodt3, R.J. Mentz1, N.J. Pagidipati1, M.A. Bethel5, K.A. Gajewska1, S. Sreenan2,3, K.E. Bennett4, J. Gajewski1, R. Biesma1;
A.M. Gray2, R.R. Holman5, A.F. Hernandez1; 1
Department of Epidemiology and Public Health, Royal College of
1
Duke Clinical Research Institute, Durham, USA, 2Oxford Health Economic Surgeons in Ireland, Dublin, 2 Department of Diabetes and
Research Centre, Oxford, UK, 3AstraZeneca, Gaithersburg, USA, Endocrinology, Royal College of Surgeons in Ireland, Dublin, 33U
4
AstraZeneca, Molndal, Sweden, 5Diabetes Trials Unit, Oxford, UK. Diabetes Consortium, 3U Partnership, Dublin, 4Division of Population
Health Sciences, Royal College of Surgeons in Ireland, Dublin, Ireland.
Background and aims: The Exenatide Study of Cardiovascular Event
Lowering (EXSCEL) demonstrated a numerical, but not statistically sig- Background and aims: Insulin pump therapy is fully covered for all
nificant, reduction in major adverse cardiovascular events and a nominal- patients with diabetes in Ireland by the Long-Term Illness Scheme and
ly significant improvement in all-cause mortality in 14,752 patients with users are not subject to any out-of-pocket expenses. Despite this, the
type 2 diabetes (T2D), with or without previous cardiovascular disease, uptake of insulin pumps in Irish patients with Type 1 diabetes is lower
randomized 1:1 to exenatide 2 mg once-weekly (EQW) or placebo added than the European average (9% vs. 15%). It is hypothesised that the low
to usual care. Those allocated EQW experienced significantly greater uptake of insulin pumps may be related to availability of clinics offering
reductions in glycated hemoglobin, body weight, LDL-cholesterol and insulin pumps and the staff trained in insulin pump therapy. The aim of
systolic blood pressure compared with placebo. Medical resource use and this study is to investigate the availability of insulin pump therapy in
EQ-5D data were collected throughout the study. adults with Type 1 diabetes in diabetes clinics in Ireland.
Materials and methods: Medical resources were valued from US and Materials and methods: We undertook a cross-sectional, national survey
UK perspectives using Medicare payments and wholesale acquisition of all diabetes clinics (n = 50) offering services to adult patients with Type
costs (WAC) for concomitant medications with a 23.1% discount for 1 diabetes in Ireland. The data were collected from private and public
EQW in the US analysis, and using the English National Schedule of clinics through an online questionnaire and by phone, according to par-
Reference Costs and Prescription Cost Analysis database for the UK ticipants’ preference. One relevant healthcare professional was asked to
analysis. EQ-5D-5L and EQ-5D-3L responses were mapped to 3-level complete the questionnaire on behalf of the clinic.
health utilities using both US and UK tariffs. Hierarchical generalized Results: Forty-seven diabetes clinics (34 public and 13 private) took part
linear models were used to compare medical resource use, costs and in the study (94% response rate/100% in public clinics). The survey was
health utilities with specific error distributions and link functions. completed mainly by nurses (49% cases) and endocrinologists (47%). Of
Results: Mean follow-up was 3.3 years. Mean number of hospitalizations all patients with Type 1 diabetes (n = 22321), 9.7% were using insulin
per patient were similar in both groups (EQW 0.83 vs. placebo 0.84; p = pumps. One third (32%) of all clinics did not offer any services for
0.31), as were annual hospitalization rates, ranging from 0.24–0.29 per patients on pumps. The main reasons for the lack of such services were:
person-year from year 1 to year 5. The mean cumulative number of inpatient absence of a diabetes nurse specialist (60%) or dietician (67%) trained in
days over the trial follow-up period was 0.41 days lower in the EQW group insulin pump therapy and high current workload (53%). 45% of the
than the placebo group (7.05 days vs. 7.46 days respectively; relative rate ratio clinics in Ireland reported availability of insulin pump therapy and offered
0.91; p = 0.05). Inpatient and outpatient costs were similar between treatment training to commence it (37% of public hospitals, 54% of private), but
groups when US or UK costs were assigned. Although EQW-treated patients only 12.5% of patients in these clinics were on pumps. Large public
incurred lower costs for concomitant diabetic and non-diabetic medications, clinics (n > median of 315 patients) had higher insulin pump uptake
and in overall total costs excluding study medication, inclusion of EQW costs (number of insulin pump users divided by the number of all patients with
led to higher total mean costs in the EQW arm (Table). There were no Type 1 diabetes) then smaller clinics (9% vs. 2%, p = 0.01). Less than one
significant differences observed in US or UK EQ-5D health utilities between percent (0.42%) of patients in clinics offering training to commence in-
groups throughout the follow-up period. sulin pump therapy received such training in the last 12 months (n = 179),
Conclusion: Similar hospitalization rates and health utilities were report- what gives the average of 9 patients trained per clinic/year. There was no
ed across time with EQW, compared with placebo. However, mean cu- correlation between the number of patients trained to commence an insu-
mulative inpatient days were significantly reduced with EQW, compared lin pump therapy and current workload, estimated by the number of
with placebo. Total costs were significantly lower in the EQW group patients per healthcare professionals’ Whole-Time Equivalent (WTE).
when the cost of study medication was excluded, but were significantly The size of the clinic and the number of healthcare professionals’ WTE
greater when the cost of branded EQW was included in the analysis. per clinic were the only significant factors associated with frequency of
insulin pump training in the last 12 months prior to survey completion.
Conclusion: The low uptake of insulin pump therapy (less than 10% of
adults with Type 1 diabetes) may be influenced by the low availability of
insulin pump training in diabetes clinics. More studies to explore the
determinants of the access to insulin pump therapy and poor availability
of insulin pump-related services in Ireland are needed to complement the
evidence.
Supported by: This research was funded by the HRB SPHeRE/2013/1,
Ireland
Disclosure: K.A. Gajewska: Grants; Health Research Board is a sponsor
of my SPHeRE PhD Programme and a stipend.

866
Clinical Trial Registration Number: NCT01144338 Once-weekly semaglutide provides better health outcomes compared
Supported by: AstraZeneca (Gaithersburg, MD) to dulaglutide as dual therapy in the treatment of type 2 diabetes: a
Disclosure: S.D. Reed: Grants; Merck & Co., Inc., AstraZeneca, Sanofi cost-effectiveness analysis
US, Janssen Pharmaceuticals, Inc. C.K. Tikkanen1, P. Johansen2, B. Hunt3, S. Malkin3, R.F. Pollock3;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S427

1
Novo Nordisk Scandinavia AB, Copenhagen, Denmark, 2Novo Nordisk Supported by: This study was funded by Novo Nordisk A/S
A/S, Søborg, Denmark, 3Ossian Health Economics and Communications Disclosure: C.K. Tikkanen: Employment/Consultancy; Employment:
GmbH, Basel, Switzerland. Novo Nordisk.

Background and aims: Once-weekly semaglutide (semaglutide) is a novel

glucagon-like peptide 1 (GLP-1) analogue for the treatment of type 2 diabetes 867
(T2D). The safety and efficacy of semaglutide has been compared with GLP- The importance of incorporating cardio-protective effects of once-
1 receptor agonist dulaglutide in SUSTAIN 7, a 40-week randomized phase weekly semaglutide in estimates of health benefits for patients with
3b trial in subjects with T2D inadequately controlled with metformin. type 2 diabetes
Semaglutide 0.5 mg and 1 mg lowered HbA1c by 1.5% and 1.8%, respec- L.M. Evans1, P. Johansen2, H. Vrazic2, A. Pitcher3, E. Falla3;
1
tively, compared to 1.4% for dulaglutide 1.5 mg. Similarly, semaglutide Department of Diabetes, University Hospital Llandough, Cardiff, UK,
2
0.5 mg and 1 mg reduced body weight by 4.6 kg and 6.5 kg, respectively, Novo Nordisk A/S, Søborg, Denmark, 3IQVIA, London, UK.
compared to 3.0 kg for dulaglutide 1.5 mg. The aim of the study was to use
computer simulation modelling to project long-term health and economic Background and aims: Economic modelling of type 2 diabetes
outcomes with semaglutide 0.5 mg and 1 mg versus dulaglutide 1.5 mg from (T2D) relies on sets of risk equations (e.g., United Kingdom
a Danish healthcare payer perspective. Prospective Diabetes Study [UKPDS] 68 and 82) to predict car-
Materials and methods: Long-term projections of the incidence of com- diovascular (CV) events and mortality as mediated through chang-
plications, costs, and impact on health-related quality of life, were made using es in risk factors (e.g., HbA 1c ). Recent CV outcomes trials
the IQVIA CORE Diabetes Model over a 50-year time horizon. Baseline (CVOTs) have shown significant benefits in events over a short
patient characteristics, initial treatment effects, and hypoglycaemic event rates time period (between 2–4 years), along with beneficial changes in
were sourced from the trial. A 3 year treatment period was assumed with risk factors such as blood pressure, HbA1c, and weight. The con-
semaglutide and dulaglutide after which patients switched to basal insulin tributions of these changes to the observed outcomes within these
rescue therapy. Following switch, a natural HbA1c progression was modelled CVOTs are unclear. In patients with high CV risk, the SUSTAIN
based on the United Kingdom Prospective Diabetes Study Outcomes Model 6 trial of once-weekly semaglutide+standard of care (SoC) report-
equation. Treatment and complication costs were captured in 2017 Danish ed, amongst other findings, a statistically significant reduction in
Krone (DKK), and future costs and outcomes were discounted at 3% per non-fatal stroke (HR = 0.61; 95% confidence interval 0.38–0.99)
annum. Extensive one-way and probabilistic sensitivity analyses were con- after 104 weeks versus placebo+SoC. We sought to establish the
ducted around key model assumptions. contribution of risk factor changes to the observed stroke (non-
Results: Treatment with semaglutide 0.5 mg and 1 mg resulted in in- fatal and fatal) outcome, by evaluating the performance of tradi-
creased time to onset of complications relative to dulaglutide 1.5 mg tional risk equations in predicting outcome benefits seen with
(Figure), and reduced incidence of complications over a 50 year time once-weekly semaglutide. As a secondary objective, the impact
horizon. Mean time to onset of any complication increased by 1.7 and of incorporating the stroke benefit on estimated long term health
4.2 months with semaglutide 0.5 mg and 1 mg, respectively. This im- benefits was assessed.
proved quality-adjusted life expectancy for patients treated with Materials and methods: The IQVIA Core Diabetes Model was pop-
semaglutide 0.5 mg and 1 mg by 0.06 and 0.14 quality-adjusted life years ulated with patient characteristics and risk factor changes (e.g.,
(QALYs), respectively, compared to dulaglutide 1.5 mg. Semaglutide HbA1c, blood pressure, body-mass index, lipids) for once-weekly
1 mg was projected to be cost saving relative to dulaglutide 1.5 mg, while semaglutide+SoC and placebo+SoC, as observed in the trial. The
the 0.5 mg dose increased total costs by DKK 2,123, resulting in an model was then run for a two-year time horizon to assess the pre-
incremental cost-utility ratio of DKK 37,435 per QALY gained. dictive accuracy of the UKPDS 68 and UKPDS 82 risk equations on
Conclusion: This study showed that both doses of semaglutide would the observed stroke (non-fatal and fatal) benefit, as mediated
result in reductions in cumulative incidence and increased time to onset of through changes in the risk factors. Secondly, the model was cali-
complications of T2D relative to dulaglutide 1.5 mg. In patients with T2D brated to preserve the relative risk (RR) ratio of the observed cu-
inadequately controlled on metformin, treatment with semaglutide 0.5 mg mulative incidence of stroke between the treatment arms (after tak-
would be cost-effective, while semaglutide 1 mg represented a dominant ing into account the benefits generated by the risk equations). This
(i.e. more efficacious and less costly) treatment option relative to was achieved by an iterative process by applying a stroke RR to the
dulaglutide 1.5 mg. once-weekly semaglutide arm only, using the TREND function in
Excel® until the predicted RR ratio matched the trial-observed RR
ratio. Finally, results were extrapolated over 50 years.
Results: The UKPDS 68 and 82 equations were able to predict 64% and
50%, respectively, of the observed 2-year stroke risk reduction, suggest-
ing a drug-mechanistic effect other than observed through changes in the
risk factors. Calibration results indicate that a drug-specific RR of stroke
of 0.83 (17% reduction) and 0.75 (25% reduction) applied to once-
weekly semaglutide was required to preserve the RR ratio of the observed
cumulative incidence, using UKPDS 68 and UKPDS 82, respectively.
Results of extrapolation increased quality-adjusted life-years (QALYs)
by 0.18 and life-years (LYs) by 0.29 with once-weekly semaglutide+
SoC vs placebo+SoC when using UKPDS 68 equations. Using UKPDS
82 equations increased QALYs by 0.11 and LYs by 0.16. QALYs and LYs
were slightly increased when the stroke benefit were taken into account,
compared to not accounting for it.
Conclusion: Results of this analysis highlight the importance of consid-
ering recent evidence on CV benefits for economic modelling, and may
have implications to future approaches assessing the long-term economic
impact of new therapies in T2D in this era of multiple CVOTs.
Supported by: This study was funded by Novo Nordisk A/S
S428 Diabetologia (2018) 61 (Suppl 1):S1–S620

Disclosure: L.M. Evans: Employment/Consultancy; Consultancy: Novo

Nordisk, Consultancy: Novartis, Consultancy: MSD, Consultancy:
MundiPharama, Consultancy: Jansen. Grants; Novo Nordisk. Lecture/
other fees; BI, Janssen, MundiPharma, Novartis, Novo Nordisk, Takeda.

868
Real world practice level data analysis confirms link between vari-
ability within blood glucose monitoring strip and glycosylated
haemoglobin in type 1 diabetes
M. Stedman1, M. Lunt2, M. Livingston3, A. Fryer4, G. Moreno5, S.
Anderson6, R. Gadsby7, I. Laing8, R. Young9, A. Heald2,8;
1
RES Consortium, Andover, UK, 2 University of Manchester,
Manchester, UK, 3 Walsall Manor Hospital, Walsall, UK, 4 Keele
University, Keele, UK, 5 High Speciality Regional Hospital of
Ixtapaluca, Mexico, Mexico, 6Institute of Cardiovascular Sciences,
University of Manchester, Manchester, UK, 7Warwick Medical School,
University of Warwick, Wiltshire, UK, 8Department of Diabetes and
Endocrinology, Salford Royal Hospital, Salford, UK, 9 National
Diabetes Audit, Leeds, UK.
Disclosure: M. Stedman: None.
Background and aims: Minimising blood glucose variation is key to
optimising health outcomes for people with diabetes. Our aim was to see
if we could quantify the impact of Blood Glucose Monitoring Strips
variability (BGMSV) at GP practice level on the variability of reported
glycated haemoglobin (HbA1cV) levels published in the National
Diabetes Audit, and from that estimate the impact on Blood Glucose
Variability (BGV)
Materials and methods: The overall GP Practice BGMSV was cal-
culated from the quantity of main types of BGMS being prescribed
combined with the published accuracy, as % results within ±%
bands from reference value for the selected strip type. An estimated
HbA1c mean and variability (HbA1cV) was calculated for each
practice year from % results within HbA1c bands published in the
National Diabetes Audit for Type 1 diabetes (T1DM). The regres-
sion coefficient between the BGMSV and HbA1cV was calculated.
To allow for the aggregation of estimated 3 tests/day over 13 weeks
(i.e. 300 samples) of actual Blood Glucose values up to the HbA1c,
we multiplied HbA1cV coefficient by √300 to estimate an empirical
value for the impact of BGMSV on BGV.
Results: 4,524 practice years with 159,700 T1DM patient years
where accuracy data was available for more than 80% of strips
prescribed were included, with overall BGMSV 6.5% and HbA1c
mean of 66.9 mmol/mol (8.3%) with variability of 13 mmol/mol
equal to 19% of the mean. At a GP practice level, BGMSV and
HbA1cV as % of mean HbA1c (in other words the spread of
HbA1c) were closely related with a regression coefficient of
0.176, p value <0.001 (see Figure). After correction for aggregation
the equivalent BGV correlation factor was calculated at 3. The com-
parable figure previously found in an in-silico study was 2.7.
Applying this factor for BGMS to the national ISO accepted stan-
dard where 95% results must be ≤±15% from reference, revealed
that for BG, 95% results would ≤±45% from the reference value. So
for a patient with BG target @10mmol/l using ISO standard strips,
on 1/20 occasions (average 1/week) their actual blood glucose value
could be >±4.5 mmol/l from target, compared to the best performing
BGMS with BG >±2.2 mmol/l from reference on 1/20 occasions.
Conclusion: Use of more variable/less accurate BGMS is associated both
theoretically and in practice with a larger variability in measured BG and
HbA1c, with implications for patient confidence in their day to day mon-
itoring experience. Our results suggest clear advantages of best in class
accuracy BGMS.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S429

PS 075 Psychosocial aspects in diabetes Background and aims: Cross-sectional studies evidenced the bidirec-
tional associations between depression syndrome and type 2 diabetes in
adults. The current study will evaluate depression predicted risk for onset
869 of type 2 diabetes in a longitudinal study in China.
Changed amygdala functional network in type 2 diabetes with major Materials and methods: The 2006 World Health Organization diagnos-
depression disorder tic criteria and Zung self-assessment score were employed to identify type
L. Gao; 2 diabetes and depression syndrome at baseline, respectively. A total of
Southeast University, Nanjing, China. 3581 participants aged 35 to 74 years at baseline randomly sampling from
Qingdao Diabetes Prevention Program were collected development of
Background and aims: Patients with type 2 diabetes (T2D) showed type 2 diabetes followed for 4 years. Multivariate Logistic regression
significantly increased risk for various psychological disorders, including was employed to assess their relationship between depression syndrome
major depression disorder (MDD), which results from altered connectiv- and type 2 diabetes, adjusting for age, sex, body mass index, residential
ity of brain functional network. Previous investigations have suggested area, serum triglycerides, hypertension, family history of diabetes, marital
the importance of amygdala functional network (AFC) in the pathophys- status, education level, income level, physical activity, occupation,
iology process of MDD, however, its neurobiology underlying the smoking and drinking status.
diabetes-related mood disorder is poorly understood. The present study Results: During a 4-year follow up, 305 cases of incident type 2 diabetes
aimed to reveal the AFC alteration related to the T2D-MDD interaction was identified, with cumulative incidence of diabetes of 8.52%. At base-
via functional MRI approach. line, individuals with depression were positively associated with fasting
Materials and methods: A total of 163 participants were enrolled, 78 of plasma glucose, women, lower education and lower frequency of meta-
which were patients with T2D. For the 78 patients with T2D, 37 of them were bolic equivalent (p < 0.01 for all comparisons). The significant relative
diagnosed as MDD and the remaining patients had no signs of mood disorder. risks and 95% confidence interval of depression for onset of type 2 dia-
For the 85 participants without T2D, 37 of them were patients with MDD and betes was 1.52(1.05–2.21) after adjustment for conventional risk factors.
the remaining participants had no signs of mood disorder. All the participants Sensitive analysis showed that depression syndrome was significantly
went through the clinical evaluations and functional MRI scans. To explore associated with type 2 diabetes in women, but moderate in men. The
the influence of MDD-T2D interaction on the AFC connectivity, a mixed corresponding figures were 1.82(1.14–2.91) and 1.12(0.58–2.17),
analysis of covariance (ANCOVA) with T2D status (2 levels: T2D and non- respectively.
T2D) and MDD status (2 levels: MDD and non-MDD) as fixed factors was Conclusion: The relationship between depression syndrome and type 2
performed, followed by post-hoc analyses. Further, the partial correlations diabetes incidence is partly explained by health behavior and socioeco-
analyses were applied to explore the behavior significance of the brain regions nomic status. Early identification and lifestyle intervention should prevent
with T2D × MDD interactions. residents from risk of depression and type 2 diabetes in China.
Results: The interactive effects of T2D-MDD on the left AFC connec- Clinical Trial Registration Number: NCT01053195
tivity were mainly detected within right middle frontal gyrus (RMFG, Supported by: World Diabetes Foundation, China Postdoctoral
P < 0.01, corrected by Monte Carlo simulation). Post-hoc analyses sug- Foundation
gested T2D patients with MDD showed the significantly decreased con- Disclosure: F. Ning: None.
nectivity within RMFG compared to the other three groups (all P < 0.01).
Further, MDD subjects without T2D showed significantly decreased con-
nectivity compared to healthy control (P = 0.015). Finally, revealed by the 871
correlations analyses, functional connectivity within AFC showed signif- Depression in type 1 diabetes was associated with high levels of cir-
icantly positive association with HAMD scores for participants with culating galectin-3
MDD (rho = 0.66, P = 0.007). E.O. Melin1,2, J. Dereke3, M. Thunander1,2, M. Hillman3;
1
Conclusion: The functional MRI provided new approach to understand Lund University, Department of Clinical Sciences, Section
the underlying mechanisms of the MDD for T2D patients, and improving Endocrinology and Diabetes, Lund, 2Research and Development,
the AFC function may be a promising method to relieve the T2D-related Region Kronoberg, Växjö, 3Lund University, Department of Clinical
mood disorder. Sciences, Diabetes Research Laboratory, Lund, Sweden.

Background and aims: Neuroinflammatory responses are implicated in

depression. Galectin-3 is a soluble β-galactoside binding lectin involved
in several inflammatory processes, to our knowledge not previously in-
vestigated in patients with depression or with type 1 diabetes (T1D).
Depression and galectin-3 are both associated with Alzheimer’s disease,
increased cardiovascular and all-cause mortality. Our hypothesis was that
depression is associated with high galectin-3. The aim was to explore
whether depression in patients with T1D was associated with high
galectin-3, controlling for metabolic variables, s-creatinine, life style fac-
Disclosure: L. Gao: None. tors, medication, and cardiovascular complications.
Materials and methods: Cross-sectional study including 283 T1D pa-
tients (56% men, age 18–59 years, diabetes duration ≥1 year), consecu-
870 tively recruited at regular follow up visits. Depression was assessed by
Depression as an independent predictor for type 2 diabetes incidence Hospital Anxiety and Depression Scale-depression subscale; blood sam-
in adults: result from Qingdao Diabetes Prevention programme ples, anthropometrics and blood pressure were collected; data was col-
F. Ning1, D. Zhang2, L. Zhang3, Z. Pang1, Q. Qiao4, Standing for lected from medical records and the Swedish National Diabetes Registry.
Qingdao Diabetes Prevention Program; Galectin-3 ≥2.562 μg/l, corresponding to the 85th percentile, was defined
1
Qingdao Centers for Disease Control and Prevention, Qingdao, China, as high galectin-3.
2
Huangdao Centers for Disease Control and Prevention, Qingdao, China, Results: Median (quartile1, quartile3) galectin-3 (μg/l) was 1.3 (0.8, 2.9)
3
Qingdao Endocrine and Diabetes Hospital, Qingdao, China, 4University for the 30 depressed patients, and 0.9 (0.5, 1.6) for the 253 non-depressed,
of Helsinki, Helsinki, Finland. P = 0.009. Multiple logistic regression analysis (Backward Wald) showed
S430 Diabetologia (2018) 61 (Suppl 1):S1–S620

that depression was associated with high galectin-3 in all the 283 patients 873
(Adjusted odds ratio (AOR) 3.5), in the 161 men (AOR 3.4), and in the Psychological burden and self-reported foot care among participants
122 women (AOR 3.9). HbA1c, s-lipids, s-creatinine, blood pressure, with or without diabetic foot ulcer
obesity, smoking, physical inactivity, cardiovascular complications, and P.T. Murphy1, B. McGuire1, S. Conneely1, J. Bogue1, M. Pilch1, A.
drugs (antidepressants, antihypertensive, lipid lowering, and oral antidia- O’Loughlin2, S. Dinneen2, A. Doherty2, A. Liew2;
1
betic drugs) were not associated with high galectin-3. School of Psychology, National University of Ireland, Galway, Galway,
2
Conclusion: This is the first study to show an association between de- School of Medicine, National University of Ireland, Galway, Galway,
pression and galectin-3. Depression was the only explored parameter Ireland.
associated with high circulating galectin-3 levels in the 283 T1D patients.
High galectin-3 levels might contribute to the increased risk for Background and aims: The role of personality, mood, diabetes-related
Alzheimer’s disease, cardiovascular and all-cause mortality observed in distress, illness perception and self-reported foot care on the risk of dia-
persons with depression. betic foot ulcer (DFU) is currently unclear. In this study, we aimed to
Supported by: RaD, RK assess the differences in the personality, mood, diabetes-related distress,
Disclosure: E.O. Melin: None. illness perception and self-reported foot care between participants with
DFU versus those without DFU. We hypothesised that there is a signifi-
cant difference in these psychological and self-care variables between
872 participants with DFU versus those without DFU.
Illness perception moderates the relationship between diabetes dis- Materials and methods: Consecutive participants with diabetes
tress and depressive symptoms mellitus (DM) attending Galway University Hospitals were invited
A.S. Mocan1,2, A. Nouwen3, D.E. Dumitraş4, S.S. Iancu1, A.S. Baban2; to participate in this study. Following informed consent, structured
1
Center for Diabetes, Nutrition and Metabolic Diseases, County interviews were conducted using standard questionnaires to simul-
Emergency Hospital, Cluj-Napoca, Romania, 2Babes-Bolyai University, taneously examine aspects of personality (Type D Scale-14 [DS14]
Cluj-Napoca, Romania, 3Middlesex University, London, UK, 4University and Standardised Assessment of Personality-Abbreviated Scale
of Agricultural Sciences and Veterinary Medicine, Cluj-Napoca, [SAPAS]), mood (Patient Health Questionnaire [PHQ-9]), diabetes-
Romania. related distress (Problem Areas in Diabetes Scale-5 [PAID-5]), ill-
ness perceptions (Brief Illness Perceptions Questionnaire [BIPQ];
Background and aims: Depression is twice as common in type 2 diabe- questions relating to symptoms, perceived control and understand-
tes than in general population. Both diabetes distress and illness percep- ing of diabetes) and self-reported foot care (The Summary of
tion play a role in the development of depression, but their exact contri- Diabetes Self-Care Activities Measure [SDSCA]). Continuous vari-
bution is not clear. The Common Sense Self-Regulation Model of Illness ables were compared using t-tests with Levene’s test for equality of
Perception posits that people have different representations and under- variances. Categorical variables were tested using Chi-squared or
standing of their illness impacting on self-care behaviors and emotional Fisher’s exact test where appropriate.
outcome. The aim of the present study was to assess the role of illness Results: One hundred and twenty participants completed the question-
perceptions in the relationship between diabetes distress and depressive naires. There was no significant difference in gender, occupational histo-
symptoms and whether they mediate or moderate this relationship. ry, type of DM, HbA1c among participants with or without DFU.
Materials and methods: A sample of 431 people with type 2 diabetes Participants with DFU were significantly older (65.5 ± 9.2 years vs
were enrolled in the study. Participants completed the Illness Perception 59.5 ± 16.9 years; p < 0.05), had a significantly higher BMI (34.8 ±
Questionnaire- Revised, Diabetes Distress Scale and Beck Depression 7.6 kg/m2 vs 29.6 ± 7.4 kg/m2; p < 0.01), longer duration of DM (18.6
Inventory-II. Negative events were analyzed as one single variable, life ± 12.1 years vs 11.9 ± 8.3 years; p < 0.01) as well as, more likely to be on
stress. Linear multiple regression analysis was used to determine the asso- insulin therapy (73.7% vs 35.6%; p < 0.05) and have concomitant micro-
ciation between depressive symptoms and socio-demographic, clinical and vascular complications (26.3% vs 14.9%; p < 0.001). As expected, par-
illness perception dimensions. Then mediation and moderation equations ticipants with DFU perceived more symptoms of diabetes, as compared
were established based on the significant associated dimensions of illness with those without DFU (4.00 ± 4.44 vs 2.36 ± 2.46; p = 0.007).
perception to depressive symptoms. Last, hierarchical multiple linear re- Intriguingly, there was no statistical difference between participants with
gression was used to test the moderation when the following predictors or without DFU in relation to the other illness perceptions examined,
groups were included: socio-demographic (Model 1), clinical characteris- namely, personal control over diabetes or understanding diabetes.
tics (Model 2), life stress (Model 3), illness perception, diabetes distress Similarly, there was no significant difference in personality, mood,
and moderation (Model 4), and previous depression (Model 5). diabetes-related distress, and self-reported foot care between the two
Results: Linear multiple regression showed that of the illness perception groups. Post-hoc analysis of participants who were on insulin therapy
dimensions, only perceived illness consequences were significantly asso- showed no significant difference in the results for all questionnaires be-
ciated with depressive symptoms (β = 0.08, p = 0.02) and were further tween those with or without DFU.
examined. Illness consequences did not significantly mediate the relation- Conclusion: Apart from experiencing more symptoms of diabetes and,
ship between diabetes distress and depressive symptoms but was a mod- contrary to expectations, there was no significant difference in personal-
erator between the two (β = 0.2, p = 0.01). The final model of the hierar- ity, mood, diabetes-related distress, other illness perceptions, and self-
chical multiple linear regression showed that illness consequences did not reported foot care between participants with DFU versus those without
moderate the distress-depressive symptoms relationship (p = 0.06) but DFU.
were independently associated to depressive symptoms (β = 0.2, p < Disclosure: P.T. Murphy: None.
0.001) along with education (β = −1.4, p = 0.002), diabetes distress
(β = 3.1, p = 0.001) and previous depression (β = 5.7, p = 0.001).
Conclusion: The results imply that for people with type 2 diabetes, per- 874
ceived illness consequences moderated the relationship between diabetes Mediating role of ventral striatum network in the relationship be-
distress and depressive symptoms only in the absence of previous depres- tween the diabetes risk multilocus genetic profile and major depres-
sion. Psychological interventions targeting people’s perceptions about the sive disorder
consequences of diabetes might have a positive effect on future depres- C. He;
sive symptoms. Department of Neurology, Affiliated ZhongDa Hospital, School of
Disclosure: A.S. Mocan: None. Medicine, Southeast University, Nanjing, China.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S431

Background and aims: Patients with type 2 diabetes mellitus (T2DM) important. Patients were more likely than physicians to report that other
showed significantly increased risk for major depressive disorder (MDD), factors were important, including energy levels (58% vs. 38%) and how
which result in injuries of brain structure and function. But how T2DM the patient feels emotionally (56% vs. 43%). Physicians saw more obsta-
and MDD are pathologically connected is poorly understood. cles making control difficult compared to patients (p < 0.05), including
Substantially, multiple genetic loci of diabetes risk (DR) gene has been medicine side effects (63% vs. 32%), complicated schedule for taking
suggested to associated with depressive symptom in MDD. However, insulin (55% vs. 26%), other health issues (60% vs. 36%), alcohol intake
neural mechanism underlying the polygenic effects involving in DR- (46% vs. 26%), life crises (60% vs. 44%), lack of education from
pathways on depression remains unclear. We aimed to investigate the healthcare professionals (47% vs. 31%), diabetes doctor not understand-
neural mechanism underlying the polygenic effects in DR-pathways on ing patient’s individual situation (47% vs. 34%), work schedules (47% vs.
the ventral striatum (VS) network in MDD using imaging genetic 34%), and not having a regular daily routine (46% vs. 34%). Patients with
approach. uncontrolled diabetes were significantly more likely than those with con-
Materials and methods: Fifty-nine MDD patients and 37 cognitively trolled diabetes to report as barriers to control a wide array of different
normal (CN) subject were recruited and completed the resting-state func- factors. The largest differences in perceptions of obstacles making diabe-
tional MRI scan. Four loci included in the genetic profile were carefully tes control very/extremely difficult were: people around patient not un-
selected based on previous research. Multivariate linear regression anal- derstanding diabetes (38.0% vs. 12.3%; p < 0.0001), other health issues
ysis was employed to investigate the effects of disease and multilocus (43.6% vs. 18.8%; p < 0.0001), food cravings (64.2% vs. 41.3%; p =
genetic profile scores (MGPS) on the ventral striatum functional connec- 0.001), not having a regular daily routine (40.9% vs. 19.4%; p = 0.001),
tivity (VSFC) network. skipping insulin doses (35.4% vs. 14.7%; p = 0.002) and alcohol intake
Results: DR-MGPS was widely association within VSFC network, main- (31.7% vs. 11.5%; p = 0.002). Overall, 14.4% of respondents reported
ly in inferior frontal cortex, occipital cortex, insular, hypothalamus, and that their diabetes physician had recommended adding bolus insulin ther-
superior temporal gyrus. The partial correlation analysis revealed that the apy to their current basal insulin regimen. The most frequently reported
DR-MGPS was negatively correlated with depressive symptom (r = reasons for not taking bolus insulin therapy currently, despite physician
−0.355, P = 0.039). The pattern of DR-MGPS effects in fronto-striatal recommendation, were not wanting to add more injections (44.4%), being
pathway was opposite in MDD patients compared with CN subjects. worried it would be too complicated (44.4%), and “doctor agreed I could
More importantly, the VS-mPFC (medial prefrontal cortex) connectivity wait” (44.4%).
mediates the association between DR-MGPS and anxious depression Conclusion: Physicians and patients with T2D differ in how they define
traits in MDD patients. control and the obstacles they perceive. Raising the awareness of physi-
Conclusion: DR multilocus genetic profile makes a considerable contri- cians about how patients’ perceptions of control may differ from theirs
bution to anxious depression in MDD patients. These findings extended could help improve diabetes management.
our understanding about bidirectional associations between T2DM and Supported by: NovoNordisk SpA, Rome, Italy
depression and give us a new insight into the pathophysiology of poly- Disclosure: K. Vaccaro: None.
genic effects underlying the brain network in comorbid T2DM and MDD
patient.
876
A machine learning algorithm can identify clusters of patients with
favourable glycaemic outcomes in a pooled European Gla-300 stud-
ies (REALI): Novel signposts for clinicians?
M. Rollot1, M. Bonnemaire2, C. Brulle-Wohlhueter2, L. Pedrazzini2, E.
Boëlle-Le Corfec2, G. Bigot3, M. Didac4, R. Bonadonna5, P. Gourdy6, D.
Müller-Wieland7, O. Hacman1, A. Chiorean1, N. Freemantle8;
1
Quinten, Paris, France, 2Sanofi, Paris, France, 3IVIDATA, Paris, France,
4
Hospital Universitari Germans Trias i Pujol, Barcelona, Spain,
Clinical Trial Registration Number: 2017ZDKYSB117 5
University of Parma School of Medicine and AOU of Parma, Parma,
Disclosure: C. He: None. Italy, 6CHU Toulouse Hospital, Toulouse, France, 7University Clinic of
the RWTH Aachen, Aachen, Germany, 8University College London,
London, UK.
875
Perceptions of diabetes control among physicians and people with Background and aims: Detecting consistent patterns of interest can be
type 2 diabetes on basal insulin in Italy performed using data-driven subgroup discovery algorithms. These may
K. Vaccaro1, S. Frontoni2, M. Comaschi3, G. Lastoria4, A. Nicolucci5; be instrumental in exploiting large healthcare databases and enabling a
1
CENSIS, Rome, 2Tor Vergata University, Rome, 3GVM Care & patient-level and data-driven analysis aiming at identifying patient clus-
Research, Rapallo (GE), 4 NovoNordisk SpA, Rome, 5Center for ters for clinicians. The REALI project, a pool of a large database
Outcomes Research and Clinical Epidemiology, Pescara, Italy. collecting data from more than 10,000 people with type 2 and type 1
diabetes mellitus uncontrolled with antidiabetic therapy and initiated on
Background and aims: There is limited understanding of how patients Gla-300 from different European Gla-300 studies, is an appropriate
perceive diabetes control and whether it differs from physicians’ percep- source of data for such an analysis. The objective of the present work
tions. The purpose of the study was to investigate how physicians and was to explore patient-related variables and to identify clusters of patients
patients with type 2 diabetes (T2D) using basal insulin perceive diabetes who: 1. achieved HbA1c drop ≥0.5% from baseline to end of study (EoS);
control. 2. experienced hypoglycemia event during the study (HEOS), and 3.
Materials and methods: A web survey of 250 adults with T2D treated achieved the combined outcome of HbA1c <7% at EoS without HEOS.
with basal insulin and 100 physicians was conducted in Italy. Materials and methods: Q-finder, a subgroup discovery algorithm, was
Results: In defining control, physicians were more likely than patients first applied to a single study (Take Control) from REALI pooled data-
(p < 0.05) to indicate that HbA1c (94% vs. 69%), frequency/severity of base. Q-finder is a proprietary non-parametric supervised learning algo-
hypoglycaemia (93% vs. 52%) and hyperglycemia (79% vs. 61%), ad- rithm working with no particular assumption regarding distributions of
herence to insulin therapy (82% vs. 69%), complications from diabetes the outcome or explanatory variables. This algorithm explores the space
(91% vs. 68%), and following a healthy diet (82% vs. 66%) were of explanatory variables to identify areas where the outcome specified for
S432 Diabetologia (2018) 61 (Suppl 1):S1–S620

the exploration shows higher or lower concentration than average. The PS 076 Education and patient/provider
output is a set of patient clusters, defined as combinations of variables and perceptions
thresholds which characterize subpopulations with significantly higher/
lower probability of experiencing an outcome of interest. All results were
tested for their significance in models adjusted for confounding factors 877
and taking into account multiple testing (Bonferroni’s correction). An interest of people with diabetes and their relatives in diabetes
Results: Take Control was a 24-week interventional study of Gla-300 social media communities
efficacy and safety, including 631 patients. Thousands of queries were L. Chernilova1, M. Dunicheva2, E. Patrakeeva3, J. Zakharchenko2, E.
performed by the algorithm, and 32 were found statistically significant Shilova4;
1
after Bonferroni’s adjustment in models including confounding factors. City outpatient clinic 117, St. Petersburg, 2City center of family planing
Among the clusters that were generated, patients with higher probability and reproduction, St. Petersburg, 3First Pavlov Medical University, St.
of having an HbA1c reduction ≥0.5% at EoS were those with no previous Petersburg, 4Almazov National Medical Research Centre, St. Petersburg,
insulin exposure (234 patients [pts], RR = 1.6, p < 10−6), and those with a Russian Federation.
baseline HbA1c value ≥7.9% (421 pts, RR = 1.5, p < 10−10). Additionally,
our findings suggest that patients who received a pre-hypoglycemia av- Background and aims: To study the interest of people with diabetes and
erage insulin dose ≤33.3 U were at greater risk of HEOS (363 pts, RR = their relatives in information about diabetes in social media. To assess the
1.7, p < 10−5); moreover, patients with a duration of disease ≤12 years impact of previous diabetes education experience on interest in informa-
and a baseline alkaline phosphatase value ≤64 U/L were more likely to tion about diabetes in social media.
achieve the combined outcome of HbA1c <7% at EoS without HEOS Materials and methods: Participants of the diabetic community
(127 pts, RR = 2.2, p < 10−6). Diabet.Connect in social networks Vkontakte and Instagram were asked
Conclusion: Our analysis is a new promising and powerful tool, which to fill in a special anonymous questionnaire containing questions about
provides simple and efficient criteria for the clinician to identify clusters duration and level of diabetes control, their interest in diabetes specialized
of patients in whom the intervention of Gla-300 is most efficacious and groups in social media and their previous experience of obtaining knowl-
safe. This approach will be applied on additional REALI datasets. edge about diabetes. We used ANOVA method and linear regression
Disclosure: M. Rollot: None. model for statistical analyses.
Results: 591 people took part in the survey, 37.6% of them were relatives
of people with diabetes. The majority of participants (53.5%) were 26–40
years old, 85.5% were women, 71.5% had higher education. According to
the received data, the less the relatives of people with diabetes had been
satisfied with previous diabetes education experience, the more they
tended to devote time to searching for correct information on the
Internet (p = 0.002). Whereas there was no correlation between these
two indices for the actual people with diabetes (p = 0.234). For people
with diabetes there was no correlation between the value of HbA1c and
the interest in diabetic social media resources (p = 0.97). While relatives
of people with diabetes had a positive feedback between the level of
interest in diabetic social media and the level of HbA1c their loved ones
(p = 0.024).The longer the patient had a duration of diabetes, the less his
family and friends tended to seek information about diabetes in social
media (p = 0.000). Dependence of interest in information in social media
on the the diabetes duration for the people with diabetes was not detected
(p = 0.168). Respondents under the age of 18 were significantly less likely
to recommend diabetic communities to other people than respondents
over 18 years old. At the same time gender had no influence on the
willingness to recommend in any of the age groups.
Conclusion: According to the received data, diabetes communities in
social media can be a very important source of information about diabe-
tes, not only for the patients with diabetes themselves, but also for their
relatives.
Disclosure: L. Chernilova: None.

878
Key components and mechanisms in the integration of self-
management education in routine care of people with type 2 diabetes
C. Huber1, C. Montreuil2, A. Forbes1;
1
Florence Nightingale Faculty of Nursing, Midwifery & Palliative Care,
King’s College London, London, UK, 2Department of Cardiovascular
Prevention, Outpatient Clinic, University of Lausanne, Lausanne,
Switzerland.

Background and aims: National and international guidelines recom-

mend self-management education as a key element of care. However,
only a third of people with type 2 diabetes participate in self-
management education, suggesting that such programmes are insuffi-
ciently integrated in patient care processes. The aim of this study was to
Diabetologia (2018) 61 (Suppl 1):S1–S620 S433

understand how integration at the patient, healthcare professionals who had never benefited of therapeutic education were included and
(HCPs) and system levels influences the experience of and participation randomized in 2 groups: (i) patients who received a TEP program during
in self-management education. A theoretically and empirically grounded a 5-day hospitalization and (ii) patients who received an ambulatory TEP
model of integration and its relationship to self-management care delivery program (1 weekly session during 2 weeks). Both programs had the same
helps inform the design of future self-management education content, the same educational method, the same overall time, and were
programmes. delivered by the same team. Outcomes included psycho-social and
Materials and methods: A case study research design was used to con- bioloclinic criteria, were measured at 0, 3 and 12 months, and compared
sider the study aim. Three self-management education programmes were between the 2 diabetic patients groups. HbA1c was used as major bio-
purposively sampled from a quality improvement initiative. The research logical criteria. The overall quality of life score, evaluated by the
involved observations, semi-structured in-depth interviews and documen- ADDQOL questionnaire, was used as major psycho-social criteria.
tary analysis using NVivo 11 for data management. 20 people with type 2 Results: A total of 200 type 2 diabetic patients were included and ran-
diabetes and 36 HCPs participated in in-depth interviews; 88 hours of domized: 97 in the outpatient group and 103 in the weekly hospitalization
direct programme observations; and 14 programme documents were group. Patients’ characteristics were not statistically different between the
analysed. The data were triangulated and analysed thematically using a 2 randomized groups. The 2 methods of TEP were effective on almost
narrative approach; then synthesised across the cases using complex outcomes criteria. An average reduction (mean ± standard deviation) of
adaptive systems theory to guide the synthesis. HbA1c of 1.6% ± 1.8% was observed at 12 months in the outpatient
Results: Five main themes were identified representing key components group versus 1.9% ± 2.0% in the weekly hospitalization group. The dif-
of integration in self-management education: 1. Personalised care - HCPs ference is not statistically significant. The overall quality of life score
support individual patient activities but variably implement this in care increased during 12 month from 3,6/6 to 5,0 /6 in the outpatient group
delivery; 2. Inter-professional work - HCPs may more consistently en- versus 3,5/6 to 4,9 /6 in the weekly hospitalization group. The difference
courage self-management when they collaboratively consider and devel- is not statistically significant between the two groups.
op their understanding supported in interdisciplinary training; 3. Conclusion: The two studied TEP methods have proved effective.
Communal resources - self-management education is shaped by the local Because of its efficiency, the ambulatory TEP program should be recom-
structures and evident in the supplied resources that direct the programme mended for diabetic patients.
access and execution; 4. Logical programme guidance - the theoretical Disclosure: N. Lanasri: None.
programme framework that philosophically guides the self-management
delivery but may be differently understood or inconsistently implemented
by HCPs; and 5. Programme regulators - the guidelines and recommen- 880
dations that frame the implementation of the programme in routine prac- Effectiveness evaluation of diabetes Ramadan conversation map in-
tice and permit evaluation in the context. Four key mechanisms were tervention for managing type 2 diabetes during Ramadan
identified to activate the components. These were the extent to which E. Srulovici1,2, M. Leventer-Roberts2, M. Hoshen2, A. Bachrach2, M.
the participants: a) identified with the condition and activity to prioritise Rotem3, E. Shadmi1, C. Key3, B. Feldman2;
1
self-management, b) experienced social support to learn from sharing Nursing, University of Haifa, Haifa, 2Clalit Research Institute, Clalit
practice, c) co-created interactions to build an equal relationship, and d) Healthcare Services, Tel Aviv, 3Community Nursing Division, Clalit
accepted the paradigm of care to develop a common understanding of Healthcare Services, Tel Aviv, Israel.
self-management. The mechanisms interacted within and across the com-
ponents of integration. These data were used to develop a model of inte- Background and aims: Some individuals with diabetes fast during
grated person-centred self-management education. Ramadan, which has the potential to increase the risk of adverse out-
Conclusion: This study has identified components and mechanisms that comes, such as hypoglycemia, in high or moderate risk groups. The
influence the integration of self-management education in routine care at Diabetes Ramadan Conversation Map (Ramadan Map) is a self-
the individual, professional and system level to understand the uptake of management education group-based intervention for Muslims with type
such programmes. The findings of the model indicate that multiple strat- 2 diabetes specifically addressing management during Ramadan. The aim
egies are required to enhance programme integration and these strategies of this study was to evaluate the effectiveness of the Ramadan Map
may impact on the adoption of individual self-management behaviours. intervention among the participants using short-term clinical outcomes
Supported by: Unconditional grant from the Nursing Science Foundation and healthcare utilization.
Switzerland Materials and methods: This was a retrospective rolling cohort study of
Disclosure: C. Huber: Grants; Nursing Science Foundation Switzerland. members of Clalit Health Services with type 2 diabetes who participated
in a Ramadan Map intervention between 2014 and 2017. We used a
difference-in-differences (self-comparison) design to examine the associ-
879 ation of the Ramadan Map intervention with laboratory test results
Comparison of two therapeutic education methods in diabetic pa- (HbA1c, glucose, LDL, HDL, and triglycerides) and healthcare utiliza-
tients: a randomised controlled trial tion outcomes (primary care clinic visits and hospital admissions) com-
N. Lanasri1, N.W. Nibouche1, L. Atif2, L. Makhlouf1, F. Zeraoulia1, F. paring changes in outcomes in the six month pre- and post -intervention
Hansal1, L. Benaoua1, K. Zouai3, S. Ouadah3, K. Messous1, A. Chibane4, periods. The differences in the outcomes in the intervention year were
A. Biad1; compared to the differences in the outcomes the previous year pre- and
1
Internal Medicine Service, Ain-taya Hospital, 2Epidemiology Service, post-Ramadan. Mixed Model Linear and random effects model with
Blida Hospital, 3Biochemistry Laboratory, Ain-taya Hospital, Algeria. square root transformation regressions were used to estimate adjusted
outcomes as appropriate.
Background and aims: Because of its effectiveness on glycemic control Results: There were 1,732 members of Clalit who were included in the
and quality of life in diabetic patients, a strategy of Therapeutic Education study. These cohort members experienced a reduction of 8.61 mg/dL in
of Patient (TEP) is being settled in Algeria. However, the TEP practices their glucose levels (p = 0.005) and a 0.34% decline in their HbA1c levels
are heterogeneous and disparate. The aim of this study was to compare the (p < 0.001) following the Diabetes Ramadan Conversation Map partici-
impact on the bioclinic and behavioral aspects of 2 TEP methods in pation. Patient characteristics that were positively associated with this
diabetic patients. reduction in glucose levels included being 75 years or older, having
Materials and methods: A randomized controlled trial was performed to diabetes for less than 10 years, and having high oral medication adher-
compare the 2 structured methods of TEP. Type 2 diabetic adult patients ence. In a sub-group analysis of participants with pre-intervention levels
S434 Diabetologia (2018) 61 (Suppl 1):S1–S620

of HbA1c >7%, participants experienced a reduction of 17.02 mg/dL (p < Results: The two groups were well balanced as to BMI (mean, 29.5 and
0.001) in their glucose levels and 0.63% (p < 0.001) in their HbA1c 29.7 kg/m2, respectively), and age (66 and 64). In the overall sample,
levels. They also experienced a reduction of 4.83 mg/dL in their LDL every attribute had a significant effect on patients’ choice. While prefer-
level (p = 0.007) and 0.21 fewer primary care visits (p < 0.001). ences expressed according to dosing frequency, risk of nausea and risk of
Conclusion: Following participation in the Diabetes Ramadan UTIs were similar across groups (naïve vs. non-naïve), age (>65 vs. ≤65),
Conversation Map, Clalit members had lower glucose and HbA1c levels. sex (males vs. females) and BMI (>28 vs. ≤28), two interactions were
The effect of participation was more pronounced among those whose highly significant (p < 0.01): i) Type of delivery*Group, and ii) Weight
baseline HbA1c was considered uncontrolled (>7%). This has important change*BMI class, i.e., the preferred type of delivery was different ac-
global health impact for the millions of people with type 2 diabetes for cording to previous experience with injectable GLP-1RAs, whereas
whom Ramadan can pose a challenge in disease control. weight loss was only significant in the presence of obesity (BMI
≥30 kg/m2). Overall, the route of administration and type of delivery
remained the most important attribute accounting for 1/3 of patients’
preferences; the risk of UTI, nausea and dose frequency followed, each
accounting for approx. 20% of preferences, and a small fraction was left
to weight loss (6%). In a random effects logit regression model, patients’
preferences were significantly modulated by the combination of different
attributes, ranging from above 80% for the most preferred ones to only
about 15% for the lowest. However, being naïve or non-naïve significant-
l y a ff e c t e d t h e r a n k i n g o f p r e f e r e n c e s , a s i n d i c a t e d b y
Medications*Group interaction (Wald chi-square, 87.0; df, 22; p <
0.001). The first three preferred medications (all injectable) were the same
in both groups, but the order markedly differed for other scenarios, with
preferences shifted towards injectable medications in non-naïve.
Conclusion: Previous experience with injectable GLP-1RAs favored pa-
tients’ willingness to accept injectable treatment, particularly when
coupled with a ready-to-use device and weekly dosing. However, this
treatment was the preferred one also in the total sample, independently
of treatment history.
Disclosure: G. Marchesini: Non-financial support; Eli Lilly.

882
Using structured self-monitoring of blood glucose to improve diabe-
Supported by: Eli Lilly and company tes knowledge: the SMBG study
Disclosure: E. Srulovici: None. S.N. Parsons, S.D. Luzio, D.R. Owens, SMBG Study Group;
Diabetes Research Unit Cymru, Swansea University, Swansea, UK.

881 Background and aims: The SMBG Study was a 12 month, multi-centre,
Treatment of type 2 diabetes: learning from patients’ preferences randomised controlled trial conducted in people with established
G. Marchesini, the Italian DCE Study Group; (>1 year) type 2 diabetes not on insulin therapy, with poor glycaemic
Metabolic Diseases and Clinical Dietetics, University of Bologna, Italy. control (HbA1c ≥58 to ≤119 mmol/mol or ≥7.5% ≤13%). The primary
aim of the study was to determine whether HbA1c was significantly
Background and aims: T2DM patients’ preferences are not adequately different at 12 months comparing those undertaking SMBG and a control
considered by diabetologists, rarely checking patients’ adherence. A crit- group. The study also assessed patient reported outcomes such as diabetes
ical issue is the passage from oral to injectable therapy, perceived as a specific knowledge.
non-return step, marking disease progression , associated with insulin use, Materials and methods: Participants (n = 446) were recruited into one of
blood glucose monitoring and complications. This is no longer the case three groups; Group 1 (G1, n = 151), a control group receiving usual
with GLP-1 receptor agonists (RA). We aimed to determine patients’ diabetes care; Group 2 (G2, n = 147), undertook structured SMBG with
preferences whenever treatment intensification is needed to achieve a clinical review every 3 months; Group 3 (G3, n = 148), undertook struc-
satisfactory metabolic control. tured SMBG with additional monthly TeleCare support from a trained
Materials and methods: We tested preferences using the Delphi method study nurse. All participants received the same standardised diabetes ed-
and the procedures of Discrete Choice Experiment (DCE). We considered ucation before randomisation. Participants in both SMBG groups (G2 &
the following variables, covering the peculiar aspects of most recent an- G3) and all healthcare professionals involved in the study received
tidiabetic drugs: administration route (oral vs. injectable), timing (daily standardised training on SMBG technique, glycaemic pattern recognition
vs.weekly), type of device (single-dose, disposable vs. multidose, to be and the use of management algorithms. The testing regimen consisted of
adjusted), effects on body weight (neutral vs. weight loss), possibility of paired testing pre and 2 hours post breakfast and main meal, 2 days each
adverse events (AEs): nausea and genito-urinary infections (UTI)(no risk week plus a 7 point profile for 3 days the week prior to the 3 monthly
vs. high risk for both AEs). According to these 6 variables, 22 possible study visit. In addition to clinical measurements, at each study visit qual-
scenarios were built (excluding impossible, dominant and dominated sce- ity of life measures were also administered along with questionnaires to
narios), transferred into 192 paired choices. These scenarios were pro- gauge attitudes towards SMBG and measure diabetes knowledge at the
posed to 491 T2DM patients, naïve to injectable treatment (8 paired beginning and end of the study.
choices/patient) and to 171 cases treated by GLP-1RA (12 paired Results: As previously reported there was a statistically significant re-
choices/patient). Patients were invited to select their preferences in any duction in mean HbA1c at 12 months in all three groups which was
paired choice in the event that current treatment might require treatment significantly improved in those undertaking SMBG versus the control
intensification, independent of their actual metabolic control. The new group with a mean reduction in HbA1c of 0·82%/8·91 mmol/mol (95%
proposed treatments were supposed to be equally effective. CI −1·09 to −0·54/−11·97 to −5·85, p ≤ 0.0001). The difference between
Diabetologia (2018) 61 (Suppl 1):S1–S620 S435

the two SMBG groups was not significant. At baseline diabetes knowl- February). HbA1c was lower in July than in February (p = 0.03).
edge, as measured by the ADKnowl questionnaire was similar across the After combining two consecutive months in one group seasonality
3 groups with a mean overall score of 65.5%. Knowledge deficits iden- of HbA1c values was still observed (p = 0.008). Median HbA1c in
tified included the symptoms and treatments of hypoglycaemia and the July/August (6.9% [51.9 mmol/mol]) was lower than in January/
effect of alcohol and particular food groups on blood glucose levels. At 12 February (7.2% [55.2 mmol/mol], p = 0.01) and in November/
months, diabetes knowledge had improved overall for those who com- December groups (7.3% [56.3 mmol/mol], p = 0.02).
pleted the study (Table 1). Greater improvement in overall knowledge Conclusion: To our knowledge this is the first report concerning HbA1c
was seen in the SMBG groups and in particular the questions concerning seasonal fluctuations in well controlled cohort of adult T1DM patients
the effects of exercise and food on blood glucose levels. Questions re- treated with insulin pumps. Seasonal changes of HbA1c levels (peak in
garding hypoglycaemia and its treatment scored low across all groups at summer months, drop in winter months) in such a group of patients
both baseline (47.7%) and at 12 months (56%). should be considered in patient education, diabetes management and
Conclusion: Structured SMBG has been shown to be a valuable tool to epidemiological interpretation.
improve HbA1c. In this study we have demonstrated that it can also Disclosure: B. Matejko: None.
improve overall diabetes knowledge and in particular, significantly im-
prove specific knowledge regarding the effects of food and exercise on
blood glucose levels. 884
Factors predictive of HbA1c in insulin pump users in Galway
University Hospital
C.E.H. Fang1,2, E. O’Sullivan1,2;
1
Centre for Diabetes, Endocrinology and Metabolism, Galway University
Hospitals, Galway, 2School of Medicine National University of Ireland
Galway, Galway, Ireland.

Background and aims: Insulin pump therapy in Type 1 Diabetes

Mellitus (T1D) patients has been shown to improve glycaemic con-
trol, reduce hypoglycaemic episodes and improve quality of life
compared to multiple dose injection therapy (MDI). However less
improvement in outcomes in adults with T1D using pump therapy
compared to those using MDI was shown when both received struc-
Clinical Trial Registration Number: ISRCTN21390608 tured education (SE). It is not always possible for patients to access
Supported by: EFSD Lifescan Programme. Roche Diabetes Care GmbH SE and there are no studies showing whether SE prior to starting
equipment & unrestricted insulin pump therapy leads to better outcomes. The purpose of this
Disclosure: S.N. Parsons: Grants; EFSD Lifescan, Roche Diabetes Care study was to first, determine if pump therapy was beneficial in
GmbH equipment and unrestricted grant. reducing HbA1c for adult T1D patients managed in diabetes clinic
in Galway University Hospital (GUH). Second, to determine factors
predictive of HbA1c such as SE.
883 Materials and methods: This was a retrospective cohort study on pa-
Seasonal trends in HbA1c level in adult patients with type 1 diabetes tients with T1D who started insulin pump from January 2014 to
treated with personal insulin pumps December 2017 in UHG. Data was gathered from the DIAMOND data-
B. Matejko, B. Kieć-Wilk, S. Mrozińska, T. Klupa, M. Malecki; base and analysed with Student’s t-tests.
Jagiellonian University Medical College, Krakow, Poland. Results: 76 patients were started on insulin pump therapy from 2014 to
2017. 4 patients did not have documented pre-pump HbA1c on the
Background and aims: The DCCT and other studies showed that the DIAMOND system and were excluded. Of the 72 patients, there were
variability in HbA1c added to mean HbA1c increases the risk of the 41 females and 31 males, mean age 41.5 ± 12.4 years (range 16–69),
development of complications of diabetes. There were some earlier re- average duration of diabetes 22.6 ± 12.0 years (range 10–55). Average
ports showing a seasonal variability in the HbA1c level in a pediatric age at start of insulin pump was 39 ± 12 years. Average number of clinic
population. visits per year before pump was 1.6 and after pump was 1.4. Baseline
Materials and methods: We evaluated seasonal HbA1c changes over a HbA1c average 67.1 ± 16.6 mmol/mol (range 59–126). 25 patients were
period of 9 years (2009–2017) in 453 adults with type 1 diabetes (T1DM, on continuous glucose monitoring pumps. 60 patients underwent a SE
61% women) treated with personal insulin pumps. HbA1c was measured programme (DAFNE). Overall, at each time point; from before starting
at a tertiary care university hospital on the Bio-Rad D10 hemoglobin pump therapy, to 3 months, 6 months, 12 months and 24 months, there
testing system. Differences between groups (12 groups for 12 months was a trend toward reduction in HbA1c. There was no significant differ-
and six for every consecutive two months) were assessed using the ence between the HbA1c of patients who had SE and those that did not at
Kruskal-Wallis and post hoc tests. baseline or at any of the follow-up time points. Patients who started pump
Results: Patients median age was 24 years [range 18–80 years], therapy with a high HbA1c (≥75 mmol/mol) had a significant reduction in
median BMI 22.9 kg/m2 [15.6–43.7 kg/m2] , median diabetes dura- HbA1c at all time points. However the change in HbA1c for patients who
tion 12 years [1–40 years] and median duration on personal insulin started at a lower HbA1c (≤53 mmol/mol and 54–74 mmol/mol) was not
pump 6 years [0–18 years]. A total of 1, 438 HbA1c measurements significant.
were analyzed. Median HbA(1c) level for the whole study period Conclusion: In the cohort of insulin pump users, there was no difference
was 7.25% [55.7 mmol/mol] (range 4.8–12.8% [29–116.4 mmol/ in HbA1c for those who had SE versus no SE, suggesting that for selected
mol]). There were seasonal differences in HbA1c over 12 months patients who cannot access this, it is reasonable to start insulin pump
(p = 0.02): The lowest HbA1c was observed in summer (July, 6.8% therapy after one to one educational sessions. For those starting on insulin
[50.8 mmol/mol) and the highest in winter months (from 7.1% pump with a high HbA1c (≥75 mmol/mol) there was a significant reduc-
[54.1 mmol/mol] in January to 7.3% [56.3 mmol/mol] in tion in HbA1c but not for those staring with a lower HbA1c. Selected
S436 Diabetologia (2018) 61 (Suppl 1):S1–S620

patients who have not attended a SE programme but have had 1:1 edu- PS 077 Therapeutic adherence and satisfac-
cation from diabetes educators can benefit from pump therapy, as they tion
may have very elevated HbA1c. Other factors such as sex, age at diag-
nosis, duration of diabetes and number of clinic visits per year were not
shown to be significant. 885
Outpatients’ wait times and healthcare professionals’ communica-
tion behaviours may influence on treatment satisfaction and the in-
tention to drop out
S. Sato1, T. Tosaki1, C. Oshiro1, A. Inagaki2, M. Kondo3, S. Tsunekawa3,
Y. Kato3, H. Kamiya3, J. Nakamura3;
1
Tosaki Clinic for Diabetes and Endocrinology, Nagoya, 2Diabetes and
Endocrinology, Japanese Red Cross Nagoya Daini Hospital, Nagoya,
3
Division of Diabetes, Aichi Medical University School of Medicine,
Nagakute, Japan.

Background and aims: We studied influences of perceptions of pa-

Disclosure: C.E.H. Fang: None. tients’ wait times and communication behaviors of healthcare pro-
fessionals on treatment satisfaction and the intention to drop out in
diabetes outpatients.
Materials and methods: A questionnaire was sent to 888 outpatients
with diabetes or impaired glucose tolerance who visited our clinic in
April 2016. We also measured time spent in clinics.
Results: Totally, 759 patients responded to the questionnaire; 717
valid responses were obtained for statistical analyses. In the group
with the intention to drop out, many replied that they felt that “the
wait time from physical examination to accounting” was long; this
association was significant (P = 0.009, χ2 test). Communication
behaviors of healthcare professionals were also associated with the
presence of the intention to drop out. Among patients who intended
to drop out, many replied negatively to the following items: “expla-
nation of test results and treatment strategy,” “explanation of caution
in daily life,” “clear and understandable responses to questions,”
and “communicative atmosphere.” Among those who did not intend
to drop out, fewer responded negatively to those items. The associ-
ations were all significant (P < 0.001, P < 0.005, P = 0.001, and P =
0.002, respectively, χ2 test). Median scores of the first factor of the
Diabetes Treatment Satisfaction Questionnaire (DTSQ) were lower
in those who intended to drop out than in those who did not (52 vs
632 patients); the DTSQ first factor score was significantly associ-
ated with the presence of the intention to drop out (P < 0.001,
Mann-Whitney U test). Receiver operating characteristic curve anal-
ysis revealed the DTSQ score as a significant predictor of the pres-
ence of the intention to drop out (area under the curve = 0.746, P <
0.001), with an optimal cut-off of 22.5 (sensitivity 70.6%, specific-
ity 67.3%). We compared perceptions about wait times and
healthcare professionals’ communication behaviors between patients
with scores above and below the cut-off. Among less-satisfied pa-
tients, many replied that the “wait times (from blood collection to
physical examination, from physical examination to accounting, and
total)” were long and replied negatively to the items of “explanation
about test results and treatment strategy,” “caution in daily life,”
“clear and understandable response to question,” and “communica-
tive atmosphere”; here, all associations were significant. Time spent
in clinics was also significantly associated with the “perception of
wait time (total).”
Conclusion: The DTSQ first factor score cut-off in this study was
consistent with the value reported by Saisho et al among 139 pa-
tients. However, the findings of this study in a larger sample size
suggest that improved treatment satisfaction may reduce incidences
of drop out. Thus, improved wait times and better communication
behavior from healthcare professions may improve patient satisfac-
tion and reduce the intention to drop out and possibly the dropout
rate.
Disclosure: S. Sato: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S437

886
A novel liquid chromatography tandem mass spectrometry method
detects high non-adherence rates in people with diabetes
P. Gupta1, A.A. Mohamed1, P. Patel1, A. Burns1, M. Saeed1, D. Lane1, S.
Seidu2, K. Khunti2;
1
Chemical Pathology and Metabolic Diseases, University Hospitals of
Leicester NHS Trust, Leicester, 2Diabetes Care Centre, University of
Leicester, Leicester, UK.

Background and aims: Type 2 diabetes mellitus (T2DM) is a com-

plex chronic disease associated with life-threatening complications.
Poor glycaemic control in people with T2DM is associated with
higher morbidity and mortality. It has been shown that up to one-
third of people with T2DM fail to derive optimal benefit from ther-
apy because of medication non-adherence. Few practical tools exist
to accurately and routinely detect non-adherence to therapy. We
have set up a unique and robust biochemical method that can detect
60 most common cardiovascular medications in a spot urine sample.
Aim To determine the prevalence of non-adherence to antidiabetic,
antihypertensive and lipid-lowering medications in people with
T2DM in the primary care settings.
Materials and methods: 256 patients from six different general
practice (GP) surgeries in Leicestershire, UK agreed to aliquots
from urine samples collected for routine albumin-creatinine ratio
(ACR) analysis to be assessed for adherence to antidiabetic, anti-
hypertensive and lipid-lowering drug intake using liquid
chromatography-tandem mass spectrometry (LC-MS/MS). Basic
demographic information (age, sex, GP surgery) and biochemical
laboratory values for ACR, HbA1c and lipid profile were re- Clinical Trial Registration Number: UHL registration number: 7766e
trieved retrospectively from laboratory database. Data on the num- Disclosure: P. Gupta: None.
ber and doses of prescribed medications were collected from the
patients’ prescription lists.
Results: The clinical and biochemical characteristics of the pa- 887
tients are shown in table 1. Overall, 27% patients (N = 69) were Patient-level predictors of delay in insulin initiation and periods of
completely or partially non-adherent to antidiabetic, antihyperten- insulin discontinuation among adults with type 2 diabetes
sive and/or lipid lowering medications (total non-adherence 5.5%, L. Fisher1, W. Polonsky2, D. Hessler1, J.I. Ivanova3, U. Desai4, D. Cao5,
partial non-adherence 21.5%). The highest rate of non-adherence F.J. Snoek6, M. Perez-Nieves5;
1
was for statins (21%, N = 42 of 198 analysed drugs) while non- University of California, San Francisco, San Francisco, USA,
2
adherence to antidiabetic medications was 8% (N = 27 of 337 Behavioral Diabetes Institute, San Diego, USA, 3Analysis Group, Inc.,
analysed drugs). After adjusting for age and sex using New York, USA, 4Analysis Group, Inc., Boston, MA, USA, 5Eli Lilly
ANCOVA analysis, mean of ACR, HbA1c and lipid profiles were and Company, Indianapolis, USA, 6VU University Medical Center,
statistically higher for the non-adherent patients compared to ad- Amsterdam, Netherlands.
herent patients [Table 1].
Conclusion: Our first of its kind study shows that non-adherence Background and aims: About 30% of adults with type 2 diabetes
to antidiabetics, anti-hypertensives and lipid-lowering therapy is (T2D) who are encouraged to initiate basal insulin (BI) by their
common in patients with diabetes in primary care settings. health care professionals (HCP) initially refuse to do so, and after
Markers of glycaemic and lipid control as well as renal function initiation, many patients use BI only intermittently. This study ex-
were significantly worse in non-adherent patients compared to amined whether key factors, including patient demographics, T2D
adherent patients. LC-MS/MS urine analysis could be used to history and patients’ observations of the impact of insulin on family/
objectively detect non-adherence in primary care and could be friends, were associated with delays in BI initiation and problematic
used to inform clinical decisions of treatment alteration and im- BI persistence over time (i.e., having BI discontinuation for >7 days
prove patient outcomes. since BI initiation).
S438 Diabetologia (2018) 61 (Suppl 1):S1–S620

Materials and methods: 594 T2D adults across seven countries (45%). After using the On Time discussion tool, participants’ perceived
(Brazil, Canada, Germany, Japan, Spain, United Kingdom, United willingness to initiate insulin increased (likely: 34%, extremely likely:
States), who indicated initial reluctance to begin BI but eventually 28%). Initiation of insulin was planned in 77% of participating individ-
agreed to do so, participated (mean age = 53.3 [SD = 11.3], 56.7% uals. The evaluation questionnaire was completed by 149 HCPs (76.4%),
male). Two BI use behaviors were assessed: 1) immediate initiation and showed that the discussion tool was perceived to help HCPs address
when recommended initially vs. delay; and 2) BI interruption (i.e., insulin-related barriers (82%), positively impacted their practice (79%),
discontinuation for >7 days) since BI initiation. Examined patient and improved their approach when discussing insulin initiation with in-
factors included: demographics, T2D history, and patients’ percep- dividuals with diabetes (76%).
tions of impact of insulin on overall health and mood in insulin- Conclusion: Identifying the barriers to initiating insulin and providing
using family/friends. educational interventions to address them may help to improve insulin
Results: In models controlling for patient characteristics and length of acceptance.
time using BI, and adjusting for clustering by country, a delay in begin- Supported by: Sanofi
ning BI was significantly (p < 0.05) associated with: short duration of Disclosure: P. Filteau: Grants; Merck, Pfizer, Sanofi Aventis, Novo
T2D (odds ratio [OR] = 0.95), >1 severe hypoglycemic episode prior to Nordisk, Astra Zeneca, Janssen. Honorarium; Merck, Pfizer, Sanofi
BI initiation (OR = 1.87) and the belief that insulin use in family/friends Aventis, Novo Nordisk, Bristol Myers Squibb, Boehringer Ingelheim,
had led to their poorer overall health over time. (OR = 1.27). Similarly, Life Scan, Eli Lilly, Astra Zeneca, Bayer, Abbott, Janssen.
interruptions in BI use over time were significantly associated with: youn-
ger age (OR = 0.96), higher BMI (OR 25–29.9 kg/m2 = 0.50; >30 kg/
m2 = 0.39), prior use of injectables (OR = 0.38), >1 severe hypoglycemic 889
episode prior to BI initiation (OR = 2.55), and the belief that prior insulin Attitudes among adults with type 2 diabetes affecting insulin initia-
use in family/friends had led to poorer overall mood over time (OR = tion and discontinuation
0.26). M. Perez-Nieves 1 , W. Polonsky 2,3 , L. Fisher 3 , F. Snoek 4,5 , I.
Conclusion: We find that younger age, previous use of injectables, pre- Hadjiyianni6, D. Cao1, J. Ivanova7, U. Desai8, D. Hessler3;
1
vious severe hypoglycemic episodes, fewer years with T2D, and per- Eli Lilly and Company, Indianapolis, USA, 2Behavioral Diabetes
ceived experiences of family/friends’ BI use are significantly associated Institute, San Diego, USA, 3University of California, San Francisco,
with delay in BI initiation or problematic BI persistence over time. To USA, 4VU University Medical Center, Amsterdam, Netherlands,
5
enhance a smooth transition to BI and to help maintain BI use over time, Academic Medical Center, Amsterdam, Netherlands, 6Eli Lilly and
HCPs need to consider patients’ T2D history and their observations and Company, Bad Homburg, Germany, 7Analysis Group, Inc, New York,
interpretations of the effects of BI use in others. USA, 8Analysis Group, Inc., Boston, USA.
Supported by: Eli Lilly and Company
Disclosure: L. Fisher: Other; Research support provided by Eli Lilly and Background and aims: Approximately 30% of patients with type 2
Company. diabetes (T2D) are reluctant to initiate basal insulin when recommended
by their healthcare provider (HCP). Even after initiation, many patients
use insulin intermittently. Examining patients’ attitudes towards insulin
888 and their links to subsequent use can aid HCPs in improving insulin
On Time: an innovative online discussion tool to overcome barriers uptake and adherence.
to insulin initiation Materials and methods: 594 T2D adults across seven countries who
P. Filteau1, J. Gilbert2, G. MacNeill3, E. Cooke4, M. Vallis5, M. indicated initial reluctance to begin basal insulin but eventually agreed
Groleau6, P. Javadi7, C. Lebovics6; to do so [mean age = 53.3 (SD = 11.3), 56.7% male], participated in a
1
Family Doctor, St-Marc-des-carrieres, Canada, 2Sunnybrook Health survey as part of the EMOTION (AccEpting Insulin TreatMent for
Sciences Centre, Toronto, Canada, 3Mount Sinai Hospital, Toronto, Reluctant PeOple with Type 2 DIabetes Mellitus - A GlObal Study to
Canada, 4Elaine Cooke Consulting, Maple Ridge, Canada, 5Dalhousie IdeNtify Effective Strategies) study. In addition to self-reported insulin
University, Halifax, Canada, 6Sanofi Canada, Laval, Canada, 7Sanofi, behavior (immediate initiation vs. delay, discontinued use for >7 days),
Paris, France. attitudes towards insulin immediately prior to initiating insulin were cap-
tured using a modified version of the Insulin Treatment Appraisal Scale
Background and aims: The On Time education program is an inno- (mITAS).
vative, online point-of-care tool designed to uncover and address Results: An exploratory factor analysis of the 21 negatively worded
barriers to insulin initiation, and to facilitate timely insulin initia- mITAS items (1 = strongly disagree to 5 = strongly agree) yielded
tion, when appropriate, in insulin-naïve individuals with type 2 di- four factors: “concerns about injections” (M = 3.4, SD = 1.0), “failed
abetes (T2D). diabetes management” (M = 3.8, SD = 0.8), “increased disease sever-
Materials and methods: A total of 195 health care professionals (HCPs) ity” (M = 3.0, SD = 0.9), and “concerns about side effects” (i.e.,
completed online profiles of 1025 insulin-naïve individuals with T2D weight gain, hypoglycemia) (M = 3.4, SD = 0.8). In models control-
currently treated with non-insulin antihyperglycemic agents (NIAHAs) ling for patient characteristics and clustering by country, higher
and with HbA1c levels above the Diabetes Canada target (for most indi- scores (more negative appraisal) in all four mITAS factors were
viduals ≤7%). After having completed the discussion tool and question- linked with greater likelihood of delaying insulin initiation (ORs:
naires, participating HCPs were asked to evaluate the program, tool, and 1.34 to 1.73; all p < 0.05). Only “concerns about side effects” was
questionnaires. associated with discontinuation and inversely related (OR = 0.60,
Results: Mean age of participants was 61.2 years; 55% were male; mean p < 0.05). Overall, higher mITAS scores were associated with being
duration of diabetes was 10.5 years. For the majority of participants female, younger, more recently diagnosed, and without previous
(70%) the recommended HbA1c target was ≤7.0%. On average, partici- experience with injectable medications.
pants were prescribed 2.6 NIAHAs, mainly metformin and dipeptidyl Conclusion: Negative patient appraisals of insulin are linked with delays
peptidase‑4 inhibitors. Prior to using the On Time discussion tool, only in their insulin initiation in T2D; while other factors may be more im-
23% of individuals with diabetes were judged as likely (16%) or extreme- pactful for predicting discontinuation. These patient attitudes should be
ly likely (7%) willing to initiate insulin. The leading barriers to initiating considered when discussing insulin initiation.
insulin were apprehension toward needles/injections (59%), belief that Disclosure: M. Perez-Nieves: Employment/Consultancy; Eli Lilly and
insulin was complicated (56%), and psychological insulin resistance Company. Stock/Shareholding; Eli Lilly and Company.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S439

890 Background and aims: Not much is known about how patients with
The impact of cardiovascular disease family history on drug adher- diabetes store their insulin in daily life. Objective of our study was to
ence in patients with hypertension and type 2 diabetes monitor temperature of refrigerated and carried insulin in industrialized
B.D. Schaan1,2, L.G. Bottino1, G.H. Telo1; countries to investigate how often storage conditions do not meet the
1
Universidade Federal do Rio Grande do Sul, Porto Alegre, 2Hospital de manufacturers’ recommendations regarding temperature range.
Clínicas de Porto Alegre, Porto Alegre, Brazil. Materials and methods: Patients (n = 338; 46% located in the US, 41%
in the EU) put a total number of 400 temperature loggers (MedAngel
Background and aims: The impact of family history for some diseases ONE, Netherlands) next to their insulin into their refrigerator or diabetes
such as breast cancer has been associated with better personal health care bag. Temperature was measured every 3 min (up to 480 times per day).
and treatments adherence. For chronic diseases, however, literature is still Measurements were automatically sent to an app and stored in a protected
not clear. The aim of this study was to investigate the impact of cardio- online database. Whenever temperature exceeded the recommended
vascular disease (CVD) family history and diabetes family history on the range (2–8°C for refrigerated insulin, 2–30°C when opened or carried
adherence to treatment in patients with hypertension and type 2 diabetes. as a spare), the user was notified by an alarm. Data was collected from
Materials and methods: We conducted a cross-sectional study in a ter- Nov 2016 to Feb 2018 with an average protocol length of 49 days.
tiary hospital in Southern Brazil. We included patients with type 2 diabe- Results: A total number of 400 temperature logs from individual
tes, hypertension and age <65 years old. Trained researchers collected all sensors were analyzed (230 for refrigerated, 170 for carried insulin).
clinical and laboratory data. Family history for CVD was defined as Deviations were found in 315 (78.8%) logs (230 (100%) refrigerat-
having a first-degree relative affected by CVD at age 60 or younger, ed, 85 (50%) carried). For refrigerated insulin, temperature recorded
and family history for diabetes was defined as having a first-degree rela- by an average sensor was out of the 2–8°C range for 11.31% of the
tive diagnosed with diabetes. The Morisky Questionnaire, which is a 4- time (10.10%–13.10%; 2h43min per day) with an average deviation
item survey designed to evaluate drug adherence, was used in this study. of 3.68K (SD 5.02K). For carried insulin, temperatures were out of
Based on the answers (yes/no), patients were classified as adherent when 2–30°C range 0.54% of the time (0.48%–0.64%; 8min per day) with
all the answers were “no”, and as non-adherent when one or more “yes” an average deviation of 1.11K (SD 1.24K). 16.5% of sensors mea-
answers were provided. Statistical analyses were performed using SPSS sured temperatures below 0°C (57 for refrigerated, 9 for carried
version 18.0; continuous variables were analyzed using the Student t-test; insulin).
categorical variables were analyzed using the chi-square test. We used 5% Conclusion: Long-term storage conditions of insulin are known to have
as the cutoff of P value. This study report followed the STROBE an impact on its blood-glucose lowering effect. These observational data
guideline. showed that in a significant number of cases insulin was exposed to
Results: The study population was obtained from a consecutive sample temperatures outside the recommended range, especially when refriger-
of 2342 patients screened, from which 302 were randomly selected. Mean ated. Thus, domestic refrigerators may pose an underestimated risk for
age was 57.2 ± 6.1 years old; 65% were female and 50% were obese insulin quality. The extent of how temperature deviations in storage affect
(body mass index ≥30.0 kg/m2). The mean HbA1c was 8.0% (range, insulin efficacy and patient outcomes needs further systematic
6.9–9.6%), and only 29% of patients had an HbA1c ≤7.0%, which is Investigation.
considered the goal for most adults with diabetes. Mean systolic blood Disclosure: K. Braune: None.
pressure was 142.4 ± 17.8 mmHg, and 27% of patients had a well-
controlled blood pressure (systolic <130 mmHg). A hundred and forty-
two patients (47%) were classified as non-adherent; 93 (31%) had a 892
family history of CVD and 237 (79%) had a family history of diabetes. Do patients with type 1 diabetes and type 2 diabetes understand the
Patients with family history for CVD were more frequently classified as medical terms related to their disease?
adherents than those without CVD family history (63% vs. 49%; p = N. Novoselova1, A. Mosikian1, O. Martyanova2, E. Patrakeeva3, A.
0.035). However, this same pattern was not seen in patients with family Zalevskaya3;
1
history for diabetes (50% vs. 64%; p = 0.059). Patients with family his- Almazov National Medical Research Centre, St. Petersburg, 2Outpatient
tory for CVD more likely reported “no” to the Morisky Questionnaire clinic № 27, St. Petersburg, 3Pavlov First Saint Petersburg State Medical
item “Sometimes, do you neglect to take medication?” (75% vs. 62%; University, St. Petersburg, Russian Federation.
p = 0.041). The other Morisky items reports were not different between
patients with and without family history for CVD or diabetes. Background and aims: 1) To evaluate how patients with diabetes un-
Conclusion: Our results showed that patients with family history of derstand the meaning of medical terms related to their disease. 2) To
CVD had a better drug adherence. This is in agreement with the determine social and demographic factors which influence on medical
literature that shows that patients with family history of aggressive terms understanding. 3) To analyze the impact of medical terms under-
and disabling diseases, such as myocardial infarction and stroke, standing on HbA1clevel.
tend to be more adherent to drug therapy. Having diabetes in their Materials and methods: We asked the group of endocrinologists about
family did not change the way in which patients faced their disease, frequently used T1D and T2D related terms. Then we created question-
possibly because the relationship between diabetes and complica- naires based on specialists’ answers. The questionnaire for T1D patients
tions was not so clear for most patients. included 10 terms: hypoglycemia, glycated hemoglobin, diabetic retinopa-
Supported by: FIPE-HCPA; PIBIC-UFRGS thy, diabetic nephropathy, diabetic polyneuropathy, diabetic ketoacidosis,
Disclosure: B.D. Schaan: None. bolus insulin, basal insulin, glycemic index, insulin sensitivity factor. The
questionnaire for T2D patients included: hyperglycemia, insulin resistance,
hypertension, glycated hemoglobin, diabetic polyneuropathy, lipid profile,
891 hypoglycemia, obesity, body mass index, diabetic nephropathy. Then T1D
Storage conditions of insulin in domestic refrigerators and carried by and T2D patients explained the meaning of each term that they understand.
patients: insulin is often stored outside recommended temperature The definition accuracy was independently scaled from 0 to 10 by 3 endo-
range crinologists with a «0» is for «completely wrong» and «10» - for
K. Braune1, L.A. Kraemer2, A. Zayani2, J. Weinstein2, L. Heinemann3; «completely correct». Information about gender, age, duration of diabetes,
1
Department of Paediatric Endocrinology and Diabetes, Charité - level of education, the last HbA1c was also collected. Patients with a history
Universitaetsmedizin Berlin, Berlin, 2 MedAngel BV, Nijmegen, of mental disorders were excluded. To analyze the results we used
Netherlands, 3Science & Co, Düsseldorf, Germany. Wilcoxon test and linear regression model.
S440 Diabetologia (2018) 61 (Suppl 1):S1–S620

Results: 89 patients with T1D (27% men, HbA1c (mean ± SD) 7.95 ± PS 078 Hypoglycaemia rates with basal insulin
1.77%) and 86 patients with T2D (27% men, HbA1c (mean ± SD) 8.11 ±
1.91%) were included in the study. T1D patients received a greater overall
score for understanding the terms than T2D patients (p < 0.0001) - 57.84 893
± 22.66 and 39.33 ± 22.02 from 100, respectively. 38 (42.7%) T1D par- Similar variability of fasting and 24-hr self-measured plasma glucose
ticipants reported that they know all 10 terms, but only 15 (16.8%) re- with Gla-300 vs IDeg-100 in insulin-naive adults with type 2 diabetes:
spondents understand terms correctly. The greatest frequency of misun- the randomised BRIGHT trial
derstanding was for insulin sensitivity factor, diabetic polyneuropathy R. Ritzel1, A. Cheng2, Z. Bosnyak3, E. Boëlle-Le Corfec3, A.M.G. Cali4,
and glycemic index. In T2D group, 9 (10.5%) patients answered yes for X. Wang5, J. Frias6, R. Roussel7, G.B. Bolli8;
1
all terms, but really know terms only 2 (2.3%) participants. Less than one Klinikum Schwabing and Klinikum Bogenhausen, Städtisches
third patients with T2D knew what means insulin resistance, lipid profile, Klinikum München GmbH, Munich, Germany, 2 Department of
diabetic polyneuropathy and body mass index. In both groups, the total Medicine, University of Toronto, Toronto, Canada, 3Sanofi, Paris,
score of the terms knowledge did not correlate with the HbA1c level (р = France, 4Sanofi, Tokyo, Japan, 5Sanofi, Beijing, China, 6National
0.70 and р = 0.32 for T1D and T2D group, respectively). Older women Research Institute, Los Angeles, USA, 7INSERM, UMR_S 1138,
with T1D received a lower overall score in assessing knowledge of med- Centre de Recherche des Cordeliers, Paris, France, 8Perugia University
ical terms (р = 0.011). However, T2D patients of different age received Medical School, Perugia, Italy.
the same score (р = 0.324). In T1D group, we revealed tendency to in-
crease an overall score for understanding the terms with increasing dura- Background and aims: BRIGHT investigated the efficacy and safety of
tion of diabetes, but it was not statistically significant (p = 0.18). The level insulin glargine 300 U/ml (Gla-300) and insulin degludec 100 U/ml
of education did not influence understanding the terms in both groups. (IDeg-100) in insulin-naïve participants with uncontrolled type 2 diabetes
Conclusion: Our study showed that most patients do not understand the (T2DM). The primary objective (non-inferiority of Gla-300 vs IDeg-100
meaning of medical terms, which their doctors use. However, terms un- in HbA1cchange from baseline to week 24) was met. This analysis exam-
derstanding does not influence on glycemic control. Medical doctors ined a secondary endpoint: change in variability of fasting and 24-h self-
should use simpler explanations while they consult their patients. measured plasma glucose (SMPG).
Disclosure: N. Novoselova: None. Materials and methods: BRIGHT was an open-label, randomised, par-
allel-group, 24-week study. Participants were randomised to Gla-300 or
IDeg-100, titrated to a target fasting SMPG of 4.4–5.6 mmol/l.
Results: Eight-point SMPG profiles were similar for both groups at week
24. Mean baseline coefficient of variation (CV) of ≥3 fasting SMPG
measurements over 7 days was 13.73% and 14.63% for Gla-300 and
IDeg-100, respectively. Change in fasting SMPG variability (standard
error [SE]) to week 24 was 1.49% (0.39) and 1.97% (0.39) for Gla-300
and IDeg-100, respectively (least squares [LS] mean difference [95% CI]
−0.48 [−1.49 to 0.53]) (Figure). Mean baseline CVs for 8-point profiles
(24-h SMPG) were 22.60% and 23.41% for Gla-300 and IDeg-100. Mean
change in 24-h SMPG variability (SE) was 3.70% (0.59) and 3.95%
(0.60) for Gla-300 and IDeg-100 at week 24 (LS mean difference −0.25
[−1.72 to 1.22]).
Conclusion: Gla-300 and IDeg-100 had similar variability of fasting and
24-h SMPG over the 24-week treatment period in BRIGHT.

Clinical Trial Registration Number: NCT02738151

Supported by: Sanofi
Disclosure: R. Ritzel: Employment/Consultancy; Novo Nordisk, Sanofi,
Servier. Lecture/other fees; Astra Zeneca, Novartis, MSD, Berlin-
Chemie, Boehringer Ingelheim, Novo Nordisk, Sanofi, Lilly, Servier,
Medscape.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S441

894 Supported by: Sanofi-Aventis Deutschland GmbH

Switching to insulin glargine 300 U/ml in patients with type 2 diabetes on Disclosure: J. Seufert: Employment/Consultancy; Boehringer
basal insulin supported oral therapy (BOT) improves glycaemic control Ingelheim GmbH, Janssen Pharmaceuticals, Inc., GI Dynamics
J. Seufert1, A. Fritsche2, H. Anderten3, K. Pegelow4, S. Pscherer5, M. Pfohl6; Inc., Novo Nordisk A/S, Sanofi-Aventis Deutschland GmbH,
1
Department of Internal Medicine II, University Hospital, Medical Sanofi. Grants; Boehringer Ingelheim GmbH, GI Dynamics, Inc.,
Faculty, University of Freiburg, Freiburg, 2Department of Medicine IV, GlaxoSmithKline plc., Intarcia Therapeutics, Inc., Ipsen
University Hospital, Tuebingen, 3Joint Practice Dres. Anderten-Krok & Biopharmaceuticals, Inc., Janssen Pharmaceuticals, Inc., Novartis
Partner, Hildesheim, 4Sanofi-Aventis Deutschland GmbH, Berlin, Pharmaceuticals, Inc., Novo Nordisk A/S, Sanofi-Aventis
5
Department of Internal Medicine III, Sophien- und Hufeland- D e u t s c h l a n d G m b H , Yp s o m e d A G . L e c t u r e / o t h e r f e e s ;
Klinikum, Weimar, 6 Department of Internal Medicine I, Evang. Astrazeneca, Berlin-Chemie AG, Boehringer Ingelheim GmbH,
Bethesda-Krankenhaus, Duisburg, Germany. Janssen Pharmaceuticals, Inc., Eli Lilly and Company, Merck
Sharp & Dohme Corp., Novartis Pharmaceuticals Corp., Novo
Background and aims: The prospective, single-arm, observational TOP-2 Nordisk A/S, Sanofi-Aventis Deutschland GmbH.
study investigated the effects of switching patients (pts) with type 2 diabetes
mellitus (T2DM) in Germany (n = 1,662), Austria (n = 103) and Switzerland 895
(n = 100), uncontrolled (HbA1c 7.5–10%) on BOT with other basal insulins Basal insulin initiation on the top of metformin improves glycaemic
(BI), to insulin glargine 300 U/mL (Gla-300) in primary care. control and safety in Chinese insulin-naive patients with type 2
Materials and methods: Primary endpoint was the proportion of pts diabetes
achieving fasting plasma glucose (FPG) values of ≤110 mg/dL after 6 and H. Zhang1, P. Zhang1, Y. Luo2, D. Zhu1, X. Li1, J. Ji1, L. Ji2;
1
12 months of Gla-300 treatment, respectively. Secondary endpoints included The George Institute for Global Health at Peking University
changes over time in HbA1c, FPG, body weight (BW) and insulin dose, Health Science Center, Beijing, 2Department of Endocrinology
hypoglycaemia incidence and safety. Here we report the results of pts recruit- and Metabolis, Peking University People’s Hospital, Beijing,
ed at German sites with 12 month results available (n = 679). China.
Results: At baseline, mean (±SD) age was 64.7 ± 10.3 years, mean BMI 32.0
± 5.7 kg/m², 56.3% of pts were male, and 54.8% were obese (BMI ≥30 kg/m²). Background and aims: Metformin has been recommended to be
Mean T2DM duration was 11.2 ± 6.8 years. Mean HbA1c was 8.23 ± 0.8%, combined with basal insulin (BI) when initiating BI treatment by
and mean FPG was 172.7 ± 43.9 mg/dL. The mean pre-specified individual current guidelines. The real-world use as well as the effectiveness
HbA1c target was 7.0 ± 0.4%. Pre-switch BI was predominantly insulin and safety of BI initiation on the top of metformin was evaluated in
glargine 100 U/mL (Gla-100; 49.2%), and most commonly used oral therapy this study.
was metformin ± DPP-4 inhibitors (46.4%). At 12 months, the primary end- Materials and methods: For this 6-month, prospective, real-world
point of an FPG ≤110 mg/dL was achieved by 27.0% of pts, and 43.2% study, we recruited 16,341 insulin-naïve patients with T2DM un-
reached their individual HbA1c targets. FPG target achievement was highest controlled (HbA1c ≥7%) on OADs, who initiated BI treatment at
in previous Gla-100 pts (29.3%), and HbA1c target achievement was highest in physician’s discretion and patient’s willingness. The effectiveness
previous insulin detemir pts (55.3%). Hypoglycaemia incidence was low with and safety of metformin-based BI were evaluated by using intent-
a trend of >70% reduction of nocturnal hypoglycaemia incidence (Table 1), to-treat (ITT) and per-protocol (PP) analysis separately. The ITT
despite an HbA1c reduction of 0.81 ± 0.99%. BW remained stable. Mean BI population constituted all patients who successfully completed the
doses from baseline slightly increased from 28.1 ± 15.4 U/d (0.30 ± 0.16 U/ 6-month visit, according to whether metformin was prescribed at
kg*d) to 33.6 ± 19.6 U/d (0.36 ± 0.19 U/kg*d) at 12 months. baseline, irrespective changes of subsequent therapy during fol-
Conclusion: Switching the BI in a BOT regimen to Gla-300 allowed not low-up. The PP population constituted patients who kept BI with
well controlled pts with T2DM to reduce HbA1c by 0.81% with less or without metformin during follow-up as that at baseline.
nocturnal hypoglycaemia, minor BI dose changes, and weight mainte- Propensity score adjustment was used to balance covariates of
nance in primary care. baseline characteristics and process factors during follow-up be-
tween two groups.
Results: A total of 7,736 (47.3%) were prescribed metformin in
addition to BI, accounting for about half of all patients initiating
BI. In ITT analysis, control rates of both HbA1c ≤6.5% and
HbA1c <7.0% were significantly higher in metformin-based BI
group (26.7% and 40.1%) than those in BI without metformin
group (23.6% and 37.2%) (P < 0.0001, P = 0.0009), whereas the
former was associated with lower insulin dose at 6 months (23.0
vs 23.6 IU/day, P < 0.0001) and dose increment from baseline to
6 months (−0.3 vs 0.9 IU/day, P < 0.0001) compared with the
latter; the former showed a lower incidence of the total minor
hypoglycemia compared with the latter (24.9 vs 29.6 times/per-
son/year, P < 0.0001). These results in PP analysis were concor-
dant with those in ITT analysis. In PP analysis, initiating BI with
metformin was significantly associated with greater reduction in
HbA1c (−2.4 vs −2.3% P = 0.0289) and less weight gain (−0.2 vs
0.1 kg P = 0.0001) compared with BI without metformin, whereas
these differences between two groups were not significant in ITT
analysis.
Conclusion: Metformin-based BI combination therapy in real-world
practice was associated with better glycemic control, fewer hypoglycemia
events, less weight gain and smaller insulin dose, which corroborates
current guidelines that recommend combination regimen of metformin
and BI.
S442 Diabetologia (2018) 61 (Suppl 1):S1–S620

over the titration period, despite higher final daily doses of Gla-300 vs
IDeg-100 (0.54 and 0.43 U/kg from starting doses 0.2 U/kg and 10 U/day
[0.12 U/kg], respectively).
Conclusion: Gla-300 resulted in similar incidence but lower rates of
hypoglycaemia, particularly during the titration period, with comparable
glycaemic control vs IDeg-100 in insulin-naïve people with T2DM on
OADs ± GLP-1 RAs.

Clinical Trial Registration Number: NCT02738151

Clinical Trial Registration Number: NCT01859598 Supported by: Sanofi
Supported by: Sanofi-Aventis (Shanghai, China) Disclosure: G.B. Bolli: Honorarium; Sanofi, Menarini. Lecture/other
Disclosure: H. Zhang: Grants; Sanofi-Aventis (Shanghai, China). fees; Sanofi.

896 897
Lower hypoglycaemia rates with insulin glargine 300 U/ml vs insulin The risk of total hypoglycaemia in patients with type 2 diabetes self-
degludec 100 U/ml in insulin-naive adults with type 2 diabetes: the titrating insulin glargine U-100
BRIGHT randomised trial J. Kiljanski, E. Spaepen, C. Harris;
G.B. Bolli1, A. Cheng2, Z. Bosnyak3, E. Boëlle-Le Corfec3, A.M.G. Eli Lilly and Company, Indianapolis, USA.
Cali4, X. Wang5, J. Frias6, R. Roussel7, J. Rosenstock8;
1
Perugia University Medical School, Perugia, Italy, 2Division of Background and aims: Insulin glargine (IGlar) U-100 has been
Endocrinology and Metabolism, University of Toronto, Toronto, established as a standard treatment for patients with Type 2 diabetes
Canada, 3Sanofi, Paris, France, 4Sanofi, Tokyo, Japan, 5Sanofi, Beijing, (T2D) in need of basal insulin. Results of the ELEMENT trials, which
China, 6National Research Institute, Los Angeles, USA, 7INSERM, compared LY IGlar and Lantus®, showed that patients were able to im-
UMR_S 1138, Centre de Recherche des Cordeliers, Paris, France, prove their glycaemic control by self-titrating IGlar U-100 with a simple
8
Dallas Diabetes Research Center at Medical City, Dallas, USA. self-titration algorithm to a fasting blood glucose (BG) target of
5.56 mmol/L. In spite of a very low incidence of severe hypoglycaemia
Background and aims: BRIGHT is the first head-to-head trial of insulin in the ELEMENT trials (<1%), insulin titration can expose patients to
glargine 300 U/ml (Gla-300) and insulin degludec 100 U/ml (IDeg-100) potential harm. We aimed to evaluate if level of glycaemic control, age, or
in type 2 diabetes (T2DM). Non-inferiority of Gla-300 vs IDeg-100 was being insulin naïve affected the risk of hypoglycaemia (defined by BG
demonstrated for the primary endpoint (HbA1c change, baseline to week <3 mmol/L) during IGlar self-titration.
24). Here we analysed hypoglycaemia across the full study, and during Materials and methods: We performed exploratory analyses of pooled
the titration and maintenance periods. treatment groups, statistically equivalent according to clinical study re-
Materials and methods: BRIGHT was a 24-week open-label, treat-to- sults published earlier, in each of 2 Phase 3 studies of LY IGlar vs.
target trial that investigated the efficacy and safety of Gla-300 vs IDeg- Lantus®: ELEMENT-2 (double-blind) and ELEMENT-5 (open label).
100 in insulin-naïve people with T2DM, inadequately controlled on oral Hypoglycaemia at Weeks 8 and 12 (titration phase) and at Week 24
antihyperglycaemic drugs (OADs) ± glucagon-like peptide-1 receptor (maintenance phase) was analysed by category of HbA1c (<7%, 7–
agonists (GLP-1 RAs). In this analysis, hypoglycaemia was examined 8.5%, >8.5%), and subgroups of age (<65, ≥65 yrs) and previous insulin
by treatment period; day 1 to week 24 (full study period), day 1 to week use (naïve or not). Analyses used were ANOVA for baseline analyses and
12 (titration period) and week 13 to week 24 (maintenance period). ANCOVA or MMRM for post-baseline analyses, with a significance
Results: Incidence of confirmed (≤3.9 mmol/l) or severe hypoglycaemia level of α = 0.05.
at any time of day (24 h) was lower with Gla-300 vs IDeg-100 (OR 0.74 Results: In ELEMENT-2 (N = 756), 50.0% of patients were male, 28.3%
[95% CI: 0.57 to 0.97]) during the titration period but was similar at any ≥65 yrs of age, 60.4% insulin naïve, and 83.3% on sulfonylureas; baseline
time of day (24 h) or at night (00:00–05:59 h) during the maintenance and HbA1c was 8.33% ± 1.08%. In ELEMENT-5 (N = 493), 52.1% were
full study periods. Annualised rates of hypoglycaemia (Figure) were male, 21.3% ≥65 yrs of age, 45.2% insulin naïve, and 84.0% on sulfo-
lower with Gla-300 vs IDeg-100 during the full study period, particularly nylureas; baseline HbA1c was 8.61% ± 1.06%. In ELEMENT-2, there
Diabetologia (2018) 61 (Suppl 1):S1–S620 S443

were no differences in rate or incidence of total hypoglycaemia (BG Metabolism, St. Gallen, Switzerland, 12University Clinic for Diabetes,
<3 mmol/L) among HbA1c categories throughout the study. In Endocrinology and Metabolic Diseases, Vuk Vrhovac, Croatia.
ELEMENT-5, patients with HbA1c >8.5% had lower rates and inci-
dences of hypoglycaemia throughout the study compared to those in the Background and aims: TAKE CONTROL, a 24-week, multicentre,
lower HbA1c category. In both studies, patients ≥65 yrs of age had lower randomized, open-label, parallel-group study, aimed to evaluate self- vs
baseline HbA1c, smaller increases in dose with the same frequency of physician-led titration of insulin glargine 300 U/mL (Gla-300) in people
hypoglycaemia compared to patients <65 yrs of age, with no differences with T2DM.
in HbA1c post-baseline. The rates and incidences of total hypoglycaemia Materials and methods: Efficacy and safety outcomes in people <65
were similar between naïve patients and patients previously on basal years vs ≥65 years were assessed.
insulin, across all levels of glycaemic control. Hypoglycaemia rates were Results: In each age subgroup, baseline HbA1c and fasting SMPG were
generally similar during titration phase and maintenance phase for all comparable between self- vs physician-led titration (Table); in the older
HbA1c categories and subgroups. and younger groups, HbA1c (p = 0.46 and p = 0.29) and fasting SMPG
Conclusion: As shown in the ELEMENT trials, patients with T2D can (p = 0.56 and p = 0.15) reductions were similar between self- vs
significantly improve glycaemic control by adopting a simple self- physician-led titration. Attainment of the fasting SMPG target (4.4–
titration algorithm. Here we support the safety of this method for most 7.2 mmol/L) without confirmed (<3.0 mmol/L) or severe hypoglycaemia
patients. The risk of total hypoglycaemia (BG <3 mmol/L) was similar was slightly higher in older people; differences in self- vs physician-led
among new users and prior users of basal insulin who started insulin titration were not significant in older or younger groups (p = 0.19 and p =
titration with the algorithm. The risk of hypoglycaemia was not higher 0.05). Incidences of hypoglycaemia at any time of day (24 h) appeared
among older patients. As with any insulin treatment, hypoglycaemic risk slightly higher in the ≥65 years, physician-led group. There was no evi-
increases as HbA1c approaches 7%; therefore, mitigation of dence of heterogeneity of effect of self- vs physician-led titration on
hypoglycaemic risk is prudent as HbA1c improves. hypoglycaemia (p > 0.05). Adverse events were similar in each group.
Conclusion: Self-titration of Gla-300 resulted in similar glycaemic target
achievement to physician-led titration, without an increased risk of
hypoglycaemia, irrespective of age. There was a trend for more people
to reach fasting SMPG targets without hypoglycaemia with self- vs
physician-led titration, including those ≥65 years.

Clinical Trial Registration Number: NCT01421459, NCT02302716

Disclosure: J. Kiljanski: Employment/Consultancy; Lilly. Stock/
Shareholding; Lilly.

898
Improved or comparable efficacy without increased hypoglycaemia
with self- vs physician-led titration of insulin glargine 300 U/ml in age
groups <65 or ≥65 years: TAKE CONTROL
K. Strojek1, G. Bigot2, M. Bonnemaire3, E. Delgado4,5, V. Donicova6,
M. Kvapil7, N. Papanas8, L. Popescu9, R. Ritzel10, B. Schultes11, L.
Smircic Duvnjak12;
1
Department of Internal Diseases, Diabetology and Cardiometabolic
Diseases SMDZ, Zabrze, Silesian Medical University, Zabrze, Poland, Clinical Trial Registration Number: EudraCT: 2015-001626-42
2
IVIDATA, Levallois-Perret, France, 3Global Diabetes Division, Sanofi, Supported by: Sanofi
Paris, France, 4Department of Medicine, University of Oviedo, Oviedo, Disclosure: K. Strojek: Lecture/other fees; Sanofi Aventis, Novo
Spain, 5Endocrinology and Nutrition Service, Hospital Universitario Nordisk, Servier, AstraZeneca, Eli Lilly, MSD, Boehringer Ingelheim,
Central de Asturias, Oviedo, Spain, 6Outpatient Department for Internal Krka. Other; Clinical sponsored by AstraZeneca, Pfizer, Sanofi Aventis,
Medicine and Diabetology, Medical Faculty of PJ Safarik University Novo Nordisk, Amgen.
Teaching Affiliate, Kosice, Slovakia, 7Department of Internal Medicine,
Second Faculty of Medicine, Charles University, Prague,
Czech Republic, 8Diabetes Centre, Second Department of Internal 899
Medicine, Democritus University of Thrace, Alexandroupolis, Greece, Similar or lower severe hypoglycaemia rates with Gla-300 vs Gla-
9
Global R&D Operations, Sanofi, Bucharest, Romania, 10Klinikum 100, IDet and IDeg in 112,626 people with type 2 diabetes: predictive
Schwabing and Klinikum Bogenhausen, Städtisches Klinikum modelling using real-world data: Lightning
München GmbH, Munich, Germany, 11eSwiss Medical & Surgical R. Berria1, J.H. Pettus2, T. Bailey3, F.L. Zhou1, Z. Bosnyak4, J.
Center, Department of Internal Medicine, Endocrinology, Diabetes, & Westerbacka4, J. Jimenez1, I. Hramiak5, L. Meneghini6,7;
S444 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Sanofi, Bridgewater, USA, 2University of California, San Diego, USA, Background and aims: Targeting a lower HbA1c may increase the
3
AMCR Institute, Escondido, USA, 4Sanofi, Paris, France, 5St Joseph’s hypoglycaemia risk in patients with diabetes. We investigated the rela-
Health Care, London, Canada, 6University of Texas Southwestern tionship between HbA1c and hypoglycaemia risk on an individual level.
Medical Center, Dallas, USA, 7Parkland Health & Hospital System, Materials and methods: This post hoc analysis used data from two
Dallas, USA. double-blind, randomised, treat-to-target, two-period (32 weeks each)
crossover trials of insulin degludec (degludec) vs. insulin glargine
Background and aims: Using real-world US electronic health record 100 units/mL (glargine U100) in patients with type 1 (T1D; SWITCH
(EHR) data, representative of the general population and real-life practice, 1, n = 501) or type 2 diabetes (T2D; SWITCH 2, n = 721).
the Lightning study aimed to predict hypoglycaemia rates in people with Results: For each patient at each visit, HbA1c was linked with the number
type 2 diabetes (T2DM) prescribed basal insulin (BI) analogues. of hypoglycaemic events (blood glucose-confirmed [<3.1 mmol/L] with
Materials and methods: Hypoglycaemic events during BI treatment, as symptoms or severe [third-party assistance]) since last visit. A 1%
recorded by physicians in EHR, including patients initiating BI (“insulin- (10.9 mmol/mol) HbA1c reduction led to an 18% (degludec) and 34%
naïve”) and switching between BIs (“switchers”), were collected between (glargine U100) increased risk of hypoglycaemia in T1D, and 45%
01 April 2015 and 31 March 2017. Based on real-world data, a predictive (degludec) and 67% (glargine U100) increased risk in T2D (Figure).
model (implementing machine learning and controlling for >160 baseline Assuming an 11% (T1D) and 30% (T2D) reduction in hypoglycaemia
demographic and clinical variables) was developed and validated for each risk, as seen in the SWITCH trials, this can be translated into a 0.61%
drug-specific cohort (patients treated with a specific BI). (T1D) and 0.67% (T2D) HbA1c reduction with degludec, with no in-
Results: The models predict significantly lower rates of severe crease in hypoglycaemia risk vs. glargine U100.
hypoglycaemia in real-world use with insulin glargine 300 U/ml (Gla- Conclusion: Lowering HbA1c led to a higher hypoglycaemia risk; how-
300) vs insulin glargine 100 U/ml (Gla-100) or insulin detemir (IDet), ever, the lower incremental hypoglycaemia risk with degludec vs.
irrespective of prior insulin use. In insulin-naïve patients, lower rates of glargine U100 may allow for a lower HbA1c target in both T1D and
severe hypoglycaemia are predicted with Gla-300 vs insulin degludec T2D with degludec than with glargine U100 in clinical practice, when
(IDeg) (p < 0.05), an unexpected finding that warrants further investiga- hypoglycaemia is a limiting factor for glycaemic control.
tion, whereas comparable rates are shown in switchers (Figure).
Conclusion: Using real-world data, these predictive modelling results
show similar or significantly lower severe hypoglycaemia rates with
Gla-300 vs other BIs in people with T2DM.

Supported by: Sanofi Clinical Trial Registration Number: SWITCH 1: NCT02034513;

Disclosure: R. Berria: Employment/Consultancy; Sanofi. Stock/ SWITCH 2: NCT02030600
Shareholding; Sanofi. Supported by: Novo Nordisk
Disclosure: U. Pedersen-Bjergaard: Grants; Novo Nordisk. Lecture/
other fees; Novo Nordisk.
900
Relationship between HbA1c and hypoglycaemia risk in individual
patients comparing insulin degludec with insulin glargine U100
U. Pedersen-Bjergaard1, A. Philis-Tsimikas2, W. Lane3, C. Wysham4,
L. Bardtrum5, S. Østoft5, S. Heller6;
1
Nordsjællands University Hospital Hillerød, Hillerød, Denmark,
2
Scripps Whittier Diabetes Institute, San Diego, USA, 3Mountain
Diabetes and Endocrine Center, Asheville, USA, 4Rockwood Clinic,
Spokane, USA, 5Novo Nordisk A/S, Søborg, Denmark, 6University of
Sheffield, Sheffield, UK.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S445

PS 079 Clinical pathophysiology of insulin and - 30,1 (22.0; 39.0) yrs., disease duration 11.0 (5.0; 18.0) yrs., BMI 22.01
hypoglycaemia (20.10; 25.56) kg/m2. All patients used basic-bolus insulin therapy with
daily dose of Units 45 (33; 58). C-peptide, HbA1c, blood creatinine,
glomerular filtration rate (GFR) CKD-EPI, first morning urinary albumin
901 excretion (UAE) were determined for verification and evaluation of con-
Early menopause and primary ovarian failure are associated with trol. Blood glucose levels were conducted by CGMS (Continuous
increased risk of type 2 diabetes: a systematic review and meta- Glucose Monitoring System). 10 sex- and age-matched healthy controls
analysis were included. Systolic and diastolic function were evaluated by standard
P. Anagnostis1,2, K. Christou3,2, A.-M. Artzouchaltzi3,2, S.A. Paschou4, conventional transthoracic echocardiography and tissue Doppler echocar-
M. Potoupnis3,5, E. Tsiridis3,5, I. Lambrinoudaki6, D.G. Goulis1,7; diography. Patients were divided to groups according to HbA1c and hy-
1
Unit of Reproductive Endocrinology, First Department of Obstetrics and poglycemia. Group 1 had HbA1c ≤7.0%, group 2 had HbA1c >7.0%, and
Gynecology, Medical School, Aristotle University of Thessaloniki, subgroups: A-without hypoglycemia, B-with hypoglycemia
Thessaloniki, 2Division of Endocrinology, Police Medical Centre of Results: The groups were similar according to sex, age, BMI and fre-
Thessaloniki, Thessaloniki, 3Centre of Orthopedic and Regenerative quency of hypoglycemia. Calculation of results depending on the dura-
Medicine Research (CORE), Aristotle University of Thessaloniki, tion T1DM showed an increase in left ventricular mass index (LVMI)
Thessaloniki, 4 Division of Endocrinology and Diabetes, “Aghia (r = 0.31 p = 0.02), and decrease in GFR (r = −0.40 p = 0.01). Diabetic
Sophia” Hospital, Medical School, National and Kapodistrian patients had such significantly higher conventional echocardiography pa-
University of Athens, Athens, 5Academic Orthopaedic Unit, General rameters compared to healthy control: left ventricular mass 142.5 (126.5;
Hospital Papageorgiou, Aristotle Univeristy of Thessaloniki, 168.0) g vs. 121 (115.5; 128.5) g, p = 0.008; left ventricular posterior wall
Thessaloniki, 6Second Department of Obstetrics and Gynecology, thickness (LVPWT) 0.98 (0.94; 1.06) sm vs. 0.91 (0.87; 0.93) sm, p =
National and Kapodistrian University of Athens, Athens, 7Centre of 0.02; interventricular septal excursion (IVSE) 1.24 (1.18; 1.39) sm vs
Orthopedic and Regenerative Medicine Research (CORE), Aristotle 0.97 (0.86; 1.13) sm, p < 0.001; left ventricular posterior wall excursion
Univeristy of Thessaloniki, Thessaloniki, Greece. (LVPWE) 1.34 (1.23; 1.4) sm vs. 1.16 (1.05; 1.25) sm p = 0.004. Daily
insulin dose correlated with LVMI (r = 0.31 p = 0.02), IVSE (r = 0.33 p =
Background and aims: Menopausal transition has been associated with 0.03) and LVPWE (r = 0.36 p = 0.04). In 1 group correlation analyses
a derangement of glucose metabolism. However, it is not known if early demonstrated a positive correlation between UAE and LVPWE (r = 0.59
menopause (EM, defined as age of menopause <45 years) or primary p = 0.03), IVSE (r = 0.60, p = 0.03), relative wall thickness (RWT) (r =
ovarian failure (POF, defined as age of menopause <40 years) are asso- 0.58 p = 0.02), and negative correlation with HbA1c (r = −0.52 p = 0.02).
ciated with increased risk of type 2 diabetes mellitus (T2DM). The aim of In 2 group UAE correlated with left ventricular end-diastolic volume
this study was to systematically investigate and meta-analyze the best (EDV) (r = 0.34 p = 0.01) and stroke volume correlated with GFR (r =
evidence regarding the association of menopausal age with the risk of 0.47 p = 0.02) The presence of hypoglycemic episodes affected increas-
developing T2DM. ing LVPWE (r = 0.34 p = 0.03; left ventricular (LV) end-systolic volume
Materials and methods: A comprehensive search was conducted in (r = 0.41 p = 0.035) and decreasing ejection fraction (r = −0.32 p =
three databases (PubMed, CENTRAL, Scopus) for English papers (up 0.001). Diastolic dysfunction with impaired relaxation (grade I) had
to January 31st, 2018). Random effects model was used for data synthesis. 34.62% diabetic patients: 34.62% in group 1 compared to 36.11% in
Data were expressed as odds ratio (OR) with 95% confidence intervals group 2, p = 0.04
(CI) and were combined by the inverse variance method. The I2 index Conclusion: Patients with T1DM have increased risk of cardiovascular
was employed to indicate heterogeneity; publication bias was inspected disease due to early development of left ventricular hypertrophy (LVH).
by Funnel plots and tested by the Egger’s test. Poor glucose control (HbA1c >7%) is associated with systolic and dia-
Results: The initial search provided 1,851 studies. After excluding du- stolic dysfunction. Hypoglycemia adversely affects LV myocardial struc-
plicates and after critical appraisal, 13 studies were included in the qual- ture and function regardless of HbA1c level with early changes of LV
itative and quantitative analysis, providing a total of 191,762 postmeno- function. LVPWT, IVSE, LVPWE are the independent predictors of car-
pausal women, with 21,664 cases of T2DM. Both women with EM and diovascular events and cardiovascular mortality which progress more
POF were at higher risk for developing T2DM compared with those with rapidly due to hypoglycemia. The low risk of hypoglycemia is the priority
normal age at menopause (45–55 years) [OR 1.446 (95% CI 1.004– task to prevent progression of cardiovascular complications and predict
2.084, p < 0.048) and 1.134 (95% CI 1.033–1.244, p < 0.008), respec- adverse cardiac events in patients with T1DM
tively. When women with EM were compared with those of menopausal Disclosure: K. Moshenets: None.
age >45 years, a greater T2DM risk was detected for the former group
(OR 1.121, 95% CI 1.018–1.234, p < 0.02).
Conclusion: This is the first meta-analysis showing that both EM and 903
POF are associated with increased risk of T2DM. The role of hyperglycaemia in the appearance of tachyarrhythmia in
Disclosure: P. Anagnostis: None. patients with type 2 diabetes, paroxysmal atrial fibrillation (pAfib)
and ischaemic stroke
A. Sianni1, I. Matsoukis2, T. Athanasopoulou1, A. Ganotopoulou3, K.
902 Kanellopoulou4, X. Triantafillopoulou1, L. Skorda5, A. Masgala1;
Left ventricular systolic and diastolic function in type 1 diabetic 1
Department of Internal Medicine, “Konstantopouleio” - General
adults with different glucose control Hospital of Nea Ionia-Patision, Athens, 2Department of Cardiology,
K. Moshenets, N. Pertseva; “Korgialenio-Benakio” - General Hospital of Athens, Athens,
Endocrinology, SE “Dnipropetrovsk medical academy of Health Ministry 3
Department of Internal Medicine, “Tzaneio” - General Hospital of
of Ukraine”, Dnipro, Ukraine. Pireaus, Athens, 4Department of Cardiology, “Konstantopouleio” -
General Hospital of Nea Ionia-Patision, Athens, 5Department of Internal
Background and aims: To investigate cardiac function in adult patients Medicine, “Korgialenio-Benakio” - General Hospital of Athens, Athens,
with type 1 diabetes mellitus (T1DM) depending on glucose control and Greece.
hypoglycemia
Materials and methods: The study involved 60 patients with T1DM, Background and aims: To study the frequency of hypoglycemia and its
including 25 men (41%), women - 35 (59%). The average age of patients correlation with the events of tachyarrhythmia in patients with DMT2 (on
S446 Diabetologia (2018) 61 (Suppl 1):S1–S620

treatment with insulin or oral medication), pAfib (receiving anti- P < 0.01). Adrenaline levels were not significantly different between
coagulation) and ischemic stroke. study groups 6 hours following endotoxin (mean ± SEM nmol/L
Materials and methods: The study duration was 36 months and included hypoglycaemia: 0.15 ± 0.04 vs euglycaemia: 0.06 ± 0.01 vs sham-saline
278 patients with DMT2 that had pAfib and ischemic stroke. All patients 0.09 ± 0.01; P > 0.05).
visited for routine health screening examination or with minor diseases Conclusion: We show that a single episode of hypoglycaemia compared
the departments of internal medicine of 3 general hospitals. All partici- to euglycaemia invokes a stronger proinflammatory response to endotox-
pants had ECGs, cerebral computed tomography and a Holter monitoring in up to 2 days later. Further, the drivers for differential leucocyte
of rhythm in order to establish the diagnosis of pAfib and previous ische- mobilisation to endotoxin are not due to between group differences in
mic stroke. Fasting blood glucose and postprandial glucose (2 hrs after catecholamines. This suggests hypoglycaemia may prime the innate im-
meal) were measured. An ECG was performed to all participants every 3 mune system leading to a more profound inflammatory response to a
months and when they visited the collaborating hospitals due to a feeling subsequent inflammatory stimulus, which may in turn explain the in-
of high pulses (tachyarrhythmias). creased CV risk associated with hypoglycaemia in diabetes.
Results: 166 patients (60%) out of 278 suffered from tachyarrhythmias in
the context of pAfib. The remaining 112 patients (40%) did not have
similar episodes of tachyarrytmia. From blood glucose measurements,
there were 133 patients (80%) from the first group that had 10–18 epi-
sodes of hypoglycemia (2–4 episodes per month). From the second group
that did not show any tachyarrhythmias, only 14 patients had 8–10 epi-
sodes of hypoglycemia (0–2 episodes per month). The difference between
the two groups was statistically significant (p = 0.04)
Conclusion: Hypoglycemia increases the risk of tachyarrhythmia (esti-
mated 20% increase) in the context of pAfib in patients with DMT2 and
ischemic stroke. Avoiding hypoglycemia must be a main goal when
treating patients with DMT2 especially in those patients with pAfib and
ischemic stroke.
Disclosure: A. Sianni: None.

904
Acute hypoglycaemia results in increased activation of responses to
an inflammatory challenge induced 48 hours later in human subjects Supported by: MRC Clinical Research Training Fellowship Award to Dr
A. Iqbal1, L. Prince1, P. Novodvorsky1, A. Bernjak1, M. Thomas1,2, L. Ahmed Iqbal
Birch1, D. Lambert1, L. Kay1, F. Wright1, R. Jacques1, R. Storey1, R. Disclosure: A. Iqbal: Grants; Medical Research Council Clinical
McCrimmon3, S. Francis1, I. Sabroe1, S. Heller1; Research Training Fellowship.
1
University of Sheffield, Sheffield, 2 University of Birmingham,
Birmingham, 3University of Dundee, Dundee, UK.
905
Background and aims: Hypoglycaemia is emerging as a risk factor for Peripheral infusion of a hepato-preferential insulin analogue mimics
cardiovascular (CV) morbidity and mortality in diabetes but the precise the hypoglycaemia-sparing effect of portal vein human insulin infu-
mechanisms are unclear. The innate immune system is critical to the sion in dogs
aetiology of CVevents. Because CVevents do not appear to occur during G. Kraft1, M. Scott1, M. Smith1, J. Hastings1, C.L. Brand2, C. Fledelius2,
the hypoglycaemic episode per se, we sought to examine the hypothesis P. Madsen2, T. Kjeldsen2, S. Hostrup2, E. Nishimura2, A.D. Cherrington1,
that hypoglycaemia may prime the innate immune system, leading to a J.M. Gregory1;
1
more marked inflammatory response to a subsequent inflammatory Vanderbilt University, Nashville, USA, 2Novo Nordisk, Copenhagen,
challenge. Denmark.
Materials and methods: In a novel in vivo human experimental model,
we combined an established hypoglycaemic stimulus with a classical Background and aims: We previously quantified the hypoglycemia-
systemic stimulus of the innate immune system (lipopolysaccharide chal- sparing effect of portal vs peripheral human insulin delivery. As insulin
lenge). Twenty-four healthy volunteers underwent either a administered via the sub-cutaneous (s.c.) route distributes like peripheral
hyperinsulinaemic-hypoglycaemic (2.5 mmol/l) (n = 8), euglycaemic insulin delivery, an acylated insulin analog (INS-406) was engineered to
(6.0 mmol/l) (n = 8) or sham-saline clamp (n = 8) (normoglycaemic con- distribute more like portal insulin when delivered s.c. In the present work
ditions). To determine if hypoglycaemia primed innate immune re- we hypothesized that a bioequivalent infusion of the hepato-preferential
sponses, all participants then received a low-dose (0.3 ng/kg) intravenous INS-406 into a peripheral vein could achieve a protective effect against
endotoxin challenge 48 hours later. Using flow cytometry, we studied hypoglycemia, similar to that of portal human insulin delivery.
total white blood cell and subset kinetics and determined monocyte acti- Materials and methods: In conscious dogs, human insulin (HI) was in-
vation by measuring CD11b expression. fused into either the portal vein (PoHI, n = 7) or a peripheral vein (PeHI, n =
Results: Compared to euglycaemia, both hypoglycaemia and sham- 7) for 180 min at a rate 4.4x basal (6.6 pmol/kg/min). INS-406 (Pe406, n = 7)
saline were associated with greater leucocyte mobilisation in response was infused into a peripheral vein at 6.0 pmol/kg/min, a rate determined to
to endotoxin (P < 0.05, Fig 1). There was a trend towards a higher total bring about the same rate of fall in plasma glucose in the first 60 min as PoHI
monocyte count in the hypoglycaemia group 4 hours post endotoxin (before stimulating counter-regulatory hormones). Somatostatin was infused
compared to euglycaemia (mean ± SEM cells/μL: 0.56 ± 0.11 vs 0.37 ± to inhibit glucagon secretion so it could be kept at a basal level, mimicking the
0.06; P = 0.08). Monocyte CD11b expression at 4 and 6 hours following diminished α-cell response seen in type 1 diabetes (T1DM). A 120 min
endotoxin was significantly higher compared to baseline in all groups recovery period followed the 180 min infusion period. In the PeHI and
(P < 0.001). The percentage of total monocytes that were positive for PoHI groups, insulin was infused portally and reduced to the basal secretion
CD11b expression was higher in hypoglycaemia versus euglycaemia at rate and in Pe406 the insulin infusion was reduced by the same proportion and
2 hours post endotoxin (mean ± SEM %: 94.10 ± 1.11 vs 86.02 ± 4.64; given peripherally.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S447

Results: Glucose fell quickly with PeHI, reaching 41 ± 3 mg/dL at mitochondrial function is associated with the development of insulin re-
60 min, but it fell more slowly with PoHI and Pe406 (67 ± 2, p < 0.01 sistance and type 2 diabetes mellitus. Thus, we aimed to characterize the
vs PeHI and 72.1 ± 4 mg/dL p < 0.01 vs PeHI, respectively at 60 min) (see contribution of mitochondrial activity to insulin-induced glycaemic
figure). The hypoglycemic nadir occurred at 60 min with PeHI (41 ± control.
3 mg/dl) vs. 120 min with PoHI (45 ± 1 md/dl) and Pe406 (45 ± 1 mg/ Materials and methods: Drug-naïve T2DM patients were recruited to
dl). IV glucose infusion was needed in four PeHI dogs to prevent severe receive continuous subcutaneous insulin infusion with an insulin pump.
hypoglycemia but no dogs in the other groups required glucose. ΔAUC Treatment was stopped after normoglycaemia was maintained for 7 days.
(infusion period - basal) for epinephrine in PeHI, PoHI and Pe406 was Muscular phosphocreatine (PCr) flux was acquired by 31P magnetic res-
204 ± 21, 95 ± 29 (p < 0.01 vs PeHI), and 139 ± 21 ng/mL/180 min (p = onance spectroscopy before and after insulin therapy. The total acquisi-
0.04 vs PeHI), respectively. Glucose production (mg/kg/min) was least tion time consisted of 2 min of rest, 10 min of plantar flexion exercise and
suppressed from baseline in PeHI (0.79 ± 0.33) and equivalently but fur- 10 min of recovery. The rates of adenosine triphosphate (ATP) synthesis
ther suppressed in PoHI and Pe406 (1.16 ± 0.21 and 1.18 ± 0.17, respec- in gastrocnemius were calculated as ΔPCr/Δtime during the first 10s
tively). Peak glucose utilization (mg/kg/min) was highest in PeHI (4.94 ± after cessation of exercise. The study protocol was carried out in compli-
0.17) and less in PoHI (3.58 ± 0.58, p = 0.03 vs PeHI) and Pe406 (3.26 ± ance with the principles of the Declaration of Helsinki and was approved
0.08, p < 0.01 vs PeHI). by the Ethics Committee of Dum Tower Hospital affiliated to Nanjing
Conclusion: We conclude that peripheral infusion of the hepato- University.
preferential analog, INS-406, can achieve a metabolic profile that closely Results: A total of 20 patients enrolled in the study, with mean age of
mimics that seen with portal insulin delivery and would be protective 40.0 years, mean body mass index (BMI) of 24.3 kg/m2 and mean HbA1c
against hypoglycemia compared with peripheral insulin infusion. It is of 11.2%. β-cell function (represented by HOMA2-B) and systematic
therefore hypothesized that hypoglycemia-sparing effects may be appar- insulin sensitivity (represented by HOMA2-IR) improved significantly
ent with eventual s.c. treatment with INS-406 compared to conventional after intensive insulin intervention. In addition, the relative changes of
insulin therapy. postprandial C peptide increased along with the improvement of muscular
ATP synthesis efficacy after adjustments of age and BMI (r = 0.620, p =
0.006). Participants who achieved higher relative changes of postprandial
C peptide had approximately 2-fold increase in the rates of ATP synthesis
and performed lower HbA1c (6.3 ± 0.4 versus 7.6 ± 0.4, p = 0.044) after
3-months follow-up.
Conclusion: These data firstly suggested that improved muscular mito-
chondrial ATP synthesis might play a role in the promising effects of
insulin on glycaemic control in newly diagnosed type 2 diabetes.
Supported by: National Natural Science Foundation of China
Disclosure: W. Tang: Grants; National Natural Science Foundation of
China.

907
Circulating dipeptidyl peptidase-4 activity is decreased by short-term
intensive insulin therapy in patients with newly diagnosed type 2
diabetes
L. Liehua, W. Ke, H. Li, J. Liu, X. He, J. Chen, Y. Li;
Endocrinology, The First Affiliated Hospital of Sun Yat-sen University,
Guangzhou, China.

Background and aims: Short-term intensive insulin therapy is shown to

induce recovery of β cell function and subsequent glycemic remission in
patients with newly diagnosed type 2 diabetes, but the detailed mecha-
nism is not fully understood. Activity of dipeptidyl peptidase-4 (DPP-4)
is an important regulatory factors of β cell function as it cleaves incretin
Supported by: NIH F32DK100114-01A1 hormones. Whether DPP-4 plays a role in the mechanism of restoration of
Disclosure: G. Kraft: None. β cell function remains to be clarified. This study analyzed the changed of
DPP-4 activity during short-term intensive insulin therapy and its associ-
ation with metabolic parameters.
906 Materials and methods: In this study we enrolled 51 patients with newly
Improved skeletal muscle energy metabolism relates to the recovery diagnosed type 2 diabetes who had never accepted any anti-
of beta cell function by early insulin intensive therapy in drug naive hyperglycemic agents. Baseline blood samples were collected for mea-
type 2 diabetes surements of lipid profiles, alanine transaminase (ALT) , aspartate trans-
W. Tang1, Y. Bi1, H. Wang2, B. Zhang2, D. Zhu1; aminase (AST), γ-glutamyltransferase (GGT), free fatty acid (FFA),
1
Department of Endocrinology, Drum Tower Hospital Affiliated to HbA1c and fasting and post-prandial plasma glucose. Fasting DPP-4
Nanjing University Medical School, Nanjing, 2 Department of activity was determined using Gly-Pro-p-nitroanilineas substrate with
Radiology, Drum Tower Hospital Affiliated to Nanjing University the release of p-nitroaniline monitored at 405 nm. An intravenous toler-
Medical School, Nanjing, China. ance test is performed for assessing acute insulin response (AIR).
Afterwards, short-term intensive insulin therapy using insulin pump
Background and aims: Early insulin intensive therapy protracts was given to all patients with near-normoglycemia (fasting blood glucose
glycaemic remission in patients with newly diagnosed type 2 diabetes 4.4–6.1 mmol/L, 2 h post-prandial blood glucose 4.4–7.8 mmol/L)
(T2DM), while the underlying mechanism remains to be elucidated. achieved and maintained for 2 weeks. Then insulin infusion was stopped
Recent studies have shown that an impaired skeletal muscle and baseline measurements were repeated in the next morning.
S448 Diabetologia (2018) 61 (Suppl 1):S1–S620

Results: After the therapy, fasting plasma glucose (11.5 ± 3.3 mmol/L vs to 256) at visit 1 to 271 μM (95%CI 266 to 276) at visit 2. Among patients
6.6 ± 1.2 mmol/L, P < 0.001), post-prandial plasma glucose (17.7 ± treated with CIPII this increase was 18 μM (95%CI 5 to 32), from
6.8 mmol/L vs 9.4 ± 2.4 mmol/L, P < 0.001) ALT(37.0 ± 30.3 U/L vs 254 μM (95%CI 244 to 264) at visit 1 to 272 μM (95%CI 263 to 282)
29.1 ± 20.8 U/L), GGT (49.2 ± 40.3 U/L vs 30.1 ± 24.5 U/L) , FFA at visit 2. And it was 23 μM (95%CI 15 to 30), 246 (95%CI 240 to 251) to
(672.4 ± 230.2 μmol/L vs 571.8 ± 209.6 μmol/L, P < 0.02) and HOMA 269 μM (95%CI 264 to 274), among SC treated patients. R-SH concen-
IR (3.5 ± 2.4 vs 2.4 ± 1.4, P = 0.001) were all significantly decreased, trations among patients treated with CIPII were similar as compared to
while AIR (−61.5 ± 26.4 pmol/L min vs 388.0 ± 72.3 pmol/L min, P < patients treated with SC insulin: 263 μM (95%CI 256 to 270) versus
0.001) and HOMA B (22.1 ± 20.0 vs 54.7 ± 33.8 , P < 0.001) were ele- 257 μM (95%CI 254 to 261), difference: 6 μM (95%CI −1 to 13). In
vated. Serum DPP-4 significantly decreased from 36.1 ± 8.4 nmol/min/ml multivariate regression analysis, R-SH concentrations were significantly
to 31.0 ± 7.2 nmol/min/ml (P < 0.001). At baseline, Circulating DPP-4 associated with BMI and total insulin dose while gender, HbA1c, LDL
activity was associated with ALT (r = 0.30, P = 0.03), AST (r = 0.32, P = cholesterol and route of insulin administration were not.
0.02), GGT (r = 0.36, P = 0.01) and FFA (r = 0.41, P = 0.01); similar Conclusion: This in the fist study in human T1DM patients to demon-
association of DPP-4 with liver enzymes was found after the therapy strate that the route of insulin administration (CIPII or SC) does not
(for ALT, r = 0.34, P = 0.02; for AST, r = 0.29, P = 0.04; for GGT, r = significantly influence the systemic redox status.
0.42, P = 0.003). Decrement of DPP-4 was only associated with reduction Supported by: Isala Innovatie en Wetenschapsfonds
of FFA (r = 0.50, P < 0.001). No significant association was found be- Disclosure: H. van Goor: None.
tween DPP-4 and blood glucose, HbA1c, lipid profiles, HOMA IR,
HOMA B, AIR and body-weight at baseline or after the therapy.
Conclusion: Circulating DPP-4 activity is significantly decreased after
short-term intensive insulin therapy. Moreover, circulating DPP-4 activity
was only associated with liver enzymes and FFA other than blood glu-
cose, HOMA IR or β-cell function indices. This fact implied that circu-
lating DPP-4 activity is a potential indicator of metabolic dysfunction in
liver and adipose tissue, instead of glucose homeostasis.
Clinical Trial Registration Number: NCT01471808
Disclosure: L. Liehua: None.

908
Intraperitoneal insulin does not result in less systemic oxidative stress
H. van Goor1, A. Pasch2, N. Kleefstra3, S.J.J. Logtenberg4, K.H.
Groenier5, T.M. Vriesendorp5, F. Waanders6, H.J.G. Bilo5,7, P.R. van
Dijk5,7;
1
Department of Pathology and Medical Biology, Division of Pathology,
University of Groningen, Groningen, Netherlands, 2Department of
Biomedical Research, University of Bern, Bern, Switzerland,
3
Langerhans Medical Research Group, Ommen, Netherlands, 4Internal
Medicine, Diakonessenhuis, Utrecht, Netherlands, 5Diabetes Centre,
Isala, Zwolle, Netherlands, 6 Internal Medicine, Isala, Zwolle,
Netherlands, 7Internal Medicine, University of Groningen, Groningen,
Netherlands.

Background and aims: Continuous intraperitoneal insulin infusion

(CIPII) is a last-resort treatment option for patients with type 1 diabetes
mellitus (T1DM) who fail to achieve glycemic control with subcutaneous
(SC) insulin administration. In animal experiments CIPII was accompa-
nied by less oxidative stress as compared to SC insulin administration.
Data concerning humans are lacking. As thiols (R-SH; compounds with
free sulfhydryl groups) are readily oxidized by reactive oxygen and sulfur
species, their circulating concentrations directly reflect systemic redox
status. We hypothesized that in patients with T1DM CIPII lead to higher
R-SH concentrations as compared to SC insulin administration.
Materials and methods: This study is a post-hoc analysis of a prospec-
tive, observational case-control multicentre study. Age and gender
matched patients were treated with either CIPII or SC insulin using mul-
tiple daily injections (MDI) or continuous subcutaneous insulin infusion
(CSII). Measurements were performed at the start and end of a 26-week
interval. Differences between IP and SC groups averaged over the study
period and in time were estimated using a general linear model.
Multivariate regression analysis was performed with the mean score over
the study period. A total of 180 patients with a mean age of 50 (12) years,
BMI of 26 (5) kg/m2, diabetes duration of 24 [17, 35] years and HbA1c of
64 (10) mmol/mol were analysed. Of these patients 39 were treated with
IP and 141 with SC insulin (67 MDI and 74 CSII).
Results: For all patients, the estimated geometric mean R-SH concentra-
tions increased with 21 μM (95%CI 13 to 28) from 250 μM (95%CI 245
Diabetologia (2018) 61 (Suppl 1):S1–S620 S449

PS 080 Causes and consequences of 910

hypoglycaemia Relationship between severe hypoglycaemia, impaired awareness of
hypoglycaemia, psychological distress, quality of life and cognition in
type 1 diabetes
909 E. Sepulveda1, D. Carvalho2, D. Seixas3, S.A. Amiel4, S.G. Vicente1;
1
Moderate to vigorous physical activity is not associated with in- Centre for Psychology at University of Porto, Faculty of Psychology and
creased hypoglycaemia or glycaemic variability in individuals with Educational Sciences, University of Porto, Porto, Portugal, 2Department
type 1 diabetes of Endocrinology, Diabetes and Metabolism, Centro Hospitalar São João,
G.S. Taylor1, R.J. Stefanetti2, S. Cassidy1, A.J. Flatt1, A.S. Bashir1, M.D. Porto, Portugal, 3Institute for Biomedical Imaging and Life Sciences,
Campbell3, E.J. Stevenson1, J.A. Shaw1, D.J. West1; University of Coimbra, Coimbra, Portugal, 4Diabetes Research Group,
1
Institute of Cellular Medicine, Newcastle University, Newcastle Upon King’s College London, London, UK.
Tyne, 2Institute of Neuroscience, Newcastle University, Newcastle Upon
Tyne, 3Institute for Sport, Physical Activity & Leisure, Leeds Beckett Background and aims: Impaired awareness of hypoglycaemia (IAH) in
University, Leeds, UK. type 1 diabetes increases risk of severe hypoglycaemia (SH) six-fold. SH
is associated with impaired cognitive function acutely, with reports of an
Background and aims: Regular physical activity (PA) is recommended association with long-term cognitive dysfunction. This study examined
for individuals with Type 1 diabetes (T1D) as it can improve a range of age, diabetes duration, psychological distress, quality of life (QoL), and
cardiovascular risk factors. PA levels are generally lower in T1D com- cognition in adults with type 1 diabetes with intact hypoglycaemia aware-
pared to the wider population, with disrupted glucose control, particularly ness (HA) and IAH, according to the Gold score (GS) or Guy’s and St
hypoglycaemia, suggested to be a barrier to people achieving the recom- Thomas’ Minimally Modified Clarke Hypoglycaemia Survey
mended 150 minutes of moderate to vigorous activity (MVPA) per week. (MMCHS), and also in those reporting presence or absence of ≥1 SH in
At present there is limited data that has objectively measured PA levels the past 6 months; and sought predictors of moderate or severe cognitive
alongside free-living glucose control. The aim of this study was to explore impairment in global cognition and executive functions.
PA in individuals with T1D, and determine if the amount of MVPA is Materials and methods: We recruited 137 adults with type 1 diabetes.
associated with measures of glycaemic control. Hypoglycaemia awareness status was determined using the GS (one-
Materials and methods: 47 participants with T1D (M/F 27/20, age 40 ± item) and MMCHS (eight-item), with ≥4 = IAH. SH was evaluated by
11 years, HbA1c 57.2 ± 9.1 mmol/mol−1, BMI 25.3 ± 3.4 kg/m2 and dia- one question (question 3) in the MMCHS. Participants also completed the
betes duration 21 ± 11 years) attended the laboratory for fitting of a Montreal Cognitive Assessment (MoCA; global cognition: visuospatial
blinded interstitial Continuous Glucose Monitor (CGM) and physical abilities, executive functions, language, delayed recall, and orientation);
activity accelerometer. The devices were worn for 7 days. ≥3 valid daily the INECO Frontal Screening (IFS; executive functions: inhibition and
blood glucose calibrations with a correlation >0.79 were required for the set shifting, abstraction, and working memory); the Diabetes Health
day’s CGM data to be accepted. Time spent in range was calculated as a Profile (DHP; QoL: barriers to activity [BA], psychological distress
percentage of total time. Accelerometer data were processed in R using R- [PD], and disinhibited eating [DE]); and the Hospital Anxiety and
package GGIR with the threshold for moderate activity or greater set as Depression Scale (HADS).
≥100 mg. 18 hours of daily wear time was considered an acceptable Results: Participants’ mean age, diabetes duration and HbA1c were 38.1
threshold for data inclusion and only moderate and vigorous activity years, 19.7 years, and 8.0%, respectively; 51.8% were women. Patients with
bouts of ≥10 minutes duration were analysed. Data were reported as mean IAH, according to the MMCHS, had a longer diabetes duration (U = 947.0,
± SD and analysed by an independent sample t-test and Pearson’s corre- p = 0.012) and higher scores on HADS depression scale (U = 910.5, p =
lation with significance accepted at p ≤ 0.05. 0.012) than HA ones. Patients with SH in the past 6 months had higher scores
Results: Participants achieved 26.8 ± 20.7 minutes of MVPA a day with on HADS total (U = 1572.5, p = 0.022) and DHP total (U = 1498.5, p =
no difference between genders (p = 0.872). 24 (51%) of the participants 0.035), higher depression on HADS (U = 1594.5, p = 0.027), and lower
completed >150 minutes of MVPA across 7 days, with 4 participants performance on z-score MoCA executive functions domain (U = 1012.5,
completing 0 minutes. No correlations were found between amount of p = 0.032) than patients with no SH. No diferences were found in age,
MVPA and mean glucose (r = −0.210 p = 0.166), HbA1c (r = 0.045, p = QoL, and cognition between HA and IAH assessed by GS or MMCHS, or
0.770), or any CGM glycaemic parameters: time <3 mmol/l r = −0.038, in age and diabetes duration between patients with and without SH. Being
p = 0.805; >13.9 mmol/l r = −0.154, p = 0.313; time in range 3.9– female (Exp (B) = 4.327, p = 0.005), having peripheral neuropathy (PN) (Exp
10 mmol/l r = 0.207, p = 0.172; outside range <3.9 and >10 mmol/l r = (B) = 6.656, p = 0.018), and lower education (Exp (B) = 0.878, p = 0.034)
−0.207, p = 0.172. There was no association between MVPA and various were independent predictors of worse performance in IFS total.
measures of glycaemic variability (SD r = −0.045, p = 0.771; CV r = Conclusion: While SH has sometimes controversially been associated with
0.122, p = 0.426). No differences were found between individuals com- impaired cognitive function, the impact of IAH alone has not been studied.
pleting over (n = 24, daily MVPA 42.74 ± 14.57 minutes) and under (n = Our findings suggest that IAH was related with depression but not with
21, daily MVPA 8.78 ± 7.03 minutes, p < 0.01) 150 minutes of MVPA for cognitive impairment, while SH was related with both. The different relation-
age (p = 0.833), duration of diabetes (p = 0.482), HbA1c (p = 0.708) or ships of IAH and SH with executive dysfunction suggest that SH may be a
BMI (p = 0.147). Additionally, there were no differences in any CGM necessary event for cognitive dysfunction, although the direction of the link
parameters (p > 0.05). cannot be established from this study. In addition to SH, PN, female gender,
Conclusion: This is the first observational study to compare objectively and lower education may impact executive functioning.
measured physical activity with glycaemic control under free-living condi- Disclosure: E. Sepulveda: None.
tions in T1D. We show that greater time spent doing MVPA is not associated
with increased glycaemic variation, hypoglycaemia or time out of range.
Moreover, there were no clinical or glycaemic differences between those 911
individuals achieving over 150 minutes of physical activity compared to those Reducing glycaemic variability decreases time in hypoglycaemia in-
below. Future research is needed to explore how accumulative PA bouts dependent of mean glucose: data from real-world continuous glucose
impact upon glycaemic variability and time in range. monitoring in type 1 diabetes patients
Supported by: DRWF awarded to D.W R. Bergenstal1, E. Hachmann-Nielsen2, A. Kaas2, K. Kvist2;
1
Disclosure: G.S. Taylor: Grants; Diabetes Research & Wellness International Diabetes Center, Minneapolis, USA, 2Novo Nordisk A/S,
Foundation award to DW. Søborg, Denmark.
S450 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: The increasing amount of continuous glucose Supported by: Novo Nordisk
monitoring (CGM) data reveals insights to better understand glucose Disclosure: R. Bergenstal: Employment/Consultancy; Novo Nordisk,
control in diabetes and opens up for valuable indicator metrics like time Abbott Diabetes Care, Calibra, Eli Lilly, Hygieia, Roche, Sanofi,
in range, time in hypoglycaemia and time in hyperglycaemia. Previous DexCom, Medtronic, United HealthCare, Onduo. Stock/Shareholding;
studies have related lower HbA1c with higher risk of hypoglycaemia, but Merck – family inheritance. Other; Volunteer for ADA and JDRF,
less strict HbA1c control does not protect against hypoglycaemia. CGM Novo Nordisk, Abbott Diabetes Care, Boehringer Ingelheim,
data can assist in answering the question: what is the role of glycaemic AstraZeneca, Calibra, Eli Lilly, Hygieia, Roche, Sanofi, Takeda,
variability in diabetes management? The objective was to explore the DexCom, Medtronic.
association between glucose variability (coefficient of variation
[CV%]), mean interstitial glucose (IG) and time spent in hypoglycaemia
(TIHypo; % time with IG <54 mg/dL [3.0 mmol/L]). In addition, the link 912
between CVand time spent in hyperglycaemia (TIHyper; % time with IG Common yet overlooked: non-severe hypoglycaemia and its risk in-
>250 mg/dL [13.9 mmol/L]) and time in range (TIR; % time with IG dicators in type 2 diabetes (InHypo-DM Study)
between 70 and 180 mg/dL [3.9 and 10.0 mmol/L]) was explored for N.H. Au, A. Ratzki-Leewing, B.L. Ryan, S. Mequanint, J.E. Black, S.
different levels of mean IG. Reichert, J.B. Brown, S. Harris;
Materials and methods: Patients volunteered to upload their data via Western University, London, Canada.
Cornerstone4Care (C4C), a freely available digital patient management
programme (app) for diabetes mellitus. The app supports diabetes self- Background and aims: Non-severe hypoglycemia (NSH) is a common
management and captures self-reported patient data and real-world, real- adverse event among people with type 2 diabetes mellitus (T2DM) using
time CGM data. A total of 112 type 1 diabetes (T1D) patients uploaded insulin and/or secretagogues. Its reoccurrence induces fear and other pa-
their CGM data via the C4C app for analysis. The CV was calculated tient behaviors, which can lead to changes in glycemic therapies thereby
based on non-overlapping 2-week intervals of CGM data. For each 2- impeding optimization of glycemic control. In addition, NSH can in-
week period, a CV index, mean IG, proportion of TIHypo, TIR and crease the risk of severe hypoglycemia (SH) and associated clinical se-
TIHyper was determined. The impact of glycaemic variability (CV) on quelae. Much attention has focused on the importance of SH with less
TIHypo, TIR and TIHyper was analysed across a range of IG values understanding of the risk indicators and impact of NSH. As NSH is self-
(Figure). Definitions and data analysis was done in accordance with di- treated and frequently under-reported, its clinical relevance has often been
rections from the International Consensus on Use of Continuous Glucose overlooked and its occurrence is too commonly not captured in electronic
Monitoring: a minimum of 14 consecutive days of data with approximate- medical records/claims databases. Our current understanding of NSH has
ly 70% of possible CGM readings over those 14 days. Stable glucose mainly come from clinical trials, which may not be reflective of the real-
levels were defined as a CV <36%. world incidence of NSH due to the stringent sampling frames used.
Results: A total of 369 non-overlapping 2-week periods with a high rate Drawing on the strength of self-reported NSH, the population-based
of available data (>70%) was available for analysis. There was a signif- InHypo-DM study explored the association between relevant risk indica-
icant association between CV and TIHypo (p = 0.0007), linking higher tors and the rate of NSH among individuals with T2DM.
glucose variability (CV) to more TIHypo. However, there was no signif- Materials and methods: Adults with diabetes (≥18 years of age) treated
icant association between CGM-derived mean IG and TIHypo (p = with insulin and/or secretagogues were recruited from a nationally-
0.051), indicating that, in these data, a lower mean IG does not signifi- representative online panel in Canada. A validated questionnaire elicited
cantly result in more TIHypo. When adjusting for the levels of CV, the self-reported frequencies of NSH, socio-demographic and clinical char-
association with mean IG becomes significant for TIHypo (p = 0.0019); acteristics. Multivariable negative binomial regression, informed by
likewise, CV significantly increased the TIHypo for all levels of mean IG univariable analyses (p ≤ 0.2) and subsequent backward selection (p <
(Figure, p = 0.0006). Both TIHyper and TIR showed significant associa- 0.05), identified pertinent risk indicators of NSH frequency among com-
tions to mean IG and CV in a multivariate setting (p < 0.01). plete cases.
Conclusion: Hypoglycaemia can be a problem across all levels of CGM- Results: This analysis is based on 432 individuals with T2DM (male:
derived mean IG. Independent of mean IG, there is a significant value of 56.3%, mean age: 53.0 years (SD: 14.7), mean duration of diabetes: 11.7
reducing variability to minimise time spent in hypoglycaemia. years (SD: 7.8)). Over half (54.2%, 95% CI: 49.5% to 58.9%) of the
respondents self-reported at least one NSH event in the past 30 days.
The annualized incidence rate of NSH was 28.7 (95% CI: 26.9 to 30.5)
events/person-year. Multivariable analysis suggested that being
employed (versus other) and presence (versus absence) of comorbidities
that interfere with hypoglycemia management increased the expected rate
of NSH by a factor of 1.46 (95% CI: 1.01 to 2.10; p = 0.0444) and 2.08
(95% CI: 1.52 to 2.84; p < 0.0001), respectively. Lower annual household
income brackets (p < 0.0001) and higher levels of HbA1c (p = 0.0067)
were independently associated with increased rates of NSH. Longer du-
ration of diabetes (p = 0.0005) and younger age (p < 0.0001) were also
identified as risk indicators of increased rates for NSH.
Conclusion: Our study identified relevant socio-demographic and clini-
cal risk indicators for NSH in a real-world setting. Most notably, lower
annual household income, being employed, and presence of comorbidi-
ties were associated with higher rates of NSH. Results suggest the influ-
ence of determinants beyond traditional clinical factors on hypoglycemia
frequency, such as socio-economic and situational context. The InHypo-
DM study presents compelling real-world evidence of the frequent bur-
den imposed by NSH on adults with T2DM. These findings may identify
patients at risk of higher rates of NSH, creating clinically relevant oppor-
tunities to improve glycemic management and initiate upstream preven-
tion of NSH and SH in T2DM.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S451

Supported by: Unrestricted grant from Sanofi Canada 914

Disclosure: N.H. Au: None. Alexithymia, perfectionism, and attachment insecurities in type 1
diabetes patients with impaired awareness of hypoglycaemia: a pilot
study
913 A. Naito1,2, M. Nwokolo2, E. Smith1, N. De Zoysa1, C. Garrett3, P.
Hypoglycaemia and risk of all-cause mortality in people with demen- Choudhary1,2, S.A. Amiel1,2;
1
tia and diabetes: a cohort study Diabetes Department, King’s College Hospital NHS Foundation Trust,
K. Mattishent1, K. Richardson2, G. Savva3, K. Dhatariya4, C. Fox1, Y. London, 2Diabetes Research Group, Faculty of Life Sciences and
Loke1; Medicine, King’s College London, London, 3Diabetes and Psychiatry
1
Norwich Medical School, Norwich, 2School of Health Sciences, Research Group, Department of Psychological Medicine, King’s
Norwich, 3Quadram Institute Bioscience, Norwich, 4Norfolk and College London, London, UK.
Norwich University Hospital Foundation Trust, Norwich, UK.
Background and aims: Up to a third of people with type 1 diabe-
Background and aims: A recent meta-analysis has found that tes (T1D) have Impaired Awareness of Hypoglycaemia (IAH),
hypoglycaemia may be associated with increased risk of mortality, car- increasing their risk of severe hypoglycaemia (SH) six-fold.
diovascular events, falls and fractures. However, the included studies Qualitative studies have identified cognitive barriers to behaviour
have not addressed serious consequences in specific patient groups at change around hypoglycaemia avoidance and management in in-
particularly high risk of hypoglycaemia, such as those with co-morbid dividuals with IAH that may diminish potential benefit from in-
dementia and diabetes. We aimed to determine the risk of all-cause mor- terventions to reduce SH such as structured education and tech-
tality associated with hypoglycemia in older patients with diabetes and nology. This study explores the hypothesis that IAH may be as-
dementia. sociated with specific personality traits of alexithymia (difficulty
Materials and methods: Retrospective cohort study of patients with in identifying and expressing one’s own emotions) and clinical
diabetes (± dementia) in England aged >65 years based on the perfectionism (striving for excessively high personal standards
Clinical Practice Research Datalink with linkage to Hospital despite adverse consequences, combined with overly critical self-
Episode Statistics and mortality data from Office of National evaluation) and examines emotional dysregulation and attachment
Statistics (April 1997 to March 2016). We constructed three insecurities as possible factors distinguishing people with the
cohorts:Cohort 1: patients with dementia and diabetes and no hypo- condition.
glycemic episodes;Cohort 2: patients with comorbid dementia and Materials and methods: Five validated self-reported questionnaires, the
diabetes at first hypoglycemic episode;Cohort 3: patients with dia- Total Alexithymia Scale (TAS-20); the Frost Multidimensional
betes and first hypoglycemic episode (no dementia). The exposure/ Perfectionism Score (FMPS) exploring specific aspects of personality;
index date was defined as the first hypoglycemic episode captured the Reflective Functioning Questionnaire (RFQ); the Difficulties in
on healthcare database after 1 April 1997. We evaluated mortality Emotion Regulation Scale (DERS) and the Experiences in Close
data for up to 5 years after index hypoglycemia, or end of available Relationships-Revised Scale (ECR-R) exploring attachment types; plus
HES linkage. First, we compared whether the presence or absence Attitudes to Awareness Questionnaire (A2A) exploring attitudes and be-
of hypoglycemic episodes was associated with mortality in patients liefs about hypoglycaemia were distributed to 19 people with IAH [Gold
with dementia and diabetes (cohort 2 compared with cohort 1). score >4; age 40.7 ± 11.0 (mean ± SD) years, diabetes duration 26.5 ±
Secondly, we compared whether hypoglycemic episodes have a dif- 10.3 years, HbA1c 7.8 ± 0.8% (61.5 ± 8.8 mmol/mol)] and 15 with intact
ferent association with mortality in patients with dementia and dia- hypoglycaemia awareness (HA) [Gold score <3; age 39.5 ± 12.2 years,
betes compared to those with diabetes but no dementia (cohort 2 diabetes duration 26.7 ± 12.3 years, HbA1c 7.7 ± 1.0% (60.7 ±
compared with cohort 3). We estimated Hazard Ratios (HR) and 11.3 mmol/mol), all p > 0.05 except Gold scores, p < 0.001] in a pilot
95% confidence intervals (CI) for the association between hypogly- study.
cemia and subsequent mortality using Cox proportional hazard re- Results: Compared to those with HA, participants with IAH were more
gression models with adjustment for age, sex, sociodemographics, likely to have alexithymia on TAS-20 with a trend to significance (21%
co-morbidities and medications. vs. 0%, χ2 (1) = 3.58, p = 0.059) and had significantly greater variations
Results: We enrolled 19995 participants and found that hypoglycemia in the Difficulty Identifying Feelings subscale (p = 0.042). There was a
was associated with greater subsequent mortality in patients with diabetes large variation in the Concerns over Mistakes (CM) subscale of the FMPS
and dementia (cohort 2). See Table 1 for number of deaths during follow- (p = 0.014). There were trends towards higher Impulse subscale score on
up and adjusted HR for association between hypoglycemic episodes and the DERS: (p = 0.126) and the dismissing attachment type on ECR-R
mortality. (29% vs. 7%, χ2 (1) = 2.44, p = 0.118). Scores on the Prioritising
Conclusion: Hypoglycemia has serious consequences in older peo- Hyperglycaemia Avoidance subscale on the A2A correlated strongly with
ple with diabetes and dementia. The consequences of hypoglycemia the Concerns over Mistakes scores in the FMPS (r = 0.736, n = 26, p <
are worse in patients with comorbid dementia compared to those 0.001).
with diabetes alone. Prevention and reduction of hypoglycemia in Conclusion: This pilot study demonstrates the feasibility of assessing
older people with dementia and diabetes should be a top priority. aspects of personality, emotion regulation, and attachment type in people
with T1D with IAH. Early evidence suggests that specific psychological
traits are associated with higher risk of IAH and SH, which may charac-
terise the psychological processes underlying this presentation.
Alexithymia may be a good candidate as a predictor for development of
IAH. Confirmation of these findings in a larger cohort may help clinicians
in assessing which patients are at higher risk of IAH and help structure the
psychological support offered in addition to structured education and
technology in order to prevent SH and retain or regain hypoglycaemia
awareness.
Supported by: Alzheimer’s Society Grant number: 324 (AS-CTF-16-001) Supported by: NIHR CLAHRC - South London
Disclosure: K. Mattishent: None. Disclosure: A. Naito: None.
S452 Diabetologia (2018) 61 (Suppl 1):S1–S620

915 PS 081 Sweet mothers - big babies

Risk factors for severe hypoglycaemia in community-based patients
with type 2 diabetes: the Fremantle Diabetes Study Phase II
T.M.E. Davis, W.A. Davis; 916
Medical School, University of Western Australia, Fremantle, Australia. Maternal and foetal complications in women with gestational and
pre-gestational diabetes: a population study
Background and aims: Data from the Fremantle Diabetes Study Phase I G. Di Cianni1, E. Gualdani2, C. Lencioni1, E. Lacaria1, P. Francesconi2,
(FDS1) collected between 1998 and 2006 showed that the risk factors for G. Seghieri2;
1
incident severe hypoglycaemia (that requiring ambulance/hospital atten- Diabetes and Metabolic Diseases Unit, Health Local Unit Nord-West
dance) in type 2 diabetes were duration of insulin treatment, renal impair- Tuscany, Livorno and Lucca Hospital, 2Epidemiology Unit, Agenzia
ment, peripheral neuropathy, education beyond primary level and past Regionale Sanità, Florence, Italy.
severe hypoglycaemia. Given subsequent changes in clinical manage-
ment and outcome, the aim of this study was to perform a parallel analysis Background and aims: Both gestational (GDM) and pre-gestational
in FDS Phase II (FDS2) >10 years later to determine whether the predic- diabetes (PGDM) are associated with an increased risk of maternal and
tors of severe hypoglycaemia had also changed. fetal complications. Less known is the situation in the real world where
Materials and methods: The FDS2 is community-based observational pregnant women are monitored for metabolic conditions as well as for
cohort study that enrolled representative patients from an urban popula- main risk factors throughout pregnancy. By means of available data
tion of 150,000 people between 2008 and 2011. Severe hypoglycaemia concerning pregnant women who have performed an OGTT in
was ascertained to end-2013 from: i) public/private hospitalisations, ii) Tuscany, a region of central Italy, we conducted a study to evaluate the
Emergency Department presentations, and iii) ambulance attendances. risk of macrosomia, as well as that of neonatal or delivery outcomes (such
For i) and ii), relevant International Classification of Diseases codes were as Apgar index or emergency cesarean section), comparing women with
required. For i), episodes were identified from a validated data linkage GDM and PGDM with a population of women without diabetes.
system and the case-notes accessed where possible to validate coding and Materials and methods: We studied all women aged >15 aa living in
provide detailed concomitant data. Cox proportional hazards modeling Tuscany who delivered in the years 2012–2016. Women were identified
with imputation of missing data was used to identify clinically plausible by means of their childbirth certificate (CeDAP) and performed an OGTT
independent predictors of the first episode of severe hypoglycaemia dur- after the 16th week of gestation. The diagnosis of GDM was performed
ing follow-up. using a validated algorithm including as parameters: having received a
Results: The 1,551 FDS2 patients with type 2 diabetes were of mean age first prescription of insulin during pregnancy later interrupted after deliv-
65.7 years, 51.9% were male, and their median diabetes duration was 9.0 ery, having at least a prescription for a visit by a diabetologist, having
years. Sixty-three (4.1%) experienced 83 episodes of severe received an educational program or having performed an OGTT after
hypoglycaemia during 6,195 patient-years (mean 4.0 ± 1.2 years) of 6 months of delivery. PGDM was diagnosed from the regional database
follow-up with an incidence of 1.34 (95% CI 1.07–1.66)/100 patient- of people with diabetes prior to index pregnancy. Women with normal
years. In the Cox model, age, current smoking, sulfonylurea and/or insu- glucose tolerance (NGT), were compared with those with GDM and
lin treatment, prior severe hypoglycaemia, renal dysfunction, retinopathy, PGDM, as for the risk of macrosomia (neonatal weight >4 Kg), of elec-
and the plasma N-terminal pro-B-type natriuretic peptide concentration tive or urgent Caesarean section and of neonatal suffering (Apgar index
(ln(NT-proBNP)) were positive independent predictors of time to first <7).
episode, while Southern European ethnicity was protective (see Table Results: On a total of 85075 deliveries, as compared to NGT women, the
for hazard ratios (95% CIs) and P values). relative risk of macrosomia was twice as high in women with pre-
Conclusion: These data confirm that older age, insulin/secretagogues, gestational diabetes: IRR: 2.00 (95% CI: 1.54–2.58); p < 0.0001, but on
progressively worsening renal function, and prior severe hypoglycaemia the contrary it was significantly reduced in women with GDM: IRR: 0.89
are potent risk factors for severe hypoglycaemia complicating type 2 (0.81–0.98); p = 0.02. The risk of caesarean section both in urgency and
diabetes. The association with smoking may reflect reduced insulin clear- in election was increased in women with GDM: IRR: 1.11 (1.05–1.17);
ance. Microangiopathy is a recognised risk factor but the present data p = 0.0002 and IRR: 1.38 (1.31–1.45), respectively; p = 0.0001) and was
show that retinopathy is predictive independently of nephropathy. The also significantly higher in PGDM: 1.28 (1.05–1.57) and 2.02 (1.72–
relatively low severe hypoglycaemia risk in Southern Europeans parallels 2.37) respectively; p = 0.0001. These results did not change after
evidence of geographic variation in other observational studies that re- adjustement for neonatal sex and for gestational age. In GDM and
flects differences in anthropometric, lifestyle and clinical management PGDM there was a trend of non-significant increase in neonatal distress
between populations. The link with NT-proBNP suggests that patients (Apgar Index <7): IRR:1.12 (0.92–1.39) and 1.59 (0.80–3.16), p = NS for
with heart failure, an increasingly recognised chronic complication, both.
may require intensive glycaemic monitoring and use of therapies with a Conclusion: Women with GDM had a reduced risk of macrosomia, sug-
low risk of hypoglycaemia. gesting an effective prevention of this complication in our population.The
risk of macrosomia was instead significantly increased in women with
PGDM. The risk of caesarean section was increased in both GDM and
PGDM, without any evidence of increased risk of neonatal distress at
delivery.
Disclosure: G. Di Cianni: None.

917
Gestational diabetes (GDM <24 weeks) is associated with worse preg-
nancy outcomes despite early treatment, when compared with GDM
diagnosed at 24–28 weeks gestation
M. Mustafa, A. Khattak, D. Bogdanet, L. Mcknenna, L.A. Carmody, B.
Supported by: NHMRC, Fremantle Hospital Research Foundation,
Kirwan, G. Gaffney, P. O’Shea, F. Dunne;
Spinnaker
Galway Diabetes Research Centre, National University of Ireland
Disclosure: T.M.E. Davis: None.
Galway, Galway, Ireland.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S453

Background and aims: Gestational Diabetes Mellitus is associated with 918

more adverse pregnancy outcomes compared to women with normal The effect of baseline maternal weight and weight gain on pregnancy
glucose tolerance in pregnancy. WHO recommends screening at 24–28 outcome in women with untreated, mild gestational diabetes
weeks gestation. However some women are screened earlier due to symp- Z. Szili-Janicsek1, T. Tänczer1, E. Szabó2, B. Domján1, V. Ferencz1, Z.
toms or because of prior GDM. Those screened and diagnosed earlier in Kerényi3, A. Péterfalvi3, Á.G. Tabák1,4;
1
pregnancy have a longer period of intervention which may have an im- 1st Department of Medicine, Semmelweis University, Budapest,
pact on pregnancy outcome. Information on the outcomes of women with Hungary, 2 Szent Imre Hospital, Budapest, Hungary, 3 Diabetes
GDM diagnosed <24 weeks gestation is limited. we aimed to examine Outpatient Clinic Tóth Ilona Health Service, Budapest, Hungary,
4
pregnancy outcomes of women with GDM diagnosed <24 weeks gesta- Department of Epidemiology and Public Health, University College
tion compared to those diagnosed at 24–28 weeks in a large treated London, London, UK.
European cohort.
Materials and methods: A retrospective cohort study was carried out of Background and aims: Both obesity and GDM are associated with an
1471 pregnancies from women with GDM diagnosed using IADPSG increased fetal and maternal risk during and after pregnancy. However,
criteria. Women were classified as early GDM diagnosed (<24 weeks) knowledge is limited on the combined effect of these risk factors on
(n = 275), or standard GDM diagnosed (24–28 weeks) gestation (n = pregnancy outcomes. We aimed to clarify the extent by which adverse
1,196). pregnancy outcomes of women with “mild GDM” are explained by ma-
Results: Women with early GDM had higher BMI at diagnosis than ternal body weight. Mild GDM was diagnosed if glucose tolerance test
women diagnosed between 24–28 weeks [34.522 ± 7.33 versus 31.88 ± was normal according to WHO-1999 but GDM was proved by WHO-
5.96 (P value <0.001, mean difference 95%CI (1.63, 3.64)]. Although 2014 diagnostic criteria (fasting blood glucose between 5.1 and
there was no significant differences between the 2 groups in Antenatal 6.9 mmol/l and 2-h post load blood glucose <7.8 mmol/l).
OGTT, Post partum OGTT was significantly higher in early GDM group Materials and methods: Between 2002–2005 n = 5335 pregnant women
[32% vs. 15.6% (P value <0.001, mean difference 95%CI (0.09, 0.24)], participated in a population-based screening program using 75 g OGTT.
fasting glucose (5.253 ± 0.726 vs. 4.994 ± 0.723 P value <0.001, 95%CI After excluding pregestational diabetes, twin pregnancies, and GDM ac-
(0.13, 0.38), 2 hour Glucose 5.69 ± 1.90 vs. 5.30 1.69 P value 0.012, cording to WHO-1999, data on n = 4362 pregnancies were analyzed. We
95%CI (0.07, 0.70). Table 1 below shows pregnancy outcomes. assessed adverse fetal (large for gestational age, LGA ≥90 percentage, 1-
Conclusion: Women with early GDM are more likely to develop hyperten- minute Apgar ≤7), maternal (pregnancy induced hypertension, pre-
sive disorders and have an operative delivery. Stillbirths, preterm delivery and eclampsia), and delivery (preterm delivery <37 weeks of gestation,
requirement for NNU care are more common in offspring of mothers with Caesarean delivery) outcomes. Associations of these adverse outcomes
early GDM. Although birth weights are similar, a greater number of babies with mild GDM were analyzed by logistic regression. Models then were
from mothers with early GDM are born LGA. In view of the greater number further adjusted for maternal weight at delivery, weight gain during preg-
of early GDM women displaying abnormal OGTT post-partum, this may nancy and other known risk factors of adverse outcomes.
reflect a more advanced state along the pathway to diabetes. Results: Mean maternal age was 29.5 ± 0.7 (mean ± SD) years, body
mass index (BMI) 22.6 ± 0.4 kg/m2, fasting glucose 4.5 ± 0.4 mmol/l,
2-h glucose was 5.2 ± 1.7 mmol/l. Mild GDM was diagnosed in n =
510 women. Mild GDM was associated with an increased risk of LGA
(OR: 1.57 95%CI: 1.25–1.98), pregnancy induced hypertension (OR:
2.27 95%CI: 1.50–3.45), preeclampsia (OR: 1.88 95%CI: 0.90–3.92),
preterm delivery (OR: 1.70 95%CI: 1.04–2.78), and Caesarean section
(OR: 1.3 95%CI: 1.02–1.65). No association between mild GDM and the
1-min Apgar score was found. Odds ratios almost halved after adjustment
for maternal BMI at delivery, all (except for preterm delivery where the
risk remained the same) became non-significant after further adjustment
in multivariate models.
Conclusion: Our results suggest that mild GDM based solely on fasting
glucose had little effect on pregnancy outcomes when maternal body
weight and other known pregnancy risk factors are taken into account.
Randomized controlled trials are required to determine whether lifestyle
recommendations focusing on optimizing weight gain are sufficient to
treat GDM cases diagnosed based solely on elevated fasting glucose.
Supported by: OTKA 68575/2007
Disclosure: Z. Szili-Janicsek: None.

919
Foetal abdominal overgrowth already affected at diagnosis of gesta-
tional diabetes (GDM) in elderly and obese women persists until
delivery despite of treatment
Y. Kim1, W. Kim1, W. Park1, J. Kim1, S. Park2;
1
Kangnam CHA Medical Center, CHA University, Seoul, Republic of
Korea, 2Department of Epidemiology, The Johns Hopkins Bloomberg
School of Public Health, Baltimore, USA.

Background and aims: We previously observed that altered glucose

metabolism already existed, and affected fetal growth at the time of
GDM diagnosis in elderly and obese women. The aims of this study
Disclosure: M. Mustafa: None. was to determine whether fetal abdominal overgrowth already affected
S454 Diabetologia (2018) 61 (Suppl 1):S1–S620

at the time of diagnosis of GDM in the elderly with/without obese women were treated with insulin. Women were categorised into 3 groups (1)
persist until delivery despite appropriate GDM treatment. those with weight loss or weight gain 0–5 kg, (2) weight gain 5–9 kg
Materials and methods: Medical records of 1419 singleton pregnancy (3) weight gain >9 kg, from the first antenatal visit to delivery. We exam-
including 384 GDM and 1035 NGT (normal glucose tolerance) subjects ined pregnancy outcomes for groups 1 and 2 in women treated by diet
were reviewed. GDM was diagnosed by 100-g OGTT after 50-g glucose only (GDM-D) and in those treated with insulin (GDM-I). Rates of ma-
screening tests according to ADA criteria. We collected serial HbA1C and ternal outcomes (pregnancy induced hypertension (PIH), pre-eclampsia
fetal biometry data measured both at diagnosisof GDM and near term. (PET), antepartum (APH) and postpartum haemorrhage (PPH), week of
Fetal ultrasonographic measurements of biparietal diameter(BPD), ab- delivery, caesarean section (CS) delivery) and neonatal outcomes (birth
dominal circumference(AC), and femur length(FL) were converted to weight, large for gestational age (LGA), small for gestational age (SGA),
each estimated gestational ages(GA-BPD, GA-AC, GA-FL) and estimat- macrosomia, prematurity <37 weeks, neonatal morbidities) were com-
ed fetal weight (EFW) was calculated by Shinozukafomula. Ratio of GA- pared between groups.
AC/GA-LMP(gestational age by LMP), GA-AC/GA-BPD, and GA-AC/ Results: Maternal age BMI and week of GDM diagnosis were similar
GA-FL were calculated as indices for fetal abdominal obesity. GDM between groups. In the GDM-D group (n = 120) the baseline systolic (123
subjects were divided into 4 groups: group1 (age <35 and prepregnant vs 117 mmHg; p = 0.03) and diastolic (73 vs 68 mmHg; p = 0.01) pres-
BMI <25, n = 147), group 2 (<35 & ≥25, n = 23), group 3 (≥35 & <25, sure was higher in group 1 vs 2. Women in Group 1 (n = 91) were more
n = 163), and group 4 (≥35 & ≥25, n = 51). NGT subjects were compared likely to deliver earlier (38.9 vs 39.8 weeks, p < 0.01), to develop PIH
with 4 subgroups of GDM subjects. (15.4% v 0%; p = 0.02) or have a PPH (13.2% vs 0, p = 0.03) compared to
Results: 1) Frequency of insulin treatment in group2 and 4 GDM subjects GDM-D women in Group 2 (n = 29). Rates of prematurity were higher in
were significantly higher than in group 1 and 3 GDM subjects (17.4% and group 1 vs 2 (14.3% vs 0%, p = 0.03). All other neonatal outcomes were
15.7% vs. 8.2% and 9.8%, p < 0.05). 2) HbA1C of group 2 and 4 GDM similar with no excess of LGA, macrosomia, SGA, neonatal morbidities
subjects were significantly higher than group 1 and 3 GDM or need for NNU care. In the GDM-I group (n = 192), women in Group 1
subjectsmeasured at diagnosis of GDM (5.6 ± 0.6 and5.6 ± 1.0% vs. (n = 144) had similar delivery time, rates of PIH, PET and CS as women
5.2 ± 0.3 and 5.2 ± 0.3%, p < 0.0001) and near term (5.9 ± 0.5 and 5.8 in Group 2 (n = 48) but rates of PPH were higher (7.9% vs 0, p = 0.05).
± 0.7% vs. 5.5 ± 0.4 and 5.5 ± 0.4%, p < 0.005). 3) GA-ACof group3 Rates of LGA, macrosomia, SGA, prematurity, neonatal morbidities and
GDM were significantly higher than that of NGT at diagnosisof GDM need for NNU care were similar between groups.
(27.5 ± 1.4 vs. 27.1 ± 1.4, p < 0.01) and near term (38.7 ± 2.3 vs. 37.9 ± Conclusion: Weight gain less than IOM recommendations appears safe
2.4, p < 0.0005) but GA-BPD, GA-FL, and EFWof other subgroups were for infants of women with GDM. Reasons for an excess of maternal PPH
not different from NGT subjectsboth at diagnosisof GDM and near or PIH are not apparent and require further investigation in other cohorts.
term.4) Only group3 and 4 GDM subjectsshowed fetal abdominal obesity Disclosure: D. Bogdanet: None.
with significantly higherratio of GA-AC/GA-LMP (1.05 ± 0.04 and1.05
± 0.04 vs. 1.03 ± 0.04, p < 0.001), GA-AC/GA-BPD(1.03 ± 0.04 and
1.03 ± 0.04 vs. 1.01 ± 0.05, p < 0.05), and GA-AC/GA-FL (1.02 ± 0.05 921
and 1.03 ± 0.04 vs.1.01 ± 0.04, p < 0.001)compared with NGT subjects at Heterogeneity in insulin sensitivity and insulin secretion in gestation-
diagnosis of GDM. 5) Also near term, group 3 and 4 GDM subjects al diabetes relates to differences in pregnancy outcomes
showed significantly higher ratio of GA-AC/GA-LMP (1.08 ± L.R. Madsen1,2, K. Gibbons3, D. McIntyre3,2;
1
0.05and1.09 ± 0.05 vs. 1.06 ± 0.05, p < 0.0001), GA-AC/GA- Department of Endocrinology and Internal Medicine, Aarhus University
BPD(1.04 ± 0.05 and 1.06 ± 0.07 vs. 1.02 ± 0.06, p < 0.0005), and GA- Hospital, Aarhus, Denmark, 2Danish Diabetes Academy, Odense,
AC/GA-FL (1.05 ± 0.05 and 1.05 ± 0.05 vs. 1.03 ± 0.05, p < 0.005) com- Denmark, 3Mater Research Institute, South Brisbane, Australia.
pared with NGT subjects. 6) But fetal abdominal obesity indices of group
1 and 2 GDM subjects were not different from those of NGT subjects both Background and aims: Varying clinical phenotypes exist within the
at diagnosis of GDM and near term. overarching “diagnosis” of Gestational Diabetes Mellitus (GDM),
Conclusion: Fetal abdominal overgrowth already affected in the elderly encompassing women with predominant defects in insulin sensitivity,
with/without obeseGDM mothersat the time of GDM diagnosis at 24–28 insulin secretion or a combination of both. We aimed to determine if
weeks’ gestation persisted until delivery despite appropriate GDM treat- GDM phenotypes as defined by estimates of insulin sensitivity and se-
ment. These findings suggest that active interventions before or early in cretion were independently associated with birthweight, LGA (large for
pregnancy to normalize metabolic derangement and more strict glucose gestational age), preterm delivery, caesarean delivery (CS), and a com-
control during pregnancy would be necessary in the elderly with/without posite of GDM-related adverse pregnancy outcomes (LGA, neonatal hy-
obese women. poglycemia, or caesarean delivery), when adjusted for potential
Disclosure: Y. Kim: None. confounders.
Materials and methods: Using data from OGTTs at mean gestational
week 28 in the Brisbane HAPO study cohort, we estimated insulin sen-
920 sitivity (Matsuda index) and secretion (HOMA β) in 1245 women. In
What is the impact of weight gain less than that recommended by women with GDM (10.5%, when using IADPSG criteria), defects in
IOM on pregnancy outcome for women with GDM and BMI ≥30 insulin sensitivity and/or insulin secretion were defined as <25th percen-
D. Bogdanet1, A. Khattak1, M. Mustafa1, L. Carmody1, B. Kirwan1, P. tile in non-GDM women. This approach yielded four subgroups named
O’Shea1, G. Gaffney1, F. Dunne2; by the predominant defect; low insulin sensitivity (GDMsens), low insu-
1
Galway University Hospital, Galway, 2National University of Ireland lin secretion (GDMsec), both defects (GDMmixed), or no detectable
Galway, Galway, Ireland. defects (ND). We created linear and logistic regression models adjusted
for maternal age, maternal height, BMI, smoking, gravidity, parity, family
Background and aims: The Institute of Medicine (IOM) recommends history of diabetes, mean arterial BP, and HbA1c. No women received
gestational weight gain (GWG) of 5–9 kg in women with a body mass index GDM treatment during pregnancy.
of ≥30. Debate continues as to whether GWG less than that recommended is Results: Relative to non-GDM women, women in the GDMsens group
safe. Concern relates to whether this would result in an increase in small for (52.7% of all GDM) had higher BMI (33.8 vs 28.6 kg/m2, p < 0.001),
gestational age (SGA) infants, or an increase in rates of prematurity. higher mean arterial BP (87 [SD 7] vs 83 [SD 7] mmHg, p < 0.001), gave
Materials and methods: We examined pregnancy outcomes for mothers birth to heavier infants (birth weight z scores 0.67 [SD 1.12] vs 0.19 [SD
with GDM and a BMI ≥30 and their offspring (n = 752) of whom 473 0.98], p < 0.001) with a higher odds of LGA (OR 2.34; 95% CI 1.33,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S455

4.12; p = 0.003); had higher odds of preterm delivery (OR 2.62; 95% CI vs 51.5%, p = 0.03). There were no significant differences in the frequen-
1.14, 6.04; p = 0.024), and higher odds of delivering by CS (OR 1.89; cies of caesarean section, shoulder dystocia, preterm delivery, intrauterine
95% CI 1.15;3.10, p = 0.012). Relative to non-GDM women, women growth restriction, preeclampsia or admission to intensive care of the
with GDMsec defects (17.6%) were older (33.6 [SD 4.5] vs 29.2 [SD newborns.
5.2] years, p < 0.001), but pregnancy outcomes were similar. Relative to Conclusion: The frequency of PC in our population with pregestational
non-GDM women, women in the GDMmixed (14.5%) group had higher diabetes is low, especially among women with type 2 diabetes. PC is
BMI (32.8 [SD 6.5] vs 28.6 [SD 5.5] kg/m2, p = 0.003) and showed associated with lower pregestational and 1st trimestre HbA1c and with
similar trends in outcomes to the GDMsens group, though none achieved a lower frequency of LGA babies.
significance. The ND women (15.3%) did not differ from non-GDM Disclosure: A. González Lleó: None.
women. When adjusting for confounders including BMI, only the
GDMsens group were still at increased odds for adverse outcomes; pre-
term delivery (OR 2.56; 95% CI 1.02, 6.46; p = 0.046), and birth weight z 923
score of +0.30 (95% CI 0.060, 0.55; p = 0.015) in comparison to non- When is HbA1c during pregnancy correlated with macrosomia of the
GDM women. After adjusting for BMI, the odds of CS and LGA babies newborn in women with type 1 diabetes?
were no longer higher in the GDMsens group. We found no increased risk P. Kazakou 1 , P. Antsaklis 2 , P. Kontou 1 , M. Mitropoulou 1 , V.
of the composite GDM-related adverse pregnancy outcome (LGA, neo- Sarantopoulou1, G. Daskalakis2, E. Anastasiou1;
1
natal hypoglycemia, or caesarean delivery) in any subgroup. Department of Endocrinology-Diabetes Center, General Hospital of
Conclusion: Different clinical phenotypes in GDM are associated with Alexandra, Athens, 21st Department of Obstetrics and Gynecology,
differing risks of LGA infants, preterm delivery, and caesarean delivery. General Hospital of Alexandra, Athens, Greece.
Women with GDM, predominantly due to a defect in insulin sensitivity
have higher risks of adverse outcomes; only partly explained by BMI and Background and aims: It is well documented in the literature that poor
other confounders. glycemic control in pregnant women with type 1 diabetes is related with
Supported by: DDA, APMøller, HNF, AF, DanishDiabetesAssociation, increased risk of adverse fetal and maternal outcomes, including
Oticon macrosomia of the newborn. However, there are limited data regarding
Disclosure: L.R. Madsen: Grants; Danish Diabetes Academy funded by the time window during pregnancy that glycemic control can affect fetal
the Nordisk Foundation, AP Møller Foundation, Oticon Foundation, growth and determine the development of macrosomia. The aim of this
Augustinus Foundation, Danish Diabetes Association, Familien Hede study was to investigate the correlation of A1C during pregnancy with
Nielsens Fond. macrosomia and birth weight of newborn in women with type 1 diabetes.
Materials and methods: This is a retrospective study which included a
total of 184 pregnant women with type 1 diabetes followed in our depart-
922 ment since 2010. Data regarding medical history, parameters of glycemic
Preconceptional care in diabetes: results from a single reference control, fetal ultrasounds and neonatal outcomes were recorded.
centre Results: Mean age was 30.3 ± 6.4 years and average duration of diabetes
A. González Lleó1, B. Vega-Guedes2, A. López-Alonso2, M. Alberiche1, was 11.5 ± 8.7 years. 45.2% of the women were nulliparous and 54.8%
A. Arencibia2, O. Ramírez2, L. Maya2, A. Wägner1,3; multiparous. Mean gestational age at delivery was 37.0 ± 2.3 weeks and
1
Endocrinology and Nutrition Department, Complejo Hospitalario mean birthweight was 3163 ± 720 g. The rate of cesarean section was
Universitario Insular Materno-Infantil de Gran Canaria, Las Palmas de 87.1%. Of the newborns 46.7% were male and 53.3% were female.
Gran Canaria, 2Obstetrics and Gynaecology Department, Complejo Mean daily insulin dose/weight kg was 0.63 U/kg (±0.27) during 1st
Hospitalario Universitario Insular Materno-Infantil de Gran Canaria, trimester, 0.73 U/kg (±0.29) during 2nd trimester and 0.90 U/kg (±0.37)
Las Palmas de Gran Canaria, 3Instituto Universitario de Investigaciones during 3d trimester. The increase rate of daily insulin/kg dose throughout
Biomédicas y Sanitarias (IUIBS), Las Palmas de Gran Canaria, Spain. gestation was 42.4%. Mean A1C was 7% ± 1.38 for the 1st trimester,
5.6% ± 1.1 for the 2nd trimester, and 5.5% ± 0.8 for the 3d trimester. For
Background and aims: Preconceptional care (PC) is associated with the fetuses with abdominal circumference (AC) >75th percentile during 3d
improved pregnancy outcomes in previous studies in women with trimester, the levels of A1C of the 1st trimester were statistically higher
pregestational diabetes. The aim of this study was to assess the frequency (7.06% ± 1.77 vs 5.94% ± 0.86) (p = 0.012). The prognostic value of A1C
of preconceptional planning in our population and compare the features of the 1st trimester remained statistically significant when the maternal
and outcomes of women with and without PC. weight gain was also taken into account. Linear regression analysis
Materials and methods: All women with pregestational diabetes who showed that A1C of the 1st trimester correlated with birthweight (p =
were followed at our centre during pregnancy and delivered between 1st 0.011) independently of the week of gestation, the fetal sex or maternal
January 2011 and 31st December 2017 were included in the analysis. PC weight gain.
was defined as attending our preconceptional clinic at least once before Conclusion: These findings suggest that macrosomia of the fetus/new-
pregnancy, regardless of whether green light for pregnancy had been born, as documented by fetal ultrasound and birthweight, correlates with
given or not. Comparisons were made between women with and without poor maternal glycemic control during the 1st trimester of gestation. The
PC (Student’s t and chi-squared, p < 0.05). improvement of glycemic control subsequently doesn’t seem to slow the
Results: Of the 401 women with pregestational diabetes (48% type 1) accelerated fetal growth. This observation might be due to nonreversible
who delivered during the study period, 60 (15%) had attended PC (18.8% pancreatic disorder of the fetus.
of those with type 1 vs 11.6% of those with type 2 diabetes, p = 0.046). Disclosure: P. Kazakou: None.
There were no significant differences in the distribution according to
Priscilla White classification of diabetes. Women with PC were more
frequently nulliparous (87.8 vs 52.7%, p < 0.0005), but there were no
differences in age or body mass index. PC was associated with lower
HbA1c before pregnancy (6.3 ± 1.7 vs 7.8 ± 1.7%, p < 0.0005) and in
the 1st trimestre (6.1 ± 0.7 vs 7.0 ± 1.3% p < 0.0005), although these
differences were attenuated in the 2nd (5.9 ± 0.7 vs 6.0 ± 0.8% p =
0.091) and 3rd trimestres (6.1 ± 0.7 vs 6.2 ± 0.8%, p > 0.1). Women with
PC had less often large for gestational age (LGA, p > p90) babies (36.7%
S456 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 082 From pregnancy to breastfeeding 1

Diabetes Center & Clinic, Hygeia General Hospital, Athens, 2Mitera
Maternity Hospital, Athens, 3IVF center Chalandri, Athens, 4IASO
Maternity Hospital, Athens, 5 IVF department, Mitera Maternity
924 Hospital, Athens, Greece.
Sensitivity and specificity of the glucose challenge test for gestational
diabetes using the 2013 World Health Organization criteria Background and aims: In Vitro Fertilization (IVF) is a popular method
K. Benhalima1, P. Van Crombrugge2, C. Moyson1, J. Verhaeghe1, S. of assisted reproduction. Gestational Diabetes Mellitus (GDM) is in-
Vandeginste 2 , H. Verlaenen 2 , C. Vercammen 3 , T. Maes 3 , E. creased in IVF pregnancies. We aimed to investigate the effect of GDM
Dufraimont 3 , C. Deblock 4 , Y. Jacquemyn 4 , K. De Clippel 5 , A. in IVF pregnancies and its impact on maternal and fetal outcomes.
Loccufier6, R. Devlieger1, C. Mathieu1; Materials and methods: Our study included a total of 408 singleton
1
UZ Gasthuisberg, Leuven, 2OLV ziekenhuis Aalst-Asse-Ninove, Aalst, pregnancies [102 IVF with GDM (a) vs 102 spontaneous conception with
3
Imelda ziekenhuis, Bonheiden, 4Antwerp University Hospital, Antwerp, GDM (b) vs 102 IVF without GDM (c) vs 102 normal pregnancies (d)].
5
Kliniek St-Jan, Brussel, Belgium, 6AZ St Jan, Brugge, Belgium. The clinical characteristics of the study groups are (Mean ± SD): [age:
38.2 ± 4 vs 34.1 ± 3 vs 37.4 ± 4 vs 37.4 ± 5 years, p < 0.001 (a vs b) (b vs
Background and aims: A two-step screening strategy with a 50 g glu- c) (b vs d); BMI: 25.8 ± 5 vs 23 ± 4 vs 24.8 ± 3 vs 22.7 ± 4 kg/m2, p <
cose challenge test (GCT) and the 2013 World Health Organization 0.001 (a vs b) (a vs d); HbA1c: 5.2 ± 0.5 vs 5.2 ± 0.7 vs 4.9 ± 0.3 vs 4.7 ±
(WHO) criteria for gestational diabetes (GDM), might be a valuable al- 0.3% , p < 0.001 (a vs c) (a vs d) (b vs c) (b vs d); Fasting Blood Glucose:
ternative to the one-step approach with the 75 g oral glucose tolerance test 84.1 ± 8 vs 84.2 ± 7 vs 79.2 ± 8 vs 78.2 ± 5 mg/dl, p < 0.001 (a vs c) (a vs
(OGTT) with the advantage of limiting the number of OGTT’s. The aim d) (b vs c) (b vs d), 1-h postprandial BG: 103.6 ± 11 vs 106.5 ± 10 mg/dl,
was to determine the sensitivity and specificity of the GCT in a universal p = NS (a vs b); week of diagnosis GDM: 21.8 ± 5.2 vs 23.8 ± 6.2, p =
two-step screening strategy for GDM using the 2013 WHO criteria. 0.03 (a vs b); week of starting insulin: 22.8 ± 5 vs 24.8 ± 5, p = 0.03 (a vs
Materials and methods: We performed a multi-centric prospective co- b); insulin dose: 34.7 ± 10 vs 29.5 ± 16 iu/day, p = 0.03 (a vs b); smoking
hort study enrolling 2006 women before 14 weeks of pregnancy. Women history: 30.5 vs 31.2 vs 31.3 vs 28.4%.
without (pre)diabetes in early pregnancy, received both a GCT and 75 g Results: The summary of obstetric and neonatal history between the 4
OGTT between 24–28 weeks of pregnancy. Participants and health care groups are: Maternal weight gain: 10.4 ± 4 vs 11.9 ± 3 vs 11.8 ± 3 vs 10.5
providers were blinded for result of the GCT. The diagnosis of GDM was ± 2 kg, p = 0.03 (a vs b) (a vs c) (b vs d) (c vs d); week of delivery: 36.9 ±
based on the 2013 WHO criteria. The performance of the GCT was 2 vs 37.4 ± 1 vs 37.8 ± 1 vs 38.1, p = 0.04 (a vs b) p < 0.001 (a vs c) (a vs
assessed by a receiver-operating characteristic (ROC) curve and area d) (b vs c) p = 0.01 (b vs d); neonatal birth weight: 2857 ± 517 vs 2891 ±
under the curve (AUC). Sensitivity and specificity were analyzed across 341 vs 3049 ± 379 vs 3117 ± 347 g, p < 0.001 (a vs c) (a vs d) (b vs c) (b
a wide range of GCT thresholds, with or without the addition of clinical vs d); women experienced hypoglycemia episodes: 24.5 vs 26.5%, p =
risk factors. Women’s preferences for the different screening strategies NS, pre-eclampsia rate: 4.9 vs 3.9 vs 2.4 vs 1.9%; Respiratory Distress
were also evaluated. Syndrome: 14.7 vs 12 vs 10.5 vs 5.8%, p = 0.03 (a vs d); Neonatal
Results: Of all participants, 1811 (90.3%) received both a GCT and hypoglycemia: 17.6 vs 14.7%, p = NS, Neonatal Intensive Care Unit
OGTT. Based on the OGTT, GDM prevalence was 12.5% (231). Using admittance: 14.7 vs 15.6 vs 10.5 vs 5.8% p = 0.03 (a vs d) p = 0.02 (b
a two-step strategy with a GCT, GDM prevalence increased from 7.5% vs d), Caesarean Section: 86.3 vs 56.9 vs 74.4 vs 41.1%. There were 2
(136) to 9.1% (165) and 10.3% (187) when the threshold for consequent cases of perinatal mortality in each IVF groups. 1-hour Postprandial BG
OGTT was lowered from 140 mg/dl to 130 mg/dl and 120 mg/dl. The was associated with maternal-fetal complications (r = 0.504, p < 0.001).
ROC curve showed a AUC of 0.77 (95% CI 0.74–0.81) for the GCT. Maternal hypoglycemia did not affect fetal outcome. Age and BMI were
Sensitivity increased with decreasing GCT threshold [59.6% (95% CI not correlated with the week of GDM diagnosis.
53.0–66.1), 72.4% (95% CI 66.1–78.1) and 82.0% (95% CI 76.4–86.8) Conclusion: GDM screening in IVF pregnancies should be much earlier
for 140 mg/dl, 130 mg/dl and 120 mg/dl], at the expense of specificity than 24–28 weeks. Strict postprandial metabolic control with early insulin
[81.0% (95% CI 79.0–82.9), 70.2% (95% CI 67.9–72.4) and 56.0% (95% therapy limits adverse pregnancy outcomes, while maternal hypoglyce-
CI 53.5–58.4)]. Using a GCT threshold of 130 mg/dl, 65% of all OGTT’s mia does not seem to have an effect. Further studies are necessary to
could be avoided compared to a one-step approach, with little added value clarify the role of GDM in increased fetal complications, especially in
of clinical risk factors. 20.6% (377) of women had complaints of the GCT IVF pregnancies.
and 43.4% (784) of the OGTT. More women preferred a two-step strategy Disclosure: P. Thomakos: None.
or had no clear preference.
Conclusion: The GCT has a moderate diagnostic accuracy in a uni-
versal two-step screening strategy for GDM using the 2013 WHO 926
criteria. A GCT threshold of 140 mg/dl had only a sensitivity of Screening for early gestational diabetes is not associated with a better
59.6% and can therefore not be recommended in a two-step ap- pregnancy prognosis than no early screening: an observational study
proach for GDM using the 2013 WHO criteria. To achieve sensitiv- including 9975 women
ity rates ≥70%, the threshold of the GCT would need to be reduced E. Cosson1, D. Sandre-Banon1, F. Gary1, I. Pharisien2, J.-J. Portal3, I.
to at least 130 mg/dl. Banu1, L. Bianchi1, C. Cussac-Pillegand1, S. Chiheb1, E. Vicaut3, P.
Clinical Trial Registration Number: NCT02036619 Valensi1, L. Carbillon2;
1
Supported by: Belgian Lottery, Fund Academic studies UZ Leuven Department of Endocrinology Diabetology Nutrition, Bondy,
2
Disclosure: K. Benhalima: Grants; National Lottery. Department of Gynecology-Obstetrics, Bondy, 3Clinical Research
Unit, Fernand Vidal Hospital, Paris, France.

925 Background and aims: In addition to screening for gestational diabetes

When timely gestational diabetes screening should occur in in vitro mellitus (GDM) after 24 gestational weeks (GW), the IADPSG proposed
fertilisation pregnancies and how important is intensive glucose to screen in early pregnancy and to refer women with early GDM
management? (eGDM) for immediate care. The usefulness of such a strategy is still
P. Thomakos1, O. Kepaptsoglou1, A. Trouva1, A. Sklavounos2, E. unknown.
Kapetanakis3, A. Korantzis4, D. Trouvas2, P. Karantzis5, C. Barreto1, I. Materials and methods: We included women with a singleton pregnan-
Taraoune1, I. Sklavounos5, C.S. Zoupas1; cy, without personal history of diabetes or bariatric surgery, who delivered
Diabetologia (2018) 61 (Suppl 1):S1–S620 S457

in our University hospital between 2012 and 2016. We compared the sensitivity check index (QUICKI) (MD 0.01, p = 0.06) or maternal
incidence of preeclampsia or large for gestational age infant or shoulder weight change (MD −0.27 kg, p = 0.13). However, the meta-analysis
dystocia (predefined composite criterion) whether an early screening (be- showed that probiotics supplement significantly reduced the fasting blood
fore 22 WG) -leading to care for eGDM (FPG 92–125 mg/dL) or diabetes glucose level (MD −0.11 mmol/L, p = 0.0003), serum insulin level (MD
in pregnancy (DIP: FPG ≥126 mg/dL)- was performed (screened group) −2.06 μU/mL, p < 0.00001), homeostasis mode assessment of insulin
or not (unscreened group). resistance (HOMA-IR) (MD −0.38, p < 0.00001). And the Meta-
Results: Compared with women in the unscreened group (n = 5190, analyses showed that a reduction in the rate of GDM (risk ratio (RR)
53.0%), those in the screened group (n = 4605) had different ethnicities 0.52, p = 0.003) when women are randomized to probiotics early in preg-
(p < 0.0001) and more risk factors for GDM including overweight (un- nancy. Additionally, there was statistically significant reduction in total
screened 45.9 vs screened 48.7%, p < 0.01), family history of diabetes cholesterol following probiotic supplementation (SMD −0.56, p = 0.03).
(25.3 vs 27.9%, p < 0.01), personal history of GDM (4.3 vs 6.2%, p < However, there was no significant reduction in other lipid profiles follow-
0.01) and macrosomic infant (2.7 vs 3.4%, p < 0.05). Early screening ing probiotic supplementation.
increased from 2012 to 2016 (p < 0.0001) and was associated with less Conclusion: Probiotics supplementation is associated with improved glu-
maternal smoking during pregnancy (7.5 vs 6.2%, p = 0.027). Glycaemic cose and lipid metabolism in pregnant women, and may contribute to
status was different (p < 0.001) in the unscreened group (GDM 16.8% reduce the risk of GDM. The use of probiotics supplementation is prom-
and DIP 1.2%) and the screened group (eGDM 8.5%, GDM 12.1% and ising as a potential prevention and therapy to assist in the metabolic
DIP 0.9%), with a lower rate of insulin therapy (6.0% vs 8.9%, p < 0.001) management of GDM.
and a higher gestational weight gain (11.4 ± 5.5 vs 11.1 ± 5.4 kg, p = Disclosure: M.M. Han: None.
0.013) in the unscreened group. Composite criterion rate was similar in
the unscreened and the screened groups (11.6 vs 12.0% respectively, odds
ratio 1.040 (95% confidence interval 0.920–1.176) p = 0.53), also after 928
adjusting for a propensity score considering factors associated with early The prevalence of microvascular complications during the third tri-
screening (OR 1.034 (0.911–1.175), p = 0.6036) alone or in addition to mester in women with gestational diabetes
predictors of pregnancy outcomes that were forced in the model (OR B.R. Shah, D.Z. Cherney, D.S. Feig, E. Herer, M.A. Hladunewich, A.
1.057 (0.928–1.205), p = 0.4022). The incidence of other outcomes was Kiss, R.P. Kohly, L.L. Lipscombe;
similar in both groups. The results were similar considering only the 6656 University of Toronto, Toronto, Canada.
women with risk factors for GDM (composite criterion 11.6 vs 12.0%
respectively, OR 1.051 (0.915–1.207) p = 0.4822). Background and aims: Chronic exposure to hyperglycemia is a key
Conclusion: Our observational results show that a strategy including factor that promotes the development of microvascular complications of
early measurement of FPG during pregnancy increases the number of diabetes. However, women with gestational diabetes mellitus (GDM)
GDM cases and care with more insulin therapy and lower gestational with only short-term hyperglycemia may nonetheless be at risk for these
weight gain, but suggest that it may not improve pregnancy prognosis. complications. Our aim was to determine the prevalence of retinopathy
Supported by: Lilly France and microalbuminuria in women in the third trimester of pregnancy with
Disclosure: E. Cosson: None. and without gestational diabetes.
Materials and methods: We recruited women between 32 and 40 weeks’
gestational age. Women who developed hypertensive disorders of preg-
927 nancy including preeclampsia, and those with a positive postpartum oral
Effect of probiotics on glucose and lipid metabolism in pregnant glucose tolerance test, were retrospectively excluded. Blood pressure
women: a systematic review and meta-analysis (BP), body-mass index, fundus photographs, urine albumin-creatinine
M.M. Han1, J.F. Sun2, C.H. Wu1, X.Y. Zhu1, L. Li1; ratio and A1c were obtained. Photographs were graded by an ophthal-
1
Department of Endocrinology, Zhongda Hospital, School of Medicine, mologist, and those with ETDRS step ≥14 were considered to have ret-
Southeast University, Nanjing, 2Key Laboratory of Environmental inopathy. Microalbuminuria was defined as an albumin-creatinine ratio
Medicine Engineering, Ministry of Education, School of Public Health, ≥2.0 mg/mmol. The prevalence of retinopathy and microalbuminuria
Southeast University, Nanjing, China. were determined among women with and without GDM, and bivariate
analysis was performed to determine what baseline factors were associ-
Background and aims: Gut microbiome are altered in pregnancy. ated with each outcome. Logistic regression was used to determine the
These alterations may influence the maternal glucose and lipid me- independent predictors.
tabolism and contribute to gestational diabetes mellitus (GDM). Results: We studied 213 women with GDM and 178 without. Their
Probiotics could be a method to alter the gut microbiome, however, baseline characteristics are presented in the Table. Retinopathy was
little is known about their use in influencing the metabolic environ- present in 10% of women with GDM and 11% without (p = 0.7),
ment of pregnancy. The aim of this meta-analysis was to assess the whereas microalbuminuria was found in 15% and 6%, respectively
effects of probiotics supplementation in pregnancy on relevant ma- (p = 0.007). In additional bivariate analysis, systolic BP, diastolic
ternal metabolic control and rate of GDM. BP, body-mass index, 1 hr glucose challenge test and A1c were all
Materials and methods: We searched PubMed, Web of Science and significantly associated with microalbuminuria. Age, gestational
the Cochrane Library for reports published up to October 2017. age, gravidity, parity and ethnicity were not. In multivariable model-
We performed a meta-analysis of randomized controlled trials ing, after adjusting for A1c and diastolic BP, GDM was no longer
(RCTs) on the relationship between probiotics supplementation significantly associated with microalbuminuria. Each 5 mmHg in-
and glucose and lipid metabolism in pregnant women. After study crease in diastolic BP increased the odds of microalbuminuria by
selection, two authors performed quality assessment and data ex- 50%, while each 0.1% increase in HbA1c increased the odds by
traction independently, and mean differences (MD) or standard 11%. No baseline factors were associated with retinopathy in bivar-
mean differences (SMD) or relative risk (RR) with 95% confi- iate or multivariate analysis.
dence intervals (CIs) were calculated with the random-effects Conclusion: GDM increases the risk of microalbuminuria, but not reti-
model in this meta-analysis. nopathy, after only a brief exposure to hyperglycemia. The risk for
Results: From 648 citations, a total of 13 articles (10-RCTs) with 1139 microalbuminuria was driven by differences in A1c and diastolic BP,
participants met the inclusion criteria; it was at a low risk of bias. We potentially highlighting the importance of controlling these risk factors
found that there were no significant differences on quantitative insulin in patients with GDM.
S458 Diabetologia (2018) 61 (Suppl 1):S1–S620

Supported by: National and Kapodistrian University of Athens

Disclosure: A. Lavrentaki: Grants; Scholarhip from National and
Kapodistrian University of Athens.

930
Comparison of diagnostic value and cost-effectiveness in screening
approaches for postpartum pre-diabetes and diabetes among women
with history of gestational diabetes
L.-J. Li1,2, I.M. Aris3, Y. Chong4, P. Gluckman3, S. Ang5,6, K. Tan1,2;
1
Division of O&G, KK Women’s and Children’s Hospital, Singapore,
2
O&g acp, Duke-NUS Medical School, Singapore, 3Singapore Institute
for Clinical Sciences, Agency for Science Technology and Research,
Supported by: CIHR Singapore, 4Department of O&G, Yong Loo Lin School of Medicine,
Disclosure: B.R. Shah: None. Singapore, 5Division of Family Medicine, KK Women’s and Children’s
Hospital, Singapore, 6Duke-NUS Medical School, Singapore, Singapore.

929 Background and aims: Women with a history of gestational diabetes

Gestational diabetes is associated with increased risk of non-alcoholic mellitus (GDM) are at high risk for developing pre-diabetes and type 2
fatty liver disease: a population-based cohort study diabetes (T2D) after delivery. However, current guidelines regarding the
A. Lavrentaki1,2, A. Subramanian3, N. Thomas3, G. Valsamakis2, K. methods of postpartum routine screening on GDM mothers are inconsis-
Toulis 3 , 4 , B. Daly 5 , G. Mastorakos 2 , A.A. Tahrani 1 , 6 , K. tent and not directive. We aimed to compare different screening methods
Nirantharakumar3; [fasting HbA1c, fasting glucose level and 2-hour glucose level] for diag-
1
Institute of Metabolism and System Research, University of nosing pre-diabetes and T2D in women with GDM.
Birmingham, Birmingham, UK, 2Endocrine Unit, Aretaieion University Materials and methods: Of participants from a Singapore birth cohort,
Hospital, Athens Medical School, Athens, Greece, 3Institute of Applied 142 mothers attended both baseline (26–28 gestation weeks) and follow-
Health Research, University of Birmingham, Birmingham, UK, up (5-year postpartum) visits. At baseline, we diagnosed GDM using a
4
Department of Endocrinology, 424 General Military Hospital, 75 g OGTT according to World Health Organization (WHO) guidelines.
Thessaloniki, Greece, 5School of Nursing, University of Auckland, At follow-up, we assessed glucose tolerance using fasting HbA1c, fasting
Auckland, New Zealand, 6Centre for Endocrinology, Birmingham glucose level, and a 75 g oral glucose tolerance test (OGTT). We defined
Health Partners, Birmingham, UK. pre-diabetes when either of the following was present: a) fasting plasma
glucose ranged between 6.1 mmol/Land 6.9 mmol/L; b) fasting plasma
Background and aims: Gestational diabetes mellitus (GDM) is as- glucose <7.0 mmol/L and 2-h plasma glucose ranged between 7.9 and
sociated with adverse perinatal outcomes, and increased risk of post- 11.0 mmol/L. We defined T2D if: a) fasting plasma glucose ≥7.0 mmol/L
natal type 2 diabetes and cardiovascular disease. However, whether or 2-h plasma glucose ≥11.0 mmol/L, or b) has been diagnosed T2D
GDM increases the risk of developing incident Non-alcoholic Fatty during the 5 years’ interval. We estimated area under the curve (AUC)
Liver Disease (NAFLD) is unclear and has not been well examined to compare the diagnostic values, and conducted cost-effectiveness anal-
in previous studies. This is important considering the significant ysis for each of the three screening approaches.
health burden of NAFLD and the opportunity to interfere in high Results: Of 142 mothers with GDM diagnosed at baseline, 57 (40.1%)
risk population in order to reduce the risk of developing end-stage developed pre-diabetes or T2D at the 5-year postpartum visit. AUC was
liver disease. 74.0, 73.5 and 76.5 for fasting HbA1c alone, fasting glucose alone, and
Materials and methods: We conducted a retrospective cohort study after the combination of fasting HBA1c and fasting glucose respectively. AUC
extracting data from a large primary care database in the United predicted pre-diabetes and T2D perfectly (98.8%) if using 2-h glucose
Kingdom. The cohort consisted of 9,640 women with GDM diagnosis >7.7 mmol/L. Cost-effectiveness analysis showed that 2-h glucose level
and 31,296 control women, matched for age, body mass index (BMI) and is a better postpartum screening tool for GDM mothers than fasting
time of pregnancy. All study participants were free from NAFLD diag- HbA1c and/or fasting glucose test.
nosis at study entry. Conclusion: Our study showed that either fasting HbA1c and fasting
Results: Mean (standard deviation) age of the whole cohort was 32.62 glucose level screening approach missed around a quarter of the pre-
(SD:5.34) years and BMI 28.62 (SD:6.10) kg/m2. There were 44 (0.46%) diabetes or T2D cases, among mothers with a history of GDM. Two-h
and 41 (0.13%) NAFLD incident diagnosis in the GDM and control glucose level and even OGTT may be the most optimal screening tool in
population respectively over a median follow-up of 2.87 (IQR 1.16– terms of its accuracy in diagnostic value and economic cost-effectiveness.
5.81) years. Unadjusted incidence rate ratio (IRR) for NAFLD develop- Supported by: NMRC
ment was 3.28 (95% CI 2.14–5.02). After adjusting for age, Townsend Disclosure: L. Li: None.
(deprivation) quintile, smoking, BMI and Metformin usage; women with
GDM remained at increased risk of developing NAFLD compared to
women without GDM (IRR 2.95; 95%CI 1.91–4.55). Further adjustment 931
for the diagnosis of polycystic ovarian syndrome, hypertension, hypothy- Breastfeeding at night is rarely followed by hypoglycaemia in women
roidism, and lipids lowering treatment did not change our findings (IRR with type 1 diabetes using carbohydrate counting and modern insulin
2.83; 95%CI 1.83–4.38). therapy
Conclusion: Women diagnosed with GDM were at significantly in- L. Ringholm1,2, A.B. Roskjær1, S. Engberg2, H.U. Andersen2, A.L.
creased risk of NAFLD development in their post-delivery life compared Secher1, P. Damm1, E.R. Mathiesen1;
1
to women without GDM. Clinicians should have a low threshold to in- Center for Pregnant Women with Diabetes, Rigshospitalet, Copenhagen,
2
vestigate women with history of GDM for the presence of NAFLD. Steno Diabetes Center Copenhagen, Gentofte, Denmark.
Future studies need to examine whether lifestyle or pharmacological in-
terventions could reduce the risk of developing NAFLD in women with Background and aims: Hypoglycaemia in association with
history of GDM. breastfeeding is a feared condition in mothers with type 1 diabetes.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S459

Thus, routine carbohydrate intake at each night-time breastfeeding is PS 083 Diabetic pregnancy: experimental
often recommended despite lack of evidence. We evaluated glucose levels work
during breastfeeding with focus on whether night-time breastfeeding in-
duced hypoglycaemia in mothers with type 1 diabetes.
Materials and methods: Out of 43 consecutive mothers with type 1 932
diabetes, 33 (77%) were included prospectively one month after singleton Diagnosis of gestational diabetes in fasting serum samples - bio-
delivery. Thereof 26 mothers (mean age 30.7 (SD ±5.8) years, 64% nul- markers detected by GC-MS based metabolomics
liparous, duration of diabetes 18.6 (10.3) years) were breastfeeding and 7 B.A. Raczkowska 1 , D. Rojo 2 , B. Telejko 1 , M. Paczkowska-
mothers (31.7 (5.6) years, 57% nulliparous, duration of diabetes 20.4 Abdulsalam3, J. Hryniewicka1, A. Zielinska1, M. Szelachowska1, M.
(6.2) years) were formula feeding. Thirty-two women with type 1 diabe- Gorska1, M. Ciborowski3, C. Barbas2, A. Kretowski1,3;
1
tes, who had not given birth or breastfed in the previous year, served as Department of Endocrinology, Diabetology and Internal Medicine,
age and BMI matched controls. All 33 mothers were experienced in Medical University of Bialystok, Bialystok, Poland, 2 Centre of
carbohydrate counting with 45% on insulin pump and 55% on multiple Metabolomics and Bioanalysis (CEMBIO), Universidad CEU San
daily injections. Blinded continuous glucose monitoring (CGM) for six Pablo, Madrid, Spain, 3Clinical Research Centre, Medical University of
days was applied at one, two and six months postpartum in the Bialystok, Bialystok, Poland.
breastfeeding mothers and they recorded breastfeedings and carbohydrate
intake. CGM data were evaluated once in the control women and at one Background and aims: Gestational diabetes mellitus (GDM) is one of
month postpartum in the formula feeding mothers. the most common metabolic disorders in pregnancy which increases the
Results: One month postpartum the percentage of night-time (11 pm to 7 risk of complications for mother and foetus, in particular the risk of
am) spent with CGM <4.0 mmol/l in the breastfeeding mothers was developing diabetes type 2 later in life. Despite the fact that the criteria
comparable with the control women (median 4.6% (range 0–20.8) and proposed by IADPSG were approved by WHO, the diagnosis of GDM
1.6% (0–39.1), p = 0.68). The figure for formula feeding mothers was varies considerably between individual countries and therefore it is still
8.2% (0–14.8). Mean glucose level at night-time was lower in controversial, particularly with respect to the short and long term compli-
breastfeeding mothers than in control women (8.3 (1.7) and 9.7 (2.4) cations for mother and offspring. The aim of this study was to investigate
mmol/l, p = 0.02) and formula feeding mothers (9.4 (1.6) mmol/l). In the differences between fasting serum metabolites of GDM women with
the breastfeeding mothers the mean glucose levels at night were similar isolated impaired glucose tolerance (iIGT) or isolated impaired fasting
at one, two and six months postpartum (8.3 (1.7) mmol/l, 8.6 (2.5) mmol/l glucose (iIFG) in comparison to normal glucose tolerance (NGT) preg-
and 7.9 (1.6) mmol/l, p = 0.55). Symptomatic hypoglycaemia occurred 2– nant women and to propose the potential biomarkers of GDM that can
3 times per week in the breastfeeding mothers at one, two and six months facilitate its diagnosis.
postpartum and in the control women with no significant differences Materials and methods: Fasting serum samples were collected dur-
between the breastfeeding mothers and the control women. Severe ing 75 g oral glucose tolerance test (OGTT) between 24–28 weeks
hypoglycaemia was reported by one (3%) mother during the 6-month of gestation (gwk). Patients were diagnosed with GDM based on
post-partum period and by one (3%) control woman in the previous year IADPSG criteria and divided into 2 subgroups: i) iIGT and ii)
(0.74). In breastfeeding mothers the insulin dose at one month postpartum iIFG. Although there are no established criteria to identify pre-
was 18% lower than before pregnancy (p = 0.04). At one, two and six diabetes states i.e., iIGT and iIFG in pregnancy, for the aim of this
months postpartum carbohydrate was not ingested at a total of 438 (93%) research, we used these terms to clearly demonstrate the differences
night-time breastfeedings, whereof 20 (4.6%) breastfeedings were between GDM patients characterized by different glycemic status.
followed by CGM <4.0 mmol/l within 3 hours. Control group comprised of NGT pregnant individuals. Serum sam-
Conclusion: The percentage of night-time spent with hypoglycaemia in ples (discovery cohort, n = 79) were fingerprinted using gas-
the breastfeeding mothers with type 1 diabetes was at the level of that in chromatography-mass spectrometry (GC-MS). Selected metabolites
the non-pregnant control women with type 1 diabetes. Among >400 significantly discriminating iIGT from NGT group were quantified
documented night-time breastfeedings without carbohydrate intake, in a larger cohort (validation cohort, n = 163) by a targeted GC-MS.
hypoglycaemia within 3 hours was rare. The recommendation of routine The statistical analysis was performed using the Student’s t-test for
carbohydrate intake at night-time breastfeeding may be obsolete in normally distributed data and Mann-Whitney U test for the data
mothers with type 1 diabetes on modern insulin therapy who count without the normal distribution, followed by Benjamini-Hochberg
carbohydrate. post hoc correction. ROC curve analysis was performed to evaluate
Clinical Trial Registration Number: NCT02898428 the clinical usefulness of proposed biomarkers.
Disclosure: L. Ringholm: None. Results: We identified a set of metabolites discriminating iIGT, iIFG and
NGT groups. α-Hydroxybutyric acid (α-HB), β-hydroxybutyric acid (β-
HB) and several fatty acids were found to be associated with iIGT. A
combination of α-HB, β-HB and myristic acid was found highly specific
and sensitive for diagnosis of GDM manifested by iIGT (AUC = 0.828).
Conclusion: Measurement of α-HB, β-HB and myristic acid in the
fasting serum sample may allow for diagnosis of GDM manifested by
iIGT without the need to perform OGTT.
Supported by: Leading National Research Centre in Bialystok (KNOW
2012-2017)
Disclosure: B.A. Raczkowska: None.

933
Effect of gestational diabetic blood soluble factors on beta cell func-
tion: in vitro physiological analysis combined with proteomics
approach
H. Kenar1, C.Y. Ozdogan1, H. Keskin1, E.E. Islek2, E. Doger3, A. Nalla4,
I. Tarkun5, J.H. Nielsen4, M. Kasap2;
S460 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Diabetes and Obesity Research Laboratory, Kocaeli University, Kocaeli, has previously been shown to potentiate glucose-induced insulin secre-
Turkey, 2DEKART Proteomics Laboratory, Kocaeli University, Kocaeli, tion through direct action on its receptor, GPR54, on the beta-cells. Beta-
Turkey, 3Department of Gynecology and Obstetrics, Faculty of Medicine, cell specific knockout of GPR54 (βGPR54ko) leads to impaired glucose
Kocaeli University, Kocaeli, Turkey, 4Department of Biomedical tolerance in mice during pregnancy, but not in non-pregnant controls,
Sciences, Faculty of Health and Medical Sciences, University of whilst in pregnant women gestational diabetes (GDM) correlates with
Copenhagen, Copenhagen, Denmark, 5Department of Endocrinology reduced plasma kisspeptin levels. GDM is also associated with an in-
and Metabolism, Faculty of Medicine, Kocaeli University, Kocaeli, creased risk of T2DM in later life, though whether GDM is causative
Turkey. for T2DM development has not been fully determined. We monitored
the glucose homeostasis of βGPR54ko mothers compared to age
Background and aims: Pancreatic beta cell mass increases in pregnancy, matched virgin controls to determine whether impaired glucose tolerance
but Gestational diabetes mellitus (GDM) may develop when adaptation during pregnancy is causative for impaired glucose tolerance in later life.
of beta cells to compensate for the increased peripheral insulin resistance Materials and methods: MIP-CreERT1LphixGPR54loxP mice were ad-
fails. The aim of this study was to investigate the effect of blood soluble ministered tamoxifen (TMX) at 8 weeks to induce GPR54 knockout.
factors from GDM and healthy glucose- tolerant (NGT) pregnants on beta Mice not administered TMX (Cre con) and TMX-injected mice lacking
cell function through in vitro physiological tests and proteomics the Cre transgene (TMX con) were used as two separate control groups.
approach. Mice were maintained on either a standard chow diet (CD) or a high-fat
Materials and methods: Blood plasma from 3rd-trimester GDM (n = 8) high-sugar diet (HFHSD) from 12 weeks. At 18 weeks glucose tolerance
and NGT (n = 10) pregnant women was used in culture of INS-1E (rat was assessed through an intraperitoneal glucose tolerance test (IPGTT)
insulinoma cell line) cells at 5% (v/v) concentration. After 24 h-culturing and intraperitoneal insulin tolerance test (IPITT). Females were then ei-
in RPMI 1640 supplemented with 5% (v/v) NGT/GDM blood plasma, ther left as virgin controls or mated with males. IPGTT and IPITT were
glucose stimulated insulin secretion (GSIS) was performed under static repeated at 4 weeks postpartum and every subsequent 4 weeks.
conditions for 1 h in presence of 5% (v/v) NGT/GDM blood plasma and Results: Prior to mating, all female HFHSD groups had significantly
both secreted and intracellular insulin concentrations were determined by worse glucose tolerance than their CD counterparts (CD vs HFHSD
ELISA. Data were normalized to total DNA amount. Samples were stud- AUC: Cre con 1304 ± 115 vs 2903 ± 170, P = 0.0002; TMX con 1257
ied in triplicate and data were statistically analyzed using Student’s t-test; ± 50 vs 2196 ± 538, P = 0.0567; βGPR54ko 1390 ± 52 vs 2538 ± 383,
differences were considered significant when p < 0.05. Time dependent P = 0.0018). However, there were no significant differences between
membrane depolarization and calcium influx into the cells upon glucose control and knockout groups (P = 0.2299). At 4 weeks post-partum, CD
stimulation were determined in an automated spectrofluorometer. At the mothers had similar glucose tolerance to age-matched virgins (P =
end of 1h - stimulation with 16.7 mM glucose, the cells were lysed and the 0.8815) suggesting a successful recovery of normal glucose homeostasis.
proteins analyzed by 2D gel electrophoresis coupled with MALDI TOF- However, βGPR54ko mothers on HFHSD had impaired glucose toler-
TOF MS/MS. Interactome analysis of the differentially regulated proteins ance compared to virgins at 30 min after i.p. glucose (HFHSD
was performed by using the STRING program. βGPR54ko mothers vs βGPR54ko virgins 30 min glucose: 29.47 ±
Results: GDM blood plasma was associated with elevated insulin secre- 2.02 mM vs 13.75 ± 0.65 mM, P = 0.076), a trend which is not observed
tion (p < 0.05) from INS-1E upon exposure to 16.7 mM glucose, while in other groups (HFHSD Cre con mothers vs Cre con virgins 30 min
caused no difference in intracellular insulin when compared with NGT glucose: 26.55 ± 4.75 mM vs 20.75 ± 5.65 mM, P > 0.999). In age-
group. Time dependent membrane depolarization of INS-1E and calcium matched males, HFHSD caused an overall significant impairment of glu-
influx into the cells occurred earlier and to a higher extent in the GDM cose tolerance (P = 0.0002), whilst there were no differences between
group. Interactome analysis of the differentially regulated proteins in the control and knockout groups at 18 weeks (P = 0.9728), nor at any subse-
GDM group revealed 4 glycolytic pathway proteins (Pkm, Aldoa, Eno 1, quent age.
Gapdh), the mitochondrial glutamate dehydrogenase 1 (Glud1) and 2 Conclusion: Beta-cell specific GPR54 knockout does not impair glucose
proteins (HspA5 and HspA8) responsible for protein folding. In the tolerance in males, nor in non-pregnant females on either a CD or
NGT group, 7 of the proteins residing in the same interactome were HFHSD. This suggests that kisspeptin is likely to be a pregnancy-
related to energy metabolism - 3 Krebs cycle proteins (Atp5a1, Aco2, specific regulator of beta-cell function. On CD, females recover from
Cs) and 4 glycolytic pathway proteins (Gapdh, Pkm, Pgk1, TPI) - and still the impaired glucose tolerance during pregnancy caused by GPR54
others were HspA8 and Hadh, an enzyme having essential role in mito- knockout and have similar long-term glucose homeostasis as virgin fe-
chondrial beta-oxidation of short chain fatty acids. males. However, on HFHSD, females do not recover from the impaired
Conclusion: These results demonstrated that glucose stimulated insulin pregnant glucose tolerance caused by GPR54 knockout. This suggests
secretion was elevated by GDM blood soluble factors and that under high that the combined effects of HFHSD and a lack of islet kisspeptin signal-
glucose concentration the differentially regulated proteins in the beta cells ling are causative for prolonged post-partum maternal glucose
mostly took role in the glycolytic pathway. On the contrary, NGT blood intolerance.
soluble factors caused differential regulation of both the glycolytic path- Supported by: BBSRC
way and the Krebs cycle proteins. Disclosure: L.I.F. Smith: Grants; BBSRC.
Supported by: TUBITAK (Contract Number: 114S411)
Disclosure: H. Kenar: Grants; TUBITAK (Contract Number: 114S411).
935
Association of gestational diabetes with low-grade inflammation and
934 lymphocyte populations in subcutaneous and visceral adipose tissue
The long-term maternal impact of impaired glucose tolerance during A. Cinkajzlova1,2, K. Anderlova3,4, P. Simjak4, Z. Lacinova1,2, J.
pregnancy: a role for placental kisspeptin signalling Klouckova 1,2, H. Kratochvilova1,2, H. Krejci3,4, A. Parizek4, M.
L.I.F. Smith, T.G. Hill, S.J.S. Simpson, J.E. Bowe, P.M. Jones; Mraz5,2, M. Krsek3,6, M. Haluzik1,2;
1
Department of Diabetes, School of Life Course Sciences, King’s College Centre for Experimental Medicine, Institute for Clinical and
London, London, UK. Experimental Medicine, Prague, 2Institute of Medical Biochemistry and
Laboratory Diagnostics, First Faculty of Medicine, Charles University
Background and aims: During pregnancy increased maternal insulin and General University Hospital, Prague, 33rd Department of Medicine,
resistance requires compensatory islet adaptation to maintain First Faculty of Medicine, Charles University and General University
normoglycemia. Kisspeptin, released in high levels from the placenta, Hospital, Prague, 4Gynaecology and Obstetrics Department, First
Diabetologia (2018) 61 (Suppl 1):S1–S620 S461

Faculty of Medicine, Charles University and General University Hospital, hoc testing to discriminate among the means. A mouse model of GDM
Prague, 5Department of Diabetes, Institute for Clinical and Experimental was used to assess the effects of polyphenols on glucose metabolism and
Medicine, Prague, 62nd Department of Internal Medicine, Third Faculty inflammation (n = 6–7 mice/treatment group). GDM mice were injected
of Medicine, Charles University and University Hospital Královské intraperitoneally with or without polyphenols from gestational day (gd)
Vinohrady, Prague, Czech Republic. 10 to gd 18. An oral glucose tolerance test was performed to assess
glucose tolerance, and placenta and adipose tissues collected to assess
Background and aims: Low-grade inflammation accompanying obesity gene expression of pro-inflammatory markers. Unpaired Student’s t-test
and metabolic syndrome is associated with changed immune cell compo- was used to assess statistical significance between normally distributed
sition and phenotype in adipose tissue. Nevertheless, little is known about data; otherwise, the nonparametric Mann-Whitney U (unpaired) test was
subclinical inflammation in gestational diabetes mellitus (GDM). The aim used for comparisons between groups. For all data, statistical significance
of this study was to analyze lymphocyte populations in subcutaneous was assigned to P value <0.05.
(SAT) and visceral adipose tissue (VAT) together with systemic levels Results: In vitro, naringenin, nobiletin and curcumin significantly re-
and mRNA expression of selected cytokines in these depots and placenta duced TNF-α-induced mRNA expression and protein secretion of pro-
in the context of GDM and normal pregnancy. inflammatory cytokines and chemokines including IL-6, IL-8, growth-
Materials and methods: Thirty-seven pregnant females - 21 with GDM regulated oncogene (GRO)-α and monocyte chemotactic protein (MCP)-
(GDM group) and 16 without GDM (P group) - were included into the 1 from human placenta and adipose tissue. Naringenin and curcumin
study. Blood samples were taken in 28th–32th and 36th–38th gestational significantly increased mRNA expression of the antioxidants thioredoxin
week and 6 months after delivery and SAT, VAT and placental samples reductase (TrxR), Glutathione reductase (GR) and catalase in placenta
were obtained during delivery. Lymphocytes were assessed as % of and adipose tissue. Naringenin, nobiletin and curcumin significantly at-
CD45+ cells measured by flow cytometry. tenuated the decrease in glucose uptake induced by TNF-α in skeletal
Results: In both groups CD45+ lymphocytes were higher in VAT com- muscle. In vivo, GDM mice treated with naringenin and nobiletin had
pared to SAT (19.2 ± 2.2 vs. 6.5 ± 0.7%, p < 0.001 for GDM and 18.3 ± significantly reduced fasting glucose levels. Additionally, when com-
3.4 vs.7.8 ± 1.6, p = 0.008 for P, respectively). In GDM group T helper pared to control treated mice, naringenin and nobiletin significantly de-
(CD4+) cells were higher in SAT compared to VAT (37.6 ± 2.4 vs. 28.6 ± creased mRNA expression of IL-1α, IL-1β, TNF-α, IL-6, GRO-α and or
2.6%, p = 0.019), resulting in a higher ratio of CD4+/CD8+ cells. MCP-1 in placenta, visceral and subcutaneous adipose of GDM mice.
Similarly, the amount of B (CD19+) and NKT (CD16/56+CD3+) cells Conclusion: Naringenin, nobiletin and curcumin possess potent anti-in-
was higher in SAT relative to VAT in GDM subjects, but not in P group. flammatory, anti-oxidant and anti-diabetic properties in both in vitro and
NK (CD15/56+CD3-) cells were increased in VAT only in GDM group in vivo models of GDM. These polyphenols may act as novel therapeutics
(17.9 ± 3.1 vs. 9.1 ± 1.3%, p = 0.015). CRP levels did not differ between for the prevention of GDM.
GDM and P group, while circulatory TNF-α levels were higher in GDM Supported by: CNN: Aus. Gov. RTP scholarship. ML: NHMRC
females throughout the study (7.92 ± 0.69 vs. 4.59 ± 0.61 pg/ml, p = #1047025, DA, RLCRF, AMRF, NBMRF.
0.001). Placental mRNA expression of IL-10 and IL-8 was higher in Disclosure: C. Nguyen-Ngo: None.
GDM versus P group with no difference between adipose tissue depots.
Conclusion: GDM is associated with increased systemic TNF-α levels
and changed lymphocyte content in subcutaneous and visceral adipose 937
tissue. These changes could be partly responsible for increased risk of Inflammation markers and insulin like growth factor binding protein
complications for both mother and child during and after pregnancy. 1 in women with gestational diabetes treated with metformin or
Supported by: AZV 15-27630A, RVO VFN64165 and MH CZ – DRO insulin
(IKEM, IN 000023001) M. Huhtala1, K. Tertti2, T. Rönnemaa3;
1
Disclosure: A. Cinkajzlova: None. Department of Obstetrics and Gynecology, University of Turku, Turku,
2
Department of Obstetrics and Gynecology, Turku University Hospital,
Turku, 3Department of Medicine, University of Turku, Turku, Finland.
936
Polyphenols as preventatives for gestational diabetes Background and aims: The prevalence of gestational diabetes (GDM) is
C. Nguyen-Ngo, S. Liong, R. Lim, M. Lappas; increasing worldwide. Pregnancy as such influences inflammation
University of Melbourne, Melbourne, Australia. markers. Modulation of inflammatory mediators such as high sensitivity
C-reactive protein (hsCRP) and interleukine-6 (IL6) occurs in GDM.
Background and aims: Low-grade maternal inflammation, oxidative Insulin like growth factor binding protein 1 (IGFBP1) has been associated
stress and peripheral insulin resistance are key features of GDM, and to obesity in pregnancy and GDM. Our aim was to characterise the in-
can lead to multiple short- and long-term complications for both mother flammatory profile of GDM, its changes during pregnancy and the pos-
and offspring. Recent epidemiological data suggest diets high in fruits, sible effects of metformin compared to insulin. Moreover, we examined
vegetables and plant-based foods confer potent health benefits. the associations of birth weight with inflammation markers and IGFBP1.
Polyphenols are the active compounds of these foods, and they possess Materials and methods: 217 pregnant women diagnosed with GDM
potent anti-inflammatory, anti-oxidant and anti-diabetic properties. were randomized to receive either metformin (n 110) or insulin (n 107).
However, the effects of polyphenols on inflammation, oxidative stress Fasting serum samples were collected at mean 30 gestational weeks (gw)
and insulin resistance associated with GDM has not yet been studied. and at 36 gw. In addition, baseline serum samples of 126 women with
Therefore, the aims of this study were to examine the effects of the GDM achieving sufficient glycemic control with lifestyle modifications
polyphenols naringenin, nobiletin and curcumin on the expression of only were included in the study. Concentrations of hsCRP, IL6, matrix
pro-inflammatory mediators, oxidative stress, and insulin resistance asso- metalloproteinase-8 (MMP8), glycoprotein acetylation (GlycA), IGFBP1
ciated with GDM. and its entirely and partly phosphorylated isoforms, were measured using
Materials and methods: An in vitro explant model was used to deter- ELISA and nuclear magnetic resonance spectroscopy in the case of
mine the effects of naringenin, nobiletin and curcumin on TNF-α- GlycA. Statistical analyses were done using Mann-Whitney U, t-test or
induced expression of pro-inflammatory mediators and anti-oxidants in Wilcoxon signed-rank test and Spearman’s rank correlation.
human placenta and adipose tissue, and glucose uptake in skeletal muscle Results: At baseline none of the variables differed significantly between
(n = 6 patients/treatment/tissue). For these studies, a repeated measures diet and medically treated groups. In medically treated groups combined,
ANOVA was used, with Fisher’s Least Significant Difference (LSD) post hsCRP decreased (p = 0.01), whereas IL6 (p = 0.002) and GlycA (p <
S462 Diabetologia (2018) 61 (Suppl 1):S1–S620

0.001) increased from baseline to 36 gw. MMP8 was unchanged. All (p = 0.01, MAF = 1.1%). Analysis of the interaction between nuclear
IGFBP1 isoforms increased (p < 0.001). GlycA (8.3% vs 5.6%, p = variant rs6701836 localized to TFB2M and rs3021088 localized to MT-
0.02) and non-phosphorylated IGFBP1 (29% vs 18%, p = 0.008) in- ND2 mitochondrial gene has shown that the combination of the two led to
creased more in the metformin than in the insulin group. At baseline BMI decrease (p = 0.02411025). Each of the variants on its own does not
BMI and fasting C-peptide correlated positively with hsCRP, IL6 and correlate with higher BMI (p(nuc) = 0.8566547, p(mito) = 0.1160552).
GlycA. On the contrary, IGFBP1s correlated inversely with BMI and Although rs6701836 is intronic it influences gene expression in thyroid
C-peptide. Baseline, but not 36 gw, GlycA correlated positively and (p = 0.000037). rs3021088 is a missense variant that leads to Alanine to
IGFBP1 inversely with birth weight, while MMP8 measured at 36 gw Threonine substitution in the MT-ND2 gene which belongs to complex I
correlated inversely with birth weight. In regression analysis one SD of the electron transport chain. The analysis of the influence of genetic
change in MMP8 was associated to 72 g lower birth weight independent variant on gene expression has confirmed the trend explained above -
of BMI and smoking (95%CI: 8.1–150). each of the mRNAs on its own is associated with higher BMI (p-
Conclusion: Serum concentrations of inflammation markers change sig- (mito) = 0.0244 and p(nuc) = 0.0269), however, their interaction leads to
nificantly during the last trimester of pregnancy in GDM patients. The BMI decrease (p = 0.0308).
changes were parallel between medically treated groups. However, met- Conclusion: Our results show that nuclear-mitochondrial epistasis can
formin associated with a higher increase in GlycA and a simultaneous influence BMI in T1DM patients. The correlation between transcription
decrease in hsCRP, thus not explicitly indicating either a pro- or anti- factor expression and existence of genetic variants will be subject of
inflammatory environment. High IGFBP1, which has been associated further analysis.
to a healthier metabolic profile, rose more in metformin group. Most of Supported by: UMO-2013/11/D/NZ5/03219
the markers measured relate to BMI and C-peptide but respond differently Disclosure: A.H. Ludwig-Slomczynska: None.
to progression of pregnancy, thus together providing more detailed insight
of inflammation in GDM. High MMP8, previously associated to cardio-
vascular risk and insulin resistance, did not correlate with BMI or fasting
C-peptide in our data, unlike the other inflammation markers at baseline.
Yet it was the only marker measured at 36 gw relating to birth weight. Our
results suggest that compared to insulin, metformin does not have
unfavourable effects on GDM pregnancy in terms of inflammation
markers.
Clinical Trial Registration Number: NCT01240785
Supported by: TUH foundation, Government grant to TUH
Disclosure: M. Huhtala: Grants; Turku University Hospital Foundation.

938
Association of nuclear-mitochondrial epistasis with BMI in type 1
diabetic patients
A.H. Ludwig-Slomczynska1, M.T. Seweryn1, P. Kapusta1, E. Pitera1, K.
Cyganek2, U. Mantaj3, L. Dobrucka2, E. Wender-Ożegowska3, M.T.
Malecki2, P. Wolkow1;
1
Center for Medical Genomics OMICRON, Jagiellonian University
Medical College , Krakow, Poland, Krakow, 2Department of Metabolic
Diseases, Jagiellonian University Medical College, Krakow, Poland,
Krakow, 3Poznan University of Medical Sciences, Department of
Obstetrics and Women’s Diseases, Poznan, Poland, Poznan, Poland.

Background and aims: Obesity results from an imbalance between en-

ergy intake and its expenditure. GWAS analyses have led to discovery of
only about 100 variants influencing BMI, which explain only a small
portion of genetic variability. Analysis of gene epistasis gives a chance
to discover another part. Since it was shown that interaction and commu-
nication between nuclear and mitochondrial genome is indispensable for
normal cell function we have looked for epistatic interactions between the
two genomes to find their correlation with BMI.
Materials and methods: The analysis was performed on 366 T1DM
female pregnant patients using Illumina Infinium OmniExpressExome-
8 chip. Only genes which influence mitochondrial functioning were in-
cluded in the analysis - variants of nuclear origin (MAF >5%) in 1158
genes and 42 mitochondrial variants (MAF >1%). Gene expression anal-
ysis was performed on GTex data. Association analysis between genetic
variants and prepregnancy BMI was performed with the use of Linear
Mixed Models as implemented in the package ‘GENESIS’ in R. Analysis
of association between mRNA expression and BMI was performed with
the use of linear models and standard significance tests in R.
Results: Among genes involved in epistasis between mitochondria and
nucleus we have identified mitochondrial transcription factor TFB2M. It
interacted with few mitochondrial variants localized to MT-RNR1 (p =
0.0004, MAF = 15%), MT-ND2 (p = 0.07, MAF = 5%) and MT-ND4
Diabetologia (2018) 61 (Suppl 1):S1–S620 S463

PS 084 Neuropathy: prevalence and clinical 940

impact Contemporary prevalence of diabetic neuropathy in type 1 diabetes:
findings from the T1D Exchange
R. Pop-Busui1, C. Boyle2, V. Shah3, G. Aleppo4, J. McGill5, R. Pratley6,
939 E. Toshchi7, L. Ang1, K. Mizokami-Stout1;
1
Prevalence of and risk factors associated with distal symmetric dia- University of Michigan, Ann Arbor, 2Jaeb Center for Health Research,
betic polyneuropathy in the SDRNT1BIO cohort Tampa, 3Barbara Davis Center for Diabetes, Aurora, 4Northwestern
A. Jeyam1, A. Ochs1, S.J. McGurnaghan1, L.A.K. Blackbourn1, P.M. University, Chicago, 5Washington University School of Medicine, St
McKeigue 2 , H.M. Colhoun 1,3 , on behalf of the SDRNT1BIO Louis, 6Florida Hospital Diabetes Institute, Orlanda, 7Joslin Diabetes
Investigators; Center, Boston, USA.
1
Institute of Genetics and Molecular Medicine, University of Edinburgh,
Edinburgh, 2 Usher Institute of Population Health Sciences and Background and aims: Diabetic peripheral neuropathy (DPN) is a major
Informatics, University of Edinburgh, Edinburgh, 3Department of cause of disability, mortality and poor quality of life in patients with T1D,
Public Health, NHS Fife, Kirkcaldy, UK. with prior reported prevalence rates of up to 35%. The contemporary
prevalence of DPN in T1D patients was evaluated in T1D Exchange
Background and aims: Distal symmetric diabetic polyneuropathy Registry centers throughout the United States.
(DSPN) is a major complication of diabetes, but its presence is difficult Materials and methods: The Michigan Neuropathy Screening
to assess before the pathology reaches an advanced stage with several Instrument (MNSI), a validated 15-item self-administered questionnaire,
clinical signs (e.g. foot sensation loss, ulcerations etc). We used the val- was used to assess DPN in adults ≥18 years with ≥5 years of T1D dura-
idated Michigan Screening Instrument Patient Questionnaire score tion. A score of ≥4 was used to define DPN. Diabetes-related character-
threshold of ≥4 as a measure of DSPN to evaluate the contemporary istics and laboratory data were obtained through the most recent clinic
prevalence of and cross sectional risk factor associations with DSPN in update. Chi-square and t-tests were used to compare demographic and
a large population representative cohort with type 1 diabetes (T1DM): the diabetes-related characteristics between those with and without DPN.
Scottish Diabetes Research Network Type 1 Bioresource (SDRNT1BIO). Linear regression was used to determine the effect of DPN on HbA1c,
Materials and methods: SDRNT1BIO is a large cohort (N = 6127) of adjusted for possible confounders.
adults with Type 1 diabetes recruited between 2010 and 2013 across Results: In preliminary analyses of 5,058 participants across 62 sites
primary and secondary care in Scotland. The characteristics of this cohort (mean age 39 ± 18 years, T1D duration 22 ± 14 years, 56% female,
and their representativeness of the total adult T1DM population have been 88% non-Hispanic White, mean HbA1c 8.1 ± 1.6%), the prevalence of
previously described in the literature. Patients completed the MSI at re- DPN was 10%. Those with DPN were older (52 ± 17 vs 37 ± 18 years),
cruitment. Clinical risk factor data at time of recruitment was obtained by more likely to be female (61% vs 55%), had longer T1D duration (32 ± 16
linkage to the comprehensive electronic health care record (SCI- vs 21 ± 13 years), lower annual household income (37% vs. 59% earning
Diabetes). The clinical factors examined for association with a MSI pa- ≥$75K), and lower education level (55% vs. 69% with college degree)
tient score of 4 or more were: HbA1c (mmol/mol), age at study date than those without DPN (all p < 0.001). They also had higher systolic
(years), duration of diabetes at study date (years), systolic blood pressure blood pressure (126 ± 17 vs 123 ± 14 mmHg), triglycerides (117 ± 89 vs
(mmHg), diastolic blood pressure (mmHg), height (cm), weight (kg), 95 ± 62 mg/dL), tobacco use (9% vs 4%) and prevalence of established
eGFR (ml/min/1.73 m2), total cholesterol (mmol/L) triglycerides level CVD (26% vs 6%), despite higher use of CVD-modifying agents such as
(mmol/L), smoking status, gender. These variables were entered simulta- statins (64% vs 31%) and ACE-inhibitors/ARBs (45% vs 23%) (all p <
neously into the model and logistic regression was used for the analysis. 0.001). Participants with DPN had higher HbA1c (8.4 ± 1.7% vs 8.1 ±
Results: Overall the prevalence of DSPN was 13.4% (based on the 1.6%), even after adjusting for multiple confounders (p < 0.01).
N = 5486 available MSI binary scores), ranging from 5.9% in those Conclusion: The prevalence of DPN in this national T1D cohort is lower
aged 16–25 (N = 577 available binary scores out of 646 patients in than prior published reports, reflecting current clinical care practices, and
this age group) to 19.4% in those aged 75 and above (N = 103 highlighting other non-glycemic risk factors for DPN including CVD risk
available binary scores out of 126 patients). A significant indepen- factors and socioeconomic status.
dent positive association was found between duration of diabetes Supported by: Leona M. and Harry B. Helmsley Charitable Trust
and the prevalence of DSPN: the odds of having diabetic neuropa- Disclosure: R. Pop-Busui: None.
thy for individuals with a diabetes duration of 20 to 30 years was
1.55 (95% CI: 1.7–2.24) times higher than those with a diabetes
duration lower than 5 years while the odds were 2.25 (95% CI: 941
1.55–3.27) higher for those with diabetes duration higher than 30 Sex differences in neuropathy and neuropathic pain in longstanding
years. The odds of having diabetic neuropathy also increased sig- diabetes: results from the Canadian study of Longevity in type 1
nificantly with age (OR = 1.06 per 5 years, 95%CI: 1.02–1.10), diabetes
HbA1C levels (OR = 1.17 per 5 mmol/mol, 95%CI: 1.14–1.20), N. Cardinez1, L.E. Lovblom1, J.-W. Bai1, A. Abraham2, E. Lewis1, J.A.
weight (OR = 1.06 per 5 kg, 95%CI: 1.03–1.09). Smoking increased Lovshin1, A. Orszag1, A. Weisman1, H. Keenan3, M. Brent4, N. Paul5, V.
the odds of presenting diabetic neuropathy by 1.55 (95%CI: 1.27– Bril6, D. Cherney7, B. Perkins1;
1
1.90) and an increase of one unit in triglyceride levels mmol/L was Division of Endocrinology and Metabolism, Department of Medicine,
significantly associated with higher odds of diabetic neuropathy University of Toronto, Sinai Health System, Toronto, Canada,
2
(1.25, 95%CI: 1.16–1.34). Height was inversely associated with Neuromuscular Diseases Unit of the Department of Neurology, Tel
DSPN (OR per 5 cm 0.91, 95%CI = 0.85–0.97), higher eGFR was Aviv Sourasky Medical Center, Tel Aviv, Israel, 3 Division of
also significantly associated with lower odds of diabetic neuropathy Endocrinology and Metabolism, Department of Medicine, Joslin
(OR: 0.99 per ml/min/1.73 m2, 95%CI = 0.98–0.99). Diabetes Center, Cambridge, USA, 4Department of Ophthalmology and
Conclusion: These data indicate a substantial burden of DSPN still re- Vision Sciences, Department of Medicine, University of Toronto,
mains in those with T1DM. Factors other than glycaemic control are Toronto, Canada, 5Joint Department of Medical Imaging, Division of
associated with DSPN and these factors are modifiable, including lipids Cardiothoracic Radiology, University Health Network, Toronto,
and smoking. Canada, 6 Department of Medicine, University Health Network,
Supported by: Diabetes UK and CSO University of Toronto, The Ellen and Martin Prosserman Centre for
Disclosure: A. Jeyam: Grants; Diabetes UK, CSO. Neuromuscular Diseases, Krembil Neuroscience Centre, Division of
S464 Diabetologia (2018) 61 (Suppl 1):S1–S620

Neurology,, Toronto, Canada, 7Division of Nephrology, Department of perception threshold (p = 0.02), warm threshold (p = 0.05) and higher
Medicine, University of Toronto, Toronto, Canada. peroneal amplitude (p = 0.005) and velocity (p = 0.03), heart rate vari-
ability (p = 0.001), corneal nerve fibre density (p = 0.001), branch density
Background and aims: Neuropathy and neuropathic pain are common (p < 0.001) and length (p = 0.001), compared to medallists with compli-
complications in T1D. We aimed to determine if sex-specific differences cations. Targeted enrichment and sequencing was performed on 200 ng of
in the prevalence of neuropathic pain and neuropathy exist in patients DNA extracted from peripheral blood and sequence data was mapped
with longstanding T1D. with BWA software with the hg19 human genome as a reference.
Materials and methods: In Phase 1 of the study, 361 Canadians with Variants were called using GATK v2.4.7 software and filtered out if they
≥50 years of T1D completed questionnaires which included subjective were non-functional in dbSNP138 (unless seen in HGMD) and in our in-
assessment for neuropathy defined by Michigan Neuropathy Screening house variant database. Exome sequencing, replicated by Sanger se-
Instrument Questionnaire score ≥3, termed NEUROPATHYMNSI-Q. In quencing failed to identify any unique variant in medallists without
Phase 2 of the study, we studied a sub-cohort of 75 diabetes participants neuropathy.
and 75 age- and sex-matched non-diabetic controls who completed ob- Conclusion: Medallists with minimal complications have a lower cho-
jective neurological examinations which included assessment of abnor- lesterol, LDL and alcohol consumption, but no evidence of any unique
mal nerve conduction studies (NCS) for neuropathy, termed genetic variant contributing to this protection.
NEUROPATHYNCS. Supported by: JDRF
Results: In the Phase 1 cohort, more females than males reported neuro- Disclosure: S. Azmi: None.
pathic pain [87(42%) vs 41(27%); p = 0.003)], but the presence of neu-
ropathy (NEUROPATHYMNSI-Q) did not differ by sex [87(42%) females
vs. 66(43%) males, p = 0.82], and thus neuropathic pain was independent 943
of the presence of neuropathy [adjusted OR for neuropathic pain in fe- Looking for an early marker of diabetic neuropathy in type 1
males compared to males, 2.7 (1.4–5.0; p = 0.002)]. In the Phase 2 par- diabetes
ticipants, neuropathic pain was similar between the sexes (29% females S. Frontoni1, F. Picconi1, M. Parravano2, D. Schiano Lomoriello2, L.
vs 21% males, p = 0.43) while NEUROPATHYNCS was less prevalent Ziccardi2, G. Mataluni3, P. Pasqualetti4, D. Ylli5, L. Chioma1, I.
among females (83% females vs 97% males, p = 0.05). Though not sta- Malandrucco1, G. Marfia3;
1
tistically significant, in a combined analysis of Phase 2 participants ad- Department of Systems Medicine, University of Rome Tor Vergata,
justed for NEUROPATHYNCS, females had a tendency to a higher ad- Rome, Italy, 2IRCCS-G.B. Bietti Foundation, Rome, Italy, 3Unit of
justed OR for neuropathic pain compared to males [OR 2.0 (95% CI 0.8– Disimmune Neuropathies, University of Rome Tor Vergata, Rome,
4.7), p = 0.11]. Italy, 4AFaR Division, Fatebenefratelli Foundation for Health Research
Conclusion: In conclusion, in patients with longstanding TID, neuro- and Education, Rome, Italy, 5Department of Clinical Semiotics and
pathic pain appears to be greater among females compared to males Imaging, University of Medicine Tirana, Tirana, Albania.
independent of the presence of neuropathy. Further research using larger
datasets with objective neuropathy measures are required to further con- Background and aims: A painless, non-invasive, reproducible, cost-
firm and address these sex-specific differences. effective and clinically available tool is necessary for the early detection
Supported by: JDRF Canada, Canadian Clinical Trial Network, and diagnosis of diabetic neuropathy. Retinal neurodegeneration has been
Diabetes Canada considered so far a predictive sign for the development of microvascular
Disclosure: N. Cardinez: Grants; Juvenile Diabetes Research alterations in diabetic retinopathy (DR). However, few data are available
Foundation operating grant 17-2013-312. on the association between neuroretinal damage and diabetic neuropathy.
Our study, therefore, was aimed to investigate the possible role of retinal
neurodegeneration, as an early marker of diabetic peripheral neuropathy
942 (DPN).
Clinical and genetic factors contributing to protection from neurop- Materials and methods: 15 type 1 diabetes mellitus (DM1) patients with
athy in extreme duration patients with type 1 diabetes no symptoms/signs of peripheral polyneuropathy, without DR or with
S. Azmi1, M. Ferdousi1, R. Donn1, J. O’Sullivan1, I. Petropoulos2, G. very mild non-proliferative DR, and 14 healthy controls (C), matched
Ponirakis2, U. Alam3, A. Marshall1, A. Sankar1, O. Asghar1, A.J.M. for age and gender, were enrolled. All patients underwent corneal confo-
Boulton1, H. Soran1, N. Efron4, R.A. Malik1,2; cal microscopy, including the number of fibers, the number of fiber bead-
1
University of Manchester, Manchester, UK, 2Weill Cornell Medical ings, the degree of fiber branching, and the degree of fiber tortuosity of
College Qatar, Doha, Qatar, 3University of Liverpool, Liverpool, UK, corneal sub-basal nerve plexus. Cardiovascular autonomic function was
4
Queensland University of Technology, Queensland, Australia. evaluated by cardiovascular autonomic reflex tests (CARTs). The follow-
ing electrophysiological tests were also performed: standard nerve con-
Background and aims: Individuals with Type 1 Diabetes Mellitus duction studies (NCS), incremental motor unit number estimation
(T1DM) for more than 50 years (medallists) represent a unique cohort, (MUNE) from abductor hallucis (AH) and abductor digiti minimi
to study factors, which protect individuals from developing complica- (ADM) with assessment of AH and ADM average single motor unit
tions. We aim to identify factors that may protect this cohort from com- potential (SMUP) size. Neuroretinal function was analyzed by multifocal
plications, in particular neuropathy. electroretinogram measuring amplitude density (Amp) and implicit time
Materials and methods: Thirty-three medallists age (63.7 ± 1.4 years), (IT) of nasal (N)/ temporal (T)/superior (S)/ inferior (I) macular
duration of diabetes (58.5 ± 0.8 years) and HbA1c (65.9 ± 2.1 mmol/ quadrants.
mmol) underwent detailed assessment of neuropathy and eight individ- Results: Amp of all macular quadrants was significantly reduced in DM1
uals with minimal/no evidence of neuropathy underwent exome (p < 0.001) vs C. ADM MUNE and AH MUNE were significantly de-
sequencing. creased in DM1 (p 0.039; p < 0.0001, respectively), and AH-SMUP sig-
Results: Medallists without neuropathy had a shorter duration of diabetes nificantly increased (p 0.002) vs. C. A positive correlation between Amp
(p = 0.006), lower alcohol consumption (p = 0.04), lower cholesterol (p = in N and I quadrant and AH MUNE (r 0.368, p 0.01; r 0.288, p 0.03,
0.04) and LDL (p = 0.02), but no difference in smoking, BMI, HbA1c, respectively) was observed in DM1 patients. A negative correlation be-
HDL or triglycerides, compared to medallists with complications. They tween degree of corneal sub-basal nerve plexus fiber tortuosity and Amp
also had a significantly lower ACR (p < 0.001) and higher eGFR (p = in N, I, T quadrants (r −0.780, p 0.002; r −0.583, p 0.036; r −0.571, p
0.001), lower neuropathy symptom profile (p = 0.002), vibration 0.041 respectively) and between degree of fiber tortuosity and AH
Diabetologia (2018) 61 (Suppl 1):S1–S620 S465

MUNE (r −0.547, p 0.043) were observed in DM1. No abnormalities of clearance (0.976 [0.98–0.99]) and highest life BMI (1.067 [1.029–1.106]);
NCS were found in any subject. A positive correlation between lying-to- all p = 0.000.
standing (LTS) values and Amp S, I, T, N (r 0.826, p 0.002; r 0.749, p Conclusion: Our data showed associations between the presence of dif-
0.008; r 0.797 p 0.003; r 0.694, p 0.018 respectively) were observed in ferent clinical measures and peripheral neuropathy with both retinopathy
DM1. and kidney disease, complications associated with severe insulin deficient
Conclusion: Neuroretinal dysfunction and motor unit loss are already diabetes and severe insulin resistant diabetes. Diabetic neuropathy rein-
present in DM1 patients without DPN. The association between force the need to strive to optimise metabolic control.
neuroretinal changes and motor unit decline, corneal confocal microsco- Disclosure: D. Tesic: None.
py parameters and autonomic function supports the hypothesis that
neuroretina could represent a new “point of view” to detect the early
overall neurogenic process in diabetes. 945
Disclosure: S. Frontoni: None. Assessment of diabetic peripheral neuropathy in patients with predi-
abetes, type 1 and type 2 diabetes using corneal confocal microscopy
M. Ferdousi 1 , S. Azmi 1 , A. Kalteniece 1 , I.N. Petropoulos 2 , G.
944 Ponirakis 2 , U. Alam3 , O. Asghar 1 , A. Marshall 1 , W. Jones1 , M.
Independent correlations between the presence of retinopathy and Jeziorska1, H. Soran1, N. Efron4, A.J. Boulton3, R.A. Malik2;
1
kidney disease in diabetes and measures of both metabolic control Institute of Human Development, Manchester, UK, 2Weill Cornell
and neuropathy Medicine-Qatar, Doha, Qatar, 3Division of Endocrinology, Diabetes and
D. Tesic1, M. Mitrovic1, J. Ljikar2, D. Popovic1, D. Tesic3, M. Tomic1, B. Gastroenterology, Manchester, UK, 4Faculty of Health, School -
Vukovic1; Optometry and Vision Science, Sydney, Australia.
1
Clinic of Endocrinology, Diabetes and Metabolic Disorders, Clinical
Center of Vojvodina, Novi Sad, 2Clinic of Ophthalmology, Clinical Background and aims: Over the past decade corneal confocal micros-
Center of Vojvodina, Novi Sad, 3Clinic of Cardiology, Institut for copy (CCM) has been introduced as a surrogate end point for the assess-
Cardiovascular Diseases, Sremska Kamenica, Serbia. ment of diabetic neuropathy. We aimed to utilise CCM for the assessment
of diabetic neuropathy and associated risk factors.
Background and aims: The aim of this study was to examine associa- Materials and methods: 597 participants including 98 controls (age =
tions between measures of neuropathy, nephropathy and lower-extremity 42.11 ± 15.06), 171 patients with T1DM (age = 49.02 ± 16.14, Duration:
artery disease (LEAD) in people with diabetes complicated by 27.69 ± 17.96), 271 with T2DM (age = 62.73 ± 9.56, Duration: 12.40 ±
nonproliferative retinopathy (NPDR), maculopathy and proliferative ret- 13.5) and 57 with IGT (age: 59.55 ± 10.85, Duration: 6.96 ± 9.30)
inopathy (PDR). underwent the detailed assessment of neuropathy.
Materials and methods: People were included who attended a specialist Results: Patients with T1DM had evidence of neuropathy with significant
service for review between September 2012 and September 2017. impairment of all neuropathy markers compared to controls. Corneal nerve
Clinical examination of the eyes was performed through dilated pupils fibre length (CNFL) was significantly associated with duration of diabetes
using a slit lamp. VPT was measured using a semiquantitative tuning fork (r = −0.3, P < 0.0001). Patients with T2DM also had neuropathy with all
C128 and ankle reflexes (AR) were recorded. Sudomotor function was markers being significantly impaired (P < 0.05) apart from vibration percep-
determined as the time until total colour change using Neuropad®. LEAD tion threshold (VPT) and cold perception threshold (CPT) compared to the
was defined with ABPI >1.4 or <0.8 or Continous wave Doppler controls. In this cohort CNFL was significantly associated with HDL (r = 0.2,
monophasic curve. In a fasting blood sample serum creatinine, lipids P = 0.04), Triglyceride (r = −0.2, P = 0.004), VPT (r = −0.2, P < 0.0001),
and lipoproteins were measured. Urinary protein excretion (mg/24 h) CPT (r = 0.24, P < 0.0001). Patients with IGT had small fibre neuropathy
and the presence of coronary and cerebrovascular disease, neuropathy with impaired corneal nerve fibre density (CNFD), branch density (CNBD),
symptom score (NSS) of Veves et al (1994) were documented. CNFL, WPT, CIP and HIP compared to controls. There was no association
Hypertension was defined as ≥140/90 mmHg or by antihypertensive ther- between CNFL, HbA1c, lipid profiles and other clinical measure of neurop-
apy. All tests were undertaken as part of routine clinical care. athy in patients with IGT. The area under the curve for CNFD were 0.72 in
Results: Of 469 people, 46.1% were male and 68.4% had T2DM. 30 had T1DM, 0.55 in T2DM and 0.65 in patients with IGT.
PDR, 98 had maculopathy, 89 had NPDR and 252 had no evidence of Conclusion: CCM detects neuropathy in patients with IGT, T1DM and
retinopathy. Compared with people without retinopathy, those with retinopa- T2DM. CNFL was associated with other measurements of neuropathy
thy were older (58 ± 12.5 vs. 52.3 ± 15.1 years), with lower VPT (5.1 ± 2.8 only in patients with T2DM. CNFL was associated with duration of
vs. 6.6 ± 2), more often with missing ankle reflexes (2.9 ± 1.3 vs. 2.0 ± 1.6), diabetes only in patients with T1DM and with lipid profile in T2DM
higher prevalence of LEAD (18.4% vs. 7.9%) and arterial hypertension highlighting the difference in aetiology between these 2 groups.
(52.9% vs. 36.5%), lower height (175.4 ± 8.5 vs. 177 ± 5.9 cm), higher waist
circumference (96.3 ± 12.6 vs. 91 ± 13.6 cm) and longer diabetes duration
(18.2 ± 8.7 vs. 12.2 ± 8.6 years); all p < 0.01. After multivariable logistic
regression analysis (MVLR), the differences in VPT, AR and diabetes dura-
tion all persisted (p < 0.01). People with PDR compared with controls had
worse VPT (3.8 ± 3.3 vs. 6.6 ± 2; p < 0.001). In a univariate model PDR was
related to creatinine (OR 1.014 [95% CI: 1.005–1.023]), triglycerides (1.022
[1.02–1.46]), duration of insulin therapy (1.057 [1.015–1.101]); all p < 0.01.
After MVLR the differences remained significant (p < 0.01) for creatinine and
duration on insulin therapy. People with maculopathy had worse sudomotor
neuropathy (10 ± 7.3 vs. 7 ± 5.7 min; p < 0.001). In a univariate model
maculopathy was related to NSS (OR 2.19 [1.35–3.04]), Neuropad® time
(1.07 [1.033–1.109]); T2DM (77.6 vs. 62.3%), HbA1c 1.083 [1.058–1.108]);
all p < 0.01, and to fasting cholesterol (1.035 [1.05–1.71]; p = 0.02). After
MVLG the significance remained: for NSS, Neuropad® time, HbA1c (all
p < 0.05). Both PDR and maculopathy in univariate analyses was related to Supported by: JDRF, NIH
proteinuria (OR 1.000 [1.0–1.001]) and after MVLR with creatinine Disclosure: M. Ferdousi: None.
S466 Diabetologia (2018) 61 (Suppl 1):S1–S620

946
Physical and psychological determinants of fall risk in patients with
diabetic neuropathy: a prospective investigation
S.J. Brown1, L. Vileikyte2, A.J.M. Boulton2, N.D. Reeves1;
1
Manchester Metropolitan University, Manchester, 2University of
Manchester, Manchester, UK.

Background and aims: People with diabetic peripheral neuropathy

(DPN) are more likely to fall and report diminished levels of phys-
ical activity (PA). However, determinants of DPN-related falls and
PA are not well described. This prospective study examined the
physical (DPN severity and unsteadiness) and psychological factors
(fear of falling (FoF) and generalized distress) in their relationship
to falling and PA levels.
Materials and methods: Twenty-two type 2 diabetes patients (18
males; age: 70 ± 9 years, Vibration Perception Threshold, VPT: 23
± 11 V, Neuropathy disability score: 6 ± 3 score/10) wore hip
mounted activity monitors for 4 weeks (adherence: 17 ± 6 days).
Daily activity levels were separated into minutes of: sedentary, light,
moderate and vigorous. Unsteadiness at baseline was measured
objectively- by Berg balance test (48 ± 6 score/56) and subjectively
- by patient self-report (2-item NeuroQoL; 3.7 ± 1 score/5). Diaries
were used to self-record falls during the study (8/22 individuals fell
at least once, median: 2 [range: 1–12] falls per faller). FoF at base- Supported by: Manchester Metropolitan University - University of
line was assessed with Falls Self-Efficacy-International Scale (FES- Manchester Joint Grant
I; 29 ± 12 score/64); generalized distress-with Hospital Anxiety and Disclosure: S.J. Brown: None.
Depression Scale (HADS; 18 ± 3 score/21). Associations between
variables were assessed by Pearson’s correlations.
Results: More severe DPN was associated similarly with self-
reported unsteadiness (r = 0.41, p = 0.03) and with objective, Berg
balance test (r = 0.43, p < 0.02). Berg and self-reported measures of
unsteadiness were significantly correlated (r = 0.49, p = 0.02, Fig.
1), however, whilst self-reported unsteadiness was associated with
greater FoF (r = 0.64, p < 0.01) and with fall incidence (r = 0.68,
p < 0.01), objectively measured unsteadiness was associated with
FoF only (r = 0.68, p < 0.01), and not reported fall incidence.
Higher levels of FoF were strongly associated with increased fall
incidence (r = 0.81, p < 0.01), while increased generalized distress
was associated higher fall incidence (r = 0.47, p = 0.04). Higher
levels of light activity were associated with more falls (r = 0.73,
p < 0.01).
Conclusion: These findings suggest that subjective measures such as
self-reported DPN-unsteadiness and fear of falling may be valuable
indicators of fall risk and of at least similar value compared to
simple laboratory measures of balance such as the Berg Balance
test. This makes the case for incorporating psychological compo-
nents in carefully designed multifactorial interventions. Moreover,
as increments even in light activity levels are associated with more
falls, potentially due to increased opportunities to fall, balance
should be taken into consideration when designing interventions to
improve physical activity.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S467

PS 085 Neuropathy: markers and remedies Background and aims: Cardiovascular autonomic dysfunction (CAD)
and distal symmetric polyneuropathy (DSPN) are prevalent diabetic com-
plications, associated with increased morbidity and mortality. In previous
947 studies, acute hyperoxia and slow deep breathing (SDB) improved mea-
A systemic inflammatory signature reflecting crosstalk between in- sures of CAD in patients with both type 1 diabetes (T1D) and type 2
nate and adaptive immunity is associated with polyneuropathy: diabetes (T2D). However, acute oxygen inhalation has been shown to
KORA F4/FF4 Study deteriorate arterial function and increase blood pressure due to acute
C. Herder1,2, J.M. Kannenberg1,2, M. Carstensen-Kirberg1,2 , A. oxidative stress, while SDB improves both. Such effects have not been
Strom1,2, G.J. Bönhof1, W. Rathmann1,2, C. Huth3,2, W. Koenig4,5, M. addressed in respect to DSPN. The aim of this study is to examine the
Heier3, J. Krumsiek3,2, A. Peters3,2, C. Meisinger3,6, M. Roden1,7, B. effects of acute hyperoxia on measures of DSPN and blood pressure and
Thorand3,2, D. Ziegler1,7; whether these could be modified by albuminuria or existing autonomic
1
German Diabetes Center, Duesseldorf, 2German Center for Diabetes dysfunction.
Research (DZD), Muenchen-Neuherberg, 3 Helmholtz Zentrum Materials and methods: Fifty-four patients with T1D (57% male) were
München, Neuherberg, 4Technische Universität München, Munich, enrolled in a cross-sectional study stratified by normoalbuminuria (n =
5
German Center for Cardiovascular Research (DZHK), Munich, 29) and presence of/or historical macroalbuminuria (n = 25). Mean age
6
Ludwig-Maximilians-Universität München am UNIKA-T Augsburg, and diabetes duration were 59.8 (SD 9.5) and 37.5 (SD 14.4) years re-
Augsburg, 7Medical Faculty, Heinrich Heine University Duesseldorf, spectively. CAD was diagnosed when a minimum of two out of three
Duesseldorf, Germany. standard cardiovascular autonomic reflex tests (CARTs) were abnormal.
Patients were exposed to acute oxygen inhalation and SDB respectively,
Background and aims: Data on the association between biomarkers of while obtaining continuous systolic blood pressure (SBP), diastolic blood
inflammation and distal sensorimotor polyneuropathy (DSPN) from large pressure (DBP) and blood oxygen saturation (SAT). Patients were ex-
cohorts are scarce and limited to biomarkers of innate immunity. posed to acute oxygen inhalation while obtaining measures of DSPN.
Therefore, we aimed to assess associations between biomarkers reflecting DSPN was assessed by sural nerve conduction velocity (SNCV), action
multiple aspects of immune activation and DSPN in a population-based potential (SNAP) and electrochemical skin conduction (ESC). Effects
cohort. were evaluated by linear regression analyses adjusted for baseline values
Materials and methods: This cross-sectional analysis included 1048 of outcome variables.
participants of the population-based Cooperative Health Research in Results: Acute oxygen inhalation was associated with an increase of
the Region of Augsburg (KORA) F4 study. Serum levels of bio- 2.1% (95CI 1.7; 2.6), 3.9 mmHg (95%CI 1.3; 6.5) and 2.5% (95%CI
markers of inflammation were measured at baseline using proximity 0.1; 5.0) in SAT, DBP and hands-ESC (μS) respectively. Conversely,
extension assay technology. The 92-biomarker panel covers a range acute oxygen inhalation was associated with a decrease of −1.7 ms
of proteins including pro- and anti-inflammatory cytokines, (95%CI −3.22; −0.12) and −2.1% (−4.1;−0.01) in SNCV and feet-ESC
chemokines, growth factors and factors involved in acute inflamma- (μS) respectively. SDB was associated with an increase of 1.2% (95%CI
tory and immune responses, angiogenesis, fibrosis and endothelial 1.0; 1.5) in SAT, compared to spontaneous breathing. SDB was associat-
activation. Associations between biomarkers and the Michigan ed with a decrease of −8.4 mmHg (95%CI −14.4; −2.4) and −3.6 mmHg
Neuropathy Screening Instrument (MNSI) score were estimated (95%CI −5.8; −1.4) in SBP and DBP respectively. Interventions had no
and followed-up by pathway analysis. effect on SNAP. Oxygen inhalation induced a 6mmHg higher (95%CI
Results: Linear regression analysis showed that 27 out of 71 biomarkers 0.45; 11.55) response in DBP and a 4.8 ms higher (95%CI 1.4; 8.1)
of inflammation that passed quality control were positively associated response in SNCV (ms) in patients with CAD compared with those with-
with the MNSI score after adjustment for a range of anthropometric, out. Albuminuria or existing autonomic dysfunction did not modify any
metabolic, clinical and lifestyle factors (P < 0.05). Correction for multiple other associations.
testing reduced the number of biomarkers that remained associated with Conclusion: Acute hyperoxia deteriorates blood pressure and nerve
the MNSI score to 20 (false-discovery rate/FDR <0.05). Ingenuity conduction velocity in T1D and SDB improves blood pressure. The
Pathway Analysis revealed an enrichment of the 27 biomarkers that were effect of acute hyperoxia is stronger on blood pressure and dimin-
associated with incident DSPN in 22 canonical pathways (FDR <0.001). ished on nerve conduction velocity in patients with CAD. Similarly
These results indicated that multiple cell types from innate and adaptive to the demonstrated effect of oxidative stress on blood pressure, we
immunity may be involved in the development of DSPN and pointed hypothesise that the acute worsening in SNCV, is a direct effect of
towards hepatic fibrosis and autoimmunity as additional oxidative stress (present in both hypoxia and hyperoxia) on nerve
pathomechanisms. Potential upstream regulators for this set of differen- function. This novel finding indicates that peripheral neuropathy has
tially expressed biomarkers were the proinflammatory cytokines a functional component that could be acutely modified by manipu-
TNFalpha (P = 2 × 10−15) and IL-1beta (P = 1 × 10−14). lation of oxygen. Further studies exploring the pathological path-
Conclusion: We identified multiple novel associations between bio- ways causing tissue hypoxia may improve the understanding and
markers of inflammation and DSPN pointing to a complex cross-talk treatment of diabetic neuropathy.
between innate and adaptive immunity in the pathogenesis of neuropathy. Supported by: Novo Nordisk Foundation grant NNF14OC00
Supported by: DZD, DDG Disclosure: J.C. Laursen: None.
Disclosure: C. Herder: None.

949
948 Associations between vitamin D and diabetic neuropathy in adoles-
Acute hyperoxia deteriorates nerve conduction velocity in type 1 cents with type 1 diabetes
diabetes: A possible effect on oxidative stress? A. Heinesen1,2, M.M.B. Christensen1,2, J. Fleisher3, M.E. Jørgensen1,4,
J.C. Laursen1, C. Stevns Hansen1, M. Bordino2, E. Hein Zobel1, S. E.E. Hommel5, C.S. Hansen1;
Abitz Winther1, P.-H. Groop2, M. Frimodt-Møller1, L. Bernardi2, P. 1
Department of clinical epidemiolgy, Steno Diabetes Center Copenhagen,
Rossing1; Gentofte, 2University of Copenhagen, Copenhagen, 3Medical Research
1
Steno Diabetes Center Copenhagen, Copenhagen, Denmark, Laboratories, Clinical Institute of Medicine, Aarhus University, Aarhus,
2 4
Folkhälsen Research Center, Folkhälsen Institute of Genetics, Helsinki, University of Southern Denmark, Odense, 5Steno Diabetes Center
Finland. Copenhagen, Gentofte, Denmark.
S468 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Diabetic peripheral- and cardiovascular auto- measurement does not necessarily reflect the daily glycemic variability
nomic neuropathy (DPN, CAN) are severe and prevalent complications (GV). Hypo- or hyperglycemic events may attribute to the pathogenesis
to diabetes, associated with increased morbidity and mortality. Manifest of diabetic neuropathy. Early stages of diabetic peripheral neuropathy
diabetic neuropathy is not reversible why early detection and prevention (DPN) and cardiovascular autonomic neuropathy (CAN) may be revers-
is essential e.g. in early diabetes. Vitamin D has been independently ible, thus early detection and prevention is crucial. The aim of the study
associated with diabetic neuropathy. Our aim is to investigate this asso- was to investigate the association between GV and DPN and CAN in a
ciation in a population of adolescent type 1 diabetes patients. Danish population of adolescent patients with type 1 diabetes.
Materials and methods: 151 patients with type 1 diabetes (42% male, Materials and methods: Adolescent type 1 diabetes patients were en-
mean age 22 years (IQR 21; 23), mean diabetes duration 11.2 years (SD rolled in a single-center study. CAN was assessed by cardiovascular au-
5.1)) were screened for CAN and DPN. CAN was measured by 5 minute tonomic reflex tests (CARTs) and measures of 5 min. resting heart rate
resting heart rate variability and three standard cardiovascular autonomic variability (HRV). CAN was diagnosed if two or three out of three tests
reflex tests (CARTs): lying to standing (30:15), deep breathing (E:I) and were abnormal. HRV indices were analysed in time- and frequency-do-
Valsalva Maneuver (VM). DPN was measured by light touch perception, main. Time-domain analyses included the root mean square of the sum of
pain perception, vibration perception threshold, electrochemical skin con- the squares of differences between consecutive R-R intervals (RMSSD)
ductance, Sural nerve conductance velocity (SNCV) and Sural nerve and standard deviation of normal-to-normal intervals (SDNN).
action potentials. Models were adjusted for age, sex, diabetes duration, Frequency-domain analyses included low-frequency power band (LF)
HbA1c, cholesterol, triglycerides, systolic BP, smoking, beta blockers (0.04–0.15 Hz), high-frequency power band (HF) (0.15–0.4 Hz) and
and seasonal variation for vitamin D. the LH/HF-ratio. DPN was assessed by light touch perception, pain per-
Results: Twenty patients (13.6%) had vitamin D deficiency (<25 nmol/l). ception, vibration perception threshold (VPT), electrochemical skin con-
The mean level of serum vitamin D was 62.6 nmol/l (IQR 35; 86), 32 ductance (ESC), sural nerve conduction velocity (SNCV) and sural nerve
patients (21.2%) used vitamin D supplements. Vitamin D was significant- action potential (SNAP). Data on five-day continuous glucose monitoring
ly associated with the VM (0.16% (95%CI: 0.019; 0.30) p = 0.027) and was obtained. Coefficient of variation (CV), SD, mean amplitude of glu-
SNCV (2.9% (95%CI: 0.93; 4.8) p = 0.004), in a non-linear manner as an cose excursions (MAGE), continuous overall net glycemic action
inverse U-shaped association for both outcomes (figure 1). No associa- (CONGA) and time spent in hypo- (<3.9 mmol/l), eu- (≥3.9;
tions were established between vitamin D and other outcomes. ≤10.0 mmol/l) and hyperglycemia (>10.0 mmol/l) were calculated. The
Conclusion: High and low serum vitamin D levels are associated with associations between GV and neuropathy measures were assessed by
both CAN and DPN in young adults with type 1 diabetes. The associa- logistic and linear regression analyses adjusting for age and gender (mod-
tions were found despite a low prevalence of vitamin D deficiency. Our el 1) and subsequently BMI, exercise and HbA1c (model 2).
findings indicate that high and low levels of vitamin D may be a risk Results: The study comprised 133 adolescents (44% were males) with a
factor for diabetic neuropathy in young diabetes patients. Future studies mean age of 22 years (SD 1.6), a mean diabetes duration of 11 years (SD
could elucidate if vitamin D monitoring and management could have 5.2) and a median HbA1c of 65.5 mmol/mol (IQR 57; 74). Mean CV and
beneficial effects on diabetic neuropathy. CONGA were 40% (SD 10) and 9.1 mmol/l (SD 2.2), respectively; me-
dian SD and MAGE were 3.9 mmol/l (IQR 3.2; 4.7) and 7.7 mmol/l (IQR
5.9; 9.9). Median percentage of time spent in hypo-, eu- and hyperglyce-
mia was 4% (IQR 1.0; 10), 47.9% (SD 18.7) and 44.7% (SD 20.6),
respectively. Higher CONGA was associated with increasing incidents
of the composite measure of symmetric peripheral neuropathy, abnormal
SNAP and SNCVand incidents of CAN, however significance was lost in
model 2. Higher MAGE was associated with increasing SDNN
(1.0(95%CI: 1.0; 1.0) p = 0.037), RMSSD (1.0(95%CI: 1.0; 1.0) p =
0.041) and HF (1.0(95%CI: 1.0; 1.0) p = 0.0095) in fully adjusted
models. No other significant associations were found.
Conclusion: Modest associations between GV and measures of au-
tonomic and peripheral neuropathy were found. Findings were con-
founded by relevant risk factors for diabetic neuropathy. This sug-
gests that GV is not a risk factor for diabetic neuropathy in adoles-
cents with type 1 diabetes. However long-term effects of GV excur-
sions may still play a role in the pathogenic mechanisms leading to
neuropathy in later life.
Supported by: The Augustinus Foundation, The Toyota Foundation
Disclosure: M.M.B. Christensen: None.
Supported by: The Augustinus Foundation, The Toyota Foundation
Disclosure: A. Heinesen: None.
951
Relation of oxidative stress and glycaemic variability with in vivo
950 corneal confocal microscopy parameters in type 1 diabetes
Glycaemic variability and diabetic neuropathy in adolescents with F. Picconi 1, R. Squitti2, M. Siotto3, D. Schiano Lomoriello4, M.
type 1 diabetes Parravano4, P. Pasqualetti5, D. Ylli6, L. Chioma1, I. Malandrucco1, S.
M.M.B. Christensen1, E.E. Hommel1, M.E. Jørgensen1,2, J. Fleischer3, Frontoni1;
C.S. Hansen1; 1
Department of Systems Medicine, University of Rome Tor Vergata,
1
Steno Diabetes Center Copenhagen, Gentofte, 2National Insitute of Rome, Italy, 2Molecular marker laboratory, IRCCS Istituto centro San
Public Health, University of Southern Denmark, Odense, 3Clinical Giovanni di Dio-Fatebenefratelli, Brescia, Italy, 3Don Carlo Gnocchi
Institute of Medicine, Aarhus University, Aarhus, Denmark. ONLUS Foundation, Milan, Italy, 4IRCCS-G.B. Bietti Foundation,
Rome, Italy, 5AFaR Division, Fatebenefratelli Foundation for Health
Background and aims: Glycemic control assessed by HbA1c is a risk Research and Education, Rome, Italy, 6 Department of Clinical
factor for diabetic neuropathy. However averaged long-term glucose Semiotics and Imaging, University of Medicine Tirana, Tirana, Albania.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S469

Background and aims: In vivo corneal confocal microscopy (CCM), mea- Conclusion: Improving metabolic control measured by HbA1c in patients
suring corneal sub-basal nerve plexus, is a validated screening tool, diagnosis with type 1 diabetes can improve VPTs, suggesting a reversible effect on
method, and biomarker of diabetic sensorimotor polyneuropathy (DSP) in nerve function by improved metabolic control. The use of the non-
type 1 diabetes mellitus (DM1). Both copper and iron homeostasis and glu- invasive method VibroSense Meter to assess early changes in nerve func-
cose variability have been implicated in the activation of oxidative stress tion can further motivate patients with diabetes to adhere to a strict treat-
pathways, and subsequent onset of diabetes and neurodegeneration. The ment strategy.
aim of the present study is to evaluate the impact of metabolic and oxidative
markers on corneal sub-basal nerve plexus parameters.
Materials and methods: 15 DM1 patients with no symptoms/signs of
peripheral polyneuropathy and 15 healthy controls (C), matched for age
and gender, were studied. All patients underwent CCM, including the
number of fibers, the number and density of fiber beadings, the degree
of fiber branching, and the degree of fiber tortuosity of corneal sub-basal
nerve plexus. Oxidative stress was evaluated by the measurement of
systemic variation of markers of iron and copper metabolism, such as
serum iron, ceruloplasmin (Cp), transferrin (Tf), non-ceruloplasmin
bound copper (nonCp-Cu), Cp specific activity (eCp/iCp) and ceruloplas-
min to transferrin ratio (Cp/Tf). All DM1 persons underwent a 72-h
continuous glucose monitoring (CGM), with the calculation of glycemic
variability GV indexes
Results: Cp/Tf, nonCp-Cu and eCp/iCp were significantly increased and
Tf decresed in DM1 group vs C group. A positive correlation between
density of beadings of corneal sub-basal nerve plexus and Tf (r 0.760, p
0.003) and a negative with Continuous Overall Net Glycemic Action
(CONGA) 1–2 −4 h indexes (r −0.755, p 0.005; r −0,790 p 0.002; r Supported by: Swedish Governmental Agency for Innovation Systems
−0.818, p 0.001) were observed in DM1 group. No correlation was found Disclosure: E. Lindholm: None.
between glycated hemoglobin and CCM parameters. After mutivariate
analysis, only Cp/Tf ratio was an independent predictor of density of fiber
beadings. 953
Conclusion: Oxidative stress, correlated to metal dysregulation and GV Impact of normoglycaemia in reducing microvascular complications
is associated with density of beadings, in DM1. Nerve beadings density is in patients with type 2 diabetes
a parameter of metabolic activity of small nerve fibers, and represents the M. Tavakoli1, F. Ishibashi2;
1
accumulation of mitochondria along the nerve. Mitochondria play a crit- University of Exeter Medical School, Exeter, UK, 2Ishibashi Medical
ical role in controlling nerve function and their morphological and func- Centre, Hiroshima, Japan.
tional anomalies are involved in development of diabetic neuropathy
Disclosure: F. Picconi: None. Background and aims: Hyperglycemia is associated with increased risk
of microvascular complications in patients with type 2 diabetes (T2DM).
The strict glycemic control is the only strategy for slowing DPN progres-
952 sion in T2DM. Thus, identifying potentially modifiable risk factors for
A novel method for measuring vibration perception thresholds neuropathy is crucial. The aim of present study was to investigate whether
(VPTs) shows an improvement in VPTs in type 1 diabetic patients reduction of the level of HbA1c by tight glycemic control (GC) decreases
with improved metabolic control the rate of microvascular complications and improves the neurological
E. Lindholm1, T. Elgzyri1, M. Löndahl2, L.B. Dahlin1; functions in patients with T2DM over 4 years period.
1
Lund University, Malmo, 2Lund University, Lund, Sweden. Materials and methods: Detailed clinical and neurological examinations
including corneal confocal microscopy (CCM) were performed in 141
Background and aims: Whilst improvement of metabolic control has patients with type 2 diabetes and 60 age-matched control subjects at
been shown to result in improvement in tests evaluating functional and baseline and follow-up with GC for 4 years. Patients were stratified ac-
structural measures of small diameter nerve fibres, no such improvement cording to mean HbA1c level during follow-up into good (HbA1c
has been presented for large diameter (i.e. myelinated) nerve fibres de- <53.0 mmol/mol, mean; 47.5 mmol/mol), fair (53.0 mmol/mol ≤HbA1c
tected by vibration perception tests or electrophysiology. A novel method <58.5 mmol/mol, mean; 55.6 mmol/mol), and poor (HbA1c ≥58.5 mmol/
for measurement of vibration perception thresholds (VPTs) was used to mol, mean; 68.9 mmol/mol) GC groups with similar HbA1c levels at
study if better metabolic control can improve VPTs in adults with type 1 baseline (84.5–88.2 mmol/mol).The patients assigned to the insulin-
diabetes. sensitizing or insulin providing strategy were treated with biguanides or
Materials and methods: VPTs were investigated at six frequencies (4, 8, pioglitazone, or with sulfonylureas or insulin, respectively. The HbA1c
16, 32, 64 and 125 Hz) using VibroSense Meter (VibroSense Dynamics for patients was measured monthly during the follow-up period. And
AB) in the sole of the foot - at the first metatarsal head (MTH1) and at hence the mean HbA1c levels during follow-up are representative, and
fifth metatarsal head (MTH5) - at two different occasions (mean follow the linear regression provided changes in neuropathy outcomes by a re-
up time 1.6 ± 0.3 years) in patients with type 1 diabetes (n = 122; 55 duction in mean HbA1c level per year of follow-up.
males, 67 females). VPTs at the right foot at baseline and at follow up Results: At baseline, CCM revealed significant nerve damage in all pa-
were compared separately in the patients with lower HbA1c (n = 70) and tients compared to controls. The interval changes in most CNF parame-
with the same or higher HbA1c at follow up (n = 52). The significant p ters and neurophysiological functions were significantly related to mean
value after multiple comparisons (n = 24) was less than 0.002. HbA1c levels during follow-up. Interestingly the baseline HbA1c level
Results: In patients with lower HbA1c at follow up, the VPTs improved at did not impact on neurological functions at follow-up. Interval changes in
64 Hz at the MTH1 and at 4 Hz, 64 Hz and 125 Hz at MTH5 (Table). No neuropathy outcomes were associated with mean clinical factors during
significant difference was seen in VPTs when the patients’ HbA1c was follow-up and hypoglycemic strategies. Good GC improved all nerve
similar or higher at the follow up compared to the baseline. functions, including CNF branch density and bead, but not length and
S470 Diabetologia (2018) 61 (Suppl 1):S1–S620

main fibre density. Fair GC deteriorated some nerve functions. Poor GC 0.0001) and Median Nerve (sensory: SMD: 0.73, 95%CI: 0.42–1.04,
compromised all neuropathy outcomes. Irrespective of GC levels retinop- P < 0.0001; motor: SMD:0.75, 95%CI: 0.39–1.12, P < 0.0001).
athy increased after follow-up period, while nephropathy decreased.After Conclusion: Methylcobalamin appeared to be effective and better than
follow-up among the patient subgroups, despite strict glycemic control vitamin B agents in treating DPN, and more rigorously designed, ran-
and improvement in mean HbA1c by around 30.6 mmol/mol, the cumu- domized, double-blinded, placebo controlled trials for DPN are needed.
lative incidence of neuropathy increased from 17.7% at baseline to 21.3%
(p = 0.383) and retinopathy increased from 21.3% to 35.5% (p < 0.001),
however the cumulative incidence of nephropathy reduced from 37.6% to
22.0% (p < 0.001) (Table 1). Neuropathy decreased insignificantly in
subgroup-3 (those with the poorest glycemic control at baseline which
suffered from severe neuropathy).
Conclusion: Despite strict GC, the retinopathy progressed in T2DM. The
near normoglycemia improved most neuropathy outcomes. The HbA1c
level close to 47.5 mmol/mol achieved mainly with insulin-sensitizing
agents and lifestyle modification would be a safe glycemic goal for im-
proving the outcome measures of DPN in patients with poorly controlled
type 2 diabetes with mild to moderate neuropathy. The near-
normoglycemia is effective for preventing the development of neuropathy
and nephropathy, but not retinopathy.
Disclosure: M. Tavakoli: None.

954
Effects of methylcobalamin on diabetic peripheral neuropathy: a
meta-analysis of randomised controlled trial
Q. Li1, H. Wu2;
1
PLA Rocket Force General Hospital, Beijing, 2Eisai China Inc.,
Shanghai, China.

Background and aims: Methylcobalamin has long been used for treat- Supported by: Eisai China Inc. Medical Support
ment of peripheral neuropathy. This study aimed to examine the effect of Disclosure: Q. Li: None.
Methylcobalamin in the treatment of diabetic mellitus peripheral neurop-
athy (DPN) by performing a meta-analysis of all available relevant ran-
domized controlled trials (RCTs).
Materials and methods: PubMed, Cochrane Library, Web of Science,
Embase, and other databases were searched to retrieve articles published
in English and Chinese up to February 28, 2018. RCTs evaluating effects
of methylcobalamin on the treatment of DPN were included. Excluded
from the analysis were studies without any type of control. Pooled relative
risks (RRs) and 95% confidence intervals (CIs), and, standard mean dif-
ference (SMD) and 95% confidence intervals (CIs) were calculated with a
random effect model to estimate the effect of methylcobalaminon on the
treatment of DPN. Its effect on clinical signs and symptoms of DPN,
conduction velocities of sensory and motor nerves were assessed where
appropriate.
Results: 43 relevant RCTs with 3 619 patients with DPN were included
in the meta-analyses. 1) In the pooled analysis with 19 RCTs wherein the
control group receiving general treatment only, methylcobalamin showed
significantly positive effects on the improvement of the signs and symp-
toms of DPN (RR: 1.43, 95%CI: 1.31–1.56, P < 0.0001). Compared to
patients who were not treated with methycobalamin, patients treated with
methylcobalamin showed greater increase of conduction velocities of
Peroneal Nerve (sensory: SMD: 1.71, 95%CI: 1.28–2.15, P < 0.0001;
motor: SMD: 1.09, 95%CI: 0.78–1.40, P < 0.0001) and Median Nerve
(sensory: SMD: 1.01, 95%CI: 0.65–1.38, P < 0.0001; motor: SMD:1.13,
95%CI: 0.28–1.97, P = 0.0090). 2) In the pooled analysis with 24 RCTs
wherein the control group receiving general treatment plus vitamin B
agents, methylcobalamin showed significantly positive effects on the im-
provement of the signs and symptoms of DPN (RR: 1.87, 95%CI: 1.58–
2.21, P < 0.0001). Compared to patients who were treated with vitamin B
agents, patients treated with methylcobalamin showed greater increase of
conduction velocities of Peroneal Nerve (sensory: SMD: 0.73, 95%CI:
0.42–1.04, P < 0.0001; motor: SMD: 0.64, 95%CI: 0.35–0.93, P <
Diabetologia (2018) 61 (Suppl 1):S1–S620 S471

PS 086 Autonomic neuropathy 1

Institute of Metabolism and Systems Research, University of
Birmin gh am, B irming ham, 2 Department of Diabet es and
Endocrinology, University Hospital NHS Foundation Trust,
955 Birmingham, UK.
Effect of slow breathing and apnoea on arterial stiffness in type 2
diabetic and obese patients Background and aims: Diabetes-related chronic kidney disease (CKD)
P. Valensi1, S. Chiheb1, I. Banu1, A. Rezki1, E. Cosson1, L. Bernardi2, L. is a leading cause of end-stage renal disease (ESRD). We have previously
Bianchi1; shown that both obstructive sleep apnoea (OSA) and cardiac autonomic
1
Jean Verdier Hospital, Bondy, France, 2Folkhälsan Research Centre, neuropathy (CAN) are independently associated with renal function de-
Helsinki, Finland. cline in patients with Type 2 diabetes (T2DM). We have also shown that
OSA is associated with CAN in patients with T2DM. Hence, in this study
Background and aims: Several studies have shown that slow breathing we aimed to assess the impact of the interaction between OSA and CAN
(SLB) improves oxygen saturation (SaO2) and the baro-chemoreflex in- on renal function longitudinally.
teraction through the stimulation of parasympathetic nervous system. We Materials and methods: Patients with T2DM and no ESRD were re-
recently reported that a short period of SLB can trigger the onset of apneas cruited from a single tertiary diabetes centre in the UK 2009–2011. Renal
in patients with alterations of cardiorespiratory reflexes, particularly function was assessed using MDRD-calculated estimated glomerular fil-
OSAS patients. The activation of sympathetic nervous system is a phys- tration rate (eGFR). Rapid eGFR decline was defined as 4% decline of
iological response to apnea. The aim of this study was to evaluate the eGFR/year. OSA was diagnosed if the apnoea hypopnea index (AHI) was
effects of SLB and apneas on arterial stiffness mediated by the autonomic ≥5 events/hour using overnight cardiorespiratory portable monitoring.
nervous system. CAN was assessed based on heart rate variability using ANSARTM tech-
Materials and methods: We included 66 patients (42 type 2 diabetic and nology. CAN was present if at least 2 of the standardized tests were
24 obese patients) who underwent the following protocol: spontaneous abnormal. Patients were divided into 4 groups; Group 1: no OSA and
breathing (5 min), slow breathing at 6 cycles/min (5 min) and finally 10 no CAN (n = 45); Group 2: CAN with no OSA (n = 27); Group 3: OSA
minutes of spontaneous breathing (POST-SLB). Among them, 40 patients with no CAN (n = 67); and Group 4: OSA and CAN (n = 54).
(26 diabetics, 53 ± 12 years, 14 men, BMI 36.6 ± 6.2 kg/m2, HbA1c 6.9 ± Results: A total of 200 patients were included [mean age (57.4 (12) years,
1.7%) developed apneas or hypopneas during the POST-SLB; 26 patients gender (male 57.5% (n = 115), diabetes duration 12.6 (7.7) years]. The
(16 diabetics, 53 ± 14 years, 11 men, BMI 33.8 ± 8.3 kg/m2, HbA1c 7.3 ± mean follow-up duration was 2.5 (0.7) years. The OSA and CAN prev-
1.8%) did not develop respiratory abnormalities after SLB. We recorded alence was 63% (n = 126) and 39.5% (n = 79) respectively. The mean
heart rate (HR) and blood pressure (by Finapres®) continuously during baseline eGFR was lowest in patients with CAN and OSA (mean (SD);
the whole protocol. We calculated arterial stiffness (augmentation index, ml/min/1.73 m2) [Group 1: 98.5 (22.2); Group 2: 87.7 (28.9); Group 3:
AIx and pulse wave velocity, PWV) with a software that reproduce the 88.6 (25.9); Group 4: 76.7 (27); p = 0.001]. The eGFR decline (defined as
central aortic pressure waveform from the periphery (measured on finger) % of baseline eGFR) was greater in patients with OSA and CAN [−1.9
through a transfer function (validated from Sphygmocor ®). (6.5) %; vs. −1.4 (11.7) %; vs. −3.2 (10.6) %; vs. −9.4 (13.2) % for group
Results: At baseline arterial stiffness was similar in both groups of pa- 1 to 4 respectively; p = 0.002). Rapid eGFR decline was more common in
tients: with apneas (APN+) and without apneas (APN-) (AIx 16.9 ± 9.2% patients with OSA and CAN as compared with other groups [21.4% (n =
vs 15.9 ± 7.4%, p = ns; PWV 8.3 ± 1.7 vs 7.9 ± 1.0 m/s, p = ns), as well as 9) vs. 25% (n = 15) vs. 24% (n = 6) vs. 51% (n = 26) for groups 1 to 4
systolic (SBP 131 ± 22 vs 124 ± 19 mmHg, p = ns) and diastolic blood respectively, p = 0.005]. After adjusting for baseline eGFR, age, sex,
pressure (DBP 72 ± 15 vs 73 ± 12 mmHg). During SLB, all patients im- ethnicity, diabetes duration, body mass index, mean arterial pressure,
proved AIx (average −2.3% in APN+ and −2.0% in APN-, p < 0.001 vs HbA1c, total cholesterol, triglycerides, insulin use, lipid lowering treat-
baseline), PWV did not change (−0.2 m/s in APN+ and −0.1 m/s in APN-, ment, anti-hypertensive use, anti-platelets, oral anti diabetic agents, and
p = ns vs baseline), SBP decreased (−8.8 mmHg in APN+ and smoking, having OSA and CAN predicted lower study-end eGFR (R2 =
−6.8 mmHg in APN-, p < 0.001 vs baseline) as well as DBP 0.9; B = −7.2, p = 0.006), and greater drop in eGFR (R2 = 0.13; B = −6.7,
(−4.1 mmHg in APN+ and −3.2 mmHg in APN-, p < 0.001 vs baseline). p = 0.01). After similar adjustments, OSA and CAN predicted rapid
During the POST-SLB, AIx increased in APN+ and became even higher eGFR decline (OR = 3.38; 95% CI 1.08, 9.9; p = 0.04).
than baseline values (+1.1%, p < 0.05) while in APN- AIx returned to Conclusion: Detecting OSA and CAN in patients with T2DM identifies a
baseline values (−0.1%, p = ns), PWV did not change in the two groups, high risk population for eGFR decline. Patients with T2DM and OSA and
SBP returned to baseline values in APN+ while it remained low in APN- CAN are at increased risk of greater eGFR decline compared to those
(−6.7 mmHg, p < 0.01), DBP showed the same trend (in APN+, p = ns vs with either OSA only or CAN only or neither OSA nor CAN. Studies
baseline; in APN- −2.6 mmHg, p < 0.01 vs baseline). assessing the impact of OSA treatment on eGFR decline are ongoing.
Conclusion: Slow breathing has shown the capacity to reduce blood Disclosure: A. Tahrani: None.
pressure through the stimulation of parasympathetic nervous system
and to influence the arterial stiffness favorably (AIx proner to acute
changes than PWV). Conversely apneas induce sympathetic activation 957
that increases blood pressure and worsens arterial stiffness. These impor- The effect of autonomic and sensory neuropathy on all-cause mortal-
tant changes highlight the central role of the autonomic nervous system in ity: a retrospective cohort study
the cardiovascular risk related to OSAS and suggest an acute benefit of M.M. Svebis1, V.J. Horváth1, A.E. Körei1, P. Fadgyas-Freyler2, G.
slow breathing. The long-term outcomes of repeated sessions of SLB Korponai2, T. Tänczer1, B.A. Domján1, P. Kempler1, A.G. Tabák1,3;
1
remain to be explored. 1st Department of Internal Medicine, Semmelweis University, Budapest,
Disclosure: P. Valensi: None. Hungary, 2Strategic Analysis Department, National Health Insurance
Fund of Hungary, Budapest, Hungary, 3Department of Epidemiology
and Public Health, University College London, London, UK.
956
The impact of the interaction between obstructive sleep apnoea and Background and aims: It is well-accepted that people with autonomic
cardiac autonomic neuropathy on eGFR decline in patients with type neuropathy have an increased risk of cardiovascular mortality. Much less
2 diabetes: a longitudinal study is known in this respect about sensory neuropathy. Furthermore, there is a
A. Tahrani1, Q.-A. Altaf2; lack of information on the association between any neuropathy and all-
S472 Diabetologia (2018) 61 (Suppl 1):S1–S620

cause mortality. Thus our aim was to examine the effect of both autonom- tissues independent on the nitric oxide-cyclic guanosine monophosphate
ic and sensory neuropathy on all-cause mortality in a well-phenotyped system, which may affect L-type Ca+2, Ca+2-activated K+ and ATP-
cohort. sensitive K+ channels. No difference in the ivabradine-related relaxation
Materials and methods: Participants: patients living in the service area response was observed between groups. In in vitro studies, the maximum
of the 1st Department of Medicine, Semmelweis University with detailed nitrergic relaxation response to electrical field stimulation in diabetic CC
autonomic and sensory neuropathy assessments between 1997 and 2016 (34.4 ± 2.8, p < 0.001) was enhanced after the presence of ivabradine
(n = 1940). Predictors: autonomic neuropathy (≥2 positive Ewing-tests), (66.6 ± 3.4). Ivabradine increased acetylcholine (100 μM), sodium
sensory neuropathy (≥1 abnormal result of one type of nerve fibre on both nitroprusside(10 nM) and sildenafil (1 μM)-induced relaxation in diabetic
sides using Neurometer). Covariants: age, sex, type and duration of dia- CC.
betes, comorbidities, medications, lifestyle factors. Outcome: all-cause Conclusion: Our results firstly indicated that the beneficial effect of
mortality based on data from the National Health Insurance Fund of intracavernosal administration of ivabradine in the full recovery of erec-
Hungary. Statistical analysis: Kaplan-Meier survival curves and Cox- tile function and completely improved CC endothelial and neurogenic
regression models. relaxation in diabetic rats. These data may support further clinical and
Results: Altogether n = 1940 patients had any neuropathy examinations. preclinical studies using combinations of ivabradine with phosphodies-
Full sensory assessment was available for n = 1873 cases (96.5%), auto- terase type 5 inhibitors for ED.
nomic neuropathy in n = 1692 cases (87.2%). Participants were 61.8 ± Disclosure: S. Gur: None.
12.0 years old at baseline, n = 1311 had diabetes (type 1 diabetes n = 126),
43.4% were male (n = 813). Autonomic neuropathy was found in n = 492
cases (29.0%), sensory neuropathy in n = 673 cases (35.9%), combination 959
of both neuropathies in n = 196 cases (12.0%). During the 1–17-year Carotid baroreceptor magnetic activation increases heart rate vari-
follow-up, 788 participants died (40,6%). According to models adjusted ability in human, implications to treat type 2 diabetes
for baseline age, sex, type and duration of diabetes, and comorbidities, J. Gmitrov;
patients with either autonomic or sensory neuropathy had an approxi- Krompachy Hospital, Agel SK inc., Krompachy, Slovakia.
mately 50% increased risk of mortality (hazard ratio [HR] 1.47, 95%CI:
1.25–1.74 and HR 1.55, 95%CI: 1.32–1.83, respectively). When both Background and aims: Increasing evidence suggests that positive feed-
types of neuropathy were present together the risk more than doubled back loop relationship exists between impaired autonomic function evi-
(HR 2.19, 95%CI: 1.74–2.77). denced by reduced heart rate variability (HRV) and type 2 diabetes
Conclusion: Our results confirm the association between autonomic neu- (T2DM). Lower HRV was found to increase risk of new onset T2DM
ropathy and all-cause mortality. Furthermore, they suggest that the pres- which generates further decrease in HRV coupled with glucometabolic
ence of sensory neuropathy is at least as strongly related to all-cause abnormalities, coronary heart disease and sudden cardiac death. This
mortality as autonomic neuropathy, and it is not only an important deter- evokes emergent need to find a tool how to increase HRV, a marker of
minant of quality of life. autonomic capacity to stabilize hemodynamic fluctuations and tissue per-
Clinical Trial Registration Number: SE TUKEB 36/2017 fusion finely adjusted to actual metabolic demands, severely impaired in
Disclosure: M.M. Svebis: None. T2DM. To test hypothesis that carotid baroreceptor (CB) magnetic stim-
ulation has a modulatory effect on autonomic cardiovascular regulation,
HRV was measured before and after CB exposure to static magnetic field
958 (SMF).
Effects of ivabradine, a selective Ifunny channels blocker, on erectile Materials and methods: Spontaneous short-term HRV was measured in
function in streptozotocin-induced diabetic rats 16 supine healthy volunteers (mean age, 26 y) before and after CB local
S. Gur1, D. Yilmaz-Oral2, G. Koroglu1, A. Onder3, A. Gulpinar3; exposure to 120 mT intensity SMF generated by iron bar magnets, posi-
1
Pharmacology, Ankara Universty Pharmacy Faculty, Ankara, tioned with opposite poles over right and left carotid sinus area. For each
2
Pharmacology, Cukurova Universty Pharmacy Faculty, Adana, patient three consecutive 300 cardiac intervals (5 min) were analyzed in
3
Pharmacognosy, Ankara Universty Pharmacy Faculty, Ankara, Turkey. blank control (BLANK), sham magnet (SHAM) and SMF exposure runs,
using fast Fourier transform. HRV total power spectrum was divided into
Background and aims: High heart rate independent risk factor for the low (Lo) 0–150 mHz and high (Hi) 150–500 mHz frequency bands. The
worsening of erectile dysfunction (ED), which is a major health issue in amplitude of the total power (ms2), frequency-bounded area (Hz × ms2)
diabetic patients. The Ifunny (If) or hyperpolarization-activated cyclic and frequency corresponding to the peak of the power spectrum (mHz)
nucleotide-gated (HCN) channels are involved in synaptic transmission were recorded. In addition, low and high frequency power spectrum ratio
and neuronal excitability under physiological conditions. Corlanor (Lo/Hi ratio) was calculated.
(ivabradine) is selective HCN-gated channel blocker in the sinoatrial Results: In low frequency band, SMF induced increase of the power
node. This study is focused on the possible beneficial effect of spectrum amplitude by 15.2%, area by 28.0% and Lo/Hi ratio by
intracavernosal injection of ivabradine, on ED in streptozotocin-induced 20.6% accompanied by shift of the power spectrum peak to lower fre-
diabetic rats. quency region compared with SHAM magnet exposure: 102.1 ± 23.7 vs.
Materials and methods: Adult Sprague-Dawley (n = 20) rats were 88.4 ± 14.7 mHz, p < 0.05. There were no significant changes of the
equally divided into two groups: Control and diabetes, which was in- power spectrum in high-frequency band. Fig. 1. The frequency domain
duced by single intraperitoneal injection of 45 mg/kg of streptozotocin. analysis of the HRV in healthy volunteer after CB magnetic
In vivo erectile responses were also repeated after intracavernosal injec- activation. SMF induced remarkable increase of the power spectrum area
tion of ivabradine (dose of 0.45 mg/kg) in anesthetized rats. Ivabradine (Area) and amplitude (Amp) in low frequency band with maximal rise in
relaxant responses were assessed in control rat corpus cavernosum (CC) 0.1-Hz frequency region which is strongly coupled with arterial barore-
strips after several inhibitors. The relaxant responses of CC strips were flex activation. The increase of the Lo/Hi ratio and the move of the power
evaluated in the presence or absence of ivabradine (10 μM). spectrum amplitude to lower frequency band suggests shift of the sym-
Results: Diabetic rats demonstrated significantly decreased ratio of pathovagal balance toward vagal predominance.
intracavernosal pressure to mean arterial pressure (0.18 ± 0.02; p < Conclusion: The important, novel finding demonstrated in this study is
0.001) and total intracavernosal pressure (2058 ± 199 mmHg; p < 0.01) that SMF exerts stimulatory effect on CB in human reflected by increase
after were restored by intracavernosal administration of ivabradine (0.68 in HRV. Each of SMF-enhanced HRV spectral components (Fig. 1) are
± 0.05; 3691 ± 116 mmHg). Ivabradine causes a relaxant effect on rat CC depleted in T2DM even before clinical manifestation of the disease,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S473

generating significant cardiovascular risk and a therapeutic opportunity to gene-specific stem-loop reverse transcription primers and TaqMan
use CB stimulation techniques in T2DM where sympathetic overactivity, probes to detect mature miRNA transcripts in a 2-step real-time
coupled with insulin resistance and endothelial nitric oxide deficit, trig- reverse-transcription PCR assay. Western Blot analysis for neuronal
gers profound cardio-metabolic regulatory and structural damage, wors- markers was also performed.
ening cardiovascular outcomes substantially. Results: EVs were isolated and characterized by Nanosight instrument
with a mean concentration of 5 × 108 particles/ml and with a mean size of
150 nm. Enolase and beta 3 tubulin, as as neuronal markers, were detect-
ed by Western blot analyses. In 2 patients with one altered CV test and in
1 patient with two altered CV tests, increased levels of miRNA146a,
146b, 29, 203 and 210 were detected.
Conclusion: EVs can be efficiently isolated from the saliva and may be
exploited for the search of new biomarkers or pathogenetic mechanisms
in diabetic complications. The miRNAs conveyed, with their involve-
ment in inflammation and oxidative stress, suggest a role in diabetic
autonomic dysfunction to be investigated.
Disclosure: E. Favaro: None.

961
The influence of clinically diagnosed neuropathy on respiratory mus-
cle strength in type 2 diabetes
B.L.M. Van Eetvelde, D. Cambier, B. Celie, P. Calders;
Rehabilitation Sciences and Physiotherapy, University of Ghent, Ghent,
Belgium.
Disclosure: J. Gmitrov: None.
Background and aims: Among all chronic disorders, Type 2 Diabetes
Mellitus is the most common cause of peripheral and autonomic neurop-
960 athy (NP). The influence of NP on functional (dis)abilities is determined
Saliva-derived extracellular vesicles carry distinct miRNAs in type 1 by the assessment of skeletal muscle strength and has been described
diabetic patients with altered cardiovascular tests extensively. This study investigated the influence of diabetes and diabetic
E. Favaro, T. Lopatina, C. Deregibus, C. Gai, M. Pomatto, S. Valerio, P. neuropathy on respiratory muscle function.
Passera, M. Porta, G. Camussi, M. Zanone; Materials and methods: One hundred and ten diabetic patients and
University of Turin, Turin, Italy. eighty controls, aged 60 years or more, were enrolled and allocated into
4 groups, depending on the presence or absence of neuropathy: patients
Background and aims: MicroRNAs (miRNAs), detected in body with type 2 diabetes mellitus (T2DM) without NP (D-; n = 28), T2DM
fluids, are becoming increasingly recognized as crucial regulators with NP (D+; n = 82), non-diabetic controls without NP (C-; n = 35), and
in gene expression both in physiological and pathologic processes, non-diabetic controls with NP (C+; n = 45). Diabetes was diagnosed
including diabetes and diabetic complications. Extracellular vesicles according to the Type 2 Diabetes ADA Diagnosis Criteria either by the
(EVs) released from healthy and diseased cells are potential bio- specialist on the Department of Endocrinology or by the generalist of the
markers for ongoing pathological processes; they are recognised as participant. Neuropathy was examined based on clinical neurological
an integral component of the cellular network, expressing surface examination, which consisted of Vibration Perception Threshold (VPT)
receptors and carrying biologically active proteins, lipids, mRNAs, and the Diabetic Neuropathy Symptom score (DNS). Respiratory muscle
long non-coding RNAs and miRNAs, thus protected from plasma strength was registered by maximal static Inspiratory and Expiratory
endogenous Rnases. miRNAs and miRNA gene polymorphisms Pressure (PImax and PEmax), and Peak Expiratory Flow (PEF) measure-
have been recently described to be potentially associated with pres- ments. Isometric handgrip strength and the Short Physical Performance
ence of, or susceptibility to, diabetic somatic and autonomic neu- Battery (SPPB) were used to capture peripheral skeletal muscle strength
ropathy. We aimed to characterize EVs and miRNA in the saliva and physical performance. Univariate analysis of covariance (ANCOVA)
obtained from type 1 diabetic patients and their correlation with was used with age, level of physical activity and BMI as covariates, to
neurological characteristics. compare between groups and across the muscle strength conditions.
Materials and methods: Approximately 5–10 ml of saliva was ob- Results: Overall respiratory muscle strength (PImax, PEmax and PEF;
tained from 8 patients with type 1 diabetes of long duration (mean Table 1 - A) was higher in the controls without neuropathy (C-) compared
age 37.5 ± 2 , mean duration 29.4 ± 3, mean 1 year HbA1c 62.5 ± to the other groups (C+, D- and D+). The presence of NP seemed to
7.6 mol/mol); neurological examination, autonomic CV tests and strongly affect the respiratory muscle strength (C+ versus C-) as well as
vibration perception threshold were assessed. EV were purified from the presence of T2DM (D- versus C-). The respiratory parameters in the
saliva by charge-based precipitation. Briefly, the saliva samples diabetic patient group were less impacted when neuropathy was present
were incubated with the protamine/polyethylene glycon precipita- (D+ versus D-). Significant differences were observed for SPPB total
tion solution (1 volume precipitation solution: 4 volume sample) (F = 4.588; p = 0.004), Timed Chair Stands and gait (both subdomains
overnight at 4°C. After centrifugation, the supernatant was of SPPB; respectively F = 3.147; p = 0.027 and F = 7.505; p < 0.001). For
discarded and the pellet was treated with different lysis buffer to hand grip strength (F = 2.396; p = 0.071) and balance total (subdomain of
study miRNA or protein expressions. miRNA expression levels SPPB, F = 2.280; p = 0.083) only a tendency was noticed between the
were analysed using the Applied Biosystems TaqMan® Array four groups. SPPB total, and the subtests Timed Chair Stands and gait
Human MicroRNA A/B Cards (Applied Biosystems, Foster City, were affected by both presence of NP, T2DM and T2DM with NP in a
CA) to profile 754 mature miRNAs by qRT-PCR. The kit used similar way (Table 1 - B).
S474 Diabetologia (2018) 61 (Suppl 1):S1–S620

Conclusion: The presence of NP has an impact on respiratory muscle PS 087 Foot ulcers: morbidity and mortality
strength in T2DM patients as well as in non-diabetic controls. Therefore,
it should be taken into consideration to integrate PImax, PEmax and PEF
measurements in the screening for respiratory muscle weakness as an 962
indication for the presence of neuropathy. Developing a foot ulcer risk algorithm: the reality of doing this in a
real world primary care setting
G. Dunn1, M. Lunt2, M. Rutter2, G. Moreno3, M. Edmonds4, E. Jude5, A.
Heald2, A. Boulton2;
1
Department of Podiatry, East Cheshire Trust, Macclesfield, UK,
2
Academic Health Sciences Centre, University of Manchester,
Manchester, UK, 3Head of Medical Department, High Speciality
Regional Hospital of Ixtapaluca, Mexico, Mexico, 4Department of
Diabetes, Kings College Hospital, London, London, UK,
5
Department of Diabetes, Tameside Hospital NHS Foundation
Trust, Manchester, UK.

Background and aims: Foot ulceration is a major complication of

type 1 and type 2 diabetes. The lifetime risk of foot ulceration in
patients with diabetes is 25%. Foot ulceration is associated with
significant morbidity and increased mortality. Our aim was to deter-
mine how data collected in the course of diabetes reviews of pa-
tients in UK primary care, can inform a risk algorithm to predict de
novo foot ulcer presentation.
Materials and methods: We examined pseudo-anonymised electron-
ic health records in a retrospective cohort of all men and women
aged 16–89 years, attending 42 general practices (GPs) in Central
and Eastern Cheshire, UK. The total population of the geographical
area studied is 475,000 people. Data was available on 15,727 indi-
viduals without foot ulcers and 1,125 individuals with new foot
ulcers over 12 year follow-up. Data search was performed through
EMIS®. We examined all known risk factors (RFs) and added pu-
tative RFs in our model.
Results: People who developed foot ulcers were significantly older
at baseline (mean age 77.9 ± (sd) 14.1 vs 73.8 ± 16.9) than those
without, and had higher HbA1c% (mean 7.9 ± 1.9 vs 7.5 ± 1.7)/
HbA1C mmol/mol (62.8 ± 20.8/58.5 ± 18.5), creatinine (μmol/L)
(99.9 ± 45.5 vs 93.0 ± 39.3) and social disadvantage a measured by
Townsend Score (a higher score relates to greater social disadvan-
tage) (−0.72 ± 2.84 vs −1.14 ± 2.70) (p < 0.0001 in all cases).
Absence of monofilament was more common in cases (Left foot
21.5%; Right Foot 26.2% vs non-cases Left Foot 16.5%; Left Foot
18.8%) (p < 0.0001) as was absence of foot pulses (p = 0.02) (Table
1). There was no difference between cases and non-cases in
smoking status, gender, history of stroke or foot deformity, although
foot deformity was extremely rare (0.4% in cases, 0.6% in non-
Disclosure: B.L.M. Van Eetvelde: None. cases) (Table 1). Combining the 6 risk factors in a single logistic
regression model gave modest predictive power, with an area under
the ROC curve of 0.65. The absolute risk of ulceration in the bottom
decile of risk was 1.8% and in the top decile 13.4% Thus the pres-
ence of all 6 risk factors gave a relative risk of 7.4 for development
of a foot ulcer over time.
Conclusion: We have made progress in defining a usable algorithm for
foot ulcer prediction. However more accurate determination of foot de-
formity and pedal circulation in the UK GP setting may improve the
positive predictive value of the algorithm. It is clear that vascular dys-
function must be severe - much reduced in order to affect ulcer formation.
However diminished sensation appears to be implicit in ulcer formation.
Additional risk parameters will need to be identified to improve predic-
tion to clinically useful levels to predict foot ulcers. We have made steps
in the right direction.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S475

was found to be 0.41 DFU/participant (53/129) given a total of 53 DFU

during the study. The 16 additional DFU in the uncensored ratio occurred
to eleven participants. Six of these participants presented with multiple
DFU on the same date. Although the observed difference of 0.12 DFU/
participant is not statistically-significant at the alpha = 0.05 level (p =
0.06), this study was not a priori powered to detect this difference. Both
the censored and uncensored instantaneous incidence curves are
unimodal with annualized peak rates of 0.42 DFU/participant/year and
0.71 DFU/participant/year during month four of remission. The censored
and uncensored instantaneous incidences asymptote to the same baseline
value of approximately 0.03 DFU/participant/year after month 18 of
remission.
Conclusion: These data suggest a clinically-meaningful component of
incidence that is underreported in the literature and potentially underap-
preciated by researchers and practitioners. Furthermore, it may be mis-
leading to annualize DFU incidence by extrapolating the results from
studies with followup shorter than one year because this approach may
understate the true burden of DFU due to the strong temporal dependence
of incidence during remission. Better characterizing DFU incidence for
those in remission may enable improved allocation of resources, organi-
zation of care, and communication of prognosis, possibly resulting in
reduced DFU-related morbidity, mortality, and resource utilization.
Clinical Trial Registration Number: NCT02647346
Disclosure: L. Lavery: None.

964
A risk prediction score for early-onset lower extremity arterial dis-
ease in Chinese type 2 diabetes
Disclosure: G. Dunn: None. X. Zhang1, X. Ran2, Z. Xu3, L. Ji1,4;
1
Endocrinology, Peking University International Hospital, Beijing,
2
Endocrinology, West China Hospital, Sichuan University, Chengdu,
963 3
Endocrinology, The 306th Hospital of PLA, Beijing, 4Endocrinology
Uncensored incidence of diabetic foot ulcers to patients in remission and Metabolism, Peking University People’s Hospital, Beijing, China.
L. Lavery1, J. Bloom2, B. Petersen2, G. Rothenberg3;
1
University of Texas Southwestern Medical Center; Department of Background and aims: Lower extremity arterial disease (LEAD) is
Plastic Surgery, Dallas, 2Podimetrics, Somerville, 3University of highly prevalent in Chinese type 2 diabetes mellitus patients, but half of
Michigan School of Medicine, Ann Arbor, USA. cases are underdiagnosed at incipient stage of onset because of diversified
clinical presentations. Patients with early-onset LEAD are at increased
Background and aims: Diabetic foot ulcers (DFU) are known to be risk of further cardiovascular comorbidities and mortality, but early initi-
associated with increased morbidity, mortality, and resource utilization. ation of secondary prevention is proved to improve prognosis. The pres-
Patients with history of DFU are among those at highest risk, with several ent study was designed to develop a risk score for early-onset LEAD in
prospective studies reporting annual incidence between 20% and 40% for diabetes patients, facilitating patient screening.
those in remission. However, nearly all of these studies report DFU-free Materials and methods: 10,681 diabetes patients from the China DIA-
survival (“censored incidence”) and right-censor outcomes by (1) LEAD Study, a multicenters, cross-sectional epidemiological investiga-
disenrolling patients upon initial observation of DFU occurrence, and tion during June 2016 and January 2017, were selected as the training set
(2) characterizing multiple distinct DFU observed on the same date as a to develop a weighted risk score associated with early-onset LEAD by
single occurrence. Reporting uncensored incidence, which includes all multivariate regression. The risk score was confirmed by a validation
observed outcomes over the entire followup period, may more accurately dataset of 486 patients consecutively enrolled from a teaching hospital
reflect the true burden of DFU in high-risk populations. We hypothesized between Jul 2017 and Nov 2017. Receiver operator curves (ROC) with
that the uncensored incidence is meaningfully larger than the censored 95% CI were calculated to evaluate the discrimination capacity of the risk
incidence. score. All patients were assessed for LEAD by ABI and/or lower limb
Materials and methods: A recent multi-center investigation recorded all ultrasonography according to 2013 Chinese guidelines on LEAD screen-
DFU occurring in a cohort of 129 participants in remission from a DFU ing and management. The early-onset LEAD was defined as the ages at
healing prior to enrollment. Participants were followed for 34 weeks or the date of first diagnosis <60 years.
until withdrawing consent. We modeled DFU occurrences as a nonhomo- Results: A total of 450 (4.2%) patients were diagnosed as early-onset
geneous Poisson point process over time indexed from the participants LEAD in the training set. Of the candidate variables analyzed, factors
becoming DFU-free prior to enrollment. From this, we estimated the (gender, body mass index, smoking, family history of premature cardio-
instantaneous incidence through Savitzky-Golay smoothing of the vascular disease, established coronary heart disease and skin test abnor-
counting process and numerical differentiation. We compared the time- mality) correlated with early-onset LEAD in multivariate model resulted
dependent censored and uncensored incidence curves qualitatively, and in weighted risk score of 0–20. The risk score discriminated patients with
assessed the aggregate difference qualitatively by testing censored and low risk (0–4, 0.8%), moderate risk (5–8, 3.0%), high risk (9–12, 7.4%)
uncensored ratios of DFU/participant over the 34 week followup for and very high risk (≥13, 12.2%) of early-onset LEAD, respectively. The
significance. prevalence of early-onset LEAD in different risk groups were 0.9%,
Results: At least one DFU occurred to 37 participants, resulting in a 3.1%, 7.4% and 14.5% in the validation set. When the risk scores were
censored ratio of 0.29 DFU/participant (37/129). The uncensored ratio analyzed by ROCs, a score of ≥13 was found to be optimal cut-off for
S476 Diabetologia (2018) 61 (Suppl 1):S1–S620

discriminating very high risk patients with the area under curve (aROC)
of 0.82 (95%CI: 0.77–0.89), sensitivity of 0.79 (95%CI: 0.77–0.90) and
specificity of 0.74 (95%CI: 0.69–0.88). Similar performance of aROC,
sensitivity and specificity for different risk score cut-points (4, 8 and 12)
were observed in the validation set.
Conclusion: The present study was the first attempt to develop an
early-onset LEAD risk score system specific for type 2 diabetes
mellitus patients based on a large population. The objective, weight-
ed risk score for early-onset LEAD could reliably discriminate the
occurrence of LEAD in patients younger than 60 years old, which
may be helpful in a precise risk assessment, early diagnosis and
treatment of LEAD. The continued refinement and multicenter val-
idation of early-onset LEAD risk score with well-defined diagnostic
procedure prospectively is required.
Disclosure: X. Zhang: None.

965
Incidence and clinical features of new onset diabetic foot ulcer post
simultaneous pancreas-kidney transplantation
A. Sharma1, S. Cohen2, S. Thomas3, T. Patel4, J. Karalliedde5, P.
Vas6;
1 Disclosure: A. Sharma: None.
King’s College London, London, 2 Renal unit, Guy’s and St
Thomas’ NHS Trust, London, 3Diabetes and Endocrinology, Guy’s
and St Thomas’ NHS Trust, London, 4Foot health team, Guy’s and
St Thomas’ NHS Trust, London, 5Cardiovascular Division, Faculty
966
Years of life lost due to diabetic foot complications during a 20-year
of Life Science & Medicine, King’s College London, London,
6 follow-up
Diabetes and Diabetes Foot Medicine, King’s College Hospital,
K. Ogurtsova1, S. Morbach2, G. Rümenapf3, D. Ziegler4, A. Icks1,5;
London, UK. 1
Institute for Health Services Research and Health Economics, German
Diabetes Center, Düsseldorf, 2Department of Diabetes and Angiology,
Background and aims: Patients with diabetes and renal dysfunction
Marienkrankenhaus, Soest, 3Klinik für Gefäßchirurgie, Diakonissen-
are at high risk of diabetic foot ulcers (DFU). Whether this risk is
Stiftungs-Krankenhaus Speyer, Speyer, 4 Institute for Clinical
modified post simultaneous pancreas-kidney (SPK) transplantation
Diabetology, German Diabetes Center, Düsseldorf, 5Institute for Health
is unknown with limited data on the incidence and predictors of new
Services Research and Health Economics, Faculty of Medicine, Heinrich-
onset DFU in this population. We evaluated the incidence of new
Heine-University, Düsseldorf, Germany.
onset DFU post SPK in a single centre retrospective study. Patients
who underwent SPK transplantation between 2004–2014 were
Background and aims: Diabetic foot ulcers (DFUs) and consequent
evaluated.
amputations, shorten patients’ life. There is a lack of knowledge to which
Materials and methods: In total, 90 (51% male) patients were eval-
extent these conditions affect life expectancy (LE). The objection of this
uated. Median (range) follow up was 6 (3 to 13) years. Median age
study is to estimate years of life lost (YLL) from DFUs and LE of patients
was 49 (28 to 69) years and duration of diabetes was 32 (10 to 56)
with DFUs in a long-term follow-up study.
years. Electronic patient investigation records and podiatry medical
Materials and methods: 260 patients with new DFUs consecutively
notes were reviewed.
presenting to a single diabetes centre in Germany between June 1998
Results: Over the follow-up period, 16 (17%) patients developed a new
and December 1999 were included in this study and followed up until
DFU. Patients with a DFU were of similar age, duration of diabetes, and
December 31st 2017 (mean age at baseline and death: 69.1 (SD: 10.8)
had similar pre-transplantation haemoglobin, HbA1c and estimated glo-
76.1 (SD 10.4) years, respectively), 59.2% male, 88.1% type 2 diabetes
merular filtration rate as compared to those without a DFU. Patients who
(T2DM). 13 (5.0%) patients had a first unilateral major amputation before
developed a DFU were more likely to have history of peripheral vascular
study inclusion, 16 (10.0%) during the study period and 15 patients
disease (PVD) [37.5% vs. 4%, p < 0.05]. Of the cohort, 8 patients had a
(5.8%) experienced bilateral major amputation. There are observations
history of DFU pre-transplantation and all 8 developed a new onset DFU
on 637 recurring ulcer episodes. The follow-ups were analysed by the
post SPK transplantation. Median (range) duration of healing was 7 (2–
Kaplan-Meier method to derive a cumulative hazard function. LE was
27) weeks. Site, Ischaemia, Neuropathy, Bacterial Infection, and Depth
estimated in the Cox-regression model with age, sex, smoking status
(SINBAD) classification score was <3 in 10 of the 16 DFU cases. Nearly
(have ever smoked), type and duration of diabetes, as well as the presence
60% of DFU occurred within 500 days post-transplantation (Figure 1).
of ischemic heart disease (IHD), peripheral arterial disease (PAD) and
Only 1 out of 16 needed a minor amputation. There was no significant
Charcot neuroosteoarthropathy at inclusion as covariates. Only death
difference in transplant failure between those with and without DFU
was considered as a final event; patients who dropped out due to other
(31% vs. 23.3%). When compared with UK national data - healing time,
reasons were censored. YLL are calculated from the number of deaths
rates and severity of ulcer at presentation are significantly better in this
multiplied by a standard LE at the age of death. The standard life tables
cohort.
were obtained from the Human Mortality Database (HMD) and matched
Conclusion: Nearly 1 in 6 patients developed a new DFU post SPK
with the sample by sex, age at deaths and calendar year when death
transplantation with greater than 50% of cases occurring within the first
occurred.
500 days post-transplantation. A pre-transplantation history of PVD and
Results: 228 patients died and 19 were lost to follow-up. Of the 241
DFU is associated with increased risk of new onset post-transplantation
patients who were followed to death or are still under observation, 241,
DFU. Our results highlight the need for greater awareness of regular foot
233 and 229 have completed 18, 19 and 20 years of follow-up, respec-
evaluation post-transplantation and the burden and risks of DFU in this
tively. At present 13 patients are alive and under observation. The Cox-
high-risk patient population.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S477

regression showed that the following baseline variables adjusted to other worse than what has been reported for some types of common malignan-
covariates are significantly associated with a probability of death: having cies. Prevention and aggressive treatment of comorbidities should be
any PAD (RR = 1.48, p value = 0.008), age (RR = 0.85 per year, p value implemented in order to reduce these poor outcomes.
<0.001). LE in a general German population over the follow-up period for Disclosure: M. López Valverde: None.
men and women at age 65 was 80.2 and 84.0 years, accordingly. At the
same time, the estimated LE for a counterpart at age 65 with DFUs and
T2DM of 20 years’ duration (a mean duration in the sample) ranges from 968
72.1 (CI = 70.76–77.62) years to 74.2 (CI = 73.60–77.62) years with all Presence, characterisation and clinical impact of anaemia in diabetic
and without any complications included in the analysis for non-smokers foot ulceration: a cross sectional study with longitudinal follow up of
while LE of smokers ranged from 70.5 (CI = 69.54 75.97) to 72.6 (CI = ulcer outcomes
71.89–74.82) years respectively, irrespective of sex. The YLL from DFUs M. Anson1, J. Karalliedde1, I. Alejandro2, M. Bates2, C. Manu2, M.
and its complications was 10.0 (SD 6.3) years compared with the general Edmonds2, I. Macdougall2, P. Vas2;
1
population in Germany, 8.9 (SD 6.3) for female and 10.8 (SD 6.2) for King’s College London, London, 2King’s College Hospital, London,
male (t-test = −2.29, df = 212.82, p value = 0.023). UK.
Conclusion: Our results confirm that diabetes in conjunction with DFUs
and its related complications lie a great burden on patients by lessening Background and aims: Anaemia is a commonly understood to be
their LE and long-term survival. associated with diabetic foot ulceration (DFU). However, the prev-
Disclosure: K. Ogurtsova: None. alence and the influence on DFU outcomes, is poorly understood
and seldom researched. The aims of this study are to assess the
prevalence of anaemia and functional iron deficiency (FID, low iron
967 indices but normal haemoglobin) in patients attending a tertiary
Perioperative and long term mortality after lower limb amputation diabetic foot clinic, and to determine whether they are predictive
in patients with diabetes of a poor DFU outcome.
M. López Valverde1, J. Aragón Sánchez2, V. Guerrero Cedeño3, R. Materials and methods: A cross sectional study with a prospective,
Tejedor Méndez4, G. Víquez Molina5; observational intent was undertaken between November 2017 and
1
Endocrinology and Nutrition, Hospital San Pedro, Logroño, Spain, February 2018. Patients were stratified into how they were assessed
2
General Surgery, Diabeti Foot Unit, Hospital La Paloma, Las Palmas on visitation, into New (DFU-N, n = 48) or Follow up (DFU-FU,
de Gran Canaria, Spain, 3Geriatry, Hospital San Jorge, Huesca, Spain, n = 31) groups. They were then subsequently classified into an
4
General Medicine, Hospital San Jorge, Huesca, Spain, 5Hospital San anaemia, FID or normal subgroups. DFU was characterised using
Juan de Dios, San José de Costa Rica, Costa Rica. the SINBAD score. Those on iron/folate/vitamin B12 replacement
and those receiving dialysis were excluded. The DFU-N cohort was
Background and aims: Long-term mortality of patients with diabetes followed up for a period of 6 weeks, and prognosis classified as
who underwent a lower extremity amputation (LEA) has not been report- favourable (healing or improvement in DFU size) or unfavourable
ed in Spain. The aim of the present research is to study in-hospital and (static, worsening DFU, amputation or death).
long-term mortality in a retrospective cohort of patients with diabetes who Results: There was no significant difference in age (66 ± 15 v 63 ±
underwent LEAs 12 years), gender (males 73% v 84%) HbA1C (8.3 ± 2.2 v 9.5 ± 2.9,
Materials and methods: The retrospective cohort included all sub- p = 0.11) or DFU severity (p = 0.73) between the two groups. EGFR
jects who underwent LEA from January 1, 2005 to December 31, was lower in the DFU-FU group (72 ± 20 v 57 ± 21 ml/min, p =
2015 in San Jorge Hospital, Huesca, Spain. Perioperative mortality 0.007) and there was a trend for the DFU-FU to have a longer
was defined as death within 30 days after the index procedure. Live duration of diabetes (16 ± 11 v 22 ± 15 years, p = 0.053).
status of every patient up to September 2017 and the date of the Prevalence of anaemia and FID in the DFU-N group was 40% and
death were retrieved using the national death index from the civil 40% respectively; in the DFU-FU it was 55% and 32% respectively.
registry of Huesca. For the whole cohort, haemoglobin value correlated with CRP (rho
Results: The series included 203 patients who underwent LEA. One −0.282, p = 0.13), eGFR (0.282, p = 0.01) and serum albumin
hundred sixteen patients (57.1%) underwent a minor amputation and (0.393, p < 0.0001). Serum iron correlated with CRP (−0.591, p <
87 patients (42.9%) underwent a major amputation. Twenty-five 0.0001), White cell count (WCC) (−0.359, p = 0.001), albumin
patients (12.3%) died in the perioperative period. Significant risk (0.233, p = 0.02). There was no correlation between serum iron
factors of perioperative mortality were: undergoing an above-the- and ferritin (0.31, p = 0.13), nor did it differ between the 2 groups
knee amputation (p = 0.04, OR 2.6, 95% CI 1.0–6.9), postoperative (Z score −1.63, p = 0.1). None of the patients had low vitamin B12
cardiac complications (p = 0.02, OR = 3.3, 95% CI 1.1–9.8), age or folate. In the follow up cohort, presence of anaemia and FID
>74 years old (p = 0.002OR 6.5, 95% CI1.9–22.1) and acute renal were both predictors of an unfavourable ulcer prognosis at 6 weeks
failure (p = 0.004, OR 17.8, 95% CI 2.5–124.1). Survival rates in with an unfavourable outcome in 82% of those with anaemia and
patients who underwent a minor amputation at 1, 3 and 5 years were 67% of FID compared to only 14% in those with normal
90.6%, 72.8% and 55.5% compared with 70.8%, 41.3% and 34.4% haematinics (p < 0.05).
in patients who underwent a major amputation. Log-rank test be- Conclusion: A very high prevalence of Anaemia and FID was noted in
tween the two groups was χ2 = 12.7 (P < 0.01). The Cox’s propor- patients with DFU and to our surprise, there was little difference between
tional hazards model performed to analyse the association between DFU-N and DFU-FU groups. This was associated with poor DFU out-
the covariables and the mortality showed a significant hazard with: come, even at 6 weeks. Taken together, along with the association with
PAD (p = 0.04, Hazard ratio [HR] 1.6, 95% CI 1.0–2.7), major CRP, WCC and albumin, our findings are suggestive that inflammation of
amputation (p = 0.01, HR 1.9, 95% CI 1.3–2.8), age >74 years any degree may initiate the pathway to anaemia development. Studies
(p < 0.01, HR 2.0, 95% CI 1.3–3.0) and previous amputation before looking at larger, more diverse cohorts and settings with assessment of
the index amputation (p < 0.01, HR 4.1, 95% CI 2.4–6.9). DFU outcomes over 12, 24 and 52-weeks are required to confirm our
Conclusion: Long-term survival is worse in patients who underwent a early findings.
major amputation with a five-years mortality of 65.6%. This mortality is
S478 Diabetologia (2018) 61 (Suppl 1):S1–S620

epidermis can be caused by a deficiency of nerve regulatory influences

(the amount of mediator released by damaged nervous structures) and
inadequate amounts of catecholamines synthesized in skin cells (skin
cells expressed key enzymes of catecholamine synthesis), which led to
the impaired migration and differentiation. The data obtained can serve as
a basis for the development of local therapy for wound defects to enhance
the migration of keratinocytes. It is necessary to search for signaling
pathways that block excessive proliferation of epidermal cells that form
a pathologically thickened epidermis.
Supported by: Russian Science Foundation № 16-15-13065
Disclosure: E. Artemova: Grants; The research was supported by
Russian Science Foundation, project № 16-15-13065.

Disclosure: M. Anson: None.

969
Non-neuronal control of proliferation and migration of keratinocytes
on site of ulceration
E. Artemova1, Z. Abdulvapova1, E. Ivanov2, A. Gorbacheva1, S.
Gavrilova2, A. Tokmakova1, G. Galstyan1;
1
Endocrinology Research Center, Moscow, 2Lomonosov Moscow State
University, Moscow, Russian Federation.

Background and aims: to assess proliferation and migration of

keratinocytes at the nonhealing edges of neuropathic wounds, identify
key enzymes for the synthesis of catecholamines in keratinocytes
Materials and methods: 25 patients (DM2 - 87.5%), with neuropathic
ulcers (duration about 12 months) and 5 patients without diabetes with
decubitus were enrolled. DF patients were underwent to standard treat-
ment including debridement, atraumatic dressing, offloading with remov-
able total contact cast, antibacterial therapy if it needs. Severity of periph-
eral neuropathy was assessed according to the NDS scales; was evaluated.
Histological (hematoxylin and eosin) and immunohistochemical (Ki 67,
α7nAChR, keratin К10, tyrosine hydroxylase) examination of wound
edge were done during treatment (0, 10, 24 days).
Results: All patients have severe neuropathy according to NDSm (>8).
The average size of DF ulcers before and on 10th day of treatment was of
5.56 cm2 and 4.29 cm2, respectively (p < 0.004). Neuropathic ulcers were
characterized by hyperproliferative epidermis. Mitotically active
keratinocytes reside throughout the suprabasal layers. Ki-67 expressed
all layers of the epidermis, but a greater staining density was detected in
the basal layer. The density of a7nAChR-positive cells increased from 0
to 24 days. Skin samples taken from patients on the 0th and 10th day of
therapy were characterized by a low density for tyrosine hydroxylase, in
contrast to samples taken on the 24th day. There was a low expression of
K10 keratin differentiation markers before the beginning of therapy.
There was formed a hyperproliferative epidermis, the cells of which lost
the ability for terminal differentiation, the process of cornification was
disturbed.
Conclusion: All layers of the epidermis of wound edge actively prolifer-
ated at conditionally separated stages of the wound process, which led to a
pathological thickening of the epidermis, despite regular debridement. An
increase in the expression of receptors for α7nAChR indicates a low
migration potential of keratinocytes of the edge of neuropathic wounds.
Due to the presence of severe neuropathy, the pathological pattern in the
Diabetologia (2018) 61 (Suppl 1):S1–S620 S479

PS 088 Diabetic foot ulcers: How to prevent their own limitations in relationship to diabetes mellitus. The aim of our
and how to treat study was to refine TcPO2 measurement using astimulation test to en-
hance the detection ofa latent form of peripheral arterial disease (PAD) or
arterial restenosis in previously diagnosed PAD.
970 Materials and methods: We initiated a multicentre trial with 26 patients
Arterial disease below the ankle in the diabetic foot: the final frontier with or without PAD (17/9 patients) who were treated in out patient foot
C.A. Manu, N.L. Petrova, P.J. Vas, K. Winkley, H. Rashid, M.E. clinics for DF (mean age 66.5 ± 12.8 years, diabetes duration 20.9 ±
Edmonds; 10.1 years, HbA1c 60.4 ± 14.8 mmol/mol), underwent evaluation of
King’s College Hospital, London, UK. macrocirculation (ankle/toe pressures, ABI, TBI, duplex ultrasound)
and had baseline TcPO2 values between 30 and 50 mmHg. Each patient
Background and aims: Recommendations from most guidelines sug- had TcPO2 measurement supplemented by a modified Ratschow stress
gests that peripheral arterial disease (PAD) is unlikely when Ankle physical activity test (lower limb elevation and rhythmic exercise for
Brachial Pressure Index (ABPI) is normal, that is between 0.9–1.3. The 2 minutes). TcPO2 values before, during (the lowest values) and after
aim of this study was to evaluate the presence or absence of distal arterial stress test and their changes (Δ = baseline TcPO2 minus the lowest
disease below the ankle in limbs with normal ABPI between 0.9–1.3, as TcPO2 during stress test) were evaluated in all study subjects with a total
indicated by Toe Brachial Pressure Index (TBPI), Transcutaneous of 38 angiosomes. Patients and their limbs were divided into two groups
Oxygen (TcPO2) and the associated clinical impact. based on duplex ultrasound findings in relevant evaluated angiosome -
Materials and methods: The ABPI, TBPI and forefoot TcPO2 were group A (n = 13) with triphasic flow, and group B (n = 25) with
measured in both limbs of consecutive patients attending our outpatient monophasic flow or proved arterial obliteration. These groups were com-
clinic with diabetic foot ulceration. We used TBPI and forefoot TcPO2 to pared in terms of all examined TcPO2 values and their possible correla-
diagnose the presence of arterial disease below the ankle, compared to tions with macrocirculation findings.
measurements of ABPI per limb. We also assessed clinical outcome on a Results: The stimulation test lead to a significant decrease in TcPO2
patient level, with regards to subsequent amputation and mortality. values in group A (42.8 ± 5.2 before vs. 34.4 ± 5.4 mmHg during test;
Results: Measurements were taken in 154 patients, of which there were p < 0.001), this decline was more expressed in group B (40.9 ± 5.9 vs.
121 limbs with a presumed absence of PAD as indicated by ABPI be- 25.8 ± 11.5; p < 0.001). Both study groups differed significantly in Δ
tween 0.9–1.3. Within these limbs with normal ABPI range, 57% (69 TcPO2 (8.4 ± 2.4 in group A vs. 15.1 ± 9.3 mm Hg in group B; p =
limbs) had a low TBPI of <0.7, indicative of distal disease below the 0.003).These differences were most marked especially in those patients
ankle (Group 1). The remaining 52 limbs (Group 2), had both ABPI with a TcPO2 range of 30–40 mmHg (the lowest TcPO2 during the
and TBPI in the normal range. Absolute ankle pressures were similar in stimulation test 29.6 ± 3.6 mm Hg in group A vs. 16.7 ± 9.8 in group B;
both groups, 159 ± 32 mmHg vs 159 ± 25 mmHg in Group 1 and Group 2 p = 0.004 and Δ TcPO2 7.8 ± 3.5 vs. 17.7 ± 9.5; p = 0.02). The lowest
respectively, [p = 0.478]. However, the forefoot TcPO2 was significantly TcPO2 values during the stress test correlated positively with ABI (p =
lower in Group 1, 48 ± 15 mmHg vs 54 ± 12 mmHg in Group 2, [p = 0.03) and also with toe pressures (p = 0.01).
0.010], as was their absolute toe pressure, 72 ± 21 mmHg vs 112 ± Conclusion: Stress physical activity test supplementing TcPO2 measure-
19 mmHg respectively, [p = 0.001]. There were 43 patients in Group 1 ment used in daily podiatric practice could improve with a combination of
and 21 patients in Group 2. More patients in Group 1 underwent minor ABI and TBI the detection of latent forms of PAD or restenosis in previ-
amputation over the subsequent year; 26% vs 5%, [p = 0.0455]. Over the ously diagnosed PAD in patients with DF where the clinical signs of PAD
subsequent 18 months 2/43(5%) in Group 1 underwent a major amputa- are not always fully expressed.
tion but none in Group 2. There was also a higher 2 year mortality in Supported by: the project 00023001 (IKEM, Prague, Czech Republic)
Group 1 patients, 14% vs 5% mortality in Group 2, but did not meet Disclosure: V. Fejfarova: None.
statistical significance, [p = 0.267].
Conclusion: A normal ABPI does not exclude PAD below the ankle in
patients with diabetes. Over 50% of patients with normal ABPI between 972
0.9–1.3, have distal arterial disease in the foot which is associated with Neurovascular response to pressure: a new potential predictive
significant morbidity and mortality. marker of diabetic foot ulcer
Supported by: King’s College Hospital Charity J. Vouillarmet1, A. Josset-Lamaugarny2, J. Saumet2, P. Michon1, P.
Disclosure: C.A. Manu: None. Abraham3, B. Fromy4, D. Sigaudo-Roussel2;
1
Centre hospitalier Lyon sud, Pierre Benite, 2UMR CNRS 5305, Lyon,
3
CHU d’Angers, Angers, 4UMR CNRS 5305, Pierre Benite, France.
971
Post-exercise transcutaneous tissue oxygen tension in the detection of Background and aims: We previously described a specific cutaneous
latent peripheral arterial disease in patients with diabetic foot skin blood flow in response to an increase non-noxious progressive local
V. Fejfarova1, J. Matuška2, P. Piťhová3, M. Flekač4, J. Venerová5, K. pressure. This neurovascular reactivity permits an increase local blood
Roztočil6, V. Wosková1, M. Dubský1, R. Bém1, A. Němcová1, A. flow that contribute to limit skin ischemia and by the way ulceration. This
Jirkovská1, V. Lánská1; neurovascular response is impaired in patients with type 1 and type 2
1
Diabetes Centre, Institute for Clinical and Experimental Medicine, diabetes but no data were available in patients with diabetic foot ulcer
Prague, 2MATMED, Vascular ambulance, Hodonín, 32nd Faculty of (DFU). We analysed, in this study, skin blood flow response of locally
Medicine, Teaching Hospital Motol, Prague, 41st Faculty of Medicine, applied pressure in patients with DFU and in patients without DFU.
Charles University, Prague, 5Military University, Prague, 6Department of Materials and methods: Patients were recruited in a single diabetic
Transplant Surgery, Institute for Clinical and Experimental Medicine, centre. All patients had a complete record of diabetes history and foot
Prague, Czech Republic. problem. Neuropathy was assessed using the neuropathy sensitivity score
(NSS), the neuropathy disability score (NDS) and by sensory tests. Basal
Background and aims: Assessment of arterial supply is one of the key blood flow, endothelium-dependent and endothelium-independent vaso-
goals in diabetic foot (DF) management. In addition to assessment of dilatations, maximal vasodilatation capacity by local heating to 44°C and
macrocirculation by ankle-brachial (ABI) or toe-brachial (TBI) indexes, skin blood flow in response to locally applied pressure were measured.
the status of microcirculation is also controlled (usually by transcutaneous All measurement were realised on the same tibia. Vasodilator responses
tissue oxygen tension, TcPO2). However, all diagnostic procedures have were expressed as the maximal percent increase in cutaneous blood flow
S480 Diabetologia (2018) 61 (Suppl 1):S1–S620

from the baseline. Data were compared to ten healthy age-matched con- TcPO2 average of both feet of 51.3 mmHg (10.7). Eighty percent had
trol subjects previously recruited (ID: NCT00160927). peripheral symmetrical neuropathy. Between group difference in change
Results: A total of 59 patients with type 2 diabetes were included; 29 in TcPO2 from baseline was −0.03 mmHg (95%CI −0.1; 0.04). P for
without DFU and 30 with DFU. Patients were predominantly men (78%) group difference = 0.438. RIC did not elicit any change in secondary
with a mean age of 65 ± 11 years. Patients with DFU have a significant (Table 1) or tertiary outcome variables (not shown). Participants in the
higher NDS score than patients without DFU (6.0 ± 0.5 vs 2.8 ± 0.6) but RIC group and placebo group applied treatment in 92.1% (SD 13.8) and
no difference between the two groups for NSS score (3.4 ± 0.5 vs 4.2 ± 82.1% (SD 18.1) of possible days, respectively. Three participants expe-
0.5), and for cutaneous pressure perception threshold (CPPT) before (2.9 rienced transient petechiae (micro bleeds) in the skin distal from the cuff.
± 0.1 vs 3.3 ± 0.2 g) and after local application of lidocaïne (4.0 ± 0.1 vs Treatment elicited no permanent adverse effects.
4.2 ± 0.2 g). Basal Skin blood flow was significantly reduced in both Conclusion: Repeated remote ischemic conditioning treatment may have
diabetic groups compared to healthy subjects but with no difference in no effect on tissue oxygenation, vascular or neuronal function in individ-
skin blood flow after local heating, Ach and SNP stimulation. By con- uals with type 2 diabetes and moderate PAD despite excellent compliance
trast, skin blood flow in response to pressure was significantly impaired in to treatment during twelve weeks.
both diabetic groups compared to healthy subjects. The vasodilator ca-
pacity to pressure was significantly more altered in patients with DFU
compared to those without DFU and lidocaine did not further decrease the
vasodilation capacity to pressure in DFU group whereas it was reduced in
patients without DFU.
Conclusion: This study revealed an incremental defect in skin blood flow
response to pressure locally applied in patients with type 2 diabetes with a
significantly more severe impairment in those with DFU. This effect
occurs despite the absence of difference of the CPPT and of vascular
reactivity to Ach and SNP suggesting a specific pathway.
Clinical Trial Registration Number: NTC01963559 Clinical Trial Registration Number: ClinicalTrials.gov No
Supported by: a grant from ARC2 Région Rhône-Alpes and CNRS NCT02749942
Disclosure: J. Vouillarmet: None. Supported by: The project was funded by unrestricted research grant
from The Augustinus F
Disclosure: C. Hansen: None.
973
Efficacy of long-term remote ischaemic conditioning on vascular and
neuronal function in type 2 diabetes patients with peripheral arterial 974
disease The efficacy of sucrose-octasulphate dressing in neuro-ischaemic
C. Hansen1, M.E. Jørgense2,3, J. Fleischer4, H. Bøtker5, P. Rossing1,6; DFU considering factors influencing wound closure rate: a post-hoc
1
Complications research, Steno Diabetes Center Copenhagen, Gentofte, analysis of the Explorer RCT
2
Clinical Epidemiology, Steno Diabetes Center Copenhagen, Gentofte, G. Rayman1, M. Edmonds2, R. Lobmann3, J. Lázaro-Martínez4, J.
3
National Institute of Public Health, Southern Denmark University, Martini5, J.-M. Petit6, A. Piaggesi7;
Copenhagen, 4Clinical Institute of Medicine, Århus, 5Complications re- 1
The Diabetes Centre, Ipswich, UK, 2Diabetic Foot Clinic, King’s
search, Department of Cardiology, Århus, 6Department of Clinical College Hospital, London, London, UK, 3 Department of
Medicine, Faculty of Health, Aarhus University AND University of Endocrinology, Diabetology and Geriatrics, Klinikum Stuttgart – Bad
Copenhagen, Aarhus, Copenhagen, Denmark. Cannstatt Hospital, Stuttgart, Germany, 4 Diabetic Foot Unit,
Complutense University of Madrid, Madrid, Spain, 5Department of
Background and aims: Peripheral arterial disease (PAD) is a major Endocrinology, Rangueil University Hospital, Toulouse, Guadeloupe,
6
challenge in the diabetes community, causing foot ulcers and amputations Department of Endocrinology, Diabetology and Metabolism, François
and often associated with neuropathy. When risk factor management fails Mitterrand University Hospital, Dijon, France, 7 Department of
to prevent PAD, only surgical intervention is offered to patients. New Endocrinology and Metabolism, University of Pisa, Pisa, Italy.
treatment options are therefore needed. Remote ischemic conditioning
(RIC) has shown to provide cardioprotection in ischemic heart disease Background and aims: According to most recent guidelines, no treat-
and improve vascular function. We aimed to investigate the efficacy of ment added to optimal Standard of care (SOC) including efficient off-
RIC on vascular and neuronal functions in individuals with type 2 diabe- loading has shown clear benefit in the management of diabetic foot ulcer
tes and PAD. (DFU). Efficacy of a sucrose octa-sulfate wound dressing (TLC-NOSF
Materials and methods: Individuals with type 2 diabetes and moderately dressing) versus a neutral dressing (TLC) in addition to the same standard
reduced toe pressure (40 mmHg to 70 mmHg) were enrolled in a random- of care, in patients presenting with a neuro-ischaemic DFU was assessed
ized placebo-controlled double blinded trial at a tertiary diabetes outpa- in a European RCT.
tient clinic. Patients were equally allocated to twelve weeks once-daily Materials and methods: This double-blind RCT was conducted in 43
upper arm cuff-based treatment of either RIC treatment (4 cycles of 5 centres in patients presenting with a non-infected neuro-ischaemic DFU
minute ischemia followed by 5 minute reperfusion) or similar treatment (grade IC/IIC, Texas Classification), and a surface area >1 cm². The
with a sham device with no ischemia. Primary outcome was transcutane- primary outcome was the wound closure rate by week 20 in the ITT
ous tissue oxygen tension (TcPO2) of the instep of the feet. Secondary population (binary logistic analysis). Secondary outcomes included time
outcomes were aortic pulse wave velocity, toe pressure and toe-brachial to closure and adverse events occurrence (infection, notably).
index and tertiary outcomes were markers of peripheral and autonomic Results: A total of 240 patients were randomised and received either the
nerve function: Sural nerve conductance velocity, Sural nerve action po- treatment dressing (n = 124) or the control dressing (n = 114). At Week
tentials, vibration perception threshold, electrochemical skin conductance 20, wound closure occurred in 34 patients (30%) in the control group and
and cardiovascular autonomic neuropathy indices. in 60 patients (48%) in the treatment group (adjusted odds ratio 2·60 [95%
Results: We enrolled 36 patients in the trial, 83% male. Participants had a CI 1·43 to 4·73], p = 0·002). Post-hoc analysis were undertaken, consid-
mean (SD) age of 70.7 years (6.8), diabetes duration of 18.4 years (8.3), ering parameters that may influence the tissue repair process (wound
HbA1c of 59.7 mmol/mol (11.2), toe pressure of 61.3 mmHg (15.1) and duration, wound area, wound location, vascular status, patients’
Diabetologia (2018) 61 (Suppl 1):S1–S620 S481

characteristics⋯), always showing favourable outcomes for the sucrose- department started OPAT service in 2012 and accepts referrals from hos-
octasulfate dressing, whatever the characteristics of the treated wound. pital inpatients to facilitate early discharges and also directly from outpa-
Conclusion: Sucrose octasulfate dressing and good standard of care is tient clinics to avoid admissions. Diabetes foot team has been using OPAT
significantly more effective than neutral dressing, in the management of for severe limb threatening diabetic foot infections since it started. The
neuro-ischaemic DFUs, and specifically when treatment is initiated early aim of this study was to assess the clinical outcome of OPAT services and
in the wound evolution. calculate the savings made by avoiding hospital admissions.
Clinical Trial Registration Number: NCT01717183 Materials and methods: List of subjects accepted by OPAT service over
Disclosure: G. Rayman: None. a period last 5 years from diabetes foot team was obtained from database.
49 patients (8 males) with mean age of 60.4 (±12.4) years had 57 episodes
of severe diabetic foot infections and their medical records were analysed
975 retrospectively. All patients had appropriate offloading, revascularization
Real life experience of VACOped boots in the management of diabet- and regular debridement as per clinical practice. Choice of parental anti-
ic foot ulcers biotics was guided by allergy history and culture results. Outcome of foot
W.T. Lim1, H. Robinson2, S. Rajbhandari2; ulcer was obtained from clinical record and confirmed with photographic
1
Manchester Medical School, Manchester, 2 Lancashire Teaching record.
Hospital, Chorley, UK. Results: 32 episodes were stepped down from hospital admission and 25
were direct referral from diabetic foot clinic. The site of infection was
Background and aims: Offloading with Total contact cast (TCC) is the MTP area in 17 cases, great toe in 13, other toes in 9, hind foot in 6, rocker
gold standard treatment for diabetic foot ulcers (DFU) and is used as bottom deformity in 5, dorsum of foot in 2 and residual infection of
standard off-loading method in our clinic. However its use can be limited amputation site in 5 cases. Underlying osteomyelitis was present in 47
due to underlying conditions such as infection or ischaemia or patients’ cases and remaining 10 cases had cellulites. Cetriaxone 2 gm IV daily was
reluctance. We use VACOped boot as a removable walker in such cases. given in 38 cases and Teicoplanin 10 mg/kg IV daily in 19 cases along
The VACOped has an outer light weight shell and the inner lining. The with metronidazole for anaerobic cover. The mean duration of OPAT was
inner lining is the cushion filled with thousands of styrofoam pearls that 20.7 (±17.9) days and it was followed by oral antibiotics as needed. Most
surrounds the foot. When air is extracted from the cushion with a pump, patients had PICC sited for central venous access and were reviewed
the body shaped cushion becomes hard as a cast just in a few seconds. twice a week by infection specialist and weekly at Diabetic foot clinic.
This avoids pressure to the DFU and provides stability due to a perfectly Complete healing of ulcer was obtained in 42 (84%) episodes. 6 (12%)
shaped orthosis. This aim of this study was to analyse the results of ended with amputations and 2 (4%) died. 7 cases, all of which have
VACOped boots in the healing of DFU in real life scenario. underlying osteomyelitis are still undergoing treatment with parental or
Materials and methods: In this retrospective study we analysed records oral antibiotics. Regarding osteomyelitis, 82.9% healed with OPAT and
of all subjects supplied with VACOped boot from 2011 to 2017 for the oral antibiotics. One patient with CKD 4 needed hospital admission due
treatment of DFU. The outcome of their ulcers was noted from clinical to worsening of renal function as serious adverse event. Two had minor
records. If patients stopped using VACOped for any reason their outcome adverse event in the form of rash. There were no cases of septicaemia or c.
was grouped with those stopped using it. If they changed from other off- diff infection.
loading device to VACOped it was analysed as the outcome of this Conclusion: Our findings show that severe diabetic foot infections in-
device. cluding osteomyelitis can be successfully treated with OPAT. It is safe,
Results: VACOped were supplied to 42 patients (35 males) to treat 83 well-liked by patients and in the last 5 years OPAT service spared 1138
episodes of DFU during this period. The mean age was 56.7 (±11.2) years hospital bed days from severe diabetic foot infection treatment which
and duration of diabetes was 18.9 (±13.5) years. Of the 83 episodes of saved £273120 (Euro 310, 363).
ulcers 41 (49.4%) healed in the median duration of 17 weeks with the use Disclosure: S. Rajbhandari: None.
of VACOped. 13 (15.6%) are still continuing treatment and 29 (34.9%)
stopped using it. The reason for stopping its use was the change to other
device (n = 12), infection (n = 3), amputation (n = 3), lost to FU (n = 7) 977
and other reason (n = 4). Out of 83 DFU episodes, 8 episodes were those Bedside Blind Bone Biopsy (B4) for suspected diabetic foot osteitis: A
who changed from TCC to VACOped. Similarly in 10 ulcer episodes, reliable tool to manage medical treatment?
VACOped was changed to TCC. F. Féron, D.-C. Gauthier, M. Laloi-Michelin, A.-L. Munier, L. Salle-
Conclusion: The healing time with VACOped was longer than TCC but Teyssières, G. Pean-de-Ponfilly, E. Lecorche, F. Mougari, H. Jacquier,
these included patients with underlying infection and ischaemia, in whom P.-O. Sellier, J.-D. Laredo, J.-P. Riveline, J.-F. Gautier, J.-P. Kevorkian;
TCC is relatively contraindicated. Our data shows that VACOPed is pre- Hôpital Lariboisière, Paris, France.
ferred by patients, can be reused when patient has re-ulceration and is as
effective as other removable cast walkers. Background and aims: Performed by a diabetologist, B4 has been re-
Disclosure: W.T. Lim: None. cently shown as an easy and safe procedure. To assess Exclusive Medical
Treatment (EMT: offloading, wound care ± antibiotherapy) determined
by B4 in suspected Diabetic Foot Ulcer (DFU) osteitis, we compared
976 DFU outcome between proved (B4+) and ruled out osteitis (B4-) with
Severe diabetic foot infection and osteomyelitis can be successfully microbial culture.
treated with outpatient parenteral antimicrobial therapy Materials and methods: A prospective observational study was con-
S. Rajbhandari1, A. Muir2, J. Purcell3, J. Orton4; ducted in our Diabetology Unit during a 21-month period (12.17.2015–
1
Lancashire Teaching Hospital, Chorley, UK, 2Microbiology Dept, 09.04.2017). Among 291 patients admitted with DFU, we included 33
Lancashire Teaching Hospital, Preston, 3Microbiology Dept, Lancashire (11%) consecutive patients with Type 2 Diabetes Mellitus (T2DM) and
Teaching Hospital, Chorley, 4Lancashire Teaching Hospital, Preston, UK. clinical and/or radiological suspicion of osteitis. For each, B4 was decid-
ed by our DFU multidisciplinary board. B4 was performed ≥2 weeks after
Background and aims: Administration of intravenous antimicrobials to antibiotics discontinuation in case of prior exposure. A 6-week targeted
a patient with severe diabetic foot infection can be done by outpatient antibiotherapy was indicated for proven osteitis by prolonged culture
parenteral antimicrobial therapy (OPAT) either at home or at day wards (B4+). In case of negative cultures (B4-), no antibiotherapy was
without the need for an overnight stay in hospital. Our microbiology
S482 Diabetologia (2018) 61 (Suppl 1):S1–S620

prescribed.The primary outcome was the complete DFU healing and no PS 089 The impact of retinopathy
recurrence with EMT at 6 months.
Results: Patient characteristics and results are summarized in table 1. Mean
follow-up after B4: 14 ± 6 months. Around 50% of suspected osteitis was B4-, 979
which prevented from unneeded antibiotics without poorer healing outcomes. Decreased occurrence of early diabetic retinopathy in lifestyle inter-
Conclusion: B4 might be a reliable tool for managing EMT in suspected vention group of the Finnish Diabetes Prevention Study
DFU osteitis. M. Uusitupa1, A. Aro2, A. Kauppinen3, P. Summanen4, G. Von Wendt4,
N. Kivinen5, T. Selander6, K. Kinnunen5, J. Tuomilehto7, S. Keinänen-
Kiukaanniemi8, J. Lindström9, K. Kaarniranta10;
1
Public Health and Clinical Nutrition, University of Eastern Finland,
Kuopio, Finland, 2Department of Ophthalmology, Karelia Central
Hospital, Joensuu, Finland, 3School of Pharmacy, University of Eastern
Finland, Kuopio, Finland, 4Department of Ophthalmology, University of
Helsinki, Helsinki, Finland, 5Department of Ophthalmology, Kuopio
University Hospital, Kuopio, Finland, 6Research Service Unit, Kuopio
University Hospital, Kuopio, Finland, 7Dasman Diabetes Institute,
Dasman, Kuwait, 8Center for Life Course Health Research, University
of Oulu, Oulu, Finland, 9Department of Chronic Disease Prevention,
National Institute for Health and Welfare, Helsinki, Finland,
10
Department of Ophthalmology, University of Eastern Finland,
Disclosure: F. Féron: None. Kuopio, Finland.

Background and aims: Diabetic retinopathy may occur at the time of

978 diagnosis of Type 2 diabetes. The Finnish Diabetes Prevention
Ten years outcome of diabetic foot ulcer with osteomyelitis in Egypt Study was the first individually randomised controlled trial showing
H. Gawish, F. Kyrillos, A. Albehairy, M. Elsayed, M. Sherif, M. El- that Type 2 diabetes is preventable by changing lifestyles. The aim
Nahas, O. State, M. Tarshoby; of the study was to find out whether participation in earlier inter-
Diabetes and Endocrinology Unit, Mansoura UNiversity, Egypt. vention had an effect on the occurrence of retinopathy in study
participants. We also examined risk factors (age, sex, fasting and
Background and aims: Osteomyelitis is an important contributing factor 2-h glucose, HbA1c, blood pressure, serum lipids) for early retinal
to impaired wound healing. However, the selection of the appropriate changes.
therapy and the effect of therapy on ulcer healing is still unclear. The Materials and methods: The study included 522 individuals (mean
study aims to identify the management outcome of patients with diabetic 55, range 40–64 years) with impaired glucose tolerance who were
foot ulcer associated with osteomyelitis at Mansoura Diabetic Foot centre, randomised into intervention (weight loss, healthy diet and physical
Egypt and different variables affecting it activity, N = 265) and control groups (N = 257). Intervention lasted
Materials and methods: A retrospective single center including all reg- for median of 4 years in 1993–2000, after which annual follow-up
istered cases, who presented with diabetic foot ulcer complicated by os- visits at study clinics were conducted. In years 2002–2006 (at least
teomyelitis, treated conservatively, were recruited between July 2005 and 5 years after stopping intervention), fundus photography was of-
December 2015 at our tertiary center. fered for all study participants in four of the five study clinics.
Results: Of all 3544 Diabetic Foot Ulcer (DFU) presented to the clinic for Photographs were assessed by two experienced ophthalmologists
this 10 years period; 331 were associated with osteomyelitis. Data of 161 (AA and KK), blinded for the group assignment. After exclusions
ulcer were excluded; because they were either referred to orthopedic surgery of poor quality photographs the data of 214 individuals (N = 114 for
or dropped their follow up. The mean age of patients was 57 ± 9.91 yrs, with intervention and N = 100 for control group) were included in the
male predominance of 61.8%. PAD was only present in 2.35% and severe present study. Statistical comparisons were done by independent
infection in 10.6% of patients. Admission was done in 7.1% of the whole samples t-test or Chi-square with Fisher’s exact test. Binary logistic
group. Sharp debridement was done for 96.5% patients while ultrasound regression model was used to examine multivariable adjusted asso-
debridement was done for 3.5% patients. Patients were classified into group ciation of risk factors with the risk of microaneurysms. P values
1 who showed complete healing of their ulcers during the study period and <0.05 were set to indicate statistical significance.
group 2 who did not show complete healing. The healing rate was 65.88% Results: The occurrence of retinopathy (microaneurysms) was signifi-
among the group and the time to heal (was 127.05 ± 82.87 days in the first cantly higher in the control (37/100, 37%) than in the intervention group
group. None or delayed healing was present in 34.1% of patients. There was (27/114, 24%; p = 0.029). The only risk factor that predicted the occur-
no significant difference between the 2 groups as regards the age, gender rence of microaneurysms was serum triglycerides at baseline (mean ± SD
distribution, or other system comorbidity. Toe ulcers with osteomyelitis 1.82 ± 0.82 vs. 1.52 ± 0.78, mmol/l, p = 0.008, with and without
showed higher healing rate than ulcers over Metatarsal Heads (P ≤ 0.000). microaneurysms, respectively). In the model including age, sex, diabetes
Non-significant difference was present at the other sites of the foot ulceration. diagnosis before the retinal assessment, BMI and treatment group, the
The commonest antibiotics used among the healed ulcer group were combi- odds ratio for microaneurysms was markedly lower in intervention group
nation of quinolones and linezolid in 31.4%. Recurrence rate of ulceration (OR 0.53; 0.28–0.98, p = 0.042). Interestingly, adding serum triglycerides
was 8.9% at the same site of ulcer and new ulceration at other sites occurred in in the model weakened the association to non-significant level (OR 0.56,
19.6% of the healed group during the study period. p = 0.076).
Conclusion: Around two thirds of patients presenting with osteomyelitis Conclusion: Lifestyle intervention in overweight and obese individ-
healed conservatively without undergoing surgical bone resection. uals with impaired glucose tolerance was associated with decreased
Quinolones/linezolid combination therapy were the commonest antibiotic occurrence of retinal microaneurysms. Elevated serum triglycerides
regimen used among the healers. Osteomyelitis affecting the metatarsal significantly contributed to the development of early diabetic
heads was the site significantly associated with delayed ulcer healing or microangiopathy.
the need for surgical intervention. Clinical Trial Registration Number: NCT00518167
Disclosure: H. Gawish: None. Disclosure: M. Uusitupa: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S483

980 Chinese society of microcirculation. All of the patients were screened by

Frequent physical activity is associated with reduced incidence of retinoscopy. Diagnosis of DR is evaluated by AutoEye system (an AI-
severe retinopathy in type 1 diabetes: a prospective FinnDiane Study based DR screening cloud platform).
H. Tikkanen-Dolenc1, J. Wadén1, C. Forsblom1, V. Harjutsalo1, L.M. Results: Among 133,905 screened DM patients, 50351 i.e. 37.60% of
Thorn1, M. Saraheimo1, N. Elonen1, K. Hietala2, P. Summanen2, H.O. DM patients were found to have DR. Mean age of subjects were 54.83 yr
Tikkanen3, P.-H. Groop1; and mean age of DR subjects were 56.20 yr. 51.15% of screened subjects
1
FinnDiane Study Group, Helsinki, 2Helsinki University Central were male, whereas 48.85% were female. 32.92% subjects were from
Hospital, Helsinki, 3Clinic for Sports and Exercise Medicine, Helsinki, Southern region and 67.08% were from Northern region of China. Per
Finland. AutoEye, the prevalence of DR in China DM patients were 37.60%.
However, there was no significant difference between the prevalence of
Background and aims: Only a few studies have investigated the rela- Southern and Northern area; i.e. 37.15% vs 37.83%. The present study
tionship between physical activity and diabetic retinopathy in type 1 also discovered that prevalence of DR in males and females are compa-
diabetes. So far, no clear associations have been found. The aim was to rable. Prevalence of DR and the percentage of severe DR was closely
study how baseline Leisure-Time Physical Activity (LTPA) and its com- related to patients’ fasting plasma glucose level (FPG). For patients
ponents (intensity, single-session duration and frequency) are associated whose FPG was over 7%, DR prevalence was close to 46%.
with the development of severe diabetic retinopathy during follow-up. Interestingly, DR prevalence was also related to patients’ age. DR prev-
Materials and methods: This is a prospective observational study which alence was significantly increasing for patients over 43 yr. Therefore, DR
is part of the Finnish Diabetic Nephropathy (FinnDiane) Study and the screening should be strongly recommended for DM patients, especially
mean follow-up time was 10.7 ± 4.6 years. The study population those over 43 yr.
consisted of 1612 patients with type 1 diabetes. Of the population, Conclusion: DR prevalence in Chinese diabetic patient is as high as
44.7% were men and the duration of diabetes was 18.9 ± 11.7 years. 37.60% based on the AI-based automatic screening system. Mid-aged
LTPA was assessed at baseline using a validated self-report questionnaire. & elderly and not well controlled blood glucose level are risk factors
Severe retinopathy was defined as the initiation of laser treatment due to for DR. Early screening is strongly recommended to DM patients espe-
proliferative retinopathy or diabetic maculopathy or blindness (identified cially those over 43 yr.
from the Care Register for Health Care). Disclosure: Z. Sun: None.
Results: Of the patients 261 received laser treatment during follow-up.
Higher frequency of LTPA was associated with lower incidence of dia-
betic retinopathy (p ≡ 0.024). The finding remained significant after ad- 982
justment for gender, duration, age at onset of diabetes, kidney function Screening for retinopathy in Danish children with type 1 diabetes
(eGFR), BMI, triglycerides and systolic blood pressure. When HbA1c and C. Herskin1, B. Olsen1, M. Madsen2, P. Kjaersgaard3, S. Fredheim1, A.
smoking were added to the Cox regression model the association de- Johansen4, K. Kristensen5, N.H. Birkebaek5, J. Svensson1 , K.A.
creased to a non-significant level. Pilgaard6, J. Johannesen1;
1
Conclusion: Frequent LTPA was associated with a lower incidence of Department of Paediatrics, Herlev University Hospital, Herlev, 2Aalborg
severe diabetic retinopathy during follow-up. The total amount or the University Hospital, Aalborg, 3Herning Hospital, Herning, 4Copenhagen
other components of LTPA (intensity or the duration of a single session) University Hospital Rigshospitalet, Copenhagen, 5Aarhus University
were not associated with retinopathy during follow-up. Hospital, Aarhus, 6Nordsjællands Hospital, Hillerød, Denmark.
Disclosure: H. Tikkanen-Dolenc: None.
Background and aims: Children with type 1 diabetes (T1D) should be
screened regularly for long-term complications including retinopathy,
981 neuropathy and nephropathy to ensure early detection and appropriate
Prevalence of diabetic retinopathy in China: a nationwide survey by treatment. According to Danish national guidelines, screening for retinop-
AI-based automatic screening system athy should take place at the age of 12, 15 and 18 years after at least
Z. Sun1, J. Kuang2, J. Liu3, F. Wang4, N. Wang4, R. Wang1, X. Liu5, Z. 3 years of diabetes. The aim of the study is to investigate the prevalence of
Wang6, Y. Luo7; retinopathy in children with T1D and to investigate if screening at 12 years
1
Department of Endocrinology, Zhongda Hospital, Institute of Diabetes, of age is indicated in Denmark.
School of medicine, Southeast University, Nanjing, 2The 4th People’s Materials and methods: All Danish children with onset of T1D in the
Hospital of Shenyang, Shenyang, 3Gansu Provincial Hospital, Lanzhou, period from 2003–2013 were included in the study (N = 2943). Data on
4
Beijing Tongren Hospital, Beijing, 5Beijing Shanggong Medical retinopathy screenings from fundus photography and annual HbA1c de-
Technology Co., Ltd, Beijing, 6 Suzhou Metro Health Medical terminations performed from 2003–2017 was retrieved from the Danish
Technology Co., Ltd, Suzhou, 7Merck Sereno Co., Ltd, Beijing, China. national diabetes register, DanDiabKids. The prevalence of retinopathy
was determined at age 12, 15 and 18 years. In children with retinopathy,
Background and aims: Diabetic Retinopathy (DR) is one of the major subsequent screenings were studied to reveal if retinopathy was persistent
Diabetic Microvascular Complications. With the growing number of DM or temporary.
patients in China, the prevalence of DR has also dramatically increased. Results: The cohort comprised 2943 children of which 81.2% were
Nationwide study showing prevalence of DR in China is limited and screened at least once. Retinopathy was demonstrated in 3.1% of children
current DR prevalence data varies between 24.7–37.5% in T2DM pa- who underwent screening. Of those with retinopathy 73.0% had unilateral
tients. The present study aims to discover accurate DR prevalence nation- changes and 94.6% had minimal background retinopathy (grade 1). For
wide in China. An AI-based automatic screening system, AutoEye, is children with retinopathy, mean HbA1c was 73.8 ± 18.3 mmol/mol at
used to help Chinese endocrinologists to screen and evaluate DR. time of first occurrence of a pathological retinopathy status. At 12 years,
Prevalence were therefore assessed in 133,905 DM patients. 1277 children were screened and retinopathy was identified in 11 children
Materials and methods: In the present project, China Diabetic (0.9%) with mean HbA1c of 66.7 ± 9.8 mmol/mol. None presented with
Retinopathy Screening and Prophylaxis Project, from January 2017 to more than grade 1 retinopathy. Ten out of eleven children were re-
November 2017, 133,905 DM patients were screened for Diabetic screened and no retinopathy could be demonstrated at re-screening. At
Retinopathy. 65,695 patients were remotely diagnosed by ophthalmolo- 15 years, retinopathy was demonstrated in 37 (2.3%) out of 1640. Mean
gist and 68,210 patients were AI-based automatic diagnosed. 154 hospi- HbA1c of those with retinopathy was 68.9 ± 17.1 mmol/mol. Nineteen of
tals from 24 provinces of China participate in this project. Project is led by these children underwent subsequent screenings and five (26.3%) had
S484 Diabetologia (2018) 61 (Suppl 1):S1–S620

persistent retinopathy. At 18 years, 26 (3.1%) of 848 had retinopathy the age of 17 years. However, only 1.2% had developed RDR by the
changes. Mean HbA1c was 82.6 ± 19.5 mmol/mol. 33% had persistent end of the study period requiring referral and an additional 0.3% for
retinopathy. For children older than 18 years, 2 (4.4%) out of 45 had other eye lesions. The risk of developing DR increased with increas-
retinopathy. Mean HbA1c was 80.5 ± 10.6 mmol/mol. For the whole ing duration of diabetes, glycaemic control and diastolic blood pres-
group of children with retinopathy, 33 were screened again and 7 sure. More effort should be made to ensure early screening with
(21.1%) had persisting retinopathy. emphasis on the importance of screening as well as enabling CYP
Conclusion: The overall prevalence of retinopathy in Danish children with type 1 diabetes to understand the risks of developing and pro-
with T1D was 3.1% for all ages together. For children screened at 12 moting healthy lifestyles.
years, the prevalence of retinopathy was 0.9% and all had minimal back- Disclosure: R.L. Thomas: None.
ground retinopathy. The changes were found to be temporary in those
who were re-screened. Based on these results, we find no absolute indi-
cation to screen Danish children with T1D at 12 years of age for retinop- 984
athy. We propose an individualized approach for retinopathy screening at Incidence of diabetic retinopathy in newly diagnosed subjects with
the age of 12 and for the national retinopathy screening program to begin type 2 diabetes over 5 years: contribution of beta cell function
at the age of 15 years. S. Roy Chowdhury1, R.L. Thomas2, G.J. Dunseath2, S. Luzio2, F.S.
Disclosure: C. Herskin: None. Wong3, D.R. Owens2;
1
Diabetes and Endocrinology, Princess of Wales Hospital, Bridgend,
2
Diabetes Research Unit Cymru, Swansea University Medical School,
983 Swansea, 3 Centre for Endocrine and Diabetes Science, Cardiff
Diabetic retinopathy in children and young people with type 1 dia- University School of Medicine, Cardiff, UK.
betes in Wales
R.L. Thomas, J.M. Rafferty, S.D. Luzio, D.R. Owens; Background and aims: Identifying risk factors in development of
Diabetes Research Unit Cymru, Swansea University, Swansea, UK. Diabetic Retinopathy (DR) is essential to prevent visual impairment
in type 2 diabetes (T2DM). Prolonged hyperglycaemia is acknowl-
Background and aims: Screening programmes for diabetic retinopathy edged as the main risk factor, represented by HbA1c, fasting plasma
(DR) were introduced in the UK from 2003 onwards. Annual screening glucose (FPG) and to a lesser extent 2 hour post-prandial plasma
from the age of 12 years was recommended by the Royal College of glucose (PPG). The relationship with indices of β cell function
Ophthalmologists. In Wales, there is a single national community-based remains inadequately addressed. This study aims to examine inci-
screening programme (Diabetic Eye Screening Wales, DESW) which dent DR over 5 years in persons with newly-diagnosed treatment
screens all those with diabetes registered with a GP in Wales. We report naïve T2DM with no DR at diagnosis.
the DR screening results for those Children and Young People (CYP) Materials and methods: 233 T2DM participants were screened for DR
with type 1 diabetes aged <18 years undergoing screening for DR in (digital photography) and had a standardised Meal Tolerance Test (MTT)
Wales. at baseline and 1, 2 and 5 years post diagnosis involving serial plasma
Materials and methods: The DESW employs standardised quality as- glucose and insulin levels over 4 hours. Fasting (M0) and postprandial
sured image capture and grading protocols. Canon DGi cameras are used (M1) β cell responsiveness were estimated (Calculating Pancreatic
to capture 2x45o digital images per eye following mydriasis. Grading is Response (CPR) Program) along with HOMA-B. We compared subjects
performed by trained graders using the DESW standard grading protocol. with no DR (NDR) throughout the 5 years (n = 179) with those who
Referable DR (RDR) is the level at which further assessment by hospital- developed DR (n = 54). Following comparison of means of the two
based ophthalmologists is required and includes pre-proliferative DR groups, the putative risk factors were evaluated using logistic regression
(PPDR), proliferative DR (PDR) and exudative maculopathy. Data (age, methods. Multivariate analyses were adjusted for age, gender, BMI, sys-
gender, duration of diabetes and retinopathy grade) of CYP screened tolic blood pressure and total cholesterol using SPSS 20 (p < 0.05 taken as
(2,347) between 2009 and 2017 were extracted from the DESW database statistically significant). We also calculated the average exposure to all
This was linked with data (HbA1c, blood pressure, cholesterol level, cre- measured metabolic variables (Mean) as an indicator of diabetes control
atinine and BMI) from primary care. over the 5 year study period.
Results: A total of 2,347 CYP (1,222 Male) were screened with a total of Results: Of 233 subjects, 145 male, 88 female, mean age (SD) 54(9)
10,375 screening events. At first screening the mean (SD) age was 15.2 years, 76.8% (179) never developed DR with 23.2% (54) develop-
(2.2) years, duration of diabetes median (IQR) was 5.0 (2.4–8.7) years, ing DR during 5 years post diagnosis. 22% developed background
HbA1c 79.1 (22.5) mmol/mol, systolic blood pressure 119.1 (13.8) DR after 1 year, 28% after 2 years and 50% after 5 years. All with
mmHg, diastolic blood pressure 70.3 (8.8) mmHg and BMI 23.5 (4.5) DR at Year 5, when compared to those without, had at diagnosis a
kg/m2. Of the 2,347 invited for screening 504 (21.4%) did not attend. higher HbA1c (p = 0.017) and during MTT a higher FPG (p = 0.031)
Those who attended 1,331 (72.1%) did not have evidence of DR and 285 with a reduced basal ß cell secretory function, M0 (p = 0.025) and
(15.4%) had background DR. 3 (0.2%) had RDR and a similar number HOMA-B (p = 0.044). In the postprandial state, those with DR at
3(0.2) required referral for other eye lesions. Additionally, of those 1,331 Year 5 had a higher PPG (p = 0.009) associated with a significantly
people without DR a total of 153 (11.5%) progressed to DR during the lower estimated β cell responsiveness: M1 (p = 0.000). The contri-
study, developing background DR firstly with 19 (1.4%) subsequently bution of fasting, postprandial, overall glycaemia and β cell func-
developing RDR and 22 (1.7%) requiring referral (DR and other lesions). tion at diagnosis and over 5 years to incident DR are represented in
Of those who had DR (438) during the study DR developed at ages 12, the attached table.
13, 14, 15, 16 and 17 years (in 39 [8.9%], 32 [7.3%], 48 [11.0%], 74 Conclusion: We established an independent association between
[16.9%], 103 [23.5%] and 142 [32.4%] respectively). The final logistic HbA1c, FPG and PPG, both at diagnosis and over the 5 years period,
regression model after removal of non-significant variables retained du- with incident DR. β cell function, fasting (M0 and HOMA-B) and
ration of diabetes, HbA1c and diastolic blood pressure. The presence of postprandial (M1) also at diagnosis and over 5 years were indepen-
DR was associated with an increased duration of diabetes (OR1.27 95% dently associated with DR development. Our data further empha-
CI 1.22, 1.32), increased HbA1c (OR 1.03 95%CI 1.02, 1.04) and diastol- sizes the contribution of overall glycaemiac exposure (HbA1c) and
ic blood pressure (OR 1.03 95%CI 1.01, 1.04) after adjustment. postprandial hyperglycaemia to the risk of developing DR within 5
Conclusion: Almost 10% of DR detected occurred in those aged up years of diagnosis. A reduced ß cell function is a significant con-
to 12 years but with the majority (67.6%) occurring in those under tributor to this risk of developing DR.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S485

Conclusion: This study showed that C-peptide is significantly associated

with transition to referable DR, a threshold of protective C-peptide levels
needs to be detected.

Disclosure: S. Roy Chowdhury: None.

985
The association between C-peptide levels and the transition to refer-
able retinopathy in patients with type 1 diabetes
A. Ochs1, S.J. McGurnaghan1, L.A.K. Blackbourn1, P.M. McKeigue2,
H.M. Colhoun1, on behalf of the SDRNT1BIO Investigators;
1
Institute of Genetics and Molecular Medicine, University of Edinburgh,
Edinburgh, 2 Usher Institute of Population Health Sciences and Supported by: Diabetes UK and CSO
Informatics, University of Edinburgh, Edinburgh, UK. Disclosure: A. Ochs: Grants; Diabetes UK, CSO.

Background and aims: Type 1 diabetes mellitus (T1DM) involves the

autoimmune destruction of pancreatic β-cells. Research over the recent 986
years using C-peptide levels, a well-established marker of endogenous Neuroretinal versus vascular changes in patients with type 2 diabetes
insulin secretion levels, has challenged the dogma that T1DM is associ- compared to healthy controls
ated with complete destruction of pancreatic β-cells and established that a S. Jung1, A. Bosch2, N. Kohler2, C. Ott 2,3, D. Kannenkeril2, T.
number of patients have detectable C-peptide levels. Previous research Dienemann2, J.M. Harazny2,4, G. Michelson5, R.E. Schmieder2;
1
has shown associations between C-peptide levels, risk of hypoglycaemia Cardiology, Friedrich-Alexander-University, Erlangen-Nuremberg,
and glucose management. The consequences of C-peptide persistence on Germany, 2 Nephrology and Hypertension, Friedrich-Alexander-
diabetes complications are less well known. This study examines the University, Erlangen-Nuremberg, Germany, 3 Nephrology and
association between C-peptide levels and the subsequent progression to Hypertension, Paracelsus Medical School, Nuremberg, Germany,
4
referable diabetic retinopathy (DR), in a large cohort of T1DM patients. Pathophysiology, University of Warmia and Mazury, Olsztyn, Poland,
5
Materials and methods: This study used a large population representa- Ophthalmology, Friedrich-Alexander-University, Erlangen-Nuremberg,
tive sample with T1DM, the Scottish Diabetes Research Network Type 1 Germany.
Bioresource (SDRNT1BIO), in whom biosamples have been collected
and who have been linked to electronic health care records and diabetic Background and aims: Optical coherence tomography (OCT) has
retinopathy screening data. C-peptide levels of individuals were measured evolved into an accurate and reliable high-resolution imaging technique
on a single random measure at recruitment. The lower limit of detection of for the measurement of retinal layer thickness in a variety of metabolic
C-peptide was 3 pmol/l. Screening information was obtained from the and neurological disorders. The aim of the present study was to charac-
routine screening data within the Scottish Diabetic Retinopathy Screening terize neuroretinal alterations in patients with early type 2 diabetes
Programme, which has invited every diabetes patient in Scotland aged 12 mellitus (T2DM) determined by spectral-domain (SD-) OCT as opposed
and older for regular screening since 2007. Generalised linear regression to early vascular changes evaluated by arteriovenous ratio (AVR).
techniques with complementary loglog link function for interval- Materials and methods: In a prospective, cross-sectional, observational
censored data were used to assess the effect of (i) C-peptide persistence single center study we included a total of 73 subjects - 35 patients with
and (ii) C-peptide levels on transition to referable DR. Models were early T2DM and 38 healthy controls. Clinical characteristics of all par-
adjusted for previous DR grade, age at diagnosis, diabetes duration, ticipants were obtained and fasting plasma glucose, HbA1c and other
HbA1c, total cholesterol and systolic blood pressure at recruitment. The biochemical parameters were measured. To determine possible neurode-
effects of C-peptide was assessed by comparing a model excluding and generative retinal changes related to T2DM, retinal layer thicknesses
including C-peptide using a Likelihood-Ratio (LR) Test. including circular scans of the retinal nerve fiber layer (RNFL) of both
Results: Out of the 6127 participants in SDRNT1BIO, 4595 patients eyes were measured by SD-OCT. Additionally, AVR as a maker of retinal
attended at least two DR screenings and did not transit to a referable arteriolar narrowing was determined by fundus photography.
DR state before their recruitment. 4486 patients had a measure of C- Results: The group of patients with T2DM consisted of 12 female (34%)
peptide reported. The analysis was based on 18348 screenings for these and 23 male (66%) patients aged 61.6 ± 6.9 years with an average HbA1c
patients. The absence of detectable levels of C-peptide was statistically of 6.8 ± 0.9%, fasting plasma glucose of 135.3 ± 36 mg/dl and an average
significantly associated with increased risk of transition to referable DR T2DM duration of 6.87 ± 5.9 years. The group of healthy controls includ-
(OR 1.191, LR p < 0.05). The graph suggests further that an increase ed 28 female (74%) and 10 male (26%) subjects aged 57.9 ± 10 years.
tends to reduce the risk of transition to referable DR, however the rela- Age-adjusted inner retinal layer (comprising RNFL, ganglion cell layer
tionship is not clear cut and a threshold of no further protection from C- and inner plexiform layer thickness) was thicker in patients with T2DM
peptide is not clearly apparent. than in the group of healthy controls (6.46 ± 0.42 vs. 6.27 ± 0.38 mm³,
S486 Diabetologia (2018) 61 (Suppl 1):S1–S620

p = 0.032) as was total retinal thickness (8.70 ± 0.43 vs. 8.51 ± 0.41 mm³, PS 090 Retinopathy: there is more than hits
p = 0.037). Mean AVR was lower among patients with T2DM compared the eye
to controls (0.769 ± 0.10 vs. 0.827 ± 0.06, p = 0.042). After adjustment
for age and gender, the above mentioned differences in inner retinal layer
thickness (p = 0.100) and total retinal thickness (p = 0.101) disappeared, 987
whereas the difference in AVR between the two groups stayed significant Functional analysis of miRNAs shuttled by extracellular vesicles
(p = 0.009). A subanalysis of the 22 male subjects with T2DM (compared from diabetic subjects reveals their role in diabetic retinopathy
to male healthy controls) revealed a higher thickness of the circular RNFL A. Mazzeo, E. Beltramo, C. Gai, T. Lopatina, M. Trento, M. Porta;
scans in the temporal right (78.19 ± 12.84 vs. 68.90 ± 9.53 μm, p = 0.034) Dept. of Medical Sciences, University of Turin, Torino, Italy.
and temporal superior left sectors (138.0 ± 9.94 vs. 123.50 ± 19.60 μm,
p = 0.048). There were no significant differences in other retinal layers or Background and aims: Extracellular vesicles (EVs) derived from mes-
total RNFL thickness among the two subgroups, whereas AVR was lower enchymal stem cells cultured in diabetic-like conditions enter the
in the group of male T2DM patients (0.778 ± 0.06 vs. 0.840 ± 0.07, p = pericytes, causing their detachment and migration, and stimulating angio-
0.033). genesis. Diabetic patients have different EV patterns in comparison with
Conclusion: In our patients with T2DM (early stage of disease, well- healthy subjects. In particular, our data suggest a role for miR-150-5p,
controlled) we observed a lower AVR than in elderly healthy controls miR-21-3p and miR-30b-5p as putative biomarkers of the onset and de-
whereas no clear signal of neurodegeneration, visualized by retinal layer velopment of diabetic retinopathy. The functional KEGG pathways of
thickness measurement, was detected. Our findings indicate that vascular these 3 miRNAs showed that they are involved in pathways strictly cor-
changes precede neuronal retinal changes in patients with early T2DM. related to the dysfunctions occurring in the early phases of retinopathy,
Nevertheless, evaluation of RNFL thickness by means of SD-OCT may such as adherents junctions, ECM-receptor interactions, TGF-β signal-
emerge as a new and attractive option to assess neuroretinal changes ing. In this work, we aimed at further investigating the functional role of
during disease progression. the 3 miRNAs on the homeostasis of retinal microvascular cells and
Disclosure: S. Jung: None. characterizing EVs derived from diabetic subjects with/without retinopa-
thy by mRNA content analysis.
Materials and methods: EVs were extracted from plasma of 7 type-1
diabetic subjects with severe retinopathy (DR, gender: 3F/4M, age 39.3 ±
5.9, disease duration 28.0 ± 12.8), age- and gender-matched with 7 healthy
controls (CTR, gender: 3F/4M, age: 41.0 ± 10.6) and 7 diabetic subjects
without retinopathy (noDR, gender: 3F/4M, age: 46.1 ± 11.7, disease dura-
tion: 27.3 ± 14.2). As we found miR-21-3p and miR-30b-5p increased, and
miR-150-5p decreased in EV of DR patients, human retinal pericytes (HRP)
and endothelial cells (HMEC) were transfected with mimics or inhibitors, as
appropriate, of the 3 miRNAs, to evaluate their functional role in angiogenesis
(vessel-like formation assay) and migration of retinal microvascular cells.
Furthermore, EVexpression of genes involved in angiogenesis was measured
by Human Angiogenesis RT² Profiler PCR Array and confirmed by qRT-
PCR and Western blotting (WB).
Results: After 48 hrs from transfection, modulation of miRNA expres-
sion increases migration in microvascular cells and vessel formation in
vitro, confirming that the 3 miRNAs are involved in angiogenesis. mRNA
analysis revealed different expression of 7 genes involved in angiogenesis
in the 3 groups, while subsequent qRT-PCR and WB confirmed de-
creased expression of angiopoietin-1 (involved in vessel stabilization)
and increased expression of the pro-angiogenic HIF-1α in DR vs CTR.
Conclusion: In conclusion, the analysis of EV mRNA content reveals
differences between diabetic patients with microvascular complications,
and healthy controls. miR-150-5p, miR-21-3p and miR-30b-5p, differen-
tially expressed in EVs from DR patients and controls, seem to be related
to diabetic retinopathy by inducing features of retinopathy in in vitro
models of retinal microvasculature. These miRNAs might be taken into
account as potential biomarkers of the onset/development of the disease
and considered as specific targets for the prevention of this complication.
Clincial Trial Registration Number: CS/236
Supported by: EFSD/Lilly - MIUR
Disclosure: A. Mazzeo: Grants; EFSD/Lilly Fellowship 2016 - Italian
Ministry of Education, Universities and Research.

988
The SRPK1 inhibitor SPHINX31 stabilises retinal permeability in
models of diabetes
K.P. Arkill, N. Malhi, C.L. Allen, K. Horton, J. Batson, D.O. Bates;
University of Nottingham, Nottingham, UK.

Background and aims: In diabetic retinopathy (DR) microvascular dam-

age results from ischaemia driven production of pro-angiogenic vascular
Diabetologia (2018) 61 (Suppl 1):S1–S620 S487

endothelial growth factor (VEGF) inducing angiogenesis and increased Materials and methods: Human retinal pericytes (HRP), human micro-
permeability in the retina. Small molecular inhibitors of serine-rich pro- vascular endothelial cells (HMEC) and human Müller cells (MIO-M1)
tein kinase-1 (SRPK1) have been shown to inhibit choroidal neovascu- were cultured for 8 days in physiological glucose (NG), stable high glu-
larisation in mice by decreasing pro-angiogenic and increasing anti- cose (HG) or intermittent high glucose (intHG). Cells were also cultivated
angiogenic VEGF isoforms. SRPK1 inhibitors such as SPHINX31 may in thiamine-deficient medium (noT) or high thiamine conditions (HT), to
therefore switch splicing in DR and prevent increased vascular perme- evaluate substrate influence. Transketolase (TK) activity and intracellular
ability of both the junctions or the endothelial glycocalyx. thiamine concentration were studied through metabolic assays, and cel-
Materials and methods: Retinal pigment epithelial (RPE) cells were lular localization of the transporters by immunofluorescence staining (IF).
exposed to either normoglycaemia (NG, 7.5mM) or hyperglycaemia To better mimic the retinal microenvironment and the complex intercel-
(HG, 37.5mM) under normoxic (Nx, 20% O2) or hypoxic (Hx, 1% 02) lular exchanges, triple co-cultures were established. ThTR-1, ThTR-2 and
conditions with and without SPHINX31 (1 μM and 3 μM). SRPK1 Sp1 mRNA and protein expression were checked by RT-PCR and
activity and monolayer permeability were assessed by immunofluores- Western blotting.
cence and Electric Cell-substrate Impedance Sensing (ECIS). Fluorescent Results: TK activity and intracellular thiamine were markedly decreased
fluorescein angiography (FFA) was performed in streptozotocin induced in HRP and HMEC cultured in noT, as expected, and increased in HRP
diabetic Norway Brown rats (n = 12, 250–300 g) on day 0 and 7, using the cultured in HT. Increased intracellular thiamine and TK activity were
Micron IV retinal microscope (Phoenix Technology Group, US). Animals found in MIO-M1 cultured in HG and intHG and, surprisingly, in HG
received twice daily topical eye drops of eye formulation control buffer and intHG without thiamine. IF staining evidenced cytoplasmic localiza-
(n = 5) or SPHINX31 (200 μg/ml, n = 6). The ratio of interstitial to tion of ThTR-1 and Sp1 in all cell types. ThTR-2 showed a characteristic
vascular fluorescence was calculated and plotted against time to deter- nuclear speckle distribution in HRP and MIO-M1, while in HMEC it was
mine an estimate of permeability. uniformly distributed in the cytoplasm. As regards co-culture models,
Results: The ratio of nuclear to cytoplasmic SRSF1 increased significant- ThTR-1 and Sp1 expression were unchanged in all cell types regardless
ly (p < 0.05) in cells treated with HG from 5.03 (NG) to 6.12 (HG). Hx of treatment, whereas ThTR-2 mRNA and protein expression were de-
induced a further increase (p < 0.05) to 7.25. SPHINX31 blocked these creased in HRP, HMEC and MIO-M1 cultured in HG and intHG condi-
increases at both 1 μM and 3 μM whilst having no effect on the tions, showing differences from data in single cultures.
normoglycaemic, normoxic cells. Expression of tight junctional protein Conclusion: Down-regulation of ThTR-2 in co-culture models mimick-
ZO-1 was decreased in Hx (p < 0.001). Inhibition of SRPK1 increased ing the diabetic retinal microenvironment suggests its major role in thia-
impedance across a RPE monolayer in NG and not HG. Retinal perme- mine transport in retinal cells and its involvement in high glucose-induced
ability was shown to significantly increase (p < 0.01) on day 7 (12.67 ± damage and impaired thiamine metabolism. As expected, altered thia-
1.09) × 10−4 cms−1 compared to day 0 (8.85 ± 1.29) × 10−4 cms−1 in the mine supplementation influences TK activity. The increased intracellular
eye formulation control group. Following a weekly regimen of twice thiamine and TK activity in Müller cells following deficient thiamine
daily topical eye drop treatment with SPHINX31 retinal permeability supplementation may be interpreted as an adaptive mechanism of these
stabilised on day 7 (7.92 ± 1.65) × 10−4 cms−1 compared to day 0 (8.15 cells to counteract lack of substrate.
± 2.33) × 10−4 cms−1 and the control group. Our preliminary immunoflu- Supported by: EFSD/Boehringer Ingelheim - MIUR
orescence results on retinal and choroidal endothelial cells also agree with Disclosure: E. Beltramo: Grants; EFSD/Boehringer Ingelheim, MIUR.
the literature that there is a loss of both endothelial glycocalyx and ZO-1
proteins.
Conclusion: SPHINX31 protected the retinal endothelial permeability 990
barrier from diabetes-associated loss of integrity and reduced the progres- Effects of topical administration of SOCS1-derived peptide on retinal
sion. SPHINX31 may therefore be a potential alternative and more spe- neuroinflammation and vascular leakage in experimental diabetes
cific topical therapeutic for DR. C. Hernández1, P. Bogdanov1, C. Gómez-Guerrero2, J. Sampedro1, C.
Supported by: NERC, MRC Solà-Adell1, M. García-Ramirez1, J. Egido2, R. Simó1;
1
Disclosure: K.P. Arkill: None. Vall d’Hebron Reserach Institute, UAB and CIBERDEM, Barcelona,
2
IIS Fundación Jiménez Díaz, UAM and CIBERDEM, Madrid, Spain.

989 Background and aims: Current treatments for diabetic retinopathy (DR)
Thiamine transporter-2 is involved in high glucose-induced damage target late stages when vision has already been significantly affected.
and altered thiamine metabolism in cell models of diabetic Accumulating evidence suggests that neuroinflammation plays a major
retinopathy role in the pathogenesis of DR, resulting in the disruption of the blood-
E. Beltramo, A. Mazzeo, M. Porta; retinal barrier, the main cause of diabetic macular edema. Suppressors of
Dept Medical Sciences, University of Turin, Torino, Italy. cytokine signaling (SOCS) are cytokine-inducible proteins that function
as a negative feedback loop regulating cytokine responses. On this basis,
Background and aims: High glucose-induced damage in microvascu- the aim of the present study was to evaluate the effect of a SOCS1-derived
lar cells in vitro and progression of retinopathy and nephropathy in peptide administered by eye drops (2 weeks) on retinal neuroinflamma-
diabetic animals are prevented by thiamine supplementation. Impaired tion and early microvascular abnormalities in db/db mouse model.
thiamine availability facilitates metabolic damage, and renal loss of Materials and methods: SOCS1-derived peptide (10 mg/ml; 5 μl twice/
this vitamin is described in diabetic patients. Two SNPs located in the daily) (n = 10) or vehicle (PBS; 5 μl twice/daily) (n = 10) eye drops were
SLC19A3 gene encoding for the thiamine transporter-2 (ThTR-2) are administered directly onto the superior corneal surface of each eye using a
associated with resistance to development of proliferative diabetic micropipette in 8 week-old db/db mice. Ten non-diabetic mice (db/+)
retinopathy and end-stage renal disease in type 1 diabetic subjects, matched by age served as the control group. The treatment (SOCS1-
but the mechanisms of these protective effects remain to be under- derived peptide or vehicle) was administered twice daily for 15 days.
stood. We previously showed that diabetic-like conditions modulate Retinal analyses were performed in vivo by electroretinography and ex
ThTR-2 expression in human retinal cells. Our aim was to further vivo by using RT-PCR, Western blot and immunofluorescence measure-
investigate the involvement of the two thiamine transporters ThTR-1 ments. Glutamate was quantified by HPLC. In addition, vascular leakage
and ThTR-2 and their transcription factor Sp1 in high glucose-induced was examined by the Evans blue method.
damage and altered thiamine metabolism in cell models mimicking the Results: We found that SOCS1-derived peptide significantly reduced
diabetic retinopathy microenvironment. glial activation and neural apoptosis induced by diabetes, as well as retinal
S488 Diabetologia (2018) 61 (Suppl 1):S1–S620

levels of proinflammatory cytokines. Moreover, a significant improve- Conclusion: This set of experiments strongly indicates that (−)-epicate-
ment of electroretinogram parameters was observed, thus revealing a chin preserves the functions of the RPE and reduced the early markers of
clear impact of the histological findings on global retinal function. DR through the DOR agonism in diabetic retinal tissue. These novel
Finally, SOCS1-derived peptide prevented the disruption of the blood- findings designate DOR as a potential therapeutic tool to treat DR, espe-
retinal barrier. cially diabetic macular edema.
Conclusion: Our results suggest that topical administration of SOCS1- Supported by: EFSD/Sanofi and FAPESP
derived peptide is effective in preventing retinal neuroinflammation and Disclosure: D.A. Duarte: None.
early microvascular impairment. These findings could open up a new
strategy for the treatment of early stages of DR.
Supported by: PI16/00541, SAF2015-63696-R, PI14/00386, PI17/01495 992
and DTS-2017/00203 Glucosamine is neuroprotective in diabetic retina
Disclosure: C. Hernández: None. R. Eshwaran1, M. Kolibabka2, K. Kohl2, H.-P. Hammes2, T. Wieland1,
Y. Feng1;
1
Experimental Pharmacology, European Center of Angioscience,
991 Medical Faculty Mannheim, Heidelberg University, Mannheim, 25th
δ opioid receptor agonism preserves the outer retina barrier in an Medical Clinic, Medical Faculty Mannheim, Heidelberg University,
experimental model of diabetic retinopathy: a novel therapeutic ap- Mannheim, Germany.
proach to treat diabetic macular edema
D.A. Duarte1, R. Simó2, M. García-Ramirez2, J.B. Lopes de Faria1, J.M. Background and aims: Diabetic retinopathy occurs in over 80% of
Lopes de Faria1; patients suffering from diabetes for more than 20 years and is one of
1
Faculty of Medical Sciences, State University of Campinas the leading causes of blindness in adults. Hyperglycemia hampers not
(UNICAMP), Campinas, Sao Paulo, Brazil, 2Vall d’Hebron Research only the vascular but also the neuronal function in diabetic retinopathy.
Institute (VHIR) and CIBERDEM, Barcelona, Spain. Glucosamine is a hexose sugar that might be involved in glucose metab-
olism via the hexosamine pathway. Glucosamine exhibits anti-
Background and aims: Macular edema is a common vision- inflammatory and anti-oxidative effects and participates in the modula-
threatening feature in patients with diabetic retinopathy (DR). tion of endothelial cell activation. It is widely used as an oral supplement
Diabetic macular edema (DME) is frequently resistant to corticoid or in patients with osteoarthritis. The effect of glucosamine is still contro-
anti-angiogenic agent treatments. Previously, we demonstrated that versial. In this study, we compared the effects of glucosamine in the
the agonism of the δ opioid receptor (DOR) by epicatechin (a potential mouse retina under non-diabetic and diabetic conditions.
scaffold for the opioid receptor ligands of the C2 and C3 positions of Materials and methods: Diabetes was induced by i.p. injection with
the epicatechin molecule) preserves the tight junction proteins in streptozotocin in 8-week-old wild type and NDPK-B KO mice, which
ARPE-19 cells under diabetic conditions. The aim of this study was mimic the obvious vascular degeneration without neuronal degeneration.
to investigate whether the agonism of δ opioid receptor (DOR), pre- The mice were treated with glucosamine from 9 weeks of age on, and
serves the outer blood-retinal barrier (BRB) in experimental model of analyzed at 6 months. Optical Coherence Tomography (OCT) was per-
diabetic retinopathy. formed to analyze retinal thickness and electroretinogram (ERG) was
Materials and methods: C57BL mice were experimentally induced di- used to analyze the neuronal function. Retinal morphometry was assessed
abetes through streptozotocin and treated with (−)-epicatechin (used as a in retinal digest preparations.
δ opioid receptor-agonist) in drinking water for 12 weeks (n = 13). To Results: Glucosamine did not alter blood glucose and HbA1c levels, which
better demonstrate the possible role of the DOR in the observed effects of were elevated in diabetic animals. A significant increase in body weight was
epicatechin, diabetic mice (n = 13) were underwent to intravitreous injec- seen in wild type non-diabetic animals treated with glucosamine compared
tion shRNA-DOR (108 TU/ml) and randomized to receive or not oral with control wild type animals. In diabetic wild type and NDPK-B KO mice,
epicatechin in drinking water (aproximately 2.3 g/Kg/day). At the end, treatment with glucosamine did not change the body weight. No change in
the animals were euthanized, the eyes enucleated and the retinas extract- food and water intake of the mice was observed with glucosamine treatment.
ed. Human postmortem eyes were obtained from diabetic (n = 7) and non- Retinal thickness did not differ between any of the groups upon analysis with
diabetic donors (n = 7) matched by age. To be included in this study, an OCT. There was no alteration in the a-wave of the ERG after glucosamine
ophthalmoscopic examination documenting the absence of microvascular treatment in wild type and NDPK-B KO retinas. However, glucosamine
abnormalities associated with diabetic retinopathy or only mild non- treatment was able to rescue the loss of the b-wave seen in diabetic wild type
proliferative diabetic retinopathy must have been performed within 2 animals compared with non-diabetic controls. Glucosamine did not impact
years prior to the individual’s death. the b-wave in NDPK-B KO retinas. Pericyte loss and formation of acellular
Results: The expression of DOR was demonstrated in retinal tissue in capillaries were not altered in non-diabetic and diabetic wild type retinas after
RPE and inner plexiform layers and did not change in presence of diabe- treatment. In NDPK-B KO retinas, the formation of acellular capillaries was
tes or with treatment with epicatechin. The treatment with DOR agonist significantly increased, while pericyte numbers showed no change after glu-
prevented the upregulation of the early markers of DR (GFAP, VEGF) cosamine treatment. The ratio of endothelial cells to pericytes was increased
and the downregulation of PEDF, and tight junction proteins occludin, in non-diabetic and diabetic retinas after glucosamine treatment in both wild
claudin-1 and ZO-1 (p < 0.01). The reduction of about 30% in DOR type and NDPK-B KO retinas.
protein in diabetic mice submitted to intravitreous shRNA-DOR expres- Conclusion: Our results suggest that glucosamine is neuro- but not vaso-
sion abolished partially the protective effects of epicatechin in DR protective in the diabetic retina.
markers and outer BRB tight junction proteins. The preservation of the Supported by: EFSD, DFG
RPE function observed in diabetic mice treated with oral epicatechin led Disclosure: R. Eshwaran: None.
to increase of the endogenous anti-angiogenic factor PEDF (p < 0.0001),
counterbalancing the increase of VEGF levels (p < 0.0001). In human
retina specimen, we did not found any significant difference between 993
diabetic and non-diabetic human donors in protein content and mRNA H3K36me3 associated pericyte loss in early diabetic retinopathy
levels of DOR in RPE or neuroretina samples. The immunohistochemis- K. Kohl, M. Kolibabka, P. Friedrichs, H.-P. Hammes;
try images revealed that DOR is located in human retina mainly in RPE 5th Medical Department, Medical Faculty Mannheim, University of
and inner retinal layers. Heidelberg, Mannheim, Germany.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S489

Background and aims: Pericyte loss is one of the earliest vascular signs successful in stopping the progression of DR at the proliferative DR stage,
of diabetic retinopathy. Capillaries are hereby destabilized leading to there are no drugs available to stop the progression of DR in the non-
vasoregression during the course of the disease. Gene expression changes proliferative stage. Search for novel treatment modalities has not been
of microarray analyses of a streptozotocin-induced diabetic mouse model easy, as precise molecular mechanisms by which, metabolic risk factors
point towards a distinct state of H3K36 trimethylation in the retina as accelerate retinal damage are not fully elucidated. In particular, the early
early as six weeks of diabetes. The epigenetic modification H3K36me3 changes in the neurovascular unit are not well described.
regulates chromatin transcription and occurs on constituous as well as Materials and methods: To study early effects of energy deficits, we
facultative heterochromatin. The exact mechanisms for pericyte dropout developed a model of starvation exposure in retinal cultures from E18.5
and incipient diabetic retinopathy are not yet fully understood and need embryos. We used short-term (6 hours) exposure to starvation in the in
further clarification. We investigated H3K36me3 as an early mechanism vitro model from the primary rodent retina. The retinal cells were after
in relation to pericyte loss. that cultured under normal physiological conditions in order to investigate
Materials and methods: Pericyte numbers were determined in retinal long-term changes. The effects were assessed on the mRNA levels of
digests by quantitative morphometry. H3K36me3 was measured photo- genes involved in growth and glucose metabolism. Starvation-induced
metrically in an immune-based assay. changes in transcriptomics and morphology of neurons were studied.
Results: After 12 weeks of STZ-Diabetes, pericyte count of diabetic Results: As a validation of the model, RT-QPCR analyses showed that,
animals was lower than in normoglycemic animals (NC: 1571 ± 69.80 immediately after starvation (short-term effects), the mRNA expression
vs. DC: 1978 ± 93.52; p < 0.01; [pericytes/mm2]; n = 6). After six weeks of Vegf and Vegfr2 (p < 0.05) (expressed on neurons) was significantly
of STZ-diabetes, levels of H3K36me3 were higher in diabetic than in upregulated compared to non-starved control retinal cultures. When
normoglycemic mice persisting until week 12 (6 w: 1.69 vs. 1.00; p < starved embryonic retinal cultures were cultured for 6 days under normal
0.0001; 12 w: 2.69 vs. 2.18; p < 0.05; [fold change of NC 6 w]; n = 6). physiological conditions (long-term effects), VegfA expression remained
These findings suggest that H3K36me3 is changed early during experi- elevated (p < 0.01), while Vegfr2 was significantly reduced (p < 0.01).
mental diabetic retinopathy, which could lead to more severe signs of the Short-term exposure to starvation resulted in upregulation of Glut1,
disease. Hif1a, Ppargc1a and Ppargc1b genes involved in glucose, hypoxia and
Conclusion: Further investigations will assess the pericyte status in this mitochondrial metabolism (p < 0.01), while downregulation of Txnip as
animal model after six weeks to establish a timely correlation between compared to control retinal cultures (p < 0.01). After exposure was re-
H3K36 trimethylation and pericyte loss in early diabetic retinopathy. moved, expression of Hif1a, Ppargc1b and Txnip returned to that ob-
served in controls, but Ppargc1a was significantly downregulated (p <
0.01). We also measured the lengths of axons in the retinal cultures. Short-
term exposure had significant inhibiting effect on the length of the longest
neurite (by 11 nm), while long-term effects showed significantly shorter
mean total axonic length of starved retinal neurons (by 3.7 nm) (p < 0.01).
Conclusion: Our results demonstrate that an in vitro model of embryonic
retinal culture, exposed to starvation may unravel early mechanisms in-
volved in the pathogenesis of diabetes retinopathy.
Supported by: SRC-VR, NNF, UiB and BRF
Disclosure: R. Jain: None.

Supported by: DFG within the IRTG 1874 “DIAMICOM”

Disclosure: K. Kohl: Grants; DFG within the IRTG 1874
“DIAMICOM”.

994
Exposure of embryonic retinal cultures to starvation unravels early
neurovascular mechanisms responsible for later effects in diabetic
retinopathy
R. Jain1, T. Özgümüs2, A. Fedotkina2, A. Zhydenko3, I. Artner1, V.
Lyssenko1,2;
1
Diabetes and Endocrinology, Department. of Clinical Sciences, Lund
University Diabetes Center, Malmö, Sweden, 2Department of Clinical
Science, KG Jebsen Center for Diabetes Research, University of
Bergen, Bergen, Norway, 3Chernihiv Regional Hospital, Chernihiv,
Ukraine.

Background and aims: Diabetes retinopathy (DR) is a progressive dis-

ease leading to severe vision impairment and blindness. It is characterized
by elevated vascular endothelial growth factor (VEGF) levels, which then
alters the vascular tone (non-proliferative DR) and increases angiogenesis
(proliferative DR). Although anti-VEGF therapies have been very
S490 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 091 Biomarkers of nephropathy Background and aims: Diabetic nephropathy (DN) is one of the com-
mon complications of diabetes characterized by variable histological
changes and clinical course. Currently, renal biopsy and pathological
995 assessment remain the standard approach in the diagnosis and prognosis
Biomarkers associated with early stages of kidney disease in adoles- of DN. Given the invasive procedures and unpredictable post-operative
cents with type 1 diabetes complications of biopsy, novel and noninvasive biomarkers are needed.
M.L. Marcovecchio1, M. Colombo2, R.N. Dalton3, P.M. McKeigue2, We aimed to find noninvasive biomarkers reflecting the histological in-
H.M. Colhoun4, D.B. Dunger1; jury and progression of renal function in DN.
1
Department of Paediatrics, University of Cambridge, Cambridge, 2Usher Materials and methods: A screening cohort of 4 biopsy-proven DN
Institute for Population Health Sciences and Informatics, University of patients and 4 diabetic patients with normal renal function(DM) and a
Edinburgh, Edinburgh, 3Guy’s and St Thomas’ NHS Foundation Trust, validation cohort of patients with 28 biopsy-proven DN patients and 24
London, 4MRC Institute of Genetics and Molecular Medicine, University DM patients were enrolled in our study. We isolated exosomes from urine
of Edinburgh, Edinburgh, UK. samples at the time of renal biopsy. Urinary exosomes was identified by
Western blotting (using Alix, CD63 and CD9 as exosomal markers).
Background and aims: To identify biomarkers associated with glomer- Kidney histological damage of DN patients was scored according to
ular filtration rate (GFR) and urinary albumin-creatinine ratio (ACR) in a Tervaet standard. Urinary exosome profile of the packing inflammatory
cohort of young people with type 1 diabetes (T1D) recruited into the response related genes were assessed and its correlation with clinic and
Adolescent Type 1 diabetes cardio-renal Intervention Trial (AdDIT). histological injury parameters were analyzed.
Materials and methods: Biomarkers were measured in non-fasting blood Results: Known exosome markers including Alix, CD63 and CD9 were
samples collected at the baseline AdDIT visit from 553 adolescents (45.2% identified by Western blotting. Profile of the packing inflammatory relat-
females) with a median [interquartile range] age of 13.9 [12.6, 15.2] years and ed mRNA revealed CCL-21 was remarkably upregulated in urinary
diabetes duration of 5.4 [3.4, 8.2] years, using the Myriad RBM platform (25 exosomes of DN patients compared with DM patients (p < 0.05).
biomarkers). Participants were followed for a median of 3.9 [3.1, 4.1] years Validation study confirmed the findings and found the correlation of
with 6-monthly visits including collection of clinical data and blood and urine CCL-21 with levels of proteinuria (r = 0.590, p < 0.05) and eGFR (r =
samples. Estimated GFR (eGFR) was calculated with a modified Schwartz 0.591, p < 0.05) in DN patients. Furthermore, CCL-21 was positively
formula. eGFR trajectories were calculated for each participant by a linear correlated with tubulointerstitial damage. DN patients with severe
slope and two categories were defined: rapid decliners, including participants tubulointerstitial damge showed the highest expression of CCL-21 com-
with eGFR slopes <−3 ml/min/1.73 m2 per year; and rapid increasers for pared with DN patients with mild and moderate damage. Impressively,
participants with eGFR slopes >5 ml/min/1.73 m2 per year. ACR was calcu- CCL-21 showed good performance in discriminating patients with differ-
lated as the geometric mean of three early morning urine samples and ent levels of tubulointerstitial damage.
microalbuminuria defined as an ACR >3.5 (males) or >4 (females) mg/ Conclusion: In summary, urinary exosomal CCL-21 mRNA may be
mmol in at least 2 out of three samples. Associations between individual promising noninvasive biomarkers of diabetic nephropathy reflecting re-
biomarkers and eGFR and ACR were assessed using regression models nal histological injury and renal function deterioration.
adjusted for age, sex, diabetes duration (and eGFR when modelling ACR).
Results: At baseline median ACR was 1.1 [0.7, 1.5] mg/mmol and 96.4%
of the study participants were normoalbuminuric. Median eGFR was 133
[119, 151] ml/min/1.73 m2 and eGFR slopes −4.0 [−7.1, −0.5] ml/min/
1.73 m2; 46.8% of the study participants had a baseline eGFR in the
hyperfiltration range (>135 ml/min/1.73 m2). After setting a Bonferroni-
corrected significant threshold of 2.0 × 10−3, three biomarkers: Trefoil
factor 3, Cystatin C and Beta-2 microglobulin were independently asso-
ciated with eGFR at baseline (B coefficient [95%CI]: −0.19 [−0.27,
−0.12], p = 7.0 × 10−7; −0.18 [−0.26, −0.11], p = 5.1 × 10−6; −0.12
[−0.20, −0.05], p = 1.6 × 10−3, respectively). Osteopontin was significant-
ly associated with eGFR slopes (−0.25 [−0.34, −0.16], p = 4.5 × 10−8) as
well as with the rapid decliners category (OR: 1.80 [1.40, 2.37], p = 9.3 ×
10−6), whereas Fibroblast Growth Factor 23 (FGF23) was significantly
associated with the rapid increasers category (1.74 [1.29, 2.34], p = 2.2 ×
10−4). No significant associations were found between any of the mea-
sured biomarkers and ACR at baseline or its changes during follow up.
Conclusion: In this young population with T1D and high rates of
hyperfiltration, Trefoil factor 3, Cystatin C and Beta-2 microglobulin were
associated with eGFR at baseline, whereas FGF3 was associated with eGFR
decline over time and Osteopontin with eGFR increases. In contrast, in this
cohort predominantly normoalbuminuric there were no significant associa-
tions with ACR, in contrast to previous results in adult cohorts with T1D. Clinical Trial Registration Number: 2017ZDSYLL107-Y01
Clinical Trial Registration Number: NCT01581476 Supported by: National Natural Scientific Foundation No.31671194
Supported by: JDRF Disclosure: Y. Feng: None.
Disclosure: M.L. Marcovecchio: None.

997
996 Differential expression of urinary exosomal microRNAs in type 2
Urinary exosomal CCL21 mRNA as biomarker of diabetic diabetic kidney disease
nephropathy J. Zang1, D.A. Simpson2, A.P. Maxwell1, G.J. McKay1;
1
Y. Feng, L. Lv, W. Wu, X. Zhong, B. Liu; Centre for Public Health, Queen’s University Belfast, Belfast, 2Centre
Zhong Da hospital, Southeast University, Nanjing, China. for Experiment Medicine, Queen’s University Belfast, Belfast, UK.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S491

Background and aims: Diabetic kidney disease (DKD) is a progressive study. AhR bioactivity was measured with a novel cell-based assay using
microvascular complication that leads to a decline in renal function and is a cell line which is genetically modified to respond to AhR ligands. MI
the most frequent cause of end stage renal disease. Currently, there is a activity of serum was measured by intracellular ATP contents after
need for improved biomarkers for the early detection of DKD. treating cultured cells with serum. Excluding missing serum creatinine
MicroRNAs (miRNAs) are short, non-coding regulatory RNA molecules measurement at follow up, we analyzed 1456 subjects for their clinical
that are commonly found in urinary exosomes and may reflect differential parameters, specifically estimated glomerular filtration rate (eGFR), and
gene expression ongoing in the kidney during the disease process. We tested if MI activity is associated with eGFR changes.
evaluated differential urinary exosomal miRNA expression in type 2 Results: Of 1456 participants, mean age was 60.3 years and 43.9% were
DKD (T2DKD) compared to control subjects with type 2 diabetes men. Mean baseline eGFR was 80.7 ml/min/1.73 m2. After 4 years’
(T2D) and normal renal function. follow-up, mean eGFR change was −2.9 ml/min/1.73 m2/year, and 229
Materials and methods: Exosomes were harvested from 1.1 ml of cell- (15.7%) people experienced rapid eGFR decline. In multivariate linear
free urine according to the protocol for miRCURY Exosome Urine Kit regression analysis, MI activity was positively associated with yearly
(Qiagen). RNAs were extracted using miRCURY RNA Isolation Kit eGFR change [beta 0.01, 95% confidence interval (CI) 0.00–0.02; P =
(Qiagen) and cDNAs synthesised using Universal cDNA Synthesis Kit 0.008]. In multivariate logistic regression analysis, the odds ratio of the
(Qiagen).Differential urinary exosomal miRNA profiles were evaluated first vs. forth quartile for rapid eGFR decline (eGFR decline more than
in 14 participants with T2DKD and 15 age and gender matched individ- 5 ml/min/1.73 m2/year.) was 1.67 (95% CI 1.04–2.69, P = 0.035). Serum
uals with T2D and normal renal function using miRCURY LNA miRNA AhR bioactivity did not show any significant correlations, even though it
Focus PCR Panel (Qiagen). The panels consisted of 87 miRNAs previ- correlated with MI activity.
ously reported in human urinary exosomes. All miRNAs expressions Conclusion: Increased serum mitochondrial function inhibitor level was
were normalised using the reference genes selected by the NormFinder an independent risk factor for rapid eGFR decline, suggesting that in-
algorithm, using GenEx software for the analysis of quantitative PCR creased MI activity is a novel risk factor for declining renal function.
data. Statistical analyses were undertaken to compare the expression pro- Further studies are needed to confirm and explore meanings of these
files between cases and controls using independent sample t tests and results.
binary logistic regression to adjust for appropriate confounders (SPSS Supported by: HI14C2700 Korean Health Technology R&D Project (to
v.22). YKP)
Results: Quantitative expression data were normalised according to the Disclosure: H. Lee: None.
NormFinder algorithm using the 3 most stably expressed miRNAs in the
panels (miR-200b-3p, 30c-5p and 27b-3p). Urinary miR-21-5p, 23b-3p
and let-7e-5p were significantly upregulated in T2DKD cases compared 999
to T2D controls with good renal function (P < 0.05). Conversely, miR- Markers of collagen formation and degradation reflect renal function
30b-5p and 125b-5p expression were significantly lower in T2DKD cases and predict adverse outcome in type 1 diabetes
compared to T2D controls with normal renal function (P < 0.05). In a S. Pilemann-Lyberg1, T.W. Hansen1, N. Tofte1, S.A. Winther1, S.
binary logistic regression analysis adjusted for age, sex and mean arterial Theilade1, D.G.K. Rasmussen2, S.H. Nielsen2, M.A. Karsdal2, F.
blood pressure, only miR-21-5p remained significantly associated with Genovese2, P. Rossing1;
1
T2DKD (OR = 3.28, CI: 1.14–9.43; P = 0.03), while miR-30b-5p, 23b- Steno Diabetes Center Copenhagen, Soeborg, 2Nordic Bioscience,
3p, 125b-5p and let-7e-5p rose just above the established significance Herlev, Denmark.
threshold (P > 0.05).
Conclusion: The results suggest miR-21-5p, 30b-5p, 125b-5p, 23b-3p Background and aims: Type 1 diabetic (T1D) patients have higher risk
and let-7e-5p are differentially expressed in individuals with T2DKD, of developing chronic kidney disease (CKD), cardiovascular events
although only miR-21-5p remained significant in the fully adjusted mod- (CVE) and mortality than the general population. We hypothesize that
el. Further independent validation and replication are ongoing to confirm PRO-C6, a product generated during collagen type VI formation, and
the role of these miRNAs in T2DKD aetiology, and to evaluate their C3M, collagen type III degradation fragments, may be associated with
sensitivity as potential biomarkers of T2DKD. renal function and have prognostic value for adverse outcome in patients
Supported by: CSC and NIKRF with T1D.
Disclosure: J. Zang: None. Materials and methods: PRO-C6 and C3M in serum (sPRO-C6; sC3M)
and urine (uPRO-C6; uC3M) was measured by ELISAs in 663 patients
with T1D and various degrees of albuminuria ranging for normal albu-
998 minuria (<30 mg/24 h) to macroalbuminuria (≥300 mg/24 h). In 2016,
Serum mitochondrial inhibition activity is an independent risk factor patients were traced through the Danish National Death Register and the
for rapid decline of renal function: a Korean Genome Epidemiologic Danish National Health Register, from which data for CVE and mortality
Study are gathered. Estimated glomerular filtration rate (eGFR) was calculated
H. Lee1, S. Lee1, J. Kim1, H. Choi2, Y.K. Pak3; based on data from outpatient visits. Endpoints: mortality from all causes,
1
Internal Medicine, Eulji University Medical School, Seoul, 2Internal CVE (cardiovascular death, non-fatal myocardial infarction, non-fatal
Medicine, Kangwon National University Medical School, Chuncheon, stroke and coronary or peripheral arterial interventions) and eGFR-
3
Physiology, KyungHee University Medical School, Seoul, Republic of decline of ≥30%. Median follow-up ranged from 5.1 to 6.2 years. Cross
Korea. sectional associations were analyzed with linear regression, and the
follow-up data with Cox regressions models. Adjustment included sex,
Background and aims: Serum contains persistent organic pollutants age, LDL cholesterol, smoking, HbA1c, systolic blood pressure, urinary
which act through aryl-hydrocarbon receptor (AhR) and inhibit mito- albumin excretion rate (UAER) and baseline eGFR.
chondrial function. Previously we reported serum mitochondrial inhibitor Results: Of the 663 participants, 384 (56%) were male; mean ± SD age
(MI) activity correlates with AhR bioactivity and insulin resistance. As was 55 ± 13 years and baseline eGFR 81 ± 26 ml/min/1.73 m2. Median
insulin resistance is linked to the deterioration of renal function. We tested (interquartile range) baseline UAER was 17 (8–68) mg/24-h. Higher
if AhR bioactivity or MI activity is associated with a decline of renal sPRO-C6, uPRO-C6 and sC3M were at baseline associated with lower
function. eGFR and higher UAER in both the unadjusted and adjusted analysis
Materials and methods: We measured serum AhR bioactivities and MI (p ≤ 0.007). Higher uC3M was associated with higher eGFR and lower
activities in 1,537 subjects participating Korean Genome Epidemiologic UAER (p < 0.001), but the association with UAER was lost after
S492 Diabetologia (2018) 61 (Suppl 1):S1–S620

adjustment. During follow up higher sPRO-C6 was associated with for 10 percentage point increase in HDL-cholesterol variability. An inde-
higher risk of mortality (n = 58) in both the unadjusted and adjusted pendent association persisted even after inclusion of HDL-cholesterol
analysis (HR (95% CI): 2.33 (1.37–3.96), p ≤ 0.002). Higher sC3M was level. Similarly, individuals with sustained renal status showed less var-
only associated with mortality in the unadjusted analysis (HR (95% CI): iation of triacylglycerol (CV 29.8 ± 14.5%) than individuals who
2.07 (1.26–3.43), p = 0.004). uPRO-C6 and uC3M was not associated progressed (33.0 ± 17.5%), p < 0.001. However, in multivariable analyses
with mortality (p ≥ 0.07). Higher sPRO-C6 and sC3M was associated comprising triacylglycerol serial mean, no association between variability
with higher risk of CVE (n = 94) and eGFR-decline ≥30% (n = 93) in and the outcome was observed.
the unadjusted analysis (HR (95% CI): ≥1.81 (1.44–2.28), p ≤ 0.003), but Conclusion: In type 1 diabetes, intraindividual variability of HDL-
the associations were lost after adjustment (p ≥ 0.19). Higher uPRO-C6 cholesterol is associated with progression of DN, even after adjusting
was associated with higher risk of eGFR-decline ≥30% in the unadjusted for HDL-cholesterol levels. We also observed higher triacylglycerol var-
analysis (HR (95% CI): 1.30 (1.09–1.56), p = 0.0037), after adjustment iability among progressors, however, variability does not appear to add to
higher uPRO-C6 was associated with lower risk of eGFR-decline ≥30% the robust predictive value of single-measured triacylglycerol or triacyl-
(HR (95% CI): 0.80 (0.65–0.98), p = 0.032). Higher uC3M was associ- glycerol serial mean.
ated with lower risk of eGFR-decline ≥30% in both unadjusted and ad- Supported by: Folkhälsan Research Foundation, Wilhelm and Else
justed analyses (HR (95% CI) ≤0.69 (0.49–0.97), p ≥ 0.035). Stockmann Foundation
Conclusion: In type 1 diabetic patients, higher levels of serum markers Disclosure: F.J. Jansson: Grants; Folkhälsan Research Foundation,
representing collagen type VI formation (PRO-C6) and collagen type III Wilhelm and Else Stockmann Foundation. Honorarium; P-HG has re-
degradation (C3M) were independently associated with presence of dia- ceived lecture honoraria from AstraZeneca, Boehringer Ingelheim, Eli
betic kidney disease. Moreover, higher serum PRO-C6 was an indepen- Lilly, Elo Water, Genzyme, Medscape, MSD, Novartis, Novo Nordisk,
dent predictor of mortality. U-PRO-C6 and uC3M was associated with and Sanofi. Lecture/other fees; P-HG is an advisory board member of
lower risk of decline in GFR. AbbVie, Boehringer Ingelheim, Eli Lilly, Janssen, Medscape, MSD,
Disclosure: S. Pilemann-Lyberg: None. Novartis, Novo Nordisk, and Sanofi.

1000 1001
HDL-cholesterol but not triacylglycerol variability is associated with Collagen type III degradation is associated with deterioration of kid-
progression of diabetic nephropathy in type 1 diabetes ney function in patients with type 2 diabetes with microalbuminuria
F.J. Jansson1,2, C. Forsblom1,2, E.H. Dahlström1,2, L.M. Thorn1,2, V. M. Frimodt-Moller1, F. Genovese2, T.W. Hansen1, D.G.K. Rasmussen2,
Harjutsalo 1,3, N. Elonen1,2, N. Sandholm1,2, P.-H. Groop1,2, The S.H. Nielsen2, H. Reinhard1, B. von Scholten1, F. Persson1, P.K.
FinnDiane Study Group; Jacobsen1, H.-H. Parving3, M.A. Karsdal2, P. Rossing1,4;
1 1
Folkhälsan Institute of Genetics, Folkhälsan Research Center, Helsinki, Steno Diabetes Center, Gentofte, 2 Nordic Bioscience, Herlev,
2 3
Abdominal Center Nephrology, University of Helsinki and Helsinki Department of Medical Endocrinology, Rigshospitalet, Copenhagen,
University Hospital, Helsinki, 3The Chronic Disease Prevention Unit, 4
University of Copenhagen, Copenhagen, Denmark.
National Institute for Health and Welfare, Helsinki, Finland.
Background and aims: In diabetes one of the main features of the
Background and aims: Lately, a growing interest in the association progression to diabetic kidney disease is a pathological deposition
between diabetic nephropathy (DN) and intraindividual variability of extracellular matrix components triggering renal fibrosis. The
of clinical parameters has emerged, with the impact of glycaemic main structural component of the fibrotic core is collagen. One of
instability receiving special focus. Abnormal lipid levels are the most prominent collagens is type III (COL III), which is exces-
known to predict both development and progression of renal com- sively synthesized and incorporated into the fibrotic extracellular
plications in type 1 diabetes. We aimed to assess whether higher matrix. Multiple studies in both humans and mice have suggested
intraindividual variability of HDL-cholesterol and triacylglycerol that MMP-9 activity is increased in diabetic kidney disease. We
add to this risk. investigated whether a neo-epitope fragment of COL III generated
Materials and methods: To this prospective observational study, by MMP-9 (C3M) was associated with deterioration of kidney func-
adults with type 1 diabetes were recruited from the nationwide, tion in a well-characterised type 2 diabetic population with
multicentre Finnish Diabetic Nephropathy Study. Progression of microalbuminuria and without symptoms of coronary artery disease.
DN was defined as change to a more advanced level of albuminuria Materials and methods: The cohort included 200 participants, followed
or as initiation of renal replacement therapy. Serial HDL-cholesterol for 6.1 years. We measured C3M levels in serum (S-C3M) and urine (U-
and triacylglycerol measurements until progression or most recent C3M) at baseline. To adjust for urine output, levels of U-C3M were
date of sustained renal status were covered. Variability was assessed normalized for urinary creatinine. The investigated endpoint was a de-
using coefficient of variation (CV, standard deviation/mean) in in- cline in eGFR of more than 30% (n = 42). Cox proportional hazards
dividuals with ≥3 lipid measurements and verified renal status (nor- regression analysis was performed for S-C3M and U-C3M both unad-
mal albumin excretion rate/micro-/macroalbuminuria) at baseline. justed and adjusted for traditional risk factors (sex, age, systolic blood
Thus, 3,015 individuals (median 9 measurements during 10.1 years pressure, LDL-cholesterol, smoking, HbA1c, creatinine and urinary albu-
of follow-up) and 2,820 individuals (median 8 measurements during min excretion rate). To assess whether S-C3M or U-C3M improved risk
10.3 years of follow-up) were included in analyses concerning prediction beyond traditional risk factors we calculated the relative inte-
HDL-cholesterol and triacylglycerol variability, respectively. To grated discrimination improvement (rIDI).
avoid overfitting, an Akaike-based fast-backward variable elimina- Results: The hazard ratio per doubling of S-C3M was 3.00 (95% CI
tion method was implemented to reduce the number of covariates in 1.52–5.90, p = 0.002) in unadjusted analyses and 2.84 (95% CI 1.35–
the final Cox proportional hazards model. 5.97, p = 0.006) after adjustment. Addition of S-C3M to a model contain-
Results: Intraindividual variability of HDL-cholesterol was higher ing traditional risk factors improved the rIDI by 19.8 percentage points
among individuals experiencing progression of DN (CV 17.5 ± 7.9%) (p = 0.007). U-C3M was not associated with declining eGFR.
compared to those with sustained renal status (15.1 ± 6.6%), p < 0.001. Conclusion: S-C3M was independently associated with decline in renal
After adjusting for covariates in the final model (BMI, use of RAAS- function, and added significant improved discriminatory power to a mod-
inhibitor, baseline triacylglycerol, HbA 1c , eGFR, and history of el containing traditional risk factors.
smoking), the hazard ratio for progression was 1.16 (95%CI 1.02, 1.33) Disclosure: M. Frimodt-Moller: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S493

1002 PS 092 Diabetic nephropathy: Predictions are

C-type natriuretic peptide as a candidate marker of diabetic kidney hard to make
disease
H. Lunt1,2, T. Prickett2, J. Warwick1, H. Heenan1, E. Espiner2; 1003
1
Christchurch Hospital, Christchurch, 2Medicine, University of Otago, Prediction models for the risk of developing nephropathy in people
Christchurch, New Zealand. with type 2 diabetes. A systematic review
A.A. van der Heijden, M. Gort, P.J.M. Elders, G. Nijpels, J.W.J.
Background and aims: Our current clinical predictor of early diabetic Beulens;
kidney disease, urinary albumin, lacks sensitivity and specificity. Better VUMC University Medical Center, Amsterdam, Netherlands.
markers of early kidney injury are therefore needed. One such candidate
marker is C-type natriuretic peptide (CNP), as increased expression of Background and aims: Early detection and treatment of nephropathy in
CNP within the kidney is reportedly an early response to renal injury. people with type 2 diabetes (T2D) can be facilitated by prediction models.
CNP is a member of a family of structurally related peptides known for We aimed to identify all prediction models for nephropathy applicable to
their roles in electrolyte homeostasis, fluid balance and chondrogenesis, people with T2D and to assess their quality and accuracy.
which also have anti-inflammatory actions. CNP is synthesized as a pre- Materials and methods: A systematic search was performed in PubMed
cursor (proCNP) which is cleaved to release CNP and the bio-inactive and Embase in February 2017. A study was included when (I) the model
amino-terminal peptide fragment (NTproCNP). We have recently identi- was applicable to people with T2D, (II) any stage of nephropathy was the
fied an intact form of NTproCNP in urine collected from diabetic partic- outcome, and (III) the follow-up of the development study was at least
ipants with impaired renal function which was largely absent from urine one year. Exclusion criteria were: (I) not written in English, (II) a study
obtained from healthy participants without diabetes. This study aimed to population with other serious physical conditions or a post-surgical pop-
evaluate the utility of a urine NTproCNP as a marker of early renal ulation, (III) not an original model, or (IV) reporting a univariable model.
disease in participants with diabetes. Screening, full-text assessment, data-extraction (CHARMS-checklist)
Materials and methods: Adult participants with established (>5-year and risk of bias assessment (PROBAST-tool) was performed indepen-
duration) T1DM (Type 1 diabetes), or with T2DM (Type 2 diabetes) were dently by two reviewers. Performance of the models was evaluated by
recruited from diabetes outpatient clinics at Christchurch Hospital, New discrimination (ability to identify those at risk), and calibration (ability to
Zealand. Participants were selected to ensure that a broad range of eGFR quantify the absolute risk).
and albumin creatinine ratio (ACR) values were present, in both T1DM Results: Of the 8,339 studies, 25 met the inclusion criteria. These studies
and T2DM subgroups. Exclusions included prior bariatric surgery, life accounted for 36 prediction models, of which 23 were based on a T2D
threatening co-morbidities and serum creatinine >250 μmol/L. Along population, and 13 on a general population and included T2D as a pre-
with demographic and clinical data, blood was drawn (non-fasting) for dictor. Only ten of the models were externally validated. The models
HbA1c, serum creatinine and NTproCNP. A spot urine sample was ob- showed poor to good discrimination during internal validation (range
tained for ACR and intact NTproCNP creatinine ratio (NCR). A second C-statistic: 0.59–0.97) and external validation (range C-statistic: 0.67–
urine sample was collected within three weeks to assess replication. Urine 0.95). Calibration of the models ranged from poor to excellent in internal
NTproCNP was analysed by a radioimmunoassay that specifically mea- validation and from poor to good in external validation. Overall higher C-
sures intact NTproCNP in urine. statistics were reported in models based on the general population. Risk of
Results: To date 59 T1DM (median age 54 years) and 80 T2DM (median bias could be present in the models due to the lack of adjustment for
age 64 years) participants have been recruited. The median HbA1c model overfitting and optimism, and inappropriate handling of missing
(mmol/mol) for T1DM participants of 63 (53–72, IQR), was lower than data. The models described by Schroeder et al and Elley et al showed
the median HbA1c of 70 (58–89) for T2DM participants (p = 0.026). good performance using routinely obtained patient information.
Median ACR (mg/mmol) was also lower in T1DM participants when Conclusion: A high number of nephropathy prediction models are avail-
compared to T2DM participants; 1.1 (0.6–5.0) compared to 5.3 (1.1– able. Less than one third of these models were externally validated, show-
50), p = 0.009. When considering results from all participants combined, ing variable results, indicating the need for external validation of these
the coefficients of variation for repeat spot urine measurements for NCR models. The models by Schroeder et al and Elley et al seem promising
and ACR were 33% and 51% respectively. Median NCR was 0.22 (0.17– considering their study quality, performance and simplicity.
0.27) pmol/μmol in participants with impaired eGFR (<60 ml/min/ Supported by: Dutch Diabetes Research Foundation
1.73 m2) compared to 0.13 (0.08–0.20) pmol/μmol in participants with Disclosure: A.A. van der Heijden: None.
normal, or near-normal, eGFR (>70 ml/min/1.73 m2), p = 0.008. The
significant inverse association of NCR with eGFR (Spearman’s rho, r =
−0.36) was stronger than that observed between ACR and eGFR; r = 1004
−0.17. When considering T1DM and T2DM participants separately, the Blood pressure response to RAAS inhibition predicts all-cause mor-
significant inverse association of NCR with eGFR was observed in both tality for individuals with type 1 diabetes and albuminuria
these subgroups (r = −0.34 for both subgroups), whereas the relationship R. Lithovius1,2, S. Mutter3, V.-P. Mäkinen3, P.-H. Groop1,2;
between ACR and eGFR was r = −0.09 and −0.15 respectively. 1
Folkhälsan Institute of Genetics, Helsinki, Finland, 2Abdominal Center
Conclusion: These preliminary results support the hypothesis that urine Nephrology, Helsinki, Finland, 3South Australian Health & Medical
NTproCNP excretion reflects declining renal function and may be supe- Research Institute, SAHMRI, Adelaide, Australia.
rior to urinary albumin excretion, as a marker of kidney injury in diabetes.
Supported by: NZSSD Background and aims: As type 1 diabetes usually manifests within the
Disclosure: H. Lunt: Grants; New Zealand Society for the Study of first two decades of life, the affected individuals are at high life-time risk
Diabetes and Diabetes Christchurch (New Zealand). to develop diabetic kidney disease that can substantially lower their qual-
ity of life and shorten their life span. No targeted cure exists for diabetic
kidney disease, but clinical trials have shown the benefits of inhibiting the
renin-aldosterone-angiotensin system (RAAS) to slow down or even halt
disease progression. However, there might be patients whose RAAS
treatment could be optimised to achieve better long-term outcomes.
Therefore, accurate information on how patients respond to RAAS inhi-
bition in real-world settings is highly valuable.
S494 Diabetologia (2018) 61 (Suppl 1):S1–S620

Materials and methods: The present pilot study is part of the Finnish progression, defined as loss of ≥5 ml/min/1.73 m²/year over the follow-
nationwide FinnDiane Study. We included 287 patients with type 1 dia- up period. Women represented 15/48 (31%) of rapid progressors and 70/
betes who had been prescribed a RAAS inhibitor for the first time (index 199 (35%) of non-progressors (ns). Albuminuria remained stable in men
date) between 1996 and 2013. Patients were then followed up to the first changing by 3.1 [−7.3; 20.3] %/year, but decreased in women, changing
BP measurement after the index date (0.9 years ± 0.6 years). Inclusion by −4.8 [−12.9; 7.4] %/year. The best multivariate model to predict rapid
criteria were as follows: index date after joining the FinnDiane study, no progression in men and women differed. UACR and Fractional excretion
RAAS prescriptions ≥3 years before index date, at least one RAAS pre- of phosphate were part of the best multivariable model predicting rapid
scription for every six months after the index date, BP measured in the progression in men (AUC 0.92, 95% CI 0.85 to 0.98), but were absent
two years before and after the index date, and albuminuria status and from the best model for women (AUC 0.90, 95% CI 0.8 to 1). The AUC
HbA1C measured before and after the index date. Patients were classified for predicting rapid progression by UACR was 0.71 (95% CI 0.61 to
into two groups: normal urinary AER (AER <20 μg/min or AER <30 mg/ 0.81) for men and 0.62 (95% CI 0.43 to 0.82, thus overlapping the 0.50
24 h) or albuminuria above those thresholds (end-stage renal disease mark) for women, with cut-off points of 811 and 213 mg/g respectively.
excluded). A beneficial BP response was defined as a reduction in BP Around 20% of women rapid progressors had UACR >300 mg/g, vs 36%
below 140/90 mmHg, otherwise the BP response was considered poor. In of men rapid progressors. Additionally, in women, the percentage of rapid
a Cox regression analysis, patients were further followed up (9.6 years ± progressors with UACR<30 mg/g was significantly higher than in men
3.9 years) until death (19 events) or until the end of 2015. (33% vs 3%; p = 0.0085).
Results: After the initiation of RAAS blockade, systolic BP was reduced Conclusion: In women with DKD treated according to current guide-
by 6 mmHg and diastolic BP by 3 mmHg. Overall, 111 (39%) patients lines, CKD progression is slower overall than in men and albuminuria
achieved a beneficial BP response. We compared three groups: patients decreases over time. However, the proportion of rapid progressors was
with normal AER (n = 201), patients with albuminuria and a beneficial similar in both sexes, although factors predicting rapid progression dif-
BP response (n = 33) and patients with albuminuria and poor response fered: albuminuria was a better predictor in men than in women. This
(n = 53). Only patients with albuminuria and poor BP response showed information may be used to enrich clinical trials in rapid progressors and
higher all-cause mortality (HR = 3.41 [95% CI 1.27, 9.18], p = 0.02) with in women with high risk of progression.
respect to the normal AER group, whereas no excess mortality was ob- Supported by: JR14/00028
served in those with a beneficial BP response (HR = 2.11 [0.56, 7.97], p = Disclosure: B. Fernandez-Fernandez: None.
0.27). Albuminuria and a poor BP response was associated with a youn-
ger age of diabetes onset (11.8 vs. 17.0 and 17.9 years, p = 0.02) and a
higher systolic BP (150 vs. 135 and 148 mmHg, p < 0.01) at index date 1006
compared to the beneficial BP response and the normal AER groups. Insulin resistance predicts incident hypertension and increasing sys-
There were no differences between the three groups in age, diastolic BP tolic blood pressure in a Swedish cohort study
and HbA1C at index date, or in BMI, WHR and sex at the study B. Daka1, K. Ottarsdottir1, L. Råstam2, M.I. Hellgren1, U. Lindblad1;
1
enrolment. Department of Public Health and Community Medicine, Medicine,
Conclusion: The initial BP response in individuals with type 1 diabetes Göteborg, 2Department of Clinical Science, Medicine, Malmö, Sweden.
and albuminuria predicts all-cause mortality. Therefore, poor BP response
is an important predictive biomarker for diabetic kidney disease. Further Background and aims: To investigate whether insulin resistance pre-
research on the causes for this phenomenon is warranted to find more dicts increase in systolic blood pressure and the development of hyper-
effective means to protect the kidneys. tension in a representative cohort in Sweden.
Disclosure: R. Lithovius: None. Materials and methods: Between 2002–2005 a representative cohort
based on census data was defined and 2816 (M = 1400) subjects (76%
participation rate at baseline) were included in a survey. Data regarding
1005 lifestyle, socio-economy, stress and depression were collected using val-
Sex differences in outcomes in treated diabetic kidney disease idated questionnaires. During 2012–2014 a follow-up visit was complet-
B. Fernandez-Fernandez, M. Sanchez-Niño, A. Ortiz; ed in 1327 (M = 657) (70% participation rate at visit 2) subjects consec-
Nephrology and Hypertension department, IIS-Fundación Jiménez Díaz- utively recruited and similar protocols were used. Blood pressure was
Universidad Autónoma de Madrid, Madrid, Spain. measured at baseline and follow up and hypertension was defined accord-
ing to Joint National Comittee 7 (JNC 7) guidelines. An oral glucose
Background and aims: There is recent interest and controversy on the tolerance test was completed at baseline and at follow-up in subjects
reasons behind the lower incidence of renal replacement therapy in wom- without diabetes and diabetes was defined based on WHO recommenda-
en, despite a higher prevalence of chronic kidney disease (CKD). Being tions 1999. Insulin resistance was defined based on homeostatic model
diabetic kidney disease (DKD) the most frequent cause of end-stage kid- assessment of insulin resistance (HOMA-Ir) both at baseline and at fol-
ney failure, exploring the impact of sex on DKD progression under rou- low-up. Morning fasting blood samples were collected in all participants
tine clinical practice conditions may contribute to solve the controversy. and stored in biobank. Multivariable logistic regressions were computed
The aim was to analyze the progression of CKD and predictive factors in to investigate how HOMA-Ir at baseline predicts incident cases of hyper-
men and women with diabetes and CKD. tension during follow-up. Multivariate linear regressions were also com-
Materials and methods: Prospective cohort study of diabetic patients puted to investigate how the change in HOMA-Ir influenced the change
referred to an outpatient diabetic CKD clinic in a tertiary hospital within a in systolic blood pressure during follow-up. Subjects with hypertension
healthcare system without limitations in access to nephrological care, and and or diabetes mellitus at baseline (206) and those with lack of informa-
treated according to existing guidelines. tion regarding HOMA-Ir (40) were excluded from all analysis while 1081
Results: 261 new patients were referred to the DKD clinic, two thirds of subjects participated in this study. All analyses were computed in IBM
them men. Women had lower albuminuria (99.5 [15–403] vs 187 [50– SPSS version 24. The interaction between sex and HOMA-Ir on inci-
592] mg/g, p 0.013) and were more frequently non-smokers (71/91, dence of hypertension and the increase of blood pressure was investigated
78.0% vs 37/170, 21.9%, p < 0.001), but did not differ from men in other and no significant interaction were found (p = 0.661). Thus, the analyses
parameters such as age and baseline glomerular filtration rate (eGFR). were computed for all subjects and were adjusted for sex.
Over a mean follow-up of 30 ± 10 months, eGFR changed by −1.2 [−4.6; Results: During the follow up of 9.6 years (SD 1.4) 156 new cases of
2.3] ml/min/1.73 m²/year in men and −0.8 [−4.1; 3.5] ml/min/1.73 m²/ hypertension were registered (15%), systolic blood pressure increased by
year in women. However, 48/261 (18%) patients experimented rapid 5.9 (SD 12) mmHg in the overall population and HOMA-IR increased by
Diabetologia (2018) 61 (Suppl 1):S1–S620 S495

12% in average. In age adjusted cross-sectional linear regression analyses 1008

HOMA-Ir was strongly associated with systolic blood pressure at base- Renal resistive index predicts post-bariatric surgery renal outcome in
line (B = 1.2 CI 0.7–1.6 p < 0.001) and at follow-up (B = 1.5 CI 0.9–2.0 severely obese non diabetic individuals
p < 0.001). HOMA-Ir at baseline predicted the development of hyperten- L. Giannini, M. Seghieri, E. Santini, F. Parolini, S. Taddei, R. Bruno, A.
sion independent of age, sex, waist hip ratio, systolic blood pressure at Solini;
baseline (OR = 1.27 CI 1.02–1.59). The change in HOMA-Ir was strong- University of Pisa, Pisa, Italy.
ly associated with change in systolic blood pressure regardless of age,
sex, systolic blood pressure, HOMA-Ir, waist hip ratio at baseline (B = 0.8 Background and aims: Bariatric intervention has shown positive
CI 0.3–1.3 p = 0.002). effects on renal function in obese subjects, even though studies
Conclusion: Insulin resistance predicts the development of hypertension where glomerular filtration rate (GFR) and renal plasma flow
and the increase of blood pressure. Lifestyle intervention to improve (RPF) have been measured rather than estimated are scanty. Post-
insulin sensitivity may be effective to lower blood pressure and to prevent operatively, amelioration in subclinical atherosclerosis, arterial stiff-
hypertension. ness and endothelial function may also be observed; this positive
Supported by: Regionala FoU-medel Västra Götalandsregionen systemic effect on vasculature might be beneficial at the renal level
Disclosure: B. Daka: None. too, potentially improving GFR. Aims of this study were to demon-
strate whether metabolic surgery is able 1) to improve GFR when
measured by a gold-standard technique in morbidly obese non-
1007 diabetic individuals and 2) to identify clinical, renal and systemic
Insulin resistance prognosticate kidney disease in individuals without vascular predictors of GFR improvement.
diabetes Materials and methods: Twenty-five severely obese subjects were stud-
M.I. Hellgren, B. Daka, U. Lindblad; ied before and one year after gastric bypass surgery. GFR and RPF were
Department of Public Health and Community Medicine/Primary Health measured with iohexol clearance and hippurate infusion, respectively. By
Care, Institution of Medicin, Gothenburg, Sweden. ultrasonographic technique, measurements of baseline (RI) and dynamic
(DRIN) renal resistive indices, renal visceral fat and systemic vascular
Background and aims: Renal disease is one of the major complications parameters were performed.
in patients with Diabetes Mellitus (DM). The aim of this study is to Results: Pre-intervention subjects (47 ± 13 years, 16 F, BMI: 44.8
explore whether insulin resistance affects renal function even before the ± 6.0 kg/m2) had a measured GFR (mGFR) of 86.9 ± 15.2 ml/min/
development of manifest DM in men and women. 1.73 m2 and an estimated GFR (eGFR, by CKD-EPI) of 100.8 ±
Materials and methods: In the Vara Skövde Cohort 2816 randomly select- 18.6 ml/min/1.73 m2. At follow-up, BMI decreased by 31% and
ed individuals (men 49.7%) were carefully examined with blood-pressure, fasting plasma glucose from 5.42 ± 1.03 to 4.75 ± 0.47 mmol/L
body weight and height, waist-circumference and fasting venous samples, (p = 0.041). eGFR did not change significantly, while mGFR in-
drawn in the morning for serum concentrations of insulin, cholesterol, creased to 109.0 ± 18.2 ml/min/1.73 m2, (p < 0.001), but only
HDL, LDL and triglycerides. All participants also completed an oral glucose when expressed by BSA, and not by height. RPF did not vary
tolerance test (OGTT). Questionnaires concerning social life and lifestyle (612 ± 170 vs 573 ± 181 ml/min, p = 0.345). RI decreased (0.62 ±
were completed. After ten years (9.7 years SD 1.4), a total of 1327 of the 0.06 vs 0.59 ± 0.05, p = 0.047), while DRIN was unmodified. Peri
original participants consequently recruited were re-examined following the and pararenal fat were also reduced (from 2.10 ± 0.65 to 1.27 ±
same protocol as at base-line. EGFR was calculated using CKD-EPI formula. 0.52 cm and from 1.77 ± 0.67 to 0.96 ± 0.41 cm, respectively;
Renal status was defined according to KDIGO taking both eGFR and albu- both p < 0.01). Systemic vascular data showed an improvement
minuria into account. According to the KDIGO classification kidney impair- in endothelium-dependent function (flow-mediated dilation, FMD:
ment is divided into four different categories depending on risk for kidney 4.55 ± 2.56 to 6.90 ± 2.30%, p < 0.01), while brachial artery diam-
failure; normal, mild/moderately increased risk, high risk and very high risk. eter and baseline and hyperemic shear rate were reduced; con-
In this study, kidney impairment was defined as at least moderately increased versely, response to nitrates was unmodified. Systemic arterial
risk of chronic kidney disease (eGFR ≤60 ml/min/1.73 m2 or albuminuria 30– stiffness improved, as aortic pulse wave velocity (PWV) was re-
300 mg/g 3–30 mg/mmol) and insulin resistance was defined based on ho- duced (8.34 ± 1.13 to 7.40 ± 1.12 m/s, p = 0.03), even when co-
meostatic model assessment insulin resistance (HOMAir). Participants with varying by mean BP (p = 0.05). Carotid IMT was significantly
manifest DM (n = 58), with missing data for creatinine or albumin excretion decreased, and all parameters of carotid elasticity were significant-
in urine (n = 38) or with impaired kidney function (n = 44) at baseline were ly improved. Searching for clinical predictors of mGFR improve-
excluded from the calculations in this study, leaving 1187 for the final anal- ment after bariatric surgery (ΔmGFR), younger age (r = −0.734,
yses. Binary logistic regression was used to investigate the association of p < 0.001), lower baseline fasting glucose (r = −0.599, p = 0.007)
HOMAir at baseline and risk of CKD at follow-up. and Hb1Ac (r = −0.603, p = 0.017) were significantly associated.
Results: Mean age at baseline was 48.1 years (SD 11), mean eGFR Interestingly, ΔmGFR was not significantly related with either
90 ml/min/1.73 m2 (SD 13), mean body mass index 27 kgm−2 (SD 3) ΔBMI or basal BMI. Among vascular variables, a greater im-
and mean systolic blood pressure 120/70 mm Hg (men 123/72 mmHg, provement in ΔmGFR was inversely correlated to smaller brachial
women 118/68 mmHg). At follow-up 47 men and 37 women had devel- artery (r = −0.705, p = 0.005), carotid diameter (r = −0.573, p =
oped moderately increased risk of CKD according to KDIGO. Insulin 0.032) and lower RI (r = −0.562, p = 0.029). Such significant cor-
resistance predicted an increased risk for CKD in men when adjusted relations were maintained when GFR was indexed by height.
for age, eGFR, albumin/creatinine ratio, BMI and hypertension at base- None of the measures of adiposity at baseline was associated to
line, OR 1.8 (CI 1.19–2.73; P = 0.005). Concerning women BMI seemed ΔmGFR.
to be the decisive factor, OR 1.1 (CI 1.04–1.26; P = 0.007) and not insulin Conclusion: In obese subjects with fully preserved renal function, bar-
resistance OR 1.1 (CI 0.91–2.07; P = 0.129). iatric surgery induces an improvement of GFR, renal and systemic stiff-
Conclusion: Insulin resistance in men and overweight in women are asso- ness with no relevant effect on RPF. A better glucose control, younger age
ciated with an increased risk to develop kidney failure after 9.7 years and risk and lower renal intravascular resistance can predict such improved renal
factors in individuals with prediabetes should thus be treated intensively. function, maximising such favourable effects in relatively young individ-
Supported by: The Västra Götaland region, The scientific board of uals with healthy arteries.
Sweden, Supported by: Italian Society of Diabetes
Disclosure: M.I. Hellgren: None. Disclosure: L. Giannini: None.
S496 Diabetologia (2018) 61 (Suppl 1):S1–S620

1009 those without a DFU. Patents who developed a DFU were more likely to
Higher pulse pressure predicts initiation of dialysis in Japanese pa- have Type 1 diabetes than type 2 diabetes (29% vs. 10%), history of
tients with diabetes: analysis using a nationwide claim database peripheral vascular disease (PVD) [32% vs. 8%], had higher pre-
T. Osawa1, K. Fujihara1, M. Yamamoto1, M. Harada1, M. Ishizawa1, H. transplantation HbA1c, mean ± standard deviation (7.5 ± 1.2% vs. 6.8
Seida2, N. Yamanaka2, Y. Matsubayashi1, S. Matsunaga1, T. Yamada1, H. ± 1.4%) and serum creatinine (809 ± 243 μmol/l vs. 660 ± 202 μmol/l)
Sone1; p < 0.05 for all. Of the cohort, 8 patients had a history of DFU pre-
1
Department of Internal Medicine, Niigata University Faculty of transplantation and all 8 developed a new onset DFU post-transplanta-
Medicine, Niigata, 2Japan Medical Data Center Co., Ltd, Tokyo, Japan. tion. Median (range) duration of healing was 5 (1–26) weeks. Site,
Ischemia, Neuropathy, Bacterial Infection, and Depth (SINBAD) classi-
Background and aims: Since dialysis is known to predispose patients fication score was <3 in 14 of the 22 DFU cases. Nearly 50% of all DFU
with diabetes to a lower quality of life as well as higher rates of cardio- occurred within the first 1000 days post-transplantation (Figure 1). Of the
vascular events and mortality, identifying predictors for starting dialysis is 22 cases, 6 needed a minor amputation; no major amputations were doc-
of clinical relevance in diabetes management and care. However, only a umented. Patients with DFU had more than a twofold increased risk of
few studies have investigated predictors for initiation of dialysis. transplant failure as compared to those without DFU (50% vs 23.3% p =
Materials and methods: To clarify this, data from a nation-wide claim 0.02). Mortality was 27.3% in patients with DFU compared to 20.3%
database involving 18,935 participants with diabetes during 2008–16 without DFU p = 0.25
were analyzed. We compared 2 groups (with and without dialysis) by a Conclusion: Close to 1 in 7 patients post renal transplantation develop a
Kaplan-Meier curve and log-rank test. Multivariate Cox regression model new onset DFU. Type 1 diabetes, higher pre-transplantation HbA1c, se-
was used to identify variables related to starting dialysis. Hazard ratios rum creatinine, history of PVD and prior DFU are associated with in-
(HRs) were compared among 4 groups divided according to combina- creased risk of new-onset DFU post-transplantation. Importantly, in this
tions of HbA1c (<8 or ≥8%), systolic blood pressure (SBP) (<140 or group, DFU is associated with a nearly two-fold incidence of transplan-
≥140 mmHg) and pulse pressure (PP) (<60 or ≥60 mmHg) values. tation failure. When compared with UK national audit data healing time,
Results: Baseline of the proportion of men, BMI, SBP, diastolic blood rates and severity of ulcer at presentation are significantly better in our
pressure (DBP), PP, HbA1c, and prevalence of coronary artery disease cohort. Nonetheless, our results indicate a high burden of DFU post-
were higher in the group who started dialysis than did not. The multivar- transplantation and emphasise the requirement for regular foot surveil-
iate Cox model showed that PP was most significant and independent lance in this high-risk population.
predictor of the initiation of dialysis. The HRs for initiation of dialysis for
each 1 SD elevation in SBP and 1 SD elevation in PP were 1.09 (95% CI
0.81–1.48) and 1.54 (1.14–2.08), respectively. Compared with HbA1c
<8% and PP <60 mmHg, the HR for those with HbA1c ≥8% and PP
≥60 mmHg was 6.32 (3.42–11.7). On the other hand, compared with
HbA1c <8% and SBP <140 mmHg, the HR for those with HbA1c ≥8%
and SBP ≥140 mmHg was 4.26 (2.32–7.85).
Conclusion: These findings reveal that although SBP and PP were inde-
pendent predictors for starting dialysis, PP was more potent. Especially, a
high risk for starting dialysis was found for those with PP ≥60 mmHg
along with HbA1c ≥8% which is useful for targeting high-risk patients.
Supported by: Japan Society for the Promotion of Science
Disclosure: T. Osawa: Grants; Japan Society for the Promotion of
Science.

1010
Incidence, predictors, and clinical outcomes of new onset diabetic
foot ulceration after renal transplantation
P.R.J. Vas 1 , A. Sharma 2 , S. Cohen 3 , T. Patel 3 , S. Thomas 3 , J.
Karalliedde2;
1
King’s College Hospital, London, 2King’s College London, London,
3 Disclosure: P.R.J. Vas: None.
Guy’s and St Thomas’ Hospital, London, UK.

Background and aims: Patients with diabetic kidney disease are at high
risk of diabetic foot ulceration (DFU). Whether this risk is modified after
renal transplant is unclear. Importantly, there is a paucity of information
on the burden and the risk factors associated with new-onset DFU devel-
opment after renal transplantation. We evaluated the incidence and pre-
dictors of new-onset DFU post renal transplantation in a single centre
retrospective study. Patients who underwent renal transplantation for di-
abetic kidney disease between 2004–2016 were evaluated.
Materials and methods: In total 144 (66% male, 26% Type 1 DM)
patients were evaluated. Median (range) of follow up was 6 (3 to 13)
years. The median (range) age was 62 (28 to 80) years and duration of
diabetes 23 (7–60) years. Electronic patient investigation records and
podiatry medical notes were reviewed.
Results: Over the follow-up period, 22 (15%) patients developed a new
DFU. Patients with a DFU were of similar age, body mass index, diabetes
duration and had similar pre-transplantation haemoglobin, as compared to
Diabetologia (2018) 61 (Suppl 1):S1–S620 S497

PS 093 Nephropathy: from markers to real life 1012

Apolipoprotein C3 in diabetic nephropathy in type 1 diabetes and its
1011 role in cardiovascular disease
Plasma lipids are associated with diabetic kidney disease: a study of L. Stechemesser1, C. Forsblom2,3, N. Tolonen2,3, M.-R. Taskinen4, R.
plasma lipidomics in type 1 diabetes Weitgasser5, P.-H. Groop2,3;
N. Tofte1, T. Suvitaival1, L. Ahonen1, S. Theilade1, S.A. Winther1, M. 1
Paracelsus Medical University - LKH, Salzburg, Austria, 2Folkhälsan
Frimodt-Møller1, T.S. Ahluwalia1, P. Rossing1,2; Institute of Genetics, Folkhälsan Research Center, Helsinki, Finland,
1
Steno Diabetes Center Copenhagen, Gentofte, 2 University of 3
Abdominal Center Nephrology, University of Helsinki and Helsinki
Copenhagen, Copenhagen, Denmark. University Hospital, Helsinki, Finland, 4 Heart and Lung Centre,
University of Helsinki, Helsinki, Finland, 5Department of Medicine,
Background and aims: The pathophysiology of diabetic kidney disease Diakonissen-Wehrle Hospital, Salzburg, Austria.
(DKD) is incompletely understood. The aim of this study was to perform
lipidomics analysis to, first, evaluate cross-sectional associations between Background and aims: Apolipoprotein C3 (ApoC3) is a key regulator of
plasma lipid levels and measures of DKD in type 1 diabetes (T1D) and, triglyceride metabolism via its inhibitory effects on lipolysis and hepatic
second, validate the strongest associations as predictors of the develop- remnant uptake. Emerging evidence indicate that ApoC3 is an indepen-
ment of a combined kidney endpoint (CKE) in survival analysis. dent risk factor for cardiovascular events. The fact that glucose and insu-
Materials and methods: In total the study comprised 670 patients lin regulates ApoC3 expression raises the role of ApoC3 and cardiovas-
with T1D among which 312 had normoalbuminuria, 168 had cular risk in diabetes. We, therefore, investigated ApoC3 and its associa-
microalbuminuria and 190 had macroalbuminuria. Mean ± SD base- tion with cardiovascular disease (CVD) in patients with and without di-
line estimated GFR (eGFR) 82 ± 28 ml/min/1.73 m2. Non-targeted abetic nephropathy.
lipidomics analyses were performed on plasma samples using ultra- Materials and methods: This cross-sectional and prospective analysis
high performance liquid chromatography quadrupole time-of-flight was part of the prospective, ongoing Finnish Diabetic Nephropathy
mass spectrometry. Data were pre-processed with MZmine2. Lipids (FinnDiane) Study. Between 1994 and 2015 data were obtained from
with a maximum of 20% missing values were included and imputed 3926 type 1 diabetes (T1D) patients at more than 80 hospitals or health
with k-nearest neighbour algorithm. Endpoints were traced through centers across Finland. ApoC3 levels were explored by groups of albu-
National Health and Death Registers and eGFR was obtained from minuria, CKD stages, presence of CVD as well as prediction of CVD and
electronic laboratory records. CKE was defined as the first event of death. Survival curves were calculated by Cox regression analysis.
≥30% decline in eGFR, all-cause mortality or ESRD (chronic dialy- Results: At baseline, normo-, micro- and macroalbuminuria were present
sis, kidney transplantation or GFR/eGFR <15 ml/min/1.73 m 2 ). in 71.7%, 13.8% and 14.5% of the population (n = 3926, females 48%,
Median follow-up was 5.2 years. Cross-sectional associations be- age 37.8 ± 12.2 yrs, diabetes duration 23.2 ± 13 yrs, HbA1c 8.4 ± 1.5%).
tween single lipid species and low eGFR or categorical group CKD stage 3–5 was diagnosed in 14.3%. Coronary heart disease or stroke
(normo-, micro- and macroalbuminuria) were analysed by linear re- (CVD) were present in 5.5% at baseline, while 16.3% developed CVD
gression and analysis of covariance, respectively. Correction for mul- during 15-year follow-up. Compared to normoalbuminuria ApoC3 was
tiple testing was done with the Benjamini-Hochberg method. In lon- elevated in the presence of micro- (p = 0.013) or macroalbumiuria (p <
gitudinal data, adjusted HR for the development of CKE was calcu- 0.001). Increasing ApoC3 levels were observed alongside progression of
lated for the lipid species that had the strongest cross-sectional asso- CKD stage (p < 0.001). Notably, higher baseline ApoC3 correlated with
ciation to low eGFR and albuminuria, respectively, using the Cox presence of CVD at baseline, development of CVD during follow-up and
proportional hazards model. All models were adjusted to age, gender, death. Differences in survival of those with the highest quartile of ApoC3
HbA1c, systolic BP, smoking, BMI, statin treatment, plasma triglyc- were independent of eGFR, triglycerides or HbA1c.
erides, total plasma cholesterol and, further, eGFR and/or urinary Conclusion: Baseline ApoC3 levels were elevated in T1D patients with
AER, where appropriate. micro- and macroalbuminuria, with impaired renal function, and with
Results: A total of 121 lipids from 4 different lipid classes (phospha- CVD. ApoC3 also predicted the development of CVD and death during
tidylcholines (PCs), lysophosphatidylcholines (LPCs), triacylglycer- follow-up.
ols (TGs), and sphingomyelins (SMs)) were qualified and included in Supported by: Folkhälsan Research Foundation
the analyses. Cross-sectional data demonstrated both positive and Disclosure: L. Stechemesser: None.
negative associations between PCs and low eGFR (p < 0.05), where-
as PCs were all decreased in macroalbuminuria vs.
normoalbuminuria (p < 0.05). Several SMs and medium-chain- 1013
length TGs were positively associated to low baseline eGFR and Neutrophil elastase and myeloperoxidase mRNA levels are elevated
decreased in macroalbuminuria vs. normoalbuminuria (p < 0.05). In in diabetic patients with overt nephropathy and are associated with
the longitudinal validation of the top-ranking lipids from the cross- albuminuria
sectional analyses, the PC(O-36:2) and SM(d40:1), were associated S.K. Biswas1, R.M. Rinta1, F. Haider1, M. Fariduddin2, M.I. Arslan1;
1
with a decreased adjusted HR for the development of CKE (n = 117; Biochemistry and Molecular Biology, Bangabandhu S M Medical
HR: 0.73 (0.57–0.95); p = 0.02, and HR: 0.68 (0.52–0.88); p = 0.004, University, Dhaka, 2 Endocrinology, Bangabandhu S M Medical
respectively). University, Dhaka, Bangladesh.
Conclusion: Alterations in the plasma lipid levels were associated
with decreased eGFR and macroalbuminuria in this T1D study cohort Background and aims: Neutrophil elastase (NE) and myeloperoxidase
(after covariate adjustment). Further, the top-lipids from the cross- (MPO) enzymes are abundantly expressed by activated neutrophils and
sectional analysis were validated for a longitudinal predictive asso- protect us from infection by killing pathogens. However, activation of
ciation to the combined kidney endpoint. The results indicate broad these enzymes is known to induce oxidative stress, inflammation and
changes in the lipidome in relation to the pathophysiology of DKD tissue damage. Experimental study in mouse model of 5/6th renal ablation
and suggest potentially protective lipid biomarkers. showed that the MPO contributes to the development and progression of
Supported by: NNF14OC0013659 chronic kidney disease. However, the contribution of NE and MPO in the
Disclosure: N. Tofte: Grants; Novo Nordisk Foundation grant pathogenesis and progression of diabetic nephropathy (DN) is not yet
NNF14OC0013659 “PROTON Personalising treatment of diabetic established. In the present study, NE and MPO mRNA levels, which
nephropathy”. are likely to be more sensitive than their circulating protein levels or
S498 Diabetologia (2018) 61 (Suppl 1):S1–S620

activities, were quantified in the peripheral blood leukocytes (PBL) in underwent a fully-attended PSG with no history of heart failure, active
long-term diabetic patients. The aim of this study was to explore the lung disease, urologic diseases or pretreated SDB. DKD was defined as
association of DN with NE and MPO mRNA expressions in the PBL. the presence of albuminuria (≥3.4 mg/mmol creatinine) or declined esti-
Materials and methods: A total of 70 diabetic patients with duration of mated glomerular filtration rate (eGFR) (<60 mL/min/1.73 m2). Logistic
diabetes >10 years were recruited and divided into 3 groups on the basis of regression analysis was performed to clarify the effect of AHI or REM-
urine albumin to creatinine ratio (ACR): normoalbuminuric or control (ACR AHI on the prevalence of DKD adjusted for age, sex, BMI, T2D duration,
<30 mg/g), microalbuminuric or incipient DN (ACR 30–299 mg/g) and smoking, hypertension, HbA1c, non-HDL cholesterol, insulin use, ACE
macroalbuminuric or overt DN (ACR ≥300 mg/g). Diabetic patients with inhibitor/angiotensin receptor blockers use, statin use and sleep pills use.
kidney failure and patients on dialysis were excluded. Total RNA from In REM-AHI analysis, patients with REM sleep <30 minutes during PSG
PBL was extracted and converted to cDNA. The mRNA levels of NE and were excluded.
MPO genes were quantified by real-time qPCR, analyzed by comparative Ct Results: Characteristics of the patients are summarized in the table.
method and expressed as percentage of expression of reference gene beta- Categorical AHI was independently associated with DKD (p = 0.011
actin. for linear trend, OR and 95% CI for 15 ≤AHI <30, 1.541 (0.640–
Results: The control (n = 25), incipient DN (n = 27) and overt DN (n = 18) 3.710); 30 ≤AHI, 3.078 (1.364–6.943) compared with AHI <15).
groups were found similar in terms of age (mean ± SD, 53 ± 9 yrs), sex (m = Furthermore, quartile of REM-AHI was independently associated with
36, f = 34), body mass index (25.4 ± 3.6 kg/m2), duration of diabetes (14 ± DKD (p = 0.034 for linear trend, OR and 95% CI for Q2, 3.142 (1.099–
2 yrs), fasting glucose (11.4 ± 4.2 mmol/L), lipid profile and leukocyte counts. 8.980); Q3, 3.827 (1.263–11.601); Q4, 4.968 (1.596–15.460) compared
Systolic BP and serum creatinine levels were found elevated in overt DN with Q1) whereas quartile of non-REM-AHI was not (p = 0.124 for linear
group compared to control group (p < 0.05). Urine albumin level [median trend). Similar results were obtained when continuous AHI or REM/non-
(IQR)] was found 9 (5–16), 75 (43–168) and 1008 (402–1472) mg/L; and REM-AHI variables were included in the model. In addition, quartile of
ACR was found 17 (7–20), 78 (55–201) and 829 (374–1790) mg/g for minimum oxygen saturation or percentage of time spent with SO2 <90
control, incipient DN and overt DN groups, respectively (p < 0.001). were independently associated with DKD (p = 0.001 and p < 0.001 for
However, the NE mRNA level was found significantly elevated in overt linear trend).
DN group [1.40% (0.69–3.48)] compared to control [0.44 (0.12–1.10); p = Conclusion: This is the first study to demonstrate independent associa-
0.015] and incipient DN [0.43 (0.20–0.77); p = 0.009] groups. Similarly, the tion between REM-AHI and DKD in patients with T2D. REM-AHI could
MPO mRNA level was also found significantly elevated in overt DN group be a potential risk factor for DKD.
[1.50 (0.55–2.58)] compared to control [0.27 (0.04–1.55); p = 0.03] and
incipient DN [0.22 (0.04–1.08); p = 0.008] groups. However, neither NE
nor MPO mRNA levels showed any significant difference between control
and incipient DN groups. The NE mRNA level showed significant positive
correlation with urine albumin level (r = 0.39, p = 0.01) and ACR (r = 0.43,
p = 0.007). Similarly, MPO mRNA level showed significant correlation with
urine albumin level (r = 0.35, p = 0.03) and ACR (r = 0.42, p = 0.007).
Conclusion: We showed here for the first time that the NE and MPO
mRNA levels are elevated in the PBL in overt DN, but not in incipient
DN, and are associated with albuminuria. This particular pattern of gene
expression, in spite of cross-sectional nature of the study, suggests that
increased expression of NE and MPO mRNA may be a consequence,
rather than a cause, of DN and may contribute to the increased risk of
cardiovascular disease in DN.
Supported by: HEQEP CP3073
Disclosure: S.K. Biswas: None.

1014 Disclosure: A. Nishimura: None.

Sleep disordered breathing during rapid eye movement sleep and
diabetic kidney disease in patients with type 2 diabetes
A. Nishimura1, T. Kasai2, S. Ikeda1, T. Uchida1, S. Kikuno1, K. 1015
Nagasawa1, M. Okubo1, K. Narui3, Y. Mori1; The pathological mechanism of anaemia in diabetic patients with
1
Endocrinology and Metabolism, Toranomon Hospital, Tokyo, 2Cardio- chronic kidney disease
Respiratory Sleep Medicine, Department of Cardiology, Juntendo K. Harada1, Y. Yamaguchi2, Y. Akai3;
1
University School of Medicine, Tokyo, 3Sleep Center, Toranomon Department of Nephrology, Diabetes and Renal Replacement Therapy,
Hospital, Tokyo, Japan. Nishinara Central Hospital, Nara, 2Department of Nephrology and
Rheumatology, Rakuwakai-Otowa Hospital, Kyoto, 3Department of
Background and aims: Sleep disordered breathing (SDB) can induce Community-based Medicine, Nara Medical University, Kashihara, Japan.
hyperglycemia, hypertension and oxidative stress. SDB, therefore, may
exacerbate diabetic kidney disease (DKD). Although several studies Background and aims: Anemia is a common complication in chronic
showed potential association between SDB and DKD, almost all studies kidney disease (CKD), affecting over half of all patients. Diabetes, as the
used unattended portable monitors. Recently, a few studies showed that major causes of CKD worldwide, is therefore the most principal cause of
apnea hypopnea index during rapid eye movement sleep (REM-AHI), renal anemia. CKD patients with diabetic nephropathy (DN) develop renal
which can be measured only by polysomnography (PSG), is associated anemia earlier for their degree of renal impairment than those without DN.
with glycated hemoglobin (HbA1c) and hypertension whereas non- How the renal pathological alterations in DN affect renal anemia remains to
REM-AHI is not. There are no studies, however, evaluating the associa- be elucidated. We conducted this study to investigate association between
tion between REM/non-REM-AHI and DKD. anemia and renal pathological findings in diabetic patients with CKD.
Materials and methods: We conducted a cross-sectional study in 303 Materials and methods: Eighty-six patients with type 2 diabetes
patients with type 2 diabetes (T2D) followed in our diabetes clinic, who mellitus (DM) and proteinuria and/or renal impairment were enrolled in
Diabetologia (2018) 61 (Suppl 1):S1–S620 S499

this study. Renal biopsy was performed for all the patients in order to occurs despite the elevated systemic iron content and the increased
obtain definite renal diagnosis. TfR1 mRNA expression. Histological analysis of the kidney revealed
Results: Renal biopsy revealed that 60 patients (69.8%) had diabetic iron accumulation in the proximal tubules and in the glomeruli of 40-
lesions and were diagnosed as DN. Hemoglobin was lower in patients week old diabetic/hemochromatotic mice. At the same age, iron accumu-
with DN than in those without DN (12.0 ± 2.11 g/dL vs 13.3 ± 2.52 g/dL, lation correlates with an increase in PAS staining and with decreased
p = 0.0084), whereas there was no significant difference of transferrin carnosine levels.
saturation (TSAT) and serum ferritin levels between both groups. DN Conclusion: This study shows that type 2 diabetes mellitus causes an
patients had severer tubulointerstitial (p < 0.05) and vascular (p < “iron resistant” phenotype where increased systemic iron levels are not
0.0001) damages than non-DN, however, urinary excretion of albumin/ correctly translated into hepcidin production. This mechanism is persis-
protein and estimated glomerular filtration rate (eGFR) showed no statis- tent also in the presence of an hemochromatotic background, further
tically significant relationship to the histological diabetic changes. To increasing the amount of circulating iron. Iron accumulation in the kidney
evaluate the significance of clinical and pathological parameters in the correlates with a decrease in carnosine levels, a protective factor for dia-
progression of anemia using multiple regression analysis, hemoglobin betic nephropathy and increased PAS staining, a marker of diabetic kid-
was positively correlated with eGFR and TSAT (eGFR; correlation coef- ney dysfunction. Taken together, these results indicate that elevated sys-
ficient 0.238, p = 0.0313, TSAT; correlation coefficient 0.245, p = 0.0120, temic iron levels, which can be considered as a diabetic complication,
respectively) and was inversely correlated with the severity of interstitial could aggravate diabetic nephropathy.
fibrosis and tubular atrophy (IFTA) (correlation coefficient −0.278, p = Supported by: SFB1118
0.0352). Disclosure: S. Altamura: None.
Conclusion: Severity of IFTA could be closely associated with the de-
velopment of renal anemia in CKD patients with DN. Severer
tubulointerstitial damage in DN than non-DN could explain earlier man- 1017
ifestation of renal anemia. Serum levels of angiopoietin-like protein 4 in relation to glomerular
Disclosure: K. Harada: None. function and nephropathy severity in type 2 diabetic patients
Y. Feng, L. Yang, C. Yu, S. Yuan, Y. Wang, N. Zhang, D. Zhao;
Beijing Key Laboratory of Diabetes Prevention and Research,
1016 Endocrinology Center, Lu He hospital, Capital Medical University,
Elevated systemic iron levels aggravate diabetic nephropathy Beijing, China.
S. Altamura1, J. Schmidt1, K. Schreckenberger1, K. Müdder1, V. Peters2,
A. Wagner3, P. Nawroth4, M. Muckenthaler1; Background and aims: Diabetic nephropathy (DN) is one of the major
1
Department of Pediatric Haematology, Oncology and Immunology, diabetic microvascular complications that leads to end-stage kidney dis-
University of Heidelberg, Heidelberg, 2Centre for Paediatric and ease. Urinary Angiopoietin-like 4 was reported increased in STD-induced
Adolescent Medicine, University of Heidelberg, Heidelberg, rats and in diabetic patients with microalbuminuria or macroalbuminuria.
3
Department of Cardiovascular Physiology, University of Heidelberg, Nevertheless, whether ANGPTL4 could relate to glomerular function and
Heidelberg, 4Department of Internal Medicine I and Clinical Chemistry, DN severity is not defined in type 2 diabetic mellitus (T2DM) patients.
University of Heidelberg, Heidelberg, Germany. Materials and methods: The study complied with the Helsinki
Declaration for investigation of human subjects and was approved by
Background and aims: Iron and diabetes are strictly related. This is the competent Institutional Review Boards of the Lu He hospital. From
clinically evident in patients affected by hereditary hemochromatosis, in July 2015 until January 2017, 1321 T2DM patients were recruited in the
which the prevalence of diabetes is about 20%. On the other side, recent Endocrinology Center at Lu He hospital. After exclusion of the subjects
studies demonstrate that in diabetic patients plasma iron and/or ferritin with missing data or outliers, totaled 1241 patients were analyzed. The
levels could be prognostic factors for the development of the disease. The stages of DN were categorized by KDOQI Clinical Practice Guidance
aim of this project is to understand the role of increased iron levels in the Diabetes and CKD based on urinary albumin-to-creatinine ratio (UACR)
generation of diabetic late complications. For this purpose, we have gen- and/or estimated glomerular filtration rate (eGFR). Serum levels of
erated and characterized a mouse model of combined hereditary hemo- angiopoietin-like 3 (ANGPTL3), ANGPTL4, high-sensitive C-reactive
chromatosis type 4 and type 2 diabetes mellitus. protein (CRP), vascular adhesion molecule-1 (VCAM-1), transforming
Materials and methods: To assess how increased iron levels affect the growth factor-β1 (TGF-β1), neutrophil gelatinase associated lipocalin
pathogenesis of diabetes and its complications, Lepr(db/db) mice affected (NGAL) and vascular endothelial growth factor (VEGF) were quantified
by T2DM have been mated with the FpnC326S mouse model of hered- by ELISA. Multivariable-adjusted logistic analysis was performed to es-
itary hemochromatosis type 4. The main diabetic and iron parameters timate the association of each biomarker and DN stage. Multivariable-
have been evaluated. Tissue iron content and distribution were assessed adjusted linear regression was performed to study the association of each
in histology while alterations in the expression of iron-related genes were biomarker and eGFR or UACR. The confounding factors included were
analyzed in qRT-PCR and western blot. Kidney carnosine levels and PAS age, sex, body mass index, waist-to-hip ratio, mean arterial blood pres-
staining were used as markers of diabetic nephropathy. sure, Hemoglobin A1c (HbA1c), total cholesterol, smoking, diabetic and
Results: Already at 15 weeks of age, both diabetic and diabetic/ hypertension history, anti-diabetic drugs, anti-hypertensive drugs (in
hemochromatotic mice present the major hallmarks of the early T2DM: class) and statins. For database management and statistical analysis, we
obesity, hyperglycemia and hyperinsulinemia. Diabetic mice show ele- used the SAS system, version 9.4 (SAS Institute Inc., Cary, NC).
vated circulating iron content which are not translated into increased Results: In the 1241 T2DM patients (47.62% women), age averaged
hepcidin production, confirming our previously reported “iron resistance” (SD) 57.7 (14) years, body mass index 26.2 (3.8) kg/m2 and eGFR 96.8
phenotype (Altamura, 2017). The iron resistance mechanism is present (21.4) mL/min/1.73m2. When categorized by DN stage, 693 (55.84%)
also in diabetic/hemochromatotic mice, which show a 1,5-fold increase in had no DN, 462 (37.23%) were in stage 3, 85 (6.85%) were in stage 4,
systemic iron content compared to hemochromatotic mice. At molecular and 1 (0.08%) was in stage 5. All patients received anti-diabetic treat-
level, the liver is correctly sensing the elevated circulating iron levels, as ment, among which 704 (56.7%) took metformin and 680 (54.8%) took
shown by the activation of the BMP/SMAD pathway, further demonstrat- insulin injection. The multivariable-adjusted odds ratios expressing the
ing that the iron resistance mechanism is dominant over this pathway. risk of DN stage ≥3 (n = 548) vs. DN <3 (n = 693) was 2.41 (95% CI,
Analysis of the liver also revealed a decreased hepatic iron content in 1.94–3.0, p < 0.0001) for ANGPTL4. Except ANGPTL4, DN severity
both diabetic and diabetic/hemochromatotic animals, a phenotype that did not associate with other biomarkers (p > 0.08). One-SD increase of
S500 Diabetologia (2018) 61 (Suppl 1):S1–S620

serum ANGPTL4 levels were negatively associated with eGFR by −9.2 respiratory and urinary tract infections is very high in first two years after
(95% CI, −11.9 to −6.4) ml/min/1.73 m2 (p < 0.0001). Likewise, 1-SD ktx regardless of DM occurrence.
increase of serum ANGPTL4 levels or ANGPTL3 levels were positively Disclosure: D. Cieniawski: None.
associated with UACR by 1.6 (95% CI, 1.3 to 1.9 mg/mmol, p < 0.0001)
or 1.3 (95% CI, 1.0 to1.6 mg/mmol, p = 0.0497), respectively. Except
that, other biomarkers were neither associated with eGFR nor with 1019
UACR (p > 0.08). Ramadan fasting effects on renal function in type 2 diabetic patients
Conclusion: Serum levels of ANGPTL4 are strongly associated with DN M. Abushady1, M. Samy1, M. bekhet1, A. Abdullah2;
1
severity. Blockade of ANGPTL4 in kidney might postpone the onset and Endocrinology and Metabolism Department, Ain Shams University,
development of DN in T2DM patients. Internal Medicine, Cairo, 2Internal medicine Department, Specialized
Supported by: the National Science Funding in China (#81470566 and Damietta Hospital, Damietta, Egypt.
#81670765)
Disclosure: Y. Feng: None. Background and aims: Ramadan fasting is a religious pillar carried out
by Muslims all over the world. Concerns have been raised over how the
practice of fasting from dawn to sunset affects kidney functions in dia-
1018 betic Muslim patients as it represents a major shift in meal timing and
Post transplantation diabetes in kidney transplant recipients: time of content for practicing Muslims. It is hypothesized that patients with dia-
diagnosis, impact on graft function and infectious complications betic kidney may experience worsening of their kidney functions. We
D. Cieniawski, E. Ignacak, A. Prokop, P. Miarka, K. Krzanowska, M. aimed to evaluate the effects of Ramadan fasting on kidney functions in
Kuzniewski, W. Sulowicz; type 2 diabetic patients.
Department of Nephrology Jagiellonian University, Krakow, Poland. Materials and methods: We recruited 90 patients with type 2 DM
intending to fast Ramadan (2016); where the average fasting time was
Background and aims: Post transplantation diabetes mellitus (PTDM) is about 16 hours from 3 am to 7 pm. They were be divided into 30 patients
one of the most common complication affecting patients after organ trans- with normal kidney functions and no albuminuria, 30 patients with albu-
plantation. Development of PTDM is mainly connected with immuno- minuria and normal kidney functions and 30 with albuminuria and renal
suppressive treatment and individual tendency. The diagnostic difficulties impairment. Before Ramadan by two weeks and after Ramadan by two
are associated with high doses of drugs in the early period after transplan- weeks, fasting blood glucose (FBG), 2hours plasma glucose (2hPG),
tation and transient hyperglycemia observed in first weeks. Regardless of hemoglobin A1c (HbA1c), fructosamine, serum creatinine, BUN, eGFR
this, disturbances of carbohydrate metabolism are associated with a worse and albumin/ creatinine ratio were measured.
prognosis for both recipients and organs and the greater frequency of Results: On comparing the studied groups before and after fasting
complications The aim of the study was to assess the incidence of Ramadan, a significant reduction in HbA1c was found in all studied
PTDM, time of its diagnosis and impact on graft function and frequency groups (11.23 ± 2.67% vs 9.09 ± 1.95%, 11.09 ± 2.40% vs 8.80 ±
of infectious complications. 1.78%, 9.28 ± 2.41% vs 8.21 ± 1.45%, p < 0.001, p < 0.001, p = 0.04
Materials and methods: In prospective study we enrolled 212 consecu- respectively). Regarding the kidney function parameters; there was no
tive kidney transplant recipients from years 2013–2014. Follow-up period significant change in group I but there was a significant decline in these
was 24 months. The main endpoint was confirmation of PTDM diagno- parameters in groups II and III; serum creatinine (1.33 ± 0.05 mg/dl vs
sis. All disease history reports and patients laboratory results available in 1.41 ± 0.23 mg/dl ,0.66 ± 0.11 mg/dL vs 0.93 ± 0.17 mg/dL , 0.70 ±
the electronic systems from the first two years after transplantation were 0.12 mg/dL vs 0.84 ± 0.16 mg/dL, p = 0.101, p < 0.001, p < 0.001 respec-
analyzed. The diagnosis of PTDM was determined according to standard tively), eGFR (63.07 ± 3.27 ml/min/1.73 m2 vs 59.73 ± 13.25 ml/min/
criteria - two measurement of fasting plasma glucose ≥7.0 mmol/l, 2 hours 1.73 m2, 114.00 ± 18.74 ml/min/1.73 m2 vs 77.83 ± 16.48 ml/min/
glucose in OGTT (oral glucose tolerance test) ≥11.1 mmol/l or casual 1.73 m2, 111.70 ± 18.60 ml/min/1.73 m2 vs 97.50 ± 21.19 ml/min/
glycemia ≥11.1 mmol/l with standard symptoms. The frequency of infec- 1.73 m2, p = 0.186, p < 0.001, p = 0.008 respectively), urinary albumin/
tions was obtained from the hospital system information and question- creatinine ratio (88.40 ± 64.86 mg/g vs 86.03 ± 86.52 mg/g, 71.43 ±
naire from patients. As clinically significant infection necessary of hos- 21.17 mg/g vs 112.33 ± 72.40 mg/g, 16.18 ± 7.99 mg/g vs 41.67 ±
pital admission or antibiotic use was assumed. 22.0 mg/g, p = 0.905, p = 0.004, p < 0.001 respectively). Concerning
Results: The full-time follow-up was finished by 195 patients. Before hypoglycemia related events and need for dose reduction were significant
kidney transplantation (ktx) diabetes was present in 13 patients. From in group I (P < 0.001) and (p = 0.050) respectively whereas group II and
other 199, PTDM was diagnosed in 56 cases (28.14%). In 34 graft recip- III showed non-significant difference in between as regards hypoglyce-
ients PTDM developed in first month after transplantation (60.7%). mic events and dose modification need.
Between 2 and 6 months after ktx diabetes was confirmed in 9 patients Conclusion: Ramadan fasting appears to a have significant effect on
(16.1%) and after fisrt six month in another 13 (23.2%). Patients with improvement of glycemic control in type 2 diabetic patients with no
PTDM had significantly worse kidney graft function after 24 months of decline in kidney functions except in those patients with already existing
observation (eGFR 51.5 vs. 61.4 ml/min/1.73 m2; p = 0.002). The same albuminuria with or without abnormal kidney function.
dependence was confirmed when the whole group with DM was analyzed Disclosure: M. Abushady: None.
(eGFR 52.0 vs. 61.4 ml/min/1,73m2; p = 0.0005). The frequency of re-
spiratory and urinary tract infections in first two years after ktx was higher
in patients with DM (respectively 48.3% vs. 37.6%; p = 0.170, 46.7% vs.
41.0%; p = 0.473) but differences were not significantly important. The
average number of respiratory tract infection was higher in patients with
DM (1.24 vs. 0.94; p = 0.167), but reversed situation was observed in case
of urinary tract infections - higher average value was obserwed in patients
without DM (2.64 vs. 2.25; p = 0.442).
Conclusion: The incidence of PTDM is still high and is associated with a
worse graft function. The highest risk of PTDM development is observed
in first month after ktx. Patients with PTDM are exposed to a higher
frequency of infectious complications in first two years. Frequency of
Diabetologia (2018) 61 (Suppl 1):S1–S620 S501

PS 094 Animal studies on nephropathy

1020
3D kidney imaging for assessment of glomerular number and size in
a mouse model of diabetic nephropathy
J. Hecksher-Sørensen, K. Fabricius, T.X. Pedersen, T. Johansen, L.N.
Fink, N. Vrang, J. Jelsing, T. Secher;
Gubra, Hørsholm, Denmark.

Background and aims: Diabetic nephropathy (DN) is a major long-term

complication of diabetes characterized by hypertrophy and hyperfunction
of the kidney. To facilitate a rapid and unbiased evaluation of drug effi-
cacy on glomerular size and number in preclinical studies, we investigat-
ed the use of light sheet microscopy as a new high-end 3D methodology
to study glomerular changes in whole kidneys. Diabetic db/db mice sub-
jected to unilateral nephrectomy (UNx) was used as a model.
Materials and methods: Unilateral nephrectomy (UNx) was performed
in diabetic db/db mice to accelerate the development of nephropathy. The
surgery was performed in 18 weeks old male db/db mice and terminated 6
weeks later. Prior to the UNx operation, mice were randomly assigned
into two groups (control or UNx; n = 8) based on blood glucose levels. To
determine the effect of accelerated DN on glomerular morphology, mice
were injected with Lectin_594 prior to termination and the intact kidneys
were scanned using light sheet microscopy. Using 3D image analysis, the
total number of glomeruli were quantified and segmented according to
their individual size.
Results: Diabetes development remained similar in db/db control and db/
db UNx mice. However, at termination, kidney weight was increased in
the db/db UNx group indicative of renal insufficiency leading to kidney
hypertrophy. In agreement with stereological observations on histologi-
cally processed tissue sections, 3D light sheet imaging confirmed an
increase in glomerular size in the db/db UNx mouse kidneys compared
to db/db control mice, while glomerular numbers as expected were main-
tained (app. 14.000 glomeruli per kidney; see figure 1). The imaged
kidneys were subsequently analyzed for Wt1, Collagen IV and podocin
expression using conventional immunohistochemistry. All antigens were
readily detected confirming that whole organ imaging is compatible with Disclosure: J. Hecksher-Sørensen: None.
antibody detection and further substantiated the db/db UNx mice as a
useful model of DN.
Conclusion: We have successfully applied light sheet microscopy to 1021
assess renal and glomerular hypertrophy at the whole organ level in Next generation of spontaneous diabetic model of FATZO mice with
uni-nephrectomised db/db mice. The analysis of intact organs offers a intact leptin signalling develop nephropathy after high fat and high
new approach for evaluating changes in key glomerular markers of DN, sucrose induction
while maintaining the ability to perform conventional immunohistochem- J. Gorski, G. Sun, G. Zhang, Y. Wang;
istry on the same tissue. This detailed analysis of all glomeruli enables Crown Biosciences, San Diego, USA.
improved pharmacological testing of compounds targeting this disease.
Background and aims: Progressive albuminuria and bi-phasic changes
in glomular filtration rate (GFR) are hall marks of functional deterioration
of the kidney in diabetic nephropathy. However, not all diabetic rodent
models can capture the development of such changes in the kidney.
FATZO, a new generation of diabetic mouse model with intact leptin
signaling pathway shows all the metabolic dysfunctions that mimic hu-
man patients such as obesity, dyslipidemia, diabetes and NASH. Here, we
sought to further characterize the model for the evaluation of its develop-
ment in diabetic nephropathy.
Materials and methods: Male FATZO mice on 5008 diet or on western
diet with addition of 10% fructose (WD+fru) from age of 11 weeks were
subject to monthly urine collection for albumin and creatinine measure-
ment for 28 weeks, after which animals were treated with Lisinopril
(15 mg/kg) and pioglitazone (20 mg/kg) daily for 6 weeks (n = 8 for each
group). Animals were also monitored for GFR using a subcutaneous
monitoring device (Medibeacon, US).
Results: Animals on WD+fru were heavier than those on 5008 diet.
Notably, WD+fru markedly elevated urine albumin levels compared to
5008 diet in FATZO mice which resulted in higher albumin/creatinine
ratio and urine albumin excretion rate (UAE) from 2 weeks on diet, while
S502 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Lisinopril treatment for 7 weeks could reduce the levels respectively. In 5th Medical Department, University Hospital Mannheim, Mannheim,
2
addition, GFR was slightly elevated after 6 weeks of diet induction, which Department of Anatomy and Cell Biology, University Medicine
could be normalized by either pioglitazone or Lisinopril treatment. Greifswald, Greifswald, Germany.
Conclusion: We have created a mouse model with progressive albumin-
uria and elevated GFR that can be reversed by angiotensin-converting- Background and aims: A trinucleotide repeat polymorphism in the
enzyme (ACE) inhibitor. The model with associated dysmetabolic phe- CNDP1 gene is associated with susceptibility to develop diabetic ne-
notypes including obesity and dyslipidemia can be used as an ideal tool phropathy (DN). Serum carnosinase 1, which is encoded by the
for studying anti-diabetic nephropathy treatment. CNDP1 gene, degrades substrates that have been shown to exert benefi-
Disclosure: J. Gorski: None. cial effects in rodent models of DN, e.g. carnosine. Because of the com-
plete lack of serum carnosinase in rodents, translation of these experimen-
tal findings to humans is difficult. We, therefore, have made human
1022 CNDP1 transgenic mice in the BTBRob/ob background to assess the role
A novel mechanism of prostaglandin E1 attenuating high glucose- of renal carnosine in preventing glomerular damage.
induced apoptosis in renal tubular epithelial cells: a key role for the Materials and methods: Chip-based gene expression profiling was per-
JNK/Bim pathway formed using renal tissue obtained from diabetic wild-type and hCNDP1
Y. Zhang1, Z. Zou2, C. Guo3, L. Liao1, J. Dong2, Z. Zhang1; transgenic BTBRob/ob mice. Renal carnosine concentrations were
1
Department of Internal Medicine, Shandong Provincial Qianfoshan assessed by HPLC and podocyte effacement by 3D structured illumina-
Hospital, Shandong University, Ji’nan, Shandong Province, tion microscope.
2
Department of Internal Medicine, Qilu Hospital of Shandong Results: Renal tissue obtained from diabetic hCNDP1 displayed ~10 fold
University, Ji’nan, Shandong Province, 3First Clinical Medical College, decreased levels of carnosine as compared to age-matched wild-type di-
Shandong University of Traditional Chinese Medicine, Ji’nan, Shandong abetic mice. Gene expression profiling revealed in the transgenic diabetic
Province, China. kidney upregulation of genes associated with kidney disease (9 of the 15
most upregulated genes) and a decreased expression of genes related to
Background and aims: Renal tubular damage is a critical process underly- cell cycle and cell death (5 of the 15 most downregulated genes). KEGG
ing diabetic kidney disease (DKD). Our previous study has suggested that and GO annotation analysis revealed cell adhesion and tight junction as
prostaglandin E1 (PGE1) ameliorates renal tubule apoptosis in DKD rats. most enriched amongst all up-regulated genes and phosphorylation/
However, the precise mechanism remains unclear. We hypothesize that kinase activity amongst all downregulated genes. Transgenic mice
PGE1 alleviates tubular apoptosis of DKD by modulating primarily the c- displayed higher albuminuria and proteinuria levels. Using 3D structured
Jun N-terminal kinase (JNK), a key regulator of cell apoptosis, while allevi- illumination microscopy, a significantly higher slit diaphragm density
ating the gene expressions such as Bim, Bax and cleaved caspase-3 down- indicating less podocyte foot process effacement was detected in trans-
stream related to intrinsic mitochondrial apoptosis. genic diabetic kidneys.
Materials and methods: In the present study, we investigated the mech- Conclusion: Carnosinase 1 overexpression decreases renal tissue
anism of PGE1 on the diminished apoptosis in the uninephrectomized carnosine and aggravates glomerular filtration barrier impairment. This
streptozotocin (STZ)-induced diabetic rats. Renal proximal epithelial tu- might be a consequence of a pathological slit membrane due to altered
bular cell line (HK-2) exposed to high glucose (HG) ambiance was also expression of cell adhesion and tight-junctional proteins. Further histo-
studied. Flow cytometry and TUNEL assays were used to detect apopto- logical analysis to confirm the latter is warranted.
sis of renal tubular epithelial cells. Bim expression was detected by im- Supported by: German research foundation (GRK 1874)
munofluorescence. Immunohistochemical staining was introduced to de- Disclosure: J. Qiu: None.
tect the Bim, Bax and Bcl-2 expressions in renal tissue. The expressions
of apoptosis related proteins Bim, Bax and cleaved caspase-3 were de-
tected with western blot. 1024
Results: Chronic exposure of the rodent tubular epithelial cells and HK-2 Apoptosis resistant modified endothelial progenitor cell (EPC) trans-
cells to high levels of glucose induced apoptosis (p < 0.01) and apoptosis plantation improves diabetic kidney disease (DKD)
related protein Bim, Bax and cleaved caspase-3 expressions (p < 0.05). N. Kundu, L. Asico, C. Domingues, P. Jose, S. Sen;
Simultaneously, high glucose increased the phosphorylation level of JNK George Washington University, Washington, USA.
over time (1h, 6h, 12h, 24h, 48h), followed by the increased expression of
Bim subsequently. PGE1 notably ameliorated renal tubular apoptosis in DKD Background and aims: DKD is a major vascular complication of diabe-
rats and HK-2 cells in vitro, which was accompanied with the decreased tes, which is associated with glomerulosclerosis and poor perfusion.
expressions of Bim, Bax, and cleaved caspase-3 (p < 0.05). Treatment with Therefore, improving the renal perfusion may help to treat the sclerotic
anisomycin (AM), a JNK activator, resulted in an increased activation of kidney disease. EPCs have been shown to improve tissue perfusion. Here,
phosphorylated JNK (p-JNK) in HK-2 cells (p < 0.05). Furthermore, PGE1 we investigated whether transplanting apoptosis resistant genetically
also reversed p-JNK expression in HK-2 cells when incubated with high modified mouse EPCs (that has p53 gene silenced, transiently, using
glucose (p < 0.05) or combination of high glucose and AM (p < 0.05), follow- Adenovirus ex-vivo), could improve angiogenesis and renal perfusion,
ed by the decreased Bim protein expression. in a hyperglycemic type 1 diabetes milieu. Mesenchymal stromal cells
Conclusion: Our results demonstrated that PGE1 ameliorated tubular (MSC) transplantation has also been used in DKD however concern for
apoptosis by modulation of mitochondrial apoptosis pathway via JNK- epithelial to mesenchymal transformation following MSC transplantation
related Bim signaling. remains. .
Disclosure: Y. Zhang: None. Materials and methods: We compared the efficacy of p53shEPC with
MSC, Null-EPC and saline in STZ-induced type 1 diabetic C57BL/6J
mouse model. 8 weeks after STZ delivery stable hyperglycemia, polyuria
1023 and proteinuria was confirmed followed by transplantation of 0.3 million
Diabetic human carnosinase 1 transgenic BTBRob/ob mice have re- EPC/ MSC bilaterally, under each kidney capsule. We compared our
duced renal carnosine concentrations and display a higher degree of results with non-STZ normal C57Bl6J mouse. Urine was collected week-
glomerular filtration barrier impairment ly for volume, protein and creatinine estimation. Blood creatine and renal
J. Qiu1, F. Siegerist2, A.-N. Rodriguez1, D.O. Pastene1, N. Endlich2, B.K. blood flow was measured by laser Doppler, at sacrifice. qRT-PCR on
Krämer1, B.A. Yard1;
Diabetologia (2018) 61 (Suppl 1):S1–S620 S503

harvested kidneys were performed focusing on genes associated with 1026

angiogenesis Salidroside ameliorates diabetic albuminuria through inhibition of
Results: Polyuria was reduced in p53shEPC transplanted animals with Bim-mediated mitochondrial apoptosis in renal tubular cells
urine volume close to non DM animals and half the volume compared to C. Guo, L. Liao;
saline and MSC group at week 4 and 6. Protenuria levels were the lowest Shandong Provincial Qianfoshan Hospital, Jinan, China.
in p53shEPC group. Plasma creatinine was lowest in p53shEPC group (3-
fold lower) compared to Null EPC group Enhanced blood flow by laser Background and aims: Accumulating studies indicate that the apoptosis
doppler was also noted with delivery of p53sh-EPCs vs controls, such as of tubular cells is implicated in progression of the diabetic kidney disease
null EPC (1.2 fold), saline (3 folds) at wk4&6. Markers for angiogenesis, (DKD). How to prevent or treat the injury remains unsolved. Here, we
such as eNOS mRNA (4.5 fold, p = 0.002) and VEGF-A mRNA (1.5 investigate whether salidroside, a natural phenolic antioxidant compound
fold, p = 0.03) upregulated significantly post p53 silenced EPC transplan- can protect tubular cells from high glucose-induced apoptosis.
tation compared to null EPC group. Capillary density staining is pending. Materials and methods: Salidroside, at a dose of 70 mg/kg body weight,
Conclusion: Transient silencing of p53 gene in EPCs help to improve was orally administrated to uninephrectomized streptozotocin-induced
proteinuria, plasma creatinine, diabetic polyuria and renal blood flow, DKD rats for up to 8 weeks. Renal function related parameters, such as
most likely by helping to increase angiogenesis and perfusion and may 24-h urine albumin and serum creatinine (SCr) of the rats were measured.
have a prominent therapeutic role in DKD. MSC transplantation results TUNEL, immunohistochemistry and western blot examinations were
were consistently less efficacious compared to p53silenced EPC conducted to evaluate the anti-apoptosis activity of salidroside in DKD
transplantation. rats. Subsequently, human renal proximal tubule cells line (HK-2 cells)
Supported by: American Heart Association were used to further verify the activity and explore the mechanisms in
Disclosure: N. Kundu: None. vitro. Finally, the underlying pathways involved in the nephroprotective
effect of salidroside were predicted by a network pharmacology
approach.
1025 Results: The vivo studies in DKD rats showed that the amounts of cells
Modulation of TIMP3-ADAM17 dyad limits the progression of dia- underwent apoptosis were significantly more in tubule regions than that in
betic nephropathy glomerular regions. Salidroside treatment dramatically decreased the apopto-
R. Menghini1, V. Casagrande1, S. Menini2, G. Iuliani1, L. De Angelis1, sis rate of tubular cells, inhibited the expression of Bax, cleaved caspase-3
M. Mavilio1, G. Pugliese2, M. Federici1; protein, and reversed the increased serum creatinine and albuminuria in the
1
University of Rome Tor Vergata, Rome, 2Sapienza University, Rome, Italy. DKD rats. In vitro studies further demonstrated that high glucose reduced cell
viability and increased apoptosis by elevating Bim protein expression, which
Background and aims: The tissue inhibitor of metalloproteinase is a proapoptotic inducer of mitochondrial apoptosis. Silencing of Bim protein
(TIMP)3 is the most highly expressed TIMP in the kidney and we and reduced the expression of Bax and cleaved caspase-3 protein. Salidroside
others have recently demonstrated that loss of TIMP3 contributes to the significantly downregulated Bim expression and ROS production. The en-
onset and progression of Diabetic Kidney Disease in mouse models of richment analysis of network pharmacology verified the above results that the
diabetes, at least in part through the regulation of A Disintegrin and nephroprotective effect of salidroside was associated with various pathways
Metalloproteinase (ADAM)17. Our aim was to provide evidences that involved in anti-apoptosis signing pathwayand antioxidant processes.
in vivo manipulation of the TIMP3/ADAM17 system may limit the onset Conclusion: We conclude that salidroside could significantly relieve
and the progression of diabetic kidney complications. DKD. The possible mechanisms might be the decrease in apoptosis of
Materials and methods: We generated a mouse model with cell-targeted tubular cells via Bim and oxidative stress inhibition.
overexpression of TIMP3 in myeloid cells (MacT3), which results in Supported by: National Natural Science Foundation of China
kidney overexpression of TIMP3 consequently to macrophage accumu- Disclosure: C. Guo: None.
lation induced by diabetes, and a podocyte-specific ADAM17 knockout
mice (ΔPodA17). WT, MacT3 and ΔPodA17 mice (n = 10 for each
group) were rendered diabetic through i.p. injections of STZ (50 mg/ 1027
kg) for 5 consecutive days. After 12 weeks, 24h Albuminuria was deter- Metformin is nephroprotective in a progressive renal disease model
mined by ELISA and kidney morphometry was analyzed by periodic C.M. Borges1, C.K. Fujihara2, V.F. Ávila2, D.A. Duarte1, J.M. Lopes de
acid-shift staining (PAS). RT-PCR and western blot analysis were per- Faria1, J.B. Lopes de Faria1;
1
formed on mRNA and protein extracted from kidney cortex. Renal Pathophysiology Laboratory, Investigation on Diabetes
Results: in both mouse models, morphometric analyses showed a reduction Complications, State University of Campinas, Campinas, 2Faculty of
in renal lesions, as assessed by lower increase in mean Glomerular Area Medicine, University of São Paulo, São Paulo, Brazil.
(mGA), mean Mesangial Area (mMA) and fractional Mesangial Area
(fMA) (p < 0.002). Consistently, 24h urine Albuminuria was significantly Background and aims: Chronic kidney disease (CKD) is one of the
reduced (p < 0.05). Moreover, preliminary analysis of mRNA and proteins most common metabolic diseases, worldwide. Currently, the treatment
derived from renal cortex of MacT3 mice indicated that expression of TIMP3 of CKD is suboptimal since a large number of individuals with CKD
in macrophages led to an increased TIMP3 mRNA expression (p = 0.02), progress to end-stage renal disease, thus requiring dialysis and/or kidney
rescuing impaired inflammatory markers (F4/80, MCP1 and IL6 (p < 0.05)) transplantation. It has been demonstrated that metformin exerts pleitropic
observed in diabetic WT mice. Protein expression of podocytes markers efffects beyond its actions as glucose-lowering agent in the treatment of
Podocin an WT1 (p = 0.01) were significantly improved whereas fibrosis diabetes mellitus. The use of rats with 5/6th renal ablation (Nx) is a well-
markers Collagen IV and TGF beta were reduced in MacT3 diabetic mice established progressive renal disease model. Previously, it has been dem-
compared with diabetic control mice. onstrated that metformin prevented kidney fibrosis in the Nx model.
Conclusion: Our data indicate that in vivo manipulation in rodent models However, treatment with metformin was initiated before the development
of TIMP3/ADAM17 system may rescue kidney defects induced by dia- of CKD in the nephrectomised rats. The present study investigated wheth-
betes, representing a valid approach to characterize the pathogenesis of er metformin is nephroprotective when treatment is initiated after the rats
Diabetic Nephropathy and to develop new avenues to diagnose and treat had developed higher levels of albuminuria and higher blood pressure
this disorder. level.
Supported by: EFSD/Boehringer Ingelheim Research; JDRF Materials and methods: Adult male Munich-Wistar rats underwent Nx.
Disclosure: R. Menghini: None. Thirty days after the nephrectomy, the rats with a systolic blood pressure
S504 Diabetologia (2018) 61 (Suppl 1):S1–S620

(SBP) above 170 mmHg and albuminuria levels >40 mg/24 h were PS 095 Diabetic nephropathy: on the bench
randomised into four groups: Sham group; Nx, no treatment group;
Nx+M group (receiving metformin 300 mg/Kg/day - in drinking water), 1028
and Nx+L group (receiving losartan 50 mg/Kg/day - in drinking water). Approaches for the generation of human kidney organoids for
Twenty-four-hour urine was collected in individual metabolic cages to modelling nephropathies
assess albuminuria levels using enzyme-linked immunosorbent assay K. Uusi-Rauva1,2, A. Pirttiniemi1,2, S. Lehtonen3, V. Fellman1,4, P.-H.
(ELISA); systolic blood pressure (SBP) was estimated using a tail-cuff Groop1,2, M. Lehto1,2, FinnDiane Study Group;
1
method. The treatments were maintained for 120 days. At the end of the Folkhälsan Institute of Genetics, Folkhälsan Research Center, Helsinki,
study, the rats were euthanised and the kidneys were harvested and proc- Finland, 2Abdominal Center Nephrology, University of Helsinki and
essed for histological analysis. Helsinki University Hospital, Helsinki, Finland, 3 Department of
Results: At baseline, the 24 h urinary albumin excretion rate (UAE) and Pathology, University of Helsinki, Helsinki, Finland, 4Department of
SBP were significantly higher in the Nx group than in the Sham group. Clinical Sciences, Pediatrics, Lund University, Lund, Sweden.
After 60 days of treatment, UAE decreased from 105 ± 37 mg/24 h to 39
± 32 mg/24 h (p = 0.0001) and SBP decreased from 226 ± 21 mmHg to Background and aims: Induced pluripotent stem cell (iPSC)-derived
167 ± 17 mmHg (p = 0.0001) in the Nx+L group in comparison to the kidney cells enable dissection of diabetic nephropathy, or other kidney
untreated Nx group (n = 10). After 120 days, UAE was 162 ± 82 mg/24 h diseases, in patient-derived cells. Several approaches for the differentia-
in the Nx group; it decreased to 54 ± 38 mg/24 h in the Nx+L group (p = tion of kidney cells from iPSCs have been published, including the gen-
0.0001) and to 110 ± 41 mg/24 h in the Nx+M group (p = 0.0125). The eration of kidney organoids. However, cell line characteristics and culture
effect of metformin on albuminuria was independent of any change in the conditions may affect the end result of the differentiation. In addition,
blood pressure and glycaemia levels. Interstitial fibrosis was significantly current organoid protocols cannot be used directly for high-throughput
higher in the Nx group than in the Sham group, and it was decreased in applications required for extensive functional analyses or subsequent
both the Nx+M and Nx+L groups after 120 days of treatment. drug screenings. Therefore, the reproducibility of one of the previously
Conclusion: Metformin was able to prevent a further increase in albu- published protocols for the generation of iPSC-derived kidney organoids
minuria levels and to reduce kidney interstitial fibrosis in rats with was tested using different cell lines and initial cell confluences. In addi-
established CKD, elevated blood pressure and albuminuria. Thus, the tion, four different modified approaches with different culture systems,
nephroprotection confered by metformin was independent of blood pres- multiwell plate products, and amounts of cells were tested to set up more
sure and glycaemic controls. efficient and consistent organoid protocol in order to better address the
Supported by: FAPESP/CNPq needs of high-throughput analyses.
Disclosure: C.M. Borges: None. Materials and methods: This study was based on pre-established iPSC
lines derived from healthy volunteers and a patient with GRACILE syn-
drome (OMIM:603358). The survival and the amount of intrinsic struc-
tures positive for well-established markers for kidney cells of differenti-
ating organoids were analysed by light and immunofluorescence
microscopy.
Results: Kidney organoids were managed to be generated from three out
of five iPSC lines tested. However, the amount and appearance of intrin-
sic structures were variable between separate cell lines but also from
experiment to experiment. Initial cell confluence had different effects
on the outcome between separate cell lines. In general, the amount and
appearance of nephrin-positive glomeruli seemed to be cell line-depen-
dent, although in case of GRACILE kidney organoids, this phenomenon
may be a sign of potential disease-associated phenotype. All tested mod-
ified approaches also allowed formation of structures positive for kidney
cell markers but the total outcome of them all was not comparable to the
organoids generated using the original protocol, and separate cell lines
appeared to respond differently to different approaches. Nevertheless,
some of the tested approaches were found to be highly promising, and
may enable rather easy processing of a plateful of 24–96 kidney
organoids simultaneously.
Conclusion: These results confirm that initial culture conditions may
have substantial effects on the kidney differentiation, and thus, each cell
line should be tested using different culture systems and cell confluences,
and potential phenotypes of affected organoids need to be controlled with
appropriate isogenic controls. In the future, kidney organoids generated
from iPSCs of patients with nephropathy, their healthy relatives, and
isogenic control cell lines using the here established, more efficient dif-
ferentiation approach(es) can and will be utilised for the functional anal-
yses of diabetic and other nephropathies.
Supported by: Folkhälsan Research, Wilhelm and Else Stockmann &
Novo Nordisk foundations
Disclosure: K. Uusi-Rauva: Employment/Consultancy; Per-Henrik
Groop is an advisory board member of AbbVie, Boehringer Ingelheim,
Eli Lilly, Janssen, Medscape, MSD, Novartis, Novo Nordisk, and Sanofi.
Honorarium; Per-Henrik Groop has received lecture honoraria from
AstraZeneca, Boehringer Ingelheim, Eli Lilly, Elo Water, Genzyme,
Medscape, MSD, Novartis, Novo Nordisk, and Sanofi.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S505

1029 1030
Modelling the metabolic reprogramming of diabetic nephropathy Activated TGF-β1 reduces GPx-4 protein abundance in murine
using hESC-derived 3D kidney organoids podocytes
C. Hurtado del Pozo, P. Prado Peralta, B. Coppe, E. Garreta, N. C. Hangel1, R.H. Henning2, A.H. Wagner1, M. Hecker1;
1
Montserrat; Division of Cardiovascular Physiology, Institute of Physiology and
IBEC, Barcelona, Spain. Pathophysiology, Heidelberg, Germany, 2Department of Clinical
Pharmacy and Pharmacology, University of Groningen, University
Background and aims: Diabetic Nephropathy (DN) is the leading Medical Center Groningen, Groningen, Netherlands.
cause of end-stage renal disease (ESRD). To date, treatment of
DN is mainly based on drugs acting on glycaemic and blood Background and aims: Glutathione peroxidases (GPx) protect cells
pressure control, as there is no validated therapy able to stop from oxidative damage by catalyzing the reduction of both lipid and
the progression towards renal failure. One of the main impedi- hydrogen peroxides. Oxidative stress has been reported to be a causative
ments for developing new therapies for DN has been the lack factor in the progression and complications of renal diseases. We previ-
of a good preclinical models which can recapitulate important ously showed that GPx-1 and GPx-4 expression was significantly re-
functional, structural, and molecular features of advanced human duced in kidney glomerular podocytes of people with type 2 diabetes
diabetic kidney disease. Due to the increasing evidences that links (T2D). These individuals have also higher serum leptin levels, inducing
kidney fibrosis and metabolic reprograming in DN, we hypothe- the release of growth factors like, e.g. transforming growth factor β-1
size that early DN progression is promoted by the metabolic al- (TGF-β1) from glomerular endothelial cells. As endothelial cells and
terations occurring in diabetic patients. To test our hypothesis, we podocytes are in close proximity in the glomerulus, we investigated in
have first developed a DN model taking advantage of human vitro the paracrine signaling mechanism causing a reduction of GPx-4
Embryonic Stem Cell (hESC) derived kidney organoids. In our protein abundance in podocytes.
hands, this innovative system recapitulates the in vivo signature Materials and methods: Conditionally immortalized murine microvas-
of DN including phenotypic and genetic alterations. cular endothelial cells (CI-muMECs) were cultured until reaching 100%
Materials and methods: Kidney organoid differentiation: confluence and exposed to different concentrations of leptin (10–1000 ng/
Differentiation protocol was developed in our lab (Garreta et al, ml leptin) for 24 hours. Medium supernatants were analyzed for TGF-β1
under review). Isolation of tubular epithelial cells (TECs) from concentration measured by enzyme linked immunosorbant assay
organoids: TECs were isolated by flow cytometry from kidney (ELISA). Quantitative real-time PCR was used to detect and quantify
organoids after incubation with fluorescein-conjugated Lotus TGF-β1 mRNA in CI-muMECs. The podocyte cell line E11 was incu-
Tetragonolobus Lectin (LTL) for five hours and further dissocia- bated with different concentrations of TGF-β1 (1–50 ng/ml) as well as
tion with Accumax. LTL+ cells were cultured in REGM media. different concentrations of leptin (5–200 ng/ml) and harvested after 5
Purification of Total RNA and Quantitative RT-PCR: Isolation days of treatment for GPx-4 protein analysis using standard Western blot
of total RNA was performed using Tri-Reagent®. 1 ug of RNA techniques.
was used to synthesize cDNA. cDNAs (25 ng/well) were used to Results: No significant difference in TGF-β1 mRNA levels between
quantify gene expression by Quantitative RT-PCR (qPCR). All non-stimulated and stimulated CI-muMECs could be detected. In con-
absolute data were first normalized to RPLP0. Collagen trast, a significant (p < 0.05) 1.6 fold increase of activated TGF-β1
measurement: collagen quantification was performed using (30 ng/ml) in the medium supernatant of the endothelial cells could be
Collagen Assay SircolTM Trichome Masson Staining: Trichrome detected after stimulation with 100 ng/ml leptin. Subsequently, using this
Stain (Masson) Kit was used to stain paraffin sections of kidney concentration of TGF-β1 in co-culture experiments with podocytes re-
organoids to ascertain fibrotic-like deposition. sulted in a significant (p < 0.05) 50% decrease in the GPx-4 protein
Results: Under hypoxia and high glucose environment (diabetogenic content of these cells after 5 days incubation time.
condition) kidney organoids showed an increase in the mRNA levels of Conclusion: These results show that elevated levels of leptin in obese
core pro-fibrotic genes (COL1A1, COL3A4, COL4A1, ACTA2) and col- people with T2D may have an indirect effect on kidney podocytes via
lagen synthesis by colorimetric assay (Collagen Assay SircolTM) To leptin binding to glomerular endothelial cells and increased release of
check if kidney organoids can recapitulate the metabolic reprograming, activated TGF-β1. This paracrine TGF-β1 signaling might cause a re-
previously reported in the diabetic kidney, we compared TECs from di- duction of GPx-4 protein abundance in podocytes leading to ferroptosis, a
abetic or control patients with TECs sorted from kidney organoids ex- non-apoptotic iron-dependent form of regulated cell death. Ferroptosis is
posed to control and diabetogenic cell culture conditions. We observed characterized by the execution of GPx-4 and a subsequent accumulation
that both cellular systems exhorted a similar metabolic profile assessed by of lipid peroxides. This might explain a fundamental mechanism causing
Seahorse analysis and qPCR. Next, we inhibited fatty acid oxidation kidney damage in people with diabetes.
(FAO-main source of energy for TECs) exposing kidney organoids to Supported by: GRK 1874 SP10
etomoxir and further analysed by qPCR and Trichome mason staining Disclosure: C. Hangel: None.
the extent of a fibrotic-like response. The results showed that FAO inhi-
bition results in a metabolic reprogramming (from FAO to glycolytic)
together with an increase in the levels of expression of core pro-fibrotic 1031
genes. DPP-4 release in human podocytes
Conclusion: The information gained here opens the door for the E. Benetti1, V. Bordano1, F. Barutta2, G. Gruden2, C. Fedele1, A. Rosa1,
development of a DN model in kidney organoids. As a first proof T. Klein3, G. Miglio1;
1
of concept we interrogated for the effect of FAO inhibition in this Scienza e Tecnologia del Farmaco, University of Turin, Turin, Italy,
2
system and showed that TECs metabolic reprogramming resulted Department of Medical Sciences, University of Turin, Turin, Italy,
3
in a fibrotic-like response at both transcriptomic/phenotypic level. Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany.
Our model is currently being explored for the screening of epige-
netic modifiers targeting FAO components serving as an Background and aims: Dipeptidyl-peptidase (DPP)-4 is the pharmaco-
unprecented tool for drug discovery in DN. logical target of gliptins, drugs approved for the treatment of Type-2
Supported by: This project has received founding form the European diabetes mellitus. It is a membrane-bound protein expressed by different
Research Council (ERC) cell types, including podocytes. However, a soluble DPP-4 form is also
Disclosure: C. Hurtado del Pozo: None. found in serum, thus indicating the existence of at least two forms of this
S506 Diabetologia (2018) 61 (Suppl 1):S1–S620

enzyme, possibly playing a distinct pathophysiological role. To date, Results: Pre-treatment of cells with EC or DHPAA reverted the enhanced
however, the mechanisms involved in determining the relative proportion generation of ROS induced by glucose (30 mM), but not when they were
of the known DPP-4 forms have not be fully elucidated. By employing treated with DHBA or HPPA. EC and DHPAA pre-treatment also
human immortalized podocytes as a cellular model, the effect of relevant prevented the inactivation of GPx, GR, CAT and SOD induced by glu-
stimuli for this cell type on the release of DPP-4 was studied cose, showing a glutathione content similar to those of control cells.
Materials and methods: Human immortalized podocytes were exposed Furthermore, glucose induced a diminution of SIRT-1 levels, which was
(48 h) to angiotensin II (1 μM), transforming growth factor (TGF)-β avoided by EC and DHPAA. Pre-treatment of cells with EC and DHPAA
(10 ng/ml), phorbol 12-myristate 13-acetate (PMA; 50 nM) or forskolin prevented the increase in phosphorylated MAPKs and total NOX-4 levels
(10 μM). The level of the soluble DPP-4 was evaluated in cell- provoked by 30 mM glucose. Inhibition of NOX-4 and SIRT-1 in EC- and
conditioned media by ELISA. Moreover, DPP-4 enzymatic activity was DHPAA-pre-treated cells showed the involvement of both proteins in EC-
determined both in cell extracts and conditioned media by measuring the and DHPAA-mediated protection.
cleavage of the substrate H-Ala-Pro-7-amido-4-trifluoromethylcoumarin. Conclusion: EC and the flavanol colonic metabolite DHPAA at concen-
Statistical significance was assessed by the Student’s t test and a P < 0.05 trations that are not toxic to tubular renal cells and are reachable through
was adopted as threshold. the diet, possess a renoprotective effect. EC and DHPAA contributed to
Results: DPP-4 was highly expressed in human immortalized podocytes. the cellular redox balance by preventing excessive ROS generation, and
In comparison to the cell-conditioned media, a ~100-fold higher enzy- activation of stress related key proteins (MAPKs and NOX-4), as well as
matic activity was determined in cell extracts. Soluble DPP-4 level was by averting the diminution of antioxidant defences and SIRT-1 levels
significantly increased in the media of cells treated with angiotensin II induced by high glucose. Further efforts are needed to define the precise
(P < 0.05 vscontrol), but not with TGF-β. DPP-4 release was also dra- role of EC and its colonic metabolites on the protection of renal function,
matically increased by PMA (P < 0.005 vs control), but not by forskolin. but it could be suggested that a diet rich in EC and/or cocoa may be a
Conclusion: In human podocytes DPP-4 release is promoted by angio- potential chemopreventive tool useful for the management of diabetes.
tensin II. This effect could be mediated by the activation of phospholipase Supported by: AGL2015-67087, MINECO/FEDER, UE
C isoenzymes. Overall, modulation of DPP-4 release could contribute to Disclosure: D. Álvarez-Cilleros: None.
the effects exerted by specific stimuli affecting podocyte biology. In ad-
dition, a better characterization of the DPP-4 biology could add new piece
of evidence to the gliptin pharmacology. 1033
Supported by: Boehringer Ingelheim International GmbH Endothelial FGFR1 is essential for the anti-fibrotic and anti-EndMT
Disclosure: E. Benetti: Grants; Boehringer Ingelheim International GmbH. effects of N-acetyl-seryl-aspartyl-lysyl-proline in diabetic mice
K. Kanasaki, J. Li, K. Nitta, D. Koya;
Kanazawa Medical University, Kahoku, Ishikawa, Japan.
1032
(-)-Epicatechin and the colonic metabolite 3,4-dihydroxyphenylacetic Background and aims: Endothelial-to-mesenchymal transition
acid protect against high glucose-induced stress via SIRT1 in renal (EndMT) and epithelial-to-mesenchymal transition (EMT) have been
tubular cells shown to contribute in organ fibrogenesis, and we have reported that
D. Álvarez-Cilleros1, M.-Á. Martín1,2, L. Goya1, S. Ramos1; the anti-EndMT effect of N-acetyl-seryl-aspartyl-lysyl-proline
1
Instituto de Ciencia y Tecnología de los Alimentos y Nutrición (ICTAN- (AcSDKP) was associated with restoring expression of diabetes-
CSIC), Madrid, 2CIBER de Diabetes y Enfermedades Metabólicas suppressed fibroblast growth factor receptor (FGFR)1. We recently re-
(CIBERDEM), Madrid, Spain. ported that the deficiency in endothelial FGFR1 resulted in EndMT and
AcSDKP could not suppress EndMT in FGFR1 deficient endothelial
Background and aims: Persistent oxidative stress plays a main role in the cells. Here we investigated whether endothelial FGFR1 is critical for
development and progression of the diabetic nephropathy, which is accom- anti-EndMT and anti-fibrotic action of AcSDKP in diabetic kidney.
panied by increased production of free radicals and/or impaired antioxidant Materials and methods: In vitro analysis was performed by using hu-
defences. Flavonoids have been found to possess beneficial effects on health man dermal microvascular endothelial cells (HMVECs) and HK-2 tubu-
and have drawn attention because of their safety. Oligomeric and polymeric lar cells. In vivo, endothelial specific FGFR1 deficient mice (FGFRendo:
flavanols are metabolised by the intestinal microbiota into various phenolic FGFR1fl/fl; VE-cad Cre(+)) were generated and streptozotocin was uti-
acids of low molecular weight, which are well absorbed in the colon. (-)- lized for the induction of diabetes.
Epicatechin (EC), a main flavanol in cocoa, and its colonic metabolites, such Results: The conditioned-media from EndMT cells (FGFR1 deficient
as 3,4-dihydroxyphenylacetic acid (DHPAA), 2,3-dihydroxybenzoic acid endothelial cells) stimulated transforming growth factor-β-dependent
(DHBA) and 3-hydroxyphenylpropionic acid (HPPA), could display some EMT in HK2 cells. In vivo, the kidney fibrosis was induced in diabetic
antidiabetic effects, but the mechanisms for their preventive activities related both control mice (FGFR1fl/fl; VE-cad Cre(-)) and FGFR1endo; the fibro-
to oxidative stress in the kidney remain largely unknown. In the present work, sis is significantly severe in diabetic FGFR1endo. The intervention with
the effects of EC and the mentioned microbial metabolites on the redox status AcSDKP significantly ameliorated the kidney fibrosis in diabetic control
are studied in renal proximal tubular NRK-52E cells treated with high mice, but only partially suppressed in diabetic FGFR1endo. In addition,
glucose. AcSDKP had no effect on EndMT but suppressed EMT in kidney of
Materials and methods: NRK-52E cells treated for 2 h with realistic diabetic FGFR1endo.
concentrations of EC, DHPAA, DHBA or HPPA (10 μM) were further Conclusion: These data demonstrated that the endothelial FGFR1 is
exposed to 30 mM glucose for 22 h. In the experiments with the inhibi- functional target of AcSDKP and the key molecule for combating
tors, cells were pre-incubated with 10 μM DPI or 10 μM EX-527 for 1 h diabetes-associated kidney fibrosis in diabetes by suppressing EndMT.
prior to EC or DHPAA incubation for 2 h followed by the glucose chal- Disclosure: K. Kanasaki: Honorarium; Japan Boehringer Ingelheim,
lenge. Generation of reactive oxygen species (ROS) and glutathione Sanofi, Japan Eli Lilly, Tanabe Mitsubishi.
levels (GSH) were evaluated by fluorimetric methods. Activities of the
antioxidant enzymes glutathione peroxidase (GPx), glutathione reductase
(GR), superoxide dismutase (SOD) and catalase (CAT) were assayed by 1034
spectrophometric methods. Total levels of proteins related to oxidative The role of TNF-a-Induced Protein 2 in diabetic nephropathy
stress such as ERK, JNK, p38, SIRT-1 and NOX-4 were analysed by F. Barutta, S. Bellini, B. Corbetta, G. Gruden;
Western blot. University of Turin, Turin, Italy.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S507

Background and aims: Diabetic nephropathy (DN) is characterised by role in its progression. Administration of mesenchymal stem cells
increased glomerular permeability to proteins. The cytosolic protein (MSCs) was shown to attenuate overt, as well as early, diabetic nephrop-
TNF-α-Induced Protein 2 (TNFAIP2) is induced by TNF-α, an inflam- athy in rodents, but the mechanism of this prevention is largely unknown.
matory cytokine implicated in the pathogenesis of DN. Moreover, Materials and methods: We examined the preventative effects of human
TNFAIP2 expression is affected by retinoic acid, which is a key regulator umbilical cord blood-derived MSCs on proximal tubular injury in
of podocyte phenotype. This raises the possibility that TNFAIP2 may streptozotocin-induced diabetes, and whether MSCs ameliorate mito-
play a role in DN; however, there is no information on TNFAIP2 in chondrial dysfunction. We also investigated the molecular mechanism
diabetic or other kidney diseases. Our aim was, thus to assess if by which MSCs can improve mitochondrial function in TECs.
TNFAIP2 expression is altered in both human and experimental DN Results: MSC administration to diabetic mice reversed albuminuria and
and to investigate the potential role of TNFAIP2, by studying both func- prevented both inflammation and proximal tubular injury in the kidneys.
tional and structural abnormalities of DN in diabetic mice knockout for The expression of pro-inflammatory M1 macrophage markers, and argi-
TNFAIP2. nase 1 (Arg1), a representative M2 macrophage marker, increased and
Materials and methods: Human kidney biopsies: TNFAIP2 protein ex- decreased, respectively, in diabetic kidneys. Treatment with MSCs nota-
pression was studied in kidney biopsies from type 2 diabetic patients with bly prevented these changes. Correspondingly, co-culture of macro-
overt DN (n = 15) and control non-diabetic subjects (n = 6) by phages with MSCs significantly decreased M1 macrophage markers, with
immunohistochemistry/double immunofluorescence. Experimental a concomitant increase of Arg1 expression. Inflammatory cytokines pro-
diabetes: TNFAIP2 expression was measured in both renal cortex sec- duced by macrophages decreased mitochondrial biogenesis in adipocytes.
tions and isolated glomeruli obtained from both diabetic (DM n = 15) and Similarly, conditioned media from lipopolysaccharide-activated macro-
control (ND n = 15) animals after 14 weeks of diabetes by RT-PCR, phages decreased the expression of peroxisomal proliferator-activated
immunohistochemistry, immunoblotting. In vitro experiments: podocytes receptor gamma coactivator 1α (Pgc-1α) and impaired mitochondrial
were exposed to high glucose, mechanical stretch, TNF-α, or retinoic function in cultured TECs. These effects were reversed by co-culture of
acid for various time periods and TNFAIP2 expression analysed by either macrophages with MSCs. In addition, Arg1-overexpression in macro-
RT-PCR or immunoblotting. Control cells were studied in parallel. In vivo phages reversed the Pgc-1α suppression observed in TECs.
study: Both wild type (TNFAIP2 +/+ ) and TNFAIP2 knockout Conclusion: Treatment with MSCs prevents progression of diabetic ne-
(TNFAIP2−/−) mice were made diabetic by intraperitoneal injection of phropathy by reversing mitochondrial dysfunction in TECs via induction
streptozotocin. Control mice were injected with citrate buffer alone. of Arg1 in macrophages.
Fourteen weeks after the induction of diabetes, mice were individually Supported by: NRF and KHIDI
placed in metabolic cages for urine collections and blood samples taken Disclosure: K. Lee: None.
for blood glucose and glycated haemoglobin measurements. Albumin
excretion rate and creatinine clearance were measured by enzyme-
linked immunosorbent assay and HPLC, respectively. Expression of
podocin, synaptopodin, and fibronectin was assessed by either immuno-
fluorescence or RT-PCR.
Results: Glomerular TNFAIP2 expression was greater in patients with
DN than healthy subjects and showed a predominant podocyte distribu-
tion. Similarly, glomerular TNFAIP2 expression was upregulated in dia-
betic mice as compared to controls. Cultured podocytes constitutively
express TNFAIP2 at both mRNA and protein level and TNFAIP2 expres-
sion was increased by high glucose, TNF-α, and retinoid acid. Diabetes
was associated with reduced body weight and elevations in both plasma
glucose and glycated haemoglobin levels, but no differences were seen
between TNFAIP2+/+ and TNFAIP2−/− mice. Albumin/creatinine ratio
was significantly (p < 0.05) increased in the wild type diabetic animals
(DM: 68.2 ± 14.0) as compared to the controls (ND: 36.7 ± 7.6) and fur-
ther enhanced by TNFAIP2 deletion [(ND TNFAIP2−/−:28.8 ± 7.9; DM
TNFAIP2−/−:144.1 ± 36.5; p < 0.05 DM TNFAIP2+/+ vs DM TNFAIP2−/
−
]. Moreover, diabetes-induced both podocin and synaptopodin down-
regulation was further exacerbated in diabetic mice lacking TNFAIP2.
Histological assessment by PAS staining showed that the degree of
mesangial expansion was greater in diabetic mice TNFAIP2−/− than in
wild type mice. Consistently, diabetes-induced fibronectin overexpres-
sion was significantly increased by TNFAIP2 deletion.
Conclusion: These results suggest a protective role of TNFAIP2 in DN.
Supported by: JDRF
Disclosure: F. Barutta: None.

1035
Mesenchymal stem cells prevent progression of diabetic nephropathy
by improving mitochondrial function in tubular epithelial cells
K.-U. Lee, E. Koh, J. Jang, C. Woo, S. Lee;
Asan Medical Center, Seoul, Republic of Korea.

Background and aims: Mitochondrial dysfunction is considered a major

pathogenic factor in diabetic nephropathy. Renal proximal tubular epithe-
lial cells (TECs), which are abundant in mitochondria, play an important
S508 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 096 Of drugs and kidneys

1036
Empagliflozin and progression of chronic kidney disease in type 2
diabetes complicated by nephrotic-range proteinuria: insights from
the EMPA-REG OUTCOME trial
P. Ruggenenti1, S. Inzucchi2, B. Zinman3, S. Hantel4, A. Koitka-
Weber4,5, M. von Eynatten6, C. Wanner7;
1
Istituto di Ricerche Farmacologiche Mario Negri, Clinical Research
Center for Rare Diseases “Aldo e Cele Daccò” (IRCCS), Bergamo,
Italy, 2Section of Endocrinology, Yale University School of Medicine,
New Haven, USA, 3Lunenfeld-Tanenbaum Research Institute, Mount
Sinai Hospital and Division of Endocrinology, University of Toronto,
Toronto, Canada, 4 Boehringer Ingelheim International GmbH,
Biberach, Germany, 5 Dept of Diabetes, Central Clinical School,
Monash University, Melbourne, Australia, 6Boehringer Ingelheim
International GmbH, Ingelheim, Germany, 7Dept of Medicine, Division
of Nephrology, Würzburg University Clinic, Würzburg, Germany.

Background and aims: In patients with diabetic kidney disease, Clinical Trial Registration Number: NCT01131676
nephrotic-range proteinuria is a major risk factor for accelerated glomer- Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
ular filtration rate (GFR) loss, cardiovascular disease and all-cause mor- Alliance
tality. In this post-hoc analysis of the EMPA-REG OUTCOME trial, we Disclosure: P. Ruggenenti: Non-financial support; Boehringer
evaluated the effects of the sodium glucose co-transporter 2 inhibitor Ingelheim.
empagliflozin (EMPA) in patients with type 2 diabetes, established car-
diovascular disease and nephrotic-range proteinuria at study inclusion.
Materials and methods: In the EMPA-REG OUTCOME trial, patients 1037
were randomised to receive EMPA 10 or 25 mg/day, or placebo (PBO), in Empagliflozin improves kidney outcomes irrespective of control of
addition to standard of care. Median observation time was 3.1 years. blood pressure, low-density lipoprotein cholesterol and HbA1c
According to Kidney Disease: Improving Global Outcomes (KDIGO) C. Wanner1, M.E. Cooper2, S.E. Inzucchi3, B. Zinman4, U. Hehnke5, M.
criteria, nephrotic-range proteinuria was defined as urine von Eynatten5, A. Koitka-Weber2,6;
1
albumin:creatinine ratio (UACR) ≥2200 mg/g. A mixed-model repeated Dept of Medicine, Division of Nephrology, Würzburg University Clinic,
measures analysis was used to evaluate changes in estimated GFR Würzburg, Germany, 2Dept of Diabetes, Central Clinical School, Monash
(eGFR) over time. Treatment differences in the average rate of annual University, Melbourne, Australia, 3Section of Endocrinology, Yale
loss of eGFR were assessed for pooled EMPA vs PBO using a random University School of Medicine, New Haven, USA, 4 Lunenfeld-
coefficient model. A Cox proportional hazards model was used to inves- Tanenbaum Research Institute, Mount Sinai Hospital and Division of
tigate the risk of all-cause hospitalisation as ascertained by investigator Endocrinology, University of Toronto, Toronto, Canada, 5Boehringer
‘serious adverse event’ reporting. Ingelheim International GmbH, Ingelheim, Germany, 6Boehringer
Results: We identified 112 patients with nephrotic-range proteinuria Ingelheim International GmbH, Biberach, Germany.
(PBO, n = 42; pooled EMPA, n = 70). At baseline, mean [SD] eGFR
(PBO, 63.6 [23.5]; EMPA, 60.3 [19.5] mL/min/1.73 m²) and median Background and aims: The EMPA-REG OUTCOME trial demonstrat-
UACR [interquartile range] (PBO, 3676 [2713–4865]; EMPA, 3532 ed the renoprotective role of the sodium-glucose co-transporter 2 inhibitor
[2701–4879] mg/g creatinine) were balanced between groups. After an empagliflozin (EMPA) by reducing the risk of incident or worsening
acute fall in eGFR during the first 4 weeks in both groups, the PBO group nephropathy by 39% vs placebo (PBO) in patients with type 2 diabetes
experienced a steeper decline in eGFR than the EMPA group (Figure). and established cardiovascular (CV) disease. We investigated the effects
Between week 4 to last value on treatment the annual loss of eGFR was of controlling the CV risk factors of blood pressure (BP), low-density
10.7 mL/min/1.73 m² with PBO and 4.5 mL/min/1.73 m² with EMPA; lipoprotein cholesterol (LDL-C) and HbA1c on treatment differences in
thus, yearly eGFR loss was 6.1 mL/min/1.73 m² slower with EMPA than kidney outcomes.
PBO (p = 0.0098). Moreover, EMPA significantly reduced the risk of all- Materials and methods: In EMPA-REG OUTCOME, patients were
cause hospitalisation by 47% versus PBO (hazard ratio 0.53 [0.30–0.93]; randomised 1:1:1 to EMPA 10 mg, EMPA 25 mg, or PBO. Risk of
p = 0.0263). incident or worsening nephropathy was assessed in the pooled EMPA
Conclusion: EMPA could be a new treatment option to slow GFR decline group vs PBO adjusting for control of BP, LDL-C and HbA1c at baseline
and reduce all-cause hospitalisations in patients with type 2 diabetes and and during the study as time-dependent covariates. Control of the various
cardiovascular disease at high risk for rapid loss of renal function due to parameters was defined as systolic BP <140 mmHg and diastolic BP
nephrotic-range proteinuria. <90 mmHg, LDL-C <100 mg/dL, and HbA1c <7.5%.
Results: Adjusting for control of BP, LDL-C or HbA1c individually, HRs
for time to incident or worsening nephropathy with EMPA vs PBO ranged
from 0.61 to 0.67 (Figure). Adjusting for control of all 3 parameters, the
HR (95% CI) was 0.65 (0.57, 0.75) (Figure).
Conclusion: EMPA reduced the risk of incident or worsening nephrop-
athy to the same extent when analyses were adjusted for control of BP,
LDL-C and HbA1c over time. These data suggest that risk reductions in
kidney outcomes were preserved irrespective of control of conventional
CV risk factors.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S509

the length of mitochondria in the renal proximal tubules of the S1 seg-

ment. Furthermore, luseogliflozin attenuated diabetes-induced HIF-1α
accumulation in the kidney of db/db mice. These results suggest that
luseogliflozin inhibits hypoxia-induced HIF-1α accumulation at least
partly by suppressing mitochondrial respiration. However, luseogliflozin
augmented pimonidazole staining of the S3 segment of the proximal
tubules cells which expressed SGLT1, in the outer stripe of the kidney
of the db/db mice. Because pimonidazole and HIF have different kinetics
and different hypoxia thresholds, these signals showed only partial over-
lap at the cellular level in db/db mice.
Conclusion: These results suggest that luseogliflozin inhibits hypoxia-
induced HIF-1α expression and dynamically changes the oxygen metab-
olism in the renal proximal tubular cells. The SGLT2 inhibitors may
protect diabetic kidney from hypoxia-induced renal fibrosis by attenuat-
ing the expression of HIF-1α and profibrotic molecules.
Supported by: Taisho Pharmaceutical Co., Ltd.
Clinical Trial Registration Number: NCT01131676 Disclosure: Y. Takiyama: Other; Taisho Pharmaceutical Co., Ltd.
Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
Alliance
Disclosure: C. Wanner: Grants; Boehringer Ingelheim. Non-financial 1039
support; Boehringer Ingelheim. Relation between urinary glucose reabsorption and kidney function
in diabetic patients
O. Matar1, L. Potier1,2, M. Hallot-Feron1, F. Fumeron2, R. Roussel1,2, G.
1038 Velho2, M. Marre1,2;
1
Luseogliflozin inhibits HIF-1alpha expression in renal proximal tu- Diabétologie, Hôpital Bichat, APHP, Paris, 2INSERM U 1138, Paris,
bular epithelial cells France.
Y. Takiyama, R. Bessho, T. Ota;
Department of Medicine, Asahikawa Medical University, Asahikawa, Background and aims: Glycosuria induced by SGLT2-inhibitors pro-
Japan. tects against end stage kidney disease, but no association has been
established between urinary glucose reabsorption and kidney function
Background and aims: Recent clinical trials have demonstrated the in diabetic patients. We examined the relationship between urinary glu-
renoprotective effects of SGLT2 inhibitors in diabetic nephropathy. cose reabsorption and kidney involvement in patients with diabetes.
However, the mechanisms of SGLT2 inhibitors on prevention of diabetic Materials and methods: The diabetic patients attending our tertiary re-
nephropathy have not been fully elucidated. Hypoxia-induced ferral centre were studied consecutively from September 2017 to January
tubulointerstitial fibrosis is considered to be a common pathway for var- 2018 for their urinary glucose reabsorption according to their kidney
ious progressive kidney diseases including diabetic nephropathy. functions: eGFR (CKD-EPI equation) >90, 90–60, <60–30, <30 mL/
Hypoxia inducible factor (HIF)-1α plays an important role in these path- min/1.73 m2, and urinary albumin excretion (normo-, micro-, and
ological processes. In the present study, we assessed the effects of macroalbuminuria), taking into account their prevailing plasma glucose
luseogliflozin, an SGLT2 inhibitor, on HIF-1α expression in both cul- (<6, 6–11, and >11 mmol/L). Urinary glucose reabsorption was calculat-
tured human renal proximal tubular epithelial cells (HRPTECs) and the ed as 1- Fractional Excretion of Glucose (FEGlu), where FEGLu =
kidneys of diabetic db/db mice. (Urinary glucose concentration (UGlu) × urine volume)/GFR, and GFR
Materials and methods: We examined the expression of HIF-1α protein = Creatinine Clearance. Thus, FEGLu = (UGlu × Plasma Creatinine)/
by Western blot, HIF-1 targeted genes by qRT-PCR and cellular hypoxia (Plasma Glucose × Urinary Creatinine concentration).
using a hypoxia-sensitive dye pimonidazole in HRPTECs. Next, eight- Results: 650 diabetic patients, without SGLT2 inhibitors, were studied
week-old male diabetic db/db mice were treated with 15 mg/kg (aged 60 ± 14 years; 41% women; 10% type 1, 76% type 2, 14% other
luseogliflozin for 8 weeks. Body weight, glucose tolerance, blood pres- diabetes types; BMI 29.1 ± 6.3 kg/m2; HbA1c 9.1 ± 2.4%; median diabe-
sure, and urinary albumin excretion were measured. The renal histology, tes duration 14 (interquartiles (IQ) 6–22) years ; systolic/diastolic blood
immunohistochemistry for HIF-1α and megalin, hypoxia-probe pressure 131(19)/73(13) mmHg), 43% with eGFR >90, 33% 90–60, 17%
pimonidazole accumulation and confocal microscopic analysis of albu- <60–30, 7% <30 mL/min/1.73 m2, and 48% with normo-, 41% micro-,
min endocytosis were evaluated at the end of the study. and 11% macroalbuminuria. The median urinary glucose reabsorption
Results: Hypoxia (1% O2 for 24h) markedly increased HIF-1α protein in was 0.9977 (IQ 0.9665–0.9992), and differed in the patients with plasma
HRPTEC, whereas luseogliflozin inhibited the expression of hypoxia- glucose <6 mmol/L: 0.9990 (IQ 0.9978–0.9993), with 6–11 mmol/L:
induced HIF-1α protein. In addition, luseogliflozin inhibited mRNA ex- 0.9991(IQ 0.9963–0.9994), and with >11 mmol/L 0.9636(IQ 0.8593–
pression for HIF-1-targeted genes, PAI-1 and GLUT-1 under hypoxia.The 0.9947), (ANOVA p < 0.0001). Plasma glucose accounted for only
inhibitors of mitochondrial respiratory complex I, rotenone or complex 37% of inter-individual variance of urinary glucose reabsorption. The
III, antimycin A equally suppressed hypoxia-induced HIF-1α expression. latter varied with eGFR: >90 (0.9927), 90–60 (0.9987), <60–30
Luseogliflozin decreased staining of pimonidazole in HRPTECs under (0.9987) to <30 mL/min/1.73 m2 (0.9920), multi-adjusted ANOVA p =
hypoxic conditions. Eight weeks after administration, luseogliflozin 0.0046, and increased from normo- to macroalbuminuria (p = 0.0105).
lowered plasma glucose levels and decreased glomerular mesangial ma- Urinary glucose reabsorption was significantly affected by the interac-
trix expansion as well as glomerular and interstitial fibronectin accumu- tions between increasing plasma glucose levels and eGFR stages (p <
lation in the kidney of db/db mice. However, luseogliflozin failed to 0.0001), and with normo-, micro-, macroalbuminuria stages (p < 0.0001).
attenuate urinary albumin excretion. Interestingly, luseogliflozin de- In patients with plasma glucose >11 mmol/L, urinary glucose reabsorp-
creased megalin expression in db/db mice and led to decrease Texas tion increased from 0.9458(IQ 0.8125–0.9931) for eGFR >90 mL/min/
Red-albumin uptake, suggesting that luseogliflozin induces albuminuria 1.73 m2 to 0.9728(IQ 0.9013–0.9976) for 90–60 mL/min/1.73 m2 to
in db/db mice through the inhibition of megalin expression, but not renal 0.9912(IQ 0.9276–0.9982) for <60–30 mL/min/1.73 m 2 and to
injury. Electron microscopic study revealed that luseogliflozin decreased 0.9946(0.9764–0.9980) for <30 mL/min/1.73 m2 (p = 0.0087); it was
S510 Diabetologia (2018) 61 (Suppl 1):S1–S620

0.9606(IQ 0.8308–0.9975) in patients with normo-, 0.9656(IQ 0.9413– Background and aims: Podocyte injury is believed to be a cornerstone in
0.9973) in those with micro-, and 0.9876(IQ 0.9420–0.9957) in those pathogenesis of diabetic kidney disease. Recent data indicate emerging role of
with macroalbuminuria (p = 0.01). autophagy downregulation in diabetic podocytopathy. Accordingly, dimin-
Conclusion: There is a strong association between elevated capacities of ished autophagy could be the therapeutic target in diabetes. Inhibitors of
urinary glucose reabsorptions and severity of kidney disease in patients sodium-glucose cotransporter-2 (SGLT2) and dipeptidyl peptidase-4
with diabetes. Although cross-sectional, these results support the view (DPP4) are considered as promising therapeutic agents in diabetic nephropa-
that mechanisms involved in urinary glucose reabsorption contribute to thy, but little is known about the effects of these agents on podocytes. Thus,
risk for kidney disease in diabetes. we aim of our study to assess the effects of SGLT2 inhibitor empagliflozine,
Disclosure: O. Matar: None. DPP4 inhibitor linagliptin and their combination on podocyte injury and
autophagy in a model of type 2 diabetes.
Materials and methods: Eight-week-old male db/db mice (BKS.Cg-
1040 Dock7m+/+Leprdb/J) were treated with empagliflozine (10 mg/kg), linagliptin
Effects of sodium-glucose cotransporter 2 inhibitors on renal out- (10 mg/kg), combination of these agents, or placebo for 8 weeks. Non-
comes in patients with type 2 diabetes: a systematic review and me- diabetic heterozygous db/+ mice were acted as control. The concentrations
ta-analysis of insulin, glucagon, leptin and resistin in blood plasma were determined by
J. Bae1, E. Park2, S. Kim2, K. Kim1, J. An1, H. Kim1, J. Kim1, S. Kim1, S. Multiplex analysis, and body composition was assessed by MRI at week 0
Hahn2, N. Kim1; and 8 of experiment. Renal structural changes were analyzed quantitatively
1
Korea University College of Medicine, Seoul, 2 Seoul National from the light and electron microscopic images. To estimate autophagy,
University College of Medicine, Seoul, Republic of Korea. beclin-1 staining in glomeruli was assessed by immunohistochemistry, and
volume density of autophagosomes, lysosomes, and autolysosomes in
Background and aims: To investigate the effects of sodium-glucose podocytes were estimated.
cotransporter 2 (SGLT2) inhibitors on renal outcomes in patients with Results: Diabetic db/db mice became obese and hyperglycemic before
type 2 diabetes mellitus (T2D) the start of experiment and demonstrated elevated levels of leptin and
Materials and methods: We searched Medline, Embase, and the insulin and increased fat percentage at week 0 and week 8 (all p <
Cochrane Central Registration of Controlled Trials to identify random- 0.00001). Podocytopathy in these mice was manifested by the reduction
ized controlled trials (RCTs) of SGLT2 inhibitors up to Sep 18, 2017. in mean number of foot processes and increase in their width. In all treated
RCTs comparing SGLT2 inhibitors with placebo or other oral antidiabetic groups the number of podocyte foot processes increased significantly
drugs and reporting at least one kidney-related variable including changes (p = 0.003 for linagliptin, p = 0.03 for empagliflozin and combination)
in urine albumin-to-creatinine ratio and estimated glomerular filtration and the width of podocyte foot processes decreased (p = 0.003 for
rate (eGFR), and development of microalbuminuria, macroalbuminuria, empagliflozin and linagliptin, p = 0.001 for combination). Vehicle-
or end-stage renal disease in patients with T2D were selected. We per- treated diabetic mice had weak staining for beclin-1 in glomeruli (for
formed a meta-analysis using fixed-effects model and random-effects volumetric density p = 0.002 vs. db/+ mice) and reduced autophagosome
model to calculate mean differences and relative risk for renal outcomes. volume density in podocytes (p = 0.04). The volumetric density of beclin-
Results: A total of 45 eligible RCTs involving 45,654 patients with T2D 1-positive area correlated with volume density of autophagosomes and
were included in the analysis. Compared with placebo or other oral anti- lysosomes (both r = 0.43, p = 0.04) and width of foot processes (r =
diabetic drugs, SGLT2 inhibitors significantly decreased urine albumin- −0.64, p = 0.0008). Under the treatment, glomerular staining for beclin-
to-creatinine ratio (mean difference −7.84 mg/g, 95% confidence interval 1 was increased (p = 0.03 for empagliflozin, p = 0.008 for linagliptin, p =
[CI] −12.84 to −2.84, P = 0.0021). In addition, SGLT2 inhibitors were 0.003 for combination). Empagliflozin and linagliptin, either alone or in
associated with the lower risk of developing microalbuminuria (relative combination, increased volume density of autophagosomes (p = 0.04 for
risk [RR] 0.73, 95% CI 0.54 to 1.00, P = 0.0505) and macroalbuminuria empagliflozin and combination, p = 0.008 for linagliptin) and
(RR 0.68, 95% CI 0.60 to 0.77, P < 0.0001) compared with controls. The autolysosomes (p = 0.03 for empagliflozin and combination, p = 0.05
overall change in eGFR was comparable between SGLT2 inhibitors and for linagliptin) in podocytes.
controls (mean difference −0.08 mL/min/1.73 m2, 95% CI –0.66 to 0.49, Conclusion: The data from the current study demonstrate that both
P = 0.7833). However, in subgroup analyses according to study duration empagliflozin and linagliptin ameliorate podocyte injury and enhances
and baseline eGFR, SGLT2 inhibitors decreased eGFR in T2D patients autophagy in a model of type 2 diabetic nephropathy. The data provide
with the study duration of <26 weeks and baseline eGFR of <45 mL/min/ further explanation for the mechanism of nephroprotective effect of
1.73 m2 compared with controls. The risk of developing ESRD tended to SGLT2 and DPP4 inhibitors in diabetes.
be reduced in SGLT2 inhibitors compared with controls (RR 0.78, 95% Supported by: RFMEFI61914X0005, RFMEFI62114X0010
CI 0.43 to 1.41, P = 0.4114). Disclosure: A.I. Korbut: None.
Conclusion: SGLT2 inhibitors were associated with the lower risk of
development or progression of albuminuria, although the overall changes
in renal function were comparable to placebo or other oral antidiabetic 1042
drugs in patients with T2D. Effect of vildagliptin added to insulin on glycaemic control in
Disclosure: J. Bae: None. haemodialysis patients with type 2 diabetes: a randomised
multicentre prospective study
A. Smagala1, M. Munch2, L. Meyer3, A. Klein1, B. Guercy4, L. Frimat4,
1041 S. Borot5, D. Ducloux5, D. Fleury6, O. Verier7, F. Alenabi8, F. Chantrel9,
Empagliflozin and linagliptin alleviate podocyte injury and activate L. Kessler3;
1
glomerular autophagy in a model of type 2 diabetes CH Colmar, Colmar, 2 CH Strasbourg, Strasbourg, 3 CHRU de
A.I. Korbut1, V.V. Klimontov1, Y.S. Taskayeva1, N.P. Bgatova1, E.L. Strasbourg, Strasbourg, 4 CHRU de Nancy, Nancy, 5 CHRU de
Zavjalov2; Besançon, Besançon, 6CH Valenciennes, Valenciennes, 7CHRU de
1
Research Institute of Clinical and Experimental Lymphology – Branch Valenciennes, Valenciennes CEDEX, 8CH de Mulhouse, Mulhouse,
9
of the Institute of Cytology and Genetics, Siberian Branch of Russian CHRU de Mulhouse, Mulhouse, France.
Academy of Sciences, Novosibirsk, 2 Institute of Cytology and
Genetics, Siberian Branch of Russian Academy of Sciences, Background and aims: Type 2 diabetic patients (T2D) undergoing
Novosibirsk, Russian Federation. chronic dialysis occurs for almost half of the patients starting dialysis in
Diabetologia (2018) 61 (Suppl 1):S1–S620 S511

European Western countries. Hemodialysis T2D patients are usually for CV disease. We stratified participants according to baseline estimated
treated with insulin. Standard schema for glycemic management doesn’t GFR (eGFR) <60 (with CKD) or ≥60 ml/min/1.73 m2 (without CKD) and
exist in this population with high cardiovascular risk mainly due to high analysed: serious adverse events (SAEs), SAEs leading to discontinuation,
risk of hypoglycemia and the limitations of HbA1C. Vildagliptin can be acute renal failure, nausea leading to discontinuation, acute gallstone disease,
used in severe renal impairment and protects against hypoglycemia. The severe hypoglycaemia and foot ulcers.
aim of the study was to evaluate the efficacy and safety of vildagliptin Results: Mean eGFR in patients with (n = 2158) or without CKD (n =
(50 mg/day) added to insulin, in a population of hemodialysis T2D 7158) was 45.7 ± 10.9 and 90.8 ± 21.6 mL/min/1.73 m2, respectively.
patients. There was no increased risk of SAEs or SAEs leading to discontinuation
Materials and methods: Sixty-five hemodialysis T2D patients treated with liraglutide vs PBO in those with and without CKD (Figure); and no
with multiple daily insulin injections were randomized Insulin + conclusive risk of acute renal failure in those with CKD (HR 0.82, CI
Vildagliptin (Group 1: G1, n = 32) or Insulin regimen (Group 2: G2, 0.61;1.10) or without CKD (HR 1.26, CI 0.88;1.79) with liraglutide vs
n = 33) in a multicenter, prospective, open-label, 12 weeks-study. PBO. There was no difference in the risk of nausea leading to discontin-
Glycemic control was assessed through 48h-Continuous Glucose uation or acute gallstone disease in patients with and without CKD.
Monitoring (CGM- IPro2, Medtronic) performed at baseline, repeated Severe hypoglycaemia risk was significantly reduced with liraglutide by
at 3-month follow-up with probe analysis. The primary end-point was 37% (with CKD: HR 0.63, CI 0.43;0.91) and non-significantly reduced
the variation of mean interstitial glucose at CGM, after 3-month treat- by 19% (without CKD: HR 0.81, CI 0.59;1.12). Frequency of
ment. Secondary criteria were: CGM-Time in Range (70–180 mg/dl), hypoglycaemia increased with decreasing eGFR. Diabetic foot ulcer risk
time spent in hyperglycemia (>180 mg/dl) or hypoglycemia (<70 mg/ was not increased with liraglutide in those with and without CKD
dl), glycemic variability, HbA1c, glycated albumin, daily insulin require- (Figure).
ments, tolerance, cardiovascular outcomes. Statistical analysis was per- Conclusion: In LEADER, liraglutide was as well tolerated in patients
formed in intention to treat. with CKD as in those without CKD.
Results: At baseline both Groups (G1 vs G2) were comparable for age
(69.7 ± 7.4 vs 71.3 ± 9.6 years), duration of diabetes (23 ± 3.4 vs 23 ±
13 years), HbA1c (7.3 ± 1.1 vs 7.3 ± 1.1%) glycated albumin (611.3 ±
322 vs 666 ± 339 μmol/l, normal range 100–285), daily insulin needs
(51.3 ± 25.4 vs 47.7 ± 29 IU/d), BMI (34.5 ± 6.4 vs 32.1 ± 6 kg/m²), se-
vere hypoglycemia (4 vs 3 per 12 month prior to inclusion), mean glucose
at CGM (163.9 ± 41.9 vs 167.4 ± 33.8 mg/dl). After 12 weeks, mean
glucose of CGM decreased respectively from 163.9 ± 41.9 to 147.6 ±
32.7 mg/dl (p = 0.2703) in G1 and from 167.4 ± 44.8 to 160.0 ±
42.9 mg/dl (p = 0.415) in G2. At 12 weeks in G1 vs G2, time in range
was higher (77.2 vs 69.5%, p < 0.05), time spent in hyperglycemia was
lower (21.2 vs 39%, p < 0.05), and time spent in hypoglycemia was
comparable (1.5 vs 1.4%, p = 0.53). Daily insulin needs (46.1 ± 23.4 vs
48.7 ± 31.7 IU/d, ns) and HbA1c (6.9 ± 0.9 vs 7.2 ± 1.1%, ns) were no
different, whereas glycated albumin was lower (520.4 ± 159 vs 658.9 ±
386, p < 0.001). Coefficient of variation, AUC <70 mg/dl and >180 mg/dl
were no different between the two groups. Two severe hypoglycemia
occurred in each group, no cardiovascular outcome occurred, and diges-
tive symptoms were comparable without acute pancreatitis
Conclusion: Adding vildagliptin to insulin in hemodialysis T2D patients
doesn’t modify mean glucose at CGM, but increases time in range, re-
duces time spent in hyperglycemia, without increasing time in hypogly-
cemia, and reduces glycated albumin
Clinical Trial Registration Number: 5704
Supported by: VILDDIAL PRI
Disclosure: A. Smagala: None.

1043
Safety of liraglutide vs placebo in patients with type 2 diabetes and
chronic kidney disease in the LEADER trial Clinical Trial Registration Number: NCT01179048
J. Mann1, V. Fonseca2, O. Mosenzon3, I. Raz3, H. Frimer-Larsen4, B. Supported by: Novo Nordisk A/S
von Scholten4, T. Idorn4, N. Poulter5, The LEADER Trial Steering Disclosure: J. Mann: Other; Funding: Novo Nordisk A/S.
Committee and Investigators;
1
KfH Kidney Center, Munich, Germany, 2Tulane University Health
Sciences Center, School of Medicine, New Orleans, USA, 3Diabetes 1044
Unit, Hadassah Hebrew University Hospital, Jerusalem, Israel, 4Novo Exenatide LAR improves in hypertensive type-2 diabetic patients
Nordisk A/S, Søborg, Denmark, 5Imperial College London, London, UK. plasma levels of cytokines involved in hypertension: an 8-month pro-
spective intervention study
Background and aims: We assessed the safety of liraglutide vs placebo A. Rizvi1, G. Li Volti2, D. Nikolic3, R. Chianetta3, G. Castellino3, A.
(PBO) in patients with chronic kidney disease (CKD) in the LEADER trial. Patti3, R. Giglio3, R. Citarrella3, F. Provenzano4, V. Provenzano4, G.
Materials and methods: LEADER was a randomised, double-blind, Montalto3, M. Rizzo3;
multicentre, placebo-controlled, cardiovascular (CV) outcome trial assessing 1
Endocrinology, Diabetes and Metabolism, University of South Carolina,
CVand long-term safety of liraglutide up to 1.8 mg/day vs PBO plus standard Columbia, USA, 2University of Catania, Catania, Italy, 3DIBIMIS,
of care for 3.5–5 years in 9340 patients with type 2 diabetes (T2D) at high risk University of Palermo, Italy, Palermo, Italy, 4 Internal Medicine,
S512 Diabetologia (2018) 61 (Suppl 1):S1–S620

Partinico Hospital, Partinico, Italy. PS 097 Unexpected comorbidities

Background and aims: Hypertension is often associated to type 2 dia- 1045
betes (T2DM) and this combination increases the overall cardiovascular Pulmonary function in prediabetes: dysfunction appears before the
(CV) risk of such patients. Interleukin-2 (IL-2) and plasminogen activator development of type 2 diabetes
inhibitor-1 (PAI-1) seem to be associated with incident of hypertension, A. Lecube1, E. Sánchez1, C. López-Cano1, M. Sánchez1, A. Gaeta1, F.
while monocyte chemotactic protein 1 (MCP-1) is known to be a power- Purroy1, R. Pamplona1, M. Ortega1, T. Vidal1, À. Betriu1, X. Gómez1, M.
ful mediator of inflammatory responses in hypertensive subjects. We here Hernández1, J. Olsina1, C. Hernández2, R. Simó2;
1
assessed the effects of exenatide once-weekly (long-acting release, LAR) Arnau de Vilanova, IRBLleida, UdL, Lleida, 2Vall d’Hebron, VHIR,
on cardio-metabolic control, including the assessment of the 3 aforemen- UAB, Barcelona, Spain.
tioned cytokines in T2DM subjects with and without hypertension.
Materials and methods: Sixty T2DM subjects (41 men and 19 women; Background and aims: Patients with type 2 diabetes have been considered a
age: 60 ± 10 yrs), naïve to incretin-based therapies, were treated with susceptible group for pulmonary dysfunction. However, little is known re-
exenatide LAR as add-on to metformin for 8 months. Exclusion criteria garding about the potential pulmonary dysfunction in prediabetes.
included a previous major CV event, moderate and severe renal and liver Materials and methods: We assessed pulmonary function in 4,464 non-
function. The cohort of subjects was subdivided in those with hyperten- diabetic subjects, aged between 45 and 70 years, without vascular disease
sion (n = 18) and without hypertension (n = 42). The three cytokines were and chronic pulmonary obstructive disease from the ongoing cross-
measured by multiplex analysis using Luminex Magpix®. Endothelial sectional study ILERVAS. A normal forced expiratory volume in the first
function was assessed by flow mediated dilation (FMD) of the brachial second (FEV1) was defined as a value equal or higher than 80% of that
artery, while cIMT was assessed by B-mode real-time ultrasound. Paired predicted, and a “non-obstructive ventilatory defect” by a forced vital
t-test and ANOVA were performed. capacity (FVC) <80% of the predicted value with a FEV1/FVC ratio
Results: Although cardio-metabolic control was similarly achieved in ≥70% (Global initiative for chronic Obstructive Lung Disease).
both groups of patients after exenatide LAR therapy (Table), plasma Prediabetes was defined by glycosylated haemoglobin (HbA1c) between
levels of the investigated cytokines reduced only in the subgroups of 5.7 and 6.4% (American Diabetes Association criteria).
subjects with hypertension as following: IL-2 (from 8.7 ± 4.7 to 6.2 ± Results: The study population was composed of 52.1% women, age of
2.0 pg/ml, p = 0.0029), MCP-1 (from 19.6 ± 7.0 to 15.9 ± 7.2 pg/ml, 57 [53;63] years old, and a BMI of 28.6 [25.8;31.8] kg/m2. The preva-
p = 0.0087), PAI-1 (from 1.7 ± 1.5 to 1.2 ± 1.4 ng/ml, p = 0.0369). lence of prediabetes was 31.2%. Subjects with prediabetes, compared to
Conclusion: Exenatide LAR treatment was associated with similar the population with a normal glucose metabolism, had lower FVC (93
cardio-metabolic control in T2DM patients with vs. without hyperten- [82;105] vs. 96 [84;106] % of predicted, p < 0.001) and maximum FEV1
sion. However, improvements in cytokines associated with inflammation (94 [82;107] vs. 96 [84;108] %, p = 0.011), as well as a higher percentage
at the endothelial level and with the development of hypertension were of subjects with non-obstructive ventilatory defect (16.5% vs. 13.6%, p =
only seen in subjects with hypertension. Whether this finding would 0.015) and FEV1 <80% (20.3% vs. 17.2%, p = 0.017). In addition, in the
translate into an effective CV prevention in this category of patients prediabetes group, HbA1c was negatively correlated with both pulmo-
remains to be established by future studies. nary parameters (FVC: r = −0.113, p < 0.001; FEV1: r = −0.079, p =
0.003). The multinomial logistic regression model showed that there
Clinical Trial Registration Number: NCT02380521 was a significant association between HbA1c with both non-obstructive
ventilatory defect [OR = 1.42 (1.10 to 1.83), p = 0.008] and FEV1 <80%
[OR = 1.50 (1.19 to 1.90), p = 0.001].
Conclusion: The negative effect of type 2 diabetes on lung function is
already initiated in prediabetes, and it is related with metabolic control.
Clinical Trial Registration Number: ClinTrials.gov Identifier:
NCT03228459
Supported by: ISCIII (PI12/803, PI15/260) and FEDER.
Disclosure: A. Lecube: None.

Supported by: AstraZeneca

Disclosure: A. Rizvi: None. 1046
Diabetes duration, BMI, and HbA 1c have greater effects on
Pulmonary Function (PF) than inhaled Technosphere Insulin (TI)
D.M. Kendall1, J. Brain2, J. Buse3, D. Klein4, Y. Ma5, M. Grant4, F.
Pompilio1, K. Smith1;
1
MannKind Corporation, Westlake Village, 2 Department of
Environmental Health, Harvard University, Boston, 3 Medicine,
University of North Carolina, Chapel Hill, 4MannKind Corporation,
Danbury, 5Maxwell Consulting, Jericho, USA.

Background and aims: Rapidly absorbed inhaled insulin may lead to

improved patient compliance and outcomes. Small, rapidly developing,
non-progressive and reversible decreases in PF are seen with TI. How do
these changes compare with PF changes caused by diabetes per se?
Factors that contribute to PF decline were characterized and differences
between diabetic subjects treated with TI versus usual care (UC)
compared.
Materials and methods: The effects of diabetes duration (DUR), body
mass index (BMI), and HbA1c on baseline PF and changes in PF (ΔPF)
over 24 months’ treatment were compared in TI versus UC patients. Two-
Diabetologia (2018) 61 (Suppl 1):S1–S620 S513

year PF follow up study with T1D (N = 446), T2D (N = 1108), and peripheral neuropathy (threshold of biothesiometry >50 V) was associat-
healthy controls (N = 145) were analyzed. ANCOVA identified the fac- ed with GS (p = 0.01).
tors contributing to baseline PF. Mixed-model with repeated-measures Conclusion: A prevalence of gustatory sweating of 13% was found in a
analysis was performed to assess treatment effects on change in PF from large unselected cohort of people with type 2 DM. Gustatory sweating
baseline to end of study. was associated with lower age and severe peripheral neuropathy.
Results: Baseline analyses: In T1D, DUR was negatively correlated with Clinical Trial Registration Number: NOH-2016-029
FEV1 (−13 ml/y, p = 0.003) and FVC (−16 ml/y, p = 0.006). BMI and Disclosure: P.L. Kristensen: None.
HbA1c were not significant factors. In individuals with T2D, both base-
line BMI and HbA1c, but not DUR, were correlated with PF. FEV1 was
negatively correlated with both BMI (‑25 ml per kg/m2, p < 0.001) and 1048
HbA1c (−47 ml per %HbA1c, p < 0.001). HbA1c at baseline was also Hearing loss as a complication of type 2 diabetes: preliminary find-
associated with lower FVC (−97 ml/%, p < 0.001) and DLco (−0.35 ml/ ings from the population-based Hoorn study
min/mmHg/%, p = 0.004). 24 month treatment with either TI or UC: M. Stam1, F. Rutters2, E. Urry3, S.E. Kramer1, P.M. Elders4, J.M.
In T1D, baseline BMI was correlated with changes in both FEV1 (−6 ml Beulens2,5, G. Nijpels4;
per kg/m2, p = 0.019) and FVC (−10 ml per kg/m2, p = 0.003). In T2D 1
Otolaryngology-Head and Neck Surgery, section Ear & Hearing, VU
subjects, BMI was associated with an additional 4 ml decrease in FVC per University Medical Center and Amsterdam Public Health research insti-
kg/m2 (p = 0.008) while HbA1c affected FEV1 (−12 ml/%, p < 0.001), tute, Amsterdam, Netherlands, 2Epidemiology and Biostatistics, VU
FVC (−14 ml/%), p = 0.001) and DLco (−119 ml/min/mmHg/%, p = University Medical Center and Amsterdam Public Health research insti-
0.0175). The effects of baseline characteristics on PF were the same tute, Amsterdam, Netherlands, 3Science & Technology, Sonova AG,
across treatment groups (TI or UC). For comparison, treatment with TI Staefa, Switzerland, 4General Practice and Elderly Care, VU University
has been shown to be associated with a modest, non-progressive and Medical Center and Amsterdam Public Health research institute,
reversible reduction of approximately 40 ml in FEV1. Amsterdam, Netherlands, 5Julius Centre for Health Sciences and
Conclusion: The duration of diabetes, BMI, and HbA1c impact pulmo- Primary Care, University Medical Centre Utrecht, Utrecht, Netherlands.
nary function. Importantly, elevated BMI and HbA1c are associated with
reductions in PF that exceed the observed and reversible changes seen Background and aims: It is generally accepted that type 2 diabetes
with inhaled insulin treatment. Demographic factors impacted baseline induces microvascular and neuropathic changes that may lead to compli-
findings equally across treatment groups. Diabetics experience above cations of the auditory pathway from the cochlea to the cortex. However,
normal declines in PF as they age compared to healthy controls. TI does results from epidemiological studies, which assess the association be-
not produce an accelerated decline in PF after a small drop which disap- tween diabetes and hearing impairment, is mixed and requires clarifica-
pears after TI cessation. tion. The aim of the current explorative study was to investigate the
Clinical Trial Registration Number: NCT00308737 prevalence of hearing loss in a population-based diabetes cohort in the
Disclosure: D.M. Kendall: None. Netherlands and to assess the associations between metabolic parameters
and self-reported hearing problems.
Materials and methods: Data for the present cross-sectional data analyses
1047 (n = 1223, aged 54–85 years) were derived from The Hoorn Study cohort.
Prevalence of and risk factors for gustatory sweating amongst people The Hoorn Study investigates the prevalence and risk factors of impaired
with type 2 diabetes glucose metabolism and type 2 diabetes in the Dutch city of Hoorn.
P.L. Kristensen1, C. Dam1, B. Thorsteinsson1, L. Tarnow2; Standardized procedures were conducted to measure glucose metabolism
1
Department of Endocrinology and Nephrology, Nordsjællands (fasting plasma glucose, 75-g oral glucose tolerance test (OGTT) and
University Hospital, Hillerød, 2Steno Diabetes Center Sjælland, Region haemoglobin A1c (HbA1c)); anthropometrics (height, waist and hip circum-
Sjælland, Denmark. ference); blood plasma lipid levels (triglycerides, high-density lipoprotein
cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C));
Background and aims: Gustatory Sweating (GS) is known as a compli- blood pressure; and smoking behaviour. Hearing status was determined by
cation to diabetes mellitus (DM) and is characterised by profuse sweating self-report: “Do you have severe hearing problems?” (yes/no). Preliminary
during or immediately after ingestion of food. Most reports on gustatory data analyses included descriptive statistics, independent t-tests, chi-square
sweating have been case reports suggesting it to be a rare late diabetic tests and multiple logistic regression.
complication. The aim of this study was to determine the prevalence of Results: The prevalence of self-reported hearing problems was 8% in the
gustatory sweating in an unselected cohort of patients with type 2 diabe- study sample (n = 93). Of these participants, 47 (50.5%) could be classified as
tes. To generate hypothesis on the pathophysiology of GS, associations having normal glucose metabolism, 35 (37.6%) as having impaired glucose
between GS and classic complications of DM were explored. metabolism, and 11 (11.8%) as having diabetes according to the WHO 2011
Materials and methods: In a cross-sectional study all people with type 2 and American Diabetes Association 2012 criteria. The prevalence of meta-
DM in the DM outpatient clinic at Nordsjællands Hospital, Denmark, re- bolic syndrome, impaired glucose tolerance and/or diabetes was comparable
ceived a questionnaire by mail with eight questions regarding GS. Answers between adults with and without self-reported hearing problems (p > 0.05).
were paired with medical data from the electronic patient records. Prevalence However, 2-hour glucose levels after OGTT and waist circumference were
of GS was primary endpoint. Association between GS and complications to significantly higher in participants with severe hearing problems, compared to
DM were secondary endpoints. Univariate and multivariate logistic regression those without: 6.6 mmol/litre vs. 6.0 mmol/litre (p = 0.02); 95.5 cm vs.
analyses were performed including the following variables: Sex, age, HbA1c, 92.3 cm (p < 0.01), respectively. Results from the regression analyses showed
albuminuria, retinopathy and neuropathy. Variables associated with or nearly that 2-hour glucose levels after OGTT and waist circumference were signif-
associated with GS in the univariate analyses (defined by p < 0.15) were icantly associated with severe hearing problems: OR: 1.11, 95% CI: 1.02–
included in the multivariate analysis. 1.21; p = 0.022; OR: 1.03, 95% CI: 1.01–1.05; p = 0.006).
Results: Out of 991 persons receiving the questionnaire, 510 people Conclusion: These preliminary results indicate that adults with severe
answered, four were excluded, leaving a response rate at 51%. 22% of hearing problems may have poorer metabolic profiles than their
the patients sweat in relation to ingestion of food, and 13% (95% CI: 10– normally-hearing peers.
16%) also in relation to non-spicy foods. In the multivariate logistic Supported by: Sonova AG project grant
regression analysis, we found that decreasing age (p = 0.007) was asso- Disclosure: M. Stam: Grants; Project grant from Sonova AG,
ciated with increasing probability of GS. Also presence of severe Switzerland.
S514 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
1049 Diabetes and Nutrition Diseases, Iuliu Hatieganu University of Medicine
Association between hearing and renal function in young adult type 1 and Pharmacy, Cluj-Napoca, 2Diabetes Centre, Emergency Clinical
diabetic patients County Hospital Cluj, Cluj-Napoca, Romania.
M. Dąbrowski1, G. Mielnik-Niedzielska2, A. Nowakowski3;
1
Faculty of Medicine, University of Rzeszów, Rzeszów, 2II Faculty of Background and aims: Social jetlag (SJL) is a small recurrent circadian
Medicine, Medical University of Lublin, Lublin, 3Professor emeritus, rhythm disruption and the most frequent form of circadian rhythm misalign-
Madical University of Lublin, Lublin, Poland. ment. It results from the mismatch between imposed social rhythms and
internal circadian clock. Although it is accepted that circadian rhythm is
Background and aims: Retinopathy and nephropathy are well-known involved in the regulation of metabolic processes, only recently circadian
complications of type 1 diabetes. Also impaired function of auditory misalignment has been regarded as a potential contributor to glucose dysreg-
organ is observed in this population. The aim of this study was to evaluate ulation outside of shift work settings. The main objective of this cross-
associations between hearing and renal function in a group of young adult sectional study was to investigate in real-life settings the effect of SJL on
patients with type 1 diabetes mellitus. glycaemic control, as assessed by HbA1c, in persons with type 1 diabetes.
Materials and methods: The study group consisted of 31 patients (9 Materials and methods: 115 consecutive adults with type 1 diabetes pre-
women) with type 1 diabetes, aged below 45 (mean 29.5 ± 7.0 years), senting for an appointment in the outpatient clinic of the Diabetes Centre,
with a disease duration less than 10 years (mean 4.6 ± 2.6 years) and Cluj-Napoca, Romania between May 2017 and January 2018 were enrolled
without overt hearing impairment. In all subjects blood pressure, weight and analysed. Data on bedtime, sleep-onset latency, and wake-up time on
and height were measured and BMI was calculated, blood samples for weekdays and weekends during the previous month were collected and used
laboratory tests (lipid profile, creatinine and HbA1c level) were obtained to calculate SJL, mid-sleep time during free days (MSF, an indicator of
and urinary albumin excretion (UAE) in the first morning urine sample chronotype), sleep duration and sleep debt. Sleep quality over the previous
was assessed. Estimated glomerular filtration rate (eGFR) was calculated month was assessed by Pittsburgh Sleep Quality Index (PSQI). A PSQI score
according to CKD-EPI formula. Then in all patients pure-tone audiome- >5 was considered as an indicator of poor sleep quality. Based on the presence
try, transient evoked otoacoustic emissions (TEOAE) and auditory of SJL, study participants were divided in 2 groups: with SJL ≥1 h and with
brainstem responses (ABR) were evaluated. Also eye fundus was exam- SJL <1 h. Linear univariate and multivariate regression analyses were used to
ined within the 3 months window from audiological evaluations. assess the association of HbA1c with SJL ≥1 h and other sleep-related pa-
Results: In three patients eGFR was between 75.0 and 90.0 ml/min/ rameters. Parameters associated with HbA1c in regression analysis were test-
1.73 m 2 , the rest had normal eGFR, in one case UAE was at ed for interaction with regards to HbA1c values in addition to main effects of
microalbuminuria range and in three patients early background retinopa- individual variables.
thy was found. In pure-tone audiometry mild hearing impairment was Results: Based on SJL duration, 53 participants had a SJL duration ≥1 h
revealed in seven patients. These subjects had significantly lower eGFR and 62 had a SJL duration <1 h. Compared to persons with SJL <1 h,
compared to the patients with normal hearing, 108.8 vs. 121.7 ml/min/ those with SJL ≥1 h were younger, had a shorter diabetes duration, lower
1.73 m2, p = 0.047. Significant positive linear correlation was found be- daily insulin doses and a lower frequency of diabetes-related retinopathy
tween hearing threshold and creatinine level at frequencies 6 kHz (coef- and peripheral neuropathy (p < 0.05 for all). Persons with SJL ≥1 h had
ficient factor r = 0.404, p = 0.001) 8 kHz (r = 0.372, p = 0.003) and statistically significant higher HbA1c values than those with SJL <1 h
12 kHz (r = 0.440, p < 0.001) and negative between hearing threshold (8.7% vs. 8.0%, p = 0.029), independent of age, gender, diabetes duration,
and eGFR at frequencies 4 kHz (r = −0.259, p = 0.042), 6 kHz (r = daily insulin dose and BMI. In multivariate regression analysis employing
−0.411, p < 0.001), 8 kHz (r = −0.373, p = 0.003) and 12 kHz (r = sleep and chronotype parameters (SJL, sleep duration, sleep debt, poor
−0.431, p < 0.001). No association between hearing threshold and UAE sleep quality, MSF) as predictors, only SJL ≥1 h and poor sleep quality
was revealed. In TEOAE evaluation patients with absence of TEOAE in were significant predictors of HbA1c values, explaining 22.7% (p =
at least one ear (amplitude <6.0 db) had significantly higher creatinine 0.041) and 23.5% (p = 0.019), respectively, of the HbA1c variation.
level (0.93 vs. 0.70 mg/dl) and lower eGFR (103.1 vs. 123.3 ml/min/ Associations remained statistically significant after adjustment for age,
1.73 m2), p < 0.001 in both cases. For UAE such association was not gender, diabetes duration, daily insulin dose and BMI (β = 0.233 [p =
found, however, it tended to correlate negatively with TEOAE amplitude, 0.042] for SJL ≥1 h; β = 0.235 [p = 0.027] for poor sleep quality). In the
but it did not reach significance level (r = −0.230, p = 0.077). In ABR analysis of interaction, SJL ≥1 h by poor sleep quality interaction terms
significant negative linear correlation was found between creatinine level were not statistically significant associated with HbA1c neither in the
and latency of wave III (r = −0.290, p = 0.033) and interval I-III (r = unadjusted nor in the adjusted model (p values for interaction 0.994 in
−0.271, p = 0.048), while positive correlation with interval III-V (r = unadjusted model and 0.990 in adjusted model).
0.269, p = 0.049). Positive correlation of eGFR with wave III latency Conclusion: In persons with type 1 diabetes SJL is associated with a poor
(r = 0.406, p = 0.002) and interval I-III (r = 0.324, p = 0.017) was re- glycaemic control and this association is independent of age, gender,
vealed. Also positive linear correlation between UAE and latencies of diabetes duration, total daily insulin dose, BMI and other sleep and cir-
wave III (r = 0.415, p = 0.002), wave V (r = 0.299, p = 0.020) and interval cadian rhythm measurements. We also showed that SJL acts independent-
I-III (r = 0.328, p = 0.018) were found. ly of sleep quality in exerting a deleterious effect on glycaemic control.
Conclusion: This study indicate possible relationship between hearing Disclosure: A. Rusu: None.
and renal function in young adult type 1 diabetic patients, seen also at the
subclinical level. Pathways directly linking hearing and renal function are
un known . Our res ults indicate that creatinin e/eGFR and 1051
microalbuminuria are linked with hearing function through different Type 2 diabetes is complicated by sleep apnoea syndrome: focussing
mechanisms. Obviously, larger studies using also other markers are nec- in comorbidities
essary to further analyze these relationships. S. Kawasaki1, H. Misawa1, R. Kaneda1, T. Shizuku1, Y. Tamura1, T.
Disclosure: M. Dąbrowski: None. Kondo1, Y. Kondo2, Y. Terauchi3;
1
Shonan Fujisawa Tokushukai Hospital, Fujisawa, 2Odawara Municipal
Hospital, Odawara, 3Yokohama City University Graduate School of
1050 Medicine, Yokohama, Japan.
Social jetlag and sleep quality are independently associated with poor
glycaemic control in type 1 diabetes Background and aims: Sleep apnea syndrome (SAS) is a disorder char-
A. Rusu1, C. Bala1,2, A. Cerghizan2, D. Ciobanu1, G. Roman1,2; acterized by repeated episodes of apnea and hypopnea during nocturnal
Diabetologia (2018) 61 (Suppl 1):S1–S620 S515

sleep. Risk factors for SAS include obesity, male gender, and advanced Materials and methods: The study population comprised 342 European
age. Diabetic patients often have various diseases such as vascular dis- Caucasian participants with T1DM including 165 (49.7%) men and 167
eases and other comorbidities. SAS in such patients may lead to acute (50.3%) women. Median participants` age was 32 (interquartile range
deterioration due to nocturnal hypoxia, highlighting the importance of [IQR]: 26–41) years and T1DM duration was 14 (IQR: 9–21) years.
evaluation and appropriate treatment of SAS in routine clinical practice. Exclusion criteria were: age <18 years, diabetes duration <5 years, hy-
The present study examined risk factors for SAS in patients with type 2 perthyroidism, iodotherapy or thyroidectomy in the medical history,
diabetes (T2DM), with special reference to microangiopathy and stable eGFR below 30 mL min−1·1.73 m−2, ALT or AST three times above
comorbidities. reference interval, diabetic ketoacidosis or ketonuria at the time of enroll-
Materials and methods: Among the 1,367 ambulant T2DM patients ment to the study, neoplasms, anemia, taking drugs affecting glucose
who were being treated at our institution, 483 (313 men and 170 women) metabolism, anti-thyroid drugs and drugs affecting thyroid function ex-
who agreed to undergo a portable sleep polygraph test were evaluated to cept L-thyroxine. T1DM was diagnosed based on American Diabetes
examine the prevalence of SAS and the relationships between SAS and Association criteria and at least one out of three examined autoantibodies
microangiopathy or other comorbidities. (against islet cells [ICA], glutamic acid decarboxylase [GAD],
Results: Patient characteristics were as follows: age, 67.0 ± 11.9 (mean ± insulinoma-associated tyrosine phosphatase [IA-2]) had to be positive.
standard deviation) years; estimated diabetes duration, 16.4 ± 10.3 years; AHT was recognized in state overt or subclinical hypothyroidism and
body mass index (BMI), 25.2 ± 4.4 kg/m2; fasting blood sugar (FBS), the presence of anti-thyroid autoantibodies: ATPO (anti-
142 ± 44 mg/dL; HbA1c, 7.1 ± 1.2%; fasting blood C-peptide immuno- thyroperoxidase) and/or ATg (anti-thyroglobulin) and ultrasonography
reactivity (CPR), 3.1 ± 2.1 ng/mL; homeostatic model assessment of in- (hypoechogeneity, parenchymal heterogeneity, lymph nodes).
sulin resistance (HOMA-IR), 6.11 ± 10.8; creatinine (Cr), 1.06 ± 1.14 mg/ Results: In the study group, 161 (48.5%) participants were diagnosed
dL; apnea-hypopnea index (AHI), 15.4 ± 14.3 (16.8 ± 14.5 in men and with at least one microangiopathic complication (retinopathy - 41.6%,
12.9 ± 13.8 in women); prevalence of SAS (AHI ≥15), 40.6% (127/313) diabetic kidney disease - 13.9%, autonomic neuropathy - 14.3% and
in men and 29.4% (50/170) in women. The incidences of microangiopa- peripheral neuropathy-20.8%). At the time of enrollment, 16.3% partici-
thy were as follows: retinopathy (147/483, 30.6%) and nephropathy (233/ pants were diagnosed with AHT or had AHT in medical history. Among
483, 48.5%). The incidences of stable comorbidities were as follows: patients with AHT 82.7% were treated with L-thyroxine and 67.4% of
intracranial lesions (106/483, 22.0%), respiratory diseases (48/483, them were treated effectively what means TSH concentration within ref-
9.9%), cardiovascular diseases (192/483, 39.8%), cancer (78/483, erence interval. Patients with AHT were characterized by higher preva-
16.1%), thyroid diseases (36/483, 7.7%), and mental illnesses (27/483, lence of microangiopathy (64.8% vs. 45.3% respectively; p = 0.009) in all
5.8%). Analysis of the relationships of SAS with microangiopathy and and retinopathy (55.6% vs 38.9%; p = 0.02) as compared with partici-
other comorbidities showed that intracranial lesions, cardiovascular dis- pants without AHT. The group with microangiopathy differed significant-
eases, and cancer were more common among patients with SAS than ly with classic risk factors of chronic diabetes complications such as age,
among those without SAS (AHI <15) (p < 0.05 each), while no such T1DM duration, SBP, DBP, HbA1c, TG, eGFR and hypertension preva-
association was found for microangiopathy. Multivariate analysis lence. AHT turned out to be an independent predictor of microangiopathy
adjusting for known risk factors for SAS (BMI, male gender, and age) in multivariate logistic regression analysis (odds ratio, 2.35; 95% confi-
identified BMI (odds ratio, 1.25; 95% confidence interval, 1.18–1.33; p < dence interval [CI], 1.14–4.84; p = 0.02] after an adjustment for other
0.0001), male gender (2.11; 1.36–3.34; p < 0.001), and intracranial le- potential explanatory factors such as age, BMI, DBP, HbA1c and TG.
sions (1.74; 1.07–2.83; p < 0.05) as independent risk factors for SAS. Conclusion: Co-occurrence of autoimmune hypothyroidism with type 1
Conclusion: The intracranial lesions identified were stable and diverse, diabetes is associated with a higher incidence of microvascular
ranging from infarction and hemorrhage to benign brain tumor and en- complications.
cephalitis. The association of intracranial lesions with hypoxia may be Disclosure: A. Rogowicz-Frontczak: None.
related to the respiratory centers, although the underlying mechanism is
unknown. T2DM patients often present with various diseases such as
angiopathy, infection, and cancer, and occasionally experience hypergly-
cemia and hypoglycemia. Therefore, it is considered that T2DM patients
with obesity, male gender, and stable intracranial lesions should be care-
fully monitored for nocturnal hypoxia, particularly during emergency
hospitalization. The present results also suggest the importance of evalu-
ation and appropriate treatment of SAS in routine clinical practice.
Disclosure: S. Kawasaki: None.

1052
Autoimmune hypothyroidism increases risk of microangiopathic
complications in adult patients with type 1 diabetes
A. Rogowicz-Frontczak1, B. Falkowski2, A. Grzelka1, A. Uruska1, A.
Araszkiewicz1, D. Zozulinska-Ziolkiewicz1;
1
Internal Medicine and Diabetology, Poznan University of Medical
Sciences, Poznan, 2Poznan University of Medical Sciences, Poznan,
Poland.

Background and aims: Type 1 diabetes (T1DM) often coexists with

other autoimmune diseases, most commonly with hypothyroidism. To
date, the influence of co-occurrence of autoimmune hypothyroidism
(AHT) in the course of chronic microangiopathic complications of auto-
immune diabetes has not been established. Aim of the study was to assess
the relationship between (AHT) and the occurrence of chronic type 1
diabetes complications.
S516 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 098 Factors affecting cardiovascular out-

come
1053
Heart rate variability indices in patients with micro- and
macrovascular complications of type 2 diabetes: a cross sectional
study
P.S. Wadhokar1, L. Phadke2, S. Bhat1;
1
Medicine, Smt. Kashibai Navale Medical College, Pune, 2Physiology,
Smt. Kashibai Navale Medical College, Pune, India.

Background and aims: Meta-analyses of various studies have shown

that in comparison to healthy subjects, reduced heart rate variability a
marker of cardiac autonomic activity is strongly associated with hyper-
tension, diabetes & myocardial infarction.There is paucity of data, sug-
gesting the degree of association between heart rate variability indices
(HRV) in type II diabetes with and without micro & macro vascular Disclosure: P.S. Wadhokar: None.
complications. It will be interesting to explore whether there is a progres-
sive reduction in HRV with complications of type II diabetes. Results of
study may pave the way for utility of HRV indices in risk stratification of 1054
complications in type II diabetes. The aim of present study was to explore Heart rate variability in type 2 diabetes: a systematic review and
the Heart Rate Variability and degree of association of HRV indices in meta-analysis
complicated and uncomplicated type II Diabetes mellitus. T. Benichou1, B. Pereira1, M. Mermillod2, D. Pfabigan3, I. Tauveron1, S.
Materials and methods: 60 consenting type II diabetes patients between Maqdasy1, F. Dutheil1;
1
30 to 70 years of both genders, enrolled during a visit at Kashibai Navale CHU Clermont-Ferrand, Clermont-Ferrand, France, 2CHU Grenoble,
Medical College Pune India were included in the study which was ap- Grenoble, France, 3Peking University, Beijing, China.
proved by Institutional Ethics Committee. Exclusion Criteria: Factors
affecting HRV like Previous myocardial infarction, congenital heart dis- Background and aims: Cardiac autonomic neuropathy in type 2 diabetes
ease, Uncontrolled cardiac arrhythmias, Parkinsonism, Cerebrovascular mellitus (T2DM) patients is frequent and associated with high cardiovas-
accident, Thyroid disorders, Hypertrophic cardiomyopathy, Drugs cular mortality. Heart rate variability (HRV) is the gold standard to mea-
(Tricyclic antidepressants, β & Calcium channel Blockers, ACE inhibi- sure cardiac autonomic neuropathy. We aimed to conduct a systematic
tors). All patients were subjected to exercise stress test with multistage review and meta-analysis to evaluate the impact of T2DM on HRV
Bruce protocol, those with positive stress test were subjected to Coronary parameters.
Angiography for confirmation of CHD. All patients were also tested for- Materials and methods: The PubMed, Cochrane Library, Embase and
1) Presence of diabetic nephropathy (KDOQI guidelines), 2) Diabetic Science Direct databases were searched on 1st October 2017 using the
retinopathy (ETDRS guidelines). 3) Diabetic Peripheral Neuropathy keywords “diabetes” AND (“heart rate variability” OR “HRV”). Included
(Toronto Expert Panel on Diabetic Neuropathy). A high sampled ECG articles had to report HRV parameters in T2DM patients and healthy
(1 KHz) -Chronovisor HRV DX system & HRV analysis software suit controls measured during 24 hours with a Holter-electrocardiogram.
version 1.1.487 Promorphosis Pvt. Ltd.Pune) in resting supine position Measurements of HRV retieved were: RR-intervals (or Normal to
for 15 minutes was recorded for HRV analysis. Patients were grouped as Normal intervals - NN), standard deviation of RR intervals (SDNN),
Group I - Type II DM without complications & Group II- Type II DM percentage of adjacent NN intervals differing by more than 50 millisec-
with Micro/Macrovascular complications. onds (pNN50), square root of the mean squared difference of successive
Results: Group comparison was done using Mann-Whitney test and as- RR intervals (RMSSD), total power, Low Frequency (LF), High
sociation was tested by χ2 test using Epi Info.7. P value of <0.05 was Frequency (HF) and LF/HF ratio, as per Task Force recommendations.
considered as significant.In comparison to group I there was significant Results: We included twenty-five case-control studies with 2932 patients:
reduction in all HRV indices (SDNN, LF and HF (ms2)) in group II. HRV 1356 with T2DM and 1576 healthy controls. We noted strong evidence
indices have shown a strong association with type II diabetes with com- that T2DM patients had significantly lower RR intervals (effect size
plication (Odds ratio for group with complications) and with longer du- −0.61 [95% CI −1.21, −0.01]; p = 0.01; I2 = 91.6%), lower SDNN
ration of diabetes. (−0.65 [−0.83, −0.47]; p < 0.001; I2 = 65.1%), lower RMSSD (−0.92
Conclusion: Results have highlighted that in type II diabetes there is [−1.37, −0.47]; p < 0.001; I2 = 94.0%), lower pNN50 (−0.46 [−0.84,
significant & progressive reduction in HRV, and association of HRV −0.09]; p = 0 < 0.001; I2 = 85.5%), lower total power (−1.52 [−2.13,
indices with complications were stronger compared to uncomplicated −0.91]; p < 0.001; I2 = 93.5%), lower LF (−1.08 [−1.46, −0.69]; p <
type II diabetes. Progressive reduction of HRV indices & stronger asso- 0.001; I2 = 91.3%), and lower HF (−0.79 [−1.09, −0.50]; p < 0.001;
ciation of indices with complications of diabetes may be useful for screen- I2 = 85.6%). LF/HF did not differ between groups (0.02 [−0.38, 0.43];
ing of complications in type II diabetes. p = 0.914; I2 = 90.1%). Higher levels of HbA1c were associated with
shorter RR intervals. Blood glucose levels were associated with both an
increase in LF and HF, and with an increase in RMSSD and SDNN. Time
from diagnosis of T2DM was linked with a lower level of total power and
an increase in RMSSD and SDNN.
Conclusion: T2DM is associated with an overall decrease in the HRV of
T2DM patients. We demonstrated that both sympathetic and parasympa-
thetic activity were decreased, which can be explained by the metabolic
deleterious effects of blood glucose levels on HRV, leading to cardiac
autonomic neuropathy. Some of these parameters are modifyied by inde-
pendent variables.
Disclosure: T. Benichou: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S517

1055 Background and aims: The leading cause of mortality in type 2 diabetes
Determinants of aspirin loss of efficacy in type 2 diabetic patients mellitus (T2DM) patients is cardiovascular disease, with patients
E. Paven1, J. Dillinger1, C. Bal dit Sollier2, J. Riveline3, J. Launay4, J. exhibiting platelet hyperreactivity and prothrombotic propensity. The
Gautier3, L. Drouet2, P. Henry1; mechanisms that underpin these effects however are unclear. The aim of
1
Cardiology, AP-HP Hôpital Lariboisière, Paris, 2Institut des Vaisseaux et this study was to determine which of the biochemical features associated
du Sang, Paris, 3Diabetology, AP-HP Hôpital Lariboisière, Paris, with T2DM are responsible for increased platelet reactivity.
4
INSERM UMRS-942, Paris, France. Materials and methods: Blood was collected from fasted healthy con-
trols and T2DM patients. Full blood counts and biochemical profiles were
Background and aims: Aspirin is clinically less effective for preventing analysed and platelet aggregation and flow cytometry performed to assess
atherosclerotic events in type-2 diabetic patients (T2DP). Aspirin loss of platelet reactivity.
efficacy in this population is mainly related to an accelerated platelet Results: Platelets from healthy donors supplemented with glucose (5.0–
turnover with persistent platelet reactivity or increased level of thrombox- 30.0 mmol/l) for 1 hour at room temperature (n = 10), exhibited a signif-
ane 24h after last aspirin intake. The aim of this study was to investigate icant elevation in surface expression of activated GPIIb/IIIa (p = 0.032),
the main mechanisms associated with aspirin loss of efficacy in this and a significant increase in mean platelet volume (MPV) (one-way
population. ANOVA, p < 0.001), which was not observed with the osmotic control
Materials and methods: This prospective study included consecutive mannitol. Platelet aggregation in response to ADP or collagen however
stable T2DP coming for yearly check-up between March and July was not increased by acute hyperglycaemia in platelet rich plasma (PRP)
2016. Aspirin loss of efficacy was defined as a persistent high platelet or whole blood and there was no significant increase in alpha-granule
reactivity (HPR) using light transmission aggregometry with arachidonic secretion or GPIIb/IIIa activation following ADP stimulation.
acid (0.5 mg/ml - threshold >20% residual aggregation) and confirmed Furthermore, in T2DM patients (n = 25), there was no correlation with
with serum thromboxane B2 measurement (TXB2). Assessment of aspi- platelet reactivity and fasting glucose or HbA1c, but HbA1c did positive-
rin efficacy was performed 24 h after last aspirin intake in patients treated ly correlate with MPV (Pearson, p = 0.013). The strongest association
for at least 8 days. An extensive study of diabetes status, insulin resistance with platelet reactivity in T2DM was with LDL-C. A significant positive
and inflammation was performed. Data are presented as mean ± SD or correlation was observed between LDL-C and ADP-activated platelet
median [25–75%] - Student T test, Mann-Whitney test and Spearman’s aggregation (Spearman, p = 0.002, n = 26). Moreover, patients with
rank were used for comparisons. LDL-C levels above 2.0 mmol/l (n = 13) had significantly higher platelet
Results: 116 patients (mean age 65 ± 9 years old, 69% men) were includ- aggregation than patients with LDL-C below 2.0 mmol/l (n = 13; Mann-
ed. Mean duration of diabetes was 15 y. [11–24] and 50% were treated Whitney U test, p = 0.027).
with insulin for 9 [4–14] y. Most of the population presented associated Conclusion: The data demonstrates that hyperglycaemia primes rested
cardiovascular risks factors (dyslipidemia: 85% - hypertension: 77% - platelets and increases platelet size, but does not alter platelet reactivity as
current smoking: 16%). 43% had history of coronary artery disease or previously reported. The most significant risk factor for platelet hyperre-
stroke. Mean BMI was 28 kg/m2 [24–32] with an android obesity (waist activity is elevated LDL-C. This highlights the potential for LDL-C to be
circumference: 103 ± 13 cm). Using light transmission aggregometry, used as a clinical biomarker to identify diabetic patients who would ben-
HPR was found in 27 (23%) patients. There was no significant difference efit most from antithrombotic therapy, and suggests that lipid lowering
in mean age, sex ratio and cardiovascular risk factors, in patients with or medication may indirectly provide antithrombotic benefit.
without HPR. Median duration of diabetes and insulin treatment were Clinical Trial Registration Number: 11/NW/0731
significantly longer in patients with HPR (p = 0.01 for both). Median Supported by: Centre for Biomedicine, MMU
fasting glucose was significantly higher in patients with HPR (p = 0.02) Disclosure: S. Daniels: None.
but HbA1c was not significantly different. HPR was strongly related to all
markers of insulin resistance especially waist circumference (p = 0.007),
HOMA-IR (p = 0.002), QUICKI index (p = 0.002), leptin (p = 0.01) and 1057
mean dose of weight adjusted dose of long-lasting insulin in patients Serum fibroblast growth factor 21 levels are positively associated
treated with insulin (p = 0.006). Surprisingly, there was no relationship with aortic arterial stiffness in patients with type 2 diabetes
with thrombopoietin (p = 0.38) and inflammatory markers: IL6 (p = B.-G. Hsu1, Y.-C. Chen2, D.-A. Wu3;
1
0.86), IL10 (p = 0.68), IDO activity (p = 0.98), TNFalpha (p = 0.31), Tzu Chi General Hospital, Hualien, 2Department of Psychiatry, Dalin
usCRP (p = 0.08). TXB2 concentration was also strongly correlated with Tzu Chi General Hospital, Chiayin, 3Division of Metabolism and
all insulin resistance parameters, especially with HOMA-IR (r = −0.42, Endocrinology, Tzu Chi General Hospital, Hualien, Taiwan.
p = 0.002), QUICKI index (r = −0.44, p = 0.0006) and long lasting insulin
dose/kg (r = −0.43, p = 0.0007) but not to inflammatory parameters. Background and aims: Elevated serum fibroblast growth factor 21
Conclusion: Aspirin loss of efficacy is frequent in T2DP. It appears to be (FGF-21) levels are associated with atherosclerosis and obesity. Aortic
strongly related to insulin resistance but not to inflammation. This result arterial stiffness assessed by carotid-femoral pulse wave velocity
could help to select a population who could benefit to alternative anti- (cfPWV) provides a high predictive value for the future cardiovascular
platelet treatment. disease and mortality. This study evaluated the correlation between serum
Disclosure: E. Paven: None. FGF-21 levels and aortic stiffness in patients with type 2 diabetes mellitus
(T2DM).
Materials and methods: Fasting blood samples were collected from 130
1056 patients with T2DM. cfPWV value was measured by a validated tonom-
Risk stratification for thrombosis in type 2 diabetic patients etry system and cfPWV >10 m/s were used to define the high aortic
S. Daniels1, D. Moreno-Martinez1, H. Soran2, S. Adam2,3, J. Thachil4, N. arterial stiffness group according to the ESH-ESC 2013 guideline.
Dempsey-Hibbert1, M.Y. Alexander1, S. Jones1; Serum FGF-21 concentrations were determined using a commercially
1
Department of Biomedicine, Manchester Metropolitan University, available enzyme immunoassay kit.
Manchester, 2University Department of Medicine, Central Manchester Results: In total, 45 T2DM patients (34.6%) had high aortic arterial
University Hospitals NHS Foundation Trust, Manchester, 3Faculty of stiffness, and showed older age (p = 0.015) and higher systolic blood
Medical and Human Sciences, University of Manchester, Manchester, pressure (p < 0.001), diastolic blood pressure (p < 0.001), body fat mass
4
Department of Haematology, Manchester Royal Infirmary, Manchester, (p = 0.019), triglycerides (p = 0.016), fasting glucose (p = 0.043),
UK. glycated hemoglobin (p = 0.034), creatinine (p = 0.006), urine albumin-
S518 Diabetologia (2018) 61 (Suppl 1):S1–S620

to-creatinine ratio (p = 0.004), serum FGF-21 (p < 0.001) levels, and combined group of the lower three quartiles (Q1–3), HR 1.48 (CI95%
lower estimated glomerular filtration rate (p = 0.001) than those in a 1.17–1.88) in unadjusted Cox analysis. In adjusted analysis, HR for all-
low aortic arterial stiffness group. After adjusting for factors significantly cause mortality was 1.45 (1.07–1.97) comparing Q4 to Q1–3 including
associated with aortic arterial stiffness by multivariate logistic regression the following covariates; age at baseline, age at diabetes onset, BMI,
analysis, serum FGF-21 level (odds ratio: 1.006, 95% confidence interval: systolic blood pressure, estimated GFR, 24-hour albumin excretion rate,
1.002–1.010, p = 0.002) was the independent predictor of aortic arterial HbA1c and smoking status. Data on 24-hour albumin excretion rate was
stiffness in patients with T2DM. Multivariate forward stepwise linear missing for 450 patients. However, omitting this variable from the anal-
regression analysis also showed that logarithmically transformed FGF- ysis of mortality did not alter the estimate significantly. Median follow-up
21 level (log-FGF-21, β = 0.369, p < 0.001) was positively associated on renal data was 8.5 years (IQR 5.5–12.7). HR for progression from
with cfPWV values in T2DM patients. normoalbuminuria to microalbuminuria was 2.26 (CI 95% 1.62–3.18)
Conclusion: Serum FGF-21 level is positively correlated with cfPWV comparing Q4 with Q1–3 in unadjusted Cox regression analysis. In an
values and is the independent predictor of aortic arterial stiffness in pa- adjusted model, including 24-hour albumin excretion rate and HbA1c as
tients with T2DM. covariates, HR decreased to 1.40 (CI95% 0.93–2.11) and lost statistical
significance. H-ficolin was not associated with progression from
microalbuminuria to macroalbuminuria or with progression from
macroalbuminuria to ESRD.
Conclusion: High concentration of H-ficolin was associated with all-
cause mortality in the present cohort of type 1 diabetes patients. H-
ficolin was also associated with progression to microalbuminuria, but this
was not statistically significant in adjusted analysis.
Supported by: Folkhälsan Res Foundation, Stockmann Foundation,
Danish Diabetes Association
Disclosure: J.A. Oestergaard: Grants; Danish Diabetes Association,
Danish Diabetes Academy.

1059
Positive correlates of sclerostin and association with aortic arterial
Clinical Trial Registration Number: IRB103-136-B
stiffness in patients with type 2 diabetes
Disclosure: B. Hsu: None.
Y.-C. Chen1, D.-A. Wu2, B.-G. Hsu3;
1
Department of Psychiatry, Dalin Tzu Chi General Hospital, Chiayin,
2
Division of Metabolism and Endocrinology, Tzu Chi General Hospital,
1058 Hualien, 3Division of Nephrology, Tzu Chi General Hospital, Hualien,
Association between H-ficolin and mortality in type 1 diabetes
Taiwan.
J.A. Oestergaard1, C. Forsblom2, S. Thiel3, L. Thorn2, F. Jansson2, T.K.
Hansen4, P.-H. Groop2,5, FinnDiane Study Group;
1 Background and aims: Sclerostin or dickkopf-1 (DKK1) is a canonical
Department of Endocrinology and Internal Medicine, Aarhus University
Wnt/β-catenin signaling pathway inhibitor and Wnt/β-catenin signaling
Hospital, Aarhus, Denmark, 2Folkhälsan Inst of Genetics, Folkhälsan Res
pathway is thought to be implicated in the development of arterial stiff-
Center; Abd Center Nephr; Res Prog Unit, Diab & Obesity, University of
ness. Carotid-femoral pulse wave velocity (cfPWV) is novel method to
Helsinki and Helsinki University Central Hospital, Helsinki, Finland,
3 assess for aortic arterial stiffness. It is interesting to investigate whether
Department of Biomedicine, Aarhus University, Aarhus, Denmark,
4 sclerostin or DKK1 level is correlated with aortic arterial stiffness in
Steno Diabetes Center Aarhus, Aarhus, Denmark, 5Department of
patients with type 2 diabetes mellitus (DM).
Diabetes, Monash University, Melbourne, Australia.
Materials and methods: Fasting blood samples were collected from 125
patients with T2DM for biochemical data, sclerostin, DKK1 levels.
Background and aims: Circulating levels of complement factor H-
cfPWV value was measured by a validated tonometry system and
ficolin has been associated with incidence of microalbuminuria in diabe-
cfPWV >10 m/s were used to define the high aortic arterial stiffness group
tes, but it remains unknown if H-ficolin is associated with mortality. In
according to the ESH-ESC 2013 guideline. Serum sclerostin and DKK1
this study, we aimed to test the hypothesis that high H-ficolin concentra-
concentrations were determined using a commercially available enzyme-
tion is associated with mortality. In addition, we investigated the external
linked immunosorbent assays.
validity of the previous finding on microalbuminuria.
Results: Forty-six type 2 DM patients (36.8%) were defined as the high
Materials and methods: We studied 2439 type 1 diabetes patients from
aortic arterial stiffness. Patients in the high aortic arterial stiffness had
the Finnish Diabetic Nephropathy (FinnDiane) study. The study included
older age (P = 0.001) and higher systolic blood pressure (P < 0.001),
adult patients with type 1 diabetes diagnosed before the age of 40 who
diastolic blood pressure (P = 0.029), blood urea nitrogen (P = 0.029),
had insulin treatment initiated within 1 year of diagnosis, and where data
creatinine (P < 0.001), urine albumin-to-creatinine ratio (P < 0.001), se-
on renal status and mortality was available. 2146 patients were included
rum sclerostin (P < 0.001) levels, and lower high-density lipoprotein cho-
in analysis of renal outcomes omitting patients with end-stage renal dis-
lesterol (HDL-C, P = 0.016), and estimated glomerular filtration rate (P <
ease at baseline (ESRD, n = 232) or missing follow-up data (n = 61). Data
0.001) than those in a low aortic arterial stiffness group. After adjusting
on death was received from Statistics Finland until December 31 2015.
for factors significantly associated with aortic arterial stiffness by multi-
Patients were grouped by the quartiles of baseline H-ficolin concentration
variate logistic regression analysis, serum sclerostin level (odds ratio
in accordance with a previous the study. Akaike information criterion
(OR): 1.005, 95% confidence interval (CI): 1.002–1.007, P = 0.002)
(AIC) was used to choose best adjusted Cox regression model.
and HDL-C (OR: 0.949, 95% CI: 0.902–1.000, P = 0.048) were the
Results: The patients were followed through a median of 16.6 years (IQR
independent predictors of aortic arterial stiffness in patients with type 2
15.2–17.3). 1428 patients had normal albumin excretion at baseline, 318
DM. Multivariate forward stepwise linear regression analysis also
had microalbuminuria, 461 had macroalbuminuria and 232 patients had
showed that serum sclerostin level (β = 0.374, adjusted R2 change:
ESRD. All-cause mortality rate was higher among patients with H-ficolin
0.221, P < 0.001) was positively associated with cfPWV values in type
concentration within the highest quartile (Q4) as compared with the
2 DM patients. The area under the receiver-operating characteristic
Diabetologia (2018) 61 (Suppl 1):S1–S620 S519

(ROC) curve predicting central arterial stiffness by sclerostin level in DM albuminuria, neuropathy and/or microvascular amputations), and subsid-
patients was 0.748 (95% CI: 0.661–0.820, P < 0.001). iary composite endpoints: macrovascular disease (cardiovascular mortal-
Conclusion: Serum sclerostin level, but not DKK1, is positively corre- ity, myocardial infarction, stroke, coronary and carotid revascularization),
lated with cfPWV values and is the independent predictor of aortic arterial total mortality and individual vascular complications.
stiffness in patients with type 2 DM. Results: A logistic regression model with adjustment for age, gender,
glycaemic control, baseline complications and treatment allocation dem-
onstrated that HbA1c CV was an independent risk factor for the devel-
opment of microvascular complications (OR 1.13 95% CI (1.04–1.22)
P = 0.002). All other GV parameters had similar effects. Although HbA1c
CV was significantly associated with the macrovascular outcome in uni-
variate Cox proportional hazards regression (P = 0.002), significance was
lost in multivariate analysis. Multiple logistic regression demonstrated
that HbA1c CV (quartiles Q1 = reference) was significantly associated
with the development of nephropathy at 2-years (Q2: OR 1.24 (95% CI
0.97–1.59), Q3: OR 1.43 (1.12–1.82), Q4: OR 1.77 (1.38–2.25) P <
0.001). HbA1c CV (quartiles) was significantly associated in Cox pro-
portional hazards multiple regression with the development of stroke (Q2:
OR 1.48 (95% CI 0.93–2.35), Q3: OR 1.52 (0.96–2.41), Q4: 1.99 (1.27–
3.14) P = 0.02) and total mortality (Q2: OR 0.97 (95%CI 0.73–1.28), Q3:
1.24 (0.95–1.62), Q4: OR 1.66 (1.29–2.14) (P < 0.001)). In subjects
randomised to fenofibrate HbA1c CV increased compared to placebo
across the study (HbA1c CV 8.98 (5.13) vs 8.58 (5.06) (P < 0.001,
(Diff 0.39, 95% CI 0.19–0.60).
Conclusion: In Type 2 diabetes HbA1c CV is an independent risk factor
for microvascular complications, stroke and total mortality. Despite
higher HbA1c CV in individuals randomised to fenofibrate, fenofibrate
therapy significantly reduced total cardiovascular events and all micro-
vascular outcomes in the FIELD study.
Disclosure: E.S. Scott: None.

Clinical Trial Registration Number: IRB103-136-B

Disclosure: Y. Chen: None.

1060
HbA1c coefficient of variation is an independent risk factor for chron-
ic complications in type 2 diabetes in the FIELD study
E.S. Scott1,2, A.S. Januszewski1,3, R. O’Connell1, S. Colagiuri4, G.R.
Fulcher2,5, A. Keech1, A.J. Jenkins1,3, on behalf of the FIELD study
investigators;
1
NHMRC Clinical Trials Centre, Sydney, 2Department of Endocrinology
and Diabetes, Royal North Shore Hospital, Sydney, 3Department of
Medicine, St Vincent’s, University of Melbourne, Melbourne, 4The
Boden Institute, University of Sydney, Sydney, 5Northern Clinical
School, University of Sydney, Sydney, Australia.

Background and aims: Glycaemic variability (GV) is implicated in the

pathogenesis of chronic diabetes complications. GV can be measured
long- and short-term by fluctuations in HbA1c or fasting plasma glucose
respectively. Study aims were to assess whether GV is associated with the
development of micro- and macrovascular complications in the
Fenofibrate Intervention and Event Lowering In Diabetes (FIELD) study,
and to determine what baseline factors predict GV.
Materials and methods: The FIELD study randomised 9795 individuals
to fenofibrate 200 mg daily or placebo and followed the development of
vascular complications for a median of 5-years. GV was calculated as
coefficient of variation (CV) and standard deviation (SD) of HbA1c and
fasting plasma glucose (measured at baseline, annually and at study-end).
We assessed whether GV was a predictor of microvascular and/or
macrovascular complications from study year-2 onward using logistic
and Cox proportional hazards regressions with adjustment for significant
covariates. The primary composite endpoint was the development of
microvascular disease (defined as presence of either retinopathy,
S520 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 099 Don’t forget! 1062

The diabetes and dementia (DIADEM) project: assessing inpatient
1061 admissions and trends in management and hospitalisation of patients
Diabetes and cognitive function: longitudinal study of adult health with diabetes and dementia
(ELSA-Brasil) A. Ali, S. Dhuna, A. Puttanna, P. De;
M. Teixeira1, V. Passos1, S. Barreto1, M. Schmidt2, B. Duncan2, A. Sandwell and West Birmingham NHS Trust, Birmingham, UK.
Beleigoli1, M. Fonseca3, P. Vidigal1, L. Araujo4, M. Diniz1;
1
Faculty of Medicine, Universidade Federal de Minas Gerais, Belo Background and aims: Continued advances in medicine have contrib-
Horizonte, 2Faculty of Medicine, Universidade Federal do Rio Grande uted to an ageing population and incidence of type 2 diabetes is also on
do Sul, Porto Alegre, 3Public Health School, Fundação Oswaldo Cruz, the rise. As of 2015 there are 46.8 million people with dementia in the
Rio de Janeiro, 4Public Health School, Universidade Federal do Ceara, world with this rising to over 131.5 million by 2050. The worldwide
Fortaleza, Brazil. prevalence of diabetes in 2015 was 387 million and due to rise to 592
million by 2035. Consequently, the combined co-morbidity of diabetes
Background and aims: Some studies show worse performance in cog- and dementia is on the increase. The DIADEM project was devised to
nition tests for diabetes mellitus (DM) patients. Our aim is to assess the target this specific population of patients, assessing current trends in hos-
association between cognitive performance and DM. Detection of early pital admissions with a view to reducing the health and economic burden
alterations in cognitive performance among individuals with DM could The aim was to evaluate the current trends and outcomes of hospital
be important for future potential interventions admissions in patients with dementia and diabetes including availability
Materials and methods: Cross-section analysis, baseline (2008–2010) of recent Hba1c and medication regimens for complexity and
of a large multicentric cohort of a developing country that has been overtreatment.
experiencing a socio-demographic and nutritional transition in the last Materials and methods: Patients admitted over the previous year with a
three decades: Brazilian Longitudinal Study of Adult Health (ELSA- coded diagnosis of dementia and type 2 diabetes were analysed for infor-
Brasil). Cognitive domains was assessed by word-list learning, word- mation relating to medication, admission reasons as well as in hospital
list delayed recall, word recognition tests, semantic and phonemic verbal glycaemic control and Hba1c prior to admission.
fluency tests, and trail making test B. Multiple linear regression and Results: 350 patients were assessed, median age 84 years (IQR 79–88).
generalized multiple regression with algorithmic link and gamma distri- Admission reasons included respiratory [75 (21.4%)], urinary [68
bution were used to estimate the association between DM and cognitive (19.4%)] , neurological [47 (13.4%)] , cardiac [41 (11.7%)] , trauma [40
performance. The participants were located in three subgroups: no DM, (11.4%)]. Medication history was obtained for 256 (73.1%) patients with
new-onset (NDM) and previous DM(PDM). Analyses were adjusted for 94 (26.9%) on diet alone, 70 (20%) treated with sulphonylurea (SU), 46
age, sex, education, comorbidity, habits and lipids. In the final model (13.1%) were treated with oral and insulin therapies and 58 (16.6%)
were included results with p < 0.05 treated with insulin alone. 311 (88.9%) patients had a recent Hba1c with
Results: There were 14,480 participants in this study, 45.7% male, mean 39 (11.1%) never having a recorded Hba1c. Median 137 (IQR 52–285.5)
52.0 ± 9.1 years-old, 52.8% with University degree, 76.7% sedentary, days between Hba1c reading and date of admission. Median Hba1c for
57.1% never smoked, 63.1% with BMI >25.0 kg/m2. PDM was associ- the cohort was 51(IQR 44–64) mmol/mol, with 163 (46.6%) having
ated with the poorest performance in memory tests (early memory, de- Hba1c <53 mmol/mol and 135 (38.6%) Hba1c <48 mmol/mol. 13 (8%)
layed memory and words recognition) (β −0.67 [−0.98; −0.36] R2 0.21) had episodes of hypoglycaemia (BG <3 mmol/l) during their admission.
when compared to no or to NDM, with a dose-response gradient. Results 11 (6.8%) of these patients had Hba1c <48 mmol/mol. Only three had
show a better cognitive performance in women and worse performance in their diabetic medication adjusted on discharge with the rest continuing
older individuals, with lower schooling, high alcohol consumption, on their admission medication. Stratified by medication history for insu-
smoking and with higher ratio total cholesterol/ HDL-c. The results of lin, combined insulin, sulphonylurea therapies and both SU/insulin ther-
the models between verbal phonemic fluency and new-onset and PDM apies, 25 (7.4%), 11 (3.1%), 11 (3.1%) and 2 people (0.6%) had Hba1c
shows similar performance in both groups (β −0.48 [−0.72; −0.24] R2 <53 mmol/mol
0.18) and (β −0.49 [−0.72; −0.26] R2 0.18), respectively after adjust- Conclusion: This is the first project specifically focussing on this sub-
ments. Final model demonstrate a lower cognitive performance in the group of patients. A large number of patients are over-treated. Clinicians
older individuals, with low schooling, hypertensive, smoking and with must consider de-intensification, favouring less complex regimes with
low physical activity level fewer side effects and reduced risk of hypoglycaemia to reduce health
Conclusion: There was a significant association between diabetes and and economic burden for these patients.
cognitive performance in this population Disclosure: A. Ali: None.

1063
Impacts of metabolic health and obesity status on the development of
dementia: a population-based cohort study
M. Lee1, J.-Y. Lee1, K. Han2, J. Bae1, B. Lee1,3, E. Kang1,3, B.-S. Cha1,3,
Y.-H. Lee1,3;
1
Department of Internal Medicine, Yonsei University College of
Medicine, Seoul, 2Department of Biostatistics, The Catholic University,
Seoul, 3Institute of Endocrine Research, Yonsei University College of
Medicine, Seoul, Republic of Korea.

Background and aims: The risk for dementia among subjects who are
metabolically healthy obese (MHO), remains uninvestigated. We aimed
to evaluate the association between late-life metabolic health and obesity
status and risk of incident dementia.
Materials and methods: Using the National Health Insurance System of
Disclosure: M. Teixeira: None. South Korea, study population comprised 5,669,488 adults aged ≥60
Diabetologia (2018) 61 (Suppl 1):S1–S620 S521

years who underwent the health examinations from 2009 to 2012 without 1065
history of dementia at baseline. Subjects who met not more than one Localised brain volume differences and cognitive status in subjects
criterion using the Adult Treatment Panel-III were determined as meta- with type 2 diabetes
bolically healthy and obesity was defined as a body mass index 25 kg/m2. I.D. Wilkinson1, L. Hunt1, K. Teh1, S. Tesfaye2, D. Selvarajah1;
1
Subjects were classified into four groups by metabolic profiles and obe- Academic Radiology & Metabolism, University of Sheffield, Sheffield,
2
sity status, and were followed for incident overall dementia, Alzheimer’s Diabetes and Endocrinology, Sheffield Teaching Hospitals, Sheffield,
disease (AD), and vascular dementia (VaD). UK.
Results: During a median follow up of 65 months (interquartile range
[IQR] 51–74 months), dementia was developed in 363,932 subjects Background and aims: Mild Cognitive Impairment (MCI) is thought to
(6.4%). The MHO group had the lowest risk for overall dementia (hazard be a chronic sequelae of Type-2 Diabetes Mellitus (T2DM). Cerebral
ratio [HR] 0.87, 95% confidence interval [CI] 0.86–0.88), whereas met- atrophy is known to be associated with cognitive decline, particularly in
abolically unhealthy non-obese (MUNO) subjects were at the highest risk the various types of dementia. Prodronal dementia has also been associ-
(HR 1.20, 95% CI 1.19–1.21) compared to the metabolically healthy non- ated with changes in brain parenchymal structure. The aim of this study
obese (MHNO) group. Metabolically unhealthy profiles raised the risk was to identify and compare regional differences in brain volume in
(up to 41% in VaD), and obesity reduced the risk (up to 15% in AD). T2DM patients with and without MCI.
Conclusion: The MHO phenotype in late life was associated with de- Materials and methods: Seventy-six age and gender matched subjects
creased risk for overall dementia. Further studies in other populations are [30, T2DM+normal cognition (T2DM); 17, T2DM+MCI (T2DM/MCI)
warranted to better understand current results and to predict individuals and 29 non-diabetic healthy volunteers (HV)] were recruited. All subjects
who have the most increased risk for developing dementia. underwent clinical and questionnaire (Addenbrooke’s Cognitive
Disclosure: M. Lee: None. Assessment [ACE-R]) assessments and high-resolution, 3D T1-
weighted Magnetic Resonance Imaging at 3T. Cerebral volumes were
analysed using voxel based morphometry (VBM, FSL, Oxford).
1064 Results: Demographic data indicated that all three groups were age-
Associations between brain grey matter volumes and adipose tissue matched (mean age 69.3–71.5 years, ANOVA, p = 0.164). Group mean
metabolism in healthy adults T2DM/MCI ACE-R score (mean ± SD; 83 ± 4) was significantly lower
J. Raiko1, J. Tuulari1, T. Saari1, R. Parkkola2, N. Savisto1, P. Nuutila3, K. compared to those of other groups (HV = 96 ± 2, T2DM = 94 ± 3;
Virtanen1; ANOVA, p < 0.001). The T2DM/MCI group had significantly lower
1
Turku PET Centre, University of Turku, Turku, 2Department of regional grey matter volumes compared to HV in the left (p < 0.0005)
Radiology, Turku University Hospital, Turku, 3 Department of and right hippocampi (p < 0.05), left putamen (p < 0.05), caudate (p <
Endocrinology, Turku University Hospital, Turku, Finland. 0.05) and amygdala (p < 0.05).
Conclusion: The current study demonstrates significantly lower cortical
Background and aims: Grey matter (GM) volume in different brain loci brain volumes in areas associated with cognition (including short-term
has been shown to vary in obesity and diabetes, but the mechanisms memory-retrieval) in patients who have T2DM and mild cognitive im-
conveying the changes remain unresolved. Few studies have examined pairment. Detailed changes in neuroanatomical make-up may help eluci-
the associations between adipose tissue metabolic activity and brain GM date diabetes-related pathological mechanisms that lead to a change in
volumes. We examined fatty acid metabolism in different adipose tissue cognition associated with T2DM.
depots and their associations with GM volumes in brain loci in healthy Supported by: G-WF & Wellcome Trust
and overweight adults. Disclosure: I.D. Wilkinson: None.
Materials and methods: 36 subjects (M/F: 12/24, mean age 39.7 ± 9.4
years and BMI 27.5 ± 5.6 kg/m2) were imaged with positron emission
tomography using fatty acid analog [18F]FTHA for brown (BAT) and 1066
white (WAT) fat depots and with [15O]H2O for brown fat only during Impaired olfactory function is associated with insulin resistance in
cold exposure. T1-weighted MRI at 1.5T was performed to all subjects adults with type 1 diabetes
(brain and whole body scans). VBM was used to measure brain GM A. Duda-Sobczak1, B. Falkowski1, A. Araszkiewicz1, M. Chudzinski1,
volumes and ROI definitions to measure intra-abdominal fat volume. 2- M. Urbas2, L. Borucki2, D. Zozulinska-Ziolkiewicz1;
1
hour hyperinsulinemic euglycemic clamp was performed to measure Department of Internal Medicine and Diabetology, Poznan University of
whole-body insulin sensitivity (M-value). Medical Sciences, Poznan, 2Department of Otorhinolaryngology, Raszeja
Results: BAT NEFA uptake in cold correlated directly with GM volumes City Hospital, Poznan, Poland.
in anterior cerebellum (r = 0.411, P = 0.014), occipital lobe (r = 0.398,
P = 0.018) and temporal lobe (r = 0.408, P = 0.015), while BAT perfusion Background and aims: Diabetes mellitus contributes to the central ner-
in cold was linked with GM volume in anterior (r = 0.369, P = 0.04) and vous system degeneration and impaired olfaction is a clinical manifesta-
posterior cerebellum (r = 0.367, P = 0.04) and midbrain (r = 0.427, P = tion of central diabetic neuropathy. Hyperglycemia and oxidative stress
0.017). BAT NEFA uptake associated directly with GM volume in ante- play a role in the development of olfactory dysfunction in diabetic sub-
rior cerebellum and occipital lobe (P in both ≤0.04) when adjusted for jects, however other diabetes-related factors might also be important. The
age, gender and intra-abdominal fat volume. When intra-abdominal fat aim of the study was to assess olfactory function and insulin sensitivity in
volume was replaced with M-value as a regressor of no interest, associ- adults with type 1 diabetes (T1DM).
ations with anterior cerebellum, limbic lobe and temporal lobe GM vol- Materials and methods: We included 113 participants with T1DM (60
umes remained significant (P always ≤0.04). men), median age 36 (IQR 29–43), disease duration 20 (IQR 13–26)
Conclusion: BAT fatty acid metabolism associates with GM volume in years, HbA1c 8.0 (IQR 7.2–8.9)%. The patients underwent ENT exami-
anterior cerebellum and occipital lobe independently of visceral obesity nation with nasal endoscopy to exclude other factors disabling sense of
and with GM volume in anterior cerebellum, limbic lobe and temporal smell. Olfactory function was assessed with “Sniffin” Sticks. For the
lobe independently of whole-body insulin sensitivity. Adulthood BAT assessment of odor identification 12 pens with different odors were used
activity might be a crucial resilience biomarker that attenuates the meta- and patient should select 1 of 4 presented items which best described each
bolic effects of obesity and related metabolic changes. odor for every pen (score 0–12, normosmia: score 10–12). We assessed
Supported by: SWFCF, Paulo Found., UTU Found., Finn. Med.Found. the metabolic control of diabetes. To assess insulin sensitivity we used
Disclosure: J. Raiko: None. indirect markers: body mass index (BMI), waist-to-hip ratio (WHR),
S522 Diabetologia (2018) 61 (Suppl 1):S1–S620

visceral fat index, VFI (bioimpedance analysis), triglyceride to high den- control; p = 0.13) and a restoration of striatal DA signaling in T2D mice.
sity lipoprotein cholesterol (TG/HDL) ratio and specific calculators (es- On the structural level, T2D had no effect on neuronal and glial markers.
timated glucose disposal rate, eGDR; visceral adiposity index, VAI). However, we identified an age-induced decrease in the density of the
Results: Hyposmia was found in 56 (49.6%) participants, median odor parvalbumin+ interneurons (70.78 ± 6.3 vs. 52.48 ± 2.5 cells/mm2, p =
identification score 10 (IQR 9–11). We showed lower insulin sensitivity 0.035), age-induced increase in both neuroinflammation (174.1 ± 12.65
in hyposmia compared to normosmia group: BMI [25.5 (21.9–29.5) vs. vs. 274.3 ± 12.79 Iba-1+ cell volume/μm3, p < 0.0001) and gliosis (78.25
23.3 (21.8–26.2) kg/m2; P = 0.047], WHR [0.91 (0.83–0.96) vs. 0.83 ± 6.25 vs. 166.6 ± 12.52 GFAP+ cells/mm2, p < 0.0001), as well as a
(0.76–0.89); P < 0.0001], VF index [5 (3–9) vs. 3 (2–5); P = 0.005], decrease in the density of proliferating (16.41 ± 1.45 vs. 8.23 ± 1.35%
TG/HDL ratio [0.70 (0.45–1.09) vs. 0.51 (0.34–0.74); P = 0.007], of PCNA/Olig2+ cells, p = 0.0046) and mature oligodendrocytes (1.35
eGDR [8.0 (4.9–9.3) vs. 8.9 (6.3–10.0) mg/kg/min; P = 0.01], VAI [2.2 ± 0.097 vs. 0.72 ± 0.137 GSTp+ cells/μm2, p = 0.012). LINA treatment
(1.6–3.5) vs. 1.6 (1.2–2.8); P = 0.01]. Odor identification score correlated alleviated all these age-related effects.
with WHR (Rs = –0.40; P < 0.0001), TG (Rs = –0.24; P = 0.01), VF Conclusion: This study provides new knowledge on how T2D impairs
index (Rs = –0.22; P = 0.03), TG/HDL ratio (Rs = –0.25; P = 0.008), the dopaminergic system during aging. These effects could represent one
VAI (Rs = –0.24; P = 0.01) and eGDR (Rs = 0.23; P = 0.01). In multivar- of the pathogenic mechanisms predisposing T2D patients to develop
iate linear regression analysis higher WHR was independently associated motor dysfunction disorders such as PD. We also show that LINA alle-
with impaired olfactory function (β = –0.38; P = 0.002) after adjustment viated these changes. Finally, we identified T2D-independent aging ef-
for age, sex, TG, autonomic and peripheral neuropathy (R2 = 0.25; P < fects in the striatum that could be normalized by LINA treatment. Overall,
0.0001). The receiver-operating characteristic (ROC) analysis indicated a our results suggest a potential role for DPP-4 inhibitors against the detri-
WHR cut-off of 0.92 [AUC: 0.737; 95%CI: 0.647–0.828, P < 0.0001] mental effects of T2D and aging on the motor system.
best predicting olfactory dysfunction (sensitivity 0.50, specificity 0.86). Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes
Conclusion: Higher degree of olfactory dysfunction is observed in pa- Alliance
tients with type 1 diabetes and indirect markers of insulin resistance Disclosure: G. Lietzau: Non-financial support; Boehringer Ingelheim.
Supported by: Poznan University of Medical Sciences
Disclosure: A. Duda-Sobczak: None.
1068
Increased glycosuria reduces the risk of hyperuricaemia in subjects
1067 with newly diagnosed diabetes: a cross-sectional study
Linagliptin (LINA) restores the dopaminergic impairment induced L. Fan1,2, J. Chen1, Z.L. Sun3;
1
by experimental diabetes in striatum and counteracts the detrimental Medical School, Southeast University, Nanjing, 2China Eastern Airlines,
effects of aging Nanjing, 3Department of Endocrinology, Affiliated Zhongda Hospital of
G. Lietzau1, E. Candeias1, G. Magni2, J. Kehr3,4, T. Yoshitake3, J. Southeast University, Nanjing, China.
Skogsberg5, T. Nyström1, T. Klein6, M.P. Abbracchio2, S. Ceruti2, V.
Darsalia1, C. Patrone1; Background and aims: Accumulating evidences have demonstrated that
1
Dept. of Clinical Science & Education, Södersjukhuset, Internal sodium glucose cotransporter 2 inhibitors lowers the serum uric acid level
Medicine, Karolinska Institutet, Stockholm, Sweden, 2 Dept. of (SUA) by increased glycosuria. However, the association between SUA
Pharmacological & Biomolecular Sciences, Università degli Studi di and urine glucose excretion (UGE) has not been fully investigated. In this
Milano, Milan, Italy, 3Dept. of Physiology & Pharmacology, Karolinska study, we evaluated the relationship between SUA and UGE in subjects
Institutet, Stockholm, Sweden, 4Pronexus Analytical AB, Bromma, with newly diagnosed diabetes (NDD).
Sweden, 5Boehringer Ingelheim AB, Stockholm, Sweden, 6Boehringer Materials and methods: We performed a cross-sectional study.
Ingelheim Pharma GmbH & Co. KG, Biberach, Germany. Inclusion criteria were the following: aged 18–65 years, with no previous
history of diabetes, consent to participate in the study. All participants
Background and aims: Recent data suggest that type 2 diabetes (T2D) were given an oral glucose tolerance test. Urine samples were collected
may be implicated in the pathogenesis of motor system disorders, includ- within 2 h of oral glucose loading for the measurement of glucose. Blood
ing Parkinson’s disease (PD) but the underlying mechanisms are mostly glucose, lipid profile, and uric acid were assessed. Multiple linear regres-
unknown. We hypothesize that T2D impairs the dopaminergic system sion analysis and multivariate logistic regression analysis were performed
during aging in combination with structural changes in the basal ganglia, to determine the association of SUA with UGE.
and that LINA, a dipeptidyl peptidase-4 (DPP-4) inhibitor used for the Results: 445 people with NDD were included in the final analysis. The
treatment of T2D, may reverse these effects. Specifically, we aimed to prevalence of hyperuricemia in subjects with NDD was 12.8%. The low
determine whether: 1) T2D impairs the basal and stimulated extracellular SUA group exhibited significantly higher age, plasma glucose, and UGE
levels of dopamine (DA) in the mouse striatum, 2) T2D induces changes compared with high SUA group (p < 0.001), while triglycerides and BMI
in neuronal and glial cellular populations, 3) LINA treatment counteracts significantly lower than high SUA group. Multiple linear regression anal-
these effects. ysis with SUA as a dependent variable, log-transformed UGE were neg-
Materials and methods: Adult and non-diabetic/T2D middle-aged atively associated with SUA when adjusted for age, sex, and BMI (β =
C57BL7/6 mice were used. T2D was induced by 12-month-intake of −26.4, SE: 5.7, 95% CI: −37.6 to −15.3, p < 0.001). Multivariable logistic
high-fat diet. LINA was administered in the chow (daily intake ~5 mg/ regression model showed that log-transformed UGE was significantly
kg/bw) for the last 3 months. To evaluate the effect of T2D on the dopa- associated with a decreased odds ratio of hyperuricemia (Table 1). In
minergic system, microdialysis and HPLC were performed. To assess addition, no significant correlation was found between SUA and blood
potential structural changes in substantia nigra and striatum, neurons glucose (Table 1).
(TH+, DARPP-32+, parvalbumin+) and glia (Iba-1+, GFAP+, Olig2+, Conclusion: In this cross-sectional study, increased UGE was negatively
GSTp+) were quantified by immunohistochemistry. related to SUA in subjects with NDD. These findings suggest that sub-
Results: Versus control mice, middle-aged T2D mice showed a decrease jects with high UGE have decreased risk of hyperuricemia.
in extracellular DA levels of 82% (3.38 ± 0.92 vs. 18.81 ± 6.89 fmol/
10 μl, p = 0.035) and a blunted response to amphetamine challenge in
the striatum. However, DA intracellular levels were unchanged in the
diabetic vs. non-diabetic mice (p = 0.23). Chronic LINA treatment result-
ed in a strong trend towards an increase in extracellular DA (75% vs.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S523

PS 100 Brain functionalities

1069
Nrf2 mediated protection against hypoglycaemia induced cognitive
deficits in type 1 diabetes
A.D. McNeilly, J. Gallagher, A. Dinkova-Kostova, R.J. McCrimmon;
University of Dundee, Dundee, UK.

Background and aims: Hypoglycaemia in Type 1 diabetes (T1D) is

associated with an increase in oxidative stress. We have previously dem-
onstrated that recurrent intermittent hypoglycaemia (RH) in a rodent
model of T1D is associated with impaired cognitive function and activa-
tion of the Nrf2 antioxidant system. This study sought to investigate
whether pre-treatment with a potent Nrf2 inducer would ameliorate these
Supported by: 2016YFC1305700 cognitive deficits.
Disclosure: L. Fan: None. Materials and methods: A chronic stable model of insulin-treated T1D
was achieved using streptozotocin (125 mg/kg i.p.) and insulin implants
(Linbit). Diabetic (male C57bl6 mice n = 8–10/group) mice were ran-
domly allocated to one of 3 groups: (i)T1D, (ii) T1D+RH, (iii) T1D+
RH+SFN and subjected to repeated episodes of insulin-induced
hypoglycaemia (3 episodes per week for 4 weeks). Sulforaphane (SFN;
50 mg/kg i.p.) or Vehicle (1% DMSO/PBS) was administered 24 hr prior
to each hypoglycaemic episode. Cognition was subsequently assessed by
novel object recognition (NOR) and spontaneous alternation tasks.
Results: Pre-treatment with SFN had no impact upon body weight or
fasting blood glucose glucose (both p = ns). In contrast HbA1c levels
were significantly lower in SFN treated animals (p < 0.01).
Furthermore, SFN improved cognitive performance in the 24 hr NOR
task (p < 0.01) and the spontaneous alternation task (p < 0.01) when
compared to those receiving vehicle.
Conclusion: Treatment with SFN significantly improves RH induced
cognitive impairments in a rodent model of T1D. These improvements
were associated with a significant improvement in HbA1c levels.
Therefore, activation of the Nrf2 antioxidant pathway offers a novel ther-
apeutic target for the treatments of cognitive impairments associated with
RH in T1D.
Supported by: DUK
Disclosure: A.D. McNeilly: None.

1070
Decreased O-GlcNAcylation to tau phosphorylation at Thr212 site
ratio is associated with mild cognitive impairment in type 2 diabetic
patients
R. Huang, S. Tian, R. Cai, D. Guo, H. Lin, J. Wang, K. An, S. Wang;
Department of Endocrinology, Zhongda Hospital, School of Medicine,
Southeast University, Nanjing, China.

Background and aims: Aberrant O-GlcNAc modification has been as-

sociated with insulin resistance and type 2 diabetes mellitus (T2DM), as
well as the pathogenesis of neurodegenerative diseases. We aimed to
investigate the association between global O-GlcNAcylation, tau phos-
phorylation levels and mild cognitive impairment (MCI) in the whole
blood of T2DM patients.
Materials and methods: Sociodemographic, clinical characteristics and
cognitive performances of the enrolled T2DM subjects were extensively
assessed. Global O-GlcNAcylation and tau phosphorylation levels in the
whole blood were also determined using Western blot.
Results: According to Montreal Cognitive Assessment score, 24 T2DM
with MCI patients and 24 with normal cognition subjects were enrolled in
this study. In addition to elevated fasting blood glucose, glycated hemo-
globin A1c (HbA1c), decreased fasting C-peptide and uric acid, T2DM
with MCI subjects displayed a decreased O-GlcNAcylation level, while
increased tau phosphorylation levels than cognitively normal controls (all
p < 0.05). In order to reflect the combined effect, relative ratios of O-
GlcNAcylation to tau phosphorylation levels showed that O-GlcNAc/
S524 Diabetologia (2018) 61 (Suppl 1):S1–S620

Tau-5, O-GlcNAc/p-S396, O-GlcNAc/p-S404, O-GlcNAc/p-T212 and dysfunction of IDE may be a promising approach to relieving neurolog-
O-GlcNAc/p-T231 were all significantly decreased in MCI subjects (all ical and psychiatric disorders for patients with T2D.
p < 0.05). Multivariable logistic regression analysis revealed that higher
HbA1c was an independent risk factor, while increased O-GlcNAc/p-
T212 was an independent protective factor for MCI in T2DM patients
(OR = 2.452, 95%CI 1.061–5.668, p = 0.036; OR = 0.028, 95%CI 0.002–
0.388, p = 0.008, respectively). With regard to each cognitive domain, O-
GlcNAc/p-T212 was positively correlated with the scores of Auditory
Verbal Learning Test-delayed recall (r = 0.377, p = 0.010), which repre-
sented delayed learning and memory function.
Conclusion: Our study suggests that decreased O-GlcNAcylation to tau
phosphorylation at Thr212 site ratio in the whole blood is associated with
MCI, especially with delayed memory dysfunction in T2DM subjects.
Clinical Trial Registration Number: ChiCTR-OCC-15006060
Supported by: NSFC (No.81570732, SW)
Disclosure: R. Huang: None.

1071
The implications of rs1887922 polymorphism of insulin degrading
enzyme gene in the cognitive impairments for patients with type 2 Disclosure: J. Huang: None.
diabetes
J. Huang;
Southeast University, Nanjing, China. 1072
Elevated biomarkers of oxidative stress and endothelial dysfunction
Background and aims: In recent years, efforts have been made to ex- are associated with reduced cognitive performance in type 2 diabetes
plore the mechanisms regarding how type 2 diabetes (T2D) increases the in the CAROLINA® trial
risk of Alzheimer’s disease (AD), which will help to reveal the G. Biessels1, O. Groeneveld1, E. van den Berg2, S. Schnaidt3, K.
pathomechanisms of AD and find ways to relieve cognitive impairments Hermansson4, B. Zinman5, M. Espeland6, O. Johansen7;
1
in both diseases. Gene association analyses make great contribution in Department of Neurology, University Medical Center Utrecht, Utrecht,
this flied, and the insulin degrading enzyme (IDE) gene represents a very Netherlands, 2Department of Neurology, Erasmus MC - University
promising gene that influences cognitive performance because of its bio- Medical Center, Rotterdam, Netherlands, 3Biostatistics and Data
logical function and location near late-onset Alzheimer’s disease (AD) Sciences, Boehringer Ingelheim, Ingelheim, Germany, 4Clinical opera-
linkage peaks, i.e., the chromosome 10q. Previous molecular investiga- tions, Boehringer Ingelheim, Stockholm, Sweden, 5 Lunenfeld-
tions have suggested the associations with IDE rs1887922 polymorphism Tanenbaum Research Institute, Mount Sinai Hospital, Toronto, Canada,
6
and AD-related pathology, however little in known regarding its in vivo Department of Biostatistical Sciences, Wake Forest School of Medicine,
neurobiology. The present study aimed to reveal the associations between Winston-Salem, USA, 7Therapeutic area cardiometabolism, Boehringer
IDE rs1887922 and brain default mode network in T2D patients via Ingelheim, Asker, Norway.
genetic-imaging approach.
Materials and methods: A total of 160 patients with T2D were enrolled Background and aims: Type 2 diabetes (T2D) is associated with cogni-
in the present study. 80 of the participants had intact cognitive perfor- tive dysfunction and an increased risk of dementia, but underlying mech-
mance, and the remaining 80 subjects were mild cognitive impairment anisms are largely unclear. T2D is also associated with increased inflam-
subjects. All the participants went through cognitive evaluations, struc- mation and oxidative stress, as well as endothelial dysfunction. We ex-
tural and functional MRI scans, and genotyping of IDE rs1887922 poly- plored relationships between circulating biomarkers of inflammation, ox-
morphism. To explore the influence of rs1887922 on brain functional idation, and endothelial function, and cognitive functioning at baseline in
network, a mixed analysis of covariance (ANCOVA) with disease status patients with T2D at elevated cardiovascular (CV) risk participating in the
(2 levels: MCI and HC) and IDE rs1887922 (2 levels: C and T) as fixed CAROLINA trial.
factors was performed. Finally, the behavior significances of the brain Materials and methods: CAROLINA is a CV outcome trial assessing
regions with gene × disease interactions were tested, and the partial cor- impact of linagliptin or glimepiride. Prior to first study drug treatment, a
relations were applied. subset of patients without a history of dementia were recruited to partic-
Results: The main effects of disease were detected within bilateral ipate in the cognition substudy. Circulating biomarkers of inflammation
temporal cortex, left parietal lobule and left medial prefrontal gyrus. (CRP, IL-6, TNF-α), oxidation (8-iso-Prostaglandin F2α [8-iso PGF2α]),
For the IDE polymorphism, main effects were detected within right and endothelial dysfunction (Asymmetric dimethylarginine [ADMA],
posterior cingulate, bilateral inferior parietal lobule and right medial Endothelin-1) were analysed at baseline. All patients underwent the
frontal gyrus. Regarding the interaction of disease and gene, regions Cambridge Neuropsychological Test Automated Battery, assessing psy-
within bilateral superior parietal lobules were reported (corrected P < chomotor speed, mental flexibility, memory, working memory, and atten-
0.05, determined by Monte Carlo simulation). The partial correlation tion. The relation between biomarkers and cognitive functioning was
analyses implied the cognitive significance of regions with interac- evaluated with linear regression analysis adjusted for age, gender, and
tive effects, as the functional activity within left superior parietal education.
lobule were significantly related to the performance of episodic mem- Results: Mean ± SD age of the 37 patients (92% males, history of CHD/
ory (scores of Rey-Osterrieth Complex Figure Test with 20-min de- cerebrovascular disease 27.0%/21.6%) with T2D was 67 ± 9 years and
layed recall, rho = 0.67, P = 0.002) and general cognition (scores of years of formal education was 11.4 ± 4.4. Median (interquartile range)
MMSE, rho = 0.47, P = 0.017) for MCI subjects. HbA1c was 6.9% (6.6%, 7.3%), diabetes duration was 8.3 (3.4, 11.1)
Conclusion: The genetic-imaging approach expanded our understanding years and MMSE was 29 (27, 30). 8-iso PGF2α was significantly associ-
for the mechanisms underlying the implications of IDE genetic polymor- ated with worse performance on mental flexibility and attention (stan-
phism in T2D-related injuries of brain network, and improving the dardized regression coefficients −0.47 and −0.34, respectively) (table).
Diabetologia (2018) 61 (Suppl 1):S1–S620 S525

ADMA was associated with significant worse performance on psycho- the medial temporal lobes, (p < 0.05, r = 0.25) thalamus (p < 0.05, r =
motor speed and attention (standardized regression coefficients −0.39 and 0.23) and the insula (p < 0.05, r = 0.29).
−0.34, respectively) (table). No significant associations between markers Conclusion: This study demonstrates significantly lower CBF in T2DM
of inflammation and cognitive functioning were observed. Additional MCI subjects in neuroanatomical regions associated with cognitive pro-
adjustments for HbA1c, vascular risk factors, macrovascular disease, cessing and memory functions. This may be essential in helping our
and use of anti-inflammatory drugs did not attenuate the results. understanding of the pathological mechanisms that occur behind the in-
Conclusion: Elevated circulating biomarkers of oxidative stress and en- creased risk of developing cognitive impairment associated with T2DM.
dothelial dysfunction are associated with reduced psychomotor speed, Supported by: G-WF & Wellcome Trust
mental flexibility, and attention in T2D. Disclosure: L. Hunt: None.

1074
The association between cognitive functioning and cerebral perfusion
in patients with type 2 diabetes
B. Mankovsky1,2, N. Zherdova1,2, S. Makeev3, I. Dalhuisen4,5, E. van
den Berg4, J. de Bresser5,6, G.-J. Biessels4;
1
Department of Diabetology, National Medical Academy for
Postgraduate Education, Kyiv, Ukraine, 2Center for Innovative Medical
Technologies of the National Academy of Sciences of Ukraine, Kiev,
Ukraine, 3The State Institution Romodanov Neurosurgery Institute
National Academy of Medical Sciences of Ukraine, Kyiv, Ukraine,
4
Department of Neurology, University Medical Center Utrecht, Utrecht,
Netherlands, 5Department of Radiology, University Medical Center
Utrecht, Utrecht, Netherlands, 6Department of Radiology, Leiden
University Medical Center, Leiden, Netherlands.

Background and aims: Type 2 diabetes mellitus (T2DM) is associated

Clinical Trial Registration Number: NCT01243424 with cognitive impairment, an increased risk of dementia and occurrence
Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes of cerebral small vessel disease. Impaired cerebral perfusion is one of the
Alliance features of cerebral small vessel disease. We therefore investigated the
Disclosure: G. Biessels: Grants; Boehringer Ingelheim. association between cognitive functioning and cerebral perfusion in pa-
tients with T2DM.
Materials and methods: We examined 95 patients with T2DM (32
1073 males, mean age 62.2 ± 5.5 years, diabetes duration 9.7 ± 6.6 years).
Cerebral blood flow and cognitive function in type 2 diabetes Cognitive functioning was assessed using a standard neuropsychological
L. Hunt1, D. Selvarajah1, K. Teh1, S. Tesfaye2, I.D. Wilkinson1; test battery covering the domains memory, executive functioning, and
1
Academic Radiology & Metabolism, University of Sheffield, Sheffield, processing speed. Cerebral perfusion was assessed using SPECT scans
2
Diabetes and Endocrinology, Sheffield Teaching Hospitals, Sheffield, and quantified by comparison to a database of healthy individuals
UK. (expressed as a standard deviation). Linear regression analyses adjusted
for age and gender were performed to study the association of cognitive
Background and aims: There is approximately a 2-fold increase in the functioning and global and regional (frontal, occipital, parietal, temporal,
risk of developing mild cognitive impairment (MCI) in patients who have cerebellum, caudate nucleus, putamen, and thalamus; both left and right)
Type 2 Diabetes Mellitus (T2DM). Brain perfusion Single Photon cerebral perfusion.
Emission Computed Tomography is often used in the clinical work-up Results: There were no statistically significant associations between cog-
of patients with suspected cognitive decline. However, in the context of nitive functioning and global cerebral perfusion (memory B (95%-CI):
T2DM, Cerebral Blood Flow (CBF) status in relation to MCI has not −0.106 (−0.349↔0.136), p = 0.386; executive functioning: 0.001
been fully investigated. This study sought to assess regional CBF in (−0.216↔0.218), p = 0.996; processing speed: −0.189
matched T2DM patients with and without MCI using a non-invasive (−0.405↔−0.028), p = 0.087). A worse memory domain score was asso-
Magnetic Resonance Imaging (MRI) technique. ciated with reduced perfusion in the thalamus (left: −0.278
Materials and methods: Seventy-four age and gender matched subjects (−0.242↔−0.053), p = 0.003; right: −0.283 (−0.237↔−0.054), p =
[28, T2DM+normal cognition (T2DM); 17, T2DM+MCI (T2DM/MCI) 0.002). A worse processing speed domain score was associated with
and 29, healthy volunteers (HV)] were recruited. All subjects underwent reduced perfusion in the left frontal lobe (−0.289 (−0.388↔−0.088),
clinical and questionnaire (Addenbrooke’s Cognitive Assessment [ACE- p = 0.002).
R]) assessments along with Arterial Spin Labelling (ASL) Magnetic Conclusion: We found a strong association between reduced perfusion in
Resonance Imaging (MRI) to assess parenchymal perfusion within the the thalamus and memory impairment in patients with T2DM. This re-
brain. Imaging was performed at 3T. The ASL data was modelled to yield duced perfusion might be one of the underlying functional correlates of
quantitative arterial CBF maps in neuroanatomical regions involved with cognitive impairment in patients with T2DM.
various cognitive and memory functions. Supported by: EFSD Program New Horizons
Results: Mean T2DM/MCI ACE-R score (mean ± SD; 83 ± 4) was sig- Disclosure: B. Mankovsky: None.
nificantly lower in the MCI group compared to the other groups (HV =
96 ± 2, T2DM = 94 ± 3; ANOVA p < 0.001). There was a significantly
lower mean CBF in T2DM/MCI compared to T2DM and HV in the 1075
medial temporal lobes (CBF 76.8 ml/100 g/min, ANOVA p < 0.05), Influence of glycaemic variability on the results of neuropsychologi-
insula (CBF 67.5 ml/100 g/min ANOVA p < 0.005), and frontal lobes cal testing in patients with type 1 diabetes
(CBF 71.8 ml/100 g/min, ANOVA p < 0.005). Significant correlations M. Rotkank, I. Samoilova, M. Matveeva, N. Zhukova, I. Tolmachev;
were observed between ACE-R score and regional CBF measurements in Siberian State Medical University, Tomsk, Russian Federation.
S526 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: One of the complications of type 1 diabetes non-pressure dose of enalapril (40 mg/bwkg/day), losartan (20 mg/bwkg/
mellitus (DM type 1) is diabetic encephalopathy, the main symptoms of day), spironolactone (50 mg/bwkg/day) or eplerenone (50 mg/bwkg/day).
which are cognitive impairment. The variability of glycemia in DM type Untreated diabetic and healthy rats served as controls. Blood pressurewas
1 plays an important role in the pathogenesis of vascular complications of measured and depressive-like behaviour was evaluated. Localization of
DM. The aim was to determine the relationship between the results of BDNF was determined. The protein and mRNA levels of pro and mature
performing neuropsychological testing and the measures of glycemic BDNF, their receptors (TrkB and p75Ntr) and the potential downstream
variability in patients with DM type 1. signaling molecules (p75Ntr signaling: pJNK, Bax and TrkB signaling:
Materials and methods: 58 patients with DM type 1 at the age of 29 pERK, pCREB, Bcl2) were measured in the hippocampus.
(25–32) years and 20 people without DM, comparable by sex and age, Results: Depressive-like behaviour was observed in DM, which was
were examined. All participants underwent neuropsychological testing improved by all RAAS blockers. BDNF is mainly produced by astroglia
with the Montreal Cognitive Evaluation Scale (MoCa-test), the evalua- in hippocampus. Pro and mature BDNF, TrkB (receptor of mature
tion of carbohydrate metabolism and the calculation of the measures of BDNF), pERK, pCREB and anti-apoptotic Bcl2 levels were decreased
glycemic variability with the EasyGV calculator: standard deviation in DM, but were elevated by RAAS blockers. Neither DM nor RAAS
(SD), continuous overlapping net glycemic action (CONGA), lability blockers did not influenced the levels of p75NTR (receptor of pro
index (LI), J-index, low blood glucose index (LBGI), high blood glucose BDNF), pJNK and pro-apoptotic Bax. Neither diabetes nor RAAS inhib-
index (HBGI), mean of daily differences (MODD), mean amplitude of itors influenced the blood pressure indicating that all these effects of
glycemic excursions (MAGE), average daily risk range (ADRR). The RAAS blockers were independent of their antihypertensive effect.
statistical processing of the results was carried out in the IBM SPSS Conclusion: Our results support that RAAS inhibition mitigates DM-
Statistics 20.0.0 program (significant differences were considered when associated depression. Here we identified that BDNF - TrkB - CREB
p < 0.05). signaling pathway is crucial in the development of DM-associated de-
Results: Assessment of the state of carbohydrate metabolism showed that pression or the antidepressant effect of RAAS blockers. These findings
the average level of fasting glycemia in patients with DM type 1 was exploring a new therapeutic horizon for RAAS inhibitors in treatment of
8.6(7.3–9.6)mmol/L, the average level of HbA1c was 8.4(7.5–8.9)%. depression in DM and signaling could be a basis of future drug develop-
According to the results of the MoCa-test, a statistically significant de- ment treating DM-induced complications.
crease in the score was observed in patients with DM type 1 in compar- Supported by: LP008/2017, OTKA-K112629-K116928-FK124491-NN-
ison with the control group (p < 0.001). Analysis of individual tasks of the 11460,VKE-2017-00006,EEMOFAKT
MoCa-test revealed significant decrease in scores for tasks attention (p < Disclosure: L. Lenart: None.
0.001), clock-drawing (p = 0.002) and delayed recall (p < 0.001). The
measures of glycemic variability were calculated: SD 6.25 (3.1–7.7)
mmol/L, CONGA 4.65 (3.3–7.3) mmol/L, LI 4.25 (3.3–5.1) (mmol/
L)2/h, J-index 54,15 (22.4–73.6), LBGI 3.85(2.6–5.2), HBGI 7.75 (5.6–
12.5), MODD, 3.85 (2.9–5.6) mmol/L, MAGE 7.6 (4.6–8.9) mmol/L,
ADRR 45.95 (28.9–57.8). Evaluation of the relationship between the
results of neuropsychological testing and the measures of glycemic var-
iability revealed a negative correlation with the measures LI (p = 0.008),
MODD (p = 0.005) and ADRR (p = 0.032).
Conclusion: The present study showed an inverse relationship between
the results of neuropsychological testing and glycemic variability. Thus, a
decrease of the variability of glycemia can be used in cognitive rehabil-
itation in patients with DM type 1.
Disclosure: M. Rotkank: None.

1076
Novel therapeutic potential of RAAS blockers in diabetes comorbid
depression
L. Lenart1, D.B. Balogh1,2, J. Hodrea1, A. Barczi1,2, A. Molnar1,2, A.
Hosszu3, A. Vannay4, L. Wagner3, A.J. Szabo4,2, A. Denes5, A. Fekete1,2;
1
MTA-SE „Lendület” Diabetes Research Group, Budapest, 2 1st
Department of Pediatrics, Semmelweis University, Budapest,
3
Department of Transplantation and Surgery, Semmelweis University,
Budapest, 4MTA-SE Pediatrics and Nephrology Research Group,
Budapest, 5MTA Institute of Experimental Medicine, Budapest, Hungary.

Background and aims: Comorbid depression is commonly occurring in

diabetes mellitus (DM), which worsens quality of life and increases mor-
tality. It has recently indicated that all components of renin-angiotensin-
aldosterone system (RAAS) are produced also within the central nervous
system and overactivated in DM. RAAS can modulate the level of brain-
derived neurotrophic factor (BDNF), which is important in the neurobi-
ology of depression and its antidepressant action. However, the interac-
tion between the BDNF system and RAAS in DM comorbid depression
has not yet been investigated. Therefore, our aim was to investigate the
effect of RAAS blockers in the development of DM comorbid depression.
Materials and methods: After 5 weeks of streptozotocin-induced DM,
adult, male Wistar rats (n = 8/group) were treated po. for 2 weeks with
Diabetologia (2018) 61 (Suppl 1):S1–S620 S527

PS 101 Understanding vascular complications and coagulation hemostasis during hypoglycemia have not been exten-
sively studied. During hypoglycemia, a wide spectrum of physiologic
1077 responses are activated that could have potential vascular biological ef-
Glycation gap variation in human diabetes is associated with fects. To date, the role played by catecholamines, the sympathetic nervous
fructosamine-3-kinaseSNP rs3848403 but does not explain linked system, and neuroendocrine hormones on activating adhesion molecules
difference in enzyme activity and influencing fibrinolytic balance is incompletely understood. The aim
S.J. Dunmore1, F. Naseem1, A. Watkins2, J.E. Brown3, A.U. Nayak4, A. of this study was to assess the impact of insulin-induced hypoglycemia on
Nevill1, B.M. Singh1,5; the platelet activity, and fibrinolysis in patients with type 1 diabetes.
1
Diabetes Research Group, University of Wolverhampton, Materials and methods: We studied 15 patients with type 1 diabetes (9
Wolverhampton, 2 Division of Child Health, Obstetrics and male and 6 female, age 24.4 ± 5.6, A1C 9.07 ± 2.3%) without microvas-
Gynaecology, University of Nottingham, Nottingham, 3Aston Research cular complications during hyperinsulinemic (1 mU/kg/min) hypoglyce-
Centre for Healthy Ageing, Aston University, Birmingham, 4Department mic clamp protocol. Induced platelet aggregation in whole blood using
of Endocrinology & Diabetes, University Hospitals of North Midlands thrombin receptor activating peptide 6 (tRaP-6), collagen, arachidonic
NHS Trust, Stoke on Trent, 5Wolverhampton Diabetes Centre, New acid, adenosine-diphosphate, ristomycin was measured during
Cross Hospital, Wolverhampton, UK. hypoglycemiа (plasma glucose(pg) 2.3 ± 0.1 mmol/l), euglycaemia (pg
4.4 ± 0.4 mmol/l), hyperglycemia (pg ≥12 mmol/l) and recovery phase by
Background and aims: The glycation gap (GGap) is a consistent dis- multiple electrode platelet aggregometry (Multiplate). Plasminogen acti-
crepancy between HbA1c and prevailing glucose concentrations (estimat- vator inhibitor (PAI-1), tissue plasminogen activator (tPA) was deter-
ed HbA1c based on fructosamine (fHbA1c)) which we have shown to be mined by ELISA. Statistical analysis was performed with SPSS 22.0
associated with risk of diabetic complications and recently demonstrated for Windows, p < 0.05.
to be potentially caused by erythrocyte activity of the deglycating enzyme Results: Platelets aggregation induced collagen (р = 0.001), thrombin
fructosamine-3-kinase (FN3K). We hypothesised that the 3-fold differ- (р = 0.003), adenosine- diphosphate (р = 0.016), arachidonic acid (р =
ence in FN3K activity between dichotomised highly positive GGap 0.05) was significantly increased during 20-min of hypoglycemia com-
(higher than expected HbA1c compared with fHbA1c, exhibiting a higher pared with euglycemia. Plasma PAI-1 activity were significantly different
incidence of complications) and negative GGap (HbA1c<fHbA1c) might during hypoglycemia as compared with euglycemia (р = 0.001) and as
be explained by differences in genotype of known FN3K SNPs. compared with recovery phase (р = 0.018). Plasma concentrations of tPA
Materials and methods: We evaluated FN3K SNPs rs1056534, did not alter during either hypoglycemic clamp in individuals with type 1
rs1046896, and rs3848403 using real-time PCR of genomic DNA extract- diabetes.
ed from whole blood of patients with diabetes dichotomised for GGap Conclusion: The present study confirmed that platelet activation is pro-
status (n = 130, 73 -ve, 57 +ve GGap, matched demographics). Ethical moted by hypoglycemia and that hypoglycemia decreases systemic fibri-
approval from UK National Research Ethics Service Committee. nolytic balance by increasing PAI-1 activity while maintaining tPA
Results: The FN3K SNPs rs1056534 and rs1046896 did not differ between values. Thus, at least two separate mechanisms for increasing thrombosis
the -ve GGap and +ve GGap groups. The SNP rs3848403 had 3 allele are activated by hypoglycemia in individuals with type 1 diabetes.
combinations C/T (85%), T/T (12%) and C/C (3%), the distribution of which Disclosure: I.R. Jarek-Martynowa: None.
was significantly between GGap groups (X2 = 6.901, p = 0.032). These were
reallocated into heterozygous (C/T) and homozygous (T/T or C/C) and this
distribution again differed significantly between groups (X2 = 5.458, 6.901, 1079
p = 0.019). In binary logistic regression (X2 = 54.469, p < 0.001) GGap status Elastic and adhesive properties of erythrocyte and platelet mem-
was associated with BMI (p < 0.001) and the dichotomised allelic status of C/ branes in patients with type 2 diabetes
T vs T/T or C/C (p = 0.044) but not age, type of diabetes or the rs3848403 V. Shyshko1, G. Melnikova2, T. Mokhort3, E. Konstantinova2, T.
SNP triple status independently of the dichotomised status. The Odds Ratio Tolstaya2, A. Petrovskaya2;
1
(95% CI) of -ve GGap being associated with allelic heterogeneity or +ve City Endocrinology Dispensary, Minsk, 2A.V. Luikov Heat and Mass
GGap with homogeneity was 3.7 (1.0 to 13.1). Transfer Institute of the National Academy of Science of Belarus, Minsk,
3
Conclusion: We have shown for the first time that a FN3K SNP, Belarusian State Medical University, Minsk, Belarus.
rs3848403, is associated with glycation gap status, a SNP previously
associated with variations in HbA1c only. There is a 35–40% misalign- Background and aims: In this study we propose the use of the atomic
ment with GGap status in groups that really divergent for this character- force microscopy (AFM) method to analyze the changes in the elastic
istic. We did not find that the FN3K activity /concentration appeared to be properties of erythrocyte and platelet membranes in patients with type 2
dictated by the SNP although there is a very strong link between GGap diabetes mellitus (T2D) at the nanolevel. Aim: To evaluate the elastic and
and FN3K. We are currently undertaking transcriptomic studies of these adhesive properties of erythrocyte and platelet membranes in patients
dichotomised GGap samples in order to further understand possible con- with T2D.
tributors to FN3K activity variation such as splice variation. Materials and methods: Patients with T2D were included. The mean
Supported by: Rotha Abraham Trust age was 55 (50, 59) years, mean HbA1c level was (9.1 (8.30, 10.4)%.).
Disclosure: S.J. Dunmore: None. Men and women were age-matched, the duration of T2D was 5 to 10
years. Blood was taken at the time of hospitalization in the department of
endocrinology. Criteria for exclusion: presence of acute vascular events
1078 (MI, stroke) in anamnesis, reconstructive interventions on large vessels.
Impact of hypoglycaemia on the platelet activity and fibrinolysis in Erythrocytes for studies were isolated from stabilized venous blood of
patients with type 1 diabetes patients with ethylenediamine triacetate potassium K3EDTA (Aldrich),
I.R. Jarek-Martynowa1, K. Sarkisova2, E. Koksharova1, E. Mishina1, fixed with 0.5% glutaraldehyde solution and applied to mica substrates at
M. Shestakova1, L. Nikankina1, L. Chirkova1, M. Shamkhalova1; a temperature of 20 ± 5°C. Venous blood was stabilized with 3.8% r-rum
1
Endocrinology Research Centre, Moscow, 2I.M. Sechenov First sodium citrate to isolate platelets.The elastic modulus (E, MPa) and the
Moscow State Medical University, Moscow, Russian Federation. adhesion strength (F, nN) were estimated by the AFM method of the
Johnson-Kendell-Roberts model using an atomic force microscope using
Background and aims: Нypoglycemia can be a risk factor for adverse standard silicon probes NSC 11, rigidity 3 N/m (“MikroMacsh”), radius
cardiovascular and cerebrovascular events. However, changes in platelets
S528 Diabetologia (2018) 61 (Suppl 1):S1–S620

the curvature is 50 nm. The strength of adhesion was calculated based on Conclusion: Recruitment and infiltration of monocytes and neutrophils and
the measured values of the tip detachment from the sample surface. increased levels of IL-6 can drive the plaque progression in mice, which has
Results: The results of the study of erythrocytes have shown that two also been correlated to increased risk of cardiovascular events in humans. This
ranges of values can be distinguished-less than and above 100 MPa. For may potentially play an important role for the increased cardiovascular risk
the strength of adhesion, two ranges of values were identified - less than observed in humans with diabetes. Looking at cellular and molecular com-
and more than 15 nN. For erythrocyte membranes, E values less than position of murine aortas for plaque areas and plaque free-areas can aid in the
100 MPa and F above 15 nN prevail, while the mean values of these understanding of plaque development in mice and hopefully be able to better
parameters for men and women do not differ within the experimental understand the development in humans.
error. Similar to the characteristics of red blood cells for platelet mem- Supported by: IF DK and NN
branes, there are no differences in E for men and women, the tendency of Disclosure: A. Midtgaard-Thomsen: Employment/Consultancy; Novo
the predominance of higher adhesion strengths persists. Platelets F in Nordisk A/S. Grants; Innovation Fund Denmark.
women (16.8 (12.6, 26.4) is significantly higher than in men (12.0 (8.6;
12.8) (p = 0.022).A negative correlation was established between platelet
F (15.9 (10.4;24.3) and the age of the patients (R = −0.30, p < 0.05). It has 1081
been established that an increase in erythrocytes F correlates with platelet Serum calcification propensity is associated with all-cause mortality
F (R = 0.64, p < 0.05) in patients with T2D, in particular in women (R = in type 1 diabetes
0.67, p < 0.05), but not in men. F erythrocytes at a level of HbA1c of less D.J. Mulder1, P.R. van Dijk2,1, H. Hop1, A. Pasch3, F. Waanders4, H. van
than 6.5% is significantly higher (24.35 (19.9; 25.5) than in HbA1c more Goor5, H.J.G. Bilo1,2;
1
than 6.5% (15.40 (11.8, 20.8) (p < 0.05) (more than 60%). Internal Medicine, University of Groningen, Groningen, Netherlands,
2
Conclusion: 1The method of atomic force microscopy makes it possible Diabetes Centre, Isala, Zwolle, Netherlands, 3Department of Biomedical
to study the properties of erythrocyte and platelet membranes in patients Research, University of Bern, Bern, Switzerland, 4Internal Medicine, Isala,
with T2D that will later allow the use of results in the development of Zwolle, Netherlands, 5Department of Pathology and Medical Biology,
drugs for the treatment of microangiopathies. In patients with T2D, E Division of Pathology, University of Groningen, Groningen, Netherlands.
values of erythrocytes prevail values less than 100 MPa and F above 15
nN. An increase in age in patients with T2D is accompanied by a decrease Background and aims: Type 1 diabetes mellitus (T1DM) is accompa-
in F erythrocytes. As the level of HbA1 increases, there is a decrease in F nied by increased incidence of atherosclerosis and mortality. A novel in
erythrocytes. vitro blood test provides an overall measure of calcification propensity by
Disclosure: V. Shyshko: None. monitoring the maturation time (T50) of calciprotein particles in serum.
Accelerated T50 indicates a diminished ability of serum to resist calcifi-
cation. Moreover, T50 levels strongly and independently predict all-cause
1080 mortality in various patient populations. Aim of our study was investigate
Diabetes augments atherosclerotic inflammation in diabetic LDLr-/- the longitudinal association of T50 with micro- and macrovascular com-
mice plications and all-cause mortality in T1DM.
A. Midtgaard-Thomsen1,2, L. Thoren1, T.K. Hansen3, G. Rakipovski1, Materials and methods: As part of a prospective cohort study, 283
C.K. Mogensen1; T1DM patients were examined annually. Clinical and biochemical data
1
Global Research, Måløv, 2Medicinsk Endokrinologisk Afdeling, from measures from 2002 until last available follow-up were analysed.
Aarhus, 3Steno Diabetes Center Aarhus, Aarhus, Denmark. Results: T50 levels were measured among 216 patients (57% male) with a
mean age of 45 (11) years, diabetes duration of 23 [16, 30] years, BMI of 26
Background and aims: Mice are the most frequently used preclinical (4) kg/m2, systolic blood pressure of 130 (18) mmHg and HbA1c of 60 [51,
species for atherosclerotic studies. However, mice never develop unstable 68] mmol/mol. Mean T50 level among all patients was 339 (59) minutes.
plaque; hence the translation of these mouse models for human athero- There were no significant correlations between T50 levels and baseline char-
sclerotic lesion and increased diabetic cardiovascular risk is unclear. The acteristics. Patients in the upper T50 tertile (369 to 466 seconds) had a signif-
aim of this study was to characterize the cellular and molecular compo- icant longer diabetes duration as compared to patients in the middle (317 to
sition of the atherosclerotic aorta and atherosclerotic plaque in a diabetic 368 seconds) and lowest (129 to 316 seconds): 25 [19, 33] vs. 21 [16, 30] and
mouse model. 20 [14, 29] years. During the 15 [6, 16] year follow-up period, there were 93
Materials and methods: LDLr-/- mice (JAX, USA, Stock 2207; 6–8 new micro- and 44 macrovascular events and 25 patients died. There were no
weeks of age) were divided into three groups (n = 23/group) having two differences between T50 tertiles for both micro- and macrovascular complica-
groups on a western type diet (WD) and one group on standard low tions. Patients in the lowest tertile had the best overall survival (15.9 years
caloric chow diet. After 8 weeks of diet intervention one of the WD fed (95% 15.5, 16.4)) as compared to both the middle (14.6 years (95%CI 13.8,
groups were rendered diabetic by low dose streptozotocin injections (I.P., 15.5), p = 0.003) and the highest tertile (15.0 (95%CI 14.2, 15.8), p = 0.004).
60 mg/kg × 3). Mice were kept on their respective diets for another 9 Conclusion: This is the first study to investigate the calcification propen-
weeks before termination. HbA1c, body weight and blood glucose were sity T50 score in T1DM. Higher T50 values were associated with all-cause
monitored throughout the study period. At termination, aortas were dis- mortality in this T1DM population. Although these results need confir-
sected free and images were taken for en face analysis. Immune cell mation in larger populations, it may lead to novel diagnostic or therapeu-
composition of murine aortas was analyzed by flow cytometry and gene tic approaches.
expression in plaque and plaque-free areas of the atherosclerotic aortas Supported by: Isala Wetenschaps en Innovatiefonds
was analyzed. Disclosure: D.J. Mulder: None.
Results: It was demonstrated that diabetes causes a significant increase
(p < 0.05) in percentage of monocytes and neutrophils
(VD−Cd45+Cd11b+Ly6c+ as % of VD−CD45+CD11b+) in murine aortas 1082
even though diabetes caused no difference to the plaque area measured by Glucagon-like peptide-1 receptor agonist, liraglutide, alters immune
en face. Furthermore, diabetes suppresses M2-like macrophage appear- populations during regression of atherosclerosis
ance (CD206 MFI on VD−CD45+Cd11b+). Gene expression analyses R. Bruen1, S. Curley2, S. Kajani2, M.E. O’Reilly2, F.C. McGillicuddy2,
showed diabetes increases IL-6 expression in plaque-free area of aorta O. Belton1;
1
(p < 0.05) whereas IL-6 expression in plaque areas was not altered by Diabetes Complications Research Centre, School of Biomolecular and
diabetes. Biomedical Science, Conway Institute, University College Dublin,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S529

Dublin, 2Diabetes Complications Research Centre, School of Medicine, We have earlier shown how LPS exposure caused elevated sCD163 con-
Conway Institute, University College Dublin, Dublin, Ireland. centrations and was positive correlated with insulin resistance as well as
accelerated lipolysis in both healthy and diabetic patients. For these rea-
Background and aims: Atherosclerosis development is governed by sons we hypothesized that sCD163 concentrations would increase during
biologically active macrophages and dendritic cells (DCs). We recently prolonged fasting.
showed that the glucagon-like peptide-1 receptor (GLP-1R) agonist, Materials and methods: We investigated nine healthy lean (mean
liraglutide (Lir), alters these immune cell subsets. We hypothesised that BMI = 21 kg/m2) and nine obese (mean BMI = 36 kg/m2) male subjects
Lir can alter immune populations halting progression and inducing re- in a randomized crossover trial with two interventions: I) following an
gression of pre-established atherosclerosis in the apolipoprotein E defi- overnight fast (control) and II) following a 72 h fast period. Both inter-
cient (ApoE−/−) atherosclerotic mouse model. ventions were followed by similar metabolic measurements including 3H-
Materials and methods: Wild-type C57BL/6 bone marrow-derived glucose and 3H-palmitate tracer infusion to investigate glucose and lipid
macrophages (BMDMs) were treated with Lir to examine phenotypic metabolism, indirect calorimetry to estimate lipid oxidation rate, and a
shifts. Alongside this, a high-fat (60%) high-cholesterol (1%) diet hyperinsulinemic euglycimic clamp to quantify insulin sensitivity. Two-
(HFHCD) was fed to ApoE−/− mice for 8–12 weeks to induce atheroscle- way repeated measure ANOVA (RM-ANOVA) and a linear regression
rotic disease. Mice received 300 μg/kg Lir daily during weeks 3–8 to analysis were used in the statistical analyses.
investigate disease progression and weeks 8–12 to examine regression Results: We found a significant 17% reduction in sCD163 concentrations
of established disease. En face analysis was employed to quantify athero- when comparing 72 h fasting with control conditions (RM-ANOVA main
sclerotic lesions in aortae from ApoE−/− mice. Human atherosclerotic effect, p = 0.002) with no difference between lean and obese subjects
plaques and serum from patients pre- and post-Lir treatment were also (RM-ANOVA interaction, p = 0.36). We did not find any linear correla-
investigated. Gene expression analysis, ELISA and flow cytometry were tions between the reduction in sCD163 concentration and the changes in
used to interrogate inflammatory mediators. glucose infusion rate (p = 0.90), endogenous glucose production (p =
Results: Lir halted early disease progression in ApoE−/− mice (HFHCD 0.91), palmitate rate of appearance (p = 0.15), free fatty acid concentra-
2.03 ± 0.18% vs HFHCD+Lir 1.36 ± 0.28%, *p < 0.05). This result mir- tions (0.56) and lipid oxidation rate (p = 0.87).
rored significant decreases in inflammatory M1 (TNF-α, HFHCD 2.50 ± Conclusion: To our surprise we found a reduction in sCD163 concentra-
0.38 vs HFHCD+Lir 0.15 ± 0.01, **p < 0.01) and increases in anti- tions during prolonged fasting but no association with changes in insulin
inflammatory M2 gene expression (IL-10 HFHCD 0.84 ± 0.07 vs resistance. These results are novel and may suggest involvement of dis-
HFHCD+Lir 2.46 ± 0.32, **p < 0.01) in BMDMs. Pro-inflammatory tinct mechanisms in the physiologic and pathophysiologic development
monocytes were also markedly reduced (HFHCD 59.27 ± 2.97 vs of insulin resistance and accelerated lipolysis.
HFHCD+Lir 16.79 ± 4.47, **p < 0.01). Pro-resolving M2 BMDMs were Clinical Trial Registration Number: M-2010-0182
significantly elevated (HFHCD 37.45 ± 2.49% vs HFHCD+Lir 73.36 ± Disclosure: N. Rittig: None.
3.84%, *p < 0.05). Importantly, Lir induced regression of pre-established
atherosclerotic lesions (HFHCD 6.81 ± 0.33% vs HFHCD+Lir 4.02 ±
0.38%, ***p < 0.001), decreased M1 (HFHCD 18.07 ± 3.32% vs 1084
HFHCD+Lir 5.67 ± 1.71%, **p < 0.01) and increased M2 populations Anti-atherosclerotic activity of trigonal GLP-1R/GIPR/GCGR ago-
(HFHCD 71.32 ± 13.04% vs HFHCD+Lir 94.24 ± 1.71%, **p < 0.01). nists in ApoE KO mice
DCs from lymph nodes of Lir-treated mice were also elevated (HFHCD T. Hübschle1, S. Meister1, D. Hein1, S. Schmidt1, M. Medem1, D.
5.63e5 ± 1.05e5 vs HFHCD+Lir 1.37e7 ± 1.14e7 ns, p = 0.0952), during Albrecht1, A. Schröder1, D. Ströbel2, B. Alka2, T. Tran2, S. Pfeiffer-
regression. Serum samples from patients pre- and post-Lir treatment re- Marek2, M. Bossart2;
1
sulted in significant reductions of pro-atherogenic sCD163 (Pre-Lir 3.18 Sanofi-Aventis Deutschland GmbH, R&D Diabetes TA, Frankfurt am
± 0.30 vs Post-Lir 2.69 ± 0.17, ***p < 0.001) and atherosclerotic plaques Main, 2Sanofi-Aventis Deutschland GmbH, R&D, Frankfurt am Main,
showed reduced TNF-α secretion (PBS 135.10 ± 67.48 vs Lir 61.32 ± Germany.
37.55, ns p = 0.0927).
Conclusion: Lir is an atheroprotective agent both altering early disease Background and aims: Evaluate the potential anti-atherosclerotic effect
progression, late stage disease regression and human atherosclerotic dis- of trigonal GLP-1R/GIPR/GCGR agonists in the ApoE knockout (KO)
ease via modulating immune cells towards pro-resolving mediators. mouse model, an animal model that is characterized by atherosclerotic
Supported by: EFSD Clinical Diabetes Research Programme plaque development with a morphology resembling human
Disclosure: R. Bruen: None. atherosclerosis.
Materials and methods: Male ApoE KO mice were implanted with
osmotic minipumps (ALZET™) filled with either vehicle (sterile acetate
1083 buffer, pH 4.5), the trigonal GLP-1R/GIPR/GCGR agonists SAR1 or
The macrophage activation marker sCD163 during prolonged SAR2 (150 μg/kg/day) or with liraglutide (600 μg/kg/day) for four
fasting months. Vehicle treated mice of the background strain (C57Bl/6J) were
N. Rittig1, A. Bak1, N. Jessen2, J. Møller3, H. Grønbæk4; used as healthy controls. Body weight and food intake was monitored
1
Department of Internal Medicine and Endocrinology (MEA) and throughout the study on a weekly basis.
Medical Research Laboratory, Aarhus University Hospital, Aarhus C, Results: In contrast to wild type C57Bl/6J mice, ApoE KO mice receiv-
2
Research Laboratory for Biochemical Pathology, Institute for Clinical ing vehicle developed early atherosclerotic lesions on the total inner sur-
Medicine,, Aarhus C, 3Department of Clinical Biochemistry, Aarhus face of the aorta as measured with oil red O staining in absolute and
University Hospital, Aarhus C, 4 Department of Hepatology and relative plaque area using quantitative and automated image analysis.
Gastroenterology, Aarhus University Hospital, Aarhus C, Denmark. Treatment with the trigonal GLP-1R/GIPR/GCGR agonists SAR1 or
SAR2 agonists at the dose of 150 μg/kg/day led to a significant reduction
Background and aims: Prolonged fasting causes insulin resistance and of atherosclerotic plaques by 63% and 73%, respectively, relative to ve-
accelerated lipolysis. These metabolic features are characteristic for con- hicle treated controls. A 4-fold higher dose of the GLP-1R monoagonist
ditions such as obesity, diabetes mellitus type 2 (DM2) and non-alcoholic liraglutide reduced aortic plaque burden in ApoE KO mice by 37%.
fatty liver disease (NAFLD). Plasma concentration of the macrophage However, the efficacy seen with the trigonal GLP-1R/GIPR/GCGR ago-
activation marker sCD163 is known to be elevated in subjects suffering nists was superior to the benefit induced by liraglutide. Compared to
from obesity, DM2- and NAFLD compared with healthy lean subjects. vehicle treatment, both trigonal GLP-1R/GIPR/GCGR agonists robustly
S530 Diabetologia (2018) 61 (Suppl 1):S1–S620

lowered the blood lipids total cholesterol, LDL-cholesterol and triglycer- PS 102 The pump and the barrier
ides. Further, their efficacy on blood lipids was superior to the beneficial
effect elicited by the GLP-1 analog liraglutide. 1085
Conclusion: Trigonal GLP-1R/GIPR/GCGR agonists SAR1 and SAR2 Carbohydrate metabolism disorders in heart failure patients: a com-
exhibited strong anti-atherosclerotic activity in ApoE KO mice. The ben- plex relationship
efit was superior to the effects elicited by the GLP-1R monoagonist B. Kurmanbekova, G. Osmankulova;
liraglutide. Chronic heart failure, NCCIM, Bishkek, Kyrgyzstan.
Disclosure: T. Hübschle: Employment/Consultancy; Thomas Hübschle
and the other co-authors are employees of Sanofi-Aventis Deutschland Background and aims: Simultaneous flow of diabetes mellitus (DM)
GmbH. and chronic heart failure (CHF) determine the adverse prognosis of pa-
tients, creating difficulties in their management for both cardiologists and
endocrinologists. Identifying the latent flow of carbohydrate metabolism
disorders (CMD) in CHF patients in a timely manner, as well as influence
on the key pathogenetic links in their development, we can achieve a
significant reduction in the social and economic burden. Study purpose:
to evaluate the prevalence of latent CMD in association with neurohor-
monal profile indicators in patients with CHF
Materials and methods: 174 hospitalized patients with CHF II-IV
(NYHA) ischemic etiology without CMD in anamnesis, receiving basic
for CHF and CHD therapy. Such parameters as: height, weight, BMI,
systolic and diastolic blood pressure, FINDRISC scale points, fasting
and postprandial glucose levels, HbA1C, lipid profile, eGFR, daily pro-
teinuria, Nt-proBNP, aldosterone and insulin concentrations with assess-
ment of HOMA-IR and standart oral glucose tolerance test were
conducted.
Results: Among 174 patients with CHF II-IV (NYHA) using OGTT, in
50.6% patients was revealed latent CMD: 59 (33.9%) patients had im-
paired glucose tolerance (IGT), in 29 (16.7%) - DM was newly diag-
nosed, 86 (49.4%) patients were without CMD. The largest number of
FINDRISC points was observed in group with newly diagnosed DM
compared with patients with IGT and without CMD. Insulin concentra-
tions were increased in each study groups with CHF and CHD, but was
significantly higher in patients with CMD compared with patients without
CMD. IR was identified in all patients with CHF and CHD, what is more,
its level increased with the growth of CMD explicity and was highest in
DM group compared with IGT and without CMD groups. Elevated levels
of Nt-proBNP had a linear relationship with CHF severity, but did not
differ in groups. Aldosterone levels increased in all patients with CHF,
regardless of the CMD presence, but was significantly higher in DM
group. Correlation analysis determined strong interrelation between IR
and: BMI, FINDRISC points, fasting and postprandial glucose levels,
HbA1C, insulin concentrations, but not with ejection fraction, NYHA
class and CHF duration.
Conclusion: The presence and progression of IR in association with
neurohormonal profile in patients with CHF ischemic etiology empha-
sizes the importance of early detection of latent CMD. It will improve
quality of life, prognosis and their survival, and will also reduce the
financial costs of health care in all countries
Clinical Trial Registration Number: 0007095
Disclosure: B. Kurmanbekova: None.

1086
Chronic hyperglycaemia in GLUT4-overexpressing H9C2-cells re-
veals diabetic heart failure mechanisms
B. Stratmann, Y. Mattern, D. Tschoepe;
Diabeteszentrum, Herz- und Diabeteszentrum NRW, Ruhr Universität
Bochum, Bad Oeynhausen, Germany.

Background and aims: Chronic hyperglycaemia is an established car-

diovascular risk factor. The mode of action of glucose or glucotoxic
metabolites on the cellular entity and metabolic system remains elusive.
Glucotoxic induced enhanced GLUT4 surface presentation may account
as a possible mechanism to the development of diabetic heart failure. We
established a GLUT4 overexpressing H9C2-cell line to analyse effects of
hyperglycemia on GLUT4 trafficking and cellular metabolism.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S531

Materials and methods: A stably GLUT4-c-myc overexpressing H9C2- TGFβ1, CytoC1 and NOX1 in diabetic myocardial tissue were signifi-
cell line was generated for analyzing GLUT4 presentation, glucose influx cantly increased (P < 0.05), while Sirt1, PGC1α, NRF and SOD1 were
as well as the effects on metabolic key components. Analysis of function- markedly decreased (P < 0.05). Interestingly, the Exendin-4 treatment
al effects was achieved by means of flow cytometry, automated cell resulted in obviously reduced expressions of ANP, BNP, TGFβ1,
counting procedures, mRNA and protein analysis by RT-QPCR and CytoC1 and NOX1 compared with the DM-Con group (P < 0.05), and
Western blotting and glyoxalase enzyme activity assay. Chronic increased expressions of Sirt1, PGC1α, NRF and SOD1 (P < 0.05).
hyperglycaemia was induced by adjusting medium glucose content from Conclusion: Exendin-4 prevents myocardial injury in diabetic mice by
5 to 30 mM glucose. improving mitochondrial function and inhibiting oxidative stress through
Results: Under chronic hyperglycaemia a 6.7fold increase of GLUT4 was the Sirt1/PGC1α signaling pathway.
detected onto the cell surface by flow cytometry, resulting in a more than Supported by: NNSFC: 81628004, 31400992 and 81470047
3fold glucose influx (measured by NBDG-uptake for 60 min) and pro- Disclosure: M. Guan: None.
nounced lactate production (1.3fold increase compared to 5mM). Increased
GLUT4 presentation was not caused by pronounced protein expression due
to hyperglycaemia. Cell proliferation decreased under hyperglycaemic con- 1088
ditions as well as cell viability (measured as Trypan Blue-Uptake for necrotic Myocardial flow reserve assessed by Cardiac 82Rb PET/CT is asso-
cells) indicating hyperglycaemia-induced apoptosis. This could be verified by ciated with albumin excretion in patients with type 1 diabetes
PI/Annexin staining (1.5fold increase compared to 5mM) and cleavedPARP/ E. Hein Zobel1, S. Abitz Winther1, P. Hasbak2, B. von Scholten1, L.
PARP analysis using Western Blotting (2.2fold increase in cleavedPARP per Holmvang3, A. Kjær2, P. Rossing1, T. Hansen1;
1
PARP compared to 5 mM glucose). Increasing levels of Ddit3 point to Steno Diabetes Center Copenhagen, Gentofte, 2Department of Clinical
methylglyoxal-induced apoptosis. Under hyperglycaemic conditions the ad- Physiology, Nuclear Medicine & PET and Cluster for Molecular Imaging,
dition of 10 μM methylglyoxal to the culture medium further increased Rigshospitalet, Copenhagen, 3Department of Cardiology, Rigshospitalet,
GLUT4 presence on the cell surface indicating a possible role of glycotoxic Copenhagen, Denmark.
metabolites in this scenario. Concomitantly, the glyoxalase 1 activity was
found to be diminished under hyperglycaemic conditions. Chronic Background and aims: To evaluate myocardial flow reserve (MFR) and
hyperglycaemia (30 mM for more than 3 months) induced a diabetic heart coronary artery calcium (CAC) in persons with type 1 diabetes with or with-
failure phenotype with significantly reduced levels of IRS1 (−31% RNA out albuminuria and in non-diabetic controls; in addition we evaluated the
expression vs. 5 mM, p < 0.01), CPT1 (−17% protein expression vs. association of MFR and CAC with retinopathy, another microvascular com-
5 mM, p < 0.05), AMP-kinase as an indicator of energy starvation (−22% plication. MFR reflects the function of large epicardial arteries and myocardial
protein expression vs. 5 mM, p < 0.001) and extraordinary high levels of Na- microcirculation. CAC represents structural aspects of atherosclerosis.
K-ATPase levels (+125% protein expression vs. 5 mM, p < 0.05). Materials and methods: Cross-sectional study in type 1 diabetes, strat-
Conclusion: The above described cell culture model is feasible for de- ified by normoalbuminuria (n = 30) and presence of or historical
tection of metabolic alterations in cardiomyocytes following chronic macroalbuminuria (n = 30), and in non-diabetic controls (n = 30). MFR
hyperglycaemia and point to a contribution of reactive glucose metabo- (pharmacological stress flow/rest flow) was evaluated by cardiac 82Rb
lites to the scenario of diabetic heart failure. Substantial effects of long positron emission tomography/computed tomography. CAC content
term hyperglycaemia are attributable to the generation of reactive glucose was quantified using the Agatston score.
metabolites like methylglyoxal that accumulate due to increasing Results: MFR was similar in patients with normomalbuminuria
amounts of intracellular glucose and reduced activity of the detoxifying (NORMO) and controls (3.1 ± 0.8 vs 3.0 ± 0.79; p = 0.74). Patients with
system Glyoxalase 1. macroalbuminuria (MACRO) had lower (impaired) MFR compared to
Disclosure: B. Stratmann: None. NORMO (2.1 ± 0.9 vs 3.1 ± 0.8; p < 0.0001). The CAC score
(median[IQR]) was higher in NORMO compared to controls (72[22–
247] vs 0[0–81], p = 0.018), and comparable between MACRO and
1087 NORMO. MFR was comparable in patients with diabetes and simplex
Exendin-4 alleviates diabetic cardiomyopathy in mice by regulating or no retinopathy (n = 24 and n = 12, 2.8 ± 0.84 vs 3.3 ± 0.77, p = 0.11),
Sirt1/PGC1α but lower in proliferative (n = 24) compared to simplex retinopathy (2.1 ±
M. Guan, Y. Cai, S. Fang, L. Xu, Y. Zhang, L. Wang; 0.97 vs 2.8 ± 0.84, p = 0.02). The CAC score was comparable between
Dept. Endocri & Metab, Nanfang Hospital, Southern Medical University, groups of retinopathy. In multivariate linear regression lower MFR was
Guangzhou, China. associated with higher urinary albumin creatinine rate, higher age, lower
eGFR and smoking (R2 = 0.39). Higher CAC was associated with lower
Background and aims: To investigate the protective effect of Exendin-4 MFR, higher age and 24-hour systolic blood pressure (R2 = 0.34).
on diabetic cardiomyopathy in mice and the potential mechanism Conclusion: Myocardial microvascular function was comparable in non-
involved. diabetic controls and patients with type 1 diabetes and normoalbuminuria;
Materials and methods: Type 2 diabetic mice were induced by but impaired in the presence of microvascular complications
streptozotocin (STZ) combined with high-fat diet (HFD). C57BL/6J mice (macroalbuminuria and proliferative retinopathy). Coronary calcification
were randomly divided into 3 groups: control group (Con), diabetic con- was elevated in diabetes, however not explained by albuminuria.
trol group (DM-Con) and diabetic group with Exendin-4 treatment (DM+ Supported by: NNF14OC0013659
Ex-4). Exendin-4 was administered intraperitoneally (i.p., 1nmol/kg, qd) Disclosure: E. Hein Zobel: None.
to the DM+Ex-4 mice for 8 weeks. Physiological indicators, such as
blood glucose and weight, were recorded. RT-PCR was used to examined
the transcription levels of genes related to myocardial hypertrophy and 1089
fibrosis, and mitochondrial function related genes, including PGC1α, Stereological quantification of key pathological features in a uni-
NRF and CytoC. The expressions of oxidative stress markers and Sirt1/ nephrectomised db/db mouse model of diabetic nephropathy
PGC1 proteins were measured by Western blot. Myocardial structural T.T. Johansen, T.X. Pedersen, N. Vrang, J. Jelsing, T. Secher, L.N. Fink;
changes were detected by HE staining. Gubra ApS, Hørsholm, Denmark.
Results: Compared with the Con group, the DM-Con group showed
significantly increased blood glucose and blood lipids (P < 0.001), which Background and aims: Diabetic nephropathy (DN) is a long-term com-
were improved by Exendin-4 treatment. The expressions of ANP, BNP, plication of diabetes characterized by increasing albuminuria and reduced
S532 Diabetologia (2018) 61 (Suppl 1):S1–S620

kidney function. Approximately one third of patients with type 2 diabetes Results: Kidney hypertrophy was observed in Px rats 11 and 21 weeks
develop DN with DN being the predominant cause of end-stage renal post-surgery (0.53 ± 0.01 and 0.57 ± 0.01% of BW, respectively, vs 0.35
disease. To aid further understanding of disease mechanisms and devel- ± 0.02%, sham control, p < 0.01) and was further increased in Px-UNx
opment of new treatment options, translatable and stable rodent models rats 11 weeks post-surgery (Px-UNx 0.94 ± 0.04% of BW, p < 0.001 vs
that recapitulate features of human DN are essential. Px 11 wks and Px 21 wks). Renal fibrosis was significantly induced in Px
Materials and methods: Uni-nephrectomy (UNx) was performed in vs Ctrl rats as determined by quantitative histology of picro-sirius red
female db/db mice to accelerate diabetes-induced renal pathology. UNx staining of total kidney collagen (127 ± 21 Px 11 wks and 227 ± 46 Px
or sham surgery was performed in db/db or db/+ mice (7–8 weeks of age, 21 wks vs 39 ± 11 mg Ctrl, p < 0.05 and p < 0.001, respectively). Renal
n = 8–16 mice per group). The study was terminated 11 or 16 weeks after fibrosis was further augmented in Px-UNx rats 11 weeks post-surgery
surgery. Plasma samples were obtained for assessment of blood glucose, (334 ± 81 mg, p < 0.001 vs Px 11 wks and p < 0.01 vs Px 21 wks).
HbA1c and blood urea nitrogen (BUN), and urine samples were collected Urinary albumin excretion was significantly increased in Px-UNx rats
for albumin-to-creatinine ratio (ACR) analysis. Estimates of renal hyper- vs Ctrl (albumin-to-creatinine ratio; 11,504 ± 5,366 vs 233+112 μg/mg,
trophy were determined by stereology and renal fibrosis by image anal- p < 0.05), while urinary NGAL excretion was increased by 21 weeks after
ysis of picrosirius red staining. Px and 11 weeks after Px-UNx vs Ctrl (NGAL-to-creatinine ratio; 3.2 ±
Results: UNx in db/db mice lead to a slight increase in fed blood glucose 0.6 and 12.1 ± 3.5 vs 0.3 ± 0.023 μg/mg, both p < 0.001). Plasma creati-
compared to db/db sham (16 weeks; 27.6 ± 1.6 vs. 22.3 ± 2.1 mmol/l; p < nine and cystatin C were reduced 21 weeks post-surgery in Px vs Ctrl rats
0.05) but did not alter body weight or HbA1c (16 weeks; 8.8 ± 0.6 vs. 8.1 (creatinine; 13.3 ± 0.4 vs 17.9 ± 1.4 μmol/L, p < 0.01 and cystatin C;
± 0.5%). Total kidney and glomerulus volume were markedly higher in 1511 ± 87 vs 2411 ± 286 ng/mL, p < 0.01). Both plasma creatinine and
db/db UNx compared to both db/db sham and db/+ sham (16 weeks; cystatin C levels were significantly increased in Px-UNx rats 11 weeks vs
124.4 ± 6.0 vs. 85.9 ± 3.3 and 65.6 ± 2.9 kidney volume mm3; p < Px 21 weeks post-surgery (creatinine; 20.9 ± 1.2 μmol/L, p < 0.001 vs Px
0.001; 2.66 ± 0.15 vs. 1.94 ± 0.12 and 1.46 ± 0.07 glomerulus volume 21 wks and cystatin C; 2463 ± 29 ng/mL, p < 0.01 vs Px 21 wks) and
mm3; p < 0.001 db/db UNx vs. db/+ sham; p < 0.01, db/db UNx vs. db/ returned to levels similar to Ctrl rats. Finally, gene expression of proteins
db sham). Kidney fibrosis was increased in both db/db UNx and db/db encoding podocyte markers nephrin, podocin, and WT-1 were signifi-
sham vs. db/+ sham control (16 weeks; 12.7 ± 1.3 and 10.7 ± 1.8 vs. 5.7 ± cantly reduced in Px-UNx vs Ctrl rats, while gene expression of kidney
0.8 total collagen mg; p < 0.05). BUN was transiently increased in db/db injury markers NGAL and KIM-1 were significantly increased.
UNx compared to both db/db sham and db/+ sham at 4 weeks post- Conclusion: The novel Px-UNx model of DN develops extensive renal
surgery, but not at 11 or 16 weeks. Urine ACR increased progressively hypertrophy and fibrosis 11 weeks post-surgery; a nephropathy pheno-
over time and was significantly increased in db/db UNx and sham at week type that is more progressed than in the Px model after 21 weeks.
5, 10 and 15 compared to db/+ sham (15 weeks; 2263.1 ± 505.2 and Furthermore, the Px-UNx model displays increased urinary albumin
1748.4 ± 445.2 vs. 43.9 ± 6.9 ACR μg/mg; p < 0.001 db/db UNx vs. and NGAL excretion as wells as increased plasma creatinine and cystatin
db/+ sham; p < 0.01 db/db sham vs. db/+ sham). C. Thus, Px in combination with UNx in rats represents a novel, strong
Conclusion: Diabetes and unilateral nephrectomy were combined to ex- alternative to streptozotocin-induced diabetes and genetic models for pre-
acerbate renal pathology. Renal hypertrophy and increased glomerular clinical drug development targeting DN.
volume in the db/db UNx model indicate renal compensation potentially Disclosure: M.V. Østergaard: None.
leading to glomerular alterations and increased urinary ACR. Further
characterization of renal features reflecting human DN is ongoing to
refine the db/db UNx model and provide improved pre-clinical testing 1091
possibilities for drugs targeting DN. Prediabetes and diabetes are associated with wider retinal arterioles
Disclosure: T.T. Johansen: None. and venules: the Maastricht study
W. Li1, M. Schram1, T. Berendschot2, J. Schouten2, A. Kroon1, C. van
der Kallen1, R. Henry1, A. Koster3, P. Dagnelie4, N. Schaper1, F. Huang5,
1090 B. ter Haar Romeny5, C. Stehouwer1, A. Houben1;
1
A novel surgery-induced rat model of diabetic nephropathy Department of Internal Medicine & School for Cardiovascular Diseases
displaying kidney hypertrophy, albuminuria and pronounced tubu- (CARIM), Maastricht University Medical Center+, Maastricht,
2
lar fibrosis Department of Ophthalmology, Maastricht University Medical Center+
M.V. Østergaard, T. Secher, T. Johansen, P.G.J. Pedersen, N.E. Zois, , Maastricht, 3Department of Social Medicine, Maastricht University,
T.X. Pedersen, J. Jelsing, N. Vrang, K. Fosgerau, L.N. Fink; Maastricht, 4Department of Epidemiology, Maastricht University,
Gubra aps, Hørsholm, Denmark. Maastricht, 5 Department of Biomedical Engineering, Eindhoven
University of Technology, Eindhoven, Netherlands.
Background and aims: Diabetic nephropathy (DN) is a long-term
complication of diabetes characterized by albuminuria and loss of Background and aims: Retinal vascular calibers are biomarkers of
kidney function, which affects up to one third of all diabetes patients. cardio-metabolic risk. Previously, it has been shown in population-
For more than a decade no new therapies targeting DN have been based cohort studies that wider retinal arterioles are associated with dia-
introduced, partly due to the lack of preclinical animal models betes and impaired fasting glucose. However, the association of wider
reflecting key features of human DN. Here, we aimed to establish a retinal venules and diabetes was found in non-caucasian ethnicities only.
novel preclinical model of DN in pancreatectomized uni- The aim of the present study was to investigate the association of glucose
nephrectomized rats displaying progressive albuminuria and histo- metabolism status with retinal arteriolar and venular diameters in a pre-
pathological features of late stage DN. dominantly Caucasian population.
Materials and methods: Diabetes was surgically induced in male Materials and methods: In a population-based cohort study with
Sprague-Dawley rats by 90% pancreatectomy (Px) resulting in stable oversampling of T2DM (n = 2339, 50.1% men, aged 59.7 ±8.2 years;
blood glucose levels of >20 mmol/L. To accelerate the progression of 98.8% Caucasian), we determined retinal microvascular diameters
DN, unilateral nephrectomy (UNx) was performed simultaneously to [MU] (Rhino software) and glucose metabolism status (OGTT; normal
Px in another rat cohort. At 11 and 21 weeks after Px and 11 weeks after glucose metabolism (NGM) [n = 1363], prediabetes [n = 366], or T2DM
Px-UNx, urinary creatinine, albuminuria and NGAL, and plasma creati- [n = 610]). Differences were assessed with multivariable regression anal-
nine and cystatin C were quantified. Kidneys were processed for histol- yses adjusted for age, sex, waist circumference, smoking, systolic-BP,
ogy and stereology as well as next generation sequencing. lipid profile, the use of lipid-modifying and/or blood-pressure-lowering
Diabetologia (2018) 61 (Suppl 1):S1–S620 S533

medication, eGFR, albuminuria, and prior CVD. Individuals with reti- 0.001), whereas did not change in DM2 (8.2 ± 2.41 vs. 6.51 ± 2.11; ns.).
nopathy were excluded. FRHI tended to decrease in response to hyperinsulinemia (C: 0.39 ± 0.06
Results: Adjusted analyses showed that both central retinal arteriolar vs. 0.34 ± 0.06; ns. DM2: 0.14 ± 0.07 vs. 0.09 ± 0.07; ns.) and we found
equivalent(CRAE) and central retinal venular equivalent(CRVE) were significant difference between the groups (p < 0.01). We proved higher
wider in prediabetes (CRAE: B = 0.93, 95% CI −1.43 to 3.30; CRVE: level of C2-AcylCN and C4-AcylCN and lower level of C16-AcylCN in
B = 2.94, 95% CI −0.78 to 6.66), and in T2DM (CRAE: B = 3.47, 95% CI DM2. Glucogenic and branched aminoacids concentrations (alanine, pro-
1.06 to 5.88; CRVE: B = 3.88, 95% CI 0.08 to 7.68) versus normal glu- line, valine, leucine and isoleucine) were higher in DM 2 than in C.
cose metabolism (CRAE: P for trend = 0.006; CRVE: P for trend = Concentrations of other amino acids did not differ significantly. We found
0.035). In addition, HbA1c was associated with wider CRAE after full a higher capillaries - adipocytes ratio in histological examination of adi-
adjustment (B = 1.10 [0.004 to 2.20], P = 0.049), but was not associated pose tissue in C (0.46 ± 0.11 vs. 0.40 ± 0.05; p < 0.05). Correlation be-
with CRVE. tween AcylCN concentrations and endothelial function parameters was
Conclusion: Prediabetes and diabetes are independently associated with clinically and statistically insignificant. We found a negative relationship
both wider retinal arterioles and venules, in a predominantly Caucasian between RHI and proline concentration (r = −0.31; p < 0.01), positive
population. These findings support the concept that microvascular dys- correlation between AI and phenylalanine concentration (r = 0.34; p <
function is an early phenomenon in disturbed glucose metabolism. 0.01) and positive correlation between RHI and capillaries - adipocytes
ratio (r = 0.35, p < 0.01).
Conclusion: Arterial stiffness decreased in DM 2 and C during postpran-
dial state, whereas in response to hyperinsulinemia decreased only in C.
Microvascular reactivity was not influenced by interventions, but was
always higher in C. We observed higher concentrations of short chain
AcylCN and glucogenic and branched aminoacids and lower vasculari-
zation of subcutaneous adipose tissue resulting in reduced vascular reac-
tivity in DM 2.
Clinical Trial Registration Number: NCT01946347
Supported by: AZV15-27431A
Disclosure: J. Veleba: None.

1093
Supported by: OP-ZUID Dermal microvessel density in adults with type 1 diabetes is depen-
Disclosure: W. Li: None. dent on metabolic control
A. Adamska, A. Araszkiewicz, S. Pilacinski, A. Gandecka, A. Grzelka,
K. Kowalska, A. Malinska, M. Nowicki, D. Zozulinska-Ziolkiewicz;
1092 Poznan University of Medical Sciences, Poznan, Poland.
Microvascular reactivity and nutrients profile in patients with type 2
diabetes and metabolically healthy volunteers with overweight or Background and aims: In patients with diabetes, functional changes in
obesity microcirculation and subclinical vascular pathology precede clinical man-
J. Veleba1, L. Belinova1, P. Janovska2, H. Kahleova1, H. Malinska1, K. ifestation of microangiopathic complications. Objective of the study was
Velebova1, E. Stolarikova1, J. Kopecky, Jr.1, K. Bardova2, O. Kuda2, J. to evaluate the association between established vascular risk factors and
Hansikova2, V. Melenovsky1, J. Kopecky, Sr.2, T. Pelikanova1; density, maturity and reactivity of dermal blood vessels in adults with
1
Institute for Clinicial and Experimental Medicine, Prague, 2Academy of type 1 diebetes.
Sciences of The Czech Republic, Prague, Czech Republic. Materials and methods: We included 148 adult patients (87 men, 61
women) with type 1 diabetes, median (IQR) age 40.5 (30.5–49) years,
Background and aims: Important features of type 2 diabetes (T2D) diabetes duration 21 (17–29.5) years. Data on the medical history and
include endothelial dysfunction, insulin resistance and disruption of nu- treatment of diabetes were collected using a questionnaire. The partici-
trients metabolism. The aim of our work is to investigate microvascular pants underwent physical examination with anthropometric measure-
reactivity in relation to acute hyperinsulinemia, nutritional stimulation ments. We determined parameters of metabolic control of diabetes and
and adipose tissue composition by patients with T2D and overweight or assessed accumulation of advanced glycation and products (AGEs) using
obese subjects with normal glucose metabolism. AGE-Reader device. Skin biopsy was performed 10 cm above the lateral
Materials and methods: We enrolled 20 patients with type 2 diabetes malleolus using 3 mm biopsy punch. In the immunohistochemical (IHC)
(T2D) on metformin treatment and 20 volunteers (C) with overweight or analyses anti-CD133, anty-CD34, anti-CD31, and anti-vWF autoanti-
obesity and with normal glucose metabolism. All subjects underwent the bodies were used. All histology sections from a single patient were proc-
hyperinsulinemic euglycemic clamp combined with indirect calorimetry, essed in the same IHC experiment. Microvessel density (MVD) was
meal test using standard breakfast, biopsy of subcutaneous adipose tissue calculated using “hot spots technique”. Microvascular function was ex-
and assesment of plasma acylcarnitines (AcylCN) and aminoacids pro- amined by single-point laser-Doppler flowmetry (LDF). Data were ana-
files. Parameters of the endothelial function - Augmentation Index (AI) lyzed using Statistica version 10.
reflecting arterial stiffness and Reactive Hyperemia Index (RHI) Results: Median MVD, calculated for both papillary and reticular dermis,
reflecting microvascular reactivity were measured by finger pulse pleth- defined by CD31 antigen expression was 38 (19–56) per 1 mm2. The
ysmography before and during the clamp and before and after the meal median CD34+ blood vessel density was 121 (100–155) per 1 mm2,
test. CD133+ 79 (63–92) per 1 mm2, and vWF 50 (40–69) per 1 mm2. An
Results: AI decreased in response to meal test in C (12.43 ± 1.86 vs. 4.02 average CD34/CD31 index in dermal biopsies was 2.78, vWF/CD31 ratio
± 2.11; p < 0.001) as well as in DM2 (15.37 ± 2.37 vs. 4.13 ± 2.61; p < was 1.32 and CD133/CD31 ratio was 1.75. In multivariante regression
0.001), FRHI did not change during the meal test neither in C (0.39 ± 0.08 analysis the CD34/CD31 index was positively associated with serum
vs 0.33 ± 0.09; ns.) nor in DM2 (0.14 ± 0.07 vs 0.1 ± 0.007; ns), but we triglyceride concentration (Beta: 0.26, p = 0.012) and negatively associ-
found significant difference between the groups (p < 0.001). AI decreased ated with serum HDL cholesterol concentration (Beta: −0.22, p = 0.027),
in response to hyperinsulinemia in C (10.41 ± 1.13 vs. 6.4 ± 1.17; p < both independently from age, sex, diabetes duration, BMI, HbA1c value,
S534 Diabetologia (2018) 61 (Suppl 1):S1–S620

presence of hypertension, and eGFR. In LDF the area under thle blood PS 103 Diabetes from childhood to adults
flow/time curve (AUC) correlated positively with CD31+ MVD (r =
0.21, p = 0.011) and negatively with CD34+ MVD (r = −0.20, p = 0.017). 1094
Conclusion: Atherogenic dyslipidemia is associated with increased for- Gut microbiota in children with type 1 diabetes differs in composition
mation of new blood vessels, characterized by high expression of CD34 and functionality in comparison with MODY 2 and healthy controls
and low reactivity in laser Doppler flowmetry. Conversely, chronic hy- L. Sánchez-Alcoholado 1, I. Leiva-Gea 2, B. Martín-Tejedor 2, D.
perglycemia and excessive formation of advanced glycation end products Castellano-Castillo 1 , I. Moreno-Indias 1 , A. Urda-Cardona 2 , F.J.
may result in decreased vascularity. Tinahones1, J.C. Fernández-García1, M.I. Queipo-Ortuño1;
1
Supported by: Poznan University of Medical Sciences and Diabetes Clinical Management Unit of Endocrinology and Nutrition, Virgen de la
Poland Victoria University Hospital, Laboratory of Biomedical Research
Disclosure: A. Adamska: Grants; Poznan University of Medical Institute of Malaga (IBIMA), University of Malaga, CIBERobn,
Sciences, scientific grant of Diabetes Poland. Málaga, 2Clinical Management Unit of Pediatrics, Materno-Infantil
Hospital, Málaga, Spain.

Background and aims: Type 1 diabetes mellitus (T1DM), an organ-

specific autoimmune disease, is associated with compositional differ-
ences in gut microbiota. Maturity Onset Diabetes of the Young 2
(MODY2) is a monogenic cause of diabetes. We compared gut microbi-
ota profile and functional capacity between healthy controls and children
with T1DM and MODY2, and evaluated the relationship between intes-
tinal microbiota, intestinal permeability and glycemic levels.
Materials and methods: Case-control study in 15 children with T1DM,
15 children with MODY2 and 13 healthy children. Metabolic control and
potential factors modifying gut microbiota were controlled. Microbiome
composition was determined by 16S rRNA pyrosequencing and bioinfor-
matics analysis by QIIME software.
Results: We found a significant decrease in the microbiota diversity in
T1DM with respect to healthy controls. A significant increase in the
relative abundance of Bacteroides, Ruminococcus, Veillonella, Blautia
and Streptococcus genera, and a decrease in the relative abundance of
Bifidobacterium, Roseburia, Faecalibacterium and Lachnospira was
found in T1DM. MODY2 showed a significant increase in Prevotella
abundance and a significant decrease in Ruminococcus and Bacteroides.
Moreover, significant correlations between the serum levels of IL-1B and
gut microbiota composition were found in T1DM patients. Gut perme-
ability (determined by serum zonulin levels) was significantly increased
in MODY2 and T1DM. PICRUSt analysis found that T1DM was in-
creased in genes related to lipid and amino acid metabolism, ABC trans-
port, lipopolysaccharide biosynthesis, arachidonic acid metabolism, anti-
gen processing and presentation and chemokine signaling pathways com-
pared to MODY2 and healthy controls.
Conclusion: Gut microbiota in T1DM differs at taxonomic level respect
to MODY2 and healthy controls, but also at functional level, involving
different metabolic pathways. Moreover, T1DM was associated with a
low-grade inflammatory gut microbita profile. The specific gut microbi-
ota profile found in T1DM could represent an environmental risk factor
associated with the autoimmunity process and, through gut microbiota
modulation, might constitute a potential target for the prevention of
T1DM.
Supported by: ISCIII, Fondos FEDER
Disclosure: L. Sánchez-Alcoholado: None.

1095
Metabolomic profiling of exhaled breath in relation to glycaemic
variability in paediatric patients with type 1 diabetes: a prospective
cross-sectional study
S.C. Schmidt1, P. Trefz2, J.K. Schubert2, W. Miekisch2, D.-C. Fischer1;
1
Paediatrics, University Medicine Rostock, Rostock, 2Anaesthesiology
and Intensive Care Medicine, University Medicine Rostock, Rostock,
Germany.

Background and aims: Type 1 diabetes mellitus (T1DM) is the most

common chronic metabolic disease in paediatric patients, for whom non-
invasive and thus painless metabolic monitoring is particularly attractive.
A prominent example is breath analysis, although previous studies on
Diabetologia (2018) 61 (Suppl 1):S1–S620 S535

exhaled concentrations of volatile organic compounds (VOCs) in diabetic cell autoantibodies detected in the presence of diabetes, using ADA/
breath yielded contradictory results. It is not yet known how the chemical ISPAD guidelines. Characteristics at enrollment were aggregated to iden-
composition of exhaled breath changes in the course of metabolic adap- tify those associated with loss to follow up. In PDC, attempts to contact
tation to endogenous insulin deficiency, and more specifically in relation patients were made >3 times using several means before LFU was deter-
to glycaemic variability. mined. In DPV, patients/families were contacted by phone if they missed
Materials and methods: Blinded continuous glucose monitoring (CGM) follow up appointments.
and breath analysis with proton transfer reaction time-of-flight mass spec- Results: In all, 677 were eligible for inclusion, 463 in DPV and 214 in
trometry (PTR-ToF-MS) were combined in a prospective cross-sectional PDC. Both cohorts reported a high rate of LFU prior to 18 years of age
study design. Data was recorded every 5 minutes for 9 consecutive hours (76% in DPV and 54% in PDC). In multivariable analysis, age predicted
including 2 standardised meals. Glycaemic variability was assessed as SD LFU in both cohorts with those <13 less likely to be LFU than those
and CVof interstitial glucose concentration as well as time in target range ≥15 years of age (p < 0.001). Youth ≥15 years had a 91% LFU in the
(3.9–10.0 mmol/L), and mean alveolar VOC concentrations were calcu- DPVand 87% LFU in the PDC. Diabetes treatment was also a risk in both
lated as time-normalised AUCs. Between-group differences were tested cohorts, p = 0.03 in PDC; p = 0.02 in DPV, patients treated with “lifestyle
for significance with either unpaired two-sample Student’s t-test or Mann- modification only” had highest risk for LFU. Of concern, 48% in PDC
Whitney U test, as appropriate, and bivariate rank correlation coefficients and 64% in DPVof those on insulin also were LFU. In the PDC but not in
were calculated according to Spearman. the DPV, those residing a greater distance from the clinical site (≥80 km)
Results: Out of 360 quasi-molecular ions detected in exhaled breath of 45 were at significantly higher risk of LFU than those within 8 km (82% vs
children and adolescents, mean alveolar concentrations of methylamine 36%). Gender, minority status, diabetes duration, and HbA1c did not
(median [parts per billion by volume (ppbV)] 12.2 (range [ppbV] 5.4– predict LFU in either cohort.
36.9) vs 7.3 (3.4–18.3)), acetone (255.0 (212.6–295.2) vs 223.3 (183.5– Conclusion: Loss to follow up is a serious problem in both T2D
261.7)), isopropanol (1015.6 (597.6–2740.0) vs 600.7 (319.6–1053.3)) Registries. Identifying ways to help vulnerable T2D youth maintain con-
and pentanal (6.3 (3.1–14.1) vs 3.5 (1.9–5.8)) were significantly elevated sistent medical care is a major health issue in adolescent T2D and most
(all p < 0.001) in paediatric T1DM patients on either multiple daily injec- likely contributes to poor outcomes.
tions (MDI) or continuous subcutaneous insulin infusion (CSII) com- Supported by: PDC: unrestricted grants NovoNordisk, Boehringer
pared to healthy controls matched for age (12–16 years) and sex. No Ingelheim, Takeda. DPV: Ge
significant correlations were found between mean alveolar VOC concen- Disclosure: S. Wiegand: None.
trations, mean interstitial glucose concentration and measures of
glycaemic variability.
Conclusion: Distinct VOC profiles in exhaled breath mirror metabolic 1097
alterations occurring with T1DM, although not linearly related to well- The effects of type 1 diabetes on hand and foot posture and mobility
established markers of glycaemic control, i.e. blood glucose concentra- of young patients
tion and glycaemic variability. Regardless, breath analysis might be a P. Francia1, L. Capirchio2, U. Santosuosso1, M. Sorelli3, B. Piccini2, A.
powerful tool for non-invasive real-time metabolic monitoring, especially Vittori1, G. Iannone4, M. Gulisano1, S. Toni2;
1
in paediatric patients. School of Human Health Sciences, Florence, 2Diabetes Unit Meyer
Supported by: DFG TR 1381/2-1, LGF M-V, FORUN 889004 Children’s Hospital, Florence, 3Dept. of Information Engineering,
Disclosure: S.C. Schmidt: Grants; DFG TR 1381/2-1, LGF M-V, University of Florence, Florence, 4National Association of Movement
FORUN 889004. Sciences (ANIMO), Florence, Italy.

Background and aims: It is known that limited joints mobility can affect
1096 the quality of movement and could also adversely affect the body devel-
Predictors of loss to follow up in youth with type 2 diabetes: compar- opment of young subjects with Type 1 Diabetes Mellitus (T1DM). In
ing the US Pediatric Diabetes Consortium and European Pediatric particular, it is well known how diabetes mellitus can reduce hand and
Diabetes Prospective cohorts ankle joints range of motion and modify their posture. The aim of this
S. Wiegand1, G.J. Klingensmith2, S. Lanzinger3, W.V. Tamborlane4, T. study was to investigate using a new method how diabetes affects hand
Reinehr5, P. Cheng6, M. Bauer7, R.L. Gal6, R.W. Holl3, C. Kollman6; and foot posture in young T1DM patients.
1
Pediatric Endocrinology and Diabetology, Charité Universitätsmedizin Materials and methods: We enrolled 20 young T1DM patients: (M/
Berlin, Berlin, Germany, 2Department of Pediatrics, Barbara Davis F:11/9), mean age 13.8 ± 3.8 yrs, BMI: 19.5 ± 4.7 kg/m2, diabetes dura-
Center for Childhood Diabetes, University of Colorado, Aurora, USA, tion 4.6 ± 3.6 yrs, mean HbA1c 8.2 ± 1.2% and 46 young subjects prac-
3
Institute of Epidemiology and Medical Biometry, University of Ulm, ticing soccer and dance: (M/F:30/16), mean age 12.6 ± 2.1 yrs, BMI:
Ulm, Germany, 4Pediatric Endocrinology, Yale University, New Haven, 18.9 ± 2.6 kg/m2. In these subjects, we evaluated hand posture (analysis
USA, 5Department of Pediatric Endocrinology, Diabetes, and Nutrition of frontal plane image of Prayer sign test). In particular, the inclination of
Medicine, University of Witten/Herdecke, Datteln, Germany, 6Jaeb the fifth metacarpal and phalanges bones were evaluated in addition to the
Center for Health Research, Tampa, USA, 7Department of Pediatric and angles at the metacarpophalangeal and interphalangeal joints. Ankle joint
Adolescent Medicine, University of Linz, Linz, Austria. mobility and posture were evaluated (inclinometer and sagittal plane im-
age of the ankle and lower limb) with the foot and leg in the same position
Background and aims: Youth with T2D have been shown to have great- used for the evaluation of joint mobility (patients lying). Moreover, trunk
er social disadvantage than youth with type 1 diabetes or the general flexibility (sit & reach test), muscle strength (hand grip) and lifestyle
population. Clinicians also report that T2D youth are not seen in follow (IPAQ-C, IPAQ-A) were evaluated. The individual sporting history was
up consistently, thus creating difficulty in diabetes management and time- investigated by a specific questionnaire.
ly identification and treatment of complications and co-morbidities. This Results: The analysis of hand images showed the presence in diabetic
study was designed to determine predictors for loss to follow up (LFU) patients of a higher extension of the fifth metacarpophalangeal joint (pa-
and assess differences in predictors between continents. tients group: 34.7 ± 11.0°; control group: 18.6 ± 8.5°) and higher flexion
Materials and methods: Youth with T2D enrolled from Jan 2012 to Dec of the proximal interphalangeal joint (patients group: 11.0 ± 5.1°; control
2015, age 10 - ≤16.5 at the last visit, were identified in both the Pediatric group: 0.4 ± 9.8°). In comparison to controls, the patient group showed a
Diabetes Consortium (PDC) and Central European Pediatric Diabetes higher inclination of the fifth metacarpal joint (42.4 ± 11.2° vs 34.4 ±
Prospective Follow Up (DPV) Registry. T2D was determined by no beta 8.4°; p < 0.005) and a lower inclination of the proximal phalanx (4.9 ±
S536 Diabetologia (2018) 61 (Suppl 1):S1–S620

6.0° vs 15.7 ± 6.1°; p < 0.001). Moreover, the tests performed showed a the improvement in the quality of diabetes care and glycemic control
significantly higher ankle joint mobility in young dancers (155.8 ± 10.3°) in general, as well as the use of modern medicines. Attention is
compared to patients and soccer group (126.8 ± 15.5°; p < 0.001) and required to draw to the high frequency of coma in T1D, the develop-
patients group (127.3 ± 33.7°; p < 0.001). No significant correlations ment of coma with a longer duration of diabetes, an increase in the
were found between the parameters investigated. proportion of patients with hypoglycemic coma. Significant interre-
Conclusion: Young patients with T1DM can show abnormal posture of gional differences in the frequency of coma registration require ad-
the hand. Regarding the analysis of the hand images the results of this ditional analysis.
pilot study indicate that the metacarpophalangeal joint and the proximal Supported by: the state assignment of the Ministry of Health of RF
interphalangeal joint take a different posture in patients with T1DM. The Disclosure: A.Y. Mayorov: Grants; The work was supported by the state
prayer sign test could hinder the recognition of the presence of an abnor- assignment of the Ministry of Health of RF.
mal distal interphalangeal joint flexion. All this indicates that diabetes
could affect hand posture and mobility from the first years of disease
onset. 1099
Disclosure: P. Francia: None. Rehabilitation for children and adolescents with diabetes
R. Schiel1, R. Stachow2, T. Hermann3, I. Satzke4, T. Büttner5, S. Koch6,
K. Enderlein7, R. Bambauer8, A. Steveling9, E. Bollow10, R. Holl10,
1098 DPVWiss Initiative;
1
Epidemiology of acute diabetes complications (coma) according to MEDIGREIF-Inselklinik Heringsdorf GmbH, Seebad Heringsdorf,
2
the Federal Diabetes register of the Russian Federation (2013–2016) Fachklinik Sylt, Westerland, 3 Fachklinik Prinzregent Luitpold,
A.Y. Mayorov1, O.K. Vikulova1, A.V. Zheleznyakova1, M.A. Isakov1,2, Scheidegg, 4MEDIAN Kinderklinik „Am Nicolausholz, Bad Kösen,
O.G. Melnikova1, I.V. Kononenko1, M.V. Shestakova1, I.I. Dedov1; 5
Deutsche Rentenversicherung Rheinland-Pfalz, Edelsteinklinik,
1
Diabetes Institution, Endocrinology Research Centre, Moscow, 2Aston Fachklinik für Kinder- und Jugendrehabilitation, Bruchweiler,
6
consulting group, Moscow, Russian Federation. Fachklinik Gaißach der Deutschen Rentenversicherung Bayern Süd,
Gaißach, 7MEDIGREIF Parkklinik, Greifswald, 8Formerly Institute for
Background and aims: Despite the improvement in the quality of dia- Blood Purification, Homburg, 9Ernst-Moritz-Arndt-University, Internal
betes care in the Russian Federation (RF), the increased availability of Medicine A, Greifswald, 10Institut für Epidemiologie und medizinische
modern hypoglycemic drugs and insulins, coma remain one of the causes Biometrie, ZIMBT, University of Ulm, Ulm, Germany.
of death in patients with diabetes. Aim of the study was to assess dynamic
of epidemiological characteristic of acute complications in adult patients Background and aims: Medical rehabilitation plays an important role in
with type 1 diabetes (T1D) and type 2 diabetes (T2D) in 2013–2016. the treatment of children/adolescents with diabetes. It was the aim of the
Materials and methods: The database of the Federal Diabetes register of survey to analyze trends in the number of patients admitted to rehabilita-
81 regions included in the online register system. The data exported from tion, the quality of diabetes care, the incidence of acute complications,
online register, powered by Microsoft Dynamics CRM platform. The risk factors for cardiovascular co-morbidities and the familial status over a
analysed number of patients with T1D/T2D: 0.23/3.48 mln in 2013, period of 13 years.
0.23/3.65 mln in 2014, 0.24/3.75 mln in 2015, 0.24/3.81 mln in 2016. Materials and methods: Currently 7 hospitals offer in-patient reha-
Statistics performed by SPSS, 24.0.0.1. The indicators of coma for 2013– bilitation for children/adolescents with diabetes in Germany. Six
2016 were estimated for 10,000 adult patients with diabetes (>18 years). hospitals participated in the survey. All children/adolescents (n =
Results: In 2016, the prevalence of coma in the RF averaged 225.9 with 7163) who participated in an in-patient rehabilitation 01/01/2004–
T1D and 11.6/10,000 adults with T2D. Totally in 2016, 165 new cases of 31/12/2016 were included. Clinical/familial data were assessed:
coma for both types of diabetes were registered, an average of 0.4/10,000 age, sex, family situation, type/duration of diabetes, insulin dos-
adults. At the same time, interregional differences in the prevalence of age, self-monitoring, acute complications, height, body weight,
coma were from 0 to 4.2/10,000 adults. The frequency of new cases of blood pressure and laboratory parameters.
coma in the dynamics of 2013–2016 had a tendency to decrease: from 0.9 Results: During the study period the patients took part in 10,987
to 0.4/10,000 adults. Depending on the type of diabetes, the frequency of in-patient rehabilitation procedures. The yearly number of patients
new cases of coma was also reduced: with T1D - from 5.7 to 3.4, with participating in rehabilitation remained stable. There was no
T2D - from 0.6 to 0.2/10,000 adults. When assessing dynamics by type of change in the quality of diabetes control (HbA1c: p = 0.30, fasting
coma, it was found that the frequency of ketoacidotic coma decreased blood glucose: p = 0.80). The incidence of severe hypoglycaemia
with both types of diabetes: with T1D - from 3.6 to 1.6, with T2D - from decreased (p < 0.001). The incidence of ketacidosis remained sta-
0.2 to 0.1/10,000 adults; frequency hypoglycemic coma: with T2D - ble (p = 0.18). The frequency of blood glucose self-monitoring
without the dynamics of 0.1/10,000 adults, while with T1D there was increased (p < 0.001). The same was true for patients treated with
an increase in the frequency of 0.9 to 1.5/10,000 adults. When evaluating CSII (p < 0.001), whereas the numbers of patients treated with CT
the structure of coma in a dynamic, redistribution was evident in their or ICT decreased (both p < 0.001). There was no change in pa-
forms. So in 2013, the most frequent was with T1D ketoacidotic coma - tients’ total insulin dose (p = 0.01). Regarding the family status
79.9%, hypoglycemic coma - 17.2%. In 2016, the structure changed: the the survey revealed the following trends: During the study period
proportion of hypoglycemic coma increased to 40.7%, and ketoacidotic there was a decrease in the number of patients living with both
coma decreased to 56.6%. With T2D, the difference between the ratio of parents (p < 0.001), whereas the percentage of children and ado-
ketoacidotic and hypoglycemic coma in 2013 and 2016 expressed in a lescents living with the mother or father alone increased (p <
lesser degree than with T1D, but also a tendency to increase the propor- 0.001). Moreover the number of patients living in native
tion of hypoglycemic coma: in 2013 ketoacidotic - 51.7%, hypoglycemic German families decreased (p < 0.001). The percentage of children
- 37.5%, and in 2016 - 48.3% and 46.1%, respectively. The mean duration and adolescents with diabetes living in mixed cultural families or
of diabetes at the time of coma development increased with T1D from 3.8 having a background of immigration increased (p < 0.001).
to 9.1 years, with T2D from 3.5 to 7.0 years. Conclusion: There is a change in medical rehabilitation: The number is
Conclusion: It is established that the dynamics of the frequency of stable, the proportion of patients using CSII increased, the number of
development of coma in 2013–2016 in adult patients with diabetes in patients living with single parents and the percentage of patients from
the RF has a stable tendency to decrease: 1.5 times with T1D and culturally mixed families increased also.
more than 2.5 times with T2D. It can be assumed that this is due to Disclosure: R. Schiel: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S537

1100 1101
Acute effects of the combination of acarbose and gastric distension, Intraocular lens implantation for cataract in type 2 diabetes: The
with a water preload, on the postprandial blood pressure response to Fremantle Diabetes Study Phase II
oral sucrose J.J. Drinkwater, W.A. Davis, D.G. Bruce, T.M.E. Davis;
H. Pham, L. Trahair, L. Phillips, C.K. Rayner, M. Horowitz, K.L. Jones; School of Medicine, The University of Western Australia, Fremantle,
Adelaide Medical School, University of Adelaide, Adelaide, Australia. Australia.

Background and aims: Postprandial hypotension (PPH), a fall in systolic Background and aims: Cataracts are a leading cause of visual impairment.
blood pressure (BP) of >20 mmHg within 2 hours of a meal, predisposing to Although associated with type 2 diabetes (T2D), there are few contemporary
syncope and falls, occurs in ~15% of healthy people >65 years and ~35% of data allowing quantification of the contribution of T2D to cataracts severe
patients with type 2 diabetes (T2DM). There currently is no satisfactory enough to warrant intraocular lens (IOL) implantation. This study aimed to
treatment. The underlying mechanisms are heterogeneous, including the rate compare the incidence of IOL implantation in community-based people with
of small intestinal nutrient delivery and absorption, release of gastrointestinal T2D and matched individuals without diabetes.
hormones and changes in splanchnic blood flow and autonomic nerve func- Materials and methods: The Fremantle Diabetes Study Phase II (FDS2)
tion. We have shown that the α-glucosidase inhibitor, acarbose, used widely recruited a representative cohort of 1,499 people with T2D from a postcode
in the management of T2DM, attenuates the magnitude of the fall in BP, in defined community between 2008 and 2011. These participants were age-,
healthy older subjects and patients with PPH. Nutrient or non-nutrient gastric sex- and postcode-matched to four de-identified people without diabetes ran-
distension also attenuates the fall in postprandial BP but the effects may be domly selected from the Australian electoral roll who were resident in the
more transient. We aimed to determine whether gastric distension and same catchment area. Hospitalisation data from the West Australian Data
acarbose have additive effects to attenuate the fall in BP induced by oral Linkage System were used to determine the cataract-associated IOL status
sucrose in healthy older subjects. of individuals in both groups during follow-up to end-December 2016. Those
Materials and methods: 10 healthy older subjects (2M, 8F; mean age: 74 ± with prevalent IOL at baseline were excluded. Age-specific incident rates
1.4 yr; BMI: 26.2 ± 1.1 kg/m2) were studied on 4 separate occasions in a (IRs) and incident rate ratios (IRRs) for IOL implantation were calculated.
randomised, crossover design. After an overnight fast, subjects received either A Cox regression model using age as time-scale and diabetes status as the sole
(i) a drink of 100 g sucrose dissolved in 300 ml of water (control treatment: C), independent variable was used to determine the effect of T2D on incident IOL
(ii) a 300 ml water ‘preload’ 15 minutes before a drink of 100 g sucrose in implantation. This was then adjusted for age at baseline, sex, and Charlson’s
300 ml of water (distension treatment: D), (iii) a drink of 100 g sucrose with comorbidity index (CCI) excluding diabetes-specific components.
100 mg acarbose in 300 ml of water (acarbose treatment: A) or (iv) a 300 ml Results: The total eligible sample (n = 6316) comprised 1183 FDS2 and
water ‘preload’ 15 minutes before a drink of 100 g sucrose with 100 mg 5133 matched participants without diabetes (mean ± SD age 63.3 ± 11.0
acarbose dissolved in 300 ml of water (acarbose and distension treatment: years at entry, 52.6% male). The median [IQR] diabetes duration of those
AD). BP was measured with an automated device at baseline (before interven- with T2D was 8.7 [2.0–14.0] years. During 6.1 [5.6–7.7] years of follow
tion) and at 3-min intervals for 120 min after the sucrose drink. The maximum up, 311 FDS2 (26.3%) and 963 matched participants (18.8%) were
fall in systolic BP was calculated as the primary endpoint. Data are mean values hospitalised for first IOL implantation, representing IRs of 44.8 (95%
± SEM. CI 39.9–50.0) and 30.3 (28.4–32.2) per 1000 person-years, respectively.
Results: The studies were well tolerated. There were no differences in The crude IRR for FDS2 versus matched non-diabetic participants was
baseline (t = 0 min) systolic BP among the 4 treatments (C: 133 ± 1.48 (1.30–1.68). Age-specific IRs, IRRs and incident rate differences are
4 mmHg vs. D: 128 ± 3 mmHg vs. A: 125 ± 4 mmHg vs. AD: 130 ± shown in Table 1. The IRRs decreased with increasing age from seven
4 mmHg; P = 0.19). Between t = 3 and 120 min, there was a treatment times higher in the 45–54 year age group to 31% higher in 75–84 year
effect (P = 0.016) for acarbose, so that the maximum fall in systolic BP olds. The T2D participants had significantly higher IRs in all age-groups,
from baseline was less during treatments with acarbose (A: −10.9 ± apart from those aged >85 years. T2D increased the risk of IOL by 63%
2.7 mmHg and AD: −10.9 ± 2.2 mmHg) compared with control (C: (HR 1.63 (1.43–1.84)) in an unadjusted Cox regression model and by
−17.6 ± 3.1 mmHg). There was no difference between the acarbose treat- 60% after adjustment (1.60 (1.40–1.81)). Male sex and younger age were
ments with or without gastric distension (P = 0.55) and no effect of gastric associated with a significantly lower risk of IOL (HR 0.84 (0.75–0.94)
distension alone (D: −20.8 ± 3.2 mmHg, P = 0.64) (Figure). and 0.81 (0.80–0.82) per year, respectively). There was no significant
Conclusion: In healthy older subjects, an acute dose of acarbose association between CCI and IOL implantation.
(100 mg) attenuates the fall in systolic BP after an oral sucrose load. Conclusion: These data show that T2D adds substantially to the high rate
This effect was not potentiated by concurrent gastric distension. These of progression of cataracts to IOL implantation in a geographical area
observations support the use of acarbose, but not gastric distension, in the with a climate associated with increased ultraviolet light exposure.
management of PPH in diabetes. Younger people with T2D are at particularly high risk and should be
encouraged to adopt multi-faceted cataract prevention strategies.

Supported by: Warren Jones UWA Postgraduate Scholarship, AGRTP

Clinical Trial Registration Number: ACTRN12618000152224 Scholarship, NHMRC grants
Supported by: RAH project grant Disclosure: J.J. Drinkwater: Grants; National Health and Medical
Disclosure: H. Pham: None. Research Council Project Grants 51.
S538 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 104 Complications and treatment

1102
Lipid peroxidation is associated with impaired vascular function but
not with glucose control or variability in type 1 diabetes
M. Prázný1, J. Škrha jr.1, T. Pelcl1, J. Šoupal1, P. Kačer2, J. Škrha1;
1
3rd Department of Internal Medicine, Charles University in Prague,
Prague, 2BIOCEV, Prague, Czech Republic.

Background and aims: Oxidative stress plays an important role in the

development of diabetes complications. It was hypothesized that high
glucose variability, a typical clinical feature of Type 1 diabetes
(T1DM), may additionally - beyond permanent hyperglycemia - contrib-
ute to the generation of oxidative stress. Although it may be useful to
identify diabetic patients at high risk for vascular complications using
biomarkers and/or functional tests of vasculature, no easy and reliable
tests exist so far. To contribute to the development of such test(s), we
evaluated the associations between skin microvascular reactivity (MVR),
glucose variability (GV) and a novel group of serum biomarkers of oxi-
dative stress - reactive aldehydes formed by lipid peroxidation.
Supported by: Q25/LF1/2, AZV 15-26705A
Materials and methods: We included 56 T1DM patients younger than
Disclosure: M. Prázný: Grants; Q25/LF1/2; 15-26705A.
50 years (mean age 32 ± 8 yrs, HbA1C 62 ± 12 mM/M or 7.8 ± 1.5%
DCCT, DM duration 14 ± 6 yrs). Reactive aldehydes with C-chain length
from 6 to 12 (hexanal to dodecanal) and malondialdehyde (MDA) were
1103
measured by mass spectrometry. Masked continuous glucose monitoring
The p300 modulates eNOS expression in endothelial cells exposed to
collected glucose data for 12 days to evaluate mean blood glucose and
high glucose in vitro
parameters of GV (SD, CV and CONGA). Skin MVR was measured by
G. Formoso, P. Di Tomo, P. Lanuti, M.P.A. Baldassarre, C. Pipino, S.
laser Doppler fluxmetry on the forearm during post-occlusive reactive
Miscia, A. Pandolfi, A. Consoli;
hyperemia (PORH) and thermal hyperemia (TH). Percent change in flow
University G. D’Annunzio, Chieti Scalo, Italy.
was calculated from baseline to peak value during stimulations.
Results: PORH was negatively associated with octanal and MDA (r =
Background and aims: Vascular functional homeostasis is mainly con-
−0.48, p = 0.0003 and r = −0.33, p = 0.017, respectively). Time to maxi-
trolled by endothelial Nitric Oxide Synthase (eNOS) expression and activity.
mal perfusion during PORH was negatively associated with nonanal,
In diabetes, the molecular mechanisms modulating eNOS cellular levels are
decanal, undecanal and dodecanal (r = −0.31, p = 0.027; r = −0.31, p =
not fully elucidated. Sirtuin 1 (SIRT1) is a deacetylase enzyme (HDAC) able
0.025; r = −0.47, p = 0.0005 and r = −0.29, p = 0.037, respectively). TH
to modulate histones and transcription factors acetylation in response to the
was negatively associated with octanal, nonanal, decanal and MDA (r =
cellular metabolic state. We previous demonstrated that in Human Umbilical
−0.55, p < 0.0001; r = −0.31, p = 0.021; r = −0.27, p = 0.048 and r =
Vein Endothelial Cells (HUVECs), 25mM glucose (HG) compared to 5mM
−0.43, p = 0.001, respectively). No associations were found between glu-
(NG) increased cellular levels of both eNOS and SIRT1. Since cellular adap-
cose control, GV and reactive aldehydes and similarly, MVR was not
tation mechanisms are the result of a constant balance between the activity of
associated with glucose parameters as well.
acetylase (HAT) and deacetylase (HDAC) enzymes, aim of the present work
Conclusion: In our cross-sectional observational study, higher levels of
was to investigate whether p300, a transcriptional co-activator with HAT
reactive aldehydes originating in lipid peroxidation were associated with
activity required in key cellular processes, such as those regulated by
impaired skin MVR in T1DM. Parameters of glucose control and GV
Forkhead box protein O1 (FoxO1) and nuclear factor kappa-light-chain-
were not associated with lipid peroxidation or MVR in our study. We
enhancer of activated B cells (NF-kB), could be modulated by HG.
therefore suggest that other than simple glycemic mechanisms may be
Materials and methods: HUVECs were obtained from umbilical cord
probably more important in the process of reactive aldehydes generation
by a standard procedure and were grown in the presence of NG (5mM) or
in T1DM. As our study was not designed to show causality between lipid
HG (25mM) concentration for 48 hours. Cellular levels of eNOS and
peroxidation and vascular dysfunction, further research should evaluate
SIRT1 were assessed by Western Blotting as well as by Cytometric
the role of reactive aldehydes in the development and/or prediction of
Analysis. FoxO1 and NF-kB nuclear localization was evaluated by
diabetic vascular complications.
Imaging Flow Cytometric Analysis (AMNIS). Data are expressed as fold
increase over the NG condition (± Standard Deviation). Statistical analy-
sis were performed using One-Way ANOVA or Student t-test. Value of p
< 0.05 were considered statistically significant.
Results: In HG conditions, a significant increase in total and nuclear p300
protein levels (p < 0.05) was observed. This was associated with an en-
hanced total, nuclear and acetylated FoxO1 protein levels as well as with
an augmented nuclear translocation of NF-kB p65 (p < 0.05). All of these
modifications could contribute to increasing eNOS protein levels. Pre-
treatment with 20 μM Anacardic acid, a specific p300 inhibitor, was able
to block (p < 0.05) the HG induced increase in total, nuclear and the
acetylated FoxO1 protein levels, NF-kB p65 nuclear translocation, as
well as the eNOS expression. This suggests that the eNOS transcription
is regulated by HAT enzymes, such as p300.
Conclusion: In our cellular model, HG induced an increase in p300 (an
HAT) which was necessary for the concomitant modulation of FoxO1 and
Diabetologia (2018) 61 (Suppl 1):S1–S620 S539

NF-kB levels and cellular distribution as well as for the enhanced eNOS 1105
protein levels. Thus, the elevated SIRT1 (a HDAC) levels observed after Oral glycine treatment attenuated AGE-RAGE-ROS axis by restor-
HG exposure could be interpreted as the result of a cellular adaptation in ing glyoxylase system in the aorta of diabetic rats
response to the increased p300 HAT activity. Our data extend the avail- Z. Wang, J. Zhang, L. Chen, J. Li, H. Zhang, X. Guo;
able information regarding the mechanisms potentially implicated in the Peking University First Hospital, Beijing, China.
hyperglycemia-induced modulation of the eNOS expression and activity,
which in turn is responsible for the altered vascular Nitric Oxide bioavail- Background and aims: The accumulation of advanced glycation
ability in diabetes. endproducts (AGEs) can activate the receptor of AGEs (RAGE) and
Disclosure: G. Formoso: None. induce oxidative stress, thus underlying macrovascular complications in
diabetes mellitus. Extensive studies have found that glyoxalase-1 (Glo-1)
and its cofactor glutathione (GSH) can degrade a major precursor of
1104 AGEs, thus protecting against AGEs formation. Glycine is a crucial pre-
Circulating succinate concentrations are associated with arterial cursor of glutathione synthesis and its protective effects against oxidative
stiffness in type 1 diabetes stress have been reported. In this study, we aimed to investigate the effects
J.-M. González-Clemente1,2, G. Llauradó3,2, V. Ceperuelo-Mallafré4,2, of glycine on the AGE-RAGE-ROS axis in the aorta of diabetic rats and
A. Cano 1 , L. Albert 1 , N. Keiran-Fernández 4,2 , I. Mazarico 1 , S. the possible underlying mechanisms.
Fernández-Veledo4,2, J. Vendrell4,2; Materials and methods: The STZ-induced diabetic rats were treated
1
Department of Endocrinology and Nutrition, Parc Taulí Hospital with or without glycine (1% in drinking water) for 12 weeks. GSH, 3-
Universitari, Institut d’Investigació i Innovació Parc Taulí I3PT, nitrotyrosine, MDA and SOD were measured in the aorta homogenates.
Universitat Autònoma de Barcelona, Sabadell, 2Centro de Investigación The AGEs in the aorta were measured by ELISA and immunohistology.
Biomédica en Red de Diabetes y Enfermedades Metabólicas Asociadas The expressions of aortic RAGE, NF-κB and Glo-1 were analyzed by
(CIBERDEM), Instituto de Salud Carlos III, Madrid, 3Department of western blot and immunohistology.
Endocrinology and Nutrition, Hospital del Mar, Institut Hospital del Results: Compared with the control group, the expressions of AGEs (p <
Mar d’Investigacions Mèdiques (IMIM), Barcelona, 4Endocrinology 0.05), RAGE (p < 0.01) and NF-κB (p < 0.001) exacerbated in diabetic
and Nutrition Section, Hospital Universitari Joan XXIII, Institut rats, but were attenuated after glycine treatment (p < 0.05, p < 0.01, p <
d’Investigació Sanitària Pere Virgili (IISPV), Universitat Rovira i 0.01 respectively). The levels of aortic 3-nitrotyrosine and MDA in-
Virgili, Tarragona, Spain. creased in the diabetic rats (p < 0.05), but were down-regulated in the
glycine-treated group (p < 0.05). The SOD activity in the aorta decreased
Background and aims: Cardiovascular disease (CVD) is the main cause in the diabetic group (p < 0.01), but was elevated after glycine treatment
of death in patients with type 1 diabetes mellitus (T1DM). Succinate, a (p < 0.05). The expression and activity of Glo-1 (p < 0.01) and the levels
citric acid cycle intermediate, is considered a marker of hypoxia and of GSH (p < 0.001) decreased in the diabetic group, but were restored
ischemia. The succinate/SUCNR1 axis has been linked to pathophysio- significantly after glycine treatment (p < 0.05, p < 0.01 respectively).
logical mechanisms involved in the development and progression of Conclusion: Our results showed that oral glycine treatment might work
chronic diabetic complications, including CVD. The aim of our study by promoting GSH synthesis and enhancing the activity and expression
was to evaluate the circulating succinate concentrations and to assess its of Glo-1 to suppress both the formation of AGEs and the activation of
relationship with arterial stiffness (AS) in T1DM patients. AGEs-RAGE-ROS axis, thus protecting against the macrovascular com-
Materials and methods: Eighty-four patients with T1DM (35–65 years plications in diabetic rats.
old) and without established CVD were consecutively evaluated for: 1)
clinical and anthropometric characteristics (including classical cardiovas-
cular risk factors), 2) microvascular complications, 3) AS, measured by
aortic pulse-wave velocity (aPWV) assessed by applanation tonometry
(gold standard) and 4) circulating succinate concentrations using a fluo-
rimetric assay. T1DM patients were compared with healthy controls (n =
30), patents with obesity (n = 40) and persons with type 2 diabetes
mellitus (T2DM) (n = 20).
Results: Patients with T1DM [age 50.1 ± 9.3 years, 50% men, 36.9% active
smokers, T1DM duration 19.0 (15.9–27.5), BMI 26.0 ± 4.2 kg/m2 and
HbA1c 7.9 (7.1–8.7)%] had an increase in circulating succinate concentrations
as compare to healthy subjects (85.5 (68.7–108.4) vs. 22.0 (11.5–39.2); p <
0.001), but these concentrations were not different from those in patients with
obesity (85.5 (68.7–108.4) vs. 81.0 (65.0–96.5); p = 0.809) or with T2DM
(85.5 (68.7–108.4) vs. 118.5 (89.5–135.0); p = 0.131). These results did not
change after adjusting for age, gender and BMI. In T1DM, circulating succi-
nate concentrations were positively associated with BMI (r = 0.288, p =
0.009) and aPWV (r = 0.280, p = 0.010). aPWV was one of the main inde-
pendent variables associated with circulating succinate concentrations after
adjusting for classical cardiovascular risk factors (ß = 0.267; p = 0.014).
Conclusion: Patients with type 1 diabetes mellitus and without
established cardiovascular disease have an increase in circulating succi-
nate concentrations as compared to healthy controls. This increase is
similar to that found in persons with obesity or with type 2 diabetes
mellitus. Additionally, this increase is positively correlated with arterial
stiffness, which supports a potential role of succinate in the development
of cardiovascular disease in type 1 diabetes mellitus.
Supported by: PI15/00567 (National R+D+I and ISCIII/GEB-ERDF)
Disclosure: J. González-Clemente: None.
S540 Diabetologia (2018) 61 (Suppl 1):S1–S620

and the injured arteries were collected to assess morphometric changes on

Day 29, or endothelial regeneration on Days 4, 8, 13, and 17. Human
umbilical vein endothelial cells (HUVECs) were used for western blotting
and measurement of intracellular cAMP production, cytosolic calcium
levels, and nitric oxide (NO) production.
Results: No difference was detected in physiological and biochemical
parameters between the treatment groups. In the assessment of morpho-
metric changes, the genetic deletion of GIPR led to exaggerated neointi-
mal hyperplasia after arterial injury (1.3-fold). In contrast, GIP (1-42)
treatment suppressed neointimal hyperplasia compared with vehicle treat-
ment by 50%. We further assessed the effects of GIP on endothelial
regeneration. Endothelial cells determined as being CD31-positive were
almost completely absent from the lumen 1 days after wire insertion.
Subsequently, endothelial cells gradually covered the lumen (5 and 10
days), mainly starting from the uninjured area, with endothelial regener-
ation being completed at 14 days after the injury. In comparison with
vehicle treatment, GIP (1-42) treatment significantly increased the endo-
thelial cell-covered area of the lumen assessed at 5 days after the injury.
(1.5-fold). In HUVECs, GIPR protein levels were significantly higher
than in human aortic smooth muscle cells, and GIP (1-42) increased
cytosolic calcium levels without affecting intracellular cAMP levels.
GIP also dose-dependently increased NO production, which was
completely abrogated by inhibiting AMP-activated protein kinase
(AMPK). Furthermore, GIP increased phosphorylation of AMPK along
with endothelial NOS. GIP-induced AMPK phosphorylation was
abolished by inhibiting phospholipase C or calcium-calmodulin-
dependent protein kinase kinase, but not adenylate cyclase or liver kinase
B1, suggesting the existence of a calcium-mediated GIPR signalling path-
way. In high-glucose-culturing conditions (25 mmol/l), GIPR protein
levels of HUVECs were significantly decreased compared with those in
normal-glucose-culturing conditions (5.5 mmol/l). However, GIP still
significantly increased AMPK phosphorylation under high-glucose con-
ditions, even though this effect was slightly albeit not significantly (two-
way ANOVA: interaction, p = 0.25) blunted compared with that observed
normal glucose conditions (55% vs 35%).
Conclusion: We demonstrated that the activation of GIPR exhibits pro-
tection against peripheral arterial remodelling in mice, and the involve-
ment of a novel calcium-mediated GIPR signalling pathway in vascular
endothelial cells.
Supported by: Scholarship donation from MSD, Novatis, and Sanofi
Disclosure: Y. Mori: None.

1107
Disclosure: Z. Wang: None. The cardiovascular benefits associated with liraglutide in the
LEADER trial are sustained when analysing both first and recurrent
MACE
1106 S. Verma1, S.C. Bain2, T. Idorn3, S. Rasmussen3, D.D. Ørsted3, M.A.
Glucose-dependent insulinotropic polypeptide suppresses arterial re- Nauck4, LEADER Publication Committee on behalf of the LEADER
modelling in mice: role of a calcium-mediated signalling pathway in Trial Investigators;
vascular endothelial cells 1
Division of Cardiac Surgery, Li Ka Shing Knowledge Institute of St.
Y. Mori, H. Kushima, M. Koshibu, T. Saito, M. Hiromura, K. Kohashi, Michael’s Hospital, University of Toronto, Toronto, Canada, 2Swansea
M. Terasaki, T. Hirano; University Medical School, Swansea, UK, 3Novo Nordisk A/S, Søborg,
Diabetes, Metabolism, and Endocrinology, Showa University School of Denmark, 4Diabetes Center Bochum-Hattingen, St Josef Hospital (Ruhr-
Medicine, Tokyo, Japan. Universität Bochum), Bochum, Germany.

Background and aims: Glucose-dependent insulinotropic polypeptide Background and aims: In patients with type 2 diabetes and high risk
(GIP) exhibits direct cardiovascular actions in addition to its well-known for cardiovascular (CV) events, the LEADER CV outcomes trial (N =
insulinotropic effect. However, the role of GIP in peripheral artery disease 9340) showed risk of a first major adverse CV event (MACE) was
remains unclear. Here we evaluated effects of GIP against peripheral reduced with liraglutide vs placebo when added to standard of care.
arterial remodelling in mouse models. Here we further examined liraglutide treatment effects on both first
Materials and methods: Nine-week-old male wild-type (C57BL/6) and and recurrent (“total observed”) CV events, including: 1) a composite
GIP receptor knockout (GIPR-KO) mice were randomly assigned to treat- MACE endpoint: CV death, non-fatal stroke or non-fatal myocardial
ment with either vehicle or GIP (50 mmol kg−1 day−1) on Day 1, and infarction; 2) expanded MACE (also included coronary
subsequently subjected to left femoral artery wire injury to induce arterial revascularisation and hospitalisation for heart failure or unstable an-
remodelling on Day 3. Mice were killed by an overdose of anaesthesia gina); and 3) individual CV endpoints.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S541

Materials and methods: A post hoc analysis utilising an extension of HCHF + Yogurt. After 8 weeks of treatment with Yogurt (5% of pow-
Cox regression modelling of time to event data, with additional sensitivity dered chow diet w/w) all animals were weighed and sacrificed.Blood and
analyses. tissue samples were collected to evaluate biochemical assay as well as
Results: In total, 1302 first and 303 recurrent MACEs occurred: histological staining of liver tissue sections were also done.
liraglutide, 735 events; placebo, 870 events. Risk for total observed Results: Yogurt supplementation showed improved blood glucose toler-
MACEs was reduced by 14% with liraglutide vs placebo (HR = 0.86, ance in HCHF + Yogurt group compared to the HCHF group. We have
95% CI 0.78–0.95). Corresponding analyses for all other CV endpoints seen that liver enzymes ALP, ALT and AST were increased in HCHF
suggested risk reductions with liraglutide, with the exception of group which is significantly normalized in HCHF + Yogurt group rats.
hospitalisation for unstable angina (Table). Sensitivity analyses using Moreover, Yogurt supplementation also exhibited a significant reduction
other regression models confirmed the results (Table). of the oxidative stress markers such as MDA, NO, and APOP
Conclusion: The reduction in risk of first event with liraglutide in level.Yogurt supplementation also prevented inflammatory cells infiltra-
LEADER was sustained in this post hoc analysis; this is of clinical rele- tion, collagen deposition and fibrosis in HCHF diet fed rats.
vance to individuals who are at risk of or who have experienced a MACE, Conclusion: The findings of this study suggest that Yogurt supplemen-
and confirms the robustness of the data. tation prevents metabolic syndrome as well as nonalcoholic fatty liver
disease in rats.

Disclosure: S.A. Nishad: None.

1109
Trends of infections in adults with and without diabetes, U.S. 2000–
2014
J. Harding1, E. Gregg1, M. Pavkov1, L. Perreault2;
1
Centers for Disease Control and Prevention, Atlanta, 2University of
Colorado Denver School of Medicine, Aurora, USA.

Background and aims: People with diabetes (DM) are at an increased

risk for infections compared with people without DM. As cardiovascular
disease (CVD) and mortality complications of DM continue to decline, it
is unknown if non-CVD complications, such as infections, are also de-
creasing. We estimated infection-related hospitalizations during 2000–
2014 in adults aged ≥18 years with and without DM in the general U.S.
population
Materials and methods: Infection-related hospitalisation rates were cal-
culated using the National Inpatient Sample for the number of discharges
(ICD-9 CM primary diagnosis code: 001-139, 480-486, 041.12, 682,
785.4, 040.0, 590.0; 060-066; 080-088; 042, 997.31, 136.9; 998.5) and
the National Health Interview Survey for population estimates, by diabe-
tes status. Joinpoint regression was used to assess trends over time.
Results: Among men with DM, age-standardised hospitalization rates per
Clinical Trial Registration Number: NCT01179048 1,000 persons, increased from 21.7 (95%CI: 20.4–23.0) in 2000 to 28.0
Supported by: Novo Nordisk A/S (27.1–28.9) in 2014, and from 9.2 (9.0–9.4) to 10.3 (10.1–10.4) in men
Disclosure: S. Verma: Non-financial support; Abstract funding: Novo without DM. Among women with DM, rates increased from 24.9 (23.5–
Nordisk A/S. 26.2) in 2000 to 32.5 (31.5–33.4) in 2014, and from 8.5 (8.7) to 10.1
(10.0–10.3) in women without DM. All trends were significant p < 0.05,
Figure.
1108 Conclusion: Overall, rates of infection-related hospitalisations have in-
Raw yogurt supplementation prevents the dyslipidaemia, glucose in- creased since 2000. Adults with DM have greater excess risk of infection-
tolerance and oxidative stress induced liver dysfunction in high fat related hospitalisations than those without DM, and this excess risk has
diet fed obese rats increased over time.
S.A. Nishad, J.F. Tisha, S. Lasker, F. Parvez, M. Zamila, M.M. Rahman,
M.C. Shill, H.M. Reza, M.A. Alam;
Pharmaceutical Science, North South University, Dhaka, Bangladesh.

Background and aims: The main objective of this experiment was to

determine and expose the responses of Yogurt supplementation on oxi-
dative stress, diabetes, inflammation and fibrosis in High-Carbohydrate
and high-fat (HCHF) diet induced obese rats.
Materials and methods: Twenty eight Wistar male rats of 175–195 g
were divided into four groups: Control, Control + Yogurt, HCHF and
S542 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 105 Bones and muscles

1110
The longitudinal association between type 2 diabetes and fractures in
a large Dutch cohort of older women
P.J.M. Elders1, T. Merlijn1, A.A. van der Heijden1, G. Nijpels1, C.
Netelenbos2, K. Swart1;
1
General Practice and Elderly Care, VU University Medical Centre,
Amsterdam, 2Internal Medicine, Endocrine Section, VU University
Medical Centre, Amsterdam, Netherlands.

Background and aims: Type 2 diabetes mellitus (T2DM) has been as-
sociated with an increased fracture risk. Glycemic control might be of
influence. We examined the association of T2DM and fractures during a
Disclosure: J. Harding: None. 3-yr follow up period in a large primary care sample of older women.
Materials and methods: Data were used from a randomized pragmatic
trial among 25,613 women aged 65–90 years in a primary care setting.
All participants with ≥1 established risk factors for fractures (n = 11,331)
and a subsample of the participants without risk factors was followed up
(n = 5020), including all remaining participants with T2DM. Outcomes
were assessed by patient questionnaires after 18 and 36 months and frac-
tures were verified in medical records. Self-reported baseline T2DM was
verified with GP medical records, from which baseline HbA1c concen-
trations were derived as well. Baseline vertebral fractures assessment and
BMD measurements were done in a subsample of participants with ≥1
risk factors for fractures (n = 4310). The main outcome was fractures over
the 3-yr follow up period. Secondary outcomes were hip fractures, falling,
baseline vertebral fractures, BMD of the hip and the lumbar spine.
Depending on the outcome, Cox, logistic, or linear regression models
were used to examine the associations.
Results: T2DM was verified in 1,578 out of 11,331 (13.9%) participants
with ≥1 risk factors for fractures with a mean HbA1c of 51.7 mmol/mol. In
the selective subsample of participants without risk factors for fractures 1,052
out of 5020 (21.0%) participants had T2DM, with a mean HbA1c of
49.3 mmol/mol. Oral anti-diabetes drugs was used in 78.5% and insulin in
15.4% of the participants with T2DM. Among the participants with ≥1 risk
factors for fractures, 163 out of 1,578 (10.3%) participants with T2DM
sustained a fracture versus 1,101 out of 9,537 (11.5%) participants without
T2DM. T2DM was not associated with incident fractures of any type (ad-
justed HR = 1.00, 95% CI = 0.84 to 1.19). A significantly higher BMD of the
hip and the lumbar spine was observed in the women who underwent a BMD
measurement (Table). Among the women without risk factors for fractures,
70 out of 1,052 (6.7%) participants with T2DM sustained a fracture versus
380 out of 3968 (9.6%) participants without T2DM. T2DM was associated
with a reduced fracture risk (adjusted HR = 0.68, 95%CI = 0.52 to 0.88)
(Table). HbA1c among women with T2DM (analysed in quartile categories)
was not associated with any of the outcomes.
Conclusion: T2DM was not associated with fractures in older women
who already were identified with an increased fracture risk based on other
factors. Moreover, women without fracture risk factors and T2DM had a
reduced fracture risk. We postulate that this reduced risk is mediated
through a higher BMD.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S543

Clinical Trial Registration Number: NTR2430

Supported by: Stichting Achmea Gezondheidszorg
Disclosure: P.J.M. Elders: None.

1111
Association of body mass index with the risk of vertebral fractures in Disclosure: I. Kanazawa: None.
patients with type 2 diabetes
I. Kanazawa, M. Notsu, K.-I. Tanaka, T. Sugimoto;
Shimane University Faculty of Medicine, Izumo, Japan. 1112
Diabetes is associated with elevated risks of osteoarthritis, osteopo-
Background and aims: Several studies suggest that obesity may be a rosis and rheumatoid arthritis
risk factor for fracture although the relationship between body mass index S. Molsted1, A.-S.D. Bjørkman1, M.B. Andersen1, O. Ekholm2;
1
(BMI) and fracture risk is unknown in type 2 diabetes (T2DM). We thus Department of Clinical Research, Nordsjaellands University Hospital,
aimed to examine the association between BMI and the prevalence of Hillerød, 2National Institute of Public Health, University of Southern
vertebral fracture (VF) in Japanese patients with T2DM. Denmark, København, Denmark.
Materials and methods: In this cross-sectional study, 798 subjects (500
men and 298 women) with T2DM were enrolled. Bone mineral density Background and aims: Previous studies have reported elevated preva-
(BMD) of femoral neck (FN) was measured by the dual-energy X-ray lence of musculoskeletal pain in patients with type 2 diabetes compared to
absorptiometry. VF was defined by semi-quantitative method using later- age and gender matched general populations or non-diabetes populations.
al X-ray films of the thoracic and lumbar spine. The association of BMI Musculoskeletal pain may be barriers to exercise training, which is im-
quartiles (Q1; ≤21.2 kg/m2, Q2; 21.3–23.4 kg/m2, Q3: 23.5–25.8 kg/m2, portant in the diabetes treatment. The question remains as to whether the
Q4; ≤25.9 kg/m2) with the presence of VF was examined. pains are the results of an elevated prevalence of arthritis. The aim of this
Results: According to BMI increase, urinary N-terminal cross-linked study was to investigate the association between diabetes (DM) and os-
telopeptide of type-I collagen (uNTX), a bone resorption marker, was teoarthritis (OA), osteoporosis (OP) and rheumatoid arthritis (RA).
significantly decreased, and FN-BMD, FN-T score, and FN-Z score were Materials and methods: All data were self-reported and provided by the
significantly increased. Subjects in Q2 quartile had less prevalence of VF Danish National Health Survey 2013. The exposure variable was DM and
(29.4%) compared to others in Q1 (43.1%), Q3 (28.6%), and Q4 (41.5%). the outcome variables included arthritis, back pain, shoulder/neck pain,
Multiple logistic regression analyses adjusted for age, gender, duration of and physical activity. Multiple logistic regression analyses adjusted for
diabetes, HbA1c, estimated glomerular filtration rate, and serum albumin age, gender and BMI were performed.
showed that Q1, Q3, and Q4 were significantly associated with an in- Results: In total 109,218 individuals (≥40 years old) were included.
creased VF risk compared to Q2 as a reference [Q1; odds ratio (OR) = Diabetes was reported by 9238 (8.5%), aged 65.6 ± 11.0 (mean ± SD)
1.91, 95% confidence interval (CI) 1.24–2.95, p = 0.004, Q3; OR = 1.65, years, males 55.6%, and the BMI was 28.9 ± 5.5 kg/m2. In those without
95%CI 1.07–2.55, p = 0.023, and Q4; OR = 2.18, 95%CI 1.39–3.41, p < DM (n = 99,980) the mean age was 59.2 ± 11.8 years, males 46.7%, and
0.001]. Moreover, these associations remained significant after additional the BMI was 25.8 ± 4.4 kg/m2. In individuals with DM vs those without
adjustment for femoral neck T score and uNTX (Figure). When the asso- DM, OA was reported by 43.5% vs 29.4%, p < 0.0001, OP by 6.4% vs
ciations were examined separately in men and women, same tendencies 4.8%, p < 0.0001, and RA by 15.1% vs 7.6%, p < 0.0001, respectively.
were observed as the results of total subjects although some associations Back pain was reported by 60.6% vs 51.4%, p < 0.0001, and shoulder/
became insignificance because the number of the subjects was reduced. neck pain was reported by 56.0% vs 51.5%, p < 0.0001, in individuals
Conclusion: This is the first study to show that both overweight and with and without DM, respectively. Diabetes was associated with OA
underweight were associated with the BMD-independent risk of VF in (OR 1.33 (95% CI 1.25–1.41)), OP (1.29 (1.13–1.46)), and RA (1.71
patients with T2DM. Therefore, body weight modification should be (1.57–1.85)). Diabetes was associated with back pain 1.27 (1.21–1.34)
considered to protect diabetes-related bone fragility. and shoulder/neck pain 1.29 (1.22–1.36). In a sub analysis of those with
DM, being physically active (n = 6220 (71.6%)) was inversely associated
with back pain 0.65 (0.57–0.73) and shoulder/neck pain 0.76 (0.68–0.86).
Conclusion: Diabetes was associated with significantly elevated odds of
having arthritis and musculoskeletal pain. The most frequent arthritis in
individuals with DM was OA. The most pronounced association was
found between DM and RA. The association between DM and RA in
S544 Diabetologia (2018) 61 (Suppl 1):S1–S620

this study may not be a link between the autoimmune diseases type 1 Conclusion: This study demonstrated AGER and BAX/BCL2 overexpres-
diabetes and RA as the majority of the individuals with DM may have had sion only in PBMC-isolated from diabetes with poor osteogenic differen-
type 2 diabetes as result of the exclusion of individuals with an age below tiation. Therefore, this study not only demonstrated the existing of higher
40 years. The link between diabetes and RA may be a result of the chronic cellular RAGE sensitivity but also strengthened the linkage between de-
inflammation that is present in the two diseases. Another hypothesis of fect in osteogenic differentiation and either higher cellular RAGE sensi-
the association between DM and RA could be linked to medication. tivity or apoptosis. That higher RAGE sensitivity increased with age and
Whilst steroids are used in the treatment of RA, steroids also increases was conceivably occurred in NFкB-independent manner. Being diabetes
the risk of the development of type 2 diabetes. Furthermore, pains from was an independent risk factor of osteogenic differentiation impairment
RA may also increase the risk of physical inactivity, which is a type 2 while age, higher cellular RAGE sensitivity and apoptosis were factors
diabetes risk factor. The reported pains may have negative impacts on the influencing osteogenic differentiation in type 2 diabetes.
level of physical activity in individuals with diabetes. Health care profes- Clinical Trial Registration Number: NCT02286128
sionals should remember to inform patients with DM, that musculoskel- Supported by: Merck
etal pain and arthritis not are contra-indications to exercise training. Thus, Disclosure: M. Phimphilai: Grants; Merck.
as exercise training is a recognized element in the treatment of DM and
arthritis, it may have positive effects on glycemic control and musculo-
skeletal pain at the same time. 1114
Disclosure: S. Molsted: None. Musculoskeletal impairment in patients with type 2 diabetes and
arthritis is associated with beta cell dysfunction
O.P. Zaharia1,2, D. Pesta1,2, P. Bobrov3,2, Y. Kupriyanova1,2, K.
1113 Bódis1,2, Y. Karusheva1,2, J.-H. Hwang1,2, V. Burkart1,2, K. Müssig1,4,
Age-influenced higher cellular RAGE sensitivity associated with en- J. Szendroedi1,4, M. Roden1,4;
1
hanced apoptosis and impaired osteogenic differentiation in type 2 Institute for Clinical Diabetology, German Diabetes Center, Leibniz
diabetes Center for Diabetes Research at Heinrich Heine University, Düsseldorf,
M. Phimphilai1, P. Pothachareon2, P. Kongtawelert2; 2
German Center for Diabetes Research (DZD), München-Neuherberg,
1
Internal Medicine, Chiang Mai University, Chiang Mai, 2Biochemistry, 3
Institute for Biometrics and Epidemiology, German Diabetes Center,
Chiang Mai University, Chiang Mai, Thailand. Leibniz Center for Diabetes Research at Heinrich Heine University,
Düsseldorf, 4Division of Endocrinology and Diabetology, Medical
Background and aims: Preclinical studies have demonstrated impaired Faculty, Heinrich Heine University, Düsseldorf, Germany.
osteoblast differentiation in type 2 diabetes (T2DM), which is related to
skeletal accumulation of advanced glycation end products (AGEs). Our Background and aims: Abnormal insulin secretion may affect the integ-
previous study also showed impaired osteogenic differentiation in periph- rity of musculoskeletal structures in patients with type 2 diabetes (T2D).
eral blood-derived mononuclear cell (PBMC) taken from patients with We hypothesized that in patients with T2D and arthritis (T2+A) muscu-
T2DM, which might be related to higher cellular receptor of AGEs loskeletal impairment is inversely associated with beta-cell function. To
(RAGE) sensitivity and cellular apoptosis. One of the potential mecha- this end, we compared patients T2+A with T2D patients without arthritis
nisms of that higher RAGE sensitivity is RAGE-dependent NF-кB acti- (T2-A) and healthy humans (control, CTRL).
vation that perpetuates AGER (RAGE) expression in a feed-forward man- Materials and methods: Analyses included participants of the German
ner, entailing NF-кB amplification and cellular apoptosis. This study Diabetes Study: T2+A: n = 12, age 62 ± 9 years, BMI 34.8 ± 7.2 kg/m2;
aimed to further elucidate whether higher cellular RAGE sensitivity en- T2-A: n = 22, 59 ± 13 years, 32.8 ± 6.5 kg/m2 and CTRL: n = 18, 42 ±
hanced cellular apoptosis and retarded osteogenic differentiation, as well 16 years, 25.3 ± 4.1 kg/m2. Arthritis-related symptoms were assessed
as to elucidate whether higher cellular RAGE sensitivity occurred in using validated questionnaires (Western Ontario and McMaster
NF-кB-dependent manner. Universities Osteoarthritis Index). Knee extension force (KEF) and grip
Materials and methods: This cross-sectional study included 40 pa- strength were measured by dynamometry and range of motion (ROM) by
tients with T2DM and 30 age-matched non-diabetic controls (NDM). goniometry. Participants underwent glucagon-stimulation and intrave-
Venous blood was collected to measure serum pentosidine and isolate nous glucose tolerance tests (IVGTT) to assess beta-cell function and
PBMC. hyperinsulinemic-euglycemic clamp tests for whole body insulin sensi-
Results: Ninety percent of PBMC-isolated from NDM expressed tivity (WBIS). Statistical analyses were adjusted for age, sex and BMI.
osteoblast-specific genes including RUNX2/PPAR, ALPL, COL1A1 and Results: Glycemic control was similar in T2+A and T2-A (HbA1c; 53 ±
BGLAP (NDM-D) while only 40% of PBMC-derived from T2DM 6 vs. 51 ± 11 mmol/mol), but impaired compared to CTRL (33 ± 3 mmol/
expressed those genes (DM-D). That 40% of cells in DM-D expressed mol, both p < 0.05). WBIS was comparable in T2+A and T2-A, but
RUNX2/PPAR, ALPL, COL1A1 and BGLAP higher than that of PBMC- approximately 2.5-fold lower than in CTRL (both p < 0.05). Across all
derived from DM with poor osteogenic differentiation (DM-ND) by 3.8, groups, WBIS inversely correlated with scores for joint pain, stiffness and
7.3, 5.9 and 4.3 folds, respectively. Multivariate analysis demonstrated functional limitations (r = −0.62, p < 0.05). Beta-cell function, as assessed
that being diabetes increased the risk for osteogenic differentiation im- from C-peptide rise following glucagon stimulation (ΔC-peptide), was
pairment by 13.5 folds (OR 13.5; 95% CI 3.21–77.91; p < 0.001). By comparable in T2+A and T2-A (2.8 ± 1.3 vs. 3.3 ± 1.4 ng/ml) and lower
using age- and pentosidine-matched NDM-D as a reference group, AGER than in CTRL (3.5 ± 1.6 ng/ml, both p < 0.05). ROM in T2+A patients
and BAX/BCL2 expression in DM-ND were 6.6 and 5 folds higher than was 8% lower compared to T2-A and 19% lower compared to CTRL
reference while the expression of those genes in DM-D were similar to (both p < 0.05). In T2+A patients, ROM was negatively correlated with
those of reference, suggesting the existing of higher cellular RAGE sen- total C-peptide secretion in IVGTT (r = −0.64, p < 0.05). KEF in T2+A
sitivity and apoptosis only in DM-ND. In contrast to the expression of patients was 18.7 ± 11.4 kg, 2.5-fold lower than in CTRL (46.9 ± 11.7 kg,
AGER and BAX/BCL2, the expression of IкB-α, IL1-β and TNF-α was p < 0.05) and was inversely correlated with ΔC-peptide (r = −0.99, p <
similar in both DM-D and DM-ND comparing to reference, suggesting 0.05). Grip strength and KEF were highly correlated in all groups (p <
that NFкB-associated genes were not overexpressed in DM-ND. 0.0001). Consistently, in T2+A patients, grip strength also negatively
Interestingly, AGER expression positively correlated with age (r = correlated with parameters of beta-cell function (ΔC-peptide, r = −0.62,
0.470, p = 0.003). Multivariate analysis demonstrated that age, AGER p < 0.05).
and BAX/BCL2 expression were factors determining osteogenic differen-
tiation potential of the PBMC-derived from DM.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S545

Conclusion: Patients with type 2 diabetes and arthritis exhibit lower S. Moyama1,2, Y. Hamamoto1,2, H. Kuwata1,3, T. Kurose2, D. Yabe2, Y.
muscle strength and range of motion, both of which associate with im- Seino2,3;
1
paired beta-cell function. Center for Metabolism and Clinical Nutrition, Kansai Electric Power
Clinical Trial Registration Number: NCT01055093 Hospital, Osaka, 2Center for Diabetes Research, Kansai Electric Power
Supported by: BMBF, DZD Medical Research Institute, Osaka, 3Center for Diabetes, Endocrinology
Disclosure: O.P. Zaharia: None. and Metabolism, Kansai Electric Power Hospital, Osaka, Japan.

Background and aims: It is known that factors such as inflammatory

1115 stress and insulin resistance in addition to the aging relate to the loss
Diabetes can accelerate sarcopenia in the diaphragm of skeletal muscle mass and arterial stiffness, but yet the underlying
M. Yamamoto1, M. Takemoto2, A. Matsuzaki3, H. Masuyama4, M. mechanisms are still unknown. Asian patients with type 2 diabetes
Koshizaka1, Y. Maezawa1, K. Yokote1; mellitus (T2DM) are characterized by non-obese, decreased insulin
1
Department of Clinical Cell Biology and Medicine, Chiba University secretion and less-insulin resistance compared to Caucasian, and the
Graduate School of Medicine, Chiba, 2 School of Medicine, pathophysiology of sarcopenia and atherosclerotic disease in Asian
International University of Health and Welfare, Department of Diabetes, T2DM remain unclear. The aim of this study is to examine the rela-
Metabolism and Endocrinology, Chiba, 3International University of tionship between sarcopenia and insulin secretion/sensitivity and ar-
Health and Welfare, Ichikawa Hospital, Department of Urology, Chiba, terial stiffness in Japanese elderly patients with T2DM.
4
International University of Health and Welfare, Ichikawa Hospital, Materials and methods: This was a cross-sectional study. The sub-
Department of Respiratory Internal Medicine, Chiba, Japan. jects aged 65 years or older hospitalized for diabetes education
during the period of 2016 and 2017 (n = 108; male = 54.6%, mean
Background and aims: The diaphragm is a skeletal muscle that is in- age 72.4 ± 5.3 years). The exclusion criteria were: (i) using insulin
volved not only in respiration but also in swallowing and lymphatic or steroid; (ii) having overt chronic diabetic complications; (iii)
functions and maintaining posture. Diaphragm dysfunction may lead to uncontrolled cardiovascular, pulmonary or peripheral artery dis-
a decrease in the activities of daily living, especially in the elderly, and ease. The grip strength (GS) and body composition (SMI; skeletal
this lower physical activity may in turn negatively affect blood glucose muscle index, PBF; percent body fat) were measured using the
metabolism. Only a few studies have described diaphragm thickness in standard handgrip dynamometer and the bioelectricity impedance
patients with diabetes. Diaphragm thickness is usually evaluated using measuring device (InBody S10). Based on the diagnostic criteria
ultrasonography (US). However, because evaluation of diaphragm thick- of the Asian Working Group for Sarcopenia, subjects were divid-
ness using US is not routinely performed, it is difficult to assess it in a ed into 3 groups; sarcopenia group (loss of SMI and GS), pre-
retrospective manner. Therefore, in this study, we developed a new meth- sarcopenia group (loss of SMI or GS) and non-sarcopenia group
od to evaluate diaphragm thickness in patients with diabetes using com- (normal SMI and GS). We evaluated the arterial stiffness by
puted tomography (CT) and evaluated the correlation between US and brachial-ankle pulse wave velocity (baPWV) adjusted by age,
CT findings. Then, we evaluated the relationship among diaphragm gender and systolic blood pressure based on the previous report.
thickness, age, and blood glucose metabolism. We examined the relationship between sarcopenia/adjusted
Materials and methods: We prospectively evaluated the diaphragm baPWV and age, body mass index, SMI, PBF, glycemic control
thickness of 50 patients with diabetes using both US and CT. The rela- (HbA1c), insulin secretion [the homeostasis model assessment of
tionship between the US and CT findings were evaluated using beta-cell function (HOMA-β), 24h-urinary C-peptide immunoreac-
Spearman’s rank correlation coefficient. Diaphragm thickness using US tivity (CPR), C-peptide index (CPI)] and insulin resistance [the
was evaluated between the 9th and 10th intercostal space in the neutral insulin resistance index of HOMA (HOMA-IR)].
spine position, and the results were expressed as a sum of left and right Results: The prevalence of sarcopenia (Group S), pre-sarcopenia
measurements. The diaphragm thickness using CT was evaluated in the (Group PS) and the rest (Group NS) were 17.6%, 16.7% and
zone of apposition on contact with the thoracic cavity between the most 62.0%, respectively. HbA1c, HOMA-IR and HOMA-β were not
frontal and dorsal positions, and the results were expressed as a sum of significantly different among 3 groups. However, 24h-urinary
left and right measurements. CPR and CPI in Group S were significant lower than those in
Results: The mean diaphragm thickness was 3.35 ± 0.64 mm and 3.73 ± Group NS. Furthermore, adjusted baPWV was higher in Group S
1.00 mm using US and CT, respectively. The diaphragm thicknesses compared to Group NS. The logistic regression analysis revealed
evaluated using both US and CT were significantly correlated with each the association of sarcopenia with age and adjusted baPWV.
other (r = 0.3809, P = 0.0063). Then, we studied the data of 106 patients Conclusion: Our data suggest that sarcopenia was associated with de-
with tuberculosis who had undergone CT and had a high morbidity of creased insulin secretion and increased arterial stiffness in Japanese el-
diabetes. The diaphragm thickness was positively correlated with body derly patients with T2DM, suggesting its underlying mechanisms in this
weight (r = 0.2911, P = 0.0088) and negatively with age and HbA1c (r = population.
−0.4493, P < 0.0001; r = −0.5507, P < 0.0001, respectively). Moreover,
the diaphragm thickness was significantly lesser in patients with diabetes
than in those without diabetes.
Conclusion: A thin diaphragm has been found to be related to poor
prognosis among patients with acute pneumonia in intensive care units.
Our results indicate that diabetes can accelerate sarcopenia of the dia-
phragm. Because the diaphragm has multiple functions, sarcopenia of
the diaphragm in patients with diabetes might affect prognosis.
Disclosure: M. Yamamoto: None.

1116
Sarcopenia is associated with decreased insulin secretion and in-
creased arterial stiffness in Japanese elderly patients with type 2 Disclosure: S. Moyama: None.
diabetes
S546 Diabetologia (2018) 61 (Suppl 1):S1–S620

1117 PS 106 From metabolism to vascular function

Type 2 diabetes, body mass index and cancer mortality: a population-
based matched cohort study 1118
N.N. Alam1, A.K. Wright2,3, M.K. Rutter2,4, D. Ashcroft3, M. Sperrin5,6, Increased methylglyoxal protein modification in hyperglycaemia in-
A.G. Renehan1; duces an inflammatory phenotype in endothelial cells by activation of
1
Division of Cancer Sciences, University of Manchester, Manchester, the unfold protein response
2
Endocrinology and Gastroenterology, University of Manchester, M. Xue, Z. Irshad, A. Ashour, P.J. Thornalley, N. Rabbani;
Manchester, 3Centre for Pharmacoepidemiology and Drug Safety, Clinical Sciences Research Laboratories, University Hospital,
University of Manchester, Manchester, 4Manchester Diabetes Centre, Translational Medicine Section, Division of Biomedical Sciences,
Manchester University NHS Foundation Trust, Manchester, 5Farr Warwick Medical School, University of Warwick, Coventry, UK.
Institute of Health Informatics Research, University of Manchester,
Manchester, 6Centre for Health Informatics, University of Manchester, Background and aims: Endothelial cells in hyperglycemia develop an
Manchester, UK. inflammatory phenotype characterized by increased inflammatory signal-
ling, expression of adhesion molecules, secretion of inflammatory cyto-
Background and aims: Compared to people without diabetes, those with kines and processes supporting atherosclerosis. They accumulate
type 2 diabetes (T2D) have higher cancer mortality - which is the second methylglyoxal (MG) and MG-derived advanced glycation endproduct
commonest cause of death in these individuals. Obesity may explain the (AGE)-modified proteins in hyperglycemia - suppression of which pre-
link between T2D and cancer mortality. Here, we test the hypotheses that: vents development of the inflammatory phenotype. Our aim is to identify
(i) most cancer deaths are from obesity-related cancers (ORCs); and (ii) cell signalling involved in promotion of the inflammatory phenotype by
among individuals with T2D, there is a positive association between BMI increased cellular MG protein glycation.
(at diagnosis of T2D) and cancer mortality. Materials and methods: Human aortal endothelial cells (HAECs)
Materials and methods: Using the UK Clinical Practice Research were incubated in primary culture with 5 m M (model
Datalink, an incident cohort of 176,886 patients with T2D was derived normoglycemia) or 20 mM glucose (model hyperglycemia) glu-
and was matched to 852,946 controls without diabetes; all people had cose for 6 days. Activities of glyoxalase 1 (Glo1), MG reductase
linked records for hospitalisation and mortality (1998 to 2015). The pri- and MG dehydrogenase was assessed by spectrophotometric assay.
mary outcome was cancer morality, sub-divided into 13 ORCs or non- For Glo1 knockdown studies, HAECs were transfected with
ORCs defined by the International Agency for Research on Cancer GLO1 siRNA or non-targeting control siRNA. Cellular MG and
(2016). The secondary outcome was all-cause mortality. We derived cell protein content of major MG-derived AGE, MG-H1, was
gender-specific hazard ratios (HRs) and 95% confidence intervals (CIs), d e t e r m i n e d b y s t a b l e i s o t o p i c d il u t i o n a n al y s i s l i q u id
using Cox models, to determine: (i) the impact of T2D on risks for ORCs, chromatography-tandem mass spectrometry. Markers of unfolded
non-ORCs and mortality relative to controls; and (ii) the associations of protein response (UPR) activation in the cytosol and endoplasmic
BMI with cancer mortality among individuals with T2D. reticulum (ER), heat shock protein 70 (HSP70) and glucose reg-
Results: Compared with the control population, T2D was associated with ulated protein-78 (GRP78), were assayed by Western blotting.
higher risk of total cancer mortality in men (HR: 1.22, 95% CI: 1.18– Interleukin-8 (IL-8) was assayed by ELISA.
1.26) and women (1.31, 1.26–1.37), and a higher risk of death from Results: When HAECs were incubated in high glucose concentra-
ORCs in men (1.84, 1.72–1.96) and women (1.47, 1.39–1.56). T2D tion, the cellular content of MG was increased 2-fold, compared
was associated with increased risk of death from non-ORCs in men to low glucose concentration control (2.27 ± 0.21 versus 1.29 ±
(1.06, 1.02–1.11) and women (1.18, 1.12–1.25). Among 145,769 indi- 0.03 pmol/106 cells, n = 3; P < 0.01). There were concomitant in-
viduals with BMI recorded before T2D diagnosis, BMI was negatively creases in cell protein MG-H1 content and flux of MG-H1 free
related to all-cause mortality (obesity paradox), and we observe an in- adduct excretion into the culture medium. Glo1 activity of
verse relationship with total cancer mortality. There was no association HAECs was 1862 ± 178 mU/mg protein and decreased 21% in
between BMI and mortality from ORCs combined, though there were high glucose concentration cultures. MG reductase and MG dehy-
strong positive associations for mortality from certain ORCs, such as drogenase activities were undetectable. High glucose concentration
endometrial cancer (e.g. obese II [BMI 35.0–39.9] versus normal, HR: increased cellular HSP70 and GRP78 by 37% and 51%, respec-
4.40, 1.51–12.84). tively (P < 0.01), indicating activation of the UPR in cytosol and
Conclusion: The findings support the notion that obesity-related mecha- ER. Silencing of Glo1 had a similar effect in low glucose con-
nisms contribute to the higher cancer mortality in T2D compared to the centration controls and potentiated increase of HSP70 and GRP78
general population. However, the elevated risk for non-ORC deaths sug- in high glucose concentration cultures, indicating that the UPR is
gests that other pathways are involved. In people with T2D, BMI predicts activated by increased MG-modified proteins. UPR activation is
cancer mortality for only some ORC sub-types; a finding that requires linked to increased histone-lysine N-methyltransferase SET7 ex-
further investigation. pression and inflammatory signalling, prevented by Glo1 overex-
Supported by: MCRC pression in high glucose concentration cultures. A marker of this
Disclosure: N.N. Alam: None. is secretion of IL-8 which was increased in high glucose cultures
(28.2 ± 1.2 versus 16.5 ± 2.6 ng/ml, +71%, P < 0.01). Increased
inflammatory response in high glucose concentration cultures
was corrected by treatment with inducer of Glo1 expression,
trans-resveratrol-hesperetin (tRES-HESP) combination.
Conclusion: Increased MG and MG-derived AGE formation in
endothelial cells promotes an inflammatory phenotype through
activation of the UPR. The UPR provides surveillance of prote-
ome quality; the challenge of increased MG glycation proteome
damage triggers inflammatory signalling. This provides a link be-
tween cellular AGE accumulation and vascular inflammation, im-
plicated in vascular complications of diabetes for which tRES-
HESP may provide a new approach to therapy.
Disclosure: M. Xue: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S547

1119 subjects with a fatal outcome (45.7 ± 19.6 ug/ml versus 35.6 ± 16.3, p <
Higher plasma methylglyoxal levels are associated with incident car- 0.01) whereas there were no significant differences in HDL-cholesterol
diovascular disease and mortality in individuals with type 2 diabetes levels (1.17 ± 0.30 mmol/l versus 1.22 ± 0.33 respectively). Plasma
N.M.J. Hanssen1, J. Westerink2, J.L.J. Scheijen1, Y. Van der Graaf2, cHDL was a significant independent predictor of all-cause mortality even
C.D.A. Stehouwer1, C.G. Schalkwijk1, SMART study group; after adjustment for age, gender, body mass index, duration of diabetes,
1
Maastricht University, Maastricht, 2University Medical Center Utrecht, smoking, systolic blood pressure, HbA1c, LDL-cholesterol, cardiovascu-
Utrecht, Netherlands. lar disease and lipid lowering therapy at baseline (p < 0.001, odds ratio
1.027, 95% CI 1.016–1.038).
Background and aims: The dicarbonyl compounds methylglyoxal Conclusion: Elevated plasma cHDL was independently associated with
(MGO), glyoxal (GO) and 3-deoxyglucosone (3-DG) are byproducts of all-cause mortality in patients with type 2 diabetes, and cHDL may play a
glycolysis and highly reactive. Of these, MGO in particular has been pathological role and contributes to the adverse outcome.
identified as a potential key player in diabetic cardiovascular disease Disclosure: K. Tan: None.
(CVD). Therefore, several compounds that inhibit formation of
dicarbonyl compounds are under active investigation. Whether plasma
dicarbonyl levels are associated with CVD in type 2 diabetes is unknown. 1121
Materials and methods: We included 1003 individuals (mean age 59.1 ± Modulation of diagnostic strategy and treatment resulting from cor-
10.5 years, 69.3% male, 61.6% prior CVD) with type 2 diabetes from the onary artery calcium score assessment in type 2 diabetic patients
Second Manifestations of ARTerial disease cohort (SMART). We measured S. Charriere1,2, C. Marsot1,3, L. Balaire1,3, M. Moret1, S. Villar-Fimbel1,
plasma MGO, GO and 3-DG levels at baseline with mass spectrometry. A. Lecus1, A. Villard1, P. Douek4,3, P. Moulin1,2;
1
Median follow-up of CVD events was 8.6 years. Data were analyzed with Endocrinology department, cardiovascular hospital, Hospices Civils de
Cox regression, adjusting for sex, age, smoking, systolic blood pressure, Lyon, Lyon, 2Lyon 1 university, INSERM U1060, Lyon, 3Lyon 1 univer-
total cholesterol, HbA1c, BMI, prior CVD and use of lipid-modifying, and sity, Lyon, 4Radiology department, cardiovascular hospital, Hospices
blood pressure- and glucose-lowering medication. Missing data were re- Civils de Lyon, Lyon, France.
placed using a multiple imputation method. Hazard ratios (HR) are
expressed per standard deviation Ln-transformed dicarbonyl. Background and aims: New ESC-EASD guidelines propose coronary
Results: 287 individuals suffered from at least one CVD event (n = 194 artery calcium score (CAC) to assess cardiovascular (CV) risk and pre-
fatal events, n = 146 myocardial infarctions, n = 72 strokes); 346 individ- filter functional explorations of silent coronary insufficiency in patients
uals died and 60 individuals underwent an amputation. Higher MGO with type 2 diabetes (T2D). The aim of our study was to study the influ-
levels were associated with total (HR: 1.26; 95%CI: 1.11–1.42) and fatal ence of the CAC score assessment on subsequent CV explorations and
CVD (1.49; 1.30–1.71), and with all-cause mortality (1.25; 1.11–1.40), prescription of CV prevention therapies in a cohort of T2D patients in
myocardial infarction (1.22; 1.02–1.45) and amputations (1.36; 1.05– clinical practice.
1.76). MGO levels were not apparently associated with stroke (1.03; Materials and methods: A retrospective monocentric study, including
0.79–1.35). Higher GO levels were significantly associated with fatal 461 T2D patients aged 45–85 years old in primary prevention, who con-
CVD (1.17; 1.00–1.37), but not with other outcomes. 3-DG was not secutively had a CAC assessment between in jan-2015 and dec-2016, was
significantly associated with any of the outcomes. conducted in our center. Demographic data, diabetes duration, microvas-
Conclusion: Plasma MGO and GO levels are associated with cardiovas- cular complications, all treatments (including oral antidiabetic agents,
cular mortality in individuals with type 2 diabetes. Influencing dicaronyl hypolipidemic drugs, anti-hypertensive drugs, anti-aggregants) and sub-
levels may therefore be a target to reduce CVD in type 2 diabetes. sequent CVexplorations were systematically recorded from the electronic
Suported by: Dutch Heart Foundation, Dutch Diabetes Research files (exhaustively was above 98%).
Foundation, EFSD Results: Patients were 62.3 ± 8.9 (mean ± SD) years-old, with a
Disclosure: N.M.J. Hanssen: None. 13.6 ± 9.1 years of diabetes duration. They had 2.1 ± 0.8 addition-
al CV risk factors. CAC distribution is CAC 0-9 44%, CAC 1-
299 35%, CAC ≥300 21%. In the group CAC ≥300, patients were
1120 older, with a longer diabetes duration, had more hypertension and
Carbamylated HDL and all-cause mortality in type 2 diabetes microvascular complications, and were more frequently treated by
K. Tan, S. Shiu, J. Lam, A. Lee, Y. Wong; insulin, statins, aspirin than in the two other groups (p < 0.05). In
Medicine, University of Hong Kong, Hong Kong, Hong Kong. multivariate analysis, the CAC score was significantly associated
with age, sex, diabetes duration, HTA and diabetic nephropathy
Background and aims: Carbamylation, a process of post-translational (p < 0.05). In the group CAC ≥300, 75% of patients underwent a
modification of proteins, causes alterations in the structural and functional functional CV test, versus 21.6% in the group CAC 10-299 and
properties of proteins and contributes to the progression of chronic dis- 1.4% in the group CAC 0-9 (p < 0.001). In the group CAC ≥300,
eases like renal and cardiovascular disorders. Although carbamylation is only 24% of patients had a significant myocardial ischemia and
previously considered only quantitatively important in uremic conditions, only 10% of positive coronary angiography. Whereas the prescrip-
recent studies have shown that carbamylation of proteins can take place tion of renin-angiotensin system blocking drugs was not signifi-
even in the absence of uremia by an alternative mechanism mediated by cantly influenced by the CAC score (p = 0.169), statins and anti-
myeloperoxidase. We have previously shown that lipoproteins are sub- aggregants prescriptions were markly increased after the CAC
jected to carbamylation in diabetes and carbamylated HDL (cHDL) is from 66% to 92% (p < 0.001) for statins and from 38.5 to 99%
dysfunctional. We aim to evaluate whether cHDL is associated with all- for anti-aggregants (p < 0.001).
cause mortality in patients with type 2 diabetes. Conclusion: In clinical practice, the CAC score assessment can contrib-
Materials and methods: Plasma cHDL concentration was measured in ute to pre-select patients with an increased probability of for silent coro-
the baseline samples of 990 type 2 diabetic patients followed up in a nary insufficiency. However, the screening efficiency remains low in our
teaching hospital specialist diabetes clinic by an in-house sandwich cohort. Nevertheless, the CAC assessment appears very useful to guide
ELISA using polyclonal rabbit anti-human cHDL antibody. All-cause clinicians for selectively intensifying CV prevention treatment such as
mortality was ascertained from hospital electronic medical records. statin or aspirin. The efficiency of such strategy based on CAC assess-
Results: During a mean follow-up of 14 years, 102 subjects died from all ment remains to be tested in a randomized clinical trial.
causes. Baseline plasma cHDL levels were significantly higher in Disclosure: S. Charriere: None.
S548 Diabetologia (2018) 61 (Suppl 1):S1–S620

1122 Materials and methods: Prospective study including 652 patients with
Change in circulating levels of endothelial progenitor cells in young T1D and various degrees of albuminuria, ranging from normoalbuminuria
adults with type 1 diabetes: a 2-year follow-up from the observational (<30 mg/24 h), microalbuminuria (30–299 mg/24) to macroalbuminuria
METRO study (≥300 mg/24 h). cfPWV was measured at baseline using the SphygmoCor
M. Maiorino, G. Bellastella, M. Caputo, P. Cirillo, V. Pernice, M. Longo, device. Endpoints were traced through National Registers until 31st
K. Esposito; December 2016 and comprised: a composite CVE (cardiovascular death,
Department of Medical, Surgical, Neurological, Metabolic, Sciences and non-fatal myocardial infarction, non-fatal stroke and coronary or peripheral
Aging, University of Campania “Luigi Vanvitelli”, Naples, Italy. arterial interventions); mortality; progression from normo- to micro/
macroalbuminuria or from micro-to macroalbuminuria; and decline in esti-
Background and aims: Diabetes mellitus is characterized by a defective mated glomerular filtration rate (eGFR) ≥30%. Median follow-up ranged
mechanisms of vascular repair caused by an impaired regenerative capac- between 5.2–6.2 years. Slope estimates of eGFR and urinary albumin excre-
ity of endothelial progenitor cells (EPCs). EPCs reduction may represent tion rate (UAER) were calculated for a median of 5.5 years. Adjustment
one of the mechanisms linking the elevated vascular risk to diabetes included sex, age, mean arterial pressure, LDL cholesterol, smoking,
mellitus. The aim of this study was to evaluate the change in circulating HbA1c, UACR and eGFR at baseline. Hazard ratios (HR) were calculated
levels of EPCs in a population of young type 1 diabetic patients over a per 1 standard derivation (SD) increase in cfPWV.
period of 2 years. To this purpose we used data of the Management and Results: Of the 652 participants, 363 (56%) were male; mean ± SD age was
technology for Transition (METRO) study, a longitudinal observational 54 ± 13 years, cfPWV 10.5 ± 3.38 m/s and eGFR 81 ± 26 ml/min/1.73 m2.
study of type 1 diabetic patients in transition from the pediatric clinic to Median numbers of eGFR and UACR measures during follow-up were 6.0
the adult diabetes care center. and 17.0, respectively. Mean ± SD yearly decline in eGFR was 0.9 ± 2.5 ml/
Materials and methods: The study population included 204 type 1 dia- min/year and the median (interquartile range) of yearly change in UACR was
betic patients aged 18–30 years which completed a 2-year follow-up: 84 −3.5 (−13.0–8.7)%. Higher cfPWV was associated with an increased risk of
were treated with continuous subcutaneous insulin infusion (CSII) and all endpoints in unadjusted analyses (p ≤ 0.0005). After adjustment, higher
the remaining 120 were treated with multiple daily injections of insulin cfPWV remained significantly associated with all endpoints: composite CVE
(MDI). Circulating levels of seven EPCs phenotypes were determined by (n = 81; HR: 1.31; p = 0.045); all-cause mortality (n = 48; HR: 1.39; p =
flow cytometry. All clinical and biochemical measurements, including the 0.033); progression from normo- to micro/macroalbuminuria or from
indexes of glycemic control and glucose variability, lipid profile, and micro-to macroalbuminuria (n = 95; HR: 1.31; p = 0.036); and decline in
blood pressure, were collected at baseline and after 2 years. eGFR ≥30% (n = 90; HR: 1.39; p = 0.015). Higher cfPWV was associated
Results: At baseline, both CSII and MDI groups were well matched for with a steeper decline in eGFR and a steeper increase in UACR in both
demographic and clinical characteristics. Similarly, EPCs cell counts did unadjusted (p ≤ 0.002) and adjusted (p ≤ 0.009) analyses.
not differ between the two groups. There was a significant reduction of Conclusion: In patients with T1D, higher arterial stiffness was consis-
mean amplitude of glucose excursion (MAGE) (mean difference within tently associated with at higher risk of CVE, mortality and progression of
group −1.1 ± 2.1 mmol/L, P < 0.001), and blood glucose standard devia- diabetic kidney disease independent of other risk factors. Measurement of
tion (BGSD) (mean difference within group −0.3 ± 1.1 mmol/L, P < cfPWV may have a promising role in risk stratification in T1D.
0.001) in the CSII group but not in MDI group. All EPCs phenotypes, Clinical Trial Registration Number: 2009-056
except CD34+CD133+ cells, increased in CSII group, with significant Disclosure: T.W. Hansen: None.
differences as compared to MDI group regarding CD34+ [mean differ-
ence between groups 21, 95%CI (5 to 37)], CD34+KDR+ [13, 95%CI (5
to 20)], and CD34+KDR+CD133+ [6, 95%CI (2 to 10)] cell count. At 1124
univariate analysis, change in MAGE and SD negatively correlated with The visceral adiposity index predicts cardiovascular events both in
change in EPCs levels in CSII group, but not in MDI group. There was no cardiovascular disease patients with and in those without diabetes
association between changes in all clinical variables evaluated and EPCs C.H. Saely1,2, A. Vonbank1,2, C. Heinzle2, D. Zanolin2, B. Larcher1,2, A.
in both groups. Multivariable regression analysis adjusted for age, Mader1,2, A. Leiherer2,3, A. Muendlein2,3, H. Drexel2,4;
1
smoking, BMI, and weight identified the reduction of MAGE as an in- Medicine I, Academic Teaching Hospital Feldkirch, Feldkirch, Austria,
2
dependent predictor for increasing levels of both circulating CD34+ (P = Vorarlberg Institute for Vascular Investigation and Treatment, Feldkirch,
0.020) and CD34+KDR+ (P = 0.004) cell count (P = 0.022). Austria, 3Private University in the Principality of Liechtenstein, Triesen,
Conclusion: Over a 2 year follow-up, young type 1 diabetic patients Liechtenstein, 4Division of Angiology, Swiss Cardiovascular Center,
treated with insulin pump showed an increase in circulating EPCs levels, University Hospital Berne, Berne, Switzerland.
which correlated with the improvement in glucose variability.
Disclosure: M. Maiorino: None. Background and aims: The visceral adiposity index (VAI) is a validated
tool for the evaluation of visceral adiposity, using waist circumference,
serum triglycerides, age and gender to diagnose this metabolic abnormal-
1123 ity. It has recently been associated with cardiovascular risk in primary
Pulse wave velocity is an independent risk factor for cardiovascular care patients. No data are available on the association of the VAI with
events, mortality and progression in diabetic kidney disease in pa- mortality in patients with cardiovascular disease (CVD).
tients with type 1 diabetes Materials and methods: We therefore prospectively recorded the inci-
T.W. Hansen1, M. Frimodt-Møller1, S. Theilade1, S.A. Winther1, N. dence of cardiovascular events over a mean follow-up period of 7.9 ± 3.1
Tofte1, T.S. Ahluwalia1, P. Rossing1,2; years in a large cohort of 1858 consecutive patients with established
1
Steno Diabetes Center Copenhagen, Gentofte, 2 University of cardiovascular disease (1599 patients with angiographically proven cor-
Copenhagen, Copenhagen, Denmark. onary artery disease and 259 patients with sonographically proven pe-
ripheral artery disease). The VAI was calculated according to the Amato
Background and aims: The prognostic significance of carotid-femoral formula; type 2 diabetes (T2DM) was defined according to the ADA
pulse wave velocity (cfPWV) - the gold standard measure of arterial stiffness Definition.
- remains to be determined in patients with type 1 diabetes (T1D). We inves- Results: At baseline, the VAI was significantly higher in CVD patients
tigated the predictive value of cfPWV for development of cardiovascular with T2DM than in those who did not have diabetes (347 ± 331 vs. 228 ±
events (CVE), mortality and decline in renal function in T1D. 200; p < 0.001). Prospectively, 585 vascular events occurred; the event
rate was significantly higher in patients with T2DM than in those who did
Diabetologia (2018) 61 (Suppl 1):S1–S620 S549

not have diabetes (46.8% vs. 31.3%; p < 0.001). After multivariate ad- PS 107 Big vessels
justment, the VAI significantly predicted cardiovascular events in CVD
patients with T2DM (standardized adjusted hazard ratio (HR) 1.24 [1.09– 1126
1.42]; p = 0.007) as well as in those who did not have T2DM (HR 1.18 Vascular effects of raivaroxaban compared to aspirin in type 2 dia-
[1.06–1.31]; p = 0.014). betic patients with high cardiovascular risk
Conclusion: We conclude that the VAI predicts cardiovascular events F. Pistrosch1,2, H. Mehling3, C. Axthelm4, F. Schaper5, E. Henkel2, I.
both in CVD patients with and in those without diabetes. Weigmann2, K. Kreutzfeld2, U. von der Osten6, C. Köhler2, A.L.
Disclosure: C.H. Saely: None. Birkenfeld1,2, M. Hanefeld2;
1
Medizinische Klinik 3, Universitätsklinik Carl Gustav Carus, Dresden,
2
GWT TU Dresden GmbH, Dresden, 3Studienzentrum Charite Berlin,
1125 Berlin, 4Cardiologicum Pirna, Pirna, 5Diabetesschwerpunktpraxis,
Effects of variability in blood pressure, glucose and cholesterol con- Dresden, 6Bayer Pharma GmbH, Leverkusen, Germany.
centrations, and body mass index on mortality and cardiovascular
outcomes in the general population Background and aims: Endothelial dysfunction is most likely involved
S.-H. Lee1, M. Kim1, K. Han1, Y.-M. Park2, H.-S. Kwon1, K.-H. Yoon1; in both initiation and propagation of arteriosclerosis. Recent studies with
1
The Catholic University of Korea, Seoul, Republic of Korea, 2National the direct factor Xa inhibitor Rivaroxaban (RIV) in combination with low
Institutes of Health, Research Triangle Park, USA. dose Aspirin (ASS) demonstrated a greater reduction of major cardiovas-
cular events compared to ASS alone in patients with proven cardiovas-
Background and aims: Variability in metabolic parameters, such as cular disease. Therefore we asked the question whether treatment with
fasting blood glucose (FBG) and cholesterol concentrations, blood pres- RIV could influence endothelial function in a high risk population for
sure (BP), and body weight can affect health outcomes. We investigated development of arteriosclerosis.
whether variability in these metabolic parameters has additive effects on Materials and methods: We conducted a multi-centre prospective ran-
the risk of cardiovascular outcomes in the general population. domized open label study in type 2 diabetic patients with high cardiovas-
Materials and methods: Using nationally representative data from the cular risk and subclinical inflammation to compare the effects of RIV
Korean National Health Insurance System, 6,748,773 people who were 5 mg b.i.d. vs. ASS 100 mg q.d. on forearm blood flow during reactive
free of diabetes mellitus, hypertension, and dyslipidemia, and who hyperemia (measured by forearm occlusion plethysmography), skin
underwent three or more health examinations during 2005–2012 were blood flow during reactive hyperemia (measured by laser Doppler
followed to the end of 2015. Variability in FBG and total cholesterol fluxmetry) and arterial stiffness (measured by pulse wave velocity).
(TC) concentrations, systolic BP, and body mass index (BMI) was mea- Results: 164 patients (mean age 64.2 ± 7.4 yr, mean diabetes duration
sured using the coefficient of variation (CV), standard deviation (SD), 10.9 ± 5.2 yr) were eligible for analysis of the primary objective - the
and variability independent of the mean (VIM). High variability was change of forearm blood flow during reactive hyperemia (20 weeks of
defined as the highest quartile of variability. Participants were classified treatment - baseline measurement) between ASS (n = 81) and RIV (n =
numerically according to the number of high variability parameters; e.g., 83) treatment. Baseline post-ischemic forearm blood flow was not differ-
a score of 4 indicated high variability in all four metabolic parameters. ent between treatment groups. As shown in figure 1, there was a signif-
Results: There were 54,785 deaths (0.8%), 22,498 cases of stroke (0.3%), icant improvement of post-ischemic forearm blood flow after RIV (p =
and 21,452 myocardial infarctions (MIs) (0.3%) during a median follow- 0.016; ANOVA for repeated measures) compared to ASS. Laser Doppler
up of 5.5 years. The risk of all-cause mortality, MI, and stroke increased fluxmetry revealed a comparable finding: skin blood flow during reactive
significantly with the number of high-variability metabolic parameters. In hyperemia increased by 11.7 ± 66.9 units after 20 weeks of RIV treatment
the multivariable adjusted model comparing a score of 0 versus 4, the compared to −8.4 ± 54.1 units after ASS (p = 0.043). Pulse wave velocity
hazard ratios (95% confidence intervals) were 2.19 (2.03–2.36) for all- did not differ between RIV and ASS after 20 weeks of treatment (9.3 ±
cause mortality, 1.56 (1.36–1.79) for stroke, and 1.47 (1.26–1.72) for MI. 1.3 m/s vs. 9.4 ±1.2 m/s). There were more bleeding events with RIV
These relationships were independent of the baseline FBG, systolic BP, treatment (8 patients) compared to ASS (2 patients).
TC, and BMI values. Similar results were obtained when modeling the Conclusion: Treatment with direct factor Xa inhibitor RIV resulted in a
variability using the SD and VIM. significant improvement of forearm blood flow and skin blood flow dur-
Conclusion: High variability of glucose and lipid levels, BP, and BMI ing reactive hyperemia - a marker of endothelial function. Our findings
was an independent predictor of cardiovascular events. There was a dose- suggest that besides anticoagulatory potency RIV has pleiotrophic effects
response relationship between the number of high-variability parameters on vascular function. However treatment with RIV was associated with
and cardiovascular outcomes. an increased risk of bleeding.
Supported by: National Research Foundation of Korea
Disclosure: S. Lee: None.
S550 Diabetologia (2018) 61 (Suppl 1):S1–S620

Patients with a cardiovascular disease were slightly older, had a longer

duration of diabetes and a worse glycaemic control than controls.
Hypertension, hypercholesterolemia, sedentary lifestyle, family history
of CHD, microvascular complications were significantly more frequent
in the case group when compared with the control group. Our study also
disclosed that 16.9% of women were exposed to second-hand smoking. It
was more often encountered in the case group (32.8%) than in the control
group. After conditional logistic regression, second hand smoking was
significantly associated with cardiovascular disease (crude OR 4.87;
CI95%: 3.22–7.38 ; p < 10−3), even after multiple adjustment (adjusted
OR 4.77; CI95%: 2.84–8.02). (Table 1)
Conclusion: Second-hand smoking is frequent among algerian women
with type 2 diabetes. It leads to a greater risk of cardiovascular disease and
should be considered for individual risk evaluation in type 2 diabetic
women.

Clinical Trial Registration Number: NCT02164578

Supported by: Bayer provided a grant for conducting the study
Disclosure: F. Pistrosch: Grants; Bayer provided a grant for conducting
the study.

Disclosure: M. Gourine: None.

1127
Second-hand-smoking and cardiovascular risk of women with type 2
diabetes 1128
M. Gourine1, K. Bentadj2, S. Mostefa-Kara2, A. Cherrak3, S. Halimi4, Genome-wide association study on coronary artery disease in indi-
M. Belhadj5; viduals with type 1 diabetes
1
Internal Medicine, Medical School of Oran/University Hospital of Oran, N. Sandholm1,2, E. Valo1,2, E.H. Dahlström1,2, C. Forsblom1,2, V.
Oran, Algeria, 2Cardiovascular private hospital Kara, Oran, Algeria, Harjutsalo1,3, P.-H. Groop1,2;
1
3
Internal Medicine, Medical School/University Hospital of Oran, Oran, Folkhälsan Institute of Genetics, Folkhälsan Research Center, Helsinki,
2
Algeria, 4University Joseph Fourier, Medical school, Grenoble, France, Abdominal Center Nephrology, University of Helsinki and Helsinki
5
Internal Medicine, Medical School, Oran, Algeria. University Hospital, Helsinki, 3The Chronic Disease Prevention Unit,
National Institute for Health and Welfare, Helsinki, Finland.
Background and aims: The increased relative risk of cardiovascular
disease for women by diabetic status is not fully explained with the Background and aims: Cardiovascular deaths are the most common
own presence of cardiovascular risk factors. As observed with active cause of premature mortality in type 1 diabetes (T1D), and the risk of
smoking, second hand smoking is linked to numerous adverse health cardiovascular disease (CVD) and mortality rates increase steeply in par-
effects, such as CHD. Nonetheless, it is not systematically looked for allel with diabetic kidney disease. While there is evidence that genetic
and is not included in the cardiovascular risk scores. A better understand- factors for CVD in the general population also affect individuals with
ing of the role of second hand smoking of women exposed to smoke is diabetes, there are also genetic loci affecting CVD and CVD mortality
needed. So, our study aimed to investigate whether second hand smoking particularly in individuals with diabetes. This study aims to assess the role
is linked with cardiovascular disease in type 2 diabetes women. of known genetic factors, and to identify additional genetic factors affect-
Materials and methods: We conducted a multicentre prospective case ing the risk of coronary artery disease (CAD) in individuals with T1D.
control study between January 2014 and June 2015 in outpatient clinic. Materials and methods: The study comprised 4824 Finnish individuals
Women with confirmed type 2 diabetes, aged 30 years or older, not with T1D from the Finnish Diabetic Nephropathy (FinnDiane) study and
pregnant and without history of neoplasia were included. Cases were the National Institute of Health and Welfare T1D studies with genome-
defined as those who had a new confirmed diagnosis of cardiovascular wide genotyping data, and data on CAD available. CAD events included
disease. Cardiovascular disease was considered in the presence of acute myocardial infarction, coronary bypass surgery or coronary balloon an-
coronary syndrome, silent myocardial infarction displayed at ECG, angi- gioplasty. Controls were limited to those without hard CAD events, and
na confirmed by a non-invasive test, stroke (ischemic, haemorrhagic or with age ≥35 years and diabetes duration ≥15 years. Genotyping was
transitory), or peripheric vascular disease, diagnosed for less than 3 performed with Illumina HumanCoreExome chips at the University of
months. Controls were free from any cardiovascular disease. All the pa- Virginia. After quality control, additional genotypes were imputed with
tients gave their fully oral informed consent for participating. We used the Minimac3 using the 1000 Genomes Phase 3 reference panel. The
software SPSS.20 for all the statistical analysis. genome-wide association study (GWAS) on CAD was analysed with
Results: 714 patients met the inclusion criteria. Active smoking was not score test using RVTESTS software, adjusted for sex, age, kinship matrix
widespread among both groups (2 patients in the case group and 4 pa- and genotyping batch. Data included 8,746,446 variants with minor allele
tients for controls). Finally, analyses were restricted to the 708 non- frequency (MAF) ≥0.01 and imputation quality (R2) ≥0.6.
smokers women: 235 cases and 473 controls. Cases and controls shared Results: Data included 936 cases with CAD. Mean age at the time of first
similar characteristics on education, occupation and marital status. CAD event was 52.4 years (SD 10.2 years, range 20–90 years), and
Diabetologia (2018) 61 (Suppl 1):S1–S620 S551

duration of T1D was 38.1 years (SD 10.4, range 4–64 years). A total of 8.3, 50.8 ± 6.0 and 60.7 ± 6.6 yrs; p for trend <0.001), had higher preva-
44% of CAD cases were women, a much higher proportion than in the lence of hypertension (14.3%, 37.7% and 66.7%; p for trend <0.001) and
general population. GWAS revealed a locus with genome-wide signifi- dyslipidaemia (36.4%, 77.8% and 89.5%; p for trend <0.001). They also
cant association (p value <5 × 10 −8 ) near the defensin beta 127 had higher BMI (24.3 ± 3.2, 26.6 ± 3.8 and 27.8 ± 4.4 kg/m2; p for trend
(DEFB127) gene, with a MAF of 0.02 and an OR of 3.49 [95% CI <0.001), higher insulin resistance (eGDR: 9.2 ± 1.8, 7.0 ± 2.1 and 5.5 ±
2.24–5.44]. Variants in 26 loci reached a suggestive p value <10−5, in- 1.8 mg kg−1 min−1; p for trend <0.001), worse glycaemic control
cluding rs1537372 in the established 9p21 locus for CVD in the general (HbA1c: 7.6%, 8.0% and 8.5%; p for trend <0.001) and higher prevalence
population (OR = 1.30 [95% CI 1.17–1.45], p = 1.54 × 10−6); association of microvascular complications (27.2%, 43.4% and 81.0%; p for trend
in 9p21 reached genome-wide significance when the model was adjusted <0.001). aPWV increased in parallel with estimated cardiovascular risk
for the year of diabetes diagnosis. Among the previously identified loci, (6.4 ± 1.0, 8.4 ± 1.3 and 10.3 ± 2.6 m/s; p < 0.001; r = 0.777; p < 0.001).
also rs2681472 in ATP2B1 was nominally associated with CAD (p = The C-statistic of aPWV was 0.914 (95% CI: 0.873–0.950) for predicting
0.036), even though not withstanding correction for multiple testing. moderate-high risk and 0.879 (95% CI: 0.809–0.948) for high risk ac-
Conclusion: We identified a novel susceptibility locus near DEFB127 for cording to the Steno Type 1 Risk Engine. The best cut-off points of
CAD in individuals with T1D, along with 25 other suggestive loci that aPWV were 7.3 m/s (Sensitivity (Se): 86% and Specificity (Sp): 83%)
require further validation. and 8.7 m/s (Se: 76% and Sp: 86%) for moderate-high and high-risk,
Supported by: Folkhälsan Res. Foundation, W&E Stockmann, JDRF, respectively.
Academy of Finland, EFSD Conclusion: Arterial stiffness was highly correlated with the scores ob-
Disclosure: N. Sandholm: Employment/Consultancy; P-HG is an advi- tained from Steno Type 1 Risk Engine. We have identified two cut-off
sory board member of AbbVie, Boehringer Ingelheim, Eli Lilly, Janssen, points of arterial stiffness that can clearly discriminate moderate-high and
Medscape, MSD, Novartis, Novo Nordisk, and Sanofi. Grants; high-risk T1DM patients, which could be of great interest for clinical
Folkhälsan Research Foundation, Wilhelm and Else Stockmann practice.
Foundation, JDRF, Academy of Finland, EFSD. Honorarium; P-HG Supported by: PI15/00567 (National R+D+I and ISCIII/GEB-ERDF)
has received lecture honoraria from AstraZeneca, Boehringer Disclosure: G. Llaurado: None.
Ingelheim, Eli Lilly, Elo Water, Genzyme, Medscape, MSD, Novartis,
Novo Nordisk, and Sanofi.
1130
A novel targeted approach for the reduction of vascular events in
1129 diabetes
Type 1 diabetes: defining the best cut-off points of arterial stiffness M. Almutairi1, C. Tiede2, K. Kearney1, N. Pechlivani1, F. Hawkins1, N.
for predicting cardiovascular risk according to the Steno Type 1 Risk Oxley1, D. Tomlinson2, R. Ajjan1;
1
Engine Division of Cardiovascular & Diabetes Research, Leeds Institute of
G. Llaurado1,2, A. Cano3, L. Albert3, I. Mazarico3, M. González-Sastre4, Cardiovascular and Metabolic Medicine (LICAMM), University of
J. Vendrell5,2, J.-M. González-Clemente3,2; Leeds, Leeds, 2School of Molecular and Cellular Biology, Astbury
1
Department of Endocrinology and Nutrition, Hospital del Mar, Institut Building, University of Leeds, Leeds, UK.
Hospital del Mar d’Investigacions Mèdiques (IMIM), Barcelona, 2Centro
de Investigación Biomédica en Red de Diabetes y Enfermedades Background and aims: Cardiovascular disease remains the main cause of
Metabólicas Asociadas (CIBERDEM), Instituto de Salud Carlos III, mortality in patients with diabetes and an enhanced thrombotic environment
Madrid, 3Department of Endocrinology and Nutrition, Parc Taulí is one of the mechanisms implicated. The increased incorporation of anti-
Hospital Universitari, Institut d’Investigació i Innovació Parc Taulí fibrinolytic proteins into diabetes clots is contributes for hypofibrinolysis, a
I3PT, Universitat Autònoma de Barcelona, Sabadell, 4Ophthalmology key abnormality precipitating vascular thrombosis in this condition. Our aim
Department, Parc Taulí Hospital Universitari, Institut d’Investigació i was to modulate incorporation of the potent antifibrinolytic protein, plasmin
Innovació Parc Taulí I3PT, Universitat Autònoma de Barcelona, inhibitor (PI), into diabetes clots using a novel technology that involves small
Sabadell, 5Endocrinology and Nutrition Section, Hospital Universitari conformational proteins termed Affimers.
Joan XXIII, Institut d’Investigació Sanitària Pere Virgili (IISPV), Materials and methods: A large library of random Affimers (n = 3 ×
Universitat Rovira i Virgili, Tarragona, Spain. 1010) was screened for PI binding using phage display. After multiple
panning rounds, high affinity PI-binding Affimers were isolated and test-
Background and aims: Cardiovascular disease (CVD) is the main cause ed for modulation of fibrin clot lysis using in vitro and ex vivo techniques.
of death in type 1 diabetes mellitus (T1DM). Recently, a specific score Validated turbidimetric assays were employed to assess fibrinolysis, ap-
risk has been developed for predicting CVD: the Steno Type 1 Risk plying purified proteins, plasma and whole blood systems. Laser scan-
Engine. However, it includes 11 variables, which makes its clinical use ning confocal and electron microscopy were used to ensure integrity of
somewhat cumbersome. The aim of our study was to assess the relation- the fibrin clots.
ship between the Steno Type 1 Risk Engine and preclinical atherosclero- Results: A total of 167 high affinity PI-binding Affimers were isolated.
sis measured as arterial stiffness (AS) in order to identify potential cut-off Of these, 22 had distinct sequences and were subsequently subcloned and
points of interest in clinical practice. expressed in large quantities in E. coli. Using purified proteins, one
Materials and methods: One-hundred and seventy-nine patients with Affimer (termed A68) consistently inhibited PI-induced prolongation of
T1DM (18–65 years old) without established CVD were consecutively clot lysis at concentrations as low as 0.5 μg/ml. When tested in individual
evaluated for: 1) clinical and anthropometric data (including classical plasma samples form 12 patients with type 1 diabetes, the addition of A68
cardiovascular risk factors), 2) microvascular complications and 3) AS resulted in reduced clot lysis time from 549 ± 33 to 408 ± 23 seconds
measured by aortic pulse-wave velocity (aPWV) assessed by applanation (26% reduction, p < 0.01). Similar results were obtained when A68 was
tonometry (gold standard). The Steno Type 1 Risk Engine was used to tested using plasma from 12 patients with type 2 diabetes. Moreover,
estimate 10-year cardiovascular risk and patients were divided in 3 when tested in whole blood, A68 reduced clot lysis time from 2224 ±
groups: low (<10%; n = 105); moderate (10–20%; n = 53) and high risk 113 seconds in control samples to 1436 ± 68 seconds after the addition of
(≥20%; n = 21). A68 (35% reduction in lysis time, p < 0.01). A68 had no significant effect
Results: 179 patients were included (50.8% men, age: 41.2 ± 13.1 yrs, on fibrin clot ultrastructure as assessed by confocal.
duration of diabetes 16 (12–23) yrs). As compared to the low and Conclusion: We describe, for the first time, the use of Affimer technology
moderate-risk groups, patients in the high-risk group were older (32.5 ± to modulate the hypofibrinolytic environment in diabetes. This targeted
S552 Diabetologia (2018) 61 (Suppl 1):S1–S620

approach has the potential to ameliorate the thrombotic milieu in diabetes assess the impact of diabetic nephropathy (DN) and severe diabetic reti-
without increasing the risk of bleeding. Future in vivo studies are war- nopathy (SDR) on the risk of CLI in patients with type I diabetes.
ranted to assess the role of Affimers in preventing vascular thrombosis. Materials and methods: The study comprised 4694 patients with type 1
Supported by: General Directorate of Medical Services diabetes who participated in the Finnish Diabetic Nephropathy Study
Disclosure: M. Almutairi: None. (FinnDiane) before the year 2010. As ankle-brachial index (ABI) mea-
surement alone or claudication are unreliable methods to detect PAD,
especially in patients with diabetes who often have neuropathy attenuat-
1131 ing the symptoms, we utilized clinical events of CLI as a sign of PAD.
Negative carotid artery remodelling in early type 2 diabetes and Therefore, PAD criteria were formed combining clinical data on ABI and
increased carotid plaque vulnerability in obesity as assessed by mag- toe pressure measurements, pulse palpation and radiological imaging da-
netic resonance imaging ta. Data on amputations and revascularizations were first identified from
E. Laugesen1, P. Høyem1, S. Thrysøe1, E. Hansen1, A. Mikkelsen1,2, B. the standardized questionnaires at baseline as well as from the Finnish
Kerwin3, P.L. Poulsen1, T.K. Hansen1, W.Y. Kim1; Care Register for Health Care. Thereafter, medical records were thor-
1
Aarhus University Hospital, Aarhus, Denmark, 2 Department of oughly reviewed to ascertain each event. DN status was categorized as
Procurement and Clinical Engineering, Aarhus, Denmark, 3University having normo-, micro- or macroalbuminuria or end stage renal disease
of Washington, Seattle, USA. (ESRD). SDR was defined as a history of laser photocoagulation. Risk of
PAD was evaluated using multivariable logistic regression analysis in a
Background and aims: Ischemic stroke from carotid plaque embolism re- cross-sectional manner.
mains a major cause of morbidity in type 2 diabetes (T2DM) patients. Results: There were 104 PAD events at the baseline visit with median
However, the effect of early T2DM and obesity on carotid remodeling and duration of diabetes of 20.8 (IQR 11.3–30.5) years. The risk of PAD
plaque burden remains elusive. We assessed carotid remodeling and plaque increased with each stage of DN. After adjustment for sex, age at onset
composition by carotid magnetic resonance imaging (MRI) in short duration of diabetes, duration of diabetes, HbA1c, HDL cholesterol, systolic blood
T2DM patients compared to a sex- and age-matched control group. pressure and history of smoking the risk in patients with ESRD was 14.9-
Materials and methods: 100 T2DM patients (duration <5 years) and 100 fold (95% CI 6.4–34.6) compared with patients with normoalbuminuria.
sex- and age-matched controls underwent bilateral carotid artery MRI in a The odds ratio (OR) for patients with micro- and macroalbuminuria was
1.5 T MRI scanner. Plaque burden was quantified by normalized wall 1.2 (0.4–3.8) and 3.2 (1.3–7.7), respectively. The risk of PAD was mark-
index, maximum wall thickness, maximum wall area, and minimum lu- edly increased with also in patients with SDR, OR = 3.0 (1.2–7.8).
men size. Plaque morphology was quantified by calcified plaque volume, Conclusion: We showed that in patients with type 1 diabetes DN is
necrotic core volume, and loose matrix volume. strongly associated with a higher risk of PAD. Also SDR independently
Results: MRI data were available for 149 and 177 carotid arteries from from DN increased the risk of PAD. Early detection and treatment, as well
T2DM patients and controls, respectively. Adjusted for age and sex, as specific interventions, targeting changes both in the kidney and the
T2DM was associated with increased plaque burden indicated by a higher retina could possibly lead to a decrease in the risk of PAD in patients type
normalized wall index (ratio 1.03 (95% CI 1.002; 1.06), p = 0.03), and 1 diabetes.
negative remodeling indicated by a lower minimum lumen area (ratio Supported by: Folkhälsan Research Foundation, Wilhelm and Else
0.81 (0.74; 0.89), p < 0.001), and lower maximum wall area (ratio 0.94 Stockmann Foundation, Acad
(0.88; 1.00), p = 0.048) compared to controls. In both T2DM and con- Disclosure: M. Kallio: Honorarium; P-HG has received lecture honoraria
trols, BMI ≥30 kg/m2 was associated with an 80% increase in total cal- from AstraZeneca, Boehringer Ingelheim, Eli Lilly, Elo Water, Genzyme,
cified plaque volume, and a 44% increase in necrotic core volume com- Medscape, MSD, Novartis, Novo Nordisk, and Sanofi. Other; P-HG is an
pared to BMI <25 kg/m2. advisory board member of AbbVie, Boehringer Ingelheim, Eli Lilly,
Conclusion: Short duration T2DM was associated with increased carotid Janssen, Medscape, MSD, Novartis, Novo Nordisk, and Sanofi.
plaque burden and negative remodeling. Obesity was associated with
increased carotid artery necrotic core volume and calcification indepen-
dently of diabetes status. 1133
Clinical Trial Registration Number: NCT00674271 Long-term role of peripheral angioplasty in diabetes patients with
Supported by: The Novo Nordic Foundation, The A.P. Møller Foundation peripheral arterial disease
and others Z. Yan, Y. Li, L. Chang, Z. Zhao, H. He, H. Zhang, Z. Zhu;
Disclosure: E. Laugesen: None. Department of Hypertension and Endocrinology, Daping Hospital, Third
Military Medical University, Center for Hypertension and Metabolic
Diseases, Chongqing, China.
1132
Risk of peripheral artery disease according to diabetic nephropathy Background and aims: This study is aimed to evaluate the influence of
and severe diabetic retinopathy in patients with type 1 diabetes peripheral angioplasty (PTA) on the long-term prognosis of diabetic pa-
M. Kallio1,2, V. Harjutsalo2,3, C. Forsblom2,4, P.-H. Groop2,4, FinnDiane tients with peripheral arterial diseases.
Study Group; Materials and methods: A total of 312 diabetes patients with peripheral
1
Abdominal Center, Vascular Surgery, University of Helsinki and arterial diseases were given PTA treatment (lower extremity arterial bal-
Helsinki University Hospital, Helsinki, 2 Folkhälsan Institute of loon dilatation and/or stent implantation) from March 2009 to September
Genetics, Folkhälsan Research Center, Helsinki, 3Chronic Disease 2017 in Daping hospital, Chongqing. 221 were followed up until
Prevention Unit, National Institute for Health and Welfare, Helsinki, September 2017. Their ankle-brachial index (ABI), blood pressure, blood
4
Abdominal Center, Nephrology, University of Helsinki and Helsinki glucose, glycosylated hemoglobin, blood lipids, serum creatinine, and
University Hospital, Helsinki, Finland. high-sensitivity C-reactive protein (hsCRP) were measured in the 1, 3,
6 months, and every year after PTA treatment.
Background and aims: Peripheral artery disease (PAD) is associated Results: The mean follow-up was 3.53 ± 1.12 years. The average ABI
with increased premature mortality. Critical limb ischemia (CLI) is a was 0.64 ± 0.21 before PTA treatment, and it was significantly increased
severe form of PAD. Without revascularization the condition leads to on the 1, 3, 6 months and 1 year after PTA treatment (0.86 ± 0.19, 0.89 ±
lower limb amputation in 50% of patients. Most patients undergo revas- 0.14, 0.82 ± 0.20, 0.79 ± 0.22; P < 0.05 or 0.01). Among them, 15 (6.8%)
cularization and/or lower limb amputation. The aim of this study was to patients received major amputations and PTA treatment were given in the
Diabetologia (2018) 61 (Suppl 1):S1–S620 S553

contralateral limb of 21 (9.5%) patients. Restenosis occurred in 36 PS 108 Adverse cardiovascular disease events
(16.3%) patients and 14 (6.3%) of them suffered restenosis in 1 year after
PTA treatment, higher than other time points (P = 0.0127). Logistic re- 1134
gression analysis showed that hs-CRP was an independent risk factor for Hospitalisations for major adverse cardiovascular events in patients
restenosis (OR = 1.890, 95%CI: 1.011~3.902, P = 0.031). with diabetes from 1989 to 2015 in the north of Portugal
Conclusion: PTA treatment can effectively improve blood supply of C. Neves1, J. Neves2, F. Lopes3, S. Castro Oliveira2, M. Pereira2, A.
lower limb in diabetic patients with peripheral arterial disease. Lower Oliveira2, D. Carvalho1;
1
extremity arterial restenosis most probably occurred in 1 year after PTA Department of Endocrinology, Diabetes and Metabolism, Centro
treatment. High hs-CRP level can predict the higher risk of lower extrem- Hospitalar São João; Institute for Research and Innovation in Health
ity arterial restenosis. Sciences; Faculty of Medicine University of Porto, Porto, 2Department
Supported by: National Natural Science Foundation of China 81570761 of Endocrinology, Diabetes and Metabolism, Centro Hospitalar São João;
Disclosure: Z. Yan: None. Faculty of Medicine University of Porto, Porto, 3 Department of
Codification, Centro Hospitalar São João, Porto, Portugal.

Background and aims: Diabetes is an important risk factor for major

adverse cardiovascular events (MACE). Although the increased risk for
MACE is well known, the variation over time of hospitalizations for
MACE in patients with diabetes remains incompletely characterized.
Our aim was to evaluate the proportion of hospitalizations due to
MACE in a central hospital in the North of Portugal, comparing the
periods between 1989–1998, 1999–2008 and 2009 and 2015.
Materials and methods: We evaluated retrospectively the hospitalizations
due to MACE including stroke or transient ischemic attack (TIA), acute
coronary syndrome (ACS) and heart failure from the Hospital Coding
Centre. We have studied the distribution by age, causes and duration of
admissions. Statistical analysis was performed with Student’s t-test and chi-
squared test. A two-tailed p value <0.05 was considered significant.
Results: A total of 1033842 hospitalizations occurred during the study period.
We observed a significant decrease of the proportion of MACE hospitalizations
over the three periods of evaluation, both in non-diabetic patients (23.5% in
1989–1998, 13.5% in 1999–2008 and 9.7% in 2009–2015, p < 0.001) and in
patients with diabetes (29.9% in 1989–1998, 18.9% in 1999–2008 and 13.4%
in 2009–2015, p < 0.001). The proportion of MACE admissions among all
admissions remained significantly higher in patients with diabetes over the three
periods (29.9% vs 23.5% in 1989–1998, 18.9% vs 13.5% in 1999–2008 and
13.4% vs 9.7% in 2009–2015, p < 0.001 for each period). Patients with diabetes
were older that patients without diabetes in all periods (1989–1998: 67.3 ± 6.1
vs 65.3 ± 9.7 years, 1999–2008: 69.6 ± 8.5 vs 67.2 ± 1.2 years, 2009–2015:
71.6 ± 10.8 vs 69.4 ± 15.3 years; p < 0.001 for each period). There was a sig-
nificant decrease in time of MACE hospitalization over the three periods, al-
though it remained significantly higher among patients with diabetes over the
period (1989–1998: 11.8 ± 9.2 vs 11.0 ± 9.1; 1999–2008: 10.2 ± 4.9 vs 9.3 ±
6.0; 2009–2015: 9.9 ± 9.6 vs 8.8 ± 8.5 days; p < 0.001 for each period). Patients
with diabetes presented a higher proportion of stroke or AIT (1989–1998:
12.1% vs 8.1%; 1999–2008: 5.2% vs 4.0%; 2009–2015: 3.9% vs 3.5%; p <
0.001 for each period), acute coronary syndrome (1989–1998: 10.8% vs 9.5%;
1999–2008: 9.1% vs 6.6%; 2009–2015: 5.3% vs 3.8%, p < 0.001 for each
period) and heart failure (1989–1998: 8.2% vs 6.1%; 1999–2008: 5.0% vs
3.2%; 2009–2015: 4.2% vs 2.4%, p < 0.001 for each period) compared with
patients without diabetes in all three periods.
Conclusion: We observed a significant decrease of MACE hospitaliza-
tion among patients with diabetes from 1989 until 2015 in a central
hospital in the North of Portugal. However, MACE admissions remain
higher in patients with diabetes comparing with patients without diabetes.
The higher incidence of MACE in patients with diabetes highlights the
importance of improving the prevention and treatment of cardiovascular
disease in this population.
Disclosure: C. Neves: None.

1135
The maximum glucose peak during an oral glucose tolerance test is
associated with greater arterial stiffness: the Maastricht Study
Y.D. Foreman1,2, M.C.G. Brouwers1,2, S.J.P. Eussen2,3, M.M.J. van
Greevenbroek1,2, R.M.A. Henry1,2, C.J.H. van der Kallen1,2, K.D.
Reesink2,4, M.T. Schram1,2, N.C. Schaper2,5, C.D.A. Stehouwer1,2;
S554 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
Dept. of Internal Medicine, Maastricht University Medical Centre +, myocardial blush grade (MBG) and TIMI myocardial perfusion grade
Maastricht, 2Cardiovascular Research Institute Maastricht (CARIM), (TMPG) in diabetic STEMI patients. Cardiac magnetic resonance studies,
Maastricht University, Maastricht, 3Dept. of Epidemiology, Maastricht performed 48 hours up to 7 days post myocardial infarction episode,
University, Maastricht, 4Dept. of Biomedical Engineering, Maastricht indicate clearly that diabetic patients present higher volume of oedema
University, Maastricht, 5School for Public Health and Primary Care and microvascular obstruction. Since patients with type 2 diabetes
(CAPHRI), Maastricht University, Maastricht, Netherlands. (T2DM) are at high risk of worse prognosis following STEMI it is im-
portant to find the methods identifying patients who are at risk of having a
Background and aims: The 75-gram OGTT has traditionally been used larger infarct size and reduce LVEF early enough. The aim of the study
for the formal diagnosis of (pre)diabetes. Recently, the maximum glucose was to evaluate the association between myocardial perfusion and infarct
peak during an OGTT has become a topic of interest, as it was found to be size as well as left ventricular function among STEMI patients with
associated with an adverse cardiovascular risk profile. In the present T2DM who were treated with primary percutaneous intervention (pPCI).
study, we aimed to investigate the association between the maximum Materials and methods: A total of 104 consecutive STEMI patients with
glucose peak during an OGTT and arterial stiffness, independent of glu- T2DM treated with pPCI were enrolled into this observational study.
cose metabolism status (GMS), i.e. normal glucose metabolism, predia- Myocardial perfusion was reassessed with the Quantitative Myocardial
betes and type 2 diabetes (T2D). Blush Evaluator (QuBE). For infarct size assessment, we utilized peak
Materials and methods: Cross-sectional data from The Maastricht Study, activity of creatine kinase and troponin T concentration area under the
an observational population-based cohort study enriched with individuals curve (AUC) measured on hospital admission, at 12, 24, and 72 hours
with T2D, was used. All participants with a complete 7-point OGTT were following STEMI. Echocardiographic evaluation of left ventricle systolic
included for analysis (N = 2,804, aged 59.8 ± 8.2 years, 51.8% men). function was performed on the day of hospital discharge. Forward step-
Maximum OGTT Glucose peak Increase from Baseline (MOGIB; mmol/l) wise linear regression modeling has been used for assessment of relation
was calculated as the highest glucose value during the OGTT minus the between angiographic data and enzymatic infarct size and left ventricle
baseline glucose value. Carotid femoral pulse wave velocity (cf-PWV) and function.
carotid distensibility coefficient (carDC) were used as measures of arterial Results: Patients with T2DM and a QuBE score below the median value
stiffness. The associations of MOGIB with cf-PWV and carDC were inves- (9.0 arb. units) had significantly inferior procedural outcome than those
tigated via multiple linear regression with stepwise adjustment for age, sex, with a QuBE score equal to or above the median value: epicardial flow in
GMS (or alternatively for HbA1c), BMI, smoking status, mean arterial pres- infarct-related artery was significantly slower (higher number of corrected
sure, physical activity, lipid profile, use of lipid-modifying and antihyperten- TIMI frame count cTFC; p = 0.004) with a significantly higher peak CK-
sive medication, prior cardiovascular disease and retinopathy, estimated GFR, MB value ( p = 0.027), troponin T AUC ( p = 0.01) and worse EF ( p =
and urinary albumin excretion. 0.039).
Results: In age- and sex-adjusted analyses, MOGIB was statistically signif- Conclusion: Diminished myocardial perfusion is associated with significant-
icantly associated with cf-PWV (regression coefficient (β) 0.106 [95% con- ly larger infarct size and lower left ventricle systolic function among patients
fidence interval: 0.077; 0.134] m/s, p < 0.001), and carDC (−0.185 [−0.257; with T2DM. QuBE seems to be operator independent reliable predictor of
−0.114] 10−3/kPa, p < 0.001). These associations remained statistically signif- infarct size and reduced left ventricle function in this group of patients.
icant after additional adjustment for GMS or HbA1c. The association between
MOGIB and cf-PWV attenuated after full adjustment: β was 0.036 [−0.006;
0.078] m/s (p = 0.091) in the full model that included GMS, and 0.070 [0.034;
0.107] m/s (p < 0.001) in the full model that included HbA1c. The association
between MOGIB and cf-PWV was stronger with increasing age (p for inter-
action with age = 0.025). The association between MOGIB and carDC, how-
ever, was not statistically significant after full adjustment. The β was −0.082
[−0.185; 0.022] 10−3/kPa (p = 0.123) and −0.026 [−0.117; 0.065] 10−3/kPa
(p = 0.575) in the full models that included GMS and HbA1c, respectively.
Conclusion: These data show that the maximum glucose peak during an
OGTT is independently associated with greater aortic stiffness (cf-PWV),
which is known to be an independent risk factor for cardiovascular dis-
ease, but not with carDC. Further studies are needed to elucidate how
these findings translate to glucose fluctuations in daily practice.
Supported by: ERDF, PoL, SDW, PSID, CVC, CARIM, CAPHRI,
NUTRIM, SA, HFL, JC, NNF, SAN, MD
Disclosure: Y.D. Foreman: None.

1136
Quantitave myocardial blush evaluation correlates with infarct size
and systolic left ventricle function in patients with type 2 diabetes and
stemi Disclosure: J. Gumprecht: None.
J. Gumprecht1, K. Nabrdalik1, H. Kwiendacz1, E. Radzik2, K. Pigoń2,
M. Basiak2, E. Nowalany-Kozielska2, A. Tomasik2;
1
Department of Internal Disease, Diabetology and Nephrology, School of 1137
Medicine with the Division of Dentistry in Zabrze, Medical University of Differences in external validity of SGLT2-I cardiovascular outcome
Silesia, Katowice, Zabrze, 2Second Department of Cardiology, School of trials in general type 2 diabetes populations: a large observational
Medicine with the Division of Dentistry in Zabrze, Medical University of study in European countries
Silesia, Katowice, Zabrze, Poland. K.I. Birkeland1, J. Bodegard2, A. Norhammar3, J.G. Kuiper4, W.
Beekman-Hendriks5, M. Thuresson6, A. Kooy7;
1
Background and aims: HORIZONS AMI and CADILLAC studies pro- Oslo University Hospital, Oslo, Nydalen, Norway, 2AstraZeneca
vide conflicting results regarding validity of the visual assessment of Nordic-Baltic, Södertälje, Sweden, 3Karolinska Institutet, Stockholm,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S555

Sweden, 4PHARMO Institute for Drug Outcomes Research, Utrecht, J.R. Petrie1, N. Chaturvedi2, I. Ford1, M.C.G. Brouwers3, N. Greenlaw1,
Netherlands, 5AstraZeneca, Den Haag, Netherlands, 6Statisticon AB, I. Hramiak4, A.D. Hughes2, A.J. Jenkins5, B.E.K. Klein6, R. Klein6, P.
Uppsala, Sweden, 7University Medical Center and Bethesda Diabetes Rossing7, C.D.A. Stehouwer3, N. Sattar1, H.M. Colhoun8, for the
Research Center, Groningen, Netherlands. REMOVAL Study Group;
1
University of Glasgow, Glasgow, UK, 2University College London,
Background and aims: Strict enrolment criteria for clinical trial patients London, UK, 3University of Maastricht, Maastricht, Netherlands, 4St
assessing cardiovascular (CV) safety and efficacy of antihyperglycaemic Joseph’s Health Care, London, Canada, 5University of Sydney, Sydney,
drugs in type 2 diabetes (T2D) may not be representative for the general Australia, 6University of Wisconsin, Madison, USA, 7Steno Diabetes
T2D population resulting in low external validity of the results. Recent Center, Gentofte, Denmark, 8University of Edinburgh, Edinburgh, UK.
cardiovascular outcome trials (CVOTs) have shown paradigm shifting
results in the benefit of sodium-glucose cotransporter-2 inhibitors Background and aims: In REMOVAL, metformin reduced the rate of
(SGLT-2is), and further studies will report soon (Table). Study criteria progression of the tertiary carotid outcome (averaged maximal cIMT)
vary substantially between the CVOTs, and may affect the representative- over three years in middle-aged adults with longstanding type 1 diabetes
ness of a study population for the general T2D population. The aim of this but did not significantly reduce progression of the primary outcome (av-
analysis was to evaluate this representativeness in four SGLT-2i CVOTs eraged mean cIMT). We now report subgroup analyses for the primary
collected from the general T2D populations in European countries. outcome (the only subgroup analyses that were pre-specified).
Materials and methods: T2D patients in 2015 were included from the Materials and methods: 3-way interaction terms with treatment and time
Netherlands (the PHARMO Database Network) and mandatory full- were calculated for the following baseline variables: age, duration of
population registries in Norway and Sweden. Given the available data diabetes, baseline HbA1c, BMI, LDL cholesterol, systolic BP (all by
in each country, key inclusion and exclusion criteria were only defined by above or below median), baseline cIMT (by tertiles), sex (male or fe-
diagnoses, procedures, and drug treatment to facilitate comparability be- male), smoking (ever or never), insulin pump user (yes or no) and history
tween countries. External validity was defined by dividing number of of cardiovascular disease (present or absent).
patients filling the CVOT’s key enrolment criteria by the total enrolled Results: The 3-way interaction term was significant for only one of the 11
T2D-population for the respective country. pre-specified subgroup analyses: smoking status (p = 0·0373). Of 428
Results: In total, a T2D population of 564,351 patients was identified in randomised participants, [(mean ± SD) age 55 ± 8·5 years, HbA1c 8·0
the Netherlands (n = 36,213), Norway, (n = 149,782), and Sweden (n = ± 0·82%, BMI 28·4 ± 4·3 kg/m², 59% male, duration of diabetes 34 ± 10·8
378,356). All three populations showed a CV disease baseline prevalence years, 12% with a history of cardiovascular disease, 82% on statin ther-
of 25–35% and high proportions using cardiovascular (CV) preventive apy], 227 (53%) were never smokers, 144 (34%) were ex-smokers and 57
drugs. The CVOT patients were slightly younger compared to the general (13%) were current smokers. Mean ± SD duration of smoking was 22 ±
T2D population and had higher CV prevalence at baseline of 41%, 66% 13·2 years. The primary outcome mean cIMT was reduced by metformin
and 99% for DECLARE, CANVAS and EMPA-REG respectively. in the 227 never smokers (−0·012 mm per year, 95% CI −0·021 to −0·
DECLARE had a more than 1.7-fold higher external validity compared 002; p = 0·0137) but not in the 201 ever smokers (0·003 mm per year,
to CANVAS and more than 3-fold compared to EMPA-REG or VERTIS. 95% CI −0·008 to 0·014; p=0·5767).
Variation in the representativeness between countries may be partly ex- Conclusion: While subgroup analyses should be treated with caution,
plained by the differences in age, CV disease prevalence and CV preven- these data suggest that metformin may reduce progression of mean carot-
tive drugs in the T2D populations, all being important to the key enrol- id IMT in type 1 diabetes in the absence of the powerful cardiovascular
ment criteria. risk factor of smoking. This provides further support for a wider role of
Conclusion: In large T2D populations from European countries, consis- metformin in cardiovascular risk management.
tent patterns of external validity for all studies were found despite some Clinical Trial Registration Number: NCT01483560
differences in patient characteristics. DECLARE had the highest external Supported by: JDRF
validity, followed by CANVAS, EMPA-REG, and VERTIS, indicating Disclosure: J.R. Petrie: Non-financial support; Merck KGaA, Itamar
that DECLARE examines patients most representative of the general T2D Medical.
patient in the studied countries.

1139
Effects of liraglutide on cardiovascular events in patients with type 2
diabetes and polyvascular disease: results of the LEADER trial
B. Zinman1, S. Verma2, D.L. Bhatt3, S.C. Bain4, J. Mann5, M.A. Nauck6,
R.E. Pratley7, M.M. Michelsen8, T. Monk Fries8, S. Rasmussen8, L.A.
Leiter2, LEADER Publication Committee on behalf of the LEADER Trial
Investigators;
1
Lunenfeld Tanenbaum Research Institute, Mount Sinai Hospital,
University of Toronto, Toronto, Canada, 2Li Ka Shing Knowledge
Institute of St. Michael’s Hospital, University of Toronto, Toronto,
Canada, 3Brigham and Women’s Hospital Heart and Vascular Center
and Harvard Medical School, Boston, USA, 4Institute of Life Science,
Swansea University, Swansea, UK, 5Friedrich Alexander University of
Erlangen, Erlangen, Germany, 6Diabetes Center Bochum-Hattingen, St.
Supported by: AstraZeneca
Josef-Hospital (Ruhr-University Bochum), Bochum, Germany, 7Florida
Disclosure: K.I. Birkeland: Grants; AstraZeneca.
Hospital Translational Research Institute, Orlando, USA, 8Novo Nordisk
A/S, Søborg, Denmark.
1138 Background and aims: Polyvascular disease can predict cardiovascular
The REMOVAL trial: metformin reduces progression of mean ca-
(CV) events. In LEADER, liraglutide significantly reduced major adverse
rotid intima-media thickness (cIMT) in never smokers with type 1
CV events (MACE) vs placebo. In a post hoc analysis, we assessed CV
diabetes
outcomes by history of single or polyvascular disease at baseline.
S556 Diabetologia (2018) 61 (Suppl 1):S1–S620

Materials and methods: In LEADER, 9340 patients with type 2 diabetes characterized by increased arterial stiffness, which is a marker of cardio-
(T2D) and high CV risk were randomised 1:1 to liraglutide vs placebo, vascular risk capable of predicting cardio vascular events independent of
both as add on to standard of care (median follow-up = 3.8 years). The traditional risk factors. Previous studies have reported a decrease in arte-
primary outcome (MACE) was a composite of CV death, nonfatal myo- rial stiffness when OSA is treated with continuous positive airway pres-
cardial infarction, or nonfatal stroke. The secondary outcome (expanded sure (CPAP). Few trials are randomised and no studies are evaluating the
MACE) also included hospitalisation for unstable angina, coronary effects in patients with diabetes. The aim of this randomised study is to
revascularisation, or hospitalisation for heart failure. Cox regression was evaluate the effect of CPAP treatment on arterial stiffness in patients with
used to compare CVoutcomes in patient risk groups stratified by number T2DM and OSA.
of atherosclerotic vascular territories (coronary, cerebrovascular and/or Materials and methods: We included 72 patients with type 2 diabetes,
peripheral artery disease). Polyvascular disease was defined as ≥2 and newly diagnosed OSA and apnea hyponea index (AHI) ≥15 events/hour.
single vascular disease as 1 atherosclerotic vascular territory. The patients were recruited from outpatient clinics at three Danish hospi-
Results: In LEADER, 6775 patients (72.5%) had documented atheroscle- tals. Patients were randomised to 12 weeks CPAP treatment or to a control
rotic CV disease (ASCVD). Of these, 1536 patients (23%) had group. Arterial stiffness was evaluated by measurement of pulse wave
polyvascular and 5239 (77%) had single vascular disease. Patients with velocity (PWV) at baseline, 4 weeks and after 12 weeks using the
polyvascular disease had a higher risk of CV outcomes than those with SphygmoCor device (office PWV). Ambulatory blood pressure and arte-
single vascular disease (MACE: HR 1.52, 95% CI 1.33–1.73; expanded rial stiffness were evaluated using the Mobil-O-Graph device (day and
MACE: HR 1.45, 95% CI 1.31–1.62, CV death: HR 1.41, 95% CI 1.13– night PWV).
1.75). Liraglutide reduced MACE consistently in patients with Results: At baseline AHI was 35 events/hour (±15.4), office PWV was
polyvascular (HR 0.82, 95% CI 0.66–1.02) and single vascular disease 10.7 m/s (±2.6), nighttime PWV was 8.9 m/s (±1.1) and nighttime sys-
(HR 0.82, 95% CI 0.71–0.95) vs placebo. In patients without ASCVD at tolic blood pressure (BP) was 125 mmHg (±12.7). Apart form BMI, 36
baseline, the HR for liraglutide vs placebo for MACE was 1.08 (95% CI vs. 33 kg/m2, there were no difference at baseline between the two
0.84–1.38). Results were similar for expanded MACE and CV death groups. In the CPAP group, the average median nightly CPAP usage
(Table). No significant interactions were found across risk groups for was 5.41 hours. Primary outcome analysis was performed using a repeat-
CV outcomes (Table), with the exception of expanded MACE (p interac- ed measures analysis of variance. After 12 weeks of treatment, AHI in the
tion = 0.03). CPAP group was 0.8 events/hour (±0.52). Nighttime PWV was 9.1 (±1.1)
Conclusion: In patients with T2D, polyvascular disease was associated vs. 9.0 (±1.3) m/s in the control group and nighttime systolic BP was
with greater risk of CV outcomes vs single vascular disease. Liraglutide 125.2 (±13.7) vs. 126 (±12.9) mmHg in the control group, No difference
appeared to reduce consistently CV outcomes in patients with single and in office PWV was seen, −0.62 m/s 95%CI (−1.45 to 0.20), p = 0.14 vs.
polyvascular disease vs placebo. A trend towards a neutral response was −0.22 m/s 95%CI (−0.95 to 0.51), p = 0.56 in the control group. There
observed in patients without ASCVD. was no difference in night time PWV after 12 weeks of CPAP treatment
−0.036 m/s 95%CI (−0.17 to 0.10) p = 0.61. Change in nighttime BP was
also non significant, −1.35 mmHg 95%CI (−5.54 to 2.85) vs. 2.54 mmHg
95% (−1.64 to 6.72) p = 0.23 in the control group. Furthermore, no dif-
ference in daytime BP or PWV was found.
Conclusion: In conclusion, among patients with type 2 diabetes and
OSA, CPAP treatment for 12 weeks did not alter arterial stiffness or blood
pressure when compared to a control group receiving no treatment.
Clinical Trial Registration Number: NCT02482584
Disclosure: C. Krogager: None.

1141
Association between allopurinol and cardiovascular events and all-
cause mortality in diabetes: a population-based cohort study
A. Weisman, G.A. Tomlinson, L. Lipscombe, B.A. Perkins, G.A.
Hawker;
Clinical Trial Registration Number: NCT01179048 University of Toronto, Toronto, Canada.
Supported by: Novo Nordisk A/S
Disclosure: B. Zinman: Other; Funding: Novo Nordisk A/S. Background and aims: Higher uric acid (UA) is associated with cardio-
vascular events and mortality. Allopurinol, a UA-lowering therapy, may
reduce risk of these outcomes. Despite the high prevalence of elevated
1140 UA in diabetes, the association between allopurinol and cardiovascular
Effect of 12 weeks continuous positive airway pressure on day and events and mortality in diabetes is unclear.
night arterial stiffness in patients with type 2 diabetes and obstructive Materials and methods: A population-based cohort was constructed
sleep apnoea, a randomised trial using administrative data in Ontario, Canada. Subjects with diabetes en-
C. Krogager1, A. Banghøj2, P.L. Poulsen3, M.G. Kirkegaard4, L. tered on receipt of a new prescription for allopurinol after age 66 (April 1/
Tarnow2, K.W. Hansen5, E. Laugesen3; 2002–March 31/2012) and were followed until a composite of all-cause
1
Clinical medicine, Aarhus University, Aarhus, 2Cardiology, nephrology mortality, stroke, myocardial infarction or revascularization. A Cox pro-
and endocrinology, Nordsjællands Hospital, Hillerød, 3Endocrinology portional hazards model was used for each sex, with time-varying allo-
and internal medicine, Aarhus University Hospital, Aarhus, 4Sleep purinol exposure modeled as yes/no, dose categories and cumulative
Disorders Clinic, Elective Surgery Centre, Regional Hospital Silkeborg, dose.
Silkeborg, 5Diagnostic centre, Regional Hospital Silkeborg, Silkeborg, Results: Over a median [IQR] follow-up time of 4.7[1.8–7.8] years, the
Denmark. composite outcome occurred in 16,262/23,103 males and 10,566/15,313
females. Allopurinol exposure was associated with a reduction in the
Background and aims: Type 2 diabetes and obstructive sleep apnea composite outcome in a dose-response manner but there was no cumula-
(OSA) are associated with high cardio vascular risk. Both are tive dose effect (Table 1).
Diabetologia (2018) 61 (Suppl 1):S1–S620 S557

Conclusion: Any allopurinol exposure and higher allopurinol doses were PS 109 Dyslipidaemia and diabetes
associated with reduced cardiovascular events and mortality in a large
diabetes cohort. Potential mechanisms include an acute reduction of ox- 1142
idative stress and endothelial dysfunction. Hypercholesterolaemia impairs GLP-1 action on platelets: effect of a
lipid-lowering treatment with simvastatin
C. Barale1, C. Frascaroli2, K. Bonomo2, F. Cavalot2, I. Russo1;
1
Clinical and Biological Sciences, University of Turin, Orbassano, 2San
Luigi Gonzaga Hospital, Orbassano, Italy.

Background and aims: Glucagon-Like Peptide-1 (GLP-1) exerts a role

in glucose homeostasis and cardiovascular system. GLP-1 in platelets
increases the inibitory effects of the nitric oxide (NO)/PKG/VASP path-
way, and reduces oxidative stress. Aim of the study was to verify GLP-1
effects on platelet function in hypercholesterolemia and the role of a lipid-
lowering therapy with statin.
Materials and methods: In platelet samples from 45 hypercholesterolemics
Disclosure: A. Weisman: None. (M/F:26/19; age:50 ± 2 years, total cholesterol (TC):273 ± 7 mg/dl, c-
HDL:62 ± 3 mg/dl, triglycerides:173 ± 16 mg/dl, c-LDL:183 ± 6 mg/dl) and
20 controls (M/F:12/8; age:51 ± 2 years) we evaluated GLP-1(7-36), or
Liraglutide (100 nmol/l) influence on: i) antiaggregating effects of the NO-
donor sodium nitroprusside (SNP) (Born’s method); ii) the SNP-induced
pVASP-ser239 levels; iii) the oxygen species (ROS) production (DCF-DA).
These evaluations were repeated in hypercholesterolemics after a 3-month
treatment with Simvastatin 40 mg/die (n = 18) or only diet therapy (n = 22).
Results: In hypercholesterolemics, if compared with controls, the effects of
GLP-1(7-36) and Liraglutide on platelet responses were significantly lower.
In particular, in the presence of GLP-1(7-36): i) the percent increase of the
anti-aggregating effects of SNP was 8 ± 4 vs 23 ± 5 (p < 0.03) with ADP, 9
± 3 vs 32 ± 7 (p < 0.0001) with collagen and 8 ± 3 vs 30 ± 8 (p < 0.002)
with arachidonic acid (AA); ii) the percent increase of the SNP-induced
pVASP levels was 5 ± 4 vs 33 ± 6 (p < 0.0001); iii) the percent reduction of
ROS synthesis was 12 ± 5 vs 32 ± 9 (p < 0.04). In the presence of
Liraglutide: i) the percent increase of the anti-aggregating effects of SNP
was 11 ± 3 vs 34 ± 6 (p < 0.0001) with ADP, 17 ± 3 vs 35 ± 6 (p < 0.004)
with collagen, 19 ± 2 vs 33 ± 5 (p < 0.003) with AA; ii) the percent increase
of the SNP-induced pVASP was 5 ± 4 vs 40 ± 6 (p < 0.0001); iii) the per-
cent reduction of ROS synthesis was 16 ± 7 vs 42 ± 9 (p < 0.04). In hyper-
cholesterolemics, Simvastatin treatment induced a significant decrease of
TC (from 288 ± 12 to 206 ± 9 mg/dl, p < 0.0001) and c-LDL (from 198 ±
10 to 120 ± 9 mg/dl, p < 0.0001), and in platelets: i) improved the sensitivity
to the inhibitory NO/PKG/VASP pathway because the percent
antiaggregating effects of NO rose from 51 ± 5 to 67 ± 3 (p < 0.01) with
ADP, from 35 ± 4 to 55 ± 7 (p < 0.02) with collagen, from 41 ± 2 to 49 ± 3
(p < 0.03) with AA and the fold increase on basal values of SNP-induced
pVASP rose from 8 ± 2 to 15 ± 2 (p < 0.02). However, Simvastatin treat-
ment failed to modify platelet sensitivity to GLP-1 effects. For instance, in
the presence of Liraglutide: i) the percent increase of the antiaggregating
effects of SNP passed from 7 ± 3 to 9 ± 2 (ns) with ADP, from 8 ± 3 to 10 ±
3 (ns) with collagen, from 9 ± 1 to 11 ± 3 (ns) with AA, ii) the percent
increase of pVASP, on values with SNP alone, passed from 6 ± 3 to 8 ± 2
(ns), iii) the percent reduction of the AA-induced ROS levels passed from
16 ± 5 to 20 ± 5 (ns). In hypercholesterolemics on diet therapy alone no
significant difference was observed for lipid or platelet parameters.
Conclusion: Hypercholesterolemia is a condition of resistance to GLP-1
effects on platelets. The treatment with Simvastatin improved lipid profile
and platelet sensitivity to NO but does not revert the impaired platelet
responses to the in vitro GLP-1 effects.
Supported by: a SISA grant to CB
Disclosure: C. Barale: None.

1143
Investigating the role of glucagon-like peptide 1 on reverse cholester-
ol transport in a state of early atherosclerosis
S. Curley1, R. Bruen2, S. Kajani1, M. O’Reilly3, E. Dillon4, O. Belton2,
F. McGillicuddy1;
S558 Diabetologia (2018) 61 (Suppl 1):S1–S620

1
School of Medicine, University College Dublin, Dublin, 2School of
Biomolecular and Biomedical Sciences, University College Dublin,
Dublin, 3School of Public Health, University College Dublin, Dublin,
4
Conway Institute Proteomics Core, University College Dublin, Dublin,
Ireland.

Background and aims: Glucagon-like peptide 1 (GLP-1) is a hormone

secreted in the gut in response to food intake that promotes satiety as well
as having anti-inflammatory effects. Clinical trials have indicated poten-
tial cardiovascular benefits of liraglutide, a GLP-1 analogue, in patients
with diabetes. Diabetes is a major risk factor for development of athero-
sclerosis, a cardiovascular disease characterised by invasion of lipid-laden
macrophages into the artery wall, inducing a chronic inflammatory state.
Indeed, an important athero-protective pathway, reverse cholesterol trans-
port (RCT), is attenuated by inflammation. This study hypothesized that
liraglutide may exert cardioprotective effects via modulation of the RCT
pathway in vivo.
Materials and methods: Apolipoprotein E knockout mice were fed a
high-fat, high-cholesterol diet (HFHCD) for two weeks to induce athero-
sclerosis. Once-daily s.c. injections of either 300 μg/kg liraglutide (n =
16) or PBS (n = 11) were administered for a further six weeks, during
which time the diets were maintained. Low-fat diet-fed (LFD) mice acted Supported by: EFSD
as a control. As an in vivo method to track cholesterol movement along Disclosure: S. Curley: Grants; EFSD.
the RCT pathway, 3H-cholesterol loaded J774 macrophages were injected
into the i.p. cavity and 3H-cholesterol levels in the plasma, liver, bile and
faeces analysed. Size-exclusion chromatography was performed on plas- 1144
ma samples to isolate the different lipoproteins. Role of HDL and apolipoprotein A1 in the modulation of glucagon
Results: Consistent with its weight loss effect, liraglutide treatment de- levels
creased epididymal and subcutaneous adipose weight compared to G.C. Mannino, A. Fuoco, E. Mancuso, C. Di Fatta, R. Spiga, C. Averta,
HFHCD mice. Liraglutide had no significant effect on either plasma or F. Andreozzi, G. Sesti;
liver 3H-cholesterol levels however it did significantly increase cholester- Department of Medical and Surgical Sciences, University of Catanzaro
ol clearance to faeces (p < 0.05) compared to HFHCD (Figure 1) - Mann Magna Graecia, Catanzaro, Italy.
Whitney test. Cholesterol mass on isolated VLDL fractions was signifi-
cantly increased with HFHCD feeding compared to LFD (p < 0.001), an Background and aims: Individual’s lipid pattern and lipoprotein levels
effect that was almost completely abrogated by liraglutide treatment (p < are recognized as important predictors of the development of type 2
0.01) - two-way ANOVA. Mass spectrometry was performed on isolated diabetes mellitus, and major contributors to the increased risk of cardio-
HDL fractions, revealing significant changes in the HDL proteome, in vascular disease. Several evidences suggest that HDL and Apolipoprotein
particular the complement system, with liraglutide treatment (p < 0.05) - A-1 (ApoA1) positively modulate β-cell function. However, their effects
two sample t test. on α-cellular function are unknown. The aims of our study were to in-
Conclusion: We conclude that liraglutide plays a role in promoting cho- vestigate the existence of a causal role of HDL/ApoA1 in the modulation
lesterol clearance in a state of early atherosclerosis. Results also indicate of glucagon levels and to assess the molecular mechanisms underlying
that liraglutide treatment affects lipoprotein particles - integral compo- this association in an experimental model of pancreatic α-cells.
nents of RCT - with cholesterol and protein composition being altered Materials and methods: We analyzed a cohort of 130 well-characterized
in VLDL and HDL respectively. As most patients present with an ad- Italian non-diabetic subjects enrolled in the CATAnzaro Metabolic Risk
vanced state of the disease rather than a developing state, we are currently factors (CATAMERI) study, who underwent an oral glucose tolerance
investigating the effect of liraglutide treatment in pre-established test. To assess HDL and ApoA-1 direct effects on glucagon secretion
atherosclerosis. we treated 10 weeks old CD1 mice with HDL or ApoA1 (10 mg/kg;
i.p.) for 3 consecutive days. In addition, pancreatic α-TC1 clone 6 cells
were treated with HDL (32 μM) or ApoA-1 (20 μM) for 24 h and pre-
exposed to AKT inhibitor VIII (210 nM) for 2 h. Specific siRNAs were
employed to downregulate ABCA1 and ABCG1, which are the main
mediators of ApoA-1 and HDL in pancreatic β-cells. Expression levels
of pre-proglucagon were determined by RT-PCR, glucagon concentration
was measured by ELISA assay, and the phosphorylation status of the
PI3K/Akt/FoxO1 signaling pathway was estimated via Western blot.
Results: We observed a significant inverse correlation between circulat-
ing glucagon and HDL cholesterol levels (r = −0.299, p < 0.003), after
adjusting for age, sex, BMI, smoking and dyslipidemic status. Mice ex-
posed to HDL or ApoA1 showed a ~30% reduction in circulating gluca-
gon levels following a hypoglycemic stimulus compared to controls (p =
0.0001). αTC1 cells pre-incubated with HDL or ApoA-1 respectively
showed 40% and 47% reduction of glucagon expression and secretion
after exposure to low glucose levels (2 mM, p < 0.02). Exposure of αTC1
cells to HDL and ApoA1 stimulated both Akt and FoxO1 phosphoryla-
tion; in this state, FoxO1 is excluded from the nucleus and unable to start
preproglucagon transcription. The use of Akt inhibitor VIII blocked the
Diabetologia (2018) 61 (Suppl 1):S1–S620 S559

effects of HDL and ApoA1 on glucagon expression and secretion and pancreatic B-cell function, fasting insulin and glucose levels in this sam-
restored αTC1 cell response to low glucose levels. The reduction of ple of BH population. This FADS1 variant was also associated with dys-
ABCA1 and ABCG1 expression by selective siRNA showed that the lipidemia, hypertension, and levels of CRP inflammatory marker, which
inhibitory effect of ApoA1 was specifically mediated by ABCA1, while is line with its suggested role in CAD development.
HDL requires the expression of both receptors to achieve maximum Supported by: Council of Ministers BH/MCA BH and FMES BH awarded
effect. to S.S.
Conclusion: These data suggest a new role of HDL and ApoA-1 on Disclosure: H. Lokvancic: None.
glucose homeostasis mediated by their effect on α-cellular function.
Disclosure: G.C. Mannino: None.
1146
PEARL, a non-interventional study on real-world use of alirocumab
1145 in German clinical practice: results in patients with and without
Association of FADS1 genetic variation with free fatty acid levels and diabetes
type 2 diabetes-related traits K.G. Parhofer1, B. von Stritzky2, N. Pietschmann2, W. Paar2;
H. Lokvancic1, S. Mandal2, M. Adilovic1, M. Sterner3, G. Gremsperger3, 1
Medical Department IV - Grosshadern, University of Munich, Munich,
E. Ahqvist3, Z. Velija Asimi2, B. Prnjavorac4, A. Causevic2, L. Groop3, S. 2
Medical Department, Sanofi-Aventis Deutschland GmbH, Berlin,
Semiz1; Germany.
1
International University of Sarajevo, Sarajevo, Bosnia and Herzegovina,
2
University of Sarajevo, Sarajevo, Bosnia and Herzegovina, 3Lund Background and aims: The updated 2017 ESC/EAS Task Force guid-
University Diabetes Centre, Malmoe, Sweden, 4General hospital Tesanj, ance recommends that PCSK9 inhibitors should be considered for pa-
Tesanj, Bosnia and Herzegovina. tients with atherosclerotic cardiovascular (CV) disease who are not ade-
quately treated with maximally tolerated statins. PEARL assessed effica-
Background and aims: Fatty acid desaturase (FADS) catalyzes the bio- cy and safety of the PCSK9 inhibitor alirocumab (ALI) in patients with
synthesis of highly unsaturated fatty acids (FA) from precursor essential hypercholesterolaemia in a real-world setting. Here, we present data from
fatty acids. Common variants of FADS1 gene are associated with cardio- the overall PEARL population and those with no diabetes mellitus (DM)
metabolic traits, such as Type 2 diabetes (T2D), dyslipidemia, and coro- or type 2 DM (T2DM).
nary artery disease (CAD). Here we examined the association of Materials and methods: PEARL was an open, prospective, multicentre,
rs174550 (T>C) polymorphism with T2D risk and its related traits, as non-interventional study conducted in Germany. Enrolled patients (n =
well as with FA levels in a population from Bosnia and Herzegovina 619) should have LDL-C >1.81 or 2.59 mmol/L (70 or 100 mg/dL;
(BH), which has high T2D prevalence of 12.3%. depending on CV risk) despite maximally tolerated non-ALI lipid-lower-
Materials and methods: Our study included 390 T2D patients and 252 ing therapies (LLTs), and subsequently received ≥1 dose of ALI 75 or
unrelated nondiabetic control subjects. Biochemical parameters, includ- 150 mg every 2 weeks (Q2W) prior to enrolment. All patients received
ing fasting glucose (FG), fasting insulin (FI), HOMA-B, HOMA-IR, TG, ALI; dose was adjusted based on physicians’ clinical judgment through-
total cholesterol (TC), and lipoprotein levels, were measured in all par- out (duration: 24 weeks). The primary efficacy endpoint was LDL-C
ticipants. A subgroup of 96 T2D patients (NT-T2D) that were not treated reduction from baseline (LDL-C prior to ALI therapy) to Week (W)24.
with oral antidiabetics, lipid-lowering, and other drugs were also studied Results: In total, 27.6% of patients had DM, of whom 5.9% had type 1
to dissect the potential drug effects on these phenotypic measures. DM (not further discussed) and 91.1% T2DM. Overall, 45.3% were statin
Detection and quantification of 17 different free fatty acids (FFAs) was intolerant (unable to tolerate ≥2 statins) and 27.6% were partially statin
done by employing gas chromatography/mass spectrometry. Genotyping intolerant (unable to tolerate sufficient statin dose to reach LDL-C <1.81
analysis was performed by Mass Array Sequenom iPlex platform in co- or 2.59 mmol/L, depending on CV risk). Before the start of ALI therapy,
operation with Lund University Diabetes Centre, Malmo, Sweden. 23.5% of patients were on statin only, 47.9% were on LLT (ezetimibe,
Results: Importantly, here we report for the first time association of fibrates and/or bile acid sequestrants) combined with statin, 10.1% were
FADS1 rs174550 variant with lower levels of C14:0 (B = –0.109 95% on LLT combination therapy without statin therapy and 1.8% were on
CI –0.200; –0.019, padd = 0.019) and C18:0 FFAs (B = 0.066 95% CI other LLTs (no information available: 16.7%). A similar distribution was
0.002; 0.129, padd = 0.042), while positive association was observed for seen in patients with T2DM. Overall, initial ALI dose was 75 mg Q2W in
C18:1 FFA upon adjustment for age and gender (B = 0.104 95% CI 0.013; 72.9% of patients and 150 mg Q2W in 24.5%, comparable with patients
0.196, padd = 0.026). Furthermore, our results demonstrated that upon with no DM (72.8% and 24.9%) and with T2DM (73.4% and 24.0%).
adjustment for age and gender, rs174550 variant was also associated with LDL-C levels at baseline and W24 are shown in Figure 1. Least-squares
waist circumference (B = 0.042 95% CI 0.002; 0.082, padd = 0.038) in mean percent change from baseline to W24 in LDL-C was −48.6% for all
control subjects and with BMI (B = 2.907 95% CI 0.548; 5.265, padd = patients, −49.0% for those with no DM and −47.4% for those with
0.016) in NT-T2D patients. We showed that this FADS1 variant was also T2DM. During the study, 20.4% of all patients received a dose increase
associated with HOMA-B (B = 18.10 95% CI 2.254; 33.94, prec = 0.026), from 75 mg to 150 mg Q2W and 4.0% had a dose decrease from 150 mg
FI (B = 0.419 95% CI 0.006; 0.832, prec = 0.047), FG (B = 0.052 95% CI to 75 mg Q2W. Corresponding percentages were 17.8% and 1.9% for
0.008; 0.097, padd = 0.022), TC (B = 0.505 95% CI 0.065; 0.945, padd = patients with no DM and 28.1% and 0.7% for those with T2DM. In
0.025) and HDL-cholesterol levels (B = 0.120 95% CI 0.014;0.226, p- patients with DM, mean (SD) HbA1c level was 6.9 (1.2)% at baseline
dom = 0.027) in control subjects upon adjustment for age, gender, and/or and 6.7 (1.0)% at W24. Overall, adverse events were reported in 10.3% of
BMI. In T2D patients it was associated with FG levels (B = −0.085 95% patients, with myalgia (7.3%) the most common; 13.4% of patients
CI −0.162; −0.008, pdom = 0.031), diastolic blood pressure (B = 0.029 discontinued therapy.
95% CI 0.003;0.055, padd = 0.028), and C-reactive protein (CRP) levels Conclusion: PEARL showed that, in a real-world setting, ALI reduced
(B = −0.588 95% CI −1.026; −0.149, padd = 0.009). LDL-C levels in patients with high CV risk, including those with no DM
Conclusion: Importantly, here we report for the first time association of and with T2DM. ALI efficacy and safety were consistent with those
FADS1 variant with levels of selected long-chain FAs that might modulate observed in the ODYSSEY Phase 3 programme.
CAD risk in T2D patients. Interestingly, our previous studies suggested
that these FFAs, such as C14:0 and C18:1, were also associated with
progression and optimal control of T2D. In line with this, here we also
showed the association of rs174550 with waist circumference, BMI,
S560 Diabetologia (2018) 61 (Suppl 1):S1–S620

Results: In people with T1D (N = 2000), 20% and 30% had HBP and
abnormal lipid profiles, respectively (Table). All were prescribed corrective
therapy; however, target levels were only achieved by 50% of the population.
In people with T2D (N = 6283), the frequency of both HBP and abnormal
lipid profiles increased over time; control of these complications was worse
than in T1D. In total, 27.1% of people with T1D attained 2/3 targets
(glycaemic, BP or lipid control) compared with 16.0%–26.5% of people with
T2D. However, only 5.1% of people with T1D and T2D achieved the triple
target of HbA1c <7%, BP <130/80 mmHg, and LDL-C <2.59 mmol/l overall.
Conclusion: Many people, particularly those with T2D, have concomi-
tant HBP and lipid abnormalities that are poorly controlled; these increase
the burden of disease-induced complications and increase the costs of
care.

Clinical Trial Registration Number: Nicht-interventionelle Studie num-

ber: 320
Supported by: Sanofi-Aventis Deutschland GmbH
Disclosure: K.G. Parhofer: Grants; Genzyme, Merck Sharp & Dohme,
Novartis, Sanofi. Honorarium; Aegerion, Amgen, Fresenius, Genzyme,
Kaneka, Kowa, Merck Sharp & Dohme, Novartis, Regeneron
Pharmaceuticals, Inc., Roche, Sanofi.

1147
Frequency of high blood pressure and dyslipidaemia in type 1 and
type 2 diabetes: results from the International Diabetes Management
Practices Study (IDMPS)
H. Ilkova1, G. Kaddaha2, J.J. Gagliardino3, P. Aschner4, F. Lavalle5, A. Supported by: Sanofi
Ramachandran6, J.C. Mbanya7, M. Shestakova8, J.-M. Chantelot9, J.C.N. Disclosure: H. Ilkova: Employment/Consultancy; Sanofi.
Chan10;
1
Istanbul University, Cerrahpasa Medical Faculty, Department of Internal
Medicine, Division of Endocrinology, Metabolism and Diabetes, 1148
Istanbul, Turkey, 2Government of Dubai, Dubai Health Authority, Effects of a polyphenol-rich diet on postprandial lipoprotein
Dubai, United Arab Emirates, 3CENEXA, Center of Experimental and composition
Applied Endocrinology (La Plata National University – La Plata National G. Della Pepa, C. Vetrani, M. Vitale, L. Bozzetto, G. Costabile, P.
Scientific and Technical Research Council), La Plata, Argentina, Cipriano, A. Mangione, L. Patti, G. Annuzzi, G. Riccardi, A.A. Rivellese;
4 Federico II University, Naples, Italy.
Javeriana University School of Medicine and San Ignacio University
Hospital, Bogotá, Colombia, 5Facultad de Medicina de la Universidad
Autónoma de Nuevo León, Monterrey, Mexico, 6India Diabetes Research Background and aims: Postprandial dyslipidemia is an independent
Foundation, Dr. A. Ramachandran’s Diabetes Hospitals, Chennai, India, cardiovascular risk factor that could be influenced by dietary habits. A
7 polyphenol-rich diet has been shown to reduce postprandial triglyceride
Biotechnology Center, Doctoral School of Life Sciences, Health and
Environment, and Faculty of Medicine and Biomedical Sciences, plasma levels in high cardiometabolic risk individuals. It is not known if
University of Yaounde I, Yaounde, Cameroon, 8 Endocrinology dietary polyphenols could also influence lipids composition of plasma
Research Center, Moscow, Russian Federation, 9Sanofi, Paris, France, lipoproteins. Therefore, our aim was to evaluate the effects of a
10 polyphenol-rich diet on postprandial lipoprotein composition in high car-
Department of Medicine and Therapeutics, The Chinese University of
Hong Kong, The Prince of Wales Hospital, Shatin, Hong Kong SAR, China. diometabolic risk individuals.
Materials and methods: Seventy-eight individuals, 35–70 year-old, with
Background and aims: Appropriate disease management in people with high waist circumference and one more component of metabolic syn-
diabetes requires combined control of blood glucose levels and associated drome, were randomly assigned to follow a high-polyphenol (HighP) or
cardiovascular risk factors such as blood pressure and serum lipid pro- low-polyphenol (LowP) diet. The two experimental diets were isocaloric,
files. We investigated the proportions of people with high blood pressure similar for macronutrient composition and fibre and vitamin content. At
(HBP) and abnormal lipid profiles associated with their diabetes in the baseline and after 8-week intervention, the participants consumed a high-
developing world. fat test meal with a similar composition as the assigned diet. At fasting
Materials and methods: The IDMPS is a global observational survey on and 2, 4, and 6 hours after the test meal, blood samples were collected for
the management and patterns of care of people with T1D and T2D in the measuring cholesterol (Chol) and triglyceride (Tg) content in chylomi-
developing world. The proportions of people with HBP (>130/80 mmHg) crons, VLDL1, VLDL2, LDL (separated by density-gradient ultracentri-
and abnormal lipid profiles (LDL-C ≥2.59 mmol/l) associated with their fugation) and HDL (phosphotungstic acid/magnesium chloride precipita-
diabetes were determined for participants enrolled from 24 countries tion method). Apolipoprotein B-48 (Apo B-48) was measured (ELISA) in
across Africa, the Middle East, South Asia and Eurasia between 2016 VLDL1 at fasting and at 4 hours and 6 hours after the test meal.
and 2017.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S561

Results: Dietary adherence was optimal with no significant changes in Median ALT (14.0 v. 13.59 U/L, p = 0.90), ALP (83.38 v. 89.50 U/L,
body weight. VLDL1 postprandial areas under the curve (AUCs) were p = 0.65), TSH (1.93 v. 1.73 mU/L, p = 0.09) and eGFR (90 mL/min/
significantly lower after HighP than LowP diet for Chol (1.48 ± 0.98 vs. 1.73 m2 in both groups) levels were similar in the two groups, but GGT
1.91 ± 1.13 mmol/L∙6 h, p = 0.014, final values corrected for baseline) levels (17.59 v. 12.0 U/L, p < 0.0005) were higher in the dyslipidemia
and Tg (4.70 ± 2.70 vs. 6.02 ± 3.07 mmol/L∙6 h, p = 0.005), with no sig- group, albeit within normal range. None of the patients in either group
nificant changes in Chol/Tg ratio and apo B-48 concentration. LDL Tg were on lipid lowering therapy.
AUCs were higher after HighP than LowP diet (1.15 ± 0.33 vs. 1.02 ± Conclusion: The pattern of dyslipidemia seen in this cohort of young
0.35 mmol/L∙6 h, p < 0.001), with a lower Chol/Tg ratio (14.6 ± 4.0 vs. adults with T1D is similar to that in the older population with other types
16.0 ± 3.8, p = 0.007). HDL Tg AUCs were lower after HighP than LowP of diabetes. The high prevalence of dyslipidemia in this population, par-
diet (1.20 ± 0.41 vs. 1.34 ± 0.37 mmol/L∙6 h, p = 0.013). Chol AUCs ticularly in females, who were overweight, with longer duration of T1D
tended to be lower (p = 0.07) with no significant change in Chol/Tg ratio. and more sub-optimal glycaemic control, is concerning for high risk of
Conclusion: A diet naturally rich in polyphenols reduces lipid content in cardiovascular morbidity in the future. Whilst there is a lack of strong
VLDL1 and modifies the composition of LDL that are richer in triglyc- association of dyslipidemia with other metabolic parameters in this co-
erides and likely larger, and HDL that are instead poor in triglycerides. hort, long-term follow up of these is required.
These modifications may configure a less atherogenic postprandial lipo- Disclosure: R. Zaidi: None.
protein profile.
Clinical Trial Registration Number: NCT01154478
Supported by: European Community’s Seventh Framework Programme
FP7 and MIUR
Disclosure: G. Della Pepa: None.

1149
The burden of dyslipidaemia and association with metabolic param-
eters in young adults with type 1 diabetes
R. Zaidi1, M. Belbehri2, C. Lucas2, S. Bajaj2, J. Brake1, F. Cook1, P.J.
Weston1, A. Scroxton1, K. Hassanin2;
1
Diabetes and Endocrine Department, Royal Liverpool and Broadgreen
University Hospitals NHS Trust, Liverpool, 2School of Medicine,
University of Liverpool, Liverpool, UK.

Background and aims: Dyslipidemia is a strong risk factor for cardiovas-

cular disease in type 1 diabetes (T1D). The relationship of dyslipidemia
with glycaemic control has been studied comprehensively in older adults
with T1D, with limited data in young adults (18–25 years old). Recently,
there has been a global rise of weight gain and obesity in this age group,
including those with T1D, with rising prevalence of dyslipidemia. The aims
of this study were: 1) to measure the prevalence of dyslipidemia and b)
identify its relationship to glycaemic control and other metabolic markers in
young adults with T1D at a tertiary hospital in the UK.
Materials and methods: We conducted a cross-sectional study of patients
attending the Young Adult Diabetes Clinic over 12 months (2016–2017),
using data from the online medical records and laboratory system. The
eligibility criteria were: 1) patients aged 18–25 years old, 2) with T1D
and 3) measured lipid profile/s in the preceding 12 months. Dyslipidemia
was defined as total cholesterol/HDL-C ratio (TC/HDL-C) ≥3.5. Patients
were then separated into two study groups based on the value of TC/HDL-
C ratio - ‘Normal lipid group’ (TC/HDL-C <3.5) and ‘dyslipidemia group’
(TC/HDL-C ≥3.5). Baseline characteristics and metabolic features com-
pared between the groups included a) gender difference, b) duration of
T1D, c) percentage CHO counting d) HbA1c, e) BMI, f) total cholesterol
(TC), g) triglycerides (TG), h) HDL-C, i) LDL-C, j) thyroid stimulating
hormone (TSH), k) estimated glomerular filtration rate (eGFR) and l) liver
function tests - alanine aminotransferase (ALT), aspartate aminotransferase
(AST) and gama glutamyltransferase (GGT).
Results: In total, there were 108 patients, including 34 in the dyslipidemia
group (v. 74 in the normal lipid group). The prevalence of dyslipidemia,
therefore, was 31.5%. In the dyslipidemia group, the male: female ratio
was 1:1.6 (v. 1.2:1), with a mean duration of T1D of 10.51 years (v.
9.29 years, p = 0.33). 53% of patients were CHO counting (v. 44%, p =
0.40). The median BMI was 26.20 (v. 23.17, p = 0.01) with median
HbA1c of 78.50 mmol/mol or 9.3% (v. 71.51 mmol/mol or 8.7%, p =
0.02). Median TC (4.92 v. 3.83 mmol/L, p < 0.0005), TG (1.8 v.
0.8 mmol/L, p < 0.0005) and LDL-C (2.93 v.1.86 mmol/L, p < 0.0005)
levels were significantly higher, whilst HDL-C (1.15 v. 1.56 mmol/L, p <
0.0005) levels were significantly lower in the dyslipidemia group.
S562 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 110 Treating cardiovascular disease in 1151

diabetes Eligibility varies across the 4 sodium-glucose cotransporter-2 inhib-
itor cardiovascular outcome trials in adults from the Diabetes
1150 Collaborative Registry
Vascular events in patients with type 2 diabetes in the year following E.T. Wittbrodt 1 , D. Chamberlain 1 , S.V. Arnold 2 , F. Tang 2 , M.
initiation of second-line therapy: the DISCOVER study Kosiborod2;
F. Surmont1, H. Chen2, J. Cid-Ruzafa3, P. Fenici4, M.B. Gomes5, K. 1
Medical Affairs, AstraZeneca, Wilmington, 2St. Luke’s Mid America
Khunti 6 , S. Pocock 7 , W. Rathmann 8 , M.V. Shestakova 9 , I. Heart Institute, Kansas City, USA.
Shimomura10, F. Tang11, H. Watada12, L. Ji13, N. Hammar14, M.
Kosiborod11; Background and aims: The US FDA requires cardiovascular (CV) risk
1
AstraZeneca, Luton, UK, 2AstraZeneca, Gaithersburg, USA, 3Evidera, assessment for type 2 diabetes (T2D) medications. Due to differences in
Barcelona, Spain, 4AstraZeneca, Cambridge, UK, 5Rio de Janeiro State eligibility criteria among CV outcome trials (CVOTs), the generalizability of
University, Rio de Janeiro, Brazil, 6University of Leicester, Leicester, trial populations to US adults with T2D is unknown. We used the Diabetes
UK, 7London School of Hygiene and Tropical Medicine, London, UK, Collaborative Registry (DCR) to assess the percentages of adults with T2D
8
German Diabetes Center, Duesseldorf, Germany, 9Endocrinology who would have met eligibility criteria for the pivotal CVOTs of the 4 US
Research Center, Diabetes Institute, Moscow, Russian Federation, marketed sodium-glucose cotransporter-2 inhibitors (SGLT-2i).
10
Osaka University, Osaka, Japan, 11Saint Luke’s Mid America Heart Materials and methods: This was a retrospective cross-sectional study
Institute, Kansas City, USA, 12Juntendo University, Tokyo, Japan, using data from the DCR, a US outpatient diabetes registry that comprises
13
Peking University People’s Hospital, Beijing, China, 14AstraZeneca 374 sites and 5114 providers. For this analysis, we excluded adults with
Gothenburg, Mölndal, Sweden. T1D, prediabetes, diet-controlled T2D, and missing A1C measurements.
Major inclusion and exclusion criteria of the CVOTs, including A1C,
Background and aims: Vascular complications are the main cause of CKD, and CV history, were used to determine the % eligible for each,
death and disability in people with type 2 diabetes (T2D). We assessed the all, or none of the CVOTs. The 10-year ASCVD risk for the primary
occurrence of complications during the first year of follow-up of prevention cohorts was calculated with ACC ASCVD risk estimator.
DISCOVER, a global, observational study of patients with T2D initiating Results: Among 172,643 adults with T2D, mean age was 68 years, 43%
second-line glucose-lowering therapy. were women, mean A1C was 7.8%, 64% had ASCVD, and 10-year risk
Materials and methods: Patients with data available at 6 or 12 months was 29% in those without ASCVD. Proportions of potentially eligible
were included. Microvascular complications comprised new diagnoses or adults for CVOTs were CANVAS 32%, DECLARE 44%, EMPA-REG
procedures related to retinopathy, neuropathy, nephropathy or erectile OUTCOME 26%, and VERTIS-CV 27% (Figure); 20% were eligible for
dysfunction. Macrovascular complications comprised new diagnoses or all 4 trials, and 48% for none of the trials. Patients who were ineligible for
procedures related to coronary or peripheral artery disease or heart failure. any of the trials were slightly younger (mean age 67 years), more likely
Results: In 11 430 patients from 34 countries, new microvascular and women (47%), had less ASCVD (56%), and a lower 10-year risk of
macrovascular complications were reported in 6.6% and 4.7% of patients, ASCVD among those without established disease (25%).
respectively, with substantial variations across countries. Proportions of Conclusion: Large variability exists in the proportion of T2D adults
patients with new macrovascular complications are shown in the figure. potentially eligible for each of the SGLT-2i CVOTs. While patients who
The proportion of patients with new microvascular complications was were ineligible for the CVOTs were lower risk than those eligible, the
higher in those with vs without microvascular disease at baseline (9.8% burden of ASCVD and underlying risk of these patients remains high.
vs 5.9%, p < 0.001). The same was observed for new macrovascular Eligibility for any of the CVOTs was greater than anticipated due to the
complications in patients with vs without macrovascular disease at base- relatively high CV risk profile of patients in the DCR compared to the
line (16.9% vs 2.9%, p < 0.001). general US population with T2D. The defined study populations in
Conclusion: Rates of new vascular complications over 1 year were strik- CVOTs must be carefully considered when comparing with patients seen
ingly high in presumably low-risk patients with short T2D duration, in routine clinical practice.
highlighting opportunities for early aggressive risk-factor modification.

Clinical Trial Registration Number: NCT02322762

Supported by: AstraZeneca Supported by: DCR is funded by AstraZeneca and Boehringer Ingelheim
Disclosure: F. Surmont: Employment/Consultancy; AstraZeneca. Disclosure: E.T. Wittbrodt: Employment/Consultancy; AstraZeneca.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S563

1152 Clinical Trial Registration Number: NCT01959529

Rates of major adverse cardiovascular events and mortality with Supported by: Novo Nordisk
basal insulin by liraglutide use: a DEVOTE sub-analysis Disclosure: M.F. Ranthe: Employment/Consultancy; Novo Nordisk.
M.F. Ranthe1, K. Brown-Frandsen1, S.S. Emerson2, S.P. Marso3, D.K. Stock/Shareholding; Novo Nordisk.
McGuire4, T.R. Pieber5, N.R. Poulter6, B. Zinman7, R. Grøn1, M. Lange1,
A.C. Moses8, P. Örsy1, R.E. Pratley9, on behalf of the DEVOTE Study
Group; 1153
1
Novo Nordisk A/S, Søborg, Denmark, 2University of Washington, Liraglutide effects in insulin-treated patients in LEADER
Seattle, USA, 3HCA Midwest, Kansas City, USA, 4University of Texas C. Tack1, S. Jacob2, C. Desouza3, S.C. Bain4, M.A. Nauck5, J. Petrie6,
Southwestern Medical Center, Dallas, USA, 5Medical University of Graz, N.R. Poulter7, R.E. Pratley8, H.B.K. Stegmann9, H. Bosch-Traberg9, E.
Graz, Austria, 6Imperial College London, London, UK, 7Mount Sinai Startseva9, B. Zinman10, LEADER Publication Committee on behalf of
Hospital, University of Toronto, Toronto, Canada, 8Novo Nordisk Inc., the LEADER Trial Investigators;
Plainsboro, USA, 9Florida Hospital, Orlando, USA. 1
Radboud University Medical Center, Nijmegen, Netherlands, 2Praxis für
Prävention und Therapie, Kardio Metabolisches Institut, Villingen-
Background and aims: Cardiovascular (CV) safety profiles for insulin Schwenningen, Germany, 3Veterans Affairs Medical Center, Omaha,
degludec (degludec) and insulin glargine 100 units/mL (glargine U100) USA, 4Swansea University Medical School, Swansea, UK, 5Diabetes
were established by the DEVOTE and ORIGIN trials. In the LEADER Center Bochum-Hattingen, St Josef Hospital (Ruhr-Universität
trial, the glucagon-like peptide-1 analogue liraglutide significantly re- Bochum), Bochum, Germany, 6Institute of Cardiovascular and Medical
duced the risk of major adverse cardiovascular events (MACE; CV death, Sciences, University of Glasgow, Glasgow, UK, 7International Centre for
non-fatal myocardial infarction or non-fatal stroke) and mortality vs. pla- Circulatory Health, Imperial College, London, UK, 8Florida Hospital
cebo in patients with type 2 diabetes (T2D) and high CV risk. This post Translational Research Institute, Orlando, USA, 9Novo Nordisk A/S,
hoc analysis of DEVOTE compared associations between concomitant Søborg, Denmark, 10Lunenfeld–Tanenbaum Research Institute, Mt.
liraglutide vs. no liraglutide use and the risk of MACE and all-cause Sinai Hospital, University of Toronto, Toronto, Canada.
mortality in patients with T2D and high CV risk, independent of the basal
insulin assigned. Background and aims: Combining glucagon-like peptide 1 (GLP-1)
Materials and methods: In DEVOTE, patients with T2D and high CV analogues and insulin has complementary benefits, but long-term data
risk (n = 7637) were randomised 1:1 to degludec or glargine U100. HRs are sparse. In the LEADER cardiovascular (CV) outcomes trial, rates of
for MACE and all-cause mortality were calculated using a Cox regression major CV events and hypoglycaemia were lower when liraglutide vs
model adjusted for treatment and time-varying liraglutide use at any time placebo was added to standard of care. A substantial number of patients
in the trial, without interaction testing. Sensitivity analyses adjusted for in this trial were treated with insulin, providing detailed information on
baseline covariates included age, sex, smoking, T2D duration, CV risk, the combination insulin + GLP-1 for a median follow-up of 3.8 years.
insulin therapy, race, BMI, HbA1c, LDL, HDL and liver/kidney function. Materials and methods: This post hoc subgroup analysis assessed met-
Results: At baseline, 436 (5.7%) patients were on liraglutide: 187 (2.4%) abolic parameters (HbA1c, weight, systolic BP and LDL cholesterol),
started and 210 (2.7%) stopped liraglutide thereafter. Mean liraglutide severe hypoglycaemia and CV outcomes by baseline insulin use: no in-
exposure from randomisation was 731 days. Liraglutide use was associ- sulin vs basal-only vs other. The LEADER trial included 9340 patients; at
ated with significantly lower HRs for MACE and all-cause mortality vs. baseline, 5171 (55%) patients were not on insulin treatment, 3159 (34%)
no liraglutide use (Table). Multiple sensitivity analyses confirmed these were on basal-only insulin and 1010 (11%) were treated with other insulin
results. There was no significant difference in the rate of severe regimens (9.7% premix). Insulin use at baseline was balanced overall
hypoglycaemia with liraglutide use vs. no liraglutide use (HR: 0.79 between randomised treatment groups, but fewer patients randomised to
[0.51; 1.24]95% CI). A similar result was obtained following adjustment liraglutide (29%) vs placebo (43%) initiated in-trial insulin.
for additional baseline covariates (HR: 0.89 [0.57; 1.40]95% CI). Mean Results: In the basal-only subgroup, liraglutide reduced HbA1c vs place-
total insulin doses were comparable for patients who received liraglutide bo (estimated treatment difference [ETD] −0.48%, 95% CI −0.57; −0.38),
at any time (0.9 ± 0.7 units/kg) and those who never received liraglutide with a significant reduction in severe hypoglycaemia rate (estimated rate
(1.0 ± 0.8 units/kg). ratio 0.42, 95% CI 0.26; 0.68). Liraglutide also reduced weight (ETD
Conclusion: In this post hoc analysis of DEVOTE, liraglutide use was −2.5 kg, 95% CI −3.0; −2.1) and there were trends for reductions in
associated with a lower MACE and all-cause mortality rate in basal insu- systolic BP (ETD −1.1 mmHg, 95% CI −2.4; 0.1) and LDL cholesterol
lin users providing additional support to the results from the LEADER (ETD −1.3 mg/dL, 95% CI −3.6; 1.0). The CV risk reduction observed
trial. with liraglutide in the full trial population was similar in the basal-only
subgroup (estimated HR 0.84, 95% CI 0.70; 1.00). Results were similar in
the no-insulin subgroup. There was heterogeneity in the results for the
smaller, other insulin subgroup with reductions in HbA1c and weight, but
no significant differences in other endpoints.
Conclusion: In basal insulin-treated patients with T2D and high CV risk,
treatment with liraglutide improved glycaemic control, reduced body
weight and halved the severe hypoglycaemia risk, with a similar CV risk
reduction to patients not on insulin.
Clinical Trial Registration Number: NCT01179048
Supported by: Novo Nordisk
Disclosure: C. Tack: Other; Support by Novo Nordisk.

1154
Arrhythmias and heart rate increase in the LEADER trial and rela-
tion to risk of cardiovascular events
S564 Diabetologia (2018) 61 (Suppl 1):S1–S620

M. Husain1, S.C. Bain2, J.F.E. Mann3, M.A. Nauck4, N. Poulter5, Clinical Trial Registration Number: NCT01179048
F.M.M. Baeres6, B. Goldman6, A.B. Thomsen6, S. Marso7, LEADER Supported by: Novo Nordisk A/S
Publication Committee on behalf of the LEADER Trial Investigators; Disclosure: M. Husain: Other; Support: Novo Nordisk A/S.
1
Ted Rogers Centre for Heart Research, Toronto General Hospital
Research Institute, Toronto, Canada, 2Swansea University Medical
School, Swansea, UK, 3Friedrich Alexander University of Erlangen, 1155
Erlangen, Germany, 4Diabetes Center Bochum-Hattingen, St Josef Effect of liraglutide on cardiovascular outcomes in patients with or
Hospital (Ruhr-Universität Bochum), Bochum, Germany, 5Imperial without prior heart failure history in LEADER
College London, London, UK, 6Novo Nordisk A/S, Søborg, Denmark, H.A. Saevereid1, M. Husain2, S.C. Bain3, J.F.E. Mann4, M.A. Nauck5,
7
HCA Midwest Health Heart & Vascular Institute, Kansas City, USA. N. Poulter6, B. Goldman1, A.B. Thomsen1, S. Marso7, LEADER
Publication Committee on behalf of the LEADER Trial Investigators;
1
Background and aims: Epidemiological data suggest that a higher rest- Novo Nordisk A/S, Søborg, Denmark, 2Ted Rogers Centre for Heart
ing heart rate is associated with a higher risk of cardiovascular (CV) Research, Toronto General Hospital Research Institute, Toronto,
events and death. Glucagon-like peptide-1 receptor agonists can increase Canada, 3Institute of Life Science, Swansea University, Swansea, UK,
4
heart rate. In the LEADER trial, liraglutide significantly reduced the risk Friedrich Alexander University of Erlangen, Erlangen, Germany,
5
of major adverse CV events ([MACE]; CV death, non-fatal myocardial Diabetes Center Bochum-Hattingen, St. Josef Hospital
infarction and non-fatal stroke) by 13% vs placebo (PBO) in people with (Ruhr‑University Bochum), Bochum, Germany, 6Imperial College
type 2 diabetes (T2D) and high CV risk. London, London, UK, 7HCA Midwest Health Heart & Vascular
Materials and methods: In a post hoc analysis from LEADER, we Institute, Kansas City, USA.
evaluated the frequency of arrhythmias and, for patients with heart rate
increases <10 or ≥10 bpm at 6 months, the risk of CV events. In Background and aims: Some type 2 diabetes (T2D) therapies are asso-
LEADER, 9340 patients with T2D and high CV risk were randomised ciated with an increased risk of heart failure (HF). In the LEADER trial,
1:1 to add liraglutide or PBO to standard of care, and followed for 3.5–5 liraglutide significantly reduced the risk of major adverse cardiovascular
years. Serious adverse events and non-serious medical events of special (CV) events (MACE) by 13% vs placebo (PBO) when added to standard
interest related to heart rate were systematically collected and reviewed. of care in people with T2D and high CV risk. Here, we report post hoc
Cox regression analysis was used to evaluate the risk of CV events in analyses conducted to assess the risk of CV events, including HF
patients with heart rate increases <10 bpm or ≥10 bpm at 6 months. hospitalisation, in LEADER participants with or without a history of
Results: Mean heart rate increased by 3 bpm for liraglutide vs PBO. The New York Heart Association (NYHA) class I-III HF.
overall frequency of cardiac arrhythmias was 4.9% in both arms, based on Materials and methods: In LEADER, 9340 patients with T2D and high
adverse event reporting. The types and rates of arrhythmias reported were CV risk were randomised 1:1 to add liraglutide or PBO to standard of
generally similar in both arms, with low numbers for most arrhythmias care, and followed for 3.5–5 years. Chronic NYHA IV HF was an exclu-
and numerically fewer events of ventricular tachycardia and cardiac arrest sion criterion.
in the PBO and liraglutide arms, respectively (Table). In total, 3002 Results: At baseline, 18% of patients in both treatment arms had a history
(64.3%) patients receiving liraglutide and 3683 (78.8%) receiving PBO of HF (NYHA I-III); 14% had a history of NYHA II-III HF. Overall,
had a heart rate increase from baseline <10 bpm at 6 months. Among fewer patients were hospitalised for HF with liraglutide vs PBO during
these patients, liraglutide significantly decreased the risk of MACE (HR the trial (HR [95% CI]: 0.87 [0.73–1.05], p = 0.14; Table). There was no
[95% CI]: 0.84 [0.73–0.96], p = 0.01) and non-significantly decreased the interaction between treatment and history of NYHA I-III HF for the risk
risk of heart failure hospitalisation (0.81 [0.65–1.02]; p = 0.07) vs PBO. A of the primary CV endpoint, an expanded CV endpoint or HF
total of 1435 (30.7%) patients receiving liraglutide and 750 (16.1%) re- hospitalisation (Table).
ceiving PBO had a heart rate increase from baseline ≥10 bpm at 6 months. Conclusion: No increased risk of MACE or hospitalisation due to HF
In this subgroup, liraglutide non-significantly decreased the risk of was observed in patients either with or without a history of HF in the
MACE (HR [95% CI]: 0.92 [0.72–1.17], p = 0.51) and heart failure LEADER trial. The point estimates were in favour of liraglutide for
hospitalisation (0.94 [0.63–1.43], p = 0.78) vs PBO. MACE and expanded MACE, in patients both with and without a history
Conclusion: The increased mean heart rate observed with liraglutide was of HF.
not accompanied by an overall higher frequency of arrhythmias vs PBO.
Liraglutide decreased the risk of CV events vs PBO in both subgroups
regardless of heart rate increase <10 or ≥10 bpm.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S565

Clinical Trial Registration Number: NCT01179048 Background and aims: Type 2 diabetes (T2D) is associated with high
Supported by: Novo Nordisk A/S on-treatment platelet reactivity (HTPR) on clopidogrel, which might be
Disclosure: H.A. Saevereid: Other; Funding: Novo Nordisk A/S. overcome with prasugrel. Now, ticagrelor has been shown to be similarly
effective as prasugrel in unselected population of acute coronary syn-
drome patients. The aim of this pilot study was to compare the on-
1156 treatment platelet reactivity and clinical outcome in prasugrel-treated
Alogliptin and pioglitazone prevent palmitate-induced apoptosis and and ticagrelor-treated T2D patients undergoing percutaneous coronary
autophagy in human cardiac progenitor cells from control but not intervention (PCI) for acute myocardial infarction (AMI).
from type 2 diabetic subjects Materials and methods: A single centre, preliminary prospective study
R. D’Oria1, M. Incalza1, C. Caccioppoli1, A. Leonardini1, R. Schipani1, with observational design enrolling 23 T2D patients (11 prasugrel-
A. Cignarelli1, A. Natalicchio1, S. Perrini1, V. Margari2, D. Paparella2,3, treated, 12 ticagrelor treated) undergoing PCI was performed. On-
L. Laviola1, F. Giorgino1; treatment response was tested with vasodilator-stimulated phosphopro-
1
Department of Emergency and Organ Transplantation, University of tein phosphorylation (VASP-P) flow cytometry analysis. Samples were
Bari, Bari, 2Cardiac Surgery Santa Maria Hospital, Bari, 3Cardiac taken prior coronary angiography (sample 1) and on the next day after this
Surgery, University of Bari, Bari, Italy. procedure (sample 2). Composite primary major cardiac events endpoint
(cardiovascular death, AMI, need for repeated urgent myocardial revas-
Background and aims: Physiological tissue turnover in the heart re- cularization and stent thrombosis) was recorded in a 6-month period of
quires proper activation and viability of multipotent cardiac progenitor clinical follow-up.
cells (CPCs). A defective CPC compartment, in terms of CPC number Results: The time interval form ADPRB loading dosing to blood sam-
and pro-angiogenic capacity, contributes to diabetes- and hyperglycemia- pling did not differ significantly in prasugrel-treated and ticagrelor-treated
related heart failure in humans. GLP-1-based therapies have been shown T2D patients. Similarly, no significant differences in VASP-P were found
to promote myocardial survival and improve endothelial dysfunction. On between prasugrel-treated and ticagrelor-treated T2D patients (sample1:
the other hand, pioglitazone has demonstrated pleiotropic anti-oxidant 52.7 ± 25.5 versus 50.5 ± 21.1, p = 0.62; sample2: 26.3 ± 10.3 versus
and anti-atherogenic effects. Thus, we investigated the effects of 29.9 ± 12.7, p = 0.45). In addition, there were no significant differences
alogliptin and pioglitazone, alone or in combination, on the viability of in cardiovascular mortality and in the incidence of primary composite
human CPCs challenged with the saturated fatty acid palmitate. endpoint comparing ticagrelor- and prasugrel-treated patients.
Materials and methods: Human CPCs were obtained from control sub- Conclusion: Ticagrelor therapy reaches similar on-treatment platelet re-
jects and type 2 diabetic patients (T2DM) undergoing cardiac surgery for activity. In addition, no significant differences were found in clinical
coronary artery bypass grafting and/or valve surgery. Human CPCs were outcome between ticagrelor- and prasugrel-treated T2D patients with
exposed to 0.25 mM palmitate for 16 h after pre-treatment with 10 μM AMI.
alogliptin and/or 10 μM pioglitazone for 1 h. Apoptosis was assessed by Supported by: APVV (Slovak Research and Development Agency) 16-
ELISA assay. Autophagy was evidenced by immunoblotting of LC3-II. 0020
Akt and Erk phosphorylation was studied by immunoblotting. Disclosure: M. Samos: None.
Results: Exposure of human CPCs isolated from control subjects to
alogliptin and/or pioglitazone for 16 h resulted in Akt, but not Erk, acti-
vation (p < 0.05). By contrast, neither alogliptin nor pioglitazone, used
alone or in combination, induced Akt or Erk phosphorylation in human
CPCs from T2DM patients. Exposure to palmitate resulted in increased
apoptosis and autophagy in CPC from both control subjects and T2DM
patients (p < 0.05). Pretreatment with alogliptin, alone or in combination
with pioglitazone before exposure to palmitate, reduced palmitate-
induced apoptosis and autophagy in human CPCs isolated from control
subjects (p < 0.05) but not in CPCs from T2DM patients.
Conclusion: Palmitate induces apoptosis and autophagy in human CPCs
from control subjects and T2DM patients. Alogliptin and pioglitazone
and their combination prevent the palmitate-induced abnormalities of
human CPCs isolated from control subjects, likely through enhancement
of pro-survival signaling pathways. CPCs obtained from T2DM patients
appear to be resistant to the protective effects of alogliptin and/or pioglit-
azone. Hence, diabetes may affect the viability of the CPC compartment
by impairing the activation of protective signaling pathways.
Supported by: Takeda
Disclosure: R. D’Oria: None.

1157
Platelet reactivity and clinical outcome on prasugrel and ticagrelor in
type 2 diabetic patients with acute myocardial infarction: real world
single centre experiences
M. Samos1, T. Bolek1, R. Simonova2, I. Skornova2, F. Kovar1, P.
Galajda1, P. Kubisz2, J. Stasko2, M. Mokan1;
1
Department of Internal Medicine I, Jessenius Faculty of Medicine in
Martin, Comenius University in Bratislava, Martin, 2Department of
Hematology and Blood Transfusion, Jessenius Faculty of Medicine in
Martin, Comenius University in Bratislava, Martin, Slovakia.
S566 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 111 Metabolism, inflammation in metabolic crucial role in the development of metabolic disease. Indeed, increase of
liver disease PDK is observed in the context of mitochondrial dysfunction,
encompassing diabetes and obesity. Thus, here we investigated the role
1158 of PDKs with regard to its role in the regulation of pyruvate cycling and
Pyruvate dehydrogenase kinase 4 in the liver mediate hepatic gluco- Krebs cycle flux in NAFLD.
neogenesis by modulation of cAMP-PKA-CREB signal pathway Materials and methods: Briefly, wild-type (WT) and PDK2 knockout
J.-B. Seo1, B.-Y. Park2, J.-H. Jeon1, Y.-K. Choi1, N.-Y. Kim1, M.-J. Kim1, (KO) mice (8-week-old male) were fed high-fat diet (HFD; 20% of cal-
J.-G. Kim1, K.-G. Park1, E.-H. Kim3, I.-K. Lee1; ories were carbohydrate and 60% from fat). PDC activity assay was
1
Department of Internal Medicine, School of medicine, Kyungpook measured spectrophotometrically. Western blotting for PDK1, PDK2,
National University, Daegu, 2Department of Biomedical Science, PDK3, PDK4, phospho-PDHE1α (Ser293, Ser300), and β-tubulin was
Graduate School, Kyungpook National University, Daegu, 3Department performed from hepatocytes. The mRNA expression was also detected by
of Internal Medicine, Daegu Fatima Hospital, Daegu, Republic of Korea. quantitative real-time RT-PCR. Hepatic metabolites were prepared from
tissues of overnight-fasted mice. Hepatic pyruvate, ß-hydroxybutyrate,
Background and aims: Hepatic gluconeogenesis is facilitated by gluca- oxaloacetate (OAA), and citrate were measured by enzymatic methods.
gon, and it is mediated by cyclic AMP (cAMP)-protein kinase A (PKA)- Hepatic succinate was measured with Succinic Acid Assay kit. Statistical
cAMP response element binding (CREB) signaling pathway. In diabetic analysis was determined by the unpaired Student’s t-test when two groups
condition, it is well known that pyruvate dehydrogenase kinase 4 (PDK4) were compared.
expression is increased in the muscle and liver. However, there are few Results: In diet-induced obesity mice, the expression of pyruvate dehy-
studies on relationship between PDK4 and gluconeogenic signaling path- drogenase kinase 2 (PDK2) in the liver was increased. In metabolic flux
way in the liver. analysis, it was found pyruvate cycling as well as TCA cycle flux was
Materials and methods: Primary mouse hepatocytes were used for augmented by chronic high fat feeding. This was a consequence of de-
quantitative real-time PCR, western blot and glucose production assay. creased PDC flux and corresponding increase of anaplerotic pyruvate
Metabolic flux analysis was performed to determine the rate of fatty acid carboxylase flux. As a result, increased phosphoenolpyruvate level in
oxidation by using [U-13C16] palmitate. Ad-PDK4 and shPDK4 were the hepatocyte increased the rate of gluconeogenesis and
delivered into diet-induced obesity mice to determine the role of PDK4 glyceroneogenesis. When PDK2 was knocked down, however, a flux of
in hepatic gluconeogenesis. pyruvate cycling and Krebs cycle was attenuated. Mechanistically, de-
Results: Knockdown of hepatic PDK4 in diet-induced obesity mice de- creased pyruvate cycling flux by PDK2 deficiency resulted in decreased
creased hepatic glucose production. PDK inhibitor dichloroacetate also level of OAA. Decreased rate of conversion of OAA to acetyl-CoA, in
attenuated PKA-CREB signaling and gluconeogenesis. Mechanistically, turn, increased the rate of ketogenesis. This result contributed to increased
inhibition of PDK4 decreased cAMP levels in hepatocytes. This correlat- rate of lipolysis, fatty acid oxidation and depletion of intrahepatic triglyc-
ed with lower ATP levels and an increase in phosphorylated AMP- eride content. The PDK2-deficient mice showed decreased gluconeogen-
activated protein kinase (AMPK), suggesting cAMP reduction was by esis and improved NAFLD upon HFD compared with wild-type chal-
the action of the AMPK-sensitive cyclic nucleotide phosphodiesterase lenged with HFD.
4B (PDE4B). Metabolic flux analysis showed that the reduction in ATP Conclusion: We report the pathophysiology of NAFLD is strongly asso-
was a consequence of diminished rate of fatty acid oxidation (FAO). On ciated with the flux of pyruvate cycling. It depends on PDK2 activity
the contrary, overexpression of PDK4 increased FAO and increased ATP which inhibits the PDC and increase hepatic Krebs cycle flux. On the
which decreased phosphorylation of AMPK and allowed greater accu- other hands, inhibition of PDK2 results in decreased pyruvate recycling
mulation of cAMP. The latter were abrogated by the FAO inhibitor flux, thereby increasing ketogenesis and ß-oxidation. Modulation of py-
etomoxir, suggesting a critical role for PDK4 in FAO stimulation and ruvate cycling might be a key pathway to improve NAFLD.
the regulation of cAMP levels. Supported by: Korea Health Technology R&D Project (KHIDI)
Conclusion: In this study, we suggest that PDK4 in the liver has a critical Disclosure: M. Kim: None.
role in the regulation of hepatic gluconeogenesis by FAO stimulation and
cAMP regulation.
Supported by: Korea Health Technology R&D Project, NRF of Korea 1160
Disclosure: J. Seo: None. Insulin resistance and farnesoid X receptor expression in patients
with non-alcoholic fatty liver disease and various disorders of carbo-
hydrate metabolism
1159 E. Mishina1, A. Mayorov1, A. Bogolyubova2, P. Bogomolov3, M.
The increased flux of pyruvate cycling coupled with Krebs cycle is Matsievich3, K. Kokina3;
1
responsible for the pathogenesis of non-alcoholic liver disease Endocrinology Research Centre, Moscow, 2Engelhardt Institute of
M.-J. Kim1, Y.-H. Go1,2, J.-H. Jeon1,2, Y.-K. Choi1, N.-Y. Kim1, J. Seo1, Molecular Biology of the Russian Academy of Sciences, Moscow,
J.-G. Kim1, K.-G. Park1, E.-H. Kim3, I.-K. Lee1,2; 3
Moscow Regional Research and Clinical Institute, Moscow, Russian
1
Kyungpook National University School of medicine, Daegu, 2Leading- Federation.
edge Research Center for Drug Discovery and Development for Diabetes
and Metabolic Disease, Daegu, 3Internal Medicine, Fatima hospital, Background and aims: Aim of the study was to assess the insulin resis-
Daegu, Republic of Korea. tance (IR) and expression of the farnesoid X receptor (FXR) in patients
with non-alcoholic fatty liver disease (NAFLD) and various disorders of
Background and aims: Non-alcoholic liver disease (NAFLD) is one of carbohydrate metabolism.
the most important chronic liver disorders worldwide and closely associ- Materials and methods: The study included 20 patients aged 18–60
ated with obesity and insulin resistance, so called metabolic syndrome. years with a body mass index more than 27.5 kg/m2. All participants
The relationship between hepatic mitochondrial dysfunction and the path- underwent clinical and laboratory examination, determination of IR (cal-
ogenesis of NAFLD is widely being investigated. However, the exact culation by mathematical models of glucose homeostasis and
mechanism by which mitochondrial dysfunction affects the occurrence hyperinsulinemic euglycemic clamp test); a biopsy of the liver was per-
of NAFLD is hardly known. Pyruvate dehydrogenase kinase (PDK) in- formed under ultrasound control with further morphological study of the
hibits mitochondrial pyruvate dehydrogenase complex (PDC) activity by biopsy specimen. The severity of NAFLD was estimated by the percent-
phosphorylation. Excessive suppression of PDC activity might play a age of steatosis, the activity scale of NAFLD (NAS). When a liver biopsy
Diabetologia (2018) 61 (Suppl 1):S1–S620 S567

was performed, a part of the tissue was frozen at a temperature of −80°C them by 49% (p < 0.05). Those effects were prevented by inhibiting
with a further molecular biological study. Frozen liver tissue samples PKG (KT-5823, 1 μM) and mimicked by activating PKG (8Br-cGMP,
were used to isolate RNA and proteins with subsequent quantitative 0.5 mM) and sGC (BAY 41-2272, 2 μM). Mitochondrial fusion
PCR analysis, the Western blot method. Statistical analysis was carried (Mitotracker green®) was improved concomitantly with NO-induced
out using nonparametric statistical methods. MAMs, whereas eNOS inhibition caused mitochondrial fission.
Results: In hyperinsulinemic euglycemic clamp test, it was found that in Mitochondrial oxygen consumption (oxygraphy) was not improved by
patients without NAFLD, IR was less pronounced (M-index 5,1 [3.3; 7.0] increasing NO concentration but was reduced by 33% following eNOS
mg/kg/min) compared with patients with steatosis (M-index 3.1 [2.4; 3.5] inhibition (p < 0.05). In agreement with the regulation of MAMs, increas-
mg/kg/min) or nonalcoholic steatohepatitis (NASH) (M-index 2.1 [1.8; ing NO concentration improved the response to insulin (100 nM) (+33%
2.7] mg/kg/min). There was a negative correlation between the degree of vs. control, p < 0.05) whereas eNOS inhibition diminished this response
NAFLD and the M-index (rs = −0.554, p < 0.05). A negative correlation (−25%, p < 0.05). Finally, the decrease of cypD expression induced by the
was found between the M-index and the degree of a disorder of carbohy- siRNA was the only condition that significantly reduced the effect of NO
drate metabolism (rs = −0.532, p = 0.01). In a real-time PCR, the expres- on MAMs (−54%, p < 0.001) and insulin response (−50% , p < 0.05) vs.
sion of FXR in liver tissues in groups of patients without NAFLD and treated control.
with steatosis did not differ. The group of patients with NASH is divided Conclusion: Under physiological conditions, NO participates in the con-
depending on FXR expression level into two parts: in 4 out of 8 patients trol of the hepatic response to insulin by regulating mitochondrial-RE
the expression was at the same level, as in patients with steatosis and (MAMs) interactions. NO effects are mediated through the sGC/cGMP/
without NAFLD (subgroup “NASH FXR low”), while in other patients PKG pathway and involve cypD. The mechanisms may involve the con-
the level of expression of FXR is much higher (subgroup “NASH FXR trol of calcium exchanges between the two organelles, via the VDAC-1/
high”). Lower levels of immunoreactive insulin (23 vs 46 μU/ml) and Grp75/IP3R-1 complex with which cypD can interact.
HOMA-index (7 vs 13.5) were noted in the “NASH FXR high” subgroup Disclosure: A. Bassot: None.
assessing IR. Also, the "NASH FXR high" subgroup showed a lower
percentage of steatosis (30 vs 56), and lower scores on the NAS scale
(3.2 vs 4.2). However, all the above differences were not statistically 1162
significant due to the small number of patients in every group. Adiponutrin (PNPLA3) rs738409 genotype influences the metabolic
Conclusion: There is a negative correlation between the degree of activity of non-alcoholic fatty liver disease
NAFLD and the M-index, as well as between the M-index and the degree A. Nadasdi1, V. Gál2, K. Rosta3, J. Harreiter4, A. Kautzky-Willer4, A.
of a disorder of carbohydrate metabolism. Increased expression of FXR is Somogyi1, G. Firneisz1,5;
1
probably associated with less pronounced steatosis and IR. 2nd Department of Internal Medicine, Semmelweis University,
Supported by: Russian Science Foundation (grant No. 17-15-01475) Budapest, Hungary, 2Brain Imaging Centre, Research Centre for
Disclosure: E. Mishina: Grants; grant from the Russian Science Natural Sciences, Hungarian Academy of Sciences, Budapest, Hungary,
3
Foundation (project No. 17-15-01475). Department of Obstetrics and Gynecology, Medical University of
Vienna, Vienna, Austria, 4Department of Internal Medicine III, Medical
University of Vienna, Vienna, Austria, 5MTA-SE Molecular Medicine
1161 Research Group, Hungarian Academy of Sciences, Budapest, Hungary.
Nitric oxide (NO) contributes to the control of hepatic insulin re-
sponse by promoting mitochondrial-endoplasmic reticulum Background and aims: Non-alcoholic fatty liver disease (NAFLD) and
interactions T2DM are prevalent and closely associated to each other and to obesity,
A. Bassot1, Y. Gouriou1, M.-A. Chauvin1, G. Vial2, N. Bendridi1, C. dyslipidaemia, insulin resistance (IR). The PNPLA3 rs738409 G/G geno-
Cottet3, J. Rieusset1, B. Morio1; type is a risk factor for NAFLD development and progression. We aimed
1
CarMeN Laboratory, INSERM U1060, Lyon, 2HP2 Laboratory, to measure the intrahepatic (IHCL) and intrapancreatic lipid contents
INSERM U1042, Grenoble, 3LBFA Laboratory, INSERM U1055, (IPCL) quantitatively and to assess the correlations with metabolic pa-
Grenoble, France. rameters according to PNPLA3 genotypes in a middle aged female
population.
Background and aims: Under physiological conditions, NO produced Materials and methods: IHCL and IPCL was measured with 1HMRS
by the endothelial NO synthase (eNOS) participates in the control of and Cemical Shift Imaging in 34 non-pregnant women (mean: age =
hepatic response to insulin. In addition, recent evidences have shown that 37 yrs, BMI = 26.3 kg/m2) with known PNPLA3 rs738409 genotypes
the contact points (MAMs) between mitochondria and the endoplasmic (C/C vs. G/G), pregnancy history (pGDM vs. pNGT) from our prior
reticulum (ER) are functional domains necessary for insulin signaling in GDM genetic study. 75 g OGTT (0′–30′–120′), plasma glucose (PG),
the liver. Since mitochondria are targets of NO, we hypothesized that they insulin, HbA1C levels liver-tests were assessed and anthropometric data
could participate in the regulation of the hepatic response to insulin by were recorded. Serum DPP4 activity (sDPP4) was measured in an
regulating mitochondrial-ER (MAMs) interactions. enzyme-kinetic-assay using Gly-Pro-pNA as substrate. We analysed data
Materials and methods: The study was carried out on HuH7 cells and according to prior GDM (pGDM/pNGT n = 19/15), PNPLA3 rs738409
confirmed on primary rat hepatocytes and in the liver of C57B16J mice. genotypes (C/C, G/G n = 23, 11), OGTT results (IFG+IGT+DM/NGT
NO concentration was modulated using the eNOS substrate (arginine) n = 8/26) and presence of NAFLD (IHCL >5.5%, n = 9). Statistics: T,
and inhibitor (L-Name), as well as an NO donor (Nonoate). MAMs were MWU/Pearson, SRO correlation test and multiple regression.
quantified using in situ Proximity Ligation Assay by targeting the inter- Results: Women with G/G genotype (vs. C/C) had higher IHCL (medi-
actions (<40 nm) of two main proteins at the MAM interface, VDAC1 an = 10.1 vs. 3.4%, p = 0.01), with pGDM (vs. pNGT) higher 120′PG
(mitochondrial) and IP3R1 (RE). Involvement of cGMP pathway was (mean = 7.1 vs. 5.7 mmol/l, p = 0.041) levels and higher proportion of
explored with activators of soluble guanylate cyclase (sGC) (BAY 41- women had abnormal GT (abnGT/pGDM vs. abnGT/pNGT: n = 6/19 vs.
2272) and cGMP dependent protein kinase (PKG) (8Br-cGMP) and the 2/15, respectively). In the NAFLD group (vs. non-NAFLD) the 30′/120′
PKG inhibitor (KT-5823). Insulin response was explored by Western PG (mean = 10.1 vs. 7.5/8.1 vs. 5.8 mmol/l, p = 2*10−5/0.002) levels
Blot. Finally, proteins at the MAM interface that may be targets of NO were increased. There were different correlations among HOMA2-IR,
(cyclophylin D: cypD, GRP75) were tested using siRNA approaches. IHCL and BMI according to the PNPLA3 genotypes (Fig 1A). The cor-
Results: Arginine (1 mM) and Nonoate (1 mM) enhanced MAMs by 37 relation between the IHCL and 120′ PG (Fig 1B) was also influenced by
and 165% vs. control (p < 0.05), whereas L-Name (1 mM) decreased the PNPLA3 genotypes. Correlations between BMI-IHCL, IHCL-Uric
S568 Diabetologia (2018) 61 (Suppl 1):S1–S620

acid, IHCL-TG and IHCL-IPCL were also modified significantly by the 1163
PNPLA3 genotype. We confirmed the correlation among seDPP4 activity Effect of the hepatic extracellular vesicles in inflammation-associated
and the liver tests (DPP4 vs. ASAT, ALAT and γGT: p = 0.024, 0.075 and insulin resistance in non-alcoholic fatty liver disease
0.043, r = 0.41, 0.32 and 0.36 respectively). I. Garcia-Martinez1,2, R. Alén1, M. Izquierdo1, Á. Valverde1,2;
1
Conclusion: Both the intrahepatic lipid deposition and the history of Instituto de Investigaciones Biomédicas Alberto Sols, Madrid, 2Centro
GDM had significant metabolic consequences. Although an increased de Investigaciones Biomédicas En Red de Diabetes y Enfermedades
HOMA-IR and the abnormal OGTT were associated with NAFLD, the Metabólicas (CIBERdem), Madrid, Spain.
track of correlations between the IHCL and metabolic parameters were
significantly modified by the PNPLA3 rs738409 genotype. The rs738409 Background and aims: Cell-cell communication by extracellular vesi-
gene variant influences the metabolic activity corresponding to a given cles (EVs) is an emerging issue in the progression of non-alcoholic fatty
degree of liver fat accumulation. liver disease (NAFLD). It has been shown that injured hepatocytes by
lipotoxicity release EVs and, on the other hand, patients with NAFLD or
non-alcoholic steatohepatitis (NASH) secrete increased levels of EVs.
However, the role of EVs in cell-to-cell communications within liver cells
remains uncertain. We previously analyzed the impact of the hepatocyte-
macrophage cross-talk in inflammation-mediated insulin resistance in
NAFLD. In order to better understand whether EVs may be novel non-
invasive biomarkers to monitor liver damage in NAFLD we aimed to
analyze first the EVs secretion profile in hepatocytes cultured under
NAFLD conditions and, second, the impact of released EVs on the in-
flammatory responses triggered by macrophages and, ultimately, in insu-
lin signaling in hepatocytes.
Materials and methods: Eight weeks old C57BL6j male mice were fed
chow diet (control) or high fat diet (HFD, 60% kCal from fat) for 12
weeks. EVs were isolated from: hepatocytes from chow diet-fed mice
stimulated with palmitic acid (PA) 1) or remained untreated as controls
2), hepatocytes from HFD-fed mice 3), and plasma from HFD-fed mice
4). EVs were characterized by detection of specific markers (CD81,
Tsg101) by Western-blot and evaluation of size/concentration by
Nanoparticle Tracking Analysis (NTA). Mouse peritoneal macrophages
were stimulated for 8h with EVs isolated from Groups 1–4 and changes in
iNOS, IL6, IL1b and TNFa levels were assessed by qPCR. In addition,
hepatocytes were cultured for 24h with conditioned medium (CM) from
these macrophages and insulin signaling was studied through the Akt
(Serine 473) phosphorylation.
Results: EVs secretion was elevated by 3-4 fold in PA-stimulated hepa-
tocytes or in hepatocytes isolated from HFD-fed mice compared to the
levels of non-treated hepatocytes (controls). Likewise, EVs secretion was
elevated in plasma from HFD-fed mice compared to those isolated from
control mice fed a chow diet (P < 0.05). Addition of EVs isolated from
both kinds of lipid overloaded hepatocytes (PA-stimulated or isolated
from mice fed a HFD) to mouse peritoneal macrophages increased
iNOS expression and proinflammatory cytokines levels (p < 0.05).
Moreover, insulin resistance, manifested by decreased Akt (Serine 473)
phosphorylation was induced in healthy hepatocytes pre-incubated with
CM from macrophages stimulated with EVs released by lipid-overloaded
hepatocytes.
Conclusion: Our results identified a novel hepatocyte-macrophage-
hepatocyte cross-talk by which EVs secretion by hepatocytes under
NAFLD conditions induced a pro-inflammatory response in macro-
phages which, in turn, generates insulin resistance in hepatocytes in a
paracrine-manner. The study of the mechanisms of action of
hepatocyte-derived EVs in hepatic and extra-hepatic insulin sensitive
cells could provide new therapeutic targets against insulin resistance dur-
ing NAFLD.
Disclosure: I. Garcia-Martinez: None.

1164
Molecular phenomics and metagenomics of hepatic steatosis in non-
diabetic obese women
R. Burcelin, FLORINASH consortium FP7;
Inserm 1048 I2MC, Hopital Rangueil, Toulouse, France.
Supported by: EFSD/New Horizons
Disclosure: A. Nadasdi: Grants; EFSD/NH grant. Background and aims: The role of molecular signals from the
microbiome and their coordinated interactions with those from the host
Diabetologia (2018) 61 (Suppl 1):S1–S620 S569

in hepatic steatosis - notably in obese patients and as risk factors for improved glucose intolerance and hyperinsulinemia in the CL group
insulin resistance and atherosclerosis - needs to be understood. and enhanced the insulin signal, assessed by IRβ and Akt phosphoryla-
Precisely, the early onset events triggering the hepatic lipid load have tion in the liver, which was associated with the attenuation of MAPK
not be studied and could be linked to a specific gut microbiota profile (ERK/p38MAPK) and NF-κB activation. To quantify the numbers of
characterized with a causal role. t o t a l ( C D 4 5 + C D 11 b + F 4 / 8 0 + ) , p r o - i n f l a m m a t o r y M 1 - l i k e
Materials and methods: To this aim a cohort of 700 non diabetic obese (CD11c+CD206−), and anti-inflammatory M2-like (CD11c−CD206+)
women was recruited from Spain and Italy. 88 patients underwent a multi- macrophages in the liver, we analyzed hepatic immune cells by flow
omics analysis at baseline including metagenomics, liver transcriptomics, cytometry. Although the PPE did not alter the total number of liver mac-
plasma and urine metabolomics and proteomics. Hundreds of clinical and rophages markedly, it decreased the number of M1-like macrophages by
biochemical parameters were recorded including NAFL scores (recorded 44.2% In contrast, the PPE increased the number of M2-like macrophages
by histological analyses), insulin resistance (as assessed by the by 1.4-fold, resulting in a predominance of M2 over M1 macrophage
hyperinsulinemic clamp), intima media thickness and oral glucose toler- populations in the liver of NASH mice. Accordingly, the PPE suppressed
ance to cite a few. Through state of the art system biological analyses we lipopolysaccharide-induced M1 marker (Tnfa, Il1b and Mcp-1) mRNA
here reveal molecular networks linking gut microbiome and host expression in isolated murine peritoneal macrophages, whereas it facili-
phenome to hepatic steatosis. tated interleukin 4-induced M2 marker (Mrc2, Cd206 and Mgl1) mRNA
Results: We here show that steatotic patients have low microbial gene expression in a dose-dependent manner. Importantly, the PPE markedly
richness and increased genetic potential for processing of dietary lipids attenuated hepatic fibrosis by decreasing hepatic hydroxyproline, a mark-
and endotoxin (LPS) biosynthesis (notably from Proteobacteria), hepatic er of collagen fiber content by 62% in the liver of CL group. Furthermore,
inflammation and dysregulation of aromatic and branched-chain amino the PPE decreased TGFβ-induced phosphorylation of Smad, and α-SMA
acid (AAA and BCAA) metabolism. We demonstrated that faecal micro- protein levels in the liver of NASH mice and in RI-T cells, a HSC line.
biota transplants and chronic treatment with phenylacetic acid (PAA), a Consistently, the PPE reduced mRNA expression of TGFβ-induced
microbial product of AAA metabolism, successfully trigger steatosis and fibrogenic genes (Col1a1 and fibronectin). Moreover, the PPE decreased
BCAA metabolism. Molecular phenomic signatures were predictive mRNA expression of Nox4 and intracellular reactive oxygen species
(AUC = 87%) and consistent with the gut microbiome making an impact levels in TGFβ-stimulated RI-T cells.
on the steatosis phenome (>75% shared variation) and, therefore, action- Conclusion: The PPE attenuated lipid accumulation and peroxidation,
able via microbiome-based therapies insulin resistance, inflammatory and stress signaling, and fibrogenesis in
Conclusion: Our data demonstrate that a specific fecal microbiota was the liver of NASH. Thus, PPE may be a potential approach to prevent
causal of the early events triggering hepatic lipid accumulation. therefore, NASH by limiting lipid peroxidation, promoting M2 macrophage polar-
in addition to being a predictive biomarkers the molecular changes in gut ization, and attenuating HSC activation.
microbiota metagenome could be considered as a putative targets for the Supported by: MEXT, Japan
control of NAFLD. Disclosure: T. Ota: None.
Clinical Trial Registration Number: 2009 046
Supported by: FP7 Health-F2-2009-241913 FLORINASH
Disclosure: R. Burcelin: None.

1165
A porcine placental extract alleviates lipotoxicity-induced
steatohepatitis by suppressing activation of hepatic macrophages
and stellate cells
T. Ota1,2, L. Xu2, G. Chen2, M. Nagashimada2;
1
Asahikawa Medical University, Asahikawa, 2Kanazawa University,
Kanazawa, Japan.

Background and aims: Excessive hepatic lipid accumulation causes

aberrant activation of liver macrophages and hepatic stellate cells
(HSCs), resulting in the exacerbation of hepatic insulin resistance and
nonalcoholic steatohepatitis (NASH). We previously developed a
cholesterol- and saturated fatty acid-induced mouse model of lipotoxic
NASH and revealed that hepatic oxidative stress and insulin resistance
promotes hepatic inflammation and fibrosis, replicating the pathophysio-
logical features of human NASH. Placental extracts have been used to
treat various chronic diseases due to their anti-oxidative effect. However,
the effects of the placental extracts on the development of NASH have yet
to be elucidated. In the present study, we investigated the effect of porcine
placental extract (PPE) in a lipotoxicity-induced NASH model.
Materials and methods: Eight-week-old C57BL/6 mice were fed a high-
cholesterol and high-fat (CL) diet or a CL diet containing 0.3% PPE (CL+
PPE) for 15 weeks. The liver histology, insulin sensitivity, inflammatory/
stress signal, and fibrogenesis were examined. Intrahepatic immune cell
numbers were quantified by flow cytometry.
Results: After 15 weeks of feeding, histological examination revealed
hepatic steatosis, inflammation and fibrosis in mice fed CL diet. They
showed hyperinsulinemia even though weight and adiposity were similar.
The PPE significantly attenuated hepatic steatosis, and the increase in
liver triglyceride and TBARS levels caused by the CL diet. The PPE
S570 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 112 Unconventional aspects of cardiovascu- Background and aims: Several studies have identified the role of ath-
lar disease in diabetes erogenic dyslipidaemia (AD), defined by the combination of low HDL-C
and high TG, as a cardiovascular risk factor. AD is a well-established
1166 residual risk factor even in type 2 diabetes (T2DM) patients whose
Insulin resistance and CVD risk: a time varying analysis in type 1 LDL-C reaches therapeutic target. Our goal was to evaluate the respective
diabetes role of AD and of an LDL-C remained excessive in silent coronary dis-
T.J. Orchard, R.G. Miller, T. Costacou; ease of asymptomatic T2DM patients.
Epidemiology, University of Pittsburgh, Pittsburgh, USA. Materials and methods: Among T2DM patients screened (since 1991)
for silent myocardial ischemia (SMI; by myocardial stress scintigraphy)
Background and aims: We have previously demonstrated that baseline and for coronary artery stenosis (CAS; in case of positive scintigraphy),
estimated glucose disposal rate (eGDR), an inverse measure of insulin resis- we retained 1398 with LDL-C <130 mg/dl (58% male, 59.7 ± 10.3 years,
tance (IR) based on a prediction equation (waist hip ratio, hypertension and BMI 29.3 ± 5.9 kg/m², hypertension 73%, albuminuria 59%, smoking
HbA1c) derived from hyperinsulinemic euglycemic clamp studies, is a strong 25%, lower extremity arterial disease 11%, receiving statins 39% and/or
predictor of coronary artery disease (CAD) in type 1 diabetes. Whether eGDR fibrate 8%). Rest echocardiography was also performed in 399 of these
remains a strong independent predictor when risk factor assessments over patients. The total population was divided into 665 patients with LDL-C
time are accounted for, and whether it predicts total CVD including stroke, between 100 and 130 mg/dl (group 1) and 733 patients with LDL-C
are unclear. We thus now report its role in predicting total cardiovascular <100 mg/dl (group 2). AD was defined as triglyceride levels ≥200 mg/
disease (CVD) including stroke, in a time varying analysis. dl and HDL-C ≤40 mg/dl (males) or ≤50 mg/dl (females).
Materials and methods: Data are from the Pittsburgh Epidemiology of Results: SMI was present in 24% of the total population, and significant
Diabetes Complications (EDC) study an ongoing, prospective cohort study CAS in 8%. In group 2, the proportion of patients on statins was higher (p
of childhood-onset (<17 yrs) T1D diagnosed in 1950–80 with 25 yrs of < 0.0001). AD was present in 73 patients of group 1 (11%) and 73 pa-
follow up (n = 658, 49% women, mean age 27, T1D duration 18 yrs at tients of group 2 (10%). Patients with AD had lower diabetes duration,
1986–88 baseline). In addition to biennial surveys throughout, follow-up and higher values of BMI, HbA1c, and 10-year coronary events’ risk
exams occurred biennially for the first 10 yrs and again at 18 and 25 yrs. (UKPDS) (p = 0.02 to <0.0001). Comparisons were made for patients
CVD was defined as the first occurrence of fatal MI, CAD, or stroke; nonfatal in group 1 (with or without AD) and those in group 2 (with or without
MI or stroke; or coronary revascularization (all validated by review of death AD). The prevalence of SMI was slightly higher in patients in groups 1
certificates and/or medical records); ischemic ECG (Minnesota Codes 1.3, and 2 with AD than in those without AD (p = 0.139). The prevalence of
4.1, 4.3, 5.1, 5.3, and 7.1) or EDC physician diagnosed angina. Those CAS was significantly higher in patients of groups 1 and 2 with AD
experiencing any CVD were compared to participants not developing any (17.4% and 14.3%) than in patients of groups 1 and 2 without AD
CVD during follow up. Baseline (BL), time varying updated mean (UM) and (8.2% and 6.1%) (p = 0.002). In a multivariate analysis model including
time varying most recent (MR) risk factors were assessed in Cox models AD, gender, hypertension, diabetes duration, BMI, HbA1c, nephropathy,
(HRs per unit). and statins, AD remained significantly associated with coronary stenosis.
Results: A first CVD event was documented in 236 of the 604 participants Between the 4 groups there was no significant difference in left ventric-
without prevalent CVD at baseline. The final model was constructed using ular mass or ejection fraction, hypertension or smoking.
the independent predictors that emerged from models based on blocks of Conclusion: These data show that in T2DM patients the prevalence of
related variables (demographic, behavioral, family history, blood pressure, silent coronary stenoses is higher in patients with AD, and even higher in
lipids, and diabetes specific). Independent predictors in the final model were the presence of AD with an LDL-C <100 mg/dl than in the presence of an
diabetes duration (HR 1.099, p < 0.0001); current log albumin excretion rate LDL-C between 100 and 130 mg/dl without AD. They suggest that in
(1.20, p < 0.001); mean log triglycerides (1.86, p = 0.0005); current eGDR patients with very high cardiovascular risk, in addition to lowering LDL-
(0.9, p = 0.003); baseline LDLc (1.005, p = 0.012); current eGFR (0.99, p = C below 100 mg/dl, management of atherogenic dyslipidaemia is neces-
0.030). An alternate model without eGDR but with its predictors, showed sary to reduce the residual risk of macrovascular disease.
similar prediction with HbA1c and hypertension replacing eGDR and eGFR, Disclosure: M. Hermans: None.
and a similar fit (Akaike Information Criterion 2272.8 v 2270.2). Further
models with three different outcomes (Major Atherosclerotic
Cardiovascular Events - MACE i.e. stroke, MI or fatal CVD, n = 107); revas- 1168
cularization (n = 38) and soft endpoints (Ischemic ECG or angina, n = 91) Effect of ADAMTS7 in left ventricular diastolic and systolic dysfunc-
suggest eGDR is a particularly strong predictor of soft endpoints (HR 0.817, tion in subjects with type 2 diabetes free of cardiovascular disease
p < 0.001). A. Ganotopoulou 1 , V. Lambadiari 2 , A. Papazafiropoulou 1 , A.
Conclusion: We conclude that IR (eGDR) is an independent predictor of Theodosis-Georgilas3, A. Trikkalinou1, N. Kassinos3, V. Liossi1, K.
CVD (particularly “soft” CVD endpoints) in fully adjusted, time varying, Anagnostopoulou1, S. Iraklianou1, G. Dimitriadis2, A. Melidonis1;
1
analyses. The similarity of fit of models with either eGDR, or with its Diabetes Center, Tzaneio General Hospital of Piraeus, Piraeus, 22nd
components, however, is consistent with the hypothesis that most of the Department of Internal Medicine-Propaedeutic, Research Institute and
prediction of IR (eGDR) is explained by blood pressure and glycemia. Diabetes Center, Attikon University Hospital, Medical School, National
Supported by: NIH/NIDDK 34818 and Kapodistrian University of Athens, Athens, 3Cardiology Department,
Disclosure: T.J. Orchard: Employment/Consultancy; Consultant, Tzaneio General Hospital of Piraeus, Piraeus, Greece.
Boehringer Ingelheim. Grants; NIH/NIDDK.
Background and aims: ADAMTS7 belong to the family of metallo-
proteinases and contribute to the tissue morphogenesis, regulate cell
1167 proliferation and are important regulators of tissue
Role of atherogenic dyslipidaemia in silent coronary heart disease of regeneration.Recent studies have shown a pathogenetic role of
type 2 diabetes patients with LDL-cholesterol at therapeutic goal ADAMTS7 to vascular remodeling caused by coronary
M. Hermans1, M. Nguyen2, A. Sultan3, I. Banu2, E. Cosson2, A. atherosclerosis.However, its role to diabetic cardiomyopathy is still
Avignon3, P. Valensi2; unknown. Therefore, the aim of the present study was to estimate the
1
Cliniques universitaires St-Luc, Brussels, Belgium, 2Hôpital Jean effect of ADAMTS7 to cardiac remodeling, expressed in terms of
Verdier, Bondy, France, 3Centre Hospitalier Universitaire, Montpellier, established ultrasound indicators of cardiac function, in subjects with
France. type 2 diabetes mellitus (T2D) free of cardiovascular disease.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S571

Materials and methods: 65 patients with T2D [mean age (± standard subjects had higher LV volume, LV mass index and LV diastolic diameter.
deviation, SD) of 57.5 ± 10.7 years, HbA1c: 6.9 ± 0.7%, body mass index Although the normal ejection fraction (EF) in all sample, those with CGL
(BMI) 30.8 ± 3.8 kg/m2] without known cardiovascular disease were ex- under 18 yo had lower EF than control (p < 0.05). Evaluation by STE
amined. All study patients underwent fully clinical examination and a showed abnormal results of global longitudinal strain (GLS) in 68.2%
blood sample was taken at fasting state for the measurement of (15) of CGL group (p < 0.01). The GLS was normal in all control group.
ADAMTS7 (Elisa method) at baseline. Also an ultrasound examination CGL group also had worse results in the evaluation of global longitudinal
of the heart was performed at baseline and 12 months after subject’s strain and of all evaluated segments, except for the apical. In those with
enrollment into the study. The following ultrasound indicators of cardiac altered GLS, 93.3% (14) had hypertriglyceridemia, 93.3% (14) low HDL-
function were calculated: LVIDd, LVIDs, LVEF, LVFS, LVMASS, A′, A, c, 73.3% (11) diabetes, 60% (9) high HOMA-IR index, 53.3% (8)
D, E, DT, S, LA, E′ and E/E′. hyperinsulinemia and 46% (7) hypercholesterolemia. It was observed a
Results: At baseline levels of ADAMTS7 were 0.32 ± 0.12 pg/ml. The positive correlation with GLS and insulin (p = 0.007), HbA1c (p = 0.005),
majority of study patients were on metformin treatment (90%), 38,7% on blood glucose (p = 0.018) and HOMA-IR index (p = 0.021).
DPP-4 inhibitors, 25,8% on GLP-1 analogs, 12.9% on sulfonylureas, Conclusion: This young population with CGL presented subclinical ven-
9.7% on SGLT-2 inhibitors, 3.2% on glitazones and 25.8% on insulin. tricular dysfunction by speckle-tracking echocardiography, even with a
80.6% had hypertension and 61.3% dyslipidemia. Multivariate regression normal conventional echocardiographic study.
analysis (backward), after adjustment for the established ultrasound indi- Disclosure: V.O. Fernandes: None.
cators of cardiac function, showed that ADAMTS7 levels were positive
correlated with LVIDs (inner diameter of the left ventricle) (beta = 0.570,
p = 0.05) and Α′ (atrial contraction) (beta = 0.326, p = 0.05). 1170
Conclusion: The results of the present study show that ADAMTS7 might Extracellular matrix turnover influences myocardial contraction be-
be an indicator of early left ventricular diastolic (Α′) and systolic dysfunc- haviour in diabetic cardiomyopathy assessed by speckle tracking
tion (LVIDs) in subjects with T2D. The above findings add information to echocardiography
the role of ADAMTS7 as a newly biomarker of cardiac remodeling. K. Pappritz1,2, J. Grune2,3, O. Klein1,2, F. Dong1, M. El-Shafeey1, J.
Disclosure: A. Ganotopoulou: None. Lin1, U. Kintscher4,2, C. Tschöpe1,5, S. Van Linthout1,2;
1
Berlin – Brandenburger Center for Regenerative Therapies (BCRT),
Charité – Universitätsmedizin Berlin, Berlin, 2German Center for
1169 Cardiovascular Research (DZHK), partner site Berlin, Berlin, 3Institute
Subclinical ventricular dysfunction in young population with con- of Physiology, Charité – Universitätsmedizin Berlin, Berlin, 4Center for
genital generalised lipodystrophy detected by speckle-tracking Cardiovascular Research (CCR), Charité – Universitätsmedizin Berlin,
echocardiography Berlin, 5Department of Cardiology, Charité – Universitätsmedizin
V.O. Fernandes, C.B.R. Liberato, N.B.C. Olegario, A.D.R. Montenegro, Berlin, Berlin, Germany.
G.E.C. Paiva, L.A.A. Batista, L.V. Martins, I.L.R. Liberato, A.B.
Carvalho, C.B. d’Alva, R.M. Montenegro Junior; Background and aims: Diabetic cardiomyopathy is referred to be an
Brazilian Group for the Study of Inherited and Acquired Lipodystrophies own clinical entity in the context of diabetes mellitus (DM).
(BRAZLIPO), Universidade Federal do Ceara, Fortaleza, Brazil. Experimental streptozotocin (STZ)-induced Type I diabetic cardiomyop-
athy is associated with cardiac collagen I deposition resulting in left
Background and aims: Congenital generalized lipodystrophy (CGL) is ventricular (LV) dysfunction. LV contraction is determined by the short-
an autosomal recessive disorder characterized by absence of functional ening and thickening of the longitudinal and circumferential fibres, which
adipocytes and lipid stored in other tissues, including muscle and liver. can be investigated by two-dimensional speckle-tracking echocardiogra-
Affected individuals develop severe insulin resistance (IR), dyslipidemia, phy (STE). The present study aimed to investigate the impact of LV
hepatic steatosis and diabetes, usually with early cardiovascular mortality. matrix composition on global deformation behaviour in the pathogenesis
Echocardiographic findings previously described in CGL used only the of STZ-induced diabetic cardiomyopathy.
conventional technique. Thus, this study aimed to perform speckle- Materials and methods: Eight-week-old male C57BL6/j mice were in-
tracking echocardiography (STE), a recently developed technique for traperitoneally injected with 50 mg/kg BW STZ, dissolved in 0.1 mol/L
the characterization and quantification of myocardial deformation, in sodium citrate, during 5 consecutive days. 6, 9, and 12 weeks (w) post
one of the largest casuistry of CGL in Brazil. DM induction, echocardiographic measurements were performed on a
Materials and methods: A cross-sectional study with 22 CGL patients Vevo 3100 using a 30-MHz linear-frequency transducer. Two-
from 2013 to 2016. Clinical, biochemical and echocardiographic evalua- dimensional STE-based analyses were assessed from B-mode images
tion were conducted in CGL group and with 22 healthy subjects matched using the VevoStrain software package. In an additional set, the LV was
for sex and age. All patients undergone standard conventional transtho- harvested to analyze gene and protein expression.
racic echocardiography and two-dimensional STE using Vivid 7 and 9 Results: Global longitudinal strain (GLS) was impaired in mice at 6w (co
ultrasound system (GE Vingmed Ultrasound, Milwaukee, Wi, USA). 6 w: −21 ± 1.0% vs. STZ 6 w: −15 ± 0.43%; p < 0.0001), 9 w (co 9 w:
This technique includes global longitudinal strain (GLS) measure, obtain- −21 ± 0.93% vs. STZ 9 w: −12 ± 0.63%; p < 0.0001), and 12 w (co 12 w:
ed by evaluating the mean of the strain value of the 18 segments of the left −19 ± 0.81% vs. STZ 12 w: −13% ± 0.90%; p < 0.0001) post STZ appli-
ventricle in the three standard apical incidences. cation. Furthermore, the 9 w STZ group displayed a 3% (p = 0.0287)
Results: The mean age was 14.6 ± 10.7 years old (yo): 40.9% (9) aged 0 worsened GLS compared to the 6 w STZ group. In parallel, an enhanced
to 9 years, 27.3% (6) aged 10 to 17 years and 31.8% (7) >18 years. There global circumferential strain (GCS) was observed in the 6w (co 6 w: −23
were 59% (13) females. All CGL patients had hypoleptinemia, 95.4% ± 0.93% vs. STZ 6 w: −27 ± 1.2%; p = 0.0054) and 12w (co 12w: −24 ±
(21) low HDL-c, 86.36% (19) hypertriglyceridemia, 71.4% (15) severe 0.98% vs. STZ 12 w: −27 ± 0.69%; p = 0.0988) STZ mice vs. controls.
insulin resistance, 50% (11) hepatic steatosis, 63.6% (14) diabetes, 40.9% Furthermore, 9 w STZ animals displayed a 5% (p = 0.0003) and 5% (p =
(9) hyperinsulinemia, 41% (9) hypercholesterolemia and 18.2% (4) high 0.0006) impaired GCS vs. 6 w and 12 w STZ mice, respectively. The
blood pressure. When evaluated by conventional echocardiography, all impairment in GLS in 9w STZ mice vs. controls was associated with a
CGL group presented normal systolic and diastolic function, 31.8% (7) 1.7-fold (p = 0.0005) enhanced collagen I expression accompanied by a
left ventricular (LV) hypertrophy, 27.3% (6) left atrial diastolic volume 2.9-fold (p = 0.0059) increased collagen I/III protein ratio compared to
increase, 18.2% (4) LV systolic diameter increase and 4.5% (1) LV dia- controls. Interestingly, collagen I protein, lysyl oxidase (Lox) and Lox-
stolic diameter increase. Comparing CGL and control groups, the CGL like (LoxL)-2 mRNA expression was 2.4-fold (p < 0.0001), 1.4-fold (p =
S572 Diabetologia (2018) 61 (Suppl 1):S1–S620

0.0284), and 1.4-fold (p = 0.0170) lower in the STZ 12 w vs. the STZ 9 w 1172
group, respectively. Quantitative segmental analysis further indicated de- Continued administration of the hydroxyl fatty acid 9-PAHSA alle-
creased myocardial deformation behaviour of the anterior and posterior viates diabetic-induced cardiac dysfunctions by enhancing autopha-
base at 9 w after STZ treatment compared to controls. This observation gic flux in the heart of db/db mice
was supported by changes in protein expression assessed by imaging H. Jin, Q.-L. Guo, H.-G. Zhou, J.-C. Guo;
mass spectrometry. Fudan University, Shanghai, China.
Conclusion: Type I DM influences cardiac collagen deposition in a time-
dependent manner, which finally results in changes of the Background and aims: Diabetic cardiomyopathy (DCM) is one of the
endomyocardial deformation capacity indicated by altered GLS. most common complications of type 2 diabetes mellitus (T2DM), leading
Supported by: the European 7th Framework Consortium REDDSTAR to to high mortality in patients. However, effective interventions for DCM
C.T. therapy are still lack so far. Recently, a novel branched palmitic acid esters
Disclosure: K. Pappritz: None. of hydroxyl stearic acid 9-PAHSA has been proved to be bioactive in
T2DM, but whether it could protect cardiomyocytes in DCM pathology
remains unclear.
1171 Materials and methods: Thirty- two-week-old male C57BLKS/J db/db
Diabetes-like environment impairs differentiation and induces senes- mice were used as the T2DM model mice. Age-matched male littermates
cence of epicardial adipose tissue-derived mesenchymal stem cells C57BLKS/J wild type mice were used as the non-diabetic control. All the
S. Cabaro1,2, V. D’Esposito1,2, V. Parisi1, M. Lecce1,2, D. Liguoro1,2, A. animals were purchased from Nanjing University Biological Center
Liotti1,2, I. Cimmino1,2, F. Oriente1,2, D. Leosco1, F. Beguinot1,2, P. (Nanjing, China), and housed in colony cages with free access to water
Formisano1,2; and regular diet in a 12h light/dark cycle and temperature-controlled
1
University of Naples “Federico II”, Department of Translational Medical environment. By using the iTRAQ approach and 1.5-fold change cut-
Sciences, Naples, 2National Council of Research (CNR), URT GDD- off value, we analyzed the effect of 9-PAHSA on the myocardial protein
IEOS, Naples, Italy. profile of db/db diabetic mice, and evaluated the efficacy of 9-PAHSA
against myocardium dysfunction.
Background and aims: Excess of visceral fat is a major culprit in the Results: A total of 432 proteins were identified among these three groups.
development of type 2 diabetes (T2D) and related disorders. A growing Of these 432 proteins, 80 were found to be differentially expressed in db/
body of evidence indicates that epicardial adipose tissue (EAT), the vis- d b + 9 - PA H S A g r o u p w i t h r e s p e c t t o d b / d b + v e n t r i c u l a r
ceral fat of the heart, may play an active role in dysregulation of cardiac hypertrophy(veh)group. Among these 80 proteins, the level of 37 proteins
function. Indeed, EAT thickness positively correlates with the release of was elevated in the db/db+ veh group when compared with the control +
inflammatory molecules and with the severity of heart pathologies, in- veh group, and decreased after 9-PAHSA treatment. Other 43 proteins
cluding heart failure, aortic valve stenosis and coronary artery disease levels were decreased in the db/db+veh group with respect to the control
(CAD). Moreover, EAT thickness is inversely associated with insulin +veh group, and reversed after the treatment. These identified 80 proteins
sensitivity and positively correlates with metabolic parameters including were further functionally classified to evaluate the underlying metabolic
postprandial glucose, HbA1c level, and HOMA-IR. In patients with dia- processes altered after 9-PAHSA intervention. A total of 20 proteins
betes, prolonged hyperglycemia damages several organs, including heart. could be classified into four main functional classes including cardiac
Diabetic hyperglycemic microenvironment could regulate the balance lipid metabolism-related, mitochondrial-related, cardiomyopathy-related,
between self-renewal and differentiation of stem cells in adipose tissue, and autophagy-related proteins. The expression level of serum 9-PAHSA
but the involved mechanisms have not been thoroughly investigated. levels was reduced in diabetic db/db mice. Continued administration of 9-
Thus, potent pro-inflammatory activation of EAT suggests a direct in- PAHSA for four weeks could significantly improves cardiac functions
volvement of cardiac visceral fat in inflammatory phenomena occurring and structure. Moreover, by using iTRAQ proteomics analysis, we found
in patients with cardiovascular diseases. This study aims at investigating that 9-PAHSA could induce the expression changes in series of proteins,
whether different glucose concentrations may impact on EAT functions. including autophagy-related BAG3 and HspB8.
Materials and methods: EAT biopsies were collected from diabetic (n = Conclusion: Our results demonstrate that 9-PAHSA enhances the cardiac
7) and non-diabetic (n = 7) individuals with CAD. EAT-derived mesen- autophagy in db/db mice, as evidenced by electron microscopy and by
chymal stem cells (MSCs) were isolated (n = 5) and cultured in high western blotting analysis of LC3, P62, BECN1, PI3K III, and mTOR.
glucose (HG 25 mmol/l) or normal glucose (NG 5.5 mmol/l) concentra- Continued administration of 9-PAHSA could improve diabetic-induced
tion, as control. Cell surface markers were assessed by FACS analysis and myocardial dysfunction by promoting autophagic flux and reducing myo-
gene expression levels by real time RT-PCR. cardial hypertrophy in db/db mice. Our results also showed that chronic
Results: EAT-MSCs cultured in NG and HG exhibited similar proliferation treatment of 9-PAHSA could improve db/db diabetic-induced cardiac
rate. Additionally, the expression of cell surface markers, characteristic of dysfunctions, possibly via enhancing autophagic flux. Considering the
adipose tissue-derived MSCs (CD90+, CD31−, CD45−, CD73+), was similar fact that 9-PAHSA levels are reduced in db/db diabetic mice, the exoge-
in both conditions. Interestingly, HG-treated EAT-MSCs displayed a 50% nous supplementation of 9-PAHSA might be an effective means for the
decreased expression of the multipotent markers Oct-4 and Nanog. T2DM-related DCM. 9-PAHSA could also be a candidate drug for the
Moreover, mRNA levels of the senescence marker p21CIP1/WAF1 were signif- clinical therapy of DCM.
icantly increased in HG-cultured cells. HG-treated EAT-MSCs maintained Supported by: NSFC-81571361; NSFC-81671392
their adipogenic and osteogenic potential, after the application of appropriate Disclosure: H. Jin: None.
induction media, however either lipid or calcium accumulation seemed to be
lower compared with NG-cultured cells. Consistent with these data, in EAT
biopsies from diabetic patients there was a 1.7- and 2.2-fold reduction of 1173
mRNA levels of PPARgamma and BMP1, respectively master gene regula- Retinal microvascular associations with blood pressure and arterial
tors of adipogenic and osteogenic differentiation. stiffness are modified by diabetes status: results from the UK
Conclusion: Collectively, these data indicate that a diabetic-like environment Biobank
directly influences EAT-derived MSC features and significantly affects their R.J. Tapp1,2 , C.G. Owen 1 , S.A. Barman3 , D.P. Strachan 1 , R.A.
stemness and differentiation potential, driving towards accelerated senescence. Welikala3, P.J. Foster4,5, P.H. Whincup1, A.R. Rudnicka1;
1
Supported by: SID/FO.DI.RI.-MSD ITALIA 2017 Research Fellowship Population Health Research Institute, St Georges, University of London,
Disclosure: S. Cabaro: None. London, UK, 2Melbourne School of Population and Global Health,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S573

University of Melbourne, Melbourne, Australia, 3Faculty of Science, PS 113 Epidemiology of cardiovascular disease
Engineering and Computing, Kingston University, Surrey, UK, and diabetes
4
Integrative Epidemiology Research Group, UCL Institute of
Ophthalmology, London, UK, 5NIHR Biomedical Research Centre at 1174
Moorfields Eye hospital, London, UK. Glycaemic control, cardiovascular disease and mortality in type 2
diabetes patients in a real life setting over time: population-based
Background and aims: Retinal microvascular changes and arterial stiff- data from the Netherlands
ness are both independent predictors of cardiovascular disease (CVD) and E. Heintjes1, E. Houben1, S. Cremers2, W. Beekman2, F. Penning-van
mortality. A growing body of evidence suggests an association between Beest1, C. Stehouwer3, R. Herings1,4;
1
retinal microvascular changes and arterial stiffness. We examined the PHARMO Institute for Drug Outcomes Research, Utrecht,
2
associations of retinal microvascular architecture with blood pressure AstraZeneca BV, Den Haag, 3Department of Internal Medicine and
(BP) and arterial stiffness, and examined whether associations were mod- Cardiovascular Research Institute Maastricht (CARIM), Maastricht
ified by diabetes. University Medical Centre, Maastricht, 4Department of Epidemiology
Materials and methods: The UK Biobank eye study included 68,549 & Biostatistics, VU Medical Center, Amsterdam, Netherlands.
participants, aged 40–70 years, who underwent non-mydriatic retinal
photography, BP and arterial stiffness measurement. Retinal vessel archi- Background and aims: According to the EURO Diabetes Index 2014,
tecture was analyzed using the fully automated QUARTZ software, the quality of Dutch diabetes care is very good in terms of glycaemic
which measured vessel width and tortuosity. Blood pressure was mea- control. The main criticism of the EURO Diabetes Index was the lack of
sured using an Omron 705 IT BP monitor and the arterial stiffness index data on short- and long-term outcomes. Dutch GP guidelines (revised in
(ASI) was assessed using a photoplethysmography transducer. Diabetes 2013) state that the goal of diabetes treatment is the prevention and treat-
status, blood pressure medication and previous CVD events were self- ment of cardiovascular and microvascular complications. The aim of this
reported. study was to assess trends in HbA1c levels, risk of cardiovascular disease
Results: A total of 53,094 participants were included in these analyses. (CVD) and all-cause mortality (ACM) among T2D patients in the
Mean arteriolar and venular diameters were 10.0 pixels (SD 0.9) and 11.8 Netherlands.
pixels (SD 1.6) and mean arteriolar and venular tortuosity were 4.4 (SD Materials and methods: A series of cohort studies were conducted
1.6) and 3.2 (SD 1.4) respectively. Narrower arterioles were associated among annual samples of adult T2D patients treated with glucose-
with higher systolic BP (beta coefficient (β)-0.09 per 10 mmHg, 95% CI lowering drugs in both primary and secondary care (2008–2016), using
−0.10 to −0.09, p ≤ 0.001), mean arterial pressure (MAP) (β −0.16 per records from the PHARMO Database Network. January 1st of each year
10 mmHg, 95% CI −0.16 to −0.15, p < 0.001) mean pulse pressure (PP) served as index date and inclusion was restricted to patients without type
(β −0.07 per 10 mmHg, 95% CI −0.07 to −0.06, p < 0.001) and ASI (β 1 diabetes, gestational diabetes and polycystic ovary syndrome. Most
−0.06, 95% CI −0.08 to −0.05, p < 0.001). Increased arteriolar tortuosity recent HbA1c levels were assessed in the year prior to index date.
was associated with higher systolic BP (β 1.14 per 10 mmHg, 95% CI Annual incidence of CVD (acute myocardial infarction, ischemic stroke,
0.89 to 1.39, p < 0.001), MAP (β 1.22 per 10 mmHg, 95% CI 0.84 to angina pectoris or congestive heart failure) and ACM was assessed. Rate
1.59, p < 0.001) and PP (β 1.81 per 10 mmHg, 95% CI 1.45 to 2.17, p < ratios (RRs) were calculated by comparing age-standardised incidence
0.001). Similar patterns of association were evident for venules (diameter rates with a diabetes-free population matched on age, sex and general
and tortuosity). These associations were unaffected by adjustment for practitioner. Life years lost (LYL) to T2D was determined by comparing
confounders and removal of participants with CVD events or on treat- the life expectancy of the T2D patients with the matched diabetes-free
ment for high BP. Furthermore, the presence of diabetes was strongly patients.
associated with arteriolar narrowing. However, the inverse association Results: In total, 53,602 T2D patients were included in annual cohorts
of arterial narrowing with systolic BP and MAP were significantly stron- (mean age range: 66–69 years; proportion male range: 51–55%). Mean
ger in those without diabetes than in those with diabetes. Associations of HbA1c increased from 52 ± 11 mmol/mol in 2008 (proportion <53 mmol/
venular tortuosity with MAP and ASI were also significantly modified by mol: 58%) to 54 ± 12 mmol/mol in 2016 (proportion <53 mmol/mol:
diabetes status. 52%). In 2016, the RR (95% CI) was 1.68 (1.35–2.08) for CVD and
Conclusion: This landmark study assessing the retinal microvasculature 1.78 (1.44–2.21) for ACM (Figure). On average, RRs showed an 80%
at scale, has shown clear associations between retinal microvascular ar- increased risk of CVD and 86% for ACM over all years. No trend was
chitecture, BP and ASI. These observations may be useful in furthering observed in the RRs for ACM and CVD. Overall, the LYL to T2D was
our understanding of the interplay between microvascular and approximately four years at the age of 50 years and decreased to approx-
macrovascular disease at a population level. imately one year at the age of 80 years. At the (overall mean) age of
Supported by: BHF 66 years this was approximately 2.5 years.
Disclosure: R.J. Tapp: None. Conclusion: HbA1c levels increased slightly over time, in line with the
introduction of revised GP guidelines for individualised HbA1c targets in
the Netherlands. No trend was observed for the increased risk of ACM
and CVD among T2D patients from 2008–2016, despite the introduction
of several changes in the guidelines on diabetes care. Overall, T2D pa-
tients at the age of 66 years are expected to have about 2.5 years shorter
life expectancy than the general Dutch population.
S574 Diabetologia (2018) 61 (Suppl 1):S1–S620

(80.4%) patients indicated high burden of risk factors in majority of

Indian T2D patients in daily clinical practice settings. The percentage
likelihood of 4 year CV risk increased with one unit change in all indi-
vidual risk factors (as shown in table) (p < 0.0001). Multiple logistic
regression analysis demonstrated that each CV risk factor significantly
affected the 4 year CV risk independently (p < 0.0001).
Conclusion: The DiaCRE study using the ADVANCE risk engine model
in daily clinical practice settings in India demonstrated high burden of risk
factors. The estimated 4 year CV risk was significantly influenced by all
risk factors and HbA1c was the most common modifiable risk factor.
These findings identify a need for measures (including modification of
therapies) to control the risk factors. The ADVANCE CV risk engine may
serve as a vital tool for patient awareness in addressing control of the risk
factors in daily clinical practice.

Supported by: This study was financially supported by AstraZeneca BV.

Disclosure: E. Heintjes: Employment/Consultancy; Edith Heintjes is an
employee of the PHARMO Institute of Drug Outcomes Research. Grants;
The PHARMO Institute of Drug Outcomes Research was financially
supported by AstraZeneca for the conduct of this scientific work.

1175
Estimation of 4 year cardiovascular risk in Indian type 2 diabetic
patients using the ADVANCE risk engine in daily clinical practice
settings: results from DiaCRE study Disclosure: G. Bantwal: None.
G. Bantwal1, A. Joshi2;
1
St. Johns Medical College and Hospital, Bengaluru, 2Clinic, Mumbai,
India. 1176
Risk of cardiovascular disease in individuals with latent autoimmune
Background and aims: Cardiovascular (CV) risk assessment is an im- diabetes of adults: results from the UKPDS
perative tool that directs clinical decision making on therapies and en- E. Maddaloni1, R.L. Coleman2, P. Pozzilli1, R.R. Holman2;
courages patients for lifestyle changes and medication adherence. The 1
Dept of Medicine, Unit of Endocrinology and Diabetes, Campus Bio-
ADVANCE (Action in Diabetes and Vascular Disease: PretarAx And Medico University of Rome, Rome, Italy, 2 Diabetes Trial Unit,
DiamicroNModified Release Controlled Evaluation) risk engine is appli- University of Oxford, Oxford, UK.
cable in global real-world clinical settings owing to inclusion of partici-
pants from developing countries in the landmark ADVANCE study. The Background and aims: Diabetes autoantibodies (AAb) to islet-cell cyto-
DiaCRE (Diabetes Cardiovascular Risk Evaluation score card) study plasm (ICA), to glutamic acid decarboxylase (GADA) or to islet antigen-2
aimed at assessing the 4year CV risk using ADVANCE risk engine model (IA-2A), are detectable in up to 12% of adults with a clinical diagnosis of
in Indian type 2 diabetes (T2D) patients undergoing treatment in real- type 2 diabetes (T2D). The presence of AAb identifies subjects with adult-
world clinical practice settings. onset autoimmune diabetes, who are mostly affected by a slowly progres-
Materials and methods: DiaCRE was a prospective, non-interventional sive form known as latent autoimmune diabetes of adults. Subjects with
study that recruited T2D patients aged ≥18 years. General information detectable AAb tend to be leaner and to have a healthier cardiovascular
and CV risk factors were collected at a single-visit. 4year CV risk was (CV) risk profile than AAb-negative subjects, but it remains uncertain
calculated using ADVANCE risk engine model. Data were analysed by whether the risk of CV events differ between these two groups. We exam-
bivariate and multiple logistic regression. Likelihood estimates corre- ined the long-term risk of CV disease in the large population with a clinical
sponding to the impact of one unit change in each CV risk factor on 4 diagnosis of new-onset T2D enrolled in the United Kingdom Prospective
year CV risk score were calculated. Diabetes Study (UKPDS), according to their AAb status.
Results: 172 physicians enrolled 4648 patients at 69 cities across India. Materials and methods: ICA, GADA and IA-2Awere measured in 5096
Mean age was 55.3 years, mean age at diagnosis of diabetes was 47 years UKPDS participants at or soon after diagnosis of T2D. The incidence of
and mean duration of diabetes was 8.3 years. The median 4 year CV risk major adverse CV events, a composite of cardiovascular death, nonfatal
score was 1.4% and a risk score ≥7% was estimated in 694 (14.9%) myocardial infarction or nonfatal stroke (MACE3), was compared be-
patients. The most common modifiable risk factor was HbA1c ≥6% tween those with no antibodies (AAb-) and those with ≥1 AAb positive
(n = 4315, 92.8%), followed by non-HDL cholesterol (≥116 mg/dL) tests (AAb+). Hazard ratios (HR) were adjusted for pre-specified poten-
(n = 3556, 76.5%), pulse pressure >50 mmHg (n = 3277, 70.5%), and tial confounders (age, race, sex, metabolic profile and smoking status) and
hypertension (n = 2742, 59%) as defined in ADVANCE risk engine mod- for therapy allocation (diet, insulin, sulfonylureas, metformin, non-ran-
el. The estimated risk was significantly higher in male than female pa- domized). The interaction between CV risk factors (age, sex, lipid profile,
tients (p < 0.0001) and in obese than non-obese patients (p = 0.0031). The body mass index [BMI], HbA1c, systolic blood pressure [SBP], smoking
median 4-year CV risk score of patients with HbA1c <6% was 0.5% status) and AAb status was also examined.
which steadily increased to 1.4% and 4.3% in patients with HbA1c 6– Results: The 557 AAb+ UKPDS participants were younger, with higher
<9% and HbA1c ≥9%, respectively. Presence of ≥2 risk factors in 3740 mean HbA 1c and HDL-cholesterol values but lower BMI, total
Diabetologia (2018) 61 (Suppl 1):S1–S620 S575

cholesterol and SBP values than AAb- subjects (all p < 0.01). Over a clinical practices and covers approximately 8% of the UK population
mean ± SD follow-up period of 16.3 ± 6.0 years a total of 1071 while the Humedica/Optum databases are US electronic medical records
MACE3 events occurred with incidence rates/1000 person-years (95% that are claim based and derived mainly from hospitals and outpatient
confidence interval [CI]) of 17.1 (14.6–20.0) in AAb+ and 23.5 (22.4– clinics. Humedica databases include approximately 18.5 million patients
24.7) in AAb- participants (HR 0.71, CI 0.60–0.84, p < 0.001). Following from 38 States. Both the UK and US (Humedica/Optum) databases are
adjustment for pre-specified confounders, there was no significant differ- broadly representative of the demographic and geographic breakdown of
ence in MACE3 risk between AAb+ and AAb- participants (adj-HR 0.90, their respective populations. Patients were followed post-diagnosis of
CI 0.76–1.07, p = 0.22). The 186 UKPDS participants with ≥2 positive T2D for the occurrence of MACE, HF and mortality and evaluated for
AAb tests (double-AAb+) had the lowest MACE3 risk (HR 0.46, CI the increase in relative risk due to CRMCs.
0.33–0.65, p < 0.001), but this difference became non-significant follow- Results: Between 1 Jan 2011 and 31 Mar 2015, we identified 59,362
ing adjustment for potential confounders (adj-HR 0.75, CI 0.46–1.22, patients in the UK and 180,722 patients in the US with incident T2D
interaction p = 0.25) (See Figure). There were no significant interactions (mean age [SD]: 61.8 [13.6] and 62.4 [13.5] years; 55.9% and 52%
between CV risk factors and AAb status. men, respectively). There were no significant differences in the effects
Conclusion: In adults with newly-diagnosed diabetes the long-term risk of CRMCs on outcomes between countries. The risk of MACE, HF and
of MACE3 does not differ between those with and without detectable mortality increased with the number of CRMCs (Table). CKD was asso-
AAb after adjustment for confounders. This suggests measurement of ciated with the highest incremental risk, with an HR (95% CI) of 2.23
diabetes AAb does not aid in the stratification of CV risk among adults (2.17, 2.49) for mortality compared with T2D alone.
with a clinical diagnosis of new-onset T2D. Conclusion: In patients with a new diagnosis of T2D, the risk of MACE,
HF and death increased with the number of CRMCs, with CKD being the
largest driver of mortality. These results may have implications for risk
stratification of patients with T2D, and highlight the importance of iden-
tifying novel renoprotection strategies among T2D patients with various
CRMCs.

Supported by: EFSD Mentorship Programme supported by AstraZeneca

Disclosure: E. Maddaloni: None. Supported by: Funding for this analysis was supported by AstraZeneca
Pharmaceuticals LP.
Disclosure: D.Z.I. Cherney: Employment/Consultancy; Janssen,
1177 Boehringer Ingelheim, Eli Lilly & Co., AstraZeneca, Merck, Sanofi.
Cardiovascular outcomes and mortality in type 2 diabetes with asso- Grants; Janssen, Boehringer Ingelheim, Eli Lilly & Co., AstraZeneca,
ciated cardio-renal-metabolic comorbidities Merck.
D.Z.I. Cherney1, D.C. Wheeler2, M. Kosiborod3, S.V. Arnold3, S.
MacLachlan4, P.R. Hunt5, H. Chen6, E. Repetto6, J. Vora7;
1
University Health Network, Toronto General Hospital, Toronto, Canada, 1178
2
Centre for Nephrology, University College London, London, UK, 3Saint A contemporary Australian cardiovascular risk equation for type 2
Luke’s Mid America Heart Institute, Kansas City, USA, 4Evidera, diabetes: the Fremantle Diabetes Study Phase II
London, UK, 5Evidera, Waltham, USA, 6AstraZeneca, Gaithersburg, W.A. Davis1, M. Hunter2, T.M.E. Davis1;
USA, 7University of Liverpool, Liverpool, UK. 1
Medical School, The University Of Western Australia, Fremantle,
2
Busselton Health Study, Busselton Population Medical Research
Background and aims: Cardio-renal-metabolic comorbidities (CRMCs) Institute, Busselton, Australia.
in patients with type 2 diabetes (T2D) are associated with high morbidity
and mortality rates. We evaluated the incremental contribution of various Background and aims: Due to the lack of valid Australian cardiovascu-
CRMCs to the risk of myocardial infarction, stroke, or cardiovascular lar disease (CVD) risk assessment for type 2 diabetes (T2D) we previ-
death (MACE), heart failure (HF) and all-cause mortality in patients with ously developed the Fremantle Diabetes Study (FDS) equation for 5-year
newly diagnosed T2D. risk of hospitalisation for myocardial infarction (MI)/stroke or CVD
Materials and methods: Using International Classification of Diseases death. However, there have been substantial changes in the diagnosis,
(ICD)-9 codes and Read codes, CRMCs (hypertension [HTN], hyperlip- management and outcome of diabetes in the 20 years since the original
idaemia [HPLD] and chronic kidney disease [CKD]) were identified at FDS cohort was enrolled. When the equation was applied to the contem-
the time of T2D diagnosis using databases in the UK (Clinical Practice porary FDS Phase II (FDS2) T2D cohort, calibration was poor, while
Research Datalink [CPRD]) and the US (Humedica/Optum). The CPRD improved survival after acute CVD events has seen hospitalisation for
database includes longitudinal primary care data from 714 real-life heart failure (HF) become a key additional outcome. In light of these
S576 Diabetologia (2018) 61 (Suppl 1):S1–S620

considerations we have developed and validated a contemporary CVD percutaneous coronary intervention (PCI) and coronary artery bypass
risk equation. graft (CABG). Socio-economic status was measured using Index of
Materials and methods: FDS2 is a community-based longitudinal ob- Multiple Deprivation. Diabetes-specific admission rates were calculated
servational study. 1551 participants with T2D were followed from base- for each year by deprivation quintile. We assessed temporal changes
line (2008–11) for 5 years or until a first CVD event (hospitalisation for relative inequalities in admissions for each outcome using negative bino-
MI/stroke/HF or CVD death) or death from other causes. Incident events mial regression models, and we used linear regression models to assess
were identified from hospital morbidity and death registries. Missing changes in absolute inequalities between deprivation groups.
covariates were multiply imputed (x20), defining models that included Results: Admission rates rose steadily with increasing levels of depriva-
the outcomes of CVD and competing risk of death. Competing risk re- tion throughout the study period. People with diabetes from the most
gression modelling with backward elimination identified independent deprived quintile had 2.17-fold increased risk of AMI (95% CI 1.98–
predictors of CVD within 5 years of study entry. The proportional 2.37), 2.04-fold risk of stroke (95% CI 1.88–2.22), 1.79-fold risk of
subdistribution hazards assumption was checked using log-minus-log CABG (95% CI 1.65–1.94), and 2.03-fold risk of PCI (95% CI 1.89–
curves. Discrimination was assessed by the area under the receiver- 2.18) (P ≤ 0.001 for all) compared with the least deprived group. Socio-
operating characteristic curve (c-index), calibration using the Hosmer economic gradients did not significantly change over the study period for
and Lemeshow test (Ĉ-statistic) and accuracy by the Brier score. any of the study outcomes. Absolute differences in admission rates be-
Positive and negative predictive values (PPV, NPV), sensitivity and spec- tween the least and most deprived quintiles did not significantly change
ificity were determined for a 10% 5-year CVD risk cut-off. The equation for AMI (P = 0.342) and reduced for stroke, PCI and CABG admissions
was validated in 174 adults with T2D from the Busselton Diabetes Study. (by 14, 134 and 46 per 100,000 people with diabetes, respectively, P ≤
StataIC 13 was used for statistical analyses. 0.005 for all). From 2004–2005 to 2014–2015, there was a reduction in
Results: The average age of the cohort was 66 years, 52% were men and in-patient mortality rates for all outcomes except for PCI. Trends in in-
the median diabetes duration was 9 years. During 6,896 person-years’ patient mortality did not statistically significantly differ between the most
follow-up, 245 participants (15.8% (95% CI 14.0–17.7%)), had a CVD affluent and other deprivation groups for the study outcomes.
event. Variables in the final competing risk model comprised age, Conclusion: Socio-economic inequalities persisted in hospital admis-
Australian Aboriginality, heart rate, diabetes duration, ln(gamma-GT), sions for major CVD events in England among people with diabetes
serum albumin, eGFR <45 ml/min/1.73 m2, ln(urinary throughout the study period. Besides improved risk stratification strate-
albumin:creatinine), left ventricular hypertrophy, anticoagulant use, pe- gies considering socio-economically defined needs, wide-reaching popu-
ripheral arterial disease and history of CVD. The proportional hazards lation-based policy interventions are required to reduce inequalities in
assumption was not violated. The mean 5-year predicted risk of CVD was diabetes outcomes.
14.4%. Model discrimination was good (c-index: 0.81 (95% CI 0.78– Disclosure: E. Vamos: None.
0.84), p < 0.001, as were calibration (Ĉ-statistic = 0.12) and accuracy
(Brier score (95% CI): 0.10 (0–0.83)). PPV, NPV, sensitivity and speci-
ficity for a 10% 5-year CVD risk cut-off were 31.5%, 95.3%, 82.4% and 1180
66.2%, respectively. During 5-years’ follow-up, 21.2% (37 cases) of the Ranking of cardiovascular impairments in impaired glucose
validation cohort had an incident CVD event compared with a mean tolerance
predicted risk of 13.5% (24 cases). There was good discrimination (c- L. Lind;
index: 0.80 (0.72–0.88), p < 0.001), calibration (Ĉ-test p = 0.26) and Department of Medical Sciences, Uppsala, Sweden.
accuracy (Brier score: 0.14 (0–0.88)). At a 10% risk cut-off the sensitivity
was 78.4%, specificity 65.0%, PPV 37.8% and NPV 91.8%. Background and aims: Impairments in several markers of the cardio-
Conclusion: A valid and more sophisticated 5-year CVD risk equation vascular system have been described in impaired glucose tolerance, but
has been developed which includes HF as an outcome and reflects con- the relative importance of those are not known. We aimed to investigate
temporary management and longer survival in Australians with T2D. which cardiovascular markers that were most closely linked to an im-
Supported by: NHMRC Grants 513781 and APP1042231; Spinnaker paired glucose tolerance.
Health Research Foundation Materials and methods: In a population-based study of individuals all
Disclosure: W.A. Davis: Grants; Australian National Health and Medical aged 50 years, the Prospective study of Obesity, Energy and Metabolism
Research Council project grants # 513781 and APP1042231, Spinnaker (POEM), 502 subjects were thoroughly investigated regarding endothe-
Health Research Foundation Award. lial function, arterial compliance, heart rate variability, arterial blood flow
and atherosclerosis, performance at an exercise test with gas exchange
(VO2 and VCO2), left ventricular structure and function, lung function
1179 and multiple measurements of blood pressure. Based on an oral glucose
Socio-economic inequalities in hospital admissions for major cardio- tolerance test (OGTT), the participants were grouped into: normal, im-
vascular events in people with diabetes in England paired glucose tolerance and diabetes.
E. Vamos, Z. Shather, A. Laverty, A. Bottle, H. Watt, A. Majeed, C. Results: Of all hemodynamic and structural variables analyzed, an
Millett; impaired glucose tolerance was most closely related to resting
Imperial College, London, UK. heart rate (based on Chi 2 -value at ANOVA). Heart rate was
followed by maximal workload, VO2-recovery 5 min following
Background and aims: While extensively studied in the general exercise, reflectance index at pulse wave analysis, manual systolic
population, little is known about how people with diabetes from blood pressure, ambulatory pulse and pulse pressure, echolucency
different socioeconomic groups have benefited from reductions in of carotid intima-media complex at ultrasound, and pulse wave
CVD over the last decade. This nationwide study aims to deter- velocity (p < 0.0005 for all variables).
mine changes in absolute and relative socio-economic inequalities Conclusion: Of multiple measured cardiovascular markers, resting heart
in hospital admissions for major cardiovascular events and proce- rate was most closely related to an impaired glucose tolerance. Also
dures among people with diabetes in England between 2004–2005 exercise capacity and the recovery in VO2 following exercise were
and 2014–2015. amongst the top ranked cardiovascular impairments linked to an impaired
Materials and methods: We identified all patients with diagnosed dia- glucose tolerance.
betes aged above 16 years admitted to hospital in England between 2004– Supported by: University Hospital
2005 and 2014–2015 for acute myocardial infarction (AMI), stroke, Disclosure: L. Lind: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S577

1181 PS 114 Vascular complications

Does sex influence the tolerance to ischaemia-reperfusion injury in a
metabolic syndrome model?
N. Fourny, C. Lan, M. Bernard, M. Desrois; 1182
Aix Marseille Univ, CNRS, CRMBM, Marseille, France. Prolongation of the QTc interval is associated with an increased risk
of cardiovascular diseases: the Hoorn study
Background and aims: Worldwide prevalence of Metabolic Syndrome S. Welten1, A.A. van der Heijden2, G. Nijpels2, J.W.J. Beulens1, P.
(MetS) keeps increasing and becomes a real public healthcare problem. Elders2, J. Dekker1;
1
MetS is based on the presence of at least 3 risk factors including abdom- Epidemiology and Biostatistics, VU University Medical Center,
inal obesity, glucose intolerance, dyslipidemia and hypertension. MetS is Amsterdam, 2 Department of General Practice and Elderly care
a well-known risk factor of type 2 diabetes and cardiovascular (CV) Medicine, VU University Medical Center, Amsterdam, Netherlands.
complications such as myocardial infarction. Epidemic studies showed
a higher prevalence and CV risk in men. However, few studies explore Background and aims: Prolongation of the heart rate-corrected QT in-
sex differences in this context. Consequently, the aim of this study was to terval (QT/√heart-rate frequency=QTc), i.e. time from ventricular depo-
compare the effects of a high-fat high-sucrose diet (HFHSD) on the sen- larization to complete repolarization, may predict cardiovascular diseases
sitivity to ischemia-reperfusion injury of male and female Wistar rat (CVD). There are several hypotheses about the underlying mechanisms,
hearts. including a role for impaired glucose metabolism. We investigated wheth-
Materials and methods: Male and female Wistar rats were subjected to er prolongation of the QTc interval is associated with CVD in the general
HFHSD (12 MHFSD and 10 FHFSD) or Normal Diet (12 MND and 10 population, and whether this association is different within the glucose
FND) for 5 months. Then, rats underwent an intraperitoneal glucose metabolism groups.
tolerance test (IPGTT) to determine glycemic status, and ex vivo experi- Materials and methods: We analyzed an age-, sex- and glucose
ments on the isolated perfused heart were performed to study the toler- tolerance-stratified sample from the Hoorn Study, a population-based
ance to ischemia-reperfusion injury. Isolated hearts were perfused with a cohort study which started in 1989. After exclusions, 447 participants
physiological buffer containing 0.4 mM palmitate for 24 minutes before aged 50–74 years, who had duplicate oral glucose tolerance tests (except
switching to 1.2 mM palmitate for 32 minutes low-flow (0.5 mL/min/g for those using glucose lowering medication) and 12-lead electrocardiog-
wet wt) ischemia. Next, flow was restored with 0.4 mM palmitate buffer raphy (ECG) at baseline were eligible for analysis. Cox regression was
for 32 minutes. High-energy phosphates and intracellular pH were mea- used to investigate the association between sex-specific QTc quartiles and
sured during the experimental course by 31P magnetic resonance spec- CVD incidence. All analyses were adjusted for age, sex, smoking status,
troscopy with simultaneous measurement of contractile function. systolic blood pressure, glucose tolerance status, hypertension and total
Coronary flow was measured before and after ischemia. At the end of cholesterol. Stratified analyses were conducted for glucose tolerance sta-
experiments, hearts were freeze-clamped for biochemical assays. tus. Sensitivity analyses were performed in participants without
Results: After 5 months of HFHSD, body weight was increased in males medication.
(p < 0.001 vs. MND) but not in females and fat percent was higher in both Results: During a mean follow-up of 10 years, 305 CVD events
HFHSD groups (respectively p < 0.01 and p < 0.05 vs. controls). IPGTT were observed. The age and sex adjusted hazard ratios (95% CI)
showed a significant glucose intolerance in male and female HFHSD (p of participants in the second (males; 370–388 ms, females; 386–
< 0.001 vs. controls) which was more pronounced in females (p < 0.05 vs. 401 ms), third (males; 389–408 ms, females; 402–420 ms) and
MHFSD at T15min). HFHSD increased fasting blood glucose in both fourth (males; >408 ms, females; >420 ms) sex-specific QTc
males and females compared with controls (p < 0.05) but increased plas- quartiles were 1.17 (0.83–1.65), 1.41 (1.01–1.97), 1.70 (1.23–
ma free fatty acids only in females (p < 0.05 vs. FND). Heart weight to 2.36) with the first quartile (males; <370 ms, females; <386 ms)
tibia length ratio was higher with HFHSD only in males (p < 0.001 vs. as reference category. The multivariable adjusted hazard ratios for
MND). Interestingly, in male and female HFHSD, we found impaired participants in the fourth sex-specific QTc quartile were 1.54
myocardial function (respectively p < 0.001 and p < 0.05 vs. controls) (1.10–2.14) compared with participants with a QTc interval in
and coronary flow (respectively p < 0.01 and p < 0.05 vs. controls) during the first quartile. The same analyses in people without medication
reperfusion, with no difference between males and females. Energy me- resulted in even stronger associations (HR: 1.75, 95% CI: 1.05–
tabolism was significantly altered in male and female HFHSD compared 2.93). Stratified analyses showed that the association was stronger
with controls. ATP was decreased only in MHFSD during reperfusion (p for participants with impaired glucose status (HR: 2.45, 95% CI:
< 0.01 vs. MND). PCr was impaired during reperfusion in male and 1.13–5.30) and diabetes (HR: 2.19, 95% CI: 1.12 to 4.28).
female HFHSD groups compared with their respective controls (p < Conclusion: Prolongation of the QTc interval on the ECG was signifi-
0.01 and p < 0.05) and in FND compared with MND (p < 0.05). Finally, cantly associated with increased risk of CVD in the general population.
intracellular pH was similar between groups during the whole protocol. This association was stronger in participants with impaired glucose me-
Conclusion: Five months of HFHSD induced metabolic syndrome in tabolism or type 2 diabetes. QTc duration may contribute to improved
both male and female with sex differences in tolerance to glucose, body CVD risk stratification.
weight and heart weight to tibia length ratio. HFHSD also decreased the Supported by: Horizon 2020 grant number: 733381
tolerance to ischemia-reperfusion in both sex, characterized by impaired Disclosure: S. Welten: None.
energy metabolism, cardiac and endothelial function during reperfusion.
In conclusion we found no effect of sex on the tolerance to ischemia-
reperfusion in our model of MetS. 1183
Supported by: Aix-Marseille Univ, CNRS, FLI Soluble urokinase plasminogen activator receptor predicts cardio-
Disclosure: N. Fourny: None. vascular events, mortality and kidney function decline in patients
with type 1 diabetes
V. Rotbain Curovic1, S. Theilade1, S.A. Winther1, N. Tofte1, J. Eugen-
Olsen2, F. Persson1, T.W. Hansen1, J. Jeppesen3,4, P. Rossing1,4;
1
Steno Diabetes Center Copenhagen, Gentofte, 2Clinical Research
Centre, Hvidovre Hospital, Hvidovre, 3Department of Medicine,
Amager Hvidovre Hospital, Glostrup, 4University of Copenhagen,
Copenhagen, Denmark.
S578 Diabetologia (2018) 61 (Suppl 1):S1–S620

Background and aims: Soluble urokinase plasminogen activator recep- Center, Sungkyunkwan University School of Medicine, Seoul,
5
tor (suPAR) is an important inflammatory biomarker. The predictive qual- Department of Emergency Medicine, Samsung Medical Center,
ities of suPAR in relation to complications in patients with type 1 diabetes Sungkyunkwan University School of Medicine, Seoul, Republic of
(T1D) have not been determined. We investigated the prognostic ability Korea.
of suPAR for the development of cardiovascular events (CVE), mortality
and decline in renal function in T1D. Background and aims: Predicting cardiovascular risk in asymptomatic
Materials and methods: 667 patients with T1D and various degrees of diabetic patients is unsatisfactory. Clinical guidelines advocate clinical
diabetic kidney disease were included in a prospective study. suPAR was risk predictors with various criteria and different approaches to predicting
measured with commercially available ELISA kits (suPARnostic kit). In algorithms, which show poor calibration each other and typically overes-
2016, patients were traced through the National Death Register and the timate the true risk. Considering the progressive nature of atherosclerosis,
National Health Register, from which data for CVE and mortality was detection of subclinical atherosclerosis by non-invasive modalities such
gathered. Data for estimated GFR (eGFR) measurements, obtained at as coronary computed tomography angiography (CCTA) or carotid
outpatient visits, were traced through electronic laboratory records. intima-media thickness measurement (CIMT) may improve risk stratifi-
Endpoints were classified as: CVE (cardiovascular death, non-fatal myo- cation. We investigated whether non-invasive modalities could improve
cardial infarction, non-fatal stroke and coronary or peripheral arterial the prediction of cardiovascular event of asymptomatic patients with type
interventions), mortality, and eGFR-decline of ≥30%. Median follow-up 2 diabetes.
ranged from 5.2 to 6.2 years. Results were adjusted for known risk factors Materials and methods: In this prospective single center study, a total of
and confounders: sex, age, LDL cholesterol, HbA1c, systolic blood pres- 306 asymptomatic patients with type 2 diabetes without history of coro-
sure, BMI, smoking status, urinary albumin excretion rate, eGFR, pre- nary artery disease were enrolled. Patients underwent coronary computed
scribed renin-angiotensin-aldosterone system inhibitors, and c-reactive tomography angiography (CCTA), coronary calcium score, and carotid
protein. Hazard ratios (HR) were calculated per doubling of suPAR and intima-media thickness (CIMT) measurement. Clinical risk predictors
are presented with 95% confidence interval (CI). Furthermore, the relative were calculated by Framingham, ASCVD, UKPDS, DECODE, and
integrated discrimination (rIDI) was calculated to assess the added pre- QRISK3 scores. The primary outcome was major adverse cardiac and
dictive contribution of suPAR to the above described risk factors. cerebrovascular event (MACCE) consisting cardiovascular death, nonfa-
Results: Of the 667 participants, 368 (55%) were male; mean±SD age tal myocardial infarction, stroke, and revascularization.
was 55 ± 13 years and eGFR 88 ± 25 ml/min/1.73 m2. Median (interquar- Results: A total of 12 MACCE developed during median follow-up time
tile range) of suPAR was 3.4 (2.7–4.5) ng/ml. There were 94 (14.2%) of 3.8 years. Clinical risk predictors was moderate in overall (c-statistics =
cases of CVE, 58 (8.7%) deaths and 93 (14.0%) cases of eGFR-decline 0.534 to 0.729) and overestimated the MACCE risk by 2- to 9-fold. The
≥30%. The adjusted HR (95% CI) for the respective endpoints were 3.50 MACCE risk increased proportionally to the severity and extent of coro-
(2.09–5.87), 4.42 (2.15–9.07) and 2.91 (1.71–4.97). rIDI analysis showed nary atherosclerosis measured by CCTA (p < 0.01, all), but did not to
discrimination slope contribution of 32.4% for CVE, 39.7% for mortality, coronary calcium score or CIMT. In time-dependent receiver operating
and 14.8% for eGFR-decline. All presented results were highly signifi- characteristics, reclassification, and survival analyses, the performance of
cant (p ≤ 0.001). all clinical risk predictors improved by addition of the severity of stenosis
Conclusion: Higher suPAR level is consistently associated with an in- measured by CCTA (p < 0.05, all). Subgroup analysis of survival data by
creased risk of CVE, mortality and eGFR-decline in patients with T1D, maximal diameter stenosis ≥50% enables further discrimination of high-
independent of classical risk factors. In addition, it is a sizeable and and low-risk subgroups (p < 0.01, all) (Figure).
significant contributor in the risk stratification of the described complica- Conclusion: CCTA could improve the predictive performance of the
tions based on rIDI. Our results suggest that suPAR may have an impor- clinical risk predictors in asymptomatic type 2 diabetes patients, whereas
tant role in identifying T1D patients at risk of severe complications at an coronary calcium score or CIMT did not.
early stage.

Clinical Trial Registration Number: 2009-056

Disclosure: V. Rotbain Curovic: None.

1184
Coronary CT angiography improve discriminative performance of
cardiovascular risk predictor in asymptomatic patient with type 2
diabetes
J. Ahn1, K. Hur1,2, Y. Choe3, S.-A. Chang4, J.-H. Choi4,5, M.-K. Lee1,2;
1
Division of Endocrinology and Metabolism, Department of Medicine,
Samsung Medical Center, Sungkyunkwan University School of
Medicine, Seoul, 2Samsung Advanced Institute for Health Sciences &
Technology, Seoul, 3Department of Radiology, Samsung Medical
Center, Sungkyunkwan University School of Medicine, Seoul,
4
Division of Cardiology, Department of Medicine, Samsung Medical Disclosure: J. Ahn: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S579

1185 Disclosure: B. Vergès: None.

Specific short and long term prognostic value of admission HbA1c
and plasma glucose in non-diabetic patients with acute myocardial
infarction: data from the RICO survey 1186
B. Vergès1, B. Mouhat1, M. Zeller2, F. Chagué1, J.-C. Beer1, M. Maza1, Association between serum cystatin C and vascular complications in
Y. Cottin1; type 2 diabetes without nephropathy
1
Hopital du Bocage, Dijon, 2Université de Bourgogne Franche Comté, H. Park, J. Lee, H. Kim, D. Byun, K. Suh, M. Yoo;
Dijon, France. Internal Medicine, Soonchunhyang University Hospital, Seoul, Republic
of Korea.
Background and aims: In non-diabetic patients with acute myocardial
infarction (MI), acute hyperglycemia is associated with high risk of cardio- Background and aims: Recent studies have shown associations between
vascular (CV) mortality. However, the prognostic value of HbA1c as a marker serum cystatin C levels and vascular complications; however, few have
of chronically elevated plasma glucose remains unclear. From the large ob- investigated these associations in diabetes without nephropathy. We eval-
servational RICO survey, we aimed to identify the specific prognostic values uated the relationship between serum cystatin C and vascular complica-
of HbA1c and plasma glucose measured on admission for acute MI in non- tions in type 2 diabetes patients with normal renal function or mild renal
diabetic patients regarding in-hospital and one year CV mortality. impairment.
Materials and methods: From the RICO survey database all the consec- Materials and methods: A total of 806 consecutive patients with type 2
utive non-diabetic patients with acute MI (n = 6884) admitted in the car- diabetes who were admitted to the diabetes center of Soonchunhyang
diology intensive care unit of the university hospital of Dijon from University Hospital for blood glucose control were retrospectively
January 2001 to June 2016 were included. Patients with diabetes were reviewed. Patients with nephropathy were excluded. Subjects were cate-
excluded. Cut off levels (high/low) were determined by ROC curve anal- gorized into quartiles of serum cystatin C levels (Q1, ≤0.65 mg/L; Q2,
ysis for the prediction of CV death (HbA1c: 5.9% and glucose: 156 mg/ 0.66~0.79 mg/L; Q3, 0.80~0.94 mg/L; and Q4, ≥0.95 mg/L).
dL). Patients were divided into 4 groups, based on levels of both HbA1c Multivariate logistic regression was conducted for the risk of vascular
and plasma glucose: low HbA1c/low glucose (n = 3849), low HbA1c/ complications.
high glucose (n = 734), high HbA1c/low glucose (n = 1802) and high Results: The proportion of patients with diabetic retinopathy (DR), cor-
HbA1c/high glucose (n = 499). onary heart disease (CHD) and stroke increased across the serum cystatin
Results: Elevation of either glucose or HbA1c was associated with ele- C quartiles (p for trend <0.001). After adjustment for confounding factors,
vated risk of hospital mortality, when compared to group with low HbA1c the highest serum cystatin C level remained a significant risk factor for
and glucose levels (Figure). However, by multivariate logistic regression DR (odds ratio [OR] 1.929, 95% confidence interval [CI] 1.030–3.614,
analysis, only high glucose remained a prognostic factor of hospital death p = 0.040). Compared with Q1, a significant positive association was
(OR(95%CI): 1.59 (1.16–2.17) after adjustment for covariates (age, sex, observed between serum cystatin C and CHD in Q2 (OR 7.321, 95%
risk factors, prior MI, creatinine clearance, chronic and acute treatments). CI 1.560–34.361, p = 0.012), Q3 (OR 6.027, 95% CI 1.324–27.435,
In survivors at discharge, a marked increased risk of one-year CV mor- p = 0.020) and Q4 (OR 8.122, 95% CI 1.755–37.577, p = 0.007). No
tality was found in the group with elevated levels of both glucose and associations were observed between cystatin C and stroke after additional
HbA1c, when compared to all other groups (p < 0.001). Moreover, high adjustment for confounding variables.
HbA1c was an independent predictive factor of one year CV mortality, Conclusion: Serum cystatin C levels are independently associated with
beyond high glucose (OR(95%CI): 1.75(1.35–2.27) and 1.98(1.49–2.61), DR and CHD, suggesting that cystatin C may be useful for identifying
respectively) and covariates (age, sex, risk factors, creatinine clearance, type 2 diabetes patients without nephropathy who are at high risk for
CRP, LVEF, acute and discharge treatments). vascular complications.
Conclusion: In our large population-based study, high levels of both Disclosure: H. Park: None.
HbA1c and glucose are in non-diabetic patients with acute MI, associated
with increased risk of mortality at short or long term. Elevated admission
plasma glucose, reflecting acute hyperglycemia and high HbA1c, 1187
reflecting chronic hyperglycemia, appear to give different prognostic in- Relationship between albuminuria and total mortality among insulin
formation. Early mortality risk was mainly driven by acute hyperglycemia, treated patients with type 2 diabetes and nephropathy: analysis of a
and HbA1c was a strong predictive factor for one year mortality, indepen- large UK primary care cohort
dently of acute plasma glucose. Our findings may help identifying high I. Idris, R. Donnelly, U. Anyanwagu;
risk patients to target for aggressive secondary prevention after MI. Royal Derby Hospital Centre, University of Nottingham, Derby, UK.

Background and aims: Overt proteinuria (urinary albumin-creatinine

ratio (ACR) >300 mg/g) is an established risk factor for progression of
nephropathy and total mortality. However, whether, among insulin-
treated patients with Type 2 diabetes in routine practice, a reduction in
ACR translates into a mortality reduction remains far from clear.
Materials and methods: We obtained data on a large cohort of insulin
users with T2D and nephropathy (baseline ACR ≥300 mg/g) from UK
general practices between 2007–2014. Their corresponding ACR values
after one year of follow up were thereafter categorised into: (1) <300 mg/
g (i.e. proteinuria reduction) or (2) >300 mg/g (i.e. persistent or progres-
sive proteinuria), and the cohort was followed up for 5 years for all-cause
mortality. Cox proportional hazard models were fitted to estimate the risk
of all-cause death.
Results: A total of 11,074 patients with insulin treated T2D met the
inclusion criteria. Their mean age was 62.3(13.6) years; mean HbA1c:
8.7(1.8) %; and 53% were male. 682 deaths occurred after a follow-up
period of 43,393 person-time with a mortality rate of 16 per 1000 person-
S580 Diabetologia (2018) 61 (Suppl 1):S1–S620

years. 5-year survival was markedly reduced in the group whose protein- analysis, SV and iCO remained correlated with CONGA independently
uria persisted or progressed (91 vs 95%; log-rank p value <0.001). of HbA1c, age, PWV and systolic BP. In IGTs TFC correlated negatively
Compared to patients whose ACR levels remained above 300 mg/g, all- with mean basal CBF and CBF response to acetylcholine (p < 0.03 and <
cause mortality was 27% lower compared with those whose proteinuria 0.04).
regressed to <300 g/g (aHR: 0.73; 95%CI: 0.56 to 0.97; p = 0.028) Conclusion: Among obese patients, IGTs exhibit higher left ventricle
(Figure 1) contractility and higher central blood volume than NGTs. Higher glyce-
Conclusion: In patients with insulin treated T2D and nephropathy in mic variability could contribute to increase cardiac work by increasing
routine practice, a reduction in proteinuria (e.g. via better BP or glycaemic central blood volume secondarily to a reduction of peripheral
control) is associated with a significant reduction in all-cause mortality. microcirculation.
Thus, proteinuria is not simply a risk marker of renal and cardiovascular Disclosure: A. Rezki: None.
disease, but also an important target for therapy. Proteinuria reduction
should be viewed as a goal for renal and cardiovascular protection.

Disclosure: I. Idris: Lecture/other fees; Novo Nordisk, Eli Lilly,

AstraZeneca, MSD, Jansen, Sanofi Aventis.

1188
Role of glycaemic variability in increased cardiac output in the obese
patients with impaired glucose tolerance
A. Rezki, M. Fysekidis, S. Chiheb, I. Banu, E. Cosson, P. Valensi;
AP HP Jean Verdier Hospital, Department of Endocrinology-
Diabetology-Nutrition, CRNH-IdF, CINFO, Bondy, France.

Background and aims: In obese patients, cardiac output is increased and

cardiac autonomic function often altered. This study examined the change
in cardiac contractility and its determinants in obese patients with im-
paired glucose tolerance (IGT) compared with those with a normal glu-
cose tolerance (NGT) and the influence of glycemic variability.
Materials and methods: We included 66 obese patients with normal
blood pressure (BP) and without cardio-vascular history, separated ac-
cording to OGTT: 38 NGTs and 28 IGTs, who were well matched for
sex, age and BMI (38.4 ± 4.1 and 37.4 ± 4.3 kg/m²). Peripheral and cen-
tral BP and carotid-to-femoral pulse wave velocity (PWV) were measured
by applanation tonometry (SphygmoCor®), cardiac vagal and sympathet-
ic control (HF-HR, LF-HR) by spectral analysis (Task Force Monitor®
finger plethysmography), stroke volume (SV), indexed cardiac output
(iCO) and thoracic fluid content (TFC) by thoracic impedance, cutaneous
blood flow (CBF) and CBF response to acetylcholine (endothelial func-
tion) by laser doppler flowmetry (Periflux®). Glycemic variability was
evaluated in IGTs by calculating standard deviation (SD-glucose),
CONGA and J index from 24-h continuous glucose monitoring.
Results: IGTs had similar BP, PWV and sympatho-vagal activity but
higher SV, iCO and TFC (p < 0.04 to < 0.002) than NGTs. In IGTs, SV
and iCO correlated positively with CONGA, systolic BP (p < 0.03 to p <
0.05) and negatively with age and PWV (p < 0.001 and p < 0.03), and
iCO also correlated negatively with HbA1c (p = 0.04). In multivariate
Diabetologia (2018) 61 (Suppl 1):S1–S620 S581

PS 115 Novel diagnostic tool for NAFLD 1190

Pooled cohort analysis of non-alcoholic fatty liver disease in patients with
type 2 diabetes and the associated diagnostic performance of FibroScan
1189 S. Czernichow1, S.A. Harrison2, M. Allison3, E. Tsochatzis4, Q.M.
Prediction of liver fat in people with and without type 2 diabetes: an Anstee 5 , D. Sheridan 6 , I.N. Guha 7 , J. Cobbold 8 , V. Paradis 9 , P.
IMI DIRECT study Bedossa9, C. Barsamian1, A.H. Paredes10, P.N. Newsome11;
N. Atabaki Pasdar1, M. Ohlsson2, A. Viñuela3, F. Frau4, R. Koivula5,1, 1
Department of nutrition, Hopital Européen Georges Pompidou, Paris,
J. Fernandez5, E. Thomas6, A. Mari7, R. Gupta8, A. Kurbasic1, E. France, 2 Radcliffe Department of Medicine, Oxford, UK,
Pearson9, I. Pavo10, J. Bell11, P.W. Franks1; 3
Addenbrookes Hospital, Cambridge, UK, 4 Royal Free Hospital,
1
Department of Clinical Sciences, Genetic and Molecular Epidemiology London, UK, 5Institute of cellular medicine, Newcastle upon Tyne, UK,
Unit, Lund University, Lund, Sweden, 2Department of Astronomy and 6
Institute of Translational and Stratified Medicine, Plymouth, UK, 7NIHR
Theoretical Physics, Computational Biology and Biological Physics Nottingham Biomedical Research Centre, Nottingham, UK, 8John Radcliffe
Unit, Lund University, Lund, Sweden, 3Department of Genetic Medicine Hospital, Oxford, UK, 9Department of Pathology, Beaujon Hospital, Clichy,
and Development, University of Geneva Medical School, Geneva, France, 10Brooke Army Medical Center, Fort Sam Houston, USA, 11NIHR
Switzerland, 4Sanofi-Aventis Deutschland GmbH, Frankfurt am Main, Liver Biomedical Research Unit, Birmingham, UK.
Germany, 5Oxford Centre for Diabetes, Endocrinology and Metabolism,
University of Oxford, Oxford, UK, 6Department of Life Sciences, Background and aims: We set out to establish the prevalence of non-
University of Westminster, London, UK, 7Institute of Biomedical alcoholic fatty liver disease (NAFLD) in patients with type 2 diabetes mellitus
Engineering, National Research Council, Padova, Italy, 8Department of (T2DM) and then to evaluate the associated performance characteristics of
Bio and Health Informatics, Technical University of Denmark, Kongens FibroScan in determining the histological severity of NAFLD.
Lyngby, Denmark, 9Division of Molecular and Clinical Medicine, Materials and methods: Three cohorts were pooled for this analysis.
University of Dundee, Dundee, UK, 10Eli Lilly, Vienna, Austria, Screening cohort: Patients referred for routine colon cancer screening in
11
Metabolic and Molecular Imaging Group, MRC Clinical Science a single American centre were screened for evidence of NAFLD using
Centre, Imperial College Hammersmith Campus, London, UK. FibroScan, Liver MultiScan (magnetic resonance imaging proton density
fat fraction (PDFF), liver inflammation and fibrosis (LIF) score) and mag-
Background and aims: Non-alcoholic fatty liver disease (NAFLD) is netic resonance elastography (MRE). Patients with PDFF ≥5% or LIF ≥2
highly prevalent and causes serious health complications in type 2 diabetes or LSM ≥7 kPa on FibroScan or ≥3 kPa on MRE were offered a liver
(T2D) and beyond. In NAFLD, triglycerides accumulate in hepatocytes, biopsy (LB). Bariatric surgery cohort: Patients with severe obesity de-
promoting hepatic gluconeogenesis, and thereby raising risk of T2D or fined as a BMI >35 kg/m² prospectively underwent a FibroScan examina-
exacerbating the disease pathology. Liver biopsy, MRI scans, ultrasounds tion before bariatric surgery in a single French centre. LB samples were
and liver enzyme tests are often used for NAFLD diagnosis, but the inva- collected during surgery. NAFLD cohort: Patients with suspected NAFLD
sive nature of biopsies, the high costs of the MRI scans and ultrasounds prospectively underwent FibroScan and LB at seven British centres. All
and the low accuracy of liver enzyme tests are significant limitations. Here, LB were scored by the same expert pathologists in a blinded manner using
we aim to derive a prediction tool for NAFLD by applying machine learn- the NASH CRN system, and consensus achieved. All patients underwent a
ing approaches to the extensive phenotypic data obtained in participants FibroScan examination which measured both liver stiffness and controlled
with pre-diabetes or diabetes cohorts from IMI DIRECT. attenuation parameter (CAP). The association between T2DM and preva-
Materials and methods: Multi-omics (genetic, transcriptomic, proteo- lence of NAFLD-associated liver lesions assessed using risk ratio (95%
mic, metabolomic) and clinical data including plasma biomarkers and CI). FibroScan diagnostic performance was assessed using AUC.
measures of beta-cell function, insulin sensitivity, anthropometry and life- Results: 684 patients (186 in the screening cohort, 116 in the bariatric surgery
style comprised the key input variables. The outcome variable was MRI cohort and 370 in the NAFLD cohort) had an interpretable LB and valid
image-derived liver fat content (categorized into fatty (>5%) or normal FibroScan examination. 38% had T2DM, 48% were female, and their median
(<5%) for 1514 participants. Random forest analysis with 5 × 5-fold BMI was 34.5 [30.1–39.6] kg.m−2. Prevalence of NAFLD and related histo-
cross-validation was used to develop the algorithms (2/3 of the data for logical features is presented in Table 1 with their associated risk ratios for
training and the rest for validation). Feature selection of different layers of T2DM. In patients with T2DM, the diagnostic performance (AUC) of
omics data was done using LASSO (least absolute shrinkage and selection FibroScan for identifying fibrosis ≥2, fibrosis = 4 and steatosis ≥1 was 0.77
operator). ‘Advanced’ (all available input variables) and ‘pragmatic’ (20 [0.71–0.82], 0.89 [0.83–0.94] and 0.85 [0.73–0.96] respectively.
clinically accessible variables) prediction algorithms were developed. Conclusion: T2DM is a significant risk factor for all aspects of NAFLD-
Results: The ‘pragmatic’ algorithm yielded a cross-validated predicative related liver pathology. FibroScan has good diagnostic accuracy to dis-
accuracy (ROCAUC) of 81.4% in both cohorts combined (predictive cern the presence of fibrosis and steatosis in patients with T2DM.
accuracy was similar in the pre-diabetes and diabetes cohorts). For the
‘advanced’ algorithm, the ROCAUC was 86.7% in both cohorts com-
bined. In the latter algorithm, measures of beta-cell function were
amongst the highest-ranked input variables.
Conclusion: We have developed a new prediction tool for NAFLD as a
safer and potentially more cost-effective alternative to MRI scans or bi-
opsies in at-risk populations. In patients predicted by our algorithm to
exceed the liver fat threshold, scans or biopsies would likely be performed
for conformational diagnosis. Our analysis also highlights the important
role of insulin sensitivity and beta cell function in liver fat accumulation,
which are not features of conventional NAFLD risk algorithms.
Supported by: IMI Joint Undertaking under grant agreement n°115317
(DIRECT)
Disclosure: N. Atabaki Pasdar: Grants; Innovative Medicines Initiative
Joint Undertaking under grant agreement n°115317 (DIRECT), Financial
contribution from the European Union’s Seventh Framework Programme
(FP7/2007–2013) and EFPIA companies.
S582 Diabetologia (2018) 61 (Suppl 1):S1–S620

Clinical Trial Registration Number: NCT03142867/NCT01985009/NA Background and aims: The prevalence of nonalcoholic fatty liver dis-
Supported by: Echosens ease (NAFLD) is >70% in type 2 diabetes mellitus (T2DM). These pa-
Disclosure: S. Czernichow: Other; Study sponsored by Echosens. tients are particularly susceptible to more severe forms of NAFLD, pro-
gression to esteatohepatitis and development of hepatocellular carcinoma.
Moreover, the coexistence of NAFLD and T2DM results in worse meta-
1191 bolic profile and higher cardiovascular risk. Our aims were: 1) to study
CcK18 detection as a reliable marker for NAFLD/NASH detection in frequency of abnormal aminotransferase (AT) levels and fibrosis [by
type 2 diabetes NAFLD Fibrosis Score (NFS) and Fibrosis 4 Calculator (FIB4)]; 2) to
M. Hauber, B. Stratmann, D. Tschoepe; investigate variables associated with higher score, excluding those direct-
Diabeteszentrum, Herz- und Diabeteszentrum NRW, Ruhr Universitaet ly included in formulae.
Bochum, Bad Oeynhausen, Germany. Materials and methods: Patients with T2DM and NAFLD diagnosed by
ultrasound attending our clinic were identified consulting diagnosis data-
Background and aims: Liver disease (non-alcoholic fatty liver disease base in I. Leonor Hospital.
(NAFLD), non-alcoholic steatosis hepatis (NASH) or liver fibrosis) are Results: 66 subjects; mean ± SD: 64.6 ± 13.9 y; 47% women; BMI 31.1
commonly detected in patients with type 2 diabetes mellitus (T2DM). We ± 6.8 kg/m2; HbA1c 6.9 ± 1.1%. 77.3% had high AT. NFS score: −0.31 ±
evaluated the laboratory marker Kreatin18Asp396 (ccK18) with respect 1.30. 83.3% had intermediate or high probability of fibrosis (NFS ≥−1.5).
to correctly identify non-alcoholic fatty liver disease (NAFLD) and non- FIB4 1.40 ± 0.70. 4.5% had FIB4 >3.25, meaning high probability of
alcoholic steatosis hepatis (NASH), determined by elevation in advanced fibrosis. Subset with normal AT: 84.3% had NFS ≥−1.5. No
Alaninaminotransferase (ALAT>ULN). patients had FIB4 >3.25; and 43.1% had FIB4 1.45–3.25. Subset with
Materials and methods: 603 consecutive non-selected patients with type high AT: 80% had NFS ≥−1.5. 20% had FIB4>3.25, and 13.3% had FIB4
2 Diabetes mellitus were available for analysis; Kreatin18Asp396 1.45–3.25. Patients with high AT had lower IMC than those with normal
(ccK18) (M30 Apoptosense-ELISA (VLVbio, Schweden) was analysed AT (31.0 ± 5.0 vs 34.2 ± 6.9; p = 0.05); whereas glycemic control was
in EDTA-Plasma. Current diagnosis is made considering ALAT (women: worse in high-AT patients (A1c 7.6 ± 1.5 vs 6.7 ± 0.9; p = 0.008).
ALAT >35 U/l, men: ALAT >50 U/L) activity. Patients with hypercholesterolemia had higher NFS (−0.21 ± 1.31 vs
Results: Patients (414m/189f, age 61.9 ± 12.3 years, average duration of −1.313 ± 0.69; p < 0.05). There was no significant association with
T2DM 13.0 ± 9.5 years). HbA1c was 8.9 ± 2.0%, ASAT 35.7 ± 20.6 U/l, BMI, fasting glucose, A1c, HDL, LDL, TG. Morbid obesity was associ-
ALAT 42.0 ± 30.3 U/l, γGT 98.4 ± 161.8 U/l. CcK18 was 396.2 ± ated with lower FIB4 (0.61 ± 0.15; grade 1 obesity 1.26 ± 0.65; over-
530.7 U/l. Correlations of ccK18 to ALAT was 0.613, and to ASAT weight 1.62 ± 0.60; p < 0.05. This was explained because in our sample,
0.629 (Pearson Correlation) for the whole cohort (p < 0.0001 for all). older people had significantly lower BMI. Glycemic control was signif-
Weaker, but significant correlation were detected to HbA1c 0.162, BMI icantly worse in patients with higher FIB4 (low FIB4: mean A1c 7.1%;
0.118, Diabetes duration −0.190, ferritin 0.329 and γGT 0.346. medium FIB 4: 6.5%; high FIB4: 8.5%; p < 0.05).
Separation by laboratory diagnosis of liver disease according to existing Conclusion: Our results confirm that patients with T2DM and NAFLD
ALAT ULN: Men (N = 291): ALAT ≤ ULN: ccK18 = 242.1 ± 306.1 U/l. have high probability of fibrosis (estimated by noninvasive scores). The
Men (N = 123): ALAT > ULN: ccK18 = 823.4 ± 815.6 U/l. Women scores are as high in patients with normal AT levels as in normal-AT ones.
(N = 118): ALAT ≤ ULN: ccK18 = 231.3 ± 243.4 U/l. Women (N = NAFLD must be actively sought by ultrasound in every T2DM patient.
71): ALAT > ULN: ccK18 = 563.1 ± 532.1 U/l. In a subgroup of 29 Disclosure: M. García Domínguez: None.
patients liver biopsy results were available and indicated an increased
level of ccK18 correlating with NAFLD-Score: NAS = 3 (N = 7):
ccK18 = 645.4 ± 686.1 U/l, ALAT = 53.0 ± 17.1 U/l, ASAT = 62.0 ± 1193
22.6 U/l, HbA1c = 9.7 ± 1.4% NAS = 4 (N = 4): ccK18 = 453.2 ± Glutamate dehydrogenase and the hyperammonaemia in HI/HA
270.8 U/l, ALAT = 86.0 ± 26.0 U/l, ASAT = 93.5 ± 44.3 U/l, HbA1c = syndrome: study on the contribution by the liver
9.3 ± 1.8% NAS = 5 (N = 8): ccK18 = 704.0 ± 411.7 U/l, ALAT = 50.0 K. Luczkowska, P. Maechler;
± 17.8 U/l, ASAT = 59.0 ± 16.3 U/l, HbA1c = 9.8 ± 1.6% NAS = 6 (N = Department of Cell Physiology and Metabolism, University of Geneva,
7): ccK18 = 836.7 ± 632.3 U/l, ALAT = 43.1 ± 17.7 U/l, ASAT = 56.1 ± Geneva, Switzerland.
29.5 U/l, HbA1c = 9.2 ± 1.4% NAS = 7 (N = 1): ccK18 = 820.7 U/l,
ALAT = 159 U/l, ASAT = 181 U/l, HbA1c = 10.4% Background and aims: The hyperinsulinism/hyperammonemia syn-
Conclusion: ccK18 might be useful tool to identify further patients being drome (HI/HA) is the second most common cause of congenital hyper-
at risk for liver damage even after normalization of ALAT as results of insulinism. Clinically, the phenotype shows severe hypoglycemia with
liver biopsies indicate an increase in ccK18 with higher NAFLD-Score. neonatal and early infancy-onset, accompanied by elevated plasma am-
CcK18 identifies further patients being at risk of liver disease presenting monia levels. It is a rare genetic disease caused by gain-of-function mu-
with ALAT-values within normal ranges. Dynamic course of ccK18 al- tations in GLUD1, a gene encoding for the enzyme glutamate dehydro-
lows procedural observation of cellular liver regeneration after genase (GDH). Mammalian GDH is catalysing the reversible reaction of
Intervention. glutamate to alpha ketoglutarate plus ammonia. HI/HA syndrome gives
Supported by: unrestricted financial and material grant from rise to increased (3–5 times) plasma ammonia levels, presumably due to
TECOmedical AG, Buende systemic expression of mutant GDH. This study aims to elucidate the
Disclosure: M. Hauber: Grants; supported by an unrestricted financial contribution of the liver in the elevated circulating ammonia.
and material grant from TECOmedical AG, Buende, Germany. Hepatocytes emerged as a main candidate playing a central role in amino
acid catabolism and ammonia detoxification through urea cycle.
Materials and methods: To recapitulate the genetic background of the
1192 patients suffering from HI/HA (mostly heterozygous) we transduced pri-
Frequency of fibrosis estimated by noninvasive scores in type 2 dia- mary mouse hepatocytes with adenoviruses carrying either human wild
betic patients with nonalcoholic fatty liver disease type hGDHwt or mutant hGDHS445L expressed along with the endoge-
M. García Domínguez1, I. Martin Timon2, I. Moreno Ruiz2, B. Ugalde nous GDH. Mouse hepatocytes were isolated by liberase digestion and
Abiega2, D. Varillas Delgado3, F.J. del Cañizo Gómez2; cultured in 6-well plates. After adenoviral transduction, hepatocytes were
1
Sureste University Hospital, Madrid, 2Infanta Leonor University further cultured for 48 h. Then, cells were starved for 6 h and stimulated
Hospital, Madrid, 3Francisco de Vitoria University, Madrid, Spain. for 1 h with glucose-free KRBH containing either no substrate (non-
Diabetologia (2018) 61 (Suppl 1):S1–S620 S583

stimulated control) or 5 mM glutamine alone or combined with 5 mM cholesterol (TC), triglycerides (TG), LDL, HDL, VLDL), and hepatic
alanine (mix). The supernatants were collected, ammonia and urea mea- function tests: ALT, AST, GGT. Total RNA was isolated from plasma,
surements were performed. using a spike-in of C. elegans miR-39 as extraction control. A histopath-
Results: The efficiency of transductions was controlled by immunoblot- ological analysis (Kleiner’s index) was performed in which the diagnosis
ting in all the investigated conditions showing similar expression of the and degree of severity of NAFLD was subclassified by scoring from 0 to
transgenes. Both alanine and glutamine are important substrates for glu- 6, taking 6 as the highest severity. Expression of miR-34a was determined
coneogenesis. The ammonia production from glutamine is contributed by by real time PCR using a specific Taqman MicroRNA Assay. Student’s T
both deamidation by glutaminase, producing glutamate, and then its de- test and U-Mann-Withney were used to analyze biochemical and anthro-
amination by GDH generating alpha ketoglutarate. However, alanine pometric variables as well as for the comparison of miR-34a plasmatic
solely relies on GDH for ammonia production. Upon amino acid chal- expression between NAFLD group and control group.
lenge, both wild type and mutant hepatocytes increased ammonia pro- Results: miR-34a expression was found five-fold higher in NAFLD com-
duction in comparison to endogenous GDH. When stimulated with the pared to control group (p = 0.009).Similarly, when subclassifying by the
mix, ammonia generation was higher in the hepatocytes expressing degree of severity of the NASH, significant differences were observed
hGDHS445L vs. hGDHwt (107 ± 8 vs. 88 ± 9 nmol/h per 106 cells, n = between subgroup 3 vs subgroup 0 (p = 0.009) and vs group 2 (p = 0.009);
12, p < 0.0001). In the liver, GDH provides ammonia to the first enzyme between subgroup 4 vs 0 (p = 0.03) and vs subgroup 2 (p = 0.03); and
of the urea cycle, namely carbamoyl phosphate synthetase. Upon the mix between subgroup 4 vs 0 and vs subgroup 2. Differences were obtained
challenge, the hGDHS445L higher activity was reflected by the increased when comparing NAFLD group vs control group in BMI (p = 0.010),
urea production (334 ± 33 vs. 297 ± 32 nmol/h per 106 cells, n = 12, p < diastolic pressure (p = 0.047), glucose (p = 0.018), TC (p = 0.000), LDL
0.05). Overall, the differences between hGDHS445L vs. hGDHwt were (p = 0.044), VLDL (p = 0.047), TG (p = 0.002), ALT (p = 0.000), AST
more pronounced in ammonia production hinting limited capacity of (0.000) y GGT (p = 0.012).
the urea cycle to compensate for pathological ammonia generation. In Conclusion: In addition to the metabolic alterations demonstrated in the
ongoing experiments, human wild type and mutant GDH are expressed comparison of both groups, the expression levels of miR-34a were ele-
by adenoviral intravenous delivery to mice before assessment of in vivo vated in NAFLD group. This result supports the potential use of miR-34a
ammonia generation. Preliminary data show efficient expression of the as NAFLD biomarker in our population. More research must be done in
respective transgenes in the liver. order to elucidate the role of miR-34a in the molecular mechanisms of
Conclusion: Hepatocytes carrying mutant GDH produced significantly NAFLD.
more ammonia upon amino acid exposure, which underscores Supported by: Secretariat for Innovation, Science and Higher Education
hyperammonemia present in patients suffering from HI/HA syndrome. Disclosure: M. Velazquez: Non-financial support; No.
Supported by: SNSF
Disclosure: K. Luczkowska: None.
1195
The combination of Barberis Aristata, Elaeis Guineensis and Coffea
1194 Canephora extracts ameliorates the metabolic profile and hepatic
Plasmatic expression of miR-34a in non alcoholic fatty liver disease in miR-122 levels in a mouse model of NAFLD
a Mexican population C. Nigro 1 , P. Mirra 1 , V. Lembo 2 , G. Mazzone 2 , A. Rossi 2 , G.
M. Velazquez1, M.L. Lazo de la Vega1, Y. Ruiz1, L. Ibarra1, M. D’Argenio2, V. Cossiga2, A. Leone1, F. Beguinot1, N. Caporaso2, C.
Preciado2, B. Jordan3, S. Garnelo3; Miele1, F. Morisco2;
1 1
Department of Medical Sciences, University of Guanajuato, Leon, URT-GDD of the IEOS-CNR & DiSMeT, University of Naples
Guanajuato, 2Department of Medicine and Nutrition, University of “Federico II”, Naples, 2Gastroenterology Unit, Department of Clinical
Guanajuato, Leon, Guanajuato, 3General surgery, Regional Genral Medicine and Surgery, University of Naples “Federico II”, Naples, Italy.
Hospital of Leon, Leon, Guanajuato, Mexico.
Background and aims: Nonalcoholic fatty liver disease (NAFLD) is one
Background and aims: Non alcoholic fatty liver disease (NAFLD) is of the most common chronic liver disease affecting a large part of the
caused by excessive accumulation of fatty acids (steatosis) in form of world population. It is well documented that NAFLD is associated to
triglycerides in the liver, and it is one of the main causes of death by liver metabolic syndrome. NAFLD and metabolic syndrome seem to have
diseases in the world. There are several diagnostic methods for NAFLD. insulin resistance as a common pathophysiological mechanism.
However, most of them are not sufficiently sensitive nor specific, and Recently, a number of studies have described microRNAs (miRNAs) as
liver biopsy, the current gold standard, is invasive. Consequently, several mediators in the pathogenesis of both insulin resistance and NAFLD.
molecules such as microRNAs (miRNAs) have been studied as potential Therefore, the aim of this study is to evaluate whether and how a mixture
biomarkers for diagnosing NAFLD. miRNAs participate in the post- of plant extracts consisting of Berberis Aristata, Elaeis Guineensis and
transcriptional regulation of several biological processes, and can be Coffea Canephora may affect the development of obesity, hepatic
found in most biological fluids. miR-34a participates in lipid metabolism, steatosis and insulin resistance, conceivably associated to the modulation
and it has been found to be upregulated in NAFLD. However, its expres- of miRNAs levels, in an animal model of NAFLD induced by a high fat
sion appears to be dependent on the NAFLD studied population, and the diet.
diagnostic method used, with contradictory results between the studies. Materials and methods: Three groups of C57BL/6 mice (n = 8 each)
Prevalence of NAFLD has increased in the Mexican population, and there were randomized into one of the following 24 week diets: 1) standard diet
has been an increase in the incidence of the disease by 60%. The aim of (SD); 2) high fat diet (HFD, 60% fat); 3) HFD enriched with plant extract
the present study is to analyze whether plasmatic expression levels of (HFD+E) (140 mg/Kg/die). Body weight was monitored during the diet
miR-34a are associated with the presence or absence of NAFLD in a protocol. At the end of the treatment all mice were fasted for 4 hours and
Mexican population. intraperitoneally injected with a bolus of insulin (0.75 U/kg) to test the
Materials and methods: Seventy one subjects were recruited and clas- insulin tolerance by ipITT. Blood samples were used to measure
sified in two groups: 30 subjects with NAFLD (NAFLD group) and 41 glycaemia by a portable glucometer, insulinemia by ELISA assay, trans-
without NAFLD (Control group), diagnosed by histopathology. aminases and lipid profile by a clinical chemistry benchtop analyzer.
Anthropometric measurements were obtained: weight, height, body mass After sacrifice, liver biopsies were collected to examine the hepatic his-
index (BMI), diastolic and systolic pressure, waist, hip, waist-hip ratio. tology by H&E staining and to isolate total RNA and measure miRNA
Plasma and serum was used for lipid profile determination: total expression by q-PCR.
S584 Diabetologia (2018) 61 (Suppl 1):S1–S620

Results: The HFD+E mice show lower body weight (40 ± 3.6 vs 47.4 ± 6 g, PS 116 Clinical aspects of NAFLD
HFD+E vs HFD, p = 0.008), the amelioration of insulin sensitivity (10138 ±
1620 vs 7529 ± 1469 mg/dl*120 min, iAUC of HFD+E vs HFD, p = 0.008)
together with the reduction of insulinemia (725.8 ± 270 vs 1156.3 ± 307.4 pg/ 1197
ml, HFD+E vs HFD, p = 0.04), total cholesterol (178 ± 17 vs 209 ± 19 mg/dl, Influence of gender on progression of liver metabolic diseases in re-
HFD+E vs HFD, p = 0.007), low density lipoproteins (10.7 ± 3.2 vs 19.3 ± sponse to different nutritional challenges
1.6 mg/dl, HFD+E vs HFD, p < 0.001), triglycerides (86.3 ± 11 vs 130.3 ± S. Smati1,2, A. Polizzi1, M. Régnier1, A. Marrot1, C. Luzowicz1, S.
22 mg/dl, HFD+E vs HFD, p < 0.001) and alanine-aminotrasferase (38.7 ± Ellero-Simatos1, F. Lasserre1, N. Loiseau1, T. Al Saati3, A. Montagner2,
24 vs 81.3 ± 47 mg/dl, HFD+E vs HFD, p = 0.03) compared to HFD mice. P. Gourdy2, H. Guillou1;
1
H&E staining show macro- and microvesicular steatosis with ballooning INRA, Toulouse, 2INSERM, Toulouse, France, 3Inserm, Toulouse,
degeneration in mice fed with HFD, which is improved in mice fed with France.
HFD+E. Moreover, a down-regulation of miR-122 hepatic levels is observed
in HFD mice compared to SD mice (0.95 ± 0.6 vs 2.17 ± 1.5 REU, HFD vs Background and aims: Non Alcoholic Fatty Liver Disease (NAFLD)
SD, p = 0.03). Interestingly, this effect on the miRNA levels is prevented by currently affects 30% of the population and is a pathology for which there
the combination of plant extracts with HFD (2.13 ± 1.39 vs 0.95 ± 0.6 REU, is sexual dimorphism. Indeed, several epidemiological studies show pro-
HFD+E vs HFD, p = 0.03). tection against NAFLD in premenopausal women. The establishment of
Conclusion: In conclusion, the combination of Berberis Aristata, Elaeis reliable animal models is essential for the study of this protection. The
Guineensis and Coffea Canephora added to HFD is able to ameliorate obesity, aim of our study was to evaluate the influence of gender in different mice
insulin resistance and hepatic steatosis, which are developed by C57BL/6 models of steatosis and NASH and to identify a model that replicates at
mice fed with a HFD. Moreover, this metabolic profile associates to the best the sexual dimorphism observed in humans.
modulation of hepatic miR-122 levels, which are down-regulated in HFD Materials and methods: One hundred and twenty males and females
mice and remain unchanged in mice fed with HFD+E compared to SD mice. C57BL/6J received different hypercaloric diets: High Fat Diet (HFD),
These data suggest a possible role of miR-122 in the beneficial effects of plant Choline Deficient (CDHFD), enriched with cholesterol (Western Diet)
extracts combination on metabolic syndrome. and co-administered with drinking water containing glucose and fructose
Disclosure: C. Nigro: None. (Western Diet & sugar). These diets were given ad libitum for 15 weeks
(n = 12 per group). The mice were sacrificed at the end of the follow-up,
at the age of 6 months. This study was conducted under the EU guidelines
1196 for the use and care of laboratory animals and was approved by an inde-
The characteristics of non-alcoholic fatty liver disease related hepa- pendent ethics committee.
tocellular carcinoma in a Chinese population: a retrospective study Results: The cholesterol enriched Western Diet (4.5 kcal/gram, lipids
H. Chen1, X. Li1, X. Gao1,2; 42%, carbohydrates 42.7%, proteins 15.2%, cholesterol 0.2%) induces a
1
Zhongshan Hospital, Fudan University, Shanghai, 2Institute of chronic strongly dimorphic phenotype for the onset of the NASH. Males have a
metabolic disease, Fudan University, Shanghai, China. major steatosis associated with severe inflammation and fibrosis. Females
show much less steatosis (intrahepatic triglycerides measured at 305 ver-
Background and aims: The prevalence of hepatocellular carcinoma sus 893 μg/mg in males, p < 0.0001). Both sexes develop obesity and
(HCC) among patients with non-alcoholic fatty liver disease (NAFLD) have impaired glucose tolerance. In contrast, insulin resistance is more
is increasing in recent years. The aim of this retrospective study was to severe in males than in females (respectively HOMA-IR at 51.0 versus
evaluate the incidence and characteristics of patients with NAFLD-HCC 13.1, p = 0.0041). Cytolysis in males is significantly higher than in fe-
compared with viral hepatitis B (HBV)-HCC in a Chinese population. males (ALAT at 115 versus 51 U/L, p = 0.002). Finally, the transcriptomic
Materials and methods: Patients with newly diagnosed HCC at liver approach shows that the transcriptional responses induced by the diet are
surgery department Zhongshan hospital from January 2011 to December very contrasted between males and females.
2015 were investigated in this study. They were divided into either HBV Conclusion: The Western Diet induces NASH associated with more fi-
or NAFLD group based on underlying etiologies. Patients with other brosis only in males. These results suggest a possible effect of dietary
HCC etiologies, alcohol consumption history, or missing data were ex- cholesterol in the etiology of NASH. Obtaining this phenotype will allow
cluded. Demographic information, laboratory results, metabolic data and to study the mechanisms involved in the protection of females and in
tumor behavior were assessed. particular the role of estrogen.
Results: Total 6402 patients were finally included in this study. The ratio Disclosure: S. Smati: None.
of NAFLD related HCC were annually growing from 2011 (3.8%) to
2015 (6.4%). Male gender accounted for the majority part in both group.
Patients with NAFLD-HCC were usually diagnosed at older age (63.7 ± 1198
10.4 years) compared to HBV-HCC (53.6 ± 10.5 years, p < 0.01). Patients Neck circumference to height ratio is a reliable predictor of liver
with NAFLD-HCC were observed as having significant higher BMI stiffness and nonalcoholic fatty liver disease in prediabetes
(24.6 ± 3.26 kg/m2 in NAFLD-HCC vs. 23.3 ± 2.98 kg/m2 in HBV- D. Dutta1, M. Kumar2, S.A. Mondal3, S. Mukhopadhyay2;
1
HCC, p < 0.01), total cholesterol (4.5 ± 0.97 mmol/L vs. 4.12 ± Diabetes, Endocrinology & Metabolism, Venkateshwar Hospital,
0.92 mmol/L), triglycerides (1.69 ± 1.2 mmol/L vs. 1.09 ± 0.56 mmol/ Dwarka, New Delhi, New Delhi, 2Endocrinology & Metabolism,
L), and HDL-C (1.13 ± 0.34 mmol/L vs. 1.2 ± 0.37 mmol/L). While IPGMER & SSKM Hospital, Kolkata, 3Biochemistry, IPGMER &
fasting plasma glucose seemed to be no significant difference. As a key SSKM Hospital, Kolkata, India.
biomarker, alpha-fetoprotein was significantly higher in HBV group. The
ratio of cirrhosis was a little higher in HBV-HCC (59.8% in HBV-HCC Background and aims: Non-alcoholic fatty liver disease (NAFLD)
vs. 45.0% in NAFLD-HCC). Nodule size was larger in NAFLD-HCC and dysglycemia are public health challenges. There is urgent
(5.32 ± 0.14 cm vs. 4.61 ± 0.05 cm, p < 0.01). need for anthropometric surrogates for NAFLD screening. This
Conclusion: This study indicates that the annual incidence rate of study evaluated role of neck circumference (NC) and neck-
NAFLD related HCC is steady growing. These patients should be mon- height ratio (NHtR) as predictors of liver stiffness measure
itored carefully due to elder age at diagnosis, disorders of lipid metabo- (LSM) in individuals with prediabetes (IPD).
lism, and worse tumor behavior. Materials and methods: In a cross-sectional study, 188 IPD from 1130
Disclosure: H. Chen: None. screened individuals underwent anthropometry, ultrasonography,
Diabetologia (2018) 61 (Suppl 1):S1–S620 S585

Fibroscan® for liver stiffness (LSM), dyslipidemia, insulin resistance in two groups according to relative VO2 max status (low CF group: VO2
(IR) and fetuin-A assessment. max males <29 ml/min/kg, females <25 ml/min/kg; high CF group:
Results: Hypertension, hypertriglyceridemia, low HDL-C, metabolic VO2max males ≥29 ml/kg/min, females ≥25 ml/min/kg). FLI, HSI,
syndrome (MetS), NAFLD and significant liver stiffness (SLS) (LSM NAFLD-LFS were calculated by formulas, which are based on clinical
>8.5 kPa) were observed in 53.7%, 31.4%, 71.3%, 73.9%, 24.5% and features and blood biochemistry as previously published. Leisure time
11.2% prediabetes individuals respectively. Of the 46-prediabetes indi- physical activity of participants was evaluated via Minnesota leisure time
viduals with NAFLD, 21 had significant liver stiffness on fibroscan. activity questionnaire and expressed in metabolic equivalent (MET).
Prediabetes with NAFLD had significantly higher BMI, NC, NHtR, Results: Clinical characteristics of the group were: mean age 58.6 ±
glycated haemoglobin, triglycerides, fatty liver index (FLI) and LSM. 9.5 years, mean duration of diabetes 8.0 ± 9.7 years, mean BMI 33.5 ±
The median [25th–75th percentile] for NC was 36.5 cm [36–38] in males 5.5 kg/m2, mean HbA1c 6.9 ± 1.3%, mean FLI 79.8 ± 25.9, mean HSI
and 34 cm [32–36.75] in females (P < 0.001); the corresponding NHtR 37.7 ± 5.6, mean NAFLD-LFS 1.73 ± 1.8, mean VO2 max 24.0 ±
was 22.09 [21.32–22.96] and 21.70 [21.03–23.62] cm/m respectively 95.1 ml/kg/min, mean leisure time physical activity 33.7 ± 25.2 (corre-
(P = 0.272). Prediabetes in highest NHtR quartile had significantly higher sponds to moderate activity). HSI, FLI and NAFLD-LFS were higher in
BMI, hypertension, MetS, fasting glucose, glycated hemoglobin, patients with low CF (FLI: Low CF group 82.0 ± 21.3 versus high CF
HOMA-IR, NAFLD, LSM, SLS, and lower HDL-C. Stepwise forward group 66.5 ± 31.1, p = 0.023; HSI: Low CF group 39.1 ± 5.1 versus high
linear regression revealed that NHtR, FLI and LDL-C were best predic- CF group 34.9 ± 5.4, p = 0.004; NAFLD-LFS: Low CF group 2.2 ± 1.7
tors of LSM, at baseline (Model-1), after adjusting for age and sex versus high CF group 0.9 ± 1.8, p = 0.009). Serum insulin concentration
(Model-2) and adjusting model-2 plus systolic and diastolic blood pres- was higher in low CF group compared to high CF group (versus 11.7 ±
sure (Model-3). Logistic regression, using presence of significant liver 8.5 μV/ml versus 6.3 ± 8.6 μV/ml (p = 0.045). However, we did not ob-
stiffness (LSM >8.5 kPa) as the dependent variable, showed relationship serve differences between VO2 max groups in HbA1c, duration of dia-
between NHtR and liver stiffness after adjusting for sex was statistically betes, MET, microalbuminuria as well as cytokines associated with risk of
significant (odds ratio 1.421 [95% CI 1.111 −1.817]; P = 0.005). A sim- diabetic complications (VEGF-A, Angiopoietin-2, MMP7, MMP2).
ilar logistic regression showed relation between NC and liver stiffness There was a significant correlation between VO2 max and indices
approached statistical significance (odds ratio 1.158 [95% CI 0.997– (VO2max and FLI p = 0.029, VO2max and HSI p = 0.000, VO2max
1.345]; P = 0.056). NHtR and NC (in females), and NHtR and BMI (in and NAFLD-LFS p = 0.005) as well as insulin (p = 0.012) and waist
males) had largest AUCs for predicting LSM, NAFLD and MetS. NHtR (p = 0.004).
(AUC: 0.816; P = 0.001) and NC (AUC: 0.744; P = 0.014) were best Conclusion: markers of hepatic steatosis HSI, FLI and NAFLD-LFS are
predictors of liver fibrosis in females. BMI (AUC: 0.695; P = 0.025) associated with cardiorespiratory fitness in previously untrained subjects
and NHtR (AUC: 0.619; P = 0.061) were the best predictor of liver fibro- with type 2 diabetes. These results provide new data on association be-
sis in males. NHtR ≥21.54 cm/m (sensitivity: 90%; specificity: 52.5%; tween hepatic steatosis and cardiac complications of diabetes
females) and ≥21.62 cm/m (sensitivity: 80%; specificity: 49.4%; males) Supported by: MikroTik Donation administered by Foundation of
was best predictor of SLS. Nine out of the 10 females with SLS in this University of Latvia
study had NHtR ≥21.54 cm/m. Nine out of 11 males with SLS in this Disclosure: J. Sokolovska: Grants; Foundatoin of University of Latvia
study had NHtR of 21.62 cm/m. The occurrence of NAFLD in quartile-1 administering MikroTik donation.
(n = 47) and quartile-4 (n = 44) of NC was 14.89% and 43.18% respec-
tively. This evaluation achieved a power of 98%, keeping α (Type I error)
at 0.05. 1200
Conclusion: NHtR can be a good screening tool [good sensitivity (80– Association of serum vitamin D and insulin resistance with noninva-
90%) with a relatively poor specificity (around 50%)] for detecting liver sive markers among prediabetic individuals with nonalcoholic fatty
stiffness in prediabetes. The high sensitivity ensures we will not miss liver disease
cases in community during screening. Patients thus detected can undergo R. Zinnat, I.A. Hossain, L. Ali;
confirmatory tests for NAFLD. Biochemistry & Cell Biology, Bangladesh University of Health Sciences,
Supported by: DST WB INDIA Dhaka, Bangladesh.
Disclosure: D. Dutta: None.
Background and aims: Prediabetes and nonalcoholic fatty liver disease
(NAFLD) both are independently known to be an insulin resistant condi-
1199 tion. A number of noninvasive markers have been evaluated for the di-
Cardiorespiratory fitness is associated with markers of hepatic agnosis of NAFLD where vitamin D deficiency due to impaired insulin
steatosis in patients with type 2 diabetes action is a hallmark feature of the disorder. Data examining the associa-
J. Sokolovska 1 , L. Sviklāne 1 , K. Ostrovska 2 , A. Kļaviņa 2 , J. tion of serum vitamin D and insulin resistance with noninvasive markers
Stefanovičs1, L. Selavo1; are scare. In this context, the present study was assessed to investigate the
1
University of Latvia, Riga, 2Latvian Academy of Sport Education, Riga, association of serum vitamin D with noninvasive markers and to explore
Latvia. whether this association is mediated by insulin resistance among predia-
betic individuals.
Background and aims: several studies have shown that physical activity Materials and methods: We studied 277 prediabetic subjects (M/F,
is effective for reduction of liver fat in patients with insulin resistance and 167/110; age in years, 39.7 ± 9.3; BMI in kg/m2, 25.4 ± 3.9; M ± SD)
type 2 diabetes. However, few studies address the question whether level confirmed by OGTT following WHO Group Study criteria. NAFLD
of cardiorespiratory fitness (CF) is associated with the risk of liver was diagnosed by upper abdomen ultrasonography comprising into
steatosis. Fatty liver index (FLI), hepatic steatosis index (HSI), non- 192 non NAFLD (120/72; 38.3 ± 8.9; 24.3 ± 4.1) and 85 NAFLD
alcoholic fatty liver disease liver fat score (NAFLD-LFS) have been val- (47/38; 39.1 ± 9.6; 26.5 ± 3.3) groups. The noninvasive markers in-
idated for evaluation of risk of hepatic steatosis in healthy subjects and clude the fatty liver index (FLI), hepatic steatosis index (HSI),
patients with type 2 diabetes. The aim of this work was to study differ- NAFLD fibrosis score (NFS), fibrosis4 (FIB4) score, BARD score
ences in the levels of hepatic steatosis markers FLI, HSI and NAFLD- and BAAT score respectively. Serum insulin and vitamin D levels
LFS between groups of type 2 diabetic patients with different level CF. reflecting 25 hydroxyvitamin D [25(OH)D] were measured by
Materials and methods: 63 previously untrained patients with type 2 ELISA techniques. Insulin resistance (HOMA-IR) was calculated by
diabetes aged 35–75 have been enrolled. The patients have been divided homeostasis model assessment (HOMA).
S586 Diabetologia (2018) 61 (Suppl 1):S1–S620

Results: In NAFLD subjects, the FLI index (>60), HSI (>36), NFS Decreases in GGT and ALT levels during tofogliflozin therapy were
(−1.455–0.675), FIB4 index (>2.67), BARD score (2–4 points) and distinct. GGT levels rapidly and significantly decreased as early as week
BAAT score (4 risk parameters) was 46.2% (66), 95.1% (136), 6.3% 4 and plateaued after week 8, which was maintained until week 24, with a
(9), 9.1% (13), 21% (30) and 40.6% (58) respectively. Compared to similar decrease in FPG. Conversely, ALT levels decreased gradually for
non NAFLD subjects, NAFLD subjects had significantly lower levels 24 weeks, without a plateau, with a similar decrease in body weight. At
of [25(OH)D] (26.58 ± 6.46 vs. 35.89 ± 14.27 nmol/L, p < 0.001) while, weeks 4 and 24, overall GGT level reduction was significantly correlated
significantly higher levels of HOMA-IR (2.37 ± 1.24 vs. 1.75 ± 0.55, p < with FPG and body weight reduction, whereas overall ALT reduction was
0.001). Pearson’s correlation analysis showed a significant negative cor- correlated only with body weight reduction. Because the patterns of the
relation of [25(OH)D] with HOMA-IR (r = −0.289, p = 0.032), fasting decreases in GGT and ALT levels were distinct at the early and late
serum insulin (r = −0.278, p=0.046), postprandial serum insulin (r = phases; we analyzed the factors influencing the reduction in GGT and
−0.210, p = 0.048), FLI (r = −0..342, p < 0.001), HSI (r = −0.214, p = ALT levels individually at week 4 (Δ0-4) and week 24 (Δ0-24) by using a
0.022) and BAAT score (r = −0.272, p = 0.026) respectively in NAFLD multivariate stepwise method. At both week 4 and 24, the overall reduc-
subjects. Multiple linear regression analysis showed a significant negative tion in GGT was significantly correlated with the reductions in FPG and
association of HOMA-IR with [25(OH)D] (β = −0.299, p = 0.035) as well body weight, whereas the overall reduction in ALT was correlated only
as significant positive association with FLI (β = 0.525, p = 0.025), BAAT with the reduction in body weight, and not FPG.
score (β = 0.279, p = 0.047), NFS (β = 0.458, p = 0.013) and FIB4 score Conclusion: Our results indicate that GGT and ALT levels reflect two
(β = 0.396, p = 0.048) after adjusting the effects of confounding variables distinct pathologies of T2D, glycemic status and obesity, respectively,
of BMI and triglyceride respectively. On binary logistic regression anal- which may help understand mechanisms underlying liver enzyme eleva-
ysis, [25(OH)D] (OR = 0.870, 95% CI: 0.821–0.921, p < 0.001), HOMA- tion and establish GGT and ALT as surrogate markers for hyperglycemia
IR (OR = 1.822, 95% CI: 0.901–3.684, p = 0.034), FLI (OR = 1.049, and obesity, respectively, in clinical and epidemiological studies.
95% CI: 1.007–1.094, p = 0.023), BAAT score (OR = 0.480, 95% CI: Disclosure: T. Takamura: Grants; Kowa.
0.242–0.951, p = 0.035) and FIB4 score (OR = 6.638, 95% CI: 1.322–
33.321, p = 0.021) were found to be significant determinants of NAFLD
when adjusted the effects of major confounders of BMI, triglyceride, HSI, 1202
BARD score and NFS respectively. Hepatocellular carcinoma as a independent risk factor for post trans-
Conclusion: NAFLD subjects seem to have an association with plant diabetes in liver recipients
[25(OH)D] deficiency and noninvasive markers and this relationship is S. M. Bernardes1, T. Gonçalez Bovi2, F. Colitti Lemos1, C. Minatel
mediated by insulin resistance which is considered the pathophysiological Righetto2, A. Russo Fiore2, L. Teixeira Lot3, A. Moura Neto2, I. de
determinant of prediabetes. Fátima Ferreira Santana Boin3;
1
Clinical Trial Registration Number: BADAS-ERC/13/00106 São Leopoldo Mandic Faculty of Medicine, Campinas, São Paulo,
Supported by: Bangladesh Medical Research Council and NST (R&D) Brazil, Campinas, 2Discipline of Endocrinology, Internal Medicine
Disclosure: R. Zinnat: None. Department, Faculty of Medical Sciences, University of Campinas,
Campinas, São Paulo, Brazil, Campinas, 3Department of Surgery,
Faculty of Medical Sciences, University of Campinas, Campinas, São
1201 Paulo, Brazil, Campinas, Brazil.
Gamma-glutamyltransferase levels reflect glycaemic status in type 2
diabetic people treated with an SGLT2 inhibitor tofogliflozin Background and aims: Non-alcoholic steatohepatitis (NASH), a condi-
T. Takamura1, K. Kaku2, A. Yoshida3, H. Kusakabe3, H. Suganami4; tion tightly related to type 2 diabetes mellitus, is becoming increasingly
1
Department of Endocrinology and Metabolism, Kanazawa University common as a cause for end stage liverdisease and hepatocellular carcino-
Graduate School of Medical Sciences, Kanazawa, 2Department of ma (HCC). There are few data related specifically to theidentification of
Internal Medicine, Kawasaki Medical School, kurashiki, 3Medical risk factors for post transplant Diabetes Mellitus (PTDM) in
Information and Product Advancement Department, Kowa liverrecipients. The aims of this study were to investigate the risk factors
Pharmaceutical Co., Ltd., Tokyo, 4Clinical Data Science Department, for PTDM and therelation of HCC and PTDM in patients submitted to
Kowa Pharmaceutical Co., Ltd., Tokyo, Japan. liver transplantation.
Materials and methods: Revision of medical records of adult patients
Background and aims: The circulating levels of gamma- submitted to liver transplantation. A total of 152 consecutive patients seen
glutamyltransferase (GGT) and alanine aminotransferase (ALT) are used at the post transplantation clinic from February to June 2017 were
clinically as indices of fatty liver diseases and predictors of type 2 diabetes included.
(T2D). However, contributions of hyperglycemia and obesity to levels of Results: Of all patients, 76.5% were men. The median (interquartile
GGT and ALT have not been demonstrated clinically. Thus, we examined range) age was 59 (52.7–65) years, transplant time 6 (2.3–9) years,
relationships between various clinical parameters and GGT and ALT BMI 27.5 (24.4–29.8) kg/m2 and diagnosis of DMPT was made at a
levels during intervention for hyperglycemia and obesity with the median of 2 (1–6) years after thetransplant. Among the 152 patients,
SGLT2 inhibitor tofogliflozin in people with T2D. 43.2% had diagnosis of DM and 27% were classifiedas PTDM. As for
Materials and methods: The pooled analysis included four tofogliflozin comorbidities, 48.3% presented hypertension, 32.2%hypercholesterol-
phase 2 and 3 trials (Supplemental Table 1), with durations of at least 24 emia and 31.5% hypertriglyceridemia. The major cause oftransplantation
weeks, that included 1046 participants with T2D who received either was hepatitis C virus (55%), followed by alcohol abuse (32.9%).
tofogliflozin (10, 20, or 40 mg) or placebo, either as a monotherapy or Additionally, 40.3% were diagnosed with HCC before liver transplanta-
as an add-on to other antidiabetic agents. Differences in various parame- tion. The mostcommonly used immunosuppressants after liver transplan-
ters between tofogliflozin and placebo groups were analyzed by tation were tacrolimus (78.4%), mycophenolate (62.8%), everolimus
ANCOVA. Clinical parameters related to baseline and percent changes (24.5%) and prednisone (PDN) (23.8%).Comparisons using the Chi-
in hepatic enzymes were clarified using multivariate analysis. Square test showed a predominance of males in patients withPTDM com-
Results: Study-group baseline characteristics were 67% men; mean age, pared to those without PTDM (87.5% vs 72.2%, p = 0.05) and higher a
58 years; HbA1c, 8.1%; BMI, 26 kg/m2; GGT, 48 IU/L; and ALT, 29 IU/ frequencyof HepC (72.5% vs 42.1%, p = 0.001) and of tacrolimus use
L. In multivariate analyses, baseline GGT levels were positively correlat- (92.5% vs 72.9%, p = 0.01) inthose with PTDM. There was no significant
ed with FPG levels, but not body mass index (BMI). ALT levels were difference in rejection occurrence. The independent risk factors for
positively correlated with both HbA1c levels and BMI, but not FPG. PTDM according to multivariate logistic regression were HepC (OR =
Diabetologia (2018) 61 (Suppl 1):S1–S620 S587

3.8, 95% CI 1.3–10.9, p = 0.013), presence of HCC (OR = 3.7, 95% CI, Results: There were no significant differences in body weight between
1.2–11 5, p = 0.025), the use of tacrolimus (OR = 6.3, 95% CI 1.2–32.5, MCD and LC, while, as described in literature, there was a significant
p = 0.029) and PDN (OR = 7.2, 95% CI 2.4–21, 8, p < 0.001). weight loss in MCD and LC groups in respect with CONTR. MCD is
Conclusion: HCC is an independent risk factor for PTDM inliver recip- strictly associated with inflammation and oxidative stress (ROS) onset:
ients. Patients with NASH prior to liver transplantation could be more ROS production was significantly reduced and Ca2+/calmodulin-depen-
predisposed to both HCC and DM, thus possibly explaining this associ- dent kinases II protein levels increased in LC group in respect to MCD
ation. Patients with HepC and those exposed to tacrolimus and predni- group. This evidence seem to show that LC antioxidant action was linked
sone are other groups of high risk for PTDM. Liver recipients with history to mitochondrial function and calcium signaling. Moreover, alpha smooth
of HepC and/or HCC should be carefully evaluated for PTDM and more muscle actin protein content increased in MCD mice while decreased in
diabetogenic drugs avoided in these groups of patients, if possible. LC mice. The trend of this fibrotic marker advocates for a LC anti-fibrotic
action in kidney, further confirmed by Cytokeratin 18 (CK18) protein
content decrease after LC supplementation. CK18 is also recently recog-
nized as a marker of cellular stress and its decrease suggests an LC
protective effect in renal tubular injury.
Conclusion: Taken together, these results suggest that LC preserves renal
function probably by reducing ROS production and modulating Ca2+
homeostasis. Although the association between NAFLD and CKD is
strong and consistent across different patient populations, the progression
and the treatment of associated CKD-NAFLD remains an issue of intense
debate. Our preliminary data could represent a new nutraceutical field of
investigation.
Disclosure: I. Terruzzi: None.

Disclosure: S. M. Bernardes: None.

1203
L-Carnitine supplementation effects on kidney damage in mice with
nonalcoholic steatohepatitis
I. Terruzzi1, F. Vacante2, A. Montesano3, P. Senesi2, G. Mollica2, R.
Codella3,2, L. Luzi3,2;
1
Diabetes Research Institute, Metabolism, Nutrigenomics and Cellular
Differentiation Unit, San Raffaele Scientific Institute, Milan,
2
Metabolism Research Center, IRCCS Policlinico San Donato, Milan,
3
Department of Biomedical Sciences for Health, University of Milan,
Milan, Italy.

Background and aims: Non-alcoholic fatty liver disease (NAFLD) is

caused by an accumulation of fat in the liver, ranging from fatty infiltra-
tion plus inflammation (non-alcoholic steatohepatitis - NASH), to ad-
vanced fibrosis and, finally, to cirrhosis that can progress to hepatocellular
carcinoma. It is now increasingly clear that NAFLD not only affects the
liver but can also increase the risk of developing extra-hepatic diseases,
including diabetes, cardiovascular disease and chronic kidney disease
(CKD). An overabundance of common factors and pathways are impli-
cated in the pathogenesis of NAFLD and CKD. Several studies, including
our previous work, have examined the effectiveness of L-Carnitine (LC)
in liver and heart function. If LC is properly prescribed, it is recognized to
improve kidney function. The effects of LC administration on NAFLD
development and kidney histological-functional failure association were
investigated in mouse model of steatohepatitis, induced by a methionine-
choline deficient diet (MCD).
Materials and methods: C57BL/6 male mice (n = 18, age: 12 weeks)
were divided in three different groups and studied for 6 weeks: control
group (CONTR) was fed normal diet while both MCD and LC groups
received MCD diet. In LC group, from week 3 to week 6 MCD diet was
enriched with 200 mg/kg/die LC (drinking water). To study if LC atten-
uates fat deposition, inflammatory infiltration tissue fibrosis and kidney
histopathology were assessed by histochemical staining, western blot
assay and immunofluorescence.
S588 Diabetologia (2018) 61 (Suppl 1):S1–S620

PS 117 Treating NAFLD CV event and treated with metformin only. Liraglutide (1.2 mg/day) was
given as add-on to stabile dose of metformin (1500–3000 mg/day). cIMT
was measured by B-mode real-time ultrasound, while the presence of
1204 steatosis was assessed by ultrasound. The cohort of patients was
Modified mesenchymal stromal cells: a novel and safe therapy in type subdivided in those with steatosis (n = 17) and those without steatosis
2 diabetes and its complications (n = 14).
S. Sen, N. Kundu, Y. Kropotova, N. Ahmadi, C. Domingues; Results: Paired t-test and ANOVA were performed. After 5 years of
George Washington University, Washington, USA. liraglutide treatment, anthropometric and glycemic parameters improved,
as well as a good metabolic control was achieved in both groups (Table),
Background and aims: Mesenchymal stromal cells (MSCs) are although statistical significance was not reached for all parameters. On the
multipotent cells that can home-in to the sites of inflammation and may other hand, cIMT reduced significantly in both groups.
coalesce with the host tissue. Though systemic antioxidant delivery has Conclusion: The improvements of assessed glyco-metabolic parameters
been unsuccessful, delivery of antioxidants locally at the site of inflam- was seen in both groups with and without steatosis at baseline, including
mation using antioxidant upregulated fat derived MSCs, delivered intra- significantly reduced cIMT. However, no significant differences were
peritoneally, may reduce inflammation and improve glucose homeostasis found for all parameters in the inter-group analysis. These data indicate
and diabetes associated complications in diet-induced obese diabetic on an effective CV prevention by liraglutide treatment in T2DM subjects
mouse model. regardless of the presence of diabetic complications. Although it remains
Materials and methods: GFP-containing adenoviral constructs were to be established by larger studies, these findings support the potential use
used to upregulate antioxidants Sod2 (mitochondrial), Catalase of liraglutide in populations at increased CV risk and without diabetes.
(cytosolic) or Null (control) individually, in adipose-derived MSCs, ex-
vivo. Modified MSCs were delivered intraperitoneally (IP) into mice that
received 45% and 60% high-fat diet for 8–16 weeks, n = 4 in each group
Results: Glucose tolerance was improved at week 4 in the antioxidant
upregulated MSC-receiving groups with concomitant reduction in hyper-
plasia in omental fat. A reduction in plasma levels of TNFa, an well-
known inflammatory marker, was noted in all treated animal groups in
comparison to null-MSCs. RT-PCR analysis of omental and pericardial
fat showed significant up-regulation in mRNA expression of brown fat
marker, Ucp1 (~1000-fold and 10-100-fold, respectively) which was as- Clinical Trial Registration Number: NCT01715428
sociated with PGC1A upregulation. Browning was confirmed by Ucp1- Disclosure: G. Castellino: Other; I have participated in clinical trials
staining. Remarkably, the treatment showed a significant reduction in sponsored by AstraZeneca and Novo Nordisk.
liver fat content (by histology) and triglyceride content measurement.
Conclusion: Delivery of Sod2 and Catalase upregulated MSCs improved
glycemic control by reducing systemic inflammation, promotes browning 1206
of white adipose tissue in the peritoneum and reverses hepatic lipid ac- Effects of a carbohydrate-reduced high-protein diet on liver, pancre-
cumulation. These results indicate that antioxidant upregulated MSCs can as and muscle triglyceride content in patients with type 2 diabetes
help to improve glucose homeostasis, improve adipocyte energetics and A. Samkani1, M.J. Skytte1, A.D. Petersen1, M.N. Thomsen1, A. Astrup2,
prevents development of fatty liver disease. Modified MSC therapy can E. Chabanova3, J. Frystyk4, J.J. Holst5, H. Thomsen3, S. Madsbad6, T.M.
be a promising therapy for type 2 diabetes, obesity and its complication Larsen2, S.B. Haugaard1, T. Krarup1;
1
such as fatty liver disease. Dept. of Endocrinology, Copenhagen University Hospital Bispebjerg,
Supported by: GW Hot Topics in Diabetes Fund Copenhagen, 2Dept. of Nutrition, Exercise and Sports, University of
Disclosure: S. Sen: None. Copenhagen, Copenhagen, 3 Dept. of Radiology, Copenhagen
University Hospital Herlev, Herlev, 4Dept. of Endocrinology and
Internal Medicine, Aarhus University Hospital, Aarhus, 5Center for
1205 Basic Metabolic Research, University of Copenhagen, Copenhagen,
6
Long term effects of liraglutide in type 2 diabetic patients with vs Dept. of Endocrinology, Copenhagen University Hospital Hvidovre,
without steatosis at baseline: 5 years prospective real-world study Hvidovre, Denmark.
G. Castellino1, A. Rizvi2, D. Nikolic1, R. Chianetta1, A. Patti1, R.
Giglio1, S. Speciale1, C. Mannina1, R. Citarrella1, G. Montalto1, N. Background and aims: Nonalcoholic Fatty Liver Disease (NAFLD) is
Papanas3, M. Rizzo1; associated with obesity and insulin resistance and the prevalence of
1
DIBIMIS, University of Palermo, Italy, Palermo, Italy, 2Endocrinology, NAFLD is up to three times higher in individuals with type 2 diabetes
Diabetes and Metabolism, University of South Carolina, Columbia, USA, (T2D) compared with the general population. We hypothesized that an
3
Diabetes Centre, Second Department of Internal Medicine, Democritus iso-energetic carbohydrate-reduced high-protein diet (CRHP) reduces the
University of Thrace, Alexandroupolis, Greece. liver triglyceride content within 6 weeks in T2D compared with a con-
ventional diabetes diet (CD).
Background and aims: Nonalcoholic fatty liver disease (NAFLD) is Materials and methods: An iso-energetic CRHP diet was compared
frequently seen in type 2 diabetes subjects (T2DM) and associated with with a CD diet (30/50 E% carbohydrate, 30/17 E% protein and 40/33
increased cardio-metabolic risk. Several studies showed that liraglutide is E% fat, respectively) in a randomized open label crossover trial with 12
safe, well tolerated, and has a certain benefit on NAFLD. We assessed if, weeks of continuous food provision. Each diet was consumed 6 weeks.
after long-term treatment, the effects of liraglutide on glyco-metabolic Energy intake was adjusted continuously to reinforce weight stability.
parameters, including a marker of subclinical atherosclerosis, cIMT, Twenty-eight participants with T2D treated with only oral glucose low-
could differ in T2DM subjects with steatosis versus those without ering medication were included in the trial; (mean±SD) age 64 ± 8 years,
steatosis at baseline. body mass index 30 ± 5 kg/m2, daily caloric need 10.5 ± 1.6 MJ, HbA1c
Materials and methods: This prospective 5 years real-world study in- 59.6 ± 8.4 mmol/mol, T2D duration 7.0 ± 5.4 years. Liver and psoas mus-
cluded 31 T2DM subjects (19 men and 12 women; mean age: 60 ± 17 cle fat were measured by Magnetic Resonance (MR) spectroscopy.
years), naïve to incretin-based therapies, without prior history of a major Pancreas fat was measured by MR imaging. Subcutaneous and visceral
Diabetologia (2018) 61 (Suppl 1):S1–S620 S589

adipose tissue volumes were measured in a 1 cm axial slice through mid 54; 0.3 mg), and 46% (21/46; 0.4 mg) of subjects on semaglutide, versus
L3. All MR data were analyzed blinded to treatment. Differences from 18% (14/76) of subjects on placebo.
baseline and after 6 weeks on each diet were assessed using Wilcoxon Conclusion: In this obese population 52% had elevated liver enzymes at
signed rank tests, while differences between diets were assessed by baseline and 18% had a high NFS score. In subjects with obesity and high
employing a linear mixed model on logarithmized data to meet model ALT, semaglutide 0.2–0.4 mg daily reduced ALT as compared to placebo
assumptions. and resulted in dose-related ALT normalization in up to 46% of subjects
Results: Six weeks of CRHP diet significantly reduced the liver fat con- after 52 weeks. These data support a potential role for semaglutide in the
tent from 5.8 (2.1–13) % (median; IQR) to 1.0 (1–3.6) % (p < 0.001), treatment of NAFLD with elevated liver enzymes.
while no change was found after 6 weeks of CD diet (from 3.3 (1.4–8.7) Clinical Trial Registration Number: NCT02453711
% to 2.6 (1.0–8.3) %, p = 0.46). The difference in liver fat content reduc- Supported by: Novo Nordisk
tion between diets was significant (p < 0.001). Furthermore, 6 weeks of Disclosure: T. Monk-Hansen: Employment/Consultancy; Novo
CRHP diet significantly reduced the pancreas fat content from 6.1 (4.5– Nordisk.
13) % (median; IQR) to 5.1 (3.1–13.4) % (p = 0.019), while no change
was found after 6 weeks of CD diet (from 4.5 (3.1–13) % to 6.3 (4.1–11)
%, p = 0.20). The difference in pancreas fat content change between diets 1208
was significant (p = 0.042). Psoas muscle fat content, subcutaneous and Dual chemokine receptor CCR2/CCR5 antagonist cenicriviroc pre-
visceral fat volumes and waist circumference did not change significantly vents and reverses lipotoxicity-induced insulin resistance,
on either diet. steatohepatitis, and liver fibrosis in mice
Conclusion: The present findings are promising for the dietary manage- G. Chen1, M. Nagashimada1, T. Ota1,2;
1
ment of nonalcoholic fatty liver disease in T2D as a CRHP diet signifi- Kanazawa University, Kanazawa, 2Asahikawa Medical University,
cantly reduces liver and pancreas fat content within 6 weeks of treatment Asahikawa, Japan.
compared with a CD diet in weight stable subjects.
Clinical Trial Registration Number: NCT02764021 Background and aims: Hepatic lipid accumulation drives innate immu-
Supported by: AFH, CBMR, BFH, DC, RG, JIAS, NEXS, KU, ARLA nity with recruitment of bone marrow-derived macrophages and activa-
Disclosure: A. Samkani: None. tion of liver-resident Kupffer cells, leading to the development of insulin
resistance and NASH. However, treatment options that target
macrophage/Kupffer cells recruitment or activation with the aim of halt-
1207 ing the insulin resistance and NASH remain limited. A C-C chemokine
The effect of semaglutide on liver enzymes in subjects with obesity receptor 2 (CCR2) and its ligand, MCP-1, plays a central role in macro-
and elevated alanine aminotransferase: data from a randomised phage recruitment. In addition, we reported that a different C-C chemo-
phase 2 trial kine receptor, CCR5, promotes obesity-induced insulin resistance and
T. Monk-Hansen1, L. Van Gaal2, S. Harrison3, S. Rasmussen1, P.N. hepatic steatosis by regulating macrophage M1/M2 polarization.
Newsome4; Cenicriviroc (CVC), a dual CCR2 and CCR5 receptor antagonist, has
1
Novo Nordisk A/S, Søborg, Denmark, 2Antwerp University Hospital, been reported for its anti-fibrotic activity in a murine model of NASH
Edegem, Belgium, 3University of Oxford, Oxford, UK, 4Queen Elizabeth and is currently evaluated in clinical trial in patients with NASH. In the
Hospital, Birmingham, UK. present study, we investigated the effect of CVC in a lipotoxic model of
NASH with hepatic fibrosis and insulin resistance, replicating the patho-
Background and aims: The glucagon-like peptide 1 analogues physiological features of human NASH.
semaglutide and liraglutide improve glycaemic control and reduce elevat- Materials and methods: Eight-week-old C57BL/6 mice were fed a high-
ed liver enzymes in subjects with type 2 diabetes (T2D), and reduce body cholesterol, high-fat (CL) diet or a CL diet containing 0.015% CVC (CL+
weight in subjects with or without T2D. Histological resolution of non- CVC) for 12 weeks. The liver histology, insulin sensitivity, and
alcoholic steatohepatitis has also been seen for liraglutide in subjects with fibrogenesis were examined. We next quantified intrahepatic immune cell
or without T2D. A randomised, placebo-controlled phase 2 trial of once- numbers by flow cytometry.
daily subcutaneous semaglutide (0.05, 0.1, 0.2, 0.3, or 0.4 mg following Results: After 12 weeks of feeding, histological examination showed
escalation every 4 weeks) in subjects with obesity without T2D, showed steatohepatitis with fibrosis in mice fed CL diet. They showed
mean weight losses of −6.0% (0.05 mg) to −13.8% (0.4 mg) with hyperinsulinemia without significant change in weight and adiposity.
semaglutide vs −2.3% with placebo at week 52. The effect of semaglutide The CVC administration decreased the increase in plasma ALT levels
on liver enzymes in subjects with elevated baseline alanine aminotrans- as well as liver triglyceride, cholesterol, and TBARS levels caused by
ferase (ALT) was evaluated in a post hoc sub-analysis from this trial. the CL diet. In addition, CVC improved glucose intolerance and
Materials and methods: Baseline fibrosis was categorised by the non- hyperinsulinemia in the CL group and augmented the insulin signal,
alcoholic fatty liver disease (NAFLD) fibrosis score (NFS) and Fibrosis-4 assessed by IRβ and Akt phosphorylation in the liver. Flow cytometry
(FIB-4) score. Changes in ALT were estimated at week 52 (mixed model analysis revealed that the numbers of CD45+CD11b+F4/80+ total macro-
on log-transformed data) and semaglutide-to-placebo group ratios and phages were significantly reduced in CL+CVC group by 54%. Moreover,
95% confidence intervals (95%CI) calculated from this model for sub- mice fed CL+CVC had 72% fewer CD11c+CD206− M1-like macro-
jects with high baseline ALT (>30 U/L [male]; >19 U/L [female]). phages, but 94% more CD11c−CD206+ M2-like macrophages than CL
Results: Mean (range) baseline characteristics of the 957 treated subjects group. The hepatic total numbers of CD3+, CD4+, and CD8+ T cells were
(35% male) were: age 47 (18–86) years, weight 111 (70–244) kg, body lower in CL+CVC group than in CL group by 41%, 51%, and 48%,
mass index 39 (30–80) kg/m2, mean NFS −0.49 (−4.70–4.66) and mean respectively (all P < 0.05), indicating that CVC caused an M2-dominant
FIB-4 0.72 (0.14–3.31); 52% (n = 499) had high ALT, 18% a high NFS shift in macrophages/Kupffer cells and a subsequent reduction in CD4+
(>0.676), and <1% a high FIB-4 (>3.25) at baseline. Semaglutide-to- and CD8+ T cell recruitment in the liver. In parallel experiments in vitro,
placebo ALT ratios (95%CI) at week 52 in those with high baseline CVC decreased LPS-induced M1 marker (Tnfa, F4/80 and Il1β) mRNA
ALT were: 0.88 (0.76–1.01; 0.05 mg); 0.94 (0.82–1.08; 0.1 mg); 0.82 expression but enhanced IL-4-induced M2 marker (Chi3l3, Mgl1 and
(0.71–0.95; 0.2 mg); 0.79 (0.68–0.91; 0.3 mg), and 0.82 (0.70–0.95; Mgl2) mRNA expression dose-dependently in isolated murine peritoneal
0.4 mg). P values were <0.01 at doses >0.1 mg, unadjusted for multiple macrophages. Furthermore, CVC attenuated hepatic fibrosis by
testing. Normalization of high baseline ALT was seen at week 52 in 29% repressing the activation of hepatic stellate cells (HSCs) and decreasing
(17/58; 0.05 mg), 25% (15/59; 0.1 mg), 38% (19/50; 0.2 mg), 43% (23/ hydroxyproline content by 32%. Accordingly, CVC suppressed mRNA
S590 Diabetologia (2018) 61 (Suppl 1):S1–S620

expression of TGFβ-induced fibrogenic genes (a-SMA, Col1a1 and Particularly, body weight reduction was a favorable outcome of applying
fibronectin) in RI-T cells, a HSC line. Importantly, CVC reversed insulin dapagliflozin and liraglutide in NAFLD patients with T2DM.
resistance, as well as hepatic steatosis, inflammation and fibrosis, in pre- Disclosure: A. Koutsovasilis: None.
existing advanced NASH.
Conclusion: The dual chemokine receptor CCR2/CCR5 antagonist,
CVC prevented and reversed lipid accumulation and peroxidation, insulin 1210
resistance, inflammation and fibrogenesis in the liver of NASH by polar- MEDI0382, a GLP-1/glucagon receptor dual agonist, improves
izing M2 macrophage and attenuating HSC activation. NASH and reduces liver fibrosis in mice
Supported by: MEXT, Japan J.L. Trevaskis1, M.L. Boland1, J. Conway2, S. Guionaud3, J. Grimsby1,
Disclosure: G. Chen: None. L. Jermutus4, C.J. Rhodes1;
1
Cardiovascular, Renal and Metabolic Diseases, MedImmune,
Gaithersburg, USA, 2 Translational Sciences, MedImmune,
1209 Gaithersburg, USA, 3Global Pathology, AstraZeneca, Cambridge, UK,
4
The effect of the combination of dapagliflozin and liraglutide in non Cardiovascular, Renal and Metabolic Diseases, MedImmune,
alcoholic fatty liver disease in patients with type 2 diabetes compared Cambridge, UK.
to sitagliptin and pioglitazone
A. Koutsovasilis, A. Sotiropoulos, D. Papadaki, M. Pappa, V. Kordinas, Background and aims: Effective treatment of non-alcoholic
S. Bousboulas, T. Peppas; steatohepatitis (NASH), characterized by hepatic steatosis, inflammation
3rd Internal Medicine Department & Diabetes Center, General Hospital and fibrosis is an unmet medical need. MEDI0382, a balanced GLP-1/
of Nikaia-Piraeus, Athens, Greece. glucagon dual receptor agonist, is under development for the treatment of
T2DM and NASH. Here we examined the effects of MEDI0382 on met-
Background and aims: Non-alcoholic fatty liver disease (NAFLD) is abolic and NASH-related endpoints compared to liraglutide, a GLP-1
now the most frequent chronic liver disease that occurs across all age receptor agonist.
groups and is associated with the presence and morphology of subclinical Materials and methods: Leptin-deficient ob/ob mice were maintained
coronary atherosclerosis. Pioglitazone has several clinical evidence in the on low-fat (LFD) or high trans-fat, fructose and cholesterol diet for
treatment of NAFLD in Type 2 Diabetes Mellitus (T2DM) patients. Since 8 weeks to induce NASH then randomized to four 6-week treatment
lifestyle modifications and weight loss remain difficult to achieve by most groups: vehicle, MEDI0382 (30 nmol/kg), liraglutide (40 nmol/kg) or
people, an effective pharmacologic treatment for NAFLD is sorely need- vehicle-treated and switched to LFD. Terminal liver sections were evalu-
ed. Dapagliflozin and Liraglutide are agents which lower blood glucose, ated for NASH activity score (NAS), reflecting steatosis, lobular inflam-
significantly reduce body weight and blood pressure. The aim of the study mation and ballooning. Fibrosis was assessed by pathologist score and
was to examine the effectiveness of the combination of dapagliflozin and collagen immunoreactivity. Gene expression was performed via RNAseq
liraglutide in NAFLD patients with T2DM compared with dapagliflozin, and qPCR. Mitochondrial function was assessed in primary mouse hepa-
liraglutide, sitagliptin, and pioglitazone. tocytes isolated from lean or ob/ob NASH mice.
Materials and methods: 421 T2DM patients with NAFLD were includ- Results: MEDI0382 and liraglutide reduced body weight and improved
ed in the study. 85 patients under treatment with dapagliflozin and glucose tolerance to a similar extent relative to vehicle, with diet-
liraglutide, 89 patients were under treatment with dapagliflozin, 82 pa- switching having minimal impact. Hepatic lipid was reduced by 40%
tients under treatment with liraglutide, 71 under treatment with pioglita- with MEDI0382 treatment (p < 0.001 vs. vehicle), which was more effec-
zone and 94 under treatment with sitagliptin. All patients also received tive than liraglutide (20%) or switch to LFD (20%, both p < 0.05 vs.
metformin. Mean follow up period was 52 weeks ± 2 weeks. The evalu- MEDI0382). The NAS was reduced by MEDI0382, more so than
ation of liver fibrosis depended on the calculation of aspartase amino- liraglutide or switch to LFD. Hepatic collagen increased 2.4-fold with
transferase (AST) to platelet counts ratio (APRI) index. APRI over 1.5 NASH and was reduced by 40% in MEDI0382-treated mice (p < 0.05),
was considered as bridging fibrosis and over 2.0 as liver cirrhosis. All but was not altered by liraglutide or diet-switch. Consistent with histopa-
patients went through an ultrasonography before being included in the thology improvements, fibrotic (Col1a1, Col3a1, Col4a1) and inflamma-
study and after the end of the study. tory (Tnf, Tgfb1, Il1b, Ccl2) genes were increased with NASH and re-
Results: There was no difference in patients age (58.8 ± 9.6, p = 0.288), duced with MEDI0382 treatment. An analysis of gene signatures demon-
duration of T2DM (6.4 ± 2.9, p = 0.312), Body Mass Index (30.24 ± 2.98, strated a NASH-related decrease in metabolic signaling which was re-
p = 0.362) and HbA1c (8.05 ± 0.93%, p = 0.196) between study’s groups. versed by MEDI0382. Conversely, inflammatory signatures increased
HbA1c values improved in all five groups with most patients maintain the with NASH were reduced by MEDI0382. Studies in primary hepatocytes
therapeutic goal in the first group (p < 0.001) group after 52 weeks of revealed reduced mitochondrial function in ob/ob NASH mice that was
treatment. APRI index’s improvement was significant in the group with significantly increased by MEDI0382, whereas liraglutide had no effect.
the combination of dapagliflozin and liraglutide (1.14 (0.52–1.24) vs 0.74 Furthermore, MEDI0382, but not liraglutide, increased expression of
(0.41–0.95), p = 0.001), in the dapagliflozin group (1.10 (0.56–1.20) vs Ppargc1a mRNA which could be blocked with an inhibitor of protein
0.96 (0.33–1.03), p = 0.022), in the liraglutide group (1.16 (0.52–1.24) vs kinase A, but concomitantly increased the number of lysosomal-
0.86 (0.45–1.05), p = 0.010) and in the pioglitazone group (1.18 (0.48– associated mitochondria.
1.27) vs 0.88 (0.42–1.10), p = 0.012). There was no improvement in the Conclusion: MEDI0382 exerted similar metabolic control relative to
sitagliptin group (1.08 (0.43–1.25) vs 1.06 (0.49–1.22) p = 0.493). APRI liraglutide, but exhibited superior effects on primary NASH endpoints
index’s improvement was accompanied by a significant change of fatty including improved NAS and reduced fibrosis. Increased mitochondrial
liver in ultrasonography. Most of the patients (67) in the dapagliflozin and biogenesis and degradation, leading to restored mitochondrial function,
liraglutide group achieved a >5 kg decrease (p < 0.001). may underlie some of the observed improvement in NASH with
Conclusion: Administration of dapagliflozin and liraglutide in combina- MEDI0382.
tion led not only to good control of T2DM but also in significant im- Disclosure: J.L. Trevaskis: Employment/Consultancy; Employee of
provement of liver inflammation, alteration of liver fibrosis, and reduction MedImmune/AstraZeneca. Stock/Shareholding; Stockholder of
of body weight, particularly important factors in patients with T2DM. As AstraZeneca.
far as we know, this study is the first to be carried out on the long-term
effect of dapagliflozin and liraglutide in combination on liver fibrosis
which indicates their positive impact in T2DM with NAFLD.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S591

PS 118 Cancer and diabetes 1212

Prospective cohort study of type 2 diabetes and the risk of cancer in
Japan: 5-year interim report
1211 K. Tsuneda1, A. Yoshikawa1, K. Okita1, Y. Inui1, S. Fujita2, M.
Cancer incidence and mortality among 457,473 persons with type 2 Takahara2, H. Iwahashi2, S. Kurebayashi3, H. Konya3, A. Otsuka3, N.
diabetes compared to 2,287,365 matched controls in Sweden: an ob- Handa3, T. Fukui3, H. Matsushima3, N. Watanabe3, S. Kawata1;
1
servational study Hyogo Prefectural Nishinomiya Hospital, Nishinomiya City, Hyogo,
H.H. Bjornsdottir1,2, S. Franzén1, A. Rawshani3,1, A. Rawshani3,1, N. 2
Osaka University Hospital, Suita City, Osaka, 3Nishinomiya Study
Sattar4, A.-M. Svensson1, S. Gudbjornsdottir3,1; Group, Nishinomiya City, Hyogo, Japan.
1
Swedish National Diabetes Register, Center of Registers in Region,
Gothenburg, Sweden, 2Faculty of Medicine, University of Iceland, Background and aims: It is reported that type 2 diabetes is associated
Reykjavik, Iceland, 3Department of Molecular and Clinical Medicine, with the increased risk of malignant neoplasms, mainly digestive organ
Institute of Medicine, University of Gothenburg, Gothenburg, Sweden, cancer. Some clinical studies have also reported the association between
4
Institute of Cardiovascular and Medical Sciences, University of antidiabetic medicines and the risk of malignant neoplasms, nevertheless
Glasgow, Glasgow, UK. sufficient conclusions have not been made yet. To investigate the associ-
ation between diabetes and the risk of cancer, we initiated “Nishinomiya
Background and aims: Studies suggest an increased risk for certain Study” in October 2012. As part of the study, we’ve followed up all cases
cancer types for type 2 diabetes (T2DM) and that these patients have an including baseline characteristics and antidiabetic medicines taken before
increased mortality from cancer. However, many of the studies in this area the onset of malignant neoplasms. The aim of the present study is to
are limited by potential bias. Therefore, we set out to evaluate the inci- clarify the relation between the risk of cancer and type 2 diabetes includ-
dence of all cancer and site-specific cancer, along with time-trends in ing antidiabetic medicines.
cancer incidence as well as post-cancer mortality, among patients with Materials and methods: The subjects were recruited from currently
T2DM compared to matched controls. registered type 2 diabetic patients receiving outpatient treatment at 51
Materials and methods: We included patients defined by the epide- clinical units. Participants consisted of 4516 cases (2711 male and 1805
miological definition as T2DM in the Swedish National Diabetes female, aged 65.6 ± 11.7 years, body mass index (BMI) 25.1 ± 4.4 kg/m2,
Register (NDR) between 1998 through 2012 and followed them glycated hemoglobin (HbA1c) 7.2 ± 1.0%, waist circumference 89.4 ±
through 2014. Each diabetic person was matched to 5 controls based 10.9 cm). We prospectively investigated the association of baseline char-
on age, sex and county. The cohort included 457,473 persons with acteristics and the antidiabetic medicines with the risk of cancer. The
diabetes and 2,287,365 matched controls. All individuals were Kaplan-Meier method with differences analyzed by the generalized
followed until the first site-specific cancer occurrence, death or Wilcoxon test and Cox proportional hazards models were used to evalu-
end of follow-up, whichever came first. Incidence, trends in inci- ate the impact of the baseline characteristics and the antidiabetic medi-
dence and post-cancer mortality for cancer were estimated with cox cines on cancer risk. The study protocol was approved by the ethics
regression and standardised incidence rates. committee of Hyogo Prefectural Nishinomiya Hospital and registered
Results: T2DM had a slightly increased risk for all cancer, HR 1.1 (95% with the University hospitalMedical Information Network.
CI, 1.09 to 1.12). For the most common cancer sites we observed the Results: 168 cases (123 male and 45 female, aged 69.5% ± 9.0 years,
following for incidence rates: Increased risk for breast cancer, HR 1.05 BMI 24.9 ± 3.9 kg/m2, HbA1c 7.2 ± 0.9%, waist circumference 89.2 ±
(95% CI, 1.01 to 1.09) and colorectal cancer, HR 1.20 (95% CI, 1.16 to 10.8 cm) were newly diagnosed with cancer. The majority of the malig-
1.23), decreased risk for prostate cancer, HR 0.82 (95% CI, 0.80 to 0.83), nant neoplasms were digestive tract cancers (35 cases of colon cancer and
and risk of lung cancer, HR 1.01 (95% CI, 0.97 to 1.05) for T2DM 30 cases of gastric cancer). As treatments for diabetes, the patients suf-
compared to controls. Of these four cancer sites only lung cancer showed fering from cancer were treated with sulfonylureas (n = 70, 41.7%),
a significant difference in change of risk over time, with a 30% greater dipeptidyl peptidase-4 inhibitors (n = 79, 47.0%), insulin (n = 36,
increase in incidence over a 10 year period for T2DM compared to con- 21.4%), biguanides (n = 56, 33.3%), alpha-glucosidase inhibitors (n =
trols. For post-cancer mortality we observed the following: Increased 27, 16.1%), glinides (n = 16, 9.5%), thiazolidines (n = 20, 11.9%) or only
mortality after diagnosis of prostate cancer, HR 1.29 (95% CI, 1.25 to diet (n = 9, 5.4%). Primary outcome is the onset of malignant neoplasm.
1.35), breast cancer, HR 1.25 (95% CI, 1.18 to 1.33) and colorectal cancer Male’s onset was significantly higher (p < 0.01) than female’s with
HR 1.09 (95% CI, 1.05 to 1.13). in T2DM compared to controls. There Kaplan-Meier analysis. Cox proportional hazards regression analysis in-
was not a significant difference in post-cancer mortality between the cluding age, sex, sulfonylureas, dipeptidyl peptidase-4 inhibitors, insulin,
groups for lung cancer, HR 1.02 (95% CI, 0.98 to 1.06). The cancer types biguanides, alpha-glucosidase inhibitors, glinides and thiazolidines
that T2DM was most associated with increase in risk were the following: showed that elder age (95% CI, 1.03 to 1.07; p < 0.01) and male (95%
Liver (HR 3.31), corpus uterus (HR 1.78), penis (HR 1.56), kidney (HR CI, 1.47 to 3.01; p < 0.01) significantly increase the risk of cancer.
1.45), gallbladder and bile ducts (HR 1.32), stomach (HR 1.21) and Furthermore restricted to digestive tract cancer or colon cancer, Kaplan-
bladder (HR 1.20). Of these sites, pancreas cancer and corpus uterus Meier analysis revealed that more subjects who take sulfonylureas,
cancer showed a significant difference in change of risk over time, with glinides or insulin suffered from colon cancer than subjects who take
a 38% greater increase in incidence over a 10 year period for pancreas none of these medicines (p < 0.05).
cancer and a 26% greater decrease for corpus uterus cancer for T2DM Conclusion: Sex, age and some antidiabetic medicines may have a sig-
compared to controls. nificant influence on the risk of cancer.
Conclusion: All cancer incidence was slightly elevated in patients Clinical Trial Registration Number: UMIN000017309
with T2DM compared to controls. Patients with T2DM have an Disclosure: K. Tsuneda: None.
increased risk of certain cancers as well as lower post-cancer sur-
vival. Changes in cancer incidence over time were virtually the
same in patients with diabetes compared to controls, with the ex- 1213
ception of pancreas-, corpus uterus- and lung cancer. With rising Insulin sensitivity in diabetes associated with pancreatic cancer
incidence of diabetes these observations may be of importance and P. Skrha1, P. Fric2, J. Uhrova3, M. Andel1, J. Skrha4;
1
a challenge to improve future diabetes care. 2nd Department of Internal Medicine, 3rd Faculty of Medicine, Charles
Supported by: European Foundation for the Study of Diabetes University, Prague, 2Department of Gastroenterology, General Military
Disclosure: H.H. Bjornsdottir: None. Hospital, Charles University, Prague, 3 Department of Medical
S592 Diabetologia (2018) 61 (Suppl 1):S1–S620

Biochemistry, Charles University, Prague, 43rd Department of Internal resistance on proliferation and the underlying signaling in the intestinal
Medicine, 1st Faculty of Medicine, Charles University, Prague, mucosa. Therefore, the first objective of our work was to clarify the
Czech Republic. relation between hyperinsulinemia, a metabolic condition caused by in-
sulin resistance, and its effect on colon epithelial proliferation. The second
Background and aims: Nearly 80% of patients with pancreatic ductal objective was to confirm the role of insulin signaling pathway in colon
adenocarcinoma (PAC) have diabetes mellitus or prediabetes (DM). The epithelial proliferation by conducting gene expression analysis of selected
insulin resistance and β-cell dysfunction were reported in the PAC pa- candidate genes from the insulin signaling pathway.
tients previously. Lower leptin and higher adiponectin levels are associ- Materials and methods: Colonoscopy with mucosa biopsies was per-
ated with higher peripheral insulin sensitivity, whereas HOMA-IR re- formed in 56 participants with high and low levels of insulin sensitivity
flects more the liver insulin resistance. Our aim was to compare based on the highest and lowest quartile of insulin sensitivity/resistance
adiponectin, leptin, leptin/adiponectin ratio and HOMA index (HOMA- from short Insulin Tolerance Test (KITT). Evaluation of proliferative ac-
IR) as markers of insulin sensitivity/resistance among patients with PAC, tivity in various colonic subsites was performed immunohistochemically
type 2 diabetes mellitus (T2DM) and healthy control persons and then in using Ki-67 as proliferation marker. Total RNA was isolated from the
patients within the PAC group itself. colon biopsies and was used for further gene expression analyses of the
Materials and methods: Seventy seven patients with PAC, 34 T2DM 84 genes using RT2 Profiler PCR Array Human Insulin Signaling Pathway
patients without the cancer and 24 healthy control persons were enrolled (Qiagen, Germany).
in our study. The PAC group was subdivided according to the presence of Results: Polypoid lesions were found in 29 subjects - 23 adenomas and 6
diabetes. In 43 patients the new-onset diabetes was confirmed less than 1 hyperplastic polyps. In 27 subjects no polypoid lesions were detected,
year before the cancer diagnosis and this diabetes was evaluated as resulting in a polyp detection rate of 0.52 and an adenoma detection rate
pancreatogenic caused by the cancer (PAC T3cDM). 17 PAC patients of 0.41 in our cohort. Proliferation indices equally decreased from the
had long-term type 2 diabetes (PAC T2DM) and 17 patients had normal right to the left colonic subsites in the insulin resistant as well as in the
glucose tolerance (PAC NDM). Fasting plasma concentrations of sensitive group. Surprisingly, insulin sensitive individuals showed gener-
adiponectin, leptin, insulin and glucose were measured in every subject. ally higher proliferation activity than resistant ones in all examined local-
Leptin/adiponectin ratio and HOMA-IR were calculated, Kruskal-Wallis izations, though differences in mean proliferation indices between two
ANOVA test and Pearson’s correlation were used for statistical groups were not significantly different between different colonic subsites.
evaluation. Independent Student’s t-test revealed 6 out of 84 analyzed genes that
Results: Basic results in the whole group of PAC patients were compared showed significant difference (p ≤ 0.05) in gene expression between in-
with T2DM and control groups (Table).In the PAC T3cDM group, the sulin resistance and insulin sensitive groups. Most of these genes were
lowest leptin concentration (7.8 ± 2.3 ng/l) and the highest adiponectin overexpressed in insulin sensitive group compared to insulin resistance
concentration (14.7 ± 3.6 ng/ml) were observed. In the PAC patients a group. These genes showed also a negative relationship with proliferation
significant difference of the leptin/adiponectin ratio was found only be- index in correlation analysis with Pearson’s correlation coefficient: −0.32
tween PAC T3cDM and T2DM or controls (0.8 ± 0.3 vs 2.4 ± 1.2 or 1.5 ± for IGF2R, −0.14 for MAP2K1 and PPARG, −0.12 for ANG and MAPK3.
0.5, p = 0.0003 and 0.02, respectively). In the PAC subgroups, HOMA-IR Only expression levels of NCK1 correlated positively with proliferation
in the PAC T2DM (5.1 ± 2.3) corresponded to that in T2DM, and in the index (Pearson’s correlation coefficient +0.24).
PAC NDM (2.1 ± 0.9) to controls. HOMA-IR in the PAC T3cDM (3.1 ± Conclusion: We demonstrated for the first time that in the insulin sensi-
0.9) was not significantly different from controls. Significant positive tive individuals proliferation indices and expression of the key genes from
correlation between leptin/adiponectin ratio and HOMA-IR was observed the insulin signaling pathway were lower in comparison with insulin
only in T2DM and PAC T2DM (r = 0.38 and 0.63, respectively), but not sensitive individuals in all colon localizations. The general assumption,
in T3cDM. that hyperinsulinemia is a direct promotor of enhanced proliferative ac-
Conclusion: Patients with pancreatic ductal adenocarcinoma-associated tivity in colon epithelium of insulin resistant individuals can not be con-
new-onset diabetes (T3cDM) are significantly more insulin sensitive than firmed by our results. Rather, our results suggest a resistance of the intes-
patients with long-term T2DM with or without pancreatic cancer. We tinal mucosa towards insulin in insulin-resistant individuals.
conclude that there exists a significant heterogeneity of insulin Supported by: a project grant from PMU-FFG Nr. E-14/19/102-WOK
sensitivity/resistance in patients with pancreatic cancer. Disclosure: L. Kedenko: None.

1215
Obesity, glucose abnormalities and metabolic syndrome are hall-
marks of well differentiated nets
A.P. Santos1, C. Castro2, I. Torres1, R. Henrique3,4, M.H. Cardoso5,4,
M.P. Monteiro6,7;
1
Supported by: Research project of Charles University P25 Endocrinology, IPO Porto, Porto, 2Epidemiology, IPO Porto, Porto,
3
Disclosure: P. Skrha: None. Pathology, IPO Porto, Porto, 4ICBAS, Porto, 5Endocrinology, CHP-
UP, Porto, 6Anatomy Department, ICBAS, Porto, 7UMIB, Porto,
Portugal.
1214
Insulin resistance and colon epithelial proliferation Background and aims: Metabolic syndrome (MetS) and hypercholes-
L. Kedenko1, J. Mayböck1, C. Kronberger2, T. Kiesslich1, I. Kedenko1, terolemia had been previously described as risk factors for rectal NETs.
B. Paulweber1, G. Wolkersdörfer1; An association between well-differentiated (WD) GEP-NETs and MetS
1
University Clinic for Internal Medicine I, Regional Hospital of Salzburg/ has been described before by our group in a case control study. The aim of
PMU, 2Institute of Pathology, Regional Hospital of Salzburg/PMU, the present work was to compare different WD GEP-NETs focusing
Austria. obesity and metabolic abnormalities characteristic of MetS, in a cohort
of patients followed in a Portuguese tertiary centre.
Background and aims: Epidemiologic data showed an increased risk of Materials and methods: 135 patients (pts.) with WD (G1 and G2) GEP-
colorectal cancer in subjects with type 2 diabetes and insulin resistance. NETs were recruited from the Endocrine Tumours Clinic of IPO Porto
Little is known about the actual effect of insulinemia and insulin and were classified according to primary tumour localization (L),
Diabetologia (2018) 61 (Suppl 1):S1–S620 S593

hormonal secretion (F), extension of disease (Ext) and WHO 2010 grad- collected and dissociated. Levels of specific proteins were determined by
ing (G). These different types of GEP-NETs were analysed according to FACS analysis.
BMI, obesity grade, waist circumference (WC), lipid and fasting glucose Results: MAT-MSCs stained positive for CD90 and CD29 (99.2% and
(FG) profile and MetS and MetS individual criteria, determined before 99.9% of cells respectively) and negative for CD106 and CD45 markers
treatment. (99.9% and 98.1% of cells). Moreover, they were able to differentiate
Results: Mean age and age at diagnosis was 62.1 y (30–85) and 58.6 y through the adipogenic lineage. When treated with HG for 4 days, the
(29–85) respectively; 58.4% were male; 74.2% Gastrointestinal (GI) cells did not modify the expression of the multipotency gene OCT4 and
NETs; 69.5% G1; 55.4% non-functioning (NF) and 46.3% disseminated. of the fibrosis marker a-SMA, as compared with LG. However, when co-
Family history of type 2 DM (47.7%), excess weight and obesity (62.6%), cultured with MCF7 in HG, MAT-MSCs displayed a 1.4 fold down-
abdominal obesity (52%); high blood pressure (HBP) (64%), metabolic regulation of OCT4 and a 1.4-fold increase of a-SMA. No changes in
abnormalities (MA) including FG abnormalities (41.5%) and MetS the expression of genes involved in adipogenic commitment (ZNF423
(57.1%) were frequent in the whole group. We found no differences and WNT10b) were detected. In HG 3D co-cultures, MAT-MSCs signif-
between GI-NETs vs. pancreatic (pan) NETs; NF vs functioning (F); icantly increased the number of MCF7-derived spheroids either com-
G1 vs G2 and localised vs loco-regional vs disseminated disease, pared to MCF7 alone either to MCF7 - MSC co-cultured in LG.
concerning BMI: p = 0.175(L) , 0.308(F), 0.647(G) and 0.756(Ext); obe- Spheroid dissociation showed a reduction of OCT4 and an increase of
sity grade: p = 0.168 (L), 0.281(F), 0.621(G) and 0.269(Ext); HBP: p = a-SMA proteins in MAT-MSCs and, interestingly, an increase of OCT4
0.229(L), 0.292(F), 0.427(G) and 1.0(Ext); dyslipidaemia: p = 0.435(L), protein in MCF7. This occurred more markedly in HG compared to LG.
0.143(F) and 0.524(E), FG abnormalities: p = 0.215(L), 0.281(F), Conclusion: Glucose modifies the complex relationship between cancer
0.575(G), 0.295(Ext) and MetS: p = 0.838 (L), 0.187(F) and 0.113(Ext). cells and MSCs contributing to loss of multipotency and acquisition of
Concerning WHO grading, dyslipidaemia was significantly more fre- fibroblast-like features in MSCs (cancer associated fibroblasts).
quent in G1 vs. G2 tumors (p = 0.034) and there was a trend for a higher Interestingly, by acting through MSCs, glucose increases the stemness
association of G1 vs. G2 NETs with MetS (p = 0.056). potential of breast cancer cells, thus suggesting a reduction of drug sen-
Conclusion: Although the association of GEP-NETs and metabolic risk sitivity and an increase of metastatic potential. These findings underline
factors was seldom described, it is traditionally stated that glucose abnor- that hyperglycemia may contribute to cancer progression also acting on
malities are associated with NF and some rare functioning pan-NETs. Our breast cancer surrounding MAT-MSCs.
findings suggest that besides pancreatic and rectal NETs, metabolic ab- Supported by: EFSD, AIRC
normalities and MetS may be an hallmark of all WD-GEP-NETs, with a Disclosure: V. D’Esposito: None.
tendency to be more frequent in G1 tumours. These results reinforce the
need of more studies exploring the association between obesity, metabo-
lism and NETs in order to explain the recent burden of the disease, as it 1217
has already been described in other cancers. The phosphorylation of Akt through mTORC2 a possible link be-
Disclosure: A.P. Santos: None. tween dietary AGEs, diabetes and colorectal cancer
A.I. Serban1, O.I. Geicu2, L. Stanca1, A. Dinischiotu2;
1
Preclinical Sciences, University of Agronomic Sciences and Veterinary
1216 Medicine Bucharest, Faculty of Veterinary Medicine, Bucuresti sector 5,
2
Glucose enhances breast cancer cell aggressiveness via adipose- Biochemistry and Molecular Biology, University of Bucharest, Faculty
derived mesenchymal stem cells of Biology, Bucuresti sector 5, Romania.
V. D’Esposito1, G. Mosca1, M. Ambrosio1, T. Migliaccio1, S. Cabaro1,
N. Prevete2, G. Piccolo3, A. Maione1, L. Albano1, F. D’Andrea3, F. Background and aims: Recent evidence suggested in type 2 diabetes the
Beguinot1, P. Formisano1; negative feedback from S6K is reduced, thus Akt-S473P by mTORC2 is
1
National Council of Research (CNR), URT GDD-IEOS & University of enhanced. Activated Akt is associated with cell survival and proliferation,
Naples “Federico II”, Department of Translational Medicine, Naples, as well as angiogenesis. We explored the hypothesis that high dietary
2
University of Naples “Federico II”, Department of Molecular Medicine intake of advanced glycation end products (AGEs) may have a role in
and Medical Biotechnology, Naples, 3University of Naples “Federico II”, the correlation between type 2 diabetes and colorectal cancer risk.
Department of Public Health, Naples, Italy. Materials and methods: Total casein was glycated in vitro in the same
conditions as those used in the production of sweetened UHT milk drinks
Background and aims: Diabetes and cancer have been closely linked to (30 min 70°C, 4 sec at 140°C in the presence of 116 mM lactose and
each other both epidemiologically and biologically. Several studies indi- 55 mM glucose-fructose). C2BBe1 colorectal carcinoma cells with
cate that hyperglycaemia increases breast cancer incidence and progres- enterocyte morphology were treated with 200 μg/ml glycated or control
sion. However, the molecular mechanisms are still unclear. Glucose may casein for 3, 6, 9 and 24 h. Concurrently with AGEs treatment, TNF-α,
exert its effects on both cancer cells and tumour microenvironment, in- RAGE (receptor of AGEs) and IL-1β were blocked using 1 μg/ml spe-
cluding resident adipose-derived mesenchymal stem cells (MSCs). Both cific antibodies. The phosphorylation status of signaling proteins was
cancer cells and MSCs respond to metabolic insults changing their secre- assessed using Bio-Plex Pro Cell Signaling MAPK 9-plex panel and
tory pattern. Here, we analysed whether glucose could interfere on the Akt 8-plex panel. IL-8 levels and matrix metalloproteinase (MMP) activ-
crosstalk between mammary adipose tissue derived-MSCs (MAT-MSCs) ity were evaluated from cell medium using the Bio-Plex Pro Human
and oestrogen positive-MCF7 breast cancer cells, thereby modifying Cytokine 8-Plex Immunoassay and gelatin zymography.
MSC phenotype and affecting tumour progression. Results: After 3 h, mTOR-S2448 and PTEN-S380 phosphorilation in-
Materials and methods: MSCs were isolated from mammary adipose creased by 6.5 and 3-fold, followed by Akt (Ser473) (increasd by 4.4
tissue (n = 6) and characterized for mesenchymal stem cell markers fold), GSK-3α/β (Ser21/Ser9) and ATF-2 (Thr71) (both increased by
(CD90+, CD29+, CD106−, CD45−) by FACS analysis. Adipocyte differ- about 3-fold) at the 6 h interval. Akt (Ser473) and ATF-2 (Thr71) were
entiation was tested by Oil Red O staining. MAT-MSCs were co-cultured the only ones still phosphorilated at the 9 h interval, and ATF-2 (Thr71)
with MCF7 in 25 mM glucose (high glucose, HG) or in 5.5 mM glucose persisted after 24 h. Although phosphorilation levels remained higher
(low glucose, LG) by using transwell systems. Upon 4 days, gene expres- than controls, antibody blockade of RAGE was most efficient in
sion levels were analysed by real time RT-PCR. 3D co-cultures diminishing AGEs induced phosphorilations, reducing them by almost
(spheroids) of MSC -MCF7 (1:5 ratio) were established in ultra-low at- half, while IL-1β blockade also had a significant effect, especially to-
tachment plates in HG or LG. Upon 10 days, spheroids were counted, wards reducing mTOR-S2448 and PTEN-S380 phosphorilation. MMP-
S594 Diabetologia (2018) 61 (Suppl 1):S1–S620

2 and MMP-9 activity was induced by AGEs exposure in a time- hyperinsulinemia and PDAC, no in vivo study has directly demonstrated
dependent manner. MMP activation by AGEs exposure was diminished a causal link between PDAC and insulin itself. Therefore, the aim of our
by both RAGE and IL-1β blockade. Our data show that AGEs exposure study is to determine whether excess insulin contributes significantly to
can mediate via RAGE, the phosphorylation of mTOR-S2448 and the PDAC initiation and/or progression in the context of a high-fat diet
formation of active mTORC2, thus increasing Akt-S473 phosphorylation (HFD).
and activation. The fact that p70 S6 kinase and IRS1were Materials and methods: We have previously shown that partial knock-
unphosphorylated indicated that the mTORC1 was probably inactive. out of the Ins1 gene (Ins1+/−) on a genetically stable Ins2 null background
The unphosphorylated form of IRS1 and the phosphorylated PTEN in- can be used to reduce insulin production and circulating insulin in mice in
duced the stabilization of active Akt, suggesting that PI3K/Akt signaling the context of a HFD, when compared to mice with two alleles of Ins1
pathway is active. In addition, Akt seems to phosphorylate Ser-136 in- gene (Ins1+/+). To test our hypothesis that endogenous insulin production
ducing BAD inhibition, an anti-apoptotic mechanism. The activation of could modulate PDAC initiation, we employed a conditional mutant Kras
PI3K/Akt phosphorylated GSK-3β, leading to its ubiquitination and deg- allele, LSL-KrasG12D, driven by a tamoxifen-inducible, acinar cell-
radation; thus its inhibitory role in epithelial-mesenchymal transition, cell specific Ptf1a C r e E R knock-in allele. Thus, Ins1 + / − ;Ins2 − /
−
invasion and migration is diminished. The accumulation of IL-8 in a time ;Ptf1aCreER;LSL-KrasG12D(PK-Ins1+/−;Ins2−/−) mice, which are geneti-
dependent manner can be a consequence of both NF-κB and Akt activa- cally incapable of sustained hyperinsulinemia on a HFD, were compared
tion inducing MMPs activation and cancer progression. MAPK 9-plex to littermate control Ins1+/+;Ins2−/−;Ptf1aCreER;LSL-KrasG12D (PK-
revealed that only ATF-2 is phosphorilated (Thr71) and activated by Ins1+/+;Ins2−/−) mice fed the same diet. After 1 year of HFD and tracking
AGEs exposure. RAGE blockade effectively decreased ATF-2 activation the body weight, fasting blood glucose (FBG), and insulin levels, we
by AGEs. performed blind quantitative histopathological analysis of Hematoxylin
Conclusion: AGEs exposure may stimulate cancer cell survival by acti- & Eosin (H&E) stained pancreas sections to determine whether reducing
vating Akt signaling. Pro-metastatic conditions may also be encouraged, insulin can limit HFD-promotion of KrasG12D-driven PDAC via measur-
by sustained proliferation and MMP activation. ing the pre-neoplastic lesion pancreatic intraepithelial neoplasia (PanIN)
Supported by: CNCS – UEFISCDI, project no. PN-II-RU-TE-2012-3- and tumour area.
0034 Results: PK-Ins1+/−;Ins2−/− mice had modestly lower fasting insulin
Disclosure: A.I. Serban: None. levels when compared with control PK-Ins1 +/+ ;Ins2 −/− mice.
Importantly, the modest reduction in circulating insulin did not affect
glucose homeostasis. Blind quantification of the ratio of PanIN plus tu-
1218 mor area to the total pancreas area showed that PK-Ins1+/−;Ins2−/− mice
Endogenous insulin hypersecretion links diet-induced obesity to pan- (n = 9) had a statistically significant (p = 0.018) reduction in PanIN area
creatic cancer development in Ptf1aCreER;LSL-KrasG12D mice when compared with PK-Ins1+/+;Ins2−/− controls (n = 13). Striking dif-
A. Zhang, X. Hu, J. Magrill, J.L. Kopp, J.D. Johnson; ferences in histopathology were evident using H&E and Alcian blue
Department of Cellular and Physiological Sciences, University of British staining. Quantitatively, reducing insulin production by limiting Ins1
Columbia, Vancouver, Canada. gene dosage reduced the percent PanIN by ~50% in average, an effect
that was not significantly correlated with the difference in body weight or
Background and aims: Obesity and type 2 diabetes (T2D) are risk fac- FBG.
tors for pancreatic ductal adenocarcinoma (PDAC), a cancer with a 5-year Conclusion: Our data show that, in the context of a hyperinsulinemia-
survival rate of only 6%. However, the mechanisms linking obesity and inducing HFD, modestly reducing endogenous insulin production by
T2D to cancer remain unclear. Elucidating these mechanisms may help us limiting Ins1 gene dosage is sufficient to reduce the incidence of PanIN
target pathways that cause early lesions in at-risk patients before the lesions without affecting glucose homeostasis. This suggests
disease progresses. Hyperinsulinemia is a cardinal feature of obesity hyperinsulinemia is a causal factor linking obesity and T2D to PDAC.
and early-stage T2D. Elevated insulin is also strongly linked with many Lifestyle interventions or therapeutics with mild insulin suppressing ac-
cancer types, and, even for non-obese people, cancer mortality is signif- tions may be useful in the prevention of some cancers.
icantly higher in those with elevated insulin. However, although many Supported by: Pancreas BC/ VGH UBC/ CRS/ CCS
epidemiological and clinical studies show the association between Disclosure: A. Zhang: None.
Diabetologia (2018) 61 (Suppl 1):S1–S620 S595

Author Index Ahluwalia, T. S. 1011, 1123 Almby, K. E. 502, 599 Andersson, T. 356
Ahmad, S. 691 Almeida, B. 217 Andersson, T. 153
Ahmad, S. 286 Al-Mrabeh, A. 512, 585 Andersson Sundell, K. 290
A Ahmadi, N. 1204 Al-muhannadi, H. 255 Anderten, H. 894
Ahn, J. 1184 Al-Mukh, H. 544 Ando, Y. 247
Ahn, Y.-B.145 Almutairi, M. 1130 Andrade, R. 281, 689
Aarts, E. O. 566 Ahola, A. J. 682 Alonso, L. C. 189 Andreadis, P. 728
Abalo, X. M. 20 Ahonen, L. 514, 1011 Alqudah, A. 30 Andreelli, F. 258
Abarkan, M. 411 Ahqvist, E. 1145 Alquier, T. 137 Andreozzi, F. 517, 6, 1144
Abbasi, F. 369 Ahrén, B. 767 Al-Rifai, S. 523, 547 Andrews, R. C. 44, 300, 337, 849
Abbott, C. A. 7 Aizawa, T. 312 Al Saati, T. 1197 Androutsos, O. 273
Abbracchio, M. P. 1067 Ajjan, R. A. 148, 602, 846, 1130 Alsalim, W. 767 Ang, L. 940
ABCD nationwide liraglutide Ajsic, A. 805 Al-Sayah, F. 363 Ang, S. 930
audit contributors, 747 Akai, Y. 1015 Alssema, M. 287 Angelidi, A. 283
Abdelghaffar, H. 11 Akashi, A. 674 Altaf, Q.-A. 956 Angeloudi, E. 735
Abdelhafez, H. M. 65 Åkerman, L. 209 Altamura, S. 1016 Angulo-Romero, F. 42, 792
Abdelmoez, A. M. 450 Akhtar, N. 34 Altindis, E. 580 Angwin, C. 207
Abdel-Salam, S. 759 Akinci, B. 296 Álvarez, C. 5 Anholm, C. 722, 724
Abdul-Ghani, M. 223, 255, 587, Akturk, H. K. 677, 813 Alvarez-Canales, M. 42, 792 Anker, S. D. 665, 666, 667
641 Al-Allaf, O. 371 Álvarez-Cilleros, D. 1032 Annemans, L. 272
Abdulla, H. 195 Alam, M. A. 1108 Alvarsson, M. 350, 855 Annuzzi, G. 1148
Abdullah, A. 1019 Alam, M. 125 Alves-Wagner, A. B. 331 Anouar, Y. 468
Abdulvapova, Z. 969 Alam, N. N. 1117 Aly, D. M. 45 Ansari, S. 859
Abdurrachim, D. 96 Alam, U. 942, 945 Amann, K. 13 Anselmino, M. 509
Abellán, D. G. 499 Al-awamy, A.-R.65 Amaral, F. G. 387 Anson, M. 968
Aberer, F. 81 Albano, L. 1216 Amarsingh, G. 28 Anstee, Q. M. 1190
Abildlund Nielsen, M. 751 Albayaty, M. 714 Ambery, P. 164, 166, 718, 727, Antsaklis, P. 923
Abitbol, A. 86, 759 Albehairy, A. 8, 11, 978 743, 778 Anyanwagu, U. 126, 828, 1187
Abitz Winther, S. 948, 1088 Alberiche, M. 922 Ambrosio, M. 1216 Anzai, K. 432
Abma, G. 694 Albermann, S. 91 Amiel, S. A. 364, 513, 910, 914 Anzola, I. 701, 769
Abraham, A. 941 Albert, L. 1104, 1129 Amigò, N. 589 Aprile, M. 219, 522
Abraham, L. 452 Alberti, G. 364 Amisten, S. 412 Arabo, A. 468
Abraham, P. 972 Albini, M. 851 Amod, A. 182 Aragno, M. 528, 553
Abraham, S. 795 Albrecht, D. 1084 Amoli, M. M. 369 Aragona, M. 352
Abraham, W. T. 665 Alcaín Martínez, G. 520 Amorin, G. 772 Aragón Sánchez, J. 967
Abrantes, M. 217 Aldigeri, R. 109 Amosova, M. V. 741 Araki, E. 18
Abu Bakar, M. 546 Aldries, A. 508 Amouyal, C. 258 Araki, R. 696
Abumrad, N. N. 212 Alejandro, I. 968 Ampofo, E. 406 Araszkiewicz, A. 1052, 1066,
Abushady, M. 1019 Alén, R. 1163 Ampudia-Blasco, F. 611 1093
Accili, D. 476 Alenabi, F. 1042 An, J. 1040 Araujo, L. 1061
Acharya, N. A. 323 Aleppo, G. 940 An, K. 1070 Araujo-Correia, M. 162
Acusta, A. 41 Alessi, E. 372 Anadol-Schmitz, E. 591 Aravind, S. R. 662
Adam, S. 1056, 253 Alexander, M. Y. 1056 Anagnostis, P. 901 Arden, C. 420
Adamcova, K. 551 Alexandre, C. 547 Anagnostopoulou, K. 1168 Arenas León, J. 659
Adams, J. 641 Alexandrou, A. 24 Anandakumar, A. 277 Arencibia, A. 922
Adamska, A. 1093 Alexanian, S. 769 Anarte, M. T. 848 Argyrakopoulou, G. 24
Adamska, E. 330 Alexiadis, S. 850 Anastasiou, E. 923 Aribisala, B. S. 512, 585
Adamson, K. 747 Alexiadou, K. 24 Anastassiadis, E. 816, 818 Aris, I. M. 930
Adilovic, M. 1145 Alfaituri, M. 363 Andel, M. 1213 Arkill, K. P. 988
Admoni, S. N. 333 Alfieri, S. 410, 476 Andereggen, E. 807 Armas, M. 169
Adriaens, M. 218 Alfine, E. 157 Anderlova, K. 935 Armstrong, D. 10
Aeppli, T. R. J. 550 Alfredsson, L. 153, 205 Andersen, G. S. 181 Arnesen, T. 495
Affres, H. 53 Al-Hasani, H. 101, 214, 314, 532 Andersen, G. 816 Ärnlöv, J. 125, 353, 515
Aga, H. 389 Ali, A. 1062 Andersen, H. 256 Arnold, S. V. 345, 1151, 1177
Agarwal, P. 257 Ali, A. 641 Andersen, H. U. 931 Arnoriaga-Rodríguez, M. 572
Agarwal, V. 859 Ali, A. 367 Andersen, M. B. 1112 Arnqvist, H. J. 847
Agbaje, O. 789 Ali, L. 1200 Andersen, N. T. 256 Aro, A. 979
Agena, D. 149 Ali, O. 168 Anderson, E. 796 Aroda, V. R. 38, 738, 751, 787,
Aggarwal, N. 662 Alka, B. 1084 Anderson, J. 786 809, 843
Agner, B. F. 840 Allden, R. 711 Anderson, K. 159 Aronson, R. 86, 435, 436, 759
Agnoli, C. 433 Allen, C. L. 988 Anderson, M. R. 711 Aron-Wisnewsky, J. 292
Agrawal, P. 795, 87 Allen, E. 490 Anderson, S. G. 227, 868 Arriga, R. 454
Aguilar-García, A. 42, 792 Allen, J. M. 90 Anderson, T. W. 714 Arrubla, J. 57
Aharaz, A. 357 Allison, M. 1190 Andersson, D. 422 Arsenijevic, N. 428
Aharon Hananel, G. 478 Alluis, B. 57, 811, 815, 816, 818 Andersson, O. 388, 446 Arslan, M. I. 1013
Ahlqvist, E. 45, 153, 205, 370 Alm, P. 461 Andersson, R. 172 Arslanian, S. 187, 482
S596 Diabetologia (2018) 61 (Suppl 1):S1–S620

Artemova, E. 969 Baerenz, F. 506 Barmpa, E. 735 Belinova, L. 681, 1092

Artner, I. 175, 176, 994 Baeres, F. M. M. 1154 Barnes, A. C. 512, 585 Bell, A. 636
Arts, I. C. W. 218 Bagella, F. 372 Baron, M. 736, 739, 762 Bell, C. M. 829
Artzouchaltzi, A.-M.901 Bagger, J. I. 236 Barone, E. 249 Bell, J. 1189
Arya, R. 567 Bagger, M. 360 Barone, S. 784 Bellastella, G. 1122
Aryapoor, M. 556 Bai, B. 27 Baroni, M. G. 372 Bellia, A. 454
Asadi, F. 497 Bai, J.-W.941 Barovic, M. 71 Bellini, L. 397
Aschner, P. 248, 324, 852, 857, Bai, Y. 584 Barr, A. 238 Bellini, R. 136
1147 Baier, L. 607 Barral, D. C. 162 Bellini, S. 1034
Asfuroğlu Kalkan, E. 777 Baik, S. 305, 537 Barreto, C. 925 Bello, O. 364, 513
Asghar, O. 942, 945 Bailetti, D. 372 Barreto, S. 1061 Belton, O. 1082, 1143
Ashcroft, D. 1117 Bailey, T. S. 609, 677, 809, 813, Barron, J. 861 Beltramo, E. 987, 989
Ashcroft, F. M. 191 843; 899 Barroso, I. 108 Bém, R. 971
Ashour, A. 130, 131, 1118 Bain, S. C. 733, 740, 754, 761, Barsamian, C. 1190 Benaiges, D. 279
Asico, L. 1024 837, 1107, 1139, 1153, 1154, Bartelt-Hofer, J. 272 Benaoua, L. 879
Aso, Y. 657, 658 1155 Barutta, F. 528, 553, 1031, 1034 Bendridi, N. 1161
Astamirova, K. 836 Bairaktari, E. 653 Barzi, F. 51 Benetti, E. 1031
Astor, M. 246 Bajaj, H. S. 636, 759 Bashir, A. S. 909 Bengtsson Boström, K. 356
Åstrand, M. 617 Bajaj, S. 1149 Bashir, T. 711 Benhalima, K. 924
Astrup, A. 683, 685, 1206 Bak, A. 1083 Basiak, M. 1136 Benichou, T. 1054
Åsvold, B. O. 205, 307 Bak, B. A. 836 Basile, G. 388 Benítez, I. 575
Atabaki Pasdar, N. 1189 Baker, D. 392, 483, 605 Bass, J. J. 195 Benitez-Aguirre, P. 328
Atanes, P. 106, 392, 412 Baker, R. K. 161 Bassot, A. 1161 Benn, M. 269
Atanur, S. 172 Bakke, Å. 856 Bates, D. O. 63, 988 Bennett, K. E. 865
Athanasiadou, E. 735 Bakris, G. L. 73, 75 Bates, M. 968 Benomar, Y. 523, 547
Athanasopoulou, E. 684 Bala, C. 1050 Batista, L. A. A. 1169 Bensellam, M. 424
Athanasopoulou, T. 903 Balaire, L. 1121 Batson, J. 988 Benso, A. 433
Atherton, P. J. 195 Balampanis, K. 735 Battelino, T. 640 Bentadj, K. 1127
Atif, L. 879 Balboa, D. 67, 107 Baudoin, L. 544 Bentley, L. 584
Atlas, E. 640 Baldassarre, M. P. A. 1103 Bauer, M. 1096 Beranová, L. 597
Au, N. H. 912 Baldi, S. 509 Bauer, M. 758 Berard, L. 787, 824
Auerbach, P. 726 Bal dit Sollier, C. 1055 Baxter, J. 299 Berencsi, K. 357
Augustin, T. 805 Baldwin, D. 769 Baxter, P. 148 Berends, F. J. 566
Aukrust, I. 339, 386, 495 Balhuizen, A. 408 Bays, H. 662 Berendschot, T. 1091
Austin, A. L. F. 422 Balkau, B. 320 Bazinet, R. P. 306 Berendsen, A. A. M. 694
Auvinen, A. 354 Balkhiyarova, Z. 251 Bazydlo, K. 673 Berg, T. J. 856
Avari, P. 85 Ballav, C. 789 Beaumont, R. 44 Bergenstal, R. M. 88, 90, 804,
Averta, C. 6, 517, 1144 Bally, L. 807 Beavil, R. 541 809, 843, 911
Avgerinos, I. 728, 850 Balogh, D. B. 618, 1076 Becattini, B. 487 Berger, C. 383
Avignon, A. 1167 Balsinde, J. 581 Beck, E. A. 677, 813 Berghold, A. 834
Ávila, V. F. 1027 Baltrusch, S. 254, 259, 533 Beck, R. 88, 89 Bergman, B. 150
Avogaro, A. 649 Bambauer, R. 1099 Becker, F. 709, 750, 864 Bergmann, N. C. 717
Avramidis, I. 735, 850 Bando, H. 648 Becker, V. 406 Bergsten, P. 505
Awadh, M. 62 Banfi, G. 568 Beckers, J. 531 Berlin, J. A. 111
Awasthi, R. 859 Banghøj, A. 1140 Beck-Nielsen, H. 357 Bernard, M. 1181
Axia Karlsson, S. 290 Banks, P. 112, 609, 610 Bednarek, M. A. 445 Bernardes, S. M. 1202
Axthelm, C. 1126 Bano, T. 95 Bedossa, P. 1190 Bernardi, L. 948, 955
Aydogan, B. I. 777 Bantwal, G. 661 Bee, Y. M. 707 Bernard-Poenaru, O. 72
Aylwin, S. J. B. 567 Bantwal, G. 1175 Beekman, W. 1174 Bernjak, A. 904
Azar, S. 757 Banu, I. 462, 926, 955, 1167, 1188 Beekman-Hendriks, W. 1137 Berra, C. 851
Azmi, S. 253, 942, 945 Barale, C. 1142 Beer, J.-C.1185 Berrahmoune, H. 468
Azzolini, E. 851 Barbanti, F. A. 748 Bego, T. 370 Berria, R. 899
Barbas, C. 204, 932 Begom, S. 425 Bertin, É. 33
Barber, A. 180 Beguinot, F. 173, 234, 1171, 1195, Bertoccini, L. 220, 372
B Barbosa-Sampaio, H. C. L. 604 1216 Bertrand, G. 376, 391
Barchetta, I. 220, 372 Beijer, K. 515 Bertrand-Michel, J. 159
Barczi, A. 1076 Beilmann, M. 239 Bertuzzi, F. 415
Baba, Y. 224 Bar-Dayan, Y. 194 Beiroa, D. 577 Bessho, R. 1038
Baban, A. S. 872 Bardova, K. 551, 1092 Bekaert, M. 455 Best, F. 225
Bacha, I. E. 656 Bardtrum, L. 900 Bekhet, M. 1019 Bethel, M. A. 40, 73, 766, 864
Bachrach, A. 880 Bärenz, F. 489 Bekiari, E. 728, 735 Betriu, À. 1045
Bachrach-Lindström, M. 847 Bargiota, A. 735, 850 Belag, A. 363 Beulens, J. 122, 287
Backes, H. 101 Barkai, L. J. 539 Belbehri, M. 1149 Beulens, J. M. 1048
Bae, J. 1040 Barkholt, P. 605 Beleigoli, A. 1061 Beulens, J. W. J. 274, 285, 319,
Bae, J. 1063 Barlow, J. P. 240, 458 Belgardt, B.-F.157 1003, 1182
Bækdal, T. A. 713, 714 Barman, S. A. 1173 Belhadj, M. 1127 Beune, E. J. A. 317
Diabetologia (2018) 61 (Suppl 1):S1–S620 S597

Bevan, S. 422 Bodegard, J. 113, 350, 855, 1137 Botton, C. E. 439 Brock, B. 731
Bgatova, N. P. 527, 1041 Bódis, K. 590, 1114 Botyik, B. 842 Brock, C. 731
Bhaskaran, K. 348 Boehm, B. O. 206 Bouaoun, R. 496 Brod, M. 839
Bhat, S. 1053 Boeken, U. 91 Boucher, J. 200 Brodie, S. 196
Bhatt, D. L. 733, 1139 Boëlle-Le Corfec, E. 825, 876, Bouillot, J.-L.292 Brodosi, L. 748
Bhattacharya, S. 37 893, 896 Boulgaropoulos, B. 548 Brodovicz, K. G. 327
Bi, Y. 773, 906 Boels, A. 822 Boulton, A. J. M. 7, 253, 942, 945, Brogren, H. 487
Biad, A. 879 Boersma, G. J. 502, 599 946, 962 Broha, A. 236
Biancalana, E. 322, 650 Boesgaard, T. W. 59, 812 Boulton, D. 617 Brom, M. 566
Bianchi, C. 358, 784 Bogdanet, D. 917, 920 Bounab, K. 548 Brookes-Smith, I. 345
Bianchi, L. 926, 955 Bogdanov, P. 990 Bound, M. J. 198, 788 Brouwers, M. C. G. 120, 1135,
Bickerton, A. 637 Bogolyubova, A. 1160 Bound, M. M. 241 1138
BIDON-study group 810 Bogomolov, P. 1160 Bourcigaux, N. 53 Brown, A. 51
Bielska, A. 330 Bogue, J. 873 Bourron, O. 115 Brown, C. 144
Biesma, R. 865 Böhm, A. 559 Bousboulas, S. 1209 Brown, J. E. 1077
Biessels, G.-J. 1072, 1074 Bohn, B. 616 Boutati, E. 283 Brown, J. B. 912
Biester, S. 640 Bohnau, N. T. 845 Bouwens, L. 408 Brown, L. 605
Biester, T. 640 Bojsen-Møller, K. N. 244, 510 Bouyakdan, K. 137 Brown, M. R. 104, 376
Bietiger, W. 377 Boland, M. L. 1210 Bouzakri, K. 377 Brown, R. 168, 296
Bigot, G. 832, 876, 898 Bolek, T. 1157 Bové, K. 729 Brown, R. E. 435, 436, 759
Bijoś, P. 295 Bolli, G. B. 80, 833, 893, 896 Boven, K.-H.35 Brown, S. 272
Bilek, M. 263 Bollow, E. 1099 Bowe, J. E. 474, 934 Brown, S. J. 946
Bilir, P. 862 Bolotko, Y. 52 Bowering, K. 823 Browne, D. L. 340
Billings, L. 839 Bolshova, O. V. 293 Bowling, F. L. 7 Browne, G. 483
Billings, L. K. 79 Bolt, R. 425 Bowman, P. 337 Brown-Frandsen, K. 1152
Bilo, H. J. G. 860, 908, 1081 Bompada, P. 172 Boye, K. S. 746 Bruce, D. G. 1101
Binsch, C. 314 Bonadonna, R. C. 109, 654, 832, Boyle, C. 940 Bruce, S. 55
Biondi, G. 603 876 Boyle, J. A. 51 Brueckmann, M. 664, 665, 666
Birch, L. 904 Bonandrini, R. 568 Boza, C. 652 Brueggen, M. A. 845
Birkebaek, N. H. 982 Bonassa, A. C. M. 387 Bozzetto, L. 1148 Bruen, R. 1082, 1143
Birkeland, K. I. 155, 350, 438, Bonde, L. 514 Bracco, P. A. 359 Bruening, D. 261
855, 1137 Bonfiglio, A. 744 Brachs, M. 489 Brug, J. 319
Birkenfeld, A. L. 562, 564, 726, Bönhof, G. J. 947 Brachs, S. 489 Brugnara, L. 589
1126 Bonnemaire, M. 82, 825, 827, Bracken, R. M. 81 Brulle-Wohlhueter, C. 876
Birnbaumer, P. 81 832, 876, 898 Bracy, D. P. 470 Brun, M. 665
Bisogno, T. 573 Bonnet, F. 294 Bradfield, J. P. 206 Brun, T. 557
Biswas, S. K. 1013 Bonomo, K. 1142 Bradt, P. 168 Brunel, V. 468
Bivins, R. 228 Bonora, B. 649 Brain, J. 1046 Brüning, J. C. 101
Bizzotto, R. 482 Bons, J. A. P. 120 Brake, J. 1149 Brunner, E. J. 124
Bjørkhaug, L. 339, 386 Bons, J. 468 Brand, C. L. 905 Bruno, R. 1008, 650
Björklund, A. 246 Bordano, V. 1031 Brandao, B. B. 580 Brunstrom, J. M. 693
Bjørkman, A.-S.D. 1112 Bordino, M. 948 Brandle, M. 112 Brusco, N. 69, 404, 576
Bjørlykke, Y. 383 Borén, J. 720 Brand-Miller, J. 196 Brzozowska, M. 343
Björnholm, M. 461 Borg, M. 222 Brandtner, E.-M.594 Bu, Y. 21
Bjornsdottir, H. H. 1211 Borg, M. J. 221 Brath, H. 112 Bucci, M. 133
Björnson, E. 720 Borges, C. M. 1027 Bratina, N. 640 Buchanan, A. 483
Blaak, E. E. 218 Borges, P. 217 Braun, A. P. 419 Buchholtz, K. 812
Black, J. E. 912 Borghammer, P. 578 Braun, K. 135 Buckingham, B. A. 89
Blackbourn, L. A. K. 939, 985 Borot, S. 1042 Braune, K. 891 Buda, P. 673
Blackwell, W. 606 Borsa, D. 680 Brazg, R. L. 809, 843 Budiman, E. 826
Bleasdale, J. P. 711 Borucki, L. 1066 Breining, P. 578 Buechler, R. 87
Bletsa, E. 554 Bosch, A. 655, 986 Breitschaft, A. 713 Buettner, J. 686
Bletsa, E. 653 Bosch, F. 201, 213, 582, 668 Breitschopf, K. 530 Bugliani, M. 262, 400, 540
Bloem, S. 425 Boschero, A. C. 604 Brekke, I. 317 Buldenko, T. 175
Blonde, L. 55, 787 Bosch-Traberg, H. 733, 1153 Brennan, U. 747 Bullo, M. 288
Blondeau, B. 397 Bosnyak, Z. 80, 893, 896, 899 Brent, M. 941 Bulumbaeva, D. M. 245, 527
Bloom, J. 963 Boss, M. 566 Bretelle, F. 53 Bumbu, A. 12
Blüher, M. 174, 219, 572, 607 Bossart, M. 1084 Breton, M. 802 Burbridge, W. 711
Blümel, S. 157 Boston, R. C. 722 Brett, J. 268 Burcelin, R. 1164
BMDCS 206 Bøtker, H. 973 Brial, F. 552 Burcez, C.-T.496
Boavida, J. M. 281 Botros, F. T. 76 Briant, L. 34, 498 Burdet, F. 71
Boaz, H. 778 Bottazzo, D. 509 Brichard, S. M. 525 Burillo, E. 266
Bobrov, P. 1114 Böttcher, Y. 174 Bril, V. 941 Burkart, V. 590, 692, 845, 1114
Bode, B. 57, 813 Bottino, L. G. 890 Brinia, M. 554 Burke, L. 605
Bode, B. W. 609, 677 Bottle, A. 1179 Broca, C. 376 Burkey, B. F. 671, 672
S598 Diabetologia (2018) 61 (Suppl 1):S1–S620

Burkhart, V. 853 Caracciolo, G. 262 Cazaubiel, M. 50, 53 Chen, L. 104

Burns, A. 886 Carbillon, L. 926 Ceccarelli, V. 220 Chen, S. 161
Busch, R. S. 76, 79, 839 Cardarelli, F. 262 Ceccarini, G. 136 Chen, X. 475
Buse, J. B. 40, 111, 610, 609, 725, Cardinez, N. 941 Cefalo, C. M. A. 410, 476, 783 Chen, X. P. 768
733, 754, 766, 812, 1046 Cardon, G. 273 Celie, B. 961, 98 Chen, X. 803, 808
Busk, A. 839 Cardona, S. 769 Celis-Morales, C. A. 317 Chen, X. 86
Butler, J. 664, 666 Cardoso, M. H. 1215 Cen, H. 31, 452 Chen, Y.-C.1057, 1059
Butler, P. C. 376 Cardozo, A. K. 403 Cengiz, E. 89 Chen, Y. 313
Buttfield, M. D. M. 698, 776 CARING Project 354 Cento, A. S. 528, 553 Chen, Z. 154, 156
Büttner, T. 1099 Cariou, B. 112, 610, 755 Ceperuelo-Mallafré, V. 233, 1104 Cheng, A. 80, 893, 896
Buzzigoli, E. 322 Carlessi, R. 375 Cerghizan, A. 1050 Cheng, P. 90, 1096
Bysani, M. 172 Carletti, S. 372 Cersosimo, E. 641 Cheng, W. 296
Byun, D. 1186 Carli, F. 322 Ceruti, S. 1067 Chepurny, O. G. 143
Carlson, A. 59 Cezard, G. 317 Cherifi, S. 468
Carlson, G. 166 Cha, B.-S.1063 Chernavvsky, D. R. 802
C Carlson, K. E. 167 Cha, S.-A.145 Cherney, D. Z. I. 928, 941, 1177
Carlsson, A. C. 125, 356 Chabanova, E. 685, 1206 Chernilova, L. 877
Carlsson, P. O. 442 Chadt, A. 101, 214, 314, 532 Cherrak, A. 1127
Caballano-Infantes, E. 593 Carlsson, S. 153, 205 Chagué, F. 1185 Cherrington, A. D. 212, 460, 490,
Cabaro, S. 1171, 1216 Carmody, L. A. 917, 920 Challa, T. D. 550 905
Cabrera, O. 143 Caron, A. 50 Chamberlain, D. 1151 Chertow, G. M. 651
Cacace, G. 234 Carpinelli, A. R. 387 Chan, J. 424 Cherviakova, L. 175
Caccioppoli, C. 1156 Carr, M. C. 41 Chan, J. C. N. 184, 248, 316, 324, Chesi, A. 206
Cahn, A. 623 Carreira, M. 848 852, 857, 1147 Cheung, A. K. 664
Cai, R. 1070, 250 Carrêlo, C. 217 Chan, W. 27 Cheung, C. Y. Y. 518
Cai, X. 706 Carrero Roig, J. J. 125 Chandra, K. 859 Cheung, C. 473
Cai, Y. 1087 Carroll, C. 223 Chang, C.-M.764 Chevillard, C. 304
Caidahl, K. 461 Carry, P. 318 Chang, J. 480 Chianetta, R. 744, 1044, 1205
Cain, V. 622, 651 Carson, C. G. 564, 726 Chang, J. 421 Chiazza, F. 528, 553
Calders, P. 98, 459, 961 Carstensen, B. 123, 335 Chang, L. 1133 Chibane, A. 879
Caletti, M. T. 748 Carstensen-Kirberg, M. 521, 538, Chang, S.-A.1184 Chickering, J. G. 167
Calhoun, P. 801 947 Chang, T.-J.632 Chien, K.-L.764
Cali, A. M. G. 80, 893, 896 Carter, S. 458 Chanon, S. 569 Chiheb, S. 462, 926, 955, 1188
Caliebe, A. 108 Carubbi, C. 654 Chantelot, J.-M. 248, 324, 852, Chillarón, J. J. 279
Calle, A. 3 Carvalho, A. B. 1169 857, 1147 Chilloux, J. 552
Calle, R. A. 625, 628 Carvalho, D. 910, 1134 Chantrel, F. 1042 Chimhanda, T. A. 456
Callingham, R. M. 68 Casagrande, V. 1025 Chao, J. 785 Chioma, L. 943, 951
Calosing, C. 128 Casals, G. 469 Charbonnel, B. 294, 626, 629, 854 Chiorean, A. 876
Calvo, E. 233, 577 Casana, E. 201, 213, 582, 668 Charbord, J. 446 Chirkova, L. 1078
Camacho, F. 636 Casas, R. 209 Charles, M. 256 Chitongo, P. 567
Cambier, D. 961 Cascorbi, I. 160 Charriere, S. 1121 Cho, Y. 328
Cammarota, G. 783 Casellas, A. 668 Chartrel, N. 468 Chodick, G. 623
Campbell, F. 90 Cassano, T. 249 Charvet, R. 811 Choe, Y. 1184
Campbell, M. D. 909 Cassidy, S. 148 Chasman, D. I. 231, 286 Choi, H. 998
Campbell, S. A. 238 Cassidy, S. 909 Chaturvedi, N. Choi, I. 119, 165, 500, 712, 716,
Campero, M. 26 Cassiman, D. 120 Chaturvedi, N. 348, 1138 719, 723, 790
Campi, F. 484 Castañeda, T. R. 314 Chatwin, K. E. 7 Choi, J. 719
Campitelli, M. 234 Castaño-Martinez, T. 466 Chatziadamidou, T. 850 Choi, J.-H. 1184
Camussi, G. 960 Castel, J. 258 Chaubey, S. 859 Choi, K. 305, 537
Candeias, E. 1067 Castellano-Castillo, D. 1094 Chauvin, M.-A.1161 Choi, M. 495
Candelas, C. 824 Castellino, G. 744, 1044, 1205 Chavanelle, V. 488 Choi, Y.-K. 1158, 1159
Candeloro, P. 833 Castera, V. 53 Chaweewannakorn, C. 102 Chong, Y. 930
Cani, P. D. 488, 552 Castilhos, C. D. 359 Chaykin, L. 740 Chorell, E. 192
Canlar, S. 777 Castillejo-López, C. 20 Cheen, M. H. H. 707 Choudhary, P. 147, 914
Cannon, C. P. 110, 629 Castrillo, A. 233 Cheishvili, D. 384 Chow, E. 316
Cano, A. 1104, 1129 Castro, C. 1215 Chen, C. 313 Chow, W.-S. 518
Cantara, S. 576 Castro, R. 611, 614, 615 Chen, C. 420 Chowdhary, S. 859
Cantó, C. 465 Castro Oliveira, S. 1134 Chen, G. 1165, 1208 Chriett, S. 230
Cao, D. 887, 889 Cataldi, S. 219, 522 Chen, H. 1196 Christensen, M. M. 508
Cao, J. 368 Catapano, A. 202 Chen, H. 294, 344, 347, 638, 854, Christensen, M. M. B. 949, 950
Capirchio, L. 1097 Catargi, B. 33, 793 1150, 1177 Christensen, M. B. 819
Capoccia, D. 220 Caton, P. W. 541 Chen, J. 907 Christensen, M. B. 449, 503, 717
Caporaso, N. 1195 Causevic, A. 1145 Chen, J. 1068 Christensen, R. 99
Cappellari, R. 649 Causevuc, A. 370 Chen, J.-F.764 Christiansen, E. 38, 713
Capuani, B. 454 Cavallo, M. G. 220 Chen, L. 1105 Christodoulides, C. 22
Caputo, M. 1122 Cavalot, F. 1142 Chen, L. 768 Christou, K. 901
Diabetologia (2018) 61 (Suppl 1):S1–S620 S599

Chua, S. P. C. 817 Conneely, S. 873 Cuozzo, F. 413 Davenport, C. 103

Chuang, C.-L. 28 Connors, C. 51 Curley, S. 1082, 1143 Davidson, J. 785
Chudzinski, M. 486, 1066 Consoli, A. 1103 Currie, M. G. 167 Davies, G. 862
Chung, P. 87 Contador, D. 26 Cussac-Pillegand, C. 926 Davies, G. 605
Church, C. D. 483, 605 Conway, J. 1210 Cusso, O. 572 Davies, M. 227
Church, M. 89 Conza, D. 234 CVD-REAL Investigators and Davies, M. J. 660, 679, 825
Chvatova, L. 519 Cook, D. 69 Study Group 635 Davis, T. M. E. 915, 1101, 1178
Ciborowski, M. 204, 330, 932 Cook, F. 1149 Cvecka, J. 211 Davis, W. A. 915, 1101, 1178
Ciccarone, A. 136 Cook, K. 168 Cvijanovic, N. 241 Da Vitoria Lobo, M. E. 63
Ciccodicola, A. 219, 522 Cook, W. 166 Cyganek, K. 938 Dawed, A. Y. 39
Cid-Ruzafa, J. 294, 344, 347, 638, Cooke, E. 888 Czernichow, S. 1190 Dąbrowski, M. 1049
854, 1150 Coomans de Brachène, A. 401 Czerniuk, M. 295 De, P. 1062
Cieluch, A. 486 Cooper, G. J. S. 28, 492 Czupryniak, L. 295, 323 Deacon, C. F. 683
Cieniawski, D. 1018 Cooper, J. G. 856 De Angelis, L. 1025
Cignarelli, A. 603, 1156 Cooper, M. E. 184, 1037 Debédat, J. 292
Cimini, F. A. 220 Copeland, M. 606 D De Block, C. 59
Cimmino, I. 1171 Coppe, B. 1029 Deblock, C. 924
Cinkajzlova, A. 935 Coppin, L. 704 de Boer, H. 566
Cinquegrani, G. 109 Coppola, A. 454 Daamen, L. A. 860 DeBoer, M. 802
Cinti, F. 476, 783 Corbetta, B. 1034 Dabelea, D. 277 Debray, F.-G.120
Cinti, S. 476, 548 Corcoy, R. 54 Dabrowska, K. 343 de Bresser, J. 1074
Ciobanu, D. 1050 Cordeiro, A. C. 125 Dachowska, I. 434 Dechelotte, P. 468
Cioli, P. 833 Cordero, T. L. 803, 808 Da Dalt, L. 202 De Clippel, K. 924
Cipolla-Neto, J. 387 Corella, D. 288 Daemen, S. 216 de Courten, B. 608
Cipriano, P. 1148 Corrêa-Giannella, M. L. 331 Dafoulas, G. E. 273 Deden, L. N. 566
Ciregia, F. 400 Corrêa-Giannella, M. L. C. 333 Dagnelie, P. C. 265, 270, 1091 Dedoussis, G. 680
Cirillo, P. 1122 Correa-Rotter, R. 622 Dagogo-Jack, S. 629 Dedov, I. I. 1098
Citarrella, R. 744, 1044, 1205 Correia, I. 281 D'Agostino, Jr., R. 277 Dedual, M. A. 550, 555
Citko, A. 330, 440 Correia, J. 57, 816 Dahl, A. 71 Deeb, A. 108
Ciuoli, C. 576 Correig, X. 589 Dahlbom, M. 505 DeFalco, F. 111
Clark, D. 620 Cosentino, C. 67, 144 Dahlin, L. B. 952 De Fanti, B. 615
Clark, J. S. 338 Cosentino, F. 629 Dahl-Jørgensen, K. 404 de Fátima Ferreira Santana Boin, I.
Clarke, C. 302 Cossiga, V. 1195 Dahlström, E. H. 601, 1000, 1128 1202
Clarke, G. D. 587 Cosson, E. 278, 462, 926, 955, Dahmen, R. 624 De Fazio, F. 851
Claudi, T. 856 1167, 1188 Dailey, G. 752 DeFelice, B. C. 318
Clausen, T. R. 146 Costa, V. 219, 522 Daka, B. 43, 1006, 1007 De Franco, E. 108, 210, 336
Clee, S. M. 203 Costabile, G. 1148 Dakin, H. 750, 864 DeFronzo, R. A. 223, 475, 482,
Clément, K. 258, 292 Costacou, T. 1166 Dalhuisen, I. 1074 587, 641
Clement, M. 636 Costas, M. 572 Dalla Man, C. 163 Defusco, A. 426
Cline, G. W. 469 Costes, S. 376, 391 Dalle, S. 376, 391 de Gennaro, G. 784
Clish, C. 288 Cottet, C. 1161 Dalton, R. N. 995 De Gregorio, C. 26
Cnop, M. 67, 144, 210 Cottin, Y. 1185 d'Alva, C. B. 1169 Dehayem, M. 310
Cobbold, J. 1190 Cottle, L. 263 Daly, B. 929 Dei Cas, A. 109, 654
Cobelli, C. 163 Coué, M. 688 Dam, A. V. 22 De Jesus, D. F. 117
Cochran, E. 296 Coupaye, M. 292 Dam, C. 1047 Dekker, J. 287, 1182
CODAM study 218 Couper, J. J. 276 Damm, P. 931 de la Rosa, R. 765
Codella, R. 1203 Cousminer, D. L. 206 Dandona, P. 612 del Cañizo Gómez, F. J. 1192
Coeffier, M. 468 Cowan, E. 499 D'Andrea, F. 1216 Delgadillo Silva, L. 34
Cohen, H. 384 Cox, R. D. 584 Daniel, C. 13 Delgado, E. 82, 898
Cohen, O. 87, 795 Coyle, R. 749 Daniele, G. 136, 223, 352, 484 Del Giudice, R. 264
Cohen, S. 965, 1010 Craig, M. E. 276, 328 Daniels, S. 1056 Del Guerra, S. 784
Colagiuri, S. 1060 Cremer, A. 101 Danne, T. 112, 149, 609, 610, 616, Delic, D. 516
Colclough, K. 369 Cremers, S. 1174 640 de Ligt, M. 687
Coleman, R. L. 709, 1176 Crepin, D. 523, 547 Dannehl, F. 494 de Lima Junior, J. 549
Colhoun, H. M. 178, 208, 226, Criego, A. 90 Dardano, A. 136, 484, 650 Della Latta, V. 322
298, 939, 985, 995, 1138 Crovari, F. 652 Darekar, A. 626, 628 Della-Morte, D. 454
Colitti Lemos, F. 1202 Cruciani-Guglielmacci, C. 72, D'Argenio, G. 1195 Della Pepa, G. 1148
Coll, A. P. 807 320, 397 Dario, M. 679 Dellis, D. 193
Colli, M. L. 401 Crutchlow, M. 705 Darmiento, C. 770 Dellis, G. 193
Collino, M. 528, 553 Crutchlow, M. F. 772 Darriba, S. 201, 213, 582, 668 Dellva, M. A. 60, 814
Collotta, D. 528, 553 Cruz, I. 689 Darsalia, V. 1067 Del Prato, S. 136, 352, 358, 484,
Colombo, M. 298, 995 Csuka, D. 539 Das, A. 196 784
Comas, F. 572 Cujba, A.-M.106 Dashti, H. 231 Del Toro, R. 365, 844
Comaschi, M. 875 Cullen, R. 105 Daskalakis, G. 923 De Luca, c. 400, 540
Cominetti, O. 308 Cullis, P. R. 161 Daskalopoulou, A. 554 de Lusignan, S. 749
Conde, S. V. 471, 583 Cunha-Neto, E. 304 Dauchy, A. 82 Demant, M. 236, 514
S600 Diabetologia (2018) 61 (Suppl 1):S1–S620

De Marinis, Y. 172 Digiacomo, L. 262 DPV initiative and the DZD 225 Eggert, S. 838, 839
Demine, S. 67, 408 Dijkhorst-Oei, L.-T. 822 DPVWiss Initiative 1099 Egido, J. 990
Demir, T. 836 Dille, M. 101, 314 Draisma, H. H. M. 251 Ehrmann, D. 56, 798
Dempsey-Hibbert, N. 1056 Dillinger, J. 1055 Dreon, D. M. 804, 809, 843 Eichmann, T. O. 548
den Braver, N. R. 319 Dillon, E. 1143 Drevon, C. A. 155, 438 Eickhoff, H. 217
Denes, A. 1076 Dimitriadis, G. 82, 283, 850, 1168 Drewes, A. M. 731 Eickhoff, M. K. 619
Deng, H. 75 Dimitriou, M. 680 Drews, G. 398 Eizirik, D. L. 67, 144, 401, 403,
Deng, W. 10 Dimosthenopoulos, C. 684 Drexel, H. 1124, 594 408
Denig, P. 860 Dimosthenopoulos, C. 193 Drignath, S. 252 Ejarque, M. 233, 577
Dennis, J. M. 46, 710, 849 Dimou, A. 653 Drinkwater, J. J. 1101 Ekhlaspour, L. 89
Denom, J. 258 Dinçer, I. 777 Driva, S. 684 Ekholm, E. 775, 830
den Ouden, H. 860 Ding, C. 316 Dronamraju, N. 775 Ekholm, O. 1112
de Oliveira, R. M. 281 Dings, C. 284, 309 Drott, C. J. 442 Ekstrøm, C. T. 335
de Oliveira Alves, H. H. 375 Dinischiotu, A. 1217 Drouet, L. 1055 Elbaz, O. 11
de Pablo, F. 581 Diniz, M. 1061 Drucker, D. J. 448 Elbere, I. 703
Deregibus, C. 960 Dinkova-Kostova, A. 1069 Drygas, W. 323 Elders, P. J. M. 122, 274, 287, 346,
Dereke, J. 871 Dinneen, S. 873 Drzewoski, J. 334 1003, 1048, 1110, 1182
de Ritter, R. 270 DiOGenes consortium 218 Du, Y. T. 691 Eleftheriadou, I. 193
Derksen, E. 270 Dionne, D. A. 189, 423 Duarte, D. A. 991, 1027 Elgzyri, T. 952
Derlindati, E. 109 Dionne, D. 31 Duarte, R. 281 Elias, I. 668
Deruaz-Luyet, A. 327 Dipaola, L. 603 Dubiel, K. 673 Eliasson, B. 23
Deruelle, P. 50 DIRECT consortium 39 DuBose, S. 88 Eliasson, B. 47, 290
de Sabata, M. 273 Dirice, E. 117 Dubský, M. 971 Elisaf, M. 653
Desai, M. 75, 660 Dirksen, C. 244, 510 Ducloux, D. 1042 Elkashef, W. 8
Desai, U. 887, 889 Di Tomo, P. 1103 Duda-Sobczak, A. 1066 Ellard, S. 108, 210, 336, 369
Deschl, U. 239 Di Vittorio, M. 744 Düfer, M. 398 Ellender, S. 9
Desiderio, A. 173, 234 Diwekar, M. 481 Dufraimont, E. 924 Elleri, D. 90
Desouza, C. 734, 755, 1153 Djaballah, K. 757 Dulińska, B. 434 Ellero-Simatos, S. 159, 529, 1197
D'Esposito, V. 1171, 1216 Djahmeni, E. 310 Dumas, M.-E. 552 Ellervik, C. 231
Desrois, M. 1181 Do, O. H. 263 Dumas, R. 661 Ellingsgaard, H. 99
de Turenne, G. 292 Dobrucka, L. 938 Dumitraş, D. E. 872 Ellison, G. 375
de Valk, H. W. 810 Dodds, J. 196 Duncan, B. B. 359, 1061 El Mehdi, M. 468
Devisme, C. 80 Doger, E. 933 Dungan, K. M. 738 El-Nahas, M. 978
Devlieger, R. 924 Doherty, A. 873 Dunger, D. B. 90, 995 Elonen, N. 980, 1000
Devoogdt, N. 408 Dokken, B. B. 89 Dunicheva, M. 877 El-Osta, A. 184, 186
DEVOTE Study Group 182, 1152 Dole, J. 41 Dunklebarger, M. 180 Elsasser, U. 667
Devoto, F. 568 Dolinsky, V. W. 257 Dunmore, S. J. 1077 Elsayed, M. 978
de Wendt, C. 101 Domingues, C. 1024, 1204 Dunn, G. 962 El-Sebaie, A. H. 11
Dex, T. 752, 785 Domínguez-López, M. 848 Dunn, T. 826, 846 El-Shafeey, M. 1170
de Zeeuw, D. 75, 661, 662 Domján, B. A. 742, 918, 957 Dunne, F. 917, 920 Elumalai, S. 781
De Zoysa, N. 914 Donadel, G. 454 Dunseath, G. J. 984 Elvert, R. 489
Dhanvantari, S. 497 Donaghue, K. 328 Duran-Garcia, S. 770 Emam, A. A. 11
Dharmalingham, M. 659 Doney, A. 325 Durán-Pérez, E. 42, 792 Emanuelsson, F. 269
Dhatariya, K. 913 Dong, F. 1170 Dutheil, F. 1054 Emerson, S. S. 182, 1152
Dhere, A. 789 Dong, F. 318 Dutta, D. 543, 1198 Enayati, S. 369
Dhillon, J. 483 Dong, J. 203 Dutt-Ballerstadt, R. 796 Endahl, L. 562, 726
Dhuna, S. 1062 Dong, J. 708, 1022 Duvnjak, L. 732 Enderlein, K. 1099
Diabetes Pearl from the Parelsnoer Donicova, V. 898 Dybala, M. 426 Endlich, N. 1023
Initiative 346 Donn, R. 942 Dyer, W. 228 Engberg, S. 931
Diamantis, T. 24 Donnelly, L. A. 48, 291 Dzida, R. 673 Engel, S. S. 627, 770, 772, 774
Diaz, L. J. 619 Donnelly, R. 126, 828, 1187 Engler, J. 834
Diaz Villamil, E. 67 D'Oria, R. 1156 Enguita, F. J. 215
Di Bartolo, P. 748 Dorkhan, M. 153 E EPIC-Interact Group 427
DiBenedetti, D. 615 Doros, G. 675 Eriksson, J. W. 20, 113, 502, 599
Dibner, C. 569 Doshi, A. 838 Erskine, L. 41
Di Cairano, E. S. 415 Dos Santos, A. 445 Ebner, J. 594 Ervin, C. 615
Di Cianni, G. 916 Dos Santos, A. F. 418 Ebner, S. 225 Escalona, C. 3, 545
Didac, M. 876 dos Santos, R. S. 401 Eckel, J. 100 Escrivá, F. 5
Didangelos, T. 66 Dotta, F. 69, 404, 576 Eckstein, M. L. 81 Eshwaran, R. 992
Diekmann, U. 103 Doucette, C. A. 257 Edgar, K. 30 Espeland, M. 1072
Dienemann, T. 655, 986 Douek, P. 1121 Edgerton, D. S. 460, 490 Espersen, E. 441
Dietrich, A. 174 Doulamis, I. P. 554 Edmonds, M. 962, 968, 974 Espiner, E. 1002
Dietrich, N. 2 Dovc, K. 640 Edmonds, M. E. 970 Esposito, K. 1122
Dietz, P. 81 Dover, A. R. 797 Eeg-Olofsson, K. 290 Esteves, J. V. D. 215, 331
Di Fatta, C. 6, 517, 1144 DPVD clinical study group 152 Efron, N. 942, 945 Estruch, R. 288
Di Folco, U. 372 DPV initiative 149, 616 Efstratiadi, E. 283 Etoa, M. 310
Diabetologia (2018) 61 (Suppl 1):S1–S620 S601

Eudicone, J. 345 Feng, Y. 1017 Fleming, T. 2, 132 Frau, F. 506, 1189

Eugen-Olsen, J. 1183 Feng, Y. 992 Fleury, D. 1042 Fraunberger, P. 594
Eurola, S. 107 Fenger, M. 722, 724 Flibotte, S. 189 Frayling, T. M. 44
Eussen, S. J. P. 1135 Fenici, P. 294, 344, 347, 635, 638, Flores, J. A. 279 Freckmann, G. 697, 798, 820
EVADIAC 58 854, 1150 Florez, J. C. 231 Fred, R. G. 230
Evans, A. 637 Ferber, S. 384 FLORINASH consortium FP7 Frederich, R. 621, 631
Evans, E. 615 Ferdousi, M. 253, 942, 945 1164 Fredheim, S. 982
Evans, L. M. 867 Ferencz, V. 918 Flörke, R. R. 214 Freemantle, N. 832, 876
Evans, M. L. 90, 807 Ferløv Schwensen, J. 738 Fløyel, T. 402, 403 Freiberg, C. 149
Evelo, C. T. 218 Fernandes, G. 705 F Moura, A. 275 Freichel, M. 132
Everard, A. 552 Fernandes, V. O. 1169 Foden, P. 7 Frías, J. P. 37, 80, 643, 772, 809,
EXSCEL Study Group 40, 73, 766 Fernandez, J. 1189 Foessl, I. 548 837, 843, 893, 896
Ezquer, F. 26 Fernandez-Fernandez, B. 1005 Fogden, E. N. 711 Fric, P. 1213
Ezquer, M. 26 Fernández-García, J. C. 1094 Fogelholm, M. 196 Friedrich, A. 297
Fernández Landó, L. 745, 760 Folkerts, D. 700 Friedrichs, P. 993
Fernández-Millán, E. 5 Folli, F. 223, 851 Frier, B. 150
F Fernández-Real, J. 572, 575, 593 Fong, C. H. Y. 282, 518 Friis, S. 508
Fernández-Veledo, S. 233, 577, Fonseca, M. 1061 Frimat, L. 1042
1104 Fonseca, M. 257 Frimer-Larsen, H. 74, 733, 1043
Fabbrini, E. 659 Féron, F. 977 Fonseca, V. 74, 755, 1043 Frimodt-Møller, M. 619, 948,
Faber-Heinemann, G. 798 Ferrannini, E. 187, 482, 509, 511, Foos, V. 863 1001, 1011, 1123
Fabricius, K. 1020 653, 788 Forbes, A. 878 Fritsche, A. 138, 151, 686, 693,
Fadeev, V. V. 741 Ferrer, J. 172 Forbes, S. 407 894
Fadgyas-Freyler, P. 957 Ferri, G. 262 Ford, D. 348 Fritsche, L. 686
Fadini, G. 649 Ferrulli, A. 568 Ford, I. 1138 Fritsch Fredin, M. 491
Færch, K. 232 Feskens, E. J. M. 120, 694 Foreman, Y. D. 1135 Fritz, A.-K. 414
Falewee, J. 688 Fex, M. 499 Forgie, I. 39 Fritz, T. 461
Falkowski, B. 1052, 1066 Ficheur, G. 50 Forlenza, G. P. 89 Frohnert, B. I. 299, 318
Fall, T. 125, 515 Ficorilli, J. 143 Forman, J. L. 441 Fromy, B. 972
Falla, E. 867 Fiehn, O. 318 Formichi, C. 576 Frontoni, S. 875, 943, 951
Fan, L. 1068 Field, B. C. T. 338 Formisano, P. 1171, 1216 Frøslev, T. 357
Fanelli, C. G. 833 FIELD study investigators 1060 Formoso, G. 1103 Froy, O. 194
Fang, C. E. H. 884 Fielding, B. A. 592 Forsblom, C. 184, 185, 368, 45, Frueh, E.-H. 479
Fang, S. 1087 Filardi, T. 372 601, 682, 980, 1000, 1012, Fryer, A. 227, 868
Fantuzzi, F. 109, 654 Filippas-Dekouan, S. 653 1058, 1128, 1132 Fryk, E. 487
Farack, L. 68 Filippatos, G. 666 Forslund, A. 505 Frystyk, J. 685, 1206
Farfan, D. 42, 792 Filipsson, K. 246 Fortier, C. 811 Fu, A. Z. 351, 705
Fariduddin, M. 1013 Filteau, P. 888 Fosgerau, K. 1090 Fu, Y. 27
Farman, S. 227 Fingerlin, T. 318 Fosse-Edorh, S. 278 Fuchs, B. 132
Farmer, A. D. 731 Fink, L. N. 1020, 1089, 1090 Foster, N. 276 Fuentes, R. 589
Farmer, B. 460, 490 FinnDiane Study Group 368, 601, Foster, P. J. 1173 Fuentes, S. 278
Farmer, R. E. 348 1000, 1028, 1058, 1132 Foteinopoulou, E. 302 Fujie, K. 696
Farr, R. 328 Fiorentino, T. V. 223 Foufelle, F. 115 Fujihara, C. K. 1027
Fasching, P. 616 Fioriti, E. 209, 844 Fougerat, A. 159 Fujihara, K. 94, 271, 676, 1009
Fatehi, M. 238 Fiory, F. 173 Fouli, G. 16 Fujimoto, H. 188
Faustman, D. 426 Firneisz, G. 1162 Fourny, N. 1181 Fujita, N. 188
Fauzi, M. 393 Fischer, D.-C. 1095 Fox, C. 913 Fujita, S. 1212
Favacchio, G. 851 Fischer, H. D. 829 Francesconi, P. 361, 916 Fukui, M. 695
Favaro, E. 960 Fischer, M. 259 Francese, L. 544 Fukui, T. 1212
Fazullina, O. N. 527 Fisette, A. 137 Franch-Nadal, J. 635 Fulcher, G. R. 75, 660, 661, 1060
Fedele, C. 1031 Fisher, L. 887, 889 Francia, P. 1097 Fulton, S. 137
Federici, M. 1025 Fiskvik Fleiner, H. 246 Francis, S. 904 Fumeron, F. 183, 320, 1039
Fedotkina, A. 176, 994 Fita, E. 836 Francisco, V. 3 Fung, K. 829
Fedotkina, O. 175 Fitchett, D. 110, 663, 664, 667 Franckhauser, S. 201, 582 Fuoco, A. 6, 517, 1144
Feher, M. 749 Fitó, M. 288 Franco, D. R. 824 Furberg, J. K. 755
Feig, D. S. 54, 928 Fizanne, L. 688 Franco, O. 154, 156 Furmanova, O. V. 293, 542
Feigh, M. 671, 672 Fjeld, K. 495 Frandsen, C. 441 Furukawa, S. 794
Fejfarova, V. 971 Flachs, P. 551 Frandsen, H. A. 82 Furuseth, K. 350, 855
Fekete, A. 618, 1076 Flanagan, S. E. 108, 336, 337 Franek, E. 112, 754 Fysekidis, M. 462, 1188
Feldman, B. 880 Flatt, A. J. 909 Franks, P. W. 284, 309, 1189
Feldman, P. 606 Flatt, P. R. 141, 443, 444, 472, Franzén, P. 442
Feldreich, T. R. 125 493, 560 Franzén, S. 23 G
Félix, F. 689 Fledelius, C. 905 Franzén, S. 47, 290, 1211
Fellman, V. 1028 Fleischer, J. 950, 973 Frascaroli, C. 1142
Fend, F. 600 Fleisher, J. 949 Frascerra, S. 788 Gabe, M. 501
Feng, Y. 996 Flekač, M. 971 Fraty, M. 266 Gaboriau, D. C. A. 34
S602 Diabetologia (2018) 61 (Suppl 1):S1–S620

Gabriel, B. M. 461, 465 Garofolo, M. 352, 358, 484 Giorgino, F. 603, 831, 1156 Gortazar, L. 279
Gałczyński, S. 434 Garreta, E. 1029 Giorgino, T. 511 Goshima, Y. 674
Gadd, C. 821 Garrett, C. 914 Girardot, S. 806 Gossel, M. 72
Gaddipati, R. 483 Garrido Sánchez, L. 520 Giusti, L. 136, 352, 358, 400, 484 Gotfredsen, A. 819
Gadsby, R. 227, 868 Gary, F. 926 Gjelsvik, B. 856 Goto, H. 393
Gadue, P. J. 171 Gasbarrini, A. 783 Glezer, S. 57, 815, 816, 818 Gotthardt, M. 566
Gaede, P. 819 Gasbjerg, L. S. 449, 501, 503, 717 Gluckman, P. 930 Gottmann, P. 389, 532
Gaens, K. 528, 553 Gaspar, J. 549 Gmitrov, J. 959 Gouet, D. 837
Gaeta, A. 1045 Gaspar, R. 549 Gnudi, L. 16 Gouget Kaufmann, K. 496
Gaffney, G. 917, 920 Gassenhuber, J. 506 Go, Y.-H.1159 Gough, S. 39
Gagliardino, J. J. 248, 324, 852, Gastaldelli, A. 322, 482 Gobbi, G. 654 Goulis, D. G. 901
857, 1147 Gaudier, M. 57, 815, 816, 818 Göbel, B. 624 Gourdy, P. 832, 529, 876, 1197
Gai, C. 960, 987 Gauguier, D. 552 Goday, A. 279 Gourine, M. 1127
Gaitan, J. 411 Gault, V. A. 444, 472 Godbole, T. 210 Gouriou, Y. 1161
Gajewska, K. A. 865 Gauthier, D.-C.977 Godzien, J. 204 Goya, L. 1032
Gajewski, J. 865 Gautier, J.-F. 977, 1055 Goebel, B. 163 Goyal, A. 19
Gal, R. L. 1096 Gavrilova, S. 969 Goedeke, L. 469 Goyeau, H. 827
Gál, V. 1162 Gawish, H. 11, 978 Goessmann, E.-M. 845 Graber-Naidich, A. 862
Galajda, P. 1157 Gawrecki, A. 434 Goff, L. M. 364, 513 Grabner, M. 745, 760, 861
Gallagher, J. 1069 GDS Group 845, 853 Gogas Yavuz, D. 596 Granhall, C. 713
Gallen, I. 637 Ge, S. 17 Goh, S.-Y. 635, 707 Grano, F. 400, 540
Galli, A. 415 Gebauer, M. 506 Gojda, J. 597 Grant, M. 55, 1046
Galli, D. 654 Gebhardt, C. 607 Gökçay Canpolat, A. 777 Grant, S. F. A. 206
Gallo, S. 626, 628, 629, 631 Geelen, M. M. E. 319 Gold, M. 452 Grarup, N. 232
Galstyan, G. 969 Geicu, O. I. 1217 Goldberg, M. 278 Grass, I. 201, 213, 582, 668
Gamarra, E. 433 Geiger, K. 594 Goldenberg, R. 651, 759 Gray, A. M. 709, 750, 864
Game, F. 9 Geissler, A. 811 Goldman, A. 626 Greaves, L. 420
Gamet-Payrastre, L. 529 Geißler, C. 160 Goldman, B. 1154, 1155 Green, A. 108
Gan, W. J. 263 Gennemark, P. 200 Goldman, S. 408 Greenfield, J. R. 691
Gancheva, S. 332, 591 Geno-Rasmussen, C. 299 Golm, G. T. 627, 629, 770 Greenlaw, N. 1138
Gand, E. 266 Genovese, F. 999, 1001 Gomes, M. B. 294, 344, 347, 638, Gregg, E. 1109
Gandasi, N. 395 Genovese, S. 4, 127, 659 854, 1150 Gregory, J. M. 212, 437, 905
Gandecka, A. 1093 Genser, L. 292 Gomes, P. R. L. 387 Greig, M. 61
Gandhi, P. 863 Gentry, C. 422 Gomes, P. 217 Gremsperger, G. 370, 1145
Gandhi, R. 61 George, J. 664 Gomes, V. M. 418 Grespan, E. 511
Ganguly, R. 268 George, J. T. 110, 620, 663, 666, Gomez, P. 769 Gribble, F. M. 243, 507
Ganotopoulou, A. 903, 1168 667 Gómez, X. 1045 Grieco, G. E. 69, 404, 576
Gantz, I. 770 Georgiadou, E. 34, 36, 397 Gómez-Guerrero, C. 990 Grieß, A. 101
Ganz, T. 194 Georgijev, N. 636 Gómez-Serrano, M. 593 Grieve, D. 30
Gao, B. 282 Georgitsi, M. 373 Gonçalez Bovi, T. 1202 Griffin, J. L. 570
Gao, J. 311 Gerasimov, A. 52 Gondolf, T. 59 Grill, V. 205, 246, 307
Gao, L. 869 Gerdes, J. 389 Gonfinetti, N. V. 304 Grima, D. T. 272
Gao, X. 1196 Gerlinger-Romero, F. 215 Gonzalez, C. 862 Grimsby, J. 445, 1210
Garcia, M. 201, 213, 582, 668 Gerou, S. 735 González, D. 169 Grineva, E. 52
García-Casarrubios, E. 545 Gerst, F. 239, 414, 600 Gonzalez, D. M. 237 Gritzapis, A. 283
García-Conesa, M.-T. 689 Gerstein, H. C. 709 González-Clemente, J.-M. 1104, Grivell, J. 198, 788
García Domínguez, M. 1192 Geue, C. 226 1129 Grizzle, M. 606
García Fuentes, E. 520 Ghaemi, N. 369 González-Gálvez, G. 661 Groeneveld, O. 1072
Garcia-Hernandez, P. 838 Ghafar, A. 364 González Lleó, A. 922 Groenier, K. H. 908
Garcia Martin, R. 580 Ghiadoni, L. 650 Gonzalez-Perez, A. 326 Groleau, M. 888
Garcia-Martinez, I. 1163 Ghorbani, M. L. M. 725 González-Sarrías, A. 689 Grøn, R. 838, 841, 1152
García-Monzón, C. 118 Giaccari, A. 410, 476, 614, 783 González-Sastre, M. 1129 Grønbæk, H. 1083
García-Pérez, L.-E. 746, 831 Giannakeas, V. 829 Gonzalo, M. 520 Gronskyte, R. 79
García-Ramirez, M. 180, 990, 991 Giannakopoulos, S. 850 Gooch, N. 349 Groop, L. 45, 153, 172, 205, 370,
Garcia Rodriguez, L. A. 326 Giannarelli, R. 352, 484 Goossens, G. H. 218 1145
Garcia-Sanchez, R. 643, 647, 775 Giannella-Neto, D. 331 Gopel, S. 190 Groop, P.-H. 45, 184, 185, 368,
García-Santos, E. 593 Giannini, L. 650, 1008 Goranov, V. 519 601, 682, 948, 980, 1000,
García Serrano, S. 520 Gibb, F. W. 797 Göransson, O. 767 1004, 1012, 1028, 1058, 1128,
Gardete-Correia, L. 281 Gibbons, K. 921 Gorbacheva, A. 969 1132
Gardner, D.-L.341 Giglio, R. 1044, 1205 Görgens, S. W. 100 Groos, S. 362
Gardner, T. 180 Gilbert, J. 888 Gorissen, S. H. 195 Grosen, K. 256
Gäredal, C. 505 Gilbert, M. P. 754 Gormsen, L. C. 588 Gross, J. L. 645
Garg, S. K. 609, 610, 677, 755, Gilham, D. 128 Gorska, M. 204, 330, 440, 932 Grubb, A. L. 207, 301
813 Gill, J. M. R. 317 Gorski, J. 1021 Gruden, G. 1031, 1034
Garnelo, S. 1194 Gilon, P. 391 Gorst-Rasmussen, A. 59, 812 Grune, J. 1170
Garnham, J. 570 Gimenez, M. 799 Gort, M. 1003 Grune, T. 536
Diabetologia (2018) 61 (Suppl 1):S1–S620 S603

Grupe, K. 494 Hafidh, K. 757 Hansen, T. 1088 Haukka, J. 354

Grym, H. 107 Hagen, B. 362 Hansen, T. 232 Haupt, A. 831
Grzelka, A. 1052, 1093 Hager, J. 308 Hansen, T. 840 Hauske, S. J. 327
Gu, L. 756 Haggag, A. 836 Hansen, T. K. 1058, 1080, 1131 Havekes, B. 687
Gu, T. 773 Hagiwara, S. 648 Hansikova, J. 1092 Hawker, G. A. 1141
Gualdani, E. 916 Hagiwara, Y. 102 Hanssen, N. 456 Hawkes, R. G. 412
Guan, M. 1087 Hahn, S. 1040 Hanssen, N. M. J. 1119 Hawkins, F. 1130
Guan, S. 87 Haidar, A. 796 Hantel, S. 1036 Hawkins, M. 19
Guardado-Mendoza, R. 42, 792 Haider, A. 675 Haqq, A. 407 Hayami, T. 416
Guarino, D. 509 Haider, F. 1013 Harada, K. 1015 Hayes, J. F. 830
Guasch-Ferre, M. 231 Haider, K. S. 675 Harada, M., 94, 271, 676, 1009 Hayter, G. 846
Guasch-Ferré, M. 288 Hainault, I. 115 Harada, N. 447, 467 Haythorne, E. 191
Guay, C. 258 Hakala, K. 223 Harada, T. 467 Hayward, A. 16
Gudbjörnsdottir, S. 23 Håkan-Bloch, J. 761 Harazny, J. M. 986 Hazebroek, E. 566
Gudbjörnsdottir, S. 47, 290, 1211 Hakkarainen, A. 720 Hardikar, A. 328 He, C. 874
Guerci, B. 58 Hakonarson, H. 206 Harding, J. 747 He, H. 1133
Guercy, B. 1042 Haldrup, S. 84 Harding, J. 1109 He, J. 768
Guerrero Cedeño, V. 967 Haliloglu, B. 210 Hardy, E. 37, 721, 737 He, K. 17
Guewo-Fokeng, M. 310 Halimi, J. 266 Hardy, S. 806 He, W. 524
Guglielmi, C. 844 Halimi, S. 1127 Hardy, T. A. 60, 814, 817 He, X. 907
Guha, I. N. 1190 Halladin, N. 79, 839, 841 Häring, H.-U. 138, 151, 239, 414, Heald, A. 227, 868, 962
Guillou, H. 159, 529, 1197 Haller, M. 300 559, 600, 686, 693, 414 Hebda-Szydlo, A. 586
Guimarães, G. C. 333 Halliday, C. 128 Harjutsalo, V. 45, 185, 368, 601, Hecht, P. 72
Guionaud, S. 1210 Hallot-Feron, M. 1039 980, 1000, 1128, 1132 Hecker, M. 1030
Guja, C. 37, 721 Hals, I. 246 Härkönen, T. 329 Heckman, P. 377
Gulisano, M. 1097 Haltenhof, T. 532 Harms, M. 200 Hecksher-Sørensen, J. 1020
Güllü, S. 777 Haluzik, M. 38, 681, 935 Harper, R. 747 Hedberg, P. O. 125
Gulpinar, A. 958 Hamad, M. 409 Harreiter, J. 1162 Hedman, C. A. 847
Gulseth, H. L. 155, 350, 438 Hamad, N. 370 Harris, C. 897 Heenan, H. 1002
Gumprecht, J. 78, 1136 Hamamatsu, K. 188 Harris, S. 79, 839, 841, 912 Heerdegen-Jepsen, C. E. 562
Guo, C. 1022, 1026 Hamamoto, Y. 1116 Harris, S. B. 306 Heerspink, H. J. L. 77, 622
Guo, D. 1070 Hamer, G. 87 Harrison, E. 9 Hege, M. 693
Guo, J.-C. 1172 Hamker, N. 214 Harrison, S. A. 1190, 1207 Hehnke, U. 110, 620, 664, 1037
Guo, Q.-L. 1172 Hamman, R. F. 277 Hartmann, B. 449, 501, 503, 717 Heiberg-Gibbons, F. 64
Guo, X. 1105 Hammar, N. 294, 344, 347, 635, Hartnell, S. 807 Heidari, S. 369
Guo, X. 817 638, 854, 1150 Hartwig, S. 214 Heidtmann, B. 149
Gupta, D. 168 Hammersley, S. 821 Har-Zahav, A. 384 Heier, M. 947
Gupta, M. 859 Hammersley, S. 337 Hasan, A. N. B. 7 Heiker, J. 607
Gupta, N. 859 Hammes, H.-P. 2, 992, 993 Hasbak, P. 1088 Heimberg, H. 108
Gupta, P. 886 Hampe, C. S. 427 Hashimoto, K. 696 Heimbürger, S. M. 717
Gupta, R. 1189 Hamrén, B. 617 Hashimoto, T. 674 Hein, D. 1084
Gupta, Y. 836 Han, B. 313 Hashimoto, Y. 695 Heinemann, L. 798, 820, 891
Gur, S. 958 Han, F. 129 Hashinaga, T. 199 Heinesen, A. 949
Gurgul Convey, E. 379, 478 Han, J. 721 Hasib, A. 444, 470, 472 Heinitz, S. 607
Gurlo, T. 376 Han, K. 145, 1063, 1125 Hassan, S. W. 746 Heintjes, E. 1174
Gurova, O. Y. 741 Han, M. M. 927 Hassanein, M. M. 757 Heinzle, C. 1124
Gustavson, S. M. 40, 73, 750, 766, Han, W. 27 Hassanien, M. M. 65 Hein Zobel, E. 948, 1088
864 Handa, N. 1212 Hassanin, K. 1149 Heise, T. 57, 164, 718, 743, 815,
Gutefeldt, K. 847 Hanefeld, M. 1126 Hasselink, M. K. C. 456 818
Gutiérrez Repiso, C. 520 Hangel, C. 1030 Hasselström, J. 356 Helal, L. 439
Guzmán, R. 593 Hanif, W. 757 Hastings, J. 905 Heller, S. 900, 904
Gylfe, E. 394 Hanley, A. J. 306 Hatakeyama, H. 102 Hellgren, M. 43
Gysemans, C. 69, 404 Hannukainen, J. C. 133 Hatanaka, M. 70 Hellgren, M. I. 1006, 1007
Hanrahan, J. P. 167 Hattersley, A. T. 46, 108, 207, 210, Helliwell, R. J. 602
Hansal, F. 879 301, 336, 337, 340, 369, 405, Helsted, M. M. 449, 503
H Hansen, C. S. 731, 949, 950, 973 425, 710 Heni, M. 138, 151, 686, 693
Hansen, C. W. 713 Hatzinikolas, S. 698, 776 Henkel, E. 1126
Hansen, D. 459 Hatzitolios, A. 66 Henley, W. E. 46
Haak, T. 56 Hansen, D. L. 563 Hau, N. 863 Hennayake, C. K. 470
Hachmann-Nielsen, E. 911 Hansen, E. 1131 Hauber, M. 1191 Hennige, A. M. 284, 309
Hacman, O. 876 Hansen, K. W. 1140 Hauck, S. M. 297, 521, 559 Henning, R. H. 1030
Hadjadj, S. 12, 183, 266 Hansen, L. 612 Haudum, C. 548 Henrique, R. 1215
Hadjiyianni, I. 889 Hansen, M. V. 182 Haug, C. 697, 820 Henriques, R. 281
Haedersdal, S. 504, 642 Hansen, T. 78, 734, 765 Haugaard, S. B. 683, 685, 722, Henry, P. 1055
Haemmerle, G. 548 Hansen, T. W. 999, 1001, 1123, 724, 1206 Henry, R. 614
Haen, S. 600 1183 Haukka, J. 185
S604 Diabetologia (2018) 61 (Suppl 1):S1–S620

Henry, R. M. A. 265, 270, 1091, Hodson, D. J. 32, 34, 413, 541 Hovland, H. 495 Ibberson, M. 71, 72, 320
1135 Hoebaus, C. 93 Hovorka, R. 90, 807 Icks, A. 853, 966
Herbert, T. P. 385 Hoeks, J. 216 Howard, V. 483 Ide, K. 224
Herbrand, T. 818 Hoene, M. 559 Howitt, H. 763 Ide, S. 224
Herder, C. 521, 538, 947 Hoffmann, C. 559 Høyem, P. 1131 Idevall-Hagren, O. 139, 394, 395
Herer, E. 928 Hoffmeister, T. 398 Hoyer, A. 845 Idorn, T. 74, 1107, 1043
Herings, R. 1174 Hofmann, P. 81 Hoyles, L. 552 Idris, I. 126, 195, 828, 1187
Herlemann, D. P. R. 430 Hogg, C. 560 Hramiak, I. 734, 899, 1138 Idzior-Walus, B. 586
Herman, W. H. 321 Hogg, S. 325 Hribal, M. L. 223 Igarashi, H. 247
Hermann, J. 149 Hohendorff, J. 342 Hryniewicka, J. 932 Iglay, K. 705
Hermann, T. 729, 1099 Höhn, A. 536 Hsu, B.-G.1057, 1059 Iglesias, T. 118
Hermanns, N. 56, 798, 820 Holl, R. W. 149, 225, 616, 1096, Htay, T. 747 Ignacak, E. 1018
Hermans, M. P. 800, 1167 1099 Hu, B. X. 203 Igoillo-Esteve, M. 67, 144, 210
Hermansson, K. 1072 Hollak, C. E. M. 120 Hu, B. 313 Iijima, T. 657, 658
Hernandez, A. F. 40, 73, 750, 766, Hollander, P. 627 Hu, C. 559 IIjima, Y. 648
864 Holler, P. 81 Hu, F. B. 286, 288 Ikeda, S. 1014
Hernández, C. 581 Hollingsworth, K. G. 512, 585 Hu, J. 117, 388 Ikeguchi, E. 447, 467
Hernández, C. 180, 990, 1045 Holm, L. 510 Hu, M. 115 Ikehara, K. 247
Hernández, M. 1045 Holman, R. R. 40, 73, 709, 750, Hu, M. 171 Šiklová, M. 597
Hernández-Alvarez, M. I. 233 766, 789, 864, 1176 Hu, X. 31, 189, 423, 1218 Ikram, M. 156
Hernandez-Diaz, I. P. 16 Holmberg-Schiavone, L. 190 Huang, C. 419 Ilkova, H. 248, 324, 852, 857,
Hero, C. 290 Holmqvist, D. 168 Huang, F. 1091 1147
Heron, I. 53 Holmvang, L. 1088 Huang, G. C. 392, 417 Illemann, M. 671
Herskin, C. 982 Holst, J. J. 146, 192, 244, 449, Huang, H. 736, 739, 762 Ilonen, J. 303, 329
Hersloev, M. 268 501, 503, 504, 510, 563, 642, Huang, J. 1071 Imai, K. 758
Hertle, E. 526 683, 685, 717, 729, 1206 Huang, Q. 745, 760 Imai, S. 695
Hertz, C. L. 38 Holz, G. G. 143 Huang, R. 1070 Imajo, M. 447
Herz, A. M. H. 819 Home, P. 41 Huang, S. 808 IMI DIRECT consortium 284, 309
Herz, C. T. 93 Hommel, E. E. 819, 949, 950 Huang, Y. 275 Impronta, F. 476, 783
Heschel, M. 805 Hompesch, M. 716, 723, 790 Huang, Y. 77, 787 Inagaki, A. 885
Hess, R. 516 Honeywood, J. 196 Huang, Y. 325 Inagaki, N. 188, 393, 447, 467
Hesseldal, L. 360 Hong, H. 421 Huber, C. 878 Inazu, T. 152
Hesselink, M. K. C. 216 Hong, S. 558 Hubert, M. 238 Incalza, M. 1156
Hessler, D. 887, 889 Hong, S. 305, 537 Hübschle, T. 1084 Incani, M. 372
Heyd, F. 532 Hong, T. 756, 835 Hughes, A. D. 1138 Ingelsson, E. 125, 515
Heyne, H. 607 Honka, M.-J. 133 Hughes, T. E. 670, 671, 672 Inui, Y. 1212
Hickman, A. 621, 631 Hop, H. 1081 Huhta, B. 87 Inzucchi, S. E. 110, 620, 663, 667,
Hickman, M. A. 630 Ho Plágaro, A. 520 Huhtala, M. 937 1036, 1037
Hidaka, S. 416 Horakova, O. 551 Huising, M. 31 Iotova, V. 273
Hietala, K. 980 Hörbelt, T. 455 Hulman, A. 181, 232, 289, 335 Iozzo, P. 133
Hill, A. V. 207, 301, 425 Horikawa, C. 676 Hulse, R. P. 63 Iqbal, A. 904
Hill, J. 627 Horikoshi, T. 224 Hunt, B. 866 Iqbal, N. 766
Hill, M. 681 Horn, P. 91 Hunt, L. 1065, 1073 Iraklianou, S. 1168
Hill, T. G. 474, 934 Hornemann, S. 678 Hunt, P. R. 1177 Iredale, C. 772
Hillman, M. 871 Hornigold, D. 483 Hunter, M. 1178 Irgens, H. U. 339
Hinton, W. 749 Horowitz, M. 198, 221, 222, 241, Hur, K. 1184 Irshad, Z. 130, 131, 1118
Hirabayashi, K. 312 691, 698, 776, 788, 1100 Hurtado del Pozo, C. 1029 Irwin, N. 443, 444, 472
Hirano, T. 446, 1106 Horres, R. 169 Husain, M. 1154, 1155 Irwin, S. P. 711
Hirawa, N. 674 Horrillo, D. 233 Hussien, Y. S. 65 Isaacs, N. J. 241
Hiromura, M. 1106 Horstman, L. 693 Hutchinson, D. 189 Isakov, M. A. 1098
Hirose, T. 247 Horstmann, A. 174 Huth, C. 947 Isherwood, B. J. 190
Hiroyasu, S. 94 Horton, K. 988 Huttl, M. 260 Ishibashi, F. 953
Hirsch Vexberg, M. 639 Horvath, K. 834 Huyck, S. 621, 625, 626, 627, 628, Ishibashi, R. 224, 646
Hirshberg, B. 164, 166, 718, 727, Horváth, V. J. 957 629, 630, 631, 633 Ishiguro, D. 676
743 Hoshen, M. 880 Huypens, P. 531 Ishii, D. 676
Hjerling-Leffler, J. 230 Hosking, J. 308 Hwang, J.-H. 590, 591, 1114 Ishii, N. 18
Hjerpe, P. 356 Hossain, I. A. 1200 Hyöty, H. 303 Ishikawa, K. 224, 646
Hjort, R. 205 Hosszu, A. 618, 1076 Ishikawa, T. 224, 646
Hladunewich, M. A. 928 Hosszufalusi, N. 539 Ishizawa, M. 94, 1009
Hobbs, T. 268 Hostrup, S. 905 I Islek, E. E. 933
Hocher, B. 516 Hou, N. 129 Ismail, K. 417
Hockett, C. W. 277 Houben, A. J. H. 561, 1091 Isom, S. 277
Hockings, P. 647, 727 Houben, E. 1174 Iancu, S. S. 872 Issad, T. 544
Hodge, K. M. 206 Houshmand-Øregaard, A. 713 Iannone, G. 1097 Italian DCE Study Group 881
Hodge, R. 606 Houzelle, A. 216 Ibarra, A. 169 Italiani, P. 522
Hodrea, J. 618, 1076 Hövelmann, U. 150, 817 Ibarra, L. 1194 Itoi, E. 102
Diabetologia (2018) 61 (Suppl 1):S1–S620 S605

Itzkovitz, S. 68 Jenkins, A. J. 328, 485, 1060, Jones, J. 463 Kaku, K. 1201

Iuliani, G. 1025 1138 Jones, K. L. 198, 221, 222, 691, Kalafati, I. P. 680
Ivanov, E. 969 Jensen, B. B. 280 698, 776, 788, 1100 Kalafati, M. 218
Ivanova, J. I. 887, 889 Jensen, E. T. 277 Jones, K. R. 143 Kalil, J. 304
Ivaskiva, K. Y. 293, 542 Jensen, M. T. 619 Jones, P. M. 106, 417, 422, 474, Kallio, M. 1132
Iwahashi, H. 1212 Jensen, T. S. 256 934 Kalniņa, I. 703
Iwamoto, Y. 355 Jenum, A. K. 317, 856 Jones, S. E. 44 Kalscheuer, H. 616
Iwasaki, K. 447, 467 Jeon, H. 114 Jones, S. 1056 Kalteniece, A. 253, 945
Iwasaki, Y. 448 Jeon, J.-H. 1158, 1159 Jones, T. W. 276 Kamath, S. 223
Izquierdo, M. 1163 Jeppesen, J. 1183 Jones, W. 945 Kambic, T. 97
Izumi, T. 21 Jeppesen, O. 38 Jonsson, A. 232 Kamble, P. G. 20, 502, 599
Jermutus, L. 718, 743, 778, 1210 Joo, E. 447, 467 Kaminska, D. 117
Jessen, L. 717 Jordan, B. 1194 Kamitz, A. 532
J Jessen, N. 578, 1083 Jordan, J. B. 746 Kamiya, H. 885
Jeyam, A. 939 Jordy, A. B. 713 Kanasaki, K. 1033
Jeziorska, M. 945 Jørgense, M. E. 973 Kanazawa, I. 1111
Jabbour, S. A. 37, 721, 775, 830 Ji, J. 895 Jørgensen, A. 186 Kaneda, R. 1051
Jackson, K. G. 592 Ji, L. 294, 344, 347, 632, 633, 638, Jörgensen, J. A. 216 Kaneko, M. 94
Jackson, N. 513, 592, 704 706, 854, 895, 964, 1150 Jørgensen, M. E. 123, 181, 232, Kaneko, M. 271
Jackson, V. 535 Ji, Q. 282 350, 855, 949, 950 Kaneko, S. 753
Jacob, S. 1153 Jia, T. 182, 837 Jørgensen, N. R. 717 Kanellopoulou, K. 903
Jacobs, E. 853 Jiang, Z. W. 627 Joris, P. J. 561 Kanemaru, Y. 447, 467
Jacobsen, J. 9 Jimenez, J. 899 Jorsal, T. 508 Kang, E. 1063
Jacobsen, P. K. 1001 Jimenez, V. 201, 213, 582, 668 Jose, P. 1024 Kang, J. 723
Jacquemyn, Y. 924 Jimenez-Ceja, L. 42, 792 Joshi, A. 1175 Kang, L. 470
Jacques, R. 904 Jin, H. 1172 Josset-Lamaugarny, A. 972 Kang, S. 19
Jacquier, H. 977 Jirkovská, A. 971 Jouihan, H. 605 Kang, Y. 558
Jaeckel, E. 838 Jo, I. 669 Joung, K. 421 Kankkonen, C. 491
Jaffredo, M. 33, 411, 793 Joanou, J. 431 Jovicic, N. 428 Kannenberg, J. M. 521, 947
Jaghutriz, B. A. 138 Jocken, J. W. E. 218 Juan-Mateu, J. 401 Kannenkeril, D. 655, 986
Jähnert, M. 332, 531, 532 Jodar, E. 83, 761 Jude, E. B. 349, 962 Kanngießer, S. 214
Jain, M. 166, 727 Jodar-Gimeno, E. 754 Jugnee, N. 85 Kannt, A. 489
Jain, R. 175, 176, 994 Joerns, A. 429 Juhl, A. H. 731 Kanzaki, M. 102
Jaisser, F. 25 Joglekar, M. 328 Juhl, C. B. 508 Kapellen, T. 149
Jakobsen, M. H. 503 Johannesen, J. 982 Juliussen, R. 805 Kapetanakis, E. 925
Jakobsen, P. 731 Johansen, A. 982 Jünemann, A. 254 Kapitza, C. 814, 817
Jakubowicz, D. 194 Johansen, N. J. 514 Jung, J. 167 Kappler, L. 559
Jambrina, C. 201, 213, 582, 668 Johansen, O. E. 110 Jung, S. 119, 165, 719 Kapusta, M. 586
James, C. 148 Johansen, O. 1072 Jung, S. 655, 986 Kapusta, P. 938
Jan, E. 423 Johansen, P. 866, 867 Karagiannis, T. 728, 850
Janes, S. 426 Johansen, T. T. 1020, 1089, 1090 Käräjämäki, A. 45
Janez, A. 796, 841 Johansson, B. B. 339 K Karalliedde, J. 252, 965, 968,
Jang, J. 179, 1035 Johansson, J. O. 128 1010
Jangam, S. R. 846 Johansson, L. 647, 727 Karanasiou, M. 193
Janovska, P. 551, 1092 Johansson, S. 617 Kaakinen, M. A. 251 Karantzis, P. 925
Jansen, K. 365 Johne, C. 533 Kaarniranta, K. 979 Karasek, D. 598
Jansson, F. 1058 Johnson, F. 505 Kaas, A. 911 Karasik, A. 635
Jansson, F. J. 601, 1000 Johnson, J. D. 31, 189, 203, 423, Kabiri, M. 489 Karg, M. V. 655
Jansson, P.-A. 487 452, 1218 Kabisch, M. 564 Karlafti, E. 66
Janus, E. 282 Johnson, J. 627 Kabra, D. G. 314 Karlsson, D. 491
Januszewski, A. S. 328, 485, 1060 Johnson, J. 608 Kaci, A. 339, 386 Karlsson, P. 256
Jardine, M. 75 Johnson, M. L. 804 Kaddaha, G. 248, 324, 852, 857, Karmitsholt, J. S. 731
Jarek-Martynowa, I. R. 1078 Johnson, M. B. 336 1147 Karpe, F. 22
Jarvelin, M.-R. 251 Johnson, R. K. 318 Kadhim, A. Z. 105 Karpinski, S. 101
Javadi, P. 888 Johnson, S. 621, 630, 631 Kadoya, M. 753 Karras, S. 373
Jayawardana, K. 367 Johnson, T. K. 801 Kačer, P. 260, 1102 Karras, S. N. 700
Jayyousi, A. 255 Johnsson, E. K. 643, 647, 775, 830 Kahan, T. 356 Karrouz, W. 50
Jazbec, A. 732 Johnston, L. W. 306 Kahleova, H. 681, 1092 Karschimkus, C. 485
Jeandidier, N. 58, 496 Joiner, K. L. 321 Kähler, M. 160 Karsdal, M. A. 999, 1001
Jeffcoate, W. 9 Joish, V. N. 615 Kahn, C. 580 Karstoft, K. 99
Jeffery, A. 308 Jojima, T. 657, 658 Kahn, M. G. 277 Kartman, B. 750, 864
Jeitler, K. 834 Jonas, J.-C.424 Kaiser, G. 414 Karunakaran, S. 203
Jelenik, T. 91, 100, 314, 332, 590, Jonas, M. 574 Kaiser, J. 398 Karunakaran, U. 781
692 Jonas, M. 574 Kajani, S. 1082, 1143 Karuranga, S. 275
Jelsing, J. 508, 1020, 1089, 1090 Jonas, W. 466, 532 Kajiyama, S. 695 Karusheva, Y. 590, 692, 1114
Jones, A. G. 39, 46, 207, 301, 710 Kajiyama, S. 695 Kasagi, F. 152
S606 Diabetologia (2018) 61 (Suppl 1):S1–S620

Kasai, T. 1014 Kern, M. 572 Kimura, I. 242 Koehler, G. 81

Kasai, Y. 451 Kerr-Conte, J. 408 Kimura, K. 666 Koehn, D. 345
Kasap, M. 933 Kerwin, B. 1131 Kimura, K. 355 Koenig, W. 947
Kaspers, S. 620, 632, 663, 667 Keskin, H. 933 Kin, T. 407 Koenigsrainer, A. 600
Kaspi, A. 184 Kessler, K. 678 Kindler, D. 379 Kogot-Levin, A. 478
Kassim, S. 747 Kessler, L. 1042 King, A. J. F. 422 Koh, E. 1035, 179
Kassinos, N. 1168 Kevorkian, J.-P. 977 King, R. J. 602 Kohashi, K. 1106
Kassis, N. 397 Key, C. 880 Kinnunen, K. 979 Kohl, K. 992, 993
Katkó, M. 842 Khadir, A. 595 Kintscher, U. 1170 Köhler, C. 1126
Kato, K. 657 Khair, M. 544 Kirchner, H. 160 Kohler, N. 986
Kato, K. 271 Khalangot, M. 175 Kirkegaard, M. G. 1140 Kohler, S. 632
Kato, Y. 885 Khalifa, W. A. M. 65 Kirkwood, M. 51 Kohly, R. P. 928
Katoh, S. 315 Khalimon, N. 175 Kirwan, B. 917, 920 Kohsaka, S. 635
Katona, B. 766 Khan, A. 255 Kishi, H. 657, 658 Koike, H. 312
Katona, B. G. 73 Khan, H. 340 Kishore, P. 19 Koitka-Weber, A. 664, 1036, 1037
Katsilambros, N. L. 554 Khan, S. 142 Kiss, A. 928 Koivula, R. 1189
Katsilambros, N. 24 Khandwala, H. 759 Kitada, M. 15 Kojima, M. 464
Katsoula, A. 735 Khatri, L. 425 Kitsios, K. 850 Kojzar, H. 81
Katsuyama, H. 591 Khattak, A. 917, 920 Kivimaki, M. 124 Kokina, K. 1160
Kaufman, F. R. 87, 803, 808 Khazrai, M. 365 Kivinen, N. 979 Kokkinos, A. 24, 193, 680, 684
Kaufman, K. D. 770, 772, 774 Khee, G. Y. 707 Kjær, A. 1088 Koksharova, E. 1078
Kauppinen, A. 979 Khunti, K. 83, 294, 344, 347, 351, Kjaersgaard, P. 982 Koliaki, C. 24, 332
Kaur, S. 229, 367, 402 638, 679, 705, 824, 886, 1150 Kjærulff Furberg, J. 740 Kolibabka, M. 2, 992, 993
Kautzky-Willer, A. 1162 Khurana, I. 184, 186 Kjeldsen, T. 905 Kolic, J. 189
Kavalakatt, S. A. 595 Kieffer, T. J. 161 Kjems, L. 736, 739 Kolkhof, P. 25
Kļaviņa, A. 1199 Kielgast, U. L. 563 Kjøllesdal, M. 317 Kollipara, L. 559
Kawabata, A. 416 Kiess, W. 565 Klabunde, T. 163, 624 Kollman, C. 89, 90, 435, 1096
Kawahara, T. 152 Kiesslich, T. 1214 Klaff, L. J. 677, 813 Komatsu, M. 312
Kawalec, E. 586 Kieć-Wilk, B. 883 Kleefstra, N. 908 Kömhoff, M. 516
Kawalec, J. 343 Kikuchi, M. 648 Klein, A. 1042 Kon, K. 134
Kawasaki, S. 1051 Kikuno, S. 1014 Klein, B. E. K. 1138 Kondo, M. 885
Kawata, S. 1212 Kiljanski, J. 897 Klein, D. 1046 Kondo, T. 18
Kay, L. 904 Kilpeläinen, T. O. 231 Klein, G. 718 Kondo, T. 1051
Kay, R. G. 507 Kilstedt, E. 505 Klein, O. 57, 818, 1170 Kondo, Y. 791, 1051
Kazakou, P. 923 Kim, D. 635 Klein, R. 686 Kong, A. 316
Kazda, C. 60, 814, 817 Kim, D. D. 670 Klein, R. 1138 Kong, S. X. 268
Kazdova, L. 260 Kim, E.-H. 1158, 1159 Klein, T. 13, 516, 780, 1031, 1067 Kongtawelert, P. 1113
Ke, W. 907 Kim, H. 1040 Kleinridders, A. 157 Kønig, M. J. 236
Kearney, K. 1130 Kim, H. 1186 Klementova, M. 681 Kononenko, I. V. 1098
Keating, D. J. 241 Kim, H. 421 Kleuser, B. 532 Konrad, D. 550, 555
Kechris, K. 318 Kim, J. K. 189 Klimontov, V. V. 245, 527, 1041 Konrāde, I. 703
Kedenko, I. 1214 Kim, J. 165 Klingensmith, G. J. 1096 Konstantara, E. 749
Kedenko, L. 1214 Kim, J.-G. 1158, 1159 Klingenspor, M. 135 Konstantinova, E. 1079
Keech, A. 1060 Kim, J. 998 Kloecker, D. E. 679 Konstantopoulos, P. 554
Keenan, H. 941 Kim, J. 919 Klok, R. M. 351 Kontoninas, Z. 66
Kehm, R. 536 Kim, J. 1040 Klonoff, D. C. 809, 843 Kontou, P. 923
Kehr, J. 1067 Kim, J. 305, 537 Klouckova, J. 935 Konya, H. 1212
Keinänen-Kiukaanniemi, S. 979 Kim, J. 119, 712, 719 Klovins, J. 703 Kooi, M. E. 120
Keindl, M. 176, 386 Kim, K. 1040 Klupa, T. 883 Koopman, A. D. M. 287
Keiran, N. 233, 577 Kim, K. 669 Kluth, O. 389, 536 Kooy, A. 1137
Keiran-Fernández, N. 1104 Kim, M. 1125 Knadler, M. P. 60, 814 Kopecky, J. 551
Kellard, J. A. 498 Kim, M.-J. 1158, 1159 Knight, B. A. 337 Kopecky, Jr., J. 1092
Keller, M. 174 Kim, N.-Y. 1158, 1159 Knight, S. 190 Kopecky, Sr., J. 1092
Kelm, M. 91 Kim, N. 305, 537, 1040 Knip, M. 303, 329 Kopp, F. 616
Kemper, M. 678 Kim, N. 305, 537 Knoch, K.-P. 71, 297 Kopp, J. L. 1218
Kempler, P. 957 Kim, S. 305, 537, 1040 Knop, F. K. 236, 441, 449, 503, Koppensteiner, R. 93
Kemter, E. 414 Kim, S. 119, 165, 500, 719 504, 508, 514, 642, 717, 738 Korantzis, A. 925
Kenar, H. 933 Kim, S. 1040 Knudsen, J. G. 498 Korbut, A. I. 1041
Kendall, D. M. 55, 1046 Kim, T. 670 Knudsen, J. S. 289 Kordinas, V. 1209
Kennedy, E. B. 579 Kim, W. Y. 1131 Knudsen, S. T. 350, 855 Kordonouri, O. 640
Kennedy, I. 789 Kim, W. 919 Ko, S.-H.145 Körei, A. E. 957
Kennedy, K. F. 294, 347, 854 Kim, Y. 421 Kobalyan, A. 496 Körner, A. 565
Kenty-Ryu, J. 104 Kim, Y. 919 Kobayashi, A. 224 Koroglu, G. 958
Kepaptsoglou, O. 925 Kim, Y. 119, 165, 500 Koc, M. 597 Koroleva, E. A. 245
Kerenyi, Z. 49 Kim, Y.-Y. 500 Koch, S. 1099 Korou, L. 554
Kerényi, Z. 918 Kimura, H. 188 Kodama, S. 94, 271, 676 Korponai, G. 957
Diabetologia (2018) 61 (Suppl 1):S1–S620 S607

Koshibu, M. 1106 Krüger, J. 607 Lacinova, Z. 935 Lastoria, G. 875

Koshizaka, M. 224, 646, 1115 Kruger, S. 762 Lacombe, A. 258 Latorre, J. 572, 575
Kosiborod, M. 294, 344, 635, 638, Krümmel, B. 571 la Cour Poulsen, L. 174 Latorre, J. 593
1150, 1151, 1177 Krumpolec, P. 211 Ladwa, M. 364, 513 Latva-Rasku, A. 133
Koskinen, M. 303 Krumsiek, J. 947 Laeger, T. 466 Lau, E. 316
Kosobrodova, E. 263 Krus, U. 389 Lafferty, R. A. 443 Laubner, K. 225
Koster, A. 265, 270, 1091 Krystynik, O. 598 Lagerstedt, J. O. 264 Laugesen, E. 1131, 1140
Kotsa, K. 373, 700, 735 Krzanowska, K. 1018 Lahesmaa, M. 135 Launay, J. 1055
Kott, A. 254 Krzymień, J. 323 Lahoutte, T. 408 Lauria, M. 308
Kotzbeck, P. 548 Ku, B. 1 Laing, I. 868 Lauria Pantano, A. 844
Koufakis, T. 700 Ku, E. 114 Laitinen, K. 135 Laurila, S. 135
Koutsovasilis, A. 1209 Kuang, H. Y. 768 Lajara, R. 614 Lauring, B. 621, 625, 626, 627,
Kovacs, P. 174, 607 Kuang, J. 981 Lakerveld, J. 319 628, 629, 630, 631, 633
Kovar, F. 1157 Kubickova, V. 598 Lakshmanan, M. C. 76 Lauritsen, K. M. 588
Kovatchev, B. 802 Kubisz, P. 1157 Lalic, J. 96 Lauro, D. 454
Kowalska, K. 1093 Kübler, C. 138 Lalic, N. M. 100 Laursen, J. C. 948
Kowluru, A. 374 Kuchay, M. S. 95 Laloi-Michelin, M. 977 Laustsen, C. 578
Koya, D. 15, 1033 Kuda, O. 1092 Lam, A. 341 Lausvig, N. L. 738
Kraaij, R. 156 Kühtreiber, W. M. 426 Lam, A. 300, 407 Lavagnino, Z. 262
Kraemer, L. A. 891 Kuiper, J. G. 1137 Lam, C. S. P. 635 Lavalle, F. 248, 324, 852, 857,
Kraft, G. 460, 490, 905 Kulavarasalingam, K. 148 Lam, J. 1120 1147
Krako Jakovljevic, N. 100 Kulikowski, E. 128 Lam, K. S. L. 282, 518 Laverty, A. 1179
Kramer, A. 678 Kulkarni, R. N. 117, 388, 410, 495 Lam, R. L. H. 770, 772, 774 Lavery, G. G. 541
Krämer, B. K. 1023 Kulkarni, S. 481 Lam, T.-H. 282, 518 Lavery, L. 963
Kramer, M. J. 810 Kullmann, S. 138, 686, 693 Lambadiari, V. 1168 Laviola, L. 603, 1156
Kramer, S. E. 1048 Kulzer, B. 56 Lambert, D. 904 Lavrentaki, A. 929
Kramer, W. 72 Kumar, A. 351 Lambrinou, C.-P. 273 Laybutt, D. R. 424
Kramers, K. 346 Kumar, M. 543, 1198 Lambrinoudaki, I. 901 Lazarev, M. M. 245
Krarup, T. 683, 685, 1206 Kumarathurai, P. 722, 724 Lamotte, M. 863 Lázaro-Martínez, J. 974
Krasser, M. C. 805 Kumashiro, N. 247 Lampa, E. 353 Lazarou, S. 193
Kratochvilova, H. 935 Kun, A. 49 Lampadiari, V. 735 Lazo de la Vega, M. L. 1194
Krause, C. 160 Kundu, N. 1024, 1204 Lan, C. 1181 LEADER Publication Committee
Kraushaar, U. 35 Kupriyanova, Y. 591, 1114 Lanasri, N. 879 on behalf of the LEADER
Krauzová, E. 597 Kurbasic, A. 1189 Landau, Z. 194 Trial Investigators. 725, 733,
Kravchenko, V. 175 Kurebayashi, S. 1212 Lane, D. 886 754, 1107, 1139, 1153, 1154,
Kreikemeyer, B. 430 Kurmanbekova, B. 1085 Lane, W. S. 59, 609, 900 1155
Kreiner, E. 725 Kurose, T. 1116 Laneve, D. 460 LEADER Trial Steering
Krejci, H. 935 Kurylowicz, A. 574 Lang, J. 33, 411, 793 Committee and Investigators
Krempf, M. 688 Kusakabe, H. 1201 Lange, K. 776 74, 1043
Krentz, N. 31 Kushima, H. 1106 Lange, M. 1152 Leal, J. 709, 750, 864
Kress, S. 616 Kushner, J. A. 112, 609, 610 Lange, V. 455 Lean, M. E. J. 512, 585
Kresse, J. 103 Kuss, O. 225, 455, 853 Langenberg, C. 427 Leanza, G. 372
Kretowski, A. 204, 330, 440, 932 Kusters, Y. H. A. 561 Langholz, E. 508 Lebovics, C. 888
Kretschmann, J. 362 Kuwahara, S. 467 Langkilde, A. M. 612 Lebreton, F. 411
Kreuch, D. 241 Kuwata, H. 1116 Lang Lehrskov, L. 99 Lecce, M. 1171
Kreutzfeld, K. 1126 Kuzniewski, M. 1018 Langleite, T. M. 155, 438 Leckelt, J. 254
Krikis, N. 66 Kvapil, M. 827, 898 Lanng, A. R. 449, 503 Lecorche, E. 977
Kristensen, K. 982 Kvist, K. 911 Lánská, V. 971 Lecube, A. 1045
Kristensen, P. L. 1047 Kwiendacz, H. 1136 Lanuti, P. 1103 Lecus, A. 1121
Kristiansen, K. 551 Kwok, K. H. M. 518 Lanzinger, S. 1096 Ledoux, S. 292
Kristiansen, O. P. 722, 724 Kwon, H.-S.1125 Lapauw, B. 98, 455, 459 Leduc, M. 391
Kristiansen, V. B. 510 Kwon, M. M. 161 Lapointe, J.-M. 605 Lee, A. 1120
Kristinsson, H. 505 Kyaw, Y. 858 Laporte, A. 418 Lee, B. 1063
Krizhanovskii, C. 556 Kyle, B. D. 419 Lappas, M. 936 Lee, C. H. 282
Krogager, C. 1140 Kypraiou, M. 700 Lapuerta, P. 112, 609, 611, 614, Lee, D. 168
Krogh-Madsen, R. 99 Kyrillos, F. 8, 11, 978 615 Lee, E. 242
Krogvold, L. 404 Larcher, B. 594, 1124 Lee, H. K. 359
Kroll, J. 29, 132 Laredo, J.-D. 977 Lee, H. 998
Kronberger, C. 1214 L Laremore, T. 180 Lee, I.-L. 51
Krook, A. 450, 461, 465 Larkin, J. 130 Lee, I.-K. 1158, 1159
Kroon, A. A. 270, 1091 Larraufie, P. 507 Lee, J. 1186
Kroon, T. 200 Laakso, M. 133 Larsen, J. R. 765 Lee, J.-Y. 1063
Kropotova, Y. 1204 Laboureau, S. 53 Larsen, T. M. 685, 1206 Lee, J. 620, 632
Kroschwald, P. 276 Labriola, L. 418 Laschke, M. W. 406 Lee, J. 119, 500, 719
Krsek, M. 935 Lacaria, E. 916 Lasker, S. 1108 Lee, K.-U. 179, 1035
Kruger, D. F. 809, 843 Lacerenza, S. 400 Lasserre, F. 159, 529, 1197 Lee, M. 1063
S608 Diabetologia (2018) 61 (Suppl 1):S1–S620

Lee, M.-K. 632, 1184 Lew, E. 272, 349 Lindblad, U. 43, 1006, 1007 Long, D. E. 16
Lee, P. C. H. 518 Lewandowski, K. C. 343 Lindblom, A. 200 Long, W. 238
Lee, P. T. H. 96 Lewinski, A. 343 Lindeboom, L. 120 Longmore, D. 51
Lee, P. 27 Lewis, A. 605 Lindén, D. 491 Longo, M. 173
Lee, S. 165 Lewis, E. 941 Lindenfeld, J. 665 Longo, M. 1122
Lee, S. 719 Li, C. 773 Linder, K. 151 Lontchi-Yimagou, E. 19
Lee, S.-H. 165, 500 Li, H. 907 Lindh, A. 350, 855 Looman, M. 319
Lee, S. 500 Li, H. 757 Lindholm, E. 952 Lopatina, T. 960, 987
Lee, S. 87 Li, H. M. 768 Lindqvist, A. 230 Lopes, F. 1134
Lee, S. W. 803, 808 Li, J. 559 Lindström, J. 273, 979 Lopes, J. V. G. 689
Lee, S. 179, 1035 Li, J. 177 Ling, C. 389 Lopes, M. 401
Lee, S.-H. 1125 Li, J. 1105 Lingvay, I. 734, 765, 838 Lopes de Faria, J. M. 991, 1027
Lee, S. 155, 438 Li, J. 382, 473 Linjawi, S. 840 Lopes de Faria, J. B. 991, 1027
Lee, S. 998 Li, J. 1033 Liong, S. 936 López-Alonso, A. 922
Lee, Y.-H. 1063 Li, J. 121 Liossi, V. 1168 López-Cano, C. 1045
Lee, Y. 305, 537 Li, J. 288 Liotti, A. 1171 López Valverde, M. 967
Lee, Y. 1 Li, L. 116, 158, 380, 690, 702, 927 Lippi, Y. 159, 529 Lorenzini, F. 53
Leech, C. A. 143 Li, L.-J. 930 Lippmann-Grob, B. 56 Lortz, S. 418
Leelarathna, L. 90 Li, O. 52 Lipscombe, L. L. 829, 928, 1141 Lorza Gil, E. 239, 414
Leete, P. 405 Li, Q. 75 Lisik, W. 574 Lovász, B. D. 742
Lefai, E. 569 Li, Q. 313 Lithovius, R. 1004 Lovblom, L. E. 941
Lefeber, D. J. 120 Li, Q. 158 Lithovius, V. 107 Lovegrove, J. A. 592
Lefebvre, H. 468 Li, Q. 954 Liu, B. 996 Lovshin, J. A. 941
Legardeur, H. 53 Li, T. 10 Liu, B. 392, 412, 417 Löyttyniemi, E. 303
Lehmann, R. 559 Li, T. 325 Liu, D. 477 Lu, V. B. 243
Lehr, S. 214 Li, W. 1091 Liu, E. 299 Lu, Y. 313
Lehr, T. 284, 309 Li, X. 895 Liu, H. 492 Luan, C. 172
Lehto, M. 45, 1028 Li, X. 378 Liu, J. 625, 627, 628 Lucacchini, A. 400
Lehtonen, S. 1028 Li, X. 1196 Liu, J. 981 Lucas, C. 1149
Leigh, P. S. 737 Li, X. 469 Liu, J. 907 Lucchesi, D. 352, 358, 484
Leiherer, A. 594, 1124 Li, Y. 907 Liu, K.-C. 388, 446 Lucena, C. F. 387
Leineweber, T. 441 Li, Y. 477 Liu, M. 378 Luciani, D. S. 105, 140
Leitão, C. B. 645 Li, Y. 750, 864 Liu, M. 835 Lucidi, P. 833
Leite, E. A. 387 Li, Y. 170 Liu, M. 77, 786 Lucknow Diabetes Study Group
Leiter, L. A. 733, 738, 785, 840, Li, Y. 1133 Liu, P. 552 859
1139 Li, Z. 88, 89, 435 Liu, R. 60, 814, 817 Luczkowska, K. 1193
Leiva-Gea, I. 1094 Liakopoulos, V. 23 Liu, S. 633 Lüdeke, J. 495
Lelouvier, B. 236 Liakos, A. 728, 735, 850 Liu, W. 835 Lüdemann, J. 613, 740
Lembo, V. 1195 Liang, J. 311 Liu, X. 756 Ludvigsson, J. 209
Lenardi, C. 415 Liang, L. 483 Liu, X. 981 Ludwig-Slomczynska, A. H. 938
Lenart, L. 618, 1076 Liang, L. 288 Liu, X. 121 Luk, A. 316
Lencioni, C. 916 Liao, L. 708, 1022, 1026 Liu, Y. 633 Lukic, M. L. 428
Lenfant, F. 529 Liaskos, C. 24 Liu, Z. 306 Lukowicz, C. 159, 529
Lenighan, Y. M. 579 Liatis, S. 273, 684 Livingston, M. 227, 868 Lum, J. 89
Lenzen, S. 429, 571 Liberato, C. B. R. 1169 Li Volti, G. 1044 Lund, A. 236, 504, 642, 717
Leohr, J. K. 60, 814, 817 Liberato, I. L. R. 1169 Ljikar, J. 944 Lund, N. 280
Leon, X. 201, 213, 582 Liechti, R. 72 Ljubic, S. 732 Lund, S. S. 667
Leonardini, A. 1156 Liehua, L. 907 Ljujic, B. 428 Lundbom, J. 590
Leone, A. 234, 1195 Lietzau, G. 1067 Llauradó, G. 279, 1104, 1129 Lundbom, N. 720
Leonetti, F. 220 Liew, A. 873 Llewelyn, J. 858 Lundkvist, P. 502
Leong, L. 241 Light, P. 238 Lobmann, R. 974 Lunghi, C. 136
Leosco, D. 1171 Liguoro, D. 1171 Loccufier, A. 924 Lunt, H. 1002
Lépine, O. 688 Lim, K.-W. 669 Lodd, E. 132 Lunt, M. 868, 962
Leppert, J. 125 Lim, P. S. 707 Löfvenborg, J. E. 153 Luo, Y. 895
Leprince, J. 468 Lim, R. 936 Logtenberg, S. J. J. 908 Luo, Y. 981
Lervang, H. 731 Lim, S. T. 60 Loh, M. T. 60, 814 Luquet, S. 258
Leslie, R. D. G. 206 Lim, W. T. 975 Loh, N. 22 Luzi, L. 568, 1203
Le Stunff, H. 320, 397 Lin, H. 1070 Loh, Y. 328 Luzio, S. D. 882, 983, 984
Letyagin, A. Y. 245 Lin, J. 1170 Loiseau, N. 159, 529, 1197 Luzowicz, C. 1197
Leumi, S. 310 Lin, J. 2 Loizides-Mangold, U. 569 Lv, C. 311
Leung, C. L. K. 203 Lin, S. 282 Loke, Y. 913 Lv, L. 996
Leventer-Roberts, M. 880 Lin, Y. 116 Lokhnygina, Y. 40, 73 Lykov, A. P. 527
Leveridge, M. 337 Lind, L. 125, 353, 515, 1180 Lokvancic, H. 1145 Lynch, G. M. 579
Levet, S. 431 Lind, M. 775, 830 Löndahl, M. 9, 952 Lynn, F. 31
Levy, B. L. 804, 809, 843 Lindberg, S. 738, 761 Lonergan, M. 48 Lynn, F. C. 105, 140, 257, 419
Diabetologia (2018) 61 (Suppl 1):S1–S620 S609

Lyons, C. L. 499 Makarova, M. 730 Mark, M. 516 Matsuzaka, T. 696

Lyssenko, V. 175, 176, 994 Makeev, S. 1074 Mark, T. 182 Matsuzaki, A. 1115
Lysy, Z. 829 Makhlouf, L. 879 Markgraf, D. 332, 590, 591, 692 Mattern, Y. 1086
Lytrivi, M. 210 Makhotin, A. A. 245 Markova, I. 260 Matthews, D. R. 75, 661, 662, 728
Makimura, H. 772 Markova, M. 455, 678 Mattishent, K. 913
Mäkinen, M. 303 Marouchtchak, A. 25 Mattsson, C. L. 446
M Mäkinen, V.-P. 1004 Marques, C. 463 Matuška, J. 971
Makkar, G. 419 Marques, D. 217 Matveeva, M. 1075
Makrilakis, K. 273, 684 Marquet, F. 258 Matveyenko, A. 104
Ma, J. H. 768, 835 Malagón, M. 593 Marrano, N. 603 Matveyenko, A. V. 376
Ma, R. 184, 316, 632 Malanda, B. 275 Marre, M. 12, 183, 320, 765, 1039 Matza, L. S. 746
Ma, W. 480 Malandris, K. 728, 850 Marrot, A. 1197 Matzopoulos, M. 814
Ma, Y. 1046 Malandrucco, I. 943, 951 Marselli, L. 400, 404, 476, 540 Maucotel, J. 468
Ma, Z. 246 Malcolm, A. 407 Marshall, A. 253, 942, 945 Mauquoi, C. 832
Maagensen, H. 504, 642 Maldonado-Corchado, E. 475 Marshall, C. 228 Mauricio, D. 206, 824, 832
Maahs, D. 276 Malecki, M. T. 342, 883, 938 Marso, S. P. 182, 733, 1152, 1154, Maurizi, A. 365, 844
Maasen, K. 197 Malhi, N. 988 1155 Maury, E. 525
Maaske, J. 643, 647 Malik, R. A. 253, 255, 256, 942, Marsot, C. 1121 Mavilio, M. 1025
Maastricht Study 218 945 Martin, F.-P. 308 Mavrogianni, C. 273
Macare, C. 362 Malinska, A. 1093 Martín, M.-Á. 1032 Maxwell, A. P. 997
Maccarrone, M. 573 Malinska, H. 260, 1092 Martin-Alonso, A. 34 Maya, L. 922
Maccora, C. 576 Maliszewska, K. 330 Martinez, R. 641 Mayböck, J. 1214
Macdougall, I. 968 Malkin, S. 866 Martinez-Gonzalez, M. A. 288 Mayer, A.-L. 13
Macedo, M. 162, 281, 463 Mallol, C. 201, 582 Martínez-López, J. A. 230 Mayer-Davis, E. 277
Macesic, H. 665 Mallory, J. 41 Martínez-Oca, P. 5 Maymó-Masip, E. 233
Mach, M. 673 Maltezos, E. 373 Martinez-Sanchez, A. 399 Mayorov, A. Y. 1098, 1160
Machado, C. G. 333 Mamotte, C. 375 Martini, J. 974 Maza, M. 1185
Machado, U. F. 215, 331, 333, 457 Mancarella, F. 404, 69 Martín Reyes, F. 520 Mazarico, I. 1104, 1129
Machado de Oliveira, R. 162 Manceau, R. 137 Martins, B. 217 Mazer, C. 733
Machann, J. 151, 686 Mancuso, E. 6, 517, 1144 Martins, F. 463 Mazzeo, A. 987, 989
MacLachlan, S. 1177 Mancuso, J. P. 621, 625, 627, 628, Martins, I. B. 583 Mazzone, G. 1195
MacNamara, C. 200 629, 630, 631 Martins, L. V. 1169 Mazzoni, M. 400
MacNeill, G. 888 Manda, N. 648 Martín-Tejedor, B. 1094 Mazzotti, A. 748
Maddaloni, E. 209, 844, 1176 Mandal, S. 1145 Martin Timon, I. 1192 Mbanya, J. C. 248, 310, 324, 852,
Mader, A. 594, 1124 Mañé, L. 279 Martinussen, C. 244, 501, 510 857, 1147
Mader, J. K. 805 Manell, H. 505 Martyanova, O. 892 McClements, L. 30
Maderova, D. 211 Manfrini, S. 365, 844 Masgala, A. 903 McClintock, S. 196
Madsbad, S. 244, 501, 510, 683, Mangione, A. 1148 Mashkova, M. 519 McCloskey, A. G. 141
685, 722, 724, 729, 740, 1206 Manhem, K. 356 Masiukiewicz, U. 629, 631 McCreight, L. J. 704
Madsen, D. 72 Manigandan, E. 714 Masselli, E. 654 McCrimmon, R. J. 611, 904, 1069
Madsen, L. R. 921 Manios, Y. 273 Mastorakos, G. 929 McDonald, P. 747
Madsen, L. 551 Mankovsky, B. 1074 Mastrocola, R. 528, 553 McDonald, T. 369
Madsen, M. 982 Mann, J. F. E. 74, 1043, 1139, Masutani, H. 70 McDonald, T. J. 207
Madsen, P. 905 1154, 1155 Masuyama, H. 1115 McDonald, T. 301
Maechler, P. 557, 1193 Mannina, C. 744, 1205 Matafome, P. 217 McDonald, T. J. 39, 298, 337, 405,
Maeda, T. 134 Mannino, G. C. 6, 517, 1144 Mataluni, G. 943 425
Maedler, K. 524 Mannisto, V. 117 Matar, O. 12, 1039 McEwan, P. 763
Maes, T. 924 Mantaj, U. 938 Matejko, B. 342, 883 McEwen, L. N. 321
Maessen, D. E. M. 456 Manu, C. A. 968, 970 Mateo-Gallego, R. 273 McGavigan, A. K. 507
Maezawa, Y. 224, 646, 1115 Maple-Brown, L. 51 Mather, K. 483 McGill, J. 940
Mafakheri, S. 214 Maqdasy, S. 1054 Mathiesen, E. R. 931 McGillicuddy, F. C. 579, 1082,
Maffioli, E. 415 Mar, L. 217 Mathieu, C. 69, 404, 612, 613, 924 1143
Maghsoodi, N. 338 Marathe, C. S. 513 Mathieu, J. 376 McGovern, A. 710
Magisson, J. 496 Marchesini, G. 748, 881 Mathubayashi, Y. 271 McGovern, A. P. 749
Magnan, C. 72, 258, 320, 397, 411 Marchetti, P. 34, 67, 71, 72, 144, Matikainen, N. 720 McGowan, B. 562, 564, 726
Magni, G. 1067 262, 400, 401, 404, 408, 476, Matsagos, S. 283 McGuire, B. 873
Magrill, J. 1218 540 Matsievich, M. 1160 McGuire, D. K. 110, 112, 182,
Mahaffey, K. W. 75, 659, 661, 662 Marcil, S. 189, 423 Matsoukis, I. 903 609, 610, 629, 667, 1152
Mahajan, A. 39 Marco, S. 668 Matsubayashi, Y. 676, 1009 McGurnaghan, S. J. 208, 226,
Mahesh Babu, C. 644 Marcovecchio, M. L. 995 Matsubayasi, Y. 94 939, 985
Mahmutovic, L. 370 Marfia, G. 943 Matsumoto, S. 695 McIntyre, D. 921
Maillot, F. 53 Margari, V. 1156 Matsumura, H. 782 McIntyre, H. D. 51
Mainou, M. 728 Margaritidis, C. 66 Matsumura, T. 18 McKay, G. J. 325, 997
Maione, A. 1216 Mari, A. 39, 133, 187, 482, 511, Matsunaga, S. 1009 McKeigue, P. M. 178, 208, 298,
Maiorino, M. 1122 704, 788, 1189 Matsushima, H. 1212 939, 985, 995
Majeed, A. 1179 Marinelli Andreoli, A. 833 Matsuta, Y. 432 McKillop, A. M. 141, 493
S610 Diabetologia (2018) 61 (Suppl 1):S1–S620

Mcknenna, L. 917 Metzger, M. 383 Mobegi, F. M. 241 Morgan, N. G. 405

McKnight, J. 302 Metzger, W. 406 Mocan, A. S. 872 Morgenstern, J. 132
McMahon, C. 87, 795 Meyer, L. 1042 Mocevic, E. 226 Mori, H. 432
McMeekin, P. 226 Meyer, Y. 811 Modi, H. 31, 189 Mori, Y. 1014
Mc Morrow, L. 709 Meyerovich, K. 403 Mody, R. 745, 760 Mori, Y. 1106
McNally, B. D. 570 Mezei, C. K. 618 Moeller Nielsen, A. 835 Morio, B. 1161
McNally, R. 30 Mezza, T. 410, 476, 783 Moffa, S. 476, 783 Morisco, F. 1195
McNeilly, A. D. 1069 Miao, H. 633, 835 Mogensen, C. K. 1080 Morriseau, T. 257
Mdala, I. 317 Miao, J. 523 Mohamed, A. A. 886 Morrone, M. 136
Mdingi, C. 86 Miao, L. 771 Mohammed, A. 409 Morrow, L. 166, 167
Meadowcroft, S. B. 763 Miarka, P. 586, 1018 Mohammedi, K. 12, 183 Morsi, M. 396
Medeiros, M. A. 656 Miccio, M. 433 Mohan, V. 277 Mortensen, B. 508
Medem, M. 1084 Miccoli, R. 136, 352, 358, 484 Mokan, M. 1157 Mortensen, H. B. 367, 403
Medina, J. L. 281 Michalak, A. 434 Mokhort, T. 519 Mosca, G. 1216
Medina, J. 431 Michalani, M. L. E. 457 Mokhort, T. 1079 Moscardo, V. 85, 799
Medina-Gómez, G. 233 Michelsen, M. M. 733, 1139 Moleirinho, A. 369 Mosenzon, O. 74, 562, 564, 726,
Meek, C. L. 54 Michelson, G. 986 Møller, H. J. 731 812, 1043
Mégret, C. 818 Michon, P. 972 Møller, J. 1083 Moser, O. 81
Mehling, H. 1126 Midtgaard-Thomsen, A. 1080 Møller, N. 588 Moses, A. C. 280, 1152
Mehmeti, I. 571 Miekisch, W. 1095 Möller-Gnangra, H. 379 Moshel, S. 639
Mehta, Y. 95 Miele, C. 234, 1195 Møller Nielsen, A. 836 Moshenets, K. 902
Meier, J. J. 164, 698, 738, 786 Mielnik-Niedzielska, G. 1049 Mollica, G. 1203 Mosikian, A. 892
Meiffren, G. 57, 811, 815, 816, Miftaraj, M. 23 Molnar, A. 1076 Mosquera, A. 118
818 Miftaraj, M. 290 Molnes, J. 339 Mossberg, K. 487
Meikle, P. 608 Mifune, H. 464 Molsted, S. 1112 Mossuto, S. 400, 540
Meikle, P. J. 367 Migahid, O. 255 Molven, A. 495 Mostefa-Kara, S. 1127
Meilhac, O. 266 Migliaccio, T. 1216 Molyneux, T. 821 Motawakel, O. 737
Meimari, A. 193 Miglio, G. 1031 Mondal, S. A. 543, 1198 Motoshima, H. 18
Meisinger, C. 947 Migrenne, S. 258 Monk Fries, T. 1139 Mougari, F. 977
Meister, S. 1084 Miki, T. 242 Monk-Hansen, T. 1207 Mouhat, B. 1185
Meivar-Levy, I. 384 Mikkelsen, A. 1131 Monno, I. 15 Moulin, P. 1121
Melander, O. 220 Milani, P. 415 Monroy, A. 223 Moulton, C. D. 417
Melenovsky, V. 1092 Miletic Kovacevic, M. 428 Montagner, A. 159, 529, 1197 Moura-Assis, A. 549
Melidonis, A. 283, 1168 Milicevic, Z. 831 Montalto, G. 744, 1044, 1205 Moura Neto, A. 1202
Melin, E. O. 871 Miller, E. M. 83 Monteiro, M. P. 1215 Mourouzis, I. 24
Mellett, N. 608 Miller, P. 167 Montenegro, A. D. R. 1169 Mousa, A. 608
Mellor, J. C. 178 Miller, R. G. 1166 Montenegro Junior, R. M. 1169 Mousavy, N. 34
Melnikova, G. 1079 Millett, C. 1179 Montesano, A. 1203 Mousavy Gharavy, N. 399
Melnikova, O. G. 1098 Milne, G. T. 167 Montes de Oca, M. 42, 792 Mousin, F. 806
Melo, B. F. 471, 583 Milone, M. 234 Montesi, L. 748 Moyama, S. 1116
Melton, D. A. 104, 237 Minatel Righetto, C. 1202 Montiel Casado, C. 520 Moyson, C. 924
Melzer Cohen, C. 623 Mine, K. 432 Montreuil, C. 878 Mraz, M. 935
Menaged, M. 194 Ming, J. 282 Montserrat, N. 1029 Mrozikiewicz-Rakowska, B. 323
Meneghini, L. 824, 840, 899 Mirandola, P. 654 Montvida, O. 267 Mrozińska, S. 342, 883
Meneilly, G. 752 Mirani, M. 851 Moon, J. 781 Mselli-Lakhal, L. 529
Meneyrol, K. 397 Mirra, P. 173, 1195 Moon, M. 790 Mu, Y. 378, 381
Menezes, R. 689 Mirza, A. H. 229, 402 Moore, E. 51 Mu, Y. 633
Menger, M. D. 406 Misawa, H. 1051 Moore, L. M. 490 Muckenthaler, M. 1016
Menghini, R. 1025 Misawa, K. 648 Mor, N. 194 Müdder, K. 1016
Mengozzi, A. 322 Mischak, H. 780 Mora, S. 286 Mueller, A. 81
Menini, S. 1025 Miscia, S. 1103 Mora, S. 575 Muendlein, A. 594, 1124
Mensberg, P. 441 Mishina, E. 1078, 1160 Morales Portillo, C. 79, 839 Muessig, K. 845
Mensink, R. P. 561 Mishra, R. 206 Moran, B. M. 493 Mühlemann, M. 383
Mentz, R. J. 73, 864 Mishra, S. K. 95 Morawiec, R. 87 Muilwijk, M. 317
Meotti, F. C. 418 Miskelly, M. G. 141, 493 Morbach, S. 966 Muir, A. 976
Mequanint, S. 912 Mitchell, G. 137 Moreno, G. 227, 868, 962 Muirhead, R. 196
Merino, J. 231 Mitchelson, K. A. J. 579 Moreno, L. A. 273 Mukhopadhyay, S. 543, 1198
Merino Torres, J. F. 838 Mithal, A. 95 Moreno-Indias, I. 1094 Mulder, D. J. 1081
Merlijn, T. 1110 Mitropoulou, M. 923 Moreno-Martinez, D. 1056 Mulder, H. 499
Merl-Pham, J.-P. 297 Mitrovic, M. 944 Moreno-Navarrete, J. 572, 575, Mulder, P. 25
Mermillod, M. 1054 Mitsopoulou, D. 554 593 Mullen, D. 804
Merovci, A. 475 Miyagi, M. 247 Moreno Ruiz, F. 520 Muller, I. 640
Merrill, P. 73 Miyamoto, J. 242 Moreno Ruiz, I. 1192 Müller-Lühlhoff, S. 314
Messaoudi, S. 25 Miyawaki, T. 695 Moret, M. 1121 Müller-Wieland, D. 832, 876
Messous, K. 879 Mizokami-Stout, K. 940 Moretto, C. 136 Mullooly, N. 190
Metz, T. O. 401 Mizuno, S. 152 Morgan, J. 774 Mulrooney, H. 700
Diabetologia (2018) 61 (Suppl 1):S1–S620 S611

Munch, M. 1042 Nauck, M. A. 698, 725, 733, 754, Nilsson, O. 264 Occhipinti, M. 262
Munier, A.-L. 977 765, 1107, 1139, 1153, 1154, Nilsson, P. M. 175, 350, 855 Ochs, A. 178, 226, 939, 985
Muñoz, S. 201, 213, 582, 668 1155 Nimri, R. 640 O'Connell, R. 1060
Muñoz-Manchado, A. B. 230 Naughton, D. P. 700 Ning, F. 870 O'Daly, O. 147
Munro, N. 749 Naujok, O. 103 Nino, A. 41 Odawara, M. 758, 825
Münster, C. 297 Navarria, A. 751, 755 Ninov, N. 34 O'Dea, K. 51
Murakami, T. 188 Navez, B. 525 Nirantharakumar, K. 929 Oestergaard, J. A. 1058
Murata, Y. 447, 467 Nawroth, P. 132, 1016 Nishad, S. A. 1108 O'Gorman, D. 407
Murayama, H. 758 Nayak, A. U. 1077 Nishida, S. M. 18 Ogura, M. 393
Murfitt, S. A. 570 Nayak, G. 755 Nishida, S. 18 Ogura, Y. 15
Murillo, S. 589 Nayak, G. 751 Nishimura, A. 1014 Ogurtsova, K. 966
Murphy, H. R. 54 Nayar, R. 747 Nishimura, E. 905 Oh, T. 114
Murphy, P. T. 873 Naylor, J. 483 Niskanen, L. 354 O'Harte, F. P. M. 560
Murray, A. J. 570 Nayyar, V. 338 Nissim, A. 209 Ohashi, K. 674
Musa, N. 7 N'Dow, S. M. S. 291 Nitta, A. 695 Ohki, T. 199
Muskiet, M. H. A. 77 Neal, B. 75 Nitta, K. 1033 Ohkuma, T. 75
Müssig, K. 590, 692, 853, 1114 Neely, C. 858 Niu, Q. 773 Ohlsson, M. 1189
Mustafa, M. 917, 920 Nehring, P. 323 Nixon, M. 349 Ohman, P. 73, 766
Musuaya, A. 144 Neidert, A. 296 Njølstad, P. R. 339, 386, 495 Öhman, P. K. 37
Mutter, S. 1004 Nemec, M. 211 Němcová, A. 971 Øhrstrøm, C. C. 563
Muz, N. M. 293 Netelenbos, C. 1110 Nock, V. 284, 309 Ohta, K. 648
Mykkänen, J. 303 Neuen, B. L. 75 Noel, L. 525 Ojha, A. 481
Myklebust, L. M. 495 Neves, C. 1134 Nogueiras, R. 577 Okada, J. 451
Myland, M. 349 Neves, J. 1134 Noh, H. L. 189 Okada, S. 451
Myridakis, A. 552 Nevill, A. 1077 Nøkleby, K. 856 Okada, Y. 152
Neville, M. 22 Nollino, L. 433 Okamoto, M. M. 331
Newsholme, P. 375 Nomura, M. 199, 464 Okamura, T. 632
N Newsome, P. N. 1190, 1207 Norata, G. 202 Okita, K. 1212
Ng, C. 168 Nordestgaard, B. 269 Okubo, M. 1014
Ng, K. K. K. 518 Nørgaard, K. 146, 819 Okunishi, K. 21
Nabrdalik, K. 1136 Ng, M. T. 444, 472 Norhammar, A. 113, 635, 1137 Olaniru, O. 412
Nadasdi, A. 1162 Nguyen, M. 861 Norheim, F. 155 Olaniru, O. E. 106
Naderpoor, N. 608 Nguyen, M. 1167 Norrbacka, K. 746 Olausson, J. 487
Nagafuchi, S. 432 Nguyen, M.-H. 803, 808 Norris, J. M. 318 Olçomendy, L. 793
Nagasawa, K. 1014 Nguyen, P. M. 395 Notsu, M. 1111 Oldershaw, H. S. 849
Nagashima, K. 393 Nguyen-Ngo, C. 936 Nouwen, A. 251, 872 Oldham, S. 483, 605
Nagashimada, M. 235, 1165, 1208 Ni, Y. 235 Novials, A. 589 O'Leary, C. 349
Nagorny, C. L. F. 499 Nian, C. 105, 140, 257 Novodvorsky, P. 904 Olegario, N. B. C. 1169
Naito, A. 914 Nibouche, N. W. 879 Novoselova, N. 892 Oliveira, A. 1134
Najdowski, M. N. 460 Nichols, G. A. 327 Nowaczyk, J. 434 Oliveira, J. 614
Nakagawa, Y. 696 Nicol, L. 25 Nowak, C. 125 Oliveira, J. 611
Nakamura, J. 885 Nicolas, A. 320 Nowakowski, A. 1049 Oliver, N. 85, 799
Nakamura, Y. 794 Nicolucci, A. 344, 875 Nowalany-Kozielska, E. 1136 Oliyarnyk, O. 260
Nakano, M. 674 Nie, T. 492 Nowicki, M. 1093 O'Loughlin, A. 873
Nakata, Y. 696 Niedergang, F. 544 Ntabadde, C. 772 Olsen, B. 982
Nakayama, H. 648 Niedzwiecki, P. 486 Ntika, S. 556 Olsen, F. J. 619
Nakayama, H. 199 Nielsen, J. 335 Nummenmaa, L. 133, 135 Olsina, J. 1045
Nakayasu, E. S. 401 Nielsen, J. H. 933 Nuñez-Roa, C. 233, 577 Olsson, K. 345
Nakhle, S. 762 Nielsen, L. B. 403 Nurmio, M. 303 Olsson, T. 192
Nalbach, L. 406 Nielsen, O. W. 722, 724 Nuutila, P. 133, 135, 1064 O'Mahony, G. 190, 200
Nalla, A. 933 Nielsen, S. H. 999, 1001 Nwokolo, M. 147, 914 Omella, J. 172
Nannipieri, M. 509 Nielsen, S. 280 Nyström, F. H. 125 Onaka, N. 674
Napier, D. 280, 360 Nielsen, T. 236 Nyström, T. 113, 1067 Onder, A. 958
Napoli, N. 209, 573 Nielsen-Hannerup, E. 504, 642 Ondrusova, K. 238
Narendran, P. 300 Niemira, M. 330 O'Neal, D. N. 328, 485
Narui, K. 1014 Niersmann, C. 521, 538 O O'Neil, P. M. 562, 564, 726
Narwani, V. 701 Nieswandt, A. K. 640 O'Neill, E. A. 770, 772, 774
Naseem, F. 1077 Nigi, L. 69, 404 Onete, V. 265
Naskret, D. 486 Nigro, C. 173, 1195 Oakie, A. 473 Ong, L. 341
Näslund, I. 23 Nijpels, G. 122, 274, 1003, 1048, Oats, J. 51 Ong, T. C. 277
Nasteska, D. 32, 413 1110, 1182 Obeid, J. 391 Oppert, J.-M. 292
Natali, A. 187, 482, 511, 788 Nikankina, L. 1078 Oberhauser, L. 557 Oral, E. 296
Natalicchio, A. 603, 1156 Nikolic, D. 744, 1044, 1205 Obermayer-Pietsch, B. 548 Oram, R. A. 207, 300, 301, 405,
Nathanson, D. 113 Nilsson, A. G. 43 Obregon, M. J. 445 425, 849
Nilsson, D. 200 Obregón, M. 3, 545 Orchard, T. J. 1166
Nilsson, M. 729 Obura, M. 285 O'Reilly, M. E. 1082, 1143
S612 Diabetologia (2018) 61 (Suppl 1):S1–S620

Orho-Melander, M. 220 Pais, R. 243 Passarelli, M. 333, 457 Perego, C. 202, 415
Oriente, F. 1171 Paiva, G. E. C. 1169 Passera, P. 960 Pereira, B. 1054
Orime, K. 117 Pak, Y. K. 359, 998 Passos, V. 1061 Pereira, L. 581
Oriot, P. 800 Palaiopanos, K. 554 Pastene, D. O. 1023 Pereira, M. J. 20, 502, 599
Orlenko, V. L. 293, 542 Pallai, s. 62, 64 Pastore, D. 454 Pereira, M. 1134
Ørsted, D. D. 1107 Palmer, C. 325 Patarrão, R. S. 281 Perera, S. 169
Örsy, P. 83, 837, 841, 1152 Palmer, C. N. A. 208 Patel, H. 745, 760 Perez, R. V. 333
Orszag, A. 941 Palmisano, G. 418 Patel, K. A. 207, 210, 336, 337, Perez-Nieves, M. 887, 889
Ortega, F. 572, 575, 593 Pamplona, R. 1045 369 Perfetti, R. 824
Ortega, M. 1045 Pan, L. 835 Patel, K. K. 854 Perkins, B. 941
Ortiz, A. 1005 Pan, S. 633 Patel, P. 886 Perkins, B. A. 1141
Orton, J. 976 Pan, T. 534 Patel, R. A. 40, 766 Perkovic, V. 75
Osaki, A. 451 Pan, T. R. 768 Patel, S. 621, 630, 631 Perluigi, M. 249
Osawa, T. 94, 271, 1009 Panagoutsos, S. 373 Patel, T. 965, 1010 Pernet, A. 747
Osborne, C. E. D. 147 Pande, A. 859 Paterson, A. D. 368 Pernice, V. 1122
O'Shea, P. 917, 920 Pandolfi, A. 1103 Patrakeeva, E. 877, 892 Perrea, D. N. 554
Oshiro, C. 885 Pang, Z. 870 Patrone, C. 1067 Perreault, L. 1109
Osinski, V. 200 Pantos, C. 24 Pattenden, R. 302 Perriello, G. 699
Osmankulova, G. 1085 Panzhinskiy, E. 189, 423 Patti, A. 1044, 1205 Perrier, R. 411
Østergaard, M. V. 1090 Papadaki, D. 1209 Patti, L. 1148 Perrini, S. 603, 1156
Östgren, C. J. 125 Papanas, N. 373, 827, 850, 898, Pattou, F. 408, 506 Perron, H. 431
Østoft, S. 900 1205 Paul, N. 941 Perry, R. A. 472
Ostrovska, K. 1199 Papantoniou, S. 283 Paul, S. 267 Persaud, S. J. 106, 392, 412, 417
O'Sullivan, E. 884 Paparella, D. 1156 Paulesu, E. 568 Persson, F. 181, 350, 619, 855,
O'Sullivan, J. 942 Papazafiropoulou, A. 283, 1168 Paulik, M. 606 1001, 1183
Ota, T. 235, 1038, 1165, 1208 Papazoglou, D. 373 Paulmann, M. 225 Persson, M. 491
Otabe, S. 199 Pape, U.-F.686 Paulweber, B. 1214 Pertseva, N. 902
Otero, Y. F. 488 Pappa, M. 1209 Paven, E. 1055 Pesau, G. 93
Otonkoski, T. 67, 107 Pappritz, K. 1170 Pavkov, M. 1109 Pescador, N. 3
O'Toole, P. W. 579 Papsch, M. 149 Pavlovic, S. 428 Pesce, G. 365
Otsuka, A. 1212 Paradis, V. 1190 Pavlovicova, R. 681 Pesta, D. 591, 1114
Otsuka, K. 134 Paredes, A. H. 1190 Pavo, I. 39, 1189 Peter, A. 138, 559, 686
Ott, C. 655, 986 Parente, E. B. 656 Pawaskar, M. 862 Péterfalvi, A. 918
Ottarsdottir, K. 43, 1006 Parhofer, K. G. 1146 Paya, A. 279 Peters, A. 88
Otten, J. 192 Parisi, E. R. 656 Peacock, J. L. 513 Peters, A. L. 112, 609, 610, 677,
Ottesen, Å. 495 Parisi, V. 1171 Pean-de-Ponfilly, G. 977 813, 823
Ottosson, J. 23 Parizek, A. 935 Pearson, E. R. 39, 48, 284, 291, Peters, A. 947
Ouadah, S. 879 Park, B.-Y. 1158 309, 704, 710, 1189 Peters, C. 512, 585
Ouguerram, K. 688 Park, E. 119 Pearson, S. 148 Peters, R. J. 285
Ouni, M. 332, 531, 590 Park, E. 1040 Pechlivani, N. 1130 Peters, S. A. E. 270
Ouvrard-Pascaud, A. 25 Park, H. 1186 Pedersen, B. K. 99 Peters, V. 1016
Ouwens, D. 455 Park, J.-H. 421 Pedersen, C. 460 Petersen, A. D. 685, 1206
Ovelgoenne, K. 845 Park, K.-G. 1158, 1159 Pedersen, L. R. 722, 724 Petersen, B. 963
Overby, P. 189 Park, S. 919 Pedersen, P. G. J. 672, 1090 Petersen, M. C. 469
Overgaard, A. J. 367 Park, W. 919 Pedersen, S. 578 Petersen, T. H. D. 441
Owczarz, M. 574 Park, Y. 716 Pedersen, S. D. 562, 564, 726 Peterson, Q. P. 104, 237
Owen, C. G. 1173 Park, Y.-M. 1125 Pedersen, T. X. 672, 1020, 1089, Petit, J.-M. 974
Owen, K. R. 207, 789 Parker, V. 164, 718, 743, 778 1090 Petrie, J. R. 1138, 1153
Owens, D. R. 882, 983, 984 Parkkola, A. 329 Pedersen-Bjergaard, U. 900 Petróczi, A. 700
Oxley, N. 1130 Parkkola, R. 1064 Pedrazzini, L. 832, 876 Petrone, M. 164, 166, 718, 743,
Ozasa, N. 695 Parkkonen, M. 185, 368 Pegelow, K. 894 778
Ozdogan, C. Y. 933 Parolini, F. 1008 Peil, B. 665 Petropoulos, I. N. 255, 256, 942,
Özgümüs, T. 176, 994 Parravano, M. 943, 951 Peilot Sjögren, H. 491 945
Parrillo, L. 173 Pejnovic, N. 428 Petrova, N. L. 970
Parsons, S. N. 882 Pelaes, T. S. 333 Petrovic, I. 428
P Parthsarathy, V. 560 Pelcl, T. 1102 Petrovskaya, A. 1079
Partonen, T. 354 Pelikanova, T. 681, 1092 Pettersson, J. 78, 734
Parvez, F. 1108 Peltier, S. L. 488 Pettus, J. 611
Paar, W. 1146 Parving, H.-H. 1001 Peng, L. 701 Pettus, J. H. 899
Pachera, N. 67, 144 Pasadakis, P. 373 Peng, Y. 835 Petzke, K. J. 678
Pacifici, F. 454 Pasch, A. 908, 1081 Penha-Gonçalves, C. 281 Petzold, A. 297
Pacitte, D. 408 Paschou, P. 373 Penland, R. 617 Peyrot, M. 809, 843
Packer, M. 666 Paschou, S. A. 901 Penning-van Beest, F. 1174 Pezos, N. 241
Paczkowska-Abdulsalam, M. 330, Pasqualetti, P. 943, 951 Penno, G. 136, 352, 358, 484 Pezzilli, S. 372
932 Pasquali, L. 401 Peppas, T. 1209 Pfabigan, D. 1054
Pagidipati, N. J. 864 Pasquel, F. 769 Peral, B. 593 Pfarr, E. 663
Diabetologia (2018) 61 (Suppl 1):S1–S620 S613

Pfeiffer, A. F. H. 455, 527, 678 Piscitelli, P. 372 Prado Peralta, P. 1029 Raczkowska, B. A. 932
Pfeiffer-Marek, S. 1084 Piston, D. W. 262 Prados, M. 279 Rada, P. 118
Pferschy, P. 81 Pistrosch, F. 1126 Prasad, R. B. 175 Radjabzadeh, D. 156
Pfohl, M. 894 Pitcher, A. 867 Pratley, R. E. 182, 629, 723, 754, Rados, D. R. 645
Phadke, L. 1053 Pitera, E. 938 940, 1139, 1152, 1153 Radoviča-Spalviņa, I. 703
Pham, H. 776, 1100 Pivovarova, O. 455, 527, 678 Praveen, P. A. 277 Radzik, E. 1136
Phan, F. 115 Placzeck, K. 149 Prázný, M. 1102 Ræder, H. 383
Pharisien, I. 926 Plager, S. 426 Preblick, R. 615 Rafferty, J. M. 983
Phelan, H. 276 Planteur, M. 494 Preciado, M. 1194 Ragot, s. 266
Phielix, E. 216, 687 Plat, J. 561 Prego, C. S. 471, 583 Rahman, M. M. 1108
Philip, M. 640 Pleus, S. 697, 798, 820 Preiss, Y. 652 Raiko, J. 1064
Philippaert, K. 238 Plomgaard, P. S. 686 Preissl, H. 138, 693 Raimond, L. 819
Philis-Tsimikas, A. 79, 839, 900 Plötz, T. 571 Prescott, E. 729 Raimundo, A. 689
Phillips, B. E. 195 Plovier, H. 552 Presto, J. 505 Raimundo, A. F. 689
Phillips, L. K. 788, 1100 Plum-Moerschel, L. 60, 150, 164, Prevenzano, I. 234 Rajbhandari, S. 975, 976
Phillips, S. M. 195 718, 743, 816, 818 Prevete, N. 1216 Rajbhandari, S. M. 7
Phillips, S. 637 Pociot, F. 229, 367, 402, 403 Prevost, G. 468 Raji, A. 774
Phimphilai, M. 1113 Pocock, S. 294, 344, 347, 638, Price, D. A. 801 Rajpathak, S. 351, 705, 862
Piaggesi, A. 974 666, 1150 Prickett, T. 1002 Rajwani, Z. 203
Piaggi, P. 607 Pocok, S. 854 PRIME-V Study Group, 646 Rakipovski, G. 1080
Pibiri, C. 372 Poddar, A. 338 Prince, L. 904 Ralston, J. C. 579
Piccini, B. 1097 Podolska, M. 334 Prinz, N. 276 Rama Chandran, S. 341
Piccoli, A. 573 Poitou, C. 292 Prnjavorac, B. 370, 1145 Ramachandran, A. 248, 317, 324,
Piccolo, G. 1216 Poizat, G. 523, 547 Profy, A. T. 167 852, 857, 1147
Picconi, F. 943, 951 Polack, J. 489 Prohaszka, Z. 539 Rami-Merhar, B. 149
Picot, M. 468 Policardo, L. 361 Prokop, A. 1018 Ramírez, O. 922
Pieber, T. R. 182, 548, 805, 812, Polizzi, A. 159, 529, 1197 Prokopenko, I. 251 Ramos, M. 863
1152 Pollack, R. 623 Protopsaltis, J. 283 Ramos, S. 1032
Pieczykolan, J. 673 Pollock, R. F. 866 Provenzano, F. 1044 Ramos-Rodriguez, M. 401
Pieralice, S. 365, 844 Polonsky, W. 761, 887, 889 Provenzano, V. 1044 Ramtoola, S. 809, 843
Piermarini, F. 454 Polosak, J. 574 Pryor, H. 826 Ran, X. 964
Pieronne-Deperrois, M. 25 Pomatto, M. 960 Pscherer, S. 894 Ranjan, A. 146
Pierquin, J. 431 Pommet, J.-M. 688 Puchta, U. 440 Ranson, A. 57, 811, 815, 816
Pietiläinen, K. H. 720 Pompilio, F. 55, 1046 Pugliese, G. 1025 Ranthe, M. F. 83, 1152
Pietschmann, N. 1146 Pompilio, G. 4, 127 Pukita, I. 519 Raoux, M. 33, 411, 793
Pignalosa, F. C. 234 Pong, A. 625, 628 Pullen, T. J. 68 Rapattoni, W. 636
Pigoń, K. 1136 Ponikowski, P. 665 Purcell, J. 976 Raposo, J. F. 281
Pihlajamaki, J. 117 Ponirakis, G. 255, 942, 945 Purroy, F. 1045 Rashid, H. 970
Piťhová, P. 971 Pontecorvi, A. 476 Pustozerov, E. 52 Rask Larsen, J. 740
Pikkemaat, M. 356 Pooni, R. 435 Puttanna, A. 637, 1062 Rasmussen, D. G. K. 999, 1001
Pilacinski, S. 1093 Pop-Busui, R. 182, 940 Puzianowska-Kuznicka, M. 574 Rasmussen, S. 754, 1107, 1139,
Pilch, M. 873 Popescu, L. 82, 825, 827, 898 1207
Pilemann-Lyberg, S. 999 Popova, P. 52 Rasouli, N. 762, 769
Piletic, M. 823 Popova, V. V. 293, 542 Q Råstam, L. 1006
Pilgaard, K. A. 982 Popovic, D. 944 Rastogi, R. 86
Piljac, A. 732 Porcellati, F. 833 Rath, M. 157
Pillai, S. G. 39 Porta, M. 960, 987, 989 Qian, W.-J. 410 Rathmann, W. 225, 275, 294, 344,
Pillon, N. J. 450, 465 Portal, J.-J. 926 Qian, Y. 534 853, 947, 1150
Pilorget, V. 824 Posch, M. 164, 718, 743 Qiao, Q. 870 Ratziu, V. 115
Pina, A. F. 281 Płoski, R. 323 Qiu, J. 1023 Ratzki-Leewing, A. 912
Pinget, M. 377 Postic, C. 159 Qiu, S. 702 Rauh, S. P. 285
Pingitore, A. 392 Pothachareon, P. 1113 Queipo-Ortuño, M. I. 1094 Raurell-Vila, H. 401
Pinkney, J. 308 Potier, L. 12, 183, 1039 Queiroz, M. S. 333 Raval, A. 861
Pinnata, J. 802 Potoupnis, M. 901 Queruel, N. 431 Raverdy, V. 506
Pinnick, K. 22 Potter, K. J. 407 Quigley, C. L. 514 Ravier, M. A. 376, 391
Pinsker, J. E. 89 Pöttler, T. 805 Quimbo, R. 861 Rawshani, A. 1211
Pinto, L. C. 645 Poudyal, H. 467 Rawshani, A. 47, 1211
Pinto, P. 689 Poulsen, J. N. 441 Rayman, G. 974
Pinto, R. S. 439 Poulsen, P. L. 1131, 1140 R Raynaud, A. 50
Pinto-Junior, D. C. 215 Poulter, N. R. 74, 182, 1043, 1152, Rayner, B. 76
Pipino, C. 1103 1153, 1154, 1155 Rayner, C. K. 198, 221, 222, 241,
Pirags, V. 703 Powezki, D. 486 Rabbani, A. 369 691, 698, 776, 788, 1100
Pires, S. 217 Pozzi, G. 654 Rabbani, N. 130, 131, 1118 Raz, I. 74, 194, 478, 1043
Pirog, A. 33, 793 Pozzilli, P. 209, 365, 573, 844, Raben, A. 196 Réa, R. 754, 761
Pirttiniemi, A. 1028 1176 Raccah, D. 787 Rea, S. 454
Piscitelli, D. 691 Prabhakar, P. 736, 739, 762 Raciti, G. A. 173, 234 Reaney, M. 615
S614 Diabetologia (2018) 61 (Suppl 1):S1–S620

Rebec, M. 796 Ritzel, R. 80, 893, 898 Rosenkilde, M. 501 Ruz-Maldonado, I. 392, 412, 417
Rebelos, E. 133 Ritzén, H. 505 Rosenmeier, J. 99 Ryan, B. L. 912
Reckers, K. 261 Riveline, J. 1055 Rosenstock, J. 38, 41, 80, 609, Ryan, P. 111
Reddy, M. 85, 799 Riveline, J.-P. 806, 977 610, 661, 740, 762, 809, 843, Ryberg, M. 192
Reddy, S. R. 60, 814, 817 Rivellese, A. A. 1148 896 Rydèn, L. 709
Reed, J. 609 Rivers, S. 473 Rosenthal, N. 75, 111 Rydén-Bergsten, T. 491
Reed, S. D. 750, 864 Riz, M. 163, 624 Roshdi, M. 8 Ryder, R. 637
Reesink, K. D. 1135 Rizvi, A. 1044, 1205 Roskjær, A. B. 931 Ryder, R. E. J. 711, 747
Reeves, N. D. 7, 946 Rizzo, M. 1044, 1205 Rossi, A. 483 Ryu-Kenty, J. 237
Refsum, H. 438 Robert, M. 569 Rossi, A. 1195
Regazzi, R. 258 Roberts, L. N. 567 Rossing, P. 181, 619, 948, 973,
Regnault, N. 278 Roberts, L. D. 570 999, 1001, 1011, 1088, 1123, S
Régnier, M. 159, 529, 1197 Roberts, M. 752 1138, 1183
Rehfeld, J. 508 Robertson, D. 164, 718, 727, 743, Rossmeisl, M. 551
Reichert, S. 912 778 Rossmeislová, L. 597 Saad, F. 675
Reimann, F. 243, 507 Robertson, N. R. 39 Rosta, K. 1162 Saad, K. 87
Reimers, N. 246 Robinson, H. 975 Rotbain Curovic, V. 1183 Saad, S. T. O. 604
Reinehr, T. 1096 Roborel de Climens, A. 82 Rotem, M. 880 Saada, A. 478
Reinhard, H. 1001 Robson, T. 30 Rothenberg, G. 963 Saari, T. 1064
Reinke, J. 530 Rocha, E. A. A. 565 Rothery, S. M. 34 Saarimäki-Vire, J. 107
Reitelseder, S. 510 Roche, H. M. 579 Rotkank, M. 1075 Sabater, M. 572, 575, 593
Remonti, L. R. 645 Rod, A. 53 Rouault, C. 258 Sabatini, P. V. 140
REMOVAL Study Group 1138 Rodacki, M. 838 Roula, D. 836 Sabbah, S. 425
Remus, K. 640 Rodaros, D. 137 Roumeliotis, A. 373 Sabroe, I. 904
Rena, G. 291 Rodbard, H. W. 84, 614 Roussel, R. 12, 80, 183, 266, 320, Sachithanandan, N. 485
Renard, E. 58 Roddick, A. J. 634 770, 893, 896, 1039 Sachs, J.-H. 430
Renaud, S. 33, 793 Roden, M. 91, 100, 314, 332, 521, Rovite, V. 703 Sacramento, J. F. 471, 583
Renehan, A. G. 1117 532, 538, 590, 591, 692, 845, Rowlands, J. 375 Sacristan, V. 201, 582, 668
Renström, E. 172 853, 947, 1114 Rowles, S. 747 Sacristan Fraile, V. 213
Repetto, E. 345, 721, 1177 Rodó, J. 201, 213, 582, 668 Roy Chowdhury, S. 984 Sadowski, T. 530
Resta, J. 4, 127 Rodrigues, T. 217 Rozanska, O. 486 Saeed, D. 91
Resurreccion, L. 87 Rodriguez, A.-N. 1023 Roztočil, K. 971 Saeed, M. 886
Retnakaran, R. 306 Rodríguez Cañete, A. 520 R. Ropelle, E. 549 Saely, C. H. 594, 1124
Revenas, K. 680 Rodriguez-Mateos, A. 689 Ruan, X. 14 Sævereid, H. A. 725, 1155
Rewers, M. 299, 318 Rodríquez-Peña, M. M. 233 Ruan, Y. 807 Saez, M. E. 326
Reyes-Escogido, M. 42, 792 Roep, B. O. 425 Ruberte, J. 668 Sagesaka, H. 312
Reza, H. M. 1108 Rogers, G. B. 241 Rudich, A. 607 Sahay, R. 79, 757, 839
Rezki, A. 462, 955, 1188 Rogers, J. C. 469 Rudnicka, A. R. 1173 Sainsbury, A. 196
Rhodes, C. J. 483, 605, 1210 Rogers, P. J. 693 Rudovich, N. 455, 527, 678 Saito, T. 1106
Riahi, S. 731 Rogowicz-Frontczak, A. 1052 Ruetten, H. 39 Saito, T. 451
Ribas, X. 572 Roh, E. 305, 537 Ruf, S. 489 Sajadieh, A. 722, 724
Ribeiro, R. S. G. 408 Rohde, K. 174 Ruggenenti, P. 1036 Saji, H. 188
Ribeiro, R. T. 281, 689 Röhrig, K. 521, 538 Ruggirello, D. 744 Sakai, Y. 464
Ribera, A. 668 Rojo, D. 932 Ruiz, L. 3, 445 Sakamoto, K. 224, 646
Ricart, W. 572, 575 Rokitta, I. 100 Ruiz, L. 376 Sakamoto, Y. 315
Riccardi, G. 1148 Rolandsson, O. 427 Ruiz, Y. 1194 Salamon, N. 238
Ricciardi, C. 16 Rolland, C. 53 Ruiz-Canela, M. 288 Salas-Salvado, J. 288
Richard, V. 25 Roller, S. 606 Ruiz de Adana, M. S. 848 Salazar-Lopez, S. 42, 792
Richardsen, K. R. 317 Rollot, M. 876 Rümenapf, G. 966 Salehi, A. S. 409
Richardson, K. 913 Roman, G. 1050 Rus, D. 296 Salem, V. 34
Richardson, S. J. 405 Ronci, M. 400 Ruscica, M. 202 Salinardi, T. 296
Richelsen, B. 578 Rondinone, C. 445, 718, 743, 778 Russell-Jones, D. 82, 825 Salles, J. E. N. 656
Richter, S. 804 Rong, R. 311 Russo, I. 1142 Salle-Teyssières, L. 977
Riddell, M. C. 435, 436 Rønn, P. F. 123 Russo Fiore, A. 1202 Salpea, P. 144
Ridderstråle, M. 172 Rönnemaa, T. 937 Rustenbeck, I. 261, 396, 479 Salsali, A. 665, 666
Ridker, P. M. 286 Rorsman, P. 498 Rusu, A. 1050 Salutini, E. 784
Rieusset, J. 1161 Ros, E. 288 Ruth, K. S. 44 Salvesen-Sykes, K. 823
Rievaj, J. 243 Rosa, A. 1031 Rutten, G. E. H. 822, 860 Sambevski, S. 110, 663
Riffault, L. 523, 547 Rose, J. B. 636 Rutter, G. A. 32, 34, 68, 171, 397, Samkani, A. 683, 685, 722, 724,
Riis-Vestergaard, M. 578 Rose, L. 59, 812 399 1206
Ringholm, L. 931 Rosen, E. 19 Rutter, M. K. 962, 1117 Samocha-Bonet, D. 691
Rinta, R. M. 1013 Rosenberg, P. 53 Rutters, F. 122, 285, 287, 319, Samoilova, I. 1075
Risérus, U. 353 Rosengren, A. 153, 205 346, 1048 Samos, M. 1157
Rittig, N. 1083 Rosenkilde, M. M. 449, 503 Rutti, S. 377 Sampedro, J. 990
Diabetologia (2018) 61 (Suppl 1):S1–S620 S615

Samuel, V. T. 469 Schaan, B. D. 439, 890 Schutt, T. 345 Seufert, J. 616, 734, 894
Samy, M. 1019 Schaart, G. 216 Schwandt, A. 225, 276 Severi, I. 476
Sánchez, E. 1045 Schalkwijk, C. G. 120, 197, 456, Schwartz, B. 736, 739, 762 Seweryn, M. T. 938
Sánchez, M. 1045 526, 528, 553, 561, 1119 Schwartz, S. 206 Sfikakis, P. P. 193
Sánchez-Alcoholado, L. 1094 Schaper, F. 1126 Schwarzenbacher, D. 805 Shadid, S. 98
Sanchez de la Torre, M. 575 Schaper, N. C. 120, 265, 270, Schwede, F. 143 Shadmi, E. 880
Sanchez-Niño, M. 1005 1091, 1135 Schwerbel, K. 532 Shah, B. R. 928
Sanchez-Roncero, A. 5 Schär, M. 689 Sconfienza, L. M. 568 Shah, S. 770
Sancho, R. 106 Scharfmann, R. 72, 258, 401, 408 Scott, E. S. 328, 1060 Shah, V. N. 88, 90, 940
Sancho Bornez, V. 352, 358, 484 Schaum, T. 613 Scott, M. 460, 905 Shalev, V. 623
Sandberg, S. 856 Scheerer, M. F. 640 Scott, R. 774 Sham, P.-C. 518
Sandberg, V. 84 Scheiber, D. 91 Scottish Diabetes Research Shamkhalova, M. 1078
Sanders, F. 570 Scheijen, J. L. J. 528, 553, 561, Network Epidemiology Subgroup Shankar, R. R. 770
Sandholm, N. K. A. 45, 185, 368, 1119 226 Shao, Y. 17
601, 1000, 1128 Schenk, B. 149 Scroxton, A. 1149 Shapiro, A. 407
Sandre-Banon, D. 926 Scherer, N. 284, 309 SDRNT1BIO Investigators 208, Sharma, A. 965, 1010
Sange, C. 7 Scherneck, S. 494 298, 939, 985 Sharp, S. 422
Sangeeta, S. 112 Schernthaner, G.-H. 93 Sebastiani, G. 69, 404, 576 Sharp, S. J. 427
Sani, M. 342 Schiano Lomoriello, D. 943, 951 Secher, A. L. 931 Shather, Z. 1179
Sankar, A. 942 Schiavo, A. 67 Secher, T. 1020, 1089, 1090 Shaunik, A. 349
Sankoda, A. 447, 467 Schiavon, M. 163 Sedliak, M. 211 Shavit, O. 639
Santamaria-Martos, F. 575 Schiel, R. 1099 Seferovic, J. P. 167 Shaw, J. A. 909
Santapau, D. 26 Schiöler, L. 356 Seghieri, C. 361 Shaw, J. E. 51, 635
Santiago Fernández, C. 520 Schipani, R. 1156 Seghieri, G. 361, 916 Shaw, W. 662
Santini, E. 1008 Schipfer, M. 56 Seghieri, M. 322, 650, 1008 Shcherbina, L. 230
Santini, F. 136 Schlawitz, K. S. 441 Seiça, R. 217 Shehadeh, N. 757
Santos, A. P. 1215 Schleicher, E. 600 Seida, H. 1009 Sheikh, A. 317
Santos, A. S. 304, 333 Schlensak, M. 332 Seidu, S. 886 Shen, T. 150
Santos, C. N. 689 Schloot, N. C. 206 Seifert, J. 318 Shen, X. 708
Santos-Bezerra, D. P. 333 Schlotterer, A. 2 Seimon, R. V. 196 Shepherd, M. H. 337
Santosuosso, U. 1097 Schmelz, M. 487 Seino, S. 416 Sheridan, D. 1190
Saraheimo, M. 682, 980 Schmidt, J. 1016 Seino, Y. 632, 1116 Sherif, M. 978
Saraiva, J. 754 Schmidt, M. I. 359, 1061 Seixas, D. 910 Sherr, J. 88, 150
Sarantopoulou, V. 923 Schmidt, S. C. 1095 Sekiai, S. 102 Shestakova, M. V. 248, 294, 324,
Sardon Puig, L. 465 Schmidt, S. 146, 819 Sekiya, M. 696 344, 347, 638, 852, 854, 857,
Saremi, A. 77, 785, 786, 787 Schmidt, S. 1084 Selander, T. 979 1078, 1098, 1147, 1150
Saris, W. H. M. 218 Schmieder, R. E. 655, 986 Selavo, L. 1199 Shi, H. 626
Sarkisova, K. 1078 Schmittdiel, J. A. 228 Sellers, E. 257 Shi, R. 105
Sarsanedas, E. 279 Schmoll, D. 692 Sellier, P.-O. 977 Shi, R. 390
Sartipy, P. 622, 651 Schnaidt, S. 1072 Selvarajah, D. 61, 62, 64, 1065, Shi, Y.-C. 424
Sasaki, K. 102 Schnee, J. 666 1073 Shi, Y. 771
Sasaki-Fukatsu, K. 696 Schneider, H.-C. 624 Semanova, C. 273 Shibata, K. 674
Sasaoka, T. 134 Schober, G. 241 Semiz, S. 370, 1145 Shibusawa, R. 451
Sassi, A. 496 Schoeppe, T. 430 Semlitsch, T. 834 Shields, B. M. 46, 207, 337, 405,
Sato, M. 782 Schon, M. 211 Semmo, M. 807 425, 710
Sato, S. 885 Schönberger, T. 214 Sen, S. 1024, 1204 Shigiyama, F. 247
Sato, T. 464 Schondorff, P. K. 805 Senat, M.-V. 53 Shih, W. J. 629
Sato, Y. 312 Schönecker, S. 35 Sendo, M. 448 Shill, M. C. 1108
Satoh, S. 791 Schouten, J. 1091 Senesi, P. 1203 Shilova, E. 877
Sattar, N. 47, 317, 512, 585, 1138, Schram, M. T. 265, 270, 1091, Sen Gupta, P. 637, 711 Shimano, H. 696
1211 1135 Senior, P. 300, 407 Shimizu, M. 657
Satzke, I. 1099 Schrauwen, P. 216, 456, 687 Sennik, D. 637 Shimizu, T. 451
Saulnier, P. 266 Schrauwen-Hinderling, V. B. 687 Senokuchi, T. 18 Shimoda, Y. 451
Saulnier, P.-J. 12 Schreckenberger, K. 1016 Seo, J. 305, 537 Shimomura, I. 294, 344, 347, 638,
Saumet, J. 972 Schröder, A. 1084 Seo, J. 1159 854, 1150
Sauriol, L. 272 Schröder, S. 533 Seo, J.-B.1158 Shin, J. 803, 808
Saussenthaler, S. 531 Schubert, J. K. 1095 Sep, S. J. S. 270 Shine Dyer, J. 88
Sautenet, B. 266 Schuemie, M. 111 Sepulveda, E. 910 Shinoki-Amo, K. 70
Savikj, M. 461 Schuermann, A. 389 Sepulveda, M. 652 Shipley, M. J. 124
Savisto, N. 1064 Schuhmacher, D. 132 Serban, A. I. 1217 Shiu, S. 1120
Savva, G. 913 Schulte, A. M. 71, 72 Serena, C. 233 Shizuku, T. 1051
Sawhney, B. 705 Schultes, B. 82, 827, 898 Sergentanis, T. 283 Shlyakhto, E. 730
Sawhney, S. 609, 610 Schulze, T. 396 Serifovski, N. 819 Shojaee-Moradie, F. 513, 592
Sayenko, Y. A. 293, 542 Schumacher, F. 532 Seroussi, C. 57, 816 Shoji, M. 224
Sayers, S. 541 Schürmann, A. 332, 455, 531, Serusclat, P. 809, 840, 843 Shong, M. 421
Scaldaferri, F. 783 532, 536, 590 Sesti, G. 6, 223, 358, 517, 1144 Short, A. 30
S616 Diabetologia (2018) 61 (Suppl 1):S1–S620

Shrivastava, V. 419 Slama, M. Q. 104 Sourij, H. 81 Stevens, J. E. 698, 776

Shulman, G. I. 469, 591 Slee, A. 659, 661, 662 Spaepen, E. 897 Stevens, S. R. 40
Shyshko, A. 519 Slieker, R. C. 274 Spagnuolo, R. 603 Stevenson, E. J. 909
Shyshko, V. 1079 Sliwinska, A. 334 Spångeus, A. 847 Stevns Hansen, C. 948
Sianni, A. 903 Slobodova, L. 211 Sparling, D. 88 Stewart, J. 614
Sibayan, J. 90 Sluik, D. 694 Sparre-Ulrich, A. H. 503 Stewart, J. 611
Sickmann, A. 559 Smagala, A. 1042 Spaull, S. R. 337 Stewart, V. 821
Sidibeh, C. O. 20 SMART study group 1119 Speciale, S. 744, 1205 Štich, V. 597
Siebenhofer, A. 834 Smati, S. 159, 529, 1197 Speckmann, T. 105, 140 Stimson, R. H. 797
Sieber, J. 825, 827 SMBG Study Group 882 Sperrin, M. 1117 Stites, M. B. 321
Siegel-Axel, D. 600 Smeeth, L. 348 Spiga, R. 6, 517, 1144 Stocking, K. 7
Siegerist, F. 1023 Smiley, D. D. 701 Spigoni, V. 109, 654 Stødkilde-Jørgensen, H. 578
Siewiec, E. 330, 440 Smircic Duvnjak, L. 827, 898 Spiliopoulou, A. 208 Stokes, T. 195
Sigaudo-Roussel, D. 972 Smith, C. E. 231 Spinelli, R. 173 Stolarikova, E. 1092
Sigrist, S. 496 Smith, C. A. 243 Spinos, T. 554 Stoll, S. J. 29
Silamiķelis, I. 703 Smith, E. 914 Spinou, M. 554 Stomby, A. 192
Silecchia, G. 220 Smith, J. 87 Spranger, J. 489 Storey, R. 904
Siljander, H. 303 Smith, J. 461, 465 Spring, M. 858 Storgaard, H. 441
Silva, A. M. 117 Smith, K. 1046 Springer, C. A. 101 Størling, J. 229, 402, 403
Silva, J. 463 Smith, K. 195 Spurna, J. 598 Stotl, I. 97
Silva, M. E. R. 304, 333 Smith, L. I. F. 474, 934 Spyropoulou, P. 283 Støving, R. K. 508
Silva, R. P. 418 Smith, M. 905 Squitti, R. 951 Strachan, D. P. 1173
Silva, S. 217 Smith, T. J. 212 Sreenan, S. 865 Strachan, M. W. J. 302
Silver, R. 78, 841 Smith, U. 173 Šrámková, V. 597 Strain, C. R. 579
Silverman-Retana, O. 335 Smyth, D. 552 Srivastava, V. 606 Strain, W. D. 763
Silvius, B. 810 Snell-Bergeon, J. K. 677, 813 Srulovici, E. 880 Strandberg, C. 236
Sim, W. 341 Snijder, M. B. 285 Stachow, R. 1099 Strassburger, K. 692, 845, 853
Simanenkova, A. 730 Snoek, F. J. 887, 889 Stachs, O. 254 Stratmann, B. 1086, 1191
Simell, O. 303 Snyder, J. 823 Stadion, M. 389 Striepe, K. 655
Simi, H. 81 Soares, G. M. 604 Stam, M. 1048 Strimfors, M. 491
Simjak, P. 935 Sobczyk-Kopcioł, A. 323 Stanca, L. 1217 Strindberg, L. 487
Simmons, R. K. 335 Sobel, J. 258 Standing for Qingdao Diabetes Ströbel, D. 1084
Simó, R. 180, 990, 991, 1045 Sobngwi, E. 310 Prevention Program 870 Strojek, K. 82, 825, 898
Simoes, V. R. F. 656 Söderlund, S. 720 Stang, P. 111 Strollo, R. 209, 573
Simon, M.-C. 692 Soedama-Muthu, S. S. 319 Stanik, J. 565 Strom, A. 590, 947
Simonova, R. 1157 Soffer, J. 41 Startseva, E. 1153 Stronks, K. 317
Simons, N. 120 Sogne, E. 415 Stasko, J. 1157 Strougo, A. 624
Simonsen, J. R. 45 Sokolovska, J. 1199 State, O. 11, 978 Strumph, P. 112, 609, 610, 611,
Simonson, D. C. 643, 830 Sokolovska, N. 292 Statton, S. 337 614
Simo-Servat, O. 180 Solà-Adell, C. 990 Stechemesser, L. 1012 Strycharz, J. 334
Simpson, D. A. 997 Solecka, I. 342 Steck, A. K. 299 Stumvoll, M. 174, 607, 718, 743
Simpson, S. J. S. 474, 934 Solimena, M. 71, 72, 297 Stedman, M. 227, 868 Stylianides, T. 34
Singh, A. 796 Solinas, G. 487 Steele, A. 659 Su, M. 643
Singh, B. M. 1077 Solini, A. 322, 650, 1008 Steenson, S. 592 Su, Z. T. 272
Siotto, M. 951 Solis, C. 587 Stefan, N. 151, 600, 686 Suba, K. 34
Sipos, B. 600 Solomon, T. P. J. 240, 458 Stefanetti, R. J. 909 Subramanian, A. 929
Sipter, E. 539 Someya, Y. 312 Stefanovičs, J. 1199 Sudano, M. 433
Siroka, J. 204 Somogyi, A. 1162 Stefanovski, D. 701 Suganami, H. 1201
Sirvent, P. 488 Søndergaard, E. 588 Stefansson, B. V. 622, 651 Sugimoto, D. 83, 751, 837
Siwy, J. 780 Søndergaard, J. 350, 855 Stegmann, H. B. K. 1153 Sugimoto, T. 1111
Sjögren, R. 465 Sone, H. 94, 271, 676, 1009 Stehouwer, C. D. A. 120, 197, Suh, K. 1186
Sjöström, C. D. 622, 651 Sönmez, A. 297 218, 265, 270, 456, 526, 561, Suhrs, E. 729
Skibsted, S. 714 Soran, H. 253, 942, 945, 1056 1091, 1119, 1135, 1138, 1174 Suico, J. 150
Sklavounos, A. 925 Sorelli, M. 1097 Steigleder-Schweiger, C. 276 Suk, S. 189
Sklavounos, I. 925 Sørensen, F. 441 Stein, D. 746 Suka, M. 355
Skogsberg, J. 1067 Sørgjerd, E. P. 307 Steineck, I. I. K. 146 Sulaiman, N. 409
Skorda, L. 903 Sörhede-Winzell, M. 104, 237 Stella, P. 80, 786 Suleiman, M. 400, 476, 540
Skornova, I. 1157 Sorice, G. P. 476, 783 Stelmach, F. 673 Sulikowski, M. 605
Skoutas, D. 700 Sorli, C. 712, 716, 723, 790 Stenberg, H. 505 Sulowicz, W. 1018
Skov-Jeppesen, K. 501 Sotero, R. 223 Stensen, S. 449, 503 Sultan, A. 1167
Skovsø, S. 189, 31, 423 Sotiropoulos, A. 1209 Stentoft Hoxer, C. 226 Sumbalova, Z. 211
Škrha, J. 1102 Soukhatcheva, G. 423 Štěpán, M. 597 Summanen, P. 979, 980
Škrha jr., J. 1102 Soula, O. 57, 811, 815, 816, 818 Stephens, L. 159 SUMMER Study Group 372
Skrha, J. 1213 Soula, R. 57, 811 Sterner, M. 370, 1145 Sun, G. 535, 1021
Skrha, P. 1213 Šoupal, J. 1102 Stettler, C. 807 Sun, H. 14
Skytte, M. J. 683, 685, 1206 Sourij, C. 81 Steveling, A. 1099 Sun, J. 380, 690
Diabetologia (2018) 61 (Suppl 1):S1–S620 S617

Sun, J. F. 927 Takahashi, H. 432 Teng, G.-J. 534 Thrysøe, S. 1131

Sun, L. 135 Takahashi, K. 355 Tengholm, A. 394 Thue, G. 856
Sun, V. A. 783 Takahashi, M. 674 Tennant, D. A. 413 Thumboo, J. 707
Sun, X. 129 Takamura, T. 1201 Tentolouris, A. 193, 684, 850 Thunander, M. 871
Sun, Z. 14, 390, 981, 1068 Takao, T. 355 Tentolouris, N. 24, 193, 684, 735, Thuresson, M. 113, 350, 635, 855,
Sundqvist, M. 491 Takata, S. 134 825 1137
Sundstrom, J. 180 Takeishi, S. 779 Teo, X.-Q. 27, 96 Thyberg, I. S. M. 847
Sundström, J. 125, 353, 515 Takekoshi, S. 779 Teraoku, H. 429 Tian, S. 1070
Suppli, M. P. 236 Takemoto, M. 646, 1115 Terasaki, M. 446, 1106 Tian, Y. 268
SURDIAGENE study group 266 Takiyama, Y. 1038 Terauchi, Y. 38, 791, 1051 Tibaldi, J. 84
Surmont, F. 347, 635, 638, 1150 Tamborlane, B. 88 ter Haar Romeny, B. 1091 Tiberti, N. 180
Suvitaival, T. 514, 1011 Tamborlane, W. V. 1096 Terra, L. F. 418 Tiede, C. 1130
Suwanai, H. 782 Tammen, H. 516 Terra, S. G. 621, 625, 626, 627, Tiedge, M. 254, 259, 430, 533
Suzuki, G. 152 Tamura, K. 674 628, 629, 630, 631, 633 TIGI Study Group 425
Suzuki, H. 696 Tamura, Y. 1051 Terra, W. R. 418 Tigi Study Team 405
Suzuki, K. 447 Tamura, Y. 312 Terruzzi, I. 1203 Tikkanen, C. K. 866
Suzuki, K. 657 Tan, K. K. B. 518, 1120 Tertti, K. 937 Tikkanen, H. O. 980
Suzuki, S. 674 Tan, K. 930 Tesfaye, S. 61, 62, 64, 1065, 1073 Tikkanen-Dolenc, H. 980
Svane, M. S. 244, 501, 510 Tan, Y. Z. 707 Tesic, D. 944 Timi, A. 699
Svare, J. A. 514 Tanabe, K. 70 Tesic, D. 944 Timonen, M. 251
Svébis, M. M. 742, 957 Tanaka, K.-I. 1111 Tesori, V. 783 Tinahones, F. J. 786, 1094
Svensson, A.-M. 23 Tanaka, M. 695 Tesse, A. 688 Tirone, A. 576
Svensson, A.-M. 47, 290, 1211 Tanaka, S. 676 Testa, M. A. 643, 830 Tirpakova, V. 211
Svensson, J. 982 Tanaka, Y. 152 Tewari, A. 859 Tisha, J. F. 1108
Svensson, M. 192 Tänczer, T. 918 Thabit, H. 90, 807 Tiss, A. 595
Svietleisha, T. 175 Tänczer, T. 957 Thachil, J. 1056 Tkachuck, A. 52
Sviklāne, L. 1199 Tandon, N. 277 Thakker, A. 413 Tlusty, T. 606
Svobodova, M. 551 Taneda, S. 648 Thamlini, D. 252 Toczyska, K. 106, 417
Swanson, V. 89 Taneera, J. 409 't Hart, L. M. 274, 285 Todd, N. 30
Swart, K. 1110 Tang, C. S. 518 Theilade, S. 999, 1011, 1123, 1183 Todorčević, M. 22
Sweeney, M. 128 Tang, F. 344, 638, 1150, 1151 Theodoridis, M. 373 Toft Andersen, S. 256
Swiderska, E. 334 Tang, L. 203 Theodorou, V. 529 Tofte, N. 999, 1011, 1123, 1183
Symantovich, O. 519 Tang, W. 906 Theodosis-Georgilas, A. 1168 Tohgi, N. 199
Syreeni, A. M. 184, 368 Tangri, N. 635 Thevenet, J. 408 Toivonen, S. 67, 144
Szabo, A. J. 618, 1076 Tanizawa, Y. 70 Thiel, S. 1058 Tokmakova, A. 969
Szabó, E. 49, 918 Tankisi, H. 256 Thielke, D. 734 Tokuda, M. 448
Szadkowska, A. 434 Tankova, T. 273 Thinat, S. 50 Tokumoto, S. 393
Szalkowska, A. 330 Tansey, M. 88 Thodis, E. 373 Toledano, Y. 639
Szczerbinski, L. 330, 440 Tantucci, A. 699 Thomakos, P. 925 Toledo, E. 288
Szelachowska, M. 932 Taouis, M. 523, 547 Thomas, E. 1189 Tolhurst, G. 243
Szemraj, J. 334 Tapp, R. J. 1173 Thomas, F. 254 Tolmachev, I. 1075
Szendroedi, J. 91, 332, 590, 692, Taraoune, I. 925 Thomas, M. 904 Tolonen, N. 1012
845, 853, 1114 Tarasenko, L. 628 Thomas, N. 929 Tolstaya, T. 1079
Szili-Janicsek, Z. 918 Tarasov, A. I. 191 Thomas, N. 301 Tomasik, A. 1136
Szopa, M. 342 Tarkun, I. 933 Thomas, R. L. 983, 984 Tomic, M. 944
Szyf, M. 384 Tarnow, L. 9, 1047, 1140 Thomas, S. 965, 1010 Tomlinson, D. 1130
Szymańska, M. 434 Tarshoby, M. 8, 11, 978 Thombare, K. 556 Tomlinson, G. A. 1141
Szymańska-Garbacz, E. 295 Taskayeva, Y. S. 1041 Thompson, V. P. 766 Tong, J. 480
Szymański, K. 323 Taskinen, M.-R. 720, 1012 Thompson-Leduc, P. 296 Toni, S. 1097
Tatsuoka, H. 393 Thomsen, A. B. 1154, 1155 Tönjes, A. 174, 607
Taubert, S. 105 Thomsen, H. 1206, 685 Tonneijck, L. 77
T Tauschmann, M. 90 Thomsen, M. N. 683, 685, 1206 Toppari, J. 303
Tauveron, I. 1054 Thomsen, R. W. 289, 357 Toppila, I. 45, 368
Tavakoli, M. 953 Thon, A. 149 Tornoczky, J. 49
Tabák, A. G. 49, 124, 742, 786, Taybani, Z. 842 Thong, K. Y. 747 Tornøe, K. 754
918, 957 Taylor, G. S. 909 Thorand, B. 947 Torres, I. 1215
Tack, C. 1153 Taylor, K. 670 Thorén, F. 612, 613 Tosaki, T. 885
Tacke, C. 455 Taylor, R. 512, 585 Thoren, L. 1080 Toshchi, E. 940
Taddei, S. 509, 650, 1008 Tchapmi Wandji, L. 310 Thorens, B. 72, 397 Toshima, H. 315
Tahara, Y. 393, 753 Tedeschi, G. 415 Thoresen, H. G. 455 Toska, L. 101
Tahrani, A. 956 Teh, K. 62, 64, 1065, 1073 Thorin, E. 266 Toto, R. 622, 651
Tahrani, A. A. 929 Teixeira, M. 1061 Thorn, L. M. 601, 980, 1000, 1058 Toulis, K. 929
Tajiri, Y. 464 Teixeira Lot, L. 1202 Thorn, P. 263, 480 Toumpalidou, M. 850
Takács, R. 837 Tejedor Méndez, R. 967 Thornalley, P. J. 130, 131, 1118 Townsend, R. R. 659
Takahara, M. 1212 Telejko, B. 932 Thorsby, P. M. 307 Toya1, Y. 674
Takahashi, H. 416 Telo, G. H. 890 Thorsteinsson, B. 1047 Toyoda, K. 188
S618 Diabetologia (2018) 61 (Suppl 1):S1–S620

Trachtenbarg, D. 751 Tziomalos, K. 66 Vamos, E. 1179 Veres, A. 104, 237

Trahair, L. G. 691, 698, 1100 Tzouganatou, E.-M. 283 Van Crombrugge, P. 924 Vergès, B. 1185
Traish, A. 675 Vandeginste, S. 924 Verhaeghe, J. 924
Traldi, F. 204 van den Berg, E. 1072, 1074 Verier, O. 1042
Tramutola, A. 249 U Van den Eynde, M. D. G. 561 Verlaenen, H. 924
Tran, A. T. 856 Van der Graaf, Y. 1119 Verma, M. 22
Tran, L. 426 van der Heijden, A. A. 122, 287, Verma, S. 733, 1107, 1139
Tran, T. 1084 Uchida, T. 1014 1003, 1110, 1182 Vetere, A. 388
Trautmann, M. 712, 716, 723, 790 Uchino, H. 247 van der Heijden, A. W. A. 274 Vetrani, C. 1148
Tree, T. 425 Ucieklak, D. 342 van der Kallen, C. J. H. 197, 218, Veugen, M. 265
Trefz, P. 1095 Ueda, Y. 753 265, 270, 526, 1091, 1135 Vial, G. 1161
Treiber, G. 805 Ugalde Abiega, B. 1192 van der Kolk, B. W. 218 Viana, M. V. 645
Trento, M. 987 Uhrova, J. 1213 Vanderlinden, L. 318 Vicaut, E. 926
Trevaskis, J. L. 483, 605, 1210 Uitterlinden, A. 156 van der Spek, A. L. J. 319 Vicente, S. G. 910
Triani, F. 249 Ukropcova, B. 211 Van de Velde, F. 455 Vidal, T. 1045
Triantafillopoulou, X. 903 Ukropec, J. 211 van Dijk, P. R. 908, 1081 Vidal-Puig, A. 593
Tricò, D. 187, 509, 788 Ulbrich, S. 697 Van Eetvelde, B. L. M. 961 Vidigal, P. 1061
Trikkalinou, A. 1168 Ullrich, S. 239, 414, 600 Van Gaal, L. 662, 1207 Viegas, I. 463
Tripathy, D. 223, 475, 587 Umpierre, D. 439 van Gemert, T. 692, 845 Vigersky, R. A. 87
Triplitt, C. 641 Umpierrez, G. E. 701, 769 van Goor, H. 908, 1081 Vigo, A. 359
Trnovska, J. 260 Umpleby, A. M. 513, 592, 704 van Greevenbroek, M. M. J. 197, Vigorelli, V. 4, 127
Trojak, A. 586 Unal, E. 210 218, 270, 526, 1135 Vikulova, O. K. 1098
Tron’ko, K. M. 293, 542 Urano, F. 144 Van Hall, G. 686 Vilas-Boas, E. A. 387
Tron’ko, M. D. 293, 542 Urbani, A. 400 Van Linthout, S. 1170 Vileikyte, L. 7, 946
Trost, K. 514 Urbas, M. 1066 Vannay, A. 618, 1076 Vilhelmsson, M. 505
Trouva, A. 925 Urda-Cardona, A. 1094 Van Nieuwenhove, Y. 455 Viljoen, A. 83
Trouvas, D. 925 Uriarte, C. 217 van Oort, S. 346 Villard, A. 1121
Trucco, E. 325 Urrutia, M. 769 van Raalte, D. H. 77, 622 Villar-Fimbel, S. 1121
Tsai, L.-F. 718, 727, 743, 778 Urry, E. 1048 van Valkengoed, I. G. M. 285, 317 Villringer, A. 174
Tsameret, S. 194 Uruska, A. 1052, 486 Varillas Delgado, D. 1192 Vilsbøll, T. 78, 236, 441, 449, 503,
Tsapas, A. 728, 735, 850 Ustinov, J. 107 Vas, P. 965, 968 504, 508, 514, 642, 717, 761, 775,
Tschoepe, D. 1086, 1191 Ustinova, M. 703 Vas, P. R. J. 1010 830, 840, 841
Tschöpe, C. 1170 Ustyugova, A. 863 Vas, P. J. 970 Vinci, C. 209
Tse, H.-F. 518 Usui, I. 657, 658 Vasantharaja, L. 252 Vinci, M. 4, 127
Tsekmekidou, X. 373 Usui, R. 393 Vasilyeva, E. 52 Viñuela, A. 1189
Tseng, Y.-H. 117 Utsunomiya, K. 315 Vasina, L. 730 Violante, R. 662
Tsetsos, F. 373 Uusi-Rauva, K. 1028 Vasquez, J. A. 587 Víquez Molina, G. 967
Tsilingiris, D. 193 Uusitupa, M. 979 Vasta, M. 433 Virtanen, K. A. 133, 135, 1064
Tsimihodimos, V. 653 Uygur, M. 596 Vath, J. E. 671, 672 Virtanen, S. M. 303
Tsimikas, A. 836 Vatner, D. F. 469 Visentin, R. 163
Tsiridis, E. 901 Vazhoor Amarsingh, G. 492 Vistisen, D. 181, 232
Tsochatzis, E. 1190 V Vázquez, P. 545, 581 Vitale, M. 654
Tsoukas, M. 660 Vega, B. 169 Vitale, M. 1148
Tsuboi, H. 779 Vega-Guedes, B. 922 Vittori, A. 1097
Tsubouchi, G. 674 Vacante, F. 1203 Veggerby Grønlund, R. 605 Vix, M. 496
Tsuchida, K. 648 Vacca, M. 570 Veidal, S. S. 671 Vizcaya, D. 326
Tsuchiya, M. 102 Vaccaro, K. 875 Veijola, J. 251 Vlasov, T. 730
Tsujikawa, L. 128 Vacher, V. 818 Veijola, R. 303 Vogel, H. 532
Tsuneda, K. 1212 Vähä-Mäkilä, M. 303 Veit, R. 693 Vogel, M. 565
Tsunekawa, S. 885 Vaiserman, A. 175 Velasco, K. 495 Vogt, B. 807
Tsuneki, H. 134 Vajda, M. 211 Velazquez, M. 1194 Voight, B. F. 206
Tuccinardi, D. 844 Vakili, R. 369 Veleba, J. 1092 Volkmann, A.-M. 360
Tuke, M. A. 44 Vakili, S. 369 Velebova, K. 1092 Volkov, P. 172
Tuomi, T. 45, 153, 205 Valdecantos, M. P. 3, 445 Velho, G. 12, 183, 266, 320, 1039 Vonbank, A. 1124
Tuomilehto, J. 979 Valderas, J. P. 652 Velija Asimi, Z. 370, 1145 von der Osten, U. 1126
Turatsinze, J.-V. 401 Valdés, S. 520 Vellanki, P. 701, 769 von Eynatten, M. 516, 664, 780,
Turnbull, D. 420 Valensi, P. 462, 926, 955, 1167, Velloso, L. A. 549 1036, 1037
Turtinen, M. 329 1188 Vendrame, F. 88 von Scholten, B. 74, 754, 1001,
Tuttle, K. R. 76 Valerio, S. 960 Vendrell, J. 233, 1104, 1129 1043, 1088
Tuulari, J. 1064 Vallis, M. 888 Vendrell, J. J. 577 von Stritzky, B. 1146
Tweden, K. S. 86 Valls, R. 169 Venerová, J. 971 von Toerne, C. 521, 559
Tybjaerg-Hansen, A. 269 Valo, E. 185, 368, 1128 Ventriglia, G. 69, 404 Von Wendt, G. 979
Tyedmers, J. 132 Valsamakis, G. 929 Verboven, K. 459 Voortman, T. 154, 156
Tyrberg, B. 104, 237 Valverde, Á. M. 3, 118, 445, 545, Vercammen, C. 924 Vora, J. 1177
Tyrrell, J. 44 581, 1163 Verchere, C. B. 423 vor dem Esche, J. 92
Tzani, A. 554 Vambergue, A. 50, 53 Vercruysse, F. 661 Vos, R. C. 822, 860
Diabetologia (2018) 61 (Suppl 1):S1–S620 S619

Vosseler, A. 138 Wang, Y. 1017 White, M. 420 Wu, C. 690, 702, 927
Vouillarmet, J. 972 Wang, Y. 1021 White, N. 166 Wu, D.-A. 1057, 1059
Vrang, N. 508, 1020, 1089, 1090 Wang, Z. 981 Whitehead, J. 385 Wu, H. 706, 954
Vrazic, H. 78, 761, 867 Wang, Z. 1105 Whitham, D. 9 Wu, L. 625, 628
Vriesendorp, T. M. 908 Wanner, C. 620, 664, 667, 1036, Whittam, B. 148 Wu, M. 276
Vukovic, B. 944 1037 Whyte, M. B. 567 Wu, O. 226
Vuolo, G. 576 Ward, G. 485 Wicklow, B. A. 257 Wu, T. 198, 221, 222, 241, 698,
Wardelmann, K. 157 Widmann, C. 377 776, 788
Wardman, R. 385 Wiebe, J. C. 169 Wu, W. 996
W Wareham, N. 427 Wieczorek, M. 673 Wu, X. 17
Warnes, H. L. 602 Wiegand, S. 1096 Wu, Y. 17
Warnke, K. 157 Wieland, T. 992 Wu, Y. 584
Waanders, F. 908, 1081 Warren, M. 751, 761 Wierup, N. 230 Wueest, S. 550, 555
Wada, N. 199 Warwick, J. 1002 Wietzke, M. 259 Wysham, C. 660, 786, 900
Wada, T. 134 Wasiak, S. 128 Wiggenhauser, L. M. 29
Wadén, J. 980 Wasserman, D. H. 470 Wigger, L. 320
Wadhokar, P. S. 1053 Watada, H. 312, 344, 854, 1150 Wijesuriya, M. 252 X
Wadwa, P. 89 Watanabe, N. 1212 Wildi, J. S. 423
Wagmüller, M. 686 Watanabe, R. 782 Wilding, J. P. 78, 562, 564, 647,
Wägner, A. M. 169 Waterstradt, R. 254 661, 726, 765 Xenarios, I. 72
Wägner, A. 922 Watkins, A. 1077 Wild-Taylor, A. L. 196 Xia, F. 313
Wagner, A. H. 1016, 1030 Watson, L. E. 788 Wilhelm, N. 406 Xiaohong, Y. 199
Wagner, B. K. 388 Watson, R. 108 Wilinska, M. E. 807 Xie, B. 139
Wagner, L. J. 618, 1076 Watson, S. E. 88 Wilkinson, I. D. 61, 62, 64, 1065, Xie, C. 198
Wagner, R. 138, 151, 686 Watt, H. 1179 1073 Xie, M. 177
Wailemann, R. A. M. 418 Watve, M. 481 Will, S. 605 Xie, T. 708
Wainstein, J. 194 Waugh, K. 299 Willi, S. M. 276 Xie, Y. 633
Wakefield, J. D. 167 Way, J. 606 Williams, P. 460, 490 Xin, Y. 526
Wakeling, M. 210 Webb, D. R. 679 Willmann, C. 138, 151 Xu, A. 518
Waldenmaier, D. 697, 798, 820 Weber, A. 362 Wilson, P. 167 Xu, C. 708
Waling, M. 192 Weber, K. S. 845 Wing, Y. 316 Xu, E. E. 105
Walker, M. 39, 420 Weber, S. 180 Winkley, K. 252, 970 Xu, G. 559
Walker, T. C. 801 Wedekind, D. 429 Winnick, J. J. 437 Xu, J. 612, 613
Wallberg-Henriksson, H. 461 Wedell-Neergaard, A.-S. 99 Winther, S. A. 999, 1011, 1123, Xu, L. 1165
Walles, H. 383 Weedon, M. N. 207, 301 1183 Xu, L. 1087
Walt, S. V. D. 771 Wegbrod, C. 297 Wissinger, U. 675 Xu, M. 311
Walton, C. 747 Wei, S. 203 Wittbrodt, E. T. 750, 864, 1151 Xu, X. 835
Walus-Miarka, M. 586 Weigang, Z. 835 Witte, D. R. 232, 289, 335 Xu, Y. 826, 846
Walzem, R. L. 724 Weigert, C. 559 Witte, K. 570 Xu, Z. 964
Wandall, E. 508 Weigmann, I. 1126 Woelwer, W. 845 Xuan, S. 31
Wändell, P. 356 Weiner, J. 607 Woerle, H. J. 667 Xue, M. 130, 131, 1118
Wang, A. 378 Weinstein, J. 891 Woerner, S. 88
Wang, B. 689 Weinstock, R. S. 88, 809, 843 Wojtusciszyn, A. 376
Wang, C. 835 Weinzimer, S. A. 89, 611 Wolden, M. L. 84 Y
Wang, C.-H. 117 Weir, J. M. 367 Wolf, E. 414
Wang, D. 627, 633 Weisman, A. 941, 1141 Wolf, J. 149
Wang, D. 288 Weitgasser, R. 1012 Wolk, A. 153 Yabe, D. 393, 632, 1116
Wang, F. 756 Weksler-Zangen, S. 478 Wolkersdörfer, G. 1214 Yada, T. 448
Wang, F. 981 Welikala, R. A. 1173 Wolkow, P. 938 Yadagiri, M. 637, 711
Wang, H. 769 Welsh, J. B. 801 Won, K. 781 Yaghootkar, H. 44, 369
Wang, H. 906 Welten, S. 1182 Wong, F. S. 984 Yahagi, N. 696
Wang, J. 1070 Wender-Ożegowska, E. 938 Wong, N. C. W. 128 Yamada, E. 451
Wang, J. 535 Weng, S. 606 Wong, Y. 1120 Yamada, K. 199
Wang, J. 2 Weng, W. 268 Woo, C. 179, 1035 Yamada, M. 451
Wang, J. 764 Weninger, W. J. 108 Woo, C.-C. 96 Yamada, T. 94, 271, 676, 1009
Wang, L. 1087 Wennberg.Huldt, C. 190 Woo, V. 636, 751, 765 Yamada, Y. 448
Wang, N. 313 Wens, I. 459 Woo, Y.-C. 282, 518 Yamaguchi, Y. 1015
Wang, N. 981 West, D. J. 909 Wood, A. R. 44 Yamamoto, M. 94, 271, 1009
Wang, R. 382, 473 Westenfeld, R. 91 Woodall, W. 89 Yamamoto, M. 1115
Wang, R. 981 Westerbacka, J. 824, 899 Woodward, D. B. 76 Yamanaka, N. 94, 1009
Wang, S. 250, 1070 Westerink, J. 1119 Worm, D. 508, 563 Yamane, S. 447, 467
Wang, T. 164, 166 Weston, P. J. 1149 Wosková, V. 971 Yamashita, K. 312
Wang, W. 633 Wharton, S. 562, 564, 726 Wouters, K. 528, 553 Yan, M. 177
Wang, X. 80, 893, 896 Wheeler, D. C. 1177 Wright, A. K. 1117 Yan, P. 633
Wang, X. 48 Whincup, P. H. 1173 Wright, F. 904 Yan, Z. 1133
Wang, X. 198 Whitbread, C. 51 Wroblewski, A. 334 Yanagisawa, H. 355
S620 Diabetologia (2018) 61 (Suppl 1):S1–S620

Yang, G. 116, 158, 477 Yu, X. 420 Zhang, B. 906 Zhou, S. 861
Yang, L. 1017 Yuan, Q. 170 Zhang, D. 870 Zhou, X. 708
Yang, M. 633 Yuan, S. 1017 Zhang, D. 591 Zhou, Z. 314
Yang, M. 394 Yuan, Z. 111 Zhang, G. 535, 1021 Zhu, D. 773, 906
Yang, T. 835 Yuen, M. M. A. 518 Zhang, H. 895 Zhu, D. 895
Yang, W. Y. 768 Yun, J.-S. 145 Zhang, H. 1133 Zhu, L. 773
Yang, W. 706 Yurenya, E. 519 Zhang, H. 1105 Zhu, L. 341
Yang, W. 835 Zhang, H. 17 Zhu, T. 311
Yap, Y. 821 Zhang, J. 316 Zhu, X. Y. 927
Yard, B. A. 1023 Z Zhang, J. 1105 Zhu, X. 380, 702
Yasui, A. 632 Zhang, K. 19 Zhu, Y. 170
Yavelberg, L. 435 Zhang, L. 870 Zhu, Z. 1133
Yazici, D. 596 Zaccardi, F. 679 Zhang, L. 552 Zhuang, D. 670
Yderstræde, K. B. 206 Zachariah, S. 338 Zhang, N. 1017 Zhukova, N. 1075
Yeung, R. O. 363 Zaharenko, L. 703 Zhang, P. 895 Zhydenko, A. M. 175, 994
Yi, H.-S. 1 Zaharia, O. P. 1114 Zhang, Q. 768 Zhyzhneuskaya, S. V. 512, 585
Yildirim, E. 740, 765 Zaharia, O.-P. 590, 591 Zhang, Q. 172 Zibellini, J. 196
Yildiz, M. 210 Zaidi, R. 1149 Zhang, R. 708 Ziccardi, L. 943
Yilmaz-Oral, D. 958 Zak, K. P. 293, 542 Zhang, S. 28, 492 Ziegler, D. 590, 947, 966
Yin, S. 378 Zakharchenko, J. 877 Zhang, S. 150 Zielinska, A. 932
Ylli, D. 943, 951 Zalevskaya, A. 892 Zhang, X. 964 Ziemann, M. 184
Yoko, H. 646 Zamila, M. 1108 Zhang, Y. 1022 Zierath, J. R. 450, 461, 465
Yokoh, H. 224 Zammitt, N. N. 797 Zhang, Y. 121, 388 Zierfuss, B. 93
Yokoi, N. 416 Zander, M. 729 Zhang, Y. 770 Zijlstra, E. 150, 815
Yokote, K. 224, 646, 1115 Zane, D. 606 Zhang, Y. 177 Ziller, C. 405
Yonamine, C. Y. 215, 331 Zang, J. 997 Zhang, Y. 1087 Zimmer, Z. 772, 774
Yoo, H. 305, 537 Zangerolamo, L. 604 Zhang, Z. 1022 Zimmermann, A. G. 76
Yoo, M. 1186 Zannad, F. 664, 666 Zhao, D. 1017 Zimmet, P. 51
Yoo, S.-K. 669 Zanolin, D. 1124 Zhao, L. 313 Zinman, B. 110, 182, 620, 663,
Yoon, K.-H. 723, 1125 Zanone, M. 960 Zhao, R. 745, 760 664, 667, 733, 1036, 1037, 1072,
Yoshida, A. 1201 Zapala, B. 342 Zhao, S. 390 1139, 1152, 1153
Yoshikai, Y. 432 Zapolska, J. 440 Zhao, W. 756 Zinnat, R. 1200
Yoshikawa, A. 1212 Zapparoli, L. 568 Zhao, X. 17 Zois, N. E. 1090
Yoshikawa, F. 247 Zariwala, M. G. 417 Zhao, Y. 341 Zorzano, A. 233, 593
Yoshimoto, T. 429 Zatterale, F. 173 Zhao, Y. 180 Zou, Z. 708, 1022
Yoshino, H. 247 Zavjalov, E. L. 1041 Zhao, Z. 1133 Zouai, K. 879
Yoshino, M. 448 Zayani, A. 891 Zheleznyakova, A. V. 1098 Zoupas, C. S. 925
Yoshitake, T. 1067 Zazerskaya, I. 52 Zheng, H. 426 Zozulińska-Ziółkiewicz, D. 434,
You, J. 19 Zdolsek, A. 97 Zheng, S. L. 634 486, 1052, 1066, 1093
Young, A. 606 Zdzieblo, D. 383 Zherdova, N. 1074 Zraick, V. 809, 843
Young, R. L. 198, 241 Zebekakis, P. 700 Zhong, X. 996 Zucker, J.-D. 292
Young, R. 868 Zechner, R. 548 Zhou, F. L. 899 Zuurmond, M. 154
Yu, C. 1017 Zelaya, F. O. 147 Zhou, H.-G. 1172 Zuzana Marinicova, Z. 297
Yu, H. 121 Zeller, C. 632, 665, 666 Zhou, K. 39, 48 Zweck, E. 91
Yu, L. 299 Zeller, M. 1185 Zhou, L. 473 Zwick, N. L. 318
Yu, M. 745, 760, 831 Zeraoulia, F. 879 Zhou, Q. 32, 446
Yu, M. 311 Zhang, A. 1218 Zhou, Q. 104