Osteoarthritis and Cartilage 30 (2022) 237e248

Review

Osteoarthritis in year 2021: biochemical markers

Y. Henrotin y z *
y musculoSKeletal Innovative research Lab (mSKIL), Institute of Pathology, Level 5, CHU Sart-Tilman, Center for Interdisciplinary Research on Medicines
(CIRM), Department of Motricity Sciences, University of Li

ege, Belgium
z Department of Physical Therapy and Rehabilitation, Princess Paola Hospital, Vivalia, Marche-en-Famenne, Belgium

a r t i c l e i n f o s u m m a r y

Article history: Objective: To summarize recent scientific advances in protein-derived soluble biomarkers of
Received 31 July 2021 osteoarthritis.
Accepted 1 November 2021 Design: A systematic search on the PubMed electronic database of clinical studies on protein-derived
soluble biochemical markers of osteoarthritis in humans that were published between January 1st 2020
Keywords: and March 31th 2021. The studies were selected on the basis of objective criteria and summarized in a
Biomarkers
table. Then they were described in a narrative review.
Protein
Results: Out of 1971 publications, 48 fulfilled all selection criteria and 16 were selected by the author for
Diagnosis
Osteoarthritis
the narrative review. The papers were classified according their clinical significance as defined in the
Prognosis BIPEDS classification. Two papers investigated the “burden of disease”, two were dedicated to “inves-
Review tigative biomarkers”, four papers question the “prognosis”, three the “efficacy of treatment” and five the
“diagnosis and phenotyping” value of protein-derived biomarkers.
Conclusions: Currently, biomarkers research is focused on their use as tools to identify molecular
endotypes and clinical phenotypes and to facilitate patient screening and monitoring in clinical trials.
This approach should allow a more targeted management of patients suffering from osteoarthritis.
© 2021 The Author(s). Published by Elsevier Ltd on behalf of Osteoarthritis Research Society
International. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).

Introduction pathophysiology obscures identification of proper treatment tar-

gets. This is complicated by the increasing knowledge that the
Although it is the most frequent musculoskeletal diseases, that physiopathological mechanisms driving the OA process differ be-
it is linked to a major handicap, that it contributes to cardiovascular tween patients, joints localization and disease stage3.
morbidity, and that it is very expensive for society, osteoarthritis A challenge for researchers is to identify markers that are linked
(OA) still suffers from a great image deficit and this, due to the to OA physiopathology, which allow early diagnosis of the disease
notable absence of therapeutic advances1. This disappointing result and predict its course, which allow among OA patients to select
likely has multiple origins. The typically slow and heterogeneous who will be the progressors and predict short-term structural
OA course makes trials often fall short in terms of size and length changes or pain/symptoms evolution and finally predict the effi-
for demonstrating treatment efficacy. This issue is further aggra- cacy and monitor the effects of a treatment at an individual level4.
vated by the use of relatively insensitive outcome measures (pa- Recently, a new challenge is trying to be taken up by scientists and
tient-reported outcome measurements), pain and radiographic health professionals, which consists in identifying and defining the
joint space changes (X-ray), required by regulatory agencies for a phenotypes of OA and the endotypes associated with it. Initially, six
drug to be certified as a Disease Modifying Osteoarthritis Drug clinical phenotypes and nine endotypes of knee OA have been
(DMOAD)2. Finally, an incomplete understanding of the OA described (Fig. 1)5e7. A challenge for the future will be to identify
the clinical phenotypes and clearly define their constituent mo-
lecular endotypes. It is more than likely that the future markers of
prognosis or efficacy of a treatment will be part of these molecular
* Address correspondence and reprint requests to: Y. Henrotin, musculoSKeletal
Innovative research Lab (mSKIL), Institute of Pathology, Level 5, CHU Sart-Tilman, pathways.
Center for Interdisciplinary Research on Medicines (CIRM), Department of Motricity To help them achieve these objectives, researchers have at their

ge, Belgium.
Sciences, University of Lie disposal clinical, biological or even medical imaging markers.
E-mail address: yhenrotin@uliege.be.

https://doi.org/10.1016/j.joca.2021.11.001
1063-4584/© 2021 The Author(s). Published by Elsevier Ltd on behalf of Osteoarthritis Research Society International. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
238 Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248

Fig. 1 OA YEAR IN REVIEW 2021 Osteoarthritis and Cartilage

Flower representation of OA phenotypes, endotypes and molecular endotypes (inspired from v6). CRPM: C-Reactive Protein Metabolite; TNF:
Tumor Necrosis Factor; IL: InterLeukin; VEGF: Vascular Endothelial Growth Factor; MMP: Matrix MetalloProteinase; TIMP: Tissue Inhibitor of
Metalloprotease; sICAM: serum InterCellular Adhesion Molecule; sVCAM: serum Vascular Cell Adhesion Molecule; MCP: Monocyte Chemo-
attractant Protein; SIRT: sirtuine.

Among all these markers, soluble biochemical markers have OR "arthrosis"[TIAB]) AND ("blood"[TIAB] OR "serum"[TIAB] OR
benefited from particular interest because they can be directly "plasma"[TIAB] OR "urine"[TIAB] OR "urinary"[TIAB] OR “synovial
detected in biological fluids like urine (u), plasma (p), serum (s) or fluid”[TIAB] OR “saliva” [TIAB] OR “SF”[TIAB] OR biomarker*[TIAB]
synovial fluids (sf). In an effort to facilitate research on biomarkers, OR marker*[TIAB]). Our search of terms was restricted to title (TI)
a working group financed by the NHI proposed a goal-based clas- or abstract (AB) of the articles.
sification, which was represented by the acronym BIPED to connote The identified publications were then stepwise selected by the
the five categories of markers: Burden of disease, Investigative, author. First, publications were required to be in English. Second,
Prognostic, Efficacy of intervention, and Diagnostic8. This classifi- publications had to present data on humans with OA. Animal or in
cation scheme helps to provide a common language and structure vitro studies were not selected for further review. Third, publication
with which to communicate knowledge and advances related to OA had to present original research work or clinical trials. Fourth, studies
biomarkers for both clinical and research applications. should present data on soluble protein-derived biochemical markers
This year ‘review aimed to summarize publication with original in blood, urine, saliva and/or synovial fluid. The markers should
data on protein-derived soluble biochemical markers and showed relate to matrix metabolism, inflammation and/or other pathobio-
them according the BIPEDS classification to allow a continuity with logical processes within joints. Genetic markers were considered
previous review papers. outside the scope of this review. Papers discussed in this review were
then selected based on the author' evaluation of its content, tabu-
lated and finally discussed following the BIPEDS classification.
Method

This narrative review of the literature was performed by Results

searching in titles and abstracts in the electronic PubMed database
for publications on clinical studies into protein-derived soluble PubMed search, paper selection and data extraction
biochemical markers for OA in humans that were published be-
tween 01-01-2020 and 01-04-2021. The following search equation The selection process of potentially relevant publications is
was used for identifying potentially relevant publications in shown in Fig. 2. Sixteen publications were selected by the author to
PubMed: ("osteoarthritis"[MeSH Terms] OR "osteoarthritis" [TIAB] be discussed in this review. Selected papers were shown in Table I.
 Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248 239

The large majority of studies included patients with knee OA (n ¼ additional predictive power over, and above established predictors
12), two with both knee and hip OA patients (n ¼ 2), one with of OA such as age, gender, Body Mass Index (BMI) and race. This
lumbar spine OA (n ¼ 1) and finally one with Temporo Mandibular suggest that use this cluster of four biomarkers may be considered
Joint (TMJ, n ¼ 1). Biomarkers were measured in serum (s, n ¼ 11), clinically useful and should be considered first in clinical trials as
urine (u, n ¼ 3), synovial fluid (sf, n ¼ 3), plasma (p, n ¼ 1) and saliva exploratory endpoint.
(n ¼ 1) using mainly immunoassays. One study investigated the In a cross-sectional study of 145 participants with knee pain,
secretome of joint tissues. Two studies investigated the association biomarkers of innate immunity were associated with MRI features
between biomarker levels and the burden of disease evaluated by of knee OA10. More precisely, serum Lipopolysaccharide Binding
X-ray or MRI, four the prognosis value, three the efficacy of treat- Protein (LBP) was associated with meniscal extrusion, and synovial
ment, and five diagnosis and molecular endotype. fluid Cluster of Differentiation 14 (CD14) was associated with
effusion. Serum Interleukin (IL)-6 was associated with osteophytes,
Burden of disease synovitis, effusion, and meniscus extrusion, while synovial fluid IL-
6 was only associated with effusion. Similarly, serum Tumor Ne-
By definition, burden of disease markers are biomarkers asso- crosis Factor alpha (TNFa) was statistically significantly associated
ciated with the severity or extent of disease. Most of the existing with osteophytes, cartilage loss, synovitis, and effusion. None of the
biomarkers have been demonstrated to be associated with knee biomarkers related to innate immunity was associated with
standard X-ray or Magnetic Resonance Imaging (MRI) features of symptoms or radiographic gradings, excepted serum IL-6 which
OA severity. Using a subset of 600 patients from Osteoarthritis was negatively associated with WOMAC function. This study
Initiative (OAI) cohort, a longitudinal cohort study sponsored by the highlighted the role of Lipopolysaccharide (LPS)/LBP pathways in
National Institute of Health (NIH), Liem et al.9 demonstrated that of association with the innate immunity in the pathogenesis of knee
the 19 currently available biomarkers only 4 (s-Coll2-1NO2, s- OA. LPS, being a Pathogen-Associated Molecular Pattern (PAMP)
CS846, s-COMP and u-CTXII) were consistently associated with enhanced in blood of OA patients secondary to increased perme-
established radiographic and/or clinical features of knee OA. These ability of the gut, binds to LBP. The CD14 biomarker is predomi-
biomarkers were independent of one another and provided nantly found on activated macrophages and serves as a receptor for

Pubmed search
From 01/01/2020 to 01/04/2021.
Total 1971 hits

43 excluded
Langage: English Yes/No
Remaining hits : 1928
1038 excluded

Human: Yes/No
Remaining hits: 890
788 excluded
Clinical trial: yes/No
Remaining hits: 102
54 excluded

Protein-deribed biomarkers: Yes/no

Remaining hits: 48
32 excluded

Papers selected by the author

Remaining hits: 16

Fig. 2 OA YEAR IN REVIEW 2021 Osteoarthritis and Cartilage

Flow diagram of the stepwise selection process of relevant paper.

 240
Author Year Joint Fluid Population Parameters Results Conclusion

Burden of disease
Liem 2020 Knee Serum Urine 600 OA patients from Biochemical: sC1, 2C, sC2C, Out of the 19 biomarkers Out of the 19 biomarkers
the OAI data set. sCPII, sPIIANP, sCol2-1NO2, only 4 (serum Coll2-1 NO2, analyzed u-CTXII, s-COMP,
sCS846, MMP-3, sCTX-I, CS846, COMP and urinary CS846 and Coll2-1 NO2
sCOMP, sHA, sNTX-I, uCTX- CTXII) were consistently appear to be the most
II, uC1, 2C, uC2C, uNTX-I, associated with established promising biomarkers for
uCTX-1a, uCTX-Ib, uColl2- radiographic and/or clinical OA as they relate to both
1NO2, creatinine features of OA. structural damage as well
Clinical: WOMAC and KOOS Serum Coll2-1 NO2, COMP as symptoms in knee
scores. and urinary CTXII were OA.
Imaging: JSN, JSW, associated with baseline
osteophytes and Kellgren K&L grade.
and Lawrence (KL) score on uCTXII had the strongest
standard X-ray. and most consistent
associations

Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248

with radiographic and
clinical features of OA.
Rajandran 2020 Knee Serum 139 patients with early Biochemical: s and sf LBP, In earlier KOA, sLBP was Biomarkers of activated
Synovial fluid stage of OA CD14, TLR4, IL-6, IL-8, TNFa statistically significantly macrophages and synovial
20 patients with late Imaging: Boston Leeds associated with meniscal inflammation are
stage of OA. Osteoarthritis Knee MRI extrusion, sfCD14 was associated with early MRI
Score associated with effusion features of KOA.
and IL-6, IL-8, and TNFa
were associated
with most MRI features
In later stage of KOA, sfLBP
was associated with
effusion. sfCD14 was
associated with cartilage
loss and BML. In earlier
stage of KOA, the
proinflammatory
biomarkers IL-6, IL-8, and
TNFa were associated
with most MRI features
Investigative
Timur 2020 Knee Secretomes of cartilage, 24 patients with end- Biochemical: Liquid Proteomic analyses of OA vs A number tissue-
synovium, Hoffa's fat pad stage knee OA. chromatography tandem non-OA knee synovial fluid dependent protein release
and meniscus. mass spectrometry, revealed 70 proteins with a from intra-articular human
Synovial fluid. followed by label free relatively higher knee OA tissues had an
quantification. abundance and 264 osteoarthritis-specific
Immunoassays for SLPI, C8, proteins with a relatively abundance in knee synovial
CLU, FN1, RARRES2, lower abundance in OA fluid.
MATN3, MMP3 and TNC, synovial fluid. These proteins may serve as
IGF2, AHSG, FN1, CFB, KNG Of the 70 higher abundance novel candidates for OA
and C8. proteins, 23 were amongst biomarker development on
the most highly expressed a tissue specific basis.
in the secretomes of a
specific intra-articular
tissue measured.
Batshon 2020 Knee Serum 28 subjects including Biochemical: immunoassays An increase in sNT/CT SIRT1 sNT/CT SIRT1 ratio is
healthy donors, early for inactive N-terminal (NT) ratio in individuals with OA indicative of early-stage OA.
OA or late OA patients. polypeptide (75SIRT1) and compared with healthy
donors, mostly due to
 a C-terminal (CT) fragment higher NT values.
of SIRT-1. NT values are increased in
synovial fluid between
healthy donors and late
OA.
Prognosis
Arnold 2020 Knee Serum 636 subjects who Biochemical: GDF-15 using GDF-15 was inversely GDF-15 represents a potent
Hip underwent hip/knee electrochemiluminescence associated with walking predictor of decreased
arthroplasty immunoassay. distance. survival over >20 years in
All-cause of mortality: death The top quartile of GDF-15 OA patients.
was ascertained by demonstrated a 2.69-fold
obtaining the survival increased risk of dying.
status via the respective
residents' registration
office. Mortality was
assessed during follow-up
at 6, 12, and 60 months
after joint replacement and
again, in the years 2014,
2015, and 2019.

Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248

Rehm 2020 Knee Serum 679 OA subjects, Biochemical: NT-proBN and Baseline u-CTX-II was Elevated cardiac biomarker
Hip undergoing hip or knee high-sensitivity hs-cTnt associated with elevated concentrations predicted
replacement followed and hs-cTnI using risk of radiographic an increased risk of long-
over 20 years. immunoassays. progression in terms of term mortality and
both JSN and KL-grade. strongest for NT-proBNP
The top quartile of NT- and hs-cTnI.
proBNP was associated
with increased risk of
mortality.
Bihlet 2020 knee Urine 1,255 knee OA patients Biochemical: uCTX-II using A prediction-model Baseline u-CTX-II was
followed for 2 years. an immunoassay. incorporating age, sex, BMI, associated with risk of
Imaging: JSN and KL score u-CTX-II and KL-grade radiographic progression.
on standard X-ray. predicted TJR within the 2-
Clinical: total joint year period.
replacement. In the absence of baseline
radiographic OA severity, u-
CTX-II independently
contributed to prediction of
TJR.
Rogers-Soeder 2020 Knee Blood 987 subjects without Biochemical: Fasting RKOA were not associated DM and higher levels of
radiographic knee OA at glucose, and free insulin with baseline DM status nor biomarkers of abnormal
baseline and stratified levels and insulin resistance with levels of fasting glucose metabolism were
by BMI and DM status. (HOMA-IR) formula at glucose and HOMA-IR not associated with
baseline. overall and in men. In increased odds of incident
Imaging: Postero-anterior women, HOMA-IR was RKOA in overall and in men.
weight bearing knee inversely associated with
radiographs. A knee was odds of incident RKOA.
defined as having incident
tibiofemoral RKOA if there
was at least one follow-up
radiograph showing RKOA
(KL grade 2 or greater) or if
there had been a TKR at the
time of the visit. The status
of every knee was followed
from baseline to 84 months.
Efficacy of treatment
Conaghan 2021 Knee Serum 244 participants with Biochemical: s-CTX-I and u- Compared with baseline, MIV-711 did not
Urine primary knee OA which CTX-II using immunoassay. levels of s-CTX-I and u-CTX- demonstrate any beneficial

241
(continued on next page)
Table I (continued )

242
Author Year Joint Fluid Population Parameters Results Conclusion

received either MIV- Clinical: 26-week change in II were reduced by MIV-711 effects on OA knee pain in
711 100 (n ¼ 82) or NRS pain score. and levels of both this study but reduced bone
200 mg (n ¼ 81) daily or Imaging: change in MRI biomarkers returned to and cartilage MRI and
matched placebo bone area and cartilage baseline values after MIV- biochemical markers.
(n ¼ 77). thickness. 711 treatment stopped.
Watt 2020 knee Synovial fluid 20 individuals Biochemical: 10 predefined 6/10 sf analytes showed 8 putative
undergoing KJD for mechanosensitive changes between baseline mechanosensitive
symptomatic molecules: Activin A, and 6 weeks KJD. molecules of the
radiographic knee OA. TGFb1, MCP-1, IL-6, FGF-2, Of these, IL-6, MCP-1, FGF- inflammatory response
LTBP2, MMP-3, TSG-6, 2, LTBP2 and TGFb-1 (activin A, LTBP2, TGFb-1,
TIMP-1 and IL-8. showed a predominant FGF-2, TIMP-1, IL-6, MCP-1,
Clinical: KOOS-4 increase in levels, while and IL-8) were seen in sf
activin A mainly over the period of KJD.
decreased
Nambi 2020 Knee Serum 60 participants with Biochemical: BMP 2, 4, 6, BMP 2, 4, 6, and 7 didn't VRT in PTOA shows
post-traumatic knee OA and 7 CRP, TNF-a, IL-2, IL-4, show any significant beneficial changes in pain,
receiving either virtual and IL-6 by immunoassays. changes between the functional disability, and
reality training (VRT) or Clinical: VAS and WOMAC groups. modification of

Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248

sensory-motor training index. Inflammatory biomarkers inflammatory biomarkers
(SMT) or control (CRP, TNF-a, IL-2, but no effect on bone
(supervised IL-4, IL-6) were reduced in morphogenic proteins.
conventional exercise VRT group compared to
programs for the knee control and sensory-motor
muscles). training groups.
Diagnosis & Phenotyping
Bianchi 2020 TMJ Saliva 46 controls. Machine learning: four For the biomolecular The final prediction model
Serum 46 TMJ OA. machine learning models markers, no differences had an accuracy of 0.823
(Logistic Regression, between OA and control (SD: 0.029) to predict TMJ
Random Forest, LightGBM, subjects were found. OA status using
XGBoost) treating 52 Top features to diagnosis LightGBM þ XGBoost with
clinical, imaging and TMJ were: mouth range 1,378 features interactions.
biochemical markers. without pain, gender and
Biochemical: serum muscle soreness, VE-
(angiogenin, BDNF, CXCL- cadherin in saliva and
16, ENA-78, MMP-3, MMP- headaches, PAI-1 in saliva
7, OPG, PAI-1, TGF-b1, and headaches, PAI-1 in
TIMP-1, TRANCE, VE- saliva and range of mouth
Cadherin, VEGF) and saliva opening without pain,
(angiogenin, BNDF, CXCL- energy, Haralick
16, ENA-78, MMP-7, OPG, correlation, entropy and
PAI-1, TGF-b1). interactions of TGF-b1 in
Clinical: mouth range of saliva and headaches, VE-
motion, age pain, crucial cadherin in serum and
facial pain, worst facial angiogenin in saliva.
pain, average pain,
headache, muscle soreness,
vertical range.
Imaging: radiomics features
from the HR-CBCT scans.
Goode 2020 Spine Serum 74 participants (37 Biochemical: N-cadherin, Significant associations Individual and
patients with Keratin-19, Lumican, were found between combinations of
radiographic disc space CXCL6, RANTES, IL-17, IL-6, radiographic DSN and OPG, biochemical biomarkers
narrowing (DNS) and BDNF, OPG and NPY. IL-6 and NPY. may reflect radiographic
low back pain - 37 Clinical: pressure-pain Relative to a cluster with DSN.
controls) threshold (PPT) low levels of biomarkers, a
measurements, using a cluster representing
standard mechanical elevated levels of OPG,
 pressure-based RANTES, Lumican, Keratin-
dolorimeter. 19 and NPY and a cluster
Imaging: the Burnett atlas representing elevated
on standard X-ray. levels of NPY were
significantly associated
with radiographic DSN.
Bay-Jensen 2021 Knee Serum 933 OA patients. Biochemical: CRPM by a The mean CRPM levels A subset of OA patients
658 RA patients. competitive solid-phase were significantly lower in appears to have tissue
ELISA and CRP by a high OA compared to the RA inflammation comparable
sensitivity C-reactive patients; however, a to that of RA.
protein assay. significant subset of OA High CRPM levels are
patients (31%) had CRPM prognostic of incident knee
levels (9 ng/mL) OA.
comparable to RA.
OA patients with CRPM
levels 9 ng/mL were more
likely to develop contra-
lateral knee OA assessed by
X-ray over a 2-year follow-
up period.
Biochemical: OSM, IL-1a, IL-

Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248

Garcia 2021 Knee Synovial fluid 27 OA patients. Synovial fluid of OA OSM might play a
1b, IL-4, IL-6, IL-7, IL-8, IL- patients with detectable prominent role in
10, IL-13, TNF-a, IFN-g, and OSM contained higher inflammatory phenotypes
IL-1Ra were measured levels of other of OA.
using a multiplex bead inflammatory cytokines,
assay. namely interferon gamma
(IFN-g), IL-1a and TNF-a,
likely indicating a more
inflammatory state.
Luo 2021 Knee Serum 253 OA patients. Biochemical: PRO-C2 Subjects with low PRO-C2 Serum/plasma level of type
Plasma Imaging: risk of levels had greater JSN II collagen formation, PRO-
radiographic medial joint compared with subjects C2, may be an objective
space narrowing (JSN) over with high PRO-C2. indicator of a low cartilage
24 months. repair endotype, displaying
radiographic progression
and superior response to a
pro-anabolic drug.

Common abbreviation: AHSG ¼ alpha-2-HS-glycoprotein; BDNF ¼ brain-derived neurotrophic factor; BML ¼ bone marrow lesion; BMP ¼ bone morphogenetic proteins; CD14 ¼ cluster of differentiation 14;
C8 ¼ complement 8; CFB ¼ complement factor B; CLU ¼ clusterin; COMP ¼ cartilage oligomeric matrix protein; CRPM ¼ metabolite of C-reactive protein, derived from protease degradation; CS846 ¼ chondroitin sulfate 846
epitope; CTXI, CTXII ¼ C-terminal cross-linked telopeptide of collagen type I and II, respectively; CXC L ¼ chemokine (CXC motif) ligand; DM ¼ diabete mellitus; DSN ¼ disc space narrowing; ENA epithelial neutrophil-
activating peptide; FN1 ¼ fibronectin 1; GDF ¼ Growth differentiation factor; HA ¼ hyaluronic acid; hsCRP ¼ high sensitivity C-reactive protein; hs-CTnI ¼ high sensitivity cardiac troponin I; hs-CTnT ¼ high-sensitive cardiac
troponin T; IL ¼ Interleukin; IFN ¼ interferon; IGF ¼ insulin growth factor; JSN ¼ joint space narrowing; JSW ¼ joint space width; KOA ¼ knee osteoarthritis; KNG ¼ kinogen; KOOS ¼ knee injury and osteoarthritis outcome
score; KL ¼ Kellgren and Lawrence grade; LBP ¼ lipopolysaccharide binding protein; LTBP ¼ latent TGF beta binding proteins; MATN3 ¼ matrilin-3; MCP ¼ Monocyte Chemoattractant Protein; MMP ¼ matrix metal-
loproteinase; MRI ¼ magnetic resonance imaging; NPY ¼ neuropeptide Y; NRS ¼ numeric rating scale; OPG ¼ osteoprotegerin; OSM ¼ oncostatin M: PAI ¼ plasminogen activator inhibitor; pro-BNP ¼ N-terminal pro-brain
natriuretic peptide type B; PIINP, PIIANP ¼ N-terminal propeptide of type II collagen, and splice variant IIA; OA ¼ osteoarthritis; OAI ¼ osteoarthritis initiative; RA ¼ rheumatoid arthritis; RANTES ¼ regulated on activation,
normal T cell expressed and secreted; RARRES2 ¼ retinoic acid receptor responder 2; RKOA ¼ radiological knee osteoarthritis; sf ¼ synovial fluid, s ¼ serum; SIRT ¼ sirtuine; SLPI ¼ Secretory Leukocyte Peptidase Inhibitor;
TIMP ¼ tissue inhibitor of metalloproteinase; TNC ¼ Tenascin; TRANCE ¼ TNF-related activation-induced cytokine; TRL ¼ toll like receptor; TGF ¼ transforming growth factor; TNF ¼ tumor necrosis factor; VAS ¼ visual
analog scale; VEGF ¼ vascular endothelium growth factor; WOMAC ¼ Western Ontario and Mcmaster Universities osteoarthritis index. Less common abbreviations are explained in the table.

Table I OA YEAR IN REVIEW 2021 Osteoarthritis and Cartilage

Extracted data from pubmed research of human clinical studies

243
244 Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248

the LPS/LBP complex. The binding of the LPS/LBP complex to CD14 predict comorbidities or mortality in OA population. Biomarkers of
would then trigger the Toll-like receptor 4 (TLR4) of macrophages, mortality should be considered in existing biomarker classification
leading to the downstream production of inflammatory mediators since it was demonstrated a reciprocal relationship between
and catabolism of chondrocytes11. This pathobiological pathways comorbidities like metabolic syndrome and OA and between OA
might explain the association between LPS and LBP increase and and cardiovascular diseases17e19. Arnold et al.20 investigated the
the severity of MRI changes12. association between growth differentiation factor-15 (GDF-15), a
novel stress-responsive cytokine, and long-term all-cause mortality
Investigative among OA patients. GDF-15 has been measured in the serum of 636
subjects, who underwent hip or knee arthroplasty between 1995
An Investigative marker is one for which there is insufficient and 1996. During a median follow-up of 19.7 years, a total of 402
information to allow inclusion into one of the BIPED categories. One deaths occurred. Compared to the bottom quartile (<780 ng/L),
way to identify new biomarker candidate is the OMIC analysis of subjects within the top quartile of GDF-15 (>1,279 ng/L) demon-
biological fluids or secretome. There is few proteomic analyses of strated a 2.69-fold increased risk of dying. They concluded that in
the synovial fluid. Further, for many proteins from OA synovial subjects with OA, GDF-15 represents a potent predictor of
fluid, their intra-articular tissue of origin remains unknown. This is decreased survival over 20 years, independently of conventional
the reason for which Timur et al.13 have performed comparative cardiovascular risk factors, renal, cardiac, and inflammatory bio-
proteomic analysis to identify OA-specific and joint tissue-depen- markers as well as walking disability, previously associated with
dent secreted proteins that may serve as candidates for OA increased mortality and lower extremity OA. This is the first pro-
biomarker development on a tissue-specific basis. Clearly, the spective study in patients with OA, demonstrating a strong prog-
originality of this study was to identify the tissue origin of protein nostic value of elevated GDF-15 on decreased survival over 20
found in synovial fluid. To do that, they performed first, mass years. Further, GDF-15 provided additional prognostic information
spectrometry proteomic analysis of the synovial fluid of non-OA on all-cause mortality, which was not captured by conventional risk
and OA patients and a comparative proteomic analysis of the factors, “maximum walking distance” as a measure of gait
secretome of Hoffa's fat pad, synovium, meniscus and articular disability, or other well-established biomarkers and might be a
cartilage of OA knees. By this way, they found 62 proteins with a useful biomarker for future risk stratification or targeting preven-
higher abundance in OA than in non-OA knee synovial fluid and tive measures in such high-risk populations like subjects with OA.
234 with a lower abundance. Interestingly, 39 out of the 62 higher In the same cohort but selecting 679 OA subjects, undergoing
protein were found in the tissue secretome and 56 of the lower hip or knee replacement during 1995 and 1996, Rehm et al.21
detected in the secretome. In total 73 proteins were tissue specific. measured in serum N-terminal pro-B type natriuretic peptide (NT-
Among the most increased in OA, they found tenascin, histidine proBNP) and high-sensitivity troponins T and I (hs-cTnT and hs-
rich glycoprotein, fibronectin and alpha 2 HS glycoprotein and cTnI) which are well-characterized cardiac markers and provide
among the most decreased Cartilage Layer Intermediate Protein prognostic information. After adjustment for age and sex and
(CLIP), aggrecan core protein, decorin and fibromodulin. It is several other established covariates (i.e., BMI, smoking status,
interesting to note that this study revealed that in addition to localization of OA, diabetes mellitus, cholesterol, and cystatin C),
cartilage and synovium, the meniscus and Hoffa's fat pad are also the highest quartile of these biomarker concentrations were all
significant contributors to the OA-specific protein composition of associated with increased mortality compared to the lowest quar-
the synovial fluid of the knee joint. This study opens new per- tile. However, after simultaneous adjustment for the cardiac bio-
spectives of diagnostic based on synovial fluid analysis and aiming markers, a statistically significant relationship for hs-cTnT was lost.
to classify knee OA according their molecular endotype. These results suggest that the assessment of cardiac biomarkers
The nicotinamide adenine dinucleotide (NAD)-dependent may be a useful complement to traditional risk factors for pre-
enzyme Silent Information Regulator 2 Type 1 deacetylase (SIRT1) dicting mortality in subjects with OA and may be used for coun-
is a critical intracellular deacetylase in maintaining adult cartilage seling subjects or trigger specific interventions to reduce relevant
health by promoting chondrocyte survival and extracellular matrix cardiovascular risk factors.
homeostasis14. Mounting data support that SIRT1 is proteolytically Previous epidemiological studies have suggested that OA is
inactivated during OA, especially in response to inflammatory associated with elevated fasting glucose and is highly prevalent
stress induced for example by cytokines15. It is well known that in among those with diabetes mellitus22,23. Proposed mechanisms
chondrocyte SIRT-1 is cleaved by cathepsin B yielding a stable but include advanced glycation end products, which reduce cartilage
inactive N-terminal (NT) polypeptide (75SIRT1) and a C-terminal mechanical properties24. Additionally, systemic inflammation
(CT) fragment. Interestingly, Batshon et al.16 found that the NT/CT could contribute to changes in cartilage metabolism and integrity,
ratio was significantly increased in OA and this increase was mainly as well as neuromuscular impairment (as a result of symmetric
due to NT fragment levels increase. As in vitro and animal studies sensory polyneuropathy and autonomic neuropathy from long-
have demonstrated that 1) IL-1b and TNFa increased NT/CT SIRT-1 standing diabetes mellitus), which could lead to muscle weakness
ratio, 2) that NT/CT SIRT-1 ratio was positively correlated with OA and joint instability. In light of these previous reports, Rogers-
severity in mice, 3) that serum SIRT1 variants were cartilage Soeder et al.25 evaluated the association of diabetes mellitus and of
derived and 4) that senolytic drugs decreased this ratio, NT/CT SIRT- biomarkers of abnormal glucose metabolism with incident radio-
1 ratio can be considered as an investigative marker of OA burden of graphic knee OA of the tibiofemoral joint among participants in the
disease and senolytic drug efficacy. Multicenter Osteoarthritis STudy (MOST)26,27. They hypothesized
that the presence of diabetes mellitus, as well as hyperglycemia and
Prognostic of OA and mortality in OA patients elevated insulin resistance in participants with and without dia-
betes, would be associated with increased odds of incident radio-
The key feature of a prognostic marker is the ability to predict graphic knee OA, independent of BMI. Baseline fasting glucose and
the future onset of OA among those without OA at baseline or the HOmeostasis Model of Assessment e Insulin Resistance (HOMA-IR)
progression of OA among those with existing disease. Since it was formula were associated with increased odds of incident radio-
demonstrated that subjects with OA are at increased risk for car- graphic knee OA in the unadjusted model, as well as in model
diovascular and all-cause mortality, some studies attempt to also adjusted for age, race, clinic site and visit. After adjustment for BMI,
 Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248 245

these associations were attenuated and lost statistical significance. molecular response in synovial fluid, and if any change was asso-
In women, elevated HOMA-IR was associated with lower odds of ciated with clinical response. They followed 20 patients candidate
incident radiographic knee OA in the model adjusted for sex, age, to KJD during 12 months and collected synovial fluid before, during
race, clinic site, visit and BMI. Contrary to their hypothesis, after and after 6 weeks of treatment with KJD. In summary, they showed
adjustment for BMI, the authors failed to find increased odds of that IL-6, IL-8, Monocyte Chemoattractant Protein (MCP)-1, Fibro-
incident radiographic knee OA in participants with diabetes mel- blast Growth Factor (FGF)-2 and Transforming Growth Factor (TGF)
litus nor in participants with higher baseline levels of fasting b-1 levels increased, while activin A mainly decreased. Those with a
glucose and HOMA-IR. Surprisingly, after adjustment for BMI, they relevant increase in IL-8 synovial fluid level during the distraction
even observed a protective association of higher levels of HOMA-IR period had a greater improvement of their algo-functional status
with the odds of incident radiographic OA in women. The author over 12 months than those with no change. This unexpected result
did not explain this result but have excluded a protection by use of was not explained by the authors. This study indicates that joint
diabetic medications. Further investigation of potential protective distraction may provide a potential opportunity in the future to
effects of diabetes mellitus is required to elucidate this unexpected validate regenerative biomarker(s) and identify pathways that
observation. drive intrinsic cartilage repair.
Using 1,255 patients with painful and radiographic knee OA Non-pharmacological modalities including exercises program
from two phase III clinical trials designed to study the efficacy and have been largely recommended to treat patients with hip and knee
safety of oral salmon calcitonin28, Bihlet et al.29 have confirmed that OA. However, few studies have investigated the structure-modi-
baseline u-CTX-II was associated with elevated risk of radiographic fying effects of these modalities in human. Recently, Nambi et al.36
progression over 2 years in terms of both joint space narrowing and have compared the effects of virtual reality training (VRT) and
Kellgren and Lawrence (KL) grade. The statistical significance of the sensory-motor training (SMT) in bone morphogenetic proteins
association between u-CTX-II and risk of total joint replacement (BMP) and inflammatory biomarkers expression in post-traumatic
was lost upon adjustment for the baseline radiographic severity, OA after the anterior cruciate ligament injury. Sixty participants
which suggests that the association between u-CTX-II and total allocated to VRT, SMT or control group underwent training pro-
joint replacement may be indirect and driven by the radiographic grams for 4 weeks. In the control group, the participants underwent
stage. Then, they studied a model integrating u-CTX-II and de- supervised conventional exercise programs for the knee muscles.
mographic and radiological items. They showed that u-CTX-II in BMP measures such as BMP 2, 4, 6, and 7 did not show any sig-
association with age, sex, BMI and cumulative knee KL-grade pre- nificant changes with time or between the groups. In both VRT and
dicted total joint replacement within the 2-year period. The weight SMT groups C-reactive Protein (CRP), TNF-a and IL-6 decreased
of the biochemical marker u-CTX-II was important. Indeed, in the with time while IL-2 and IL-4 increased. In contrast, in the control
absence of baseline radiographic OA severity, u-CTX-II indepen- conventional exercise program, only TNF-a was significantly
dently contributed to prediction of total joint. The authors modified. Further, VRT and SMT groups significantly decreased
concluded that a composite model combining baseline age, sex, more pro-inflammatory mediators (CRP, TNF-a, IL-6) and increased
BMI, u-CTX-II and KL-grade predicted total joint replacement dur- more IL-2 and IL-4 than the control group. This indicates that ex-
ing a 2-year period. ercises program may modulate inflammation in OA.

Efficacy of intervention Diagnosis and phenotyping

An efficacy of intervention biomarker chiefly provides infor- Diagnosis

mation about the efficacy of treatment among those with OA or Diagnostic markers are defined by the ability to classify in-
those at high risk of developing OA. Efficacy of intervention dividuals as either diseased or non-diseased. Most of the existing
markers may be measured prior to therapy to predict treatment biomarkers have been demonstrated to statistically distinguish
efficacy, or may be measured more than once to assess short-term groups of patients with radiological or symptomatic knee OA from
changes that occur as a result of pharmacologic or other those without these characteristics. Rare are those that discrimi-
interventions. nate groups of patients with spine, hip, ankle or TMJ OA.
CMIV-711 is a novel selective cathepsin K inhibitor with bene- TMJ disorders are the second most common musculoskeletal
ficial effects on bone and cartilage in preclinical OA models30,31. condition affecting 5e12% of the population. Chronic disability in
Two-hundred forty-four participants with primary knee OA, were TMJ OA increases with aging. Bianchi et al.37 have analyzed fifty-
included in a 26-week randomized, double-blind, placebo- two clinical, biological and high-resolution Cone-beam computed
controlled phase IIa study and received MIV-711 100 or 200 mg Tomography (CBCT, radiomics) markers from TMJ OA patients and
daily or matched placebo. MIV-711 was not more effective than controls. They screened the diagnostic performance of each feature
placebo for pain, but it significantly reduced bone and cartilage and built a machine learning models based on the most relevant
lesion progression compared to placebo32. Significant reductions in features. They identified top features to diagnosis TMJ. including
bone resorption and cartilage degradation biomarkers were also mouth range without pain, gender and muscle soreness, VE-cad-
observed. Compared with baseline, levels of s-CTX-I and u-CTX-II herin in saliva and headaches, Plasminogen Activator Inhibitor
were reduced by 27.8% and 34.4%, respectively in the MIV-711 100 (PAI)-1 in saliva and headaches, PAI-1 in saliva and range of mouth
mg group and by 50.3% and 51.6%, respectively in the MIV-711 200 opening without pain, energy, Haralick correlation, entropy and
mg group. However, no difference between MIV-711 and placebo interactions of TGF-b1 in saliva and headaches, VE-cadherin in
group has been reported. serum and angiogenin in saliva. Interestingly and even no differ-
Surgical knee joint distraction (KJD) is a technique where, under ences in biochemical markers levels were found between OA and
anesthesia, an external fixation frame is placed on both sides of the control subjects, prediction models showed that the interaction
joint, allowing distraction (gradual pulling apart of the joint's bony between VE-cadherin, PAI-1, TGF-b1 in saliva and some clinical
ends by ~5 mm for 6 weeks). KJD leads to clinical improvement in features like headaches or range of motion without pain are among
knee OA and also apparent cartilage regeneration by MRI33,34. Watt the top features with mean contribution to the information gain in
et al.35 investigated if alteration of the joint's mechanical environ- the predictive models. As markers of inflammation, VE-cadherin,
ment during the 6-week period of KJD was associated with a angiogenin, TGF-b1 and PAI-1 have been previously shown to be
246 Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248

expressed in the TMJ synovial fluid and plasma and to be correlated Cartilage-driven phenotype. One study investigated the hypothesis
with the condylar morphology in OA patients. that a low cartilage repair endotype exists and that such endotype
Recently, osteoprotegerin (OPG) was found to be more elevated is more likely to progress radiographically. In this purpose, Luo et
in patient with radiographic intervertebral disc narrowing than in al.46 examined the associations between the level of PRO-C2, the N-
those without intervertebral discs narrowing. OPG is a member of terminal propeptide of collagen type IIB reflecting type II collagen
the TNF receptor superfamily and has been found to be associated formation cartilage formation, in serum of knee OA subjects and
with intervertebral disc degeneration in mice and human tissue radiological OA severity and progression. They found that patients
samples38,39. In propensity score matched regression analyses with the lowest PRO-C2 levels at baseline were those with the more
(matched for age, sex, BMI, knee OA, and hip OA), these authors important joint space narrowing progression. They had 3.4 times
have also observed significant associations of OPG, IL-6 and neu- more chance to progress over a 2-year period than the patient with
ropeptide-Y with disc space narrowing. Finally, they conducted a the highest PRO-C2 level. This suggests that low cartilage formation
cluster analysis to group together participants with similar may be associated with an endotype of higher structural loss.
biomarker profiles, ignoring case/control status. With cluster rep-
resenting lower levels of biomarkers as the referent, a significant Discussion
association was found among cases with disc space narrowing
compared to controls without disc narrowing for the cluster with The method used to write this review of the papers published
higher levels of OPG, Regulated on Activation, Normal T Cell between January 2020 and April 2021 was inspired by that used by
Expressed and Secreted (RANTES), keratin-19 and lumican and van Spil and Szilagyi for the year 201947. This approach ensures
neuropeptide-Y. A significant association was also found with cases continuity in the reflection on this theme. However, our work
with disc space narrowing compared to controls without disc space suffers from some limitations. Our search was limited to articles
narrowing in the cluster represented with a higher levels of neu- published on PubMed and the paper' selection was made by a single
ropeptide Y40. investigator according to arbitrary criteria. This is one reason for
which this review can not be considered as a systematic review.
This narrative review of the literature shows that interest in bio-
Phenotyping logical markers is still very much present in the scientific com-
Another important observation is that during the last year the munity. However, despite many efforts made by academic
majority of the papers dedicated to OA biomarkers aimed to better researchers but also by industry, there are still neither marker
define OA phenotype and clearly identify their constituent molec- allowing routine diagnosis, nor biological markers approved by
ular endotypes with multiple biomarkers. The two main investi- drug agencies such as drug development tools. The main reasons
gated phenotypes during the last year were the inflammatory/ are the absence of an “on-off” biological marker for OA, the lack of
immune/synovitis driven phenotype and the cartilage driven sensitivity to changes in existing markers, but also the great
phenotype. phenotype heterogeneity of the subjects included in the cohorts
used to validate or qualify biological markers. The absence of sur-
rogate biochemical markers for standard X-ray strongly slows
Inflammatory-driven phenotype. CRP is mainly expressed and down the development of DMOAD. This is probably one of the
released from the liver in response to injury and infection, where reasons that prompted researchers to search biological markers
from it is released in its pentameric form and binds to the site of that allow to select rapid OA progressor patients. Identifying these
injury followed by binding to complement and Fc receptors. At this patients would reduce the sample size and duration of clinical
site, CRP may be metabolized by proteases, such as matrix metal- trials. During the last 80 months some studies were focused on the
loproteinases (MMPs), resulting in the release of the CRP metabo- use of biochemical markers to better characterized OA patients'
lites41,42. One such metabolite (M) is the MMP degradation product phenotypes. Among them, some are actually cellular biomarkers
CRPM. Interestingly, Bay-Jensen et al.43 compared the levels of CRP and their clinical relevance in biological fluids needs to be deter-
and CRPM in blood of OA and Rheumatoid Arthritis (RA) patients to mined. Anyway, this approach by phenotype has been demon-
investigate if OA patients have inflammation levels comparable to strated in other disease area to be helpful to better select subjects to
that of RA patients. They used the serum of patients coming from include in clinical trials according the therapeutic target of the item.
three different cohorts including either early RA, moderate to se- This can also lead to significant savings in healthcare management
vere RA or OA subjects. They found that the levels of serum CRP and by administrating the right treatment to the right patient.
CRPM were significantly higher in the RA cohort than in the OA The second key message coming from this year's review is that
cohort but a third of the OA patients had CRPM and half had CRP there are now models integrating multiple demographic, clinical,
levels corresponding to the levels observed in RA. This indicated imaging and biochemical markers allowing to better predict OA
that as significant subset of patients may have an inflammatory incidence and progression. In the coming months, important data
signature. Interestingly, only CRPM, and not CRP, was prognostic for will be disseminated thanks to the European IMI-APPROACH con-
OA incidence. Indeed, OA patients with CRPM levels 9 ng/mL were sortium including forty Universities and research organizations and
more likely to develop contra-lateral knee OA assessed by X-ray six industries48,49. The aim of APPROACH project is creating a
over a 2-year follow-up period. platform comprising data on large cohorts of patients and healthy
Oncostatin M (OSM) is a cytokine from the IL-6 family that has people. The APPROACH cohort is unique in its attempt to select
been shown to be detectable in synovial fluid in up to 30% of the OA patients primarily from existing cohorts using machine learning
patients44. In line with these findings, Garcia et al.45 demonstrated models trained using patient data from the CHECK cohort to in-
synovial fluid of knee OA patients with detectable OSM contained crease the likelihood of radiographic joint space width (JSW) loss
higher levels of other inflammatory cytokines, namely interferon and/or knee pain progression during a limited, 2-year follow-up
gamma (IFN-g), IL-1a and TNF-a, likely indicating a more inflam- period. In addition to this unique preselection of patients, the
matory state. APPROACH cohort combines a very broad spectrum of conventional
Taken together these data indicate that OSM and CRPM are and novel, explorative, imaging, biochemical, clinical and de-
candidate biomarker for identification of OA patients with an in- mographic markers. Using data science techniques suitable to
flammatory phenotype. analyze these ‘big data’, algorithms of biomarkers should identify
 Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248 247

and predict phenotypes/endotypes of OA that share distinct un- 7. Mobasheri A, Saarakkala S, Finnil€ a M, Karsdal MA, Bay-
derlying pathobiological mechanisms with their structural and Jensen AC, van Spil WE. Recent advances in understand-
function consequences, relevant for practical and targeted clinical ing the phenotypes of osteoarthritis. F1000Res 2019;12:8.
trials. F1000 Faculty Rev-2091, https://doi.org/10.12688/
In conclusion, this year provides very interesting new data on f1000research.20575.1; PMID: 31885861; PMCID:
biochemical markers, mainly on the identification of the pheno- PMC6913225.
types of patients with OA. The identification and characterization of 8. Bauer DC, Hunter DJ, Abramson SB, Attur M, Corr M, Felson D,
these phenotypes are important steps towards understanding the et al, Osteoarthritis Biomarkers Network. Classification of
disease, its diagnosis and its treatment. This research approach is osteoarthritis biomarkers: a proposed approach. Osteoarthritis
very promising and will probably make it possible to better specify Cartilage 2006;14(8):723e7.
the indications for current treatments and to accelerate the 9. Liem Y, Judge A, Kirwan J, Ourradi K, Li Y, Sharif M. Multi-
development of DMOAD by allowing a better selection of patients variable logistic and linear regression models for identification
to be included in clinical trials. of clinically useful biomarkers for osteoarthritis. Sci Rep
2020;10(1):11328e35.
Contributions 10. Rajandran SN, Ma CA, Tan JR, Liu J, Wong SBS, Leung YY.
YH has performed papers research and written the narrative Exploring the association of innate immunity biomarkers with
summary. MRI features in both early and late stages osteoarthritis. Front
Med (Lausanne) 2020;7:554e669.
11. Nair A, Kanda V, Bush-Joseph C, Verma N, Chubinskaya S,
Conflict of interest
Mikecz K, et al. Synovial fluid from patients with early osteo-
YH is the founder and the president of Artialis SA. He also received
, Tilman SA, Naturex, Laboratoire arthritis modulates fibroblast-like synoviocyte responses to
consulting fees from Nestle
toll-like receptor 4 and toll-like receptor 2 ligands via soluble
Expanscience, Immunobio and Genequine. The authors also
CD14. Arthritis Rheum 2012;64(7):2268e77.
received funding from the Belgian Walloon Region under grant
12. Huang Z, Kraus VB. Does lipopolysaccharide-mediated
agreement N 1320131, 7781 and 6905.
inflammation have a role in OA? Nat Rev Rheumatol
2016;12(2):123e9.
Role of the funding source 13. Timur UT, Jahr H, Anderson J, Green DC, Emans PJ,
The funding source did not have any influence on study design, Smagul A, et al. Identification of tissue-dependent proteins
collection, analysis and interpretation of data, in the writing of the in knee OA synovial fluid. Osteoarthritis Cartilage
manuscript and in the decision to submit the manuscript for 2021;29(1):124e33.
publication. 14. Gabay O, Sanchez C, Dvir-Ginzberg M, Gagarina V, Zaal KJ,
Song Y, et al. Sirtuin 1 enzymatic activity is required for
Acknowledgments cartilage homeostasis in vivo in a mouse model. Arthritis
Rheum 2013;65(1):159e66.
None. 15. Dvir-Ginzberg M, Gagarina V, Lee EJ, Booth R, Gabay O, Hall DJ.
Tumor necrosis factor a-mediated cleavage and inactivation of
References SirT1 in human osteoarthritic chondrocytes. Arthritis Rheum
2011;63(8):2363e73.
1. Kraus VB, Karsdal MA. Osteoarthritis: current molecular bio- 16. Batshon G, Elayyan J, Qiq O, Reich E, Ben-Aderet L, Kandel L,
markers and the way forward. Calcif Tissue Int 2020, https:// et al. Serum NT/CT SIRT1 ratio reflects early osteoarthritis
doi.org/10.1007/s00223-020-00701-7. Epub ahead of print. and chondrosenescence. Ann Rheum Dis 2020;79(10):
PMID: 32367210. 1370e80.
2. Karsdal MA, Michaelis M, Ladel C, Siebuhr AS, Bihlet AR, 17. Swain S, Sarmanova A, Coupland C, Doherty M, Zhang W.
Andersen JR, et al. Disease-modifying treatments for osteoar- Comorbidities in osteoarthritis: a systematic review and meta-
thritis (DMOADs) of the knee and hip: lessons learned from analysis of observational studies. Arthritis Care Res (Hoboken)
failures and opportunities for the future. Osteoarthritis Carti- 2020;72(7):991e1000.
lage 2016;24(12):2013e21. 18. Courties A, Sellam J, Berenbaum F. Metabolic syndrome-asso-
3. Henrotin Y, Sanchez C, Cornet A, Van de Put J, Douette P, ciated osteoarthritis. Curr Opin Rheumatol 2017;29(2):
Gharbi M. Soluble biomarkers development in osteoarthritis: 214e22.
from discovery to personalized medicine. Biomarkers 19. Hall AJ, Stubbs B, Mamas MA, Myint PK, Smith TO. Association
2015;20(8):540e6. between osteoarthritis and cardiovascular disease: systematic
4. Bay-Jensen AC, Henrotin Y, Karsdal M, Mobasheri A. The need review and meta-analysis. Eur J Prev Cardiol 2016;23(9):
for predictive, prognostic, objective and complementary 938e46.
blood-based biomarkers in osteoarthritis (OA). EBioMedicine 20. Arnold N, Rehm M, Büchele G, Peter RS, Brenner RE,
2016;7:4e6. Günther KP, et al. Growth differentiation factor-15 as a potent
5. Dell'Isola A, Allan R, Smith SL, Marreiros SS, Steultjens M. predictor of long-term mortality among subjects with osteo-
Identification of clinical phenotypes in knee osteoarthritis: a arthritis. J Clin Med 2020;9(10):3107e16.
systematic review of the literature. BMC Musculoskelet Disord 21. Rehm M, Büchele G, Peter RS, Brenner RE, Günther KP,
2016;17(1):425e33. Brenner H, et al. Relationship between cardiac biomarker
6. Mobasheri A, van Spil WE, Budd E, Uzieliene I, Bernotiene E, concentrations and long-term mortality in subjects with
Bay-Jensen AC, et al. Molecular taxonomy of osteoarthritis for osteoarthritis. PLoS One 2020;15(12):e0242814.
patient stratification, disease management and drug develop- 22. Rehling T, Bjørkman AD, Andersen MB, Ekholm O, Molsted S.
ment: biochemical markers associated with emerging clinical Diabetes is associated with musculoskeletal pain, osteoar-
phenotypes and molecular endotypes. Curr Opin Rheumatol thritis, osteoporosis, and rheumatoid arthritis. J Diabetes Res
2019;31(1):80e9. 2019 Dec 6;2019:6324348.
248 Y. Henrotin / Osteoarthritis and Cartilage 30 (2022) 237e248

23. Dell'Isola A, Vinblad J, Lohmander S, Svensson AMRN, PhD, proteins and inflammatory biomarkers in post-traumatic
Turkiewicz A, et al. Understanding the role of diabetes in the osteoarthritis. Sci Rep 2020 Sep 28;10(1):15864.
osteoarthritis disease and treatment process: a study protocol 37. Bianchi J, de Oliveira Ruellas AC, Gonçalves JR, Paniagua B,
for the Swedish Osteoarthritis and Diabetes (SOAD) cohort. Prieto JC, Styner M, et al. Osteoarthritis of the Temporoman-
BMJ Open 2019;9(12):e032923. dibular Joint can be diagnosed earlier using biomarkers and
24. Chanchek N, Gersing AS, Schwaiger BJ, Nevitt MC, Neumann J, machine learning. Sci Rep 2020;10(1):8012.
Joseph GB, et al. Association of diabetes mellitus and 38. Takegami N, Akeda K, Yamada J, Sano T, Murata K, Huang J,
biochemical knee cartilage composition assessed by T2 relax- et al. RANK/RANKL/OPG system in the intervertebral disc.
ation time measurements: data from the osteoarthritis initia- Arthritis Res Ther 2017;19(1):121.
tive. J Magn Reson Imaging 2018;47(2):380e90. 39. Liang QQ, Li XF, Zhou Q, Xing L, Cheng SD, Ding DF, et al. The
25. Rogers-Soeder TS, Lane NE, Walimbe M, Schwartz AV, expression of osteoprotegerin is required for maintaining the
Tolstykh I, Felson DT, et al, Multicenter Osteoarthritis (MOST) intervertebral disc endplate of aged mice. Bone 2011;48(6):
Study Group. Association of diabetes mellitus and biomarkers 1362e9.
of abnormal glucose metabolism with incident radiographic 40. Goode AP, Schwartz TA, Kraus VB, Huebner JL, George SZ,
knee osteoarthritis. Arthritis Care Res (Hoboken) 2020;72(1): Cleveland RJ, et al. Inflammatory, structural, and pain
98e106. biochemical biomarkers may reflect radiographic disc space
26. Segal NA, Torner JC, Felson DT, Niu J, Sharma L, Lewis CE, et al. narrowing: the Johnston County Osteoarthritis Project.
Knee extensor strength does not protect against incident knee J Orthop Res 2020;38(5):1027e37.
symptoms at 30 months in the multicenter knee osteoarthritis 41. Marnell L, Mold C, Du Clos TW. C-reactive protein: ligands,
(MOST) cohort. PM R 2009;1(5):459e65. receptors and role in inflammation. Clin Immunol
27. Segal NA, Anderson DD, Iyer KS, Baker J, Torner JC, Lynch JA, 2005;117(2):104e11.
et al. Baseline articular contact stress levels predict incident 42. Skjøt-Arkil H, Schett G, Zhang C, Larsen DV, Wang Y, Zheng Q,
symptomatic knee osteoarthritis development in the MOST et al. Investigation of two novel biochemical markers of
cohort. J Orthop Res 2009;27(12):1562e8. inflammation, matrix metalloproteinase and cathepsin
28. Karsdal MA, Byrjalsen I, Alexandersen P, Bihlet A, Andersen JR, generated fragments of C-reactive protein, in patients with
Riis BJ, et al, CSMC021C2301/2 Investigators. Treatment of ankylosing spondylitis. Clin Exp Rheumatol 2012;30(3):
symptomatic knee osteoarthritis with oral salmon calcitonin: 371e9.
results from two phase 3 trials. Osteoarthritis Cartilage 43. Bay-Jensen AC, Bihlet A, Byrjalsen I, Andersen JR, Riis BJ,
2015;23(4):532e43. Christiansen C, et al. Serum C-reactive protein metabolite
29. Bihlet AR, Bjerre-Bastos JJ, Andersen JR, Byrjalsen I, (CRPM) is associated with incidence of contralateral knee
Karsdal MA, Bay-Jensen AC. Clinical and biochemical factors osteoarthritis. Sci Rep 2021;11(1):6583.
associated with risk of total joint replacement and radio- 44. Manicourt DH, Poilvache P, Van Egeren A, Devogelaer JP,
graphic progression in osteoarthritis: data from two phase III Lenz ME, Thonar EJ. Synovial fluid levels of tumor necrosis
clinical trials. Semin Arthritis Rheum 2020;50(6):1374e81. factor alpha and oncostatin M correlate with levels of markers
30. Lindstro €m E, Rizoska B, Tunblad K, Edenius C, Bendele AM, of the degradation of crosslinked collagen and cartilage
Maul D, et al. The selective cathepsin K inhibitor MIV-711 at- aggrecan in rheumatoid arthritis but not in osteoarthritis.
tenuates joint pathology in experimental animal models of Arthritis Rheum 2000;43(2):281e8.
osteoarthritis. J Transl Med 2018;16(1):56. 45. Garcia JP, Utomo L, Rudnik-Jansen I, Du J, Zuithoff NPA,
31. Lindstro €m E, Rizoska B, Henderson I, Terelius Y, Jerling M, Krouwels A, et al. Association between oncostatin M expres-
Edenius C, et al. Nonclinical and clinical pharmacological sion and inflammatory phenotype in experimental arthritis
characterization of the potent and selective cathepsin K in- models and osteoarthritis patients. Cells 2021;10(3):508.
hibitor MIV-711. J Transl Med 2018 May 9;16(1):125. 46. Luo Y, Samuels J, Krasnokutsky S, Byrjalsen I, Kraus VB, He Y,
32. Conaghan PG, Bowes MA, Kingsbury SR, Brett A, Guillard G, et al. A low cartilage formation and repair endotype predicts
Rizoska B, et al. Disease-modifying effects of a novel cathepsin radiographic progression of symptomatic knee osteoarthritis.
K inhibitor in osteoarthritis: a randomized controlled trial. J Orthop Traumatol 2021;22(1):10.
Ann Intern Med 2020;172(2):86e95. 47. van Spil WE, Szilagyi IA. Osteoarthritis year in review 2019:
33. Intema F, Van Roermund PM, Marijnissen AC, Cotofana S, biomarkers (biochemical markers). Osteoarthritis Cartilage
Eckstein F, Castelein RM, et al. Tissue structure modification in 2020;28(3):296e315.
knee osteoarthritis by use of joint distraction: an open 1-year 48. van Helvoort EM, van Spil WE, Jansen MP, Welsing PMJ,
pilot study. Ann Rheum Dis 2011;70(8):1441e6. Kloppenburg M, Loef M, et al. Cohort profile: the Applied
34. Wiegant K, van Roermund PM, Intema F, Cotofana S, Eckstein F, Public-Private Research enabling OsteoArthritis Clinical
Mastbergen SC, et al. Sustained clinical and structural benefit Headway (IMI-APPROACH) study: a 2-year, European, cohort
after joint distraction in the treatment of severe knee osteoar- study to describe, validate and predict phenotypes of osteo-
thritis. Osteoarthritis Cartilage 2013;21(11):1660e7. arthritis using clinical, imaging and biochemical markers. BMJ
35. Watt FE, Hamid B, Garriga C, Judge A, Hrusecka R, Custers RJH, Open 2020;10(7):e035101.
et al. The molecular profile of synovial fluid changes upon joint 49. Taylor J, Dekker S, Jurg D, Skandsen J, Grossman M,
distraction and is associated with clinical response in knee Marijnissen AK, et al. APPROACH research consortium and
osteoarthritis. Osteoarthritis Cartilage 2020;28(3):324e33. APPROACH Principal Investigators. Making the patient voice
36. Nambi G, Abdelbasset WK, Elsayed SH, Khalil MA, heard in a research consortium: experiences from an EU
Alrawaili SM, Alsubaie SF. Comparative effects of virtual reality project (IMI-APPROACH). Res Involv Engagem 2021 May
training and sensory motor training on bone morphogenic 10;7(1):24.