International Journal of Obesity www.nature.

com/ijo

ARTICLE OPEN

Clinical Research

Impact of baseline adipose tissue characteristics on change in

adipose tissue volume during a low calorie diet in people with
obesity—results from the LION study
✉
Daniela Junker 1 , Mingming Wu1, Anna Reik 2, Johannes Raspe 1, Selina Rupp1, Jessie Han1, Stella M. Näbauer 1
,
2
Meike Wiechert , Arun Somasundaram1, Egon Burian1,3, Birgit Waschulzik 4, Marcus R. Makowski1, Hans Hauner 2,5
,
Christina Holzapfel 2,6,9 and Dimitrios C. Karampinos 1,7,8,9

© The Author(s) 2024

BACKGROUND/OBJECTIVES: Weight loss outcomes vary individually. Magnetic resonance imaging (MRI)-based evaluation of
adipose tissue (AT) might help to identify AT characteristics that predict AT loss. This study aimed to assess the impact of an 8-week
1234567890();,:

low-calorie diet (LCD) on different AT depots and to identify predictors of short-term AT loss using MRI in adults with obesity.
METHODS: Eighty-one adults with obesity (mean BMI 34.08 ± 2.75 kg/m², mean age 46.3 ± 10.97 years, 49 females) prospectively
underwent baseline MRI (liver dome to femoral head) and anthropometric measurements (BMI, waist-to-hip-ratio, body fat), followed by
a post-LCD-examination. Visceral and subcutaneous AT (VAT and SAT) volumes and AT fat fraction were extracted from the MRI data.
Apparent lipid volumes based on MRI were calculated as approximation for the lipid contained in the AT. SAT and VAT volumes were
subdivided into equidistant thirds along the craniocaudal axis and normalized by length of the segmentation. T-tests compared baseline
and follow-up measurements and sex differences. Effect sizes on subdivided AT volumes were compared. Spearman Rank correlation
explored associations between baseline parameters and AT loss. Multiple regression analysis identified baseline predictors for AT loss.
RESULTS: Following the LCD, participants exhibited significant weight loss (11.61 ± 3.07 kg, p < 0.01) and reductions in all MRI-based AT
parameters (p < 0.01). Absolute SAT loss exceeded VAT loss, while relative apparent lipid loss was higher in VAT (both p < 0.01). The
lower abdominopelvic third showed the most significant SAT and VAT reduction. The predictor of most AT and apparent lipid losses was
the normalized baseline SAT volume in the lower abdominopelvic third, with smaller volumes favoring greater AT loss (p < 0.01 for SAT
and VAT loss and SAT apparent lipid volume loss).
CONCLUSIONS: The LCD primarily reduces lower abdominopelvic SAT and VAT. Furthermore, lower abdominopelvic SAT volume was
detected as a potential predictor for short-term AT loss in persons with obesity.
International Journal of Obesity; https://doi.org/10.1038/s41366-024-01568-6

INTRODUCTION Outcomes of a short-term weight-reducing intervention have a

One strategy to prevent obesity-related comorbidities such as significant impact on long-term weight maintenance success
diabetes, cardiovascular disease, and certain cancers is a [3–5]. This short-term effect, in turn, varies substantially between
lifestyle intervention aiming at weight loss, ideally by reducing persons, with varying loss of AT in the different AT depots. The
excess adipose tissue (AT) and ectopic fat [1]. The various fat ability to accomplish short-term weight loss is dependent on
depots have distinct metabolic profiles due to substantial many factors, including adherence to the intervention [6], insulin
functional differences [2]. Weight loss is also accompanied by resistance [7, 8], genetics and epigenetics [9, 10], gut microbiota
loss of organ- and muscle mass; thus, the success of a weight [11], sleeping habits [12, 13] or basal metabolic rate [14, 15]. Age
loss intervention is best assessed by evaluating changes in and sex are also factors often discussed, but findings from several
the AT. studies are inconsistent [16].

1
Institute of Diagnostic and Interventional Radiology, School of Medicine and Health, Technical University of Munich, Munich, Germany. 2Institute for Nutritional Medicine, School
of Medicine and Health, Technical University of Munich, Munich, Germany. 3Department of Diagnostic and Interventional Radiology, University Hospital Ulm, Ulm, Germany.
4
Institute of AI and Informatics in Medicine, School of Medicine and Health, Technical University of Munich, Munich, Germany. 5Else Kroener-Fresenius-Center of Nutritional
Medicine, School of Life Sciences, Technical University of Munich, Freising, Germany. 6Department of Nutritional, Food and Consumer Sciences, Fulda University of Applied
Sciences, Fulda, Germany. 7Munich Institute of Biomedical Engineering, Technical University of Munich, Garching, Germany. 8Munich Data Science Institute, Technical University
of Munich, Garching, Germany. 9These authors jointly supervised this work: Christina Holzapfel, Dimitrios C. Karampinos. ✉email: daniela.junker@tum.de

Received: 12 January 2024 Revised: 4 June 2024 Accepted: 12 June 2024

 D. Junker et al.
2
The distribution of AT is also relevant for the success of a weight Anthropometric measurements
loss intervention. People with abdominal obesity and thus more Waist and hip circumferences were measured by trained study
visceral adipose tissue (VAT) benefit more from weight loss than staff according to standard operating procedures [39]. Waist-to-
those with gluteal-femoral obesity, who tend to have more hip ratio (WHR) was calculated as the ratio between waist- and hip
subcutaneous adipose tissue (SAT) [17]. Such AT characteristics circumference.
might be predictors for an intervention’s success but are mainly Weight was assessed in a fasted state in light clothing without
assessed by anthropometric measurements [16, 18, 19]. A few shoes and with an emptied bladder using a body composition
studies used imaging techniques and found that increased VAT, scale (BC-418MA, Tanita Europe B.V., Netherlands) for bioimpe-
VAT/SAT ratio, or total AT at baseline are associated with greater dance analysis (BIA). Since the participants were allowed to wear
weight- and VAT loss [20–25]. However, most studies performing a light clothing, one kilogram was deducted for the assessment.
weight loss intervention while using imaging methods focused on Weight (kg) and body fat (%, kg) were recorded. Height was
the changes in AT instead of evaluating the association between measured in a standing position without shoes using a
baseline AT characteristics and the outcome of the intervention stadiometer (Seca 214, Seca, Hamburg, Germany). BMI was
[8, 26–29]. Additionally, studies are limited by small sample sizes calculated as the quotient of weight in kilograms and height in
[23], by investigating single AT depots [30], or by using single-slice meter squared (kg/m²) [39]. Participants underwent baseline and
measurements leading to cross-sectional areas instead of measur- follow-up appointments at the Institute for Nutritional Medicine.
ing volumes for AT quantification [20–22, 24, 25, 31].
In general, methods that can assess the abdominopelvic fat MRI measurements
content and distribution range from simple anthropometric The MRI examinations of the abdomen and pelvis were performed
measures to more complex 3D imaging technologies, such as on a 3T MR scanner (Ingenia Elition X, Philips Healthcare, Best, The
Computed Tomography (CT) or Magnetic Resonance Imaging Netherlands; software release 5.6). Participants were placed in
(MRI). These methods highly differ regarding accuracy, preci- supine position, head first, and a 16-channel torso coil and the
sion, and the amount of information gathered. 3D imaging build-in-table 12-channel posterior coil were used. For PDFF and
technologies offer the highest precision and amount of volume measurements of AT, a 6-echo multi-echo gradient echo
collected data. MRI is the preferred method for AT imaging in sequence with bipolar gradients was used in four stacks, covering
healthy individuals since it does not require ionizing radiation. the abdomen and pelvis from the liver dome to the center of the
Single-slice measurements have poorly predicted VAT and SAT femoral heads. Each stack was acquired during a breath-hold scan
changes during weight loss [32]. In groups of people of different of 10.3 seconds (see sequence parameters in Table S1). The mean
ages and sexes, more accurate results are achieved by time interval between the baseline clinical assessment (e.g.,
measuring AT volumes in a volumetric approach [33]. Examining anthropometric measurements) at the Institute for Nutritional
the whole abdomen and pelvis also allows assessment of the Medicine [39] and the baseline MRI scan was 8 days (range,
regional AT distribution in the craniocaudal axis. The current 0–43 days). For the follow-up measurements after the LCD, the
gold standard of MRI methods for spatially resolved tissue fat time interval was kept to a minimum with a mean of 1.7 days
quantification are Dixon-based chemical shift encoding-based (range, 0–21 days).
fat quantification techniques [34–37]. After considering multiple
confounding factors, these techniques measure a tissue’s PDFF mapping. PDFF, the state-of-the-art chemical shift
proton density fat fraction (PDFF) [38], offering a method to encoding-based fat quantification technique [38], is defined as
further characterize AT phenotypes. the proportion of mobile proton density in fat tissue attributable
The present study aimed to (1) assess AT and apparent lipid to fat. PDFF maps were generated using the vendor´s online
volume changes as meaningful parameters for the success of a complex-based fat quantification algorithm (Philips mDIXON
weight loss intervention using an MRI-based approach and to (2) Quant package, for scan parameters see Supplementary Table
evaluate baseline parameters concerning their correlation with S1), accounting for the presence of multiple fat peaks, a single T2*
and their prognostic value for those changes. To achieve this, data correction, and phase errors [40].
from the lifestyle intervention (LION) study in people with obesity
undergoing an 8-week formula-based weight loss intervention Adipose tissue segmentation. VAT and SAT were segmented using
[39] was analyzed. a deep learning-based automated segmentation pipeline after
[41, 42]. The segmented region extended from the liver dome to
the middle of the femoral heads. For details and code availability,
STUDY COHORT AND METHODS see [43, 44]. VAT and SAT volumes and mean AT PDFF values (%)
Study design and participants were extracted. To investigate regional variations of VAT and SAT
Between October 2019 and October 2021, a subgroup of 127 distribution in the craniocaudal axis, the segmentations were
persons with obesity (73 females, mean age 45 years, BMI subdivided into equidistant thirds, starting at the liver dome and
30.0–39.9 kg/m2) were recruited from the LION study [39] to ending at the middle of the femoral head. A Python algorithm
undergo an MRI examination of the abdomen and pelvis on a 3T (Version 3.8.0, Python Software Foundation, Beaverton, USA) was
scanner. Of those, 81 participants (49 females, mean age 46.3 established for these calculations. This resulted in SAT and VAT
years) completed a follow-up MRI scan after an 8-week formula- subvolumes for the segmented abdominopelvic region’s upper,
based low-calorie diet (LCD) of 800 kcal with an optional middle, and lower third. The middle third included the periumbi-
additional daily intake of 200 g non-starchy vegetables. The lical abdominal fat, and the lower third contained the pelvic and
analysis presented in the following is based on the data obtained gluteal fat down to the hip joint. The lipid volume was calculated
from the 81 participants who completed the study. as PDFF ´ Volume for both AT compartments, thereby not taking
The study protocol and procedures were approved by the into account MR-invisible components [45], thus the term
ethical committee of the School of Medicine and Health of the “apparent lipid volume” is employed hereinafter. Absolute loss
Technical University of Munich, Germany (Project Number 69/19S; of total AT and apparent lipid volume was calculated as
ClinicalTrials.gov Identifier: NCT04023942). Written informed con- Volumefollow
up
Volumebaseline and expressed in L. Relative AT
sent was obtained from all participants. Inclusion and exclusion and apparent lipid volume losses were calculated as
criteria were defined as described elsewhere [39]; for the MRI ðVolumefollow
up
Volumebaseline Þ=Volumebaseline and expressed
examinations, additional exclusion criteria were standard MRI in %. Note that this method of volume loss calculation may result
contraindications. in negative values. VAT/SAT ratio was calculated as Volume
VolumeSAT in the
VAT

International Journal of Obesity

 D. Junker et al.
3
segmented region. Normalization of AT volumes (expressed in of −24.08 ± 7.25% SATTV (−26.27 ± 8.14% SATLV) and
L/cm) was performed using the length of the segmented region in −23.75 ± 7.24% VATTV (−27.95 ± 8.62% VATLV) (Table S3), while
cm: Volume
Length . This was done in order to account for different
women showed a loss of −19.75 ± 5.82% SATTV (−21.03 ± 6.20%
physiques, as the height of the participants did not necessarily SATLV) and −20.50 ± 6.58% VATTV (−24.45 ± 7.50% VATLV) (Table
correlate with the length of the abdominopelvic region. S4). The absolute volume loss of both total volume and apparent
lipid volume was significantly higher in SAT compared to VAT
(p < 0.01) in males and females. The sex-specific results for the
Statistical analysis
relative AT- and apparent lipid volume losses are shown in Fig. S2.
Statistical analysis was performed using MedCalc Statistical
When dividing SAT and VAT into equidistant thirds from the
Software (version 20.118; MedCalc Software bvba, Ostend,
liver dome down to the middle of the femoral head, a significant
Belgium; https://www.medcalc.org; 2022). Normal distribution of
change in all subvolumes for both SAT and VAT was observed
data was verified using cumulative frequency distribution plots.
(Table 1). However, the strongest effects were observed for both
Data is expressed as mean ± standard deviation, if not otherwise
SAT and VAT in the lower third, with a large effect for SAT (0.81)
denoted. Differences between baseline and follow-up measure-
and a medium effect for VAT (0.71). The evaluation in females
ments were tested by means of the paired samples t-test. Sex
revealed a large effect for the lower third SAT (0.89), while in
differences and differences between AT depots were evaluated
males, large effects were seen in lower SAT and middle and lower
using the independent samples t-test. In order to compare the
VAT (0.81, 0.85 and 0.86, respectively; see Table 2).
effect size of the intervention on the normalized AT volumes in
each subvolume (upper, middle, lower third), standardized mean
Association analyses
difference (SMD, Cohen’s d) was calculated. SMD can be
The Δ SATTV (%) correlated the strongest with the normalized
interpreted based on [46], with an absolute value of >│0.2│
baseline SAT volume (L/cm) in the lower third, with a higher loss
representing a small effect, >│0.5│ representing a medium effect,
(indicated by a negative number) being associated with smaller
and >│0.8│ representing a large effect. Due to the typical
baseline volumes (r = 0.52, p < 0.01, Fig. 1). Other strong correla-
distribution patterns of AT in males and females, separate SMD
tions of Δ SATTV (%) were found with normalized total SAT volume
analyses for males and females were performed. Correlation
(L/cm), normalized SAT apparent lipid volume (L/cm), and BIA-
analyses were carried out using Pearson correlation. To determine
based body fat (%) at baseline (r = 0.47, r = 0.47, r = 0.49,
which baseline parameter had the strongest effect on each
respectively; p < 0.01) (Table 3). Δ SATLV (%) showed associations
change in AT (relative total volume loss Δ ATTV and relative
similar to those of the aforementioned Δ SATTV (%) (see Table 3
apparent lipid volume loss Δ ATLV of SAT and VAT, respectively),
and Fig. 1).
stepwise multiple linear regression analyses were performed,
The Δ VATTV (%) correlated best with normalized SAT volume in
including the five strongest associated parameters according to
the lower third and normalized total SAT volume at baseline (both
the Pearson correlation coefficient, and with the addition of age
r = 0.48, p < 0.01). Other correlations were found with normalized
and sex as covariates. All analyses were conducted using two-
SAT apparent lipid volume, normalized SAT volume in the upper
tailed tests with a significance level of 5%, and no correction was
third, and BIA-based body fat (%) at baseline (r = 0.47, r = 0.43,
made for multiple testing because of the explorative character of
r = 0.41, respectively; p < 0.01 for all). Δ VATLV (%) correlated best
the study.
with normalized total SAT volume, normalized SAT apparent lipid
volume, and normalized SAT volume in the lower as well as the
upper third at baseline (r = 0.53, r = 0.53, r = 0.52 and r = 0.50,
RESULTS respectively; p < 0.01) (Table 3 and Fig. 1). Notably, baseline SAT
Characteristics of the study cohort and changes following PDFF correlated with all AT and apparent lipid losses (each
the LCD p < 0.01), while baseline VAT PDFF did not show any correlations
The population characteristics at baseline and follow-up are (Table 3). The correlation analyses were also performed in males
shown in Table 1. Characteristics stratified by sex are presented in and females separately, as shown in Supplementary Tables
the Supplementary Material (Tables S2–S4). S5 and S6.
After the LCD intervention, a mean weight loss of Correlation analyses of weight loss and BMI reduction (both
−11.61 ± 3.07 kg, a mean BMI decrease of −3.89 ± 0.88 kg/m², relative to baseline) with baseline MRI and anthropometric
and a BIA-based body fat reduction of −4.08 ± 2.04% was achieved parameters revealed no correlations (data not shown). When
(Table 1). The imaging data revealed an absolute loss of SAT total males and females were considered separately, no correlation was
volume (Δ SATTV) of −3.24 ± 1.07 L and an absolute loss of SAT found in males. In females, both relative weight- and BMI loss
apparent lipid volume (Δ SATLV) of −3.14 ± 1.03 L. Absolute loss of correlated with baseline body fat (%) (r = 0.38 for weight loss and
VAT total volume (Δ VATTV) was −1.24 ± 0.66 L, and of VAT r = 0.37 for BMI loss, both p < 0.01), waist circumference (r = 0.3,
apparent lipid volume (Δ VATLV) −1.17 ± 0.64 L (Table 1). Relative p = 0.04 for both), and SAT volume in the lower third (total SAT:
to baseline, loss of SAT total volume (Δ SATTV) was −21.46 ± 6.73% r = 0.32, p = 0.03 for weight loss and r = 0.32, p = 0.02 for BMI loss;
(Δ SATLV −23.10 ± 7.44%), and Δ VATTV −21.79 ± 6.99% (Δ VATLV normalized SAT: r = 0.32, p = 0.03 for weight loss and r = 0.32,
−25.83 ± 8.09%) (Table 1). The Δ SATTV (L) was significantly higher p = 0.02 for BMI loss), with smaller baseline values always favoring
than Δ VATTV (L) (p < 0.01), while the Δ SATTV (%) and Δ VATTV (%) greater losses.
were not significantly different (Table 1). However, there was a
significantly higher Δ VATLV (%) compared to Δ SATLV (%) (p < 0.01) Multiple regression models
(Fig. S1). The PDFF decreased both in SAT and in VAT, with a Multiple linear regression models for the prediction of AT and
stronger decrease in VAT (p < 0.01). apparent lipid volume losses (indicated by a negative number)
The levels of significance of the differences between males and showed that the normalized SAT volume in the lower third and
females at baseline and follow-up are marked by an asterisk in body fat % at baseline are the predictors for Δ SATTV (%) (b = 0.24,
Table 1, and the corresponding numbers separated by sex are p < 0.01 and b = 0.003, p = 0.02), with less SAT at baseline being
shown in Tables S2–S4. For males, Δ SATTV was −3.50 ± 1.23 L (Δ beneficial for SAT loss. The same was true for Δ SATLV (%)
SATLV −3.41 ± 1.18 L) and Δ VATTV was −1.76 ± 0.68 L (Δ VATLV (b = 0.28, p < 0.01 and b = 0.003, p = 0.01). For Δ VATTV (%), the
−1.68 ± 0.67 L) (Table S3). For females, Δ SATTV was −3.07 ± 0.93 L predictor was normalized SAT volume in the lower third (b = 0.32,
(Δ SATLV −2.96 ± 0.88 L) and Δ VATTV was −0.91 ± 0.37 L (Δ VATLV p < 0.01), and for Δ VATLV (%), the predictors were normalized total
−0.84 ± 0.35 L) (Table S4). Relative to baseline, men showed a loss SAT volume and normalized SAT volume in the middle third

International Journal of Obesity

 D. Junker et al.
4
Table 1. Characteristics of the study population (n = 81) at baseline and follow-up.
Parameter Baseline Follow-Up Difference baseline vs.
follow-up
Characteristics & Females 49 (60.49%) 49 (60.49%) –a
anthropometry Age, years 46.30 (10.97, 21–65)** 46.48 (10.95, 21–66)** –a
Height, m 1.72 (10.24, 1.49–1.98)** 1.72 (10.24, 1.49–1.98) –a
**
Weight, kg 101.56 (15.17, 89.95 (13.76, p < 0.01
71.30–149.80)** 63.40–132.20)**
BMI, kg/m2 34.08 (2.75, 30.19 (2.62, p < 0.01
30.20–39.70) 26.10–36.00)
Waist circumference, cm 105.75 (11.54, 96.30 (10.21, p < 0.01
76–135.20)** 77.70–119.20)**
Hip circumference, cm 118.00 (8.07, 110.37 (7.20, p < 0.01
101.00–136.00) 97.30–128.40)
WHR 0.90 (0.09, 0.72–1.12)** 0.87 (0.08, 0.71–1.05)** p < 0.01
Body fat, % 38.61 (7.05, 34.53 (7.98, p < 0.01
22.80–52.30)** 18.90–51.00)**
SAT PDFF, % 90.39 (1.33, 88.43 (2.38, p < 0.01
86.18–92.61)** 80.61–92.20)**
Total volume, L 15.55 (3.96, 7.47–24.93) 12.31 (3.62, p < 0.01
5.37–20.27)
Upper third 4.03 (1.18, 1.43–6.89) 3.17 (1.10, 1.31–5.92)* p < 0.01
Middle third 5.25 (1.51, 2.49–8.98) 4.24 (1.35, 1.95–7.22) p < 0.01
Lower third 6.27 (1.60, 2.87–10.48) 4.90 (1.48, 1.86–8.40)* p < 0.01
Normalized volumeb, L/cm 0.35 (0.09, 0.18–0.60)** 0.29 (0.08, 0.13–0.51)** p < 0.01
Upper third 0.27 (0.08, 0.11–0.53)** 0.22 (0.08, 0.09–0.45)** p < 0.01
Middle third 0.35 (0.10, 0.18–0.58) 0.29 (0.09, 0.13–0.53) p < 0.01
Lower third 0.42 (0.10, 0.20–0.68)** 0.34 (0.10, 0.12–0.59)** p < 0.01
Apparent lipid volume, L 14.09 (3.69, 6.44–22.77) 10.95 (3.39, p < 0.01
4.38–18.52)
Normalized apparent lipid 0.32 (0.08, 0.15–0.55)** 0.25 (0.08, 0.10–0.46)** p < 0.01
volumeb, L/cm
VAT PDFF, % 78.85 (4.10, 74.67 (4.83, p < 0.01
67.43–85.66)** 62.18–81.94)**
Total volume, L 5.70 (2.32, 1.91–12.93)** 4.46 (1.85, 1.56–9.81)** p < 0.01
Upper third 1.12 (0.64, 0.25–3.41)** 0.97 (0.58, 10.12–2.98) p < 0.01
**
Middle third 2.92 (1.29, 0.79–6.75)** 2.22 (0.99, 0.65–4.96)** p < 0.01
Lower third 1.66 (0.57, 0.70–3.97)** 1.27 (0.42, 0.60–2.44)* p < 0.01
Normalized volumeb, L/cm 0.13 (0.05, 0.04–0.28)** 0.10 (0.04, 0.04–0.22)** p < 0.01
Upper third 0.07 (0.04, 0.02–0.20)** 0.07 (0.04, 0.01–0.17)** p < 0.01
Middle third 0.20 (0.08, 0.05–0.44)** 0.15 (0.06, 0.05–0.33)** p < 0.01
Lower third 0.11 (0.04, 0.05–0.26)* 0.09 (0.03, 0.05–0.17) p < 0.01
Apparent lipid volume, L 4.57 (2.04, 1.29–11.08)** 3.40 (1.57, 0.97–8.04)** p < 0.01
Normalized apparent lipid 0.10 (0.04, 0.03–0.24)** 0.08 (0.03, 0.02–0.18)** p < 0.01
volumeb, L/cm
VAT/SAT ratio 0.39 (0.21, 0.12–1.15)** 0.39 (0.20, 0.13–1.07)** p = 0.61
The significance level for sex differences is marked as * for p < 0.05 and ** for p < 0.01. Data are shown as mean (SD, range) or as n (%).
MRI magnetic resonance imaging, SAT subcutaneous adipose tissue, VAT visceral adipose tissue, PDFF proton density fat fraction, BMI body mass index, SD
standard deviation, WHR waist-to-hip ratio.
a
Not changing with intervention.
b
Normalized by the length of the abdominopelvic region in cm.

(b = 1.03, p < 0.01 and b = −0.49, p = 0.04, respectively). Overall, a loss compared to a woman with less SAT in the lower third at
smaller SAT volume in the lower third at baseline was baseline exhibiting greater VAT loss. The reported multiple
advantageous for SAT- and VAT loss (Table 4). The impact of regression analyses were not performed separated by sex as the
SAT distribution on VAT loss is visible in Fig. 2, showing a woman number of independent variables in the model (n = 7) was too
with more baseline SAT in the lower third exhibiting smaller VAT large for the respective groups (n = 32 and n = 49).

International Journal of Obesity

 D. Junker et al.
5
Table 2. Standardized mean differences (SMD) for comparison of normalized SAT and VAT subvolumes (L/cm) before and after weight loss
intervention (upper, middle, and lower third of the segmented region liver dome to femoral head) in the complete cohort and in males and females.
Compartment Subvolume SMD 95% CI
All (n = 81) SAT, L/cm Upper third −0.68 −0.81 to −0.54
Middle third −0.65 −0.77 to −0.55
Lower third −0.81 −0.95 to −0.66
VAT, L/cm Upper third −0.22 −0.33 to −0.12
Middle third −0.59 −0.68 to −0.49
Lower third −0.71 −0.86 to −0.55
Females (n = 49) SAT, L/cm Upper third −0.68 −0.88 to −0.49
Middle third −0.73 −0.93 to −0.54
Lower third −0.89 −1.13 to −0.65
VAT, L/cm Upper third −0.22 −0.43 to −0.02
Middle third −0.62 −0.76 to −0.49
Lower third −0.62 −0.82 to −0.42
Males (n = 32) SAT, L/cm Upper third −0.78 −0.95 to −0.61
Middle third −0.58 −0.68 to −0.48
Lower third −0.81 −1.02 to −0.63
VAT, L/cm Upper third −0.36 −0.54 to −0.17
Middle third −0.85 −1.06 to −0.64
Lower third −0.86 −1.12 to −0.62
SMD > │0.8│ based on standardized mean differences represent a large effect and are printed in bold.
CI confidence interval, SAT subcutaneous adipose tissue, SMD standardized mean differences, VAT visceral adipose tissue.

Fig. 1 Scatter plots depicting correlations between relative AT and AT apparent lipid volume losses and respective baseline parameters.
Blue dots represent male cases, open red diamonds represent female cases. A Correlation between Δ SATTV (%) and baseline normalized lower
third SAT volume. B Correlation between Δ VATTV (%) and baseline normalized lower third SAT volume. C Correlation between Δ SATLV (%) and
baseline normalized lower third SAT volume. D Correlation between Δ VATTV (%) and baseline normalized total SAT volume. SAT subcutaneous
adipose tissue, VAT visceral adipose tissue, Δ SATTV (%) relative loss of SAT total volume, Δ SATLV (%) relative loss of SAT apparent lipid volume,
Δ VATTV (%) relative loss of VAT total volume, Δ VATLV (%) relative loss of VAT apparent lipid volume.

International Journal of Obesity

 D. Junker et al.
6
Table 3.Pearson correlation coefficients for VAT and SAT adipose tissue and apparent lipid volume loss with anthropometric and imaging
parameters at baseline.
Parameters at baseline Δ SATTV (%) Δ SATLV (%) Δ VATTV (%) Δ VATLV (%)
Anthropometry BMI, kg/m² 0.32** 0.33** 0.29** 0.34**
Body fat, % 0.49** 0.53** 0.41** 0.42**
Waist circumference, cm 0.09 0.10 0.19 0.23*
Hip circumference, cm 0.36** 0.39** 0.28* 0.32**
WHR −0.15 −0.16 0.003 0.03
SAT PDFF, % 0.33** 0.37** 0.39** 0.43**
Normalized volumea, L/cm
• Total 0.47** 0.52** 0.48** 0.53**
• Upper 0.39** 0.43** 0.43** 0.50**
• Middle 0.36** 0.40** 0.39** 0.43**
• Lower 0.52** 0.56** 0.48** 0.52**
Normalized apparent lipid volumea, L/cm 0.47** 0.51** 0.47** 0.53**
VAT PDFF, % 0.05 0.04 0.09 0.14
Normalized volumea, L/cm
• Total 0.04 0.03 0.02 0.07
• Upper −0.02 −0.04 −0.08 −0.03
• Middle 0.001 −0.01 0.01 0.06
• Lower 0.20 0.21 0.18 0.24*
a
Normalized apparent lipid volume , L/cm 0.05 0.04 0.03 0.08
VAT/SAT ratio −0.24* −0.28* −0.25* −0.24*
Significance is marked as * for p < 0.05 and ** for p < 0.01.
SAT subcutaneous adipose tissue, VAT visceral adipose tissue, WHR waist-to-hip ratio, Δ SATTV (%) relative loss of SAT total volume, Δ SATLV (%) relative loss of SAT
apparent lipid volume, Δ VATTV (%) relative loss of VAT total volume, Δ VATLV (%) relative loss of VAT apparent lipid volume.
a
Normalized by the length of the abdominopelvic region in cm.

DISCUSSION Several weight loss intervention studies have shown that the
The present analysis shows that both anthropometric and MRI- absolute loss of SAT volume is greater than that of VAT [49–51],
measured parameters of AT significantly decrease in people with which is also evident in the current LION study. There was,
obesity after an 8-week formula-based LCD as weight loss however, no significant difference between the relative losses of
intervention. The absolute volume change Δ SATTV (L) was VAT and SAT volume. When evaluating the changes in apparent
significantly higher than Δ VATTV (L). In contrast, relative volume lipid volumes in SAT and VAT, VAT exhibited a significantly higher
loss (in % to baseline) was statistically different for apparent lipid relative loss of apparent lipids compared to SAT. Studies in rodent
volumes, with a higher apparent lipid volume loss in VAT models may provide possible explanations for these findings:
compared to SAT. When dividing SAT and VAT into equidistant under fasting conditions, the responsiveness of genes related to
thirds in the craniocaudal axis from the liver dome to the femoral lipid metabolism is more pronounced in VAT than in SAT [52].
heads, the largest volume changes were detectable in the lower Further, the activity of the sympathetic nervous system toward
third for both SAT and VAT. The strongest predictor for nearly all distinct adipose depots is known to be dynamic and potentially
AT- and apparent lipid volume losses was the normalized SAT hierarchical, with a switch in lipolytic activity from VAT to SAT in
volume in the lower third at baseline. The best predictor for VAT the course of a calorie restriction [53]. Thus, in the relatively short
apparent lipid volume loss was normalized total SAT volume. follow-up period of 8 weeks, lipids in VAT were potentially the
Considering the study design, decreases in obesity markers preferred energy source, however, the total volume loss in SAT
(anthropometric and MRI-based) were expected since participants compared to VAT was much greater.
underwent an 8-week LCD. A decrease in PDFF in SAT in this Changes in AT volume were most prominent in the lower third
cohort has been reported before [47]. Potential explanations are of the segmented region, i.e. around the pelvis. It should be
the depletion of lipids from the adipocytes leading to a relative emphasized that the location of the AT in MRI studies is, to some
PDFF decrease or an increase in tissue hydration due to the extent, different from anthropometric measurements due to the
metabolic processes associated with weight loss. Correlations of participants’ position during the measurements (lying vs. standing
obesity markers with AT hydration and with AT PDFF are known position). In MRI studies, AT is distributed more cranially due to the
[36, 48]. The observed correlation of smaller baseline SAT PDFF lack of gravity pulling the AT caudally. Thus, comparisons to
with greater AT loss in the present longitudinal setting adds a new studies using waist and hip circumferences are difficult. Previous
aspect to these relationships: a lower baseline fat fraction in SAT weight-loss studies found larger changes in abdominal AT than in
might be beneficial for AT loss success. PDFF is a good approach lower body AT (hip, leg, or gluteofemoral) [23, 54–56]. A study
to measure MR-visible AT lipid content. Given that lipids are including an overfeeding/underfeeding protocol found that the
primarily found inside the adipocytes, PDFF potentially serves as upper body AT returned to pre–weight-gain levels more rapidly
an indicator of adipocyte fat content, possibly even offering than lower body AT during underfeeding [57]. Differences in study
insights into adipocyte size. Nonetheless, confirming this hypoth- populations and methods could partially explain the discrepancies
esis in vivo poses a considerable challenge. with the present results. Furthermore, it has to be mentioned that

International Journal of Obesity

 D. Junker et al.
7
Table 4. Multiple regression analysis for VAT and SAT adipose tissue and apparent lipid volume loss.
Dependent variable R² Independent variables Coefficient b Std. Error t p
Δ SATTV (%) 0.32 Constant −0.42
Normalized baseline lower third SAT volume, L/cm 0.24 0.07 3.22 <0.01
Body fat, % 0.003 0.001 2.41 0.02
a
Age, years –
a
Sex –
a
Normalized baseline SAT volume, L/cm –
a
Normalized baseline SAT app. lipid volume, L/cm –
a
Normalized baseline upper third SAT volume, L/cm –
Δ SATLV (%) 0.38 Constant −0.47
Normalized baseline lower third SAT volume, L/cm 0.28 0.08 3.54 <0.01
Body fat, % 0.003 0.001 2.76 0.01
a
Age, years –
a
Sex –
a
Normalized baseline SAT volume, L/cm –
a
Normalized baseline SAT app. lipid volume, L/cm –
a
Normalized baseline upper third SAT volume, L/cm –
Δ VATTV (%) 0.23 Constant −0.35
Normalized baseline lower third SAT volume, L/cm 0.32 0.07 4.87 <0.01
a
Age, years –
a
Sex –
a
Body fat % –
a
Normalized baseline SAT volume, L/cm –
a
Normalized baseline SAT app. lipid volume, L/cm –
a
Normalized baseline upper third SAT volume, L/cm –
Δ VATLV (%) 0.32 Constant −0.44
Normalized baseline SAT volume, L/cm 1.03 0.26 3.90 <0.01
Normalized baseline middle third SAT volume, L/cm −0.49 0.23 −2.11 0.04
a
Age, years –
a
Sex –
a
Normalized baseline SAT app. lipid volume, L/cm –
a
Normalized baseline upper third SAT volume, L/cm –
a
Normalized baseline lower third SAT volume, L/cm –
Method: stepwise, enter variable if p < 0.05, remove variable if p > 0.1.
SAT subcutaneous adipose tissue, VAT visceral adipose tissue, Δ SATTV (%) relative loss of SAT total volume, Δ SATLV (%) relative loss of SAT apparent lipid volume,
Δ VATTV (%) relative loss of VAT total volume, Δ VATLV (%) relative loss of VAT apparent lipid volume, app. apparent.
a
Variables not included in the model.

there is no clear definition regarding “lower body fat” or gluteal AT) is advantageous for weight loss [17], and that an
“gluteofemoral fat”, and if and how much of the leg is included increased VAT or VAT/total AT ratio (thus less gluteal SAT) is
varies from study to study. beneficial for the success of a weight loss intervention [20–22].
It is well established that the pattern of AT distribution is a AT and apparent lipid volume losses were associated with
strong determinant of AT functioning (e.g., lipolytic function) and baseline anthropometric parameters (most strongly to body fat %)
that AT deposition differs between females and males [58, 59]. The and with MRI-measured parameters at baseline. The parameters
present results also show this sex dimorphism in accordance with that correlated best (five highest r-values) were included in the
the classic gynoid and android AT distribution [59]. Men lost more multiple regression models to narrow down the key predictors for
AT than women, especially in the middle and lower VAT, which is AT loss. The models revealed that the baseline normalized volume
in line with previous findings [22, 60]. In contrast, general AT loss of SAT in the lower abdominopelvic third predicts both SAT and
was pronounced in the lower SAT in both males and females. To VAT loss, with a smaller volume being associated with greater AT
the best of our knowledge, this pronounced SAT loss in the lower loss. The detected correlations of relative BMI- and weight loss
third for both males and females has not been reported before, with baseline anthropometric parameters in females- lower body
presumably because the MRI methodology applied in the present fat % and smaller waist circumference correlated with higher
work, dividing the section from liver dome to femoral heads into losses- contradict previous findings, where a higher waist
thirds, has not been used for AT volume measurements. The circumference was associated with success of a lifestyle interven-
finding is somewhat surprising, but the results are corroborated by tion for weight loss and body fat % showed no association with
previous findings showing that a higher waist-to-hip ratio (i.e., less success [19]. This could be attributed to differences in study

International Journal of Obesity

 D. Junker et al.
8

Fig. 2 Segmented coronal fat images of two female participants at baseline and follow-up showing less VAT loss in a woman with more
SAT in the lower third at baseline (participant A) compared to more VAT loss in a woman with fewer SAT in the lower third at baseline
(participant B). VAT is marked in green, and SAT is marked in blue. The upper left image shows the concept of division into equidistant thirds
(upper, middle, lower third). Participant A (female, 49 years, baseline BMI 32.1 kg/m²) with a SAT distribution with emphasis in the lower third
(0.43 L/cm) lost 12.62% VAT after the 8-week calorie restriction. Participant B (female, 56 years, baseline BMI 32.2 kg/m²) with a more balanced SAT
distribution with less SAT in the lower third (0.39 L/cm) lost 31.45% VAT after the 8-week calorie restriction. The biggest change in SAT is visible in
the lower third of the segmented region in both participants. SAT (blue) subcutaneous adipose tissue, VAT (green) visceral adipose tissue.

design, the limited BMI range (30.0–39.9 kg/m2) of the present intervention, so the 8 weeks of the intervention were identical for
study, or the fact that the reported findings are limited to females. all study participants. Lastly, different approaches for normalizing
However, data on these parameters as predictors for weight loss AT volumes could be considered, including BMI, height, or body
are scarce [16]. Nevertheless, BMI and weight loss consistently surface area. The length of the segmented region (from the liver
show correlations with SAT volume in the lower third in the dome to the femoral head) was selected as the parameter of
present analysis. choice as it best accounted for differences in physique with regard
to the torso and it yielded the clearest results compared to BMI
Limitations and strengths and body surface area (data not shown).
Some limitations of this study have to be considered. Firstly, Some strengths of this study should be mentioned. The use of
partial volume effects need to be taken into account when 3D imaging technologies for AT measurements, as has been
interpreting PDFF measurements with MRI, as PDFF cannot applied here, eliminates the limitations of single-slice measure-
differentiate between intracellular water content and non-lipid ments [32]. Furthermore, we used 3D imaging data in an
tissue portions (e.g., from adjacent organs) within a voxel interventional setting with a relatively large sample size compared
(3 × 3 × 6 mm³). Secondly, the present calculation of apparent to other studies [23]. The chemical shift encoding-based fat
lipid volumes does not take MR-invisible components (non-free- quantification method used here has the advantage to be
water and non-fat fractions) such as water bound to macro- relatively fast, allowing for breath-hold scans minimizing motion
molecules into account; thus, the term “apparent” was employed. artifacts. Using this technique, scans of larger body parts or even a
However, PDFF provides a good approximation for the actual lipid whole body scan could be performed. The segmentation of AT
content [45]. Thirdly, there was a time gap between anthropo- depots was achieved through an automated segmentation
metric (Institute for Nutritional Medicine) and MRI (Institute of pipeline based on deep learning methods after [41, 42], leading
Diagnostic and Interventional Radiology) measurements. Thirdly, to high accuracy and independence from different readers [44].
compared to studies that measured gluteofemoral AT (measured Moreover, employing this approach results in a noteworthy
by thigh circumference, hip circumference, or leg AT mass [61]), decrease in segmentation time, surpassing the efficiency of
the present analysis used segmentations as low as the middle of manual or semi-manual segmentation methodologies, as semi-
the femoral head. However, due to the lying position of the manual segmentation of similar datasets in an earlier study [36]
participants during the MRI-scan, the AT can be expected to be took around 25 min per case. Lastly, the weight loss intervention
distributed more cranially in contrast to a standing position. was highly standardized, increasing the comparability of the
Furthermore, there was no control group for the weight loss results between participants.

International Journal of Obesity

 D. Junker et al.
9
CONCLUSION 17. Wabitsch M, Hauner H, Bockmann A, Parthon W, Mayer H, Teller W. The rela-
In conclusion, a sex- and depot-specific decrease in AT in people tionship between body fat distribution and weight loss in obese adolescent girls.
with obesity undergoing an 8-week LCD, measured by anthropo- Int J Obes Relat Metab Disord. 1992;16:905–11.
metrics and MRI was observed. Results indicated a greater 18. Batterham M, Tapsell LC, Charlton KE. Baseline characteristics associated with
different BMI trajectories in weight loss trials: a case for better targeting of
reduction of SAT in the lower third of the abdominopelvic region.
interventions. Eur J Clin Nutr. 2016;70:207–11.
The baseline normalized SAT volume in the lower third of the 19. Cresci B, Castellini G, Pala L, Bigiarini M, Romoli E, Poggiali R, et al. Fit and
abdominopelvic region predicted both SAT and VAT volume loss, motivated: outcome predictors in patients starting a program for lifestyle change.
where a smaller volume was associated with greater AT loss. Obes Facts. 2013;6:279–87.
Consequently, measuring SAT volume in the lower abdominopel- 20. Hallgreen CE, Hall KD. Allometric relationship between changes of visceral fat and
vic third might help to predict the success of short-term AT loss total fat mass. Int J Obes. 2008;32:845–52.
after an LCD as a weight loss intervention. However, with regard to 21. Smith SR, Zachwieja JJ. Visceral adipose tissue: a critical review of intervention
long-term success and weight cycling, data collected after strategies. Int J Obes Relat Metab Disord. 1999;23:329–35.
12 months will have to be evaluated, and further studies beyond 22. Pinho CPS, Diniz ADS, Arruda IKG, Leite A, Rodrigues IG. Effects of weight loss on
adipose visceral and subcutaneous tissue in overweight adults. Clin Nutr.
12 months duration might be necessary.
2018;37:1252–8.
23. Busetto L, Tregnaghi A, Bussolotto M, Sergi G, Beninca P, Ceccon A, et al. Visceral
fat loss evaluated by total body magnetic resonance imaging in obese women
DATA AVAILABILITY operated with laparascopic adjustable silicone gastric banding. Int J Obes Relat
The network results from the deep learning-based automated segmentation pipeline Metab Disord. 2000;24:60–9.
are available at: https://github.com/BMRRgroup/lion-abd-seg-nnunet and https:// 24. Fujioka S, Matsuzawa Y, Tokunaga K, Kawamoto T, Kobatake T, Keno Y, et al.
github.com/BMRRgroup/lion-abd-seg-3dunet. The anthropometric and imaging Improvement of glucose and lipid metabolism associated with selective reduc-
datasets analyzed during the current study are available from the corresponding tion of intra-abdominal visceral fat in premenopausal women with visceral fat
author on reasonable request. obesity. Int J Obes. 1991;15:853–9.
25. Leenen R, van der Kooy K, Deurenberg P, Seidell JC, Weststrate JA, Schouten FJ,
et al. Visceral fat accumulation in obese subjects: relation to energy expenditure
REFERENCES and response to weight loss. Am J Physiol. 1992;263:E913–9.
1. Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA, 26. Brochu M, Tchernof A, Turner AN, Ades PA, Poehlman ET. Is there a threshold of
et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or visceral fat loss that improves the metabolic profile in obese postmenopausal
metformin. N Engl J Med. 2002;346:393–403. women? Metabolism. 2003;52:599–604.
2. Karpe F, Pinnick KE. Biology of upper-body and lower-body adipose tissue-link to 27. Chowdhury B, Kvist H, Andersson B, Bjorntorp P, Sjostrom L. CT-determined
whole-body phenotypes. Nat Rev Endocrinol. 2015;11:90–100. changes in adipose tissue distribution during a small weight reduction in obese
3. Unick JL, Neiberg RH, Hogan PE, Cheskin LJ, Dutton GR, Jeffery R, et al. Weight males. Int J Obes Relat Metab Disord. 1993;17:685–91.
change in the first 2 months of a lifestyle intervention predicts weight changes 8 28. Giannopoulou I, Ploutz-Snyder LL, Carhart R, Weinstock RS, Fernhall B, Goulo-
years later. Obesity. 2015;23:1353–6. poulou S, et al. Exercise is required for visceral fat loss in postmenopausal women
4. Astrup A, Rossner S. Lessons from obesity management programmes: greater with type 2 diabetes. J Clin Endocrinol Metab. 2005;90:1511–8.
initial weight loss improves long-term maintenance. Obes Rev. 2000;1:17–9. 29. Gallagher D, Heshka S, Kelley DE, Thornton J, Boxt L, Pi-Sunyer FX, et al. Changes
5. Handjieva-Darlenska T, Holst C, Grau K, Blaak E, Martinez JA, Oppert JM, et al. in adipose tissue depots and metabolic markers following a 1-year diet and
Clinical correlates of weight loss and attrition during a 10-week dietary inter- exercise intervention in overweight and obese patients with type 2 diabetes.
vention study: results from the NUGENOB project. Obes Facts. 2012;5:928–36. Diabetes Care. 2014;37:3325–32.
6. Alhassan S, Kim S, Bersamin A, King AC, Gardner CD. Dietary adherence and 30. Thamer C, Machann J, Stefan N, Haap M, Schafer S, Brenner S, et al. High visceral
weight loss success among overweight women: results from the A TO Z weight fat mass and high liver fat are associated with resistance to lifestyle intervention.
loss study. Int J Obes. 2008;32:985–91. Obesity. 2007;15:531–8.
7. Vidal-Ostos F, Ramos-Lopez O, Blaak EE, Astrup A, Martinez JA. The triglyceride- 31. Park HS, Lee K. Greater beneficial effects of visceral fat reduction compared with
glucose index as an adiposity marker and a predictor of fat loss induced by a low- subcutaneous fat reduction on parameters of the metabolic syndrome: a study of
calorie diet. Eur J Clin Invest. 2022;52:e13674. weight reduction programmes in subjects with visceral and subcutaneous obe-
8. Kantartzis K, Machann J, Schick F, Rittig K, Machicao F, Fritsche A, et al. Effects of a sity. Diabet Med. 2005;22:266–72.
lifestyle intervention in metabolically benign and malign obesity. Diabetologia. 32. Shen W, Chen J, Gantz M, Velasquez G, Punyanitya M, Heymsfield SB. A single MRI
2011;54:864–8. slice does not accurately predict visceral and subcutaneous adipose tissue
9. Papandonatos GD, Pan Q, Pajewski NM, Delahanty LM, Peter I, Erar B, et al. changes during weight loss. Obesity. 2012;20:2458–63.
Genetic predisposition to weight loss and regain with lifestyle intervention: 33. Demerath EW, Sun SS, Rogers N, Lee M, Reed D, Choh AC, et al. Anatomical
analyses from the Diabetes Prevention Program and the Look AHEAD Rando- patterning of visceral adipose tissue: race, sex, and age variation. Obesity.
mized Controlled Trials. Diabetes. 2015;64:4312–21. 2007;15:2984–93.
10. Delahanty LM, Pan Q, Jablonski KA, Watson KE, McCaffery JM, Shuldiner A, et al. 34. Hu HH, Kan HE. Quantitative proton MR techniques for measuring fat. NMR
Genetic predictors of weight loss and weight regain after intensive lifestyle Biomed. 2013;26:1609–29.
modification, metformin treatment, or standard care in the Diabetes Prevention 35. Reeder SB, Cruite I, Hamilton G, Sirlin CB. Quantitative assessment of liver fat with
Program. Diabetes Care. 2012;35:363–6. magnetic resonance imaging and spectroscopy. J Magn Reson Imaging.
11. Hjorth MF, Roager HM, Larsen TM, Poulsen SK, Licht TR, Bahl MI, et al. Pre- 2011;34:729–49.
treatment microbial Prevotella-to-Bacteroides ratio, determines body fat loss 36. Franz D, Weidlich D, Freitag F, Holzapfel C, Drabsch T, Baum T, et al. Association of
success during a 6-month randomized controlled diet intervention. Int J Obes. proton density fat fraction in adipose tissue with imaging-based and anthropo-
2018;42:580–3. metric obesity markers in adults. Int J Obes. 2018;42:175–82.
12. Chaput JP, Tremblay A. Sleeping habits predict the magnitude of fat loss in adults 37. Burian E, Syvari J, Holzapfel C, Drabsch T, Kirschke JS, Rummeny EJ, et al. Gender-
exposed to moderate caloric restriction. Obes Facts. 2012;5:561–6. and age-related changes in trunk muscle composition using chemical shift
13. Kline CE, Chasens ER, Bizhanova Z, Sereika SM, Buysse DJ, Imes CC, et al. The encoding-based water-fat MRI. Nutrients. 2018;10:1972.
association between sleep health and weight change during a 12-month beha- 38. Reeder SB, Hu HH, Sirlin CB. Proton density fat-fraction: a standardized MR-
vioral weight loss intervention. Int J Obes. 2021;45:639–49. based biomarker of tissue fat concentration. J Magn Reson Imaging.
14. Krotkiewski M, Garellick G, Sjostrom L, Persson G, Bjuro T, Sullivan L. Fat cell 2012;36:1011–4.
number, resting metabolic rate, mean heart rate, and insulin elevation while seeing 39. Reik A, Holzapfel C. Randomized controlled lifestyle intervention (LION) study for
and smelling food as predictors of slimming. Metabolism. 1980;29:1003–12. weight loss and maintenance in adults with obesity-design and methods. Front
15. Stroeve JH, Saccenti E, Bouwman J, Dane A, Strassburg K, Vervoort J, et al. Weight Nutr. 2020;7:586985.
loss predictability by plasma metabolic signatures in adults with obesity and 40. Eggers H, Brendel B, Duijndam A, Herigault G. Dual-echo dixon imaging with
morbid obesity of the DiOGenes study. Obesity. 2016;24:379–88. flexible choice of echo times. Magn Reson Med. 2011;65:96–107.
16. Chopra S, Malhotra A, Ranjan P, Vikram NK, Sarkar S, Siddhu A, et al. Predictors of 41. Isensee F, Jaeger PF, Kohl SAA, Petersen J, Maier-Hein KH. nnU-Net: a self-
successful weight loss outcomes amongst individuals with obesity undergoing configuring method for deep learning-based biomedical image segmentation.
lifestyle interventions: a systematic review. Obes Rev. 2021;22:e13148. Nat Methods. 2021;18:203–11.

International Journal of Obesity

 D. Junker et al.
10
42. Estrada S, Lu R, Conjeti S, Orozco-Ruiz X, Panos-Willuhn J, Breteler MMB, et al. the segmented region, MRI data collection, revision of the manuscript. SR, JH and SMN:
FatSegNet: a fully automated deep learning pipeline for adipose tissue seg- study coordination on the part of Radiology, MRI procedures and data collection, data
mentation on abdominal dixon MRI. Magn Reson Med. 2020;83:1471–83. cleaning, data extraction and data analysis. AR: study coordination on the part of
43. Wu M, Somasundaram A, Rupp S, Han J, Naebauer S, Junker D, et al. Assessing sex Nutritional Medicine, recruitment and screening of study participants, data collection,
differences in abdominal fat depots of people with obesity under weight loss analysis and interpretation, revision of the manuscript. MWi: study coordination on the
using automated segmentation. 2023 Annual Scientific Meeting of ISMRM. Tor- part of Nutritional Medicine, recruitment and screening of study participants, data
onto, Canada; 2023. collection and extraction. AS: setting up the segmentation pipeline, data extraction and
44. Somasundaram A, Wu M, Reik A, Rupp S, Han J, Naebauer S, et al. Evaluating sex- data cleaning. EB: medical supervision and supervision of data collection and
specific differences in abdominal fat volume and proton density fat fraction on segmentation. BW: data analysis and interpretation, providing detailed advice on all
MRI scans using automated nnU-net-based segmentation. Radiol Artif Intell. statistical issues. MRM: conception and design of the study. HH: conception and design
2024:e230471. of the study, data interpretation, revision of the manuscript. CH: conception, design and
45. Hu HH, Li Y, Nagy TR, Goran MI, Nayak KS. Quantification of absolute fat mass by supervision of the study, data collection, extraction and analysis, data interpretation,
magnetic resonance imaging: a validation study against chemical analysis. Int J revision of the manuscript. DCK: conception and design of the study, supervision of data
Body Compos Res. 2011;9:111–22. collection, extraction and analysis, data interpretation, revision of the manuscript.
46. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale,
N.J: L. Erlbaum Associates; 1988.
47. Han J, Wu M, Rupp S, Junker D, Reik A, Wiechert M, et al. Subcutaneous adipose FUNDING
tissue water fraction increases with weight loss in persons with obesity. 2022 This study is funded by the German Federal Ministry of Education and Research (BMBF,
annual scientific meeting of ISMRM. London, UK; 2022. grant number: 01EA1709) within the framework of the Junior Research Group
48. Michael N, Sadananthan SA, Yaligar J, Lee SS, Leow MK-S, Khoo CM, et al. Adipose “Personalized Nutrition & eHealth (PeNut)” of the enable Competence Cluster of Nutrition
tissue hydration as a potential non-invasive marker for adipose tissue hyper- Research. In addition, the analysis was part of the project “Imaging, metabolic risk, and
trophy. 2015 annual scientific meeting of ISMRM. Toronto, Canada; 2015. genetics: Algorithms based on Artificial Intelligence to predict metabolic changes through
49. Abe T, Song JS, Bell ZW, Wong V, Spitz RW, Yamada Y, et al. Comparisons of calorie Lifestyle Intervention (IMaGENE)” funded by BMBF (grant number: 16DKWN075). Further,
restriction and structured exercise on reductions in visceral and abdominal sub- the present work was supported by the German Research Foundation (project number
cutaneous adipose tissue: a systematic review. Eur J Clin Nutr. 2022;76:184–95. 450799851 and project number 455422993/FOR5298-iMAGO-P1). The authors from the
50. Gray DS, Fujioka K, Colletti PM, Kim H, Devine W, Cuyegkeng T, et al. Magnetic- Institute of Radiology also acknowledge research support from Philips Healthcare. Open
resonance imaging used for determining fat distribution in obesity and diabetes. Access funding enabled and organized by Projekt DEAL.
Am J Clin Nutr. 1991;54:623–7.
51. Leenen R, van der Kooy K, Droop A, Seidell JC, Deurenberg P, Weststrate JA, et al.
Visceral fat loss measured by magnetic resonance imaging in relation to changes in
serum lipid levels of obese men and women. Arterioscler Thromb. 1993;13:487–94. COMPETING INTERESTS
52. Li Y, Bujo H, Takahashi K, Shibasaki M, Zhu Y, Yoshida Y, et al. Visceral fat: higher DCK receives grant support from Philips Healthcare. HH is a member of the scientific
responsiveness of fat mass and gene expression to calorie restriction than sub- advisory board of Oviva AG (Zurich, Switzerland) and CH of 4sigma GmbH
cutaneous fat. Exp Biol Med. 2003;228:1118–23. (Oberhaching, Germany). HH and CH received speaker honoraria from Novo Nordisk
53. Sipe LM, Yang C, Ephrem J, Garren E, Hirsh J, Deppmann CD. Differential sym- (Copenhagen, Denmark). The other authors declare no conflicts of interest.
pathetic outflow to adipose depots is required for visceral fat loss in response to
calorie restriction. Nutr Diabetes. 2017;7:e260.
54. Hendler RG, Welle SL, Statt MC, Barnard R, Amatruda JM. The effects of weight ETHICAL APPROVAL
reduction to ideal body weight on body fat distribution. Metabolism. 1995;44:1413–6. All procedures were performed following the ethical standards of the institutional
55. Clifton PM. Relationship between changes in fat and lean depots following research committee and with the Helsinki Declaration and its later amendments or
weight loss and changes in cardiovascular disease risk markers. J Am Heart Assoc. comparable ethical standards. Written informed consent was obtained from all
2018;7:e008675. participants prior to data collection. Study protocols and procedures were approved
56. Bosy-Westphal A, Schautz B, Lagerpusch M, Pourhassan M, Braun W, Goele K, by the ethical committee of the School of Medicine and Health of the Technical
et al. Effect of weight loss and regain on adipose tissue distribution, composition University of Munich, Germany.
of lean mass and resting energy expenditure in young overweight and obese
adults. Int J Obes. 2013;37:1371–7.
57. Singh P, Somers VK, Romero-Corral A, Sert-Kuniyoshi FH, Pusalavidyasagar S, ADDITIONAL INFORMATION
Davison DE, et al. Effects of weight gain and weight loss on regional fat dis- Supplementary information The online version contains supplementary material
tribution. Am J Clin Nutr. 2012;96:229–33. available at https://doi.org/10.1038/s41366-024-01568-6.
58. Camilleri G, Kiani AK, Herbst KL, Kaftalli J, Bernini A, Dhuli K, et al. Genetics of fat
deposition. Eur Rev Med Pharmacol Sci. 2021;25:14–22. Correspondence and requests for materials should be addressed to Daniela Junker.
59. Palmer BF, Clegg DJ. The sexual dimorphism of obesity. Mol Cell Endocrinol.
2015;402:113–9. Reprints and permission information is available at http://www.nature.com/
60. Doucet E, St-Pierre S, Almeras N, Imbeault P, Mauriege P, Pascot A, et al. Reduction reprints
of visceral adipose tissue during weight loss. Eur J Clin Nutr. 2002;56:297–304.
61. Manolopoulos KN, Karpe F, Frayn KN. Gluteofemoral body fat as a determinant of Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims
metabolic health. Int J Obes. 2010;34:949–59. in published maps and institutional affiliations.

ACKNOWLEDGEMENTS
The authors would like to thank Sandhanakrishnan Ravichandran for his help and
input in the data cleaning process and Lisa Patzelt for her help with the MR scanning. Open Access This article is licensed under a Creative Commons
Furthermore, the authors are grateful to all participants of the LION Study and to all Attribution 4.0 International License, which permits use, sharing,
members of the LION study team, especially Vincent Winkler, Miriam Neidhardt, Bea adaptation, distribution and reproduction in any medium or format, as long as you give
Klos, Sandra Bayer, Judith Bodensteiner, Christine Reimers, Christina Ikkert, Andrea appropriate credit to the original author(s) and the source, provide a link to the Creative
Stiglmeier, Alexandra Sandner, Bärbel Huber, and Kurt Rack. We thank the Munich Commons licence, and indicate if changes were made. The images or other third party
Study Center for support in data management. material in this article are included in the article’s Creative Commons licence, unless
indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly
AUTHOR CONTRIBUTIONS from the copyright holder. To view a copy of this licence, visit http://
DJ: conception and design of the study protocol, medical supervision, data extraction, creativecommons.org/licenses/by/4.0/.
analysis and interpretation, drafting and revision of the manuscript. MWu: setting up the
MRI protocol, collecting and extracting data, setting up the segmentation pipeline,
cleaning data, revision of the manuscript. JR: setting up the algorithm for subdivision of © The Author(s) 2024

International Journal of Obesity