MR-based fat fraction changes in subcutaneous adipose tissue in people with obesity undergoing a weight loss intervention: results from the LION study

Introduction

Obesity has reached pandemic levels, with 60% of the adult population in Europe being identified as overweight or obese (1). This chronic condition is associated with adipocyte hypertrophy and subacute inflammation, which promotes systemic insulin resistance and metabolic dysfunction (2,3). Individuals are thereby predisposed to the development of type 2 diabetes and cardiovascular disease (4). Therefore, critical attention should be paid to the shifting trend toward an obesity-dominant demographic. More efforts should be aimed at mitigating the development of metabolic syndrome and co-morbidities (5). Achieving and maintaining weight loss is one of the most impactful strategies to reduce the risk of obesity-related chronic diseases (2,6,7). Specifically, reducing abdominal adipose tissue has been shown to significantly lower systemic inflammation and improve cardiometabolic mortality (5).

Body fat is primarily stored in the subcutaneous adipose tissue (SAT). Both visceral adipose tissue (VAT) and SAT are significant predictors of metabolic risk in opposite directions. An enlargement of SAT depots may have protective benefits by acting as a reservoir for excess triglycerides and reducing fat accretion in other organs among people with obesity (8). Abdominopelvic SAT can be divided anatomically into two compartments—superficial SAT (SSAT) and deep SAT (DSAT). These compartments are separated by a thin fibrous fascia (9). While SSAT has been associated with beneficial metabolic processes, the more saturated DSAT exhibits an inflammatory molecular profile, suggesting that these sub-compartments have distinct functional and morphological properties (10-12). A dedicated investigation of SSAT and DSAT cross-sectionally and longitudinally during weight loss may therefore provide a better insight into the heterogeneity of obesity.

There is a growing interest in developing ideal methods for monitoring health risk that can be used in the design of personalized weight loss interventions (13,14). Imaging methods such as computed tomography or magnetic resonance imaging (MRI) can be utilized as a non-invasive assessment tool of obesity through the evaluation of fat parameters such as SAT and VAT volume (15-17). MRI adipose tissue volume analysis has previously relied primarily on longitudinal relaxation time (T1) weighted imaging (18), and more recently, on chemical shift encoding (CSE)-based water fat separation combined with deep learning-based segmentation techniques (19-22). By calculating the VAT/SAT ratio, it has been possible to monitor an individual’s cardiometabolic risk and glycemic status (16,23). In addition, MRI CSE water-fat separation has been used to quantify the proton density fat fraction (PDFF) of gluteal fat. Gluteal fat PDFF was found to be positively associated with anthropometric parameters such as body mass index (BMI) and waist circumference (24). PDFF, defined as the ratio of the density of mobile protons from triglycerides to that from both triglycerides and water, is a highly reliable magnetic resonance (MR) based biomarker that is routinely used to quantify tissue fat concentration and is the current method of choice to estimate tissue fat content using MRI (25-29).

An alternative approach for estimating adipose tissue PDFF involves spectrally resolving the lipid and water peaks within a volume of interest using single-voxel magnetic resonance spectroscopy (MRS). Multiple studies have compared MRI and MRS for measuring PDFF across organs and found a close agreement between the two methods (29-31). PDFF mapping is already widely applied to assess ectopic fat changes in liver and muscle after weight loss interventions (23,32). A PDFF mapping acquisition typically anatomically covers both the abdominal organs and SAT. To our knowledge, no clinical studies have yet compared these techniques for estimating SAT PDFF in an obesity cohort undergoing a weight loss intervention. Using SAT PDFF mapping, valuable information on SAT composition can be obtained in addition to the information regarding ectopic fat changes, which may facilitate targeted and personalized obesity interventions.

The purpose of this study was to (I) compare MRI and MRS methods while evaluating the physiological variations in the SAT PDFF in people with obesity before and after an 8-week weight loss intervention; (II) compare the PDFF variation between the SSAT and DSAT; and (III) analyze the PDFF changes in correspondence to standard anthropometric markers such as weight, BMI and body fat. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-7/rc).

Methods

Study design and participants

Between October 2019 and October 2021, 127 persons with obesity were recruited from the LION study (33) to undergo an MRI examination of the abdomen and pelvis on a 3T scanner at baseline and after an 8-week formula-based low-calorie diet of 800 kilocalories with an optional additional daily intake of 200 g non-starchy vegetables. The following analysis is based on the data obtained from abdominopelvic MRI scans of 69 participants at the beginning of dietary intervention (baseline) and 63 participants who received a second scan at the end of the intervention (after weight loss). An overview of the participant MRI and MRS data used in the final analyses is presented in Figure 1.

Figure 1 Flow chart of the MRI and MRS data gathering process for this study. Not all participants who received an MRI scan at baseline completed a scan at the end of intervention due to attrition (participant withdrawal, illness, or COVID-19 pandemic-related restrictions). Additionally, for certain participants, MRS data at baseline and/or after weight loss were not collected due to shortened scan protocols caused by participant discomfort or scanning issues. Technical difficulties during data export, post-processing, or analysis led to partial or complete exclusion of some MRI/MRS data. For anthropometric analyses, participants were excluded if either (I) weight measurements taken at different departmental appointments varied by more than 2.5 kg, beyond expected daily variation, and/or (II) SSAT PDFF measurements were not possible due to very low subcutaneous adipose tissue volume, which made ROI placement unfeasible. The final number of participants included in the MRI and MRS analyses is indicated in the outlined boxes in the chart. COVID-19, coronavirus disease 2019; DSAT, deep subcutaneous adipose tissue; MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; PDFF, proton density fat fraction; ROI, region of interest; SSAT, superficial subcutaneous adipose tissue.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study protocol and procedures were approved by the ethical committee of the School of Medicine and Health of the Technical University of Munich, Germany (project number 69/19S; ClinicalTrials.gov identifier: NCT04023942). Written informed consent was obtained from all participants. Inclusion and exclusion criteria were defined as outlined in the LION study (33), while standard MRI contraindications served as additional exclusion criteria.

Anthropometric measurements

Body weight was assessed in a fasted state in light clothing without shoes prior to each MRI scan using a scale (MPD 250K1 00M, Kern and Sohn, Germany) at the radiology department. To account for the light clothing, one kilogram was deducted from the analyses in determining weight (kg). Body fat (%) was measured using bioelectrical impedance analysis based on a body composition scale (BC-418MA, Tanita Europe B.V., Netherlands) during separately scheduled appointments at baseline and after weight loss at the Institute for Nutritional Medicine (33). Weight (kg) was also recorded separately at these appointments by the Institute for Nutritional Medicine. Height was measured in a standing position without shoes using a stadiometer (Seca 214, Seca, Hamburg, Germany) (33). The BMI was calculated as the weight in kilograms divided by height in meters squared (kg/m²) (33).

Magnetic resonance measurements

Participants were scanned in a supine position on a 3T MR scanner (Ingenia, Philips Healthcare, Netherlands) using a 16-channel torso coil and the built-in-table 12-channel posterior coil over the abdomen and pelvis at baseline and at the end of dietary intervention. The imaging protocol included Dixon acquisitions for MRI-based SAT PDFF assessment and single-voxel MRS acquisitions for MRS-based SAT PDFF assessment.

For MRI-based SAT PDFF assessment, a six-echo multi-echo three-dimensional (3D) bipolar gradient echo imaging sequence under breath-hold at end-expiration was used with bipolar readout gradients and the following sequence parameters: Repetition Time (TR) =7.0 ms; Echo Time (TE)₁ =1.14 ms; ΔTE =0.8 ms; flip angle =3°; bandwidth =2367.4 Hz; 132×180 matrix size; Field of view (FOV) =400×543.4×144 mm³; voxel size =3.0×3.0×6.0 mm³; combination of compressed sensing (CS) and sensitivity encoding (SENSE) CS-SENSE with R=3.5. Four stacks of axial images (the upper stack beginning at the liver dome) were acquired and merged. The acquisition time for each stack was 10.3s. PDFF maps were generated based on the fat quantification algorithm (mDixon quant) provided by the vendor (Philips Healthcare). The online complex-based water-fat quantification algorithm used a single effective transverse relaxation time (T2*) correction, a pre-calibrated fat spectrum accounting for the presence of multiple fat peaks and a phase correction to address eddy-current-induced phase errors (34).

For MRS-based SAT PDFF assessment, MRS was performed using a single-voxel stimulated echo acquisition mode (STEAM) sequence in the DSAT using the following acquisition parameters: repetition time (TR) =5,000 ms; mixing time (TM) =17 ms; TE =11, 15, 20, 25 ms; flip angle =90°; spectral bandwidth =5,000 Hz; 4,096 samples; 8 averages; scan time =3 min. A 15×15×15 mm³ voxel was placed in the DSAT depot at the level of the 5th lumbar spine vertebra (L5), excluding major vessels. The long TR was used to reduce T1-weighting effects. The multi-echo acquisition allowed for the correction of transverse relaxation time (T2) weighting effects.

MRI and MRS data analysis

To compare MRI-derived SAT PDFF with MRS-derived SAT PDFF values, the corresponding region of interest (ROI) from the imaging-based PDFF maps was automatically matched with the location of the voxel from MRS using Matlab (R2019b, MathWorks, Natick, Massachusetts, USA) (Figure 2). The mean imaging-based PDFF in the ROI was computed to compare with the MRS-derived PDFF value. The imaging-based PDFF of SSAT and DSAT was obtained using the brush tool in the free open-source software ITK-SNAP (Version 3.8.0, www.itksnap.org) (35). Two 3D isotropic spherical ROIs (DSAT: total volume, 22.9 cm³; SSAT: total volume, 0.8 cm³) were manually placed bilaterally on the posterior sides of the fat depots in the axial MRI PDFF maps. The voxels were positioned at the level of the L5 in SSAT and DSAT, respectively (Figure 2). Large vessels, fascia, and other surrounding tissue types were avoided to minimize partial volume effects. The manual segmentations were reviewed to ensure a precise definition of the target area with a board-certified radiologist. The mean PDFF value of each fat depot was taken from the PDFF maps at baseline and after weight loss.

Figure 2 PDFF maps of subcutaneous adipose tissue regions with ROI placement for fat fraction quantification. (A) Extraction of MRI-based SAT PDFF values matching the corresponding MRS voxel in one participant (female, 31 years) at baseline (left) and after weight loss (right). The mean PDFF as measured by the voxel of interest in the posterior deep SAT (red) matching the corresponding MRS voxel location decreased by 0.6% (from 95.4% to 94.8%) at the end of dietary intervention. The patient lost a total of 9.5 kg (baseline weight: 79.0 kg, post-intervention weight: 69.5 kg). (B) Extraction of PDFF values from deep SAT and superficial SAT regions based on axial MRI-based PDFF maps of one participant (female, 93.1 kg, 50 years) at baseline. Two three-dimensional circular ROI voxels (deep SAT: total volume, 22.9 cm³; superficial SAT: total volume, 0.8 cm³) represented here as red voxels were placed posteriorly at the level of the 5^th lumbar spine vertebra in the deep SAT (left) and superficial SAT (right), respectively. The mean PDFF measured by the ROIs was 94.5% in the deep SAT and 94.7% in the superficial SAT. The fascia separating the superficial from the deep fat depot can be seen as a faint green outline as indicated by the red arrow. MRI, magnetic resonance imaging; SAT, subcutaneous adipose tissue; PDFF, proton density fat fraction; MRS, magnetic resonance spectroscopy; ROI, region of interest.

The MRS data of progressively longer echo times (11, 15, 20, 25 ms) were jointly fitted using a frequency domain-based in-house written Matlab routine. Fat peaks were characterized based on an adapted triglyceride model from Hamilton et al. (36). All fat peaks and the water peak at 4.7 parts per million (ppm) were modeled as single Gaussian resonances. Lipid peaks were delineated at 0.9, 1.3, 1.6, 2.02, 2.24, 2.75, 4.1, 4.27, 5.29 ppm. All fat peaks and the water peak were constrained to have equal linewidths (MRS constrained linewidth) to match the single T2* assumption of the Dixon water-fat separation algorithm that was applied in imaging (Figure 3). The T2 and T2-corrected peak areas were estimated by a nonlinear least-squares fitting approach, using the data from the multi-TE acquisition. The MRS-based PDFF was computed as a ratio of the T2-corrected fat peak area over the sum of the T2-corrected water peak area and T2-corrected fat peak area.

Figure 3 Example of the measured spectra (here showed spectrum at TE =11 ms) of one patient (female, 77.6 kg, 37 years) at baseline (77.6 kg) and after weight loss (67.8 kg). The water peak energy is increased from baseline to after dietary intervention (blue arrows). Lipid peaks were delineated at 0.9 ppm (A), 1.3 ppm (B), 1.6 ppm (C), 2.02 ppm (D), 2.24 ppm (E), 2.75 ppm (F), 4.1 ppm (G), 4.27 ppm (H) and 5.29 ppm (J). ppm, parts per million; TE, echo time.

Statistical analysis

Statistical analyses were performed using the Python programming language (Software Version 3.6, Python Software Foundation, United States) using SciPy (37). Statistical figures were generated using the Matplotlib and Seaborn libraries (38,39) in Python and formatted in Microsoft PowerPoint (Version 16.83, 2024). The absolute PDFF agreement between the MRS and MRI methods was analyzed with Bland-Altman plots, including mean difference and limits of agreement. Study data distribution was assessed for normality using the Shapiro-Wilk test. The distribution type (normal or non-normal) for each subgroup is reported in Tables 1-3. A paired-sample t-test was used to evaluate the changes in outcome measures at the end of intervention for normally distributed data. Non-normally distributed data was evaluated with the Wilcoxon one-sample sized rank test. For DSAT- and SSAT PDFF comparison, independent t-tests were applied for normally distributed data. Otherwise, the Mann-Whitney U test was used. Correlation analyses were performed using Pearson correlation coefficient when both parameters showed a normal distribution. When one or both parameters were not normally distributed, Spearman’s Rank correlation coefficient was used. For descriptive statistics, normally distributed data were reported as mean ± standard deviation (SD), whereas non-normally distributed data were presented as median and interquartile range (IQR). Additional partial correlation analyses were performed, adjusting for age, sex, and time intervals between measurements. These analyses are available in Appendix 1. A P value ≤0.05 was considered statistically significant.

Results

Study cohort and subsamples

The initial study cohort consisted of 127 participants: 73 females (57.5%) and 54 males (42.5%). Participants had a median age of 46.6 years (IQR, 15.6 years) and a median weight of 99.7 kg (IQR, 19.2 kg). The mean BMI was 34.4±2.9 kg/m² and the median body fat percentage was 40.2% (IQR, 11.6%).

The number of participants with MRI- and MRS-based PDFF measurements varied across time points due to study attrition/scan-related/post-processing exclusions (Figure 1). For each imaging modality, the final study subsamples were as follows:

• MRI-based PDFF:
  • Baseline: 57 participants;
  • Post-weight loss: 52 participants;
  • Matched baseline and post-weight loss measurements: 37 participants.
• MRS-based PDFF:
  • Baseline: 69 participants;
  • Post-weight loss: 63 participants;
  • Matched baseline and post-weight loss measurements: 59 participants.

Based on the final participant data included in the analyses, the median time interval between the MRI scan and the appointment at the nutritional department was 7 days (IQR, 5 days) for the 69 participants at baseline, and 2 days (IQR, 3 days) for the 63 participants after weight loss.

The demographic and clinical characteristics of the study participants in the subsamples are described in Table 1 (participants scanned at baseline), Table 2 (participants scanned after weight loss), and Table 3 (participants scanned at both time points) using measurements of central tendency and dispersion (mean, median; SD, IQR). Participants lost an average of 10.5±2.8 kg. After weight loss, BMI decreased by 3.9±0.8 kg/m², and median body fat percentage decreased by 4.0% (IQR, 2.1%).

The analysis comparing MRI- and MRS-based PDFF before and after dietary intervention was conducted using the data of 48 participants. For the analysis of DSAT PDFF, a subset of 40 participants (29 females, 11 males; mean age, 46 years) was analyzed. As anthropometric analyses were also conducted, participants were excluded if the weight measurements recorded at the separate appointments by the radiology department and Institute for Nutritional Medicine differed by more than 2.5 kg, considering standard weight fluctuations. Furthermore, ROI placement was not feasible in cases where SSAT thickness was less than the ROI voxel diameter, resulting in the exclusion of these individuals. Therefore, SSAT PDFF measurements were limited to 35 participants (29 females, 6 males; mean age, 46 years).

Agreement between MRI- and MRS-based SAT PDFF

The SAT PDFF quantified by MRS was reflective of the PDFF measured by MRI both at baseline (r=0.67, P<0.01) and after weight loss (r=0.81, P<0.01) (Figure 4). The SAT PDFF change measured with MRS correlated strongly with MRI-based SAT PDFF change (r=0.75, P<0.01) (Figure 5). The Bland-Altman plots revealed a mean bias of +3.42% (baseline) and +3.75% (after weight loss) using MRS in comparison to MRI over the measured SAT PDFF range (Figure 6).

Figure 4 Scatter plots with linear regression lines comparing MRI- and MRS-based subcutaneous adipose tissue proton density fat fraction measurements at baseline and after weight loss (baseline R²=0.45, after weight loss R²=0.66). MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy.

Figure 5 Scatter plot with linear regression line comparing MRI- and MRS-based SAT PDFF measurement changes from baseline and after weight loss. The SAT PDFF change measured with MRS was reflective of the PDFF change of MRI (R²=0.56). MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; PDFF, proton density fat fraction; SAT, subcutaneous adipose tissue.

Figure 6 Bland-Altman plots of SAT PDFF measurements comparing MRI and MRS methods at baseline (A) and after weight loss (B) (n=48). Horizontal lines indicate LOA (green) and the mean difference (red) between the two methods, with LOA at ±1.96 SD. MRS tends to overestimate PDFF when compared to MRI. The Bland-Altman plots show a PDFF bias of −3.42% with a 95% LOA ranging from −5.17% to −1.67% at baseline and a bias of −3.75% with a 95% LOA ranging from −5.47% to −2.03% after weight loss. LOA, limits of agreement; MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; PDFF, proton density fat fraction; SAT, subcutaneous adipose tissue; SD, standard deviation.

PDFF analysis in SSAT and DSAT

Both SSAT and DSAT PDFF values decreased after the intervention. At the end of the dietary intervention, participants had a mean reduction of SSAT PDFF of 1.0%±0.9%, and a median reduction of DSAT PDFF of 0.9% (range: −2.4% to −0.04%) (Figure 7). The absolute PDFF decrease was statistically significant for SSAT and DSAT (P<0.01). Furthermore, correlation analyses found a positive association between SSAT and DSAT PDFF at baseline (r=0.39, P=0.02) and after weight loss (r=0.46, P<0.05). The comparison of PDFF difference between DSAT and SSAT was statistically significant after weight loss (P<0.01), but not at baseline (P=0.08).

Figure 7 Boxplots comparing distribution of MRI-based DSAT and SSAT region of interest PDFF measurements as absolute measurements at baseline and after weight loss (A) and as absolute change at the end of dietary intervention (B). The mean PDFF measured 94.2%±1.2% in SSAT and 93.7%±0.8% in DSAT at baseline. This decreased to 93.2%±1.3% and 92.7%±0.8% respectively after weight loss (values are denoted as mean ± standard deviation). There was a significant mean reduction of SSAT PDFF by 1.0%±0.9% (t-test: P<0.05) and median reduction of DSAT PDFF by 0.9% (−2.4% to −0.04%, Mann-Whitney test: P<0.001). DSAT, deep subcutaneous adipose tissue; MRI, magnetic resonance imaging; PDFF, proton density fat fraction; SSAT, superficial subcutaneous adipose tissue.

SAT PDFF change and correlation with weight change

After weight loss, a decrease in SAT PDFF was observed, which was associated with weight loss (kg). There was a significant decrease in the MRI-based PDFF by an average of 1.4%±1.6%, and MRS-based PDFF by 1.1%±0.9% (P<0.01). Body weight loss (kg) correlated significantly with decreases in both MRI- and MRS-based PDFF (MRI: r=0.41, P=0.01; MRS: r=0.38, P=0.01) (Figure 8). Body weight loss also significantly correlated with decreases in DSAT- and SSAT PDFF (DSAT: r=0.34, P<0.01; SSAT: r=0.48, P<0.05).

Figure 8 Scatter plots comparing weight change with MRI- and MRS-based SAT PDFF change (A), and with deep- and superficial SAT PDFF change measurements after dietary intervention (B). Linear regression lines are shown only for superficial SAT in (B). Weight change positively correlated with SAT PDFF change. MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; PDFF, proton density fat fraction; SAT, subcutaneous adipose tissue.

SAT PDFF change and correlation with other anthropometric markers

The decrease in SAT PDFF after the 8-week dietary intervention significantly correlated with reductions in other anthropometric parameters, including BMI (r=0.44, P<0.01) and body fat percentage (r=0.59, P<0.01). For participants with a SAT PDFF loss of more than 2%, a greater variability in BMI or body fat percentage decrease was observed, as seen in the broad scattering above this range (Figure 9).

Figure 9 Scatter plots comparing MRI-based SAT PDFF change with change in BMI (kg/m²) and body fat (%) after dietary intervention. MRI-based PDFF correlated significantly with changes in both anthropometric markers. BMI, body mass index; MRI, magnetic resonance imaging; PDFF, proton density fat fraction; SAT, subcutaneous adipose tissue.

Discussion

The present study employs MRI and MRS techniques to quantify SAT PDFF in people with obesity undergoing an 8-week formula-based weight loss intervention. Results demonstrated an inter-method agreement between SAT PDFF measurements obtained with MRI and MRS. Both the MRI-based and the MRS-based measurements showed a reduction in SAT PDFF after weight loss, with stronger SAT PDFF changes in participants who lost the most weight. Similar PDFF changes were observed in both the superficial and the deep layers of SAT after weight loss. The findings support the use of MRI and MRS as clinically reliable approaches for assessing SAT PDFF changes in the context of obesity and weight loss interventions.

The reported correlation in SAT between MRI-based and MRS-based PDFF measurements is consistent with previous literature comparing the two methods in humans across several organs, including the liver (29), the pancreas (40), and the bone marrow (31). Our findings also indicate that MRS tends to overestimate PDFF relative to MRI measurements. This difference may be attributable, at least in part, to the adoption of a model with a single linewidth for the water peak, equal to the linewidth of the fat peak, in the employed MRS fat spectrum fitting, which may not have captured the full energy of the actual water peak. This limitation in modeling could have led to a slight underestimation of the water fraction and consequently, to an overestimation of the fat fraction, hence the higher PDFF values. On the one hand, the employed water-fat signal model in the MRI-based PDFF quantification approach assumed a common T2* decay for both the water and fat signals, as typically performed in the analysis of multi-echo acquisitions with a limited number of echoes (41,42). The assumption of a common T2* value has been debated for certain experimental scenarios (43,44). Neglecting to resolve for independent T2* decay rates of water and fat could induce a bias in the PDFF, which is expected to be limited while using a low number of echoes in MRI-based PDFF quantification. On the other hand, a recent study has shown that the water component within adipose tissue is affected by pronounced microscopic field inhomogeneities, which induce a broad spectral distribution of water signals (45). It has been postulated that the water signal within adipose tissue shows a fast, non-exponential signal decay (45). However, it remains unclear what the optimal composite peak shape would be to best fit the water signal within an in vivo adipose tissue spectrum, especially in a multi-TE MRS acquisition while correcting for T2 effects. In addition, multi-echo gradient echo imaging and single-voxel MRS inherently sample data at different echo times. Therefore, all fat and water peaks were constrained to have equal linewidths in the MRS processing to match the single T2* assumption of the MRI-based fat quantification. Whether the small disparities in PDFF between the MRI-based and the MRS-based methods, relative to the true PDFF, hold clinical relevance in the context of intervention monitoring remains uncertain. Nevertheless, our findings indicate that MRI- and MRS-derived PDFF could be a potential marker for monitoring SAT PDFF changes. PDFF mapping techniques are already included in MR protocols used to track weight loss changes in ectopic fat depots (e.g., liver and muscle). The SAT region is typically included in the field-of-view of MRI scans, focusing on liver and paraspinal muscle PDFF changes, and SAT PDFF could therefore be easily extracted from the same data.

A statistically significant decrease in MRI- and MRS-based SAT PDFF after dietary intervention was observed, and the decrease in MRI- and MRS-based SAT PDFF was associated with body weight loss. It is important to note that SAT volume changes have previously been reported from the same cohort (21,46). The present analysis shows that not only SAT volume but also SAT composition changes with weight loss. Several previous weight loss studies have reported an increased adipose tissue hydration post-intervention, which corresponds to our observation of decreased PDFF. In one study, O₂ extraction [deoxygenated hemoglobin (HbR)] and water content increased after weight loss in SAT (47). Laaksonen et al. attributed the increased adipose tissue water content after a 9-week caloric restriction to improved blood flow and insulin sensitivity (48). Additionally, higher partial pressure of oxygen (pO₂) levels were found in the SAT of individuals with obesity in contrast to non-obese counterparts, suggesting an impaired capacity for O₂ extraction (49). This was also associated with lower blood flow, lower capillarization, insulin resistance, and increased expression of inflammatory cell markers within adipose tissue. Caloric restriction-induced adipocyte shrinkage results in a shortened diffusion distance to mitochondria, facilitating more efficient O₂ extraction and consequently a relative decrease in lipid-to-water volume ratio (47). Furthermore, caloric restriction leads to a greater relative loss of lipid volume, which may potentially contribute to increased water hydration or a decrease in PDFF (47).

Given the structural and metabolic differences between DSAT and SSAT (10,11), differences in PDFF between both SAT compartments following weight loss were studied. Our results show that SSAT PDFF is significantly higher than that of DSAT after the dietary intervention. This is likely reflective of the distinctive characteristics of DSAT and SSAT at both macroscopic and microscopic levels. Anatomical dissections of cadavers have reported small, well-arranged and tightly packed fat lobules in SSAT, in contrast to the larger, irregularly packed lobules found in DSAT (9). Abdominal DSAT also exhibits higher metabolic activity, a higher inflammatory profile and a greater degree of saturation compared to abdominal SSAT (11,12). Elevated adipocyte saturation, indicating a higher ratio of saturated to monounsaturated fatty acids, has been linked to insulin resistance (50).

As expected, the anthropometric markers, including weight, BMI and body fat percentage, significantly correlated with changes in SAT PDFF. Among these, body fat loss showed the strongest association with PDFF decrease. The significant correlations observed between SAT PDFF and anthropometric variables provide additional validation for the application of PDFF measurements in weight loss interventions.

The disproportionately higher representation of female participants in our study population presents challenges in delineating gender-specific differences and therefore, limits the generalizability of our findings. The role of age and sex in influencing the outcomes of lifestyle interventions remains uncertain, necessitating larger-scale studies for more conclusive insights. Given the well-documented sexual dimorphism in fat distribution and sex hormone profiles between males and females (51), it is worth further investigating the potential influence of age and sex on changes in adipose tissue fat fraction during weight loss interventions.

This study has several limitations. First, the use of an identical T2* for fat and water in MRI may have caused inaccuracies in PDFF measurements of adipose tissue, as discussed in detail above. Second, the results of this study were based on manually drawn ROIs in one two-dimensional (2D) slice at the level of L5, which may not fully account for the heterogeneity of cell components within the entire SAT depot, and the small size of the ROI used in SSAT may have led to skewing of the mean PDFF value. In contrast, the DSAT is thicker and primarily dictates adipose thickness at the L5 region, making ROI measurements in this area more accurate. While it was possible to place a ROI in the DSAT of all participants, several individuals had fat distributions with an extremely thin or low volume of SSAT. This made placing an ROI in this area either challenging, due to partial voluming effects at the borders of the ROI, or impossible, resulting in the omission of the respective SSAT PDFF measurement. The use of whole volume segmentation is a possible solution that may provide a more representative measure of mean PDFF. Further, the time interval between the scans at baseline and after weight loss did not meet the targeted 8-week duration as intended for every participant (time interval for 48 participants with successful baseline/post-weight loss measurements of MRI- and MRS PDFF: mean, 60.5 days; median, 59 days; range, 49–84 days). This was caused by delays due to scanning restrictions enforced due to temporary hospital measures enacted in response to the coronavirus disease 2019 (COVID-19) pandemic or participant illness. This variation may have contributed to the broader ranges observed in recorded weight losses and PDFF changes. Additionally, not all subjects received their MRI and anthropometric measurements on the same day, which may have introduced minor discrepancies between PDFF and anthropometric measurements. Another limitation of this study was the absence of a control group. However, the goal of the present study was to study the effect of weight loss in an obese group. Finally, while the present study offers valuable insights based on a relatively small sample size, additional studies are needed to reinforce these findings and fully explore the usefulness of SAT PDFF for better metabolic phenotyping and effective risk stratification in people with obesity.

Conclusions

In summary, our study shows that SAT PDFF decreases in individuals with obesity undergoing a standardized weight loss intervention, based on both MRI and MRS measurements. Our present findings also show that weight loss leads to the reduction of both SSAT and DSAT PDFF in persons with obesity. The SSAT and DSAT PDFF correlated strongly at baseline and after weight loss, and SSAT had a significantly higher PDFF than DSAT after weight loss. The decrease in SAT PDFF after weight loss indicates an increase in adipose tissue hydration, which may be explained in part by improvements in blood flow and tissue microstructural changes. The SAT PDFF is also a strong predictor of anthropometric markers of obesity. Based on this study, the measurement of adipose tissue fat fraction has the potential to serve as an additional biomarker for monitoring the effects of weight change and metabolic response in people with obesity, especially while employing already MRI-based PDFF mapping methods to assess ectopic fat changes.

Acknowledgments

The authors would like to thank Arun Somasundaram, Sandhanakrishnan Ravichandran and Johannes Raspe for their help and input in the data cleaning process and Lisa Patzelt for her help with the MRI scanning. Furthermore, the authors are grateful to all participants of the LION Study and to all members of the LION study team, especially Meike Wiechert, Vincent Winkler, Miriam Neidhardt, Bea Klos, Sandra Bayer, Judith Bodensteiner, Christine Reimers, Christina Ikkert, Andrea Stiglmeier, Alexandra Sandner, Bärbel Huber, and Kurt Rack. We thank the Munich Study Center for support in data management.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-7/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-7/dss

Funding: The LION-Study was funded by the German Federal Ministry of Education and Research (BMBF) (No. 01EA1709) within the framework of the Junior Research Group “Personalized Nutrition & eHealth (PeNut)” of the enable Competence Cluster of Nutrition Research. In addition, the analysis was part of the project “Imaging, metabolic risk, and genetics: Algorithms based on Artificial Intelligence to predict metabolic changes through Lifestyle Intervention (IMaGENE)” funded by BMBF (No. 16DKWN075). Further, the present work was supported by the German Research Foundation (Nos. 450799851 and 455422993/FOR5298-iMAGO-P1). The authors from the Department of Radiology also acknowledge research support from Philips Healthcare.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-7/coif). D.C.K. receives grant support from Philips Healthcare. H.H. is a member of the scientific advisory board of Oviva AG (Zurich, Switzerland) and C.H. of 4sigma GmbH (Oberhaching, Germany). H.H. and C.H. received speaker honoraria from Novo Nordisk (Copenhagen, Denmark). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.The study protocol and procedures were approved by the ethical committee of the School of Medicine and Health of the Technical University of Munich, Germany (project number 69/19S; ClinicalTrials.gov identifier: NCT04023942). Written informed consent was obtained from all participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. WHO European Regional Obesity Report 2022 [cited 2024 Mar 8]. Available online: https://www.who.int/europe/publications/i/item/9789289057738
2. Kawai T, Autieri MV, Scalia R. Adipose tissue inflammation and metabolic dysfunction in obesity. Am J Physiol Cell Physiol 2021;320:C375-91. [Crossref] [PubMed]
3. Honecker J, Ruschke S, Seeliger C, Laber S, Strobel S, Pröll P, Nellaker C, Lindgren CM, Kulozik U, Ecker J, Karampinos DC, Claussnitzer M, Hauner H. Transcriptome and fatty-acid signatures of adipocyte hypertrophy and its non-invasive MR-based characterization in human adipose tissue. EBioMedicine 2022;79:104020. [Crossref] [PubMed]
4. Mozaffarian D, Benjamin EJ, Go AS, Arnett DK, Blaha MJ, Cushman M, et al. Heart disease and stroke statistics--2015 update: a report from the American Heart Association. Circulation 2015;131:e29-322. Erratum in: Circulation 2015;131:e535. Erratum in: Circulation 2016;133:e417. [Crossref] [PubMed]
5. Paley CA, Johnson MI. Abdominal obesity and metabolic syndrome: exercise as medicine? BMC Sports Sci Med Rehabil 2018;10:7. [Crossref] [PubMed]
6. Hamman RF, Wing RR, Edelstein SL, Lachin JM, Bray GA, Delahanty L, Hoskin M, Kriska AM, Mayer-Davis EJ, Pi-Sunyer X, Regensteiner J, Venditti B, Wylie-Rosett J. Effect of weight loss with lifestyle intervention on risk of diabetes. Diabetes Care 2006;29:2102-7. [Crossref] [PubMed]
7. Wing RR, Lang W, Wadden TA, Safford M, Knowler WC, Bertoni AG, Hill JO, Brancati FL, Peters A, Wagenknecht LLook AHEAD Research Group. Benefits of modest weight loss in improving cardiovascular risk factors in overweight and obese individuals with type 2 diabetes. Diabetes Care 2011;34:1481-6. [Crossref] [PubMed]
8. Taksali SE, Caprio S, Dziura J, Dufour S, Calí AM, Goodman TR, Papademetris X, Burgert TS, Pierpont BM, Savoye M, Shaw M, Seyal AA, Weiss R. High visceral and low abdominal subcutaneous fat stores in the obese adolescent: a determinant of an adverse metabolic phenotype. Diabetes 2008;57:367-71. [Crossref] [PubMed]
9. Markman B, Barton FE Jr. Anatomy of the subcutaneous tissue of the trunk and lower extremity. Plast Reconstr Surg 1987;80:248-54. [Crossref] [PubMed]
10. Cancello R, Zulian A, Gentilini D, Maestrini S, Della Barba A, Invitti C, Corà D, Caselle M, Liuzzi A, Di Blasio AM. Molecular and morphologic characterization of superficial- and deep-subcutaneous adipose tissue subdivisions in human obesity. Obesity (Silver Spring) 2013;21:2562-70. [Crossref] [PubMed]
11. Lundbom J, Hakkarainen A, Lundbom N, Taskinen MR. Deep subcutaneous adipose tissue is more saturated than superficial subcutaneous adipose tissue. Int J Obes (Lond) 2013;37:620-2. [Crossref] [PubMed]
12. Marinou K, Hodson L, Vasan SK, Fielding BA, Banerjee R, Brismar K, Koutsilieris M, Clark A, Neville MJ, Karpe F. Structural and functional properties of deep abdominal subcutaneous adipose tissue explain its association with insulin resistance and cardiovascular risk in men. Diabetes Care 2014;37:821-9. [Crossref] [PubMed]
13. Chooi YC, Ding C, Magkos F. The epidemiology of obesity. Metabolism 2019;92:6-10. [Crossref] [PubMed]
14. Blüher M. Obesity: global epidemiology and pathogenesis. Nat Rev Endocrinol 2019;15:288-98. [Crossref] [PubMed]
15. Fox CS, Massaro JM, Hoffmann U, Pou KM, Maurovich-Horvat P, Liu CY, Vasan RS, Murabito JM, Meigs JB, Cupples LA, D'Agostino RB Sr, O'Donnell CJ. Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham Heart Study. Circulation 2007;116:39-48. [Crossref] [PubMed]
16. Storz C, Heber SD, Rospleszcz S, Machann J, Sellner S, Nikolaou K, Lorbeer R, Gatidis S, Elser S, Peters A, Schlett CL, Bamberg F. The role of visceral and subcutaneous adipose tissue measurements and their ratio by magnetic resonance imaging in subjects with prediabetes, diabetes and healthy controls from a general population without cardiovascular disease. Br J Radiol 2018;91:20170808. [Crossref] [PubMed]
17. Crabtree CD, LaFountain RA, Hyde PN, Chen C, Pan Y, Lamba N, Sapper TN, Short JA, Kackley ML, Buga A, Miller VJ, Scandling D, Andersson I, Barker S, Hu HH, Volek JS, Simonetti OP. Quantification of Human Central Adipose Tissue Depots: An Anatomically Matched Comparison Between DXA and MRI. Tomography 2019;5:358-66. [Crossref] [PubMed]
18. Machann J, Thamer C, Stefan N, Schwenzer NF, Kantartzis K, Häring HU, Claussen CD, Fritsche A, Schick F. Follow-up whole-body assessment of adipose tissue compartments during a lifestyle intervention in a large cohort at increased risk for type 2 diabetes. Radiology 2010;257:353-63. [Crossref] [PubMed]
19. Haueise T, Schick F, Stefan N, Schlett CL, Weiss JB, Nattenmüller J, et al. Analysis of volume and topography of adipose tissue in the trunk: Results of MRI of 11,141 participants in the German National Cohort. Sci Adv 2023;9:eadd0433. [Crossref] [PubMed]
20. Küstner T, Hepp T, Fischer M, Schwartz M, Fritsche A, Häring HU, Nikolaou K, Bamberg F, Yang B, Schick F, Gatidis S, Machann J. Fully Automated and Standardized Segmentation of Adipose Tissue Compartments via Deep Learning in 3D Whole-Body MRI of Epidemiologic Cohort Studies. Radiol Artif Intell 2020;2:e200010. [Crossref] [PubMed]
21. Somasundaram A, Wu M, Reik A, Rupp S, Han J, Naebauer S, Junker D, Patzelt L, Wiechert M, Zhao Y, Rueckert D, Hauner H, Holzapfel C, Karampinos DC. Evaluating Sex-specific Differences in Abdominal Fat Volume and Proton Density Fat Fraction at MRI Using Automated nnU-Net-based Segmentation. Radiol Artif Intell 2024;6:e230471. [Crossref] [PubMed]
22. Estrada S, Lu R, Conjeti S, Orozco-Ruiz X, Panos-Willuhn J, Breteler MMB, Reuter M. FatSegNet: A fully automated deep learning pipeline for adipose tissue segmentation on abdominal dixon MRI. Magn Reson Med 2020;83:1471-83. [Crossref] [PubMed]
23. Vogt LJ, Steveling A, Meffert PJ, Kromrey ML, Kessler R, Hosten N, Krüger J, Gärtner S, Aghdassi AA, Mayerle J, Lerch MM, Kühn JP. Magnetic Resonance Imaging of Changes in Abdominal Compartments in Obese Diabetics during a Low-Calorie Weight-Loss Program. PLoS One 2016;11:e0153595. [Crossref] [PubMed]
24. Franz D, Weidlich D, Freitag F, Holzapfel C, Drabsch T, Baum T, Eggers H, Witte A, Rummeny EJ, Hauner H, Karampinos DC. Association of proton density fat fraction in adipose tissue with imaging-based and anthropometric obesity markers in adults. Int J Obes (Lond) 2018;42:175-82. [Crossref] [PubMed]
25. Reeder SB, Hu HH, Sirlin CB. Proton density fat-fraction: a standardized MR-based biomarker of tissue fat concentration. J Magn Reson Imaging 2012;36:1011-4. [Crossref] [PubMed]
26. Scotti A, Tain RW, Li W, Gil V, Liew CW, Cai K. Mapping brown adipose tissue based on fat water fraction provided by Z-spectral imaging. J Magn Reson Imaging 2018;47:1527-33. [Crossref] [PubMed]
27. Hamilton G, Smith DL Jr, Bydder M, Nayak KS, Hu HH. MR properties of brown and white adipose tissues. J Magn Reson Imaging 2011;34:468-73. [Crossref] [PubMed]
28. Reeder SB, Cruite I, Hamilton G, Sirlin CB. Quantitative Assessment of Liver Fat with Magnetic Resonance Imaging and Spectroscopy. J Magn Reson Imaging 2011;34:729-49. [Crossref] [PubMed]
29. Yokoo T, Serai SD, Pirasteh A, Bashir MR, Hamilton G, Hernando D, Hu HH, Hetterich H, Kühn JP, Kukuk GM, Loomba R, Middleton MS, Obuchowski NA, Song JS, Tang A, Wu X, Reeder SB, Sirlin CBRSNA-QIBA PDFF Biomarker Committee. Linearity, Bias, and Precision of Hepatic Proton Density Fat Fraction Measurements by Using MR Imaging: A Meta-Analysis. Radiology 2018;286:486-98. [Crossref] [PubMed]
30. Vu KN, Gilbert G, Chalut M, Chagnon M, Chartrand G, Tang A. MRI-determined liver proton density fat fraction, with MRS validation: Comparison of regions of interest sampling methods in patients with type 2 diabetes. J Magn Reson Imaging 2016;43:1090-9. [Crossref] [PubMed]
31. Karampinos DC, Melkus G, Baum T, Bauer JS, Rummeny EJ, Krug R. Bone marrow fat quantification in the presence of trabecular bone: initial comparison between water-fat imaging and single-voxel MRS. Magn Reson Med 2014;71:1158-65. [Crossref] [PubMed]
32. Feng X, Lin Y, Zhuo S, Dong Z, Shao C, Ye J, Zhong B. Treatment of obesity and metabolic-associated fatty liver disease with a diet or orlistat: A randomized controlled trial. Am J Clin Nutr 2023;117:691-700. [Crossref] [PubMed]
33. Reik A, Holzapfel C. Randomized Controlled Lifestyle Intervention (LION) Study for Weight Loss and Maintenance in Adults With Obesity-Design and Methods. Front Nutr 2020;7:586985. [Crossref] [PubMed]
34. Eggers H, Brendel B, Duijndam A, Herigault G. Dual-echo Dixon imaging with flexible choice of echo times. Magn Reson Med 2011;65:96-107. [Crossref] [PubMed]
35. Yushkevich PA, Piven J, Hazlett HC, Smith RG, Ho S, Gee JC, Gerig G. User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability. Neuroimage 2006;31:1116-28. [Crossref] [PubMed]
36. Hamilton G, Schlein AN, Middleton MS, Hooker CA, Wolfson T, Gamst AC, Loomba R, Sirlin CB. In vivo triglyceride composition of abdominal adipose tissue measured by (1) H MRS at 3T. J Magn Reson Imaging 2017;45:1455-63. [Crossref] [PubMed]
37. Virtanen P, Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, et al. SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat Methods 2020;17:261-72. [Crossref] [PubMed]
38. Hunter JD. Matplotlib: A 2D graphics environment. Comput Sci Eng 2007;9:90-5.
39. Waskom M. seaborn: statistical data visualization. J Open Source Softw 2021;6:3021.
40. Hu HH, Kim HW, Nayak KS, Goran MI. Comparison of fat-water MRI and single-voxel MRS in the assessment of hepatic and pancreatic fat fractions in humans. Obesity (Silver Spring) 2010;18:841-7. [Crossref] [PubMed]
41. Bydder M, Yokoo T, Hamilton G, Middleton MS, Chavez AD, Schwimmer JB, Lavine JE, Sirlin CB. Relaxation effects in the quantification of fat using gradient echo imaging. Magn Reson Imaging 2008;26:347-59. [Crossref] [PubMed]
42. Yu H, McKenzie CA, Shimakawa A, Vu AT, Brau AC, Beatty PJ, Pineda AR, Brittain JH, Reeder SB. Multiecho reconstruction for simultaneous water-fat decomposition and T2* estimation. J Magn Reson Imaging 2007;26:1153-61. [Crossref] [PubMed]
43. Chebrolu VV, Hines CD, Yu H, Pineda AR, Shimakawa A, McKenzie CA, Samsonov A, Brittain JH, Reeder SB. Independent estimation of T*2 for water and fat for improved accuracy of fat quantification. Magn Reson Med 2010;63:849-57. [Crossref] [PubMed]
44. Karampinos DC, Ruschke S, Dieckmeyer M, Eggers H, Kooijman H, Rummeny EJ, Bauer JS, Baum T. Modeling of T2* decay in vertebral bone marrow fat quantification. NMR Biomed 2015;28:1535-42. [Crossref] [PubMed]
45. Fischer A, Schick F. Towards detection of inflammation in adipose tissue: Microscopic field simulations to estimate water signal properties. Z Med Phys 2021;31:394-402. [Crossref] [PubMed]
46. Junker D, Wu M, Reik A, Raspe J, Rupp S, Han J, Näbauer SM, Wiechert M, Somasundaram A, Burian E, Waschulzik B, Makowski MR, Hauner H, Holzapfel C, Karampinos DC. 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. Int J Obes (Lond) 2024;48:1332-41. [Crossref] [PubMed]
47. Ganesan G, Warren RV, Leproux A, Compton M, Cutler K, Wittkopp S, Tran G, O'Sullivan T, Malik S, Galassetti PR, Tromberg BJ. Diffuse optical spectroscopic imaging of subcutaneous adipose tissue metabolic changes during weight loss. Int J Obes (Lond) 2016;40:1292-300. [Crossref] [PubMed]
48. Laaksonen DE, Nuutinen J, Lahtinen T, Rissanen A, Niskanen LK. Changes in abdominal subcutaneous fat water content with rapid weight loss and long-term weight maintenance in abdominally obese men and women. Int J Obes Relat Metab Disord 2003;27:677-83. [Crossref] [PubMed]
49. Goossens GH, Bizzarri A, Venteclef N, Essers Y, Cleutjens JP, Konings E, Jocken JW, Cajlakovic M, Ribitsch V, Clément K, Blaak EE. Increased adipose tissue oxygen tension in obese compared with lean men is accompanied by insulin resistance, impaired adipose tissue capillarization, and inflammation. Circulation 2011;124:67-76. [Crossref] [PubMed]
50. Funaki M. Saturated fatty acids and insulin resistance. J Med Invest 2009;56:88-92. [Crossref] [PubMed]
51. White UA, Tchoukalova YD. Sex dimorphism and depot differences in adipose tissue function. Biochim Biophys Acta 2014;1842:377-92. [Crossref] [PubMed]