Lifestyle-based weight loss interventions frequently demonstrate long-term inefficiency and weight regain. Identification of underlying mechanisms and predictors to identify subjects who will benefit from lifestyle-based weight loss strategies is urgently required. We analyzed 143 adults of the randomized Maintain trial (Maintain-Adults) after intended weight loss to identify mechanisms contributing to the regulation of body weight maintenance. Unbiased RNA sequencing of adipose and skeletal muscle biopsies revealed fatty acid metabolism as a key pathway modified by weight loss. Variability of key enzymes of this pathway, estimates of substrate oxidation, and specific serum acylcarnitine (AC) species, representing a systemic snapshot of in vivo substrate flux, predicted body weight maintenance (defined as continuous or dichotomized [< or ≥3% weight regain] variable) 18 months after intended weight loss in the entire cohort. Key results were confirmed in a similar randomized controlled trial in 137 children and adolescents (Maintain-Children), which investigated the same paradigm in a pediatric cohort. These data suggest that adaption of lipid utilization in response to negative energy balance contributes to subsequent weight maintenance. Particularly a functional role for circulating ACs, which have been suggested to reflect intracellular substrate utilization, as mediators between peripheral energy stores and control of long-term energy homeostasis was indicated.

Although weight loss can be achieved by lifestyle interventions, numerous trials failed to demonstrate a long-term improvement of body weight (1,2). No or temporary delay of subsequent weight regain was achieved by a prolonged weight maintenance intervention as demonstrated by us and others (2,3). This might partly explain the lack of effect of multimodal weight loss strategies on cardiovascular events as recently shown in the Look AHEAD (Action for Health in Diabetes) trial (4). Notably, the impact of sustained weight loss on cardiovascular outcomes was demonstrated by a subanalysis of successful weight losers within this trial (5), emphasizing the necessity to identify the underlying mechanism of weight regain. Despite its clinical relevance, the phenomenon of weight regain is currently not well understood.

The role of the neuroendocrine system in the homeostatic regulation of body weight is well described. Although the counter-regulatory neuroendocrine adaption after weight loss may promote weight regain in humans (6), the specific circuits predicting body weight regain in humans are not clear. Inconsistent findings have been reported indicating positive, negative, or no associations between, for example, leptin or ghrelin and weight regain (711). Recently we demonstrated the impact of adrenergic activity on body weight maintenance within a randomized controlled trial (RCT) (3). However, circulating parameters reflect only limited aspects. Given the effects of endocrine circuits on local energy metabolism and substrate utilization, tissue-specific analyses may further improve our understanding of body weight regain. Exemplarily, we recently demonstrated perturbation of the local atrial natriuretic peptide system by weight loss and its potential impact on fat mass (12).

We here analyzed whether weight loss–associated alterations in skeletal muscle and adipose tissue predict body weight maintenance and contribute to long-term body weight regulation. We applied unbiased RNA sequencing to adipose and skeletal muscle tissue biopsies and analyzed metabolism and substrate utilization on a functional level. Finally, we confirmed major findings in a weight maintenance trial in children and adolescents.

Participants

Two RCTs were performed. The study in adults (Maintain-Adults) was performed between 2010 and 2016. A total of 223 overweight or obese subjects were screened for participation. Details of the screening process are provided in the Supplementary Data. Subjects (156 total; 120 female and 36 male; BMI ≥27 kg/m2) were enrolled in the pretrial weight loss phase, and 143 participants (Supplementary Table 1A) who lost at least 8% of their body weight during the weight loss phase were randomized.

A similar trial was performed in children and adolescents (Maintain-Children) between 2009 and 2015. Children and adolescents (n = 137) within an age range from 10 to 17 years were randomized after a guided weight reduction program into a weight maintenance intervention or control group for 12 months (Supplementary Table 1B). Details including the protocols of both trials were already reported (12,13) and are provided in the Supplementary Data.

Study Design

The characteristics of both trials are shown in Supplementary Fig. 1A and B.

Pretrial Weight Loss Phase

All participants of Maintain-Adults underwent a standardized weight reduction program for 12 weeks. Caloric restriction was realized by a meal replacement strategy, nutritional counseling, and physical exercises. A weight loss of at least 8% was intended by this intervention, as this is known to be associated with significant changes in hormonal and metabolic function in humans (11,14,15). The weight reduction program of Maintain-Children was performed in a combination of outpatient and inpatient settings using a specialized residential weight loss center (AHG-Klinik für Kinder und Jugendliche-Beelitz-Heilstätten) using dietary advice, increased physical activity, and psychological support.

Twelve-Month Randomized Weight Maintenance Phase

Subjects of Maintain-Adults (n = 143; weight loss ≥8%) were enrolled in the weight maintenance period and were divided into an intervention and control group. Subjects in the control group were no longer involved in any form of counseling. In contrast, a continuous multimodal counseling focusing on caloric restriction, nutritional counseling, physical exercises, and psychological support was performed for the next 12 months in the intervention group. Further details of the intervention were reported previously (3) and are provided in the Supplementary Data.

Subjects of Maintain-Children who achieved a weight loss of −0.2 BMI SD score (BMI-SDS) were randomized into the 12-month intervention (n = 65) or control (n = 72) group. Participants of the intervention group attended 10 intervention modules throughout a 12-month period. Nutritionists and psychologists addressed the issue of healthy eating and healthy lifestyle. The intervention modules were linked to the BABELUGA lifestyle monitoring map concept (16). In addition, all participants were recommended to increase physical activity into specialized fitness studios (Fidelio), which are local providers of all-around activity and exercise for overweight and obese children and adolescents. Participants assigned to the control group received usual medical care without any further support throughout the whole year.

Follow-up Period

The 12-month randomized intervention period was followed by a free living period of 6 months without any further active intervention in all subjects. This follow-up was performed in both Maintain-Adults and Maintain-Children.

Procedure

A comprehensive phenotyping was performed before (T-3) and after (T0) weight loss, 12 months (T12) after randomization, and after a further 6 months without active intervention (T18). Each subject participated in a 5-day protocol. This included anthropometric, hormonal, and metabolic evaluation, oral glucose tolerance test, and indirect calorimetry (using Vmax ENCORE; CareFusion Germany 234 GmbH, Höchberg, Germany) in a fasting state. In addition, 24-h urine collection, hyperinsulinemic-euglycemic clamps, and biopsies from subcutaneous adipose tissue and skeletal muscle were performed at T-3 and T0. All participants received a dietary recommendation of a balanced energy intake during the 3 days before phenotyping. Phenotyping of Maintain-Children included anthropometric, hormonal, and metabolic evaluation and oral glucose tolerance test. Phenotyping procedures are described in detail in the Supplementary Data.

Laboratory Tests

Laboratory analyses were performed using established methods. Details including the inter- and intra-assay coefficients of variation are provided in the supplement. Acylcarnitine (AC) species were assessed in plasma samples by liquid chromatography—tandem mass spectrometry using a commercially available MS kit (MassChrom Amino Acids and Acylcarnitines; Chromsystems, Munich, Germany) as previously described (17,18).

Tissue samples were analyzed by RNA sequencing using the HiSeq 2000 system (TruSeq SBS Kit-Hs 200 cycles; Illumina, San Diego, CA) (for details see Supplementary Data).

Statistical Analysis

The data reported here are based on per protocol analysis including data of all available participants at the corresponding time point. We calculated the continuous variable regainBMI for all major analyses investigating the predictive value of specific parameters on body weight regain. Therefore, regainBMI was calculated as absolute change of BMI from T0 to T18 (kg/m2). Moreover, regainBMI was also dichotomized for group comparison. As recommended previously (19), a clinically significant weight regain was defined as an increase of BMI between T0 and T18 of at least 3% of the BMI at T0 (weight regain group [WR]). Subjects who gained <3% or lost body weight after T0 were classified as the weight maintenance group (WM).

Weight loss–induced changes (T-3 to T0) of specific parameters were expressed as percentage of baseline values at T-3. ACs were grouped according to their chemical structure in short- (C2- and C4-AC), medium- (C6- to C14-AC), and long-chain ACs (C16- to C22-AC).

Statistical procedures were performed using SPSS version 22.0 (SPSS Inc., Chicago, IL), SAS software, version 9.4 (SAS Institute), and the R software package. Raw values were reported and plotted unless otherwise mentioned.

The time course of BMI and specific AC levels between T-3 and T18 was analyzed using mixed-model, repeated-measures analyses of variance. An independent effect of ACs on regainBMI was analyzed by correlations between ACs and regainBMI after the effects of age, sex, BMI after weight loss (at T0), and randomization state were eliminated. Details of all statistical analyses are described in the Supplementary Data.

Study Approval

The study protocols were approved by the Institutional Review Board of the Charité Medical School and all subjects gave written informed consent. Both trials were registered at ClinicalTrials.gov using the same trial number (NCT00850629).

Body Weight Changes During Weight Loss and Maintenance Period

The primary outcome of the RCT, weight regain after 18 months, was reported previously (3). We observed a weight loss of −4.67 ± 1.47 kg/m2 (−12.6 ± 3.7% of the initial BMI) during the weight loss period. This was accompanied by the already reported (12) improvement of glucose (HOMA of insulin resistance 2.88 ± 2.65 vs. 1.58 ± 1.00; P = 1.2 × 10−15; ISIClamp 0.06 ± 0.03 vs. 0.08 ± 0.04 mg ⋅ kg−1 ⋅ min−1/(mU ⋅ L−1); P = 2.2 × 10−20) and lipid metabolism, including a decreased respiratory coefficient (RQ) (0.82 ± 0.07 vs. 0.78 ± 0.08; P = 3.3 × 10−9), which indicated an improved lipid utilization. Even if a lower weight regain (regainBMI) was observed in the intervention group until month 12, these subjects regained more weight than control subjects between T12 and T18. This results in a similar moderate regainBMI after 18 months within the intervention and control group (+1.42 kg/m2 [0.74; 2.09] vs. +1.96 kg/m2 [1.22; 2.70]; P = NS). Thus BMI at T18 was comparable between the intervention and control group after adjustment for sex, age, and BMI at baseline (32.77 kg/m2 [31.98; 33.55] vs. 33.51 kg/m2 [32.66; 34.36]; P = 0.20). Individual BMI course of the participants is shown in Supplementary Fig. 2.

Tissue-Specific Changes During Weight Loss

Unbiased whole transcriptome profiling was performed in subcutaneous adipose tissue of 75 participants and skeletal muscle of 87 participants before and after the weight loss period. This analysis revealed weight loss induced changes of 8,388 transcripts in adipose tissue and 134 transcripts in skeletal muscle (Fig. 1). Fatty acid desaturase 1 was the most significantly altered gene in adipose tissue, and notably 4 of the top 10 differentially expressed adipose genes were involved in fatty acid metabolism (Table 1).

Figure 1

Weight loss–induced changes of gene expression in skeletal muscle (A) and adipose tissue (B). Differentially expressed genes in skeletal muscle (A) and adipose tissue (B) during the 12-week weight loss period. Log fold change indicates the log fold change of differentially expressed genes (red points), and logCPM indicates the log of counts per million. Black points indicate genes that were not differentially expressed.

Figure 1

Weight loss–induced changes of gene expression in skeletal muscle (A) and adipose tissue (B). Differentially expressed genes in skeletal muscle (A) and adipose tissue (B) during the 12-week weight loss period. Log fold change indicates the log fold change of differentially expressed genes (red points), and logCPM indicates the log of counts per million. Black points indicate genes that were not differentially expressed.

Table 1

Weight loss–induced changes of adipose gene expression (top 10)

Gene nameLog fold changeAdjusted P
Fatty acid desaturase 1 −1.278 4.65 × 10−27 
Stathmin-like 2 −1.263 2.24 × 10−18 
Unknown transcript 1.252 1.08 × 10−18 
Stearoyl-CoA desaturase (delta-9-desaturase) −1.176 3.25 × 10−17 
Hypothetical protein LOC286411 −1.131 6.68 × 10−23 
Fatty acid desaturase 2 −1.099 4.99 × 10−23 
ELOVL family member 6, elongation of long-chain fatty acids (FEN1/Elo2, SUR4/Elo3-like, yeast) −1.089 5.96 × 10−15 
Ribosomal protein L19 pseudogene 9 1.073 9.46 × 10−14 
Hypothetical protein FLJ37543 −1.043 1.62 × 10−17 
UDP-Gal:βGlcNAc, β-1,4-galactosyltransferase, polypeptide 6 −1.001 2.12 × 10−16 
Gene nameLog fold changeAdjusted P
Fatty acid desaturase 1 −1.278 4.65 × 10−27 
Stathmin-like 2 −1.263 2.24 × 10−18 
Unknown transcript 1.252 1.08 × 10−18 
Stearoyl-CoA desaturase (delta-9-desaturase) −1.176 3.25 × 10−17 
Hypothetical protein LOC286411 −1.131 6.68 × 10−23 
Fatty acid desaturase 2 −1.099 4.99 × 10−23 
ELOVL family member 6, elongation of long-chain fatty acids (FEN1/Elo2, SUR4/Elo3-like, yeast) −1.089 5.96 × 10−15 
Ribosomal protein L19 pseudogene 9 1.073 9.46 × 10−14 
Hypothetical protein FLJ37543 −1.043 1.62 × 10−17 
UDP-Gal:βGlcNAc, β-1,4-galactosyltransferase, polypeptide 6 −1.001 2.12 × 10−16 

The top 10 differentially expressed genes in adipose tissue determined using DESeq2.

Hierarchical clustering of gene expression after weight loss of genes highly altered by weight loss revealed a substantial clustering of several transcripts (Fig. 2). This phenomenon was not clearly observable in skeletal muscle. As the observed clustering might reflect weight loss–induced changes of specific pathways, we aimed to identify pathways modified by body weight reduction. Therefore, gene set enrichment analysis using the complete expression data set was performed. In line with the above-mentioned results of the single gene analysis, the KEGG pathway “biosynthesis of unsaturated fatty acids” was among the top five altered pathways in adipose tissue (log fold change = −0.45; P = 2.53 × 10−11) (Table 2). Other pathways found by this analysis are not known to be involved in energy metabolism, and more importantly, no cluster of single genes involved in these pathways could be identified by single gene analysis. Notably, we did not detect significantly altered pathways in skeletal muscle tissue using gene set enrichment analysis.

Figure 2

Hierarchical clustering of Spearman rank correlations between pairs of adipose (A) and muscle (B) transcripts after weight reduction (T0), which were modified by weight loss. Transcripts are ordered according to the clustering performed. The blue (positive) and red (negative) colors indicate the strength and direction of the transcript-transcript correlation. In adipose tissue, only genes with log fold change >0.5 were selected to improve visualization. Hierarchical clustering was applied to group transcripts based on the similarity of their expression profile across this data set.

Figure 2

Hierarchical clustering of Spearman rank correlations between pairs of adipose (A) and muscle (B) transcripts after weight reduction (T0), which were modified by weight loss. Transcripts are ordered according to the clustering performed. The blue (positive) and red (negative) colors indicate the strength and direction of the transcript-transcript correlation. In adipose tissue, only genes with log fold change >0.5 were selected to improve visualization. Hierarchical clustering was applied to group transcripts based on the similarity of their expression profile across this data set.

Table 2

Weight loss–induced changes of KEGG pathways (top five)

KEGG pathwayLog fold changeAdjusted P
RIBOSOME 0.638 4.74 × 10−15 
TERPENOID_BACKBONE_BIOSYNTHESIS −0.484 1.84 × 10−12 
OOCYTE_MEIOSIS −0.323 1.52 × 10−11 
BIOSYNTHESIS_OF_UNSATURATED_FATTY_ACIDS −0.439 2.53 × 10−11 
DNA_REPLICATION 0.322 3.13 × 10−08 
KEGG pathwayLog fold changeAdjusted P
RIBOSOME 0.638 4.74 × 10−15 
TERPENOID_BACKBONE_BIOSYNTHESIS −0.484 1.84 × 10−12 
OOCYTE_MEIOSIS −0.323 1.52 × 10−11 
BIOSYNTHESIS_OF_UNSATURATED_FATTY_ACIDS −0.439 2.53 × 10−11 
DNA_REPLICATION 0.322 3.13 × 10−08 

The top five pathways significantly modified by weight loss in adipose tissue determined using gene set enrichment analysis.

Prediction of Regain After Eighteen Months

Tissue-Specific Gene Expression

Given the accumulation of altered genes involved in fatty acid metabolism, we next focused on this metabolic pathway and investigated whether those genes are related to body weight maintenance after 18 months. Using all genes of the previously identified KEGG pathway “biosynthesis of unsaturated fatty acids,” the predictive information of weight loss–induced changes of gene expression to regainBMI was analyzed. Stepwise linear regression revealed that weight loss–induced changes of HADHA (α subunit of the mitochondrial trifunctional protein), ACAA1 (acetyl-CoA acyltransferase 1), and HSD17B12 (17β-hydroxysteroid dehydrogenase) explained ∼26% of the variability of regainBMI, suggesting that the response of fatty acid metabolism to weight loss may influence subsequent weight regain (Table 3).

Table 3

Adipose tissue transcripts predicting regainBMI

Gene nameStandardized βP value
ΔHADHA 0.721 0.000028 
ΔACAA1 −0.511 0.002 
ΔHSD17B12 0.291 0.016 
Gene nameStandardized βP value
ΔHADHA 0.721 0.000028 
ΔACAA1 −0.511 0.002 
ΔHSD17B12 0.291 0.016 

Weight loss–induced change (Δ) was calculated as transcript number at T-3 minus transcript number at T0.

Functional Estimates of Substrate Utilization

To confirm this hypothesis, we examined the relationship of functional estimates of altered lipid metabolism and body weight regain. In line with a globally improved lipid utilization, we observed a reduction of RQ in the entire cohort during weight loss (0.817 ± 0.069 vs. 0.775 ± 0.078; P < 0.001). However, RQ reduction was less pronounced in subjects regaining body weight (WR) compared with those who maintained their body weight reduction (WM) (−3.43 ± 11.69 vs. −8.07 ± 7.48%; P < 0.05). Consequently, a lower RQ after weight loss was seen in the WM group (0.75 ± 0.06 vs. 0.79 ± 0.09; P < 0.05), suggesting a stronger total lipid utilization in these subjects. These results were confirmed after adjustment for age, sex, BMI after weight loss, and randomization state (F = 5.06; P < 0.05).

The lower RQ in the WM group was accompanied by higher free fatty acid (FFA) levels after weight loss compared with the WR group (0.74 ± 0.03 vs. 0.64 ± 0.03 mmol/L; P < 0.05). These results were again robust to adjustment for age, sex, BMI after weight loss, and randomization state (F = 6.86; P < 0.05). Changes in adipose insulin sensitivity seem not to be involved in elevation of FFAs in WM, as insulin-mediated suppression of FFA was comparable between WM and WR after weight loss (92.98 ± 5.79 vs. 92.14 ± 10.92%; P = NS). Interestingly, serum glycerol levels after weight loss were not different between WM and WR (127.6 ± 55.5 vs. 143.2 ± 12.2 µmol/L; P = NS). As these data indicate a potential role of adipose lipid metabolism, we assessed AC species derived from FFAs for in-depth analysis of lipid utilization. Changes of AC species during weight reduction are shown in Supplementary Table 2. Although most of the ACs declined during weight loss, some long-chain ACs were elevated after weight reduction. Interestingly, weight loss–induced changes of numerous AC species were positively associated with changes of FFAs and the decrease of RQ (Supplementary Table 3). According to those findings, higher circulating FFAs and lower RQ after weight loss were associated with higher levels of several AC species at this time point (Supplementary Table 4). In this context, we also investigated the relationship of FFA levels and myocellular mRNA expression of carnitine palmitoyl transferase 1 (CPT-1), which regulates the rate-limiting step in the transfer of long-chain FFAs from cytosol into the mitochondria for β-oxidation. Thereby a positive association between muscular CPT-1 expression and elevated FFA levels could be revealed after weight loss (r = 0.255; P < 0.05).

Next, we analyzed the predictive effect of ACs on body weight regain. Notably, grouped long-chain ACs after weight loss were inversely associated with regainBMI after the effect of age, sex, BMI after weight loss, and randomization state was regressed out (r = −0.191; P < 0.05). Such a relationship was not found for grouped short- or medium-chain ACs. Analysis of long-chain AC subspecies revealed a significant association between low regainBMI and high postweight loss levels of C18:1-AC (r = −0.253; P < 0.005), C20:1-AC (r = −0.197; P < 0.05), and C20:2-AC (r = −0.203; P < 0.05), adjusted for age, sex, BMI after weight loss, and randomization state. The weight loss–induced increase of C18:1-AC (r = −0.192; P < 0.05) was also associated with a lower regainBMI (Fig. 3A–D). Interestingly, the weight loss–induced changes of those ACs disappeared over time (Supplementary Fig. 3). Similar ratios of short-chain to long-chain ACs as well as C2-AC to (C16 + C18)-ACs were observed in the WM and WR group after weight loss (28.34 ± 1.28 vs. 27.74 ± 0.56 and 28.60 ± 1.31 vs. 27.77 ± 0.58; P = NS).

Figure 3

Long-chain ACs and regainBMI. Correlation of regainBMI with long-chain ACs within adults (AD) and children/adolescents (EG). R and P values were adjusted for age, sex, BMI after weight loss, and randomization state.

Figure 3

Long-chain ACs and regainBMI. Correlation of regainBMI with long-chain ACs within adults (AD) and children/adolescents (EG). R and P values were adjusted for age, sex, BMI after weight loss, and randomization state.

To confirm our findings, we analyzed those AC species in a second RCT evaluating the efficacy of a comparable weight maintenance intervention in children and adolescents (13). Within this trial, BMI-SDS decreased from 2.03 ± 0.31 to 1.66 ± 0.14 during weight loss (P < 0.001). Afterward, a substantial increase was observed until T18 (1.88 ± 0.45; n = 111; P < 0.001 vs. T0). A comparable pattern of ACs was identified that predicted weight regain. C18:1-AC (r = −0.209; P < 0.05), C20:1-AC (r = −0.255; P < 0.01), and C20:2-AC (r = −0.199; P < 0.05) were significantly correlated with changes of BMI-SDS during the maintenance period within this cohort. Adjustment for age, sex, BMI-SDS after weight loss, and randomization state slightly attenuated the association between C18:1-AC (r = −0.161; P = 0.087), C20:2-AC (r = −0.161; P = 0.085), and weight regain, whereas the relation was robust for C20:1-AC (r = −0.222; P < 0.05) (Fig. 3E–G).

This study aimed to reveal novel mechanisms regulating human body weight maintenance. We combined unbiased mRNA deep sequencing of skeletal muscle and adipose tissue with functional metabolic estimates in a randomized human weight maintenance trial. We observed for the first time that the expression of genes involved in lipid biosynthesis (HSD17β12 mediates elongation of FFAs) or utilization (HADHA and ACAA1 encode proteins involved in β-oxidation) (20) within the adipose tissue predict weight maintenance after initial weight loss.

To further investigate these findings of unbiased whole transcriptome analysis, which indicated that the individual response of adipose tissue function to weight loss might contribute to long-term outcome, we analyzed functional estimates of lipid metabolism. Higher FFA levels and lower RQ were observed in individuals maintaining a lower body weight. Elevated fasting FFAs suggest increased FFA supply after weight loss. This might be driven by different mechanisms such as increased lipolysis or reduced re-esterification. Comparable glycerol levels between WM and WR seem to contradict differences in lipolysis. However, glycerol levels are also influenced by changes in other metabolic pathways, such as increased gluconeogenesis in the fasting state (21). Thus, our data do not mandatorily exclude differences in adipose tissue lipolysis. Unfortunately, data of tissue-specific methods like microdialysis, which indicate an increased lipolysis despite unchanged circulating glycerol levels (22), were not measured within the current trial. Nevertheless, lower RQ in WM supports preferred lipid utilization in subjects maintaining lower body weight. We next analyzed ACs as a specific circulating lipid species. Formation of ACs from acyl-CoAs, the activated form of fatty acids, permits the transfer into mitochondria, where generation of ATP within the respiratory chain is enabled by repeated shortening of acyl-CoAs within β-oxidation. Conversion of resulting acyl-CoAs with shorter chain length into corresponding ACs enables their export into circulation. Thus, ACs have been suggested to provide an integrated systemic snapshot of in vivo substrate flux through specific steps of β-oxidation and amino acid degradation and are classically measured to diagnose fatty acid oxidation disorders such as LCHAD deficiency (23). Our data confirmed previous findings indicating a reduction of medium-chain ACs and an elevation of selected long-chain ACs (e.g., C18:1-AC) by weight reduction (24,25). Interestingly, a higher weight loss–induced increase of long-chain ACs (especially 18:1-AC) and higher circulating plasma levels of long-chain AC (C18:1-AC, C20:1-AC, and C20:2-AC) levels after weight loss were observed in subjects who maintained the achieved body weight reduction over 18 months. This indicates that weight loss–induced adaption of lipid metabolism might be important in the regulation of body weight maintenance. Considering the exploratory nature of our finding, we aimed for confirmation of these data in another independent cohort. We therefore measured these AC levels in a cohort of obese children and adolescents, who underwent a similar weight loss–weight maintenance intervention. Those data validated our findings and confirmed that circulating long-chain ACs predict weight maintenance after intended weight loss, even in different age-groups.

Actually, increased levels of short-, medium-, and long-chain ACs were repeatedly reported as estimates of impaired β-oxidation in type 2 diabetes (18,2628). But this may also represent a state of energy deficiency, as caloric restriction is known to be associated with increased lipolysis and lipid oxidation, indicated by increase of FFAs and glycerol as well as an elevation of ACs and ketone bodies (29). Thus, lower RQ after weight loss, seen in WM subjects, and the negative association of RQ with almost all ACs during and after weight loss may suggest a stronger switch to lipid oxidation than impaired β-oxidation. Comparable ratios of short- to long-chain ACs in the WM and WR group also indicate an appropriate utilization of ACs by the tricarboxylic acid cycle.

Vice versa, stimulated lipolysis in the fasting state (30) and higher oral FFA intake can result in an elevation of corresponding ACs (31). Thus, enhanced FFA availability seems to be another potential factor increasing AC levels. This is supported by the positive association of FFAs and several AC species found in our analyses. Therefore, it is tempting to speculate that an increased FFA supply by adipose tissue during weight loss and appropriate adaption of myocellular β-oxidation might induce elevated long-chain AC levels and weight maintenance. The observed association of elevated FFA levels after weight loss and higher myocellular CPT-1 expression may support such a mechanism. Especially FFA 18:1 levels might be relevant in this context, as high FFA 18:1 levels also predict improvement of estimates of obesity during weight loss (32).

Even if the exact nature of FFA-mediated AC elevation was not identified until now, mitochondrial efflux of ACs via OCTN2 and monocarboxylate transporter 9 was already suggested previously (28). Export of ACs seems to depend on increased intramitochondrial AC content, a phenomenon usually observed during higher flux in β-oxidation. It was suggested that such an AC efflux would liberate mitochondrial CoA via transfer of acyl residuals from acyl-CoA to ACs to prevent a decline of β-oxidation due to CoA depletion (28). Therefore, increased circulating AC levels may reflect a mechanism to preserve β-oxidation capacity and to facilitate lipid utilization. As both processes consume energy, they would be supportive for body weight maintenance. Given the known effects of the sympathetic nervous system regarding lipolysis, substrate utilization, and energy expenditure (33,34), the ability described here to shift substrate oxidation during weight loss might be a consequence of the recently demonstrated predictive role of a higher sympathetic drive for weight maintenance (3). Moreover, it is tempting to speculate that this is driven by individual differences in the known weight loss–induced increase of β1 adrenergic lipolysis in subcutaneous adipose tissue (22). Currently we can only speculate whether elevated long-chain AC species just reflect increased lipid utilization or whether they also act as critical interorgan mediators in the regulation of long-term energy homeostasis. Particularly long-chain ACs can diminish insulin sensitivity (28). Although the impaired antilipolytic effect of insulin would result in higher adipose tissue breakdown and could thereby increase lipid utilization, our results do not suggest differences in adipose insulin sensitivity between WM and WR. On the other hand, limited data indicate a dose-dependent induction of myocellular FFA oxidation by several medium-chain ACs themselves (35). Beside these peripheral effects, central administration of oleic acid results in marked inhibition of food intake (36). Taken together, a specific function of ACs is likely to exist, and future studies are required to better understand the mechanisms beyond the predictive information of specific ACs.

The interpretation of our data is limited by some factors. Behavioral, social, as well as environmental factors may have influenced body weight regain (3739). Even if we performed group sessions under a standardized setting, we cannot exclude that these factors may have influenced our results. Participants of the intervention group were advised to consume a balanced diet, advocating the preferential intake of specific foods (like high intake of vegetables, cereals, and fat-reduced foods and lean meat consumption) to achieve the intended macronutrient distribution. As this was an outpatient study, we cannot rule out that other food than reported was consumed. Moreover, weight reduction based on a low-fat diet is known to decrease FFAs and ketone bodies due to impaired lipid utilization (40). This may have influenced our findings. However, the dietary component of our weight loss intervention was primarily based on a standardized formula diet to avoid huge differences in the fat content during the weight loss period. Moreover, tissue-specific analysis included only transcriptional data. As activity of several involved proteins is regulated by phosphorylation and posttranscriptional modifications, our data could only be indicative for involved pathways. However, concomitant changes of metabolites and confirmation of metabolite findings in a second independent cohort support the functional relevance of lipid metabolism in predicting weight maintenance.

The study has some strengths. We analyzed individual adipose and skeletal muscle samples at two different time points and related the findings of our deep sequencing analysis to detailed clinical phenotypes in a significant number of participants. To the best of our knowledge, this is one of the largest RCTs on body weight maintenance with such detailed biopsy and clinical phenotyping. The confirmation of our key findings in a second, independent randomized trial is an exceptional strength of our study and makes false-positive results rather unlikely. Particularly the combination of mRNA deep sequencing results with functional findings and metabolite data supports that the regulation of lipid metabolism contributes significantly to body weight maintenance after 18 months.

In summary, our data demonstrated that a weight loss–induced metabolic switch to increase FFA supply as well as intact β-oxidation might be important elements in the regulation of body weight maintenance. Several long-chain ACs, which mirror the substrate flux through specific steps of β-oxidation, were characterized as predictors of long-term success of weight loss interventions in two independent RCTs. Future studies are warranted to clarify the mechanism underlying the predictive role of ACs.

Clinical trial reg. no. NCT00850629, clinicaltrials.gov

Acknowledgments. The authors thank K. Simon, B. Horchler, N. Huckauf, and C. Kalischke (all Charité-Universitätsmedizin Berlin) for excellent technical assistance as well as A. Reisshauer (Charité-Universitätsmedizin Berlin) for the support regarding the physical activity intervention.

Funding. The authors thank Nestlé HealthCare Nutrition GmbH, Frankfurt am Main, Germany, for the opportunity to purchase the Optifast 2 diet at a reduced price. This research was supported by the Deutsche Forschungsgemeinschaft (DFG KFO 218/1) and the German Ministry for Education and Research (BMBF) by support of the Berlin Institute of Health (BIH) and the German Centre for Cardiovascular Research (DZHK, BER5.1).

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the manuscript.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.M., L.L., and M.B. analyzed and researched data, contributed to interpretation of the results, and wrote the manuscript. S.W. contributed to interpretation of the results. V.L., P.K., N.H., W.C., and H.K. researched data and contributed to interpretation of the results. A.E. and P.R. analyzed data and contributed to interpretation of the results. J.S. researched data, contributed to interpretation of the results, and wrote the manuscript. All authors read and approved the submitted version of the manuscript. K.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Coughlin
JW
,
Brantley
PJ
,
Champagne
CM
, et al.;
Weight Loss Maintenance Collaborative Research Group
.
The impact of continued intervention on weight: five-year results from the weight loss maintenance trial
.
Obesity (Silver Spring)
2016
;
24
:
1046
1053
[PubMed]
2.
Pekkarinen
T
,
Kaukua
J
,
Mustajoki
P
.
Long-term weight maintenance after a 17-week weight loss intervention with or without a one-year maintenance program: a randomized controlled trial
.
J Obes
2015
;
2015
:
651460
[PubMed]
3.
Mai
K
,
Brachs
M
,
Leupelt
V
, et al
.
Effects of a combined dietary, exercise and behavioral intervention and sympathetic system on body weight maintenance after intended weight loss: results of a randomized controlled trial
.
Metabolism
2018
;
83
:
60
67
[PubMed]
4.
Wing
RRBP
,
Bolin
P
,
Brancati
FL
, et al.;
Look AHEAD Research Group
.
Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes
.
N Engl J Med
2013
;
369
:
145
154
[PubMed]
5.
Gregg
EW
,
Jakicic
JM
,
Blackburn
G
, et al.;
Look AHEAD Research Group
.
Association of the magnitude of weight loss and changes in physical fitness with long-term cardiovascular disease outcomes in overweight or obese people with type 2 diabetes: a post-hoc analysis of the Look AHEAD randomised clinical trial
.
Lancet Diabetes Endocrinol
2016
;
4
:
913
921
[PubMed]
6.
Sumithran
P
,
Prendergast
LA
,
Delbridge
E
, et al
.
Long-term persistence of hormonal adaptations to weight loss
.
N Engl J Med
2011
;
365
:
1597
1604
[PubMed]
7.
Mavri
A
,
Stegnar
M
,
Sabovic
M
.
Do baseline serum leptin levels predict weight regain after dieting in obese women
?
Diabetes Obes Metab
2001
;
3
:
293
296
[PubMed]
8.
Crujeiras
AB
,
Goyenechea
E
,
Abete
I
, et al
.
Weight regain after a diet-induced loss is predicted by higher baseline leptin and lower ghrelin plasma levels
.
J Clin Endocrinol Metab
2010
;
95
:
5037
5044
[PubMed]
9.
Wing
RR
,
Sinha
MK
,
Considine
RV
,
Lang
W
,
Caro
JF
.
Relationship between weight loss maintenance and changes in serum leptin levels
.
Horm Metab Res
1996
;
28
:
698
703
[PubMed]
10.
Napolitano
A
,
Miller
SR
,
Murgatroyd
PR
, et al
.
Prediction of weight loss and regain following dietary, lifestyle, and pharmacologic intervention
.
Clin Pharmacol Ther
2012
;
91
:
1027
1034
[PubMed]
11.
Wang
P
,
Holst
C
,
Andersen
MR
, et al
.
Blood profile of proteins and steroid hormones predicts weight change after weight loss with interactions of dietary protein level and glycemic index
.
PLoS One
2011
;
6
:
e16773
[PubMed]
12.
Brachs
M
,
Wiegand
S
,
Leupelt
V
, et al
.
ANP system activity predicts variability of fat mass reduction and insulin sensitivity during weight loss
.
Metabolism
2016
;
65
:
935
943
[PubMed]
13.
Bau
AM
,
Ernert
A
,
Krude
H
,
Wiegand
S
.
Hormonal regulatory mechanisms in obese children and adolescents after previous weight reduction with a lifestyle intervention: maintain - paediatric part - a RCT from 2009-15
.
BMC Obes
2016
;
3
:
29
[PubMed]
14.
Wang
P
,
Holst
C
,
Astrup
A
, et al.;
Diogenes Consortium
.
Blood profiling of proteins and steroids during weight maintenance with manipulation of dietary protein level and glycaemic index
.
Br J Nutr
2012
;
107
:
106
119
[PubMed]
15.
Larsen
TM
,
Dalskov
SM
,
van Baak
M
, et al.;
Diet, Obesity, and Genes (Diogenes) Project
.
Diets with high or low protein content and glycemic index for weight-loss maintenance
.
N Engl J Med
2010
;
363
:
2102
2113
[PubMed]
16.
Ernst
M
,
Wiegand
S
.
Adipositas bei Kindern und Jugendlichen einmal anders. Die BABELUGA-Methode: Prävention
.
Bern, Huber
,
Therapie, Selbstmanagement
,
2010
, p.
1
310
17.
Metz
TF
,
Mechtler
TP
,
Merk
M
, et al
.
Evaluation of a novel, commercially available mass spectrometry kit for newborn screening including succinylacetone without hydrazine
.
Clin Chim Acta
2012
;
413
:
1259
1264
[PubMed]
18.
Mihalik
SJ
,
Goodpaster
BH
,
Kelley
DE
, et al
.
Increased levels of plasma acylcarnitines in obesity and type 2 diabetes and identification of a marker of glucolipotoxicity
.
Obesity (Silver Spring)
2010
;
18
:
1695
1700
[PubMed]
19.
Stevens
J
,
Truesdale
KP
,
McClain
JE
,
Cai
J
.
The definition of weight maintenance
.
Int J Obes
2006
;
30
:
391
399
[PubMed]
20.
Saloniemi
T
,
Jokela
H
,
Strauss
L
,
Pakarinen
P
,
Poutanen
M
.
The diversity of sex steroid action: novel functions of hydroxysteroid (17β) dehydrogenases as revealed by genetically modified mouse models
.
J Endocrinol
2012
;
212
:
27
40
[PubMed]
21.
Neeland
IJ
,
Hughes
C
,
Ayers
CR
,
Malloy
CR
,
Jin
ES
.
Effects of visceral adiposity on glycerol pathways in gluconeogenesis
.
Metabolism
2017
;
67
:
80
89
[PubMed]
22.
Barbe
P
,
Stich
V
,
Galitzky
J
, et al
.
In vivo increase in beta-adrenergic lipolytic response in subcutaneous adipose tissue of obese subjects submitted to a hypocaloric diet
.
J Clin Endocrinol Metab
1997
;
82
:
63
69
[PubMed]
23.
Das
AM
,
Illsinger
S
,
Lücke
T
, et al
.
Isolated mitochondrial long-chain ketoacyl-CoA thiolase deficiency resulting from mutations in the HADHB gene
.
Clin Chem
2006
;
52
:
530
534
[PubMed]
24.
Tan
HC
,
Khoo
CM
,
Tan
MZ
, et al
.
The effects of sleeve gastrectomy and gastric bypass on branched-chain amino acid metabolism 1 year after bariatric surgery
.
Obes Surg
2016
;
26
:
1830
1835
[PubMed]
25.
Schooneman
MG
,
Napolitano
A
,
Houten
SM
, et al
.
Assessment of plasma acylcarnitines before and after weight loss in obese subjects
.
Arch Biochem Biophys
2016
;
606
:
73
80
[PubMed]
26.
Adams
SH
,
Hoppel
CL
,
Lok
KH
, et al
.
Plasma acylcarnitine profiles suggest incomplete long-chain fatty acid beta-oxidation and altered tricarboxylic acid cycle activity in type 2 diabetic African-American women
.
J Nutr
2009
;
139
:
1073
1081
[PubMed]
27.
Mihalik
SJ
,
Michaliszyn
SF
,
de las Heras
J
, et al
.
Metabolomic profiling of fatty acid and amino acid metabolism in youth with obesity and type 2 diabetes: evidence for enhanced mitochondrial oxidation
.
Diabetes Care
2012
;
35
:
605
611
[PubMed]
28.
Schooneman
MG
,
Vaz
FM
,
Houten
SM
,
Soeters
MR
.
Acylcarnitines: reflecting or inflicting insulin resistance
?
Diabetes
2013
;
62
:
1
8
[PubMed]
29.
Collet
TH
,
Sonoyama
T
,
Henning
E
, et al
.
A metabolomic signature of acute caloric restriction
.
J Clin Endocrinol Metab
2017
;
102
:
4486
4495
[PubMed]
30.
Costa
CC
,
de Almeida
IT
,
Jakobs
C
,
Poll-The
BT
,
Duran
M
.
Dynamic changes of plasma acylcarnitine levels induced by fasting and sunflower oil challenge test in children
.
Pediatr Res
1999
;
46
:
440
444
[PubMed]
31.
Kien
CL
,
Everingham
KI
,
D Stevens
R
,
Fukagawa
NK
,
Muoio
DM
.
Short-term effects of dietary fatty acids on muscle lipid composition and serum acylcarnitine profile in human subjects
.
Obesity (Silver Spring)
2011
;
19
:
305
311
[PubMed]
32.
Smilowitz
JT
,
Wiest
MM
,
Watkins
SM
, et al
.
Lipid metabolism predicts changes in body composition during energy restriction in overweight humans
.
J Nutr
2009
;
139
:
222
229
[PubMed]
33.
Tentolouris
N
,
Liatis
S
,
Katsilambros
N
.
Sympathetic system activity in obesity and metabolic syndrome
.
Ann N Y Acad Sci
2006
;
1083
:
129
152
[PubMed]
34.
Spraul
M
,
Ravussin
E
,
Fontvieille
AM
,
Rising
R
,
Larson
DE
,
Anderson
EA
.
Reduced sympathetic nervous activity. A potential mechanism predisposing to body weight gain
.
J Clin Invest
1993
;
92
:
1730
1735
[PubMed]
35.
Lehmann
R
,
Zhao
X
,
Weigert
C
, et al
.
Medium chain acylcarnitines dominate the metabolite pattern in humans under moderate intensity exercise and support lipid oxidation
.
PLoS One
2010
;
5
:
e11519
[PubMed]
36.
Obici
S
,
Feng
Z
,
Morgan
K
,
Stein
D
,
Karkanias
G
,
Rossetti
L
.
Central administration of oleic acid inhibits glucose production and food intake
.
Diabetes
2002
;
51
:
271
275
[PubMed]
37.
Svetkey
LP
,
Stevens
VJ
,
Brantley
PJ
, et al.;
Weight Loss Maintenance Collaborative Research Group
.
Comparison of strategies for sustaining weight loss: the weight loss maintenance randomized controlled trial
.
JAMA
2008
;
299
:
1139
1148
[PubMed]
38.
Del Corral
P
,
Bryan
DR
,
Garvey
WT
,
Gower
BA
,
Hunter
GR
.
Dietary adherence during weight loss predicts weight regain
.
Obesity (Silver Spring)
2011
;
19
:
1177
1181
[PubMed]
39.
Christensen
BJ
,
Iepsen
EW
,
Lundgren
J
, et al
.
Instrumentalization of eating improves weight loss maintenance in obesity
.
Obes Facts
2017
;
10
:
633
647
[PubMed]
40.
Lien
LF
,
Haqq
AM
,
Arlotto
M
, et al
.
The STEDMAN project: biophysical, biochemical and metabolic effects of a behavioral weight loss intervention during weight loss, maintenance, and regain
.
OMICS
2009
;
13
:
21
35
[PubMed]
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Supplementary data