Introduction

Obesity relapse is commonly observed after diet-induced weight loss being the major drawback of nonsurgical treatment.1 Because weight cycling itself is believed to have adverse health consequences,2, 3, 4, 5 some health professionals even advice against short-term dieting, although more recent scientific evidence remains controversial (see Field et al.6 also for review). Chronic dieting may be harmful for biological mechanisms of weight control because repeated cycles of weight loss and regain appear to enhance subsequent weight gain and may therefore predispose to obesity.7, 8, 9, 10 In addition, there could be detrimental effects of weight cycling on cardiometabolic risk independent of obesity.11, 12

Weight cycling-associated changes in body composition may contribute to both a predisposition for weight gain and an increased cardiometabolic risk. Byrne et al.13 reported that after weight loss, lean body mass of the trunk was not regained in proportion to the regain in limb lean mass, thus implying a loss in high metabolically active organ tissue. A lower proportion of organ mass would result in a lower specific metabolic rate of total lean mass. However, previous results of our group have shown that the loss in organ and tissue masses did not fully explain the observed decrease in resting energy expenditure (REE) after diet-induced weight loss.14, 15 Adaptive thermogenesis after weight loss has also been found by other studies that did not consider changes in the composition of lean mass.16, 17, 18, 19, 20, 21, 22 A persistence of adaptive thermogenesis is considered a risk factor for weight regain. There was however no significant change in REE after weight regain following diet-induced weight loss in obese women.23, 24 So far, the contribution of changes in the composition of lean mass (in particular the proportion of high metabolically active organ mass) to weight cycling-associated changes in REE remains unknown.

In postmenopausal women25 and elderly people from the Health Ageing and Body Composition Study,26 proportionally more lean mass was lost during the weight-loss period than was regained during the weight-regain period. Although this could be a side effect of dieting that is confined to elderly people, a disproportional regain in fat mass (FM) has also been observed in young people recovering from severe starvation (for a review, see Dulloo et al.27). In addition, cross-sectional data in normal and overweight women,28, 29 re-feeding studies in anorectic patients30, 31, 32, 33 and results from the Minnesota semi-starvation study16 suggest that weight cycling may also adversely affect body fat distribution, leading to an accumulation of trunk or visceral fat that would explain an increased metabolic risk independent of total adiposity. However, longitudinal studies in overweight patients do not support this association.34, 35, 36, 37

Claims that weight cycling adversely affects REE and body composition in overweight subjects can be ruled out using state-of-the-art whole-body magnetic resonance imaging (MRI) to assess body fat distribution (visceral and subcutaneous fat) as well as the composition of lean mass (skeletal muscle and organ mass). The present study sets out to compare the composition of weight loss and weight regain in overweight and obese subjects using MRI. The impact of weight regain-associated changes in body composition on REE is analyzed.

Materials and methods

Study protocol and subjects

The study group was a subsample of 103 overweight and obese participants who were recruited for a dietary-weight loss intervention trial. Inclusion criteria and recruitment of participants for the dietary intervention trial have been described previously.14 Briefly, healthy overweight and obese Caucasian men and women aged 22–45 years with a body mass index >29 kg m−2 were recruited from the general public. All subjects had a normal physical examination and electrocardiogram recording, no use of lipid-lowering, hypoglycemic or antihypertensive medication, no history of cardiovascular or metabolic disease and a normal thyroid function. Women were premenopausal, nonpregnant and nonlactating. All measurements were performed at baseline (T0), after weight loss intervention (T1) and 6 months thereafter (T2, follow-up). The subsample (n=47) was selected on the basis of their weight regain 6 months after weight loss. Only participants who either regained ⩾30% of their weight loss (weight regainers) or participants who were weight stable (weight change±20% of weight loss) were included in the database. Data of 56 subjects were disregarded for this analysis because of unsuccessful weight loss (n=11), nonattendance at follow-up (n=14), further weight loss at follow-up (n=12), weight regain between 20 and 30% of weight loss (n=10) and missing data (n=9). The selected subsample of 47 subjects did not significantly differ from the remaining study population with respect to age and initial body mass index or weight loss (all P>0.05).

The weight loss program consisted of weekly individual counseling by a registered dietician and a 13±3-week low-calorie, nutritionally balanced self-selected diet containing 800–1000 kcal day−1 whereof 434 kcal day−1 were supplied as a very-low-energy diet (BCM-Diät, PreCon, Darmstadt, Germany; ingestion of 2 shakes per day provided all nutrients according to the Recommended Dietary Allowance (RDA): 37.3 g protein, 38.8 g carbohydrate and 13.5 g fat). The study was approved by the medical ethics committee of the Christian-Albrechts-University Kiel. All subjects provided their fully informed and written consent before participation.

Anthropometric measurements and body composition analysis

Body weight (±100 g) was measured on an electronic Tanita scale coupled to the BOD-POD system (Life Measurement Instruments, Concord, CA, USA). Height was measured on a stadiometer (seca, Hamburg, Germany) to the nearest 0.5 cm.

Magnetic resonance imaging

Volumes of adipose tissue, skeletal muscle and internal organs were assessed by MRI (Magnetom Avanto 1.5-T, Siemens Medical Systems, Erlangen, Germany). Subjects were examined in a supine position with their arms extended above their heads. Transversal images were obtained from wrist to ankle by using a contiguous axial T1 weighted gradient-echo sequence (TR 157 ms, TE 4 ms, flip angle 70 °, voxel size 3.9 × 2 × 8 mm3). The protocol for the brain comprised contiguous 4 mm slices with 1 mm interslice gaps (TR 313 ms, TE 14 ms). For the rest of the body, images were obtained with 8 mm slice thickness and 2 mm interslice gaps. Image acquisition for volumetric assessment of the thoracic and abdominal region was obtained in breath-hold, and heart mass was assessed using a breath-navigated and pulse-triggered T2-weighted HASTE sequence (imaging parameters: TR 700 ms, TE 24 ms, flip angle 160 °, voxel size 2.2 × 1.3 × 8 mm3, turbo factor 106). The volume of visceral adipose tissue (VAT) was acquired from top of the liver or the base of the lungs (T10) to the femur heads. All images were segmented manually (Slice-O-Matic, Tomovision 4.3 Software, Montreal, Canada). Briefly, the software employed knowledge-based image processing to label pixels as fat and nonfat components using a threshold for adipose tissue on the basis of the gray-level histograms of the images. Each organ/tissue was analyzed by the same observer who was blinded to the time point and subject identity (intraobserver coefficients of variation based on comparison of repeated segmentations were 1.8% for brain, 0.07% for liver, 1.7% for heart, 1.0% for kidneys, 1% for VAT, 0.9% for subcutaneous adipose tissue (SAT) and 1.7% for skeletal muscle). Total organ/tissue volume was determined from the sum of all areas (cm2) multiplied by the slice thickness. Volume data were transformed into organ mass using the following densities: 1.036 g cm−3 for brain, 1.06 g cm−3 for heart and liver, 1.05 g cm−3 for kidneys28 and 1.04 g cm−3 for skeletal muscle.38

Densitometry

Air-displacement plethysmography was performed using the BOD-POD device (Life Measurement Instruments; software version 1.69). Subjects were measured in tight-fitting underwear and a swimming cap. Two repeated measurements of body volume were performed and averaged. Measured thoracic lung volume was subtracted from body volume. In order to reduce measurement variability caused by repeated lung volume measurements, thoracic lung volume was only measured at T0 and this value was subsequently used for body composition analysis at T1 and T2, respectively. BOD-POD software was used to calculate body density as body weight divided by body volume and percent fat mass (%FM) using Siri’s equation.39 Fat-free mass (kg) was calculated accordingly: weight (kg) −FM (kg). The coefficient of variation for repeated measurements of %FM was 2.4%.

Dual-energy X-ray absorptiometry was performed to measure bone mineral content using a Hologic Discovery A densitometer and the whole-body-software 12.6.1:3 (Hologic, Inc., Bedford, MA, USA).

Resting energy expenditure

Indirect calorimetry was performed in the morning between 0730 and 0900 h after an overnight fast on a metabolic ward at constant temperature and humidity (ventilated hood system: Vmax Spectra 29n; SensorMedics BV, Bilthoven, The Netherlands; software Vmax, version 12-1 A). The minimum duration of measurement was 45 min and the first 10 min were discarded. Flow calibration was performed by a 3L syringe, and gas analyzers were calibrated before and every 5 min during the run. Data were collected every 20 s and acquired VO2 and VCO2 were converted to REE (kcal 24 h) using the abbreviated equation of Weir. The coefficients of variation for repeated REE measurements were 5.2%.

REE was normalized for detailed body composition by subtracting REE calculated from organ and tissue masses (REEc) from measured REE (REEmeasured−calculated).14 Calculation of REE was based on the sum of eight body compartments (brain, heart, liver, kidneys, skeletal muscle mass, bone mass, adipose tissue and residual mass) × the corresponding tissue respiration rate, using the specific tissue metabolic rates as reported by Elia.40 For bone mass, a specific metabolic rate of 9.63 kJ kg × day−1 was assumed.41 Residual mass was calculated as body mass minus the sum of brain, heart, liver, kidneys, skeletal muscle mass, bone mass and adipose tissue. The metabolic activity of residual mass was assumed to be 30 kJ kg × day−1.14

REEc (kJ day−1)=(1008 × brain mass)+(840 × liver mass)+(1848 × heart mass)+(1848 × kidney mass)+(55 × skeletal muscle mass)+(9.63 × bone mass)+(19 × adipose tissue)+(30 × residual mass).

Adipose tissue was calculated from FM assuming a fat content of 90%. Bone mass was calculated by multiplication of bone mineral content × 1.85 based on Reference man data.38

Clinical and metabolic variables

Fasting serum concentrations of thyroid-stimulating hormone (TSH), free triiodothyronine (fT3) and free thyroxine (fT4) were measured by radioimmunoassay (DiaSorin, Dietzenbach, Germany); the intra- and inter-assay coefficients of variation were 2.5 and 5.7% (TSH), 4.6 and 6.5% (fT3) and 2.4 and 6.8% (fT4). The sensitivity limits were 0.02 mIU ml−1 (TSH), 0.35 pg ml−1 (fT3) and 1 pg ml−1 (fT4), respectively. Plasma insulin was measured by radioimmunoassay showing no crossreactivity with C-peptide and only 14% with proinsulin (Adaltis, Rome, Italy). Plasma glucose was assayed using a hexokinase enzymatic method. The homeostasis model assessment was used to calculate insulin resistance (IR) as HOMA-IR=fasting insulin (μU ml−1) × fasting glucose (mmol l−1)/22.5.42

Statistical methods

Data are expressed as means±s.d. A paired t-test was used to compare within-group differences in the changes of outcome measures with weight loss (ΔT1−T0) and weight regain (ΔT2−T1). Comparisons between genders or between weight regainers and weight-stable subjects were analyzed by Mann–Whitney U-test. Relationships between variables were sought by correlation analysis (Pearson’s and Spearman’s r). Two-tailed P-values of <0.05 were considered to indicate statistical significance. Data analyses were performed with SPSS statistical software (SPSS 15.0, Inc., Chicago, IL, USA).

Results

Mean weight loss was higher in men when compared with women (P<0.01) and did not differ between weight-stable and weight-regaining men (−14.2±3.1 vs −13.3±3.5 kg, P>0.05), whereas weight-stable women had a higher weight loss (−12.5±2.0 vs −7.8±3.7 kg, P<0.001). Because of selection criteria, weight regain was significantly lower in the weight-stable group when compared with weight regainers (1.2±1.9 vs 5.6±1.5 kg, P<0.001, in weight-stable and weight-regaining men and 0.3±1.4 vs 6.5±2.2 kg, P<0.001, in weight-stable and weight-regaining women). Weight regain ranged from −2.2 to 2.6 kg in the weight-stable groups and from 4.0 to 13.0 kg in weight regainers. No differences in baseline parameters of body composition, REE, thyroid hormones, basal glucose levels and insulin sensitivity were observed between weight-stable and weight-regaining participants, except for a higher heart mass in weight-regaining men (Table 1). Because of gender differences in body composition, differences between weight loss and weight regain were evaluated separately for men and women.

Table 1 Baseline (T0) body composition and metabolic characteristics of the study participants stratified by weight change after weight loss (ΔT2−T1).
Full size table

Comparing the composition of weight loss and weight regain

Changes in body composition with weight loss and weight regain are given in Table 2. Weight regain was incomplete, accounting for 83 and 42% of weight loss in women and men. In women, regain in %FM, visceral and SAT trunk was proportional to weight regain (83–85%), whereas regain in adipose tissue of the extremities was higher (102% for arms and 102% for legs, Figure 1a). In men, %FM and SAT trunk were also regained in proportion to weight loss (50%, 51%), whereas regain in visceral and extremity adipose tissue was lower (24%, VAT; not significant, SAT arms; and 20%, SAT legs; Figure 1b).

Table 2 Absolute changes in body composition and parameters of glucose metabolism with weight loss and follow-up in weight regainers stratified by gender
Full size table
Figure 1
figure 1

Regain as a percentage of loss for body weight, total fat mass and different adipose tissue compartments, that is, (mean regain in VAT × 100)/mean loss in female (a) and male (b) weight regainers. (c) Regain as a percentage of loss for body weight and regional skeletal muscle (SM) in both genders combined is shown.

Full size image

In women, loss in skeletal muscle mass occurred predominately at the trunk, whereas during weight regain an increase in muscle mass was observed mainly at the extremities. The increase in muscle mass at the extremities with weight regain was also found for men, whereas skeletal muscle at the trunk remained unchanged during weight loss and weight regain. Thus, in both genders, reconstitution of skeletal muscle at the trunk seems to lag behind the extremities (Figure 1c). Organ mass did not change with weight loss and regain in both genders, except for a significant increase during weight regain in women (this was explained by an increase in liver mass with weight regain 184±188 g, P<0.01).

Consistent with the absence of an adverse effect of weight regain on body fat distribution, basal insulin levels and homeostatic model assessment (HOMA) index improved with weight loss and remained unchanged with weight regain in both genders (Table 2). However, in the total study population, weight change at follow-up was associated with changes in HOMA index (r=0.30; P<0.05).

Body composition during weight loss and follow-up in weight-stable subjects

For weight-stable subjects, changes in body fat distribution, the composition of lean mass and parameters of glucose metabolism during weight loss and follow-up are shown in Table 3. Because of a higher weight loss in weight-stable when compared with weight-regaining women (see Table 2, P<0.001), losses in organ mass (P<0.05), visceral fat (P<0.05) and SAT trunk (P<0.01) were higher. In contrast, in men, no differences in weight loss or loss of fat and lean compartments were observed between weight-stable and weight-regaining participants. Despite weight stability, there was a significant increase in organ mass in both genders and SAT trunk in women (Table 3).

Table 3 Absolute changes in body composition and parameters of glucose metabolism with weight loss and follow-up in weight-stable-subjects stratified by gender
Full size table

Similar to weight regainers, HOMA index improved with weight loss and remained unchanged after the weight-stable phase.

Impact of weight loss and weight regain on REE

In weight-stable participants, REE adjusted for body composition was unaffected during weight loss and follow-up (Table 4). In contrast, in the group of weight regainers, REE estimated from organ and tissue masses × their specific organ metabolic rates (REEcalculated) revealed that a 100% recovery of REE would have been expected from the regain in organ and tissue mass. However, REE adjusted for changes in organ and tissue masses (REEmeasured–REEcalculated) was reduced with weight loss (P<0.01) and remained reduced with weight regain. However, between-group differences in metabolic adaptation with weight loss were not significant (P=0.132 for ΔT1–T0 REEmeasured–REEcalculated between weight losers and regainers).

Table 4 Changes in REE and thyroid hormones with weight loss and weight regain in weight-stable and weight-regaining subjects
Full size table

In weight regainers, fT3 levels significantly decreased during weight loss and failed to rise again with weight regain (Table 4). No correlation was found between changes in REE adjusted for body composition and the change in fT3 during weight loss and weight regain (data not shown). However, in weight regainers, weight regain was associated with a further T1–T2 decrease in fT3 levels (r=−0.40; P<0.05).

Discussion

Adverse effects of weight cycling are of great concern in conventional obesity therapy that is frequently associated with a failure to sustain weight loss.1 To the best of our knowledge, this is the first study that investigates the effect of weight loss and weight regain on body fat distribution, the composition of lean mass and its effect on REE in younger overweight subjects using whole-body MRI. The results show that weight regain does not adversely affect body fat distribution or the composition of lean mass. However, in contrast to weight-stable subjects, weight regainers were characterized by a decrease in REE with weight loss that is likely to add to the lower weight loss (in women) and the propensity for weight regain.

Effect of weight loss and regain on adiposity and body fat distribution

In the present study, weight regain did not adversely affect body fat distribution in both men and women (Table 2). There may however be sex differences in the reconstitution of adipose tissue depots. When compared with weight regain, regain in adipose tissue of the extremities was disproportionally higher in women and lower in men. However, the impairment of fat storage in gluteofemoral depots in men was not compensated by an increased reconstitution in visceral fat because regain in VAT was only 24% at a 42% weight regain. SAT trunk was regained in proportion to weight regain in both genders. In contrast to our results, a preferential trunk or visceral fat regain has been shown in most30, 31, 32, 33 but not all studies on weight recovery in patients with anorexia nervosa.43 However, comparison of weight loss and regain in obese patients did not confirm an effect of weight cycling on visceral fat accumulation36 or even found an increased gynoid pattern of fat redistribution after weight regain.34, 35, 37 Interestingly, the latter finding is supported by our results in females showing a disproportionately lower loss in leg fat (−10%) compared with visceral or subcutaneous trunk fat (both −14%) after dieting, coupled with a higher regain in leg fat (102% of loss for leg fat, 85% and 83% of loss for VAT and SAT trunk, Table 2). In an animal model, adipocyte hyperplasia occurred during weight regain.44 Similarly, overfeeding in normal-weight adults was associated with lower-body adipocyte hyperplasia, whereas abdominal adipocytes showed hypertrophy but no increase in cell number.45, 46 Because lower-body adipocyte cellularity remained higher after subsequent weight loss,46 and our results show an increased regain of gluteofemoral adipose tissue, it is tempting to speculate that weight cycling leads to a more gynoid body fat distribution with an increased capacity for fat storage.

Discrepant results of previous studies regarding the change in fat distribution may be because of age differences, differences in endocrine parameters (for example, sex steroids or cortisol levels) or a varying length in follow-up (because changes in fat distribution could be transient and therefore be missed in the long-term follow-up). Not only does the redistribution of adipose tissue differ between weight-reduced normal-weight and overweight subjects, weight regain in underweight subjects is also commonly associated with a preferential reconstitution of FM termed the catch-up fat phenomenon (Keys et al.16 reviewed in Dulloo et al.27). Because there was no significant change in fat-free mass in our study (Table 2), we could not compare the recovery of FM and fat-free mass. This is likely explained by a higher initial FM in overweight subjects that allows a higher reliance on fat mobilization during caloric restriction so that the composition of weight loss and weight regain is equal (both mainly consisting of fat). However, two recent studies in elderly people have shown that proportionally more lean mass was lost during the weight loss period than was regained during the weight regain period.25, 26

In summary, current evidence suggests that changes in body fat and body fat distribution with weight regain depend on age and weight status with a preferential regain of fat only occurring with re-feeding in the elderly and in weight-reduced normal-weight subjects.

Effect of weight loss and regain on the composition of lean mass

Using dual-energy X-ray absorptiometry, Byrne et al.13 found that after weight loss, lean body mass of the trunk was not regained in proportion to the regain in limb lean mass. The authors therefore assumed a loss in high metabolically active organ tissue. Consistent with the observation by Byrne et al.,13 we found that the reconstitution of skeletal muscle at the trunk seems to lag behind the extremities in both genders (Table 2). Autopsy data from people who died of starvation revealed a reduction in organ mass that was mainly because of a loss in liver and heart mass (reviewed in Keys et al.16). In contrast, organ mass did not significantly change with moderate weight loss except in the group of weight-stable women who had lost 11% of their body weight (Table 3). A 75% regain in organ mass occurred in this group despite weight stability. There was also an increase in organ mass during follow-up in weight-stable men (Table 3) and weight-regaining women (Table 2) that was mainly explained by a significant increase in liver mass (158±178 g, P<0.001 in all subjects). These results argue against a loss in high metabolically active organ mass with weight loss and regain. Further studies need to investigate if the seemingly preferential regain in liver mass is associated with an accumulation of ectopic liver fat. However, the persistent improvement in basal glucose metabolism (Table 2) and γ-glutamyltransferase levels (T0: 35.0±21.6 U l−1, T1: 25.6±7.1 U l−1 P<0.01, T2: 28.4±13.6 U l−1, P=0.49) do not support an increase in intrahepatic lipids during follow-up.

Effect of weight loss and regain on REE

Previous studies have shown that weight loss may lead to an adaptive thermogenesis, that is, a reduction in REE beyond that explained by losses in fat-free mass and FM16, 17, 18, 19, 20, 21, 22 or organ and tissue masses.14 Because this metabolic adaptation favors resistance to further weight loss,22 it may also be the reason for the 30% lower weight loss in weight-regaining when compared with weight-stable women (see Results). There is however conflicting evidence, with some studies in favor47, 48, 49, 50 and others not supporting a long-term persistence of adaptive thermogenesis.51, 52, 53, 54 Methodological issues certainly contribute to the discrepant findings and may arise from the absence of weight stability at the time of testing, a too short period of weight stability, heterogeneity in physical activity and the amount of energy imbalance, macronutrient composition of the diet (that is, carbohydrate and protein content), inadequate normalization of REE for changes in body composition as well as the inclusion of an appropriate control group. Beyond these potential sources of error, extent and duration of metabolic adaptation in response to energy restriction may be subject to interindividual variation (for example, in the amount of brown adipose tissue or in triiodothyronine levels). In a previous publication we found that women with an adaptive thermogenesis were characterized by a greater decline in fT3 during weight loss.14 However, in the present study the decrease in fT3 with weight loss was similar between weight regainers and weight-stable participants (Table 4), suggesting other causes of metabolic adaptation like differences in autonomic nervous system activity.

The a priori identification of individuals who are prone to weight regain would facilitate an intensive support tailored to the phase of weight maintenance. In the present study, baseline parameters like age, obesity, body composition and REE did not differ between weight regainers and weight-stable subjects. A lower weight loss (in women only) and an adaptive thermogenesis were the only characteristics that discriminated weight regainers from weight-stable individuals. A lower weight loss than expected from the prescribed energy deficit may be a first, although an imprecise, approach for identification because poor compliance was responsible for approximately half of the difference between measured and predicted weight loss, whereas 38% was because of adaptive thermogenesis (that occurred in 54% of dieting women) and 14% was explained by the higher proportion of FM in weight loss.15

Study limitation

Unfortunately, information on physical activity throughout the study is missing. Because increased physical activity is known to facilitate weight maintenance after weight loss, our weight-stable group might be characterized by a higher physical activity level that may also have contributed to a higher weight loss and a lower decrease in REE. Interindividual differences in physical activity between weight regainers may have also contributed to the variance in the regain of muscle and fat compartments.

In conclusion, we found no evidence for an adverse effect of weight loss and regain on body fat distribution and the composition of lean mass. In contrast to weight-stable subjects, weight regainers showed an adaptive thermogenesis with weight loss that might explain the lower weight loss (in women) and the propensity to weight regain.