Effects of multiple cycles of weight loss and regain on the body weight regulatory system in rats

Abstract

We studied the effects of multiple cycles of weight loss and regain on the defended body weight in rats. Thirty-six male Wistar rats were divided into three weight-matched groups: weight cyclers (n = 18), ad libitum-fed controls (n = 9), and maturity controls (n = 9). Cyclers underwent four rounds of 20% weight loss from 50% caloric restriction, each cycle followed by recovery to stable plateau weight on ad libitum feeding. Controls ate ad libitum. Maturity controls ate ad libitum and then weight cycled the final two rounds to evaluate the effect of age in later cycles. Cyclers’ postdiet plateau weight became progressively lower than that of controls. With each weight loss, ghrelin increased, while insulin and leptin decreased; the magnitude of these changes did not differ across cycles. After four rounds, cyclers’ weight (504 ± 7 vs. 540 ± 22 g; P < 0.05) and percent body fat (11.7 vs. 15.2%; P < 0.05) were lower than in controls. After a 4-mo follow-up period of ad libitum feeding, cyclers maintained a lower total fat-pad mass versus controls (8.6 ± 0.5 vs. 15.9 ± 3.6 g; P < 0.01) and a lower glucose area-under-the-curve on oral glucose tolerance tests (P < 0.05). Repeated weight-loss cycles exerted positive effects, durably lowering defended levels of body adiposity and improving glucose tolerance.

INTRODUCTION

High rates of obesity and overweight persist in the United States (30), even though Americans in all overweight/obesity categories report trying to lose weight (5, 36). Unfortunately, these efforts rarely yield durable success (25, 47). Repeated episodes of intentional weight loss followed by unintended weight regain have been termed “weight cycling.” Although some studies have warned of health risks such as hypertension, hyperlipidemia (7, 17, 24, 42), and weight gain (14, 23) associated with weight cycling, some experts believe that the known risks of obesity outweigh the potential harms of weight cycling (3a, 28).

The metabolic consequences of weight cycling have been debated. Some human studies have shown decreases in resting energy expenditure following weight cycling (22, 37, 48), whereas others have not (7, 44, 45). Animal studies examining repeated weight-loss cycles have found that in later cycles there is a decreased rate of weight loss (21) and increased rates of weight regain (21, 31), feed efficiency (2, 6, 32, 40), and adiposity (32). Other findings have been contradictory (33). Little is known about the response of important adiposity and appetite-regulating hormones such as leptin, insulin, and ghrelin to multiple cycles of weight loss and regain.

When animals lose weight from caloric restriction, leptin, insulin, and ghrelin levels change adaptively to help restore baseline weight. Leptin is an adipocyte-derived hormone that regulates body adiposity through effects on appetite and thermogenesis (8). Leptin decreases disproportionally to adiposity in diet-induced weight loss (34), decreasing satiety when the body is weight reduced. Insulin, also an important central-acting mediator of energy homeostasis, falls similarly with weight loss (34). Acting in the opposite manner, ghrelin is the only known circulating orexigenic (appetite-stimulating) hormone and has repeatedly been shown to increase in conditions of weight loss and negative energy balance (3, 34, 35, 49), increasing hunger when the body is weight reduced. It is not known whether multiple cycles of weight loss and regain might alter the magnitude of these adaptive hormonal responses, perhaps either progressively augmenting or blunting them. We studied the effects of multiple rounds of weight cycling in rodents on these compensatory responses and the defended level of body weight.

METHODS

Animals.

Thirty-six male Wistar rats were obtained at age 11 wk (vendor: Simonsen) and individually housed in a temperature- and humidity-controlled room and maintained on a 12:12-h reversed light-dark cycle. All procedures were approved by the Institutional Animal Care and Use Committee at the University of Washington and were in accordance with the National Institutes of Health Guide for the Care and Use of Animals.

Study protocol.

Animals were fed a mildly high-fat diet (13.3% fat, 4.14 kcal/g; Laboratory Diet Breeder Chow) throughout the study. Baseline food intake was measured for 10 days, and then, animals were divided into two body weight-matched groups: weight cyclers (n = 18) and ad libitum-fed controls (n = 18). For each weight-cycling animal, average daily food intake during the baseline period was used to calculate a caloric restriction of 50% of average daily calories consumed. Cyclers were provided their individual 50% restricted caloric load daily, and weight was measured daily. When an individual animal achieved 20% weight loss from its baseline weight, deemed the nadir weight, blood samples were obtained via the lateral saphenous vein before restoring ad libitum feeding. During the ad libitum-feeding period, daily weight was monitored until each animal recovered to a stable body weight. The plateau weight was defined as a stable body weight over 3 days with daily weight fluctuations within 50% of the mean weight change of animals in the control group over the previous 2 wk. Once all animals in the cycling group attained plateau weight, the cycle procedures were repeated, beginning with 1 wk of monitored food intake and then 50% restricted caloric daily feeding until the animal reached its 20% weight-reduction nadir and then resumption of ad libitum feeding. Animals were weight cycled in this manner a total of four times over ~1 yr. Controls were fed ad libitum throughout. Before the third restriction period, the controls (n = 18) were subdivided into two weight-matched groups of maturity controls (n = 9), which underwent the same cycling protocol to achieve 20% weight loss followed by weight regain for the last two cycle phases and controls (n = 9) that continued ad libitum feeding. The maturity control group was created to evaluate if changes in weight loss were related to previous episodes of weight loss or age. Study design is shown in Supplemental Fig. S1; Supplemental Materials for this article are available at: https://zenodo.org/record/3255069#.Xa3luxGWyUk. Weight cycling started at age 14 wk, after rats had reached early adulthood. Following the last weight cycle, all animals were fed ad libitum for a 4-mo observation period to assess the long-term impact of weight cycling.

Body weight and food intake were measured daily during the weight-cycling phases of the experiment for the cyclers and every other day for controls. After all four weight cycles were completed and all animals had reached plateau weight, food intake and body weight were measured weekly until they were euthanized. Blood samples collected during the first two cycles were performed at baseline, when each animal reached nadir weight, and again at the end of the plateau phase before each new food-restriction cycle. All animals were fasted for at least 12 h before blood draws. Blood samples were tested for glucose (via chemstick), insulin (via Multiplex assay), leptin (via Multiplex assay), and ghrelin (via RIA assay) levels. Considerable care was taken with the timing of blood draws, because cycling animals at 20% weight loss consumed all their food rapidly and were essentially meal fed. Thus blood draws were timed to occur 3–6 h before the onset of the dark cycle to avoid the distinctive preprandial surges of ghrelin that occur in meal-fed rats (12) and the similar, but smaller, preprandial ghrelin surge in ad lib-fed rats (12).

Body composition and metabolic rate measurements.

Body composition (percentage and total fat and lean mass) was determined by quantitative magnetic resonance (Echo MRI Whole Body Composition Analyzer; Echo Medical Systems, Houston, TX) after all animals had returned to stable body weight following the final weight cycle. A subset of six cycling and six control rats had metabolic rates measured by indirect calorimetry (Oxymax Deluxe Calorimetry System; Columbus Instruments, Columbus OH). The system used a closed chamber where oxygen and carbon dioxide concentrations were measured by volume at inlet and outlet ports to calculate average V̇o₂.

Oral glucose tolerance test and tissue collection.

After the 4-mo observational ad libitum period, all rats underwent an oral glucose tolerance test. Animals were fasted overnight, and blood samples were obtained at 0, 15, 30, 60, 90, and 120 min following oral administration via gavage of 1 g/kg of D50 glucose solution. Two-hour area-under-the-curve was calculated.

Animals were anaesthetized, and then, body weight and snout-to anus length were measured before collection of trunk blood and of fat pad tissue. Retroperitoneal, epididymal, and mesenteric fat pads were weighed upon excision. During the 4-mo follow-up phase, one cycler, three controls, and two maturity controls died. Final measurements were not collected on these animals.

Statistical analyses.

Descriptive statistics were calculated as means ± SE. Time to weight loss and time to weight regain were defined as the number of days from baseline to 20% weight loss and the number of days from nadir weight to recovery of stable plateau body weight. Mixed model tests including all available data points were performed using STATA 13.1 (StataCorp, College Station, TX) to test main effects as well as interactions and group differences through formal post hoc testing.

RESULTS

Body weight, time to weight loss, and weight regain.

Control animals gained weight steadily throughout the study period. After each weight-loss phase, the cycling animals regained weight such that it surpassed their pre-weight loss body weight (Fig. 1A). Weight at the plateau of each cycle showed a main effect of time [χ²(5) = 2127, P < 0.0001] but no main effect of group [χ²(2) = 3.94, P = 0.14]. However, there was a group-by-time interaction [χ²(10) = 42.1, P < 0.0001]. By the third cycle, the plateau body weights of the cycling animals were significantly lower than the control animal body weights (Fig. 1B). The lower body weight of cycling animals persisted through cycle 4 and the subsequent long-term ad libitum-feeding period until euthanized. During the plateau phase after weight-loss cycles 3 and 4, the body weights of maturity control animals were not significantly different than those of controls. However, by the time they were euthanized, both maturity control and cycling rats were significantly lighter than control animals (Fig. 1B).

Fig. 1.

Changes in mean (±SE) body weight over time for cycling, control, and maturity control groups. A: percent change in body weight by group throughout 4 weight-loss/weight-regain cycles and during subsequent long-term follow up. B: weights during plateau phases by group as body weight in grams. Main effect of time and group. C: weights during plateau phases as percentage of initial body weight. Main effect of time and group. #P < 0.08, *P < 0.05, **P < 0.01.

For cycling animals, the mean number of days to 20% weight loss increased with each of the four weight-loss episodes (cycle 1 = 17.3 ± 0.6 days; cycle 2 = 23.8 ± 1.3 days; cycle 3 = 25.3 ± 1.1 days; cycle 4 = 29.0 ± 1.4 days; Fig. 1A). The increased durations were significantly different between cycles 1 and 2 (P < 0.01) and cycles 3 and 4 (P < 0.01). Maturity controls took a mean of 27.3 ± 1.6 and 28.9 ± 2.2 days to reach 20% weight loss during cycles 3 and 4, respectively. Time to 20% weight loss did not differ during cycles 3 and 4 between the cycling and the maturity control groups (P = 0.35, P = 0.88 respectively). Similarly, the time to stable weight regain increased over the course of the study in cycling animals (cycle 1 = 16.9 ± 0.8 days; cycle 2 = 16.1 ± 0.6 days; cycle 3 = 18.3 ± 0.7 days; cycle 4 = 20.5 ± 0.8 days). However, only the increase from cycle 2 to 3 was significant (P < 0.01), and there was no difference in time to weight regain in cycles 3 and 4 between cycling animals and maturity controls (maturity controls cycle 3 = 17.0 ± 1.03 days; cycle 4 = 19.4 ± 1.21 days; P = NS).

Food intake during each recovery weight plateau phase relative to baseline food intake showed a main effect of time [χ²(4) = 72.82, P < 0.01)] and group [χ²(2) = 12.77, P < 0.01]. There was a group-by-time interaction [χ²(8) = 40.89, P < 0.01]. Weight-cycling rats ate less during the post-dieting plateau weight than did control rats after the first cycle, and that difference continued for the duration of the study (Fig. 2). Weight-cycling rats had significantly lower food intake than the maturity control rats during the first two cycles, but that difference went away when the maturity control rats underwent a weight reduction-cycle during cycle 3 (Fig. 2).

Fig. 2.

Food intake as a change from baseline. Mean food intake during last 8 days of weight-recovery plateau periods relative to mean of last 10 days of baseline food intake. Food intake during weight-recovery plateaus was significantly less in weight-cycling rats compared with control rats and maturity control rats after cycles 1 and 2. After cycle 3, food intake was significantly different between the control animals and both the maturity controls and weight-cycling groups. At cycle 4, food intake only differed between the control rats and the weight-cycling rats. **P < 0.01.

Body composition and oral glucose tolerance.

Body composition analysis was performed after all cycling animals had regained a stable weight following completion of cycle 4 (Fig. 3). At that time, the weight-cycling group had significantly lower mean body weight than did controls (Fig. 3, top left). Higher body weights in control animals were explained by greater total body fat mass compared with the cycling and maturity control groups (Fig. 3, top middle). Compared with weight cyclers, controls also had significantly higher percentage of body fat (Fig. 3, top right), without notable differences in lean mass (Fig. 3, bottom left) or percentage of lean mass (Fig. 3, bottom middle). There was a corresponding improvement in glucose tolerance among cyclers, as shown by oral glucose tolerance tests (Fig. 3, bottom right).

Fig. 3.

Differences in body composition and insulin resistance in weight-cycling and control rats after recovery of stable weight following the weight-loss cycle 4. Body weight and composition were measured in all groups once all individual cycling and maturity control animals had returned to stable body weight following the cycle 4. Body composition was determined by quantitative magnetic resonance. AUC, area under the curve; OGTT, oral glucose tolerance test. P values were determined by mixed model. *P < 0.05.

Differences in adiposity persisted to the end of the 4-mo ad libitum-feeding period. Compared with ad libitum-fed controls, mean total fat pad mass (Fig. 4A) was less in the cycling group (P < 0.01) and less in maturity controls (Fig. 4A; P < 0.01). The difference in adipose mass was especially notable in visceral fat pads, which were 113% larger in controls than in the cycling group (P < 0.001) and 94% larger than the maturity controls (P < 0.05). Epididymal pads were 70% larger in controls than cyclers (P < 0.01), and retroperitoneal fat pads were 86% larger (P < 0.01). There were no differences in body length among the groups (Fig. 4B) nor in basal metabolic rate (BMR) as assessed in a subset of weight-cycling and control rats that underwent indirect calorimetry (Fig. 4C).

Fig. 4.

Differences in adiposity, length, and metabolic rate based on prior weight cycling measured after subsequent long-term ad libitum feeding. A and B: body composition (A) and body length (B) measurement taken at death. P value determined by mixed model. C: metabolic rate as measured by indirect calorimetry in 4 cyclers and 4 controls during the long-term ad libitum-feeding observational phase. **P < 0.01.

Appetite-regulating hormone responses to weight cycling.

There was no effect of time or group on ghrelin levels; however, there was an interaction between group and time [χ²(4) = 12.01, P = 0.02]. Although the weight-cycling rats largely had significant changes in ghrelin levels between baseline, nadir, and plateau phases, during the plateau phases there was no difference between the ghrelin levels of the weight-cycling rats and the combined control group rats (cycle 1 P = 0.47, cycle 2 P = 0.11; Fig. 5A). There was an effect of time on insulin [χ²(4) = 39.54, P < 0.001], but there was no effect of group (P = 0.18). However, an interaction between group and time was present [χ²(4) = 12.01, P = 0.02], such that by the plateau after the second cycle, insulin levels were significantly lower in the weight-cycling group (P = 0.02; Fig. 5B). There was an effect of time [χ²(4) = 22.80, P < 0.001] and group [χ²(1) = 4.66, P < 0.03] on leptin levels; however, there was no interaction between group and time [χ²(4) = 5.68, P = 0.22]. Leptin levels were significantly different between the cycling rats and the control rats at the second plateau phase (P < 0.01; Fig. 5C).

Fig. 5.

Hormone levels through weight-loss cycles 1 and 2. A–C: hormone levels [ghrelin (A), insulin (B), and leptin (C)] throughout weight-loss cycles 1 and 2. Maturity controls are included with control animals during these cycles. Markings between time points indicate changes within a group. Markings above a single time point denote differences between groups. #P < 0.08, *P < 0.05, **P < 0.01.

DISCUSSION

Weight cycling in male rats resulted in several beneficial metabolic effects. Episodic weight loss followed by weight regain elicited by ad libitum feeding of a relatively high-calorie diet resulted in a long-term reduction in body weight, body fat percentage, and body fat mass compared with control rats who ate the same high-calorie diet unrestricted throughout the experiment. The benefit appeared to accrue with increasing numbers of episodes of weight loss, and a benefit was derived even when the first weight loss began at later ages. Along with a lower fat mass long term, rats that had previously undergone weight cycling demonstrated improved oral glucose tolerance. There was no evidence for adaptation to weight loss. Although each subsequent weight-loss cycle took longer to achieve a 20% reduction among cycling animals, this same increase in duration was seen in age-matched maturity control animals who had never undergone caloric restriction previously, implying that the longer time to achieve 20% weight loss during later cycles may have been a function of aging or greater weight (older rats had more grams of weight to lose to achieve 20% weight loss), rather than a consequence of previous weight-loss cycles.

Furthermore, there was evidence that weight cycling durably lowered the defended level of body weight and fat mass. Weight cyclers’ body and fat mass remained lower than that of controls even after a prolonged period of unrestricted feeding. Moreover, during weight plateau periods of ad libitum feeding, weight-cycling rats consumed less than the control rats. This fact reflects that the animals appear to have been in energy balance but maintaining a lower body adiposity. In addition, leptin, ghrelin, and insulin concentrations at plateau weights did not differ to a significant degree as compared with control animals, It has previously been shown that a high-fat diet can increase the defended body weight in mice (18) whereas Roux-en-Y gastric bypass surgery lowers it (19) as does seasonal adaptation (27). It is possible that the lower body weights compared with control animals during plateau periods were due to the relatively short duration during which the cyclers were allowed to regain weight, but two factors argue against that interpretation. First, the pace of weight gain of the cycled rats stabilized, and its slope equaled that of controls (as per the protocol design) before the subsequent weight loss cycle was initiated; thus it seems unlikely that they would have attained the weight of the controls over time. Second, the 4-mo observational period of ad libitum feeding that followed all weight-loss cycles should have been sufficient for body mass to reach that of the control rats. Body composition and postmortem analyses were performed at two separate time points, and both support the conclusion that prior weight cycling rendered specific, long-term effects on body adiposity, demonstrating a lower percentage of body fat and significantly less fat pad mass, with no differences in length or lean mass. Calorimetry data showed that despite their reduced weight, the cycling and noncycling rats did not show significant differences in metabolic rate. Since control animals were heavier than cycled animals and would be anticipated to have higher resting metabolic rate, this finding could actually reflect a detrimental effect of prolonged high-fat feeding on metabolism in the control animals.

Humans are prone to regaining weight after weight loss. Subsequent attempts at weight loss take people longer than initial weight-loss attempts (21). This is consistent with our findings that subsequent weight-loss cycles took longer to achieve 20% reduction than the initial cycles. However, the longer cycle time for age-matched control rats indicates that this might be more a function of aging than of previous weight-loss attempts. The rat model indicated that hormone levels including leptin and ghrelin returned to pre-weight loss levels once the rats regained weight, and this is not consistently seen in human studies (16). Some studies have shown that changes in important appetite and adiposity-regulating hormone levels, including decreases in leptin and increases in ghrelin, persist long after weight loss (39). However, this was not seen in our study. Additionally, the weight-cycling rats had the same BMR as the control rats at the end of the study, despite repeated episodes of weight loss. This has not been consistently demonstrated in human studies (16, 22, 37), with some suggesting reduction in BMR in weight-reduced humans that perpetuates after weight regain, whereas others show weight-appropriate BMR after weight loss.

Although human studies have shown a long-term decrease in leptin levels after dieting (16), our rat model indicated that hormone levels return to pre-weight cycling levels over time. Similarly, ghrelin levels returned to pre-weight loss values in the cycling rats. The increased ghrelin in weight-reduced rats was expected and is consistent with homeostatic responses to negative energy balance, thereby promoting return to the higher defended body weight. Once weight was regained, there was no evidence that the rats had persistent homeostatic drive because they ate less than the control rats while maintaining a lower body mass. This is the first study to evaluate hormone levels after two cycles of caloric restriction over such a long duration, showing the ultimate stability of the hormone levels.

One limitation of this study is that the experimental conditions imposed upon rats are difficult to generalize to humans. The rats were subjected to a dramatic caloric restriction as a means to lose weight. Sustaining that degree of caloric restriction is typically challenging for humans. The rats were not exposed to the pervasive food cues that humans experience in our uncontrolled environment. Additionally, although the cycling rats demonstrated a lower long-term level of defended body weight, it is impossible to parse what is secondary to caloric restriction versus weight loss per se. Furthermore, these rats were all a genetically identical male population on a matched diet that might not accurately reflect female rats or other dietary strategies. Future studies evaluating sexual dimorphism in weight-cycling are required, and other peripheral hormones such as PYY or GLP-1 (9) should also be assessed. Genetic or pharmacological manipulations would be needed to more fully assess if the action of peripheral hormones is necessary or sufficient to produce the observed weight effects. Beyond the changes in weight, future studies should evaluate other health implications of cyclic dieting, such as cardiovascular risks. Despite likely health benefits of weight loss (22a), the risks of weight cycling on cardiovascular health are still debated (4).

Overall, the results of this study are very reassuring. They also raise the question of whether weight loss, even when followed by weight regain, could have a positive long-term impact on long-term body weight and lower the defended level of adiposity. The current findings in rodents are consistent with several large human intervention trials, including the Diabetes Prevention Program and the Look AHEAD studies, which both showed that after weight loss people tend to regain most but not all of their lost weight. At 10 and 8 yr of follow-up after the Diabetes Prevention Program and Look AHEAD interventions, respectively, both showed subjects maintaining ~2 kg of weight loss compared with control groups that did not undergo the intensive lifestyle program and experience weight loss (22b, 24a). These data raise the critical question of what the mechanism behind such findings could be. Neither hormone levels nor metabolic rate appeared to be altered after weight cycling to explain why cycling rats were eating less and maintaining a lower body weight when fed ad libitum over time. Caloric restriction is known to have beneficial effects on age-related cardiovascular and metabolic changes and increase longevity (1, 48). Alternatively, a central mechanism controlling weight and energy homeostasis, such as prevention of central leptin or insulin resistance, could have occurred. Finally, arcuate nucleus inflammation and gliosis are known to occur in obesity in rodents (41, 43) and humans (41). It is tempting to speculate that structural changes in hypothalamic tissue, and even loss of critical neurons (41), could participate in resetting body weight over time and, hence, amelioration or avoidance of hypothalamic inflammation and gliosis via episodic caloric restriction prevented the upward drift in body adiposity observed in control animals.

Although there are some studies that indicate that the periods of stress associated with weight cycling may confer negative cardiovascular risks (4), from a purely weight and metabolic perspective this was not shown in our rat experiment. The improvement in fat mass as well as improvement in glucose tolerance seen in our rats that had undergone weight cycling implies metabolic benefits to the periods of caloric restriction, despite the stress of the weight-gain times. Our study uniquely demonstrates these benefits over a prolonged period of time. Additionally, given the apparent increased rate of death in the control group compared with cyclers, there might be some mortality benefit from weight cycling. This agrees with other studies of rodent weight cycling (36). It is unknown whether the benefit is due to caloric restriction, reduced weight, or other factors. Future research should focus on the health implications of weight cycling, including whether there is beneficial impact on metabolic syndrome, such as amelioration of fatty liver and/or hypothalamic gliosis. Additionally, further human research is needed to confirm whether the body defends a lower long-term level of adiposity after periods of prior weight loss, as was indicated by our rat model. If our findings do apply to humans, then patients and clinicians can take heart that it may be better to try and eventually regain weight after weight loss by caloric restriction, even repeatedly, than not to try at all.

GRANTS

This study was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-070826, DK-089036, and DK-098466 and National Heart, Lung, and Blood Institute Grant T32-HL-007028 (to E. A. Schur) and the University of Washington's Nutrition Obesity Research Center (P30-DK-035816).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.S.F., D.E.C., and E.A.S. conceived and designed research; J.L.R., S.J.M., and E.A.S. analyzed data; J.L.R. and E.A.S. interpreted results of experiments; J.L.R. prepared figures; J.L.R. drafted manuscript; J.L.R., S.J.M., D.E.C., and E.A.S. edited and revised manuscript; R.S.F. and E.A.S. performed experiments; R.S.F., S.J.M., D.E.C., and E.A.S. approved final version of manuscript.