Impact of Sustained Calorie Restriction and Weight Cycling on Body Composition in High-Fat Fed Male and Female C57BL/6J Mice

Abstract

Objective:

To investigate body composition changes with weight cycling among adult, diet-induced obese C57BL/6J mice.

Design and Methods:

A total of 555 singly-housed mice were fed ad libitum (AL) a high-fat diet from 8 to 43 weeks of age. The 200 heaviest mice of each sex were randomized to four groups: ever obese (EO, continued AL feeding); obese weight loser (OWL, calorie restricted); obese weight loser moderate (OWLM, body weight halfway between EO and OWL); and weight cycling (WC, diet restricted to OWL followed by AL refeeding cycles). Body weight and composition data were collected. A linear regression was used to calculate residuals between predicted and observed fat mass. Linear mixed models were used to compare diet groups.

Results:

While weight loss and regain resulted in changes in body weight and composition, fat mass, body weight, and relative body fat were not significantly greater for the WC compared to the EO group. During long-term calorie restriction, male (but not female) OWLM remained relatively fatter than EO.

Conclusion:

Weight cycling did not increase body weight or relative fat mass for middle-aged, high-fat-fed adult mice. However, long-term moderate calorie restriction results in lower body weight but greater “relative” fat in male mice.

INTRODUCTION

Nearly 40% of US adults are classified as obese (1), which is associated with an increased risk of numerous adverse health conditions including hypertension, type 2 diabetes, and cancer (2). Sustained weight loss, even at a modest level (5–10% of initial body weight), has been shown to reduce obesity-related disorders (3). National survey data indicates that approximately one quarter to one half of the adult population were trying to lose weight, and the most common strategies were eating fewer calories or less fat and exercising more (4–6). Despite the known ability to reduce body weight through diet, exercise, and/or medical intervention (medications, surgery), the majority of individuals who successfully lose weight regain that weight during the subsequent 2–5 years. In fact, less than half of those who try to lose weight are successful maintainers (7–9). Many individuals undergo repeated bouts of weight loss and regain, also known as “weight cycling” or “yo-yo dieting.”

In most cases of intentional weight loss, the goal is to lose fat. However, it has been demonstrated that body mass reduction via dieting often includes both fat and lean mass (i.e., muscle mass) losses, with weight regain disproportionally as fat mass (10). Accordingly, a popular theory (Canon’s paradox) emerged that “dieting makes you fat” (11), which offered an explanation for weight regain despite dieting. As much of the data was associative in nature, subsequent theory challenged the direction of effect by suggesting that “being fat makes you diet” rather than vice versa (12). It was subsequently hypothesized that sustained energy restriction followed by eating ad libitum (AL, or without restriction), in particular of energy-dense (high-calorie) diets, together with a physically inactive lifestyle, causes additional deposition of body fat (13), which over multiple weight cycles led to ever different body weights and fat levels. This was addressed in a twin study in which subjects who had frequent intentional weight loss (IWL) also had greater weight gain and risk of being overweight than did their non-weight cycling co-twins over the course of 25 years, emphasizing that the association is independent of genetic effects (14). In contrast, an article that reviewed the evidence from studies of weight cycling caused by periodic food insufficiency in both animals and humans suggested that weight cycling is not associated with deleterious effects on body composition (10).

Calorie restriction (CR) decreases fat mass in people who successfully maintain reduced weight through dieting in both the short- and long-term periods of follow-up (15, 16). Reduction of calorie intake may be continuously practiced or periodic in nature (intermittent) and accomplished by a number of different paradigms to reduce energy intake (e.g. intermittent fasting, time restricted eating, alternate day fasting) (17, 18). In pre-clinical rodent models, acute or sustained CR at levels traditionally used for lifespan extension (ranging from 30% to 40% reduction below AL) results in significantly lower body weight, body fat, and lean mass (19–21). Fewer studies have included body composition assessments with less restriction (5–20% CR), with different sexes included in the same design, or analyses of body composition to consider “relative” body fat (adjusting for concomitant lean mass changes) versus absolute fat mass relative to AL-fed controls. Initially, in unexpected results from our previous study, researchers showed that short-term mild CR (5%) in female C57BL/6J mice can cause increased absolute fat mass coincident with lower lean mass without significant changes in total body mass (22), highlighting the potential importance of measuring fat mass and lean mass in addition to body weight with weight modifying interventions. In a clinical study, no significant difference was observed in fat mass or fat-free mass among subjects consuming a very-low-calorie diet and those consuming a balanced-calorie deficit diet in a 12-week controlled trial (23).

Based on these results, the effects of CR might vary due to levels or duration of CR. Furthermore, most pre-clinical models of CR are initiated at younger ages, and those studies performed in older animals are most often in the context of a low-fat diet where an obese phenotype is not present. Understanding the impact of different levels of energy restriction (CR) and weight cycling on adult body composition, including relative fatness, is of translational relevance.

To investigate the impact of weight cycling and CR on body composition in adult organisms after the establishment of obesity, body composition was measured non-invasively in a longevity study with weight cycling and different levels of CR under high-fat diet feeding in adult male and female C57BL/6J mice.

METHODS

Animals and Diets

Five hundred fifty-five C57BL/6J mice (n=277 males and n=278 females) were purchased (Jackson Laboratory, Bar Harbor, Maine) and ad libitum (AL) fed a high-fat diet (D12451, 45% of calories from fat and 20% from protein; Research Diets, New Brunswick, NJ) from 8 to 47 weeks of age. From week 47, the animals with the heaviest body weights (heaviest two-thirds within each sex; n=200/199 for males and female, respectively) were subsequently randomized into four groups (based on the body weight data from week 43): ever obese (EO, n=35/35 [m/f]), obese weight loser (OWL, n=35/34), obese weight loser moderate (OWLM, n=57/58), and weight cycling (WC, n=73/72) (Table 1).

Table 1.

Sample size at each time point of measurement.

	Males (week)					Females (week)
	43a	56	71	84	99	43a	57	73	85	99
EO	35	33	23	20	15	35	35	24	17	9
OWL	35	33	27	26	19	34	33	26	25	21
OWLM	57	57	50	40	31	58	57	49	42	32
WC	73	73	61	39	32	72	70	62	51	42
Total	200	196	161	125	97	199	195	161	135	104

EO: ever obese; OWL: obese weight loser; OWLM: obese weight loser moderate; WC: weight cycling C57BL/6J mice.

^aSample sizes at weeks 8 and 23/24 were identical to week 43.

The EO group received continued AL feeding with the same diet throughout their lifespan. The OWL group was fed the high-fat diet (HFD) restricted in amount to a level sufficient to achieve a group mean body weight similar to that of the reference group of C57BL/6J mice of the same sex fed a low-fat diet (LFD; D12450B, 10% of calories from fat and 20% from protein; Research Diets). These LFD control animals (n=15 for each sex) were acquired 1 month ahead of the HFD cohort and received AL LFD from 8 weeks of age until the end of life, serving as a reference benchmark body weight for each sex of this strain under the equivalent husbandry conditions within the same facility.

The average body weight of the OWLM group was maintained approximately midway between the EO and OWL groups through moderate CR of the same HFD. The WC group was food restricted to a level sufficient to achieve a group mean body weight of the OWL group and then allowed AL feeding until a plateau weight was reached. The weight loss phases were implemented by gradual restriction of about 10% less than the food intake of the previous week, with no more than 40% restriction, until they reached targeted weight maintenance status (for OWL and OWLM groups) or restriction-refeeding (for WC groups) to reduce negative side effects associated with acute, severe CR implementation (24).

To ensure that the mice received sufficient micronutrients during restriction, the mice were fed the HFD with supplementation (same 45% of calories as fat as the D12451 diet but with 67% added vitamins and minerals; D11022101, Research Diets) when they were restricted by more than 20% of the amount being consumed by the EO group. Therefore, the OWL group (around 25~30% of restriction overall) received this special high-fat diet over their lifespan after randomization, and the WC animals received it during the weight loss phase when restriction was more than 20% of EO food intake.

Body weight and body composition data were collected prior to randomization to groups at 8, 23/24, and 43 weeks of age and among the randomized groups among the ages of 43 and 99 weeks (week 43, baseline randomization; week 56/57, first weight trough for WC group; week 71/73, first weight regain peak for WC group; week 84/85, second trough for WC group; week 99, second peak for WC group). The LFD group was not included in the analysis because they were 1) fed a different macronutrient composition known to influence body composition, 2) not CR at any point in the study design, 3) from a separate cohort, and 4) included in the study simply as a body-weight benchmark. The experiment was performed over an extended duration where the numbers of animals diminished with age in the study due to natural death or moribund status. The actual sample sizes included for analysis at different time points are listed in Table 1. All mice were single housed for the duration of the study; housing conditions were as previously described (12:12 light:dark cycle, 20–22°C ambient temperature) (25).

All procedures were performed in accordance with the Institutional Animal Care and Use Committee at the University of Alabama at Birmingham.

Measurements

From 43 weeks-of-age, the body weight of all animals was measured weekly. Body composition (fat and lean mass) was determined in vivo in all groups by quantitative magnetic resonance (QMR) (EchoMRI™ 3-in-1,V2.1; Echo Medical Systems, Houston, TX) (26) at peak and trough points of the WC group (weeks 43, 56/57, 71/73, 84/85, and 99). There were differences in the time of measurement between male and female groups of mice because of the time needed to reach the targeted weight and/or staggering measurements to accommodate the number of animals.

The food intake of the EO group and of the WC group with AL refeeding (weight regain phase) was measured weekly. During restriction phases, the OWL group, the OWLM group, and the WC group were provided a daily ration in the afternoon (around 15:00). The amount was manually prepared to a specific weight (± 0.05g) on the basis of the animals’ weight maintenance needs. Any leftover food was removed and recorded daily.

Statistical Analyses

Descriptive statistics (mean ± standard deviation (SD)) were used to describe the body weight, fat mass, and lean mass at each measurement.

For relative fat mass, the use of ratios (percentage fat as fat mass divided by body weight, or fat mass to lean mass) is common in obesity research, but there are sometimes drawbacks to the use of ratios to adjust data and regression-based approaches to controlling for covariates often mitigates these (27). Instead, residuals from the regression of lean mass on fat mass among AL-fed groups were used to compare the relative fat mass. Body composition (fat and lean mass) data of all animals at weeks 24 and 43 (when all animals were AL- and HFD-fed), and EO group only (because they were AL- and HFD-fed throughout the study) at weeks 56/57, 71/73, 84/85, and 99 were used to generate reference regression lines separately for males and females. Relative fat mass was then calculated as the residuals (observed values for fat mass minus predicted values for fat mass using lean mass as predictor) for all animals.

To compare differences in body weight, fat mass, lean mass, and relative fat mass among diet groups at weeks 56/57, 71/73, 84/85, and 99, we ran linear mixed models with group, week (as a factor), and a group by week interaction. We used data from weeks 43 to 99 and included a random effect for animal ID to account for repeated measures. Reported p-values in Table 2 are pairwise comparisons of estimated marginal means.

Table 2.

Pairwise comparisons between diet groups for body weight, fat mass, lean mass, and relative fat mass at different time points

Group comparison	Outcome	Males Week				Female Week
		56_	71ˉˉ	84_	99ˉˉ	57_	73ˉˉ	85_	99ˉˉ
EO - WC	Body Weight	<0.001	0.021	<0.001	0.696	<0.001	0.041	<0.001	0.001
	Fat mass	<0.001	0.157	<0.001	0.154	<0.001	0.047	<0.001	0.003
	Lean mass	<0.001	<0.001	<0.001	0.006	<0.001	0.332	<0.001	0.027
	Relative fat mass	0.041	0.247	0.040	<0.001	<0.001	0.290	0.936	0.351
EO - OWLM	Body Weight	<0.001	<0.001	<0.001	0.114	<0.001	<0.001	<0.001	<0.001
	Fat mass	<0.001	<0.001	0.009	0.786	<0.001	<0.001	<0.001	<0.001
	Lean mass	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	Relative fat mass	0.009	0.010	0.001	<0.001	0.011	0.790	0.610	0.908
OWL - OWLM	Body Weight	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	Fat mass	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	Lean mass	<0.001	0.713	<0.001	0.007	<0.001	<0.001	<0.001	<0.001
	Relative fat mass	0.001	<0.001	<0.001	<0.001	0.038	0.059	0.001	0.061
OWL - WC	Body Weight	>0.999	<0.001	0.539	<0.001	0.509	<0.001	0.083	<0.001
	Fat mass	0.588	<0.001	0.297	<0.001	0.702	<0.001	0.115	<0.001
	Lean mass	0.181	<0.001	0.106	<0.001	0.140	<0.001	0.077	<0.001
	Relative fat mass	0.161	<0.001	0.036	<0.001	0.439	0.246	0.005	0.377
EO - OWL	Body Weight	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	Fat mass	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	Lean mass	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	Relative fat mass	0.582	0.017	0.935	0.599	<0.001	0.063	0.021	0.143
OWLM - WC	Body Weight	<0.001	<0.001	<0.001	0.014	<0.001	<0.001	<0.001	<0.001
	Fat mass	<0.001	<0.001	<0.001	0.153	<0.001	<0.001	<0.001	<0.001
	Lean mass	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	Relative fat mass	<0.001	0.056	<0.001	0.020	0.104	0.330	0.425	0.216

EO: ever obese; OWL: obese weight loser; OWLM: obese weight loser moderate; WC: weight cycling C57BL/6J mice

^“_”: body weight trough

^“ˉˉ”: body weight peak. Displayed are p-values from post-hoc pairwise comparisons after running a linear mixed model with group, week, and grou-p*week interaction with a random effect for animal ID. Reported p-values are not adjusted for multiple comparisons; p<0.05 highlighted in bold.

To compare differences in body weight, fat mass, lean mass, and relative fat mass within the weight cycling diet group between peak-peak and trough-trough a linear mixed model with week as a factor and a random effect for animal ID was run. Comparisons were done between peaks (week 43 and 71/73, and week 71/73 and 99) and between troughs (week 56/57 and 84/85). Table 3 reports p-values from those pairwise comparisons.

Table 3.

Pairwise comparisons of weight cycling (WC) groups between peak-peak and trough-trough measurements in body weight, fat mass, lean mass, and relative fat mass.

	WC Males			WC Females
	43:71 \/	71:99 \/	56:84 /\	43:73 \/	73:99 \/	57:85 /\
Body Weight	Δ=7.2 g p<0.001	Δ=1.1 g p=0.234	Δ=4.5 g p<0.001	Δ=7.1 g p<0.001	Δ=1.5 g p=0.036	Δ=3.6 g p<0.001
Fat Mass	Δ=4.6 g p<0.001	Δ=0.6 g p=0.432	Δ=1.5 g p=0.034	Δ=5.8 g p<0.001	Δ=0.2 g p=0.801	Δ=0.6 g p=0.322
Lean Mass	Δ=2.0 g p<0.001	Δ=0.7 g p=0.002	Δ=1.5 g p<0.001	Δ=1.2 g p<0.001	Δ=1.2 g p<0.001	Δ=1.4 g p<0.001
Relative Fat Mass	Δ=0.7 g p=0.169	Δ=−0.7 g p=0.273	Δ=−1.4 g p=0.021	Δ=1.8 g p=0.009	Δ=−3.9 g p<0.001	Δ=−4.0 g p<0.001

WC: weight cycling C57BL/6J mice groups

^\/: body weight decrease and increase (peak-peak comparison)

^/\: body weight increase and decrease (trough-trough comparison). Displayed are differences between time points (positive values indicate increase between timepoints) and p-values from post-hoc pairwise comparisons after running a linear mixed model with week and with a random effect for animal ID in the weight cycling diet group. Reported p-values are not adjusted for multiple comparisons; p<0.05 highlighted in bold.

Model assumptions for both sets of mixed models were checked by plotting raw data and standardized residuals. Standardized residuals were further evaluated with regards to outliers, normality, and homoscedasticity. Normality of residuals was not violated (skewness statistic < 2 for all outcomes), but we identified a few outliers (standardized residuals larger than ± 3 SD) that we excluded in sensitivity analyses (see Supplemental Tables S1 and S2). We also observed heteroscedasticity among groups and/or weeks for some outcomes, however, we account for those unequal variances using an unstructured covariance matrix.

All reported p-values were unadjusted for multiple comparisons. All statistical analyses were performed with R version 4.1.2 and RStudio version 2022.07.1 (RStudio Team (2020)).

RESULTS

As the study was designed and implemented, the OWLM group was maintained halfway between that of the EO and OWL groups (Figure 1). The body composition measurements for all groups at each trough and peak of the WC group are shown in Figure 2 (for statistical differences among groups, please refer to Table 2). The average body weight, fat mass, and lean mass kept increasing with age in the female EO group but dropped at the last measurement (week 99) in the male EO group.

Figure 1. Body weight.

The mean (± standard deviation) body weights of all animals included in the analyses (week 8 to week 99) are shown for C57BL/6J males (A) and for females (B). EO, ever obese; OWL, obese weight loser; OWLM, obese weight loser moderate; WC, weight cycling. [See Table 1 for sample size at each time point].

Figure 2.

Body composition. Fat mass (A, B) and lean mass (C, D) measures in all groups of C57BL/6J mice at each time point by sex. Data shown are mean ± standard deviations. EO, ever obese; OWL, obese weight loser; OWLM, obese weight loser moderate; WC, weight cycling. See Table 2 for statistical comparisons. [See Table 1 for sample size at each time point].

The individual body composition data of all animals at week 23 and 43 and EO group from week 56 to week 99 were used to create the regression line (red), that are shown in Figure 3-A for male (fat mass = −23.92 + 1.84*lean mass, R² = 0.71, p<0.001, total n = 490) and Figure 3-B for female animals (fat mass = −48.02 +3.35*lean mass, R² = 0.66, p<0.001, total n = 482) .The average relative fat mass (residuals to the regression line, represented as unit fat mass for a given lean mass) of each sex are shown in Figure 3-C for male and Figure 3-D for female animals. The relative fat mass of male WC group shifted with body weight at either body weight peak or trough points, while the relative fat mass of female WC showed a different pattern of an increase up to week 56 and a decrease thereafter. The relative fat mass of both male and female EO groups first increased and then decreased with age.

Figure 3.

Relative fat mass. Reference regression line of lean mass on fat mass from all diet groups for weeks 23/24 and 43 and the EO group after week 43 of male (A) and female C57BL/6J mice (B); relative fat mass (residuals to the regression line) for each group of mice of males (C) and females(D). Data shown in (C) and (D) are mean ± standard deviation. EO, ever obese; OWL, obese weight loser; OWLM, obese weight loser moderate; WC, weight cycling. See Table 2 for statistical comparisons for relative fat mass. [See Table 1 for sample size at each time point].

Overall, the monitoring of body weight for the WC groups satisfied our goal of cycling between EO and OWL groups. After the first weight loss and regain cycle at week 71, the body weight and lean mass of the male EO group were significantly different than that of the male WC group (p=0.021, p<0.001, respectively), but there was no significant difference between these two groups in fat mass or relative fat mass (p=0.157, p=0.247, respectively) (Table 2). Although, the female WC group regained most of their body weight and fat mass as compared to the EO group, both remained significantly lower (p=0.041, p=0.047, respectively) at week 73 after the first cycle, and there was no significant difference between them in lean mass and relative fat mass (p=0.332, p=0.290, respectively). After another weight loss and regain cycle at week 99, there was no significant difference in body weight (p=0.696) and fat mass (p=0.154), between male EO and WC group, however the EO group had significantly lower lean mass (p=0.006) and higher relative fat mass (p<0.001) than the WC group. Between the female WC and EO group after the second weight cycle at week 99, they were not significantly different in relative fat mass (p=0.351), though there were significant differences in body weight (p=0.001), fat mass (p=0.003), and lean mass (p=0.027).

The relative fat mass of male OWLM group was significantly higher than the male EO group at all four timepoints (at week 56:p=0.009, 71: p=0.010, 84: p=0.001, 99: p<0.001), while the body weight, fat mass and lean mass of male OWLM group were generally significantly lower than male EO group except for body weight and fat mass at week 99 (p=0.114, p=0.786, respectively, Table 2).

For females, however, the relative fat mass of both OWLM and OWL groups decreased across time, especially for the OWL group (Figure 3D). The relative fat mass of the female WC group was not significantly different than female EO group at weeks 73, 85, and 99 (p=0.290, p=0.936, p=0.351, respectively) but was significantly different at week 57 (p<0.001). The body weight, fat mass, lean mass, and relative fat mass of female OWLM group were all significantly higher than the female OWL group (p<0.05) with the exception of relative fat mass at weeks 73 and 99 (p=0.059, p=0.061, respectively).

Weight cycling group

Comparisons of body weight, body composition, and relative fat mass between peak-peak or trough-trough of the WC group are shown in Table 3. After the first weight cycle (peak-peak weeks 43:71/73), both male and female WC groups had significantly higher body weight (p<0.001 for both sexes), higher fat mass (p<0.001 for both sexes), higher lean mass (p<0.001 for both sexes) than they did before weight loss. Relative fat mass was significantly higher among females (p=0.009) in week 73, but not among males (p=0.169) in week 71, compared to week 43.

Compared to the first regain peak (week 71/73), body weight, lean mass, and relative fat mass were significantly higher in WC females (p=0.036, p<0.001, p<0.001, respectively) at the second regain peak (week 99), whereas in WC males only lean mass was significantly higher (p=0.002). Body weight, fat mass, and relative fat mass were not significantly different in males between week 71 and 99 (p=0.234, p=0.432, p=0.273, respectively).

The results from the trough-trough comparison between weeks 56/57:84/85 shows that WC males and females had significantly higher body weight (p<0.001 for both sexes), fat mass (males only, p=0.034), and lean mass (p<0.001 for both sexes) and significantly lower relative fat mass (p=0.021, p<0.001, respectively) at the later time point (Table 3).

DISCUSSION

The study on which the present analysis was based was a randomized controlled trial designed to test the impacts of weight cycling, CR, and body composition on mortality in both male and female C57BL/6J mice (25). In this study, we evaluated the impacts of weight cycling and long-term CR on changes in body weight and body composition after diet-induced obesity between 43 and 99 weeks of age.

In an earlier study of long-term weight cycling in rats, researchers suggested that repeated weight loss and regain reduces final body weight and fat-free mass but not total fat content (28). However, a review of the literature for human (29) and animal (10) studies did not support the theory that weight cycling permanently alters body composition or body fat distribution. Based on our study, if comparisons are made within the same group before and after a completed weight cycle, mice are heavier with higher amounts of body fat and lean mass following an initial weight cycle in both males and females. Yet, considering the fact that animals get heavier with normal aging, as shown by the EO groups in the present study, it is not sufficient to conduct self-comparisons before and after weight cycles. Based on the results of our study, we found that weight cycling did not make male or female mice relatively fatter when compared with AL-fed EO groups after initial weight regain or low-bodyweight control OWL groups after initial weight loss. The age of the animals at the completion of the second full weight cycle overlapping with the increase in mortality and thus smaller sample sizes observed in the groups present challenges for separating age-related effects on body composition and potential morbidity or impending mortality-related effects. Researchers who have conducted other studies suggest that weight cycling without exercise might contribute to the loss of lean mass in an aging population (28, 30–33), while age-related loss of lean mass was not observed for any of the groups after the initial weight reduction within the duration of the study. To our knowledge, ours was the first animal study to look at the effect of weight cycling on body weight and body composition incorporating an age-matched, AL-fed (high-fat diet) comparison group in both sexes.

CR normally reduces body weight and body fat, but in our previous study, we observed that short-term mild CR (5%) can lead to increased absolute fat mass in female C57BL/6J mice (22). In the present study, we found that the moderately calorie restricted OWLM group had a lower absolute fat mass, but with a higher relative fat mass than the EO group at most weeks for males though not for females (Table 2, Figure 3B, Figure 3D), albeit both male and female OWLM groups have significantly increased lifespans (~14% and 18% mean increase) and improved health-related markers as previously reported (25). However, the EO group had significantly higher absolute body weight, fat mass, and lean mass than the OWLM group at most measurements for both sexes. Similarly, OWLM males were relatively fatter than WC males at all weeks based on all measurements except week 71 in relative fat mass. One possible explanation is that reduced meal frequency is associated with different and more sustained elevations of plasma glucose and delayed insulin response (34). We observed that most of the calorie-restricted mice would gorge at daily feeding, consuming almost the entire daily food allotment within the first hour after the food was provided, which might impair insulin regulation. A recent report of weight cycling achieved by switching between high-fat and low-fat diet feeding periods in mice has shown reduced glucose tolerance, reflecting insulin secretion impairment more so than whole-body insulin action (35). Similarly, time-restricting access (6 hours per day) to a high-fat diet resulted in lower weight gain and body fat percentage, despite elevated levels of systemic inflammatory markers (36). In light of these findings, we speculate that impaired glucose and insulin regulation, as well as metabolic adaptations in mice during long-term moderate CR, could affect the utilization of adipose tissue. However, this phenomenon might differ with varying dietary compositions and in humans. In one study, 1 year of CR (~12%) by overweight healthy adults led to significantly less fat mass and percentage fat mass compared with non-restricted controls (37). In another study, there was a significant reduction in fat mass of individuals consuming 1 meal/day without CR compared to those consuming 3 meals/day with the same amount of calories, with both groups maintaining their body weight during the 6 month study (38). It is noted that few dietary strategies to limit calorie intake in humans achieve a similar level of metabolic alteration with daily feeding/fasting events as is observed in the reported murine model.

In most studies, researchers have examined the effects of weight cycling on body composition within only one sex. However, different sexes might have different responses to weight cycling and CR. Sex differences in body composition are observed in U.S. adults at all ages (older than 8 years), with females having different fat mass or percentage body fat than males (39). In this study, the reference regression line of the EO groups used for the relative fat mass comparison had a greater slope in females than in males (Figure 3A & Figure 3c). These differences are possibly determined by a complex interplay of genetic, epigenetic, and hormonal factors (40). We also observed sex differences in body weight and body composition from the 8-month high-fat diet feeding in these mice (41). In the present study, sex differences were observed in the regain of body weight, fat mass, and lean mass after the first weight cycle. WC males were not able to reach the levels of the EO group, whereas WC females returned to a close level of the EO group. This might partially be the result of the relatively smaller body mass in females. In the longitudinal self-comparison of WC groups before and after the second weight cycle (Table 3), WC females had significantly different lean mass and relative fat mass (between weeks 73 and 99), whereas WC males had different fat mass (between weeks 71 and 99). Additionally, in OWLM groups, which experienced sustained durations of energetic restriction, the relative fat mass of males was significantly greater than that of the WC group at most timepoints, whereas relative fat mass of females was not significantly different between OWLM and WC across timepoints. The relationship(s) among restricted food availability and lower body weight with greater relative adiposity and longevity gains in the male OWLM mice provide evidence in support of hypotheses related to perceptions of energetic uncertainty and adiposity. As we and others previously hypothesized and recently conceptualized, perception of the threat to energetic certainty may result in differential partitioning of energetic resources through physiologic and metabolic responses, ultimately contributing to the retention or expansion of body adiposity stores under isocaloric or nutritional deficit states (42, 43).

The present study benefits from a design utilizing animals that were individually housed and fed daily provisions of specific amounts of food for CR, eliminating the concern regarding ‘diet’ cycling versus weight cycling. The use of validated, longitudinal, non-invasive, in vivo body composition assessments via QMR further enhanced the ability to measure individual animal body composition over the time-period of the study with minimal disturbance to the animals or feeding behavior interference that can result from other body composition methodology assessments that require anesthetic immobilization during the measurement. Importantly, the inclusion of the proper controls (AL fed the HFD for the duration of the study) permitted the comparison and interpretation of ‘relative’ changes in body composition among groups. In addition to these strengths, multiple limitations are noted. The use of a single diet composition contrasts with the variability of dietary intake in free-living individuals and the social isolation of single-housing may have further impacts on food intake. Similarly, feeding once/day for CR periods, which may have circadian impacts on timing of restriction, may not reflect normal daily dietary intake patterns even in those that practice time restricted eating or other calorie restriction paradigms. Finally, the age at induction and duration of obesity establishment (adolescent versus adult-onset) with limited number of weight cycles able to be completed within the duration of the current study prior to the increase in unintentional, age-related changes in body composition when nearing mortality should be considered in assessing life stage dependent effects and the translational generalizability of these robust pre-clinical findings. Future studies utilizing similar models could be designed to address some of the hypotheses regarding the method of CR and AL feeding implementation, age of onset of obesity phenotypes, number of weight cycle events and other aspects related to body composition and health outcomes.

In conclusion, weight loss and regain resulted in heavier and fatter mice when compared with baseline but not heavier or fatter mice within the period seen in this study than AL-fed animals that were never weight cycled. Moderate CR results in lower absolute fat mass, but increased relative fat mass, especially in male mice. Sex differences need to be taken into consideration when investigating weight cycling and CR in obesity research.

What is already known about this subject?

• Weight loss is often followed by weight regain.

• Weight cycling is proposed to increase fat mass after regain.

• Calorie restriction decreases fat mass (and lean mass).

What does this study add?

• Cycles of weight loss and regain do not lead to increased body weight or relative fat mass in mice compared to appropriate controls.

• Long-term, moderate calorie restriction results in lower absolute fat mass, but greater relative fat mass than sustained obesity or weight cycling in male mice.

How might your results change the direction of research or the focus of clinical practice?

• Provides perspective regarding the use of relevant control groups for assessment of body weight and body composition changes, including relative fat mass, in pre-clinical models of weight loss or weight cycling.

• Identifies sex-differential outcomes and evidence to consider how the context and conditions (unique versus translatable) within the framework of energetic uncertainty influence nutrient partitioning in pre-clinical versus clinical models.