The Effect of a Leptin Phenotype on Weight Change and Energy Expenditure Responses to Acute and Prolonged Energetic Stressors

Abstract

Leptin is a hormone produced by adipocytes that plays a crucial role in regulating energy homeostasis and body mass. Despite its close correlation with body fat, up to ~40% of variation in plasma leptin concentration remains unexplained, allowing for the classification of a distinct “leptin phenotype”. This leptin phenotype - characterized by either relatively high or relatively low leptin concentration relative to an individual’s level of body fat – presents an intriguing opportunity to test whether relatively higher (compared to lower) leptin concentrations differentially affect energy expenditure, metabolic adaptation, and susceptibility to weight change in response to energy balance perturbations. To test this hypothesis, we characterized the energy expenditure and weight change response between the two leptin phenotypes (relatively high vs low) using three distinct experimental contexts; a cross-sectional analysis (N=104), acute (24-hour) perturbations with fasting and overfeeding (N=77), and chronic perturbations with 24-month caloric restriction (N=144) or 8-week overfeeding (N=28). Leptin phenotype did not explain variations in energy expenditure responses either in cross-sectional analyses or in response to acute or prolonged energetic stressors. Moreover, leptin phenotype was not a determinant of weight change in response to energy restriction or surplus, or subsequent weight recovery. These results suggest that classifying individuals based upon a leptin phenotype does not allow to detect differential susceptibility to energy expenditure adaptations or weight change.

NEW & NOTEWORTHY

Leptin is linked to body fat, but unexplained variation remains. This study challenges the idea that distinct leptin phenotypes – characterized by relatively high or low leptin concentration for a given level of body fat – affects energy expenditure or weight change in response to acute or prolonged energy stressors. We found no association between leptin phenotypes and energy expenditure or weight change either cross-sectionally or in response to acute or prolonged over- or underfeeding.

Graphical Abstract

INTRODUCTION

Leptin is a hormone predominantly produced by adipocytes that has a well-established role in the regulation of energy expenditure, energy intake, and – as a result – body mass (1,2). During weight loss, plasma leptin concentrations decrease (3,4), resulting in reduced energy expenditure and stimulation of hunger or appetite (5). Combined, these mechanisms are thought to contribute to plateauing of weight loss and weight regain that is often observed following a voluntarily initiated reduction in body mass (6). Moreover, it has been suggested that the phenotypical response of the weight-reduced state characterized by reduced energy expenditure coupled with a hedonic drive to increase energy intake mirrors that of animal models that are deficient in (or unresponsive to) leptin (7). Clinically, administration of recombinant leptin to replace circulating concentrations to pre-weight loss levels can assist with weight maintenance and attenuation of the adaptive responses to energy expenditure (8,9).

Leptin is closely related to body fat (10,11), and changes in body fat are related to changes in leptin (3,4). However, despite this strong correlation, there is considerable variability in leptin at any given level of body fat. Body fat percentage accounts for approximately 60–80% of the variance in plasma leptin concentrations including within healthy, weight-stable individuals (10,12). This implies that the remaining variance is explained by factors other than body fat per se. Recently, it has been suggested that this variability allows for the characterization of a distinct “leptin phenotype”; or identification of individuals that differ by fasting leptin relative to body fat, but are matched across other characteristics (e.g., overall body mass or body fat) (12). A “relatively high leptin” phenotype – defined sex-specifically as having relatively higher leptin than predicted from body fat percentage – has been associated with insulin resistance and hyperinsulinemia, compared to individuals with “relatively low leptin” and independent of overall body mass. Although a nascent concept, these results suggest that a given leptin phenotype confers differential susceptibility to cardiometabolic risk which is hypothesized to result from increased exposure to leptin and/or the underlying adipocyte physiology contributing to such.

Given the well-established role of leptin in regulation of body mass (7) and energy expenditure (e.g., in the form of weight change-induced metabolic adaptations) (4,13), the characterization of this distinct leptin phenotype presents a novel opportunity to further our understanding of individual differences in energy expenditure cross-sectionally as well as responses to imposed energy imbalance and resulting body mass regulation. Specifically, the identification of this leptin phenotype raises the question of whether chronic exposure to relatively higher leptin concentrations affects energy expenditure responses as well as resistance to weight change and subsequent recovery of body mass. Demonstrating such differential relationships would have direct clinical implications. Measuring fasting plasma leptin in addition to body fat could be a simple means to identify people more likely to be resistant to weight loss and/or prone to weight (re)gain. To test this working hypothesis, we analyzed leptin, body mass, and energy expenditure data from several unique studies: (1) cross-sectional data of the Pennington Nutrition and Obesity Research Center (NORC) repository, (2) prospective data following an acute over- and underfeeding stimulus, and (3) prospective data following a prolonged over- and underfeeding stimulus. We aimed to characterize the energy expenditure responses in the “relatively high leptin” (RHL) compared to “relatively low leptin” (RLL) phenotype and investigate whether this phenotype would confer differential susceptibility to weight change.

MATERIALS AND METHODS

Approach

This study involved data from three unique experimental paradigms. Study 1 used repository data within the Pennington NORC repository to evaluate the relationship between leptin concentrations and energy expenditure cross-sectionally. Study 2 employed data from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK in Phoenix, AZ) to explore energy expenditure changes in leptin phenotype in response to both acute 24-hour fasting and 200% overfeeding with diets of differing macronutrient proportions. Study 3 included data from the 2-year Comprehensive Assessment of the Long-term Effect of Reducing Intake of Energy (CALERIE) Phase 2, and from an 8-week overfeeding study conducted at Pennington Biomedical Research Center (PBRC), again comparing weight and energy expenditure changes in study participants classified into the leptin phenotype. An overview of the study design of this analysis is provided in Figure 1.

Figure 1.

Study design. The effects of a relatively high (RHL) vs relatively low (RHL) “leptin phenotype” on energy expenditure were evaluated cross-sectionally in a large cohort (Study 1); on energy expenditure in response to a variety of acute nutrient stimuli in the form of 24-hour under- and overfeeding (Study 2), and on energy expenditure and body mass changes longitudinally in response to prolonged under- and overfeeding (Study 3). All studies were secondary analyses on deidentified data and exempt from IRB approval.

Experimental protocols and study measures

For Study 1, we queried the NORC repository for cross-sectional data that included fasting plasma leptin, body composition (i.e., fat mass, fat-free mass) assessed via dual-energy X-ray absorptiometry (DXA), and a measure of energy expenditure in the form of either resting metabolic rate (RMR), total daily energy expenditure (TDEE) assessed via doubly labeled water (DLW), and/or 24-hour energy expenditure (24hEE) assessed in a whole-room indirect calorimeter measured at baseline or prior to any intervention. The query yielded 104 participants for which the following deidentified data was obtained: Sex and age at the time of the visit were collected at baseline. Height was measured in duplicate without shoes using a wall-mounted stadiometer. Body mass was measured in duplicate following an overnight fast using a calibrated scale with participants wearing a hospital gown and no shoes. The weight of the gown was subtracted to obtain true metabolic body mass. Body mass index (BMI) was calculated as body mass divided by height in meters squared. Circulating leptin concentration in samples obtained after an overnight fast was assessed by the PBRC Clinical Chemistry Laboratory via ELISA according to manufacturer’s instructions (EMD Millipore, Billerica, MA).

Study 2 included data from 77 individuals partaking in a series of metabolic assessments as inpatients at the Phoenix NIDDK Branch that were designed to characterize energy expenditure responses to multiple short-term dietary interventions in a crossover design (14). The interventions consisted of 24-hour fasting and 24-hour 200% overfeeding with varying macronutrient content, including low-protein high-fat, high-fat normal-protein), high-carbohydrate normal-protein, high-protein low-carbohydrate, and balanced conditions. Study eligibility criteria included weight stability over the preceding 6 months (i.e., changes in body mass of ≤5%); good health based on medical assessment, history, and lab tests; and normal glucose regulation based on the results of an oral glucose tolerance test (15). Included in this analysis are data on plasma leptin concentration and 24hEE assessed in a whole-room indirect calorimeter as previously described (16,17). Prior to the first dietary intervention, participants were placed on a weight-maintaining diet that consisted of 20% protein, 30% fat, and 50% carbohydrates. The same diet was used for two energy balance assessments in the calorimeter. During the second assessment, the 24-hour energy intake provided was equal to the 24hEE from the first assessment to achieve energy balance with greater precision. The 24hEE measured in this second eucaloric session was then doubled to derive the prescribed 24-hour energy intake for the overfeeding conditions (i.e., 200% of energy requirement); only chamber data from those who consumed at least 95% of the study food were included. All dietary interventions, namely fasting and the varying 200% overfeeding diets provided for 24 hours while in the calorimeter were conducted in a random order and with a 3-day period on the weight-maintaining diet between each assessment to minimize carry-over effects. The change in 24hEE during each dietary intervention was calculated as the absolute difference from the baseline 24hEE under eucaloric conditions. Plasma leptin measurements were obtained from blood samples obtained after an overnight fast before and after entering the whole-room indirect calorimeter (14); leptin values used in this analysis were calculated as the average of all pre-chamber measurements within each individual. Samples were collected using EDTA-containing tubes and were frozen at −70°C to be processed at the NIDDK Core Laboratory in Bethesda, MD, with leptin measured using ELISA following manufacturer’s instructions (EMD Millipore, Billerica, MA) (14).

Study 3 included deidentified data from individuals enrolled in the CALERIE Phase 2 study that tested the efficacy of prescribed caloric restriction (CR) compared to an ad libitum group (18–20), and from individuals completing an 8-week overfeeding study (21). For the underfeeding paradigm, only data from the intervention group (N=144) were included. The CR intervention targeted an immediate and sustained 25% reduction in energy intake from energy requirements as determined by DLW at baseline for the first 12 months followed by 12 months of subsequent weight maintenance, both timepoints of which were used in this analysis. To guide adherence, a mathematical model was used to predict and visualize expected weekly changes in body mass, and participants followed an intensive behavioral intervention (22). Included in this analysis are data on plasma leptin concentration, body mass and body composition, and TDEE assessed using DLW. For the overfeeding paradigm, baseline energy requirements were determined using DLW at study start, whereafter participants commenced an overfeeding protocol that provided 140% of baseline energy requirements for 8 weeks. Participants remained free-living but all meals were consumed under supervision. The study included a small number of females (N=6) and therefore only male participants (N=28) are included in this analysis to be able to properly define the two leptin phenotypes. Included in this analysis are data on plasma leptin concentration, body mass and body composition, and 24hEE measured in a whole-room indirect calorimeter.

Energy expenditure assessments

Energy expenditure assessments included measures of RMR, 24hEE determined in a whole-room indirect calorimeter, TDEE assessed using DLW. RMR was assessed using a metabolic cart according to standard operating procedures following 30 minutes of rest prior to testing. 24hEE was assessed while participants stayed in a whole-room indirect calorimeter, with provided meals tailored to the caloric target of the specific intervention (see above for study details). Energy expenditure was calculated from VO2 and VCO2 according to the Lusk (23) and Weir (24) equations, respectively; with 24hEE extrapolated from the average O2 consumption and CO2 production per minute during the time spent in the chamber. TDEE via DLW (2 g of 10% enriched H218O and 0.12 g of 99.9% enriched 2H2O per kg of estimated total body water as assessed via DXA) was determined based on 18O and 2H enrichment in urine samples collected at baseline and following ingestion. The enrichment was used to determine isotope elimination rates, and isotope dilution spaces were calculated by extrapolation to the baseline zero time point. Production of CO2 was calculated with an estimated respiratory exchange ratio of 0.86 and used to determine TDEE using the Weir equation (24).

Statistical analysis

Data are presented as mean (standard deviation) or N (%) for continuous or categorical variables, respectively. For each study, the leptin phenotype was defined based on the residuals of the sex-specific (male, female) cubic fit between body fat percentage and plasma leptin concentration; individuals were classified as having relatively low (RLL) or relatively high (RHL) leptin if they fell below or above the median value of the residuals’ distribution, respectively (12). Baseline anthropometric and clinical outcome characteristics across the median split within each sex separately (i.e., across the leptin phenotype) were compared using a linear model with leptin phenotype included as a fixed effect. Differences in anthropometric and clinical outcomes over time in the intervention studies (Study 2 and 3) between the leptin phenotypes within each sex were assessed using a linear mixed model including fixed effects for leptin phenotype (i.e., RLL, RHL) and time point (e.g., baseline, follow-up) as well as the interaction, and a random effect for participant to account for repeated measures within the individual. Additionally, to assess the robustness of any derived relationships found via this approach, we performed sensitivity analyses using leptin residuals as a continuous predictor of the outcome instead of comparing across the leptin phenotype categorically. To assess energy expenditure in response to body mass changes over time, a prediction equation of TDEE (for the underfeeding paradigm) or 24hEE (for the overfeeding paradigm) at baseline was developed using a multiple linear regression model including fat-free mass, fat mass, sex, and age. At subsequent time points, energy expenditure was predicted based on the equation developed at baseline, with the difference between measured and predicted energy expenditure changes reported as a measure of metabolic adaptation (also termed adaptive thermogenesis) in response to body mass change induced via under- or overfeeding.

RESULTS

Cross-sectional data: Leptin phenotype does not explain variation in energy expenditure (Study 1)

We examined 104 participants (50% male, 32±12 years of age) to explore cross-sectional differences in measures of energy expenditure between leptin phenotypes. Participants were identified as having “relatively low leptin” (RLL) or “relatively high leptin” (RHL) depending on whether they fell above or below the median value of the residuals of the sex-specific cubit fit between body fat percentage and plasma leptin concentration. Individuals were equally split into sex-specific leptin phenotypes (N=26 each), and distributions of leptin concentration in relation to body fat percentage separated by leptin phenotype are presented in Supplementary Figure S1. There were no differences in age, body mass, or body fat between the leptin phenotypes, though leptin concentration differed by design (Table 1), with an effect estimate of −4.4 ng/mL (95% CI: −7.7, −1.0) between the leptin phenotype in males (P=0.02) and −18.5 ng/mL (95% CI: −26.4, −10.6) in females (P<0.001). There were no observed differences in measures of energy expenditure (RMR, 24hEE, or TDEE) between individuals having relatively low or relatively high leptin levels in either sex (Table 1).

Table 1.

Cross-sectional comparison of energy expenditure between leptin phenotypes (Study 1)

Male
	RLL	N	RHL	N	Effect Estimate (95% CI), P Value
Leptin, ng/mL	5.4 (3.6)	26	9.8 (7.7)	26	−4.4 (−7.7, −1.0), P=0.02
RMR, kcal/d	1700 (228)	26	1716 (232)	25	−16 (−145, 114), P=0.81
24hEE, kcal/d	2279 (370)	26	2314 (213)	26	−35 (−225, 156), P=0.72
TDEE, kcal/d	2786 (644)	22	2943 (651)	24	−156 (−542, 229), P=0.42
Female
	RLL	N	RHL	N	Effect Estimate (95% CI), P Value
Leptin, ng/mL	19.7 (38.2)	26	38.2 (16.5)	26	−18.5 (−26.4, −10.6), P<0.001
RMR, kcal/d	1393 (200)	23	1494 (242)	24	−101 (−231, 30), P=0.13
24hEE, kcal/d	2173 (315)	14	2031 (498)	22	142 (−162, 446), P=0.35
TDEE, kcal/d	2083 (156)	15	2132 (346)	17	−50 (−248, 149), P=0.61

Summary data presented as mean (SD). Individuals were classified as having relatively low (RLL) or relatively high (RHL) leptin levels if they fell below or above the median value of the residuals’ distribution of the sex-specific cubic fit between body fat percentage and fasting plasma leptin concentration, respectively. Effect estimates alongside corresponding 95% confidence intervals derived from linear model including leptin phenotype as a fixed effect. RMR, resting metabolic rate; TDEE, total daily energy expenditure assessed via doubly labeled water; 24hEE, 24-hour energy expenditure assessed in whole-room indirect calorimeter.

Acute interventional data: No differences in 24hEE change between leptin phenotypes in response to acute under- or overfeeding (Study 2)

A total of 77 participants (81% male, 37±10 years of age) were included in this analysis to explore responses to acute underfeeding (i.e., 24 hours of fasting) and overfeeding (i.e., 24 hours of 200% overfeeding) (12), though not all participants completed all conditions (Supplementary Table S1). Using the median value of the residuals of the sex-specific cubic fit between body fat percentage and plasma leptin concentration, Participants were identified as having RLL (N=31 male, N=7 female) or RHL (N=31 male, N=8 female) based on whether they fell above or below the median, respectively. There were no differences in age, body mass, or body fat between the RHL and RLL phenotype, except for leptin concentrations differing between the groups by design (Supplementary Table S1). The effect estimate for leptin concentration in males (N=31 each) was −5.2 ng/mL (95% CI: −8.5, −1.9) between leptin phenotypes (P=0.003) and in females (N=7 RLL; N=8 RHL), it was −24.5 ng/mL (95% CI: −41.0, −8.0) between the phenotypes (P=0.007). There were no significant differences in 24hEE between sex-specific leptin phenotypes within the different dietary conditions (Figure 2), except for in females during the 200% CNP overfeeding condition (−139 kcal/d; 95% CI: −241, −36 kcal/d; P=0.01). However, the sample size for females was small and this association disappeared in a sensitivity analysis examining the continuous association between leptin residuals and 24hEE (P>0.05).

Figure 2.

Energy expenditure changes to 24-hour fasting or 24-hour 200% overfeeding between “leptin phenotypes”. Individuals were classified as having relatively low (RLL) or relatively high (RHL) leptin levels if they fell below or above the median value of the residuals’ distribution of the sex-specific cubic fit between body fat percentage and fasting plasma leptin concentration, respectively.

Chronic interventional data: No differences in body mass, body composition, or energy expenditure changes between leptin phenotypes in response to prolonged under- or overfeeding (Study 3)

In the underfeeding paradigm (Study 3), 144 participants (31% male, 38±7 years of age) from the CALERIE 2 trial (18–20) were included to explore changes in body mass and composition as well as energy expenditure responses to prolonged underfeeding (25% prescription) over 12-month followed by 12 months of subsequent weight maintenance between the RHL and RLL phenotype. Participants were split into leptin phenotypes (N=22 in each phenotype for males, N=50 each for females), and there were no differences in age, body mass, or body fat between the leptin phenotypes except plasma leptin concentration as expected (Table 2). Effect estimates for leptin concentration between phenotypes were −4.8 ng/mL (95% CI: −7.4, −2.2) in males (P<0.001) and −15.4 ng/mL (95% CI: −20.2, −10.7) in females (P<0.001). Within males and females separately, the change in body weight and body fat throughout the intervention of 12 months of caloric restriction followed by 12 months of weight maintenance was related to the relative change in leptin concentration, except during weight maintenance in males which may have been due to limited variability in the data (Supplementary Table S2). However, leptin phenotype was not a determinant of the loss of bodyweight, body fat, lean body mass, or overall body mass during weight loss or subsequent weight loss maintenance/weight regain (Table 2). Furthermore, the leptin phenotype did not moderate the extent of metabolic adaptation (Figure 3, Supplementary Table S3). Additionally, adding leptin to the equation predicting energy expenditure at baseline did not improve the predictive ability (<1% additional variance explained).

Table 2.

Body mass and composition response to chronic under- and overfeeding between leptin phenotypes (Study 3)

Male
	RLL	N	RHL	N	Effect Estimate (95% CI), P Value
Leptin, ng/mL	4.3 (2.5)	22	9.1 (5.4)	22	−4.8 (−7.4, −2.2), P<0.001
Weight Loss
Δ Body Mass, kg	−9.6 (2.9)	21	−9.2 (3.1)	18	−0.4 (−2.2, 1.4), P=0.67
Δ Fat-Free Mass, kg	−2.9 (1.0)	21	−2.9 (1.5)	18	0.1 (−0.9, 1.0), P=0.91
Δ Fat Mass, kg	−6.5 (2.5)	21	−6.2 (2.8)	18	−0.3 (−1.8, 1.2), P=0.67
Δ Fat Percentage, %	−5.8 (2.4)	21	−5.1 (2.9)	18	−0.7 (−2.2, 0.8), P=0.37
Weight Maintenance
Δ Body Mass, kg	0.3 (1.7)	20	1.3 (2.0)	17	−1.2 (−3.1, 0.7), P=0.21
Δ Fat-Free Mass, kg	−0.1 (1.1)	20	0.3 (1.3)	17	−0.5 (−1.4, 0.4), P=0.29
Δ Fat Mass, kg	0.4 (0.9)	20	1.1 (1.7)	17	−0.9 (−2.4, 0.7), P=0.26
Δ Fat Percentage, %	0.5 (1.0)	20	1.1 (1.8)	17	−0.7 (−2.3, 0.8), P=0.34
Weight Gain
Δ Body Mass, kg	7.6 (2.1)	14	7.0 (2.0)	14	0.6 (−0.9, 2.2), P=0.42
Δ Fat-Free Mass, kg	3.4 (1.8)	14	2.9 (1.1)	14	0.5 (−0.6, 1.6), P=0.38
Δ Fat Mass, kg	4.2 (1.0)	14	4.1 (1.7)	14	0.1 (−0.9, 1.2), P=0.80
Δ Fat Percentage, %	3.1 (1.0)	14	2.9 (1.4)	14	0.1 (−0.7, 1.0), P=0.77
Female
	RLL	N	RHL	N	Effect Estimate (95% CI), P Value
Leptin, ng/mL	13.6 (7.4)	50	29.0 (15.0)	50	−15.4 (−20.2, −10.7), P<0.001
Weight Loss
Δ Body Mass, kg	−7.6 (3.2)	46	−7.9 (2.9)	44	0.3 (−0.8, 1.5), P=0.58
Δ Fat-Free Mass, kg	−1.3 (1.2)	46	−1.5 (1.2)	44	0.3 (−0.3, 0.8), P=0.35
Δ Fat Mass, kg	−5.9 (2.4)	46	−6.0 (2.5)	44	0.1 (−0.8, 1.1), P=0.79
Δ Fat Percentage, %	−5.5 (2.2)	46	−5.5 (2.6)	44	0.1 (−0.9, 1.0), P=0.91
Weight Maintenance
Δ Body Mass, kg	1.1 (1.6)	43	0.9 (1.7)	39	0.0 (−1.2, 1.3), P=0.95
Δ Fat-Free Mass, kg	0.0 (1.2)	43	0.1 (1.1)	39	−0.2 (−0.7, 0.4), P=0.56
Δ Fat Mass, kg	1.0 (1.2)	43	0.8 (1.4)	39	0.2 (−0.8, 1.2), P=0.69
Δ Fat Percentage, %	1.1 (1.3)	43	0.9 (1.8)	39	0.3 (−0.7, 1.3), P=0.57

Summary data presented as mean (SD). Individuals were classified as having relatively low (RLL) or relatively high (RHL) leptin levels if they fell below or above the median value of the residuals’ distribution of the sex-specific cubic fit between body fat percentage and fasting plasma leptin concentration, respectively. Effect estimates alongside corresponding 95% confidence intervals derived from linear model including leptin phenotype as a fixed effect. Leptin values for males included in the weight gain paradigm were 5.2 (3.4) ng/mL and 8.0 (5.9) ng/mL for the RLL and RHL phenotype, respectively (effect estimate: −2.8 (−6.5, 0.9), P = 0.14).

Figure 3.

A comparison of metabolic adaptation in response to weight loss and subsequent weight maintenance as well as overfeeding-induced weight gain between leptin phenotypes. Individuals were classified as having relatively low (RLL) or relatively high (RHL) leptin levels if they fell below or above the median value of the residuals’ distribution of the sex-specific cubic fit between body fat percentage and fasting plasma leptin concentration, respectively. Metabolic adaptation is defined as the unexplained change in energy expenditure calculated as the difference between measured and predicted energy expenditure changes.

Finally, we explored whether body mass, body composition, and energy expenditure responses to 8 weeks of 140% overfeeding in 28 male participants (27±5 years of age) differed based across leptin phenotypes (N=14 each). There were no differences in age, body mass, or body fat between leptin phenotypes at baseline. Even though numerically trending in the expected direction by design (−2.8 ng/mL; 95% CI: −6.5, 0.9; P=0.14), there was no statistically significant difference in plasma leptin values between leptin phenotypes at baseline, likely due to the smaller sample size and the narrow range of observed body fat percentage in this population (Table 2). There were no differences between leptin phenotypes in the changes in body mass or body composition with overfeeding (Table 2). Similarly, baseline leptin phenotype was not a determinant of the degree of weight gain-induced metabolic adaptation (also termed adaptive thermogenesis) during overfeeding (Figure 3, Supplementary Table S3). In this population, adding fasting leptin concentration to the predictive equation at baseline was a statistically significant contributor and explained an additional 7% of the variation in baseline 24hEE. However, calculating metabolic adaptation based on the prediction equation that included fasting leptin did not affect the conclusions.

DISCUSSION

This study explored the potential role of a leptin phenotype – defined as relatively high or relatively low leptin concentrations in relation to body fat levels within an individual (12) – in moderating changes in energy expenditure and body mass and/or composition cross-sectionally as well as in response to varying energetic perturbations, including acute (24-hour) under- and overfeeding and prolonged caloric restriction or excess. To this end, we leveraged multiple datasets utilizing robust methodologies including the gold-standard assessments of energy expenditure via whole-room indirect calorimetry and doubly labeled water. Collectively, our findings indicate that there was no evidence of this leptin phenotype explaining variation in energy expenditure cross-sectionally, and leptin phenotype was similarly not associated with differential metabolic adaptation or susceptibility to weight change when explored longitudinally across both acute and prolonged interventions. These results suggest that chronic exposure to elevated leptin concentrations (relative to a given level of body fat) does not meaningfully contribute to variability observed in energy balance in eucaloric conditions as well as in response to caloric restriction or overfeeding.

The proposed leptin phenotype has been reasoned to be reflective of differences in underlying adipose tissue biology (12), and experimentally, leptin has been shown to induce insulin resistance in adipose tissue (25,26). Specifically, leptin impairs insulin signaling in adipocytes by directly interfering with the insulin signaling pathway, resulting in reduced glucose uptake and impaired insulin-stimulated lipogenesis. In addition, leptin promotes the release of FFAs from adipose tissue into the bloodstream, further exacerbating insulin resistance, as elevated levels of FFAs are known to interfere with insulin signaling in various tissues, including adipose tissue. As a result, chronically elevated exposure to this hormone could be hypothesized to result in differential weight and fat mass changes independent of any effect of adiposity per se. Furthermore, based on data in rodents it has been suggested that chronic leptin-melanocortin signaling promotes shortening of melanocortin-4 receptor-bearing primary cilia of hypothalamic neurons. This is an important driver of age-related obesity, thereby offering a potential mechanism via which a phenotype of relatively high leptin could be detrimentally affecting weight change in response to energy stressors, resulting in predisposition to development of obesity (27). However, we did not find evidence of such effects in the present analyses, which highlights the complexity of the energy homeostatic system and the role of leptin in individuals with sufficient endogenous leptin production (9,28). Despite the strong correlation between leptin and body fat and its established role in energy balance (1,2,10), its clinical utility for guiding weight management appears limited based on our findings.

There are several potential mechanisms that might underlie the failure of this leptin phenotype to explain energy expenditure and body mass responses. First, compensatory adaptations within leptin signaling pathways (e.g., via centrally integrated feedback loops) may buffer the effects of chronically elevated or reduced leptin concentrations (29). Second, other relevant hormones such as adiponectin or insulin might contribute to the coordinated response to energetic stressors that is aimed at maintaining (or reestablishing) energy homeostasis, which may obscure any potential effects of a single hormonal agent (30,31). Third, the characterization of this leptin phenotype is based on hormone concentrations assessed in the overnight fasted state; however, leptin concentrations display a distinct circadian pattern (13,32) and as such, using an assessment of leptin at a single point in time might be insufficient to allow for characterization of distinct phenotypes. It is possible that other measurements of leptin concentration (e.g., those in response to a meal) may show a differential relationship to energy expenditure responses and weight regulation. Lastly, leptin concentrations in the peripheral circulation may not be representative of local hormonal actions and signaling effects, such as transport across the blood brain barrier (33,34), variations in kinetics and flux between leptin production and clearance (35). Hence, the combined effect of such alterations on the background of possible genetic variations (e.g., leptin receptor sensitivity) as well as environmental or lifestyle exposures require more mechanistic, translational research approaches.

The present analysis benefits from the integration of diverse datasets that employed both acute and prolonged intervention paradigms. Combining these tq2qrials thereby allowed for a comprehensive characterization of energy expenditure and body mass responses across the proposed leptin phenotypes and in the context of varying energy stressors. Furthermore, we included studies conducted during weight stability as well as during sustained weight loss/gain – hence, we explored the explanatory power of leptin phenotypes under both static and dynamic conditions. However, several limitations warrant discussion. Although the sample sizes were robust for some of the included data sets, others such as the 8-week overfeeding paradigm, were relatively small, which may have limited statistical power to detect an effect. Similarly, datasets were not balanced for sex and although the leptin phenotype was defined based on the sex-specific association between body fat and leptin, we were not able to explore potential interactions between sex and leptin phenotypes. Additionally, even though a significant body of research exists on the role of leptin in the regulation of energy homeostasis, the concept of this leptin phenotype has only recently emerged (12). As such, its utility in explaining interindividual variability in energy expenditure and resulting body mass changes in response to nutrient stimuli remains to be explored. This is of particular importance given the development of leptin resistance which occurs at higher levels of adiposity and acts as a potential confounder. Additionally, the binary categorization across the median of the continuous distribution of leptin residuals into two distinct “phenotypes” ignores variability within each group and carries the risk of amplifying small differences near the cutoff point to create spurious results in subsequent analyses (36). This is of particular concern in smaller data sets such as those employed herein, and we have therefore performed sensitivity analyses using the continuous spectrum of data distributions to address the robustness of any statistically significant findings in our analyses. However, because we did not find statistical evidence of any differences across the categorical leptin phenotypes, it is unlikely that we missed any effects that would be of clinical significance. Finally, the population included in our study was comprised of generally healthy adults and as such, generalizability of these findings to individuals with severe obesity or cardiometabolic conditions in whom the underlying (patho)physiology and background leptin exposure as well as leptin signaling is presumably different remains uncertain. Similarly, the datasets included females across life stages (i.e., pre-, peri-, and postmenopausal) and it has been suggested that the loss of estrogen associated with menopause reduces leptin concentrations (37), which our study was unable to account for or tease out.

In conclusion, our findings suggest that there is no evidence of the proposed leptin phenotype affecting energy expenditure or susceptibility to weight change in response to energetic stress. These results challenge the hypothesis that chronic exposure to relatively higher or lower leptin concentrations than would be predicted based on any given level of body fat drives meaningful interindividual differences in regulation of energy expenditure and resulting body mass. Given the complex and multifaceted nature of energy homeostasis, future research is warranted to elucidate other determinants of variability, including in more diverse and/or clinical populations, as well as potential contexts in which variability in leptin concentrations may exert more pronounced effects.

DATA AVAILABILITY

Data are available from the corresponding author upon reasonable request.