Association between the FTO rs9939609 single nucleotide polymorphism and dietary adherence during a 2-year caloric restriction intervention: Exploratory analyses from CALERIE™ phase 2

Abstract

Caloric restriction (CR) improves markers of aging in humans, but it is not known if the fat mass and obesity-associated gene (FTO) rs9939609 single nucleotide polymorphism (SNP), which is associated with increased appetite and energy intake, influences adherence to prolonged CR. Utilizing data from the two-year Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy (CALERIE™) phase 2 randomized controlled trial, we tested whether the FTO rs9939609 SNP was associated with adherence to CR in healthy adults without obesity. As secondary aims, we assessed whether the FTO rs9939609 SNP was associated with changes in body composition, biomarkers of aging, and eating behaviors. Participants were randomized into either a CR group that targeted a 25% reduction in energy intake compared to the habitual energy intake at baseline, or an ad libitum (AL) control group. Participants were genotyped for the FTO rs9939609 SNP. Dietary adherence was determined through changes in energy intake using doubly labeled water and changes in body composition at baseline, month 12, and month 24 in both the CR and AL condition. Weight, body composition, resting metabolic rate (RMR), adiponectin, insulin, leptin, and eating behaviors were measured at the same timepoints. A total of 144 participants (91 CR and 53 AL; age: 38.6 ± 7.1 years; body mass index: 25.3 ± 1.7 kg/m²) were studied. Of these, 27 were homozygous for the ‘obesity-risk’ A allele (AA), while 44 were homozygous for the T allele (TT) and 73 were heterozygotes (AT). By design, the CR group exhibited greater percent CR compared to the AL group during the trial (P < 0.01), but no genotype-by-treatment interaction was observed for change in energy intake or percent CR (P ≥ 0.40). The FTO rs9939609 SNP was also negligibly associated with change in most other endpoints (P ≥ 0.13), though AAs showed a reduction in RMR adjusted for body composition change over the 24 months relative to TTs (genotype-by-treatment interaction: P = 0.03). In a two-year CR intervention delivered to healthy individuals without obesity, the FTO rs9939609 SNP was not associated with adherence to CR and did not alter improvements in most aging biomarkers.

1. INTRODUCTION

Caloric restriction (CR) is an intervention that is defined as a reduction in energy intake without malnutrition [1]. Evidence from numerous non-human species indicates that sustained CR which begins early in life or during mid-life enhances longevity, with greater benefits seen at higher levels of CR [2,3]. Clinical trials in humans also show that CR improves markers of aging. Most notably, evidence from the Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy (CALERIE™) phase 2 study indicates that two years of CR improve biological markers of aging, longevity, and age-related diseases relative to ad libitum (AL) eating [4–6].

Within western societies that are characterized by excess availability of energy-rich foods, there is heterogeneity in energy intake and adherence to CR that may be explained by genetic polymorphisms [7]. In this regard, a cluster of SNPs in intron one of the fat mass and obesity-associated gene (FTO) are associated with obesity-related traits and in turn age-related diseases [8]. The most examined SNP within this intronic region is the FTO rs9939609 SNP. At this SNP, the minor A allele is associated with higher weight in a dose-dependent manner, with people homozygous for the A allele weighing 2-3 kg more than those homozygous for the T allele [9]. Homozygous carriers of the A allele additionally exhibit differences in eating behaviors, greater appetite, and higher energy intake compared to TTs, suggesting that increased food intake mediates FTO-related variations in weight [10–12].

The relationship between FTO, weight, and energy intake has led to studies examining if FTO SNPs affect adherence and weight loss during energy restriction regimens in individuals with obesity. Presently, these studies have yielded equivocal findings: some have shown that carriers of the FTO rs9939609 SNP A allele experienced poorer weight loss outcomes than individuals who do not carry the risk allele [13], while others report that the risk allele is not associated with weight change [14] or may be related to greater weight loss [15]. However, to our knowledge, no study has tested the association between FTO risk alleles and long-term energy intake changes during energy restriction using the gold standard intake-balance method [16]. In addition, no study has investigated whether, within healthy adults without obesity, the FTO rs9939609 SNP influences adherence to CR and subsequently alters improvements in body composition and notable biomarkers of aging, such as the decrease in resting metabolic rate (RMR) per unit of fat and fat-free mass (i.e., metabolic adaptation), an increase in adiponectin, and reductions in leptin and insulin. Such studies would provide insights into whether FTO rs9939609 genotype should modulate the delivery of CR interventions to optimize dietary adherence and CR-induced attenuations in aging within healthy adults.

The primary aim of these analyses was to test whether the FTO rs9939609 SNP is related to adherence to CR during a two-year CR intervention provided to healthy individuals without obesity. Our secondary aim was to see if the changes in weight, body composition, RMR, and eating behaviors is associated with the FTO rs9939609 SNP.

2. MATERIALS AND METHODS

The CALERIE™ phase 2 randomized controlled trial was a two-year study (clinicaltrials.gov registration: NCT00427193) performed at Pennington Biomedical Research Center (Baton Rouge, LA, USA), Washington University School of Medicine (St. Louis, MO, USA), and Tufts University (Boston, MA, USA), with the Duke Clinical Research Institute (Durham, NC, USA) serving as the coordinating center [6,17]. Comprehensive details of the participants, study design, intervention, and outcome measures have been reported [6,17–19]. All institutions received approval from their respective institutional review boards, and participants provided written informed consent and received financial compensation.

2.1. Participants

Participants were eligible if they were aged 20-50 years (men) or 20-47 years (women) and if their body mass index was ≥ 22.0 and < 28.0 kg/m². They were excluded if they had significant medical conditions (including history or clinical manifestation of cardiovascular disease or diabetes), psychiatric or behavioral problems (including history or clinical manifestation of any eating disorders) or had engaged in drug abuse in the previous two years. Individuals with high levels of physical activity (≥ 30 mins on ≥ 5 days/week) were additionally excluded, as well as women who were pregnant or planned to become pregnant during the trial.

Akin to previous research [20], only completers who identified themselves as White were incorporated in the analyses, because the effects of the FTO rs9939609 SNP on appetite [11] and weight [8] have been robustly identified in Whites, and because of differences in allele frequency and linkage distribution patterns in other races [21]. Most participants who were randomized self-identified as White (76.4%), so there was low power to perform analyses in other racial groups; however, results were similar when other races were included in analyses (data not shown).

2.2. Study design

Prior to randomization, participants completed a 6-week baseline period, which included two 14-day measures of total daily energy expenditure (TDEE) through doubly labeled water (DLW) to ascertain energy requirements [16]. Participants were then randomized in a 2:1 ratio into the CR group or the AL group using a permuted block randomization technique. Randomization was stratified by study site, sex, and body mass index (BMI; normal weight: 22.0 ≤ BMI < 25.0 kg/m²; overweight: 25.0 ≤ BMI < 28.0 kg/m²).

The CR intervention aimed to induce an immediate and sustained 25% decrease in daily energy intake relative to baseline. During the first 27 days of the trial, all food was supplied to participants, with three different diets (low glycaemic load, Mediterranean, and low fat) provided during this period [6]. Participants were thereafter able to alter their dietary composition whilst striving to maintain 25% CR. Dietary adherence was primarily determined at 6-month intervals by measuring TDEE via DLW and adjusting for changes in body composition [16]. Moreover, weight change was measured frequently and compared to estimated weight change from mathematical models to provide a frequent metric of adherence between DLW assessments [22,23]. The AL group was instructed to sustain their habitual energy intake.

Interventionists delivered the intervention to participants in the CR group during the trial via periodic individual counselling and group sessions [19]. Several techniques were covered during sessions to assist participants during CR [19]. Each participant, for instance, was instructed to self-monitor food intake with the assistance of food scales, measuring cups and spoons, and portion size training. Other topics of these sessions included maintaining motivation, managing food cravings, managing hunger, goal setting, and social support [19]. AL group participants did not receive any counseling and were instructed to maintain their usual dietary intake. Neither group received an exercise prescription, but both were provided with multivitamin and calcium supplements.

2.3. Genotyping

The CALERIE™ genotype database was developed by the Kobor Lab at the University of British Columbia and the Genomics Analysis Shared Resource at Duke University from CALERIE™ baseline-blood-sample DNA obtained from the CALERIE™ Biorepository at the University of Vermont. Genotyping was conducted using the Illumina Global Screening Array-24 v3.0 (GSA) BeadChips containing 654,027 markers, with ~30,000 add-on markers from Infinium PsychArray-24 focused content panel (Illumina, San Diego, CA). Briefly, 200ng DNA was processed and hybridized to the GSA chips according to the manufacturer’s instructions and scanned using the Illumina iScan platform. Genotypes were called for SNPs matched to dbSNP (v151) [24] and for which valid calls were made in > 98% of participants using GenomeStudio v2.0 (Illumina, San Diego, CA). Additional SNPs were imputed using the IMPUTE2 software suite [25] and the 1000 Genomes Phase 3 reference panel [26].

2.4. Outcomes

In the current analyses, for all outcome measures, data at baseline, month 12, and month 24 were used for both the CR group and the AL group.

2.4.1. Dietary adherence

Dietary adherence was assessed specifically via energy intake and percent CR. Energy requirements at baseline (baseline energy intake) was equal to the mean of the two measures of TDEE that were derived from DLW during the 6-week baseline period [16]. Energy intake and average percent CR from baseline were calculated retrospectively in both groups over 12-month intervals via the intake-balance method, which uses measures of TDEE and changes in body composition and is considered the gold standard assessment of energy intake [16].

2.4.2. Weight and body composition

After an overnight fast of ≥ 8 hours, clinic weight was measured using a calibrated scale (Scale Tronix 5200; Welch Allyn). Fat mass and fat-free mass were measured using dual-energy X-ray absorptiometry (Hologic 4500A, Delphi W, or Discovery A Scanners).

2.4.3. Resting metabolic rate and physical activity

RMR was measured in duplicate using indirect calorimetry over 30 minutes (Vista-MX, VacuMed, Ventura). Further, the RMR residual was calculated as the difference between measured RMR and predicted RMR at each time point, with the latter estimated from a baseline regression of RMR as a function of fat mass and fat-free mass. The change in residual values (follow-up minus baseline) was then used to represent change in RMR adjusted for fat mass and fat-free mass changes and to determine if metabolic adaptation occurred (negative residuals indicating metabolic adaptation) [27]. Physical activity level (PAL) was computed as TDEE/RMR to provide a relative measure of physical activity whilst body composition changes occurred [28].

2.4.4. Blood biomarkers

Fasting concentrations of high-molecuar weight adiponectin, insulin and, leptin were measured using enzyme-linked immunosorbent assays (Alpco, Salem), chemiluminescent immunoassays (Elecsys, Roche Diagnostics), and multiplex immunoassays (Bio-Plex 200, Bio-Rad Laboratories), respectively.

2.4.5. Eating behaviors and attitudes

Eating behaviors and attitudes were measured with the Eating Inventory, which is a 51-item questionnaire that measures dietary restraint, disinhibition, and hunger. Dietary restraint is defined as the intent and ability to restrict food intake, disinhibition is the tendency to overeat, and hunger is the susceptibility to feelings of hunger. Scores for restraint, disinhibition, and hunger range from 0-21, 0-18, and 0-14, respectively, and a greater number indicates greater levels of each respective eating component [29].

2.5. Statistical analysis

Baseline characteristics and outcome measures at baseline were assessed using general linear models and chi-squared tests to determine the effects of FTO rs9939609 genotype and treatment group. Three-way general linear models were used to examine the change in outcome measures from baseline, with genotype (TT vs. AT vs. AA), treatment (CR vs. AL), and time (month 12 vs. month 24) used as fixed factors. For general linear models, sex, study site, and (for change scores) baseline values of the respective outcome were used as covariates. When significant effects of interest occurred in the general linear model, post-hoc comparisons were performed with Bonferroni adjustments to observe where variations lay. To supplement comparisons, partial eta squared (ηρ²) values were calculated as estimates of effect size, with values of 0.01, 0.06, and 0.14 considered small, moderate, and large, respectively [30]. Analyses were performed using SPSS version 25, and the significance level was set to P < 0.05. Unless noted otherwise, baseline characteristics and outcome measures are presented as mean and standard deviation (SD), whereas change scores are presented as estimated marginal means and standard error (SE).

3. RESULTS

3.1. Participants

In total, 220 participants (145 CR, 75 AL) were randomized at baseline, with 168 participants (76.4%; 111 CR, 57 AL) self-identifying as White. Of those 168 participants, 144 (91 CR, 53 AL; 96 [66.7%] females; age: 38.6 ± 7.1 years) completed the trial and were genotyped for the FTO rs9939609 SNP; thus, this sample was included in the analyses.

Overall, 100 participants (69.4%) carried at least one rs9939609 risk A allele, with 27 (18.8%) AAs identified (Table 1). The distribution of males and females was similar amongst genotypes (P = 0.96), and no significant differences were observed for baseline weight and BMI (P ≥ 0.14; ηρ² ≤ 0.03). However, there was a significant effect of genotype on baseline energy requirements (P = 0.03; ηρ² = 0.05), although post-hoc comparisons showed no significant effects following Bonferonni adjustments (P ≥ 0.07; Table 1).

Table 1.

Baseline characteristics of CALERIE™ phase 2 participants according to FTO rs9939609 genotype (TT vs. AT vs. AA) in those who self-identified as White and completed the trial.

		All (N = 144)		TT (N = 44)		AT (N = 73)		AA = (N = 27)
		Mean/Number	SD/%	Mean/Number	SD/%	Mean/Number	SD/%	Mean/Number	SD/%
Age (years)		38.6	7.1	38.6	6.4	39.0	7.2	37.2	8.1
Sex
	Male	48	33.3	14	31.8	25	34.2	9	33.3
	Female	96	66.7	30	68.2	48	65.8	18	66.7
Weight (kg)		73.3	8.9	72.5	7.9	72.9	9.6	75.4	8.6
BMI (kg/m2)		25.3	1.7	25.1	1.5	25.2	1.8	25.8	1.7
Energy requirements (kcal/day)		2503	395	2406	357	2544	398	2553	432

Abbreviations: BMI, body mass index; FTO, fat mass and obesity-associated gene.

Values are mean (±SD), except for Sex, which is number (%).

Significant main effect of genotype from general linear model (P = 0.03); no differences between groups after Bonferroni adjustment (P ≥ 0.07). No other significant main effects (P ≥ 0.14).

3.2. Outcomes

3.2.1. Baseline

At baseline, there was no main effect of genotype and treatment (P ≥ 0.06; ηρ² ≤ 0.04) and no genotype-by-treatment interactions (P ≥ 0.18; ηρ² ≤ 0.02) for energy requirements, weight, and body composition (Table 2). There were likewise no main genotype and treatment effects (P ≥ 0.05; ηρ² ≤ 0.04) and no genotype-by-treatment interactions for RMR, PAL, insulin, leptin, and questionnaire outcomes (P ≥ 0.21; ηρ² ≤ 0.02). However, a main effect of genotype was revealed for adiponectin (P < 0.01; ηρ² = 0.08), with post hoc comparisons showing that TAs displayed higher concentrations than TTs (P < 0.01; Table 2).

Table 2.

Outcome measures at baseline in CALERIE™ phase 2 participants according to FTO rs9939609 genotype (TT vs. AT vs. AA) and treatment (CR vs. AL) in those who self-identified as White and completed the trial.

	TT (N = 44)				TA (N = 73)				AA (N = 27)
	CR (N = 26)		AL (N = 18)		CR (N = 45)		AL (N = 28)		CR (N = 20)		AL (N = 7)
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Energy requirements (kcal/day)	2447	342	2348	380	2537	393	2555	412	2612	445	2392	379
Weight (kg)	72.8	8.6	72.1	6.8	72.8	9.7	73.1	9.6	76.8	8.0	71.3	9.6
Fat mass (kg)	23.3	4.2	25.2	5.6	23.9	4.1	23.2	4.7	25.5	4.5	24.5	3.1
Fat-free mass (kg)	49.5	8.9	46.8	7.8	48.9	9.7	49.9	9.6	51.2	8.5	46.9	9.7
RMR (kcal/day)	1453	187	1421	194	1448	175	1443	218	1501	220	1381	199
PAL	1.69	0.16	1.66	0.23	1.75	0.17	1.76	0.16	1.75	0.19	1.73	0.09
Adiponectin (ng/mL)	3758*	1996	4130*	1630	5031	2440	5973	3937	5041	3943	5971	4076
Insulin (uIU/mL)	6.2	3.0	6.2	2.8	5.0	2.1	5.4	2.5	5.6	1.9	6.2	1.7
Leptin (pg/mL)	13923	10724	19624	14310	14872	12155	12954	9398	14597	8424	16993	12067
Dietary Restraint	10	4	10	4	10	4	10	3	10	4	11	4
Disinhibition	5	3	5	3	5	3	5	3	5	3	6	3
Hunger	4	4	4	3	4	3	3	3	4	4	2	2

Abbreviations: AL, ad libitum; CR, caloric restriction; PAL, physical activity level; RMR, resting metabolic rate.

All values are mean (±SD).

^*Significant main effect of genotype from general linear model (P < 0.01); TAs different than TTs (Bonferroni adjusted P < 0.01). No other significant main or interaction effects (P ≥ 0.05).

3.2.2. Change scores

In accordance with results from the full sample, the CR group exhibited a reduction in energy intake and greater percent CR than the AL group at month 12 and month 24 (main effect of treatment: P < 0.01; ηρ² ≥ 0.30). However, the effect of the CR intervention on energy intake change and percent CR did not vary by genotype, with no interactions involving both genotype and treatment observed and small to negligible effect size estimates seen (P ≥ 0.40; ηρ² ≤ 0.01; Figure 1a–1f). Weight change was also similar between FTO rs9939609 genotypes in response to CR (three-way and genotype-by-treatment interaction: P ≥ 0.13; ηρ² ≤ 0.03), despite a main effect of treatment indicating the CR group lost more weight than the AL group over both time points (main effect of treatment: P < 0.01; ηρ² = 0.54; Figure 1g–1i). Similarly, while the CR group showed a decrease in fat mass and fat-free mass relative to the AL group at months 12 and 24 (main effect of treatment: P < 0.01; ηρ² ≥ 0.34), no interactions involving both genotype and treatment were observed (P ≥ 0.25; ηρ² ≤ 0.20; Figure 1j–1o).

Figure 1.

Change scores in energy intake, percent CR, weight, and body composition from baseline to months 12 and 24 according to FTO genotype. Solid lines and filled circles represent AL group; dashed lines and open circles represent CR group. Values are mean ± SE.

Though a reduction in RMR was seen at months 12 and 24 in response to CR (main effect of treatment: P < 0.01; ηρ² = 0.17), no significant interactions between genotype, treatment, and time were revealed for RMR change (P ≥ 0.09; ηρ² ≤ 0.04; Figure 2a–2c). For the RMR adjusted for fat mass and fat-free mass (RMR residual), no three-way interaction was observed (P = 0.09; ηρ² = 0.04), but there was a significant two-way genotype-by-treatment effect for change in RMR residual over 24 months (P = 0.03; ηρ² = 0.05; Figure 2d–2f). Post-hoc comparisons revealed the RMR residual decreased (i.e., there was metabolic adaptation) over the intervention in CR relative to AL in AAs compared to TTs (P = 0.01). PAL was not altered during the trial (main effect of treatment: P = 0.86; ηρ² < 0.01), and FTO genotype did not interact with the treatment group to modify PAL (P ≥ 0.27; ηρ² ≤ 0.02; Figure 2g–2i).

Figure 2.

Change scores in RMR, RMR residual, and PAL from baseline to months 12 and 24 according to FTO genotype. Solid lines and filled circles represent AL group; dashed lines and open circles represent CR group. Values are mean ± SE.

There was an increase in adiponectin and a decrease in insulin and leptin in the CR group compared to the AL group (main effect of treatment: P < 0.01; ηρ² ≥ 0.18), yet no interactions were observed between genotype, treatment, and time (P ≥ 0.07; ηρ² ≤ 0.02; Figure 3a–3i).

Figure 3.

Change scores in adiponectin, insulin, and leptin from baseline to months 12 and 24 according to FTO genotype. Solid lines and filled circles represent AL group; dashed lines and open circles represent CR group. Values are mean ± SE.

There was no significant three-way interaction for restraint or disinhibition and no two-way interactions involving both genotype and treatment for these constructs (P ≥ 0.09; ηρ² ≤ 0.04; Figure 4a–4f). Restraint was nonetheless greater in CR compared to AL at both time points (main effect of treatment: P < 0.01; ηρ² = 0.52), and the CR group exhibited an increase in disinhibition compared to the AL group over the trial (main effect of treatment: P = 0.02; ηρ² = 0.04). Three-way general linear models revealed a significant genotype-by-treatment-by-time interaction for hunger (P = 0.03; ηρ² = 0.05), although post-hoc comparisons were not significantly different following Bonferroni adjustment (Figure 4g–4i).

Figure 4.

Change scores in restraint, disinhibition, and hunger from baseline to months 12 and 24 according to FTO genotype. Solid lines and filled circles represent AL group; dashed lines and open circles represent CR group. Values are mean ± SE.

4. DISCUSSION

Our primary finding was that the FTO rs9939609 SNP was not associated with dietary adherence during a two-year CR intervention. Likewise, we showed that in response to two years of CR, the FTO rs9939609 SNP was not related to reductions in weight, body composition, insulin, and leptin. The FTO rs9939609 SNP was nonetheless related to the change in RMR adjusted for fat mass and fat-free mass, with AAs displaying a decline compared to TTs.

Current evidence indicates that the FTO rs9939609 obesity-risk A allele increases energy intake [10,12], possibly through greater postprandial concentrations of acylated ghrelin [11,12], and this subsequently increases body weight and risk for age-related diseases [31]. It is less clear, however, whether the FTO rs9939609 SNP influences weight loss and the ability to adhere to a CR diet, and no study has examined such questions utilizing robust assessments of energy intake that overcome limitations linked to traditional measures of free-living energy intake [32]. We found that CR and associated decreases in weight, fat mass, and fat-free mass achieved over two years were similar amongst FTO rs9939609 genotypes. Moreover, consistent with previous findings [33], we found that the FTO rs9939609 SNP did not affect PAL during CR in our cohort. Our observations are novel because, at odds with previous regimens [14,15], the CALERIE™ phase 2 trial exclusively recruited individuals who were normal weight or overweight. It is difficult to speculate whether our findings can be generalized to individuals with obesity and other non-white populations, because this cohort was relatively homogenous. However, it must be noted that CR is safe, improves multiple biomarkers of aging, and could delay the onset of multiple age-related diseases in individuals without obesity [6]. Therefore, we feel our findings are noteworthy for healthy populations without obesity who seek to acquire CR-induced benefits, as they imply that CR regimens do not have to be modified based on the common FTO rs9939609 SNP in order to optimize dietary adherence.

It is possible that any FTO-mediated alterations in energy intake and CR were offset by the systematic, sustained, and highly controlled delivery of the CR intervention. The relative effectiveness of the interventional strategies across FTO rs9939609 genotypes is perhaps illustrated by the negligible between-genotype variations in eating behaviors and attitudes in response to CR. Indeed, during the whole two-year intervention, the CR group received sessions focusing on strategies related to increased restraint (e.g., portion control, healthy meal replacements) and the management of hunger (e.g., maintaining hunger) [19]. Such strategies, therefore, could be critical in negating any FTO-related predispositions to poor adherence during CR. Further work is now needed to determine if other heritable factors and predictors impact the effectiveness of similar interventions in individuals without obesity.

Amongst several theoretical mechanisms of aging, one theory posits that CR slows biological aging via a decrease in energy expenditure or resting metabolic rate (RMR) per unit of fat and fat-free mass [34,35]. This reduction in energy expenditure is termed metabolic adaptation and is hypothesized to attenuate tissue, cellular, and DNA damage and slow aging [35]. In our analyses, despite similar CR exhibited by FTO rs9939609 genotypes, AAs showed a decrease in RMR adjusted for the loss of fat mass and fat-free mass compared to TTs. Our results are broadly supportive of those by Grau and colleagues who observed that carriers of the FTO rs9939609 A allele presented a decline in resting energy expenditure compared to TTs, in spite of no between-genotype differences in body composition change [36]. Further studies are needed to elucidate any implications of these between-genotype variations. Since metabolic adaptation may be a key mediator in the CR-induced attenuations in biological aging [34,35], longer-term observations are required to determine if the same degree of CR slows aging and the development of aging-related diseases to a greater extent in AAs. Similarly, given the impact of resting RMR on appetite and energy balance [37], studies are needed to assess whether the FTO rs9939609 SNP modulates long-term weight change after the cessation of regimented interventions. These investigations are important because the effects of FTO SNPs [13,38] and metabolic adaptation [39] on weight loss maintenance are equivocal.

An increase in adiponectin [40] and a reduction in leptin [35,41] and insulin [42] have been associated with improvements in longevity markers. We observed a CR-induced increase in adiponectin, as well as decrease in leptin and insulin during CR; yet the changes in concentrations were similar in FTO rs9939609 genotypes. While some have suggested that obesity-related SNPs within intron one of FTO alter leptin concentrations [43], our results bolster most cross-sectional observations [11,12,44] and short-term interventional studies [45] in adults showing that no differences in leptin concentrations occur between FTO rs9939609 genotypes when weight and fat mass variations are similar or statistically controlled. Moreover, the between-genotype variations we saw in adiponectin and insulin support the results of others who showed that the same SNP did not alter changes in these hormones during dietary interventions delivered to individuals with overweight or obesity [46,47].

To our knowledge, the current study is the first to examine the effect of the FTO rs9939609 SNP on energy intake during a well-structured CR intervention using gold-standard assessments of energy intake, but limitations are present. Our analysis was exploratory and consisted of white participants without obesity. Hence, though the addition of other races did not affect our primary findings (data not shown), additional studies in other racial groups are required. We additionally did not measure ghrelin, which could be affected by the FTO rs9939609 SNP risk allele [11,12]. Future studies should assess ghrelin and other appetite-related peptides because of this and in light of the increased adiponectin seen in TAs compared to TTs. Indeed, this association may suggest FTO-related interactions between appetite-related peptides and adipokines occur [48]. It must also be acknowledged that the sample size was relatively small, limiting our ability to detect significant FTO-related interactions. Nevertheless, our estimates of effect size indicate that any modulating influence of the FTO rs9939609 SNP is, at most, small.

In summary, we found that during a two-year CR intervention, the common FTO rs9939609 SNP was not associated with adherence to CR. The decreases in weight, body composition, adiponectin, leptin, and insulin were also similar between FTO rs9939609 genotypes; however, the metabolic adaptation that was observed in homozygous carriers of the risk A allele compared to homozygous carriers of the T allele requires further study. Our findings suggest that the FTO rs9939609 polymorphism per se should not alter the delivery of a comprehensive, two-year CR intervention for individuals without obesity looking to obtain aging-related benefits.

ABBREVIATIONS:

AL

ad libitum

CALERIE™

Comprehensive Assessment of Long-term Effect of Reducing Intake of Energy

CR

caloric restriction

DLW

doubly labeled water

FTO

fat mass and obesity-associated gene

RMR

resting metabolic rate

SD

standard deviation

SE

standard error

SNP

single nucleotide polymorphism

TDEE

total daily energy expenditure