Low Physical Activity Accentuates the Effect of the FTO rs9939609 Polymorphism on Body Fat Accumulation

RESEARCH DESIGN AND METHODS

The FTO rs9939609, rs8050136, and rs7193144 polymorphisms were genotyped in 17,508 Danes comprising five study groups: 1) the population-based Inter99 study sample (clinical trial reg. no. NCT00289237, clinicaltrials.gov) (n = 6,104), sampled at the Research Centre for Prevention and Health (10); 2) unrelated type 2 diabetic patients (n = 2,015), sampled through the outpatient clinic at Steno Diabetes Center (SDC) (hereafter referred to as the SDC type 2 diabetes study group); 3) a population-based group of unrelated middle-aged glucose-tolerant subjects (n = 661) examined at SDC (the SDC control group); 4) the ADDITION (Anglo-Danish-Dutch Study of Intensive Treatment in People with Screen Detected Diabetes in Primary Care) study cohort (clinical trial reg. no. NCT00237548) (n = 8,382) sampled by the Department of General Practice at the University of Aarhus, Aarhus, Denmark (11); and 5) a population-based sample of young healthy Danish Caucasians (n = 346) recruited from the Research Centre for Prevention and Health (12). All participants in the Inter99 and SDC control group study samples underwent a standard 75-g oral glucose tolerance test, and all participants in the young study group (group 5) underwent a tolbutamide-modified intravenous glucose tolerance test (12).

All study participants were Danes by self-report. Informed written consent was obtained from all subjects before participation. The studies were approved by the regional ethical committees and were in accordance with the principles of the Declaration of Helsinki. Type 2 diabetes, impaired glucose tolerance, impaired fasting glycemia, and normal glucose tolerance were defined according to the World Health Organization (13). Overweight and obesity were defined as BMI ≥25 and ≥30 kg/m², respectively. Interaction studies with physical activity and insulin sensitivity were performed in study group 1, including subjects with normal glucose tolerance, impaired fasting glycemia, impaired glucose tolerance, and screen-detected diabetes. Further information on the study groups can be found in an online appendix (available at http://dx.doi.org/10.2337/db07-0910).

Biochemical and anthropometrical measurements.

Obesity-related measures, fasting serum lipids, plasma glucose, and serum insulin were measured as described (10–12,14–16). Further information can be found in the online appendix. The level of physical activity was self-reported by questionnaire (14) and divided into categories as physically passive, light or medium physically active, and hard or very hard physically active. The BIGTT insulin sensitivity index (BIGTT-S_i) and the BIGTT acute insulin response (BIGTT-AIR) were calculated as described (15).

Genotyping.

The FTO rs9939609, rs8050136, and rs7193144 polymorphisms were genotyped using Taqman allelic discrimination (KBioscience, Herts, U.K.). Discordances between 1,464 random duplicate samples were 0.27, 0.14, and 0.14%, respectively, and the genotyping success rates were 97.4, 97.8, and 97.4%, respectively. All genotype groups obeyed Hardy-Weinberg equilibrium.

Statistical analyses.

Fisher's exact test was applied to examine differences in genotype distributions and allele frequencies between affected and unaffected subjects, and logistic regression was used, assuming an additive model, when adjustments for sex, age, and BMI were introduced. A general linear model was used to test quantitative variables for differences between genotype groups, assuming an additive, dominant, or recessive model. Adjustment for sex, age, and BMI was applied when appropriate. The Benjamini and Hochberg method was used to correct for multiple testing, considering both the number of traits and the genetic models tested. Correction for multiple testing was performed separately in the two studies of quantitative traits. Linear models extended with environmental parameters were used to test for interaction using an ANOVA test, assuming an additive model. BIGTT-S_i was included as a covariate, while physical activity and glucose tolerance status were treated as categorical variables. A weighted analysis of the interactions was also performed, wherein variance of BMI in the subgroups of physical activity and insulin sensitivity was estimated from the residuals of a linear model with sex and age included. All analyses were performed with RGui, version 2.5.0 (http://www.r-project.org). P values <0.05 were considered significant.

RESULTS

Due to near-perfect linkage disequilibrium between the three genotyped SNPs (mean r² = 0.99), we excluded rs8050136 and rs7193144 from further analyses. The overall minor allele frequency for FTO rs9939609 was 41.6%.

We validated the previous observation of a strong unadjusted association between the FTO rs9939609 A-allele and type 2 diabetes (odds ratio 1.13 [95% CI 1.06–1.20]) (Table 1). The association between rs9939609 and type 2 diabetes was abolished when adjusting for BMI (1.06 [0.97–1.16], P_additive = 0.2). The A-allele was associated with overweight and obesity in the population-based Inter99 study sample, the ADDITION study cohort, and the SDC type 2 diabetes study group separately but not in the SDC control group (supplementary Table B). When combining these four study groups, we found a strong association with both overweight (1.19 [1.13–1.24], P_{allele frequency} = 1 × 10⁻¹²) and obesity (1.27 [1.20–1.34], P_{allele frequency} = 2 × 10⁻¹⁶) (Table 1).

In the population-based Inter99 study sample, the FTO rs9939609 A-allele was highly associated with obesity-related measures including BMI (P_additive = 1 × 10⁻⁹), body weight (P_additive = 2 × 10⁻⁹), and waist circumference (P_additive = 1 × 10⁻⁷). No convincing association with fasting levels of serum triglyceride or cholesterol or with post–oral glucose load levels of serum insulin or plasma glucose were shown. However, decreased BIGTT-S_i (P_additive = 0.004) and increased BIGTT-AIR (P_additive = 0.001) were observed in homozygous carriers of the FTO rs9939609 A-allele (Table 2).

In a study of 346 healthy young Danish whites, the FTO rs9939609 A-allele was associated with elevated BMI (P_additive = 0.002), body weight (P_additive = 0.008), fat mass (P_additive = 0.001), body fat percentage (P_additive = 3 × 10⁻⁴), and fasting serum leptin concentrations (P_additive = 0.003) but not with height or lean body mass. No association with birth weight, birth length, or the ponderal index at birth was observed (Table 3). To ensure the robustness of the quantitative trait analyses, we corrected for multiple testing using the Benjamini and Hochberg method.

The effect of the FTO rs9939609 genotype on BMI, body weight, and waist circumference in the population-based Inter99 study sample and the SDC type 2 diabetes study group, stratified according to glucose tolerance status, is shown in supplementary Fig. 1. We found no interaction between glucose tolerance status and the FTO rs9939609 genotype effect on BMI, body weight, or waist circumference (data not shown).

We found an interaction between the FTO rs9939609 genotype and self-reported physical activity on BMI levels in the population-based Inter99 study sample (P_interaction = 0.007). The FTO rs9939609 genotype effect on BMI for physically passive, light or medium physically active, and hard or very hard physically active subjects was 0.38, 0.37, and −0.11 kg/m², respectively, when comparing homozygous T-allele carriers and heterozygous carriers and 1.95, 0.69, and 0.47 kg/m², respectively, when comparing homozygous T-allele carriers and homozygous A-allele carriers (Fig. 1A).

Finally, we found an interaction between the FTO rs9939609 genotype and measures of insulin sensitivity. The FTO rs9939609 genotype effect on BMI in the highest, medium, and lowest insulin sensitivity groups, stratified by BIGTT-S_i tertiles, was −0.14, 0.31, and 0.54 kg/m², respectively, between homozygous T-allele carriers and heterozygous carriers, and 0.25, 0.60, and 1.34 kg/m², respectively, comparing homozygous T-allele carriers and homozygous A-allele carriers (P_interaction = 2 × 10⁻⁴) (Fig. 1B). Since the variance of BMI in the different physical activity and insulin sensitivity subgroups was substantial, we also performed a weighted analysis for the interactions. The subgroups were weighted by the reciprocal variance, and the interactions remained significant for both physical activity (P_interaction = 0.003) and insulin sensitivity (P_interaction = 0.03).

DISCUSSION

In the present study, we validated that FTO predisposes to type 2 diabetes. As previously observed (7), this association seems to be mediated by the effect of increased fat mass, since it abolishes when adjusting for BMI. The variant strongly associates with overweight and obesity and with quantitative traits such as BMI, body weight, and waist circumference. Homozygous carriers of the A-allele in the population-based Inter99 study sample weighed on average 3.3 kg (SE 2.1–4.6) more than noncarriers, which is reflected in a BMI increased by 1.1 kg/m² (0.7–1.4) and a waist circumference increased by 2.3 cm (1.3–3.3). The association with obesity-related measures was not affected by glucose tolerance status.

BMI is by definition influenced by measures of body weight and height, but in the present study, the FTO rs9939609 genotype only affected body weight. BMI is also influenced by lean body mass and fat mass, but in studies of 346 young healthy Danish whites, we only demonstrated an increase in fat mass. Analyses of BMI adjusted for waist circumference remained significant, whereas waist circumference adjusted for BMI did not (data not shown). This indicates that the observed increase in BMI is due to a global increase in fat mass rather than an intra-abdominal fat accumulation.

Finally, the FTO rs9939609 genotype was associated with increased fasting serum leptin levels, which are considered a result of increased adiposity. We observed an increase in body weight estimates at all ages except among newborns, which is in accordance with previous findings (7,8), suggesting that body fat accumulation takes place in early childhood.

In the study of 5,722 middle-aged individuals from the population-based Inter99 study sample, we found no differences in postoral glucose load measures of serum insulin or plasma glucose. However, whole-body insulin sensitivity, estimated by the BIGTT-S_i, was significantly decreased in homozygous carriers of the FTO rs9939609 risk A-allele. Furthermore, we found that the impact of the FTO rs9939609 genotype on BMI levels was highly influenced by insulin sensitivity. We only noticed a modest FTO rs9939609–induced increase in BMI levels among participants with a high insulin sensitivity index, whereas low insulin sensitivity index enhanced the genotype effect, particularly among homozygous A-allele carriers.

Thus, this is the first study implying interactions between the FTO rs9939609 genotype and insulin sensitivity. FTO is ubiquitously expressed, and numerous mechanisms leading to decreased insulin sensitivity exist. Since FTO is relatively abundantly expressed in muscle (7), it is feasible that the FTO rs9939609 genotype might affect insulin-mediated glucose uptake in muscle. Obviously, this hypothesis needs to be tested experimentally.

Interestingly, we showed that the impact of the FTO rs9939609 genotype is influenced by the habitual level of physical activity in the population-based Inter99 study sample. Physical inactivity was associated with a BMI increase of 1.95 ± 0.33 kg/m² in homozygous FTO rs9939609 A-allele carriers, whereas no major effect of sedentary lifestyle was found comparing noncarriers and those heterozygous for the FTO rs9939609 A-allele. Obviously, this finding needs replication in independent study populations to be used in a public health context. Also, since physical activity in our study has been assessed by questionnaire, it would be important to validate the finding with more direct measures of physical activity.

In conclusion, our study validates that variation in FTO associates with an overall increase in body fat accumulation as reflected by BMI, body weight, and waist circumference. Moreover, in middle-aged individuals, the FTO rs9939609 genotype may be associated with a decrease in estimates of whole-body insulin sensitivity, and in homozygous carriers of the FTO A-allele, physical inactivity associates with a relatively large increase in BMI compared with that in noncarriers and those heterozygous for the A-allele.