Adiposity
In the past year, the list of obesity susceptibility loci grew only slightly, with one large-scale GWAS for body fat percentage identifying two new loci, near the insulin receptor substrate 1 (IRS1) and the sprouty homolog 2 (SPRY2) genes, and confirming the fat mass and obesity associated (FTO) locus. As such, the total number of obesity susceptibility loci stands currently at 52, of which 32 were primarily identified for BMI, 2 were identified for body fat percentage, 14 were identified for waist-to-hip ratio (adjusted for BMI), and 4 were identified for extreme and early-onset obesity. The two-stage design (i.e., discovery stage, followed by replication stage), the stringent significance threshold (P < 5 × 10), and the large scale that characterizes GWAS ensure that these 52 loci are unequivocally established. Despite these convincing observations, few loci have been studied in more in-depth functional experiments or detailed association analyses to gain greater insight into the mechanisms by which they contribute to obesity risk.
Because of the ever-growing scale of GWAS (e.g., the discovery stage of the latest meta-analysis for BMI included 123,865 individuals), the follow-up of these loci in epidemiological studies has become a challenging undertaking. To have sufficient statistical power, follow-up studies require sample sizes of a magnitude similar to those of the original GWAS, particularly if the outcomes of interest are impeded by inaccurate measurements. As a consequence, few studies have been able to consistently replicate the associations identified in the GWAS that counted more than 30,000 individuals, and so far, no studies have been able to show evidence that any of the obesity susceptibility loci influence body weight through association with levels of physical activity or energy expenditure.
The locus most extensively studied continues to be FTO, which was the first obesity susceptibility locus to be identified in 2007 by two independent studies with a discovery stage of around 4000 individuals each. The relatively small sample size needed for FTO's discovery, coupled with the fact that its BMI-increasing allele is common and has the largest effect among all established obesity susceptibility loci, has greatly facilitated its replication. Variants in the first intron of FTO have been consistently shown to associate with a variety of obesity-related traits across all age groups and across various ethnic groups. Experimental research has suggested that FTO might contribute to obesity risk through peripheral mechanisms influencing energy expenditure, as well as through CNS influencing the regulation of food intake, with recent data from humans and animal models providing more evidence for the latter.
Of particular interest are the findings of a recent largescale epidemiological study that examined the role of daily physical activity in the association between FTO and obesity risk. Since the discovery of the FTO locus, a growing number of studies have reported that physical activity attenuates the association between variants in the FTO locus and obesity-related traits. However, doubts arose with the publication of studies that could not confirm such an interaction between FTO and physical activity on obesity risk. To firmly confirm or refute this observation, Kilpeläinen et al. conducted a meta-analysis based on 45 studies in adults (n = 218,166), as well as nine studies in children and adolescents (n = 19,268). Although appropriate for the study of main-effect associations, literature-based meta-analyses are not an option for interaction analyses because definitions of physical activity, statistical analyses of interaction, and reported interaction results typically differ widely across studies. Furthermore, interaction analyses are often secondary to main-effect analyses and are typically reported only when results are statistically significant, such that this literature is prone to publication bias. Therefore, in this recent study, all groups with genotype data for the FTO locus as well as with data on physical activity and obesity-related outcomes were invited to participate in a de novo meta-analysis. More specifically, because physical activity was assessed in different ways, it was first standardized centrally and categorized as a binary trait (physically active vs inactive) for each participating study. Subsequently, the data of each study were analyzed according to a standardized plan to achieve the greatest consistency possible across studies and to facilitate the final meta-analysis.
Consistent with early observations, the results in adults confirmed that a physically active lifestyle significantly attenuates the effect of FTO variation on the risk of obesity. Each additional FTO risk allele increases the odds of obesity by 23% (P = 7 × 10) and the BMI by 0.36 kg·m (equivalent to 1040 g for a person 1.70 m tall) (P = 2 × 10) in the total adult population. However, the effect of FTO on obesity risk was significantly (Pinteraction = 0.001) reduced by physical activity; i.e., each FTO risk allele increases the odds of obesity by 30% in the inactive individuals but only by 22% in the active individuals (Fig. 1). Consistent with the observations for obesity risk, the BMI-increasing effect of FTO was reduced by 30% (P = 0.005) in the physically active individuals (0.32 kg·m per allele or 925 g for a 1.70-m-tall person) compared with the inactive individuals (0.46 kg·m per allele or 1330 g for a 1.70-m-tall person) (Fig. 1). Similar interaction effects were observed for waist circumference, body fat percentage, and risk of being overweight.
(Enlarge Image)
Figure 1.
—Effect of the rs9939609 variant or a proxy (r > 0.8) on risk of obesity (Pinteraction = 0.001) (A) and on BMI (Pinteraction = 0.005) (B) by activity status in a random-effects meta-analysis (27). Drawn from the data reported in PLoS Medicine.
Most notable was the observation that physical activity was three times more effective (P = 0.001) in attenuating the BMI-increasing effects of FTO in North Americans than in Europeans, with similar trends for other obesity-related traits. In contrast to the adult data, physical activity had no effect on the association between FTO variants and obesity-related traits in children and adolescents.
These findings emphasize the importance of physical activity in body weight regulation in adults, in particular in those who are genetically predisposed to obesity, and they oppose the often-held fatalist view that a genetic susceptibility is nonmodifiable. The reason for the difference in effectiveness between continents is unclear, but the often-reported lower physical activity levels and higher obesity prevalence in North Americans as compared with Europeans may play a role. Another diverging factor may be the differences in the measurement of physical activity, with more often continuous data recorded in North America and more categorical data in Europe. It remains unclear what biological mechanisms are behind the observed interaction between physical activity and FTO and whether the effect attenuation is observed only with physical activity or also with other lifestyle factors. Variants in the first intron of FTO have been shown to be associated with methylation capability, such that some have speculated that this region might be sensitive to epigenetic effects. Of interest are recent studies suggesting that not only physical activity but also dietary habits and energy intake might attenuate the effects of FTO on obesity susceptibility.
An important secondary observation from this large-scale interaction meta-analysis is the fact that variants in FTO are not associated with physical activity per se in adults (P = 0.20) or in children (P = 0.6), thus providing convincing evidence that physical activity is not the mediating factor in the association between FTO and obesity risk.