In this unique study of a large and well standardized European cohort of children aged 2–9 years, we confirmed that the A allele of FTO rs9939609 polymorphism was associated with higher BMI and central body fat distribution and with a greater prevalence of overweight/obesity. Moreover, we showed that this variant was related to a greater increase in body fat and adiposity indices over the two-year follow-up. We did not observe any significant interaction between the FTO gene and the community-oriented intervention program implemented as part of the IDEFICS project. Finally, we reported for the first time that the A allele of FTO rs9939609 was associated to an increased incidence of overweight/obesity in children.
The cross-sectional analysis of the IDEFICS cohort replicated, in a large and well-characterized population sample of children of European origin, the association of this locus with childhood obesity. The magnitude of the association and the risk for overweight/obesity conferred by FTO
rs9939609 was comparable with that previously reported in other association studies in children 
, thus confirming this as a childhood obesity susceptibility gene. Of interest, the magnitude of the association is also similar to that observed in adult studies 
, suggesting a relatively stable gene-phenotype association over time for this variant, that may be considered a marker of early onset obesity.
The present study confirms the now established obesity risk A allele, but, more importantly, adds the novel findings obtained by the longitudinal observation of the IDEFICS cohort. In particular, the carriers of the A allele experienced a greater body mass increase and possessed a higher risk of developing overweight/obesity during the period of observation. This finding is intriguing given the relatively short observation period (two years). Other longitudinal studies that explored the genetic influences of FTO
on the variability of body mass over time have reported heterogeneous findings 
. However, most of these studies were retrospective analyses of already established cohorts, with large variation in sample size, often relying on self-reported heights and weights and with vastly different study design. In the present study, the genotyping strategy was part of the study protocol and was agreed before the study commenced, thus allowing the extraction from the whole IDEFICS population of a country-, sex- and age-balanced random sample of children, whose size was a priori
defined in order to have enough power to detect biologically and clinically relevant differences in the main outcomes.
An additional strength of the present study is the use of precisely standardized phenotypic measurements within the eight European countries participating in the survey. In fact, all measurements were conducted according to detailed standard operation procedures. In particular, subsamples of study subjects were examined repeatedly to calculate the inter- and intra-observer reliability of anthropometric measurements 
We also analyzed the effect of the genetic variation in FTO
on body mass change after the lifestyle intervention. In theory, a lifestyle intervention designed for the primary prevention of obesity might attenuate the risk conferred by the genetic background. Based on preliminary analyses of the effects of the IDEFICS intervention, 16% of the overweight children became normal weight in the intervention region while only 12% in the control region after the two-year intervention. In both regions 9% of the non-overweight became overweight (unpublished data). At the individual level, a significant intervention effect was only observed when looking at BMI in girls (unpublished data). We did not found any interaction between FTO
and the lifestyle intervention, with the carriers of the risk allele experiencing a significantly greater increase in body mass over time, independently of the study group (control versus intervention). Few studies on relatively small samples previously explored the association between this genotype and the effect of lifestyle intervention programs both in adults 
and in children 
, in general with negative results. Our study on a large and well-characterized sample of European children was in principle powered enough to detect the possible differential effect of the intervention according to the genotype. However, due to the relatively short duration of the intervention and its apparently small effect, we cannot exclude that the early genotyping of FTO
may be effective in identifying individuals genetically predisposed to obesity that are going to benefit more from a lifestyle intervention than less genetically susceptible individuals. In fact, we cannot rule out the possibility that the mechanism by which FTO
modulates body mass over time needs a longer period to elicit a differential response to an intervention program.
Despite the many strengths, there are a number of limitations. First, we cannot rule out the possibility that our results may be marginally influenced by population stratification, since no neutral markers were genotyped. Second, the study participants were not nationally representative of the eight European countries involved in the IDEFICS study, as this was never the study objective. We randomly extracted from the whole IDEFICS cohort an age-, sex- and country-matched sample of white European children. In our analysis, however, all data were country-adjusted. Indeed, the adjustment for the eight geographic regions influences the strength of the FTO
association with BMI only in a minor way (adjusted: p
0.0002; unadjusted p
0.0008). Third, although we did not find any interaction between FTO
and the intervention, the possibility that the intervention program conducted in half of the population could have marginally affected the relationship between FTO
and the phenotypes of interest over the follow-up cannot be excluded. This possibility was taken into account by adjusting for the control/intervention group in the longitudinal analysis. Once again, this adjustment did not influence the strength of the association between FTO
and the variation of BMI over the two-year follow-up (adjusted: p
0.0002; unadjusted p
0.0002). Finally, the duration of the follow-up was only two year and a 28% loss at follow-up was observed. However, the longitudinal association of FTO
rs9939606 with adiposity indices clearly emerged even after the relatively short follow-up period and notwithstanding the expected reduction of the sample size, mainly due to the burden of the extensive examination protocol. The planned follow-up of the IDEFICS cohort in subsequent years will allow further evaluation of the association of the FTO
locus with body mass during the transition from childhood to adolescence.
In summary, we confirmed associations between the FTO rs9939609 and higher body mass and overweight/obesity risk in European children. The main finding of the study is that the children with the A risk allele of this variant had a greater increase in body mass and central adiposity over time and therefore were at higher risk of developing overweight/obesity during growth, independently from intervention measures.