In a recently accepted article in PNAS, entitled "Cohort of birth modifies the association between FTO genotype and BMI," the association of FTO variant rs993609 with body mass index is described as having essentially zero influence for study participants born before 1942 and increasing influence on this obesity phenotype as participants were born in increasingly more recent years. That long-range enhancers within the FTO region recapitulate aspects of IRX3 expression implies that the obesity-associated interval serves to regulate IRX3. Consistent with this, obesity-associated SNPs are associated with expression of IRX3, but not FTO, in human brains. Nonetheless, this is an important obesity locus, be it FTO or IRX3 as the functional unit.
The authors rightly suggest that gene-environment interactions (GxEs) coupled with changes to the environment of the participants could alter the FTO-BMI association.
FTO is subject to exercise-induced changes in DNA methylation. See, for example, table 5 (and reference 3) of Rönn, Volkov, et al. We have cataloged a large number of genetic variants that show the type of GxEs suggested by the recent PNAS article. That catalog shows that some nine different studies observed modulating effects of physical activity on the FTO-BMI association. (In most populations of European ancestry, for example, in which many of these studies were conducted, the variants analyzed are in relatively strong to very strong linkage disequilibrium.) Other lifestyle choices also modulated the effects of FTO variants, including macronutrient intakes of carbohydrate, and fatty acids such as saturated fat, MUFA (mono-unsaturated fatty acid) and PUFAs (poly-unsaturated fatty acids). Whether time spent engaged in physical activity shrank as the birth cohorts became more recent, or diet changed, or some combination of this, is difficult to ascertain. But a list of known FTO-BMI GxEs would be a good place to begin such an analysis.