A high intake of saturated fatty acids strengthens the association between the fat mass and obesity-associated gene and BMI

Abstract

Evidence that physical activity (PA) modulates the association between the fat mass and obesity-associated gene (FTO) and BMI is emerging; however, information about dietary factors modulating this association is scarce. We investigated whether fat and carbohydrate intake modified the association of FTO gene variation with BMI in two populations, including participants in the Genetics of Lipid Lowering Drugs and Diet Network (GOLDN) study (n = 1069) and in the Boston Puerto Rican Health (BPRHS) study (n = 1094). We assessed energy, nutrient intake, and PA using validated questionnaires. Genetic variability at the FTO locus was characterized by polymorphisms rs9939609 (in the GOLDN) and rs1121980 (in the GOLDN and BPRHS). We found significant interactions between PA and FTO on BMI in the GOLDN but not in the BPRHS. We found a significant interaction between SFA intake and FTO on BMI, which was stronger than that of total fat and was present in both populations (P-interaction = 0.007 in the GOLDN and P-interaction = 0.014 in BPRHS for categorical; and P-interaction = 0.028 in the GOLDN and P-interaction = 0.041 in BPRHS for continuous SFA). Thus, homozygous participants for the FTO-risk allele had a higher mean BMI than the other genotypes only when they had a high-SFA intake (above the population mean: 29.7 ± 0.7 vs. 28.1 ± 0.5 kg/m²; P = 0.037 in the GOLDN and 33.6. ± 0.8 vs. 31.2 ± 0.4 kg/m²; P = 0.006 in BPRHS). No associations with BMI were found at lower SFA intakes. We found no significant interactions with carbohydrate intake. In conclusion, SFA intake modulates the association between FTO and BMI in American populations.

FIGURE 1

BMI in participants in the GOLDN study (A) and the BPRHS (B) depending on the SFA intake (2 levels, based on the population mean: 27.6 g/d in GOLDN and 22.7 g/d in BPRHS) and the FTO polymorphisms (recessive model; rs9939609 in GOLDN and rs1121980 in BPRHS). Values are adjusted means ± SEM. Models were adjusted for gender, age, tobacco smoking, alcohol drinking, PA, and total energy intake. P values for mean comparison in each saturated fat strata were also adjusted for covariates. *Different from CC+TT, P < 0.05. BPRHS, Boston Puerto Rican Health Study; GOLDN, Genetics of Lipid Lowering Drugs and Diet Network; PA, physical activity.

FIGURE 2

Predicted values of BMI by the FTO polymorphisms (recessive model) in the GOLDN study (A) and the BPRHS (B) plotted against the SFA intake (n = 881 TT+TA and n = 188 AA in GOLDN and n = 917 CC+CT and n = 177 TT in BPRHS). Predicted values were calculated from the regression models containing the SFA intake (as continuous), the FTO polymorphism, their interaction term, and the potential confounders (gender, age, smoking, drinking, PA, and total energy intake). The P value for the interaction term between SFA intake and the corresponding FTO polymorphism (rs9939609 in GOLDN and rs1121980 in BPRHS participants) was obtained in the hierarchical multivariate adjusted interaction model in which SFA intake was logarithmically transformed. BPRHS, Boston Puerto Rican Health Study; GOLDN, Genetics of Lipid Lowering Drugs and Diet Network; PA, physical activity.