Gene-obesogenic environment interactions in the UK Biobank study

Abstract

Background: Previous studies have suggested that modern obesogenic environments accentuate the genetic risk of obesity. However, these studies have proven controversial as to which, if any, measures of the environment accentuate genetic susceptibility to high body mass index (BMI).

Methods: We used up to 120 000 adults from the UK Biobank study to test the hypothesis that high-risk obesogenic environments and behaviours accentuate genetic susceptibility to obesity. We used BMI as the outcome and a 69-variant genetic risk score (GRS) for obesity and 12 measures of the obesogenic environment as exposures. These measures included Townsend deprivation index (TDI) as a measure of socio-economic position, TV watching, a 'Westernized' diet and physical activity. We performed several negative control tests, including randomly selecting groups of different average BMIs, using a simulated environment and including sun-protection use as an environment.

Results: We found gene-environment interactions with TDI (Pinteraction = 3 × 10 -10 ), self-reported TV watching (Pinteraction = 7 × 10 -5 ) and self-reported physical activity (Pinteraction = 5 × 10 -6 ). Within the group of 50% living in the most relatively deprived situations, carrying 10 additional BMI-raising alleles was associated with approximately 3.8 kg extra weight in someone 1.73 m tall. In contrast, within the group of 50% living in the least deprivation, carrying 10 additional BMI-raising alleles was associated with approximately 2.9 kg extra weight. The interactions were weaker, but present, with the negative controls, including sun-protection use, indicating that residual confounding is likely.

Conclusions: Our findings suggest that the obesogenic environment accentuates the risk of obesity in genetically susceptible adults. Of the factors we tested, relative social deprivation best captures the aspects of the obesogenic environment responsible.

Keywords: UK Biobank; body mass index; gene–environment; obesogenic environment; social deprivation.

Figure 1.

Forest plot demonstrating the change in BMI per-allele increase in BMI genetic risk score (GRS) for the 12 different obesogenic environments and the negative control on a standardized inverse normalized scale. BMI was corrected for age, sex, ancestry principal components and assessment centre location prior to calculating residuals. The analyses were further adjusted for genotype platform.

Figure 2.

Association between the BMI GRS (by decile) and BMI in (a) the most socially deprived (black circles) and least socially deprived (white circles); (b) high and low self-reported physical activity, (c) high and low TV watching and (d) high and low composite score, (e) high and low use of sun protection in the summer, (f) individuals randomly selected to be of high BMI (black circles) and individuals randomly selected to be of low BMI (white circles) and (g) individuals in the high obesogenic simulated environment (black circles) and individuals in the low obesogenic simulated environment (white circles). Note that, for the simulated environment, we used the median BMI GRS BMI association after 1000 simulations. For (f), it was not possible to use a continuous measure in the calculation of the interaction term. This figure is based on a similar way of showing interaction data with a BMI GRS from . SEP, socioeconomic position.

Figure 3.

Histograms showing the -log10(P-values) for the interactions from (a) the 100 iterations of the individuals selected to be of different BMIs at random and (b) the 10 000 iterations of a simulated environment with a similar association to BMI as TDI. The dashed line represents the median value and the solid line represents the P-value obtained from the real interactions with TDI.