Genetic sum score of risk alleles associated with Body-Mass-Index interacts with socioeconomic status in the Heinz Nixdorf Recall Study
Past research has shown that obesity is influenced by both, environmental factors and genetic variants. Twin studies suggested that the impact of some genetic variants depends on certain environmental exposures such as the socioeconomic status (SES), indicating gene by SES interaction. Empirical studies examining GxSES on body mass index (BMI), used as a surrogate marker for the risk of obesity, incorporating genome-wide molecular genetic information are, however, limited and do not include underlying risk factors potentially responsible for the interaction effects detected. This thesis aimed to investigate whether a sum score (GRS) of 97 BMI-associated variants interacts with income and education, used as contextual markers of socially differing environmental conditions. Within the study population inequalities in health were seen for both SES indicators on BMI and an increase of BMI per additional risk allele was observed. Negative GRSxSES interaction was found, showing stronger genetic effects in groups of lower SES. To explore whether these interactions were due to underlying interactions with SES-related health behaviours, smoking, physical activity and alcohol consumption were included into analyses, not affecting the magnitude and direction of the observed GRSxSES interaction. Independent of the SES effect, a positive GRSxPA interaction was found. Further, income did not have a confounding effect on the GRSxPA, GRSxS and GRSxAlc interactions, while GRSxPA and GRSxAlc interactions might be party confounded by education. Sex-stratified analyses showed overall slightly stronger effect size estimates in women than in men. Interactions between individual SNPs and both SES indicators revealed effect size estimates close to each other, indicating that no single variant alone substantially triggered the observed GRSxSES interactions. Also, overlaps among the highest SNPxSES interaction effects were found to be more frequent than expected, suggesting that the GRSxEducation and GRSxIncome interactions are at least partly mediated by the same genes. The fact that overlaps were also seen with health behaviors could suggest that for some individual SNPs the GRSxSES interaction is mediated by health behaviors. Present results support the hypothesis that individuals in higher SES groups are enabled to reduce their genetic susceptibility to disease through better material, behavioral and psychosocial factors. Although the found GRSxSES interaction seems not to be explained by physical activity, smoking or alcohol consumption, further research must consider other factors that might modify the effect, e.g. dietary patterns or levels of stress. Further analyses could also be extended by other examples of complex conditions such as diabetes or coronary artery diseases.