Genetic ancestry influences gene-environment interactions with sociocultural factors: Results from the Hispanic Community Health Study/Study of Latinos

Abstract

Often, studies will aggregate all participants identified as Hispanic/Latino, despite genetic and environmental substructures, preventing the meaningful interrogation of the roles of genetics and environment in human health. Using the Hispanic Community Health Study/Study of Latinos (HCHS/SOL), we examined how self-identified background group and genetic ancestry influence gene-environment interactions between body mass index (BMI) and a polygenic score for BMI (PGS_BMI). Participants (n = 7,075) identified with six background groups: Central American, Cuban, Dominican, Mexican, Puerto Rican, and South American. Generalized linear models incorporating complex survey weighting were used to model BMI through joint and stratified (background group, estimated Amerindigenous [AME] ancestry) analyses including PGS_BMI and other health-related variables. Interaction effects were modeled between PGS_BMI and diet and age at immigration. Comparing pooled to background group-stratified analyses, we observe heterogeneous distributions of environmental and sociocultural variables, as well as differing associations with AME ancestry. Within the multivariate model, PGS_BMI performance decreased with increasing AME ancestry. After stratification, PGS-age-at-immigration interactions remained statistically significant in some strata: Mexican background individuals born in the US (50 states/DC) (β = 1.33, p < 0.01), Dominican background individuals 6-12 years old (β = 4.38, p < 0.001), and Cuban background individuals 0-5 years old (β = 2.20, p = 0.015) relative to those ≥ 21 years old at migration. It is vital to understand populations of interest to model them appropriately and prevent possible confounding or misinterpretation. While this work focuses specifically on Hispanic/Latino groups, these lessons are relevant to other groups as we diversify work to better understand gene-environment interactions.

Keywords: cardiometabolic; confounding; gene-environment interaction; genetic ancestry; polygenic score.

Figure 1

Conceptual framework of full analytic model

Figure 2

Distributions of selected variables by AME ancestry proportion aggregated and by background Graphical depictions of relationships between AME ancestry proportion and (A) BMI, (B) PGS_BMI, (C) immigrant generation, (D) diet score (1 = under 60th sex-specific percentile of JAMA healthy diet score, 2 = top 40th sex-specific percentile of JAMA healthy diet score), and (E) prevalent cardiovascular disease in the complete sample (left; 1 = yes, 0 = no) and stratified by self-identified background group identity (right). ∗ indicates statistically significant differences (p < 0.05) between categories.