Gene-environment dependencies lead to collider bias in models with polygenic scores

Abstract

The application of polygenic scores has transformed our ability to investigate whether and how genetic and environmental factors jointly contribute to the variation of complex traits. Modelling the complex interplay between genes and environment, however, raises serious methodological challenges. Here we illustrate the largely unrecognised impact of gene-environment dependencies on the identification of the effects of genes and their variation across environments. We show that controlling for heritable covariates in regression models that include polygenic scores as independent variables introduces endogenous selection bias when one or more of these covariates depends on unmeasured factors that also affect the outcome. This results in the problem of conditioning on a collider, which in turn leads to spurious associations and effect sizes. Using graphical and simulation methods we demonstrate that the degree of bias depends on the strength of the gene-covariate correlation and of hidden heterogeneity linking covariates with outcomes, regardless of whether the main analytic focus is mediation, confounding, or gene × covariate (commonly gene × environment) interactions. We offer potential solutions, highlighting the importance of causal inference. We also urge further caution when fitting and interpreting models with polygenic scores and non-exogenous environments or phenotypes and demonstrate how spurious associations are likely to arise, advancing our understanding of such results.

Gene-Environment Pathways to Cognitive Intelligence and Psychotic-Like Experiences in Children

“…In linear regression, unobserved confounding variables may bias estimates, especially when those unobserved confounders and genetic factors are jointly associated with environmental factors and target traits 71 . To overcome this limitation, we used a novel framework for causal inference based on probabilistic modeling, called the deconfounder 41, 72 .…”

Childhood Socioeconomic Status and Polygenic Scores for Cognition Have Independent Associations with Cognitive Performance During Middle Childhood

“…Fourth , the inclusion of PGS and heritable environmental factors (e.g., SES) in the same regression can create spurious associations. 41 In that case, the effect of the environment should increase, while the effect of genetics should decrease. We did not find this to be the case in our models, suggesting that the associations we observed are not spurious.…”

“…We did not find this to be the case in our models, suggesting that the associations we observed are not spurious. Relatedly, prior work shows that gene x environment interactions may be inflated in the case of gene-environment correlations (i.e., between Cognition PGS and SES [41][42][43] ). Because we did not detect any interaction between PGS and SES in our models, we expect little inflation of these estimates.…”

Nonlinear effects of environment on childhood asthma susceptibility

“…This may merely be a product of the inherent complexity of PRS-environment effects being explored or the need to account for additional complex influences such as collider bias. 11 The conflicting findings in MAAS in comparison to ALSPAC and IOWBC may stem from antenatal parental smoking rather than parental smoking within the first year of the child's life (as in the IOWBC and ALSPAC) being evaluated in MAAS. However, this may also be explained by inherent cohort differences; MAAS conducted selective recruitment from specific regions of South Manchester and Cheshire (see Supplementary Methods) whilst the IOWBC and ALSPAC are unselected cohorts representative of the general population.…”