Calculation of polygenic risk scores for common traits and their interaction with exposure to maternal smoking in the HELIX project

Abstract

Introduction: Complex traits are determined by both genetic and environmental factors, as well as by their interactions (GxE). Maternal smoking during pregnancy is one of the most relevant exposures with both short- and long-term health effects in the mother and the new-born. The identification of genetic factors of complex traits has been transformed during the last decades thanks to genome-wide association studies (GWAS), which can be used to predict individuals’ disease risk through the calculation of polygenic risk scores (PRSs). A PRS is defined as a number that summarises the estimated effect of many genetic variants on an individual’s phenotype. Objective: In this study we aimed, first, to compute and validate PRSs for common traits for children of the Human Early Life Exposome (HELIX) project, and second, to test the interaction between maternal smoking during pregnancy and these PRSs (GxE). Methods: PRSs for a total of 42 common traits were calculated for 1155 children of European ancestry from the HELIX project using the PRSice-v2 tool accounting for linkage disequilibrium. 10 PRSs with a different number of single nucleotide polymorphisms (SNPs) were calculated. For 23 traits measured in HELIX, we tested their association with the PRS, also using PRSice-v2. The PRS with the highest R2 fit of the model was considered the best PRS and followed in the GxE analyses. The association between maternal smoking during pregnancy and 10 selected traits was tested using linear regression models adjusted for covariates. GxE interactions between maternal smoking during pregnancy and the PRSs were tested by including and interaction term in the models. Results: Out of the 23 PRSs tested for validation, 17 showed significant relationships with the phenotypic trait. However, the proportion of the variance explained by the PRS was very small (maximum 4.9%). In line with other studies, mothers that smoked during pregnancy had new-borns with lower birth weight, higher body mass index, and higher waist circumference. Effects were stronger in offspring of sustained smoker mothers than in non-sustained smokers. We did not detect any significant GxE interaction between maternal smoking in pregnancy and the PRSs, likely due to the limited sample size and variance explained by the PRSs. However, a marginal trend was found for birth weight, where children within the third tertile of the birth weight PRS were protected against the adverse effects of sustained smoking during pregnancy in comparison to children in the other two tertiles. Conclusion: A total of 42 PRSs were calculated for HELIX children, which 17 of them were associated with their phenotypic trait explaining a small proportion of it. We could not find any interaction between maternal smoking and the PRSs, except for a marginal trend for birth weight.