Polygenic Risk Scores and Physical Activity

Abstract

Purpose Polygenic risk scores (PRS) summarize genome-wide genotype data into a single variable that produces an individual-level risk score for genetic liability. PRSs have been used for prediction of chronic diseases and some risk factors. As PRSs have been studied less for physical activity (PA), we constructed PRSs for PA and studied how much variation in PA can be explained by these PRSs in independent population samples. Methods We calculated PRSs for self-reported and objectively measured PA using UK Biobank genome-wide association study summary statistics, and analyzed how much of the variation in self-reported (MET-hours/day) and measured (steps and moderate-to-vigorous PA minutes/day) PA could be accounted for by the PRSs in the Finnish Twin Cohorts (FTC, N = 759-11,528) and the Northern Finland Birth Cohort 1966 (NFBC1966, N = 3,263-4,061). Objective measurement of PA was done with wrist-worn accelerometer in UK Biobank and NFBC1966 studies, and with hip-worn accelerometer in the FTC. Results The PRSs accounted from 0.07% to 1.44% of the variation (R2) in the self-reported and objectively measured PA volumes (P-value range 0.023 to < 0.0001) in FTC and NFBC1966. For both self-reported and objectively measured PA, individuals in the highest PRS deciles had significantly (11 to 28%) higher PA volumes compared to the lowest PRS deciles (P-value range 0.017 to < 0.0001). Conclusions PA is a multifactorial phenotype and the PRSs constructed based on UK Biobank results accounted for statistically significant but overall small proportion of the variation in PA in the Finnish cohorts. Using identical methods to assess PA and including less common and rare variants in the construction of PRSs may increase the proportion of PA explained by the PRSs.