Polygenic risk analysis in physical activity and health: why are the same results interpreted differently?

Abstract

Polygenic risk scores (PRSs), designed to capture genetic predisposition to specific traits,1 are becoming increasingly accessible at scale and are being used in physical activity research. PRSs are typically calculated by aggregating the effect sizes of single-nucleotide polymorphisms (SNPs) associated with a particular trait or disease, usually derived from genome-wide association studies (GWASs), into a score for each individual to reflect their genetic liability to that trait or disease. Various methodologies exist for constructing PRSs, ranging from simple to more complex.1 The simplest approach often involves selecting a subset of SNPs based on their associated p values from the GWAS, while more sophisticated methods may incorporate additional data, such as linkage disequilibrium patterns or functional genomic information.1 Regardless of the approach, the appeal of PRSs lies in their simplicity and versatility, especially as they can be readily computed in cohorts with existing genetic data. This makes PRSs valuable both for controlling confounding and as a research focus in their own right. This editorial aims to discuss some key challenges in using PRSs for analysing physical activity and health, focusing on the difficulty of distinguishing mechanisms behind associations and the limited clinical interpretability of effect estimates. We also offer some practical recommendations for future research. Recent studies have employed PRSs related to physical activity to investigate various outcomes, finding that high scores are associated with lower risk of cardiometabolic risk factors, coronary heart disease, stroke, hypertension, type 2 diabetes, obesity and all-cause mortality.2–4 However, the interpretation of these results is hampered by several challenges. A fundamental challenge is that several potential explanations exist for why a PRS for physical activity might be associated with health outcomes. These include a true causal effect of physical activity on health outcomes and the influence of shared genetics. Unfortunately, …