Polygenic risk scores (PRS) are valuable for stratifying genetic susceptibility, yet there is often a frequent mismatch between inherited predisposition and observed clinical outcomes. This project will investigate the phenomena of “resilience” (high-PRS individuals remaining disease-free) and “discordance” (low-PRS individuals developing disease). Our objective is to identify protective and risk-enhancing factors by applying a harmonised analytical framework across a wide spectrum of chronic diseases. We will begin by targeting key domains including, but not limited to, cardiometabolic conditions (e.g., type 2 diabetes, cardiovascular disease), respiratory disease (e.g., asthma), and selected cancers. We hypothesise that modifiable factors-including lifestyle behaviours, clinical markers, and molecular profiles-can buffer genetic risk on an additive scale, thereby informing prevention strategies.
For progressive conditions such as diabetic complications or advancing chronic kidney disease, landmark analyses will be applied to assess how exposures measured prior to defined time points shape short-term progression risk. This secondary analysis of UK Biobank will integrate PRS with deep phenotyping from accelerometry, primary care records, NMR metabolomics, and Olink proteomics, linked to outcomes from algorithmically defined fields and national registries.
Analyses will adopt prospective modelling with Cox and competing-risks regression, quantify additive interactions (e.g., RERI, AP), and apply false discovery rate control for multiple testing. To complement these conventional approaches, we will also explore machine learning methods to assist in identifying key multidimensional features.
By systematically characterising resilience and discordance within each disease, this research will advance understanding of the gene-environment interplay, refine communication of genetic risk, and provide evidence to guide targeted prevention and early intervention.
