Improving prediction of atrial fibrillation: the impact of polygenic risk scores over conventional risk factors amongst 270,000 individuals in UK Biobank

Abstract

Funding Acknowledgement: Type of funding source: Foundation. Main funding source(s): British Heart Foundation Background: Previous studies have suggested only modest benefits of adding genetic information to conventional risk factors for prediction of atrial fibrillation (AF). However, these studies have been based on limited numbers of AF cases and pre-date recent AF genetic discoveries. Purpose: To examine the independent relevance of common genetic risk factors over and above established non-genetic risk factors for predicting AF amongst 270,000 participants from UK Biobank, and to determine potential clinical utility. Methods: UK Biobank (UKB) is a large prospective study of over 500,000 British individuals aged 40 to 69 years at recruitment. Incident AF was ascertained using hospital episode statistics and death registry data. The CHARGE-AF score, which combines the relevance of age, height, weight, blood pressure, use of antihypertensives, diabetes, heart failure, and my-ocardial infarction (MI) was used to estimate 5-year risk of AF at baseline. A polygenic risk score (PRS) was constructed based on 142 independent variants previously associated with AF in a genome-wide meta-analysis of 60,620 AF cases from the AFGen Consortium, weighted by their published effect sizes. A total of 270,254 individuals were analysed after exclusions for genetic QC, non-White British ancestry, and prevalent AF. Cox proportional hazard models were used to estimate associations between risk scores (based on standard deviation [SD] units) and incident AF. Standard methods were used to assess predictive value.