Combining Genetic Proxies of Drug Targets and Time-to-event analyses From Longitudinal Observational Data To Identify Target Patient Populations

Abstract

BackgroundHuman genetics is an important tool for identifying genes as potential drug targets, and the extensive genetic study of cardiovascular disease provides an opportunity to leverage genetics to match specific patient populations to specific drug targets to improve prioritization of patient selection for clinical studies.MethodsWe selected well described genetic variants in the region ofPCSK9(rs11591147 and rs562556),ADRB1(rs7076938),ACE(rs4968782 and rs4363) andBAG3(rs2234962) for use as proxies for the effects of drugs. Time-to-event analyses were utilized to evaluate their effects on atrial fibrillation (AF) and heart failure (HF) death and/or re-hospitalization using real-world longitudinal dataset. To mitigate the effect of confounding factors for cardiovascular (CV) outcomes, we employed propensity score matching.ResultsAfter matching, a genetic proxy for PCSK9 inhibition (rs11591147) improved survival from CV death/heart transplant in individuals following a diagnosis of ischemic heart disease (Hazard Ratio (HR) 0.78,P=0.04). A genetic proxy for beta-blockade (rs7076938) improved freedom from rehospitalization or death in individuals with AF (HR 0.92,P=0.001), and a genetic proxy of ACE inhibition (rs7076938) improved freedom from death or rehospitalization for HF (HR 0.84,P=0.017) and AF (HR 0.85,P=0.0014). A protective variant in BAG3 (rs2234962) showed decreased risk of HF rehospitalization or CV death/heart transplant composite outcome within 10 years in HF patients (HR=0.96,P=0.033). Notably, despite smaller cohort sizes after matching, we often observed numerically smaller HRs and reduced P, indicating more pronounced effects and increased statistical association. However, not all genetic proxies replicated known treatment effects.ConclusionsGenetic proxies for well-known drugs corroborate findings from clinical trials in cardiovascular disease. Our results may demonstrate a novel analytical approach that leverages genetic evidence from a large cohort to effectively select patient populations where specific drug targets may be most effective.