Abstract 12228: Development of a Novel Polygenic Risk Score to Predict Risk of Aortic Stenosis Events

Abstract

Introduction: Aortic stenosis (AS) is the most common valvular heart disease. Polygenic risk scores (PRS) use common genetic variants to enhance risk prediction for disease but have not yet been applied in AS. Methods: We generated an AS genetic risk score using summary data from a recently performed GWAS in the Million Veteran Program and score selection performed in the UK Biobank. We analyzed investigator reported adverse events with either newly identified AS or worsening of AS (including heart failure in the setting of AS or aortic valve replacement) from 5 TIMI trials (FOURIER, PEGASUS, SAVOR, SOLID, and ENGAGE). Cox proportional hazard models (adjusted for age, sex, diabetes, BMI, hyperlipidemia, and the first 5 principal components) were used to generate adjusted hazard ratios for risk of AS evaluating normalized PRS as a continuous variable. Results: The best performing AS PRS used a p-value threshold of 5x10 -8 and consisted of 27 independent single nucleotide polymorphisms which together predicted AS in the UKBB with a p-value of 2.6x10 -5 . In the TIMI trials, there were a total of 52,183 individuals (median age 66 years [interquartile range 60-72], 71.7% male) with a median follow up of 2.4 years. Of these, 215 (0.4%) had an AS-related adverse event. Age was the strongest clinical risk factor for an AS-related adverse event (HR 1.08 [1.06-1.10] per 1-SD, p-value 1.8x10 -19 ), followed by diabetes and BMI. After adjustment for all clinical risk factors there was a 16% increase in the risk of an AS-related adverse event with each standard deviation (SD) increase in the AS PRS (HR 1.16 [1.02-1.31] per 1-SD, p-value 0.03). Conclusions: We developed a novel polygenic risk score that predicts clinical risk of aortic stenosis-related adverse events in patients with cardiometabolic disease, even after adjustment for traditional risk factors.