Abstract 10621: Prediction of Incident Atherosclerotic Cardiovascular Disease Using Traditional and Polygenic Risk Score Modeling: The Million Veteran Program Experience

Abstract

The prediction of myocardial infarction (MI), acute ischemic stroke (AIS), cardiovascular (CVD) death, and 3-point MACE (MI, AIS, or CVD death) in Million Veteran Program (MVP) participants over 8 years of follow up was evaluated using environmental/traditional risk factors (“E”), polygenic risk scores (“G”), and a combination of both approaches (“GxE”). Individuals free of atherosclerotic cardiovascular disease (ASCVD) at baseline were included. Risk factor levels for the participants close to the time of MVP enrollment for age, sex, systolic blood pressure, cholesterol, HDL cholesterol, smoking status, and diabetes status were analyzed. Outcomes were determined from VA electronic record data, Medicare/Medicaid data, and the National Death Index. Analyses were undertaken separately for non-Hispanic European (EUR), African American (AFR), and Hispanic (HIS) participants. There were 157,941 veterans at risk with 8,157 MI events, 2,024 AIS events, 1,778 CVD deaths, and 9,350 3-pt MACE events over 8 years of follow-up. The overall results showed good performance with the environmental model for all three racial-ethnic groups, with C-statistics ranging from 0.69 to 0.77. The G models showed very modest prediction capabilities and similarly modest improvement in the combined GxE models. In conclusion, traditional risk factor modeling has been shown to be highly effective and, in the MVP experience, the additional impact of genetic or genetic interaction information was small.