A Polygenic Risk Score to Predict Incident Heart Failure Across the Spectrum of Cardiovascular Risk

Background The clinical utility of a heart failure (HF) polygenic risk score (PRS) is uncertain. Objectives The purpose of this study was to investigate the ability of an HF PRS to predict new-onset HF in individuals across the spectrum of cardiovascular risk. Methods An HF PRS (>1 million single nucleotide variations) was used to stratify individuals from 7 clinical studies to low (quintile [Q] 1), intermediate (Q2-Q4), or high (Q5) genetic risk. In 6 trials of patients at elevated cardiovascular risk, HRs adjusted for clinical factors were derived for the risk of hospitalization for HF using Cox-PH models, and discrimination and calibration was explored. Analyses were conducted in a separate data set of individuals at low cardiovascular risk. Results We studied 74,897 patients without HF, of which 51,627 were at elevated cardiovascular risk (median age 65 years; 71% men; median follow-up 2.5 years). After adjusting for clinical factors, individuals with intermediate and high HF PRS carried a 2- and 5-fold increased rate of new-onset HF, respectively (intermediate risk: HRadj: 2.01 [95% CI: 1.62-2.49]; P < 0.001; high-risk: HRadj: 5.00 [95% CI: 3.99-6.27]; P < 0.001). Addition of the HF PRS to a clinical model improved the AUC (0.787 [95% CI: 0.775-0.798] to 0.822 [95% CI: 0.811-0.832]). Results were consistent in 23,270 individuals at low cardiovascular risk over 20-year follow-up. Conclusions An HF PRS is a strong and independent predictor for new-onset HF in individuals across the spectrum of cardiovascular risk. The polygenic contribution to HF is complementary to established clinical risk assessment.