Abstract 14071: Plasma Protein Concentrations and Risk of Incident Cardiovascular Diseases in the UK Biobank

Abstract

Introduction: Profiling of plasma proteins has the potential to improve diagnosis, predict disease and reveal new biologic pathways for cardiovascular disease (CVD). Objectives: We tested the associations of circulating proteins with new-onset coronary artery disease (CAD), atrial fibrillation (AF), heart failure (HF), and aortic stenosis (AS) in the UK Biobank. Methods: The initial dataset consisted of 52,705 individuals and 1,463 proteins measured using the Olink platform. A total of 47,665 unrelated individuals without prevalent CVD and 1,459 proteins were retained after quality control. Cox proportional hazards models were used to evaluate associations of protein levels at baseline with incident CAD, AF, HF, and AS. Models were adjusted for age, age 2 , race, first ten principal components of genetic ancestry, sex, smoking, BMI, SBP, antihypertensive and lipid-lowering medication use, DMII, serum creatinine, LDL and HDL cholesterol, and Townsend deprivation index. The four outcomes were incorporated as time-varying covariates. Results: Among the 47,665 individuals, mean (SD) age was 56.8 (8.2) years, 25,674 (53.9%) were female, and 44,691 (93.8%) white. Over a median follow-up of 11.1 years, there were 2,941 (6.2%), 2,502 (5.2%), 1,470 (3.1%) and 420 (0.9%) incident cases of CAD, AF, HF, and AS, respectively. Multivariable-adjusted Cox regression revealed 270, 156, 385, and 21 significant associations (P < 8.6 x 10-6) for CAD, AF, HF, and AS, respectively. Fifteen biomarkers overlapped across all outcomes. The strongest associations for CAD, AF, HF, and AS were with GDF-15 (HR = 1.33, 95% CI = 1.28 - 1.39), NT-proBNP (HR = 1.83, 95% CI = 1.75 - 1.91), HE4 (encoded by WFDC2; HR = 1.67, 95% CI = 1.57 - 1.77), and GDF-15 (HR = 1.51, 95% CI = 1.43 - 1.59), respectively. NT-proBNP, IGFBP-4 and HE4 were the strongest overlapping associations. Conclusions: High-throughput proteomics may offer new mechanistic insights and improve risk prediction for common CVDs.