Abstract 12100: Exploration of Proteomic Risk Markers for Incident Heart Failure and Cardiac Remodeling in Three European Population-Based Cohorts

Abstract

Introduction: Several biomarkers have been linked with incident heart failure (HF), but the pathophysiology of HF is complex and multifactorial. Hypothesis: We explored 92 biomarkers’ associations with left ventricular hypertrophy (LVH), prevalent LV diastolic dysfunction (LVDD), and incident HF in three European population-based cohorts. Methods: Profiling of 92 biomarkers was carried out in 1737 participants from the echocardiography subsample of the population-based cohort study, the Malmö Preventive Project (mean age 67 years, 29% women). The final study sample consisted of 1681 subjects for analyses of incident HF (n=130), and 1530 subjects for analyses of LVDD (n=109) and LVH (n=498). Multivariable Cox regression analyses excluding subjects with prevalent HF were carried out for incident HF. Multivariable logistic regression analyses excluding subjects with reduced or mid-range LV ejection fraction (<50%) were carried out for LVDD and LVH. Replication analyses of significant findings were carried out in the HOMAGE (562 cases of incident HF, 871 controls) and STANISLAS cohorts (304 cases of LVDD and 1668 controls; and 245 cases of LVH and 1681 controls). Results: In the multivariable model adjusted for clinical risk factors, six proteins were associated with incident HF (median follow-up time 12.6, interquartile range 10.9-13.6 years): ST2 (hazard ratio (HR) 1.40; 95% confidence interval (1.03-1.91); p=0.031), GDF-15 (HR 1.33 (1.00-1.77); p=0.048), AZU1 (HR 1.42 (1.10-1.82); p=0.006), MPO (HR 1.48 (1.02-2.15); p=0.041), Gal-4 (HR 1.37 (1.02-1.85); p=0.038) and NT-proBNP (HR 1.98 (1.66-2.36); p=2.1x10 -14 ) in the discovery cohort. In the replication cohort, ST2, GDF15, NT-proBNP and Gal-4 also displayed significant associations with incident HF. In analyses of DD, none of the associated proteins in the discovery cohort (AZU1; TR; PCSK9 and NT-proBNP) were replicated in the STANISLAS cohort. Among AZU1, PGLYRP1, and NT-proBNP, NT-proBNP was the only protein that was significantly associated with LVH in both cohorts. Conclusions: We replicated the association of ST2, GDF-15, NT-proBNP and Gal-4 with incident HF, whereas NT-proBNP was the only biomarker that was associated with LVH as well as incident HF in discovery and replication cohorts.