Proteomic biomarkers associated with incident heart failure and frailty in late life

Abstract

Abstract Background Heart failure (HF) and frailty are highly prevalent in late life and commonly co-exist, but the mechanisms underlying their bi-directional relationship are incompletely understood. This study aimed to identify shared molecular pathways associated with incident HF and frailty in late life. Methods Among participants in the Atherosclerosis Risk in Communities (ARIC) study, a communit-based cohort study in the United States, 4,877 plasma proteins were measured using an aptamer-affinity assay (Somascan v4) at study Visit 3 (V3; 1993–1994; n=10,368, age 60±6 years; 822 incident HF events) and at study Visit 5 (V5; 2011–2013; n=3,908, age 75±5 years; 336 incident HF events). Frailty was assessed at V5 using Fried criteria, which incorporates gait speed, grip strength, low energy expenditure, weight loss, and exhaustion. We examined the association of proteins at V3 with incident HF after V3 with Bonferroni corrected P<0.05 using multivariable Cox proportional hazard regression models. For HF-associated proteins at V3, we assessed the association of V5 protein levels with incident HF after V5. For the resulting HF-associated proteins, multivariable logistic regression was used to assess associations of V5 protein values with prevalent frailty at V5 (n=223 cases) and with incident frailty by study Visit 6 (2016–2018; n=152 incident cases). All models adjusted for age, sex, race, hypertension, diabetes, cardiovascular disease, BMI, atrial fibrillation, and stroke. The set of HF-related proteins that associated with incident frailty at FDR <0.05 using Benjamini-Hochberg correction was tested for pathway enrichment using the Reactome database. Results Of 289 proteins associated with incident HF post-V3 at p<1.0x10–5 (0.05/4,877), 84 were significantly associated with incident HF post-V5 at p<1.7x10–4 (0.05/289). Among 4,131 HF-free participants at V5, 48 of these 84 HF-associated proteins associated with prevalent frailty at p<5.9x10–4 (0.05/84). Among Visit 5 participants who completed a frailty assessment and were free of both prevalent HF and frailty (n=3,908), 31of 48 candidate proteins were also significantly associated with incident frailty at FDR 0.05, 18 of which were significantly associated with incident frailty at p<1.0x10–3 (0.05/48; Figure 1). The 31 proteins associated with incident frailty at FDR 0.05 enriched for collagen biosynthesis, formation, and trimerization (COL28A1, COL6A3, EFEMP1), and cytokine immune pathways and TNF receptor binding (TNFRSF1A and B, VEGFA, B2M, and HAVCR2) in pathway enrichment analysis. Conclusions Collagen metabolism and immune pathways may be shared biologic pathways between HF and frailty in late-life. Funding Acknowledgement Type of funding sources: Public Institution(s). Main funding source(s): The Atherosclerosis Risk in Communities study has been funded in whole or in part with Federal funds from the National Heart, Lung, and Blood Institute, National Institutes of Health, Department of Health and Human Services.