Protein Biomarkers of New-Onset Heart Failure: Insights From the Heart Omics and Ageing Cohort, the Atherosclerosis Risk in Communities Study, and the Framingham Heart Study

Abstract

Background: We sought to identify protein biomarkers of new-onset heart failure (HF) in 3 independent cohorts (HOMAGE cohort [Heart Omics and Ageing], ARIC study [Atherosclerosis Risk in Communities], and FHS [Framingham Heart Study]) and assess if and to what extent they improve HF risk prediction compared to clinical risk factors alone.

Methods: A nested case-control design was used with cases (incident HF) and controls (without HF) matched on age and sex within each cohort. Plasma concentrations of 276 proteins were measured at baseline in ARIC (250 cases/250 controls), FHS (191/191), and HOMAGE cohort (562/871).

Results: In single protein analysis, after adjusting for matching variables and clinical risk factors (and correcting for multiple testing), 62 proteins were associated with incident HF in ARIC, 16 in FHS, and 116 in HOMAGE cohort. Proteins associated with incident HF in all cohorts were BNP (brain natriuretic peptide), NT-proBNP (N-terminal pro-B-type natriuretic peptide), eukaryotic translation initiation factor 4E-BP1 (4E-binding protein 1), hepatocyte growth factor (HGF), Gal-9 (galectin-9), TGF-alpha (transforming growth factor alpha), THBS2 (thrombospondin-2), and U-PAR (urokinase plasminogen activator surface receptor). The increment in C-index for incident HF based on a multiprotein biomarker approach, in addition to clinical risk factors and NT-proBNP, was 11.1% (7.5%-14.7%) in ARIC, 5.9% (2.6%-9.2%) in FHS, and 7.5% (5.4%-9.5%) in HOMAGE cohort, all P<0.001), each of which was a larger increase than that for NT-proBNP on top of clinical risk factors. Complex network analysis revealed a number of overrepresented pathways related to inflammation (eg, tumor necrosis factor and interleukin) and remodeling (eg, extracellular matrix and apoptosis).

Conclusions: A multiprotein biomarker approach improves prediction of incident HF when added to natriuretic peptides and clinical risk factors.

Keywords: biomarkers; cardiovascular diseases; heart failure; protein interaction maps; proteomics.

Figure 1.

Venn diagram with the significant biomarkers identified in each cohort. Protein names are available in Table S1.

Figure 2.

Venn diagram for the biomarkers in each cohort that improved the C-index when added to the clinical model. Protein names are available in Table S1.

Figure 3.

Predictive value of multiprotein biomarkers approach for incident heart failure (HF) in addition to routine clinical markers of incident heart failure in the 3 cohorts. Biomarkers retained in multivariable model in each cohort (including NT-proBNP [N-terminal pro-B-type natriuretic peptide]) included the following: ARIC (Atherosclerosis Risk in Communities study; P=55): all significant BMs, except BNP (brain natriuretic peptide)+AXIN1, CASP-3, PDGF subunit A, PDGF subunit B, SRC, STAMPB; FHS (Framingham Heart Study; P=15): all significant BMs, except BNP; HOMAGE (Heart Omics and Ageing cohort; P=109): all significant BMs, except BNP, EPHB4, IL-18BP, OSM, TNF-R1, TNF-R2, TNFRSF14. The clinical model included the following variables: age, sex, smoking, diabetes, history of coronary artery disease, serum creatinine, body mass index, systolic blood pressure, use of antihypertensive medication, heart rate, hypertension, history of atrial fibrillation, and ratio total/high-density lipoprotein cholesterol.

Figure 4.

Complex network analysis of significant protein biomarkers. Overrepresented pathways are symbolized by white triangles and biomarkers are ellipses. Biomarker color depends on the study where it was identified. Protein names are available in Table S1.