Background: Despite major advances in pharmacological treatment for patients with heart failure, residual mortality remains high. This suggests that important pathways are not yet targeted by current heart failure therapies.
Objectives: We sought integration of genetic, transcriptomic, and proteomic data in a large cohort of patients with heart failure to detect major pathways related to progression of heart failure leading to death.
Methods: We used machine learning methodology based on stacked generalization framework and gradient boosting algorithms, using 54 clinical phenotypes, 403 circulating plasma proteins, 36,046 transcript expression levels in whole blood, and 6 million genomic markers to model all-cause mortality in 2,516 patients with heart failure from the BIOSTAT-CHF (Systems BIOlogy Study to TAilored Treatment in Chronic Heart Failure) study. Results were validated in an independent cohort of 1,738 patients.
Results: The mean age of the patients was 70 years (Q1-Q3: 61-78 years), 27% were female, median N-terminal pro-B-type natriuretic peptide was 4,275 ng/L (Q1-Q3: 2,360-8,486 ng/L), and 7% had heart failure with preserved ejection fraction. During a median follow-up of 21 months, 657 (26%) of patients died. The 4 major pathways with a significant association to all-cause mortality were: 1) the PI3K/Akt pathway; 2) the MAPK pathway; 3) the Ras signaling pathway; and 4) epidermal growth factor receptor tyrosine kinase inhibitor resistance. Results were validated in an independent cohort of 1,738 patients.
Conclusions: A systems biology approach integrating genomic, transcriptomic, and proteomic data identified 4 major pathways related to mortality. These pathways are related to decreased activation of the cardioprotective ERBB2 receptor, which can be modified by neuregulin.
Keywords: heart failure; machine learning; omics; systems biology.
Copyright © 2023 American College of Cardiology Foundation. Published by Elsevier Inc. All rights reserved.