The overarching goal of this project is to improve care for patients with heart failure (HF). HF, whether with reduced (HFrEF) or preserved (HFpEF) ejection fraction, is associated with significant morbidity, mortality, and cost. In the U.S. alone, HF affects over 5 million adults, and the prevalence is projected to exceed 8 million by 2030. HF is the most frequent cause of hospitalization among Medicare recipients and results in over $30 billion in health care expenditures each year. Advances in management, especially for HFrEF, have modestly reduced death rates over time, but mortality continues to be high, with approximately half of patients dying within 5 years of diagnosis. Moreover, the pace of drug discovery has been slow, and there are no proven therapies for patients suffering with HFpEF. Among patients with established HF there is substantial variation in illness severity, degree of cardiac remodeling, disease progression, and response to therapy. These observations highlight the heterogeneity of the HF syndrome and suggest existence of subtypes with differing clinical and potentially genetic profiles, with subsequent differences in downstream disease mechanisms, overall risk, and therapeutic response. However, the understanding of the phenotypic, genetic, and pathophysiological heterogeneity of HF is incomplete. This project investigates the phenotypic substructure and genetic architecture of HF by leveraging a unique collection of interrelated datasets from Vanderbilt University Medical Center (VUMC), including the de- identified electronic health record (EHR) and BioVU, a linked DNA biobank. The EHR contains ~2.6 million patients, including ~35,000 with HF, and BioVU currently houses >225,000 DNA samples. Dense genotype data are available in >28,000 subjects and an institutional genotyping project will increase this to >125,000 by mid- 2017; this includes >13,000 subjects with HF. The proposed research will: 1) identify HF subtypes from dense clinical data alone using advanced, unbiased, deep learning algorithms (Aim 1), 2) define the genetic architecture of HF and HF subtypes by using inferred gene expression, general linear mixed models, genetic risk scores, and traditional association testing to quantify heritability of and genetic correlations among HF subtypes, define the contribution of established risk factors to HF subtypes, and 3) discover subtype-specific genetic risk factors (Aim 2), and discover HF subtype-specific clinical outcomes, disease associations, and drug response phenotypes using advanced phenome scanning and network analysis (Aim 3).

Public Health Relevance

Heart failure (HF) is a complex, debilitating syndrome associated with significant morbidity and mortality. The heterogeneity of HF has limited success of prior efforts to understand HF pathobiology and develop effective interventions. By defining clinical and genetic modifiers of HF risk, disease course, and treatment response for clinically recognized and novel, data-driven HF subtypes, results from this work could result in a more sophisticated HF classification system based on underlying biology, and ultimately facilitate precision risk stratification, tailoring of therapeutic strategies, and rational HF clinical trials.

National Institute of Health (NIH)
National Heart, Lung, and Blood Institute (NHLBI)
Research Project (R01)
Project #
Application #
Study Section
Clinical and Integrative Cardiovascular Sciences Study Section (CICS)
Program Officer
Papanicolaou, George
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Vanderbilt University Medical Center
United States
Zip Code
Mosley, Jonathan D; Levinson, Rebecca T; Brittain, Evan L et al. (2018) Clinical Features Associated With Nascent Left Ventricular Diastolic Dysfunction in a Population Aged 40 to 55 Years. Am J Cardiol 121:1552-1557