Integrative Omics in Failing Myocardium Highlights Metabolic Substrate Utilization Defects in Human Heart Failure

Introduction: Changes in substrate utilization may be a biomarker of, or a contributor to, heart failure with reduced ejection fraction (HFrEF). Previous work has described circulating metabolites, proteins and transcripts that reflect myocardial bioenergetics, but direct assessment in relevant tissues is lacking. Here, we leverage integrative omics to identify tissue-level biomarkers and pathways of HFrEF that may play a role in pathogenesis. Methods: Myocardial tissue was obtained from 27 HFrEF donors and 21 non-failing hearts (noHF) from Duke University and the University of Colorado Medical Centers. Targeted (n=139 metabolites) and nontargeted (Metabolon, n=817) metabolic profiling was performed; in addition, 496 proteins were assayed (Olink) and ~31,000 transcripts were quantified via RNA sequencing. Individual metabolites were tested for association with HFrEF in models adjusted for age, sex and diabetes. Metabolomics, proteomics and RNA-seq data were integrated using block sparse partial least squares discriminant analysis (sPLS-DA) to identify correlated biomarkers that discriminate HFrEF from noHF hearts. Results: The targeted metabolomics platform primarily assayed metabolic fuel substrate pathways; 71 analytes (51%) were associated with HFrEF after controlling the false discovery rate (FDR) at 5%. Medium- and long-chain acylcarnitines (C10s-C18s) were significantly lower in tissue from failing hearts, while branched-chain amino acids (BCAA) and ketoacids were higher. In the nontargeted set, 293 metabolites (36%) differed between HFrEF and noHF. BCAA and acylcarnitine associations were recapitulated along with a decrease in malonylcarnitine. The sPLS-DA model selected 25 targeted and 50 nontargeted metabolites, 35 proteins and 35 transcripts, and successfully discriminated HFrEF from no HF (balanced error rate = 3.2%). 39/186 tested KEGG pathways were overrepresented in these analytes, including fatty acid metabolism (FDR q=0.002). Conclusions: Analysis of tissue-level metabolites suggests both decreased use of fatty acids as a fuel and disrupted enzymatic breakdown of BCAAs in the failing heart, while integrative omics pathway analysis also supports the association of disrupted fatty acid metabolism with HF.