Quantitative Metabolomics Reveals Heart Failure With Midrange Ejection Fraction as a Distinct Phenotype of Heart...

Heart failure with midrange ejection fraction (HFmrEF) has been recently acknowledged as a separate phenotype, but metabolomics evaluation of this subtype remains largely unexamined.

A quantitative metabolomics study on amino acids and acylcarnitines was performed to characterize different states of heart failure (HF) in 628 participants. Both multivariate orthogonal partial least squares- discriminant analysis and univariate Mann-Whitney U test were used to explore reliable metabolic profiles associated with different HF states. The resulting metabolites were further refined to obtain diagnostic metabolite scores (DMSs) with the use of ordinal logistic regression. Lasso-penalized regression was applied to produce a survival-associated prognostic metabolite score (PMS). The Cox proportional hazards model, Kaplan-Meier curves, and time-dependent receiver operating characteristics were used for a comprehensive assessment of prognostic value using PMS versus traditional clinical biomarkers.

The optimized models identified a panel of 15 differential metabolites that were shared across different HF states, whereas some metabolites were associated with a specific state. PMS consisting of 9 metabolites demonstrated an appreciably better prognostic value (hazard ratio [HR] 1.62, 95% confidence interval [CI] 1.25-2.1) vs the natural logarithm of N-terminal pro-B-type natriuretic peptide (Ln[NT-proBNP]; HR 1.23, 95% CI 0.94-1.61; P < 0.001). The overall area under the receiver operating characteristic curve value of PMS was superior to that of Ln(NT-proBNP) in risk prediction for patients with HFmrEF and HF with reduced ejection fraction (HFrEF) subtypes (P < 0.001).

Targeted metabolomics has provided a novel understanding of the molecular mechanism underlying HF. Both DMS and PMS clearly demonstrated HFmrEF as a distinct phenotype between a mild HF with preserved ejection fraction state and a severe HFrEF state. PMS exhibited superior prognostic value than Ln(NT-proBNP). Further investigation is needed with independent large-scale validation.