Chronic heart failure (CHF) is an ongoing clinical syndrome with cardiac dysfunction that can be traced to alterations in cardiac metabolism. The identification of metabolic biomarkers in easily accessible fluids to improve the early diagnosis of CHF has been elusive to date. In this study, we took multidimensional analytical techniques to discover potentially new diagnostic biomarkers by focusing on the dynamic changes of metabolites in serum during the progression of CHF. Using mass-spectrometry-based untargeted metabolomics, we identified 23 cardiac metabolites that were altered in a rat model of myocardial infarction induced CHF. Among these differential metabolites, branched-chain amino acids (BCAAs) in serum, especially leucine and valine, showed a high capability to differentiate between CHF and sham-operated rats, of which area under the receiver operating characteristic curve was greater than 0.75. Combining with targeted analysis of the amino acids and related proteins and genes, we confirmed that BCAA metabolic pathway was significantly inhibited in rat failing hearts. On the basis of the time series data of serum samples, we characterized the fluctuation pattern of circulating BCAAs by the disease progression model. Finally, the time-resolved diagnostic potential of serum BCAAs was evaluated by the machine-learning-based classifier, and high diagnostic accuracy of 93.75% was achieved within 3 weeks after surgery. These findings provide a promising metabolic signature that can be further exploited for CHF early diagnostic development.
Keywords: branched-chain amino acid; classifier; disease progression model; heart failure; metabolomics.