Untargeted metabolomics identifies novel metabolites and pathways in Finnish men with heart failure

Abstract Early predictors of heart failure (HF) remain poorly defined. We measured plasma metabolites in a population-based Finnish cohort to identify metabolites that independently predict incident HF. We also performed an exploratory analysis of whether selected metabolites improved discrimination beyond traditional clinical and laboratory factors. Untargeted LC–MS/MS quantified 1009 known metabolites in 10,115 men from the METSIM study. A total of 663 participants developed HF during 14.2-year follow-up. We applied Cox regression analysis to associate the clinical factors and metabolites alone and combined with the risk of HF. Genetic variants that were associated with metabolites and increased the risk of HF were identified. Forty-five metabolites (33 were novel) increased significantly the risk of HF, including markers of catecholamine turnover, polyamine catabolites, very-long-chain acyl-carnitines, and bile-acid sulfates. ROC analysis showed that adding 12 metabolites to the clinical model increased significantly the AUC from 0.775 to 0.795. The metabolites identified converged to the adrenergic/mitochondrial redox stress, nitrogen-polyamine flux, and membrane-lipid/bile-acid remodeling pathways. Ten metabolite-quantitative-trait loci included signaling, nitrogen handling, lipid remodeling, and nucleotide detoxification pathways to specific genes. We also showed that these metabolites add prognostic value to conventional risk factors and highlight pathways of early myocardial injury in patients with HF.