Metabolic Profiling Associates with Disease Severity in Non-Ischemic Dilated Cardiomyopathy

ABSTRACT Aims Metabolomic profiling may have diagnostic and prognostic value in heart failure. This study investigated whether targeted blood and urine metabolomics reflects disease severity in non-ischemic dilated cardiomyopathy (DCM) patients and compared its incremental value on top of NT-proBNP. Methods and results A total of 149 metabolites were measured in plasma and urine samples of 273 DCM patients with different stages of disease (DCM patients with LVRR (normal LVEF), n=70; asymptomatic DCM, n=72 and symptomatic DCM, n=131). Acylcarnitines, sialic acid, and glutamic acid are the most distinctive metabolites associated with disease severity, as repeatedly revealed by uni-biomarker linear regression, sPLSDA, Random Forest and conditional Random Forest analyses. However, the absolute difference of the metabolic profile among groups was marginal. A decision tree model based on the top metabolites did not surpass NT-proBNP in classifying stages. However, a combination of NT-proBNP and the top metabolites improved the decision tree to distinguish DCM patients with LVRR from symptomatic DCM (AUC 0.813±0.138 versus 0.739±0.114; p=0.02). Conclusion Functional cardiac recovery is reflected in metabolomics. These alterations reveal potential alternative treatment targets in advanced symptomatic DCM. The metabolic profile can complement NT-proBNP in determining disease severity in non-ischemic DCM.