Purpose: The pathobiochemical mechanisms underlying the development of acute heart failure (AHF) on the level of metabolomics are poorly understood. Therefore, the present study aims to investigate AHF disease specific metabolomic profiles associated with the clinical presence of the AHF syndrome in affected patients. 

Methods: Consecutive AHF patients presenting to the emergency department as well as healthy subjects serving as controls, were included prospectively in the MEMORIA study. Blood samples including EDTA plasma were drawn and stored. EDTA plasma from healthy subjects served as the control group. 630 different metabolites were measured by the MxP® Quant 500 kit (Biocrates life sciences ag) based on a targeted mass spectrometry approach. Statistical analysis was performed by metaboanalyst (5.0) and R statistics (The R Foundation ©, r-project.org) including fold change analyses, univariable t-test and false discovery rating (FDR), multivariate principal component analysis (PCA), partial least squares regression with discriminant analysis (PLS-DA), permutation testing, heatmaps and C-statistics with calculation of area under the receiver operating characteristic curves (AUC).

Results: A total of 150 consecutive AHF patients were compared to 84 healthy controls. Mean age of AHF patients was 75 years (range 51-93 years), 57% males and 43% females. Coronary artery disease was present in 64%, valvular heart disease in 96%, median left ventricular function (LVEF) was 40% (interquartile range (IQR) 31-49%), E/E´ 14 (IQR 11-20), and tricuspid annular systolic excursion (TAPSE) was 17 mm (IQR 14-21 mm). Univariable t-tests revealed 28 significantly altered metabolites. After applying FDR, multivariable PCA and PLS-DA including permutation testing 10 metabolites remained significantly altered between AHF and controls: phosphatidylcholine (PC aa 24:0) (p=2.0394e-21), medium-chain acylcarnitines (C3-DC (C4-OH) (p=0.000004), xanthine (p=0.0000004), anserine (p=0.00004), trigonelline (p=0.0004), cholesterol esters (CE 20:1) (p=0.0001) and the triglycerides (18:3 36:2) (p=0.0005), (18:3 36:3) (p=0.0001), and (18:3 36:4)(p=0.00005). Applying multivariable C-statistics AHF patients were significantly discriminated from healthy controls by these metabolites (AUC = 0.910 (95% confidence interval (CI) 0.843-0.959; p=0.0001).