Enhanced diagnosis and prognosis of severe alcoholic hepatitis using novel metabolomic biomarkers.

Abstract

Differentiating alcoholic hepatitis (AH) from acute decompensation of alcoholic cirrhosis (DC) is challenging, as the presentation and biochemistry are similar. We aimed to identify potential metabolomic biomarkers to differentiate between AH and DC, and to predict short-term mortality. We included consecutive biopsy proven AH and DC patients, which were managed according to current guidelines and followed up until the end of the study. Untargeted metabolomics was assessed in all patients at baseline. Specific analyses were successively performed to identify potential biomarkers, which were further semi-quantitatively analysed against relevant clinical endpoints. Thirty-four patients with AH and 37 with DC were included. UHPLC-MS analysis identified 83 molecules potentially differentiating between AH and DC. C16-Sphinganine-1P (S1P) was the most increased, whereas Prostaglandin E2 (PGE2) was the most decreased. The PGE2/S1P ratio < 1.03 excellently discriminates between AH and DC: AUC 0.965 (p < 0.001), Se 90%, Sp 100%, PPV 0.91, NPV 1, and diagnostic accuracy 95%. This ratio is not influenced by the presence of infection (AUC 0.967 vs. 0.962), correlates with the Lille score at 7days (r= -0.60; P= 0.022) and tends to be lower in corticosteroid non-responders as compared with patients who responded [0.85(±0.02) vs. 0.89(±0.05), P= 0.069]. Additionally, decreased ursodeoxycholic acid levels are correlated with MELD and Maddrey scores and predict mortality with a 77.27% accuracy (NPV = 100%). This study suggests the PGE2 (decreased)/S1P (increased) ratio as a biomarker to differentiate AH from DC. The study also finds that low levels of ursodeoxycholic acid could predict increased mortality in AH.