A Novel Scoring System to Predict Outcomes in Severe Alcoholic Hepatitis

Current tools to predict survival in patients with severe alcoholic hepatitis (SAH), such as the model for end-stage liver disease (MELD) and Lille scores, do not perform consistently. Better models are needed to identify which patients with SAH might benefit from medical therapy or liver transplantation.

Researchers developed and tested a scoring system to predict transplantation-free survival at 90 and 180 days in SAH (Maddrey's score ≥32) using baseline clinical variables and molecular variables (gene expression patterns from fixed liver biopsy samples) collected from 71 patients in Belgium. The score's accuracy was tested in two validation cohorts of patients with SAH: 48 patients from Belgium and Switzerland and 20 patients referred for liver transplantation evaluation in the U.S.

The analysis identified 123 genes plus the MELD score as comprising the highest-performing prognostic model, in which a poor prognosis group had 90-day and 180-day survival rates of 29% and 26%, respectively, and a good prognosis group had 90-day and 180-day survival rates of 76% and 65%, respectively. In the first validation cohort, the poor prognosis group had 90-day and 180-day survival rates of 43% and 34%, respectively, and the good prognosis group's rates were 96% and 84%, respectively. In the second validation cohort, those with poor prognosis had respective 90-day and 180-day survival rates of 24% and 12%, and those with good prognosis had 100% survival. Overall predictive performance was high (area under the curve, 0.86 for 90-day survival and 0.83 for 180-day survival).