A novel score of IL-13 and age predicts 90-day mortality in severe alcohol-associated hepatitis: A multicenter plasma biomarker analysis.

Background: Severe alcoholic hepatitis (AH) has a high short-term mortality rate. The MELD assesses disease severity and mortality; however, it is not specific for AH. We screened plasma samples from patients with severe AH for biomarkers of multiple pathological processes and identified predictors of short-term mortality.

Methods: Plasma was collected at baseline from 85 patients with severe AH (MELD≥20, Maddrey's discriminant function≥32) enrolled in the Defeat Alcoholic Steatohepatitis clinical trial (investigating IL-1 receptor antagonist+pentoxifylline+zinc vs. methylprednisolone+placebo). Samples were analyzed for 43 biomarkers and the markers' association with 28- and 90-day mortalities was assessed.

Results: Thirty-one (36.5%) patients died during the 90-day follow-up with similar ratios in the treatment groups. Eight biomarkers showed an association with mortality. IL-6, IL-22, interferon-α2, soluble TNF receptor 1, lipocalin-2, and α-fetoprotein levels were associated with 28-day mortality, while IL-6, IL-13, and endotoxin levels with 90-day mortality. In multivariable Cox regression, encephalopathy, lipocalin-2, and α-fetoprotein levels were independent predictors of 28-day mortality, and IL-6, IL-13, international normalized ratio levels, and age were independent predictors of 90-day mortality. The combination of IL-13 and age had superior performance in predicting 90-day mortality compared with MELD in the total cohort and the individual treatment groups.

Conclusions: We identified predictors of short-term mortality in a cohort exclusively involving patients with severe AH. We created a composite score of IL-13 and age that predicts 90-day mortality regardless of the treatment type with a performance superior to MELD in severe AH.