Prognosis of treated severe alcoholic hepatitis in patients with gastrointestinal bleeding.

Background & Aims: All trials on severe alcoholic hepatitis (AH) have included patients with "pure" AH, i.e., without concomitant gastrointestinal bleeding (GIB). Severe AH is often suspected in cirrhotic patients with GIB. We aimed at (1) assessing the prevalence of AH in patients with GIB and Maddrey discriminant function (DF) ⩾32; (2) comparing the outcome in AH patients with or without GIB (AH-GIB+, AH-GIB-); and (3) assessing the performance of the Lille model for survival in AH-GIB+ patients.

Methods: We retrospectively included all patients with alcoholic cirrhosis admitted between January 2005 and March 2011 with the following: (1) jaundice <3 months; (2) DF ⩾32 at admission; (3) bilirubin level >50 μmol/L; and (4) active drinking. Exclusion criteria were advanced hepatocellular carcinoma, other etiology of cirrhosis, severe comorbidities and DF <32 after stabilization. In our centre, we systematically plan a liver biopsy for these patients. Patients with severe AH received prednisolone.

Results: We screened 161 patients (86 GIB+, 75 GIB-), and analyzed data for 58 and 47 patients in each group, respectively. The 2 groups did not differ in prevalence of AH (77.3% vs. 81%), demographic data, MELD/Child-Pugh score, or DF. The 2 groups were similar in 6-month probability of survival (73.9 ± 6.0% vs. 69.9 ± 7%, p=0.49). The probability of developing infection was lower for AH-GIB+ patients (24.1% vs. 44.7%, p=0.04). The AUC for the Lille model in predicting 6-month survival was 0.71 ± 0.06 for all patients and 0.74 ± 0.06 for AH-GIB+ patients (p>0.05).

Conclusions: Prevalence of AH is 80% for patients with cirrhosis and GIB, recent jaundice and DF ⩾32. Infection was lower for AH-GIB+ patients, which suggests a beneficial role of antibiotic prophylaxis treatment. Survival among subjects with GIB was the same as among subjects without GIB.