The Diagnostic and Prognostic Significance of Liver Histology in Alcoholic Hepatitis

Ewan Forrest; Gemma Petts; Andrew Austin; Kirsty Lloyd; Mark Wright; Nikhil Vergis; Stephen Atkinson; Steven Masson; David Patch; Alberto Quaglia; Mark Thursz; Robert Goldin


Aliment Pharmacol Ther. 2021;53(3):426-431. 

In This Article



Patients recruited to the Steroids or Pentoxifylline for Alcoholic Hepatitis (STOPAH) trial were studied. Recruitment to the trial was based on a clinical diagnosis of AH and a Maddrey's Discriminant Function (DF) greater than or equal to 32. Full details of eligibility criteria and the clinical outcomes from the trial have recently been published[7,8] but all patients had recent onset of jaundice with a serum bilirubin greater than 80μmol/l. A liver biopsy was encouraged although not a requirement for inclusion in the trial. The Glasgow alcoholic hepatitis score (GAHS), Model for End Stage Liver Disease (MELD: pre-2016 UNOS variant) score and Discriminant function (DF) were calculated according to previously published protocols.[9–11]

Liver Biopsy and Histology

Liver biopsies were performed via transjugular route by an experienced hepatologist or interventional radiologist. Biopsies were formalin fixed and paraffin embedded with two histological sections from each patient stained with Haematoxylin & Eosin and Sirius Red at St. Mary's Hospital. Two experienced liver histopathologists, blinded to treatment arm and patient outcome, independently recorded histological features of alcoholic hepatitis including those described in the AHHS as well as the Laennec fibrosis[12] and Kleiner steatosis scores.[13] Biopsies were considered adequate when there were at least five portal tracts available for assessment and ASH diagnosed when hepatocyte steatosis, ballooning of hepatocytes and inflammatory infiltration were sufficiently evident. Only those biopsies obtained before or within 48 hours of randomisation were used to relate histological features with subsequent outcome.

Statistical Analysis

Analyses were performed using MedCalc Statistical Software version 17.6 (MedCalc Software bvba, Ostend, Belgium;; 2017). Comparison of the discriminatory strength of scores was performed by area under the Receiver Operating Curve (AUC) analysis. Kaplan-Meier analysis was used to assess survival and survival curves were compared using the Log-Rank test with hazard ratio (HR) described where relevant. Results are presented with 95% confidence intervals (95% CI). Direct comparisons between data were performed using t-tests and between proportions using Odds Ratio (OR) calculation. Spearman's rho correlation co-efficient (ρ) was calculated for associations between variables.