Defining Risk in Alcohol-Associated Liver Disease Using the Model for End-Stage Liver Disease.

Abstract

INTRODUCTION: Alcohol-associated liver disease (ALD) is a common cause of morbidity and premature mortality. Most prognostic scores have been defined in the short term. We used a large retrospective cohort of patients with ALD to describe the natural history of ALD and to define risk prediction in the longer term, taking nonliver mortality into account. METHODS: The WALDO cohort includes 734 patients with biopsy-proven ALD. Prognostic scores were assessed with dynamic area under the curve and C-index. Risk estimates for morbidity and mortality were derived for the model for end-stage liver disease (MELD) and validated in an external cohort. RESULTS: During a median follow-up of 4.9 years, 240 patients died from liver disease or underwent liver transplantation (LT), and 114 patients died from nonliver causes. Outcomes varied across the spectrum of ALD: The cumulative incidence of liver-related death or LT in people with decompensated cirrhosis or alcohol-associated hepatitis was 47% and 40%, respectively, compared with 7.4% in patients without cirrhosis and 13% in compensated cirrhosis. MELD was the best predictor of outcomes: (area under the curve for mortality/LT at 1 year was 0.853), although MELD3.0 and the Child-Turcotte-Pugh score performed similarly. Risk of liver-related outcomes were tabulated for integer values of the MELD score. Risk estimates based on the MELD were well calibrated in an external cohort. DISCUSSION: These data illustrate the natural history of ALD and define the risks of outcomes based on the MELD score across the spectrum of disease.