The Enhanced Liver Fibrosis test maintains its diagnostic and prognostic performance in alcohol-related liver disease: a cohort study

Abstract

Abstract Background Alcohol is the main cause of chronic liver disease. The Enhanced Liver Fibrosis (ELF) test is a serological biomarker for fibrosis staging in chronic liver disease, however its utility in alcohol-related liver disease warrants further validation. We assessed the diagnostic and prognostic performance of ELF in alcohol-related liver disease. Methods Observational cohort study assessing paired ELF and histology from 786 tertiary care patients with chronic liver disease due to alcohol (n = 81) and non-alcohol aetiologies (n = 705). Prognostic data were available for 64 alcohol patients for a median of 6.4 years. Multiple ELF cut-offs were assessed to determine diagnostic utility in moderate fibrosis and cirrhosis. Survival data were assessed to determine the ability of ELF to predict liver related events and all-cause mortality. Results ELF identified cirrhosis and moderate fibrosis in alcohol-related liver disease independently of aminotransferase levels with areas under receiver operating characteristic curves of 0.895 (95% CI 0.823–0.968) and 0.923 (95% CI 0.866–0.981) respectively, which were non-inferior to non-alcohol aetiologies. The overall performance of ELF was assessed using the Obuchowski method: in alcohol = 0.934 (95% CI 0.908–0.960); non-alcohol = 0.907 (95% CI 0.895–0.919). Using ELF < 9.8 to exclude and ≧ 10.5 to diagnose cirrhosis, 87.7% of alcohol cases could have avoided biopsy, with sensitivity of 91% and specificity of 85%. A one-unit increase in ELF was associated with a 2.6 (95% CI 1.55–4.31, p < 0.001) fold greater odds of cirrhosis at baseline and 2.0-fold greater risk of a liver related event within 6 years (95% CI 1.39–2.99, p < 0.001). Conclusions ELF accurately stages liver fibrosis independently of transaminase elevations as a marker of inflammation and has superior prognostic performance to biopsy in alcohol-related liver disease.