Non-invasive models for predicting histology in patients with chronic hepatitis B

Abstract

Background and aim: In contrast to chronic hepatitis C (CHC), few studies had been performed in assessing non-invasive models for predicting significant fibrosis or cirrhosis in chronic hepatitis B (CHB) patients. We aimed to evaluate non-invasive markers for diagnosing significant fibrosis/cirrhosis in patients with CHB, and to evaluate accuracy of models from CHC in CHB patients.

Patients and methods: Liver biopsies from consecutive treatment-naïve CHB patients were evaluated histologically by a pathologist blindly, using the Ishak score. Patients were divided randomly into a training (65%) and a validation sets (35%). Markers of fibrosis were evaluated by univariate followed by multivariate analysis in the training set. Area under receiver operating characteristics curve (AUROC) was assessed and validated in the training set. AUROC of aspartate aminotransferase (AST), AST/alanine aminotransferase (ALT) ratio, and AST-platelets ratio index (APRI) (derived from studies from CHC) in diagnosing significant fibrosis/cirrhosis were also assessed.

Results: Two-hundred and eighteen CHB patients were evaluated: 83% male, 86% Chinese, 47% having significant fibrosis, 19% having cirrhosis. Platelets were the only factor significantly associated with significant fibrosis and cirrhosis at multivariate analysis but the AUROC was only modest at 0.63 and 0.73, respectively. Models derived from studies from CHC were even less accurate.

Conclusion: Models with non-invasive markers in predicting histology from CHC patients were unsuitable for CHB patients. No variables consisting of simple and readily available markers were able to predict cirrhosis accurately in patients with CHB.