Background/aims: Noninvasive markers for predicting significant fibrosis and inflammation have not yet been validated in an unselected group of chronic hepatitis B virus (HBV) carriers. The aim of this study was to create noninvasive models to predict significant fibrosis and inflammation in chronic HBV carriers.
Methods: A total of 276 (229 HBeAg negative, 47 HBeAg positive) unselected consecutive treatment naïve patients chronically infected with HBV who attended our center over a 36-month period underwent liver biopsy. HBeAg negative patients were randomly divided into two cohorts: training group (N = 130) and validation group (N = 99). HBeAg positive patients were analyzed as a whole without separation. Thirteen parameters were analyzed separately in HBeAg negative and HBeAg positive patients to predict significant fibrosis (Ishak stage >or=3) and inflammation (Ishak grade >or=7).
Results: In HBeAg negative patients significant liver fibrosis was best predicted using the variables HBV DNA levels, alkaline phosphatase, albumin, and platelet counts with an area under ROC curve (AUC) of 0.91 for the training group and 0.85 for the validation group. Using the low cutoff probability of 4.72, significant fibrosis could be excluded with negative predictive value of 99% in the entire cohort, and liver biopsy would have been avoided in 52% of patients. The best model for predicting significant inflammation included the variables age, HBV DNA levels, AST, and albumin with an AUC of 0.93 in the training and 0.82 in the validation group. In HBeAg positive patients no factor could predict accurately stages of liver fibrosis, but the best factor for predicting significant inflammation was AST with an AUC of 0.87.
Conclusions: Significant hepatic fibrosis and necroinflammation can reliably be predicted using routinely checked tests and HBV DNA levels.