A novel system for predicting liver histopathology in patients with chronic hepatitis B

    loading  Checking for direct PDF access through Ovid


There is currently a lack of reliable, reproducible, and easily applied methods for assessing changes in liver histology in patients in the gray zone phase of chronic hepatitis B (CHB). Therefore, we aimed to develop a novel predictive scoring system to detect significant liver histological changes in these patients.

A total of 388 patients in the gray zone phase of CHB who underwent liver biopsy were divided into a training group and a validation group, and their clinical and routinely available laboratory parameters were analyzed using univariate analysis, Spearman correlation analysis, and logistic modeling. A novel scoring system, termed the Significant Histological Model (SHM), was constructed using logistic modeling. The diagnostic accuracy of our novel scoring system was evaluated by the receiving operating characteristic (ROC) method, sensitivity, specificity, and positive and negative predictive values (NPVs).

We established the novel SHM scoring system using serum aspartate transaminase (AST), platelet counts (PLTs), albumin (ALB), and hepatitis B virus (HBV) DNA (log10 IU/mL) levels. The area under the ROC curve of the SHM scoring system was 0.763 in the training group and 0.791 in the validation group. For patients with a score of −1.0 or less and no significant histological changes, the sensitivity was 78.9%, specificity was 51.5%, positive predictive value (PPV) was 46.4%, and NPV was 82.0%. In the validation set, the sensitivity, specificity, PPV, and NPV were 80.0%, 66.6%, 56.3%, and 86.2%, respectively.

This novel scoring system using AST, PLT, ALB, and HBV DNA (log10 IU/mL) levels identifies patients in the gray zone phase of CHB with and without histological changes with a high degree of accuracy. Here, we provide the experimental basis for the initiation of clinical antiviral treatment without the need for liver biopsy.

Related Topics

    loading  Loading Related Articles