Abstract
BACKGROUND: Traditionally part of chronic hepatitis B (CHB) patients with normal alanine aminotransferase (ALT) are recommended to antiviral therapy referring to liver biopsy. However, liver biopsy is an invasive method with various potential complications. A noninvasive model was established in the study to evaluate liver histology and to identify the need of antiviral therapy. METHODS: A total of 614 liver biopsied CHB patients with ALT less than upper limit of normal from 2 centers were retrospectively analyzed. They were divided into a training cohort and a validation cohort. A noninvasive model to predict the significant liver histological changes was established and validated. RESULTS: The results of analysis showed that ALT, Age, platelet (PLT) and liver stiffness (LS) were independent risk factors for significant liver injury. The model was established based on the 4 indexes, with the area under the curve of 0.85 and 0.87 in training cohort and validation cohort. Meanwhile, 2 cut-off scores were selected. By applying the low cut-off score (- 0.207), patients without significant liver injury could be identified with high accuracy, with negative predictive value of 72.7% and 73.7% in training and validation cohorts. By applying the high cut-off score (0.537), the presence of significant liver injury could be diagnosed with high accuracy, with positive predictive value of 90.3% and 88.8% in the training and validation cohorts. By applying the model, liver biopsy would have been avoided in 87.6% (538/614) patients, with correct prediction in 87.9% (473/538). CONCLUSION: The novel noninvasive model composed of ALT, Age, PLT, LS can correctly assess liver histology in CHB patient with normal ALT, which helps to determine the need of antiviral therapy without liver biopsy.