Abstract
BACKGROUND/AIMS: Noninvasive indexes can be used to diagnose and stage liver fibrosis caused by chronic hepatitis B (CHB). We aimed to evaluate whether changes in the liver stiffness measurement (LSM) and serum biomarkers can predict liver fibrosis regression in CHB patients based on triple liver biopsies. METHODS: This multicenter cohort study was based on triple liver biopsies and lasted for 260 weeks. Liver fibrosis regression was defined as Ishak decreased ≥1 stage or predominantly regressive by P-I-R classification with stable Ishak stage. Twelve noninvasive models were validated externally and yielded area under the receiver operating characteristic curve (AUROC) values ≥0.700 for predicting significant fibrosis in the training set. RESULTS: A total of 175 CHB patients were included (median age: 38 years, 76.6% male). A total of 69.2% (117/169) and 79.6% (78/98) patients achieved liver fibrosis regression at week 78 and week 260, respectively. The mixed effects model revealed significant group×time interactions between the regression and non-regression groups for aminotransferase to platelet ratio index (APRI; p=0.041), new algorithm attributed to age, alanine aminotransferase, gamma-glutamyl transferase algorithm (p=0.022), and King's score (p=0.016) from baseline to week 78 as well as for APRI (p=0.046) from baseline to week 260. The AUROC values for model changes were all <0.750 for predicting liver fibrosis regression. Additionally, the changes in the LSM and most noninvasive models were significantly correlated with the changes of Ishak-histology activity index score. CONCLUSIONS: Changes in the LSM and noninvasive models were not strong predictors of liver fibrosis regression after 78 weeks and 260 weeks of treatment among CHB patients. It is critical to develop a dynamic noninvasive model for assessing liver fibrosis regression (ClinicalTrials.gov identifier NCT02849132).