Predictive tool for evident histological liver injury in chronic hepatitis B patients: Development and validation.

BACKGROUND

Chronic hepatitis B (CHB) is a leading cause of liver-related mortality, progressing to fibrosis, cirrhosis, and hepatocellular carcinoma. Existing noninvasive tools (e.g., aspartate aminotransferase to platelet ratio index, fibrosis-4 index, liver stiffness measurement) and invasive liver biopsy have limitations in assessing evident histological liver injury (EHLI), highlighting the need for novel predictive models.

AIM: To develop and validate a predictive model for EHLI in CHB patients using a cohort from Hunan Province, China, to facilitate early risk identification and optimize resource allocation.

METHODS

This observational real-world study enrolled 223 CHB patients (August 2020 to March 2022) from the Second Xiangya Hospital, divided into development (n = 159) and validation (n = 64) cohorts (7:3 ratio). EHLI was defined as Ishak fibrosis stage ≥ 3 and/or histologic activity index ≥ 9. Variables were screened via univariable logistic regression and least absolute shrinkage and selection operator regression, and a multivariable logistic regression model and nomogram were constructed. Performance was evaluated using area under the curve (AUC), calibration plots, Hosmer-Lemeshow test, and decision curve analysis (DCA). Gene expression profiles were analyzed to identify immune-related pathways.

RESULTS

L59, platelet count (PLT), alanine transaminase (ALT), and aspartate transaminase (AST) were identified as independent predictors of EHLI. The model showed high discriminative ability, with AUC of 0.921 [95% confidence interval (CI): 0.880-0.963] in the development cohort and 0.959 (95%CI: 0.910-1.0) in the validation cohort, demonstrating a 20%-32% relative improvement in AUC over conventional noninvasive scores. Calibration plots demonstrated good agreement between predicted and observed EHLI, and DCA confirmed clinical utility (threshold probabilities: 20%-80%). Transcriptomic analysis identified 210 differentially expressed genes, with hub genes (e.g., COL1A2) and transforming growth factor-β/Smad pathway involvement linked to liver injury.

CONCLUSION

A novel nomogram incorporating L59, PLT, ALT, and AST robustly predicts EHLI in CHB patients. This model, using routinely measured variables, aids clinical decision-making and optimizes resource allocation.