Development and internal validation of a nomogram for predicting significant liver fibrosis in primary biliary cholangitis: a multicenter retrospective study

Accurate non-invasive staging of significant liver fibrosis is vital for managing primary biliary cholangitis (PBC). Existing serum models often lack accuracy in this cholestatic disease. This study aimed to develop and validate a disease-specific nomogram for diagnosing significant fibrosis in PBC. In this multicenter retrospective study, 118 PBC patients who underwent liver biopsy were enrolled and stratified by fibrosis stage (significant fibrosis: Scheuer ≥S2, n = 69; non-significant fibrosis, n = 49). All enrolled patients were treatment-naïve to ursodeoxycholic acid (UDCA) prior to liver biopsy and baseline blood collection. Independent predictors were identified using least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression to construct a nomogram. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC), calibration curves, decision curve analysis (DCA), and clinical impact curves (CIC). Internal validation was performed via bootstrap resampling and 5-fold cross-validation. Performance was compared against aspartate aminotransferase-to-platelet ratio index (APRI), fibrosis-4 (FIB-4), AST/ALT ratio (AAR), and gamma-glutamyl transpeptidase-to-platelet ratio (GPR). Platelet count, albumin, gamma-glutamyl transferase, and immunoglobulin G were integrated into the nomogram. The model yielded an AUROC of 0.885 (95% CI: 0.827–0.942) with good calibration (Hosmer-Lemeshow test, P = 0.9485). Internal validation confirmed robustness (bootstrap AUROC = 0.869). At the optimal cutoff, the nomogram achieved 73.9% sensitivity and 87.8% specificity, significantly outperforming all conventional indices (all P < 0.05, DeLong’s test). Based on routine laboratory parameters , this nomogram provides an accurate tool for non-invasive risk stratification of significant fibrosis in PBC, demonstrating superior diagnostic performance compared to traditional serum models.