Development of a nomogram for predicting 2-year native liver survival in biliary atresia using dynamic liver function indicators

To develop and validate a nomogram based on dynamic liver function indexes for predicting native liver survival (NLS) in children with biliary atresia (BA) at 2 years post-Kasai surgery, providing clinicians with a basis for individualized treatment decisions and optimizing early intervention strategies for high-risk children.

Children with type III BA were categorized by their 2-year NLS status. Univariate and multivariate logistic regression analyses were performed to identify predictors of NLS and to construct a nomogram model. Model performance was evaluated using internal bootstrap validation (1,000 resamples) and a training-test split (7:3), with discrimination assessed by the area under the receiver operating characteristic curve (AUC) and calibration by calibration curves.

A total of 134 children with type III BA were included. Univariate analysis identified significant associations between prognosis and the following: age at surgery, jaundice clearance failure, liver fibrosis stage, and 3-month postoperative levels of gamma-glutamyl transpeptidase (GGT), serum albumin (ALB), and aspartate aminotransferase and platelet ratio index (APRI) (all P < 0.05). Multivariate analysis established these independent predictors: liver fibrosis stage F4 (OR = 3.418, 95% CI: 1.745-6.695), APRI at 3 months (OR = 2.285, 95% CI: 1.175-4.445), age at surgery (OR = 1.773, 95% CI: 1.192-2.637), GGT at 3 months (OR = 1.942, 95% CI: 1.211-3.117), ALB at 3 months (OR = 0.948, 95% CI: 0.916-0.981), and jaundice clearance failure (OR = 2.437, 95% CI: 1.275-4.657). The resulting nomogram demonstrated stable performance across age subgroups (AUC = 0.926 for ≤60 days; AUC = 0.867 for >60 days). In the training set, the AUC was 0.872 (95% CI: 0.813-0.931), with sensitivity of 90.7% and specificity of 78.8%. The model showed excellent generalizability in the independent test set (AUC = 0.971).

This study developed and validated a nomogram integrating dynamic liver function indicators, effectively predicting 2-year NLS in children with BA. The model provides a reliable quantitative basis for individualized treatment decisions and early intervention, with strong clinical applicability, particularly in resource-limited settings. It offers a foundation for optimizing early intervention strategies for high-risk children.