Peer-reviewed article

Construction and validation of a prediction model for graft failure in pediatric liver transplant recipients based on the APRI-ALBI score: a retrospective study of 289 cases.

Quanhai Zhang, Haifeng Tang, Xiaoke Dai, Mingman Zhang · 8 May 2026

Abstract

OBJECTIVE: • • •. BACKGROUND: To develop and validate a preoperative APRI-ALBI-based nomogram for predicting graft failure (GF) after pediatric liver transplantation. METHODS: A single-center, retrospective cohort of 289 children who underwent primary liver transplantation for end-stage liver disease between April 2017 and September 2024 was enrolled. Forty-two recipient, donor and intra-operative variables were collected. Candidate predictors were first screened by univariate Logistic/Cox analysis, further reduced by LASSO regression with 10-fold cross-validation, and finally entered into multivariable models. A nomogram integrating APRI-ALBI and other independent risk factors was constructed. Model discrimination, calibration and clinical utility were assessed with ROC analysis, calibration plots, Hosmer-Lemeshow test, decision-curve analysis (DCA) and restricted cubic splines (RCS). Kaplan-Meier curves and time-dependent ROC were used to validate 1- and 3-year GF prediction. RESULTS: Multivariable Logistic/Cox analysis identified six independent predictors of GF: MELD score, male donor, APRI-ALBI score, donor height, operative time and intra-operative blood loss. The resulting nomogram achieved an AUC of 0.706, outperforming PELD, MELD and Child-Pugh scores, with good calibration (Hosmer-Lemeshow p > 0.05) and favorable net benefit on DCA. RCS revealed a dose-dependent increase in GF risk once APRI-ALBI exceeded 10. Kaplan-Meier analysis showed significantly lower GF-free survival in the high APRI-ALBI group (p < 0.01). Time-dependent ROC curves demonstrated excellent predictive accuracy at 30 days and 1 year (AUC 0.899-0.977). CONCLUSION: APRI-ALBI is an independent risk factor for GF in pediatric liver transplantation. The APRI-ALBI-based nomogram offers robust discrimination, calibration and clinical utility for individual risk stratification.