Performance of eight predictive models for hepatocellular carcinoma recurrence after liver transplantation: A comparative study

Abstract

Introduction and ObjectivesLiver transplantation is the optimal treatment for patients with early hepatocellular carcinoma and cirrhosis. However, hepatocellular carcinoma recurs in approximately 15 % of individuals. This study aimed to assess the efficacy of predictive models for hepatocellular carcinoma recurrence after liver transplantation. Patients and MethodsThis retrospective study included 381 patients with HCC and evaluated the performance of the following models: R3-AFP score, alpha-fetoprotein (AFP) model, University of California, Los Angeles (UCLA) nomogram, Pre-Model of Recurrence after Liver Transplantation (MORAL), Post-MORAL, and Combo MORAL models, Risk Estimation of Tumor Recurrence (RETREAT) model and Platelet to Lymphocyte Ratio (PLR) model. ResultsThe R3-AFP score, UCLA nomogram, AFP model, RETREAT, Combo MORAL, and Post-MORAL models exhibited comparable AUROCs, ranging from 0.785 to 0.733. The AUROCs for the R3-AFP model and AFP model were superior to those of the Pre-MORAL and PLR models. The UCLA nomogram, RETREAT score, Combo MORAL model, and Post-MORAL model performed similarly to the first two models, but were only superior to the PLR model. ConclusionsThe R3-AFP model, UCLA nomogram, AFP model, RETREAT, Combo MORAL, and Post-MORAL models demonstrated a moderate predictive capacity for hepatocellular carcinoma recurrence following transplantation. No significant differences were observed among these models in their ability to predict recurrence.