α-Fetoprotein model versus Milan criteria in predicting outcomes after hepatic resection for hepatocellular carcinoma: multicentre study.

Background: The Milan criteria and the French α-fetoprotein (AFP) model have both been validated for predicting outcomes after liver transplantation for hepatocellular carcinoma, with the Milan criteria also used for predicting outcomes after hepatic resection. The aim of this study was to evaluate the AFP model's predictive value for recurrence and survival following hepatocellular carcinoma resection and compare its performance with that of the Milan criteria.

Methods: Data for patients who underwent hepatocellular carcinoma resection between 2002 and 2021 were analysed. For both the AFP model and Milan criteria, patients were divided into two groups: those with hepatocellular carcinoma within and beyond the AFP model (scores ≤ 2 and > 2 points, respectively) and the Milan criteria. Cumulative recurrence and overall survival rates were compared between patients within and beyond the AFP model. Predictions of recurrence and overall survival by the AFP model and Milan criteria were compared using net reclassification improvement and area under the receiver operating characteristic curve analyses.

Results: Among 1968 patients evaluated, 1058 (53.8%) and 940 (47.8%) were classified as beyond on the AFP model and Milan criteria, respectively. After controlling for competing factors on multivariable analyses, being beyond the AFP model was independently associated with recurrence and worse overall survival after resection of hepatocellular carcinoma. Time-dependent net reclassification improvement and area under the receiver operating characteristic curve analyses demonstrated that the AFP model was superior to the Milan criteria in predicting recurrence. Of note, patients who were classified as beyond both the Milan criteria and AFP model had an even higher risk of postoperative recurrence and mortality (hazard ratios 1.51 and 1.47, respectively).

Conclusion: The French AFP model demonstrated superior prognostic accuracy to the Milan criteria in predicting recurrence and survival after hepatocellular carcinoma resection. The AFP model not only effectively stratified patient risk but also identified a subgroup of high-risk patients among those beyond the Milan criteria.