AFP Model Predicts Outcomes After Liver Resection: A Multicenter Retrospective Cohort Study

The AFP model, originally developed to refine biological selection for liver transplantation, integrates tumor size, number, and serum AFP level. It was shown to be superior to the Milan criteria for predicting hepatocellular carcinoma (HCC) after liver transplantation. Although widely validated in the transplant setting, its prognostic value after hepatic resection for HCC remains insufficiently defined. This study aimed to assess the performance of the AFP score in predicting recurrence and survival after curative hepatectomy in a contemporary French multicenter cohort.

We retrospectively analyzed consecutive adults undergoing curative-intent liver resection for pathologically confirmed HCC across five high-volume French hepatopancreatobiliary centers between 2012 and 2021. Patients were classified according to AFP score (≤ 2 vs. > 2). Survival outcomes were estimated by using Kaplan-Meier methods, and prognostic factors were evaluated by using univariable and multivariable Cox regression analyses.

Among 424 patients, median overall survival (OS) was 112.8 months, and median recurrence-free survival (RFS) was 37.1 months. In univariable analysis, the AFP score was significantly associated with OS (hazard ratio [HR] 1.22, p = 0.001) and RFS (HR 1.25, p < 0.001). In multivariable analysis, the AFP score remained independently predictive of recurrence (HR 1.18, p = 0.004). Patients with AFP score > 2 had significantly lower 24-month RFS compared with those with a score ≤ 2 (log-rank p = 0.0005).

In this large French multicenter cohort, the AFP model, originally validated in transplantation, retains strong prognostic value after hepatectomy, independently predicting recurrence. Integrating the AFP model into preoperative assessment may improve biological risk stratification and guide postoperative surveillance and future neoadjuvant and adjuvant strategies.