Background: This study was designed to validate the α-fetoprotein model for predicting recurrence after liver transplantation in Korean hepatocellular carcinoma patients.
Methods: Patients who underwent liver transplantation for hepatocellular carcinoma at Samsung Medical Center between 2007 and 2015 were included. Recurrence, overall survival, and disease-specific survival of patients divided by both the Milan criteria and the α-fetoprotein model were compared using Kaplan-Meier log-rank test. The predictability of the α-fetoprotein model compared with the Milan criteria was tested by means of net reclassification improvement analysis applied to patients with a follow-up of at least 2 years.
Results: A total of 400 patients were included in the study. Patients within Milan criteria had 5-year recurrence, and overall survival rates of 20.9% and 76.3%, respectively, compared with corresponding rates of 50.3% and 55.7%, respectively, for patients who were beyond Milan criteria. α-Fetoprotein model low-risk patients had 5-year recurrence and overall survival rates of 21.1% and 76.2%, respectively, compared with corresponding rates of 57.7% and 52.2%, respectively, in high-risk patients (P < 0.001, all). Although overall net reclassification improvements were statistically nonsignificant for recurrence (net reclassification improvements [NRI] = 1.7%, Z = 0.30, P = 0.7624), and overall survival (NRI = 9.0%, Z = 1.60, P = 0.1098), they were significantly better for predicting no recurrence (NRI = 6.6%, Z = 3.16, P = 0.0016) and no death. (NRI = 7.7%, Z = 3.65, P = 0.0003).
Conclusions: The α-fetoprotein model seems to be a promising tool for liver transplantation candidacy, but further investigation is needed.