Development of a Risk Classifier to Predict Tumor Recurrence and Lenvatinib Benefits in Hepatocellular Carcinoma After Liver Transplantation

Abstract

Current selection tools were not precise enough to predict recurrence of hepatocellular carcinoma (HCC) and benefit of adjuvant lenvatinib for patients who received liver transplant (LT) for HCC. Thus, we aim at developing a risk classifier to predict recurrence of HCC and benefit of adjuvant Lenvatinib for those who underwent LT for HCC. Cox regression model was applied to selected predictors and created the final model in a training cohort of 287 patients who underwent LT for HCC, which was tested in an internal validation cohort of 72 patients by using C-statistic and net classification index (NRI) compared with the following HCC selection criteria: the Milan criteria, the Up-to-7 criteria, and the University of California, San Francisco criteria. We built a Risk Classifier of South China Cohort (RCOSC) based on 4 variables: the maximum diameter plus number of viable tumors, alpha-fetoprotein, microvascular invasion, and highest alanine aminotransferase in 7 days after LT. In validation analyses, our RCOSC showed good predictive performance (C-statistic, 0.866; 95% confidence interval [CI], 0.833-0.899) and had better prognostic value than Milan criteria (NRI, 0.406; P < .001), University of California, San Francisco (NRI, 0.497; P < .001), and Up-to-7 (NRI, 0.527; P < .001). By applying the RCOSC, we were able to accurately categorize patients into high-risk and low-risk groups. Further survival analysis revealed that the patients in the high-risk group might have a better therapeutic response to preventive regimen of lenvatinib after LT for HCC (hazard ratio, 0.38; 95% CI, 0.161-0.871, P=.018). Our RCOSC presented favorable predictive performance for HCC recurrence. It might be capable of sifting out patients who benefit from adjuvant therapy after LT for HCC, providing a reliable tool for precise clinical decision-making of patients with HCC with LT.