A non-invasive decision tree predicting recurrence-free survival after liver transplantation for hepatocellular...

Liver transplantation (LT) provides an option to radically eliminate hepatocellular carcinoma (HCC) on the premise that recipients are thoroughly evaluated for posttransplant tumor recurrence risk before operation. This study aimed to optimize the Milan criteria (MC) by combining circulating prognostic markers, including interleukin-6 (IL-6) and alpha-fetoprotein (AFP).

We retrospectively enrolled 449 HCC cases receiving LT in three medical centers in China and divided them into the training cohort (n = 254) and the validation cohort (n = 195). Cox regression analysis was applied to identify recurrence-related risk factors based on patients' clinical and pathological characteristics, pretransplant plasma IL-6 and AFP levels. The collaborative prediction method was presented in the form of a decision tree. Kaplan-Meier analysis showed the predictive results of independent risk factors and the newly established prediction tree.

In the training cohort (n = 254), we established a predictive decision tree based on three independent risk factors: pretransplant plasma IL-6 (> 15 pg/mL), AFP (> 60 ng/mL), and the MC (beyond the MC), and named it as MIA (MC-IL-6-AFP) tree. According to the MIA tree, patients beyond the MC but with IL-6 ≤ 15 pg/mL and AFP ≤ 60 ng/mL (n = 39) had comparable 3-year recurrence-free survival (RFS) rates as patients within the MC (n = 121): 66.8% vs. 74.7% (P = 0.520). Likely, in the validation cohort (n = 195), the 3-year RFS rates of patients beyond the MC but with IL-6 ≤ 15 pg/mL and AFP ≤ 60 ng/mL (n = 26) were close to those of patients within the MC (n = 87): 71.8% vs. 76.6% (P = 0.660).

We proposed that pretransplant plasma IL-6 and AFP were valid prognostic biomarkers for HCC-related LT. The MIA tree could refine the MC by combining IL-6 and AFP and define an extra subset of eligible candidates without significant sacrifice in RFS.