OBJECTIVE To develop and validate a model to predict tumor recurrence after living donor liver transplantation (LDLT) (MoRAL) for hepatocellular carcinoma (HCC) beyond the Milan criteria (MC).

BACKGROUND Some subgroups of HCC exceeding the MC experience substantial benefit from LDLT.

METHODS This multicenter study included a total of 566 consecutive patients who underwent LDLT in Korea: the beyond-MC cohort (n = 205, the derivation [n = 92] and validation [n = 113] sets) and the within-MC cohort (n = 361). The primary endpoint was time-to-recurrence.

RESULTS Using multivariate Cox proportional hazard model, we derived the MoRAL score using serum levels of protein induced by vitamin K absence-II and alpha-fetoprotein, which provided a good discriminant function on time-to-recurrence (concordance index = 0.88). Concordance index was maintained similarly on both internal and external validations (mean 0.87 and 0.84, respectively). At cut off of 314.8 (75th percentile value), a low MoRAL score (≤314.8) was associated with significantly longer recurrence-free (versus > 314.8, HR = 5.29, P < 0.001) and overall survivals (HR = 2.59, P = 0.001) in the beyond-MC cohort. The 5-year recurrence-free and overall survival rates of beyond-MC patients with a low MoRAL score were as high as 66.3% and 82.6%, respectively. The within-MC patients with a high MoRAL score showed a higher risk of recurrence than beyond-MC patients with a low MoRAL score (HR = 2.56, P = 0.035). The MoRAL score was significantly correlated with explant histology.

CONCLUSIONS This new model using protein induced by vitamin K absence-II and alpha-fetoprotein provides refined prognostication. Among beyond-MC HCC patients, those with a MoRAL score ≤314.8 and without extrahepatic metastasis might be potential candidates for LDLT.