Objectives: European liver transplant (LT) centers have moved away from using the Milan Criteria (MC) for hepatocellular carcinoma (HCC) patient selection, turning to models including tumor biological indices, namely alpha-fetoprotein (AFP). We present the first US model to incorporate an AFP response (AFP-R), with comparisons to MC and French-AFP models (F-AFP).
Methods: AFP-R was measured as differences between maximum and final pre-LT AFP in HCC patients undergoing LT at 3 US centers (2001 to 2013). Cox and competing risk-regression analyses identified predictors of recurrence-free survival (RFS).
Results: Of 1450 patients, 235 (16.2%) were outside MC. Tumor size, number, and AFP-R were independent predictors of RFS and were assigned weighted points based on Cox-regression analysis. An AFP-R consistently < 200 ng/mL predicted the best outcome. A 3-tiered competing-risk RFS model, the New York/California (NYCA) score, was developed, accurately discriminating between groups (P < 0.001), and correlating with overall survival (P < 0.001). Two hundred one of 235 patients outside MC (85.5%) would be recategorized into NYCA low/acceptable-risk groups. The c-statistic for our NYCA score is 0.731 compared with 0.613 for MC and 0.658 for F-AFP (P < 0.0001).
Conclusion: Incorporation of AFP-R into HCC selection criteria allows for MC expansion. As United Network for Organ Sharing considers adding AFP to selection algorithms, the NYCA score provides an objective, user-friendly tool for centers to appropriately risk-stratify patients.