Recurrent hepatocellular carcinoma after transplantation: use of a pathological score on explanted livers to predict recurrence.

Milan and University of California at San Francisco (UCSF) criteria are used to select patients with hepatocellular carcinoma (HCC) for liver transplantation (LT). Recurrent HCC is a significant cause of death. There is no widely accepted pathological assessment strategy to predict recurrent HCC after transplantation. This study compares the pathology of patients meeting Milan and UCSF criteria and develops a pathological score and nomogram to assess the risk of recurrent HCC after transplantation. All explanted livers with HCC from our center over the 18-yr period 1985 to 2003 were assessed for multiple pathological features and relevant clinical data were recorded; multivariate analysis was performed to determine features associated with recurrent HCC. Using pathological variables that independently predicted recurrent HCC, a pathological score and nomogram were developed to determine the probability of recurrent HCC. Of 75 cases analyzed, 50 (67%) met Milan criteria, 9 (12%) met only UCSF criteria and 16 (21%) met neither criteria based on explant pathology. There were 20 cases of recurrent HCC and the mean follow-up was 8 yr. Recurrent HCC was more common (67 vs. 12%; P < 0.001) and survival was lower (15 vs. 83% at 5 yr; 15 vs. 55% at 8 yr; P < 0.001) with those who met only UCSF criteria, compared to those who met Milan criteria. Cryptogenic cirrhosis (25 vs. 5%; P = 0.015), preoperative AFP >1,000 ng/mL (20 vs. 0%; P < 0.001) and postoperative OKT3 use (40 vs. 15%; P = 0.017) were more common among patients with recurrent HCC. While microvascular invasion was the strongest pathological predictor of recurrent HCC, tumor size >or=3 cm (P = 0.004; odds ratio [OR] = 7.42), nuclear grade (P = 0.044; OR = 3.25), microsatellitosis (P = 0.020; OR = 4.82), and giant/bizarre cells (P = 0.028; OR = 4.78) also predicted recurrent HCC independently from vascular invasion. The score and nomogram stratified the risk of recurrent HCC into 3 tiers: low (<5%), intermediate (40-65%), and high (>95%). In conclusion, compared to patients meeting Milan criteria, patients who meet only UCSF criteria have a worse survival and an increased rate of recurrent HCC with long-term follow-up, as well as more frequent occurrence of adverse histopathological features, such as microvascular invasion. Application of a pathological score and nomogram could help identify patients at increased risk for tumor recurrence, who may benefit from increased surveillance or adjuvant therapy.