Evaluation of Model for End Stage Liver Disease (MELD)-based systems as prognostic index for hepatocellular carcinoma

Background: The Cancer of Liver Italian Program (CLIP) and Japan Integrated Scoring System (JIS) used the Child-Turcotte-Pugh (CTP) score to evaluate the liver function. Aim: We aimed to evaluate the performance of Model for End Stage Liver Disease (MELD) based CLIP and JIS to predict the prognosis of hepatocellular carcinoma (HCC). Methods: Consecutive patients with HCC who presented to our Hepatoma Clinic from January 2003 to April 2005 were studied. MELD-based CLIP and JIS were generated by replacing the original CTP score with MELD score at three categories (< 10, 10-14 and > 14). Results: Among 471 HCC patients (85.1% males; aged 58.8 +/- 12.2 years), 73% had chronic hepatitis B, 37.4% had > 1 nodule, 84.1% had tumor size > 2 cm, 55.0% had Child's B cirrhosis, 12.7% underwent tumor resection and 20.6% received locoregional therapy. The cumulative survival at 3 and 6 months were 67% and 55%, respectively. For 3-month survival, the area under the receiver operating characteristic curves (AUC) of MELD-CLIP (0.69) and MELD-JIS ( 0.69) were superior to the original systems (0.64, P = 0.004 and 0.64, P = 0.0018, respectively). For 6-month survival, AUC of MELD-CLIP ( 0.64) and MELD-JIS (0.62) were also superior to the original systems (0.54, P = 0.003 and 0.59, P = 0.002, respectively). The MELD-based systems performed best among patients who received locoregional therapy to HCC. Advanced cirrhosis (hypoalbuminemia, hyperbilirubinemia, ascites, coagulopathy and elevated creatinine), and cancer (portal vein thrombosis, elevated alpha-fetoprotein, large and multiple tumors) were associated with higher mortality. Conclusions: MELD-based systems performed better than Child-Pugh based systems as prognostic indexes for HCC.