Comparison of MELD and Child-Pugh scores to predict survival after chemoembolization for hepatocellular carcinoma

PURPOSE: To compare the value of the Child-Pugh and Model for End-stage Liver Disease (MELD) scores to predict patient survival rates after transarterial chemoembolization (TACE) for hepatocellular carcinoma (HCC). MATERIALS AND METHODS: Eighty-seven patients underwent 169 TACE sessions. Child-Pugh and MELD values were calculated before initial treatment. Survival length was tracked from the date of the first TACE procedure. Transplant recipients were censored from the study at the time of surgery. Child-Pugh and MELD scores as well as bilirubin and albumin levels and International Normalized Ratio were placed in high and low categories defined by their respective medians. Patient survival was compared at 3 months, 6 months, 12 months, and 24 months, and patterns were tested with chi(2) or Fisher exact tests. Survival over the entire period was examined with Kaplan-Meier analysis and differences were tested with log-rank tests. RESULTS: Mean and median survival times for all patients were 24 and 17 months, respectively. Sixteen patients were censored for transplantation at a mean of 12.9 months. MELD and Child-Pugh scores correlated well with each other (r = 0.68). Child-Pugh score (r = -0.35, P = .04) correlated more strongly with 12-month survival than did MELD score (r = -0.26, P = .12). After high/low score category division, a significantly greater survival difference was predicted by Child-Pugh score (27.2 months vs 10.3 months; P = .03) versus MELD score (27.5 months vs 15.8 months; P = .19). An albumin level greater than 3.4 g/dL was also associated with significantly improved survival (29.3 months vs 10.1 months; P = .0032). Survival differences between high-risk and low-risk groups at the 3-, 6-, 12-, and 24-month intervals were significant for low Child-Pugh scores and for albumin levels greater than 3.4 g/dL. Statistical significance was not approached at any of the time lengths with MELD scores. CONCLUSIONS: Child-Pugh score correlates better than MELD score to overall patient survival and is a better predictor than MELD score of survival at specific time points. Of the components of the Child-Pugh and MELD systems, albumin level is the most useful predictor of survival.