Background: A major problem in assessing the likelihood of survival of patients with hepatocellular carcinoma (HCC) arises from a lack of models capable of predicting outcome accurately.
Aim: To compare the ability of the Italian score (CLIP), the French classification (GRETCH) and the Barcelona (BCLC) staging system in predicting survival in patients with HCC.
Methods: We included 406 consecutive patients with cirrhosis and HCC. Seventy-eight per cent of patients had hepatitis C. Independent predictors of survival were identified using the Cox model.
Results: One-hundred and seventy-eight patients were treated, while 228 were untreated. The observed mortality was 60.1% in treated patients and 84.9% in untreated patients. Among treated patients, albumin, bilirubin and performance status were the only independent variables significantly associated with survival. Mortality was independently predicted by bilirubin, alpha-fetoprotein and portal vein thrombosis in untreated patients. CLIP achieved the best discriminative capacity in the entire HCC cohort and in the advanced untreatable cases, while BCLC was the ablest in predicting survival in treated patients.
Conclusions: Overall predictive ability of BCLC, CLIP and GRETCH staging systems was not satisfactory, and was not uniform for treated patients and untreated patients. None of the scoring systems provided confident prediction of survival in individual patients.