A prognostic model for patients with hepatocellular carcinoma within the Milan criteria undergoing non-transplant therapies, based on 1106 patients.

Background: The Milan criteria are used to select candidates with small hepatocellular carcinoma (HCC) for liver transplantation. Due to severe shortage of donors, majority of patients within the Milan criteria need to seek alternative treatments.
Aim: To propose a prognostic model for these patients undergoing non-transplant therapies.
Methods: A total of 1106 HCC patients, who were within the Milan criteria and received non-transplant therapies were retrospectively analysed. Patients were randomly assigned to the derivation and validation set according to treatments. A prognostic model was constructed from independent predictors of survival identified in the multivariate Cox model of the derivation set and was confirmed in the validation set.
Results: In the Cox model, serum bilirubin ≥1.5 mg/dL [risk ratio (RR): 1.525, P = 0.016], α-fetoprotein (AFP) ≥100 ng/mL (RR: 1.728, P < 0.001), mild ascites (RR: 1.705, P = 0.025) and moderate/severe ascites (RR: 4.163, P < 0.001) were independent predictors of poor survival in the derivation set (n = 553). A prognostic model with a total of 0-4 points was derived with the sum of three variables: 1 point each for bilirubin ≥1.5 mg/dL, AFP ≥100 ng/mL and mild ascites, and 2 points for moderate/severe ascites. This scoring system accurately predicted the survival in the validation set (n = 553; P < 0.001). The model consistently discriminated the survival in patients stratified by curative and noncurative treatments (both P values <0.001).
Conclusion: The newly proposed prognostic scoring model, based on serum bilirubin and AFP level, and severity of ascites, is informative to predict the survival in non-transplant HCC patients within the Milan criteria.
(© 2012 Blackwell Publishing Ltd.)