Survival simulation of hepatocellular carcinoma derived from follow-up studies of 450 patients.

A simulation model to predict the survival probability of individual patients with hepatocellular carcinoma (HCC) after therapy was derived from the results of various therapies and follow-up studies of 450 HCC patients. Twenty-two prognostically important variables were analyzed by Cox's proportional hazards model. The 9 significant variables that were extracted were used to build the simulation. In this model, S(t), the expected estimated survival rate for individual patient at time t (month), is calculated by the following equation: S(t) = (exp (-0.03655t) (exp [0.9479 ([portal vein invasion]-0.222) + 0.3846 ([tumor number]-2.00) + 0.2578 ([tumor size]-3.231) + 0.0742 ([loge AFP]-5.647) + 0.8184 ([metastasis]-0.036) + 0.2810 ([Child's class]-1.689)-0.7088 ([transcatheter arterial embolization]-0.578)-0.9746 ([percutaneous ethanol injection]-0.153)-0.5377 ([hepatectomy]-0.109)]) The validity of the model was assessed using a split-sample technique. This paper does not discuss the superiority or inferiority of the therapies, because some selection bias for prognostic factors among the therapies can not be completely excluded. But this model is proposed as a practical model to predict the survival of patients with HCC.