A MODEL TO PREDICT SURVIVAL IN PATIENTS WITH CIRRHOSIS AND HEPATOCELLULAR CARCINOMA TREATED WITH RESECTIVE SURGERY

ntroduction: Which patients with cirrhosis and hepatocellular carcinoma (C-HCC) should undergo surgical resection of the tu- mour is debated, partly because predictive models allowing easy and accurate prediction of survival in this subgroup of patients are lacking. In the present study, our aims were to identify the main factors associated with survival after resective surgery, and to provide clinicians with a simple tool enabling them to estimate survival probabilities of these patients. Methods: One hundred-forty-eight consecutive patients with C- HCC who underwent resective surgery in two Italian Centres from January 1994 through December 2002 were investigated ret- rospectively. Seven patients (4.7%) died within one month from the operation; the remaining 141 patients were followed for a mean of 31 months. Survival probabilities from the time of diag- nosis were calculated by the Kaplan-Meier method. Patient-re- lated variables (gender, age, alcohol consumption), cirrhosis- related variables (HBsAg,anti-HCV Ab, prothrombin time, serum sodium, blood urea nitrogen, gamma glutamyl transpeptidase, albumin, bilirubin), tumour-related variables (tumour size, uni- nodular or multinodular tumour, serum alpha-fetoprotein (AFP), histological grading of the tumour according to the Edmonson’s score, toxic syndrome), and three prognostic scores (Child-Pugh’s stage, Okuda’s stage, CLIP score) were evaluated by univariate and multivariate analysis with Cox proportional-hazards regres- sion. Results: The percentages of patients in the series surviving 1, 3 and 5 years after surgery were 65%, 40% and 23%,respectively. At multivariate analysis, among the 141 patients who survived the pen-operative period, the variables independently associated with survival were serum albumin, serum AFP,Edmonson’s score and presence of the toxic syndrome (fever >38”C, body weight loss >lo%, asthenia), allowing to estimate individual survival probabilities (R) according to the following formula: R = 1.024 * toxic syndrome + 0.1602 * log (AFP) - 2.0325 * log (albumin) + 0.2894 * Edmonson’s score. Survival probabilities three years after surgery were explored by means of a logistic model (area under ROC curve = 0.7’7)and graphically expressed in a nomogram (Figure). Conclusions: The prognosis of patients with cirrhosis undergoing resective surgery for hepatocellular carcinoma depends both on cirrhosis-related and tumor-related variables, and can be pre- dicted by a simple four-variable model.