Nomograms for prediction of long-term survival in elderly patients after partial hepatectomy for hepatocellular carcinoma

Background. Partial hepatectomy is an important treatment for elderly patients with hepatocellular carcinoma. However, prediction of long-term outcomes of an individual elderly patient after partial hepatectomy still is lacking. This study aimed to develop 2 nomograms to pre- or postoperatively predict overall survival for these patients. Methods. Of the 528 elderly patients (aged 65 years) who underwent partial hepatectomy for hepatocellular carcinoma at the Eastern Hepatobiliary Surgery Hospital between 2008 and 2011, 425 patients served as a training cohort to develop pre- and postoperative nomograms, and the remaining 103 patients comprised a validation cohort. The Cox proportional hazards model was used for univariate and multivariable analyses of tumor recurrence and overall survival. Discrimination and calibration of the models were measured using the concordance index, calibration plots, and Kaplan Meier curves. Results. Based on preoperative data, the independent risk factors of overall survival were age 75 years, Charlson score, alpha-fetoprotein >= 20 mu g/L, hepatitis B virus-deoxyribonucleic acid >= 10(4) IU/mL, and tumor diameter. Based on postoperative data, nonanatomic hepatectomy, absence of tumor encapsulation, and presence of microvascular invasion were additional independent risk factors. These independent predictors were incorporated into the pre- and postoperative nomograms, respectively. The concordance indexes of the 2 nomograms for overall survival prediction were 0.70 (95 % confidence interval, 0.67-0.74) and 0.72 (0.69-0.78), respectively. Both nomograms accurately predicted 1-, 3-, and 5-year overall survival probability, and their predictive performances were optimally validated. Conclusion. The proposed 2 nomograms showed good individualized predictive performance in elderly patients with hepatocellular carcinoma before and after partial hepatectomy.