A Nomogram to Predict Long-term Survival After Resection for Intrahepatic Cholangiocarcinoma

IMPORTANCE: Intrahepatic cholangiocarcinoma (ICC) is a primary cancer of the liver that is increasing in incidence, and the prognostic factors associated with outcome after surgery remain poorly defined. OBJECTIVE: To combine clinicopathologic variables associated with overall survival after resection of ICC into a prediction nomogram. DESIGN, SETTING, AND PARTICIPANTS: We performed an international multicenter study of 514 patients who underwent resection for ICC at 13 major hepatobiliary centers in the United States, Europe, and Asia from May 1, 1990, through December 31, 2011. Multivariate Cox proportional hazards regression modeling with backward selection using the Akaike information criteria was used to select variables for construction of the nomogram. Discrimination and calibration were performed using Kaplan-Meier curves and calibration plots. INTERVENTIONS: Surgical resection of ICC at a participating hospital. MAIN OUTCOMES AND MEASURES: Long-term survival, effect of potential prognostic factors, and performance of proposed nomogram. RESULTS: Median patient age was 59.2 years, and 53.1% of the patients were male. Most patients (74.7%) had a solitary tumor, and median tumor size was 6.0 cm. Patients were treated with an extended hepatectomy (202 39.3%), a hemihepatectomy (180 35.0%), or a minor liver resection (<3 segments) (132 25.7%). Most patients underwent R0 resection (87.9%), and 35.7% of patients had N1 disease. Using the backward selection of clinically relevant variables, we found that age at diagnosis (hazard ratio HR, 1.31; P < .001), tumor size (HR, 1.50; P < .001), multiple tumors (HR, 1.58; P < .001), cirrhosis (HR, 1.51; P = .08), lymph node metastasis (HR, 1.78; P = .01), and macrovascular invasion (HR, 2.10; P < .001) were selected as factors predictive of survival. On the basis of these factors, a nomogram was created to predict survival of ICC after resection. Discrimination using Kaplan-Meier curves, calibration curves, and bootstrap cross-validation revealed good predictive abilities (C index, 0.692). CONCLUSIONS AND RELEVANCE: On the basis of an Eastern and Western experience, a nomogram was developed to predict overall survival after resection for ICC. Validation revealed good discrimination and calibration, suggesting clinical utility to improve individualized predictions of survival for patients undergoing resection of ICC.