To establish a Nomogram prognostic model for the cancer-specific survival (CSS) of patients with early intrahepatic cholangiocarcinoma (ICC) and to validate the predictive efficacy.

Clinical data of 943 patients with early ICC were retrieved from the Surveillance, Epidemiology, and End Results (SEER) database of American National Cancer Institute (NCI). According to the ratio of 7∶3, they were randomly divided into the training (n=663) and validation groups (n=280). In the training group, the independent risk factors were screened by univariate and multivariate Cox's regression analyses, anda Nomogram prognostic model was established based on these independent risk factors. The accuracy of this model was validated by C-index, receiver operating characteristic (ROC) curve and calibration curve. The clinical utility of the Nomogram model was assessed by decision curve analysis (DCA) and compared with the American Joint Committee on Cancer (AJCC) staging. According to the total score calculated by Nomogram model, all patients were subjected to the risk stratification. The effect of risk stratification was evaluated by Kaplan-Meier survival curve.

Cox's regression analysis showed that age, sex, marital status, tumor diameter, histological grading, surgery and radiotherapy were the independent influencing factors for the CSS of patients with early ICC (HR=1.364, 1.237, 0.555, 1.269, 1.350, 0.244, 0.587; P<0.05). Based on these independent risk factors, the Nomogram prediction model of CSS patients was established. In the training group, the C-index was 0.724, and 0.676 in the validation group. The area under the ROC curve (AUC) at 1, 3 and 5 years exceeded 0.7 in two groups. The calibration curve analysis demonstrated that the predictive results of Nomogram model were in good agreement. DCA analysis revealed that Nomogram model had high clinical utility. Compared with the AJCC staging, this model yielded higher accuracy and clinical application value. A risk stratified system was established, and all patients were divided into the high, middle and low risk groups. Kaplan-Meier survival curve showed that the 1-year CSS in the low, middle and high risk groups were 88.4%, 65.5% and 35.5%, 63.4%, 32.0% and 7.6% for the 3-year CSS, and 48.2%, 20.4% and 4.5% for the 5-year CSS, respectively. Significant differences were observed in the CSS among three groups (χ²=332.27, P<0.05).

Based on the SEER database, the Nomogram prognostic model for early ICC has been successfully established. Compared with conventional AJCC staging, this model yields higher predictive efficiency, which can also deliver stratified analysis of survival risk.