Risk Factors for Lymph Node Metastasis in Hilar Cholangiocarcinoma: A SEER-based Prediction Model by Nomogram

Abstract Background: Lymph node metastasis (LNM) is an important independent factor affecting the prognosis of patients with hilar cholangiocarcinoma. The correct evaluation of lymph node status is very important in the clinic. This study aimed to investigate the risk factors for LNM in patients with hilar cholangiocarcinoma and establish a nomogram model that can effectively predict LNM. Methods: A total of 2683 patients diagnosed with hilar cholangiocarcinoma from 2000 to 2019 in the Surveillance, Epidemiology, and End Results (SEER) database were retrospectively analyzed. Their clinicopathological data were extracted and randomly divided into a training cohort (n = 1879) and a validation cohort (n = 804) with a ratio of 7 : 3. Univariate and multivariate logistic regression analyses were used to evaluate the significant predictors of LNM. Based on the above prediction factors, a nomogram prediction model was constructed. Calibration maps, receiver operating curves (ROC), and the area under ROC curve (AUC) were used to validate the working power of the nomogram model using the two cohorts mentioned above. Results: LNM occurred in 1244 (46.37%) Of the total 2683 patients. Univariate regression analysis showed that age, T stage, tumor size, and histological grade were associated with LNM ( p < 0.05). Multivariate regression analysis showed that T stage, tumor size, and histological grade were independent risk factors for LNM ( p < 0.05). The nomogram prediction model showed good predictive power for LNM. The C-indices of the training and validation cohort were 0.725 (95% CI: 0.702-0.747) and 0.711 (95% CI: 0.676-0.746), respectively. The AUC value was 0.736 (95% CI: 0.713-0.758). The calibration curve showed high consistency between the prediction of the model and the actual transition situation, which verified the accuracy and discriminative ability of the nomogram. Decision curve analysis(DCA) showed that the nomogram model could predict clinical outcomes with satisfied accuracy. Conclusion: As predicted, T stage, tumor size, and histological grade were independent factors influencing LNM in patients with hilar cholangiocarcinoma. The visualized nomogram model can effectively predict the risk of LNM in patients with hilar cholangiocarcinoma and will help physicians make individualized treatment decisions.