Nomogram model of survival prediction for nasopharyngeal carcinoma with lung metastasis: developed from the SEER database and validated externally.

Objective: Distant metastasis occurs in some patients at the first diagnosis of nasopharyngeal carcinoma (NPC), the prognosis is poor, and there are significant individual differences. This study established a nomogram model of lung metastasis of NPC as a supplement to TNM staging. Methods: The training cohort is used to build the nomogram model, and the validation cohort is used to evaluate the model. The training cohort of 177 patients is from the Surveillance, Epidemiology, and End Results (SEER) database. Factors affecting overall survival (OS) in patients with lung metastasis of NPC analysis by Cox regression analysis and then a nomogram were established. 122 patients from the Affiliated Tumor Hospital of Guangxi Medical University were selected as the external validation cohort. The concordance index (C-index), the area under the curve (AUC), and the calibration curve were used to assess the accuracy of the nomogram and used the decision curve analysis (DCA) curve to measure the clinical benefit capacity of the model. The patients were separated into two groups with different risks, and the "Kaplan-Meier (KM)" survival analysis was used to evaluate the differentiation ability of the model. Results: Age, T-stage, radiation, chemotherapy, and brain metastases can affect the OS in NPC with lung metastasis. A nomogram was developed according to the above five factors. The C-index of the training cohort and the validation cohort were 0.726 (95% CI: 0.692-0.760) and 0.762 (95% CI: 0.733-0.791). The AUC of the nomogram was better than that of the TNM staging. In the training cohort, the nomogram predicted OS AUC values of 0.767, 0.746, and 0.750 at 1, 2, and 3 years, TNM stage of 0.574, 0.596, and 0.640. In the validation cohort, nomogram predictions of OS AUC values of 0.817, 0.857, and 0.791 for 1, 2, and 3 years, TNM stage of 0.575, 0.612, and 0.663. DCA curves suggest that nomogram have better clinical net benefits than TNM staging. The KM survival analysis shows that the nomogram has a reasonable risk stratification ability. Conclusion: This study successfully established a nomogram model of NPC lung metastasis, which can be used as a supplement to TNM staging and provide reference for clinicians.