关键词: 鼻咽癌, SEER, 生存周期, 预测模型

Abstract:

Objective　To analyze the survival cycle and prediction model of patients with nasopharyngeal carcinoma based on the Surveillance Epidemiology and End Results (SEER). Methods　A total of 1 000 patients diagnosed with nasopharyngeal carcinoma from 2014 to 2022 were collected through the SEER, and the patients' data were retrospectively analyzed. The extracted data were analyzed using the R 3.6.3 software; the patients were divided into a training set and a verification set according to the ratio of 7∶3. A model was built with the training set, and the Cox regression analysis was performed. On this basis, a nomogram survival prediction model was established. The C-index was used to evaluate the differentiation between the training set and the verification set, and the calibration curve was used to detect the nomogram fit. Results　Among the 1 000 patients, 74.1% (741/1 000) were male, and 25.9% (259/1 000) female; the patients with Keratinized squamous cell carcinoma accounted for 6.0% (60/1 000), the patients with non-keratinized squamous cell carcinoma 66.1% (661/1 000), the patients with basal-like squamous cell carcinoma 5.8% (58/1 000), and the patients with other pathological types 22.1% (221/1 000); 32.4% (324/1000) were undifferentiated, 33.0% (330/1 000) poorly differentiated, 24.4% (244/1 000) moderately differentiated, and 10.2% (102/1 000) highly differentiated. The overall survival rate was 33.9% (339/1 000), and the mortality rate was 66.1 (661/1 000). To ensure the stability of the subsequent model, χ² test was used to confirm that there were no statistical differences in baseline characteristics between the training set and the verification set (all P>0.05). Age, gender, pathological type, tumor grade, TNM stage, number of metastatic organs, surgical treatment, radiotherapy, and chemotherapy all affected the patients' overall survival (all P<0.05), while marital status, education level, and differentiation degree had no effect on the patients' overall survival (all P>0.05). Multivariate Cox regression analysis was performed on the factors with P < 0.05 in the results of univariate analysis. The results showed that age, gender, pathological type, tumor grade, TNM stage, number of metastatic organs, surgical treatment, radiotherapy, and chemotherapy were all independent risk factors for the patients' overall survival. The patients aged > 50 years old, female patients, the patients with non-keratinized squamous cell carcinoma, high tumor grade, late local stage, and more organ metastases, and the patients who did not take surgery, radiotherapy, and chemotherapy had poor prognosis. In order to be more convenient and intuitive for clinical application, a nomogram of the patients was constructed based on the results of multi-factor Cox regression analysis to predict the 3-year survival level. In the training set, the C index of the nomogram prediction model to judge the prognosis was 0.828 (95%CI 0.791-0.863); in the verification set, the C index of the nomogram prediction model to judge the prognosis was 0.826 (95%CI 0.77-0.862); the RFS prediction effect of the nomogram was good. Conclusions　Age, gender, chemotherapy, TNM stage, pathological type, and number of metastatic organs are the risk factors affecting the survival cycle of patients with nasopharyngeal carcinoma. The model is successfully constructed, and the model has good differentiation and fit.

Key words: Nasopharyngeal carcinoma,  , SEER,  , Survival cycle,  , Prediction model