关键词: 胸腺上皮肿瘤； Masaoka-Koga分期； 预后分析； 列线图； SEER数据库

Abstract: Objective To explore the influencing factors affecting the overall survival (OS) of thymic epithelial tumors（TETs） patients,draw a nomogram and evaluate the differentiation,calibration and clinical application value,to develop personalized postoperative diagnosis and treatment plans for monitoring and follow-up.Methods We extracted clinical data from the Surveillance,Epidemiology,and End Results (SEER) database,encompassing patients with pathologically confirmed thymic epithelial tumors (TETs) who had complete follow-up records.They were randomly divided into training set and validation set at a ratio of 7∶3.The Cox proportional hazards regression model (Cox analysis) was used to obtain the independent influencing factors affecting the prognosis of TETs patients,and a nomogram model was constructed.The model was evaluated from three aspects:discrimination,calibration and clinical application value.Results Multivariate Cox regression analysis was performed after univariate Cox analysis of the training set to determine that age,pathological type,Masaoka-Koga stage,surgery and chemotherapy were independent prognostic factors related to OS in TETs patients (all P<0.05).The C-index of the nomogram is 0.765.The Area Under the Curve of ROC for 3-year,5-year and 10-year OS were 0.791,0.81 and 0.803,respectively.Calibration plots were used to evaluate that the predicted values of the nomogram model were in good agreement with the actual OS,and decision curve analysis showed that the nomogram model had higher clinical net benefits at 3,5 and 10 years.The results of internal validation with the validation set are basically consistent.Conclusion The nomogram model constructed in this study has good discrimination,accuracy and good clinical application value,and can be used to predict the OS-related individualized survival time of TETs patients in clinical practice.

Key words: thymic epithelial tumor; Masaoka-Koga stage； prognostic analysis; nomogram; SEER database