Development and external validation of a nomogram for choosing postoperative adjuvant therapies in uterine sarcoma patients from real-world data

Ze Ouyang¹Weili Tao²Man Zou^3*

• ¹Department of Medical Oncology, Hubei Cancer Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China
• ²Department of Ultrasound, First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan Province, China
• ³Department of Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China

Objective. To develop and validate a prognostic nomogram to identify uterine sarcoma (US) patients who may not require adjuvant therapy after surgery, based on SEER and external Asian cohort data.: Firstly, data from eligible uterine sarcoma patients in the United States (n = 1626) meeting the criteria of this study were collected from the Surveillance, Epidemiology, and End Results (SEER) database and randomly divided into a training cohort (n = 1138) and an internal validation cohort (n = 488). Multivariate Cox regression, Lasso regression, and cross-validation were performed to select the optimal variables associated with survival. A nomogram-driven model was then constructed to stratify the recurrence risk thresholds for the assessed patients. Finally, an external dataset from another cohort of our hospital (n = 90) was used to validate the accuracy and specificity of the nomogram model in discriminating patient risks, employing the C-index, receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA).Results: Using the aforementioned classification aggregation methods, analysis of the training cohort identified diagnostic age, FIGO stage, grade, tumor size, and peritoneal cytology as independent predictors of overall survival (OS). The subsequent risk model demonstrated that patients with a threshold below 55 had a 10-year survival rate exceeding 80%, suggesting no need for postoperative adjuvant therapy. Internal validation confirmed the reliability of this multiparameter model, as evidenced by a Cindex of 0.77 and ROC AUC values of 0.812, 0.824, and 0.839 for 1-year, 3-year, and 5-year OS. Similarly, the accuracy and specificity were confirmed by the external validation cohort, as evidenced by a C-index exceeding 0.83, reaching a peak of 0.9, and ROC AUC values greater than 0.876, highlighting that the stratified thresholds displayed by our nomogram outperformed FIGO staging in identifying low-risk recurrence patients.Our constructed multiparameter nomogram model appears to be superior to the FIGO staging system for identifying low-risk patients who do not require adjuvant therapy after surgery, although prospective data are required for further validation.