Establishment and validation of nomogram models for overall survival (OS) and cancer-specific survival (CSS) of patients with spindle cell sarcoma (SCS): A SEER database analysis.

Abstract

Abstract Background Spindle cell sarcoma (SCS) is rare in clinical practice. The purpose of this study was to establish the nomograms to predict the OS and CSS prognosis of patients with SCS based on the Surveillance, Epidemiology, and End Results (SEER) database. Methods The data of patients with SCS were extracted from the SEER database between 2004 and 2020, and randomly allocated to the training cohort and validation cohort. Univariate and multivariate Cox regression analysis are used to screen for independent risk factors both in overall survival (OS) and cancer-specific survival (CSS). Nomograms for OS and CSS were established for patients with SCS based on the results of multivariate cox analysis. Then we validated the nomograms by Concordance index (C-index), receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). Finally, the Kaplan-Meier curve and log-rank test were performed to compare between the patients with SCS in three different levels and different treatment groups. Results A total 1369 patients with SCS were included and randomly divided into the training cohort (n = 961, 70%) and validation cohort (n = 408, 30%). Age, M, tumor size, tumor location, surgery and radiation were independent prognostic factors for OS, while Age, N, M, tumor size, tumor location and surgery for CSS by Cox regression analysis. The nomogram models were established based on the result of the Multiple Cox analysis both in OS and CSS. The C-index of the OS model was 0.79 and 0.77 in the training and validation group, while 0.80 and 0.78 for CSS. The 3/5-year AUCs were 0.817 and 0.824 for the training group and 0.798 and 0.792 for the validation group for OS, while 0.829 and 0.831 in the training group, 0.814 and 0.791 in the validation group for CSS. calibration curves showed high consistencies between the observed survival and the predicted survival both in OS and CSS. In addition, DCA analyzed the clinical practicality of OS and CSS nomogram models have good net benefit. Conclusion The two nomograms we have established are expected to accurately predicting personalized prognosis of SCS patients, which may beneficial for clinical decision-making.