Abstract / Summary
Our study aimed to develop nomograms to accurately predict the 1-, 3-, and 5-year overall survival (OS) and cancer-specific survival (CSS) of patients with rare histological variant prostate cancer (PCa) based on clinicopathological characteristics. We gathered clinicopathological and prognostic data from the Surveillance, Epidemiology, and End Results database. Independent prognostic factors were identified using univariate and multivariate Cox regression models, after which nomograms were developed. The predictive values were evaluated using several parameters, including the receiver operating characteristic (ROC) curve, areas under the ROC curve (AUC), concordance index (C-index), calibration curve, net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision curve analysis (DCA). A total of 3538 patients with rare histological variant PCa diagnosed from 2000 to 2021 were randomly assigned to the training cohort ( n = 2476) and the validation cohort ( n = 1062) at a ratio of 7:3. Univariate and multivariate Cox regression analysis identified age, marital status, Gleason score, tumor grade, summary stage, chemotherapy, surgery, and histology as independent prognostic factors for OS and CSS (all P < 0.05). We developed nomograms to predict patients’ 1-, 3-, and 5-year OS and CSS. The C-index for OS prediction was 0.831 and 0.821 in the training and validation cohorts, respectively, while they were 0.873 and 0.868 for CSS, respectively. For OS, the 1-, 3-, and 5-year AUCs were 0.918, 0.908, and 0.892, respectively, in the training cohort and 0.913, 0.902, and 0.882, in the validation cohort; for CSS, they were 0.929, 0.940, and 0.930, respectively, in the training cohort and 0.933, 0.932, and 0.921, in the validation cohort. Furthermore, the calibration curves demonstrated a high consistency between the actual observed survival probabilities and those predicted by the nomograms. The ROC curves, AUC, C-index, NRI, IDI, and DCA showed that the nomograms exhibited better prognostic discrimination than the TNM staging system within the present SEER-derived cohort. Our study developed nomograms showing favourable predictive performance in internal validation to accurately predict the 1-, 3-, and 5-year OS and CSS for patients with rare histological variant PCa. These tools may assist clinicians in optimizing clinical management and providing individualized survival forecasts for these patients, pending further external validation.