Abstract
BACKGROUND: Nonmetastatic castration-resistant prostate cancer (nmCRPC) is a clinical challenge due to the high progression rate to metastasis and mortality. To date, no prognostic model has been developed to predict the metastatic probability for nmCRPC patients. In this study, we developed and externally validated a machine-learning model capable of calculating risk scores and predicting the likelihood of metastasis in nmCRPC patients. PATIENTS AND METHODS: A total of 2716 nmCRPC patients were included in this study. The training and testing datasets were derived from SPARTAN (NCT01946204) and ARAMIS (NCT02200614), respectively. Regarding metastasis-free survival as the endpoint, we subjected 13 clinical features to 10 machine-learning models and their combinations to predict metastasis. Model performance was assessed through accuracy (AUC), calibration (slope and intercept), and clinical utility (DCA). The risk score calculated by the model and risk factors based on eight identified variates were used for metastatic risk stratification. RESULTS: The final prognostic model included eight prognostic factors, including novel hormone therapy application, Gleason score, previous treatments received (both surgery and radiotherapy, or neither), Race (White), PSA doubling time (PSADT), hemoglobin (HGB), and lgPSA. The prognostic model resulted in a C-index of 0.724 (95% CI 0.700-0.747) in internal validation and relatively good performance through tAUC (>0.70 at 3-month intervals between 6 and 39 months) in external validation. In the risk score stratifying strategy, compared with the low-risk group, the metastasis HRs for medium- and high-risk groups were 1.72 (95% CI 1.39-2.12) and 4.43 (95% CI 3.66-5.38); as for risk factor count, the HRs are 1.98 (95% CI 1.50-2.61) and 4.17 (95% CI 3.16-5.52), respectively. CONCLUSIONS: In this study, we developed and validated a machine learning prognostic model to predict the risk of metastasis in nmCRPC patients. This model can assist in the risk stratification of nmCRPC patients, guide follow-up strategies, and aid in selecting personalized treatment intensities.