Development and validation of an online prognostic risk stratification tool to assist decisions on perioperative chemotherapy for men with luminal early breast cancer.

BACKGROUND: Male breast cancer (MBC) is rare, and indications for chemotherapy (ChT) in luminal early MBC remain understudied. We aimed to develop a model to predict outcomes for these patients and identify candidates for ChT omission. METHODS: Luminal early MBC patients from the Surveillance, Epidemiology, and End Results (SEER) program (diagnosed in 2010–2018, N = 2845) and a Chinese cohort (diagnosed in 2000–2020, N = 586) were included. SEER-derived patients (diagnosed in 2010–2016) were split into training (N = 1405) and test (N = 758) sets in a 65:35 ratio; others (diagnosed in 2017–2018) formed the US validation set (N = 682). Chinese patients constituted the CHN validation set. The response variable and primary outcome was overall survival (OS). Four machine learning models and a Cox-based nomogram were evaluated. Low- and high-risk groups were stratified utilizing the 5-year risk scores predicted by the optimal model. Survival analysis employed Kaplan-Meier method and Cox regression. RESULTS: Patients in two risk groups exhibited significantly different OS across all datasets (all P < 0.001). ChT conferred no OS benefit in low-risk patients [ChT-free as reference: training set, hazard ratio (HR) = 1.19, 95% confidence interval (CI), 0.88 to 1.60, P = 0.249; test set, HR = 1.19, 95% CI, 0.81 to 1.76, P = 0.375; US validation set, HR = 1.47, 95% CI, 0.73 to 2.94, P = 0.276; CHN validation set, HR = 0.95, 95% CI, 0.45 to 1.99, P = 0.891. Multivariable: comparable results]. For high-risk patients, ChT was significantly associated with superior OS [ChT-free as reference: training set, HR = 0.59, 95% CI, 0.47 to 0.74, P < 0.001; test set, HR = 0.68, 95% CI, 0.49 to 0.94, P = 0.019; US validation set, HR = 0.49, 95% CI, 0.29 to 0.84, P = 0.007; CHN validation set, HR = 0.42, 95% CI, 0.19 to 0.93, P = 0.027. Multivariable: comparable results]. The web-based tool has been deployed. CONCLUSIONS: Our model accurately predicts and stratifies OS in American and Chinese male patients with luminal early breast cancer. This risk stratification may aid in identifying low-risk candidates to omit ChT.