Qiu Y, Wang L, Zhong X, et al. - Researchers sought to construct a prognostic model to predict recurrence based on the prevalence of breast cancer stem cells (BCSCs) in breast cancer. For this purpose, they quantified the stem cells of breast cancer patients via immunohistochemistry and dual-immunohistochemistry. They used the holdout methods, where the dataset was randomly split into two exclusive sets (70% training and 30% testing sets), to evaluate the performance of Cox proportional hazard regression model. Four groups of BCSCs showed a link with relapse-free survival: ALDH1A3, CD44⁺/CD24⁻, integrin alpha 6, and protein C receptor. A relapse risk score (RRS) was calculated using a prognostic panel of integrated correlated biomarkers. RRS classified 67.81 of patients into low-risk group and 32.19% of patients into high-risk group. By Kaplan-Meier method, a significantly lower relapse rate at 5 years was seen in the low-risk group vs high-risk group. Independent of age at diagnosis or tumour size, the RRS was established as a powerful classifier in the multiple Cox model. For the prediction of the relapse risk in early stage breast cancer, the applicability of the RRS model was suggested. Hormonal therapy treatment is not beneficial for high RRS score oestrogen receptor-positive patients.