Predicting the survival benefits of postoperative radiotherapy for breast cancer with lymph node micrometastasis: A Machine Learning model

Objective: This study aims to assess the impact of post-mastectomy radiotherapy (PMRT) on the survival outcomes of patients with T1-T3N1miM0 breast cancer. Additionally, we seek to develop an interpretable machine learning model to predict the individualized 5-year survival benefits for patients undergoing postoperative radiotherapy in comparison to those who do not receive such treatment. Methods: Based on data from the SEER database spanning from 2010 to 2022, a cohort of 17,994 breast cancer patients diagnosed with T1-3N1miM0 was analyzed. Propensity score matching was employed to mitigate baseline discrepancies between the radiotherapy and non-radiotherapy cohorts. The overall survival and breast cancer-specific survival between the two groups were compared using the Kaplan-Meier method. A predictive model utilizing XGBoost was developed to estimate the 5-year survival advantages associated with postoperative radiotherapy, with the model's outcomes elucidated through SHAP analysis. Results: A significant difference in survival was observed between the postoperative radiotherapy and non-radiotherapy groups (HR: 0.85, 95% CI 0.75-0.96, p < 0.01). However, no significant differences in overall survival (OS) were found among the T1 and stage I radiotherapy and non-radiotherapy groups (HR: 0.87, 95% CI 0.75-1.08, p = 0.195; HR: 0.96, 95% CI 0.77-1.19, p = 0.704). Although breast cancer-specific survival (BCSS) was assessed, the difference was not significant (HR: 0.87, 95% CI 0.74-1.02, p = 0.84). The XGBoost model we developed exhibited exceptional predictive performance and was identified as the most effective model for predicting the 5-year survival outcomes of T1-T3pN1miM0 breast cancer patients (AUC = 0.770). Conclusion: Breast cancer patients diagnosed with T1N1miM0 or Stage I（pN1miM0） may potentially forgo PMRT. An XGBoost machine learning model was created to forecast the 5-year survival advantage of radiotherapy for postoperative individuals with T1-T3N1miM0 breast cancer.