Postoperative radiation therapy in localized breast malignant phyllodes tumors: RPA-derived risk model identifies high-benefit subgroups with local control improvement.

INTRODUCTION: Breast malignant phyllodes tumors (MPTs) are rare, aggressive neoplasms with high locoregional recurrence rates despite wide-margin resection. Though NCCN recommends postoperative radiation therapy (PORT), its efficacy remains controversial. This study aims to evaluate PORT's survival benefits in non-metastatic MPT and refine individualized radiation therapy strategies in clinical practice. METHODS: We retrospectively analyzed 108 non-metastatic MPT patients treated with R0 resection. Comparative analyses employed Kaplan-Meier methods, multivariable Cox models, and recursive partitioning analysis (RPA) to identify high-benefit PORT subgroups for local recurrence-free survival (LRFS). RESULTS: PORT was identified as an independent prognostic factor for LRFS (P < 0.01), while showing no protective effect on distant metastasis-free survival or overall survival. The final RPA model established three prognostic groups: low-risk (tumor size ≤5.0 cm with N0, and underwent mastectomy), intermediate-risk (tumor size >5.0 cm with N0, and underwent mastectomy), and high-risk (underwent mastectomy but with N+, or underwent BCS with any tumor size). Significant LRFS benefits were observed in intermediate-risk (P = 0.01) and high-risk groups (P < 0.05) compared to non-PORT counterparts, whereas low-risk patients showed no significant improvement (P = 0.29). CONCLUSIONS: PORT significantly improved LRFS for MPT patients, particularly benefiting intermediate- and high-risk subgroups. This proposed risk stratification provided evidence to guide individualized PORT decision making.