Predictive Modeling of Ipsilateral Mean Lung Dose (MLD) in Breast Cancer Radiotherapy: The Synergistic Impact of Planning Target Volume and Surgical Paradigm (MRM vs. BCS) A High-Density Analytical Study Integrating 100 Clinical, Anatomical, and Biological Variables

Purpose Radiation-induced pulmonary toxicity (RIPT) remains the primary dose-limiting constraint in adjuvant breast cancer radiotherapy (RT), particularly as clinical indications for regional nodal irradiation (RNI) expand. This study develops and validates a high-fidelity predictive model for Ipsilateral Mean Lung Dose (MLD) using Planning Target Volume (PTV) and surgical technique (Modified Radical Mastectomy [MRM] vs. Breast Conserving Surgery [BCS]) as the primary determinants. Methods We analyzed a prospective cohort of 150 patients undergoing radiotherapy at a tertiary oncology center. The baseline dataset was expanded to incorporate 100 multi-dimensional criteria. All patients were planned using advanced techniques (IGRT, IMRT, DIBH). Dosimetric data were extracted using the Acuros XB algorithm. Multiple linear regression and machine learning recursive feature elimination (RFE) were employed to build the predictive framework. Toxicity-free survival (TFS) was estimated via Kaplan-Meier analysis. Results Analysis identifies a profound linear correlation between PTV and MLD (R^2 = 0.88, p < 0.001). However, the surgical paradigm acts as a critical dosimetric modifier; MRM patients exhibited an 18.5% higher MLD compared to BCS patients for matched PTV volumes. This "surgical penalty" is attributed to the anatomical reduction in the glandular buffer. A critical MLD threshold of 12.5 Gy was identified as a prognostic tipping point for Grade 2 + radiation pneumonitis. Conclusion We established a robust predictive model enabling the prospective estimation of MLD. This provides an automated quality assurance framework for the personalization of breast cancer RT, ensuring locoregional control is not compromised by pulmonary morbidity.