Cardiac Substructure Dosimetry and Major Adverse Cardiac Events After Breast Radiotherapy: A 12-Year Cohort Study with NTCP Modeling.

Abstract

To identify cardiac substructure dosimetric predictors of major adverse cardiac events (MACE) following breast radiotherapy, and to develop a robust normal tissue complication probability (NTCP) model for personalized risk assessment using a cohort with over a decade of follow-up. This retrospective study included 633 female breast cancer patients receiving postoperative radiotherapy (2011-2013) at two XXXX public hospitals. The heart and six cardiac substructures were retrospectively contoured on planning CTs, and dose-volume metrics were extracted and converted to EQD2. The primary endpoint was the first occurrence of MACE (cardiogenic death, myocardial infarction, coronary revascularization, unstable angina, or heart failure). Fine-Gray regression was used to identify predictors of MACE, with non-cardiogenic death as the competing event. Based on these predictors, an NTCP model was developed to quantify the 10-year excess MACE risk attributable to radiation. Model performance was evaluated using bootstrap optimism-corrected Harrell's C-index and 10-year time-dependent AUC. Over a median follow-up of 11.8 years, 29 patients (4.6%) developed MACE. In multivariable analysis, left anterior descending artery (LAD) V40 in EQD2 emerged as the strongest dosimetric predictor (sHR = 1.023 per %; p = 0.010). A threshold of LAD V40 ≥18% identified patients at more than 2-fold increased risk of MACE (sHR = 2.333; p = 0.033). The final NTCP model (incorporating LAD V40, age, and cardiac history) demonstrated excellent discrimination with an optimism-corrected C-index of 0.865. Compared with 3D-CRT, VMAT re-planning of high-risk cases substantially reduced LAD V40, translating to an average 5.7% absolute reduction in projected 10-year excess MACE risk. In this multicenter cohort with long-term follow-up, LAD V40 was identified as a superior predictor of MACE compared to whole-heart dosimetry. Pending external validation, the developed NTCP model offers a promising framework for personalized risk stratification to guide the use of cardiac-sparing techniques.