Cardiac calcification and radiation dose to cardiac substructures as predictors of survival in breast cancer patients undergoing thoracic radiotherapy

Cardiac toxicity remains a relevant concern in breast cancer radiotherapy. Both incidental dose to specific cardiac substructures and pre-existing coronary calcifications may impact long-term outcomes. This study evaluated the prognostic significance of these two complementary predictors in a Chilean cohort of breast cancer patients undergoing thoracic radiotherapy. This retrospective cohort study included 375 breast cancer patients treated at UC-Christus Health Network (1985–2022), comprising 61 deceased and 314 non-deceased patients. Cardiac substructures were automatically segmented using an open-source algorithm. Cardiac calcium burden was quantified from non-contrast planning computed tomography (CT) using a 130 HU threshold. Dose distributions were converted to equivalent dose in 2 Gy fractions. Univariate Cox regression with false discovery rate correction and graph-based collinearity selection were applied, followed by multivariable Cox models adjusted for relevant clinical covariates and cardiac calcium burden. Cardiac calcium burden was independently associated with OS after adjustment for clinical covariates. Among dosimetric parameters, three metrics survived collinearity elimination and remained independently associated with overall survival (OS) in separate multivariable models: composite cardiac base region (C1) mean dose, right ventricle (RV) mean dose, and RV-V10Gy. Kaplan–Meier analyses confirmed progressive survival differences across calcium burden catefum burden and radiation dose to the C1 region and RV were independently associated with OS in breast cancer patients receiving thoracic radiotherapy. Cardiac calcium quantification from routine planning CT represents a clinically actionable tool for cardiovascular risk stratification within the existing radiotherapy workflow. This study also provides the first Latin American validation of an open-source automated cardiac segmentation tool in a breast cancer radiotherapy cohort. gories and RV-based dosimetric risk groups, while no statistically significant separation was observed for C1 mean dose groups, when restricting the minimum group size. Automated segmentation demonstrated high accuracy for whole-heart delineation, while geometric accuracy for left anterior descending artery segmentation was limited. Pre-treatment cardiac calcium burden and radiation dose to the C1 region and RV were independently associated with OS in breast cancer patients receiving thoracic radiotherapy. Cardiac calcium quantification from routine planning CT represents a clinically actionable tool for cardiovascular risk stratification within the existing radiotherapy workflow. This study also provides the first Latin American validation of an open-source automated cardiac segmentation tool in a breast cancer radiotherapy cohort.