Evidence of different predictors for different types of cardiotoxicities after hypo-fractionated whole breast radiotherapy
TL;DR
CAC_corr events were mainly associated with CAC burden and age, independently of tumor laterality, as most occurred after left-sided irradiation, while MHD was not independently associated with cardiac events.
Abstract
Background and purpose Cardiac calcifications (CAC) are emerging as relevant predictors of cardiac toxicity after breast cancer radiotherapy. This study aimed to investigate the association between quantitative CAC scores and specific subgroups of cardiac events in patients treated with moderate hypofractionation, and to assess the interplay between dosimetric and clinical predictors for different families of events. Materials and methods Data from 1172 consecutive breast cancer patients treated at a single institution with 40 Gy in 15 fractions were analyzed. Mean heart dose (MHD) was extracted after automatic heart segmentation. Agatston score (AS), CAC volume, and maximum heart HU value (Max_HU) were quantified from planning CT scans. Available clinical variables included laterality, chemotherapy, age, smoking status, and respiratory toxicity occurring within 3 years after radiotherapy. Cardiac events were classified into two groups: CAC_corr (coronary artery disease, acute myocardial infarction, congestive heart failure, and valvular disease) and CAC_uncorr (conduction abnormalities, atrial premature beats, atrial fibrillation, pericarditis, and tachyarrhythmia). Associations between predictors and outcomes were tested using logistic regression. Results After excluding patients with pacemakers/electrodes, 1091 patients were available. With a median follow-up of 6.5 years, 29 cardiac events were recorded. Among them, 15/29 were CAC_corr and were similarly distributed between right-and left-sided tumors (8 vs 7). Conversely, 13/14 CAC_uncorr events occurred in left breast cancer patients; therefore, analyses for this subgroup were restricted to left-sided cases (13/561). For CAC_corr events, the combination of AS/CAC_volume/Max_HU with age showed the best predictive performance (AUC=0.80/0.79, p<0.0001), while respiratory toxicity had no impact. For CAC_uncorr events, age and anthracycline-based chemotherapy (FEC) were the strongest predictors (AUC=0.73, p=0.0125). Including respiratory toxicity significantly improved model performance (AUC=0.77, p=0.0003). MHD was not independently associated with cardiac events. Conclusion CAC_corr events were mainly associated with CAC burden and age, independently of tumor laterality. In contrast, CAC_uncorr events appeared more related to treatment-related factors, as most occurred after left-sided irradiation. Within left breast cancer patients, these events were mainly associated with age, anthracycline-based chemotherapy, and respiratory toxicity rather than with MHD.