The cohort included a development set (n = 4305) and an external set (n = 1503) of BC patients treated with postoperative RT (2005-2015). Cardiac dose-volume histogram (DVH) parameters were obtained from three-dimensional CT-based planning. Multivariable cause-specific Cox regression models predicted ACE, accounting for overall mortality risk. Model 1 included seven clinical parameters, whole heart (WH)-Dmean, and left ventricle (LV)-V5. Model 2 replaced WH-Dmean and LV-V5 with cardiac DVH parameters offering higher prognostic value. Model 3 was built by testing whether the better-performing model was enhanced by adding left anterior descending artery DVH parameters, coronary artery calcium (CAC) volumes, or CAC DVH parameters. The incremental value of adding a parameter (or to compare models) was assessed using the likelihood ratio (LR) test. Internal validation was done via bootstrapping. An exploratory assessment of the models' performance in an external population was performed.
At ten years, the cumulative incidence of ACE was 4.8 % and 1.4 % in the development and external sets, respectively. Model 2, which included WH-D1, LV-V6, and right ventricle (RV)-V6, did not outperform Model 1 (LR p = 0.422). Consequently, further model improvement was pursued by expanding Model 1 rather than Model 2. The resulting Model 3 incorporated the parameters from Model 1 along with the total CAC volume, significantly enhancing predictive performance (LR p < 0.001). At 10 years post-RT, each model had c-indexes above 0.73 at internal validation, with Model 3 performing best (c-index 0.78; 95 % CI, 0.72-0.83) CONCLUSION: A model including seven clinical parameters, WH-Dmean, LV-V5, and total CAC volume best predicted ACE risk and may guide RT dose optimisation strategies.
This study aimed to develop prediction models for acute coronary events (ACE) following breast cancer (BC) radiation therapy (RT).
This website uses cookies to ensure you get the best experience on our website.