ESTRO 2025 - Abstract Book

S2795

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2025

Excluding patients with electrodes/pacemaker, 1097 patients were analyzed. MHD mean values were 0.8±0.1/2.4±0.7Gy for right/left. With a median follow-up of 6.5y, 29 patients experienced cardiac events. AS/CAC_volume/Max_HU were all strongly significant predictors (p<0.0001). Age, laterality (left/right), obesity and hypertension were also significant at univariate analysis. MHD was not associated, in both overall/left cohorts: once encoded using the best cut-off (1Gy, mostly representing laterality) resulted to be predictive. The best multivariate model combined MHD>1Gy, age and CAC scores with similar performances for the three different scores, as shown in Table-1. The model using CAC_volume, promising for “easy” clinical use, showed AUC=0.77, p<0.0001 and almost perfect calibration (R2=0.98), being CAC_volume the strongest predictor (p<0.0001). Figure-2 shows the risk against CAC_volume, stratified according to MHD above/below 1Gy.

Conclusion: CAC scores extracted from planning CT were the most important predictors of cardiotoxicity after breast RT in a modern series. Despite low MHDs, a residual interplay between CAC scores and MHD was found, showing that left breast patients with large scores (i.e.:CAC_volume>0.5-1cc) are the candidates to focus efforts in further reducing heart dose.

Keywords: Biomarkers, predictive models, cardiac toxicity

References: Ref-1: https://github.com/wasserth/TotalSegmentator?tab=readme-ov-file Ref-2: Daniel S. Berman, Yael Arnson, and Alan Rozanski. Coronary artery calcium scanning: the Agatston score and beyond. JACC Cardiovascular Imaging, 9:1417–1419, 2016.

1901

Digital Poster Optimizing Dosimetry in Stereotactic Radiosurgery for Multiple Brain Metastases: The Advantage of Dual Isocenter Technique Cristina Teixeira, Thierry Gevaert, Marlies Boussaer, Sven Van Laere, Mark De Ridder Radiotherapy, Universitair Ziekenhuis Brussel (UZB), Vrije Universiteit Brussel (VUB), Brussels, Belgium