ESTRO 2025 - Abstract Book

S554

Clinical - Breast

ESTRO 2025

2978

Digital Poster Limited impact of breast densitometry in predicting toxicities after hypo-fractionated (40Gy/15fr) whole breast Radiotherapy Alfonso Belardo 1 , Martina Mori 1 , Andrei Fodor 2 , Paola Mangili 1 , Gabriele Palazzo 1 , Maria Giulia Ubeira Gabellini 1 , Marcella Pasetti 2 , Roberta Tummineri 2 , Miriam Torrisi 2 , Antonella del Vecchio 1 , Nadia G Di Muzio 2,3 , Claudio Fiorino 1 1 Medical Physics, IRCCS San Raffaele Scientific Institute, Milan, Italy. 2 Radiation Oncology, IRCCS San Raffaele Scientific Institute, Milan, Italy. 3 Faculty of Medicine and Surgery, Vita-Salute San Raffaele University, Milan, Italy Purpose/Objective: The study aimed to define the role of breast densitometry in the prediction of toxicities after the delivery of moderate hypo-fractionated whole breast post-operative Radiotherapy. Material/Methods: Data of a large mono-Institutional cohort (n=1127) of patients treated between 2009-2017 was available; patients were all treated with tangential fields delivering 40Gy/15fractions to the whole breast. Moderate-severe acute (ACUG2+), edema (EDMG2+), Fibrosis/Telangiectasia (FTELG2+) and Liponecrosis (LIPNEC) reactions developed within 3, 6, 42 and 42 months respectively after RT were scored using RTOG criteria. Planning CTs and the segmented Clinical Target Volume (CTV) were available: CTV HU-histograms were extracted using a home-made Python code. CTVs with a silicon implant were previously excluded. Only the range [-200,70 HU] was investigated to avoid clips and other high-density materials. Selected parameters as mean/median HU, SD, kurtosis, skewness, 10 th /25 th /75 th and 90 th percentiles were extracted to characterize the relative weight of fat and fibroglandolar tissue. For each endpoint the association with HU-histograms parameters were tested with the Mann-Whitney test (MW). The population was randomly split into training/validation cohort (n=800/327). Univariate (ULR) and Multivariate Logistic Regression (MLR) using minimum redundancy and a bootstrap-based machine learning approach were performed, including the CTV volume as a potential predictor. Results: ACUG2+/EDMG2+/FTELG2+/LIPNEC rates were 14.5%/4.5%/4.3%/11% with 164/51/49/121 events out of 1127 patients respectively. From MW, for all endpoints, CTV Volume resulted to be significantly different among populations of patients with and without toxicity (p_MW=0.0002/<0.0001/<0.0001/ 0.0017) (Table-1).