Main Session
Sep
28
PQA 04 - Breast Cancer, Patient Reported Outcomes/QoL/Survivorship, Functional Radiation Medicine, Hematologic Malignancies, Palliative Care, and International/Global Oncology
Presenter(s)
Raleigh Anderson, MD - Thomas Jefferson University, Philadelphia, PA
R. Anderson, R. Simon, N. L. Simone, and Y. Chen; Dept. of Radiation Oncology, Sidney Kimmel Medical College and Comprehensive Cancer Center, Thomas Jefferson University, Philadelphia, PA
Purpose/Objective(s):
With increasing obesity and metabolic dysfunction in breast cancer patients, minimizing skin toxicity is imperative. While there are general management strategies and dosimetric constraints that can limit toxicities, individualized frameworks to predict skin toxicities in breast cancer patients are not commonly used and further research is needed to identify patient specific risk factors both clinically and based on individualized treatment planning data. Artificial intelligence is well suited to develop risk models from diverse multivariate variables. We sought to evaluate dosimetric and clinical differences between patients who had grade 2 or higher and patients with grade 1 or 0 skin toxicity as a first step in creating a machine learning toolkit to identify patients who are at higher risk for skin side effects.Materials/Methods:
Consecutive breast cancer patients treated at a single institution between December 2024 and December 2025 were identified who had either a grade 0/1 (coded as no/minimal toxicity) or a grade 2 or greater (coded as toxicity) CTCAE skin toxicity at the end of treatment. The patients could be treated to the whole breast or chest wall to a dose of 42.56 Gy in 16 fractions or 50Gy in 25 fractions. Patients' demographics, tumor characteristics and treatment characteristics were evaluated including age, BMI, DM2, blood glucose, smoking/alcohol use, menopausal status, T and N stage, tumor grade, type of surgery, axillary surgery, systemic therapy, radiation targets/dose, and presence of a surgical cavity boost. Plans were evaluated for the dose received by a skin contour that was created with a 5 mm interior margin from body edge and values of V30Gy, V35Gy, V40Gy, V45Gy, V50Gy, and Dmax. Numeric clinical and dosimetric differences between the patient groups were assessed with a two-tailed t-test. Categorical clinical differences were calculated with Chi-square test of independence.Results:
Plans from 21 patients were identified with grade 2 or higher skin toxicity (Tox), and 11 with grade 0 or 1 skin toxicity (No Tox). While no differences were noted between the groups with regards to demographic, clinical or tumor characteristics, N staging (p=0.082), and post-operative systemic therapy (p=0.074) trended towards significance. Additionally, a difference was noted between groups for the V30Gy skin dose(median 273.77 cc Tox group vs 221.89 cc No Tox; p< 0.05) and V35Gy skin dose (median 247.36 cc Tox group vs 182.47 cc in the No Tox; p< 0.05) with trends to significance in V40Gy (p=0.062) and V45Gy (p=0.084) to the skin.Conclusion:
These findings demonstrate that higher intermediate skin dose volumes are associated with grade =2 radiation dermatitis. Therefore, skin dosimetric parameters may serve as key predictive features in developing an AI-based model to identify breast radiation patients at increased risk for clinically significant skin toxicity.