ESTRO 2025 - Abstract Book

S1333

Clinical - Lung

ESTRO 2025

but little exists for NSCLC. The objective of this study was to develop a multivariate CT-based radiomics model to predict response to definitive chemoradiation in patients with lung adenocarcinoma.

Material/Methods: Patients diagnosed with locally advanced, unresectable adenocarcinoma of the lung who had undergone chemoradiation followed by at least one dose of maintenance durvalumab between 2018 and 2023 were included in this study. Treatment response of the primary lung tumor (excluding the nodes) to the chemoradiation component only was classified into responders (R) and non-responders (NR) using the Response Evaluation Criteria in Solid Tumors (RECIST 1.1). The Python library, PyRadiomics, was used to extract statistical, morphological, and textural features from the patients’ pre-treatment CT images and their wavelet-filtered versions. A nested leave one-out cross-validation was used for model building and evaluation. Results: Fifty-seven patients were included in this study. The clinical stage was IIIA-C in 98% of patients. All but one participant received 6,000-6,600 cGy of radiation in 30-33 fractions. All participants received concurrent platinum based chemotherapy. The average pretreatment tumor size was 35 mm (range 11 - 130). The average post treatment size was 31 mm (range 8 - 83). Based on RECIST 1.1, 20 (35%) patients were classified as responders (R) to chemoradiation and 37 (65%) patients as non-responders (NR) and defined the ground truth label for model evaluation. A 3-feature model based on the k-nearest neighbors (KNN) k = 1 machine learning classifier was found to have the best performance, achieving a recall, specificity, accuracy, balanced accuracy, precision, negative predictive value (NPV), F1-Score, and area under the curve (AUC) of 84%, 70%, 79%, 77%, 84%, 70%, 84%, 0.77, respectively. The validation performance of this model was 84%, 93%, 88%, 88%, 93%, 87%, 89% and 0.88, respectively. Conclusion: Our results suggest that a CT-based radiomics model may be able to predict chemoradiation response for lung adenocarcinoma patients. Further studies are needed to validate our findings. Digital Poster Optimizing postoperative proton radiotherapy in thymic epithelial tumors: added value of breath hold? Esther Kneepkens, Marije Velders, Judith van der Stoep, Maud Cobben, Maud van den Bosch, Nicole Hendrix, Giorgio Cartechini, Ilaria Rinaldi, Dirk de Ruysscher, Dianne Hartgerink, Judith van Loon, Stéphanie Peeters Department of Radiation Oncology, Maastro, GROW School for Oncology, Maastricht University Medical Center+, Maastricht, Netherlands Purpose/Objective: Thymic epithelial tumors (TET) are rare neoplasms originating from the thymus' epithelial cells. At our center, patients receive postoperative proton therapy based on the Dutch model-based approach for reduction in the estimated normal tissue complication probability (NTCP) for cardiac toxicity, pneumonitis and dysphagia. 1–3 Inspiratory breath hold (BH) may increase lung volume and beneficially change heart position/shape. This study aimed to refine proton plans by exploring the potential benefits of BH for these patients. Material/Methods: Due to the absence of BH CT scans for TET patients, we studied the effect of inspirational BH on CT scans of a cohort of 10 lymphoma patients. The FB CTVs were derived from matched TET patients using deformable image Keywords: radiomics, lung adenocarcinoma, predictive marker 1750