S2460
Physics - Radiomics, functional and biological imaging, and outcome prediction
ESTRO 2026
Digital Poster 3141
A Clinically Implementable Model for Predicting complications after Stereotactic Radiosurgery of Brain lesions Arina A Nadina 1 , Natalia А Antipina 2 , Elena R Vetlova 2 , Andrey V Golanov 2 1 Department of Medical Physics, National Research Nuclear University MEPhI (Moscow Engineering Physics Institute), Moscow, Russian Federation. 2 Department of Radiotherapy, National Medical Research Center of Neurosurgery named after N.N. Burdenko, Moscow, Russian Federation Purpose/Objective: Radiation necrosis (RN) is a serious complication following stereotactic radiosurgery (SRS) for brain metastases, often causing prolonged neurological symptoms and significantly diminishing patient quality of life. Identifying high-risk patients is therefore critical in neuro-oncology to facilitate proactive mitigation of severe complications and preserve neurological function. This study aimed to develop a clinically implementable model that predicts individual RN risk to improve treatment safety. Material/Methods: We conducted a retrospective study of 96 patients with 107 metastatic lesions treated using CyberKnife and Novalis systems. We included patients with complete dosimetric data and a minimum follow-up of 6 months. Treatment utilized fractionation schemes of 3 × 8 Gy, 5 × 6 Gy, and 7 × 5 Gy. All doses were converted to a biologically effective dose (BED) using an α / β ratio of 3 Gy for normal brain tissue before analysis. The cohort exhibited a 33.5% incidence of radiation necrosis. We analyzed clinical parameters—including age, sex, primary tumour, lesion location, and peritumoral edema—alongside dosimetric parameters such as target volume, equivalent dose, dose-volume metrics, and plan quality indices (coverage, homogeneity). Logistic regression and random forest models were developed and evaluated using cross- validation and an independent test set. Results: Our analysis identified four key predictors of RN risk: V14 of the brain (excluding the target), target coverage index, homogeneity index, and the presence of peritumoral edema. The logistic regression model demonstrated high clinical utility, achieving a sensitivity of 88%, a specificity of 43%, and a ROC-AUC of 0.688 on the test set. The random forest model yielded a test set ROC-AUC of 0.665, with a sensitivity of 75% and a specificity of 57%.
Conclusion: MRI-based radiomic and clinical nomograms
accurately predicted 2-year PFS and DMFS in LACC treated with VMAT and MR-IGABT, demonstrating excellent calibration, good discrimination, and meaningful clinical benefit, highlighting their potential utility for individualized risk stratification and treatment decision-making in clinical practice. References: Sittiwong W et al. Pre-treatment and pre- brachytherapy MRI first-order radiomic features by a commercial software as survival predictors in radiotherapy for cervical cancer objectives. Clinical and Translational Radiation Oncology. 2025 Jul;53:100965.Dankulchai P, et al. Pre-treatment T2- weighted magnetic resonance radiomics for prediction of loco-Regional Recurrence after image-guided adaptive brachytherapy for locally advanced cervical cancer. Journal of Contemporary Brachytherapy. 2024;16(3):193–201. Lucia F et al. Prediction of outcome using pretreatment 18F - FDG PET/CT and MRI radiomics in locally advanced cervical cancer treated with chemoradiotherapy. European Journal of Nuclear Medicine and Molecular Imaging. 2018;45(10):1900 - 1912. Keywords: MR Radiomic, Cervical cancer, Nomogram validation
Made with FlippingBook - Share PDF online