Computerized Risk Modeling of Imaging Changes and Radionecrosis After Proton Beam Therapy for Childhood Brain Tumors
F Heinzelmann1,2,3*, C Baeumer1,2,3,4, S Peters1,5, A Stock6, S Schulze-Schleithoff1, S Frisch1, B Timmermann1,3,4,5,7, (1) West German Proton Therapy Center Essen (WPE), Essen, NW, DE, (2) TU Dortmund University, Department of Physics, Dortmund, NW, DE, (3) University Hospital Essen, West German Cancer Center (WTZ), Essen, NW, DE, (4) German Cancer Consortium (DKTK), Heidelberg, NW, DE, (5) University Hospital Essen, Department of Particle Therapy, Essen, NW, DE, (6) University Hospital Wuerzburg, Department of Neuroradiology, Wuerzburg, BY, DE, (7) University of Duisburg-Essen, Faculty of Medicine, Essen, NW, DE
Presentations
PO-GePV-T-125 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: In childhood brain tumors, long-term side effects of radiation therapy have to be limited to a minimum. Quantitative descriptions of risk factors for the occurrence of radiation-induced brain injuries (RIBIs) in proton beam therapy such as radionecrosis are under development. Therefore, the correlations of post-therapeutic imaging changes with biological and physical parameters, e.g., linear energy transfer (LET) values, variable relative biological effectiveness (RBE), and proximity to the ventricular system (VS), were investigated in this retrospective study.
Methods: To forecast and validate imaging changes, published logistic regression models for the prediction of normal tissue complication probability (NTCP) based on adult glioma patients and a combined cohort with benign brain tumors were applied. The quantities of constant and variable RBE weighted dose, and unrestricted dose-averaged LET were evaluated in a research version of the treatment planning system RayStation with a built-in Monte Carlo dose engine. While the voxel-wise NTCP model predicts the RIBI location for every voxel in the planning CT, the patient-level model calculates the event of a RIBI for each patient.
Results: The analyzed cohort included 37 cases with follow-up MRI images. The respective NTCPs were calculated. In three cases, asymptomatic imaging changes were identified. The post-therapeutic findings were in the brainstem (n=2) or in the frontal lobe (n=1) but outside the 4 mm region surrounding the VS. For these cases, the calculated NTCPs were not significantly increased within any applied model but the estimates for the imaging change within the frontal lobe were more consistent with the radiobiological model calculation.
Conclusion: The comparison between the predicted and occurred RIBIs in our cohort showed only partial agreement when applying the published NTCP models. Consequently, improvements and adaptations of the NTCP models for pediatric patients are of utmost importance to provide reliable forecasts for side-effects.
Funding Support, Disclosures, and Conflict of Interest: Funding Support from Mercator Research Center Ruhr (MERCUR) and Barbara und Hubertus Trettner Stiftung
Keywords
Taxonomy
TH- External Beam- Particle/high LET therapy: Proton therapy – computational dosimetry-deterministic
Contact Email