Dosimetric effects of Onyx embolization on Gamma Knife arteriovenous malformation dose distributions

OBJECTIVE Patients with arteriovenous malformations (AVMs) treated with Gamma Knife radiosurgery (GKRS) subsequent to embolization suffer from elevated local failure rates and differences in adverse radiation effects. Onyx is a common embolic material for AVMs. Onyx is formulated with tantalum, a high atomic number (Z = 73) element that has been investigated as a source of dosimetric uncertainty contributing to the less favorable clinical results. However, prior studies have not modeled the complicated anatomical and beam geometries characteristic of GKRS. This study investigated the magnitude of dose perturbation that can occur due to Onyx embolization using clinically realistic anatomical and Gamma Knife beam models. METHODS Leksell GammaPlan (LGP) was used to segment the AVM nidus and areas of Onyx from postcontrast stereotactic MRI for 7 patients treated with GKRS postembolization. The resulting contours, skull surface, and clinically selected dose distributions were exported from LGP in DICOM-RT (Digital Imaging and Communications in Medicine-radiotherapy) format. Isocenter locations and dwell times were recorded from the LGP database. Contours were converted into 3D mesh representations using commercial and in-house mesh-editing software. The resulting data were imported into a Monte Carlo (MC) dose calculation engine (Pegasos, Elekta Instruments AB) with a beam geometry for the Gamma Knife Perfexion. The MC-predicted dose distributions were calculated with Onyx assigned manufacturer-reported physical constants (MC-Onyx), and then compared with corresponding distributions in which Onyx was reassigned constants for water (MC-water). Differences in dose metrics were determined, including minimum, maximum, and mean dose to the AVM nidus; selectivity index; and target coverage. Combined differences in dose magnitude and distance to agreement were calculated as 3D Gamma analysis passing rates using tolerance criteria of 0.5%/0.5 mm, 1.0%/1.0 mm, and 3.0%/3.0 mm. RESULTS Overall, the mean percentage differences in dose metrics for MC-Onyx relative to MC-water were as follows; all data are reported as mean (SD): minimum dose to AVM = -0.7% (1.4%), mean dose to AVM = 0.1% (0.2%), maximum dose to AVM = 2.9% (5.0%), selectivity = 0.1% (0.2%), and coverage = -0.0% (0.2%). The mean percentage of voxels passing at each Gamma tolerance were as follows: 99.7% (0.1%) for 3.0%/3.0 mm, 98.2% (0.7%) for 1.0%/1.0 mm, and 52.1% (4.4%) for 0.5%/0.5 mm. CONCLUSIONS Onyx embolization appears to have a detectable effect on the delivered dose distribution. However, the small changes in dose metrics and high Gamma passing rates at 1.0%/1.0 mm tolerance suggest that these changes are unlikely to be clinically significant. Additional sources of delivery and biological uncertainty should be investigated to determine the root cause of the observed less favorable postembolization GKRS outcomes.