Three-dimensional assessment of the effects of high-density embolization material on the absorbed dose in the target for Gamma Knife radiosurgery of arteriovenous malformations

OBJECTIVE Arteriovenous malformation (AVM) is an intracranial vascular disorder. Gamma Knife radiosurgery (GKRS) is used in conjunction with intraarterial embolization to eradicate the nidus of AVMs. Clinical results indicate that patients with prior embolization tend to gain less benefit from GKRS. The authors hypothesized that this was partly caused by dosimetric deficiency. The actual dose delivered to the target may be smaller than the intended dose because of increased photon attenuation by high-density embolic materials. The authors performed a phantom-based study to quantitatively evaluate the 3D dosimetric effect of embolic material on GKRS. METHODS A 16-cm-diameter and 12-cm-long cylindrical phantom with a 16-cm-diameter hemispherical dome was printed by a 3D printer. The phantom was filled with radiologically tissue-equivalent polymer gel. To simulate AVM treatment with embolization, phantoms contained Onyx 18. The material was injected into an AVM model, which was suspended in the polymer gel. The phantom was attached to a Leksell frame by standard GK fixation method, using aluminum screws, for imaging. The phantom was scanned by a Phillips CT scanner with the standard axial-scanning protocol (120 kV and 1.5-mm slice thickness). CT-based treatment planning was performed with the GammaPlan treatment planning system (version 10.1.1). The plan was created to cover a fictitious AVM target volume near the embolization areas with eleven 8-mm shots and a prescription dose of 20 Gy to 50% isodose level. Dose distributions were computed using both tissue maximum ratio (TMR) 10 and convolution dose-calculation algorithms. These two 3D dose distributions were compared using an in-house program. Additionally, the same analysis method was applied to evaluate the dosimetric effects for 2 patients previously treated by GKRS. RESULTS The phantom-based analyses showed that the mean dose difference between TMR 10 and convolution doses of the AVM target was no larger than 6%. The difference for GKRS cases was 5%. There were small areas where a large dose difference was observed on the isodose line plots, and those differences were mostly at or in the vicinity of the embolization materials. CONCLUSIONS The results of both the phantom and patient studies showed a dose reduction no larger than 5% due to the embolization material placed near the target. Although the comparison of 3D dose distributions indicated small local effects of the embolic material, the clinical impact on the obliteration rate is expected to be small.