2431 - Dosimetric Comparison of Photon and Proton Radiosurgery for Large Arteriovenous Malformations

R. Travis¹, M. Chadband¹, M. Soike¹, P. G. R. Schmalz², and M. C. Dobelbower¹; ¹University of Alabama at Birmingham Department of Radiation Oncology, Birmingham, AL, ²University of Alabama at Birmingham Department of Neurosurgery, Birmingham, AL

Purpose/Objective(s): Radiosurgery for large arteriovenous malformations can be limited by toxicity due to dose fall off into normal brain. Proton radiosurgery offers a theoretical benefit in dose fall off at the cost of greater uncertainty from robust planning requirements and fewer beam angles compared to photon arc therapy. This dosimetric study explores whether for the largest AVMs, pencil beam scanning radiosurgery (PBSR) would provide improved normal brain dosimetry and similar coverage compared to standard photon radiosurgery plans in treated patients. Materials/

Methods: The ten largest AVMs treated at a single institution with linear accelerator based photon radiosurgery were re-planned for PBSR in Varian Eclipse. Per institutional preference, the new proton plans assumed a 2% CT density correction and a 2 mm range uncertainty. Plans used 2-4 beams based on standard institutional practice for intracranial lesions and practical resource limitations at the proton unit. To maintain clinical comparability between radiosurgery and PBSR plans, isotoxicity concerning organs at risk (OARs) was first prioritized followed by equivalent target coverage. Several metrics for plan quality and surrogates for toxicity were compared using a paired 2-tailed T-test.
Results: Ten AVMs with median size 36.6cc (24.3cc - 60.1cc) were re-planned for PBSR and compared to their respective photon plans. All AVMs were prescribed 1750 cGy in one fraction. Calculated metrics for PBSR and photon plans with T-test p-values are listed in the table below. For one AVM abutting the chiasm, PBSR and photon V95% were 77.6% and 98.6% respectively when planned for optic isotoxicity. For another AVM abutting the brainstem, PBSR and photon V95% were 91.1% and 97.4% respectively when planned for brainstem isotoxicity.
Conclusion: In comparison to photon radiosurgery, PBSR planned with a 2% CT density correction and 2mm range uncertainty spills greater high dose into normal brain with higher gradient indices and poorer target coverage near OARs. While the normal brain mean dose and V5Gy significantly improved with PBSR, V12Gy appeared greater with PBSR. This is a known predictor of adverse events in several publications. In its current state, PBSR does not compare favorably with linac-based radiosurgery. Future work in PBSR should focus on mitigating uncertainty to allow for tighter robust proton plans. This could include implementation of dual energy CT simulation, greater number of proton fields, LET based planning, and evaluation of decreased range uncertainty.

Metric	Proton Mean Value	Photon Mean Value	Significance
Gradient Index	3.00	2.60	p=0.05
RTOG Conformity Index	0.99	1.01	p=0.70
Normal Brain V12Gy	32.0 cc	23.6 cc	p=0.06
Normal Brain V5Gy	84.0 cc	114.6 cc	p=0.01
Normal Brain Dmean	1.17 Gy	2.22 Gy	p<0.01
GTV V95%	95.8%	98.6%	p=0.24
GTV V100%	89.7%	95.0%	p=0.18