Despite the success of stereotactic radiosurgery, large inoperable arteriovenous malformations (AVMs) of 14 cm3 or more have remained largely refractory to stereotactic radiosurgery, with much lower obliteration rates. We review treatment of large AVMs either previously untreated or partially obliterated by embolization with fractionated stereotactic radiotherapy (FSR) regimens using a dedicated linear accelerator (LINAC).

Before treatment, all patients were discussed at a multidisciplinary radiosurgery board and found to be suitable for FSR. All patients were evaluated for pre-embolization. Those who had feeding pedicles amenable to glue embolization were treated. LINAC technique involved acquisition of a stereotactic angiogram in a relocatable frame that was also used for head localization during treatment. The FSR technique involved the use of six 7-Gy fractions delivered on alternate days over a 2-week period, and this was subsequently dropped to 5-Gy fractions after late complications in one of seven patients treated with 7-Gy fractions. Treatments were based exclusively on digitized biplanar stereotactic angiographic data. We used a Varian 600SR LINAC (Varian Medical Systems, Inc., Palo Alto, CA) and XKnife treatment planning software (Radionics, Inc., Burlington, MA). In most cases, one isocenter was used, and conformality was established by non-coplanar arc beam shaping and differential beam weighting.

Thirty patients with large AVMs were treated between January 1995 and August 1998. Seven patients were treated with 42-Gy/7-Gy fractions, with one patient lost to follow-up and the remaining six with previous partial embolization. Twenty-three patients were treated with 30-Gy/5-Gy fractions, with two patients lost to follow-up and three who died as a result of unrelated causes. Of 18 evaluable patients, 8 had previous partial embolization. MeanAVM volumes at FSR treatment were 23.8 and 14.5 cm3, respectively, for the 42-Gy/7-Gy fraction and 30-Gy/5-Gy fraction groups. After embolization, 18 patients still had AVM niduses of 14 cm3 or more: 6 in the 7-Gy cohort and 12 in the 5-Gy cohort. For patients with at least 5-year follow-up, angiographically documented AVM obliteration rates were 83%for the 42-Gy/7-Gy fraction group, with a mean latency of 108 weeks (5 of 6 evaluable patients), and 22% for the 30-Gy/5-Gy fraction group, with an average latency of 191 weeks (4 of 18 evaluable patients) (P = 0.018). For AVMs that remained at 14 cm3 or more after embolization (5 of 6 patients), the obliteration rate remained 80% (4 of 5 patients) for the 7-Gy cohort and dropped to 9% for the 5-Gy cohort. A cumulative hazard plot revealed a 7.2-fold greater likelihood of obliteration with the 42-Gy/7-Gy fraction protocol (P = 0.0001), which increased to a 17-fold greater likelihood for postembolization AVMs of 14 cm3 or more (P = 0.003).

FSR achieves obliteration for AVMs at a threshold dose, including large residual niduses after embolization. With significant treatment-related morbidities, further investigation warrants a need for better three-dimensional target definition with higher dose conformality.