To report a single-institution experience with intracranial arteriovenous malformations (AVMs) treated utilizing a linear accelerator-based patient rotator (PR) or BrainLAB (BL) radiosurgery system (BrainLAB AG, Heimstetten, Germany).

Since 1989, 84 evaluable patients were treated. PR patients (n = 45) were planned/localized on the basis of biplane angiography and treated between 1989 and 2000. BL patients (n = 39) were planned/localized on the basis of CT/MRI and treated since 2000. Kaplan-Meier analyses of survival, nidus obliteration (NO), and any radiographic improvement were undertaken with Cox regression of dose and volume effects.

No significant complication, survival, previous embolization incidence, AVM location or size differences existed between BL/PR patients. The groups differed significantly in prescribed dose (PR: 16.2 Gy, BL: 17.3 Gy, p = 0.004) and isodose (PR: 62%, BL: 79%, p < 0.0001). Estimated 2-year NO rate was 87% for BL patients, 12% for PR patients (p < 0.0001). Ultimate PR NO rate was 67% at 6 years. Dose (p = 0.037) and isodose (p = 0.014) significantly affected PR NO outcome; volume was of borderline significance (p = 0.069). No factors significantly affected BL outcome. Analyses of small (< or = 4.0 cm3), high-dose (> or = 17.0 Gy) PR patients (PR1 group) vs. BL patients still demonstrated greater NO (p = 0.04) and radiographic improvement (p = 0.0004) rates for the BL group. PR1 patients had a 76% 3-year NO rate.

BL-based radiosurgery achieved a high NO rate, the PR method did not. Differences in outcomes between PR/BL groups may be due to localization methods or an inherent advantage with the BL system.