Objective: In this study, we developed an a priori system to stratify surgical intervention of perisylvian arteriovenous malformations (AVMs) in 20 patients. We stratified the patients into three categories based on preoperative functional magnetic resonance imaging (fMRI) language activation pattern and relative location of the AVM.
Methods: In Group I (minimal risk), the AVM was at least one gyrus removed from language activation, and patients subsequently underwent asleep resection. In Group II (high risk), the AVM and language activation were intimately associated. Because the risk of postoperative language deficit was high, these patients were then referred to radiosurgery. In Group III (indeterminate risk), the AVM and language were adjacent to each other. The risk of language deficit could not be predicted on the basis of the fMRI alone. These patients underwent awake craniotomy with electrocortical stimulation mapping and optical imaging of intrinsic signals for language mapping.
Results: All patients from Group I (minimal risk) underwent asleep resection without deficit. All Group II (high-risk) patients tolerated radiosurgery without complication. In Group III (indeterminate risk), three patients underwent successful resection, whereas two underwent aborted resection after intracranial mapping.
Conclusion: We advocate the use of fMRI to assist in the preoperative determination of operability by asleep versus awake craniotomy versus radiosurgery referral. In addition, we advocate the use of all three functional mapping (fMRI, electrocortical stimulation mapping, and optical imaging of intrinsic signals) techniques to clarify the eloquence score of the Spetzler-Martin system before definitive treatment (anesthetized resection versus radiosurgery versus intraoperative resection versus intraoperative closure and radiosurgery referral).