Unanticipated memory loss during superselective Wada testing for right cerebral arteriovenous malformation.

Abstract

120 Background: A major concern in the treatment of cerebral AVMs is whether so-called eloquent regions occupy normal anatomic locations. Recently, we have shown with selective anesthetic injections that language has a wide and unusual distribution in patients with left cerebral AVM. Our Objective here was to determine whether superselective Wada testing in medial right-hemisphere arteries would produce verbal memory loss in patients with right cerebral AVM. Methods: Nine patients, 8 right- and 1 left-handed, had superselective injections of amobarbital sodium plus lidocaine into vessels near or feeding right medial AVMs. Five patients had AVMs supplied by PCA feeders and four had ACA feeders. Memory testing occurred in a no-anesthetic baseline, 1 minute after anesthetic injection, and 12 minutes after injection. A memory test consisted of 5 words presented on a computer screen which the patient had to read aloud. After 3 minutes, recall was tested. A maximum score was 15, with uncued recall of a word = 3 points, prompted recall = 2 points, and multiple choice = 1 point. Based on clinical considerations, various numbers of vessels were tested in each patient. Results: Superselective anesthetic testing in most vessels yielded memory scores no different than baseline, and embolization took place. In each patient, however, there was at least one vessel feeding the AVM in which significant short-term verbal memory loss took place. In these instances, the mean memory score at baseline was 14 (s.d. = 1.5). Following injection of anesthetics, the mean score fell significantly to 7.1 (t-test, p<.00001). Attention was unaltered. After dissipation of the anesthesia at 12 minutes, memory returned to baseline levels in every case. Conclusions: In nine patients, superselective Wada injection into arteries feeding right medial AVMs produced unanticipated verbal memory deficits. Traditional assumptions about eloquent regions would have suggested that such testing would have been unnecessary, but embolization here would have yielded significant morbidity. Our findings also question established notions of verbal memory restricted to the left hemisphere.