Abstracts

Rationale: Cortical stimulation via intracranial electrodes is performed to help localize sites essential for language function in epileptic patients undergoing surgical resection. Yet, stimulation carries an inherent risk of electrically-induced seizures and sometimes fails to detect language-related regions in uncooperative or young patients. We hypothesize that measurement of language-related gamma-oscillations on ECoG can serve as an alternative mapping technique. We determined the sensitivity of ECoG- and stimulation-based methods for detecting language-related sites in patients with left-side language dominance as determined by Wada test.Methods: We studied 10 patients (age range: 8-37 years; mean age: 15.2 years) who underwent extraoperative ECoG, stimulation mapping, and Wada test. ECoG was recorded during a question-and-answer session, where each patient overtly verbalized an answer (e.g. Yellow ) to a given auditory question (e.g.: What is the color of banana? ). After time-frequency analysis, sites showing significant augmentation of gamma-oscillations (50-150 Hz) were determined and defined as language-related ECoG sites. Each patient subsequently underwent conventional stimulation mapping. Sites at which stimulation specifically resulted in auditory perception changes, failure to verbalize a correct answer, or sensorimotor symptoms involving the mouth were determined and defined as language-related stimulation sites. We determined how frequently stimulation- and ECoG-based methods detected language-related sites in the superior temporal region (Brodmann Area [BA] 22/41/42), dorsolateral prefrontal region (BA 44/45), dorsolateral premotor region (BA 6), and Rolandic region (BA 4/3/1/2). Results: ECoG and stimulation detected language-related sites in the superior temporal region with sensitivity of 10/10 (i.e.: 10 out of 10 patients) and 5/10, respectively. Likewise, sensitivity of ECoG and stimulation was 10/10 and 2/10 in the inferior dorsolateral prefrontal region, 9/10 and 9/10 in the dorsolateral premotor region, and 10/10 and 9/10 in the Rolandic region. Language-related sites in the superior temporal and dorsolateral prefrontal regions were detected by ECoG more frequently than stimulation (p<0.05 on Fisher's exact probability test). Gamma-augmentation reached the maximum level in the superior temporal gyrus during a question, in the dorsolateral prefrontal region around the offset of a question, in the dorsolateral premotor region prior to vocalization, and in the Rolandic region immediately following the onset of vocalization. Two patients required in-patient speech therapy due to aphasic symptoms following cortical resection of language-related ECoG sites not detected by stimulation. An unwanted generalized tonic-clonic seizure was induced by stimulation of sites outside of seizure onset zones in a patient.Conclusions: This study supports the hypothesis that measurement of language-related gamma-oscillations on ECoG can supplement cortical stimulation in preoperative localization of language-related sites.