Abstracts

Rationale: The goal of early epilepsy surgery is closely linked with improving cognitive outcome. Thus, identification of eloquent language cortex prior to resection is essential. The high occurrence of cognitive deficits in pediatric patients creates more challenges to accurate language mapping than in adults. Subdural ESM can invoke seizures and may be sparsely applied, yielding incomplete results. fMRI is a noninvasive alternative but it indirectly measures neuronal activity, motion artifacts lead to unreliable data, and activation patterns reveal networks that may be non-essential to language function. Despite their routine use, comparison of fMRI and ESM is limited to a few adult case series. Here we report concordance data between fMRI and ESM in a pediatric sample. Methods: Ten patients (6 F 4M, 11-18 yo, 6 RH) underwent fMRI and ESM procedures. fMRI language mapping involved Verb Generation (VG) and Story Listening (SL) paradigms performed on a 3-T GE scanner. ESM was conducted using an Ojemann stimulator with amperages ranging from 4-10 milliamps (50 Hz; 5 sec.). fMRI and post-implantation CT images were fused to allow for direct comparison. Results: fMRI during VG elicited left inferior frontal gyrus (IFG) activation in right handed patients. Bilateral or right IFG activation was observed in left-handed patients. A total of 346 contacts were stimulated across patients; 43 produced a positive response. In the sample, 43/117 positive contacts were located within an active fMRI cluster using a p <.01 threshold (Bonferroni). Among the 229 negative contacts, 41 were located within an active fMRI cluster and 46 were not. The overall sensitivity of language fMRI was .36 and the specificity was .82. The false positive rate (+fMRI activation; - ESM finding) was .17, and the false negative rate was .63. Conclusions: We employed a novel co-registration method allowing for analysis of concordance between ESM and fMRI in pediatric epilepsy patients. Language fMRI did not activate many critical regions displayed in ESM, and there was significant inter-patient variability. Concordance was slightly higher in frontal regions but less than acceptable. These data are noteworthy since a high false negative rate in patients with "dominant" hemisphere epilepsy was found, a group known to be at risk for post-surgical language deficits. Our results suggest that fMRI is a complementary tool to ESM and is helpful in guiding intracranial electrode placement to ensure coverage of language cortex but may not be adequate alone to guide surgical decisions. We safely detected language cortex in all patients employing a low level intensity during ESM, which is contrary to data that suggests that ESM in children is ineffective; only provokes responses at high amperage levels; or readily induces seizures (1/10 patients had a seizure). Future aims include fMRI analyses at different thresholds to better determine agreement between modalities. Determining language outcomes based on predictions made by fMRI and ESM is needed.