Abstracts

Functional MRI has shown promise for lateralizing language in pre-surgical epilepsy patients. Previous studies found a strong correlation between Wada language scores and fMRI brain activation generated from a Semantic Decision (SD) task. Such findings indicated these techniques have similar lateralization value. A potential drawback of relying exclusively on SD as an activation paradigm for presurgical language mapping is the minimal activation of the anterior temporal lobe produced by SD. As a result, the ability of SD to predict postoperative decline may be limited in individual subjects, despite being effective in language lateralization. A paradigm that has shown promise in producing fMRI activation of the anterior temporal lobe involves having subjects generate names to aurally delivered definitions. Such Definition Naming (DN) paradigms may have potential for better prediction of postoperative language outcome than SD. To better understand the value of DN in determining language dominance and activation of language zones, DN and SD are compared to a Wada language index. Forty-three patients with epilepsy were selected who had valid Wada testing and fMRI studies using DN and SM. Wada testing used a standardized administration and was quantitatively scored to yield an index of language dominance. Two fMRI language activation protocols were completed. In the SD protocol, patients heard the name of an animal and pressed a button if the animal was native to the United States and used by humans. In the DN protocol, patients were asked to speak the name of each of 128 aurally delivered definitions. Appropriate control tasks were contrasted with the experimental paradigms. Laterality indices for fMRI tasks were calculated by tabulating the number of task-contrast voxels of activation in left and right hemispheres and selected ROIs: LI = (V[sub]L [/sub]- V[sub]R[/sub])/(V[sub]L[/sub] + V[sub]R[/sub]). The Wada LI was found to significantly correlate with SD, r(41)=.67, p[lt].001, and the DN, r(41)=.72, p[lt].001. Despite DN requiring speech production but SD having no speech component, the LI[rsquo]s produced by the two tasks were significantly correlated with each other, r(43)=.78, p[lt].001. When voxel counts from various brain regions were examined, DN produced greater activation in dominant anterior temporal regions than the SM, t(42)= 7.1, p[lt].001. As found in previous studies, SD and Wada language laterality index were significantly correlated. A new finding was that a DN task that requires speech generation was also significantly correlated with the Wada. In addition, DN produced greater activation of the dominant anterior temporal lobe. These findings indicate that a definition naming task may be a more useful method in predicting postoperative language functioning than a semantic decision task. (Supported by 1RO1 NS35929-04, 1RO1 NS33576, NIH M01-RR00058)