Abstracts

Rationale: Functional Magnetic Resonance Imaging (fMRI) has been used for preoperative brain mapping in patients with epilepsy due to hippocampal sclerosis (HS). The use of more than one language paradigm allows a more robust identification of language areas. We evaluated language activation patterns in patients with left temporal lobe epilepsy with HS, with four different language paradigms assessing different language aspects. Methods: Seven right handed patients with epilepsy secondary to left HS and seven right handed normal controls underwent an fMRI study using four distinct paradigms: word generation (WG), visual confrontation (CN) and written sentence responsive naming (RN), as well as phonologic and semantic language decision (LD). Data were collected with a 3T MRI scanner with compressed GRE EPI BOLD images (40 ACPC oriented slices, 3.3mm isometric voxels, TR=4s, TE30ms/ FA90). All paradigms were block designed, each with a five minutes duration. Data analysis was performed with XBAM (http://www.brainmap.co.uk), using a nonparametric statistical inference approach. Patients and controls activation maps were compared with ANOVA, with a significance level of p<0.05. Results: Both patients and controls group maps showed increased BOLD effect in the frontal gyri (bilateral for WG; and left for all others), temporal-occipital (bilateral for WG and RN and left for all others), SMA/cingulate gyrus, and left parietal cortex. On WG, the patients group showed an increased number of activated regions in the right hemisphere, and an overall decreased BOLD effect in the homologous regions on the left side. On VCN, the patients group showed increased activation in the right temporal-parietal and right parietal regions, posterior cingulate and left middle frontal gyrus, and a reduced activation in the left middle frontal, left inferior parietal gyri, right inferior temporal gyrus, as well as right anterior cingulus and right frontal-basal region. On RN, the patients group map showed decreased number of clusters in all the above listed regions; and an increased number of clusters in right inferior and middle frontal gyri, bilateral temporal regions and left cerebellum. On LD, the patients group map showed increased number of clusters in the right hemisphere (inferior and middle frontal, superior temporal and inferior parietal regions) and an overall decrease in the BOLD effect at the homologous clusters, in the left temporal-parietal regions and right temporal pole, right frontal-basal region and right inferior frontal gyrus. Conclusions: We found a consistent pattern of language network reorganization in TLE-MTS throughout the tasks, with decreased activation of left hemisphere language related regions, with increased activation of areas neighboring language areas and extensive activation of contralateral homologous brain regions. Mesial temporal lobe epilepsy associated with mesial temporal sclerosis appears to cause extensive at-distance reorganization of the language network.