Abstracts

Rationale: Temporal lobe epilepsy (TLE) presents with a high percentage of refractoriness and for these patients a thorough investigation is necessary to define the epileptogenic zone for surgical treatment[1]. Simultaneous EEG-fMRI has been proposed as a non-invasive pre-surgical evaluation tool for patients with focal refractory epilepsy[2,3]. However, the technique's specific role in different types of focal epilepsy remains to be determined. In this study we tried to evaluate the value of EEG-fMRI as a pre-surgical evaluation tool in patients with refractory TLE and assess its potential post-surgical outcome prediction value. Methods: Fifty-eight patients with diagnosis of refractory TLE and investigated for epilepsy surgery were invited to undergo EEG-fMRI between July/2007 and March/2012. All patients underwent 20-48 minutes EPI sequences in 1.5 or 3T MRI scanners. EEG was acquired using a cap with 64 MR-compatible electrodes and corrected for gradient and pulse-related artifacts. fMRI data was analyzed with SPM8. Spikes were identified on the scanner EEG and used to build a model of BOLD changes in a general linear model (GLM). In addition the cross-correlation between the EEG topographic map calculated from the average of spikes recorded during long term video telemetry outside the scanner and the map at each scanner EEG time point was added as a condition[4], following convolution with a haemodynamic kernel. Thus it allowed the analysis of all data irrespective of the presence or absence of spikes on the scanner EEG. The degree of concordance of the EEG-fMRI maps was defined according to the presence of the global statistical maximum BOLD voxel in relation to the temporal lobe ipsilateral to the resected area (Concordant); the presence of a significant cluster in the temporal lobe ipsilateral to the resected area but with the global maximum in a different region (Some Concordance); or no significant cluster in the temporal lobe ipsilateral to the resected area (Discordant). Surgical outcome was accessed according to ILAE classification, ILAE criteria 1-3 being considered good surgical outcomes and ILAE 4-6, poor outcomes. Results: From the 58 TLE patients who underwent EEG-fMRI, 23 (40%) underwent surgery. All were submitted to unilateral temporal lobe resection of different extent according to the pathology. The mean follow-up period after surgery was 22 months (range: 1-53 months). Twelve patients undergoing surgery had spikes inside the scanner and 11 did not. Significant spike-related BOLD changes were revealed in all but one case. Four patients (22%) were classified as Concordant results, and of those, 3 (75%) had a good surgical outcome (ILAE 1-3). Nine patients were classified as Some Concordance and six of these (67%) had a good surgical outcome. Two of the nine patients (22%) classified as Discordant had a good surgical outcome. Conclusions: EEG-fMRI can provide useful information for the pre-surgical evaluation of patients with TLE and resection of the temporal lobe including spike-related significant BOLD changes may lead to better postoperative outcome.