Abstracts

Rationale: EEG/fMRI can noninvasively measure the hemodynamic response related to interictal epileptic discharges (IEDs) on scalp EEG. The clinical significance of EEG/fMRI is still under assessment. This study aims to correlate BOLD response related to IEDs with the localization of surgical resection to determine the conditions in which surgical outcome can be predicted better. Methods: All consecutive refractory focal epilepsy patients who underwent surgical treatment after EEG/fMRI from June 2006 to December 2010 and with 12 months postsurgical follow up were included. Each patient underwent a 2hr 3T scan. IEDs were visually identified. A statistical t map of BOLD response was generated using general linear model. EEG/fMRI results were not used for surgical decision making. BOLD response with the maximum t-value (t-max, activation or deactivation) was correlated with the resection. The concordance between the resection and BOLD response was evaluated by co-registering individual t map with postoperative MRI or, when not available, with postoperative CT or operative protocol. Surgical outcome was defined according to Engel's classification: Class 1 & 2 was considered a good outcome and Class 3 & 4 a poor outcome. Results: Forty-six patients had surgery after EEG/fMRI study, 13 were excluded: 4 had no IEDs in the scanner and 9 showed no significant BOLD response during IEDs. Thirty-three patients were included (8 non-lesional) and classified in 4 groups defined by the concordance between resection and t-max. Group 1, fully concordant (Fig. 1): 10 patients with t-max concordant and response mostly confined to resection area. 8 of these 10 patients had a good surgical outcome. Group 2, partially concordant: 8 patients with t-max at the immediate edge of the resection, and response usually as part of a larger cluster overlapping or adjacent to the resection. 4 (50%) with well confined cluster had a good outcome, and 4 had a poor outcome but in those the resection included only a small portion of a widespread BOLD response. Group 3, partially discordant: 6 patients with t-max remote from resection, but with an additional less significant BOLD cluster in the resection. Half of the patients had a good and half a poor outcome. Group 4, fully discordant (Fig. 2): 9 patients with no significant BOLD response in the resection. Only 1 patient from this group had a good outcome. Conclusions: In patients with fully concordant or fully discordant BOLD response and resection (19/33, or 58%), EEG/fMRI predicted the surgical outcome. These findings suggest that complete removal of the maximum BOLD response should lead to a seizure-free condition. The study also indicates that a widespread BOLD response is probably associated with a large epileptogenic zone, in which condition a small resection should not lead to a favorable outcome. Finally, the concomitant occurrence of less significant clusters but localized in the suspected epileptogenic zone should not be neglected as their removal sometimes also led to a good outcome. Support by Canadian Institutes of Health Research.