Abstracts

Rationale: Magnetoencephalography (MEG) has several advantages over scalp electroencephalography (EEG) for the localization of epileptic sources, one of which is MEG's insensitivity to skull defects and other cranial anatomic alterations. Patients with epilepsy who have had previous neurosurgery are difficult to evaluate with non-invasive EEG because of the alterations in the electrical field produced by cranial anomalies from the prior surgery, or invasively due to post-surgical adhesions and the risk of hemorrhage. Our study aims :1) to quantify, in the presence of skull defects, the ability of MEG to localize sources by comparison to intracranial recording and surgical outcome; and 2) to investigate the possibility that such patients can proceed directly to resective surgery guided only by MEG and one other non-invasive test, such as Ictal SPECT or PET. Methods: We included all patients with prior craniotomies, who then had MEG studies, and later underwent resective epilepsy surgery. We reviewed the non-invasive data, intracranial EEG, and seizure outcome after a minimum 6 months follow-up using Engel classification. Patients whose only previous neurosurgery was a SEEG not followed by resection were excluded. Results: Among 736 patients undergoing MEG from 2008 to 2013, 76 patients had previous neurosurgery involving craniotomy with or without resection and were re-operated to remove the putative seizure focus after another evaluation which included MEG, scalp video EEG (VEEG) and MRI , and in some patients SPECT, PET and invasive VEEG. 47/76 patients achieved Engel class I (for follow up duration) and were seizure free during follow-up from 6 months to 4 years, and 8/76 patients had Engel class II epilepsy surgery outcome (total 55 patients with "good outcome"). Among these patients localization by MEG was concordant but more precise than scalp VEEG in 23/55 patients; provided localizing information not available from EEG in 12/55 patients; scalp VEEG was more accurate in 16/55 patients (13 of them because of a normal MEG). Four patients had normal or non-localizable both MEG and scalp EEG. Thirty nine of 55 patients with Engel class I and II (good outcome) had positive MEG; 31 of them were evaluated with Ictal SPECT and/or PET with concordant results in 27(87%) patients, partially concordant in 1 and discordant in 3 patients with multiregional MEG and MEG with bilateral clusters. Twenty one of 39 MEG-positive patients (Engel class I and II) underwent invasive VEEG evaluation, which was concordant with MEG in 20(95%) and discordant in only 1 patient. Conclusions: A MEG study with only the conventional duration of one hour demonstrated high concordance with non-invasive imaging tests and multi-day scalp VEEG in patients with previous neurosurgery who achieved good surgery outcome. In these patients who also underwent intracranial evaluation, the invasive localization was highly concordant with the MEG results. These findings suggest an important role of MEG in the pre-surgical evaluation process, and that MEG may reduce the need for invasive evaluation in selected cases.