Abstracts

Rationale: Magnetoencephalography is a non-invasive diagnostic tool used in the presurgical evaluation of patients with medically refractory epilepsy. The Equivalent Current Dipole (ECD) method is the gold standard for localizing interictal spikes on MEG. We recently described SAMepi, a modification of SAM(g2), both beamforming methods for spike localization.1 In this study we evaluate localization of interictal spikes using each method in comparison with results of presurgical evaluation, and lobe of surgical resection. Methods: 44 subjects were retrospectively selected for inclusion in this study based on availability of a good quality MEG study, epilepsy surgery and at least one year seizure outcome data. SAMepi, SAM(g2) and ECD MEG results were compared to MRI findings, interictal and ictal scalp EEG localizations, ictal intracranial EEG localizations, and lobe of surgical resection. We then evaluated the effect of concordance of the MEG result with area of surgical resection to predict seizure outcome at one year postoperatively. Results: In our study, SAMepi, SAM(g2) and ECD have similar sensitivities for detection of interictal epileptiform activity (ECD and SAM(g2) 27/44(61%), SAMepi 26/44(59%)). Concordance of localizations using each method with lobe of surgical resection, interictal and ictal scalp EEG, ictal invasive EEG, and MRI are detailed in Figure 1 and were relatively similar. SAMepi, SAM(g)2 and ECD showed similar rates of concordance with areas of surgical resection (76.9%, 81.5%, and 77.8% respectively). Of note, a positive MEG result concordant with lobe of surgical resection was not predictive of seizure freedom, nor was a non-localizing result. However, when a positive MEG localization was obtained, resective surgery that did not include at least one of the identified areas was a predictor of poor surgical outcome (SAMepi 6/6, SAM(g2) 4/5, and ECD 5/6). Conclusions: Identification of epileptiform activity for clinical use traditionally involves localization of dipoles, a process that is labor and time intensive, as well as highly reliant on the experience of the interpreter. Beamforming methods such as SAMepi and SAM(g2) have a similar sensitivity to ECD and require significantly less time and expertise for interpretation. In this study, we found that each of these methods of MEG localization had a similar sensitivity for detection of interictal epileptiform activity, and had similar rates of concordance with seizure zone localization based on MRI lesions, interictal and ictal scalp EEG, ictal invasive EEG recordings and area of eventual surgical resection. Most importantly, surgical resection that was completely discordant with MEG localization was predictive of a lack of seizure freedom at one year. This should be taken into account when making surgical decisions in patients with positive MEG localization results. Of note, like many similar studies, this analysis is limited due to the relatively small sample size. Funding: NINDS Intramural Research Program