Abstracts

Rationale: Patients with suspected supplementary motor area (SMA) epilepsy often require intracranial video electroencephalogram (vEEG) studies utilizing large interhemispheric subdural electrodes, both for localization of the seizure onset zone, as well as detailed mapping of eloquent cortex for operative planning. We gathered detailed brain mapping data from these patients to map the human SMA. We also tested the hypothesis that patients with seizure onset zone in the SMA have altered brain mapping of the SMA compared to control patients who did not have SMA epilepsy. Methods: Following institutional review board approval, a prospectively maintained database of all patients who underwent long-term intracranial electrode vEEG monitoring as part of a phase II epilepsy work-up from 1992 to 2017 was queried for all cases in which interhemispheric electrodes were implanted. Patients who did not have adequate records of awake brain mapping were excluded from the study. Twenty-five patients met inclusion criteria. Pertinent demographic, clinical and imaging variables were extracted for analysis including age, gender, duration of epilepsy, histopathology, post-operative course, duration of follow-up and Engel class outcome, among others. Detailed cortical mapping data from bipolar stimulation (50 Hz, 300-µs pulse width, 5-second train) were normalized to the Montreal Neurologic Institute atlas and the complex motor/sensory responses were analyzed and used to construct an SMA homunculus. Patients were then divided into two categories based on seizure onset zone – SMA (n=13) and non-SMA (n=12). These subgroups were subjected to further analysis. All motor responses to cortical simulation were grouped into five categories: head, trunk, upper extremity, lower extremity, or complex lower extremity (lower extremity movement and movement of another body part). Functional centers for four somatotopic categories (head, trunk, upper extremity and lower extremity) were computed by pooling data across subjects for SMA and non-SMA groups. In order to determine whether the underlying dispersion of mapped electrodes was similar between SMA and non-SMA patients, MANOVA analysis was performed for the geometric center between all mapped interhemispheric electrodes in each group. The Kolmogorov-Smirnov test was also performed between SMA and Non-SMA electrode locations to test for sampling bias. Differences in the distribution of each category of mapped responses on the cortical surface were tested in SMA and Non-SMA patients using a two-tailed student’s t-test. Differences between the cortical locations of each of the functional centers computed above, as well as the relative location of the functional centers with respect to each other (i.e. the somatotopy of the functional centers) for SMA versus non-SMA patients were analyzed using a stratified bootstrap method to mitigate the issues that arise due to the different sample sizes of the two groups. We also weighted mapped responses by the inverse of the stimulation current to generate current-weighted functional centers. To test whether the homunculus of the SMA patients changed significantly compared to the non-SMA patients, the variance of both the y-coordinates and z-coordinates were calculated for the current-weighted functional responses relative to each other and a fisher’s T-test was used to assess for significance. For all tests, a p-value of <0.05 was considered significant, except for the student’s t-test for which we used a Bonferroni corrected value of p <0.0083. Results: The SMA group was composed of 13 individuals who had a total of 527 interhemispheric electrodes implanted with 284 mapped responses. The Non-SMA group had 12 individuals with a total of 409 interhemispheric electrodes with 168 mapped responses. Univariate analysis of baseline demographic, clinical and imaging characteristics showed that complex partial seizures were significantly more common in the non-SMA group (p=0.039), and that the number of AEDs at the time of surgery were significantly higher for the SMA group (p= 0.002). Age at onset, duration of epilepsy, gender, normal MRI, and presence of focal cortical dysplasia were similar between the two groups. The distribution of electrodes and the sampling rate between the two groups was not significantly different (p=0.532). Analysis of the clustering and geographic center of functional responses between subjects demonstrated that the distribution of only lower extremity responses was significantly changed between SMA and non-SMA (p=0.00012). Further, MANOVA analysis of mapping of current-weighted functional centers demonstrated that there was a significant difference in the locations of each of the current-weighted functional centers between the SMA and Non-SMA groups (p < 0.01). These differences were further validated using stratified bootstrapping to permute the coordinates of positive responses between the two groups. Lastly, we illustrated the relative positions of each functional group, with regard to the other functional groups, for both the SMA and Non-SMA responses to create a homunculus. We discovered that the SMA group tended to have dispersed functional group clusters, while the Non-SMA group tended to have overlapping functional group clusters. The relative positions of each functional group were significantly different in SMA versus non-SMA patients, thus providing evidence that there is functional re-mapping of the SMA in patients with focal SMA epilepsy. Conclusions: Seizure onset in the SMA changes the underlying mapping of functional regions in the SMA compared to patients who have seizure onset outside of the SMA. This is a novel observation of functional plasticity in eloquent cortex in the setting of juvenile-onset epilepsy. Funding: No funding