Abstracts

Rationale: Seizures exhibit great diversity in their region of onset, their electrographic patterns, and their underlying pathology. It is natural to consider, then, whether certain types of seizures result from distinct neuronal mechanisms. Our current work explores the potential role of interactions between local and more distant networks during seizure generation. We measured network interactions by quantifying changes in cross-frequency coupling (CFC) in seizures with different electrographic patterns, arising from different regions to understand how the strength of these values might reflect clinically relevant information about the seizure’s neurophysiological dynamics. We then examined how closely CFC strength is associated with onset channel activity, and whether the clinically identified onset zones in recorded seizures reflect the actual onset zones. Methods: Seizures (n=378 from 43 patients) recorded from patients with medical refractory epilepsy who underwent presurgical evaluation with intracranial electrodes were analyzed. CFC between low (delta and theta) and high (ripples and fast ripples) frequency bands at seizure onset was measured using a generalized linear modeling (GLM) framework in MATLAB. To evaluate seizure dynamics in the onset channels, CFC was measured only in the seizure onset channel around the start of the seizure. In a subsequent analysis, a subset of seizures (n=48 from 15 patients) were selected to identify whether CFC strength has a stronger correlation with onset channels around the time of seizure initiation compared to the rest of channels. Results: We identified five different electrographic patterns in seizures arising from six different categories of cortical and subcortical regions. We found that while each region could give rise to seizures with multiple kinds of patterns, certain patterns were more likely to be associated with particular brain regions. Intriguingly, we found that changes in CFC dynamics between different bands were closely related to seizure onset region. Further, we found that in all patients with a beneficial surgery outcome (Engel classes I-II) and only one onset focus (5/15 patients), the channels that were clinically identified as the onset channels had much greater CFC values in their signature coupling pattern suggested by the first analysis (ie., ripple or fast ripple coupling). In cases where the clinical reviewer identified more than one onset focus, the CFC analysis could only identify the correct onset zone if the onset contact was observed to be in a region with a higher CFC value in our first finding (5/15 patients). For the remaining 5 patients CFC failed to identify the onset since the focus of interest was in a region with relatively lower CFC strength. Conclusions: Through our findings, we demonstrate the importance of seizure onset region as a factor that should be considered along with other clinical factors, such as electrographic pattern and pathology, in any mechanistic inference made about seizure generation. This study highlights the substantial differences between seizures arising from different regions in terms of what should be considered pathologic, and provides evidence that oscillatory networks recruited during seizure initiation can be distinct between regions and in order to employ relevant spectral information a normalization to account for these differences should be taken into account. Funding: NIH grants R01-NS062092, 1K24NS088568-01A1, R01-NS079533, R01_NS072023MGH-ECORFonds de Recherche Santé Québec (FRSQ)