Abstracts

Rationale: Human focal seizures produce clinical symptoms by disrupting neural coding. However, the precise nature of this disruption remains unclear. Recent evidence has suggested that the temporal order, or sequence, of firing among a population of neurons may also play a fundamental role in information processing in the human brain. We hypothesized that seizures would disrupt these sequences, providing a possible mechanism by which seizures create symptoms.

Methods: Five patients at our institution were implanted with micro-electrode arrays (MEAs). We recorded simultaneous intracranial EEG (iEEG), multi-unit spiking activity (MUA), and micro-scale local field potential (LFP) signals. We detected bursts by taking the mean smoothed spiking rate across all channels, and capturing instances in which this rate exceeded five standard deviations from the mean. We additionally used a previously-published algorithm to detect interictal and ictal discharges. Therefore, we captured three types of bursts: seizure bursts (bursts related to an ictal discharge), interictal bursts (bursts related to an interictal discharge) and baseline bursts (bursts related to no discharge).

Results: We found strong self-similarity among seizure bursts (using Spearman's rho), but not among IED bursts or baseline bursts. We then used PCA to reduce our high-dimensional sequence data to a low-dimensional manifold. We found that, at the time of seizure onset, seizure bursts reside in the same sub-region as baseline bursts. Shortly after seizure onset, however, seizure bursts abruptly travel to a distinct region of the subspace, where they persist through seizure termination. Next, we were interested in capturing the extent to which burst sequences are directional. Because our recorded neurons occupy different spatial locations within a small patch of cerebral cortex, sequences of spiking activity can also be characterized by their spatial organization. We found that, as seizure bursts diverge from baseline activity, they take on directional properties, indicated by a high goodness of fit (R²). Finally, we examined the directionality of ictal discharges in the LFP. We found that the LFP direction was strong and stable from the outset. MUA direction, on the other hand, was variable at seizure onset. Over the course of the seizure, however, MUA sequences changed their direction to match the direction of the LFP.

Conclusions: There is a wealth of recent work suggesting that ictal discharges travel over the brain surface as traveling waves (Diamond, et al 2021, Diamond, et al 2023). In the present work, we extend these findings to examine the influence of these macroscopic traveling waves on microscopic networks. We find evidence that ictal discharges entrain neuronal spiking sequences, by causing them to become dissimilar from baseline activity; stereotyped; and directional, where that direction is given by the direction of waves in the LFP. Our results may help explain why seizures create symptoms. Further, our results could help with decision-making and prognostication in clinical epileptology.

Funding: This work was supported by the Intramural Research Program of the National Institute for Neurological Disorders and Stroke.