Seizure onset location shapes dynamics of initiation

Highlights

• High-frequency oscillations and phase-amplitude coupling have region-specific dynamics.

• Infraslow activity occurs throughout the brain during seizures regardless of seizure onset zone.

• Onset region is best considered with pattern and other aspects when studying seizure mechanism.

Abstract

Objective

Ictal electrographic patterns are widely thought to reflect underlying neural mechanisms of seizures. Here we studied the degree to which seizure patterns are consistent in a given patient, relate to particular brain regions and if two candidate biomarkers (high-frequency oscillations, HFOs; infraslow activity, ISA) and network activity, as assessed with cross-frequency interactions, can discriminate between seizure types.

Methods

We analyzed temporal changes in low and high frequency oscillations recorded during seizures, as well as phase-amplitude coupling (PAC) to monitor the interactions between delta/theta and ripple/fast ripple frequency bands at seizure onset.

Results

Seizures of multiple electrographic patterns were observed in a given patient and brain region. While there was an increase in HFO rate across different electrographic patterns, there are specific relationships between types of HFO activity and onset region. Similarly, changes in PAC dynamics were more closely related to seizure onset region than they were to electrographic patterns while ISA was a poor indicator for seizure onset.

Conclusions

Our findings suggest that the onset region sculpts neurodynamics at seizure initiation and that unique features of the cytoarchitecture and/or connectivity of that region play a significant role in determining seizure mechanism.

Significance

To learn how seizures are initiated, researchers would do well to consider other aspects of their manifestation, in addition to their electrographic patterns. Examination of onset pattern in conjunction with the interactions between different oscillatory frequencies in the context of different brain regions might be more informative and lead to more reliable clinical inference as well as novel therapeutic approaches.

Introduction

Understanding seizure generation is a crucial step in tailoring pharmacological, surgical and neuromodulatory approaches to epilepsy. Yet, seizures are extraordinarily diverse, resulting from a wide range of disorders, and their underlying mechanism or mechanisms remain unknown. Comparing the underlying neural dynamics that precede seizure activity has been one of the key approaches to trying to delineate underlying mechanisms. Changes in spectral content within defined frequency bands that occur around the time of a seizure indicate abnormal neuronal network activity (Imamura et al., 2011, Worrell et al., 2012, Wu et al., 2014) yet, our knowledge of the relationship between these spectral modulations and their specific roles in seizure generation is limited. Furthermore, our understanding of how these types of activities at seizure onset might depend on the originating brain structure itself. Such specificity would be useful in seizure localization and better understanding mechanism and ultimately tailoring therapy. Among these various changes, high-frequency oscillations (HFOs), further delineated as ripples (80–200 Hz) and fast ripples (250–500 Hz), as well as changes in the power of infraslow activity (ISA, 0.02–0.5 Hz), have attracted particular attention (Vanhatalo et al., 2003, Jacobs et al., 2009, Zijlmans et al., 2012).

Since their first description, HFOs are believed to be associated with epileptogenesis (Bragin et al., 1999). Findings that (a) the appearance of HFOs is strongly associated with seizure onset, and (b) removing areas with higher HFO rates leads to better surgical outcomes (Jacobs et al., 2010, Zijlmans et al., 2011, Zijlmans et al., 2012, van’t Klooster MA, van Klink NEC, Zweiphenning WJEM, Leijten FSS, Zelmann R, Ferrier CH, et al., 2017, Cimbalnik et al., 2018), are particularly interesting and hold clinical relevance. Despite this and other evidence for the diagnostic utility of HFOs, only a few studies have investigated HFO incidence during ictogenesis. Quite a few of these studies have reported that the power of HFOs are augmented during a seizure (Jacobs et al., 2009, Khosravani et al., 2009, Nariai et al., 2011). Animal model studies found specific changes in particular frequency bands associated with different seizure onset patterns, with ripples presenting more often in seizures with low-voltage fast activity (LVF) pattern onsets and fast ripples predominant in seizures with hypersynchronous spiking (HYP) onsets (Lévesque et al., 2012, Salami et al., 2015). Although these investigations suggest distinct HFO band-pattern associations with different seizure onset patterns, such studies in patients have yielded conflicting results (Perucca et al., 2014, Ferrari-Marinho et al., 2016, Frauscher et al., 2017).

There is evidence to suggest that the occurrence rate of HFOs may depend on the regions of the brain being studied (Cimbalnik et al., 2018). Researchers have addressed the epileptogenicity of certain regions by assessing their tendency to generate rapid discharges (Bartolomei et al., 2008). Furthermore, electrographic patterns from recordings of the scalp appear to encode properties about the seizure onset region, depending on which specific region is generating the seizure. (Ebersole and Pacia, 1996, Tanaka et al., 2018). In the presence of such reports, the extent to which HFO changes in different regions can be used to inform seizure diagnostics remains an open question.

On the other end of the spectrum, ISA may also be informative of the epileptogenic zone (Rodin et al., 2014, Wu et al., 2014, Thompson et al., 2016) and indicate important circuit activity underlying seizure initiation. Although its appearance in the field predates that of HFOs (Gumnit and Takahashi, 1965), it has not been widely studied. ISA changes may reflect fluctuations in extracellular potassium concentration or glial cell depolarization in response to such concentration changes (Kanazawa et al., 2015), and these might be necessary for generating seizures (Jirsa et al., 2014). However, reports investigating ISA at seizure onset have had conflicting findings with either widespread (Rodin and Modur, 2008) or localized (Ikeda et al., 1999) occurrence of the ISA. Further, while ISA has been reported at the onset of some seizures, ISAs have not been found in all seizures (Bragin et al., 2007, Kanazawa et al., 2015). These divergent reports leave doubts as to the importance of ISA as a biomarker of seizure onset zone or its role in seizure generation.

Most importantly, for both high and low frequency activity, the heterogeneity of existing findings might reflect confounds of different seizure etiologies, electrographic patterns, interactions of different oscillations or anatomic substrates. To disambiguate these potential conflicts, we explored changes in HFOs and ISA at the onset of different seizures and the potential role of interactions between different spectral components in seizure generation. HFOs tend to be relatively focal (Jefferys et al., 2012), and so may not fully reflect the spatial extent of neuronal activity. To assess the interactions between local and more distant networks, alongside an investigation of ISA and HFOs at seizure onset, we quantified changes in cross-frequency coupling (CFC) of low (delta and theta) frequency bands corresponding to ripple and fast ripple frequency ranges in seizures with different onset patterns arising from different regions. The strength in CFC is believed to reflect the level of interaction between large-scale brain networks and more local networks (Canolty and Knight, 2010). Since different regions possesses a variety of neuronal populations, the interactions between them can vary from a region to another (Motoi et al., 2019), we hypothesize that the features we measure in this study will be region specific. We then used the results of this analysis to explore how these changes are correlated with origin of seizures and the role they may play in seizure onset.