Abstracts

Rationale: Neocortical and temporal limbic areas typically exhibit at least one narrowband local field potential (LFP) rhythm (e.g., 9-11 Hz) in the 1-35 Hz range. Simultaneous single unit and LFP recordings in animals and humans have often found an antagonistic relationship between the magnitude of these LFP oscillations and neural firing. At a macroscopic level, the power of the dominant 1-35 Hz rhythm is generally anti-correlated with broadband intracranial electroencephalography (iEEG) power (a proxy for local neural spiking rate). This suggests that LFP rhythms may reflect modulations in circuit excitability that gate inter-areal communication in normal brain function. Here we investigate the possibility that this "regulatory" effect of 1-35 Hz rhythms influences a region's susceptibility to ictal activity. Specifically, we hypothesize that pre-ictal regulatory patterns of activity (as measured via anti-correlation of low-frequency and broadband iEEG power) reflect successful gating of incoming drive to a circuit and that the strength of this anti-correlation will predict the susceptibility of a region to ictal activity. Methods: We recorded iEEG data from 7 patients with medial temporal lobe epilepsy and focal unilateral seizure onsets. Seizure onsets were manually marked in each channel. In the pre-onset data, we measured the correlation between the power of oscillations in the 1-35 Hz range and a broadband power signal estimated across a range of higher frequencies (60+ Hz). The specific 1-35 Hz frequency band with the most negative correlation with broadband power was defined as the putative regulatory frequency for that channel. The post-onset maximum broadband iEEG power was defined as the extent to which a region participated in ictal activity. Pre-onset interictal discharge (IID) rate was also estimated for each channel via a logistic regression model trained on a subset of manually marked data. This was done to control for the confounding effect of IIDs on 1-35 Hz vs. broadband power correlations. Results: Multiple linear regression found that both the pre-onset regulatory frequency strength (t(84)=7.98, p < .0001) and IID rate were independently predictive of the subsequent degree of ictal activity (t(84)=3.57, p=.0003). In each of the seven patients individually, we observed weaker ictal activity in channels with stronger pre-ictal regulatory activity. Conclusions: Our results suggest a novel means for predicting a region's susceptibility to ictal activity by measuring the degree of correlation between 1-35 Hz and broadband iEEG power modulations. If this relationship is causal, it suggests that strengthening the degree of naturalistic rhythmic activity in an area via techniques such as electrical stimulation would help prevent ictal activity. Funding: Natural Sciences and Engineering Research Council of Canada grant RGPIN-2014-04465 to CJH.