Abstracts

Rationale: Memory impairment in epilepsy is conventionally attributed to the functional disruption of the mesial temporal regions. The disruption is thought to be at least two-fold: static effects from tissue damages caused over the disease duration, and dynamic/transient effects on the normal neural processing caused by epileptiform activity including seizures and interictal spikes (IIS; Gelinas et al., 2016). The latter in particular is supported by rodent studies reporting transient impairment of retrieval accuracy when hippocampal inter-ictal spikes occurred during memory retrieval (Kleen et al., 2010). Although this has been implicated in human data as well, the results are rather mixed. While Kleen et al. (2013) found that while polyspikes impaired memory retrieval, canonical IIS (high-amplitude isolated spikes) had no effect on memory retrieval. However, the distribution of different subtypes of inter-ictal activity was not elucidated, making these findings less definitive. In addition, they could not differentiate between hippocampal vs. propagating IIS due to limited coverage. Thus, in spite of the widely accepted clinical significance of inter-ictal spikes, the effects of their occurrence on memory retrieval are still unknown. Here, using recordings from human intracranial sEEG electrodes, we tested the effects of occurrence and spatial distribution of of IIS on memory retrieval. Methods: Intracranial data were collected from 8 patients with intractable epilepsy while patients performed episodic memory tasks. The IIS were detected within each retrieval trial ([0-2200ms] post retrieval-cue) using conventional algorithms to facilitate reproducibility (number of IIS in individuals ranges 4-1,735, M = 571, SD = 590). We tested whether memory retrieval is impaired by IIS occurrence regardless of the regions [approach 1], IIS occurrence in the hippocampus [approach 2], and IIS prevalence (number of channels with IIS weighted by number of spikes within the trial) and spread over the memory network [approach 3]. Using logistic regression, we tested whether these IIS measures predict successful memory retrieval. Results: Occurrence of IIS do not appear to affect memory retrieval regardless of the brain region involved (approach 1 & 2, all p > .05). In approach 3, IIS prevalence alone did not predict successful retrieval (p > .05), and adding the IIS spread did not improve the model (p >.05). These results were consistent in both the individual and group level analyses. Conclusions: IIS alone do not appear to disrupt memory retrieval. Although it is a possible that analysis using a larger retrieval window instead of first 2.2s may reveal different results, the lack of significant effects of our measures may suggest that there are aspects to IIS (e.g., amplitude, latency, laterality) that have larger effects than just the occurrence of spikes. Additionally, it implies that episodic memory is robust to perturbation by spiking activity. This may imply redundancies in the systems involved in memory retrieval, specifically in the temporal domain, wherein a brief episode of perturbation is insufficient to degrade processing. Funding: NSF 1406556, NIH NS087527