Abstracts

Rationale: Interictal spikes (IIS) are highly correlated with the presence of seizures, and this is the basis of their use as biomarker for epilepsy. Unfortunately the mechanisms underlying IIS initiation, propagation and termination, their role in epilepsy, and the basis for correlation with seizures remain uncertain. In this study we used recordings of human epileptic networks, calcium imaging in vitro, and computational modeling to explore one of these questions, namely the origin of the variance in IIS morphology. Methods: To study the source of IIS variance, we analyzed the initiation and spread of IIS in patients with long standing pharmaco-resistant epilepsy. We used electrocortigraphic data from subdural grids to map the propagation of IIS through cortex by stacking spatial averages of ?V at all electrodes for each time point. To study the network determinants of IIS propagation, we used calcium imaging to map IIS-like trajectories in chronically epileptic hippocampal organotypic slices. We also used a large-scale computational model of the CA3 region of hippocampus based on Traub and Miles ( 1991 ) with added constraints to make the network scale-free. Connectivity between cells was random and decreased with distance . A limited number of hub-cells were strongly connected with other cells. Results: Even for multiple spikes originating from a single location, IIS had variable onset trajectories. The recruitment of the same cortical areas occurred in a unique sequence for each spike. Calcium imaging also supported the idea that activation of different areas of the network occurred in unique sequences for each spike in individual hippocampal cultures. To test the influence of interneurons on the pattern of network activation during spikes, we blocked GABA-mediated synaptic inhibition, resulting in most of the trajectory variance being removed. We used the computational model of CA3 hippocampal network to simulate and extend these results. With intact inhibition series of locally synchronized activity occurred randomly throughout the network, and was extinguished by GABAergic inhibition. This activity created localized refractory areas, through which subsequent large-scale synchronous activity propagated poorly, influencing future spikes locations and trajectories. After blocking inhibition, local events were no longer quenched, thus every initiation could spread to involve the entire network, and the pathway variance due to local refractory areas was lost. Conclusions: In networks with intact GABAergic transmission, inhibition is sufficient to quench nascent IIS, leading to formation of local refractory areas. These refractory areas strongly affect propagation trajectories of subsequent IIS leading to high degree of pathway variability. On the other hand in networks with impaired inhibition the amount of quenching is decreased, resulting in low variability. Thus pathway variance could be used as a noninvasive measure of the functional integrity of the inhibitory system.