Authors :
Presenting Author: Samuel Tomlinson, MD – University of Pennsylvania
Patrick Davis, MD PhD – Boston Children's Hospital
Caren Armstrong, MD PhD – UC Davis
Benjamin Kennedy, MD – Children's Hospital of Philadelphia
Eric Marsh, MD PhD – Children's Hospital of Philadelphia
Rationale: Interictal spikes are paroxysmal discharges observed specifically in epilepsy, yet their relationship with seizure generation remains incompletely understood. Spikes frequently co-occur across multiple brain regions simultaneously, suggesting that spikes can engage distributed networks of functionally interconnected brain regions. Characterizing spike networks may elucidate mechanisms through which the epileptogenic brain becomes pathologically synchronized. In this study, we extracted spike networks from passively-acquired invasive EEG recordings and probed them experimentally using cortico-cortical evoked potentials (CCEPs). We assessed whether the extent of concordance between spontaneous (i.e., spike-derived) and exogenous (i.e., CCEP-derived) networks varied based on epileptogenicity, anatomic location, and other parameters.
Methods: Pediatric patients (n=28) were prospectively enrolled to undergo systematic 1Hz cortical stimulation during stereo-EEG investigation (30 consecutive pulses, 2-8 mA amplitude). Spikes were detected from two 30-minute interictal EEG segments per patient. For each contact (“index”), we extracted the spike network by time-locking EEG waveforms in reference to spikes at the index contact and thresholding the spike-aligned event-related potential (Spike-ERP). Similarly, the CCEP network was computed for each index contact based on the amplitude of evoked responses. Concordance between thresholded spike and CCEP networks was quantified using a similarity metric (Jaccard Index, JI; range: 0-1).
Results: Spike-CCEP concordance was assessed at 1,001 index contacts across 28 subjects (35.8 contacts/subject). Spike-CCEP concordance was poorly discriminative of seizure-onset (SOZ) versus non-SOZ contacts (p=0.30) and varied significantly across anatomical regions, with significantly higher concordance observed in the mesial temporal structures (hippocampus, amygdala). Spatially, spike networks were more diffusely distributed than CCEP networks in all regions except the mesial temporal lobe and exhibited weaker concordance with resting-state functional connectivity networks. Typical clinical seizures triggered by 1Hz stimulation (n=9) were analyzed in a subset of patients. Spike-CCEP concordance at locations of stimulated seizures significantly exceeded values observed across all contacts (0.43±0.12 vs. 0.29±0.14, p=0.002) and across spontaneous SOZ contacts (p=0.001), even when the index contact was not involved in spontaneous seizure onset.
Conclusions: Interictal spikes provide a unique window into the behavior of the epileptogenic network. The strength of concordance between spike networks and other connectivity measures was surprisingly low and heterogeneous across the brain, suggesting that spike networks contribute distinct information about the organization of the epileptogenic network. Stimulated seizures were triggered preferentially at locations with high spike-CCEP concordance. The ability to exogenously activate the interictal spike network using stimulation may represent a biomarker for regions of heightened epileptogenicity, regardless of their involvement in spontaneous seizure onset.
Funding:
NIH 5-T32-NS-091006–10 (SBT); NIH 5-T32-NS-091008 (CA)