Abstracts

Rationale: Epileptic spasms (ES) are a brief and repetitive seizure semiology. ES are more often seen in young patients and may be associated with severe developmental complications, therefore proactive treatment approaches including resective surgery may be taken. However, it is challenging to localize foci of ES activity because high-frequency ictal discharges rapidly involve widespread cortical regions. Moreover, the relationship between ES propagation dynamics and anatomical white matter pathways, both which can be estimated during pre-resection workup, is not well understood. A better understanding of the associated propagation networks in ES may ultimately allow estimates of seizure onset and symptomatogenic zones to be refined based on propagated activity and the anatomy that constrains it. Methods: T1-weighted and diffusion MRI were preoperatively collected from 7 children (1.9 to 3.6 years old, Table 1) undergoing 2-stage resective surgery with subdural EEG monitoring sampled at 1 kHz. CT and x-ray were collected following electrode implantation and used to determine each contact's corresponding location on the preoperative brain. Clusters of ES were identified, and individual spasms were marked approximately 1 second prior to visually-defined EEG signal onset. For each spasm, an electrode's onset time was then calculated as the first 25-fold z-score increase in power across multiple wavelet-convolved frequency bands. Probabilistic tractography seeded from a sphere centered on the electrode with the earliest onset times generated streamlines connecting to other electrode locations. Results: In general, we found a correlation between an electrode's average latency to spasm onset in the 90-110 Hz range and the average tractography length between it and the electrode with the earliest average spasm onset. Since each spasm signal was seen only in a subset of electrodes, this correlation varied depending on which electrodes were considered: electrode-wide variance in spasm onset latency, normalized by maximum latency across spasms, was minimized when considering electrodes involved in 50% or more tagged spasms. In this set of electrodes, a patient's relationship between tractography length and onset latency varied between r = .35-.59 (p = .031-.001), with r = .42 (p < .001) for patient-normalized pooled data (Figure 1). Such a relationship was not significant when utilizing tractography length along corticothalamic routes between electrodes. No significant differences in this relationship were seen across patients with varying seizure frequency or history. In our 7-patient cohort the relationship was slightly weaker in all patients with lesional etiologies on pathology (dysplasia or TSC, Table1) as compared to those with only gliosis (average r = .41 vs. average r = .59, respectively). Replication with larger sample size is needed to verify the possible role of such pathological differences in the structural-functional relationship of epileptic spasm propagation. Conclusions: Within patients, epileptic spasm activity spreads across the brain with relatively consistent dynamics, and these dynamics appear proportional to the length of direct, corticocortical white matter connections on diffusion tractography. Further understanding of anatomical-functional relationships in seizure propagation will benefit not only seizure modeling efforts but could also help develop markers of epilepsy subtypes during clinical workup. Funding: This study was funded by grants from the National Institute of Health, R01-NS089659 to Dr. Jeong-Won Jeong and R01-NS064033 to Dr. Eishi Asano.