Abstracts

The unpredictability of ictal onset has traditionally been viewed as a fundamental characteristic of epilepsy. Attempts to find order in the chaotic processes that generate seizures have been the focus of much research in the last decade, with most of this work focused on adult temporal lobe epilepsy. This work examines the utility of adult seizure prediction algorithms in pediatric, extra-temporal epilepsy.
Eight consecutive patients seen at CHOP who underwent continuous video and intracranial-EEG recording were screened for seizures of extra-temporal origin. Seven patients met criteria, and a single patient with an exquisitely focal ictal onset zone and concordant, pathologically-confirmed cortical dysplasia was first used to assess two different seizure prediction algorithms developed for adult epilepsy. Ictal onset zones and times were annotated by a clinical epileptologist. Of 27 total seizures, 8 had at least 1 hour of uninterrupted inter-ictal activity prior to seizure onset. These 8 seizures were analyzed for the presence of high frequency epileptiform oscillations (HFEOs) by time-frequency and spectral analysis. HFEO frequency and duration were calculated. The frequency and amplitude characteristics of the seizures were also tested for the presence an accumulation of energy in the EEG prior to ictal onset.
All eight seizures analyzed had a high-frequency ictal onset (25 hz [plusmn] 5 hz) in 1 to 4 contiguous sub-dural electrodes. Two observations were made. First, HFEOs (75 hz [plusmn] 25 hz) were seen in the inter-ictal EEG in a single electrode within the area of cortical dysplasia, but HFEOs were not seen in the contacts demonstrating ictal-onset on intra-cranial EEG. The inter-ictal HFEOs were significantly faster than the frequency at ictal onset. HFEOs were also infrequently seen in the surrounding normal cortex. Second, an increase in lower frequency, high amplitude activity (accumulated energy) was seen in the contacts where the ictal onset occurred. This change was seen as early as 90 minutes prior to the seizure onset (range 30-90 min) and often coincided with a decline in the HFEOs in the adjacent electrode.
These data suggest that HFEOs may be useful in identifying areas of abnormal cortex and that changes in frequency content or amplitude of the intracranial- EEG may be early signs of an emerging seizure. Together, this information may allow for better localization prior to epilepsy surgery and for the development of algorithms that will reliably predict seizure onset in children with extra-temporal epilepsy.
[Supported by: National Institutes of Health Grants # RO1NS041811-01]