Abstracts

RATIONALE: Recent work shows that the pre-ictal IEEG in humans with MTLE is characterized by sustained increases in total energy compared to baseline IEEG far removed in time from ictal activity¹. Other research suggests that seizure precursors within minutes of EEG seizure onset may be due primarily to activity in the low frequency ([lt] 5 Hz) range². Understanding the frequency specificity of preictal energy increases is an important step in elucidating the mechanisms of ictogenesis, and may also increase the sensitivity of EEG-based seizure prediction methods.
METHODS: 10 patients with MTLE were chronically implanted with electrodes as part of the evaluation procedure for surgical resection. IEEG was continuously monitored throughout their hospital stay, which lasted from 4 days to 2 weeks. Seizures were marked visually by an expert epileptologist and 40 minute records prior to each seizure, from an electrode in or near the ictal onset zone (IOZ), were extracted and analyzed. Pre-ictal records were compared to a baseline record from the same recording site, but at least 4 hours removed in time from the onset or end of any ictal activity. Frequency specificity was measured by defining a spectral deviance function, Dev([omega]) [definedas] |S[sub]pre-ictal[/sub] ([omega])|²/|S[sub]baseline[/sub]([omega])|², where |S([omega])|² is the periodogram of the IEEG record at frequency [omega]. A deviance value greater than 1 indicates an increase in pre-ictal signal energy compared to baseline. A flat spectral deviance function would indicate a change in energy that has no frequency specificity.
RESULTS: In all 10 patients there was a peak in the spectral deviance in the frequency range [lt] 5 Hz. 8 of the 10 patients showed additional peaks at higher frequencies that generalized across most seizures. The shape of the spectral deviance function appeared similar across seizures from a given patient, but differed from patient to patient.
CONCLUSIONS: Pre-ictal IEEG energy changes have a frequency-specific component that generalizes across all patients studied (0 [ndash] 5 Hz band), and in most patients one or more higher frequency-specific components, which were patient specific. Understanding these findings may provide important clues to mechanisms of ictogenesis. Changes in patient-specific preictal activity may be a function of electrode placement, or each patient[scquote]s unique physiology. In-vivo recording in animals in which the location of the epileptic focus and recording electrodes can be controlled will likely be necessary to determine the significance of these findings and their relationship to ictogenesis.
1. Litt B, Esteller R, Echauz J, et al. Epileptic Seizures May Begin Hours in Advance of Clinical Onset: A Report of 5 Patients. Neuron 2001;29(4).
2. Le Van Quyen M, Martinerie J, Navarro V, et al. Anticipation of epileptic seizures from standard EEG recordings. The Lancet 2001;357(9251):183.
Support: National Institutes of Health Training Grant [pound]T32-GM07517 (Graduate Training in Systems and Integrative Biology) and the National Science Foundation Graduate Fellowship (SDC); Whitaker Foundation Graduate Fellowship (APR); Whitaker Foundation, Epilepsy Foundation, American Epilepsy Society, University of Pennsylvania Research Foundation, and the National Institutes of Health grant [pound]MH-62298RO1 (BL).