Abstracts

Rationale: Intracranial EEG (iEEG) is performed in pharmacoresistant partial epilepsy patients when non-invasive tests are incongruent or putative ictal onset zone is near eloquent cortex. Determining ictal onset zone using iEEG has been conventionally based on the identification of specific ictal patterns in the bandwidth of 1- 70 Hz. High frequency oscillations (HFOs, > 60 Hz) have been recently recognized as highly correlated with epileptogenic zones. However, HFOs can be difficult to detect in the time domain because of small amplitude and low sampling rate (256-512 Hz) of current clinical EEG system. Therefore, true incidence of ictal HFOs and its role in localization of epileptogenic zone on iEEG remain elusive. Methods: From September 2008 to Dec 2009, we identified 47 patients (mean age = 10 yrs, range 9 mo to 25 yrs; M:F = 28:19) who had iEEG, high frequency domain analysis, and subsequent resective surgery. iEEGs were recorded with 2000 Hz sampling rate and examined visually in the 1-70 Hz bandwidth. iEEG during a patient s habitual seizure was further analyzed with time-frequency decomposition in three additional frequency bandwidths: 80-150, 150-300, 300-500Hz. Each individual electrode was also carefully evaluated for the leading edge of HFO (primary ictal onset zone, PIZ) in the fast Fourier Transformation power spectrum and compared with the visually apparent onset in the time domain. In addition to the seizure onset, the spread of HFOs during a seizure (secondary ictal spread zone, SIZ) before generalization was also analyzed in the same bandwidths. The completeness of resection was determined based on the inclusion of PIZ and SIZ in the resection margin. Surgical outcome was scored using Engel s classification at 6 mo, 1 yr, and 2 yr post-operatively. Results: Overall seizure-free outcome in 47 patients was 46% (22/47) (mean follow-up duration 10 mo, 6 - 18 mo). Forty-four (94%) of 47 patients had ictal HFO by time-frequency decomposition analysis on iEEG. Ictal HFO frequency was distributed in all three bandwidths: 11, 80 - 150; 16, 150-300; 7, 300 - 500 Hz bandwidth. Unlike the HFOs with two lower bandwidths, the highest frequency (300 - 500 Hz) ictal HFOs tended to remain in the same electrodes during a seizure until it generalized. Thirty four patients (77%, 34/44) had complete resection including PIZ and SIZ while the other 10 patients (23%, 10/44) had incomplete resection because of involvement of eloquent cortex. Complete resection of ictal HFOs resulted in significantly higher seizure freedom (56%, 19/34) than incomplete resection (10%, 1/10) (Fisher s Exact Test, p = 0.013). Among the patients who had complete resection of ictal HFOs, different bandwidths did not affect surgical outcome. Conclusions: Our study showed that ictal HFO is commonly found in iEEG and has a localizing value. The presence of ictal HFOs and its complete resection may be one of the favorable prognostic indicators for surgical outcome.