Abstracts

Rationale: High frequency oscillations (HFOs) have been linked as a biomarker for seizure genesis. Determining the spatial and temporal relationship of HFO to spike frequency in patients with tumor-related refractory epilepsy may further clarify the epileptogenic onset zone in these patients. Methods: Three 10-minute interictal intracranial (IC) EEG samples were taken from six patients with tumor-related refractory epilepsy undergoing 2-stage surgery. IC-EEG samples were taken daily roughly at the same time of the day, during awake, resting phase, and >6 hrs following a seizure. Automatic spike detection software (Stellate Systems) was used to identify the individual spikes. Each file was manually corrected by a single EEGer to remove artifacts and manually mark the spikes missed by the software. Each intracranial electrode (ICE) location was blindly identified by type as tumoral (T), peritumoral (PT), or non-tumoral (N) based on preoperative and post-implantation neuroimaging. In addition, each ICE was labeled by zone as seizure onset (SO), seizure spread (SS), or neither (N). Seizure onset and spread together were defined as the epileptogenic zone. Each IC-EEG sample was separately marked by a single EEGer for HFOs. EEG was sampled at 200Hz or 1,000Hz and filtered with a high-pass filter of 50Hz. Automatic computer program analysis was performed in Matlab (MathWorks) to calculate the frequency at which HFOs and spikes occurred at each ICE. Two-hundred milliseconds of data were searched on either side of a marked HFO to look for concurrent spike activity (SHFO). Data analysis included comparing spike frequency, HFO occurrence, and SHFO frequency to both electrode type and zone using ANOVA with subject blocking and Tukey tests as needed. Analysis of frequency and Chi-square of electrode type by electrode zone was also preformed. Results: A total of six patients were included in the analysis and the number of ICEs ranged from 41 to 96. ANOVAs looking at electrode zone with spike, HFO, and SHFO were significant for HFO (p<0.0001) and SHFO (p<0.0001). Post-hoc comparison of means by Tukey tests showed significant positive differences for seizure onset zone to spread or neither, but not when comparing spread to neither. There was no relationship for spike and electrode zone. ANOVAs looking at electrode type were significant only for HFO, with PT versus T and N being significant. Total percentages of electrode type by electrode zone showed onset electrodes outside of the tumor or peritumor region half of the time. Conclusions: Electrodes with a high frequency of spiking activity are not necessarily good predictors of the seizure onset zone, while electrodes with a high frequency of HFOs and HFOs with concurrent spikes, especially outside the tumoral region, are good predictors of the seizure onset zone.