Abstracts

Rationale: To investigate the ictal high frequency oscillations (HFOs: ?70 Hz) in temporal lobe epilepsy.Methods: We studied patients who underwent temporal resection at University of Louisville Epilepsy Center after intracranial recording with uni-/bilateral subdural grids/strips and posteroanteriorly-inserted hippocampal depth electrodes. EEG was acquired at 1000 Hz sampling rate. Presurgically, the seizure onset zone (SOZ) was determined traditionally (1 70 Hz filter, 10-s window) by reviewing the EEG in a bipolar montage to identify the earliest rhythmic activity. We then looked for discrete HFOs at onset (53 300 Hz filter, 2-s window), and if present, re-defined the SOZ in terms of HFOs (Modur et al., Epilepsia 2011, in press). The SOZ and 1 cm of surrounding tissue were resected. For purposes of this study, we performed a post-hoc analysis using the BESA software (MEGIS, Germany). To identify the ictal HFOs, we filtered the EEG with a 50 Hz high-pass filter and obtained fast Fourier transform (FFT)-derived spectra over 0 330 Hz on a 512-ms epoch placed at seizure onset. We classified the seizure onsets into 3 subtypes (Fig): neocortical temporal (NCT: subdural electrodes involved); amygdalo-hippocampal (AHC: depth electrodes involved); diffuse temporal (DT: simultaneous NCT and AHC onset). For NCT and AHC seizures, we marked the time of seizure spread to the corresponding AHC and NCT channels respectively. Based on peak FFT frequency, we classified the frequencies at seizure onset and seizure spread as: conventional frequency activity (CFA: 1 69 Hz); HFOs-1 (70 199 Hz); HFOs-2 (200 330 Hz). We analyzed the data using linear mixed models and Tukey's test.Results: Of 8 patients, 5 had unilateral seizures concordant with the intended surgical side (Table). Three patients had independent, contralateral seizures, felt to be insufficient to contraindicate the intended surgery. Outcome was class I/II in 7 patients over a mean follow-up of 33 months. At onset, frequencies of NCT (mean 67 Hz, 95% CI 62 71 Hz) and DT (mean 61 Hz) seizures were in the high gamma or HFOs-1 range, significantly higher than the beta-range AHC seizures (mean 23 Hz). At onset, frequencies of NCT seizures and NCT channels of DT seizures were similar. At onset, frequencies in AHC channels of DT seizures were higher than the AHC seizures (mean 53 vs. 31 Hz, p=0.001). Frequencies in NCT channels were similar regardless of whether they were involved at seizure onset or by spread (mean 67 vs. 72 Hz, p=0.39). Power in the CFA, HFOs-1 and HFOs-2 bands was 2-, 6- and 2-fold higher respectively in NCT channels at seizure onset compared to spread. Frequencies in AHC channels were lower when involved at seizure onset compared to spread (mean 23 vs. 42 Hz, p=0.004). In AHC seizures, the power in CFA band was higher at seizure onset compared to spread (p=ns) whereas the power in the HFOs-1 and HFOs-2 bands was lower at seizure onset compared to spread (p=ns).Conclusions: Ictal HFOs occur in NCT rather than AHC seizures. Frequency patterns at seizure onset and spread support the notion that distinct epileptic networks underlie seizure genesis in the subtypes of TLE.