Abstracts

RATIONALE: Cortical resection is an effective treatment for selected children with drug-resistant partial epilepsy of neocortical origin. Ictal EEG findings on subdural electrodes are currently utilized as the gold standard to define epileptogenic foci. Longer subdural EEG monitoring to capture habitual seizures, however, is expensive and increases the risk of complications such as infection. We studied the relationship between quantitative interictal subdural EEG data and visually-defined ictal subdural EEG findings and determined whether interictal EEG findings are predictive of ictal EEG onset zones.
METHODS: Thirteen children (age: 1.2-15.4 years) underwent chronic intracranial EEG recording, using 48 to 120-channel subdural grid-electrodes. Three distinct 10-minute segments of the continuous interictal EEG recording were selected for each patient, and the spike frequency for each channel was determined using an automatic spike detection program. Subsequently, the average spike frequency of each electrode was compared to ictal assessment (seizure onset, seizure spread and no early ictal involvement). In addition, 50 distinct regional interictal spikes were averaged for each patient, and the amplitude as well as latency after the leading spike (averaged spike showing the earliest peak) were measured for each electrode and analyzed with respect to ictal EEG findings. For each electrode in each patient, spike frequency and spike amplitude were normalized to the maximal value recorded, and these normalized values were further analyzed.
RESULTS: Reproducibility of the spike frequency pattern derived from three 10-minute segments was high (Kendall[ssquote]s W: 0.85 [plusminus] 0.08). Electrodes with the highest spike frequency were found to be a part of the seizure onset in all 13 cases. Seizure onset electrodes showed higher spike frequency than spread electrodes (Mann-Whitney[ssquote]s U test: Chi2(13) = 101; p [lt] 0.001), and spread electrodes showed higher spike frequency than normal electrodes (Chi2(13) = 288; p [lt] 0.001). A receiver operating characteristics analysis showed that a cutoff threshold at 25% of the maximal spike frequency resulted in an accuracy of 0.87 for the detection of seizure onset electrodes. Electrodes with the highest spike amplitude were found to be a part of the seizure onset zone in 12/13 cases. The spike amplitude of seizure onset electrodes was higher than that of seizure spread electrodes (Chi2(13) = 93; p [lt] 0.001), and seizure spread electrodes showed higher spike amplitudes than normal electrodes (Chi2(13) = 214; p [lt] 0.001). Furthermore, electrodes showing the leading spike were found to be a part of seizure onset zone in 10/13 patients.
CONCLUSIONS: Quantitative interictal subdural EEG may predict the ictal onset zones in children with intractable neocortical epilepsy, and may potentially reduce intracranial EEG monitoring periods. Further correlation with surgical outcome is the logical next step to determine the clinical significance of quantitative interictal EEG analyses in pediatric epilepsy surgery.
[Supported by: NS34488, NS38324]