Abstracts

Rationale: Intracranial ictal and interictal electrophysiology have been used in tumor-related refractory epilepsy surgery to better define the epileptogenic zone. Interictal spike variables (amplitude, duration, and slope) may further characterize this epileptogenic zone in tumor-related epilepsy. Methods: Seven patients with tumor-related refractory epilepsy who underwent 2-stage surgery were studied. Long-term intracranial EEG was recorded over 4-6 days using 40 to 96 electrodes per patient. Three 10-minute interictal awake EEG samples taken ≥ 6 hours from the last clinical seizure were analyzed using automatic spike detection software (Stellate Systems) and corrected by a single EEGer. The corrected files were used to calculate average spike frequency for each electrode. Each electrode was labeled by its ictal activity as seizure onset (EEG involvement at earliest point in seizure), seizure spread (EEG involvement within 10 seconds of seizure onset), or neither. Electrode location was blindly identified as tumoral (within or overlying tumor), peritumoral (adjacent to tumor), or non-tumoral based on preoperative and post-implantation neuroimaging. Automatic computer program analysis was performed in Matlab (MathWorks) dividing each spike into 2 half waves at its peak (see figure). Eighty milliseconds of data were searched on either side of the peak to find the lowest value which was designated the spike's trough. Distance to the peak value was used to calculate duration (dur) while the difference in voltage was used to calculate amplitude (amp). The parameters preceding trough to peak are defined as amp1 and dur1 while parameters succeeding trough are defined as amp2 and dur2. Slope was calculated as amplitude/duration for each half wave (slope1 and slope2). Statistical analysis was done to standardize the automatic spike variables generated for each electrode (n=433) and tests used were the Spearman correlation coefficient between ranks, and the Kruskal-Wallis test for rank order differences across a categorical variable (i.e, nonparametric equivalent of one-way ANOVA and t-test). Results: The Spearman correlations between the seizure onset zone (onset, spread, neither) and spike variables were significant for spike frequency (r=-0.34, p<0.01), dur1 (r=0.14, p <0.01), and slope1 (r=-0.17, p<0.01). For tumor location, the correlations were with spike frequency (r=-0.14, p<0.01) only. The Kruskal-Wallis tests were significant for seizure onset by spike frequency (χ2=57.83, p <0.01; onset and spread were significantly different than neither) and slope 1 (χ2 = 12.57, p <0.01; onset was significantly different than neither). The Kruskal-Wallis tests for tumor location were also significant for spike frequency (χ2=59.78, p<0.01). Conclusions: Analysis of interictal spike variables show that the initial phase of the spikes correlates better with the epileptogenic zone, but not with tumor location suggesting the generation of spikes by more synchronized neuronal population is associated with epileptogenicity.