Abstracts

Rationale: Pre-surgical intracranial monitoring in epilepsy is used to determine the region of cortex for resection, ablation or modulation. It has been proposed that an increased amplitude of corticocortical evoked potentials (CCEPs) could be a marker for the seizure onset zone. Proving this requires correction of the known confound of distance: CCEP amplitudes are higher nearby the site of stimulation. We compare models for the CCEP-amplitude relationship and explore the impact of choice of recording montage. We then illustrate how correcting CCEP amplitude for distance changes an analysis correlating CCEP amplitudes with the clinically determined ablation zone.

Methods: Single pulse electrical stimulation (1 Hz, 20 second train) was performed at multiple sites in 9 patients with refractory epilepsy undergoing stereo-EEG monitoring. Resulting CCEPs were averaged, and amplitude was defined as the root mean squared deviation from the end of the stimulus artifact to 100 ms after the stimulus. We computed the Euclidean distance between each stimulated and recording electrode, then compared the effect of distance of the CCEP amplitude using the traditional referential montage, or the alternatives of a bipolar or Laplacian montage. We compared an exponential (semi-log) or power law (log-log) relationship. Finally, we defined a metric called CCEP-ictal overlap which is positive if CCEPs are higher in areas of early seizure-spread compared to control areas, and explored the effect of recording montage and correcting CCEP amplitudes for distance.

Results: The CCEP amplitude versus distance relationship was better fit with a log-log than a semi-log relationship: for the bipolar montage, R² was 0.33 for log-log vs. 0.30 for semi-log (p=0.001, paired t-test). There was an apparent and statistically significant relationship between positive CCEP-ictal overlap and the ablation zone when using the raw, uncorrected CCEP amplitudes and any recording montage, but this disappeared when distance-corrected CCEP amplitudes were used. For example, when using a bipolar montage to measure the raw CCEPs amplitude, we found that 18/37 (49%) of contacts inside the ablation zone showed +CCEP-ictal overlap, significantly more (p < 0.001, Fisher exact) than 10/91 (11%) of contacts outside the ablation zone. However, when we used the log-log model to Z-score correct CCEP amplitude, we found that only 5/32 (14%) of contacts in the ablation zone showed +CCEP-ictal overlap, which was not significantly different (p=0.32, Fisher exact) than 7/91 (8%) of contacts outside the ablation zone.