Abstracts

Rationale: Cortico-cortical evoked potentials (CCEPs) measured by low-frequency electrical stimulation allow in-vivo measures of network interactions in the brains of patients undergoing invasive presurgical stereo-electroencephalography (SEEG) monitoring for epilepsy (Matsumoto et al. Brain, 2004, Matsumoto et al. Brain, 2007). In combination with other modalities, the CCEPs technique provides additional insight into brain regions involved in epileptic networks, thus helping in localization of the epileptogenic zone (EZ). Recently, we identified a “fingerprint” pattern from ictal SEEG signals that can be used as a reliable biomarker of the EZ (Grinenko et al. Brain 2018). In this study, we explore the relationship between CCEP responses and propagation of the ictal fingerprint. We show that the strength of the CCEP responses is strongly correlated with the fingerprint prediction scores when the EZ areas are stimulated. Methods: Five patients who underwent SEEG evaluation and had CCEPs recorded at the Cleveland Clinic were recruited for this retrospective study. All five patients had surgical resection and were seizure-free 12 months after surgery. From the CCEP recordings, we averaged trials corresponding to the responses for each stimulated contact pair for each polarity (odd and even) to improve the SNR. We then computed the root-mean-square (RMS) values of the averaged signals at all other contacts for three canonical periods: Early (10ms – 60ms), Late (60ms – 250ms) and VeryLate (>250ms) to represent the strength of CCEP responses. We also applied the fingerprint method (Grinenko et al. Brain 2018) to the ictal data and generated a prediction (epileptogenicity) score for each contact. A positive score indicates higher probability of a contact being an EZ contact; a negative score indicates a very low probability. We then computed the Pearson correlation coefficients between the CCEP’s RMS values from each stimulated contact pair to all other electrodes and the fingerprint prediction scores at the corresponding electrodes. We repeated this for each period for all five patients. In addition, we tested whether the (mean) correlation values were significantly higher when EZ contacts were stimulated than that when other non-EZ contacts were stimulated across all patients using the Wilcoxon rank sum test. Results: Fig. 1 (a) – (e) shows the correlation matrices under the odd polarity for each period of CCEP response (x-axis) for all five patients when different contacts were stimulated (y-axis). The blue contacts in the y-axis were identified as the EZ contacts by the fingerprint method (Grinenko et al. Brain 2018). In contrast to the non-EZ contacts, higher correlations were observed consistently across patients when the EZ contacts were stimulated, indicating the concordance between the network identified using CCEPs and the network along which the fingerprint pattern propagated. Examples of the CCEP responses are shown in Fig. 2. Higher RMS values were observed at contacts that have higher fingerprint prediction scores and vice versa. Moreover, Fig. 1 (f) shows box-plots of the mean correlation values across five patients and two polarities when the EZ and non-EZ contacts were stimulated in different period of CCEP responses. The correlations were significantly higher (p-value was 9.13 E-5, 1.23 E-4, 2.91 E-4, 9.13 E-5 for the four periods, respectively) when the EZ contacts were stimulated than when the non-EZ contacts were stimulated, regardless of the period of responses. Conclusions: Higher correlations between the CCEP response and fingerprint scores were observed when the EZ contacts were stimulated, suggesting that interictal propagation of evoked potentials follows a similar propagation network as does ictal propagation as reflected in the characteristic ‘fingerprint’. This result is consistent with the use of CCEPs for mapping ictal propagation networks as part of a presurgical evaluation of candidates for targeted ablation. Funding: National Institutes of Health Grants: R01-NS074980, R01-EB026299, and R01-NS089212