Abstracts

Rationale: Sevoflurane anesthesia is suggested to activate high-frequency oscillations (HFO) preferentially in epileptogenic brain regions. We determined if the resultant HFO effective connectivity measured by intracranial electroencephalography (iEEG) is congruent with white matter connectivity described by diffusion weighted imaging (DWI) tractography. We also determined if HFO spectral amplitude, HFO effective connectivity, or DWI white matter tractography could classify the epileptogenic sites.

Methods: This is an observational study of eight pediatric patients (age range: 4-22 years; 5 male) who underwent two-stage surgery for drug-resistant focal epilepsy. We analyzed intraoperative iEEG data collected during an oxygen baseline, at three time points while sevoflurane was dynamically increased from 0 to 2 minimum alveolar concentration (MAC), and at a plateau of 2 MAC. The iEEG signals were time-frequency transformed to attain HFO spectral amplitudes and then fed into a transfer entropy (TE) algorithm that calculates effective connectivity. To quantify DWI white matter connectivity, a whole-brain tractography template was created via standardized data from 1065 Human Connectome Project (HCP) participants. For each patient, their electrode coordinates were used as regions of interest on the template to compute a generalized fractional anisotropy (GFA) connectivity matrix. Binary logistic mixed model analysis was carried out to determine if HFO spectral amplitude, HFO effective connectivity, or GFA could classify the epileptogenic status of electrodes. To elucidate possible congruency between HFO effective connectivity and GFA, the matrices were randomly permuted and subject to correlational analysis. Finally, linear mixed model analysis tested whether GFA classified HFO spectral amplitude or effective connectivity values. All statistical analyses were run for each anesthetic condition.

Results: Both HFO spectral amplitude and effective connectivity accurately classified the epileptogenic status of electrodes, yet GFA failed to do so. These effects were most prominent at sevoflurane 2 MAC, and effective connectivity was more capable than spectral amplitude. Permuting GFA and HFO effective connectivity matrices revealed a positive correlation at sevoflurane 2 MAC (mean rho across 8 patients: 0.066; 95%CI: 0.01 to 0.12). Linear mixed model analysis similarly suggested that higher GFA was associated with higher HFO spectral amplitude at sevoflurane 2 MAC (t: +3.03; p: 0.0026).

Conclusions: Sevoflurane-induced neurophysiology measures such as intraoperative HFO spectral amplitude and effective connectivity are promising epileptogenicity biomarkers. In addition, these HFO signals may propagate along white matter pathways delineated by DWI tractography. Considering such dual effective, anatomical connectivity hubs could improve planning for epilepsy surgery, ensuring removal or disconnection of implicated grey and white matter structures.

Funding: This work was supported by NIH grants NS064033 (to E.A.) and NS089659 (to J.W.J.), as well as KAKENHI Grant JP19K09494 (to M.I.).