Sevoflurane-induced high-frequency oscillations, effective connectivity and intraoperative classification of epileptic brain areas

Abstract

Objective: To determine how sevoflurane anesthesia modulates intraoperative epilepsy biomarkers on electrocorticography, including high-frequency oscillation (HFO) effective connectivity (EC), and to investigate their relation to epileptogenicity and anatomical white matter.

Methods: We studied eight pediatric drug-resistant focal epilepsy patients who achieved seizure control after invasive monitoring and resective surgery. We visualized spatial distributions of the electrocorticography biomarkers at an oxygen baseline, three time-points while sevoflurane was increasing, and at a plateau of 2 minimum alveolar concentration (MAC) sevoflurane. HFO EC was combined with diffusion-weighted imaging, in dynamic tractography.

Results: Intraoperative HFO EC diffusely increased as a function of sevoflurane concentration, although most in epileptogenic sites (defined as those included in the resection); their ability to classify epileptogenicity was optimized at sevoflurane 2 MAC. HFO EC could be visualized on major white matter tracts, as a function of sevoflurane level.

Conclusions: The results strengthened the hypothesis that sevoflurane-activated HFO biomarkers may help intraoperatively localize the epileptogenic zone.

Significance: Our results help characterize how HFOs at non-epileptogenic and epileptogenic networks respond to sevoflurane. It may be warranted to establish a normative HFO atlas incorporating the modifying effects of sevoflurane and major white matter pathways, as critical reference in epilepsy presurgical evaluation.

Keywords: Acute electrocorticography (ECoG); Diffusion tensor imaging (DTI) tractography; General anesthesia; Modulation index; Subdural grid electroencephalography (EEG); Transfer entropy (TE).

Figure 1.. Methodological summary.

A. Patients were implanted with either surface and/or depth intracranial electrodes to map epileptogenic and eloquent cortex. B. Raw intraoperative electrocorticography (ECoG) traces from a representative patient as sevoflurane was increased from an oxygen baseline (bottom) to 2 minimum alveolar concentration (MAC; top). C. ECoG data was then mathematically transformed from the time-voltage domain into the time-frequency domain; representative time-frequency transformed electrode under oxygen baseline (left) and sevoflurane (right). D. Electrodes were co-registered on each patient’s three-dimensional magnetic resonance image (MRI) reconstructed cortical surface for further analysis (patient 6 with right frontal distribution shown). Left and right images show lateral and medial views, respectively. For patient 6, A-B are depth electrodes and C-H are surface electrodes. Yellow dotted line represents the resection margin.

Figure 2.. Delta effective connectivity responds to sevoflurane anesthesia.

A. Delta transfer entropy values of pooled individual electrode sites (n = 621), as a function of sevoflurane concentration. The black trend line represents the equation predicted from the linear mixed model analysis. B. Each patient’s (n = 8) average delta transfer entropy for epileptogenic (yellow; n = 202 total sites) and non-epileptogenic (green; n = 419 total sites) sites, at each anesthetic stage. Yellow and green dots represent the average delta transfer entropy value of all epileptogenic or non-epileptogenic electrode sites, respectively, for individual patients. Lines connect a given patient’s non-epileptogenic and epileptogenic values. Asterisks denote binary logistic mixed model significance (p < 0.005) for classifying epileptogenic sites. For each patient and anesthetic stage, effect size was estimated using Cohen’s d defined as the absolute value of: [(the mean biomarker value in the epileptogenic sites) – (the mean biomarker value in the non-epileptogenic sites)] / (standard deviation for all sites). The average Cohen’s d across all eight patients is shown under the corresponding anesthetic stage. O2 = oxygen baseline; SI1 = sevoflurane increasing first third; SI2 = sevoflurane increasing mid third; SI3 = sevoflurane increasing last third; Sev2 = sevoflurane 2 minimum alveolar concentration (MAC).

Figure 3.. Sevoflurane activates high-frequency oscillation (HFO) effective connectivity most in the epileptogenic zone.

A. HFO transfer entropy (TE) for all pooled individual electrodes (n = 621) as a function of sevoflurane concentration. The black trend line represents the equation predicted from the linear mixed model analysis. B. Each patient’s (n = 8) average HFO transfer entropy for epileptogenic (yellow; n = 202 total sites) and non-epileptogenic (green; n = 419 total sites) sites, at each anesthetic stage. Yellow and green dots represent the average HFO transfer entropy value of all epileptogenic or non-epileptogenic electrode sites, respectively, for individual patients. Lines connect a given patient’s non-epileptogenic and epileptogenic values. Asterisks denote binary logistic mixed model significance (p < 0.005) for classifying epileptogenic sites. For each patient and anesthetic stage, effect size was estimated using Cohen’s d defined as the absolute value of: [(the mean biomarker value in the epileptogenic sites) – (the mean biomarker value in the non-epileptogenic sites)] / (standard deviation for all sites). The average Cohen’s d across all eight patients is shown under the corresponding anesthetic stage. C. Right hemisphere of patient 6 with electrodes mapped to reconstructed three-dimensional cortical surface. HFO (red arrows) and delta (blue arrows) effective transfer entropy connections above 6 standard deviations between electrode sites, at each anesthetic condition. Yellow dotted lines outline the resection margin (i.e., the epileptogenic zone defined in the present study). O2 = oxygen baseline; SI1 = sevoflurane increasing first third; SI2 = sevoflurane increasing mid third; SI3 = sevoflurane increasing last third; Sev2 = sevoflurane 2 minimum alveolar concentration (MAC).

Figure 4.. White matter – high-frequency oscillation (HFO) correlation.

(Top) Diffusion weighted imaging tractography. Electrode sites with a right frontal distribution from patient 6 converted to Lausanne brain atlas regions-of-interest (ROIs) and overlayed on a standardized whole-brain tractography template derived from 1,065 patients in the Human Connectome Project. (Bottom) The white matter tractography network from patient 6’s electrode ROIs, with HFO transfer entropy (TE) values superimposed on corresponding tracts as color, for each sevoflurane stage. Light and dark blue tracts denote HFO transfer entropy values above and below the oxygen baseline TE maximum, respectively. The top row is a right-lateral-sagittal view, the middle row is a superior-axial view, and the bottom row is an anterior-coronal view. P = posterior; A = anterior; R = right; L = left.

Figure 5.. A working hypothesis: sevoflurane activates high-frequency oscillations (HFOs) preferentially in epileptogenic brain regions.

During oxygen baseline there are low-level background HFOs and delta waves resonating in the brain. Increasing the concentration of sevoflurane anesthesia activates widespread spike-and-wave epileptiform activity, along with the HFO and delta wave components. During the ‘sev increasing last third’ epoch, the sevoflurane-induced HFOs and delta waves become hyper-synchronized in epileptogenic tissue by an unknown process that may involve resistance to delta effective connectivity augmentation. This phase-amplitude coupling could create windows of disinhibition at sevoflurane 2 minimum alveolar concentration (MAC), when HFO effective connectivity is most amplified in epileptogenic neural tissue, and these signals may partially utilize white matter tracts to propagate throughout the brain.