Consciousness and Nociception During Anesthesia

Overview

The goal of this preliminary prospective and observational study is to use hd-EEG (high density electroencephalogram) to investigate how functional and network markers of consciousness are associated to nociception during general anesthesia. More specifically, the aim of this study is to characterize whole-brain functional connectivity and network changes induced by noxious stimulation, and adapt these findings to a clinically-applicable EEG (electroencephalogram) montage.

Background: Until now, the clinical neuromonitoring of consciousness in general anesthesia has relied on the electrophysiological assessment of isolated frontal regions of the brain, providing only partial insight into the neural mechanisms involved in sustaining or suspending consciousness. What is more, functional connectivity and graph theory markers of consciousness have not yet been investigated in response to nociception during surgical anesthesia, and it remains unclear how nociception affects anesthetic-induced unconsciousness. Aims: This study will investigate how hd-EEG (high density electroencephalogram) functional markers of consciousness are associated to nociception during general anesthesia. More specifically, the aim of this study is to characterize whole-brain functional connectivity and network changes induced by noxious stimulation, and adapt these findings to a clinically-applicable EEG (electroencephalogram) montage. To this end, three specific objectives have been set: 1. Compare the functional connectivity and network characteristics of general anesthesia before, during and after tetanic stimulation of the ulnar nerve during general anesthesia 2. Investigate the association between functional connectivity and network properties and the changes in hypnotic depth and nociception calculated by standard clinical neuromonitoring tools. 3. Identify the combination of 20 EEG channels that optimally detect changes in the functional network induced by noxious stimulation. Study design: This study will be a preliminary prospective and observational study. It will be conducted on thirty patients (n = 30) scheduled for surgery under general anesthesia. Patients will be recruited and tested from Hôpital Maisonneuve-Rosemont, with the hd-EEG (high density electroencephalogram) system of the Center for Advanced Research in Sleep Medicine (CARSM) of the Centre intégré universitaire de santé et de services sociaux du Nord-de-l'Île-de-Montréal (CIUSSS-NIM). Protocol: No premedication will be administered. When entering the operating room, the investigators will place a 128-channel EEG saline net (MagstimEGI) referenced to Cz on the patient's head (10 min application time). Impedances will be verified and maintained \<50 kΩ for the duration of the study. The patient will be positioned supine and all routine monitors will be installed (electrocardiography, non-invasive blood pressure and pulse oximetry, nasopharyngeal temperature and end-tidal partial pressure of carbon dioxide). The PMD200TM monitor (a finger sensor device which reflects low to high nociception) as well as the Bispectral Index (BIS), will be connected to the anesthesia machine and the patient, recording the Nociception Level (NOL) index and BIS continuously. A 5-min baseline resting-state EEG with eyes closed will be performed before induction (Awake). Baseline values of NOL, BIS, mean arterial pressure (MAP) and heart rate (HR) will be recorded for 1 min before induction.General anesthesia will be induced with propofol and remifentanil. Propofol will be administered with a target-controlled infusion (TCI) pump, using the Marsh pharmacokinetic model,46 with initial target predicted effect-site concentration of propofol at 2.0 mcg/ml in flash mode. Propofol targets will be adjusted by increments of 0.2 mcg/ml until a BIS between 45-55 is reached. Simultaneously, remifentanil will be administered for induction of anesthesia at the dose of 1 mcg.Kg-1 over 30 sec, with a targeted NOL index below a threshold of 10. To eliminate potential electromyographical contamination of BIS and EEG, rocuronium (0.8 mg/kg bolus followed by infusion) will be given once the patient is no longer responding and a BIS 45-55 is achieved. The patient will be then manually ventilated with 100% inspired O2 until the response to electrical stimulation of the forearm shows that muscle relaxants are fully active. The patient will be intubated with a video laryngoscope and a stylet and mechanical ventilation will be initiated. Hypnotic depth will be continuously monitored with a BIS value between 45 and 55, and propofol administration will be adjusted to maintain this target. Remifentanil will be paused once no more painful stimulation occurs. At this point, noxious stimulation will be delayed at least 10 minutes after remifentanil is paused to allow for the remifentanil to be cleared out. During this period, time will be judiciously used to position the patient, apply sterile drape and prepare surgical equipment. Finally, a "no pain period" of 5 min, during which the patient faces no stimulation or manipulation, will take place. During this 5-min Pre-Stimulation period, resting-state EEG recording will be recorded. Measurements of the NOL Index, BIS, HR and MAP will start 1 min before applying noxious stimulus (means will constitute the basal values for NOL and HR for the Pre-Stimulation condition). Electric stimulation will then be applied. This will consist in a standardized tetanic stimulation to the ulnar nerve of the non-dominant forearm delivered by a routine nerve stimulator (100 Hz, 70 mA) for a duration of 30 sec.47 Hd-EEG recording and simultaneous measurements of the NOL Index, BIS, HR and MAP will take place during the 30 sec of tetanic stimulation (Stimulation), and will continue during the 5 min after (Post-Stimulation). This window of recording will be kept free of any other external stimulation. Once this period over, the EEG net will be removed and the investigators will exit the operating room. Surgery will begin and anesthetic procedures will continue as standard of care. Data analyses: Data analyses will be led at CIUSSS-NIM. Hd-EEG data will be bandpass filtered from 0.1 to 50 Hz, and non-scalp channels will be discarded. Noisy epochs and channels, muscle and non-physiological artifacts will be removed. EEG data will be re-referenced to an average reference. For each patient-condition, the investigators will construct functional connectivity networks using the amplitude envelope correlation (AEC), weighted phase lag index (wPLI)48 and directed phase lag index (dPLI),49 in delta (0.5-4Hz), theta (4-7 Hz), alpha (8-13 Hz) and beta (14-30 Hz) frequency bands for all pairwise combinations of electrode channels on 10-sec windows. From AEC and wPLI networks, the investigators will calculate graph theoretical properties (e.g. binary small-worldness, clustering coefficient, modularity, global efficiency) to appraise overall functional integration and differentiation. dPLI matrices will be used to assess degree and direction of frontoparietal connectivity across conditions. Variations in the BIS (delta-BIS) and NOL (delta-NOL) across conditions, as well as the peak value of the NOL (peak-NOL) during the Stimulation and Post-Stimulation conditions will be recorded. Statistical analyses: The investigators will compare the Awake, Pre-Stimulation, Stimulation and Post-Stimulation conditions on the various functional connectivity and network properties using independent-samples t-tests and ANOVA with age and sex as covariates, in order to appraise differences across conditions. The investigators will then use k-means clustering analysis to identify the measures most capable of distinguishing conditions with and without noxious simulation. The most salient features identified using k-means will then be correlated with the induced delta-BIS and delta-NOL, and peak-NOL, to investigate the relationship between these variables. Finally, the investigators will use a deep machine learning approach to identify the combination of 20 EEG channels that can most strongly detect pain-induced alterations in the functional connectivity and network properties. P-values will be corrected for multiple comparisons with familywise error correction. If the chosen functional connectivity and network markers do not yield statistical differences in response to noxious stimulation in SA1, the investigators will proceed to a deep machine learning approach with feature learning, using convolution architecture, in order to identify the EEG features (spectral, temporal and spatial) that show strongest alterations in response to noxious stimulation. Risks and inconveniences : There are no risks anticipated with participation in this study. The inconvenience associated to electrode placement are minimal, and include potential irritation of the scalp, which disappears on its own. No experimental drugs will be used

Conditions

Interventions

Eligibility

Locations (1)

Outcomes

Central Contacts