Relationship Between Sleep EEG, Intraoperative EEG and Postoperative Cognitive Dysfunction

Overview

The challenge of new recommendations and better adapted practices is pushing to operate patients who are getting older and more fragile. In this context, there is an inevitable increase in the risks associated with care and in particular perioperative neurological complications, of which postoperative cognitive dysfunction (POCD) is the most common. The interest of perioperative cerebral monitoring and in particular electroencephalography (EEG) to reduce neurological and cognitive damage in surgery has been the subject of abundant research and corresponds to a crucial issue. From the literature and preliminary results obtained in our clinical research unit, it appears that there is also a relationship between certain characteristics of the peroperative EEG (signal strength and burst suppression) and the occurrence of postoperative cognitive dysfunction (PCOD). In this context, quantitative analysis of the preoperative sleep EEG and the EEG obtained during general anesthesia could allow the identification of a simple to use biomarker of susceptibility or fragility. To our knowledge, there is no preoperative evaluation strategy using EEG analysis to detect a predisposition to POCD. The main goal of this observational clinical study is to extend the traditional use of per-operative EEG with pre-operative and post-operative sleep EEG for the detection and prediction of early post-operative cognitive dysfunction.

The interest of perioperative brain monitoring to reduce neurological and cognitive damage in surgery has been the subject of much research and corresponds to a crucial issue. The notion of quantitative EEG covers all the electrical signal processing techniques (fast Fourier transformation, evoked potentials, EEG mapping, etc.) made possible through the digitization of EEG data and the development of software to process these signals. In this context, EEG-derived indices have been set up to monitor hypnosis during general anesthesia. Between 0 (deep hypnosis) and 100 (awakening), the Bispectral Index (BIS) integrates in its calculation the percentage of fast waves, the degree of wave synchronization and the periods of EEG suppression (or "Burst suppression") and allows an adaptation of the dosage of anesthetic products according to its value. Avoiding too deep anesthesia through neuro-monitoring could help to reduce postoperative delirium by reducing the burst suppression pattern. Pre-operative cognitive status assessment : The investigators observe a great delay in the means implemented to have an objective evaluation of cognitive disorders in preoperative care. One of the pitfalls of our anesthetist specialty is the lack of continuity of medical care in the perioperative period. The geriatric evaluation including a MMSE (mental mini state examination) and a MoCA (Montreal cognitive assessment) is performed only occasionally. Interest of the resting pre-operative EEG : Sleep EEG analysis is already used in cognitive studies. Slow Wave Activity (SWA) is a parameter known to be inversely correlated with chronological age and lowered more significantly in the context of cognitive dysfunction. The power of spindles or sleep spindles and particularly fast spindles have been shown to be potentially predictive of poor cognitive development in patients already showing symptoms (Mild Cognitive Impairment). It is likely that per operative susceptibility to Burst suppression and the appearance of postoperative cognitive dysfunction are linked to a set of per operative factors (anesthesia, surgery, inflammation, pain...) but also preoperative factors (cognitive state, physiological age and cerebral fragility...). The investigators hypothesize that there is an association between preoperative resting EEG abnormalities and the risk of postoperative cognitive dysfunction. The goal of this observational clinical study is to assess the fragility of a patient under general anesthesia and based on the perioperative quantitative analysis EEG. Thus, this project should help specialists in perioperative medicine to identify patients at risk of cognitive dysfunction and could lead to proposing strategies for its prevention. Measures are based on neuro-psychological tests (Montreal cognitive assesment -MoCA) and EEG signal and subfrequency power spectral densities including alpha band as well as the amount of Propofol, each measured for EEG periods where SEF 95 = 8-13Hz. Analysis includes the characteristics of the preoperative sleep EEG (SWA, fast spindles). Experimental design: This is a single-center, observational prospective study. Population concerned : Patients eligible for interventional neuroradiology or surgery performed under General Anesthesia are selected to participate in this prospective, observational, single-center, routine care study. Patients will be included if they received, for a non-urgent scheduled procedure. Patients over 18 years of age are eligible to participate in this protocol. During the anesthesia consultation, they will be given a hand-delivered letter of information on the objectives and the course of the study. Their non-opposition to participating in this study will be collected at the latest during the pre-anesthetic visit, on the eve of the operation after a period of reflection. The protocol will start the day before the intervention. The interventions will be performed under general anaesthesia. The protocol does not interact with the care procedure. All the measures are obtained non-invasively. The day before the procedure: performance of a sleep EEG (DREEM®) and a MoCA test. On the day of the procedure, monitoring of the depth of anesthesia using the quantitative EEG (Sedline Masimo®) ensures continuous recording of intraoperative EEG data. Postoperatively the same investigator performs MoCA at Day1, Day2 and Day5. A sleep EEG (DREEM®) is performed at Day2 and Day5 if the patient is still hospitalized. During their general anesthesia, all patients' routine monitoring will consist in collecting the following parameters: electrocardiogram, pulsated oxygen saturation, endtidal CO2, respiratory rate, tidal volume and neuromuscular function. For all patients whatever the comorbidities, anesthesia induction will be performed using a target-controlled infusion (Orchestra® Base Primea - Fresenius Kabi France). No additional additional examination will be performed. The anesthetic strategy is decided by the anesthetist in charge of the procedure. The doctor in charge of the study collecting the data does not interfere at any time in the management of the patient. The measurements cannot influence the prescribing physician since at this stage the data are not yet analysed and available. Number of selected subjects: Assuming an incidence of DCPO of 30%, with power at 80% and alpha risk at 5%, it is necessary to include 100 patients to demonstrate an area under the ROC curve of 80% with a confidence interval width of 0.1 Statistics: Continuous data will be expressed in median \[interquartile\] and qualitative data in n (%). Categorical variables will be compared by Mann-Whitney test and continuous variables by Wilcoxon test. In this study, the primary endpoint was the occurrence of post-operative cognitive dysfunction, defined as a decrease in MoCA score of more than 4 points between preoperative and postoperative periods (worst value of the score measured at D1, D2 and D5). The main exposure is the quantitative score obtained from the analysis of the pre-operative sleep EEG. The ability of the various parameters of interest of the preoperative sleep EEG to predict the occurrence of DCPO will be estimated by constructing an ROC curve. The area of these ROC curves will be measured and the corresponding 95% confidence interval estimated. The areas under the ROC curve of the different parameters will be compared by DeLong test. A uni- and then multivariate analysis will be performed to identify the clinical parameters associated with the occurrence of a DCPO. The possible added value of the sleep EEG parameters on the clinical parameters will be evaluated by reclassification techniques. The statistical analysis will be carried out using the statistical analysis software "R" (The R Foundation, Vienna, Austria)