Prediction Models for Cardiovascular and Neurocognitive Disease Risk in the General Population

More accurate and earlier identification of people at risk for cardiovascular disease (CVD) and neurodegenerative diseases (memory, cognition, dementia) through the appropriate use of biomarkers could lead to earlier initiation of preventive therapies and potentially avoid sometimes fatal events and complications.

Biomarkers Biomarkers are useful for determining disease risk, but also for making a diagnosis. Although there is an overabundance of predictive models of CVD and/or neurodegenerative disease risk for the general population, few have been externally validated, so they are currently not widely used by practitioners, policy makers, and designers of public health guidelines. High inter-individual variability hinders the establishment of general laws for use in predictive medicine.

Screening In addition to the lack of validation, other limitations are the low participation rate in screening campaigns (regardless of disease) and the relative difficulty, accuracy, cost, and time taken to perform the measurements.

Why anaesthesia

The perioperative period is a very good time to screen for cardiovascular and neurodegenerative diseases for several reasons:

Patients come to their anesthesia consultation and to the operating room because they have a direct visible benefit. The anaesthesia consultation is therefore often a medical consultation with quantification of the different risks and the performance of numerous complementary explorations.

• The indications for procedures under anesthesia have multiplied in recent years (digestive endoscopy, interventional cardiology, interventional neuroradiology, etc.).
• Every year, more than 12 million people in France receive anesthesia to perform surgical or interventional procedures. The criteria of quality and requirement are increasing in an aging population with numerous comorbidities and therefore more vulnerable to cardiovascular and neurodegenerative complications. More and more patients, especially the elderly and frail, have recourse to general anesthesia for scheduled surgical procedures, but they are not always well monitored because they do not have an attending physician or because they do not participate in screening or public health monitoring campaigns. In France, there are 7,400 surgical rooms in just over 1,000 public and private health institutions.
• The physiological data collected intraoperatively during systematic monitoring are very "rich" and of very good quality because they are not very noisy. Thus they lend themselves perfectly to numerical and statistical analysis, in particular via "machine learning".
• The induction of general anesthesia or the installation of a locoregional anesthesia and its maintenance represents a strong and reproducible physiological "test" for the cardiovascular and cerebral systems. An alteration of the values of the function parameters on the EEG and/or on the arterial stiffness can be considered as a symptom. The investigators believe that the analysis of the parameters during this test is very informative about the real physiological state of both the cardiovascular and cerebral systems of the patient.
• The patients are regularly re-convoked postoperatively for the follow-up of their pathology and the possible complications are recorded in their file, allowing a short and medium term follow-up.

The project aims to validate a biomarker predictive of cardiovascular complications, the pulse wave velocity (PWV), and a biomarker predictive of cognitive disorders, the power of the Alpha wave on the electroencephalogram (EEG), from the data usually collected during each anesthesia during the perioperative period.

These 2 parameters are obtained from the physiological monitoring parameters used routinely during anesthesia without any additional monitoring not required for anesthetic management.

Biomarker 1 Pulse wave velocity (PWV): Arterial stiffness is currently one of the best predictive, early and independent biomarkers of cardiovascular events. The best parameter for the assessment of arterial stiffness is the measurement of pulse wave velocity (PWVcf) between the carotid and femoral arteries. This index has been widely used and measured during general anesthesia by our team in several studies as one of the determining parameters of left ventricular afterload. Non-invasive measurement of arterial stiffness by calculation of pulse wave velocity (PWV) and systolic pulsatility index (SPI) is obtained with an automatic digital sphygmomanometer MESI mTablet (MESI®, Slovenia). Our team has developed a method for continuous estimation of PWVcf in the operating room from data obtained by the usual monitoring: an electrocardiogram (ECG), a digital photoplethysmography (PPG) and an oscillometric brachial sphygmomanometer for non-invasive blood pressure measurement (NIBP). In addition, the hemodynamic variations inherent to anesthesia allow us to explore the PWVcf as a function of arterial pressure variations, and thus to obtain a true arterial distensibility index expressed in m/s per mmHg.

Biomarker 2 Electroencephalogram EEG Dementia is a disease that affects memory and thinking to the point of disrupting normal daily activities. Neuropsychological tests (Montreal Cognitive assessment (MoCA) or biomarkers (blood, CSF, MRI, EEG) are useful to make the diagnosis but too late to detect patients at risk or not of cognitive complications because of the inter-individual variations of brain aging. The interest of perioperative brain monitoring and in particular electroencephalography (EEG) to reduce neurological and cognitive damage in surgery has been the subject of much research and is a crucial issue. It is likely that the per operative susceptibility to Burst suppression and the appearance of postoperative cognitive dysfunction are related to a set of per operative factors (anesthesia, surgery, inflammation, pain...) but also pre operative factors (cognitive state, physiological age and cerebral fragility...). Currently this susceptibility or fragility could be detected by a parameter of cerebral function monitored continuously to estimate the depth of anesthesia during anesthesia by the intraoperative EEG - frontal EEG. From the literature and preliminary results obtained in our clinical research unit, it appears that there is a relationship between the occurrence of burst suppression (BS) (%), the evolution of the 95% spectral frequency front (SEF95) and the power of the alpha band (dB) on the frontal EEG and the appearance of postoperative cognitive dysfunction.

In this context of monitoring the depth of anesthesia by EEG, quantitative EEG analysis could be a simple to use and objective bio-marker to determine physiological age and/or personalized early brain aging of anesthetized subjects.

Thanks to the acquisition media, the data will be available and of good quality guaranteeing that the tools/algorithms developed from them will not be biased and will be effective.

Patients over 45 years old are eligible to participate in this protocol. During the anaesthesia consultation, they will be given a letter of information on the objectives and progress of the study. Their non-opposition to participating in this study will be collected at the latest during the pre-anesthetic visit, the day before the operation, after a reflection period.

The protocol will start the day before the operation and will last up to 5 years. On the day of the anesthesia consultation or the day before the procedure: MoCA test, MemScreen, Finger tapping test (FTT), muscle grip strength, pulse wave velocity measurement.

The procedures will be performed under general anesthesia or locoregional anesthesia with sedation. The protocol does not interact with the care procedure.

On the day of the procedure a non-invasive monitoring of the depth of anesthesia and/or sedation from the quantitative EEG (Sedline Masimo®) ensures a continuous recording of intraoperative EEG data.

Routine monitoring includes continuous blood pressure, SpO2 and ECG measurements.

Post-operatively The same investigator performs MoCA, MemScreen and the Finger tapping test (FTT) at Day 2 and Day 5. The postoperative monitoring protocol requires twice-daily assessment of delirium by the Confusion Assessment Method (CAM) administered by the ward nurses. To avoid learning bias due to repetition of the MoCA different versions will be used during neuropsychological assessments.

Assessments at 3 months and at 1, 2, 3, 4 and 5 years include the MoCA and MemScreen neuropsychometric tests as well as FTT, the measure of the I.A.D.L (Instrumental Activities of Daily Living). They are performed during follow-up consultations or by telemedicine in the absence of a scheduled consultation. If the patient is seen during the surgical follow-up consultation, a PWV measurement will be performed before the neuropsychometric tests. This strategy does not add any additional risk and thus guarantees the safety of the patients included in the protocol.

No invasive devices are used in addition to those required for anesthesia. All of the monitoring instruments described above are already in routine use for routine care in our department. The monitoring data are currently available and are made available in the clinical research unit of our department thanks to

• A data extraction system from the PHILIPS IntelliVue monitoring solutions. The Data Warehouse Connect software solution allows the collection of all these data with a fine sampling (2ms for the tracings, 1s for the numerical data), which considerably widens the capacities in terms of data analysis and exploitation.
• The retrieval system is temporally coupled to the patient's events, medications (doses) administered through an IntelliSpace Critical Care and Anesthesia (ICCAA) information system that is operational in all operating rooms. Thus the data provided are coupled with the digital traces and measurements of the multiparametric monitor, transmitted in HL7 to the server of the hospital's IT department.

The duration of the anaesthesia is not prolonged for the study. No additional examination will be performed. The anesthetic strategy is decided by the anesthesiologist in charge of the operation.

The physician in charge of the study collecting the data does not intervene at any time in the management of the patient. The measures cannot influence the prescribing physician since at this stage the data are not yet analyzed and available.

No added risk.o invasive devices are used in addition to those required for anesthesia. All of the monitoring instruments described above are already used routinely for routine care in our department.

The investigators consider two objectives, the first aiming to demonstrate a significant association at 2 years between the occurrence of cardiovascular complications and pulse wave velocity on the one hand; the occurrence of cognitive dysfunctions and the power of the alpha band (EEG marker) on the other hand. These two objectives concern the same patient population. The investigators consider a significance level α=0.05 as well as a statistical power 1-β=0.85. The investigators consider logistic regression as a statistical test with H0: β_1=0. The two tests are considered independent and not multiple. To include comorbidities and demographics including age as possible confounders, The investigators make the moderate assumption of a multiple correlation coefficient R^2=0.5 (multivariable adjustment). The probability under the null hypothesis is 0.17 and 0.15 for the first and second test respectively. Considering in both cases an odds ratio of 2, it is obtain n_1=281 and n_2=304 patients. The sample n=max〖(n_1,n_2)〗=304. Finally, in view of the previous studies, the investigators consider that 30% of patients will be lost. Therefore, it is conclude that 396 patients should be included in this study.

Continuous data will be expressed as median [interquartile] and categorical data as n (%). Categorical variables will be compared by Mann-Whtney test and continuous variables by Wilcoxon test.

A full multivariate analysis will also be performed. All statistical analyses will be performed using R statistical software (The 'R' Foundation for Statistical Computing, Vienna, Austria). Results will be expressed as means (± standard deviation). A p-value of less than 0.05 is considered significant.

Inclusion criteria

• Patients > 45 years old
• Eligible for outpatient or scheduled surgery or interventional procedures under general anesthesia or locoregional anesthesia with sedation.
• Patient having expressed no objection to participation in this research.
• Patient who is not subject to a legal protection measure

Exclusion criteria

• Patients under 45 years of age.
• Patient opposed to participation in the protocol
• Pregnant woman
• Patient under judicial protection
• Patient not affiliated to a social health system