Cerebral Perfusion Variation During Blood Pressure Changes in ICU

The risk associated with arterial hypotension during anesthesia for intensive care sedation has been demonstrated, but the threshold at which consequences for perfusion of one or more organs appear varies according to the mechanism of hypotension, associated abnormalities (HR, cardiac output and oxygen transport) and the patient's terrain.

Currently, a mean arterial pressure greater than 60 mm Hg (1) and a reduction of less than 30-50% from the value measured before sedation are commonly used to ensure good perfusion of all organs. In intensive care, it is recommended to maintain a MAP between 60 and 70 mmHg and a Cerebral Perfusion Pressure (CPP) > 50 mmHg for neurocompromised patients with Intra Cranial pressure (ICP) measure. Normally, cerebral blood flow is self-regulated, allowing adaptation of cerebral blood flow to oxygen requirements at different levels of high and low blood pressure. However, this protective mechanism may fail for a degree of hypotension that depends on several factors such as the age or vascular status of the patient.

The aim of the study is to measure non-invasively, easily and reliably the variations of cerebral perfusion in patients with or without cardiovascular risk factors during controlled variations performed during routine care to set the blood pressure level within the recommended safety standards during sedation in intensive care unit. Cerebral perfusion and cerebral O2 adequacy will be compared and assessed by continuous and simultaneous measurements of cerebral blood flow by transcranial Doppler (CTD), cerebral O2 adequacy by near-infrared spectroscopy (NIRS) and a parameter of cerebral function by electroencephalogram (EEG) - frontal EEG.

The objective for a given patient is to detect a possible threshold pressure value that could impact blood flow velocity (Doppler) and possibly EEG and cerebral oximetry.

This clinical research is carried out on adult patients hospitalized in intensive care. Their non-opposition to participate in this study will be collected during the hospitalization or, if necessary, the trusted person/relative of a patient unable to express his/her will. The protocol starts after at least 12 hours of stablisation in the intensive care unit.

All measurements are obtained non-invasively. Monitoring includes routine continuous electrocardiogram and blood pressure monitoring by non-invasive (Clearsight®, EV1000®, Edwards Sciences) or invasive means if already in place, pulsed O2 saturation (SpO2), ventilatory parameters (RR, VC, FiO2), and depth of sedation by frontal EEG.

In the ICU, patients are sedated according to the usual protocol with midazolam or propofol and sufentanil with vasopressor support by norepinephrine. Adjustment of the blood pressure level during routine care allows to obtain mean arterial pressure levels adapted to the patient's situation and comorbidities.

In our current practice

• when blood pressure stability is obtained, the dose of norepinephrine is progressively decreased to a MAP target ≥ 70% of the baseline value without never going below 60 mmHg for patients without low cardiovascular risk factors and 80 mmHg for patients with cardiovascular risk factors or a cerebral perfusion pressure (CPP) > 50 mmHg for neurocerebral injury patients monitored by an intracranial pressure ICP

Patients over 18 years old are eligible to participate in this protocol. For ICU patients, monitoring of the depth of anesthesia using a quantitative frontal EEG (Sedline Masimo®) ensures continuous recording of intraoperative EEG data. Placement of a frontal O3® Sensor rSO2 Masimo® or Foresight sensor rSO2 Edwards® electrode for continuous measurement of cerebral tissue O2 saturation (StO2). Continuous measurement of blood velocity in the middle cerebral artery by transcranial pulsed Doppler (TCD) Atys Medical TCD-X®. The probe attachment system on the patient's head is similar to a pair of glasses; it is light and comfortable. The orientation of the robotic probe is automatically readjusted to ensure stable recording quality over time.

No additional examinations will be performed. The physician in charge of the study collecting the data is not involved in the patient's management at any time.

The monitoring data are currently available and are made available in the clinical research unit of our department using

• 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 extraction system is temporally coupled to the events that occur to the patient, the drugs (doses) administered thanks to an information system IntelliSpace Critical Care and Anesthesia (ICCAA)

The physician in charge of the study collecting the data does not participate 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.

Our main objective is to identify a significant linear relationship between the percentage change in mean velocity (Vm) and the percentage change in the 95% spectral frequency front (SEF95) during the MAP change.

We seek to reject the null hypothesis based on the R2 measure of goodness of fit: H0: R2 = 0.

Considering a type I error probability = 0.05 and a power of 85%, and an effect size of 0.3 based on preliminary data. For a single predictor, we obtain a sample size of n = 32 patients. To adjust for possible confounding factors, we assume a partial correlation coefficient rho=0.5 between Vm and the covariates. This correction gives us a sample size of 32*2 = 64 patients . Finally, preliminary data indicate that 30% of the included patients have a poor signal or defective measurements resulting from artifacts or unsuitable experimental conditions. The final number of patients to be included is estimated at N = 92.

Rejection of the primary null hypothesis will be established by a linear regression model.

As a secondary objective, linear regression will also be evaluated between Vm and EEG markers such as alpha band power, time spent in burst-suppression, and delta-to-alpha ratio.

Descriptive statistics will be reported as n (%) for categorical variables, mean(standard deviation) for continuous variables, and median[IQR] for ordinal variables that do not have a normal distribution.

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