210 L . Andersen et al .: J Extra Corpor Technol 2023 , 55 , 209 – 217
Material and methods Patients
This study was based on data from patients admitted for cardiac surgery at the Heart Centre of Umeå University Hospital between August 2019 and March 2020 . The cohort comprised a subset of 22 patients selected at the discretion of the anesthesiologist as suitable for online monitoring of the regional cerebral oxygen saturation . Patients ( n = 12 ) requiring deep hypothermic circulatory arrest were however excluded . The remaining cohort ( n = 10 ) formed the final dataset . Descriptive characteristics are listed in Table 1 .
Ethics
The Regional Ethical Review Board in Umeå approved the study protocol ( DNr-2018 / 436-31 ). Patient consent was waived .
Intraoperative indices
CPB was conducted in a standard fashion maintaining MAP > 50 mmHg , mixed venous oxygen saturation > 70 %, non-pulsatile roller pump perfusion ( Stöckert S5 , LivaNova , London , UK ), and membrane oxygenation . Normocapnia ( 5 – 6 kPa ) and normal arterial oxygen tension ( 15 – 20 kPa ) were aimed for and verified by intermittent blood gas analyses .
Patient monitoring included at least one arterial blood pressure preferably from the radial artery , central venous pressure via the internal jugular vein , 5-lead electrocardiography , peripheral oxygen saturation , and nasal and / or bladder temperature . Induction of anesthesia was performed using propofol and fentanyl combined with rocuronium for skeletal muscle relaxation . Sevoflurane was used to maintain anesthesia throughout the intraoperative period . Blood pressure was controlled by norephedrine or intermittent boluses of phenylephrine .
Cerebral oximetry
The INVOS 5100C TM ( Medtronic Inc , Solna , Sweden ) monitor measures the regional cerebral oxygen balance ( rSO 2 ) in the prefrontal cortex by emitting infrared light at 730 nm and 810 nm captured by photodetector sensors placed on the left and right side of the forehead . The degree of light absorption at the specified frequencies translates to the actual hemoglobin oxygen saturation level in the arterial , capillary , and venous compartments , where the contribution from the venous saturation is approximately 75 % [ 14 , 20 ]. Median normative rSO 2 in conscious adult cardiac surgical patients has previously been reported at 66 % [ 21 ]. It has been established that changes in the rSO 2 level serve as a good estimate of the CBF when using transcranial Doppler assessments ( TCA ) as a reference [ 13 , 15 – 17 , 22 ]. MAP and rSO 2 were sampled at 0.2 Hz .
Cerebral oxygenation index
The COx index can either be analyzed in the frequency or time domain [ 23 ]. In this study , we focused on analysis in the
Table 1 . Patient characteristics ( n = 10 ).
Characteristics |
Age ( years ) |
65 ( 34 ) |
Male gender ( n ) |
10 |
Body surface area ( m 2 ) |
1.9 ( 0.4 ) |
Hypertension ( n ) |
1 |
Diabetes mellitus ( n ) |
3 |
Current or prior smoker ( n ) |
2 |
Prior carotid artery disease ( n ) |
0 |
Prior stroke ( n ) |
1 |
Prior transient ischemic attack ( n ) |
1 |
Type of surgery |
Coronary artery bypass grafting ( n ) |
2 |
Composite aortic grafting ( n ) |
5 |
Aortic valve replacement ( n ) |
3 |
Laboratory values |
Haemoglobin ( g / L ) |
147 ( 18 ) |
Creatinine ( lmol / L ) |
84 ( 24 ) |
Pre CPB-haematocrit (%) |
43.0 ( 5.8 ) |
Nadir CPB-haematocrit (%) |
32.5 ( 4.3 ) |
Intraoperative notations |
Surgery ( min ) |
254 ( 106 ) |
Cardiopulmonary bypass ( min ) |
155 ( 64 ) |
Aortic cross clamp ( min ) |
86 ( 54 ) |
Nadir body temperature (° C ) |
33.8 ( 7.2 ) |
COx |
0.02 ( 0.11 ) |
Mean arterial pressure CPB ( mmHg ) |
58 ( 6.5 ) |
Cerebral oxygen saturation balance (%) |
72 ( 6.0 ) |
Scale variables : Median ( IQR ). CPB : Cardiopulmonary bypass .
time domain . The COx index was defined by the calculated Pearson correlation coefficient ( r ) using 30 pairs of MAP and rSO 2 database recordings updated every 5th second ( 0.2 Hz ). Several different thresholds have been suggested in the literature to identify the loss of CA [ 24 ]. In this study , a COx of > 0.4 indicated that CBF is controlled by MAP , whereas COx 0.4 indicates that CA is preserved and not influenced by MAP variations [ 2 – 4 , 18 ]. Collected data were manually filtered to exclude erroneous values following flushing of blood pressure catheters or invalid COx calculations , mostly caused by loss of nominator-denominator variance . Left and right rSO 2 channels were merged using the mean value for analysis .
The correlation coefficient ( r ) estimates the linear relation between two signals . To estimate if a high correlation reflects a “ true ” co-variation or only is a random or spurious finding , surrogate data analysis can be applied [ 25 ]. The analysis is then repeated after replacing one of the signals with a set of synthesized signals , where the generated signals normally have the same frequency components as the replaced signal . The results can then be used to construct confidence intervals that can be used to evaluate the significance of the results from the interpatient registrations .
To test how often high positive or negative COx values occur in data that are not correlated , we applied surrogate data analysis as follows [ 25 ]. For each case , we calculated the COx index based on rSO 2 and MAP from the same subjects , but for each subject , we also created nine artificial COx values after replacing MAP with data from the other nine subjects . The variation over time in the COx index was then calculated for