The Effect of Changes in Cerebral Oximeter Values During Cardiac Surgery on the Incidence of Postoperative Cognitive Dysfunction (POCD): a Retrospective Study Based on Propensity Score Matched Analysis


 Background The occurrence of postoperative cognitive dysfunction(POCD) after major cardiac surgery is associated with an increase in perioperative mortality and morbidity. The increased regional cerebral oxygen saturation(rSO2) value is likely related to increased POCD due to oxidative stress. The incidence of POCD in relation to an increase in rSO2 values is not clinically known yet.Methods Among 3482 patients underwent cardiac surgery, 976 patients were allocated for this retrospective study. Of these, 230 patients (32.5%) were observed to have postoperative neurologic symptoms. After propensity score 1:2 ratio matching, a total of 690 patients were included in the analysis. Recorded data on occurrence of POCD from the postoperative period to predischarge were collected from the electronic records. Results The mean baseline rSO2 value was higher in the POCD (–) group than in the POCD (+) group (61.9 ± 9.6 vs 55.3 ± 11.4; P < 0.001). The mean overall minimum rSO2 value was lower in the POCD (+) group (52.2 ± 8.3 vs 48.3 ±10.5, P < 0.001). As compared with baseline, the POCD (+) group had significantly higher values in the CPB preweaning and CPB post-weaning periods (P < 0.001, respectively). There was a greater decrease in the overall minimum rSO2 values compared with baseline in the POCD (–) group (P < 0.001). The maximum mean rSO2 changes compared with baseline were higher in the POCD (+) group (adjusted odds ratio, 1.08; 95% confidence interval [CI], 1.04–1.11; P < 0.001). The AUROC for delta values of minimal compared with baseline values was 0.62 and the AUROC for the delta values of maximal compared with baseline values was 0.72.Conclusions Increased cerebral oximeter levels relative to baseline values has an effect on the development of POCD. Intraoperative oxidative damage was associated with increased POCD, and increased cerebral oximeter levels during cardiac surgery can also be a risk factor for POCD.


Introduction
The incidence of postoperative cognitive dysfunction (POCD) after major cardiac surgery ranges from 25-80%. [1][2][3] Alterations in systemic and regional perfusion and oxygenation, exposure to anesthetic agents, cardiopulmonary bypass (CPB), and rapid changes in plasma pH and cellular metabolism during cardiac surgery induce oxidative stress that can contribute to postoperative delirium. [4][5][6] POCD is associated with an increase in perioperative mortality and morbidity, longer hospital stay, and a prolonged process of rehabilitation. 7 For the early detection of POCD, regional cerebral oxygen saturation (rSO 2 ) is monitored using near-infrared spectroscopy (NIRS) during cardiac surgery. 8,9 The cerebral oximeter is widely used in cardiac surgery to detect changes in cerebral oxygen saturation and cerebral ischemia during surgery. 10 Previous studies have demonstrated a correlation between POCD, intensive care unit, and increased length of hospital stay when brain oxygen saturation levels are reduced by more than 30% relative to baseline during cardiac surgery. [10][11][12][13] However, these previous studies focused on the occurrence of cognitive dysfunction after surgeries in which the cerebral oximeter values decreased, and recent studies have focused on and proven that oxidative stress caused by oxygen can affect neuronal damage. 5,6 Lopez et al reported that oxidative damage causes POCD as a result of direct neuronal damage as well as destruction of the blood-brain barrier. 14 The extent of the oxidative stress that occurs during cardiac surgery, the changes in rSO2 values that re ect it, and the occurrence of POCD are not exactly known. Therefore, the purpose of this study is to investigate the effect of changed rSO2 on POCD in cardiac surgery through retrospective analysis.

Patients
This retrospective study was approved by the institutional review board (IRB) of Samsung Medical Centre (approval No. 2019-02-083, March 5, 2019). Because of the retrospective nature of the study, in which the medical records of patients who had already undergone the operation were analyzed, informed consent requirements were waived by the IRB. We examined the medical electronic records of adult patients aged 20 years or older who underwent cardiac surgery on CPB between January 2015 and December 2018. For patients who underwent two or more surgeries, only the nal surgery was included in the analysis. We excluded pediatric patients, patients who did not receive CPB, and patients for whom intraoperative data were missing.

Anesthesia and CPB
Anesthesia was induced, and total intravenous anesthesia was maintained with remifentanil, propofol, and rocuronium via peripheral line during operation. The anesthetic agent was titrated for a bispectral index between 40 and 60. CPB was performed using bicaval cannulation with mild hypothermia. For myocardial protection, both intermittent cold blood cardioplegia via antegrade or retrograde route and topical cooling with ice slush were used. Retrograde cardioplegia was infused using direct coronary sinus cannulation in patients with signi cant aortic regurgitation. PaCO 2 and pH were managed in accordance with the a-stat strategy during the CPB period.
Regional cerebral oximetry monitoring Regional cerebral oximetry was obtained with NIRS monitoring performed using sensors placed bilaterally on the patient's forehead during cardiac surgery (sensor and monitor: Medtronic/Covidien INVOS cerebral/somatic oximetry adult sensors, Somanetics Corporation, Troy, MI, USA). The rSO 2 baseline value was obtained while the patient was breathing room air, and subsequent rSO 2 values were recorded at ve-minute intervals throughout the duration of the surgical period beginning one minute after the sensor was placed.
Propensity score matching All patients were divided according to the presence or absence of postoperative neurologic symptoms.
We performed propensity score matching between the two groups to compensate for demographic imbalance. This method consisted of determining cases and controls and then selecting the rst case and nding the control with the closest propensity score. A logistic regression model was built given the covariates of age, sex, presence of cerebrovascular accident history, and kind of operation. We applied 1:2 nearest-neighbor matching without replacement to ensure that conditional bias was minimized.

Assessment of neurologic problem
Recorded data on occurrence of POCD from the postoperative period to predischarge were collected from the electronic records and were categorized as stroke, delirium, confusion, agitation, change in mental status, seizure, coma, new focal lesion or slowness to awaken after surgery. 15 Assessment of delirium was performed using the confusion assessment method for ICU (CAM-ICU). 16 When patients were discharged from ICU, delirium was assessed using the confusion assessment method (CAM). The CAM-ICU and CAM testers were assessed and recorded by who were trained and had been performing these tests for a number of years. Patients were assessed for delirium for the until discharge.

Statistical analysis
We performed 1:2 propensity score matching, which is a method used to reduce confounding effects in observational studies. 17 Continuous variables were compared using the t-test or Mann-Whitney U test, as appropriate. Categorical variables were analyzed using Pearson's chi-square test or Fisher's exact test, as appropriate. Data are presented as means (standard deviations [SDs]). Logistic regression analysis was performed to obtain the crude odds ratio and adjusted odds ratio of the values for each operation period's rSO 2 on POCD. A receiver-operating characteristic (ROC) curve was constructed using cerebral oximeter values, and the area under the ROC (AUROC) was calculated to assess the prediction of POCD. Statistical analyses were performed using SPSS version 22 (SPSS Inc.) and R statistical software version 3.5.3 (Vienna, Austria; https://www.r-project.org/). A P-value less than 0.05 was considered to be statistically signi cant.

Results
We assessed the study eligibility of 3482 patients (Fig. 1). Of these, 2506 patients were excluded because of the following reasons: (1) 1021 patients were pediatric patients, (2) 515 patients underwent cardiac surgery without CPB, and (3) 970 patients were excluded due to lack of intraoperative cerebral oximeter data. Therefore, 976 patients were allocated for this retrospective study. Among these, 230 patients (32.5%) were observed to have postoperative neurologic symptoms. After propensity score 1:2 ratio matching, a total of 690 patients were included in the nal analysis (Fig. 1).
Patient characteristics are shown in Table 1. Table 2 shows the rSO2 values for each operation period. Baseline rSO 2 values and overall minimum rSO 2 values were signi cantly different between the two groups (P < 0.001, respectively). The mean baseline rSO 2 value was higher in the POCD (-) group than in the POCD (+) group (61.9 ± 9.6 vs 55.3 ± 11.4; P < 0.001). The mean overall minimum rSO 2 value was lower in the POCD (+) group (52.2 ± 8.3 vs 48.3 ± 10.5, P < 0.001). As compared with baseline, the POCD (+) group had signi cantly higher values in the CPB preweaning and CPB post-weaning periods as well as maximum values (P < 0.001, respectively). There was a greater decrease in the overall minimum rSO 2 values compared with baseline in the POCD (-) group (P < 0.001). The rSO 2 values in the pre-CPB weaning and post-CPB weaning periods and the overall maximum values were not statistically signi cantly different between the two groups. Data are presented as the number of patients with percentage (%) or the mean ± SD. Data are presented as the mean ± SD. Table 3 shows the results of multivariate logistic regression analysis on the variables that were signi cant in Table 2. The maximum mean rSO 2 changes compared with baseline were higher in the POCD (+) group (adjusted odds ratio, 1.08; 95% con dence interval [CI], 1.04-1.11; P < 0.001).

Discussion
The results of this retrospective study con rmed that increased cerebral oximeter values are associated with increased POCD due to intraoperative oxidative damage. To the best of our knowledge, this is the rst study to prove clinically that that not only a decrease in rSO 2 but also an increase in rSO 2 may affect the occurrence of POCD. The use of cerebral oximeter monitoring during major cardiac surgery can reduce oxidative damage, demonstrating that it can reduce brain dysfunction and injury after surgery.
The mechanism underlying the occurrence of POCD in cardiac surgery is still unclear. However, the trend of research on the relationship between rSO 2 and the occurrence of POCD in cardiac surgery has been changing. There have been several studies on the correlation between rSO 2 and occurrence of POCD using cerebral oximeter values in cardiac surgery. Previously, most studies have focused on the desaturation event on rSO 2 , which could adversely affect the incidence of POCD and other postoperative outcomes. Cerebral desaturation is generally de ned as a decrease in saturation values to less than 70% of baseline for one minute or longer. 12 The application of an intervention algorithm to prevent a desaturation event of rSO 2 , including PaCO 2 ≥40 mm Hg, mean arterial pressure ≥60 mm Hg, and maintaining a su cient level of cerebral perfusion, was also associated with a decrease in the incidence of major organ dysfunction. 10 Among patients aged 70 years or older who underwent coronary artery bypass graft surgery, those with a reduction of ≥50% compared with baseline rSO 2 had a high incidence of early-onset POCD, and those with a reduction of ≥30% compared with baseline rSO 2 had a high incidence of late-onset POCD. 18 In contrast, the ndings of a prospective randomized trial indicated there was no difference between intraoperative rSO 2 values and POCD after cardiac surgery. 19 That is, the decrease in rSO 2 was not absolutely predictive of POCD, which is still controversial.
Meanwhile, there is no controversy about the possibility of organ injury after the transition from the ischemic state to hyperoxic reperfusion during surgery. 20 In a similar context, clinicians are not considering the potential for brain tissue injury during hyperoxygenation during reperfusion after conversion from CPB to spontaneous circulation. 21 The CPB weaning period is the period during which CPB support and mechanical ventilation support are combined, and the highest PaO 2 levels are observed in the systemic circulation and the rSO 2 values are highest during the entire operation period.
Possible mechanisms for the association between hyperoxic cerebral reperfusion and POCD include hyperoxic reperfusion-induced vasoconstriction or oxidative injury. 22,23 During hyperoxic reperfusion, oxidative neuronal damage increases, F2-isoprostanes and isofurans in plasma increase as markers of systemic oxidative neuronal damage. 24,25 F2-isoprostanes are associated with brain arteriole vasoconstriction, and isofurans have been found to be mediators between hyperoxia and POCD. 26,27 In addition to F2-isoprostanes and isofuran, S100 calcium-binding protein B, an indicator of blood-brain barrier disruption by oxidative injury, was also higher in the POCD group during cardiac surgery. 6 That is, the possibility that the blood-brain barrier can by destroyed by systemic oxidative damage during surgery and cause POCD by neuronal injury was proven.
CPB reduces platelet counts by about 50% and causes platelet dysfunction as well as reduced levels of clotting factors and von Willebrand factors. 28,29 A large volume of heparin is administered during CPB, and in the CPB weaning process, it reverses to protamine. This process may have a rebound effect of heparin, which causes bleeding after CPB weaning. 26 Therefore, the possibility of POCD by microbleeds as well as cerebral hypoxia or hyperoxia cannot be excluded. In a study of pediatric patients to detect intracerebral hemorrhage (ICH) using NIRS, 21 patients showed an increase in the NIRS value of 45% or greater, among whom 12 patients had actual ICH (sensitivity = 1.0, speci city = 0.8). 30 When acute hemorrhage occurs, hemoglobin aggregation temporarily increases the cerebral oximeter level; that is, the possibility that POCD occurred as a result of intracranial hemorrhage or microbleeds during surgery cannot be excluded. In our study, ICH was observed in a total of nine patients, and the maximal cerebral oximeter value increased by an average of 47% compared with baseline. The rapid increase in rSO 2 during conversion from CPB to spontaneous ow also means that the possibility of ICH cannot be excluded. However, an increase in the cerebral oximeter value is not an absolute ICH marker during cardiac surgery because of the effects of anesthesia, ventilator, CPB ow, and so forth. Thus, further prospective studies are needed in the future.
There are several limitations to our study. First, we did not analyze the patients by age group. Previous studies evaluating POCD have mainly focused on elderly patients, who may have increased vulnerability to neurological deterioration. However, our study has signi cance in that the results were obtained using propensity score matching to compare the occurrence of POCD according to the increase in rSO 2 regardless of age. Second, we cannot con rm the speci c time point exactly. Most of the preweaning periods and post-weaning periods in which most cerebral oximeter values were increased during surgery had maximal cerebral oximeter values, but the exact time point could not be determined. Third, we could not provide the exact cutoff value for the risk of POCD when rSO 2 increases. The second and third limitations are common limitations of all retrospective studies, and further prospective studies are needed to con rm a clear time point and cutoff value.
In conclusion, increased cerebral oximeter levels relative to baseline values has an effect on the development of POCD. Intraoperative oxidative damage was associated with increased POCD, and increased cerebral oximeter levels during cardiac surgery can also be a risk factor for POCD.
Con icts of Interest:

None
Author Contributions: Jin Hee Ahn contributed study conception, study design, data acquisition, data analysis and drafting manuscript.
Eun kyung Lee helped data interpretation and critical comments.
Do-yeon Kim and Sehee Kang helped study conception, study design and data acquisition.