Perioperative Hyperoxia and Delirium after On-pump Cardiac Surgery: A Mediation Analysis ======================================================================================== * Kwame Wiredu * Stefana Voicu * Heba Naseem * Ariel L Muller * Myles D Boone * Scott A. Gerber * Shahzad Shaefi ## Abstract **Background** Neurologic and neurobehavioural complications are common after cardiac surgery with cardiopulmonary bypass (CPB). Exposure to the artificial bypass surface, conversion to laminar flow and hypothermia likely contribute to systemic inflammation observed after CPB. To ensure adequate systemic oxygenation, the CPB patient is often exposed to supraphysiologic levels of oxygen. Relative to normoxia, perioperative hyperoxia during CPB has not been shown to impact neurocognition in the long-term. Whether this holds true for the immediate post-operative neurocognitive function is the question of this nested case-control study. **Methods** 46 age-and sex-matched subjects, aged ≥65 years, selected for this study were randomized to receive normoxia or hyperoxia during CABG with CPB in the parent trial. Levels of four neuroinflammatory biomarkers (S100B, ENO2, CHI3L1, UCHL1) were measured at baseline and at post-bypass. Baseline neurocognition was established with the Montreal Cognitive Assessment tool and patients were assessed on each post-operative day for delirium using the confusion assessment method. Mediation analyses was conducted for the conditional effect of perioperative oxygen treatment on the occurrence of delirium, assuming mediation effect from change in biomarker levels. **Results** 26 subjects (*n* = 12) demonstrated delirium. Of the four biomarkers, only S100B levels were differentially abundant post-bypass regardless of treatment (8.18 versus 10.15pg/mL, *p* value < 0.001). We found significant direct effects of treatment on the occurrence of delirium (effect size = -2.477, *p* = 0.004). There was no statistically significant mediating effect by S100B. **Conclusion** While perioperative hyperoxia may not be associated with neurocognitive dysfunction in the long-term, its immediate effects may contribute significantly to the occurrence of post-operative delirium. Taken together, our findings suggest a dose-response-time relationship between hyperoxia and neurocognitive function. Keywords * biomarkers * cardiopulmonary bypass * delirium * neuroinflammation * plasma ## Introduction The effects of perioperative hyperoxia on myocardial damage, acute kidney injury and long-term neurocognitive dysfunction are well documented,[1-6] but the impact on the immediate post-operative neurocognitive function is less well-characterized.[7] Globally, over 1.25 million patients undergo cardiac surgery on cardiopulmonary bypass (CPB) each year.[8] Perioperatively, the CPB population are often exposed to supraphysiologic concentrations of oxygen.[9] This practice is premised on the primary goal of maintaining end-organ perfusion as hypothermia, microcirculatory heterogeneity and interstitial fluid shifts during CPB all contribute to poor tissue oxygenation.[10] At the tissue level, hyperoxia is beneficial in the ischemic preconditioning of the myocardium, reduces overall gas microemboli and provides significant oxygen reserves in the event of interrupted ventilation.[11-13] In fact, the ability to monitor regional cerebral oxygenation in real-time has provided unequivocal evidence linking cerebral desaturations during CPB to worse clinical outcomes,[14-17] further emphasizing the need for higher oxygen concentrations. On the other hand, hyperoxia may also trigger vasoconstriction that further compromises perfusion, may instigate inflammation and worsen ischemia-reperfusion injury through increased oxidative stress.[18-21] Notwithstanding these, the prevailing observation is that hyperoxia during CPB is not associated with poor long-term neurocognitive outcomes.[2, 5, 22] Delirium is etiologically heterogeneous and lower pre-surgical cognitive reserve is a recognized risk factor.[23-25] The extent to which perioperative oxygen treatment modifies the occurrence of delirium in a typical CPB demographic with suboptimal baseline neurocognition, however, remains largely unknown. In this nested case-control study, we hypothesized that peri-operative oxygen administration significantly impacts the occurrence of post-operative delirium. We estimated the conditional effect of perioperative hyperoxia in an elderly cohort who underwent cardiac surgery on CPB. Further, we measured a panel of neuroinflammatory markers and ascertained their possible role in mediating the hyperoxia-delirium relationship. Finally, we proposed a conceptual model regarding the interaction between baseline neurocognition and perioperative hyperoxia as they relate to post-operative delirium. ## Methods ### Study design and Ethics approval Subjects in this nested case-control study were selected from the parent clinical trial[5, 26] that examined the effects of intra-operative oxygen therapy on neurocognitive outcomes among cardiac surgical patients at the Beth Israel Deaconess Medical Center, Boston MA (Trial registration number [NCT02591589](http://medrxiv.org/lookup/external-ref?link_type=CLINTRIALGOV&access_num=NCT02591589&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom), principal investigator: Shahzad Shaefi, registration date: October 29, 2015). Subjects were enrolled between July 2015 and July 2017, and all patients provided informed consent. Institutional review board (IRB) approval 2014P000398/33 was amended for the purposes of this current study on 09/17/2021 by the Committee on Clinical Investigations (CCI) at the BIDMC. **Figure 1** summarizes the design of the current study. ![Figure 1:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2022/06/08/2022.06.07.22276112/F1.medium.gif) [Figure 1:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/F1) Figure 1: Study design and schematic illustration of the nested case-control study. Baseline plasma samples (**1**) were collected prior to CABG on CPB during which patients were randomized to normoxia or hyperoxia treatment groups (**2**). Plasma samples were collected at post-bypass (**3**). Following surgery, patients were followed with daily assessments for delirium using CAM (**4**). Biomarker concentrations in plasma were measured in duplicates by bead-based multiplex assay (**5**), for subsequent data analyses (**6**). CAM: confusion assessment method; CPB: cardiopulmonary bypass; FiO2: fraction of inspired oxygen; O2 Rx: intra-operative oxygen treatment ### Patient population, Exclusion and Inclusion Criteria Details of enrollment, exclusion criteria and treatment allocation are published previously.[5, 26] Briefly, eligible participants included patients 65 years or older who were booked for elective CABG requiring CPB. Neurocognitive assessment was achieved using the telephonic Montreal Cognitive Assessment (tMoCA) as the primary endpoint. Post-operatively, subjects were assessed daily for delirium as a secondary endpoint using the confusion assessment method (CAM). Patients were excluded if they were undergoing emergent CABG, if they required single-lung ventilation, CABG without CPB, intraoperative balloon counter-pulsation or mechanical circulatory support. Subjects with MoCA scores below 10 were also excluded. ### Sample size calculation Because quantitative studies on the selected biomarkers in the context of delirium are largely unexplored, Cohen’s estimation of effect size[27] was used to determine the optimal sample size. Here, we used a standardized effect size by Cohen’s estimation, *d* = 0.8 for a paired design that provides a statistical power of 80% at a significance level of 0.05. Further, delirium cases were also matched to non-delirium controls at approximately 1:3 ratio, to a total of 46 subjects in the current study. ### Conduct of Study and Biomarker Measurements Whole blood samples at baseline and post-bypass (P-BP) were collected into 4mL EDTA-treated tubes (BD Diagnostics) and centrifuged immediately. Resulting plasma was stored at -80ºC until analyses. Limits of detection, limits of quantification and linearity of biomarker signal were established using serial dilutions of patient samples and laboratory standards. Biomarker measurements were made using a custom R&D Human Premixed Multi-Analyte Panel, a magnetic bead-based multiplex assay (Catalog Number: LXSAHM-04, Lot Number: L140030). Analyte concentrations were determined by a 5-parameter logistic (PL) regression computed from the standard curves. All biomarkers were measured in duplicates. For quality control, intra-assay variability was assessed at a cut-off of 20%. ### Statistical analyses Descriptive statistics are presented as mean (standard deviations) or count (proportion) for continuous and categorical variables, respectively. Biomarker concentrations, which were assayed in duplicates, were summarized as averages and compared between groups using Student’s *t*-test (paired, unequal variance). Structural equation modeling (SEM) was employed to ascertain the conceptual and statistical mediation models, hypothesizing a direct effect of oxygen treatment on delirium outcome, and an indirect (mediation) effect via the measured biomarkers. (**Figure 4A** and **Supplemental Figure 1**). Mediation analysis was then performed to estimate the total, indirect and direct effects of intraoperative oxygen treatment on the occurrence of delirium. Here, the direct effect of perioperative oxygen treatment was computed as the change in the odds of developing delirium in patients receiving hyperoxia versus the odds of developing delirium in patients receiving normoxia, when baseline neurocognition is fixed (i.e., holding tMoCA scores constant). Average tMoCA score was also defined as the arithmetic mean of tMoCA scores for subjects in each arm of this nested cohort. Significance of the mediation effects were computed by bootstrapping.[28] All analyses were performed in R environment for statistical computing, v4.1.1.[29] at a significance level ⍰ = 0.05. ## Results 26% of subjects (n = 12) in the nested cohort demonstrated delirium postoperatively. At baseline, delirium cases and non-delirium controls were matched by age, sex and race (**Table 1**). Details of subjects’ comorbidities and preoperative medications are reported previously.[5] Intraoperatively, there was no statistically significant differences in aortic cross-clamp or cardiopulmonary bypass times between cases and non-cases. Baseline neurocognition, as measured by the Montreal Cognitive Assessment (MoCA) tool, was generally low in this cohort, and significantly lower among cases relative to controls (*p* = 0.02). Of the non-delirium controls, 21% (*n* = 7) met the criteria for subsyndromal delirium. View this table: [Table 1:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/T1) Table 1: Baseline characteristics of study participants We measured four biomarkers well-documented to be markers of neuro-inflammation. For quality control, we set an intra-assay variability cut-off of 20%. Biomarker measurements with 20% coefficient of variation (%CV) or higher between duplicate runs were removed from all downstream analyses, although including them did not change results of our analyses. Levels of ubiquitin carboxyl-terminal hydrolase isozyme L1 (UCHL1) in all samples were below the limits of quantification (**Supplemental table 1**). Of the remaining biomarkers, only S100B levels were significantly higher at post-bypass relative to baseline levels (*p* value < 0.001, **Figure 2**). We found no significant differences in S100B levels by sex or body mass index (BMI) at baseline or post-bypass (**Table 2**). Stratified analyses also showed that the absolute increase in S100B levels is not confounded by sex, BMI, patient outcomes or intra-operative oxygen treatment (**Table 2, Supplemental figure 2**). We, however, observe that for patients with the longest CPB times (> 141 mins), the absolute change in S100B levels was not significant (*p* = 0.076, **Table 2**). View this table: [Table 2:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/T2) Table 2: Stratified analysis of baseline to post-operative change in S100B levels ![Figure 2:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2022/06/08/2022.06.07.22276112/F2.medium.gif) [Figure 2:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/F2) Figure 2: Comparison of neuro-inflammatory biomarker levels. Boxplots show the quantile distribution (median and inter-quantile ranges) of each biomarker, and a comparison of biomarker levels at baseline and at post-bypass. Statistical tool used is the Student’s *t*-test. Asterisks (*) represent statistically significant difference (i.e., *p* value < 0.05); *ns*: not significant; S100B: protein S100-B; ENO2: gamma-enolase; CHI3L1: chitinase-3-like protein 1. There were significant conditional and total effects of intra-operative oxygen treatment on the occurrence of delirium (effect estimate = -2.477, *p* = 0.004, **Tables 3** and 4). We, however, did not find post-bypass changes in S100B levels to mediate the oxygen-delirium relationship (indirect effect = 0.002, *p* = 0.584, **Table 3** and **Figure 3**). View this table: [Table 3:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/T3) Table 3: Conditional effects and regression coefficients of predictors in mediation modeling View this table: [Table 4:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/T4) Table 4: Causal Mediation Analysis ![Figure 3:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2022/06/08/2022.06.07.22276112/F3.medium.gif) [Figure 3:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/F3) Figure 3: Mediation modeling and analyses. **A**: Conceptual model, and **B**: Statistical model of the relationship between delirium occurrence and intra-operative oxygen treatment, assuming a mediation role by post-bypass increases in S100B biomarker levels. Dashed lines = statistically non-significant; MoCA: Montreal Cognitive Assessment tool used to establish baseline neurocognition; O2: oxygen ## Discussion We have presented an exploratory, nested case-control study that tested the hypothesis that the amount of intraoperative oxygen administered likely contributes to the occurrence of delirium. We found a significant direct effect between hyperoxia and the odds of developing delirium. We, however, did not find a significant mediating role of neuroinflammatory biomarkers measured in our study. The effects of perioperative oxygen treatment on the occurrence of delirium, however small the effect size may be, reechoes the fact that delirium is etiologically heterogenous with likely many other possible pathomechanistic pathways besides our observation. Despite the paucity of literature on the relationship between perioperative hyperoxia and post-operative delirium, result of our study is consistent with recent studies investigating the matter.[7, 30] In one of the aforementioned studies, Lopez, Pandharipande [7] monitored the duration of cerebral hyperoxia with oximetry monitors intraoperatively and found that despite the considerable fluctuations in cerebral oxygenation, hyperoxia after a period of hypoxia was most strongly associated with the occurrence of delirium. Authors also observed that in delirium subjects, there was an increase in plasma concentrations of markers of oxidative stress (F2-isoprostanes and isofurans), suggesting a possible mediation role. In the remaining study, Kupiec, Adamik [30] established a maximum PaO2 cut-off of 33.2 kPa, beyond which post-operative delirium was more likely to occur. Notable differences between these studies and ours is the relatively smaller sample size in our study, the choice of the baseline neurocognitive assessment tool and the frequency of assessments for post-operative delirium. Traditionally, ascertaining the mediation effect of a variable ***M*** (e.g., biomarker levels) on the relationship between a predictor ***X*** (e.g., peri-operative oxygen treatment) and outcome ***Y*** (delirium) has required that a relationship already exists between ***X*** and ***Y***. This approach is heavily debated, and proposed alternatives suggest that a prior relationship between ***X*** and ***Y***, or the lack thereof, neither proves nor rules out causal associations[31-33]. This was the observation in our study, in which we found that the relationship between ***X*** and ***Y*** is moderated by the interaction of another variable W (baseline neurocognition). The clinical significance of the post-bypass increases in S100B levels without any significant associations with delirium remains to be determined. This is also the observation by Jönsson, Johnsson [34] and Nguyen, Huyghens [35]. Aptly described by authors as the “controversial significance of early S100B levels after cardiac surgery”, Jönsson, Johnsson [34] measured S100B levels at defined intervals from end of bypass until 48 hours post-surgery, and concluded that the predictive significance of the S100B biomarker is limited.[34] These findings are in sharp contrast to many other studies in which S100B levels were consistently increased in delirium cases regardless of the sampling time after surgery[36-38]. These conflicting findings about the S100B-delirium relationship highlights three possibilities: (1) that plasma levels of S100B do not necessarily reflect CSF levels, (2) the relatively short half-life of S100B (60 to 120 mins) requires that blood sampling is appropriately timed, or (3) that increases in S100B levels only signal neuronal response to the surgical insult, and not because of the occurrence of delirium. With regards to the gap in plasma-to-CSF levels, S100B is also secreted by extra-neuronal tissues (e.g., adipocytes)[39, 40], although the predominant source remains in mature, perivascular astrocytes[41, 42]. To ascertain that the post-bypass increases in S100B levels in our cohort were not confounded by body fat, we performed stratified analyses and found no differences in S100B levels by sex or by BMI. Further, although there was no association between CPB duration and incidence of delirium, we observe a downtrend in the absolute change in S100B levels with increasing CPB times and found no significant change in S100B levels for subjects with the longest CPB times (> 141 mins). We intimate that this is likely due to the short half-life of the S100B protein, although this hypothesis will require a formal interrogation in future studies. Our study is, however, without limitations, notable among them is sample size and selection bias. Our relatively small sample size (*n* = 46) did not allow for the statistical adjustment of covariates such as age, sex, BMI in the structural modeling and mediation analyses. We, however, controlled for possible confounding during the study designing stage by matching cases to controls by age, sex, BMI and race (**Table 1**), and do not expect the lack of statistical adjustments to have any significant impact on the strength of associations in our findings. To prevent selection bias, subjects in the present study were selected to reflect the distribution of treatment (hyperoxia versus normoxia) and outcome (delirium case versus non-delirium control) in the parent trial. We acknowledge that our cohort had considerable deficits in neurocognition at baseline. While the ideal choice of controls would be subjects without any baseline deficits, controls are more appropriately sampled from, and in terms of risk, should be representative of the very population that gave rise to the cases being investigated[43, 44]. Nonetheless, we excluded subjects with extremely low tMoCA scores (< 10). In addition, patients’ baseline tMoCA scores were included, and statistically adjusted for, in our models. There are several mechanisms proposed to underlie the possible neurological damage after cardiac surgery on CPB [45]. In our study, we focused on the neuroinflammatory mechanism as a possible mediator of the exposure-to-outcome relationship. It is likely that our choice of biomarkers, albeit their recognized associations with neuroinflammation, may not be directly involved in the pathogenesis of post-bypass delirium. Given the half-lives and turnover of many inflammatory proteins, it is equally likely that our timing of blood sampling did not permit detection of biomarker level that accurately reflects any possible neuroinflammatory process that may have been at play. To date, our study remains the only one that has investigated the effects of perioperative hyperoxia on both the immediate and long-term neurocognitive functions [5] in the same cohort of patients. Taken together, the findings that hyperoxia increases the risk of post-operative delirium, yet with no association with long-term cognitive decline, may best be explained by a dose-response-time relationship, although further studies are required to definitely interrogate these hypotheses. ## Data Availability All data produced in the present study are available upon reasonable request to the authors ## Figure Captions and Table Legends ![Supplemental figure 1:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2022/06/08/2022.06.07.22276112/F4.medium.gif) [Supplemental figure 1:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/F4) Supplemental figure 1: Structural equation model diagram. Delirium is the dependent latent variable; intra-operative oxygen treatment is the independent variable and post-bypass change in S100B biomarker levels is the assumed mediator of the oxygen-delirium relationship ![Supplemental Figure 2](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2022/06/08/2022.06.07.22276112/F5.medium.gif) [Supplemental Figure 2](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/F5) Supplemental Figure 2 View this table: [Supplemental Table 1:](http://medrxiv.org/content/early/2022/06/08/2022.06.07.22276112/T5) Supplemental Table 1: Measured Biomarkers and their Dynamic Range Table 1: Baseline characteristics of study participants Table 2: Stratified analysis of baseline to post-operative change in S100B levels Table 3: Conditional effects and simultaneous regression coefficients of predictors in mediation modeling. β: regression coefficients; CE: estimate of conditional effects; SE: bootstrap standard error; *p*: p values; *: significant association Table 4: Causal Mediation Analysis. ACME: average causal mediation effects; ADE: average direct effects ## Acknowledgements Funding for this work was supplied partly by the Burroughs Wellcome training grant to KW and via K08 GM134220 and R03 AG060179 to SS. * Received June 7, 2022. * Revision received June 7, 2022. * Accepted June 8, 2022. * © 2022, Posted by Cold Spring Harbor Laboratory This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/) ## References 1. 1.Belboul, A., The effects of hyperoxia during cardiopulmonary bypass on blood cell rheology and post operative morbidity associated with cardiac surgery. J Extra-Corp Tech, 1991. 23: p. 43–48. 2. 2.Fontes, M.T., et al., Arterial hyperoxia during cardiopulmonary bypass and postoperative cognitive dysfunction. Journal of cardiothoracic and vascular anesthesia, 2014. 28(3): p. 462–466. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1053/j.jvca.2013.03.034&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=23972739&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) 3. 3.Smit, B., et al., Moderate hyperoxic versus near-physiological oxygen targets during and after coronary artery bypass surgery: a randomised controlled trial. Critical Care, 2016. 20(1): p. 1–10. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26728475&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) 4. 4.Abou-Arab, O., et al., Hyperoxia during cardiopulmonary bypass does not decrease cardiovascular complications following cardiac surgery: the CARDIOX randomized clinical trial. Intensive Care Medicine, 2019. 45(10): p. 1413–1421. 5. 5.Shaefi, S., et al., Intraoperative Oxygen Concentration and Neurocognition after Cardiac Surgery: A Randomized Clinical Trial. Anesthesiology, 2021. 134(2): p. 189–201. 6. 6.Bae, J., et al., Association between intraoperative hyperoxia and acute kidney injury after cardiac surgery: a retrospective observational study. Journal of Cardiothoracic and Vascular Anesthesia, 2021. 35(8): p. 2405–2414. 7. 7.Lopez, M.G., et al., Intraoperative cerebral oxygenation, oxidative injury, and delirium following cardiac surgery. Free Radic Biol Med, 2017. 103: p. 192–198. 8. 8.McGuinness, S.P., et al., A multicenter, randomized, controlled phase IIb trial of avoidance of hyperoxemia during cardiopulmonary bypass. Anesthesiology, 2016. 125(3): p. 465–473. 9. 9.Grocott, B.B., et al., Oxygen Management During Cardiopulmonary Bypass: A Single-Center, 8-Year Retrospective Cohort Study. Journal of Cardiothoracic and Vascular Anesthesia, 2021. 35(1): p. 100–105. 10. 10.Kirklin, J.K., E.H. Blackstone, and J.W. Kirklin, Cardiopulmonary bypass: studies on its damaging effects. Blood purification, 1987. 5(2-3): p. 168–178. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1159/000169466&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=3497647&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) 11. 11.Georgiadis, D., et al., Influence of oxygen ventilation on Doppler microemboli signals in patients with artificial heart valves. Stroke, 1997. 28(11): p. 2189–2194. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToic3Ryb2tlYWhhIjtzOjU6InJlc2lkIjtzOjEwOiIyOC8xMS8yMTg5IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjIvMDYvMDgvMjAyMi4wNi4wNy4yMjI3NjExMi5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 12. 12.Tähepõld, P., et al., Hyperoxia elicits myocardial protection through a nuclear factor κB-dependent mechanism in the rat heart. The Journal of Thoracic and Cardiovascular Surgery, 2003. 125(3): p. 650–660. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1067/mtc.2003.36&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12658209&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000181949800030&link_type=ISI) 13. 13.Karu, I., et al., Exposure to sixty minutes of hyperoxia upregulates myocardial humanins in patients with coronary artery disease-a pilot study. Journal of Physiology and Pharmacology, 2015. 66(6): p. 899–906. 14. 14.Murkin, J.M., et al., Monitoring brain oxygen saturation during coronary bypass surgery: a randomized, prospective study. Anesthesia & Analgesia, 2007. 104(1): p. 51–58. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1213/01.ane.0000246814.29362.f4&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17179242&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000243040100011&link_type=ISI) 15. 15.Slater, J.P., et al., Cerebral oxygen desaturation predicts cognitive decline and longer hospital stay after cardiac surgery. The Annals of thoracic surgery, 2009. 87(1): p. 36–45. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.athoracsur.2008.08.070&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19101265&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000261834000006&link_type=ISI) 16. 16.Zorrilla-Vaca, A., et al., Intraoperative cerebral oximetry-based management for optimizing perioperative outcomes: a meta-analysis of randomized controlled trials. Canadian Journal of Anesthesia/Journal canadien d’anesthésie, 2018. 65(5): p. 529–542. 17. 17.Deschamps, A., et al., Cerebral oximetry monitoring to maintain normal cerebral oxygen saturation during high-risk cardiac surgery: a randomized controlled feasibility trial. Anesthesiology, 2016. 124(4): p. 826–836. 18. 18.Ihnken, K., et al., Normoxic cardiopulmonary bypass reduces oxidative myocardial damage and nitric oxide during cardiac operations in the adult. The Journal of thoracic and cardiovascular surgery, 1998. 116(2): p. 327–334. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0022-5223(98)70134-5&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=9699587&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000075104300023&link_type=ISI) 19. 19.Gerritsen, W., et al., Off-pump versus on-pump coronary artery bypass grafting: oxidative stress and renal function. European journal of cardio-thoracic surgery, 2001. 20(5): p. 923–929. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S1010-7940(01)00941-1&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=11675176&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) 20. 20.Farquhar, H., et al., Systematic review of studies of the effect of hyperoxia on coronary blood flow. American heart journal, 2009. 158(3): p. 371–377. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.ahj.2009.05.037&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19699859&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000269641200008&link_type=ISI) 21. 21.Spoelstra-de Man, A., et al., Cardiovascular effects of hyperoxia during and after cardiac surgery. Anaesthesia, 2015. 70(11): p. 1307–1319. 22. 22.Toraman, F., et al., Adjusting oxygen fraction to avoid hyperoxemia during cardiopulmonary bypass. Asian Cardiovascular and Thoracic Annals, 2007. 15(4): p. 303–306. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/021849230701500407&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17664202&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) 23. 23.Knaak, C., et al., Presurgical cognitive impairment is associated with postoperative delirium and postoperative cognitive dysfunction. Minerva anestesiologica, 2020. 86(4): p. 394–403. 24. 24.Kapoor, P., et al., Prevalence of preoperative cognitive impairment in older surgical patients.: A systematic review and meta-analysis. Journal of Clinical Anesthesia, 2022. 76: p. 110574. 25. 25.Fritz, B.A., et al., Preoperative cognitive abnormality, intraoperative electroencephalogram suppression, and postoperative delirium: a mediation analysis. Anesthesiology, 2020. 132(6): p. 1458–1468. 26. 26.Shaefi, S., et al., Intraoperative oxygen concentration and neurocognition after cardiac surgery: study protocol for a randomized controlled trial. Trials, 2017. 18(1): p. 600. 27. 27.Cohen, J., Statistical power analysis. Current directions in psychological science, 1992. 1(3): p. 98–101. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/1467-8721.ep10768783&link_type=DOI) 28. 28.Preacher, K.J. and A.F. Hayes, SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior research methods, instruments, & computers, 2004. 36(4): p. 717–731. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3758/BF03206553&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15641418&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000225674500016&link_type=ISI) 29. 29.R Core Team, R: A Language and Environment for Statistical Computing, in R Foundation for Statistical Computing. 2013: Vienna, Austria. 30. 30.Kupiec, A., et al., Intra-operative hyperoxia and the risk of delirium in elderly patients after cardiac surgery. Aging (Albany NY), 2020. 12(8): p. 7006. 31. 31.Shrout, P.E. and N. Bolger, Mediation in experimental and nonexperimental studies: new procedures and recommendations. Psychological methods, 2002. 7(4): p. 422. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1037/1082-989X.7.4.422&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12530702&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000180256800002&link_type=ISI) 32. 32.Bollen, K.A., Structural equations with latent variables. Vol. 210. 1989: John Wiley & Sons. 33. 33.Hayes, A.F., Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. 2017: Guilford publications. 34. 34.Jönsson, H., et al., Controversial significance of early S100B levels after cardiac surgery. BMC neurology, 2004. 4(1): p. 1–7. 35. 35.Nguyen, D.N., et al., Serum S100B protein could help to detect cerebral complications associated with extracorporeal membrane oxygenation (ECMO). Neurocritical care, 2014. 20(3): p. 367–374. 36. 36.van Munster, B.C., et al., Markers of cerebral damage during delirium in elderly patients with hip fracture. BMC neurology, 2009. 9(1): p. 1–7. 37. 37.van Munster, B.C., et al., Cortisol, interleukins and S100B in delirium in the elderly. Brain and cognition, 2010. 74(1): p. 18–23. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.bandc.2010.05.010&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20580479&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000280944000003&link_type=ISI) 38. 38.Taylor, J., et al., Postoperative delirium and changes in the blood–brain barrier, neuroinflammation, and cerebrospinal fluid lactate: a prospective cohort study. British Journal of Anaesthesia, 2022. 39. 39.Steiner, J., et al., S100B serum levels are closely correlated with body mass index: an important caveat in neuropsychiatric research. Psychoneuroendocrinology, 2010. 35(2): p. 321–324. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.psyneuen.2009.07.012&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19674846&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) 40. 40.Gonçalves, C.A., M.C. Leite, and M.C. Guerra, Adipocytes as an important source of serum S100B and possible roles of this protein in adipose tissue. Cardiovascular psychiatry and neurology, 2010. 2010. 41. 41.Ahlemeyer, B., et al., S-100β protects cultured neurons against glutamate-and staurosporine-induced damage and is involved in the antiapoptotic action of the 5 HT1A-receptor agonist, Bay x 3702. Brain research, 2000. 858(1): p. 121–128. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0006-8993(99)02438-5&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=10700604&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000085638700016&link_type=ISI) 42. 42.Steiner, J., et al., Evidence for a wide extra-astrocytic distribution of S100B in human brain. BMC neuroscience, 2007. 8(1): p. 1–10. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/1471-2202-8-1&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17199885&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) 43. 43.Wacholder, S., et al., Selection of controls in case-control studies: I. Principles. American journal of epidemiology, 1992. 135(9): p. 1019–1028. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/oxfordjournals.aje.a116396&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=1595688&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1992HY09200008&link_type=ISI) 44. 44.Lewallen, S. and P. Courtright, Epidemiology in practice: case-control studies. Community eye health, 1998. 11(28): p. 57. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17492047&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom) 45. 45.Nah, H.W., et al., New brain infarcts on magnetic resonance imaging after coronary artery bypass graft surgery: lesion patterns, mechanism, and predictors. Annals of neurology, 2014. 76(3): p. 347–355. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/ana.24238&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25111072&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F06%2F08%2F2022.06.07.22276112.atom)