Perioperative Hyperoxia and Delirium after On-pump Cardiac Surgery: A Mediation Analysis

BackgroundNeurologic 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. Methods46 age-and sex-matched subjects, aged [&ge;]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. Results26 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. ConclusionWhile 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.