Cognitive function after sevoflurane- vs propofol-based anaesthesia for on-pump cardiac surgery: a randomized controlled trial

British Journal of Anaesthesia 106 (6): 840 50 (2011) Advance Access publication 25 April 2011. doi:10.1093/bja/aer091 Cognitive function after sevoflurane- vs propofol-based anaesthesia for on-pump cardiac surgery: a randomized controlled trial J. Schoen*, L. Husemann, C. Tiemeyer, A. Lueloh, B. Sedemund-Adib, K.-U. Berger, M. Hueppe and M. Heringlake Department of Anesthesiology, University of Luebeck, Ratzeburger Allee 160, D-23538 Luebeck, Germany * Corresponding author. E-mail: julika.schoen@uk-sh.de Editor s key points Cognitive dysfunction after cardiac surgery might be associated with decreases in cerebral oxygen saturation. Patients developing intraoperative cerebral desaturation showed worse early postoperative cognitive test results than patients without cerebral desaturation. Hypnotic drug selection might be one of the factors attenuating the effects of cerebral desaturation on cognitive outcome after on-pump cardiac surgery. Background. Cognitive dysfunction is a frequent complication after cardiac surgery and has been found to be associated with decreases in cerebral oxygen saturation (Sc O2 ) measured with near-infrared spectroscopy. Sevoflurane has neuroprotective properties in vitro and in animal models. This study was designed to determine cognitive and clinical outcomes after sevoflurane- compared with propofol-based anaesthesia for on-pump cardiac surgery and the impact of decreases in Sc O2 under different anaesthesia regimens. Methods. One hundred and twenty-eight patients were randomly assigned to either i.v. anaesthesia with propofol- (PROP) or sevoflurane-based anaesthesia (SEVO). An intraoperative Sc O2,50% was defined as desaturation. The Abbreviated Mental Test, Stroop Test, Trail-Making Test, Word Lists, and mood-assessment tests were performed before, 2, 4, and 6 days after cardiac surgery. Markers of general outcome were obtained. Results. The analysis groups had differences in baseline cognitive performance. Analysis of variance for repeated measures (incorporating covariance of baseline scores) showed that in three of four cognitive tests, patients with cerebral desaturation showed worse results than patients without desaturation. Patients assigned to sevoflurane-based anaesthesia showed better results in all cognitive tests than patients after propofol. Interactions between the anaesthetic regimen and desaturation were found in all four cognitive tests. There were no differences in markers of organ dysfunction or general clinical outcome. Conclusions. Patients with impaired cognitive performance before operation may be at particular risk for intraoperative cerebral insult. A sevoflurane-based anaesthesia was associated with better short-term postoperative cognitive performance than propofol. Keywords: anaesthetics volatile, sevoflurane; brain, injury; clinical trials; surgery, cardiovascular Accepted for publication: 26 February 2011 Cognitive alterations after cardiac surgery are of growing importance in an ageing population. After coronary artery bypass grafting (CABG), 1 2 35 63% of patients show cognitive dysfunction on discharge with one-third still suffering from cognitive decline 5 months after surgery. 3 4 Even though some data suggest that long-term effects of surgery and anaesthesia on cognitive function might be superimposed by effects of normal ageing, 5 a reduction in the incidence of postoperative cognitive dysfunction is of primary importance. Several attempts have been made to reduce cerebral damage during cardiac surgery, focusing on either the reduction in macro- and microembolism 3 or the optimization of the cerebral oxygen delivery/demand ratio. A non-invasive method for estimation of the cerebral oxygenation is the measurement of the regional cerebral oxygen saturation Sc O2 by near-infrared spectroscopy. 6 Deterioration of the Sc O2 during cardiac or non-cardiac surgery has been shown to be associated with postoperative cognitive dysfunction, focal cerebral deficits, 78 longer hospital stay, and increased postoperative morbidity. 9 11 Maintaining Sc O2 above a critical value by structured interventions has led to a lower postoperative morbidity and shorter hospital stay in cardiac 9 and non-cardiac patients. 12 A possibility of attenuating cerebral injury after cardiopulmonary bypass (CPB) could be anaesthetic pre- and postconditioning. The application of inhalation anaesthetics before and immediately after an ischaemic period has been shown to attenuate the ischaemia reperfusion injury of several organs. Neuronal anaesthetic preconditioning has been shown in vitro and in several animal studies, 13 14 but clinical data are scarce. 15 & The Author [2011]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Cognitive function after sevoflurane- vs propofol-based anaesthesia The present study was designed to determine whether patients after sevoflurane-based anaesthesia differ from patients after propofol-based anaesthesia in postoperative cognitive test results and the incidence of major organ dysfunction after on-pump cardiac surgery. To control for intraoperative changes in cerebral perfusion, the second objective was the impact of intraoperative cerebral desaturation on postoperative cognitive and major organ function. The third objective was whether the impact of cerebral desaturation on postoperative function was dependent on the anaesthesia regimen used. The question whether patients are able to complete cognitive testing at all as early as 2 days after cardiac surgery was another secondary objective. Methods Patients and study design The study is registered in the European Clinical Trials Database no. 2005-004928-39 and the ISRCTN Register (ISRCTN44821042). After approval of the local ethical committee and written informed consent, 128 patients undergoing elective cardiac surgery with CPB were enrolled in this prospective randomized study. Exclusion criteria were age below 18 yr, overt neurological diseases or dementia, significant stenosis of the carotid arteries, pregnancy, contraindications for sevoflurane, insufficient knowledge of the German language, and emergency indication. The randomization was performed after written informed consent was obtained. Multiple randomization lists stratified by age (,65 and 65 yr) and type of operative procedure (CABG, valve, or combined procedures) were used to provide equal groups. Justification of sample size The size and direction of a possible difference in cognitive function between different anaesthesia regimens cannot be determined on the basis of empirical data. According to Cohen, 16 an effect size of d¼0.50 is a low-to-median effect and should be clinically relevant. Assuming an a error of 5% and a b error of 20%, a sample size of N¼126 with n¼63 in each group is considered sufficient to identify relevant group differences. Intervention Oral premedication followed a standardized institutional protocol. All patients were equipped with a radial artery catheter, central venous catheter, and pulmonary artery catheter. Additionally, all patients were equipped with bi-hemispherical near-infrared spectroscopy sensors (INVOS Cerebral Oximeter 5100, Somanetics, Troy, MI, USA) (see below) and a bispectral index (BIS) probe on the forehead. Before CPB, all patients received 400 IU kg 21 heparin, achieving an activated clotting time above 500 s. Surgery was performed in moderate hypothermia using antegrade blood cardioplegia according to Buckberg and a-stat ph management. After weaning from CPB, protamine was applied as appropriate. Anaesthesia protocol for the volatile group (SEVO) Anaesthesia was induced with etomidate 0.2 0.3 mg kg 21 and sufentanil 1 mg kg 21 and maintained with remifentanil 0.2 0.25 mg kg 21 min 21 and sevoflurane 0.6 1 mean alveolar concentration aiming at a BIS of 40 50. Pancuronium bromide 0.07 0.1 mg kg 21 was used for relaxation. During the study period, there was no approval of the technical inspection authority to apply sevoflurane during CPB. During CPB, propofol 3 5 mg kg 21 h 21 was applied according to BIS (aim 40 50). After the release of the aortic cross-clamp, sevoflurane was continued and the propofol infusion stopped. During sternal closure, 1 g of metamizol and piritramid 15 mg were given i.v. For transport to the intensive care unit (ICU), the remifentanil infusion was stopped, and propofol 2 mg kg 21 h 21 was started and maintained until normothermia, haemodynamic stability, and sufficient spontaneous breathing were achieved. Piritramide and pethidin were applied for analgesia as required. An overview of the two different protocols is given in Table 1. Anaesthesia protocol for the i.v. group (PROP) Induction of anaesthesia and postoperative treatment in the i.v. group was identical to the volatile anaesthesia group, but anaesthesia was maintained with remifentanil 0.2 0.25 mg kg 21 min 21 and propofol 3 5 mg kg 21 h 21 as required to achieve a BIS of 40 50. Neurocognitive and psychometric tests All patients performed a set of psychometric and neurocognitive tests on the day before surgery and 2, 4, and 6 days after surgery. For all cognitive tests, parallel versions were used at random at the different measurements. The tests were selected on the basis of the Statement of Consensus on Assessment of Neurobehavioral Outcomes After Cardiac Surgery 17 and adapted to a preceding study of our group. 18 Most of the tests are taken from the Nuremberg Geriatric Table 1 Anaesthesia protocol in the study groups. CPB, cardiopulmonary bypass; MAC, mean alveolar concentration; BIS, bispectral index Induction of anaesthesia Maintenance of anaesthesia before and after CPB During CPB I.V. group (PROP) Volatile group (SEVO) Etomidate 0.2 0.3 mg kg 21, sufentanil 1 mg kg 21, pancuronium 0.07 0.1 mg kg 21 Remifentanil 0.2 0.25 mg kg 21 min 21, propofol 3 5 mg kg 21 h 21 achieving a BIS of 40 50 Remifentanil 0.2 0.25 mg kg 21 min 21, sevoflurane 0.6 1 MAC (age-adapted) achieving a BIS of 40 50 Propofol 3 5 mg kg 21 h 21 according to BIS (aim 40 50) 841

Schoen et al. Inventory (NGI). 19 All tests can be requested in printed form from the corresponding author. The German self-report inventory BSKE (Befindlichkeitsskalierung anhand von Kategorien und Eigenschaftswörtern) 20 was used to measure the positive and the negative mood. The Abbreviated Mental Test (AMT) 21 was used to assess cognitive impairment, dementia, or postoperative delirium. 22 Information processing was assessed by a German Trail- Making Test (TMT). 23 A modified version of the Stroop Test was used to measure directed attention and interference. 23 For the assessment of memory ability, patients were required to reproduce 10 previously read words without a time restriction [Word List (WL-N)]. All neurocognitive tests were performed by three trained investigators who instructed the patients in a standardized manner. The investigators were blinded to the anaesthesia protocol. If patients were still on respirator or haemodynamically unstable in the ICU or if they were disorientated or unresponsive, they were referred to as cannot perform test. Patients who had to stop testing due to fatigue or other discomfort were documented as did not complete test. These two groups were compiled in the group not able to be tested. Cerebral oxygen saturation measurement Regional cerebral oxygen saturation (Sc O2 ) measurement was started before induction of anaesthesia with the patient breathing room air, using this value as the baseline. Cerebral desaturation was defined as intraoperative Sc O2,50%. Measurement was continued throughout the surgical procedure and ICU stay until extubation. The INVOS monitor was covered to the attending physician throughout the data collection. The analysis of the collected data was performed after cessation of the study period. Markers of organ function and clinical outcome parameters On the day before surgery, after admission to the ICU, and days 2, 4, and 6 after surgery, blood tests included electrolytes, creatinine, liver enzymes, white and red blood cell counts, creatine kinase, troponin, S100-b, and C-reactive protein. Haemodynamic variables, temperature, and blood gas analyses were obtained several times intraoperatively at defined stages of cardiac surgery. The duration of surgery and CPB, aortic cross-clamp time, intraoperative need for cardiac assist devices, inotropic agents, vasopressors, blood components, and insulin were documented. The clinical outcome was compiled in the four major organ systems: brain, kidney, heart, and lung, as described in Table 2. One point was added for each developing complication, comprising a Major complication Score (MaCS) with a minimum of 0¼no complication and a maximum of 4¼complications in major organ systems. Table 2 Definition of major organ system complications. AMT, Abbreviated Mental Test; CI, cardiac index; ICU, intensive care unit Organ system complication Brain Kidney Heart Lung Statistical analysis If not stated otherwise, data are given as mean (SD). The Kolmogorov Smirnov test was used for identification of the normality of distribution. Univariate statistics were performed by Student s t-test for independent samples or the x 2 test (nominal data) as appropriate. Binary logistic regression was performed to determine predictors for the ability to perform cognitive tests 2 days after cardiac surgery. The analysis was performed with a 2 2-factorial plan with the randomized factor anaesthesia-protocol and the factor desaturation. Reasons for omitting the tests were documented and analysed separately. Missing data were then amended by the means of the respective groups. Group differences were analysed by analysis of variances for repeated measures with the factors, time, anaesthesia protocol, and desaturation (MANOVA). In the case of preoperative group differences, the preoperative test result was incorporated as the covariate (MANCOVA). Statistical significance was assessed at the 5% level. The statistical analysis was performed without a-adjustment; therefore, the results are considered mainly explanatory. Results Definition AMT,6 at any of the 3 postoperative measurements and/or focal cerebral deficit Increase in creatinine.0.3 mg dl 21 within 48 h CI,2.5 for more than 2 h during ICU stay and/or need for noradrenalin.0.5 mg h 21 and/or dobutamine.25 mg h 21 and/or milrinone.1.2 mg h 21 at any time during ICU stay Need for mechanical ventilation for more than 12 h and/or need for reintubation Recruitment A total of 153 patients fulfilled the inclusion criteria during the study period, but 25 patients refused to participate in the study, leaving 128 patients for randomization. Sixty-four patients were randomized for each anaesthesia regimen; 11 patients could not be analysed for various reasons (Fig. 1), leading to the final n¼60 patients in the propofol group and n¼57 in the sevoflurane group. In the propofol group, n¼14 (25%) patients had intraoperative cerebral desaturation and n¼42 (75%) patients had no desaturation. In n¼4 patients, the Sc O2 measurement could not be analysed correctly. In the sevoflurane group, n¼20 (37%) patients had desaturation and n¼34 (63%) had no desaturation; in n¼3 patients, the measurement could not be analysed. The incidence of desaturation was not significantly different in the anaesthesia groups (P¼0.172). 842

Cognitive function after sevoflurane- vs propofol-based anaesthesia Assessed for eligibility n=153 Refused to participate n=25 Randomized n=128 Allocated to propofol-based anaesthesia n=64 Received allocated intervention n=62 Unexpected off-pump procedure n=1 Accidentally wrong type of anaesthesia n=1 Allocated to sevofluranebased anaesthesia n=64 Received allocated intervention n=62 Operation cancelled n=2 Lost to follow up: operative re-exploration n=2 Lost to follow up: operative re-exploration n=3 patient died n=2 Analysed n=60 Analysed n=57 Fig 1 Flow chart of patient recruitment and reasons for exclusion from the analysis of patients after a sevoflurane- or propofol-based anaesthesia during on-pump cardiac surgery. Cognitive function Analysis of the baseline cognitive tests on the day before surgery showed a main effect of cerebral desaturation in the AMT and in the WL-N. Patients with subsequent intraoperative cerebral desaturation had lower scores in the AMT and remembered fewer words in the WL-N (Table 3). To adjust for the differences in the baseline measurements, an analysis of covariance with repeated measures was performed for the AMT and WL-N with the preoperative test result as the covariate (MANCOVA). Table 4 summarizes the results of the analysis of variance for repeated measures. A main effect for intraoperative desaturation was present in three of the four cognitive tests. Patients with intraoperative desaturation showed worse results in the AMT, Stroop Test, and TMT. A main effect was found for the anaesthesia regimen in all four cognitive tests. Patients with sevofluranebased anaesthesia showed better results in all tests. Further, a main effect for the time of measurement was found in all cognitive tests. Worst results were found on day 2 after surgery. Interactions were found between the anaesthesia regimen and desaturation in all four tests; patients with desaturation under propofol showed worse results than those with desaturation under sevoflurane. The interaction between anaesthesia and time reached significance only in one test, whereas the interaction between time and desaturation was significant in three tests; patients with and without desaturation showed time course differences in the linear trend. The interaction between time, anaesthesia, and desaturation was significant in all four tests. Figure 2A D illustrates the results. Organ function There were no relevant differences in patient characteristic data, baseline laboratory results, and surgery-related data (Table 3). There was a significant main effect regarding cerebral desaturation in the baseline Sc O2. Patients who subsequently developed cerebral desaturation had lower baseline Sc O2 immediately before surgery (Table 3). The analysis of variance for repeated measures showed no relevant main effects or interaction between the anaesthesia regimen and intraoperative cerebral desaturation in creatinine, troponin I, S100-b, or C-reactive protein (Fig. 3A D). 843

Schoen et al. Table 3 Patient characteristics, baseline cognitive tests, surgery-related data, and baseline laboratory results. If not mentioned otherwise, data are shown as mean (SD) (range) for age, or mean (SD). Sc O2, regional cerebral oxygen saturation; AMT, Abbreviated Mental Test; Stroop, Stroop Test; TMT, Trail-Making Test; WL-N, Word List; CRP, C-reactive protein; Trop I, troponin I; CK, creatine kinase; CABG, coronary artery bypass grafting; Valve, valve replacement or repair; CPB, cardiopulmonary bypass; AoX, aortic cross-clamp Propofol group Sevoflurane group Anova No desaturation (n542) Age (yr) 62.7 (8.5) (43 76) Desaturation (n514) 67.6 (8.5) (53 79) No desaturation (n534) 65.1 (7.3) (44 76) Desaturation (n520) 64.2 (8.7) (52 82) Main effect desaturation Main effect anaesthesia 0.380 0.104 0.027 EuroScore 3.2 (2.5) 4.4 (3.0) 4.0 (3.1) 4.3 (3.5) 0.241 0.579 0.518 Gender Male, n¼34 Male, n¼9 Male, n¼27 Male, n¼11 (81%) (64.3%) (79.4%) (55%) Female, n¼8 (19%) Female, n¼5 (35.7%) Missing n¼1 Female, n¼7 (20.6%) Female, n¼9 (45%) Baseline Sc O2 66.0 (4.6) 57.6 (8.6) 66.2 (6.0) 60.2 (7.8) 0.000 0.073 0.107 Baseline AMT 8.7 (0.9) 8.0 (1.2) 8.8 (0.9) 8.4 (0.9) 0.007 0.178 0.470 (score) Baseline Stroop 51.4 (15.9) 58.3 (21.0) 51.7 (16.6) 45.55 (7.3) 0.904 0.062 0.049 (s) Baseline TMT (s) 33.9 (14.9) 44.07 (26.5) 34.2 (15.0) 36.5 (20.0) 0.294 0.070 0.058 Baseline WL-N 5.6 (1.1) 4.7 (1.5) 5.3 (1.5) 4.9 (1.0) 0.019 0.876 0.393 (words) Creatinine 102.5 (27.9) 91.1 (37.3) 91.9 (28.4) 93.7 (28.7) 0.447 0.528 0.293 clearance (ml min 21 ) CRP (mg dl 21 ) 7.6 (7.5) 16.0 (28.9) 8.2 (9.1) 14.1 (16.6) 0.019 0.826 0.686 Trop I (mg 0.04 (0.16) 0.06 (0.16) 0.02 (0.06) 0.15 (0.49) 0.166 0.608 0.373 litre 21 ) CK (mg litre 21 ) 82.1 (45.3) 102.1 (79.8) 102.4 (67.9) 83.8 (48.1) 0.954 0.938 0.121 CK-MB (mg 10.0 (5.4) 11.5 (5.4) 9.4 (3.1) 12.1 (9.3) 0.084 0.988 0.604 litre 21 ) S100-b (mg 0.06 (0.03) 0.07 (0.02) 0.06 (0.04) 0.08 (0.05) 0.286 0.670 0.482 litre 21 ) Type of surgery CABG, n¼29 (69%) CABG, n¼5 (35.7%) CABG, n¼21 (61.8%) CABG, n¼14 (70%) Valve, n¼8 (19%) Combination, n¼5 (11.9%) Duration of operation (min) Duration of CPB (min) Duration of AoX (min) Valve, n¼3 (21.4%) Combination, n¼6 (42.9%) Valve, n¼5 (14.7%) Combination, n¼8 (23.5%) Valve, n¼2 (10%) Combination, n¼4 (20%) 226 (52) 244 (83) 251 (53) 245 (66) 0.650 0.310 0.329 88 (20) 109 (27) 114 (40) 105 (38) 0.380 0.104 0.027 73 18 87 (28) 94 (37) 87 (35) 0.593 0.077 0.103 Interaction There were no significant group differences in the MaCS (Table 2). However, more patients with desaturation under propofol had one or more complications (P¼0.018; Fig. 4). There were no group differences regarding ventilation time or length of stay in the ICU, in the high dependency unit, or the hospital (data not shown). Ability to perform tests Two days after surgery, 62 patients (53.0%) completed the cognitive tests. Twenty (17.1%) patients refused to do the tests, and in seven cases (6.0%), the data were incomplete or missing for unknown reasons. Twenty-eight (23.9%) patients were labelled not able to be tested (see the Methods section). The characteristics of patients who were not able to be tested 2 days after cardiac surgery are shown in Table 5. These patients were older, had more often a lower educational level, and suffered from diabetes mellitus compared with the patients who did complete the tests. Patients who had refused testing were excluded from this analysis. More patients who were not able to be tested had received a propofol-based anaesthesia (P¼0.046). The type and duration of operation, CPB time, and aortic 844

Cognitive function after sevoflurane- vs propofol-based anaesthesia Table 4 Analysis of variance with repeated measures for the cognitive tests. *Analysis of covariance. AMT, Abbreviated Mental Test; WL-N, Word List AMT* Stroop Test Trail-Making Test WL-N* F P F P F P F P Main effect Desaturation 6.37 0.013 5.21 0.024 5.53 0.020 0.27 0.602 Anaesthesia 24.47 0.000 12.16 0.001 18.45 0.000 15.09 0.000 Time 4.18 0.017 12.71 0.000 41.51 0.000 11.25 0.000 Interaction Anaesthesia desaturation 12.36 0.001 15.18 0.000 19.20 0.000 9.78 0.002 Time anaesthesia 2.12 0.122 2.56 0.069 3.00 0.050 6.92 0.001 Time desaturation 0.31 0.734 4.06 0.013 3.27 0.038 8.02 0.000 Time anaesthesia desaturation 3.78 0.025 3.06 0.040 10.50 0.000 6.11 0.003 cross-clamp were comparable and the groups did not differ in any intraoperative haemodynamic parameter or requirements of vasopressors or inotropic drugs (data not shown). Patients who were not able to be tested had lower baseline Sc O2 (P¼0.008) and had more often suffered intraoperative cerebral desaturation with Sc O2,50% (P¼0.025) or Sc O2,45% (P¼0.001). Both groups had comparable ventilation times, but patients who could not complete the tests had a longer ICU stay. The parameters of age, educational level, baseline Sc O2, intraoperative cerebral desaturation below Sc O2,50%, and the anaesthesia regimen were inserted in a binary logistic regression model. It identified age above 65 yr, low educational level, and desaturation with Sc O2,50% as independent predictors of the inability to accomplish cognitive tests 2 days after surgery (Table 6). Discussion In the present study, patients who developed intraoperative cerebral desaturation showed worse early postoperative cognitive test results than those without cerebral desaturation. Cerebral desaturation has been defined as intraoperative regional cerebral oxygen saturation below 50% of the absolute value measured with near-infrared spectroscopy. However, patients who subsequently developed cerebral desaturation did not only start with lower baseline cerebral oxygen saturation but also showed worse baseline test results in two of the four cognitive tests used. The differences in baseline cognitive status could be eliminated statistically for the analysis of the postoperative tests results, but the fact that patients suffering desaturation had lower baseline cognitive performance by chance allows investigation of a known cerebral insult (subsequent desaturation events) in the setting of decreased cognitive performance pre-insult. Stern 24 has introduced the model of cognitive reserve. It refers to the individual differences in the cognitive processes that allow some people to cope better with brain pathology than others and has been related to intelligence, age, educational level, and social activity, many of which were significant predictors of postoperative cognitive performance in our study (Table 5). The association of desaturation events with subsequent poor cognitive test results might therefore be understood as a physiological insult which has its greatest effect in patients with low cognitive reserve. Intraoperative cerebral desaturation measured with nearinfrared spectroscopy has been shown to be associated with postoperative central nervous system dysfunction. The definition of which threshold is regarded as potentially harmful desaturation varies. Yao and colleagues 7 described a cerebral desaturation below 40% of the absolute values to be predictive of postoperative cognitive impairment. Slater and colleagues 10 showed that a desaturation of more than 3000 s % below 50% of the absolute values was associated with cognitive decline and prolonged hospital stay. In the present study, only a total of nine patients in both groups showed desaturations below 40% of the absolute value. We therefore decided to choose the cut-off value of Sc O2,50% described by Slater and colleagues. In our cohort, the mean area under the curve below 50% was 2295 s % for both hemispheres added. The declines in cerebral oxygenation observed in the present study must therefore be regarded as moderate desaturation. The differences in cognitive test results in the present study between patients with and without cerebral desaturation support the tighter threshold of Sc O2. The first postoperative cognitive tests were conducted 2 days after surgery. According to Silbert and colleagues, 25 cognitive testing should be possible as early as 18 h after cardiac surgery with a completion rate above 60%. However, in the present study, only 53% completed the cognitive tests on day 2 after surgery. We did not discriminate the reasons for not being able to conduct the tests. This could be ongoing sedation in the ICU or inability to communicate with the investigators due to disorientation. Interestingly, the patients who could or could not conduct the cognitive testing had a comparable preoperative EuroScore, type of surgery, and duration of surgery and CPB and comparable intraoperative haemodynamics but differed in the baseline Sc O2. Patients who could not completely accomplish 845

Schoen et al. A 9.5 B 100 Propofol without desaturation Propofol with desaturation Sevoflurane without desaturation Sevoflurane with desaturation 9.0 90 AMT scores 8.5 8.0 7.5 7.0 6.5 Stroop Test (s) 80 70 60 50 6.0 Before surgery POD 2 POD 4 POD 6 40 Before surgery POD 2 POD 4 POD 6 C 120 D 6.0 100 5.5 5.0 TMT (s) 80 60 WL-N Words 4.5 4.0 3.5 40 3.0 20 Before surgery POD 2 POD 4 POD 6 2.5 Before surgery POD 2 POD 4 POD 6 Fig 2 Results of the cognitive tests in patients after a sevoflurane- or propofol-based anaesthesia during on-pump cardiac surgery in patients with and without cerebral desaturation. (A) AMT. Means and standard error of the mean. (B) Stroop Test. Means and standard error of the mean. (C) TMT. Means and standard error of the mean. (D) Word List. Means and standard error of the mean. AMT, Abbreviated Mental Test; POD, postoperative day; TMT, trail-making-test; WL-N, Word List. the cognitive tests had a longer ICU stay even though the group differences in ventilation time did not reach statistical significance. Old age, low educational level, and intraoperative cerebral desaturation were identified as independent risk factors for not being able to perform a cognitive test 2 days after surgery. The predictive power of age, educational level, and low Sc O2 on the ability to take part in cognitive testing supports the theory that these factors characterize a group of patients with low cognitive reserve. Even though the distinct effects in animal and in vitro studies are suggestive of a clinical significance of anaesthetic preconditioning of the brain, 13 clinical data on neuroprotective effects of volatile anaesthetics are scarce. 15 26 A small retrospective study showed no effects of sevoflurane on long-term cognitive function after CABG; 26 another small prospective study showed no benefit of isoflurane compared with propofol regarding cognitive function. The present study shows prospectively that patients after sevoflurane-based anaesthesia showed better results than patients after propofol-based anaesthesia in four independent cognitive tests, despite a comparable pre- and intraoperative risk profile. This finding suggests neuroprotective properties of sevoflurane. As the differences in cognitive performance did not affect ventilation times, or length of stay in the ICU or the hospital, they could be interpreted as a short-term effect on recovery after anaesthesia 27 with questionable clinical relevance. But early cognitive disturbances have been shown to be associated with a higher risk of long-term cognitive impairment with severe impact on the quality of life. 428 846

Cognitive function after sevoflurane- vs propofol-based anaesthesia A 100 B 0.8 Propofol without desaturation Propofol with desaturation Sevoflurane without desaturation Sevoflurane with desaturation Creatinine (mmol litre 1 ) 95 90 85 80 75 70 65 Troponin T (mg litre 1 ) 0.6 0.4 0.2 0.0 60 Before surgery POD 1 POD 2 POD 4 POD 6 Before surgery POD 1 POD 2 POD 4 POD 6 C 0.22 D 250 S100-b (mg litre 1 ) 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 C-reactive protein (mg dl 1 ) 200 150 100 50 0 0.04 Before surgery POD 1 POD 2 POD 4 POD 6 Before surgery POD 1 POD 2 POD 4 POD 6 Fig 3 Course of markers of organ function in patients after a sevoflurane- or propofol-based anaesthesia during on-pump cardiac surgery in patients with and without cerebral desaturation. (A) Creatinine. Means and standard error of the mean. (B) Troponin T. Means and standard error of the mean. (C) S100-b. Means and standard error of the mean. (D) C-reactive protein. Means and standard error of the mean. The most interesting hypothesis of the present study was the question whether a sevoflurane-based in comparison with a propofol-based anaesthesia regimen might attenuate the effects of cerebral desaturation on cognitive outcome after on-pump cardiac surgery. And indeed we found a significant interaction between the anaesthesia regimen and cerebral desaturation in all cognitive tests. This result has to be interpreted carefully, as the groups showed preoperative interaction with regard to age, with patients receiving propofol and suffering desaturation being older. This means that the group might have lower cognitive reserve and therefore be more vulnerable to cerebral damage. The great number of more complex surgical procedures in this group indicates that patients receiving propofol and having desaturation might sustain greater surgical trauma and might therefore be prone to greater cerebral damage. However, the longer duration of CPB in the sevoflurane group likewise means cerebral hazard. Keeping in mind these limitations, this finding might be carefully interpreted as a preconditioning effect of sevoflurane on cerebral tissue. In an animal study by Zhu and colleagues, 29 sevoflurane has recently been shown to precondition the brain to inflammatory changes and ischaemia. 30 The results of the present study are at least suggestive of a clinical relevance of the preconditioning effects of sevoflurane on neuronal tissue. Whether these effects are based on reduction in inflammatory response or ischaemiaprotective mechanisms cannot be differentiated by the present data. However, this assumption has to be verified in a more homogeneous population with a completely comparable preoperative condition and surgical trauma. 847

Schoen et al. 100 80 80 P=0.018 70 60 % 50 40 30 20 10 0 Propofol without desaturation Propofol with desaturation Sevoflurane without desaturation Sevoflurane with desaturation No major complications One or more major complications Fig 4 Proportion of patients with no versus one or more major organ complications in the postoperative course. We found no impact of the anaesthetic regimen on major outcome criteria, like major organ dysfunction, ventilation time, or length of stay in the ICU or the hospital, or laboratory results. However, these findings are in line with recent observations of De Hert and colleagues, 31 who could not show any differences in troponin T in an unselected group of patients after on-pump cardiac surgery between i.v. and volatile anaesthesia. Interestingly, we did find a disproportionately high number of patients with one or more major organ complications in the group with cerebral desaturation under propofol-based anaesthesia. Previously, our own group could show an association between preoperative cerebral oxygenation and postoperative morbidity and mortality. 32 A low cerebral saturation can be interpreted either as a sign of temporarily inadequate regional or global haemodynamic condition. An association between intraoperative cerebral desaturation and postoperative morbidity would therefore make sense even though the documented intraoperative parameters did not indicate any group differences in the global haemodynamic condition. The alternative interpretation would be the theory of Sc O2 being an indicator of underlying cerebral small vessel disease that might reduce cognitive reserve. The major limitation of the present study is the group size. Despite the power analysis to calculate the sample size, the effective group size was much smaller due to the inability of patients to perform cognitive tests and a large number of missing values having to be amended. This certainly weakens the information given by the present study. On the other hand, the differences in the number of dropouts in the study groups have to be interpreted as results of the intervention. Another limitation is the lack of long-term outcome testing. The effect of early cognitive function on long-term cognitive performance was not subject to the present study. But it has been shown that early cognitive deficits can be associated with long-term functional decline and impaired quality of life. 4 A major limitation in this context is that we had no standardized test on delirium as a global cerebral deficit. The CAM-ICU would have been a suitable tool to detect hyperand hypoactive delirium. 33 We only used the AMT, which can be interpreted as a global cognitive test 22 but is not specific for delirium. So, we cannot state any differences between sevoflurane and propofol regarding the incidence of postoperative delirium. The data seem to be weakened by the fact that patients of both groups received propofol during CPB and after operation for sedation in the ICU. This interference in the treatment could not be avoided due to technical reasons. The characteristic of anaesthetic preconditioning is that the protective effect outlasts the duration of application of the volatile anaesthetic. 34 The fact that we applied propofol during the potentially harmful period of CPB, therefore, does not contradict the assumption of the preconditioning effects of sevoflurane. In conclusion, the present study shows that an intraoperative cerebral desaturation is associated with a worse early cognitive outcome after on-pump cardiac surgery, 848

Cognitive function after sevoflurane- vs propofol-based anaesthesia Table 5 Characteristics of patients who could or could not perform the cognitive testing 2 days after cardiac surgery. If not mentioned otherwise, data are shown as mean (SD) (range) for age, or mean (SD). *Student s t-test for independent samples; x 2 test. CABG, coronary artery bypass grafting; Valve, valve replacement or repair; CPB, cardiopulmonary bypass; rsc O2, regional cerebral oxygen saturation; ICU, intensive care unit. Educational level: 0, no graduation; 1, compulsory school; 2, secondary school; 3, vocational diploma; 4, general qualification for university entrance; 5, technical college; 6, university degree Patients who did perform tests, n562 (53.0%) Patients who could not completely perform tests, n528 (23.9%) Age (yr) 62.2 (8.44) (42 82) 66.8 (6.60) (53 78) 0.016* Height (m) 173.1 (8.38) 169.7 (9.11) 0.110* Weight (kg) 80.9 (4.12) 85.4 (14.30) 0.178* Gender Male, n¼48 (77.4%) Male, n¼18 (64.3%) 0.192 Female, n¼14 (22.6%) Female, n¼10 (35.7%) EuroScore 3.6 (2.93) 4.3 (2.72) 0.251* Educational level 2.5 (1.87) 1.8 (1.51) 0.029* Diabetes n¼7 (11.3%) n¼8 (28.6%) 0.042 Chronic kidney disease n¼27 (43.5%) n¼15 (53.6%) 0.378 Anaesthesia-protocol Propofol n¼28 (45.2%) Propofol: n¼19 (67.9%) 0.046 Sevoflurane n¼34 (54.8%) Sevoflurane: n¼9 (32.1%) Operation CABG: n¼34 (54.8%) CABG: n¼17 (60.7%) 0.595 Valve: n¼12 (19.4%) Valve: n¼3 (10.7%) Combination: n¼16 (25.8%) Combination: n¼8 (28.6%) Duration of operation (min) 234.4 (53.04) 253.9 (75.17) 0.235* Duration of CPB (min) 102.1 (32.04) 106.2 (31.79) 0.553* Duration of aortic cross-clamp 83.4 (29.06) 87.4 (30.60) 0.604* (min) Mean baseline rsc O2 (%) 65.5 (5.28) 60.7 (7.86) 0.008* Minimal rsc O2 below 50% n¼14 (23.3%) n¼12 (48.0) 0.025 Minimal rsc O2 below 45% n¼4 (6.7%) n¼9 (36.0%) 0.001 Ventilation time (h) 7.6 (3.3) 9.1 (7.5) 0.331* ICU length of stay (h) 23.4 (10.7) 38.4 (29.1) 0.013* Hospital length of stay (days) 7.9 (3.8) 7.9 (3.3) 0.995* P-value Table 6 Binary logistic regression (Model: Simultaneous, including all variables) analysis with preoperative predictors on the ability to perform cognitive tests 2 days after surgery Parameter P-value Specification Prevalence could not perform tests (%) Odds ratio 95% CI Age 0.026 65 yr 23.4 Reference.65 yr 42.5 4.70 1.20 18.36 Educational level 0.010 More than compulsory school 18.4 Reference Compulsory school or no graduation 42.9 7.40 1.62 33.91 Baseline rsc O2 0.438.62% 23.8 Reference 62% 40.0 0.481 0.08 3.06 Minimal rsc O2 0.038 50% 22.0 Reference,50% 46.2 5.83 1.10 30.79 Anaesthesia 0.052 Propofol 40.4 Reference Sevoflurane 20.9 0.27 0.07 1.01 and this effect may be exaggerated in patients who have evidence of diminished cognitive reserve before operation. Further, the possibility exists that sevoflurane-based volatile anaesthesia regimen might be associated with better cognitive function compared with a propofol-based anaesthesia regimen. Further investigation is needed to confirm the impact of sevoflurane on global cerebral deficits and to identify possible subgroups of patients who particularly benefit from volatile anaesthesia regarding cognitive function. Conflict of interest J.S. and M.H. received honoraria for lectures from Covidien Germany. 849

Schoen et al. Funding The study has been funded by a scientific grant by Abbott, Wiesbaden, Germany, and technical support by Covidien Germany. References 1 Arrowsmith J, Grocott H, Reves J, Newman M. Central nervous system complications of cardiac surgery. Br J Anaesth 2000; 84: 378 93 2 van Dijk D, Keizer AMA, Diephuis JC, Durand C, Vos LJ, Hijman R. Neurocognitive dysfunction after coronary artery bypass surgery: a systematic review. J Thorac Cardiovasc Surg 2000; 120: 632 9 3 Hogue CW Jr, Palin CA, Arrowsmith JE. Cardiopulmonary bypass management and neurologic outcomes: an evidence-based appraisal of current practices. Anesth Analg 2006; 103: 21 37 4 Newman MF, Kirchner JL, Phillips-Bute B, et al. Longitudinal assessment of neurocognitive function after coronary-artery bypass surgery. N Engl J Med 2001; 344: 395 402 5 Selnes OA, Pham L, Zeger S, McKhann GM. Defining cognitive change after CABG: decline versus normal variability. Ann Thorac Surg 2006; 82: 388 90 6 Edmonds HL Jr, Ganzel BL, Austin EH 3rd. Cerebral oximetry for cardiac and vascular surgery. Semin Cardiothorac Vasc Anesth 2004; 8: 147 66 7 Yao FS, Tseng CC, Ho CY, Levin SK, Illner P. Cerebral oxygen desaturation is associated with early postoperative neuropsychological dysfunction in patients undergoing cardiac surgery. J Cardiothorac Vasc Anesth 2004; 18: 552 8 8 Goldman S, Sutter F, Ferdinand F, Trace C. Optimizing intraoperative cerebral oxygen delivery using noninvasive cerebral oximetry decreases the incidence of stroke for cardiac surgical patients. Heart Surg Forum 2004; 7: E376 81 9 Murkin JM, Adams SJ, Novick RJ, et al. Monitoring brain oxygen saturation during coronary bypass surgery: a randomized, prospective study. Anesth Analg 2007; 104: 51 8 10 Slater JP, Guarino T, Stack J, et al. Cerebral oxygen desaturation predicts cognitive decline and longer hospital stay after cardiac surgery. Ann Thorac Surg 2009; 87: 36 45 11 Schoen J, Serien V, Hanke T, et al. Cerebral oxygen saturation monitoring in on-pump cardiac surgery: a 1-year experience. Appl Cardiopulm Pathophysiol 2009; 13: 243 52 12 Casati A, Fanelli G, Pietropaoli P, et al. Monitoring cerebral oxygen saturation in elderly patients undergoing general abdominal surgery: a prospective cohort study. Eur J Anaesthesiol 2007; 24: 59 65 13 Kitano H, Kirsch JR, Hurn PD, Murphy SJ. Inhalational anesthetics as neuroprotectants or chemical preconditioning agents in ischemic brain. J Cereb Blood Flow Metab 2007; 27: 1108 28 14 Kapinya KJ, Lowl D, Futterer C, et al. Tolerance against ischemic neuronal injury can be induced by volatile anesthetics and is inducible NO synthase dependent. Stroke 2002; 33: 1889 98 15 Kanbak M, Saricaoglu F, Avci A, Ocal T, Koray Z, Aypar U. Propofol offers no advantage over isoflurane anesthesia for cerebral protection during cardiopulmonary bypass: a preliminary study of S-100{beta} protein levels: [L anesthesie au propofol, compare a l isoflurane, n a pas d avantage pour la protection cerebrale pendant la circulation extracorporelle: une etude preliminaire des niveaux de proteines S-100{beta}]. Can J Anesth 2004; 51: 712 7 16 Cohen J, editor. Statistical Power Analysis for Behavioral Sciences, 2nd Edn. Hillsdale, NJ: Lawrence Erlbaum, 1988 17 Murkin JM, Newman SP, Stump DA, Blumenthal JA. Statement of consensus on assessment of neurobehavioral outcomes after cardiac surgery. Ann Thorac Surg 1995; 59: 1289 95 18 Gameiro M, Eichler W, Schwandner O, et al. Patient mood and neuropsychological outcome after laparoscopic and conventional colectomy. Surg Innov 2008; 15: 171 8 19 Aufdembrinke B, Ott H, Rohloff H. Measures of memory and information processing in elderly volunteers. Pschopharmacol Ser 1988; 6: 48 64 20 Hueppe M, Uhlig T, Vogelsang H, Schmucker P. Personality traits, coping styles, and mood in patients awaiting lumbar-disc surgery. J Clin Psychol 2000; 56: 119 30 21 Hodkinson HM. Evaluation of a mental test score for assessment of mental impairment in the elderly. Age Ageing 1972; 1: 233 8 22 Ni Chonchubhair A, Valacio R, Kelly J, O Keefe S. Use of the abbreviated mental test to detect postoperative delirium in elderly people. Br J Anaesth 1995; 75: 481 2 23 Oswald WD, Fleischmann UM. Das Nürnberger Altersinventar (NAI). Göttingen: Hogrefe, 1997 24 Stern Y. Cognitive reserve. Neuropsychologia 2009; 47: 2015 28 25 Silbert BS, Scott DA, Doyle TJ, et al. Neuropsychologic testing within 18 hours after cardiac surgery. J Cardiothorac Vasc Anesth 2001; 15: 20 4 26 Kadoi Y, Goto F. Sevoflurane anesthesia did not affect postoperative cognitive dysfunction in patients undergoing coronary artery bypass graft surgery. J Anesth 2007; 21: 330 5 27 Gupta A, Stierer T, Zuckerman R, Sakima N, Parker SD, Fleisher LA. Comparison of recovery profile after ambulatory anesthesia with propofol, isoflurane, sevoflurane and desflurane: a systematic review. Anesth Analg 2004; 98: 632 41, table of contents 28 Newman MF, Grocott HP, Mathew JP, et al. Report of the substudy assessing the impact of neurocognitive function on quality of life 5 years after cardiac surgery. Stroke 2001; 32: 2874 81 29 Zhu J, Jiang X, Shi E, Ma H, Wang J. Sevoflurane preconditioning reverses impairment of hippocampal long-term potentiation induced by myocardial ischaemia-reperfusion injury. Eur J Anaesthesiol 2009; 26: 961 8 30 Sanders RD, Maze M. Neuroinflammation and postoperative cognitive dysfunction: can anaesthesia be therapeutic? Eur J Anaesthesiol 2010; 27: 3 5 31 De Hert S, Vlasselaers D, Barbe R, et al. A comparison of volatile and non volatile agents for cardioprotection during on-pump coronary surgery. Anaesthesia 2009; 64: 953 60 32 Heringlake M, Garbers C, Kaebler J-K, et al. Preoperative cerebral oxygen saturation and clinical outcomes in cardiac surgery. Anesthesiology 2011; 114: 58 69 33 Klugkist M, Sedemund-Adib B, Schmidtke C, Schmucker P, Sievers HH, Huppe M. Confusion Assessment Method for the Intensive Care Unit (CAM-ICU): diagnosis of postoperative delirium in cardiac surgery. Anaesthesist 2008; 57: 464 74 34 Kehl F, Payne RS, Roewer N, Schurr A. Sevoflurane-induced preconditioning of rat brain in vitro and the role of KATP channels. Brain Res 2004; 1021: 76 81 850