| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
During the early postoperative period after coronary artery bypass grafting (CABG) surgery, many patients experience jugular bulb oxyhemoglobin desaturation (SjO2 < 50%). We sought to determine whether SjO2 during cardiopulmonary bypass and the early postoperative period influenced long-term cognitive performance after CABG surgery. One hundred two patients completed a battery of cognitive tests the day before and 3 mo after CABG surgery. A General Cognitive Score was generated from these tests as an overall measure of cognitive function. Intraoperatively, SjO2 was determined by intermittent blood sampling, and postoperatively, the area under the curve of SjO2 < 50% and time was calculated from continuous reflectance oximetry. No significant correlations between cognitive performance and either intra- or postoperative SjO2 were found. Preoperative cognitive performance was the main determinant of cognition at 3 mo (r2 = 0.83, P < 0.001), and palpable atheroma of the ascending aorta made a small, but significant, contribution to a decline in cognition (r2 = 0.018, P = 0.001).
Implications: Postoperative cerebral hypoperfusion, indicated by oxyhemoglobin desaturation < 50%, occurs in many patients after coronary artery bypass graft surgery. However, this measurement of hypoperfusion does not appear to influence long-term cognition after coronary artery bypass graft surgery.
Cognitive decline may be detected in up to 80% of patients one week after coronary artery bypass grafting (CABG) surgery (1). Although some of this decrement is caused by permanent "brain injury," some patients recover with time (1). Probably of greater socioeconomic importance is assessment of cognition three months after surgery, which detects long-term decline and occurs in almost one third of patients (1,2). Memory, attention, and psychomotor speed are the domains of cognition that are most commonly affected. However, no study has looked specifically at overall cognitive ability after CABG surgery (1).
The etiology of this cerebral injury is multifactorial, and thought to include macro- and microscopic emboli, inflammation, and nitric oxide-mediated glutamate excitotoxicity (3,4). Genetic predisposition in the form of apolipoprotein E- Jugular bulb oxyhemoglobin desaturation (SjO2 < 50%) and widened AVDO2 (Appendix 1) are estimates of cerebral hypoperfusion that are both significantly, although weakly, associated with short-term cognitive decline (6). The brain may not only be susceptible to hypoperfusion during CPB but also in the immediate postoperative period when cerebral edema can occur (8). Previously, we have reported that many patients experience frequent and prolonged episodes of cerebral hypoperfusion as estimated by SjO2 < 50% in the early postoperative period after CABG surgery (9). In addition, we have found evidence of cerebral hypoperfusion using an anerobic estimate, the cerebral arteriovenous difference in lactate content (AVDL) (Appendix 1), and a combination estimate, the lactate-oxygen index (LOI) (Appendix 1) (7,9). SjO2, AVDL, and LOI are related to factors that influence cerebral blood flow (CBF) including PaCO2, arterial saturation (SaO2) and arterial hemoglobin concentration (Hba) (7,9,12,13). Therefore, deranged levels of these factors may cause cerebral hypoperfusion. However, AVDL, AVDO2, and LOI are mathematically coupled and not statistically independent from Hba and SaO2 (14,15). SjO2 and jugular bulb lactate concentrations (Lj) are robust and independent estimates of adequate cerebral perfusion. The aim of this study was to determine whether individual differences in the duration of SjO2 < 50%, SjO2 during CPB, and Lj are associated with individual differences in long-term cognitive outcome after CABG surgery. A subsidiary aim was to investigate whether factors that may influence CBF, PaCO2, SaO2, Hba, and mean arterial pressure (MAP) affect cognitive outcome.
A number of illnesses and their associated treatments/interventions are suspected of causing cognitive deficits. Among these are cardiac arrest (16), type II diabetes (17), hypoglycemia (18), liver failure (19), and cardiac surgery. In the latter case, a large body of evidence documents cognitive decrements after CABG surgery (20). Typically, cognitive decrements were demonstrated by using case control studies. However, to determine factors causing these decrements, another research design is necessary. A hypothetical causative factor must be identified, measured, and then correlated with the degree of cognitive decrement. For example, it has long been suspected that people with type I diabetes perform less well on cognitive tests than matched controls as demonstrated by using case control studies (21). More recent research has identified repeated episodes of severe hypoglycemia as one possible cause of cognitive under-performance in the diabetic group. Studies that test this hypothesis measure individual differences in the amount of severe, repeated hypoglycemia and correlate these with individual differences in cognitive change (18,2123). The general approach, then, in investigating hypotheses about the etiology of cognitive decrements (as opposed to demonstrating the existence of cognitive decrements) is: 1) to identify and measure individual differences in a nominated causal factor, and 2) correlate these with individual differences in measured cognitive change/decrement. The simple hypothesis is that the greater the insult/causative factor, the more severe the cognitive decrement. This design and analytic approach studies the association between putative cause and effect within the illness group, and does not require a control group. We use a similar approach to design and analysis in the present study. We measured individual differences in a number of factors associated with cerebral hypoperfusion and hypothesize that those patients who experience the increased amounts of these putative brain insults will have increased cognitive decrements. Because putative insult factors and cognitive decrements are variables on a continuum, the most efficient analysis is a method that preserves these continua rather than one that imposes an arbitrary threshold, which merely produces a spurious binary division of subjects into "cognitively damaged" and "noncognitively damaged" people. In addition, cognitive change is not most effectively assessed by using the arithmetic difference between pre- and postoperative scores. More important is identification of the contributing factors to variance in postoperative cognitive test scores. Clearly, the first contributing factor will be the preoperative test score, which can be "removed," more effectively than using change scores between baseline and outcome, by entering it as the first independent variable in a multiple regression equation (15). This is an efficient method of construing the difference between pre- and postoperative test scores. Effectively, the remaining variance delivers a measure of the individual differences in cognitive change postoperatively. It provides "target," outcome variance to which we can aim our putative causal factors. Thus, variables that enter the multiple regression model (in which the postoperative cognitive test score is the dependent variable) after the preoperative test score has been entered have thereby, in effect, accounted for some variance in the cognitive change between the pre- and postoperative states. It is necessary to explore different domains of cognitive ability to discover whether CABG surgery affects some functions and not others (22). However, it is well established that all cognitive domains share much common variance; that is, all cognitive tests correlate positively (24,25). It is necessary, therefore, to explore whether the effect of a brain insult on a given cognitive domain operates via this "general" cognitive variance. For example, the effect of age on various specific cognitive domains is entirely a result of the effect of age on general mental ability (26). Therefore, cognitive ability is best examined as a hierarchy with general ability at the pinnacle and separable (correlated) domains underneath. An example of this approach is our reanalysis (22) of the United States multicentre Diabetes Control and Complications Trial which has been deemed a more informative analysis than the original analysis (23). This hierarchical approach was adopted for the cognitive tests used in the present study, as is explained in the statistics section. The study was approved by the local ethics committee. Exclusion criteria included a history of stroke, transient ischemic attack, insulin-dependent diabetes mellitus, depressive illness, alcohol abuse, and long-term sedative or psychiatric medication. Written, informed consent was obtained from patients the day before surgery. Between January 1998 and July 1999, 178 patients scheduled for CABG surgery were approached and 135 (117 men, 18 women) agreed to participate. A battery of 11 cognitive tests, including those recommended by the Statement of Consensus 1995 (27), were administered the day before and 3 mo after CABG surgery. Where available, parallel forms of a test were administered at follow-up. Mood was assessed by the Hospital Anxiety and Depression Scale (28). A single investigator (SY) performed the cognitive assessments. Domains of cognitive function assessed and the tests used were as follows:
Tests were administered in the following order: ReyAVLT trials 15, Trail-Making Test A and B, WAIS-R Block Design, National Adult Learning Test, WAIS-R Object Assembly, Controlled Oral Word Association Test, ReyAVLT trial 6 and recognition, WAIS-R Digit Symbol, Paced Auditory Serial Addition Task (4-s then 2-s trial), Grooved Pegboard (dominant and nondominant), and WAIS-R Picture Completion. Although our primary objective was to avoid categorizing patients into "deficit," "no deficit" groups, we analyzed the data using two recognized definitions of cognitive decline (34,35). To that end, we used a 1-SD and a 0.5-SD decline from the presurgical raw cognitive scores in two or more tests as definitions of decline. One-and-a-half hours before surgery, patients were premedicated with temazepam 2040 mg or lorazepam 12 mg by mouth either alone or in combination with morphine (1015 mg) and atropine (0.30.6 mg) IM. Anesthesia was induced with thiopentone (13 mg/kg), etomidate (0.10.2 mg/kg) or propofol (12 mg/kg) IV in addition to fentanyl (410 µg/kg) or remifentanil (12 µg/kg) IV. Neuromuscular blockade was obtained with pancuronium (0.1 mg/kg) or rocuronium (0.9 mg/kg). Anesthesia was maintained with isoflurane (1%2%), propofol target controlled infusion (23 µg/mL) and remifentanil (0.020.1 µg · kg-1 · min-1) or morphine (0.25 mg · kg-1 · h-1) and midazolam (0.04 mg · kg-1 · h-1) IV. Aortic and right heart cannulation was performed after systemic heparinization (300 IU/kg), and CPB was established once the activated clotting time was greater than 450 s. The surgeon commented on the presence of palpable atheroma of the ascending aorta. Aprotinin was administered at the discretion of the surgeon and anesthetist. The CPB circuit consisted of a membrane oxygenator I-3500-2ATM AVECOR Cardiovascular Inc., Plymouth, MN) and a nonpulsatile roller pump (Stockert Instruments, Munich, Germany). Pump flows were maintained at 2.4 L · min-1 · m-2. Distal graft anastomoses were performed during single-application of an aortic cross-clamp and proximal anastomoses by using a side-biting technique. Cold anterograde St Thomass cardioplegia was used. Because a change in clinical practice, the first 49 patients had the CPB circuit primed with 2 L of lactated Ringers solution and sodium bicarbonate 50 mmole. In all subsequent cases, the circuit was primed with 2 L nonlactated Ringers and sodium bicarbonate 50 mmole (n = 53).
Acid base management was alpha-stat and moderate to mild hypothermia (25°36°C) was used. Rewarming was achieved by setting the heat exchanger no more than 10°C above the nasopharyngeal temperature (NPT) and never more than 42°C. Hypovolemia was corrected with packed red cells (Hb < 7 g/dL) and Ringers solution (Hb After the induction of anesthesia, a Baxter Edslab 4FTM (Baxter Healthcare Corporation, Edwards Critical-Care Division, Irvine, CA) double-lumen oximetric catheter was placed in the right internal jugular bulb by using a previously described technique (8). The left internal jugular site was used twice: once when cannulation of the right internal jugular vein failed and once when the right was found to be very small on ultrasound examination Site-RiteTM Dymax Corporation, Pittsburgh, PA). Catheter patency was maintained with a continuous infusion of heparin-saline, and correct placement was confirmed in the intensive care unit (ICU) by a lateral cervical spine radiograph. Catheters were calibrated in vivo. The hemoglobin concentration, used for calibration, was updated when it differed by more than 0.5 g/dL from the previously entered value. Catheters was recalibrated at 36°C during rewarming from hypothermic CPB and whenever the difference between catheter and bench values exceeded 5%. Arterial (radial artery) and jugular venous blood were sampled before CPB (baseline); 10 min into CPB; before rewarming; when NPT reached 36°C; on arrival at the ICU, and 1, 2, and 6 h thereafter. Jugular venous samples were drawn at a rate of 1 mL/min into a preheparinized gas syringe. Samples were analyzed for oxyhemoglobin saturation and concentration on an IL 482 CO-OximeterTM (Instrumentation Laboratory, Lexington, MA); oxygen tension and PaCO2 on a Radiometer ABL4TM machine (Radiometer Ltd., Copenhagen, Denmark); and lactate concentration on a YSI 2300 StatTM machine (YSI Inc, Yellow Springs, OH). CO-Oximeter and the lactate analyzers were calibrated and accuracy confirmed against reference standards before each study. MAP, NPT, and SjO2 were measured by using a Kontron Kolormon 7250TM (Kontron Instruments Ltd, Watford, Herts, UK). Data were logged each minute by using the Edinburgh Monitor/Browser ProgramTM (MRC Head Injuries Clinical Research Initiative, Western General Hospital, Edinburgh, UK). Mean values of the predictor variables (Hb, SaO2, PaCO2, SjO2, Ljv) were determined during CPB (three measurements) and postoperatively (four measurements). The assessment of MAP involved two approaches. First, mean values of individual MAP readings recorded at the time of blood sampling were calculated. Second, the duration with a MAP < 50 mm Hg during CPB and a MAP < 60 mm Hg during the postoperative period were calculated. These levels were chosen because a MAP < 50 mm Hg during rewarming from CPB influences the incidence of SjO2 < 50% (13) and because no patient had a MAP < 50 mm Hg postoperatively, a threshold of 60 mm Hg was used. The area under the curve for SjO2 < 50% was the primary estimate of cerebral hypoperfusion in the postoperative period. This was not used for the intraoperative period because we have previously found a poor agreement between fiberoptic and bench oximetry measurement at this time (36), so that absolute Co-Oximeter values were used during CPB.
Analgesia and sedation in the ICU were achieved with 2-mg IV bolus doses of morphine and/or midazolam as required. Hypovolemia was corrected with plasma protein substitute (Hb
Statistical analysis was performed by using SPSS Version 9.0 for WindowsTM (SPSS Inc, Chicago, IL). Power calculation, based on the actual sample size obtained (102), showed that the study had an 80% power to detect a moderate effect (f2 = 0.15 [f2 represents variance]), assuming multiple regression with up to 7 predictor variables with Data reduction of the highly intercorrelated cognitive test scores was achieved in the following manner. First, the individual subscores of AVLT, WAIS, PASAT, Verbal fluency, Grooved pegboard, and Trails tests were examined and found to be highly correlated. Subscores were transformed to Z-scores and summed to generate overall scores for each test. Directional correction of timed tests (Trail making, Grooved pegboard) was performed to ensure that higher scores on all tests equated to better performance. Factor analysis of the cognitive test battery found that 52% of the presurgery test score variance on these six tests was explained by one component, validating the use of a General Cognitive Factor. The cognitive tests Z-scores were then summed to obtain an overall General Cognitive Score that had a very high correlation with the General Cognitive Factor (Figure 1). For this reason, we elected to use the General Cognitive Score derived from the individual Z-scores as our measure of cognitive ability. Pearsons correlation was used to assess the association of the cognitive Z-scores (the General Cognitive Score and individual tests) with predictor variables. Stepwise multiple regression was used to find predictors of the 3-mo follow-up scores. Criteria for entry to and removal from the model were P < 0.05 and P > 0.10, respectively.
One hundred two (75.6%) patients (90 men, 12 women) completed the study while 11 (8%) had incomplete cognitive data and a further 11 (8%) refused follow-up. Two (1.5%) patients died, two had off-pump CABG surgery, six failed to start the study, and one was withdrawn after sustaining a cardiac arrest in the ICU. Baseline characteristics and operative details of patients who completed the study, those with incomplete follow-up data, and those who declined follow-up are presented in Table 1.
The bias and limits of agreement between the catheter and CO-Oximeter measurements of SjO2 during CPB were 0.5% and -11% to 12%, respectively. The inaccuracy of the fiberoptic catheter during CPB is most likely the result of a wall artifact arising from surgical pressure on the catheter and relatively lower cerebral blood flow at this time (36). Postoperatively, the bias was 0.0% and limits of agreement -5.8% to 5.8%. During the postoperative period, 49 (48%) patients experienced a SjO2 < 50%, and the mean total duration of SjO2 < 50% was 44 (SD 97) min. The median number of episodes of SjO2 < 50% was 2 (range, 114), and the median duration of a single episode was 20 (range 5104) min. No interventions were taken to treat SjO2 < 50%. Candidate predictor variables of cognitive decrement are presented in Table 2. CPB and postoperative values of Hb, SaO2, PaCO2, and MAP were all significantly (P < 0.05) decreased compared with pre-CPB values.
Raw scores for all the cognitive tests are listed in Table 3. Correlation coefficients between the putative predictor variables and the six individual Z-scores (AVLT, WAIS, PASAT, Grooved pegboard, and Trail-Making) and the postoperative General Cognitive Score are presented in Table 4.
The preoperative cognitive score accounted for the majority of the postoperative variance for all cognitive scores (47% to 82%) while palpable aortic atheroma accounted for between 1% and 5% additional, independent variance. Priming the CPB circuit with lactated Ringers was detrimental to Verbal Fluency, and verbal memory was adversely affected by Hospital Anxiety and Depression scores (Table 5).
The incidence of cognitive decline was dependent on the definition used, being either 7% (1 SD definition) or 24% (0.5 SD definition). The between group differences for age, area under the curve of SjO2 < 50%, and aortic atheroma are tabulated (Table 6).
We confirms that SjO2 < 50% occurs in many patients during the early postoperative period after CABG surgery (8). However, cognition three months after CABG surgery is not associated with postoperative SjO2 < 50% or SjO2 values during CPB. Moreover, although a preliminary study suggested that anerobic estimates of cerebral hypoperfusion after CABG surgery (9) might predict cerebral damage, no association was found between jugular lactate levels and long-term cognitive outcome. There are several possible explanations for these negative associations and why our findings differ from those of Croughwell et al. (6). With 102 subjects, our study had 80% power to detect a medium effect size in a multiple regression with seven predictor variables and set at 0.05. If the true association between SjO2 < 50% and cognitive change is smaller than this (i.e., approximately 9% shared variance), then a larger powered study is needed. However, our study group showed frequent postoperative SjO2 < 50% (48%), and anything larger than a small effect of SjO2 < 50% should have been apparent. Second, we studied patients at three months compared with 48 days after cardiac surgery, and it may be that SjO2 < 50% during CPB and postoperatively affects short- but not long-term cognition. Notwithstanding the above limitations, the most likely explanation for our results is that postoperative SjO2 < 50% does not influence long-term cognitive outcome. Categorizing patients into arbitrary groups of "deficit" and "no deficit" results in a three-fold difference in the incidence of deficit and potentially groups patients incorrectly because a larger proportion of excellent performers are categorized as showing deficit (37). Our approach has the advantage that it does not suffer from regression toward the mean and accommodates learning effect. However, for comparison, the categorical incidence of "cognitive deficit" in our study group was 7%24%. This is similar to recent cardiac surgery literature as well as the International Study of Postoperative Cognitive Dysfunction that found an incidence of 9.9% three months after major noncardiac surgery (34,35,37,38). Interestingly, there was no significant difference in SjO2 < 50% between "deficit" and "no deficit" groups, although the 1-SD deficit group showed a significant incidence of aortic atheroma (Table 6). We postulated that a pathological process, such as cerebral edema and vasoconstriction (7,8), underlay the occurrence of postoperative SjO2 < 50%. However, hypocapnea resulting from mechanical hyperventilation may be the cause because SjO2 is highly related to PaCO2 (9,12). The importance of SjO2 < 50% in Croughwell et al.s study (6) and neurotrauma (10) may reflect other pathophysiological influences on SjO2. During the course of the study, a clinical decision was made to prime the CPB circuit with nonlactated rather than lactated Ringers solution to prevent postoperative lactemia (9). A lactated prime correlated significantly with AVLT and Verbal fluency subscores. However, after multiple regression, only verbal fluency was significantly influenced. The mechanism for this effect is difficult to explain, and its importance on cognitive outcome is questionable, given that it only influenced a single test and arose from post hoc analysis. As other investigators have found that age, aortic atheroma, and the duration of CPB influence cognitive outcome after CABG surgery (1,2), we entered these variables into our analysis. Unlike previous studies (1,2), no significant correlation between the duration of CPB and cognition was found, and this is in keeping with a study by Selnes et al. (5). In contrast, age and the surgeons assessment of the aorta were both significantly correlated with the General Cognitive Score as well as several of the cognitive subscores in the present study. However, after multiple regression, aortic atheroma was significantly predictive of the General Cognitive Score and three cognitive subscores while age was related to only one subscore. Therefore, ages correlation with cognitive outcome may largely be an expression of its association with the preoperative cognitive score and increasing incidence of aortic atheroma and not a direct interaction of age and CABG surgery (5). The surgeons assessment of the aorta accounted for approximately 2% of the variance in the three-month General Cognitive Score after preoperative score was entered. Epiaortic scanning is more sensitive than palpation (39), and if it had been used, the importance of aortic atheroma within the model might have been greater. To improve patient outcome, research must identify those factors underlying the mechanisms of cognitive change. Future studies may examine other putatively causal factors using a similar design as ours. However, there is a limitation to the present design. When a factor is common to all subjects and affects cognition equally, our individual differences approach would be inappropriate. It is appropriate, as here, where the hypothesized factors show individual differences, and it is legitimate to ask whether variance among subjects in the "causal" variable contributes to variance in cognitive outcomes. In conclusion, many patients experience episodes of SjO2 < 50% during the early postoperative period after CABG surgery. However, SjO2 < 50% does not appear to make an important contribution to long-term cognition.
AVDO2 (mL/L) = CaO2 - CjO2 CaO2 (mL/L) = [Hb (gl-1) x 1.34 x SaO2/100 (%)] + [PaO2 (kPa) x 0.225] CjO2 (mL/L) = [Hb (gl-1) x 1.34 x SjO2/100 (%)] + [PjO2 (kPa) x 0.225] CO2 (mmol/L) = CO2 (mL/L-1)/22.4 LOI = -AVDL (mmol/L)/AVDO2 (mmol/L) Where AVDO2 = cerebral arteriovenous difference in oxygen concentration, Cao2 = arterial oxygen content, CjO2 = jugular bulb oxygen content, Hb = hemoglobin concentration, SaO2 = arterial oxyhemoglobin saturation with oxygen, SjO2 = jugular bulb oxyhemoglobin saturation with oxygen, PaO2 = arterial oxygen tension, PjO2 = jugular bulb oxygen tension, and LOI = lactate oxygen index, AVDL = cerebral arteriovenous difference in lactate concentration.
Supported by Welcome Trust Grant 050190. We would like to thank Timothy P. Howells, BA, Msc, for supplying and modifying the computer-based data collection system.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|