| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Click here to read this article as a CME activity
Department of Anesthesiology, Duke University Medical Center, Durham, North Carolina
Accepted for publication June 6, 2007.
* Address correspondence to Dr Bar-Yosef, Department of Anesthesiology, Duke University Medical Center, Box 3094, Durham, NC 27710. (Email: baryo001{at}mc.duke.edu).
| Adult cardiac surgery:
The Annals of Thoracic Surgery CME Program is located online at http://cme.ctsnetjournals.org. To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal.
|
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentReferences |
|---|
Methods: A total of 525 patients having on-pump coronary artery bypass graft (CABG) procedures underwent cognitive testing at baseline and 6 weeks postoperatively. Multivariable linear regression was used to determine the relationship between NCD and intraoperative hyperglycemia (glucose 
200 mg/dL). Diabetic and nondiabetic patients were analyzed separately to eliminate a possible confounding effects between diabetes and hyperglycemia.
Results: In the nondiabetic patients, even after controlling for age, years of education, and baseline cognitive function, hyperglycemia was associated with a decrease in cognitive function at 6 weeks (p = 0.0351). Hyperglycemia had no effect on cognitive function in diabetic patients, however.
Conclusions: These findings suggest that in nondiabetic patients undergoing CABG operations, intraoperative hyperglycemia is associated with an increased risk of NCD.
Hyperglycemia occurs commonly during cardiac operations, particularly during cardiopulmonary bypass (CPB). Peripheral insulin resistance resulting from stress response–induced increases in circulating catecholamines and cortisol [1], decreased cell metabolism from hypothermia [2], and increased renal tubular reabsorption of glucose [3] all play a role. In addition, exogenous administration of dextrose, most notably in the cardioplegia solution as well as variably in the pump prime [4], also contributes to intraoperative hyperglycemia.
To date there is no consensus on the definition of intraoperative hyperglycemia, although most commonly poor intraoperative glucose control is defined as a maximal intraoperative glucose level greater than 200 mg/dL [5–7]. Similarly, a retrospective study of 4701 nondiabetic patients undergoing cardiac operations found that 41% had peak intraoperative glucose levels of 181 to 272 mg/dL [8]. A recent study in diabetic patients undergoing on-pump cardiac operations found a 36% incidence of glucose level higher than 180 mg/dL [5].
The occurrence of hyperglycemia in cardiac surgical patients is not benign. Several studies have shown that tight glycemic control in the perioperative period decreases mortality after cardiac procedures and reduces the incidence of postoperative adverse events, including infectious complications (pneumonia and mediastinitis), myocardial complications (low cardiac output, atrial fibrillation, recurrent ischemia), and prolonged mechanical ventilation and intensive care unit (ICU) length of stay [6–8].
These studies, however, have not made a distinction between the intraoperative and postoperative periods. More recent studies have shown that intraoperative hyperglycemia is associated with an increased ICU length of stay and an increased risk of postoperative morbidity and mortality, even when controlling for postoperative glucose levels [5, 9].
Hyperglycemia may be especially harmful to the ischemic brain. Multiple studies have demonstrated the association between peri-insult hyperglycemia and a worse clinical outcome in settings of stroke [10], subarachnoid hemorrhage [11], traumatic brain injury [12], and cardiac arrest [13]. Cerebral injury, manifested by stroke or neurocognitive dysfunction (NCD), is also a common problem after cardiac operations.
Recently, we have shown a 36% prevalence of cognitive dysfunction 6 weeks after coronary artery bypass grafting (CABG) [14]. Both ischemic injury and inflammation have been proposed as etiologic factors for NCD in cardiac surgical patients [15]. Because hyperglycemia is known to worsen cerebral ischemic injury and to exacerbate inflammation [16], it could have the potential to worsen neurocognitive function as well. Data are lacking on its potential effects on adverse cerebral outcome after cardiac operations. This study tests the hypothesis that intraoperative hyperglycemia during CABG is associated with postoperative cognitive dysfunction.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Anesthetic and Surgical Management
Anesthesia and CPB management were relatively standardized during the study period. Anesthesia was induced and maintained with intravenous (IV) midazolam (0.02 to 0.1 mg/kg), IV fentanyl (10 to 25 µg/kg), and isoflurane (up to 1%) with muscle relaxation provided by IV pancuronium (0.1 mg/kg). All patients underwent nonpulsatile hypothermic (30° to 34°C) CPB using a membrane oxygenator, 40-µm arterial filter, and roller pumps. Perfusion was maintained at pump flow rates of 2.0 to 2.4 L/(min · m2). The pump was routinely primed with nonglucose-containing crystalloid solutions, and serial hematocrits were kept at 18% or higher with packed red blood cell transfusion as necessary. Intraoperatively, shed mediastinal blood was returned to the venous reservoir using cardiotomy suction.
Arterial blood gases were measured every 15 to 30 minutes to maintain arterial carbon dioxide partial pressures at 35 to 40 mm Hg (unadjusted for temperature) and oxygen partial pressures at 150 to 250 mm Hg. A mean arterial pressure between 50 and 90 mm Hg during CPB was achieved using IV phenylephrine or sodium nitroprusside, or both, as required. Myocardial protection was achieved with intermittent cold anterograde or retrograde hyperkalemic blood cardioplegia (4:1 blood-crystalloid ratio), or both. The crystalloid portion of the cardioplegia contained dextrose (50 g/L). An intravenous infusion of insulin was started and titrated to effect at the discretion of the attending anesthesiologist.
Intraoperative Data Collection
Intraoperative patient data, including medications given and laboratory results, were recorded using automated anesthesia record keeping (Arkive, Diatek, San Diego, CA, and SATURN, North American Draeger, Telford, PA). Blood glucose levels were measured before and after CPB as well as every 30 minutes during CPB. The highest level measured was chosen for analysis, and hyperglycemia was defined by a highest glucose level of 200 mg/dL or more.
Cognitive Testing
A previously reported and validated standard neurocognitive test battery was administered by experienced psychometricians one day before surgery (baseline) and again 6 weeks after surgery. Tests used included the short story module of the Randt Memory Test, the Digit Span and Digit Symbol subtests of the Wechsler Adult Intelligence Scale-Revised, the Trail Making Test, Part B, and the Modified Visual Reproduction Test from the Wechsler Memory Scale [14].
Statistical Analysis
The methodology used to analyze the neurocognitive data has been described previously in detail [14]. For neurocognitive data, four distinct cognitive domains were identified using factor analysis: (1) verbal memory and language comprehension (short term and delayed); (2) attention, psychomotor processing speed and concentration; (3) abstraction and visuospatial orientation; and (4) figural memory. An overall cognitive index score at each test time point was determined by averaging the independent factor scores. A continuous outcome variable—a cognitive change score—was then defined as the difference between the 6-week and the baseline cognitive indices and was used as the primary end point in linear regression analysis.
Multivariable models were constructed, controlling for age, years of education, baseline cognitive function, and preoperative diabetes status. As a secondary end point, a dichotomous outcome variable for cognitive deficit was defined as a decline of at least one standard deviation (SD) calculated from the baseline scores in at least one of the four cognitive domains. This was analyzed using logistic regression. Although cognitive change score and cognitive deficit are both measures of cognitive dysfunction, they are not necessarily equivalent: a patient might demonstrate a deficit in one cognitive domain but still show average improvement on the cognitive change score.
In addition to this core model, we also tested for the possible confounding effect of those variables that were significantly different between the hyperglycemic and nonhyperglycemic group and between the diabetic and nondiabetic group. These variables included gender, body mass index (BMI), ejection fraction (EF), CPB time, number of grafts, and the use of insulin intraoperatively (a yes/no dichotomous variable). Also tested was the effect of year of operation on the postoperative cognitive outcome.
Hyperglycemia was defined by the occurrence of at least one glucose level of 200 mg/dL or more during the operation. Patients were further categorized as diabetic and nondiabetic according to a preoperative medical diagnosis of diabetes mellitus requiring therapy with oral agents or insulin.
Data are reported as mean ± SD, unless otherwise stated. Demographic comparisons between groups were done by Wilcoxon rank sum for continuous variables and the Fisher exact test for dichotomous variables. All statistical tests were two-tailed, and p < 0.05 was considered significant. Analysis was done using SAS 9.1 software (SAS Institute, Cary, NC).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
|
|
A total of 380 patients (72% of the study population) were nondiabetic. Compared with the diabetic patients, these patients were leaner (BMI, 28.5 ± 5.1 kg/m2 versus 31.5 ± 5.9 kg/m2, p < 0.0001), had a higher EF (0.551 ± 0.112 versus 0.518 ± 0.113, p = 0.0029), had a shorter CPB time (104.6 ± 30 minutes versus 112.9 ± 27.5 minutes, p = 0.0018), and included less women (24.5% versus 38.6%, p = 0.0017). In nondiabetic patients, hyperglycemia was significantly associated with a decrease in cognitive change score (continuous outcome variable) at 6 weeks in both univariate analysis (p = 0.04; Fig 1A) and after adjustment for baseline cognitive score, age, and educational level (p = 0.035; Table 3). Gender, BMI, EF, CPB time, and the date of surgery were nonsignificant and were omitted from the final model. In the diabetic patients, no association of cognitive change with hyperglycemia was found (p = 0.74, Fig 1A).
|
|
Intravenous insulin was used intraoperatively in 44% of the patients. There was a strong association between diabetes and insulin: 84% of diabetic patients received some insulin compared with 25% of nondiabetic patients (p < 0.0001). Not surprisingly, hyperglycemia was also associated with insulin use: 49% of hyperglycemic patients received some insulin compared with 27% of patients without hyperglycemia (p < 0.0001). When diabetic and nondiabetic patients were analyzed separately, we found that 31% of the nondiabetic patients with hyperglycemia received some insulin compared with 11% of nondiabetic patients without hyperglycemia (p < 0.0001). In our multivariables models, the use of insulin had no association with cognitive dysfunction in diabetic and nondiabetic patients (p = 0.18 and 0.33, respectively).
Of the original 703 patients who had a baseline cognitive testing, 178 (25%) were lost to follow-up. The reasons for their attrition are summarized in Table 4. Overall, patients lost to follow-up tended to be older, sicker (lower EF and higher incidence of diabetes), had a lower baseline cognitive status, and had undergone a longer operation (Table 1). There was no difference, though, in the degree or incidence of hyperglycemia between the patients who were lost to follow-up and those who completed the study (p = 0.4479). Death accounted for 11 of the nonreturners and was evenly distributed between diabetic (1.4%) and nondiabetic patients (1.6%): all 11 experienced hyperglycemia. A "worst-case" analysis assigning the worst observed cognitive change score to these 11 deaths confirms our primary findings (multivariable hyperglycemia effect p = 0.0067 in nondiabetic patients; p = 0.6201 in diabetic patients).
|
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
Since the original finding by Myers and Yamaguchi [17], hyperglycemia has been repeatedly shown to have deleterious effect on the brain in animal models of both focal and global cerebral ischemia [18, 19]. Several mechanisms were proposed by which hyperglycemia may adversely affect the brain. Increased intracellular glucose accumulation might lead, under anaerobic conditions of ischemia, to increased lactate production and lactic acidosis [20]. Neuronal acidosis has been shown to increase lipid peroxidation [21] and intracellular calcium accumulation [22], resulting in expansion of ischemic penumbral regions into frank infarction [20].
Another theory revolves around the role of the excitatory amino acid neurotransmitter glutamate, which is known to play a pivotal role in neuronal death through increasing cellular calcium influx. A study by Li and colleagues [23] demonstrated that hyperglycemia increased glutamate accumulation in the ischemic rat brain and was associated with exaggerated neuronal damage.
A third possible mechanism relates to the tendency of hyperglycemia to increase corticosteroid production, which may have its own neurotoxic effect [24]. Acute hyperglycemia, both in diabetic and nondiabetic subjects, has also been shown to induce inflammatory cytokines production [16], which by themselves might have injurious neuronal effects [25].
Finally, hyperglycemia in animal models of brain ischemia was found to aggravate edema formation through disruption of the blood–brain barrier [26]. This edema, combined with a described decrease in regional cerebral blood flow induced by hyperglycemia [27], might also contribute to increasing ischemia of the penumbra regions.
An increasing number of clinical studies have found that hyperglycemia is associated with increased morbidity and mortality after stroke [10], subarachnoid hemorrhage [11], traumatic brain injury [12], and neurologic morbidity after cardiac arrest [13]. However, it is not clear whether this is a causative effect or simply an association of hyperglycemia as a marker of a worse clinical situation without having a primary pathogenetic role. Experimental studies of focal ischemia, however, have found that reducing the glucose level by using insulin does eliminate the increased brain damage associated with hyperglycemia [28].
The effect of hyperglycemia on cerebral injury related to cardiac operations has been incompletely investigated. For many years, the use of solutions containing glucose for priming the CPB circuit was a common practice. In a study of 70 patients, Newman [29] randomly allocated patients to either electrolyte or glucose-containing prime and found that neuropsychologic outcome was worse in the group that received glucose-containing prime [29]. Metz and Keats [30], however, found that adding glucose to the priming solution had no effect on the incidence of postoperative stroke. In a large retrospective study of more than 2600 patients undergoing CABG, Wermeskerken and colleagues [31] found no association between maximal intraoperative glucose levels and the incidence of adverse neurologic outcome, defined as the occurrence of postoperative stroke, coma, or transient ischemic attack.
A much smaller study of 29 patients undergoing aortic reconstruction under hypothermic circulatory arrest did show that hyperglycemia (>250 mg/dL) was a risk factor for neurological morbidity [32]. Furthermore, in a study in 200 diabetic patients undergoing on-pump cardiac operations, intraoperative hyperglycemia (>200 mg/dL) was associated with an increased risk of neurologic morbidity, defined as stroke or irreversible encephalopathy [5].
Two previous studies have specifically examined the association between cognitive dysfunction after CABG and intraoperative glucose concentrations and found no relationship. One was a retrospective, relatively small study of 60 patients, and NCD was evaluated only at discharge from the hospital and not at 6 weeks, a more clinically relevant time point [33]. A more recent study randomized 381 nondiabetic patients undergoing CABG to receive insulin or placebo, aiming at a glucose level during CPB of less than 100 mg/dL [34]. Neurocognitive function and neurologic morbidity were examined at 1 week, 6 weeks, and 6 months; no difference was found between the groups.
Although contradictory to our results, two possible explanations exist. First, Butterworth and colleagues’ [34] definition of neurocognitive deficit differs from ours and is based on a 20% decrease from baseline in at least two neuropsychiatric tests. This definition might exclude less severe degrees of cognitive dysfunction, especially because repeated testing is expected to result in better scores due to a learning effect. Indeed, they have found only a 20% incidence of cognitive dysfunction at 6 weeks versus 30% to 40% in our study. Second, it seems that many patients in the placebo group had a maximal glucose level of less than 200 mg/dL, the threshold used to define the hyperglycemic group in our study.
Of interest is that hyperglycemia in our study was associated with NCD only in nondiabetic patients (Table 3, Fig 1). There are several possible explanations for this difference between diabetic and nondiabetic patients, including that the hyperglycemia threshold (200 mg/dL) may be inadequate for diabetic patients because their brain might be "accustomed" to higher glucose levels. To test this hypothesis, we did examine other glucose threshold values in the diabetic group, ranging from 200 to 350 mg/dL. None of the threshold values affected the lack of association with NCD (data not shown).
Second, and also related to the choice of a threshold for defining hyperglycemia, we may have been underpowered to detect an association in diabetic patients because their total number was relatively small, at 145 patients, and the number of nonhyperglycemic diabetic patients was even smaller, at only 27, when the glucose level of 200 mg/dL was chosen as the hyperglycemia threshold.
A third possible explanation is that more patients with diabetes were treated with insulin compared with nondiabetic patients. Insulin might have a beneficial effect on global brain ischemia independent from its hypoglycemic effects [35]. An intriguing finding is that diabetes has been shown to nullify the effects of acute hyperglycemia also in the setting of ischemic stroke [10]. Further studies are required to investigate the possibility that the lack of deleterious neurologic effects of hyperglycemia in diabetic patients is reflective of some yet unknown pathophysiologic mechanism.
Our study has some potential limitations. Our rate of nonreturn for cognitive testing follow-up was 25%, a common problem in studies of postoperative NCD. In addition, the nonreturners had lower baseline cognitive scores and were sicker in several respects compared with returners (Table 1), making them more likely to have had lower 6-week scores than the returners. However, the rate of perioperative hyperglycemia was not related to return status (75% versus 78%, p = 0.4196), giving us no reason to believe that its effect would be different in nonreturners than in returners.
The loss to follow-up rate usually increases with a longer follow-up time. In our patients, follow-up was performed at 6 weeks after surgery, assuming that at this time period, the short-lived confounding effects of anesthesia, pain, metabolic abnormalities, and some medications used immediately after operation should be minimal. It is, however, possible that with even longer follow-up, some cognitive recovery will still occur, which is not captured by our study design.
We deliberately chose to analyze the glucose level only intraoperatively, assuming that this is the period with the highest risk of cerebral damage from emboli and hemodynamic changes. The possible role of delayed postoperative hyperglycemia was not examined in this study.
Another limitation derives from the length of time over which patients were recruited: our study spans 11 years. During this period, the management of intraoperative hyperglycemia has evolved to the almost routine use of intravenous insulin in most patients in an effort to reduce hyperglycemia, resulting in increased insulin usage and decreased mean glucose levels (data not shown). Also related to this is the possibility that as mean glucose levels have decreased during these years, other factors that have changed over time and affected NCD might have confounded our results. However, we found no effect of the operation date on the degree of cognitive dysfunction or on the relationship between hyperglycemia and NCD.
Another possible confounding factor is CPB duration: longer CPB might both exacerbate hyperglycemia (due to cardioplegia solutions containing glucose) and affect other factors that might increase NCD (hypoperfusion, microemboli, inflammatory response). However, we have found no relationship between CPB time and NCD after CABG. Several additional factors might affect cognitive outcome, including carotid atherosclerotic disease and the surgical technique itself, and there might be other as yet unknown factors that were not tested in the current study.
Another limitation is that only the presence or absence of insulin—but not the total dose—was available for analysis. We were therefore limited in the ability to analyze the possible separate effects of glucose control and insulin.
In summary, we have shown that in nondiabetic patients, even after controlling for age, years of education, and baseline cognitive function, the cognitive deficit rate was significantly higher in patients experiencing hyperglycemia. As in any retrospective study, our results should be interpreted mainly as a demonstration of an association, not necessarily causation. Future prospective studies should strive to define the exact role of hyperglycemia in post-CABG cognitive outcome, the optimal intraoperative glucose level, and the threshold for instituting insulin treatment.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. P. Grocott Hyperglycemia and postoperative cognitive dysfunction: another call for better glycemic control?/L'hyperglycemie et la dysfonction cognitive postoperatoire : faut-il un meilleur controle de la glycemie? Can J Anesth, March 1, 2008; 55(3): 140 - 145. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
