Glucose Control Regimens

Practice patterns among anesthesiologists regarding intraoperative blood glucose control have undergone substantial changes. A survey of British anesthesiologists in 1993 demonstrated that a greater proportion were likely to intervene to maintain the perioperative blood glucose levels in their diabetic patients at less than 180 mg/dL and that they were more likely to do so with separate infusions of insulin and glucose rather than glucose–insulin–potassium (GIK) solutions then they had when similarly surveyed in 1985. The dramatic results of the 2001 Leuven trial of tight glycemic control in critically ill patients encouraged many to extrapolate their findings to the management of perioperative and intraoperative blood glucose levels, in an attempt to aggressively target a blood glucose level of 80 to 110 mg/dL while in the operating room during anesthesia for cardiac surgery. Adoption of this practice has been controversial, and it has become clear from investigations among critically ill patients that such tight glycemic control is associated with a substantial risk of hypoglycemia, with its attendant morbidity and mortality. Some groups have extrapolated the Leuven results to the intraoperative period but have had difficulty achieving such tight control. Additionally, achieving and maintaining tight glucose control in the intensive care unit (ICU) has proved to require a substantial outlay of resources: an average of 4.72 minutes (range, 3.13 to 8.15) per hour was devoted to measuring blood glucose and adjusting insulin infusions, which is time that could be spent on other aspects of patient care.

Given the risk of hypoglycemia associated with tight glycemic control in the ICU as well as the increased resources needed, there is a relative paucity of clinical data to support its use in the perioperative period. Two randomized trials have attempted to address this issue for cardiac and vascular surgery, and one meta-analysis has attempted to address the issue among a more heterogeneous population of surgical patients. In the first trial, 400 patients undergoing cardiac surgery were randomly assigned to intensive insulin control (target blood glucose level, 80 to 100 mg/dL) versus a conventional algorithm, in which patients did not receive insulin unless their blood sugar exceeded 200 mg/dL. Although the study size was small and the investigators were unable to reach their target blood glucose levels in the experimental group, they did find a statistically insignificant increase in mortality, stroke, and heart block requiring a pacemaker in the intensive insulin control group despite having equivalently low rates of hypoglycemia between the two groups. In a subsequent meta-analysis, the same authors attempted to examine the effect of insulin in the perioperative period. Analysis of the pooled results in which the authors compared a heterogeneous group of patients undergoing a variety of interventions demonstrated an improvement in the 30-day mortality rate with intensive insulin therapy (relative risk [RR], 0.69; 95% confidence interval [CI], 0.51 to 0.94) but a significantly increased incidence of hypoglycemia (RR, 2.07; 95% CI, 1.29 to 3.32) in the 20 trials that included data on hypoglycemia. These authors concluded that perioperative insulin use may decrease mortality rates and increase postoperative hypoglycemia but that their mortality data were too unreliable to draw definitive conclusions. In a subsequent trial, 236 patients undergoing major vascular procedures were randomly assigned to a continuous insulin infusion targeting a less aggressive blood glucose level of 100 to 150 mg/dL versus intermittent intravenous insulin boluses for blood glucose levels more than 150 mg/dL. The authors evaluated a composite primary endpoint that included death, myocardial infarction (MI), and congestive heart failure and noted a significant decrease (3.5% versus 12.3% [ p = 0.013]) among the insulin infusion group. Although the authors noted that the incidence of hypoglycemia (8.8% versus 4.1% [ p = 0.18]) was similar in both groups, there did appear to be a trend toward more frequent hypoglycemia in the insulin infusion arm. In light of these studies, as well as the abundant literature on glycemic control among the critically ill, the American Diabetes Association/American Association of Clinical Endocrinologists revised their recommendations for critically ill patients to include maintenance of blood glucose between 140 and 180 mg/dL. Similarly, the Society for Thoracic Surgeons recommends a target blood glucose range of less than 180 mg/dL for patients undergoing cardiac surgery.

Perioperative Hyperglycemia and the Outcome of Critically Ill Patients

Hyperglycemia associated with insulin resistance is common in critically ill patients regardless of whether they have previously been given a diagnosis of diabetes and is associated with adverse outcomes in this group. The management of hyperglycemia in critically ill patients has undergone revision with the publication of conflicting results from large-scale randomized controlled trials. Initial enthusiasm for intensive insulin therapy, targeting a blood glucose level of 110 mg/dL, was based on the results of a study of surgical ICU patients at a single center in Leuven, Belgium. At the time of this study the standard of care in most centers was to tolerate hyperglycemia that did not reach the threshold for glycosuria (serum glucose level of 215 mg/dL). Thus the investigators used this threshold as the control arm in their trial. In patients randomly assigned to intensive insulin therapy (maintenance of blood sugar at or below 110 mg/dL), the authors demonstrated a 34% reduction in the in-hospital mortality rate, a 46% reduction in bloodstream infections, and a 41% reduction in acute renal failure requiring hemodialysis or hemofiltration when compared with patients receiving conventional therapy (insulin infusion if the blood glucose level exceeded 215 mg/dL with maintenance between 180 and 200 mg/dL). The results of this trial were so compelling that they were rapidly incorporated into guidelines issued by professional organizations, as well as practice patterns in ICUs internationally.

However, the authors of the Leuven trial were unable to reproduce the survival benefit of intensive insulin control in a follow-up study of medical ICU patients comparing the same blood glucose control strategies as in their earlier work. Although they were able to demonstrate a survival advantage for the subset of patients who required a prolonged (greater than 3 day) ICU stay, they also found a disturbing association between intensive insulin therapy and increased mortality rate in the subset of patients that stayed for shorter periods in the ICU. A large meta-analysis carried out at the same time was similarly unable to demonstrate a survival benefit, but did find a significant association between intensive insulin therapy and the development of hypoglycemia. This association between intensive insulin therapy and hypoglycemia was again demonstrated in two subsequent randomized controlled trials involving tight glycemic control, both of which were terminated because of unacceptably high rates of hypoglycemia, which has led many to question the safety of intensive glucose control regimens.

Coincidently, the Normoglycemia in Intensive Care Evaluation—Surviving Using Glucose Algorithm Regulation (NICE-SUGAR) investigators were completing a large multicenter randomized controlled trial comparing intensive insulin therapy versus standard therapy. This trial was not designed to address the safety concerns raised by the interceding trials but rather was powered to detect an absolute difference in mortality favoring intensive insulin therapy as compared with standard glucose control at the time (140 to 180 mg/dL). In stark contrast to the Leuven group’s findings, the NICE-SUGAR investigators found that the intensive insulin group had a significantly higher mortality rate (27.9% versus 24.9%; p = 0.02) than did control subjects and was also associated with much higher rates of severe hypoglycemia.

While tight glycemic control with intensive insulin therapy was widely and rapidly embraced after the publication of the Leuven trial, the results of more recent studies demonstrating a significant association with hypoglycemia and particularly increased mortality rates with intensive insulin therapy have largely led to its abandonment. Currently, the American Diabetes Association recommends targeting a blood glucose range of 140 to 180 mg/dL for the majority of critically ill patients, reserving a more stringent target of 110 to 140 mg/dL for select patients as long as this can be achieved without significant hypoglycemia.

Effect of Stress-Related Hormonal Changes on Metabolic Changes in Diabetic and Nondiabetic Patients

The effects of surgery and the stress response on the development of hyperglycemia have been recently reviewed. Activation of the sympathetic pituitary and adrenal systems in response to acute stress (e.g., injury in the perioperative period) leads to the secretion of counter-regulatory hormones such as epinephrine, norepinephrine, glucagon, cortisol, and growth hormone, which stimulate increased hepatic glucose production and peripheral insulin resistance. Increased hepatic glucose synthesis through gluconeogenesis is responsible for the majority of stress-induced hyperglycemia and is primarily a response to the increased glucagon secreted by the pancreas in response to high levels of circulating catecholamines, particularly epinephrine. Insulin secretion from the pancreas is inhibited by circulating catecholamines, particularly norepinephrine through alpha-2 dependent pathways. The mechanism of peripheral insulin resistance is poorly understood but is mediated, in part, by cortisol and epinephrine and involves reduced insulin-mediated glucose uptake (IMGU) through the insulin-inducible facilitated glucose transporter GLUT-4 found in skeletal muscle and adipose tissue, as well as reduced skeletal muscle glycogen synthesis. Insulin resistance also promotes lipolysis with the formation of excessive free circulating fatty acids. The net effect of these pathways is elevated levels of glucose and free fatty acids. Hyperglycemia and hyperlipidemia induce oxidative stress through increased free radical generation and reduce the bioavailability of nitric oxide, which leads to vasoconstriction and platelet aggregation. Hyperglycemia also stimulates inflammation through increased cytokine production and causes increased release of tissue factor, which activates the coagulation cascade.

The degree of stress-induced hyperglycemia has been shown to be proportional to surgical stress in postoperative patients, to the degree of myocardial stress in patients with acute MI and with the severity of neurologic injury in patients with trauma. Although it has been well-documented that stress-induced hyperglycemia is associated with morbidity and mortality, several lines of evidence indicate that the mortality effect of stress-induced hyperglycemia is different for patients with diabetes and without diabetes. It appears that patients without diabetes incur a higher degree of morbidity and mortality for a given elevation of blood glucose level. Among ICU patients it has been demonstrated that ICU mortality is greater for patients without diabetes than for patients with diabetes at any blood glucose level. In patients who have sustained a MI the mortality effect of short-term blood glucose elevation was greater for patients without diabetes in a large observational study and a meta-analysis. Similar results have been demonstrated in patients with ischemic stroke and intracranial hemorrhage.

Effect of Perioperative Hyperglycemia on Wound Healing and Postoperative Infections

Observational studies have established that diabetic patients are at greater risk of developing a variety of infections, including pneumonia, cystitis, and surgical site infections. Acute hyperglycemia frequently accompanies severe physiologic stress (e.g., surgery) and has also been shown to be associated with an increased risk of infectious complications. Hyperglycemia is a potent immunomodulator, leading to significant and sustained decreases in the function of neutrophils, which decreases chemotaxis, adherence, phagocytosis, and bacteriocidal activity, all of which culminate in an increased susceptibility to bacterial infections at blood glucose levels greater than 200 mg/dL. In experimental models, many of these effects could be reversed with improved glycemic control. The beneficial effects of insulin and glycemic control on wound healing have been demonstrated in animal studies.

Although it is clear that hyperglycemia is associated with an increased risk of infection in the perioperative period, data to support a causal relationship between the two are inconclusive. The Leuven trial, which is still one of the largest randomized clinical trials that used infection as a primary endpoint, demonstrated a 46% reduction in bloodstream infections with tight glycemic control (blood glucose target, <110 mg/dL) in a predominantly postsurgical population. However, subsequent data have called into question the safety of such a target in critically ill patients. Five randomized trials have been conducted in perioperative patients in order to determine the effect of perioperative insulin infusion and blood glucose control on infectious complications, and their results have been summarized in a recent Cochrane review. Ghandi et al investigated the effect of intensive intraoperative insulin therapy in a randomized trial of 400 cardiac surgery patients targeting a blood glucose level of 80 to 100 mg/dL and was unable to demonstrate an association with surgical site infections. Bilotta et al examined strict glycemic control (blood glucose level, 80 to 120 mg/dL) versus conventional control (blood glucose level, 80 to 200 mg/dL) in 180 patients undergoing emergency cerebral aneurysm clipping and found a statistically insignificant trend toward lower infection rates in the tight glycemic control group, although there was also a trend toward hypoglycemia in this group. Grey and Perdrizet randomly assigned 61 hyperglycemic surgical ICU patients to strict glycemic control (blood glucose level, 80 to 120 mg/dL) versus standard control (blood glucose level, 180 to 220 mg/dL) and found more nosocomial infections in the conventional control group; however, they did not report rates of hypoglycemia. The authors of the Cochrane review stated that there was no evidence to support the use of tight glycemic control, below the standard blood glucose target of less than 200 mg/dL for prevention of infections in postoperative patients.

Perioperative Hyperglycemia and Outcome after Cardiovascular Surgery

Cardiac surgery, particularly hypothermic cardiopulmonary bypass, presents several unique challenges to the management of perioperative blood glucose levels. The etiology of hyperglycemia during cardiopulmonary bypass is multifactorial. Hypothermia during bypass suppresses insulin secretion in the face of hyperglycemia. Furthermore, insulin resistance may be profound as a result of increased levels of counter-regulatory hormones such as epinephrine, cortisol, glucagon, and growth hormone, which, in conjunction with increased glucose reabsorption by the kidneys, may lead to profound hyperglycemia. Multiple lines of evidence have indicated that the resulting hyperglycemia is an independent factor associated with increased short- and long-term morbidity and mortality after cardiac surgery; however, many conflicting results exist in the literature, and the magnitude of this risk and the ideal blood glucose level for risk reduction remain unclear.

Many of the earliest reports suggesting an association between hyperglycemia and morbidity or mortality after cardiac surgery were retrospective studies using historical control subjects, used a variety of definitions for hyperglycemia, and were subject to the confounding influence of the many other improvements in the care these patients were exposed to during the study period. However, these trials demonstrated significant associations between elevated perioperative blood glucose levels and mortality, particularly among higher risk individuals, and morbidity, including sternal wound infection, hospital length of stay, and new onset atrial fibrillation. The results of observational studies comparing hyperglycemia and the incidence of morbidity and mortality among cohorts of cardiac surgery patients have been less consistent: there have been positive trials demonstrating such an association and negative trials that were unable to demonstrate associations between perioperative hyperglycemia and morbidity or mortality. These trials used a heterogeneous definition of hyperglycemia from 150 mg/dL to greater than 360 mg/dL, making their aggregate results somewhat difficult to interpret. Gandhi et al used logistic regression to demonstrate that a 20-mg/dL increase in intraoperative blood glucose level was associated with a 30% increase in negative perioperative outcomes.

The results of the few randomized trials of intraoperative blood glucose control in cardiac surgery have examined a diverse group of endpoints but have been generally negative. An early pilot study attempting to achieve normoglycemia during cardiopulmonary bypass demonstrated a significant association with postoperative hypoglycemia, even though the investigators were unable to achieve adequate blood glucose control intraoperatively. A larger study examining the effect of an insulin infusion to maintain a target blood glucose level of 100 mg/dL versus placebo was unable to demonstrate an association between decreased intraoperative blood glucose levels and neurocognitive events after cardiopulmonary bypass. One small randomized trial of 40 cardiac surgery patients compared a GIK solution to maintain a target blood glucose level of 100 to 180 mg/dL versus intravenous insulin boluses when blood glucose levels exceeded 180 mg/dL and demonstrated decreased blood glucose levels, improvement in lactate clearance, and a decreased requirement for postoperative inotropic support, although they provided no data on intraoperative blood glucose control. The largest randomized trial specifically designed to examine the effects of intraoperative blood glucose control on morbidity and mortality in cardiac surgery involved 400 patients randomly assigned to receive intensive insulin therapy with a target blood glucose level of 80 to 100 mg/dL intraoperatively (similar to the Leuven trial) versus intermittent intravenous insulin for blood glucose levels greater than 200 mg/dL. Although the investigators were unable to reach their blood glucose targets in their treatment group, and the study was unable to demonstrate an association between intraoperative hyperglycemia and a composite outcome of death and major morbidity, the investigators did note a trend toward higher rates of death ( p = 0.061) and stroke ( p = 0.02) in the intensive treatment group, findings which, although statistically insignificant and not part of their a priori hypothesis, do agree with those of the NICE-SUGAR investigators and support the conclusion that intensive blood glucose control as defined in the Leuven trial is associated with an unacceptable risk of major adverse outcomes in a variety of settings—in this case, intraoperative blood glucose control during cardiac surgery. Using these results and extrapolating from the conclusions of trials in critically ill patients, the Society for Thoracic Surgeons Practice Guidelines for Blood Glucose Management during Adult Cardiac Surgery currently recommend maintenance of blood glucose levels at less than 180 mg/dL intraoperatively and postoperatively, reserving a tighter postoperative threshold of less than 150 mg/dL for patients who are expected to be in the ICU for longer than 3 days.

Glycemic Control in the Setting of Acute Myocardial Infarction

Hyperglycemia is a common feature of MI, present in up to 50% of patients with an ST segment elevation MI. Cardiovascular stress from MI is a potent stimulus for release of the counter-regulatory hormones that act to increase circulating levels of glucose and free fatty acids. Hyperglycemia and hyperlipidemia are associated with increased generation of free radicals and oxidative damage and have been shown to be associated with an increase in inflammation and arrhythmias as well as decreased rates of successful thrombolysis in acute MI. Hyperglycemia leads to poorer 12-month survival rates in MI regardless of diabetic status, and there appears to be a dose-response relationship between increases in blood glucose level and mortality, at least among patients without diabetes. This relationship between increasing blood glucose levels and mortality has also been demonstrated prospectively in the CREATE-ECLA trial. However, determining whether hyperglycemia is simply a marker of more severe injury (i.e., an epiphenomenon) or is causally related to negative outcomes after MI requires further prospective evaluation.

The results of prospective trials of blood glucose control in MI have been mixed. One early trial demonstrated a significant reduction in 1-year mortality among patients randomly assigned to receive an insulin infusion with a modest target blood glucose level of approximately 120 to 180 mg/dL followed by subcutaneous insulin therapy. Early reports of GIK infusions in patients with MI, infused for cardiomyocyte metabolic substrate therapy without the intention of controlling blood glucose levels, indicated that insulin may have salutatory effects in MI other than blood glucose regulation. A randomized trial of insulin infusion to maintain blood glucose levels at less than 180 mg/dL for 24 hours after MI found no association between this therapy and the primary endpoint (heart failure or reinfarction); however, subgroup analysis showed that the mortality rate was lower among patients with mean blood glucose levels less than 144 mg/dL during the first 24 hours. A trial of intensive insulin therapy (blood glucose target, 72 to 108 mg/dL) compared with standard therapy (blood glucose target, 108 to 144 mg/dL) in survivors of out-of-hospital cardiac arrest was unable to demonstrate a difference in 30-day mortality rates, although, similar to many other studies, an increased incidence of moderate hypoglycemia in the intensive insulin arm was noted. The largest trial evaluating blood glucose control in MI was to be the 3000-patient DIGAMI 2 trial comparing three arms. The target blood glucose range of 126 to 180 mg/dL was initiated for the first 24 hours in the first two experimental groups. After 24 hours the first group continued at this target, while the second reverted to standard therapy administered by “local routines” (not defined). The third group had standard therapy throughout. The trial was stopped early after recruiting only 1253 patients because of slow enrollment and was unable to demonstrate a significant difference in its primary endpoint, mortality. The low recruitment, differences in baseline variables, lack of definition of standard therapy, and crossover between treatment groups make the results of this trial difficult to interpret. It is important to note, however, that among this study population hyperglycemia was still one of the most important prognostic predictors.

Perioperative Hyperglycemia and Neurologic Outcome after Brain Injury

Hyperglycemia is independently associated with the development of secondary brain injury and portends a poor clinical prognosis in stroke, subarachnoid hemorrhage, and traumatic brain injury. Neurologic injury is a potent stimulator of the stress response with activation of the sympathetic nervous system and release of counter-regulatory hormones that stimulate hyperglycemia. However, it is now evident that this hyperglycemia contributes to and propagates secondary brain injury through a variety of mechanisms, including synthesis of reactive oxygen species, promotion of intracellular acidosis, inflammation, and changes in endothelial cell function and nitric oxide metabolism, all of which favor a milieu of vasoconstriction and thrombosis. While hyperglycemia potentiates secondary brain injury, the brain relies almost exclusively on glucose for metabolism. Thus the injured brain, with a compensatory increase in glucose demand, is particularly sensitive to the effects of hypoglycemia. This has been demonstrated in patients who have sustained a stroke in which there is a J -shaped association between blood glucose levels and mortality. Similarly, in patients with subarachnoid hemorrhage, moderate hypoglycemia is associated with vasospasm and cerebral infarction, and in traumatic brain injury, evidence demonstrates a significant imbalance in energy substrate and requirements when intensive insulin therapy is used.

Although it is clear that both hyperglycemia and hypoglycemia can potentiate acute brain injury by a variety of mechanisms, clinicians have only sparse data from prospective trials to help guide them in the management of blood glucose levels in the brain-injured patient. The majority of prospective trials have been conducted in stroke patients and includes one large trial and several pilot studies, all evaluating the effects of intensive insulin therapy on outcomes in this population. The largest of the trials, the Glucose in Stroke Trial (GIST) randomly assigned 933 patients with stroke and hyperglycemia on admission to receive intensive blood glucose control (target, 72 to 126 mg/dL) with the use of a GIK infusion compared with normal saline infusion in control subjects. The trial was stopped well short of its calculated 2355 patient sample size at 933 patients because of slow enrollment and was unable to demonstrate an effect on either morbidity or mortality. Importantly, however, the investigators did demonstrate in post-hoc subgroup analysis that there was a significant increase in mortality rates among patients in the GIK group that had a greater than 36-mg/dL decrease in blood glucose levels. Several small pilot studies have evaluated the feasibility of tight glycemic control using intensive insulin therapy in stroke patients and, although each individually was underpowered to detect a mortality difference, a recent Cochrane review has attempted to consolidate their findings with those of the GIST trial. The Cochrane review concluded that maintenance of tight glycemic control did not provide any benefit in terms of morbidity or mortality and exposed patients to a greater risk of hypoglycemia. Much less data exist from prospective randomized trials regarding blood sugar control in other types of neurologic injury. In a small, randomized trial after aneurysm clipping in patients with subarachnoid hemorrhage, it was shown that intensive insulin therapy (target blood glucose level, 80 to 120 mg/dL) was associated with decreased rates of postoperative infection, but no difference was seen in the rates of vasospasm, mortality, or neurologic recovery. Importantly, the investigators did not report rates of hypoglycemia in the two groups other than to note that 10.5% of the intensive insulin therapy group had blood glucose values below 80 mg/dL versus 3.5% in the control group.

Although there is sufficient evidence to state that hyperglycemia worsens secondary neurologic injury in stroke, subarachnoid hemorrhage, and traumatic brain injury, analysis of available evidence does not provide any clear guidance as to appropriate blood glucose targets in this patient population. Additionally, the significant risks of further neurologic injury from hypoglycemia and the lack of evidence of efficacy from tight glycemic control would argue against adopting intensive forms of insulin therapy in this group.