September 24, 2002

Glucose Control in the Acutely Ill

Intensive insulin therapy in critically ill patients.
Van den Berghe G, Wouters P, Weekers et al. N Engl J Med. 2001; 345:1359-1367.

Commentary by Douglas Coursin, M.D.

The incidence of diabetes mellitus (DM) in the United States continues to grow at an alarming rate with a doubling of the number of diabetics in the US over the past decade. Approximately 9% of Americans (20 million) are diabetic. This may be related, in part, to increasing age, inactivity and chronic compromise of the population, but DM also occurs in a greater number of younger patients, not all of whom are inactive or obese [1].

Despite continued therapeutic advances, particularly with newer insulin preparations, oral agents and combination therapy, diabetes remains a significant cause of acute and long-term sequelae. Data from large multicenter, randomized trials over the past 10 years demonstrate the benefits of tighter glucose control in limiting microvascular damage of the eye, nerves and kidney [2-4]. However, long-term improved control of glucose does not appear to alter progression of macrovascular pathology nor lower diabetic mortality [2,4].

There is a growing body of literature on preventing the development of DM and slowing the progression of cardiac and renal pathology in diabetics through the chronic administration of angiotensin converting enzyme inhibitors (ACEIs) [5]. There is also a body of data reporting the acute and long-term benefits of glucose control in diabetic patients sustaining an acute myocardial infarction (AMI). Over the past 10 years or so, the DIGAMI (diabetes and insulin-glucose infusion in acute myocardial infarction) study group from Scandinavia reported that tighter control of glucose for the first 24 hours or so after AMI had a positive effect on survival. This was associated with a prolonged effect for up to 3.4 years after infarction [6]. This sustained benefit is probably polyfactorial and not uniquely reflective of early glucose control.

Potentially sentinel trial
This past fall, Van Den Berghe, et al. reported an interesting single-center prospective, randomized, controlled trial from Belgium that described impressive benefits in a critically ill population that had tighter glucose control during its ICU course. The basis for what may become a sentinel trial was the observation that hyperglycemia and insulin resistance are common in the critically ill and potentially harmful to these patients. Since hyperglycemia occurs frequently in non-diabetic patients in the ICU, the authors queried whether normalization of blood glucose levels with insulin improved outcome for diabetic and non-diabetic hyperglycemic patients.

1548 surgical ICU patients were randomly assigned to receive intensive insulin therapy (study group goal was maintenance of blood glucose between 80 and 110 mg/dL) or conventional insulin therapy (goal blood glucose between 180 and 200 mg/dL). All of the patients were mechanically ventilated and a large percent were s/p cardiac surgery (~63% in each treatment group). Only 13% of the patients were known diabetics (5% were on insulin, 8% were on oral agents) at the time of inclusion in the study.

The patients were well matched in regards to demographics, severity of illness, presence of diabetes and ICU management. Patients in the intense glucose therapy group had significantly decreased:

1. ICU mortality (8% vs. 4.6%) if they remained in the ICU for > 5 d
2. hospital mortality
3. ICU and hospital length of stay
4. duration of mechanical ventilation
5. incidence of multiple organ failure, renal failure, infection, hyperbilirubinemia, and received fewer transfusions

Study caveats
Some caveats about interpreting the study - first, the investigators were not able to completely maintain a blind since glucose levels were measured regularly and goals for management were known while the study was on-going. The authors attempted to alleviate some of this potential bias by separating investigators from direct individual patient care. Secondly, the population studied was overwhelmingly surgical (two thirds underwent cardiac surgery) and not very sick. Additional data are needed before these findings are extrapolated to a wider and potentially sicker medical ICU patient population.

Significant implications
Nonetheless, this study has significant potential implications for anesthesia and intensive care providers if the findings are accurate and can be generalized to a wider population. We need to pay attention to blood glucose even in non-diabetics since normalization of blood glucose appears to be good and not overly risky. About 5% of patients in the intensely treated group developed hypoglycemia as defined by a blood sugar less than 70 mg/dL, but none had major complications from hypoglycemia. With the advent of newer continuous non-invasive glucose monitoring techniques-integrated systems that allow repetitive measurement of glucose and rapid adjustment of insulin controller devices-improved glucose control should result.

The impact of glucose normalization
Why should normalization of glucose have such impact? The answer to this question remains unknown or incompletely understood, but experts suggest that the benefits are related to both elimination of some of the deleterious effect of hyperglycemia itself and a positive humeral and anti-inflammatory effect of supplementation with insulin. It is well described that glucose levels above approximately 220 mg/dL result in compromised white cell aggregation, chemotaxis, killing and overall function. These levels are also associated with impaired fibroblast function and impaired wound healing, which may also result in increased nosocomial infection. Patients who develop such infections are at greater risk for resistant infection, which results in prolonged hospitalization, cost and mortality. Hyperglycemia results in osmotic diuresis and may lead to dehydration, decreased renal function, hemoconcentration and increased thromboembolic events. Some of these deleterious sequelae of hyperglycemia can be ameliorated with insulin therapy and maintenance of improved glucose control.

Furthermore, insulin has been shown in animal and human trials to modulate several key pro-inflammatory mediators, particularly blunting the production or action of tumor necrosis factor-alpha (TNF-a) and macrophage migration-inhibitory factor (MMIF) [7,8]. TNF is a major cytokine mediator in systemic inflammation and sepsis. Among other things, it has negative effects on cardiac contractility, peripheral vascular function, and interacts with a host of other immunomodulating cytokines and chemokines. TNF may be increased in AMI, and the anti-TNF effect of glucose/insulin/potassium infusions is hypothesized to be an important mechanism in improving the outcome after MI.

The anti-MMIF effect of insulin may blunt some of the deleterious effects associated with hyperglycemia in patients with systemic inflammatory responses and/or sepsis. Some hypothesize that insulin also may have salutary effects in other chronic inflammatory processes such as inflammatory bowel diseases and connective tissue pathologies.

However, far more experience and data need to be gathered to guide clinicians in the optimal use of glucose and insulin therapy. Higher levels of glucose and insulin may be associated with electrolyte abnormalities, particularly with potassium and phosphorus, and increased hepatic uptake of glucose that over time may lead to fatty infiltration of the liver. Very high plasma insulin levels may result in excessive norepinephrine release and activation of the sympathoadrenal response to surgery.

It seems reasonable to take a more aggressive approach to perioperative glucose control using regular insulin infusions, electrolyte supplementation and careful glucose monitoring. Hopefully, additional studies will provide information on the wider applicability of "tight" glucose control and help elucidate the mechanisms of the apparent benefit of insulin therapy in hyperglycemic, acutely ill patients, whether they are diabetic or not.