Introduction
Critical illness is associated with many endocrine and metabolic changes, including changes in the glucose homeostasis [
1‐
7]. Both hypoglycemia and hyperglycemia may lead to adverse outcome as expressed in length of pediatric intensive care unit (PICU) stay and mortality rates [
6‐
16].
A follow-up study in patients who survived meningococcal septic shock in childhood showed that severe mental retardation was associated with hypoglycemia during admission [
17]. Children who died from meningococcal septic shock appeared to have significantly lower levels of blood glucose on admission to the PICU in comparison with those who survived, in whom levels were moderately increased [
4,
5]. The most severely ill children had signs of (relative) adrenal insufficiency on admission. Deficiency of substrate, reduced activity of adrenal enzymes because of endotoxins, cytokines, or medication, and shock with disseminated intravascular thrombosis can cause necrosis of the adrenal glands and result in (relative) adrenal insufficiency in children with meningococcal disease [
5].
Many children with meningococcal septic shock suffer from hyperglycemia [
12,
18,
19]. The pathophysiological mechanism leading to hyperglycemia in critically ill children with meningococcal disease may be different from that in adults. Recently, it was shown that the acute phase of sepsis in children is quite different from that in adults [
18]. It was suggested that hyperglycemia associated with β-cell dysfunction rather than insulin resistance may be the normal pathophysiological response in children with meningococcal septic shock. It was also suggested that treatment of hyperglycemia with exogenous insulin may not be supportive and may even be potentially detrimental in critically ill children [
18].
Better insight into pathophysiological mechanisms leading to hyperglycemia is crucial to improve treatment strategies. The gold standard for quantifying insulin sensitivity
in vivo is the hyperinsulinemic euglycemic clamp technique [
20]. This is a complex and invasive technique and therefore is not easily applied in studies with critically ill children. The search for uncomplicated and inexpensive quantitative tools to evaluate insulin sensitivity has led to the development of other assessments. The fasting glucose-to-insulin ratio and homeostasis model assessment (HOMA) of insulin resistance have been proven to be useful estimates of insulin sensitivity, also in critical illness [
21‐
24]. There is a good correlation between estimates of insulin resistance derived from HOMA and from the hyperinsulinemic euglycemic clamp [
24]. The assessment of β-cell function is difficult because the β-cell response to the secretory stimuli is complex. There is no gold standard for β-cell function. The HOMA method for assessing β-cell function (HOMA-%B) is based on measurements of fasting insulin or C-peptide concentration to calculate pre-hepatic insulin secretion in relation to blood glucose levels [
24]. The objective of the present study was to investigate the occurrence of hyperglycemia in relation to the insulin response and exogenous factors, such as glucose intake and drug use, in a homogenous group of critically ill children with meningococcal sepsis or meningococcal septic shock or both.
Discussion
Thirty-three percent of all children in the present study were hyperglycemic on admission, and one child was hypoglycemic. Blood glucose levels in shock and sepsis survivors were higher than in shock non-survivors. Hyperglycemic children had significantly higher insulin and C-peptide levels in comparison with normoglycemic children. HOMA showed a predominance of insulin resistance in hyperglycemic children, although β-cell insufficiency or a combination of insulin resistance and β-cell insufficiency was also seen. Multiple regression analysis revealed that both age and plasma insulin levels on admission were significantly related to blood glucose level.
Hyperglycemia is a common finding in critically ill children, and our results are in line with those of previous studies [
8,
11,
14]. Whereas others have reported an association between hyperglycemia and mortality [
8‐
14], we showed, in the present study, that shock non-survivors had the lowest blood glucose levels. This study concerns children with meningococcal sepsis and septic shock, whereas the other studies included children with mixed diagnoses. Only Branco and colleagues [
12] studied children with septic shock (various causes) and showed that a peak glucose level of greater than 9.8 mmol/L was independently associated with an increased risk of death (relative risk of 2.59).
In our study, insulin levels on admission were the lowest in children who did not survive and were closely related to the low blood glucose levels. The association between a lower blood glucose level on admission and mortality in the present study might be explained by the specific features of meningococcal disease, like the high risk for relative adrenal insufficiency [
5]. This could also explain the positive correlation between blood glucose levels and age, as the youngest children showed the highest mortality rate in combination with the lowest blood glucose levels on admission. Previously, we showed that the concomitant use of therapeutic drugs such as etomidate, which was used in almost half of the studied children, influenced blood glucose levels as well [
5]. In accordance with previous findings, children intubated with etomidate showed lower glucose and cortisol levels on admission in comparison with those without etomidate. Hyperglycemia was associated with elevated insulin levels in half of the children. HOMA showed that insulin resistance as well as β-cell dysfunction resulting in a hypoinsulinemic response resulted in hyperglycemia. Insulin resistance, caused by high levels of counter-regulatory hormones and cytokines, oxidative stress, and therapeutic interventions (such as glucocorticoid and catecholamine administration), is the main pathophysiological mechanism of hyperglycemia in critically ill patients [
32].
Concerning therapeutic interventions, glucocorticoid and catecholamine use in insulin-resistant hyperglycemic children was more frequent than in those without insulin resistance. However, the numbers were too small to detect significant differences. Cortisol level on admission was positively correlated with plasma glucose level in children without previous glucocorticoid treatment, indicating that endogenous cortisol release is a causative factor for hyperglycemia. Sepsis guidelines recommend glucocorticoids for the treatment of vasopressor-dependent septic shock [
15]. Glucocorticoids stimulate hepatic glucose production, mainly by mobilizing substrate for hepatic gluconeogenesis and activation of key hepatic gluconeogenic enzymes. Furthermore, glucocorticoid excess reduces glucose uptake and utilization by peripheral tissues, owing in part to direct inhibition of glucose transport into the cells [
33]. Hyperglycemic episodes were more common in adult septic shock patients who received hydrocortisone in bolus therapy as compared with those who received a continuous infusion with an equivalent dose [
34]. This important side effect of glucocorticoid treatment has not yet been addressed in studies in critically ill children.
Another important causative factor of hyperglycemia might be the amount of glucose intake. In the present study, children were considered to be fasting on admission, because they received only a continuous glucose infusion without enteral intake. Glucose intake did not differ between normoglycemic and hyperglycemic children. In critically ill adults, an association between hyperglycemia and a high glucose infusion rate (greater than 5 mg/kg per minute) was shown [
35]. On the other hand, low-caloric parenteral nutrition in adult surgical trauma patients resulted in fewer hyperglycemic events and lower insulin requirements [
36]. Maximum glucose oxidation rates in severely burned children approximate 5 mg/kg per minute [
37]. Exogenous glucose in excess of this amount enters non-oxidative pathways and is unlikely to improve energy balance and lipogenesis and may result in hyperglycemia [
38,
39].
Two studies have suggested that a hypoinsulinemic response in critically ill children might result in hyperglycemia [
18,
40]. First, van Waardenburg and colleagues [
18] studied 16 children with meningococcal disease on the third day of admission (10 shock survivors and 6 sepsis survivors). Whereas most children were normoglycemic, shock survivors had lower insulin levels (50 pmol/L) and insulin-to-glucose ratios (8 pmol insulin per mmol glucose) in comparison with sepsis survivors (130 pmol/L and 24 pmol insulin per mmol glucose, respectively), suggesting normal or enhanced insulin sensitivity in shock survivors. Second, Preissig and Rigby [
40] showed relatively low C-peptide levels (1.5 nmol/L, 4.4 ng/mL) within 48 hours after admission in hyperglycemic critically ill children with respiratory and cardiovascular failure. Accordingly, the present study also showed relatively low C-peptide levels for shock survivors and sepsis survivors during admission (1.0 to 1.7 nmol/L, 3.0 to 5.1 ng/mL). HOMA-%B based on paired C-peptide, insulin, and glucose levels showed β-cell dysfunction of the pancreas in 38% of hyperglycemic children who were either shock or sepsis survivors. The cause of pancreatic dysfunction could have many factors, including elevations in pro-inflammatory cytokines, catecholamines, and glucocorticoids. It was hypothesized that β-cells become dysfunctional if physiological changes occur acutely. When the same changes occur more gradually, this might allow β-cells to adapt and function at supraphysiological levels over time, resulting in insulin resistance. Also, β-cell exhaustion is a known phenomenon characterized by an ability to increase secretion up to a certain level and thereafter fail in response to further demand.
Finally, proinflammatory cytokines are important mediators of the hyperglycemic stress response. We did not find correlations between cytokines and insulin levels or HOMA-%S in hyperglycemic children, presumably because of the relatively small sample size.
Forty-eight hours after admission, the percentage of children with hyperglycemia had decreased from 33% to 8% without insulin therapy. In contrast, in critically ill adult patients, hyperglycemia may persist for days to weeks with or without insulin therapy [
41]. This difference might be due to the rapid resolution of the acute stress response that is seen in severely ill children with meningococcal disease [
5]. The present data also show that the elevated cortisol and cytokine levels on admission decrease to normal values within 24 hours.
There are several limitations to this study. The hyperinsulinemic euglycemic clamp technique is the 'gold standard' for quantifying insulin sensitivity
in vivo because it directly measures the effects of insulin to promote glucose utilization under steady-state conditions. It is not easily implemented, however, in large studies with critically ill children. In the present study, therefore, insulin sensitivity was indirectly assessed by investigating the insulin response to glucose and by HOMA. Diabetes studies and epidemiological studies on glucose tolerance have frequently used HOMA, and recent reports have shown its value for assessment of insulin sensitivity in the critically ill [
22,
23]. Nevertheless, as we are the first to use HOMA analysis to describe insulin resistance and β-cell dysfunction in critically ill children, there are no control data for HOMA for sick children and we have to be careful in our conclusions. Under basal conditions, the product of β-cell responsivity and insulin sensitivity is assumed to be a constant, and different values of tolerance are represented by different hyperbolas [
42]. We have shown that, in critically ill children with impaired glucose tolerance, β-cells can be dysfunctional, resulting in an inadequate compensatory increase in insulin release to the decreased insulin sensitivity.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JV performed literature searches and statistical analysis and wrote this paper under the direct supervision of KJ. MdB participated in the coordination of the study and carried out the data collection. AH-K participated in the design of the study and helped to edit and revise the paper critically. JH participated in the design and coordination of the study and helped to draft the manuscript. KJ conceived of the study, participated in its design and coordination, and helped to draft the manuscript. All authors read and approved the final manuscript.