Pearls
- •
Cortisol is a key mediator of the stress response—influencing immunity, metabolism, and modulating the transcription of perhaps 25% of the entire genome.
- •
Relative adrenal insufficiency and critical illness–related corticosteroid insufficiency are poorly understood concepts in terms of both pathophysiology and therapeutic intervention.
- •
Critical illness hyperglycemia is the result of inflammation-mediated increased endogenous glucose production and decreased utilization secondary to insulin resistance, the latter of which promotes catabolism and lipolysis, ultimately leading to lipotoxicity, glucotoxicity, and further inflammation.
- •
Sick euthyroid syndrome, common among critically ill patients, is characterized by a rapid decrease in T3 and variable increase in rT3 that appears to be proportional to the intensity of illness severity and concentration of tumor necrosis factor-α.
The endocrine system is closely aligned with the neurogenic and inflammatory systems, particularly as an aspect of the stress response. Multiple servo feed-forward and feed-back signals normally provide precise control over this integrated system. There are several important primary endocrine emergencies that the pediatric intensivist must understand, as their monitoring and treatment require critical care. Additionally, critical illness of all types and their treatment frequently generate unique secondary endocrinopathies affecting the multiple endocrine axes. While these acute responses of the endocrine axes are considered an evolutionarily driven adaptive response to stress, chronic hormonal changes are likely the result of advances in critical care and are the basis of much intensive care unit (ICU)-related morbidity.
Hypothalamic-pituitary-adrenal axis
A schematic overview of the classic regulation of cortisol synthesis and secretion is provided in Fig. 84.1 . A variety of stimuli converge on the paraventricular nucleus of the hypothalamus, resulting in release of corticotropin-releasing hormone (CRH). CRH is transported via hypophyseal portal capillaries to the anterior pituitary, facilitating the production of pro-opiomelanocortin, a peptide that includes primary protein sequences for adrenocorticotropic hormone (ACTH), β-lipotropin, β-endorphin, and melanocyte-stimulating hormone. ACTH is then transported by the blood to the zona fasciculata of the adrenal gland, where it binds to type 2 melanocortin receptors (MC2R). MC2R is a seven-transmembrane domain protein coupled with G proteins that uses intracellular cyclic adenosine monophosphate signaling to modulate adrenal steroidogenesis. Single nucleotide polymorphisms in MC2R may be responsible for some of the variability in cortisol response to pediatric critical illness, and familial cortisol deficiency—a lethal condition associated with hypoplastic adrenal glands—is caused by another mutation in the MC2R gene.
During critical illness, activation of the hypothalamic-pituitary-adrenal (HPA) axis frequently reflects activation of the systemic inflammatory response syndrome (e.g., by interleukin-6). , Interleukins 1, 2, and 6 are generally thought to stimulate cortisol production, whereas tumor necrosis factor–alpha (TNF-α), macrophage inhibitory protein (MIP), and corticostatin, a protein with anti-ACTH activity, are generally considered to be inhibitory for cortisol production.
Cortisol biochemistry and biology
Within the adrenal cortex, several enzymes lead to the conversion of cholesterol to the steroid hormones cortisol, aldosterone, and androstenedione. Normal adrenal production of cortisol is equivalent to hydrocortisone administration of approximately 8 to 12 mg/m per day. Stress production of cortisol may reach 200 to 300 mg/day, resulting in a plasma total cortisol concentration occasionally exceeding 60 µg/dL. In addition, reduced cortisol breakdown, related to suppressed expression and activity of cortisol catabolic enzymes, contributes significantly to hypercortisolemia and associated ACTH suppression. , If the latter is maintained during prolonged critical illness, adrenal lipid depletion and reduced ACTH-regulated gene expression may occur in conjunction with adrenal atrophy and insufficiency. Hypercortisolemia in critical illness is correlated with severity of illness. Mean and range of plasma cortisol concentrations as a function of age have been reported for normal children and may be found in Table 84.1 .
| 2 h | 7 d | 2 wk–3 mo | 3 mo–1 y | 1–3 y | 3–5 y | 5–7 y | 7–11 y | 11–15 y | |
|---|---|---|---|---|---|---|---|---|---|
| Mean | 6.80 | 1.14 | 3.34 | 6.34 | 7.52 | 7.37 | 7.83 | 7.98 | 6.41 |
| Range | 1.25–29.80 | 0.13–9.64 | 0.92–8.68 | 2.12–17.30 | 1.71–13.70 | 3.24–12.98 | 3.98–12.90 | 3.54–18.40 | 2.17–17.40 |
| N | 12 | 20 | 11 | 23 | 20 | 17 | 19 | 27 | 25 |
Although the plasma half-life for cortisol ranges from 80 to 115 minutes, biological duration of action of cortisol is approximately 8 hours. A diurnal rhythm of cortisol production is noted among healthy (but not critically ill) patients with peak cortisol production typically from 8 am to 9 am and a nadir of cortisol production typically around midnight. Cortisol exhibits mineralocorticoid activity approximately 1% that of aldosterone. Cortisol is transported from the adrenal gland to various tissues via cortisol binding proteins, namely, transcortin with high affinity but low capacity and albumin with low affinity but high capacity. Cortisol concentration is locally increased at sites of inflammation through degradation of transcortin by neutrophil elastase, as well as local upregulation of 11β-hydroxysteroid dehydrogenase, an enzyme key to cortisol synthesis.
Cortisol diffuses through the plasma membrane binding to the glucocorticoid receptor (GCR). Commonly occurring GCR polymorphisms may explain individualized patient responses to corticosteroids. Children with septic shock demonstrate a transient depression of GCR messenger ribonucleic acid (mRNA) in their neutrophils that may reflect an adaptive cortisol resistance response. On the other hand, septic patients may demonstrate a transient increased or decreased expression of GCR on mononuclear cells. GCRs represent a potential drug target for treatment of severe inflammation and the consequences of excess endogenous or exogenous corticosteroid.
Ultimately, cortisol modulates the transcription of thousands of genes, perhaps 25% of the entire genome. , Although the majority of corticosteroid action is related to changes in gene transcription mediated through chromatin remodeling, corticosteroids also affect protein synthesis by decreasing the stability of mRNA. A number of inflammatory proteins appear to be regulated by this posttranscriptional mechanism. Even with appropriate cortisol production by the adrenal gland, cortisol resistance can occur by multiple mechanisms, including depletion of corticosteroid-binding globulins, activation of 11β-hydroxysteroid dehydrogenase, decreased glucocorticoid receptor density and activity, and elevated antiglucocorticoid compounds and receptors.
Actions of cortisol
Cortisol affects three general areas of physiology: immunity, metabolism, and hemodynamics.
Immunity
Much of critical illness may involve disturbance of the balance between systemic inflammatory responses versus compensatory antiinflammatory responses. , , As depicted in Fig. 84.2 , cortisol biochemistry represents a key regulatory mechanism for virtually all aspects of antiinflammation.
In this respect, cortisol increases the synthesis of IκB, trapping nuclear factor-κB (NF-κB) in the cytoplasm and effectively thwarting synthesis of a variety of proinflammatory mediators. In addition, cortisol increases the production of annexin, which inhibits phospholipase A 2 , resulting in modulation of both cyclooxygenase and lipoxygenase inflammatory lipid pathways. Sepsis induces widespread immunosuppression. , Among children with sepsis who receive corticosteroids, further repression of gene networks related to adaptive immunity has been reported.
Metabolism
A schematic summary of the effects of cortisol on metabolism is displayed in Fig. 84.3 .
As a key mediator of the stress response, cortisol facilitates lean muscle catabolism to provide substrate for hepatic gluconeogenesis, synthesis of acute-phase reactants, and expansion of the immune system. Hypercortisolemia represents a key mediator in hyperglycemia of critical illness discussed later. Both protein catabolism and hyperglycemia, if prolonged in the ICU, can become maladaptive and associated with increased risk for adverse outcomes, including death.
Hemodynamics
Cortisol’s permissive effects on maintaining a normal hemodynamic status include augmenting cardiac contractility, maintaining vascular tone, promoting endothelial integrity, upregulating vasoactive receptors, and increasing catecholamine synthesis. , , Specifically, cortisol suppresses inducible nitric oxide synthetase, leading to increased vascular tone and blood pressure. Additionally, high intraadrenal concentrations of cortisol during critical illness induce the cascade of enzymatic activity that produces epinephrine via methylation of norepinephrine by phenyl-ethanolamine N-methyltransferase.
Assessing the cortisol stress response
Historically, adrenal function during critical illness has been quantified by random total plasma cortisol concentrations or by calculating the difference between a corticotropin-stimulated plasma cortisol concentration minus a baseline cortisol concentration. In the latter case, a so-called delta value less than 9 µg/dL was previously considered evidence of adrenal insufficiency (AI) or inadequate adrenal reserve. Depending on the cutoff value chosen (10, 15, 18, 20, 25 µg/dL), critically ill patients may demonstrate a wide range of adrenal insufficiency occurrence when random baseline total plasma cortisol concentrations are evaluated.
A detailed evaluation of critically ill adults with sepsis indicated that those patients with a high baseline plasma cortisol concentration but delta cortisol concentration less than 9 µg/dL exhibited the highest risk for hemodynamic instability and mortality. A subsequent interventional trial of adjunctive hydrocortisone for adult septic shock was designed on the basis of these findings. However, a confirmatory trial failed to replicate this finding and concluded that corticotropin stimulation testing provided no information regarding which patients with septic shock would benefit from hydrocortisone replacement therapy.
Serial low-dose corticotropin stimulation testing in addition to comprehensive ACTH, dehydroepiandrosterone, and cytokine measurements in a cohort of critically ill Turkish children revealed that 28% exhibited AI that was largely resolved by 2 weeks. Baseline cortisol levels of the patients were significantly higher than those of healthy children. Patients with multiple-organ dysfunction syndrome (MODS) had significantly higher concentrations of baseline and stimulated cortisol, as well as higher procalcitonin, TNF-α, and interleukin-6 (IL-6) concentrations compared with those without MODS. Cortisol, ACTH, and dehydroepiandrosterone concentrations were higher among the children with AI compared with those without AI. Interestingly, sepsis as an antecedent of critical illness was not associated with an increased risk of AI. Similarly, among 381 critically ill Canadian children also evaluated by low-dose corticotropin stimulation testing, AI was noted in 30.2%. Children with AI exhibited higher baseline cortisol concentrations, were significantly older, and demonstrated an increased need for volume resuscitation and vasoactive-inotropic support.
Free cortisol
Although most cortisol is bound to transcortin or serum albumin, the free fraction comprising 10% to 15% of the total is actually responsible for the protean effects of cortisol. Accordingly, it has been suggested that perhaps free cortisol rather than total cortisol concentrations might be more reliable in terms of identifying a population that would most benefit from cortisol replacement therapy. For most critically ill patients, cortisol-binding globulin is typically decreased and the percent of cortisol as the free fraction increased. Moreover, with corticotropin adrenal stimulation, free cortisol increases substantially more than total cortisol. Assessing total cortisol concentrations may be especially problematic in critically ill patients with low albumin. In an investigation that measured both total and free cortisol concentrations in a general population of critically ill children, the majority exhibited low total and free cortisol concentrations. However, none of these children demonstrated clinical evidence of corticosteroid insufficiency (hypotension, hyponatremia, hypoglycemia). These results question the use of current thresholds for assigning diagnoses of AI or critical illness–related corticosteroid insufficiency in critically ill children. The results further suggest that clinicians currently are unable to reliably define adequacy of the adrenal response to the stress of critical illness either with total or free cortisol measurements.
Adrenal insufficiency in the intensive care unit
AI may be classified under two major categories: primary, in which direct malformation or destruction of the adrenal glands occurs, and secondary, in which there is typically loss of HPA axis integrity. The latter situation is most often encountered among critically ill patients.
Primary adrenal insufficiency
Primary adrenal insufficiency, or Addison disease, is more typically encountered in adult patients and is included in a group of disorders termed autoimmune adrenalitis . , Signs and symptoms of Addisonian crisis include intercurrent illness with a history of chronic weight loss and anorexia, dizziness, lethargy, and chronic pigmentation. Symptoms more likely to be associated with admission to the ICU include sudden hypovolemic shock, hyperkalemia, vomiting, diarrhea, abdominal pain, and coma. Other causes of primary adrenal failure include congenital adrenal hyperplasia, an autosomal recessive disorder, with approximately 90% of cases secondary to 21 hydroxylase deficiency ( CYP21A2 mutation). , Primary congenital adrenal failure can also occur among infants with adrenoleukodystrophy associated with a metabolic defect in metabolism of very-long-chain fatty acids. Other causes of congenital adrenal failure include Wolman disease and familial unresponsiveness to ACTH involving altered MC2R , as previously discussed.
Because of its precarious circulation associated with a subcapsular arteriolar plexus, the adrenal glands are subject to hemorrhage and infarction, particularly in the setting of septic shock. Such events, termed the Waterhouse-Friderichsen syndrome , were initially described as adrenal apoplexy. ,
Although a variety of drugs inhibit the sequential steps in cortisol synthesis, etomidate is particularly notable, as it is increasingly being used as a sedative to facilitate endotracheal intubation without adversely affecting hemodynamics. A single bolus of etomidate suppresses 11β-hydroxylase activity, one of the critical enzymes in the cortisol synthetic pathway. , A meta-analysis of seven studies concluded that administration of etomidate for endotracheal intubation was associated with higher rates of AI and mortality among patients with sepsis. Further studies, however, have not confirmed consistently worse outcomes with its use. Given the known effects of etomidate on cortisol production, and until further definitive prospective studies have been performed, the authors recommend empiric administration of stress-dose hydrocortisone for 48 hours after etomidate use in patients with septic shock.
Secondary adrenal insufficiency
Secondary AI in the pediatric ICU (PICU) can be seen with pituitary disorders (e.g., among children following surgical resection of a craniopharyngioma). In addition, long-term corticosteroid administration among patients with recalcitrant asthma, patients with oncologic or rheumatologic diagnoses, or transplantation patients results in suppression of ACTH release, occasionally leading to adrenal atrophy. ,
Probably the most controversial cause of secondary adrenal failure seen in the ICU is so-called relative adrenal insufficiency or critical illness–related corticosteroid insufficiency (CIRCI). A seemingly inadequate adrenal response relative to the magnitude of stress characterizes CIRCI. This situation is a dynamic, typically reversible state that is thought to be secondary to both decreased cortisol production and tissue resistance to cortisol, as discussed previously. Multiple pediatric observational studies, most associated with sepsis, have examined associations with random baseline serum cortisol concentrations with various outcomes. , These studies ascertain cortisol circadian rhythm among children with sepsis. As sepsis severity increased (e.g., sepsis → septic shock → sepsis death), proinflammatory mediator concentrations (e.g., IL-6, TNF-α) and ACTH increased, while cortisol concentrations decreased. Both serum cortisol and ACTH concentrations correlated with illness severity per the Pediatric Risk of Mortality study, organ dysfunction scores, lactate, and C-reactive protein. Pediatric observational studies have also examined corticotropin stimulation testing among children with severe sepsis. These studies in general demonstrated that, like adults, low delta cortisol is common among children with sepsis. Among children with a low delta cortisol, illness severity was higher, as was requirement for vasoactive-inotropic resuscitation. Such children also more frequently demonstrate vasoactive-inotropic resistance shock and MODS. Chronic illness and the degree of organ dysfunction at presentation, as well as low delta cortisol concentration, predicted risk of mortality.
Treatment of adrenal insufficiency
Guidelines for steroid replacement in the acute and long-term management of primary AI are well established with utilization of stress-dose hydrocortisone during times of illness and replacement of maintenance glucocorticoids and mineralocorticoids when well. , The 2017 American College of Critical Care Medicine Clinical Practice Parameters for Hemodynamic Support of Pediatric and Neonatal Septic Shock additionally recommend use of hydrocortisone in children with catecholamine-resistant septic shock at risk for absolute AI, such as those with purpura fulminans, chronic steroid exposure, sedation with etomidate, and congenital adrenal hyperplasia. The same group did not make recommendations for how or when to treat CIRCI. Treatment of CIRCI remains controversial, as the diagnosis of CIRCI remains controversial.
Two recent adult randomized controlled trials again looked at mortality outcomes after use of steroids in adult patients with septic shock. The ADRENAL study found no difference in mortality among intubated adults with septic shock treated with a continuous infusion of hydrocortisone compared with those who did not received hydrocortisone. The APROCCHSS trial reported lower 90-day mortality in adults with septic shock who received hydrocortisone plus fludrocortisone compared with those who received placebo. Multiple pediatric observational cohort studies have shown no benefit or possible harm with adjunctive corticosteroids used in pediatric septic shock. No large prospective pediatric randomized clinical trial has yet to examine this issue, although one is currently underway at the time of publication of this chapter.
Corticosteroid side effects
Main side effects of corticosteroid administration include immunosuppression and subsequent increased risk for secondary infection, protein catabolism with resultant myopathy, hyperglycemia, and hypernatremia.
Corticosteroid therapy has been associated with increased risk for hospital-acquired infection among a general population of critically ill children and in children following surgery for congenital heart disease. In the latter investigation, children who received hydrocortisone were nearly 30 times more likely to develop a central catheter–associated infection as compared with children who did not receive hydrocortisone. With increasing recognition of the key role of immunosuppression in the pathogenesis of sepsis, it is noteworthy that children with sepsis demonstrate widespread repression of gene programs associated with adaptive immunity and that this gene repression is further enhanced with concurrent corticosteroid administration. Even if administering only a single dose of a corticosteroid, it is important to recognize that such a dose alters the expression of about 25% of the human genome. ,
Dissolution of lean body mass mediated by both endogenous and exogenous corticosteroids represents a key element of the stress response. Although this may be beneficial in the short term, prolonged corticosteroid-mediated muscle catabolism can be associated with ICU weakness (including the diaphragm) and hyperglycemia. Muscle weakness has been associated with prolonged mechanical ventilation weaning. , Hyperglycemia has been associated with a variety of adverse events in the PICU, as detailed later. Additional important clinical consequences of protein catabolism that may be exaggerated by exogenous corticosteroid administration include impaired wound healing, hypoalbuminemia, disordered coagulation, and impaired gut function with bacterial translocation.
In the CORTICUS sepsis interventional trial, transient hypernatremia was also noted among patients receiving hydrocortisone. Although gastrointestinal hemorrhage represented an important side effect in previous clinical trials examining the potential utility of high-dose methylprednisolone as adjunctive therapy in sepsis, this side effect has not been problematic in later investigations using low-dose hydrocortisone. Among adult critically ill patients with acute lung injury, corticosteroid administration is associated with transition to delirium. However, among critically ill adults with mixed diagnoses in a separate large prospective study, systemic corticosteroid use was not associated with transition to delirium. Several important clinical differences in patient characteristics between the two studies may explain the opposing findings. Further research is needed to elucidate the relationship between exogenous steroid administration and ICU delirium.
Long-term exposure to corticosteroids results in characteristic cushingoid side effects (e.g., as manifested in children with bronchopulmonary dysplasia). In the absence of obvious exogenous steroid administration, Cushing disease is diagnosed as a high 24-hour urine-free cortisol concentration that is not suppressed by administration of dexamethasone. Clinical characteristics of Cushing syndrome include hypertension, hypokalemic alkalosis, proximal myopathy, hyperglycemia, osteoporosis, opportunistic infections, psychiatric problems, and central obesity with characteristic striae.
Alterations of glucose homeostasis
Glucose homeostasis in health
Under normal conditions, glucose concentration is tightly regulated by the neuroendocrine system. Control of blood glucose is complex, involving interaction among the liver, pancreas, muscle, adipose tissue, pituitary, adrenals, and bone. The brain and periphery conduct a constant biochemical conversation by which the periphery informs the brain about its metabolic needs and the brain addresses these needs through its control of somatomotor, autonomic, and neurohumoral pathways involved in energy intake, expenditure, and storage. ,
Glucose is obtained from three sources: intestinal absorption of food, glycogenolysis, and gluconeogenesis. Once transported into cells, glucose can be stored as glycogen or it can undergo glycolysis to pyruvate. Pyruvate can be reduced to lactate, transaminated to form alanine, or converted to acetyl coenzyme A (CoA). Acetyl CoA can be oxidized in the mitochondrial tricarboxylic acid cycle to carbon dioxide and water, converted to fatty acids for storage as triglyceride, or serve as substrate for ketone body or cholesterol synthesis. Although glycogenolysis can occur in most tissues in the body, only the liver and kidneys express the enzyme glucose-6-phosphatase, which is required for release of cellular glucose into the bloodstream. The liver and kidneys also contain the enzymes required for gluconeogenesis. Of the two organs, the liver is responsible for the bulk of glucose output; the kidney supplies only 10% to 20% of glucose production during fasting.
Glucose homeostasis is centered on glucose-induced secretion of insulin from pancreatic β cells and insulin effect on glucose metabolism in peripheral tissues. The switch from glycogen synthesis during and immediately after meals to glycogen breakdown and gluconeogenesis is orchestrated by hormones, of which insulin is centrally important. Plasma insulin concentrations peak after meals. This surge in insulin activates glycogen synthesis, enhances peripheral glucose uptake, and inhibits glucose production. In addition, lipogenesis is stimulated while lipolysis and ketogenesis are suppressed. During fasting, the plasma insulin level falls to less than or equal to 5 μU/mL. Hormones—including glucagon, catecholamines, cortisol, and growth hormone (GH)—counteract the effects of insulin and promote glycogenolysis, gluconeogenesis, lipolysis, and ketogenesis.
Currently, obesity is an increasingly prevalent disease in the pediatric population. Glucose metabolism in patients with obesity is altered in health with chronic inflammation and dysregulated immune pathways, leading to impaired glucose metabolism. This subset of patients may require special attention when dealing with stress hyperglycemia.
Hyperglycemia
Hyperglycemia has been defined by the World Health Organization as blood glucose levels greater than 126 mg/dL when fasting. There are no specific age-adjusted levels for infants and children. No guidelines specifically define stress hyperglycemia, although a common definition describes it as transient hyperglycemia resolving spontaneously after dissipation of acute severe illness. Stress hyperglycemia (SHG) was first described by Thomas Willis in the seventeenth century. Initially identified as an appropriate adaptive harmless or even beneficial short-term physiologic response to critical illness, it is now considered a risk factor, especially because advanced critical care has improved survival and extended ICU length of stay (LOS). Hyperglycemia is common in nondiabetic critically ill children and occurs in up to 86% of patients.
Stress hyperglycemia and outcomes
Studies in children have challenged the assertion that hyperglycemia is beneficial or inconsequential by observing that SHG is associated with worse clinical outcomes. A retrospective study from a single PICU examined 1173 admissions. Hyperglycemia was prevalent and associated with increased morbidity, as characterized by increased LOS and mortality. Association between hyperglycemia and worse outcome in children was also demonstrated in specific disease processes, such as traumatic brain injury, general trauma, severe burns, septic shock, meningococcal sepsis, and postoperative congenital heart disease. , Hyperglycemia is common in infants with necrotizing enterocolitis admitted to the neonatal ICU (NICU) and is associated with increased late mortality and NICU LOS. SHG is also associated with specific complications, such as increased risk of venous thromboembolism in nondiabetic children, increased risk of mortality from central catheter–associated bloodstream infections, and more frequent nosocomial infections, including surgical site infection.
While clearly prevalent in critically ill children, some studies were unable to show a clear association between hyperglycemia and worse outcomes. One retrospective study found that hyperglycemia within 24 hours of PICU admission was associated with increased rate of mechanical ventilation, ICU LOS, and mortality, but when controlled for disease severity, it was not independently associated with increased morbidity or mortality. Another retrospective study evaluated the association between blood glucose level and duration of mechanical ventilation and ICU LOS in mechanically ventilated patients with bronchiolitis. Hyperglycemia was a frequent event in this patient population, yet results failed to show an independent association with worse outcomes. Such studies support the notion that hyperglycemia may be a marker of severity of illness rather than a cause.
Preexisting nutritional status may also impact outcomes associated with hyperglycemia in the critically ill child. A prospective study from Brazil reported that malnourished patients with hyperglycemia were at greater risk of mortality independent of severity of illness. Obesity is prevalent among critically ill pediatric patients, but the available literature on the relationship between obesity and clinical outcome is limited and conflicting. A systematic review of recent observational studies showed increased mortality in obese critically ill pediatric patients, while two separate retrospective single-center analyses revealed no difference in mortality by weight status. ,
With current knowledge, it is clear that hyperglycemia is prevalent in critically ill children and is associated with worse outcome in some disease processes but not in others. Many studies are not controlled for severity of illness, making it difficult to interpret results. Different definitions for SHG used by different authors (range of 126–250 mg/dL in nine key pediatric studies of association of SHG and mortality) make it even harder to compare results and create thresholds for treatment. Moreover, thus far, all studies have failed to establish cause-and-effect relationships, yet many suggest that aggressive maneuvers should be used to normalize plasma glucose levels. Younger patients (<1 year) are at higher risk for spontaneous hypoglycemia and require special consideration when treating hyperglycemia with insulin.
Pathophysiology of stress hyperglycemia
During stress, the normal mechanisms that counteract hyperglycemia are overwhelmed, causing a persistent unchecked state of high glucose blood levels. These changes help the body provide glucose to meet increased metabolic demands. SHG results from increased levels of counterregulatory hormones (epinephrine and norepinephrine, glucagon, cortisol, and GH) and proinflammatory cytokines (TNF-α, IL-1, and IL-6). The overall effect is increased hepatic and renal glucose production. Peripheral and hepatic insulin resistance is a well-recognized phenomenon in critically ill patients. It is characterized by organ-specific alteration in glucose utilization/production and impaired insulin-mediated uptake. Insulin concentrations may be elevated or decreased. Use of carbohydrate-based feeds, glucose-containing fluids, and drugs such as epinephrine and corticosteroids may exacerbate the situation.
High hepatic output of glucose, especially through gluconeogenesis, is the most important contributor to SHG. Glycogen stores are limited and rapidly depleted during stress. Lactate, pyruvate, and alanine are the main precursors used by the liver for gluconeogenesis. Often overlooked, renal-derived gluconeogenesis is a significant contributor (up to 40%) of glucose production during stress, mostly in response to epinephrine. The principal precursors for renal gluconeogenesis are lactate and glycerol ( Fig. 84.4 ).
Mechanisms underlying insulin resistance occur at several levels. Insulin receptor levels are unchanged in most short-term animal models of sepsis. However, reduction in insulin receptors is seen in longer-term models. One known mechanism for hepatic insulin resistance is associated with an increase in GH and reduction in insulin growth factor-1 (IGF-1). During critical illness, GH levels increase significantly, but there is a fall in hepatic GH receptors with disruption of downstream signaling. IGF-1 levels drop in response to proinflammatory cytokines, especially TNF-α with decreased IGF-1 synthesis. In addition, low IGF-1 may result from upregulation and increased affinity of its binding protein IGFBP-3, reducing the free active protein.
Elevated insulin levels are common in critically ill adults. The critically ill child may present with insulin deficiency due to β-cell dysfunction and impaired insulin production (direct effect of proinflammatory cytokines), which contributes to SHG. ,
Insulin resistance ultimately promotes a catabolic state in which lipolysis is activated. Lipotoxicity, glucotoxicity, and inflammation are key components of global SHG. Hyperglycemia itself further exacerbates the inflammatory and oxidative stress response and proinflammatory cytokine storming, promoting a vicious cycle whereby SHG leads to further SHG.
During critical illness, specific interventions can mediate development of SHG. Use of vasoactive-inotropic drugs—such as epinephrine, norepinephrine, and dopamine—is frequently associated with SHG. Epinephrine stimulates β 2 -receptors, promoting glycogenolysis and gluconeogenesis, and increases insulin resistance by release of glucagon and cortisol. It also reduces insulin secretion via stimulation of α 2 -receptors. Effects of dopamine and norepinephrine are less prominent due to lesser activity at the β 2 -receptors. Several medications commonly used in the ICU setting may result in development of SHG (e.g., patients frequently receive antifungal and antibiotic medications in large volumes of dextrose-containing fluids). Corticosteroids can increase the risk of SHG, especially when given in bolus doses. Thiazide diuretics are also associated with the occurrence of SHG. Calcineurin inhibitors, such as tacrolimus and cyclosporine, can result in SHG and posttransplant diabetes due to decreased insulin biosynthesis and release.
Nutritional support practices strongly influence SHG. Critically ill children are frequently prescribed parenteral nutrition (PN). Provision of excess carbohydrate calories in PN can result in SHG, which has been associated with increased risk of mortality in critically ill pediatric patients. Overfeeding is common in critically ill children, regardless of whether PN or enteral nutrition is employed. Studies have shown that commonly used predictive equations to calculate caloric needs are inferior to targeted indirect calorimetry and frequently result in overprescription of calories.
Mechanisms of stress hyperglycemia adverse outcomes
Postulated mechanisms by which SHG causes harm include direct cellular damage and alterations of essential organ function. In diabetic patients, four main molecular mechanisms have been identified in glucose-mediated complications: (1) increased polyol pathway flux, (2) increased advanced glycation end-product formation, (3) activation of protein kinase C isoforms, and (4) increased hexosamine pathway flux. Each of the four pathogenic mechanisms reflects a hyperglycemia-induced process, namely, overproduction of superoxide anion by the mitochondrial electron-transport chain. Diabetes-related injury develops over years, but some common mechanisms parallel acute stress-related hyperglycemia. During SHG, overexpression of insulin-independent transporters (glucose transporter types 1 to 3 [GLUT-1 to GLUT-3]) results in glucose overload and toxicity essentially in every organ. Cells damaged by hyperglycemia are primarily those unable to effectively control their intracellular glucose concentration—notably, neuronal, capillary endothelium, and renal mesangial cells. Glucose overload results in excessive glycolysis and oxidative phosphorylation with increased production of reactive oxygen species such as superoxide anion, the same toxic end product involved in the injury pathway for diabetic patients. Reactive oxygen species cause mitochondrial dysfunction and altered metabolism with subsequent apoptosis and cellular and organ system failure in the critically ill child. Glucose overload can also lead to glycation, the reaction of glucose with the amine group of proteins, which may impair the function of these proteins.
Hyperglycemia is a risk factor for infection in acute illness. The relative bacterial overgrowth witnessed in hyperglycemia may be partly due to altered host defenses. Acute, short-term hyperglycemia impairs macrophage activity, reduces polymorphonuclear leukocyte chemotaxis and bactericidal capacity, and alters complement fixation in critically ill patients. SHG affects all major components of innate immunity and impairs the ability of the host to combat infection. Furthermore, hyperglycemia is associated with poor gut motility, a factor that may be important in bacterial overgrowth and translocation. A raised blood glucose level is also recognized as being proinflammatory and pro-oxidant. Mononuclear cells isolated from healthy volunteers exhibited higher levels of NF-κB binding activity, raised reactive oxygen species, and increased levels of TNF-α mRNA following exposure to hyperglycemia.
Hyperglycemia also results in a hypercoagulable state partly through the increased expression of tissue factor, which is both pro-coagulant and proinflammatory. SHG is implicated in other abnormalities, such as endothelial dysfunction and alteration in vascular smooth muscle tone, commonly observed during critical illness. Likewise, hyperglycemia has been associated with deleterious effects on the nervous system. Underlying mechanisms in critical illness remain largely speculative and are often extrapolated from knowledge in diabetic patients. Hyperglycemia-induced blood-brain barrier permeability, oxidative stress, and microglia activation may play a role and compromise neurons and glial cell integrity.
Clinical trials examining management of critical illness hyperglycemia
In 2001, a single-center adult study reported reduction of hospital mortality by more than 30% using a tight glycemic control (TGC) protocol. The effect was attributed to the actual glycemic control rather than the infused insulin dose. These impressive results brought the issue of SHG to clinical attention and fostered considerable discussion. Previously seen as an aspect of a normal stress response, physicians now started viewing SHG differently, considering it to be a major cause or contributor to pathophysiology that must be aggressively addressed and treated. The appeal of such straightforward intervention was too great to resist. Subsequent studies failed to reproduce these results, yet guidelines generated by professional societies initially recommended TGC for adult critically ill patients. , A large international randomized multicenter study involving more than 6000 adult patients and many other smaller studies reported that TGC increased mortality and risk for severe hypoglycemia among adults in the ICU. In the setting of such consistent negative results, guidelines were revised and currently recommend using a higher glucose threshold for initiation of insulin therapy at 150 mg/dL, with the goal of keeping blood glucose <180 mg/dL, focusing on close monitoring and safety margins to avoid hypoglycemia and minimize glucose variability. While TGC became the standard of care for adults, the pediatric critical care community was hesitant to adopt any guidelines or consistent standard approach. There are, however, several prospective randomized clinical trials to guide practice.
In very-low-birth-weight neonates, a multicenter trial of insulin with continuous 20% dextrose infusion was terminated prematurely for concerns of futility and potential harm associated with hypoglycemia. This study took a proactive approach to glycemic control in that the treatment group received insulin with glucose infusion regardless of their blood glucose level before the intervention. Considering the study population and the fact that the study was not designed to treat hyperglycemia, it is difficult to compare the results with any other pediatric study.
Five randomized prospective studies have examined the association between TGC and clinical outcomes in critically ill children. The first study, published in 2009, was derived from a single center in Belgium. The study enrolled a mixed medical/surgical patient population, although 75% of the subjects had undergone cardiac surgery. The study targeted age-adjusted glycemic range for infants and children, 50 to 80 mg/dL in children less than 1 year old, and 70 to 100 mg/dL in the remainder. Results showed improved short-term outcome, including mortality, in the TGC group despite the fact that the TGC group had severe hypoglycemia (<40 mg/dL) at unacceptable rates (25% overall and 44% in neonates).
A two-center prospective randomized trial published in 2012 enrolled 980 children below age 3 years who underwent cardiac surgery with cardiopulmonary bypass. The authors reported that TGC (target range, 80–110 mg/dL) could be achieved with a low hypoglycemia rate. However, the study found no clinical benefit for TGC in terms of infection rate, mortality, LOS, or measures of organ failure when compared with standard care. A post hoc analysis of this study demonstrated that TGC may, in fact, lower the rate of infection in children older than 60 days of age at the time of cardiac surgery when compared with standard care. In a secondary analysis of this trial, insulin appeared to have no discernible impact on skeletal muscle degradation.
The concern for the impact of hypoglycemia on neurocognitive long-term outcome was addressed by both investigator teams. Four-year neurocognitive follow-up in the single-center study ascertained that insulin-induced hypoglycemia caused by TGC was not associated with worse neurocognitive outcome. However, the outcomes of both treatment groups were similar to the few patients who developed moderate or severe hypoglycemia in the two-center, cardiac-only trial. The group that had no hypoglycemia, as reported by continuous glucose monitoring, had a markedly better neurocognitive outcome than the other three groups, raising the possibility that the group in the first trial with no hypoglycemia detected may have had undetected hypoglycemia leading to the moderately impaired outcomes. Subsequent studies have confirmed the dangers of hypoglycemia in this population. Taken together, these data suggest that, in order to ensure optimal outcome, hypoglycemia should be assiduously avoided.
In 2014, a large multicenter randomized trial involving 13 centers in the United Kingdom reported that TGC in critically ill children had no significant effect on major clinical outcomes (number of days alive and free from mechanical ventilation at 30 days after enrollment), but patients in the TGC arm had lower need for renal replacement therapy and reduced 12-month healthcare costs. These effects were mostly notable in the noncardiac patient population.
In 2010, a single-center study focusing on pediatric patients with severe burns concluded that intensive insulin therapy significantly decreased infections and sepsis and improved organ function by decreasing inflammation.
After five randomized clinical trials of TGC to low versus high target ranges, this area of critical care therapeutics has become one of the most well studied in the field. The end result in general pediatric ICU and cardiac ICU patients, although initial promising findings were noted in the original single-center trial, is that low targets produce little to no benefit yet increase hypoglycemia, which is becoming increasingly associated with harm. Consensus has evolved that insulin infusion should be initiated when glucose levels reach 150 mg/dL, with the goal of keeping blood glucose less than 180 mg/dL. In the burn population, which has repeatedly been shown to be physiologically different from other critically ill children, the single-center study that has been completed stands in support of targeting a low range of 80 to 110 mg/dL.
Glucose measurement
Until recently, intermittent blood glucose levels (using point of care, blood gas analyzer, or central laboratory measurement) were the only means of blood glucose monitoring. Accuracy is probably the most important metric in selecting the best glycemic management device for critically ill children, but rapidity/turnaround time, cost, and sample volume are also important factors. Intermittent measurements are limited by the workload associated with the sampling process and with the potential that “between measurements” events will be missed. A simulation study modeling adult patients on TGC protocol demonstrated that increasing the frequency of glucose measurements reduced the adverse impact of glucose measurement imprecision on glycemic control. A mathematical simulation in a cohort of critically ill patients suggested that glycemic control is more optimal with a blood glucose measurement interval of no longer than 1 hour, with further benefit obtained with use of a measurement interval of 15 minutes. These findings have important implications for the development of glycemic control standards and future studies. With growing interest in glycemic control and the possible beneficial effect of frequent glucose measurements, continuous glucose monitoring systems have been developed. Although termed continuous , current systems still sample glucose intermittently with a measurement interval of a few milliseconds up to 15 minutes. The Clinical and Laboratory Standard Institute uses 15 minutes as the cutoff for definition of continuous measurement.
One of the key advantages of continuous glucose monitoring is the ability to identify and display trends in blood glucose measurements. High-quality continuous glucose monitoring devices enable clinicians to assess the complexity of the glycemic signal—how one point in time changes relative to neighboring measurements. Continuous glucose monitoring using subcutaneous sensors measuring interstitial fluid has been validated in the pediatric population, , and sensor performance has improved exponentially over the past decade. Although not yet approved by the US Food and Drug Administration for use in the inpatient setting, we expect such approvals to be forthcoming.
Hypoglycemia
Although the determination of which glucose levels represent hypoglycemia is controversial, a glucose level less than 40 mg/dL is generally accepted to represent severe hypoglycemia. However, this concentration is well below the level at which counterregulatory responses occur. As plasma glucose levels reach 80 to 85 mg/dL, insulin secretion decreases, and as levels approximate 65 mg/dL, glucagon, epinephrine, cortisol, and GH are released. In addition, a decrease in mental efficiency may be seen when levels fall below 50 to 60 mg/dL. Because a delay in the recognition and management of hypoglycemia may lead to long-term neurologic sequelae, it is important to make a distinction between the laboratory diagnosis of hypoglycemia (<40 to 50 mg/dL) and an interventional threshold at which therapies to raise serum glucose should be applied. Setting the interventional threshold at a level similar to that which elicits counterregulatory responses seems appropriate; as such, treatment should be offered for hypoglycemia when levels fall below 60 mg/dL to prevent complications, especially in young children. An even higher interventional threshold (<70 mg/dL) is warranted for children who are at increased risk of hypoglycemia.
Clinical manifestations
Diaphoresis, tremor, tachycardia, anxiety, weakness, hunger, nausea, and vomiting are all autonomic manifestations caused by the adrenergic stress response that occurs with a rapid decline in blood glucose levels. Other symptoms associated with hypoglycemia are a result of a deficiency of the brain’s primary energy substrate, which are known as neuroglycopenic symptoms. These symptoms include headache, visual disturbances, lethargy, restlessness, irritability, dysarthria, confusion, somnolence, stupor, coma, hypothermia, seizures, and motor and sensory disturbances. The glycemic ranges at which these symptoms manifest vary, as critically ill patients cannot recognize or communicate symptoms. The picture is further masked by sedation and analgesia.
Pathogenesis
Imaging studies of infants who sustained neonatal hypoglycemic brain injury display diffuse cortical and subcortical white matter damage that is most prominent in the parietal and occipital lobes. This pattern differs from the neuroimaging features of other neonatal insults, including hypoxic-ischemic encephalopathy. Interestingly, this pattern does not resemble the glucose uptake pattern of neonatal brains by positron emission tomography, which may indicate that neuronal damage is not simply due to cerebral deprivation of its primary substrate for energy production. Evidence indicates that hypoglycemia activates receptors for excitatory amino acids within the brain and causes cell depolarization, with subsequent cellular edema and apoptosis.
Fasting adaptation
Consumption of glucose is largely dependent on the brain-to-body ratio. This phenomenon explains the reduced fasting tolerance of infants whose glucose utilization rate (approximately 6 mg/kg per minute) is significantly higher than that of older children and adults (1 to 2 mg/kg per minute). This reality places younger patients at increased risk of hypoglycemia. In addition, their ability to maintain euglycemia through glycogenolysis and gluconeogenesis is reduced because glycogen stores and muscle bulk are small, thus reducing the pool of available gluconeogenic substrates. Within the brain, astrocytes, but not neurons, are capable of storing glycogen. The brain contains less than 1 mmol/kg of free glucose reserve. Fasting tolerance increases rapidly in the first days of life. Neonates may fast up to 18 hours after 1 week of age. By 1 year, a 24-hour fast is tolerated; by 5 years, a child may fast for up to 36 hours without experiencing hypoglycemia.
Understanding fasting physiology is crucial to the logic and methodologic approach required for diagnosing the etiology of hypoglycemia. Normally in the postabsorptive state, metabolism is governed primarily by counterregulatory hormones. In the first 4 hours of a fast in infants or in the first 8 hours in older children, glucagon is released and euglycemia is maintained primarily by glycogenolysis. Following glycogen store depletion, gluconeogenesis gains importance in the maintenance of normal glucose levels (see Fig. 84.4 ).
Muscle provides amino acids, particularly alanine and glutamine, as gluconeogenic substrates. Glycerol 3-phosphate derived from triglyceride hydrolysis is also a gluconeogenic precursor. Fatty acids resulting from triglyceride hydrolysis are transported to the liver, where they are oxidized to generate acetyl CoA and ketones. The latter may then be used as alternative fuel by skeletal and cardiac muscle to help ensure availability of glucose to the brain and to erythrocytes that are strictly dependent on glucose for energy production. The brain may also use ketones as an alternative fuel source, but it does so only during a prolonged fast.
Hypoglycemia that occurs early during fasting may indicate hormonal imbalance or a primary disorder of glycogenolysis. Disorders of gluconeogenesis (see also Chapter 81 ) will not manifest during early fast. They become apparent only after glycogen stores have been depleted; hence, typically, they present later in infancy once feeding intervals become increasingly prolonged. The same is true for fatty oxidation disorders. These disorders generally require a more prolonged fast to manifest, nearing 12 to 18 hours in infants and 18 to 24 hours in older children. However, the most common cause of childhood hypoglycemia is ketotic hypoglycemia. This illness most frequently occurs in toddlers and preschoolers and is uncommon after 8 to 9 years. It is typically triggered by intercurrent infection and caloric restriction, both common events in the PICU. A defect in protein catabolism, transamination, or amino acid efflux from skeletal muscle, as well as impaired autonomic regulation of epinephrine secretion, has been postulated.
Hypoglycemia has been observed in association with a variety of critical illness diagnoses, including sepsis, congestive heart failure, renal failure, liver failure, and pancreatitis, and it has been associated with increased mortality among critically ill children. , Critically ill patients are at risk of hypoglycemia not only because of their underlying illness but also because of factors unique to their hospitalization, such as muscular atrophy from prolonged immobilization and gluconeogenic substrate depletion, undernutrition often resulting from the limitation of caloric intake because of fluid restriction, increased glucose consumption, AI, loss of IV access or inadvertent disconnection of infusion lines, or iatrogenic factors related to drugs and therapies, including the practice of TGC.
Hypoglycemia treatment
After obtaining the “critical” blood/urine samples, administration of 2 mL/kg of 10% dextrose water solution (or an equivalent dose of dextrose) is indicated for patients with hypoglycemia. Subsequently, an IV maintenance fluid regimen should be considered to provide a glucose infusion rate of 6 to 8 mg/kg per minute. Serum glucose should be rechecked 15 minutes after the initial bolus; if hypoglycemia persists, a repeat bolus of 2 to 4 mL/kg of 10% dextrose water (or an equivalent dose of dextrose) should be administered and the glucose infusion rate increased by 25% to 50%. If the volume of fluid required to maintain glucose concentrations greater than 70 mg/dL is excessive, a higher dextrose concentration should be used. Glucagon (0.03 mg/kg for patients <30 kg, or 1 mg for patients >30 kg) can reverse hypoglycemia in patients with adequate glycogen stores and normal glycogenolytic pathways. Definitive treatment will depend on the underlying etiology.
In summary, hypoglycemia is a manifestation of iatrogenic, intentional, or accidental drug ingestion or administration or the manifestation of an underlying disorder. All critically ill patients with hypoglycemia should raise a high index of suspicion because many defects that cause hypoglycemia remain silent until an intercurrent illness or stress overwhelms the compensatory capacity of the individual. Unless certitude of the etiology of the hypoglycemia exists before therapy, a “critical” blood/urine sample should be obtained to guide diagnosis and further management. Prompt recognition and treatment are necessary to prevent neurologic injury. A multidisciplinary approach, including endocrine and/or metabolism consultation, is often necessary.
Alterations of thyroid hormone in critical illness
Thyroid biochemistry
Thyroid-stimulating hormone (TSH) derived from the anterior pituitary is a pleotropic hormone that modulates all aspects of thyroid hormone synthesis. TSH action within the thyroid follicular cells facilitates the sodium iodide symporter, resulting in (1) enhanced iodine concentration in the thyroid gland; (2) increased synthesis of thyroglobulin, the site of tyrosine residues destined for iodination; and (3) activated thyroid peroxidase, which catalyzes iodination of tyrosine residues as well as tyrosine coupling. It is important to note that autoantibodies may bind to TSH receptors and stimulate a response similar to TSH, resulting in a hyperthyroid state. Leptin is likely to mediate an important role in the regulation of the thyroid axis, as suggested by the close correlation between the circadian rhythm of leptin secretion and TSH.
An overview of thyroid hormone biosynthesis and secretion is provided in Fig. 84.5 . In this schematic diagram, iodide is transported into the thyroid follicular cell by the action of the sodium-iodide symporter (NIS). Subsequently, this iodide diffuses passively through the iodide channel termed pendrin (P). Thyroglobulin (TG) is synthesized within the rough endoplasmic reticulum (ER) and subsequently packaged by the Golgi apparatus into thyroglobulin secretory vesicles that are released into the follicular cell lumen. Thyroid oxidase (TO) produces hydrogen peroxide that is subsequently used by thyroid peroxidase (TPO) to oxidize iodide to iodine, which subsequently reacts with the tyrosine residues within thyroglobulin to produce monoiodotyrosine (MIT) and diiodotyrosine (DIT) residues within the thyroglobulin peptide.
