Key points

Adrenal steroidogenesis

With major involvement of cytochrome P450 isoforms, 3 classes of steroids are produced from cholesterol in the adrenal cortex. Generally mineralocorticoids are synthesized in the zona glomerulosa, glucocorticoids in the zona fasciculata, and gonadocorticoids in the zona reticularis. Synthesis of mineralocorticoids depends primarily on the renin-angiotensin-aldosterone (RAA) axis; synthesis of glucocorticoids is governed by activity of the hypothalamic-pituitary-adrenal (HPA) axis; and synthesis of gonadocorticoids is regulated by hypothalamic-derived gonadotropin-releasing hormone and pituitary-derived follicle stimulating hormone and luteinizing hormone with critical involvement of steroidogenic acute regulatory protein. Steroid hormones are not stored at their sites of biosynthesis; rather, release of steroid hormones is controlled almost entirely through regulation of their synthesis. Although critical care medicine practitioners frequently prescribe glucocorticoids for their favorable hemodynamic as well as immunosuppression properties to patients with septic shock, the purpose of this review is to suggest that mineralocorticoids as well as gonadocorticoids may also represent potentially useful adjuncts for treatment of sepsis.

Adrenal steroidogenesis

With major involvement of cytochrome P450 isoforms, 3 classes of steroids are produced from cholesterol in the adrenal cortex. Generally mineralocorticoids are synthesized in the zona glomerulosa, glucocorticoids in the zona fasciculata, and gonadocorticoids in the zona reticularis. Synthesis of mineralocorticoids depends primarily on the renin-angiotensin-aldosterone (RAA) axis; synthesis of glucocorticoids is governed by activity of the hypothalamic-pituitary-adrenal (HPA) axis; and synthesis of gonadocorticoids is regulated by hypothalamic-derived gonadotropin-releasing hormone and pituitary-derived follicle stimulating hormone and luteinizing hormone with critical involvement of steroidogenic acute regulatory protein. Steroid hormones are not stored at their sites of biosynthesis; rather, release of steroid hormones is controlled almost entirely through regulation of their synthesis. Although critical care medicine practitioners frequently prescribe glucocorticoids for their favorable hemodynamic as well as immunosuppression properties to patients with septic shock, the purpose of this review is to suggest that mineralocorticoids as well as gonadocorticoids may also represent potentially useful adjuncts for treatment of sepsis.

Mineralocorticoids

A schematic summary of aldosterone synthesis and regulation is provided in Fig. 1 .

RAA axis. ACE, angiotensin-converting enzyme; DCT, renal distal convoluted tubule; JGA, renal juxtaglomerular apparatus. In addition to monitoring local perfusion pressure and environmental Na ⁺ and Cl concentrations, the JGA receives β ₁ -adrenergic neural input from the brain.

As an aspect of the RAA axis activation, angiotensin II production is enhanced by endothelial angiotensin-converting enzyme, particularly within the pulmonary vasculature. Angiotensin II (a peptide not steroid hormone) mediates multiple activities also essential to the (sepsis) stress response:

• •
  

  Increases systemic vascular resistance and blood pressure

  
  
• •
  

  Stimulates aldosterone release

  
  
• •
  

  Increases plasminogen activator inhibitor-1, facilitating a prothrombotic state

  
  
• •
  

  Enhances thirst and salt craving

  
  
• •
  

  Increases antidiuretic hormone, adrenocorticotropic hormone, and norepinephrine release

  
  
• •
  

  Facilitates sodium reabsorption at proximal convoluted tubule

  
  
• •
  

  Stimulates renal afferent/efferent vasoconstriction

  
  
• •
  

  Increases nuclear factor κB (NF-κB), resulting in increased proinflammatory cytokine release

The principal actions of aldosterone, the end product of the RAA axis, include the following :

• •
  

  Sodium reabsorption

  
  
• •
  

  Potassium and hydrogen ion excretion

  
  
• •
  

  Insulin resistance

  
  
• •
  

  Hypertension

  
  
• •
  

  Activation of nuclear transcription factor, NF-κB activation

  
  
• •
  

  Arginine vasopressin release

More than 4 decades ago, hyperreninemic hypoaldosteronism was identified as a novel diagnostic entity among critically ill adults. This syndrome was observed to occur primarily among patients with infection and hypotension who exhibited a normal serum potassium, increased plasma cortisol (averaging 40.1 ± 10.1 μg/dL), increased plasma renin, low concentrations of plasma aldosterone, and elevated concentrations of aldosterone’s immediate precursor, 18-hydroxycorticosterone. Mortality with this constellation of findings was 78%. These patients were generally unresponsive to either adrenocorticotropic hormone or angiotensin II and were postulated to have a defect in the adrenal zona glomerulosa. Similar discordance between plasma renin and aldosterone concentrations has also been demonstrated in children with meningococcal sepsis : among 29 nonseptic critically ill children, mean Pediatric Risk of Mortality (PRISM) score was 9.4 and mean aldosterone concentration, 1489 ± 2.44 pg/mL. Among 231 children with meningococcal sepsis with a mean PRISM score of 32.3, however, mean aldosterone concentration was only 428 pg/mL ± 88 pg/mL. In another study focused on children with meningococcemia, aldosterone concentrations tended to be higher among nonsurvivors compared with survivors, but plasma renin activity did not significantly differ.

In ex vivo cell culture experiments, tumor necrosis factor (TNF)-a and interleukin-2 (IL-2) have been shown to inhibit synthesis of angiotensin II–induced aldosterone synthesis in a dose-response fashion. Aldosterone production is known to be under tonic dopaminergic inhibition that can be overridden by infusion of angiotensin II. An infusion of dopamine can abrogate aldosterone production whereas metoclopramide, a dopamine antagonist, can reverse dopaminergic inhibition. These findings suggest one reason why dopamine may not be the preferred vasoactive-inotropic agent among critically ill children.

Aldosterone’s role in hemodynamics and inflammation, as well as fluid and electrolyte balance, may be clinically relevant in the pediatric ICU, in view of the initial clinical trial that demonstrated a benefit of hydrocortisone in hastening resolution of septic shock as well as decreasing mortality, at least among patients demonstrating a seemingly inadequate (<9 μg/dL) increase in serum cortisol after corticotropin stimulation. This trial included both hydrocortisone and fludrocortisone, an aldosterone agonist, in the treatment group. Although hydrocortisone also possesses mineralocorticoid activity, it may be that aldosterone plays a more protean role in maintaining hemodynamics than is currently acknowledged. In addition, perhaps aldosterone mediates a systemic proinflammatory response to balance to the broad systemic anti-inflammatory activity characteristic of sepsis and potentiated by endogenous and exogenous cortisol, perhaps modulating away from immune paralysis.

Gonadocorticoids

Two major classes of gonadocorticoids have been characterized for potential utilization in critical illness, namely the anabolic steroids exemplified by oxandrolone and the estrogens typified by 17β-estradiol.

Anabolic steroids

Critically ill patients, in particular those with severe sepsis, are at risk for proteolysis of lean body muscle as a result of a catabolic state induced by endogenous and exogenous corticosteroids as well as proinflammatory cytokines, IL-1, TNF-α, and IL-6, and the effect of immobilization. These biochemical and clinical antecedents, common in the ICU, activate atrogene as well as a variety of protease pathways that ultimately release amino acids from skeletal muscle proteins. These amino acids are used for 3 key aspects of the acute stress response :

• •
  

  Gluconeogenesis to generate carbohydrate energy substrate

  
  
• •
  

  Protein synthesis focused on acute phase reactants

  
  
• •
  

  Expansion of immune system components.

This catabolic state is perpetuated as long as the initiating proinflammatory stimuli remain in place. Electron microscopy of muscle biopsies from critically ill patients has demonstrated loss of the a-band due to loss of myosin thick filaments. Computerized tomography and ultrasonography imaging have been used to document critical illness-associated lean body mass loss in real time.

Prolonged muscle catabolism has been extensively described among thermal burn patients, including children. Muscle catabolism affects both skeletal and diaphragmatic muscle and may be associated with need for prolonged weaning from mechanical ventilation as well as long-term impaired functional status and health-related quality of life in the physical domain. Patients with extensive loss of lean body mass may require prolonged rehabilitation.

During the first year after large thermal burns, both the fractional breakdown rate and the fractional synthesis rate for proteins have been documented. This skeletal muscle protein turnover remains elevated for up to a year after burn injury and has been attributed to simultaneous increases in both protein anabolism and catabolism. Muscle breakdown generally exceeds synthesis during this time interval, however, resulting in a persistent negative net protein balance.

Several potential therapeutic interventions have been suggested to counteract protein catabolism associated with prolonged critical illness, including

• •
  

  Activation of phosphoinositide 3-kinase

  
  
• •
  

  Antiatrophy drugs specifically targeting muscle-specific ubiquitin ligases (MURF-1 and atrogin 1)

  
  
• •
  

  Insulin and insulin-like growth factor 1

  
  
• •
  

  Myostatin (growth factor differentiation factor 8)

  
  
• •
  

  Electrical muscle stimulation

  
  
• •
  

  Anabolic steroids

Early clinical investigation demonstrated the benefits of both growth hormone and oxandrolone in promoting net positive nitrogen balance among thermal burn patients. Similar anabolic effects of oxandrolone have been reported for children with severe burn injury with oxandrolone providing a stimulus for new protein synthesis with corresponding correction of negative protein balance. Positive protein balance effects of oxandrolone translate clinically into improved weight gain after burn injury. In addition, administering oxandrolone for up to 2 years after severe burn injury results in greater improvements in bone mineral content, bone mineral density, and height velocity.

Estrogens

In multiple preclinical as well as human clinical studies, preferential survival of patients of the female gender has been reported. For example, in a clinically relevant mouse model of septic shock using cecal ligation and puncture, administration of an estrogen agonist markedly improved survival in a dose-response fashion. Corresponding analysis of gene expression data derived in this study demonstrated reversal of proinflammatory gene expression in estrogen agonist–treated animals comparable to animals that underwent a sham operation.

Estrogen mediates regulation of a family of genes related to respiration, including

• •
  

  Aerobic glycolysis

  
  
• •
  

  Respiratory efficiency

  
  
• •
  

  ATP generation

  
  
• •
  

  Calcium loading tolerance

  
  
• •
  

  Antioxidant defenses

  
  
• •
  

  Mitochondrial function in general

If the concept of mitochondrial dysfunction and widespread energy failure as a precursor for multiple organ dysfunction syndrome associated with sepsis is embraced, estrogen agonist therapy seems attractive for additional investigation.

Glucocorticoids, corticosteroids

Cortisol (hydrocortisone) represents the quintessential glucocorticoid that mediates alteration in gene expression for approximately 25% of the genome . As summarized in Fig. 2 , cortisol synthesis is under control of the HPA axis. Multiple proinflammatory agents signal the hypothalamus to release corticotropin-releasing hormone, which stimulates the anterior pituitary to release adrenocorticotropin hormone (ACTH). Corticotropin is released from the pituitary as proopiomelanocortin that is subsequently proteolytically cleaved to multiple small peptides. Corticotropin subsequently binds to class 2 melanocortin G protein–coupled receptors in the zona fasciculate cells of the adrenal cortex (and other sites, including leukocytes ), stimulating synthesis and release of cortisol. Typical stimulants of the HPA axis include IL-1, IL-2, and IL-6. In addition to TNF-α, other inhibitors of cortisol synthesis include macrophage inhibition factor and corticostatin, a peptide defensin with anticorticotropin activity. Cortisol is transported to peripheral tissues by cortisol binding protein as well as albumin.

Overview of the HPA axis and its bidirectional communication with the immune system. CRH, corticotropin-releasing hormone; INF, interferon.

Corticosteroids regulate gene expression through 2 major pathways. After nonfacilitated transport across the plasma membrane and binding to the cytosol glucocorticoid receptor (composed of heat shock proteins), the cortisol-receptor complex may: (1) traverse the nuclear membrane and directly bind to glucocorticoid responsive elements or (2) bind to NF-κB with subsequent binding of the complex to NF-κB responsive elements of the genome. Corticosteroids have the general effect of decreasing proinflammatory mediators and enhancing anti-inflammatory mediators. For example, corticosteroids are known to inhibit chemotaxis; inhibit expression of adhesion molecules, such as ELAM-1 and ICAM-1; and inhibit inducible nitric oxide synthase, NADP oxidoreductase, and cyclooxygenase resulting in decreased production of nitric oxide, superoxide anion, and other reactive oxygen species and prostaglandins. In addition, corticosteroids increase the production of lipocortin-1 (annexin) that similarly inhibits NADPH oxidoreductase and cyclooxygenase. Corticosteroids up-regulate the expression of IL-10 and IκB that both mediate anti-inflammatory activity. Finally, corticosteroids increase the production of macrophage inhibition factor that provides negative feedback for cortisol production. As summarized in Fig. 3 , based on in vitro animal and human studies, corticosteroids would be expected to favorably modulate all aspects related to unstable hemodynamics in septic shock.

Hemodynamic instability in sepsis and the biological plausible beneficial effects of adjunctive corticosteroids at multiple points along the natural history of septic shock. NO, nitric oxide (and other nitrogen and oxygen active species).

Cortisol can be assessed as either total cortisol or free cortisol, with the free form representing typically only approximately 10% of the total but accounting for all of the biological activity. Furthermore, cortisol can be assessed as a random concentration or a delta concentration change comparing baseline and stimulated states after administration of corticotropin. True adrenal insufficiency is typically associated with random total cortisol concentrations less than 5 μg/dL. Patients exhibiting an increase in circulating total cortisol from baseline less than 9 μg/dL after corticotropin administration have been viewed as having inadequate adrenal reserve and hence potential for relative adrenal insufficiency. Neither random cortisol concentration nor the corticotropin stimulation test, however, has consistently identified a group of patients who might benefit from hydrocortisone replacement therapy. In a general population of critically ill children, neither free cortisol nor total cortisol predicted signs or symptoms of adrenal insufficiency, such as hypotension, hypoglycemia, or hyponatremia. In this study, 33% of children exhibited total cortisol less than 10, 57% exhibited free cortisol less than 2%, and 30% exhibited free cortisol less than 0.8 μg/dL, but none demonstrated clinical evidence of critical illness-related cortisol insufficiency.

The 2014 Clinical Practice Parameters for Hemodynamic Support of Pediatric and Neonatal Septic Shock suggest adjunctive corticosteroid treatment of catecholamine-resistant shock if a patient is at risk for absolute adrenal insufficiency. What constitutes catecholamine resistant shock and absolute adrenal insufficiency remains unclear. Patients receiving acute or chronic corticosteroid dosing, patients with disorders of the HPA axis, children with congenital adrenal hyperplasia and multiple endocrinopathies, and patients who have had recent treatment with ketoconazole and etomidate are at risk for adrenal insufficiency. As discussed previously, however, laboratory testing of cortisol either with a random sample or corticotropin stimulation testing does not accurately predict which patients will respond favorably to hydrocortisone replacement.

In the twenty-first century, the practice of critical care medicine benefits from widespread vaccination for bacterial pathogens as well as mandatory newborn screening for congenital adrenal hyperplasia. Accordingly, the prevailing paranoia regarding risk for Addisonian crisis in pediatric sepsis is unfounded. Sepsis treatment guidelines previously recommended maintaining equipoise regarding the question of adjunctive steroid therapy for pediatric sepsis pending prospective randomized clinical trials. This recommendation is appropriate given the contradicting results of the 2 large, high-quality adult trials examining adjunctive corticosteroid therapy for septic shock. Currently, 2 large follow-up adult randomized controlled trials examining adjunctive corticosteroids for septic shock are in progress in Europe and Australia/New Zealand. In an important editorial related to prescription of corticosteroids for septic shock, clinicians were reminded that those who treat their patients with corticosteroids because they have observed a rapid reduction in the need for vasoactive inotropic support should be aware that more rapid weaning from this chemical hemodynamic support is an unreliable surrogate for clinically meaningful outcomes, because this intervention does not also improve survival.

In addition to the 2 large adult clinical trials examining adjunctive corticosteroids for septic shock, multiple observational studies have also concluded that corticosteroids provide no benefit and are potentially harmful when used for this indication. No high quality prospective randomized controlled clinical trials have examined the potential utility of adjunctive corticosteroids for pediatric septic shock. However, 7 good-quality, observational, cohort investigations have concluded no benefit or potential harm (increased mortality) related to adjunctive corticosteroids administered for pediatric septic shock:

• •
  

  PHIS (Pediatric Health Information System) database investigation

  
  
• •
  

  Researching Severe Sepsis and Organ Dysfunction in Children: A Global Perspective (RESOLVE) database investigation

  
  
• •
  

  PERSEVERE (PEdiatRic SEpsis biomarkEr Risk modEl) database investigation

  
  
• •
  

  Pediatric septic shock personalized medicine

  
  
• •
  

  SPROUT (Sepsis Prevalence, Outcomes and Therapies Study) point prevalence investigation

  
  
• •
  

  STRIPES (Steroid Use in Pediatric Fluid and Vasoactive Infusion Dependent Shock) investigation

  
  
• •
  

  PALICC (Pediatric Acute Lung Injury Consensus Conference) database investigation

As an example, one of these studies involved a follow-up investigation of the RESOLVE (NCT00049764) trial database. In this retrospective analysis, 193 children who received adjunctive corticosteroids for septic shock were compared with 284 children who did not receive adjunctive corticosteroids. At baseline, the 2 groups were similar in terms of age, gender, illness severity per PRISM III scores, and number of organ dysfunctions as well as baseline Pediatric Overall Performance Category score. Several outcomes, including mortality, days of vasoactive-inotropic infusion, duration of mechanical ventilation, time to composite organ dysfunction resolution, Pediatric Overall Performance Category score, and duration of pediatric ICU and hospital lengths of stay were similar in children who did or did not receive corticosteroid intervention as adjunctive therapy in this largest pediatric sepsis interventional clinical trial conducted to date.

Although most clinicians readily cite the beneficial hemodynamic and anti-inflammatory properties of corticosteroids, the adverse effects of this drug class are typically underappreciated. Relevant corticosteroid side effects include widespread immunosuppression, hypertension, hyperglycemia, reduced somatic growth, impaired wound healing, neuromuscular weakness, hospital acquired infection, and possible altered neurodevelopment. When administering even a single dose of corticosteroids, it is important to realize that although the intervention may (or may not) improve clinical hemodynamics, the corticosteroid also alters expression of approximately 25% of the human genome. Of particular relevance to treatment of pediatric septic shock with adjunctive corticosteroids is the observation that this intervention results in repression of multiple elements of adaptive immunity, including T-cell receptor signaling, T-helper cell signaling, overall regulation of the immune response, and signaling in T-lymphocytes and macrophages and cytotoxic T-lymphocytes as well as glucocorticoid receptor signaling. With mounting evidence for an alternative paradigm for sepsis pathogenesis in which both proinflammatory and anti-inflammatory responses may be simultaneously operative, it makes little sense to administer potent anti-inflammatory agents to treat sepsis patients who are in a state of immune suppression and may well benefit from immune reconstitution. Risks for ICU-acquired infections associated with hydrocortisone dosing have been clearly documented for both children and adults.

Although a pediatric intensivist’s gut reaction is to prescribe corticosteroids when a patient with (septic) shock is deteriorating, this instinctive approach may not be appropriate. The disconnect between improved hemodynamics after administration of the corticosteroids but lack of a benefit in terms of clinically meaningful outcome measures represents one of the most important research questions in pediatric critical care medicine, and, based on current evidence, it is difficult to argue either for or against the use of this drug class in children with septic shock. Clinicians should be aware that the actions of corticosteroids are not just like those of vasoactive-inotropic drugs that are also commonly used to treat unstable hemodynamics in septic shock; the notion that “it can’t hurt” is a misconception. There are passionate individuals on both sides of this important clinical controversy, but current guidelines are not supported by evidence in terms of whether corticosteroid therapy is beneficial for patients with severe sepsis. Because corticosteroids may produce either benefit or harm, there is a scientific, ethical, and health economic imperative to conduct a randomized trial of adjunctive corticosteroids for pediatric septic shock.

Design of such a trial is challenged by frequent (albeit unwarranted) lack of equipoise regarding the research question. Pediatric intensivists may currently be practicing therapeutic illusion, wherein physicians believe that their actions or tools are more effective than they actually are. Therapeutic illusion facilitates continued use of inappropriate tests and treatments. The results of such actions may be unnecessary and costly or dangerous care. Before concluding that a treatment is effective, physicians should consider other explanations, including natural history of the disease. If there exists some evidence for benefit, it is also important to look for evidence of risk.

Pediatric intensivists need to acknowledge that evidence is not the plural of anecdotes and the need to practice with intellectual honesty. A randomized controlled trial of adjunctive corticosteroids for pediatric septic shock has been waiting offstage for decades. Science requires testing of ideas, including the potential risks and benefits of all types of steroid treatment of sepsis. This testing will determine if predictions are supported by the experiment. This testing of theories and generating evidence are what distinguish science, including medicine, from other creative fields and ultimately provide a platform for best practice.