Historical Review

In 1564, the Roman anatomist, Bartholomeus Eustachius (1520-1570), discovered the adrenal gland as “glandulae quae renibus incumbent”—glands with an unknown function. Multiple hypotheses on its possible role were posited over centuries, such as that by the anatomist Adrianus Spigelius (1570-1625), who described the adrenal gland as the “upholstering space holder” between kidney and diaphragm. In 1855, Thomas Addison (1793-1860) first described a phenomenon wherein the only pathologic finding in some deceased patients was a morphologic destruction of the adrenal gland. He concluded that this organ must have a crucial function, and he called the syndrome Morbus Addison (Addison’s disease). One year later, Brown-Séquard confirmed his hypothesis after performing a series of bilateral adrenalectomies in cats, demonstrating that these endocrine glands were necessary for life. Addison’s conclusions, however, were not accepted, and even 2 years after his death, the famous pathologist, Rudolf Virchow, declared that he had never heard such an illogical statement.

In the 19th and early 20th century, several key findings were made. In 1856, von Koelliker described the anatomic division of the adrenal gland into cortex and medulla; and in 1903, Biedl confirmed that the adrenal cortex is the essential part. In 1894, epinephrine (adrenaline) was isolated from the adrenal medulla as the first hormone. Its chemical structure was described 3 years later, and in 1901, epinephrine was synthesized. In patients with Addison’s disease, however, the administration of epinephrine had no success, whereas the use of an animal extract of the adrenal cortex was life saving. Purification techniques were rapidly improved, and the resulting “Cortin” was the first-choice drug for treatment of Addison’s disease until the middle of the 20th century. Three independent groups of biochemists (Kendall, Winterstein, and Reichstein) successfully isolated 17-hydroxy-11-dehydrocorticosterone (later called cortisone) from the adrenal cortex; the physiologic compound, cortisol, was first described by Reichstein in 1937. The extraction of cortisone, however, remained arduous and uneconomical. Bovine adrenal glands of more than 20,000 animals were necessary to produce 1 kg of cortisone. The first synthesis and pharmaceutical preparation of cortisone was described in 1947 by an industrial company. Until that time, cortisone was only used in patients with Addison’s disease.

In the same decade, Hench and Kendall, two rheumatologists at the Mayo Clinic, found that in patients with different forms of rheumatism, the symptoms showed temporary remissions during pregnancy and inflammatory diseases like hepatitis. They speculated that this might be due to a general stimulation of the endocrine system and concluded that the use of cortisone might be beneficial in patients with acute rheumatoid arthritis. In September 1948, a female bedridden patient with severe and painful rheumatism that was resistant to all standard therapies at the time was the first documented case of cortisone treatment for inflammatory disease. After 3 days, the patient was able to stand up; 1 week later, she left the clinic without pain and on her own feet. Retrospectively, the speculations regarding pregnancy and hepatitis were obviously wrong, but the antiinflammatory character of cortisone was a key finding in pharmaceutical research. In contrast to Selye, who described cortisone as a crucial promoter of the physiologic stress response, the aforementioned finding that the adrenal gland cortex is the location for endogenous production of cortisone, an important inhibitor of stress and inflammation, has been confirmed. In 1950, Kendall, Hench, and Reichstein received the Nobel Prize in Medicine for their historical findings on the physiologic role of the adrenal gland.^1–4

Anatomy of the Adrenal Gland

The two paired adrenal glands are located in the retroperitoneal soft tissue near the top of each kidney. In neonates, the adrenal glands are relatively large (approximately one-third of the kidney’s size) compared with other organs. In the postnatal period, the cortex portion shrinks, leading not only to a relatively but also an absolutely smaller size of the organ. In adults, each adrenal gland weighs 4 to 5 g, has a flat form with a sagittal diameter of less than 1 cm, a transverse diameter of 3 cm, and a crani-caudal diameter of 4 to 5 cm. The right gland has a triangle/pyramid-like shape, whereas the left organ is shaped like a half-moon. The adrenal gland is composed of two embryologically distinct tissues. The adrenal cortex develops during the 5th week of gestation from a clump of mesodermal cells within the urogenital ridge known as the adrenal primordium. Later, during the 12th week of gestation, the adrenal medulla develops from neuroectodermal cells of the embryonic neural tube. In the fetal period, the cortex surrounds the medullar cells, resulting in the typical “sandwich” structure, consisting of a flat grey medulla with a yellow cortex.

The circulatory supply, with a flow rate of about 5 mL per minute, is maintained by up to 50 arterial branches from the aorta, renal arteries, and inferior phrenic arteries for each adrenal gland. Blood flow is directed from the capsule into the subcapsular arteriolar plexus through the cortex towards the medulla, where a single vein drains the blood entering the vena cava or the renal vein, respectively. A direct blood supply to the medulla is maintained by medullary arteries.

The adrenal cortex receives afferent and efferent innervation. Direct contact of nerve terminals with adrenocortical cells has been suggested, and chemoreceptors and baroreceptors present in the adrenal cortex infer efferent innervation. Diurnal variation in cortisol secretion and compensatory adrenal hypertrophy are influenced by adrenal innervation. Splanchnic nerve innervation has an effect in regulating adrenal steroid release. The adrenal medulla secretes the catecholamines, epinephrine and norepinephrine, that affect blood pressure, heart rate, sweating, and other activities also regulated by the sympathetic nervous system. The adrenal cortex is divided into three layers: (1) the zona glomerulosa, just under the capsule, (2) the zona fasciculata, the middle layer, and (3) the zona reticularis, the innermost net-like patterned area with reticular veins draining into medullary capillaries. The zona glomerulosa exclusively produces the mineralocorticoid, aldosterone; the zonae fasciculate and reticularis produce glucocorticoids and androgens.⁵

Physiology of the Hypothalamic-Pituitary-Adrenal Axis

The adrenal glands are part of a complex system that produces interacting hormones to maintain physiologic integrity, especially during the stress response.^6,7 This system, the hypothalamic-pituitary-adrenal (HPA) axis, includes the hypothalamic region which produces corticotropin-releasing hormone (CRH), triggering the pituitary gland. The pituitary gland is composed of two major structures: the adenohypophysis (anterior pituitary) and neurohypophysis (posterior pituitary). The anterior pituitary is responsible for the secretion of corticotropin (adrenocorticotropic hormone [ACTH]), thyroid-stimulating hormone (TSH), growth hormone (GH), β-lipotropin, endorphins, prolactin, luteinizing hormone (LH), and follicle-stimulating hormone (FSH). The posterior pituitary secretes vasopressin (antidiuretic hormone [ADH]) and oxytocin. Corticotropin regulates the production of corticosteroids by the adrenal glands. Hypothalamic neurons receive input from many areas within the CNS; they integrate these inputs and initiate an output to the anterior pituitary via the median eminence. The median eminence secretes releasing hormones into a hypophyseal portal network of capillaries that connect the median eminence with the pituitary hormones.

The anterior pituitary gland secretes corticotropin (ACTH) under stimulation from hypothalamic CRH. ACTH in turn stimulates the synthesis and release of glucocorticoids, mineralocorticoids, and androgenic steroids from the adrenal gland. In terms of a feedback loop, ACTH release is inhibited by glucocorticoids, which act on both the pituitary corticotropic cells and hypothalamic neurons. ACTH is also released during stress, independent of the circulating serum cortisol level. CRH, vasopressin, and norepinephrine act synergistically to increase ACTH release during stress. Endorphinergic pathways also play a role in ACTH regulation. Acute administration of morphine stimulates release of ACTH, while chronic administration blocks ACTH secretion. ACTH and cortisol are secreted normally in a diurnal pattern, with lowest concentrations between 10 PM and 2 AM and highest levels around 8 AM. From a practical point of view, it is important to know about rhythms, because inadequate assessment of endocrine function must take into account the variability of hormone levels in the blood. Samples obtained at different times can provide useful dynamic information regarding hypothalamic-pituitary-adrenal function. Loss of diurnal rhythm may indicate hypothalamic dysfunction.

The HPA axis is stimulated not only by physical or psychic stress but also by peptides like ADH and cytokines. Thus, the HPA axis plays an important role during infections and immunologic disorders.^8,9 By interaction with the renin-angiotensin-aldosterone system (RAAS) regulating fluid and salt balance, synthesis of androgens (e.g., dehydroepiandrosterone) with possible impact on immunomodulation, and the sympathoadrenergic system, the HPA axis is probably the most important organ of stress response. Stimulation of the immune system by infections induces the release of proinflammatory cytokines like tumor necrosis factor alpha (TNF-α), interleukin (IL)-1β, or IL-6. Following a cascade, these cytokines stimulate both the hypothalamus and the anterior pituitary gland, which finally leads to the release of glucocorticoids. IL-6 is also able to induce steroid release directly from the adrenal gland. The adequate increase of glucocorticoid levels during inflammation is a crucial factor for appropriate stress response. In acute infections, this release maintains metabolic and energy integrity. If the process is chronic, the HPA axis develops an adaptation which induces typical clinical manifestations such as hypercatabolic states, hyperglycemia, and suppression of androgens and growth and thyroid hormones. These changes, however, may increase the risk for secondary infections. Increased cortisol levels suppress higher regulatory levels of the HPA axis in terms of a negative feedback loop. Hence, after major surgery or during sepsis and septic shock, high cortisol and low ACTH levels are detectable.^10,11 Even the infusion of dexamethasone or CRH is not able to suppress increased cortisol levels in these patients.^12,13 This phenomenon leads to the question of how cortisol release is induced. Several investigations demonstrated that adrenal cortisol synthesis in critically ill patients is not regulated by ACTH, but by paracrine pathways via endothelin, atrial natriuretic peptide (ANP), or cytokines like IL-6.^14–16 IL-6 directly induces the adrenal cortex to release cortisol, which in chronic courses, can worsen the prognosis.¹⁷

Cellular Response to Adrenocortical Hormones and Related Drugs

Cortisol, the major free circulating adrenocortical hormone, is a hydrophobic hormone; being a steroid, it circulates bound to protein. Complexed cortisol-binding globulin (CBG, or transcortin) accounts for about 95% of circulating cortisol, but only the free form is biologically active. Its plasma half-life is 60 to 120 minutes. Cortisol is metabolized by hydroxylation in the liver, and the metabolites are excreted in urine. Steroid hormones enter the cytoplasm of cells where they combine with a receptor protein. Metabolic, immunologic, and hemodynamic responses to adrenocortical steroid hormones are regulated in a very complex manner that includes transactivation, transrepression, posttranscriptional/translational regulation, and nongenomic effects. The immediate nongenomic effects of steroid hormones were primarily attributed to mineralocorticoids (aldosterone). Rapid activation of the sodium-proton exchanger, increase of intracellular Ca⁺⁺, and activation of second messenger pathways were described.^18,19 A randomized trial in patients during cardiac catheterization revealed that within minutes after aldosterone injection, cardiac index and arterial pressure increased significantly for 10 minutes and returned to baseline afterwards.²⁰ Interestingly, the genomic effects of aldosterone seemed to be mediated by binding to glucocorticoid (GC) receptors and not to mineralocorticoid receptors.²¹ There is evidence that GC, like cortisol, also modulates immune functions by rapid nongenomic effects via nonspecific interactions with cellular membranes and specific binding to membrane-bound GC receptors (GR).²² Nonspecific membrane effects have been demonstrated for inhibition of sodium and calcium cycling across plasma membranes by impairing Na⁺/K⁺-ATPase and Ca⁺⁺-ATPase. Moreover, the rapid activation of lipocortin-1 and inhibition of arachidonic acid release after GC was independent from GR translocation. Finally, high-sensitivity immunofluorescence staining revealed membrane-bound GR on circulating B lymphocytes and monocytes.²²

The multiple mechanisms by which GCs modulate cellular responses include mainly genomic pathways.^23–25 Nongenomic effects are thought to account for immediate immune effects of high doses of GC, whereas membrane-bound receptors probably mediate low-dose GC effects. The classic model is that GCs bind to the cytoplasmatic ligand-regulated GC receptor alpha (GRα), which is an inactive multiprotein complex consisting of two heat shock proteins (hsp90) acting as molecular chaperones and other proteins (Figure 165-1). Upon GC binding to GRα, conformational change causes dissociation of hsp90, with subsequent nuclear translocation of GRα homodimers, binding of GRα to GC response elements (GRE) of DNA, and transcription of responsive genes (transactivation) such as lipocortin-1 and β₂-adrenoreceptors. Alternatively, GRα may bind to negative GRE (nGRE) and repress transcription of genes (transrepression) such as pro-opiomelanocortin (POMC). More importantly, transrepression without direct binding of GRα to GRE by protein-protein interactions of GRα with transcription factors, nuclear factor kappa B (NF-κB) and AP-1, has been recognized as a key step by which GC suppress inflammation,²⁶ inhibiting synthesis of TNF-α, IL-1β, IL-2, IL-6, IL-8, inducible nitric oxide synthase (iNOS), cyclooxygenase (COX)-2, cell adhesion molecules, and growth factors, and promoting apoptosis.²⁷ In addition, NF-κB repression may be mediated by GC-induced up-regulation of the cytoplasmatic NF-κB inhibitor, IκBα (see Figure 165-1) which prevents translocation of NF-κB.²⁸ Clinical investigations provide support for the presence of endogenous GC inadequacy in the control of inflammation and peripheral GC resistance.²⁹ With GC treatment, the intracellular relations between the NF-κB and GRα signaling pathways change from an initial NF-κB-driven and GRα-resistant state to a GRα-sensitive one. However, data are conflicting and probably do not explain early (<2 hours) suppressive effects of GC but may account for the longer-term dampening effect of GC on inflammatory processes.²³

Figure 165-1 Cellular mechanisms of glucocorticoid effects (right) and glucocorticoid resistance (left).

After passive transport through the cell membrane, glucocorticoids (GC) bind to the intracellular GC receptor alpha (GRα), which is sequestered in the cytoplasm, bound to the heat-shock protein (HSP) complex that comprises chaperone molecules HSP70 and HSP90. Binding of GC to GRα allows formation of a homodimer which is transported into the nucleus. GR-mediated transcription induces inhibitor kappa B alpha (IκBα), which binds to and inhibits nuclear factor kappa B (NF-κB). Thus, GC inhibits the NF-κB-mediated synthesis of proinflammatory cytokines like tumor necrosis factor alpha (TNFα). Impaired GC sensitivity (GC resistance) includes three major pathways (dotted arrows): (1) decreased cytoplasmatic GC concentrations secondary to increased P-glycoprotein-mediated efflux of GC due to overexpression of the MDR-1 gene; (2) increased expression of a truncated splice variant of the GR which is unable to transactivate GC-sensitive genes (GRβ); and (3) activation of proinflammatory mediators via upstream kinases (JNK), which can directly inhibit GR transcription activity. GC, glucocorticoids; GR, glucocorticoid receptor; HSP, heat shock protein; JNK, c-Jun N-terminal kinase; MDR-1, multidrug resistance gene 1.

Besides transcriptional regulation, posttranscriptional, translational, or posttranslational processes have been described for GC-induced modulation of COX-2, TNF-α, GM-CSF, IL-1β, IL-6, IL-8, and interferon gamma (IFN-γ).²³ Furthermore, GCs act at multiple levels to regulate iNOS expression by decreased iNOS gene transcription and mRNA stability; reduced translation and increased degradation of the iNOS protein by the cysteine protease, calpain³⁰; limitation of the availability of the NOS cofactor, tetrahydrobiopterin; reduced transmembranous transport and de novo synthesis of the NOS substrate, L-arginine; and lipocortin-1-induced inhibition of iNOS.^31,32 Together, these complex mechanisms result in the considerable effect of GC to inhibit inflammation and to stabilize hemodynamics. Finally, GC receptors have been found in nearly every nucleated cell in the body, and since each cell type has its own expression of GC effect, it follows that GCs have many effects in the body, which is equally true of endogenously produced GC hormones or exogenously administered GC medications. Both increase hepatic production of glucose and glycogen and decrease peripheral use of glucose. Steroids also affect fat and protein metabolism. They increase lipolysis both directly and indirectly by elevating free fatty acid levels in the plasma and enhancing any tendency to ketosis. GCs further stimulate peripheral protein metabolism, using the amino acid products as gluconeogenic precursors.

Definitions of Adrenal Insufficiency

Adrenal glands may stop functioning when the HPA axis fails to produce sufficient amounts of the appropriate hormones. Primary adrenal insufficiency is defined by the inability of the adrenal gland to produce steroid hormones even when the stimulus by the pituitary gland via corticotropin is adequate or increased. Primary adrenal insufficiency affects 4 to 6 out of 100,000 people. The disease can strike at any age, with a peak between 30 and 50 years, and affects males and females about equally. In 70%, the cause is a primary destruction of the adrenal glands by an autoimmune reaction (“classical” Addison’s disease or autoimmune adrenalitis), with about 40% of patients having a history of associated endocrinopathies. Most adult patients have antibodies against the steroidogenic enzyme, 21-hydroxylase,³³ but their role in the pathogenesis of autoimmune adrenalitis is uncertain. In the other 30%, the adrenal glands are destroyed by cancer, amyloidosis, antiphospholipid syndrome, adrenomyeloneuropathy, acquired immunodeficiency syndrome (AIDS), infections (e.g., tuberculosis, cytomegaly, fungi), or other identifiable diseases (Box 165-1). In these cases, the typical morphologic changes of the adrenal cortex are atrophy, inflammation, and/or necrosis. In primary adrenal insufficiency, the whole adrenal cortex is involved, resulting in a deficiency of GCs, mineralocorticoids, and adrenal androgenes.^34,35

Secondary adrenal insufficiency is characterized by adrenal hypofunction due to the lack of pituitary ACTH or hypothalamic CRH. Diseases of the anterior pituitary that can cause secondary adrenal insufficiency include neoplasms (e.g., craniopharyngiomas, adenomas), infarction (e.g., Sheehan’s syndrome, trauma), granulomatous disease (e.g., tuberculosis, sarcoidosis), hypophysectomy, and infection.³⁶ Causes also include hypothalamic dysfunction, such as after irradiation or surgical interventions (see Box 165-1). Because aldosterone secretion is more dependent on angiotensin II than on ACTH, aldosterone deficiency is not a problem in secondary adrenal insufficiency. Selective aldosterone deficiency can occur as a result of depressed renin secretion and angiotensin II formation.³⁴ Rare patients have an isolated deficiency of CRH,³⁷ and lymphocytic hypophysitis with subsequent adrenal insufficiency was described in women.³⁸ These disorders may lead to an isolated ACTH deficiency.³⁴

The so-called tertiary adrenal insufficiency, which is often summarized together with secondary forms, commonly occurs after withdrawal of exogenous GCs. Many of these patients do well during normal activities but are unable to mount an appropriate GC response to stress. This effect depends on the dose and duration of treatment and varies greatly from person to person. It should be anticipated in any patient who has been receiving more than 30 mg of hydrocortisone per day (or 7.5 mg of prednisolone or 0.75 mg of dexamethasone per day) for more than 3 weeks.³⁵ If supraphysiologic doses of GCs have been administered to a patient for more than 1 to 2 weeks, the drug should be tapered to allow for adrenal gland recovery. It may take 6 to 12 months for the adrenal glands to recover fully after prolonged use of exogenous GCs.³⁹ Since ACTH is not a major determinant of mineralocorticoid production, the basic deficit in adrenal insufficiency is that of deficient GC production. It is important that neither the dose of applied glucocorticoids, nor the time of treatment, nor the basal plasma level of cortisol allow sufficient assessment of the function of the HPA axis. Some drugs have also been described to induce adrenal insufficiency, either by directly affecting adrenocortical steroid release (e.g., fluconazole, etomidate)^40,41 or by enhanced hepatic metabolism of cortisol (e.g., rifampicin, phenytoin).³⁵

Isolated hypoaldosteronism is very rare and should be suspected in cases of hyperkalemia in the absence of renal insufficiency. The main causes for isolated deficiency of aldosterone secretion are congenital deficiency of aldosterone synthetase, hyporeninemia due to defects in the juxtaglomerular apparatus, or treatment with angiotensin-converting enzyme inhibitors that lead to loss of angiotensin stimulation. Other forms of hypoaldosteronism usually occur in patients with chronic renal disease and/or diabetes mellitus.

Relative Adrenal Insufficiency

The aforementioned forms of adrenal insufficiency which lead to an absolute deficiency of steroids are rare in critically ill patients (0%-3%).⁴² They are mostly characterized by morphologic changes of the HPA axis. To reflect the notion that subnormal adrenal corticosteroid production during acute severe illness can also occur without obvious structural defects in the HPA axis, deficiency syndromes due to a dysregulation have been termed functional adrenal insufficiency.⁴³ Functional adrenal insufficiency can develop during the course of an illness and is usually transient.³⁵ Decreased levels of GCs are registered much more often; these levels might be sufficient in normal subjects but are too low for stress situations, owing to higher need, and are associated with a worse outcome.⁴⁴ This led to the concept of relative adrenal insufficiency (RAI). The major cause for RAI is inadequate synthesis of cortisol due to cellular dysfunction. Hence, in contrast to absolute adrenal insufficiency, the morphologic changes in RAI may be minor, sometimes characterized by cellular hyperplasia within the adrenal cortex. This is often combined with peripheral GC resistance of the target cells, which is caused by inflammatory events and aggravates the clinical course, although the absolute cortisol serum levels might be normal.⁴⁵ In septic shock, RAI may be due to impaired pituitary corticotropin release, attenuated adrenal response to corticotropin, and reduced cortisol synthesis (Figure 165-2).^35,46,47 In addition, cortisol transport capacity to effect sites may be reduced, and response to cortisol may be impaired at the tissue level by cytokines modulating GC receptor affinity to cortisol and/or GC response elements.^48,49 In clinical trials, it was demonstrated that prolonged treatment of systemic inflammation in patients with severe acute respiratory distress syndrome (ARDS) with methylprednisolone can improve the decreased GC response by increasing the GC receptor affinity and reducing the NF-κB-mediated DNA binding and transcription of proinflammatory cytokines.²⁹ Thus, if RAI can be identified, treatment with supplemental corticosteroids may be of benefit.³⁵ Prevalence of RAI in the critically ill varied from 0% to 77% with different definitions, cutoff values, study populations, and adrenal function tests^{34,35,46,50,51} and may be as high as 50% to 75 % in severe septic shock.⁵²

Figure 165-2 Concept of relative adrenal insufficiency (RAI).

Unlike in adequate stress response (left), RAI may occur when causal or additional factors impair the function of the hypothalamic-pituitary-adrenal (HPA) axis. This may be due to microcirculatory failure, additional drugs like antibiotics, anesthetic drugs, infections, long-term use of steroids, or hemorrhages. Impaired HPA axis function results in insufficient antiinflammatory response and increased inflammatory response. Plus (+) denotes activation; minus (−), inhibition. CRH, corticotropin-releasing hormone; CBG, cortisol-binding globulin; GC, glucocorticoids.

Evaluation of Adrenal Insufficiency

In clinical practice, assessment of adrenal function is difficult, especially in critically ill patients, since the diurnal rhythm is lost. Values indicating normal adrenocortical function are listed in Box 165-2. Normally, serum cortisol concentrations in the morning (8 AM) of less than 3 µg/dL (80 nmol/L) are strongly suggestive of absolute adrenal insufficiency,⁵³ while values below 10 µg/dL (275 nmol/L) make the diagnosis likely. Basal urinary cortisol and 17-hydroxycorticosteroid excretion is low in patients with severe adrenal insufficiency but may be low-normal in patients with partial adrenal insufficiency. Generally, baseline urinary measurements are not recommended for the diagnosis of adrenal insufficiency. To differentiate between primary, secondary, and tertiary adrenal insufficiency in cases of low cortisol, it is recommended to measure plasma ACTH concentrations simultaneously. Inappropriately low serum cortisol concentrations in association with increased ACTH concentrations are suggestive of primary adrenal insufficiency, whereas the combination of low cortisol and ACTH concentrations indicates secondary or tertiary disease. This, however, should be confirmed by stimulation of the adrenal gland with exogenous ACTH. In secondary or tertiary adrenal insufficiency, the adrenal glands release cortisol, whereas in primary adrenal insufficiency, the adrenal glands are partially or completely destroyed and do not respond to ACTH.

ACTH stimulation tests usually consist of administering 250 µg (40 International Units) of ACTH (so-called high-dose ACTH stimulation test). For long-term stimulation tests, which are preferred for differentiating between secondary and tertiary adrenal insufficiency, 250 µg of ACTH are infused either over 8 hours or over 2 days.⁵⁴ Serum cortisol and 24-hour urinary cortisol and 17-hydroxycorticosteroid (17-OHCS) concentrations are determined before and after the infusion. This test may be helpful in distinguishing primary from secondary/tertiary adrenal insufficiency. In primary adrenal insufficiency, there is no or a minimal response of plasma or urinary cortisol and urinary 17-OHCS. Increases of these values in the 2 to 3 days of the test are indicative of a secondary/tertiary cause of adrenal insufficiency. In normal subjects, the 24-hour urinary 17-OHCS excretion increases 3- to 5-fold above baseline. Serum cortisol concentrations reach 20 µg/dL (550 nmol/L) at 30 to 60 min and exceed 25 µg/dL (690 nmol/L) at 6 to 8 hours post initiation of the infusion. Today this is not very often used, because clinical manifestations of adrenal insufficiency combined with basal cortisol levels, short-term ACTH stimulation tests, and CRH tests (see later) usually provide sufficient information.

A short-term stimulation test with 250 µg ACTH, mostly used for patients who are not critically ill, determines basal serum cortisol levels and the induced-response concentration 30 and 60 minutes after intravenous (IV) administration of ACTH. The advantage of the high-dose test is that pharmacologic plasma ACTH concentrations can be achieved by either IV or intramuscular injection.⁵⁵ This way of application, however, may be too high to identify mild cases of secondary adrenal insufficiency or chronic deficiencies.⁵⁶ Furthermore, it should not be used when acute secondary adrenal insufficiency (e.g., Sheehan’s syndrome) is presumed, since it takes several days for the adrenal cortex to atrophy, and it will still be capable of responding to ACTH stimulation normally. In these cases, a low-dose ACTH test or an insulin-induced hypoglycemia may be required to confirm the diagnosis.^57,58 A rise in serum cortisol concentration after 30 or 60 minutes to a peak of 18 to 20 µg/dL (500 to 550 nmol/L) or more is considered a normal response to a high-dose ACTH stimulation test and excludes the diagnosis of primary adrenal insufficiency and almost all cases of secondary adrenal insufficiency except those of recent onset.^59–61

To further differentiate between secondary and tertiary adrenal insufficiency, laboratory investigations may be augmented by a CRH stimulation test. In both conditions, cortisol levels are low at baseline and remain low after CRH. In patients with secondary adrenal insufficiency, there is little or no ACTH response, whereas in patients with tertiary disease, there is an exaggerated and prolonged response of ACTH to CRH stimulation which is not followed by an appropriate cortisol response.^62,63 Formerly, the HPA axis was also tested by a stimulated hypoglycemia test. After administering 0.1 units of insulin per kilogram bodyweight, inducing a hypoglycemic state of less than 40 mg/dL serum glucose, an intact HPA axis induces a serum cortisol concentration of more than 20 µg/dL. Nowadays, this procedure is considered obsolete because of the high risk of hypoglycemia.

In critically ill patients, primary causes of absolute or relative adrenal insufficiencies are multiple and often undetectable if no specific hypothesis exists. Volume-resistant septic shock or any other form of life-threatening hypotension with increasing need for catecholamines should give reason to evaluate adrenal function. Formerly, a serum cortisol value less than 20 µg/dL was suggestive for the diagnosis. Meanwhile, it is acknowledged that several factors complicate investigations of the HPA axis in patients with critical illness. A short-term ACTH stimulation test may be performed in critically ill patients suspected of having adrenal insufficiency. However, in most patients, RAI will be present, especially in patients with severe sepsis and septic shock. A clear definition of RAI is still missing, and the pathophysiology is rather complex, which makes it difficult to define clear cutoffs for both basal serum cortisol concentrations and incremental increases after short-term ACTH stimulation tests. Proposed cutoff points may depend on different methods used to measure cortisol, with variations when compared to high-performance liquid chromatography (HPLC) as the reference method.⁶⁴ In addition, considering free cortisol or increase in free cortisol in response to ACTH could increase accuracy of adrenocortical function tests.⁴⁸ Furthermore, extrapolating the diagnosis from reference values obtained from healthy people or patients with hypothalamic-pituitary-adrenal disorders may be misleading, since normal or high-normal cortisol concentrations in septic shock may indicate inadequate adrenal response to stress. In a large series of patients, receiver operating characteristic curve (ROC) analysis reached highest sensitivity (68%) and specificity (65%) for a reference value of less than 9 µg/dL (incremental increase) to detect nonresponders.⁵² Basal cortisol of 34 µg/dL and incremental increase of 9 µg/dL after stimulation were the best cutoff points to discriminate between survivors and nonsurvivors. The higher the basal plasma cortisol and the weaker the cortisol response to corticotropin, the higher was the risk of death. Some investigators have questioned the discriminative power of the incremental increase of cortisol after stimulation in patients with high basal cortisol values, as increases may reflect adrenal reserve more than adrenal function. Hence, RAI was defined based on the hemodynamic response when a randomly measured cortisol was less than 25 µg/dL.⁴⁶

Routine use of the low ACTH stimulation test in critically ill patients cannot be recommended at present, although the low-dose test is preferred in patients with secondary or tertiary adrenal insufficiency.⁶⁵ After stimulation with 250 µg ACTH, circulating corticotropin concentrations are 40 to 200 pg/mL during stress but may be as high as 60,000 pg/mL.³⁵ Stimulation of the adrenal gland with low doses of ACTH (1 µg) was shown to increase sensitivity and specificity to detect adrenal insufficiency in patients with hypothalamic-pituitary-adrenal disorders who respond normally to traditional high-dose stimulation.^35,66–69 The test is performed by measuring serum cortisol concentrations immediately before and 30 minutes after IV injection of ACTH in a dose of 1 µg (160 mIU) per 1.73 m² body surface.³⁴ This dose stimulates maximal adrenocortical secretion up to 30 minutes post injection, and in normal subjects results in a peak plasma ACTH concentration about twice that of insulin-induced hypoglycemia.⁷⁰ A value of 18 µg/dL (500 nmol/L) or more at any time during the test is indicative of normal adrenal function. The advantage of this test is that it can detect partial adrenal insufficiency that may be missed by the standard high-dose test.^57,58

Using the 1-µg ACTH stimulation test to more precisely uncover patients with RAI in septic shock has been proposed, but the 1-µg stimulation test has not been well validated in critically ill patients and patients with septic shock.^34,35 In addition, studies evaluating low-dose and high-dose ACTH stimulation tests in septic shock may have been flawed by methodological problems. At present, using the 1-µg ACTH stimulation test cannot be recommended routinely until further data from well-designed randomized studies in septic shock patients are available. The current recommendation is to use a three-level therapeutic guide for evaluating RAI in critically ill patients, especially those with septic shock. Patients with a random basal cortisol below 15 µg/dL will likely profit from low-dose corticosteroid therapy, whereas corticosteroid replacement is unlikely to be helpful when basal cortisol is above 34 µg/dL. When a random basal cortisol value is between 15 and 34 µg/dL, adrenocortical stimulation with 250 µg ACTH should discriminate responders (incremental increase ≥ 9 µg/dL) from nonresponders (<9 µg/dL). However, it has been pointed out that no cutoff values will be entirely reliable.³⁵

Clinical Symptoms

About 25% of patients with adrenal insufficiency present with adrenocortical crisis.³⁴ The symptoms are nonspecific and include sudden dizziness, weakness, dehydration, hypotension, and shock (Box 165-3). In many cases, the clinical picture may be indistinguishable from shock due to loss of intravascular fluid volume. Other features such as anorexia, nausea, vomiting, diarrhea, abdominal pain, and delirium may be present, but they are also common in patients with other acute illness. Hence, these symptoms may not be helpful in establishing the diagnosis of adrenal insufficiency and are often misleading. Hypoglycemia is rare in acute adrenal insufficiency but more common in secondary adrenal insufficiency; it is a common manifestation in children and thin women with the disorder. Especially in patients in the intensive care unit (ICU), it remains extremely difficult to recognize an acute, absolute adrenal insufficiency based on clinical symptoms. However, if the diagnosis is missed, the patient will probably die, so the threshold for laboratory investigations in cases of unexplained catecholamine-resistant hypotension should be low. It is important to be mindful that the onset of an acute adrenocortical crisis is not necessarily an acute beginning of the underlying disease itself. The preceding course is often gradual and may go undetected until an acute illness, stress, trauma, pregnancy, or other conditions precipitate adrenal crisis.^34,71

In most cases, primary adrenal insufficiency is the underlying disorder. Typical symptoms such as hyperpigmentation, scanty axillary and pubic hair, hyponatremia, or hyperkalemia may be diagnosed in the acutely ill patient. Adrenal crisis can occur in patients receiving appropriate doses of GCs if their mineralocorticoid requirements are not met.⁷² After spontaneous events (e.g., hemorrhage, myocardial infarction, adrenal vein thrombosis), these signs are absent. If an acute adrenal crisis is suspected, a blood sample should be obtained to confirm the diagnosis. The main clinical problem is hypotension and shock due to acute mineralocorticoid deficiency. This is one reason for the fact that an acute adrenal crisis after secondary adrenal insufficiency is not so typical. However, GC deficiency may also contribute to hypotension by decreasing vascular responsiveness to angiotensin II, norepinephrine, and other vasoconstrictive hormones, reducing the synthesis of renin substrate and increasing production and effects of prostacyclin and other vasodilatory hormones.^73,74 Finally, panhypopituitarism may be associated with symptoms, owing not only to lack of corticotropin but also TSH, gonadotropin, and growth hormone.

In chronic adrenal insufficiency, the major clinical features (see Box 165-3) may be detected but may also be absent if adrenal gland insufficiency develops over a prolonged period of time. There is a stage characterized by normal basal steroid secretion but an inability to respond to stress. Hence, the patient may be asymptomatic. In other cases, there may also be signs and symptoms suggestive of other hormone deficiency such as decreased thyroid and gonadal function. Independent from the underlying cause, the most common clinical manifestations are general malaise, fatigue, weakness, anorexia, weight loss, nausea, vomiting, abdominal pain, arthralgia, postural syncope, diarrhea that may alternate with constipation, hypotension, electrolyte abnormalities (hyponatremia, hyperkalemia, metabolic acidosis), decreased axillary and pubic hair, and loss of libido and amenorrhea in women.^34,71

In primary adrenal insufficiency, hyperpigmentation and autoimmune manifestations (vitiligo) are typically due to increased ACTH concentrations, whereas this is not seen in secondary or tertiary adrenal insufficiency. Soon after the disease develops, the skin becomes dark, which may appear to be tanning but appears on both sun-exposed and nonexposed areas. Black freckles develop on the forehead, face, and shoulders; a bluish-black discoloration may develop around the lips, mouth, rectum, scrotum, or vagina. Another specific symptom of primary adrenal insufficiency is a craving for salt.³⁵ Typical laboratory abnormalities are hyponatremia, hyperkalemia, acidosis, slightly elevated creatinine concentrations, mild normocytic anemia, and rarely, hypercalcemia.³⁵

In secondary adrenal insufficiency, since production of mineralocorticoids by the zona glomerulosa is mostly preserved, dehydration and hyperkalemia are not present, and hypotension is less prominent than in primary disease. Especially in the early stages, the onset of chronic adrenal insufficiency is often insidious, and the diagnosis may be difficult. Some patients initially present with gastrointestinal symptoms such as nausea, vomiting, diarrhea, and abdominal cramps.^35,75 In other patients, the disease may be misdiagnosed as depression or anorexia nervosa.^76,77 Hyponatremia and increased intravascular volume may be the result of “inappropriate” increase in vasopressin secretion. Decreased libido and potency as well as amenorrhea may occur. Hypoglycemia is more common in secondary adrenal insufficiency, possibly due to concomitant growth hormone insufficiency, and in isolated ACTH deficiency. Clinical manifestations of a pituitary or hypothalamic tumor, such as symptoms and signs of deficiency of other anterior pituitary hormones, headache, or visual field defects, may also be present.^34,71 Finally, in young patients suspected of having adrenal insufficiency, delayed growth and puberty would point to the presence of hypothalamic-pituitary disease, as would headaches, visual disturbances, or diabetes insipidus in patients of any age.^35,36 Laboratory screening in patients with chronic adrenal insufficiency usually reveals hyponatremia, hypoglycemia, lymphocytosis, and eosinophilia.³⁵

Therapeutic Strategies

Treatment of adrenal insufficiency involves eradication of the precipitating cause (e.g., tumor, infection) and hormone replacement. In acutely ill patients, if the diagnosis of adrenal crisis is suspected but not known, blood should be obtained for measurement of cortisol concentrations, followed by the administration of 250 µg of ACTH in patients with unknown history. Therapy should be started immediately while awaiting results of testing.⁷⁸ Dexamethasone (1 mg every 6 hours) may be given as the initial GC replacement, since it does not cross-react with cortisol in the plasma while adrenal testing is being performed. Patients are usually treated with IV fluids in the form of isotonic saline to restore intravascular volume and replace urinary salt losses. Dextrose infusion may be added to prevent hypoglycemia. Hydrocortisone (100 mg IV bolus or over 30 min, followed by continuous infusion of 10 mg/h, or 50 mg every 4 hours, or 75 to 100 mg every 6 hours, resulting in a total daily dose of 240-300 mg hydrocortisone) is frequently given for hormonal replacement.^34,78 However, equivalent GC doses of methylprednisolone or dexamethasone may also be used. Typically, mineralocorticoid replacement therapy is not required in adrenal crisis so long as the patient is receiving isotonic saline. Prophylactic use of antibiotics is not beneficial, but specific infections should be treated aggressively with appropriate antibiotic therapy.

Once the patient is stable, or in cases of chronic adrenal insufficiency, GCs can be tapered to maintenance doses. Long-term replacement doses consist of hydrocortisone, 30 mg/d, with two-thirds (20 mg) given in the morning and one-third (10 mg) given at night; or prednisone, 7.5 mg in a similar regimen (5 and 2.5 mg, respectively). The daily dose may be decreased to 20 or 15 mg of hydrocortisone as long as the patient’s well-being and physical strength are not reduced.³⁴ The goal should be to use the smallest dose that relieves the patient’s symptoms, in order to prevent weight gain and osteoporosis.^34,78,79 If the patient continues to experience weakness or other symptoms of GC deficiency, the dose can be increased. Excessive GC therapy should be avoided so as to minimize complications of this therapy. In addition, a mineralocorticoid effect is provided with fludrocortisone (50-100 µg PO daily) to prevent sodium loss, intravascular volume depletion, and hyperkalemia, especially when the dose of hydrocortisone decreases below 100 mg/d. Therapy can be guided by monitoring blood pressure, serum potassium, and plasma renin activity, which should be in the upper normal range.^34,61 Clinical response, however, is the best indicator of adequacy of replacement. The optimal dosage of mineralocorticoids remains stable over long periods. Excessive mineralocorticoid replacement may cause congestive heart failure, alkalosis, hypokalemia, or hypertension. Patients receiving prednisone or dexamethasone may require higher doses of fludrocortisone to lower their plasma renin activity to the upper normal range, whereas patients receiving hydrocortisone, which has some mineralocorticoid activity, may require lower doses. The mineralocorticoid dose may have to be increased in the summer, particularly if patients are exposed to temperatures above 29°C (85°F). In cases of isolated hypoaldosteronism, treatment includes liberal sodium intake and daily administration of fludrocortisone. In patients with secondary adrenal insufficiency due to panhypopituitarism, replacement with other hormones may also be necessary. In women, the adrenal cortex is the primary source of androgen in the form of dehydroepiandrosterone and dehydroepiandrosterone sulfate. Although the physiologic role of these androgens in women has not been fully elucidated, their replacement is being increasingly considered in the treatment of adrenal insufficiency.^80,81

Once the patient is stable and on maintenance doses of steroids, ACTH testing can be repeated to document adrenal recovery. Patients with primary adrenal insufficiency require lifelong GC and mineralocorticoid replacement therapy and should carry a card containing information on current therapy, as well as some type of MedicAlert bracelet or necklace with recommendations for treatment in emergency situations. One of the important aspects of the management of chronic primary adrenal insufficiency is patient and family education. Patients should understand the reason for lifelong replacement therapy, the need to increase the dose of GCs during minor or major stress, and how to inject hydrocortisone, methylprednisolone, or dexamethasone in emergencies. Patients should also have supplies of dexamethasone sodium phosphate and should be educated about how and when to administer them. The survival rate for patients with chronic primary adrenal insufficiency has gone from 2 years or less before the availability of steroid replacement to that of the normal population now that GCs are readily available. In acute adrenal insufficiency, prompt recognition and treatment usually result in a favorable outcome, provided the underlying disease process can be treated.

Glucocorticoid Replacement in Patients with Septic Shock

In patients with severe sepsis and septic shock, the individual clinical course is extremely varied. The impact of the primary disease, as well as immunologic factors (cytokines), affect the HPA axis, and functional testing is aggravated. In contrast to the early phase of septic shock, adrenal cortisol release may recover, thus leading to RAI with absolute steroid levels around or even above normal range.⁸² In refractory septic shock, prevalence of RAI may be as high as 50% to 75%.⁵² Furthermore, dynamic testing is not always available in ICUs, which makes it difficult for the physician considering hormone replacement therapy, because decisions have to be made within hours in severe forms of septic shock to improve prognosis. Rationale for the use of high-dose GCs in infection, sepsis, and shock can be attributed to well-defined antiinflammatory and hemodynamic effects recognized for decades. Proposed mechanisms of protection include improvement of hemodynamic, metabolic, endocrine, and phagocytic functions, resulting in maintenance of normal morphologic-functional status of tissues including brain, liver, heart, kidneys, and adrenals.⁸³ In addition, GCs were recognized to inhibit key features of inflammation: endothelial cell activation and damage, capillary leakage, granulocyte activation, adhesion and aggregation, complement activation, and formation and release of eicosanoid metabolites, oxygen radicals, and lysosomal enzymes.^84–89

However, only in one long-term prospective study in humans receiving high doses of methylprednisolone (30-60 mg/kg) or dexamethasone (2-4 mg/kg), including 179 bacteremic septic shock patients over a period of 8 years, were experimental results confirmed and mortality reduced from 38% to 10%.⁹⁰ Evidence from another study suggested that prolongation of treatment might have been beneficial, since shock reversal and improved survival occurred after bolus GC application in an early time window but vanished after several days.⁹¹ Two meta-analyses included 9 and 10 randomized trials, respectively, of patients with severe sepsis and septic shock who received up to 42 g of hydrocortisone equivalent or more; both concluded that high doses of corticosteroids were ineffective⁹² or harmful.⁹³ This was confirmed by a large randomized trial in 1987.⁹⁴ Patients with proven gram-negative infections probably benefited more from GCs.⁹² In one analysis, studies with the highest quality demonstrated worse outcomes with corticosteroids.⁹³ High-dose GCs were associated with increased risk of secondary infections, mortality,⁹³ and increased incidence of renal and hepatic dysfunction.⁹⁵ Taken together, these results suggest that high-dose GCs failed to be effective in septic shock in the long run, most probably owing to immune system breakdown.

Similar to studies of high-dose GC treatment, numerous randomized controlled trials with low-dose corticosteroids in patients with septic shock also confirmed shock reversal and reduction of vasopressor support within few days after initiation of therapy in most patients.^96–101 In a crossover study, mean arterial pressure and systemic vascular resistance increased during low-dose hydrocortisone treatment, and heart rate, cardiac index, and norepinephrine requirement decreased significantly.¹⁰² All effects were reversible with cessation of hydrocortisone. Some studies indicate that corticosteroid-induced increase of sensitivity to norepinephrine is more pronounced in patients with RAI than in patients without RAI.^46,101 There are multiple potential mechanisms by which corticosteroids may modulate vascular tone. Considerable evidence confirms that cytokine-induced formation of nitric oxide (NO) plays a central role in vasodilation, catecholamine resistance, maldistribution of blood flow, and mitochondrial and organ dysfunction, and that the amount of NO production correlates with shock severity and outcome.^103,104 In a crossover trial, norepinephrine requirement could be reduced by low-dose hydrocortisone in nearly all patients within 1 to 2 days. Hydrocortisone treatment also induced a significant and prolonged decline of nitrite/nitrate levels, which significantly correlated with reduction of norepinephrine requirement during hydrocortisone infusion.¹⁰² Considering the complex genomic and nongenomic actions of corticosteroids described earlier, it is probable that NO is not the only target. However, inhibition of NO synthesis by hydrocortisone at least contributes to shock reversal.

It is recognized that GCs modulate the stress response in a very complex manner that includes not only antiinflammatory and immunosuppressive actions to protect the host from overwhelming inflammation, but also immune-enhancing effects.²⁷ The final effect of corticosteroids may be dependent on multiple factors such as the dose, type of cell or tissue, time point of action, and the balance of pro-inflammatory and antiinflammatory cofactors. Markers of the inflammatory response, antiinflammatory response, granulocyte, monocyte, and endothelial activation, antigen-presenting capacity, and innate immune response were investigated in septic shock patients.¹⁰² Hydrocortisone significantly attenuated inflammatory and antiinflammatory responses as well as granulocyte, monocyte, and endothelial activation. Monocyte HLA-DR expression was depressed, but receptor down-regulation was limited and followed by a rebound increase after drug withdrawal.¹⁰² One could thus conclude that the immune effects of low-dose hydrocortisone treatment in septic shock may be characterized as immunomodulatory rather than immunosuppressive. Attenuation of a broad spectrum of the inflammatory response without causing severe immunosuppression might be a promising therapeutic approach, which goes far beyond hemodynamic stabilization.

Although data on outcome in septic shock patients after low-dose corticosteroid treatment are limited, up to 300 mg hydrocortisone per day may improve survival. In most trials with low-dose corticosteroids,^96–100 28-day all-cause mortality was reduced, whereas in high-dose trials, there was no significant effect. In a multicenter trial in 300 patients with severe volume and catecholamine-refractory septic shock, survival time was significantly increased in patients with RAI but not in responders to ACTH.⁹⁷ Similar results were obtained for ICU and hospital mortality, but not for 1-year follow-up. Significant increases of serious adverse events during treatment with low-dose hydrocortisone have not been reported. The incidence of gastrointestinal bleeding, superinfections, or hyperglycemia has not been different in patients treated with corticosteroids or placebo, and wound infections were even less frequent in patients treated with low-dose hydrocortisone.⁹⁷ These findings were not confirmed by another large randomized trial, the Corticosteroid Therapy of Septic Shock (CORTICUS) trial¹⁰⁵ which, however, used different inclusion criteria. Only patients who were successfully resuscitated by volume therapy plus applied vasopressor were included.¹⁰⁵ These contradictory results led the Surviving Sepsis Campaign to redefine their guidelines in 2008,¹⁰⁶ recommending the use of low-dose GCs only for patients who are not responding adequately to volume plus vasopressor therapy—that is, those who are still hypotensive.¹⁰⁶

Treatment with low-dose hydrocortisone may induce an increase of sodium levels within a few days, and hypernatremia with values over 155 mmol/L have been reported during prolonged treatment.¹⁰⁰ Nevertheless, the indication for low-dose corticosteroids should be weighed against possible risks, and treatment should be limited to the duration of volume- and vasopressor-restrictive hypotension.

Dosing of hydrocortisone in septic shock is similar to adrenal crisis (100 mg initial bolus, followed by 200-300 mg per day), and the dose should be tapered when the patient stabilizes. Hydrocortisone should be preferred, although a comparative study of different corticosteroids has not been performed in septic shock, since most experience of low-dose corticosteroid treatment in septic shock was derived from studies using hydrocortisone (see earlier). Furthermore, hydrocortisone is the synthetic equivalent to the physiologic final active compound, cortisol, so treatment with hydrocortisone directly replaces cortisol independently from metabolic transformation. Finally, in contrast to dexamethasone, hydrocortisone has intrinsic mineralocorticoid activity. A recent randomized trial demonstrated that the addition of oral fludrocortisone to low-dose hydrocortisone has no benefit in septic shock patients.¹⁰⁷ It has not been established whether a weight-adjusted regimen (e.g., 0.18 mg/kg/h)⁵⁶ of continuous hydrocortisone infusion is superior to a fixed regimen; moreover, a comparative study of bolus versus infusion regimens has not been performed so far. Patients should be weaned from low-dose hydrocortisone over several days to avoid hemodynamic and immunologic rebound effects. In patients with septic shock, abrupt cessation of low-dose hydrocortisone was followed by significant reversal of many hemodynamic and immunologic effects observed during corticosteroid therapy, even after a short treatment period of 3 days.¹⁰² Adrenal function tests with 250 µg ACTH can be performed in patients with septic shock; however, at present it cannot be recommended to exclude responders or patients with high random cortisol values from low-dose corticosteroid therapy.³⁵ When basal serum cortisol concentrations are less than 15 µg/dL in septic shock, low-dose hydrocortisone replacement is recommended; levels of over 34 µg/dL are considered sufficient. Between 15 and 34 µg/dL, an incremental increase of less than 9 µg/dL serum cortisol makes relative adrenal insufficiency likely, and therapy may be considered according to the clinical state.³⁵ Other recommendations prefer a randomly assigned cutoff level of below 25 µg/dL serum cortisol.⁴⁶ The routine use of the low ACTH stimulation test (1 µg ACTH) cannot be recommended at present until further data from well-designed randomized studies in septic shock patients are available. Most importantly, it has to be realized that all the aforementioned studies were performed in patients with catecholamine-resistant septic shock. So far there are no data justifying the use of low-dose steroids in patients with sepsis and severe sepsis. Significant effects on outcome have been observed only in patients with systolic blood pressure below 90 mm Hg despite vasopressor therapy.⁹⁷ It is not yet known whether low-dose corticosteroids are also effective in patients with less severe shock. Sufficient data on the dose-response characteristics of GCs in septic patients are still lacking, and the current recommended strategy using 200 to 300 mg hydrocortisone per day is based on empirical recommendations; further investigations are needed.

Further Implications for Anesthesia and Critical Care

Surgical stress increases serum cortisol levels five- to sixfold postoperatively, with return to normal at 24 hours unless stress continues. Patients who have received GCs equivalent to 30 mg/d cortisol for longer than 3 weeks may have impairment in this stress response, and steroid supplementation should be considered. However, short-term treatment of heterogeneous groups of patients with critical illness is controversial, and supraphysiologic doses of GCs are not beneficial and may even be harmful.¹⁰⁸ Hence, outside the situations in which benefit has been proved, supraphysiologic doses of GCs (e.g., 30 mg methylprednisolone per kilogram of body weight per day) in patients with critical illness are not indicated. Some successful indications, however, have been described: in patients with unresolving ARDS, pharmacologic doses (2 mg methylprednisolone/kg/d) reduced mortality and improved organ function.²⁹ Furthermore, early treatment with dexamethasone may decrease morbidity in bacterial meningitis,^109,110 although a recent meta-analysis was less enthusiastic.¹¹¹ The positive effects of steroid treatment on tissue-specific resistance to GCs have already been described. However, despite the frequent suggestion that unexplained intraoperative hypotension and even death reflect unrecognized hypocortisolism, there is no evidence that primary adrenal insufficiency is a likely explanation for this response.

Patients with known chronic adrenal insufficiency must be advised to double or triple the dose of hydrocortisone temporarily whenever they have any febrile illness or injury.³⁴ In stressful situations or during major surgery, trauma, burns, or medical illness, high doses of GCs up to 10 times the daily production are required to avoid an adrenal crisis, although no data from randomized trials are available. A continuous infusion of 10 mg of hydrocortisone per hour or the equivalent amount of dexamethasone or prednisolone eliminates the possibility of GC deficiency. This dose can be halved the second postoperative day, and the maintenance dosage can be resumed the third postoperative day. However, it is important that with regard to possible detrimental effects and the possibility of decreased resistance to infections, this treatment should not be used for prolonged periods in the absence of evidence of corticosteroid insufficiency. General perioperative management should include avoidance of etomidate as an anesthetic drug (selection of other drugs and muscle relaxants is not influenced by the presence of treated hypocortisolism), infusion of sodium-containing fluids, minimal doses of any anesthetic drugs to avoid increased sensitivity to drug-induced myocardial depression, invasive monitoring of hemodynamics, glucose, and electrolytes, and decreased initial doses of muscle relaxant, monitoring the effect using a peripheral nerve stimulator. Especially when acute adrenal insufficiency has been detected in a critically ill patient with a previously unknown disorder, thorough diagnostics are demanded even after improvement.

Observations suggest that control of cortisol secretion in response to stress is more complex than originally thought. Interactions between corticotropin-releasing factor (CRF), vasoactive intestinal polypeptide, arginine vasopressin, catecholamines, and other hormones in the control of cortisol secretion have been described.¹¹² α₂-Adrenergic receptor antagonists (e.g., clonidine), which are widely used in ICUs, may suppress the cortisol response to surgical stress. On the other hand, increases in intracranial pressure stimulate cortisol release without increasing ACTH levels, and adrenalectomy but not adrenal demedullation increased the permeability of brain tissue to macromolecules.¹¹³ Further evidence also suggests that white blood cells may release ACTH-like peptides that can stimulate adrenal gland secretion of cortisol, and that primary adrenal insufficiency is associated with increases in serum levels of angiotensin-converting enzyme.¹¹⁴

There are multiple interactions between drugs and the HPA axis that have to be considered if absolute or relative adrenal insufficiency is suspected. Moreover, in patients with hepatic dysfunction, GC doses should be tapered, especially when using prednisone, since hydroxylation to the active component needs considerable metabolic capacity. Special attention is required in the concomitant use of GCs with other drugs, because of potential interactions and because some drugs may affect the metabolism of steroids, which may lead to a decreased or increased GC effect on their target tissues.^115,116 Glucocorticoids decrease blood levels of aspirin, coumarin anticoagulants, isoniazid, insulin, and oral hypoglycemic agents, whereas cyclophosphamide and cyclosporine levels may be increased. Inversely, antacids, carbamazepine, cholestyramine, colestipol, ephedrine, mitotane, phenobarbitone, phenytoin, and rifampicin decrease GC blood concentrations, whereas they are increased by cyclosporine, erythromycin, oral contraceptives, and troleandomycin. Furthermore, the combination of exogenous GC administration and amphotericin B, digitalis glycosides, and potassium-depleting diuretics may induce or worsen hypokalemia, warranting frequent monitoring of potassium levels. Finally, the general risk for immunosuppression by GCs precludes any use of vaccines from live attenuated viruses to avoid severe generalized infections.^115,116

Conclusions

Underproduction of adrenal hormones can lead to serious illness. Glucocorticoids play a critical permissive role in intermediary metabolism, are counter-regulatory in relation to insulin, modulate inflammatory and immune responses, and optimize cardiovascular and central nervous system function. Therefore, diseases with a primary adrenocortical dysfunction or those leading to secondary adrenal insufficiency may have severe sequelae, which often are life threatening. The concept of relative adrenal insufficiency in critically ill patients with functional disorders of the HPA axis has gained attention during recent years. Especially in patients with severe sepsis and septic shock, this phenomenon is suspected of having a major impact on severity of illness and prognosis. Both absolute adrenal insufficiency and RAI should be diagnosed by using adequate laboratory investigations. In most cases, testing the basal level of cortisol, combined with a short-term stimulation test with 250 µg ACTH, can identify the disease. In patients with critical illnesses, however, it continues to be difficult to diagnose RAI.

In cases of severe volume- and catecholamine-resistant shock with suspected adrenal crisis, immediate replacement therapy is indicated. If the diagnosis is questionable, dexamethasone should be administered to allow functional diagnostics. Once the diagnosis is made, hydrocortisone is the preferred drug, since it provides both gluco- and mineralocorticoid effects. After stabilization, the dose of GCs should be tapered down to a total of 20 to 35 mg hydrocortisone per day or equivalent analogs. The fundamental role of GCs in the stress response to infection, and increasing knowledge of the antiinflammatory and immunosuppressive pharmacodynamic profile, have been the rationale for its use in sepsis trials for decades. Timing, dosage, and duration of GC administration were adapted to different disease pathophysiologic models and had a major impact on outcome. Randomized controlled trials of high-dose GCs failed to improve outcome, leading to skepticism and avoidance of any GCs in septic patients by most ICU physicians over the years, with the exception of some special indications. However, recent randomized controlled trials with low doses of hydrocortisone in septic shock evoked a corticosteroid renaissance. Based on current data, an incremental increase of less than 9 µg/dL after a 250-µg ACTH stimulation test may be used in patients with severe septic shock to determine relative adrenal insufficiency, although this is still under discussion. Meanwhile, prolonged treatment of septic shock with low doses of corticosteroids is considered a therapeutic option to promote shock reversal if the patient does not respond to volume replacement and vasopressor therapy.

Key Points