Chapter 79. Adrenocortical Insufficiency

• Activation of the hypothalamic-pituitary-adrenal (HPA) axis with increased glucocorticoid activity is an essential component of the general adaptation to stress.
• Mediators released in patients with sepsis may either stimulate or impair the synthesis and action of cortisol via actions on the HPA axis and glucocorticoid receptor.
• A random cortisol level in a highly stressed ICU patient is currently the most useful test to assess the integrity of the HPA axis.
• A stress cortisol level of less than 25 μg/dL in a highly stressed patient (hypotensive, respiratory failure) is highly suggestive of adrenal failure.
• The incidence of adrenal failure may be as high as 60% in patients with septic shock.
• All patients with suspected HPA axis dysfunction should be treated with stress doses of hydrocortisone (100 mg q 8 IV) pending the results of diagnostic testing.
• A random cortisol level of less than 15 μg/dL or a level of less than 20 μg/dL post low-dose corticotrophin stimulation testing (low-dose ACTH test) in a non-hypotensive patient with unexplained fever, eosinophilia or altered mental status warrants a trial of treatment with stress doses of hydrocortisone.

The maintenance of life depends on the capacity of the organism to sustain its equilibrium via allostasis—the ability to achieve stability through change. Allostasis is critical to survival in stress situations.1 The three major regulatory systems of the body responsible for allostasis include the central nervous system (CNS), the endocrine system, and the immune system. These three regulatory systems are closely interrelated and orchestrate the stress response. Importantly, activation of the hypothalamic-pituitary-adrenal (HPA) axis with the release of cortisol is the cornerstone of the stress response.

The qualitative and quantitative aspects of the stress response depend on the nature and severity of the stressor and are time-dependent. With mild to moderate short-term stress there is general activation of the CNS, the endocrine system, and the immune system in an attempt to restore homeostasis. However, with severe and/or prolonged stress, this coordinated allostatic response may fail with the ultimate demise of the organism. Teleologically, the latter may be viewed as a mechanism to spare society of those overwhelmed by disease. However, these are the very patients who are treated in ICUs and who are trying to “cheat death.”

Cortisol, the primary glucocorticoid hormone in humans, is secreted by the adrenal glands. In healthy, unstressed persons, cortisol is secreted in a diurnal pattern under the influence of corticotropin (ACTH) released from the pituitary gland. Corticotropin secretion, in turn, is under the influence of hypothalamic corticotropin-releasing hormone (CRH), and both hormones are subject to negative-feedback control by cortisol. Aldosterone, the primary mineralocorticoid hormone in the body, is also secreted by the adrenal glands. Aldosterone secretion, however, is regulated primarily by the renin-angiotensin system and potassium levels. Circulating cortisol is bound to corticosteroid-binding globulin, with less than 10% in the free bioavailable form. With acute stress in the form of infection, surgery, trauma, burns, or illness, cortisol production increases by a factor that is roughly proportional to the severity of the stress. The increase in cortisol is due to the increased production of CRH and corticotropin and a reduction in the negative feedback from cortisol. Diurnal variation of cortisol secretion is lost. The increased adrenal glucocorticoid production is associated with a shift away from mineralocorticoid and androgen production (i.e., reprioritization).2 Indeed, while there is activation of the renin-angiotensin system in an attempt to maintain blood pressure, aldosterone secretion is decreased.3–6 The dissociation between renin-angiotensin and aldosterone may represent an additional adaptive mechanism designed to promote cortisol production in critically ill patients.3–6 Sustained high levels of ACTH may explain this finding because in normal subjects ACTH causes a biphasic aldosterone response with an initial increase followed by diminished production.3–5

Glucocorticoids exert their effects by binding to and activating a 90-kDa intracellular glucocorticoid receptor (GR) protein.7 All cells appear to have appreciable levels of GR. The GR is localized in the cytoplasm of the cell and translocates into the nucleus on ligand binding. In the absence of hormone, cytoplasmic GR is associated with a large protein complex that includes heat shock protein 90 (hsp-90) and hsp-56.8 This protein complex functions to maintain the GR in an inactive conformation that is competent for glucocorticoid binding. When activated by a ligand, GRs bind as dimers to glucocorticoid response elements (GREs) in target genes that then activate or repress transcription of the associated genes. Via this mechanism, glucocorticoids regulate gene transcription in every cell in the body. In addition to increased secretion of cortisol, a number of other mechanisms may increase cellular glucocorticoid activity. An acute decline in the concentration of cortisol-binding globulin, as well as decreased cortisol binding to cortisol-binding globulin, results in an increase in the free biologically active fraction of the hormone.9–13 In addition, the number of GRs may be upregulated acutely during stress. Sun and colleagues14 demonstrated increased expression of GR mRNA, increased expression of GRs, and increased binding activity of the GRs in skeletal muscle in a murine cecal ligation and perforation model. In this study, pretreatment of the animals with the GR antagonist RU-38486 prevented the increase in GR protein and mRNA levels seen in sepsis and abolished the sepsis-induced increases in GR binding activity. These data suggest that cortisol may increase synthesis of its own receptor directly via a positive-feedback loop. The collective result of adrenocortical adaptation to severe stress is a dramatic increase in cellular glucocorticoid activity. This activation is an essential component of the general adaptation to illness and stress and contributes to the maintenance of cellular and organ homeostasis. For example, adrenalectomized animals succumb rapidly to hemorrhagic and septic shock, whereas steroid replacement is protective against these challenges.15,16

In the acutely stressed individual, the metabolic, cardiovascular, and immunologic actions of cortisol play a critical role in mediating the stress response. Glucocorticoids increase blood glucose levels, facilitating the delivery of glucose to cells during acute and chronic stress. Glucocorticoids increase blood glucose concentrations by increasing the rate of hepatic gluconeogenesis and inhibiting adipose tissue glucose uptake.17 Hepatic gluconeogenesis is stimulated by increasing the activities of phosphoenolpyruvate carboxykinase and glucose-6-phosphatase as a result of binding of glucocorticoids to the GREs of the genes for these enzymes. Glucocorticoids also stimulate free fatty acid release from adipose tissue and amino acid release from body proteins. Proteolysis is activated via glucocorticoid upregulation of the calpain and ubiquitin-proteasome pathways. Major roles of these processes are to supply to the cell energy and substrate that are required for response to stress and repair from injury.

Glucocorticoids are required for normal cardiovascular reactivity to angiotensin II, epinephrine, and norepinephrine, contributing to the maintenance of cardiac contractility, vascular tone, and blood pressure. These effects are mediated partly by the increased transcription and expression of the receptors for these hormones.18,19 Glucocorticoids are required for the synthesis of N+,K+-ATPase and catecholamines. Glucocorticoid effects on the synthesis of catecholamines and catecholamine receptors are partially responsible for the positive inotropic effects of these hormones.20 Glucocorticoids also decrease the production of nitric oxide and vasodilatory prostaglandins, major vasorelaxants and modulators of vascular permeability.21–24

Glucocorticoids possess anti-inflammatory and immunosuppressive effects that are mediated through specific receptor mechanisms.25–28 Glucocorticoids influence most cells that participate in immune and inflammatory reactions, including lymphocytes, natural killer cells, monocytes, macrophages, eosinophils, neutrophils, mast cells, and basophils. Glucocorticoids decrease the accumulation and function of most of these cells at inflammatory sites. Most of the suppressive effects of glucocorticoids on immune and inflammatory reactions appear to be a consequence of the modulation of production or activity of cytokines (e.g., interleukin 1 [IL-1], IL-2, IL-3, IL-6, interferon-γ [IFN-γ], tumor necrosis factor α [TNF-α]), chemokines, eicosanoids, complement activation, and other inflammatory mediators (e.g., bradykinin, histamine, macrophage migration inhibitory factor). Glucocorticoids control mediator production predominantly through inhibition of transcription factors such as nuclear factor κ β (NF-κ β).27,28 This inhibition is mediated by induction of the I-κ β α inhibitory protein.29 Glucocorticoids also produce anti-inflammatory effects by enhancing release of anti-inflammatory factors such as IL-1 receptor antagonist, soluble tumor necrosis factor receptor, and IL-10.30,31 In addition, glucocorticoids block the transcription of mRNA for enzymes required for the synthesis of some mediators (e.g., cyclooxygenase 2, inducible nitric oxide synthase).

The HPA axis and the immune response are linked together via a feedback loop in which activated immune cells produce cytokines that signal the brain. Activation of the HPA axis by specific cytokines increases the release of cortisol, which, in turn, feeds back and suppresses the immune reaction (and further cytokine release).32 IL-1 and IL-6 administered peripherally increase HPA axis activity, increasing levels of CRH, ACTH, and glucocorticoids.33,34 Cytokines also affect the pituitary and adrenal cortex directly. IL-1, IL-6 and TNF-α stimulate the secretion of ACTH from cultured pituitary preparations, and IL-1 and IL-6 stimulate glucocorticoid production in cultured adrenal preparations.33,34

Cytokines, however, also suppress the HPA axis and impair GR function. Whereas acute elevation of IL-6 levels stimulates ACTH release, chronic IL-6 elevation may blunt ACTH release.35 In addition, TNF-α impairs CRH-stimulated ACTH release,36,37 and a number of clinical studies have reported inappropriately low ACTH levels in patients with severe sepsis and the systemic inflammatory response syndrome (SIRS).36,38,39 Indeed, Schroeder and coworkers40 reported similar circulating levels of ACTH in healthy controls and in patients with severe sepsis. In addition, plasma from patients with septic shock impairs synthesis of corticosteroids by adrenocortical cells.41–43 TNF-α and corticostatin have been demonstrated to inhibit adrenal gland function.37,44,45 Corticostatin (defensin) is a peptide produced by immune cells.46 Concentrations of corticostatin increase over 20-fold in animals with infection but not other forms of stress.47 TNF-α has been shown to reduce adrenal cortisol synthesis by inhibiting the stimulatory actions of ACTH and angiotensin II on adrenal cells.44,48,49 Proinflammatory cytokines have been shown to influence the number, expression, and function of GRs. IL-1 has been demonstrated to decrease GR translocation and transcription.50 The half-life of cortisol has been demonstrated to be prolonged in sepsis; this may reflect a decreased number of GRs or decreased affinity of the receptor for its ligand.43,51,52 While Sun and colleagues14 demonstrated increased expression of GRs in skeletal muscle in the early phase of experimental sepsis, Liu and colleagues53 demonstrated decreased expression of GRs following burn injury. It is particularly noteworthy that in the latter study the decreased expression of GRs was attenuated by treatment with both anti-TNF-α and anti-IL-1 monoclonal antibodies. Ali and colleagues54 reported a 40% decline in the number of glucocorticoid receptors in the liver of septic rats. The decline in hormone-binding activity was associated with a fall in glucocorticoid receptor mRNA. Decreased affinity of the GR from mononuclear leukocytes of patients with sepsis also has been reported.55,56 Reduced activity of gluconeogenic enzymes during endotoxemia despite elevated circulating glucocorticoid levels provides further evidence to support impaired intracellular actions of glucocorticoids during sepsis.57 In total, these data support the idea that mediators released in patients with sepsis may either stimulate or impair the synthesis and action of cortisol via actions on the HPA axis and the GR. The net effect of these opposing actions on the HPA axis and GR may be time-dependent and in addition depend on the severity of illness and extent of mediator production.

Since there are no clinically useful tests to assess the cellular actions of cortisol (i.e., end-organ effects), the diagnosis of adrenal insufficiency is based on measurement of serum cortisol levels; this has resulted in much confusion and misunderstanding.39,58–69 Traditionally, the “integrity” of the HPA axis has been assessed by the short corticotropin stimulation test (also known as the Cosyntropin stimulation test). This test is usually performed by administering 250 μg of synthetic corticotropin intravenously and obtaining a serum cortisol determination before and 30 and 60 minutes following administration.58,70 A 30- to 60-minute serum cortisol level below 18 μg/dL or an increase in the cortisol concentration of less than 9 μg/dL has been regarded by many as diagnostic of adrenal insufficiency.58 However, these criteria were developed to assess adrenal reserve in patients with destructive diseases of the adrenal glands and are based on responses in normal, nonstressed, healthy controls.58,70 We believe that the standard corticotropin stimulation test lacks sensitivity for the diagnosis of adrenal insufficiency58,71–74 (Table 79-1).

High doses of corticotropin (e.g., 250 μg) are supraphysiologic (over 1000-fold higher than normal maximal stress ACTH levels).39,62–65 The very high levels of corticotropin obtained with 250 μg can override adrenal resistance to ACTH and result in a normal cortisol response. Patients with normal responses to high-dose corticotropin may fail to respond normally to stress61,71 (Fig. 79-1). Owing to the decreased sensitivity of the high-dose ACTH stimulation test (HD-ACTH) for the diagnosis of adrenal insufficiency, many investigators evaluated the use of lower doses of ACTH that approximate stress levels of ACTH (i.e., 1 to 2 μg) for the diagnosis of adrenal insufficiency. A number of studies have demonstrated that a 1-μg dose (LD-ACTH) of corticotropin is more sensitive and specific for diagnosing primary and secondary adrenal insufficiency than the 250-μg dose.38,71,73,75–79

The change in cortisol level following corticotropin stimulation (Δmax) is used by some clinicians to diagnose adrenal insufficiency.59,63,64,80 However, the Δmax is a measure of adrenal reserve and not adrenal function. The increase in cortisol following administration of corticotropin should not be used as a criterion for the diagnosis of adrenal insufficiency. A maximally stressed patient may be secreting all the cortisol that his or her adrenal glands can synthesize. This patient may have an appropriately high serum cortisol level but be unable to respond further following corticotropin injection (no reserve). For example, a critically ill patient with a basal stress cortisol level of 54 μg/dL that increases to 57 μg/dL with corticotropin does not have adrenal insufficiency. It is the absolute level that is of importance rather than the Δmax. On the other hand, it is clear from a number of studies that impaired response to ACTH (i.e., small Δmax) has prognostic value and generally indicates a poor prognosis. The exact reasons for this are unclear but may relate to high levels of circulating factors that impair adrenal response to maximal stress.

Most important, the administration of exogenous ACTH bypasses the CNS-hypothalamic-pituitary axis and tests the integrity of the adrenal glands directly. It is essential that one evaluate the entire axis because defects in the hypothalamic-pituitary components frequently cause adrenal insufficiency. Endogenous stresses such as hypotension, hypoxemia, fever, and hypoglycemia are superior stimuli for testing the integrity of the HPA axis than is ACTH testing. These endogenous stressors test the function of the entire HPA axis and therefore should be regarded as the “gold standard” for adrenal testing. In other words, the HPA axis evolved to respond to stress and is best evaluated using stress that tests all components of the axis. Many inappropriately assume that the ACTH stimulation test is capable of detecting the presence of adrenal insufficiency because they assume that significant adrenal atrophy has occurred and results in diminished ACTH responsiveness. This concept is flawed for a variety of reasons. First, it takes 1 to 2 weeks for the adrenal gland to atrophy following pituitary removal when all ACTH secretion from the pituitary is lost. Thus the ACTH stimulation test may easily miss acute adrenal insufficiency (the most common form of adrenal insufficiency in critically ill patients). Second, adrenal atrophy may take longer in patients with partial adrenal insufficiency (common in critically ill patients). Third, the response to ACTH may be normal in patients with partial secondary adrenal insufficiency or ACTH resistance. ACTH testing is not required to diagnose adrenal insufficiency in severely stressed patients because the CNS-HPA axis already should be maximally activated. In such patients, a random stress cortisol level provides adequate information on the integrity of the entire HPA axis. In patients in whom the level of stress is less intense (not hypotensive, hypoxemic, or in pain), the LD-ACTH simulation test should be used to assess adrenal reserve (Fig. 79-2). One also may evaluate the pituitary-adrenal axis by administering CRH.40 This test bypasses the hypothalamus but does require the integrity of the pituitary and adrenal glands. However, the sensitivity and specificity of the test for detecting adrenal insufficiency in critically ill patients have not been determined.

There is much controversy regarding levels of circulating cortisol that are considered to be an adequate response to stress.58 Many textbooks and published manuscripts state that the normal circulating cortisol response to stress is a level greater than 15 to 20 μg/dL. However, the choice of 15 to 20 μg/dL is based primarily on the response to exogenous high-dose ACTH (250 μg) and the response to insulininduced hypoglycemia in nonstressed, non-critically ill patients.70 During surgical procedures such as laparotomy, serum corticotropin and cortisol concentrations rise rapidly, peaking in the immediate postoperative period, and then decline to baseline levels over the next 72 hours.81 The magnitude of the postoperative increase in serum cortisol concentration is correlated with the extent of the surgery, with a peak between 30 and 45 μg/dL in patients undergoing major surgery.82–87 During severe illness, serum cortisol concentrations tend to be higher than those of patients undergoing major surgery.81 In patients with multiple trauma, the serum cortisol level remains greater than 30 μg/dL for at least a week, with peak values between 40 and 50 μg/dL.88 Cortisol levels are increased in critically ill ICU patients, with the highest values being reported in patients with the highest illness severity scores and those with the highest mortality.89,90 We have found cortisol levels to be elevated in all forms of shock (Fig. 79-3). Interestingly, cortisol levels are significantly lower in patients with septic shock compared with patients with hypovolemic and cardiogenic shock, suggesting the presence of mediators that impair HPA axis function. Overall, the available data suggest that the degree of activation of the HPA axis and elevation in serum cortisol level is related to the severity and duration of the stressor. Animal and human studies demonstrate increasing serum levels of epinephrine and cortisol with increasing severity of stress, with hypotension and sepsis being two of the most intense stressors.91,92 Based on the available data,74 we believe that a random cortisol level in severely stressed patients (e.g., those with hypotension, hypoxemia, burn, high fever, multiple trauma) should be above 25 μg/dL. Higher levels may be appropriate in patients with septic shock due to “tissue cortisol resistance.”

The cortisol threshold of 25 μg/dL for an adequate adrenal response to severe stress is supported by two clinical studies that evaluated the hemodynamic response of replacement doses of hydrocortisone in vasopressor-dependent patients. While the hemodynamic response to steroids may be a crude marker of acute adrenal insufficieny, we have no other end-organ or biochemical marker of cellular glucocorticoid activity at this time. Furthermore, this end point has been used in other studies evaluating the role of corticosteroids in vasopressor-dependent patients with sepsis and may serve as a surrogate marker of inadequate glucocorticoid activity during severe illness.93–96 Rivers and colleagues97 studied the HPA axis in a group of vasopressor-dependent surgical patients. In a subgroup of patients treated with corticosteroids, the basal serum cortisol concentration was 49 μg/dL in the steroid nonresponders and 20 μg/dL in patients who were weaned from vasopressors within 24 hours of the initiation of steroid treatment. Only one patient in the steroid-responsive group had a baseline serum cortisol level of greater than 25 μg/dL, and only 2 nonresponders had a baseline level of less than 25 μg/dL. In this study, the diagnostic sensitivity of a threshold of 25 μg/dL was 90%, with a specificity of 92%, a positive predictive value of 60%, and a negative predictive value of 66%. Similarly, we investigated the hemodynamic response following treatment with replacement doses of hydrocortisone in 59 medical patients in septic shock.71 The baseline serum cortisol level was 14.1 ± 5.2 μg/dL in the steroid-responsive patients compared with 33.3 ± 18 μg/dL in the steroid-nonresponsive patients. The sensitivity of a baseline cortisol level of less than 25 μg/dL in predicting steroid responsiveness was 96%, this compared with 22% for the standard HD-ACTH stimulation test. The area under the receiver operating characteristic (ROC) curve of the stress (baseline) cortisol level was 0.84; a stress cortisol level of 23.7 μg/dL had the best discriminating power, with a sensitivity of 86%, a specificity of 66%, a likelihood ratio of 2.6, a positive predictive value of 62%, and a negative predictive value of 88%. Clearly, there is no absolute serum cortisol level that distinguishes an adequate from an insufficient adrenal response.

Based on current evidence, we believe that a random cortisol level should be interpreted in conjunction with the severity of illness, and 25 μg/dL is a useful threshold value for an appropriate response to severe stress (see Fig. 79-2). We believe that a random cortisol level should be above 25 μg/dL in severely stressed ICU patients with normal adrenal function. It is not necessary to obtain cortisol levels at a specific time of the day because critically ill patients lose the diurnal variation in their cortisol levels.82 In nonhypotensive critically ill patients, the normal cortisol response (30 to 60 minutes) after 1 to 2 μg corticotropin (LD-ACTH) should be a level greater than 20 μg/dL. We do not recommend performing the unphysiologic high-dose corticotropin stimulation test. Importantly, a random cortisol level of less than 15 μg/dL or a level less than 20 μg/dL post LD-ACTH test in a nonhypotensive critically ill patient with unexplained fever, eosinophilia or, metal status changes may warrant a trial of replacement doses of corticosteroids.

The incidence of adrenal insufficiency in critically ill patients is variable and depends on the underlying disease and severity of the illness. The reported incidence varies widely (0% to 77%) depending on the population of patients studied and the diagnostic criteria used to diagnose adrenal insufficiency.39,51,62–65,71,97–101 However, the overall incidence of adrenal insufficiency in critically ill patients approximates 30%, with an incidence as high as 50% to 60% in patients with septic shock.72 The following incidences of adrenal insufficiency apply to patient groups with no previous history suggestive of adrenal insufficiency (e.g., previous use of glucocorticoids). For example, using the criteria cited earlier (stress cortisol level <25 μg/dL), we diagnosed adrenal insufficiency in 36 of 59 (61%) patients with septic shock.71 Only 5 (9%) of these patients met the “classic criteria” (cortisol level <18 μg/dL 60 minutes after 250 μg corticotropin) for adrenal insufficiency. Rydvall and colleagues102 reported a 47% incidence of adrenal insufficiency in a general ICU population (using a stress cortisol threshold of 25 μg/dL). Briegel and colleagues99 reported 13 of 20 (65%) patients with septic shock having a stress cortisol level of less than 25 μg/dL. Sibbald and colleagues65 reported 20 of 26 (77%) septic patients with stress cortisol levels of less than 25 μg/dL. Moran and colleagues103 reported a 49% incidence of adrenal insufficiency in patients with septic shock. We reported a 50% to 75% incidence of adrenal insufficiency in critically ill HIV-infected patients.73. Interestingly, we found a 0% incidence of adrenal insufficiency in nonhypotensive septic patients, which contrasted with a 61% incidence in septic patients with hypotension.71 Patients with closed head injury have an incidence of adrenal insufficiency of approximately 15%, whereas patients with cardiogenic shock following myocardial infarction have an incidence of adrenal insufficiency of approximately 5% to 10%. In a subset of patients with acute gastrointestinal hemorrhage and hypotension, we found an incidence of adrenal insufficiency of only 2% to 4%. Unfortunately, the incidence of adrenal insufficiency has not been evaluated reliably in all groups of critically ill patients and clearly varies with the underlying disease.

Patients with chronic adrenal insufficiency usually present with a history of weakness, weight loss, anorexia, lethargy, slowed mentation, and general failure to thrive. Some patients complain of nausea, vomiting, abdominal pain, and diarrhea. Clinical signs include orthostatic hypotension and, rarely, hyperpigmentation (primary adrenal insufficiency). Laboratory testing may demonstrate hyponatremia, hyperkalemia, hypoglycemia, and a normocytic anemia.74,104 This presentation contrasts with the features of acute adrenal insufficiency (Table 79-2). Hypotension refractory to fluids and requiring vasopressors is the most common feature of acute adrenal insufficiency in critically ill patients.71,74,104 Patients usually have a hyperdynamic circulation, which may compound the hyperdynamic profile of septic patients.71 However, the systemic vascular resistance, cardiac output, and pulmonary capillary wedge pressure can be low, normal, or high.72 The variability in hemodynamics reflects the combination of adrenal insufficiency and the underlying disease. Importantly, acute adrenal insufficiency always should be excluded in critically ill patients requiring vasopressor support. CNS dysfunction is common, frequently compounded by the underlying disease. However, adrenal failure may present with altered mental status, particularly unexplained confusion, as the major clinical finding. Lethargy, weakness, anorexia, and abdominal complaints are common. Furthermore, in critically ill patients with unexplained fever, adrenal failure should be excluded. Laboratory assessment may demonstrate eosinophilia and hypoglycemia. Hyponatremia and hyperkalemia are uncommon. Many patients have a mild metabolic acidosis.

Acute adrenal insufficiency (Table 79-3) occurs in patients who are unable to increase their production of cortisol in the face of acute stress. This includes patients with hypothalamic and pituitary disorders (secondary adrenal insufficiency) and patients with destructive diseases of the adrenal glands (primary adrenal insufficiency). Secondary adrenal insufficiency is common in patients who have been treated with exogenous corticosteroids within the past year. Other causes of secondary adrenal insufficiency include pituitary infarction, hypothalamic disease, and sepsis.

Synthetic glucocorticoids are commonly used drugs. The use of these drugs is associated with suppression of the HPA axis. The degree of suppression depends on many factors, including the glucocorticoid potency, the dose, the dosing schedule, and the duration of use. The degree of suppression, however, generally is not predictable in any individual patient.105 The use of inhaled corticosteroids in asthmatics also has been associated with varying degrees of adrenal suppression.75,106 Systemic glucocorticoids probably do not cause significant HPA axis suppression when used for less than 5 days. When these drugs are used for between 5 and 30 days, the HPA axis will recover in most patients within 14 days of stopping treatment.107 However, when used for longer than 30 days, it may take up to a year for the HPA axis to recover. The recovery of the HPA axis can be assessed most reliably by measurement of a random cortisol level in a stressed patient. The degree of adrenal recovery in the unstressed patient can be determined by the response to 1 μg corticotropin.115 However, it is important to note that supraphysiologic doses of glucocorticoids suppress both CRH production in the hypothalamus and ACTH production in the pituitary gland. This suppression can outlast the duration of adrenal suppression.108 Therefore, a normal cortisol response to high-dose corticotropin does not predict conclusively a normal response to stress (which involves the hypothalamic-pituitary components of the axis).109 The properties of the corticosteroids in common use are listed in Table 79-4.

The most common cause of chronic primary adrenal insufficiency (Addison's disease) in the past was tuberculosis, followed by autoimmune disease. However, human immunodeficiency virus (HIV) infection and other infections in immunosuppressed patients (e.g., tuberculosis, CMV, fungal) are currently the most important causes of primary adrenal insufficiency. The adrenal gland is the endocrine organ most commonly involved in patients with the acquired immunodeficiency syndrome (AIDS).110–114 Human CMV has been demonstrated in the adrenal glands of 33 to 88 percent of patients who died of AIDS. Less commonly, tubercle bacilli, Cryptococcus neoformans, Toxoplasma gondii, Histoplasma capsulatum, lymphoma, hemorrhage, or Kaposi's sarcoma may involve the adrenal gland.113 In addition, a number of drugs used in patients with HIV infection, most notably ketoconazole and megestrol acetate, can impair adrenal function.115,116 Although the adrenal gland is commonly affected by opportunistic infections and tumor infiltration in AIDS, adrenal insufficiency in the outpatient setting is uncommon.113,117 However, these patients may be unable to increase the synthesis of cortisol in the face of stress. Using the revised diagnostic criteria (i.e., stress cortisol level <25 μg/dL), we reported adrenal insufficiency in 75% of critically ill HIV-infected patients who had not been treated with corticosteroids.73

Sepsis and the systemic inflammatory response syndrome (SIRS) are the most common causes of adrenal failure in the critically ill. There is increasing evidence of HPA axis insufficiency in critically ill septic patients, which appears to result from circulating suppressive factors released during systemic inflammation.39,62–65,71,80 It should be emphasized that in most cases adrenal failure results from suppression of the HPA axis and GR expression and is completely reversible, with restoration of normal adrenal function once the patient has recovered from the septic episode.99 In addition to sepsis, we have noted a high incidence of HPA axis failure in patients with liver failure. The exact cause remains unclear. However, we recommend screening all critically ill cirrhotic patients for adrenal failure.

We have reported three patterns of adrenal insufficiency in patients with septic shock based on their response to low-dose (LD) and high-dose (HD) corticotropin.71 Forty-two percent of patients failed to respond to both LD and HD corticotropin and most likely represent patients with primary adrenal gland failure. Twenty-eight percent of patients had low stress cortisol levels but responded to both LD and HD corticotropin; we postulate that these patients lacked adequate secretion of CRH or ACTH and represent patients with secondary adrenal insufficiency. In a previous study we demonstrated that 85% of the patients with adrenal insufficiency (baseline cortisol of less than 25 μg/dL) had inappropriately low (<40 pg/mL) ACTH levels.118 A third group of patients (30%) failed to respond to LD corticotropin but responded normally to HD corticotropin. Due to the ability of pharmacologic doses of corticotropin to produce a normal response, when physiologic stress levels of corticotropin were ineffective, we postulate that this group represents patients with corticotropin resistance. It is possible that these various patterns of HPA axis failure during critical illness represent the differential effects of inhibitory mediators at various levels of the HPA axis.

Patients with chronic critical illness may develop adrenal insufficiency while in the ICU. Although not evaluated in prospective trials, we have observed patients who had normal adrenal function when admitted to the ICU but who later developed adrenal insufficiency (e.g., long term ventilated patient with ARDS) (Fig. 79-4). The only apparent cause was a prolonged systemic inflammatory response. The adrenal insufficiency may have resulted from chronic secretion of systemic cytokines and other HPA axis suppressive substances.119 These patients have a higher mortality compared with patients who maintain adequate cortisol levels (see Fig. 79-4) and illustrate the importance of serial follow-up of adrenal function in long-term critically ill patients.

We believe that there is a bimodal distribution of mortality in relationship to the random cortisol level during sepsis. Patients with low cortisol levels (i.e., <25 μg/dL) who are not treated with corticosteroids and patients with very high levels (i.e., >45 μg/dL) have the highest mortality. This hypothesis may explain the apparent contradictory reports in the literature. Annane and colleagues59 reported a mean random cortisol level of 34 μg/dL in 189 patients with septic shock, with the nonsurvivors having higher levels than survivors (39 versus 28 μg/dL, respectively). However, in the study by Schroeder and colleagues,40 the mean random cortisol level was only 19 μg/dL, with nonsurvivors having a lower cortisol level than survivors (10 versus 17 μg/dL). Most of the patients included in the study by Schroeder and colleagues would have met our criteria for adrenal insufficiency. However, none of the patients was treated with corticosteroids.

Impaired responses to corticotropin and CRH are also associated with increased mortality.40,103 However, it remains unclear as to whether the impaired response is a direct contributor to the increased mortality or is secondary to hypothalamic-pituitary dysfunction or suppression (i.e., from circulating mediators or elevated cortisol levels).

Deficiency of cortisol is associated with increased morbidity and mortality during critical illness. Mckee68 randomized 18 critically ill patients with adrenal insufficiency to glucocorticoid treatment or placebo. One of eight (13%) steroid-treated patients died compared with 9 of 10 (90%) placebo patients. Evidence for high mortality from adrenal insufficiency in critically ill patients also comes from the report of Ledingham and Watt,120 who noted increased mortality from use of etomidate (a sedative agent that causes adrenal insufficiency) in multiple-trauma patients (44% etomidate versus 27% other sedatives). The report by Ledingham and Watt emphasizes that even slight impairment of the adrenal response during severe illness can be lethal. Rivers and colleagues97 reported faster weaning from vasopressors and improved survival in hydrocortisone-treated critically ill patients (79% versus 55%). We found that patients with septic shock and inappropriately low cortisol levels had improved survival when treated with hydrocortisone (65% versus 35%).

Further evidence to support the benefit of glucocorticoid treatment of acute adrenal insufficiency in patients with septic shock comes from the studies of Bollaert and colleagues94 and Briegel and coworkers.93 Bollaert and colleagues94 randomized 41 patients with septic shock to hydrocortisone (100 mg IV every 8 hours) or placebo. Although random cortisol levels were obtained, treatment with hydrocortisone was not stratified based on the levels. However, the glucocorticoid-treated patients had a significantly greater reversal of shock at 7 and 28 days and reduced 28-day mortality (30% versus 70%, p = 0.09) compared with the placebo group. Similarly, Briegel and colleagues93 randomized 40 critically ill septic shock patients to intravenous hydrocortisone or placebo. Hydrocortisone treatment was associated with improved shock reversal and decreased days of vasopressor support. There also was earlier resolution of organ dysfunction, shorter ventilator time, and shorter ICU stay. Annane and colleagues80,96 randomized 200 septic shock patients to steroid replacement or placebo. There was a significant 30% decrease in death in a subset of the steroid-treated patients. Oppert and colleagues95 treated 20 septic shock patients with hydrocortisone (10 mg/h for 7 days). Patients with “inadequate” steroid production were weaned from vasopressors significantly faster than patients with “adequate” steroid production. These studies of physiologic stress doses of glucocorticoids administered for many days contrast with earlier studies of high-dose glucocorticoids (i.e., 30 mg/kg methylprednisolone) administered for one to two doses. The short-term pharmacologic dose studies of glucocorticoids failed to report benefit in patients with septic shock.121,122

Interestingly, Schelling and colleagues123 evaluated the effect of hydrocortisone treatment during septic shock on the incidence of posttraumatic stress disorder. The administration of hydrocortisone at stress levels during septic shock reduced the incidence of posttraumatic stress disorder and improved emotional well-being in survivors.

The stress of major surgery may precipitate acute adrenal insufficiency in patients with inadequate adrenal reserve (i.e., adrenal crisis). This is especially true in patients with secondary adrenal insufficiency maintained on exogenous glucocorticoids. Prospective, randomized trials have failed to adequately evaluate the dose of glucocorticoid required in various perioperative settings. Thus recommendations for glucocorticoid coverage are based on a risk-benefit evaluation, published studies in the literature, and clinical experience. Importantly, patients with adrenal insufficiency undergoing surgery should receive appropriate stress doses of glucocorticoids before and after surgery. If surgery is prolonged (i.e., >4 to 6 hours), we recommend an additional dose of glucocorticoid during surgery. Patients with a high likelihood of impaired gastric emptying or impaired gut absorption should receive glucocorticoid repletion via the intravenous route until gut function has returned to relatively normal levels. We caution against the use of intramuscular glucocorticoid preparations in patients in whom hypoperfusion of muscle is likely. Patients also should receive sufficient steroid to control their underlying disease (e.g., asthma, autoimmune disease). Minimal doses of glucocorticoid for treatment of patients with suspected adrenal insufficiency based on type of surgery are as follows: First, patients undergoing minor surgery (e.g., hernia repair, laparoscopic cholecystectomy, knee surgery) should receive a minimal dose of 25 mg hydrocortisone equivalent prior to and 8 hours following surgery. This dose may be given orally if gut function is intact or intravenously. If patients demonstrate normal recovery, they may be weaned quickly from steroids or returned to maintenance levels. Second, patients undergoing moderate surgical stress (e.g., open cholecystectomy, partial colon resection, uncomplicated back surgery, hip replacement) should receive 25 to 50 mg hydrocortisone equivalent every 8 hours intravenously for 1 to 2 days. The dose then may be tapered to baseline levels based on clinical response. Third, patients undergoing major surgical stress (e.g., pancreatoduodenectomy, esophagectomy, total colectomy, repair for perforated bowel, cardiopulmonary bypass, ileofemoral bypass) should receive 50 to 100 mg hydrocortisone equivalent every 8 hours intravenously for 2 to 3 days. The dose then can be tapered to baseline doses based on clinical response. Importantly, the dose should be increased to maximal stress doses (100 mg hydrocortisone equivalent every 8 hours intravenously) in patients who remain hypotensive or deteriorate during recovery from surgery.124

In patients with severe stress (e.g., hypotension, hypoxemia, pain), a random (stress) serum cortisol level should be obtained (see Fig. 79-2). The patient should be started empirically on hydrocortisone (100 mg IV every 8 hours) pending results of testing. If the serum cortisol level returns less than 25 μg/dL, the hydrocortisone should be continued. In addition, if the patient has improved clinically with hydrocortisone and the cortisol level is greater than 25 μg/dL, we favor continuing the hydrocortisone for a few days (unless there is a specific contraindication). The dose of hydrocortisone should be tapered down toward maintenance doses as the patient's clinical status improves. If the level of stress is uncertain and adrenal insufficiency is suspected, we favor adrenal testing with the low-dose (1 μg) corticotropin stimulation test (see Fig. 79-2). If the corticotropin stimulation test cannot be performed immediately, administer dexamethasone (2 mg) and perform the test within the next 12 hours. Dexamethasone does not significantly crossreact with cortisol in the assay for cortisol and can be given to patients pending adrenal testing.

The currently recommended dosage schedule in adults includes 100 mg hydrocortisone intravenously every 8 hours or 50 mg intravenously every 6 hours.69,71,80,94,97 While a continuous infusion of hydrocortisone has been recommended, there are no data at this time to suggest that this dosing regimen is more efficacious than intermittent doing. In the study reported by Annane and colleagues,80 patients were treated with hydrocortisone (50 mg every 6 hours) and fludrocortisone (50 mg daily). However, as demonstrated previously, it appears that low aldosterone levels are part of the “normal” stress response, and there are no data to suggest that fludrocortisone provides any additional benefits over hydrocortisone alone. Furthermore, as illustrated in Table 79-4, hydrocortisone possesses mineralocorticoid activity. When given at high doses (e.g., 200 to 300 mg daily), hydrocortisone has adequate mineralocorticoid activity to treat primary adrenal failure.

HPA axis dysfunction is common in severely ill patients. Even slight impairment of the adrenal response to severe illness can increase morbidity and mortality, and we believe that low serum cortisol levels may be the cause rather than the consequence of poor outcome in these patients. Therefore, a high index of suspicion for adrenal insufficiency is required in all critically ill patients, particularly those with refractory hypotension. All patients with suspected HPA axis dysfunction should be treated with stress doses of corticosteroids pending the results of diagnostic testing.