SAN DIEGO—The overall incidence of adrenal insufficiency (AI) in critically ill patients is about 30% to 40% and may be as high as 60% in those with septic shock, although statistics vary with the severity of illness and the diagnostic criteria used. The condition remains underrecognized largely because current diagnostic approaches are not sensitive enough to detect it, according to Paul Marik, MD, who gave a presentation on the subject at the American College of Chest Physicians’ recent annual meeting.[1]

The gold standard for testing for AI in critically ill patients should be a random serum cortisol, and that level should be above 25 µg/dL, asserted Dr. Marik, Professor of Critical Care Medicine at the University of Pittsburgh. In his presentation, Dr. Marik reviewed much of the controversy surrounding the definition of and appropriate tests for AI, as well as symptoms of the condition and glucocorticoid treatment.

SERUM CORTISOL: HOW LOW IS TOO LOW?

When the hypothalamic-pituitary-adrenal (HPA) axis is activated by a stressor, such as hypotension, fever, or pain, cortisol is released as part of the “fight or flight” stress response, which also serves to limit host-mediated tissue damage. Precisely what constitutes a sufficient stress cortisol level is, however, controversial at present.

Many published sources list a serum cortisol level above 18 to 20 µg/dL as a normal response to stress.[2] According to Dr. Marik, though, a cutoff value of 18 is too low for an accurate diagnosis of AI. “Studies show that during the stress of surgery,” he argued, “patients consistently increase serum cortisol to above 25 and often above the 30 µg/dL threshold.” In major surgery, levels peak between 30 and 45 µg/dL.[2]

He continued by pointing out that in patients with severe illness, such as sepsis, “serum cortisol concentrations tend to be higher than those [in] patients undergoing major surgery.” In severe sepsis, cortisol levels often remain elevated for the duration of the illness, instead of declining over the course of a few days as they do after surgery.

Circulating cortisol levels therefore reflect the severity of the stressor, and hypotension and severe sepsis as seen in critical illness are two of the most intense stressors. Thus, Dr. Marik urged using a random cortisol stress level of 25 µg/dL for the diagnosis of AI. “Stress cortisol is the gold standard for assessing adrenal function,” he said.

DEBUNKING THE 18 µG/DL CUTOFF

Many traditional methods of testing for AI are either inappropriate or not sensitive enough to detect the condition. Historically, the cutoff of 18 µg/dL for diagnosis was based on patients’ responses to the exogenous high-dose adrenal corticotropic hormone (HD-ACTH) stimulation test, which requires administration of 250 µg of synthetic corticotropin. Cortisol levels below 18 µg/dL or a change in cortisol levels of less than 9 µg/dL were considered indicative of AI.

Several problems exist with this test, Dr. Marik contended. First, these criteria were originally established in nonstressed patients, most of whom had tuberculosis.

Most notably, the high dose of synthetic corticotropin does not reflect any pathophysiologic condition—the amount exceeds the maximum possible ACTH levels achieved during stress by 1,000 times. “It is akin to giving a patient 20,000 units of insulin,” Dr. Marik stipulated.

Because of these extremely high levels of corticotropin, ACTH resistance may be masked by a cortisol response to the megadose given in the test. Patients may be able to respond to these extreme levels but still may have an insufficient response to normal stress levels.

In addition, measurement of a change in cortisol levels of less than 9 µg/dL is “clinically meaningless,” according to Dr. Marik. Cortisol measurements need to be evaluated in a situational context depending on the baseline level. In a severely stressed patient, for example, if the initial cortisol reading is 45 µg/dL and that level increases following HD-ACTH testing to 48 µg/dL, the patient does not have AI but rather is responding appropriately. “A stressed patient may not be able to increase those levels further,” he said.

On the other hand, misdiagnosis can also occur when the requisite change in cortisol level of 9 is met. For example, a patient’s cortisol levels increase from 9 to 18 µg/dL. According to the classic definition, a change in cortisol of 9 or more indicates that the patient was responding appropriately. However, in this case, the baseline level itself would be insufficient for a critically ill patient.

In all, “the ACTH test is a physiologically meaningless test to perform in critically ill patients,” stated Dr. Marik. He added, “Hopefully, after this lecture, none of you will do this test anymore. If you do, then I’ve failed in my mission.”

HPA AXIS FAILURE, NOT SINGLE GLAND FAILURE

Perhaps the most important limitation of the HD-ACTH test is that it bypasses the hypothalamus and pituitary to test the adrenal gland directly. In patients with sepsis, “the entire axis fails, not just the adrenal gland,” Dr. Marik explained.

He outlined three distinct patterns of adrenal failure, denoting different types of HPA axis malfunction. In primary adrenal failure, patients with low cortisol levels don’t respond to either the HD-ACTH or the low-dose (LD) ACTH test. (In the LD-ACTH test, 1 µg of corticotropin is given instead of 250 µg.) This indicates a problem with the adrenal gland.

In the second group, patients with low baseline serum cortisol levels do respond to both the HD- and LD-ACTH tests, which suggests a failure of the HPA axis. Finally, ACTH resistance is characterized by low initial cortisol levels and a patient response only to the HD-ACTH test. “Adrenal insufficiency is a heterogeneous disease,” Dr. Marik summarized.

SYMPTOMS AND TREATMENT

It is essential to recognize the signs of AI as “even slight impairment of the adrenal response to severe illness can increase morbidity and mortality.”[2] Because sepsis is one of the most common causes of AI, those patients are at high risk and should be closely monitored.

“Hypotension is the key indicator of adrenal failure,” stressed Dr. Marik. Other clinical features include hypoxia, unexplained fever, altered mental state, and eosinophilia. This condition is often reversible with treatment of the underlying disease.

Glucocorticoids have proven to be beneficial in reversing the effects of AI and in decreasing mortality. In one study, hydrocortisone administration resulted in faster weaning from vasopressors and improved survival (79% vs 55% in nontreated patients).[3] In another study, 40 patients in septic shock received either hydrocortisone or placebo. Hydrocortisone administration was associated with reversal of shock, fewer days on vasopressor support, earlier resolution of organ dysfunction, and shorter time spent on mechanical ventilation and in the ICU.[4]

Thus, when critically ill patients are hypotensive or display other symptoms of AI, Dr. Marik advised obtaining a baseline serum cortisol level and administering hydrocortisone pending the results of the test. The dose can be tapered as the patient’s condition improves and symptoms resolve. But he stressed that recognition is the key to successful treatment, “The index of suspicion for adrenal insufficiency should always be high in critically ill patients.”

—Lisa Pallatroni

References
1. Marik PE. Adrenal insufficiency in critical illness. Presented at: American College of Chest Physicians Conference; November 6, 2002; San Diego, Calif.
2. Marik PE, Zaloga GP. Adrenal insufficiency in the critically ill: a new look at an old problem. Chest. 2002;122:1784-1796.
3. Rivers EP, Gaspari M, Abi Saad G, et al. Adrenal insufficiency in high-risk surgical ICU patients. Chest. 2001;119:889-896.
4. Briegel J, Forst H, Haller M, et al. Stress doses of hydrocortisone reverse hyperdynamic septic shock: a prospective, randomized, double-blind, single-center study. Crit Care Med. 1999;27:723-732.

Return to table of contents