Key Point:

• Adrenal function in critically ill patients can be evaluated with the ACTH stimulation test or with a random cortisol level; at present which test is preferable is the subject of much debate.

ORLANDO—Is the adrenocorticotropin hormone (ACTH) stimulation test better than a random plasma cortisol measurement for evaluating adrenal function in the critically ill? Or is the much simpler random measurement more practical for use in the ICU? This question was debated during the recent annual meeting of the Society of Critical Care Medicine.[1]

Speaking in favor of the ACTH test was Djillali Annane, MD, who is Head of the Intensive Care Unit at the Raymond Poincare Hospital in Garches, France. “Increased [plasma] cortisol levels in acutely ill patients only mean that either cortisol clearance is impaired or there is tissue resistance to cortisol,” argued Dr. Annane.

By eliciting a cortisol response to stimulation, the ACTH test can identify patients with septic shock who are unable to tolerate any further stress, he said. In studies of such patients, those who responded with a cortisol increase of 9 µg/dL or higher were more likely to survive than were those with a less robust cortisol response; a direct correlation was also seen between the size of the posttest change in cortisol level and the probability of survival. In contrast, a wide range of baseline cortisol values have failed to discriminate survivors from nonsurvivors in critically ill populations.

In such populations, a less than 9-µg/dL increase in cortisol concentrations after an ACTH test appears to be linked both to extremely high systemic levels of inflammatory cells and to impaired mean arterial pressure responsiveness to norepinephrine. Furthermore, in large trials of acutely ill patients, only those who had no cortisol response to ACTH experienced marked improvement after corticosteroid administration.

Thus, with the use of this diagnostic test, “you can identify patients who are more likely to be vasopressor dependent, more likely to have overwhelming systemic inflammation, and more likely to respond to steroid replacement,” Dr. Annane asserted. A cutoff value between 4 and 9 µg/dL is 100% specific and clearly rules out adrenal insufficiency regardless of the baseline cortisol measurement, he said.

Dr. Annane’s final recommendations: “Take a random cortisol level. If a random cortisol is less than 15 µg/dL, you do not need an ACTH test,” he advised. “In stressed patients, lower cortisol levels are likely due to adrenal insufficiency; we have to treat these patients,” he added.

An ACTH test may be necessary, however, when a random cortisol level is above 15 µg/dL; this is particularly true in a patient who is not responding to therapy. Adrenal insufficiency is likely in this setting if the ACTH stimulation produces a cortisol increase of less than 9 µg/dL. Corticosteroid administration should be considered, noted Dr. Annane, in critically ill patients with adrenal insufficiency whose response to the ACTH test is blunted.

IN DEFENSE OF RANDOM CORTISOL MEASUREMENTS

“Neither the absolute nor the percentage changes [in cortisol levels] from baseline are useful in the diagnosis of adrenal insufficiency,” related Paul Marik, MD, who is currently a Professor of Critical Care Medicine at the University of Pittsburgh School of Medicine.

Furthermore, assessing these changes with the ACTH test has certain problems. “First of all, it is intuitively obvious that [the test] bypasses the hypothalamus and the pituitary,” Dr. Marik pointed out. Second, he added, the serum cortisol levels produced by 250 µg of ACTH, the amount used in the standard version of the test, is a thousand-fold higher than the maximum levels that can be achieved with intense physiologic stress.

Most important, about half of healthy volunteers and stressed patients without evidence of hypothalamo-pituitary-adrenal disease respond to the ACTH test with less than a 9-µg/dL change in cortisol levels. “So, my colleague Dr. Annane is asking us to use a test which has a diagnostic specificity similar to flipping a coin,” said Dr. Marik.

In patients who survive critical illness, a linear (albeit weak) relationship has been reported between cortisol measurements and APACHE scores, whereas such a relationship between those two variables has not been found in patients who die.

PUTTING THE MEASUREMENTS TO THE TEST

In a recent study, Dr. Marik assessed the ability of random cortisol measurements to detect adrenal insufficiency in 59 septic shock patients who were receiving 100 mg of hydrocortisone every eight hours. The diagnostic accuracy of a baseline cortisol level below 25 µg/dL was compared with that of the standard 250-µg and low-dose 1-mg ACTH tests. Patients were considered to be corticosteroid responsive if pressor therapy could be discontinued within 24 hours of the first dose of hydrocortisone.

According to Dr. Marik, when random cortisol levels were measured, 61% of the patients met the criteria for adrenal insufficiency; only 22% did so when the low-dose ACTH test was used, and just 8% did so when the standard ACTH test was employed. Of the 22 patients who were found to be responsive to corticosteroid administration, 95% had had a baseline cortisol level of less than 25 µg/dL. In contrast, the low-dose and standard ACTH tests produced positive results in only 54% and 22% of these patients, respectively.

In a statistical analysis, a 23.7-µg/dL cortisol level was the most accurate random measurement for determining the hemodynamic response to corticosteroids. The study findings were consistent with the results of other research, Dr. Marik noted.

In highly stressed critically ill patients who are hypotensive, hypoxic, and septic, a random cortisol level should be above 25 µg/dL, concluded Dr. Marik. “In mildly to moderately stressed patients, a random level of less than 15 µg/dL would be abnormal,” he added. “However, if one was uncertain as to the level of stress, one could do the low-dose [ACTH] test just to test adrenal responsiveness.”

—Timothy Begany

Reference
1. Farmer C, Annane D, Marik P. Pro/con: Diagnosing adrenal failure in the critically ill. Presented at: annual meeting of the Society of Critical Care Medicine; February 24, 2004; Orlando, Fla.

Return to table of contents