Published ahead of print on September 14, 2006, doi:10.1164/rccm.200509-1369OC
© 2006 American Thoracic Society doi: 10.1164/rccm.200509-1369OC
Diagnosis of Adrenal Insufficiency in Severe Sepsis and Septic ShockService de Réanimation and Service de Biochimie-Pharmacologie, Hôpital Raymond Poincaré, Faculté de Médecine Paris Ile de France Ouest, Garches; and Service de Biochimie Hormonale, Hôpital Saint Louis, Faculté de Médecine Saint Louis Lariboisière, Paris, France Correspondence and requests for reprints should be addressed to Djillali Annane, M.D., Ph.D., Service de Réanimation, Hôpital Raymond Poincaré (AP-HP), Faculté de Médecine Paris Ile de France Ouest (UVSQ), 104 Boulevard Raymond Poincaré, 92380 Garches, France. E-mail: djillali.annane{at}rpc.aphp.fr
Rationale: Diagnosis of adrenal insufficiency in critically ill patients has relied on random or cosyntropin-stimulated cortisol levels, and has not been corroborated by a more accurate diagnostic standard. Objective: We used the overnight metyrapone stimulation test to investigate the diagnostic value of the standard cosyntropin stimulation test, and the prevalence of sepsis-associated adrenal insufficiency. Methods: This was an inception cohort study.
Measurements and Results: In two consecutive septic cohorts (n = 61 and n = 40), in 44 patients without sepsis and in 32 healthy volunteers, we measured (1) serum cortisol before and after cosyntropin stimulation, albumin, and corticosteroid-binding globulin levels, and (2) serum corticotropin, cortisol, and 11
Conclusions: In sepsis, adrenal insufficiency is likely when baseline cortisol levels are less than 10 µg/dl or delta cortisol is less than 9 µg/dl, and unlikely when cosyntropin-stimulated cortisol level is 44 µg/dl or greater or
Key Words: corticol injection • free cortisol • corticotropin
Almost one century after the original description of apoplexy of the adrenal glands in septic shock (1), consensus is lacking on diagnostic criteria to define adrenal insufficiency in critical illness (2). In unstressed subjects, adrenal insufficiency is defined by a cosyntropin-stimulated cortisol level less than 18–20 µg/dl (3, 4). In critical illness, the diagnostic criteria for adrenal insufficiency have included a random cortisol level lower than 15 (2) or 25 µg/dl (5), or a cortisol increment after cosyntropin stimulation of 9 µg/dl or less (2, 6). In patients with severe hypoproteinemia, adrenal insufficiency may be defined by serum free cortisol level of less than 2.0 µg/dl at baseline or less than 3.1 µg/dl after cosyntropin stimulation (7). We have recently provided data underscoring the clinical significance of adrenal insufficiency in patients with septic shock (8–11). After 250 µg cosyntropin stimulation, patients with a cortisol increment of 9 µg/dl or less (nonresponders to cosyntropin) had vasopressor hyporesponsiveness (8), higher risk of death (9), and improved response to prolonged corticosteroid supplementation (10, 11). However, that nonresponders to cosyntropin had adrenal insufficiency remains controversial. In fact, the use of the cosyntropin stimulation test to assess adrenal function may present some variability (12), and may lead to misdiagnosing secondary adrenal insufficiency (13). More sensitive and cumbersome reference tests, such as insulin tolerance and metyrapone stimulation, have not been evaluated in intensive care unit (ICU) patients (2). Today, the use of insulin tolerance is impractical, because intensive insulin therapy has become a standard of care for ICU patients (14, 15), and septic shock is commonly associated with peripheral insulin resistance. For this reason, we used the overnight single-dose metyrapone stimulation test to investigate the diagnostic value of the standard 250-µg cosyntropin stimulation, and the prevalence of sepsis-associated adrenal insufficiency.
Study Population The CCPPRB de St. Germain en Laye approved the study, and healthy volunteers and patients or their relatives provided written informed consent before enrolment.
Patients with Sepsis
ICU Patients with No Sepsis Exclusion criteria for patients with sepsis and those without sepsis were as follows: age of less than 18 yr; pregnancy or breast-feeding; history of infection with human immunodeficiency virus; any known preexisting endocrine or liver disease (including any stage of cirrhosis, acute or chronic viral hepatitis, alcoholic liver disease, or hepatic tumors); any treatment with etomidate, glucocorticoids, estrogen, or any drug interfering with the hypothalamic–pituitary adrenal axis in the preceding 6 mo (4, 17). At study entry, the following parameters were recorded: time from ICU admission; age and sex; past medical history and estimated prognosis of any underlying disease, stratified according to the criteria of McCabe and Jackson (0, nonfatal; 1, ultimately fatal [i.e., < 5 yr]; or 3, rapidly fatal [i.e., < 1 yr]) (18); severity of illness, as assessed by the Simplified Acute Physiology Score (SAPS) II (scores can range from 0 to 163, with higher scores indicating higher risk of death) (19); the Sepsis-Related Organ Failure Assessment (SOFA) score (score can range from 0 to 24, with scores for each organ system [respiratory, hematological, hepatic, cardiovascular, neurological, and renal] ranging from 0 [normal] to 4 [most abnormal]) (20); and vital signs. Laboratory measurements included arterial blood gas, with the ratio of the partial pressure of arterial oxygen to the fraction of inspired oxygen, lactate levels, blood and urinary electrolytes concentration, total white blood cells, eosinophil and platelet counts, and serum levels of albumin and glucose.
On ICU admission, patients had blood samples drawn before and 60 min after a 250-µg cosyntropin administration. Later that day (> 8 h), patients received an overnight single-dose metyrapone stimulation test (30 mg/kg of the adrenal 11
Reference Standard for the Diagnosis of Adrenal Insufficiency and Nonresponders to Cosyntropin
Hormonal Assays
Statistical Analysis
Patients Characteristics A total of 69 patients with sepsis were eligible for inclusion in the study from February 2002 to May 2004, and 43 from December 2005 to April 2006. A total of 11 patients (8 and 3 in the first and second study periods, respectively) were excluded: 2 with refractory shock died before a cosyntropin-stimulation test could be performed, 8 did not tolerate enteral nutrition and could not receive metyrapone, and 1 did not consent (Figure 1). A total of 44 ICU patients without sepsis were enrolled, including 11 patients with drug overdoses, 9 with acute cardiogenic pulmonary edema, 5 with smoke inhalation–induced acute lung injury, 5 with status epilepticus, 4 with acute myocardial infarction, 4 with acute exacerbations of chronic obstructive pulmonary disease, 3 with keto-acidosis, and 3 with acute pulmonary embolism. As compared with patients without sepsis, patients with sepsis were older, were more likely to have fatal underlying comorbidities, mechanical ventilation, or vasopressor therapy, had higher SAPSII and SOFA scores, temperature, and hospital mortality rates, and greater lengths of stay (Table 1).
Hormonal Investigations Metyrapone was well absorbed in all subjects, and plasma metyrapone levels (see Figure E1 in the online supplement) and 8:00 A.M. cortisol levels (Table 2) were comparable between patients and healthy control subjects. The two groups with sepsis had similar albumin, corticotropin, and hormonal levels (Table 2). In comparison with healthy controls, both septic and patients without sepsis had lower albumin (p < 0.001), CBG (p < 0.01), and corticotropin (p = 0.02) levels, higher baseline total (p = 0.03) and free (p = 0.04) cortisol levels, and higher stimulated free cortisol (p = 0.02). In comparison with patients without sepsis, more patients with sepsis had low albumin levels (p = 0.01).
More patients with sepsis (31/61 [59% from the original cohort with sepsis] and 24/40 [60% of the validation cohort with sepsis]) met the criteria for adrenal insufficiency with the metyrapone test than patients without sepsis (3/44; 7%; p < 0.001 for both comparisons) (Table 3). In addition, 29/36 (80%) and 16/24 (67%) patients with sepsis had secondary adrenal insufficiency, as did 1/3 (33%) patients without sepsis (p = 0.02 for both comparisons). Similarly, more patients with sepsis had total cortisol levels of less than 15 µg/dl than did patients without sepsis.
Among the original cohort with sepsis, in comparison with patients without adrenal insufficiency, those with adrenal insufficiency were more often vasopressor dependent (p < 0.01), had higher SOFA cardiovascular scores (p = 0.02), positive blood cultures (p = 0.001), gram-negative sepsis (p < 0.01), and greater risk of in-hospital death (relative risk, 2.66; 95% confidence interval [CI], 1.17–6.06; see Table E1). At baseline, patients with sepsis with adrenal insufficiency had lower free cortisol levels than patients with sepsis without adrenal insufficiency (p = 0.04) and patients without sepsis (p = 0.04; Table E2). After cosyntropin stimulation, patients with sepsis with adrenal insufficiency had lower free cortisol levels (p = 0.03) and lower  total (p = 0.03) and free (p = 0.04) cortisol than those without adrenal insufficiency.
Diagnostic Value of Baseline, Cosyntropin-stimulated, and Absolute Increment of Total and Free Cortisol Levels
A baseline total cortisol level less than 10 µg/dl, or cosyntropin-stimulated increments in total cortisol less than 9 µg/dl and free cortisol less than 2 µg/dl, were strong predictors of adrenal insufficiency, having positive likelihood ratios equal to infinity, 8.46 (95% CI, 1.19–60.25), and 9.50 (95% CI, 1.05–9.54), respectively (Table E3). Furthermore, the combination "baseline total cortisol level less than 10 µg/dl or total cortisol less than 9 µg/dl" was the stronger predictor of the presence of adrenal insufficiency. A cosyntropin-stimulated total cortisol level of 44 µg/dl or greater, and a cosyntropin-stimulated increment in total cortisol of 16.8 µg/dl or greater were predictive of the absence of adrenal insufficiency (Table 4). In the "validation cohort with sepsis," this combination correctly classified the adrenal function in 34/40 patients, with a sensitivity of 0.83 (95% CI, 0.74–0.95), a specificity of 0.88 (95% CI, 0.74–1.00), and a positive likelihood ratio of 6.67 (95% CI, 1.80–24.68) (Table 5 and Figure 3).
Finally, in a multiple logistic regression analysis, the strongest independent predictor of adrenal insufficiency was positive blood cultures, with an odds ratio of 10.2 (95% CI, 1.8–57.2).
In the present study, adrenal insufficiency was identified in 60% of patients with sepsis, and was associated with a greater likelihood of vasopressor dependency, severe cardiovascular dysfunction, and higher risk for in-hospital death. Using the overnight single-dose metyrapone stimulation test as a reference, the combination of "baseline cortisol level less than 10 µg/dl or a cosyntropin-stimulated total cortisol increment less than 9 µg/dl" was the best predictor of adrenal insufficiency. By contrast, the combination "cosyntropin-stimulated cortisol level of 44 µg/dl or greater, and an increment in total cortisol of 16.8 µg/dl or greater" excluded the diagnosis of adrenal insufficiency. The observed prevalence of adrenal insufficiency in sepsis was higher than previously thought for critically ill patients (23). However, this was the first time that adrenal function in ICU patients was investigated using a test that assesses the whole hypothalamic–pituitary–adrenal axis. All previous studies were based on random cortisol levels or rapid corticotropin tests (2) that may underdiagnose adrenal failure in comparison with metyrapone testing (13). It is known that sepsis-induced cytokines may blunt the hypothalamic pituitary axis (17). In addition, recent data suggests that septic shock is associated with inducible nitric oxide synthase–induced neuronal apoptosis in the hypothalamus, which, in turn, may result in secondary adrenal failure (24). All healthy control subjects, and most of the ICU patients without sepsis, had normal metyrapone tests. In addition, patients with septic shock without adrenal insufficiency had responses to metyrapone that mimicked those of both healthy and critically ill control subjects. The time window of more than 8 h left between the ACTH and metyrapone tests allowed the avoidance of interference between the two tests, as previous studies have shown that cortisol levels returned to baseline values around 6 h after a 250-µg dose of ACTH (25). The overnight metyrapone test was feasible in almost all of the 112 screened patients with sepsis, with full absorption of the drug and sufficient inhibition of cortisol synthesis, excluding false-negative tests. All patients subsequently received corticosteroid replacement for at least 24 h, and tolerated the metyrapone test well. In agreement with others (2), we found that fever, tachycardia, hypotension, multiple organ dysfunction, hyponatremia, hypoglycemia, or increased eosinophil count were inadequate to diagnose adrenal insufficiency. To the best of our knowledge, the strong association between bacteremia and the presence of adrenal insufficiency has not been previously reported. Until additional studies are available, clinicians should consider adrenal function testing in patients with bacteremia.
Previous studies proposing a baseline cortisol level less than 15 µg/dl as a diagnostic criterion for adrenal insufficiency in critically ill patients included a limited number of patients with sepsis (2, 26). In the present study, on a larger and more homogenous group of patients with sepsis, a total cortisol level less than 10 µg/dl more accurately predicted adrenal insufficiency, in keeping with findings obtained from a cohort of ICU patients with confirmed adrenal insufficiency (27). The present study confirms the diagnostic value of a
Although most of our patients had septic shock, our findings suggest that adrenal insufficiency might be underappreciated in patients with severe sepsis. At present, little data are available in the literature on the incidence of impaired adrenal function in patients without septic shock. Similar to patients with septic shock and adrenal insufficiency (10), a recent randomized study reported a positive response to prolonged glucocorticoid supplementation in patients with severe community-acquired pneumonia; however, adrenal function was not tested (28). Another recent, randomized, placebo-controlled, double-blind trial showed a high prevalence of adrenal failure, as defined by a In summary, physicians should systematically search for adrenal insufficiency in severe sepsis or septic shock, especially when blood cultures are positive. Patients with a baseline total cortisol level less than 10 µg/dl, or a cortisol increment after cosyntropin less than 9 µg/dl, are very likely to have adrenal insufficiency. Conversely, in patients with a cosyntropin-stimulated total cortisol level of 44 µg/dl or greater, or a cortisol increment after cosyntropin stimulation of 16.8 µg/dl or greater, adrenal insufficiency can be ruled out. When the baseline cortisol level is between 10 and 44 µg/dl, and the cortisol increment after cosyntropin stimulation is between 9 and 16.8 µg/dl, assessment of adrenal function requires metyrapone testing.
Supported by a grant from Délégation Régionale à la Recherche Clinique, Ile de France–Assistance Publique–Hôpitaux de Paris (Ger-Inf-05R2 from Groupe d'Etude et Recherche sur le Médicament [GERMED]). The study sponsor had no responsibility for the study design, data analysis and interpretation, or decision to submit this manuscript for publication. This article has online supplement, which is accessible from this issue's table of contents at www.atsjournal.org Originally Published in Press as DOI: 10.1164/rccm.200509-1369OC on September 14, 2006 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form September 3, 2005; accepted in final form September 8, 2006
Related articles in AJRCCM:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
-deoxycortisol levels before and after an overnight metyrapone stimulation. Adrenal insufficiency was defined by postmetyrapone serum 11
2 (or Fisher's exact tests) for categorical data, and the Mann-Whitney U test for numerical data. The groups were compared with the use of analysis of variance (with subsequent Bonferroni tests where appropriate) for numerical data, and the Pearson 
