| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Editorial |
National Institutes of Health, Bethesda, Maryland 20892
Address all correspondence and requests for reprints to: Lynnette Nieman, National Institutes of Health, Building 10, Clinical Research Center, 1 East, Room 1-3140, 10 Center Drive, Mail Stop Code 1109, Bethesda, Maryland 20892-1109. E-mail: NiemanL{at}nih.gov.
Although Cushing’s syndrome is uncommon, many of its characteristics are common in the general population, including obesity, depression, hypertension, and reproductive abnormalities, raising the question of who should be screened for the disorder. Elevated urine free cortisol (up to 4-fold over normal levels) also may be present in patients with psychiatric disease, obesity, alcohol abuse, and pain. Recently the term "pseudo-Cushing’s" has been used to identify such individuals with mild clinical and/or biochemical features of Cushing’s syndrome (1).
Because urine free cortisol cannot establish pathological hypercortisolism unless it is quite elevated, there is a need for additional tests to confirm the diagnosis of Cushing’s syndrome. The dexamethasone suppression test (DST) takes advantage of the fact that Cushing’s syndrome patients with primary adrenal disease or ectopic ACTH-secreting tumors do not respond to dexamethasone, whereas those with corticotrope adenomas generally show some suppression of cortisol, albeit less than normal individuals. The 1-mg DST (administration of dexamethasone between 2300 and 0000 h, with measurement of serum cortisol the next morning) has a high false positive rate but relatively high sensitivity (2). The 2-mg or low-dose DST (LDDST) (administration of 500 µg dexamethasone every 6 h with measurement of serum cortisol 2 or 6 h after the last dose) has high sensitivity and specificity (3). The utility of dexamethasone suppression is limited in individuals with altered metabolism of dexamethasone because of hepatic or renal disease or in those who take medications that alter CYP3A4 activity and dexamethasone clearance (4). Elevated corticosteroid-binding globulin (CBG) also can give a false positive result. Demonstration of a normal serum cortisol at midnight can exclude Cushing’s syndrome but is impractical (5). Measurement of late-night salivary free cortisol is more convenient, but there is considerable assay variability and no single diagnostic threshold has been validated (6, 7).
In 1993, Yanovski et al. (1) developed a combined low-dose dexamethasone suppression-CRH stimulation (Dex-CRH) test to distinguish patients with Cushing’s syndrome from those with pseudo-Cushing’s states. Using a serum cortisol of more than 1.4 µg/dl (38 nmol/liter), the test demonstrated high specificity, sensitivity, and diagnostic accuracy (8, 9). A later report from the same group showed that the test also distinguished healthy volunteers from individuals with mild Cushing’s disease (8, 9). Importantly, the LDDST had 100% specificity only in individuals with an appropriate dexamethasone level. Altogether eight of 59 patients with proven Cushing’s disease showed suppression of pre-CRH cortisol to less than 1.4 µg/dl (38 µg/dl) (sensitivity 86%) and each was properly characterized after CRH administration. These data suggested that the Dex-CRH test was useful in patients with Cushing’s disease who have falsely normal responses to dexamethasone, and that dexamethasone metabolism might play a role in false positive results.
In this issue, Erickson et al. (10) show that the Dex-CRH test at the Mayo clinic had maximal sensitivity and specificity of 90% using a 15-min cortisol threshold of 2.5 µg/dl (70 nmol/liter). An ACTH criterion gave slightly better diagnostic accuracy, but confidence intervals overlapped (10). This report of 21 patients with proven Cushing’s syndrome and 30 patients with presumed pseudo-Cushing’s syndrome follows three reports that challenge the nearly 100% diagnostic accuracy noted above. In 2006, Martin et al. (11) reported that 12 patients with Cushing’s syndrome were appropriately characterized by LDDST and the Dex-CRH test. Of 24 patients with suspected Cushing’s syndrome, three were misdiagnosed by the LDDST, and five by the Dex-CRH test. Using a cortisol diagnostic threshold of 1.9 µg/dl (53 nmol/liter), Gatta et al. (12) reported 100% sensitivity and 76% specificity with the Dex-CRH test. Subsequently, Pecori Giraldi et al. (13) reported 94% sensitivity and 90% specificity of the LDDST in 32 patients with Cushing’s syndrome and 23 patients with pseudo-Cushing states. However specificity after CRH was only 62% at the 1.4-µg/dl (38 nmol/liter) cutoff point, and 80% at 4-µg/dl (110 nmol/liter) point, whereas sensitivity fell from 100% to 94% when the threshold was increased (13). Overall, in 92 patients without Cushing’s syndrome in these reports, the specificity of the LDDST was 79% (95% confidence interval, 70–86%) compared with a 70% specificity for the Dex-CRH test (confidence interval, 60–78%). In 59 patients with Cushing’s syndrome, sensitivity was 96% for the LDDST and 98% for the Dex-CRH test.
Why do the results differ between these studies, and how do they affect how the test should be used in clinical practice? It is not uncommon for diagnostic accuracy to decrease as tests move from tightly controlled, often inpatient, research settings to clinics in academic centers and in the community (14). Other factors that must be considered include confounding variables, diagnostic endpoints, similarity of test protocols, and whether there is sufficient statistical power to draw conclusions.
What Are the Confounding Variables for Dexamethasone Tests?
A major confounder for these tests is the large variability in interpersonal metabolism of dexamethasone. For example, in healthy volunteers, dexamethasone levels 2 h after the last dose were 13.0 ± 6.1 µmol/liter (469.5 ± 220.4 µg/dl) (8). Dexamethasone is metabolized by the CYP3A4 enzymes, which can be induced by smoking and alcohol use, and by many medications including carbamazepine, nifedipine, phenobarbital, phenytoin, rifampin, St. John’s wort, tamoxifen, and topiramate (4). Measurement of plasma dexamethasone levels is vital to know whether levels are adequate. In our experience, low concentrations of dexamethasone frequently are associated with an abnormal result, and when the dexamethasone dose is increased to achieve adequate concentrations the test response normalizes. Additionally, missed doses or imperfect timing might cause an abnormal result and may be more likely with outpatient testing. Measurement of plasma dexamethasone is available in the United States from commercial laboratories and would help in these problem areas. Unfortunately, none of the recent reports provide this information.
Is the Test Endpoint Reliable?
The original serum cortisol criterion of 38 nmol/liter (1.4 µg/dl) falls in the less reliable part of the assay. The functional detection limit of the best cortisol assays is around 0.8 µg/dl (22 nmol/liter), and some assays cannot reliably measure levels less than 2 µg/dl (55 nmol/liter). Additionally, antibody-based assays generally result in higher values than structurally based assays such as HPLC, so that a cutoff point developed in one assay would not necessarily be the threshold derived from the same samples measured in a different assay (15). Thus, values near any published cutoff point should be viewed with some skepticism. The potential influence of assay differences is illustrated in the report of Martin et al. (11), in which five of the eight individuals with falsely abnormal results had cortisol values of 1.5–1.8 µg/dl (41–49 nmol/liter).
Theoretically, elevated CBG may cause a falsely abnormal response to dexamethasone, but this has not been formally evaluated. Although women taking estrogen were excluded from analysis in the present report, the CBG status of subjects in other reports is unknown.
Does the Study Have Sufficient Power?
Confidence intervals provide a rough guide to the power of the study and should be included when comparing the various tests within a study and in meta-analysis. For example, use of confidence intervals in the Erickson study showed that the higher diagnostic accuracy using the ACTH endpoint was not statistically different from results using the cortisol criterion. However, it is possible that a larger study might demonstrate a difference.
Is the Study Generalizable?
Are the Subjects Similar?
As noted above, the definitions of pseudo-Cushing’s states vary, so that subjects in studies may differ. For example, subjects in the Erickson study had either biochemical and/or clinical features of hypercortisolism, whereas those in the Yanovski study had both (1, 10). Some studies included more patients with psychiatric disorders than others. It is possible that the Dex-CRH test may perform differently in various conditions.
Are the Protocols of Similar Design?
Apart from potential cortisol assay differences, the amount of dexamethasone and the type of CRH used were different in the Martin study compared with the others (11). It is unclear whether this may have affected the results.
Were the Same Diagnostic Thresholds Used?
The optimal comparison of different studies of the Dex-CRH test would eliminate confounding influences of metabolism, assays, and protocol design and then evaluate the results. However, comparison of the diagnostic assessments (e.g. sensitivity, specificity) depends on the diagnostic threshold used; these criteria differed among the published studies. Overall, the original cortisol criterion of 1.4 µg/dl (138 nmol/liter) performed poorly compared with higher cutoff points.
How Do these Data Affect Use of the Dex-CRH Test in Clinical Practice?
Perhaps the most useful lesson is that diagnostic test performance tends to change over time as more patients are studied and the variables noted above influence the outcome. From a clinician’s perspective, this means that reports should be carefully scrutinized for possible confounders. The assay used in a study should be compared with the assay available to the clinician to judge whether the diagnostic criterion is applicable. Results close to the diagnostic cutoff point should be regarded with skepticism. The characteristics of the population studied should be considered when evaluating the results from any given patient. Intrinsic confounders (e.g. drug metabolism) should be kept in mind and ways to evaluate their effect (e.g. measurement of drug concentration) should be incorporated into the test protocol. Subsequent studies should be compared with each other in an ongoing analysis of the utility of a test.
The publication of these studies in this issue indicates that the diagnosis of Cushing’s syndrome continues to vex endocrinologists. The current epidemic of obesity suggests that this problem will continue, if not increase in magnitude. Ongoing investigation of the utility of diagnostic tests is important, particularly in populations that may be at increased risk, such as hypertensive, obese, diabetic, hirsute, or osteopenic individuals. Multisite studies should be considered to evaluate new promising tests for uncommon disorders so that they can be validated (or not) in less time.
Footnotes
This work was supported by the Intramural Program of the National Institute of Child Health and Human Development.
The author has nothing to disclose.
Abbreviations: CBG, Corticosteroid-binding globulin; Dex-CRH, combined low-dose dexamethasone suppression-CRH stimulation; DST, dexamethasone suppression test; LDDST, low-dose DST.
Received June 11, 2007.
Accepted June 19, 2007.
References
This article has been cited by other articles:
 |
|
 |
|
  M. Boscaro and G. Arnaldi Approach to the Patient with Possible Cushing's Syndrome J. Clin. Endocrinol. Metab., September 1, 2009; 94(9): 3121 - 3131. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Reimondo, S. Bovio, B. Allasino, S. De Francia, B. Zaggia, I. Micossi, A. Termine, F. De Martino, P. Paccotti, F. Di Carlo, et al. The combined low-dose dexamethasone suppression corticotropin-releasing hormone test as a tool to rule out Cushing's syndrome Eur. J. Endocrinol., November 1, 2008; 159(5): 569 - 576. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. K. Nieman Screening for Reversible Osteoporosis: Is Cortisol a Culprit? Ann Intern Med, October 16, 2007; 147(8): 582 - 584. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
