The Diagnosis of Cushing’s Syndrome: An Endocrine Society Clinical Practice Guideline

The clinical and biochemical diagnosis of Cushing’s syndrome (CS) presents a challenge to both clinicians and laboratories. In recent years, there have been several attempts to reach consensus on the management of CS, either by literature review or by drafting consensus guidelines. In 2003, the European Neuroendocrine Association and the Italian Society of Endocrinology reached a consensus on the diagnostic criteria, management and complications of CS at a workshop in Ancona, Italy.¹ Three years later, Findling and Raff reviewed the literature from 1996 to 2006 on the important issues in the diagnosis and management of CS.²

Current Clinical Practice Guideline

The American Endocrine Society and the European Society of Endocrinology jointly published an evidence-based clinical practice guideline on the diagnosis of CS in 2008.³ This guideline was developed by a six-member task force and focused on the role of biochemical investigations in the diagnosis of CS. The main investigations discussed were urinary free cortisol (UFC), late-night salivary cortisol, overnight 1 mg and two-day 2 mg/day low dose dexamethasone suppression tests (DST). Although the complete article comprises 14 pages, a convenient summary of recommendations is listed on the first two pages, and an algorithm for testing is included. The quality of evidence supporting each “recommendation” is shown, and where applicable, rationale behind the decision making and “remarks” on the technical aspects of testing are provided.

This guideline acknowledges there is no single best test for the screening of CS. However, a tailored investigative approach is suggested for specific clinical scenarios such as pregnancy, epilepsy, renal failure, cyclical CS and adrenal incidentaloma.

The proposed biochemical investigations involve the measurement of cortisol in serum, urine or saliva matrices. The authors highlighted the problems created by differences in assay methodologies, sensitivities and specificities and potential interferences from corticosteroid metabolites and synthetic glucocorticoids. When assisting clinicians in the choice of laboratory investigations and subsequent interpretation of results with respect to diagnostic cut-offs or reference intervals, it is prudent that chemical pathologists and biochemists are mindful of the pitfalls discussed in the guidelines.

Who Should be Tested?

Iatrogenic CS arising from the use of exogenous corticosteroids (oral, rectal, inhaled, topical or injections) must be excluded prior to testing. In addition, it should be remembered that illnesses other than CS are associated with hypercortisolism, notably alcoholism, poorly controlled diabetes mellitus, depression and other psychiatric conditions.

Endogenous CS is uncommon, hence widespread testing is not encouraged. Testing is recommended in patients with multiple and progressive features of CS, especially those with a high discriminatory value such as easy bruising, facial plethora, proximal muscle weakness and reddish-purple striae. Patients with unusual features for their age such as early onset osteoporosis or hypertension also warrant testing, as do children with increasing weight with concomitant low height for age percentile. Patients with an adrenal incidentaloma should also be tested.

Initial Testing

The use of random serum cortisol or plasma ACTH to screen for CS is not recommended. Tests directed towards identifying the aetiology of CS should also be avoided in the initial screening phase.

The guideline offers a choice of the following four tests for the initial investigation of CS: (1) 24 h UFC; (2) late-night salivary cortisol; (3) 1 mg overnight DST or (4) longer low dose DST (2 mg/day for 48 h). Individual patient characteristics may help to determine test choice. For example, DST is not recommended for patients on oestrogen therapy or the oral contraceptive pill as oestrogens increase cortisol binding globulin (CBG), often giving rise to a false positive result.

Due to the variability in cortisol secretion, at least two measurements are stipulated for UFC and late-night salivary cortisol. Although specific cut-off levels are proposed for each test to optimise diagnostic accuracy, these cut-offs vary with different assay methodologies and are highly dependent on the sensitivity of the assays at the limit of detection. The reference intervals are also method specific and the upper reference limit is higher in assays that are affected by cross-reactivity with other corticosteroid medications or metabolites. It is well established that UFC measured by antibody-based immunoassays grossly overestimates urine cortisol compared to structurally based reference methods such as HPLC and tandem mass spectrometry (LC-MS/MS).

The four initial tests recommended by the guideline review are summarised below. A normal result is likely to exclude CS (unless the pre-test probability is high or cyclical CS is suspected). Any abnormal result may require further testing and referring the patient for specialist endocrine assessment.

UFC

Measurement of 24 h UFC provides an integrated measure of free cortisol secretion. It relies on a complete 24 h urine collection which can be verified by analysis of urine creatinine. Although at least two urine collections are advocated, UFC may still be normal in cyclical and mild CS. Urine free cortisol is not affected by conditions and medications that affect CBG. However, falsely low results can be seen in renal impairment when creatinine clearance is below 60 mL/min. False positive results may occur with over-collection and excessive fluid intake of more than 5 L/day. It is important to appreciate the effect of cross-reactivity of interfering cortisol metabolites on different assay methodologies, and method specific reference intervals should be adopted.

Late-night Salivary Cortisol

The loss of circadian rhythm with absence of a late-night cortisol nadir is a feature in CS and this abnormality underlies the basis of late-night salivary or midnight serum cortisol testing. Salivary cortisol reflects the unbound biologically active form of serum cortisol. It is not affected by salivary flow rate, is stable at room temperature, non-invasive (e.g. ideal in children) and convenient to the patient (e.g. sample can be mailed in to the laboratory). However, it may not be appropriate for shift workers or patients with variable sleep patterns. False positive results may be seen in tobacco users (either smokers or those chewing tobacco) or patients using licorice. Both tobacco and liquorice contain 11β-hydroxysteroid dehydrogenase type 2 inhibitor glycyrrhizic acid which inhibits the conversion of cortisol to inactive cortisone. Further evaluation is required to determine whether gender, age, and co-existing medical conditions such as diabetes, hypertension and obesity affect late-night salivary cortisol measurement.

Although the diagnostic sensitivity and specificity is said to be excellent (92–100%), evidence suggests that salivary cortisol is no better than UFC in the diagnosis of CS in adults. Measurement of salivary cortisol by ELISA and LC-MS/MS has been well validated in the United States but is not routinely available in Australia. The guideline does not mention the performance of salivary cortisol on commonly available automated platforms such as the Roche Elecsys.

1 mg Overnight DST

This involves the outpatient administration of 1 mg dexamethasone between 2300 and 2400 h and blood sampling between 0800 and 0900 h the following morning for the measurement of serum cortisol. A normal response is adequate suppression of serum cortisol as determined by the cut-off. The guideline recommends a diagnostic cut-off of 50 nmol/L (1.8 μg/dL) rather than 140 nmol/L (5 μg/dL). The lower cut-off achieves a sensitivity of more than 95% with a compromise lower specificity of 80%. False positives are seen in women on oestrogen therapy, malabsorption, non-compliance, anticonvulsants (e.g. phenytoin, carbamazepine) or medications (e.g. rifampicin) which induce hepatic enzyme metabolism through cytochrome P450 3A4. False negative results may occur with reduced clearance of dexamethasone as seen in liver and renal failure. Caution should also be exercised in using this test in paediatric patients where it has not been fully validated.

48 hour, 2 mg/day DST

This test requires the patient to take 0.5 mg dexamethasone six hourly for 48 hours beginning at 0900 and serum cortisol taken at six hours after the last dose. CS is excluded by a normal response, which is considered to be an adequate suppression of serum cortisol of less than 50 nmol/L. A variation of this protocol is also discussed in the guideline.

This test is not widely used as an initial screening test in Australia and its use in outpatients is highly dependent on patient selection and ability to co-operate. The guideline acknowledges that this test has diagnostic accuracy similar to, if not lower than, the three initial tests mentioned above. However, it may still be a useful tool to differentiate true CS from other conditions associated with non-autonomous hypercortisolism (e.g. alcoholism, poorly controlled diabetes mellitus and depression).

Interpretation of results from any of the initial tests is influenced by clinical pre-test probability. A normal result in a patient with high pre-test probability for CS warrants further investigation by an Endocrinologist. Patients who test positive on the initial screening tests should also be referred to an Endocrinologist for further assessment.

Subsequent Evaluation

Physiological causes of hypercortisolism should be excluded. Another confirmatory test from one of the suggested initial tests may be undertaken or the guideline suggests performing either dexamethasone - corticotropin - releasing hormone (CRH) test or midnight serum cortisol as discussed below.

48 hour, 2 mg/day Low Dose DST with CRH

The guideline suggests this test may be useful in patients with equivocal results for UFC. Following 48 hours of 2 mg/day dexamethasone, CRH (1 μg/kg) is administered intravenously two hours after the last dose of dexamethasone. Serum cortisol is measured before and 15 minutes after CRH administration. The guideline suggests the measurement of serum dexamethasone at the time of CRH administration to exclude false positive results. However, neither CRH nor dexamethasone assays are readily available in Australia. The guideline acknowledges varying reports of diagnostic sensitivity and specificity.

In the opinion of the present authors, to consider this test suggests that the diagnosis of CS is in doubt and it may be worthwhile to repeat one of the initial investigations 3 to 6 months later.

Midnight Serum Cortisol

This investigation requires the patient to be admitted to hospital, and the test should be performed at least 48 hours after admission. The guideline discusses the sensitivity and specificity of various diagnostic cut-offs for both “sleeping” (sample taken within 5–10 minutes of waking, or undisturbed through an indwelling catheter) and “awake” serum cortisol levels. The authors acknowledge the inconsistent findings in the literature and, until better evidence is available, it appears that serum midnight cortisol of less than 207 nmol/L (7.5μg/dL), whether it be “sleeping” or “awake”, is a reasonable cut-off to exclude the diagnosis of CS.

Special Populations/Considerations

If CS is suspected in pregnancy, the guideline recommends screening with UFC. UFC more than three times the upper reference limit in second and third trimesters is considered abnormal, and greater than that attributable to the physiological increase in urinary cortisol excretion. DST is not recommended, as the increase in circulating CBG in pregnancy gives rise to false positives.

Midnight salivary or serum cortisol is recommended in patients on anticonvulsants (e.g. phenytoin, phenobarbitone and carbamazepine) that could induce the hepatic cytochrome catalysed clearance of dexamethasone.

In patients with chronic renal failure (creatinine clearance below 60 mL/min), the guideline suggests that a normal response to 1 mg DST makes CS unlikely. However, the converse does not hold true; an abnormal response is not necessarily diagnostic. Salivary or serum midnight cortisol may be useful but the diagnostic thresholds for these tests are not well validated.

Multiple measurements of UFC or salivary cortisol are often required to diagnose cyclic CS. The guideline does not recommend the use of DST.

In order to exclude subclinical or mild CS associated with adrenal incidentaloma, the guideline recommends 1 mg DST or late-night salivary cortisol as UFC is less sensitive.

Conclusion

This guideline offers a reasonable approach to screen for CS with a choice of predominantly three tests (UFC, late-night salivary cortisol and 1 mg DST). As the fourth test, 48 hour, 2 mg DST, should be reserved for consideration in special circumstances. The guideline acknowledges that the choice of tests is dependent on the patient’s individual characteristics and the experience of the clinicians and laboratories. It emphasises the importance of excluding exogenous corticosteroid use prior to testing and, in addition, certain psychiatric conditions (e.g. depression, anxiety), morbid obesity, alcoholism, and diabetes mellitus may give rise to false positive on the initial screening tests. A normal test result makes CS unlikely and any abnormal response should be referred to an Endocrinologist for further evaluation.

In Australia, salivary cortisol is not routinely available and our experience with this test is limited. If salivary cortisol is offered, the clinician should be aware of the methodology employed and the limitations of the assay. This also applies to UFC as different analytical methods, procedures in extraction versus non-extraction and cross-reactivity with other corticosteroid metabolites, will all affect the reference intervals.

This guideline highlights the difficulties in providing quality evidence for their recommendations, the main impediments being the rarity of CS, the lack of clinical outcome data and difficulties with cortisol assay standardisation. The need for appropriately powered randomised clinical trials to investigate outcomes in patients treated for CS is highlighted. It is particularly pertinent to validate these diagnostic tests in patients with mild or subclinical CS in order to optimise their management.