This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Newell-Price, J. Articles by Grossman, A. Search for Related Content PubMed PubMed Citation Articles by Newell-Price, J. Articles by Grossman, A.

The Diagnosis and Differential Diagnosis of Cushing’s Syndrome and Pseudo-Cushing’s States

Department of Endocrinology, St. Bartholomew’s Hospital, West Smithfield, London, EC1A 7BE, United Kingdom

	Abstract

Top Abstract I. Introduction II. Definitions and Etiology III. Diagnostic Overview IV. Clinical Features V. Biochemical Diagnosis of... VI. ACTH-Dependent vs. ACTH... VII. Differential Diagnosis of... VIII. Other Causes of... IX. Imaging X. Differentiation from Pseudo... XI. Conclusions References

 

I. Introduction

II. Definitions and Etiology

III. Diagnostic Overview

IV. Clinical Features

V. Biochemical Diagnosis of Cushing’s Syndrome

A. Cardinal features

B. Urinary free cortisol (UFC)

C. Low-dose dexamethasone testing

D. Circadian rhythm assessment

E. Cyclical Cushing’s syndrome

VI. ACTH-Dependent vs. ACTH-Independent Cushing’s Syndrome

VII. Differential Diagnosis of ACTH-Dependent Cushing’s Syndrome

A. Overview

B. Basal testing

C. Dynamic noninvasive testing

D. Invasive testing

VIII. Other Causes of Cushing’s Syndrome

IX. Imaging

A. Pituitary

B. Adrenal

C. Ectopic secretion

X. Differentiation from Pseudo-Cushing’s States

XI. Conclusions

	I. Introduction

Top Abstract I. Introduction II. Definitions and Etiology III. Diagnostic Overview IV. Clinical Features V. Biochemical Diagnosis of... VI. ACTH-Dependent vs. ACTH... VII. Differential Diagnosis of... VIII. Other Causes of... IX. Imaging X. Differentiation from Pseudo... XI. Conclusions References

 
THREE quarters of a century have passed since Harvey Cushing’s original descriptions of the clinical syndrome characteristic of glucocorticoid excess (1, 2). Interest in Cushing’s syndrome, especially in recent years, continues to gain momentum from both a clinical and basic science viewpoint, as witnessed by the ever increasing literature on the subject. More than in any other area of clinical endocrinology, diagnosis, differential diagnosis, and management continue to challenge the physician and occasionally cause considerable controversy. This is reflected by our limited understanding, despite much study, of the biology of the possible causes. Recent advances, especially in molecular biology, have given some insight as to the basis of the biochemical tests that are in common clinical practice; yet in the majority of cases, diagnosis is an extremely pragmatic affair involving the utilization of diagnostic strategies that have been developed and validated over the last 30 yr or so. Cushing’s syndrome has been the subject of numerous original papers and reviews. Due to the relative rarity of the condition, many years of experience are required before a given diagnostic approach can be formally validated, this often being in the form of retrospective analyses. The last few years have seen the introduction of several new approaches and the validation of existing ones in larger series. It thus seems a pertinent time to review our current understanding, address areas of debate and, at the risk of further controversy, to suggest some diagnostic approaches.

It is our belief that the clinical spectrum of Cushing’s syndrome is shifting as this rare diagnosis is increasingly being considered by astute physicians in more common settings, such as the diabetic clinic. Indeed, recent work has suggested that up to 3–4% of individuals with poorly controlled diabetes mellitus with an obese phenotype may have Cushing’s syndrome (3). Therefore, our ability to make the diagnosis of Cushing’s syndrome is becoming an increasing challenge, since the pathological process of glucocorticoid excess is being considered at an earlier stage in its natural history. In this paper, we critically review the clinical features of the syndrome and the biochemical tests that confirm or refute clinical suspicion. There then follows a critique of the biochemical tests and imaging investigations employed in the differential diagnosis of Cushing’s syndrome, with particular emphasis on more recent approaches, large-series validations and modifications of existing protocols. Finally, consideration is given to the vexed issue of the differentiation of Cushing’s syndrome from pseudo-Cushing’s states.

	II. Definitions and Etiology

Top Abstract I. Introduction II. Definitions and Etiology III. Diagnostic Overview IV. Clinical Features V. Biochemical Diagnosis of... VI. ACTH-Dependent vs. ACTH... VII. Differential Diagnosis of... VIII. Other Causes of... IX. Imaging X. Differentiation from Pseudo... XI. Conclusions References

 
Endogenous Cushing’s syndrome is a clinical state resulting from prolonged, inappropriate exposure to excessive endogenous secretion of cortisol and hence excess circulating free cortisol, characterized by loss of the normal feedback mechanism of the hypothalamo-pituitary-adrenal axis and the normal circadian rhythm of cortisol secretion (4). Other situations in which there is biochemical evidence of excess cortisol secretion, without the development of a ’Cushingoid state,’ however subtle, such as in the setting of a long period in the intensive care unit, will not be considered here. The etiology of Cushing’s syndrome may be excessive ACTH production from the pituitary gland, ectopic ACTH secretion by a nonpituitary tumor, or excessive autonomous secretion of cortisol from a hyperfunctioning adrenocortical tumor (5, 6, 7, 8, 9). Other than these broad ‘ACTH-dependent’ and ‘ACTH-independent’ categories, the syndrome may, in addition, be caused by ectopic CRH secretion (10, 11, 12), bilateral primary pigmented nodular adrenal hyperplasia and macronodular adrenal hyperplasia (13), the ectopic actions of gastric-inhibitory peptide or catecholamines (14, 15, 16), and other adrenal-dependent processes associated with adrenocortical hyperfunction such as McCune-Albright syndrome and Carney’s complex (13, 17) (Tables 1 and 2). Pseudo-Cushing’s states, which may have similar clinical presentations together with evidence of hypercortisolemia, may be caused by alcohol dependence (18, 19, 20, 21, 22) and depression (23, 24, 25, 26). It may also be necessary to differentiate Cushing’s syndrome from other clinical presentations with ‘Cushingoid’ features, such as cases of simple obesity in which some Cushingoid clinical features may be present (27, 28, 29, 30, 31, 32). The incidence of pituitary-dependent Cushing’s disease and adrenal adenomas in women is 3 to 4 times that of men: since this is the most common form of Cushing’s syndrome, as a whole, therefore, women easily outnumber men (Table 1).

	III. Diagnostic Overview

Top Abstract I. Introduction II. Definitions and Etiology III. Diagnostic Overview IV. Clinical Features V. Biochemical Diagnosis of... VI. ACTH-Dependent vs. ACTH... VII. Differential Diagnosis of... VIII. Other Causes of... IX. Imaging X. Differentiation from Pseudo... XI. Conclusions References

 
The clinical signs of Cushing’s syndrome are protean, providing the stimulus for further biochemical evaluation and imaging. The diagnosis of Cushing’s syndrome must be established before any attempt at differential diagnosis, since the tests employed in the differential diagnosis may be misleading, or uninterpretable, unless there is biochemical confirmation of the hypercortisolemic state. This latter criterion is vital and applicable to all of the tests detailed below, since the reported sensitivity, specificity, and diagnostic accuracy are only valid during periods of active and sustained hypercortisolism. In some circumstances diagnosis may be relatively straightforward, while in others, and especially at times when the degree of hypercortisolism is only mild and variable, diagnosis and differential diagnosis may remain elusive.

	IV. Clinical Features

Top Abstract I. Introduction II. Definitions and Etiology III. Diagnostic Overview IV. Clinical Features V. Biochemical Diagnosis of... VI. ACTH-Dependent vs. ACTH... VII. Differential Diagnosis of... VIII. Other Causes of... IX. Imaging X. Differentiation from Pseudo... XI. Conclusions References

 
Symptoms associated with hypercortisolemia include weight gain, lethargy, weakness, menstrual irregularities, loss of libido, depression, hirsutism, acne, purplish skin striae, and hyperpigmentation (6, 36, 37, 38). Associated problems such as diabetes mellitus or hypertension may also bring the patient to medical attention. The signs associated with Cushing’s syndrome are extremely varied and differ in severity (Table 3). Signs that differentiate Cushing’s syndrome from pseudo-Cushingoid states most reliably include the presence of proximal myopathy, easy bruising, and thinness and fragility of the skin (36, 37). In our experience signs such as buffalo hump, obesity, and hirsutism are poor discriminators. In the case of children, gain in weight associated with growth retardation are particularly prominent features that should alert clinical suspicion to the diagnosis (39, 40, 41, 42).

The original descriptions of the ‘overt’ ectopic ACTH syndrome reported the effects of high levels of ACTH and cortisol, usually of rapid onset and most often due to ACTH secretion from small-cell lung cancers (50). Symptoms include profound weakness as a direct result of high circulating levels of cortisol and associated hypokalemia, while there is often little weight gain and absence of classically Cushingoid appearance. Pigmentation frequently appears as a result of the high circulating levels of ACTH. In contrast, ACTH-secreting carcinoid tumors, most frequently bronchial in origin, may present with clinical features indistinguishable from pituitary-dependent or primary adrenal disease; this clinical situation is now referred to as the ‘occult’ ectopic ACTH syndrome (9). Thus, the clinical history and examination may be extremely helpful in differentiating between pituitary and ectopic causes in cases of ‘overt’ ectopic ACTH syndrome, whereas they have poor discriminating power in the ‘occult’ ectopic ACTH syndrome.

	V. Biochemical Diagnosis of Cushing’s Syndrome

Top Abstract I. Introduction II. Definitions and Etiology III. Diagnostic Overview IV. Clinical Features V. Biochemical Diagnosis of... VI. ACTH-Dependent vs. ACTH... VII. Differential Diagnosis of... VIII. Other Causes of... IX. Imaging X. Differentiation from Pseudo... XI. Conclusions References

 
A. Cardinal features
The cardinal biochemical features of Cushing’s syndrome are excess endogenous integrated secretion of cortisol, loss of the normal feedback of the hypothalamo-pituitary-adrenal axis (HPA), and disturbance of the normal circadian rhythm of cortisol secretion. The biochemical tests that are used in the diagnosis of Cushing’s syndrome rely upon these parameters. In the investigation of Cushing’s syndrome, initial biochemical tests of high sensitivity should be employed, although the diagnosis may later be refuted by tests of higher specificity. This is an acceptable trade off, since tests that are highly specific but relatively insensitive will inevitably miss individuals with mild disease. In the interpretation of published series, caution is required, since diagnostic criteria that provide discrimination between groups under study are inherently reliant on the assays on which they are based. Therefore, the responses for a given test require validation in the locally used assay before they may be reliably interpreted in a given patient. Supraregional and nationwide interassay quality control assurance provides a means of achieving this and is widely practiced.

B. Urinary free cortisol (UFC)
Collection of urine for estimation of cortisol and cortisol metabolites is a noninvasive procedure and is widely used as a screening test for the diagnosis of Cushing’s syndrome. Under normal conditions, approximately 10% of serum cortisol is unbound and physiologically active. Free unbound cortisol passes through the kidneys, and although the majority is reabsorbed in the tubules, the remainder is excreted unaltered (4). Excess cortisol saturates circulating cortisol-binding globulin, resulting in an increase in the urine cortisol-UFC. UFC measurements have superseded the measurement of urinary 17-hydroxycorticosteroids (17-OHCS) or 17-oxogenic steroids, which are metabolites of cortisol and cortisone. In his review of 14 separate studies assessing the utility of 17-OHCS measurement for the diagnosis of Cushing’s syndrome, Crapo (51) reported that the false negative rate was 11% of 315 individuals with Cushing’s syndrome, while in 173 obese controls the false positive rate was 27%. Similarly poor results were obtained utilizing 17-ketogenic steroids (KGS) with a false negative rate in 24% of 235 patients with Cushing’s syndrome (51). In contrast, 24-h UFC measurements by RIA should reflect the integrated cortisol secretion, with a raised level being consistent with Cushing’s syndrome. The upper normal range in most assays is between 220–330 nmol/24 h (80–120 µg/24 h). The majority of problems associated with this test relate to the adequacy of collection, although in some assays there may be cross-reactivity with exogenous glucocorticoids. Adequate written instructions will improve collections (52), but there was a false negative rate of 5.6% and false positive rate of 3.3% in the combined data on 479 obese, lean, and chronically ill individuals (51). Expressing the UFC over creatinine allows the adequacy of collection to be established and improves the specificity (53), although it should be noted that creatinine may vary with changes in lean body mass. More recently, UFC measurement was shown to have a diagnostic sensitivity and specificity of 100% and 98%, respectively, in the differentiation of 48 patients with Cushing’s syndrome from 98 normal subjects and 95 obese individuals (54). However, although 24-h UFC measurement in 146 patients with Cushing’s syndrome was shown to have a sensitivity of 95%, it was noted that 11% had at least one of four 24-h collections with values within the normal range (55). Furthermore, ‘raised’ 24-h UFC levels have been documented in 40% of 60 depressed inpatients (56) and in 50% of 45 women with the polycystic ovarian syndrome (57); by definition, almost complete overlap in levels is seen in the differentiation from various causes of pseudo-Cushing’s states (58), emphasizing the potential for diagnostic confusion. The problem of cross-reactivity becomes a particular issue if the possibility of exogenous glucocorticoid administration exists (57). HPLC has recently been compared with RIA for the measurement of cortisol and cortisone in the assessment of endogenous Cushing’s syndrome and Cushing’s syndrome due to exogenous glucocorticoid ingestion (59). Using this method on single urine samples, 19 of 29 patients with histologically proven ACTH-dependent or ACTH-independent Cushing’s syndrome had an increase in both cortisone and cortisol, while a further 8 of 29 had an increase in one or the other. Overall, 27 of 29 (93%) had HPLC measurements comparable to a competitive binding assay, with levels of cortisol, cortisone, or both that were higher than the normal range; in 6 patients with Cushing’s syndrome due to exogenous glucocorticoids, the UFC and cortisone were suppressed, and prednisolone and prednisone were detected. Utilization of the competitive binding assay in this latter group resulted in every individual having a falsely elevated ‘UFC’ measurement. Since RIAs also suffer from problems of cross-reactivity, this approach may occasionally be useful in difficult cases where doubt exists as to the origin of glucocorticoid. Thus, if replicated in further studies, this approach may prove to be of value, as in this small number the sensitivity approaches that for four 24-h collections measured by RIA, but data on the values seen in pseudo-Cushing’s states are needed for full evaluation. Nevertheless, similar discrimination might be made by a single 0900 h plasma cortisol, since this should be suppressed in conditions in which exogenous glucocorticoids have resulted in Cushing’s syndrome, as long as the plasma cortisol RIA has little cross-reactivity for synthetic glucocorticoids and hydrocortisone is not being administered. Finally, a low dihydroepiandrosterone sulfate, because of suppressed plasma ACTH, may commonly be found in, and is a useful additional indicator of, exogenous glucocorticoid administration (57).

Overall, UFC estimations have a high sensitivity, but relatively low specificity; therefore, if several UFC collections are normal, Cushing’s syndrome is highly unlikely.

C. Low-dose dexamethasone testing
Since the original description by Liddle in 1960 (60) of the 48-h 2 mg/day low-dose dexamethasone suppression test (LDDST), this diagnostic tool has remained an important part of the evaluation of suspected Cushing’s syndrome. Administration of dexamethasone, which is not measured in most cortisol RIAs, results in suppression of the HPA axis in normal individuals and a fall in plasma and urinary cortisol. A variety of regimens exist for the dexamethasone administration, and a range of diagnostic ‘cut-offs’ that classify responses have been reported.

The overnight test involves the oral administration of 0.5–2.0 mg dexamethasone (most commonly 1 mg) at 2300 or 2400 h, after which a plasma cortisol sample is obtained at 0800 h or 0900 h the next morning (61). There appears to be no better discrimination with 1.5 mg or 2 mg of dexamethasone than with 1 mg administration (51). Because of its ease of administration as an outpatient, it has been widely advocated as a screening test. The reported cut-off values for suppression of serum cortisol in studies utilizing modern RIAs, but with relatively small numbers of individuals with Cushing’s syndrome, range from 100–200 nmol/liter (3.6–7.2 µg/dl) (62, 63). Some patients with Cushing’s syndrome, however, demonstrate unusual suppressibility to dexamethasone (64), and thus cut-offs at this level are likely to result in a significant number of false negative responses. Therefore, to enhance sensitivity, a recent extensive review assessing the utility of low-dose dexamethasone testing in the diagnosis of Cushing’s syndrome suggested that suppression of the postdexamethasone plasma cortisol to 50 nmol/liter (1.8 µg/dl) or less effectively excludes Cushing’s syndrome (65). At this level false positive rates will be higher, but the main value of the overnight test is that of ease of outpatient screening to exclude Cushing’s syndrome, which may be an acceptable price to pay for enhanced sensitivity.

The original description of the 48-h 2 mg/day dexamethasone suppression test (60) reported the suppression of urinary 17-OHCS as an indicator of cortisol suppression. Serum cortisol RIAs provide a more simple measurement, with test sensitivities of 97–100% (58, 66, 67), comparable to the overnight test (51). In our own experience, testing in 150 individuals with proven Cushing’s syndrome and measuring the 0900 h serum cortisol before and after the administration of 0.5 mg dexamethasone strictly every 6 h for 48 h and a cut-off value for suppression of 50 nmol/liter (1.8 µg/dl), resulted in a sensitivity of 98% (three patients with histologically proven Cushing’s disease suppressed to less than 50 nmol/liter) (68). In a direct comparison of the responses of 39 patients with Cushing’s syndrome compared with 19 with pseudo-Cushing’s syndrome, a plasma cortisol concentration at 0800 of 38 nmol/liter (1.4 µg/dl), exactly 2 h after the last dose of dexamethasone that had been administered for 48 h as above, gave a specificity of 100% and a sensitivity of 90% for the diagnosis of Cushing’s syndrome, while measurement of urinary steroids provided a sensitivity of only 50–60% (58). Increased sensitivity was achieved by analyzing the plasma cortisol after the administration of CRH (see Section X). More recently, and using the same criterion, these same authors have demonstrated the utility of this combined test in differentiating mild Cushing’s disease (three demonstrated suppression of plasma cortisol on the LDDST limb of the test) from normal individuals (69). The 1-mg overnight test has a specificity of 87.5% (63), while the reported specificity for the 2 mg/day 48-h test is 97–100% (67). Thus, the standard Liddle 2 mg/day 48-h LDDST has the same sensitivity and higher specificity than overnight dexamethasone testing, but should be performed by measuring plasma cortisol rather than urinary steroids. It is our routine practice to use the Liddle 48-h, 2 mg/day dexamethasone suppression test in both inpatient and outpatient settings, since, with adequate written instructions, compliance is extremely high and the results are reproducible.

In all the variations of the oral dexamethasone tests, variable absorption and metabolism of dexamethasone will influence the result of the test (70, 71). Thus, in an effort to reduce false positive responses, simultaneous measurement of plasma cortisol and dexamethasone has been advocated for the overnight test to ensure adequate plasma dexamethasone concentrations of 5.6 nmol/liter (0.22 µg/dl) or greater and to confirm compliance (70). This does, however, require access to a costly dexamethasone assay, which is often unnecessary, although it may be particularly useful in cases of suspected malabsorption. To overcome this type of problem, the intravenous dexamethasone suppression test has been proposed (29). In this study, an infusion of dexamethasone at 1 mg/h between 1100 h and 1500 h caused a sustained suppression day of plasma cortisol to less than 83 nmol/liter (3 µg/dl) until at least 0900 h the next day in normal and obese subjects, while in patients with Cushing’s syndrome the plasma cortisol was greater than 276 nmol/liter (10 µg/dl) at this 0900 h time point. Interestingly, distinction between Cushing’s disease and either the ectopic ACTH syndrome or ACTH-independent Cushing’s syndrome could be made on the basis of a 50% fall in plasma cortisol during the dexamethasone infusion. An alternative regimen involved intravenous administration of dexamethasone at a dose of 5 µg/kg/h for 5 h between 1000 h and 1500 h, which resulted in suppression of plasma cortisol to less than 38 nmol/liter (1.4 µg/liter) at 1900 h in 19 patients with simple obesity, while in 12 patients with Cushing’s syndrome the plasma cortisol was 68 nmol/liter (2.5 µg/dl) or greater at this time point; at 0800 h the next day, the obese group had a sustained suppression of plasma cortisol, while those with Cushing’s syndrome had values of 136 nmol/liter (5 µg/dl) or more (72). Such tests are clearly more complex to perform, but in certain circumstances their application may be useful.

Drugs such as phenytoin, phenobarbitone, carbamezepine, and rifampicin will induce hepatic enzymatic clearance of dexamethasone, thereby reducing the plasma dexamethasone concentration (73, 74) and resulting in false positive responses to dexamethasone testing. Estrogens increase the cortisol-binding globulin concentration in the circulation; since RIAs measure total cortisol, false positive rates are seen in 50% of women on the oral contraceptive pill (75). It is our routine practice, where possible, and particularly in mild cases, to stop such estrogen-containing drugs and delay investigation for 6 weeks to allow the cortisol-binding globulin to return to baseline. However, this may not be necessary in the case of transdermal estrogens.

D. Circadian rhythm assessment
In normal circumstances the level of serum cortisol begins to rise at 0300–0400 h and reaches a peak at 0700–0900 h, with levels then falling for the remainder of the day. Loss of the normal circadian rhythm in patients with Cushing’s syndrome was first reported by Doe et al. in 1960 (76) and has been confirmed in several studies (77, 78, 79, 80). In contrast, other reports have suggested that the rhythm may persist in certain patients but with levels that are set abnormally high (81, 82, 83). There is a large overlap in 0900 h serum cortisol values between patients with Cushing’s syndrome and normal subjects (36, 84, 85); therefore, this sampling time affords poor discrimination. The overlap between patients with Cushing’s syndrome and the normal range diminishes with clock time such that at time points in the range 1600–2100 h, 17% of patients with Cushing’s syndrome have values within the normal range, falling to 3.4% at 2300 h (51). Urine cortisol samples have also been used for this purpose. Clearance of urine cortisol collected between 2200 h and 2300 h, and expressed as a ratio of nanograms/mg creatinine, ranged from 76 to 905 in 14 patients with Cushing’s syndrome but only from 6 to 43 in 20 normal subjects, affording discrimination between groups (53). Similarly, a timed urine collection between 2000 h and 2400 h revealed a ratio (expressed as micromoles/mol creatinine) ranging from 27.5 to 855 in 20 patients with Cushing’s syndrome, 1.1 to 9.4 in nonobese control subjects, and 9.4 to 27.8 in 34 obese subjects, thus giving an overlap with the Cushing’s syndrome group in one individual with extreme obesity (86). In our own series a single sleeping plasma cortisol was greater than 50 nmol/liter (1.8 µg/dl) in 150 individuals with Cushing’s syndrome (three of which had suppressed on a 2 mg/day 48-h LDDST), while in control subjects the sleeping midnight serum cortisol was less than 50 nmol/liter (1.8 µg/dl) in all (68) (Fig. 1). The test does, however, require inpatient admission for a period of 48 h or more to avoid false positive responses due to the stress of hospitalization, and the blood sample needs to be drawn within 5–10 min of waking the patient. To avoid false positive results due to anticipation, the patients should not be warned that the test is to be performed; if the patient is awake, the test is not readily interpretable. It is apparent from Fig. 1 that the test affords no discriminatory power between any of the causes of Cushing’s syndrome. Moreover, this reported sensitivity is reliant, as stated above, on an active state of cortisol hypersecretion.

View larger version (19K): [in this window] [in a new window]   Figure 1. Sleeping plasma midnight cortisol values in 150 patients with Cushing’s syndrome and 20 normal volunteers. [Reproduced with permission from J. Newell-Price et al.: Clin Endocrinol (Oxf) 43:545–550, 1995 (68 ).]

E. Cyclical Cushing’s syndrome
As stated above, for unknown reasons some ACTH-secreting tumors of any type causing Cushing’s syndrome exhibit cyclical and intermittent secretion (47, 48), and this may be reflected by a history of variable and intermittent depression with anxiety (92), an alteration in the level of prevailing glycemia, or indeed any of the plethora of symptomatology and signs outlined in Table 3. This may cause considerable diagnostic confusion, but careful documentation of the history is of paramount importance. Inpatient admission, sometimes on repeated occasions, with sampling for sleeping midnight plasma cortisol is one means of getting around this diagnostic conundrum, and proceeding to further investigation if documented hypercortisolemia is present. If inpatient admission cannot be justified, salivary cortisol estimations may be used to establish the diagnosis as an outpatient (93). Multiple and repeated 24-h UFC collections may also prove useful. Clearly, the diagnostic dimension of time is often needed to establish the diagnosis.

	VI. ACTH-Dependent vs. ACTH-Independent Cushing’s Syndrome

Top Abstract I. Introduction II. Definitions and Etiology III. Diagnostic Overview IV. Clinical Features V. Biochemical Diagnosis of... VI. ACTH-Dependent vs. ACTH... VII. Differential Diagnosis of... VIII. Other Causes of... IX. Imaging X. Differentiation from Pseudo... XI. Conclusions References

 
Only once the diagnosis of Cushing’s syndrome is established may the differential diagnosis be entertained. The initial step is to establish whether ACTH is detectable in the plasma; if it is consistently undetectable, the diagnosis of ACTH-independent Cushing’s syndrome may be made, and focus should turn to adrenal imaging. Usually this is a fairly easy discrimination to make: at our institution a plasma ACTH level of less than 10 pg/ml was seen in all patients with Cushing’s syndrome due to a cortisol-secreting adrenal adenoma. It is, however, our experience that a few patients with pituitary-dependent Cushing’s disease have the occasional undetectable plasma ACTH, as assessed by a conventional RIA with a lower limit of detection of 10 pg/ml (94). Thus, if measured by this means, we would recommend the collection of several plasma ACTH collections to avoid this potential problem, or the measurement of ACTH after CRH stimulation. Nevertheless, we have not seen any patient with ACTH-dependent Cushing’s syndrome that has had more than one or two ACTH levels of less than 10 pg/ml. Immunoradiometric assays (IRMA) provide increased speed of assay, high reproducibility, and sensitivity for ACTH measurement. Using IRMA, ACTH levels consistently below 5 pg/ml are found in patients with cortisol-producing adrenal adenomas, autonomous bilateral adrenal hyperplasia, and Cushing’s syndrome due to the administration of exogenous glucocorticoids (95). The use of such sensitive assays will provide very good discrimination, and when the levels are this low the diagnosis is clearly ACTH-independent; in contrast, when plasma ACTH is unequivocally measurable, ACTH dependence exists. As such, they provide optimal screening at the critical decision limb as to whether the clinician is dealing with ACTH-dependent or ACTH-independent disease. A gray area does, however, exist and low-detectable ACTH levels need cautious interpretation and repeated measurement. IRMA measurements are more specific than RIA measurements, but it is this specificity that may, in theory, result in ‘normal levels’ being recorded in patients with the ectopic ACTH syndrome, since in this situation ACTH precursors may be present in large quantities, and these are not detected by ACTH IRMA (4, 96).

The high-dose dexamethasone suppression test (HDDST) was originally introduced to distinguish adrenal causes of Cushing’s syndrome from Cushing’s disease, and in the original report allowed an accurate discrimination in all those tested (60). The advent of reliable ACTH assays has facilitated the discrimination between adrenal causes and pituitary or ectopic ACTH secretion, although the HDDST remains useful in demonstrating functional autonomy (independent secretion of cortisol) of an adrenal adenoma or carcinoma disclosed on abdominal scanning. Some centers advocate the use of CRH testing (see below) to confirm a lack of ACTH response in this situation.

	VII. Differential Diagnosis of ACTH-Dependent Cushing’s Syndrome

Top Abstract I. Introduction II. Definitions and Etiology III. Diagnostic Overview IV. Clinical Features V. Biochemical Diagnosis of... VI. ACTH-Dependent vs. ACTH... VII. Differential Diagnosis of... VIII. Other Causes of... IX. Imaging X. Differentiation from Pseudo... XI. Conclusions References

 
A. Overview
In contrast to differentiating between ACTH-dependent and ACTH-independent etiologies, the differential diagnosis of ACTH-dependent Cushing’s syndrome is far more taxing. Both corticotroph adenomas and tumors causing the ectopic ACTH syndrome are frequently small and thus difficult to visualize, and strenuous efforts at localization are required to allow correct management: these, in turn, rely heavily on biochemical testing to direct the imaging to the appropriate site. There is, however, no such thing as the single, simple, perfect, noninvasive test that will, in every case, allow the differentiation between Cushing’s disease and the ectopic ACTH syndrome. Indeed, the existence of such a test would presuppose an invariant difference in the biology of these etiological entities that would allow a completely reliable classification based on the responses seen during biochemical testing, and this does not seem to be the case (97). The highest degree of accuracy is most likely to be obtained, therefore, by using a variety of tests that assess the spectrum of different physiological responses to a variety of agents.

In most centers, during the investigation of ACTH-dependent Cushing’s syndrome the a priori probability that a patient has pituitary disease is usually between 85% and 90%. Therefore, statistically, the endocrinologist has a far better than even chance of getting the correct diagnosis with almost no investigation whatsoever, once the presence of detectable plasma ACTH has been established. It is widely held that pituitary surgery is the optimal management of Cushing’s disease (98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110). One extreme approach might consist of proceeding directly to pituitary surgery after the sustained detection of plasma ACTH. Because of the inherent risks of the operation and the potential for hypopituitarism, especially in individuals of child-bearing age, this ultimate reductionist approach is unacceptable. Furthermore, such an approach would not improve the condition of a patient with ectopic ACTH secretion; indeed, delay in the correct localization and appropriate management of these tumors may result in metastatic disease (111). Biochemical testing is used in an attempt to improve upon the pretest likelihood and to direct the physician to the appropriate imaging and sampling modalities before formal management. Assuming that the default mode of treatment is pituitary surgery, the peripheral noninvasive tests in the literature are reported such that specificity is optimized (this inevitably being at the cost of reduced sensitivity) so that patients do not undergo inappropriate pituitary surgery. A major problem with all analyses, however, is the ascertainment of diagnosis. Is the ‘gold standard’ for pituitary disease a positive ‘central’ ACTH gradient on bilateral inferior petrosal sinus sampling (BIPSS), the neurosurgeon who ‘sees’ the tumor, the patient cured after a microadenomectomy but with negative histology, or the tumor immunostaining with anti-ACTH antibody? Depending on classification, the sensitivity and specificities for any test can be radically altered: no consensus currently exists, and it appears unlikely that one will emerge. This is further compounded by selection bias, intention to test/treat variables, and ‘excluded cases.’ We are left with recommendations ranging from ‘do all the tests in all the patients’ to suggestions not far from the one outlined above. Bearing these caveats in mind, we will review the tests currently employed in the differentiation of the causes of ACTH-dependent Cushing’s syndrome. We must again emphasize that these tests are only interpretable in the context of sustained and current hypercortisolemia.

B. Basal testing
1. Plasma ACTH. Although the levels of plasma ACTH tend to be higher in the ectopic ACTH syndrome than in Cushing’s disease, there is a large overlap between values, as assessed by RIA and IRMA (4, 85, 96), and therefore this affords poor discrimination between groups. The presence of POMC precursors, due to partial processing of this peptide and incomplete cleavage to ACTH, ‘big ACTH’ (112), is documented by a specific two-site IRMA in the ectopic ACTH syndrome, particularly when caused by small-cell lung carcinoma (113). Such ‘overt’ ectopic ACTH secretion is usually clinically obvious, unlike the ‘occult’ secretion due to carcinoid tumors, most often bronchial in origin. Recently, POMC precusors have also been documented in 12 patients with histologically proven ACTH-secreting carcinoid tumors, albeit at lower levels than seen in small-cell lung cancer, but higher than in any of the 27 patients with Cushing’s disease caused by a pituitary microadenoma (114). In contrast, pituitary corticotroph macroadenomas may also exhibit poor processing of POMC (115, 116) causing diagnostic confusion. Furthermore, overlap has been documented in the levels of POMC present in 20 patients with the ectopic ACTH syndrome and 42 patients with pituitary-dependent Cushing’s disease, with the values reflecting the aggressive nature of the tumors regardless of origin (117). Therefore, such analysis may prove to be helpful in discriminating Cushing’s disease from causes of ‘occult’ ectopic ACTH secretion, but the assays are not widely available and the data are currently conflicting.

2. Serum potassium. Serum potassium is usually low in the ectopic ACTH syndrome; therefore, this may be an extremely helpful discriminator, although up to 10% of patients with Cushing’s disease exhibit hypokalemia (4, 85). The apparent reason for the hypokalemia is the saturation of 11ß-hydroxysteroid dehydrogenase by excessive cortisol, which under normal physiological circumstances protects the mineralocorticoid receptor from the effects of cortisol (118). However, this generally reflects the prevailing levels of cortisol rather than the specific etiology. Thus, hypokalemia has high sensitivity for the ectopic ACTH syndrome, but a specificity that only approaches the pretest likelihood. In our experience we have seen only one patient with the ectopic ACTH syndrome (with a bronchial carcinoid) who did not have hypokalemia and associated alkalosis.

3. Ectopic cosecretion. In up to 70% of cases, occult ectopic tumors may express and cosecrete one or more additional peptides such as calcitonin, somatostatin, gastrin, pancreatic polypeptide, vasoactive intestinal peptide, glucagon, hCG-ß, -fetoprotein, -subunit, neuron-specific enolase, GHRH, CRH, and carcinoembryonic antigen (119, 120). Thus, measurement of these specific peptides may sometimes be useful. The presence of an additional peptide provides stronger evidence for the ectopic ACTH syndrome, may also serve as a tumor marker during follow-up, and may occasionally be useful during venous sampling for localization (120).

C. Dynamic noninvasive testing
1. High dose dexamethasone testing. For more than 30 yr HDDST has remained one of the main biochemical tools used in the differential diagnosis of ACTH-dependent Cushing’s syndrome. The basis of the test relies on the fact that, in most situations, the corticotroph tumor cells in Cushing’s disease retain some responsiveness to the negative feedback effects of glucocorticoids while tumors ectopically secreting ACTH do not. The standard test was performed on 24-h collections of urine for the measurement of 17-OHCS or UFC, calculating the degree of suppression from day 1 to day 3 after the administration of oral dexamethasone at a dose of 2 mg every 6 h for 48 h. As with the LDDST, several suppression cut-offs that optimally define pituitary disease, in addition to combinations of measurements of different steroids, have been reported. Originally, suppression of urinary 17-OHCS by 50% or greater was reported as being consistent with Cushing’s disease (60). Although this criterion has no intrinsic legitimacy, subsequent studies have confirmed the utility of this cut-off using more easily obtained plasma cortisol estimations, and calculating the suppression of values at 0800 h or 0900 h before and after 48 h of dexamethasone administration (85, 121, 122, 123). Recently, data from the National Institutes of Health indicate that measurement of UFC as an end point is as accurate a test as when 17-OHCS measurement is used, and that suppression of UFC by 90% or 17-OHCS by 64% in a given patient results in 100% specificity and 83% sensitivity for the diagnosis of pituitary disease (124). This series included 118 patients with surgically proven Cushing’s syndrome: 94 patients with Cushing’s disease, 14 with primary adrenal disease, and 10 with ectopic ACTH secretion. However, 2 yr later a further report from this center indicated that an increase in the level of suppression of 17-OHCS to 69% was required to maintain a specificity of 100%, albeit with a reduced sensitivity of 79%, and highlighted the utility of plasma cortisol measurement (123). Such results reflect the problems encountered in attempting to develop increasingly sophisticated cut-off criteria that maximize specificity to define disease etiology, since only one outlying responder is required to drastically alter the specificity of a given test; this inevitably results in a fall in sensitivity.

The 48-h HDDST is somewhat cumbersome; as an alternative, the 8-mg overnight dexamethasone suppression test has been developed, which involves the administration of a single 8-mg dose of dexamethasone orally at 2300 h with measurement of plasma cortisol at 0800 h before and after administration (125, 126). This test has a sensitivity ranging from 57% to 92% and a specificity ranging from 57% to 100% (123, 125, 126, 127). The time points and cut-offs that result in these reported figures vary. After the original description of the 8-mg high-dose overnight test, a 50% suppression of plasma cortisol at the second 0800 h sampling point resulted in 100% specificity and 92% sensitivity (125). More recently, a report from the NIH, including seven individuals with bronchial carcinoid tumors causing the ectopic ACTH syndrome, demonstrated 88% sensitivity and 57% specificity using these points and cut-off criteria (123). In the same study, 100% specificity and 71% sensitivity was achieved by a pre-dexamethasone sampling time of 0830 h and post-dexamethasone sampling time of 0900 h and increasing the suppression criterion to more than 68%. These minor time changes appear to have significant effects on the reported results, and further larger scale data are needed for confirmation of these revised criteria. Even so, it is apparent that they still do not appear to be as discriminatory as standard 48-h high-dose dexamethasone testing. This lowering of test sensitivity is an inevitable result of increasing experience and numbers, in addition to the increasing identification of small carcinoid tumors that may have previously gone undiagnosed.

Part of the failure of suppression during high-dose dexamethasone testing in patients with Cushing’s disease may relate to inadequate levels of plasma dexamethasone due to inadequate absorption, increased clearance, or poor compliance (128). To circumvent these problems, the use of a 5-h intravenous infusion of dexamethasone at a rate of 1 mg/h has been advocated, with suppression of plasma cortisol by 50% or more being consistent with Cushing’s disease (129). A slight modification of this test, with an infusion of dexamethasone for 7 h, demonstrated a fall in plasma cortisol of 190 nmol/liter or more in all of 90 patients with Cushing’s disease, but only in 2 of 7 with the ectopic ACTH syndrome, giving a sensitivity of 100% and a specificity of 90% (130). Both false positive responders were ectopic secretors of CRH, and since this is a rare cause of Cushing’s syndrome, more data on the responses seen in the ectopic ACTH syndrome are needed. Overall, the sensitivity of the HDDST ranges from 65% to 100%, and the specificity ranges from 60% to 100% (51, 85, 122, 123, 124, 131). Combining the results of the 48-h and overnight HDDST tests performed in every individual, and using the revised criteria for the 8-mg overnight test and conventional HDDST, resulted in a sensitivity of 92% and specificity of 100% (123). Needless to say, this raises the philosophical question of when further refinements should be abandoned, since the published criteria are becoming ever more complex and the benefit of test simplicity begins to be lost. It is hardly surprising that repeat testing, or combining the results of overnight and 48-h HDDST (123), improves results, but a logical extension of this argument is that multiple HDDST should get the correct answer if repeated a sufficient number of times. Clearly, active management of patients precludes this, and, as such, adopting the simplest approach (48-h standard HDDST using plasma cortisol samples) and combining the results with tests that act in a physiologically distinct manner would appear to be a more rational approach.

A more reductionist suggestion by Findling et al. (132) has been to call for the abandonment of the HDDST altogether, as it has been shown to provide little diagnostic advantage over clinical assessment in the differential diagnosis of ACTH-dependent Cushing’s syndrome. The most recent study from these authors examined the effectiveness of the HDDST in clinical practice, whether conventional or overnight, and used simultaneous BIPSS as the gold standard for diagnosis of the origin of ACTH secretion (133). Rather than reporting extensive single-center experience, this thought-provoking work illustrated the results of dexamethasone suppression tests performed by physicians referring patients to the authors for BIPSS, and the rigor with which these tests had been performed was deliberately ignored in the study design. In their consecutive series of 112 patients referred for BIPSS, and using the standard response criterion of suppression of the postdexamethasone cortisol by greater than 50% as being consistent with Cushing’s disease, the HDDST had a sensitivity of 81% and a specificity of 67% and as such is less accurate than the pretest likelihood of Cushing’s disease. At no response level was it possible to achieve 100% specificity. Their analysis is uniquely based on the HDDST in practice and is, in effect, a meta-analysis, but clearly such data would be radically altered if some of the testing was being performed suboptimally for any number of the reasons that they highlight. Furthermore, the classification into pituitary or ectopic secretion of ACTH is based on the results of BIPSS as the ‘gold standard,’ rather than surgical confirmation, although this in itself is not a perfect test (see below). The HDDST has been shown to have better performance elsewhere, particularly when the standard 48-h HDDST is used (85, 121, 122, 123), and one conclusion might be that it should only be performed in centers with large experience. It is probably too early to call for the abandonment of the HDDST, but it certainly highlights the importance of the complete rigor required in all endocrinological assessment.

2. Metyrapone testing. Liddle and co-workers (134) introduced the ‘long metyrapone test’ to differentiate primary adrenal causes of Cushing’s syndrome from other causes (134). This test is based upon the fact that metyrapone inhibits the synthesis of cortisol by inhibiting the cleavage of cholesterol to form pregnenolone (135) and, through the inhibition of 11ß-hydroxylase, preventing the hydroxylation of 11-deoxycortisol to form cortisol (135). In patients with primary adrenal pathology, administration of metyrapone should not result in a rise in 17-OHCS excretion; in Cushing’s disease, as a result of lowering of plasma cortisol and hence decreased negative feedback at both hypothalamic and pituitary levels, this should result in a compensatory increase in plasma ACTH. This will overcome the early step of metyrapone inhibition, producing a rise in urinary 17-OHCS secretion and an increase in plasma 11-deoxycortisol (135). Test protocols involve the collection of 24-h urine specimens for the estimation of 17-OHCS or 11-deoxycortisol excretion and/or the determination of plasma 11-deoxycortisol. Metyrapone at a dose of 750 mg is administered orally at 4-h intervals beginning at 0800 h for six doses, and urine and blood samples are collected on the day before, the day of, and the day after metyrapone administration. Crapo (51) has analyzed the data from 15 separate studies and shown that 101 of 110 (98%) patients with Cushing’s disease demonstrated an increase in urinary 17-OHCS or 17-KGS, while only 8 of 49 patients with adrenal tumors showed such an increase. It should be noted, however, that 6 of 13 patients with ectopic ACTH secretion also had a rise in their urinary 17-OHCS or 17-KGS levels, affording very poor differentiation between ACTH-dependent groups. As is the case for the HDDST, the main use of the metyrapone test in more recent years has been in the differential diagnosis of ACTH-dependent Cushing’s syndrome (136, 137, 138, 139, 140, 141, 142). The largest and most recent study again comes from the NIH (143). In this series, using surgical cure as the gold standard for pituitary disease, a rise in urinary 17-OHCS of more than 70% or a rise in plasma 11-deoxycortisol of more than 400-fold from baseline was seen in 71% of 170 patients with Cushing’s disease, but not in any of the 15 patients with the ectopic ACTH syndrome; these data indicate a test sensitivity of 71% and specificity of 100% for the diagnosis of Cushing’s disease, although one patient who was ultimately classified as having a unilateral hyperfunctioning adrenal nodule exhibited a rise in plasma 11-deoxycortisol of 730-fold (a rise, but probably not of this magnitude, would be predicted from enzymatic blockade). These authors were able to achieve a greater sensitivity, while maintaining 100% specificity, by combining the results of high-dose dexamethasone suppression with the results of the metyrapone test; a rise in urinary 17-OHCS of more than 70% or a rise in plasma 11-deoxycortisol of more than 400-fold on metyrapone testing, or on high-dose dexamethasone testing a suppression of urinary 17-OHCS by 69% or UFC by 90%, was seen in 88% of 170 patients with Cushing’s disease but not in any of the patients with ectopic ACTH secretion (the independent results of the HDDST of this study have been discussed above). Such a result is comparable to the far more easily administered and interpreted CRH test (see below). This combined HDDST/metyrapone test is far more cumbersome than CRH testing and appears to be less accurate than combining the results of HDDST and CRH (4, 122, 144). Metyrapone and dexamethasone are, however, inexpensive and widely available; therefore, if CRH cannot be obtained, testing with metyrapone in this fashion may be a reasonable, although inferior, option.

To simplify testing with metyrapone, a shorter test was also developed to distinguish Cushing’s disease from primary adrenal pathology (145, 146, 147, 148). More recently, its use has been compared with the standard long metyrapone test in its ability to distinguish between 57 patients with Cushing’s disease and 6 patients with the ectopic ACTH syndrome (149). Administration of metyrapone at a dose of approximately 30 mg/kg at 2400 h and analysis of 0900 h plasma values before and after metyrapone administration showed that suppression in plasma cortisol of more than 40%, or an increase of plasma 11 deoxycortisol by more than 220-fold, was seen in 37 of 57 patients with Cushing’s disease (sensitivity, 65%), but not in any of the patients with the ectopic ACTH syndrome (specificity, 100%). The sensitivity, at a specificity of 100%, was improved to 84% by combining the results of the long and short test, but to achieve this the authors had to revise their previously documented criteria for the long metyrapone test (143). By itself, the short metyrapone test has very poor sensitivity and should be abandoned.

3. Testing with CRH. CRH was identified by Vale and co-workers in 1981 (150), and since this time it has been extensively used in the differential diagnosis of ACTH-dependent Cushing’s syndrome. It had been hoped that this test would allow complete discrimination between pituitary and ectopic ACTH secretion; in the majority of patients with Cushing’s disease the intravenous administration of CRH causes an excessive rise in plasma ACTH and cortisol, while in patients with the ectopic ACTH syndrome, such an effect is seen only rarely (24, 30, 94, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161). It seems likely that this disparity in response relates to the relatively greater expression of the CRH receptor in corticotroph adenomas compared with tumors ectopically secreting ACTH. It is important to note that the majority of reports documenting the use of this peptide in this context have used the ovine (oCRH) rather than the human-sequence peptide (hCRH).

Testing with CRH has been performed in the morning (122) and evening (158), and the most recent reports highlight the clinical utility of morning testing (156, 161). Since the circadian rhythm of cortisol secretion is lost in Cushing’s syndrome, it is unnecessary to go to the added inconvenience of testing at 2000 h (127). The test is performed with the patient in a rested, fasted, and recumbent state. Most test protocols take samples for plasma ACTH and cortisol at basal samples between -15 and 0 min, and stimulated samples at 15, 30, 45, 60, 90, and 120 min after the intravenous administration of CRH, 1 µg/kg body weight, or a total dose of 100 µg. The test is well tolerated, with side effects consisting of mild short-lived mild facial flushing and a metallic taste in the mouth.

As a group, the responses to these peptides seen in patients with Cushing’s disease and the ectopic ACTH syndrome differ in a quantitative rather than a qualitative fashion, and thus the absolute responses are of less value than percentage changes from basal values. In their meta-analysis of 10 studies reporting the use of peripheral CRH testing in 129 patients with Cushing’s disease, 21 with the ectopic ACTH syndrome and 29 with primary adrenal disease, Kaye and Crapo (127) suggested diagnostic criteria consistent with Cushing’s disease as being a rise from basal in peak plasma cortisol of 20%, or a rise in peak plasma ACTH of 50% after the administration of CRH. When these criteria are used for the plasma ACTH responses, the test has a sensitivity of 86% and a specificity of 95%, while plasma cortisol responses give an improved sensitivity of 91% and a similar specificity of 95%. Plasma cortisol samples are far more easily handled and analyzed and thus have advantages over and above the improved sensitivity compared with ACTH sampling. In contrast, in the largest reported series from the NIH, the responses to oCRH, administered at 0800 h, that best discriminated between pituitary and nonpituitary origins of ACTH secretion was a rise of 35% or more in the mean plasma ACTH concentrations at 15 and 30 min above the mean basal value at -5 and -1 min; this was seen in 93 of 100 patients with Cushing’s disease, while a response less than this was observed in all 16 patients studied with the ectopic ACTH syndrome (13 of which had carcinoid tumors) giving a sensitivity of 93% and a specificity of 100% (156). Nevertheless, analysis of the responses utilizing the other basal time points employed (-15 and 0 min, or combinations thereof) revealed that some of the patients with the ectopic ACTH syndrome would have been misclassified. The plasma cortisol responses were less impressive, with a rise of 20% or more at the mean of the levels at 30 and 45 min giving a sensitivity of 91% and a specificity of 88%: no combinations of time points allowed the achievement of 100% specificity without sensitivity being severely compromised. The earlier time points used for ACTH reflect the time course in the response to CRH that tends to peak at 15–30 min. The authors themselves recommend more cautious cut-offs to guarantee specificity and also suggest the use of cortisol responses, unless the Hazelton RIA for ACTH, as used in their study, is employed. The number of patients studied in this latter single study (156) is comparable in size to the total number analyzed by Kaye and Crapo (127), and yet differing response criteria are recommended. This may be explained, in part, by differences in study protocols, since the meta-analysis included studies utilizing evening testing, and in part by different assays employed. The sensitivity of the response criteria set out by Nieman and co-workers (158) has also been validated by others (151).

The human sequence peptide has similar effects to the oCRH in normal individuals and patients with Cushing’s disease (162, 163, 164, 165). Although some reports indicate that testing with hCRH is less accurate than when using oCRH (152), we have found that the responses are qualitatively similar, albeit with a quantitatively lower response to hCRH, in patients with Cushing’s disease, Cushing’s syndrome due to adrenal adenoma, and in obese and lean volunteers (155). Published large series are needed that report the responses to hCRH that best discriminate between patients with the ectopic ACTH syndrome and Cushing’s disease.

Combined analysis of all published series reveals that between 7% and 14% of all patients with Cushing’s disease fail to respond to CRH if the best discriminating criteria are applied. This is somewhat disappointing compared with the initial hopes for the clinical utility of CRH as a discriminating agent. Although most of the Cushing’s disease nonresponders exhibit suppression on a HDDST (122, 131), there are rare cases in which an ACTH-secreting bronchial carcinoid tumor may suppress on a HDDST and exhibit responsiveness to CRH (166). Such cases are very uncommon, and the use of high-dose dexamethasone testing and CRH stimulation will, in the vast majority of cases, allow correct classification.

4. Testing with vasopressin. For many years it has been known that vasopressin also stimulates ACTH release and, in particular, that it potentiates the ACTH-releasing effects of CRH (167, 168). These actions are thought to occur via the specific corticotroph vasopressin receptor, the V₃ (also known as V_1b) receptor, which has recently been cloned (169, 170). The lysine or arginine vasopressin (AVP) test has been used in the differential diagnosis of ACTH-dependent Cushing’s syndrome but has a false negative response in 27% of patients with Cushing’s disease (171, 172, 173, 174, 175, 176, 177, 178). Increases in urinary cortisol excretion have been observed in patients with Cushing’s disease after the administration of 10 U of intraperitoneal AVP (176), while serum cortisol responses to this dose of lysine vasopressin (LVP) in patients with Cushing’s disease have been noted to be less than that after 100 µg of CRH (179). Tabarin and co-workers (180) have shown that 18/21 patients with Cushing’s disease responded to CRH while 17 showed similar responses to vasopressin. In this same study 2 of 7 patients with the ectopic ACTH syndrome responded to LVP while none showed a response to CRH. Thus, CRH appears to discriminate better than LVP between ectopic ACTH secretion and Cushing’s disease. Furthermore, side effects consisting of abdominal pain, nausea, and flushing have precluded the routine clinical use of vasopressin for diagnostic testing, although it has been suggested that it may be better tolerated when used as a low-dose infusion, or as small bolus doses, in combination with CRH (181, 182). Overall, the LVP or AVP test appears to be inferior to CRH testing.

5. Testing with desmopressin. Desmopressin, a long-acting analog of vasopressin (183), has relative specificity for the renal V₂ receptor with little V₁-mediated pressor activity (184). While its specific V_1b receptor activity is uncertain, it has been shown previously to have no intrinsic in vivo ACTH-releasing characteristics when given as an infusion in man (185). Desmopressin has, however, been shown to cause a rise from baseline in peak plasma cortisol of more than 4 times the intraassay coefficient of variation (158) in 15 of 16 patients tested with Cushing’s disease when given as an intravenous bolus dose of 5–10 µg, but not in one patient with an ACTH-secreting pheochromocytoma; a complete data set for ACTH responses was not reported (186). Since this peptide appears to be free of the V₁ receptor-mediated pressor side effects, it has been suggested that it may be used to aid the differential diagnosis of the causes of ACTH-dependent Cushing’s syndrome. More recently, we have shown that a response, defined as a 20% rise in serum cortisol (156), after the administration of 10 µg desmopressin iv was seen in 14 of 17 patients with Cushing’s disease and 1 of 5 patients with histologically proven ‘occult’ ACTH-secreting ectopic tumors (an ACTH-secreting medullary cell carcinoma of the thyroid) (187). Using ACTH response criteria of a 35% rise or more (156), 12 of 17 patients with Cushing’s disease and 3 patients with the ectopic ACTH syndrome showed a response. These findings have been confirmed, using response criteria for CRH testing as defined by Kaye and Crapo (127), with responses being seen in 14 of 17 patients with Cushing’s disease, while only 1 patient with the ectopic ACTH syndrome was studied, and no response was seen to either desmopressin or CRH (188, 189). Combining the data of all published series (Table 4) reveals that for the desmopressin test the cortisol responses have a sensitivity of 84% and specificity of 83%, while ACTH responses provide poorer discrimination with a sensitivity of 77% and specificity of 73%. Therefore, testing with desmopressin is inferior to testing with CRH in terms of sensitivity and specificity, although this peptide is cheaper and more easily available worldwide. Although the total numbers are small, the results are in keeping with those for LVP and AVP testing (151, 180). A possible explanation for the relatively poorer specificity of the desmopressin test is the more common expression of the V_1b (or V₃) receptor in ACTH-secreting nonpituitary tumors (190, 191). As such, it seems likely that testing with desmopressin alone will result in poorer discrimination than testing with CRH, although more studies are needed to confirm this impression (192). Nevertheless, some patients with Cushing’s disease respond only to one peptide or the other (186, 187, 188, 192), and thus in certain circumstances testing with desmopressin may be useful.

7. Recent developments: testing with hexarelin. Administration of the synthetic peptide hexarelin, a member of the GH-releasing peptides (GHRPs) family, has recently been shown to have far greater ACTH- and cortisol-releasing effects than hCRH in 10 patients with Cushing’s disease (193). The responses seen were remarkable in terms of the absolute levels of stimulated plasma cortisol, and even more impressively plasma ACTH, both of which were far higher than those seen even when testing with oCRH or hCRH in patients with Cushing’s disease. In contrast, no ACTH rise was seen in two patients with the ectopic ACTH syndrome, while the cortisol response in one overlapped those of the Cushing’s disease group. No ACTH or cortisol response was seen in 5 patients with cortisol-secreting adrenal adenomas. These results are very promising, and if replicated in larger studies, this may prove to be an extremely useful tool: even so, one cortisol responder in the ectopic ACTH group suggests that GHRP receptors may have also been present in these tumors, and indeed GHRP receptor expression in ACTH-secreting nonpituitary tumors has recently been demonstrated (194, 195). It seems likely, therefore, that with further experience more responders in the ectopic ACTH syndrome group will be reported.

D. Invasive testing
1. Inferior petrosal sinus sampling for ACTH. In the discrimination between pituitary and ectopic sources of ACTH, none of the noninvasive tests discussed above have been validated as providing 100% diagnostic accuracy in large series or meta-analyses. Therefore, alternative strategies have been developed. Although in many circumstances peripheral biochemical tests will provide evidence of pituitary disease, the presence of a pituitary lesion on imaging, especially if less than 4 mm in diameter, does not necessarily confirm functionality, since 10% of the general population harbor pituitary incidentalomas (196). Moreover, in at least 40–50% of cases of Cushing’s disease, no abnormality will be disclosed on pituitary imaging (96, 197, 198). Thus, venous sampling from the inferior petrosal sinuses and/or cavernous sinuses is now widely practiced 1) to confirm or refute a central (pituitary) source of ACTH, especially when pituitary imaging is negative, and 2) to attempt to lateralize the site of a pituitary tumor to guide neurosurgical approaches.

The technique was originally described by Corrigan and co-workers (199) and involves the placement of venous sampling catheters in the inferior petrosal sinuses that drain the pituitary venous effluent (200, 201). A significant gradient between the pituitary (central) and peripheral values of plasma ACTH, obtained by simultaneous sampling, is indicative of Cushing’s disease. Although in early reports the procedure was performed with sequential catheterization of each of the petrosal sinuses (199, 202, 203), it was soon realized that simultaneous bilateral inferior petrosal sampling was required as the drainage of the pituitary tends to have dominant and nondominant drainage to the inferior petrosal sinuses, and therefore unilateral sampling may miss a central source (204, 205, 206, 207). The basal ratio of the values of plasma ACTH obtained from central and peripheral samples that have been taken as indicative of Cushing’s disease have been reported as greater than 1.4 or 1.5 (206, 208, 209, 210), or greater than 2.0 (4, 39, 40, 132, 211, 212, 213, 214, 215, 216, 217, 218). Since patients with the ectopic ACTH syndrome have been documented to have maximal basal ratios of 1.7 or 2.0 (211, 219), it would seem prudent to take this more conservative ratio of 2.0 (4). Although a baseline ratio of more than 2.0 is consistent with Cushing’s disease, ACTH secretion is intermittent, and a significant minority of patients with Cushing’s disease have a ratio less than this on the basal samples. For this reason, after basal samples are obtained, a stimulating agent, most commonly CRH, is usually administered to increase the sensitivity of the test. Both hCRH and oCRH have been used for this purpose with g