Introduction

Cushing’s Syndrome results from chronic exposure to excessive circulating levels of glucocorticoids. It is now one hundred years since Harvey Cushing reported the classical clinical syndrome that bears his name. Even now its investigation and management can vex the most experienced endocrinologist. It may be difficult to miss the diagnosis in its most florid form. But, given the high prevalence of many of its non-specific symptoms such as obesity, muscle weakness and depression, clinicians are now considering the diagnosis in its earlier manifestations. The plethora of investigations often needed for the diagnosis and differential diagnosis has grown over the intervening century, and require careful interpretation. In its severe form and when untreated, the metabolic upset of Cushing's syndrome is associated with a high mortality. Yet, more subtle excesses of cortisol may also have significant effects on glycaemic control and blood pressure, and may therefore be an important cause of morbidity. Treatment is often complex and may require all the modalities of surgery, radiotherapy and medical management.

Pathophysiology , Aetiology & Epidemiology

In normal physiology the end product of the hypothalamo-pituitary-adrenal (HPA) axis is the glucocorticoid cortisol, secreted from the zona fasciculata and reticularis of the adrenal gland under the stimulus of adrenocorticotropin (ACTH) from the pituitary gland. ACTH in turn is secreted in response to corticotropin releasing hormone (CRH) and vasopressin from the hypothalamus. Cortisol exerts negative feedback control on both CRH and vasopressin in the hypothalamus, and ACTH in the pituitary. In normal individuals, cortisol is secreted in a circadian rhythm; levels fall during the day from a peak at 07.00h-08.00h to a nadir at around midnight: they then begin to rise again at 02.00h. It is the loss of this circadian rhythm, together with loss of the normal feedback mechanism of the hypothalamo-pituitary-adrenal (HPA) axis, which results in chronic exposure to excessive circulating cortisol levels and that gives rise to the clinical state of endogenous Cushing's syndrome (1). Any of the numerous synthetic steroids that have glucocorticoid activity, administered in excessive quantities can give rise to exogenous Cushing's syndrome. This is the commonest cause of Cushing's syndrome seen in general clinical practice, usually due to treatment for chronic conditions such as asthma or rheumatological disease.

The etiology of Cushing's syndrome can broadly be divided into two categories; ACTH-dependent and ACTH-independent (Table 1).

Table 1 . Aetiology of Cushing's syndrome

ACTH-dependent Pituitary-dependent Cushing's syndrome (Cushing's disease) Ectopic ACTH syndrome Ectopic CRH syndrome Exogenous ACTH administration

ACTH-independent Adrenal adenoma Adrenal carcinoma ACTH-independent bilateral macronodular adrenal hyperplasia (AIMAH) AIMAH secondary to abnormal hormone receptor expression/function Primary pigmented nodular adrenal disease (PPNAD), associated with Carney complex or sporadic McCune-Albright syndrome Exogenous glucocorticoid administration

ACTH-dependent Cushing's syndrome.

Cushing's Disease

Pituitary-dependent Cushing's syndrome, better known as Cushing's disease, is the most common at least in endocrine practice, accounting for 60-80% of all cases. Epidemiologic studies from Europe suggest an incidence of between 0.7 and 2.4 per million per year (2;3). It presents much more commonly in women, and usually between the ages of 25 to 40 years of age. It is almost always due to a monoclonal corticotroph adenoma (4;5). Although apparent nodular corticotroph hyperplasia (in the absense of an CRH-producting tumour) has been described, it is rare in large surgical series (6;7), and its existence is still debated. The majority of tumors are intrasellar microadenomas (<1 cm in diameter), although macroadenomas account for approximately 5-10% of tumors, and extrasellar extension or invasion may occur. True pituitary corticotroph carcinomas with extra-pituitary metatstases causing Cushing's syndrome have also rarely been described (8;9). Despite much research the molecular pathogenesis of corticotroph adenomas remains unknown (10).

Ectopic ACTH syndrome & ectopic CRH tumors

Ectopic sources of ACTH derive from a diverse group of tumor types. The most frequent cause is probably small-cell lung carcinoma where it is estimated up to 12% of cases will have Cushing's syndrome (11). This may not be evident from series at endocrine centres where often more occult tumours are investigated (Table 2), and carcinoid tumours tend to predominate. Ectopic ACTH syndrome is more common in men, and usually presents after the age of 40 years.

Table 2. Aetiology of the ectopic ACTH syndrome in patients seen at St. Bartholomew's Hospital 1969-2001
Site of secretion	Female	Male
Bronchial carcinoid tumor	11	2
Small cell lung carcinoma	1	5
Medullary thyroid carcinoma		3
Pancreatic carcinoid tumor	1	2
Thymic carcinoid tumor		1
Disseminated carcinoid tumor		1
Mesothelioma	1
Pancreatic carcinoma		1
Colonic carcinoma	1
Phaechromocytoma		1
Gall bladder carcinoma	1
Total	16	16

The mechanism by which these non-corticotroph tumours express the proopiomelanocortin (POMC) gene is not fully understood but may be related to hypomethylation of the POMC promotor in some tumours (12;13). As well as producing ACTH these tumours may also produce the precursor POMC, and other pre-ACTH peptides, so -called "big" ACTH (14;15), which may potentially be helpful in the differential diagnosis of these tumors (16). However, assays for these are not routinely available in clinical practice.

Isolated Ectopic CRH production is difficult to diagnose and exceedingly rare, with just over 20 cases described in the literature (17).

ACTH-independent Cushing Syndrome

ACTH-independent causes of Cushing syndrome apart from exogenous glucocorticoids encompass a heterogeneous group of diseases.

The most common pathology is an adrenal adenoma or carcinoma. The latter may lack some classic histological features of malignancy, but can usually be differentiated on the basis of weight (more than 100g), nuclear pleomorphism, necrosis, mitotic figures, and vascular or lymphatic invasion. Adrenal adenoma occurs most often around 35 years of age and is significantly more common in women with an incidence of approximately 0.6 per million per year (18). The incidence of adrenal cancer is approximately 0.2 per million per year (18). It is one and a half times more common in women, and has a bimodal age distribution, with peaks in childhood and adolescence and late in life (1;19).

ACTH-independent bilateral macronodular adrenal hyperplasia (AIMAH) is a rare form of Cushing’s syndrome with sometimes huge nodular adrenal glands on imaging. Most cases are sporadic, but a few familial cases have been reported (20). In most the etiology is unknown, but in a few cases the nodules have been shown to express increased numbers of receptors normally found on the adrenal gland, or ectopic receptors that then can stimulate cortisol production. The best described example is food-dependent Cushing’s syndrome: in which ectopic gastric inhibitory peptide (GIP) receptors on the adrenal glands respond to GIP released after a meal causing hypercortisolaemia (21). In one patient, treatment with octreotide ameliorated the syndrome (22). Abnormal expression of vasopressin,  -adrenergic, luteinizing hormone/human chorionic gonadotropin, serotonin, angiotensin, leptin, glucagon, IL-1, and TSH have also been described and functionally linked to cortisol production (23). AIMAH tissue may express more than one of these aberrant receptors (24).

Cushing’s syndrome due to bilateral nodular adrenal disease can also be an unusual feature of McCune-Albright syndrome (25). This condition is caused by an activating mutation at codon 201 of the  -subunit of the G protein stimulating cyclic adenosine monophosphate (cAMP) formation. This occurs in a mosaic pattern in early embryogenesis (26). The characteristic features are fibrous dysplasia of bone, café-au-lait skin pigmentation, and endocrine dysfunction (usually precocious puberty). However, if this affects some adrenal cells the constitutive activation of adenylate cyclase leads to nodule formation and glucocorticoid excess. The normal adrenal cortex, where the mutation is not present, becomes atrophic (27).

Primary pigmented nodular adrenal disease (PPNAD), otherwise known as micronodular adrenal disease, is another rare form of Cushing’s syndrome. It is characterised by small or normal-size adrenal glands with cortical micronodules (average 2–3 mm) that may be dark or black in color. The internodular cortex is usually atrophic (28). Most cases of PPNAD occur as part of the Carney complex in association with a variety of other abnormalities, including myxomas of the heart, skin or breast; hyperpigmentation of the skin, and other endocrine disorders (sexual precocity; Sertoli cell, Leydig cell, or adrenal rest tumors; and acromegaly). Cushing’s syndrome occurs in approximately 30% of cases of Carney Complex. The tumor suppressor gene PRKAR1A (type 1A regulatory subunit of protein kinase A) has been shown to be mutated in approximately half of patients with Carney complex. In isolated PPNAD, mutations in PRKAR1A and also the phosphodiesterase 11A (PDE11A) gene have demonstrated (29).

A missense mutation of the ACTH receptor resulting in its constitutive activation and ACTH-independent Cushing’s syndrome has also been reported (30).

Other very rare causes of Cushing's syndrome have been reported: Adrenal rest tissue in the liver, in the adrenal beds, or in association with the gonads may produce hypercortisolaemia, usually in the context of ACTH-dependent disease after adrenalectomy (31-34). Ectopic cortisol production by an ovarian carcinoma has been reported (35).

Pseudo Cushing's syndrome

Pseudo-Cushing's states are conditions in which a patient presents with clinical features suggestive of true Cushing's syndrome and with some biochemical evidence of hypercortisolaemia. Both settle after resolution of the pre-disposing condition. The pathophysiology has not clearly been established. Depression and alcohol abuse are the two most common such states (1).

Clinical & Laboratory Features

The clinical manifestations associated with hypercortisolaemia are variable and differ widely in severity. The classical impression of the disease as the association of gross obesity of the trunk with wasting of the limbs, facial rounding and plethora, hirsutism with frontal balding, muscle weakness, spontaneous bruising, vertebral fractures, hypertension and diabetes mellitus, is rarely seen nowadays in its most obvious form. More commonly the clinical diagnosis may be more equivocal. Many symptoms common in Cushing's syndrome including lethargy, depression, obesity, hypertension, hirsuitism, and menstrual irregularity are also very common in the general population, and it is useful to have a strategy looking for more specific clues when considering the diagnosis.

The distribution of fat can be useful as typically in Cushing's syndrome there is increased visceral adiposity giving rise to truncal obesity, fat deposition in the face ("moon face"), dorsocervical area ("buffalo hump"), supraclavicular , and temporal fossae. The signs that are most discriminatory are proximal myopathy, osteoporosis, and thin skin and easy bruising (36;37). When assessing for myopathy it is useful to ask questions about function typically affected by proximal muscle weakness such as climbing stairs or getting up from a chair. Formal testing can be of leg extension whilst sitting, or rising unaided from a squatting position. Osteoporosis occurs in approximately 50% of adult patients (38) and can be gleaned from formal bone densitometry, or from a history of fractures, typically vertebral due to the preferential loss of trabecular bone induced by glucocorticoids.

Skin thickness is best tested over the dorsum of the hand, but it is helpful to consider the age and gender of the patient as natural atrophy increases with age and female gender. In addition, skin thickness may be preserved in women with hyperandrogenaemia related to the Cushing's syndrome.

Purple coloured "Violaceous" striae greater than 1cm across are almost pathognomonic of Cushing's syndrome. Typically seen on the abdomen, they can also occur in other areas. Less wide and coloured striae are more commonly present, and should be differentiated from the typical healed silvery striae seen most commonly post-partum.

Increased fine non-pigmented vellus hair on the upper cheeks or forehead may be seen in Cushing’s syndrome, as well as more typical terminal hair hirsuitism on the face and body, reflecting increased androgens.

Sequential photographs of the patient over many years can be extremely helpful in demonstrating progression to a Cushingoid state. In children, weight gain associated with growth retardation should highlight the possibility of the diagnosis (39).

The rapid onset of the symptoms of profound weakness, associated with hyperpigmentation, often with little or no weight gain and an absense of a gross Cushingoid appearance, is almost certainly due to the ectopic ACTH syndrome, most often from small cell lung carcinoma. However, other forms of the ectopic ACTH syndrome, particularly associated with carcinoid tumors, may be clinically indistinguishable from patients with other forms of hypercortisolism (40). Severe hirsutism and virilisation strongly suggest an adrenal carcinoma (41).

Some cases of ACTH-dependent Cushing's syndrome occur in a periodic or cyclical form, with intermittent and variable cortisol secretion, the symptoms and signs waxing and waning according to the active periods of the disease. These patients can cause particular diagnostic difficulty, as it is imperative that the diagnostic tests are performed in the presence of hypercortisolaemia to allow accurate interpretation. Patients may 'cycle in' or 'cycle out' over periods of months or years; if at presentation they are eucortisolaemic,they will need regular re-evaluation usually with urinary free cortisol or late-night salivary cortisol to allow full investigation at the appropriate time. Cyclicity can occur with all causes of Cushing’s syndrome (42)..

Although, the hallmark of Cushing's syndrome is hypercortisolaemia, other laboratory abnormalities may co-exist.

Hypokalemic metabolic alkalosis is related to the degree of hypercortisolaemia and represents a mineralocorticoid action of cortisol at the renal tubule due to saturation of the enzyme 11  -hydroxysteroid dehydrogenase type 2, which inactivates cortisol to cortisone (43). It occurs when urine free cortisol excretion is greater than about 4100 nmol per day (44). Therefore, although a more common feature of ectopic ACTH secretion, it may also occur in approximately 10% of patients with Cushing’s disease.

Hypercortisolism suppresses other pituitary hormones. In both men and women, hypogonadotrophic hypogonadism is common and correlates with the degree of hypercortisolemia (45;46). In addition, the coexistence of polycystic ovarian syndrome in Cushing’s syndrome is more common (47). There is reduced GH secretion during sleep and blunted GH responses to dynamic stimulation tests (48). Thyrotropin- releasing hormone and thyroid-stimulating hormone release has been shown to be disturbed and in particular the nocturnal surge of thyroid-stimulating hormone is lost (49). This may not have a significant effect on free thyroid hormone levels during active hypercortisolaemia, but there is a significantly increased prevalence of autoimmune thyroid disease in patients sucessfully treated for Cushing’s syndrome, and therefore it is important to follow them with serial thyroid function tests (50;51).

Cushing’s syndrome is characterized by insulin resistance and hyperinsulinaemia with glucose intolerance evident in 20% to 30%, and overt diabetes mellitus in 30% to 40% of patients, (52;53). It has been suggested that as many as 2% of overweight, poorly-controlled patients with diabetes may have occult Cushing’s syndrome if fully investigated (54). However, in the absence of clinical suspicion the yield is probably lower (55) (56).

There is an increase in total cholesterol and triglyceride levels, and variable effect on high-density lipoprotein (HDL). These changes are multifactorial, including cortisol effects on increased hepatic synthesis of very low density lipoprotein (VLDL), lipolysis, and free fatty acid metabolism (57).

Hypercortisolaemia increases clotting factors including factor VIII, fibrinogen, and von Willebrand factor, and reduces fibrinolytic activity. This along with other risk factors such as obesity, surgery and invasive investigative procedures results in a significant increased risk of thrombotic events in patients with Cushing's syndrome (58).

The major cause of mortality in Cushing's disease is cardiovascular events, and patients exhibit direct markers of accelerated cardiovascular disease, including increased carotid artery intima-media thickness and atherosclerotic plaques. These markers and other cardiovascular risk factors such as hyperinsulinaemia continue to be present long after cure of the hypercortisolemia (59). Sympathetic autonomic function is also abnormal in patients with Cushing's syndrome (60), and the ECG abnormalities of a prolonged QTc dispersion (QTcd) and left ventricular hypertrophy have been identified as characteristic features in patients with Cushing's disease (61).

Biochemical confirmation of Cushing's syndrome

As stated above, hypercortisolaemia together with the loss of the normal circadian rhythm of cortisol secretion, and disturbed feedback of the HPA axis, are the cardinal biochemical features of Cushing's syndrome. Tests to confirm the diagnosis are based upon these principles. To screen for Cushing's syndrome, tests of high sensitivity should be used initially, so as to avoid missing milder cases. Tests of high specificity can then be employed to exclude false positives. It is important to realise that the validation of the published test criteria employed have been on specific assays, and thus test responses should ideally be validated on the local assay used before the results can be interpreted in particular patients. This is aided by supra-regional and nationwide inter-assay quality control assurance (1). Exogenous oestrogens will increase cortisol-binding globulin and therefore total cortisol levels. Hence, in all investigations relying on a serum cortisol assay that measures total cortisol, hormone replacement therapy or the oral contraceptive pill should be stopped some time prior to investigation. We currently wait for a period of 6 weeks, although it is likely that a shorter time off treatment may still be effective.

Urinary free cortisol

Measurement of urinary free cortisol (UFC) is a non-invasive test that is widely used in the screening of Cushing's syndrome. Under normal conditions, 10% of plasma cortisol is 'free' or unbound and physiologically active. Unbound cortisol is filtered by the kidney, with the majority being reabsorbed in the tubules, and the remainder excreted unchanged. Thus, 24-hour UFC collection produces an integrated measure of serum cortisol, smoothing out the variations in cortisol during the day and night. In a series of 146 patients with Cushing's syndrome, UFC measurement was shown to have a sensitivity of 95% for the diagnosis (62). However, within this series 11% had at least one of four UFC collections within the normal range, which confirms the need for multiple collections. The 24 hour UFC is of little value in the differentiation from pseudo-Cushing's states (63;64), although obesity per se does not appear to confound the results (65). The major drawback of the test is the potential for an inadequate 24-hour urine collection, and written instructions must be given to the patient. In addition, creatinine excretion in the collection should be measured to assess completeness, and should equal approximately 1g/ 24 hours in a 70kg patient (variations depend on muscle mass). This should not vary by more than 10% between collections in the same individual (41). High-performance liquid chromatography or tandem mass spectrometry are now used to measure UFC, which overcomes the previous problem with conventional radioimmunoassay of cross-reactivity of some exogenous glucocorticoids and other structurally similar steroids (66). Drugs such as carbamazepine, digoxin and fenofibrate may coelute with cortisol during high- performance liquid chromatography and cause falsely elevated results (67;68).

In summary, UFC measurements have a high sensitivity if collected correctly, and several completely normal collections make the diagnosis of Cushing's syndrome very unlikely. Values greater than fourfold normal are rare except in Cushing's syndrome. For intermediate values the specificity is somewhat lower, and thus patients with marginally elevated levels require further investigation (1).

Low-dose dexamethasone suppression test

This test works on the principle that in normal individuals administration of an exogenous glucocorticoid results in suppression of the HPA axis, whilst patients with Cushing's syndrome are resistant, at least partially, to negative feedback. Dexamethasone is a synthetic glucocorticoid that is 30 times more potent than cortisol, and with an extremely long duration of action. It does not cross-react with most cortisol assays. The original low-dose dexamethasone test (LDDST) described by Liddle in 1960 measured urinary 17-OHCS after 48 hours of dexamethasone 0.5mg 6 hourly (69). However, the simpler measurement of a single plasma or serum cortisol has been validated in various series and gives the test a sensitivity of between 95% and 100% (70-72). The overnight LDDST was first proposed by Nugent et al in 1965; this measured a 09.00h plasma cortisol after a single dose of 1mg dexamethasone taken at midnight (73), and is thus considerably easier to perform. Since then, various doses have been suggested for the overnight test, between 0.5 and 2mg, and various diagnostic cut-offs have been used (74-76). There appears to be no advantage in discrimination between 1mg and 1.5mg or 2mg (77). Although higher doses have been tried, the increased suppression in some patients with Cushing's syndrome significantly decreases the sensitivity of the test (78). In a comprehensive review of the LDDST, both the original 2 day test and the overnight protocol appear to have comparable sensitivities (98%-100%) using the criteria of a post-dexamethasone serum cortisol of <50nmol/l (1.8 g/dL) (79). However, the specificity is greater for the 2 day test (95%-100%) compared to the overnight test (88%) (79). Despite the greater simplicity of the overnight test, we would still advocate the 48 hour test, even as an out-patient, as long as written instructions are given to the patient. If the overnight test is used, we suggest that a dose of dexamethasone 1mg at midnight and a threshold of <50nmol/l (1.8 g/dL) at 09.00h will rarely lead to the diagnosis being missed, but false positives remain significant.

It should be noted that patients with PPNAD may show a paradoxical rise in cortisol levels to dexamethasone (80).

Factors such as variable absorption and increased metabolism can influence dexamethasone test results. A history of symptoms of malabsorption and a careful drug history should be taken prior to using the test in a patient. Hepatic enzyme inducers such as carbamazepine, phenytoin, phenobarbitone and rifampicin will reduce plasma dexamethasone concentrations (81), and will usually render the test uninterpretable. Causes of increased or decreased corticosteroid-binding globulin (CBG) (such as pregnancy, exogenous estrogens and the nephrotic syndrome) should also be excluded because these may result in false-positive and false-negative tests(82).

Late night salivary cortisol

Salivary cortisol measurement accurately reflects the plasma free cortisol concentration. Loss of the circadian rhythm of cortisol secretion by measuring night-time salivary cortisol can be utilised as a screening test for Cushing’s syndrome. Due to the simple non-invasive collection procedure which can conveniently be performed at home, and the fact that salivary cortisol is stable for days at room temperature, it offers a number of attractive advantages over blood collection, particularly in children. Assays using a modification of the plasma cortisol radioimmunoassay, enzyme-linked immunosorbent assay, or liquid chromatography tandem mass spectrometry are now widely available. Over the past decade there has been considerable increasing interest in this test, and it has been evaluated at a large number of centres worldwide. In a recent meta-analysis of these studies, in adult patients the sensitivity and specificity of this test appears to be relatively consistent at different centres, and overall is 92% and 96% respectively (83). However, it should be noted that the diagnostic value cut-off varies between studies because of different assays and the comparison groups studied. Normal values also differ between adults and paediatric populations, and may be affected by other comorbidities such as diabetes (84), and the method by which the saliva is collected (85). Not surprisingly, this test performs less well in patients with subclinical Cushing's syndrome (86).

Salivary rather than serum cortisol has been evaluated as the endpoint for the LDDST. This has potential benefit in terms of convenience but requires further evaluation(87;88).

Salivary cortisol has also been advocated as a sensitive tool to detect recurrence or treatment failure in patients post-pituitary surgery for Cushing's disease (89;90).

In summary, late-night salivary cortisol appears to be a useful and convenient screening test for Cushing's syndrome, particularly in the outpatient setting. However, local normal ranges need to be validated based on the assay used and population studied.

2nd line tests

In some patients with equivocal results other tests may be needed. The most useful of these are a midnight serum cortisol, and the dexamethasone-CRH test. Less reliable tests such as the insulin tolerance test (91), the loperamide test (92) , and the desmopressin test (37)are not advocated.

Midnight serum cortisol

Before the introduction of salivary cortisol measurement a midnight serum cortisol was the only reliable method used to determine loss of the circadian rhythm of cortisol secretion. It is still useful as a second line test in cases of diagnostic difficulty. However, it is a burdensome test that requires that the patient should have been an in-patient for at least 48 hours to allow acclimatisation to the hospital environment. The patient should not be forewarned of the test, and should be asleep prior to venepuncture, which must be performed within 5-10 minutes of waking the patient. A single sleeping midnight plasma cortisol <50nmol/l (1.8 g/dL) effectively excludes Cushing's syndrome (72), but false positive results do occur, particularly in the critically ill, in acute infection, heart failure, and in the pseudo-Cushing's state associated with depression (93). An awake midnight cortisol of greater than 207 nmol/l (7 g/dL) shows 94% sensitivity and 100% specificity for the differentiation of Cushing's syndrome from pseudo-Cushing's states (94).

The dexamethasone-CRH test

In 1993 a combined dexamethasone-CRH (Dex-CRH) test was introduced for the difficult scenario of the differentiation of pseudo-Cushing’s states from true Cushing’s syndrome in patients with only mild hypercortisolemia and equivocal physical findings (63). The theory being that a small number of patients with Cushing's disease as well as normal individuals will show suppression to dexamethasone, but those with Cushing's disease should respond to CRH with a rise in ACTH and cortisol. In the original description of the test, dexamethasone 0.5 mg every 6 hours was given for eight doses, ending 2 hours before administration of ovine CRH (1 µg/kg intravenously) to 58 adults with UFC less than 1000 nmol/day (360 µg/day). Subsequent evaluation proved 39 to have Cushing’s syndrome and 19 to have a pseudo-Cushing’s state. The plasma cortisol value 15 minutes after CRH was less than 38 nmol/L (<1.4 µg/dL) in all patients with pseudo-Cushing’s states and greater in all patients with Cushing’s syndrome. A prospective follow-up study by the same group in 98 patients continued to show the test to have an impressive sensitivity and specificity of 99% and 96%, respectively (95). Importantly, in these two studies although eight of 59 patients with proven Cushing's disease showed suppression to dexamethasone, all were correctly characterised after CRH. However, results from a number of other smaller studies have challenged the diagnostic utility of this test over the standard LDDST. Overall, in these reports the specificity of the LDDST in 92 patients without Cushing's syndrome was 79%, versus 70% for the Dex-CRH. The sensitivity in 59 patients with Cushing's syndrome was 96% for the LDDST versus 98% for the Dex-CRH (96). It is perhaps not surprising that the diagnostic utility of the Dex-CRH has altered with further studies at more centers. There are a number of reasons why there might be the case: variable dexamethasone metabolism in individuals; different definitions of patients with pseudo-Cushing's; different protocols and assays; and variable diagnostic thresholds. It is recommended that if this test is used, a dexamethasone level is measured at the time of CRH administration and the serum cortisol assay is accurate down to these low levels (97).

The Differential Diagnosis of Cushing's syndrome

Once Cushing's syndrome has been diagnosed, the next step is to differentiate between ACTH-dependent and ACTH-independent causes. Nowadays, the simplest method of differentiation is the measurement of plasma ACTH. Modern two-site immunoradiometric assays are more sensitive than the older radioimmunoassays and therefore provide the best discrimination. Rapid collection and processing of the sample is essential as ACTH is susceptible to degradation by peptidases so that the sample must be kept in an ice water bath and centrifuged, aliquoted, and frozen within a few hours to avoid a spuriously low result. It is useful to duplicate this test because patients with ACTH-dependent Cushing’s disease have been shown to have on occasion ACTH levels less than 10 ng/L (2 pmol/L) on conventional radioimmunoassay (98). Consistent ACTH measurements of less than 10 ng/L (2pmol/L) essentially confirm ACTH-independent Cushing's syndrome. Conversely, if levels are consistently greater than 20 ng/L (4pmol/L), Cushing's syndrome is ACTH-dependent. Intermediate levels are less discriminatory, but a lack of ACTH response to the CRH test (see below) may be particularly helpful. It is important to stress that all investigations into the differential diagnosis of Cushing's syndrome must be carried out in the presence of hypercortisolaemia and it is useful to confirm this with concomitant UFC or late-night salivary cortisol.

Investigating ACTH-independent Cushing's syndrome

Imaging of the adrenal glands is the mainstay in differentiating between the various types of ACTH-independent Cushing's syndrome. High-resolution computed tomography (CT) scanning of the adrenal glands is the investigation of choice and is accurate for masses greater than 1 cm and allows evaluation of the contralateral gland (99). In certain circumstances MRI may be useful for the differential diagnosis of adrenal masses; the T2-weighted signal is progressively less intense in phaeochromocytoma, carcinoma, adenoma, and finally normal tissue (100). Adrenal tumours typically appear as a unilateral mass with an atrophic contralateral gland (101). If the lesion is greater than 5 cm in diameter it should be considered to be malignant until proven otherwise, and imaging characteristics should not be relied upon. Bilateral adenomas can be present (101). In PPNAD the adrenal glands appear normal or slightly lumpy from multiple small nodules but are not generally enlarged (102). Exogenous administration of glucocorticoids results in adrenal atrophy and very small glands may be a clue as to this entity. AIMAH is characterized by bilaterally huge (>5 cm) adrenals with a nodular pattern (103;104). Confusion can arise as the CT appearance of the adrenals in AIMAH may be similar to the appearance seen in ACTH-dependent forms of Cushing's syndrome, where adrenal enlargement is present in 70% of cases (105), but the two can usually be distinguished by the ACTH level and the degree of adrenal enlargement. Some patients with Cushing's disease can also develop a degree of adrenal autonomy which can cause biochemical confusion (106).

Identifying the source in ACTH-dependent Cushing's syndrome

This has been one of the most significant challenges in the investigation of Cushing's syndrome in the past, although advances over the last 15 years have greatly improved our diagnostic capability. Cushing's disease accounts for by far the majority of cases of ACTH-dependent Cushing's syndrome, between 85% and 90% in most series. This depends on gender, and in our series of 115 patients with ACTH-dependent Cushing's syndrome, of the 85 women, 92% had Cushing's disease; this percentage was 77% in the 30 men (107). Therefore, even before one starts investigation, the pretest probability that the patient with ACTH-dependent Cushing’s syndrome has Cushing's disease is very high, and any investigation must improve on this pretest likelihood. However, as transsphenoidal pituitary surgery is widely accepted as the primary treatment of Cushing's disease, testing should be designed to avoid inappropriate pituitary surgery in patients with ectopic ACTH production. Thus, any test should ideally be set with 100% specificity for the diagnosis of Cushing's disease.

Levels of serum cortisol and ACTH tend to be higher in the ectopic ACTH syndrome, but there is considerable overlap of values, producing poor discrimination (Fig 1) (107;108).

Invasive testing

Bilateral inferior petrosal sinus sampling

This is the "gold standard" test for distinguishing between Cushing's disease and an ectopic source of ACTH. The procedure involves placement of sampling catheters in the inferior petrosal sinuses that drain the pituitary. Blood for measurement of ACTH is obtained simultaneously from each sinus and a peripheral vein at two time points before and at 3-5 minutes and possibly also 10 minutes after the administration of ovine or human CRH (IV 1 μg/kg or 100μg respectively). A central (inferior petrosal) to peripheral plasma ACTH gradient of 2:1 or greater pre-CRH, or a gradient of 3:1 post-CRH is consistent with Cushing's disease. Results from early series show these criteria to be 100% sensitive and specific for Cushing’s disease (109;110). However, it is now clear that false negative tests and to a smaller degree false positive test results do occur (111-113). In order to minimise these it is important to ensure the patient is actively hypercortisolaemic (as above) at the time of the study (114), and that catheter position is confirmed as bilateral and any anomalous venous drainage noted by venography before sampling (115). For the diagnosis of Cushing's disease this test has a sensitivity of approximately 94% and a specificity fractionally short of 100%. There appears to be no discriminatory difference between ovine or human sequence CRH. Recent data suggest that where is CRH is unavailable desmopressin 10 μg may be used instead (116).

It should be noted that the procedure is technically difficult, and should only be performed by radiologists experienced in the technique. The most common complications are transient ear discomfort or pain, and local groin haematomas. More serious transient and permanent neurological sequelae have been reported, including brainstem infarction, although these are rare (<1%), and most have related to a particular type of catheter used (117) (118); if there are any early warning signs of such events the procedure should be immediately halted. Patients should be given heparin during sampling to prevent thrombotic events (119). CRH itself is generally tolerated well, although patients may experience brief facial flushing and a metallic taste in the mouth. One case of CRH inducing pituitary apoplexy in a patient with Cushing’s disease has been reported (120). There appears to be no advantage in trying to sample the cavernous sinus. Sampling of the internal jugular veins is a simpler procedure but is not as sensitive as BIPSS (121).

BIPSS is also useful to lateralise microadenomas within the pituitary using the inferior petrosal sinus ACTH gradient (IPSG), with a basal or post-CRH inter-sinus ratio of at least 1.4 being the criteria for lateralisation used in all large studies (110;111;122;123). In these studies the diagnostic accuracy of localisation as assessed by operative outcome varied between 59% and 83%. This is improved if venous drainage is assessed to be symmetric (124). The accuracy of lateralisation appears to be higher in children (90%), a situation where imaging is often negative (125). There is some discrepancy between studies as to whether CRH improves the predictive value of the test (126). If a reversal of lateralisation is seen pre- and post-CRH, the test cannot be relied upon (127).

Non-invasive tests

High dose dexamethasone suppression test

As with the LDDST, the high dose dexamethasone suppression test (HDDST) was originally proposed by Liddle to differentiate between cortisol-secreting adrenal tumors and Cushing's disease (69). The HDDST’s role in the differential diagnosis of ACTH-dependent Cushing’s syndrome is based on the same premise: that most pituitary corticotroph tumors retain some albeit reduced responsiveness to negative glucocorticoid feedback, whereas ectopic ACTH-secreting tumours like adrenal tumours typically do not. The test is performed according to the same protocol as the LDDST either as 2mg 6 hourly for 2 days, or as an overnight using a single dose of 8mg of dexamethasone at 23.00h. Overall, only about 80% of patients with Cushing's disease will show suppression of cortisol to less than 50% of the basal value and there are high number of false positives tests (~10-30%) seen in ectopic Cushing’s syndrome (128-130) (131-133). Shifting the criteria can only increase sensitivity with a loss of specificity, and vice-versa. Therefore, the test achieves worse discrimination than the pretest probability of Cushing's disease. In addition, a recent study has shown that suppression to HDDST can be inferred by a > 30% suppression of serum cortisol to the 2-day LDDST (134). Therefore, we do not recommend routine use of the HDDST except when bilateral inferior petrosal sinus sampling is not available, and then only as part of a combined testing strategy with the CRH test.

The CRH test

The use of the CRH (corticotrophin-releasing hormone) test for the differential diagnosis of ACTH-dependent Cushing's syndrome is based on the premise that pituitary corticotroph adenomas retain responsivity to CRH, while ectopic ACTH tumours lack CRH receptors and therefore do not respond to the agent. CRH either 1 µg/kg or 100 µg synthetic ovine (oCRH) or human sequence CRH (hCRH) is given as a bolus injection and the change in ACTH and cortisol measured. Human sequence CRH has qualitatively similar properties to oCRH, although it is shorter acting with a slightly smaller rise in plasma cortisol and ACTH in normal and obese patients, and in those with Cushing's disease (135); this may be related to the more rapid clearance of the human sequence by endogenous CRH-binding protein (136). The availability differs worldwide with oCRH predominant in North America but hCRH elsewhere.

Different centres have used differing protocols, including type of CRH and sampling time-points, and thus there is little consensus on a universal criterion for interpreting the test. In the largest published series of the use of oCRH, an increase in ACTH by at least 35% from a mean basal (-5 and -1 minutes) to a mean of 15 and 30 minutes after oCRH in 100 patients with Cushing's disease and 16 patients with the ectopic ACTH syndrome gave the test a sensitivity of 93% for diagnosing Cushing’s disease, and was 100% specific. The best cortisol criterion proved less discriminatory (137). Conversely, in the largest series of the use of hCRH in 101 patients with Cushing's disease and 14 with the ectopic ACTH syndrome, the best criterion to differentiate Cushing's disease from ectopic ACTH syndrome was a rise in cortisol of at least 14% from a mean basal (-15 and 0 minutes) to a mean of 15 and 30 minutes, giving a sensitivity of 85% with 100% specificity. The best ACTH response was a maximal rise of at least 105%, giving 70% sensitivity and 100% specificity (107). In a multicentre analysis from Italy, both hCRH or oCRH were used in 148 patients with Cushing's disease and 12 with the ectopic ACTH syndrome. A maximal 50% increase in ACTH and cortisol levels were considered as consistent with Cushing's disease, excluding all patients with the ectopic ACTH syndrome and thus giving 100% specificity. The sensitivity and specificity for the ACTH response were comparable for the two types of CRH (sensitivity: 85% vs 87% for oCRH and hCRH respectively). However, the sensitivity for the cortisol response was significantly greater with oCRH than with hCRH (sensitivity: 67% vs 50% for oCRH and hCRH respectively) (138). The authors do not report in this paper or an associated publication (19) whether time-point combinations other than the maximal were analysed for the rise in cortisol. Indeed, our data show that if the maximal rise in cortisol is used the sensitivity falls to 71% (107). These results again demonstrate that specific criteria need to be developed for each test, and cannot readily be extrapolated from other similar but non-identical agents.

In summary, the CRH test is a useful discriminator between causes of ACTH-dependent Cushing's syndrome, particularly in a combined testing strategy with the HDDST or LDDST when diagnostic accuracy greater than that of either test alone, yielding 98% to 100% sensitivity, and an 88% to 100% specificity (129;134;139) Which cut-off to use should be evaluated at individual centres, and caution should be exercised as there will undoubtedly be patients with the ectopic ACTH syndrome who respond outside these cut-offs. However, it should be remembered that responses to both CRH and high-dose dexamethasone are more frequently discordant in Cushing's disease due to a macroadenoma (140).

Testing with other peptides

Both vasopressin and desmopressin (a synthetic long-acting vasopressin analogue without the V1-mediated pressor effects) stimulate ACTH release in Cushing’s disease, probably through the corticotroph-specific V3 (or V1b) receptor. Hexarelin (a growth hormone secretagogue) stimulates ACTH release probably occurs through stimulation of vasopressin release in normal subjects (141), and by stimulation of aberrant growth hormone secretagogue receptors in corticotroph tumors (142). These peptides have been utilised in a similar manner to CRH to try and improve the differentiation of ACTH-dependent Cushing’s syndrome, but have unfortunately proved inferior (143-145). However, in centres with no availability of CRH the desmopressin test may be an alternative. A combined desmopressin and hCRH stimulation test initially looked promising (146), but a further study of this combined test showed significant overlap in the responses (147). The inferior discriminatory value of these stimulants is most likely due to the expression of both vasopressin and growth hormone secretagogue receptors by some ectopic ACTH-secreting tumours (148) (119).

Imaging

Pituitary

Imaging of the pituitary is an important part of the investigation of ACTH-dependent Cushing's syndrome to identify a possible pituitary lesion and to aid the surgeon during exploration. However, the results must be used in conjunction with the biochemical assessment as approximately 10% of normal subjects may have pituitary incidentalomas on MRI (149). Modern MRI techniques using T1-weighted spin echo and/or spoiled gradient recalled acquisition (SPGR) techniques will identify an adenoma in up to 80% of patients with Cushing’s disease (19;150;151) . On MRI, 95% of microadenomas exhibit a hypointense signal with no post-gadolinium enhancement (Fig 2); however, as the remaining 5% have an isointense signal post-gadolinium, pre-gadolinium images are essential (152). CT has a sensitivity of only approximately 40%-50% for identifying microadenomas, and is thus significantly inferior to MRI (sensitivity 50%-60%)(19;153;154), and it should therefore be reserved for patients in whom MRI is contraindicated or unavailable. Computerised tomography (CT) imaging typically shows a hypodense lesion that fails to enhance post-contrast.

Preoperative localisation to one side of the pituitary gland by MRI had been advocated as better than BIPSS with a positive predictive value of 93% (112;155). Other groups have found MRI less effective (156), (111). In addition, as noted above, we have found MRI often to be unhelpful in the paediatric age group, and BIPSS to be of significant value in these patients (125).

Ectopic tumours

Visualising an ectopic ACTH source can be a challenge, but in general patients should begin with imaging of the chest and abdomen with CT and/or MRI, bearing in mind likely sites (Table 2). The most common site of the secretory lesion is the chest, and although small cell lung carcinomas are generally easily visualised, small bronchial carcinoid tumors that can be less than 1cm in diameter often prove more difficult. Fine-cut high-resolution CT scanning with both supine and prone images can help differentiate between tumors and vascular shadows (1). MRI can identify chest lesions that are not evident on CT scanning, and characteristically show a high signal on T2-weighted and short-inversion-time inversion-recovery images (STIR) (157).

The majority of ectopic ACTH secreting tumours are of neuroendocrine origin and therefore may express somatostatin receptor subtypes. Therefore the radiolabelled somatostatin analogue ( In-pentetreotide) scintigraphy may be useful to show either functionality of identified tumours, or to try and localise radiologically unidentified tumours (158). Undoubedly this is a useful technique, but to date there are only sporadic reports that it identifies lesions not apparent using conventional imaging (159;160). However, a lesion of uncertain pathology is more likely to represent a neuroendocrine tumour, and hence an ectopic source of ACTH, if somatostatin scintigraphy is positive.

Unless the tumours are metabolically active, which is not usually the case, 18-flurodeoxyglucose positron-emission tomography (PET) does not generally offer any advantage over conventional CT or MRI (161) (162).

Investigative strategy for the diagnosis and differential diagnosis of Cushing’s syndrome

There have been a number of international consensus statements published for the diagnosis and differential diagnosis of Cushing's syndrome, the latest on the diagnosis in 2008 (97;119). It is recommended that UFC (at least two measurements), the LDDST, or late night salivary cortisol (two measurements) are used as the first line screening test. Abnormal results should be confirmed by one other of these tests. In patients with discordant results second-line tests should be used as necessary for confirmation. Once the diagnosis of Cushing’s syndrome is unequivocal, ACTH levels, the CRH test (combined with LDDST or HDDST), together with appropriate imaging, are the most useful non-invasive investigations to determine the aetiology. BIPSS is recommended in cases of ACTH-dependent Cushing’s syndrome where the clinical, biochemical, or radiological results are discordant or equivocal. However, in many centres where BIPSS is available and validated, the practice is to use this test in almost all cases of ACTH-dependent Cushing’s syndrome.

Treatment of Cushing’s syndrome

Surgical Management

Transphenoidal surgery

Transsphenoidal surgery is widely regarded as the treatment of choice for Cushing’s disease (163). The overall remission rate in various large series is in the order of 70%-75%, although higher rates of approximately 90% can be achieved with microadenomas (6;19;164;165). Of the patients achieving remission, about 25% of these will have a recurrence by 10 years (166), and this emphasises the need for long term follow-up. Prognostic markers of long term remission are patients aged over 25 years, a microadenoma detected by MRI, lack of invasion of the dura or cavernous sinus, histological confirmation of an ACTH-secreting tumor, low postoperative cortisol levels and long lasting adrenal insufficiency (167). Where an adenoma is apparent at transsphenoidal exploration, a selective adenomectomy is performed, and the surgeon may be guided by pre-operative imaging. However, where no tumour is obvious a hemi-hypophysectomy as guided by the BIPSS results is often the best course of action, hopefully achieving cure without panhypopituitarism (168). Where remission is not achieved at the first operation, a re-operation may be attempted, but appears to offer prolonged remission in only around 50% of cases, and with a high risk of hypopituitarism (169). There is limited data on endoscopic pituitary surgery for Cushing's syndrome although more centres are now offering this technique (170). It does appear that it may be useful in patients with persistent or recurrent disease (171).

The procedure is not without risks, and in the European Cushing’s disease survey group of 668 patients, the perioperative mortality was 1.9%, with other major complications occurring in 14.5% (164). The frequency of reported adverse events varies widely: diabetes insipidus (either temporary or permanent) (3%-46%); hypogonadism (14%-53%); hypothyroidism (14%-40%); cerebrospinal fluid rhinorrhoea (4.6%-27.9%); severe growth hormone deficiency (13%); bleeding (1.3%-5%); and meningitis (0-2.8%) (164;165;172;173). Transsphenoidal surgery is also a useful procedure in patients with Nelson’s syndrome to reduce tumour size, and ameliorate hyperpigmentation. (174). Thromboprophylaxis with low molecular weight heparin should be used perioperatively in all surgical procedures for Cushing's syndrome (58).

Postoperative Evaluation and Management

It is usual to give glucocorticoid cover for transsphenoidal surgery at initial daily doses of upto 400 mg hydrocortisone (4 mg dexamethasone), tapering off within 1 to 3 days. Morning (9:00 a.m .) serum cortisol measurements are then obtained for 3 days starting 24 hours after the last glucocorticoid administration, during which time the patient should be observed for development of signs of adrenal insufficiency. Postoperative hypocortisolemia (<50 nmol/L [1.8 µg/dL] at 9:00 a.m .) is probably the best indicator of the likelihood of long-term remission. However, detectable cortisol levels of less than 140 nmol/L (<5µg/dl) are also compatible with sustained remission (175;176). Higher postoperative cortisol levels are more likely to be associated with failed surgery; however, cortisol levels may sometimes gradually decline over 4-6 weeks reflecting either gradual infarction of remnant tumor or some degree of adrenal semiautonomy. Persistent cortisol levels greater than 140 nmol/L (>5µg/dL) six weeks after surgery require further investigation.

The CRH test in the early postoperative period in patients with hypocortisolaemia and apparent "cure" may provide a useful index of the risk of recurrence of Cushing’s disease, the rationale being that responsiveness may indicate residual tumor (19;177;178).

Persistent hypercortisolaemia after transsphenoidal exploration should prompt reevaluation of the diagnosis of Cushing’s disease, especially if previous diagnostic test results were indeterminate or conflicting, or if no tumor was found on pathological examination.

The treatment options for patients with persistent Cushing’s disease include repeat surgery, radiation therapy, and adrenalectomy. If immediate surgical remission is not achieved at the first exploration, early repeat transsphenoidal surgery including the endoscopic technique may be worthwhile in a significant proportion of patients, at the expense of increased likelihood of hypopituitarism (169) (171;179). Repeat sellar exploration is less likely to be helpful in patients with empty sella syndrome or very little pituitary tissue on MRI scans. Patients with cavernous sinus or dural invasion identified at the initial procedure are not candidates for repeat surgery to treat hypercortisolism and should receive radiation therapy.

Patients who are hypocortisolaemic should be started on glucocorticoid replacement. Hydrocortisone 20 mg total daily dose in three divided doses is the preferred choice. The largest dose (10mg) should be taken before getting out of bed, and the last 5mg dose should be taken no later than 6:00 p.m . because later administration of glucocorticoids may result in disordered sleep. This low dose of hydrocortisone should be used to avoid long-term suppression of the HPA axis. All patients receiving chronic glucocorticoid replacement therapy should be instructed that they are “dependent” on taking glucocorticoids as prescribed, and that failure to take or absorb the medication will lead to adrenal crisis and possibly death. They should be prescribed a 100-mg hydrocortisone (or other high dose glucocorticoid) intramuscular injection pack for emergency use. They should also obtain a medical information bracelet or necklace that identifies this requirement (Medic Alert Foundation). Education should stress the need for compliance with the daily dose of glucocorticoid; the need to double the oral dose for nausea, diarrhea, and fever; and the need for parenteral administration and medical evaluation during emesis, trauma, or severe medical stress.

The patient should be told to expect desquamation of the skin, and flu-like symptoms (malaise, joint aching, anorexia, and nausea) during the postoperative months, and that these are signs that indicate remission. Some of these symptoms have been related to high levels of circulating interleukin-6 (180). Most patients tolerate these symptoms of glucocorticoid withdrawal much better if they are forewarned and alerted to their positive nature. The glucocorticoid dose should not be increased in the absence of intercurrent illness based on these symptoms alone, but, signs of adrenal insufficiency, such as vomiting, electrolyte abnormalities, and postural hypotension should be excluded (181).

Recovery of the HPA axis can be monitored by measurement of 9:00 a.m . serum cortisol after omission of hydrocortisone replacement. Because recovery after transsphenoidal surgery rarely occurs before 3 to 6 months and is common at 1 year, initial testing at 6 to 9 months is reasonable (182). If the cortisol is undetectable on 2 consecutive days, then recovery of the axis has not occurred and glucocorticoid replacement can be restarted. If the cortisol is measurable, adequate reserve of the HPA axis can be assessed using the insulin tolerance test (183), with a peak cortisol value of greater than 500 nmol/L (>18 µg/dL), indicating adequate reserve on modern assays (184). Many centers use the cortisol response to 250 µg synthetic (1-24) ACTH as an alternative means of assessing HPA reserve (185;186), but there is some controversy as to its reliability in this situation (187;188). If it is used instead of the insulin tolerance test, a 30-minute cortisol of greater than 600 nmol/L (>22 µg/dL) is probably more reliable (184). Glucocorticoid replacement can be discontinued abruptly if the cortisol response is shown to be normal.

Where recovery of the HPA axis is only partial on dynamic testing, but the 9:00 a.m . cortisol levels are above the lower limit of the normal range (200 nmol/L [7 µg/dL]), it is reasonable to taper and stop the hydrocortisone unless symptoms of adrenal insufficiency occur. Patients need to continue to be aware of the continuing need for additional glucocorticoids at times of stress or illness and should be given a supply of oral hydrocortisone and an intramuscular injection pack. For patients with detectable but low 9:00 a.m . cortisol levels, the hydrocortisone replacement dose should be adjusted down if weight loss has occured, and a slightly lower dose may be given. Assessment of adequate replacement of hydrocortisone dosing by measuring serum cortisol at various points throughout the day, ensuring that levels are always sufficient (>50 nmol/L [>1.8 µg/dL]) before each dose is useful. This may mean that the peak levels after each dose appear to be unphysiological, but there is a tradeoff between mirroring a normal physiologic rhythm as far as possible and the inconvenience of multiple dosing. In the future modified release hydrocortisone may provide more physiological replacement (189).

Two late conundrums may arise: the questions of recurrence and permanent lack of recovery of the axis. Patients who articulate that the Cushing’s syndrome has returned are often correct, even before physical and biochemical evidence are unequivocal. Investigation is warranted in a patient with these complaints or with recurrent physical signs characteristic of hypercortisolemia. UFC can be measured initially on dexamethasone 0.5 mg/day, if not yet weaned from glucocorticoids. Measurement of late-night salivary cortisol having omitted the afternoon dose of hydrocortisone may also be useful. However, ideally assessment of a cortisol circadian rhythm can be done as an inpatient having stopped the hydrocortisone completely. If recurrent Cushing’s disease is diagnosed, the therapeutic options are the same as for persistent disease. It should be remembered when investigating recurrence that long-standing ACTH stimulation by a pituitary adenoma causing macronodular adrenal hyperplasia may subsequently involve autonomous cortisol production (190).

If the UFC result is subnormal or low, the patient should be questioned about the actual dose of glucocorticoid that has been taken. Often, patients take additional hydrocortisone, either because they discover that this decreases the symptoms of glucocorticoid withdrawal or because they have increased the dose “for stress,” often without following strict guidelines. These patients have a suppressed axis and very slow regression of Cushingoid features because of exogenous hypercortisolism. They require education and support along with reduction in the daily dose of hydrocortisone to recommended levels. The patient who has a subnormal cortisol response to ACTH 2 years after transsphenoidal surgery (in the absence of overreplacement) is likely to proceed to life-long ACTH deficiency.

Post-operatively, assessment for deficiencies of other pituitary hormones should also be sought, and the appropriate replacement regimen initiated as necessary.

Diuresis is common after transsphenoidal surgery and may result from intraoperative or glucocorticoid-induced fluid overload or may be due to diabetes insipidus. For these reasons, assessment of paired serum and urine osmolality and the serum sodium concentration is essential. It is advisable to withhold specific therapy unless the serum osmolality is greater than 295 mOsm/kg, the serum sodium is greater than 145 mmol/L, and the urine output is greater than 200 mL/hour with an inappropriately low urine osmolality. Desmopressin (DDAVP, Ferring) 1 µg given subcutaneously will provide adequate vasopressin replacement for 12 hours or more. Hyponatraemia may occur in as many as 20% of patients within 10 days of surgery. This may be due to injudicious fluid replacement or the syndrome of inappropriate antidiuretic hormone secretion (SIADH) as is frequently seen after extensive gland exploration, and fluid intake should be restricted (191). A small minority of patients proceed to (apparently) permanent diabetes insipidus, requiring long-term treatment with a vasopressin analogue.

Some glucocorticoid-induced abnormalities, including hypokalemia, hypertension, and glucose intolerance, may be normalised during the postoperative period so that preoperative treatments for these need to reassessed.

Adrenalectomy

Adrenalectomy is the treatment of choice for all cases ACTH-independent Cushing’s syndrome. This is either unilateral in the case of an adrenal adenoma or carcinoma, or bilateral in cases of bilateral hyperplasia. In adrenal adenomas cure following surgery in skilled hands approaches 100% (192), and is associated with low morbidity and mortality (193). In adrenal cancer; more aggressive surgical approaches probably account for the increase in life span reported in this disease (194;195). This approach may require multiple operations to resect primary lesions, local recurrences, and hepatic, thoracic, and, occasionally, intracranial metastases, and is usually accompanied with adjuvant mitotane as discussed below.

Laparoscopic adrenalectomy, both unilateral and bilateral, has been shown in experienced hands to be a safe procedure and in many centres has become the approach of choice for non malignant disease. Its complication rate is lower than with the open approach, and the in-patient stay is significantly reduced (196).

Bilateral adrenalectomy is also an important therapeutic option in patients with ACTH-dependent Cushing’s syndrome not cured by other techniques, particularly in young patients desiring fertility where there are concerns over radiotherapy induced hypopituitarism. However, it has the disadvantages of life long glucocorticoid and mineralocorticoid replacement therapy, and increased peri-operative morbidity and mortality. In addition, the development of Nelson’s syndrome in patients with ACTH-secreting pituitary adenomas occurs in between 8% and 38% of cases (197). The chance of developing Nelson’s syndrome appears to be greater if adrenalectomy is performed at a younger age, and if a pituitary adenoma is confirmed at previous pituitary surgery (166;197). Prophylactic pituitary radiotherapy probably reduces the risk of developing Nelson’s syndrome (198). However, it may be best to hold radiotherapy in reserve and undertake regular MRI scanning of the pituitary, especially whern imaging has originally not shown any clear tumour (199). Others have advocated unilateral adrenalectomy plus pituitary irradiation as an alternative to bilateral adrenalectomy, as it gives similar remission rates to primary transsphenoidal surgery (200), but this should be reserved for selected cases.

Patients after unilateral surgery will require glucocorticoid replacement whilst the suppressed HPA axis reawakens. This can take from between three months to two years (182).

Surgery for the ectopic ACTH syndrome

If the ectopic ACTH-secreting tumour is benign and amenable to surgical excision, such as in a lobectomy for a bronchial carcinoid tumour, the chance of cure of Cushing’s syndrome is high. However, if significant metastatic disease is present, surgery is not curative although it may still be of benefit in selected cases. In this situation bilateral adrenalectomy may be an option if medical management fails.

Radiotherapy

Pituitary

Conventional pituitary radiotherapy is delivered at a total dose of 4500 to 5000 cGy in 25 fractional doses over 35 days using a three-field (opposed lateral fields and vertex field) technique. Stereotactic conformal field planning is used to optimize the tumor dose and minimize radiation to other areas. This approach ensures that the daily dose to neural tissue does not exceed 180 cGy and avoids the complications of optic neuritis and cortical necrosis associated with larger total and fractional doses (201). Primary pituitary radiotherapy for the treatment of Cushing’s disease has been shown to produce poor long-term remission rates of around 50% (202) (166). In contrast, as a second line therapy to failed pituitary surgery better results are achieved with around 80% showing long-term remission as defined by the normalisation of the clinical state and biochemical parameters (203) (204). Remission usually occurs by two years although it can take considerably longer and patients should be reassessed at least yearly. Medical therapy is usually utilised in the interim. The major side effect is other anterior pituitary hormone deficiency in over 50%, commonly growth hormone, with a lesser incidence of hypogonadism and hypothyroidism. There is some evidence of an increased risk of cerebrovascular complications, which is of concern particularly in younger patients. The risk of optic neuropathy is low and probably less than 1% as long as low-dose fractions are used. Although meningiomas and gliomas have been reported after pituitary radiotherapy, it is not clear whether the incidence is significantly greater than the background risk of developing such tumors (205;206).

Steriotactic radiotherapy has been less well investigated but has a number of theoretical advantages including reduction in risk of cerebrovascular disease. It is not suitable for large lesions near the optic chiasm or optic nerves. Gamma knife radiosurgery is probably the most widely used of these techniques. As adjunctive therapy after failed transsphenoidal surgery it achieves biochemical remission in about 55%, although followup times have not been as long as for conventional radiotherapy (207). It can also be used as salvage therapy in difficult tumours (208). Radiosurgery of the pituitary gland using proton beams has similar efficacy as second line therapy (209). Linear accelerator radiotherapy for Cushing’s disease is less well described, but there are reports of some success in small numbers of patients (210).As with other forms of radiotherapy, new hormone deficiencies are the major side-effect.

Other Tumours

Overall, there is no significant evidence that radiotherapy improves survival in adrenocortical carcinoma, although in the literature there are sporadic reports that it may be helpful adjuvant treatment to radical surgery in selected cases and will decrease local recurrence (211-213). Local radiotherapy following surgical resection of an ectopic ACTH-secreting source may also be beneficial, particularly in non-metastatic thoracic carcinoid tumours (214;215).

Medical Management

The role of medical treatment of Cushing’s syndrome is an important one. It is the routine practice of many groups to pre-treat Cushing's syndrome patients prior to surgical treatment to reverse the hypercortisolaemia and its metabolic sequelae, and to hopefully reduce the complications of the definitive procedure. Similarly, medical treatment is desirable in patients with Cushing's disease whilst waiting for pituitary radiotherapy to take effect. In patients where surgery and/or radiotherapy has failed, medical management is often essential prior to (or long-term as an alternative to) bilateral adrenalectomy. It may not always be possible to identify the source of ACTH in certain cases of ACTH-dependent Cushing's syndrome, and therefore medical management is desirable pending re-investigation. Finally, medical therapy is helpful as a palliative modality in patients with metastatic disease causing Cushing's syndrome.

Adrenolytic Therapy

These agents are primarily used as inhibitors of steroid biosynthesis in the adrenal cortex, and thus can be utilised in all cases of hypercortisolemia regardless of cause, often with rapid improvement in the clinical features of Cushing's syndrome. The most commonly used agents are metyrapone, ketoconazole, mitotane and in certain circumstances etomidate.

Metyrapone

Metyrapone acts primarily to inhibit the enzyme 11β-hydroxylase, thus blocking the production of cortisol from 11-deoxycortisol in the adrenal gland (216). The subsequent elevation of 11-deoxycortisol can be monitored in the serum of patients treated with metyrapone. It should be noted that there may be some cross-reactivity from 11-deoxycortisol with some cortisol radioimmunoassays: this may result in an unnecessary increase in the metyrapone dose and subsequent clinical hypoadrenalism (217). It is preferable to measure the serum cortisol via liquid chromatography tandem mass spectrometry in patients treated with metyrapone (218). The fall in cortisol is rapid, with trough levels at 2 hours post-dose, and in our unit we usually administer a test dose of 750mg with hourly cortisol estimation for 4 hours (219). Maintenance therapy is 500-6000 mg/day in 3-4 divided doses daily. Metyrapone has been used to good effect to reduce the hypercortisolaemia in patients with Cushing's syndrome from adrenal tumours, the ectopic ACTH syndrome, and Cushing's disease. In the former, patients can be very sensitive to low doses of this agent, whilst in Cushing’s disease higher doses are often required. In Cushing's disease this can be due to the compensatory rise in ACTH in patients not having received pituitary radiotherapy.

The principal side-effects with metyrapone are hirsutism and acne (as predicted by the rise in adrenal androgens), dizziness and gastrointestinal upset. However, it is hypoadrenalism that remains the most important potential problem, and careful monitoring of treatment and education of the patient is required. Hypokalaemia, oedema and hypertension due to raised mineralocorticoids are infrequent (219), but can require cessation of therapy (220).

Ketoconazole

Ketoconazole is an imidazole derivative which was originally developed as an oral anti-fungal agent. It is an inhibitor of sex steroids production by its action on C17-20 lyase, and cortisol secretion by 11β-hydroxylase inhibition (221-223). It also inhibits 17-hydroxylase and 18-hydroxylase activity, amongst other enzymes (224). It has also been reported to have a direct effect on ectopic ACTH secretion from a thymic carcinoid tumor (225). Treatment for Cushing's syndrome is usually started at a dose of 200mg twice daily, with an onset of action that is slower than metyrapone. It has been used sucessfully to lower cortisol levels in patients with Cushing's syndrome of various aetiologies including adrenal carcinoma, the ectopic ACTH syndrome, and invasive ACTH-producing pituitary carcinoma, with doses required between 200-1200mg/day in up to 4 divided daily doses (226-228).

The principal side effect of ketoconazole is hepatotoxicity. Reversible elevation of hepatic serum transaminases occurs in approximately 5%-10% of patients, with the incidence of serious hepatic injury at around 1 in 15,000 patients (229). The hepatotoxicity appears to be idiosyncratic, but has been reported within 7 days of the start of treatment in a patient with Cushing's syndrome (230). Other adverse reactions of ketoconazole include skin rashes and gastrointestinal upset (231), and one must always be wary of causing adrenal insufficiency (230;232). Due to its C17-20 lyase inhibition and consequent anti-androgenic properties, ketoconazole is particularly useful in female patients where hirsutism is an issue, which may be worsened with metyrapone. Conversely, gynaecomastia and reduced libido in male patients may be unacceptable and require alternative agents. One further advantage of ketoconazole is its inhibition of cholesterol synthesis, particularly LDL cholesterol (233), and in 34 patients with Cushing's syndrome the mean total cholesterol was reduced from 6.1 to 5.0 mmol/l on ketoconazole (231).

Triazole antifungals can also be effective, such as fluconazole (234), and itraconazole.

Mitotane

Mitotane (o’p'DDD), an isomer of the insecticide DDD (belonging to the same family of chemicals as the insecticide DDT), was developed following the observation of adrenal atrophy in dogs administered DDD. It reduces cortisol and aldosterone production by blocking cholesterol side-chain cleavage and 11β-hydroxylase in the adrenal gland (235). Mitotane is used as a treatment for adrenal carcinoma and causes tumour regression and improved survival in some patients (236;237), and has a beneficial effect on endocrine hypersecretion in approximately 75% of patients (238). It is also utilised in Cushing's syndrome of non-maliganant origin, and in this regard lower doses can be utilised (up to 4 g/day versus 12 g/day), thus reducing the incidence of side effects, particularly gastrointestinal (217;239). At these lower doses the onset of the cortisol lowering effect takes longer (6-8 weeks) than with higher doses. One major problem even with lose dose mitotane is the hypercholesterolaemia (principally an increase in LDL-cholesterol), which appears to be due to the impairment of hepatic production of oxysterols, normally a brake on the enzyme HMG Co A reductase (240). However, simvastatin a HMG Co A reductase inhibitor, can reverse the hypercholesterolaemia, and it or a similar agent should be used if necessary in patients treated with mitotane (240). Other side effects of mitotane include: neurological disturbance; elevation of hepatic enzymes; hypouricaemia; gynaecomastia in men; and a prolonged bleeding time (238;241). In the long-term, measurement of blood levels can allow dose titration and reduction as appropriate. A therapeutic level of 14-20 μg/ml has been recommended. In general, its use has been limited outside of adrenal carcinoma, in which cases it has recently been shown to prolong life (237).

Etomidate

Etomidate is an imidazole-derived anaesthetic agent which was reported to have an adverse effect on adrenocortical function in 1983 (242). Compared to the other imidazole derivative ketoconazole, etomidate more potently inhibits adrenocortical 11β-hydroxylase, has a similar inhibition of 17-hydroxylase, but has less of an effect on C17-20 lyase (243). At higher concentrations it also appears to have an effect on cholesterol side-chain cleavage (244;245). Following their initial report in 1983 (246), Allolio et al have shown that low-dose intravenous non-hypnotic etomidate (2.5mg/hour) normalised cortisol levels in 5 patients with Cushing's syndrome of various aetiologies (247). Since then, there have been a number of case reports on the use of mitotane in successfully reducing hypercortisolaemia in seriously ill patients with either Cushing's disease or the ectopic ACTH syndrome (248-251). It is usually given at a dose of 2.5 - 3.0mg/hour which is adjusted based on the serum cortisol levels. Etomidate is an effective adrenolytic agent that acts rapidly, but is limited in its use by the fact it has to be given parenterally. However, in this situation it may be life-saving. The preparation available in the USA contains the vehicle propylene glycol with the potential for nephrotoxicity, as opposed to the preparation available in Europe which contains alcohol.

Other medical agents

Somatostatin receptors have been demostrated on both corticotroph adenomas, and some ectopic ACTH-secreting tumours. However, although octreotide has been helpful in reducing ACTH and cortisol levels in selected case reports of ectopic ACTH-secreting tumours there has been much more limited success in patients with Cushing's syndrome probably through downregulation of receptor sub-type 2 in these tumors by hypercortisolaemia (252). Recently, there has been renewed interest with the introduction of pasireotide (SOM230), a somatostatin analog with a broader spectrum of activity for somatostatin receptor sub-types, including type 5, which is not downregulated during hypercortisolaemia. This agent was shown in vitro to reduce human corticotroph proliferation and ACTH secretion (253), and there have now been a number of clinical trials published. In an initial phase II trial, pasireotide 600µg injected twice daily for 15 days reduced UFC levels in 76% of 29 patients and normalised levels in 17% (254). A multicentre phase III dose-randomised trial in 162 patients with either new, persistent, or recurrent Cushing's disease has presented 12 month results. At six months there was a reduction in UFC levels in 91 of 103 evaluable patients, with a median UFC reduction of 48%. Normalisation of UFC levels was achieved in 14.6% of patients on the 600µg dose twice daily, and 26% of patients on the 900µg twice daily dose. Patients who showed <50% reduction in UFC levels from baseline by month two were unlikely to show improvement by month six or 12. The most clinically relevant adverse events were hyperglycaemia (39%), with 18% developing frank diabetes mellitus, and hypocortisolaemia (8%) (255).

Pasireotide at a lower dose of 250µg three times daily has also been used in stepwise combination therapy with the dopamine agonist cabergoline (previously been demonstrated to have modest but variable efficacy as monotherapy in Cushing's disease (256;257)), and ketoconazole. pasireotide monotherapy induced normalisation of UFC levels in 5 of 17 patients (29%). The addition of cabergoline normalised UFC levels in an additional 4 patients (24%). The further addition of ketoconazole in the remaining 8 patients induced normalisation of UFC levels in 6 of these. Thus in total, remission was achieved in 88% of patients using combination therapy out to 80 days treatment (258).

Therefore pasireotide represents an exciting potential new treatment for Cushing's disease, although the frequency of hyperglycaemia is a concern.

Temozolomide is an oral alkylating prodrug that is converted in vivo to the DNA repair inhibitor dacarbazine. Traditionally this chemotherapy agent has been used in the treatment of malignant gliomas, but recent evidence suggests it is also useful in selected aggressive pituitary tumors including corticotroph pituitary carcinomas (259;260). Although, some reports suggested that response to temozolomide in pituitary tumors can be predicted by low expression of the DNA repair enzyme O6-methylguanine-DNA-methyltransferase (MGMT) probably related to MGMT gene promotor methylation (261;262), not all studies have confirmed this (263;264). However, the therapeutic response can usually be determined after three cycles of chemotherapy.

Retinoic acid has been found to inhibit ACTH-secretion and cell proliferation both in vitro in ACTH-producing tumor cell lines, and cultured human corticotroph adenomas, and in vivo in nude mice (265). However, we are not aware that the potential anti-secretory and anti-proliferative activities of this agent in Cushing’s syndrome has been investigated further.

The thiazolidinedione rosiglitazone, a PPAR-γ agonist, has been shown in supra-pharmacological doses to suppress ACTH secretion in human and murine corticotroph tumor cells. In addition, the development of murine corticotroph tumours, generated by subcutaneous injection of ACTH-secreting AtT20 cells, were prevented (266). It appears this is not specific to corticotroph adenomas, but also applies to other forms of pituitary tumor (267). However, results in human subjects with Cushing's disease have been disappointing (268-271). This may be because doses used in the animal studies were much higher than the equivalent licensed dose in humans. Its use cannot be recommended, and indeed for other reasons it has now been withdrawn from the market.

In the rare causes of Cushing’s syndrome due to bilateral macronodular adrenal hyperplasia (AIMAH) and aberrant receptor expression of GIP, β-adrenergic and LH/hCG receptors, specific receptor antagonists may prove to be useful (271a). Although octreotide has been shown to have cause a therapeutic reponse in GIP related AIMAH as mentioned above (22), others have found neither this somatostatin analog or pasireotide to be helpful in inducing a sustained response (272).

The glucocorticoid antagonist mifepristone (RU 486) is a potent antagonist of glucocorticoid and progesterone receptors (273). In humans, mifepristone blocks glucocorticoid-induced negative feedback at the hypothalamo-pituitary level, inducing a rise in ACTH, arginine-vasopressin (AVP) and hence cortisol (274). Although, it has proven effective in the treatment of hypercortisolaemia (275;276), the major drawback is the lack of biochemical markers to monitor over-treatment, and its long half-life and minimal agonist activity leaves the patient open to hypoadrenalism.

Monitoring Treatment

It is important to monitor all patients on medical therapy for Cushing’s syndrome, to assess the effectiveness of treatment, and in particular to avoid adrenal insufficiency. We use the mean of five serum cortisol measurements across the day, although others favour measurement of urinary free cortisol (UFC). A mean serum cortisol between 150 and 300 nmol/l (5.5-11 ug/dl) corresponds to a normal cortisol production rate (277), and this range should be the aim of therapy. As mentioned above a liquid chromatography tandem mass spectrography cortisol assay is preferable in patients on metyrapone.

Cushing's syndrome in specific groups

Chronic Renal Failure

Cushing’s syndrome in the setting of chronic renal failure is poorly described but may pose diagnostic difficulties. In chronic renal failure plasma levels of cortisol are generally normal but with some radioimmunoassays may be increased (278;279). ACTH levels are increased (280). Glomerular filtration rates of less than 30 mL/min result in decreased cortisol excretion and spuriously low UFC values (281). The ACTH and cortisol responses to ovine CRH may be suppressed in patient with renal failure except for those undergoing continuous ambulatory peritoneal dialysis (282). The metabolism of dexamethasone is normal in chronic renal failure, but the oral absorption can be altered in some patients There is reduced degree of suppression of cortisol by dexamethasone suggesting a prolonged half-life of cortisol. Normal suppression to the overnight 1-mg LDDST is uncommon, and the 2-day LDDST does better in this regard (278;283).

Paediatric Cushing’s Syndrome

The most common presentation of Cushing’s syndrome in children is growth retardation, whilst weight increases (39). However, the catch is that patients with virilising adrenal tumors may show growth acceleration (284). Other virilizing signs such as acne and hirsutism are seen in approximately 50% of patients regardless of etiology (39). Hypertension and striae are seen in approximately 50% of cases (285). Muscle weakness may be less common in the paediatric patient due to increased exercise (286). Psychiatric and cognitive changes may affect school performance, however, children may show “compulsive diligence” and actually do quite well academically (287). Headaches and fatigue are common (39). Cushing’s disease accounts for the between 75% and 80% of Cushing’s syndrome in older children, but before the age of 10 years ACTH-independent causes of Cushing’s syndrome are more common. Cushing’s disease has a male predominance in pre-pubertal children . Two causes of ACTH-independent Cushing’s syndrome, McCune-Albright syndrome and PPNAD, are typically diseases of childhood or young adults. Signs of virilization in the very young (<4 years) suggest adrenal carcinoma. Ectopic secretion of ACTH occurs rarely in the pediatric population and is usually due to bronchial or thymic carcinoids (288).

As mentioned previously, late night salivary cortisol measurement has particular logistical benefits in children (289) (87). Serum midnight cortisol measurements in inpatients has high sensitivity (290). UFC should be corrected for body surface area (x1.72 m ) (291). The standard 2 day LDDST adult protocol can be used in children weighing 40kg or more, otherwise the dexamethasone dose is adjusted to 30µg/kg/day (292). As in adults there is good correlation between the cortisol suppression on the LDDST and the HDDST for the differential diagnosis and thus the latter is not necessary (293). Although it can be argued that the ectopic ACTH syndrome is so rare in children that BIPSS is not necessary, they do add reasurance in those with a negative pituitary MRI which is the case in more than 50% of cases. In addition, BIPSS has arguably better accuracy in lateralisation of the pituitary tumor (288). MRI is at least as useful as CT in the evaluation of adrenal causes (294).

Transsphenoidal surgery is the treatment of choice in children with Cushing's disease, with similar rates of remission as in adults (295). Conventional radiotherapy after non-curative transsphenoidal surgery performs even better than in adults with reported remission rates as high as 100%, with remission occurring within 12 months (296). Following pituitary surgery, plus or minus radiotherapy, the incidence of growth hormone deficiency is high, but prompt diagnosis and treatment with human growth hormone ensure acceptable growth acceleration and catch-up growth, although an abnormal body composition often persists (297). Normalization of reduced bone mineral density can also be achieved (298). Adrenalectomy is first-line therapy in ACTH-independent Cushing's syndrome.

Cushing’s Syndrome in Pregnancy

Cushing’s syndrome in pregnancy is fortunately rare in pregnancy but its diagnosis is challenging because of the symptoms and signs that are common to both conditions, such as weight gain, fatigue, striae, hypertension and glucose intolerance, and the normal physiological changes in pregnancy. This topic has been comprehensively reviewed by Lindsay and colleagues (299). Total serum cortisol levels increase in pregnancy, beginning in the first trimester and peaking at 6 months, with a decrease only after delivery, probably reflecting increased induction of corticosteroid-binding globulin production by estrogen. UFC excretion is normal in the first trimester and then rises upto three-fold by term. Suppression to dexamethasone testing is blunted. The cortisol diurnal rhythm is maintained in pregnancy but with a higher nadir. Although adrenal adenomas are often the cause of Cushing's syndrome in pregnancy (40-50%), ACTH levels may not be suppressed in these patients, possibly due to placental CRH stimulation of the pituitary corticotrophs or placental ACTH secretion. The HDDST may be useful to distinguish these patients from Cushing's disease. The CRH test has also been used to identify patients with Cushing's disease, and there is no evidence of harm both in animal studies and the small number of pregnant patients studied with this drug. MRI without gadolinium enhancement is considered safe in the third trimester, and its use in combination with the non-invasive tests above should be able to resolve most diagnostic issues. BIPSS with approriate additional radiation protection for the fetus should be reserved only for cases where diagnostic uncertainty remains.

Maternal hypercortisolism is associated with a poor outcome, both maternal and fetal, thus definitive surgical treatment of adrenal or pituitary disease is recommended to achieve eucortisolemia. The second trimester is probably the safest time for operative intervention, although adverse fetal outcomes may still persist: Medical treatment carries potential risk to the fetus and should be considered only as second line therapy when the benefit outweighs the risk, and then generally only as an interim measure. Metyrapone is probably the adrenolytic agent of choice, although its association with pre-eclampsia has been reported. Ketoconazole has been utilised sucessfully in a small number of patients but is teratogenic in animals and therefore should be used with caution.

Prognosis

The life expectancy of patients with non-malignant causes of Cushing's syndrome, once a uniformly fatal illness, has improved dramatically with effective surgical and medical treatments. From a number of studies in patients with Cushing’s disease treated in the era of transsphenoidal surgery, it appear that after curative transsphenoidal surgery long-term mortality is not significantly different from that in the general population (18;168;300). However, another population based study suggested that mortality is marginally increased (3). This is perhaps not surprising given that increased cardiovascular risk markers and evidence of atherosclerotic disease persist when measured 5 years after remission of Cushing’s disease, and hence it is important to aggressively treat associated conditions such as hypertension and diabetes (301). There is also some evidence that the outcome from Cushing's disease may be worse in males (302). The outcome of paediatric Cushing’s disease is excellent if treated at centres with appropriate experience (285) (303).

Cushing's syndrome results in significant impairment in quality of life. Unfortunately, in the long term this is only partially improved with treatment. The affective and cognitive changes associated with Cushing’s syndrome are particularly slow to resolve, and may not normalize (304).

There is some evidence that deficits in bone mass may be partially reversed after treatment of hypercortisolemia (305;306). Bisphosphonate treatment may induce a more rapid improvement in bone mineral density (307), and should be considered (along with calcium and vitamin D supplements) in patients with osteoporosis.

The prognosis of the potentially malignant causes of Cushing's syndrome is more variable. Adrenal cancer associated with Cushing's syndrome has an extremely poor prognosis. Tumors that produce ectopic ACTH tend to have a poorer prognosis, compared with tumors from the same tissue that do not produce ACTH. Small cell lung cancer, islet cell tumours and thymic carcinoids (308) illustrate this phenomenon. Up to 82% of patients with small cell lung cancer and Cushing’s syndrome die within 2 weeks from the start of chemotherapy (11).