5.7 Cushing’s syndrome

I would like to see the day when somebody would be appointed surgeon somewhere who had no hands, for the operative part is the least part of the work

Harvey Cushing: Letter to Dr Henry Christian, 20 November 1911

Harvey Cushing described the first case of Cushing’s syndrome with a severe phenotype in 1912. Since that time, investigation and management of Cushing’s syndrome has remained a significant clinical challenge (1, 2) and patients suspected of this diagnosis warrant referral to major centres.

Endogenous Cushing’s syndrome is due the chronic, excessive, and inappropriate secretion of cortisol. When presentation is florid diagnosis is usually straightforward, but in modern practice Cushing’s syndrome is frequently and increasingly considered in mild cases in the absence of the classical signs in the context of osteoporosis, diabetes, hypertension, gynaecology, and psychiatric clinics, and achieving a diagnosis can be difficult. Appropriate management of Cushing’s syndrome is dependent on correctly identifying the cause of excess cortisol. Separating non-ACTH-dependent causes (adrenal tumours) from ACTH-dependent causes (pituitary or ectopic secretion of ACTH) is usually simple. However, many ectopic sources are occult and the differentiation of the source of ACTH secretion may require meticulous and repeated investigation to enable the appropriate surgery to be undertaken.

In most circumstances the mainstay of therapy remains surgery to either an ACTH-secreting tumour or directly to the adrenal glands, but additional treatment with cortisol-lowering drugs and tumour-directed radiotherapy is often needed.

Endogenous Cushing’s syndrome is usually sporadic and divided into ACTH-dependent, and ACTH-independent causes (Table 5.7.1). Overall, ACTH-dependent causes account for approximately 80% of cases, and of these 80% are due to corticotroph pituitary adenomas (Cushing’s disease) with an excess female predominance, and the remaining 20% due to the ectopic ACTH syndrome (2). Cushing’s disease, the ectopic ACTH syndrome, and adrenal adenomas may also be found in the context of multiple endocrine neoplasia 1 (MEN 1).

Most cases of Cushing’s disease are due to corticotroph microadenomas, a few millimetres in diameter, only being larger than 1 cm (macroadenoma) in 6% of cases (1, 3). These tumours express the proopiomelanocortin gene (POMC 176830), the peptide product of which is subsequently cleaved to ACTH. POMC-processing is usually efficient in corticotroph microadenomas, but less so in macroadenomas, which may secrete relatively large amounts of unprocessed POMC. Some pituitary macroadenomas are ‘silent corticotroph adenomas’, and may present with tumour mass effects (e.g. optic chiasm compression) alone; on follow-up, initial absence of cushingoid features may progress to overt clinical Cushing’s syndrome. Approximately 90% of tumours express the corticotropin-releasing hormone (CRH)-1 receptor, as evidenced by the release of ACTH in response to exogenously administered CRH. Tumours also express the vasopressin-3 receptor, and respond to vasopressin and desmopressin.

Tumours causing Cushing’s disease are relatively resistant to the effects of glucocorticoids, but POMC expression and ACTH secretion are reduced by higher doses of dexamethasone in 80% of cases (2, 4). This may be caused by ‘miss-expression’ of the ‘bridging protein’ Brg1 (which is important for glucocorticoid inhibitory feedback on POMC expression) found in corticotroph tumours, and may be one event determining tumourogensis (5). Corticotroph tumours also show overexpression of cyclin E, low expression of the cyclin-dependent inhibitor, p27, and a high Ki-67 expression, all indicative of a relatively high proliferative activity (4). The excess number of reproductive-aged women with Cushing’s disease, and the fact that there is a male preponderance in prepubertal cases (6) suggest a potential aetiological role for oestrogens.

Carcinoid tumours causing the ectopic ACTH syndrome, most frequently bronchial, show a molecular phenotype close to that of pituitary corticotroph tumours. In contrast, data in small cell lung cancer cells have shown that POMC is activated by transcription factors distinct from those in the pituitary, including E2F factors (7), which are able to bind the promoter when it is in an unmethylated state (8), suggesting a different pathogenesis.

In ACTH-independent macronodular hyperplasia excess cortisol secretion may be associated with either ectopically-expressed receptors or increased eutopic receptor expression (9), and activation by ligands not usually associated with adrenal steroidogenesis: gastric inhibitory peptide (food-dependent Cushing’s); vasopressin; interleukin-1; lutenizing hormone; and serotonin. Activation of receptors increasing intracellular cAMP is thought to cause hyperplasia over many years, and hence Cushing’s syndrome.

Primary pigmented nodular adrenal disease (PPNAD) causes small ACTH-secreting nodules on the adrenal, often not visualized on imaging. PPNAD can be sporadic or part of the Carney’s complex and most cases occur in late childhood or in young adults, often with a mild or cyclical presentation (10, 11). Germ line mutations of the regulatory subunit R1A of PKA (PRKAR1A) are present in approximately 45% of patients with Carney’s complex (12, 13) and as well as in sporadic PPNAD. Interestingly, these patients show a paradoxical increase in cortisol secretion in response to dexamethasone.

McCune–Albright syndrome is due to a postzygotic activating mutation in the GNAS1 gene. The resulting tissue mosaicism results in a varied phenotype, and the disease may present in the first few weeks of life. These mutations lead to constitutive steroidogenesis in the affected adrenal nodules (14). Mutations of GNAS1 have also been found in ACTH-independent macronodular hyperplasia.

The true prevalence of Cushing’s syndrome is difficult to quantify. Earlier data suggest an incidence from 0.7 to 2.4/million population per year depending on the population studied (1). More recently, biochemical Cushing’s syndrome with no clear clinical features has been shown to be common. Incidental adrenal lesions found on CT scans are now a very common clinical problem and approximately 1% of the population aged 70 or more will have evidence of low-grade hypercortisolaemia from such a lesion. In addition, Cushing’s syndrome is found in 1–5% of obese patients with type 2 diabetes, and up to 10.8% of older patients with osteoporosis and vertebral fracture (15). The difficulty here, however, is whether detection of mild Cushing’s syndrome in these populations is of clinical value as the outcomes of small and uncontrolled intervention studies are mixed. These data indicate that formal intervention trials are needed before widespread screening in these populations can be recommended, and there is a need for clinical decision-making tools to allow stratification for intervention on an individualized basis.

Glucocorticoid receptors are present in virtually all cells, reflecting the diverse actions of cortisol, and hence the symptoms and signs of hypercortisolaemia encompass all organ systems. Many of the symptoms associated with hypercortisolaemia are common and of little specificity, such as weight gain, lethargy, weakness, menstrual irregularities, loss of libido, hirsutism, acne, depression, and psychosis (Table 5.7.2). Whilst each symptom itself may be mild, the presence of a greater number of features in any given patient increases the likelihood of Cushing’s syndrome. The signs most useful in differentiating Cushing’s syndrome include the presence of proximal myopathy, and easy bruising, purplish striae, thinness, and fragility of the skin (2). The sign of proximal weakness is most easily demonstrated by asking the patient to stand from sitting position without the use hands; an initial backwards movement of the buttocks is present in early myopathy, whilst in more severe cases rising from a chair may not be possible.

Presentation differs between genders, with purple striae, muscle atrophy, osteoporosis, and kidney stones being more common in men (16). Gonadal dysfunction is common in both sexes. The adverse effects of glucocorticoids on bone metabolism are evidenced by decreased bone mineral density. Over 70% of patients with Cushing’s syndrome may present with psychiatric symptoms ranging from anxiety to frank psychosis; if present, depression is often agitated in nature, and some degree of psychiatric disturbance often persists following remission of Cushing’s syndrome (17). Impairment in short-term memory and cognition is common and can persist for at least a year following treatment. Cortisol excess predisposes to hypertension and glucose intolerance.

Classically, the ectopic ACTH syndrome due to small cell lung cancer may have a rapid onset with severe features: profound weakness, myopathy, hyperpigmentation, diabetes mellitus, and hypokalaemic alkalosis, while there is often neither weight gain nor the classical cushingoid appearance. In contrast, the clinical phenotype and biochemical features of carcinoid and other neuroendocrine tumours (of any tissue origin) may be indistinguishable from that of Cushing’s disease, causing diagnostic difficulty.

Clinical and biochemical features may commonly vary in a ‘cyclical fashion’, causing diagnostic difficulty. Signs and symptoms fluctuate with the cortisol, such as facial plethora, myopathy, mood, blood pressure, and blood glucose, and all investigations may be normal when hypercortisolaemia is absent. Great care is needed to seek for evidence of ‘cyclicity’ in the clinical history.

Most patients initially suspected of possibly having Cushing’s syndrome will not have this condition. The complete assessment of a patient known to have some form of Cushing’s syndrome is complex, expensive, and often stressful for the patient, who is usually already significantly ill emotionally, psychologically, and physically. Thus efficient screening procedures are needed to identify the minority who will need intensive and expensive investigation leading to an accurate and precise differential diagnosis (1, 17).

It is recommended that clinical judgement is used to select patients for testing, which should be considered in: (1) patients with features that are unusual for age, such as hypertension and osteoporosis; (2) those with multiple and progressive features, especially if these include the signs that most reliably distinguish Cushing’s syndrome: the presence of thin skin in the young, easy bruising, proximal myopathy, and purple striae; (3) in children with increasing weight percentile and decreased linear growth; and (4) patients with adrenocortical lesions consistent with an adenoma found on CT scans performed for other reasons, so-called adrenal ‘incidentaloma’ (15).

It is essential that a careful drug history is taken prior to any biochemical testing seeking to exclude exogenous sources of glucocorticoids that may be present in prescribed oral, rectal, inhaled, topical, or parenteral medication as well as in many ‘over the counter’ preparations, including skin creams, ‘skin-whitening’ agents, and various ‘tonics’ and herbal preparations.

The biochemical hallmark of the condition is inappropriate cortisol secretion not subject to the normal negative feedback effects of circulating glucocorticoids. The tests are based on demonstration of excessive cortisol secretion, loss of its circadian rhythm, and the abnormal feedback regulation of the hypothalamic–pituitary− adrenal axis (Fig. 5.7.1).

In florid cases of Cushing’s syndrome the diagnosis may be obvious, but biochemical confirmation is still needed. Inves-tigation of Cushing’s syndrome is a two-step process. Hyper-cortisolaemia must be confirmed and then the cause identified. Failure to follow this approach will result in inappropriate treatment and management.

Several tests are usually needed. Investigation should be performed when there is no acute concurrent illness, such as infection or heart failure, as these may cause false-positive results. The three main tests in use are: 24-h urinary free cortisol; ‘low-dose’ dexamethasone-suppression tests; and assessment of midnight plasma or late-night salivary cortisol. The best approach is to perform at least two different tests; if concordantly positive or negative, Cushing’s syndrome is either likely or unlikely, respectively (2, 15). When there are discrepancies between tests further evaluation and repeated testing is often required. Hypercortisolaemia is also found in some patients with depression, alcohol dependence, anorexia nervosa, and late pregnancy. However, in contrast to true endogenous Cushing’s syndrome, the biochemistry improves when the underlying condition has resolved.

Urinary cortisol is a direct assessment of circulating free (biologically active) cortisol. Excess circulating cortisol saturates the binding proteins (cortisol binding globulin) and is excreted in urine as free cortisol, and when collected for 24 h gives an integrated estimation of the level of hypercortisolaemia. A single measurement has low sensitivity, and three 24-h collections should be performed (2, 17). Values fourfold greater than the upper limit of normal are rare except in Cushing’s syndrome. In contrast, if values are normal on repeated occasions Cushing’s syndrome is unlikely. Specificity is a common problem with antibody-based assays, but (2, 17) high performance liquid chromatography (HPLC) and tandem mass spectrometry improves diagnostic accuracy, although substances such as digoxin and carbamazepine may produce peaks in the HPLC assay that give falsely high values (17). Moreover, if there is renal impairment with a GFR of less than 30.0 ml/min, or an incomplete collection, the urinary free cortisol may be falsely low (15, 17). Review of the collection volume and correction for creatinine concentration may be helpful in assessing whether the collection is complete. Use of urinary free cortisol is advised in the very rare situation of Cushing’s syndrome being considered during pregnancy.

Two tests are in common use. In the overnight dexamethasone-suppression test, 1 mg of dexamethasone is administered at 23.00 hours and serum cortisol measured the next day at 08.00–09.00 hours. In the 48-h dexamethasone-suppression test, dexamethasone is administered at the dose of 0.5 mg every 6 h for 2 days at 09.00, 15.00, 21.00, and 03.00 hours with measurements of serum cortisol at 09.00 hours at the start and end of the test. To exclude Cushing’s syndrome the serum cortisol value should be less than 50 nmol/l following either test (1, 2, 15, 17). The 48-h test, though more cumbersome, is more specific and with adequate regular instructions can easily be performed by outpatients. In both tests, caution needs to be exercised if there is potential malabsorption of dexamethasone or if patients are on drugs that increase hepatic clearance of dexamethasone, including carbamazepine, phenytoin, phenobarbital, or rifampicin. Patients taking oestrogen therapy, or who are pregnant, may have an increase in the cortisol binding globulin. As commercial cortisol assays measure total cortisol, this may give a false-positive result on dexamethasone-suppression testing. Oral oestrogens need to be stopped for a period of 4–6 weeks so that cortisol binding globulin may return to basal values. Even transdermal oestrogens may cause false-positive results, and tests should be repeated off transdermal oestrogens if positive results are obtained. In renal failure, suppression on dexamethasone testing is likely to exclude Cushing’s syndrome.

It is important to note that 3–8% of patients with proven Cushing’s disease show suppression of serum cortisol to less than 50 nmol/l on either test. Thus, if clinical suspicion remains high, repeated tests and other investigations are indicated.

The normal circadian rhythm of cortisol secretion is lost in patients with Cushing’s syndrome. A single sleeping midnight plasma cortisol of less than 50 nmol/l effectively excludes Cushing’s syndrome at the time of the test. This is one of the harder tests to perform as it requires hospitalization for at least 48 h, and lack of intercurrent illness, but it can be of great utility to exclude Cushing’s syndrome, especially when the patient is on drugs known to enhance metabolism of dexamethasone causing a false positive on dexamethasone testing. Values above 50 nmol/l when asleep or above 207 nmol/l when awake are found in Cushing’s syndrome, even in those who suppress on dexamethasone (18, 19). An elevated midnight plasma cortisol does not provide additional information if clinical signs are florid and there is clear lack of suppression on dexamethasone testing.

Salivary cortisol reflects free circulating cortisol and its ease of collection and stability at room temperature make it a highly suitable screening tool for outpatient assessment. The diagnostic ranges vary between reports due to the different assays and the comparison groups used to set cut-off points. The test has a sensitivity and specificity of between 95% and 98% (15, 17). As the values of salivary cortisol are an order of magnitude lower than serum cortisol, it is essential that the performance of the local assay be known and that the appropriate cut-off point is utilized. The test is of particular use in the assessment of cyclical Cushing’s syndrome, and in children. Despite these advantages, salivary cortisol is not yet used widely in the UK.

In cases of doubt the best option is to repeat the tests at a later date, or seek further opinion. The dexamethasone-suppressed CRH test, and the desmopressin test have been proposed as useful diagnostic tools but more recent data confirm that the dexamethasone-suppressed CRH test is not more accurate than the 48-h low-dose dexamethasone-suppression test (20).

Once a diagnosis of Cushing’s syndrome is established the next step is establish the cause. Investigation will vary depending upon the availability of the biochemical tests and imaging and expertise detailed below.

The first key procedure is to measure plasma ACTH. The plasma should be separated rapidly and stored at −40 °C to avoid degradation and a falsely low result. Levels consistently below 5 ng/l indicate ACTH-independent Cushing’s syndrome and attention can be turned to imaging the adrenal with CT. Levels of ACTH persistently above 15 ng/l almost always reflect ACTH-dependent pathologies and require investigation, as detailed below. The values between these two need cautious interpretation as patients with Cushing’s disease and adrenal pathologies may have intermediate values (2, 17, 21). A positive CRH test (see below) can identify an occasional patient with Cushing’s disease with low baseline ACTH plasma levels.

In established Cushing’s syndrome, when plasma ACTH has been sampled and handled carefully, and levels are persistently undetectable, the cause is of adrenal origin. The next diagnostic procedure is to proceed to imaging with CT or MRI, which will most likely show an adrenocortical adenoma or carcinoma. If imaging is negative the diagnosis may either be PPNAD, or due surreptitious hydrocortisone absorption.

Localization of the source of ACTH secretion in ACTH-dependent Cushing’s syndrome can constitute one of the most formidable challenges of clinical endocrinology. Carcinoid tumours may be clinically indistinguishable from Cushing’s disease, and are frequently difficult to identify with imaging, especially if radiological (pituitary, thoracic, pancreatic) ‘incidentalomas’ complicate interpretation. As a result, biochemical evaluation rather than imaging is used to differentiate between pituitary and nonpituitary causes. In women with ACTH-dependent Cushing’s syndrome, 9 out of 10 cases will be due to Cushing’s disease. It is against this pretest likelihood that the performance of any test needs to be judged. On occasion, despite all investigation, in some patients it may not be possible to locate the source of ACTH with confidence, and management of hypercortisolaemia may be needed without a precise diagnosis being reached.

Whilst very high levels of plasma ACTH may be seen in ectopic ACTH, the values frequently overlap those seen in Cushing’s disease. High levels of cortisol of any aetiology may overwhelm the 11β-hydroxysteroid dehydrogenase type II enzyme in the kidney, allowing cortisol to act as a mineralocorticoid; approximately 70% of patients with ectopic ACTH syndrome due to carcinoid tumours have hypokalaemia, but it is also present in approximately 10% of patients with Cushing’s disease with extremely high cortisol production (2).

The relative merits of each investigation will be discussed, but ultimately local experience of a given investigation, dependent on assays and radiological skill, will be an important determinant of the overall diagnostic success.

The high-dose dexamethasone-suppression tests (2 mg given every 6 h for 48 h and serum cortisol measure at 09.00 h at the beginning and end, or a single 8 mg dose given at 23.00 h and serum cortisol measured the next day at 09.00 h) have been in widespread use for many years. The test relies upon the relative sensitivity of pituitary corticotroph adenomas to the effects of glucocorticoids, compared to the resistance exhibited by nonpituitary tumours. Approximately 80% of patients with Cushing’s disease will demonstrate suppression of the serum cortisol to a value of less than 50% of the basal level (2). This is less than the pretest likelihood of Cushing’s disease and, thus, by itself the high-dose dexamethasone-suppression test has little diagnostic utility. Moreover, when utilizing the 48-h low-dose dexamethasone-suppression test, if there has already been the demonstration of suppression of serum cortisol by more than 30%, there is no further advantage to utilizing the high-dose dexamethasone-suppression test. Therefore, continued routine use of the high-dose dexamethasone-suppression test can no longer be recommended except when bilateral inferior petrosal sinus sampling (BIPSS) is not available. The positive predictive value for Cushing’s disease is, however, high if there is a positive response (suppression of serum cortisol <50%) and a positive response on CRH testing (see below and Fig. 5.7.2), but the negative predictive value for exclusion of Cushing’s disease when both tests are negative, is low.

CRH was identified and sequenced in 1981, and is available for clinical practice as either the ovine (oCRH) or human sequence (hCRH) which differ by seven amino acid residues. oCRH has a longer duration of action and is the form available in North America, while the experience of hCRH dominates in Europe. In practice, the value of the test is the same (22, 23). CRH is well tolerated, with side effects from systemic administration consisting of mild, short-lived facial flushing, a sensation of a metallic taste, and a transient sinus tachycardia. A single intravenous bolus of CRH (100 μg or 1 μg/kg) administered at 09.00 hours stimulates pituitary ACTH and cortisol release in healthy individuals, excessively in patients with pituitary-dependent Cushing’s syndrome, but generally not in patients with ectopic ACTH secretion or adrenal tumours. The very variable baseline cortisol and ACTH levels in patients with Cushing’s syndrome means that a response to corticotropin-releasing hormone is defined in terms of the increment rather than the peak values.

Since the vasopressin-3 receptor is expressed in pituitary and many ectopic tumours secreting ACTH, the desmopressin test is of limited utility in the differential diagnosis of ACTH-dependent Cushing’s syndrome.

If a patient has ACTH-dependent Cushing’s syndrome, with responses both on dexamethasone-suppression and CRH testing suggesting pituitary disease, and the pituitary MRI scan shows an isolated lesion of 6 mm or more, most will regard the diagnosis of Cushing’s disease to have been made. A major problem is that up to 40% of patients with proven Cushing’s disease have normal pituitary MRI scans (21). In these cases, sampling of the gradient of ACTH from the pituitary to the periphery is the most reliable means for discriminating between pituitary and nonpituitary sources of ACTH, and is strongly recommended for most cases of ACTH-dependent Cushing’s syndrome. Since the pituitary effluent drains via the cavernous sinuses to the petrosal sinuses and then jugular bulb, there is a gradient of the value of plasma ACTH compared to the simultaneous peripheral sample when there is a central source of ACTH. BIPSS is a highly skilled and invasive technique, requiring placement of catheters in both inferior petrosal sinuses. Plasma ACTH levels in peripheral blood fluctuate spontaneously by up to a factor of two, and hence a central to peripheral ratio greater than 2 is required to have confidence that ACTH secretion is pituitary and not the result of random variation from either a pituitary or ectopic source. Via a needle in a femoral vein, two catheters are passed up the inferior and superior vena cavae into the neck. One each is then placed in a jugular vein and advanced into the inferior petrosal sinus. Catheter position and venous anatomy require confirmation by venography, as nonuniform drainage is not uncommon. The diagnostic accuracy of the test is improved with the administration of CRH. A basal central: peripheral ratio of more than 2:1 or a CRH-stimulated ratio of more than 3:1 is consistent with Cushing’s disease. The combined data for many series indicate a sensitivity and a specificity of 94% (24). Where CRH is unobtainable or too costly, desmopressin offers a reasonable alternative, but few patients with ectopic ACTH secretion have been studied in this way.

False-positive results may be caused by inadequate suppression of the normal corticotrophs; the duration and amount of hypercortisolism should be assessed prior to the test. For this reason pretreatment with cortisol-lowering agents prior to BIPSS is to be strongly discouraged as this increases the likelihood of a false-positive response in a patient with ectopic disease. ACTH secretion will always be localized to the pituitary in normal individuals and hence it is crucial to establish that all patients truly have ACTH-dependent Cushing’s syndrome before undertaking this procedure. A false-negative result may be found in patients with cyclical Cushing’s disease if the procedure is undertaken when the disease is inactive, and thus it is imperative to measure serum cortisol in the 24 h prior to sampling to establish activity.

In adults, BIPSS is only 70% accurate for lateralization of the source of ACTH within the pituitary gland (2, 17), but in children it may have greater accuracy for this purpose than MRI. False negatives may also occur if there is atrophic or plexiform venous drainage of the petrosal sinuses and this possibility should be checked for by venography at the time of BIPSS.

Multidetector CT gives the best resolution of adrenal anatomy, whilst MRI and sequence manipulation can give information on the probability of malignancy. Cortisol-secreting adenomas are typically less than 4 cm in diameter and associated with atrophy of the unaffected adrenal tissue, and the contralateral adrenal gland. Malignancy is more common with increasing tumour size and radiological evidence of vascular invasion, or cosecretion of sex steroids. In ACTH-dependent Cushing’s syndrome nodules may occur and adrenal hyperplasia is not always symmetrical, causing diagnostic confusion with a unilateral primary adrenal cause if the biochemistry is not strictly assessed; in 30% of Cushing’s disease the adrenal glands appear normal, whilst in ectopic ACTH the adrenals are virtually always homogeneously enlarged (25).

Up to 40% of corticotroph adenomas causing Cushing’s disease in adults are not visible on MRI scanning (21). Those that are visible usually fail to enhance following gadolinium on T1-weighted imaging. The use of dynamic MRI, with the administration of intravenous contrast media and rapid sequence acquisition following this, does not improve the overall diagnostic rate. There is also a 10% rate of pituitary incidentalomas in the normal population (26), emphasizing the need for careful biochemical discrimination of pituitary from nonpituitary sources of ACTH. In the absence of a pituitary macroadenoma, an abnormal MRI alone is not conclusive evidence in favour of Cushing’s disease.

Small cell lung cancer may be obvious, but in most cases thoracic and abdominal imaging by fine-cut CT is needed to identify small neuroendocrine tumours, which may be extremely hard to localize, as a source of ACTH (27, 28). Other than small cell lung cancer, bronchial carcinoid tumours are the most common sources of ectopic ACTH secretion, and are usually less than 1 cm in diameter. High-resolution dynamic CT scanning is need with 1 mm cuts and studies early after intravenous contrast administration. However, small, typically enhancing carcinoid tumours may be confused with pulmonary vascular shadows, but bronchial carcinoid tumours usually have high signal intensity on T2-weighted and short-inversion-time inversion recovery on MRI. ACTH-secreting thymic carcinoid tumours are generally larger than 2 cm and readily visualized by CT. Although ectopic ACTH-secreting tumours often express somatostatin receptors and can be seen on radiolabelled octreotide scintigraphy, they are also almost always identified by CT.

To deliver high-quality treatment to patients with Cushing’s syndrome requires a team that includes specialized surgeons and physicians, radiologists, cytologists, histopathologists, and radiotherapists. The sustained hypercortisolaemia of Cushing’s syndrome, of any aetiology, suppresses ACTH secretion from healthy corticotrophs and hence hypoadrenalism will be the consequence of complete excision of any tumour causing Cushing’s syndrome, be it adrenal, pituitary, or an ectopic source of ACTH secretion, and this may be prolonged.

Management is aimed at lowering cortisol levels, removing tumour tissue, and, in the case of Cushing’s disease, causing the least harm to remaining pituitary function. Some centres use medical therapy to control hypercortisolaemia prior to surgery, and this makes intuitive sense, but there are no published data that this affects overall outcome. Hypertension and diabetes require treatment on their own merits, but both tend to improve, often dramatically, with control of hypercortisolaemia. Severe hypokalaemia secondary to Cushing’s syndrome is extremely difficult to treat unless hypercortisolaemia is corrected. If it persists, or while cortisol control is being effected, trimaterone or high-dose amiloride is helpful.

The only consistently effective drugs for controlling hypercortisolaemia are those that act on the adrenal glands to inhibit cortisol secretion: metyrapone and ketoconazole. These are not curative, and cortisol oversecretion will recur when they are discontinued. Drugs are used to regulate cortisol secretion in very specific circumstances, namely: in preparation for surgery, in patients not cured by surgery, while waiting for radiotherapy to be effective, after chemotherapy, and to correct acute severe physical or psychiatric consequences of hypercortisolaemia. Cortisol-induced psychosis usually responds rapidly to lowering of circulating cortisol levels.

A potential side effect of all drugs used to control cortisol secretion is hypoadrenalism, particularly in patients with cyclical Cushing’s syndrome, and hence all patients require close monitoring. Although urinary free cortisol is easy to use for monitoring therapy, it has major limitations in that hypoadrenalism may be difficult to establish accurately, and failure to ensure a complete 24-h collection will result in spuriously low results. Calculation of the mean of five serum cortisol measurements obtained via a cannula between 09.00 h and 21.00 h in a single day offers both the ability to identify transient hypoadrenalism and allows accurate monitoring, since a mean serum cortisol between 150 and 300 nmol/l has been demonstrated to equate to a normal cortisol production rate (29).

Metyrapone is effective in controlling hypercortisolaemia in 80% of patients with Cushing’s disease and adrenal tumours, and in 70% of cases with the ectopic ACTH syndrome. It inhibits cortisol secretion by blocking the final step in cortisol synthesis, namely conversion from 11-deoxycortisol by the cytochrome P450 enzyme 11-hydroxylase. Serum cortisol levels fall within 2 h of instigating therapy, but the effect is short lived and metyrapone requires to be taken three times daily. Treatment is initiated at 500 mg thrice daily and the dose titrated against mean serum cortisol, with dose increments being every 72 h to a maximum dose of 6 g/day. The average daily dose in patients with Cushing’s disease is approximately 2 g/day, while in the ectopic ACTH the average dose required to control cortisol secretion is 4 g/day. Hypoadrenalism is the major unwanted effect of metyrapone and can occur for several reasons: overtreatment, inability to mount a cortisol response to intercurrent infection, cyclical Cushing’s syndrome (see above), and problems with cortisol assays. Metyrapone therapy results in gross elevation of circulating levels of the cortisol precursor 11-deoxycortisol. A small amount of cross-reactivity of 11-deoxycortisol in some cortisol assays will produce artificially elevated apparent serum cortisol, potentially masking hypoadrenalism. Hirsutism and acne, if present in women patients before treatment, may worsen due to the accumulation of androgenic precursors secondary to the blockade of cortisol synthesis. Gastrointestinal upset is frequently attributed to metyrapone but is rare in the absence of hypoadrenalism. Mean serum cortisol levels through the day of between 150 and 300 nmol/l should be the aim, but cross-reactivity with 11-deoxycortisol must be excluded (29).

Ketoconazole is an orally active antimycotic but in larger doses is an inhibitor of cortisol synthesis. It is important to note that achlorhydria and antacid therapy interfere with ketoconazole absorption. Ketoconazole acts at several points in adrenal steroidogenesis to inhibit cortisol synthesis; however, its principal site of action is early in corticosteroidogenesis. In contrast to metyrapone, adrenal androgen concentrations fall with treatment. An additional desirable characteristic is that ketoconazole lowers serum cholesterol concentrations, which are characteristically raised in Cushing’s syndrome. Treatment is initiated with 200 mg three times per day and adjusted depending on serum cortisol concentrations; with between 200 and 1200 mg/day required to normalize cortisol secretion rates in patients with Cushing’s disease. Ketoconazole is of slower onset of action than metyrapone, and dose adjustments should only be made every 2 to 3 weeks, although in patients with adrenal adenomas responsiveness is more rapid and hypoadrenalism has occurred within 24 h. Ketoconazole consistently induces a reversible rise in liver transaminase and γ-glutamyltransferase levels, and rarely fulminant hepatic failure has been seen. Liver function must be monitored on initiation of treatment and closely thereafter. Hypoadrenalism can occur, but is less common than with metyrapone. Ketoconazole is teratogenic to male fetuses and is contraindicated in pregnancy. Ketoconazole and metyrapone may be given in combination, allowing doses to be used that are lower than required as monotherapy, with ketoconazole lowering androgen levels and thereby greatly increases the acceptability of metyrapone in women.

Etomidate, is an imidazole, and its principal clinical use is as an anaesthetic agent. At low, subhypnotic doses intravenous etomidate is a potent inhibitor of cortisol secretion. The use of intravenous etomidate in an intensive care situation is reported when oral adrenolytic therapy is not possible. Doses between 1.2 and 2.5 mg/h lower serum cortisol, sometimes to undetectable levels, when the patient needs to be maintained on a ‘block and replace’ regimen with the concomitant use of intravenous hydrocortisone (1–2 mg/h) (30).

High-dose o,p′DDD (ortho,para′dichlorodiphenyl dichloroethane, mitotane) has been used widely in the treatment of inoperable adrenocortical carcinoma, but when given at a lower dose is effective in controlling cortisol secretion in Cushing’s syndrome. It has a direct adrenolytic action, destroying adrenocortical cells, but also blocks cortisol synthesis by inhibiting 11-βhydroxylation and cholesterol side-chain cleavage. It is of slow onset of action, with changes in dose requiring 6 weeks to be fully effective. Use of mitotane in adrenocortical cancer is addressed in Chapter 5.4.

Low-dose treatment with mitotane for benign Cushing’s syndrome, with a starting dose of 0.5 to 1 g/day, with gradual dose titration is well tolerated, with rare gastrointestinal upset and few neurological side effects, and is used more frequently in mainland Europe. Currently, it is mainly used for Cushing’s disease only when metyrapone and ketoconazole cannot be used effectively. The major limitation of treatment is that it consistently causes hypercholesterolaemia, but if mitotane therapy is necessary, then the hypercholesterolaemia can be reversed by the use of a statin or ketoconazole.

RU 486 (mifepristone) is a potent glucocorticoid receptor antagonist that blocks cortisol action and reverses the consequences of hypercortisolaemia. A trial of its use in ectopic ACTH syndrome is in progress. It is reported to have reversed cortisol-induced psychosis in a patient with Cushing’s syndrome. Its use, however, depends on clinical assessment only, as cortisol levels remain high in the blood, and this has limited its widespread use.

Over the past 30 years many agents have been used in an attempt to inhibit the secretion of ACTH by corticotroph tumours, including sodium valproate and cyproheptadine, but to date none has been shown to consistently lower plasma ACTH. If a compound were to be developed for the treatment Cushing’s disease with the equivalent efficacy that dopamine agonists have for prolactinomas, this would be a huge step forwards.

Recently, the PPAR-γ agonist rosiglitazone has been tried in mouse models of Cushing’s disease, but data in humans is disappointing. Corticotroph tumours may also express the dopamine-2 receptor and short-term administration of cabergoline at a dose of 1–3 mg/week may reduce hypercortisolism in up to 40% of case (31), but often with escape after this and larger studies are needed. The newer multiligand somatostatin analogue, pasireotide, appears to lower cortisol levels in some patients with Cushing’s disease (32), but larger studies are awaited.

Transsphenoidal selective microadenectomy, by an experienced pituitary surgeon, is the treatment of choice for Cushing’s disease, as it offers the prospect of a dramatic, rapid, and longlasting cure without other hormonal deficiency (31) (Fig. 5.7.4a,b).

In most cases, control of tumour volume is not a priority as the majority have either microadenomas (Fig. 5.7.4c) or no visible tumour on MRI. Numerous series have reported the results and long-term follow-up following trans-sphenoidal surgery for Cushing’s disease. Taking all series in the world literature together, the initial remission rate is between 60 and 80%, but with a relapse rate of up to 20% when followed for many years, emphasizing the need for lifelong follow-up (Fig. 5.7.5) (31). It is likely that these variations reflect surgical skill as well as the controversy regarding the characterization of remission or continuing disease in the postoperative period. Overall, with careful and prolonged follow-up (10 years) the long-term remission rate is approximately 60%; series suggesting rates higher than this either have shorter follow-up or less stringent criteria for remission. Patients who are hypocortisolaemic (low 09.00 h serum cortisol) in the immediate postoperative period require glucocorticoid therapy until the hypothalamic–pituitary–adrenal axis recovers, usually 6–18 months postoperatively. While long-term remission is most likely when postoperative serum cortisol is low (<50 nmol/l), there is no threshold value that fully excludes possible recurrence (31). Care needs to be taken in the interpretation of postoperative serum cortisol in those patients who have received high-dose perioperative glucocorticoids, as these may suppress the level of cortisol in any remaining corticotroph tumour cells, with the patient appearing to be in remission, but then for the tumour cells to grow slowly and relapse appear years later. Similarly, suppression of serum cortisol on dexamethasone testing in the postoperative period is a poor indicator of long-term remission. Levels of postoperative serum cortisol of 100–200 nmol/l do not necessarily indicate failure of surgery, as some patients may remain in long-term remission. On the other hand levels above 200 nmol/l will almost always indicate failure of surgery. Prompt postoperative assessment of the hypothalamic–pituitary−adrenal axis is important, as in patients in whom hypercortisolaemia persists after an initial operation, repeat surgery, within 10 days, will allow remission in a further 50% of patients having a second operation. There is no agreement as to whether the presence or absence of a microadenoma on MRI makes remission more likely, but remission for macroadenomas is less than 15%.

Complications of pituitary surgery include cerebrospinal fluid leakage (less than 5%) or meningitis (under 2%), but are unusual in experienced hands. Hypopituitarism may occur, but successful microadenectomy will leave pituitary function intact in more than 50%. Pituitary function needs to be tested in full, pre- and postoperatively. The importance of preserving pituitary function has to be balanced against fitness for surgery and the consequences of deficiency, such as future fertility plans. It is important to note that functional deficiencies of growth hormone secondary to hypercortisolaemia may remain for 2 years after achieving remission by surgery. Thus the frail elderly patient, in whom a second operation would not be possible, might need an attempted total hypophysectomy at first operation, whereas in a fit young patient the tumour may be treated by a more limited procedure to attempt selective removal of the apparent local microadenoma, on the understanding that a second operation may be necessary if cure does not follow the first attempt, with almost inevitable hypopituitarism afterwards.

For patients with an adrenocortical adenoma causing Cushing’s syndrome the treatment of choice is a laparoscopic adrenalectomy by an experienced surgeon, as this is a safe and well-tolerated procedure. The adrenal contralateral to a cortisol-secreting adrenal tumour will be atrophic, and glucocorticoid, but not mineralocorticoid, replacement therapy may be required for months or sometimes years. In any cause of ACTH-dependent Cushing’s syndrome, total bilateral adrenalectomy induces a rapid resolution of the clinical features. Following bilateral surgery, patients require lifelong treatment with glucocorticoid and mineralocorticoid. With the low morbidity associated with laparoscopic adrenal surgery, this approach is being considered more frequently, and possibly even as primary therapy in some patients with Cushing’s disease, especially when disease is severe or because of patient preference. A major concern following bilateral adrenalectomy in patients with Cushing’s disease is the development of Nelson’s syndrome—a locally aggressive pituitary tumour that secretes high levels of ACTH, resulting in pigmentation. It remains controversial as to whether the tumour progression is a result of the lack of cortisol feedback following adrenalectomy, or whether the progression reflects corticotroph tumours that were always programmed to behave in an aggressive manner (33). If no tumour is visible on pituitary MRI at the time of adrenalectomy the likelihood of Nelson’s syndrome is much less. Monitoring is by MRI and measurement of plasma ACTH. The tumour itself may be treated with further surgery or radiotherapy. Some advocate pituitary radiotherapy at the time of adrenalectomy to reduce the risk of this syndrome (34), but others have not confirmed this (33).

Conventional external-beam radiotherapy has been available for 40 years, with large amounts of data demonstrating it to be safe and effective at controlling tumour growth and hormone secretion (31). In Cushing’s disease, its use is reserved for patients not cured by surgery, those in whom surgery is deemed inappropriate, and in the treatment of Nelson’s syndrome. While waiting for the effect of radiotherapy to happen, which appears to be quicker in children, patients will usually require continued treatment with cortisol-lowering drugs, and regular biochemical assessment. Long-term hypopituitarism is likely in most cases.

Stereotactic radiotherapy, the most widely used variety being the ‘gamma knife’, is a means of delivering a high dose of radiation to a small volume in a single session without surrounding tissues being exposed to significant radiation. The main advantages of stereotactic over conventional radiotherapy are the convenience of a single session, more rapid correction of hypersecretion, and preservation of the function of surrounding healthy pituitary tissue. Rare, larger tumours are less easily treated, and the dose to the optic chiasm limited to less than 6–8 Gy. Despite enthusiasm for the ‘gamma knife’, there appears to be a relapse rate of up to 20% following treatment (31), which does not compare favourably to conventional radiotherapy. It is likely that this poorer outcome reflects case selection. In some circumstances gamma knife radiotherapy can be extremely effective, even as primary therapy, and may be more rapid in onset and in efficacy. This depends on absolute confidence in diagnosis, and an anatomically favourable lesion, especially if not approachable by surgery (Fig. 5.7.6). Except in highly selected cases such as this, gamma knife radiosurgery is not yet recommended.

Uncontrolled and severe Cushing’s syndrome has a 5-year survival of just 50%, with death due mainly to vascular and infective complications. With modern management control of hypercortisolaemia is associated with a normalization of the standard mortality ratio (15). Patients are, however, still left with features of cardiovascular risk for years after remission, and quality of life is frequently significantly impaired (17, 35).

Cushing’s syndrome should be considered in any child with obesity in combination with short stature, as children with simple obesity tend to be tall whereas hypercortisolaemia stunts growth. The elevated circulating androgens seen in Cushing’s syndrome, particularly in adrenal adenomas, can result in apparent puberty and virilization without gonadal enlargement (pseudoprecocious puberty). Contrary to earlier reports, bone age is normal in 80% of children with Cushing’s syndrome as although androgens accelerate bone maturation, with a consequent loss of linear growth potential, hypercortisolaemia appears to delay maturation.

The specific features that may be present on examination include: inappropriate axillary or pubic hair, penile enlargement, scrotal pigmentation, and clitoral hypertrophy. Acne vulgaris can develop in children of any age. These features of hyperandrogenism are likely to be more pronounced with androgen-secreting adrenal tumours, but also occur with ACTH-dependent Cushing’s syndrome. Primary or secondary amenorrhoea may be a feature in girls. Blood pressure is often mildly or moderately elevated, but diabetes mellitus is very unusual. School performance can suffer and psychiatric and emotional symptoms may occur.

The distribution of causes in childhood is different from that encountered in adults; there is a male predominance in Cushing’s disease (6). Fetal- and neonatal-onset Cushing’s syndrome have been described, but the diagnosis remains exceptionally rare until approximately 8 years of age (36, 37). Under 2 years of age, adrenal carcinoma accounts for 80% of cases of Cushing’s syndrome, of which 80% occur in females.

Cushing’s syndrome presenting in childhood is rare but, if considered, the diagnosis should be relatively straightforward to confirm. Any child with weight gain and growth failure should be investigated but, as described above, the presenting symptoms may vary. The investigative algorithm is as described for adults, and experience shows that it is not necessary to alter the dose of dexamethasone or corticotropin-releasing hormone used in adults. Although more technically challenging in children, inferior petrosal sinus sampling is, as in adults, vital for localizing the source of ACTH secretion. (37)

The principles of treatment of hypercortisolaemia are the same in children as adults. Ketoconazole is preferable to metyrapone in children, as the former lowers, rather than increases, circulating androgen levels, but both are safe. Trans-sphenoidal surgery achieves remission in the majority of children with Cushing’s disease. Pituitary radiotherapy is reserved for surgical failures, or those whose surgery is impossible because of the small size of the pituitary fossa; however, when required it controls ACTH secretion more promptly in children than in adults with Cushing’s disease, but residual pituitary function requires close monitoring to ensure normal pubertal development and growth. Growth hormone deficiency occurs early after radiotherapy in children, but may recover in some (38).

Once hypercortisolaemia has been controlled, the management of growth and puberty is a major challenge. Glucocorticoids both inhibit growth hormone secretion and induce epiphyseal insensitivity to growth hormone action, and correction of hypercortisolaemia is a prerequisite to re-establishing linear growth. Adrenal androgen-induced pseudoprecocious puberty causes premature true gonadotropin-dependent puberty, and hence, even once adrenal androgen secretion has been controlled, bone age will continue to advance and potential for linear growth diminish. These factors can be regulated by the combined use of gonadotropin-releasing hormone analogues to inhibit gonadotropin secretion and control puberty, and growth hormone treatment to induce linear growth. With effective treatment of hypercortisolaemia and careful management of puberty and growth, children with Cushing’s syndrome will achieve a normal final height (39) (Fig. 5.7.7).

There is a clear need for multinational databases to better establish the prevalence and complications Cushing’s syndrome, especially in at risk populations, and these are being established. Formal intervention studies are needed in mild Cushing’s syndrome in the context of hypercortisolaemia in adrenal incidentaloma and osteoporosis. The outcome of treatment for Cushing’s disease remains disappointing in many patients, and further developments are needed in this area, especially novel approaches to medical therapy to lower ACTH.