Subclinical Cushing’s syndrome (CS)
Although the majority of AI are nonhypersecretory adenomas, many patients present isolated or multiple mild hypothalamic-pituitary-adrenal (HPA) abnormalities, such as elevated or high normal urinary-free cortisol excretion, impaired cortisol rhythm, partial cortisol suppression to dexamethasone administration, low plasma adrenocorticotropic hormone (ACTH) levels and/or poor responses to corticotropin-releasing hormone [14,26,36–38]. Adrenal insufficiency following surgical excision of presumed nonsecretory adrenal adenomas has been described in 18–20% of cases, suggesting the presence of mild hypercortisolism that is called subclinical CS (SCS) . This term is used to define a condition characterised by alterations of HPA function in keeping with a subtle cortisol hypersecretion in the absence of signs and/or symptoms specific of overt hypercortisolism . Rarely, cortisol secretion can be under the control of one or more aberrant hormone receptors in patients with unilateral adenomas or incidental bilateral ACTH-independent macronodular hyperplasia .
Subclinical CS is the most frequent hormonal abnormality detected in patients with AI, with a prevalence that ranges from 5% to 47% in various studies [3,14,20,23,26,29,36,40,41]. This wide variation is mainly attributed to the different work-up protocols and variable criteria used to define subclinical cortisol excess, as well as in different inclusion criteria and size of reporting series. However, it is difficult to characterise this endocrine disorder as there is a continuum from normality to autonomy, and the degree of cortisol excess may be only slightly higher than the physiological daily cortisol output [3,42]. Following the suggestion that even subclinical hormone overproduction by AI may be associated with a greater morbidity if left untreated, the threshold for treating SCS has been lowered in the last decade [1,3,43].
Because many patients with clinically nonfunctioning incidentalomas are exposed to a chronic, even if only minimal to mild, cortisol excess, it is biologically plausible to anticipate that they should suffer, at least to some extent, from the classic long-term consequences of overt CS [24,44]. Although AI patients with SCS lack many of the usual stigmata of overt CS, they may exhibit one or more of the effects of continuous autonomous cortisol secretion . Indeed, patients with SCS are more likely to have hypertension, dyslipidaemia, IGT or type 2 DM and evidence of atherosclerosis . Furthermore, they exhibit increased waist-to-hip ratio, increased indices of IR and significant changes in carotid intimal-medial thickness . These data are in agreement with the view that SCS may be associated with the clinical phenotype of the IR syndrome as the subtle autonomous cortisol secretion may cause IR in otherwise normoglycaemic and nonobese subjects . An alternative hypothesis is that AI may be a consequence rather than cause of the metabolic syndrome ; however, a causal link between SCS and IR remains the most plausible explanation . In SCS, the cortisol secretion is probably a continuum between normal and clear-cut cortisol excess and may be intermittent. Therefore, this subtle cortisol hypersecretion may be not reliably revealed by the commonly employed biochemical markers [3,38,49].
Many of the features encountered in patients with AI, such as central fat deposition, hypertension and low high-density lipoprotein (HDL) levels, are important predisposing factors for the development of cardiovascular disease putting subjects with AI at greater risk than the general population. However, studies specifically designed to investigate the possible benefits from treatment of SCS are partially conflicting, although all suggested an improvement in hypertension [14,44,50–52]. Although some authors have described an improvement of all features of the metabolic syndrome in patients with SCS following the removal of the adrenal mass [14,36,52,53], others have not [44,50,51]. These discrepancies may be because of the different criteria adopted to define SCS, insufficient sample size and differences in the duration of follow-up and timing of postoperative evaluation.
Obesity in patients with SCS. Hypercortisolism is characterised by a redistribution of adipose tissue from peripheral to central sites of the body, mainly in the truncal region and visceral depots . Central abdominal obesity determined by waist-to-hip ratio and dual-energy x-ray absorptiometry is more frequent in patients with SCS [14,28]. Surgical treatment of subclinical hypercortisolism in patients with AI is associated with a significantly higher probability of improving body weight .
Diabetes mellitus type 2 and glucose metabolism in patients with SCS. Several observations have suggested that DM type 2 in patients with SCS may be more frequent than previously appreciated [56,57]. In a controlled study, the proportion of SCS in patients with DM type 2 was estimated to be higher than controls ; however, a subsequent study failed to support the validity of screening diabetic patients without clinical features of CS . The presence of SCS in diabetic patients is associated with poor metabolic control [14,25,37,58]. However, surgical treatment of SCS in patients with AI has yielded conflicting results. In some series, adrenalectomy did not seem to improve glucose metabolism [44,51], whereas in one study it was associated with a significant improvement in fasting glucose levels in 52·5% of patients .
Hypertension in patients with SCS. Blood pressure values are higher in patients with AI compared to healthy matched individuals . Furthermore, approximately 50% of patients with AI are hypertensive . Surgical excision of AI in patients with SCS is associated with normalisation of blood pressure and/or a significantly higher probability of improving blood pressure levels .
Dyslipidaemia in patients with SCS. HDL cholesterol levels are lower, and triglyceride levels are higher, in AI patients with SCS compared to controls . However, the effects of surgical treatment on lipid profile are conflicting as in one study adrenalectomy did not show any effect , whereas in others it led to restoration of lipid levels to normal in 37·5% of patients .
Bone metabolism in patients with SCS. Overt endogenous glucocorticoid excess is a well-recognised cause of bone loss and osteoporotic fractures . Cortisol excess inhibits bone formation, increases bone resorption, impairs calcium absorption from the gut and affects the secretion of several hormones (particularly gonadotrophins and GH), cytokines and growth factors influencing bone metabolism . Subclinical hypercortisolism has been shown to be associated with increased bone resorption, bone loss and a high prevalence of vertebral fractures [45,61,62]. The degree of clinical consequences of SCS on bone metabolism varies with the extent of hormone overproduction . In a recent study, patients with AI and SCS exhibited reduced bone density, architectural deterioration of trabecular bone and an increased prevalence of vertebral fractures . In a further study, it was shown that subclinical hypercortisolism is a common and underrated finding in patients with established osteoporosis . Although no improvement in bone parameters was documented in patients with SCS after adrenalectomy [52,64], antiresorptive treatment resulted in significant increase in lumbar BMD and bone turnover markers . Further longitudinal studies of adequate statistical power are needed to estimate the risk of osteoporotic fractures and their attendant impact on outcome and quality of life.
Phaeochromocytoma, though usually histologically mostly benign, is a potential lethal disorder with an unpredictable course . Approximately 0·01–0·1% of all hypertensive patients have been shown to harbour a pheochromocytoma but the majority lack the classic clinical triad of phaeochromocytoma (headache, palpitations and diaphoresis), and the diagnosis is easily overlooked or delayed . Furthermore, hypertension is constant in only about half of the patients, paroxysmal in a third and absent in about 20% . The mean interval between initial presentation and diagnosis is estimated to be 3·5 years with a reported delay as long as 30 years in extreme cases . Imaging findings consistent with phaeochromocytoma include increased attenuation on unenhanced CT, prominent vascularity of the mass and delayed washout of intravenous contrast medium  (Fig. 1). CT can identify phaeochromocytomas above 1 cm in diameter with a 77–98% and 29–92% sensitivity and specificity, respectively . CT scans of suboptimal technology confer limited diagnostic accuracy and make biochemical evaluation mandatory. Phaeochromocytmas exhibit a high signal intensity on T2-weighted magnetic resonance imaging (MRI), and MRI T2-weighted imaging has a sensitivity and specificity in identifying phaeochromocytoma of approximately 92% and 88%, respectively .
Clinically silent phaeochromocytoma is not rare, and its prevalence in patients with AI has been estimated between 1·5% and 13% . In the larger series of AI, phaeochromocytoma was the second most prevalent form of hyperfunctioning tumour occurring in 4·2% of all masses . Silent phaeochromocytomas can carry a significant morbidity and mortality if not diagnosed early. This is highly relevant as in autopsy series phaeochromocytoma was found in 0·13% of cases, and the tumour had not been suspected in 75% of the patients while alive, although it contributed to their death in approximately 55% of cases . About half of the patients with AI that proved to harbour phaeochromocytomas were normotensive, while the others had mild-to-moderate hypertension .
The most feared diagnostic possibility for AI is ACC, which has a mean survival of approximately 18 months and a 5-year overall survival of around 16% . ACC is rare, with an incidence ranging from 0·6 to 2 cases per million population per year [43,73]. Although ACC can develop at any age, there is a bimodal age distribution, with disease peaks before the age of 5 years and in the 4th to 5th decade of life . Adrenocortical carcinomas are increasingly detected as incidentalomas during abdominal imaging . The prevalence of primary adrenal carcinoma in clinically inapparent adrenal masses correlates with the size of the mass . Adrenal cortical carcinoma accounts for 2% of tumours up to 4 cm in size, 6% of tumours between 4·1 cm and 6 cm and 25% of tumours that are larger than 6 cm; the larger the diameter of the mass, the greater the risk of it being malignant . These tumours can be either functioning or nonfunctioning, the former accounting for approximately 60%, with CS either alone or in association with virilisation is the most frequent presentation . In a series of 1004 AI, the relative rate of malignancy was 4·6% .
Imaging features on adrenal CT are used to distinguish adenomas from malignant lesions. Adrenal adenomas are usually small, well-defined homogeneous lesions with clear margins and high lipid content, whereas malignant lesions are larger, have an irregular border, vague contour, invade into surrounding structures and exhibit high signal intensity . Signal intensity lower than 10 Hounsfield units is highly indicative of a benign adrenal lesion  (Fig. 2). Benign adenomas are also characterised by rapid absolute and relative washout of intravenous contrast; a relative washout value of more than 40% has a sensitivity and specificity of 96% and 100%, respectively, in identifying adenomas [77,78]. The combination of unenhanced CT intensity and washout values can distinguish adenomas from other adrenal tumours with 98% sensitivity and 92% specificity [77,79]. The accuracy of MRI in the differentiation between benign and malignant tumours is comparable to that of CT scanning, particularly using in-and out-of-phase imaging . Normal adrenal glands have T1 and T2 signal intensity equal or slightly lower than that of the normal liver, whereas malignant masses are hypointense on T1 and hyperintense on T2-weighted images and exhibit strong enhancement after contrast injection and delayed washout . Most ACCs accumulate and retain [18F]-fluorodeoxyglucose (FDG) and thus can be visualised by FDG positron emission tomography (FDG PET) . In a recent prospective multicenter study that included 77 patients who underwent surgery, the degree of preoperative FDG PET uptake predicted with high precision the pathological findings . Using a cut-off value above 1·45 for adrenal to liver maximum standardised uptake value (maxSUV) ratio, the sensitivity and specificity to distinguish adrenocortical adenomas from ACCs were 100% and 88%, respectively . Although the imaging phenotype does not predict hormonal function, it does not necessarily predict the underlying pathology either, and therefore, surgical resection should be considered for patients who have AI with a suspicious imaging phenotype.