Imaging of incidentally discovered adrenal masses.

Abstract

In the last decade the increased use of cross-sectional imaging, especially computed tomography (CT) and magnetic resonance imaging (MRI), has led to an increase in the number of incidentally discovered adrenal masses. Even in patients with a known malignancy, the majority of these will be benign, nonhyperfunctioning adenomas ( 48). Rapidly evolving imaging techniques used to evaluate these masses has meant that intervention is now indicated in only a minority of patients. The approach to their investigation must include a consideration of several factors including: diagnostic costs, discomfort to the patient, risks from biopsy, consequences of false-positive results, and disease prevalence. This article reviews the imaging appearances of the most common causes of an incidentally discovered adrenal mass, and suggests a strategy for the investigation of such masses. Adrenal masses of varying pathology are found in 0.35-4.4% of patients imaged with CT for reasons other than suspected adrenal pathology ( 25; 51; 1; 2; 30), and are usually referred to as adrenal ‘incidentalomas’. Large autopsy series of more than 1000 patients report a slightly higher incidence of 1.4-5.7% ( 57; 15; 36; 26; 56; 13). It is likely that with the increased use of abdominal CT and MRI, improving spatial resolution, and the use of thinner slices, there will be a further increase in the prevalence of such incidentally discovered adrenal masses. The vast majority of truly incidental adrenal masses (i.e. those discovered in patients with no other pathology) are adrenal adenomas. The incidence of other adrenal masses such as myelolipomas, adrenal cysts, phaeochromocytomas, carcinoma and metastases varies considerably in the literature depending on the size of the series ( 35). Even in patients known to have malignant disease, adrenal adenomas are often more common than metastases ( 48). The first step in investigating an adrenal mass detected incidentally is to establish whether it is biochemically active. If shown to be biochemically active the mass is dealt with accordingly. If the mass is not hyperfunctioning a substantial differential diagnosis remains. Some of these masses, such as myelolipomas and cysts, have specific imaging features which may allow characterization of the mass on imaging alone. In patients with an adrenal mass which cannot be characterized on imaging alone, the problem is then to differentiate benign from malignant lesions. In patients with a known malignancy the differential diagnosis is essentially between adrenal metastases and an adrenal adenoma, whilst in patients without a known malignancy adrenal carcinoma must also be considered. The major hyper-secretory syndromes associated with adrenal masses are Cushing's syndrome, Conn's syndrome and phaeochromocytomas. Cushing's syndrome may be clinically obvious, and can be confirmed by grossly elevated urinary free cortisol or, with greater sensitivity, by a failure of serum cortisol at 0900 h to suppress to < 50 nmol/l in the standard low-dose dexamethasone suppression test ( 50). However, it may be noted that some, possibly the majority, of ‘nonfunctioning’ adenomas may also show a failure of suppression, although presumably in these cases the overall secretory rate remains below or within the normal range ( 65). Patients with Conn's syndrome will demonstrate hypokalaemic alkalosis, although this may be masked by a low sodium diet: further investigation will reveal a suppressed plasma renin and elevated aldosterone:renin ratio ( 69). Such tumours are usually small and have the signal characteristics of adenomas. Finally, secretory phaeochromocytomas can be identified by elevated urinary catecholamine excretion using either GCMS or HPLC with electrochemical detection: the sensitivity of this technique currently approaches 100% ( 55; 8). The majority of incidentally discovered adrenal masses are nonhyperfunctioning benign adrenal adenomas, representing 36-94% of all adrenal masses in patients without a known malignancy ( 1, 51, 45; 28; 4; 31; 62). The reported prevalence of adrenal adenomas at autopsy varies widely, but is as high as 68% ( 9; 60). Adrenal adenomas consist of cords of clear cells separated by fibrovascular tissue. The number and size of these nodules increase with age ( 16), and they occur with increased frequency in obese diabetic patients and elderly women ( 27). On CT these nodules are characteristically round in shape, less than 3 cm in diameter, with smooth borders and homogeneous internal architecture. Calcification, necrosis and haemorrhage are uncommon ( 21). The typical attenuation values range from −10 to 30 Hounsfield units. They enhance only minimally after injection of intravenous contrast medium ( Fig. 1a). On MRI the signal characteristics of nonhyperfunctioning adenomas are similar to those of the normal adrenal gland. They are typically hypointense on T1 and isointense or slightly hyperintense on T2-weighted images ( 11) ( Fig. 1b). However, atypical adenomas exist that can be considerably hyperintense, relative to liver, on T2-weighted images ( 44). Adrenal adenoma. a, CT scan following injection of intravenous contrast medium showing a 1 cm, incidentally detected left adrenal mass (arrows). b, T1-weighted image showing the normal medial limb of the left adrenal (small arrow) and a mass of intermediate signal in the lateral limb (large arrow). c, In-phase chemical shift imaging showing the resultant signal intensity. d, Out-of-phase chemical shift imaging showing marked loss of signal intensity indicating the presence of intracellular lipid. Tumours of the lung, breast, gastrointestinal tract, thyroid and kidney are the most common primaries to metastasise to the adrenals. Metastases tend to be larger than adenomas, less well-defined and of inhomogeneous density, and occasionally have a thick enhancing rim after intravenous contrast medium ( 24). On MR they are typically hypointense compared to liver on T1-weighted images and relatively hyperintense on T2-weighted images. Some adrenal metastases are atypical and either isointense or hypointense relative to liver on T2-weighted images ( 54; 34). In addition, some metastases have very long T2 relaxation times and can mimic phaeochromocytomas, although phaeochromocytomas can usually be differentiated on clinical grounds. As discussed below, CT attenuation and chemical shift MR imaging can be helpful in distinguishing between adenomas and metastases ( Fig. 2). Chemical shift imaging to evaluate the incidentally discovered adrenal mass. Adrenal metastasis from a lung cancer. a, In-phase chemical shift imaging shows a left adrenal mass of intermediate signal intensity (arrow). b, Out-of-phase imaging showing there has been no loss of signal intensity and that the presence of intracellular fat has not been demonstrated. Myelolipomas are benign neoplasms of the adrenal cortex composed of mature fat and haemopoietic tissue in varying proportions. Most are functionally inactive but there are isolated reports of endocrine abnormalities associated with myelolipomas, including Cushing's and Conn's syndromes, and disorders associated with excess androgens or oestrogens ( 72; 75; 32). Large tumours may cause pain or may manifest with a retroperitoneal haemorrhage. The diagnosis of myelolipoma is made by demonstrating the presence of fat within an adrenal mass. This can be accomplished with either CT or MRI although the presence of haemorrhage or infarction can complicate the diagnosis ( 40). The proportion of fat detected on CT is variable; in some patients most of the tumour consists of fatty tissue, in others the use of narrow slices on CT will usually confirm discrete regions of fat attenuation (− 30 to −100 HU). On MR the presence of fat is best demonstrated on T1-weighted images with and without fat suppression. The fat-containing area in a myelolipoma should be equal in signal intensity to that of subcutaneous and retroperitoneal fat at all pulse sequences ( 14) ( Fig. 3a-d). Myelolipoma. a, Longitudinal ultrasound section showing an incidentally detected large uniformly hyperechoic mass (arrows) above the right kidney (curved arrow). b, CT scan after intravenous injection of contrast medium showing the mass consists predominantly of fat (large arrow). A small amount of soft tissue is demonstrated on the medial aspect of the mass (small arrows). c, T1-weighted image corresponding to the CT showing the mass consists predominately of fat of similar signal intensity to the surrounding perirenal fat. d, STIR sequence showing suppression of the signal from the fat within the mass. Adrenal cysts occur more often in women than men and are unilateral in more than 80% of cases. Endothelial cysts account for 45% of all adrenal cysts, the remainder being epithelial, parasitic or pseudocysts ( 23; 12). Pseudocysts are thought to occur as a result of adrenal haemorrhage ( 49). Adrenal cysts have a similar appearance on CT and MR to cysts elsewhere in the body: they are of fluid density with a clearly defined margin and a thin wall on CT. On MR they are usually markedly hypointense on T1-weighted images and markedly hyperintense on T2-weighted images ( Fig. 4a,b). The presence of proteinaceous fluid, infectious debris or haemorrhage within a cyst can cause increased signal intensity on T1-weighted images, and the presence of a soft tissue component or nodularity may make distinction from neoplasms difficult ( 33). Incidentally detected right endothelial adrenal cyst. a, CT scan performed after injection of intravenous contrast medium shows the typical appearance of a cyst within the right adrenal gland (curved arrow). b, T2-weighted MRI showing the mass to be of high signal consistent with fluid contents. Approximately 90% of phaeochromocytomas originate in the adrenal medulla and 10% show no hormonal activity ( 52; 63). On CT, phaeochromocytomas are usually discrete, rounded or oval masses of similar density to liver on uncontrasted scans. Central necrosis is frequent, which occasionally can be so marked as to result in the appearance of cystic change ( 74). In sporadic cases phaeochromocytomas may be large (average 5 cm) ( 40), although the lesions are often smaller in patients with MEN or the von Hippel-Lindau syndrome, presumably because, unlike incidentally discovered masses, these tumours are usually being actively sought in patients who may have presented with other manifestations of their disease ( 21). Most tumours enhance markedly after intravenous contrast medium ( Fig. 5a). Injection of ionic contrast medium can precipitate a hypertensive crisis in some patients who have not received α-adrenergic blockade ( 53). However, this does not appear to be the case with nonionic contrast medium ( 47). Phaeochromocytoma. a, CT scan following injection of intravenous contrast medium shows a large, enhancing adrenal mass (arrow). b, T2 weighted image corresponding to a showing the mass is of high signal intensity typical of a phaeochromocytoma. On MRI most phaeochromocytomas are hypointense on T1-weighted images and markedly hyperintense on T2-weighted images ( 19; 71) ( Fig. 5b). Whilst typical, these appearances on T2-weighted images are not specific, and there is some overlap in appearance with oedematous or necrotic adrenal metastases. A recent study reported that 35% of phaeochromocytomas had atypical signal intensity on T2-weighted images and may not be of high signal on T2-weighted images ( 39). Although the use of intravenous gadolinium is rarely necessary, as with CT phaeochromocytomas enhance markedly following injection ( 71; 39). Primary adrenal cortical carcinomas are rare, highly malignant tumours which often present with abdominal pain or a palpable mass, but can occasionally present incidentally, and are usually large at the time of diagnosis. Ninety percentage produce excess steroids and about half cause symptoms related to excess hormone production, the majority producing Cushing's syndrome ( 5). They occur more commonly on the left than on the right and approximately 10% are bilateral ( 17). On CT the tumours generally exceed 6 cm and are heterogeneous where there is necrosis and calcification ( 18). The MR appearances of carcinoma are nonspecific: they are usually hypointense, relative to liver, on T1-weighted images, and hyperintense, relative to liver, on T2-weighted images. However, the multiplanar capability of MR is useful in demonstrating the invasion of the carcinoma, particularly into the inferior vena cava. Up to 24% of tumours are less than 6 cm and at CT some are homogeneous and morphologically resemble nonhyperfunctioning adenomas ( 20). Histology from percutaneous biopsy of these lesions is often unreliable, and in view of this, in many institutions all adrenal masses between 3-5 cms will also be surgically removed for diagnosis. In patients without a known primary malignancy metastases account for up to 21% of incidentally discovered adrenal masses ( 2); conversely even in patients with a known malignancy many masses are benign ( 48). Determination of the nature of an adrenal mass in a patient with known malignancy can be crucial to the decision as to whether an attempt at curative therapy of the primary neoplasm is warranted. The specificity for diagnosis of an adenoma by noninvasive imaging needs to be very high to ensure that a patient with adrenal metastases does not undergo an unnecessary attempt at curative resection of the primary tumour because of misdiagnosis of the adrenal lesion as an adenoma. The sensitivity is much less critical as the only consequences of a false negative diagnosis of an adenoma on noninvasive imaging is that a percutaneous biopsy will be necessary to establish the diagnosis. In patients with known malignant disease, adrenal masses greater than 3 cm are malignant in 90-95% of cases, while 78-87% of lesions less than 3 cm are benign ( 42; 10). However, several studies have shown that the size alone is poor at discriminating between adenomas and nonadenomas ( 38; 39; 70). 39) found, using a threshold of 1.5 cm, that the specificity for the diagnosis of adenoma was reasonably high (93%), but the sensitivity was only 16%. In the same series, using 2.5 cm as the size cut off, the specificity was 79% and the sensitivity 84%. The density of a mass as determined by measurement of the attenuation value on CT is more discriminatory than assessment of size. Showing that an adrenal mass represents an adenoma is based on confirming the presence of intracellular lipid within the mass. Fat is of low attenuation on CT, and its presence within adenomas can be demonstrated by measuring the attenuation value in Hounsfield Units. In a large study by 38), all adrenal masses with a CT attenuation value of less than 18 Hounsfield units on scans performed without intravenous contrast medium were adenomas (specificity 100%, sensitivity 85%). At attenuation values less than 10 Hounsfield units the specificity was still 100% but the sensitivity dropped to 68%: therefore, clinicians can be reassured they are dealing with a benign mass if attenuation values of 10 HU or less are obtained on precontrast scans. This supported earlier work by 39) who reported a specificity of 96% and sensitivity of 79% using a threshold of 10 Hounsfield units. However, many patients, particularly when being screened for metastatic disease, are only scanned after administration of intravenous contrast medium, and in these patients the attenuation values on unenhanced scans cannot be measured. Recent work suggests that if CT densitometry is performed on delayed images obtained between 15 min and 1 h after intravenous injection of contrast medium, a threshold of between 24 and 37 Hounsfield units is useful in characterizing an adrenal mass as an adenoma ( 38, 7; 64). Thus, CT should help resolve the aetiology in the great majority of solitary nonfunctioning adrenal lesions. A variety of MRI protocols using different pulse sequences have been advocated in an attempt to distinguish between benign and metastatic lesions. Techniques include conventional spin-echo, gadolinium-enhanced imaging, and chemical shift imaging. The latter is now the most widely used technique. Chemical shift imaging utilizes the fat content of adrenal masses for their characterization ( 44, 67; 3; 41, 37; 59). Benign, nonfunctioning adenomas generally contain large lipid-laden cells, in contrast to malignant lesions which contain little or none. Chemical shift imaging relies on the fact that protons in water molecules precess at a slightly different rate to the protons in lipid molecules in a magnetic field. As a result, water and fat protons cycle in and out-of-phase with respect to one another. By selecting an appropriate echo time (TE), one can acquire an in-phase and an out-of-phase image. The signal intensity of a pixel on an in-phase image is derived from the signal of water plus fat protons. On out-of-phase images the signal intensity is derived from the difference of the signal of water and fat protons. Therefore, adenomas lose signal intensity on out-of-phase images compared with in-phase images, whereas metastases remain unchanged ( Fig. 1 and 2). There are several ways of assessing the degree of loss of signal intensity. Quantitative analysis can be made using a variety of ratios, essentially comparing the loss of signal in the adrenal with that of liver, paraspinal muscle or spleen on in-phase and opposed phase images. Fatty infiltration of the liver (particularly in oncology patients receiving chemotherapy) and iron overload make the liver an unreliable internal standard. Fatty infiltration may also affect skeletal muscle to a lesser extent. The spleen has been shown to be the most reliable internal standard ( 41). Simple visual assessment of relative signal intensity loss, in comparison with the reference organ, is just as accurate as quantitative methods, although quantitative methods may be useful in equivocal cases ( 41, 37). When loss of signal is demonstrated in an adrenal mass on chemical shift imaging, an adrenal adenoma can be diagnosed with a high degree of certainty. However, although specificities of 100% have been reported, metastatic lesions from hepatocellular carcinomas, renal cell carcinoma and liposarcomas can contain lipid, and two cases of adrenocortical carcinomas containing microscopic amounts of fat showing areas of loss of signal intensity on chemical shift imaging have recently been reported ( 58). However, in these cases signal loss was heterogeneous and not uniform. Conversely, it is probable that some functioning adenomas may contain insufficient lipid to result in loss of signal on out-of-phase imaging ( 66), although these would presumably be identified biochemically. Nuclear scintigraphy using iodine-131-6-iodomethyl-19-norcholeserol (NP-59) has shown some potential in separating benign from malignant masses ( 40, 22). Benign adenomas take up cholesterol analogues such as NP-59, whereas metastases do not. However, haemorrhage or inflammatory masses do not take up NP-59 either, resulting in considerable overlap between benign and malignant processes. Another promising technique for differentiating benign from malignant adrenal masses is 2-[F-18]-fluoro-2-deoxy-D-glucose (FDG) PET scanning. In a study of 20 patients FDG PET scanning correctly differentiated all patients with benign nonhyperfunctioning adenomas from patients with metastases ( 6). However, this technique is not generally available, and may prove prohibitively expensive for routine use. Even with improved imaging and new techniques such as chemical shift MRI, a small percentage of adrenal masses, particularly in patients with malignancy, cannot be accurately characterized and require percutaneous biopsy for diagnosis ( 43). Minor complications include abdominal pain, haematuria, nausea and small pneumothoraces. Major complications, generally regarded as those requiring treatment, occur in 3-4% of cases and include pneumothoraces requiring intervention, and haemorrhage, with isolated reports of adrenal abscesses, pancreatitis and seeding of metastases along the needle track ( 29; 61; 73; 46). The type of complication varies with the approach used, but does not appear to be related to needle size ( 73; 46). The reported accuracy ranges from 90 to 96%. One study showed accuracy was increased with the use of larger needles ( 73). The development of chemical shift MRI and delayed contrast enhanced CT means that most patients found to have a truly incidental adrenal mass (with no evidence of underlying malignancy) can now be investigated with noninvasive techniques. A proposed algorithm for these patients is suggested ( Fig. 6). Biochemical analysis should always be undertaken first. If the patient has an underlying biochemical abnormality further imaging is not required to characterize the mass which will be surgically removed. Imaging may, however, be required to facilitate surgery by demonstrating the exact relationship of the mass to adjacent structures. If the mass is not shown to be biochemically active, in clinical practice it is size that principally determines further investigation. If the mass is more than 3cms in maximal diameter many institutions recommend surgical removal or biopsy because of the relatively high risk of malignancy, recognizing that on biopsy the distinction between adenoma and carcinoma can be extremely difficult. If the mass has specific imaging features to allow confident characterization, such as fat confirming a myelolipoma or the features of an adrenal cyst, then no further action is required. The remaining masses will be less than 3 cms in diameter and have no associated biochemical abnormality. If these are shown to have a density of < 10 HU on unenhanced CT, or < 30 HU on delayed enhanced CT, they can be dismissed as benign adenomas. If the CT criteria are not met, chemical shift MRI is appropriate. If there is loss of signal on out-of-phase chemical shift imaging compared with in-phase imaging the lesion can be dismissed as a benign adenoma. This strategy will characterize more than 90% of incidentally discovered masses, thus markedly reducing the need for percutaneous biopsy ( 43). Algorithm for radiological investigation of incidentally discovered adrenal masses. In patients without clinical or biochemical evidence of a hypersecretory adrenal syndrome, incidental adrenal masses are usually detected by CT scan. Although cysts, myelolipomas and nonfunctioning phaeochromocytomas have characteristic imaging findings, and can be readily diagnosed with high accuracy, the morphological features of many adrenal masses are nonspecific. The diagnostic dilemma most often concerns the differentiation of the very common adenoma from an incidental deposit or a small carcinoma. Recent work on non invasive techniques using attenuation values on CT and chemical shift imaging on MRI will allow us to make the diagnosis with 100% specificity. However, it must be understood that the sensitivity is not 100%, so that even when lesions lack these characteristics they may still be adenomas. These lesions will still require biopsy, while larger lesions (above 3cms) will require biopsy and very often surgical removal because of the increasing risk of malignancy with size.