ELSEVIER
EJR EUROPEAN JOURNAL OF RADIOLOGY
Differential diagnosis of incidentally detected adrenal masses revealed on routine abdominal CT
Filip Čtvrtlík ª, Miroslav Heřman a,*, 1, Vladimír Študent b,2, Vlastislava Tichác,3, Jiří Minařík d,4
a Department of Radiology, University Hospital Olomouc, I.P.Pavlova 6, 775 20 Olomouc, Czech Republic
b Department of Urology, University Hospital Olomouc, I.P.Pavlova 6, 775 20 Olomouc, Czech Republic
” Department of Pathology, University Hospital Olomouc, I.P.Pavlova 6, 775 20 Olomouc, Czech Republic
d Department of Internal medicine, University Hospital Olomouc, I.P.Pavlova 6, 775 20 Olomouc, Czech Republic
Received 5 October 2007; received in revised form 23 November 2007; accepted 27 November 2007
Abstract
Purpose: The aim of this study was to compare CT findings of adrenal incidentalomas with a definitive histological diagnosis in order to establish CT features characteristic for individual types of lesions.
Patients and methods: The retrospective study comprised of patients with adrenal lesions detected on abdominal CT. The patients with these incidental findings subsequently underwent adrenalectomy. The adrenalectomy was followed by a histological assessment of the expansion process. The study consisted of 62 adrenal expansions found in 55 patients (in seven patients bilateral lesions were found). According to the definitive histological diagnosis after adrenalectomy, the lesions were divided into the following six groups: primary adrenocortical carcinoma (n=4), metastasis (n=7), adrenocortical adenoma (n=37), pheochromocytoma (n=9), myelolipoma (n=2), and others (n=3). CT observations were categorized as follows: size, shape, margins, density, side of the expansion, homogeneous or heterogeneous density before and after contrast application, presence of central hypodensity, presence of central calcifications and fat deposits.
Statistical analysis was carried out using the x2-test, Kruskal-Wallis test and Mann-Whitney test. To estimate the differences between the subgroups, the t-test was used. For the evaluation of the mutual relations of maximum size, mean size, and volume, regression analysis (coefficient of determination R2) was used.
Results: The correlation and regression analysis suggest that there will be no statistically significant error if the maximum size measurements are used instead of the mean size or volume measurements of the lesion. Sensitivity, specificity, accuracy, positive predictive value, negative predictive value for distinguishing adenomas and non-adenomas using a cut-off diameter of 41.5 mm were 81.1%, 70.0%, 77.2%, 83.3%, 66.7%, respectively; using the non-contrast density threshold of 23 HU, they were 89.2%, 100%, 93%, 100%, 83.3%, respectively; using the post-contrast density threshold of 47.5 HU, they were 80.6%, 100%, 88.2%, 100%, 76.9%, respectively; using the increase of density threshold of 34.5 HU, they were 74.2%, 70.0%, 72.5%, 79.3%, 63.6%, respectively. A study of receiver operating characteristics (ROC) analyses resulted in the following conclusions: (a) the most accurate parameter for distinguishing adenomas from non-adenomas is the value of non-contrast density, (b) the second most accurate parameter is the post-contrast density, (c) the least suitable parameters are the size of the lesion and increase of density, (d) therefore, in practice, the value of non-contrast density parameter should be used.
Conclusion: Standard CT of the abdomen (not specifically aimed at adrenal glands) is a suitable method for distinguishing adrenal lesions which need to be operated on from those which are probably benign but need to be monitored. @ 2007 Elsevier Ireland Ltd. All rights reserved.
Keywords: Adrenal glands; Incidental findings; Computed tomography; Adenoma; Pheochromocytoma; Carcinoma; Metastasis
* Corresponding author. Tel .: +420 588443480; fax: +420 588442508. E-mail addresses: ctvrtlikf@seznam.cz (F. Čtvrtlík), miroslav.herman@fnol.cz (M. Heřman), studentv@seznam.cz
(V. Študent), tichym@tunw.upol.cz (V. Tichá), abretina@email.cz (J. Minařík). ☒
1 Tel .: +420 588443480; fax: +420 588442508.
2 Tel .: +420 588442894; fax: +420 588442514.
3 Tel .: +420 588442453; fax: +420 588223907.
4 Tel .: +420 588443362; fax: +420 588442502.
1. Introduction
Adrenal glands may be affected by several pathological structural changes, including adrenal incidentalomas which are defined as incidentally found structures in the gland dur- ing abdominal examination with the use of imaging methods [1-5]. The rising accessibility of ultrasound examination (US), computed tomography (CT) or magnetic resonance imaging (MRI) increases the likelihood of an incidentally found tumour. Imaging methods play a primary role in the diagnostics of inci- dentalomas whereas in the case of endocrine active tumours, they usually follow the clinical and laboratory examination of the patient.
The essential objective in the case of incidentaloma finding is to exclude malignancy and secretory activity of the lesion. Exclusion of malignancy with the use of imaging methods is a task for radiologists. Confirmation or exclusion of secretory activity is done using hormonal testing [6]. If secretory activity is confirmed, the most likely cause is subclinical Cushing’s syn- drome [1,2,7]. However, there is mostly no secretory activity. The most common histological finding in these cases is benign adenoma [1,2].
Differentiation of the biological character of an inciden- taloma is of more importance than the confirmation of secretory activity from the point of view of the patient’s prognosis. During clinical evaluation, it is important to confirm the lesions which require surgical intervention [5]. Thus, a precise morphological description of an incidentally found structure with a radiolo- gist’s diagnostic conclusion determine further diagnostic and especially therapeutic procedures [8-18].
The aim of this study was to compare CT findings of adrenal incidentalomas with a definitive histological diagnosis in order to establish CT features characteristic for individual types of lesions.
2. Patients and methods
2.1. Study group
The retrospective study comprised of patients with adrenal expansions that were found on abdominal CT. The patients with these incidental findings subsequently underwent adrenalec- tomy at the Department of Urology, University Hospital Olomouc between the years 2003 and 2007. The adrenalectomy was followed by a histological assessment of the expansion pro- cess. The adrenalectomies were performed using laparoscopic technique in all but 17 patients, where an open abdominal access was used. The study consisted of 62 adrenal expansions found in 55 patients (in seven patients bilateral lesions were found). The age distribution was from 22 to 77 years, with an average of 57.5 years. There were 29 men and 26 women.
2.2. Technique of the CT scan
CT examinations in 47 patients were performed at the Depart- ment of Radiology, University Hospital Olomouc, using the Hi-Speed CTi-Pro CT scanner (General Electric, Milwaukee,
USA). These examinations were carried out with a spiral tech- nique with 5- or 7-mm-thick slices, pitch 2, before and after the intravenous administration of contrast medium. The con- trast medium was applied using an automatic injector with a speed of 1.5-2 ml/s at a total amount of 80-120 ml according to the patient’s weight. In cases where a small adrenal expan- sion was noted during examination, supplementary scans with a 3-mm thickness were added. (The remaining 15 examinations were carried out in other hospitals using various devices) The slice thickness ranged from 1 to 7 mm, the techniques of con- trast medium application varied. In eight cases, only unenhanced scans were obtained.
In all the cases, CT examinations covered the whole abdomen; they were not specifically aimed at the adrenals and were indicated because of other problems.
2.3. Histological findings
According to the definitive histological diagnosis after adrenalectomy, expansions were divided into the following six groups: primary adrenocortical carcinoma (n=4), metastasis (n=7), adrenocortical adenoma (n=37), pheochromocytoma (n=9), myelolipoma (n=2), and others (n=3). The latter group contained two old haematomas and one adrenal ganglioneu- roma. The adenoma group also comprised cases of adrenal hyperplasia-due to the fact that the histological analysis can- not always definitively distinguish adenoma from hyperplasia [15,19] and because these entities have a similar clinical impact.
2.4. Observed CT characteristics
CT observations were categorised as follows.
2.4.1. Side of the expansion
Lesions were classified as left-sided, right-sided, and bilat- eral.
2.4.2. Size
To assess the size of the lesion, primary scans in the transverse plane and reconstructed images in other planes were used. Three diameters were measured according to the following procedure: the longest diameter in any plane was identified and two other diameters mutually perpendicular to the first and to each other were used for the measurement. For statistical estimation, three values were used: maximum diameter, calculated mean size (the arithmetic mean of the three diameters measured) and volume of the lesion calculated according to the formula for a triaxial ellipsoid: V=4/3Tabc, where a, b and c are the lengths of the ellipsoid semi-axes.
2.4.3. Shape
The lesions were divided according to their shape into three groups-regular oval, regular spherical, and irregular-shaped. Lesions which did not fit into the former two groups were included in the group of irregular-shaped lesions.
26
(a)
26
(b)
W 400 : L 40
W 400 : L 40
2.4.4. Margins
Margins of the lesions were assessed as smooth, lappeted or lobulated.
2.4.5. Density
Average values of density on unenhanced scans (non-contrast density) and, where possible, on enhanced scans (post-contrast density) were measured using circular region-of-interest (ROI). Cystic or necrotic areas or calcifications were excluded from the measurement; the enhancing rim was used to measure the density. The increase in density after contrast administration was calculated as a difference between non-contrast and post- contrast densities.
Other features assessed included: homogeneous or heteroge- neous density before and after contrast application, the presence of central hypodensity corresponding to necrosis which was defined as a portion inside the lesion with a non-contrast density of 0-20 HU, the presence of central hyperdensity corresponding to bleeding which was defined as a non-enhancing portion with density of 60-90 HU, and the presence of calcifications with the density above 120 HU and of deposits of fat demarcated as areas with negative densities from -20 to -150 HU within the soft-tissue mass.
2.5. Statistical analysis
Statistical analysis was carried out using the x2-test, Kruskal-Wallis test, and Mann-Whitney test. To estimate the
differences between the subgroups, the t-test was used. P-values of less than 0.05 were considered as statistically significant. For the evaluation of the mutual relations of maximum size, mean size, and volume, regression analysis (coefficient of determi- nation R2) was used. Receiver operating characteristics (ROC) curves from the acquired data were generated to assess the ability of size and unenhanced and enhanced CT attenua- tion values to distinguish between adenomas and other groups of expansions-non-adenomas (the groups of others and of myelolipomas were omitted).
For statistical estimation of the differences between the individual histological groups, the groups of others and of myelolipomas were omitted because of a small number of cases or unequivocal radiological findings of myelolipomas (Fig. 1).
3. Results
3.1.1. Gender and age
The mean age in the group of patients with primary carci- noma was 51.25 years (±6.85), in the group with metastases, it was 64.00 years (±8.74), in the group of adenomas 56.84 years (±9.06), in the group of pheochromocytomas 57.33 years (±12.88), in the group of myelolipomas 67.50 years, and in the group of others 51.67 years (±18.77). No statistically signifi- cant difference in the age and gender distribution between the histological groups of lesions was found.
| Histological group | Number of lesions | Maximum diameter (mm) | Mean size (mm) | Volume (mm3) |
|---|---|---|---|---|
| Adenoma | 37 | 33.9±15.1(11-70) | 28.7± 12.4(11-58) | 18,200 ± 24,092 (666-99,093) |
| Primary carcinoma | 4 | 109.5±3.3(105-113) | 99.9±4.5(95-106) | 491,302 ± 77,805 (414,936-599,500) |
| Metastases | 7 | 55.7 ± 34.6 (32-125) | 46.8±29.5(26-105) | 111,563 ± 20,0876 (9,152-550,000) |
| Pheochromocytoma | 9 | 71.7±33.1(24-112) | 64.2±31.8 (16-103) | 212,067 ±205,417 (1,680-550,000) |
| Myelolipoma | 2 | 68.5 (47-90) | 64.3 (44-84) | 169,807 (440,415-295,200) |
| Others | 3 | 60.3 ± 12.7 (46-70) | 50.4±13(37-63) | 70,589 ± 51,398 (24,472-126,000) |
Numbers represent mean ± S.D. with corresponding range in parentheses.
3.1.2. Side of the expansion
Twenty-nine expansions (46.8%) were found on the left side and 33 (53.2%) on the right side. There was no significant differ- ence between the number of left and right-sided lesions. Bilateral lesions occurred in seven patients, in five patients with adenoma (or hyperplasia) and in two patients with metastases.
3.1.3. Size
The maximum diameter, mean size, and volume of lesions of different histological groups are shown in Table 1.
The correlation and regression analysis proved that the parameters of maximum size, mean size, and volume of the tumour correlated highly, which was confirmed by the coeffi- cient of determination R2 that was within the range of 0.96-0.99. The correlation and regression analysis suggest that there will be
no statistically significant error if the maximum size measure- ments are used instead of the mean size or volume measurements of the lesion. Because of this, only the maximum size which is an easily detectable parameter was used for further statistical estimation.
The maximum diameter of adenomas was significantly smaller than the size of primary carcinomas (p=0.001), metastases (p=0.023), and pheochromocytomas (p=0.004) (Figs. 2-5). There was also some difference in the maximum diameter between primary carcinomas and pheochromocy- tomas, but slightly below the border of statistical significance (p=0.050). There was no statistically significant difference in the maximum diameter between primary carcinomas and metastases (p=0.058) and between metastases and pheochro- mocytomas (p=0.427).
The differences in the maximum diameter between the indi- vidual histological groups are shown in Fig. 6.
14
(a)
5
(b)
W 352 : 1.28
W 400 : 1. 40
(a)
8
(b)
W 400 : 1. 40
W 400 : 1. 40
13
(a)
13
(b)
W 400 : 1. 40
W 400 : 1. 40
14
(a)
13
(b)
W 400 : L 40
W 400 : 1.40
When adenomas and non-adenomas were divided according to a cut-off diameter of 41.5 mm, the following results were obtained: a sensitivity of 81.1%, specificity of 70.0%, accu- racy of 77.2%, positive predictive value of 83.3%, and negative predictive value of 66.7% (Table 3).
3.1.4. Shape
Spherical shape was found in 25 lesions (42.4%), oval shape in 34 (57.6%), and irregular shape in one case (2.7%) only. There was no statistically significant difference between the shape and histological finding.
3.1.5. Margins
Forty-seven expansions had smooth margins (75.8%), ten (16.1%) had lappeted margins and five (8.1%) had lobulated margins. No statistically significant difference between the type of margins and the histological types of expansions was found.
3.1.6. Density
Values of non-contrast and post-contrast densities and increase of density after contrast administration in the indi- vidual histological groups are shown in Table 2. (Groups of myelolipoma and of others were omitted from statistical analysis because of a small number of cases or unequivocal radiological findings of myelolipomas.)
Non-contrast density in the group of adenomas was sig- nificantly smaller than in the group of primary carcinomas (p=0.003), metastases (p=0.0001), and pheochromocytomas (p=0.00002). There was no significant difference in non- contrast density between primary carcinomas and metastases
Carcinoma
Metastasis
0
Adenoma
Pheochromocytoma
0 20 40 60 80 100 120 140 ☐ ☐ ☐
☐ ☐
Size (mm)
| Histological group | Number | Non-contrast (HU) | Contrast (HU) | Increase of density (HU) |
|---|---|---|---|---|
| Adenoma | 37 | 6± 12.6(-10-42) | 36.7 ± 20.1 (3-95) | 28.7± 16.2(1-82) |
| Primary carcinoma | 4 | 35.8±6.9(28-43) | 70.8 ± 11.3 (60-86) | 35± 17.8 (20-58) |
| Metastases | 7 | 35.7±6.5(28-45) | 100.4 ±35.1 (50-152) | 64.7±30.9(22-112) |
| Pheochromocytoma | 9 | 39.6±9.3(25-60) | 78.7±17.5(50-111) | 39.1±9.9(25-54) |
| Myelolipoma | 2 | -76.5 [-63-(-90)] | -53 | 10 |
| Others | 3 | 25.3 ± 6.4 (18-30) | 37 (35-39) | 13 (17-9) |
Numbers are mean ± S.D. with corresponding range in parentheses.
Carcinoma
Metastasis
Adenoma
0
Pheochromocytoma
0
*
-20
0
20
40
60
80
Unenhanced CT value (HU)
(p=0.848), primary carcinomas and pheochromocytomas (p=0.634), and between metastases and pheochromocytomas (p=0.307). The distribution of non-contrast densities is demon- strated in Fig. 7.
For the differentiation of adenomas from non-adenomas using a non-contrast density threshold of 23 HU the following were obtained: a sensitivity of 89.2%, specificity of 100%, accu- racy of 93%, positive predictive value of 100%, and negative predictive value of 83.3% (Table 3).
Post-contrast density in the group of adenomas was sig- nificantly smaller than in the groups of primary carcinomas (p=0.005), metastases (p=0.0002), and pheochromocytomas (p=0.00008). There were no statistically significant differences in contrast densities between primary carcinomas and metas-
Carcinoma
Metastasis
Adenoma
0
0
Pheochromocytoma
0
0
*
-50
0
50
100
150
200
Enhanced CT value (HU)
tases (p=0.185), primary carcinomas and pheochromocytomas (p=0.393), and between metastases and pheochromocytomas (p=0.202). The differences in post-contrast density between the individual groups of expansions are shown in Fig. 8.
For the differentiation of adenomas from non-adenomas using a post-contrast density threshold of 47.5 HU the following were obtained: a sensitivity of 80.6%, specificity of 100%, accu- racy of 88.2%, positive predictive value of 100%, and negative predictive value of 76.9% (Table 3).
The increase of density in the group of adenoma was signif- icantly smaller than that in the groups of metastases (p=0.008) and of pheochromocytomas (p=0.040). There were no statisti- cally significant differences in the increase of density between adenomas and primary carcinomas (p=0.659), primary car-
| Sensitivity (%) | Specificity (%) | Accuracy (%) | PPV (%) | NPV (%) | |
|---|---|---|---|---|---|
| Tumour size 41.5 mm | 81.1 (71.9-87.9) | 70.0 (53.1-82.6) | 77.2 (65.3-86.1) | 83.3 (73.9-90.3) | 66.7 (50.5-78.7) |
| Non-contrast density 23 HU | 89.2 (74.6-96.9) | 100.0 (87.9-100.0) | 93.0 (84.5-93.0) | 100.0 (92.7-100.0) | 83.3 (73.3-83.3) |
| Post-contrast density 47.5 HU | 80.6 (62.5-92.6) | 100.0 (87.6-100.0) | 88.2 (78.5-88.2) | 100.0 (90.1-100.0) | 76.9 (67.4-76.9) |
| Increase of density 34.5 HU | 74.2 (63.3-82.5) | 70.0 (53.2-82.9) | 72.5 (59.4-82.7) | 79.3 (67.7-88.2) | 63.6 (48.4-75.4) |
Numbers in parentheses are 95% CI.
| Histological group | Homogenous non-contrast | Homogenous post-contrast | Central hypodensity |
|---|---|---|---|
| Adenoma/hyperplasia | 32 (86.5%) | 18 (58.1%) | 1 (2.7%) |
| Primary carcinoma | 0 (0%) | 0 (0%) | 2 (50%) |
| Metastases | 2(28.6%) | 0 (0%) | 2 (28.6%) |
| Pheochromocytoma | 3 (33.3%) | 1 (11.1%) | 6 (66.7%) |
| Myelolipoma | 1 (50%) | 0 (0%) | 0 (0%) |
| Others | 1 (33.3%) | 0 (0%) | 0 (0%) |
Numbers present number of cases with percentages in parentheses.
cinomas and metastases (p=0.072), primary carcinomas and pheochromocytomas (p=0.481), and between metastases and pheochromocytomas (p=0.100).
The differences in the increase of density between the indi- vidual groups of expansions are shown in Fig. 9.
For the diagnosis of adenoma, at the increase of density threshold of 34.5 HU, the following were obtained: a sensitivity of 74.2%, specificity of 70.0%, accuracy of 72.5%, positive pre- dictive value of 79.3%, and negative predictive value of 63.6% (Table 3).
Fig. 10 shows the comparison of receiver operating char- acteristics (ROC) curves for differentiation of adenomas from non-adenomas for the size, non-contrast, post-contrast density, and increase of density. The largest area under the curve was for non-contrast density 0.964 ± 0.03, for post-contrast density it was 0.944 ±0.03, for the size of the mass 0.835 ±0.06, and for increase of density 0.730 ±0.07. A study of receiver oper- ating characteristics (ROC) analyses resulted in the following conclusions: (a) The most accurate parameter for distinguish- ing adenomas from non-adenomas is the value of non-contrast density, (b) The second most accurate parameter is the post- contrast density, (c) the least suitable parameters are the size of the tumour and increase in density, (d) therefore, in prac-
Carcinoma
Metastasis
Adenoma
0
*
Pheochromocytoma
0
-20
0
20
40
60
80
100
120
Increase of density (HU)
tice, the value of non-contrast density parameter should be used.
The frequency of lesions with homogeneous/heterogeneous appearance on non-contrast and post-contrast scans in the individual histological groups is shown in Table 4. Homoge- neous structure of adenomas was present significantly more frequently than in primary carcinomas (p=0.001), metastases (p=0.004), and pheochromocytomas (p=0.003). The frequency of homogeneous post-contrast structure in adenomas was signifi- cantly higher than in primary carcinomas (p=0.045), metastases (p=0.009), and pheochromocytomas (p=0.021).
The percentage of central hypodensities in adenomas was significantly lower in comparison with primary carcinomas (p=0.021) and pheochromocytomas (p<0.0001). There was no significant difference in the presence of central hypodensity between adenomas and metastases (p=0.058).
We only found one lesion with central hyperdensity corre- sponding to fresh bleeding which was present in an adenoma. It was not found in any of the other masses.
In the whole study group, six adrenal masses (9.7%) con- taining calcifications were found-two in adenomas, two in metastases, one in a pheochromocytoma, and one in a gan- glioneuroma. The occurrence of calcifications in the individual histological groups was not statistically significant.
There was only one case of limited fat deposits in the lesion. This was found in an adenoma.
ROC curve
1.0
0.8
sensitivity
0.6
0.4
Size
Unenhanced attenuation
0.2
Enhanced attenuation
Increase of density
0.0
0.0
0.2
0.4
0.6
0.8
1.0
1 - specificity
4. Discussion
Adrenal incidentalomas are clinically unapparent adrenal masses incidentally detected on abdominal examination using imaging methods for non-adrenal disease. When an inciden- taloma is found, it is essential to exclude both malignancy and secretory activity of the mass. Radiologists participate in the assessment of a malignancy using imaging methods. The eval- uation of secretory activity is a task for an endocrinologist. The result is of crucial importance to the urologist who based on the results of imaging methods and hormonal activity testing decides on surgery. If hormonal activity is not confirmed and if a malignant expansion is not suspected, the decisive criterion for evaluation is the size of the tumour. Recently, there has been a decline in the use of bioptic confirmation of the lesion aetiol- ogy and patients are referred for radical surgery earlier. This is also facilitated by the use of laparoscopic technique with fewer postoperative complications.
In our study, we compared the CT appearance detected on routine CT examination of the abdomen with a definitive histological diagnosis with the aim to establish CT features char- acteristic for individual kinds of expansions. A limiting factor of our study was a small sample size in some of the histological groups.
In comparison with the published results [1,2], we had a dif- ferent frequency of cases in the individual histological groups with a greater number of adenomas.
We proved that adenomas had significantly smaller size in comparison with primary carcinomas, metastases, and pheochromocytomas. This result is in accordance with the literature [2,3]. If the discriminative size of 41.5 mm for distin- guishing adenomas from non-adenomas was taken into account, the sensitivity was 81.1%, specificity 70%, accuracy 77.2%, positive predictive value 83.3%, and negative predictive value 66.7%. These results suggest that using the size as a single parameter for differentiation of adenomas from other histolog- ical groups is not suitable and other characteristics need to be considered [3]. If the sizes of primary carcinomas, metastases, and pheochromocytomas are compared, the situation is even more complicated. The sizes of tumours in these groups over- lapped, as evidenced by other authors [1-4].
In accordance with the recent literature, we have proved the fact that adenomas have significantly smaller non-contrast density in comparison with the other assessed histological groups, which allows their distinction from primary carcino- mas, metastases, and pheochromocytomas [3,11,20]. Sensitivity and specificity for the diagnosis of adenoma with a non-contrast density of 23 HU or smaller was 89.2% and 100%, respectively, with an accuracy of 93%, positive predictive value of 100%, and negative predictive value of 83.3%. Myelolipomas have similar densities as adenomas, but they are very well distinguishable from the latter (Fig. 1). On the contrary, the densities within other histological groups are not significantly different [3].
A similar situation occurred following the intravenous appli- cation of contrast medium. Corresponding to the findings of Szolar et al. [20], the adenomas had a significantly lower den- sity on the post-contrast scan in comparison with the other
groups, but the other groups were not distinguishable. When the values of non-contrast and post-contrast density between adenomas and primary carcinomas, metastases, and pheochro- mocytomas were compared, there was a more frequent overlap in the contrast values. At the determination level of the post- contrast density of 47.5 HU, the sensitivity for distinguishing adenomas and non-adenomas was 80.6%, specificity 100%, accuracy 88.2%, positive predictive value 100%, and negative predictive value 76.9%. Thus, the value of contrast density is not an applicable single parameter for differentiating adenomas from non-adenomas.
It had been described that the application of contrast medium should have a better benefit for differentiation of adenomas from other adrenal masses when delayed scans are performed and contrast medium wash-out is assessed. Using the contrast density value measured 1 min after the contrast medium applica- tion and then again after 10 or, according to some studies, even more minutes, it is possible to calculate the absolute or relative decrease in the density which shows the rate of contrast medium wash-out. Adenomas have faster wash-out rates in comparison with primary carcinomas, metastases, and pheochromocytomas [12,13,20]. Szolar et al. [20] state that 10 min after the appli- cation of contrast medium, the value of density in adenomas is significantly lower than that in primary carcinomas, metastases, and pheochromocytomas. However, the technique of delayed scans can only be performed on a CT specifically aimed at characterising an adrenal mass. In common practice, when an incidentaloma is found, the possibility to perform delayed scans after the administration of contrast medium is limited.
We also tried to find out whether other morphological cri- teria could characterise the tumour. Differences in shape and margins of the tumour were not statistically significant between the individual histological groups. This was also found by oth- ers [3,20]. Comparison of the structure of the lesions showed a significantly higher percentage of homogeneous structure in the group of adenomas when compared to primary carcinomas, metastases, and pheochromocytomas on both non-contrast and post-contrast scans. These parameters can therefore contribute to the differentiation of adenomas from non-adenomas.
Central hypodensity representing necrosis was found in a significantly smaller number of cases of adenomas compared to primary carcinomas and pheochromocytomas. There was no such difference in the case of metastases, which may be a limit- ing factor for practical use of this parameter. We only found one case with central hyperdensity corresponding to fresh bleeding which was present in an adenoma. Similarly, there was only one case of limited fat deposit in the expansion, also found in an ade- noma. The low occurrence of these features was not adequate for statistical estimation. The last parameter studied was the pres- ence of calcifications. When the individual histological groups were compared, no statistically significant difference was found in the presence of this feature.
By comparing our ROC curves of importance of individual parameters for the definitive diagnosis, we came to a similar conclusion by Korobkin et al. [12]: the most suitable method for distinguishing adenomas from non-adenomas is the value of non-contrast density, followed by post-contrast density, with the
size of the tumour being the least suitable. Unlike Korobkin et al. [12], the curve of contrast density in our study was much closer to that of non-contrast density and the difference between the areas under the curves of both parameters was not so marked. This suggests that the use of the post-contrast density values for differentiation of adenomas from non-adenomas is not of minor importance. The increase of density appeared to be the least suitable parameter.
A combination of various parameters can be useful. For instance, a combination of non-contrast density and the size of the lesion increases sensitivity and specificity notably according to Hamrahian et al. [3].
Our study was limited by the interpretation of results obtained by using different devices and slightly different techniques of CT examination, which on the other hand reflects the common daily practice as the patients are routinely examined at different radiological departments.
5. Conclusion
In conclusion, standard CT of the abdomen (not specifically aimed at adrenal glands) is a suitable method for distinguishing adrenal lesions which need to be operated on from those which are probably benign but need to be monitored.
Acknowledgement
We would like to thank Dr. Jana Zapletalová for statisti- cal consultation. This work was supported by the grant No. ND/7652-3 from the Ministry of Health of the Czech Republic.
References
[1] Mansmann G, Lau J, Balk E, Rothberg M, Miyachi Y, Bornstein SR. The clinically inapparent adrenal mass: update in diagnosis and management. Endocr Rev 2004;25(2):309-40.
[2] Mantero F, Terzolo M, Arnoldi G, et al. A survey on adrenal incidentaloma in Italy. J Clin Endocrinol Metab 2000;82(2):637-44.
[3] Hamrahian AH, Ioachimescu AG, Remer EM, et al. Clinical utility of noncontrast computed tomography attenuation value (hounsfield units) to differentiate adrenal adenomas/hyperplasias from nonadenomas: cleveland clinic experience. J Clin Endocrinol Metab 2005;90(2):871-7.
[4] Guo YK, Yang ZG, Li Y, et al. Uncommon adrenal masses: CT and MRI features with histopathologic correlation. Eur J Radiol 2007;62(3):359-70.
[5] Prager G, Heinz-Peer G, Passler C, et al. Surgical strategy in adrenal masses. Eur J Radiol 2002;41(1):70-7.
[6] Wang J. Endocrine tumor abdomen. Eur J Radiol 2007;62(3):315-6.
[7] Lockhart ME, Smith JK, Kenney PJ. Imaging of adrenal masses. Eur J Radiol 2002;41(2):95-112.
[8] Dunnick NR, Korobkin M. Imaging of adrenal incidentalomas: current status. AJR Am J Roentgenol 2002;179(5):559-68.
[9] Grumbach MM, Miller BM, Braunstein GD, et al. Management of clinical inapparent adrenal mass (incidentaloma). Ann Intern Med 2003;138(5):424-9.
[10] Korobkin M, Brodeur FJ, Francis I, Quint LE, Dunnick NR, Londy F. CT time-attenuation washout curves of adrenal adenomas and nonadenomas. AJR Am J Roentgenol 1998;170(3):747-52.
[11] Korobkin M, Giordano TJ, Brodeur FJ, et al. Adrenal adenomas: rela- tionship between histologic lipid and CT a MRI findings. Radiology 1996;200(3):743-7.
[12] Korobkin M, Brodeur FJ, Yutzy GG, et al. Differentiation of adrenal adeno- mas from nonadenomas using CT attenuation values. AJR Am J Roentgenol 1996;166(3):531-6.
[13] Caoili EM, Korobkin M, Francis IR, Cohan RH, Dunnick NR. Delayed enhanced CT of lipid-poor adrenal adenomas. AJR Am J Roentgenol 2000;175(5):1411-5.
[14] Israel GM, Korobkin M, Wang Ch, Hecht EN, Krinsk GA. Comparison of unenhanced CT and chemical shift MRI in evaluating lipid-rich adrenal adenomas. AJR Am J Roentgenol 2004;183(1):215-9.
[15] Gopan T, Remer E, Hamrahian AH. Evaluation and managing adrenal incidentalomas. Cleve Clin J Med 2006;73(6):561-8.
[16] Mayo-Smith WW, Boland GW, Noto RB, Lee MJ. State-of-the-art adrenal imaging. Radiographics 2001;21(4):995-1012.
[17] Dunnick NR. Imaging of adrenal incidentalomas: current states. AJR Am J Roentgenol 2002;179(3):559-68.
[18] Kurtaran A, Traub T, Shapiro B. Scintigraphic imaging of the adrenal glands. Eur J Radiol 2002;41(2):123-30.
[19] Lloyd WR, Douglas RB, Yuong WF. Atlas of nontumor pathology endocrine diseases. Washington: AFIP; 2002. pp. 207-18.
[20] Szolar DH, Korobkin M, Reittner P, et al. Adrenocortical carcinomas and adrenal pheochromocytomas: mass and enhancement loss evaluation at delayed contrast-enhanced CT. Radiology 2005;234(2):479-85.