KIDNEYS, URETERS, BLADDER, RETROPERITONEUM
Check for updates
Differentiation between heterogeneous adrenal adenoma and non-adenoma adrenal lesion with CT and MRI
Justine Lanoix1D . Manel Djelouah1 . Lea Chocardelle1 . Sophie Deguelte2 . Brigitte Delemer3 . Anthony Dohan4,5 . Philippe Soyer4,5 . Maxime Barat4,5 . Christine Hoeffel1,6
Received: 22 October 2021 / Revised: 2 January 2022 / Accepted: 5 January 2022 / Published online: 17 January 2022 @ The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022
Abstract
Purpose To assess whether heterogeneous adrenal adenomas can be distinguished from heterogeneous non-adenomas with Computed Tomography (CT) and/or Magnetic Resonance Imaging (MRI).
Method From 2009 to 2019, 980 consecutive adrenalectomies were retrospectively identified. Patients without adequate CT/MRI, with homogeneous and/or < 1 cm lesions were excluded. Differences between adenomas and non-adenomas were analyzed using Chi-square, Student t or Fischer tests, and interobserver agreement using weighted kappa test or intraclass correlation coefficient. Independent variables associated with adenomas were searched for using multivariable analysis. Area under the receiver operating characteristic curve (AUC) of the final model and its diagnostic performances were calculated. Results Final population comprised 183 patients (106 women, 77 men, mean age 53.2+14.4 years) with 124 non-adenomas and 59 heterogeneous adenomas. Macroscopic or microscopic fat on CT/MRI allowed diagnosis of adenoma with 98% specificity and 63% sensitivity. Interobserver agreement was almost perfect for macroscopic fat (k=0.82; 95% CI 0.66; 0.94) and substantial for microscopic fat (k=0.75; 95% CI 0.62; 0.86). A multivariable model including micro- or macroscopic fat [Odds ratio (OR) 81.19; 95% CI 20.17; 572.27], diameter <5.5 cm (OR 7.32; 95% CI 2.17; 31.28), calcifications (OR 5.68; 95% CI 2.08; 16.18), and hemorrhage (OR 3.10; 95% CI 0.70; 15.35) had an AUC of 0.91 (95% CI 0.86; 0.96), 71% (42/59, 95% CI 58; 82) sensitivity, 93% (115/124; 95% CI 87; 97) specificity, and 86% (157/183; 95% CI 79; 90) accuracy for the diagnosis of adenoma.
Conclusion A multivariable model enables CT/MR diagnosis of heterogeneous adenomas. Presence of microscopic fat, even if partial, in a heterogeneous mass is highly specific of adenoma.
Justine Lanoix and Manel Djelouah have contributed equally to this article and share first authorship.
| ☒ Justine Lanoix justine.lanoix@gmail.com | Christine Hoeffel choeffel-fornes@chu-reims.fr | |
|---|---|---|
| Manel Djelouah mdjelouah@chu-reims.fr | 1 | Department of Radiology, Reims University Hospital, 45 rue Cognacq-Jay, 51092 Reims, France |
| Lea Chocardelle lchocardelle@chu-reims.fr | 2 | Department of Hepatic and Digestive Surgery, Reims University Hospital, 51092 Reims, France |
| Sophie Deguelte sdeguelte@chu-reims.fr | 3 | Department of Endocrinology, Reims University Hospital, 51092 Reims, France |
| Brigitte Delemer bdelemer@chu-reims.fr Anthony Dohan | 4 | Department of Abdominal & Interventional Radiology, Hôpital Cochin, Assistance Publique-Hôpitaux de Paris, 27, rue du Faubourg-Saint-Jacques, 75014 Paris, France |
| anthony.dohan@aphp.fr Philippe Soyer | 5 | Faculté de Médecine, Université de Paris, 75006 Paris, France |
| philippe.soyer@aphp.fr Maxime Barat maxime.barat@aphp.fr | 6 | CRESTIC, University of Reims Champagne-Ardenne, 51100 Reims, France |
Graphical abstract
DIFFERENTIATION BETWEEN HETEROGENEOUS ADRENAL ADENOMA AND NON-ADENOMA ADRENAL LESION WITH CT AND MRI
42 mm-large incidentally discovered heterogeneous right adrenal lesion in a 54 year-old woman. A. Unenhanced CT B. Portal phase enhanced CT C. 10 min enhanced CT D. In-Phase and E. Opposed-phase Chemical-shift MRI E. T2 Unenhanced global attenuation is 22 HU, and absolute washoutis 10 %, thus not in favour of adrenal adenoma. Presence of microscopic fat (arrowhead ; partial signal drop on OP CSI), area of chronic hemorrhage (star (area with mean attenuation 41 HU, not further enhancing)), of a tiny calcification (arrow) suggest heterogeneous adrenal adenoma with a specificity of 93 %. Pathological examination of the surgically removed lesion revealed an adrenal adenoma with hemorrhagic changes.
LANOIX J et al; 2021
Abdominal Radiology The Official Journal of the Society of Abdominal Radiology www.abdominalradiology.org
Abbreviations
ACC Adrenocortical carcinoma
AUC Area under the receiver operating characteristic curve
CI Confidence interval
CT Computed tomography
CSI Chemical shift imaging
HU Hounsfield unit
ICC Intraclass correlation coefficient
MRI Magnetic resonance imaging
OP Opposed phase
OR Odds ratio
ROI Region of interest
SD Standard deviation
Introduction
Adrenal lesions are increasingly being fortuitously discov- ered on either computed tomography (CT) or magnetic res- onance imaging (MRI) examinations [1]. Although some adrenal lesions such as cysts, myelolipomas, hematomas, and lipid-rich homogeneous adenomas are easy to charac- terize, distinguishing leave-alone lesions (i.e., adenomas) from non-adenomas that may warrant further management may still be a diagnostic challenge for the radiologist. Most lipid-rich adenomas display typical features and usually are round or oval shaped, homogeneous, well limited, and small (<4 cm), with unenhanced attenuation value ≤ 10 Hounsfield unit (HU) at CT or with a homogeneous drop in signal on opposed-phase chemical shift MRI (CSI)
[1-4]. Recently, the iodine density-to-virtual non-contrast attenuation ratio obtained from portal phase dual-energy CT has been shown to allow differentiation of adenomas from metastases with superior sensitivity and equivalent specificity compared with true unenhanced attenuation [5]. When these diagnostic methods do not allow discrimina- tion of adenomas from non-adenomas, most homogene- ous adrenal adenomas can be diagnosed based on high washout rate calculated using multiphase adrenal CT [1, 6, 7]. A small proportion of adrenal adenomas, however, are atypical on imaging because they are markedly heteroge- neous [1, 8, 9]. Heterogeneous adenomas still represent a diagnostic challenge. Their presentation on CT, let alone on MRI, has been poorly studied. The imaging features of 24 large heterogeneous degenerated adrenal adenomas as well as their differentiation from adrenocortical carcino- mas (ACC) have been reported in the late 1990s, but no evaluation of the presence of microscopic or macroscopic fat on CT/MRI was performed [10]. Since then, Gabriel et al. suggested that an adrenal lesion with heterogeneous signal drop on CSI, thus in part containing microscopic fat, was highly likely to be a benign adrenal lesion [11]. More recently, Thomas et al. compared 23 pathologically proven heterogeneous adrenal adenomas > 4 cm to ACC, using CT and concluded that evidence of micro- or mac- roscopic fat, even if only present in the solid part of the lesion, was highly specific for a benign condition [12]. Yet, some scarce case reports have described a variety of heterogeneous non-adenomas containing evidence of fat at CT or MRI and thus mimicking adenomas [13-19].
The purpose of this study was to assess whether heteroge- neous adrenal adenomas could be distinguished from hetero- geneous non-adenomas on CT or MRI and to identify CT or/ and MRI signs associated with the diagnosis of heterogene- ous adenomas.
Materials and methods
This study was performed in accordance with the ethical standards of the two institutions. Due to the retrospective design of the study, informed consent from the patients was not required according to national policy.
Study population
A search of the pathological databases from two academic medical centers from 2009 to 2019 retrieved 980 patients who underwent adrenalectomy. Patients with adrenal hyperplasia (n=74) or with extra-adrenal lesions (n=27), simple cysts (n=14), myelolipomas (n=12), and spon- taneous hematomas (n=6) were excluded. Patients with abdominal CT or MRI performed within two months of the
adrenalectomy including at least unenhanced CT, or CSI- and fat-suppressed T1-weighted MRI sequences were eligi- ble for inclusion. CT and MRI examinations of the remain- ing 463 patients were reviewed in consensus by a radiologist (J.L.) and a neuroendocrine surgeon (S.D.) with, respec- tively, 2 and 15 years of experience in abdominal imaging, to exclude patients with limited imaging quality (n=4), nodules < 1 cm (n=35), and non-heterogeneous lesions (n=241) (Fig. 1). A heterogeneous lesion was defined as containing regions of variable attenuation or signal inten- sity, or heterogeneously suppressing on CSI, or containing evidence of necrosis, cystic, or hemorrhagic changes, or a significant amount of calcifications, hindering reliable place- ment of a region of interest (ROI) on at least two-thirds of the lesion excluding the adrenal cortex.
Imaging evaluation
CT examinations
A variety of CT scanners were used, including Sensation® (16 or 64 rows of detector), Definition® (64 rows), Somatom Drive® (128 rows) (Siemens Healthineers), Discovery 750 HD® (64 rows), and Revolution Evo® (128 rows) (GE Healthcare). Imaging protocols varied over the review period because of changes in CT technology and the num- ber of units used.
Multidetector row CT acquisition parameters were as follows: tube voltage, 120 kVp; section collimation, 80×0.5 mm-64×1.25 mm; helical pitch, 0.81-1.37; scan time per spiral, 0.5-0.7 s; and image reconstruction thick- ness, 2.5-3 mm. Effective tube current values differed between patients due to automatic tube current modulation. A volume of 2-3 mL/kg body weight of non-ionic contrast material was injected into an antecubital vein at a flow rate of 3.5 mL/s. Images obtained at portal and late phases were acquired 70-80 s and 10 min after contrast material admin- istration, respectively.
One hundred and twenty-four patients underwent abdomi- nal CT with and without intravenous administration of iodi- nated contrast material during the portal venous phase and 32 had multiphasic imaging allowing measurement of abso- lute percentage of washout at 10 min following injection.
MRI examinations
One hundred and sixteen patients underwent abdominal MRI using a 1.5-T (Magnetom Avanto ®, Sola®, or Aera®; Siemens Healthineers) or 3-T unit (Achieva®; Philips, and Magnetom Skyra®; Siemens Healthineers) and multichan- nel phased array coils. Fifty-seven patients underwent both abdominal MRI and CT.
Patients with adrenalectomies from 2009 to 2019 (n=980)
Excluded patients (n = 517; 52.8%)
74 adrenal hyperplasia
27 extra-adrenal lesion
14 simple cysts
12 myelolipomas 6 spontaneous hematomas
Remaining patients with adrenalectomies from 2009 to 2019 (n = 463)
384 CT/MRI not available or incomplete examinations
Excluded patients (n =280; 60.5%)
4 poor quality imaging
35 small lesions (< 1cm)
241 homogeneous lesions
Study population (n = 183)
Adrenal heterogeneous adenomas (n =59)
Adrenal heterogeneous non-adenomas (n= 124)
MRI protocol included axial fat-suppressed T2 sequences (TRs 2500-5740, TEs 92-121, flip angle 122-147, slice thickness 4-5 mm with intersection gap 0-1 mm, matrix size 256-320x256-320, one signal), unenhanced breath-hold in- and out-of-phase gradient echo T1-weighted sequences (CSI) (TRs 120-160 ms, TEs 4-4.8 ms, flip angle 70-90° for in-phase and TEs 2-2.4 ms, flip angle 55-70° for opposed- phase, with a size matrix of 260-320x 195-320, one signal acquired, 3-7 mm slice thickness, and 0-1 mm intersection gap), and axial fat-suppressed three-dimensional breath-hold gradient echo T1-weighted MRI sequences before and after intravenous administration of a gadolinium chelate during the portal venous phase.
Diffusion-Weighted Imaging (DWI) sequences, when per- formed, were not considered in the evaluation.
Image analysis
All images were independently reviewed on a picture archiv- ing and communications system (Synapse solution, Fujifilm) by two abdominal radiologists with, respectively, 1 and 27 years of experience in abdominal imaging (L.C., C.H.) blinded to clinical and biological data and final diagnosis. Analysis was combined for the patients who had both CT and MRI examinations for all criteria except for the pres- ence of micro- or macroscopic fat and for calcification which were only analyzed on CT.
Qualitative analysis included margins (regular, lobu- lated, or irregular), calcification (on CT), macroscopic fat (visually detectable area(s) of attenuation < - 20 HU, signal loss after fat suppression on T1-weighted MRI), microscopic fat (area(s) of attenuation ≤ 10 HU visu- ally detectable on CT, marked signal drop relative to the spleen on opposed-phase (OP) compared with in-phase CSI sequences, as assessed visually), signal intensity (hyper-, iso-, or hypointense relative to the renal cortex) on T2-weighted MRI, necrotic and/or cystic changes, hem- orrhage (area(s) of unenhanced attenuation value > 40 HU without any contrast enhancement after contrast injection [20] or non-suppressing areas of high signal intensity on T1/fat-suppressed T1-weighted MRI compared to renal parenchyma).
Distribution of the microscopic fatty component of the lesion on CT or MRI was defined as “diffuse” (in the major part of the lesion), “multiple foci” (intermingled puncti- form spots), “partial” (at least one third of the size of the lesion), and “crescent-like” (thin peripheral).
Quantitative analysis included lesion size (largest diam- eter on axial images expressed in cm), whole-lesion atten- uation (HU) on pre-contrast CT (excluding the necrotic or cystic portions and the edge of the lesion when possible), and attenuation (HU) of the microscopic fatty part, if pre- sent. CT absolute washout measurements were calculated when possible, using the validated formula, at 10 min [1, 6].
Histopathology
Pathological examination of the resected specimens was per- formed by mainly two pathologists with more than 15 and 25 years of experience in adrenal gland pathology, respec- tively. The presence of mature adipocytes, myelolipomatous metaplasia, and hemorrhage were recorded from the histo- pathological reports of the resected specimens. Weiss Score was used for differentiation of adenomas from ACC: score of 0 indicating benign adenoma, score 1 to 2 adrenocortical neoplasm with uncertain malignant potential, and a score higher than 3 highly likely to be malignant [21].
Statistical analysis
Categorical variables were expressed as raw numbers, proportions, and percentages and continuous variables as means ± standard deviations (SD) and ranges. Differences between adenomas and non-adenomas were analyzed using the Chi2 test for categorical variables and Student t test or Fischer exact test for continuous variables. Results of the most experienced reader were kept for analysis.
Interobserver agreement was evaluated using the weighted kappa (k) test for non-continuous variables [12, 22] and intraclass correlation coefficient (ICC) for continu- ous variables [23]. Sensitivity, specificity, and accuracy of combined CT and MRI features (CT/MRI) for the diagno- sis of heterogeneous adrenal adenoma versus adrenal non- adenoma were determined and threshold size was defined using Youden’s index. Sensitivity, specificity, PPV, NPV, and accuracy of CT, MRI, and combined CT/MRI for the independent analysis of microscopic, macroscopic, and micro- or macroscopic fat were also calculated.
Multivariable logistic regression with forward and back- ward selection was used to identify the independent char- acteristics predicting adenomas. The accuracy of the final multivariable model was calculated and the corresponding sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUC) were calculated. Sta- tistical analyses were performed with R software (version 3.6.1, R Development Core Team, 2019). P values <0.05 were considered statistically significant.
Results
Patient demographics
One hundred and eighty-three patients with 59 adrenal ade- nomas and 124 non-adenomas constituted the final popula- tion. There were 106 women and 77 men with a mean age of 53.2±14.4 [SD] years (age range 19-83 years). For the patient with bilateral adrenal lesions, both had the same CT/
MRI characteristics and only the most heterogeneous one was considered for evaluation. Figure 1 shows the study flow chart. Demographics characteristics of patients are reported in Table 1. Non-adenomas included pheochromocytomas (28/124; 22.6%), ACC (72/124; 58.1%), metastases [18/124; 14.5%, from lung cancer (n=9/124; 7.3%), skin melanoma (n=2/124; 1.6%), hepatocellular carcinoma (n=2/124; 1.6%), kidney clear cell adenocarcinoma (n=1/124; 0.8%), colorectal cancer (n=2/124; 1.6%), and unknown primary (n=2/124, 1.6%)], ganglioneuromas (3/124; 2.4%), sarco- mas (2/124; 1.6%), and a composite tumor with a combi- nation of pheochromocytoma and ganglioneuroma (1/124; 0.8%) (Table 2). At histopathological analysis, areas of myeloid metaplasia/myelolipomatous foci were present in 22 (22/59; 37.3%) adenomas and in no (0/124; 0%) non-adeno- mas. Interspersed mature adipocytes were found without any myeloid metaplasia in one ACC and one ganglioneuroma. Adenomatous lesions all had a Weiss score ≤ 1.
CT findings
Table 3 reports the CT findings in 124/183 patients. Sig- nificant differences in mean overall unenhanced attenuation value, macroscopic fatty component, and microscopic fatty component were found between adenomas and non-adeno- mas (31 ±9 [SD] HU vs. 23 ±11 [SD] HU; 18/5135.3% vs. 0/73 0%; 18/51 35.3% vs. 2/73 2.7%, respectively; P<0.001 for all) (Table 3). No differences in “absolute washout greater than 60%” were found between adenomas (7/23; 30.4%) and non-adenomas (1/9; 11.1%) (P=0.386) (Fig. 2).
MRI findings
Table 4 summarizes the MRI findings in 116/183 patients. Significant differences in macroscopic fatty component, microscopic fatty component, and micro- or macroscopic fatty component were found between adenomas and non- adenomas (11/40 27.5% vs 0/76 0%; 24/40 60.0% vs 0/76 0%; and 25/40 62.5% vs 0/76 0%, respectively; P <0.001).
Combined CT and MRI findings (CT/MRI)
A hundred and eighty-three lesions in 183 patients were analyzed (Table 1). On CT, no non-adenomas con- tained macroscopic fat versus 22/59 (37.3%) adenomas (P<0.001). Microscopic fat was more frequently observed in adenomas (32/59; 54.2%) than in non-adenomas (2/124; 1.6%) (P<0.001). The two (2/124; 1.6%) non-adenomas with microscopic fat seen on CT were a ganglioneuroma and an ACC, with a “multiple foci pattern” distribution (Figs. 3, 4). When combining both micro- and macro- scopic fat, 37 (37/59; 62.7%) adenomas contained micro- or macroscopic fat versus 2 (2/124; 1.6%) non-adenomas
| Variable | NAA (n=124) | AA (n=59) | Kappa [95% CI] | OR [95% CI]* | P value |
|---|---|---|---|---|---|
| Sex | N.A | 0.92 [0.49; 1.72] | 0.792 | ||
| Man | 53 (42.7%) | 24 (40.7%) | – | ||
| Woman | 71 (57.3%) | 35 (59.3%) | |||
| Age (year) | 52.2±15.2 [19-83] | 55.4±12.4 [2-81] | N.A | 1.02 [0.99; 1.04] | 0.079 |
| Side | N.A | – | 0.790 | ||
| Right | 58 (46.8%) | 28 (47.5%) | |||
| Left | 65 (52.4%) | 31 (52.5%) | |||
| Bilateral | 1 (0.8%) | 0 (0%) | |||
| Functioning | N.A | 0.40 [0.21; 0.74] | 0.004 | ||
| Yes | 74 (60.2%) | 22 (37.3%) | – | ||
| No | 49 (39.8%) | 37 (62.7%) | |||
| Macroscopic fat | 0.82 [0.66; 0.94] | – | <0.001 | ||
| Present | 0 (0%) | 22 (37.3%) | |||
| Absent | 124 (100%) | 37 (62.7%) | |||
| Microscopic fat | 0.75 [0.62; 0.86] | 72.30 [20.27; 463.47] | <0.001 | ||
| Present | 2 (1.6%) | 32 (54.2%) | – | ||
| Absent | 122 (98.4%) | 27 (45.8%) | |||
| Micro- or macroscopic fat | 1 [1; 1] | 102.59 [28.61; 660.24] | <0.001 | ||
| Present | 2 (1.6%) | 37 (62.7%) | – | ||
| Absent | 122 (98.4%) | 22 (37.3%) | |||
| Diameter <5.5 cm | N.A | 6.68 [3.26; 14.70] | <0.001 | ||
| Yes | 49 (39.5%) | 48 (81.4%) | – | ||
| No | 75 (60.5%) | 11 (18.6%) | |||
| Hemorrhage | 0.76 [0.63; 0.86] | 0.65 [0.27; 1.44] | 0.302 | ||
| Present | 27 (21.8%) | 9 (15.3%) | – | ||
| Absent | 97 (78.2%) | 50 (84.7%) | |||
| Calcification | 0.77 [0.66; 0.87] | 4.97 [2.45; 10.32] | <0.001 | ||
| Yes | 18 (14.5%) | 27 (45.8%) | – | ||
| No | 106 (85.5%) | 32 (54.2%) | |||
| Necrotic/cystic transformation | 0.50 [0.36; 0.63] | 0.20 [0.09; 0.41] | <0.001 | ||
| Yes | 18 (14.5%) | 32 (54.2%) | – | ||
| No | 106 (85.5%) | 27 (45.8%) | |||
| Margin | 0.48 [0.36; 0.60] | 13.11 [5.35; 39.61] | <0.001 | ||
| Irregular/lobulated | 68 (54.8%) | 5 (8.5%) | – | ||
| Regular | 56 (45.2%) | 54 (91.5%) |
NAA non-adrenal adenoma, AA adrenal adenoma
Quantitative variables are expressed as means ± standard deviations; numbers in brackets are ranges
Qualitative variables are expressed as raw numbers; numbers in parentheses are percentages. N.A. not applicable OR = odds ratio followed by 95% confidence interval in brackets. Bold indicates significant P value
*Odds ratios and 95% CIs are not shown for some variables because a zero value for prevalence led to unstable estimate of these parameters
(P <0.001). Distribution of fat in the 32 adenomas with microscopic fat was as follows: diffuse (3/32; 9.3%), mul- tiple (10/32; 31.3%), partial (9/32; 28.1%), and “cres- cent-like” (10/32; 31.3%). Adrenal lesions smaller than 5.5 cm were more prevalent in adenomas (48/59; 81.4%) than in non-adenomas (49/124; 39.5%) (P <0.001). The optimal size threshold for distinguishing adenomas from
non-adenomas was 5.5 cm, yielding 60% specificity and 81% sensitivity for the diagnosis of adenoma.
No differences in prevalence of hemorrhage were found between adenomas (9/59; 15.3%) and non-adeno- mas (27/124; 21.8%) (P=0.30). Calcifications were more frequently observed in adenomas (27/59; 45.8%) than in non-adenomas (18/124; 14.5%) (P <0.001). Necrotic and/
| Pathological examination diagnosis | NAA (n=124) |
| Pheochromocytoma | 28 (22.6%) |
| Adrenocortical carcinoma | 72 (58.1%) |
| Metastases | 18 (14.5%) |
| Lung cancer | 9 (7.3%) |
| Melanoma | 2 (1.6%) |
| HCC | 2 (1.6%) |
| Kidney clear cell carcinoma | 1 (0.8%) |
| Colorectal | 2 (1.6%) |
| Unknown | 2 (1.6%) |
| Ganglioneuroma | 3 (2.4%) |
| Sarcoma | 2 (1.6%) |
| Composite tumor | 1 (0.8%) |
NAA non-adrenal adenoma
or cystic changes were more often present in adenomas (32/59; 54.2%) than in non-adenomas (18/124; 14.5%) (P <0.001) (Fig. 5). Irregular or lobulated margins were more frequently observed in non-adenomas (68/124; 54.8%) than in adenomas (5/59; 8.5%) (P <0.001).
Interobserver agreement
Interobserver agreement (Table 1) for combined CT/MRI features was almost perfect regarding the presence of mac- roscopic fat (k=0.82; 95% CI 0.66; 0.94) and substantial for the presence of hemorrhage (k=0.76; 95% CI 0.93; 0.86), calcification (k=0.77; 95% CI 0.66; 0.87), and microscopic fat (k=0.75; 95% CI 0.62; 0.86). Agreement was moderate regarding the assessment of margins (k=0.48; 95% CI 0.36; 0.60) and necrotic and/or cystic changes (k=0.50; 95% CI 0.36; 0.63).
Diagnostic performances of CT/MRI for the diagnosis of heterogeneous adenoma
Sensitivity and specificity for the diagnosis of heterogene- ous adenoma were 37% (22/59; 95% CI 25; 51) and 100% (124/124; 95% CI 97; 100), respectively, regarding the pres- ence of macroscopic fat and 54% (32/59; 95% CI 41; 67) and 98% (122/124; 95% CI 94; 100), respectively, for the pres- ence of microscopic fat (Table 5). When combining micro- and macroscopic fat, sensitivity for adenoma was 63% (37/59; 95% CI 49; 75), with 98% specificity (122/124; 95% CI 94; 100). Hemorrhage had a specificity of 78% (97/124; 95% CI 70; 85) and calcifications had a specificity of 85% (106/124; 95% CI 78; 91) for the diagnosis of adenoma. Table 6 shows diagnostic performances of CT, MRI, and
| CT variable | NAA (n=73) | AA (n=51) | Kappa or ICCt [95% CI] | OR [95% CI]* | P value |
|---|---|---|---|---|---|
| Mean overall unenhanced density (HU) | 31 ±9 [12-83] | 23 ±11 [-12-44] | 0.52+ [0.38; 0.64] | 0.92 [0.87; 0.96] | <0.001 |
| Macroscopic fat | 0.94 [0.83; 1] | – | <0.001 | ||
| Present | 0 (0%) | 18 (35.3%) | |||
| Absent | 73 (100%) | 33 (64.7%) | |||
| Microscopic fat | 0.77 [0.62; 0.90] | 35.5 [9.7; 230.2] – | <0.001 | ||
| Present | 2 (2.7%) | 18 (35.3%) | |||
| Absent | 71 (97.3%) | 33 (64.7%) | |||
| Micro- or macroscopic fat | 1 [1; 1] | 52.4 [14.3; 340.9] – | <0.001 | ||
| Present | 2 (2.7%) | 26 (51.0%) | |||
| Absent | 71 (97.3%) | 25 (49.0%) | |||
| Mean microscopic fatty component density (HU) | NA | 2±6 [-15 to 8] | 0.32+ [0; 0.68] | – | |
| Absolute washout (n=32) | 9 | 23 | 1 [1; 1] | 0.386 | |
| >60% | 1 (11.1%) | 7 (30.4%) | 3.50 [0.49; 71.35] | ||
| <60% | 8 (88.9%) | 16 (69.6%) | – |
NAA non-adrenal adenoma, AA adrenal adenoma
Qualitative variables are expressed as raw numbers; numbers in parentheses are percentages
Quantitative variables are expressed as means ± standard deviations; numbers in brackets are ranges
HU Hounsfield unit, OR odds ratio followed by confidence interval in brackets. Bold indicates significant P value
*Odds ratios and 95% CIs are not shown for some variables because a zero value for prevalence led to unstable estimate of these parameters
a
b
C
combined CT/MRI-independent analysis of microscopic or/and macroscopic fat for the diagnosis of heterogeneous adrenal adenoma.
Multivariable analysis
Table 7 shows the most relevant qualitative and quantitative model related to heterogeneous adrenal adenomas, using a threshold size <5.5 cm. Due to poor interobserver agree- ment, necrotic/cystic changes and margins were not kept as variables to build the model. Presence of hemorrhage was included in the multivariate model, even if no statistical dif- ferences were showed using univariate analysis owing to its presumed clinical relevance. The model included micro- or macroscopic fat (OR 81.19; 95% CI 20.18; 572.27), size <5.5 cm (OR 7.32; 95% CI 2.17; 31.28), calcification (OR 5.68; 95% CI 2.08; 16.18), and hemorrhage (OR 3.10; 95% CI 0.70; 15.35). This model yielded an AUC of 0.91 (95% CI 0.86; 0.96), with 86% accuracy (95% CI 79; 90), 71% sensitivity (95% CI 58; 82), and 93% specificity (95% CI 87; 97) (Fig. 6).
Discussion
Our study demonstrates that a diagnostic model including size below 5.5 cm, calcifications, hemorrhagic changes, and micro- or macroscopic fat inside the adrenal lesion allows differentiation of heterogeneous adenoma versus non-ade- noma with an area under the curve of 0.91. Moreover, the presence of microscopic fat inside a heterogeneous adrenal lesion at CT/MRI, even if partial, allows diagnosis of het- erogeneous adenoma versus non-adenoma with a specificity of 98% and a sensitivity of 54%. Besides, it is noteworthy that interobserver agreement regarding the presence of fat was almost perfect and was improved using the combination of CT and MRI.
CT criteria defined in the late 1990s that are now widely accepted to diagnose adrenal adenomas only apply to non- significantly heterogeneous lesions [24, 25]. They indeed imply placement of a circular region of interest on at least two-thirds of the mass, avoiding areas of necrosis or calcifi- cations [26]. Such a measurement may be hazardous, if not impossible in case of a heterogeneous lesions. Moreover, endocrinologists are generally reluctant to accept a diagnosis of benign lesion when a lesion is markedly heterogeneous. As far as MRI is concerned, the criterion for an adrenal adenoma is complete and homogeneous suppression of sig- nal on opposed-phase CSI [25, 27].
Characterization of a heterogeneous adrenal lesion thus still remains a diagnostic challenge, particularly if fortui- tously discovered. If such lesions could be classified as being a probable adenoma or as a lesion that has to be resected,
| MRI variable | NAA (n=76) | AA (n=40) | Kappa [95% CI] | OR [95% CI]* | P value |
|---|---|---|---|---|---|
| Macroscopic fat | 0.74 [0.52; 0.91] | - | <0.001 | ||
| Present | 0 (0%) | 11 (27.5%) | |||
| Absent | 76 (100%) | 29 (72.5%) | |||
| Microscopic fat | 0.83 [0.70; 0.94] | - | <0.001 | ||
| Present | 0 (0%) | 24 (60.0%) | |||
| Absent | 76 (100%) | 16 (40.0%) | |||
| Micro- or macroscopic fat | 1 [1; 1] | – | <0.001 | ||
| Present | 0 (0%) | 25 (62.5%) | |||
| Absent | 76 (100%) | 15 (37.5%) | |||
| T2 signal | 0.92 [0.85; 0.98] | <0.001 | |||
| Hypointense | 8 (10.5%) | 8 (20.0%) | 3.93 [1.25; 12.59] | ||
| Hyperintense | 13 (17.1%) | 18 (45.0%) | 5.44 [2.20; 14.07] | ||
| Isointense | 55 (72.4%) | 14 (35.0%) | – |
NAA non-adrenal adenoma, AA adrenal adenoma
Qualitative variables are expressed as raw numbers; numbers in parentheses are percentages OR odds ratio followed by confidence interval in brackets. Bold indicates significant P value
*Odds ratios and 95% CIs are not shown for some variables because a zero value for prevalence led to unstable estimate of these parameters
this would aid in management, particularly in frail patients. Indeed, adrenal operations, despite the generalization of laparoscopic procedures and the development of mini-inva- sive approaches, are still associated with 5-10% periopera- tive complications since they increasingly involve geriatric patients who may have significant comorbidities [28, 29].
Literature regarding CT and MRI evaluation of heteroge- neous adenomas is rather poor. Thomas et al. reviewed the CT characteristics of 25 adrenal adenomas larger than 4 cm, of which 23 were heterogeneous and 33 ACC [12]. These researchers found that unenhanced attenuation value < 10 HU on unenhanced images measured on a solid-appearing enhancing portion of the tumor had 100% specificity for the diagnosis of adenoma [12]. Similarly, presence of macro- scopic fat (attenuation ← 20 HU) yielded 93% specificity [12]. However, these two features yielded sensitivities of only 41% and 33%, respectively [12], lower than 54% and 37%, respectively, found in our series at CT/MRI.
In our series, mean whole-lesion attenuation measured on the largest possible area, excluding necrotic or cystic changes, when possible, was greater in non-adenomas than in adenomas; however, the associated OR was close to 1, indicating that this criterion was poorly discriminative.
It is noteworthy that 69.6% of adenomas did not show an absolute percentage washout diagnostic of adenoma. Thomas et al. also reported 44% of their adenomas dis- playing washout less than 60%, suggesting that washout measurements may be limited for the diagnosis of het- erogeneous adenomas. Washout measurement is indeed likely to be modified in case of diffuse more or less chronic hemorrhagic changes which are frequent in this population
of heterogeneous adenomas, as the unenhanced attenua- tion of the lesion is then increased and the lesion poorly enhancing after contrast administration.
In our study, interspersed adipocytes within the two non-adenomas were diagnosed at CT as microscopic fat by both readers. Measured attenuations in the fatty areas were between - 5 and 0 HU for the ganglioneuroma and between - 20 and 0 HU for the ACC. Our threshold used to define macroscopic fat was rigorous ( ← 20 HU) and a substantial amount of clustered adipocytes is probably required to reach such low attenuation values, such as those found in myelolipomas [30]. Unlike Thomas et al. [12] who reported only moderate agreement for this fea- ture on CT, we found an almost perfect agreement for macroscopic fat both on CT and on combined CT/MRI. Of note, the so-called “lipomatous ganglioneuromas” located in other anatomic areas than the adrenal gland have already been reported [13, 31, 32].
Lesions displaying heterogeneous signal drop on opposed-phase CSI were all adenomas in our study, consist- ent with previous findings of eighteen reported heterogene- ously suppressing lesions at CSI that proved to be benign [11].
Fat on imaging, either microscopic or macroscopic, has been reported, although anecdotally, in a variety of non- adenomatous lesions [33], including metastases [15], phe- ochromocytomas [19], and ACC [34]. A systematic review reviewed the three case reports and one short series report- ing macroscopic fat in ACC detected on imaging studies. These authors showed that the proportion of gross fat was under 5% in lesions all above 6 cm and that the reliability
a
b
of the reported cases was questionable, thus emphasizing the rarity of fat-containing ACC [34].
Up to 37% of adenomas in our series had areas of mye- loid/lipomatous metaplasia at histopathological analysis. This finding has been previously reported and it has been suggested that these interspersed foci of myelolipoma- tous metaplasia may appear as areas of suppression on OPCSI, unlike what is usually described in myelolipomas
that display areas of fat suppression on fat-suppressed T1-weighted MRI [11, 30].
Lastly, our findings of a high proportion of calcifica- tions in heterogeneous adenomas (45.8%) are consistent with two other studies reporting a frequency of 40% of cal- cifications in large adenomas [10, 12]. We assume that cal- cifications represent the eventual evolution of underlying
chronic hemorrhagic changes that may not be easy to diag- nose at CT/MRI.
Our study has limitations inherent to its retrospective nature, which hindered prospective accurate matching of areas of micro- or macroscopic fat at imaging with their counterparts at pathology. Yet, the presence of fat iden- tified at imaging was always confirmed in the reports. Another bias is that our study was limited to patients who underwent adrenalectomy. Specifically, it excluded malignant lesions that may not undergo resection such as, specifically, adrenal metastases from clear cell renal cell carcinoma that have been reported to contain fat in up to 33% of cases [15]. However, Song et al. [35] have shown that in a population of more than 1000 consecu- tive patients with no history of malignancy, no metastases were detected, thus emphasizing that the use of fat as a criteria for the diagnosis of heterogeneous adenoma has to be considered in view of the history of the patient. Lastly, a delayed phase after injection was lacking in 75% of our patients, so that absolute percentage of washout could
| CT/MRI variable | TP | TN | FP | FN | Sensitivity (%) [95% CI] | Specificity (%) [95% | CI] Accuracy (%) [95% CI] |
|---|---|---|---|---|---|---|---|
| Macroscopic fat | 22 | 124 | 0 | 37 | 37 [25; 51] | 100 [97; 100] | 80 [73; 85] |
| Microscopic fat | 32 | 122 | 2 | 27 | 54 [41; 67] | 98 [94; 100] | 85 [79; 90] |
| Micro- or macroscopic fat | 37 | 122 | 2 | 22 | 63 [49; 75] | 98 [94; 100] | 87 [81; 91] |
| Hemorrhage | 9 | 97 | 27 | 50 | 15 [7; 27] | 78 [70; 85] | 58 [50; 65] |
| Calcification | 27 | 106 | 18 | 32 | 46 [33; 59] | 85 [78; 91] | 73 [66; 79] |
| Necrosis | 32 | 18 | 106 | 27 | 54 [41; 67] | 14 [9; 22] | 27 [21; 34] |
| Regular marginª | 54 | 68 | 56 | 5 | 92 [81; 97] | 55 [46; 64] | 67 [59; 73] |
TP true positive, TN true negative, FP false positive, FN false negative, CI confidence interval aNon-irregular or non-lobular margin
| Variable | TP | TN | P | FN | Sensitivity (%) [95% CI] | Specificity (%) [95% | CI] PPV (%) [95% | CI] NPV (%) [95% CI] | Accuracy (%) [95% CI] |
|---|---|---|---|---|---|---|---|---|---|
| CT Macroscopic fat | 18 | 73 | 33 | 0 | 35 [22; 49] | 100 [95; 100] | 100 [81; 100] | 68 [59; 77] | 73 [64; 80] |
| MRI Macroscopic fat | 11 | 76 | 29 | 0 | 27 [14; 43] | 100 [95; 100] | 100 [71; 100] | 72 [62; 80] | 75 [66; 82] |
| CT/MRI Macroscopic fat | 22 | 124 | 0 | 37 | 37 [25; 51] | 100 [97; 100] | 100 [85; 100] | 77 [70; 83] | 80 [73; 85] |
| CT Microscopic fat | 18 | 71 | 33 | 2 | 35 [22; 49] | 97 [90; 99] | 90 [68; 98] | 68 [58; 77] | 71 [62; 79] |
| MRI Microscopic fat | 24 | 76 | 16 | 0 | 60 [43; 75] | 100 [95; 100] | 100 [85; 99] | 82 [73; 89] | 86 [78; 91] |
| CT/MRI Microscopic fat | 32 | 122 | 2 | 27 | 54 [41; 67] | 98 [94; 100] | 94 [80; 99] | 82 [75; 88] | 85 [79; 90] |
| CT Micro- or macroscopic fat | 26 | 71 | 25 | 2 | 50 [36; 65] | 97 [90; 99] | 92 [76; 99] | 73 [64; 82] | 78 [69; 85] |
| MRI Micro- or macroscopic fat | 25 | 76 | 15 | 0 | 62 [45; 77] | 100 [95; 100] | 100 [86; 99] | 83 [74; 90] | 87 [79; 92] |
| CT/MRI Micro- or macro- scopic fat | 37 | 122 | 2 | 22 | 63 [49; 75] | 98 [94; 100] | 95 [83; 99] | 85 [78; 90] | 87 [81; 91] |
TP true positive, TN true negative, FP false positive, FN false negative, PPV positive predictive value, NPV negative predictive value, CI confi- dence interval
aNon-irregular or non-lobular margin
| Variableª | OR | [95% CI] | P value |
|---|---|---|---|
| Micro- or macroscopic fat | 81.19 | [20.18; 572.27] | <0.001 |
| Threshold size <5.5 cm | 7.32 | [2.17; 31.28] | 0.002 |
| Calcification | 5.68 | [2.08; 16.18] | <0.001 |
| Hemorrhage | 3.10 | [0.70; 15.35] | 0.142 |
a Adjusted on age
OR odds ratio, CI confidence interval. Bold indicates significant P value
not be calculated. In addition, the 10-min delayed adrenal enhancement washout test that we used in the present study has been reported to have decreased sensitivity to adenomas compared with a 15-min delay protocol [36, 37].
In conclusion, we propose a multivariable model based on the presence of fat, even if partial, inside a heterogene- ous lesion. This model allows highly specific diagnosis of heterogeneous adrenal adenomas which may be then left in place and followed up in a patient without a context of neoplasia, in whom a heterogeneous lesion is fortuitously discovered and for whom surgery is at risk.
Acknowledgements The authors gratefully acknowledge all the inves- tigators for their contributions to the trial.
ROC curve
Sensitivity
0.0 0.2 0.4 0.6 0.8 1.0
AUC
0.908
0.0
0.2
0.4
0.6
0.8
1.0
1-Specificity
Author contributions Guarantor of the integrity of the entire study: CH. Conceptualization and methodology: AD, PS, MB, and CH. Lit- erature research: JL, MD, MB, and CH. Investigation: JL, MD, LC, SD, BD, and MB. Data and statistical analysis: MD. Writing original draft: JL, MD, AD, and CH. Writing and reviewing of the manuscript: JL, LC, SD, BD, AD, PS, MB, and CH. Supervision: MD, PS, MB, and CH.
Funding No funds, grants, or other support were received.
Declarations
Conflict of interest: The authors have no relevant financial or non- financial interests to disclose.
Ethical approval Due to the retrospective design of the study, informed consent from the patients was not required according to national policy.
Research involving human and rights The authors declare that the work described has been carried out in accordance with the Declaration of Helsinki of the World Medical Association revised in 2013 for experi- ments involving humans.
References
1. Fassnacht M, Arlt W, Bancos I et al (2016) Management of adre- nal incidentalomas: European Society of Endocrinology Clinical Practice Guideline in collaboration with the European Network for the Study of Adrenal Tumors. Eur J Endocrinol 175:G1-G34.
2. Korobkin M, Giordano TJ, Brodeur FJ et al (1996) Adrenal adeno- mas: relationship between histologic lipid and CT and MR find- ings. Radiology 200:743-747.
3. Mayo-Smith WW, Song JH, Boland GL, et al (2017) Manage- ment of Incidental Adrenal Masses: A White Paper of the ACR Incidental Findings Committee. J Am Coll Radiol 14:1038-1044.
4. d’Amuri FV, Maestroni U, Pagnini F, et al. (2019) Magnetic resonance imaging of adrenal gland: state of the art. Gland Surg.8:S223-S232.
5. Nagayama Y, Inoue T, Oda S, et al (2020) Adrenal adenomas versus metastases: diagnostic performance of dual-energy spectral CT virtual noncontrast imaging and iodine maps. Radiology;296:324-332.”
6. Lattin GE, Sturgill ED, Tujo CA et al (2014) From the radio- logic pathology archives: adrenal tumors and tumor-like con- ditions in the adult: radiologic-pathologic correlation. Radio- graphics 34:805-829.
7. Liu T, Sun H, Zhang H, Duan J, Hu Y, Xie S (2019) Distinguish- ing adrenal adenomas from non-adenomas with multidetector CT: evaluation of percentage washout values at a short time delay triphasic enhanced CT. Br J Radiol 92:20180429.
8. Elsherif SB, Javadi S, Blair KJ et al (2020) Preresection radio- logic assessment and imaging features of 156 pathologically proven adrenal adenomas. J Comput Assist Tomogr 44:419-425.
9. Sahdev A (2017) Recommendations for the management of adrenal incidentalomas: what is pertinent for radiologists? Br J Radiol 90:20160627.
10. Newhouse JH, Heffess CS, Wagner BJ, Imray TJ, Adair CF, Davidson AJ (1999) Large degenerated adrenal adenomas: radiologic-pathologic correlation. Radiology 210:385-391.
11. Gabriel H, Pizzitola V, McComb EN, Wiley E, Miller FH (2004) Adrenal lesions with heterogeneous suppression on chemical shift imaging: clinical implications. J Magn Reson Imaging 19:308-316.
12. Thomas AJ, Habra MA, Bhosale PR et al (2018) Interobserver agreement in distinguishing large adrenal adenomas and adreno- cortical carcinomas on computed tomography. Abdom Radiol 43:3101-3108.
13. Shaaban AM, Rezvani M, Tubay M, Elsayes KM, Woodward PJ, Menias CO (2016) Fat-containing retroperitoneal lesions: imaging characteristics, localization, and differential diagnosis. Radiographics 36:710-734.
14. Sydow BD, Rosen MA, Siegelman ES (2006) Intracellular lipid within metastatic hepatocellular carcinoma of the adrenal gland: a potential diagnostic pitfall of chemical shift imaging of the adrenal gland. AJR Am J Roentgenol 187:W550-W551.
15. Schieda N, Krishna S, McInnes MDF et al (2017) Utility of MRI to differentiate clear cell renal cell carcinoma adrenal metastases from adrenal adenomas. AJR Am J Roentgenol 209:W152-W159.
16. Shinozaki K, Yoshimitsu K, Honda H et al (2001) Abdominal cystic tumors containing small amount of fat in the septa: report of two cases. Abdom Imaging 26:315-318.
17. Yoo JY, McCoy KL, Carty SE et al (2015) Adrenal imaging fea- tures predict malignancy better than tumor size. Ann Surg Oncol 22(S3):721-727.
18. Schieda N, Davenport MS, Pedrosa I et al (2019) Renal and adre- nal masses containing fat at MRI: proposed nomenclature by the society of abdominal radiology disease-focused panel on renal cell carcinoma. J Magn Reson Imaging 49:917-926.
19. Blake MA, Krishnamoorthy SK, Boland GW et al (2003) Low- density pheochromocytoma on CT: a mimicker of adrenal ade- noma. AJR Am J Roentgenol 181:1663-1668.
20. Silverman SG, Mortele KJ, Tuncali K et al (2007) Hyperattenuat- ing renal masses: etiologies, pathogenesis, and imaging evalua- tion. RadioGraphics 27:1131-43.
21. Aubert S, Wacrenier A, Leroy X et al (2002) Weiss system revis- ited: a clinicopathologic and immunohistochemical study of 49 adrenocortical tumors. Am J Surg Pathol 26:1612-1619.
22. Benchoufi M, Matzner-Lober E, Molinari N, Jannot AS, Soyer P (2020) Interobserver agreement issues in radiology. Diagn Interv Imaging 101(10):639-641.
23. Shrout PE, Fleiss JL (1979) Intraclass correlations: uses in assess- ing rater reliability. Psychol Bull 86:420-428.
24. Schieda N, Siegelman ES (2017) Update on CT and MRI of adre- nal nodules. AJR Am J Roentgenol 208:1206-1217.
25. Garrett RW, Nepute JC, Hayek ME, Albert SG (2016) Adrenal incidentalomas: clinical controversies and modified recommenda- tions. AJR Am J Roentgenol 206:1170-1178.
26. Sherlock M, Scarsbrook A, Abbas A et al (2020) Adrenal inciden- taloma. Endocr Rev 41:775-820.
27. Bilbey JH, McLoughlin RF, Kurkjian PS et al (1995) MR imag- ing of adrenal masses: value of chemical-shift imaging for dis- tinguishing adenomas from other tumors. AJR Am J Roentgenol 164:637-642.
28. Brandao LF, Autorino R, Laydner H et al (2014) Robotic versus laparoscopic adrenalectomy: a systematic review and meta-anal- ysis. Eur Urol 65:1154-1161.
29. Economopoulos KP, Mylonas KS, Stamou AA et al (2017) Lapa- roscopic versus robotic adrenalectomy: a comprehensive meta- analysis. Int J Surg 38:95-104.
30. Kenney PJ, Wagner BJ, Rao P, Heffess CS (1998) Myelolipoma: CT and pathologic features. Radiology 208:87-95.
31. Duffy S, Jhaveri M, Scudierre J, Cochran E, Huckman M (2005) MR imaging of a posterior mediastinal ganglioneuroma: fat as a useful diagnostic sign. AJNR Am J Neuroradiol 26:2658-2662.
32. Guan YB, Zhang WD, Zeng QS, Chen GQ, He JX (2012) CT and MRI findings of thoracic ganglioneuroma. Br J Radiol 85:e365-372.
33. Lam AK (2017) Lipomatous tumours in adrenal gland: WHO updates and clinical implications. Endocr Relat Cancer 24:R65-79.
34. Ranathunga DS, Cherpak LA, Schieda N, Flood TA, McInnes MD (2020) Macroscopic fat in adrenocortical carcinoma: a systematic review. AJR Am J Roentgenol 214:390-394.
35. Song JH, Chaudhry FS, Mayo-Smith WW (2008) The incidental adrenal mass on CT: prevalence of adrenal disease in 1,049 con- secutive adrenal masses in patients with no known malignancy. AJR Am J Roentgenol 190:1163-1168.
36. Sangwaiya MJ, Boland GWL, Cronin CG et al (2010) Inciden- tal adrenal lesions: accuracy of characterization with contrast- enhanced washout multidetector CT-10-minute delayed imaging protocol revisited in a large patient cohort. Radiology 256:504-510.
37. Expert Panel on Urological Imaging, Mody RN, Remer EM, Nikolaidis P et al (2021) ACR Appropriateness Criteria® Adrenal Mass Evaluation: 2021 Update. J Am Coll Radiol. 18:S251-S267.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.