ELSEVIER
Surgery
journal homepage: www.elsevier.com/locate/surg
SURGERY
NOWEMBER 2018
$
=
Differentiating between adrenocortical carcinoma and lipid-poor cortical adenoma: A novel cross-sectional imaging-based score
Tal Yalon, MDª,*, Mariana Yalon, MDb, Dan Assaf, MDC, Karina Lenartowiczª, Trenton Foster, MDª, Melanie Lyden, MDª, Benzon Dy, MDª, Irina Bancos, MDª, Travis Mckenzie, MDª
a Endocrine Surgery, Mayo Clinic, Rochester, MN
b CT Clinical Innovation Center, Department of Radiology, Mayo Clinic, Rochester, MN
” Chaim Sheba Medical Center, Tel Hashomer, Ramat Gan, Israel
d Division of Endocrinology, Mayo Clinic, Rochester, MN
Check for updates
ARTICLE INFO
Article history: Accepted 19 July 2022 Available online 14 October 2022
ABSTRACT
Background: Discrimination between adrenocortical carcinoma and lipid-poor cortical adenoma preop- eratively is frequently difficult as these two entities have overlapping imaging characteristics. Differenti- ation will allow for the selection of the most appropriate operative approach and may help prevent over-treatment. We aimed to identify imaging features that could preoperatively differentiate adreno- cortical carcinoma from lipid-poor cortical adenoma and use them in a novel imaging-based score. Methods: We conducted a retrospective analysis of patients with pathologically proven adrenocortical carci- noma and lipid-poor cortical adenoma who underwent resection in a single tertiary referral center between March 1998 and August 2020. The inclusion criteria were diameter >1 cm, attenuation >10 Hounsfield units on nonenhanced computed tomography, and histopathologic diagnosis. Patients with metastatic or locally advanced adrenocortical carcinoma adenoma (stages 3-4) were excluded. We developed a score using binary logistic multivariate regression model in 5-fold derivation (~70%) cohorts with stepwise backward conditional regression as feature selection. Standardized mean regression weight was used as variable score points.
Results: We identified 232 adrenals resected across 211 patients. By comparing the imaging character- istics of adrenocortical carcinoma (n = 56) and lipid-poor cortical adenoma (n = 156), we revealed statistically significant differences between the groups in 9 parameters: size, attenuation, thin and thick rim enhancement patterns, heterogeneity, calcification, necrosis, fat infiltration, and lymph node prominence. The score mean performance was 100% sensitivity for the exclusion of adrenocortical car- cinoma, 80% specificity (95% confidence interval, 68.3-91.5), 66% positive predictive value (95% confi- dence interval, 52.3-78.7), and 100% negative predictive value with area under the curve of 0.974. Conclusion: We defined and evaluated a novel 9-variable, imaging-based score. This score outperformed any single variable and could facilitate safe preoperative discrimination of adrenocortical carcinoma and lipid-poor cortical adenoma.
@ 2022 Elsevier Inc. All rights reserved.
Introduction
Adrenal incidentalomas constitute a common clinical dilemma, with a prevalence of 4% to 7% reported on radiological series of
Tal Yalon and Mariana Yalon contributed equally to this article.
Presented at of the 42nd Annual Meeting of the American Association of Endocrine Surgeons, May 22-24, 2022, Cleveland, OH (podium oral presentation)
* Reprint requests: Tal Yalon, MD, General and Endocrine Surgery, Mayo Clinic, Rochester, MN E-mail address: yalon.tal@mayo.edu (T. Yalon);
Twitter: @tal_yalon, @YalonMariana
patients >40 years, increasing ≤10% in older age groups.1-4 Upon encountering an adrenal mass, a physician must consider whether the tumor is secretory and determine its malignant potential. Although the presence of excess hormonal production can be detected by a series of laboratory investigations, discriminating between a benign and malignant lesion can be challenging. This is especially true for differentiating adrenocortical carcinoma (ACC), a rare and aggressive primary adrenal cancer,5-7 from lipid-poor cortical adenoma (LPCA), which is a benign lesion of the adrenal gland. With an overall 5-year survival of 45%, the ideal treatment for ACC is early diagnosis and en bloc surgical resection of the
tumor, followed by systemic treatment as indicated by stage and completeness of resection.4,7 Overestimation of the malignancy risk of LPCA is a common clinical dilemma and often results in unnec- essary resections and overtreatment. It is therefore critical to be able to differentiate these diagnoses in order to provide optimal surgical treatment for ACC while avoiding overtreatment for benign tumors.4 Preoperative laboratory evaluation for hormonal secretion and steroid metabolites through blood or urine analysis may pro- vide some evidence for or against malignancy but often remain inconclusive, especially in smaller tumors.6,8 To date, radiologic evaluation by means of cross-sectional imaging remains critical for characterizing adrenal tumors as benign or suspicious for malig- nancy. Adrenal lesions with homogenous appearance, with lipid reach composition, and demonstrated by attenuation ≤10 Houns- field units (HU) on noncontrast computed tomography (NCCT) are clearly benign and, similarly, signal drop on opposed phase with chemical shift (CS) magnetic resonance imaging (MRI) is also indicative of a lipid-rich lesion and consistent with a benign ade- noma, although the level of evidence on this is low.9-14 On the other hand, discriminating between ACC and LPCA remains chal- lenging as both lesions can have a density >10 HU on NCCT and can have indeterminant characteristics on CS MRI analysis.15,16 The role of fludeoxyglucose F 18 positron emission tomography-CT in the diagnosis of primary adrenal malignancy remains limited.4
Preoperative differentiation between ACC and LPCA based on imaging characteristics may allow for the selection of the most appropriate surgical approach and may help prevent overtreatment in selected cases. In this study, we aimed to identify imaging fea- tures that could preoperatively differentiate ACC from LPCA and use these features in a novel cross-sectional imaging-based score.
Materials and Methods
Study population
A retrospective review of a prospectively maintained adrenal database performed at a single tertiary referral center, Mayo Clinic, Rochester, MN, was performed. The study was conducted with approval by the Institutional Review Board of the Mayo Clinic. We reviewed the electronic medical records for all patients who un- derwent open, laparoscopic, or robotic-assisted adrenalectomy with ACC or LPCA reported on final pathology between March 1998 and August 2020. Study inclusion criteria were age ≥18 years old, adrenal masses >1 cm in largest diameter, attenuation >10 HU on NCCT in the LPCA group, and a conclusive histopathologic diagnosis after surgical resection. Patients with known locally advanced or metastatic ACC (stages 3-4) were excluded. As adrenal metastases are a heterogeneous group of diseases, each with its own unique characteristics, we chose to focus on the primary adrenal tumors mentioned above.
Clinical data
Demographic patient data recorded were date of birth, age, and sex. Other data collected included the presence or absence of a clinical endocrine syndrome at presentation, endocrine laboratory work-up panel when available, preoperative imaging data, and the laterality of adrenal mass or masses. Operative details recorded include the surgical date and laterality. Final pathology reports were also reviewed and documented. In addition, patient records were screened for the presence of other malignancies, including the primary tumor location and tumor oncological status.
Image acquisition technique
The imaging studies were performed over a 22-year period us- ing a variety of scanners, platforms, and scanning protocols. How- ever, CT sections through the upper abdomen were obtained with 3- or 5-mm thickness. Post-contrast-enhanced portal venous phase and delayed-phase CT images when available were obtained at 65 to 70 seconds and 10 to 15 minutes, respectively. MRI studies were acquired on variable platforms with variable protocol sequences.
Image analysis
All imaging studies were reviewed by an abdominal radiologist. The adrenal lesions and their surrounding retroperitoneal struc- tures were evaluated for tumor spread. Adrenal masses in this cohort were characterized using qualitative and quantitative im- aging parameters. The examined parameters were: 1) size (the single largest lesion diameter recorded on the axial plane). 2) change in size, when available. 3) attenuation on NCCT (determined by placing the region of interest centrally, to include 75% of a lesion, avoiding lesion edges and tissues beyond the adrenal mass); this measurement was performed only in homogenous masses. If a lesion had heterogeneous attenuation on NCCT, it was termed as such, without HU measurements. 4) Intralesional calcifications (determined as the presence of calcific component with “attenua- tion similar to the attenuation of other calcified or osseous struc- tures in the same data set,” single or multiple, and of any size or shape)17. 5) macroscopic fat (determined as the presence of intra- lesional low attenuating component). 6) cystic change (low atten- uating, nonenhancing component, subjectively demonstrating fluid density,18 occupying a majority of the mass). 7) To determine lack of enhancement, a cutoff of 10-HU difference between the non- enhanced and the contrastenhanced scan was used. 8) Evidence of necrosis (similar to the definition of cystic change, the presence of low attenuating nonenhancing component; but here, the compo- nent is relatively smaller and less well defined). However, cystic change and necrosis share common imaging findings, limiting the differentiation between the two. 9) differentiation between sharp or nonsharp (infiltrative) margins and distinguishing between globular shape and irregular lesion shape, with globular configu- ration signifying oval or rounded lesions.
Contrast-enhanced studies were evaluated in the portal venous imaging phase, to generate the following 6 distinct enhancement patterns (1) hairline-thin rim enhancement with enhancing inter- nal structures (some appearing as blood vessels traversing the lesion) (Figure 1); (2) diffusely heterogenous (Figures 2 and 3); (3) heterogenous with well-demarcated internal components (Figure 4); (4) homogenous mild enhancement (Figure 5); (5) ho- mogenous significant enhancement (Figure 6); and (6) heteroge- nous thick rim enhancement with a central hypodense nonenhancing component (Figures 7 and 8).
Postcontrast enhancement characteristics were evaluated subjectively on the portal venous phase, for heterogeneity and magnitude of enhancement (significant or mild). Delayed enhancement was defined as subjective lesion attenuation in the delayed imaging phase. For the adrenal washout imaging, both the absolute washout percentage (AWP) and the relative washout percentage (RWP) were calculated, depending on the availability of unenhanced imaging phase. The calculated AWP and RWP values suggestive of adrenal adenoma were >60% and >40%, respectively.19
Retroperitoneal tissue immediately surrounding the mass was examined for fat stranding, evidence of prominent lymph nodes (subjective evaluation of lymph nodes short and long axis), and
radiologic signs suggesting adjacent organ involvement by a direct extension (determined by close proximity between the mass and the organ and by the impression of the mass). When available, focal parenchymal perfusion difference within the adjacent organ was considered positive for involvement.
The following lesional features were evaluated on MRI when available: size, heterogeneity, intralesional macroscopic fat, cystic component, and hemorrhage. For measurement of intracellular lipid, both qualitative and quantitative analyses on CS MRI were used, with the use of visual estimation for the signal drop and calculation of signal intensity index (SII) and adrenal-to-spleen ratio (ASR), when available.
Statistical analysis
Data analysis was performed using IBM SPSS, version 27 (SPSS IBM, Inc, Armonk, NY) software with 3-sided significance level of
a = 0.05. Descriptive statistics are presented using prevalence and percentage values for the categorical variables, whereas the continuous variables are presented with means and SD and the skewed distributed variables are presented by median and range. Group comparisons were tested using Student’s t test for continuous normally distributed variables and the Mann- Whitney U test for nonparametric comparisons. Categorical comparisons were tested using the x2 analysis or the Fisher exact test, as appropriate.
The ACC score components were selected using a univariate binary logistic regression model, identifying all features associated with ACC with P <0.1. The ACC score was developed using a binary logistic multivariate regression model in 5-fold derivation (~70%) cohorts with stepwise backward conditional regression as feature selection. Standardized mean regression weight was used as vari- able score points. The score was validated using 5-fold cross vali- dation on the test cohorts (~30%). The performance parameters were computed, including sensitivity, specificity, positive predic- tive value (PPV), negative predictive value (NPV), and accuracy. Receiver operating characteristic (ROC) analysis was used to eval- uate the associations by calculating the ROC area under the curve (AUC).
Results
A total of 232 adrenal masses meeting inclusion criteria were resected from 211 patients between March 1998 and April 2020. All masses had an NCCT density >10 HU; 42 tumors were bilateral (18%), 106 left sided (46%), and 84 right sided (36%). Patients were divided into 2 groups based on final surgical pathology: ACC group (n = 56) and LPCA group (n = 155) (Figure 9).
The mean age in the ACC group was 67 vs 57 years in the ACC versus LPCA group; 35 patients (63%) were female in the ACC group vs 118 patients (67%) in the LPCA group (P = . 46). History of other malignancy was present in 13 patients (23%) of the ACC group and in 50 patients (32%) in the LPCA group (P = . 53). None of the pa- tients had an active known malignancy. No patient presented with bilateral adrenal masses in the ACC group.
A
B
C
D
E
Laboratory hormonal profile
There were significant differences between groups regarding hormone secretion. The ACC group had higher levels of dehydro- epiandrosterone sulfate (366.2 vs 52.32 µg/dL; P =. 011) and higher morning serum cortisol (73 vs 57.6 µg/dL; P = . 035). There were no significant differences between groups in comparing 24-hour urine cortisol, serum aldosterone, low-dose dexamethasone suppression test (morning cortisol levels), and corticotropin (P > .5).
Radiological findings
Tumors in the ACC group were larger than those in the LPCA group (95 mm [28-262 mm] vs 34 mm [10-95 mm]; P < . 0001). The ACC group had higher average tumor attenuation values on
NCCT (33.5 vs 20.7 HU; P < . 0001) and had significantly lower AWP values (31.2% vs 49.1%; P = . 005).
The ACC group had higher rates of significant hyperenhance- ment (54.2% vs 29.5%; P =. 002), enhancement patterns (AUC = 0.8; P < . 0001), heterogeneity on NCCT (79.1% vs 26.9%, respectively; AUC = 0.76; P < . 0001), postcontrast heterogeneity (98% vs 53.6%, respectively; AUC = 0.72; P < . 0001), tumor calcification (39.6% vs 15.5%, respectively; AUC = 0.621; P <. 0001), tumor necrosis (76% vs 0.57%, respectively; AUC = 0.88; P < . 0001), nonsharp margins (13.2% vs 1.1%, respectively; AUC = 0.56; P = . 001), irregular shape (79.2% vs 29.3%, respectively; AUC = 0.75; P < . 0001), fat infiltration (58.5% vs 4.6%, respectively; AUC = 0.77; P < . 0001), lymph node prominence (41.5% vs 3.4%, respectively; AUC = 0.69; P < . 0001), and radiologic suggestion of adjacent organ involvement (37.5% vs 0, respectively; AUC = 0.69; P < . 0001).
There was no difference between the groups in delayed enhancement (AUC = 0.57; P = . 08), macroscopic fat (AUC = 0.54; P =. 12), or presence of cystic component (AUC = 0.52; P = . 31).
For a total of 47 adrenal lesions (33 LPCAs and 14 ACCs), non- dedicated MRI studies were available. Size, heterogeneity, evidence of intralesional macroscopic fat, and cystic component corre- sponded to the observations on CT when a comparison study was available. Hemorrhage was observed in 3 LPCAs and 11 ACCs. In the LPCA group, the presence of intracellular fat was detected on visual estimation in 24 cases, whereas, when using the quantitative methods, it was detected in 16 cases with ASR and in 22 cases with SII. Evaluation for intracellular fat in the ACC group was partial because of mass heterogeneity and was available for a single ACC, indicating the presence of intracellular fat on both qualitative and quantitative measurements.
Total patients (n=211)
Adrenal tumors (n=232)
ACC group (stage 1 and 2) (n=56)
Non ACC group (LPCA) (n=155)
Predicting adrenocortical carcinoma
Binary logistic regression univariate analysis
Univariate analysis revealed the following variables to be significantly associated with ACC diagnosis: tumor size (odds ratio [OR] = 1.085; 95% CI, 1.06-1.11; P < . 0001), tumor attenuation (OR = 1.115; 95% CI, 1.07-1.16; P < . 0001), significant tumor hyperenhancement (OR = 2.82; 95% CI, 1.45-5.5; P = . 002), het- erogeneity (OR = 42.35; 95% CI, 5.69-314.63; P < . 0001), tumor calcification (OR = 3.57; 95% CI, 1.79-7.09; P < . 0001), tumor radiologic necrosis (OR = 547.83; 95% CI, 69.13-4341.457; P < . 0001), sharp margins (OR = 0.076; 95% CI, 0.015-0.38; P = .002), fat infiltration (OR = 31.43; 95% CI, 12.8-77.15; P < . 0001), and lymph node prominence (OR = 21.47; 95% CI, 8.07-57.13; P < . 0001).
Creating the adrenocortical carcinoma score
Using ~70% derivation by 5-fold cross validation (n range = 152-166), including only the variables associated with ACC (P < . 1), a binary logistic multivariate regression model was used in the 5- fold derivation cohorts with stepwise backward conditional regression as feature selection. When selecting 11 variables, with macroscopic fat and sharp margin excluded due to inconsistency contribution in the validation cohorts, 9 variables were consistent in their contribution to the 5-fold cohorts.
A 9-variable score was generated based on the mean score of each variable in the 5-fold derivation cohorts, yielding a score ranging from -9 to 22 points, calculated by size >4 cm (+2 points), attenuation >20 HU (+1 point), peripheral septal enhancement pattern (-4 points), heterogenous thick peripheral with central hypodensity enhancement pattern (+5 points), heterogeneity (+1 point), calcification (-5 points), the presence of radiologic necrosis (+8 points), tumor fat infiltration (+3 points), and lymph node prominence (+2 points) (Table I).
Testing the adrenocortical carcinoma score performance
A sensitivity of 100% was selected as the cutoff criterion through all the 5-fold derivation cohorts, yielding a mean cutoff of 3 (95% CI, 1.34-4.66). The mean performances of the 5-fold derivation cohort were 100% sensitivity as was defined, 79.94% specificity (95% CI, 68.3-91.5), 65.55% PPV (95% CI, 52.3-78.7), and 100% NPV, with an AUC of 0.974 (Table II).
| 1 | Attenuation >10 HU | 1 point |
| 2 | Heterogeneity | 1 point |
| 3 | Size >4 cm | 2 points |
| 4 | Lymph node prominence | 2 points |
| 5 | Tumor fat infiltration | 3 points |
| 6 | Heterogenous thick peripheral with central hypodensity enhancement pattern | 5 points |
| 7 | Presence of radiologic necrosis | 8 points |
| 8 | Peripheral septal enhancement pattern | -4 points |
| 9 | Calcifications | -5 points |
HU, Hounsfield unit.
| Score | Sensitivity (FN) | Specificity (FP) | PPV | NPV |
|---|---|---|---|---|
| <1 | 100 | 50.6 (89) | 37.8 | 100 |
| <2 | 100 | 76.7 (41) | 56.8 | 100 |
| <3 | 100 | 86.9 (23) | 70.1 | 100 |
| <4 | 94.4 (3) | 93.7 (11) | 82.3 | 98.2 |
| <5 | 83.3 (9) | 97.7 (4) | 91.8 | 95.03 |
| <6 | 79.6 (11) | 99.4 (1) | 97.7 | 94.09 |
| <7 | 72.2 (15) | 99.4 (1) | 97.5 | 92.1 |
| <8 | 70.4 (16) | 100 | 100 | 91.7 |
| <9 | 70.4 (16) | 100 | 100 | 91.7 |
| <10 | 64.8 (19) | 100 | 100 | 90.3 |
| <11 | 61.1 (21) | 100 | 100 | 89.3 |
| <12 | 57.4 (23) | 100 | 100 | 88.4 |
| <13 | 53.7 (25) | 100 | 100 | 87.6 |
| <14 | 50 (27) | 100 | 100 | 86.7 |
| <15 | 38.9 (33) | 100 | 100 | 84.2 |
| <16 | 37 (34) | 100 | 100 | 83.8 |
| <17 | 33.3 (36) | 100 | 100 | 83.02 |
| <18 | 20.4 (43) | 100 | 100 | 80.4 |
| <19 | 18.5 (44) | 100 | 100 | 80 |
| <20 | 11.1 (48) | 100 | 100 | 78.6 |
| <21 | 7.4 (50) | 100 | 100 | 77.9 |
| <22 | 3.7 (52) | 100 | 100 | 77.2 |
FN, false-negative value; FP, false-positive value; NPV, negative predictive value; PPV, positive predictive value.
We compared the ACC score performance by the implementa- tion of the common cutoffs or classification criteria to our cohort and compared all the performance measures.
Tumor necrosis as a single classification variable scored sensi- tivity of 76%, specificity of 99.4%, PPV of 97.4%, NPV of 93.5%, and 0.877 AUC. Tumor size >4 cm scored sensitivity of 88.9%, specificity of 74.4%, PPV of 51.6%, NPV of 95.6%, and 0.817 AUC. Attenuation >20 HU scored sensitivity of 71.9%, specificity of 92.8%, PPV of 50.9%, NPV of 96.7%, and 0.719 AUC (Figure 10).
Discussion
The preoperative need to determine the malignancy potential of an adrenal mass is a common clinical dilemma. This dilemma is especially apparent when trying to discriminate between benign LPCA and ACC, due to their overlapping radiological features.
Benign nonsecretory adrenal masses do not require resection, whereas, when there is a suspicion of ACC, the tumor should be approached in a manner that ensures complete resection without disruption, generally requiring open adrenalectomy.4,15 Accurate preoperative differentiation between these 2 pathologies may help determine the need for surgical intervention and guide toward the most appropriate surgical approach.
Chart Title
ACC score
Necrosis
Size>4
Attenuation > 20
AUC
1
0.8
0.6
NPV
0.4
sensitivity
0.2
0
PPV
specificity
To date, there is no known single radiographic feature that can consistently differentiate ACC without clear invasive features or metastasis from LPCA. We, therefore, aimed to define radiologic features observed in a pathologically confirmed large adrenal lesion cohort that could potentially be combined to produce a scoring algorithm differentiating ACC from LPCA better than any individual feature.
In previous studies, association with malignancy was found to correspond with lesion attenuation on NCCT, lesion size, and washout characteristics on a dedicated adrenal protocol.12,16,20-22
In recent studies, the presence of tumor necrosis was evaluated as a predictor for ACC. Garay-Lechuga et al,23 in their cohort of 18 ACC of all stages and 41 adrenocortical adenomas, investigated the diagnostic efficiency of the presence of necrosis and other radio- logical features of CT scan that are related to ACC. In their study, they found that the presence of necrosis was the single most important feature significantly associated with ACC, whereas Thomas et al,24 in their study investigating the CT features of large (>4 cm) adrenal tumors with 31 ACCs of various stages and 25 adrenocortical adenomas, found necrosis as a less reliable predictor of malignancy.
Among the multiple tumor imaging characteristics observed in our cohort, we found radiological evidence of necrosis to be a very strong predictor for ACC, which was present in only 1 LPCA (0.6%) as opposed to 38 ACCs (68%).
In this study, we compared the preoperative adrenal mass im- aging characteristics of 56 patients with histopathologically proven stages 1 and 2 ACCs and 155 patients with LPCA. In addition to the multitude of morphologic tumor characteristics evaluated in our cohort, we have introduced the concept of evaluating adrenal lesion enhancement, generating 6 distinct patterns of enhancement. Our results found a set of 9 statistically significant radiological param- eters in ACC and LPCA. Although radiological features, such as attenuation >10 HU, heterogeneity, size >4 cm, lymph node prominence, peritumoral fat stranding, heterogenous thick rim enhancement pattern with a central hypodense nonenhancing component, and the presence of radiological necrosis were all associated with a higher risk of malignancy with various degrees of statistical individual strength, the presence of the hairline-thin rim with enhancing internal structures enhancement pattern and calcification were found to be predictors of benignity (Table I). By combining each of the radiological features’ individual statistical strengths, we were able to use them in a novel score, ranging be- tween -9 to +22, that is able to differentiate between benign and
malignant lesions based on preoperative abdominal CT. In order to avoid underdiagnosis of ACC, we set the bar to a 100% sensitivity (score of 3), which resulted in a specificity of 79.94%, PPV of 65.55%, and NPV of 100%. In simpler terms, by using this scoring system in our present cohort, 31 patients (20%) out of the 155 patients from the LPCA arm could have been safely spared unnecessary adrenal- ectomy, without the risk of missing even a single ACC.
Our study had several inherent limitations to be acknowledged. First, the retrospective nature and the rarity of the investigated pathology contributed to nonuniform imaging study protocols and hence incomplete imaging data. Moreover, because CT is the most used imaging modality for the evaluation of adrenal masses, data acquired on MRI were frequently incomplete, not allowing for the performance of a comparison of diagnostic accuracy between the two modalities. Prospective studies with dedicated uniform data acquisition on both cross-sectional imaging modalities would be thus beneficial. Second, the radiological evaluation of our cohort of adrenal masses was performed by a single, nonblinded abdominal radiologist, who might have increased the chance for an observer bias. Our results hence merited prospective validation and inter- observer agreement evaluation. From a hormonal evaluation standpoint, there is variability in how adrenal masses are evalu- ated, depending on institutional and provider preferences. Given this, not all patients have the same preoperative laboratory, which limited our ability to obtain a meaningful analysis of this data. Lastly, our cohort did not include less common adrenal lesions that share similar imaging characteristics, such as ganglioneuromas; thus, this scoring system could not be applied to such lesions.
In conclusion, we defined and evaluated a multitude of imaging features of adrenal masses in a large cohort. Nine of these variables were combined to generate an imaging-based ACC exclusion score. This score outperformed any single variable and presents a low-risk model with a 0 false-positive rate that can facilitate preoperative discrimination of ACC and LPCA. This scoring system has preoper- ative implications and can potentially be used along with other clinical data to aid in the selection of surgical approaches and may help prevent unnecessary surgery in select patients.
Funding/Support
This research did not receive any specific funding from any agencies in the public, commercial, or not-for-profit areas.
Conflict of interest/Disclosure
The authors have no conflicts of interests or disclosures to report.
References
1. Ebbehoj A, Li D, Karu RJ, et al. Epidemiology of adrenal tumours in Olmsted County, Minnesota, USA: a population-based cohort study. Lancet Diab Endo- crinol. 2020;8:894-902.
2. Reimondo G, Castellano E, Grosso M, et al. Adrenal incidentalomas are tied to increased risk of diabetes: findings from a prospective study. J Clin Endocrinol Metab. 2020; 105:e974-981.
3. Song JH, Chaudhry FS, Mayo-Smith WW. The incidental adrenal mass on CT: prevalence of adrenal disease in 1,049 consecutive adrenal masses in patients with no known malignancy. AJR Am J Roentgenol. 2008; 190:1163-1168.
4. Fassnacht M, Dekkers OM, Olaf M, et al. European Society of Endocrinology Clinical Practice Guidelines on the management of adrenocortical carcinoma in adults, in collaboration with the European Network for the Study of Adrenal Tumors. Eur J Endocrinol. 2018;179:G1-G46.
5. Sharma E, Dahal S, Sharma P, et al. The characteristics and trends in adreno- cortical carcinoma: a United States population based study. J Clin Med Res. 2018;10:636-640.
6. Bancos I, Taylor AE, Chortis V, et al. Urine steroid metabolomics for the dif- ferential diagnosis of adrenal incidentalomas in the EURINE-ACT study: a prospective test validation study. Lancet Diab Endocrinol. 2020;8:773-781.
7. James BC, Aschebrook-Kilfoy B, Cipriani N, Kaplan EL, Angelos P, Grogan RH. The incidence and survival of rare cancers of the thyroid, parathyroid, adrenal, and pancreas. Ann Surg Oncol. 2016;23:424-433.
8. Hines JM, Bancos I, Bancos C, et al. High-resolution, accurate-mass (HRAM) mass spectrometry urine steroid profiling in the diagnosis of adrenal disorders. Clin Chem. 2017;63:1824-1835.
9. Boland GW, Lee M, Gazelle GS, Halpern EF, McNicholas MM, Mueller PR. Characterization of adrenal masses using unenhanced CT: an analysis of the CT literature. AJR Am J Roentgenol. 1998;171:201-204.
10. Israel GM, Korobkin M, Wang C, Hecht EN, Krinsky GA. Comparison of unen- hanced CT and chemical shift MRI in evaluating lipid-rich adrenal adenomas. AJR Am J Roentgenol. 2004;183:215-219.
11. Korobkin M, Giordano TJ, Brodeur FJ, et al. Adrenal adenomas: relationship between histologic lipid and CT and MR findings. Radiology. 1996;200: 743-747.
12. Lee MJ, Hahn PF, Papanicolaou N, Egglin TK, Saini S, Mueller PR, Simeone JF. Benign and malignant adrenal masses: CT distinction with attenuation co- efficients, size, and observer analysis. Radiology. 1991;179:415-418.
13. Ream JM, Gaing B, Mussi TC, Rosenkrantz AB. Characterization of adrenal le- sions at chemical-shift MRI: a direct intraindividual comparison of in- and opposed-phase imaging at 1.5 T and 3 T. AJR Am J Roentgenol. 2015;204: 536-541.
14. Dinnes J, Bancos I, Di Runnano LF, et al. Management of endocrine disease: Imaging for the diagnosis of malignancy in incidentally discovered adrenal masses: a systematic review and meta-analysis. Eur J Endocrinol. 2016;175: R51-R64.
15. Fassnacht M, Arit W, Bancos I, et al. Management of adrenal incidentalomas: European Society of Endocrinology Clinical Practice Guideline in collaboration with the European Network for the Study of Adrenal Tumors. Eur J Endocrinol; 016;175:G1-G34.
16. Schieda N, Siegelman ES. Update on CT and MRI of adrenal nodules. AJR Am J Roentgenol. 2017;208:1206-1217.
17. Davenport MS, Neville AM, Ellis JH, Cohan RH, Chaudhry HS, Leder RA. Diag- nosis of renal angiomyolipoma with hounsfield unit thresholds: effect of size of region of interest and nephrographic phase imaging. Radiology. 2011;260: 158-165.
18. Corwin MT, Mitchell AS, Wilson M, Campbell MJ, Fananapazir G, Loehfelm TW. Accuracy of focal cystic appearance within adrenal nodules on contrast- enhanced CT to distinguish pheochromocytoma and malignant adrenal tu- mors from adenomas. Abdom Radiol. 2021;46:2683-2689.
19. Pena CS, Boland GW, Hahn PF, Lee MJ, Mueller PR. Characterization of inde- terminate (lipid-poor) adrenal masses: use of washout characteristics at contrast-enhanced CT. Radiology. 2000;217:798-802.
20. Mantero F, Terzolo M, Arnaldi G, et al. A survey on adrenal incidentaloma in Italy. Study Group on Adrenal Tumors of the Italian Society of Endocrinology. J Clin Endocrinol Metab. 2000;85:637-644.
21. Iñiguez-Ariza NM, Kohlenberg JD, Delivanis DA, et al. Clinical, biochemical, and radiological characteristics of a single-center retrospective cohort of 705 large adrenal tumors. Mayo Clin Proc Innov Qual Outcomes. 2018;2:30-39.
22. Mayo-Smith WW, Song JH, Boland GL, et al. Management of incidental adrenal masses: a white paper of the ACR incidental findings committee. J Am Coll Radiol. 2017;14:1038-1044.
23. Garay-Lechuga D, Pérez-Soto RH, Hernández-Acevedo J, et al. Computed to- mography (CT) scan identified necrosis, but is it a reliable single para.meter for discerning between malignant and benign adrenocortical tumors? Surgery. 2022;171:104-110.
24. Thomas AJ, Habra MA, Bhosale PR, et al. Interobserver agreement in dis- tinguishing large adrenal adenomas and adrenocortical carcinomas on computed tomography. Abdom Radiol. 2018;43:3101-3108.
Discussion
Check for updates
Dr. Barbra Miller (Columbus, OH): Many of the factors that you’ve listed involve interpretation and are subjective, such as shape, margins, heterogeneity, fat infiltration. This is similar to the American Thyroid Association criteria and Thyroid Imaging Reporting and Data Systems when looking at thyroid nodules. Have you looked at any measures that can be characterized in an objective fashion? There are the terms such as sphericity undula- tion, eccentricity, entropy, and/or compactness?
Dr. Tal Yalon: Thank you for this very valid question. Indeed, one of the limitations of this study is the fact that it is based on a single abdominal radiologist’s observations which can be identified as subjective analysis. We did try to define those parameters in a more objective manner which is elaborated in more in our paper how- ever I don’t think this time frame is enough to address this matter.
All the radiological parameters in our cohort are defined very rigidly in order to try to eliminate the subjectivity of a single operator.
Dr. Quan Duh (San Francisco, CA): Very nice study. The current standard for radiological diagnosis is to check Hounsfield unit first, then if it is higher than 10 the next step is to check washout. I saw that you have the washout data, but didn’t use them for your evaluation at the end. How much improvement is there beyond using the washout data?
Dr. Tal Yal: I think that the main component of this score is the synergetic effect that we have between different parameters. Even though washout was statistically significant as a single parameter when we tried to investigate the differences between the groups when we pulled it into the same score eventually the
math didn’t really add up. It wasn’t as strong enough as the other parameters. I think interestingly, for example, we do rely quite often on the household units in order to try to differen- tiate between cancer or any other suspicious lesion, in our weighted parameters this was statistically significant, but it had the lowest strength of significance as opposed for example to necrosis that had eight times more predictive value towards an adrenocortical carcinoma. I think the main component of our study and what I think is a bit more novel are the enhancement characteristics rather than the washout itself.
Dr. Dhaval Patel (Edison, New Jersey): Do you have any data on hormonal evaluation and your scoring system? And can you comment on what impact this would have had on the rate of surgery?
Dr. Tal Yalon: In terms of impact, if we would have utilized this score, we would have spared 20% of the unnecessary adrenalec- tomies for benign disease. So in our cohort, it would be something like 30 patients which I think is quite substantial, preventing this amount of unnecessary surgeries. As for your first question regarding the hormonal evaluation, we did have quite a lot of data in regard to hormonal evaluation unfortunately, it was a bit lacking. I think that mainly since we were looking at a period of 20 years and there is quite an operator variability in terms of ordering those tests, so it wasn’t as strong as a predictor as we thought it going to be, or we didn’t have significant enough of data in order to use it. This is the reason why eventually we chose to focus our attention on the imaging findings, which were very consistent. In other words, 100% of our patients did have this data as opposed to about 60% that had full of hormonal workup.