ANZJSurg.com
Predicting malignancy in adrenal incidentaloma and evaluation of a novel risk stratification algorithm
Elizabeth Foo,* Robin Turner,t Kuan-Chi Wang,* Adam Aniss,* Anthony J. Gill,#§ Stanley Sidhu,* Roderick Clifton-Bligh1 and Mark Sywak*
*Endocrine Surgical Unit, The University of Sydney, Sydney, New South Wales, Australia
tSchool of Public Health, The University of New South Wales, Sydney, New South Wales, Australia
Department of Anatomical Pathology, Royal North Shore Hospital, Sydney, New South Wales, Australia
§ Cancer Diagnosis and Pathology Research Group, Kolling Institute of Medical Research, Sydney, New South Wales, Australia and
1Department of Endocrinology, Royal North Shore Hospital, Sydney, New South Wales, Australia
Key words
adrenal incidentaloma, adrenocortical carcinoma, laparoscopic adrenalectomy.
Correspondence
Clinical Associate Professor Mark Sywak, Endocrine Surgical Unit, The University of Sydney, AMA House Suite 202, 69 Christie Street, St Leonards, NSW 2065, Australia. Email: marksywak@nebsc.com.au
E. Foo MBBS; R. Turner MD; K .- C. Wang MBBS; A. Aniss PhD; A. J. Gill MD; S. Sidhu PhD; R. Clifton-Bligh PhD; M. Sywak MMedSci (ClinEpi).
This study was presented at the RACS Annual Scientific Congress, May 2015, Perth.
Accepted for publication 1 November 2016.
doi: 10.1111/ans.13868
Abstract
Background: Incidentally discovered adrenal lesions known as adrenal incidentalomas (AI) are being encountered with increasing frequency due to the widespread use of abdomi- nal computed tomography (CT). The aim of this study was to identify the clinical predictors of malignancy in AI and to evaluate the accuracy of a recently proposed risk stratification algorithm.
Methods: A retrospective analysis of 96 patients presenting with AI between 2004 and 2014 was undertaken; 66 patients underwent adrenalectomy, and 30 were managed non- operatively. Univariate analysis including patient demographics, CT features of tumour size, density and heterogeneity was performed. Hormonal parameters including 24-h urinary-free cortisol and serum dehydroepiandrosterone sulphate (DHEAS) were also included. A Cleve- land Clinic risk stratification model utilizing adrenal size and density was evaluated.
Results: The overall rate of malignancy was 8%. On univariate analysis, the following pre- operative variables were predictive of malignancy - tumour size on pathology (P = 0.0031) and CT (P = 0.0016), heterogeneity on CT imaging (P = 0.0036), a relative percentage washout of less than 40% (P = 0.0178), elevated 24-h urinary-free cortisol levels (P = 0.0176), elevated DHEAs (P = 0.0061) and younger age at presentation (P < 0.0001). Evaluation of the Cleveland Clinic algorithm found an area under the receiver operating characteristic curve of 0.81 (95% confidence interval 0.52-1.00).
Conclusion: CT characteristics of tumour size, density and heterogeneity are significantly associated with malignancy in AI and applied together reliably exclude malignancy. The risk stratification algorithm utilizing size and density alone may fail to identify some smaller adrenal cancers.
Introduction
The rate of identification of incidentally discovered adrenal lesions is increasing as the use of cross-sectional imaging for the investiga- tion of abdominal symptoms becomes more widespread. These incidentally discovered adrenal lesions are termed ‘incidentalomas’. Adrenal incidentalomas (AI) are reported in up to 5% of computed tomography (CT) scans performed for a variety of indications.1,2 The majority of these lesions are benign and do not require inter- vention; however, up to 5% of these AIs may be malignant.3-5 In addition, over 10% of AIs may be functional, giving rise to
endocrinopathies such as Cushing’s syndrome, hyperaldosteronism and phaeochromocytoma. Adrenocortical carcinoma (ACC) is a rare malignant tumour associated with a rapidly progressive course and generally poor prognosis. Metastatic disease to the adrenal glands typically occurs in patients with a past history of a malig- nancy and accounts for up to 2.5% of AIs.
The challenge for clinicians is to discriminate between benign and malignant AIs. Definitive preoperative diagnosis of ACC is not possible on the basis of cytology or biochemical testing, and surgi- cal decision making is heavily influenced by radiological character- istics. The likelihood of ACC depends on the size of the lesion,
with 6% of lesions between 4.1 and 6 cm and 25% of lesions greater than 6 cm being malignant. Current guidelines recommend surgery for AIs greater than 40 mm.6-8 In addition, attenuation values on non-contrast CTs can help to distinguish between adrenal adenomas and carcinomas.8-11 The majority of ACC have attenua- tion on non-contrast CT scans, which is greater than 30 Hounsfield units (HU). Magnetic resonance imaging has similar limitations to CTs, being unable to differentiate between benign and malignant in 10-30% of tumours.7,12 Contrast-enhanced CT washout character- istics may be effective in differentiating adenomas from non-adeno- mas.6,13,14 The 2016 European guidelines (ESE and ENSAT) recommend the use of non-contrast CTs in the first instance to determine if the incidental adrenal mass is benign. Should the mass be indeterminate, and no excess hormone secretion is detected, three options can be considered: (i) imaging using another modality such as magnetic resonance imaging, (ii) repeat imaging in 6-12 months or (iii) immediate surgery.15
Given the rarity of ACC, most research is limited by patient numbers and performed retrospectively. Birsen and colleagues recently proposed a risk stratification algorithm based on tumour size and HU to help guide surgical decision making.16 The aim of this study was to identify the preoperative tumour characteristics that predict malignancy in AIs and to evaluate the accuracy of the recently proposed risk stratification algorithm based on an inde- pendent data set.
Methods
A retrospective analysis of data stored within a prospectively main- tained Endocrine Surgery Database in the setting of a tertiary refer- ral centre was undertaken. Consecutive patients referred for evaluation of AI in the period 2004-2014 were included in the study. These patients underwent CT imaging and were tested for
evidence of active hormone secretion before either undergoing sur- gery for resection of the AI or being managed by observation and serial radiological surveillance and hormonal testing. Hormonal evaluation comprised the measurement of serum electrolytes, ratio of plasma aldosterone concentration to plasma renin activity (PRA ng/mL/h), 24-h total urinary metanephrines and fractio- nated catecholamines, 24-h urinary-free cortisol and dehydroepi- androsterone sulphate (DHEAS). Radiological imaging was performed using high resolution thin-slice CT imaging with measurement of HU on non-enhanced scans. In addition, washout characteristics following the administration of contrast material were recorded.
Symptomatic patients presenting for investigation of adrenal tumours were not included in the study. Similarly, patients with a known extra-adrenal primary cancer being screened for metastatic disease were excluded from our study. Final histopathology reports were reviewed for surgical cases. Non-operative cases were classi- fied as having benign adenoma on the basis of progression-free sta- tus after at least 6 months of radiological surveillance.
The following factors were analysed for association with malignancy: older age,10,17 male gender,10 a previous history of malignancy,10,17 a larger tumour size,8-11 higher density,8-11 percentage washout on contrast CT, a heterogeneous appearance,9,17 calcification9 on CT and elevated cortisol levels. Fisher’s exact test was used, and variables with a P-value of less than 0.05 were considered significant.
Lastly, an independent evaluation of a scaled score algorithm recently published by a group from the Cleveland Clinic was per- formed.16 A total of 73 tumours with complete size and density measurements were used in our analysis. In accordance with their model, we excluded all hormonally active tumours, following which we accorded each non-secreting tumour a total score based on size and density. Tumour size <40, 40-60 or >60 scored 1, 2 or 3, respectively. Tumour density on non-contrast CT <10 HU,
| Variables | Incidence of cancer | Fisher's exact test (P-value) | |
|---|---|---|---|
| Tumour size on CT | ≥40 mm: 22% | <40 mm: 2% | 0.0066#§ |
| Heterogeneity | Heterogenous: 24% | Homogenous: 3% | 0.0036# |
| Relative percentage washout | <40% washout: 25% | ≥40% washout: 0% | 0.0178# |
| Cortisol levels | Increased: 67% | Normal: 6% | 0.0176# |
| Dehydroepiandrosterone levels | Increased: 100% | Normal: 6% | 0.0061 |
| Younger age | <50 years: 27% | 0.0001# | |
| 50-59 years: 0% | |||
| 60-69 years: 0% | |||
| >70 years: 6% | |||
| Density on non-contrast CT | <10HU: 0% | >10 HU: 10% | 0.3282 |
| Density on contrast CT | 10-29.9HU: 10% | 1 | |
| 30-49.9HU: 8% | |||
| 50-69.9HU: 10% | |||
| ≥70HU: 7% | |||
| Previous history of cancer | Yes: 8% | No: 9% | 1 |
| Gender | Male: 9% | Female: 8% | 1 |
| Absolute percentage washout | <60%: 18% | ≥60%:6% | 0.1649 |
| Catecholamine levels | Increased: 0% | Normal: 9% | 1 |
| Aldosterone to renin ratio | Increased: 0% | Normal: 9% | 1 |
| Calcification on CT | Present: 23% | Absent: 6% | 0.0735 |
tResults of univariate analysis and Fisher’s exact test. All variables analysed are in the extreme leftcolumn. In the middle column, the incidence of cancer expressed as a percentage of total tumours in each category is reported. For example, there is a 22% incidence of cancer in tumours measuring greater or equal to 40 mm on CT. The right column contains P-values from Fisher’s exact test. +P-value < 0.05 is considered significant. §P = 0.0016 when tumour size is ana- lysed as a categorical variable. CT, computed tomography.
10-20 HU or >20 HU scored 1, 2 or 3, respectively. Scores for size and density of each tumour were then added to give a total com- bined score. Sensitivity and specificity values, as well as positive likelihood ratios, were calculated for total score cut-off from 2 to 6. Area under the receiver operating characteristic (ROC) curve was plotted to determine the discriminating power of the scoring system.
Results
In the period 2004-2014, 96 consecutive patients with a total of 98 incidentally discovered adrenal lesions were evaluated. Of these, 66 cases underwent surgical management, and 30 patients were observed. The median age at presentation was 59 years (range 25-77 years), and mean tumour diameter on preoperative CT was 34 ± 18.8 mm. The overall rate of malignancy was 8.2%, with eight ACC confirmed at final histopathology. The median tumour diameter on preoperative imaging for ACC was 61.5 mm (range: 34-100 mm). Demographic details for the surgical and observed cases are available as supporting information (Table S1). As part of the clinical workup, 11 of the 98 tumours (11%) were found to be hormonally active.
The following variables were significantly associated with malig- nancy: tumour size on CT (P = 0.0016), heterogeneity on CT ima- ging (P = 0.0036), a relative percentage washout of less than 40% (P = 0.0178), elevated 24-h urinary-free cortisol levels (P = 0.0176), elevated DHEAs (P = 0.0061) and younger age at presentation (P < 0.0001) (Table 1). The characteristics of patients with ACC are shown in Table 2. The final histopathology of all tumours resected is shown in Table 3.
We applied the algorithm published by the Cleveland Clinic group to our data set. A total of 73 tumours had both size and den- sity measurements on CT. First, we excluded all hormonally active tumours (nine of the 73 tumours). Second, we scored tumours on size and density (see Methods section). A total of 25% of tumours (n = 16) had a score of 2; 23% of tumours (n = 15) had a score of 3; 34% of tumours (n = 22) had a score of 4; and the remaining 17% of tumours (n = 11) had a score of 5 or 6. There were no malignancies identified in tumours scoring 2 or 4; however, one ACC scored 3, and there was 27% incidence of ACCs in tumours that scored either 5 or 6. The results are represented in Table 4. Applying a cut-off combined score of 5 for size and density yielded
| Pathology | Quantity | Mean size ± SD (mm) | Mean density ± SD (HU) |
|---|---|---|---|
| Benign adrenocortical adenoma | 39 | 35 ± 10.8 | 17.1 ± 18.6 |
| Benign cyst | 8 | 57 ± 20.2 | 29.1 ± 22.6 |
| Adrenocortical carcinoma | 8 | 64 ± 21.8 | 31.2 ± 13.1 |
| Pheochromocytoma | 2 | 22 ± 3.5 | 46 |
| Ganglioneuroma | 1 | 59 | 31.5 |
| Myelolipoma | 1 | 28 | n.a.+ |
| Splenunculus | 1 | 34 | n.a.+ |
| Other | 6 | 29 ± 11.8 | 9.5 ± 27.3 |
tList of all adrenal incidentalomas resected in the current study according to histopathology. Included are the number of tumours in each category, their mean size ± SD (mm) and mean density ± SD (HU). Tumour density on non-contrast CT was not reported. CT, computed tomography; HU, Hounsfield unit on non-contrast computed tomography.
the highest sensitivity (75%) and specificity (87%) in identifying cancers in our data set. The scoring system yielded an area under the ROC curve of 0.81, 95% confidence interval (0.52, 1.00).
Discussion
Current Australasian surgical practice is in line with American,6,7 Canadian12 and European15 guidelines recommending CT imaging as part of diagnostic workup. Indications for surgery are a tumour size greater than 40 mm, density greater than 10 HU, atypical radi- ological characteristics or a rapid increase in size. For non- suspicious lesions, follow-up with interval imaging is individua- lized.18 This retrospective analysis of AI confirms the association of an increased risk of malignancy with larger tumour size (P = 0.0016), heterogeneity (P = 0.0036), a relative percentage washout of less than 40% (P = 0.0178), elevated 24-h urinary-free cortisol levels (P = 0.0176) and elevated DHEAs (P = 0.0061). Notably, the risk of malignancy decreased with increasing patient age, possibly reflecting the increasing incidence of benign lesions in older patients.
| Patient | Gender | Age (years) | Tumour size (mm) | Density (HU) | Features on CT | Hormone secretion | Prior history of malignancy |
|---|---|---|---|---|---|---|---|
| 1 | M | 37 | 100 | n.a.+ | Heterogeneous, calcifications present | Cortisol and DHEAS | No |
| 2 | F | 39 | 60 | 28 | Heterogeneous, calcifications present | Cortisol and DHEAS | No |
| 3 | M | 25 | 40 | 35 | Heterogeneous | None | No |
| 4 | M | 71 | 34 | 10 | Homogeneous | None | Yes |
| 5 | F | 26 | 73 | 43 | Heterogeneous | None | No |
| 6 | F | 25 | 60 | 40 | Heterogeneous, calcifications present | None | No |
| 7 | M | 40 | 85 | n.a.+ | Homogeneous | None | No |
| 8 | F | 46 | 63 | n.a.+ | Heterogeneous | None | No |
tList of characteristics of all patients with adrenocortical carcinoma in the study, including gender, age, tumour size and density, radiological features on CT, hor- monal profile and previous history of cancer. Tumour density on non-contrast CT was not reported. CT, computed tomography; DHEAS, dehydroepiandrosterone sulfate; F, female; HU, Hounsfield unit on non-contrast computed tomography; M, male.
| Score | Number of tumours | Percentage of total tumours scored (%) | Number of cancers | Incidence of cancer in each score category (%) |
|---|---|---|---|---|
| 2 | 16 | 25 | 0 | 0 |
| 3 | 15 | 23 | 1 | 7 |
| 4 | 22 | 34 | 0 | 0 |
| 5 or 6 | 11 | 14 | 3 | 27 |
tScoring of a total of 64 non-secreting tumours with both size and density measurements. According to the Cleveland Clinic group’s algorithm 1, 2 or 3 points were allocated for tumours <40, 40-60 or >60 mm, respectively, and an additional 1, 2 or 3 points were given for density on non-contrast CT of HU <10, 10-20 or >20, respectively. A total of 25% of tumours (n = 16) had a combined score of 2, with no cancers in the score category; 23% of tumours (n = 15) had a score of 3, with 1 cancer identified; 34% of tumours (n = 22) had a score of 4, with no cancer identified; the remaining 17% of tumours (n = 11) had a score of 5 or 6, with a 27% incidence of adrenocortical carcinomas in the score category.
Algorithms with graded parameters have also been proposed as tools for risk stratification in AI. A recent large South China study has evaluated malignancy risk in terms of size but not density.19 It is likely that taking into account an additional variable would pro- vide more accurate assessment of tumour malignancy. The most recent scoring system proposed by the Cleveland Clinic group scores non-secreting tumours according to size and density. A score of 2-4 with no suspicious features on imaging would be categor- ized as low risk and warrants observation with a follow-up CT in 3 months. A score of 5 or 6 would point to immediate surgery. An area of 0.81 under the ROC curve reflects moderate discrimination of the classification algorithm, although there was a wide confi- dence interval partly attributed to our limited sample size. Our find- ings agreed with the Cleveland Clinic group’s study, demonstrating that a cut-off score of 5 offered the best sensitivity (80%) and speci- ficity (85%). However, one case in our data set scored a 3 with no suspicious features on imaging.
Except for a few,8,20 most studies are limited by the rarity of ACC and small sample sizes.21-23 We have been able to demon- strate that the Cleveland Clinic’s scoring method had moderate dis- criminating power; however, we overlooked 20% (n = 1) of ACC in our data set. While adrenal carcinomas are usually large when detected, and those greater than 50 mm tend to be heterogeneous on CT, cancers smaller than 50 mm tend to appear homogenous.9,24 Our independent assessment demonstrates that screening based on size and density alone might exclude smaller, lower-density, homo- geneous cancers.25
This study was conducted in a tertiary surgical centre where selection bias is likely to be a significant limitation. Due to selec- tion for surgery and the size of the study, these data reflect likely outcomes for a surgical practice rather than a community medical practice. However, the prevalence of 8% ACC as a proportion of AI tumours in this study is reflective of ACC rates obtained in other surgical series (range 0-14%).26 These data are therefore relevant to the large majority of specialist surgeons who undertake adrenal surgery in a similar setting.
Extrapolating the results of our univariate analysis with the eval- uation of the Cleveland Clinic’s risk stratification algorithm sug- gests that, in our cohort, the addition of an additional parameter such as a relative percentage washout of less than 40% seen on pre- operative CT scan may have avoided the misclassification of one patient who presented with a small ACC. Contrast-enhanced wash- out characteristics have been demonstrated to effectively identify adrenal adenomas,13,14 although we need to be mindful that
enhancement washout has limited value in the assessment of tumours with large areas of necrosis or haemorrhage.27 Using a rel- ative percentage washout of 40% at 15 min yielded 96% sensitivity and 100% specificity.2 28,29
We propose that a combination of variables, including size, den- sity and percentage washout on contrast CT, need to be included in order to improve on current risk stratification methods for the man- agement of AI.
References
1. Kloos RT, Gross MD, Francis IR, Korobkin M, Shapiro B. Incidentally discovered adrenal masses. Endocr. Rev. 1995; 16: 460-84.
2. 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. Am. J. Roentgenol. 2008; 190: 1163-8.
3. Terzolo M, Ali A, Osella G, Mazza E. Prevalence of adrenal carcinoma among incidentally discovered adrenal masses. A retrospective study from 1989 to 1994. Gruppo Piemontese Incidentalomi Surrenalici. Arch. Surg. 1997; 132: 914-9.
4. Thompson GB, Young WF Jr. Adrenal incidentaloma. Curr. Opin. Oncol. 2003; 15: 84-90.
5. Young WF Jr. Management approaches to adrenal incidentalomas. A view from Rochester, Minnesota. Endocrinol. Metab. Clin. North Am. 2000; 29: 159-85.
6. Zeiger MA, Thompson GB, Duh Q-Y et al. American Association of Clinical Endocrinologists and American Association of Endocrine Sur- geons medical guidelines for the management of adrenal incidentalo- mas. Endocr. Pract. 2009; 15 (Suppl 1): 450-3.
7. NIH state-of-the-science statement on management of the clinically inapparent adrenal mass (“incidentaloma”). NIH Consens. State Sci. Statements 2002; 19: 1-25.
8. Sturgeon C, Shen WT, Clark OH, Duh Q-Y, Kebebew E. Risk assess- ment in 457 adrenal cortical carcinomas: how much does tumor size predict the likelihood of malignancy? J. Am. Coll. Surg. 2006; 202: 423-30.
9. Ozturk E, Onur Sildiroglu H, Kantarci M et al. Computed tomography findings in diseases of the adrenal gland. Wien. Klin. Wochenschr. 2009; 121: 372-81.
10. Bulow B, Ahren B, Swedish Research Council Study Group of Endo- crine Abdominal Tumours. Adrenal incidentaloma - experience of a standardized diagnostic programme in the Swedish prospective study. J. Intern. Med. 2002; 252: 239-46.
11. Fassnacht M, Kenn W, Allolio B. Adrenal tumors: how to establish malignancy? J. Endocrinol. Invest. 2004; 27: 387-99.
Predicting malignancy in adrenal incidentaloma
12. Kapoor A, Morris T, Rebello R. Guidelines for the management of the incidentally discovered adrenal mass. Can. Urol. Assoc. J. 2011; 5: 241-7.
13. Korivi BR, Elsayes KM. Cross-sectional imaging work-up of adrenal masses. World J. Radiol. 2013; 5: 88-97.
14. Korobkin M, Brodeur FJ, Francis IR, Quint LE, Dunnick NR, Goodsitt M. Delayed enhanced CT for differentiation of benign from malignant adrenal masses. Radiology 1996; 200: 737-42.
15. Fassnacht M, Arlt W, Bancos I et al. Management of adrenal incidenta- lomas: European Society of Endocrinology Clinical Practice Guideline in collaboration with the European Network for the Study of Adrenal Tumors. Eur. J. Endocrinol. 2016; 175: G1-34.
16. Birsen O, Akyuz M, Dural C et al. A new risk stratification algorithm for the management of patients with adrenal incidentalomas. Surgery 2014; 156: 959-65.
17. Lumachi F, Borsato S, Tregnaghi A et al. High risk of malignancy in patients with incidentally discovered adrenal masses: accuracy of adre- nal imaging and image-guided fine-needle aspiration cytology. Tumori 2007; 93: 269-74.
18. Glover A, Sidhu S. Approach to the Adrenal Incidentaloma. Agency for Clinical Innovation. Available from URL: https://www.aci.health. nsw.gov.au/resources/surgical-services/clinical-guidelines/approach-to- the-adrenal-incidentaloma
19. Ye Y-l, Yuan X-x, Chen M-k, Dai Y-p, Qin Z-k, Zheng F-f. Management of adrenal incidentaloma: the role of adrenalectomy may be underesti- mated. BMC Surg. 2016; 16: 41.
20. Kasperlik-Zaluska AA, Otto M, Cichocki A et al. 1,161 patients with adrenal incidentalomas: indications for surgery. Langenbecks Arch. Surg. 2008; 393: 121-6.
21. Murai M, Baba S, Nakashima J, Tachibana M. Management of inciden- tally discovered adrenal masses. World J. Urol. 1999; 17: 9-14.
22. Mohamed AM, Moftah SG, El-lithy MA. Value of combined CT para- meters in distinguishing benign from malignant adrenal masses in can- cer patients. Egypt. J. Radiol. Nucl. Med. 2012; 43: 275-83.
23. Strosberg JR, Hammer GD, Doherty GM. Management of adrenocorti- cal carcinoma. J. Natl. Compr. Canc. Netw. 2009; 7: 752-8.
24. Oliveira Caiafa R, Salvador Izquierdo R, Buñesch Villalba L, Sebastià Cerqueda MC, Nicolau MC. Diagnosis and management of adrenal incidentaloma. Radiologia 2011; 53: 516-30.
25. Favia G, Lumachi F, Basso S, D’Amico DF. Management of inciden- tally discovered adrenal masses and risk of malignancy. Surgery 2000; 128: 918-24.
26. Mansmann G, Lau J, Balk E, Rothberg M, Miyachi Y, Bornstein SR. The clinically inapparent adrenal mass: update in diagnosis and man- agement. Endocr. Rev. 2004; 25: 309-40.
27. Dunnick NR, Korobkin M. Imaging of adrenal incidentalomas: current status. Am. J. Roentgenol. 2002; 179: 559-68.
28. Korobkin M, Brodeur FJ, Francis IR, Quint LE, Dunnick NR, Londy F. CT time-attenuation washout curves of adrenal adenomas and nonade- nomas. AJR Am. J. Roentgenol. 1998; 170: 747-52.
29. Caoili EM, Korobkin M, Francis IR et al. Adrenal masses: characteriza- tion with combined unenhanced and delayed enhanced CT. Radiology 2002; 222: 629-33.
Supporting information
Additional Supporting Information may be found in the online ver- sion of this article at the publisher’s web-site:
Table S1. Characteristics of patients and tumours in the surgery group versus the observation group.