Adrenocortical adenomas with regression and myelolipomatous changes: urinary steroid profiling supports a distinctive benign neoplasm

Adrenocortical neoplasms are classically divided into adenomas (ACA) and carcinomas (ACC). Heterogeneous appearance and greater size are criteria to suggest malignancy, along with the urinary steroid profile (USP). The presence of regression and myelolipomatous changes in adenomas (ACA‐RML) can contribute to confusion with ACC and its USP remains unknown.

Background Adrenocortical neoplasms are classically divided into adenomas (ACA) and carcinomas (ACC). Heterogeneous appearance and greater size are criteria to suggest malignancy, along with the urinary steroid profile (USP). The presence of regression and myelolipomatous changes in adenomas (ACA-RML) can contribute to confusion with ACC and its USP remains unknown. Objective To evaluate the features of ACA-RML in comparison with other adrenocortical neoplasms. Methods We selected consecutive ACA (11), ACA-RML (7) and ACC (13) cases for which USP analysis was performed before surgery and tissue was available for histological evaluation (King's College Hospital, 2005-2012). Cases were classified according to WHO and Armed Forces Institute of Pathology criteria. USPs were obtained by gas chromatography/mass spectrometry. Total excretion of individual steroids and indices (sums and ratios chosen to reflect steroid metabolic activity) were compared between ACA-RML, ACA and ACC. Results In comparison with ACA, tumours in ACA-RML were significantly larger (8Á5 AE 2Á4 vs 3Á5 AE 1Á0, P = 0Á002), presented in older patients and showed relatively higher incidence in males. Mitotic figure counts were significantly lower (0Á39 AE 0Á04 vs 0Á93 AE 0Á11 in ACA, P = 0Á001) and revealed higher frequency of apoptotic cells (100% vs 9% in ACA, P = 0Á001). The USP of ACA-RML showed no diagnostic features of ACC. No differences from ACA were significant, but Introduction Adrenocortical proliferative lesions represent a heterogeneous group comprising nodular hyperplasias, adenomas (ACA) and carcinomas (ACC). [1][2][3][4] Clues that should alert clinicians to the possibility of malignancy include the following: tumour size >4-5 cm in diameter, imaging features of calcification, blurred margins, an irregular shape, heterogeneous contrast enhancement and distant metastases on computed tomography, together with hormonal overproduction in more than one adrenocortical axis. 5,6 In contrast, adenomas are smaller, smooth shaped and with homogeneous enhancement.
A minority of adenomas resemble carcinomas in being large, with heterogeneous appearance frequently related to regressive/ degenerative changes, which contribute to a more difficult preoperative diagnosis, especially when the changes are extensive. The criteria used to identify these changes are not clearly defined but include haemorrhage, fibrin deposition, interstitial hyalinization, cystic changes, adipose tissue infiltration and extramedullary hematopoiesis, with or without calcification or ossification. 7,8 To date, adrenocortical adenomas with extensive regressive changes are extremely rare and are reported as single cases or very short series only.
The adrenal cortex produces cortisol, corticosterone and C19 steroids. These are metabolized in the liver and the kidney and then excreted in the urine either in the form of free steroids or conjugated with sulphuric or glucuronic acid. The urinary steroid profile (USP) has proved to be very useful in the identification of carcinomas, 9,10 but findings in adenomas with regression and myelolipomatous changes (ACA-RML) have not been previously reported. Here, we described clinical, histological and USP features of seven cases of this very rare adenoma subtype in comparison with those of conventional adrenocortical adenomas and carcinomas. We aimed to determine if the steroid profile in the ACA-RML group could be securely distinguished from ACC. We further compared this group with ACA and explored whether any differences found might be related to the histological changes.

Selection and description of participants
All adrenocortical proliferative lesions available at King's College Hospital (Dept Histopathology, 1988-2012) were reviewed, including nodular hyperplasias (44), adenomas (118) and carcinomas (28). From this series, extensive regressive myelolipomatous changes (present in at least one-third of the tumour) were observed in 7 ACA (2005-2012); erythroblasts and red blood cells closely collocated with macrophages containing hemosiderin with or without megakaryocytes were required in the focus to identify the myeloid component. During this period, all consecutive ACA (11), ACA-RML (7) and ACC (13) had urine steroid profiles performed before surgery and histopathology available for evaluation and were selected for this analysis. Myelolipomas in the context of hyperplastic conditions (such as congenital adrenocortical hyperplasia or Cushing's disease) were excluded from this study. Conventional ACA and ACC were used as benign and malignant controls, respectively.
The clinical syndrome was investigated in each case by standard clinical and laboratory approaches, including basal hormone determination, along with stimulation and suppression tests if necessary and imaging methods. 4,[11][12][13] The bare adrenal weight was also recorded in each case.

Histologic criteria and classification
All surgical specimens were serially sectioned and routinely processed for histological diagnosis and sampled with at least 1 block per cm of maximum lesion diameter. The cases were classified according to the World Health Organization and Armed Forces Institute of Pathology criteria 12,13 ; if there were proven metastases, they were classified as malignant. All haematoxylin and eosin-stained sections were evaluated histologically and graded by two of the authors (both pathologists, FAH and SDC) independently, using criteria previously described. 3 In case of grading disagreement, the lesions were discussed during simultaneous inspection before final categorization. Reproducibility data were not recorded. Standard scores for the classification of these lesions 4,14,15 were also registered; as the benign group scores low figures (up to 2), 14,15 no further analysis was performed with these scores in the distinction of ACA-RML from ACA and ACC.

Urinary steroid profiling
Examination of 24-h urinary steroid metabolite excretion was carried out using gas chromatography/mass spectrometry (GC/ MS). A detailed description of this methodology has been published. 16 In summary, free and conjugated steroids were extracted from 10 ml urine by solid-phase extraction. Steroid conjugates were enzymatically hydrolysed, reextracted, and chemically derivatized to form methyloxime-trimethylsilyl ethers. GC/MS was carried out on a Clarus 500 â instrument (Perkin-Elmer, Waltham, MA, USA).
The steroid metabolites quantified in our study are listed in supplemental Table 1 and the calculated sums and ratios in supplemental Table 2. Levels of steroid metabolite excretion were expressed as lg/day. Supplemental Figure 1 shows the metabolic relationships of the studied steroids. Other steroids that have proved to be useful markers of ACC were also compared empirically between groups. These included tetrahydro-11-deoxycortisol (16), pregnene-3,16,20-triols (15), 16a-and 21hydroxypregnenolone and tetrahydro-11-deoxycorticosterone.

Statistics
Total excretion of individual steroids and steroid indices were compared between ACA-RML, ACA and ACC groups using the Kruskal-Wallis nonparametric test; for significance, values of P < 0Á05 and pair-wise comparisons were carried out using Dunn's post hoc test.

Results
Clinical characteristics of the patient groups are listed in Table 1.

Histological features
ACA-RML. (Figure 1)After removing the adipose tissue, the adrenal glands weighed 291Á4 AE 366Á3 g. Histologically, the peritumoral gland was atrophic in five cases (71%) and unremarkable in two cases (29%). The tumours showed an alveolar growth pattern in four cases (57%) and a trabecular pattern in 3 (43%). Cytologically, tumours composed of foamy clear cells predominated in this series (5 cases, 71%) and no pigment-containing cells were observed.
Small nonconfluent necrotic foci were present in two cases (29%) and isolated apoptotic cells in all seven cases (100%). No or minimal inflammatory response was found, predominantly lymphocytic (6 cases, 86%). Myxoid stromal reaction was observed in five cases (71%) and no desmoplastic tissue was observed. No capsular, lymphovascular, sinusoidal or perineural invasion was identified. Haemorrhage and stromal hyalinization with myelolipomatous changes as inclusion diagnostic criteria were present in at least 1/3 of the neoplasm in all seven cases.
Small nonconfluent necrotic foci were present in one case (9%) and occasional isolated apoptotic cells in one case (9%). No or minimal inflammatory response was found, predominantly lymphocytic (10 cases, 91%). A myxoid stromal reaction was observed in two cases (18%) and no desmoplastic tissue was observed. No capsular, lymphovascular, sinusoidal or perineural invasion was identified.

Urinary steroid profiling
The USP of ACA-RML revealed no unusual patterns and the cases clustered with ACA and not with ACC. When compared with ACA (regardless of the inclusion or not of ACA with Conn's syndrome), no statistically significant differences for any individual steroid or index could be demonstrated. However, there appeared to be a trend for ACA-RML towards decrease in dehydroepiandrosterone (DHA) and DHA metabolites (Fig. 2) whether graphed as individual steroids, expressed as a total of DHA and DHA metabolites or as a ratio related to total cortisol metabolites (Fig. 3). 16a-Hydroxy DHA showed a greater difference than DHA, suggesting a decrease in 16a-hydroxylation activity in ACA-RML. However, a ratio of 16a-hydroxylated vs. non-16a-hydroxylated DHA metabolites was not different between ACA-RML and ACA.
Urinary steroid profiling analysis revealed statistically significant differences between both ACA groups (i.e.: ACA and ACA-RML) and the ACC group. ACC patients had higher excretion of numerous steroids, especially metabolites of steroid pathway intermediates. Those with the greatest differences are shown in Fig. 4, while the differences for DHA and DHA metabolites were also all significant (Fig. 3).

Discussion
ACA-RML represents a distinctive benign adrenocortical neoplasm for which the USP findings cluster with ACA. Although both the variability and rarity of this condition preclude firmer conclusions, the USP suggests an impaired DHA metabolic pathway with relatively preserved cortisol biosynthesis. Both age and ACTH (adrenocorticotroph hormone) suppression can contribute to this finding. Secretion of adrenal steroids, especially DHA and its sulphate, declines with age. 10,17 Given that the ACA-RML group is older than the ACA group (Table 1), it is possible that the decrease in total C19 steroid sulphate (C19SS) metabolites and the increased cortisol metabolites (FM)/C19SS ratio are age related. However, sample numbers were too few to be able to apply procedures such as stepwise regression analysis to examine the influence of age. No glucocorticoid suppression testing was undertaken in the ACA-RML group, so the contribution of the tumour tissue to steroid secretion cannot be estimated. Likewise any contribution of tumour tissue to further metabolism of steroids arising in normal adrenocortical tissue cannot be gauged. In our series, the majority of the ACA and ACA-RML had ACTH-independent Cushing's syndrome. As DHA is secreted from the normal cortex and the abnormal secretion is usually along a single pathway (glucocorticoid, mineral cor-ticoid or androgen), low DHA and DHA metabolites levels may reflect ACTH suppression; however, ACTH independent mechanisms must also be considered from secretory adrenal neoplasms. Conversely in the ACC, it is common for multiple steroidogenic pathways to be upregulated independently of ACTH, hence the different profile. However, the values are Fig. 3 Comparison of patient groups: C19 steroid sulphates metabolites: -ACA vs ACC: *P < 0Á05, **P < 0Á01. -ACA-RML vs ACC: **P < 0Á01, ***P < 0Á001. ▲, ▲▲, ▲▲▲ : Not significant for the comparison ACA vs ACA-RML. Only asterisks comparison were significant, all remaining (triangles) were found not significant. Fig. 4 Comparison of patient groups: steroid precursor metabolites: -ACA vs ACC: ***P < 0Á01, ***P < 0Á001. -ACA-RML vs ACC: *P < 0Á05, **P < 0Á01, ***P < 0Á001. -▲, ▲▲, ▲▲▲ : Not significant for the comparison ACA vs ACA-RML. Only asterisks comparison were significant, all remaining (triangles) were found not significant. much lower for ACA-RML and they cannot be explained by ACTH suppression only. Low secretion in DHA and metabolites in ACA-RML have a multifactorial origin. It cannot be argued that all DHA is from the normal cortex and that it will be due to ACTH suppression because Cushing's syndrome was not universally identified and cortical atrophy was not present in all cases: two patients presented with adrenal mass only. In addition, any regression analysis of this very rare subtype of ACA (with a reduced number of cases) will not result in any statistically significant differences. Considering the proportionally big size of these neoplasms, the ischaemic changes associated with the mass effect (stromal hyalinization, oedema) will be a major contributor to this low secretion, especially in combination with age, 17,18 incomplete ACTH intratumoral functional activity 19 and oxidative effect of mononuclear bone marrow cells. [20][21][22] ACA is a proliferative lesion with low cellular turnover, along with heterogeneous kinetic and clonal profiles. [23][24][25] The mitotic figure counting was significantly lower in ACA-RML than in ACA and revealed a higher frequency of apoptotic cells. This kinetic pattern corresponds with the right tail of regressive monoclonal ACA (low proliferation and high apoptosis), which frequently shows interstitial haemorrhage and myelolipomatous changes. 23,24,26 These added interstitial features give rise to lesions greater in size and with heterogeneous appearance. 8 ACA-RML tumours were of significantly larger size than ACA, presented in older patients and showed a similar incidence in males and females, unlike the common epidemiology in ACA.
USP profiles of ACA-RML do not show any of the ACC markers previously reported, such as increased steroid precursor metabolites (pregnenediol and pregnenetriol) or increased DHA metabolites. 27 The evaluation of USP in adrenocortical neoplasms includes the identification of abnormal metabolites and/ or the presence of a heterogeneous background associated with partial enzymatic deficiencies; the former findings being more accurate predictors of a malignant diagnosis. The pathological heterogeneity contributes to this complexity, but no single marker or value on its own is pathognomonic of this disease. Although USP will be more widely used in the future, considering the low ACC incidence and its heterogeneity with several patterns, it is more likely that accurate interpretations would be more reliable from specialized centres.
Although no significant differences between the ACA-RML and the ACA groups were found, there is an apparent decrease in the C19SS metabolic pathway in the group of adenomas with regressive changes. ACA-RML typically presented in older patients (as compared to both ACA and ACC), with large adrenal masses that reveal some degree of radiological heterogeneity. These features lead to investigate a potentially malignant condition, for which the urine steroid profile can be valuable. Our results probed the absence of positive malignant markers to support a benign lesion. Like with other diagnostic tests, it cannot be expected 100% specificity and/or sensitivity and that is our experience with the USP for the diagnosis of ACC. For that reason, observation only would be advised with caution as we can be dealing with the odd case, and the surgical intervention must be individually evaluated and, if necessary, a less radical excision should always be considered. Although the absence of markers will not definitely exclude a malignant lesion, a less radical surgical intervention can be planned in advance; this is a more desirable option for elderly patients who are the most frequently affected by this tumour type.
In conclusion, ACA-RML reveals distinctive histological features, along with a regressive kinetics profile that may co-segregate with decreased DHA synthesis. Importantly, they lack USP markers of malignancy, so that clear distinction from ACC is possible without surgical intervention. These conditions are radiologically similar, so USP findings can be critical in determining treatment strategy.