The Reticulin Algorithm for Adrenocortical Tumor Diagnosis A Multicentric Validation Study on 245 Unpublished Cases
Eleonora Duregon, MD,* Ambrogio Fassina, MD,; Marco Volante, MD, PhD,* Gabriella Nesi, MD,¿ Raffaella Santi, MD,¿ Gaia Gatti, MD,* Rocco Cappellesso, MD,¡ Paolo Dalino Ciaramella, MD,§ Laura Ventura, PhD, | Marcello Gambacorta, MD,| Angelo Paolo Dei Tos, MD,#
Paola Loli, MD,§ Massimo Mannelli, MD, ** Franco Mantero, MD,tt Alfredo Berruti, MD,## Massimo Terzolo, MD,§§ and Mauro Papotti, MD*
Abstract: The pathologic diagnosis of adrenocortical carcinoma (ACC) still needs to be improved, because the renowned Weiss Score (WS) system has a poor reproducibility of some para- meters and is difficult to apply in borderline cases and in ACC variants. The “reticulin algorithm” (RA) defines malignancy through an altered reticulin framework associated with 1 of the 3 following parameter: necrosis, high mitotic rate, and vascular invasion. This study aimed at validating the interobserver re- producibility of reticulin stain evaluation in an unpublished series of 245 adrenocortical tumors (61 adenomas and 184 car- cinomas) from 5 Italian centers, classified according to the WS. Eight pathologists reviewed all reticulin-stained slides. After training, a second round of evaluation on discordant cases was performed 10 weeks later. The RA reclassified 67 cases (27%) as adenomas, including 44 with no reticulin alterations and 23 with an altered reticulin framework but lacking the subsequent pa- rameters of the triad. The other 178 cases (73%) were carcino- mas according to the above-mentioned criteria. A complete (8/8 pathologists) interobserver agreement was reached in 75% of cases (K = 0.702), irrespective of case derivation, pathologists’ experience, and histologic variants, and was further improved when only those cases with high WS and clinically malignant
From the Departments of *Oncology; §§Clinical and Biological Sciences, University of Turin at San Luigi Hospital, Orbassano, Torino; *Department of Medicine (DIMED), Pathology and Cytopathology Unit; ttDepartment of Medicine (DIMED), Endocrinology Unit; | Department of Statistics, University of Padua, Padova; Divisions of ¿Pathology; ** Endocrinology, University of Florence at Careggi Hospital, Firenze; Divisions of §Endocrinology; |Pathology, Niguarda Ca’ Granda Hospital, Milano; Department of Pathology and Molecular Genetics, General Hospital, Treviso; and #Department of Medical Oncology, University of Brescia, Italy. Conflicts of Interest and Source of Funding: Supported by grants from the Italian Association for Cancer Research (AIRC, Milan, grant no. IG/10795/2010 to MP) and Cariparo Foundation Excellence grant (Padova, 2011/2012). The authors have disclosed that they have no significant relationships with, or financial interest in, any commercial companies pertaining to this article.
Correspondence: Marco Volante, MD, PhD, Department of Oncology, University of Turin at San Luigi Hospital, Regione Gonzole 10, 10043 Orbassano, Torino, Italy (e-mail: marco.volante@unito.it) Copyright @ 2013 by Lippincott Williams & Wilkins
behavior were considered. After the training, the overall agree- ment increased to 86%. We conclude that reticulin staining is a reliable technique and an easy-to-interpret system in adreno- cortical tumors; moreover, it has a high interobserver reprodu- cibility, which supports the notion of using such a method in the proposed 2-step RA approach for ACC diagnosis.
Keywords: adrenal cortex, carcinoma, reticulin stain, reprodu- cibility study, diagnosis
(Am J Surg Pathol 2013;37:1433-1440)
T he pathologic diagnosis of adrenocortical carcinoma (ACC) is based on the recognition of several mor- phologic parameters, none of which is per se pathogno- monic of malignancy. For this reason, they were combined in scoring systems that include up to 12 mi- croscopic and macroscopic criteria.14 These procedures are time consuming and hardly reproducible, unless used by experienced and specifically trained pathologists.5,6 In the last decade, simplified versions of these scoring sys- tems were proposed, grounded either on different aspects of a single parameter [eg, mitotic rate and clear cells comprising 25% or less of the tumor in the Aubert modified Weiss System (WS)]7 or on an algorithmic ap- proach centered on mitotic figure variability, such as the stepwise discriminant diagnostic system.8 The WS is the most widely used diagnostic system nowadays, but its reliability is challenged in borderline cases, having only few criteria represented. Moreover, it is scarcely re- producible in the ACC morphologic variants. In fact, specifically for the pure oncocytic neoplasms, a modified system has been proposed,9 whereas for the myxoid var- iant, some Weiss criteria are difficult to identify because of the peculiar morphology of such tumors.10
Some of us have recently proposed a different approach, the “reticulin algorithm” (RA), which defines malignancy in adrenocortical tumors relying on a 2-step process: first, the analysis of the reticulin framework (highlighted by reticulin silver-based histochemical
staining), then, if a disruption is recognized, malignancy is eventually defined through the identification of at least 1 of 3 malignancy-related criteria-that is, necrosis, high mitotic rate (>5/50 HPF), and venous invasion. This algorithm showed a similar diagnostic performance as compared with the WS, but it was faster and easier to apply. The efficacy of the RA was also confirmed in the rare oncocytic variant,12 whose diagnostic criteria are even more complex than those of classical ACC.
To date, the reticulin staining technique has only been used in a single report of 40 adrenocortical tumors13 and quoted in a review article.14 It still needs to be vali- dated, at least with regard to reticulin staining, which is the first step of the algorithm. In fact, the second step is based on the identification of 3 parameters, which are part of the WS and have been extensively reappraised in a recent reproducibility study by a French group.6
To this purpose, a multicentric validation study has been designed, aimed at assessing the reproducibility of the reticulin stain interpretation in a series of 245 newly col- lected and unpublished adrenocortical tumors from 5 Italian centers, mostly including classical ACC forms and special variants, as well as a consistent number of benign tumors.
We here show that reticulin staining is a fast and cheap technique and an easy-to-interpret system in adre- nocortical tumors, in which both quantitative and qual- itative changes are considered in the evaluation of the reticulin framework disruption; it has a high interobserver reproducibility, which supports the use of such a sim- plified and fast method in the proposed 2-step RA ap- proach for ACC diagnosis.
MATERIALS AND METHODS
Case Collection
A total of 245 adrenocortical tumors were collected from the archives of the Pathology Divisions of 5 Italian Institutions and classified as adenomas or carcinomas according to the WS,3,4 except for the 21 pure oncocytic lesions, which were classified according to the system proposed by Bisceglia et al,9 as follows (Table 1):
(a) 117 consecutive adrenocortical tumors having a WS ≥ 1 were collected between 2009 and 2012 from the pathology files of the University of Turin at San Luigi Hospital; this series included 19 cases resected at San Luigi Hospital and 98 surgical cases received in consultation from different hospitals. Among them, 13 cases were adrenocortical adenomas (ACAs) with
WS 1 or 2 (one had myxoid and another oncocytic features), whereas the remaining 104 were ACCs with WS ≥3 (8 had myxoid changes, 17 oncocytic, and 1 sarcomatoid features). The majority of these patients were treated at San Luigi Hospital in Orbassano, Torino, which serves as one of the referral centers for ACC in Italy. All previously published oncocytic and myxoid cases observed in this same period10,12 and the 139 adrenal tumor cases originally described in the previous study on the RA11 were excluded from the present investigation;
(b) 61 adrenocortical tumors, including 26 ACAs (2 oncocytic) and 35 ACCs (9 oncocytic and 1 myxoid) were recovered between 1993 and 2011 from the University of Florence at Careggi Hospital;
(c) 26 ACAs (4 oncocytic) and 24 ACCs (7 oncocytic and 3 myxoid) were retrieved from the pathology files of the University of Padua between 2000 and 2008;
(d) 1 ACA and 14 ACCs (3 oncocytic and 1 myxoid) were retrieved from 1994 to 2007 at the Division of Pathology of Niguarda Ca’Granda Hospital, Milan;
(e) 2 ACAs and 7 ACCs (1 oncocytic and 1 myxoid) were recovered from 1998 to 2012 at the General Hospital of Treviso, Italy.
Two cases, both carcinomas according to the WS, were of pediatric age (a 9-mo-old girl and a 3-y-old boy). These cases were also reclassified according to the criteria of Wieneke et al,15 as follows: 1 cases was classified as benign having 1/9 criteria (necrosis only) and the second as indeterminate for malignancy (3/9 parameters, in- cluding size, capsular invasion, and necrosis).
The study received ethical approval from the local Review Board of our Institutions.
Histochemistry
To define the status of the reticulin framework, each Institution performed a reticulin histochemical staining on its respective cases, using a commercially available silver impregnation-based kit (Bio Optica, Milan, Italy). The block for reticulin staining was selected on the basis of the architecture of the tumor in hematoxylin and eosin sections. Areas with a diffuse/nodular/trabecular (if present) rather that alveolar (normal adrenal-like) ar- rangement, which are morphologically consistent with reticulin disruption, were selected for reticulin staining.
All 245 reticulin-stained slides were originally re- vised by a local pathologist (M.P.) in order to verify that a representative block was selected for each case and were
| TABLE 1. Description of Adrenocortical Tumor Cases Enrolled in the Study According to the Center of Collection | ||||
|---|---|---|---|---|
| Total #245 | Years of Collection | ACA #61 (Onco/Myx) | ACC #184 (Onco/Myx) | |
| Turin | 117 | 2009-2012 | 13 (1/1) | 104 (17/8) |
| Florence | 61 | 1993-2011 | 26 (2/0) | 35 (10/1) |
| Padua | 42 | 2000-2008 | 18 (4/0) | 24 (7/3) |
| Milan | 15 | 1994-2007 | 1 (0/0) | 14 (3/1) |
| Treviso | 10 | 1998-2012 | 3 (0/0) | 7 (1/1) |
Myx indicates myxoid variant of adrenocortical tumors; onco, oncocytic variant.
A
B
C
D
E
F
reclassified according to the RA. In each slide, the retic- ular fiber framework was evaluated first at low (x 100) and then at high magnification (×400). The positive control was the normal adrenal gland on which we de-
fined as “intact” the mesh appearance of the reticular fibers, all with the same thickness, completely surround- ing adrenocortical cells in nests or cords, a structure re- calling an intact fishing net (Figs. 1A, B). All patterns that
| TABLE 2. Comparison of Clinical and Pathological Features of the Present Series and of That of a Previous Study on RA Application | ||||
|---|---|---|---|---|
| 139 Adrenocortical Tumors11 | 245 Adrenocortical Tumors (Present Series) | |||
| ACA (47) | ACC (92) | ACA (61) | ACC (184) | |
| F/M ratio | 3.2 | 1.5 | 1.9 | 1.45 |
| Age (year), mean (range) | 53 (15-77) | 46 (20-85) | 54 (16-85) | 48 (9 mo-97) |
| Location L/R | 23/24 | 48/45* | 28/31 | 92/79+ |
| Size, (range) (cm) | 4 (1-10) | 11.4 (1.6-25) | 3.9 (1-12.9) | 10.5 (2-30) |
| Weight, (range) (g) | 39 (4-150) | 428 (8-3100) | 46 (2.2-201) | 365 (5-2500) |
| Functional status | ||||
| Not functioning | 13 | 50 | 11 | 54 |
| Cortisol | 18 | 13 | 13 | 30 |
| Aldosteron | 15 | 6 | 8 | 4 |
| Androgens | — | 8 | — | 8 |
| Cortisol + androgens | — | — | — | 14 |
| Estrogens | — | — | — | 3 |
| Not known | 1 | 15 | 29 | 71 |
| Status (outcome)} | ||||
| NED | 47 | 32 | 39 | 63 |
| AWD | — | 15 | — | 44 |
| DOD | — | 28 | — | 39 |
| Recent cases | — | 7 | — | 2 |
| WS | — | — | — | — |
| 0-2 | 47 | — | 61 | — |
| 3-5 | — | 38 | — | 43 |
| 6-9 | — | 54 | — | 141 |
| ACC variants | — | — | — | — |
| Myxoid | — | 6 | 1 | 16 |
| Oncocytic | 4 | 12 | 7 | 37 |
| Sarcomatoid | — | — | — | 1 |
*1 case had bilateral tumor.
¡For 13 cases location was not specified in the original report.
Follow-up data were available for 137 cases in the previous report11 and for 187 cases in the current series.
AWD, alive with disease; DOD, died of disease; F, female; L, left; M, male; NED, no evidence of disease; R, right.
differed from the reticulin network of a normal adrenal gland were recorded as “altered.” Both quantitatively and qualitatively altered patterns were recorded. In the former case, the reticular fiber framework continuity was lost in more or less extensive tumor areas (Figs. 1C, D), as previously described.11 Qualitative disruptive changes were characterized by an apparently intact reticulin network made of fibers having variable and irregular thickness, with a frayed appearance, surround- ing single cells or, more rarely, small groups of cells (Figs. 1E, F).
To assess reproducibility of the reticulin stain in- terpretation, the same set of slides was circulated among 8 pathologists with different experience in adrenal patho- logy (E.D., M.V., G.G., A.F., R.C., A.P.D.T., R.S., M.G.), in 2 rounds. During the first, participants (blinded of the original diagnoses) were separately asked to review and categorize the same 245 slides, choosing between 2 possible options (normal or altered), based only on the description of reticulin patterns provided in our previous study.11 In the second round of evaluation performed 10 weeks later, all discordant cases at the first screening (ir- respective of the number of discordant observers) were reassessed, without knowledge of the previous evaluation, after a training specifically designed on selected photo- micrographs of 22 representative cases of different retic- ulin patterns at different magnifications.
Statistical Analysis
Concordant diagnoses among the 8 pathologists were computed, and nonparametric tests (Wilcoxon and Mann-Whitney) were used to test for differences in con- cordant diagnoses between subgroups and between the first and the second rounds. Fleiss K statistics were used to assess agreement among the 8 pathologists. Internal consistency of data was measured with the Cronbach a statistic. Bootstrap methods were used to calculate 95% confidence intervals for the Cronbach a statistic. Sub- group Fleiss k and Cronbach a statistics were also cal- culated. Statistical analyses were performed using the free software R (http://www.r-project.org/). A significance level of 0.05 was used.
RESULTS
Clinicopathologic Data
The main clinical and pathologic features of the whole series of 245 tumors are summarized in Table 2. Follow-up information was available for 187 cases (in- cluding 39 ACAs and 148 ACCs). Thirty-nine patients died of disease, 44 are alive with disease, and 104 are alive with no evidence of disease (65 ACCs and 39 ACAs). These patients are currently on follow-up at our In- stitutions, with a mean follow-up of 40 months (range, 6 to 195 mo for ACC and 6 to 96 mo for ACA).
A
S
B
Central Reticulin Stain Analysis and RA Application
Centralized evaluation of reticulin-stained slides showed that the stromal framework was similar to that of the normal adrenal cortex in 44 cases (18%), whereas it was altered in the remaining 201 cases (82%). The latter mostly included a more or less extensive framework dis- ruption, whereas a minority of them (13 cases, 6%; 6 of which were of the oncocytic type) were rather charac- terized by the qualitative alterations described in the Materials and methods section. Interestingly, 3 cases (1 oncocytic and 2 conventional) presented heterogenous reticulin patterns attributable to subclonal expansion within otherwise normal adrenal-like areas (Fig. 2). Comparison with available Ki-67 immunohistochemical reactions of the same areas in 2 cases showed a significantly increased proliferation in reticulin-disrupted as compared with reticulin-preserved areas. All cases with an altered network entered the second step of the algorithm, and at least 1 parameter among high mitotic rate (>5/50 HPF), necrosis, and venous invasion was found in 178 cases. In the other 23 cases, no malignancy-related parameters were found in any of the available hematoxylin and eosin-stained slides (from 5 to 13 per case). As compared with the WS and Lin-Weiss-Bisceglia scoring systems, 6 cases (all but 2 purely oncocytic) were classified as ACAs by RA rather than ACCs as in the original diagnostic scoring, and all
these patients are alive with no evidence of disease at cur- rent follow-up (6 to 195 mo, mean 49 mo).
Interobserver Reproducibility of Reticulin Stain Evaluation
In the first round, reticulin stain evaluation was performed on 245 cases by 8 pathologists (Table 3).
On the whole series and in the subgroup with fol- low-up information available, the agreement rate among 8 pathologists was 75%. Stratifying cases for WS, the full agreement was 24% in cases with WS < 3 and 88% in those with WS≥3 (P < 0.0001), whereas for clinical behavior the agreement was 61% in patients without evidence of disease and 92% for those who had a re- currence or died of disease (P < 0.0001). The “dis- crepancy” group (61 cases) mainly constituted tumors with a WS < 3 (38/61, 62%) (Fig. 3). The K score calcu- lated among all pathologists for the 2 possible options (intact vs. altered reticulin framework) was substantial (K = 0.702), whereas the internal consistency of data was excellent using the Chronbach a (x = 0.954). A sub- division of cases according to the original Institution, pathologists’ experience, ACC histologic variant, WS, and clinical behavior indicated a satisfactory internal consistency of data through the Chronbach a, whereas the K score may assume low values despite the high agreement because of the very small variability between
| # Pathologist | # Cases | Fleiss K | Cronbach a | Overall Agreement (%) | |
|---|---|---|---|---|---|
| First round | 8 | All (#245) | 0.702 | 0.954 (0.938, 0.956) | 75 |
| 8 | FI cases (#61) | 0.602 | 0.931 (0.894, 0.956) | 59 | |
| 8 | PD cases (#42) | 0.753 | 0.964 (0.941, 0.981) | 71.4 | |
| 8 | TO cases (#117) | 0.722 | 0.965 (0.938, 0.980) | 83 | |
| 4 junior | All (#245) | 0.661 | 0.90 (0.868, 0.926) | 88 | |
| 4 senior | All (#245) | 0.73 | 0.917 (0.881, 0.946) | 78 | |
| 7 excluding the one outlier | All (#245) | 0.754 | 0.958 (0.941, 0.971) | 84 | |
| 8 | WS≥3 (#184) | 0.325 | 0.816 (0.773, 0.883) | 88 | |
| 8 | WS <3 (#61) | 0.384 | 0.923 (0.887, 0.945) | 24 | |
| 8 | Classical ACT (#185*) | 0.54 | 0.956 (0.939, 0.969) | 77 | |
| 8 | Purely oncocytic ACT (#26) | 0.467 | 0.943 (0.883, 0.969) | 58 | |
| 8 | Mixed oncocytic ACT (#18) | 0.486 | 0.933 | 83 | |
| 8 | Myxoid ACT (#16) | 0 | 0.546 | 81 | |
| 8 | DOD+AWD cases (#84) | 0.281 | 0.816 (0.557, 0.890) | 92 | |
| 8 | NED cases (#103) | 0.529 | 0.954 (0.938, 0.967) | 61 | |
| Second round | 8 | Discordant in 1st round (#61) | 0.62 | 0.935 (0.912, 0.954) | 44 |
*Including the only one sarcomatoid ACC.
ACT indicates adrenocortical tumor; AWD, alive with disease; DOD, died of disease; FI, Florence cases; NED, not evidence of disease; PD, Padua cases; TO, Turin cases.
and within raters. The 2 pediatric cases, which were discordant in terms of classification when considering Weiss or Wieneke criteria, had an altered reticulin pattern with a concordance at the first round of 7/8 and 8/8, respectively.
During the second round that was restricted to discordant cases at the first screening, performed blindly from the original assessment and after a specific training, a complete agreement among all 8 pathologists was ach- ieved in 44% of cases (K = 0.62, x = 0.935). Among them, all but 4 cases with a WS ≥ 3 reached complete agree- ment. The overall concordance rate after the second round increased to 86%. Concerning cases with qual- itative reticulin framework alterations, 5/13 were dis- cordant in the first round, and an agreement of 13/13 was reached in the second round.
DISCUSSION
In this interobserver reproducibility study, the pro- posed RA approach for the diagnosis of ACC11 has been validated with regard to the first step of this algorithm, specifically that based on the histochemical reticulin stain to detect the presence of reticulin fiber disruptive changes. The results of this study indicate that the RA: (a) is a fast technique and easy-to-interpret system in adrenocortical tumors, when both quantitative and qualitative changes are considered in the evaluation of reticulin framework disruption; (b) has a high interobserver reproducibility, which supports its routine use in the proposed 2-step RA approach for ACC diagnosis.
Although RA for ACC diagnosis is based on a 2-step procedure (reticulin disruption evaluation, fol- lowed by the recognition of 3 Weiss criteria, namely ne- crosis, mitotic count, and venous invasion), we decided to restrict our reproducibility study to the first step of the reticulin histochemical procedure, as it represents a novel
tool in the diagnostic procedure of adrenocortical tumors. In fact, although reticulin staining is an easy-to-perform ancillary staining technique recommended in the diag- nostic practice of several pathologic conditions including endocrine tumors (ie, pituitary16 and parathyroid17), it has not been validated in this specific setting, after the original observation by our group.
Therefore, the ability of different observers to identify reticulin framework alterations was assessed in a large series of unpublished adrenocortical tumors selected from 5 centers and including 184 ACCs (WS 3 to 9, with conventional, myxoid, and oncocytic features) and 61 ACAs (WS 0 to 2). To reproduce daily practice con- ditions, the 5 participant Institutions performed the re- ticulin histochemical staining in-house and contributed the slides to build 1 single set of cases. This was circulated and reviewed by 8 pathologists with a different experience in adrenal pathology (2 residents, 2 junior, and 4 senior consultant histopathologists, all working in one of the participating centers). A substantial overall interpretation agreement was observed (75%, with a k value of 0.702) when each case was categorized between the 2 options of intact versus altered reticulin framework, which is the base of the RA. Interestingly, the overall agreement and K values did not change significantly when a separate stat- istical analysis was performed according to pathologists’ experience or when the set of cases originated from dif- ferent centers, thus confirming an acceptable reproduci- bility of the judgments among observers, among different sources of the tumor tissues, and among reticulin-stained slides prepared in different laboratories.
A detailed analysis of the 61 cases that proved to be discordant (even in only 1 of 8 observers) raised some issues that prompted a second blind review by each par- ticipant, preceded by a training using digital photo- micrographs of representative cases and illustrating the intact reticulin framework of the normal adrenal gland,
pathologist
1
2
3
4
5
6
7
8
0-2 — >
3-9 — >
Weiss score
cases with the “normal adrenal-like” mesh, cases with either quantitatively or qualitatively altered mesh, and finally those with an extensively disrupted reticulin net- work.
The revision process allowed to confirm that cases assigned to the group of “normal adrenal-like” intact reticulin framework eventually were all adenomas ac- cording to the WS and were all clinically benign tumors and that cases having an extensive loss of reticulin fibers with marked disruption and fragmentation of the retic- ulin mesh were associated with the presence of 1 or more of the 3 Weiss criteria and ultimately were reported as ACC. However, within these 2 conditions, a fraction of cases (approximately 6%) had more complex reticulin patterns that during the first round caused some dis- crepancy in the various observers’ interpretation, includ- ing (1) focally disrupted reticulin fibers in an otherwise intact tumor; and (2) an apparently intact reticulin net- work, surrounding single cells or small cell groups, but made of irregularly thickened fibers with markedly frayed borders, superficially resembling the reticulin framework of so called “hemangiopericytoma.”18 The former con- dition was more commonly observed in oncocytic adre- nocortical tumors, whereas the latter was a feature of malignant tumors, in which qualitative (rather than quantitative) changes of the reticulin mesh were repre- sentative of the network disruption. This could reflect different growth and invasive properties of individual tumors in terms of connective tissue support to epithelial cells with either the absence or the reduction of the re- ticulin framework on one side or the production of a largely abnormal reticulin mesh occasionally surrounding single cells, thus mimicking a benign condition, but re-
sulting in an irregularly thickened and frayed network. These observations expanded the proposed criteria for reticulin disruption evaluation11 introducing both quan- titative and qualitative analysis of the reticulin fibers. Incidentally, when the diagnostic algorithm was com- pleted and compared with the classical WS-based diag- nosis and outcome, all such qualitatively disrupted tumors were classified as ACC. Moreover, although het- erogeneity of reticulin staining profiles is frequent but usually presenting a mixture of different patterns (ie, qualitative and quantitative changes), all ascribable to disruption of the reticulin framework, 3 cases presented heterogenous patterns of staining suggestive of central subclonal expansion of a neoplastic population within otherwise normal adrenal-like areas at the periphery of the lesion, thus possibly supporting the hypothesis of an adenoma-to-carcinoma progression, as previously reported.19
The interobserver agreement after the second round increased from 75% to 86% (K = 0.63, x = 0.935), a fig- ure significantly different from the original performance (P < 0.0001) and supporting the notion that an accurate training in reticulin and Weiss criteria evaluation might help to further improve the diagnostic accuracy of ACC both in referral centers and in general surgical pathology divisions. In this respect, it is important to note that a significant reproducibility was maintained also consider- ing junior pathologists only, neither specifically trained nor experienced in adrenal pathology.
Moreover, the reticulin histochemical stain worked well irrespective of the histologic type. In fact, interob- server agreement was not significantly reduced in ACC variants, with special reference to the oncocytic one that was consistently represented in this series. This finding is of relevance because using classical scoring systems, myxoid tumors may be underdiagnosed because of the difficulty of identifying some Weiss parameters, whereas oncocytic adrenal tumors are easily over- diagnosed, because of the intrinsic occurrence of 3 Weiss parameters (eosinophilic cytoplasm, diffuse growth, and atypias). 10,12,20,21
Our detailed analysis of reticulin alterations and reproducibility of staining interpretation in a large cohort of adrenocortical tumors nicely correlates with a recent French reproducibility study of Weiss criteria designed on 45 adrenocortical tumors of the classical type (neither oncocytic nor pediatric or myxoid cases were apparently included). In fact, all 3 Weiss criteria incorporated in the RA system proved highly reproducible among the dif- ferent observers involved in the study.6 In particular, necrosis was the most reproducible criterion with a “substantial” agreement (K = 0.78) at first screening, which increased to an “almost perfect” agreement (K = 0.83) at the second reading round (after a dedicated training). The mitotic count performance significantly increased from a moderate to a substantial agreement after the training (K values of 0.54 and 0.65, respectively). Finally, with regard to the third parameter, venous in- vasion assessment remained a slightly more complicate
task, with an observed moderate agreement (K = 0.54), with no improvement after the second reading.6
In conclusion, the pathologic definition of malig- nancy in adrenocortical tumors plays a relevant role in stratifying patients for further molecular studies and/or for specific treatments. Scoring systems or algorithms for assessing malignancy are the current popular diagnostic tools but need to be refined in order to have homogenous interpretation criteria worldwide. This goal can be reached through the application of strictly controlled parameters, as demonstrated by the good reproducibility of Weiss parameter assessment after precise pathologists’ training.6 The same conclusion was reached in the current study in which good reproducibility of the reticulin staining procedure and interpretation was obtained among 8 observers from 5 centers with different experi- ence in endocrine pathology. These findings encourage the use of the RA, which is a fast, reproducible, cheap, and easy-to-apply method proposed for ACC diagnosis.
REFERENCES
1. Hough AJ, Hollifield JW, Page DL, et al. Prognostic factors in adrenal cortical tumors. A mathematical analysis of clinical and morphologic data. Am J Clin Pathol. 1979;72:390-399.
2. van’t Sant HP, Bouvy ND, Kazemier G, et al. The prognostic value of two different histopathological scoring systems for adrenocortical carcinomas. Histopathology. 2007;51:239-245.
3. Weiss LM. Comparative histologic study of 43 metastasizing and nonmetastasizing adrenocortical tumors. Am J Surg Pathol. 1984;8: 163-169.
4. Weiss LM, Medeiros LJ, Vickery AL Jr. Pathologic features of prognostic significance in adrenocortical carcinoma. Am J Surg Pathol. 1989;13:202-206.
5. Lau SK, Weiss LM. The Weiss system for evaluating adrenocortical neoplasms: 25 years later. Hum Pathol. 2009;40:757-768.
6. Tissier F, Aubert S, Leteurtre E, et al. Adrenocortical tumors: improving the practice of the Weiss system through virtual microscopy: a National Program of the French Network INCa- COMETE. Am J Surg Pathol. 2012;36:1194-1201.
7. Aubert S, Wacrenier A, Leroy X, et al. Weiss system revisited: a clinicopathologic and immunohistochemical study of 49 adreno- cortical tumors. Am J Surg Pathol. 2002;26:1612-1619.
8. Blanes A, Diaz-Cano SJ. Histologic criteria for adrenocortical proliferative lesions: value of mitotic figure variability. Am J Clin Pathol. 2007;127:398-408.
9. Bisceglia M, Ludovico O, Di Mattia A, et al. Adrenocortical oncocytic tumors: report of 10 cases and review of the literature. Int J Surg Pathol. 2004;12:231-243.
10. Papotti M, Volante M, Duregon E, et al. Adrenocortical tumors with myxoid features: a distinct morphologic and phenotypical variant exhibiting malignant behavior. Am J Surg Pathol. 2010; 34:973-983.
11. Volante M, Bollito E, Sperone P, et al. Clinicopathological study of a series of 92 adrenocortical carcinomas: from a proposal of simplified diagnostic algorithm to prognostic stratification. Histo- pathology. 2009;55:535-543.
12. Duregon E, Volante M, Cappia S, et al. Oncocytic adrenocortical tumors: diagnostic algorithm and mitochondrial DNA profile in 27 cases. Am J Surg Pathol. 2011;35:1882-1893.
13. Sanad S. Histopathological and immunohistochemical studies on the adrenal cortical tumors of Egyptian patients. Life Sci J. 2011;8:970-978.
14. de Krijger RR, Papathomas TG. Adrenocortical neoplasia: evolving concepts in tumorigenesis with an emphasis on adrenal cortical carcinoma variants. Virchows Arch. 2012;460:9-18.
15. Wieneke JA, Thompson LD, Heffess CS. Adrenal cortical neoplasms in the pediatric population: a clinicopathologic and immunopheno- typic analysis of 83 patients. Am J Surg Pathol. 2003;27:867-881.
16. Horvath E. Pituitary hyperplasia. Pathol Res Pract. 1988;183:623-625.
17. Papotti M, Volante M, Bussolati G. Applied histochemistry in disorders of the endocrine system. In: Wick MR, ed. Diagnostic Histochemistry. University of Virginia, Charlottesville: Cambridge University Press; 2008.
18. Ambrosini-Spaltro A, Eusebi V. Meningeal hemangiopericytomas and hemangiopericytoma/solitary fibrous tumors of extracranial soft tissues: a comparison. Virchows Arch. 2010;456:343-354.
19. Bernard MH, Sidhu S, Berger N, et al. A case report in favor of a multistep adrenocortical tumorigenesis. J Clin Endocrinol Metab. 2003;88:998-1001.
20. Volante M, Buttigliero C, Greco E, et al. Pathological and molecular features of adrenocortical carcinoma: an update. J Clin Pathol. 2008;61:787-793.
21. Papotti M, Libe R, Duregon E, et al. The Weiss score and beyond- histopathology for adrenocortical carcinoma. Horm Cancer. 2011;2:333-340.