Differentiation between basal cell carcinoma and trichoepithelioma by immunohistochemical staining of the androgen receptor: an overview

ARTICLE

ejd.2011.1504

Auteur(s) : Aimée H.M.M. Arits^1,2 a.arits@MUMC.nl, Arienne M.W. Van Marion³, Bjorn G.P.M. Lohman⁴, Monique R.T.M. Thissen^1,2, Peter M. Steijlen^1,2, Patty J. Nelemans⁵, Nicole WJ Kelleners-Smeets^1,2

¹ Department of Dermatology, Maastricht University Medical Centre, P.O. box 5800, 6202 AZ Maastricht, The Netherlands

² GROW Research Institute for Oncology and Developmental Biology, Maastricht University; Maastricht, The Netherlands

³ Department of Pathology, VieCuri Medical Centre, Venlo, The Netherlands

⁴ Department of Pathology, Maastricht University Medical Centre, Maastricht, The Netherlands

⁵ Department of Epidemiology, Maastricht University Medical Centre, P.O. box 5800, 6202 AZ Maastricht, The Netherlands

Reprints: A. H.M.M. Arits

Basal cell carcinoma (BCC) is the most common cancer in Caucasians. Recently evidence has become available that BCC is an epithelial tumour arising from the progenitor cells of the interfollicular epidermis and the upper infundibulum [1]. Histologically there are three main types of the BCC: Superficial, nodular and infiltrative [2]. The highest incidence of BCC is seen in the middle-aged, with preference for the sun exposed body sites [3]. Trichoepithelioma (TE) is another epithelial tumour originating from the outer root sheath of the hair follicle, but in contrast to BCC, this is a benign tumour with clear follicular differentiation. There are two types: the classic type and the desmoplastic type [4]. TE mainly occurs in young adults with predilection sites such as the nose, the upper lip and the cheeks [4]. Clinical distinction between TE and BCC is difficult in some cases and this dilemma may extend to a microscopic level as well. Distinction between the two neoplasms is important because they have different biological behaviour and need different treatment. On haematoxylin & eosin (H&E) stained slides, both tumours are composed of nests of basaloid cells with a degree of follicular differentiation. The histological differentiation between BCC and TE has been predominantly based on this degree of follicular differentiation. A high degree of follicular differentiation within the tumour, favours a benign tumour like TE. Other typical histological TE characteristics are the presence of primitive epithelial structures resembling hair papillae (known as papillary-mesenchymal bodies), the presence of small keratinous cysts and cleft formation in the peritumoral stroma [4]. Conversely, BCC is characterized by cleft formation between tumour and stroma, peripheral palisading of basaloid keratinocytes, ulceration, inflammatory response, mitotic figures, necrosis and peritumoral mucin production [4, 5]. Histological confusion is mostly seen between a nodular BCC and a classic TE (cTE) or between an infiltrative BCC and desmoplastic TE (dTE). In the past decade many studies have shown the relevance of several additional immunomarkers in differentiating between the two entities. As we discussed in an earlier case study, the usefulness of most of these antibodies, like oncogenes (Bcl-2), cell membrane glycoproteins (Ber-EP4, CD10 and CD34) and cytokeratins, was disappointing [5].

Human sebaceous glands and hair follicles are target structures in the skin for androgen action. They contain steroid enzymes, capable of transforming weak androgens into the target-tissue-active androgens testosterone and dihydrotestosterone, which bind to the androgen receptor (AR) to regulate cellular transcription [6]. The detection of AR antibodies seems to be a promising tool in the differentiation between BCC and TE [5]. In normal skin, AR is expressed in sebaceous glands, pilosebaceous duct keratinocytes, interfollicular epidermal keratinocytes, dermal fibroblasts and certain cells of the secretory coils of eccrine sweat glands [7-9]. Recently it has been assumed that AR is also expressed in some cutaneous neoplasms like BCC [7]. In contrast, AR expression seems to be absent in mature hair follicles and the epidermis, thus also in benign hair follicle tumours like TE [9]. We discuss the usefulness of AR immunohistochemistry in differentiating BCC from TE, based on our own findings and the results of previous studies.

Materials and methods

Case selection

This study was performed based on the code of proper behaviour of tissue. This means that in the process of collecting patient specimens, material was de-identified and was untraceable to the patient himself. From the Pathologic Anatomical National Automated Archive of the Department of Pathology in the Maastricht University Medical Centre (MUMC), we randomly selected 75 H&E slides of BCC and TE. This archive is a database recording all reports of pathological diagnoses and marks them with a unique number. By using a combination of search terms, reports of specific histological diagnosis can be retrieved. With this unique number, the representive paraffin blocks can be found in the pathology archives of the MUMC. New H&E-stained slides and immunohistochemical slides with AR antibody staining were created from these blocks.

The available 75 H&E slides were reviewed by three independent investigators [i.e. a dermato-pathologist (AM), an oncologic-dermatologist (MT) and a resident in dermato-pathology (BL), using predetermined characteristics. Four slides were excluded because no representive tumour could be found in the H&E slide. This was due to further cutting up of the paraffin blocks to create two new consecutive slides for H&E and immunostaining. The seventy-one slides left were reviewed based on the presence or absence of the following characteristics: tumour-stroma cleft formation, ulceration, epithelial primitive structures, small keratinous cysts, inflammatory response, mitosis, necrotic tumour cells, papillar mesenchymal bodies, stromal oedema and peritumoral mucin production. Fifty-six slides (38 BCC and 18 TE) with unequivocal histological characteristics of either tumour were used for immunohistochemistry with AR antibodies. Based on the morphological classification mentioned by Rippey 1998, the investigators distinguished three subtypes of BCC: superficial, nodular and infiltrative. The infiltrative BCC included the morphoeic type. (2) TE was divided in a classical and a desmoplastic type.In addition we searched PubMed, a service of the US National Library of Medicine including citations of MEDLINE and other life science journals and online books, to find other publications on this topic. To date we found three studies examining the role of AR expression in differentiating between BCC and TE by using immunohistochemical staining [10-12]. We will also discuss the results of these studies.

Immunostaining protocol

We included biopsy as well as excision specimens. First, these were fixed in 4% formalin and embedded in paraffin blocks of which several four micron sections were cut. Staining with AR was based on the standard immunoperoxidase technique after deparaffinisation. Sections were fixed on Dako silanized slides to accommodate alkaline retrieval by heat treatment at PH 9.5 for three minutes in Borg Decloker (Biocare Medical, Walnut Creek, CA). The tissue was stained for 20 minutes with a monoclonal antibody AR441 (Dako, Carpinteria, CA) (dilution 1:50). Afterwards, the sections were incubated for 20 seconds with a secondary reagent; envision Flex with mouse (linker) followed by labelled polymer detection (envision Flex/HRP). Colour reaction was developed by immersing slides in a solution of substrate chromogen (substrate working solution) up to 10 minutes, with the reaction being monitored microscopically. Sections of human prostate were used as a positive control. The positive staining of sebaceous tissue was used as internal control. Any nuclear expression of AR was interpreted as positive staining. Previous studies on this topic used the same clone of AR antibody and used almost the same immunohistochemical staining procedure as explained above.

Results

Eighteen TE and thirty-eight BCC were included. Eight TE were of the classic type and ten of the desmoplastic type. Ten superficial BCC, thirteen nodular BCC and fifteen infiltrative BCC were included. The other patient and tumour characteristics are mentioned in table 1.

Table 1 Tumour characteristics.

^* Data are presented as percentages (no.).

Androgen receptor expression

Analyses comparing AR expression in nodular/superficial BCC versus cTE showed a sensitivity of 96% (22/23) for detection of BCC (figure 1A-D), a specificity of 38% (3/8) (figure 2A, B) and a positive and negative predictive value of 82% (22/27) and 75% (3/4) respectively (table 2).

Table 2 Relation between androgen receptor immunostaining and histological diagnosis of basal cell carcinoma and trichoepithelioma.

BCC: basal cell carcinoma; (c/d)TE: (Classic/desmoplastic) trichoepithelioma; sBCC: superficial BCC; nBCC: nodular BCC; iBCC: infiltrative BCC.

The data in parentheses are numbers used to calculate proportion.

Analyses comparing AR expression in infiltrative BCC versus dTE showed a sensitivity of 67% (10/15) for detection of BCC (figure 1E, F), a specificity of 100% (10/10) (figure 2C, D) and a positive and negative predictive value of 100% (10/10) and 67% (10/15) respectively (table 2).

One of the superficial BCC and five of the infiltrative BCC were negative for AR immunostaining. No correlation between the presence of trichodifferentiation (i.e. epithelial primitive structures, small keratinous cysts and papillar mesenchymal bodies) and negative AR immunostaining in BCC was found.

Results of previous studies

Izikson et al. were very conclusive. None of the cTE they studied showed positive AR immunoreactivitiy, whereas, in 78% of the BCC, at least focal AR expression was detected [11]. In two other papers the role of AR antibodies in differentiating between a dTE and morpheaform BCC (mBCC) (i.e. one of the aggressive subtypes collectively referred as infiltrative) was studied. AR expression was seen in 13% of the dTE, compared to 65% of the mBCC in the study by Katona et al. [12]. Whereas Costache et al. showed consistent AR expression in mBCC and no expression in dTE [10]. The results of these three studies are shown in table 3.

Table 3 Review of studies assessing the use of the androgen receptor in differentiation between trichoepithelioma and basal cell carcinoma.

AR: androgen receptor; BCC: basal cell carcinoma; cTE: classic trichoepithelioma; dTE: desmoplastic trichoepithelioma.

The data in parentheses are numbers used to calculate proportion.

Discussion

This study shows that the presence or absence of antibodies against AR in the tumour, as shown by immunohistochemistry, can be helpful in differentiating between BCC and TE.

First we need to explain how this difference in AR immunostaining between BCC and TE can happen. Probably because the two tumours originate from different stem cells. Youssef et al. studied a mouse model to identify cells at the origin of BCC [1]. They found that BCC arises from long-term resident progenitor cells of the interfollicular epidermis and the upper infudibulum rather than the hair follicle bulge stem cells, as suggested earlier [1]. These resident progenitor cells are also responsible for the maintenance of sebaceous glands. As mentioned in the introduction, AR is expressed in normal skin among other things in interfollicular epidermal keratinocytes, which arise from the same stem cells as BCC [1, 7]. This may explain AR presence in BCC. Instead, TE is an epidermal tumour originating from the outer root sheath of the hair follicle which arises from the multipotent stem cells in the bulge [13]. Choudhry et al. showed AR absence in the germinative matrix, outer root sheath (including the bulge region), inner root sheath, hair shaft and hair bulb [8], suggesting the absence of AR in TE.

We have already mentioned that we used a panel of three independent investigators as a gold standard test to evaluate the H&E slides of the tumours. An important thing to notice is that in 1/5 cases diagnosis could not be made unanimously by these observers. Lack of this gold standard H&E test in previous studies on this topic could imply that the results of those three studies[10-12] are less reliable. Furthermore, it justifies our study on this topic. Our study showed, comparable to the studies by Costache et al. and Katona et al., that AR immunostaining is a test with a 100% positive predictive value (PPV) and a 100% specificity for the detection of dTE [10-12]. This means that there are no false positive cases. On the other hand, we showed AR immunostaining to be a diagnostic test with a high sensitivity (96%) for the detection of superficial/nodular BCC. This outcome is consistent with the data shown by Izikson et al., showing a high sensitivity in combination with a high specificity as well [11].

In most cases the histopathological diagnosis of a BCC is clear, however in 20% of cases, confusion with a TE occurs. In these cases immunohistochemistry with AR antibodies can be helpful. Our results suggest a potential role for AR immunohistochemical staining in differentiating TE from BCC; especially desmoplastic TE from infiltrative BCC.

Disclosure

Acknowledgments: All authors participated sufficiently in the work and take public responsibility. Financial support: Drs A.H.M.M. Arits and N.W.J. Kelleners-Smeets are financed by a grant of The Netherlands Organization for Scientific Research ZONMW (08-82310-98-08626). Conflicts of interest: none.

References

1. Youssef KK, Van Keymeulen A, Lapouge G, et al. Identification of the cell lineage at the origin of basal cell carcinoma. Nat Cell Biol; 12: 299-305.

2. Rippey J.J. Why classify basal cell carcinomas?. Histopathology 1998 ; 32 : 393-398.

3. de Vries E, Louwman M, Bastiaens M, de Gruijl F, Coebergh J.W. Rapid and continuous increases in incidence rates of basal cell carcinoma in the southeast Netherlands since 1973. J Invest Dermatol 2004 ; 123 : 634-638.

4. Weedon D. Skin Pathology. Tokyo : Churchill Livingstone, 2001.

5. Arits AH, Parren LJ, van Marion AM, Sommer A, Frank J, Kelleners-Smeets N.W. Basal cell carcinoma and trichoepithelioma: a possible matter of confusion. Int J Dermatol 2008 ; 47 : Suppl. 113-17.

6. Sawaya M.E. Purification of androgen receptors in human sebocytes and hair. J Invest Dermatol 1992 ; 98 : 6 Suppl.92S-96S.

7. Bayer-Garner IB, Givens V, Smoller B. Immunohistochemical staining for androgen receptors: a sensitive marker of sebaceous differentiation. Am J Dermatopathol 1999 ; 21 : 426-431.

8. Choudhry R, Hodgins MB, Van der Kwast TH, Brinkmann AO, Boersma W.J. Localization of androgen receptors in human skin by immunohistochemistry: implications for the hormonal regulation of hair growth, sebaceous glands and sweat glands. J Endocrinol 1992 ; 133 : 467-475.

9. Shikata N, Kurokawa I, Andachi H, Tsubura A. Expression of androgen receptors in skin appendage tumors: an immunohistochemical study. J Cutan Pathol 1995 ; 22 : 149-153.

10. Costache M, Bresch M, Boer A. Desmoplastic trichoepithelioma versus morphoeic basal cell carcinoma: a critical reappraisal of histomorphological and immunohistochemical criteria for differentiation. Histopathology 2008 ; 52 : 865-876.

11. Izikson L, Bhan A, Zembowicz A. Androgen receptor expression helps to differentiate basal cell carcinoma from benign trichoblastic tumors. Am J Dermatopathol 2005 ; 27 : 91-95.

12. Katona TM, Perkins SM, Billings S.D. Does the panel of cytokeratin 20 and androgen receptor antibodies differentiate desmoplastic trichoepithelioma from morpheaform/infiltrative basal cell carcinoma?. J Cutan Pathol 2008 ; 35 : 174-179.

13. Demirkesen C, Hoede N, Moll R. Epithelial markers and differentiation in adnexal neoplasms of the skin: an immunohistochemical study including individual cytokeratins. J Cutan Pathol 1995 ; 22 : 518-535.