The rate and pattern of bcl-2 and cytokeratin 15 expression in trichoepithelioma and nodular Basal cell carcinoma: a comparative study.

CONTEXT: Histopathological evaluations can differentiate between clinically resembling trichoepithelioma (TE) and basal cell carcinoma (BCC) unless the biopsy specimens are small or superficial. Previous studies used immunohistochemical evaluation for Bcl-2 and cytokeratin 15 (CK15), in attempts to differentiate between these two entities, with heterogeneous findings. AIMS: This study intended to compare the rate and patterns of Bcl-2 and CK15 expressions between specimens of TE and nodular BCC. SETTINGS AND
DESIGN: Case-series including formalin-fixed, paraffin-embedded cutaneous biopsies. SUBJECTS AND METHODS: Twenty-two BCC and 12 TE specimens were stained for Bcl-2 and CK15 and examined microscopically. The rate and patterns of expressions were compared between the two groups. STATISTICAL ANALYSIS USED: Statistical analysis was performed using the statistical software (SPSS version 16.0; SPSS Inc., Chicago, IL, USA), Pearson Chi-square, or Fisher's exact tests, wherever appropriate.
RESULTS: The two groups were comparable for the expression rate and patterns of Bcl-2 (86.4%: 5 central, 14 diffuse in BCC vs. 83.3%: 2 central, 8 diffuse in TE; P = 0.59 and 0.54 for rate and pattern, respectively). The rate of CK15 expression was significantly higher in TE specimens (66.7%: 4 central, 3 diffuse, 1 peripheral vs. 4.5%: 1 central; P < 0.001). The positive likelihood ratio in distinguishing the two neoplasms was 14.7 (95% confidence interval: 2.1-103.7).
CONCLUSIONS: CK15 but not Bcl-2 staining may help in differentiating between BCC and TE even in BCCs with follicular differentiation.

What was known?

CK15 and Bcl-2 may express differently in basal cell carcinoma (BCC) and trichoepithelioma (TE).

The clinical importance of CK15 and Bcl-2 expression in differentiation between BCC and TE is not definite.

The pattern of CK15/Bcl-2 expression in differentiation between BCC and TE is not well understood.

Introduction

Trichoepithelioma (TE) is a benign cutaneous neoplasm, but it may bear a resemblance to another malignant tumor of skin, the basal cell carcinoma (BCC). They both are flesh-colored, slowly growing, pearl-like nodules with a tendency to occur on the face.[12]

On histopathological examination, however, they are usually distinctive entities. A well-circumscribed lesion composed of islands of basaloid cells with smooth margins, clefting within the surrounding stroma, keratinous microcysts, and abortive hair papilla is suggestive of TE. Quite the contrary, a BCC lesion has ill-defined asymmetrical borders and is consisted of clusters of basaloid cells with peripheral palisaded nuclei, extensions to the epidermis, a mixture of single cell or massive necrosis, apoptosis, and brisk mitotic rate as well as the presence of loose fibromyxoid stroma with peritumoral retractions and clefting between tumor islands, and the surrounding mucin-rich stroma.[23] However, histological differentiating between TE and BCC might be occasionally difficult,[45] particularly when the microscopic features are not prominent, or the biopsies are too superficial or small.

In addition, each of TE and BCC expresses diverse histological subtypes and sometimes the microscopic distinction between them is a major diagnostic challenge.[6]

It should be born in mind that the differentiation between TE and BCC is of great clinical significance because their treatment and outcomes are totally different. So, some investigators have attempted to use various markers to distinguish these two neoplasms.[2]

Bcl-2 is a proto-oncogene with a role in apoptosis, expressed in basal keratinocytes, mesenchymal cells of follicular papilla, clear cells of eccrine glands, and melanocytes and lymphocytes reside in skin. High Bcl-2 expression may reflect an increase in cell growth caused by the inhibition of apoptosis.[5]

Keratins, on the other hand, are a group of fibrous proteins that form the cell cytoskeleton. It is assumed that the degree of differentiation of each individual epithelial cell determines the expression rate of keratins.[7]

A 50 kD molecular weight acidic (or type I) keratin polypeptide is cytokeratin 15 (CK15) which is believed to be expressed in the basal layer of the outer root sheath, in basal epidermal keratinocytes, and sebaceous glands, in both human and animal studies.[89] However, it is also claimed that this marker is rarely detected in basal keratinocytes outside the hair follicle.[10111213] TE is composed of follicular stem cells, and so, CK15 expression is expected in the majority of these cases.[1213]

However, variability in the rate and staining pattern of both Bcl-2 and CK15 has been described in the literature among TE and BCC cases, impeding a straightforward implementation of these two markers for differentiation between these two neoplasms.[14151617] We aimed to compare immunohistochemical expression of Bcl-2 and CK15 in cases with facial nodular BCC and TE.

Subjects and Methods

Case selection

Formalin-fixed, paraffin-embedded cutaneous biopsies were selected from the archive of the dermatopathology service at Shahid Beheshti Medical University of Tehran from September 2009 to September 2011.

Twenty-two cases of nodular BCC and 12 cases of TE were randomly selected from these files. The diagnosis in each case was confirmed by an experienced dermatopathologist and a pathologist through routine microscopic examination [H and E, Figure 1] and inter-observer agreement was reached on all cases involved in the study. The histological criteria for BCC and TE were as described above.[23] All the lesions were located on the face. This study was approved by the Ethics Committee of Shahid Beheshti Medical University of Tehran.

Figure 1

Trichoepithelioma (TE) (a) and nodular BCC (b) (H and E, original magnification ×4)

Immunohistochemistry

Three micrometer thick sections were first cut, deparaffinized, and rehydrated. After antigen retrieval following the manufacturer's instructions, these tissue sections were incubated with antibodies against Bcl-2 (Clone: Bcl-2/100/D5; Novocastra, Newcastle, UK) for 20 min at room temperature and CK15 (Lyophilized Mouse Monoclonal Antibody Cytokeratin 15, Clone: LHK15; Novocastra) for 60 min at 25°C. The working dilution of CK15 antibody was 1/20, whereas Bcl-2 was a ready-to-use preparation. Reactivity was detected by avidin – biotin complex technique.[18] Diaminobenzoic acid was used as the chromogen for color development, following with hematoxylin staining.

Positive controls were run in concordance with each antibody to ensure adequacy of immunostaining techniques, including tonsillar lymphoid tissue for Bcl-2 and bulge region of the scalp normal hair follicles for CK15. The negative tissue control was cerebellum.

The patterns of expression of Bcl-2 and CK15 proteins were determined by one dermatopathologist and a pathologist blinded to the previous diagnosis and accordingly, all cases were categorized into four groups based on the dominant pattern of tumor staining: Centrally positive, peripherally positive, diffusely positive, and negative [Figure 2]. The dermatopathologist and pathologist were also blinded to the markers used for each specimen.

Figure 2

Bcl-2 and cytokeratin 15 (CK15) expressions in basal cell carcinoma (BCC) and trichoepithelioma (TE) [H]. A central pattern of Bcl-2 expression in (a) BCC and (b) TE. A diffuse pattern of Bcl-2 expression in (c) BCC and (d) TE. A central pattern of CK15 expression in (e) BCC and (f) TE. (g, h) Diffuse and peripheral patterns of CK15 expression in TE (original magnification ×40 in (g) and ×10 in others)

Statistics

Differences in Bcl-2 and CK15 staining patterns in TE and BCC groups were tested using Pearson Chi-square or Fisher's exact tests, wherever appropriate. For all statistical tests, exact P values were computed. Statistical analysis was performed using the statistical software SPSS version 16.0 (SPSS Inc.). P ≤ 0.05 were considered statistically significant. Sensitivity and specificity as well as positive and negative likelihood ratios were also calculated for both parameters.

Results

In this study, 19 cases with nodular BCC (86.4%) were positively stained for Bcl-2. In the TE group, 10 cases (83.3%) expressed positivity for this marker [Figure 3]. The rate of Bcl-2 staining was comparable between the two groups (P = 0.59; Fisher's exact test).

Figure 3

Rate of Bcl-2 expression in histological groups of trichoepithelioma (TE) and BCC

The patterns of Bcl-2 expression in the BCC group were central in 5 cases (26.3%) and diffuse in 14 cases (73.7%). In the TE group, two cases (20%) expressed Bcl-2 centrally and eight cases (80%) diffusely [Table 1]. The two groups were comparable in this regard (P = 0.54; Fisher's exact test).

Table 1

Status and patterns of Bcl-2 and cytokeratin 15 expression in histological groups of trichoepithelioma and basal cell carcinoma

Using the Bcl-2 expression as a tool to differentiate between nodular BCC and TE, the sensitivity was 86.4% (95% confidence interval: 65.1-97.1), the specificity was 16.7% (95% confidence interval: 2.1-48.4), the positive likelihood ratio was 1.0 (95% confidence interval: 0.8-1.4), and the negative likelihood ratio was 1.2 (95% confidence interval: 0.2-6.3).

One case out of 22 nodular BCC specimens (4.5%) was positively stained for CK15, whereas eight cases out of 12 TE specimens (66.7%) expressed positivity for this marker [Figure 4]. The rate of CK15-stained cases was significantly higher in the TE group than in the nodular BCC group (P < 0.001, Pearson's Chi-square test).

Figure 4

Rate of cytokeratin 15 expression in histological groups of trichoepithelioma (TE) and BCC

The only positively CK15-stained case in the BCC group showed this expression centrally. In the TE group, the patterns were central in four cases (50%), peripheral in one case (12.5%), and diffuse in three cases (37.5%) [Table 1].

Considering the CK15 expression to distinguish nodular BCC from TE, the sensitivity was 66.7% (95% confidence interval: 34.9-90.1), the specificity was 95.5% (95% confidence interval: 77.2-99.9), the positive likelihood ratio was 14.7 (95% confidence interval: 2.1-103.7), and the negative likelihood ratio was 2.9 (95% confidence interval: 1.3-6.4).

The only CK15-stained case in the BCC group was also positive for Bcl-2. In the TE group, seven cases were stained positively for both Bcl-2 and CK15 simultaneously [Table 2].

Table 2

Status of simultaneous expression of Bcl-2 and cytokeratin 15 in histological groups of trichoepithelioma (TE) and basal cell carcinoma

Discussion

A usual histopathological profile of TE comprises distinctive islands of basaloid cells, clefting within the surrounding stroma, abortive hair papilla, and keratinous microcysts. In a typical BCC lesion, however, the clusters of basaloid cells are irregular, within each, there are traces of epidermis, as well as a mixture of single cells, necrosis, apoptosis, frequent mitosis, loose fibromyxoid stroma with peritumoral retractions and clefting between tumor islands, and the surrounding mucin-rich stroma.[23]

Although it is not difficult to distinguish between BCCs and TEs based on these histopathological features, in some cases, especially when obtained biopsies are superficial or inadequate, both entities may be extremely similar. Due to the quite different management and prognosis of BCC and TE, it is pivotal to search for a way that could reliably differentiate between the two neoplasms.

Immunohistochemical assessment, especially for Bcl-2, has been tried by previous studies in an attempt to differentiate between BCC and TE; but the results have been conflicting.[36192021]

Although some investigations have claimed that this marker could be efficiently used for distinguishing between the two entities,[26] others pointed out that their findings provided little support to the contention that Bcl-2 immunostains are useful for adjunctive histological differentiation of TE from BCC.[36]

Smoller et al.[6] showed a significant difference in the pattern of immunohistochemistry staining of Bcl-2 reactivity, which was predominantly peripheral in TE versus diffuse in BCC.

Poniecka and Alexis[2] evaluated 10 specimens of TE and 20 specimens of BCC. Bcl-2 stained all but one BCC specimen in a diffuse pattern, whereas all TEs showed staining of the outermost epithelial layer. They concluded that Bcl-2 may be of some value in distinguishing BCC from TE, limited by the quantitative nature of the difference in staining.

Diffuse pattern of Bcl-2 expression in BCC and presence of strongly stained peripheral cells with Bcl-2 staining in TE has been reported by other investigations, as well.[1519]

Abdelsayed et al.[22] compared cellular expression of Bcl-2 between 20 cases of TE and 20 cases of BCC occurred on the skin of the face and neck. They also showed similar quantitative and qualitative expressions of Bcl-2 in these two neoplastic groups. The positive cells were distributed generally in the neoplastic islands in both entities. The lack of consistency between findings in these studies reflects the complexity in the distinction between these neoplasms and prompts our investigation. Furthermore, this examination has not been substantiated with immunohistochemical evaluation of other markers such as CK15.

Based on our findings, neither the rate of Bcl-2 expression (83.3% in the TE group vs. 86.4% in the BCC group, P > 0.05; sensitivity = 86.4%, specificity = 16.7%, positive likelihood ratio = 1.0, and negative likelihood ratio = 1.2) nor its pattern was significantly different between the two neoplasms.

Considering the role of gene products of Bcl-2 which prolong cell survival, the similarity between BCC and TE in terms of the expression of Bcl-2 suggests that they possess similar pathobiological alterations of cellular apoptotic mechanisms.

Simultaneously, we showed that the rate of CK15 expression was significantly higher in the cases with TE than those with BCC (66.7% vs. 4.5%, P < 0.001; sensitivity = 66.7%, specificity = 95.5%, positive likelihood ratio = 14.7, and negative likelihood ratio = 2.9). It should be noted that in statistics, a positive likelihood ratio > 10 is a convincing diagnostic evidence and a value between 5 and 10 is regarded to provide high diagnostic evidence.[23]

To the best of our knowledge, this study is the first to concomitantly evaluate Bcl-2 and CK15 expressions in TE and nodular BCC. There is still ongoing debate on the usefulness of CK15 in differentiating TEs from BCCs.

In a study by Kanitakis et al.,[12] the reactivity of an antibody to the CD8 antigen, recognizing CK15, was tested in eight specimens of TE and 17 specimens of BCC. Based on the findings, the rate of reactivity was 62.5% in the first group, whereas all BCCs were completely unreactive. They suggested that BCCs do not differentiate toward hair bulge cells, and from a practical point of view, immunostaining for CK15 is apparently an additional useful adjunct for the differential diagnosis between BCC and TE.

Jih et al.[13] screened a series of TEs (n = 13) and BCCs (n = 37) by an anti-CD8 monoclonal antibody that cross-reacts with CK15. All TE (100%) and a subset of BCC specimens (27%) were reported to be CK15-positive. They concluded that expression of CK15 in TEs and some BCCs suggests that these tumors are related to hair follicle stem cells in the bulge, which are thought to be important in tumorigenesis.

In contrast, Choi et al.[24] showed that although there was no significant difference in the CK15 positivity rate between TE (73% out of 11 specimens) and BCC (71% out of 17 specimens), the staining pattern was significantly different between the two groups (peripheral and central patterns: 87.5% and 12.5% in the TE group vs. 16.7% and 83.3% in the BCC group, respectively). They assumed that not only TE but also a considerable number of BCCs originate from hair follicle bulge cells.

Our findings are in line with the first two reports. The pattern of CK15 expression, however, was not assessable because of the low rate of CK15-positive cells in the BCC group.

It is still a major challenge to identify the exact origin of neoplasms in the field of cancer biology.[25]

Among these neoplasms, the non-melanoma skin cancers are very common, and the great majority of these tumors are BCCs.[26]

The staining differences between TE and BCC may relate to the distinct origin and biological behavior of the two tumors and may therefore be of value in subsequent patient management.[19]

Different patterns of staining may originate from cells arising from other areas, skin adnexa, or the epidermis; or may be simply indicative of lost expression of that marker during tumor development and progression.[24]

Some studies underscore the morphological and biochemical similarities between human hair follicles and BCCs,[131427] to some extent that others have proposed BCCs as follicle-derived tumors.[2829]

Unlike superficial BCC, histological sections of nodular BCC have shown no connection to the overlying epidermis, suggesting that this subtype may arise from epithelial cells in hair follicles with deep extension into the dermis.[3031] However, these studies have not excluded the possibility of tumor expansion where tumor initiation occurs in another compartment (s), but proliferative expansion into a detectable tumor can only take place when cells have migrated into the hair follicle. As mentioned before, all our patients in the BCC group suffered from nodular subtype. In contrast to superficial BCCs, several previous reports point to the follicle as a potential site of origin for nodular BCC in mouse[3031] and rat,[32] but the involvement of follicle stem cells in the development of these tumors was not addressed. In addition, Córdoba et al.[33] showed that Bcl-2 can distinguish between BCC and trichoblastoma only if the former is not with follicular differentiation. This indirectly suggests that at least many of our cases might be with follicular differentiation, and the findings in terms of the CK15 presentation indicate that a follicular differentiation would not hinder its usefulness in differentiating these tumors from TEs.

We did not use a normal control group in this study and it can be acknowledged as a limitation of this study. However, it was previously shown that the basal layer of normal epidermis is usually negative for CK15.[11] Finally, like in any other dermatological pathology, the role of possible confounding factors such the skin type[3435] and age of the patients[36] should be investigated in further studies

In summary, our results suggest that CK15 but not Bcl-2 staining may be an additional useful adjunct for the differentiation between BCC and TE, since CK15 expression would strongly favor the diagnosis of TE versus BCC, even in those with follicular differentiation. The pattern of expression is not of clinical importance.

What is new?

CK15 but not Bcl-2 staining is possibly useful for the differentiation between basal cell carcinoma (BCC) and trichoepitheliomas (TE).

CK15 expression strongly favors the diagnosis of TE versus BCC, even in those with follicular differentiation.

The pattern of CK15 expression is not of clinical importance.