Trichoblastoma of the Breast Detected by Screening Mammography: A Diagnostic Pitfall

Abstract

To the Editor: We report 2 cases of trichoepithelioma/trichoblastoma detected at breast screening mammography. The first patient was a 50-year-old woman with a mammographically detected 6 mm, well-defined, nodule with “benign looking” calcification, situated in the lower medial quadrant of the left breast. Needle core biopsies showed a cribriform and partly adenoidal, mitotically inactive, basaloid tumor set in a relatively abundant collagenous stroma. Foci of rounded microcalcification were associated with keratocysts. A provisional diagnosis of basal cell carcinoma was later reviised to a probable trichoepithelioma (cribriform trichoblastoma) and the diagnosis was confirmed after conservative excision biopsy. A second, 58-year-old, woman had a mammographically “suspicious but probably benign” superficial lesion in the left breast. Needle cores showed a mitotically active basaloid tumor comprising nodules and strands in a collagenous stroma. The provisional diagnosis of invasive ductal adenocarcinoma was revised to possible basal cell carcinoma after unusual oestrogen receptor staining. Dermatopathologic review of the cores showed the presence of papillary mesenchymal bodies, and reactive Merkel cells supporting a diagnosis of trichoblastoma. The excision specimen, initially reported as a malignant trichoblastic carcinoma, was revised (on dermatopathologic review) to a benign subcutaneous trichoblastoma with implantation in the needle biopsy tract. Our cases highlight the need to consider cutaneous tumors in mammographically detected, suspicious, superficial lesions, and the importance of expert dermatologic review. Trichoblastoma (including trichoepithelioma) is a relatively uncommon benign skin tumor that shows overlapping histologic features with basal cell carcinoma (BCC) and, in our experience, is a frequent cause of diagnostic difficulty.7 We report 2 cases of mammographically detected trichoblastoma that caused diagnostic difficulty in needle core biopsies, leading initially to a false impression of malignancy. We discuss the diagnosis of trichoblastoma and the usefulness of immunohistochemistry with particular reference to the difficulty of diagnosis in needle core specimens. REPORTS Case 1 A 50-year-old woman was found, at screening mammography, to have a 6-mm well-defined mass in the lower medial quadrant of the left breast, 6 cm from the nipple. The mass contained radiographically “benign looking” microcalcification. Fine needle aspiration biopsy was acellular (noncontributory). Stereoscope-guided needle core biopsies showed a mitotically inactive basaloid tumor with cribriform and adenoidal areas set in a relatively abundant paucicellular collagenous stroma (Fig. 1). Cellular pleomorphism was minimal. Several foci of rounded microcalcification were present in association with keratin-filled microcysts (keratocysts). Initially the lesion was interpreted as a possible basal cell carcinoma, but the diagnosis was revised to a probable trichoepithelioma (cribriform trichoblastoma) that seemed to be centered in the dermis. After discussion with the breast surgeon, we advised a conservative excision. Excision biopsy confirmed a residual relatively small trichoepithelioma in the dermis and upper subcutis that “shelled-out” at surgery. The patient is well, with no evidence of recurrence at 8 years follow-up.FIGURE 1.: Tumor buds in a desmoplastic stroma.Case 2 A 58-year-old woman with learning difficulties underwent screening mammography and a superficial “suspicious but probably benign” mass was identified in the 6 o'clock position of the left breast. A needle core biopsy was initially reported as an “invasive adenocarcinoma, probably grade 2, ductal of no special type.” The reporting pathologist noted an odd distribution of oestrogen receptor staining and similarity to basal cell carcinoma and asked for dermatopathologic review. The needle core consisted of a basaloid tumor with variable large and small nodules and irregular random branching and budding set in an evenly cellular collagenous stroma (Figs. 2A, B). The stroma in places became more cellular around the epithelial lobules and several well-formed papillary mesenchymal bodies were noted (Figs. 2D, E, 4B, C), confirmed with CD10 staining (Fig. 3C). The epithelium composed of small basaloid cells with marked nuclear crowding and foci of mitotic “hot spots,” but overall a relatively low mitotic count (1 to 2 per 10 high-power fields, Fig. 2C). Apoptotic debris was scattered in most lobules. There were no well-formed keratocysts, but distinctive small rounded balls of paler cells (trichogerminomatous pattern) were noted. No microcalcification, and no matrical or inner root sheath keratinisation (trichohyaline granules) was evident.FIGURE 2.: Case 2: Trichoblastoma in needle core biopsy from breast. A and B, Low power of needle core biopsies shows variably sized and irregular basaloid tumor nests and cords set in a collagenous stroma (A, H&E ×20. B, H&E ×50). C, Crowded basaloid cells with 3 mitotic figures in this high-power field but a relative lack of cellular pleomorphism. Individual tumor cell necrosis/apoptotic cells clearly evident (H&E×400). D (H&E×200) and E (H&E×400), Evenly cellular stroma with condensations of small dark papillary mesenchymal cells (arrows) indenting basaloid nests.FIGURE 3.: Case 2: Trichoblastoma in needle core biopsy from breast -immunohistochemistry. A, BerEP4 showing rather weak focal staining only (×50). B, CK20. Numerous reactive Merkel cells throughout the basaloid epithelium (×50) with dendritic processes evident (Inset ×400). C, CD10 negative in the basaloid epithelium except for weak staining adjacent to a well-formed papillary mesenchymal body (arrow) that stains strongly (×100, Inset ×400).Oestrogen receptor staining showed moderate expression in 10% of nuclei mostly aligned at the periphery of the tumor lobules. BerEP4 staining was focally moderate in only 10% of the tumor (Fig. 3A), but reactive Merkel cells were widely distributed throughout the lesion (CK20 Fig. 3B, and Cam5.2). Although the diagnosis of trichoblastoma (benign skin adnexal tumor) was rendered, a mastectomy and axillary node clearance was undertaken after discussion. The lesion in the mastectomy specimen was initially reported as favoring a malignant trichoblastic tumor with focal infiltration. On expert dermatopathological review the diagnosis of a benign subcutaneous trichoblastoma, having some trichogerminomatous areas, was reaffirmed. There was a focus of tumor implantation beyond the circumscribed border of the lesion within the reaction to the needle biopsy tract (Figs. 4A, D).FIGURE 4.: Case 2. Trichoblastoma of breast. Excision specimen with implantation of tumor in the needle biopsy tract. A, Montage from excision specimen showing a subcutaneous trichoblastoma composed of irregular basaloid lobules in a collagenous stroma (H&E×20). Note sharp demarcation between tumor stroma and adjacent dermis and subcutis. B, Papillary mesenchymal body [arrow (H&E×400)]. C, CD10 highlighting tumor stroma and papillary mesenchymal body (arrow) but negative in the tumor epithelium (H&E×200). D, Closer view of tumor implanted in the needle core tract [seen also at the bottom in A (H&E×100)].DISCUSSION Trichoblastomas typically occur on the face in adults but other favored sites include the perianal area where giant forms up to 10 cm in diameter may occur. Clinically lesions form flesh-colored papules, nodules, or plaques usually less than 2 cm in diameter. Our cases are unusual in being clinically asymptomatic, subcutaneous nodules in the breast, detected radiographically at screening mammography. Trichoblastoma has only rarely been reported in the skin of the breast.20 Until recently the detection of trichoblastoma in the breast by mammography has not been reported in the radiology literature, although the finding of superficial epidermal lesions by this technique is well recognized and can be confirmed by oblique views.8 The term “trichoblastoma” was introduced by Headington to describe a range of benign hair follicle tumors.9 Histologically, trichoblastoma is a basaloid tumor that has a circumscribed profile generally with pushing or rounded borders. The basaloid epithelium of this tumor may show a great variation in architectural pattern. Ackerman et al classified trichoblastomas by their histologic pattern as: cribriform (sieve-like, classical trichoepithelioma); small nodular, large nodular, retiform (irregular cords and trabeculae), racemiform (interconnected grape-like formations), columnar (desmoplastic trichoepithelioma), and adamantinoid (including cutaneous lymphadenoma). Like some others we consider that the fibroepithelial tumor of Pinkus is a fenestrated variant of trichoblastoma.5,7 The so-called rippled pattern of trichoblastoma2,18 is now considered to be a variant of sebaceoma.12,14 This variability in tumor architecture in a needle core biopsy can easily be confused with the irregular growth of a carcinoma. The tumor lobules of trichoblastoma are usually set in a quite even, although often rather collagenous or myxoid, stroma, recapitulating the perifollicular mesenchyme of hair follicles. The prominent collagenous stroma might also add to confusion with a carcinoma unless attention is paid to the rather “organized” or evenly cellular appearances. Dermal tumors may be polypoid and have a collarette, but are not normally connected with the epidermis (in contrast to most BCC). Trichoblastomas usually arise, at least in part, within the dermis but may be entirely subcutaneous as in our second patient. A helpful diagnostic clue, and almost invariable feature, is the presence of clefts in the stroma between adjacent lobules of tumor and at the stromal interface with the adjacent dermis or subcutis.1,7 This important diagnostic feature cannot be assessed in a needle core biopsy so that careful attention to other features is required for preoperative diagnosis. Papillary mesenchymal bodies, collections of small dark stromal nuclei abutting with the basaloid epithelium, are a highly characteristic feature of trichoepithelioma/trichoblastoma4,6 and are present in most cases. The presence of papillary mesenchymal cells supported a confident diagnosis in both our cases. Additional features that may be present include inner root sheath differentiation (bright orange trichohyaline granules), matrical keratinisation, prominent infundibular keratocysts (seen in case 1 but not case 2) and sebaceous differentiation. All these features may be seen occasionally in basal cell carcinoma. Trichoblastoma does not usually display widespread pronounced peripheral palisading, retraction artifact, and stromal mucin as seen in many nodular BCC. However, these features may be present focally and we have seen several trichoblastomas with adenoidal pattern and a mucinous stroma resembling BCC. The latter cases should really be subjected to immunohistochemistry. Mitotic activity in trichoblastoma can vary from absent (most conventional superficial trichoepitheliomas) to moderately high (3 or more per mm2)4 and may add to diagnostic confusion if one is unfamiliar with the diagnosis. We have noticed that mitotic activity is often present in “hot-spots” of proliferation on nuclear proliferation stains (Ki67), presumably recapitulating the hair germ. Other histologic features may lead to diagnostic confusion with malignancy including central comedo necrosis (occasionally present in tumors composed of large nodules) and rarely pleomorphic giant or so-called “monster” cells.11 Melanin pigmentation in trichoblastoma is not uncommon and reactive, usually dendritic, melanocytes may be shown with S100 and MelanA staining; this may also be a feature of some BCCs. The combination of BCC and trichoepithelioma is exceedingly rare, but has been reported particularly in the setting of multiple trichoepitheliomas.10,15,19,22 There are exceedingly rare reports of local recurrence zand even malignant transformation of longstanding benign trichoblastomas.17 In view of the diagnostic difficulty and rare instances of local recurrence and malignant transformation, we usually advise complete excision of any mitotically active trichoblastoma.7 As discussed, immunohistochemistry can be helpful in the histologic diagnosis of trichoblastoma but needs to be interpreted with some caution with an awareness of pitfalls.7 BerEP4 has been reported to be quite variable in trichoepithelioma/trichoblastoma21 and this is also our personal observation. In some cases tumors may be diffusely positive, similar to most BCC3,7 but a lack of diffuse strong BerEP4 staining in a basaloid tumor should prompt one to consider a diagnosis other than conventional BCC.7 CD10 staining, usually expressed in the perilesional stroma and papillary mesenchymal cells of trichoepithelioma but absent in the basaloid epithelium (the reverse pattern in most nodular basal cell carcinomas)16 is in our opinion quite characteristic and was helpful in case 2 (Figs. 3C, 4C) but more publications are required with this marker.7 Reactive Merkel cells are highly characteristic of benign hair follicle tumors and their presence within the deeper basaloid lobules is very useful to distinguish trichoepithelioma from basal cell carcinoma (Merkel cells absent) provided one is aware that occasional Merkel cells may be seen in entrapped follicles of a BCC.7,13,19 Unfortunately, Merkel cells may be absent in trichoepithelioma and trichoblastoma in up to 50% of cases so their absence is less helpful. The identification of Merkel cells with CK20 staining was most helpful in the needle core biopsy in our second case (Fig. 3C) and permitted a confident diagnosis. Our first case predated our detailed knowledge of the usefulness of immunohistochemistry for this differential diagnosis and there was insufficient material for additional studies. Cases such as ours need to be distinguished from primary breast carcinoma. The first case, having relatively well-separated circumscribed lobules, an adenoidal/cribriform pattern, and calcification, superficially resembles a ductal carcinoma in situ (cribriform type). The second case having more irregular “undifferentiated” basaloid masses with moderate mitotic activity was confused with a poorly differentiated breast carcinoma. To recap, the lack of cellular pleomorphism, lack of atypical mitotic figures, evenly cellular stroma, and prominent papillary mesenchymal cells are the most useful features in the differential diagnosis with a breast carcinoma. We are unaware of any description of Merkel cells in breast carcinoma. Very superficial breast lesions may be infrequently encountered in core needle biopsies. Furthermore as explained above, the diagnosis of benign hair follicle tumors, as with many cutaneous adnexal lesions, relies heavily on the architecture of the tumor interface with surrounding tissues. It can be difficult to fully appreciate the circumscribed architecture of such superficial lesions using core needle biopsy and these factors likely contributed to the difficulty in diagnosis in our cases. Conservative excision in the first case confirmed the favored diagnosis of trichoepithelioma. The second case was initially reported as a malignant breast tumor and then possibly a BCC. Immunohistochemistry proved helpful in supporting a confident diagnosis of trichoblastoma. There was tumor implantation in the second case needle core biopsy tract as evidenced by local fibrosis and recent hemorrhage (Figs. 4A, D). In the presence of a relatively undifferentiated mitotically active basaloid tumor, this was misconstrued as an area of invasion leading to a provisional diagnosis of trichoblastic carcinoma that was only reversed on further dermatopathologic review. Pathologists and clinicians need to consider the possibility of skin adnexal tumors being detected at screening mammography. These cases show the importance of awareness of the existence of non-breast tumors such as those of the skin when assessing either the histology or radiologic features of breast lesions. They highlight significant potential pitfalls in the diagnosis of trichoblastoma and helpful diagnostic clues. This is particularly relevant when dealing with basaloid skin tumors that can resemble tumors of the underlying breast or BCC. A conservative approach should be adopted in these circumstances and referral to an expert dermatopathologist is essential in informing decisions made at multidisciplinary team meetings concerning the optimal treatment offered to individual patients. Andrew C. Sherley-Dale, MRCGP, MRCP* Naresh Chachlani, MD, FEBP, FIAC† D. Scott A. Sanders, MD, FRCPath† Richard A. Carr, FRCPath, DipRCPath† Departments of *Dermatology †Pathology, Warwick Hospital Warwick, UK