Immunohistochemical Distinction of Primary Sweat Gland Carcinoma and Metastatic Breast Carcinoma

AJCP / Original Article Immunohistochemical Distinction of Primary Sweat Gland Carcinoma and Metastatic Breast Carcinoma Can It Always Be Accomplished Reliably? Mark J. Mentrikoski, MD, and Mark R. Wick, MD From the Division of Surgical Pathology and Cytopathology, University of Virginia Medical Center, Charlottesville. CME/SAM Key Words: Sweat gland carcinoma; Metastatic cutaneous breast carcinoma; Immunohistochemistry; Differential diagnosis Am J Clin Pathol March 2015;143:430-436 ABSTRACT Objectives: Even with adequate history, the distinction of cutaneous metastatic breast carcinoma from primary sweat gland carcinoma can be difficult. Although previous studies have attempted to separate these tumors with various immunohistochemical panels, those series have been limited by small numbers of patients as well as the inclusion of benign sweat gland tumors. Methods: In this analysis, stains for p63, CK5/6, and D2-40 were included, as well as GATA3 and mammaglobin, in an evaluation of 21 primary sweat gland carcinomas and 33 examples of cutaneous metastatic breast carcinoma. Results: Immunoreactivity for p63, CK5/6, D2-40, GATA3, and mammaglobin was respectively observed in 81%, 71%, 52%, 71%, and 5% of sweat gland carcinomas compared with 6%, 6%, 6%, 91%, and 45% of metastatic breast carcinomas. These differences were statistically significant for p63, CK5/6, and D2-40. For the diagnosis of metastatic breast carcinoma, GATA3 was the most sensitive marker (91%), but its sensitivity was substantially lower. Mammaglobin was 95% specific for breast carcinoma but again suffered from limited sensitivity (45%) in this context. Conclusions: These data suggest that p63 and CK5/6 are specific determinants for sweat gland carcinoma in the stated setting. In the absence of those analytes, metastatic breast carcinoma cannot always be identified to the exclusion of a primary tumor. This diagnostic scenario continues to require the procurement of a detailed clinical history regarding the number and duration of skin lesions in any given case. Upon completion of this activity you will be able to: list immunohistochemical stains that may be useful in distinguishing metastatic breast carcinoma from primary cutaneous adnexal carcinomas. discuss the importance of obtaining clinical information when evaluating cutaneous metastases. The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 Credit per article. Physicians should claim only the credit commensurate with the extent of their participation in the activity. This activity qualifies as an American Board of Pathology Maintenance of Certification Part II Self-Assessment Module. The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose. Questions appear on p 461. Exam is located at www.ascp.org/ajcpcme. Morphologic mimicry between two clinicopathologic entities can produce substantial diagnostic challenges. Indeed, that fact largely accounts for the important role that immunohistochemistry (IHC) currently plays in diagnostic anatomic pathology. One particular example of this phenomenon in dermatopathology is the microstructural similarity between metastatic ductal mammary carcinoma (MDMC) and primary cutaneous sweat gland carcinoma (SGC). Although metastasis of internal malignancies to the skin is relatively rare, it may occasionally be the first sign of systemic disease in patients with breast cancer. 1-3 Because both SGC and MDMC comprise cytologically atypical dermal nests of epithelioid cells, often containing duct-like structures, a distinction between those two tumor types can be virtually impossible on the basis of morphologic analysis alone. That fact is not surprising, because the breasts and the sweat glands are functional and structural homologues of 430 Am J Clin Pathol 2015;143:430-436

AJCP / Original Article one another. Both of them derive from ectodermal appendages, and although the mammary ducts arise from specialized mammary ridges, they are essentially comparable to modified sweat glands. Therefore, MDMC and SGC can bear a striking resemblance to one another at histologic and immunohistochemical levels of analysis. In the past, clinical features such as the location and number of the neoplasms, a history of malignancy, and lesional duration have been stressed as important diagnostic determinants. When such information is lacking, ancillary studies such as IHC may be pursued, in hopes that they may lead to an accurate pathologic classification of the tumor in question. Over the past 2 decades, several studies have examined with variable success selected individual immunostains, or panels of them, to make the specified distinction. 4-13 Although SGC often labels for so-called basal cell markers such as p63 and cytokeratin (CK) 5/6, that is not always true. Moreover, a suitably sensitive, specific, and exclusive marker for MDMC has not yet emerged. Shortcomings have existed in prior studies that compared SGC and MDMC. The first problem relates to the rarity of malignant sweat gland neoplasms and limited numbers of those lesions that have been available for evaluation. In response to that drawback, some authors also have included morphologically benign appendageal neoplasms of the skin for comparison with MDMC a step that unquestionably distorts the resulting data. 11,12 The current study was undertaken in an effort to extend the possible role for IHC in distinguishing between metastatic mammary and primary sudoriferous neoplasms in the skin. The authors included only morphologically malignant lesions that were bona fide microscopic imitators of one another. The immunohistologic markers that were used partially recapitulated the structure of prior publications, but they also included GATA3 gene product. The latter protein has been demonstrated predominantly in tumors of the urothelial tract, breast epithelium, and surface integument. Materials and Methods After obtaining approval from the institutional review board at the University of Virginia, its electronic surgical pathology database was searched between 1990 and 2012 to retrieve examples of MDMC in the skin and SGC. Benign sweat glandular neoplasms, such as syringomas, acrospiromas, spiradenomas, and poromas, were explicitly excluded. Fifty-four cases were ultimately identified, including 33 examples of MDMC and 21 of primary SGC. H&E-stained slides and the clinical records from each case were reviewed to confirm the diagnoses. Archival formalin-fixed, paraffinembedded tissue from the 54 cases was included in a tissue Table 1 Summary of Immunohistochemical Stains and Antigen Retrieval Techniques Antibody Manufacturer Clone Dilution microarray (TMA), which was constructed with four 0.6- mm cores of tumor per case using a manual microarrayer (Beecher Instruments, Sun Prairie, WI). Immunohistochemical analysis was performed of 4-µm thick TMA sections with the DAKO EnVision Plus Dual Link System horseradish peroxidase technique, using commercial antibodies and a DAKO Autostainer Plus (DAKO North America, Carpinteria, CA). Immunostains included GATA3, mammaglobin, p63, CK5/6, and D2-40 Table 1. Nuclear labeling of tumor cells was considered to represent a positive result for GATA3 and p63. Cytoplasmic and/or membranous staining was recorded as a positive result for CK5/6, D2-40, and mammaglobin. Appropriate positive and negative control sections were stained concurrently. Immunolabeling was scored semiquantitatively as follows: 0/negative (<5% of cells stained), 1+ (6%-25% of tumor cells), 2+ (26%-50% of tumor cells), and 3+ (>51% of tumor cells). Staining intensity was also graded as weak, moderate, or strong. Any case with moderate to strong intensity and at least 1+ staining was considered positive. Cases showing weak staining intensity were considered positive only if the staining was in a diffuse pattern (>51% of tumor cells staining). The sensitivity and specificity of each stain, vis-à-vis the diagnoses of primary cutaneous adnexal carcinoma and metastatic breast carcinoma, were calculated using Bayesian methods with a standard 2 2 contingency table. The statistical associations of each stain were analyzed by the Fisher exact test, in which a two-tailed P value of less than.05 was considered statistically significant. Results Demographic Data and Morphologic Composition Antigen Retrieval p63 Biocare (Concord, CA) 4A4 1:200 Citrate, ph 9.0 CK5/6 Biocare CK5/6.007 1:100 ph 9.0 GATA3 Biocare L50-823 1:250 Citrate, ph 6.0 D2-40 Covance D2-40 1:100 None (Princeton, NJ) Mammaglobin Zeta (Sierra Madre, CA) 304-1A5 1:50 None Primary SGCs The study cases were derived from 13 men and 8 women in the SGC cohort, with a mean age of 66.7 years Am J Clin Pathol 2015;143:430-436 431

Mentrikoski and Wick / Adnexal Carcinoma vs Metastatic Breast Carcinoma Table 2 Immunohistochemical Staining Results of p63, CK5/6, D2-40, Mammaglobin, and GATA3 in Primary Sweat Gland Carcinomas and Metastatic Breast Carcinoma Grade of Reactivity, No. (%) Primary Sweat Gland Carcinoma (n = 21) Metastatic Breast Carcinoma to Skin (n = 33) Marker Total Positive 0 1+ 2+ 3+ Total Positive 0 1+ 2+ 3+ p63 17 (81) 4 (19) 1 (5) 0 16 (76) 2 (6) 31 (94) 1 (3) 1 (3) 0 CK5/6 15 (71) 6 (29) 0 3 (14) 12 (57) 2 (6) 31 (94) 0 1 (3) 1 (3) D2-40 11 (52) 10 (48) 0 5 (24) 6 (29) 2 (6) 31 (94) 1 (3) 0 1 (3) Mammaglobin 1 (5) 20 (95) 1 (5) 0 0 15 (45) 18 (55) 0 1 (3) 14 (42) GATA3 15 (71) 6 (29) 1 (5) 8 (38) 6 (29) 30 (91) 3 (9) 0 3 (9) 27 (82) Table 3 Sensitivity and Specificity, With P Values, for Individual Immunohistochemical Stains Stain p63 CK5/6 D2-40 Mammaglobin GATA3 SGC (n = 21), No. (%) 17 (81) 15 (71) 11 (52) 1 (5) 15 (71) MBCS (n = 33), No. (%) 2 (6) 2 (3) 2 (6) 15 (45) 30 (91) P value <.0001 <.0001.0002.007.1306 Sensitivity, % 81 a 71 a 52 a 45 b 91 b Specificity, % 94 a 94 a 94 a 95 b 29 b a Sensitivity and specificity for the diagnosis of primary sweat gland carcinoma (SGC). b Sensitivity and specificity for the diagnosis of metastatic breast carcinoma to skin (MBCS). (range, 36-89 years). Of the 21 SGC cases, 19 were classified as eccrine adnexal carcinomas and two as primary cutaneous apocrine adenocarcinomas. Two of the 21 SGC lesions were recurrent tumors; the remaining 19 were primary neoplasms. Primary sites for the SGC cohort included five scalp, four axilla, three chest, two face, two leg, one ear, one abdomen, and one scrotum. When available, the clinical history revealed presence of a lesion for 3 weeks to 5 months (mean, 2.7 months). Metastatic Breast Carcinomas The MDMC group was composed entirely of women with a mean age of 64.3 years (range, 30-90 years). All examples were associated with a known history of primary breast carcinoma of the ductal type, not further specified prior to diagnosis of metastasis. Thirty cases arose on the skin of the chest ipsilateral to the primary lesion and three in the skin of the ipsilateral axilla. The duration of disease for the MDMC cohort ranged from 1 to 4 weeks from initial presentation of metastatic skin lesion to clinical biopsy/ excision. Immunohistochemical Results Primary Sweat Gland Carcinomas Immunohistochemical staining results in cases of SGC are summarized in Table 2. The sensitivity and specificity of each marker, as well as their P values for the distinction of primary SGC vs MDMC, are listed in Table 3. p63 was expressed most commonly in SGC, with strong, diffuse staining seen in most cases (17/21 [81%]). CK5/6 was also commonly seen in SGCs, whereas podoplanin (D2-40) was detected in a smaller number of SGC cases. All three of these markers were specific for SGC in the specified differential diagnosis (Table 3), but a moderate degree of background staining of dermal collagen was seen in D2-40 preparations Image 1. Among p63, CK5/6, and podoplanin, at least one marker was expressed in 19 (90%) of 21 SGCs, yielding a conjoint sensitivity of 90% and a specificity of 81%. Concurrent labeling for at least two of the three markers increased diagnostic specificity to 100%, but mandating that result for an interpretation of SGC decreased the composite sensitivity of the stains to 71%. Nuclear GATA3 immunoreactivity was apparent in 71% of cases of SGC, with a staining intensity ranging from weak to strong. More heterogeneous intratumoral labeling was observed with GATA3 compared with p63 and CK5/6. Both cases of primary cutaneous apocrine adenocarcinoma were strongly positive for GATA3, but they were negative for all other markers used in this study Image 2. Weak, diffuse cytoplasmic reactivity for mammaglobin was seen in only one example of SGC. Metastatic Breast Carcinomas MDMC rarely stained for p63, CK5/6, or podoplanin, with only two (6%) of 33 cases being labeled for each of those determinants (Image 1). No individual case expressed more than one of the three markers. GATA3 was the most sensitive immunostain for the identification of MDMCs, in which strong, diffuse nuclear positivity was seen in 91%. The specificity of GATA3 for the diagnosis of MDMC was 432 Am J Clin Pathol 2015;143:430-436

AJCP / Original Article compromised by its similar expression in 71% of primary SGCs, although labeling of those lesions was more patchy and tended to have at most moderate staining intensity. As an individual marker, mammaglobin was specific (95%) for MDMC, but it lacked sensitivity. Cytoplasmic staining was detected in only 15 (45%) of 33 cases. Both GATA3 and mammaglobin were positive in 42% of cases, with a respective sensitivity and specificity of 42% and 100% for the diagnosis of MDMC. Discussion Although they are relatively uncommon, cutaneous metastases may be seen in up to 10% of all patients with visceral carcinomas.3,14,15 In women, adenocarcinoma of the breast is the most common source for metastasis to the skin, typically presenting with nodular lesions on the trunk, axillae, or chest wall, usually near prior lumpectomy or mastectomy sites.16,17 Many observers still consider the clinical presentation to be of paramount importance in distinguishing MDMC from primary neoplasms of the skin.3,18 Metastatic carcinomas typically manifest with multiple, grouped, rapidly growing lesions, whereas solitary, slowly enlarging (>6 months) lesions are typical of SGC. When adequate clinical information is lacking, immunohistology has been applied to this particular differential diagnosis, with variable success. Over the past decade, several studies have attempted to delineate the optimal panel of markers for the separation of MDMC and SGC.4-13 In general, most of those studies have suffered from a lack of statistical power because of the rarity of primary SGC, with limited numbers of specimens. The current series admittedly has that same limitation. Another issue making the prior literature difficult to interpret is the common inclusion of embryologically related but benign tumors in several studies, potentially causing skewed results.11,12 For instance, although eccrine spiradenomas doubtlessly show sudoriferous differentiation, their morphologic images look nothing like those of metastatic breast carcinomas, and they should not be included in comparative analyses. In this evaluation, we used only primary SGCs that structurally imitated MDMCs. The basal cell markers p63 and CK5/6 were often present in SGCs in our assessment, with p63 being the most sensitive. p63 is a nuclear transcription factor that is involved in keratinocyte differentiation. It is normally expressed in basal epithelial cells, including those of the skin, but is downregulated as the cells mature. In addition to being a marker of squamous differentiation, p63 can be used in breast pathology to highlight myoepithelium surrounding ductal adenocarcinoma in situ; however, it is not Image 1 In a case of metastatic ductal mammary carcinoma (MDMC) (H&E, 10), the tumor cells are negative for p63 and CK5/6. D2-40 highlights stromal cells but is negative in tumor cells. Strong nuclear positivity is seen with GATA3, as is strong cytoplasmic positivity with mammaglobin. In contrast, in a case of primary sweat gland carcinoma (SGC) that arose on the scalp of a 78-year-old woman (H&E, 10), p63 and CK5/6 are strongly positive. Podoplanin is absent in the tumor cells but seen in background stromal cells. GATA3 immunostaining shows diffuse, moderate nuclear positivity. Mammaglobin is lacking. Am J Clin Pathol 2015;143:430-436 433

Mentrikoski and Wick / Adnexal Carcinoma vs Metastatic Breast Carcinoma A B C D Image 2 A case of primary cutaneous apocrine adenocarcinoma in the scalp of a 39-year-old man. Microscopic examination revealed infiltration of the dermis by nests and acini of cells with prominent eosinophilic cytoplasm (A, H&E, 10; B, H&E, 20). Like metastatic breast carcinoma, the tumor cells are strongly positive for GATA3 (C, 40) but negative for p63 (D, 40). usually expressed by invasive mammary carcinomas. Several prior analyses have shown that p63 reactivity generally distinguishes MDMC from SGC,5,6,8,9,11,12 and our study recapitulates that finding. One cautionary note regarding the use of p63 is that it has also been observed in metaplastic mammary carcinomas,19 but the high-grade, sarcomatoid nature of those lesions makes it unlikely that they would enter into differential diagnosis with SGC. CK5/6 is a mixture of cytokeratins that are also considered basal cell markers. Again, previous publications have shown CK5/6 reactivity in most SGCs but not in metastatic adenocarcinomas.10,11 We found that CK5/6 was slightly less sensitive for sweat gland tumors than was p63 (71% vs 81%), but both markers had robust specificity (94%). Although CK5/6 can be expressed in basal-like or 434 Am J Clin Pathol 2015;143:430-436 triple-negative breast carcinomas, none of the metastatic cases in our series had that phenotype. D2-40, the antibody that recognizes podoplanin, has classically been used to label lymphatic endothelium. Nonetheless, it has also been detected in a variety of unrelated tumors, including mesotheliomas, seminomas, and primary cutaneous adnexal carcinomas.7 Although Liang and colleagues7 showed variable podoplanin positivity in a spectrum of cutaneous neoplasms other than SGCs, it was nonetheless sensitive and specific for primary lesions of the skin as opposed to metastatic adenocarcinomas. One issue with that particular study was that cellular labeling of only 25% was defined as diffuse, potentially skewing the number of positive cases.7 It is possible that this effect augmented the reported sensitivity of podoplanin in the identification

AJCP / Original Article of SGCs, because our series as well as others showed lesser levels of expression in SGC. 8,9 Nevertheless, our results support the premise that podoplanin is a specific marker for SGC and is rarely expressed in MDMC. Unfortunately, because of its limited sensitivity, podoplanin can be considered a thirdline marker for primary skin tumors, behind CK5/6 and p63. As mentioned earlier, a biomarker that is both sensitive and specific for MDMC has yet to be clearly defined. Estrogen and progesterone receptor proteins are often used in surgical pathology as surrogate markers for mammary epithelial differentiation, because most breast carcinomas express those moieties. However, both receptors also may potentially be seen in SGCs, 20-22 a finding that decreases their usefulness in the specific differential diagnostic setting being considered here. Gross cystic disease fluid protein 15 (GCDFP-15) is another marker of breast carcinoma that has been studied in MDMC. Its comparative lack of sensitivity and specificity, compared with the immunophenotypes of SGCs, has limited the value of GCDFP-15 in this setting. 12,13,21,23 Mammaglobin is another mammary epithelial determinant that is more sensitive but less specific than GCDFP-15. 24 To date, only a single study has evaluated mammaglobin in the differential diagnosis of MDMC and primary SGC. 12 In that series, mammaglobin was observed in two-thirds of MDMCs, whereas it was lacking in all SGCs. Our results showed that mammaglobin had a lower level of sensitivity for MDMC, in line with another publication on primary and metastatic breast carcinomas. 25 GATA3 is a transcription factor that is involved in embryologic development of the mammary glands and urinary bladder, as well as neoplasms in those organs. To our knowledge, GATA3 has not been evaluated in MDMC and SGC. Although it was a sensitive marker for MDMC, it lacked specificity and was also seen in 71% of SGC cases in our series. In context, that is not surprising because other recent publications have shown that GATA3 also participates in the development of the skin. 26,27 Cutaneous apocrine adenocarcinomas are rare sweat gland neoplasms; they most commonly arise in the axilla and anogenital skin. 28 The literature pertaining to these cutaneous tumors is sparse because of their limited frequency, and the current series likewise included only two cases seen over a 20-year period. Apocrine sweat gland carcinomas are essentially indistinguishable from mammary apocrine carcinomas, and those seen in the axillary region are particularly challenging to separate from MDMCs. Interestingly, both tumors in this series occurred in men, with one on the scrotum and the other on the scalp. In summary, primary SGCs are rare neoplasms that can be morphologically comparable to cutaneous metastases from breast carcinomas. Both p63 and CK5/6 were sensitive and specific for the diagnosis of SGC, but a similarly effective breast carcinoma marker has yet to be identified. The clinical setting must continue to play a central role in the appropriate classification of MDMC and SGC. Address reprint requests to Dr Wick: University of Virginia Hospital, 1215 Lee St, Rm 3020, Charlottesville, VA 22908-0214; mrwick1@usa.net. References 1. Alcaraz I, Cerroni L, Rutten A, et al. Cutaneous metastases from internal malignancies: a clinicopathologic and immunohistochemical review. Am J Dermatopathol. 2012;34:347-393. 2. Krathen RA, Orengo IF, Rosen T. Cutaneous metastasis: a meta-analysis of data. South Med J. 2003;96:164-167. 3. Lookingbill DP, Spangler N, Sexton FM. Skin involvement as the presenting sign of internal carcinoma: a retrospective study of 7316 cancer patients. J Am Acad Dermatol. 1990;22:19-26. 4. Fernandez-Flores A. Podoplanin immunostaining in cutaneous apocrine carcinoma and in cutaneous metastasis from the breast. Appl Immunohistochem Mol Morphol. 2010;18:573-574. 5. Ivan D, Hafeez Diwan A, Prieto VG. Expression of p63 in primary cutaneous adnexal neoplasms and adenocarcinoma metastatic to the skin. Mod Pathol. 2005;18:137-142. 6. Ivan D, Nash JW, Prieto VG, et al. Use of p63 expression in distinguishing primary and metastatic cutaneous adnexal neoplasms from metastatic adenocarcinoma to skin. J Cutan Pathol. 2007;34:474-480. 7. Liang H, Wu H, Giorgadze TA, et al. Podoplanin is a highly sensitive and specific marker to distinguish primary skin adnexal carcinomas from adenocarcinomas metastatic to skin. Am J Surg Pathol. 2007;31:304-310. 8. Mahalingam M, Nguyen LP, Richards JE, et al. The diagnostic utility of immunohistochemistry in distinguishing primary skin adnexal carcinomas from metastatic adenocarcinoma to skin: an immunohistochemical reappraisal using cytokeratin 15, nestin, p63, D2-40, and calretinin. Mod Pathol. 2012;23:713-719. 9. Plaza JA, Ortega PF, Stockman DL, et al. Value of p63 and podoplanin (D2-40) immunoreactivity in the distinction between primary cutaneous tumors and adenocarcinomas metastatic to the skin: a clinicopathologic and immunohistochemical study of 79 cases. J Cutan Pathol. 2010;37:403-410. 10. Plumb SJ, Argenyi ZB, Stone MS, et al. Cytokeratin 5/6 immunostaining in cutaneous adnexal neoplasms and metastatic adenocarcinoma. Am J Dermatopathol. 2004;26:447-451. 11. Qureshi HS, Ormsby AH, Lee MW, et al. The diagnostic utility of p63, CK5/6, CK 7, and CK 20 in distinguishing primary cutaneous adnexal neoplasms from metastatic carcinomas. J Cutan Pathol. 2004;31:145-152. 12. Rollins-Raval M, Chivukula M, Tseng GC, et al. An immunohistochemical panel to differentiate metastatic breast carcinoma to skin from primary sweat gland carcinomas with a review of the literature. Arch Pathol Lab Med. 2012;135:975-983. Am J Clin Pathol 2015;143:430-436 435

Mentrikoski and Wick / Adnexal Carcinoma vs Metastatic Breast Carcinoma 13. Wick MR, Ockner DM, Mills SE, et al. Homologous carcinomas of the breasts, skin, and salivary glands: a histologic and immunohistochemical comparison of ductal mammary carcinoma, ductal sweat gland carcinoma, and salivary duct carcinoma. Am J Clin Pathol. 1998;109:75-84. 14. Lookingbill DP, Spangler N, Helm KF. Cutaneous metastases in patients with metastatic carcinoma: a retrospective study of 4020 patients. J Am Acad Dermatol. 1993;29:228-236. 15. Schwartz RA. Cutaneous metastatic disease. J Am Acad Dermatol. 1995;33:161-186. 16. De Giorgi V, Grazzini M, Alfaioli B, et al. Cutaneous manifestations of breast carcinoma. Dermatol Ther. 2010;23:581-589. 17. Pipkin CA, Lio PA. Cutaneous manifestations of internal malignancies: an overview. Dermatol Clin. 2008;26:1-15, vii. 18. Brownstein MH, Helwig EB. Patterns of cutaneous metastasis. Arch Dermatol. 1972;105:862-868. 19. Koker MM, Kleer CG. p63 expression in breast cancer: a highly sensitive and specific marker of metaplastic carcinoma. Am J Surg Pathol. 2004;28:1506-1512. 20. Swanson PE, Mazoujian G, Mills SE, et al. Immunoreactivity for estrogen receptor protein in sweat gland tumors. Am J Surg Pathol. 1991;15:835-841. 21. Wallace ML, Longacre TA, Smoller BR. Estrogen and progesterone receptors and anti gross cystic disease fluid protein 15 (BRST-2) fail to distinguish metastatic breast carcinoma from eccrine neoplasms. Mod Pathol. 1995;8:897-901. 22. Kariya Y, Moriya T, Suzuki T, et al. Sex steroid hormone receptors in human skin appendage and its neoplasms. Endocr J. 2005;52:317-325. 23. Wallace ML, Smoller BR. Differential sensitivity of estrogen/ progesterone receptors and BRST-2 markers in metastatic ductal and lobular breast carcinoma to the skin. Am J Dermatopathol. 1996;18:241-247. 24. Bhargava R, Beriwal S, Dabbs DJ. Mammaglobin vs GCDFP- 15: an immunohistologic validation survey for sensitivity and specificity. Am J Clin Pathol. 2007;127:103-113. 25. Luo MH, Huang YH, Ni YB, et al. Expression of mammaglobin and gross cystic disease fluid protein-15 in breast carcinomas. Hum Pathol. 2013;44:1241-1250. 26. Kaufman CK, Zhou P, Pasolli HA, et al. GATA-3: an unexpected regulator of cell lineage determination in skin. Genes Dev. 2003;17:2108-2122. 27. Sellheyer K, Krahl D. Expression pattern of GATA-3 in embryonic and fetal human skin suggests a role in epidermal and follicular morphogenesis. J Cutan Pathol. 2010;37:357-361. 28. Robson A, Lazar AJ, Ben Nagi J, et al. Primary cutaneous apocrine carcinoma: a clinico-pathologic analysis of 24 cases. Am J Surg Pathol. 2008;32:682-690. 436 Am J Clin Pathol 2015;143:430-436