NOTE Pathology

Cutaneous Clear Cell Adnexal Carcinoma in a Dog: Special Reference to Cytokeratin Expression

Kyohei YASUNO1), Shoko NISHIYAMA1), Fumio SUETSUGU2), Kikumi OGIHARA3), Hiroo MADARAME4) and Kinji SHIROTA1,5)*

1)Research Institute of Biosciences, 3)Laboratory of Environmental Pathology, 4)Veterinary Teaching Hospital and 5)Laboratory of Veterinary Pathology, Azabu University, 1–17–71 Fuchinobe, Sagamihara, Kanagawa 229–8501 and 2)Maria Pet Clinic, 103–3 Kitayacho, Nakahara, Kawasaki, Kanagawa 211–0015, Japan

(Received 25 May 2009/Accepted 21 June 2009)

ABSTRACT. A 5-year-old, male Bichon-Frise dog presented with a cutaneous mass in the basal region of the auricle. Histologically, thecutaneous neoplasm was comprised of lobules with solid cellular proliferation separated by thin fibrous septa. Neoplastic cells variedin size, with moderate to abundant amounts of PAS-positive cytoplasm, large nuclei and prominent nucleoli. Immunohistochemicalexaminations showed that tumor cells were positive for pan-cytokeratin (CK) (AE1/AE3 and CAM5.2), CK8 and CK18, but negativefor pan-CK (KL1), CK7, CK14, CK16 and CK20. Double-labeled immunofluorescence testing indicated that neoplastic cells frequentlyco-expressed CK and vimentin, suggesting divergent differentiation of tumor cells. Based on these findings, the tumor was diagnosedas canine clear cell adnexal carcinoma.KEY WORDS: canine, cytokeratin, immunohistochemistry, neoplasm, skin.

J. Vet. Med. Sci. 71(11): 1513–1517, 2009

Malignant cutaneous tumors in dog with clear or vacu-olated cytoplasm are uncommon, and include balloon cellmelanoma, clear cell basal carcinoma, sebaceous carcinomaand clear cell adnexal carcinoma [3, 4, 12]. Canine cutane-ous clear cell adnexal carcinoma was first proposed bySchulman et al. [12], and seems to be the same neoplasmknown as clear cell hidradenocarcinoma or follicular stemcell carcinoma [12]. Clear cell adnexal carcinoma is a raremalignant neoplasm characterized by divergent differentia-tion into cutaneous adnexa, such as apocrine sweat glandsand hair follicles. Clear cell adnexal carcinoma has beenreported in dogs with a mean age of 6.9 years, with noapparent predilections for breed, sex or site [4, 9, 12]. Thisneoplasm shows a moderate prognosis following surgicalexcision, but infrequent recurrence and metastasis to locallymph nodes has been reported [12]. The origin of thistumor is likely to be follicular stem cells, as some tumorcells immunohistochemically show double-positivity forcytokeratin (CK) and vimentin [9, 12].

CKs are classified according to molecular weight and iso-electric points, and are numbered from 1 to 20 [1]. Exami-nation of CK isoforms in clear cell adnexal carcinoma maybe useful to identify divergent differentiation and tumor ori-gin, as expression of CK isoforms differs among the epithe-lial cells of various cutaneous adnexa. We present herein acase of canine clear cell adnexal carcinoma and describe theimmunohistological properties, particularly in terms of CKisoform expressions.

A 5-year-old, male Bichon-Frise dog was brought in witha red papule approximately 8 mm in diameter (Fig. 1). The

mass was located on the skin at the basal region of the leftauricle, and no swelling or inflammatory signs were appar-ent in surrounding tissues. Six months later, cytoscreeningwas performed and squamous cell carcinoma or malignanttrichoepithelioma was suspected by a diagnostic laboratory.The tumor mass was then surgically excised and submittedto our laboratory for histopathological examination. On cutsection, the mass was grey-white and well-defined, withpartial hemorrhage. During the 11 months since the surgicalexcision, no local recurrence or metastasis has been recog-nized.

Excised tissues were fixed in 10% neutral-buffered for-malin, embedded in paraffin, sectioned and stained usinghematoxylin and eosin (HE), Periodic acid-Schiff (PAS),Sudan III and Fontana-Masson silver impregnation. Immu-nohistochemical staining was performed by immunoenzymepolymer method using the primary antibodies shown inTable 1. Peroxidase-conjugated anti-mouse (Histofine Sim-ple Stain MAX-PO(M); Nichirei, Tokyo, Japan) or peroxi-dase-conjugated anti-rabbit (Histofine Simple Stain MAX-PO(R); Nichirei) immunoglobulin (Ig)G were used as sec-ondary antibodies. These immunohistochemical stains wereincubated with diaminobenzidine and counterstained withMayer’s hematoxylin. Nine samples of canine trichoblas-toma were used for immunohistochemical comparison andintact skin samples taken from several dogs were used asnormal control tissues. Neoplastic cells in trichoblastomaswere divided into palisade/basaloid cells and spindle/elon-gated cells by their histological features. To determine co-expression of CK and vimentin in neoplastic cells, double-labeled immunofluorescence testing was performed usinganti-CK (clone AE1/AE3) and anti-vimentin antibodies asprimary antibodies. In this assay, affinity-purified goat anti-mouse IgG fluorescein isothiocyanate (FITC) (EY Labora-

* CORRESPONDENCE TO: SHIROTA, K., Laboratory of VeterinaryPathology, School of Veterinary Medicine, Azabu University, 1–17–71 Fuchinobe , Sagamihara, Kanagawa 229–8501, Japan.

e-mail: shirota@azabu-u.ac.jp

K. YASUNO ET AL.1514

Fig. 1. The neoplastic mass in the skin of the basal region of the left auricle.Fig. 2. The cutaneous neoplasm is well-demarcated and densely cellular. The mass is comprised of variably sized lobules and a large cystic

area. Hematoxylin and Eosin (HE) stain. Bar=1,000 m.Fig. 3. a) The neoplastic lobule with a follicular papilla-like structure in the periphery of the lobule (arrow). HE. Bar=50 m. b) Neoplastic

cells vary in size, showing a moderate to abundant amount of eosinophilic cytoplasm with large nuclei and nucleoli. Spindle cells andmulinucleated cells are also present (arrow). HE. Bar=25 m. c) The majority of tumor cells includes Periodic acid-Schiff (PAS)-positivefine granules in the cytoplasm. PAS. Bar=15 m.

1515CANINE CUTANEOUS CLEAR CELL ADNEXAL CARCINOMA

tories, San Mateo, California, U.S.A.) and affinity-purifiedgoat anti-mouse IgG Rhodamine (Chemicon, Temecula,California, U.S.A.) were used to label anti-CK and anti-vimentin antibody, respectively. In addition, a part of theformalin-fixed tissue specimen was cut into 1-mm3 cubes,re-fixed in 2.5% glutaraldehyde, post-fixed in 1% OsO4 andembedded in epoxy resin. Ultrathin sections were double-stained with uranyl acetate and lead citrate, then examinedusing a JEOL 1210 transmission electron microscope(JEOL, Tokyo, Japan) at 80 kV.

Histologically, the cutaneous neoplasm was well-demar-cated and densely cellular. The mass was comprised of vari-ably sized lobules separated by thin fibrous bands, and alarge cystic structure was also present (Fig. 2). The cystwall was composed of layers of neoplastic cells and the cystwas filled with blood and defoliated tumor cells. Neoplasticcells varied in size, and ranged in shape from round topolygonal to spindle-shaped. Moderate to abundantamounts of pale eosinophilic cytoplasm were present alongwith a large nucleus and prominent nucleolus (Fig. 3a).Multinucleated tumor cells were occasionally observed,with some including more than ten nuclei (Fig. 3b). Mosttumor cells had vacuolated or clear cytoplasm, which wasnegative for Sudan III and positive for PAS with a fine gran-ular appearance (Fig. 3c). Most PAS-positive granules werediastase-labile, indicating glycogen. Fontana-Masson stain-ing did not identify melanin in tumor cells and no neuroen-

docrine granules were identified in the cytoplasmultrastructually. Occasionally, clusters of small, dark, roundto spindle-shaped nuclei suggestive of the papilla of a hairfollicle were observed adjacent to neoplastic lobules (Fig.3a). Mitotic rate was approximately 4 per 10 high-poweredfield.

The results of immunohistochemical examinations aresummarized in Table 2. Neoplastic cells were constantlypositive for pan-CK (AE1/AE3 and CAM 5.2), and also CK8 and CK18, but negative for other CK markers (Fig. 4).Neoplastic cells positive for CAM 5.2 were often spindle-shaped (Fig. 4c), although round or polygonal-shaped cellswith vacuoles were negative. A monoclonal antibody,CK19 (Ks19.1), did not react with any cutaneous cells fromnormal dogs or the present case. Double-labeled immuno-fluorescence testing revealed CK-vimentin double-positivecells among the neoplastic cells, particularly within thespindle-shaped cell population (Fig. 5). Follicular papilla-like structures (follicular papillary mesenchymal bodies)were negative for CK, but positive for vimentin. Several S-100a-positive tumor cells were present in peripheral areas ofneoplastic lobules (Fig. 4k), but no neuron specific enolase(NSE)-positive cells were seen within the tumor mass.MHC class II-positive cells were also present within theneoplastic lobules, and formed cell processes, indicatingdendritic cells (Fig. 4l). Results of immunohistochemicalexaminations of trichoblastomas and normal tissues are also

Fig. 4. Immunostaining of tumor tissues. Immunostaining for cytokeratin (CK). a) CK clone AE1/AE3. b) CK clone KL1. c) CK cloneCAM5.2. The majority of positive cells is spindle-shaped. d) CK 7. e) CK 8. f) CK 14. g) CK 16. h) CK 18. i) CK 20. Most neoplastic cellsare positive for CK clone AE1/AE3, CAM5.2, CK8 and CK18. Many neoplastic cells are also positive for vimentin. j) Vimentin. k) S-100a. Several positive cells are observed peripherally in neoplastic lobules (arrow). l) MHC class II. Dendritic cells forming cell processesare observed within the tumor. Bar=25 m.

Fig. 5. Double-labeled immunofluorescence microscopy of tumor cells. Green color indicates positive staining for CK clone AE1AE3. Redcolor shows positive staining for vimentin. Nuclei are colored blue with 4,6-diamino-2-phenylindole. a) CK clone AE1/AE3. FITC. b)Vimentin. Rhodamin. c) Merge. Yellow or orange colored cells show co-expression of CK and vimentin.

Table 1. Immunohistochemistry

Antibody Clone Dilution Antigen retrievala) Antibody sourceb)

Cytokeratin AE1/AE3 1:50 trypsin DakoCytokeratin KL1 1:100 MW ImmunotechCytokeratin CAM 5.2 prediluted proteinase K Becton DickinsonCytokeratin 7 OC-TL 12/30 prediluted proteinase K DakoCytokeratin 8 Ks 8.7 prediluted MW Progen BiotechnikCytokeratin 14 LL002 1:100 MW SerotecCytokeratin 16 LL025 1:50 MW ThermoCytokeratin 18 Ks 18.4 prediluted proteinase K Progen BiotechnikCytokeratin 19 Ks 19.1 prediluted MW Progen BiotechnikCytokeratin 20 Ks 20.8 prediluted proteinase K DakoVimentin V9 1:25 MW Dako-SMA 1A4 1:50 NT DakoS-100a polyclonal 1:200 MW DakoNSE NSE-1G4 1:100 MW ZymedMelan A A103 1:50 MW DakoMHC class II TAL.1B5 1:100 MW Dako

a) MW=microwave/citrate buffer (pH 6.0); NT = no treatment; b) Dako, Copenhergen, Denmark;Immunotech, Marseille, France; Becton Dickinson, Heidelberg, Germany; Progen Biotechnik,Heidelberg, Germany; Serotec, Wiesbaden, Germany; Thermo, California, U.S.A.; Zymed, California,U.S.A.

K. YASUNO ET AL.1516

described in Tables 2 and 3. Finally, the cutaneous tumorwas diagnosed as clear cell adnexal carcinoma on the basisof morphological and immunohistochemical characteriza-tion.

Differential diagnoses for canine clear cell adnexal carci-noma include balloon cell melanoma, clear cell basal carci-noma and sebaceous carcinoma, due to var iousmorphological similarities. On immunohistochemicalexamination, CK-positive results rule out melanocytic neo-plasms. Basal cell carcinoma is reportedly positive for KL1and CK14 [5], which were negative in our case. Sebaceouscarcinoma was also excluded based on histological andimmunohistochemical properties.

In the present case, cellular differentiation into a hair fol-licle was suggested from the presence of follicular papilla-like structures and expression of CK8. CK8 is expressed inthe outer root sheath cells, but also expressed in the normalsweat gland [6, 8, 10] (Table 3). The outer root sheath cellincludes PAS-positive granules in the cytoplasm, which wasalso seen in our case. These results suggest differentiationof neoplastic cells into outer root sheath cells, but we did notdetect any differentiation to inner root sheath cells. Innerroot sheath cells are positive for CK16 and KL1, but thesewere negative in our case (Table 2). Formation of subtletubular structures suggesting differentiation into a sweatgland has been reported in other cases of clear cell adnexal

Table 2. Results of immunostains of tumor and tricoblastoma tissues

Tumor Trichoblastoma (n=9)

Palisade/basaloid Spindle/elongatedAntibody Clone cells cells

Cytokeratin AE1/AE3 + (> 75%)a) + (> 10%) + (> 10%)Cytokeratin KL1 – + (0–10%) + (0–10%)Cytokeratin CAM 5.2 + (> 50%) – –Cytokeratin 7 OC-TL 12/30 – – –Cytokeratin 8 Ks 8.7 + (> 75%) + (> 10%) + (> 50%)Cytokeratin 14 LL002 – + (> 10%) + (> 10%)Cytokeratin 16 LL025 – – + (0–10%)Cytokeratin 18 Ks 18.4 + (> 75%) – –Cytokeratin 20 Ks 20.8 – – –Vimentin V9 + (> 50%) – –-SMA 1A4 – – –S-100a polyclonal + (1–10%) – –NSE NSE-1G4 – – –Melan A A103 – – –MHC class II TAL.1B5 (< 1%) (< 1%)

– = negative; = few positive cells within tumor tissues; + = positive. a) The percentage indicatethe estimated proportion of the positive cells.

Table 3. Results of immunostains of normal tissuesa)

Normal tissues

Inner root Outer root Sebaceous Apocrine FollicularAntibody Clone Epidermis sheath sheath gland gland papilla

Cytokeratin AE1/AE3 + + + – + –Cytokeratin KL1 +b) + – – – –Cytokeratin CAM 5.2 – + – – + –Cytokeratin 7 OC-TL 12/30 – – – – + –Cytokeratin 8 Ks 8.7 – – + – + –Cytokeratin 14 LL002 +c) + + + +g) –Cytokeratin 16 LL025 – + – – – –Cytokeratin 18 Ks 18.4 – – – – + –Cytokeratin 20 Ks 20.8 –d) – – – – –Vimentin V9 –e) – – – – +-SMA 1A4 – – – – +g) –S-100a polyclonal – – – – – –NSE NSE-1G4 – – – – – –Melan A A103 –f) – – – – –MHC class II TAL.1B5 –e) – – – – –

a) – = negative; + = positive, b) basal cell layer was negative, c) granular and horny cell layers were negative, d)Merkel cells were positive, e) Langerhans cells were positive, f) melanocytes were positive, g) only myoepithelialcells showed positive reaction.

1517CANINE CUTANEOUS CLEAR CELL ADNEXAL CARCINOMA

carcinoma [9, 12], but was not found in our case. The largecystic structure observed in the present case may suggestdifferentiation into a tubular structure. Moreover, expres-sion of CK18 suggested differentiation into sweat gland, asCK18 expression is specifically observed in the normalsweat gland [6, 8, 10] (Table 3). In contrast, CK7 expres-sion is also reportedly specific to normal sweat glands [2],but our case was negative for CK7 (Table 2). Combiningthese findings, these features indicate that neoplastic cells inour case showed inconclusive differentiation to sweat glandcells, because of the positivity for CK18. Differentiation tomelanocytes has also been suggested in the majority of clearcell adnexal carcinomas on the grounds of positivity withFontana-Masson staining and melan-A [9, 12], both ofwhich were negative in our case. Tumor differentiation toneuroendocrine cells has also been suggested on the basis ofultrastructual findings in several cases [9], but we did notdetect neuroendocrine granules. A lack of CK20 expressionexcluded differentiation of the neoplasm to Merkel cells.

The origin of clear cell adnexal carcinoma is consideredto involve follicular stem cells, with tumor cells later show-ing divergent differentiation into cutaneous adnexa [9, 12].In mice, follicular stem cells can form hair follicles, sweatgland, sebaceous gland and epidermis [11]. As in this case,divergent differentiation to hair follicle and sweat gland,and co-expression of CK and vimentin strongly suggest thatthe tumor was likely to have originated as a follicular stemcell. Recently, follicular stem cell markers such as CK15,CK19, CD34 and CD200 have been reported in humans [7],but no precise markers have been described for canine folli-cular stem cells. Further research is required to clarify theorigins of canine clear cell adnexal carcinoma.

Trichoblastoma must be considered as a differential diag-nosis, given the lower proportion of the clear cell populationin the present case as compared with previous cases [9, 12],and the absence of tubular structures. Some types of tricho-blastoma, including granular cell or clear cell trichoblas-toma, show presence of a PAS-positive clear cell populationas a clear cell adnexal carcinoma [3, 4, 13]. Clear cell tri-choblastoma is a rare variant for which only a single case inhumans and 2 cases in dogs have been reported [13, 14],which histologically resembles our case. Spindle cells andfollicular papilla-like structures could also be present in tri-choblastoma [3, 4]. To differentiate trichoblastoma fromclear cell adnexal carcinoma, immunohistochemical exami-nation for CK8 and CK18 could be useful. Trichoblastomais considered to originate from hair germ cells [4], so tumorcells are positive for CK8, but negative for CK18, which isonly expressed in sweat glands (Tables 2, 3). This immuno-histological property of trichoblastoma was independent ofhistopathological subtype as ribbon, trabecular and spindlecell type. Conversely, clear cell adnexal carcinoma showspositive results for both CK8 and CK18 according to diver-gent differentiation to cutaneous adnexa. Furthermore, novimentin-positive tumor cells are present in trichoblastoma,and this could represent the crucial difference between thesetumors.

This paper is the first report of detailed immunohis-tochemical examination for cytokeratin expression in a caseof canine clear cell adnexal carcinoma.

ACKNOWLEDGMENT. We wish to thank Dr. IsaoNarama from the Department of Pathology, Faculty of Phar-maceutical Sciences at Setsunan University, for criticalcomments regarding the histopathological findings.

REFERENCES

1. Cooper, D., Schermer, A. and Sun, T. T. 1985. Classificationof human epithelia and their neoplasms using monoclonal anti-bodies to keratins: strategies, applications, and limitations.Lab. Invest. 52: 243–256.

2. Espinosa De Los Monteros, A., Fernandez, A., Millan, M. Y.,Rodriguez, F., Herraez, P. and Martin De Las Mulas, J. 1999.Coordinate expression of cytokeratins 7 and 20 in feline andcanine carcinomas. Vet. Pathol. 36: 179–190.

3. Goldschmidt, M. H., Dunstan, R. W., Stannard, A. A., vonTscharner, C., Walder, E. J. and Yager, J. A. 1998. HistologicalClassification of Epithelial and Melanocytic Tumors of theSkin of Domestic Animals, 2nd ser., vol. 3. Armed ForcesInstitute of Pathology, Washington, DC.

4. Gross, T. L., Ihrke, P. J., Walder, E. J. and Affolter, V. K.2005. Skin Diseases of the Dog and Cat. 2nd ed., BlackwellPublishing, Oxford.

5. Hoinghaus, R., Hewicker-Trautwein, M. and Mischke, R.2007. Immunocytochemical differentiation of neoplastic andhyperplastic canine epithelial lesions in cytologic imprint prep-arations. Vet. J. 173: 79–90.

6. Kato, K., Uchida, K., Nibe, K. and Tateyama, S. 2007. Immu-nohistochemical studies on cytokeratin 8 and 18 expressions incanine cutaneous adnexus and their tumors. J. Vet. Med. Sci.69: 233–239.

7. Kloepper, J. E., Tiede, S., Brinckmann, J., Reinhardt, D. P.,Meyer, W., Faessler, R. and Paus, R. 2008. Immunophenotyp-ing of the human bulge region: the quest to define useful in situmarkers for human epithelial hair follicle stem cells and theirniche. Exp. Dermatol. 17: 592–609.

8. Kozaki, M., Nakamura, Y., Iguchi, M., Kano, R., Watanabe,S., Fujiwara, K. and Hasegawa, A. 2001. Immunohistochemi-cal analysis of cytokeratin expression in dog skin. J. Vet. Med.Sci. 63: 1–4.

9. Mikaelian, I. and Wong, V. 2003. Follicular stem cell carci-noma: histologic, immunohistochemical, ultrastructural, andclinical characterization in 30 dogs. Vet. Pathol. 40: 433–444.

10. Nibe, K., Uchida, K., Itoh, T. and Tateyama, S. 2005. A case ofcanine apocrine sweat gland adenoma, clear cell variant. Vet.Pathol. 42: 215–218.

11. Oshima, H., Rochat, A., Kedzia, C., Kobayashi, K. and Barran-don, Y. 2001. Morphogenesis and renewal of hair folliclesfrom adult multipotent stem cells. Cell 104: 233–245.

12. Schulman, F. Y., Lipscomb, T. P. and Atkin, T. J. 2005. Caninecutaneous clear cell adnexal carcinoma: histopathology, immu-nohistochemistry, and biologic behavior of 26 cases. J. Vet.Diagn. Invest. 17: 403–411.

13. Sharif, M. and Reinacher, M. 2006. Clear cell trichoblastomasin two dogs. J. Vet. Med. A Physiol. Pathol. Clin. Med. 53:352–354.

14. Tronnier, M. 2001. Clear cell trichoblastoma in associationwith a nevus sebaceus. Am. J. Dermatopathol. 23: 143–145.