DNA measurement and immunohistochemical characterization of epithelial and mesenchymal cells in canine mixed mammary tumours: putative evidence for a common histogenesis.

DNA measurement by image cytometry, and a detailed immunohistochemical study using monoclonal antibodies directed against different human cytokeratin types, muscle-specific actin, vimentin and S100 protein were carried out on normal canine mammary tissue (n =4), benign canine mammary mixed tumours (n =20) and malignant canine mammary mixed tumours (n =13). The results showed that ductal and alveolar luminal cells in normal and neoplastic tissue were immunoreactive with CAM5.2 and AE1/AE3 antibodies recognizing human keratins.Basal/myoepithelial cells were clearly differentiated from ductal and alveolar epithelial cells, since the latter only expressed cytokeratins, whereas the former also expressed vimentin and muscle-specific actin. This immunohistochemical study showed that there is loss of expression of muscle-specific actin and cytokeratins in areas of myoepithelial proliferation, and enhanced expression of vimentin and S100 protein in proliferative areas with osseous and/or chondroid metaplasia. The ploidy studies revealed that 20% (4/20) of benign and 54% (7/13) of malignant mixed tumours of canine mammary gland were aneuploid and that the epithelial and myoepithelial components of the mixed tumours had identical DNA content.Our results reinforce the role of myoepithelial cells in mesenchymal metaplasia in mixed mammary tumours and suggest the possibility of a common origin of both components from a totipotential stem cell with capacity for divergent differentiation.     

Studies on histogenesis of mixed tumours of the mammary gland in bitches

The aim of the study was the immunohistochemical assessment of the reactivity of mixed tumours in bitches with special regard to myoepithelial cells and cartilage. The materials for study were 20 mixed tumours collected during surgery or autopsy. Paraffin-embedded sections were routinely stained with hematoxylin and eosin and HID method was also used. Immunohistochemical assays were made using the AB-Complex method with the use of monoclonal antibodies against: cytokeratin MNF-116, cytokeratin 19, vimentin, actin and S-100 protein. On the basis of the reactions, 5 types of myoepithelial origin cells transformed during the formation of cartilage cells were diagnosed.     

Immunolocalization of the smooth muscle-specific protein calponin in complex and mixed tumors of the mammary gland of the dog: assessment of the morphogenetic role of the myoepithelium

The immunohistochemical expression of the smooth muscle-specific protein calponin was studied to assess the contribution of myoepithelial cells to the histogenesis of spindle cells of complex and mixed tumors of the mammary gland of the dog and the origin of cartilage and bone in mixed tumors. Formalin-fixed tissues from 55 benign and malignant tumors (49 also containing surrounding normal mammary gland) were evaluated. Periacinar and periductal myoepithelial cells of all the 49 normal mammary glands were diffusely stained by the anti-human calponin monoclonal antibody. Calponin was found in 53 (98%) of the tumors studied, reacting with the myoepithelium-like cells of 86% of benign tumors and their remnants in 85% of malignant tumors. Five different types of calponin-immunoreactive myoepithelial cells were identified: hypertrophic myoepithelial cells. fusiform cells, stellate myoepithelial cells, rounded (myoepithelial) cells, and chondroblasts. Differences in staining intensity and staining pattern among these five types of cells suggested a transition of myoepithelial cells to chondroblasts. Stromal myofibroblasts also showed calponin immunoreactivity, but they did not react with a cytokeratin 14 monoclonal antibody, which recognizes myoepithelial cells in mammary gland. Calponin appears to be a very sensitive marker of normal and neoplastic myoepithelium in the canine mammary gland, and its identification in different cell types of complex and mixed tumors of the mammary gland of the dog suggests a major histogenetic role for myoepithelial cells.     

Immunohistochemical evaluation of MMP-2 and TIMP-2 in canine mammary tumours: a survival study

Canine mammary tumours (CMTs) are a very heterogeneous group of neoplasms with variable prognosis. Their aggressiveness is mainly due to their ability to invade locally and to metastasize. The degradation of extracellular matrix components is an important determinant of the invasive phenotype. The aims of this study were to analyse by immunohistochemistry and double immunofluorescence the expression of metalloproteinase 2 (MMP-2) and tissue inhibitor of metalloproteinase 2 (TIMP-2) in eight normal canine mammary glands and 118 CMTs (24 benign, 94 malignant) and to investigate relationships with metastatic disease and survival.MMP-2 and TIMP-2 expression was higher in malignant tumours than in normal canine mammary tissue and benign tumours. The main difference between benign and malignant CMTs was the pattern of expression of both molecules: benign tumours presented TIMP-2 and MMP-2 immunoreactivity in the myoepithelial cells lining the basement membrane of tubuloalveolar structures, while malignant tumours showed mainly diffuse expression in neoplastic cells. In malignant tumours, increased TIMP-2 expression was significantly associated with the development of distant metastases, lower overall survival and lower disease-free survival. MMP-2 expression was not significantly associated to any of these parameters. These results suggest that the immunohistochemical expression of TIMP-2 is a useful prognostic factor in CMTs.     

Immunohistochemistry with keratin,vimentin, desmin,and α‐smooth muscle actin monoclonal antibodies in canine mammary gland: Benign mammary tumours and duct ectasias

Summary

Duct ectasias (n=2) and different types of benign canine mammary tumours (n=19) were studied immunohistochemically with monoclonal antibodies (MoAbs) directed against various human keratin types (K), α‐smooth muscle actin, vimentin, and desmin. In the duct ectasias and in most tumours the epithelial structures revealed an inner and outer cell layer. The inner cell layer was characterized by labelling with K 7, 8, 18, 19 and mostly also with K 4 and/or K 10 MoAbs. The outer cell layer was almost invariably labelled by K 14, K 14 and 17, and a‐smooth muscle actin MoAbs. The labelling patterns of both duct ectasias and tumours corresponded largely to the patterns observed in normal mammary gland tissue, although a more distinct heterogeneity was seen. Tumours histomorphologically assumed to be of a myoepithelial origin did not show immunohistochemical features of myoepithelial cells. The myoepithelial nature of the vast majority of spindle‐shaped cells present in the adenomas of the complex type and in the fibroadenomas of the benign mixed type could not be confirmed immunohistochemically. These cells, however, unequivocally expressed vimentin, suggesting proliferation of stromal cells in these tumours, which in the fibroadenomas of the benign mixed type may show metaplasia to bone or cartilage.

In the duct ectasias and in some tumours, a fraction of elongated stromal cells, probably representing myofibroblasts, was labelled with the α‐smooth muscle actin MoAb.     

Immunohistochemical expression of Epidermal Growth Factor Receptor (EGFR) in canine mammary tissues

Epidermal Growth Factor Receptor (EGFR) has been extensively studied in human breast cancer; however, systematic studies of EGFR protein expression in canine mammary gland tumours are lacking. Therefore, we evaluated its immunohistochemical expression in a series of 136 canine mammary tumours and representative areas of adjacent normal and hyperplastic mammary tissue and investigated a possible correlation between EGFR overexpression and several clinicopathological parameters and survival. In normal and hyperplastic canine mammary glands, EGFR expression was consistently observed in myoepithelial cells, with luminal cells usually negative. In tumour tissues, EGFR overexpression was found in 9 benign (19.6%) and 38 malignant (42.2%) lesions, with EGFR positivity significantly related with malignancy. Besides animal age and tumour size, there were no significant associations between other clinicopathological parameters and EGFR overexpression. On survival analysis, tumours with EGFR overexpression showed a reduced disease-free and overall survival; however these associations failed to reach statistically significant levels.     

An osteoid variant of cutaneous melanoma in a dog detected by S100 and melan a markers

Osteoid malignant melanoma is a rare type of melanoma described in humans and dogs with some areas of bone differentiation. In this tumour, the origin of the bone matrix remains unclear. We report one case of this variant with, for the first time, a cutaneous origin in a dog. Malignant melanomas are aggressive tumours. Amelanotic tumours are sometimes difficult to recognize as they require immunohistochemical evaluation for an adequate diagnosis and we have used anti-vimentin, S100, and melan A antibodies for identification. Melan A is less sensitive but more specific than S100 in identifying amelanotic melanomas. This tumour was positive for vimentin, S100 and melan A, including the areas of osteoid. These results suggest osteoid differentiation of tumour cells rather than induced stromal metaplasia.     

p63: a novel myoepithelial cell marker in canine mammary tissues

Several immunohistochemical markers have been used to demonstrate the presence of myoepithelial cells in order to determine their role in the histogenesis of mammary tumors. p63, a recently characterized p53 homologue, is consistently expressed in myoepithelial cells of the human breast; however, no assessment of its immunoreactivity has been reported so far in canine mammary tissues. We investigated p63 immunohistochemical expression, as a novel myoepithelial cell nuclear marker, in 81 samples of normal (n = 2), hyperplastic (n = 11), and neoplastic (n = 68) canine mammary tissues. Myoepithelial phenotype was confirmed by using complementary monoclonal antibodies: alpha-smooth muscle actin, cytokeratin 14, cytokeratin AE1/AE3, and vimentin. p63 expression was observed in 91.4% (74/81) of the samples evaluated. Normal mammary glands, mammary hyperplasias, and benign tumors showed 100% immunoreactivity, with p63 expression restricted to myoepithelial cell nuclei. In general, benign mixed tumors showed a basal cell compartment immunoreactive to p63, with a gradual decrease of its expression during myoepithelial transformation. p63 expression was found in 72% of malignant tumors, allowing myoepithelial or basal cell identification in spindle-cell carcinomas (2/2), tubulopapillary carcinomas (8/9), solid carcinomas (7/10), and carcinosarcomas (1/3). The osteosarcoma analyzed was p63 negative. In our series, stromal components were consistently nonreactive to p63. In conclusion, the present study reveals p63 as a sensitive and highly specific marker of myoepithelial cells in canine mammary tissues, and the authors suggest p63 as an additional marker for defining myoepithelial histogenesis.     

Histopathological and immunohistochemical characteristics of two canine lipid-rich mammary carcinomas

Clinicopathological and immunohistochemical findings of two uncommon canine lipid-rich mammary carcinomas are described. The predominant histological feature in both tumours was the presence of at least 80% of cells with intracytoplasmic vacuoles which stained positively with Sudan IV but not with alcian-blue periodic acid-schiff method. In both tumours, small groups of non-vacuolated cells were identified among the vacuolated cells. However, histological and immunohistochemical differences were also found between these tumours. Thus, one of them was composed of tumour cells with a large and single vacuole, which were arranged in lobular pattern, while the other neoplasm showed an intraductal growth of tumour cells with a fine vacuolated cytoplasm. Immunohistochemically, in the first tumour most vacuolated cells were positive for CK (cytokeratin)8-7, indicating a secretory epithelial immunophenotype while CK5 and CK8-7-expressing non-vacuolated cells were associated with luminal duct immunophenotype. However, in the second tumour the expression of CK14 in most of vacuolated cells and alpha-smooth muscle actin (alpha-SMA) in non-vacuolated cells, alone or in combination with CK5 suggested a myoepithelial immunophenotype for both cell types. These results suggest heterogeneity of the cell type and growth pattern for this type of canine tumour as has been described in women but not in dogs.     

Calponin expression and myoepithelial cell differentiation in canine, feline and human mammary simple carcinomas

Calponin is a 34‐kDa smooth muscle‐specific protein that has been shown to be a highly sensitive marker of myoepithelial cells in canine, feline and human mammary tissue and tumours. The expression of calponin was studied in 15 canine, 32 feline and 28 human simple mammary carcinomas using a monoclonal mouse antihuman calponin antibody and the avidin–biotin peroxidase complex (ABC) immunohistochemical technique. Calponin expression was compared with the expression of cytokeratin 14, a marker of normal mammary myoepithelial cells in the three species. Four different types of calponin‐positive cells were identified: (1) Type 1: cytokeratin‐14‐positive pre‐existing myoepithelial cells forming a continuous layer with images of focal disruptions; (2) Type 2: cytokeratin‐14‐positive isolated nests of fusiform, polygonal or round cells without atypia; (3) Type 3: cytokeratin‐14‐positive atypical cells indistinguishable from non‐reactive atypical cells, which should have never been detected in haematoxylin and eosin‐stained sections and (4) Type 4: cytokeratin‐14‐negative stromal fusiform cells around the neoplastic growth or cell nests, identified as myofibroblasts. Calponin‐negative and cytokeratin‐14‐positive atypical neoplastic cells were observed in three canine, 28 feline and two human carcinomas. The latter were indicative of altered expression of high‐molecular‐weight cytokeratins in luminal epithelial‐type simple carcinomas. Our findings show that calponin is a good marker of myoepithelial cell differentiation in feline, human and, particularly, canine simple carcinomas. The high number (six out of 15) of canine tumours with type 3 cells points to the need of both introducing calponin examination in the routine diagnostic schedule and performing further studies on its prognostic significance.     

Alkaline phosphatase reaction of canine mammary mixed tumours: a light and electron microscopic study

Alkaline phosphatase (ALK-P) reactions at the light and electron microscopic levels were performed on eight canine mammary mixed tumours. In the morphologically normal gland next to the tumours, the myoepithelial cells and the stromal tissues near the acinar basement membranes both showed a positive ALK-P reaction by light microscopy. At the electron microscopic level, those parts of the myoepithelial cell membrane that were in contact with other cells--myoepithelial or lumenal epithelial--showed an ALK-P-positive reaction, as did the adjacent stromal tissue. The characteristic tumour cells, at sites of early proliferation, were ALK-P-positive, while in the mucoid and chondroid areas of the mixed tumours they were largely ALK-P-negative by light microscopy. By electron microscopy, those neoplastic cells which were in contact with other cells typically showed a positive ALK-P reaction, while those cells which were in contact with mucoid or chondroid matrix were largely negative. Although the cytoplasmic filaments of the neoplastic cells were 10 nm thick, and those of normal myoepithelial cells were 6 to 8 nm thick, the observations reported provide further evidence for the view that the essential cells of the canine mammary mixed tumour are derived from myoepithelial cells.     

Expression of p63 normal canine skin and primary cutaneous glandular carcinomas

p63, a recently identified homologue of the p53 protein, is expressed consistently in basal cells of several human multilayered epithelia. In this study, expression of p63 was determined in 31 primary cutaneous glandular carcinomas, including sebaceous, perianal (hepatoid) gland, apocrine and ceruminous carcinomas, as well as their adjacent normal skin. Similar to humans, p63 is a reliable marker for basal and myoepithelial cells in canine epidermis, cutaneous appendages and malignant apocrine and ceruminous gland neoplasms. In sebaceous carcinomas, not only basal cells, but also some sebocytes, showed nuclear staining for p63. Most mature epithelial cells in perianal gland carcinomas exhibited strong p63 expression. Based on these findings, basal/myoepithelial cells could be involved in the oncogenesis of these tumours and p63 might be used as a diagnostic marker in these lesions.     

Total serum alkaline phosphatase activity in dogs with mammary neoplasms: a prospective study on 79 natural cases

Increased total alkaline phosphatase (TALP) activity in the serum, long noticed in canine mammary tumours among other neoplasms, has not been yet associated with malignancy, osseous transformation of neoplastic tissue or histopathological typing. Therefore, the purpose of this study was to correlate this biochemical abnormality with the above-mentioned parameters, in 79 adult to elderly female dogs with mammary neoplasms, without evidence of metastatic or any other disease. Histopathology disclosed that 64 (81%) of these neoplasms were malignant and 15 (19%) benign, belonging to various histological types. Radiology and histopathology revealed the presence of osseous tissue in 18 (22.8%) cases. The malignant neoplasms were subsequently allocated into group A including 46 (74.2%) of epithelial origin and group B with 16 (25.8%) neoplasms of both epithelial and mesenchymal origin ('malignant mixed' tumours). In addition, their benign counterparts were divided into group C (adenomas, fibroadenomas) and group D (benign mixed tumours) that included seven (46.7%) and eight (53.3%) tumours, respectively. Almost 55% of the dogs with malignant and 47% with benign tumours had increased serum-TALP activity. However, no significant difference in serum-TALP activity was found between the dogs with malignant (mean +/- SE: 243.7 +/- 37.4 U/l) and benign (167.9 +/- 38.4 U/l) neoplasms, with (238.9 +/- 45.3 U/l) and without (226.5 +/- 38.3 U/l) osseous transformation, with (298.5 +/- 85.6 U/l) or without (201.2 +/- 30.5 U/l) myoepithelial cell proliferation and with different tumour size (T1/T2: 175.1 +/- 34.9 and T3: 254.5 +/- 42.5 U/l). In histopathological typing, the only difference noticed involved the malignant neoplasms of group A (190.5 +/- 25.5 U/l) compared with group B (378 +/- 124.6 U/l) dogs. The higher increase of serum-TALP activity in 'malignant mixed' tumours could not be attributed to osseous transformation or new ALP isoenzyme production by myoepithelial cells. Increased serum-TALP activity is of no apparent diagnostic (as to tumour type) or prognostic value.     

Characterization of spindle cell component of ferret (Mustela putorius furo) adrenal cortical neoplasms – correlation to clinical parameters and prognosis

Adrenal cortical epithelial tumours are common in ferrets. A variant tumour type with prominent spindle cell proliferation has been identified. We characterized these variant tumours with light microscopy and immunohistochemical analysis and correlate these features to clinical parameters and prognosis. We classified 24 ferret adrenal cortical masses with recognizable spindle cell proliferation obtained from the AMC and AFIP databases, based on percentage of spindle cells present and features of malignancy. These masses were separated into hyperplastic nodules and adenomas (both with 1–24% spindle cells), ‘mixed’ adenomas (≥25% spindle cells), adenocarcinomas (1–24%) and ‘mixed’ adenocarcinomas (≥25% spindle cells). Tumours were evaluated immunohistochemically for smooth muscle actin (SMA) and estrogen receptor (ER) expression. Disease‐free interval (DFI) and survival time (ST) were calculated using Kaplan–Meier product limit method. Of 24 cases of spindle cell variant adrenal tumours, one was a hyperplastic nodule, 10 were adenomas, three were ‘mixed’ adenomas, six were adenocarcinomas and four were ‘mixed’ adenocarcinomas. The proliferative spindle cell cytoplasm was SMA‐positive (smooth muscle myocyte origin). ER positivity, seen in nine of 24 cases, was restricted to adenocarcinomas, ‘mixed’ adenomas and ‘mixed’ adenocarcinomas. DFI and ST were significantly reduced in ‘mixed’ adenocarcinomas or tumours with ER expression. DFI was significantly reduced in tumours with marked smooth muscle. The spindle cell component of these variant adrenal tumours is smooth muscle in origin. The presence of abundant smooth muscle, a more malignant histologic grade (‘mixed’ adenocarcinomas) and ER expression are significantly positively correlated to both decreased DFI and decreased ST.     

Expression of epidermal growth factor receptor and vascular endothelial growth factor in malignant canine epithelial nasal tumours*

Epidermal growth factor (EGF) and vascular endothelial growth factor (VEGF) signalling pathways play a role in carcinogenesis. Inhibition of EGF receptor (EGFR) and of VEGF is effective in increasing the radiation responsiveness of neoplastic cells both in vitro and in human trials. In this study, immunohistochemical evaluation was employed to determine and characterize the potential protein expression levels and patterns of EGFR and VEGF in a variety of canine malignant epithelial nasal tumours. Of 24 malignant canine nasal tumours, 13 (54.2%) were positive for EGFR staining and 22 (91.7%) were positive for VEGF staining. The intensity and percentage of immunohistochemically positive neoplastic cells for EGFR varied. These findings indicate that EGFR and VEGF proteins were present in some malignant epithelial nasal tumours in the dogs, and therefore, it may be beneficial to treat canine patients with tumours that overexpress EGFR and VEGF with specific inhibitors in conjunction with radiation.     

Expression of maspin in mammary gland tumors of the dog

Maspin is a serine protease inhibitor that inhibits tumor invasion and metastasis in human breast cancer and is consistently expressed by mammary myoepithelial cells (MECs). To analyze the value of maspin as a marker of the MEC layer of the normal and tumoral canine mammary gland, the immunohistochemical expression of maspin was studied in formalin-fixed tissues from 55 benign and malignant tumors (40 tumors also contained the surrounding normal mammary gland) using a commercially available monoclonal antibody. Periacinar and periductal MECs of all 40 normal mammary glands were stained by the anti-human maspin monoclonal antibody, and immunoreactivity was observed in the nucleus and cytoplasm of these cells. In addition, maspin was found in 53 (98%) of the tumors studied, reacting with the MECs in 100% of benign tumors and 93% of malignant tumors and to the epithelial cells of 16% of benign and 73% of malignant tumors. In the MEC compartment, immunoreactivity was observed in the cytoplasm of hypertrophic MECs, fusiform MECs, stellate MECs, rounded (myoepithelial) cells, and chondroblasts. In the epithelial cell compartment, immunoreactivity was observed in the cytoplasm of cells with and without squamous differentiation. Stromal myofibroblasts were unreactive. Maspin appears to be a very sensitive marker of the normal and neoplastic myoepithelium that, contrary to smooth muscle differentiation markers, does not stain stromal myofibroblasts. In addition, a subset of neoplastic epithelial cells reacted with the maspin antibody. The relationship between maspin expression in different cellular compartments of canine mammary carcinomas and the biologic aggressiveness of the disease remains to be elucidated.     

The contribution of stem cells to epidermal and hair follicle tumours in the dog

Background –  Although cutaneous stem cells have been implicated in skin tumourigenesis in humans, no studies have been conducted to elucidate the presence and the possible role of stem cells in hair follicle tumours in the dog. Hypothesis –  Stem cell markers are expressed in canine epidermal and follicular tumours and can be used to better understand the biology and origin of these tumours. Animals and Methods –  In the present study, normal skin sections and 44 follicular tumours were retrospectively investigated for the immunohistochemical expression of keratin 15 (K15) and nestin. In addition, 30 squamous cell carcinomas were evaluated for K15 expression. Results –  In normal skin, K15 and nestin were expressed in the outer root sheath cells of the isthmic portion of the hair follicle (bulge region), and K15 expression was also scattered in the basal cell layer of the epidermis. Infundibular keratinizing acanthomas, pilomatricomas and squamous cell carcinomas were mostly negative for K15, trichoblastomas were moderately to strongly positive, tricholemmomas were either negative or strongly positive, and trichoepitheliomas had heterogeneous staining. Nestin expression was generally faint in all follicular tumours. Conclusions and clinical importance –  Our results show that K15 can be a reliable marker for investigating the role of stem cells in hair follicle tumours of the dog, while nestin was judged to be a nonoptimal marker. Furthermore, our study suggests that hair follicle stem cells are present in the bulge region of hair follicles and could possibly play a role in tumourigenesis of canine tumours originating from this portion of the follicle, namely trichoblastomas, tricholemmomas and trichoepitheliomas. The loss of K15 expression in squamous cell carcinomas compared with normal skin suggests that this event could be important in the malignant transformation.     

The expression of p63 and cytokeratin 5 in mixed tumors of the canine mammary gland provides new insights into the histogenesis of these neoplasms

Cytokeratin 5 and p63 have been described as basal and myoepithelial cell markers in human breast. Mixed tumors of the canine mammary gland have been associated with a myoepithelial origin. Cytokeratin 5 expression has not been evaluated in these tumors. We investigated the relation between cytokeratin 5 and p63 double-immunohistochemical expression in 23 mixed tumors of the canine mammary gland (10 benign mixed tumors and 13 carcinomas arising from benign mixed tumors) and their origin. Cytokeratin 5 and p63 co-expression was observed in myoepithelial cells of benign mixed tumors, as well as in squamous differentiation of carcinoma arising from benign mixed tumors. Though a few interstitial spindle cells of the mesenchymal components expressed both p63 and cytokeratin 5, the basal epithelial cells were labeled only by cytokeratin 5. The co-expression of p63 and cytokeratin 5 in myoepithelial cells and squamous differentiation suggest that, like in human breast, cytokeratin 5 can also be considered a myoepithelial- and squamous-cell differentiating marker in canine tumors. The presence of some interstitial spindle cells stained for p63 and cytokeratin 5 might be associated with a myoepithelial origin of the mesenchymal component of mixed tumors of the canine mammary gland. Moreover, contrary to p63, basal epithelial cells were labeled by cytokeratin 5, indicating that cytokeratin 5 may not represent an exclusive myoepithelial cell marker but also a basal epithelial cell marker in canine mixed tumors. According to these data, basal epithelial cells may be related to the origin of the epithelial component of mixed tumors of the canine mammary gland.     

Expression of bone morphogenetic protein-6 (BMP-6) in myoepithelial cells in canine mammary gland tumors.

Seventy-three mammary tumors and three mammary tissue specimens were examined to elucidate the expression of bone morphogenetic protein (BMP)-6 in the myoepithelial cells of canine mammary gland tumors. Morphologically, the myoepithelial cells were classified into four types: resting and proliferating cells inside the basement membrane, and spindle- and star-shaped cells proliferating in the outer area of the basement membrane. The characteristics of these myoepithelial cells were confirmed by immunohistochemistry using antibodies raised against keratin, cytokeratin 19, alpha-smooth muscle actin, and vimentin. In simple adenoma, a small number of resting myoepithelial cells was immunopositive for BMP-6. In complex adenomas and benign mixed tumors, all types of myoepithelial cells, depending in some cases on their specific location within the tumor, were immunopositive for BMP-6, but almost all of the tubular epithelial cells were immunonegative. Foci consisting of a proliferation of BMP-6-positive star- and spindle-shaped cells had mucinous stroma with marked hyaline and chondroid changes. In contrast, the foci with BMP-6-negative spindle- and star-shaped cells tended to have mucinous stroma without chondroid change. Several types of mesenchymal cells including chondrocytes, osteoblasts, and fibroblastlike cells in the mixed tumors, showed an intense immunopositive reaction for the BMP-6 antibody, and were located close to the ectopic cartilage and bone matrix. No significant immunoreactivity for BMP-6 was observed in most of the malignant mammary tumors; only one malignant mixed tumor was examined. All of these findings indicate that BMP-6 expression in myoepithelial cells may increase in complex adenomas and benign mixed tumors in canine mammary glands, and that BMP-6 expression is most intense in the vicinity of chondroid matrix in these tumors.