Spontaneous Mammary Intraepithelial Lesions in Dogs—A Model of Breast Cancer

Introduction

Breast cancer is the second leading cause of cancer deaths in women (1). Breast intraepithelial lesions (IEL), such as usual hyperplasia, atypical hyperplasia, and carcinoma in situ, are frequently diagnosed nowadays due to the success of mammographic screening, education programs, and awareness by women. These lesions are recognized as risk factors for invasive cancer and their presence affects patient management decisions (2, 3). Alarmingly, about a half million cases of breast IELs are diagnosed every year (4). Surgery, plus tamoxifen therapy, is recommended for women with estrogen receptor (ER)-positive ductal carcinoma in situ (DCIS); however, a “wait and watch” approach is usually recommended for lower-grade lesions and for lobular carcinoma in situ. Therefore, these precancerous lesions are proposed targets for development of new chemotherapeutic or preventive agents and as end points for clinical trials (5, 6). Establishing an animal model of spontaneous mammary IELs is paramount for testing new treatment and preventive modalities before human clinical trials.

Unlike laboratory rodents, dogs share a common environment with people and, therefore, may be exposed to some of the same carcinogens (7). They also are more outbred than laboratory rodents, yet certain breeds are at increased risk for developing mammary tumors (7). Because dogs have a shorter life span than people, it is possible to study mammary IELs and invasive tumors that develop after a few years instead of decades. Importantly, dogs with mammary adenocarcinoma are appropriate subjects for the study of breast cancer because the mammary gland is the most common site of neoplasia in female dogs (8, 9) and dysplasias develop before tumors in canine mammary tissue (10-13).

Estimates of annual incidence of canine mammary neoplasia depend on the population studied and vary widely from 198 (14) or 205 per 100,000 dogs (15) to 111 per 10,000 female dogs between 3 and 10 years of age (16). Prepubertal or pubertal ovariohysterectomy is more commonly practiced now that veterinarians and pet owners are keenly aware of the predilection of the unspayed female dog for mammary tumors and has resulted in significant reduction in the incidence of canine mammary tumors in the United States. In countries where ovariohysterectomy is less routine, mammary tumors are more common (17). One of four non-ovariohysterectomized female dogs over 4 years of age is expected to develop mammary neoplasia (18). Thirty percent to 50% of canine mammary tumors are malignant and 50% to 75% of these recur or metastasize within 1 to 2 years (7, 10, 17, 18).

The risk of canine mammary neoplasia is affected by exposure to estrogen in early mammary development. Ovariohysterectomy before the first, second, or third estrous cycle reduces the relative risk of developing mammary tumors to 0.5%, 8%, or 26%, respectively (19). Ovariohysterectomy later in life had no significant effect on risk of developing mammary tumors in that study (19). These data not only implicate ovarian hormones in the development of mammary tumors in dogs but also suggest that the hormonal effect is exerted in the first years of life and that continuous hormonal influence is unnecessary. However, results of a more recent study (20) suggest that ovariohysterectomy is an effective adjunct to tumor removal in dogs with mammary carcinoma. Interestingly, dogs in that study that were spayed <2 years before excision of the mammary carcinoma survived longer than those that were spayed >2 years before surgery for the mammary carcinoma. As in humans, advancing age, progesterone treatment, obesity in early life, and diet also increased the risk of mammary tumors in the dog (21).

A genetic alteration commonly found in human breast cancer is amplification of HER-2 genes (22). HER-2 protein expression has been used to predict patient response to treatment. Expression of HER-2/neu (c-erbB-2) in canine mammary carcinoma is similar to that in human breast carcinoma, suggesting a possible role in carcinogenesis and value as a prognostic indicator (23, 24).

A model of spontaneous mammary IELs with pathologic similarity to those in humans is needed. This study was designed to determine the prevalence of IELs in mastectomy specimens from dogs with palpable mammary tumors; to correlate the type of IEL with the type of mammary tumor; to determine the expression of ER-α, progesterone, HER-2, and Ki67 in IELs by immunohistochemistry; and to compare canine mammary IELs with those in the human breast.

Materials and Methods

Pathology Evaluation

Formalin-fixed, paraffin-embedded specimens of 212 mammary tumors from 200 female dogs were retrieved from the archives of the Purdue University Animal Disease Diagnostic Laboratory and Veterinary Teaching Hospital (West Lafayette, IN) and the Institute of General Pathology and Anatomical Pathology, Sassari University (Sardinia, Italy). Tumors were classified according to WHO Histological Classification of Mammary Tumors of the Dogs and Cats by two veterinary pathologists (E. Antuofermo and M.A. Miller; ref. 11), and adjacent mammary tissue was evaluated for the presence of IELs.

Mammary IELs were classified (25) in consultation with a pathologist (S. Badve) and compared with human mammary IELs in biopsy specimens submitted to the Department of Pathology, Indiana University School of Medicine. Adenosis was diagnosed when proliferation of epithelial and myoepithelial cells resulted in expansion of mammary lobules with an increased number of ductules and acini. Nuclear atypia was not obvious and mitotic figures were generally absent. Sclerosing adenosis was diagnosed when adenosis was associated with fibrosis and distortion of the lobular architecture. Intraductal papilloma was distinguished from ductal hyperplasia by its solitary nature. These lesions consisted of papillary fronds arising from a segment of the ductal lining and projecting into the ductal lumen. Papillomas had proliferation of ductal epithelium and myoepithelium without pleomorphism or nuclear atypia. When stromal fibrosis was prominent, these were classified as sclerosing papillomas.

Usual ductal hyperplasia was diagnosed when ductal epithelial cellularity was increased without atypia. Affected ducts were partially filled by a mixed population of epithelial and myoepithelial cells; irregularly shaped fenestrations were lined by epithelial cells without polarity. Cords of epithelial cells sometimes bridged the ductal lumina. Epithelial cells often seemed to overlap because cytoplasmic boundaries were indistinct and nuclei were often oriented in a “streaming” fashion with the long axes of cells parallel to fenestrations.

Atypical ductal hyperplasia (ADH) was diagnosed when some but not all features of DCIS were identified (26). ADH was distinguished from low- or intermediate-grade DCIS by the rudimentary and irregular shape of fenestrations and by the lack of polarity of the proliferative epithelial cells such that they were generally oriented with their long axes parallel to the fenestrations. In lesions with low-grade cellular atypia, size of the proliferation was also considered as recommended by Schnitt et al. (27). Whereas high-grade cellular atypia always warranted the classification of high-grade DCIS, if features of low- or intermediate-grade DCIS were observed, but did not involve more than one duct cross-section, the IEL was classified as ADH.

DCIS was diagnosed when ductal epithelial proliferation fully involved at least two ductal cross-sections, when the proliferation was of one cell type (epithelial) with cellular and architectural atypia, such as formation of discrete cribriform spaces lined by polarized epithelial cells. DCIS was further subdivided into low grade (monomorphic cells, no increase in nuclear size; diffuse, fine chromatin, few prominent nucleoli or mitotic figures), intermediate grade (features intermediate between low and high grade), and high grade (pleomorphic cells, nuclear diameter >2.5 that of normal ductal epithelial cells, vesicular chromatin, prominent or multiple nucleoli, and mitotic figures). Patterns observed in DCIS included cribriform, papillary, micropapillary, solid, and solid with comedo-type necrosis. The cribriform pattern was common with low-grade DCIS, whereas solid, especially solid with comedo necrosis, was a feature of high-grade DCIS.

Canine mammary IELs were categorized according to the relative risk of developing invasive mammary carcinoma ascribed to their human counterparts (25). The invasive canine mammary carcinomas were not graded because a grading system that correlates strongly with biological behavior or prognosis has not been established for mammary tumors in dogs. However, multiple studies have shown that overall survival is longest in complex carcinomas, intermediate in simple carcinomas, and shortest in sarcomas (10, 11). Within the group of simple carcinomas, aggressiveness increases from tubulopapillary carcinoma to solid carcinoma to anaplastic carcinoma (10, 11).

Immunohistochemistry

Immunohistochemistry was done according to the standard avidin-biotin-peroxidase complex method. Briefly, histologic sections (5 μm) with IELs and adjacent neoplastic and nonneoplastic mammary tissue were cut from formalin-fixed, paraffin-embedded tissues and mounted on positively charged Superfrost slides (Fisher Scientific). Tissue sections were deparaffinized and rehydrated in graded ethanol. Antigens were retrieved by steaming in 0.01 mol/L citrate buffer (pH 6.0) in a water bath at 95°C temperature for 10 min. Tissues were then blocked for endogenous peroxidase in 3% hydrogen peroxide in methanol and for nonspecific binding in TBS containing 0.3% Triton X-100, 0.2% saponin, and 0.5% blocking agent (NEN Life). Tissues were incubated overnight at 4°C in the following antisera: ER-α monoclonal mouse anti-human antibody clone CC4-6 (28-30) at 1:25 dilution, progesterone receptor (PR) NCL-PGR antibody clone 1A6 (29, 30) at 1:40 dilution (Novocastra Ltd.), Ki67 monoclonal mouse anti-human antibody clone 7B11 (31) at 1:100 dilution (Zymed Laboratories), and HER-2/neu polyclonal rabbit anti-human antibody (23, 24) at 1:100 dilution (DakoCytomation) followed by biotinylated horse anti-mouse or goat anti-rabbit secondary antibodies (Vector Laboratories). The chromogen was 3,3′-diaminobenzidine (Vector Laboratories). Sections were counterstained in hematoxylin (Richard Allen Scientific) and then coverslipped in 50:50 xylene/Permount (Fisher Scientific). Negative control slides were treated either with no serum or with isotype-matched IgG serum. The positive controls consisted of sections from human breast carcinoma known to overexpress ER, PR, and HER-2/neu receptors and Ki67 nuclear antigen (DakoCytomation). Some IELs could not be evaluated immunohistochemically because insufficient lesional tissue remained in the paraffin blocks.

All slides were reviewed independently by two authors (E. Antuofermo and S.I. Mohammed) and questionable cases were reviewed simultaneously to reach consensus. Nuclear staining was scored for expression of ER, PR, and Ki67. ER, PR, and Ki67 expression in IELs and tumor tissues was analyzed and reported as the percentage of epithelial cells with positive staining. The intensity of ER, PR, and Ki67 immunostaining was graded on a scale of 0 to 3 where 0 = no staining, 1 = weak staining, 2 = moderate staining, and 3 = strong staining. The expression of HER-2 was defined as cell membrane staining of epithelial cells and scored according to the HercepTest kit guide; scores of 0 or 1+ were considered negative for HER-2 expression, 2+ was weak-positive, and 3+ was strong-positive complete membrane staining in >10% of the epithelial cells.

Statistical Analysis

The statistical analysis was done using Student's t test, Tukey standardized test, and one-way ANOVA test. Statistical Analysis System software (SAS) was used for the statistical analysis. P < 0.05 was considered significant in all analyses.

Results

Ninety-three benign and 119 malignant mammary tumors were found in excisional specimens from the 200 dogs in this study. Nonneoplastic mammary parenchyma was included in most mastectomy specimens and was histologically similar to nonneoplastic human mammary tissue (Fig. 1A and B ). Intralobular ductules and interlobular ducts were lined by a single layer of columnar cells supported by a row of myoepithelial cells. Mammary IELs were detected in excisional specimens from 60 (30%) of the 200 dogs. These canine mammary IELs resembled their human counterparts histologically and are listed in order of increasing relative risk (as ascribed to human IELs; ref. 25) in Table 1 . An IEL unassociated with increased risk of invasive carcinoma in humans, adenosis, was found in four dogs and associated with benign tumors. Low-risk IELs included 2 cases of sclerosing adenosis, 3 of intraductal papilloma, 3 of sclerosing papilloma, and 10 of usual (without atypia) ductal hyperplasia. All dogs with intraductal papilloma and 8 of 10 dogs with ductal hyperplasia had benign mammary tumors; however, 2 of 10 dogs with ductal hyperplasia and 3 of 3 dogs with sclerosing papilloma had malignant tumors. Canine (Fig. 1C) and human (Fig. 1D) ductal hyperplasia were histologically similar.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

IELs in canine and human mammary biopsies. Canine (A, C, and E) and human (B, D, and F) mammary biopsies with nonneoplastic gland (A and B), usual ductal hyperplasia (C and D), and ADH (E and F). H&E stain. Magnification, ×200.

Figure 1E and F depicts micropapillary fronds of ADH in the canine and human mammary gland, respectively. Five of six (83%) dogs with ADH had malignant mammary tumors.

Of the 32 dogs with DCIS, 29 (91%) had malignant mammary tumors. Figure 2A and B depicts low-grade DCIS with cribriform pattern in a dog and woman, respectively. Four of six dogs (67%) with low-grade DCIS had malignant tumors. Figure 2C and D depicts intermediate-grade DCIS with papillary pattern in a dog and woman, respectively. Proliferative epithelial cells that lined papillary fronds were more pleomorphic and had larger nuclei than those of low-grade DCIS. Nine of 10 dogs (90%) with intermediate-grade DCIS had malignant tumors. Figure 2E and F depicts high-grade DCIS with solid pattern and comedo necrosis in a dog and woman, respectively. Epithelial cells were nonoverlapping with vesicular nuclei and prominent nucleoli. All 16 dogs (100%) with high-grade DCIS had malignant tumors.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Canine (A, C, and E) and human (B, D, and F) mammary biopsies with DCIS: low (A and B), intermediate (C and D), and high grade (E and F). H&E stain. Magnification, ×200.

We have not followed these 200 dogs to document the behavior of IELs. However, to correlate the IEL grade with the invasiveness of the mammary tumors, we evaluated each mastectomy specimen for evidence of lymphatic or nodal invasion by tumor cells. Lymphatic involvement was found in 0 of 22 tumor specimens associated with low-risk IELs (see Table 1), 1 of 6 specimens with ADH, 1 of 6 specimens with low-grade DCIS, 0 of 10 specimens with intermediate-grade DCIS, and 8 of 16 specimens with high-grade DCIS.

Expression of ERs in IELs

In nonneoplastic canine mammary tissue, 80% to 100% of epithelial cells had strong ER immunoreactivity. Positive immunoreactivity to ER was also noted in 10 of 10 (100%) lesions of ductal hyperplasia, 1 of 5 (20%) ADH, 0 of 6 low-grade DCIS, 5 of 6 (83%) intermediate-grade DCIS, and 1 of 11 (9%) high-grade DCIS. The percentage of ER-positive cells in ductal hyperplasia lesions ranged from 50% to 90%, whereas the percentage of ER-positive cells in DCIS ranged from 0% in low-grade to 0% to 5% in high-grade lesions (Fig. 3A1-E1 ). Only one high-grade DCIS lesion had ER expression in 90% of the cells. This decrease in the percentage of ER immunoreactivity as the IEL lesion increased in grade was significant (P < 0.0001). The ER immunoreactivity of benign and malignant mammary tumors was similar to that of their associated IELs and immunoreactivity for ER was significantly decreased (P < 0.0009), with 90% to 95% cells showing no staining as the tumor increased in architectural or cytologic atypia.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Expression of ER, Ki67, and HER-2 proteins in IELs in canine mammary tissues. A1 to E1, ER. A2 to E2, Ki67. A3 to E3, HER-2. A1 to A3. Nonneoplastic mammary gland. B1 to B3. Usual ductal hyperplasia. C1 to C3. ADH. D1 to D3. Low-grade DCIS. E1 to E3. High-grade DCIS. Very strong nuclear staining is observed for ER in nonneoplastic mammary tissue (A1) and usual ductal hyperplasia (B1), whereas very weak nuclear staining is observed in ADH (C1) and low-grade (D1) and high-grade (E1) DCIS. In contrast, staining intensity for Ki67 is opposite to that of ER, weak nuclear staining is seen in nonneoplastic (A2) and usual hyperplasia (B2), whereas strong nuclear staining is seen in basilar cells of low-grade DCIS (D2) and in all cells in high-grade DCIS (E2). Similarly, strong membranous staining of HER-2 is not seen with nonneoplastic (A3) or usual ductal hyperplasia (B3), is weak in atypical hyperplasia (C3), but is clearly seen in high-grade DCIS (E3). Immunohistochemistry. Magnifications, ×200 (ER and Ki67) and ×400 (HER-2).

Expression of PRs in IELs

PR staining varied among IELs and their associated tumors and did not correlate with the type of lesion or risk for malignancy.

Expression of Ki67 Nuclear Antigen in IELs

Nonneoplastic mammary tissue had immunoreactivity for Ki67 in a few cells (Fig. 3A2). All IELs except low-grade DCIS were somewhat immunoreactive to Ki67 (Fig. 3A2-E2). The percent positive epithelial cells ranged from 1 to 10 in ductal hyperplasia; in all other IELs, there was a trend for increased numbers of positive cells as the lesion increased in grade, with the highest range of Ki67-positive cells in high-grade DCIS. Similarly, all IEL-associated tumors were immunoreactive for Ki67 with increasing positive cells (10-90%; P < 0.0041) in more histologically aggressive tumors.

Expression of HER-2 in IELs

A few HER-2 positive cells (10-20%) with 2+ staining intensity were seen in 9 of 10 ductal hyperplasias and 2 of 6 ADH lesions, whereas all cells in low-grade DCIS were negative. HER-2 staining was weakly positive (2+) in 8 of 11 high-grade DCIS, with 50% to 90% of the epithelial cells showing membranous staining (Fig. 3A3-E3). On the other hand, 3 of 11 DCIS scored 0 and were negative for HER-2. HER-2 expression in adjacent mammary tumors was similar to that in the IEL with which they were spatially associated.

Discussion

In 30% (60 of 200) of dogs with palpable mammary tumors, archived formalin-fixed, paraffin-embedded tissues contained IELs in mammary tissue adjacent to the tumors. Because the histologic sections were prepared from mastectomy or lumpectomy specimens solely for diagnostic purposes and included minimal nonneoplastic tissue, the prevalence of IELs may have been underestimated by this retrospective study. Of these IELs, 39 (65%) were associated with malignant mammary tumors. DCIS was the most common mammary IEL, accounting for 32 cases, of which 29 (90%) were associated with malignant tumors. DCIS is also the most frequently diagnosed IEL in human mammary biopsy specimens and is usually heralded by the mammographic detection of calcifications (25). In the dogs in this study, microscopic calcifications were observed in only 4 of 32 cases. This difference could reflect interspecies variation or only the difference in sampling techniques. The canine mammary specimens were collected because of the presence of a palpable mammary mass and only examined retrospectively for IELs. Peritumoral areas were not specifically sampled. It is possible that sampling of peritumoral areas and specimen radiographs will increase the detection of IELs.

ADH and sclerosing papilloma were detected less commonly than DCIS. These lesions were usually associated with malignant tumors, whereas usual ductal hyperplasia, adenosis, sclerosing adenosis, and (nonsclerosing) papilloma were usually associated with benign mammary tumors.

Complex (both epithelial and myoepithelial) carcinomas and simple (almost always epithelial) carcinomas are common malignant mammary tumors in dogs (10, 11). Sarcomas and carcinosarcomas are relatively rare (10, 11). Sarcomas were not observed in association with canine mammary IELs. The malignant tumors associated with IELs included complex carcinoma, several types of simple carcinoma (tubulopapillary, solid, and anaplastic), and carcinosarcoma.

Because IELs are considered a risk factor for invasive cancer, an AACR task force has recommended accelerated development of chemotherapeutic modalities to treat, prevent, or delay their formation (5). However, because the current standard therapy for DCIS is complete excision, either by lumpectomy or mastectomy, phase III trials with eradication of DCIS as an end point are not feasible (5). Furthermore, few breast IELs progress to cancer, a process that can take 2 decades or more (5, 6). Thus, studies to examine progression and invasion of these lesions in humans are not feasible and development and evaluation of noninvasive therapy for breast IELs require an animal model.

Mouse models of human breast cancer include xenografts; tumors induced by chemicals, radiation, or mouse mammary tumor virus; and genetically engineered mice (32). Spontaneous mammary tumors are rare in mice (32). Most mouse mammary tumors are histologically distinct, although some genetically engineered mouse mammary tumors resemble human breast cancers (33); mammary IELs have not been well characterized in genetically engineered mouse models (34).

As in humans and in contrast to mice and rats, spontaneous mammary tumors are the most common neoplasm of female dogs (10, 11). Spontaneous mammary IELs are common in dogs well before the age of onset of palpable mammary tumors. Warner (13) found that none of 16 sexually intact female beagles <2 years of age had mammary dysplasias, whereas 7 of 11 females 2 to 3 years of age and 10 of 12 females 3 to 4 years of age had numerous “dysplasias,” detected by dissecting microscopic examination of whole-mount mammary specimens; unfortunately, the histologic appearance of these dysplasias was not described. Gilbertson et al. (12) analyzed 232 mastectomy specimens from dogs with benign or malignant epithelial tumors for ductal epithelial changes, which were categorized as normotypic and normoplastic epithelium, normotypic and hyperplastic epithelium, moderately atypical epithelial proliferation, markedly atypical epithelial proliferation, and severely atypical epithelial proliferation limited to the boundaries of the duct (in situ carcinoma). They identified precancerous atypical and noninvasive malignant neoplasms in 50 of 74 mastectomy specimens from dogs with malignant epithelial tumors that had morphologic features that would be considered precursor lesions in women. Additionally, it was noted that nuclear differentiation was a prognostically significant variable in predicting the development of de novo or recurrent invasive carcinoma within 2 years. Although the precancerous ductal proliferative lesions were not associated with increased incidence of malignancy in the first 2 years, dogs that were followed longer than 2 years had 9-fold increase in risk of invasive cancer (12). Although these earlier studies do not provide direct evidence that canine mammary IELs progress to invasive carcinoma, they have shown that IELs develop at an earlier age than mammary tumors in dogs and that the degree of nuclear atypia in canine IELs correlates with the risk of developing invasive carcinoma. In the study reported here, the fact that IELs of higher grade were more commonly adjacent to malignant mammary tumors supports their value as prognostic indicators.

There is considerable disagreement among pathologists on the classification of human breast IELs (4, 25, 27, 35-39). There is even less consensus in veterinary medicine, and although mammary IELs have been reported (10-13), and veterinary pathologists have attempted to classify canine mammary tumors according to criteria used for human mammary tumors as much as possible (10, 11), IELs have not been compared directly with human IELs. The current study documents the similarity between human and canine spontaneous mammary IELs and supports use of the dog as a model of spontaneous mammary carcinogenesis. One limitation of previous and the current study is the lack of data on the biological properties of canine mammary IELs. Despite their histologic and immunohistochemical similarity, we do not know whether canine mammary IELs have the same biological behavior or prognostic value as their human counterparts. Future transplantation studies or therapeutic/chemopreventive trials could assess their biological potential (40).

ER and PR expression in breast epithelial cells is considered a marker for risk of developing breast cancer and for preventive intervention (41-43); HER-2 is a prognostic marker (44-46). Therefore, in this study, we have evaluated the expression of ER, PR, and HER-2 in adjacent nonneoplastic canine mammary tissue, IELs, and tumors by immunohistochemistry. In addition, we measured Ki67 by immunohistochemistry because, in human breast cancer, expression of ER is usually inversely related to the proliferation index of the tumor (47, 48). ER expression in nonneoplastic mammary tissues and in IELs varied in the proportion of positive cells and intensity of immunoreactivity. Adjacent nonneoplastic mammary tissue was positive for ER with strong immunoreactivity in >70% of cells. This prominent ER expression in nonneoplastic tissue adjacent to a mammary tumor or IEL could represent a precancerous change (49). However, the expression of ER protein decreased as the IELs increased in grade. Most ductal hyperplasia specimens expressed ER protein, whereas high-grade DCIS was almost always negative for ER protein. Similar results have been reported in normal, dysplastic, and tumor from canine mammary gland (50, 51) and some human breast cancers (52, 53). Most human breast cancers are ER positive and ER expression usually increases with progression from normal to DCIS to invasive carcinoma. However, ∼30% of human breast cancers do not express ER. Precursor lesions for ER-negative breast cancers have not yet been identified. Canine mammary IELs could be a model for ER-negative tumors.

Previous studies established the association between tumor growth fractions measured by Ki67 index as a prognostic indicator (47). In humans, Ki67 expression is associated with increased tumor grade, decreased overall survival rates, and poor response to hormonal therapy. In this study, the expression of Ki67 and ER was inversely associated in that highly proliferative lesions tended to lose ER expression. Similar findings were reported in canine mammary dysplasias and tumors (54).

Lesions with increased Ki67 expression, although ER negative, were usually HER-2 positive. HER-2 protein overexpression is an important prognostic marker as well as a target for molecular-directed therapy. Fifteen percent to 25% of human breast cancers (55, 56) and 17% of canine mammary carcinomas (24) overexpress HER-2 protein. The distribution of expression of HER-2 in canine IELs was similar to that reported in various human IELs (57). HER-2 expression was significant in proliferative lesions with high risk.

Sexually intact dogs with mammary tumors are often ovariohysterectomized at the time of mastectomy. Although prepubertal or pubertal oophorectomy markedly reduces the risk of developing mammary carcinoma (19), there is less evidence of preventive or therapeutic effect of oophorectomy in middle-aged or older dogs. However, late oophorectomy may reduce the risk of developing benign mammary tumors (10). This differing effect of oophorectomy may reflect the low prevalence of ER expression in canine mammary carcinomas in comparison with that in benign canine mammary tumors. In women with BRCA mutations, postpubertal oophorectomy significantly decreased the risk of developing breast cancer in the subsequent 15 years, particularly if done before 40 years of age (58). The canine BRCA1 gene has 84% homology with the human gene, but its role in mammary cancer has not been determined (59).

Few studies have examined the efficacy of tamoxifen in canine mammary carcinoma. Twenty-three dogs with mammary tumors that were treated by surgery and ovariohysterectomy were additionally treated empirically (ER status of those tumors was unknown) with tamoxifen (60). Pet owner compliance was poor probably because of undesirable estrogenic effects of tamoxifen in the dogs and no conclusions were drawn by these authors about the efficacy of tamoxifen.

With routine mammography, detection of noninvasive IELs, such as usual hyperplasia, atypical hyperplasia, and DCIS, is much more frequent. However, management of these lesions is a problem. Although not all invasive cancers are preceded by DCIS and not all DCIS lead to invasive cancer, there is a real chance that invasive cancer may develop in a patient with DCIS despite surgical excision. From our experience, which is supported by preliminary data from our ongoing studies of mammary IELs in healthy dogs without palpable tumors, a good fraction of dogs will have ER-negative and HER-2 positive IELs and another fraction will have ER-negative, PR-negative, and HER-2 negative IELs. Animal models of spontaneous precursor lesions for ER-negative and HER-2-positive and for triple-negative breast cancer are needed for chemopreventive trials.

Our goal is to evaluate the dog as a model for human breast DCIS and other IELs. When fully characterized, this model could be examined to answer important questions in human as well as comparative medicine. Preliminary studies reported here show that the canine mammary gland bears significant similarity to the human breast. Canine IELs are morphologically identical to their human counterparts and may undergo calcifications akin to human lesions. Additional studies are being done to analyze the frequency of microcalcifications in canine mammary glands and whether these can be used to detect and follow IELs and provide a model for testing noninvasive modalities for the treatment of DCIS.