Breed-related differences in altered BRCA1 expression, phenotype and subtypein malignant canine mammary tumors

Keum-Soon Im a,1, Il-Hwan Kim a,1, Na-Hyun Kim a, Ha-Young Lim a, Jong-Hyuk Kim b,c, Jung-Hyang Sur a,!a Department of Veterinary Pathology, Small Animal Tumor Diagnostic Center, College of Veterinary Medicine, Konkuk University, 1 Hwayang-dong, Kwangjin-gu, Seoul 143-701,Republic of Koreab Department of Veterinary Clinical Science, College of Veterinary Medicine, University of Minnesota, St. Paul, MN 55108, USAc Masonic Cancer Center, University of Minnesota, Minneapolis, MN 55455, USA

a r t i c l e i n f o

Article history:Accepted 11 July 2012

Keywords:Mammary tumorsShih TzuBRCA1ImmunohistochemistryBasal-like subtypeDog

a b s t r a c t

BRCA1 is a high-penetrance breast cancer susceptibility gene and BRCA1-associated breast cancer has ahigh familial prevalence that is more common among certain populations of humans. A similar high prev-alence also exists for canine mammary tumors (CMTs) and the objective of this study was to determinethe breed-related differences in malignant CMTs. Comparative analyses of the expression of various prog-nostic factors for CMTs, including BRCA1, estrogen receptor (ER), progesterone receptor (PR), and humanepidermal growth factor receptor-2 (HER-2) were conducted on 139 malignant CMT cases from fivebreeds with the highest prevalence of CMTs in Korea.

Significant breed-related differences were observed in the expression of BRCA1 (P = 0.003), histologicalgrade (P = 0.038), and extensive lymphatic invasion (P = 0.042). The Shih Tzu breed had the highest pro-portion of dogs with malignant CMT and strong overexpression of BRCA1. Cytoplasmic and membranousexpression of BRCA1 was associated with the ER negative (P = 0.004), PR negative (P = 0.046), and triplenegative (ER, PR, and HER-2 negative; P = 0.016) phenotype and the basal-like molecular subtype(P = 0.019) in Shih Tzu dogs. Since these features are similar to BRCA1-related human breast cancer, dogswith BRCA1-associated CMT, particularly Shih Tzu dogs, may serve as a suitable spontaneous model,although additional molecular studies are needed.

! 2012 Elsevier Ltd. All rights reserved.

Introduction

Mammary tumors are the most common neoplasms in women(Yassaee et al., 2002) and female dogs (Morrison, 2002). However,the mechanism of mammary tumor development has not yet beenwell defined because of the morphological and biological heteroge-neity of these tumors. Numerous studies have determined prog-nostic factors underlying mammary tumors in humans(Fitzgibbons et al., 2000; Harris et al., 2007) and dogs (Hellménet al., 1993; Klopfleisch et al., 2011).

Environmental, hereditary, and biological factors, such as age,hormonal profile, and obesity, may be involved in the developmentof mammary neoplasms (Sorenmo et al., 2011). High-penetrancebreast cancer susceptibility genes, including BRCA1, BRCA2, CHEK2,TOX3, TP53, and PTEN, are responsible for 5–10% of human breastcancers (HBCs) (Rivera et al., 2009). Among these, BRCA1 andBRCA2 are the most well-known genes associated with a high inci-

dence of HBCs (Da Silva and Lakhani, 2010; Vargas et al., 2011). Theprevalence of BRCA1 mutations is high in individuals with a familyhistory of HBC (van der Groep et al., 2011). Furthermore, the de-gree and locus of the BRCA1 mutation differs across races or ethnic-ities, as found in the studies conducted on Swedes (Johannssonet al., 1996), Hispanics (Lagos-Jaramillo et al., 2011), AshkenaziJews (Offit et al., 1996), African–Americans (Gao et al., 1997), andKoreans (Ahn et al., 2007).

Several studies (Nieto et al., 2003; Klopfleisch and Gruber,2009; Rivera and von Euler, 2011) have described BRCA1 mRNAor protein expression and BRCA1 mutation in canine mammary tu-mors (CMTs). BRCA1 is involved in DNA repair and recombination,cell cycle checkpoints, apoptosis, and transcriptional regulation(Deng, 2006). Although its role in the pathogenesis of CMTs is stillunclear, expression levels and localization of BRCA1 are signifi-cantly different in normal mammary glands compared to benignand malignant CMTs (Nieto et al., 2003; Kim et al., 2010).

Tracing the family history of dogs is difficult, but CMTs are com-mon in certain breeds (Sorenmo et al., 2011) and the objective ofthis study was to determine whether the characteristics of malig-nant CMTs differ across dog breeds, just as the incidence of HBCs

1090-0233/$ - see front matter ! 2012 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.tvjl.2012.07.014

! Corresponding author. Tel.: +82 2 450 4153.E-mail address: jsur@konkuk.ac.kr (J.-H. Sur).

1 These authors contributed equally to the work.

The Veterinary Journal 195 (2013) 366–372

Contents lists available at SciVerse ScienceDirect

The Veterinary Journal

journal homepage: www.elsevier .com/ locate / tv j l

differs across races or ethnicities. To this end, we analyzed the pat-tern of BRCA1, estrogen receptor (ER) and progesterone receptor(PR) expression; human epidermal growth factor receptor-2(HER-2) overexpression; triple-negative phenotype and molecularsubtype; and clinicohistopathological diversity across five breedswith the highest prevalence of CMTs in Korea.

Material and methods

Samples

The histopathological database of the Department of Veterinary Pathology, Kon-kuk University Animal Teaching Hospital, Seoul, Korea, was analyzed to identify thefive breeds with the highest prevalence of CMTs between 2007 and 2010. The totalnumber of CMT cases selected for this study was 292. All tissues were fixed in 10%neutral buffered formalin, processed routinely, and embedded in paraffin wax. Sec-tions of 4-lm thickness were stained using hematoxylin and eosin (HE).

Histopathology

The tumors were classified as benign (n = 153) or malignant (n = 139), and his-tological types were evaluated in malignant tumors on the basis of the classificationproposed by Goldschmidt et al. (2011), which uses HE staining and immunohisto-chemical (IHC) analysis.

At the time of surgical removal, 46.2% of 292 cases had multiple masses. Caseswith concomitant benign and malignant tumors were considered malignant. Ifthere was more than one malignant mass (15.6% of 292 cases), the most malignanttumor was selected based on histological grade and the existence of lymphaticinvasion (Matos et al., 2012). The mean age of the dogs with malignancies was cal-culated. The histological grade was assigned on the basis of the scheme by Clementeet al. (2010). The density of lymphocyte infiltration within and around tumoralareas was graded on a scale of 0–3: 0, no infiltration; 1, a few and scattered; 2, mod-erate density with several foci; and 3, intensively dense with almost continuouslyextensive infiltration (Black et al., 1956). Tumor necrosis and microscopic evidenceof lymphatic invasion were identified.

Immunohistochemical staining

The tissue specimens that were determined to be malignant (n = 139) wereimmunostained to assess the expression of BRCA1, ER, PR, and HER-2. To confirmthe histological type of the tumor or to determine if it belonged to the basal-likemolecular subtype, staining for cytokeratin 14 (CK14), p63, or smooth muscle actinwas also performed.

IHC analysis was performed using formalin-fixed tissue sections with primaryantibodies (Table 1). The sections were dewaxed in xylene, hydrated through agraded ethanol series, and washed three times in phosphate-buffered saline (PBS;pH 7.4, 0.1 M). Endogenous peroxidase was blocked by incubation of the sectionsin 3% hydrogen peroxide in PBS for 20 min at room temperature (RT). The sectionswere washed three times in PBS. Heat-induced epitope retrieval for primary anti-bodies, except for the anti-HER-2 antibody, was performed using a microwave ovenat high power (650 W). After pretreatment, the serial sections were washed threetimes in PBS, subsequently overlaid with the primary antibody and incubated. Iso-type-matched immunoglobulins were used as negative controls.

Normal mammary glands and normal to hyperplastic lesions adjacent to tumorswere used as positive controls for studying the expression of all antibodies (exceptthat of BRCA1 and HER-2) as well as for comparing BRCA1 expression in mammarymalignant tumors. We used a two-step EnVision (DAKO) method, in which theEnVision rabbit/mouse reagent conjugated to horseradish peroxidase was appliedfor 20 min at RT to visualize the immunolabeling. The slides were subsequentlywashed four times in PBS and incubated with the supplied substrates until the de-

sired color intensity was achieved. The reaction was terminated by washing theslide with distilled water. The sections were counterstained with Gill’s hematoxylin,dehydrated, and a cover slip was applied.

Immunohistochemical scoring

Entire sections were examined to estimate the percentage and intensity of po-sitive cells. Two investigators (KSI and IHK) evaluated and scored the sections in ablind fashion and a third investigator (JHS) re-scored the sample in cases of discrep-ancy between the scorers (BRCA1, n = 10; ER, n = 7; PR, n = 12; HER-2, n = 15; CK14and p63, n = 8). Final scoring was agreed by the majority.

On the basis of the findings of our previous study (Kim et al., 2010), we consid-ered only cytoplasmic and membranous expression of BRCA1 as positive. Immu-nolabeling intensity for BRCA1 was graded on a scale of 0 (no labeling) to 3(strong intensity). The reactivity scores were calculated by summing up the multi-plication products of the intensity grades and each grade’s area. These scores weregraded as 0 (no labeling), 1+ (score 6 100), 2+ (score 6 200), and 3+ (score 6 300).The normal mammary tissues or the tissues adjacent to the tumor were graded as 0or 1+; therefore, we considered similar expression with normal gland and normal tohyperplastic lesions (0 and 1+) as normal expression, and 2+ (weak) and 3+ (strong)as overexpression in malignant tumors.

Nuclear ER and PR immunoreactivity >10% in tumor cells was considered a po-sitive result (Gama et al., 2008). Membrane HER-2 immunolabeling was measuredon four-point scale according to the HercepTest scoring system (Dako): scores of 2+and 3+ indicate HER-2 overexpression. Immunoreactivity for CK14 and p63 was de-fined as positive if >1% of invasive neoplastic cells stained positive for membraneCK14 and nuclear p63 (Kim et al., 2006). The molecular subtype was determinedaccording to the scheme by Gama et al. (2008): luminal A (ER-positive/HER-2 neg-ative), luminal B (ER-positive/HER-2 overexpression), HER-2 overexpressing (ER-negative/HER-2 overexpression), basal-like (ER/HER-2 negative and CK14- or p63-positive) and normal-like (all-negative) subtype.

Statistical analyses

The Statistical Package for Social Sciences (SPSS) software program for Win-dows, v.17 was used for all statistical analyses. Continuous variables were ex-pressed as means ± standard deviation (SD), and categorical data, as frequenciesand percentages. Fisher’s exact test and the Kruskal–Wallis test were applied. Sta-tistical significance was defined as P < 0.05.

Results

Breeds

The five breeds with the highest prevalence of CMT from 2007to 2010 were the Maltese (n = 102), Yorkshire terrier (n = 69), ShihTzu (n = 50), Poodle (n = 43), and mixed breed (n = 28). Malignan-cies were found in 139 of these CMT cases (47.6%; Table 2): Mal-tese (44/102, 43.1%), Yorkshire terrier (33/69, 47.8%), Shih Tzu(29/50, 58%), Poodle (17/43, 39.5%), and mixed breed (16/28,57.1%). The mean age of the dogs with malignancies was9.6 ± 3.0 years (range, 3–17 years), and the mean ages of eachbreed are given in Table 2. The Shih Tzu had the highest malig-nancy rate and a relatively younger mean age, and the mixed breedhad the second highest malignancy rate and a relatively oldermean age (no significant difference).

Table 1Primary antibodies and immunostaining protocols.

Antibody Source Supplier Clone Isotype Dilution Ag retrieval Incubation

BRCA1 M3606 Dakocytomation GLK-2 Mouse IgM, kappa 1:500 Citric acid (pH 6.0; 10 min) 3 h, RTER MU368-UCE BioGenex ER88 Mouse IgG1 1:60 Tris–EDTA (pH 9.0; 20 min) 3 h, RTPR PN IM1546 Immonotech SAS PR10A9 Mouse IgG2a 1:500 Citric acid (pH 6.0; 20 min) 4 "C, overnightHER-2 MU134-UCE BioGenex CB11 Mouse IgG1 1:100 ! 3 h, RTCK14 Ab7800 Abcam LL002 Mouse IgG3 1:300 Tris–EDTA (pH 9.0; 10 min) 3 h, RTp63 sc-8431 Santa Cruz Biotechnology 4A4 Mouse IgG2a 1:100 Tris–EDTA (pH 9.0; 20 min) 4 "C, overnightSMA sc-32251 Santa Cruz Biotechnology 1A4 Mouse IgG2a 1:200 Tris–EDTA (pH 9.0; 10 min) 40 min, RT

ER, estrogen receptor; PR, progesterone receptor; HER-2, human epidermal growth factor receptor-2; CK14, cytokeratin 14; SMA, smooth muscle actin; RT, room temperature.

K.-S. Im et al. / The Veterinary Journal 195 (2013) 366–372 367

Features of canine mammary tumors in different breeds

Table 2 summarizes the features of CMTs in the differentbreeds. No association was found between any breed and histolog-ical classification, necrosis status, and lymphocyte infiltration sta-tus. However, the Maltese and Poodle breeds showed a higherpercentage of carcinoma arising in a complex adenoma/mixed tu-mor (12/44, 27.3% and 7/17, 41.2%, respectively), and carcinoma-complex type (13/44, 29.5% and 4/17, 29.4%, respectively) tumors,which are known to have a better prognosis (Karayannopoulouet al., 2005). The different histological grades (P = 0.038) were sig-nificantly associated with certain breeds. We found that 44.4% (8/18) of the dogs with histological grade 3 belonged to the Shih Tzubreed, whereas none of the Poodles had grade 3 cancers. The pro-portion of dogs with grade 3 cancer was higher in the Shih Tzu (8/29, 27.6%) and mixed breed (4/16, 25%) than in the other breeds. Ofthe 12 cases of lymphatic invasion, 7 (58.3%) occurred in Shih Tzu(Fig. 1A and B), whereas none occurred in the mixed breed. Lym-phatic invasion was not present in >93% of the samples from allthe breeds, except in the Shih Tzu (22/29, 75.9%). Therefore, weconcluded that the rate of lymphatic invasion tends to be higherin Shih Tzus (P = 0.042).

Analysis of BRCA1 expression

Staining for BRCA1 demonstrated weak overexpression in25.9% of all of our dogs, and there was strong overexpression

in 7.9% (Table 3). The association between breed and expressionof BRCA1 was statistically significant (P = 0.003). Of the 29 ShihTzu dogs, 12 (41.4%) showed normal expression and 5 (17.2%)showed strong overexpression, whereas of the 16 mixed-breeddogs, 15 (93.8%) showed normal expression and none showedstrong overexpression. Among the purebred dogs, Poodles hadthe lowest rate of strong BRCA1 overexpression (1/17, 5.9%;P = 0.025).

Relationship of ER, PR, HER-2, triple-negative phenotype, andmolecular subtype with breeds

Of all our dogs, 57.6% were ER positive, 58.3% were PR positive,and 25.9% showed HER-2 overexpression (Table 3). The Poodlegroup had the highest percentage of dogs with positive stainingfor ER (13/17, 76.5%), PR (12/17, 70.6%), and HER-2 overexpression(6/17, 35.3%), whereas the mixed breed had the lowest percentagesof dogs with positive staining for ER (6/16, 37.5%) and HER-2 over-expression (1/16, 6.3%), and the Shih Tzu had the lowest percent-age of dogs that were PR positive (14/29, 48.3%). No significantbreed-related differences were found in ER and PR expression,and in HER-2 overexpression.

The luminal subtype (ER positive) tended to be dominant in allthe breeds (not significant), but the basal-like subtype was alsoprominent in the Shih Tzu and mixed breed (9/29, 31.0% and 6/16, 37.5%, respectively). Among the dogs with strong BRCA1 over-expression, 60% (3/5) of the Shih Tzu, 50% (1/2) of the Yorkshire

Table 2Histopathological features of malignant canine mammary tumors according to breeds (n = 139).

Maltese(n = 44)

Yorkshire terrier(n = 33)

Shih Tzu(n = 29)

Poodle(n = 17)

Mixed breed(n = 16)

P

Age (mean ± SD) 9.1 ± 3.1 10.5 ± 2.6 8.8 ± 3.0 9.6 ± 3.2 10.7 ± 3.0 n.s.

Histological Carcinoma-in situ (n = 1) 0 0 1 (3.4%) 0 0classification Tubular (n = 14) 5 (11.4%) 6 (18.2%) 1 (3.4%) 1 (5.9%) 1 (6.3%)

Tubulopapillary (n = 6) 2 (4.5%) 2 (6.1%) 0 1 (5.9%) 1 (6.3%)Cystic-papillary (n = 6) 2 (4.5%) 2 (6.1%) 0 2 (11.8%) 0Cribriform (n = 2) 0 1 (3.0%) 0 0 1 (6.3%)Carcinoma-micropapillary invasive (n = 1) 1 (2.3%) 0 0 0 0Carcinoma-solid (n = 5) 1 (2.3%) 1 (3.0%) 2 (6.9%) 0 1 (6.3%)Comedocarcinoma (n = 1) 0 0 1 (3.4%) 0 0Carcinoma-anaplastic (n = 2) 0 1 (3.0%) 1 (3.4%) 0 0 n.s.Carcinoma arising in a complex adenoma/mixedtumor (n = 36)

12 (27.3%) 8 (24.2%) 7 (24.1%) 7 (41.2%) 2 (12.5%)

Carcinoma-complex type (n = 33) 13 (29.5%) 6 (18.2%) 5 (17.2%) 5 (29.4%) 4 (25.0%)Carcinoma and malignant myoepithelioma (n = 2) 1 (2.3%) 0 0 0 1 (6.3%)Ductal carcinoma (n = 5) 2 (4.5%) 2 (6.1%) 1 (3.4%) 0 0Intraductal papillary carcinoma (n = 3) 1 (2.3%) 0 1 (3.4%) 0 1 (6.3%)Squamous cell carcinoma (n = 3) 0 1 (3.0%) 2 (6.9%) 0 0Adenosquamous carcinoma (n = 2) 0 1 (3.0%) 1 (3.4%) 0 0Lipid-rich carcinoma (n = 1) 0 0 1 (3.4%) 0 0Inflammatory carcinoma (n = 1) 0 0 1 (3.4%) 0 0Osteosarcoma (n = 7) 3 (6.8%) 1 (3.0%) 2 (6.9%) 0 1 (6.3%)Chondrosarcoma (n = 1) 0 0 1 (3.4%) 0 0Fibrosarcoma (n = 3) 0 0 1 (3.4%) 0 2 (12.5%)Carcinosarcoma (n = 4) 1 (2.3%) 1 (3.0%) 0 1 (5.9%) 1 (6.3%)

Histological grade 1 (n = 82) 31 (70.5%) 20 (60.6%) 14 (48.3%) 11 (64.7%) 6 (37.5%) 0.0382 (n = 39) 11 (25.0%) 9 (27.3%) 7 (24.1%) 6 (35.3%) 6 (37.5%)3 (n = 18) 2 (4.5%) 4 (12.1%) 8 (27.6%) 0 4 (25.0%)

Lymphatic invasion No (n = 127) 42 (95.5%) 31 (93.9%) 22 (75.9%) 16 (94.1%) 16 (100%) 0.042Yes (n = 12) 2 (4.5%) 2 (6.1%) 7 (24.1%) 1 (5.9%) 0

Necrosis No (n = 75) 28 (63.6%) 15 (45.5%) 15 (51.7%) 10 (58.8%) 7 (43.8%) n.s.Yes (n = 64) 16 (36.4%) 18 (54.5%) 14 (48.3%) 7 (41.2%) 9 (56.3%)

Lymphocyte 0 (n = 85) 21 (47.7%) 21 (63.6%) 17 (58.6%) 12 (70.6%) 14 (87.5%) n.s.infiltration 1 (n = 25) 15 (34.1%) 1 (3%) 5 (17.2%) 3 (17.6%) 1 (6.3%)

2 (n = 15) 4 (9.1%) 7 (21.2%) 3 (10.3%) 1 (5.9%) 03 (n = 14) 4 (9.1%) 4 (12.1%) 4 (13.8%) 1 (5.9%) 1 (6.3%)

SD, standard deviation; HPF, high-power field; n.s., no significance; statistical significance, P < 0.05.

368 K.-S. Im et al. / The Veterinary Journal 195 (2013) 366–372

terrier, 33.3% (1/3) of the Maltese, and none of the Poodles andmixed-breed dogs showed the basal-like subtype. Moreover, a tri-ple-negative phenotype (ER, PR, and HER-2 negative) with strongBRCA1 overexpression was observed only in the Shih Tzu (3/5,60%) and Yorkshire terrier (1/2, 50%) breeds.

Relationship of ER, PR, HER-2, triple-negative phenotype, andmolecular subtype with BRCA1

ER and BRCA1 expression were found to be inversely related,since dogs with normal BRCA1 expression had a greater tendency

Fig. 1. Serial sections of canine mammary carcinoma in the Shih Tzu. (A) Mammary inflammatory carcinoma. Mitoses (arrows). Hematoxylin and eosin (HE), scale bar, 35 lm.(B) Lymphatic invasion of mammary inflammatory carcinoma. HE, scale bar, 35 lm. (C) Strong overexpression of BRCA1. Immunohistochemistry (IHC), scale bar, 70 lm. (D)Estrogen receptor negative expression. IHC, scale bar, 35 lm. (E) Progesterone receptor negative expression. IHC, scale bar, 35 lm. (F) Negative for human epidermal growthfactor receptor-2 overexpression. IHC, scale bar, 35 lm. (G) Cytokeratin 14 positive expression. IHC, scale bar, 70 lm. (H) p63-negative expression. IHC, scale bar, 35 lm.

K.-S. Im et al. / The Veterinary Journal 195 (2013) 366–372 369

to show ER positive expression (52/92, 56.5%; Table 4). Similarly,dogs that showed strong BRCA1 overexpression tended to be ERnegative (8/11, 72.7%; P = 0.047). PR expression also showed an in-verse relationship with the expression of BRCA1, but this relation-ship was not significant. HER-2 overexpression was negative in ahigh proportion of dogs with normal BRCA1 expression (77/92,83.7%), whereas HER-2 overexpression was increased in dogs withstrong BRCA1 overexpression (4/11, 36.4%; P = 0.001). There wasno significant relationship between the expression of BRCA1 andthe triple-negative phenotype. However, in the dogs with normalBRCA1 expression, the luminal A subtype was the most dominant(45/92, 48.9%), and in those with strong BRCA1 overexpression,the basal-like subtype was dominant (5/11, 45.5%; P = 0.003).

Relationship of ER, PR, HER-2, triple-negative phenotype andmolecular subtype with BRCA1 in Shih Tzu dogs

The proportion of Shih Tzu dogs with ER (51.7%) and PR (48.3%)positive expression, and HER-2 overexpression (17.2%) was lower

than the corresponding proportion in the other breeds. Shih Tzudogs that showed strong BRCA1 overexpression (Fig. 1C) weremore likely to show ER (100%; P = 0.004; Fig. 1D) and PR (100%;P = 0.046; Table 5 and Fig. 1E) negative expression, but HER-2 over-expression was not significant (Fig. 1F). Although, in Shih Tzu dogs,only 5 cases showed strong BRCA1 overexpression, the proportionof those with triple-negative phenotype (3/5, 60%; P = 0.016;Fig. 1D–F) and basal-like subtype (3/5, 60%; P = 0.019; Fig. 1F andH) were higher than those with the other molecular subtypes.

Discussion

The loss of BRCA1 function can occur as a result of frame shift,non-sense, and mis-sense mutations, chromosomal rearrange-ments, splice-site alterations, loss of heterozygosity (LOH) of thewild-type allele, or altered location of wild-type BRCA1 proteinwith abnormality of molecules for transport. This loss of functioncan lead to biological instability that promotes abnormal cell cyclearrest, apoptosis, and tumorigenesis (Lee et al., 1999; Szabo et al.,

Table 3BRCA1, estrogen receptor (ER) and progesterone receptor (PR) expression, human epidermal growth factor receptor-2 (HER-2) overexpression, triple-negative phenotype, andmolecular subtype according to breeds.

Maltese(n = 44)

Yorkshire terrier(n = 33)

Shih Tzu(n = 29)

Poodle(n = 17)

Mixed breed(n = 16)

P

BRCA1 expression Normal expression (n = 92) 32 (72.7%) 20 (60.6%) 12 (41.4%) 13 (76.5%) 15 (93.8%) 0.003Weak overexpression (n = 36) 9 (20.5%) 11 (33.3%) 12 (41.4%) 3 (17.6%) 1 (6.3%)Strong overexpression(n = 11)

3 (6.8%) 2 (6.1%) 5 (17.2%) 1 (5.9%) 0

ER expression Negative (n = 59) 15 (34.1%) 16 (48.5%) 14 (48.3%) 4 (23.5%) 10 (62.5%) n.s.Positive (n = 80) 29 (65.9%) 17 (51.5%) 15 (51.7%) 13 (76.5%) 6 (37.5%)

PR expression Negative (n = 58) 15 (34.1%) 16 (48.5%) 15 (51.7%) 5 (29.4%) 7 (43.8%) n.s.Positive (n = 81) 29 (65.9%) 17 (51.5%) 14 (48.3%) 12 (70.6%) 9 (56.3%)

HER-2overexpression

Negative (n = 103) 31 (70.5%) 22 (66.7%) 24 (82.8%) 11 (64.7%) 15 (93.8%) n.s.Positive (n = 36) 13 (29.5%) 11 (33.3%) 5 (17.2%) 6 (35.3%) 1 (6.3%)

Triple-negative No (n = 113) 39 (88.6%) 27 (81.8%) 23 (79.3%) 14 (82.4%) 10 (62.4%) n.s.Yes (n = 26) 5 (11.4%) 6 (18.2%) 6 (20.7%) 3 (17.6%) 6 (37.5%)

Molecular subtype Luminal A (n = 60) 21 (47.7%) 11 (33.3%) 14 (48.3%) 8 (47.1%) 6 (37.5%) n.s.Luminal B (n = 20) 9 (18.2%) 6 (18.2%) 1 (3.4%) 5 (29.4%) 0HER-2 overexpressing(n = 16)

5 (11.4%) 5 (15.2%) 4 (13.8%) 1 (5.9%) 1 (6.3%)

Basal-like (n = 34) 8 (18.2%) 8 (24.2%) 9 (31.0%) 3 (17.6%) 6 (37.5%)Normal-like (n = 9) 2 (4.5%) 3 (6.8%) 1 (3.4%) 0 3 (18.8%)

Triple-negative phenotype, ER-/PR-/HER-2!; luminal A, ER+/HER-2!; luminal B, ER+/HER-2+; HER-2 overexpressing, ER-/HER-2+; basal-like, ER-/HER-2!, and CK14 or p63+;normal-like, all-; n.s., no significance; statistical significance, P < 0.05.

Table 4Relationship between estrogen receptor (ER) and progesterone receptor (PR) expression, human epidermal growth factor receptor-2 (HER-2) overexpression, triple-negativephenotype, and molecular subtype and BRCA1 expression.

BRCA1 (n = 139) P

Normal expression (n = 92) Weak overexpression (n = 36) Strong overexpression (n = 11)

ER expression Negative (n = 59) 40 (43.5%) 11 (30.6%) 8 (72.7%) 0.047Positive (n = 80) 52 (56.5%) 25 (69.4%) 3 (27.3%)

PR expression Negative (n = 58) 37 (40.2%) 14 (38.9%) 7 (63.6%) n.s.Positive (n = 81) 55 (59.8%) 22 (61.1%) 4 (36.4%)

HER-2 overexpression Negative (n = 103) 77 (83.7%) 19 (52.8%) 7 (63.6%) 0.001Positive (n = 36) 15 (16.3%) 17 (47.2%) 4 (36.4%)

Triple-negative No (n = 113) 75 (81.5%) 31 (86.1%) 7 (63.6%) n.s.Yes (n = 26) 17 (18.5%) 5 (13.9%) 4 (36.4%)

Molecular subtype Luminal A (n = 60) 45 (48.9%) 13 (36.1%) 2 (18.2%) 0.003Luminal B (n = 20) 7 (7.6%) 12 (33.3%) 1 (9.1%)HER-2 overexpressing (n = 16) 8 (8.7%) 5 (13.9%) 3 (27.3%)Basal-like (n = 34) 24 (26.1%) 5 (13.9%) 5 (45.5%)Normal-like (n = 9) 8 (8.7%) 1 (2.8%) 0

Triple-negative phenotype, ER-/PR-/HER-2!; luminal A, ER+/HER-2!; luminal B, ER+/HER-2+; HER-2 overexpressing, ER-/HER-2+; basal-like, ER-/HER-2!, and CK14 or p63+;normal-like, all-; n.s., no significance; statistical significance, P < 0.05.

370 K.-S. Im et al. / The Veterinary Journal 195 (2013) 366–372

2004; Deng, 2006; Troudi et al., 2007). Most studies indicate thatBRCA1 is a nuclear protein, and loss of BRCA1 function could leadto abnormal cytoplasmic distribution of BRCA1 in HBCs (Thakuret al., 1997; Troudi et al., 2007). A similar finding was observedin CMTs in our previous study (Kim et al., 2010), as well as in thestudy by Nieto et al. (2003).

CMTs occur mainly in breeds such as the English Springer Span-iel, German shepherd, Poodle, Maltese, and Beagle (Sorenmo et al.,2011). However, breed predispositions vary according to regionaldifferences. Furthermore, familial CMT within the same breedhas been observed in Beagles (Schafer et al., 1998), and hereditaryrisk factors such as BRCA1 and BRCA2 mutations were confirmed inEnglish Springer Spaniels (Rivera et al., 2009).

Since it is difficult to track the family history of CMT in dogs, weanalyzed the characteristics of five breeds with a high prevalencefor CMT according to various parameters: clinicohistopathologicalfeatures; BRCA1, ER, and PR expression; HER-2 overexpression; tri-ple-negative phenotype, and molecular subtypes. Compared toother breeds, the Shih Tzu had a higher proportion of dogs withmalignancies (58%; not significant), younger age (8.8 ± 3.0 years;not significant), a histological grade of 3 (27.6%; P = 0.038), andextensive lymphatic invasion (24.1%; P = 0.042). Furthermore, theproportion of dogs showing strong BRCA1 overexpression was17.2%, which was threefold higher than that observed in the otherbreeds (P = 0.00). These differences suggest that both tumor fea-tures with prognostic values and the altered expression of BRCA1are different among dog breeds.

ER and PR expression, HER-2 overexpression, triple-negativephenotype, and molecular subtype were not associated withbreeds, but ER expression, HER-2 overexpression, and molecularsubtype were significantly associated with the expression ofBRCA1 (P = 0.047, 0.001, and 0.003, respectively). Furthermore,Shih Tzu dogs with strong BRCA1 overexpression tended to showER (P = 0.004) and PR (P = 0.046) negative expression, a triple-neg-ative phenotype (P = 0.016), and basal-like subtype (P = 0.019), asin BRCA1-related HBCs (Lakhani et al., 2002). However, not all ofthese factors were related to BRCA1 in other breeds (data notshown).

Numerous studies (Lakhani et al., 2002; Musolino et al., 2007)have shown that 60–90% of BRCA1-related HBCs are negative forER and that 50–80% are negative for PR. Consistent with these find-ings, our study showed that 100% of the Shih Tzu dogs with strongBRCA1 overexpression showed ER and PR negative expression.HER-2 overexpression is very rare in BRCA1-related HBCs because

BRCA1 and HER-2 are located close to each other on chromosome17. Therefore, loss of HER-2 may occur along with LOH at theBRCA1 locus (Grushko et al., 2002). However, in our study, HER-2overexpression tended to increase with the expression of BRCA1(P = 0.001) in CMTs, and was not significant in the Shih Tzu breed.In dogs, BRCA1 is located on chromosome 9 (Rivera et al., 2009) andHER-2 on chromosome 1 (Ordás et al., 2007). The distance betweenthe BRCA1 and HER-2 loci in dogs may explain why HER-2 overex-pression is different between BRCA1-related HBCs and CMTs. CMTswith strong BRCA1 overexpression have a tendency to show thetriple-negative phenotype (ER, PR, and HER-2 negative) or basal-like subtype (negative for ER and HER-2, and positive for any basalmarker), just as BRCA1-related HBC is associated with the triple-negative phenotype or basal-like subtype (Sørlie et al., 2001; Leeet al., 2011).

Conclusions

This study showed that Shih Tzu dogs had a higher percentageof altered strong overexpression of BRCA1, which related to ER andPR negative expression, triple-negative phenotype, and basal-likesubtype as in BRCA1-related HBCs. Therefore, dogs with CMT, par-ticularly the Shih Tzu breed, may provide a good spontaneousmodel for studying BRCA1-related HBCs. In the future, additionalmolecular studies focusing on specific breeds are warranted.

Conflict of interest statement

None of the authors of this paper has a financial or personalrelationship with other people or organisations that could inappro-priately influence or bias the content of the paper.

Acknowledgements

The authors thank Ms. Rae-Hwa Jang for her excellent technicalassistance and the participating private veterinary clinics for pro-viding access to tissue samples of canine mammary tumors. Thisreport represents a part of a PhD Thesis by Il-Hwan Kim. The re-search was supported by the Basic Science Research Program ofthe National Research Foundation of Korea (NRF) and was fundedby the Ministry of Education, Science and Technology (2011-0021337).

Table 5Relationship of estrogen receptor (ER) and progesterone receptor (PR) expression, human epidermal growth factor receptor-2 (HER-2) overexpression, triple-negative phenotype,and molecular subtype with BRCA1 in the Shih Tzu.

BRCA1 (n = 29) P

Normal expression (n = 12) Weak overexpression (n = 12) Strong overexpression (n = 5)

ER expression Negative (n = 14) 7 (58.3%) 2 (16.7%) 5 (100%) 0.004Positive (n = 15) 5 (41.7%) 10 (83.3%) 0

PR expression Negative (n = 15) 6 (50.0%) 4 (33.3%) 5 (100%) 0.046Positive (n = 14) 6 (50.0%) 8 (66.7%) 0

HER-2 overexpression Negative (n = 24) 11 (91.7%) 10 (83.3%) 3 (60.0%) n.s.Positive (n = 5) 1 (8.3%) 2 (16.7%) 2 (40.0%)

Triple-negative No (n = 23) 9 (75.0%) 12 (100%) 2 (40.0%) 0.016Yes (n = 6) 3 (25.0%) 0 3 (60.0%)

Molecular subtype Luminal A (n = 14) 5 (41.7%) 9 (75.0%) 0 0.019Luminal B (n = 1) 0 1 (8.3%) 0HER-2 overexpressing (n = 4) 1 (8.3%) 1 (8.3%) 2 (40.0%)Basal-like (n = 9) 5 (41.7%) 1 (8.3%) 3 (60.0%)Normal-like (n = 1) 1 (8.3%) 0 0

Triple-negative phenotype, ER-/PR-/HER-2!; luminal A, ER+/HER-2!; luminal B, ER+/HER-2+; HER-2 overexpressing, ER-/HER-2+; basal-like, ER-/HER-2!, and CK14 or p63+;normal-like, all-; n.s., no significance; statistical significance, P < 0.05.

K.-S. Im et al. / The Veterinary Journal 195 (2013) 366–372 371

References

Ahn, S.H., Son, B.H., Yoon, K.S., Noh, D.Y., Han, W., Kim, S.W., Lee, E.S., Park, H.L.,Hong, Y.J., Choi, J.J., Moon, S.Y., Kim, M.J., Kim, K.H., Kwak, B.S., Cho, D.Y., 2007.BRCA1 and BRCA2 germline mutations in Korean breast cancer patients at highrisk of carrying mutations. Cancer Letters 245, 90–95.

Black, M.M., Speer, F.D., Opler, S.R., 1956. Structural representations of tumor–hostrelationships in mammary carcinoma; biologic and prognostic significance.American Journal of Clinical Pathology 26, 250–265.

Clemente, M., Pérez-Alenza, M.D., Illera, J.C., Peña, L., 2010. Histological,immunohistological, and ultrastructural description of vasculogenic mimicryin canine mammary cancer. Veterinary Pathology 47, 265–274.

Da Silva, L., Lakhani, S.R., 2010. Pathology of hereditary breast cancer. ModernPathology 23, S46–S51.

Deng, C.X., 2006. BRCA1: Cell cycle checkpoint, genetic instability, DNA damageresponse and cancer evolution. Nucleic Acids Research 34, 1416–1426.

Fitzgibbons, P.L., Page, D.L., Weaver, D., Thor, A.D., Allred, D.C., Clark, G.M., Ruby,S.G., O’Malley, F., Simpson, J.F., Connolly, J.L., Hayes, D.F., Edge, S.B., Lichter, A.,Schnitt, S.J., 2000. Prognostic factors in breast cancer. College of AmericanPathologists Consensus Statement 1999. Archives of Pathology and LaboratoryMedicine 124, 966–978.

Gama, A., Alves, A., Schmitt, F., 2008. Identification of molecular phenotypes incanine mammary carcinomas with clinical implications: Application of thehuman classification. Virchows Archives 453, 123–132.

Gao, Q., Neuhausen, S., Cummings, S., Luce, M., Olopade, O.I., 1997. Recurrent germ-line BRCA1 mutations in extended African American families with early-onsetbreast cancer. American Journal of Human Genetics 60, 1233–1236.

Goldschmidt, M., Peña, L., Rasotto, R., Zappulli, V., 2011. Classification and gradingof canine mammary tumors. Veterinary Pathology 48, 117–131.

Grushko, T.A., Blackwood, M.A., Schumm, P.L., Hagos, F.G., Adeyanju, M.O., Feldman,M.D., Sanders, M.O., Weber, B.L., Olopade, O.I., 2002. Molecular-cytogeneticanalysis of HER-2/neu gene in BRCA1-associated breast cancers. CancerResearch 62, 1481–1488.

Harris, L., Fritsche, H., Mennel, R., Norton, L., Ravdin, P., Taube, S., Somerfield, M.R.,Hayes, D.F., Bast Jr., R.C., 2007. American Society of Clinical Oncology 2007update of recommendations for the use of tumor markers in breast cancer.Journal of Clinical Oncology 25, 5287–5312.

Hellmén, E., Bergström, R., Holmberg, L., Spångberg, I.B., Hansson, K., Lindgren, A.,1993. Prognostic factors in canine mammary tumors: A multivariate study of202 consecutive cases. Veterinary Pathology 30, 20–27.

Johannsson, O., Ostermeyer, E.A., Håkansson, S., Friedman, L.S., Johansson, U.,Sellberg, G., Brøndum-Nielsen, K., Sele, V., Olsson, H., King, M.C., Borg, A., 1996.Founding BRCA1 mutations in hereditary breast and ovarian cancer in southernSweden. American Journal of Human Genetics 58, 441–450.

Karayannopoulou, M., Kaldrymidou, E., Constantinidis, T.C., Dessiris, A., 2005.Histological grading and prognosis in dogs with mammary carcinomas:Application of a human grading method. Journal of Comparative Pathology133, 246–252.

Kim, J.H., Yu, C.H., Yhee, J.Y., Im, K.S., Sur, J.H., 2010. Lymphocyte infiltration,expression of interleukin (IL)-1, IL-6 and expression of mutated breast cancersusceptibility gene-1 correlate with malignancy of canine mammary tumours.Journal of Comparative Pathology 142, 177–186.

Kim, M.J., Ro, J.Y., Ahn, S.H., Kim, H.H., Kim, S.B., Gong, G., 2006. Clinicopathologicsignificance of the basal-like subtype of breast cancer: A comparison withhormone receptor and Her2/neu-overexpressing phenotypes. Human Pathology37, 1217–1226.

Klopfleisch, R., Gruber, A.D., 2009. Increased expression of BRCA2 and RAD51 inlymph node metastases of canine mammary adenocarcinomas. VeterinaryPathology 46, 416–422.

Klopfleisch, R., von Euler, H., Sarli, G., Pinho, S.S., Gärtner, F., Gruber, A.D., 2011.Molecular carcinogenesis of canine mammary tumors: News from an olddisease. Veterinary Pathology 48, 98–116.

Lagos-Jaramillo, V.I., Press, M.F., Ricker, C.N., Dubeau, L., Mai, P.L., Weitzel, J.N., 2011.Pathological characteristics of BRCA-associated breast cancers in Hispanics.Breast Cancer Research and Treatment 130, 281–289.

Lakhani, S.R., Van De Vijver, M.J., Jacquemier, J., Anderson, T.J., Osin, P.P., McGuffog,L., Easton, D.F., 2002. The pathology of familial breast cancer: Predictive value ofimmunohistochemical markers estrogen receptor, progesterone receptor, HER-

2, and p53 in patients with mutations in BRCA1 and BRCA2. Journal of ClinicalOncology 20, 2310–2318.

Lee, E., McKean-Cowdin, R., Ma, H., Spicer, D.V., Van Den Berg, D., Bernstein, L.,Ursin, G., 2011. Characteristics of triple-negative breast cancer in patients witha BRCA1 mutation: Results from a population-based study of young women.Journal of Clinical Oncology 29, 4373–4380.

Lee, W.-Y., Jin, Y.-T., Chang, T.-W., Lin, P.-W., Su, I.-J., 1999. Immunolocalization ofBRCA1 protein in normal breast tissue and sporadic invasive ductal carcinomas:A correlation with other biological parameters. Histopathology 34, 106–112.

Matos, A.J.F., Baptista, C.S., Gärtner, M.F., Rutteman, G.R., 2012. Prognostic studies ofcanine and feline mammary tumours: The need for standardized procedures.The Veterinary Journal 193, 24–31.

Morrison, W.B., 2002. Canine and feline mammary tumors. In: Cancer in Dog andCats: Medical and Surgical Management. Teton New Media, Jackson, WY, USA,p. 565.

Musolino, A., Bella, M.A., Bortesi, B., Michiara, M., Naldi, N., Zanelli, P., Capelletti, M.,Pezzuolo, D., Camisa, R., Savi, M., Neri, T.M., Ardizzoni, A., 2007. BRCAmutations, molecular markers, and clinical variables in early-onset breastcancer: A population-based study. Breast 16, 280–292.

Nieto, A., Pérez-Alenza, M.D., Del Castillo, N., Tabanera, E., Castaño, M., Peña, L.,2003. BRCA1 expression in canine mammary dysplasias and tumours:Relationship with prognostic variables. Journal of Comparative Pathology 128,260–268.

Offit, K., Gilewski, T., McGuire, P., Schluger, A., Hampel, H., Brown, K., Swensen, J.,Neuhausen, S., Skolnick, M., Norton, L., Goldgar, D., 1996. Germline BRCA1185delAG mutations in Jewish women with breast cancer. The Lancet 347,1643–1645.

Ordás, J., Millán, Y., Dios, R., Reymundo, C., de Las Mulas, J.M., 2007. Proto-oncogeneHER-2 in normal, dysplastic and tumorous feline mammary glands: Animmunohistochemical and chromogenic in situ hybridization study. BMCCancer 7, 179.

Rivera, P., Melin, M., Biagi, T., Fall, T., Häggström, J., Lindblad-Toh, K., von Euler, H.,2009. Mammary tumor development in dogs is associated with BRCA1 andBRCA2. Cancer Research 69, 8770–8774.

Rivera, P., von Euler, H., 2011. Molecular biological aspects on canine and humanmammary tumors. Veterinary Pathology 48, 132–146.

Schafer, K.A., Kelly, G., Schrader, R., Griffith, W.C., Muggenburg, B.A., Tierney, L.A.,Lechner, J.F., Janovitz, E.B., Hahn, F.F., 1998. A canine model of familialmammary gland neoplasia. Veterinary Pathology 35, 168–177.

Sorenmo, K.U., Rasotto, R., Zappulli, V., Goldschmidt, M.H., 2011. Development,anatomy, histology, lymphatic drainage, clinical features, and celldifferentiation markers of canine mammary gland neoplasms. VeterinaryPathology 48, 85–97.

Sørlie, T., Perou, C.M., Tibshirani, R., Aas, T., Geisler, S., Johnsen, H., Hastie, T., Eisen,M.B., van de Rijn, M., Jeffrey, S.S., Thorsen, T., Quist, H., Matese, J.C., Brown, P.O.,Botstein, D., Eystein Lønning, P., Børresen-Dale, A.L., 2001. Gene expressionpatterns of breast carcinomas distinguish tumor subclasses with clinicalimplications. Proceedings of the National Academy of Sciences of the UnitedStates of America 98, 10869–10874.

Szabo, C.I., Worley, T., Monteiro, A.N., 2004. Understanding germ-line mutations inBRCA1. Cancer Biology and Therapy 3, 515–520.

Thakur, S., Zhang, H.B., Peng, Y., Le, H., Carroll, B., Ward, T., Yao, J., Farid, L.M., Couch,F.J., Wilson, R.B., Weber, B.L., 1997. Localization of BRCA1 and a splice variantidentifies the nuclear localization signal. Molecular and Cellular Biology 17,444–452.

Troudi, W., Uhrhammer, N., Ben Romdhane, K., Sibille, C., Mahfoudh, W., Chouchane,L., Ayed, F.B., Bignon, Y.-J., Ben Ammar Elgaaied, A., 2007. Immunolocalization ofBRCA1 protein in tumor breast tissue: Prescreening of BRCA1 mutation inTunisian patients with gereditary breast cancer? European Journal ofHistochemistry 51, 219–226.

van der Groep, P., van der Wall, E., van Diest, P.J., 2011. Pathology of hereditarybreast cancer. Cell Oncology (Dordr) 34, 71–88.

Vargas, A.C., Reis-Filho, J.S., Lakhani, S.R., 2011. Phenotype–genotype relationship infamilial breast cancer. Journal of Mammary Gland Biology and Neoplasia 16,27–40.

Yassaee, V.R., Zeinali, S., Harirchi, I., Jarvandi, S., Mohagheghi, M.A., Hornby, D.P.,Dalton, A., 2002. Novel mutations in the BRCA1 and BRCA2 genes in Iranianwomen with early-onset breast cancer. Breast Cancer Research 4, R6.

372 K.-S. Im et al. / The Veterinary Journal 195 (2013) 366–372