expression of cox-1 and cox-2 in canine mammary tumours

of 9 /9
J. Comp. Path. 2007, Vol. 136, 177^185 Expression of Cox-1 and Cox-2 in Canine Mammary Tumours F. L. Queiroga, A. Alves, I. Pires and C. Lopes * Department ofVeterinary Sciences, University ofTra Ł s-os-Montes and Alto Douro, 5000-911,Vila Real, and * Institute of Biomedical Sciences Abel Salazar (ICBAS), Largo Prof. Abel Salazar, 4099-003 Porto, Portugal Summary The aim of this study was to investigate immunohistochemically the expression of cyclooxygenase-1 (Cox-1) and cyclooxygenase-2 (Cox-2) in canine mammary tumours of di¡erent histological types. Cox-1 and Cox-2 enzyme expression was evaluated in 70 mammary samples (four normal, six hyperplastic, 60 neoplastic [21benign and 39 malignant]). Cox-1 expression was identi¢ed in all the samples, and Cox-2 in all the mammary lesions except ductal hyperplasia.Two of the four normal mammary gland samples showed focal immunoreactivity for Cox-2. Cox-1 immunoexpression did not di¡er signi¢cantly between benign and malignant lesions (P ¼ 0.272). Cox-2 immunoexpression was higher in malignant tumours than in benign counterparts (Po0.001). Of the malignant tumours, carcinosarcomas and tubulopapillary and squamous cell carcinomas had the highest Cox-2 scores.The study showed that malignant tumours had the highest values of Cox-2 expression, and Cox-2 immunolabelling was particularly intense in histological types classically associated with high malignancy.This suggests that non- steroidal anti-in£ammatory drugs (NSAIDs), particularly Cox-2 inhibitors, may have a useful role to play in the treatment of canine malignant mammary tumours. r 2007 Elsevier Ltd. All rights reserved. Keywords: Cox-1 and -2; cyclooxygenase; dog; mammary tumours; tumours Introduction Mammary tumours are the most common neoplasms in female dogs (Misdorp, 2002 ). Studies on morpholo- gical characterization and prognosis include those of Yamagami et al. (1996) , Pena et al. (1998) , Sarli et al. (2002) and Lee et al. (2004) , but studies on the biochem- ical characterization of benign and malignant lesions are needed to provide a ¢rmer basis for therapy. The cyclooxygenase (COX) enzyme, also known as the prostaglandin H-synthase, catalyses the rate-limit- ing step in prostanoid biosynthesis (Hwang et al.,1998 ). Two isoforms of Cox have been identi¢ed (Cox-1 and Cox-2). The two isoenzymes are similar in protein structure but are produced by di¡erent genes and have di¡erent biological functions (Williams and DuBois, 1996 ; Hla et al.,1999; Wilson et al., 2004 ). Cox-1 is consti- tutively expressed in many tissues and plays an impor- tant role in the regulation of normal physiological functions such as cytoprotection of the stomach, control of platelet aggregation, renal function and reproduc- tion (Dubois et al.,1998 ). Cox-2 is usually absent in nor- mal cells but can be induced by growth factors, in£ammatory reactions, tumour promoters and onco- genes (Howe et al., 2001 ). Much evidence suggests that Cox-2 plays an impor- tant role in mammary tumour initiation and progres- sion in human patients (Soslow et al., 2000 , Davies et al., 2002 ; Shim et al., 2003 ) and rodents (Bauer et al., 2000; Liu et al., 2001 ). In the dog, Cox-1 and Cox-2 ‘‘up-regu- lation’’ has been investigated in di¡erent neoplastic tis- sues (bladder, skin, intestinal, mammary and nasal) (Khan et al., 2000 , Pestili de Almeida et al., 2001 ; McEn- tee et al., 2002 ; Dore Ł et al., 2003 ; Kleiter et al., 2004 ). In canine mammary tumours, Cox-1 expression was re- cently identi¢ed immunohistochemically in a small series of mammary carcinomas (n ¼ 11) (Mohammed et al., 2004 ). Likewise, Cox-2 up-regulation has been www.elsevier.com/locate/jcpa ARTICLE IN PRESS 0021-9975/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi: 10.1016/j.jcpa.2007.01.010 Correspondence to: F.L. Queiroga (e-mail: [email protected]).

Upload: utad

Post on 25-Apr-2023

2 views

Category:

Documents


0 download

TRANSCRIPT

ARTICLE IN PRESS

J. Comp. Path. 2007,Vol.136,177^185

0021-9975/$ - see fdoi:10.1016/j.jcpa.

Correspondence

www.elsevier.com/locate/jcpa

Expression of Cox-1 and Cox-2 inCanine Mammary Tumours

F. L. Queiroga, A. Alves, I. Pires and C. Lopes*

Department ofVeterinary Sciences, University ofTraŁ s-os-Montes and Alto Douro, 5000-911,Vila Real, and*Institute of Biomedical Sciences Abel Salazar (ICBAS), Largo Prof. Abel Salazar, 4099-003 Porto, Portugal

Summary

The aim of this study was to investigate immunohistochemically the expression of cyclooxygenase-1 (Cox-1) andcyclooxygenase-2 (Cox-2) in canine mammary tumours of di¡erent histological types. Cox-1and Cox-2 enzymeexpressionwas evaluated in 70 mammary samples (four normal, six hyperplastic, 60 neoplastic [21benign and 39malignant]). Cox-1 expression was identi¢ed in all the samples, and Cox-2 in all the mammary lesions exceptductal hyperplasia.Two of the four normal mammary gland samples showed focal immunoreactivity for Cox-2.Cox-1 immunoexpression did not di¡er signi¢cantly between benign and malignant lesions (P ¼ 0.272). Cox-2immunoexpression was higher in malignant tumours than in benign counterparts (Po0.001). Of the malignanttumours, carcinosarcomas and tubulopapillary and squamous cell carcinomas had the highest Cox-2 scores.Thestudy showed that malignant tumours had the highest values of Cox-2 expression, and Cox-2 immunolabellingwas particularly intense in histological types classically associated with high malignancy.This suggests that non-steroidal anti-in£ammatory drugs (NSAIDs), particularly Cox-2 inhibitors, may have a useful role to play in thetreatment of canine malignant mammary tumours.

r 2007 Elsevier Ltd. All rights reserved.

Keywords:Cox-1and -2; cyclooxygenase; dog; mammary tumours; tumours

Introduction

Mammary tumours are the most common neoplasmsin female dogs (Misdorp, 2002). Studies on morpholo-gical characterization and prognosis include those ofYamagami et al. (1996), Pena et al. (1998), Sarli et al.(2002) and Lee et al. (2004), but studies on the biochem-ical characterization of benign and malignant lesionsare needed to provide a ¢rmer basis for therapy.The cyclooxygenase (COX) enzyme, also known as

the prostaglandin H-synthase, catalyses the rate-limit-ing step in prostanoid biosynthesis (Hwang et al.,1998).Two isoforms of Cox have been identi¢ed (Cox-1 andCox-2). The two isoenzymes are similar in proteinstructure but are produced by di¡erent genes and havedi¡erent biological functions (Williams and DuBois,1996; Hla et al.,1999;Wilson et al., 2004). Cox-1is consti-tutively expressed in many tissues and plays an impor-

ront matter2007.01.010

to: F.L. Queiroga (e-mail: [email protected]).

tant role in the regulation of normal physiologicalfunctions suchas cytoprotection of the stomach, controlof platelet aggregation, renal function and reproduc-tion (Dubois et al.,1998). Cox-2 is usually absent in nor-mal cells but can be induced by growth factors,in£ammatory reactions, tumour promoters and onco-genes (Howe et al., 2001).Much evidence suggests that Cox-2 plays an impor-

tant role in mammary tumour initiation and progres-sion in humanpatients (Soslow et al., 2000, Davies et al.,2002; Shim et al., 2003) and rodents (Bauer et al., 2000;Liu et al., 2001). In the dog, Cox-1 and Cox-2 ‘‘up-regu-lation’’ has been investigated in di¡erent neoplastic tis-sues (bladder, skin, intestinal, mammary and nasal)(Khan et al., 2000, Pestili deAlmeida et al., 2001;McEn-tee et al., 2002; DoreŁ et al., 2003; Kleiter et al., 2004). Incanine mammary tumours, Cox-1 expression was re-cently identi¢ed immunohistochemically in a smallseries of mammary carcinomas (n ¼ 11) (Mohammedet al., 2004). Likewise, Cox-2 up-regulation has been

r 2007 Elsevier Ltd. All rights reserved.

ARTICLE IN PRESS

F.L. Queiroga et al.178

studied in canine mammary adenocarcinomas (DoreŁet al., 2003; Mohammed et al., 2004) and adenomas(DoreŁ et al., 2003).Canine mammary tumours vary considerably both

pathologically and clinically (Perez-Alenza et al.,2000).The purpose of the present study was to examineCox-1 and -2 expressions in di¡erent histological typesof these neoplasms.

Materials and Methods

Histopathology

The expression of Cox-1and Cox-2 was evaluated in 70canine mammary samples obtained from the histo-pathological archives of theUniversity ofTraŁ s-os-Mon-tes and Alto Douro (UTAD),Vila Real, Portugal.Thesamples for histopathology were ¢xed in 10% bu¡eredformalin, embedded in para⁄n wax and sectioned at3 mm.Mammary tumourswere diagnosedby two indepen-

dent pathologists according to theWHO classi¢cationsystem of canine mammary tumours (Misdorp et al.,1999).

Table 1Caninemammary samples (n ¼ 70) examined

Histological diagnosis� Number of samples

Normal mammary gland 4Mammary hyperplasia 6

Benign tumoursSimple adenoma 5Complex adenoma 8Papillary adenoma 3Benign mixed tumour 5

Malignant tumoursIn-situ carcinoma 2Complex carcinoma 9Squamous cell carcinoma 2Solid carcinoma 3Tubulopapillary carcinoma 14Tubular carcinoma 6Carcinosarcoma 3

�WHOtumour classi¢cation (Misdorp et al.,1999).

Immunohistochemistry (IHC)

This was performed with Cox-1 and Cox-2 antibodieswhich, in previous studies (Khan et al., 2000; McEnteeet al., 2002) had shown reactivity with canine species.Dewaxed sections were examined by the streptavidin-biotin-peroxidase complex method. For Cox-1, no pre-treatment was used. For Cox-2, however, antigen wasretrieved by microwave treatment for 20min at 750Win a solution of dextran 0.05% in distilled water, fol-lowed by cooling for 20min at room temperature.To block endogenous peroxidases, the slides were im-

mersed for 30min in 0.3%H2O2. Non-speci¢c stainingwas eliminated by incubation for 30min with normalpig serum (for Cox-1) or normal rabbit serum (forCox-2). Excess serum was removed and the sectionswere incubated with primary antibody overnight at4 1C. Antisera speci¢c for Cox-1 (Cayman ChemicalCo, Ann Arbor, MI, USA; cat. no. 160108, polyclonal)and Cox-2 (Transduction Laboratories, Lexington,KY, USA; clone 33, monoclonal) were diluted to 1 in100 and 1 in 40 in phosphate-bu¡ered saline (PBS), re-spectively. Control sections were incubated with PBS.Sections were then incubated with biotin-labelled anti-rabbit or anti-mouse secondary antibody followed bystreptavidin-biotin complex (ABC) for 30min. Label-ling reactions were developed for 5minwith a solutionof 3,30-diaminobenzidine tetrahydrochloride andH2O2 in PBS bu¡er. Slides were counterstained withGill’s haematoxylin.

Immunohistochemical Evaluation

Positivity was indicated by the presence of distinctbrown cytoplasmic labelling. Immunoreactivity wasevaluated ‘‘blind’’ by two observers using an immuno-histochemical score (IHS) method, described pre-viously (Soslow et al., 2000; DoreŁ et al., 2003). Brie£y,the IHSwasbasedon estimates of the percentage of im-munoreactive cells (‘‘grade’’) and of labelling intensity.The former was evaluated as follows: no cells labelled(grade 0);1^10% (grade 1);11^50% (grade 2); 51^80%(grade 3); 81^100% (grade 4). The labelling intensitywas recorded as: negative (0); weak (+); moderate(++); strong (+++).

Statistical Analyses

The Mann^Whitney test was used to analyse di¡er-ences between benign and malignant tumours in re-spect of frequencies of Cox-1 and Cox-2 expression.Statistical analyses were performed with the SPSS sys-tem (SPSS Inc., Chicago, USA), version 12.0, Po0.05being considered signi¢cant.

Results

In all, 70 mammary samples (four normal, six hyper-plastic, 60 neoplastic [21 benign and 39 malignant])were examined (Table1).Immunoexpression of Cox-1andCox-2 inmammary

lesions is summarized inTables 2 and 3. Cox-1 expres-sion was not signi¢cantly di¡erent in benign and ma-lignant mammary tumours (P ¼ 0.272). Cox-2immunoexpression, however, was higher in malignanttumours than in the benign counterparts (Po0.001).

ARTICLE IN PRESS

Table 2Cox-1 and -2 expression in 27 mammary samples (hyperplasia or benign tumour)

Sample no. Histological diagnosis (27 samples) Cox-1 Cox-1score� Cox-2 Cox-2 score�

Grade� Intensity� Grade� Intensity�

1 Lobular hyperplasia 3 + 3 2 ++ 42 Lobular hyperplasia 2 ++ 4 3 + 33 Lobular hyperplasia 3 ++ 6 3 + 34 Lobular hyperplasia 3 +++ 9 3 ++ 65 Ductal hyperplasia 3 ++ 6 0 0 06 Ductal hyperplasia 3 ++ 6 0 0 07 Complex adenoma 4 +++ 12 3 + 38 Complex adenoma 3 +++ 9 4 + 49 Complex adenoma 3 ++ 6 2 +++ 610 Complex adenoma 3 +++ 9 3 ++ 611 Complex adenoma 3 ++ 6 2 ++ 412 Complex adenoma 4 +++ 12 3 ++ 313 Complex adenoma 3 +++ 9 4 + 414 Complex adenoma 3 +++ 9 3 ++ 615 Complex adenoma 3 +++ 9 2 ++ 416 Papillary adenoma 3 +++ 9 2 ++ 417 Papillary adenoma 2 + 2 2 + 218 Duct papilloma 3 +++ 9 3 + 319 Simple adenoma 4 + 4 2 ++ 420 Simple adenoma 3 ++ 6 3 + 321 Tubular adenoma 2 +++ 6 2 ++ 422 Tubular adenoma 2 ++ 4 2 ++ 423 Tubular adenoma 2 + 2 4 ++ 824 Benign mixed tumour 3 ++ 6 2 ++ 425 Benign mixed tumour 3 +++ 9 2 ++ 426 Benign mixed tumour 3 +++ 9 3 +++ 927 Benign mixed tumour 3 +++ 9 4 ++ 8

�SeeMaterials andMethods.The sample nos do not correspond with those inTable 3.

Cox-1 and Cox-2 in Canine Mammary Tumours 179

Immunoexpression in Normal Mammary Gland

In all four samples, Cox-1 immunoexpression was de-tected inmore than 50%of epithelial cells, asmoderateto strong cytoplasmic labelling. Cox-2 immunolabel-ling of weak intensity was observed in the cytoplasmof some (o10%) epithelial cells in two of the foursamples.

Immunoexpression in Hyperplastic Mammary Lesions andBenign Tumours

The IHSvalues for hyperplasticmammary lesions ran-ged from 3 to 6 (Cox-1) and 0 to 6 (Cox-2). Labelling ofboth Cox-1andCox-2 was observed in the cytoplasm ofepithelial cells, with a heterogeneous labelling pattern.Ductal hyperplasia was the only mammary lesion thatdid not express Cox-2. In two cases of lobular hyperpla-sia, Cox-2 expressionwas moderate, with focal areas ofstrong immunoreactivity.The IHS values of benign tumours ranged from 2 to

12 (Cox-1) and 2 to 9 (Cox-2). Both Cox-1 and Cox-2immunoexpression was observed in the cytoplasm ofepithelial cells. Cox-2 expression varied between be-

nign tumours of di¡erent histological type. Thus, insimple adenomas, occasional small foci (o10% ofcells) were labelled (with strong intensity), while in tu-bulopapillary and mixed benign tumours 10^20% ofcells were immunoreactive.In benign tumours of complex type (n ¼ 9) and in

benign mixed tumours (n ¼ 4), Cox-1 overexpressionpresented a fairly uniform pattern in epithelial, myoe-pithelial, chondroblast and stromal cells (Fig.1). Cox-2expression, however, was particularly intense in theepithelial components.When cartilage formation waspresent (n ¼ 7), the chondroblasts were positive forCox-2, withweak to moderate intensity.In benign tumours of simple type (n ¼ 8), Cox-1

overexpression presented a uniform pattern, but Cox-2immunoexpression was particularly intense in theepithelial components and in cells with a luminal loca-tion (Fig. 2). The myoepithelial basal cells were nega-tive for Cox-2.In 15 tumour cases, normal mammary lobules were

present adjacent to the neoplastic areas. In this appar-ently normal tissue, Cox-1 immunoexpression was ob-served in more than 50% of the epithelial cells. The

ARTICLE IN PRESS

Table 3Cox-1 and -2 expression in 39 malignant tumours

Sample no. Histological diagnosis (39 samples) Cox-1 Cox-1score� Cox-2 Cox-2 score�

Grade� Intensity� Grade� Intensity�

1 Complex carcinoma 2 +++ 6 3 ++ 62 Complex carcinoma 4 +++ 12 3 +++ 93 Complex carcinoma 3 +++ 9 3 ++ 64 Complex carcinoma 2 ++ 4 3 +++ 95 Complex carcinoma 3 +++ 9 3 +++ 96 Complex carcinoma 2 ++ 4 4 + 47 Complex carcinoma 3 +++ 9 2 ++ 48 Complex carcinoma 3 ++ 6 2 +++ 69 Complex carcinoma 3 +++ 9 3 +++ 910 Squamous cell carcinoma 3 ++ 6 4 +++ 1211 Squamous cell carcinoma 4 +++ 12 4 +++ 1212 Solid carcinoma 2 ++ 4 3 ++ 613 Solid carcinoma 2 +++ 6 4 +++ 1214 Solid carcinoma 3 +++ 9 3 +++ 915 Tubulopapillary carcinoma 3 ++ 6 3 ++ 616 Tubulopapillary carcinoma 2 +++ 6 3 +++ 917 Tubulopapillary carcinoma 3 +++ 9 4 ++ 818 Tubulopapillary carcinoma 2 +++ 6 2 +++ 619 Tubulopapillary carcinoma 4 +++ 12 4 ++ 820 Tubulopapillary carcinoma 4 +++ 12 3 +++ 921 Tubulopapillary carcinoma 3 +++ 9 3 ++ 622 Tubulopapillary carcinoma 3 +++ 9 3 +++ 923 Tubulopapillary carcinoma 3 +++ 9 3 ++ 624 Tubulopapillary carcinoma 2 ++ 4 3 +++ 925 Tubulopapillary carcinoma 3 +++ 9 3 +++ 926 Tubulopapillary carcinoma 3 ++ 6 3 +++ 927 Tubulopapillary carcinoma 3 +++ 9 3 ++ 628 Tubulopapillary carcinoma 3 ++ 6 3 +++ 929 Tubular carcinoma 3 +++ 9 3 +++ 930 Tubular carcinoma 3 +++ 9 4 +++ 1231 Tubular carcinoma 4 ++ 8 3 ++ 632 Tubular carcinoma 2 +++ 6 3 +++ 933 Tubular carcinoma 4 +++ 12 3 +++ 934 Tubular carcinoma 3 ++ 6 4 ++ 835 Carcinosarcoma 4 +++ 12 4 +++ 1236 Carcinosarcoma 3 +++ 9 4 +++ 1237 Carcinosarcoma 4 ++ 8 3 +++ 938 Ductal in-situ carcinoma 3 +++ 9 4 ++ 839 Ductal in-situ carcinoma 3 +++ 9 3 +++ 9

�SeeMaterials andMethods.The sample nos do not correspond with those inTable 2.

F.L. Queiroga et al.180

immunolabelling was cytoplasmic and identical withthat observed in neoplastic cells. Cox-2 immunolabel-ling was observed in epithelial cells in ¢ve of these15 cases. This Cox-2 immunoexpression was focal(o10% of cells) and less intense than in epithelialneoplastic cells.

Immunoexpression in Malignant Tumours

Immunoexpression of Cox-1 and Cox-2 was observedin all malignant mammary tumours (n ¼ 39).Cox-1 labelling was cytoplasmic with a homoge-

neous distribution in epithelial, myoepithelial, chon-droblast and stromal cells. There were no di¡erences

in the immunolabelling pattern between di¡erent his-tological types. Multifocal areas of apparently normalmammary gland adjacent to malignant tumoursshowed weak to moderate Cox-1 labelling in 16 out of19 cases.Cox-2 immunolabelling was predominantly cyto-

plasmic and granular (Fig. 3). Additionally,14 of the 39samples showed strong perinuclear labelling, and inthree cases (two carcinosarcomas and one solid carci-noma) nuclear labelling was evident (Fig. 4). In malig-nant tumours of simple type (n ¼ 26), Cox-2immunoreactivity was more intense in cells with a lu-minal location (Fig. 5). Fusiform stromal componentsoccasionally presented a di¡use or focal reactivity for

ARTICLE IN PRESS

Fig. 2. Case 19 (Table 2) Simple adenoma. Cox-2 labelling is parti-cularly intense in luminal epithelial cells. IHC. Bar, 30 mm.

Fig. 3. Case 30 (Table 3). Tubular carcinoma. Neoplastic epithelialcells show intense Cox-2 immunoreactivity. Perinuclear la-belling is also seen. IHC. Bar,12 mm.

Fig.4. Case14 (Table 3). Solidcarcinoma. Several neoplastic epithe-lial cells show Cox-2 nuclear labelling. IHC. Bar, 60 mm.

Fig. 1. Case 7 (Table 2). Complex adenoma. Epithelial, myoepithe-lial and stromal cells show moderate to intense Cox-1 immu-noreactivity. IHC. Bar, 30 mm.

Cox-1 and Cox-2 in Canine Mammary Tumours 181

Cox-2. The two cases of ductal in-situ carcinomashowed a high Cox-2 score with cytoplasmic granularexpression. In complex carcinomas, Cox-2 expressionwas present both in simple and complex areas.Whencartilage formation was present (n ¼ 6), the chondro-blasts showed moderate to strong immunolabelling. Incarcinosarcomas, both carcinomatous and sarcoma-tous areas showed strong and homogeneous Cox-2 ex-pression (Figs 6 and 7). Cox-2 was also expressed innon-neoplastic mammary lobules adjacent to the ma-lignant tumour tissue in10 of 19 cases. Labelling in ap-parently normal epithelial cells was often multifocal(10^50% of epithelial cells) and with an intensity simi-lar to or less than that of neoplastic cells.

Discussion

This study demonstrated the expression of both Cox-1and Cox-2 in a large sample of benign andmalignant ca-ninemammary lesions. Immunoexpression of Cox-2 wasdescribedearlierbyDoreŁ et al. (2003) incaninemammaryadenomas and adenocarcinomas, and Cox-1 and Cox-2immunoexpressionwas studied in a small series of caninemammary carcinomas by Mohammed et al. (2004). Thepresent study provides evidence of similar overexpressionin other histological types of canine mammary tumour,such as benignmixed and carcinosarcoma.Cox-1 expression was detected in all the samples ex-

amined, including normal mammary gland used as a

ARTICLE IN PRESS

Fig. 5. Case 17 (Table 3).Tubulopapillary carcinoma. Luminal neo-plastic epithelial cells show strong Cox-2 cytoplasmic label-ling. IHC. Bar, 60 mm.

Fig. 6. Case 35 (Table 3). Carcinosarcoma (carcinomatous area).Strong Cox-2 immunoreactivity is seen in neoplastic epithe-lial cells. IHC. Bar, 30 mm.

Fig.7. Case 35 (Table 3). Carcinosarcoma (sarcomatous area). Neo-plastic cells show strong and homogeneous Cox-2 immunola-belling. IHC. Bar, 30 mm.

F.L. Queiroga et al.182

control. Cox-2 labellingwas identi¢ed in 96% (n ¼ 25)of benign neoplastic and hyperplasia cases and in100% (n ¼ 39) of the malignant tumours. Ductal hy-perplasia was the only mammary lesion not to expressCox-2.Two of four samples of normal mammary gland

showed focal reactivity for Cox-2 in apparently normalepithelial cells, as previously reported by Mohammedet al. (2004) and, in canine nasalmucosa, byKleiter et al.(2004). This focal immunoreactivity possibly indicatesthat Cox-2 has a constitutive function in canine tissues,but further studies are necessary to clarify its role innormal canine mammary gland.

Cox-1 expression was similar in benign and malig-nantmammary lesions, but Cox-2 expressionwasmorestriking in malignant tumours than in benign counter-parts. In benign tumours, however, Cox-2 immunola-belling varied, some histological types (e.g., twobenign mixed tumours and one tubular adenoma)showing high IHS values. Some benign mammary tu-mours progress to become malignant (Perez-Alenzaet al., 2000; Sorenmo, 2003), and Cox-2 overexpressionmay prove to be of value in identifying such tumours.In human breast cancer, Cox-2 overexpression has

been linked to the presence of distant metastasis andpoor prognosis (Denkert et al., 2004). In the presentstudy, Cox-2 expression was particularly striking inmalignant tumours of histological types classically de-¢ned as more aggressive, e.g., squamous cell carcino-mas and carcinosarcomas, both of which are highlyaggressive in the dog and have a poor prognosis. Thismay indicate a link between high Cox-2 expressionand malignancy in canine mammary tumours, as hasbeen observed in several human (Costa et al., 2002;Kirkpatrick et al., 2002; Hazar et al., 2004) and animal(Liu et al., 2001; Beam et al., 2003; Borzacchiello et al.,2003) tumours.It was noteworthy that two cases of mammaryductal

carcinoma in situ were associated with elevated Cox-2scores, suggesting that Cox-2 overexpressionmaybe anearly event in canine mammary carcinogenesis, as ob-served in human breast cancer (Half et al., 2002; Shimet al., 2003;Yoshimura et al., 2003; Boland et al., 2004).Some of the malignant tumour samples showed a

Cox-2 perinuclear (14 of 39 cases) or nuclear (twocarcinosarcomas and one solid carcinoma) labellingpattern, a ¢nding not previously described in caninemammary tumours but reported in human breast

ARTICLE IN PRESS

Cox-1 and Cox-2 in Canine Mammary Tumours 183

cancer (Ristimaki et al., 2001; Sucic et al., 2003). More-over, perinuclear immunolabelling was detected in ca-nine intestinal tumours by McEntee et al. (2002).Subcellular localization of Cox-2 in the nuclear envel-ope may explain perinuclear immunolabelling (Mori-ta et al., 1995). A nuclear labelling pattern may berelated to the ¢nding that Cox-2 nuclear localizationre£ects the role of Cox-2 in the regulation of gene ex-pressionby its tra⁄cking-import fromcytoplasmto nu-cleus in cells with activated DNA transcription(Parfenova et al., 2001). In future studies, not only theintensity of Cox-2 labelling but also its type (cytoplas-mic, perinuclear, nuclear) should be borne in mind.Normal mammary gland adjacent to malignant tu-

mours expressed Cox-2 with an immunolabelling in-tensity similar to that observed in neoplastic cells.Thismay indicate an increased probability of neoplasia inthe neighbourhood of the primary tumour, and thatapparently normal epithelial cells have already suf-fered neoplastic activation (Soslow et al., 2000).Cox-1 and Cox-2 are the main targets of non-steroi-

dal anti-in£ammatory drugs (NSAIDs) (Howe et al.,2001; Dang et al., 2002; Davies et al., 2002). Inhibition ofCox-2 by NSAIDs would appear to be of value in rela-tion to human colon and breast cancer (Harris et al.,1999, 2000; Fosslien, 2000; Moran, 2002). Clinical trialsin dogs with invasive bladder tumours showed that pir-oxicam (an NSAID and Cox inhibitor) induced tu-mour remission through apoptosis and reduction oftumour angiogenesis (Knapp et al.,1994,2002). It seemspossible that NSAIDs, particularly Cox-2 inhibitors,may prove useful in treating canine mammary tu-mours.In conclusion, the study demonstrated that malig-

nant tumours, especially those showing histologicalfeatures associated with high malignancy, expressedCox-2 with particular intensity. Further studies will benecessary to con¢rm the biological implications ofthese ¢ndings.

Acknowledgments

A PhD Grant (SFRH/BD/10883/2002) from the Foun-dation of Science andTechnology (FCT), Portugal sup-ported this study.The authors thank Prof. JoseŁ Aranha,Department of Forest Sciences, University of TraŁ s-os-Montes and Alto Douro (UTAD), for assistance withthe statistical work and Mrs L|¤ gia Bento, UTAD, forexpert technical assistance.

References

Bauer, A. K., Dwyer-Nield, L. D. and Malkinson, A. M.(2000). High cyclooxygenase1 (COX-1) and cyclooxygen-ase 2 (COX-2) contents in mouse lung tumours. Carcino-genesis, 21, 543^550.

Beam, S. L., Rassnick, K. M., Moore, A. S. and McDo-nough, S. P. (2003). An immunohistochemical study of cy-clooxygenase-2 expression in various feline neoplasms.Veterinary Pathology, 40, 496^500.

Boland, G. P., Butt, I. S., Prasad, R., Knox,W. F. and Bun-dred, N. J. (2004). COX-2 expression is associated withan aggressive phenotype in ductal carcinoma in situ. Brit-ishJournal of Cancer, 90, 423^429.

Borzacchiello, G., Ambrosio, V., Galato, P., Perillo, A. andRoperto, F. (2003). Cyclooxygenase-1 and -2 expressionin urothelial carcinomas of the urinary bladder in cows.Veterinary Pathology, 40, 455^459.

Costa, C., Soares, R., Reis-Filho, J. S., Leita� o, D., Amen-doeira, I. and Schmitt, F. C. (2002). Cyclo-oxygenase 2 ex-pression is associated with angiogenesis and lymph nodemetastasis in human breast cancer. Journal of Clinical

Pathology, 55, 429^434.Dang, C.T., Shapiro, C. L. andHudis, C. A. (2002). Potential

role of selective COX-2 inhibitors in cancer management.Oncology, 16, 30^36.

Davies, G., Martin, L. A., Sacks, N. and Dowsett, M. (2002).Cyclooxygenase-2 (COX-2), aromatase and breast can-cer: a possible role for COX-2 inhibitors in breast cancerchemoprevention. Annals of Oncology, 13, 669^678.

Denkert, C.,Winzer, K. J. and Hauptmann, S. (2004). Prog-nostic impact of cyclooxygenase-2 in breast cancer. Clini-cal Breast Cancer, 4, 428^433.

DoreŁ , M., Lanthier, I. and Sirois, J. (2003). Cyclooxygenase-2 expression in canine mammary tumours. VeterinaryPathology, 40, 207^212.

Dubois, R. N., Abramson, S. B., Cro¡ord, L., Gupta, R. A.,Simon, L. S.,Van De Putte, L. B. and Lipsky, P. E. (1998).Cyclooxygenase inbiologyanddisease.FASEBJournal,12,1063^1073.

Fosslien, E. (2000).Molecular pathology of cyclooxygenase-2in neoplasia. Annals of Clinical and Laboratory Science, 30,3^21.

Half, E.,Tang, X. M., Gwyn, K., Sahin, A.,Wathen, K. andSinicrope, F. A. (2002). Cyclooxygenase-2 expression inhuman breast cancers and adjacent ductal carcinoma insitu. Cancer Research, 62,1676^1681.

Harris, R. E., Alsha¢e, G. A., Abou-Issa, H. and Seibert, K.(2000). Chemoprevention of breast cancer in rats by cele-coxib, a cyclooxygenase 2 inhibitor. Cancer Research, 60,2101^2103.

Harris, R. E., Kasbari, S. and Farrar,W. B. (1999). Prospec-tive study of nonsteroidal anti-in£ammatory drugs andbreast cancer.Oncology Reports, 6,71^73.

Hazar, B., Ergin, M., Seyrek, E., Erdogan, S.,Tuncer, I. andHakverdi, S. (2004). Cyclooxygenase-2 (Cox-2) expres-sion in lymphomas. Leukemia&Lymphoma, 45,1395^1399.

Hla, T., Bishop-Bailey, D., Liu, C. H., Schaefers, H. J. andTrifan, O. C. (1999). Cyclooxygenase-1and -2 isoenzymes.International Journal of Biochemistry & Cell Biology, 31,551^557.

Howe, L. R., Subbaramaiah, K., Brown, A.M. andDannen-berg, A. J. (2001). Cyclooxygenase-2: a target for the pre-vention and treatment of breast cancer. Endocrine RelatedCancer, 8, 97^114.

ARTICLE IN PRESS

F.L. Queiroga et al.184

Hwang, D., Scollard, D., Byrne, J. and Levine, E. (1998). Ex-pression of cyclooxygenase-1and cyclooxygenase-2 in hu-man breast cancer. Journal of the National Cancer Institute,90, 455^460.

Khan, K., Knapp, D., Denicola, D. and Harris, R. (2000).Expression of cyclooxygenase-2 in transitional cell carci-noma of the urinary bladder in dogs. AmericanJournal ofVeterinary Research, 61, 478^481.

Kirkpatrick, K., Ogunkolade,W., Elkak, A., Bustin, S., Jen-kins, P., Ghilchik, M. andMokbel, K. (2002).ThemRNAexpression of cyclo-oxygenase-2 (COX-2) and vascularendothelial growth factor (VEGF) in human breast can-cer. CurrentMedical Research and Opinion, 18, 237^241.

Kleiter, M., Malarkey, D. E., Ruslander, D. E. andThrall, D.E. (2004). Expression of cyclooxygenase-2 in canineepithelial nasal tumours.Veterinary Radiology&Ultrasound,45, 255^260.

Knapp, D.W., Glickman, N.W., Mohammed, S. I., DeNico-la, D. B.,Widmer,W. R., Bonney, P. L. and DeGortari, A.E. (2002). Antitumour e¡ects of piroxicam in spontaneouscanine invasive urinary bladder cancer, a relevant modelof human invasive bladder cancer.Advances in ExperimentalMedicine and Biology, 507, 377^380.

Knapp, D.W., Richardson, R. C., Chan,T. C., Bottoms, G.D.,Widmer,W. R., DeNicola, D. B., Teclaw, R., Bonney,P. L. and Kuczek,T. (1994). Piroxicam therapy in 34 dogswith transitional cell carcinoma of the urinary bladder.Journal ofVeterinary InternalMedicine, 8, 273^278.

Lee, C. H., Kim,W. H., Lim, J. H., Kang, M. S., Kim, D.Y.andKweon,O.K. (2004).Mutationandoverexpressionofp53 as a prognostic factor in canine mammary tumours.Journal ofVeterinary Science, 5, 63^69.

Liu, C. H., Chang, S. H., Narko, K.,Trifan, O. C.,Wu,M.T.,Smith, E., Haudenschild, C., Lane, T. F. and Hla, T.(2001). Overexpression of cyclooxygenase-2 is su⁄cient toinduce tumorigenesis in transgenic mice. Journal of Biolo-gical Chemistry, 276,18563^18569.

McEntee, M. F., Cates, J. M. and Neilsen, N. (2002). Cy-clooxygenase-2 expression in spontaneous intestinal neo-plasia of domestic dogs.Veterinary Pathology, 39, 428^436.

Misdorp,W. (2002).Tumours of the mammary gland. In:Tu-mours in Domestic Animals. 4th Edit., D.J. Meuten, Ed.,Iowa State Press, Iowa, USA, pp. 575^606.

Misdorp, W., Else, R.W., HellmeŁ n, E. and Lipscomb, T.P.(1999). Histological Classi¢cation of MammaryTumours of the

Dogand the Cat, 2nd series,Vol.VII, Armed Forces Instituteof Pathology, American Registry of Pathology,Washing-ton, DC, and theWorld Health Organization Collaborat-ing Center for Worldwide Reference on ComparativeOncology, pp.1^59.

Mohammed, S. I., Khan, K. N., Sellers, R. S., Hayek,M. G.,DeNicola, D. B.,Wu, L., Bonney, P. L. and Knapp, D.W.(2004). Expression of cyclooxygenase-1and 2 in naturally-occurring canine cancer. Prostaglandins Leukotrienes and Es-sential FattyAcids, 70, 479^483.

Moran, E. M. (2002). Epidemiological and clinical aspects ofnonsteroidal anti-in£ammatory drugs and cancer risks.Journal of Environmental Pathology, Toxicology and Oncology,21,193^201.

Morita, I., Schindler,M., Regier,M.K., Otto, J. C., Hori,T.,DeWitt, D. L. and Smith,W. L. (1995). Di¡erent intracel-lular locations for prostaglandin endoperoxide Hsynthase-1 and -2. Journal of Biological Chemistry, 270,10902^10908.

Parfenova, H., Parfenov, V. N., Shlopov, B. V., Levine, V.,Falkos, S., Pourcyrous, M. and Le¥er, C.W. (2001). Dy-namics of nuclear localization sites for Cox-2 in vascularendothelial cells. AmericanJournal of PhysiologyçCell Phy-

siology, 281C,166^178.Pena, L. L., Nieto, A. I., PeŁ rez-Alenza, D., Cuesta, P. and

Castano, M. (1998). Immunohistochemical detection ofKi-67 and PCNA in canine mammary tumours: relation-ship to clinical and pathologic variables. Journal ofVeterin-ary Diagnostic Investigation, 10, 237^246.

Perez-Alenza, M. D., Pena, L., del Castillo, N. and Nieto, A.I. (2000). Factors in£uencing the incidence and prognosisof caninemammary tumours. Journal of Small Animal Prac-tice, 41, 287^291.

Pestili de Almeida, E., Piche, C., Sirois, J. and DoreŁ , M.(2001). Expression of cyclo-oxygenase-2 in naturally oc-curring squamous cell carcinomas in dogs. Journal of His-

tochemistry and Cytochemistry, 49, 867^875.Ristimaki, A., Nieminen, O., Saukkonen, K., Hotakainen,

K., Nordling, S. andHaglund, C. (2001). Expression of cy-clooxygenase-2 in human transitional cell carcinoma ofthe urinary bladder. American Journal of Pathology, 158,849^853.

Sarli, G., Preziosi, R., Benazzi, C., Castellani, G. and Mar-cato, P. S. (2002). Prognostic value of histologic stage andproliferative activity in canine malignant mammary tu-mours. Journal of Veterinary Diagnostic Investigation, 14,25^34.

Shim,V., Gauthier, M. L., Sudilovsky, D., Mantei, K., Chew,K. L.,Moore, D.H., Cha, I.,Tlsty,T. D. and Esserman, L.J. (2003). Cyclooxygenase-2 expression is related to nucle-ar grade in ductal carcinoma in situ and is increased in itsnormal adjacent epithelium. Cancer Research, 63,2347^2350.

Sorenmo, K. (2003). Canine mammary gland tumours.Veter-inary Clinics of North America: Small Animal Practice, 33,573^596.

Soslow, R. A., Dannenberg, A. J., Rush, D.,Woerner, B. M.,Khan, K. N.,Masferrer, J. andKoki, A.T. (2000). COX-2is expressed in human pulmonary, colonic, and mam-mary tumours. Cancer, 89, 2637^2645.

Sucic, M., Boban, D., Markovic-Glamocak, M., Jakic-Razu-movic, J., Vrbanec, D., Ries, S., Gjadrov-Kuvezdic, K.,Cobanov, A.M., Podolski, P.,Martinovic, I., Erak,V., Bu-disic, Z., Ivrlac, R. and Unusic, J. (2003). Expression ofcyclooxygenase-2 in ¢ne-needle aspirates frombreast car-cinoma and benign breast diseases. Breast, 12, 51^57.

Williams, C. S. and DuBois, R. N. (1996). Prostaglandin en-doperoxide synthase: why two isoforms? AmericanJournalof Physiology, 270, 393^400.

Wilson, J. E., Chandrasekharan, N.V.,Westover, K. D., Ea-ger, K. B. and Simmons, D. L. (2004). Determination ofexpression of cyclooxygenase-1 and -2 isozymes in caninetissues and their di¡erential sensitivity to nonsteroidal

ARTICLE IN PRESS

Cox-1 and Cox-2 in Canine Mammary Tumours 185

anti-in£ammatory drugs. AmericanJournal ofVeterinary Re-search, 65, 810^818.

Yamagami,T., Kobayashi,T.,Takahashi, K. and Sugiyama,M. (1996). Prognosis for canine malignant mammary tu-moursbasedonTNMandhistologic classi¢cation. JournalofVeterinaryMedical Science, 58,1079^1083.

Yoshimura, N., Sano, H., Okamoto, M., Akioka, K., Ushi-gome, H., Kadotani,Y.,Yoshimura, R., Nobori, S., Higu-

chi, A., Ohmori,Y. and Nakamura, K. (2003). Expressionof cyclooxygenase-1and -2 in humanbreast cancer. SurgeryToday, 33, 805^811.

Received, January 3rd, 2006

Accepted, January 30th, 2007

� �