High Intrinsic Expression of P-glycoprotein and Breast Cancer Resistance Protein in Canine Mammary Carcinomas Regardless of Immunophenotype and Outcome

Simple Summary Multidrug resistance of neoplastic cells to chemotherapeutic drugs is a phenomenon mediated by several molecular mechanisms. Among these, P-glycoprotein (P-gp) and Breast Cancer Resistance Protein (BCRP) counteract the intracellular load of multiple drugs, preventing their efficacy. The basal (intrinsic) cellular expression can be further stimulated by drug exposure. P-gp and BCRP are a subject of intense investigation both in human and veterinary oncology since a better understanding of how their expression is distributed across different tumors allows planning alternative therapeutic strategies. In canine mammary carcinomas, a phenotypic classification similar to the one widely adopted for breast cancer is currently employed. For Basal- and Normal-like phenotypes, chemotherapy is still the main option. In this study, we observed that canine mammary carcinomas bear a high intrinsic expression of both P-gp and BCRP, regardless of their molecular phenotype, and their presence does not influence the outcome. Abstract P-glycoprotein (P-gp) and breast cancer resistance protein (BCRP) are major actors in multidrug resistance (MDR) phenomenon in both human and canine mammary carcinomas (CMCs). The aim of this study was to investigate an association between the intrinsic expression of P-gp and BCRP compared to the immunophenotypes and outcome in CMCs. Fifty CMCs were evaluated at immunohistochemistry (IHC) for P-gp, BCRP, Estrogen receptor alpha (ER), Progesterone receptors (PR), Human Epidermal Growth Factor Receptor type 2 (HER2), basal cytokeratins 5/6 (CK5/6), Epidermal Growth Factor Receptor 1 (EGFR), and Ki67 proliferation index. P-gp and BCRP positive cases were, respectively, 52% and 74.5%, with a significantly higher expression of BCRP than P-gp. Five immunophenotypes were defined in 37 out of 50 CMCs: 9 (24.3%) Luminal A, 5 (13.5%) Luminal B, 9 (24.3%) HER2 overexpressing, 9 (24.3%) Triple-negative basal-like, and 5 (13.5%) Triple-negative non-basal-like. In all CMCs at least one marker was expressed. Follow-up data were available for 25 animals. The average cancer-specific survival was 739 ± 444 days. A number of CMCs bear a high expression of P-gp and BCRP but no significant association was found between their expression and the immunophenotypes, Ki67 index, the histological grade, and tumor-related death.


Introduction
Multidrug resistance (MDR) of neoplastic cells to multiple chemotherapeutic drug is a complex phenomenon which remains a major challenge in the treatment of cancer [1].

Sample Selection, History and Histological Analysis
Samples were retrieved from the archive of the Department of Veterinary Medical Sciences, University of Bologna, of the Ospedale Veterinario "I Portoni Rossi" Bologna and of the Clinica Veterinaria Malpensa, Italy. Fifty formalin-fixed and paraffin-embedded tissue samples of mammary glands from 45 dogs were selected at the optic microscope based on the histopathological diagnoses of mammary carcinoma. The anamnestic and clinical data available from the archive, such as age, sex and neutering status, size of the tumor measured by the pathologist at the trimming of the surgical excised mass, presence of lymph node or systemic metastases and clinical stage (TNM system) were collected and the follow-up period was defined as 24 months between histologic diagnosis and last data collection. According to the reports, the dogs had not received chemotherapy at the time of biopsy/surgery. Cancer-specific survival (CSS) was defined as the period between surgery and tumor-related death, which was clinically outlined as spontaneous death or euthanasia due to tumor-related issues. Follow-up data were retrieved by phone calls to the referent veterinarian or to the owner of the dogs.
The original histologic diagnosis and histological grade was reviewed for each slide and updated, when necessary, according to the current histologic classification [21].
The study was conducted in accordance with guidelines and regulation and in compliance with the current national legal treatment of animal tissue samples.

Immunohistochemistry
A combination of 6 immunohistochemical markers: Estrogen Receptor alpha (ER), Progesterone Receptor (PR), HER2, Cytokeratins 5/6 (CK5/6), Epidermal Growth Factor Receptor 1 (EGFR), and Ki67, was used to define the immunophenotypes of CMCs as proposed for human breast cancer immunophenotypes [22]. P-gp and BCRP were assessed to establish the chemoresistance potential of CMCs. Formalin-fixed and paraffin waxembedded tissues were sectioned (3 µm). The primary antibody types, dilutions, antigen retrieval methods, and tissues used as positive (internal and external) controls are reported in Table 1. Prior to antigen retrieval, endogenous peroxidase was blocked by immersion in H 2 O 2 3% in methanol for 30 . Blocking of non-specific antigenic sites was achieved by incubating the slides in a solution of 10% goat serum and PBS for 30 at room temperature. Slides were then rinsed with TRIS for reagent removal and were incubated overnight at 4 • C with the primary antibody diluted in a solution of 10% goat serum and PBS. The slides were rinsed in TRIS buffer and then incubated with secondary anti-mouse antibody (biotinylated goat anti-mouse immunoglobulins; Dako, Glostrup, Denmark) diluted 1 in 200 in 10% NGS in PBS. The reaction was revealed by a commercial streptavidin-biotin-peroxidase technique (ABC Kit Elite, Vector, Burlingame, CA, USA) and visualized with 3,3 -Diaminobenzidine in tablets (DAB chromogen/substrate kit; Diagnostic BioSystem, Pleasanton, CA, USA). Slides were counterstained with Harris' hematoxylin and permanently mounted with DPX mountant.
External positive controls, and positive control internal to the examined tissue, when available, were examined. Corresponding negative control slides were processed in parallel by replacing the primary antibody with an isotype control non-reactive antibody, purchased at the same provider for each primary antibody.
All the markers except Ki67 were scored by semiquantitative evaluation of 10 representative high-power fields at the optical microscope. P-gp and BCRP were considered positive when ≥20% and ≥10% of cells were labelled for P-gp and BCRP, respectively, as suggested by previous studies [14,18,23]. A further semiquantitative evaluation of the immunostaining was performed, and the positive carcinomas were assigned to 2 subgroups according to the percentage of positive cells: intermediate positivity (range of 20-50% of P-gp-positive cells; range of 10-50% of BCRPpositive cells) or high positivity (≥50% of P-gp-positive cells; ≥50% of BCRP-positive cells).
ER and PR immunolabelling were evaluated according to the consensus on stand-ard guidelines for hormone receptor assessment using immunohistochemistry [24,25] and the Allred score for the epithelial and myoepithelial component of each carcinoma was performed. According to the leading publication by Nguyen et al. (2018), a cut-off of ≥10% of positive nuclei in the epithelial component of carcinomas was adopted to assign hormones receptor positive status [26].
HER2 was scored based on the 2018 ASCO/CAP guidelines for the assessment of HER2 status in breast cancer. Carcinomas were considered HER2 positive only for a 3+ IHC score [26,27].
EGFR and CK5/6 were considered positive at a threshold ≥10% of positive cells as suggested by previous studies on canine mammary tumors [25,26].
A semiquantitative evaluation was performed for each marker, assessing the percentage of positive cells at the optic microscope in 10 high power fields, representative of the histological subtype and grade of the carcinomas, as suggested by the literature [25,26].
Ki67 proliferation index was assessed by manual image analysis based on the number of positive nuclei among >500 neoplastic cells and expressed as a percentage (Image J software, National Institute of Health, Bethesda, MA, USA).
Cases showing no reactivity of the internal positive control, for the examined marker, presumably because of a deterioration of the antigen due to formalin fixation, were excluded.
Immunophenotypes were classified into five groups according to the scheme proposed for breast carcinomas by Nielsen et al. (2004) and translated to canine mammary carcinomas by Abadie et al. (2018) [22,28]

Statistical Analysis
Comparison between groups was analyzed by Chi square test with Yates correction. Correlations between categorical variables were analyzed using the Pearson χ 2 test. Survival curves were computed using the Kaplan and Meier estimate and compared by log-rank test. For all statistical tests, a p-value < 0.05 was considered significant.

Results
All the data collected for this study are reported in Supplementary Materials Table S1.

Animal Data and Histopathological Characteristics of Tumors
The selected samples included 50 CMCs from 45 female dogs, 16 of which were spayed. Age ranged from 4.7 to 13.6 years, with median age of 10 years and average value of 9.8 ± 2.6 years. Maximum tumor size was 50 mm diameter and minimum 5 mm, with an average value of 25.24 ± 12.24 mm.
Follow-up data at 24 months post-surgery were available for 25 animals, 13 of which were alive, and 15 were dead, 4 for causes unrelated to the mammary carcinoma. Minimum CSS was 59 days, corresponding to a 14-year-old spayed dog with inflammatory mammary carcinoma. Excluding dogs that died of causes other than mammary carcinoma, the average CSS was 739 ± 444 days.
Seven out of 14 carcinomas, for which information was available, presented lymph node metastasis at the time of diagnosis.
Four out of 16 carcinomas, for which information was available, presented systemic metastases (pulmonary and disseminated to multiple organs) at the time of diagnosis.
Fourteen carcinomas were classified according to their clinical stage (TNM system): 2 were stage I, 1 was stage II, 7 were stage IV and 4 were stage V.
There were 19 carcinomas that were histological grade I; 13 carcinomas were grade II; and 16 were grade III. Inflammatory carcinomas (2 in this caseload) were not graded by this system.

Immunohistochemistry
The results for each immunohistochemical marker are reported in Table 2.  For each marker, the cases with inconsistent staining of the internal positive control were excluded from the study and comprised: 2 cases for P-gp, 3 cases for BCRP, 5 cases for ER, 14 cases for PR, 7 cases for EGFR, 4 cases for CK5/6, and 4 cases for Ki67. This allowed to summarize a molecular phenotype in only 37 out of the 50 CMCs available. Relevant pictures of the IHC external and/or internal positive CTR are reported in Figure 1. For each marker, the cases with inconsistent staining of the internal positive control were excluded from the study and comprised: 2 cases for P-gp, 3 cases for BCRP, 5 cases for ER, 14 cases for PR, 7 cases for EGFR, 4 cases for CK5/6, and 4 cases for Ki67. This allowed to summarize a molecular phenotype in only 37 out of the 50 CMCs available. Relevant pictures of the IHC external and/or internal positive CTR are reported in Figure  1. ER and PR immunostaining were nuclear in the luminal epithelial, and myoepithelial cells (Figure 2c,d, respectively) and were evaluated with the Allred score as reported in Supplementary Materials Table S2; the percentage of hormones receptor positive carcinomas are reported in Table 1. For both ER and PR, the most intense immunostaining was seen in the myometrium (positive external CTR, Figure 1c,d).
EGFR immunostaining was present multifocally to diffusely, with mild to intense positive reaction at the cell membrane mainly in neoplastic luminal epithelial cells and occasionally in myoepithelial cells (Figure 2f). CK5/6 positivity was seen in myoepithelial and luminal epithelial cells with a cytoplasmic, discontinuous, multifocal, mild to intense staining (Figure 2g). P-gp immunolabelling was strong at the cellular membrane and cytoplasmic staining was found in a minority of positive neoplastic cells (Figure 2a). BCRP immunolabelling was membranous and cytoplasmic in most positive neoplastic cells (Figure 2b).
ER and PR immunostaining were nuclear in the luminal epithelial, and myoepithelial cells (Figure 2c,d, respectively) and were evaluated with the Allred score as reported in Supplementary Materials Table S2; the percentage of hormones receptor positive carcinomas are reported in Table 1. For both ER and PR, the most intense immunostaining was seen in the myometrium (positive external CTR, Figure 1c,d).
EGFR immunostaining was present multifocally to diffusely, with mild to intense positive reaction at the cell membrane mainly in neoplastic luminal epithelial cells and occasionally in myoepithelial cells (Figure 2f). CK5/6 positivity was seen in myoepithelial and luminal epithelial cells with a cytoplasmic, discontinuous, multifocal, mild to intense staining (Figure 2g).
Considering the two markers investigated, BCRP positive cases were significantly higher than P-gp positive tumors (Chi square with Yates correction, p = 0.016).
We were able to classify 37 (74%) carcinomas by their immunophenotype. The remaining 13 (26%) carcinomas lacked consistent immunoreactivity for one of the markers required for the classification, possibly due to deterioration of the antigen caused by delayed or excessive long-lasting formalin fixation.
Luminal A, HER2-overexpressing and Triple-negative carcinomas were equally frequent in this caseload (9/37; 24%). An equal percentage of CMCs were of Luminal B and Triple-negative immunophenotypes (5/37, 13.5%). A significant association between immunophenotypes and Ki67 index (Spearman test, p < 0.05) was evident, with higher values in non-luminal non-HER2 overexpressing phenotypes (i.e., Triple-negative basal-like and non basal-like); while no correlation was revealed with grade (Spearman test, p > 0.05). P-gp, BCRP, and the coexpression of P-gp and BCRP in the different immunophenotypes is reported as graphs in Figure 3. The 2 HER2-overexpressing CMCs were not included in this group because of the lack of reliable BCRP expression in one CMC and of BCRP and P-gp expression in another CMC by IHC. A total of 37 carcinomas out of 50 (74%) expressed at least one of the chemoresistance markers P-gp and/or BCRP. P-gp and BCRP were both expressed by the same carcinoma in 46% (23/50) of the tumors. BCRP alone was expressed in 12/50 carcinomas (24%), whereas P-gp alone was expressed only in 2/50 tumors (4%).
Considering the two markers investigated, BCRP positive cases were significantly higher than P-gp positive tumors (Chi square with Yates correction, p = 0.016).
We were able to classify 37 (74%) carcinomas by their immunophenotype. The remaining 13 (26%) carcinomas lacked consistent immunoreactivity for one of the markers required for the classification, possibly due to deterioration of the antigen caused by delayed or excessive long-lasting formalin fixation.
Carcinomas in our cohort were therefore subdivided as follows: • Luminal A, HER2-overexpressing and Triple-negative carcinomas were equally frequent in this caseload (9/37; 24%). An equal percentage of CMCs were of Luminal B and Triple-negative immunophenotypes (5/37, 13.5%). A significant association between immunophenotypes and Ki67 index (Spearman test, p < 0.05) was evident, with higher values in non-luminal non-HER2 overexpressing phenotypes (i.e., Triple-negative basal-like and non basal-like); while no correlation was revealed with grade (Spearman test, p > 0.05). P-gp, BCRP, and the coexpression of P-gp and BCRP in the different immunophenotypes is reported as graphs in Figure 3. The 2 HER2-overexpressing CMCs were not included in this group because of the lack of reliable BCRP expression in one CMC and of BCRP and P-gp expression in another CMC by IHC. No significant association was found among the expression of P-gp or BCRP and the immunophenotypes (p > 0.05).
All HER2-overexpressing CMCs expressed at least one of the two chemoresistance markers (7/7), and no case was BCRP negative.
No statistical correlation was found among P-gp or BCRP expression or their co-expression and the Ki67 index (p > 0.05) or the histological grade (p > 0.05). The Ki67 index No significant association was found among the expression of P-gp or BCRP and the immunophenotypes (p > 0.05).
All HER2-overexpressing CMCs expressed at least one of the two chemoresistance markers (7/7), and no case was BCRP negative.
No statistical correlation was found among P-gp or BCRP expression or their coexpression and the Ki67 index (p > 0.05) or the histological grade (p > 0.05). The Ki67 index and the histological grade were significantly correlated (R = 0.67, p < 0.00001). The results of the correlation analyses are reported in Table 3.
Survival analysis did not reveal any differences in the outcome of female dogs bearing tumors positive to P-gp or BCRP singularly or combined with respect to female dogs bearing negative tumors (survival analysis p < 0.05, Figure 4).

Discussion
Breast cancer can be classified according to an immunohistochemical panel of molecular markers, which helps to predict the prognosis and guides the therapy in routine clinical practice [29,30]. Overall carcinomas are classified as Luminal if they express ER and/or PR while the amplification of the HER2 gene drives an increased expression of HER2 that defines the HER2-overexpressing subtype [30]. Luminal and HER2 overexpressing carcinomas can benefit from specific therapies targeting their oncogenic pathways [31]. The risk of recurrence of Luminal carcinomas can be predicted more accurately with a further evaluation of the Ki67 labelling index and the expression of CK5/6 and EGFR [22,29]. Mammary carcinomas lacking the expression of ER, PR, and HER2 are known as Triple-negative breast cancers and are furtherly subclassified into basal-like carcinomas when expressing basal/myoepithelial markers: cytokeratins CK5/6, or CK14 and/or the EGFR [22,32]. Triple-negative carcinomas lacking the expression of basal markers are classified as Fivenegative [29]. Currently no targeted therapy exists for Triple-negative and Five-negative subtypes, which bear the worst prognosis and require the administration of conventional chemotherapeutic protocols, leading to the emergence of chemoresistance in many patients [5,29,33]. Thus, there is a particular need to elucidate drug resistance mechanisms for this subtype, especially in triple negative breast cancer of which the core-basal subtype, which responds poorly to cytotoxic chemotherapy, has the worst prognosis [6].
Research on CMCs has successfully translated this human-based molecular classification to the bitch [25,28,[34][35][36][37][38][39][40]. Even if major differences exist between human and canine mammary carcinomas with regard to the Luminal and HER2-overexpressing subtypes, the bitch has been confirmed as a useful spontaneous model for studying triple negative mammary carcinomas and the prognostic value of molecular subtyping has been demonstrated in both women and female dogs [28,41].
In this study, a high number of HER2-overexpressing carcinomas (24.3%) were detected. There is an ongoing longstanding controversy about the expression of HER2 in CMCs: some studies have reported significant levels of HER2 expression in CMCs [36,42,43], whereas others have questioned the feasibility of detecting HER2 overexpression by IHC [25,28,44].
Few studies have investigated a correlation between ABC-transporters overexpression in the subtypes of breast cancer. Increased expression of BCRP in invasive ductal carcinoma cells and its significant correlation with HER2 expression were found to be strongly correlated with tumor progression, invasion, and metastasis in two studies; no association with PR and ER status was found in one of these studies [45,46]. In vitro studies have shown a more frequent and intense BCRP expression in HER2-enriched mammary cancer cultured cells [47,48]. Other aggressive subtypes such as basal carcinomas had a high expression of BCRP/ABCG2 [6]. Interestingly, it has been suggested that BCRP/ABCG2 may affect the important role of cancer stem cells in drug resistance [49]. P-gp expression has been detected in a high percentage of breast cancers and was found to be increased after exposure to chemotherapeutic drugs (particularly those known to be P-gp substrates), and correlated with a worse response to treatment in both the adjuvant and neoadjuvant settings, but a direct role of P-gp as a cause of clinical drug resistance has not been adequately tested, even in breast cancer [7]. In one study, basallike breast carcinomas were found to bear a higher expression of P-gp associated with the reduction or loss of estrogen receptor, progesterone receptor, and HER2 [50]. The expression of P-gp/MDR1 gene in cancer stem cells was found to be related with the molecular subtypes of breast cancer tissue: basal-like subtype and normal-like subtype had both significantly higher P-gp expression than both Luminal subtypes and HER2 overexpressing subtype, while HER2-overexpressing subtype has shown a significantly higher P-gp/MDR1 expression than Luminal subtypes [51].
At present there are no published studies investigating the expression of P-gp and BCRP in the immunophenotypes of CMCs. This led us to investigate the expression of the two most important ABC transporters that are associated with MDR, in the context of the different CMCs immunophenotypic subtypes. We hypothesized that this information could be relevant for therapeutic implications.
In the present study, a high percentage of CMCs was found to express at least one MDR marker, with half and 72% of carcinomas positive for P-gp and BCRP, respectively. The high intrinsic expression of chemoresistance markers P-gp and BCRP among malignant mammary carcinomas is a consistent finding in both human and canine mammary tumors, being reported in several studies [7,[9][10][11][12]14,15,18]. Intrinsic expression of these membrane pumps in mammary glands, especially at the ductal epithelium, has been related to the physiological activity of this excretory organ and can be retained in neoplastic mammary cells [52,53].
Subgrouping mammary carcinomas into immunophenotypes was found not to be related to the expression of MDR markers in this study. This may be attributed to a loss of significance due to the small number of the caseload or to other factors. The most important can be that the retrospective collection of archived cases has various limits including, in this study, the unreliability of the IHC staining for some delicate antibodies (especially anti-PR) in formalin-fixed and paraffin-embedded tissue, an issue reported in other studies [54,55]. Despite this, an interesting finding in the present study was that all HER2-overexpressing CMCs expressed BCRP, which was consistent with studies regarding breast cancer; it could be related to unfavorable prognostic factors and suggests to administer the therapeutic protocol targeting HER2 avoiding conventional chemotherapeutic drugs that are BCRP substrates [47,48].
No significant association was found between P-gp, BCRP expression, and coexpression and any other variable investigated in this study, including Ki67 proliferation index, death due to the tumor, and histological grade. Histological grade, known to be associated with Ki67 [41], result confirmed also in this investigation, was previously found to be associated with P-gp expression in CMCs [14], even though opposite findings have been published [11]. This discrepancy could be related to the limited numerosity of the caseload in each mentioned study. However, it is likely that no correlation exists between P-gp and BCRP expression and immunophenotype of CMCs and no useful prognostic information can be extrapolated by a sole IHC analysis of their intrinsic expression. The biological significance of MDR associated pumps is better assessed with functionality assays [7]. In fact, de novo or intrinsic MDR occurs in a tumor when ABC transporters are expressed and functional in neoplastic cells before induction chemotherapy, whereas secondary or acquired expression of ABC transporters appear after the first chemotherapy treatment [4]. In this study we have been able to assess only the intrinsic (i.e., basal) expression of P-gp and BCRP in dogs which had not undergone chemotherapy at the time of surgical excision of the tumor. An explanation could be that ABC transporters other than P-gp and BCRP (i.e., MRP1/ABCC1, MRP3/ABCC3, MRP5/ABCC5, MRP6/ABCC6, MRP7/ABCC7 and/or ABCC11/MRP8) may have an important role in drug resistance in CMCs [6,10,19,20].

Conclusions
A relevant number of CMCs bear a high expression of P-gp and BCRP MDR markers, which could have therapeutic and prognostic implications. However, in the present study neither associations nor correlations were discovered between the intrinsic IHC expression of ABC transporters and immunophenotypes of CMCs or their relevance for survival. Nevertheless, BCRP is to a great extent expressed in CMCs and all HER2-overexpressing CMCs expressed at least one of the two chemoresistance markers.
Supplementary Materials: The following are available online at https://www.mdpi.com/2076-261 5/11/3/658/s1. Table S1: Excel file reporting all the data collected for this caseload study, Table S1: Excel file reporting the results of the Allred scoring system for Estrogen and Progesterone receptors, Table S2: Excel file reporting all the data collected for this caseload study. Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Institutional Review Board Statement: This study did not require ethical approval. Data were retrieved from records of client-owned pets. The pets were treated and housed according to standard hospital protocols for management of client-owned pets and no tests or treatments were conducted for research purposes. The tissue samples utilized in this study were the same used for diagnostic purpose.
Data Availability Statement: All data generated or analyzed during this study are included in this published article and its supplementary information files. The raw datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.