Low Stromal Mast Cell Density in Canine Mammary Gland Tumours Predicts a Poor Prognosis

J. Comp. Path. 2020, Vol. 175, 29e38

Available online at www.sciencedirect.com

ScienceDirect www.elsevier.com/locate/jcpa

NEOPLASTIC DISEASE

Low Stromal Mast Cell Density in Canine Mammary Gland Tumours Predicts a Poor Prognosis H. Ariyarathna, N. Thomson, D. Aberdein and J. S. Munday School of Veterinary Science, Massey University, Palmerston North, New Zealand

Summary Tumour histological classification and grade are frequently used to predict the prognosis of canine mammary gland tumours. While these techniques provide some information about tumour behaviour, it is currently difficult to predict which tumours will metastasize. Mast cell density has been shown to predict metastasis in human breast cancer. The present study investigated whether the average mast cell density in 10 high-power (
400) microscopical fields (10 HPFs), evaluated by toluidine blue staining, similarly predicted the behaviour of canine mammary gland tumours. Mast cell density was evaluated in 53 canine mammary neoplasms for which the clinical outcome was known. Stromal mast cell density in malignant tumours that had subsequently developed radiographical evidence of metastasis (n ¼ 21) was significantly lower (P <0.001) than in malignant tumours that did not show evidence of metastases (n ¼ 20) or in benign tumours (n ¼ 12). The density of stromal mast cells that best predicted the disease outcome was #10/10 HPFs. Eighty-one percent of malignant tumours with #10 stromal mast cells/10 HPFs subsequently metastasized, while only 9.5% of malignant tumours with >10 stromal mast cells/10 HPFs developed metastases. There was a positive correlation between stromal mast cell density and survival time (rs ¼ 0.50, P <0.001). These findings suggest that assessing stromal mast cell density using toluidine blue staining may represent an easy to perform and cost-effective histopathological measure that, in conjunction with classification and grading, could better predict the behaviour of canine mammary neoplasms. Ó 2019 Elsevier Ltd. All rights reserved. Keywords: dog; mammary gland tumour; mast cell; toluidine blue stain

Introduction Canine mammary gland tumours (CMGTs) are one of the most common neoplasms of entire female dogs (Sorenmo, 2003; Goldschmidt et al., 2011). These tumours can display a highly variable clinical behaviour that ranges from benign to highly malignant with rapid local and distant metastases (Misdorp and Hart, 1976; Sorenmo, 2003). Due to this variable clinical behaviour, it is important to be able to predict accurately which tumours are more likely to show aggressive clinical behaviour, so that Corresponding author: H. Ariyarathna (e-mail: H.Ariyarathna@massey. ac.nz). 0021-9975/$ - see front matter https://doi.org/10.1016/j.jcpa.2019.12.004

an optimal treatment plan can be devised for each patient. Currently, the histological classification and tumour grade of a CMGT have been shown to predict prognosis (Canadas et al., 2019); however, differentiation between histological subtypes and grades can be subjective, increasing the potential for interpathologist variability (Meuten et al., 2018). Any uncertainty around tumour classification or grade decreases the accuracy of the prognosis. Therefore, the identification of easy to interpret, objective prognostic factors could improve the treatment of these common canine neoplasms. Immune cells infiltrate many different cancer types and, due to their ability to significantly influence the Ó 2019 Elsevier Ltd. All rights reserved.

30

H. Ariyarathna et al.

behaviour of a tumour, have been identified as prognostic markers for many different human cancer types (Faustino-Rocha et al., 2017; Aponte-L
opez et al., 2018). Mast cells (MCs) are frequently present in many different human and animal cancers and numerous studies of human cancers have investigated whether tumour-associated MCs could be used as prognostic markers (Dyduch et al., 2012; Varricchi et al., 2017; Aponte-L
opez et al., 2018). These studies have shown that the presence of MCs within a tumour influences tumour behaviour differently in different cancer types. For example, in gastric, pancreatic and colorectal cancers in people, a high intratumoural MC density correlates with a poor prognosis (Varricchi et al., 2017). Conversely, oral squamous cell carcinomas and ovarian carcinomas with a high MC density have a more favourable prognosis (Attramadal et al., 2016; Varricchi et al., 2017). Furthermore, MCs in renal carcinoma and pulmonary small cell carcinoma do not appear to influence the behaviour of the neoplasms (Varricchi et al., 2017). Considering this seemingly variable role of MCs in different types of human cancers, the ability of MC density to predict prognosis cannot be generalized and appears to be specific for each cancer type (Varricchi et al., 2017). The variable influence of MCs on tumour behaviour has been suggested to be due to the high variety of chemical mediators produced by these cells, which have both proand antitumourigenic properties (Khazaie et al., 2011; Dyduch et al., 2012). The prognostic significance of MCs in human breast cancer is currently not fully resolved (Dabiri et al., 2004; Varricchi et al., 2017), although most studies have revealed that high density of MCs within a breast cancer is associated with a more favourable prognosis (Varricchi et al., 2017). In dogs, four studies have investigated the presence of MCs in CMGTs. One study showed significant differences in MC density between non-neoplastic mammary glands and neoplastic or preneoplastic mammary glands (Sfacteria et al., 2011). In the other three studies, a positive correlation was observed between the MC density and tumour microvessel density (Woldemeskel and Rajeev, 2009; Lavalle et al., 2010; Im et al., 2011). While these previous studies suggested that MCs could have a role in CMGT development and progression, to the authors’ knowledge this is the first time that the presence of MCs has been evaluated as a potential prognostic marker for CMGTs. Therefore, the aim of the present study was to investigate the density of MCs in the peripheral and stromal compartments of benign and malignant CMGTs. As the clinical outcome of each case was

known, the MC densities of the tumours could be correlated with clinical outcome and survival times of the dogs to determine whether MC density was prognostic.

Materials and Methods Case Selection

Histology specimens of CMGTs were identified using the surgical biopsy archive of IDEXX diagnostic laboratories, New Zealand. All neoplasms had been submitted for histopathology between 2012 and 2015, and in all cases, tumour excision had been performed with curative intent rather than for diagnostic purposes. A questionnaire was sent to the submitting veterinarians to obtain the post-surgical disease outcome of the patients. The information requested in the questionnaire included details of pre- and postsurgical clinical examination findings, any additional treatments that had been used, evidence of mammary tumour metastasis, diagnostic methods used to detect tumour metastasis and the cause of death for dogs that died. Cases were excluded if adjunct therapies including anti-inflammatory drugs were used to alter the behaviour of the tumour. The cases were followed for a minimum of 3 years from the date of mammary tumour excision. Disease-specific overall survival time was calculated from the date of tumour excision to the date of the dog’s death or humane destruction due to the mammary tumour metastasis. Survival rates were determined using the number of dogs alive at the end of the follow-up period. Dogs that died or were humanely destroyed due to causes other than mammary tumour metastasis during the follow-up period were excluded from survival rate analysis and were censored at the point of death during the survival time analysis. Histology and Mast Cell Quantification

Sections were prepared from formalin-fixed and paraffin wax-embedded tissue and stained with haematoxylin and eosin (HE) or toluidine blue (0.1% toluidine blue solution in 30% ethanol). The HEstained sections were examined to determine the histological subtypes and grades of the tumours, following the guidelines of Goldschmidt et al. (2011) and Pe~ na et al. (2014), respectively. Mast cell quantification was carried out using toluidine blue-stained sections, using a modification of a previously described method to evaluate MC density in human tissues (Sang et al., 2016). Briefly, tumour peripheral and stromal areas with the highest MC density were identified by scanning the sections at low power

31

Mast Cell Density in Canine Mammary Tumours

(
100 magnification). Tumour periphery was defined as the area at the periphery of the tumour capsule in encapsulated tumours or the area adjacent to the tumour margins in non-encapsulated tumours. Tumour stroma was defined as the inter- and intralobular and interductal regions within the tumour. Individual MCs were then counted in 10 nonoverlapping high-power fields (HPFs) at
400 magnification, where each microscopical field corresponded to an area of 0.785 mm2. This procedure was repeated twice for each tumour and then mean MC density per 10 HPFs was calculated. When non-neoplastic mammary tissues were available in addition to the neoplastic gland, peripheral and stromal MC densities were also assessed in the nonneoplastic tissue following the same method described above. When multiple wax blocks were available for a single large neoplasm, MCs were counted in all sections and the average was taken. Tumours were classified into three groups: ‘malignantemetastatic’ if the neoplasm was confirmed to have metastasized by microscopical examination or believed to have metastasized due to the presence of radiographical lesions consistent with pulmonary metastases; ‘malignantenon-metastatic’ if the neoplasm was classified as malignant using histological criteria, but no clinical evidence of metastases developed during the follow-up period; and ‘benign’ if the neoplasm was histologically consistent with a benign neoplasm. In addition to these three groups, a separate group of non-neoplastic mammary tissue was also included in the analysis. This group was comprised of nonneoplastic mammary gland tissue that had been submitted together with mammary gland neoplasms of some dogs. Statistical Analysis

Peripheral and stromal MC densities of the malignantemetastatic, malignantenon-metastatic, benign tumours and non-neoplastic mammary tissues were compared using the KruskaleWallis H test to identify whether there were any significant differences between groups. When significant differences were identified, post hoc analysis was performed by mean rank test to identify which group or groups were significantly different from the others. The association between stromal or peripheral MC density and survival time was analyzed using Spearman’s rankorder correlation. Other group comparisons were performed using Chi-square or Fisher’s exact tests. Survival times were investigated by KaplaneMeier curves, and significant differences were determined by Log rank test. These statistical analyses were performed using SPSS software (IBM, version 25.0)

and P values <0.05 were considered to indicate significant differences. X-tile software (https://medicine. yale.edu/lab/rimm/research/software.aspx) was used to identify the optimal cut-off point for the stromal MC density that would best predict prognosis in dogs with malignant CMGTs (Camp et al., 2004).

Results Selected Cases

From a total of 100 CMGTs included in the archives, additional information was available for 63 cases. Ten cases were excluded; the cause of death was not known in five dogs, there was insufficient follow-up time in three dogs and two dogs had received antiinflammatory drugs. Therefore, a total of 53 cases were included in the study. Histological Subtypes and Grades

The 53 CMGTs included seven cases from 2012, 13 from 2013, 21 from 2014 and 12 from 2015. Fortyone (77.3%) CMGTs were histologically classified as malignant while 12 (22.7%) were benign. The malignant mammary tumours were classified into nine different histological subtypes with simple carcinomas further subclassified into tubular, tubulopapillary, cribriform and cystic papillary carcinomas (Table 1). Grading of the malignant tumours Table 1 Histological classification of canine mammary gland tumours Histological subtype Malignant tumours Simple carcinoma Tubular Tubulopapillary Cribriform Cystic papillary Intraductal papillary carcinoma Adenosquamous carcinoma Ductal carcinoma Carcinoma e mixed Carcinoma e complex Solid carcinoma Anaplastic carcinoma Comedo carcinoma Benign tumours Complex adenoma Simple adenoma Intraductal papillary adenoma Ductal adenoma Papillary adenoma

10 5 3 1 1 7 6 6 4 3 2 2 1 5 3 2 1 1 53

H. Ariyarathna et al.

32

revealed that 13 were grade I, 23 were grade II and five were grade III (Table 2). Non-neoplastic mammary tissues were present adjacent to the mammary gland tumours in 44 (83%) cases. Of the 53 neoplasms, 21 subsequently metastasized (therefore were classified as malignantemetastatic), 20 were malignant with no clinical evidence of metastasis (malignantenon-metastatic) and 12 were benign. Mast Cell Distribution

In most tumours MCs were scattered along the pericapsular area and throughout the tumour stroma. Focal aggregates of MCs were observed rarely in the neoplastic or non-neoplastic mammary tissues. In malignant CMGTs, MCs were frequently found in the stromal and peripheral compartments of malignantenon-metastatic tumours, while they were scarce in malignantemetastatic tumours (Figs. 1 and 2). The lowest mean stromal MC density (MCs/10 HPFs) was observed in malignantemetastatic tumours (20.3  37.4) followed by malignantenonmetastatic tumours (94.3  72.7), benign tumours (108  70.6) and non-neoplastic mammary tissues (115  51.3; Table 3 and Fig. 3), with significant differences in stromal MC density identified between the groups (Z ¼ 38.2, P <0.001, KruskaleWallis H test). Post hoc analysis revealed that mean rank stromal MC density in the malignantemetastatic CMGTs was significantly lower than the MC density in the other groups (17.8 versus 51.8, 57.4 and 55.2, mean rank test). In contrast, there was no difference in peripheral mast cell density between the four groups (Z ¼ 2.7, P ¼ 0.45, KruskaleWallis H test). Cut-Off Analysis

By X-tile analysis, a strong, direct and continuous association between stromal MC density and survival times of dogs with malignant CMGTs was identified. Furthermore, an optimal cut-off was identified as 10 MCs/10 HPFs (c2 ¼ 20.13), which best predicted the disease outcome. Therefore, a low MC density was defined as #10 MCs/10 HPFs while a high stromal MC density was defined as >10 MCs/10 HPFs. Table 2 Histological grades of malignant canine mammary gland tumours Histological grade Grade I Grade II Grade III Total

Total 13 23 5 41

Fig. 1. A grade II simple carcinoma. Note the abundant mast cells visible within the stroma (arrows) in this tumour, which did not develop subsequent metastases during the follow-up period. Toluidine blue.

Fig. 2. A grade II simple carcinoma. Note the lack of mast cells within the stroma of this tumour, which metastasized during the follow-up period. Toluidine blue.

Risk of Metastasis

In this study, 21/53 (39.6%) of dogs with malignant mammary tumours died after developing evidence of CMGT metastasis. In two of the 21 dogs, tumour metastasis was confirmed by post-mortem examination. Histology from one of these dogs revealed variably sized clusters of anaplastic epithelial cells arranged in glands within sections of lung, spleen, heart and regional lymph nodes. Similar neoplastic cells were visible within sections of lung in the other dog in which necropsy examination was performed. In a further two of the 21 dogs in which metastasis was diagnosed, the diagnosis was confirmed by cytology of fine needle aspirates (FNAs) of pulmonary

Mast Cell Density in Canine Mammary Tumours Table 3 Peripheral and stromal MC densities in mammary gland tumours and non-neoplastic mammary tissue Number Peripheral mast cell density Malignantemetastatic 21 Malignantenon20 metastatic Benign 12 Non-neoplastic 44 mammary tissue Stromal mast cell density Malignantemetastatic 21 Malignantenon20 metastatic Benign 12 Non-neoplastic 44 mammary tissue

Average

Z

P value

129.3  81.5 144.8  49.1

2.7

0.45

38.2

<0.001

129.8  37.7 122.4  42.6 20.3  37.4 94.3  72.7 108  70.6 115  51.3

masses identified in thoracic radiographs. Cytology from these cases revealed a population of anaplastic epithelial cells consistent with a malignant epithelial neoplasm. In the other 17 dogs in which metastasis was diagnosed, this diagnosis was made based on thoracic radiographical findings that revealed the presence of multiple solid masses in the pulmonary parenchyma accompanying an interstitial pattern. While these were not confirmed to be metastases microscopically, the radiographical findings in this case were consistent with metastases and these neoplasms were classified as malignantemetastatic in this study. Five dogs (9.4%) died due to unrelated causes within the 3-year follow-up period while 27

33

(51%) dogs survived until the end of the follow-up period without any clinical evidence of mammary tumour metastasis. The percentage deaths due to tumour metastasis 2 and 3 years after surgical excision of a malignant CMGT were 41.6% (15/36) and 58% (21/36), respectively. Rates of tumour metastasis were not significantly different between different histological subtypes (P ¼ 0.13, Fisher’s exact test) or between different histological grades (P ¼ 0.29, Chisquare test). Using the optimal cut-off from Xtile analysis, 16 (43%) malignant mammary tumours had a low stromal MC density while 21 (57%) had a high stromal MC density. Thirteen of 16 (81%) dogs with malignant CMGTs with a low stromal MC density died of tumour metastasis during the follow-up period, while just 2 (9.5%) of the 21 dogs that had malignant CMGTs with high stromal MC density died due to tumour metastasis. Therefore, in this study, dogs with malignant CMGTs with low stromal MC density were eight times more likely to die due to neoplasm metastasis than dogs with malignant CMGTs with a high stromal MC density. Survival Time Analysis

The overall mean survival time (MST) of the 41 dogs with malignant CMGTs was 721 days (95% CI 609e833). There was a significant, moderate positive correlation between the stromal MC density and

Fig. 3. Boxplot of stromal and peripheral mast cell densities in different canine mammary gland tumour categories. Note the significantly low stromal mast cell density in malignantemetastatic tumours (P <0.001). The box represents the first to third quartiles with the median indicated by the horizontal line. The vertical lines indicate the minimum and maximum values. Outliers are indicated by a circle, or an asterisk for extreme outliers greater than three times the interquartile range from first or third quartile.

34

H. Ariyarathna et al.

survival time (rs ¼ 0.50, P <0.001, Spearman’s rankorder correlation), but no correlation between peripheral MC density and survival time (rs ¼ 0.13, P ¼ 0.41, Spearman’s rank-order correlation). As part of the cut-off analysis using X-tile software, the survival times between dogs with malignant mammary tumours with low and high MC densities were compared. Dogs that had malignant mammary tumours with a low stromal MC density had a significantly shorter survival time (497 days 95% CI 342e651) than dogs with CMGTs that had a high stromal MC density (973 days, 95% CI 879e1,068, P <0.001, X-tile analysis, Fig. 4). There were no significant differences between the MST of dogs with CMGTs of different histological subtypes (P ¼ 0.08, Log rank test, Table 4). However, there were significant differences in the MST of dogs with CMGTs of different grades (P ¼ 0.001, Log rank test, Table 4). Post hoc analysis showed that the MST of dogs with grade I tumours (971 days, 95% CI 812e1,130) was significantly longer than the MST of dogs with grade II (729 days, 95% CI 585e872, p ¼ 0.03) and grade III tumours (429 days, 95% CI 201e656, P < 0.001). Furthermore, the MST of dogs with grade II tumours was significantly longer than that of dogs with grade III tumours (P ¼ 0.04). To exclude the possible confounding effect of comparing grade I and grade III tumours, which have been shown previously to have different survival times, only grade II tumours (n ¼ 23) were

Table 4 Survival times of dogs with mammary gland tumours Number

Histological type Total Simple carcinoma Intraductal papillary carcinoma Adenosquamous carcinoma Ductal carcinoma Histological grade Total Grade I Grade II

Estimated mean survival time (95% CI) days

29 10 7

809 (565e1,053) 1040 (943e1,138)

6

610 (427e794)

6

764 (457e1,071)

41 13 23

971 (812e1,130) 729 (585e872)

Grade III 5 429 (201e656) Mast cell density in malignant tumours Stromal MCD #10/10 HPFs 16 497 (342e651) 10/10 HPFs 25 973 (879e1,068) Mast cell density in grade II carcinomas Total 23 Stromal MCD #10/10 HPFs 9 449 (309e590) >10/10 HPFs

14

P value

0.08 (Log-rank test)

0.001 (Log-rank test)

<0.001 (X-tile analysis)

<0.001 (Logrank test)

933 (785e1,081)

used to evaluate the association between MST and MC density. This revealed that the MST of the nine dogs with low stromal MC density was 449 days (95% CI 309e590), which was

Fig. 4. KaplaneMeier survival curve of the dogs with malignant canine mammary gland tumours with low and high stromal mast cell densities. SMCD-L, stromal mast cell density low (#10 stromal MCs/10 HPFs); SMCD-H, stromal mast cell density high (>10 stromal MCs/10 HPFs). Mean survival times between dogs with malignant mammary tumours with low stromal mast cell density was significantly different from high stromal mast cell density (P <0.001).

Mast Cell Density in Canine Mammary Tumours

significantly shorter than the MST of the 14 dogs with high stromal MC density (933 days, 95% CI 785e1,081, P <0.05, Log rank test).

Discussion In the present study, the behaviour of the CMGTs could be predicted by the density of the MCs in the tumour stroma. Stromal MC density was significantly lower in malignant tumours where the patient later developed metastasis, compared with malignant tumours that did not subsequently metastasize during the follow-up period. Additionally, there was a significant correlation between stromal MC density and the survival time of dogs with malignant mammary gland tumours. Furthermore, dogs with malignant CMGTs classified as having low stromal MC density were eight times more likely to die due to tumour metastasis than dogs with malignant CMGTs with a high stromal MC density. These results therefore suggest that stromal MC density is an important prognostic indicator for CMGTs. Similar to the results in the present study of CMGTs, MC density has been found to be prognostic in the majority of human breast cancer studies (Aaltomaa et al., 1993; Amini et al., 2007; Rajput et al., 2008; Glajcar et al., 2017; Jana et al., 2017). Interestingly, while most studies of human breast cancer have reported that a high stromal MC density was indicative of favourable prognosis, a small number of studies have associated high stromal MC density with an unfavourable prognosis (Kanakkunen et al., 1997; Ranieri et al., 2009; Keser et al., 2017). These apparently contradictory results suggest that stromal MCs may influence tumour behaviour differently in some circumstances. Three studies have investigated whether stromal MC density in breast cancer is associated with other molecular prognostic factors including the presence of oestrogen or progesterone hormone receptors or human epidermal growth factor receptor 2 (Amini et al., 2007; Sang et al., 2016; Glajcar et al., 2017). These studies produced conflicting results and it is currently unclear whether or not stromal MC density is a prognostic factor that is independent of the other molecular factors currently recognized as prognostic for human breast cancer. There are currently no studies investigating an association of hormone or growth factor receptor expression and MC density in canine tumours. The mean overall survival time of the dogs with malignant CMGTs included in this study was 721 days. This was consistent with the MSTs reported by previous studies on these neoplasms, which ranged between 359 and 720 days (Santos et al., 2013;

35

Rasotto et al., 2017; Canadas et al., 2019). Previous studies have reported that 56e80% of dogs with malignant CMGTs survive for at least 2 years after neoplasm excision (Hellm
en et al., 1993; Karayannopoulou et al., 2005; Sassi et al., 2010; Pe~ na et al., 2013). Similarly, the 2-year survival rate in the present study was 59%. The similarities in survival times and survival rates between the present and previous studies of CMGTs suggest that the 41 malignant neoplasms included in the present study were representative of malignant CMGTs in the wider population of dogs. Therefore, although the present study contained comparatively small numbers of CMGTs, it is expected that stromal MC density will be also associated with CMGT behaviour in larger samples of these tumours. As has been reported previously (Vinothini et al., 2009; Rasotto et al., 2012, 2017; Santos et al., 2013; Cassali et al., 2014; Gundim et al., 2016; Canadas et al., 2019) tumour grade was predictive of survival time of dogs with CMGTs in the present study. It was hypothesized that stromal MC density could have been predictive simply because CMGTs of a higher grade were more likely to have a lower MC density. However, as stromal density remained prognostic when CMGTs of a single grade were evaluated, it appears unlikely that stromal MC density is simply dependent on the tumour grade. Further studies with sufficiently large numbers of CMGTs in each grade are necessary to fully determine the relationship between stromal MC density and tumour grade. In the present study, survival times were not significantly different between different histological subtypes of mammary tumours. This was probably due to the small numbers of some subtypes included in this study. Furthermore, the small numbers of some subtypes meant that it could not be determined whether stromal MC density is an independent prognostic factor to histological classification. Further studies with sufficiently large numbers of tumours in each histological subtype are necessary to investigate this matter. Currently, tumour grade and histological classification are recommended for prognostic determination in CMGTs. However, the subjectivity of tumour classification and grading has been identified as a disadvantage due to high interobserver variability (Chu et al., 2011). Therefore, histological classification and grading need to be complemented with other reliable histochemical or molecular methods to improve the prognostic accuracy of CMGTs. Advantages of assessing stromal MC density include the low cost of toluidine blue-stained sections, the ease of recognizing mast cells within the toluidine bluestained sections and the ability to objectively count

36

H. Ariyarathna et al.

the numbers of cells in histological fields. Due to the ease of identification of MCs, using these cells to predict prognosis would appear to be readily adaptable to automated counting of the cells within a histological field. The present study revealed a continuous positive correlation between stromal MC density and survival times of dogs with malignant CMGTs. Despite the continuous nature of the association, it is practical to have cut-offs that can be communicated to clients in a commercial setting. In this study, #10/10 HPFs was identified as the optimal cut-off for stromal MC density with 81% of the dogs with malignant CMGTs with a stromal MC density #10/10 HPFs developing tumour metastasis, while metastasis developed in only 9.5% of the dogs with malignant CMGTs that had a mast cell density of >10/10 HPFs. Therefore, determining whether a malignant CMGT has more or less than 10 stromal MCs/10 HPFs would be comparatively easy for pathologists, but appears to be a powerful predictor of prognosis and therefore which dogs are most likely to benefit from post-surgical adjuvant therapy to prevent subsequent tumour metastasis. The toluidine blue staining method used in the present study is relatively rapid and inexpensive compared with the other immunohistochemical or molecular prognostic methods described in previously studies (Pe~ na et al., 1998, 2014; Queiroga et al., 2011; Santos et al., 2013). Moreover, this method is performed routinely and does not need special technical expertise to perform. Considering the simplicity of identifying MCs, the present study suggests that measuring stromal MC density to determine prognosis could be easily incorporated into the routine assessment of CMGTs. The results of the present study revealed that stromal MC density predicted the biological behaviour of CMGTs. However, it is unknown whether the stromal MC density directly influences tumour metastasis or whether both the MC density and the behaviour of the neoplasm are determined by the properties of the neoplastic cells. If MCs influence tumour behaviour, they could do it by producing antitumour compounds that prevent tumour metastasis (Theoharides and Conti, 2004; Faustino-Rocha et al., 2017; Varricchi et al., 2017; Aponte-L
opez et al., 2018). For example, chondroitin sulphate secreted by MCs may increase adhesion between tumour cells and the extracellular matrix and therefore inhibit tumour metastasis (Faustino-Rocha et al., 2017). In addition, heparan sulphate proteoglycans secreted by MCs inhibit neovascularization in tumours, minimizing the possibility of tumour metastasis (Theoharides and Conti, 2004).

In the present study, peripheral MC density was not associated with disease outcome. It is not certain what mechanisms operate differently between the tumour stromal and peripheral compartments to produce this discrepancy. However, stromal MCs are located within the tumour and therefore more closely associated with the tumour cells and tumour microenvironment than the MCs scattered along the tumour periphery. Typically, the disease outcome of a tumour is determined by the properties of tumour cells and tumour microenvironment (Quail and Joyce, 2013). Therefore, stromal MCs which are in close association with tumour cells and tumour microenvironment are more likely to influence tumour behaviour and be prognostic of the disease outcome than peripheral MCs. Only two previous studies have investigated the MC density within normal or neoplastic canine mammary glands. In contrast to the present study, MC density of non-neoplastic mammary tissues in both previous studies was lower than that of neoplastic mammary gland (Im et al., 2007; Sfacteria et al., 2011). The reasons for this difference is unclear, although both previous studies contained small numbers of samples and only determined the overall MC density, rather than distinguishing between peripheral and stromal compartments. These findings suggest that measuring stromal MC density using toluidine blue staining may represent an easy to perform and cost-effective histopathological parameter that, in conjunction with classification and grading, could better predict the behaviour of canine mammary neoplasms.

Acknowledgments The authors thank the veterinarians who participated in the survey, together with IDEXX laboratories and the technicians in the histology laboratory at the School of Veterinary Science, Massey University, New Zealand. The current study was funded by a Massey University Postgraduate Research Grant (grant number 20195) and a Massey University Foundation Grant (grant number 21307).

Conflict of Interest Statement The authors declare no conflicts of interest with respect to the research, authorship or publication of this article.

References Aaltomaa S, Lipponen P, Papinaho S, Kosma VM (1993) Mast cells in breast cancer. Anticancer Research, 13, 785e788.

Mast Cell Density in Canine Mammary Tumours

Amini R-M, Aaltonen K, Nevanlinna H, Carvalho R, Salonen L et al. (2007) Mast cells and eosinophils in invasive breast carcinoma. BMC Cancer, 7, 165. Aponte-L
opez A, Fuentes-Panan
a EM, CortesMu~ noz D, Mu~ noz-Cruz S (2018) Mast cell, the neglected member of the tumor microenvironment: role in breast cancer. Journal of Immunological Research, 2018, 9627806. Attramadal CG, Kumar S, Gao J, Boysen ME, Halstensen TS et al. (2016) Low mast cell density predicts poor prognosis in oral squamous cell carcinoma and reduces survival in head and neck squamous cell carcinoma. Anticancer Research, 36, 5499e5506. Camp R, Dolled-Filhart M, Rimm D (2004) X-tile: a new bioinfomatic tool for biomarker assessment and outcome-based cut-point optimization. Clinical Cancer Research, 10, 7252e7259. Canadas A, Franc¸a M, Pereira CVilaca R, Hugo V et al. (2019) Canine mammary tumors: comparison of classification and grading methods in a survival study. Veterinary Pathology, 56, 208e219. Cassali GD, Lavalle GE, Ferreira E, Estrela-Lima A (2014) Consensus for the diagnosis, prognosis and treatment of canine mammary tumors - 2013. Brazilian Journal of Veterinary Pathology, 7, 38e69. Chu P-Y, Liao AT, Liu C-H (2011) Interobserver variation in the morphopathological diagnosis of canine mammary tumors among veterinary pathologists. International Journal of Applied Research in Veterinary Medicine, 9, 388e391. Dabiri S, Huntsman D, Makretsov N, Cheang M, Gilks B et al. (2004) The presence of stromal mast cells identifies a subset of invasive breast cancers with a favorable prognosis. Modern Pathology, 17, 690e695. Dyduch G, Kaczmarczyk K, Oko
n K (2012) Mast cells and cancer: enemies or allies? Polish Journal of Pathology, 63, 1e7. Faustino-Rocha AI, Gama A, Neuparth MJ, Oliveira PA, Ferreira R et al. (2017) Mast cells in mammary carcinogenesis: host or tumor supporters? Anticancer Research, 37, 1013e1021. Glajcar A, Szpor J, Pacek A, Tyrak KE, Chan F et al. (2017) The relationship between breast cancer molecular subtypes and mast cell populations in tumor microenvironment. Virchows Archiv, 470, 505e515. Goldschmidt M, Pe~ na L, Rasotto R, Zappulli V (2011) Classification and grading of canine mammary tumors. Veterinary Pathology, 48, 117e131. Gundim LF, De Araujo CP, Blanca WT, Guimar~ aes EC, Medeiros AA (2016) Clinical staging in bitches with mammary tumors: influence of type and histological grade. Canadian Journal of Veterinary Research, 80, 318e322. Hellm
en E, Bergstr€om R, Holmberg L, Sp angberg IB, Hansson K et al. (1993) Prognostic factors in canine mammary tumors: a multivariate study of 202 consecutive cases. Veterinary Pathology, 30, 20e27. Im K-S, Kim J-H, Yhee J-Y (2011) Tryptase-positive mast cells correlate with angiogenesis in canine mammary

37

carcinoma. Journal of Comparative Pathology, 144, 157e163. Jana S, Ghosh S, De A, Pal S, Sengupta S, Ghosh T (2017) Quantitative analysis and comparison of mast cells in breast carcinomas and axillary lymph nodes. Clinical Cancer Investigation Journal, 6, 214e218. Kanakkunen J-P, Harvima IT, Naukkarinen A (1997) Quantitative analysis of tryptase and chymase containing mast cells in benign and malignant breast tissues. International Journal of Cancer, 72, 385e388. Karayannopoulou M, Kaldrymidou E, Constantinidis T, Dessiris A (2005) Histological grading and prognosis in dogs with mammary carcinomas: application of a human grading method. Journal of Comparative Pathology, 133, 246e252. Keser SH, Kandemir NO, Ece D, Gecmen GG, Gul AE et al. (2017) Relationship of mast cell density with lymphangiogenesis and prognostic parameters in breast carcinoma. The Kaohsiung Journal of Medical Sciences, 33, 171e180. Khazaie K, Blatner NR, Khan MW, Gounari S, Gounaris E et al. (2011) The significant role of mast cells in cancer. Cancer and Metastasis Reviews, 30, 45e60. Lavalle G, Bertagnolli A, Tavares W, Ferreira MAND, Cassali GD (2010) Mast cells and angiogenesis in canine mammary tumor. Arquivo Brasileiro de Medicina Veterin
aria e Zootecnia, 62, 1348e1351. Meuten D, Munday JS, Hauck M (2018) Time to standardize? Time to validate? Veterinary Pathology, 55, 195e199. Misdorp W, Hart A (1976) Prognostic factors in canine mammary cancer. Journal of the National Cancer Institute, 56, 779e786. Pe~ na L, Andr
es PD, Clemente M, Cuesta P, P
erezAlenza MD (2013) Prognostic value of histological grading in non-inflammatory canine mammary carcinomas in a prospective study with two-year follow-up: relationship with clinical and histological characteristics. Veterinary Pathology, 50, 94e105. Pe~ na L, Gama A, Goldschmidt M, Abadie J, Benazzi C et al. (2014) Canine mammary tumors: a review and consensus of standard guidelines on epithelial and myoepithelial phenotype markers, HER2, and hormone receptor assessment using immunohistochemistry. Veterinary Pathology, 51, 127e145. Pe~ na LL, Nieto AI, P
erez-Alenza D, Cuesta P, Casta~ no M (1998) Immunohistochemical detection of Ki-67 and PCNA in canine mammary tumors: relationship to clinical and pathologic variables. Journal of Veterinary Diagnostic Investigation, 10, 237e246. Quail DF, Joyce JA (2013) Microenviornmental regulation of tumor progression and metastasis. Nature Medicine, 19, 1423e1437. Queiroga FL, Pires I, Parente M, Greg
orio H, Lopes CS (2011) COX-2 over-expression correlates with VEGF and tumour angiogenesis in canine mammary cancer. Veterinary Journal, 189, 77e82. Rajput AB, Turbin DA, Cheang MC, Voduc DK, Leung S et al. (2008) Stromal mast cells in invasive breast cancer

38

H. Ariyarathna et al.

are a marker of favourable prognosis: a study of 4,444 cases. Breast Cancer Research, 107, 249e257. Ranieri G, Ammendola M, Patruno R, Celano G, Zito FA et al. (2009) Tryptase-positive mast cells correlate with angiogenesis in early breast cancer patients. International Journal of Oncology, 35, 115e120. Rasotto R, Berlato D, Goldschmidt MH, Zappulli V (2017) Prognostic significance of canine mammary tumor histologic subtypes: an observational cohort study of 229 cases. Veterinary Pathology, 54, 571e578. Rasotto R, Zappulli V, Castagnaro M, Goldschmidt MH (2012) A retrospective study of those histopathologic parameters predictive of invasion of the lymphatic system by canine mammary carcinomas. Veterinary Pathology, 49, 330e340. Sang J, Yi D, Tang X, Zhang Y, Huang T (2016) The associations between mast cell infiltration, clinical features and molecular types of invasive breast cancer. Oncotarget, 7, 81661. Santos AA, Lopes CC, Ribeiro JR, Martins LR, Santos JC et al. (2013) Identification of prognostic factors in canine mammary malignant tumours: a multivariable survival study. BMC Veterinary Research, 9, 1. Sassi F, Benazzi C, Castellani G, Sarli G (2010) Molecularbased tumour subtypes of canine mammary carcinomas assessed by immunohistochemistry. BMC Veterinary Research, 6, 5.

Sfacteria A, Lanteri G, Grasso G, Marci B (2011) Mast cells in canine mammary gland tumour: number, distribution and EPOR positivity. Veterinary and Comparative Oncology, 9, 310e315. Sorenmo K (2003) Canine mammary gland tumors. Veterinary Clinics of North America: Small Animal Practice, 33, 573e596. Theoharides TC, Conti P (2004) Mast cells: the Jekyll and Hyde of tumor growth. Trends in Immunology, 25, 235e241. Varricchi G, Galdiero MR, Loffredo S, Marone G, Iannone R et al. (2017) Are mast cells MASTers in cancer? Frontiers of Immunology, 8, 424. Vinothini G, Balachandran C, Nagini S (2009) Evaluation of molecular markers in canine mammary tumors: correlation with histological grading. Oncolology Research, 18, 193e201. Woldemeskel M, Rajeev. (2009) Mast cells in canine hemangioma, hemangiosarcoma and mammary tumors. Veterinary Research Communications, 34, 153e160.

September 10th, 2019 ½ Received, Accepted, December 17th, 2019