Canine mammary carcinomas can become fatal due to the development of distant metastases. One of the most important prognostic factors in the diagnosis is the acknowledgment of metastases to the regional lymph node that represents an early step in metastatic spread
. Klopfleisch et al.
 and Lu et al.
 demonstrated that metastatic spread of canine mammary tumor to the lymph nodes is associated with a gene expression profile of increased cell cycle progression, altered cell differentiation and decreased growth factor signaling. Metastasis development is a complex process involving invasion, intravasation, survival in the bloodstream, extravasation and homing and proliferation at the site of metastasis
. Although some phenotypes showed greater aggressiveness and metastatic capability, only a selected subpopulation was able to metastatize in the multiple and heterogeneous tumor cell population. In this case the phenotype may have been transient and these selected cells have had an intrinsic program to transition to a phenotype enhancing their ability for heterotypic interaction and survival proliferation in distant organs
 as Darwinian clonal evolution. Conversely, the metastatic process has also been described as a stochastic event, the primary tumor cells having equal metastatic capability, characterized by a phenotypic overlap between the primary tumor and its metastases
[33, 34]. Thus the identification of molecular phenotypes in primary tumors and metastases can provide predictive information on the most likely metastatic profile, not the condition in the primary tumor.
Sassi et al.
 identified three phenotypes out of the four detected by Gama et al.
, demonstrating that basal-like subtypes were associated with a better outcome than luminal A and luminal B tumors, in contrast with the findings of Gama et al.
. The prognostic role of c-erbB-2 overexpression remains controversial despite what is known in human medicine. According to a study by Hsu et al.
, the relationship between the clinical course and protein expression of c-erbB-2 in dogs with malignant mammary neoplasia indicated a greater survival rate in tumors overexpressing c-erbB-2 compared to those having non-overexpressed levels of antigen. Certainly, c-erbB-2 plays an important role in carcinogenesis, but does not seem to be directly correlated with progression to malignancy
. In the present investigation it seems that luminal A or B concordance should be considered a positive prognostic factor, whereas concordance for the other molecular types or discordance should not, even if these results await confirmation in a larger number of cases and proper statistical analysis.
This study revealed four out of the five protein expression phenotypes of breast cancer in primary tumors (20 cases): luminal A (40%), luminal B (35%), c-erbB-2 overexpressing (10%) and basal-like (15%). The prevalence of luminal phenotypes (75%) over the others (25%) is in accordance with findings both in human
[11, 36, 37] and veterinary
[8, 9] medicine.
Based on the present study and in agreement with Brunetti et al.
, labeling for CK and p63 would only appear necessary when a tumor is negative for ER, PR and c-erbB-2.
With regard to luminal A and B phenotypes, the expression profiles of ER and PR are essential to decide on the application of endocrine therapy
 in breast cancer and canine mammary neoplasia, and also seem to play a minor role in predicting tumor biological behaviour
[40, 41]. Wu et al.’s study
 in breast cancer confirmed the observation that ER and/or PR could be lost when carcinomas metastasizes, thereby resisting endocrine therapy. The present study shows almost overlapping results, losing hormone receptors by moving from luminal A to basal-like (2 cases) and/or to normal-like (1 case) phenotypes and from luminal B to c-erbB-2 overexpressing (2 case) and/or to basal-like (1 case), and/or to normal-like (1 case), confirming that the gene expression profile in canine mammary tumors may prove a helpful tool in clinical practice. Chang et al.
 indicated that the ER or PR expression in dogs was associated with tumor size, clinical stage, and lymph node metastasis or distant metastases. Dogs with malignant mammary neoplasia and expression of both ER and PR had a longer survival rate than dogs with malignant mammary tumors that were ER positive but PR negative. This latest information on PR suggested that the receptor was a better outcome predictor than ER status alone and that its positive or negative expression could serve as a prognostic factor, especially in dogs with malignant neoplasia with ER expressed
. The present results disclosed a high prevalence of hormone receptor expression in the primary tumor, whose positivity was ensured by reactivity to least one of the two markers (ER and/or PR) (Table
3). The latest results indicate that there are grounds for the use of anti-hormone therapy in dogs, administering molecules other than those hitherto used in veterinary medicine (tamoxifen) as their side-effects are already well-known
. A similar analysis in lymph node showed a net loss of hormone receptor expression, namely ER. ER loss is a known adverse prognostic factor
, and therefore its lack of expression in metastases is indicative. In this study, only five out of the 20 cases showed positivity to both ER and PR in the primary tumor, with persistent positivity in the lymph node metastasis in only two cases. The remaining three cases showed loss of one or both hormone receptor staining in the lymph node metastasis: loss of ER (case n° 1), loss of PR (case n° 14) and concomitant loss of ER and PR (case n° 9). According to the literature, these cases should have a poor prognosis, justified by our phenotypes, but with maintained luminality in the first two cases, and a shift to c-erbB-2 overexpression in the third meaning a even worse evolution by the complete loss of ER and PR.
Interestingly, cases n° 3 and n° 13 (Table
3) in which in lymph node metastasis occurred, lost all the markers expressed in the primary tumor, likely due to the selection of a significantly aggressive cell subpopulation. The normal-like or multiple markers negative (MMN) subtype tumors have been shown to be negative for basal markers, such as CK5/6 and CK14 in our cases, and negative for other molecular markers. The majority of normal-like subtype tumors express CK8/18, CK19 (used in this study), with an absence of CK5/6 suggesting that these cells were most probably derived from a luminal gland cell
. The normal-like subtype is also included in the triple negative breast cancer group (TNBC) characterized by an aggressive clinical course, poor survival rate and, unlike the overexpressing hormone receptors or c-erbB-2-overexpressing tumors, is not amenable to hormone therapy or c-erbB-2-directed agents
. Although no correlation has been found between histological type and phenotype, the normal-like cases represent an exception.
The present study identified five phenotypes in the lymph node metastases: luminal A (25%), luminal B (15%), c-erbB-2 overexpressing (20%), basal-like (30%) and normal-like (10%). The novel aspect of this study is the evaluation of the lymph node metastasis phenotypes and their correlation with the primary tumor, never hitherto applied to canine species. The relationship between the primary tumor and metastatic phenotype is defined by a concordance in 65% of cases and a discordance in the remaining 35%, suggesting the two main metastatic capability theories coexist. All seven discordant cases showed a progressive behavior, according to the prognostic value of molecular phenotypes reported by Gama et al.
, suggesting phenotypic evolution with a worse prognosis from the primary tumor to lymph node metastasis.