The availability of predictive preclinical animal models for human breast tumours represents a major challenge in breast cancer research. In viv
o mouse models such as xenografts and transgenics, although widely used, have been demonstrated to fail in recapitulating essential features of human breast cancers such as heterogeneity, tumour microenvironment and dependence on steroid hormones [1
]. Besides the intrinsic evolutionary distance between mice and humans, additional differences can originate from induced genetic modifications (transgenic mice) or from the altered presence of adjacent normal tissue, stromal cells, vasculature and immune system components (xenografts) [3
]. Together these factors translate into a limited value of these mouse models for the study of cancer pathogenesis, progression and therapy, and represent a major obstacle to the identification of reliable predictive molecular biomarkers and the development of effective therapeutic agents [1
However, during the last few years, along with the sequencing of the entire dog genome (99% complete, ~2.5 billion base pairs) and the clear evidence of its close similarity with the human genome [9
], the dog has emerged as an attractive alternative model for cancer research [9
]. For many gene families and in particular those associated with cancer, the similarities between dog and human gene sequences have been found to be much closer than the respective counterparts in mouse [10
]. Molecular cytogenetic analysis of canine tumour cells derived from haematological malignancies revealed the retention of ancestral chromosomal aberrations in comparable cancers of human and dog [11
]. In mammary carcinomas, altered expression of the ERBB2
genes were observed to be similar in the two species, suggesting similar roles in carcinogenesis and potential use as prognostic indicators [13
]. It was also observed that similar mutations in oncogenes can result in different cancers in humans and dogs. For example, similar mutations in KIT
, a tyrosine kinase growth factor receptor, have been identified in both human gastrointestinal stromal tumours (GIST) and dog mast-cell cancers [16
]. Moreover, comparative histological analyses indicate that the intratumoural (cell-to-cell) heterogeneity observed in human breast tumours also occurs in the cognate dog tumors [17
]. The natural consequences of this heterogeneity underlie the deadly features of human cancers, namely acquired resistance to therapy, recurrence and metastasis.
Additional and more general considerations have contributed to the increased interest in the dog as a preclinical model. Dogs have a large body size and are relatively outbred as compared to other laboratory animals, therefore providing a genetic diversity similar to that seen in humans [9
]. Spontaneous cancers in dogs develop in the context of a natural immune system where the tumor and its microenvironment are syngeneic. Unlike mice, dogs share a common environment with humans and are exposed to some of the same carcinogens [18
]. Moreover, compared to humans, the shorter life span of dogs facilitates the study of mammary tumours that develop after a few years instead of decades. Finally, as in humans, progesterone treatment, advancing age, obesity and diet, represent risk factors for mammary tumour development in dogs [9
In this study we present the first genome-wide analysis of transcriptional changes occurring in the mammary tumours of dog, including a comparative analysis with respect to human breast tumors. Starting from independent human and dog microarray studies, changes in gene expression levels were compared following the mapping of the orthologous genes represented across both array platforms. By unsupervised analysis, we show that clustering is predominantly driven by the origin of the samples (tumour/normal) rather than by species (dog/human), indicating a close overall transcriptional similarity between tumours in both systems. A significant number of genes differentially expressed in human breast tumours, compared to normal human samples, were also found deregulated in the canine model. Moreover, a pathway-focused analysis of these genes revealed a large degree of similarity in the up- and down-regulation of several cancer related pathways.
We have also addressed the comparison of human and dog tumours from a more systemic perspective, by exploring the networks of transcriptional relationships existing between different gene signatures. Various prognostic and oncogenic signatures, derived from independent human breast cancer studies, were collected from the literature and their expression relationships were examined in human breast tumours. We show that many of these signatures, although developed within different experimental settings and contexts, exhibit coordinated patterns of expression. By performing the same analysis on the dog tumour samples, we observed that these relationships are largely maintained, thus suggesting a close interspecies similarity in the network of cancer signalling circuitries governing the establishment and the progression of the tumour.
Collectively our data confirm and strengthen the value of the dog as a suitable model for studying breast cancer, including the development of prognostic molecular biomarkers and the evaluation of novel cancer therapeutics.