Breast cancer subtypes, mouse models, and microarrays

Breast cancer can no longer be viewed as a single disease. Molecular profiling studies have altered the way we consider breast cancer, showing us that there are several subtypes, each with their own unique biology. There are many model systems available in which to study breast cancer, however, each of these comes with advantages and disadvantages. We chose the mouse as a model to investigate breast cancer biology because it gives us the ability to study tumor progression and response to therapy in vivo. Numerous mouse models of breast carcinomas have been developed. The extent to which any faithfully represent clinically significant human phenotypes was unknown. Analogous to our human studies, we characterized mammary tumor gene expression profiles from a large number of murine models using DNA microarrays and compared the resulting data to our human breast tumor dataset. Two major applications of across-species tumor comparisons surfaced from these studies. First, we were able to determine that mouse models contain many of the global characteristics of particular classes or subtypes of human tumors. This included basal versus luminal distinctions, a proliferation/cell cycle signature, and a fibroblast signature. Second, the mouse models were able to inform the human disease; for example, we identified an amplicon that included the K-ras gene present in both mouse and human basal tumors. The high proliferation seen in common between mouse models of Rb loss and human basal-like breast tumors hinted that there is an Rb defect in this human subtype. And finally the mouse spindloid tumors shared significant gene overlap with a new molecular subtype of breast cancer. Although no single murine model recapitulated all the expression features of a given human subtype, these shared expression features have provided us a common framework so that we can now integrate these murine mammary tumor models into our studies of human breast cancer.