Breast Cancer Risk Associated with High Mammographic Density

Institution: University of California, San Francisco
Investigator(s): Thea  Tlsty , Ph.D. -
Award Cycle: 2005 (Cycle 11) Grant #: 11IB-0039 Award: $148,163
Award Type: IDEA
Research Priorities
Prevention & Risk Reduction>Risk reduction and identification

Initial Award Abstract (2005)
While increased breast density is one of the strongest predictors of breast cancer risk, we know little about its biological basis or how it is associated with increased cancer risk. We do not know if increased breast density is causal or reflective of the mechanisms that lead to increased breast cancer risk. There are influences by genetic, physiologic and environmental factors that generate the tissue phenotypes characteristic of high breast density (excess collagen, tissue remodeling, etc.). These phenotypic characteristics have been observed in conditions where response to tissue remodeling or damage is occurring, such as in mammary gland development (branching morphogenesis), wound healing or the desmoplastic reaction (i.e., the pervasive growth of dense fibrous tissue around the tumor). In these processes, "activated stroma" results in increased levels of collagen and tenascin, stromal remodeling and altered cell cycles for cellular components located within.

We hypothesize that increased breast density may be the end result of biologic processes that result in altered cell-cell and/or cell-extracelluar matrix (ECM) interactions and that these are causal for increased breast cancer risk. Prior work from our laboratory has shown that such altered stroma can dramatically influence tumorigenesis in human models. Proper stromal-epithelial interactions within the ECM can actually suppress the expression of pre-neoplastic phenotypes in epithelial cells and conversely, altered stromal-epithelial interactions can promote the probability that pre-neoplastic lesions progress to malignancy.

If fibroblasts were activated in tissue with high breast density, we would predict that fibroblasts from these tissues would (1) manifest many of the same phenotypes observed in activated stroma (increased collagen, etc.) and (2) possess similar functional properties in facilitating tumor progression of pre-malignant lesions. We will test these predictions within the studies of the proposed application. We will use recombinant grafts of different pairs of fibroblasts and epithelial cells implanted in the mammary fat pad of murine hosts to evaluate cell to cell interactions. Standard tumorigenenic parameters (growth, latency metastasis), cellular characteristics (proliferation and apoptosis) as well as tissue morphology (collagen content, MMP expression, etc.) will be assessed using methodology ongoing in our lab.

Our long-term goal is to identify clinically relevant biomarkers for the early genetic and epigenetic events in carcinogenesis that reflect stromal-epithelial interactions. Our purpose in this study is to determine if fibroblasts from tissue which exhibit high or low mammographic density modulate tissue appearance, tumorigenicity or metastasis. The translational applications of these studies are significant: first, predictive tests could be designed using a molecular imaging ligand attached to an antibody against one or more of the stromal marker gene products for a novel imaging test that may facilitate the detection of tissue at risk for breast cancer, hopefully earlier, less painfully, and more specifically than current mammographic techniques that rely on changes in tissue architecture and density. Second, markers may be identified, tested and then developed into a tissue diagnostic for testing on core needle biopsies that are currently benign but have a high likelihood of developing cancer. Third, a chemical signal may be identified that (in the future) could be assayed in the patient's serum or nipple aspiration fluid leading to the development a simple test for the very early detection of breast cancer risk. Fourth, targets may be identified for the development of novel chemoprevention strategies. Fifth, a mouse model would provide a biologically relevant system to study chemoprevention and/or modulation of human breast density. Although these types of studies are beyond the scope of this grant application, the development a model that would be clinically useful is not. Sixth, novel imaging methodologies may provide improved assessments of biological factors that contribute to breast cancer risk, whether through compositional or metabolic measurements.

Final Report (2006)
Present estimates suggest that for every 1% increase in mammographic density (MD), there is a 2% increase in risk for breast cancer. Other than age, MD is one of the strongest predictors of breast cancer risk; yet the biological basis for the increased risk of breast cancer that accompanies increased MD is unknown. We hypothesized that increased MD may be caused by biological processes that result in altered cell-cell and/or cell-extracellular matrix (ECM) interactions and that these changes are causal for increased breast cancer risk. These altered cell-ECM interactions are influenced by genetic, physiological and environmental factors and generate tissue phenotypes that are characteristic of high MD (excess collagen, tissue remodeling, decreased fat and increased cell number etc.). These MD phenotypic characteristics are reminiscent of an 'activated' stroma. Activated stroma is similar to stroma formed during normal, non-pathological processes such as morphogenesis and wound healing, and can also be found in pathological states such as desmoplasia (i.e., formation of adhesions or fibrosis in the vascular stroma of a tumor). Our current results supported by the CBCRP have identified discrete molecular differences between low density and high density associated fibroblasts that have the potential to link MD to cancer risk. In these studies we compared fibroblasts cultured from benign breast biopsies isolated from women with high and low MD. RNA was purified, amplified, and gene expression profiles were analyzed by Affymetrix 47,000 gene microarrays. Preliminary results demonstrate that mammary fibroblasts from tissues of high MD express lower levels of CD36 (also known as GP88, GPIV and GP IIIb) compared to mammary fibroblasts from breast samples of low MD. CD36 is a receptor whose ligands include collagen and thrombospondin and has been shown to be involved in various biological processes relevant to MD (apoptosis, collagen signaling, adipocyte differentiation, etc.). We have also demonstrated that TGFβ , a protein known to be involved in modulating the phenotypes of activated stroma, including increased collagen and adipocyte differentiation, is increased in mammary fibroblasts from tissues of high MD. Intriguingly, we have shown that CD36 expression is modulated in mammary fibroblasts by TGFβ . We hypothesize that increased TGFβ activity, working through CD36-dependent and/or CD36-independent pathways, will influence phenotypes of increased MD and cancer risk. Interventions to lower MD could facilitate the detection of breast cancer and alter epithelial-stromal functions to alter disease initiation and progression.