Breast Stromal Genes Act as Early Markers of Malignancy

Institution: University of California, San Francisco
Investigator(s): Stefanie  Jeffrey , M.D. - Thea  Tlsty , Ph.D. - Stefanie  Jeffrey , M.D. - Thea  Tlsty , Ph.D. -
Award Cycle: 2002 (Cycle VIII) Grant #: 8EB-0106 Award: $249,624
Award Type: TRC Full Research Award
Research Priorities
Imaging, Biomarkers, & Molecular Pathology>Biomarkers and novel screening approaches: unmasking the hidden signs

This is a collaboration with: 8EB-1106 -

Initial Award Abstract (2002)
At present, early detection of breast cancer depends on the radiological and pathological properties (density, architectural distortion, calcifications) of the tumor tissue. These are often non-specific and relatively late changes, and they depend on a tumor of fairly large size (>0.5 cm for mammography) and a late stage of progression. The connective tissue and supporting cells in the breast, called stroma, represent an alternative opportunity to detect the early stages breast cancer. This is because the breast epithelial cells interact with the stroma, and stromal change may occur in parallel or even precede epithelial transformation. The stromal cells do not actually become cancerous themselves, since they remain genetically normal. However, we have reasons to believe that they become influenced by nearby cancer cells to alter their gene expression, such that abundance and presence of certain stromal cellular and secreted proteins could change. Thus, our hypothesis is that examination of stromal gene expression changes resulting from epithelial cell signaling should identify a set of markers that may be used for future molecular diagnosis. This might more accurately reveal tissue that is transitioning into invasive breast cancer.

Our project is a collaboration of expertise and perspectives on breast cancer. Dr. Tlsty is experienced in the basic science signaling processes between epithelial cells and stromal fibroblasts. Dr. Jeffrey is a clinician with research background in breast cancer imaging, treatment, and genetic analysis. Some underlying paradigms for our interest in the stroma include, (i) fibroblasts appear to "instruct" epithelial cells during development, (ii) the stroma and fibroblasts appear to undergo senescence with human aging and this creates a "permissive" state for tumor development, (iii) tumor cells may "signal" stromal fibroblasts in a, so-called, paracrine manner. We hope to address these issues by employing two fundamental approaches. First, we will co-culture fibroblasts and mammary epithelial cells. As part of this approach we will use both "normal" fibroblast and "carcinoma-associated fibroblasts" (CAF). This will allow us to establish a model system to study underlying genetic changes and signaling events associated with fibroblast-epithelial-cancer cell interactions. Secondly, we will employ the method of cDNA microarrays (i.e., "gene chips") to comprehensively analyze the gene "expression profile" that underlie differences in the fibroblast-epithelial-cancer cell interactions associated with breast cancer progression.

Our aim of this research is to identify specific genetic changes in breast stromal fibroblasts that are associated with specific stages of breast cancer, such as DCIS, various hyperplasia, and different invasive phenotypes. This will lead to molecular "signatures" in the stroma that could be used to both detect breast cancer at an earlier stage and predict disease progression risk. A potential clinical advantage is that the stroma would be expected to be more accessible to specific detection reagents (e.g., antibodies), because it is better exposed to the blood supply and not encased in a basement membrane, which is often the case with pre-invasive tumors.


Final Report (2006)
A primary goal in breast cancer diagnosis is early detection. When an invasive breast cancer is discovered at a small size, it is less likely to be associated with lymph node metastases and more likely to carry a better prognosis. It is estimated, however, that by the time a tumor is identified mammographically or by physical examination, the cancer has been already been growing for about 5-8 years.

Breast epithelium differences and grows within a matrix of breast stroma. The stroma is made up of fibroblasts that secrete collagen and other substances. As cancers grow, cell-to-cell signaling between stromal fibroblasts and malignant breast epithelium has been demonstrated to contribute to the growth and spread of breast tumor. We believe that the tissue following malignant transformation and may therefore serve as a phenotypic marker of breast cancer in its earliest stage. We hypothesize that specific changes in stromal gene expression may identity a developing malignant lesion at the detectable tumor.

A total of 36 fibroblast lineages were isolated and grown in culture from breast cancer specimens and normal mammary tissue samples. Thirteen of these samples were obtained directly from within tumor tissue and are referred to as carcinoma associated fibroblasts (CAF), while 11 samples were obtained from tissue more than 5 cm away from a tumor, but within the same breast as the tumor, and are referred to as normal associated fibroblasts (NAF). An additional 2 samples were obtained from tissue from the contra-lateral breast of women with cancer and are referred to as contra-lateral mastectomy fibroblasts (CMF). One sample was obtained from mastectomy tissue from a woman who did not have cancer, but had a family history of breast cancer, and was therefore referred to as high risk fibroblasts (HRF). Finally, 9 of these samples are from reduction mammoplasty tissue and are referred to as reduction mammoplasty fibroblasts (RMF).

Our preliminary results, utilizing both the 2-dimensional and 3-dimentional in vitro assays, indicate that mammary CAFs can stimulate the growth and alter the morphology of non-tumorigenic, immortalized mammary epithelial cells. These in vitro assays can readily be utilized to dissect the critical signals from stromal fibroblasts that are necessary for promoting tumorigenic phenotypes.

Affymetrix HU133 plus 2 chips, which contain over 47,000 transcripts (Affymetrix, Santa Clara, CA), were used for all microarray analyses. The microarray hybridizations, scans and analyses were preformed by the Gladstone Genomics Core Laboratory (GGCL) located at the Mission Bay campus of UCSF. The dataset was preprocessed by the GGCL using the following procedures: (1) background correction; (2) normalization using the quantile method; and (3) summarization of probe set values using the RMA (Robust Multi-array Average) method, which fits a specified robust linear model to the probe level data. These procedures were performed using the Bioconductor package “affyPLM”. After preprocessing, quality control assessments were performed and no arrays were discarded. Using the Bioconductor package, differential gene expression was determined between sample sets. Our results showed that the expression of several chemokines and a chemokine receptor is repressed in CAFs when compared to either NAFs or RMFs. In addition, the expression of several MMPs is repressed in CAFs when compared to RMFs, indicating an alteration in stromal remodeling.

With potential CAF markers identified, we are now in a position to determine when in the progression of a tumor the stroma begins to acquire “CAF” characteristics. Preliminary data has demonstrated our ability to stain tissue samples with antibodies against several of the candidate gene products. We are currently gathering clinical samples of tissue blocks containing pre-malignant lesions so that we can analyze the distribution of these markers during the entire continuum of tumor progression.