Stroma Expression Patterns in Breast Cancer

Institution: Palo Alto Veterans Institute for Research
Investigator(s): Robert West, M.D., Ph.D. - Robert West, M.D., Ph.D. -
Award Cycle: 2009 (Cycle 15) Grant #: 15NB-0156 Award: $164,403
Award Type: IDEA Competitive Renewal
Research Priorities
Biology of the Breast Cell>Pathogenesis: understanding the disease

This is a collaboration with: 15NB-0156A -

Initial Award Abstract (2009)

The tissue that forms the stroma surrounding epithelial breast carcinoma cells (tumor microenvironment, TME) is a complex structure that consists of a wide variety of cell types, many of which are poorly described. Recent studies have revealed the importance of the interaction between the TME and tumor cells by showing that the TME actively promotes tumor growth and possibly even metastasis. However, the complexity of the TME makes it difficult to study. For example, it is difficult to determine how the expression patterns of a multitude of different stromal cells vary from one tumor to the next or to determine to which degree these variances represent distinct stromal signatures. We have developed a novel approach to discover new types of stromal reaction patterns through gene expression profiling of soft tissue tumors (STT). In this approach STTs (that are derived from distinct normal connective tissue precursor cells) function as a “discovery tool” for the various types of stromal reaction patterns. Using this approach and previous CBCRP funding, we have to date described four novel distinct stromal reaction patterns in invasive breast carcinoma. The biological significance of the identification of these various stromal reaction patterns is borne out by the fact that several of these stromal expression patterns have prognostic significance independent from traditional prognosticators such a tumor size, tumor grade and even lymph node status.

The stromal response patterns that we have discovered for breast cancer can be identified in clinical samples through the development of a small panel of antibodies. The current project is focused on stromal responses, or other yet to be discovered stromal responses, present in both invasive cancer and ductal carcinoma in situ (DCIS). Our aims are to: (1) develop clinically useful biomarkers of stromal expression patterns in invasive cancer, and (2) discover stromal response patterns associated with pre-invasive breast cancer. Our experimental approach involves tissue microarrays of 283 cases to screen for antibody markers for different stromal biomarker profiles. We will ultimately reduce each panel down to the best 2-3 markers which will serve as a clinical test for the presence of these stromal signatures in breast cancer. Next, using this data we will analyze a combination of pre-existing DCIS gene expression profiling datasets and generate a novel stromal-based biomarker response profile.

If our research is successful, we will have created clinically practical panels of antibodies that can discriminate key stromal signatures in clinical breast cancer specimens. These panels will have both prognostic utility and may predict response to therapy. Finally, we will have insight into the stromal signatures associated with DCIS, which will assist in clinical decision-making on the best course of therapy.




Progress Report 1 (2010)

In a previous CBCRP IDEA grant (13IB-0076), we characterized stromal expression signatures in invasive breast cancer (we define stromal expression signatures as gene expression patterns of cells in the stroma surrounding the breast cancer cells, including “fibroblasts”, endothelial cells, and leukocytes). We hypothesize that the stromal signatures provide critical functions for the survival and progression of the neoplastic breast cancer cells. These stromal signatures are associated with specific clinicopathologic features of breast cancers as identified in five independent breast cancer gene expression profiling datasets.

With the specific continuation funding from the CBCRP in the current grant (15NB-0156) we have pursued two specific goals. Our first goal is to generate a clinically useful test to assess the presence of the desmoid-type fibromatosis (DTF) fibroblast stromal response signature in breast cancer samples. We have taken three approaches. The first is to identify a panel of gene markers that might accurately assess the presence of the stromal signatures in a clinical laboratory. We have analyzed our existing library of markers, for both the DTF fibroblast and colony stimulating factor 1 (CSF1) macrophage stromal signatures, to assess the discriminatory power of individual markers in a set of 30 cases. We found that two markers from each signature, MMP11 and osteonectin in the DTF fibroblast signature, and cathepsin L and CD163 in the CSF1 macrophage signature, are candidate surrogate markers of response. We plan on testing these markers on additional independent tumor sets and discovering for new markers to test by mining gene expression profiling data. A second approach is to generate a quantitative image analysis method for identifying the stromal signatures. As morphologic features correlate with molecule features, in general, and that the digitalization of tissue slides is an emerging technology in clinical surgical pathology (with the aim of using the images instead of the physical slides to make diagnoses), we think that it is possible that computer assisted image analysis to determine the presence of stromal signatures could be both feasible and practical in the future. We have started to generate quantitative image analysis algorithms to identify the stromal signatures in H&E images. We have analyzed H&E images for a large set (~400) of morphologic features (like nuclear size) identified by quantitative image analysis and looked for correlations with the matching breast cancer’s gene expression profile. To date, we have achieved a correlation of 0.65 which we expect to improve on as we gain more experience with image analysis algorithms. A third approach is to identify the driving biology that generates the stromal signatures. In a study of these stromal signatures in patients with synchronous breast cancers, we find that the DTF fibroblast stromal response signature shows concordance in patients with matched dual primaries, while the CSF1 macrophage response stromal signature does not. This suggests that that the DTF fibroblast response is host-specific, and that the CSF1 response is tumor-elicited. Based on these studies we plan to examine whether any single nucleotide polymorphisms (minute genetic changes that naturally vary in the general population and are thought to be responsible for some variation in cancer incidence and progression) in the genes involved in these signatures may correlate with the presence of these signatures.

The second goal is assess whether the stromal signatures that we have identified in invasive breast cancer are also present in pre-invasive lesions, such as ductal carcinoma in situ, and if so, determine whether these signatures correlate with clinicopathologic features and predict the natural history of the disease or response to treatment. Our recent results demonstrate that distinct subsets of ductal carcinoma in situ have the DTF fibroblast and CSF1 macrophage response signatures that can also be found in invasive breast cancer. We have shown this with two different methodologies: by analysis of a published gene expression profiling dataset, which provides measurements of all the gene transcripts in the stromal signatures, and by immunohistochemistry, which looks at a limited number of protein products from the genes but allows for localization within specific compartments of the tumor microenvironment. We are now examining whether these stromal signatures are present in even earlier breast neoplasias. To do this, we have developed a high through put sequencing technique for expression profiling of archival material. This allows us to accurately profile invasive ductal carcinoma, ductal carcinoma in situ, early neoplasias (like atypical ductal hyperplasia, etc), and normal breast tissue within the same patient. We have generated sets of these profiles for over 15 patients with breast cancer and are analyzing the data for the presence of the stromal signatures and other gene pathways that are associated with them.




Progress Report 2 (2011)

In the first year of this grant, we have pursued two specific goals. Our first goal has been to generate a clinically useful test to assess the presence of the DTF fibroblast stromal response signature in breast cancer samples. We have identified a panel of gene markers that might accurately assess the presence of the stromal signatures in a clinical laboratory. We have identified DTF and CSF1 “core signatures” (which contain 66 and 112 genes respectively), which include genes chosen because they showed highly coordinate expression across multiple independent breast cancer datasets and were the most highly representative genes of the DTF fibroblast and CSF1 macrophage signatures, as previously described. We now have identified a corollary signature of cases that have no or very little expression of the DTF core genes. This “DTF negative” signature is not simply the cases that are not classified as “DTF positive”. In fact, the majority of cases do not fall into either category. These “DTF negative” cases have their own characteristic gene signature and clinicopathologic features.

We are now examining whether these stromal signatures are present in even earlier breast neoplasias. To do this, we have developed a high through put sequencing technique for expression profiling of archival material. We designed a novel RNA-seq protocol, 3SEQ, to specifically expression-profile polyA+RNA extracted from formalin-fixed, paraffin-embedded archival tissue. This allows us to accurately profile invasive ductal carcinoma, ductal carcinoma in situ, early neoplasias (like atypical ductal hyperplasia, etc), and normal breast tissue within the same patient. We have now generated sets of these profiles for 24 patients with and without breast cancer. We have found that the DTF signature can be identified at very early stages of breast neoplasia, including columnar cell change. Cases of columnar cell change that develop invasive cancer also tend to demonstrate the DTF signature.



Determination of Stromal Signatures in Breast Carcinoma
Periodical:PLoS Biology
Index Medicus: PLoS Biol
Authors: West RB, Nuyten DSA, Subramanian S, Nielsen TO, Corless CL, et al.
Yr: 2005 Vol: 3 Nbr: 6 Abs: Pg:187

Analysis of stromal signatures in the tumor microenvironment of ductal carcinoma in situ.
Periodical:Breast Cancer Research and Treatment
Index Medicus: Breast Cancer Res Treat
Authors: Sharma M, Beck AH, et al and West RB
Yr: 2010 Vol: 123 Nbr: 2 Abs: Pg:397-404