D. Anderson Cancer Center, Orlando, FL, USA, 4 University of California at Irvine, Irvine, CA, USA We have developed a linear model of prostate tissue that describes gene expression changes as a sum of contributions of four major cell types in tumor enriched samples including tumor cells, stroma cells, epithelial cells of BPH, and dilated cystic glands. When combined with knowledge of the cell type distribution as estimated by pathologists, the model provides estimates of gene expression for each cell type (1). By comparing the expression of stroma cells Selleck RO4929097 in low (<15%) tumor samples
with normal volunteer biopsy samples, we derived 417 significant gene expression differences which were further filtered JQ1 to remove genes with significant expression in tumor
cells. The resulting 17 genes, which appeared to have high expression in stroma only when in the presence of tumor, were applied to a training set of 18 PCa cases and 17 noncancer tissues of the same cases all measured on U133plus2 Affymetrix arrays. The program PAM yielded 97% accuracy for discriminating tumor cases vs. non tumor cases. The classifier was then tested on multiple independent prostate samples including 65 tumor cases and a separate 79 case set both measured on U133A arrays and both publically available, and 55 independent cases measured on U133plus2 arrays in house which yielded an accuracy of 96–100% for the three sets. To exclude performance that may be based on recognition of tumor cells, we tested the classifier on 9 additional independent normal volunteer biopsy cases and 7 normal rapid autopsy cases that were histologically PRKACG confirmed to be tumor free which yielded 100% accuracy as nontumor cases for both series. Thus a classifier based on tumor-adjacent stroma is highly accurate for discrimination of tumor and nontumor. A significant number of the million prostate biopsies in the U.S. per year have equivocal pathological readings, therefore, methods for augmenting diagnostic accuracy based on stroma may be helpful. 1.Stuart
et al. PNAS 2004;101:615–20. O76 Bone Marrow Endothelial Progenitor Cells are Systemic Sensors of Breast Cancer Robert Suriano 1, Andrea George1, Shilpi Rajoria 1, Erin Lambers 2, Raj Kishore 2, Raj Tiwari 1 1 Department of Microbiology and Immunology, New York Medical College, Valhalla, NY, USA, 2 Feinberg Cardiovascular Institute, Northwestern University, Chicago, IL, USA Circulating bone marrow derived endothelial progenitor cells (BM-EPC) have been observed to contribute to neo-vascularization of breast cancers and the identification of its systemic mediators will impact clinical care. We discovered a crucial role for BM-EPCs in breast cancer progression with estradiol (E2) as a major modulator. We utilized TEK2/GFP-Balb/c ± ovariectomized ± estrogen supplementation as our experimental mouse model.