Abstract Background: Cellular heterogeneity within all breast cancer subtypes remains a major cause of treatment failure and development of metastatic disease. Currently, both preclinical studies and drug development efforts focus almost exclusively on the epithelial component of breast cancers. Despite strong preclinical data novel therapies often fail in clinical testing. We propose that this failure is, in part, due to taking tumor cells out of their context and using pre-clinical models that fail to capture the complexity of human disease. We hypothesize that tumors hijack normal components of the tissue microenvironment and use it to their advantage. Here we demonstrate that similar to the normal hematopoetic niche, two major subtypes of breast cancer stroma can be defined by CD146 expression. We further show that the ratio of the stromal subtypes alters the response to therapy and increases the metastatic potential of breast cancer cells (BCC). Results: Tumor associated stroma, from all breast cancer subtypes, contains a mixture of CD146+ and CD146- fibroblasts. We isolated and derived pure human CD146+ and CD146- clonal lines. Both subtypes expressed markers of activated fibroblasts and clustered by gene expression profiling with normal stromal cell lines HS27A (CD146+) or HS5 (CD146-) according to CD146 expression. Although both stromal subtypes were derived from tumor associated tissue, CD146+ breast cancer stroma clustered with normal breast associated stroma and correlated with good clinical outcome (Finak et. al. Nature Med 2008). CD146- stroma clustered with breast cancer associated stroma and predicted worse outcome. Using cell line and patient-derived xenograft models of estrogen receptor (ER) positive breast cancer, we demonstrated that CD146+ compared to CD146- stroma supported significantly higher ER expression in BCCs. BCC co-cultured with CD146+ stroma responded more robustly to estrogen treatment and anti-endocrine therapy with tamoxifen. Next we used expression profiling data to predict stromal influence on treatment response. CD146+ stroma expressed 3-fold more TGFβ than CD146- stroma. Inhibiting TGFβ decreased proliferation 2-fold in BCCs grown in media conditioned by CD146+ stroma, but not by CD146- stroma. Conversely, HBEGF expression was 3-fold higher in CD146- stroma compared to CD146+ stroma. Inhibiting EGFR decreased proliferation 1.5-fold in BCCs grown in conditioned by CD146- stroma, but not by CD146+ stroma. In addition, intracardiac injection of stroma resulted in distant metastases in our primary orthotopic PDX model of breast cancer. Lastly we confirmed our preclinical observation using a small set of clinical samples. Patients with high CD146+ to CD146- stromal ratio had better clinical outcomes than patients with high CD146- to CD146+ stromal ratio. Conclusion: We conclude that stromal subtypes defined by CD146 expression direct the heterogeneity of ER expression, response to therapy and the metastatic potential of breast cancer cells. Citation Format: Heather M Brechbuhl, Jessica J Finlay-Schultz, Tomomi M Yamamoto, Austin E Gillen, Anthony Elias, Carol A Sartorius, Peter Kabos. CD146 positive and negative stroma direct breast tumor estrogen receptor levels, therapeutic response and metastatic potential [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P4-04-06.