However, 3D co-cultures to study heterotypic interactions are less widely used. Hence, we developed and employed a series of 3D co-culture systems to investigate the impact of fibroblasts on tumor cell phenotype and response to endocrine therapy. Fibroblasts are the most abundant cell type in the breast GDC-0834 stroma and while they play a role in the endocrine regulation of normal breast differentiation, it is not well understood how they affect the response of breast cancer cells to targeted endocrine therapy. In order to characterize the influences of cancer cell-stromal interactions on therapeutic response, we profiled the conditioned medium of the co-culture and defined a molecular interaction signature. The iSig provides mechanistic insight into tumor progression and the dynamics of cancer cell behavior by identifying specific secreted proteins involved in cancer cell-stromal cross-talk. When we separated breast cancer patient microarray data based on iSig expression levels, we were able to predict patient outcome that was comparable to available gene expression profiling methods. Although uncovering genomic and proteomic dynamics of tumor behavior are crucial for understanding the pathophysiology of cancer, imaging techniques remain a gold standard of determining diagnosis and prognosis in many solid tumors, including breast cancer. Here, we used Fourier Transform infrared spectroscopic imaging for rapid and labelfree profiling of co-culture samples, integrating molecular information about cellular phenotypes with microscopy. As opposed to antibody-based imaging approaches such as immunohistochemistry or immunofluorescence, emerging chemical imaging technologies use spectroscopy to provide functional and molecular information Chenodiol within cells and tissues without staining or the requirement for a priori knowledge of molecular transformations. By visualizing the chemistry inherent within a sample, cell activity changes and disease states in vitro and cell types within complex tissues can be monitored.Although a number of studies have reported the development of high performance imaging, rapid data processing and classification algorithms, as well as applications to histologic analyses of tissues, few studies have related spectral data to underlying molecular transformations.