Please note that we have stopped the regular imports of Gene Expression Omnibus (GEO) data into ArrayExpress. This may not be the latest version of this experiment.
E-GEOD-18331 - A phenotype-based model for rational selection of novel targeted therapies in treating aggressive breast cancer
Released on 1 June 2011, last updated on 13 June 2011
Treating unselected cancer patients with new drugs dilutes proof of efficacy when only a fraction of patients respond to therapy. We conducted a meta-analysis on eight primary breast cancer microarray datasets representing diverse breast cancer phenotypes. We present a high-throughput protocol which incorporates drug sensitivity signatures to guide preclinical testing for effective therapeutic agents. Specifically, we focus on drug classes currently undergoing early phase clinical testing. Our genomic and experimental results suggest that the majority of basal-like breast cancers should respond to inhibitors of the phosphatidylinositol-3-kinase pathway, and that a relatively low toxicity histone deacetylase inhibitor, valproic acid, may target aggressive breast cancers. For a subset of drugs, prediction of sensitivity associates with tumor recurrence, suggesting clinical relevance. Preclinical studies using both cell lines and patient tumors grown in 3-dimensional in vitro and orthotopic in vivo preclinical models provide an efficient and highly relevant assessment of drug sensitivity in tumor phenotypes, and validate our genomic analyses. Together, our results show that high-throughput transcriptional profiling can significantly impact drug selection for breast cancer patients. Pre-identification of patient response may not only improve therapeutic response rates, it can also assist in quickly identifying the optimal inclusion criteria for clinical trials. Our model facilitates personalized drug therapy for cancer patients and may be generalized for study of drug efficacy in other diseases. Breast cancer pleural effusion samples from triple negative patients. Compared samples that are computationally predicted to be sensitive to valproic acid and those that are not predicted to be sensitive.
transcription profiling by array
Adam Gustafson <firstname.lastname@example.org>, Adam M Gustafson, Andrea H Bild, Bryan E Welm, Jeffrey T Chang, Stefanie S Jeffrey