Predictive value of a gene signature evaluated in multiple chemotherapy regimens for breast cancer.

Abstract

545 Background: Approximately 12,800 U.S. women present with locally advanced breast cancer each year. Primary (or neoadjuvant) chemotherapy is a treatment to shrink a large tumor to a size that allows it to be removed with breast conserving surgery. A woman who would otherwise need a mastectomy (surgical removal of the breast) may instead elect to receive primary chemotherapy and breast- conserving lumpectomy. Unfortunately, primary chemotherapy has a low response rate (30%-50%) and a significant negative impact on a patient's quality of life. If a tumor does not respond to primary chemotherapy, a mastectomy may be necessary after all. Currently, there is no widely used clinical test to determine before treatment if a given tumor will respond. We have developed a signature of 22 genes whose expression pattern accurately predicts response to primary chemotherapy in breast cancer. This 22-gene signature was discovered by a unique approach using an in vitro model of normal breast epithelial cell differentiation. Here we present data showing that the 22-gene signature predicts response to multiple chemotherapies. Methods: We performed analysis of two published microarray datasets to test whether the 22- gene signature would be predictive of response to primary chemotherapy by hierarchical clustering, Fischer's exact tests, and ROC analysis. The dataset of Chang et al. includes 24 breast cancer core biopsies (Lancet. 2003;362:362-369) and Hess et al. includes 133 fine-needle aspirates from breast cancer patients (JCO. 2006;24:4236-4244) taken before primary treatment with docetaxel or combination treatment respectively and response was assessed after chemotherapy. Results: The 22-gene signature accurately predicted chemotherapy response in both independent datasets (ROC p = 0.0001; Fisher's Exact p < 0.0001). ROC analysis values for sensitivity and specificity showed that the 22-gene signature was superior to clinical parameters and other published gene signatures. Conclusions: This gene signature has the potential to fill the existing need for an in vitro diagnostic to provide accurate personalized information to guide primary chemotherapy treatment. Author Disclosure Employment or Leadership Position Consultant or Advisory Role Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration Bioarray Therapeutics Inc. Bioarray Therapeutics Inc. Bioarray Therapeutics Inc. Bioarray Therapeutics Inc.