Abstract PR12: Directing personalized breast cancer treatment with pathway signatures

Abstract

Abstract One of the major limitations in the treatment of breast cancer is the patient to patient heterogeneity of the tumors. This is evidenced by the well-known intrinsic subtypes of breast cancer, including luminal A and B, basal, claudin-low, HER2+ and normal like. If one considers the hormonal status (ER / PR) of the breast cancers in addition to the intrinsic subtypes, then additional classifications are apparent. Indeed, recent work has suggested that there are approximately 17 subtypes of breast cancer. With these various types of breast cancer, clinicians are faced with the challenge of determining the optimal treatment strategy for the individual breast cancer patient. When treatment options are considered in this way, it becomes readily apparent that breast cancer, and cancer in general, needs to be treated with a personalized approach. We believe that one mechanism to address the requirement for personalized therapy is to employ cell signaling pathway signatures based on gene expression data to differentiate the tumors into treatment groups. As a proof of principle experiment, we have used pathway signatures in a mouse model of breast cancer. An extensive gene expression database was constructed for 23 major mouse models of breast cancer encompassing over 1100 breast cancer samples. Predictions for key signaling pathway activation were calculated using genomic data and pathway signatures. This dataset demonstrated heterogeneity in the vast majority of mouse model systems with important relations to human breast cancer. Analysis of this data prompted a focus on the MMTV-Myc model as the genomic signatures clearly predicted various subtypes of cancer in the MMTV-Myc model system. Indeed, in one subtype of breast cancer in the Myc model we noted elevation of the EGFR, TNF and Ras pathways whereas these pathways are not predicted to be active in a second subtype of cancer from the same model system. In the second tumor subtype, there is a high predicted activity for other pathways, including Myc, Stat3 and AKT. To demonstrate that these predictive signatures can be used to guide clinical therapy, we then took an approach where we targeted each of the pathways in a combinatorial approach with small molecule inhibitors in transplantable tumors from the initial tumor subtypes. Thus, we have two therapies, each tailored to a specific subtype of a mouse model breast cancer. Safety trials for both weight and liver enzymes for the combination of the three drugs in each treatment arm revealed that there were no significant side effects when inhibitors were used at the low doses designed for use in therapy. Importantly, we have shown that the predicted combination therapy is effective in blocking tumor growth in the tumor type that is predicted to be responsive while it is ineffective in blocking tumor growth in the other subtype of tumors. These results clearly demonstrate in a proof of principle experiment that using pathway signatures is a viable mechanism for identifying and directing treatment for breast cancer. This abstract is also presented as Poster A02. Citation Format: Jing-Ru Jhan, Daniel Hollern, Eran R. Andrechek. Directing personalized breast cancer treatment with pathway signatures. [abstract]. In: Proceedings of the AACR Special Conference: The Translational Impact of Model Organisms in Cancer; Nov 5-8, 2013; San Diego, CA. Philadelphia (PA): AACR; Mol Cancer Res 2014;12(11 Suppl):Abstract nr PR12.