Effective personalized therapy for breast cancer based on predictions of cell signaling pathway activation from gene expression analysis
Current therapeutic outcomes for breast cancer underscore the complexity of treating a heterogeneous disease. Indeed, studies have shown that differences in gene expression among patients with the same subtype of breast cancer are correlated with the response to treatment. This strongly suggests that there is an urgent need to treat breast cancer with a personalized approach. Here we employed cell signaling pathway signatures to predict pathway activity in subtypes of MMTV-Myc mammary tumors. We then split tumors into subsets and developed individualized combinatorial treatments for two subtypes with distinct pathway activation patterns. Elevation of the EGFR, RAS and TGFβ pathways was observed in one subtype whereas these pathways were not predicted to be active in the other subtype that had high predicted activity of the Myc, Stat3 and Akt pathways. In a proof-of-principle experiment, treatment of these two subtypes with targeted therapies inhibited tumor growth only in the subtype of tumor where the therapy was designed to be active. We then analyzed gene expression profiles of human breast cancer patients and patient-derived xenograft (PDX) samples to predict pathway activity, and validated our approach of developing individualized treatments in mice with PDX tumors. Importantly, our combinatorial therapy resulted in tumor regression, including regression in PDX samples from triple-negative breast cancer. Together our data is a proof-of-principle experiment that demonstrates that cell signaling pathway signature-guided treatment for breast cancer is viable.