Abstract 1078: TGF-β and TNF-α activate gene transcription programs associated with poor breast cancer prognosis

Abstract

Abstract Background: Tumor Growth Factor-β (TGF-β) and Tumor Necrosis Factor-α (TNF-α) are cytokines, which induce the local inflammation and contribute to tumor invasion and metastasis. Recent reports show that TGF-β and TNF-α induce epithelial-to-mesenchymal transition (EMT) in both non-cancer and cancer cells. EMT is a cellular program, through which cancer cells transform from epithelial to mesenchymal phenotype. It is known that EMT induces cancer stem cell (CSC), which is a main culprit in tumor progression. However, it is not well understand what molecular signature drives EMT and CSC induction in the downstream of TGF-β and TNF-α. We hypothesize that TGF-β and TNF-α promote tumor progression and contribute to poor cancer prognosis. The objective of this study is to identify the downstream gene signature of TGF-β and TNF-α stimulation, which allows us to predict poor cancer prognosis. Methods and Results: We previously established in vitro EMT model, in which we stimulated ARPE-19 retinal epithelial cells with TGF-β and TNF-α and observed cellular focuses and fibronectin buildups, hallmarks of in vitro EMT (J Bio Chem 2010). We analyzed microarray transcription profiles of EMT progression at different time points (0, 1, 6, 16, 24, 48, 60 hours). To quantify the likeliness of EMT and CSC, we analyzed EMT and CSC gene scores of the EMT progression by correlation analysis (PNAS 2009). We observed that EMT gene score reached the highest at 16 hours and followed by elevations of CSC gene score, suggesting a dynamic change of gene signature that drives the progression from triggering EMT to inducing CSC by TGF-β and TNF-α. In order to determine which time point of EMT timeframe is more representative for poor cancer prognosis, the transcriptional profile at each time point was compared with the one at 0 hour using t-test and the differentially expressed genes (p<0.05, fold change> 2) were used for further analysis. The expression pattern of each time point was compared with each cancer patient from two breast cancer data sets (Mainz and Transbig) and was used to predict patient prognosis. We discovered that the gene signature at 6-hour (early phase of EMT) was the only predictable signature for the poor prognosis of primary breast cancer patients (logrank p-value of 0.045), with a 5-year distance metastasis free (DMF) survival rate of 0.77 vs. 0.94, and a hazard ratio of 1.49. The same gene signature also predicts poor prognosis in ER-positive primary breast cancer patients (p-value = 0.028, hazard ratio = 1.74). We further performed Ingenuity Pathway Analysis using 6-hour gene signature and observed that interferon response and NF-κ pathway activation were the top ranked pathways. Conclusion: Using time-course transcriptional profiles stimulated by TGF-β and TNF-α, we discovered gene signature that precedes EMT and CSC induction, represented by NF-κ pathway. This signature was significant in predicting prognosis of breast cancer patients. Citation Format: Yuan Qi, Kazuharu Kai, Chad J. Creighton, Bedrich Eckhardt, Hideyuki Saya, Debu Tripathy, Naoto T. Ueno. TGF-β and TNF-α activate gene transcription programs associated with poor breast cancer prognosis. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1078. doi:10.1158/1538-7445.AM2015-1078