Abstract 1040: An integrated analysis of EMT across diverse cancer types identifies new potential therapeutic targets

Abstract

Abstract Background: Epithelial-to-mesenchymal transition (EMT) is a marker of resistance to specific targeted drugs in non-small cell lung cancer (NSCLC), head and neck squamous cell carcinoma (HNSC), and other epithelial cancer and may be associated with poor patient outcomes. Here we performed an integrated molecular analysis in 1934 tumors representing 11 different cancer types in the Cancer Genome Atlas (TCGA) and 1264 representative cell lines to investigate new potential therapeutic targets (PTTs) in mesenchymal tumors. Methods: For each tumor, we correlated the EMT score (based on expression of an EMT gene signature) with mutation status, copy number alterations (CNA), mRNA, miRNA, and protein expression. PTTs expressed in mesenchymal tumors were investigated in vitro using publically available drug sensitivity datasets (DSDs) by correlating IC50s with EMT score. Results: Most cancer types exhibited a range of EMT scores that included both epithelial and mesenchymal tumors, with the exception of kidney clear cell carcinoma (KIRC) (all mesenchymal) and ovarian serous cystadenocarcinoma (OV) (primarily epithelial). Patients with mesenchymal OV and HNSC experienced shorter overall survival (p=0.029 and 0.076, respectively). BAP1 and PP2R1A mutations were more frequent in mesenchymal tumors, independent of primary site (adjusted p value <0.05). Conversely, NSD1 and KEAP1 mutations were more frequent in tumors with low EMT scores (epithelial) (p<0.05). Overall mutation burden and CNAs were not associated with mesenchymal status. A striking association between miR expression and EMT score was observed across most tumor types, including low miR 200 family and high miR 199a and 214 expression in mesenchymal cancers (r>0.3). PTTs commonly expressed in mesenchymal tumors included AXL, PDGFRA/B, MMP2 and DDR2 (r>0.6). Of these, investigation of the DSDs supported in vitro sensitivity of mesenchymal cell lines overall to PDGFR and GSK3 inhibitors, but resistance to drugs targeting ErbB family members, including the dual VEGF/EGFR inhibitor vandetanib (p<0.001) and drugs targeting Her2 and/or EGFR (p<0.001). In contrast, other EMT-associated drug resistance was tissue specific, such as resistance of mesenchymal breast cancer lines to mTOR inhibition (p 0.03). Conclusions: Application of the EMT signature to diverse tumor types identifies highly conserved patterns of gene and miRs expression, including high miR-199a expression in mesenchymal cancers. While mutational landscape and CNAs were most often driven by tumor type, protein expression and drug sensitivity were influenced by a combination of tumor type and EMT status. New associations between drug response and EMT were identified, such as the sensitivity of mesenchymal cancers to PDGFR and GSK3 inhibitors, that warrant further investigation. Citation Format: Pan Tong, Milena P. Mak, Lixia Diao, Jing Wang, Patrick Kwok-Shing Ng, You-hong Fan, William N. William, John V. Heymach, Kevin R. Coombes, Lauren Averett Byers. An integrated analysis of EMT across diverse cancer types identifies new potential therapeutic targets. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 1040. doi:10.1158/1538-7445.AM2014-1040