Epithelial-mesenchymal gene expression signature defines clinically relevant subtypes in epithelial ovarian cancer
Journal
CANCER RESEARCH
Journal Volume
72
Date Issued
2012
Author(s)
Abstract
Abstract Epithelial ovarian cancer (EOC) represents a broad and heterogeneous entity which includes different invasive behavior as well as distinct histopathological subtypes. The heterogeneity of EOC is further complicated by the fact that EOC is characterized by a high degree of genetic damages and multiple genetic alterations resulting from different mechanisms. Even within the seemingly similar histo-pathological group, molecular subtypes with different gene/pathway activations have been identified. In this study, we described a classification scheme based on in silico meta-analysis of gene expression profiling of 1,538 EOC microarray data. and identified five major robust subtypes termed Epithelial-A, Epithelial-B, Mesenchymal, Stem-A, and Stem-B that were distinctive in gene expression patterns. The epithelial subtype is hallmarked by genes known to confer epithelial characteristics, such as Keratin genes (KRT17, 14, 19, and 7), E-cadherin (CDH1), and Ep-CAM (EPCAM) and show the second best prognosis in Kaplan-Meier analysis. On the other hand, the mesenchymal subtype consists of genes such as fibronectin (FN1) and known EMT inducers ZEB1 and TWIST1. The Stem-A subtype consists of genes which confer stemness (LGR5, NCAM1). The Mesenchymal and Stem-A subtypes are linked with poorer outcomes. We further utilized a panel of 42 ovarian cancer cell lines to model EMT in vitro by performing phenotypic EMT characterization. Four EMT subgroups (epithelial, intermediate epithelial, intermediate mesenchymal, mesenchymal) representing different epithelial-mesenchymal compositions on the EMT spectrum were identified. We also demonstrated that this EMT spectrum can be used to refine the analysis of data generated in cell-based functional studies. By performing EMT-related functional assays, we found that ovarian cancer cells harbouring intermediate phenotypes are more migratory, invasive, anoikis resistant, and more spheroidogenic. We further utilized this cell line model to apply a cell line-based EMT scoring system to re-classify the 1,538 EOC microarray data. It revealed that the Mesenchymal and Stem-A subtypes were both enriched in higher mesenchymal scores. This supported the hypothesis of EMT's contribution to the aggressiveness of solid tumors. In conclusion, our results showed that EOC is sub-classified into distinct molecular subtypes correlated with clinical outcomes that are linked to EMT. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 2979. doi:1538-7445.AM2012-2979
Publisher
AMER ASSOC CANCER RESEARCH
Type
other