Epithelial-mesenchymal transition (EMT) is a reversible and dynamic process hypothesized to be co-opted by carcinoma during invasion and metastasis. Yet, there is still no quantitative measure to assess the interplay between EMT and cancer progression. Here, we derived a method for universal EMT scoring from cancer-specific transcriptomic EMT signatures of ovarian, breast, bladder, lung, colorectal and gastric cancers. We show that EMT scoring exhibits good correlation with previously published, cancer-specific EMT signatures. This universal and quantitative EMT scoring was used to establish an EMT spectrum across various cancers, with good correlation noted between cell lines and tumours. We show correlations between EMT and poorer disease-free survival in ovarian and colorectal, but not breast, carcinomas, despite previous notions. Importantly, we found distinct responses between epithelial- and mesenchymal-like ovarian cancers to therapeutic regimes administered with or without paclitaxelin vivo and demonstrated that mesenchymal-like tumours do not always show resistance to chemotherapy. EMT scoring is thus a promising, versatile tool for the objective and systematic investigation of EMT roles and dynamics in cancer progression, treatment response and survival.

The phenotypic transformation of well-differentiated epithelial carcinoma into a mesenchymal-like state provides cancer cells with the ability to disseminate locally and to metastasise. Different degrees of epithelial–mesenchymal transition (EMT) have been found to occur in carcinomas from breast, colon and ovarian carcinoma (OC), among others. Numerous studies have focused on bona fide epithelial and mesenchymal states but rarely on intermediate states. In this study, we describe a model system for appraising the spectrum of EMT using 43 well-characterised OC cell lines. Phenotypic EMT characterisation reveals four subgroups: Epithelial, Intermediate E, Intermediate M and Mesenchymal, which represent different epithelial–mesenchymal compositions along the EMT spectrum. In cell-based EMT-related functional studies, OC cells harbouring an Intermediate M phenotype are characterised by high N-cadherin and ZEB1 expression and low E-cadherin and ERBB3/HER3 expression and are more anoikis-resistant and spheroidogenic. A specific Src-kinase inhibitor, Saracatinib (AZD0530), restores E-cadherin expression in Intermediate M cells in in vitro and in vivo models and abrogates spheroidogenesis. We show how a 33-gene EMT Signature can sub-classify an OC cohort into four EMT States correlating with progression-free survival (PFS). We conclude that the characterisation of intermediate EMT states provides a new approach to better define EMT. The concept of the EMT Spectrum allows the utilisation of EMT genes as predictive markers and the design and application of therapeutic targets for reversing EMT in a selective subgroup of patients.

Evolutionarily conserved across eukaryotic cells, macroautophagy (herein autophagy) is an intracellular catabolic degradative process targeting damaged and superfluous cellular proteins, organelles, and other cytoplasmic components. Mechanistically, it involves formation of double-membrane vesicles called autophagosomes that capture cytosolic cargo and deliver it to lysosomes, wherein the breakdown products are eventually recycled back to the cytoplasm. Dysregulation of autophagy often results in various disease manifestations, including neurodegeneration, microbial infections, and cancer. In the case of cancer, extensive attention has been devoted to understanding the paradoxical roles of autophagy in tumor suppression and tumor promotion. In this review, while we summarize how this self-eating process is implicated at various stages of tumorigenesis, most importantly, we address the link between autophagy and hallmarks of cancer. This would eventually provide a better understanding of tumor dependence on autophagy. We also discuss how therapeutics targeting autophagy can counter various transformations involved in tumorigenesis. Finally, this review will provide a novel insight into the mutational landscapes of autophagy-related genes in several human cancers, using genetic information collected from an array of cancers.