Epithelial-to-mesenchymal transition signature assessment in colorectal cancer quantifies tumour stromal content rather than true transition
Session type: Poster / e-Poster / Silent Theatre session
Epithelial-to-mesenchymal transition (EMT) is a well-described process in cancer, whereby epithelial tumour cells transition to a mesenchymal biology. Gene expression signatures have been developed to characterise the EMT process and assign samples in cohorts a relative EMT score. In this study, we will examine the use of EMT signatures in data from both colorectal cancer cell lines and tumour samples with a view to delineating the cellular source of this signalling.
We used single-sample gene set enrichment analysis (ssGSEA) EMT signature scores assigned to transcriptional data from 3 cohorts; the 61 colorectal cancer (CRC) cell lines within the Cancer Cell Line Encylocpedia (CCLE), 215 stage II colon cancer patient samples (E-MTAB-863) and 156 stage II/III colorectal cancer patient samples (GSE103479). H&E slides from the GSE103479 cohort were quantified for histological features and percentage of epithelial/stromal cells and aligned to EMT signature scores.
From in silico analysis of CRC cell lines from the CCLE, we demonstrate that EMT signatures preferentially identify cell lines with a fibroblast origin, which have higher relative EMT scores than any epithelial cell line in this collection. Furthermore, we observe that increased EMT scores significantly correlate with higher fibroblast, cancer-associated fibroblast and general stroma scores when assessing tumour tissue samples using both in silico analyses and histological assessment (in all cases p < 0.00001).
These results emphasise the importance of combining histology assessment, informatics and mechanistic biology to the interpretation of signatures aimed at characterising EMT, or other signalling biology. In the era of precision medicine, an interdisciplinary approach incorporating pathological, bioinformatic and molecular biology is essential to advance our understanding of cancer from high-throughput molecular data.