Comparative study of transcriptomics-based scoring metrics for the epithelial-hybrid-mesenchymal spectrum

Abstract

The Epithelial-mesenchymal transition (EMT) is a cellular process implicated in embryonic development, wound healing, and pathological conditions such as cancer metastasis and fibrosis. Cancer cells undergoing EMT exhibit enhanced aggressive behavior characterized by drug resistance, tumor-initiation potential, and the ability to evade immune system. Recent in silico, in vitro, and in vivo evidence indicates that EMT is not an all-or-none process; instead, cells stably acquire one or more hybrid epithelial/ mesenchymal (E/M) phenotypes which often can be more aggressive than purely epithelial or mesenchymal cell populations. Thus, the EMT status of cancer cells can prove to be a critical estimate of patient prognosis. Recent attempts have employed different transcriptomics signatures to quantify EMT status in cell lines and patient tumors. However, a comprehensive comparison of these methods, including their accuracy in identifying cells in the hybrid E/M phenotype(s), is lacking. Here, we compare three distinct metrics that score EMT on a continuum, based on the transcriptomics signature of individual samples. Our results demonstrate that these methods exhibit good concordance among themselves in quantifying the extent of EMT in a given sample. Moreover, scoring EMT using any of the three methods discerned that cells undergo varying extents of EMT across tumor types. Separately, our analysis also identified tumor types with maximum variability in terms of EMT and associated an enrichment of hybrid E/M signatures in these samples. Moreover, we also found that the multinomial logistic regression (MLR) based metric was capable of distinguishing between 'pure' individual hybrid E/M vs. mixtures of epithelial (E) and mesenchymal (M) cells. Our results, thus, suggest that while any of the three methods can indicate a generic trend in the EMT status of a given cell, the MLR method has two additional advantages: a) it uses a small number of predictors to calculate the EMT score, and b) it can predict from the transcriptomic signature of a population whether it is comprised of 'pure' hybrid E/M cells at the single-cell level or is instead an ensemble of E and M cell subpopulations.

A mechanistic model captures the emergence and implications of non-genetic heterogeneity and reversible drug resistance in ER+ breast cancer cells

“…Next, for a more comprehensive analysis of such correlations, we investigated pairwise correlations among expression levels of ESR1, canonical mesenchymal (SLUG, VIM, ZEB1) and epithelial (CDH1) genes, three EMT scoring metrics (76GS, KS, MLR) (39), and four gene set activity estimation via ssGSEA scores (ERα early response, ERα late response, Tamoxifen resistance and Hallmark EMT). Across patient samples irrespective of whether the samples were micro-dissected tumour biopsies or whole tissue tumour biopsies, we observed an expected positive correlation among EMT metrics, mesenchymal markers and ssGSEA scores for hallmark EMT ( Fig 2D ).…”

Analysis of immune subtypes across the epithelial-mesenchymal plasticity spectrum

“…Higher MLR or KS scores represent more mesenchymal samples while this is the inverse for GS76 scores (lower = more mesenchymal). Thus, KS and MLR scores of samples in a given dataset correlate positively with one another, and both KS and MLR correlate negatively with GS76 scores, as observed across multiple datasets (Chakraborty et al 2020).…”

“…TCGA datasets: TCGA gene expression datasets were obtained from https://xenabrowser.net/datapages/ CCLE dataset: CCLE gene expression data was downloaded from https://portals.broadinstitute.org/ccle/data EMT scoring: Three different EMT scoring methods -KS, MLR, GS76 -were used to score samples separately in all the datasets as previously described (Chakraborty et al 2020).…”

“…In this study, we combined three novel EMT scoring metrics (Chakraborty et al 2020) with immune subtype analysis from the iATLAS algorithm (Thorsson et al 2018; Eddy et al 2020) to understand the relationships between EMT and immune signature within and between cancer types. Our analyses demonstrate heterogeneity in both E/M phenotype and immune signature within a single cancer type.…”

Investigating epithelial-mesenchymal heterogeneity of tumors and circulating tumor cells with transcriptomic analysis and biophysical modeling

“…We calculated the EMT scores of multiple publicly available datasets associated with CTCs, using three different EMT metrics – 76GS, KS and MLR [27]. These three methods use different gene lists and algorithms; the more epithelial a sample is, the lower its KS score (on a scale of [−1, 1]) or MLR score (on a scale of [0, 2]) and the higher is its 76GS score (no a priori defined range of values).…”