Comparative Study of Transcriptomics-Based Scoring Metrics for the Epithelial-Hybrid-Mesenchymal Spectrum

Area of Science:

• Cellular Biology
• Cancer Research
• Bioinformatics

Background:

• Epithelial-mesenchymal transition (EMT) is a cellular process involved in development, wound healing, and cancer metastasis.
• Cancer cells undergoing EMT display increased drug resistance, tumor-initiation potential, and immune evasion.
• Hybrid epithelial/mesenchymal (E/M) phenotypes can be more aggressive than pure epithelial or mesenchymal states, impacting patient prognosis.

Purpose of the Study:

• To compare existing transcriptomics-based metrics for quantifying EMT status in cancer cells.
• To evaluate the accuracy of these metrics in identifying hybrid E/M phenotypes.
• To identify the most effective method for assessing EMT continuum and hybrid states.

Main Methods:

• Comparison of three distinct transcriptomics signature-based metrics for scoring EMT on a continuum.
• Analysis of EMT variability across different tumor types.
• Utilizing multinomial logistic regression (MLR) to distinguish pure hybrid E/M cells from mixtures.

Main Results:

• The three compared EMT scoring methods showed good concordance in quantifying EMT extent.
• Significant variability in EMT extent was observed across tumor types, with some enriched for hybrid E/M signatures.
• The MLR-based metric effectively distinguished pure hybrid E/M cells from mixed E/M populations.

Conclusions:

• All three methods provide a general trend of EMT status, but MLR offers superior resolution for hybrid E/M phenotypes.
• The MLR method is advantageous due to its use of fewer predictors and its ability to differentiate pure hybrid E/M cells at the single-cell level from mixed populations.