Exon-Skipping-Based Subtyping of Colorectal Cancers

Area of Science:

• Molecular biology
• Genomics
• Cancer research

Background:

• Colorectal cancer (CRC) molecular subtyping is crucial for prognosis and treatment but current methods are complex and costly.
• Existing Consensus Molecular Subtype (CMS) classification relies on gene-level expression, facing challenges in clinical application due to uncertainty and expense.
• Alternative splicing (AS), a key driver of transcriptome diversity, has been underexplored for CRC subtyping.

Purpose of the Study:

• To develop and validate an alternative splicing-based framework for colorectal cancer subtyping.
• To create a clinically adaptable classifier for CRC subtypes using differential exon usage.
• To demonstrate the efficacy of AS-based subtyping compared to traditional gene expression methods.

Main Methods:

• Utilized unsupervised clustering to associate alternative splicing categories with Consensus Molecular Subtypes (CMS).
• Developed a classifier using gene-level expression data for ground truth and selected features via bootstrapping and L1-penalized estimation.
• Validated the subtype prediction framework on independent colorectal cancer datasets (Indivumed and The Cancer Genome Atlas).

Main Results:

• A colorectal cancer (CRC) subtype identifier was developed based on 29 exon-skipping events.
• The AS-based classifier demonstrated accurate classification of unseen tumors.
• Achieved more precise differentiation of subtypes with distinct biological and prognostic features compared to gene expression classifiers.

Conclusions:

• A novel framework utilizing a small set of exon-skipping events reliably classifies colorectal cancer (CRC) subtypes.
• The developed method is suitable for clinical application, enabling classification from individual patient specimens.
• Alternative splicing offers a promising avenue for improved and clinically relevant CRC subtyping.