[Paper Review] A Common Gene Expression Signature Analysis Method for Multiple Types of Cancer

Yingcheng Sun, Xiangru Liang+1 more|arXiv (Cornell University)|Dec 28, 2019

Bioinformatics and Genomic Networks30 references4 citations

TL;DR

This study proposes a network-based method to identify a common gene expression signature across multiple cancer types using TCGA data from rectum (READ), breast (BRCA), and colon (COAD) adenocarcinomas. By integrating gene co-expression networks and clinical outcomes, the approach reveals conserved biological mechanisms linked to cancer progression, suggesting shared molecular pathways across cancers with potential for pan-cancer diagnostics and therapeutics.

ABSTRACT

Mining gene expression profiles has proven valuable for identifying signatures serving as surrogates of cancer phenotypes. However, the similarities of such signatures across different cancer types have not been strong enough to conclude that they represent a universal biological mechanism shared among multiple cancer types. Here we describe a network-based approach that explores gene-to-gene connections in multiple cancer datasets while maximizing the overall association of the subnetwork with clinical outcomes. With the dataset of The Cancer Genome Atlas (TCGA), we studied the characteristics of common gene expression of three types of cancers: Rectum adenocarcinoma (READ), Breast invasive carcinoma (BRCA) and Colon adenocarcinoma (COAD). By analyzing several pairs of highly correlated genes after filtering and clustering work, we found that the co-expressed genes across multiple types of cancers point to particular biological mechanisms related to cancer cell progression , suggesting that they represent important attributes of cancer in need of being elucidated for potential applications in diagnostic, prognostic and therapeutic products applicable to multiple cancer types.

Motivation & Objective

• To identify common gene expression signatures across multiple cancer types that may reflect shared biological mechanisms.
• To overcome the limitation of inconsistent signatures across cancer types by leveraging gene-to-gene connections in co-expression networks.
• To improve the identification of robust, clinically relevant gene signatures by integrating network topology with survival outcomes.
• To explore whether co-expressed genes across different cancers point to conserved pathways in cancer progression.
• To develop a method applicable to multiple cancer types for diagnostic, prognostic, and therapeutic applications.

Proposed method

• A network-based approach is used to model gene-to-gene interactions across multiple cancer datasets.
• The method maximizes the association between subnetworks and clinical outcomes, particularly survival data.
• Gene expression profiles from TCGA were analyzed for READ, BRCA, and COAD to identify co-expressed gene pairs.
• Filtering and clustering techniques were applied to extract highly correlated gene pairs across cancer types.
• Subnetworks were constructed based on co-expression patterns and validated for clinical relevance.
• The approach integrates multi-omics network topology with survival outcomes to prioritize biologically meaningful signatures.

Experimental results

Research questions

• RQ1 Can a common gene expression signature be identified across multiple cancer types using network-based integration of gene co-expression and clinical data?
• RQ2 Do co-expressed genes across different cancers point to shared biological mechanisms related to cancer progression?
• RQ3 To what extent do these common signatures correlate with clinical outcomes such as survival?
• RQ4 Can a unified method identify robust, pan-cancer gene signatures suitable for diagnostic or therapeutic use?
• RQ5 What specific biological pathways are enriched in the identified common gene expression signatures?

Key findings

• Highly correlated gene pairs were identified across READ, BRCA, and COAD, indicating conserved co-expression patterns.
• The co-expressed genes in the identified subnetworks are significantly associated with cancer cell progression mechanisms.
• The network-based method successfully linked subnetworks to clinical outcomes, enhancing signature reliability.
• Biological pathway analysis revealed that the common signatures are enriched in key cancer-related processes.
• The findings suggest that shared molecular mechanisms may underlie multiple cancer types, supporting pan-cancer applications.
• The method demonstrates potential for identifying universal biomarkers applicable across diverse cancer types.