Tumour-specific Causal Inference Discovers Distinct Disease Mechanisms Underlying Cancer Subtypes

Area of Science:

• Oncology
• Computational Biology
• Genomics

Background:

• Cancer arises from somatic genome alterations (SGAs) disrupting cellular signaling.
• Tumor heterogeneity in progression and therapy response is linked to distinct pathway aberrations.
• Identifying shared disease mechanisms is crucial for advancing precision oncology.

Purpose of the Study:

• To develop a novel computational framework for uncovering distinct combinations of aberrant signaling pathways in tumors.
• To apply this framework to identify clinically relevant cancer subtypes based on pathway aberrations.

Main Methods:

• Utilized a tumor-specific causal inference algorithm (TCI) to link SGAs and differentially expressed genes (DEGs) in TCGA data.
• Employed a network-based method to identify DEG modules representing co-regulated pathways.
• Used module gene expression as a proxy for pathway activation status.

Main Results:

• Successfully classified breast cancers (BRCAs) into five subgroups and glioblastoma multiformes (GBMs) into six subgroups.
• These subgroups demonstrated distinct combinations of aberrant signaling pathways.
• Identified patient groups with significantly different survival patterns, validating the clinical relevance of the subtypes.

Conclusions:

• The developed computational framework effectively identifies distinct combinations of aberrant signaling pathways in cancer.
• This approach can stratify tumors into clinically meaningful subtypes, offering potential for improved precision oncology strategies.