Tumor collagens predict genetic features and patient outcomes

Area of Science:

• Oncology
• Cancer Biology
• Genomics

Background:

• The tumor extracellular matrix (ECM) critically influences cancer progression and patient outcomes.
• Collagens are the primary components of the tumor ECM, but their role in cancer is not fully understood.
• Understanding collagen composition changes and their impact on cancer is essential for biomarker and therapeutic development.

Purpose of the Study:

• To investigate the role of collagen gene expression in classifying solid tumors.
• To explore the association between collagen composition and patient survival, tumor microenvironment, and molecular alterations.
• To develop predictive models for cancer characteristics based on collagen expression.

Main Methods:

• Clustering of RNA expression data for 43 collagen genes across solid tumors from The Cancer Genome Atlas (TCGA).
• PanCancer analysis to assess collagen's ability to identify tissue of origin.
• Cancer-type-specific clustering to correlate collagen expression with survival, immunoenvironments, somatic mutations, copy number variations (CNVs), and aneuploidy.
• Development of a machine learning classifier to predict aneuploidy and chromosome arm copy number alteration (CNA) status using collagen expression.

Main Results:

• Collagen expression patterns alone can identify the tissue of origin for solid tumors.
• Collagen clustering within cancer types strongly correlates with patient survival, immune cell infiltration, somatic mutations, and chromosomal instability.
• A machine learning model accurately predicts aneuploidy and CNA status based solely on collagen expression in numerous cancer types.
• These findings highlight a significant link between the collagen ECM and specific molecular alterations in cancer.

Conclusions:

• Collagen gene expression serves as a powerful intrinsic classifier for solid tumors, revealing tissue of origin and predicting clinical outcomes.
• The tumor collagen ECM is intricately linked to the cancer's molecular landscape, including genetic mutations and chromosomal abnormalities.
• Targeting the collagen ECM or utilizing collagen expression signatures offers promising avenues for improving cancer prognosis and developing novel therapeutic strategies.