Rational Approach to Finding Genes Encoding Molecular Biomarkers: Focus on Breast Cancer.

Area of Science:

• Biomarker Discovery
• Computational Biology
• Oncology

Background:

• Early cancer detection significantly improves patient survival and treatment efficacy.
• Molecular biomarkers, particularly extracellular ones, are crucial for non-invasive cancer diagnosis.
• Existing methods for biomarker discovery can be enhanced by integrating transcriptional regulation and gene co-expression data.

Purpose of the Study:

• To develop and test a data mining strategy for predicting novel molecular cancer biomarkers.
• To identify subtype-specific biomarkers for breast cancer using a 'guilt by association' approach.
• To uncover extracellular protein markers suitable for early, non-invasive cancer detection.

Main Methods:

• Analysis of transcription factor networks, functional pathways, and protein co-expression data for 23 known breast cancer proteins.
• Utilizing a 'guilt by association' strategy starting with intracellular and transmembrane protein 'seeds'.
• Prediction of novel extracellularly targeted or secreted protein biomarkers associated with specific breast cancer molecular subtypes.

Main Results:

• Prediction of 150 novel biomarker genes associated with three major breast cancer molecular subtypes.
• Identification of 114 markers linked to basal, 48 to luminal, and 7 to Her2-positive breast cancer.
• Demonstration that the approach can identify extracellular markers from intracellular starting points.

Conclusions:

• The 'guilt by association' data mining approach effectively predicts novel extracellular cancer biomarkers.
• This strategy is applicable beyond breast cancer, offering a widely applicable method for biomarker mining.
• The identified markers hold potential for improving early, non-invasive diagnosis of breast cancer subtypes.