Random forest algorithm identifies miRNA signatures for breast cancer detection and classification from patient urine...

Area of Science:

• Oncology
• Genomics
• Biomarker Discovery

Background:

• Breast cancer is a leading cause of cancer in women, with low mammography screening participation due to pain.
• Liquid biopsies offer a less invasive alternative for early cancer detection.
• MicroRNAs (miRNAs) in bodily fluids are potential biomarkers for various cancers.

Purpose of the Study:

• To develop and validate a urine-based microRNA sequencing method for breast cancer detection.
• To identify subtype-specific microRNA signatures for luminal A, luminal B, HER2-enriched, and triple-negative breast cancer.
• To implement a novel random forest algorithm for analyzing microRNA expression patterns.

Main Methods:

• MicroRNA sequencing was performed on 82 urine samples (32 from breast cancer patients, 50 from healthy controls).
• A random forest algorithm was utilized to identify a diagnostic microRNA signature.
• Analysis included patients with different intrinsic breast cancer subtypes.

Main Results:

• A signature of 275 microRNAs in urine was identified, enabling invasive breast cancer detection.
• Distinct microRNA expression patterns were found for major intrinsic breast cancer subtypes.
• The study validated microRNA sequencing in urine as a viable technique for breast cancer detection.

Conclusions:

• Urine microRNA sequencing is a validated, non-invasive method for breast cancer detection.
• This approach can differentiate between major breast cancer subtypes.
• It presents a new, painless procedure for breast cancer screening and treatment monitoring.