Leveraging ML for profiling lipidomic alterations in breast cancer tissues: a methodological perspective

Area of Science:

• Biochemistry
• Oncology
• Computational Biology

Background:

• Breast cancer exhibits complex metabolic alterations.
• Understanding lipid dysregulation is crucial for cancer research.

Purpose of the Study:

• To investigate metabolite profile alterations in breast cancer tissues using machine learning and statistical analysis.
• To identify a lipid signature associated with breast cancer and its subtypes.

Main Methods:

• Combined machine learning and statistical analyses for feature selection.
• Univariate and multivariate analyses of lipid profiles.
• Lipidomics analysis on original and validation datasets.

Main Results:

• Identified a significant lipid signature in breast cancer tissues.
• Elevated saturated and monounsaturated phospholipids observed.
• Reduced triacylglycerol levels in cancer tissues compared to non-cancerous tissues.
• Specific alterations in phosphatidylcholine (PC 30:0) levels linked to breast cancer subtypes (HER2, ER, PR status).

Conclusions:

• Metabolomic analysis provides critical insights into breast cancer development.
• Identified lipid alterations may serve as biomarkers for early detection.
• Findings can inform treatment evaluation strategies for breast cancer.