Breast cancer patient stratification using a molecular regularized consensus clustering method.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Breast cancer exhibits significant heterogeneity, impacting patient prognosis and treatment response.
• Accurate subtype determination is crucial for personalized medicine in breast cancer care.
• Integrative genomics approaches using multi-modal data are emerging for patient stratification.

Purpose of the Study:

• To develop a novel method for integrating diverse genomic data (numerical and categorical) for breast cancer patient stratification.
• To address limitations of existing consensus clustering algorithms in handling mixed data types.
• To improve the accuracy of breast cancer subtyping for personalized treatment strategies.

Main Methods:

• Developed a mathematical formulation for integrative clustering of multi-source data.
• Introduced Molecular Regularized Consensus Patient Stratification (MRCPS), a novel consensus clustering method.
• Applied MRCPS to TCGA breast cancer datasets, utilizing numerical and categorical data with flexible similarity metrics.

Main Results:

• MRCPS effectively integrates both numerical and categorical data for clustering.
• The method demonstrated superior performance in data aggregation compared to traditional approaches.
• MRCPS showed significant clinical relevance in differentiating patient outcomes and predicting prognosis.

Conclusions:

• MRCPS offers a universal and effective solution for integrative genomics studies requiring mixed-data clustering.
• The developed method enhances breast cancer patient stratification, leading to better prediction of clinical outcomes.
• This approach holds promise for advancing personalized medicine in oncology and beyond.