Breast Cancer Prognostics Using Multi-Omics Data.

Area of Science:

• Oncology
• Bioinformatics
• Genomics

Background:

• Breast cancer is a significant global health challenge, impacting numerous women.
• Accurate prediction of patient survival is crucial for effective clinical management.

Purpose of the Study:

• To evaluate the predictive power of different omics data types for ten-year breast cancer survival.
• To investigate if data fusion of multiple omics datasets enhances survival prediction accuracy.

Main Methods:

• Collected copy number variation, gene expression, proteome, and phosphoproteome data from 77 breast cancer patients.
• Developed predictive models for ten-year survival using individual omics data types.
• Applied ten data fusion techniques to combined multi-omics datasets.

Main Results:

• Proteome data models achieved the highest predictivity (mean AUC = 0.725), outperforming other individual omics data types.
• Data fusion techniques did not yield superior predictive performance compared to models using only proteome data.

Conclusions:

• High-quality proteomic data is highly effective for breast cancer survival prediction.
• Integrating multiple omics data types does not necessarily improve predictive accuracy for this clinical outcome.