Capturing biomarkers associated with Alzheimer disease subtypes using data distribution characteristics.

Area of Science:

• Biomedical research
• Genomics
• Statistical modeling

Background:

• Late-onset Alzheimer disease (AD) is heterogeneous with diverse risk factors and clinical presentations.
• Current biomarker evaluation methods (e.g., fold change, AUC) are suboptimal for subtype analysis.
• Identifying specific AD subtypes is vital for understanding disease mechanisms and developing targeted therapies.

Purpose of the Study:

• To introduce a novel statistical metric for identifying biomarkers associated with disease subtypes.
• To address the limitations of existing methods in analyzing heterogeneous disease data.
• To reveal potential biomarkers for subtypes of complex diseases like Alzheimer disease.

Main Methods:

• Developed and applied the Bimodality Coefficient Difference (BCD) metric.
• BCD assesses differences in data distribution bimodality between cases and controls.
• Validated BCD using large-scale synthetic data and applied it to Alzheimer disease gene expression data.

Main Results:

• Demonstrated the weaknesses of traditional fold change and AUC methods for subtype evaluation.
• BCD effectively identified analytes associated with specific disease subsets in synthetic data.
• Applied BCD to gene expression data, confirming its utility in revealing novel AD biomarkers for heterogeneous subtypes.

Conclusions:

• The Bimodality Coefficient Difference (BCD) is a robust metric for identifying biomarkers in heterogeneous diseases.
• BCD offers a significant advancement over traditional methods for subtype biomarker discovery.
• This approach can accelerate the understanding and treatment of complex diseases like Alzheimer disease.