Detection for pathway effect contributing to disease in systems epidemiology with a case-control design

Area of Science:

• Systems epidemiology
• Statistical inference
• Network topology analysis

Background:

• Identifying specific pathway effects is crucial for understanding disease etiology.
• Existing methods for distinguishing pathway topology between disease states require further development for population-level inference.
• Advanced statistical approaches are needed to capture complex relationships among risk factors for disease.

Purpose of the Study:

• To develop powerful statistical methods for identifying specific disease-contributing pathways.
• To capture complex interrelationships among risk factors.
• To enhance statistical inference in systems epidemiology.

Main Methods:

• Development of Pathway Effect Measures (PEM).
• Utilized simulation studies to assess method stability and power.
• Employed bootstrap-based methods, including bias-corrected bootstrap confidence intervals (CI).
• Applied the method to real-world Acute Myeloid Leukemia (AML) patient data.

Main Results:

• Pathway Effect Measures (PEM) demonstrated stability in simulations.
• Bootstrap-based methods, particularly bias-corrected bootstrap CI, showed superior power.
• Successfully identified a specific pathway (Treg→TGFβ→Th17) associated with AML (p < 0.05).
• Revealed impaired Th17-Treg correlation balance in AML patients, suggesting a role in pathogenesis.

Conclusions:

• Bootstrap-based PEM are validated as powerful tools for detecting disease-specific pathway effects.
• These methods offer novel insights into disease mechanisms.
• Provides a more comprehensive approach to studying pathway effects in disease.