Analysis of secondary phenotypes in multigroup association studies

Area of Science:

• Genetics
• Biostatistics
• Medical Informatics

Background:

• Case-control studies are limited for complex diseases with multiple subtypes.
• Multigroup studies collect extensive secondary outcomes alongside primary disease status.
• Existing methods may not adequately account for multigroup sampling in genetic analyses.

Purpose of the Study:

• Develop a general regression framework for secondary phenotypes in multigroup association studies.
• Address limitations of standard analyses in the presence of disease heterogeneity.
• Improve the accuracy of genome-wide association studies for complex diseases.

Main Methods:

• A conditional regression model for secondary outcomes given multigroup status and covariates.
• Generalized estimation equations (GEE) for parameter estimation.
• Validation through simulations and analysis of Alzheimer's Disease Neuroimaging Initiative (ADNI) data.

Main Results:

• The proposed framework appropriately adjusts for multigroup sampling schemes.
• Demonstrates the impact of sampling bias on standard genome-wide association analyses.
• Provides a more robust method for analyzing secondary phenotypes in complex diseases.

Conclusions:

• The developed regression framework enhances the analysis of secondary outcomes in multigroup studies.
• Accounting for multigroup sampling is crucial for accurate genetic association findings.
• This approach offers improved statistical power and reliability for complex disease research.