Implicating causal brain imaging endophenotypes in Alzheimer's disease using multivariable IWAS and GWAS summary data

Area of Science:

• Neurogenetics
• Computational Biology
• Alzheimer's Disease Research

Background:

• Heritable brain phenotypes are increasingly implicated in Alzheimer's Disease (AD) pathogenesis.
• Discovering causal brain changes requires integrating genetic and neuroimaging data.
• Existing methods like univariate Imaging Wide Association Study (UV-IWAS) have limitations, including inconsistent effect estimation and inflated Type I errors due to genetic pleiotropy.

Purpose of the Study:

• To develop and implement a multivariate extension to the Imaging Wide Association Study (IWAS) model, termed MV-IWAS.
• To consistently estimate and test causal effects of multiple brain imaging endophenotypes on AD, accounting for pleiotropic and correlated SNPs.
• To extend MV-IWAS to incorporate variant-specific direct effects on AD, allowing for testing of residual pleiotropy.

Main Methods:

• Implementation of a multivariate extension to the IWAS model (MV-IWAS).
• Incorporation of variant-specific direct effects, analogous to Egger regression Mendelian Randomization.
• Development of a convenient approach using publicly available Genome-Wide Association Study (GWAS) summary data and a reference panel.
• Validation through simulations using individual-level or summary data.
• Application to 1578 heritable imaging derived phenotypes (IDPs) from the UK Biobank.

Main Results:

• MV-IWAS demonstrated well-controlled Type I errors and superior statistical power compared to UV-IWAS in the presence of pleiotropic SNPs.
• The method identified numerous imaging derived phenotypes (IDPs) flagged as potential false positives by UV-IWAS.
• MV-IWAS uncovered additional causal neuroimaging phenotypes associated with AD, strongly supported by existing literature.

Conclusions:

• MV-IWAS provides a robust and powerful method for discovering causal neuroimaging phenotypes in Alzheimer's Disease.
• The approach effectively handles genetic pleiotropy and correlated SNPs, overcoming limitations of previous univariate methods.
• MV-IWAS, utilizing publicly available GWAS summary data, offers a practical tool for advancing AD research by identifying novel imaging-based risk factors.