MR-EILLS: an invariance-based Mendelian randomization method integrating multiple heterogeneous GWAS summary...

Area of Science:

• Genetics
• Statistical Genetics
• Epidemiology

Background:

• Genetic structure diversity causes heterogeneity in Genome-Wide Association Studies (GWAS) summary datasets.
• Integrating heterogeneous GWAS data complicates the inference of causal exposure-outcome effects.

Purpose of the Study:

• To introduce Mendelian randomization using Environment Invariant Linear Least Squares (MR-EILLS) for robust causal inference across heterogeneous populations.
• To develop a method capable of handling both univariate and multivariate scenarios, including invalid instrumental variables.

Main Methods:

• MR-EILLS utilizes environment invariant linear least squares to detect causal relationships invariant across populations.
• The method accommodates violations of instrumental variable assumptions like exchangeability and exclusion restriction.
• It provides unbiased causal effect estimations for single or multiple exposures, regardless of instrument validity.

Main Results:

• MR-EILLS demonstrates superior estimation accuracy, stable Type I error rates, and higher statistical power compared to traditional Mendelian randomization and meta-analysis methods.
• Application to 11 blood cell traits and 20 disease outcomes across five ancestries identified expected and novel causal links.
• Results revealed biologically interpretable causal relationships and additional associations supported by observational studies.

Conclusions:

• MR-EILLS is a powerful and accurate Mendelian randomization tool for analyzing heterogeneous GWAS data.
• The method enhances the reliability of causal inference in multi-population genetic studies.
• This approach facilitates the discovery of genetic causal relationships between complex traits and diseases across diverse ancestries.