Extending the MR-Egger method for multivariable Mendelian randomization to correct for both measured and unmeasured...

Area of Science:

• Epidemiology
• Statistical Genetics
• Causal Inference

Background:

• Mendelian randomization (MR) uses genetic variants as instrumental variables to infer causal relationships.
• Standard MR relies on strong instrumental variable assumptions, often violated by pleiotropy (genetic variants affecting outcomes through unintended pathways).
• Existing methods like multivariable MR and MR-Egger address measured and unmeasured pleiotropy, respectively.

Purpose of the Study:

• To extend the MR-Egger method to a multivariable setting.
• To develop a method that corrects for both measured and unmeasured pleiotropy simultaneously.
• To improve causal estimation and statistical power in genetic analyses with complex pleiotropic effects.

Main Methods:

• Development of the multivariable MR-Egger method.
• Theoretical derivations and a simulation study to evaluate method performance.
• Application to investigate the causal effect of high-density lipoprotein cholesterol on coronary heart disease risk.

Main Results:

• The multivariable MR-Egger method offers advantages over univariable MR-Egger.
• Demonstrated improved plausibility of assumptions for consistent causal estimation.
• Showcased enhanced power to detect causal effects when assumptions are met.

Conclusions:

• The multivariable MR-Egger method effectively corrects for both measured and unmeasured pleiotropy.
• This approach is valuable for analyzing high-dimensional data with highly related risk factors.
• It serves as a robust sensitivity analysis tool when genetic variants exhibit known pleiotropic effects.