Effect of selection bias on two sample summary data based Mendelian randomization.

Area of Science:

• Genetics
• Biostatistics
• Epidemiology

Background:

• Mendelian randomization (MR) is increasingly used to infer causal relationships between exposures and traits.
• Current MR practices often involve selecting instrument single nucleotide polymorphisms (SNPs) based on exposure Genome-Wide Association Studies (GWAS) summary statistics.
• This selection process can introduce bias, potentially invalidating MR findings.

Purpose of the Study:

• To highlight the selection bias inherent in common MR instrument selection practices.
• To demonstrate how this bias affects popular MR methods like summary data-based MR (SMR) and two-sample MR Steiger.
• To propose a novel, more powerful, and simpler alternative to the SMR method.

Main Methods:

• Analysis of selection bias in instrument SNP selection for Mendelian randomization.
• Evaluation of the impact of selection bias on the SMR and MR Steiger methods.
• Development and proposal of a new MR method as an alternative to SMR.

Main Results:

• The common practice of selecting instrument SNPs from GWAS summary statistics can lead to selection bias.
• This bias can cause the SMR method to be conservative and the MR Steiger method to be either conservative or liberal.
• A simple and more powerful alternative to the SMR method is presented.

Conclusions:

• Selection bias is a critical issue in Mendelian randomization that can compromise the validity of causal inference.
• Existing popular MR methods are susceptible to this bias.
• The proposed novel MR method offers a more robust and powerful approach compared to SMR.