MRBEE: A bias-corrected multivariable Mendelian randomization method

Area of Science:

• Genetics
• Epidemiology
• Biostatistics

Background:

• Mendelian randomization (MR) infers causality using genetic variants as instrumental variables.
• Existing MR methods face challenges like weak instruments, horizontal pleiotropy, and sample overlap, potentially biasing results.
• Genome-wide association studies (GWAS) provide summary statistics, making MR increasingly popular for causal inference.

Purpose of the Study:

• To introduce MRBEE (MR using bias-corrected estimating equation), a novel multivariable MR method.
• To address and simultaneously correct for weak instrument bias, sample overlap bias, and horizontal pleiotropy.
• To improve the accuracy and robustness of causal effect estimation in genetic epidemiology.

Main Methods:

• MRBEE employs bias-corrected estimating equations for multivariable MR analysis.
• The method is designed to simultaneously handle weak instrument bias, sample overlap, and horizontal pleiotropy.
• Simulations and real-world data analyses were conducted to evaluate MRBEE's performance.

Main Results:

• MRBEE demonstrated nearly unbiased causal effect estimates in simulations and real data.
• The method showed well-controlled type I error rates and superior power compared to existing robust methods.
• MRBEE is computationally efficient, making it practical for large-scale genetic studies.

Conclusions:

• MRBEE offers a robust and efficient approach for multivariable Mendelian randomization studies.
• The method provides valuable insights into pleiotropy and enhances causal inference accuracy.
• Real data analyses identified 42 novel horizontal pleiotropic loci associated with myopia, schizophrenia, and coronary artery disease.