METEOR: a data-adaptive Mendelian randomization method for powerful detection of shared and specific exposures...

Area of Science:

• Genetics and Epidemiology
• Statistical Genetics
• Causal Inference

Background:

• Accurate identification of causal exposures for multimorbidity is essential for effective co-prevention and co-management strategies.
• Existing multi-outcome Mendelian randomization (MR) methods have limitations, including data input restrictions, inability to handle sample overlap, reliance on pre-selected instrumental variables (IVs), and failure to account for horizontal pleiotropy.

Purpose of the Study:

• To propose METEOR, a novel MR method for jointly modeling one exposure and multiple outcomes to identify shared and outcome-specific causal exposures.
• To address limitations of existing multi-outcome MR methods by accounting for sample overlap, utilizing diverse GWAS datasets, adaptively selecting IVs, and modeling horizontal pleiotropy.

Main Methods:

• METEOR employs a joint-likelihood framework with a scalable, sampling-based algorithm to infer causal effects using summary statistics.
• The method integrates multi-outcome analysis, sample overlap correction, adaptive instrumental variable selection from correlated single-nucleotide polymorphisms, and explicit horizontal pleiotropy modeling.

Main Results:

• Simulations demonstrate that METEOR provides well-calibrated P-values and achieves significant power improvements (55.33% global, 56.50% single-outcome tests) over existing MR methods.
• Real-data applications show METEOR yields accurate causal effect estimates, reduces false positives by 18.75%, and identifies potential causal exposures for co-managing diseases.

Conclusions:

• METEOR is a powerful and flexible tool for multi-outcome MR analysis, capable of identifying shared and specific causal exposures.
• The findings suggest that interventions targeting BMI could benefit co-management of cardiovascular and gastrointestinal diseases, while blood pressure control may aid multimorbidity across cardiovascular, gastrointestinal, and mental disorders.