Robust Mendelian randomization in the presence of residual population stratification, batch effects and horizontal pleiotropy

Carlos Cinelli,Eleazar Eskin,Nathan Lapierre,Sriram Sankararaman,Brian L Hill

doi:10.1038/s41467-022-28553-9

Carlos Cinelli, Eleazar Eskin + Show 3 more

Open Access

https://doi.org/10.1038/s41467-022-28553-9

Copy DOI

Abstract

Mendelian Randomization (MR) studies are threatened by population stratification, batch effects, and horizontal pleiotropy. Although a variety of methods have been proposed to mitigate those problems, residual biases may still remain, leading to highly statistically significant false positives in large databases. Here we describe a suite of sensitivity analysis tools that enables investigators to quantify the robustness of their findings against such validity threats. Specifically, we propose the routine reporting of sensitivity statistics that reveal the minimal strength of violations necessary to explain away the MR results. We further provide intuitive displays of the robustness of the MR estimate to any degree of violation, and formal bounds on the worst-case bias caused by violations multiple times stronger than observed variables. We demonstrate how these tools can aid researchers in distinguishing robust from fragile findings by examining the effect of body mass index on diastolic blood pressure and Townsend deprivation index.

Highlights

Mendelian Randomization (MR) studies are threatened by population stratification, batch effects, and horizontal pleiotropy
Let X denote a set of observed “control” covariates which accounts for potential violations of the MR assumptions, such as population stratification, batch effects and traits that could block putative horizontal pleiotropic pathways[20]
An example for which this is the case is depicted in the directed acyclic graph (DAG) of Fig. 1a—in this example there are no pleiotropic pathways, and there is confounding due to population structure, adjusting for X is sufficient for eliminating all biases

Summary

Introduction

Mendelian Randomization (MR) studies are threatened by population stratification, batch effects, and horizontal pleiotropy. The technique of MR has become a standard tool for inferring causal relationships, with numerous applications published in medical, genetic and epidemiological journals[6–14] This growth has been accelerated by the availability of large genetic databases[15] and Genome-Wide Association Studies (GWAS) linking many genetic variants to complex phenotypes[8]. In the case of horizontal pleiotropy, researchers are advised to perform alternative analyses that rely on modified identification assumptions (such as MREgger[39], MR-PRESSO40, MR-MBE41, MR-Mix[42], MR-GENIUS43, among a plethora of variations) These methods have proved useful for partially mitigating these problems, residual biases may still remain[9,44]. Since those biases are impervious to sample size, they may lead to highly statistically significant false findings with large genomic data, as we demonstrate later via simulations

Methods

Results

Conclusion