Sensitivity Analyses for Robust Causal Inference from Mendelian Randomization Analyses with Multiple Genetic Variants

Abstract Background Two-sample Mendelian randomization (MR) allows the use of freely accessible summary association results from genome-wide association studies (GWAS) to estimate causal effects of modifiable exposures on outcomes. Some GWAS adjust for heritable covariables in an attempt to estimate direct effects of genetic variants on the trait of interest. One, both or neither of the exposure GWAS and outcome GWAS may have been adjusted for covariables. Methods We performed a simulation study comprising different scenarios that could motivate covariable adjustment in a GWAS and analysed real data to assess the influence of using covariable-adjusted summary association results in two-sample MR. Results In the absence of residual confounding between exposure and covariable, between exposure and outcome, and between covariable and outcome, using covariable-adjusted summary associations for two-sample MR eliminated bias due to horizontal pleiotropy. However, covariable adjustment led to bias in the presence of residual confounding (especially between the covariable and the outcome), even in the absence of horizontal pleiotropy (when the genetic variants would be valid instruments without covariable adjustment). In an analysis using real data from the Genetic Investigation of ANthropometric Traits (GIANT) consortium and UK Biobank, the causal effect estimate of waist circumference on blood pressure changed direction upon adjustment of waist circumference for body mass index. Conclusions Our findings indicate that using covariable-adjusted summary associations in MR should generally be avoided. When that is not possible, careful consideration of the causal relationships underlying the data (including potentially unmeasured confounders) is required to direct sensitivity analyses and interpret results with appropriate caution.

Download Full-text

Guidelines for performing Mendelian randomization investigations

Wellcome Open Research ◽

10.12688/wellcomeopenres.15555.2 ◽

2020 ◽

Vol 4 ◽

pp. 186 ◽

Cited By ~ 14

Author(s):

Stephen Burgess ◽

George Davey Smith ◽

Neil M. Davies ◽

Frank Dudbridge ◽

Dipender Gill ◽

...

Keyword(s):

Statistical Methods ◽

Genetic Variants ◽

Mendelian Randomization ◽

Data Sources ◽

Sensitivity Analyses ◽

Primary Analysis ◽

Data Presentation ◽

Journal Editors ◽

Robust Statistical Methods

This paper provides guidelines for performing Mendelian randomization investigations. It is aimed at practitioners seeking to undertake analyses and write up their findings, and at journal editors and reviewers seeking to assess Mendelian randomization manuscripts. The guidelines are divided into nine sections: motivation and scope, data sources, choice of genetic variants, variant harmonization, primary analysis, supplementary and sensitivity analyses (one section on robust statistical methods and one on other approaches), data presentation, and interpretation. These guidelines will be updated based on feedback from the community and advances in the field. Updates will be made periodically as needed, and at least every 18 months.

Download Full-text

The association between serum iron status and risk of asthma: a 2-sample Mendelian randomization study in descendants of Europeans

American Journal of Clinical Nutrition ◽

10.1093/ajcn/nqz162 ◽

2019 ◽

Vol 110 (4) ◽

pp. 959-968 ◽

Cited By ~ 1

Author(s):

Lulu Huang ◽

Longman Li ◽

Xiaoyu Luo ◽

Sifang Huang ◽

Qingzhi Hou ◽

...

Keyword(s):

Genetic Variants ◽

Mendelian Randomization ◽

Iron Status ◽

Age At Onset ◽

Transferrin Saturation ◽

Sensitivity Analyses ◽

Multidisciplinary Study ◽

Asthma Risk ◽

Environmental Causes ◽

Liberal Approach

ABSTRACT Background Observational studies present conflicting results about a possible association of iron status with asthma risk, pointing to potential modifiable targets for prevention. Objective The aim of this study was to use Mendelian randomization (MR) to estimate associations between iron status and asthma risk. Methods We used the Genetics of Iron Status consortium to identify genetic variants that could be used as instrumental variables for the effect of systemic iron status. The following sets of instruments were used: a conservative set (instruments restricted to variants with concordant relations to 4 iron status biomarkers) and a liberal set (instruments selected using variants associated with at least 1 of 4 iron status biomarkers). Associations of these genetic variants with asthma risk were estimated in data from the Trans-National Asthma Genetics Consortium (TAGC) and the GABRIEL consortium (A Multidisciplinary Study to Identify the Genetic and Environmental Causes of Asthma in the European Community). Data on the association of genetic variants with iron status and with asthma were combined to assess the influence of iron status on asthma risk. Results In the conservative approach, the MR OR of asthma was 1.00 (95% CI: 0.91, 1.10) per SD increase in iron, 0.96 (95% CI: 0.78, 1.18) in log-transformed ferritin, 0.99 (95% CI: 0.93, 1.06) in transferrin saturation, and 1.03 (95% CI: 0.93, 1.14) in transferrin in the TAGC dataset (none of the values were statistically significant). An age at onset–stratified analysis in the GABRIEL dataset suggested no effect of iron status in childhood onset, later onset, or unknown age at onset asthma. Findings from the liberal approach were similar, and the results persisted in sensitivity analyses (all P > 0.05). Conclusions This MR study does not provide evidence of an effect of iron status on asthma, suggesting that efforts to change iron concentrations will probably not result in decreased risk of asthma.

Download Full-text

Cannabis use and risk of schizophrenia: a Mendelian randomization study

10.1101/092015 ◽

2016 ◽

Cited By ~ 1

Author(s):

Julien Vaucher ◽

Brendan J. Keating ◽

Aurélie M. Lasserre ◽

Wei Gan ◽

Donald M. Lyall ◽

...

Keyword(s):

Genetic Variants ◽

Mendelian Randomization ◽

Cannabis Use ◽

Sensitivity Analyses ◽

P Value ◽

Tobacco Exposure ◽

Increased Risk ◽

Show Evidence ◽

Public Health Message ◽

Genetically Determined

ABSTRACTCannabis use is observationally associated with an increased risk of schizophrenia, however whether the relationship is causal is not known. To determine the nature of the association between cannabis use on risk of schizophrenia using Mendelian randomization (MR) analysis, we used ten genetic variants previously identified to associate with cannabis use in 32,330 individuals. Genetic variants were used in a MR analyses of the association of genetically determined cannabis on risk of schizophrenia in 34,241 cases and 45,604 controls from predominantly European descent. Estimates from MR were compared to a metaanalysis of observational studies reporting effect estimates for ever use of cannabis and risk of schizophrenia or related disorders. Genetically determined use of cannabis was associated with increased risk of schizophrenia (OR of schizophrenia for users vs. non-users of cannabis: 1.37; 95%CI, 1.09 to 1.67; P-value=0.007). The corresponding estimate from observational analysis was 1.50 (95% CI, 1.10 to 2.00; P-value for heterogeneity = 0.88). The genetic instrument did not show evidence of pleiotropy on MR-Egger (Egger test, P-value=0.292) nor on multivariable MR accounting for tobacco exposure (OR of schizophrenia for users vs. nonusers of cannabis, adjusted for ever vs. never smoker: 1.41; 95% CI, 1.09-1.83). Furthermore, the causal estimate remained robust to sensitivity analyses. These findings strongly support a causal association between genetically determined use of cannabis and risk of schizophrenia. Such robust evidence may inform public health message about the risks of cannabis use, especially regarding its potential mental health consequences.

Download Full-text

Iron Status and Risk of Stroke

Stroke ◽

10.1161/strokeaha.118.022701 ◽

2018 ◽

Vol 49 (12) ◽

pp. 2815-2821 ◽

Cited By ~ 14

Author(s):

Dipender Gill ◽

Grace Monori ◽

Ioanna Tzoulaki ◽

Abbas Dehghan

Keyword(s):

Genetic Variants ◽

Serum Iron ◽

Detrimental Effect ◽

Stroke Risk ◽

Mendelian Randomization ◽

Iron Status ◽

Transferrin Saturation ◽

Cardioembolic Stroke ◽

Sensitivity Analyses ◽

Ratio Method

Background and Purpose— Both iron deficiency and excess have been associated with stroke risk in observational studies. However, such associations may be attributable to confounding from environmental factors. This study uses the Mendelian randomization technique to overcome these limitations by investigating the association between genetic variants related to iron status and stroke risk. Methods— A study of 48 972 subjects performed by the Genetics of Iron Status consortium identified genetic variants with concordant relations to 4 biomarkers of iron status (serum iron, transferrin saturation, ferritin, and transferrin) that supported their use as instruments for overall iron status. Genetic estimates from the MEGASTROKE consortium were used to investigate the association between the same genetic variants and stroke risk. The 2-sample ratio method Mendelian randomization approach was used for the main analysis, with the MR-Egger and weighted median techniques used in sensitivity analyses. Results— The main results, reported as odds ratio (OR) of stroke per SD unit increase in genetically determined iron status biomarker, showed a detrimental effect of increased iron status on stroke risk (serum iron OR, 1.07; 95% CI, 1.01–1.14; [log-transformed] ferritin OR, 1.18; 95% CI, 1.02–1.36; and transferrin saturation OR, 1.06; 95% CI, 1.01–1.11). A higher transferrin, indicative of lower iron status, was also associated with decreased stroke risk (OR, 0.92; 95% CI, 0.86–0.99). Examining ischemic stroke subtypes, we found the detrimental effect of iron status to be driven by cardioembolic stroke. These results were supported in statistical sensitivity analyses more robust to the inclusion of pleiotropic variants. Conclusions— This study provides Mendelian randomization evidence that higher iron status is associated with increased stroke risk and, in particular, cardioembolic stroke. Further work is required to investigate the underlying mechanism and whether this can be targeted in preventative strategies.

Download Full-text

Investigating causal pathways between liability to ADHD and substance use, and liability to substance use and ADHD risk, using Mendelian randomization

10.1101/524769 ◽

2019 ◽

Cited By ~ 1

Author(s):

Jorien L Treur ◽

Ditte Demontis ◽

George Davey Smith ◽

Hannah Sallis ◽

Tom G Richardson ◽

...

Keyword(s):

Substance Use ◽

Alcohol Use ◽

Genetic Variants ◽

Mendelian Randomization ◽

Smoking Initiation ◽

Cannabis Use ◽

Causal Effects ◽

Sensitivity Analyses ◽

Genome Wide Association Studies ◽

Causal Pathways

ABSTRACTBackgroundAttention-deficit hyperactivity disorder (ADHD) has consistently been associated with substance (ab)use, but the nature of this association is not fully understood. In view of preventive efforts, a vital question is whether there are causal effects, from ADHD to substance use and/or from substance use to ADHD.MethodsWe applied bidirectional Mendelian randomization using summary-level data from the largest available genome-wide association studies (GWASs) on ADHD, smoking (initiation, cigarettes/day, cessation, and a compound measure of lifetime smoking), alcohol use (drinks/week and alcohol use disorder), cannabis use (initiation and cannabis use disorder (CUD)) and coffee consumption (cups/day). Genetic variants robustly associated with the ‘exposure’ were selected as instruments and then identified in the ‘outcome’ GWAS. Effect estimates from individual genetic variants were combined with inverse-variance weighted regression and five sensitivity analyses were applied (weighted median, weighted mode, MR-Egger, generalized summary-data-based MR, and Steiger filtering).ResultsWe found strong evidence that liability to ADHD increases likelihood of smoking initiation and also cigarettes per day among smokers, decreases likelihood of smoking cessation, and increases likelihood of cannabis initiation and CUD. In the other direction, there was evidence that liability to smoking initiation and CUD increase ADHD risk. There was no clear evidence of causal effects between liability to ADHD and alcohol or caffeine consumption.ConclusionsWe find evidence for causal effects of liability to ADHD on smoking and cannabis use, and of liability to smoking and cannabis use on ADHD risk, indicating bidirectional pathways. Further work is needed to explore causal mechanisms.

Download Full-text

The effect of angiotensin-converting enzyme levels on Covid-19 susceptibility and severity: a Mendelian randomization study

International Journal of Epidemiology ◽

10.1093/ije/dyaa229 ◽

2020 ◽

Author(s):

Guillaume Butler-Laporte ◽

Tomoko Nakanishi ◽

Vincent Mooser ◽

Alessandra Renieri ◽

Sara Amitrano ◽

...

Keyword(s):

Angiotensin Converting Enzyme ◽

Ace Inhibitors ◽

Genetic Variants ◽

Genome Wide Association Study ◽

Mendelian Randomization ◽

Severe Disease ◽

Sensitivity Analyses ◽

European Ancestry ◽

Converting Enzyme ◽

Discontinue Therapy

Abstract Background There has been uncertainty about the safety or benefit of angiotensin-converting enzyme (ACE) inhibitors during the Covid-19 pandemic. We used Mendelian randomization using genetic determinants of serum-ACE levels to test whether decreased ACE levels increase susceptibility to SARS-CoV-2 infection or Covid-19 severity, while reducing potential bias from confounding and reverse causation in observational studies. Methods Genetic variants strongly associated with ACE levels, which were nearby the ACE gene, were identified from the ORIGIN trial and a separate genome-wide association study (GWAS) of ACE levels from the AGES cohort. The ORIGIN trial included 4147 individuals of European and Latino ancestries. Sensitivity analyses were performed using a study of 3200 Icelanders. Cohorts from the COVID-19 Host Genetics Initiative GWAS of up to 960 186 individuals of European ancestry were used for Covid-19 susceptibility, hospitalization and severe-disease outcome. Results Genetic variants were identified that explain between 18% and 37% of variance in ACE levels. Using genetic variants from the ORIGIN trial, a standard-deviation decrease in ACE levels was not associated with an increase in Covid-19 susceptibility [odds ratio (OR): 1.02, 95% confidence interval (CI): 0.90, 1.15], hospitalization (OR: 0.86, 95% CI: 0.68, 1.08) or severe disease (OR: 0.74, 95% CI: 0.51, 1.06). Using genetic variants from the AGES cohort, the result was similar for susceptibility (OR: 0.98, 95% CI: 0.89, 1.09), hospitalization (OR: 0.86, 95% CI: 0.66, 1.11) and severity (OR: 0.75, 95% CI: 0.50, 1.14). Multiple-sensitivity analyses led to similar results. Conclusion Genetically decreased serum ACE levels were not associated with susceptibility to, or severity of, Covid-19 disease. These data suggest that individuals taking ACE inhibitors should not discontinue therapy during the Covid-19 pandemic.

Download Full-text

Guidelines for performing Mendelian randomization investigations

Wellcome Open Research ◽

10.12688/wellcomeopenres.15555.1 ◽

2019 ◽

Vol 4 ◽

pp. 186 ◽

Cited By ~ 26

Author(s):

Stephen Burgess ◽

George Davey Smith ◽

Neil M. Davies ◽

Frank Dudbridge ◽

Dipender Gill ◽

...

Keyword(s):

Genetic Variants ◽

Mendelian Randomization ◽

Data Sources ◽

Sensitivity Analyses ◽

Robust Methods ◽

Primary Analysis ◽

Data Presentation ◽

Journal Editors

This paper provides guidelines for performing Mendelian randomization investigations. It is aimed at practitioners seeking to undertake analyses and write up their findings, and at journal editors and reviewers seeking to assess Mendelian randomization manuscripts. The guidelines are divided into nine sections: motivation and scope, data sources, choice of genetic variants, variant harmonization, primary analysis, supplementary and sensitivity analyses (one section on robust methods and one on other approaches), data presentation, and interpretation. These guidelines will be updated based on feedback from the community and advances in the field. Updates will be made periodically as needed, and at least every 18 months.

Download Full-text

Using Mendelian randomization to determine causal effects of maternal pregnancy (intrauterine) exposures on offspring outcomes: Sources of bias and methods for assessing them

Wellcome Open Research ◽

10.12688/wellcomeopenres.10567.1 ◽

2017 ◽

Vol 2 ◽

pp. 11 ◽

Cited By ~ 50

Author(s):

Deborah A. Lawlor ◽

Rebecca Richmond ◽

Nicole Warrington ◽

George McMahon ◽

George Davey Smith ◽

...

Keyword(s):

Genetic Variants ◽

Mendelian Randomization ◽

Causal Effect ◽

Strong Association ◽

Simulated Data ◽

Real Data ◽

Causal Effects ◽

Sensitivity Analyses ◽

Specific Situation ◽

Exclusion Restriction

Mendelian randomization (MR), the use of genetic variants as instrumental variables (IVs) to test causal effects, is increasingly used in aetiological epidemiology. Few of the methodological developments in MR have considered the specific situation of using genetic IVs to test the causal effect of exposures in pregnant women on postnatal offspring outcomes. In this paper, we describe specific ways in which the IV assumptions might be violated when MR is used to test such intrauterine effects. We highlight the importance of considering the extent to which there is overlap between genetic variants in offspring that influence their outcome with genetic variants used as IVs in their mothers. Where there is overlap, and particularly if it generates a strong association of maternal genetic IVs with offspring outcome via the offspring genotype, the exclusion restriction assumption of IV analyses will be violated. We recommend a set of analyses that ought to be considered when MR is used to address research questions concerned with intrauterine effects on post-natal offspring outcomes, and provide details of how these can be undertaken and interpreted. These additional analyses include the use of genetic data from offspring and fathers, examining associations using maternal non-transmitted alleles, and using simulated data in sensitivity analyses (for which we provide code). We explore the extent to which new methods that have been developed for exploring violation of the exclusion restriction assumption in the two-sample setting (MR-Egger and median based methods) might be used when exploring intrauterine effects in one-sample MR. We provide a list of recommendations that researchers should use when applying MR to test the effects of intrauterine exposures on postnatal offspring outcomes and use an illustrative example with real data to demonstrate how our recommendations can be applied and subsequent results appropriately interpreted.

Download Full-text

Genetic predictors of participation in optional components of UK Biobank

Nature Communications ◽

10.1038/s41467-021-21073-y ◽

2021 ◽

Vol 12 (1) ◽

Author(s):

Jessica Tyrrell ◽

Jie Zheng ◽

Robin Beaumont ◽

Kathryn Hinton ◽

Tom G. Richardson ◽

...

Keyword(s):

Sensitivity Analysis ◽

Genetic Variants ◽

Genetic Factors ◽

Mendelian Randomization ◽

Genetic Correlations ◽

Mr Studies ◽

Uk Biobank ◽

Factors Affecting ◽

Participation Bias ◽

Genetic Predictors

AbstractLarge studies such as UK Biobank are increasingly used for GWAS and Mendelian randomization (MR) studies. However, selection into and dropout from studies may bias genetic and phenotypic associations. We examine genetic factors affecting participation in four optional components in up to 451,306 UK Biobank participants. We used GWAS to identify genetic variants associated with participation, MR to estimate effects of phenotypes on participation, and genetic correlations to compare participation bias across different studies. 32 variants were associated with participation in one of the optional components (P < 6 × 10−9), including loci with links to intelligence and Alzheimer’s disease. Genetic correlations demonstrated that participation bias was common across studies. MR showed that longer educational duration, older menarche and taller stature increased participation, whilst higher levels of adiposity, dyslipidaemia, neuroticism, Alzheimer’s and schizophrenia reduced participation. Our effect estimates can be used for sensitivity analysis to account for selective participation biases in genetic or non-genetic analyses.

Download Full-text