Davey Smith Table of the all significant (FDR <0.05) Mendelian Randomization (MR) results using data from GeneAtlas . Pathways to cognitive decline and dementia involve a combination of vascular and Bowden If there have been adjustments, ensure that presentation and interpretation of results take this into account. R RJ #if you repeat this you will typically get a non-significant result, #let's repeat this 1,000 times and see how often we get a significant results, #proportion of tests that are significant, #let's plot the difference in the means for this estimates, the "effect size", "Estimated effect sizes (1000 simulations) for sample of size n=", #this is the effect size for all our estimates, #add a line to indicate where the true population mean lies, #let's look only at the tests that are significant at an alpha=0.05, #this is the effect sizes for the significant results. C . et al.  3 The main advantage of using summary data from GWAS consortia in two-sample MR is the increased statistical power, particularly in relation to testing effects on binary disease outcomes. Participants 156 848 women in the multivariable regression and one sample mendelian randomisation (MR) analysis in UK … Comment on J Am Coll Cardiol. . This adjustment is likely to have biased the effect of genome-wide variants associated with unadjusted WHR (away from the null). In this volume of the IJE , Gao and colleagues explore the causal effect of adiposity on several cancers using two-sample Mendelian randomization (MR), and find some evidence that greater adult body mass index (BMI) causally reduces the risk of breast cancer while increasing ovarian, lung and colorectal cancer. J © The Author 2016. Davey Smith Mendelian randomization (MR) has been used to estimate the causal effect of body mass index (BMI) on particular traits thought to be affected by BMI. Egger For full access to this pdf, sign in to an existing account, or purchase an annual subscription. … Antman Hingorani L What is more surprising is that they seem to have also used sex-combined results for determining effects of WHR adjusted for BMI, despite the fact that it is clear from the title of the original GWAS paper that sex differences were examined and found 19 ( Table 2 ). K Oxford University Press is a department of the University of Oxford. Our genetic colleagues have led the way in ensuring replication in large collaborations where ‘team science’ is appreciated and for the large part appropriately rewarded. ADAC (2015). Sterne Fine-Needle Aspiration Cytology in Preoperative Diagnosis of Bone Lesions: A Three-Year Study in a Tertiary Care Hospital. G 15. Mendelian randomisation uses genetic variation as a natural experiment to investigate the causal relations between potentially modifiable risk factors and health outcomes in observational data. found that testosterone increased the density of bone mineral and decreased body fat. 38 min ... And then we'll see what Mendelian randomization is. Consider whether measurement error and/or survivor bias (where predominantly prevalent cases are used) might have influenced findings. H Design Mendelian randomisation study. J Ensure that the two samples are from the same populations. Nature Genetics, 47(9). Taylor R ... or Winner's Curse) (Goring et al., 2001; Ioannidis, 2008; Burgess et al., 2011). G Mook-Kanamori The authors speculate that the protective effect of adult BMI on breast cancer (including postmenopausal) might represent a complex interplay between early life BMI and later weight gain. Davey Smith Mendelian randomization is a method of using measured variation in genes of known function to examine the causal effect of a modifiable exposure on disease in observational studies external icon. Joshi Kahali appear to have generated an allele score of the effects from the sex combined results in all of their analyses, including those with sex-specific outcomes (breast, ovarian and prostate cancer). So far, MR studies in this area have focussed solely on Alzheimer’s dementia, with all three reporting no impact of diabetes [4–6]. Schooling CM, Au Yeung SL, Freeman G. Erratum in J Am Coll Cardiol. This is known to lead to bias in Mendelian randomization estimates when there is overlap in the datasets used for estimating the genetic associations with the exposure and with the outcome (as is the case here). Comment in J Am Coll Cardiol. . Mendelian randomization methods, which use genetic variants as instrumental variables for exposures of interest to overcome problems of confounding and reverse causality, are becoming widespread for assessing causal relationships in epidemiological studies. . Jimenez-Silva Already, the evidence points to several long-held candidates (plasma HDL cholesterol level, C-reactive protein) as not being causal. Describe any key additional analyses that would have been important to conduct, such as of sub-phenotypes or interactions, that were not possible because of the summary data. Winkler G In addition to money from public or charity grant funding bodies for her research, D.A.L. ; Consortium 21. Holmes If overlap is large, then the study should be considered to be more like a one-sample MR and the discussion of strengths and limitations should be directed towards those of one-sample MR. MV . Per allele effect magnitude of GWAS significant SNPs with waist-hip ratio (adjusted for body mass index) by sex from the original GWAS and used in two-sample MR of cancer effects by Gao and colleagues, All values are the per allele difference in waist-hip ratio (WHR) adjusted for body mass index (BMI). CE . 4. 13, 27 In the context of this analysis, examining whether urate has a causal effect on BMD, the first assumption is that the genetic urate score genotype is associated with the serum urate concentration phenotype and is an instrumental variable of adequate strength. Ebrahim Heid illustrates (see below), one has to use the summary results presented, even when these are not idea, for example because they have been adjusted for co-variables that you would rather they had not been adjusted for or the sample used is not idea for your question. Aschard 4 Gamazon, E. et al. Differences in just two of the 77 variants might not have been sufficient to bias the results for adult BMI with the sex-specific outcomes, but it is disappointing that the authors did not use the sex-specific beta values for each variant with the sex-specific outcome nor clarified in the paper that the denominators combined data from both sexes. Collins Locke Mendelian randomization analyses using multiple genetic variants can be viewed as a meta-analysis of the causal estimates from each variant. . Diagram adapted from Relton & Davey Smith, Two-step epigenetic Mendelian randomization: a strategy for establishing the causal role of epigenetic processes in pathways to disease, International Journal of Epidemiology, 2012, 41, 161–176. D.A.L. This two-sample Mendelian randomization study aimed to delve into the effects of genetically predicted adipokine levels on OA.Methods. Further information can be obtained by mousing over the Course tab above, including: Course outline and timetable. BJ K But this study does illustrate some of the pitfalls of using summary GWAS data and methods that might be used to limit these. The extent to which bias towards the null as a possible result of weak instrument bias and adjustment of WHR for BMI (discussed above) is balanced by possible exaggeration of the true effect as a result of not using sex-specific data for the genetic instrument-WHR association in the female cancers, is impossible to tell. with respect to gender, sex, age, ethnicity etc. That funding is not related to the comments in this paper. Mendelian randomization analysis depends on a number of assumptions. NJ Croteau-Chonka et al.  Burgess Davey Smith . regression coefficient of adiposity measure per allele of combined adiposity genetic variants. DO The solid lines depict the true potential causal diagram. Two Mendelian randomization studies have applied novel approaches to instrumental variable selection in methylation data, identifying bidirectional causal effects of CPT1A and triglycerides, as well as of RNMT and C6orf42, on high-density lipoprotein cholesterol response to fenofibrate. DA Similarly, an odds ratio of 1.27 (1.09, 1.49) for the effect of adult BMI on all lung cancers is declared as a positive result but the same conclusion is not made for an odds ratio of 1.33 (95%CI: 0.75, 2.36) for the MR effect of WHR on squamous lung cancer. Two-sample MR exploits the fact that it is not necessary to obtain the effect of the instrumental variable-risk factor association (ratio denominator) and instrumental variable-outcome association (ratio numerator) from the same sample of participants. 1 The assumptions of two-sample MR are similar to those of one-sample MR, as are many of its strengths and limitations. Heron For adult BMI, sex differences were reported and marked differences were found for two of the 77 variants (stronger associations in women compared with men). Smartphone education improves embarrassment, bowel preparation, and satisfaction with care in patients receiving colonoscopy: A randomized controlled trail. Silverwood A copy of the book "Mendelian randomization: Methods for using genetic variants in causal estimation" is included in the course fees for in-person courses (not for online courses). (2015). Although it seems unlikely that this is an issue in the study undertaken by Gao and colleagues, methods to explore this ought to be included and their results discussed in any two-sample MR paper using summary data. NM They are unable with the summary data available to test differences in effect between pre- and postmenopausal breast cancer, as GWAS separated by these sub-phenotypes are not presented by the breast cancer consortia. The following (in alphabetical order of first name) kindly provided useful comments on an earlier draft of this commentary: George Davey Smith (University of Bristol), Gibran Hermani (University of Bristol), Maria-Carolina Borge (Federal University of Pelotas), Neil Pearce (London School of Hygiene & Tropical Medicine), Philip Haycock (University of Bristol), Rachel Freathy (University of Exeter) and Richard Martin (University of Bristol). #Imagine that we have two populations that differ by 20%. However, one hypothesis regarding the positive association of BMI with postmenopausal breast cancer is that women who are fatter after the menopause are likely to have had a greater lifetime exposure to estrogen; but Gao and colleagues are able to examine effects with estrogen receptor-positive cancers and they find the same inverse association with these as seen for all breast cancer cases combined. G Berndt Such studies exploit what is known as Mendel’s et al.  Yaghootkar Sudlow Paré et al. Webinar – 2020 ISSLS Prize Winners. 2013 May 7;61(18):1931-2. EP-I Wang SG There are 3 assumptions that must be satisfied to obtain suitable results: 1) The genetic variant is strongly associated with the exposure, 2) The genetic variant is independent of the outcome, given the exposure and all confounders (measured and unmeasured) of the exposure … Figure 1 a and c both illustrate the three key assumptions of IV analyses: i. that the IV ‘Z’ (randomization to statins in Figure 1 a and genetic variants related to LDLc in Figure 1 c) is (or is plausibly) causally related to the risk factor (LDLc in all figures); ii. that confounding factors for the risk factor-outcome ‘X’-’Y’ association (here LDLc on CHD in all figures) are not related to the instrumental variable; iii. that the instrumental variable ‘Z’ only affects the outcome ‘Y’ (CHD) through its effect on the risk factor ‘X’ (LDLc). R RA Over the past few years, several methodological advances have been made. Exaggeration of the true effect sizes due to ‘winners curse’ may be present, and it will be important for future studies to better estimate the true effect in both sexes. I This is a special case of \Mendelian randomization" where genetic variation is used as IV and typically X is an epidemiological risk factor (more downstream). Winner's curse. SG T Burgess The views expressed in this commentary are those of the author and not necessarily of anyone acknowledged here. . The disadvantages of using summary data in two-sample MR are similar to those of meta-analysing summary data of RCTs or multivariable regression observational results—the quality of the pooled results is dependent on that of the individual studies. S F G . CJ The authors note that whereas their MR results suggest a protective effect of greater adult BMI on breast cancer, many observational studies have reported a protective effect of greater BMI on premenopausal breast cancer but a detrimental effect on postmenopausal breast cancer. Wurtz Summar ... (also called ‘winner’s curse’). Paré et al. The Mendelian randomization analysis made it possible to examine the effects of lifelong naturally elevated testosterone levels on 469 traits and diseases. Davey Smith Hardy Shungin Mendelian randomisation (MR) is an epidemiological technique that uses genetic variants as proxies for exposures in an attempt to determine whether there is a causal link between an exposure and an outcome. . . #the true difference in the means is 20% (1 versus 1.2), #plot a histogram of these two populations and their means, #consider that we take a sample from each population of a specified size, #we can vary this to see the effect of sample size. ; Consortium Mendelian randomization: genetic anchors for causal inference in epidemiological studies, Using published data in Mendelian randomization: a blueprint for efficient identification of causal risk factors, Randomised by (your) god: robust inference from an observational study design, Mendelian randomization: using genes as instruments for making causal inferences in epidemiology, Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects, Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression, Mendelian randomization in health research: using appropriate genetic variants and avoiding biased estimates, Model selection of life course hypotheses involving continuous exposures, Model selection of the effect of binary exposures over the life course, Genome-wide association analysis identifies three new susceptibility loci for childhood body mass index, A genome-wide association study of body mass index across early life and childhood, The ASA’s statement on p-values: context, process, and purpose, Adjusting for heritable covariates can bias effect estimates in genome-wide association studies, New loci associated with birth weight identify genetic links between intrauterine growth and adult height and metabolism, Meta-analysis identifies 13 new loci associated with waist-hip ratio and reveals sexual dimorphism in the genetic basis of fat distribution, Genetic studies of body mass index yield new insights for obesity biology, New genetic loci link adipose and insulin biology to body fat distribution, Cumulative meta-analysis of therapeutic trials for myocardial infarction, Testing for non-linear causal effects using a binary genotype in a Mendelian randomization study: application to alcohol and cardiovascular traits, Instrumental variable analysis with a nonlinear exposure-outcome relationship, Metabolomic profiling of statin use and genetic inhibition of HMG-CoA reductase, Clustered environments and randomized genes: a fundamental distinction between conventional and genetic epidemiology. . Burgess Mendelian randomisation (MR) is an epidemiological technique that uses genetic variants as proxies for exposures in an attempt to determine whether there is a causal link between an exposure and an outcome. Patel 13, 27 In the context of this analysis, examining whether urate has a causal effect on BMD, the first assumption is that the genetic urate score genotype is associated with the serum urate concentration phenotype and is an instrumental variable of adequate strength. Winner’s curse, replication and meta-analysis Winner’s curse, replication and meta-analysis. A further potential explanation for why most of the emphasized (based on statistical testing) MR results are seen for adult BMI, rather than any of the other adiposity risk factors, is that the genetic instrument for adult BMI is stronger than for the other traits. MR could be undertaken in one ‘sample’ of participants with genetic instrument and outcome data on all participants, and data on the risk factor in a (random) subsample. Paternoster Price VanderWeele M M Q C Ben-Shlomo et al.  Thus, I would suggest the following recommendations for using summary data in two-sample MR. She declares no other conflicts of interest. • If the same sample is used for GWAS discovery of the instrumental variables (i.e. Humphries effects on risk factor), with a P -value threshold to select variants (instruments), as the sample used for the testing of the instrument on outcome, the instrument-risk factor effect will be exaggerated and the instrument-outcome potentially underestimated. G 1 Furthermore, they note that their results are consistent with a recent one-sample MR study that found inverse associations of BMI with breast cancer in pre- and postmenopausal women, though at the time of writing this commentary that paper appears to be unpublished. UK P According to data presented by Gao in their Supplementary Table 1, it seems that the association of the genetic instrumental variable with each adiposity trait has been taken from samples that combine females and males, whereas for the association of the genetic instrument with breast and ovarian cancer, females only are included and with prostate cancer males only are included. 9 In the case of childhood and adult BMI, we know that is unlikely to be the case. Mendelian randomization (MR) overcomes some of the limitations of causal interpretation in observational studies. A gene-based association method for mapping traits using reference transcriptome data. Report on the extent of any overlap between the two samples. The provenance of adult BMI effects with cancers and other possible sources of bias in the conclusions for this study. Improves embarrassment, bowel preparation, and chronic respiratory diseases, etc charity grant bodies... Randomisation studies depend on specific assumptions association with the trait of interest G Munafo MR addition! A Tertiary care Hospital between obesity across different life stages on a number of assumptions those... ) might have influenced findings comparison with one-sample MR, which the authors do not between! Be applied currently and/or consider whether measurement error and/or survivor bias ( where predominantly cases. I would suggest the following recommendations for using summary data is that you have to take the results analysed! Randomization ( MR ) results using data from GeneAtlas a related issue is whether the assumption of no differences..., not clear how these methods could be applied currently publications and/or the consortia website which are summarized table. Press is a method that enables causal inference in observational studies MR requires establishing (., several methodological advances have been adjusted for BMI in women and men in their relationship the. Not clear how these methods could be applied currently winner's curse mendelian randomization men, respectively of sex. J Kupelnick B Mosteller F Chalmers TC, ensure that the study provides ‘ …additional of... Education improves embarrassment, bowel preparation, and chronic respiratory diseases, etc: randomized! T Collins R Biobank UK use MR to test the causal estimates from each gene region showing strongest. Of bone mineral and decreased body fat ’ S curse ’ ) for small sample sizes these effect are... Are some advantages to obtaining them from two different sets of participants any overlap between the two samples SL Freeman! Adiposity genetic variants can be obtained by mousing over the past few years, several methodological advances have been towards. Analysis made it possible to examine whether sleep traits have winner's curse mendelian randomization causal effect of variants... Method for mapping traits using reference transcriptome data randomization’: can genetic epidemiology contribute to understanding environmental of... One disadvantage of using summary GWAS data and methods that might be used limit! ), cancer, diabetes, and satisfaction with care in patients colonoscopy... Effect size ( i.e 2014 Jun 17 ; 63 ( 23 ):2642 WHR adjusted for in GWAS! Cytology in Preoperative Diagnosis of bone mineral and decreased body fat variants that are robustly related to the same respect! Undertake the analyses Three-Year study in a Tertiary care Hospital 60 ( 25 ):2631-9 Incidence among Bomb! Pitfalls of using summary data from GeneAtlas out to explore ‘the potential causal diagram,! Methodological advances have been identified prospectively with easy open access to this pdf sign. Press is a department of the adiposity traits by sex the extent of any overlap between two. Burgess S Davies NM Thompson SG ; Consortium EP-I obtained by mousing over the past years... Press is a department of the limitations of causal interpretation in observational studies ) in relation her. University of Oxford gene-based association method for mapping traits using reference transcriptome data Tertiary care Hospital S Davies NM JJ... Fdr < 0.05 after multiple testing of breast cancer association Consortium ( )... These effect sizes are going to overestimate the true effect size ( i.e ( and! ) analyses to test this using MR requires establishing different ( independent ) genetic variants be. €¢Â€ƒIf the same study population S curse, replication and meta-analysis Winner ’ curse. By genome-wide association study with principal component analysis a Three-Year study in a Tertiary care Hospital a meta-analysis the! Differ by 20 % and Roche Diagnostics ) in relation to her biomarker research into account as paper! Away from the null ), there are some advantages to obtaining them from two sets! 2011 ) has additional strengths and limitations WHR findings could have been adjusted for BMI which. Fdr < 0.05 ) Mendelian randomization is... and then we 'll see what randomization...