winner's curse mendelian randomization

. Yeung, Shiu Lun [corrected to Au Yeung, Shiu Lun]. In addition to money from public or charity grant funding bodies for her research, D.A.L. 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. Design Mendelian randomisation study. CE #If we did a t-test from a random sample drawn from population 1 and population 2 using a sample size of 3 we are underpowered to detect a difference and so won't usually get a p value < 0.05. Mosteller Egger Mendelian randomization has emerged as a valuable approach in investigating whether an association of a biomarker with CAD is casual or not. P (2015). Setting UK Biobank prospective cohort study and Breast Cancer Association Consortium (BCAC) case-control genome-wide association study. Comparison of one-sample and two-sample MR, • Can check this within the population with exposure and outcome (as both in same population), • Use F-statistic and R 2 of genetic instrument-risk factor association as measure of strength, • Weak instrument biases towards the confounded regression analysis result, • Can check this within the population with exposure but need to be careful that the population used for testing genetic instrument-outcome association is the same as that testing instrument-risk factor (e.g. NE S J . As this is the denominator of the MR ratio estimate, it means that the estimated effect of WHR adjusted for BMI for female cancers (breast and ovarian) may be exaggerated and those for prostate cancers underestimated. JJ 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. NJ G Board Certified or Board Eligible AP/CP Full-Time or Part-Time Pathologist, Chief of ID, VA Ann Arbor Healthcare System, Instrumental variable is related to risk factorÂ, Confounders of the risk factor-outcome association are not related to the genetic instrumentÂ, Genetic instrument only related to the outcome through its effect on the risk factorÂ, Copyright © 2020 International Epidemiological Association. 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. JP Timpson 1 Concepts of MR and Instrumental variable (IV) methods motivation, assumptions, inference goals, merits and limitations two-stage least squares (2SLS) method from econometrics literature 2,3,9 It is notable, for example, that Gao et al. In particular ‘winners’ curse’, which can underestimate true causal effects in one-sample MR, 10 is unlikely to happen in two-sample MR, and unlike the impact of weak instrument bias in one-sample MR (which biases effects towards the confounded multivariable regression result), in two-sample MR weak instrument bias is towards the null. Dale Humphries Nature Genetics, 47(9). As with all epidemiological approaches, findings from Mendelian randomisation studies depend on specific assumptions. . (2015). 21 In both GWAS, the results of the per allele effect of the genetic instrument on WHR adjusted for BMI is notably stronger in females than males. Scott • With summary data from large GWAS consortia, not clear how these methods could be applied currently. 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). Winner's Curse. ; Consortium Pathways to cognitive decline and dementia involve a combination of vascular and Mendelian Randomization results from GeneAtlas. Holmes ADAC IM R Biobank 1The authors conclude that the study provides ‘…additional understanding of the complex relationship between adiposity and cancer risks’. . The comments made in this paper are those of the author and not necessarily of the MRC or NIHR. Consider whether measurement error and/or survivor bias (where predominantly prevalent cases are used) might have influenced findings. Collins 4 The paper by Gao and colleagues illustrates this with the large numbers of cancer cases that they analyse: 15 748, 5100, 12 160, 4369 and 14 160, respectively, for breast, colorectal, lung, ovarian and prostate cancer. Davey Smith 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. G 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. Davies Harbord Wang Did you miss the ISSLS Prize webinar or wish to view it again? P 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. The dashed lines represent the parameters that need to be estimated, which are equal to the multiplication of the respective effects represented by the solid lines. 2013 May 7;61(18):1931-2. Mendelian randomization phenome-wide scan of AMH-associated variants. regression coefficient of adiposity measure per allele of combined adiposity genetic variants. . Taylor 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. Those developing MR as a method have from the start been very open about its limitations and have worked at developing methods to test and limit sources of bias. DO From the original papers it can be seen that the instrument for adult BMI is stronger than for the other traits (the respective R 2 for adult BMI, birthweight, child BMI and WHR, are: 0.027, 0.008, 0.020 and 0.013; the R 2 for child BMI is only for the three novel SNPs but, as the authors of the original paper point out, it was calculated on a relatively small sample and needs to be treated with some caution). S Lau Report on the extent of any overlap between the two samples. N For two-sample MR to be valid, the two samples have to be from the same underlying population, but for the sex-specific cancers in the paper by Gao et al. It also has additional strengths and limitations in comparison with one-sample MR, which are summarized in Table 1 . 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. . CJ 36 However, genetic associations with SBP and DBP from the replication analyses only were not reported in the original study, limiting the possibility to distinguish whether the asymmetry in … . Identification of genetic factors controlling phosphorus utilization efficiency in wheat by genome-wide association study with principal component analysis. . Now you have the chance to listen to the presentations and you can also contact the presenters via e-mail if you wish to ask questions or discuss the papers. . 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. . VanderWeele The winner's curse is the phenomenon that the association estimate of the variant with the strongest association from a GWA study tends to be overestimated [Göring et al., 2001]. Figure shows DAGs of instrumental variable (IV) analyses to test the causal effect of low-density lipoprotein cholesterol (LDLc) on CHD. Kahali Greater adult BMI, but not waist-hip ratio (WHR), is concluded to decrease breast cancer and increase ovarian, lung and colorectal cancer risk. Hemani et al.  the outcome) can be done on separate sets of studies; so called ‘two-sample Mendelian randomization’ • Know the basics of how to undertake a two-sample Mendelian randomization study Future availability A Mendelian randomization short course will run as a … Oxford University Press is a department of the University of Oxford. Mendelian randomization (MR) Use inherited genetic variants to infer causal relationship of an exposure and a disease outcome. 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. 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 … Table of the all significant (FDR <0.05) Mendelian Randomization (MR) results using data from GeneAtlas . Heron J G Nature Genetics, 47(9). Mendelian randomization analyses using multiple genetic variants can be viewed as a meta-analysis of the causal estimates from each variant. NM Mendelian randomization (MR) is one of such approaches whose reliability has been established in epidemiology and is gaining popularity among health economists in testing causal research statements and obtaining consistent evidence in a cost effective manner. 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. Search for other works by this author on: Mendelian randomization study of adiposity-related traits and risk of breast, ovarian, prostate, lung and colorectal cancer. 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). Aschard Kraft Patel 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. Already, the evidence points to several long-held candidates (plasma HDL cholesterol level, C-reactive protein) as not being causal. works in a unit that receives funding from the University of Bristol and UK Medical Research Council (grant ref: MC_UU_12013/5) and she is a National Institute of Health Research Senior Investigator (NF-SI-0611‐10196). Smartphone education improves embarrassment, bowel preparation, and satisfaction with care in patients receiving colonoscopy: A randomized controlled trail. NA 21. Determine whether any covariables have been adjusted for in the original GWAS and report on this. Gao Davey Smith has received support from industry (Medtronic and Roche Diagnostics) in relation to her biomarker research. with summary data from large GWAS consortia, likely to have more power for these analyses which tend to be statistically inefficient, • Possible if large sample sizes and data on the relevant stratifying risk factors (and genetic instruments for these) available, • Possible if individual participant data on the two samples and large sample sizes and data on the relevant stratifying risk factors (and genetic instruments for these) available, • In general, with summary data from large GWAS consortia, it is unlikely to be able to test these. Indeed, there are some advantages to obtaining them from two different sets of participants. et al.  Now genetic epidemiologists have shown us how to provide complete open-access summary data, and it is likely that over the coming decade important and impactful use will be made of these data. However, this is a one-sample MR, as the subgroup ‘belong’ to the same study population. Price Winner’s curse, replication and meta-analysis Winner’s curse, replication and meta-analysis. For example, UK Biobank will soon release GWAS data on all 500 000 participants and has already amassed large numbers of incident cases of cardiovascular disease and common cancers such as breast cancer. Over the past few years, several methodological advances have been made. P Thompson I Evidence from a two-sample Mendelian randomization analysis, High-throughput multivariable Mendelian randomization analysis prioritizes apolipoprotein B as key lipid risk factor for coronary artery disease. 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). C G Davey Smith EM J We attempted to minimize the risk of bias due to population stratification by restricting to individuals of European ethnicity and adjusting for genetic principal components. Risk of Prostate Cancer Incidence among Atomic Bomb Survivors: 1958-2009. Gao et al. ; Consortium Heid 4. 11 Collection of unique imaging data on a subsample of 100 000 of those participants has begun, and thus MR to determine the causal effect of novel imaging biomarkers on common chronic disease outcomes, in which the genetic instrument-disease outcome association in 500 000 participants is divided by the genetic instrument-imaging biomarker association in the 100 000 subgroup, will soon be possible. this is a Mendelian randomization study). Shungin Sterne Schooling CM, Au Yeung SL, Freeman G. Erratum in J Am Coll Cardiol. If there have been adjustments, ensure that presentation and interpretation of results take this into account. JC . 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. 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. with respect to gender, sex, age, ethnicity etc. So far, MR studies in this area have focussed solely on Alzheimer’s dementia, with all three reporting no impact of diabetes [4–6]. On the other hand, the known “winner’s curse” in GWAS studies might also lead to an underestimation of causal effect estimates using Mendelian randomization . SI RM Harbord The Mendelian randomization analysis made it possible to examine the effects of lifelong naturally elevated testosterone levels on 469 traits and diseases. Gao and colleagues set out to explore ‘the potential causal relationship between obesity across different life stages and risk of multiple cancers’. I looked at the four original GWAS papers to explore whether there was any difference in the GWAS of the adiposity traits by sex. As a result the one-sample MR effect estimate will be an underestimate of the true causal effect 10, • Using two non-overlapping samples avoids this. 14,18–20 Interestingly, although Gao et al. Lawlor TC A related issue is whether the WHR findings could have been biased towards the null more than BMI findings. All variants combined explained 1.34% and 0.46% of the variation in WHR adjusted for BMI in women and men, respectively. found that testosterone increased the density of bone mineral and decreased body fat. AE M One disadvantage of using summary data is that you have to take the results as analysed in the original study. A decade before the first paper proposing the use of MR, 2 Lau and colleagues demonstrated that, had a cumulative meta-analysis been regularly updated, the beneficial effect of streptokinase in patients who had experienced an acute myocardial infarction would have been established by 1973, with the further randomization of over 35 000 patients after that date simply confirming the original results and delaying widespread implementation of a life-saving treatment. Lazar Yaghootkar … Timpson BP indicates blood pressure. 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. et al.  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. Davey Smith Wurtz 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. 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. This is because MR-Egger is only valid if the effect of the genetic instrument on the risk factor of interest is independent of its effect on any other phenotypes that might violate that assumption. Heron Chalmers Methods and findings. 22 In 1997, Egger and Davey Smith showed the same with respect to beta-blockers and mortality after acute myocardial infarction. 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). Participants 156 848 women in the multivariable regression and one sample mendelian randomisation (MR) analysis in UK … Beyond Mendelian randomization—what can we learn from genetic epidemiology? Furthermore the MR-Egger test, 9 which the authors used to test violation of the exclusion restriction criteria, cannot be used to differentiate effects of adult from child BMI, as Gao and colleagues acknowledge. Warrington 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). 2012 Dec 25;60(25):2631-9. 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). Mendelian randomization analysis depends on a number of assumptions. 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). Horikoshi Summar Peakman S et al.  United Nations' Sustainable Development Goals (SDG, 2015) has specified NCDs as one of their important health related targets (Target-3.4) for improving overall wellbeing of human populations (2). G 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. 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. AU Kupelnick Paré et al. 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. ... or Winner's Curse) (Goring et al., 2001; Ioannidis, 2008; Burgess et al., 2011). Thus, it is concluded BMI reduces breast cancer risk {odds ratio[OR] 0.66 [95% confidence interval (CI): 0.57, 0.77)]}, but the same is not concluded for WHR [0.73 (0.53, 1.00)]. • Not relevant as can decide within the one sample with genetic instrument, risk factor and outcome, what to adjust for. Thus, as the paper by Gao et al. F K Design Mendelian randomisation study. Smith . 3 Gao and colleagues do not provide any information on the strength of the different instrumental variables, such as the F-statistic or R 2 for the genetic instrument-adiposity trait associations. Report how risk factors and outcomes were assessed, including whether disease cases were prevalent, incident or a mixture. Randall For full access to this pdf, sign in to an existing account, or purchase an annual subscription. But this study does illustrate some of the pitfalls of using summary GWAS data and methods that might be used to limit these. This issue is not discussed by Gao et al. http://creativecommons.org/licenses/by/4.0/, Receive exclusive offers and updates from Oxford Academic, Using Mendelian randomization to investigate a possible causal relationship between adiposity and increased bone mineral density at different skeletal sites in children, The effects of eight serum lipid biomarkers on age-related macular degeneration risk: a Mendelian randomization study, Is disrupted sleep a risk factor for Alzheimer’s disease? The basic assumption—that genetic variants which can proxy for a potentially modifiable exposure are essentially unrelated to confounding factors—has been demonstrated to have widespread plausibility.25 The connection between the standard Mendelian randomization experiment and the theory of instrumental variables has been elaborated upon.26,27 Extensions to use multiple genetic variants for increasing power and investigatin… Objective To examine whether sleep traits have a causal effect on risk of breast cancer. Felix Davey Smith Davey Smith Thompson 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. 9 In the case of childhood and adult BMI, we know that is unlikely to be the case. DA L 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. . Locke The provenance of adult BMI effects with cancers and other possible sources of bias in the conclusions for this study. If this is not the case (as in this paper for the sex-specific cancers), check the original paper publications and/or contact the original authors to see if it is possible to obtain results from the same population (here sex-specific results). MRC Integrative Epidemiology Unit at the University of Bristol and School of Social and Community Medicine, University of Bristol, Oakfield House, Oakfield Grove, Bristol BS8 2BN, UK. Antman Thompson SG Setting UK Biobank prospective cohort study and Breast Cancer Association Consortium (BCAC) case-control genome-wide association study. Ebrahim . UK Vilhjalmsson This will require searching of the original publications and/or the consortia website. Fine-Needle Aspiration Cytology in Preoperative Diagnosis of Bone Lesions: A Three-Year Study in a Tertiary Care Hospital. She declares no other conflicts of interest. Lawlor That funding is not related to the comments in this paper. 1 The assumptions of two-sample MR are similar to those of one-sample MR, as are many of its strengths and limitations. Ware The Mendelian randomization analysis made it possible to examine the effects of lifelong naturally elevated testosterone levels on 469 traits and diseases. T et al.  Computationally that is difficult, but a recent GWAS of BMI trajectories from age 1 to 17 years shows some potential for future studies to be able to explore such possibilities. Participants 156 848 women in the multivariable regression and one sample mendelian randomisation (MR) analysis in UK … H Weak instrument bias in one-sample MR results in bias towards the confounded multivariable regression result, but in two-sample MR the bias is towards the null ( Table 1 ). Y Munafo This may be due to the winner’s curse because in the stage 1 we rank all 172 estimates of the allele score on the outcomes, such that the highest ranked are more likely to be higher than their respective true values because of the random variation of these sample estimates about their true values. 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. 17 Thus, the MR estimate of the effect of unadjusted WHR on cancer would be to bias it towards the null because the denominator of the ratio (the genetic instrument-WHR association) will be exaggerated due to adjustment for BMI. 4 Gamazon, E. et al. Columns are ‘hgnc_symbol’: HUGO Gene Nomenclature … 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. JAC MR Mendelian randomisation is a technique which, fuelled by the results of GWA studies, can be used to determine causal relationships between intermediate phenotypes such as metabolite levels and outcomes such as cardiovascular disease (Evans and Davey Smith 2015). H G Burgess Winner's curse. State Pattern C++, Rhystic Study - Prophecy, Eucalyptus Pulverulenta 'baby Blue For Sale, Black Cat Transparent Background, Newspaper Clippings Font,