Static.cambridge.org



Supplementary MaterialsSupplementary NotePage 2Construction of the lifetime smoking indexPage 2Genotyping and exclusion procedure.Instrument discovery for lifetime smoking indexInstrument validation – positive control outcomesInstrument validation – replication in an independent samplePage 2Page 3Page 3Page 5Genetic correlationsPage 6FiguresFigure S1. Quantile-quantile plot of lifetime smoking index Page 7Figure S2. Scatter plot of IVW and sensitivity analyses of lifetime smoking on coronary artery diseasePage 8Figure S3. Scatter plot of IVW and sensitivity analyses of lifetime smoking on lung cancerPage 9Figure S4. Single SNP analysis of lifetime smoking on coronary artery diseasePage 10Figure S5. Single SNP analysis of lifetime smoking on lung cancerPage 11Figure S6. Leave-one-out analysis of lifetime smoking on coronary artery diseasePage 12Figure S7. Leave-one-out SNP analysis of lifetime smoking on lung cancerPage 13Figure S8. Scatter plot of IVW and sensitivity analyses of lifetime smoking on schizophreniaPage 14Figure S9. Scatter plot of IVW and sensitivity analyses of lifetime smoking on major depressive disorderPage 15Figure S10. Scatter plot of IVW and sensitivity analyses of schizophrenia on lifetime smokingPage 16Figure S11. Scatter plot of IVW and sensitivity analyses of depression on lifetime smokingPage 17Figure S12. Single SNP analysis of lifetime smoking on schizophreniaPage 18Figure S13. Single SNP analysis of lifetime smoking on depressionPage 19Figure S14. Single SNP analysis of schizophrenia on lifetime smokingPage 20Figure S15. Single SNP analysis of depression on lifetime smokingPage 21Figure S16. Leave-one-out analysis of lifetime smoking on schizophreniaPage 22Figure S17. Leave-one-out analysis of lifetime smoking on depressionPage 23Figure S18. Leave-one-out analysis of schizophrenia on lifetime smokingPage 24Figure S19. Leave-one-out analysis of depression on lifetime smokingPage 25TablesTable S1. SNPs associated with lifetime smoking index at the genome-wide level of significancePage 26Table S2. Positive control two-sample MR to validate the lifetime smoking instrumentTable S3. Tests of the unweighted and weighted regression dilution I2GX for the lifetime smoking exposureTable S4. Tests of heterogeneityPage 30Page 31Page 32Table S5. Test of directional pleiotropy using the MR Egger interceptTable S6. Mendelian randomisation results following Steiger filtering. Page 34Page 35Table S7. Bi-directional two-sample Mendelian randomisation analyses of the effect of lifetime smoking on depression (2013). Table S8. Sensitivity analysis without genotyping chip used as a covariate. Table S9. Bi-directional two-sample Mendelian randomisation analyses of the effect of smoking on schizophrenia (2018).Table S10. Bi-directional two-sample Mendelian randomisation analyses of the effect of smoking on schizophrenia (CHRNA5-A3-B4 Variants Removed)Table S11. Bi-directional two-sample Mendelian randomisation analyses of lifetime smoking on depression (Howard et al., 2019 – ex UKBB and 23andMe)Table S12. Comparison of Mendelian randomisation sensitivity methods Page 36Page 37Page 38Page 39Page 40Page 41ReferencesPage 43Supplementary NoteConstruction of the lifetime smoking indexWe simulated values of two constants: half-life (τ), and lag time (δ). Together these constants capture the non-linear risk of smoking on health. Half-life captures the exponentially decreasing effect of smoking at a given time on health outcomes. Lag time accounts for the observation that smokers are more at risk of certain diseases (e.g., lung cancer) immediately after stopping smoking than current smokers ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"3mnk8qEz","properties":{"formattedCitation":"(Leffondr\\uc0\\u233{}, Abrahamowicz, Siemiatycki, & Rachet, 2002)","plainCitation":"(Leffondré, Abrahamowicz, Siemiatycki, & Rachet, 2002)","noteIndex":0},"citationItems":[{"id":2211,"uris":[""],"uri":[""],"itemData":{"id":2211,"type":"article-journal","title":"Modeling smoking history: a comparison of different approaches","container-title":"American journal of epidemiology","page":"813–823","volume":"156","issue":"9","source":"Google Scholar","title-short":"Modeling smoking history","author":[{"family":"Leffondré","given":"Karen"},{"family":"Abrahamowicz","given":"Michal"},{"family":"Siemiatycki","given":"Jack"},{"family":"Rachet","given":"Bernard"}],"issued":{"date-parts":[["2002"]]}}}],"schema":""} (Leffondré, Abrahamowicz, Siemiatycki, & Rachet, 2002). This is likely due to the prodromal consequences of disease being felt by the individual. Simulations were run for possible values of (between 2 and 50 varying in increments of 1)?and (between 0 and 5, varying in increments of 0.1) to find the best fitting model to explain the effects of lifetime smoking on lung cancer and all-cause mortality. Akaike information criterion (AIC) was used to select the best fitting model. The best fitting value for half-life was 18 for both lung cancer and all-cause mortality. We did not see any improvement of fit for changes in the value of lag time, therefore this was set to 0 as has been done previously ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a1gd0l7laeb","properties":{"formattedCitation":"(Lewer, McKee, Gasparrini, Reeves, & de Oliveira, 2017)","plainCitation":"(Lewer, McKee, Gasparrini, Reeves, & de Oliveira, 2017)","noteIndex":0},"citationItems":[{"id":1886,"uris":[""],"uri":[""],"itemData":{"id":1886,"type":"article-journal","title":"Socioeconomic position and mortality risk of smoking: evidence from the English Longitudinal Study of Ageing (ELSA)","container-title":"The European Journal of Public Health","page":"1068–1073","volume":"27","issue":"6","source":"Google Scholar","title-short":"Socioeconomic position and mortality risk of smoking","author":[{"family":"Lewer","given":"Dan"},{"family":"McKee","given":"Martin"},{"family":"Gasparrini","given":"Antonio"},{"family":"Reeves","given":"Aaron"},{"family":"Oliveira","given":"Cesar","non-dropping-particle":"de"}],"issued":{"date-parts":[["2017"]]}}}],"schema":""} (Lewer, McKee, Gasparrini, Reeves, & de Oliveira, 2017). This removes the effects of lag time from the model. These values were used to fit the final model which is:tsc* = max(tsc - , 0)dur* = max(dur + tsc - , 0) – tsc*lifetime smoking = (1 – 0.5dur*/) (0.5tsc*/) ln(int+1)…where = half-life, = lag time, int = cigarettes per day, tss = time started smoking, tsc = time since cessation, dur = duration of smoking (either age-tss for current smokers or [age-tsc]-tss for former smokers). So, in our case, where = 0, tsc* = tsc and consequently, our lifetime smoking calculation can be simplified to: lifetime smoking = (1 – 0.5dur/) (0.5tsc/) ln(int+1)Individuals who have never smoked and have no smoking exposure will have a value of 0. In our sample, values for smokers ranged from 0.007 (an individual who smoked 1 cigarette per day for 1 year) to 4.169 (an individual who currently smokes 140 a day and started smoking at age 11 years of age). Values of lifetime smoking were treated as continuous in subsequent analysis. Genotyping and exclusion procedureUK Biobank participants provided blood samples at initial assessment centre. Genotyping was performed using the Affymetrix UK BiLEVE Axiom array for 49,979 participants and using the Affymetrix UK Biobank Axiom? array for 438,398 participants. The two arrays share 95% coverage, but chip is adjusted for in all analyses because the UK BiLEVE sample is over represented for smokers. Imputation and initial quality control steps were performed by the Wellcome Trust Centre for Human Genetics resulting in over 90 million single nucleotide polymorphisms (SNPs) and indels ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a1tga3sb0ct","properties":{"formattedCitation":"(Bycroft et al., 2017)","plainCitation":"(Bycroft et al., 2017)","noteIndex":0},"citationItems":[{"id":1889,"uris":[""],"uri":[""],"itemData":{"id":1889,"type":"article-journal","title":"Genome-wide genetic data on ~500,000 UK Biobank participants","container-title":"bioRxiv","page":"166298","source":"","abstract":"The UK Biobank project is a large prospective cohort study of ~500,000 individuals from across the United Kingdom, aged between 40-69 at recruitment. A rich variety of phenotypic and health-related information is available on each participant, making the resource unprecedented in its size and scope. Here we describe the genome-wide genotype data (~805,000 markers) collected on all individuals in the cohort and its quality control procedures. Genotype data on this scale offers novel opportunities for assessing quality issues, although the wide range of ancestries of the individuals in the cohort also creates particular challenges. We also conducted a set of analyses that reveal properties of the genetic data (such as population structure and relatedness) that can be important for downstream analyses. In addition, we phased and imputed genotypes into the dataset, using computationally efficient methods combined with the Haplotype Reference Consortium (HRC) and UK10K haplotype resource. This increases the number of testable variants by over 100-fold to ~96 million variants. We also imputed classical allelic variation at 11 human leukocyte antigen (HLA) genes, and as a quality control check of this imputation, we replicate signals of known associations between HLA alleles and many common diseases. We describe tools that allow efficient genome-wide association studies (GWAS) of multiple traits and fast phenome-wide association studies (PheWAS), which work together with a new compressed file format that has been used to distribute the dataset. As a further check of the genotyped and imputed datasets, we performed a test-case genome-wide association scan on a well-studied human trait, standing height.","DOI":"10.1101/166298","language":"en","author":[{"family":"Bycroft","given":"Clare"},{"family":"Freeman","given":"Colin"},{"family":"Petkova","given":"Desislava"},{"family":"Band","given":"Gavin"},{"family":"Elliott","given":"Lloyd T."},{"family":"Sharp","given":"Kevin"},{"family":"Motyer","given":"Allan"},{"family":"Vukcevic","given":"Damjan"},{"family":"Delaneau","given":"Olivier"},{"family":"O'Connell","given":"Jared"},{"family":"Cortes","given":"Adrian"},{"family":"Welsh","given":"Samantha"},{"family":"McVean","given":"Gil"},{"family":"Leslie","given":"Stephen"},{"family":"Donnelly","given":"Peter"},{"family":"Marchini","given":"Jonathan"}],"issued":{"date-parts":[["2017",7,20]]}}}],"schema":""} (Bycroft et al., 2017). Individuals were excluded if there were sex-mismatches between reported and chromosomal sex or aneuploidy (N=814). Individuals were restricted to European ancestry based on the first four principal components of population structure and related individuals were removed following MRC Integrative Epidemiology Unit filtering steps ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a5eb2puk9a","properties":{"formattedCitation":"(Mitchell, Hemani, Dudding, & Paternoster, 2017)","plainCitation":"(Mitchell, Hemani, Dudding, & Paternoster, 2017)","noteIndex":0},"citationItems":[{"id":1786,"uris":[""],"uri":[""],"itemData":{"id":1786,"type":"webpage","title":"UK Biobank Genetic Data: MRC-IEU Quality Control, Version 1","container-title":"data.bris","abstract":"This is a full description of the quality control procedure undertaken and the derived files produced by the MRC-IEU associated with the full UK Biobank (July 2017) genetic data.","URL":"","note":"DOI: 10.5523/bris.3074krb6t2frj29yh2b03x3wxj","title-short":"UK Biobank Genetic Data","language":"en_GB","author":[{"family":"Mitchell","given":"R."},{"family":"Hemani","given":"G"},{"family":"Dudding","given":"T"},{"family":"Paternoster","given":"Lavinia"}],"issued":{"date-parts":[["2017",11,6]]},"accessed":{"date-parts":[["2018",1,16]]}}}],"schema":""} (Mitchell, Hemani, Dudding, & Paternoster, 2017). After excluding individuals who had withdrawn consent, 463,033 of the participants remained ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"ar18ijpu0","properties":{"formattedCitation":"(Mitchell et al., 2017)","plainCitation":"(Mitchell et al., 2017)","noteIndex":0},"citationItems":[{"id":1786,"uris":[""],"uri":[""],"itemData":{"id":1786,"type":"webpage","title":"UK Biobank Genetic Data: MRC-IEU Quality Control, Version 1","container-title":"data.bris","abstract":"This is a full description of the quality control procedure undertaken and the derived files produced by the MRC-IEU associated with the full UK Biobank (July 2017) genetic data.","URL":"","note":"DOI: 10.5523/bris.3074krb6t2frj29yh2b03x3wxj","title-short":"UK Biobank Genetic Data","language":"en_GB","author":[{"family":"Mitchell","given":"R."},{"family":"Hemani","given":"G"},{"family":"Dudding","given":"T"},{"family":"Paternoster","given":"Lavinia"}],"issued":{"date-parts":[["2017",11,6]]},"accessed":{"date-parts":[["2018",1,16]]}}}],"schema":""} (Mitchell et al., 2017). We restricted our analysis to autosomes only and used filtering thresholds for SNPs of minor allele frequency (MAF) >0.01 and info score (measure of imputation uncertainty) >0.8. Instrument discovery for lifetime smoking indexThe results of our GWAS of lifetime smoking (N = 462,690) are presented in Figure 1 and Supplementary Figure S1. The most strongly associated regions on chromosome 15 (in the CHRNA5-A3-B4 gene complex) and chromosome 9 (near the DBH gene) have been previously shown to be associated with smoking heaviness and cessation respectively ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"7ojvKD0b","properties":{"formattedCitation":"(Furberg et al., 2010)","plainCitation":"(Furberg et al., 2010)","noteIndex":0},"citationItems":[{"id":1907,"uris":[""],"uri":[""],"itemData":{"id":1907,"type":"article-journal","title":"Genome-wide meta-analyses identify multiple loci associated with smoking behavior","container-title":"Nature genetics","page":"441","volume":"42","issue":"5","source":"Google Scholar","author":[{"family":"Furberg","given":"Helena"},{"family":"Kim","given":"YunJung"},{"family":"Dackor","given":"Jennifer"},{"family":"Boerwinkle","given":"Eric"},{"family":"Franceschini","given":"Nora"},{"family":"Ardissino","given":"Diego"},{"family":"Bernardinelli","given":"Luisa"},{"family":"Mannucci","given":"Pier M."},{"family":"Mauri","given":"Francesco"},{"family":"Merlini","given":"Piera A."}],"issued":{"date-parts":[["2010"]]}}}],"schema":""} (Furberg et al., 2010). We identified 10,415 SNPs at the genome-wide level of significance (P < 5 × 10-8). After clumping and filtering, 126 independent SNPs remained. A full list of these SNPs and their z-scored effect sizes can be found in Supplementary Table S1. Instrument validation – positive control outcomesWe tested our genetic instrument using the positive controls of lung cancer, CAD and DNA methylation at the AHRR locus. We conducted these analyses using GWAS summary data in a two-sample MR framework using our exposure instrument for lifetime smoking from our GWAS in the UK Biobank. Outcome GWAS samples for instrument validationFor lung cancer, we used the summary data from the ILCCO consortium GWAS, which comprises 11,348 cases and 15,861 controls of European ancestry ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"gvU09ceN","properties":{"formattedCitation":"(Wang et al., 2014)","plainCitation":"(Wang et al., 2014)","noteIndex":0},"citationItems":[{"id":1900,"uris":[""],"uri":[""],"itemData":{"id":1900,"type":"article-journal","title":"Rare variants of large effect in BRCA2 and CHEK2 affect risk of lung cancer","container-title":"Nature genetics","page":"736","volume":"46","issue":"7","source":"Google Scholar","author":[{"family":"Wang","given":"Yufei"},{"family":"McKay","given":"James D."},{"family":"Rafnar","given":"Thorunn"},{"family":"Wang","given":"Zhaoming"},{"family":"Timofeeva","given":"Maria N."},{"family":"Broderick","given":"Peter"},{"family":"Zong","given":"Xuchen"},{"family":"Laplana","given":"Marina"},{"family":"Wei","given":"Yongyue"},{"family":"Han","given":"Younghun"}],"issued":{"date-parts":[["2014"]]}}}],"schema":""} (Wang et al., 2014). Lung cancer cases were classified by tumour type using either the International Classification of Diseases for Oncology (ICD-O) or the World Health Organisation (WHO) coding. Tumours with overlapping histologies were classified as mixed. Most samples in the GWAS meta-analysis only included adenocarcinomas (AD) or squamous carcinomas (SQ) but classification was different for each contributing cohort. The GWAS was run with a binary outcome, case control status for any primary lung cancer tumour. For more details see Wang et al. (2014)7. For CAD, we used the GWAS summary data from CARDIoGRAMplusC4D which comprises 60,801 cases and 123,504 controls of mixed ancestry ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"FttyCmp4","properties":{"formattedCitation":"(CARDIoGRAMplusC4D Consortium & others, 2015)","plainCitation":"(CARDIoGRAMplusC4D Consortium & others, 2015)","noteIndex":0},"citationItems":[{"id":1035,"uris":[""],"uri":[""],"itemData":{"id":1035,"type":"article-journal","title":"A comprehensive 1000 Genomes-based genome-wide association meta-analysis of coronary artery disease","container-title":"Nature genetics","page":"1121-1149","volume":"47","issue":"10","source":"Google Scholar","author":[{"family":"CARDIoGRAMplusC4D Consortium","given":""},{"literal":"others"}],"issued":{"date-parts":[["2015"]]}}}],"schema":""} (CARDIoGRAMplusC4D Consortium & others, 2015). Case status was defined as any CAD diagnosis including myocardial infarction, acute coronary syndrome, chronic stable angina or coronary stenosis of >50%. This GWAS was conducted with a binary outcome of CAD cases compared with controls. We conducted a GWAS of AHRR (cg05575921) methylation in the Accessible Resource for Integrated Epigenomic Studies (ARIES) subset of ALSPAC ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"omxnwHv2","properties":{"formattedCitation":"(Relton et al., 2015)","plainCitation":"(Relton et al., 2015)","noteIndex":0},"citationItems":[{"id":2116,"uris":[""],"uri":[""],"itemData":{"id":2116,"type":"article-journal","title":"Data resource profile: accessible resource for integrated epigenomic studies (ARIES)","container-title":"International journal of epidemiology","page":"1181–1190","volume":"44","issue":"4","source":"Google Scholar","title-short":"Data resource profile","author":[{"family":"Relton","given":"Caroline L."},{"family":"Gaunt","given":"Tom"},{"family":"McArdle","given":"Wendy"},{"family":"Ho","given":"Karen"},{"family":"Duggirala","given":"Aparna"},{"family":"Shihab","given":"Hashem"},{"family":"Woodward","given":"Geoff"},{"family":"Lyttleton","given":"Oliver"},{"family":"Evans","given":"David M."},{"family":"Reik","given":"Wolf"}],"issued":{"date-parts":[["2015"]]}}}],"schema":""} (Relton et al., 2015). Genome-wide DNA methylation profiling in ARIES was performed using the Illumina Infinium HumanMethylation450 BeadChip (450K) array for ~1000 mother-offspring pairs ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"LDrJzUUo","properties":{"formattedCitation":"(Relton et al., 2015)","plainCitation":"(Relton et al., 2015)","noteIndex":0},"citationItems":[{"id":2116,"uris":[""],"uri":[""],"itemData":{"id":2116,"type":"article-journal","title":"Data resource profile: accessible resource for integrated epigenomic studies (ARIES)","container-title":"International journal of epidemiology","page":"1181–1190","volume":"44","issue":"4","source":"Google Scholar","title-short":"Data resource profile","author":[{"family":"Relton","given":"Caroline L."},{"family":"Gaunt","given":"Tom"},{"family":"McArdle","given":"Wendy"},{"family":"Ho","given":"Karen"},{"family":"Duggirala","given":"Aparna"},{"family":"Shihab","given":"Hashem"},{"family":"Woodward","given":"Geoff"},{"family":"Lyttleton","given":"Oliver"},{"family":"Evans","given":"David M."},{"family":"Reik","given":"Wolf"}],"issued":{"date-parts":[["2015"]]}}}],"schema":""} (Relton et al., 2015). For this analysis, we used methylation data derived from whole blood which was collected from mothers (846 smokers and non-smokers) when the index offspring were around 18 years of age. Methylation data were normalised in R with the wateRmelon package ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"hcA8Tazk","properties":{"formattedCitation":"(Pidsley et al., 2013)","plainCitation":"(Pidsley et al., 2013)","noteIndex":0},"citationItems":[{"id":2296,"uris":[""],"uri":[""],"itemData":{"id":2296,"type":"article-journal","title":"A data-driven approach to preprocessing Illumina 450K methylation array data","container-title":"BMC genomics","page":"293","volume":"14","issue":"1","source":"Google Scholar","author":[{"family":"Pidsley","given":"Ruth"},{"family":"Wong","given":"Chloe CY"},{"family":"Volta","given":"Manuela"},{"family":"Lunnon","given":"Katie"},{"family":"Mill","given":"Jonathan"},{"family":"Schalkwyk","given":"Leonard C."}],"issued":{"date-parts":[["2013"]]}}}],"schema":""} (Pidsley et al., 2013) using the Touleimat and Tost ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"eKzWduoS","properties":{"formattedCitation":"(Touleimat & Tost, 2012)","plainCitation":"(Touleimat & Tost, 2012)","noteIndex":0},"citationItems":[{"id":2298,"uris":[""],"uri":[""],"itemData":{"id":2298,"type":"article-journal","title":"Complete pipeline for Infinium? Human Methylation 450K BeadChip data processing using subset quantile normalization for accurate DNA methylation estimation","container-title":"Epigenomics","page":"325–341","volume":"4","issue":"3","source":"Google Scholar","author":[{"family":"Touleimat","given":"Nizar"},{"family":"Tost","given":"J?rg"}],"issued":{"date-parts":[["2012"]]}}}],"schema":""} (Touleimat & Tost, 2012)algorithm to reduce the non-biological differences between probes. As was done previously ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"oSqegrED","properties":{"formattedCitation":"(Gaunt et al., 2016)","plainCitation":"(Gaunt et al., 2016)","noteIndex":0},"citationItems":[{"id":2301,"uris":[""],"uri":[""],"itemData":{"id":2301,"type":"article-journal","title":"Systematic identification of genetic influences on methylation across the human life course","container-title":"Genome biology","page":"61","volume":"17","issue":"1","source":"Google Scholar","author":[{"family":"Gaunt","given":"Tom R."},{"family":"Shihab","given":"Hashem A."},{"family":"Hemani","given":"Gibran"},{"family":"Min","given":"Josine L."},{"family":"Woodward","given":"Geoff"},{"family":"Lyttleton","given":"Oliver"},{"family":"Zheng","given":"Jie"},{"family":"Duggirala","given":"Aparna"},{"family":"McArdle","given":"Wendy L."},{"family":"Ho","given":"Karen"}],"issued":{"date-parts":[["2016"]]}}}],"schema":""} (Gaunt et al., 2016), AHRR methylation (cg05575921) was rank-normalised to remove outliers and regressed on the following covariates: age, the top ten ancestry principal components, bisulphite conversion batch and estimated white blood cell counts (using an algorithm based on differential methylation between cell types ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"wzS39ViG","properties":{"formattedCitation":"(Houseman et al., 2012)","plainCitation":"(Houseman et al., 2012)","noteIndex":0},"citationItems":[{"id":2303,"uris":[""],"uri":[""],"itemData":{"id":2303,"type":"article-journal","title":"DNA methylation arrays as surrogate measures of cell mixture distribution","container-title":"BMC bioinformatics","page":"86","volume":"13","issue":"1","source":"Google Scholar","author":[{"family":"Houseman","given":"Eugene Andres"},{"family":"Accomando","given":"William P."},{"family":"Koestler","given":"Devin C."},{"family":"Christensen","given":"Brock C."},{"family":"Marsit","given":"Carmen J."},{"family":"Nelson","given":"Heather H."},{"family":"Wiencke","given":"John K."},{"family":"Kelsey","given":"Karl T."}],"issued":{"date-parts":[["2012"]]}}}],"schema":""} (Houseman et al., 2012)). Residuals were then taken forward and SNP association effects were obtained in PLINK1.07 using exact linear regression. Full details of the GWAS methods used for AHRR locus methylation are described elsewhere ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"EiE5XLTY","properties":{"formattedCitation":"(Relton et al., 2015)","plainCitation":"(Relton et al., 2015)","noteIndex":0},"citationItems":[{"id":2116,"uris":[""],"uri":[""],"itemData":{"id":2116,"type":"article-journal","title":"Data resource profile: accessible resource for integrated epigenomic studies (ARIES)","container-title":"International journal of epidemiology","page":"1181–1190","volume":"44","issue":"4","source":"Google Scholar","title-short":"Data resource profile","author":[{"family":"Relton","given":"Caroline L."},{"family":"Gaunt","given":"Tom"},{"family":"McArdle","given":"Wendy"},{"family":"Ho","given":"Karen"},{"family":"Duggirala","given":"Aparna"},{"family":"Shihab","given":"Hashem"},{"family":"Woodward","given":"Geoff"},{"family":"Lyttleton","given":"Oliver"},{"family":"Evans","given":"David M."},{"family":"Reik","given":"Wolf"}],"issued":{"date-parts":[["2015"]]}}}],"schema":""} (Relton et al., 2015).Two-sample Mendelian randomisation of positive controlsAnalyses were conducted using MR Base, an R ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"afoq4tg85o","properties":{"formattedCitation":"(R. Core Team, 2014)","plainCitation":"(R. Core Team, 2014)","noteIndex":0},"citationItems":[{"id":133,"uris":[""],"uri":[""],"itemData":{"id":133,"type":"book","title":"R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2013","publisher":"ISBN 3-900051-07-0","source":"Google Scholar","title-short":"R","author":[{"family":"R. Core Team","given":""}],"issued":{"date-parts":[["2014"]]}}}],"schema":""} (R. Core Team, 2014) package for two-sample MR ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"n4XloeLp","properties":{"formattedCitation":"(Hemani et al., 2018)","plainCitation":"(Hemani et al., 2018)","noteIndex":0},"citationItems":[{"id":2078,"uris":[""],"uri":[""],"itemData":{"id":2078,"type":"article-journal","title":"The MR-Base platform supports systematic causal inference across the human phenome","container-title":"eLife","page":"e34408","volume":"7","source":"Google Scholar","author":[{"family":"Hemani","given":"Gibran"},{"family":"Zheng","given":"Jie"},{"family":"Elsworth","given":"Benjamin"},{"family":"Wade","given":"Kaitlin H."},{"family":"Haberland","given":"Valeriia"},{"family":"Baird","given":"Denis"},{"family":"Laurin","given":"Charles"},{"family":"Burgess","given":"Stephen"},{"family":"Bowden","given":"Jack"},{"family":"Langdon","given":"Ryan"}],"issued":{"date-parts":[["2018"]]}}}],"schema":""} (Hemani et al., 2018). We compared results across five different MR methods: inverse-variance weighted, MR Egger ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"4jzZpNmS","properties":{"formattedCitation":"(Bowden, Davey Smith, & Burgess, 2015)","plainCitation":"(Bowden, Davey Smith, & Burgess, 2015)","noteIndex":0},"citationItems":[{"id":1031,"uris":[""],"uri":[""],"itemData":{"id":1031,"type":"article-journal","title":"Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression","container-title":"International journal of epidemiology","page":"512–525","volume":"44","issue":"2","source":"Google Scholar","title-short":"Mendelian randomization with invalid instruments","author":[{"family":"Bowden","given":"Jack"},{"family":"Davey Smith","given":"George"},{"family":"Burgess","given":"Stephen"}],"issued":{"date-parts":[["2015"]]}}}],"schema":""} (Bowden, Davey Smith, & Burgess, 2015), weighted median ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"wP74zaPB","properties":{"formattedCitation":"(Bowden, Davey Smith, Haycock, & Burgess, 2016)","plainCitation":"(Bowden, Davey Smith, Haycock, & Burgess, 2016)","noteIndex":0},"citationItems":[{"id":1033,"uris":[""],"uri":[""],"itemData":{"id":1033,"type":"article-journal","title":"Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator","container-title":"Genetic epidemiology","page":"304–314","volume":"40","issue":"4","source":"Google Scholar","author":[{"family":"Bowden","given":"Jack"},{"family":"Davey Smith","given":"George"},{"family":"Haycock","given":"Philip C."},{"family":"Burgess","given":"Stephen"}],"issued":{"date-parts":[["2016"]]}}}],"schema":""} (Bowden, Davey Smith, Haycock, & Burgess, 2016), weighted mode ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"DVOoTSnS","properties":{"formattedCitation":"(Hartwig, Smith, & Bowden, 2017)","plainCitation":"(Hartwig, Smith, & Bowden, 2017)","noteIndex":0},"citationItems":[{"id":1783,"uris":[""],"uri":[""],"itemData":{"id":1783,"type":"article-journal","title":"Robust inference in two-sample Mendelian randomisation via the zero modal pleiotropy assumption","container-title":"bioRxiv","page":"126102","source":"Google Scholar","author":[{"family":"Hartwig","given":"Fernando Pires"},{"family":"Smith","given":"George Davey"},{"family":"Bowden","given":"Jack"}],"issued":{"date-parts":[["2017"]]}}}],"schema":""} (Hartwig, Smith, & Bowden, 2017) and MR RAPS ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"3TKoOxop","properties":{"formattedCitation":"(Zhao, Wang, Hemani, Bowden, & Small, 2018)","plainCitation":"(Zhao, Wang, Hemani, Bowden, & Small, 2018)","noteIndex":0},"citationItems":[{"id":2085,"uris":[""],"uri":[""],"itemData":{"id":2085,"type":"article-journal","title":"Statistical inference in two-sample summary-data Mendelian randomization using robust adjusted profile score","container-title":"arXiv","source":"","abstract":"Mendelian randomization (MR) is a method of exploiting genetic variation to unbiasedly estimate a causal effect in presence of unmeasured confounding. MR is being widely used in epidemiology and other related areas of population science. In this paper, we study statistical inference in the increasingly popular two-sample summary-data MR design. We show a linear model for the observed associations approximately holds in a wide variety of settings when all the genetic variants satisfy the exclusion restriction assumption, or in genetic terms, when there is no pleiotropy. In this scenario, we derive a maximum profile likelihood estimator with provable consistency and asymptotic normality. However, through analyzing real datasets, we find strong evidence of both systematic and idiosyncratic pleiotropy in MR, echoing some recent discoveries in statistical genetics. We model the systematic pleiotropy by a random effects model, where no genetic variant satisfies the exclusion restriction condition exactly. In this case we propose a consistent and asymptotically normal estimator by adjusting the profile score. We then tackle the idiosyncratic pleiotropy by robustifying the adjusted profile score. We demonstrate the robustness and efficiency of the proposed methods using several simulated and real datasets.","URL":"","note":"arXiv: 1801.09652","author":[{"family":"Zhao","given":"Qingyuan"},{"family":"Wang","given":"Jingshu"},{"family":"Hemani","given":"Gibran"},{"family":"Bowden","given":"Jack"},{"family":"Small","given":"Dylan S."}],"issued":{"date-parts":[["2018",1,29]]},"accessed":{"date-parts":[["2018",6,26]]}}}],"schema":""} (Zhao, Wang, Hemani, Bowden, & Small, 2018). Each method makes different assumptions and therefore a consistent effect across multiple methods strengthens causal evidence ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"apdr1trcat","properties":{"formattedCitation":"(Lawlor, Tilling, & Davey Smith, 2016)","plainCitation":"(Lawlor, Tilling, & Davey Smith, 2016)","noteIndex":0},"citationItems":[{"id":1771,"uris":[""],"uri":[""],"itemData":{"id":1771,"type":"article-journal","title":"Triangulation in aetiological epidemiology","container-title":"International journal of epidemiology","page":"1866–1886","volume":"45","issue":"6","source":"Google Scholar","author":[{"family":"Lawlor","given":"Debbie A."},{"family":"Tilling","given":"Kate"},{"family":"Davey Smith","given":"George"}],"issued":{"date-parts":[["2016"]]}}}],"schema":""} (Lawlor, Tilling, & Davey Smith, 2016). If a SNP was unavailable in the outcome GWAS summary statistics, then proxy SNPs were searched for with a minimum LD r2 = 0.8. We aligned palindromic SNPs with MAF<0.3.Results We validated the genetic instrument for lifetime smoking exposure using two-sample MR of smoking on positive control outcomes: lung cancer, CAD and hypomethylation at the AHRR locus. All five MR methods indicated the expected direction of effect (see Supplementary Table S2) increasing risk of disease outcomes and decreasing AHRR methylation, with the exception of the MR Egger SIMEX adjusted estimates for CAD. However, these should be interpreted with caution given the low I2GX (see Supplementary Table S3) ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"bbCYGP7X","properties":{"formattedCitation":"(Bowden, Del Greco M, et al., 2016)","plainCitation":"(Bowden, Del Greco M, et al., 2016)","noteIndex":0},"citationItems":[{"id":1751,"uris":[""],"uri":[""],"itemData":{"id":1751,"type":"article-journal","title":"Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I 2 statistic","container-title":"International journal of epidemiology","page":"1961–1974","volume":"45","issue":"6","source":"Google Scholar","title-short":"Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression","author":[{"family":"Bowden","given":"Jack"},{"family":"Del Greco M","given":"Fabiola"},{"family":"Minelli","given":"Cosetta"},{"family":"Davey Smith","given":"George"},{"family":"Sheehan","given":"Nuala A."},{"family":"Thompson","given":"John R."}],"issued":{"date-parts":[["2016"]]}}}],"schema":""} (Bowden, Del Greco M, et al., 2016). There was strong evidence of an effect of lifetime smoking on increased odds of lung cancer and CAD. There was weaker evidence of an effect on AHRR methylation, possibly due to lower power. Sensitivity analyses are presented in Supplementary Figures S2-S7. There was evidence of significant heterogeneity in the SNP-exposure effects (see Supplementary Table S3); however, tests of MR Egger intercepts generally indicated weak evidence of directional pleiotropy (see Supplementary Table S4), with the exception of CAD. Instrument validation – replication in an independent sample ALSPAC sample description and measuresTo test prediction in an independent sample, we used 2,712 mothers from Avon Longitudinal Study of Parents and Children (ALSPAC). ALSPAC is a longitudinal birth cohort, which recruited 14,541 pregnant women between April 1991 and December 1992 with detailed descriptions reported elsewhere ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a1psr7u8bf6","properties":{"formattedCitation":"(Boyd et al., 2013; Fraser et al., 2012)","plainCitation":"(Boyd et al., 2013; Fraser et al., 2012)","noteIndex":0},"citationItems":[{"id":1436,"uris":[""],"uri":[""],"itemData":{"id":1436,"type":"article-journal","title":"Cohort profile: the ‘children of the 90s’—the index offspring of the Avon Longitudinal Study of Parents and Children","container-title":"International journal of epidemiology","page":"111–127","volume":"42","issue":"1","source":"Google Scholar","title-short":"Cohort profile","author":[{"family":"Boyd","given":"Andy"},{"family":"Golding","given":"Jean"},{"family":"Macleod","given":"John"},{"family":"Lawlor","given":"Debbie A."},{"family":"Fraser","given":"Abigail"},{"family":"Henderson","given":"John"},{"family":"Molloy","given":"Lynn"},{"family":"Ness","given":"Andy"},{"family":"Ring","given":"Susan"},{"family":"Davey Smith","given":"George"}],"issued":{"date-parts":[["2013"]]}}},{"id":1433,"uris":[""],"uri":[""],"itemData":{"id":1433,"type":"article-journal","title":"Cohort profile: the Avon Longitudinal Study of Parents and Children: ALSPAC mothers cohort","container-title":"International journal of epidemiology","page":"97–110","volume":"42","issue":"1","source":"Google Scholar","title-short":"Cohort profile","author":[{"family":"Fraser","given":"Abigail"},{"family":"Macdonald-Wallis","given":"Corrie"},{"family":"Tilling","given":"Kate"},{"family":"Boyd","given":"Andy"},{"family":"Golding","given":"Jean"},{"family":"Davey Smith","given":"George"},{"family":"Henderson","given":"John"},{"family":"Macleod","given":"John"},{"family":"Molloy","given":"Lynn"},{"family":"Ness","given":"Andy"},{"literal":"others"}],"issued":{"date-parts":[["2012"]]}}}],"schema":""} (Boyd et al., 2013; Fraser et al., 2012). The study website contains details of all the data that is available through a fully searchable data dictionary and variable search tool (bristol.ac.uk/alspac/researchers/our-data/). Self-reported measures of smoking status, age at initiation, age at cessation and cigarettes per day were collected from these women when their offspring were 18 years of age (mean mothers age = 48 years, SD = 4.3). Statistical AnalysisPLINK was used to generate a polygenic risk score in ALSPAC ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"XyezWYXV","properties":{"formattedCitation":"(Purcell et al., 2007)","plainCitation":"(Purcell et al., 2007)","noteIndex":0},"citationItems":[{"id":594,"uris":[""],"uri":[""],"itemData":{"id":594,"type":"article-journal","title":"PLINK: a tool set for whole-genome association and population-based linkage analyses","container-title":"The American Journal of Human Genetics","page":"559–575","volume":"81","issue":"3","source":"Google Scholar","title-short":"PLINK","author":[{"family":"Purcell","given":"Shaun"},{"family":"Neale","given":"Benjamin"},{"family":"Todd-Brown","given":"Kathe"},{"family":"Thomas","given":"Lori"},{"family":"Ferreira","given":"Manuel AR"},{"family":"Bender","given":"David"},{"family":"Maller","given":"Julian"},{"family":"Sklar","given":"Pamela"},{"family":"De Bakker","given":"Paul IW"},{"family":"Daly","given":"Mark J."},{"literal":"others"}],"issued":{"date-parts":[["2007"]]}}}],"schema":""} (Purcell et al., 2007). Linear regression was used to estimate the percentage variance of lifetime smoking explained by the polygenic score. ResultsIn the ALSPAC independent sample, the 126 SNPs explained 0.36% of the variance in lifetime smoking (P = 0.002).Genetic CorrelationsMethodsGenetic correlations were first calculated between lifetime smoking, smoking initiation ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"apoai3rc0e","properties":{"formattedCitation":"(Liu et al., 2019)","plainCitation":"(Liu et al., 2019)","noteIndex":0},"citationItems":[{"id":2356,"uris":[""],"uri":[""],"itemData":{"id":2356,"type":"article-journal","title":"Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use","container-title":"Nature genetics","page":"1","source":"Google Scholar","author":[{"family":"Liu","given":"Mengzhen"},{"family":"Jiang","given":"Yu"},{"family":"Wedow","given":"Robbee"},{"family":"Li","given":"Yue"},{"family":"Brazel","given":"David M."},{"family":"Chen","given":"Fang"},{"family":"Datta","given":"Gargi"},{"family":"Davila-Velderrain","given":"Jose"},{"family":"McGuire","given":"Daniel"},{"family":"Tian","given":"Chao"}],"issued":{"date-parts":[["2019"]]}}}],"schema":""} (Liu et al., 2019), depression ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a13sapmagdl","properties":{"formattedCitation":"(Wray, Sullivan, & others, 2018)","plainCitation":"(Wray, Sullivan, & others, 2018)","noteIndex":0},"citationItems":[{"id":1276,"uris":[""],"uri":[""],"itemData":{"id":1276,"type":"article-journal","title":"Genome-wide association analyses identify 44 risk variants and refine the genetic architecture of major depression","container-title":"Nature Genetics","page":"668-681","volume":"50","source":"Google Scholar","author":[{"family":"Wray","given":"Naomi R."},{"family":"Sullivan","given":"Patrick F."},{"literal":"others"}],"issued":{"date-parts":[["2018"]]}}}],"schema":""} (Wray, Sullivan, & others, 2018) and schizophrenia ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"aiof5osren","properties":{"formattedCitation":"(Ripke et al., 2014)","plainCitation":"(Ripke et al., 2014)","noteIndex":0},"citationItems":[{"id":663,"uris":[""],"uri":[""],"itemData":{"id":663,"type":"article-journal","title":"Biological insights from 108 schizophrenia-associated genetic loci","container-title":"Nature","page":"421-427","volume":"511","issue":"7510","source":"Google Scholar","author":[{"family":"Ripke","given":"Stephan"},{"family":"Neale","given":"Benjamin M."},{"family":"Corvin","given":"Aiden"},{"family":"Walters","given":"James TR"},{"family":"Farh","given":"Kai-How"},{"family":"Holmans","given":"Peter A."},{"family":"Lee","given":"Phil"},{"family":"Bulik-Sullivan","given":"Brendan"},{"family":"Collier","given":"David A."},{"family":"Huang","given":"Hailiang"},{"literal":"others"}],"issued":{"date-parts":[["2014"]]}}}],"schema":""} (Ripke et al., 2014) using the summary statistics outlined in the main text. Full summary statistics were only available for depression and smoking initiation excluding 23andMe. Genetic correlations were calculated using the LD Score Regression software ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a20f0prijcc","properties":{"formattedCitation":"(Bulik-Sullivan et al., 2015)","plainCitation":"(Bulik-Sullivan et al., 2015)","noteIndex":0},"citationItems":[{"id":2571,"uris":[""],"uri":[""],"itemData":{"id":2571,"type":"article-journal","title":"LD Score regression distinguishes confounding from polygenicity in genome-wide association studies","container-title":"Nature Genetics","page":"291-295","volume":"47","issue":"3","source":"PubMed","abstract":"Both polygenicity (many small genetic effects) and confounding biases, such as cryptic relatedness and population stratification, can yield an inflated distribution of test statistics in genome-wide association studies (GWAS). However, current methods cannot distinguish between inflation from a true polygenic signal and bias. We have developed an approach, LD Score regression, that quantifies the contribution of each by examining the relationship between test statistics and linkage disequilibrium (LD). The LD Score regression intercept can be used to estimate a more powerful and accurate correction factor than genomic control. We find strong evidence that polygenicity accounts for the majority of the inflation in test statistics in many GWAS of large sample size.","DOI":"10.1038/ng.3211","ISSN":"1546-1718","note":"PMID: 25642630\nPMCID: PMC4495769","journalAbbreviation":"Nat. Genet.","language":"eng","author":[{"family":"Bulik-Sullivan","given":"Brendan K."},{"family":"Loh","given":"Po-Ru"},{"family":"Finucane","given":"Hilary K."},{"family":"Ripke","given":"Stephan"},{"family":"Yang","given":"Jian"},{"literal":"Schizophrenia Working Group of the Psychiatric Genomics Consortium"},{"family":"Patterson","given":"Nick"},{"family":"Daly","given":"Mark J."},{"family":"Price","given":"Alkes L."},{"family":"Neale","given":"Benjamin M."}],"issued":{"date-parts":[["2015",3]]}}}],"schema":""} (Bulik-Sullivan et al., 2015).ResultsThere was evidence of positive genetic correlations between both smoking phenotypes (rG = 0.868, <0.001), between lifetime smoking and depression (rG = 0.404, p<0.001) and between lifetime smoking and schizophrenia (rG = 0.160, p<0.001). Genetic correlations between smoking initiation and mental health are reported elsewhere ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a1hb14cjo1u","properties":{"formattedCitation":"(Liu et al., 2019)","plainCitation":"(Liu et al., 2019)","noteIndex":0},"citationItems":[{"id":2356,"uris":[""],"uri":[""],"itemData":{"id":2356,"type":"article-journal","title":"Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use","container-title":"Nature genetics","page":"1","source":"Google Scholar","author":[{"family":"Liu","given":"Mengzhen"},{"family":"Jiang","given":"Yu"},{"family":"Wedow","given":"Robbee"},{"family":"Li","given":"Yue"},{"family":"Brazel","given":"David M."},{"family":"Chen","given":"Fang"},{"family":"Datta","given":"Gargi"},{"family":"Davila-Velderrain","given":"Jose"},{"family":"McGuire","given":"Daniel"},{"family":"Tian","given":"Chao"}],"issued":{"date-parts":[["2019"]]}}}],"schema":""} (Liu et al., 2019). There was a positive genetic correlation between smoking initiation and schizophrenia (rG = 0.137, p<0.001) and between smoking initiation and major depression (rG = 0.186, p<0.001). Figure 1. Quantile-quantile plot of SNP associations with lifetime smoking index. The x-axis indicates the -log10 p-value expected under the null hypothesis and the y-axis represents the observed -log10 p-value.Testing inflation by population structureThe QQ plot suggests evidence of inflation due to population stratification. To check this, we calculated the LD score intercept (1.051, SE =?0.0094) and mean chi-square?statistic (1.907) which give an attenuation ratio of 0.056, minimal evidence of inflation. The pattern seen here is therefore likely due to the large sample size giving power to detect high numbers of associations. Figure S2. Scatter plot of IVW and sensitivity analyses of lifetime smoking on coronary artery disease.Figure S3. Scatter plot of IVW and sensitivity analyses of lifetime smoking on lung cancer.Figure S3 shows evidence of an outlier, however this plot does not account for SE in the SNP effect. Therefore, we followed up this outlier using radial MR. Radial MRA radial MR analysis of lifetime smoking on lung cancer using first order weights and an alpha level of 0.01 identified 5 outliers for the IVW method (in order of how large an outlier they are):IVW Outliers1. rs10918701 2. rs12244388 3. rs3291204. rs6562474 5. rs8042849 Outliers are usually removed in an incremental fashion, beginning with the largest. However, we can see from our leave-one-out analysis (see Figure S7) that the two largest outliers for both methods (rs10918701 and rs12244388) do not affect the estimate once removed. Figure S4. Single SNP effects of lifetime smoking on coronary artery disease.Figure S5. Single SNP effects of lifetime smoking on lung cancer.The SNP with the largest effect is rs8042849, which is an intron variant of the HYKK gene, has previously been associated with nicotine dependence and forced expiratory volume (FEV). Figure S6. Leave-one-out effects of lifetime smoking on coronary artery disease using the inverse-variance weighted method.Figure S7. Leave-one-out effects of lifetime smoking on lung cancer using the inverse-variance weighted method.Figure S8. Scatter plot of IVW and sensitivity analyses of lifetime smoking on schizophrenia.Figure S9. Scatter plot of IVW and sensitivity analyses of lifetime smoking on major depressive disorder.Figure S10. Scatter plot of IVW and sensitivity analyses of schizophrenia on lifetime smoking.Figure S11. Scatter plot of IVW and sensitivity analyses of depression on lifetime smoking.Figure S12. Single SNP analysis of lifetime smoking on schizophrenia.Figure S13. Single SNP analysis of lifetime smoking on depression.Figure S14. Single SNP analysis of schizophrenia on lifetime smoking.The SNP with the largest effect on smoking is from rs8042374 located in the CHRNA3 gene (known to be associated with Nicotine dependence ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"dhJVOMWA","properties":{"formattedCitation":"(Munaf\\uc0\\u242{} et al., 2012; Thorgeirsson et al., 2008; Tobacco Consortium, 2010; Ware, van den Bree, & Munaf\\uc0\\u242{}, 2011)","plainCitation":"(Munafò et al., 2012; Thorgeirsson et al., 2008; Tobacco Consortium, 2010; Ware, van den Bree, & Munafò, 2011)","noteIndex":0},"citationItems":[{"id":2025,"uris":[""],"uri":[""],"itemData":{"id":2025,"type":"article-journal","title":"Association Between Genetic Variants on Chromosome 15q25 Locus and Objective Measures of Tobacco Exposure","container-title":"JNCI: Journal of the National Cancer Institute","page":"740-748","volume":"104","issue":"10","source":"academic.","abstract":"BackgroundTwo single-nucleotide polymorphisms, rs1051730 and rs16969968, located within the nicotinic acetylcholine receptor gene cluster on chromosome 15q25 locus, are associated with heaviness of smoking, risk for lung cancer, and other smoking-related health outcomes. Previous studies have typically relied on self-reported smoking behavior, which may not fully capture interindividual variation in tobacco exposure.MethodsWe investigated the association of rs1051730 and rs16969968 genotype (referred to as rs1051730–rs16969968, because these are in perfect linkage disequilibrium and interchangeable) with both self-reported daily cigarette consumption and biochemically measured plasma or serum cotinine levels among cigarette smokers. Summary estimates and descriptive statistical data for 12?364 subjects were obtained from six independent studies, and 2932 smokers were included in the analyses. Linear regression was used to calculate the per-allele association of rs1051730–rs16969968 genotype with cigarette consumption and cotinine levels in current smokers for each study. Meta-analysis of per-allele associations was conducted using a random effects method. The likely resulting association between genotype and lung cancer risk was assessed using published data on the association between cotinine levels and lung cancer risk. All statistical tests were two-sided.ResultsPooled per-allele associations showed that current smokers with one or two copies of the rs1051730–rs16969968 risk allele had increased self-reported cigarette consumption (mean increase in unadjusted number of cigarettes per day per allele = 1.0 cigarette, 95% confidence interval [CI] = 0.57 to 1.43 cigarettes, P = 5.22 × 10?6) and cotinine levels (mean increase in unadjusted cotinine levels per allele = 138.72 nmol/L, 95% CI = 97.91 to 179.53 nmol/L, P = 2.71 × 10?11). The increase in cotinine levels indicated an increased risk of lung cancer with each additional copy of the rs1051730–rs16969968 risk allele (per-allele odds ratio = 1.31, 95% CI = 1.21 to 1.42).ConclusionsOur data show a stronger association of rs1051730–rs16969968 genotype with objective measures of tobacco exposure compared with self-reported cigarette consumption. The association of these variants with lung cancer risk is likely to be mediated largely, if not wholly, via tobacco exposure.","DOI":"10.1093/jnci/djs191","ISSN":"0027-8874","journalAbbreviation":"J Natl Cancer Inst","language":"en","author":[{"family":"Munafò","given":"Marcus R."},{"family":"Timofeeva","given":"Maria N."},{"family":"Morris","given":"Richard W."},{"family":"Prieto-Merino","given":"David"},{"family":"Sattar","given":"Naveed"},{"family":"Brennan","given":"Paul"},{"family":"Johnstone","given":"Elaine C."},{"family":"Relton","given":"Caroline"},{"family":"Johnson","given":"Paul C. D."},{"family":"Walther","given":"Donna"},{"family":"Whincup","given":"Peter H."},{"family":"Casas","given":"Juan P."},{"family":"Uhl","given":"George R."},{"family":"Vineis","given":"Paolo"},{"family":"Padmanabhan","given":"Sandosh"},{"family":"Jefferis","given":"Barbara J."},{"family":"Amuzu","given":"Antoinette"},{"family":"Riboli","given":"Elio"},{"family":"Upton","given":"Mark N."},{"family":"Aveyard","given":"Paul"},{"family":"Ebrahim","given":"Shah"},{"family":"Hingorani","given":"Aroon D."},{"family":"Watt","given":"Graham"},{"family":"Palmer","given":"Tom M."},{"family":"Timpson","given":"Nicholas J."},{"family":"Davey Smith","given":"George"}],"issued":{"date-parts":[["2012",5,16]]}}},{"id":2037,"uris":[""],"uri":[""],"itemData":{"id":2037,"type":"article-journal","title":"A Variant Associated with Nicotine Dependence, Lung Cancer and Peripheral Arterial Disease","container-title":"Nature","page":"638-642","volume":"452","issue":"7187","source":"PubMed Central","abstract":"Smoking is a leading cause of preventable death, causing approximately five million premature deaths world-wide each year, . Evidence for genetic influence on smoking behaviour and nicotine dependence (ND)- has prompted a search for susceptibility genes. Furthermore, assessing the impact of sequence variants on smoking-related diseases is important for public health reasons, . Smoking is the major risk factor for lung cancer (LC)-, and one of the main risk factors for peripheral arterial disease (PAD)-. We have identified a common variant in the nicotinic acetylcholine receptor gene cluster on chromosome 15q24 with an effect on smoking quantity, ND and the risk of two smoking-related diseases in populations of European descent. The variant has an effect on the number of cigarettes smoked per day in 15,771 smokers (P=6×10?20). The same variant associated with ND in a previous genome-wide association study using low quantity smokers as controls (OR=1.3, P=1×10?3), , and with a similar approach we observe a highly significant association with ND (OR =1.40, P=7×10?15). Comparison of LC (N=1,024) and PAD (N= 2,738) cases with about 30,000 population controls each showed that the variant confers risk of LC (OR=1.31, P=1.5×10?8) and PAD (OR=1.19, P=1.4×10?7). The findings highlight the role of nicotine addiction in the pathogenesis of other serious diseases and provide a case study of the role of active gene-environment correlation in the pathogenesis of disease.","DOI":"10.1038/nature06846","ISSN":"0028-0836","note":"PMID: 18385739\nPMCID: PMC4539558","journalAbbreviation":"Nature","author":[{"family":"Thorgeirsson","given":"Thorgeir E."},{"family":"Geller","given":"Frank"},{"family":"Sulem","given":"Patrick"},{"family":"Rafnar","given":"Thorunn"},{"family":"Wiste","given":"Anna"},{"family":"Magnusson","given":"Kristinn P."},{"family":"Manolescu","given":"Andrei"},{"family":"Thorleifsson","given":"Gudmar"},{"family":"Stefansson","given":"Hreinn"},{"family":"Ingason","given":"Andres"},{"family":"Stacey","given":"Simon N."},{"family":"Bergthorsson","given":"Jon T."},{"family":"Thorlacius","given":"Steinunn"},{"family":"Gudmundsson","given":"Julius"},{"family":"Jonsson","given":"Thorlakur"},{"family":"Jakobsdottir","given":"Margret"},{"family":"Saemundsdottir","given":"Jona"},{"family":"Olafsdottir","given":"Olof"},{"family":"Gudmundsson","given":"Larus J."},{"family":"Bjornsdottir","given":"Gyda"},{"family":"Kristjansson","given":"Kristleifur"},{"family":"Skuladottir","given":"Halla"},{"family":"Isaksson","given":"Helgi J."},{"family":"Gudbjartsson","given":"Tomas"},{"family":"Jones","given":"Gregory T."},{"family":"Mueller","given":"Thomas"},{"family":"Gotts?ter","given":"Anders"},{"family":"Flex","given":"Andrea"},{"family":"Aben","given":"Katja K.H."},{"family":"Vegt","given":"Femmie","non-dropping-particle":"de"},{"family":"Mulders","given":"Peter F.A."},{"family":"Isla","given":"Dolores"},{"family":"Vidal","given":"Maria J."},{"family":"Asin","given":"Laura"},{"family":"Saez","given":"Berta"},{"family":"Murillo","given":"Laura"},{"family":"Blondal","given":"Thorsteinn"},{"family":"Kolbeinsson","given":"Halldor"},{"family":"Stefansson","given":"Jon G."},{"family":"Hansdottir","given":"Ingunn"},{"family":"Runarsdottir","given":"Valgerdur"},{"family":"Pola","given":"Roberto"},{"family":"Lindblad","given":"Bengt"},{"family":"Rij","given":"Andre M.","non-dropping-particle":"van"},{"family":"Dieplinger","given":"Benjamin"},{"family":"Haltmayer","given":"Meinhard"},{"family":"Mayordomo","given":"Jose I."},{"family":"Kiemeney","given":"Lambertus A."},{"family":"Matthiasson","given":"Stefan E."},{"family":"Oskarsson","given":"Hogni"},{"family":"Tyrfingsson","given":"Thorarinn"},{"family":"Gudbjartsson","given":"Daniel F."},{"family":"Gulcher","given":"Jeffrey R."},{"family":"Jonsson","given":"Steinn"},{"family":"Thorsteinsdottir","given":"Unnur"},{"family":"Kong","given":"Augustine"},{"family":"Stefansson","given":"Kari"}],"issued":{"date-parts":[["2008",4,3]]}}},{"id":241,"uris":[""],"uri":[""],"itemData":{"id":241,"type":"article-journal","title":"Genome-wide meta-analyses identify multiple loci associated with smoking behavior","container-title":"Nature genetics","page":"441–447","volume":"42","issue":"5","source":"Google Scholar","author":[{"literal":"Tobacco Consortium"}],"issued":{"date-parts":[["2010"]]}}},{"id":2022,"uris":[""],"uri":[""],"itemData":{"id":2022,"type":"article-journal","title":"Association of the CHRNA5-A3-B4 gene cluster with heaviness of smoking: a meta-analysis","container-title":"Nicotine & Tobacco Research","page":"1167–1175","volume":"13","issue":"12","source":"Google Scholar","title-short":"Association of the CHRNA5-A3-B4 gene cluster with heaviness of smoking","author":[{"family":"Ware","given":"Jennifer J."},{"family":"Bree","given":"Marianne BM","non-dropping-particle":"van den"},{"family":"Munafò","given":"Marcus R."}],"issued":{"date-parts":[["2011"]]}}}],"schema":""} (Munafò et al., 2012; Thorgeirsson et al., 2008; Tobacco Consortium, 2010; Ware, van den Bree, & Munafò, 2011). However, removing this SNP did not remove the effect (see Supplementary Figure S18). Figure S15. Single SNP analysis of depression on lifetime smoking.Figure S16. Leave-one-out analysis of lifetime smoking on schizophrenia.Figure S17. Leave-one-out analysis of lifetime smoking on depression.Figure 18. Leave-one-out analysis of schizophrenia on lifetime smoking.Figure S19. Leave-one-out analysis of depression on lifetime smoking.Table S1. SNPs associated with lifetime smoking index at the genome-wide level of significance (p<5x10-8) and clumped for independence at kb=10000 and r2=0.001, in ascending order of p-value. SNPCHRBPEANonEAEAFBetaSEp-valuers80428491578817929CT0.3420.0280.0021.80E-39rs11338241991.36E+08CA0.889-0.0410.0033.00E-37rs60117792061984317CT0.1910.0280.0032.30E-27rs9919670111.13E+08GA0.612-0.0220.0027.60E-27rs289077221.46E+08GT0.413-0.0200.0022.10E-22rs351758341547680815GA0.788-0.0240.0024.60E-22rs12244388101.05E+08GA0.661-0.0190.0021.40E-19rs11783093827425349CT0.8390.0230.0031.20E-16rs11210229173860028AG0.3840.0170.0022.00E-16rs6215587421.06E+08AG0.873-0.0240.0035.20E-16rs10226228732315613AG0.630-0.0160.0022.00E-15rs61198972031145415GA0.762-0.0180.0023.60E-15rs28671122651349TG0.8350.0210.0034.80E-15rs98639151.67E+08GA0.3670.0160.0029.40E-15rs3742365141.04E+08TC0.595-0.0160.0022.50E-14rs240192471.15E+08GC0.5020.0150.0022.70E-14rs780701971.18E+08AG0.540-0.0150.0026.70E-14rs549845144076469GA0.3010.0160.0028.30E-14rs10922907191193049AT0.4510.0150.0023.00E-13rs7569203245154418AC0.689-0.0160.0027.40E-13rs173098741127667236GA0.740-0.0160.0029.70E-13rs6778080349317338TC0.2670.0160.0021.30E-12rs80421341597514404TG0.541-0.0140.0021.30E-12rs1757659441.48E+08GA0.7240.0160.0021.70E-12rs776661061.12E+08CA0.1830.0180.0032.20E-12rs192201873560401CT0.3640.0140.0023.00E-12rs7553348175005067GA0.4380.0140.0025.20E-12rs7528604166407352GA0.5660.0140.0025.70E-12rs32912051.34E+08CT0.5810.0140.0026.30E-12rs1262370222.03E+08AG0.613-0.0140.0027.70E-12rs1329651991.28E+08GT0.606-0.0140.0028.10E-12rs6935954626255451AG0.4210.0140.0028.20E-12rs4671357260136176TC0.519-0.0140.0021.10E-11rs3896224101.06E+08AG0.5850.0140.0021.10E-11rs32634131.08E+08GA0.5250.0140.0021.20E-11rs43918021128674592AG0.7070.0150.0021.40E-11rs7267886441.12E+08GA0.8290.0180.0031.60E-11rs1122822191146632809GA0.959-0.0330.0053.80E-11rs108798711275380511TG0.343-0.0140.0025.00E-11rs8893981669556715CT0.5880.0130.0026.30E-11rs447334821.82E+08AT0.250-0.0150.0026.40E-11rs122114891.22E+08CG0.5870.0130.0027.30E-11rs317021435418368TA0.814-0.0170.0031.10E-10rs1933270149977965TG0.3640.0130.0021.50E-10rs86141727588806CA0.817-0.0170.0031.80E-10rs11255908108802912TG0.743-0.0150.0022.30E-10rs1315339351.68E+08AG0.884-0.0200.0032.50E-10rs267867021.04E+08AT0.4860.0130.0023.10E-10rs7333559131.01E+08GA0.2120.0150.0023.20E-10rs76608582194474725CA0.9530.0310.0053.20E-10rs421983384892866TC0.5190.0130.0023.30E-10rs454359293014254TC0.520-0.0120.0024.50E-10rs11948770513246336TC0.768-0.0150.0024.90E-10rs7039819982430418GA0.4270.0130.0025.10E-10rs1028229271.11E+08CT0.3620.0130.0025.90E-10rs28388342146665208CT0.699-0.0130.0026.30E-10rs62483342881256TG0.6950.0130.0026.60E-10rs62135536244326028CT0.9680.0350.0068.00E-10rs381103821.13E+08TC0.724-0.0140.0028.90E-10rs359243260475509TC0.393-0.0130.0029.50E-10rs11768481796629103CA0.6660.0130.0029.90E-10rs6779302316859710GT0.633-0.0130.0021.20E-09rs3516960689604066TG0.6120.0130.0021.20E-09rs675960671750333733GA0.649-0.0130.0021.20E-09rs26756381063576286GA0.5810.0120.0021.30E-09rs757424061117070365GA0.7390.0140.0021.30E-09rs713675451877576337GA0.791-0.0150.0021.40E-09rs71627581543161351GA0.8890.0190.0031.60E-09rs13016665257995348CA0.577-0.0120.0021.80E-09rs3692301689645437GT0.308-0.0130.0021.80E-09rs10052591550812738TC0.5730.0120.0022.10E-09rs376994921.66E+08TA0.528-0.0120.0022.50E-09rs71555951477502546AC0.674-0.0130.0022.50E-09rs7077678101.04E+08CT0.6230.0120.0022.60E-09rs8603261457342912CT0.4280.0120.0022.70E-09rs12202536667475273AG0.513-0.0120.0022.80E-09rs48148732019616429CT0.7670.0140.0022.90E-09rs1474126942140702786GA0.850-0.0170.0032.90E-09rs984294731.57E+08CT0.326-0.0130.0023.10E-09rs2894808652861990TA0.922-0.0220.0043.50E-09rs127086651624728227AG0.285-0.0130.0023.50E-09rs2026452241798520AG0.203-0.0150.0023.90E-09rs620980131850863861GA0.640-0.0120.0024.10E-09rs495752851.06E+08AC0.208-0.0150.0024.20E-09rs1246265986761745TC0.305-0.0130.0024.20E-09rs65985391599204483TC0.489-0.0120.0024.50E-09rs1300900821.74E+08AG0.3280.0120.0024.60E-09rs175532621092912773AC0.885-0.0180.0035.30E-09rs72971751256473808TC0.431-0.0120.0026.60E-09rs24577451.71E+08AG0.272-0.0130.0027.40E-09rs6962772799081730AG0.8460.0160.0037.80E-09rs124812822044761377GC0.722-0.0130.0027.80E-09rs353433441918471610CA0.7330.0130.0028.80E-09rs65624741367332812CG0.6510.0120.0021.00E-08rs775758377582005AT0.4330.0120.0021.10E-08rs2062882891839576GA0.587-0.0120.0021.10E-08rs7519626199514554CT0.3240.0120.0021.20E-08rs943534011.08E+08TA0.3440.0120.0021.20E-08rs348660951116377356AG0.686-0.0120.0021.20E-08rs3488092059032097AG0.348-0.0120.0021.30E-08rs10508471687443734CT0.4260.0110.0021.40E-08rs7322054431.31E+08AC0.842-0.0160.0031.50E-08rs4571506587756918CT0.5400.0110.0021.50E-08rs7320831737834367GA0.3330.0120.0021.50E-08rs6741228222548774TC0.4330.0110.0021.60E-08rs4949465132178489TC0.870-0.0170.0031.70E-08rs6217597221.61E+08TC0.9660.0310.0061.70E-08rs1362332231212410AG0.809-0.0140.0031.80E-08rs128316171284758368CT0.764-0.0130.0021.90E-08rs1186121416746611GT0.7840.0140.0022.00E-08rs1091870111.62E+08GA0.3720.0120.0022.10E-08rs108239681074738269AT0.6330.0120.0022.10E-08rs740869111250015942GA0.9250.0210.0042.10E-08rs473192571.33E+08CT0.316-0.0120.0022.60E-08rs284853051574044197CT0.6310.0120.0022.60E-08rs119323717526486GC0.439-0.0110.0022.80E-08rs609524281675640521TC0.9090.0190.0033.00E-08rs99042881747031973TC0.7080.0120.0023.10E-08rs129678551835138245AG0.3310.0120.0023.10E-08rs2254710637477000CA0.2360.0130.0023.50E-08rs72674867895578201AT0.7650.0130.0023.80E-08rs1931263196175101GT0.510-0.0110.0024.00E-08rs576115031631165795GA0.4850.0110.0024.00E-08rs61796681423678196AT0.912-0.0190.0044.20E-08rs695789671.32E+08CT0.503-0.0110.0024.50E-08rs2080870560388313AT0.2580.0120.0024.90E-08Note. CHR = chromosome, BP = base position, EA = effect allele, Non EA = non-effect allele, EAF = effect allele frequency. Table S2. Two-sample Mendelian randomisation analyses of the effect of lifetime smoking on coronary artery disease, lung cancer and AHRR Methylation.OutcomeMethodOR (95% CI)P-valueCoronary Artery DiseaseInverse-Variance Weighted1.56 (1.34, 1.82)1.19 × 10-08MR Egger (SIMEX)0.76 (0.44, 1.34)0.35Weighted Median1.65 (1.36,?2.00)5.40 × 10-07Weighted Mode1.79 (1.06,?3.03)0.03MR RAPS1.63 (1.40, 1.90)5.65 × 10-10Lung CancerInverse-Variance Weighted4.21 (2.98,?5.96)3.49 × 10-16MR Egger (SIMEX)16.64 (3.88, 71.42)9.61 × 10-05Weighted Median2.77 (1.91,?4.03)8.88 × 10-08Weighted Mode6.19 (2.07, 18.54)0.001MR RAPS3.71 (2.75, 5.00)8.65 × 10-18Beta (95% CI)P-valueAHRR MethylationInverse-Variance Weighted-0.098 (-0.168, -0.028)0.006MR Egger (SIMEX)-0.176 (-0.443, 0.102)0.217Weighted Median-0.125 (-0.228, -0.021)0.02Weighted Mode-0.207 (-0.511, 0.097)0.18MR RAPS-0.095 (-0.171, -0.013)0.01Of the 126 genome-wide significant SNPs associated with the lifetime smoking index, 126 were available from the GWAS of coronary artery disease (Nikpay et al., 2015), 120 from the GWAS of lung cancer (Wang et al., 2014) and 119 from our GWAS of AHRR locus methylation. SIMEX-corrected estimates are unweighted. Due to low regression dilution I2GX (see Supplementary Table S3), MR Egger SIMEX estimates should be interpreted with caution and MR Egger estimates could not be calculated. SIMEX = simulation extrapolation, MR RAPS = robust adjusted profile score. Table S3. Tests of the unweighted and weighted regression dilution I2GX.I2GX UnweightedI2GX WeightedmFLifetime smoking > CAD0.6440.45444.05Lifetime smoking > Lung cancer0.6430.34044.36Lifetime smoking > AHRR methylation0.6350.45544.08Lifetime smoking > Schizophrenia (2014)0.6340.40544.08Lifetime smoking > Schizophrenia (Steiger filtered)0.6670.44144.25Lifetime smoking > Schizophrenia (2018)0.6420.41744.27Lifetime smoking > Schizophrenia (CHRNA5 removed)0.590.2543.05Lifetime smoking > Depression (2018)0.6440.42944.05Lifetime smoking > Depression (2018) (Steiger filtered)0.6460.43644.22Lifetime smoking > Depression (2013)0041.06Lifetime smoking > Depression (2019)0.6170.36343.46Schizophrenia (2014) > Lifetime smoking0.429037.90Schizophrenia (2018) > Lifetime smoking0.487040.86Schizophrenia (CHRNA3 removed) > Lifetime smoking0.43037.78Depression (2018) > Lifetime smoking0036.72Depression (2018) > Lifetime smoking (Steiger filtered)0036.49Depression (2013) > Lifetime smoking0.358019.05Depression (2019) > Lifetime smoking0.3220.11128.79Smoking initiation > Schizophrenia0.6030.40144.98Smoking initiation > Depression (2018)0.6130.56144.93 Schizophrenia > Smoking initiation0.4220.76237.62Depression (2018) > Smoking initiation0036.83CAD: coronary artery disease. Unweighted estimates only take into account dilution in the SNP-exposure effects, whereas weighted estimates account for the SE of the SNP-outcome effects ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"Efj9w6vJ","properties":{"formattedCitation":"(Bowden, Del Greco M, et al., 2016)","plainCitation":"(Bowden, Del Greco M, et al., 2016)","noteIndex":0},"citationItems":[{"id":1751,"uris":[""],"uri":[""],"itemData":{"id":1751,"type":"article-journal","title":"Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I 2 statistic","container-title":"International journal of epidemiology","page":"1961–1974","volume":"45","issue":"6","source":"Google Scholar","title-short":"Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression","author":[{"family":"Bowden","given":"Jack"},{"family":"Del Greco M","given":"Fabiola"},{"family":"Minelli","given":"Cosetta"},{"family":"Davey Smith","given":"George"},{"family":"Sheehan","given":"Nuala A."},{"family":"Thompson","given":"John R."}],"issued":{"date-parts":[["2016"]]}}}],"schema":""} (Bowden, Del Greco M, et al., 2016). The unweighted I2 estimates were larger for both positive control outcomes so in Table 1 (main text), unweighted MR Egger SIMEX estimates are presented. For the main analysis unweighted MR Egger SIMEX corrections are presented unless I2 estimates < 0.6 which is too low to conduct either MR Egger analysis. All estimates show evidence of high dilution in the SNP-exposure effects, so MR Egger estimates should be interpreted with caution. Table S4. Tests of Heterogeneity in the SNP-exposure associationMethodQdfP-valueLifetime smoking > CADInverse-variance weighted184.72125<0.001MR Egger175.121240.002Q’9.6010.002Lifetime smoking > Lung cancerInverse-variance weighted235.82119<0.001MR Egger228.96118<0.001Q’6.8710.009Lifetime smoking > AHRR methylationInverse-variance weighted122.071180.38MR Egger121.351170.37Q’0.7210.40Lifetime smoking > SchizophreniaInverse-variance weighted573.03124<0.001MR Egger571.23123<0.001Q’1.8110.18Lifetime smoking > Schizophrenia (2018)Inverse-variance weighted592.94121<0.001MR Egger591.77120<0.001Q’1.1710.28Lifetime smoking > Schizophrenia (CHRNA5 variant removed)Inverse-variance weighted566.23123<0.001MR Egger565.96122<0.001Q’0.2710.60Lifetime smoking > Depression (2018)Inverse-variance weighted259.98125<0.001MR Egger256.69124<0.001Q’3.2910.07Lifetime smoking > Depression (2013)Inverse-variance weighted33.014330.46MR Egger33.07320.41Q’0.0810.78Lifetime smoking > Depression (2019)Inverse-variance weighted497.29124<0.001MR Egger488.88123<0.001Q’8.4110.003Schizophrenia > Lifetime smokingInverse-variance weighted773.92101<0.001MR Egger770.30100<0.001Q’3.6110.05Schizophrenia (2018) > Lifetime smokingInverse-variance weighted869.67135<0.001MR Egger864.25134<0.001Q’5.4210.20Schizophrenia (CHRNA3 variant removed) > Lifetime smokingInverse-variance weighted628.72100<0.001MR Egger622.4999<0.001Q’6.2310.01Depression (2018) > Lifetime smokingInverse-variance weighted255.6133<0.001MR Egger250.2932<0.001Q’5.3210.02Depression (2013) > Lifetime smokingInverse-variance weighted59.75360.008MR Egger59.48350.006Q’0.2710.60Depression (2019) > Lifetime smokingInverse-variance weighted481.9794<0.001MR Egger481.7793<0.001Q’0.2010.66Smoking Initiation > SchizophreniaInverse-variance weighted1468.61370<0.001MR Egger1467.30369<0.001Q’1.3210.25Smoking Initiation > Depression (2018)Inverse-variance weighted704.94369<0.001MR Egger699.86368<0.001Q’5.0710.02Schizophrenia > Smoking initiationInverse-variance weighted409.42106<0.001MR Egger402.35105<0.001Q’7.061<0.001Depression (2018) > Smoking initiationInverse-variance weighted153.6133<0.001MR Egger150.9532<0.001Q’2.6610.10Note: df = degrees of freedom where degrees of freedom is equal to the number of SNPs -1. Q = Rucker’s Q ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"xvSqh1eC","properties":{"formattedCitation":"(Bowden et al., 2017)","plainCitation":"(Bowden et al., 2017)","noteIndex":0},"citationItems":[{"id":2262,"uris":[""],"uri":[""],"itemData":{"id":2262,"type":"article-journal","title":"A framework for the investigation of pleiotropy in two-sample summary data Mendelian randomization","container-title":"Statistics in medicine","page":"1783–1802","volume":"36","issue":"11","source":"Google Scholar","author":[{"family":"Bowden","given":"Jack"},{"family":"Del Greco M","given":"Fabiola"},{"family":"Minelli","given":"Cosetta"},{"family":"Davey Smith","given":"George"},{"family":"Sheehan","given":"Nuala"},{"family":"Thompson","given":"John"}],"issued":{"date-parts":[["2017"]]}}}],"schema":""} (Bowden et al., 2017), a test of heterogeneity or dispersion in the SNP-exposure effects. The Q’ indicates the extent to which MR Egger is a better fit than the inverse-variance weighted method. Table S5. MR Egger test of directional pleiotropyOutcomeIntercept (95% CI)P-valueLifetime smoking > CAD0.012 (0.003, 0.022)0.010Lifetime smoking > Lung cancer-0.021 (-0.042, -0.001)0.062Lifetime smoking > AHRR methylation-0.001 (-0.004, 0.002)0.406Lifetime smoking > Schizophrenia (2014)-0.006 (-0.025, 0.013)0.531Lifetime smoking > Schizophrenia (2018)-0.004 (-0.022, 0.013)0.627Lifetime smoking > Schizophrenia (CHRNA5 removed)-0.002 (-0.022, 0.017)0.808Lifetime smoking > Depression (2018)0.006 (-0.003, 0.015)0.210Lifetime smoking > Depression (2013)0.009 (-0.055, 0.073)0.787Lifetime smoking > Depression (2019)0.005 (-0.002, 0.012)0.148Schizophrenia (2014) > Lifetime smoking0.002 (-0.004, 0.007)0.495Schizophrenia (2018) > Lifetime smoking0.002 (-0.002, 0.006)0.360Schizophrenia (CHRNA3 removed) > Lifetime smoking0.002 (-0.002, 0.007)0.32Depression (2018) > Lifetime smoking0.006 (-0.008, 0.019)0.416Depression (2013) > Lifetime smoking-0.001 (-0.004, 0.003)0.693Depression (2019) > Lifetime smoking0.0004 (-0.004, 0.005)0.845Smoking initiation > Schizophrenia-0.003 (-0.013, 0.007)0.565Smoking initiation > Depression (2018)0.004 (-0,001, 0.010)0.103Schizophrenia > Smoking initiation0.002 (-0.001, 0.006)0.177Depression (2018) > Smoking initiation0.003 (-0.012, 0.006)0.458CAD: coronary artery disease. Table S6. Bi-directional two-sample Mendelian randomisation of lifetime smoking on schizophrenia and depression following Steiger filtering. ExposureOutcomeMethodN SNPOR (95% CI)p-valueSmokingSchizophreniaInverse-Variance Weighted105/125 (84%)1.73 (1.39, 2.16)9.12 x 10-07MR Egger (SIMEX)6.19 (2.75, 13.95)2.64 x 10-05Weighted median1.72 (1.34, 2.22)2.78 x 10-05Weighted mode1.84 (0.73, 4.65)0.20MR RAPS1.83 (1.45, 2.32)3.97 x 10-07SmokingDepressionInverse-Variance Weighted124/126 (98%)1.94 (1.67, 2.25)3.46 x 10-18MR Egger (SIMEX)1.17 (0.67, 2.04)0.59Weighted median1.95 (1.65, 2.31)4.70 x 10-15Weighted mode1.81 (1.17, 2.79)0.008MR RAPS1.95 (1.68, 2.27)3.81 x 10-18ExposureOutcomeMethodN SNPBeta (95% CI)p-valueSchizophreniaSmoking-102/102 (100%)--DepressionSmokingInverse-Variance Weighted32/34 (94%)0.063 (0.007, 0.120)0.028MR Egger (SIMEX)--Weighted median0.084 (0.042, 0.126)8.64 x 10-05Weighted mode0.111 (0.034, 0.187)0.008MR RAPS0.065 (0.008, 0.123)0.026Note: For each SNP in the instrument, Steiger filtering calculates how much of the variance the SNP explains in the exposure and how much it explains in the outcome. The number of SNPs which explain more variance in the exposure are presented in the N SNP column. Analysis is then repeated using only these SNPs to ensure that results are not due to reverse causation. For the effect of schizophrenia on lifetime smoking, all SNPs comprising the instrument for schizophrenia were better instruments for schizophrenia than lifetime smoking, therefore, that analysis was not repeated. All SIMEX corrections are unweighted due to greater unweighted I2GX (see Supplementary Table S2). Given low I2GX, all MR Egger results should be interpreted with caution. Table S7. Bi-directional two-sample Mendelian randomisation analyses of the effect of lifetime smoking on depression (2013). ExposureOutcomeMethodOR (95% CI)p-valueLifetime SmokingDepressionInverse-Variance Weighted3.66 (2.08, 6.43)6.46 x 10-06Weighted median3.69 (1.72, 7.94)0.001Weighted mode3.18 (0.63, 16.06)0.17MR RAPS4.23 (2.33, 7.67)2.04 x 10-06ExposureOutcomeMethodBeta (95% CI)p-valueDepressionLifetime SmokingInverse-Variance Weighted0.002 (-0.006, 0.011)0.572Weighted median0.003 (-0.007, 0.012)0.602Weighted mode0.002 (-0.016, 0.019)0.844MR RAPS0.003 (-0.005, 0.012)0.435Note: When depression is the exposure, a relaxed p-value threshold of p<5x10-5 was used because there were no SNPs associated at the genome wide level of significance. The direction of effect is consistent with the more recent GWAS for MDD despite sample overlap. There is weaker statistical evidence, possibly due to reduced sample size (N = 18?759). Both MR Egger and MR Egger (SIMEX) estimates could not be conducted due to low regression dilution I2GX (see Table S2).Table S8. Bidirectional two-sample Mendelian randomisation analyses of the effect of lifetime smoking (from a sensitivity GWAS without chip as a covariate) on schizophrenia and major depression.ExposureOutcomeMethodN SNPOR (95% CI)P-valueLifetime SmokingSchizophreniaInverse-Variance Weighted1372.23 (1.69, 2.95)2.03 x 10-08MR Egger (SIMEX)1373.71 (1.47, 3.39)0.006Weighted median1372.07 (1.61, 2.66)1.24 x 10-08Weighted mode1371.56 (0.63, 3.86)0.34MR RAPS1372.42 (1.86, 3.15)6.09 x 10-11Lifetime SmokingDepressionInverse-Variance Weighted901.88 (1.57, 2.24)5.26 x 10-12MR Egger (SIMEX)904.39 (0.60, 323.15)0.51Weighted median901.70 (1.39, 2.07)1.53 x 10-07Weighted mode901.49 (0.99, 2.26)0.06MR RAPS901.85 (1.55, 2.21)9.00 x 10-12Beta (95% CI)P-valueSchizophreniaLifetime SmokingInverse-Variance Weighted1020.020 (0.003, 0.037)0.02MR Egger (SIMEX)102--Weighted median1020.008 (-0.004, 0.020)0.19Weighted Mode1020.004 (-0.027, 0.035)0.80MR RAPS1020.014 (-0.001, 0.030)0.06DepressionLifetime SmokingInverse-Variance Weighted360.102 (0.036, 0.168)0.002MR Egger (SIMEX)---Weighted median360.101 (0.056, 0.147)1.29 x 10-05Weighted mode360.124 (0.044, 0.204)0.004MR RAPS360.091 (0.025, 0.157)0.007Note: There were 139 genome-wide significant SNPs associated with lifetime smoking index when genotype chip was not included as a covariate in the GWAS. SIMEX-corrected estimates are unweighted. MR Egger regression was not conducted for schizophrenia or major depression as exposures because regression dilution I2GX was below 0.3 (see Supplementary Table S6). Due to low regression dilution I2GX for lifetime smoking as the exposure (see Supplementary Table S6), MR Egger and MR Egger SIMEX estimates should be interpreted with caution. SIMEX = simulation extrapolation, MR RAPS = robust adjusted profile score.Table S9. Bi-directional two-sample Mendelian randomisation analyses of the effect of lifetime smoking on schizophrenia (2018). ExposureOutcomeMethodOR (95% CI)p-valueLifetime SmokingSchizophreniaInverse-Variance Weighted2.64 (1.99, 3.52)2.36 x 10-11MR Egger (SIMEX)4.03 (1.38, 11.72)0.01Weighted median2.23 (1.76, 2.82)4.02 x 10-11Weighted mode2.20 (1.11, 4.36)0.02MR RAPS2.74 (2.09, 3.58)2.55 x 10-13ExposureOutcomeMethodBeta (95% CI)p-valueSchizophreniaLifetime SmokingInverse-Variance Weighted0.025 (0.011, 0.038)0.0003MR Egger (SIMEX)--Weighted median0.016 (0.005, 0.026)0.004Weighted mode0.029 (-0.011, 0.068)0.159MR RAPS0.017 (0.004, 0.03)0.009Note: MR Egger estimates could not be conducted due to low regression dilution I2GX (see Table S2).Table S10. Bi-directional two-sample Mendelian randomisation analyses of the effect of lifetime smoking on schizophrenia with the CHRNA5-A3-B4 Variants removed ExposureOutcomeMethodN SNPOR (95% CI)p-valueLifetime SmokingSchizophreniaInverse-Variance Weighted1242.19 (1.61, 2.99)6.32 x 10-07Weighted median1242.02 (1.57, 2.61)6.23 x 10-08Weighted mode1241.67 (0.85, 3.28)0.14MR RAPS1242.36 (1.77, 3.15)5.53 x 10-09ExposureOutcomeMethodBeta (95% CI)p-valueSchizophreniaLifetime SmokingInverse-Variance Weighted1010.018 (0.003, 0.032)0.02Weighted median1010.015 (0.003, 0.026)0.01Weighted mode1010.016 (-0.012, 0.044)0.28MR RAPS1010.016 (0.003, 0.030)0.02Note: This analysis was run using Ripke et al ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a1si5f15hjq","properties":{"formattedCitation":"(2014)","plainCitation":"(2014)","noteIndex":0},"citationItems":[{"id":663,"uris":[""],"uri":[""],"itemData":{"id":663,"type":"article-journal","title":"Biological insights from 108 schizophrenia-associated genetic loci","container-title":"Nature","page":"421-427","volume":"511","issue":"7510","source":"Google Scholar","author":[{"family":"Ripke","given":"Stephan"},{"family":"Neale","given":"Benjamin M."},{"family":"Corvin","given":"Aiden"},{"family":"Walters","given":"James TR"},{"family":"Farh","given":"Kai-How"},{"family":"Holmans","given":"Peter A."},{"family":"Lee","given":"Phil"},{"family":"Bulik-Sullivan","given":"Brendan"},{"family":"Collier","given":"David A."},{"family":"Huang","given":"Hailiang"},{"literal":"others"}],"issued":{"date-parts":[["2014"]]}},"suppress-author":true}],"schema":""} (2014) GWAS of schizophrenia. When lifetime smoking was the exposure, rs8042849 (located in the CHRNA5 gene was removed). This variant is in high LD (r2 = 0.83) with rs16969968, the variant previously associated with smoking heaviness ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a1bnlub96gv","properties":{"formattedCitation":"(Furberg et al., 2010)","plainCitation":"(Furberg et al., 2010)","noteIndex":0},"citationItems":[{"id":1907,"uris":[""],"uri":[""],"itemData":{"id":1907,"type":"article-journal","title":"Genome-wide meta-analyses identify multiple loci associated with smoking behavior","container-title":"Nature genetics","page":"441","volume":"42","issue":"5","source":"Google Scholar","author":[{"family":"Furberg","given":"Helena"},{"family":"Kim","given":"YunJung"},{"family":"Dackor","given":"Jennifer"},{"family":"Boerwinkle","given":"Eric"},{"family":"Franceschini","given":"Nora"},{"family":"Ardissino","given":"Diego"},{"family":"Bernardinelli","given":"Luisa"},{"family":"Mannucci","given":"Pier M."},{"family":"Mauri","given":"Francesco"},{"family":"Merlini","given":"Piera A."}],"issued":{"date-parts":[["2010"]]}}}],"schema":""} (Furberg et al., 2010). When schizophrenia was the exposure, we removed rs8042374. This is located in the gene CHRNA3, part of the CHRNA5-A3-B4 gene complex which has been shown to affect nicotine intake and consequently smoking heaviness ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"a1eqiufl7f9","properties":{"formattedCitation":"(Furberg et al., 2010)","plainCitation":"(Furberg et al., 2010)","noteIndex":0},"citationItems":[{"id":1907,"uris":[""],"uri":[""],"itemData":{"id":1907,"type":"article-journal","title":"Genome-wide meta-analyses identify multiple loci associated with smoking behavior","container-title":"Nature genetics","page":"441","volume":"42","issue":"5","source":"Google Scholar","author":[{"family":"Furberg","given":"Helena"},{"family":"Kim","given":"YunJung"},{"family":"Dackor","given":"Jennifer"},{"family":"Boerwinkle","given":"Eric"},{"family":"Franceschini","given":"Nora"},{"family":"Ardissino","given":"Diego"},{"family":"Bernardinelli","given":"Luisa"},{"family":"Mannucci","given":"Pier M."},{"family":"Mauri","given":"Francesco"},{"family":"Merlini","given":"Piera A."}],"issued":{"date-parts":[["2010"]]}}}],"schema":""} (Furberg et al., 2010). However, this particular variant is in low LD with rs16969968 (r2 = 0.18). Results were highly consistent with those including these two variants. As the I2GX statistic was low (indiciative of large regression dilution bias), MR Egger results were not reliable and so are not reported (see Table S2).Table S11. Bi-directional two-sample Mendelian randomisation analyses of the effect of lifetime smoking on depression (Howard et al., 2019 with UKBB and 23andMe removed)ExposureOutcomeMethodN SNPOR (95% CI)p-valueLifetime SmokingDepression (2019)Inverse-Variance Weighted1251.63 (1.46, 1.83)6.23 x 10-17MR Egger (SIMEX)1250.89 (0.58, 1.37)0.59Weighted median1251.59 (1.43, 1.75)3.06 x 10-19Weighted mode1251.57 (1.15, 2.16)0.006MR RAPS1251.61 (1.44, 1.81)9.14 x 10-17ExposureOutcomeMethodN SNPBeta (95% CI)p-valueDepression (2019)Lifetime SmokingInverse-Variance Weighted950.165 (0.126, 0.204)2.03 x 10-16MR Egger (SIMEX)---Weighted median950.131 (0.099, 0.164)1.69 x 10-15Weighted mode950.114 (0.035, 0.194)0.006MR RAPS950.171 (0.133, 0.201)7.13 x 10-19Note: This analysis has been conducted using the most recent PGC GWAS of depression ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"ajr1j425id","properties":{"formattedCitation":"(Howard et al., 2019)","plainCitation":"(Howard et al., 2019)","noteIndex":0},"citationItems":[{"id":2574,"uris":[""],"uri":[""],"itemData":{"id":2574,"type":"article-journal","title":"Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions","container-title":"Nature Neuroscience","page":"343","volume":"22","issue":"3","source":"","abstract":"The authors conducted a genetic meta-analysis of depression and found 269 associated genes. These genes highlight several potential drug repositioning opportunities, and relationships with depression were found for neuroticism and smoking.","DOI":"10.1038/s41593-018-0326-7","ISSN":"1546-1726","language":"En","author":[{"family":"Howard","given":"David M."},{"family":"Adams","given":"Mark J."},{"family":"Clarke","given":"Toni-Kim"},{"family":"Hafferty","given":"Jonathan D."},{"family":"Gibson","given":"Jude"},{"family":"Shirali","given":"Masoud"},{"family":"Coleman","given":"Jonathan R. I."},{"family":"Hagenaars","given":"Saskia P."},{"family":"Ward","given":"Joey"},{"family":"Wigmore","given":"Eleanor M."},{"family":"Alloza","given":"Clara"},{"family":"Shen","given":"Xueyi"},{"family":"Barbu","given":"Miruna C."},{"family":"Xu","given":"Eileen Y."},{"family":"Whalley","given":"Heather C."},{"family":"Marioni","given":"Riccardo E."},{"family":"Porteous","given":"David J."},{"family":"Davies","given":"Gail"},{"family":"Deary","given":"Ian J."},{"family":"Hemani","given":"Gibran"},{"family":"Berger","given":"Klaus"},{"family":"Teismann","given":"Henning"},{"family":"Rawal","given":"Rajesh"},{"family":"Arolt","given":"Volker"},{"family":"Baune","given":"Bernhard T."},{"family":"Dannlowski","given":"Udo"},{"family":"Domschke","given":"Katharina"},{"family":"Tian","given":"Chao"},{"family":"Hinds","given":"David A."},{"family":"Trzaskowski","given":"Maciej"},{"family":"Byrne","given":"Enda M."},{"family":"Ripke","given":"Stephan"},{"family":"Smith","given":"Daniel J."},{"family":"Sullivan","given":"Patrick F."},{"family":"Wray","given":"Naomi R."},{"family":"Breen","given":"Gerome"},{"family":"Lewis","given":"Cathryn M."},{"family":"McIntosh","given":"Andrew M."}],"issued":{"date-parts":[["2019",3]]}}}],"schema":""} (Howard et al., 2019) with the UK Biobank and 23andMe samples removed. This is not the primary analysis given the less stringent definition of depression. However, the results are relatively consistent. Effects are smaller when depression is the outcome and slightly larger when depression is the exposure. SIMEX correction for lifetime smoking as the exposure was unweighted due to greater unweighted I2GX (see Supplementary Table S2). Given low I2GX, all MR Egger SIMEX results should be interpreted with caution. Table S12. A comparison of each of the MR sensitivity analyses and their assumptions. MethodDescriptionAdditional AssumptionsPowerInvalid variants allowedIV2IV3Random effects Inverse‐variance weighted (IVW) A meta-analysis of the Wald ratios for each SNP (ZYZX) weighted by the inverse of the variance of the SNP-outcome association. Any horizontal pleiotropy must be balancedHas the most power if the assumptions are satisfied. 0% (or 100% if all horizontal pleiotropy is balanced)??MR‐Egger regression ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"HSEBiQVm","properties":{"formattedCitation":"(Bowden et al., 2015)","plainCitation":"(Bowden et al., 2015)","noteIndex":0},"citationItems":[{"id":1031,"uris":[""],"uri":[""],"itemData":{"id":1031,"type":"article-journal","title":"Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression","container-title":"International journal of epidemiology","page":"512–525","volume":"44","issue":"2","source":"Google Scholar","title-short":"Mendelian randomization with invalid instruments","author":[{"family":"Bowden","given":"Jack"},{"family":"Davey Smith","given":"George"},{"family":"Burgess","given":"Stephen"}],"issued":{"date-parts":[["2015"]]}}}],"schema":""} (Bowden et al., 2015)An extension of the IVW which relaxes the assumption that any pleiotropy must be balanced. A significant intercept term suggests bias from directional pleiotropy, i.e. the average pleiotropic effect is not zero. MR-Egger regression provides consistent estimates even if all genetic instrumental variables are invalid as long as the INSIDE assumption is met. InSIDEaNOMEbHas the lowest power100%??Weighted median ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"xG4FIl9U","properties":{"formattedCitation":"(Bowden, Davey Smith, et al., 2016)","plainCitation":"(Bowden, Davey Smith, et al., 2016)","noteIndex":0},"citationItems":[{"id":1033,"uris":[""],"uri":[""],"itemData":{"id":1033,"type":"article-journal","title":"Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator","container-title":"Genetic epidemiology","page":"304–314","volume":"40","issue":"4","source":"Google Scholar","author":[{"family":"Bowden","given":"Jack"},{"family":"Davey Smith","given":"George"},{"family":"Haycock","given":"Philip C."},{"family":"Burgess","given":"Stephen"}],"issued":{"date-parts":[["2016"]]}}}],"schema":""} (Bowden, Davey Smith, et al., 2016)The weighted median estimate is obtained by first calculating the Wald ratio causal estimate for each SNP and then taking the estimate with the median inverse variance weight. Consistent when 50% of weight contributed by genetic variants is valid.Similar to that of IVW method.50%??Weighted mode ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"rK05AESn","properties":{"formattedCitation":"(Hartwig et al., 2017)","plainCitation":"(Hartwig et al., 2017)","noteIndex":0},"citationItems":[{"id":1783,"uris":[""],"uri":[""],"itemData":{"id":1783,"type":"article-journal","title":"Robust inference in two-sample Mendelian randomisation via the zero modal pleiotropy assumption","container-title":"bioRxiv","page":"126102","source":"Google Scholar","author":[{"family":"Hartwig","given":"Fernando Pires"},{"family":"Smith","given":"George Davey"},{"family":"Bowden","given":"Jack"}],"issued":{"date-parts":[["2017"]]}}}],"schema":""} (Hartwig et al., 2017)Finds the largest cluster of Wald ratio estimates. The majority of the genetic instruments can be invalid providing the ZEMPA assumption is satisfied. In the weighted mode method, the mode is calculated using the inverse variance weights of the Wald ratios. ZEMPAcLess powerful than IVW and weighted median.50%??Robust adjusted profile score (RAPS) ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"GcS0Td0m","properties":{"formattedCitation":"(Zhao et al., 2018)","plainCitation":"(Zhao et al., 2018)","noteIndex":0},"citationItems":[{"id":2085,"uris":[""],"uri":[""],"itemData":{"id":2085,"type":"article-journal","title":"Statistical inference in two-sample summary-data Mendelian randomization using robust adjusted profile score","container-title":"arXiv","source":"","abstract":"Mendelian randomization (MR) is a method of exploiting genetic variation to unbiasedly estimate a causal effect in presence of unmeasured confounding. MR is being widely used in epidemiology and other related areas of population science. In this paper, we study statistical inference in the increasingly popular two-sample summary-data MR design. We show a linear model for the observed associations approximately holds in a wide variety of settings when all the genetic variants satisfy the exclusion restriction assumption, or in genetic terms, when there is no pleiotropy. In this scenario, we derive a maximum profile likelihood estimator with provable consistency and asymptotic normality. However, through analyzing real datasets, we find strong evidence of both systematic and idiosyncratic pleiotropy in MR, echoing some recent discoveries in statistical genetics. We model the systematic pleiotropy by a random effects model, where no genetic variant satisfies the exclusion restriction condition exactly. In this case we propose a consistent and asymptotically normal estimator by adjusting the profile score. We then tackle the idiosyncratic pleiotropy by robustifying the adjusted profile score. We demonstrate the robustness and efficiency of the proposed methods using several simulated and real datasets.","URL":"","note":"arXiv: 1801.09652","author":[{"family":"Zhao","given":"Qingyuan"},{"family":"Wang","given":"Jingshu"},{"family":"Hemani","given":"Gibran"},{"family":"Bowden","given":"Jack"},{"family":"Small","given":"Dylan S."}],"issued":{"date-parts":[["2018",1,29]]},"accessed":{"date-parts":[["2018",6,26]]}}}],"schema":""} (Zhao et al., 2018)An extension of IVW into a general framework which allows very many weak instruments. Requires no sample overlap in the exposure and outcome SNP effect estimates. InSIDEaPleiotropy is additive. Pleiotropic effects are balanced (have mean zero). 100%??Note. Where Z = the genetic instrument, X = the exposure and Y = the outcome. IV2 = assumption 2, that all instruments (Z) must not be associated with confounders. IV3 = assumption 3, that all instruments (Z) must only be associated with the outcome (Y) through the exposure (X). These two columns have a cross if that method requires the assumption to be met and a tick if that assumption can be relaxed. Throughout the table, invalid refers to instruments that do not meet the required assumptions for MR. Power of the methods might differ under different models of pleiotropy. aThe InSIDE assumption = pleiotropic effects of Z are independent of the effects of Z on the exposure. bThe no measurement error assumption (NOME) = assumes that the ZX associations are known, rather than estimated. cThe ZEMPA assumption = the largest subset of genetic instrumental variables with the same ratio estimate will contain the valid instruments. References ADDIN ZOTERO_BIBL {"uncited":[],"omitted":[],"custom":[]} CSL_BIBLIOGRAPHY Bowden, J., Davey Smith, G., & Burgess, S. (2015). Mendelian randomization with invalid instruments: Effect estimation and bias detection through Egger regression. International Journal of Epidemiology, 44(2), 512–525.Bowden, J., Davey Smith, G., Haycock, P. C., & Burgess, S. (2016). Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genetic Epidemiology, 40(4), 304–314.Bowden, J., Del Greco M, F., Minelli, C., Davey Smith, G., Sheehan, N. A., & Thompson, J. R. (2016). Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: The role of the I 2 statistic. International Journal of Epidemiology, 45(6), 1961–1974.Bowden, J., Del Greco M, F., Minelli, C., Davey Smith, G., Sheehan, N., & Thompson, J. (2017). A framework for the investigation of pleiotropy in two-sample summary data Mendelian randomization. Statistics in Medicine, 36(11), 1783–1802.Boyd, A., Golding, J., Macleod, J., Lawlor, D. A., Fraser, A., Henderson, J., … Davey Smith, G. (2013). Cohort profile: The ‘children of the 90s’—the index offspring of the Avon Longitudinal Study of Parents and Children. International Journal of Epidemiology, 42(1), 111–127.Bulik-Sullivan, B. K., Loh, P.-R., Finucane, H. K., Ripke, S., Yang, J., Schizophrenia Working Group of the Psychiatric Genomics Consortium, … Neale, B. M. (2015). LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nature Genetics, 47(3), 291–295. , C., Freeman, C., Petkova, D., Band, G., Elliott, L. T., Sharp, K., … Marchini, J. (2017). Genome-wide genetic data on ~500,000 UK Biobank participants. BioRxiv, 166298. Consortium, & others. (2015). A comprehensive 1000 Genomes-based genome-wide association meta-analysis of coronary artery disease. Nature Genetics, 47(10), 1121–1149.Fraser, A., Macdonald-Wallis, C., Tilling, K., Boyd, A., Golding, J., Davey Smith, G., … others. (2012). Cohort profile: The Avon Longitudinal Study of Parents and Children: ALSPAC mothers cohort. International Journal of Epidemiology, 42(1), 97–110.Furberg, H., Kim, Y., Dackor, J., Boerwinkle, E., Franceschini, N., Ardissino, D., … Merlini, P. A. (2010). Genome-wide meta-analyses identify multiple loci associated with smoking behavior. Nature Genetics, 42(5), 441.Gaunt, T. R., Shihab, H. A., Hemani, G., Min, J. L., Woodward, G., Lyttleton, O., … Ho, K. (2016). Systematic identification of genetic influences on methylation across the human life course. Genome Biology, 17(1), 61.Hartwig, F. P., Smith, G. D., & Bowden, J. (2017). Robust inference in two-sample Mendelian randomisation via the zero modal pleiotropy assumption. BioRxiv, 126102.Hemani, G., Zheng, J., Elsworth, B., Wade, K. H., Haberland, V., Baird, D., … Langdon, R. (2018). The MR-Base platform supports systematic causal inference across the human phenome. ELife, 7, e34408.Houseman, E. A., Accomando, W. P., Koestler, D. C., Christensen, B. C., Marsit, C. J., Nelson, H. H., … Kelsey, K. T. (2012). DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics, 13(1), 86.Howard, D. M., Adams, M. J., Clarke, T.-K., Hafferty, J. D., Gibson, J., Shirali, M., … McIntosh, A. M. (2019). Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions. Nature Neuroscience, 22(3), 343. , D. A., Tilling, K., & Davey Smith, G. (2016). Triangulation in aetiological epidemiology. International Journal of Epidemiology, 45(6), 1866–1886.Leffondré, K., Abrahamowicz, M., Siemiatycki, J., & Rachet, B. (2002). Modeling smoking history: A comparison of different approaches. American Journal of Epidemiology, 156(9), 813–823.Lewer, D., McKee, M., Gasparrini, A., Reeves, A., & de Oliveira, C. (2017). Socioeconomic position and mortality risk of smoking: Evidence from the English Longitudinal Study of Ageing (ELSA). The European Journal of Public Health, 27(6), 1068–1073.Liu, M., Jiang, Y., Wedow, R., Li, Y., Brazel, D. M., Chen, F., … Tian, C. (2019). Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use. Nature Genetics, 1.Mitchell, R., Hemani, G., Dudding, T., & Paternoster, L. (2017, November 6). UK Biobank Genetic Data: MRC-IEU Quality Control, Version 1. ò, M. R., Timofeeva, M. N., Morris, R. W., Prieto-Merino, D., Sattar, N., Brennan, P., … Davey Smith, G. (2012). Association Between Genetic Variants on Chromosome 15q25 Locus and Objective Measures of Tobacco Exposure. JNCI: Journal of the National Cancer Institute, 104(10), 740–748. , R., Wong, C. C., Volta, M., Lunnon, K., Mill, J., & Schalkwyk, L. C. (2013). A data-driven approach to preprocessing Illumina 450K methylation array data. BMC Genomics, 14(1), 293.Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M. A., Bender, D., … others. (2007). PLINK: A tool set for whole-genome association and population-based linkage analyses. The American Journal of Human Genetics, 81(3), 559–575.R. Core Team. (2014). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2013. ISBN 3-900051-07-0.Relton, C. L., Gaunt, T., McArdle, W., Ho, K., Duggirala, A., Shihab, H., … Reik, W. (2015). Data resource profile: Accessible resource for integrated epigenomic studies (ARIES). International Journal of Epidemiology, 44(4), 1181–1190.Ripke, S., Neale, B. M., Corvin, A., Walters, J. T., Farh, K.-H., Holmans, P. A., … others. (2014). Biological insights from 108 schizophrenia-associated genetic loci. Nature, 511(7510), 421–427.Thorgeirsson, T. E., Geller, F., Sulem, P., Rafnar, T., Wiste, A., Magnusson, K. P., … Stefansson, K. (2008). A Variant Associated with Nicotine Dependence, Lung Cancer and Peripheral Arterial Disease. Nature, 452(7187), 638–642. Consortium. (2010). Genome-wide meta-analyses identify multiple loci associated with smoking behavior. Nature Genetics, 42(5), 441–447.Touleimat, N., & Tost, J. (2012). Complete pipeline for Infinium? Human Methylation 450K BeadChip data processing using subset quantile normalization for accurate DNA methylation estimation. Epigenomics, 4(3), 325–341.Wang, Y., McKay, J. D., Rafnar, T., Wang, Z., Timofeeva, M. N., Broderick, P., … Han, Y. (2014). Rare variants of large effect in BRCA2 and CHEK2 affect risk of lung cancer. Nature Genetics, 46(7), 736.Ware, J. J., van den Bree, M. B., & Munafò, M. R. (2011). Association of the CHRNA5-A3-B4 gene cluster with heaviness of smoking: A meta-analysis. Nicotine & Tobacco Research, 13(12), 1167–1175.Wray, N. R., Sullivan, P. F., & others. (2018). Genome-wide association analyses identify 44 risk variants and refine the genetic architecture of major depression. Nature Genetics, 50, 668–681.Zhao, Q., Wang, J., Hemani, G., Bowden, J., & Small, D. S. (2018). Statistical inference in two-sample summary-data Mendelian randomization using robust adjusted profile score. ArXiv. Retrieved from ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download