Advertisement

Contribution of metabolic risk factors and lifestyle behaviors to cardiovascular disease: A mendelian randomization study

  • Author Footnotes
    1 These authors contributed equally to this article.
    Yiming Jia
    Footnotes
    1 These authors contributed equally to this article.
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Author Footnotes
    1 These authors contributed equally to this article.
    Rong Wang
    Footnotes
    1 These authors contributed equally to this article.
    Affiliations
    Department of Dermatology, The First Affiliated Hospital of Soochow University, Suzhou, China
    Search for articles by this author
  • Daoxia Guo
    Affiliations
    School of Nursing, Medical College of Soochow University, Suzhou, China

    Department of Epidemiology, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA
    Search for articles by this author
  • Lulu Sun
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Mengyao Shi
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Kaixin Zhang
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Pinni Yang
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Yuhan Zang
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Yu Wang
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Fanghua Liu
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Yonghong Zhang
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China
    Search for articles by this author
  • Zhengbao Zhu
    Correspondence
    Corresponding author. Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, 199 Renai Road, Industrial Park District, Suzhou, Jiangsu Province 215123, China.
    Affiliations
    Department of Epidemiology, School of Public Health and Jiangsu Key Laboratory of Preventive and Translational Medicine for Geriatric Diseases, Medical College of Soochow University, Suzhou, China

    Department of Epidemiology, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA
    Search for articles by this author
  • Author Footnotes
    1 These authors contributed equally to this article.

      Highlights

      • This study provided novel insight into the causal risk factors for cardiovascular diseases from a genetic standpoint.
      • This study further developed actionable prevention strategies against cardiovascular diseases.
      • Blood pressure, glucose, lipids, body mass index, smoking, alcohol intake, and education were promising preventive targets.

      Abstract

      Background and aims

      Etiologic associations between some modifiable factors (metabolic risk factors and lifestyle behaviors) and cardiovascular disease (CVD) remain unclear. To identify targets for CVD prevention, we evaluated the causal associations of these factors with coronary artery disease (CAD) and ischemic stroke using a two-sample Mendelian randomization (MR) method.

      Methods and results

      Previously published genome-wide association studies (GWASs) for blood pressure (BP), glucose, lipids, overweight, smoking, alcohol intake, sedentariness, and education were used to identify instruments for 15 modifiable factors. We extracted effects of the genetic variants used as instruments for the exposures on coronary artery disease (CAD) and ischemic stroke from large GWASs (N = 60 801 cases/123 504 controls for CAD and N = 40 585 cases/406 111 controls for ischemic stroke). Genetically predicted hypertension (CAD: OR, 5.19 [95% CI, 4.21–6.41]; ischemic stroke: OR, 4.92 [4.12–5.86]), systolic BP (CAD: OR, 1.03 [1.03–1.04]; ischemic stroke: OR, 1.03 [1.03–1.03]), diastolic BP (CAD: OR, 1.05 [1.05–1.06]; ischemic stroke: OR, 1.05 [1.04–1.05]), type 2 diabetes (CAD: OR, 1.11 [1.08–1.15]; ischemic stroke: OR, 1.07 [1.04-1.10]), smoking initiation (CAD: OR, 1.26 [1.18–1.35]; ischemic stroke: OR, 1.24 [1.16–1.33]), educational attainment (CAD: OR, 0.62 [0.58–0.66]; ischemic stroke: OR, 0.68 [0.63–0.72]), low-density lipoprotein cholesterol (CAD: OR, 1.55 [1.41–1.71]), high-density lipoprotein cholesterol (CAD: OR, 0.82 [0.74–0.91]), triglycerides (CAD: OR, 1.29 [1.14–1.45]), body mass index (CAD: OR, 1.25 [1.19–1.32]), and alcohol dependence (OR, 1.04 [1.03–1.06]) were causally related to CVD.

      Conclusion

      This systematic MR study identified 11 modifiable factors as causal risk factors for CVD, indicating that these factors are important targets for preventing CVD.

      Graphical abstract

      Keywords

      Abbreviations:

      BMI (body mass index), BP (blood pressure), CAD (coronary artery disease), CARDIOGRAMPLUSC4D (Coronary Artery Disease Genome-wide Replication and Meta-analysis plus the Coronary Artery Disease Genetics), CVD (cardiovascular disease), GWAS (genome-wide association study), HDL-C (high-density lipoprotein cholesterol), IVW (inverse-variance weighted), LD (linkage disequilibrium), LDL-C (low-density lipoprotein cholesterol), MEGASTROKE (Multiancestry Genome-Wide Association Study of Stroke), MR (Mendelian randomization), MR-PRESSO (MR Pleiotropy Residual Sum and Outlier), OR (odds ratio), RCT (randomized controlled trial), RSS (residual sum of squares), SNP (single nucleotide polymorphism), STROBE-MR (Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization)

      1. Introduction

      Cardiovascular disease (CVD), including coronary artery disease (CAD) and ischemic stroke, is a major public health issue globally [
      • Who
      Global status report on noncommunicable diseases 2014.
      ]. In 2012, CVD accounted for 17.5 million deaths worldwide, among which 7.4 million were due to CAD and 6.7 million were due to stroke [
      • Who
      Global status report on noncommunicable diseases 2014.
      ]. CVD is a consequence of multiple etiologies and a series of modifiable factors (metabolic risk factors and lifestyle behaviors) are implicated in the development of CVD, such as hypertension, diabetes, hyperlipidemia, obesity, smoking, physical inactivity and unhealthy diet [
      • Roth G.A.
      • Mensah G.A.
      • Johnson C.O.
      • et al.
      Global burden of cardiovascular diseases and risk factors, 1990-2019: update from the GBD 2019 study.
      ]. In recent decades, randomized controlled trials (RCTs) have established the ability of BP control [
      • Bohm M.
      • Schumacher H.
      • Teo K.K.
      • et al.
      Cardiovascular outcomes and achieved blood pressure in patients with and without diabetes at high cardiovascular risk.
      ], lipid lowering [
      • Costa J.
      • Borges M.
      • David C.
      • Vaz Carneiro A.
      Efficacy of lipid lowering drug treatment for diabetic and non-diabetic patients: meta-analysis of randomised controlled trials.
      ], modest weight loss [
      • Seimon R.V.
      • Espinoza D.
      • Ivers L.
      • et al.
      Changes in body weight and blood pressure: paradoxical outcome events in overweight and obese subjects with cardiovascular disease.
      ], and sedentary behavior reduction [
      • Fenton S.A.M.
      • Veldhuijzen van Zanten J.
      • Kitas G.D.
      • et al.
      Sedentary behaviour is associated with increased long-term cardiovascular risk in patients with rheumatoid arthritis independently of moderate-to-vigorous physical activity.
      ] to reduce the risk of CVD. With regard to the other modifiable factors, there is a lack of evidence due to the methodological limitations to circumvent the bias inherent in observational studies to demonstrate whether they are causally associated with CVD or whether they are related merely as a consequence of a shared risk factor profile [
      • Shah A.D.
      • Langenberg C.
      • Rapsomaniki E.
      • et al.
      Type 2 diabetes and incidence of cardiovascular diseases: a cohort study in 1·9 million people.
      ,
      • Lawes C.M.
      • Parag V.
      • Bennett D.A.
      • et al.
      Blood glucose and risk of cardiovascular disease in the Asia Pacific region.
      ,
      • Mons U.
      • Muezzinler A.
      • Gellert C.
      • et al.
      Impact of smoking and smoking cessation on cardiovascular events and mortality among older adults: meta-analysis of individual participant data from prospective cohort studies of the CHANCES consortium.
      ,
      • Zheng Y.L.
      • Lian F.
      • Shi Q.
      • et al.
      Alcohol intake and associated risk of major cardiovascular outcomes in women compared with men: a systematic review and meta-analysis of prospective observational studies.
      ,
      • Kubota Y.
      • Heiss G.
      • MacLehose R.F.
      • Roetker N.S.
      • Folsom A.R.
      Association of educational attainment with lifetime risk of cardiovascular disease: the atherosclerosis risk in communities study.
      ]. To identify potential additional targets in the prevention of CVD and further reduce CVD burden, a better understanding of which risk factors have causal effects on the risk of CVD will be important for public health.
      Mendelian randomization (MR) is an emerging genetic method leveraging single nucleotide polymorphisms (SNPs) as instrumental variables to address causal hypotheses between exposures and diseases and this method is less vulnerable to residual confounders and reverse causation [
      • Burgess S.
      • Butterworth A.
      • Thompson S.G.
      Mendelian randomization analysis with multiple genetic variants using summarized data.
      ]. Some previous MR studies have demonstrated the causal effect of some modifiable factors (e.g., type 2 diabetes [
      • Gan W.
      • Bragg F.
      • Walters R.G.
      • et al.
      Genetic predisposition to type 2 diabetes and risk of subclinical atherosclerosis and cardiovascular diseases among 160,000 Chinese adults.
      ] and body mass index [BMI] [
      • Larsson S.C.
      • Back M.
      • Rees J.M.B.
      • Mason A.M.
      • Burgess S.
      Body mass index and body composition in relation to 14 cardiovascular conditions in UK Biobank: a Mendelian randomization study.
      ]) on the risk of CVD. Recently, genome-wide association studies (GWAS) with larger sample sizes discovered several novel genetic loci for diabetes and BMI [
      • Mahajan A.
      • Taliun D.
      • Thurner M.
      • et al.
      Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps.
      ,
      • Yengo L.
      • Sidorenko J.
      • Kemper K.E.
      • et al.
      Meta-analysis of genome-wide association studies for height and body mass index in approximately 700000 individuals of European ancestry.
      ] which could contribute greater power for making causal inferences in MR studies. For other modifiable factors, including fasting glucose, smoking heaviness and alcohol dependence, the evidence for their causal associations with CVD obtained from MR studies is limited. Fortunately, GWASs for these metabolic risk factors and lifestyle behaviors are now available [
      • Scott R.A.
      • Lagou V.
      • Welch R.P.
      • et al.
      Large-scale association analyses identify new loci influencing glycemic traits and provide insight into the underlying biological pathways.
      ,
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ,
      • Walters R.K.
      • Polimanti R.
      • Johnson E.C.
      • et al.
      Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders.
      ].
      Herein, we conducted a two-sample MR study to explore the causal associations of 15 modifiable risk factors with the risks of CAD and ischemic stroke based on the most recent and largest GWAS data (Graphical Abstract).

      2. Methods

      2.1 Study design

      The genetic variants used as instrumental variables in this study were previously published and were based on several cohorts of predominantly European ancestry (Table 1) [
      • Mahajan A.
      • Taliun D.
      • Thurner M.
      • et al.
      Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps.
      ,
      • Yengo L.
      • Sidorenko J.
      • Kemper K.E.
      • et al.
      Meta-analysis of genome-wide association studies for height and body mass index in approximately 700000 individuals of European ancestry.
      ,
      • Scott R.A.
      • Lagou V.
      • Welch R.P.
      • et al.
      Large-scale association analyses identify new loci influencing glycemic traits and provide insight into the underlying biological pathways.
      ,
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ,
      • Walters R.K.
      • Polimanti R.
      • Johnson E.C.
      • et al.
      Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders.
      ,
      • Elsworth B.
      • Lyon M.
      • Alexander T.
      • et al.
      The MRC IEU OpenGWAS data infrastructure.
      ,
      • Hemani G.
      • Zheng J.
      • Elsworth B.
      • et al.
      The MR-Base platform supports systematic causal inference across the human phenome.
      ,
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ,
      • Doherty A.
      • Smith-Byrne K.
      • Ferreira T.
      • et al.
      GWAS identifies 14 loci for device-measured physical activity and sleep duration.
      ,
      • Lee J.J.
      • Wedow R.
      • Okbay A.
      • et al.
      Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
      ,
      • Nikpay M.
      • Goel A.
      • Won H.H.
      • et al.
      A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease.
      ,
      • Malik R.
      • Chauhan G.
      • Traylor M.
      • et al.
      Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes.
      ]. There were three core assumptions for these instrumental variables: (1) the genetic variants were associated with exposures; (2) the genetic variants were independent of potential confounders; and (3) the genetic variants affected outcomes only through their effects on exposures. The participant selection, genotyping and imputation steps of these population-based cohorts and the corresponding baseline characteristics were described in detail in previous studies [
      • Mahajan A.
      • Taliun D.
      • Thurner M.
      • et al.
      Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps.
      ,
      • Yengo L.
      • Sidorenko J.
      • Kemper K.E.
      • et al.
      Meta-analysis of genome-wide association studies for height and body mass index in approximately 700000 individuals of European ancestry.
      ,
      • Scott R.A.
      • Lagou V.
      • Welch R.P.
      • et al.
      Large-scale association analyses identify new loci influencing glycemic traits and provide insight into the underlying biological pathways.
      ,
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ,
      • Walters R.K.
      • Polimanti R.
      • Johnson E.C.
      • et al.
      Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders.
      ,
      • Elsworth B.
      • Lyon M.
      • Alexander T.
      • et al.
      The MRC IEU OpenGWAS data infrastructure.
      ,
      • Hemani G.
      • Zheng J.
      • Elsworth B.
      • et al.
      The MR-Base platform supports systematic causal inference across the human phenome.
      ,
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ,
      • Doherty A.
      • Smith-Byrne K.
      • Ferreira T.
      • et al.
      GWAS identifies 14 loci for device-measured physical activity and sleep duration.
      ,
      • Lee J.J.
      • Wedow R.
      • Okbay A.
      • et al.
      Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
      ,
      • Nikpay M.
      • Goel A.
      • Won H.H.
      • et al.
      A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease.
      ,
      • Malik R.
      • Chauhan G.
      • Traylor M.
      • et al.
      Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes.
      ]. The protocol and data collection were approved by the ethics committee of the original GWASs, and written informed consent was obtained from each participant before data collection. This MR study followed the guidelines of Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization (STROBE-MR).
      Table 1Description of the risk factor-related genetic instruments used in the MR Study.
      Risk factorUsed SNPs
      SNPs used in the present MR analysis (coronary artery disease/ischemic stroke).
      Proxy SNPs
      Proxy SNPs used in the present MR analysis (coronary artery disease/ischemic stroke).
      Sample sizePopulationYearVariance
      Phenotypic variance explained by the genetic variants used in the present MR analysis (coronary artery disease/ischemic stroke).
      F-statistic
      Average F-statistic calculated with all included genetic instruments (coronary artery disease/ischemic stroke).
      Blood pressure
       Hypertension [
      • Elsworth B.
      • Lyon M.
      • Alexander T.
      • et al.
      The MRC IEU OpenGWAS data infrastructure.
      ,
      • Hemani G.
      • Zheng J.
      • Elsworth B.
      • et al.
      The MR-Base platform supports systematic causal inference across the human phenome.
      ]
      464/4679/9462 933European20185.06%/5.09%53/53
       Systolic BP [
      • Elsworth B.
      • Lyon M.
      • Alexander T.
      • et al.
      The MRC IEU OpenGWAS data infrastructure.
      ,
      • Hemani G.
      • Zheng J.
      • Elsworth B.
      • et al.
      The MR-Base platform supports systematic causal inference across the human phenome.
      ], mmHg
      1338/133820/20757 601European20189.47%/9.47%59/59
       Diastolic BP [
      • Elsworth B.
      • Lyon M.
      • Alexander T.
      • et al.
      The MRC IEU OpenGWAS data infrastructure.
      ,
      • Hemani G.
      • Zheng J.
      • Elsworth B.
      • et al.
      The MR-Base platform supports systematic causal inference across the human phenome.
      ], mmHg
      1404/140425/24757 601European201810.22%/10.22%61/61
      Glucose
       Type 2 diabetes [
      • Mahajan A.
      • Taliun D.
      • Thurner M.
      • et al.
      Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps.
      ]
      277/2780/0898 130European20182.12%/2.12%70/70
       Fasting glucose [
      • Scott R.A.
      • Lagou V.
      • Welch R.P.
      • et al.
      Large-scale association analyses identify new loci influencing glycemic traits and provide insight into the underlying biological pathways.
      ], mmol/L
      32/320/0133 010European20123.12%/3.12%134/134
      Lipids
       LDL-C [
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ], mg/dL
      53/531/1188 577Mostly European20134.22%/4.22%157/157
       HDL-C [
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ], mg/dL
      64/640/0188 577Mostly European20133.68%/3.68%113/113
       Triglycerides [
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ], mg/dL
      37/370/0188 577Mostly European20133.32%/3.32%175/175
      Overweight
       BMI [
      • Yengo L.
      • Sidorenko J.
      • Kemper K.E.
      • et al.
      Meta-analysis of genome-wide association studies for height and body mass index in approximately 700000 individuals of European ancestry.
      ], kg/m2
      830/8310/0≈700 000European2018≈7.43%/7.44%68/68
      Smoking
       Smoking initiation [
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ]
      355/3552/21 232 091European20191.29%/1.29%45/45
       Smoking heaviness [
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ], cigarettes smoked/day
      46/461/1337 334European20191.17%/1.17%87/87
      Alcohol intake
       Alcohol consumption [
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ], alcoholic drinks/week
      89/890/0941 280European20190.60%/0.60%64/64
       Alcohol dependence [
      • Walters R.K.
      • Polimanti R.
      • Johnson E.C.
      • et al.
      Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders.
      ]
      2/20/046 568European20180.18%/0.18%41/41
      Sedentariness
       Sedentary time [
      • Doherty A.
      • Smith-Byrne K.
      • Ferreira T.
      • et al.
      GWAS identifies 14 loci for device-measured physical activity and sleep duration.
      ], hours/day
      4/40/091 105European20180.15%/0.15%33/33
      Education
       Educational level [
      • Lee J.J.
      • Wedow R.
      • Okbay A.
      • et al.
      Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
      ], years
      1209/12120/01 131 881European20184.87%/4.89%48/48
      Abbreviations: BMI, body mass index; BP, blood pressure; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; MR, mendelian randomization; SNP, single nucleotide polymorphism.
      a SNPs used in the present MR analysis (coronary artery disease/ischemic stroke).
      b Proxy SNPs used in the present MR analysis (coronary artery disease/ischemic stroke).
      c Phenotypic variance explained by the genetic variants used in the present MR analysis (coronary artery disease/ischemic stroke).
      d Average F-statistic calculated with all included genetic instruments (coronary artery disease/ischemic stroke).

      2.2 Genetic instruments for metabolic risk factors and lifestyle behaviors

      An overview of the data concerning SNPs used as instruments in the MR study is presented in Table 1, and details are available in Tables S1-S15. Regarding unconfounded proxies for metabolic risk factors and lifestyle behaviors, we selected relevant SNPs identified in previously published GWASs as having reached genome-wide significance (p < 5.0 × 10−8) and being independent (r2 < 0.1 within a clumping window of 10 000 kb). In the case of SNPs exhibiting linkage disequilibrium (LD) above a threshold of r2 = 0.1, only the SNP with the lowest p was retained. If a specified exposure-associated SNP could not be found in the CAD or ischemic stroke dataset, a proxy variant (r2 > 0.7) was selected for the MR analysis based on a 1000 Genomes European reference panel via the LDlinkR package (the ‘LDproxy’ command) in the statistical software R. We excluded SNPs that were not available and had no proxy variant in the CAD and ischemic stroke GWASs. We incorporated 464/467 SNPs (CAD/ischemic stroke) for hypertension (based on self-reported diagnosis and medications; MRC IEU OpenGWAS data accessible at the website: https://gwas.mrcieu.ac.uk/datasets/ukb-b-14057/) [
      • Elsworth B.
      • Lyon M.
      • Alexander T.
      • et al.
      The MRC IEU OpenGWAS data infrastructure.
      ,
      • Hemani G.
      • Zheng J.
      • Elsworth B.
      • et al.
      The MR-Base platform supports systematic causal inference across the human phenome.
      ], 1338/1338 SNPs for systolic BP (https://gwas.mrcieu.ac.uk/datasets/ieu-b-38/) [
      • Elsworth B.
      • Lyon M.
      • Alexander T.
      • et al.
      The MRC IEU OpenGWAS data infrastructure.
      ,
      • Hemani G.
      • Zheng J.
      • Elsworth B.
      • et al.
      The MR-Base platform supports systematic causal inference across the human phenome.
      ], 1404/1404 SNPs for diastolic BP (https://gwas.mrcieu.ac.uk/datasets/ieu-b-39/) [
      • Elsworth B.
      • Lyon M.
      • Alexander T.
      • et al.
      The MRC IEU OpenGWAS data infrastructure.
      ,
      • Hemani G.
      • Zheng J.
      • Elsworth B.
      • et al.
      The MR-Base platform supports systematic causal inference across the human phenome.
      ], 277/278 SNPs for type 2 diabetes (based on self-reported or hospital diagnosis) [
      • Mahajan A.
      • Taliun D.
      • Thurner M.
      • et al.
      Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps.
      ], 32/32 SNPs for fasting glucose [
      • Scott R.A.
      • Lagou V.
      • Welch R.P.
      • et al.
      Large-scale association analyses identify new loci influencing glycemic traits and provide insight into the underlying biological pathways.
      ], 53/53 SNPs for low-density lipoprotein cholesterol (LDL-C) [
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ], 64/64 SNPs for high-density lipoprotein cholesterol (HDL-C) [
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ], 37/37 SNPs for triglycerides [
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ], 830/831 SNPs for BMI [
      • Yengo L.
      • Sidorenko J.
      • Kemper K.E.
      • et al.
      Meta-analysis of genome-wide association studies for height and body mass index in approximately 700000 individuals of European ancestry.
      ], 355/355 SNPs for smoking initiation [
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ], 46/46 SNPs for smoking heaviness [
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ], 89/89 SNPs for alcohol consumption [
      • Liu M.
      • Jiang Y.
      • Wedow R.
      • et al.
      Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
      ], 2/2 SNPs for alcohol dependence [
      • Walters R.K.
      • Polimanti R.
      • Johnson E.C.
      • et al.
      Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders.
      ], 4/4 SNPs for sedentary time [
      • Doherty A.
      • Smith-Byrne K.
      • Ferreira T.
      • et al.
      GWAS identifies 14 loci for device-measured physical activity and sleep duration.
      ], and 1209/1212 SNPs for educational level [
      • Lee J.J.
      • Wedow R.
      • Okbay A.
      • et al.
      Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
      ] (Fig. 2). Subsequently, we applied the package gtx in R (version 3.4.3; R Development Core Team) to calculate the phenotypic variance of each modifiable risk factor explained by the corresponding instrumental variables, and the observed phenotypic variance ranged from 0.15% for sedentary time to 10.22% for diastolic BP (Table 1). Finally, we calculated the F-statistic to estimate the strength of the genetic instruments for each metabolic risk factor and lifestyle behavior. A higher F-statistic suggests a stronger instrument, and a cutoff of 10 was used to distinguish between strong instruments and weak instruments [
      • Brion M.J.
      • Shakhbazov K.
      • Visscher P.M.
      Calculating statistical power in Mendelian randomization studies.
      ].

      2.3 Data sources for CAD and ischemic stroke

      Summary statistics for CAD originated from the Coronary Artery Disease Genome-wide Replication and Meta-analysis plus the Coronary Artery Disease Genetics (CARDIOGRAMPLUSC4D) consortium's 1000 Genomes–based genome-wide association meta-analysis of 48 studies, involving 60 801 CAD cases and 123 504 controls of European (77%), South Asian (13%), East Asian (6%), and Hispanic and African American ancestry (4%) [
      • Nikpay M.
      • Goel A.
      • Won H.H.
      • et al.
      A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease.
      ]. Case subjects were those with diagnoses of myocardial infarction (≈70% of the total number of cases), acute coronary syndrome, chronic stable angina, or coronary artery stenosis of at least 50%.
      Summary statistics for ischemic stroke originated from the dataset released by the Multiancestry Genome-Wide Association Study of Stroke (MEGASTROKE) project, which was launched by the International Stroke Genetics consortium [
      • Malik R.
      • Chauhan G.
      • Traylor M.
      • et al.
      Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes.
      ]. They performed a meta-analysis using summary data from 29 European-descent GWASs, including 40 585 cases and 406 111 controls. Cases were defined based on the clinical and imaging criteria.

      2.4 Statistical analyses

      For the main analyses, we performed a random-effect inverse-variance weighted (IVW) method to compute the association estimates, by which the causal linkage between exposure and outcome can be assessed when all of the genetic instruments are valid (Fig. 1) [
      • Burgess S.
      • Butterworth A.
      • Thompson S.G.
      Mendelian randomization analysis with multiple genetic variants using summarized data.
      ]. The heterogeneity among the SNPs utilized in the main analyses was assessed by Cochran's Q statistic (Table S16) [
      • Bowden J.
      • Davey Smith G.
      • Haycock P.C.
      • Burgess S.
      Consistent estimation in mendelian randomization with some invalid instruments using a weighted median estimator.
      ]. Subsequently, we leveraged the online web tool (https://shiny.cnsgenomics.com/mRnd/) to estimate the power for the MR analyses (Table S17) [
      • Brion M.J.
      • Shakhbazov K.
      • Visscher P.M.
      Calculating statistical power in Mendelian randomization studies.
      ].
      Figure 1
      Figure 1In the premise of meeting with three assumptions, Mendelian randomization (MR) analysis can be performed to infer the causality for the associations between exposures (metabolic and lifestyle risk factors) and outcomes (coronary artery disease and ischemic stroke). The assumptions of MR were as following: first, the genetic instruments are associated with the modifiable risk factors; second, the genetic instruments are independent of confounders; third, the genetic instruments impact on coronary artery disease and ischemic stroke exclusively via their effects on these risk factors.
      Figure 2
      Figure 2Overview of the main findings of this Mendelian randomization (MR) study on the effects of metabolic and lifestyle risk factors on coronary artery disease (CAD) and ischemic stroke, including the number of genetic variants; results from the inverse-variance weighted (IVW), weighted median (WM), MR-pleiotropy residual sum and outlier (MR-PRESSO) methods; as well as our conclusions. Association estimates in the results displayed here could be found in and . Abbreviations: BMI, body mass index; BP, blood pressure; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SNP, single nucleotide polymorphisms.
      The inclusion of multiple genetic instruments can increase the statistical power, while some genetic variants may not meet the criteria for instrumental variables, and their inclusion will most likely to lead to results with biased causalities given the presence of pleiotropy. Therefore, we further conducted a series of sensitivity analyses to assess the robustness of the primary conclusion. First, we employed the weighted median approach, in which the MR estimates were robust when up to 50% of genetic variants were invalid [
      • Bowden J.
      • Davey Smith G.
      • Haycock P.C.
      • Burgess S.
      Consistent estimation in mendelian randomization with some invalid instruments using a weighted median estimator.
      ]. Second, we applied MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) analysis to identify outlying SNPs by comparing the observed residual sum of squares (RSS) to the expected RSS and obtained a robust estimate with outlier correction via a leave-one-out approach [
      • Ong J.S.
      • MacGregor S.
      Implementing MR-PRESSO and GCTA-GSMR for pleiotropy assessment in Mendelian randomization studies from a practitioner's perspective.
      ]. Third, the MR-Egger regression method was performed to evaluate the potential effect of confounding by pleiotropic pathways on the results [
      • Hemani G.
      • Bowden J.
      • Davey Smith G.
      Evaluating the potential role of pleiotropy in Mendelian randomization studies.
      ]. Fourth, we utilized Steiger filtering to eliminate the SNPs which explained more of the phenotypic variance of the outcome compared to that of the exposure, and further repeated the IVW MR analyses after filtering [
      • Hemani G.
      • Tilling K.
      • Davey Smith G.
      Orienting the causal relationship between imprecisely measured traits using GWAS summary data.
      ]. In addition, we also applied the directionality Steiger test to determine whether the direction of causality was oriented from exposure to outcome through comparing the phenotypic variance of the exposure and the outcome explained by all included genetic variants. If the exposure had higher phenotypic variance than that of the outcome and the pSteiger reached the significance threshold, the direction of the identified association was deemed correct [
      • Hemani G.
      • Tilling K.
      • Davey Smith G.
      Orienting the causal relationship between imprecisely measured traits using GWAS summary data.
      ]. Fifth, taking into account the pleiotropy of instruments used for BP and lipids, we conducted 5 multivariable MR analyses to evaluate the relatively direct associations for these traits [
      • Hemani G.
      • Bowden J.
      • Davey Smith G.
      Evaluating the potential role of pleiotropy in Mendelian randomization studies.
      ]. In brief, we performed multivariable MR analyses for systolic BP and diastolic BP with adjustment for the genetic overlap among the 2 BP traits and performed multivariable MR analyses for LDL-C, HDL-C, and triglycerides with adjustment for the genetic overlap between the studied lipid trait and the other 2 lipid traits. Finally, to further assess the impact of BMI on the causal estimates of the effect of genetically determined type 2 diabetes on CAD and ischemic stroke, we studied that effect using genome-wide significant SNPs from the GWAS for type 2 diabetes adjusted for BMI (185 SNPs for CAD; and 185 SNPs for ischemic stroke) [
      • Mahajan A.
      • Taliun D.
      • Thurner M.
      • et al.
      Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps.
      ].
      The results for the outcomes (CAD and ischemic stroke) are presented as odds ratios (ORs) and their 95% confidence intervals (CIs). We assessed the strength of evidence for associations adjusting for multiple comparisons using a Bonferroni-corrected significance level of p < 1.67 × 10−3 (0.05/30 [15 risk factors∗2 outcomes] = 1.67 × 10−3). Values of p between 1.67 × 10−3 and 0.05 were considered nominally significant. All analyses were performed in R (version 3.4.3; R Development Core Team) with the packages gtx, LDlinkR, MendelianRandomization, MRPRESSO and TwoSampleMR.

      3. Results

      3.1 Strength and heterogeneity of genetic instruments

      The F-statistic for the genetic instruments of the 15 metabolic risk factors and lifestyle behaviors ranged from 33 to 175, suggesting that there was no evidence of weak instrument bias in this study (Table 1). The heterogeneity analysis suggested the presence of heterogeneity among the genetic variants for the majority of the metabolic risk factors and lifestyle behaviors (all p < 0.05), except for alcohol dependence and sedentary behavior (Table S16).

      3.2 Power calculation for the MR analysis

      As shown in Table S17, our MR analyses had adequate power to detect small effect sizes (e.g., OR = 1.1) for most of the metabolic risk factors and lifestyle behaviors.

      3.3 Relationship between modifiable risk factors and cardiovascular disease

      The metabolic and lifestyle risk factors, along with the strength and magnitude of their associations with CAD and ischemic stroke from the main analysis, are illustrated in Fig. 3. Among these modifiable factors, 9 of them had causal roles in higher odds of CVD: genetically predicted hypertension (CAD: OR, 5.19 [95% CI, 4.21–6.41]; ischemic stroke: OR, 4.92 [4.12–5.86]), systolic BP (CAD: OR per 1-mmHg increase, 1.03 [1.03–1.04]; ischemic stroke: OR per 1-mmHg increase, 1.03 [1.03–1.03]), diastolic BP (CAD: OR per 1-mmHg increase, 1.05 [1.05–1.06]; ischemic stroke: OR per 1-mmHg increase, 1.05 [1.04–1.05]), type 2 diabetes (CAD: OR, 1.11 [1.08–1.15]; ischemic stroke: OR, 1.07 [1.04–1.10]), smoking initiation (CAD: OR, 1.26 [1.18–1.35]; ischemic stroke: OR, 1.24 [1.16–1.33]), LDL-C (CAD: OR per 1-SD increase, 1.55 [1.41–1.71]), triglycerides (CAD: OR per 1-SD increase, 1.29 [1.14–1.45]), BMI (CAD: OR per 1-SD increase, 1.25 [1.19–1.32]), and alcohol dependence (CAD: OR, 1.04 [1.03–1.06]). In contrast, a genetic liability of educational attainment (CAD: OR per 1-SD increase, 0.62 [0.58–0.66]; ischemic stroke: OR per 1-SD increase, 0.68 [0.63–0.72]) and HDL-C (CAD: OR per 1-SD increase, 0.82 [0.74–0.91]) was inversely associated with the risk of CVD. In addition, there was suggestive evidence for the potential harmful impact of genetically higher fasting glucose (CAD: OR per 1-mmol/L increase, 1.26 [1.06–1.51]; ischemic stroke: OR per 1-mmol/L increase, 1.25 [1.00–1.55]) and smoking heaviness (CAD: OR per 1-SD increase, 1.17 [1.00–1.37]) on the risk of CVD. No significant associations with the risk of CVD were observed for genetically determined alcohol consumption and sedentary time (Fig. 3).
      Figure 3
      Figure 3Forest plots on the effects of metabolic risk factors and lifestyle behaviors on coronary artery disease (CAD) and ischemic stroke in the inverse-variance weighted method of Mendelian randomization analyses. Odds ratios (ORs) with 95% confidence intervals (CIs) represent the association estimates with CAD and ischemic stroke risks of: hypertension; 1-mmHg increase in systolic blood pressure (BP); 1-mmHg increase in diastolic BP; type 2 diabetes; 1-mmol/L increase in fasting glucose; 1-SD increase in low density lipoprotein cholesterol (LDL-C); 1-SD increase in high-density lipoprotein cholesterol (HDL-C); 1-SD increase in triglycerides; 1-SD increase in body mass index (BMI); ever smoked regularly compared with never smoked; 1-SD increase in number of cigarettes smoked per day; 1-SD increase in log-transformed alcoholic drinks/week; alcohol dependence; 1-SD increase in sedentary time; 1-SD increase in years of educational attainment.

      3.4 Sensitivity analyses

      We found directional pleiotropy for the associations of type 2 diabetes, fasting glucose, HDL-C, BMI, and smoking intensity with disease (CAD or ischemic stroke) based on the intercept of the MR-Egger regression model (all p < 1.67 × 10−3; Table 2). In addition, the weighted median analysis showed significant associations for hypertension (CAD: OR, 4.26 [3.40–5.35]; ischemic stroke: OR, 4.27 [3.37–5.42]), systolic BP (CAD: OR per 1-mmHg increase, 1.03 [1.02–1.03]; ischemic stroke: OR per 1-mmHg increase, 1.03 [1.03–1.03]), diastolic BP (CAD: OR per 1-mmHg increase, 1.05 [1.04–1.06]; ischemic stroke: OR per 1-mmHg increase, 1.05 [1.04–1.06]), type 2 diabetes (CAD: OR, 1.07 [1.03–1.12]), LDL-C (CAD: OR per 1-SD increase, 1.61 [1.49–1.76]), HDL-C (CAD: OR per 1-SD increase, 0.87 [0.80–0.94]), triglycerides (CAD: OR per 1-SD increase, 1.24 [1.13–1.37]), BMI (CAD: OR, 1.18 [1.09–1.27]), smoking initiation (CAD: OR, 1.22 [1.12–1.33]; ischemic stroke: OR, 1.22 [1.11–1.34]), and educational level (CAD: OR per 1-SD increase, 0.63 [0.58–0.69]; ischemic stroke: OR per 1-SD increase, 0.72 [0.65–0.78]) (Table 2). The MR-PRESSO method further confirmed the significant associations of hypertension (CAD: OR, 5.16 [4.21–6.32]; ischemic stroke: OR, 4.90 [4.04–5.95]), systolic BP (CAD: OR per 1-mmHg increase, 1.03 [1.03–1.04]; ischemic stroke: OR per 1-mmHg increase, 1.03 [1.03–1.03]), diastolic BP (CAD: OR per 1-mmHg increase, 1.06 [1.05–1.06]; ischemic stroke: OR per 1-mmHg increase, 1.05 [1.04–1.05]), type 2 diabetes (CAD: OR, 1.13 [1.10–1.17]; ischemic stroke: OR, 1.08 [1.06–1.11]), LDL-C (CAD: OR per 1-SD increase, 1.61 [1.45–1.78]), HDL-C (CAD: OR per 1-SD increase, 0.81 [0.73–0.89]), triglycerides (CAD: OR per 1-SD increase, 1.36 [1.20–1.55]), BMI (CAD: OR, 1.26 [1.20–1.32]), smoking initiation (CAD: OR, 1.26 [1.17–1.34]; ischemic stroke: OR, 1.24 [1.16–1.33]), and educational level (CAD: OR per 1-SD increase, 0.62 [0.58–0.66]; ischemic stroke: OR per 1-SD increase, 0.68 [0.63–0.72]) with the risk of CVD (Table 2). Although a nonsignificant link was observed between alcohol consumption and CAD in the principal analysis, the MR-PRESSO approach revealed a suggestive positive relationship (OR per 1-SD increase, 1.19 [1.00–1.40]) (Table 2). After removing the SNPs explaining more of the phenotypic variance of the outcome than that of the exposure by performing Steiger filtering, the repeated IVW MR analyses yielded similar findings as the main analysis (Table S18). Furthermore, the directionality Steiger test confirmed that the direction of the identified associations was correct (all p < 1.67 × 10−3; Table S18). In the sensitivity analyses adjusting for the effects of BMI, the relationships between genetically predicted type 2 diabetes and the risk of CAD and ischemic stroke were partly attenuated but still achieved statistical significance (CAD: OR, 1.12 [1.07–1.16]; ischemic stroke: OR, 1.07 [1.04–1.10]) (Table S19). In the multivariable MR model with adjustment for the genetic overlap among the 2 BP traits, the associations of systolic BP with risks of CAD and ischemic stroke remained significant (CAD: OR per 1-mmHg increase, 1.02 [1.01–1.03]; ischemic stroke: OR per 1-mmHg increase, 1.03 [1.02–1.04]), while the multivariable MR analyses adjusting for the genetic overlap among the 3 lipid traits suggested a significant association between LDL-C and the risk of CAD (OR per 1-SD increase, 1.52 [1.37–1.69]) (Table S20).
      Table 2Sensitivity analyses for the impact of modifiable risk factors on CAD and ischemic stroke.
      Risk factorOutcomeSNPsWeighted medianMR-EggerMR-PRESSO
      OR (95% CI)p valueIntercept (95% CI)p valueSNPsOR (95% CI)p value
      Blood pressure
       HypertensionCAD4644.26 (3.40–5.35)4.41 × 10−361.000 (0.995–1.005)0.98449
      Remaining SNPs after excluding outliers.
      5.16 (4.21–6.32)2.89 × 10−56
      Ischemic stroke4674.27 (3.37–5.42)4.53 × 10−331.000 (0.996–1.004)0.99464
      Remaining SNPs after excluding outliers.
      4.90 (4.04–5.95)4.39 × 10−58
       Systolic BPCAD13381.03 (1.02–1.03)6.63 × 10−431.000 (0.997–1.002)0.841 313
      Remaining SNPs after excluding outliers.
      1.03 (1.03–1.04)3.11 × 10−92
      Ischemic stroke13381.03 (1.03–1.03)3.99 × 10−470.997 (0.995–0.999)0.011 333
      Remaining SNPs after excluding outliers.
      1.03 (1.03–1.03)1.43 × 10−76
       Diastolic BPCAD14041.05 (1.04–1.06)1.37 × 10−421.000 (0.997–1.002)0.891 385
      Remaining SNPs after excluding outliers.
      1.06 (1.05–1.06)6.89 × 10−81
      Ischemic stroke14041.05 (1.04–1.06)5.94 × 10−420.997 (0.995–1.000)0.021 398
      Remaining SNPs after excluding outliers.
      1.05 (1.04–1.05)9.15 × 10−70
      Glucose
       Type 2 diabetesCAD2771.07 (1.03–1.12)1.98 × 10−41.011 (1.008–1.015)7.23 × 10−9272
      Remaining SNPs after excluding outliers.
      1.13 (1.10–1.17)5.57 × 10−16
      Ischemic stroke2781.06 (1.02–1.10)5.83 × 10−31.004 (1.001–1.008)8.30 × 10−3276
      Remaining SNPs after excluding outliers.
      1.08 (1.06–1.11)3.35 × 10−10
       Fasting glucoseCAD321.21 (1.00–1.45)0.041.006 (0.996–1.016)0.2331
      Remaining SNPs after excluding outliers.
      1.24 (1.05–1.46)0.01
      Ischemic stroke320.94 (0.75–1.18)0.611.019 (1.009–1.029)2.30 × 10−4321.25 (1.00–1.56)0.05
      Lipids
       LDL-CCAD531.61 (1.49–1.76)1.72 × 10−290.988 (0.979–0.997)8.87 × 10−349
      Remaining SNPs after excluding outliers.
      1.61 (1.45–1.78)4.61 × 10−19
      Ischemic stroke531.01 (0.93–1.10)0.820.989 (0.981–0.998)0.0151
      Remaining SNPs after excluding outliers.
      1.01 (0.94–1.08)0.84
       HDL-CCAD640.87 (0.80–0.94)5.13 × 10−40.988 (0.980–0.996)2.69 × 10−356
      Remaining SNPs after excluding outliers.
      0.81 (0.73–0.89)9.07 × 10−6
      Ischemic stroke641.02 (0.94–1.11)0.670.991 (0.986–0.995)2.74 × 10−5640.95 (0.90–1.01)0.08
       TriglyceridesCAD371.24 (1.13–1.37)8.08 × 10−61.009 (0.999–1.020)0.0830
      Remaining SNPs after excluding outliers.
      1.36 (1.20–1.55)2.32 × 10−6
      Ischemic stroke370.93 (0.84–1.02)0.131.007 (1.000–1.013)0.04371.00 (0.93–1.09)0.92
      Overweight
       BMICAD8301.18 (1.09–1.27)4.08 × 10−51.003 (1.001–1.005)6.14 × 10−4823
      Remaining SNPs after excluding outliers.
      1.26 (1.20–1.32)1.10 × 10−20
      Ischemic stroke8311.07 (0.99–1.16)0.091.002 (1.000–1.003)0.04830
      Remaining SNPs after excluding outliers.
      1.07 (1.03–1.13)2.46 × 10−3
      Smoking
       Smoking initiationCAD3551.22 (1.12–1.33)8.38 × 10−60.999 (0.993–1.005)0.73352
      Remaining SNPs after excluding outliers.
      1.26 (1.17–1.34)4.98 × 10−11
      Ischemic stroke3551.22 (1.11–1.34)6.29 × 10−51.003 (0.998–1.009)0.243551.24 (1.16–1.33)4.16 × 10−10
       Smoking heavinessCAD460.96 (0.80–1.16)0.681.010 (1.004–1.017)1.19 × 10−3461.17 (1.00–1.38)0.05
      Ischemic stroke461.03 (0.83–1.27)0.811.006 (0.999–1.014)0.10461.14 (0.96–1.36)0.15
      Alcohol intake
       Alcohol consumptionCAD891.25 (0.99–1.58)0.060.996 (0.992–1.001)0.1087
      Remaining SNPs after excluding outliers.
      1.19 (1.00–1.40)0.04
      Ischemic stroke891.07 (0.76–1.51)0.701.002 (0.997–1.008)0.39891.21 (0.97–1.50)0.09
       Alcohol dependence
      Sensitivity analyses could not be performed since the number of SNPs was less than 3.
      CAD2NANANANANANANA
      Ischemic stroke2NANANANANANANA
      Sedentariness
       Sedentary timeCAD40.84 (0.54–1.31)0.450.830 (0.621–1.108)0.2140.76 (0.43–1.35)0.36
      Ischemic stroke40.96 (0.61–1.51)0.870.887 (0.693–1.135)0.3440.88 (0.60–1.30)0.53
      Education
       Educational levelCAD12090.63 (0.58–0.69)5.62 × 10−260.999 (0.996–1.002)0.471 208
      Remaining SNPs after excluding outliers.
      0.62 (0.58–0.66)2.16 × 10−45
      Ischemic stroke12120.72 (0.65–0.78)8.40 × 10−131.002 (1.000–1.005)0.1012120.68 (0.63–0.72)1.26 × 10−30
      Odds ratios (ORs) with 95% confidence intervals (CIs) represent the association estimates with coronary artery disease (CAD) and ischemic stroke risks of: hypertension; 1-mmHg increase in systolic blood pressure (BP); 1-mmHg increase in diastolic BP; type 2 diabetes; 1-mmol/L increase in fasting glucose; 1-SD increase in low density lipoprotein cholesterol (LDL-C); 1-SD increase in high-density lipoprotein cholesterol (HDL-C); 1-SD increase in triglycerides; 1-SD increase in body mass index (BMI); ever smoked regularly compared with never smoked; 1-SD increase in number of cigarettes smoked per day; 1-SD increase in log-transformed alcoholic drinks/week; alcohol dependence; 1-SD increase in sedentary time; 1-SD increase in years of educational attainment.
      Abbreviations: MR-PRESSO, Mendelian randomization pleiotropy residual sum and outlier; NA, not applicable; SNPs, single nucleotide polymorphisms.
      a Remaining SNPs after excluding outliers.
      b Sensitivity analyses could not be performed since the number of SNPs was less than 3.

      4. Discussion

      This two-sample MR study provided genetic evidence of causality between some metabolic risk factors and lifestyle behaviors and CVD, showing that hypertension, systolic BP, diastolic BP, type 2 diabetes, LDL-C, triglycerides, BMI, smoking initiation, and alcohol dependence adversely impact the cardiovascular system and that HDL-C and educational attainment are inversely related to the risk of CVD. Additionally, fasting glucose and smoking intensity appeared to be potential causal risk factors for CVD.
      In the present study, the associations of BP, lipids, BMI, smoking initiation, excessive drinking, and educational attainment with CAD as well as the effects of smoking initiation and educational attainment on ischemic stroke were identical to those in previous observational studies [
      • Mons U.
      • Muezzinler A.
      • Gellert C.
      • et al.
      Impact of smoking and smoking cessation on cardiovascular events and mortality among older adults: meta-analysis of individual participant data from prospective cohort studies of the CHANCES consortium.
      ,
      • Kubota Y.
      • Heiss G.
      • MacLehose R.F.
      • Roetker N.S.
      • Folsom A.R.
      Association of educational attainment with lifetime risk of cardiovascular disease: the atherosclerosis risk in communities study.
      ,
      • Fuchs F.D.
      • Whelton P.K.
      High blood pressure and cardiovascular disease.
      ,
      • Hu G.
      • Cui Y.
      • Jousilahti P.
      • et al.
      Joint effect of high-density lipoprotein cholesterol and low-density lipoprotein cholesterol on the risk of coronary heart disease.
      ,
      • Sarwar N.
      • Danesh J.
      • Eiriksdottir G.
      • et al.
      Triglycerides and the risk of coronary heart disease: 10,158 incident cases among 262,525 participants in 29 Western prospective studies.
      ,
      • Rimm E.B.
      • Stampfer M.J.
      • Giovannucci E.
      • et al.
      Body size and fat distribution as predictors of coronary heart disease among middle-aged and older US men.
      ,
      • Corrao G.
      • Rubbiati L.
      • Bagnardi V.
      • Zambon A.
      • Poikolainen K.
      Alcohol and coronary heart disease: a meta-analysis.
      ] and MR studies [
      • Wan E.Y.F.
      • Fung W.T.
      • Schooling C.M.
      • et al.
      Blood pressure and risk of cardiovascular disease in UK Biobank: a mendelian randomization study.
      ,
      • Allara E.
      • Morani G.
      • Carter P.
      • et al.
      Genetic determinants of lipids and cardiovascular disease outcomes: a wide-angled mendelian randomization investigation.
      ,
      • Lyall D.M.
      • Celis-Morales C.
      • Ward J.
      • et al.
      Association of body mass index with cardiometabolic disease in the UK Biobank: a mendelian randomization study.
      ,
      • Rosoff D.B.
      • Davey Smith G.
      • Mehta N.
      • Clarke T.K.
      • Lohoff F.W.
      Evaluating the relationship between alcohol consumption, tobacco use, and cardiovascular disease: a multivariable Mendelian randomization study.
      ,
      • Carter A.R.
      • Gill D.
      • Davies N.M.
      • et al.
      Understanding the consequences of education inequality on cardiovascular disease: mendelian randomisation study.
      ]. This evidence indicated that these modifiable factors were causally associated with the risk of CVD. For type 2 diabetes, the intercept term in the MR-Egger regression model reflected a potential pleiotropic effect on the risk of CAD and ischemic stroke. However, the associations of type 2 diabetes with CAD and ischemic stroke remained causal in the further sensitivity analysis. Moreover, the finding for a deleterious role of type 2 diabetes on CVD is in concordance with previous prospective studies [
      • Shah A.D.
      • Langenberg C.
      • Rapsomaniki E.
      • et al.
      Type 2 diabetes and incidence of cardiovascular diseases: a cohort study in 1·9 million people.
      ,
      • Rosengren A.
      Cardiovascular disease in diabetes type 2: current concepts.
      ].
      The potential association between smoking intensity and CVD has been supported by previous prospective observational studies [
      • Price J.F.
      • Mowbray P.I.
      • Lee A.J.
      • Rumley A.
      • Lowe G.D.
      • Fowkes F.G.
      Relationship between smoking and cardiovascular risk factors in the development of peripheral arterial disease and coronary artery disease: edinburgh Artery Study.
      ]. In the present study, IVW MR analysis showed causal associations between smoking initiation (ever smoked regularly) and increased risks of CAD and ischemic stroke. However, these associations were attenuated in sensitivity analyses, and the MR-Egger regression demonstrated potential pleiotropy. Despite the uncertainty of the evidence of association between smoking and CVD, smoking is still discouraged due to its well-known adverse effects on CVD [
      • Mons U.
      • Muezzinler A.
      • Gellert C.
      • et al.
      Impact of smoking and smoking cessation on cardiovascular events and mortality among older adults: meta-analysis of individual participant data from prospective cohort studies of the CHANCES consortium.
      ].
      The role of fasting glucose in the development of CAD and ischemic stroke along with the effect thresholds remain controversial. Based on a prospective cohort study with 698 782 participants, nonlinear associations between fasting glucose and CAD and ischemic stroke were observed, with glucose levels between 3.90 and 5.59 mmol/L being associated with low vascular risk [
      • Sarwar N.
      • Gao P.
      • et al.
      The Emerging Risk Factors Collaboration
      Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies.
      ]. However, another observational study involving 237 468 subjects from the Asia Pacific region reported continuous associations of fasting glucose levels with the risks of CAD and stroke, which extended down to approximately 4.9 mmol/L, well below the usual fasting glucose threshold for the diagnosis of diabetes and impaired glucose tolerance [
      • Lawes C.M.
      • Parag V.
      • Bennett D.A.
      • et al.
      Blood glucose and risk of cardiovascular disease in the Asia Pacific region.
      ]. In the present study, we observed suggestive positive correlations between genetically determined fasting glucose and the predisposition to CAD and ischemic stroke. The results should only be interpreted as the average association at the population level because a two-sample MR study could not be performed to readily assess nonlinear associations. Despite the uncertainty of the relationship between fasting glucose and CAD and ischemic stroke, fasting glucose should still be kept within the normoglycemic range for individuals, since type 2 diabetes has been demonstrated to be highly comorbid with CVD [
      • Shah A.D.
      • Langenberg C.
      • Rapsomaniki E.
      • et al.
      Type 2 diabetes and incidence of cardiovascular diseases: a cohort study in 1·9 million people.
      ].
      BMI has been implicated in the risk of stroke, but the direction and magnitude of the associations are incongruous across earlier studies and at times appear to be nonlinear. For instance, a systematic review with 102 466 cases among 4 432 475 participants observed a J-shaped association between BMI and stroke, suggesting that the nadir of the dose-response curve was at BMI = 23–24 kg/m2 [
      • Liu X.
      • Zhang D.
      • Liu Y.
      • et al.
      A J-shaped relation of BMI and stroke: systematic review and dose-response meta-analysis of 4.43 million participants.
      ]. On the other hand, the literature has indicated sex differences in the correlation between BMI and ischemic stroke [
      • Kawate N.
      • Kayaba K.
      • Hara M.
      • Kotani K.
      • Ishikawa S.
      Jichi Medical School Cohort Study Group
      Body mass index and stroke incidence in Japanese community residents: the Jichi Medical School (JMS) Cohort Study.
      ,
      • Kroll M.E.
      • Green J.
      • Beral V.
      • et al.
      Adiposity and ischemic and hemorrhagic stroke: prospective study in women and meta-analysis.
      ] and these differences are likely to be driven by the different distributions of muscle and fat mass among men and women [
      • Toss F.
      • Wiklund P.
      • Franks P.W.
      • et al.
      Abdominal and gynoid adiposity and the risk of stroke.
      ]. Herein, we performed causal inference using the genetic variants that had been adjusted for sex as a proxy for BMI [
      • Yengo L.
      • Sidorenko J.
      • Kemper K.E.
      • et al.
      Meta-analysis of genome-wide association studies for height and body mass index in approximately 700000 individuals of European ancestry.
      ]. In contrast to previous MR studies with smaller scales [
      • Larsson S.C.
      • Back M.
      • Rees J.M.B.
      • Mason A.M.
      • Burgess S.
      Body mass index and body composition in relation to 14 cardiovascular conditions in UK Biobank: a Mendelian randomization study.
      ], BMI was reported as a potential causal risk factor for ischemic stroke in this large-sample MR study. Taken together, given that obesity frequently leads to CVD, there is evidence that a high BMI might need to be reduced moderately to a normal stratum [
      • Guh D.P.
      • Zhang W.
      • Bansback N.
      • Amarsi Z.
      • Birmingham C.L.
      • Anis A.H.
      The incidence of co-morbidities related to obesity and overweight: a systematic review and meta-analysis.
      ].
      In the present MR study, the lack of causal associations between alcohol consumption and sedentary time and risk of CVD suggested that the relationships found in previous observational studies [
      • Zheng Y.L.
      • Lian F.
      • Shi Q.
      • et al.
      Alcohol intake and associated risk of major cardiovascular outcomes in women compared with men: a systematic review and meta-analysis of prospective observational studies.
      ,
      • Wilmot E.G.
      • Edwardson C.L.
      • Achana F.A.
      • et al.
      Sedentary time in adults and the association with diabetes, cardiovascular disease and death: systematic review and meta-analysis.
      ] might be distorted by residual confounders and reverse causation, which were difficult to exclude completely. On the other hand, it might also be a consequence of inadequate power due to the relatively limited phenotypic variance explained by the included genetic instruments. The latter rationale explained the key reason for the null findings for sedentary time, which had been reported as a cardiovascular risk factor independent of physical activity in a previous randomized intervention study [
      • Fenton S.A.M.
      • Veldhuijzen van Zanten J.
      • Kitas G.D.
      • et al.
      Sedentary behaviour is associated with increased long-term cardiovascular risk in patients with rheumatoid arthritis independently of moderate-to-vigorous physical activity.
      ]. Additionally, the nonsignificant causal relationship between lipids and ischemic stroke disagreed with the findings from a previous randomized controlled trial [
      • Costa J.
      • Borges M.
      • David C.
      • Vaz Carneiro A.
      Efficacy of lipid lowering drug treatment for diabetic and non-diabetic patients: meta-analysis of randomised controlled trials.
      ], presumably due to bias as a result of the discrepancy in genetic background among the participants of the GWASs on ischemic stroke (European only) and lipids (mostly European with a small number of non-European).
      CAD shares common atherogenesis mechanisms with ischemic stroke. BP and lipids have been linked to the degree of arterial stiffness [
      • Frohlich E.D.
      • Susic D.
      Blood pressure, large arteries and atherosclerosis.
      ,
      • Podolecka E.
      • Grzeszczak W.
      • Zukowska-Szczechowska E.
      Correlation between serum low-density lipoprotein cholesterol concentration and arterial wall stiffness.
      ,
      • Lebrun C.E.
      • van der Schouw Y.T.
      • Bak A.A.
      • et al.
      Arterial stiffness in postmenopausal women: determinants of pulse wave velocity.
      ,
      • Kawasoe S.
      • Ide K.
      • Usui T.
      • et al.
      Association of serum triglycerides with arterial stiffness in subjects with low levels of low-density lipoprotein cholesterol.
      ] while arterial stiffness has emerged as a valid vascular marker of subclinical atherosclerosis. In addition, diabetes was reported to have the ability to induce abnormalities of the vascular wall and in turn to accelerate atherosclerotic lesions [
      • Galicia-Garcia U.
      • Benito-Vicente A.
      • Jebari S.
      • et al.
      Pathophysiology of type 2 diabetes mellitus.
      ]. As pivotal risk factors for atherosclerosis, obesity, smoking, and excessive drinking were implicated in multiple pathophysiologic pathways, including BP elevation [
      • Sarzani R.
      • Salvi F.
      • Dessi-Fulgheri P.
      • Rappelli A.
      Renin-angiotensin system, natriuretic peptides, obesity, metabolic syndrome, and hypertension: an integrated view in humans.
      ,
      • daLuz P.L.
      • Coimbra S.R.
      Alcohol and atherosclerosis.
      ] and endothelial dysfunction [
      • Virdis A.
      • Giannarelli C.
      • Neves M.F.
      • Taddei S.
      • Ghiadoni L.
      Cigarette smoking and hypertension.
      ]. In contrast, pursuing a healthy lifestyle might be responsible for the beneficial effects of a high educational level on the risk of CAD and ischemic stroke. However, due to the limited evidence for the associations between some risk factors and CAD and ischemic stroke from RCTs, causality is still unclear in this field. Our study has important scientific significance and public health implications, providing not only novel insight into the causal risk factors for CVD from a genetic standpoint, but also targets for the prevention of CVD. This MR study suggests that BP, glucose, lipids, BMI, smoking, alcohol intake, and education are promising targets for the prevention of CVD. Further clinical trials are needed to confirm the preventive effect of interventions targeting these metabolic risk factors and lifestyle behaviors.
      A strength of this study is the use of the MR design. Taking advantage of the random assortment of alleles during gametogenesis, MR studies are able to explore the potential causality for the associations between exposures and diseases. Moreover, we evaluated the effects of a variety of modifiable risk factors on CVD risk using several large-scale GWAS datasets, which enabled us to perform MR analyses with high statistical power. However, several limitations should be discussed here. First, this two-sample MR study assumed linear associations between exposures and diseases (CAD and ischemic stroke), whereas the associations for certain modifiable risk factors (e.g., fasting glucose level [
      • Sarwar N.
      • Gao P.
      • et al.
      The Emerging Risk Factors Collaboration
      Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies.
      ]) might be “J” or “U” shaped. Second, for the risk factors for which the variance explained by genetic instruments is low, we noted that causal inference might be hindered by limited power and accuracy (Table S17). Third, alcohol consumption, smoking, and sedentary behavior were highly confounded, and the variance explained by the MR instruments was generally low for these exposures. Therefore, further studies are warranted to investigate the effects of residual confounding and pleiotropy on the associations of metabolic risk factors and lifestyle behaviors with CVD. Fourth, the SNP-exposure and SNP-outcome estimates used in our study were mostly based on European subjects, leading to reduced reliability when extrapolating the findings to individuals of non-European descent. However, as we focused on European ancestry, considerable population stratification bias is not expected in the current study. Nonetheless, we emphasized the importance of incorporating other race-based populations in further studies for more conclusive results. Finally, there might be an overlap between participants in exposure GWASs from UK Biobank and the two outcome GWASs, which may lead to weak instrument bias [
      • Burgess S.
      • Davies N.M.
      • Thompson S.G.
      Bias due to participant overlap in two-sample Mendelian randomization.
      ]. Although the F-statistic suggested that instrument bias was minimal in this MR study, further two-sample MR studies based on independent cohorts without overlapping participants are needed to better understand the role of metabolic risk factors and lifestyle behaviors in the etiology of CVD.
      In conclusion, this MR study provided novel insight into the causal risk factors for CVD from a genetic standpoint, suggesting that BP, glucose, lipids, BMI, smoking, alcohol intake, and education are promising targets for the prevention of CVD. Further studies are warranted to confirm these findings in other populations and to elucidate the underlying mechanisms.

      Funding

      This study was supported by the National Natural Science Foundation of China (grant: 82103917, 82103921 and 82020108028), the high level personnel project of Jiangsu Province (grant: JSSCBS20210712), and the Natural Science Research Project of Jiangsu Provincial Higher Education (grant: 21KJB330006), and a Project of the Priority Academic Program Development of Jiangsu Higher Education Institutions, China.

      Authors’ contributions

      The study was conceived and designed by YJ, RW, YZh and ZZ. YJ, RW, YZh and ZZ coordinated the study. YJ, RW, DG, LS, MS, KZ, PY, YZa, YW, FL, YZh, and ZZ contributed to data collection. YJ and RW performed the statistical analysis and prepared the first draft of manuscript. YZh and ZZ revised the paper and helped to write the final draft of manuscript. ZZ is guarantor.

      Consent for publication

      The authors confirmed that all participants provided informed consent for publication.

      Availability of data and material

      All summary statistics used in this two-sample Mendelian randomization are available online from each genome-wide association study. Statistical code is available on the request by directly contacting the corresponding author (email: [email protected] ).

      Declaration of competing interest

      The authors report no conflicts of interest.

      Acknowledgements

      We thank the authors and participants of all GWASs that we have used, for making their results publicly available. Full acknowledgement and funding statements for each of these resources are available via the relevant cited reports.

      Appendix A. Supplementary data

      The following is/are the supplementary data to this article:

      References

        • Who
        Global status report on noncommunicable diseases 2014.
        2014
        • Roth G.A.
        • Mensah G.A.
        • Johnson C.O.
        • et al.
        Global burden of cardiovascular diseases and risk factors, 1990-2019: update from the GBD 2019 study.
        J Am Coll Cardiol. 2020; 76: 2982-3021
        • Bohm M.
        • Schumacher H.
        • Teo K.K.
        • et al.
        Cardiovascular outcomes and achieved blood pressure in patients with and without diabetes at high cardiovascular risk.
        Eur Heart J. 2019; 40: 2032-2043
        • Costa J.
        • Borges M.
        • David C.
        • Vaz Carneiro A.
        Efficacy of lipid lowering drug treatment for diabetic and non-diabetic patients: meta-analysis of randomised controlled trials.
        BMJ. 2006; 332: 1115-1124
        • Seimon R.V.
        • Espinoza D.
        • Ivers L.
        • et al.
        Changes in body weight and blood pressure: paradoxical outcome events in overweight and obese subjects with cardiovascular disease.
        Int J Obes. 2014; 38: 1165-1171
        • Fenton S.A.M.
        • Veldhuijzen van Zanten J.
        • Kitas G.D.
        • et al.
        Sedentary behaviour is associated with increased long-term cardiovascular risk in patients with rheumatoid arthritis independently of moderate-to-vigorous physical activity.
        BMC Muscoskel Disord. 2017; 18: 131
        • Shah A.D.
        • Langenberg C.
        • Rapsomaniki E.
        • et al.
        Type 2 diabetes and incidence of cardiovascular diseases: a cohort study in 1·9 million people.
        Lancet Diabetes Endocrinol. 2015; 3: 105-113
        • Lawes C.M.
        • Parag V.
        • Bennett D.A.
        • et al.
        Blood glucose and risk of cardiovascular disease in the Asia Pacific region.
        Diabetes Care. 2004; 27: 2836-2842
        • Mons U.
        • Muezzinler A.
        • Gellert C.
        • et al.
        Impact of smoking and smoking cessation on cardiovascular events and mortality among older adults: meta-analysis of individual participant data from prospective cohort studies of the CHANCES consortium.
        BMJ. 2015; 350: h1551
        • Zheng Y.L.
        • Lian F.
        • Shi Q.
        • et al.
        Alcohol intake and associated risk of major cardiovascular outcomes in women compared with men: a systematic review and meta-analysis of prospective observational studies.
        BMC Publ Health. 2015; 15: 773
        • Kubota Y.
        • Heiss G.
        • MacLehose R.F.
        • Roetker N.S.
        • Folsom A.R.
        Association of educational attainment with lifetime risk of cardiovascular disease: the atherosclerosis risk in communities study.
        JAMA Intern Med. 2017; 177: 1165-1172
        • Burgess S.
        • Butterworth A.
        • Thompson S.G.
        Mendelian randomization analysis with multiple genetic variants using summarized data.
        Genet Epidemiol. 2013; 37: 658-665
        • Gan W.
        • Bragg F.
        • Walters R.G.
        • et al.
        Genetic predisposition to type 2 diabetes and risk of subclinical atherosclerosis and cardiovascular diseases among 160,000 Chinese adults.
        Diabetes. 2019; 68: 2155-2164
        • Larsson S.C.
        • Back M.
        • Rees J.M.B.
        • Mason A.M.
        • Burgess S.
        Body mass index and body composition in relation to 14 cardiovascular conditions in UK Biobank: a Mendelian randomization study.
        Eur Heart J. 2020; 41: 221-226
        • Mahajan A.
        • Taliun D.
        • Thurner M.
        • et al.
        Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps.
        Nat Genet. 2018; 50: 1505-1513
        • Yengo L.
        • Sidorenko J.
        • Kemper K.E.
        • et al.
        Meta-analysis of genome-wide association studies for height and body mass index in approximately 700000 individuals of European ancestry.
        Hum Mol Genet. 2018; 27: 3641-3649
        • Scott R.A.
        • Lagou V.
        • Welch R.P.
        • et al.
        Large-scale association analyses identify new loci influencing glycemic traits and provide insight into the underlying biological pathways.
        Nat Genet. 2012; 44: 991-1005
        • Liu M.
        • Jiang Y.
        • Wedow R.
        • et al.
        Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use.
        Nat Genet. 2019; 51: 237-244
        • Walters R.K.
        • Polimanti R.
        • Johnson E.C.
        • et al.
        Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders.
        Nat Neurosci. 2018; 21: 1656-1669
        • Elsworth B.
        • Lyon M.
        • Alexander T.
        • et al.
        The MRC IEU OpenGWAS data infrastructure.
        bioRxiv, 2020 (08.10.244293)
        • Hemani G.
        • Zheng J.
        • Elsworth B.
        • et al.
        The MR-Base platform supports systematic causal inference across the human phenome.
        Elife. 2018; 7
        • Willer C.J.
        • Schmidt E.M.
        • Sengupta S.
        • et al.
        Discovery and refinement of loci associated with lipid levels.
        Nat Genet. 2013; 45: 1274-1283
        • Doherty A.
        • Smith-Byrne K.
        • Ferreira T.
        • et al.
        GWAS identifies 14 loci for device-measured physical activity and sleep duration.
        Nat Commun. 2018; 9: 5257
        • Lee J.J.
        • Wedow R.
        • Okbay A.
        • et al.
        Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
        Nat Genet. 2018; 50: 1112-1121
        • Nikpay M.
        • Goel A.
        • Won H.H.
        • et al.
        A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease.
        Nat Genet. 2015; 47: 1121-1130
        • Malik R.
        • Chauhan G.
        • Traylor M.
        • et al.
        Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes.
        Nat Genet. 2018; 50: 524-537
        • Brion M.J.
        • Shakhbazov K.
        • Visscher P.M.
        Calculating statistical power in Mendelian randomization studies.
        Int J Epidemiol. 2013; 42: 1497-1501
        • Bowden J.
        • Davey Smith G.
        • Haycock P.C.
        • Burgess S.
        Consistent estimation in mendelian randomization with some invalid instruments using a weighted median estimator.
        Genet Epidemiol. 2016; 40: 304-314
        • Ong J.S.
        • MacGregor S.
        Implementing MR-PRESSO and GCTA-GSMR for pleiotropy assessment in Mendelian randomization studies from a practitioner's perspective.
        Genet Epidemiol. 2019; 43: 609-616
        • Hemani G.
        • Bowden J.
        • Davey Smith G.
        Evaluating the potential role of pleiotropy in Mendelian randomization studies.
        Hum Mol Genet. 2018; 27: R195-R208
        • Hemani G.
        • Tilling K.
        • Davey Smith G.
        Orienting the causal relationship between imprecisely measured traits using GWAS summary data.
        PLoS Genet. 2017; 13e1007081
        • Fuchs F.D.
        • Whelton P.K.
        High blood pressure and cardiovascular disease.
        Hypertension. 2020; 75: 285-292
        • Hu G.
        • Cui Y.
        • Jousilahti P.
        • et al.
        Joint effect of high-density lipoprotein cholesterol and low-density lipoprotein cholesterol on the risk of coronary heart disease.
        Eur J Prev Cardiol. 2013; 20: 89-97
        • Sarwar N.
        • Danesh J.
        • Eiriksdottir G.
        • et al.
        Triglycerides and the risk of coronary heart disease: 10,158 incident cases among 262,525 participants in 29 Western prospective studies.
        Circulation. 2007; 115: 450-458
        • Rimm E.B.
        • Stampfer M.J.
        • Giovannucci E.
        • et al.
        Body size and fat distribution as predictors of coronary heart disease among middle-aged and older US men.
        Am J Epidemiol. 1995; 141: 1117-1127
        • Corrao G.
        • Rubbiati L.
        • Bagnardi V.
        • Zambon A.
        • Poikolainen K.
        Alcohol and coronary heart disease: a meta-analysis.
        Addiction. 2000; 95: 1505-1523
        • Wan E.Y.F.
        • Fung W.T.
        • Schooling C.M.
        • et al.
        Blood pressure and risk of cardiovascular disease in UK Biobank: a mendelian randomization study.
        Hypertension. 2021; 77: 367-375
        • Allara E.
        • Morani G.
        • Carter P.
        • et al.
        Genetic determinants of lipids and cardiovascular disease outcomes: a wide-angled mendelian randomization investigation.
        Circ Genom Precis Med. 2019; 12e002711
        • Lyall D.M.
        • Celis-Morales C.
        • Ward J.
        • et al.
        Association of body mass index with cardiometabolic disease in the UK Biobank: a mendelian randomization study.
        JAMA Cardiol. 2017; 2: 882-889
        • Rosoff D.B.
        • Davey Smith G.
        • Mehta N.
        • Clarke T.K.
        • Lohoff F.W.
        Evaluating the relationship between alcohol consumption, tobacco use, and cardiovascular disease: a multivariable Mendelian randomization study.
        PLoS Med. 2020; 17e1003410
        • Carter A.R.
        • Gill D.
        • Davies N.M.
        • et al.
        Understanding the consequences of education inequality on cardiovascular disease: mendelian randomisation study.
        BMJ. 2019; 365: l1855
        • Rosengren A.
        Cardiovascular disease in diabetes type 2: current concepts.
        J Intern Med. 2018; 284: 240-253
        • Price J.F.
        • Mowbray P.I.
        • Lee A.J.
        • Rumley A.
        • Lowe G.D.
        • Fowkes F.G.
        Relationship between smoking and cardiovascular risk factors in the development of peripheral arterial disease and coronary artery disease: edinburgh Artery Study.
        Eur Heart J. 1999; 20: 344-353
        • Sarwar N.
        • Gao P.
        • et al.
        • The Emerging Risk Factors Collaboration
        Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies.
        Lancet. 2010; 375: 2215-2222
        • Liu X.
        • Zhang D.
        • Liu Y.
        • et al.
        A J-shaped relation of BMI and stroke: systematic review and dose-response meta-analysis of 4.43 million participants.
        Nutr Metabol Cardiovasc Dis. 2018; 28: 1092-1099
        • Kawate N.
        • Kayaba K.
        • Hara M.
        • Kotani K.
        • Ishikawa S.
        • Jichi Medical School Cohort Study Group
        Body mass index and stroke incidence in Japanese community residents: the Jichi Medical School (JMS) Cohort Study.
        J Epidemiol. 2017; 27: 325-330
        • Kroll M.E.
        • Green J.
        • Beral V.
        • et al.
        Adiposity and ischemic and hemorrhagic stroke: prospective study in women and meta-analysis.
        Neurology. 2016; 87: 1473-1481
        • Toss F.
        • Wiklund P.
        • Franks P.W.
        • et al.
        Abdominal and gynoid adiposity and the risk of stroke.
        Int J Obes. 2011; 35: 1427-1432
        • Guh D.P.
        • Zhang W.
        • Bansback N.
        • Amarsi Z.
        • Birmingham C.L.
        • Anis A.H.
        The incidence of co-morbidities related to obesity and overweight: a systematic review and meta-analysis.
        BMC Publ Health. 2009; 9: 88
        • Wilmot E.G.
        • Edwardson C.L.
        • Achana F.A.
        • et al.
        Sedentary time in adults and the association with diabetes, cardiovascular disease and death: systematic review and meta-analysis.
        Diabetologia. 2012; 55: 2895-2905
        • Frohlich E.D.
        • Susic D.
        Blood pressure, large arteries and atherosclerosis.
        Adv Cardiol. 2007; 44: 117-124
        • Podolecka E.
        • Grzeszczak W.
        • Zukowska-Szczechowska E.
        Correlation between serum low-density lipoprotein cholesterol concentration and arterial wall stiffness.
        Kardiol Pol. 2018; 76: 1712-1716
        • Lebrun C.E.
        • van der Schouw Y.T.
        • Bak A.A.
        • et al.
        Arterial stiffness in postmenopausal women: determinants of pulse wave velocity.
        J Hypertens. 2002; 20: 2165-2172
        • Kawasoe S.
        • Ide K.
        • Usui T.
        • et al.
        Association of serum triglycerides with arterial stiffness in subjects with low levels of low-density lipoprotein cholesterol.
        Circ J. 2018; 82: 3052-3057
        • Galicia-Garcia U.
        • Benito-Vicente A.
        • Jebari S.
        • et al.
        Pathophysiology of type 2 diabetes mellitus.
        Int J Mol Sci. 2020; 21
        • Sarzani R.
        • Salvi F.
        • Dessi-Fulgheri P.
        • Rappelli A.
        Renin-angiotensin system, natriuretic peptides, obesity, metabolic syndrome, and hypertension: an integrated view in humans.
        J Hypertens. 2008; 26: 831-843
        • daLuz P.L.
        • Coimbra S.R.
        Alcohol and atherosclerosis.
        An Acad Bras Cienc. 2001; 73: 51-55
        • Virdis A.
        • Giannarelli C.
        • Neves M.F.
        • Taddei S.
        • Ghiadoni L.
        Cigarette smoking and hypertension.
        Curr Pharmaceut Des. 2010; 16: 2518-2525
        • Burgess S.
        • Davies N.M.
        • Thompson S.G.
        Bias due to participant overlap in two-sample Mendelian randomization.
        Genet Epidemiol. 2016; 40: 597-608