Advertisement

A genome-wide search for gene-by-obesity interaction loci of dyslipidemia in Koreans shows diverse genetic risk alleles

  • Moonil Kang
    Affiliations
    Division of Genome and Health Big Data, Department of Public Health Sciences Graduate School of Public Health, Seoul National University, Seoul, Republic of Korea
    Search for articles by this author
  • Joohon Sung
    Correspondence
    To whom correspondence should be addressed.
    Affiliations
    Division of Genome and Health Big Data, Department of Public Health Sciences Graduate School of Public Health, Seoul National University, Seoul, Republic of Korea

    Institute of Health and Environment, Graduate School of Public Health, Seoul National University, Seoul, Republic of Korea
    Search for articles by this author
Open AccessPublished:October 29, 2019DOI:https://doi.org/10.1194/jlr.P119000226
      Dyslipidemia is a well-established risk factor for CVD. Studies suggest that similar fat accumulation in a given population might result in different levels of dyslipidemia risk among individuals; for example, despite similar or leaner body composition compared with Caucasians, Asians of Korean descent experience a higher prevalence of dyslipidemia. These variations imply a possible role of gene-obesity interactions on lipid profiles. Genome-wide association studies have identified more than 500 loci regulating plasma lipids, but the interaction structure between genes and obesity traits remains unclear. We hypothesized that some loci modify the effects of obesity on dyslipidemia risk and analyzed extensive gene-environment interactions (G×Es) at genome-wide levels to search for replicated gene-obesity interactive SNPs. In four Korean cohorts (n = 18,025), we identified and replicated 20 gene-obesity interactions, including novel variants (SCN1A and SLC12A8) and known lipid-associated variants (APOA5, BUD13, ZNF259, and HMGCR). When we estimated the additional heritability of dyslipidemia by considering G×Es, the gain was substantial for triglycerides (TGs) but mild for LDL cholesterol (LDL-C) and total cholesterol (Total-C); the interaction explained up to 18.7% of TG, 2.4% of LDL-C, and 1.9% of Total-C heritability associated with waist-hip ratio. Our findings suggest that some individuals are prone to develop abnormal lipid profiles, particularly with regard to TGs, even with slight increases in obesity indices; ethnic diversities in the risk alleles might partly explain the differential dyslipidemia risk between populations. Research about these interacting variables may facilitate knowledge-based approaches to personalize health guidelines according to individual genetic profiles.
      Dyslipidemia is a well-established risk factor for CVD. Genome-wide association studies (GWASs) have identified more than 500 loci influencing plasma lipid levels in European ancestry cohorts (
      • Teslovich T.M.
      • Musunuru K.
      • Smith A.V.
      • Edmondson A.C.
      • Stylianou I.M.
      • Koseki M.
      • Pirruccello J.P.
      • Ripatti S.
      • Chasman D.I.
      • Willer C.J.
      • et al.
      Biological, clinical and population relevance of 95 loci for blood lipids.
      ,
      • Asselbergs F.W.
      • Guo Y.
      • van Iperen E.P.
      • Sivapalaratnam S.
      • Tragante V.
      • Lanktree M.B.
      • Lange L.A.
      • Almoguera B.
      • Appelman Y.E.
      • Barnard J.
      • et al.
      Large-scale gene-centric meta-analysis across 32 studies identifies multiple lipid loci.
      ,
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • Peloso G.M.
      • Gustafsson S.
      • Kanoni S.
      • Ganna A.
      • Chen J.
      • Buchkovich M.L.
      • Mora S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ,
      • Surakka I.
      • Horikoshi M.
      • Magi R.
      • Sarin A.P.
      • Mahajan A.
      • Lagou V.
      • Marullo L.
      • Ferreira T.
      • Miraglio B.
      • Timonen S.
      • et al.
      The impact of low-frequency and rare variants on lipid levels.
      ,
      • Below J.E.
      • Parra E.J.
      • Gamazon E.R.
      • Torres J.
      • Krithika S.
      • Candille S.
      • Lu Y.
      • Manichakul A.
      • Peralta-Romero J.
      • Duan Q.
      • et al.
      Meta-analysis of lipid-traits in Hispanics identifies novel loci, population-specific effects, and tissue-specific enrichment of eQTLs.
      ,
      • Spracklen C.N.
      • Chen P.
      • Kim Y.J.
      • Wang X.
      • Cai H.
      • Li S.
      • Long J.
      • Wu Y.
      • Wang Y.X.
      • Takeuchi F.
      • et al.
      Association analyses of East Asian individuals and trans-ancestry analyses with European individuals reveal new loci associated with cholesterol and triglyceride levels.
      ,
      • Lu X.
      • Peloso G.M.
      • Liu D.J.
      • Wu Y.
      • Zhang H.
      • Zhou W.
      • Li J.
      • Tang C.S.
      • Dorajoo R.
      • Li H.
      • et al.
      Exome chip meta-analysis identifies novel loci and East Asian-specific coding variants that contribute to lipid levels and coronary artery disease.
      ,
      • Liu D.J.
      • Peloso G.M.
      • Yu H.
      • Butterworth A.S.
      • Wang X.
      • Mahajan A.
      • Saleheen D.
      • Emdin C.
      • Alam D.
      • Alves A.C.
      • et al.
      Exome-wide association study of plasma lipids in >300,000 individuals.
      ,
      • Southam L.
      • Gilly A.
      • Suveges D.
      • Farmaki A.E.
      • Schwartzentruber J.
      • Tachmazidou I.
      • Matchan A.
      • Rayner N.W.
      • Tsafantakis E.
      • Karaleftheri M.
      • et al.
      Whole genome sequencing and imputation in isolated populations identify genetic associations with medically-relevant complex traits.
      ,
      • Davis J.P.
      • Huyghe J.R.
      • Locke A.E.
      • Jackson A.U.
      • Sim X.
      • Stringham H.M.
      • Teslovich T.M.
      • Welch R.P.
      • Fuchsberger C.
      • Narisu N.
      • et al.
      Common, low-frequency, and rare genetic variants associated with lipoprotein subclasses and triglyceride measures in Finnish men from the METSIM study.
      ,
      • Kanai M.
      • Akiyama M.
      • Takahashi A.
      • Matoba N.
      • Momozawa Y.
      • Ikeda M.
      • Iwata N.
      • Ikegawa S.
      • Hirata M.
      • Matsuda K.
      • et al.
      Genetic analysis of quantitative traits in the Japanese population links cell types to complex human diseases.
      ). The total variance explained by these loci in the Framingham Heart Study was 15.0% for total cholesterol (Total-C) levels, 13.7% for HDL-C levels, 14.6% for LDL-C levels, and 11.7% for triglyceride (TG) levels, corresponding to 25–30% of the heritability of each lipid trait (
      • Teslovich T.M.
      • Musunuru K.
      • Smith A.V.
      • Edmondson A.C.
      • Stylianou I.M.
      • Koseki M.
      • Pirruccello J.P.
      • Ripatti S.
      • Chasman D.I.
      • Willer C.J.
      • et al.
      Biological, clinical and population relevance of 95 loci for blood lipids.
      ,
      • Asselbergs F.W.
      • Guo Y.
      • van Iperen E.P.
      • Sivapalaratnam S.
      • Tragante V.
      • Lanktree M.B.
      • Lange L.A.
      • Almoguera B.
      • Appelman Y.E.
      • Barnard J.
      • et al.
      Large-scale gene-centric meta-analysis across 32 studies identifies multiple lipid loci.
      ,
      • Willer C.J.
      • Schmidt E.M.
      • Sengupta S.
      • Peloso G.M.
      • Gustafsson S.
      • Kanoni S.
      • Ganna A.
      • Chen J.
      • Buchkovich M.L.
      • Mora S.
      • et al.
      Discovery and refinement of loci associated with lipid levels.
      ).
      Previous GWASs on lipids have been successful in terms of both the richness of robustly replicated loci and the genetic variances explained by these loci, but the current list in the GWAS Catalog is based on marginal association models assuming a lack of gene-gene or gene-environment interactions (G×Es). Ethnic differences in lipid abnormalities reactive to obesity, however, suggest that the interplay between genetic background and obesity traits may play a role in regulating lipid concentrations (
      • Frank A.T.
      • Zhao B.
      • Jose P.O.
      • Azar K.M.
      • Fortmann S.P.
      • Palaniappan L.P.
      Racial/ethnic differences in dyslipidemia patterns.
      ). Compared with non-Hispanic whites, for example, non-Hispanic blacks have a lower dyslipidemia subtype prevalence despite a higher obesity risk (
      • Frank A.T.
      • Zhao B.
      • Jose P.O.
      • Azar K.M.
      • Fortmann S.P.
      • Palaniappan L.P.
      Racial/ethnic differences in dyslipidemia patterns.
      ,
      • Hyre A.D.
      • Muntner P.
      • Menke A.
      • Raggi P.
      • He J.
      Trends in ATP-III-defined high blood cholesterol prevalence, awareness, treatment and control among US adults.
      ). Conversely, despite similar or even leaner body compositions than Caucasians of the same age and sex, the prevalence of dyslipidemia is increased in Asians (
      • Frank A.T.
      • Zhao B.
      • Jose P.O.
      • Azar K.M.
      • Fortmann S.P.
      • Palaniappan L.P.
      Racial/ethnic differences in dyslipidemia patterns.
      ,
      • Karthikeyan G.
      • Teo K.K.
      • Islam S.
      • McQueen M.J.
      • Pais P.
      • Wang X.
      • Sato H.
      • Lang C.C.
      • Sitthi-Amorn C.
      • Pandey M.R.
      • et al.
      Lipid profile, plasma apolipoproteins, and risk of a first myocardial infarction among Asians: an analysis from the INTERHEART Study.
      ,
      • Huxley R.R.
      • Barzi F.
      • Lam T.H.
      • Czernichow S.
      • Fang X.
      • Welborn T.
      • Shaw J.
      • Ueshima H.
      • Zimmet P.
      • Jee S.H.
      • et al.
      Asia Pacific Cohort Studies Collaboration and the Obesity in Asia Collaboration.
      ), particularly those of Korean descent (
      • Frank A.T.
      • Zhao B.
      • Jose P.O.
      • Azar K.M.
      • Fortmann S.P.
      • Palaniappan L.P.
      Racial/ethnic differences in dyslipidemia patterns.
      ,
      • Goff Jr., D.C.
      • Bertoni A.G.
      • Kramer H.
      • Bonds D.
      • Blumenthal R.S.
      • Tsai M.Y.
      • Psaty B.M.
      Dyslipidemia prevalence, treatment, and control in the Multi-Ethnic Study of Atherosclerosis (MESA): gender, ethnicity, and coronary artery calcium.
      ,
      • Truesdale K.P.
      • Stevens J.
      • Cai J.
      Impact of body mass index levels on lipid abnormalities in Chinese Asians, American Blacks and American Whites: the People's Republic of China (PRC) and Atherosclerosis Risk in Communities (ARIC) studies.
      ).
      A growing body of evidence from candidate-gene studies also supports the presence of gene-obesity interactions in lipid profiles, as the effects of lipid-associated loci are modified by obesity traits such as BMI, waist circumference (WC), and waist-hip ratio (WHR) (
      • Turner P.R.
      • Talmud P.J.
      • Visvikis S.
      • Ehnholm C.
      • Tiret L.
      DNA polymorphisms of the apoprotein B gene are associated with altered plasma lipoprotein concentrations but not with perceived risk of cardiovascular disease: European Atherosclerosis Research Study.
      ,
      • Mailly F.
      • Fisher R.M.
      • Nicaud V.
      • Luong L.A.
      • Evans A.E.
      • Marques-Vidal P.
      • Luc G.
      • Arveiler D.
      • Bard J.M.
      • Poirier O.
      • et al.
      Association between the LPL-D9N mutation in the lipoprotein lipase gene and plasma lipid traits in myocardial infarction survivors from the ECTIM Study.
      ,
      • Gerdes C.
      • Fisher R.M.
      • Nicaud V.
      • Boer J.
      • Humphries S.E.
      • Talmud P.J.
      • Faergeman O.
      Lipoprotein lipase variants D9N and N291S are associated with increased plasma triglyceride and lower high-density lipoprotein cholesterol concentrations: studies in the fasting and postprandial states: the European Atherosclerosis Research Studies.
      ,
      • Jemaa R.
      • Elasmi M.
      • Naouali C.
      • Feki M.
      • Kallel A.
      • Souissi M.
      • Sanhaji H.
      • Hadj Taieb S.
      • Souheil O.
      • Kaabachi N.
      Apolipoprotein E polymorphism in the Tunisian population: frequency and effect on lipid parameters.
      ,
      • Stojkovic I.A.
      • Ericson U.
      • Rukh G.
      • Riddestrale M.
      • Romeo S.
      • Orho-Melander M.
      The PNPLA3 Ile148Met interacts with overweight and dietary intakes on fasting triglyceride levels.
      ). However, few studies have been performed at a genome-wide scale to identify SNPs influencing lipid levels based on obesity status, and even these studies were carried out using two-step analyses and genetic risk scores rather than exhaustive methods (
      • Surakka I.
      • Isaacs A.
      • Karssen L.C.
      • Laurila P.P.
      • Middelberg R.P.
      • Tikkanen E.
      • Ried J.S.
      • Lamina C.
      • Mangino M.
      • Igl W.
      • et al.
      A genome-wide screen for interactions reveals a new locus on 4p15 modifying the effect of waist-to-hip ratio on total cholesterol.
      ,
      • Lamina C.
      • Forer L.
      • Schonherr S.
      • Kollerits B.
      • Ried J.S.
      • Gieger C.
      • Peters A.
      • Wichmann H.E.
      • Kronenberg F.
      Evaluation of gene-obesity interaction effects on cholesterol levels: a genetic predisposition score on HDL-cholesterol is modified by obesity.
      ,
      • Justesen J.M.
      • Allin K.H.
      • Sandholt C.H.
      • Borglykke A.
      • Krarup N.T.
      • Grarup N.
      • Linneberg A.
      • Jorgensen T.
      • Hansen T.
      • Pedersen O.
      Interactions of lipid genetic risk scores with estimates of metabolic health in a Danish population.
      ,
      • Ali A.
      • Varga T.V.
      • Stojkovic I.A.
      • Schulz C.A.
      • Hallmans G.
      • Barroso I.
      • Poveda A.
      • Renstrom F.
      • Orho-Melander M.
      • Franks P.W.
      Do genetic factors modify the relationship between obesity and hypertriglyceridemia? Findings from the GLACIER and the MDC studies.
      ). Moreover, current findings of gene-obesity interactions are predominantly based on European populations (
      • Turner P.R.
      • Talmud P.J.
      • Visvikis S.
      • Ehnholm C.
      • Tiret L.
      DNA polymorphisms of the apoprotein B gene are associated with altered plasma lipoprotein concentrations but not with perceived risk of cardiovascular disease: European Atherosclerosis Research Study.
      ,
      • Mailly F.
      • Fisher R.M.
      • Nicaud V.
      • Luong L.A.
      • Evans A.E.
      • Marques-Vidal P.
      • Luc G.
      • Arveiler D.
      • Bard J.M.
      • Poirier O.
      • et al.
      Association between the LPL-D9N mutation in the lipoprotein lipase gene and plasma lipid traits in myocardial infarction survivors from the ECTIM Study.
      ,
      • Gerdes C.
      • Fisher R.M.
      • Nicaud V.
      • Boer J.
      • Humphries S.E.
      • Talmud P.J.
      • Faergeman O.
      Lipoprotein lipase variants D9N and N291S are associated with increased plasma triglyceride and lower high-density lipoprotein cholesterol concentrations: studies in the fasting and postprandial states: the European Atherosclerosis Research Studies.
      ,
      • Stojkovic I.A.
      • Ericson U.
      • Rukh G.
      • Riddestrale M.
      • Romeo S.
      • Orho-Melander M.
      The PNPLA3 Ile148Met interacts with overweight and dietary intakes on fasting triglyceride levels.
      ,
      • Surakka I.
      • Isaacs A.
      • Karssen L.C.
      • Laurila P.P.
      • Middelberg R.P.
      • Tikkanen E.
      • Ried J.S.
      • Lamina C.
      • Mangino M.
      • Igl W.
      • et al.
      A genome-wide screen for interactions reveals a new locus on 4p15 modifying the effect of waist-to-hip ratio on total cholesterol.
      ,
      • Lamina C.
      • Forer L.
      • Schonherr S.
      • Kollerits B.
      • Ried J.S.
      • Gieger C.
      • Peters A.
      • Wichmann H.E.
      • Kronenberg F.
      Evaluation of gene-obesity interaction effects on cholesterol levels: a genetic predisposition score on HDL-cholesterol is modified by obesity.
      ,
      • Justesen J.M.
      • Allin K.H.
      • Sandholt C.H.
      • Borglykke A.
      • Krarup N.T.
      • Grarup N.
      • Linneberg A.
      • Jorgensen T.
      • Hansen T.
      • Pedersen O.
      Interactions of lipid genetic risk scores with estimates of metabolic health in a Danish population.
      ,
      • Ali A.
      • Varga T.V.
      • Stojkovic I.A.
      • Schulz C.A.
      • Hallmans G.
      • Barroso I.
      • Poveda A.
      • Renstrom F.
      • Orho-Melander M.
      • Franks P.W.
      Do genetic factors modify the relationship between obesity and hypertriglyceridemia? Findings from the GLACIER and the MDC studies.
      ), and the G×E effects could have been underestimated if non-Caucasians are more susceptible to obesity-reactive dyslipidemia genes.
      We hypothesized that some genetic susceptibility loci modify the effects of obesity on the risk of dyslipidemia and assumed that these G×E loci might include both novel and known genes with different reactive effect sizes. We also hypothesized that the novel gene-obesity interactions underlying lipid parameters explain more of the total and genetic variances of each lipid trait than do the genetic loci with only marginal effects. G×E analyses are generally believed to suffer from weak study power rather than type 1 errors or false-positives (
      • Murcray C.E.
      • Lewinger J.P.
      • Conti D.V.
      • Thomas D.C.
      • Gauderman W.J.
      Sample size requirements to detect gene-environment interactions in genome-wide association studies.
      ). Partly due to this, replicating the findings has not been widely accepted as a prerequisite for reporting G×Es, but we believe replications in G×E studies are as important as in GWASs. To search for and replicate gene-obesity interactive SNPs influencing the risk of dyslipidemia, we carried out a genome-wide interaction scan (GWIS) of four independent genome cohort studies in Korea. We believe that the identification of genes interacting with potentially modifiable risk factors will facilitate knowledge-based approaches to personalize health guidelines according to an individual's genetic profile.

      MATERIALS AND METHODS

      Participants

      A total of 18,025 individuals from four independent Korean cohorts with a genome-wide set of variants were included in this study: 4,637 individuals from the Ansan cohort, 4,205 individuals from the Ansung cohort, 3,700 individuals from the urban cohort, and 5,483 individuals from the rural cohort. The baseline characteristics of each cohort are described in Table 1. Participants in the Ansan and Ansung cohorts were aged 40–69 years and recruited from industrialized suburban and rural areas in Korea. The urban and rural cohorts were population-based cohorts aged over 40 years and were respectively recruited from urban medical institutions and rural regions in Korea: Gangwha, Goryeong, Namwon, Pyeongchang, Wonju, and Yangpyeong. These cohorts are part of the Korean Genome and Epidemiology Study (KoGES), an ongoing population-based cohort study initiated in 2001 to understand chronic diseases in Korea. The research was performed according to the Declaration of Helsinki principles. The research protocol and data in this study were approved by the Institutional Review Board of Seoul National University (Institutional Review Board number: E1805-003-001).
      TABLE 1.Basic characteristics of the participants in each Korean cohort
      Reference SetReplication Set
      Ansan CohortAnsung CohortUrban CohortRural Cohort
      Participants4,2363,6063,4364,736
      Age (years)50.1 ± 7.756.9 ± 8.852.7 ± 8.260.2 ± 9.3
      Sex, male (%)2,136 (50.4)1,531 (42.5)1,494 (43.5)1,914 (40.4)
      BMI (kg/m2)24.5 ± 2.824.4 ± 3.223.9 ± 2.923.9 ± 3.2
      WC (cm)80.4 ± 7.985.6 ± 8.582.2 ± 8.883.5 ± 8.8
      HC (cm)94.1 ± 4.791.0 ± 5.595.3 ± 5.893.2 ± 6.6
      WHR0.85 ± 0.060.94 ± 0.060.86 ± 0.060.90 ± 0.06
      Total-C (mmol/l)5.11 ± 0.824.87 ± 0.825.13 ± 0.895.13 ± 0.96
      HDL-C (mmol/l)1.18 ± 0.241.17 ± 0.241.42 ± 0.341.18 ± 0.29
      LDL-C (mmol/l)3.18 ± 0.752.96 ± 0.753.08 ± 0.833.20 ± 0.88
      TG (mmol/l)1.64 ± 0.981.63 ± 1.001.37 ± 1.001.62 ± 1.09
      Remnant-C (mmol/l)0.75 ± 0.450.75 ± 0.460.63 ± 0.460.74 ± 0.50
      LDL-C was calculated using the Friedewald's formula for individuals with TG under 4.52 mmol/l; Remnant-C was determined as the level of Total-C minus HDL-C minus LDL-C. Detailed features stratified by obesity status into subgroups based on BMI, WC, and WHR are presented in supplemental Table S2.

      Measurements

      Data on health status, health-related behaviors, and medical and medication histories were collected through a standardized questionnaire. Trained experts at clinical centers performed anthropometric measurements, specimen collection, and laboratory tests. All participants provided informed consent for the baseline data and specimens; the detailed protocols are described in previous reports (
      • Cho Y.S.
      • Go M.J.
      • Kim Y.J.
      • Heo J.Y.
      • Oh J.H.
      • Ban H.J.
      • Yoon D.
      • Lee M.H.
      • Kim D.J.
      • Park M.
      • et al.
      A large-scale genome-wide association study of Asian populations uncovers genetic factors influencing eight quantitative traits.
      ,
      • Kim Y.J.
      • Go M.J.
      • Hu C.
      • Hong C.B.
      • Kim Y.K.
      • Lee J.Y.
      • Hwang J.Y.
      • Oh J.H.
      • Kim D.J.
      • Kim N.H.
      • et al.
      Large-scale genome-wide association studies in East Asians identify new genetic loci influencing metabolic traits.
      ). Total-C, HDL-C, and TG were measured using traditional enzymatic methods in blood samples drawn after an 8 h fast. For individuals with TG under 4.52 mmol/l, LDL-C was calculated using the Friedewald's formula (
      • Friedewald W.T.
      • Levy R.I.
      • Fredrickson D.S.
      Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge.
      ); remnant cholesterol (Remnant-C) was determined as the level of Total-C minus HDL-C minus LDL-C (
      • Varbo A.
      • Benn M.
      • Tybjaerg-Hansen A.
      • Jorgensen A.B.
      • Frikke-Schmidt R.
      • Nordestgaard B.G.
      Remnant cholesterol as a causal risk factor for ischemic heart disease.
      ,
      • Jørgensen A.B.
      • Frikke-Schmidt R.
      • West A.S.
      • Grande P.
      • Nordestgaard B.G.
      • Tybjaerg-Hansen A.
      Genetically elevated non-fasting triglycerides and calculated remnant cholesterol as causal risk factors for myocardial infarction.
      ). BMI and WHR were calculated using directly measured height, weight, WC, and hip circumference (HC).

      Phenotypes

      We defined dyslipidemia based on thresholds of high-risk CVD groups reported by the National Cholesterol Education Program (
      • National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III)
      Third Report of the National Cholesterol Education Program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III) final report.
      ): Total-C over 6.21 mmol/l or LDL-C over 4.14 mmol/l or TG over 2.26 mmol/l or the lowest quintile of HDL-C or the highest quintile of Remnant-C. Individuals with a medical history of hyperlipidemia or the use of any lipid-lowering drugs like statins were considered as cases of dyslipidemia in the statistical analyses. Obesity traits were defined using clinical guidelines of the National Institutes of Health and the Korean Society for the Study of Obesity (KSSO) (
      Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults–the evidence report. National Institutes of Health.
      ,
      • Seo M.H.
      • Lee W.Y.
      • Kim S.S.
      • Kang J.H.
      • Kang J.H.
      • Kim K.K.
      • Kim B.Y.
      • Kim Y.H.
      • Kim W.J.
      • Kim E.M.
      • et al.
      Committee of Clinical Practice Guidelines, Korean Society for the Study of Obesity (KSSO).
      ): overweight class 1 (BMI ≥23.0 kg/m2), overweight class 2 (BMI ≥25.0 kg/m2), obesity (BMI ≥30.0 kg/m2), abdominal obesity class 1 (WC >90 cm for males, 80 cm for females), abdominal obesity class 2 (WC >102 cm for males, 88 cm for females), and abdominal obesity due to WHR (WHR >0.90 for males, 0.85 for females). Individuals with inaccurate lipid levels or any history of cancer or diabetes were excluded from this study.

      Genotype information

      The following genome-wide dense SNP arrays were used to generate genetic data: the Affymetrix Genome-Wide Human SNP Array 5.0 for the Ansan and Ansung cohorts, the Affymetrix Genome-Wide Human SNP Array 6.0 for the urban cohort and part of the rural cohort (n = 1,816), and the Illumina HumanOmni1-Quad BeadChip for the rest of the rural cohort (n = 3,667). All variants violating Hardy-Weinberg equilibrium (P-value <1 × 10−6), with genotype call rates below 95%, and with minor allele frequency (MAF) values below 0.01 were excluded. After quality control, the remaining markers were imputed using the 1000 Genomes Project's haplotypes phase I in NCBI build 37 (GRCh37/hg19) of the Asian reference panels. SHAPEIT2 and IMPUTE2 were used for phasing and imputation, respectively (
      • Howie B.N.
      • Donnelly P.
      • Marchini J.
      A flexible and accurate genotype imputation method for the next generation of genome-wide association studies.
      ,
      • Delaneau O.
      • Zagury J.F.
      • Marchini J.
      Improved whole-chromosome phasing for disease and population genetic studies.
      ). Only SNPs with imputation quality scores higher than 0.6 were retained, yielding 4,780,608 SNPs for the Ansan and Ansung cohorts and 5,729,661 SNPs for the urban and rural cohorts. After comparing each cohort, 3,914,038 overlapping SNPs were selected as the final genetic markers.

      Statistical analyses for the GWIS

      The risk of dyslipidemia was adjusted for age, age2, sex, and each obesity trait one by one; the logarithm of the odds ratio (OR) of dyslipidemia was corrected using a logit model. Before the GWIS, we conducted marginal disease-gene association (DG) and environment-gene correlation (EG) tests for the 3.9 million SNPs. Genetic markers associated with both lipid and obesity traits (P-value <1 × 10−3) were excluded to reduce potential pleiotropy. Exhaustive scans, including case-control (CC), case-only (CO) (
      • Piegorsch W.W.
      • Weinberg C.R.
      • Taylor J.A.
      Non-hierarchical logistic models and case-only designs for assessing susceptibility in population-based case-control studies.
      ), and empirical Bayesian (EB) tests (
      • Mukherjee B.
      • Chatterjee N.
      Exploiting gene-environment independence for analysis of case-control studies: an empirical Bayes-type shrinkage estimator to trade-off between bias and efficiency.
      ), are direct one-step methods; CC tests the null hypothesis βG×E = 0 using a standard G×E model, CO checks the EG in affected individuals, and EB integrates the results from CC with those of CO. Two-step methods, on the other hand, comprise screening and hypothesis testing. We carried out emerging two-step approaches in parallel: DG|EB (DG1), DG|CC (DG2) (
      • Kooperberg C.
      • Leblanc M.
      Increasing the power of identifying gene x gene interactions in genome-wide association studies.
      ), EG|CC (EG2) (
      • Murcray C.E.
      • Lewinger J.P.
      • Gauderman W.J.
      Gene-environment interaction in genome-wide association studies.
      ), hybrid (H2) (
      • Murcray C.E.
      • Lewinger J.P.
      • Conti D.V.
      • Thomas D.C.
      • Gauderman W.J.
      Sample size requirements to detect gene-environment interactions in genome-wide association studies.
      ), Cocktail I (CT1), Cocktail II (CT2) (
      • Hsu L.
      • Jiao S.
      • Dai J.Y.
      • Hutter C.
      • Peters U.
      • Kooperberg C.
      Powerful cocktail methods for detecting genome-wide gene-environment interaction.
      ), and EDG×E (
      • Gauderman W.J.
      • Zhang P.
      • Morrison J.L.
      • Lewinger J.P.
      Finding novel genes by testing G x E interactions in a genome-wide association study.
      ). H2, Cocktail, and EDG×E adopt both DG and EG information to screen genetic markers in step-1; H2 uses these tests in parallel, Cocktail applies the information flexibly depending on the P-values of each DG and EG test, and EDG×E combines DG and EG statistics to generate new screening statistics. For testing in step-2, H2 and EDG×E adopt the results from CC; Cocktail applies EB when step-1 is based on DG and adopts CC if it uses EG to screen genetic markers.
      After finding novel G×E SNPs, we applied the standard genome-wide significance level (P-value <5 × 10−8) for exhaustive scans. We assumed a screening threshold of 1 × 10−4 for step-1 of the DG1, DG2, EG2, and H2 methods; for step-2, the subset of SNPs passing step-1 were tested at a more liberal cut-off (0.05 divided by the number of screened SNPs) (
      • Murcray C.E.
      • Lewinger J.P.
      • Conti D.V.
      • Thomas D.C.
      • Gauderman W.J.
      Sample size requirements to detect gene-environment interactions in genome-wide association studies.
      ). We applied weighted hypothesis testing in step-2 of the CT1, CT2, and EDG×E methods rather than testing only SNPs passing step-1; stepwise penalties according to the marginal P-value were applied for each SNP in step-2 (
      • Ionita-Laza I.
      • McQueen M.B.
      • Laird N.M.
      • Lange C.
      Genomewide weighted hypothesis testing in family-based association studies, with an application to a 100K scan.
      ). We removed one of a pair of the identified SNPs if the linkage disequilibrium (LD) was greater than 0.5 (variance inflation factor greater than 2) for informed LD pruning (LD clumping). We identified novel G×E loci using a reference set consisting of the Ansan and Ansung cohorts; all detected G×E loci were reconfirmed using a replication set composed of the urban and rural cohorts. We considered the effect size, magnitude of standard error, P-value, and the direction of effect to calculate the final results of meta-analyses. We used PLINK (
      • Purcell S.
      • Neale B.
      • Todd-Brown K.
      • Thomas L.
      • Ferreira M.A.
      • Bender D.
      • Maller J.
      • Sklar P.
      • de Bakker P.I.
      • Daly M.J.
      • et al.
      PLINK: a tool set for whole-genome association and population-based linkage analyses.
      ), METAL (
      • Willer C.J.
      • Li Y.
      • Abecasis G.R.
      METAL: fast and efficient meta-analysis of genomewide association scans.
      ), and R in the analyses.
      We carried out additional continuous and categorical analyses for TG to clarify the scale and categorization issues; we tested whether the different scales of phenotypes (continuous vs. dichotomous) and the different categorization methods (quantiles vs. clinical guidelines) had affected the results. We conducted GWISs using TG levels as continuous traits; TG, a nonnormalized trait, was transformed into a logarithmic scale and adjusted for age, age2, sex, and each obesity trait one by one. For categorical analyses, we used quantiles as cut-offs to determine outcomes and environmental factors instead of using clinical guidelines. We conducted GWISs 1) using the highest quintile of TG as a categorical phenotype and 2) using the highest quintile of BMI or WC or WHR as obesity traits. We also adjusted the risk of hypertriglyceridemia for age, age2, sex, and each obesity trait one by one. All the results of additional analyses were identified using the reference set and replicated using the replication set.

      Methods of evaluating impacts

      We estimated the genetic variances due to the genetic susceptibility SNPs using the simplified equation 2p(1 − p)[log(OR)]2, where p is the MAF of a variant and OR is the estimated odds ratio from a logistic regression model for marginal associations (
      • Witte J.S.
      • Visscher P.M.
      • Wray N.R.
      The contribution of genetic variants to disease depends on the ruler.
      ). The contribution of each G×E marker was estimated using 2p(1 − p)[log(ORG)]2/VP + 2ep[(1 − p) + 2p(e − 1)2][log(ORG×E)]2/VP. In this equation, e is the prevalence of an environmental factor, VP is the phenotypic variance, and ORG and ORG×E are estimated additive and gene-obesity interactive ORs from a logistic regression model for G×Es. We used GenABEL, the R package for genome-wide association analyses (
      • Aulchenko Y.S.
      • Ripke S.
      • Isaacs A.
      • van Duijn C.M.
      GenABEL: an R library for genome-wide association analysis.
      ), to estimate the total heritability of dyslipidemia from the Healthy Twin Study, a family-based cohort study in Korea (supplemental Table S2.e) (
      • Sung J.
      • Cho S-I.
      • Lee K.
      • Ha M.
      • Choi E-Y.
      • Choi J-S.
      • Kim H.
      • Kim J.
      • Hong K.S.
      • Kim Y.
      • et al.
      Healthy twin: a twin-family study of Korea–protocols and current status.
      ,
      • Gombojav B.
      • Song Y.M.
      • Lee K.
      • Yang S.
      • Kho M.
      • Hwang Y.C.
      • Ko G.
      • Sung J.
      The Healthy twin study, Korea updates: resources for omics and genome epidemiology studies.
      ). GCTA, an analysis tool for genome-wide complex traits (
      • Yang J.
      • Lee S.H.
      • Goddard M.E.
      • Visscher P.M.
      GCTA: a tool for genome-wide complex trait analysis.
      ), was also used to estimate the SNP-based heritability attributable to all GWAS variants genotyped on a microarray. To transform the estimate of variance explained on the observed scale to that on the underlying scale, we assessed the prevalence of dyslipidemia using data from the Korean National Health and Nutrition Examination Survey (KNHANES) (
      • Lee M.H.
      • Kim H.C.
      • Ahn S.V.
      • Hur N.W.
      • Choi D.P.
      • Park C.G.
      • Suh I.
      Prevalence of Dyslipidemia among Korean adults: Korea National Health and Nutrition Survey 1998–2005.
      ).

      RESULTS

      Table 1 shows the baseline characteristics of the participants in each Korean genome cohort. We observed the age, sex, obesity-related traits, and age- and sex-standardized plasma lipid levels of the cohorts; all features were stratified by obesity status into subgroups based on BMI, WC, and WHR (supplemental Tables S1, S2). We focused on the adjusted lipid concentrations to assess the trends of lipids in each obesity and abdominal obesity subgroup. As expected, age- and sex-adjusted lipid levels significantly worsened as the degree of obesity status increased in the combined Korean cohort (supplemental Fig. S1).
      We identified 55 SNPs showing genome-wide significant G×E effects on the risk of abnormal lipid profiles with at least one of the six obesity traits (supplemental Table S3). By conducting LD clumping based on the genetic contribution to the risk of dyslipidemia, we detected 20 gene-obesity interactions due to novel SNPs near SCN1A and SLC12A8 and to lipid-associated SNPs near APOA5, BUD13, ZNF259, and HMGCR that were reported in previous GWASs. Table 2 shows the marginal and gene-obesity interactive effects of the newly identified variants on the risk of dyslipidemia; we summarized the novel G×Es according to the discriminators of obesity traits, such as BMI, WC, and WHR. Figure 1 (supplemental Table S4) describes the risk of abnormal lipid profiles for each genetic and environmental factor; we estimated the OR as the ratio of the probability of dyslipidemia occurring in each exposed group (G ≠ 0 or E ≠ 0) to the probability in a nonexposed group (G = 0 and E = 0).
      TABLE 2.Novel gene-obesity interactive loci modifying the risk of dyslipidemia identified from the meta-analysis of the Korean cohorts
      GWASG×E Interaction
      TraitEnvironmentGeneMarkerCHRPositionMAFA1/A2ORD (P)ORG (P)ORG×E (P)Test
      Total-CBMICOL4A3BPrs77334365746664920.48C/T0.81 (3.88E-09)0.79 (4.26E-10)1.47 (7.32E-03)CT1/CT2/EDG×E
      WCHMGCRrs28784175746172620.48G/A0.81 (4.04E-09)0.76 (3.07E-08)1.18 (1.62E-02)CT1/CT2/EDG×E
      COL4A3BPrs77334365746664920.48C/T0.81 (5.38E-09)0.78 (1.97E-10)1.27 (1.13E-02)CT1/CT2/EDG×E
      WHRHMGCRrs77028955746128930.48G/A0.81 (7.29E-09)0.72 (3.15E-08)1.22 (5.40E-03)CT1/CT2/EDG×E
      HDL-CBMILOC101929680/SCN1Ars1189002821669432770.09G/T0.96 (4.19E-01)0.92 (8.78E-02)2.30 (2.79E-08)CO
      LDL-CBMILOC101928271rs116930762211400330.20C/T0.75 (1.25E-09)0.96 (6.54E-01)0.70 (6.79E-04)CT1/CT2/EDG×E
      WCANKDD1Brs77032825749069630.46A/C0.77 (7.72E-10)0.74 (6.12E-11)1.30 (2.45E-02)CT1/CT2/EDG×E
      WHRANKDD1Brs77032825749069630.46A/C0.77 (9.08E-10)0.70 (4.41E-08)1.19 (4.57E-02)EDG×E
      TGBMILOC105374079/SLC12A8rs7700880831248681730.06T/C0.97 (6.37E-01)0.89 (9.29E-02)2.70 (4.33E-08)CO
      BUD13rs1558860111166073680.22A/C1.55 (5.10E-35)1.76 (1.01E-14)0.80 (3.27E-03)EDG×E
      WCAPOA5rs651821111166625790.29C/T1.87 (1.56E-73)2.01 (5.00E-47)0.81 (4.06E-04)CT1/CT2/EDG×E
      BUD13rs918144111166338250.47T/C0.71 (4.22E-28)0.66 (1.47E-19)1.17 (7.75E-0)3EDG×E
      WHRBUD13rs180378111165889090.32A/G1.37 (8.00E-221.65 6.04E-180.76 1.26E-05DG1/CT1/CT2/EDG×E
      APOA5/ZPR1(ZNF259)rs2075291111166613920.08A/C1.98 (8.80E-42)2.23 (4.26E-21)0.82 (2.43E-02)CT1/CT2
      APOA5rs651821111166625790.29C/T1.86 (6.82E-73)2.26 (8.52E-41)0.74 (3.92E-06)DG1/CT1/CT2/EDG×E
      Remnant-CBMIBUD13rs7926828111165864230.31C/T1.35 (1.76E-18)1.33 (5.07E-16)1.37 (4.88E-02)EDG×E
      WCBUD13rs2075295111166284010.47C/T0.75 1.53E-270.73 2.21E-271.16 1.16E-02EDG×E
      WHRBUD13rs180378111165889090.32A/G1.35 (5.63E-26)1.47 (1.92E-15)0.84 (1.07E-03)EDG×E
      APOA5rs651821111166625790.29C/T1.82 (1.52E-87)1.99 (4.07E-41)0.82 (1.76E-04)CT1/CT2/EDG×E
      The results for each gene-obesity interaction were summarized according to the discriminators of obesity traits: BMI, WC, and WHR. Detailed results are presented in supplemental Table S3.
      Figure thumbnail gr1
      Fig. 1.Gene-obesity interactive effects on the risk of dyslipidemia. The bar plots on the lower side of each graph describe the OR as the ratio of the probability of dyslipidemia occurring in each exposed group (G ≠ 0 or E ≠ 0) to the probability in a nonexposed group (G = 0 and E = 0). The upper plots, on the other hand, show multiplicative effects of obesity traits for each genetic group. The figure above describes the estimated OR of each lipid abnormality due to the interplay between HMGCR and abdominal obesity based on WHR (A), LOC101928271 and overweight class 1 (B), BUD13 and abdominal obesity class 1 (C), BUD13 and abdominal obesity class 2 (D), LOC101929680/SCN1A and obesity (E), APOA5 and abdominal obesity based on WHR (F), and APOA5 and abdominal obesity based on WHR (G). Further details are provided in .
      We identified three novel SNPs interacting with obesity traits to modify the risk of abnormal elevation of Total-C: rs2878417, rs7702895, and rs7733436. In particular, COL4A3BP exhibited synergistic effects with BMI and WC on the risk of abnormalities in Total-C. For the interplay between HMGCR and WHR, the marginal odds ratio (ORD) was 0.81 (95% CI, 0.78–0.84); ORG and ORG×E were 0.72 (95% CI, 0.68–0.77) and 1.22 (95% CI, 1.13–1.30), respectively. As shown in Fig. 1A (supplemental Table S4.a), the multiplicative effect of abdominal obesity was 1.12 (95% CI, 0.96–1.31) for individuals with two wild-type alleles at rs7702895. The magnitude of the effect of abdominal obesity, however, increased with the number of minor alleles, with values of 1.46 (95% CI, 1.28–1.67) for heterozygous and 1.57 (95% CI, 1.26–1.95) for homozygous minor alleles.
      SCN1A marked by rs11890028 was detected as a novel locus interacting with obesity to change the risk of abnormal reduction of HDL-C. Although the marginal effect of this variant was negligible (P = 4.19 × 10−1), a noticeable G×E effect (P = 2.79 × 10−8) was observed in an exhaustive CO analysis. The estimated ORD, ORG, and ORG×E for the interplay between SCN1A and BMI were 0.96 (95% CI, 0.91–1.01), 0.92 (95% CI, 0.87–0.96), and 2.30 (95% CI, 1.98–2.67). Figure 1E (supplemental Table S4.e) describes the gene-obesity interactive effect on the risk of abnormal HDL-C reduction; the multiplicative effect for common homozygous or heterozygous and rare homozygous genotypes was 1.42 (95% CI, 1.22–1.65), 1.99 (95% CI, 1.57–2.52), and 6.24 (95% CI, 4.03–9.64), respectively.
      LOC101928271 showed antagonistic effects on the risk of abnormal elevation of LDL-C due to BMI. For the gene-obesity interaction, the estimated ORD, ORG, and ORG×E were 0.75 (95% CI, 0.71–0.78), 0.96 (95% CI, 0.88–1.05), and 0.70 (95% CI, 0.63–0.78), respectively. As shown in Fig. 1B (supplemental Table S4.b), the multiplicative effect of overweight class 1 was 1.00 (95% CI, 0.54–1.82) for rare homozygous genotypes. For common homozygous or heterozygous genotypes, on the other hand, obesity acted as a risk factor for abnormalities in LDL-C; the multiplicative effect was 1.82 (95% CI, 1.61–2.06) and 1.34 (95% CI, 1.12–1.61), respectively.
      We identified six novel SNPs with modifiable effects on the risk of abnormalities in TG due to obesity indices: rs1558860, rs180378, rs2075291, rs651821, rs918144 on chromosome 11, and rs77008808 on chromosome 3. BUD13 and APOA5 have previously been reported as lipid-associated loci, and the marginal effects of these loci on the risk of abnormal TG elevation were also markedly significant in this study. BUD13 marked by rs918144 showed a risky G×E effect attributable to WC; ORD, ORG, and ORG×E were 0.71 (95% CI, 0.69–0.73), 0.66 (95% CI, 0.64–0.70), and 1.17 (95% CI, 1.10–1.24), respectively. As shown in Fig. 1C (supplemental Table S4.c), the multiplicative effect of abdominal obesity for common homozygous or heterozygous and rare homozygous genotypes was 1.56 (95% CI, 1.40–1.76), 1.68 (95% CI, 1.52–1.86), and 2.12 (95% CI, 1.76–2.56), respectively. Conversely, APOA5 marked by rs651821 showed a protective G×E effect due to WHR; ORD, ORG, and ORGxE were 1.86 (95% CI, 1.80–1.93), 2.26 (95% CI, 2.13–2.41), and 0.74 (95% CI, 0.69–0.79), respectively. The multiplicative effect of abdominal obesity for each genotype is illustrated in Fig. 1F (supplemental Table S4.f), with values of 2.56 (95% CI, 2.20–2.99) for common homozygous, 1.89 (95% CI, 1.68–2.12) for heterozygous, and 1.39 (95% CI, 1.16–1.66) for rare homozygous genotypes.
      BUD13 and APOA5 also showed significant marginal associations with the risk of abnormal elevation of Remnant-C. BUD13 interacted with WC to modify the risk of dyslipidemia; ORD, ORG, and ORG×E were 0.75 (95% CI, 0.73–0.77), 0.73 (95% CI, 0.71–0.75), and 1.16 (95% CI, 1.10–1.23), respectively. The multiplicative effect of abdominal obesity class 2 is illustrated in Fig. 1D (supplemental Table S4.d), with values of 1.31 (95% CI, 1.17–1.47), 1.41 (95% CI, 1.28–1.55), and 1.77 (95% CI, 1.51–2.07) for individuals with common homozygous or heterozygous and rare homozygous genotypes, respectively. APOA5, on the other hand, had an antagonistic effect due to WHR; ORD, ORG, and ORG×E were 1.82 (95% CI, 1.77–1.88), 1.99 (95% CI, 1.89–2.09), and 0.82 (95% CI, 0.78–0.86), respectively. As shown in Fig. 1G (supplemental Table S4.g), the multiplicative effect of abdominal obesity was 2.20 (95% CI, 1.96–2.46) for individuals with two wild-type alleles at rs651821. The effects decreased as the number of minor alleles increased, with values of 1.85 (95% CI, 1.69–2.03) for heterozygous and 1.48 (95% CI, 1.28–1.71) for rare homozygous genotypes.
      We ascertained the identified G×Es from different points of view; Fig. 2 (supplemental Table S5) shows the trends of lipid levels due to changes in BMI for each subgroup stratified by the number of risk alleles of G×E SNPs. In normal weight (18.5 kg/m2 ≤ BMI < 25.0 kg/m2) individuals with no risk alleles (low-risk group), HDL-C levels decreased by 0.032 mmol/l (95% CI, 0.030–0.034 mmol/l) for each unit (1 kg/m2) increase in BMI; the HDL-C decrement was 0.039 mmol/L (95% CI, 0.035-0.043 mmol/L) for individuals with at least one risk allele (high-risk group) and 0.038 mmol/l (95% CI, 0.033–0.043 mmol/l) for the upper 50% of the high-risk group (higher-risk group). Differences in the HDL-C decrement for each genetic subgroup were far clearer in obese individuals; for individuals with BMI over 25 kg/m2, for example, the HDL-C decrements for the low-, high-, and higher-risk groups were 0.004 mmol/l (95% CI, 0.002–0.006 mmol/l), 0.014 mmol/l (95% CI, 0.010–0.018 mmol/l), and 0.017 mmol/l (95% CI, 0.012–0.022 mmol/l). Similarly, we ascertained that the TG increment was different for each genetic subgroup, and the trends of change were clearer in individuals with obesity.
      Figure thumbnail gr2
      Fig. 2.Changes in lipid levels due to increments in BMI for each risk group. The participants in each study population were classified into three groups by the number of risk alleles on G×E markers: the low-risk group (individuals with no risk alleles), the high-risk group (individuals with at least one risk allele), and the higher-risk group (the upper 50% of the high-risk group). The figure above describes the trends of lipid levels due to an increment of 1 kg/m2 in BMI for each subgroup. A: The differences in the decrement of HDL-C for each genetic subgroup were far clearer in the obese group than in the group with normal BMI. B: The differences in the increment of TG for each risk group were far clearer in the obese group than in the group with normal BMI; further details are presented in .
      We conducted GWISs using TG levels as continuous traits in supplemental Table S6 and newly found five genetic markers near APOA5, BUD13, and ZNF259 interacting with BMI. All the identified variants had strong marginal effects (P ≤ 9.70 × 10−25) on the traits of interest and were in LD with our previous finding, BUD13 marked by rs1558860, except for rs2041967 (r2 = 0.20). We could not detect any interactions of G×E markers with WC or WHR. In supplemental Table S7, we compared the identified gene-obesity interactive variants of hypertriglyceridemia defined by two different methods: by clinical guidelines and by the quantile distribution in our data. All the interactions of loci (APOA5 and BUD13) with WC or WHR were replicated with our previous findings. On the other hand, we could not detect any interactions of G×E markers with BMI; only the variants, previously detected with multiple analytical methods for testing G×Es (Table 2), were consistently identified. Similarly, we identified only one G×E SNP of dyslipidemia, located on APOA5, by using the alternative definition of obesity, the highest quintile of BMI or WC or WHR (supplemental Table S8); we could not find any loci interacting with BMI or WC.
      Table 3 shows the contributions of marginal associations and gene-obesity interactions to abnormal lipid profiles; we present the proportion of total heritability for each lipid explained by GWAS-identified SNPs, novel G×E loci, and the combined set of both lipid-associated and gene-obesity interactive variants (total genetic impact). The total and SNP-based heritabilities of the risk of abnormalities in Total-C were approximately 35.5% and 17.7–24.9%, respectively, after adjusting the risk of dyslipidemia by age, age2, and sex. The genetic contributions increased when we considered both marginal associations and gene-obesity interactions, with differences between the GWAS-identified and total genetic impact of 1.1–1.9%. The total and SNP-based heritabilities of the risk of LDL-C abnormalities, on the other hand, were approximately 31.7% and 17.2–25.6%, respectively. For each obesity trait, the total genetic contributions including gene-obesity interactions to the risk of dyslipidemia were 0.9–2.4% higher than the marginal impact due only to direct associations.
      TABLE 3.Contributions of gene-obesity interactive loci to abnormal lipid profiles
      TraitHeritability (%) (SNP-Based Heritability)EnvironmentGeneMarkerA1/A2Contribution (%) of Genetic VariantsTotal Genetic Contribution in KOR Population
      Contribution of GWAS-Identified Loci for KOR (Number of SNPs)Additional Contribution of Gene-Obesity Interaction (Allele Frequency in Each Ethnic Group
      Allele frequencies of each genetic marker in various ethnic groups were referred to the 1000 Genomes Project database (GRCh37/hg19).
      )
      KOREASSASEURAMRAFR
      Total-C35.5 (17.7–24.9)BMICOL4A3BPrs7733436C/T9.5 (20)1.2 (0.48)1.2 (0.49)1.1 (0.41)1.2 (0.61)1.2 (0.54)0.3 (0.06)10.7
      WCHMGCRrs2878417G/A8.4 (17)1.3 (0.48)1.3 (0.50)1.2 (0.40)1.3 (0.58)1.3 (0.53)1.3 (0.57)10.8
      COL4A3BPrs7733436C/T1.1 (0.48)1.2 (0.49)1.1 (0.41)1.2 (0.61)1.2 (0.54)0.2 (0.06)
      WHRHMGCRrs7702895G/A7.2 (15)1.9 (0.48)1.9 (0.49)1.8 (0.40)1.9 (0.56)1.9 (0.46)0.2 (0.03)9.1
      HDL-C34.8 (18.4–26.2)BMILOC101929680/SCN1Ars11890028G/T19.5 (31)0.3 (0.09)0.4 (0.13)0.8 (0.28)0.8 (0.29)0.6 (0.21)0.4 (0.16)19.8
      LDL-C31.7 (17.2–25.6)BMILOC101928271rs11693076C/T5.5 (11)0.9 (0.20)0.9 (0.20)1.0 (0.22)1.6 (0.46)1.6 (0.48)1.7 (0.85)6.4
      WCANKDD1Brs7703282A/C6.3 (14)1.8 (0.46)1.8 (0.48)1.7 (0.43)1.8 (0.62)1.8 (0.54)0.4 (0.06)8.1
      WHRANKDD1Brs7703282A/C5.6 (12)2.4 (0.46)2.4 (0.48)2.4 (0.43)2.3 (0.62)2.4 (0.54)0.5 (0.06)8.0
      TG38.3 (18.4–26.4)BMILOC105374079/SLC12A8rs77008808T/C48.2 (53)0.3 (0.06)0.3 (0.07)0.2 (0.04)0.0 (0.00)0.0 (0.00)0.0 (0.00)52.7
      BUD13rs1558860A/C4.2 (0.22)4.5 (0.24)3.8 (0.19)2.0 (0.09)2.8 (0.13)0.2 (0.01)
      WCAPOA5rs651821C/T36.6 (40)7.4 (0.29)7.3 (0.29)5.5 (0.19)2.6 (0.08)4.5 (0.15)4.8 (0.16)47.1
      BUD13rs918144T/C3.2 (0.47)3.0 (0.38)3.2 (0.56)3.2 (0.48)3.1 (0.60)2.9 (0.68)
      WHRBUD13rs180378A/G39.3 (48)4.6 (0.32)4.7 (0.32)5.4 (0.60)5.3 (0.43)5.2 (0.40)5.4 (0.46)58.0
      APOA5/ZPR1(ZNF259)rs2075291A/C3.4 (0.08)1.8 (0.04)0.5 (0.01)0.0 (0.00)0.0 (0.00)0.0 (0.00)
      APOA5rs651821C/T10.7 (0.29)10.7 (0.29)7.9 (0.19)3.8 (0.08)6.6 (0.15)6.9 (0.16)
      Remnant-C48.6 (11.3–14.2)BMIBUD13rs7926828C/T35.8 (55)1.3 (0.31)1.3 (0.32)1.5 (0.52)1.0 (0.22)1.1 (0.25)1.3 (0.31)37.0
      WCBUD13rs2075295C/T36.7 (56)1.8 (0.47)1.7 (0.38)1.8 (0.46)1.4 (0.27)1.8 (0.44)1.6 (0.34)38.5
      WHRBUD13rs180378A/G38.5 (59)2.4 (0.32)2.4 (0.32)2.7 (0.60)2.7 (0.43)2.7 (0.40)2.8 (0.46)47.8
      APOA5rs651821C/T6.9 (0.29)6.8 (0.29)5.1 (0.19)2.4 (0.08)4.2 (0.15)4.5 (0.16)
      KOR, Korean; EAS, East Asian; SAS, South Asian; EUR, European; AMR, American; AFR, African.
      a Allele frequencies of each genetic marker in various ethnic groups were referred to the 1000 Genomes Project database (GRCh37/hg19).
      The contributions of the combined set of both GWAS-identified and GxE variants were markedly higher when several independent gene-obesity interactive loci were present for each pair of lipid traits and environmental factors. Figure 3A, B, and C (Table 3) present the risk of abnormal elevation of TG. Genetic factors accounted for approximately 38.3% of the total variance of the risk of abnormalities in TG after adjusting the risk by age, age2, and sex. Genetic markers located on the genome-wide dense SNP microarray accounted for 18.4–26.4% of the overall variance for the risk of hypertriglyceridemia. Approximately 36.6% of the total heritability was due to 40 independent GWAS-identified SNPs only; the genetic contribution increased to 47.1% when we considered the interactions of APOA5 or BUD13 with WC. Similarly, the total genetic impact increased from 39.3% to 58.0% when we considered both marginal associations and newly found genetic interactions attributable to WHR. For Caucasians, the additional TG heritability due to the interactions of G×E variants with WC or WHR was 5.8% and 9.1%; the gain was 10.6% and 18.7% for Koreans, respectively.
      Figure thumbnail gr3
      Fig. 3.Contributions of marginal associations and gene-obesity interactions to the risk of dyslipidemia. The pie plots describe the proportion of phenotypic variation attributable to the overall genetic variation (total heritability), genetic markers assayed by SNP arrays (SNP-based heritability), and the combined set of both GWAS-identified and novel G×E variants. The bar plots, on the other hand, show the proportion of genetic variation explained by marginal and gene-obesity interactive effects. Parts (A) to (C) describe the genetic contributions to abnormal TG due to the interplay between genes and obesity indices classified by BMI (A), WC (B), and WHR (C). D: The genetic contributions to abnormal Remnant-C due to the interplay between genes and obesity indices classified by WHR; further details are presented in .
      The genetic contributions to the risk of abnormal elevation of Remnant-C are described in Fig. 3D (Table 3). Genetic factors explained approximately 48.6% of the total variance after adjusting for age, age2, and sex. Genotyped loci on the SNP microarray accounted for 11.3–14.2% of the overall variance for Remnant-C. Approximately 38.5% of the total heritability was explained by 59 independent GWAS-identified SNPs only. When both marginal associations and interactions of APOA5 or BUD13 with WHR that modify the risk of abnormal Remnant-C were considered, the genetic contribution increased to 47.8%; the difference between marginal and total genetic impact was 9.3%. For Caucasians, on the other hand, the additional heritability for Remnant-C was just 5.1%.

      DISCUSSION

      One of the main purposes of human genome studies is to personalize treatment and health guidelines according to an individual's genetic constitution. GWISs are approaches intended for achieving this end, particularly when genetic loci interacting with modifiable risk factors are examined at a genome-wide level. Such studies permit the identification of higher- or lower-risk individuals depending on changes in known risk factors. In this study, we identified novel and known genes interacting with obesity indices to modify the risk of dyslipidemia. We also replicated our findings using independent genome cohorts and assessed how much phenotypic variance or heritability was additionally explained by considering the gene-obesity interactions.
      Our study focused on increasing power to detect gene-obesity interactions by applying a variety of strategies for testing G×Es. We carried out emerging exhaustive scans and two-step methods in parallel because each analytical model provided differential power to detect G×Es, mainly according to marginal genetic and G×E effects. We tested interactions of SNPs at a genome-wide scale with several obesity traits, including Korean-specific parameters defined by additional ranges of BMI and WC. Besides, we adopted liberal cut-offs and stepwise penalties due to marginal P-values as well as the standard genome-wide significance level to find gene-obesity interactive loci influencing the risk of dyslipidemia. Type 1 errors are generally considered to be less problematic than possible underpowered findings (
      • Murcray C.E.
      • Lewinger J.P.
      • Conti D.V.
      • Thomas D.C.
      • Gauderman W.J.
      Sample size requirements to detect gene-environment interactions in genome-wide association studies.
      ); it is recommended to use multiple models for G×Es. Further verification can be done by replication and stratified analyses for candidate G×E regions.
      Our findings reveal a genome-wide set of variants with a wide range of marginal effects on the risk of dyslipidemia. We identified novel G×E markers near SCN1A and SLC12A8 with little or no direct association with lipid parameters as well as gene-obesity interactions related to lipid-associated loci reported in previous GWASs on lipids: APOA5, BUD13, ZNF259, and HMGCR. We identified SCN1A and SLC12A8 through exhaustive CO analyses, while all other G×Es due to lipid-associated loci were detected using two-step methods due to the marginal effects of each locus in the first step. These trends are consistent with the results of an earlier simulation study of statistical power for G×E detection, which showed that exhaustive CO analysis is more powerful than other two-step methods when the marginal effects of genetic variants are small (
      • Gauderman W.J.
      • Zhang P.
      • Morrison J.L.
      • Lewinger J.P.
      Finding novel genes by testing G x E interactions in a genome-wide association study.
      ). To our knowledge, SCN1A and SLC12A8 have not been previously associated with any lipid parameters.
      We replicated the novel findings in four Korean genome cohorts; one strength of using cohorts formulated on identical protocols is the ability to examine gene-obesity interactions with high-quality health outcomes, genetic and environmental factors. In addition, conducting meta-analyses with the independent Korean cohorts permitted the estimation of more precise effects of susceptibility loci interacting with obesity traits. We also classified individuals in this study into three groups according to the number of risk alleles at G×E loci and compared the changes in lipid levels when BMI, WC, and WHR increased by one unit between the three groups. This comparison reconfirmed the identified gene-obesity interactions from different points of view, as the changes in lipids due to the elevation of obesity indices worsened as the number of risk alleles increased.
      Although generating interesting findings, our approaches for testing gene-obesity interactive effects on lipid profiles are not free of limitations. Our study did not include G×Es due to loci marked by rare variants (MAF <0.01) and other essential obesity indices, such as body fat percentage and visceral fat level. We focused only on the G×E effects due to a set of common variants because our study populations did not include enough information for rare genetic variants. In addition, current analytical methods do not provide adequate power for detecting G×E effects on rare variants; the latest approaches using gene-set analyses and sum of powered score tests are also limited to testing G×Es at a variant-by-variant level (
      • Su Y.R.
      • Di C.Z.
      • Hsu L.
      • Genetics C.
      A unified powerful set-based test for sequencing data analysis of GxE interactions.
      ,
      • Yang T.
      • Chen H.
      • Tang H.
      • Li D.
      • Wei P.
      A powerful and data-adaptive test for rare-variant-based gene-environment interaction analysis.
      ). Some rare variants in NPC1L1, however, are known to be associated with cholesterol absorption and LDL-C levels (
      • Cohen J.C.
      • Pertsemlidis A.
      • Fahmi S.
      • Esmail S.
      • Vega G.L.
      • Grundy S.M.
      • Hobbs H.H.
      Multiple rare variants in NPC1L1 associated with reduced sterol absorption and plasma low-density lipoprotein levels.
      ), and the interplay between rare variants and obesity traits might play a role in regulating lipid levels.
      Furthermore, our new findings primarily concern G×Es based on indirect obesity indices, as BMI, WC, and WHR are surrogate measures of overall and abdominal adiposity (
      • Després J.P.
      Body fat distribution and risk of cardiovascular disease: an update.
      ). Other vital indices measuring adipose tissue distribution, such as body fat percentage and visceral fat level, have been reported to be related to higher risks of CVD and metabolic syndrome in large-scale epidemiological studies (
      • Yusuf S.
      • Hawken S.
      • Ounpuu S.
      • Bautista L.
      • Franzosi M.G.
      • Commerford P.
      • Lang C.C.
      • Rumboldt Z.
      • Onen C.L.
      • Lisheng L.
      • et al.
      Obesity and the risk of myocardial infarction in 27,000 participants from 52 countries: a case-control study.
      ,
      • Després J.P.
      • Lemieux I.
      Abdominal obesity and metabolic syndrome.
      ,
      • Konieczna J.
      • Abete I.
      • Galmes A.M.
      • Babio N.
      • Colom A.
      • Zulet M.A.
      • Estruch R.
      • Vidal J.
      • Toledo E.
      • Diaz-Lopez A.
      • et al.
      PREDIMED-Plus Investigators.
      ) but were not addressed in this study. Considering the evidence that cardio-metabolic abnormalities are more closely linked with body shape and fat distribution than with conventional obesity indices (
      • Coutinho T.
      • Goel K.
      • Correa de Sa D.
      • Kragelund C.
      • Kanaya A.M.
      • Zeller M.
      • Park J.S.
      • Kober L.
      • Torp-Pedersen C.
      • Cottin Y.
      • et al.
      Central obesity and survival in subjects with coronary artery disease: a systematic review of the literature and collaborative analysis with individual subject data.
      ,
      • Dallongeville J.
      • Bhatt D.L.
      • Steg P.H.
      • Ravaud P.
      • Wilson P.W.
      • Eagle K.A.
      • Goto S.
      • Mas J.L.
      • Montalescot G.
      • Investigators R.R.
      Relation between body mass index, waist circumference, and cardiovascular outcomes in 19,579 diabetic patients with established vascular disease: the REACH Registry.
      ), interactions of genetic susceptibility loci of dyslipidemia with indicators that directly reflect adiposity warrant greater concern.
      We used dyslipidemia as a dichotomous outcome variable instead of using quantitative lipid levels because the statistical power would be improved by the following reasons. 1) Some analytical approaches for testing G×Es use dichotomous traits as a prerequisite, such as CO and EB, and these methods are optimized to detect genetic markers with weak marginal effects. 2) Some methods, such as CC, DG2, EG2, H2, and EDG×E, could be extended to quantitative outcomes and these analyses have more power for G×E markers with strong marginal genetic effects (
      • Gauderman W.J.
      • Zhang P.
      • Morrison J.L.
      • Lewinger J.P.
      Finding novel genes by testing G x E interactions in a genome-wide association study.
      ). 3) In the analysis of statistical interactions, quantitative scales are sensitive to distribution; it is well-known that some nonnormal distribution would generate false-positive results of interactions. Despite concerns about the scale issues, our approaches could cover a broader range of gene-obesity interactive variants, having strong marginal effects as well as weak marginal effects, than the G×E analyses using TG levels quantitatively (supplemental Table S6). For GWISs, it is more important to use multiple analytical models as possible to generate a consistent and powerful result.
      For categorical analyses, the cut-offs were decided by clinical guidelines for managing hyperlipidemia and obesity. For HDL-C, on the other hand, we used quantiles because the clinical cut-off points (HDL-C <1.03 mmol/l for males, 1.29 mmol/l for females) resulted in too many dyslipidemia cases (46.8%) in our study population; the arbitrary cut-offs could affect the results for interactions. To clarify the issue, we conducted GWISs using two different methods of categorization: by clinical guidelines and by the quantile distribution in our datasets. Compared with using quantiles to determine dyslipidemia and obesity traits, the categorical G×E analyses using clinical thresholds could find out a more extensive range of gene-obesity interactions (supplemental Tables S7, S8). The clinical cut-off points, commonly accepted and ascertained by several epidemiological studies, were more appropriate to detect interactions at a genome-wide scale than quantiles dividing the study population into equal-sized bins for phenotypes.
      Our ability to extend these novel findings from Korean populations to other ethnic groups is limited by differences in the MAFs of genetic markers, distributions of obesity traits, and prevalences of each lipid abnormality. Our results were estimated and reconfirmed in four independent Korean cohorts sharing phenotyping and genotyping protocols, and the identified gene-obesity interactions in the risk of dyslipidemia might not be supported when racial differences in lipid profiles and the distribution of genetic and environmental factors are considered. ZNF259 marked by rs2075291, for example, could be a useful therapeutic target for managing TG in the Korean population; the ZNF259-WHR interactive impact on the risk of hypertriglyceridemia was 3.4%. This genetic variant, however, is not a suitable target for the other ethnic groups; the minor allele of rs2075291 is infrequent for South Asians and too rare for Europeans, Americans, and Africans (Table 3). Conversely, some minor findings in this study could be useful for managing lipids and obesity traits in other ethnic groups.
      Many human traits and complex diseases are known to be a consequence of both genetic and environmental factors, and thus G×E analyses may hold the key to further insights on disease biology and the development of better prediction models. Our exploration of G×Es at a genome-wide level in Koreans revealed novel genetic susceptibility loci of dyslipidemia interacting with modifiable obesity traits. Our results were replicated in independent genome cohorts and confirmed by comparing changes in lipid levels due to an increment of obesity for each genetic subgroup. Compared with lipid-associated loci with only marginal effects, the inclusion of gene-obesity interactive loci clearly explained more of the total and genetic variances of each lipid. Based on the different allele frequencies between Caucasians and Asians, besides, Asians have a higher risk of dyslipidemia, particularly for TG, even with a small increase in obesity indices. These newly identified gene-obesity interactions can be used to classify individuals into higher- or lower-risk groups and to develop personalized guidelines for managing lipid and obesity traits according to the genetic constitution.

      Acknowledgments

      This study was provided with bioresources from National Biobank of Korea, Centers for Disease Control and Prevention, Ministry for Health and Welfare, Republic of Korea. Data in this study were from the Korean Genome and Epidemiology Study (4851-302), National Research Institute of Health, Centers for Disease Control and Prevention, Ministry for Health and Welfare, Republic of Korea.

      REFERENCES

        • Teslovich T.M.
        • Musunuru K.
        • Smith A.V.
        • Edmondson A.C.
        • Stylianou I.M.
        • Koseki M.
        • Pirruccello J.P.
        • Ripatti S.
        • Chasman D.I.
        • Willer C.J.
        • et al.
        Biological, clinical and population relevance of 95 loci for blood lipids.
        Nature. 2010; 466: 707-713
        • Asselbergs F.W.
        • Guo Y.
        • van Iperen E.P.
        • Sivapalaratnam S.
        • Tragante V.
        • Lanktree M.B.
        • Lange L.A.
        • Almoguera B.
        • Appelman Y.E.
        • Barnard J.
        • et al.
        Large-scale gene-centric meta-analysis across 32 studies identifies multiple lipid loci.
        Am. J. Hum. Genet. 2012; 91: 823-838
        • Willer C.J.
        • Schmidt E.M.
        • Sengupta S.
        • Peloso G.M.
        • Gustafsson S.
        • Kanoni S.
        • Ganna A.
        • Chen J.
        • Buchkovich M.L.
        • Mora S.
        • et al.
        Discovery and refinement of loci associated with lipid levels.
        Nat. Genet. 2013; 45: 1274-1283
        • Surakka I.
        • Horikoshi M.
        • Magi R.
        • Sarin A.P.
        • Mahajan A.
        • Lagou V.
        • Marullo L.
        • Ferreira T.
        • Miraglio B.
        • Timonen S.
        • et al.
        The impact of low-frequency and rare variants on lipid levels.
        Nat. Genet. 2015; 47: 589-597
        • Below J.E.
        • Parra E.J.
        • Gamazon E.R.
        • Torres J.
        • Krithika S.
        • Candille S.
        • Lu Y.
        • Manichakul A.
        • Peralta-Romero J.
        • Duan Q.
        • et al.
        Meta-analysis of lipid-traits in Hispanics identifies novel loci, population-specific effects, and tissue-specific enrichment of eQTLs.
        Sci. Rep. 2016; 6: 19429
        • Spracklen C.N.
        • Chen P.
        • Kim Y.J.
        • Wang X.
        • Cai H.
        • Li S.
        • Long J.
        • Wu Y.
        • Wang Y.X.
        • Takeuchi F.
        • et al.
        Association analyses of East Asian individuals and trans-ancestry analyses with European individuals reveal new loci associated with cholesterol and triglyceride levels.
        Hum. Mol. Genet. 2017; 26: 1770-1784
        • Lu X.
        • Peloso G.M.
        • Liu D.J.
        • Wu Y.
        • Zhang H.
        • Zhou W.
        • Li J.
        • Tang C.S.
        • Dorajoo R.
        • Li H.
        • et al.
        Exome chip meta-analysis identifies novel loci and East Asian-specific coding variants that contribute to lipid levels and coronary artery disease.
        Nat. Genet. 2017; 49: 1722-1730
        • Liu D.J.
        • Peloso G.M.
        • Yu H.
        • Butterworth A.S.
        • Wang X.
        • Mahajan A.
        • Saleheen D.
        • Emdin C.
        • Alam D.
        • Alves A.C.
        • et al.
        Exome-wide association study of plasma lipids in >300,000 individuals.
        Nat. Genet. 2017; 49: 1758-1766
        • Southam L.
        • Gilly A.
        • Suveges D.
        • Farmaki A.E.
        • Schwartzentruber J.
        • Tachmazidou I.
        • Matchan A.
        • Rayner N.W.
        • Tsafantakis E.
        • Karaleftheri M.
        • et al.
        Whole genome sequencing and imputation in isolated populations identify genetic associations with medically-relevant complex traits.
        Nat. Commun. 2017; 8: 15606
        • Davis J.P.
        • Huyghe J.R.
        • Locke A.E.
        • Jackson A.U.
        • Sim X.
        • Stringham H.M.
        • Teslovich T.M.
        • Welch R.P.
        • Fuchsberger C.
        • Narisu N.
        • et al.
        Common, low-frequency, and rare genetic variants associated with lipoprotein subclasses and triglyceride measures in Finnish men from the METSIM study.
        PLoS Genet. 2017; 13: e1007079
        • Kanai M.
        • Akiyama M.
        • Takahashi A.
        • Matoba N.
        • Momozawa Y.
        • Ikeda M.
        • Iwata N.
        • Ikegawa S.
        • Hirata M.
        • Matsuda K.
        • et al.
        Genetic analysis of quantitative traits in the Japanese population links cell types to complex human diseases.
        Nat. Genet. 2018; 50: 390-400
        • Frank A.T.
        • Zhao B.
        • Jose P.O.
        • Azar K.M.
        • Fortmann S.P.
        • Palaniappan L.P.
        Racial/ethnic differences in dyslipidemia patterns.
        Circulation. 2014; 129: 570-579
        • Hyre A.D.
        • Muntner P.
        • Menke A.
        • Raggi P.
        • He J.
        Trends in ATP-III-defined high blood cholesterol prevalence, awareness, treatment and control among US adults.
        Ann. Epidemiol. 2007; 17: 548-555
        • Karthikeyan G.
        • Teo K.K.
        • Islam S.
        • McQueen M.J.
        • Pais P.
        • Wang X.
        • Sato H.
        • Lang C.C.
        • Sitthi-Amorn C.
        • Pandey M.R.
        • et al.
        Lipid profile, plasma apolipoproteins, and risk of a first myocardial infarction among Asians: an analysis from the INTERHEART Study.
        J. Am. Coll. Cardiol. 2009; 53: 244-253
        • Huxley R.R.
        • Barzi F.
        • Lam T.H.
        • Czernichow S.
        • Fang X.
        • Welborn T.
        • Shaw J.
        • Ueshima H.
        • Zimmet P.
        • Jee S.H.
        • et al.
        Asia Pacific Cohort Studies Collaboration and the Obesity in Asia Collaboration.
        Circulation. 2011; 124 (Isolated low levels of high-density lipoprotein cholesterol are associated with an increased risk of coronary heart disease: an individual participant data meta-analysis of 23 studies in the Asia-Pacific region): 2056-2064
        • Goff Jr., D.C.
        • Bertoni A.G.
        • Kramer H.
        • Bonds D.
        • Blumenthal R.S.
        • Tsai M.Y.
        • Psaty B.M.
        Dyslipidemia prevalence, treatment, and control in the Multi-Ethnic Study of Atherosclerosis (MESA): gender, ethnicity, and coronary artery calcium.
        Circulation. 2006; 113: 647-656
        • Truesdale K.P.
        • Stevens J.
        • Cai J.
        Impact of body mass index levels on lipid abnormalities in Chinese Asians, American Blacks and American Whites: the People's Republic of China (PRC) and Atherosclerosis Risk in Communities (ARIC) studies.
        Atherosclerosis. 2011; 218: 517-523
        • Turner P.R.
        • Talmud P.J.
        • Visvikis S.
        • Ehnholm C.
        • Tiret L.
        DNA polymorphisms of the apoprotein B gene are associated with altered plasma lipoprotein concentrations but not with perceived risk of cardiovascular disease: European Atherosclerosis Research Study.
        Atherosclerosis. 1995; 116: 221-234
        • Mailly F.
        • Fisher R.M.
        • Nicaud V.
        • Luong L.A.
        • Evans A.E.
        • Marques-Vidal P.
        • Luc G.
        • Arveiler D.
        • Bard J.M.
        • Poirier O.
        • et al.
        Association between the LPL-D9N mutation in the lipoprotein lipase gene and plasma lipid traits in myocardial infarction survivors from the ECTIM Study.
        Atherosclerosis. 1996; 122: 21-28
        • Gerdes C.
        • Fisher R.M.
        • Nicaud V.
        • Boer J.
        • Humphries S.E.
        • Talmud P.J.
        • Faergeman O.
        Lipoprotein lipase variants D9N and N291S are associated with increased plasma triglyceride and lower high-density lipoprotein cholesterol concentrations: studies in the fasting and postprandial states: the European Atherosclerosis Research Studies.
        Circulation. 1997; 96: 733-740
        • Jemaa R.
        • Elasmi M.
        • Naouali C.
        • Feki M.
        • Kallel A.
        • Souissi M.
        • Sanhaji H.
        • Hadj Taieb S.
        • Souheil O.
        • Kaabachi N.
        Apolipoprotein E polymorphism in the Tunisian population: frequency and effect on lipid parameters.
        Clin. Biochem. 2006; 39: 816-820
        • Stojkovic I.A.
        • Ericson U.
        • Rukh G.
        • Riddestrale M.
        • Romeo S.
        • Orho-Melander M.
        The PNPLA3 Ile148Met interacts with overweight and dietary intakes on fasting triglyceride levels.
        Genes Nutr. 2014; 9: 388
        • Surakka I.
        • Isaacs A.
        • Karssen L.C.
        • Laurila P.P.
        • Middelberg R.P.
        • Tikkanen E.
        • Ried J.S.
        • Lamina C.
        • Mangino M.
        • Igl W.
        • et al.
        A genome-wide screen for interactions reveals a new locus on 4p15 modifying the effect of waist-to-hip ratio on total cholesterol.
        PLoS Genet. 2011; 7: e1002333
        • Lamina C.
        • Forer L.
        • Schonherr S.
        • Kollerits B.
        • Ried J.S.
        • Gieger C.
        • Peters A.
        • Wichmann H.E.
        • Kronenberg F.
        Evaluation of gene-obesity interaction effects on cholesterol levels: a genetic predisposition score on HDL-cholesterol is modified by obesity.
        Atherosclerosis. 2012; 225: 363-369
        • Justesen J.M.
        • Allin K.H.
        • Sandholt C.H.
        • Borglykke A.
        • Krarup N.T.
        • Grarup N.
        • Linneberg A.
        • Jorgensen T.
        • Hansen T.
        • Pedersen O.
        Interactions of lipid genetic risk scores with estimates of metabolic health in a Danish population.
        Circ Cardiovasc Genet. 2015; 8: 465-472
        • Ali A.
        • Varga T.V.
        • Stojkovic I.A.
        • Schulz C.A.
        • Hallmans G.
        • Barroso I.
        • Poveda A.
        • Renstrom F.
        • Orho-Melander M.
        • Franks P.W.
        Do genetic factors modify the relationship between obesity and hypertriglyceridemia? Findings from the GLACIER and the MDC studies.
        Circ Cardiovasc Genet. 2016; 9: 162-171
        • Murcray C.E.
        • Lewinger J.P.
        • Conti D.V.
        • Thomas D.C.
        • Gauderman W.J.
        Sample size requirements to detect gene-environment interactions in genome-wide association studies.
        Genet. Epidemiol. 2011; 35: 201-210
        • Cho Y.S.
        • Go M.J.
        • Kim Y.J.
        • Heo J.Y.
        • Oh J.H.
        • Ban H.J.
        • Yoon D.
        • Lee M.H.
        • Kim D.J.
        • Park M.
        • et al.
        A large-scale genome-wide association study of Asian populations uncovers genetic factors influencing eight quantitative traits.
        Nat. Genet. 2009; 41: 527-534
        • Kim Y.J.
        • Go M.J.
        • Hu C.
        • Hong C.B.
        • Kim Y.K.
        • Lee J.Y.
        • Hwang J.Y.
        • Oh J.H.
        • Kim D.J.
        • Kim N.H.
        • et al.
        Large-scale genome-wide association studies in East Asians identify new genetic loci influencing metabolic traits.
        Nat. Genet. 2011; 43: 990-995
        • Friedewald W.T.
        • Levy R.I.
        • Fredrickson D.S.
        Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge.
        Clin. Chem. 1972; 18: 499-502
        • Varbo A.
        • Benn M.
        • Tybjaerg-Hansen A.
        • Jorgensen A.B.
        • Frikke-Schmidt R.
        • Nordestgaard B.G.
        Remnant cholesterol as a causal risk factor for ischemic heart disease.
        J. Am. Coll. Cardiol. 2013; 61: 427-436
        • Jørgensen A.B.
        • Frikke-Schmidt R.
        • West A.S.
        • Grande P.
        • Nordestgaard B.G.
        • Tybjaerg-Hansen A.
        Genetically elevated non-fasting triglycerides and calculated remnant cholesterol as causal risk factors for myocardial infarction.
        Eur. Heart J. 2013; 34: 1826-1833
        • National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III)
        Third Report of the National Cholesterol Education Program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult Treatment Panel III) final report.
        Circulation. 2002; 106: 3143-3421
      1. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults–the evidence report. National Institutes of Health.
        Obes. Res. 1998; 6: 51S-209S
        • Seo M.H.
        • Lee W.Y.
        • Kim S.S.
        • Kang J.H.
        • Kang J.H.
        • Kim K.K.
        • Kim B.Y.
        • Kim Y.H.
        • Kim W.J.
        • Kim E.M.
        • et al.
        Committee of Clinical Practice Guidelines, Korean Society for the Study of Obesity (KSSO).
        J. Obes. Metab. Syndr. 2019; 28 (2018 Korean Society for the Study of Obesity guideline for the management of obesity in Korea): 40-45
        • Howie B.N.
        • Donnelly P.
        • Marchini J.
        A flexible and accurate genotype imputation method for the next generation of genome-wide association studies.
        PLoS Genet. 2009; 5: e1000529
        • Delaneau O.
        • Zagury J.F.
        • Marchini J.
        Improved whole-chromosome phasing for disease and population genetic studies.
        Nat. Methods. 2013; 10: 5-6
        • Piegorsch W.W.
        • Weinberg C.R.
        • Taylor J.A.
        Non-hierarchical logistic models and case-only designs for assessing susceptibility in population-based case-control studies.
        Stat. Med. 1994; 13: 153-162
        • Mukherjee B.
        • Chatterjee N.
        Exploiting gene-environment independence for analysis of case-control studies: an empirical Bayes-type shrinkage estimator to trade-off between bias and efficiency.
        Biometrics. 2008; 64: 685-694
        • Kooperberg C.
        • Leblanc M.
        Increasing the power of identifying gene x gene interactions in genome-wide association studies.
        Genet. Epidemiol. 2008; 32: 255-263
        • Murcray C.E.
        • Lewinger J.P.
        • Gauderman W.J.
        Gene-environment interaction in genome-wide association studies.
        Am. J. Epidemiol. 2009; 169: 219-226
        • Hsu L.
        • Jiao S.
        • Dai J.Y.
        • Hutter C.
        • Peters U.
        • Kooperberg C.
        Powerful cocktail methods for detecting genome-wide gene-environment interaction.
        Genet. Epidemiol. 2012; 36: 183-194
        • Gauderman W.J.
        • Zhang P.
        • Morrison J.L.
        • Lewinger J.P.
        Finding novel genes by testing G x E interactions in a genome-wide association study.
        Genet. Epidemiol. 2013; 37: 603-613
        • Ionita-Laza I.
        • McQueen M.B.
        • Laird N.M.
        • Lange C.
        Genomewide weighted hypothesis testing in family-based association studies, with an application to a 100K scan.
        Am. J. Hum. Genet. 2007; 81: 607-614
        • Purcell S.
        • Neale B.
        • Todd-Brown K.
        • Thomas L.
        • Ferreira M.A.
        • Bender D.
        • Maller J.
        • Sklar P.
        • de Bakker P.I.
        • Daly M.J.
        • et al.
        PLINK: a tool set for whole-genome association and population-based linkage analyses.
        Am. J. Hum. Genet. 2007; 81: 559-575
        • Willer C.J.
        • Li Y.
        • Abecasis G.R.
        METAL: fast and efficient meta-analysis of genomewide association scans.
        Bioinformatics. 2010; 26: 2190-2191
        • Witte J.S.
        • Visscher P.M.
        • Wray N.R.
        The contribution of genetic variants to disease depends on the ruler.
        Nat. Rev. Genet. 2014; 15: 765-776
        • Aulchenko Y.S.
        • Ripke S.
        • Isaacs A.
        • van Duijn C.M.
        GenABEL: an R library for genome-wide association analysis.
        Bioinformatics. 2007; 23: 1294-1296
        • Sung J.
        • Cho S-I.
        • Lee K.
        • Ha M.
        • Choi E-Y.
        • Choi J-S.
        • Kim H.
        • Kim J.
        • Hong K.S.
        • Kim Y.
        • et al.
        Healthy twin: a twin-family study of Korea–protocols and current status.
        Twin Res. Hum. Genet. 2006; 9: 844-848
        • Gombojav B.
        • Song Y.M.
        • Lee K.
        • Yang S.
        • Kho M.
        • Hwang Y.C.
        • Ko G.
        • Sung J.
        The Healthy twin study, Korea updates: resources for omics and genome epidemiology studies.
        Twin Res. Hum. Genet. 2013; 16: 241-245
        • Yang J.
        • Lee S.H.
        • Goddard M.E.
        • Visscher P.M.
        GCTA: a tool for genome-wide complex trait analysis.
        Am. J. Hum. Genet. 2011; 88: 76-82
        • Lee M.H.
        • Kim H.C.
        • Ahn S.V.
        • Hur N.W.
        • Choi D.P.
        • Park C.G.
        • Suh I.
        Prevalence of Dyslipidemia among Korean adults: Korea National Health and Nutrition Survey 1998–2005.
        Diabetes Metab. J. 2012; 36: 43-55
        • Su Y.R.
        • Di C.Z.
        • Hsu L.
        • Genetics C.
        A unified powerful set-based test for sequencing data analysis of GxE interactions.
        Biostatistics. 2017; 18: 119-131
        • Yang T.
        • Chen H.
        • Tang H.
        • Li D.
        • Wei P.
        A powerful and data-adaptive test for rare-variant-based gene-environment interaction analysis.
        Stat. Med. 2019; 38: 1230-1244
        • Cohen J.C.
        • Pertsemlidis A.
        • Fahmi S.
        • Esmail S.
        • Vega G.L.
        • Grundy S.M.
        • Hobbs H.H.
        Multiple rare variants in NPC1L1 associated with reduced sterol absorption and plasma low-density lipoprotein levels.
        Proc. Natl. Acad. Sci. USA. 2006; 103: 1810-1815
        • Després J.P.
        Body fat distribution and risk of cardiovascular disease: an update.
        Circulation. 2012; 126: 1301-1313
        • Yusuf S.
        • Hawken S.
        • Ounpuu S.
        • Bautista L.
        • Franzosi M.G.
        • Commerford P.
        • Lang C.C.
        • Rumboldt Z.
        • Onen C.L.
        • Lisheng L.
        • et al.
        Obesity and the risk of myocardial infarction in 27,000 participants from 52 countries: a case-control study.
        Lancet. 2005; 366: 1640-1649
        • Després J.P.
        • Lemieux I.
        Abdominal obesity and metabolic syndrome.
        Nature. 2006; 444: 881-887
        • Konieczna J.
        • Abete I.
        • Galmes A.M.
        • Babio N.
        • Colom A.
        • Zulet M.A.
        • Estruch R.
        • Vidal J.
        • Toledo E.
        • Diaz-Lopez A.
        • et al.
        PREDIMED-Plus Investigators.
        Clin. Nutr. 2018; 38 (Body adiposity indicators and cardiometabolic risk: cross-sectional analysis in participants from the PREDIMED-Plus trial): 1883-1891
        • Coutinho T.
        • Goel K.
        • Correa de Sa D.
        • Kragelund C.
        • Kanaya A.M.
        • Zeller M.
        • Park J.S.
        • Kober L.
        • Torp-Pedersen C.
        • Cottin Y.
        • et al.
        Central obesity and survival in subjects with coronary artery disease: a systematic review of the literature and collaborative analysis with individual subject data.
        J. Am. Coll. Cardiol. 2011; 57: 1877-1886
        • Dallongeville J.
        • Bhatt D.L.
        • Steg P.H.
        • Ravaud P.
        • Wilson P.W.
        • Eagle K.A.
        • Goto S.
        • Mas J.L.
        • Montalescot G.
        • Investigators R.R.
        Relation between body mass index, waist circumference, and cardiovascular outcomes in 19,579 diabetic patients with established vascular disease: the REACH Registry.
        Eur. J. Prev. Cardiol. 2012; 19: 241-249