genomewide association studies in cardiovascular disease—an … · 2017-07-16 · genomewide...

Genomewide Association Studies in CardiovascularDisease—An Update 2011

Tanja Zeller,1 Stefan Blankenberg,1 and Patrick Diemert1*

BACKGROUND: Genomewide association studies haveled to an enormous boost in the identification of sus-ceptibility genes for cardiovascular diseases. This re-view aims to summarize the most important findingsof recent years.

CONTENT: We have carefully reviewed the current liter-ature (PubMed search terms: “genome wide associa-tion studies,” “genetic polymorphism,” “genetic riskfactors,” “association study” in connection with the re-spective diseases, “risk score,” “transcriptome”).

SUMMARY: Multiple novel genetic loci for such impor-tant cardiovascular diseases as myocardial infarction,hypertension, heart failure, stroke, and hyperlipidemiahave been identified. Given that many novel geneticrisk factors lie within hitherto-unsuspected genes orinfluence gene expression, these findings have inspireddiscoveries of biological function. Despite these suc-cesses, however, only a fraction of the heritability formost cardiovascular diseases has been explained thusfar. Forthcoming techniques such as whole-genome se-quencing will be important to close the gap of missingheritability.© 2011 American Association for Clinical Chemistry

The widespread use of genomewide association studies(GWASs)2 in the last 5 years has led to an enormousacceleration in discoveries across the entire spectrumof cardiovascular phenotypes. In fact, the multitude ofnovel gene loci for cardiovascular disease identified inrecent years through GWASs has been compared to a“gold rush” in cardiovascular genomics. This impres-

sion might have been triggered by the fact that in the 20years before GWAS became widely available, most ge-netic associations were based on candidate genes. Al-though such candidate-gene studies have reportedmultiple positive findings for cardiovascular pheno-types since the early 1990s, few have actually been rep-licated and have withstood the test of time (1 ). There-fore, up until about 2006, the search for causal genes incardiovascular disease could be compared to a searchfor the proverbial needle in a haystack.

The current picture in 2011 is vastly different, andmany new loci have been identified over the last 5 yearsfor most important cardiovascular diseases and pheno-types, including coronary artery disease (CAD), hyper-tension, heart failure, hyperlipidemia, and stroke.

Building on a substantive body of work that hasbeen described over the past few years, we discuss thecurrent knowledge of genomics in cardiovascular dis-ease. We provide an overview of the promises and pit-falls of the GWAS approach and systematically reviewthe current knowledge of GWASs in cardiovasculardisease. Table 1 provides an overview of importantterms used in this review.

GWASs—Technical Aspects

Genetic-association studies gather a large number ofindividuals affected by a particular trait, such as CAD,myocardial infarction (MI), or related quantitativetraits, and search for individual marker alleles (or ge-notypes) that are more frequent in individuals withdisease than in healthy control individuals (2 ). The re-sults of the Human Genome Project (3 ) and the Inter-national HapMap Project (4 ) have facilitated the de-velopment of a public catalogue of the most commonform of human genomic variation, single-nucleotidepolymorphisms (SNPs). In addition, this SNP infor-mation is being used in the characterization of linkagedisequilibrium (LD) structures of the genome (5 ),thereby providing the foundation for large-scale geno-typing and computational technologies, an approachcalled a GWAS.

GWASs are facilitated by the availability of afford-able whole-genome SNP panels and large-scale geno-typing technologies (e.g., Affymetrix or Illumina), andpowerful association-analysis methods and software.

1 Department of General and Interventional Cardiology, The University HeartCenter at the University Medical Center Hamburg-Eppendorf, Hamburg, Ger-many.

* Address correspondence to this author at: University Heart Center Hamburg,Martinistrasse 52, 20246 Hamburg, Germany. Fax �49-40-741054818; [email protected].

Received October 4, 2011; accepted November 1, 2011.Previously published online at DOI: 10.1373/clinchem.2011.1704312 Nonstandard abbreviations: GWAS, genomewide association study; CAD, cor-

onary artery disease; MI, myocardial infarction; SNP, single-nucleotide polymor-phism; LD, linkage disequilibrium; PROVE IT–TIMI 22, Pravastatin or Atorvasta-tin Evaluation and Infection Therapy – Thrombolysis in Myocardial Infarction 22(trial); LDL-C, LDL cholesterol; RAP, LDL receptor–associated protein; C/EBP,CCAAT enhancer–binding protein; eQTL, expression quantitative trait loci.

Clinical Chemistry 58:192–103 (2012) Reviews

92

Association studies, which can be performed onunrelated individuals, rely on the LD between testedloci and disease-predisposing variants at the level of thetotal population. The rationale of this approach is thatif unknown disease-predisposing variants are presentsomewhere in the genome, they might be detected viatheir LD with tagging SNPs represented on the geno-typing array (5 ).

By design, GWASs provide an unbiased survey ofthe effects of common genetic variants. SNPs chosenfor GWASs typically have a minor-allele frequency�0.05 and are selected to “tag” the most common hap-lotypes. The power of detection depends directly on thesample size of the study population, the minor-allelefrequency, the strength of LD between the SNP and thecausal variants, and the effect sizes of the alleles (6 ).

The effect sizes associated with SNPs identifiedthrough GWASs are rather small, with odds ratiosusually between 0.7 and 1.3 (7 ). Therefore, largesample sizes are necessary to uncover the geneticcontributions of common variants (8 ). Another keyaspect for a successful GWAS is the quality of phe-notypic characterization. Because complex diseasessuch as cardiovascular disease are characterized byphenotypic heterogeneity (e.g., phenotypes mayvary with respect to age or disease onset), detailedand standardized phenotyping is crucial (9, 10 ).Similar to the effects of genotyping errors, evenmodest levels of errors in the phenotypic character-ization may diminish the power to detect (9 ). Max-imum attention to individual/phenotype character-ization by means of standardized, reproducible, and

quality-controlled measurements needs to be en-sured (11 ), and patients and controls should un-dergo the same phenotyping protocol to establishboth the presence of disease in patients and the ab-sence of disease in controls (12 ). Accordingly, stud-ies with smaller sample sizes and precise phenotypicmeasurement may be as powerful as studies that areeven 20 times larger but use less precise measure-ments (13 ).

A large number of statistical tests and a large sam-ple size are required in GWASs to detect modestly sizedassociations. Typically, millions of statistical testsspread across the spectrum of known common SNPshave to be performed; thus, GWASs require extremelevels of statistical significance (14 ). The primarymethod used to address the issue of spurious statisticalsignificance arising from multiple comparisons involv-ing up to 1 million SNPs is the Bonferroni correction(15 ). In such evaluations, the conventional � level of P(�0.05) is divided by the number of tests performed(e.g., 0.05/1 000 000, or 5 � 10�8).

The imputation of unmeasured genotypes by us-ing information about human LD patterns revealed bythe HapMap Project has become an essential tool inconducting GWASs (4 ). Use of that informationmakes it possible to supplement directly genotypedSNPs obtained from commercially available arrayswith millions of “free” genotypes spanning the humangenome at the relatively low costs of computation timeand additional statistical tests, leading to an increasedpower to detect associations (16 ). Furthermore, impu-tation tools simplify the combination of association re-

Table 1. Legend: glossary of important terms in cardiovascular genetics.

Glossary

Genomewide association study (GWAS) Study investigating statistical association between observable traits or the presence/absenceof a disease and a large number of genetic markers. GWASs do not rely on a priorihypotheses.

Single-nucleotide polymorphism (SNP) Variation of single nucleotide in a genetic sequence; common form of human variation.

Minor-allele frequency (MAF) Proportion of the less common of 2 alleles in a population.

Linkage disequilibrium (LD) Nonrandom correlation between 2 alleles located near each other on the samechromosome.

International HapMap Project Effort to identify and catalog genetic similarities and differences in humans. Genomewidedatabase of common human variations among multiple ancestral population samples.

Haplotype Combination of specific alleles at different loci on the chromosome that tend to beinherited together.

Expression quantitative trait loci (eQTL) Genomic loci that show statistically significant correlation to gene expression level. In eQTLstudies, genetic variants at specific genomic loci are associated with gene expressionlevels to pinpoint polymorphic genetic regions affecting gene expression (in cis or trans).

cis-/trans-acting Spatial designation of the location of a genetic variant in relation to a locus (gene) affectedby this variation, e.g., an eQTL. cis-acting indicates a location near the affected locus(e.g. within 1 MB upstream and downstream); trans-acting indicates a location elsewherein the genome.

Genetic Studies in Cardiovascular Disease—Update 2011 Reviews

Clinical Chemistry 58:1 (2012) 93

sults across studies that use different genotyping plat-forms and thus have relatively few SNPs in common(16 ).

Alleles identified in GWASs are often not causativebut are likely in LD with the true causative ones (17 ).Therefore, results obtained in discovery GWASs areonly a first step. They require replication in indepen-dent study populations and ultimately experimentalvalidation.

Review of GWAS Results for the Most ImportantCardiovascular Phenotypes

CORONARY ARTERY DISEASE

CAD is the most common cause of death in industrial-ized countries, and its prevalence is rapidly increasingin developing countries. CAD has a complex and het-erogeneous etiology involving numerous environmen-tal and genetic factors of disease risk (18 ).

In 2007, three landmark studies used the GWAapproach successfully and simultaneously reported theidentification of variants on chromosome 9p21.3 asso-ciated with risk for MI and CAD (19 –21 ). The SNPs inthe region were in strong LD and defined a haplotypeassociated with a 15%–20% increased risk in heterozy-gous individuals and a 30%– 40% increased risk in ho-mozygous individuals (5 ). Since then, several studieshave confirmed the role of the chromosome 9p21.3region in the risk of CAD and have postulated, on thebasis of fairly persuasive data, associations of this re-gion with other disease phenotypes, including type 2diabetes, (22 ), ischemic stroke (23 ), aortic aneurysm(24 ), glioma (25 ), malignant melanoma (26 ), and ag-gressive periodontitis (27 ).

It is particularly striking that the genomic regionof interest contains no annotated genes and SNPs tag-ging this region that are associated with any establishedrisk factor for CAD. Subsequently, the results of furtherGWASs for MI and CAD have been published (28 –33 ).A large metaanalysis of GWASs on CAD that incorpo-rated 22 000 cases and 60 000 controls was recentlypublished (34 ), bringing the total number of GWA lociwith genomewide significance for CAD to 26 (Fig. 1).

There are no obvious links to function or tradi-tional risk factors for the majority of the CAD risk lociidentified by GWASs. It is notable, however, that 6 ofthese loci are associated with blood lipid concentra-tions, an observation that underscores the pathophys-iological importance of this risk factor.

HYPERTENSION

Hypertension is the most widespread cardiovasculardisease, with a global burden estimated at 25% to 30%of the population. All genetic loci discovered thus farexplain only about 2% of the overall variation in blood

pressure. The identification of genetic loci for hyper-tension through GWASs has proved to be exceptionallydifficult, however, with the first GWA analysis (35 )having yielded no loci with genomewide significance.This problem may be due to the particular difficulty indefining this phenotype: Single blood pressure mea-surements in population-based cohorts are often unre-liable, and many individuals receive medications, suchas �-blockers and angiotensin-converting enzyme in-hibitors, that influence blood pressure. Several large-scale GWASs and metaanalyses have recently identifiedbroadly replicated risk loci for hypertension withgenomewide significance (12, 35– 40 ). The odds ratiosassociated with risk loci for hypertension identified inGWASs are very modest and do not exceed 1.05. Withsome authors arguing that the genetic contribution todeveloping hypertension might be as large as 50%, it isevident that there is still a huge amount of missing her-itability for this important disease.

HYPERLIPIDEMIA

The first GWASs for blood lipid concentrations werepublished in 2008 (41– 43 ). Several aspects of GWASsfor blood lipids are particularly noteworthy. First, asubstantial number of the genes well known to causerare mendelian hyperlipidemia disorders, such asLDLR3 (low density lipoprotein receptor), APOB [apo-lipoprotein B (including Ag(x) antigen)], and PCSK9(proprotein convertase subtilisin/kexin type 9) in fa-milial hypercholesterolemia, have been rediscoveredthrough GWASs. In fact, hyperlipidemia is better thanmost other phenotypes in that there are substantialnumber of gene loci for which common genetic varia-tion and rare mendelian mutations overlap. Second, itis intriguing that a large number of GWA loci for hy-perlipidemia are also significantly associated with CADrisk, proving the unequivocal link between blood lipidconcentration and CAD risk. Finally, several newly dis-

3 Genes: LDLR, low density lipoprotein receptor; APOB, apolipoprotein B (includ-ing Ag(x) antigen); PCSK9, proprotein convertase subtilisin/kexin type 9;SLCO1B1, solute carrier organic anion transporter family, member 1B1; ABCB1,ATP-binding cassette, sub-family B (MDR/TAP), member 1; CYP2C19, cyto-chrome P450, family 2, subfamily C, polypeptide 19; CYP2C9, cytochrome P450,family 2, subfamily C, polypeptide 9; VKORC1, vitamin K epoxide reductasecomplex, subunit 1; CYP4F2, cytochrome P450, family 4, subfamily F, polypep-tide 2; CDKN2A, cyclin-dependent kinase inhibitor 2A (melanoma, p16, inhibitsCDK4); CDKN2B, cyclin-dependent kinase inhibitor 2B (p15, inhibits CDK4);CDKN2B-AS, CDKN2B antisense RNA (non-protein coding) [previously known asANRIL (antisense noncoding RNA in the INK4 locus)]; MTAP, methylthioadenos-ine phosphorylase; Mus musculus gene Cdkn2a, cyclin-dependent kinase inhib-itor 2A; Mus musculus gene Cdkn2b, cyclin-dependent kinase inhibitor 2B; Musmusculus gene Mtap, methylthioadenosine phosphorylase; Mus musculus geneDmrta1, doublesex and mab-3 related transcription factor like family A1;CELSR2, cadherin, EGF LAG seven-pass G-type receptor 2 (flamingo homolog,Drosophila); PSRC1, proline/serine-rich coiled-coil 1; SORT1, sortilin 1; Musmusculus gene Sort1, sortilin 1; Mus musculus gene Ldlr, low density lipoproteinreceptor; LIPA, lipase A, lysosomal acid, cholesterol esterase.

Reviews

94 Clinical Chemistry 58:1 (2012)

covered GWAS loci for blood lipids have already beenfunctionally elucidated in animal studies [e.g., (44 –46 )], providing good evidence for how GWAS findingscan inspire functional experiments. A large GWASmetaanalysis for blood lipid concentrations in�100 000 individuals published in 2010 has identified95 loci showing genomewide significance for associa-tion with blood lipids (46 ).

STRUCTURAL HEART DISEASE AND HEART FAILURE

Echocardiography constitutes the most reliable toolfor diagnosing structural changes of the heart. Suchstructural changes as diastolic dysfunction, cardiac

hypertrophy, dilation of the atria, and valvular dis-ease are detectable in their early stages with echocar-diography and may precede the phenotype of con-gestive heart failure by many years. A metaanalysis ofGWASs on echocardiographic data for left ventricu-lar function and dimension, atrial size, and aorticroot diameter was recently published (47 ). Thisstudy identified many genes that might provide afunctional link between intermediate phenotypessuch as hypertension or obesity and resulting struc-tural changes of the heart.

Congestive heart failure constitutes a spectrum ofheterogeneous conditions, such as dilated cardiomy-

Fig. 1. CAD GWAS loci identified as of September 2011.

The 26 CAD-risk loci with genomewide significance identified by mid-2011 are highlighted. PCSK9, proprotein convertasesubtilisin/kexin type 9; PPAP2B, phosphatidic acid phosphatase type 2B; SORT1, sortilin 1; MIA3, melanoma inhibitory activityfamily, member 3; WDR12, WD repeat domain 12; MRAS, muscle RAS oncogene homolog; PHACTR1, phosphatase and actinregulator 1; ANKS1A, ankyrin repeat and sterile alpha motif domain containing 1A; TCF21, transcription factor 21; LPA,lipoprotein, Lp(a); ZC3HC1, zinc finger, C3HC-type containing 1; CDKN2A, cyclin-dependent kinase inhibitor 2A (melanoma, p16,inhibits CDK4); ABO, ABO blood group (transferase A, alpha 1–3-N-acetylgalactosaminyltransferase; transferase B, alpha1–3-galactosyltransferase); CXCL12, chemokine (C-X-C motif) ligand 12 [formerly known as SDF1]; LIPA, lipase A, lysosomalacid, cholesterol esterase; CYP17A1, cytochrome P450, family 17, subfamily A, polypeptide 1; APOA5, apolipoprotein A-V;SH2B3, SH2B adaptor protein 3; COL4A1, collagen, type IV, alpha 1; HHIPL1, HHIP-like 1; ADAMTS7, ADAM metallopeptidasewith thrombospondin type 1 motif, 7; SMG6, smg-6 homolog, nonsense mediated mRNA decay factor (C. elegans); RASD1, RAS,dexamethasone-induced 1; UBE2Z, ubiquitin-conjugating enzyme E2Z; LDLR, low density lipoprotein receptor; MRPS6, mito-chondrial ribosomal protein S6.



opathy, ischemic cardiomyopathy, hypertrophic ob-structive cardiomyopathy, and others. A familial oc-currence and a simple mendelian inheritance has beendemonstrated for several of these diseases, whereas thegenetic risk factors that commonly lead to heart failurehave been unknown.

Two GWASs have recently identified 3 novel geneloci for dilated cardiomyopathy (48, 49 ).

RHYTHM DISORDERS

Disorders of heart rhythm and electrical conductanceof the heart have a strong heritable component. Re-cently, several metaanalyses of GWASs have been pub-lished on electrocardiographic components of electri-cal conductance in the heart, including PR interval(50 ), QT interval (51 ), QRS duration (52 ), and restingheart rate (53 ). It is notable that many of the genesidentified by GWASs for these subphenotypes of car-diac or electrical conductance include those for ionchannels or proteins that regulate the intracellular con-centrations of electrolytes, such as phospholamban forcalcium. Of particular clinical interest are severalrecent GWASs on atrial fibrillation [e.g., (50, 54 –57 )]and ventricular fibrillation after MI (58 ). Genetic riskfactors associated with the risk of ventricular fibrilla-tion and/or sudden cardiac death after MI might beclinically useful in helping to identify individuals whowould benefit from antiarrhythmic therapy or implan-tation of an automated defibrillator. Prospective trialswill be important for determining whether such ge-netic variants carry clinical implications. For atrial fi-brillation, it is worth noting that the 4q25 and 16q22GWAS loci are also associated significantly with therisk of embolic stroke (59, 60 ), results that fit the un-derlying etiology linking atrial fibrillation with embolicstroke.

Pharmacogenomics

The importance of the pharmacogenomics of drugs inthe treatment of cardiovascular disease has been ad-dressed in several GWAS analyses.

Several prospective observational studies havesuggested an association between the SNP encodingthe Trp719Arg substitution in kinesin-like protein 6(rs20455) and the development of clinical CAD. Inaddition, results from the PROVE IT–TIMI 22(Pravastatin or Atorvastatin Evaluation and Infec-tion Therapy – Thrombolysis in Myocardial Infarc-tion 22) trial suggest that carriers of this variant willshow an incremental benefit of risk reduction whentreated with statins (61 ). Rather prematurely, acommercial genetic test for statin response was thenmade available; however, recent large-scale meta-analyses of the SNP encoding Trp719Arg could not

confirm a role for this variant in CAD risk and statinresponse (62 ).

Statin-induced rhabdomyolysis is a rare but po-tentially devastating side effect of statin therapy. A re-cent GWAS comparing 85 individuals with statin-induced myopathy with control individuals taking thesame medication identified a common risk-factor SNPin the SLCO1B1 (solute carrier organic anion trans-porter family, member 1B1) gene, which encodes atransmembrane transporter already known to regulateintracellular statin concentrations (63 ). Homozygosityfor this SNP is associated with a 17-fold increased riskof myopathy under statin treatment. This associationhas thus far been demonstrated only for simvastatin,but the same mechanism likely applies to other staindrugs as well. Genetic testing for this risk factor couldgreatly increase the safety of statin treatment, particu-larly in situations in which myopathy risk is increased,as in cotreatment with amiodarone, cyclosporine, andother drugs.

Currently, clopidogrel is the most extensivelyused drug for inhibiting platelet aggregation, partic-ularly after coronary artery stenting and/or treat-ment for acute coronary syndrome. The clopidogrelprodrug is metabolized and bioactivated in the liverby various cytochrome oxidases. The intracellularconcentration is also regulated by ABC transporterproteins, such as P-glycoprotein [encoded byABCB1, ATP-binding cassette, sub-family B (MDR/TAP), member 1]. Several GWASs have linked ge-netic variation in CYP2C19 (cytochrome P450,family 2, subfamily C, polypeptide 19) andP-glycoprotein with the bioavailability of clopi-dogrel, the degree of platelet inhibition, and cardio-vascular events after stent implantation (64 – 66 ). Incontrast to clopidogrel, the metabolism of prasugrel,a newly developed platelet inhibitor, is not affectedby these genetic polymorphisms (67 ). In the nearfuture, genetic testing for polymorphisms associatedwith clopidogrel metabolism may have an impact onboth the choice and dose of drug in high-riskindividuals.

GWASs have also led to considerable progress inthe pharmacogenetics of oral anticoagulants. Ge-netic polymorphisms in CYP2C9 (cytochrome P450,family 2, subfamily C, polypeptide 9), VKORC1 (vi-tamin K epoxide reductase complex, subunit 1), andCYP4F2 (cytochrome P450, family 4, subfamily F,polypeptide 2) have been identified as major deter-minants of warfarin response (68, 69 ). Applyingthese genetic polymorphisms in a genetic scoremakes it possible to predict effective warfarin dosesmore precisely and safely than with a clinical orfixed-dose algorithm (70 ).

Reviews


Clinical Utility of GWAS Findings inCardiovascular Disease—Risk Prediction withSNP Scores

The enormous progress made in the field of genomiccardiovascular research, as well as the continuously in-creasing availability of genetic testing to the generalpublic, leads inevitably to the question of whethergenomic variants, singly or in combination, have thediscriminatory power to substantially improve the pre-diction of clinically important cardiovascular out-comes and, even more importantly, therapy discrimi-nation. Given that most of the genetic variants explainonly a modest fraction of the variance (34, 71 ), a com-bined multilocus risk score might improve risk predic-tion (72 ).

Therefore, recent efforts have evaluated the risk-predicting ability of validated lipid-modulating SNPs(71 ) and SNPs associated with type 2 diabetes (73 ),hypertension (37 ), coronary heart disease (74, 75 ),and cardiovascular disease, including intermediatephenotypes and biomarkers (76 –78 ). In most analyses,assessment of genetic risk scores did not offer any sub-stantial improvement in risk discrimination with c sta-tistics or net reclassification (71, 73, 75, 76, 78 ). Thus,the clinical usefulness of genetic risk scores based onvariants identified by GWASs currently remains un-clear, and the use of genetic testing to improve riskstratification is clearly too premature and remains to bedefined. However, novel sequencing techniques withthe potential to lead to the identification of susceptibil-ity variants with larger effect sizes and improved statis-tical analyses will provide a framework for futureresearch.

The Next Big Challenge: Linking Association Signalswith Biological Function

GWASs have the power to detect phenotype-associatedloci and to delineate the genetic basis of common com-plex disorders (6 ); however, the functional conse-quences of the identified association signals are scant.The majority of the associations reside outside of cod-ing regions, and the identified alleles are seldom trulycausative. Furthermore, GWAS results mostly have aminimal to modest impact on risk stratification, diag-nosis, and genetic prevention and treatment at the in-dividual level. Consequently, a substantial gap exists inour understanding of how genetic variants affect thepathophysiological mechanisms through which thesesusceptibility loci contribute to disease (33 ). Hence,the next challenge is to link association signals withbiological function, and extensive additional geneticand molecular studies are needed to identify causativevariants responsible for disease susceptibility. In the

next section, we provide an overview of recent majorattempts to unravel the basis for CAD susceptibility.

Gene Desert with Regulatory Function?The 9p21 Locus

When the first large-scale GWASs on CAD identifiedthe chromosome 9p21.3 region in 2007 (19 –21 ), itcame as a surprise that this particular genomic region islocated in a gene desert with no annotated genes. Sub-sequently, a large number of genetic and experimentalanalyses have been performed to elucidate the effects ofthis locus.

The identified region is adjacent to the INK4/ARFlocus, which comprises CDKN2A [cyclin-dependentkinase inhibitor 2A (melanoma, p16, inhibits CDK4)],CDKN2B [cyclin-dependent kinase inhibitor 2B (p15,inhibits CDK4)], and CDKN2B-AS [CDKN2B anti-sense RNA (non-protein coding); previously known asANRIL (antisense noncoding RNA in the INK4 locus)],a gene encoding a non–protein-coding RNA. This lo-cus also contains the MTAP (methylthioadenosinephosphorylase) gene (Fig. 2). Several groups have in-vestigated the expression of transcripts by the INK4/ARF locus with respect to SNPs conferring CAD risk(79 – 82 ) and demonstrated an effect of the 9p21 locuson the expression of INK4/ARF transcripts, includingthe ANRIL transcript, but not on MTAP gene expres-sion (79 ). Interestingly, multiple sites in the 9p21 re-gion seem to independently influence the productionof INK4/ARF transcripts (82 ), but CAD-associatedSNPs within this region are all highly associated withCDKN2B-AS (i.e., ANRIL) expression, suggesting aprominent role for ANRIL as the prime susceptibilitycandidate gene of the 9p21 locus (82, 83 ). That conclu-sion has been underlined by results demonstratingANRIL expression in different cells and tissues relevantfor atherosclerosis (80, 81, 83 ). Additionally, ANRILtranscripts of different lengths have been detected, withproduction of the short transcript increased and thelarge transcript decreased in carriers of the 9p21 riskhaplotype (80, 81 ). Jarinova et al. hypothesized that byreducing production of the long ANRIL transcript orincreasing production of the short ANRIL transcripts,the risk allele may modify CDKN2B expression,thereby promoting proliferative phenotype cell typesrelevant to atherosclerosis (80 ). In line with these re-sults, Visel et al. (84 ) showed in a transgenic mousemodel that deletion of the orthologous region onmouse chromosome 4 decreased the expression of theneighboring Cdkn2a (cyclin-dependent kinase inhibi-tor 2A) and Cdkn2b (cyclin-dependent kinase inhibitor2B) genes [but not Mtap (methylthioadenosine phos-phorylase) and Dmrta1 (doublesex and mab-3 relatedtranscription factor like family A1)] in heart tissue, in-



dicating that the CAD risk interval is required for ap-propriate expression of Cdkn2a and Cdkn2b in theheart and thus might impact pathophysiological pro-cesses. Furthermore, an influence on cell proliferationand senescence, as well as an increased mortality undera diet containing high fat and high cholesterol (butwithout differences in aortic fatty-lesion formation),was observed in chr4�70kb/�70kb mice (84 ). The regionon mouse chromosome 4 shows only 50% homologywith the human 9p21 region, however, and no ANRILortholog exists in the mouse (85), thus raising additionalquestions about the discrepancies and commonalities ofthe different 9p21 studies. A recent study (85) investi-gated the cellular expression pattern of the CDKN2B,CDKN2A, and MTAP genes in human vascular lesionsand found abundant production of the encoded proteinsin the vessel walls, thus confirming these genes as interest-ing candidates potentially implicated in atherogenesis(but independent of the 9p21 atheroscleroticsusceptibility).

Recently, Harismendy et al. took a further step to-ward elucidating the molecular mechanism of the 9p21locus (86 ). This group investigated the functions of theunderlying variants of the CAD association in detail byusing cellular assays and population sequencing. Thestudy predicted the 9p21 interval to be the seconddensest gene desert for predicted enhancers in the hu-man genome and to contain the most disease-associated variants, indicating an important regulatoryfunction for the 9p21 gene desert. The group identified33 predicted enhancers within 9p21, and by sequencingthe region and computational determination of the vari-

ants disrupting consensus transcription factor–bindingsites in these predicted enhancers, they found CAD-associated SNPs (rs10811656 and rs10757278) to be lo-cated in one of these enhancers. In the presence of the riskhaplotype, a binding site for STAT1 (a transcription fac-tor involved in inflammatory responses) was disrupted.This finding was experimentally confirmed bychromatin-immunoprecipitation experiments. Further-more, this group observed an effect of small interferingRNA–mediated STAT1 knock-down on CDKN2B-AS(ANRIL) and CDKN2B expression, highlighting an im-portant role of STAT1/enhancer interaction in the regu-lation of the INK4/ARF locus and CAD susceptibility.

Despite this major progress made in the elucida-tion of the 9p21 susceptibility mechanisms, further invitro and in vivo studies are required to fully establishthe involvement of the INK4/ARF locus and MTAP inthe pathologic process and to translate the GWAS sig-nal into a biological meaning.

The 1p13 Sortilin Story

In 2007, another promising locus was identified onchromosome 1p13.3 through a genomewide approachin CAD patients (19 ). These results have been widelyreplicated, and this locus has consistently been associ-ated in several studies with LDL cholesterol (LDL-C)concentrations (41– 43, 46, 87, 88 ). Most studies haveidentified rs599839 as the lead SNP with the strongestcorrelation with CAD and LDL-C (89 ). In Europeanpopulations, the less frequent alleles have been associ-

Fig. 2. Overview of the chromosome 9p21 locus and regions associated with different traits.

The graph depicts the chromosomal location of the 9p21 locus encompassing the CDKN2A, CDKN2B, CDKN2B-AS (ANRIL), andMTAP genes. Encoded transcripts are displayed as gray arrows according to their location. A more detailed overview ofsubregions of the 9p21 locus associated with different traits is shown in the lower part of the figure. T2D, type 2 diabetes.

Reviews


ated with decreased LDL-C concentrations and a de-creased risk of CAD (44 ).

The 1p13.3 locus maps to a region encompassing 3tightly linked genes, namely CELSR2 [cadherin, EGFLAG seven-pass G-type receptor 2 (flamingo homolog,Drosophila)], PSRC1 (proline/serine-rich coiled-coil1), and SORT1 (sortilin 1). Different cell types showeddifferences in the regulation of the expression of thesegenes. In human liver, expression of all 3 genes showedregulation by the rs599839 genotype (42, 90). In humanblood cells, Linsel-Nitschke et al. observed a significantassociation between rs599839 and levels of SORT1 ex-pression but detected no association with CELSR2 orPSRC1 (44). However, using the largest population-basedhuman monocyte biobank available, Zeller et al. could notidentify the cis association between SORT1 mRNA levelsand the 1p13.3 locus, whereas they did prove a strongassociation between PSRC1 mRNA levels and the less fre-quent rs629301 allele (a tag of rs599839) (91). Despitethese discrepancies, recent investigations have focused onthe SORT1 gene to provide functional characterization ofthe association signals (44, 92, 93).

Sortilin 1 is a member of a VPS10P-domain recep-tor family, acts as a multiligand receptor, and binds theLDL receptor–associated protein (RAP), the lipolyticenzyme lipoprotein lipase, and apolipoprotein A-V(44, 94 ). Linsel-Nitschke et al. reported overproduc-tion of sortilin in transfected human embryonic kidney293 cells and found a significant increase in the uptakeof radiolabeled LDL and RAP particles into transfectedcells. These data suggest a potential function of sortilinas a lipoprotein receptor that mediates the uptake of

LDL particles into cells (44 ). In 2010, Musunuru et al.reported on an elegant series of experimental studies.These investigations identified a SNP at 1p13.3 thatcreated a CCAAT enhancer– binding protein (C/EBP)transcription factor– binding site that had functionalconsequences. The less frequent allele (that correlateswith reduced LDL-C concentrations) introduced theC/EBP binding site and increased the expression of lu-ciferase reporter constructs in human hepatoma cells.In liver-specific experiments in a humanized mousemodel, sortilin overproduction decreased total plasmacholesterol and LDL-C, results that were consistent withthe genetic findings in human cohorts. Conversely, smallinterfering RNA–meditated knockdown of Sort1 (sortilin1) expression in a mouse model led to an increase in totalplasma cholesterol. Musunuru et al. concluded that theirresults identified SORT1 as the causal gene at the 1p13.3locus for LDL-C and CAD/MI (93).

In contrast to the studies of Linsel-Nitschke et al.and Musunuru et al. that reported a negative correla-tion between Sort1 expression and LDL-C concentra-tions, Kjolby et al. reported a positive correlation thatwas supported by experiments with whole-animalSort1�/�Ldlr�/� double knock-out mice (92 ).

Apparently, discrepancies exist in the current results forthe SORT1 story. It is likely that the use of different in vivoand in vitro models is responsible for these discrepant out-comes; thus, further studies are needed to fully reconcile andexplaintheSORT1story.Nevertheless,currentdatahaveun-covered regulatory pathway(s) in hepatic lipoprotein exportby logical, hypothesis-based experiments and have providedthe first functional explanations for cardiovascular risk being

Table 2. Legend: summary of different theories/models of mode of action of sequence variations on thechromosome 1p13 CAD risk locus based on experimental data from 3 different studies.

Models of sortilin function andrelationship to 1p13 variants

Model 1 [Linsel-Nitschke et al.(44 ), Dube et al. (94 )]

G allele of SNP (rs599839) is associated with increased SORT1 expression andincreased LDL-C uptake into liver (lowering blood LDL-C concentrations),presumably through increased interactions with the LDL receptor and increasedreceptor-mediated lowering of circulating LDL-C concentrations consistent withexpected clinical biochemical consequences (i.e., decreased LDL-C concentrations)of increased SORT1 expression.

Model 2 [Musunuru et al (93 ),Dube et al. (94 )]

Identification of SNP (rs12740374) within binding site for C/EBP. Disruption of C/EBPbinding by the risk allele (G) of rs12740374, leading to reduced transcription ofSORT1 and increased hepatic LDL precursor VLDL secretion.

Lowering of circulating LDL-C concentrations consistent with the expected clinicalbiochemical consequence (i.e., decreased LDL-C concentrations) of increasedSORT1 expression.

Model 3 [Kjolby et al. (92 ), Dubeet al. (94 )]

Identification of colocalization of sortilin with apolipoprotein B-100, which facilitatesmaturation of pre-VLDL and hepatic secretion of VLDL. Increased SORT1expression stimulates hepatic release of VLDL and increases plasma LDL-C.

Increased circulating LDL-C concentrations are contrary to the clinical biochemicalconsequence (i.e., decreased LDL-C concentrations) expected on the basis ofearlier GWAS and eQTL studies.



associated with 1p13.3. Table 2 summarizes the models andrecent results of functional analyses of the 1p13/SORT1locus.

Linking GWAS Results with the Transcriptome

Variation in gene expression is important in mediatingdisease susceptibility. The identification of geneticvariants that potentially affect regulatory elements andthereby regulate gene transcription is sought as an aidin mapping human disease genes (95 ). eQTL (expres-sion quantitative trait loci) mapping, a widely used toolfor identifying genetic variants that affect gene regula-tion, provides a means for detecting transcriptionalregulatory relationships on a genomewide scale(87, 91, 92, 95–98 ).

In a study conducted to assess the overall variabil-ity of the monocytic transcriptome for nearly 1500healthy individuals, Zeller et al. (91 ) detected 2745eQTLs (of 12 808 expressed genes), most (90%) ofwhich were cis modulated. The authors further inves-tigated whether monocytic gene expression might me-diate the effects of loci recently identified by GWASs ofcardiovascular risk factors. Loci identified in previousGWASs of lipids, blood pressure, and body mass indexwere chosen, and associations between gene expressionand the lead or tag SNPs, as well as the respective riskfactor, were analyzed. For the phenotypic traits, severalSNPs showed strong correlation to gene expression,mainly in cis; however, when the association of the ex-pression trait with the risk factor under considerationwas assessed, only a few significant findings were ob-served. One was the above-mentioned association be-tween the 1p13.3 locus and LDL-C (91 ).

Within the frame of the “CADomics” study, thesame group performed a GWAS via a case– control ap-proach with more than 20 000 CAD cases and 38 000controls. They subsequently analyzed the effect on geneexpression for each CAD-associated SNP (33 ). Thisapproach revealed a novel CAD-susceptibility locus onchromosome 10q23.31, with the lead SNPs, rs1412444and rs2246833, located in intronic regions of the LIPA(lipase A, lysosomal acid, cholesterol esterase) gene,which encodes lysosomal acid lipase A. Using theabove-mentioned data set of global monocytic gene ex-pression, the authors assessed the influence of bothSNPs on gene expression. A strong cis effect on LIPAtranscript levels was observed, with the risk alleles be-ing associated with increased LIPA expression. Thisfind provides the first evidence for the functionality ofthis locus. Furthermore, LIPA transcript levels weresignificantly associated with prevalent cardiovascularrisk factors and phenotypes of subclinical disease. Inline with the results from the CADomics study, theCoronary Artery Disease (C4D) Genetics Consortium

confirmed LIPA as a novel CAD-associated locus in ametaanalysis of European and South Asian popula-tions and confirmed the association with increasedLIPA expression in a subsequent eQTL analysis (99 ).

These examples demonstrate the benefit of us-ing eQTL analyses to further evaluate the potentialfunctional relevance of genetic variants identified inlarge-scale association studies and present a first stepin the translation of genetic-association signals intobiological function. Combining eQTL and DNA se-quence–variation data sets in an analysis of coex-pressed gene networks, e.g., as shown by Heining etal. (100 ), will provide even more relevant informa-tion for unraveling the genetic underpinnings of car-diovascular disease.

Beyond GWAS: What Comes Next?

The broad availability of GWASs has led to an enor-mous boost in the discovery of risk genes for cardio-vascular disease. Genes identified by GWASs haveinspired biology and enabled the discovery of novelpathways or mechanisms, as exemplified by the find-ings for chromosome 9p21/ANRIL, chromosome1p13/SORT1, and—just recently—the chromosome10q23 LIPA locus. However, only a very small fractionof the heritability for the most important cardiovasculardiseases, such as CAD or hypertension, can be explainedby the common genetic risk factors identified by GWASsthus far. A huge gap of missing heritability remains. Thefield of individual risk prediction for cardiovascular dis-ease using common genetic factors is still in its infancy.Most genetic risk factors identified thus far will show onlysmall odds ratios, rendering the positive predictive valuesof such tests minimal.

For the most important cardiovascular diseases, suchas CAD, hypertension, and hyperlipidemia, large-scalemetaanalyses have already been conducted. Because thedensity of coverage and the number of participants in-volved in these studies have been optimized, it seems un-likely at present that substantially more discoveries willarise from GWASs.

To prioritize GWAS findings and to translate theminto biologically meaningful functions require post-GWAS analyses including bioinformatics-based ap-proaches, epigenetic approaches, and follow-up exper-imental studies (101, 102 ). With the rapid advancesin next-generation technologies, whole-exome and-genome sequencing is feasible, and data from the 1000Genomes Project (103 ) will provide a comprehensivemap of novel, rare variants.

For the identification of variants contributing to dis-ease susceptibility and possibly mediating gene expressioncaused by environmental influences, assessments of epi-genetic mechanisms (i.e., histone modification, methyl-

Reviews


ation, and microRNAs) are promising approaches (102).For example, a recent study described by Richardson et al.(104) used bioinformatics to identify variants that disturbor create microRNA-recognition elements in a genome-wide fashion, thus providing functional hypotheses forobserved GWAS associations.

The task remaining for the future is to establish cau-sality and functionality and ultimately to move from genediscovery to translation (101). This road will require in-depth post-GWAS analyses across multiple disciplines in-volving genetics, molecular biology, and bioinformatics.

Author Contributions: All authors confirmed they have contributed tothe intellectual content of this paper and have met the following 3 re-quirements: (a) significant contributions to the conception and design,acquisition of data, or analysis and interpretation of data; (b) drafting

or revising the article for intellectual content; and (c) final approval ofthe published article.

Authors’ Disclosures or Potential Conflicts of Interest: Upon man-uscript submission, all authors completed the Disclosures of PotentialConflict of Interest form. Potential conflicts of interest:

Employment or Leadership: S. Blankenberg, guest editor, ClinicalChemistry, AACC.Consultant or Advisory Role: None declared.Stock Ownership: None declared.Honoraria: None declared.Research Funding: P. Diemert, National Genomic Research Net-work Germany.Expert Testimony: None declared.Other Remuneration: P. Diemert, travel costs for lectures ongenetics.

Role of Sponsor: The funding organizations played no role in thedesign of study, choice of enrolled patients, review and interpretationof data, or preparation or approval of manuscript.

References

1. Freely associating [Editorial]. Nat Genet 1999;22:1–2.

2. Cardon LR, Bell JI. Association study designs forcomplex diseases. Nat Rev Genet 2001;2:91–9.

3. Lander ES, Linton LM, Birren B, Nusbaum C,Zody MC, Baldwin J, et al. Initial sequencing andanalysis of the human genome. Nature 2001;409:860–921.

4. International HapMap Consortium. The Interna-tional HapMap Project. Nature 2003;426:789–96.

5. Cambien F, Tiret L. Genetics of cardiovasculardiseases: from single mutations to the wholegenome. Circulation 2007;116:1714–24.

6. Marian AJ, Belmont J. Strategic approaches tounraveling genetic causes of cardiovascular dis-eases. Circ Res 2011;108:1252–69.

7. Pe’er I, Yelensky R, Altshuler D, Daly MJ. Esti-mation of the multiple testing burden forgenomewide association studies of nearly allcommon variants. Genet Epidemiol 2008;32:381–5.

8. Arking DE, Chakravarti A. Understanding cardio-vascular disease through the lens of genome-wide association studies. Trends Genet 2009;25:387–94.

9. Founti P, Topouzis F, van Koolwijk L, TraversoCE, Pfeiffer N, Viswanathan AC. Biobanks andthe importance of detailed phenotyping: a casestudy—the European Glaucoma Society Glauco-GENE project. Br J Ophthalmol 2009;93:577–81.

10. Padmanabhan S, Hastie C, Prabhakaran D,Dominczak AF. Genomic approaches to coro-nary artery disease. Indian J Med Res 2010;132:567–78.

11. Padmanabhan S, Melander O, Hastie C, MenniC, Delles C, Connell JM, et al. Hypertension andgenome-wide association studies: combininghigh fidelity phenotyping and hypercontrols.J Hypertens 2008;26:1275–81.

12. Padmanabhan S, Melander O, Johnson T, DiBlasio AM, Lee WK, Gentilini D, et al. Genome-wide association study of blood pressure ex-tremes identifies variant near UMOD associated

with hypertension. PLoS Genet 2010;6:e1001177.

13. Wong MY, Day NE, Luan JA, Chan KP, WarehamNJ. The detection of gene-environment interac-tion for continuous traits: Should we deal withmeasurement error by bigger studies or bettermeasurement? Int J Epidemiol 2003;32:51–7.

14. McCarthy MI, Abecasis GR, Cardon LR, Gold-stein DB, Little J, Ioannidis JPA, et al. Genome-wide association studies for complex traits: con-sensus, uncertainty and challenges. Nat RevGenet 2008;9:356–69.

15. Witte JS. Genome-wide association studies andbeyond. Annu Rev Public Health 2010;31:9–20.

16. Bis JC, Glazer NL, Psaty BM. Genome-wide as-sociation studies of cardiovascular risk factors:design, conduct and interpretation. J ThrombHaemost 2009;7(Suppl 1):308–11.

17. Nica AC, Montgomery SB, Dimas AS, StrangerBE, Beazley C, Barroso I, et al. Candidate causalregulatory effects by integration of expressionQTLs with complex trait genetic associations.PLoS Genet 2010;6:e1000895.

18. Lloyd-Jones D, Adams R, Carnethon M, De Si-mone G, Ferguson TB, Flegal K, et al. Heartdisease and stroke statistics—2009 update: areport from the American Heart Association Sta-tistics Committee and Stroke Statistics Subcom-mittee. Circulation 2009;119:480–6.

19. Samani NJ, Erdmann J, Hall AS, Hengstenberg C,Mangino M, Mayer B, et al. Genomewide asso-ciation analysis of coronary artery disease.N Engl J Med 2007;357:443–53.

20. McPherson R, Pertsemlidis A, Kavaslar N, Stew-art A, Roberts R, Cox DR, et al. A common alleleon chromosome 9 associated with coronaryheart disease. Science 2007;316:1488–91.

21. Helgadottir A, Thorleifsson G, Manolescu A,Gretarsdottir S, Blondal T, Jonasdottir A, et al. Acommon variant on chromosome 9p21 affectsthe risk of myocardial infarction. Science 2007;316:1491–93.

22. Zeggini E, Weedon MN, Lindgren CM, FraylingTM, Elliott KS, Lango H, et al. Replication ofgenome-wide association signals in UK samples

reveals risk loci for type 2 diabetes. Science2007;316:1336–41.

23. Gschwendtner A, Bevan S, Cole JW, Plourde A,Matarin M, Ross-Adams H, et al. Sequence vari-ants on chromosome 9p21.3 confer risk for ath-erosclerotic stroke. Ann Neurol 2009;65:531–9.

24. Helgadottir A, Thorleifsson G, Magnusson KP,Gretarsdottir S, Steinthorsdottir V, Manolescu A,et al. The same sequence variant on 9p21 as-sociates with myocardial infarction, abdominalaortic aneurysm and intracranial aneurysm. NatGenet 2008;40:217–24.

25. Shete S, Hosking FJ, Robertson LB, Dobbins SE,Sanson M, Malmer B, et al. Genome-wide as-sociation study identifies five susceptibility locifor glioma. Nat Genet 2009;41:899–904.

26. Bishop DT, Demenais F, Iles MM, Harland M,Taylor JC, Corda E, et al. Genome-wide associ-ation study identifies three loci associated withmelanoma risk. Nat Genet 2009;41:920–5.

27. Schaefer AS, Richter GM, Groessner-Schreiber B,Noack B, Nothnagel M, El Mokhtari NE, et al.Identification of a shared genetic susceptibilitylocus for coronary heart disease and periodon-titis. PLoS Genet 2009;5:e1000378.

28. Tregouet D-A, König IR, Erdmann J, MunteanuA, Braund PS, Hall AS, et al. Genome-widehaplotype association study identifies theSLC22A3-LPAL2-LPA gene cluster as a risk locusfor coronary artery disease. Nat Genet 2009;41:283–5.

29. Erdmann J, Grosshennig A, Braund PS, König IR,Hengstenberg C, Hall AS, et al. New suscepti-bility locus for coronary artery disease on chro-mosome 3q22.3. Nat Genet 2009;41:280–2.

30. Myocardial Infarction Genetics Consortium.Genome-wide association of early-onset myo-cardial infarction with single nucleotide poly-morphisms and copy number variants. NatGenet 2009;41:334–41.

31. Reilly MP, Li M, He J, Ferguson JF, Stylianou IM,Mehta NN, et al. Identification of ADAMTS7 asa novel locus for coronary atherosclerosis andassociation of ABO with myocardial infarction inthe presence of coronary atherosclerosis: two



genome-wide association studies. Lancet 2011;377:383–92.

32. Erdmann J, Willenborg C, Nahrstaedt J, PreussM, König IR, Baumert J, et al. Genome-wideassociation study identifies a new locus for cor-onary artery disease on chromosome 10p11.23.Eur Heart J 2011;32:158–68.

33. Wild PS, Zeller T, Schillert A, Szymczak S, Sin-ning CR, Deiseroth A, et al. A genome-wideassociation study identifies LIPA as a suscepti-bility gene for coronary artery disease. Circ Car-diovasc Genet 2011;4:403–12.

34. Schunkert H, König IR, Kathiresan S, Reilly MP,Assimes TL, Holm H, et al. Large-scale associa-tion analysis identifies 13 new susceptibility locifor coronary artery disease. Nat Genet 2011;43:333–8.

35. Wellcome Trust Case Control Consortium.Genome-wide association study of 14,000 casesof seven common diseases and 3,000 sharedcontrols. Nature 2007;447:661–78.

36. Newton-Cheh C, Johnson T, Gateva V, TobinMD, Bochud M, Coin L, et al. Genome-wideassociation study identifies eight loci associatedwith blood pressure. Nat Genet 2009;41:666–76.

37. Levy D, Ehret GB, Rice K, Verwoert GC, LaunerLJ, Dehghan A, et al. Genome-wide associationstudy of blood pressure and hypertension. NatGenet 2009;41:677–87.

38. Fox ER, Young JH, Li Y, Dreisbach AW, KeatingBJ, Musani SK, et al. Association of geneticvariation with systolic and diastolic blood pres-sure among African Americans: the CandidateGene Association Resource study. Hum MolGenet 2011;20:2273–84.

39. Kato N, Takeuchi F, Tabara Y, Kelly TN, Go MJ,Sim X, et al. Meta-analysis of genome-wideassociation studies identifies common variantsassociated with blood pressure variation in eastAsians. Nat Genet 2011;43:531–8.

40. Johnson AD, Newton-Cheh C, Chasman DI, Eh-ret GB, Johnson T, Rose L, et al. Association ofhypertension drug target genes with blood pres-sure and hypertension in 86,588 individuals.Hypertension 2011;57:903–10.

41. Sandhu MS, Waterworth DM, Debenham SL,Wheeler E, Papadakis K, Zhao JH, et al. LDL-cholesterol concentrations: a genome-wide as-sociation study. Lancet 2008;371:483–91.

42. Kathiresan S, Melander O, Guiducci C, Surti A,Burtt NP, Rieder MJ, et al. Six new loci associ-ated with blood low-density lipoprotein choles-terol, high-density lipoprotein cholesterol ortriglycerides in humans. Nat Genet 2008;40:189–97.

43. Willer CJ, Sanna S, Jackson AU, Scuteri A, Bon-nycastle LL, Clarke R, et al. Newly identified locithat influence lipid concentrations and risk ofcoronary artery disease. Nat Genet 2008;40:161–9.

44. Linsel-Nitschke P, Heeren J, Aherrahrou Z, BruseP, Gieger C, Illig T, et al. Genetic variation atchromosome 1p13.3 affects sortilin mRNA ex-pression, cellular LDL-uptake and serum LDLlevels which translates to the risk of coronaryartery disease. Atherosclerosis 2010;208:183–9.

45. Musunuru K, Pirruccello JP, Do R, Peloso GM,Guiducci C, Sougnez C, et al. Exome sequenc-ing, ANGPTL3 mutations, and familial combined

hypolipidemia. N Engl J Med 2010;363:2220–7.46. Teslovich TM, Musunuru K, Smith AV, Edmond-

son AC, Stylianou IM, Koseki M, et al. Biologi-cal, clinical and population relevance of 95 locifor blood lipids. Nature 2010;466:707–13.

47. Vasan RS, Glazer NL, Felix JF, Lieb W, Wild PS,Felix SB, et al. Genetic variants associated withcardiac structure and function: a meta-analysisand replication of genome-wide associationdata. JAMA 2009;302:168–78.

48. Stark K, Esslinger UB, Reinhard W, Petrov G,Winkler T, Komajda M, et al. Genetic associa-tion study identifies HSPB7 as a risk gene foridiopathic dilated cardiomyopathy. PLoS Genet2010;6:e1001167.

49. Villard E, Perret C, Gary F, Proust C, Dilanian G,Hengstenberg C, et al. A genome-wide associ-ation study identifies two loci associated withheart failure due to dilated cardiomyopathy. EurHeart J 2011;32:1065–76.

50. Pfeufer A, van Noord C, Marciante KD, ArkingDE, Larson MG, Smith AV, et al. Genome-wideassociation study of PR interval. Nat Genet2010;42:153–9.

51. Pfeufer A, Sanna S, Arking DE, Muller M, GatevaV, Fuchsberger C, et al. Common variants at tenloci modulate the QT interval duration in theQTSCD Study. Nat Genet 2009;41:407–14.

52. Sotoodehnia N, Isaacs A, de Bakker PIW, DörrM, Newton-Cheh C, Nolte IM, et al. Commonvariants in 22 loci are associated with QRSduration and cardiac ventricular conduction. NatGenet 2010;42:1068–76.

53. Eijgelsheim M, Newton-Cheh C, Sotoodehnia N,de Bakker PIW, Muller M, Morrison AC, et al.Genome-wide association analysis identifiesmultiple loci related to resting heart rate. HumMol Genet 2010;19:3885–94.

54. Ellinor PT, Lunetta KL, Glazer NL, Pfeufer A,Alonso A, Chung MK, et al. Common variants inKCNN3 are associated with lone atrial fibrilla-tion. Nat Genet 2010;42:240–4.

55. Kaab S, Darbar D, van Noord C, Dupuis J,Pfeufer A, Newton-Cheh C, et al. Large scalereplication and meta-analysis of variants onchromosome 4q25 associated with atrial fibril-lation. Eur Heart J 2009;30:813–9.

56. Sinner MF, Pfeufer A, Akyol M, Beckmann B-M,Hinterseer M, Wacker A, et al. The non-synonymous coding IKr-channel variant KCNH2–K897T is associated with atrial fibrillation: resultsfrom a systematic candidate gene-based analysisof KCNH2 (HERG). Eur Heart J 2008;29:907–14.

57. Benjamin EJ, Rice KM, Arking DE, Pfeufer A, vanNoord C, Smith AV, et al. Variants in ZFHX3 areassociated with atrial fibrillation in individualsof European ancestry. Nat Genet 2009;41:879–81.

58. Bezzina CR, Pazoki R, Bardai A, Marsman RF, deJong JSSG, Blom MT, et al. Genome-wide asso-ciation study identifies a susceptibility locus at21q21 for ventricular fibrillation in acute myo-cardial infarction. Nat Genet 2010;42:688–91.

59. Gudbjartsson DF, Holm H, Gretarsdottir S, Thor-leifsson G, Walters GB, Thorgeirsson G, et al. Asequence variant in ZFHX3 on 16q22 associateswith atrial fibrillation and ischemic stroke. NatGenet 2009;41:876–8.

60. Lemmens R, Buysschaert I, Geelen V,Fernandez-Cadenas I, Montaner J, Schmidt H, et

al. The association of the 4q25 susceptibilityvariant for atrial fibrillation with stroke is lim-ited to stroke of cardioembolic etiology. Stroke2010;41:1850–7.

61. Iribarren C, Go AS, Husson G, Sidney S, Fair JM,Quertermous T, et al. Metabolic syndrome andearly-onset coronary artery disease: Is the wholegreater than its parts? J Am Coll Cardiol 2006;48:1800–7.

62. Assimes TL, Holm H, Kathiresan S, Reilly MP,Thorleifsson G, Voight BF, et al. Lack of associ-ation between the Trp719Arg polymorphism inkinesin-like protein-6 and coronary artery dis-ease in 19 case-control studies. J Am Coll Car-diol 2010;56:1552–63.

63. SEARCH Collaborative Group. SLCO1B1 variantsand statin-induced myopathy—a genomewidestudy. N Engl J Med 2008;359:789–99.

64. Mega JL, Close SL, Wiviott SD, Shen L, HockettRD, Brandt JT, et al. Cytochrome P-450 polymor-phisms and response to clopidogrel. N EnglJ Med 2009;360:354–62.

65. Mega JL, Close SL, Wiviott SD, Shen L, WalkerJR, Simon T, et al. Genetic variants in ABCB1and CYP2C19 and cardiovascular outcomes af-ter treatment with clopidogrel and prasugrel inthe TRITON-TIMI 38 trial: a pharmacogeneticanalysis. Lancet 2010;376:1312–9.

66. Mega JL, Simon T, Collet J-P, Anderson JL, Ant-man EM, Bliden K, et al. Reduced-functionCYP2C19 genotype and risk of adverse clinicaloutcomes among patients treated with clopi-dogrel predominantly for PCI: a meta-analysis.JAMA 2010;304:1821–30.

67. Mega JL, Close SL, Wiviott SD, Shen L, HockettRD, Brandt JT, et al. Cytochrome P450 geneticpolymorphisms and the response to prasugrel:relationship to pharmacokinetic, pharmacody-namic, and clinical outcomes. Circulation 2009;119:2553–60.

68. Caldwell MD, Awad T, Johnson JA, Gage BF,Falkowski M, Gardina P, et al. CYP4F2 geneticvariant alters required warfarin dose. Blood2008;111:4106–12.

69. Cooper GM, Johnson JA, Langaee TY, Feng H,Stanaway IB, Schwarz UI, et al. A genome-widescan for common genetic variants with a largeinfluence on warfarin maintenance dose. Blood2008;112:1022–7.

70. International Warfarin Pharmacogenetics Con-sortium. Estimation of the warfarin dose withclinical and pharmacogenetic data. N EnglJ Med 2009;360:753–64.

71. Kathiresan S, Melander O, Anevski D, GuiducciC, Burtt NP, Roos C, et al. Polymorphisms asso-ciated with cholesterol and risk of cardiovascu-lar events. N Engl J Med 2008;358:1240–9.

72. Morrison AC, Bare LA, Chambless LE, Ellis SG,Malloy M, Kane JP, et al. Prediction of coronaryheart disease risk using a genetic risk score: theAtherosclerosis Risk in Communities Study. Am JEpidemiol 2007;166:28–35.

73. Meigs JB, Shrader P, Sullivan LM, McAteer JB,Fox CS, Dupuis J, et al. Genotype score in ad-dition to common risk factors for prediction oftype 2 diabetes. N Engl J Med 2008;359:2208–19.

74. Ripatti S, Tikkanen E, Orho-Melander M, Havu-linna AS, Silander K, Sharma A, et al. A multi-locus genetic risk score for coronary heart

Reviews


disease: case-control and prospective cohortanalyses. Lancet 2010;376:1393–400.

75. Qi L, Ma J, Qi Q, Hartiala J, Allayee H, CamposH. Genetic risk score and risk of myocardialinfarction in Hispanics. Circulation 2011;123:374–80.

76. Paynter NP, Chasman DI, Pare G, Buring JE,Cook NR, Miletich JP, et al. Association betweena literature-based genetic risk score and cardio-vascular events in women. JAMA 2010;303:631–7.

77. Peterson RE, Maes HH, Holmans P, Sanders AR,Levinson DF, Shi J, et al. Genetic risk sum scorecomprised of common polygenic variation isassociated with body mass index. Hum Genet2011;129:221–30.

78. Dehghan A, Dupuis J, Barbalic M, Bis JC, Eiriks-dottir G, Lu C, et al. Meta-analysis of genome-wide association studies in �80 000 subjectsidentifies multiple loci for C-reactive protein lev-els. Circulation 2011;123:731–8.

79. Liu Y, Sanoff HK, Cho H, Burd CE, Torrice C,Mohlke KL, et al. INK4/ARF transcript expressionis associated with chromosome 9p21 variantslinked to atherosclerosis. PLoS One 2009;4:e5027.

80. Jarinova O, Stewart AFR, Roberts R, Wells G,Lau P, Naing T, et al. Functional analysis of thechromosome 9p21.3 coronary artery disease risklocus. Arterioscler Thromb Vasc Biol 2009;29:1671–7.

81. Holdt LM, Beutner F, Scholz M, Gielen S, GabelG, Bergert H, et al. ANRIL expression is associ-ated with atherosclerosis risk at chromosome9p21. Arterioscler Thromb Vasc Biol 2010;30:620–7.

82. Cunnington MS, Santibanez Koref M, MayosiBM, Burn J, Keavney B. Chromosome 9p21 SNPsassociated with multiple disease phenotypescorrelate with ANRIL expression. PLoS Genet2010;6:e1000899.

83. Broadbent HM, Peden JF, Lorkowski S, Goel A,Ongen H, Green F, et al. Susceptibility to coro-nary artery disease and diabetes is encoded bydistinct, tightly linked SNPs in the ANRIL locuson chromosome 9p. Hum Mol Genet 2008;17:806–14.

84. Visel A, Zhu Y, May D, Afzal V, Gong E, Atta-nasio C, et al. Targeted deletion of the 9p21

non-coding coronary artery disease risk intervalin mice. Nature 2010;464:409–12.

85. Holdt LM, Sass K, Gabel G, Bergert H, Thiery J,Teupser D. Expression of Chr9p21 genesCDKN2B (p15INK4b), CDKN2A (p16INK4a, p14ARF)and MTAP in human atherosclerotic plaque.Atherosclerosis 2011;214:264–70.

86. Harismendy O, Notani D, Song X, Rahim NG,Tanasa B, Heintzman N, et al. 9p21 DNA vari-ants associated with coronary artery diseaseimpair interferon-� signalling response. Nature2011;470:264–8.

87. Aulchenko YS, Ripatti S, Lindqvist I, BoomsmaD, Heid IM, Pramstaller PP, et al. Loci influenc-ing lipid levels and coronary heart disease risk in16 European population cohorts. Nat Genet2009;41:47–55.

88. Muendlein A, Geller-Rhomberg S, Saely CH,Winder T, Sonderegger G, Rein P, et al. Signif-icant impact of chromosomal locus 1p13.3 onserum LDL cholesterol and on angiographicallycharacterized coronary atherosclerosis. Athero-sclerosis 2009;206:494–9.

89. Kleber ME, Renner W, Grammer TB, Linsel-Nitschke P, Boehm BO, Winkelmann BR, et al.Association of the single nucleotide polymor-phism rs599839 in the vicinity of the sortilin 1gene with LDL and triglyceride metabolism, cor-onary heart disease and myocardial infarction.The Ludwigshafen Risk and CardiovascularHealth Study. Atherosclerosis 2010;209:492–7.

90. Schadt EE, Molony C, Chudin E, Hao K, Yang X,Lum PY, et al. Mapping the genetic architectureof gene expression in human liver. PLoS Biol2008;6:e107.

91. Zeller T, Wild P, Szymczak S, Rotival M, SchillertA, Castagne R, et al. Genetics and beyond—thetranscriptome of human monocytes and diseasesusceptibility. PLoS One 2010;5:e10693.

92. Kjolby M, Andersen OM, Breiderhoff T, FjorbackAW, Pedersen KM, Madsen P, et al. Sort1, en-coded by the cardiovascular risk locus 1p13.3, isa regulator of hepatic lipoprotein export. CellMetab 2010;12:213–23.

93. Musunuru K, Strong A, Frank-Kamenetsky M,Lee NE, Ahfeldt T, Sachs KV, et al. From non-coding variant to phenotype via SORT1 at the1p13 cholesterol locus. Nature 2010;466:714–9.

94. Dube JB, Johansen CT, Hegele RA. Sortilin: anunusual suspect in cholesterol metabolism:From GWAS identification to in vivo biochemicalanalyses, sortilin has been identified as a novelmediator of human lipoprotein metabolism.Bioessays 2011;33:430–7.

95. Dixon AL, Liang L, Moffatt MF, Chen W, HeathS, Wong KCC, et al. A genome-wide associationstudy of global gene expression. Nat Genet2007;39:1202–7.

96. Göring HHH, Curran JE, Johnson MP, Dyer TD,Charlesworth J, Cole SA, et al. Discovery ofexpression QTLs using large-scale transcrip-tional profiling in human lymphocytes. NatGenet 2007;39:1208–16.

97. Emilsson V, Thorleifsson G, Zhang B, Leonard-son AS, Zink F, Zhu J, et al. Genetics of geneexpression and its effect on disease. Nature2008;452:423–8.

98. Stranger BE, Forrest MS, Clark AG, MinichielloMJ, Deutsch S, Lyle R, et al. Genome-wide as-sociations of gene expression variation in hu-mans. PLoS Genet 2005;1:e78.

99. Coronary Artery Disease (C4D) Genetics Consor-tium. A genome-wide association study in Eu-ropeans and South Asians identifies five newloci for coronary artery disease. Nat Genet 2011;43:339–44.

100. Heinig M, Petretto E, Wallace C, Bottolo L,Rotival M, Lu H, et al. A trans-acting locusregulates an anti-viral expression network andtype 1 diabetes risk. Nature 2010;467:460–4.

101. Glazer NL. Moving beyond genome-wide asso-ciation studies. Circ Cardiovasc Genet 2011;4:91–3.

102. Wang X, Prins BP, Sober S, Laan M, Snieder H.Beyond genome-wide association studies: newstrategies for identifying genetic determinantsof hypertension. Curr Hypertens Rep 2011;13:442–51.

103. 1000 Genomes Project Consortium. A map ofhuman genome variation from population-scalesequencing. Nature 2010;467:1061–73.

104. Richardson K, Lai CQ, Parnell LD, Lee YC, Ordo-vas JM. A Genome-wide survey for SNPs alter-ing microRNA seed sites identifies functionalcandidates in GWAS. BMC Genomics 2011;12:504.



genomewide association studies in cardiovascular disease—an … · 2017-07-16 · genomewide...

Documents