Original Article - COnnecting REpositories vascular disease are found at higher rates among people...

498

Abdominal aortic aneurysm (AAA) is a pathological dila-tation of the aorta that can result in rupture and death. It

is a common disease among older people. A national screen-ing program for 65-year-old men is being introduced in the United Kingdom, with the aim of reducing mortality from rup-tured AAA by offering surgery to those with large aneurysms.

However, surgery is not without risk, and the goals of pri-mary prevention and medical treatment to slow progression are actively pursued through efforts to better understand the pathogenesis of AAA.

Clinical Perspective on p 504

Background—Abdominal aortic aneurysm (AAA) is a common cardiovascular disease among older people and demonstrates significant heritability. In contrast to similar complex diseases, relatively few genetic associations with AAA have been confirmed. We reanalyzed our genome-wide study and carried through to replication suggestive discovery associations at a lower level of significance.

Methods and Results—A genome-wide association study was conducted using 1830 cases from the United Kingdom, New Zealand, and Australia with infrarenal aorta diameter ≥30 mm or ruptured AAA and 5435 unscreened controls from the 1958 Birth Cohort and National Blood Service cohort from the Wellcome Trust Case Control Consortium. Eight suggestive associations with P<1×10−4 were carried through to in silico replication in 1292 AAA cases and 30 503 controls. One single-nucleotide polymorphism associated with P<0.05 after Bonferroni correction in the in silico study underwent further replication (706 AAA cases and 1063 controls from the United Kingdom, 507 AAA cases and 199 controls from Denmark, and 885 AAA cases and 1000 controls from New Zealand). Low-density lipoprotein receptor (LDLR) rs6511720 A was significantly associated overall and in 3 of 5 individual replication studies. The full study showed an association that reached genome-wide significance (odds ratio, 0.76; 95% confidence interval, 0.70–0.83; P=2.08×10−10).

Conclusions—LDLR rs6511720 is associated with AAA. This finding is consistent with established effects of this variant on coronary artery disease. Shared causal pathways with other cardiovascular diseases may present novel opportunities for preventative and therapeutic strategies for AAA. (Circ Cardiovasc Genet. 2013;6:498-504.)

Key Words: aneurysm ◼ cholesterol, LDL ◼ genome-wide association study ◼ lipids

© 2013 American Heart Association, Inc.

Circ Cardiovasc Genet is available at http://circgenetics.ahajournals.org DOI: 10.1161/CIRCGENETICS.113.000165

Received April 9, 2013; accepted September 16, 2013.*Drs Bradley and Hughes contributed equally to this work.The online-only Data Supplement is available at http://circgenetics.ahajournals.org/lookup/suppl/doi:10.1161/CIRCGENETICS.113.000165/-/DC1.Correspondence to Declan T. Bradley, MBChB, MA, PhD, Institute of Pathology, Centre for Public Health, School of Medicine, Dentistry and Biomedical

Sciences, Queen’s University Belfast, Grosvenor Rd, Belfast BT12 6BL, Northern Ireland, United Kingdom. E-mail [email protected]

A Variant in LDLR Is Associated With Abdominal Aortic Aneurysm

Declan T. Bradley, MBChB, MA, PhD*; Anne E. Hughes, PhD*; Stephen A. Badger, MCh, MD; Gregory T. Jones, PhD; Seamus C. Harrison, MD;

Benjamin J. Wright, PhD; Suzannah Bumpstead, BSc; Annette F. Baas, PhD; Sólveig Grétarsdóttir, PhD; Kevin Burnand, MBBS, MS; Anne H. Child, MD; Rachel E. Clough, PhD; Gillian Cockerill, PhD; Hany Hafez, MBBCh, PhD;

D. Julian A. Scott, MD; Robert A.S. Ariëns, PhD; Anne Johnson, RGN; Soroush Sohrabi, MD, PhD; Alberto Smith, PhD; Matthew M. Thompson, MD;

Frank M. van Bockxmeer, PhD; Matthew Waltham, MA, PhD; Stefán E. Matthíasson, MD, PhD; Gudmar Thorleifsson, PhD; Unnur Thorsteinsdottir, PhD;

Jan D. Blankensteijn, MD; Joep A.W. Teijink, MD, PhD; Cisca Wijmenga, PhD; Jacqueline de Graaf, MD, PhD; Lambertus A. Kiemeney, PhD; John B. Wild, MBChB;

Sarah Edkins, PhD; Rhian Gwilliam, PhD; Sarah E. Hunt, PhD; Simon Potter, PhD; Jes S. Lindholt, MD, DMSci, PhD; Jonathan Golledge, MChir; Paul E. Norman, DS;

Andre van Rij, MD; Janet T. Powell, MD, PhD; Per Eriksson, PhD; Kári Stefánsson, MD, PhD; John R. Thompson, PhD; Steve E. Humphries, PhD; Robert D. Sayers, MD;

Panos Deloukas, PhD; Nilesh J. Samani, MD; Matthew J. Bown, MD

Original Article

by guest on February 13, 2017http://circgenetics.ahajournals.org/

Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from

http://circgenetics.ahajournals.org/lookup/suppl/doi:10.1161/CIRCGENETICS.113.000165/-/DC1

mailto:[email protected]

http://circgenetics.ahajournals.org/









Bradley et al LDLR and Abdominal Aortic Aneurysm 499

AAA shares many features with other cardiovascular dis-eases. Lifestyle risk factors include tobacco use1,2 and inactiv-ity.3 Coronary artery disease, cerebral vascular disease, and peripheral vascular disease are found at higher rates among people with AAA, suggesting that there are shared exposures or shared disease pathways.1,4–6 AAA exhibits significant heri-tability and is observed to cluster in families.7 A family history of AAA confers a risk of disease that is comparable in effect size with cigarette smoking.4,5 The role of cholesterol and ath-erosclerosis in AAA is not well understood. Atherosclerosis often accompanies AAA, but whether the association is causative is controversial.8 Higher levels of high-density lipoprotein-cholesterol are associated with protection against AAA,9,10 and raised serum total cholesterol is associated with increased risk.3,4 The effect of serum low-density lipoprotein (LDL) remains unclear, with some reports of increased risk but inconclusive meta-analyses.9–12 Genetic polymorphisms of LRP113, DAB2IP,14 and SORT115 were discovered to be associ-ated with susceptibility to AAA in genome-wide association studies. CDKN2B-AS16–18 and IL6R19 were identified and con-firmed with a high level of certainty in candidate-gene studies. Relatively few genetic risk factors for AAA have been identi-fied to date, which is unexpected considering its heritability, complex pathogenesis, and similarity to other cardiovascular diseases for which many risk associations have been found. We reanalyzed our genome-wide association study with par-ticular emphasis on limiting possible sources of experimen-tal method-related false associations and carried through to replication newly identified putative associations at a lower threshold of discovery association.

Materials and MethodsStudy Design and ParticipantsAll participating centers had approval from appropriate local research ethics committees to conduct their studies, and all participants gave written informed consent. Identifiable information was not provided to collaborating researchers. The study adheres to the tenets of the Declaration of Helsinki. The study design has been described previ-ously.13 A case–control genome-wide association study was carried out using 1830 cases from the United Kingdom, New Zealand, and Australia who had an infrarenal aorta diameter of ≥30 mm on trans-abdominal ultrasound scan or a ruptured AAA and 5435 unscreened population controls from the 1958 Birth Cohort and National Blood Service cohort from the Wellcome Trust Case Control Consortium. The mean age of the cases was 73 years, and the median infrarenal aortic diameter was 4.6 cm. Median age of the Blood Transfusion Service cohort was 45 years, and participants from the 1958 Birth Cohort were aged 52 years at the time of genotyping. Associations with genomic inflation-corrected P<5×10−8 in additive model logistic regression were considered to be significant at a genome-wide level, and P values <1×10−4 were considered to be suggestive of an associa-tion. The threshold of P<10−4 was chosen to be an order of magnitude greater than the original analysis to investigate putative associations that might be apparent after stringent quality control aimed at exclud-ing artifactual false associations. A 2-stage replication study was con-ducted. First, a representative single-nucleotide polymorphism (SNP) from each associated locus was carried through to an in silico replica-tion study that comprised 452 cases and 27 712 controls from Iceland and 840 cases and 2791 controls from The Netherlands. Associations from known loci at CDKN2B-AS1, LRP1, and IL6R, and the previ-ously investigated GPC6, were excluded from further analyses. SNPs were chosen for predicted compatibility with a Sequenom multiplex reaction. SNPs that showed significant association at P<0.05 after Bonferroni correction for multiple testing in stage 1 were carried

through to a laboratory replication study of 706 cases and 1063 con-trols from the United Kingdom (Belfast, Leeds, and Leicester) and 507 cases and 199 controls from Viborg, Denmark. The same diag-nostic criteria were used for cases in the replication studies as in the discovery study. Statistical significance at stage 2 also had P<0.05 af-ter Bonferroni correction for multiple testing. Finally, to confirm any positive findings from stage 2, we used in silico data from a recent genome-wide association study of AAA.15 This New Zealand study consisted of 885 additional cases of AAA and 1000 additional con-trols, none of which was included in the original genome-wide study.

Discovery Study Microarray Experiment and GenotypingCases were genotyped at the Wellcome Trust Sanger Institute us-ing the Illumina Human670 Quad Custom version 1 bead chip, and the WTCCC2 common controls were genotyped using the Illumina 1.2M Duo Custom version 1a bead chip. The raw idat data produced at the Sanger Institute were normalized and genotyped using the corrected robust linear mixture model crlmm package20 in R ver-sion 2.14.0 on the Queen’s University Belfast High-Performance Computing Dell Cluster using SOCK parallelization in 4 batches of ≈1900 individuals. Quality control steps are described in the online-only Data Supplement.

Discovery Study Association AnalysisThe genome-wide study was analyzed using an additive model which assumes that each extra copy of an allele contributes the same change in odds and that 2 copies of an allele confer twice the odds of 1 copy. The ancestry of participants was analyzed by principal components analysis (Methods in the online-only Data Supplement). The associa-tion analysis was conducted using logistic regression with ancestry eigenvectors included as covariates for the 5 statistically significant axes to correct the analysis for residual population stratification. The analysis was performed using PLINK version 1.07.

Replication Stage Genotyping and AnalysisA Sequenom iPLEX was designed using MySequenom software and conducted according to the manufacturer’s protocols. Primer se-quences are shown in Table III in the online-only Data Supplement. Briefly, a multiplex polymerase chain reaction was conducted on an ABI 9700 Dual 384-well thermal cycler, unincorporated dNTPs were neutralized using a shrimp alkaline phosphatase reaction, primer ex-tension was conducted using the iPLEX termination mix and further cycling, and conditioning resin was added before nanodispensing and measurement of extension product on a Sequenom MassArray. Cluster plots were inspected using Typer 4.0 Software. In silico analyses were provided from Icelandic (452 cases and 27 712 con-trols) and Dutch (840 cases and 2791 controls) AAA genome-wide association studies. Association P values from the Icelandic analyses were adjusted for population relatedness and potential population stratification by the genomic control factor (λ=1.143). The Dutch study did not require correction (λ=1)14 and was, therefore, used without correction in our study. LDL receptor (LDLR) rs6511720 was genotyped as part of a Sequenom iPLEX in a UK replication group (729 cases and 1104 controls) and a replication group from Viborg, Denmark (496 cases and 176 controls). Twelve UK cases, 25 UK controls, 6 Viborg cases, and 4 Viborg controls were excluded because of low genotyping rate (<90%). The remaining genotyping rate was 99.9%.

The association analyses were conducted using additive model lo-gistic regression in PLINK for the laboratory study with no covari-ates. Data from the Iceland and Netherlands studies were analyzed in NEMO.21

The additional New Zealand genome-wide study (885 cases and 1000 controls) was conducted using the Affymetrix SNP6 GeneChip array and tested for association between AAA and LDLR rs6511720 using additive model logistic regression in PLINK. Full details of this study have been described previously.15


Dow

nloaded from


500 Circ Cardiovasc Genet October 2013

Logistic regression analyses with adjustment for age (United Kingdom, Denmark, and New Zealand replication studies) and LDL-cholesterol (adjusted for statin use; New Zealand) were also conducted.

Meta-analysisThe results from the individual studies were combined using inverse variance-weighted meta-analysis in METASOFT. Between-study heterogeneity was investigated by estimating the heterogeneity index (I2) also in METASOFT. Fixed-effects and Han Eskin random-effects meta-analyses were conducted. In the absence of heterogeneity, a fixed-effects model was used, and in its presence, Han Eskin ran-dom-effects model was used.22 Han Eskin meta-analysis is a form of random-effects meta-analysis for the use in genetic studies in the presence of heterogeneity.

ImputationTargeted imputation of LDLR was carried out using IMPUTE223 with reference to all populations in The 1000 Genomes Project Phase 1 integrated version 3 data release. Thirty Markov chain Monte Carlo iterations, including a burn in of 10 iterations and 40 conditioning states, were used. Population size (-Ne) was set to 20 000. Imputed regions were each 2 Mb in size. Visualization was performed using Locus Zoom.

ResultsIn the final discovery study, 1755 cases (1554 men and 201 women) and 5314 controls (2706 men and 2608 women) were genotyped for 503 892 SNPs that passed all quality control thresholds. The final genotyping rate was 99.85%. A quantile–quantile plot is shown in Figure II in the online-only Data Supplement.

No SNPs had association P values reaching genome-wide significance level in the discovery study. Twenty-three SNPs in 12 loci had suggestive association P values of <1×10−4 (Figure 1; Table II in the online-only Data Supplement). Eight SNPs were carried through to in silico replication, one

of which remained significant after Bonferroni correction for multiple testing (LDLR rs6511720 A; odds ratio [OR], 0.79 [0.69–0.89]; P=2.34×10−4; Figure 2). The same allele was associated significantly in an independent laboratory repli-cation (OR, 0.73 [0.57–0.94]; P=0.016; Table). Han Eskin random-effects meta-analysis of combined replication stages 1 and 2 produced an OR of 0.78 (0.69–0.87; P=1.95×10−5) for rs6511720 A, and both the discovery and the replication steps give an OR of 0.78 (0.72–0.85), an association that reached genome-wide significance (P=6.97×10−9).

The additional New Zealand genome-wide study data set provided further independent replication of the association (OR, 0.68 [0.50–0.93]; P=4.8×10−3). The combination of the New Zealand results with the rest of the study by Han Eskin random-effects meta-analysis resulted in an OR of 0.76 (0.70–0.83; P=2.08×10−10).

Age-adjusted analyses were performed for Denmark, United Kingdom, and New Zealand replication groups. The associations found in Denmark and New Zealand were statistically signifi-cant after adjustment for age (Denmark: OR, 0.52 [0.36–0.77; P=8.9×10−4]; New Zealand: OR, 0.68 [0.50–0.93; P=0.016]). The UK replication group (which did not show association in unadjusted analysis) did not show association after adjustment for age (OR, 1.03 [0.82–1.29; P=0.80]; Table IV in the online-only Data Supplement). Adjustment of the New Zealand analy-ses for LDL-cholesterol (corrected for statin use) demonstrated that the association of LDLR and AAA was independent of LDL-cholesterol (adjusted OR, 0.69 [0.52–0.92]; P=0.012).

DiscussionWe report that rs6511720 in LDLR is a novel genetic risk fac-tor for AAA. The effect associated with this SNP is similar to that reported for LRP1, DAB2IP, IL6R, and SORT1. LDLR is

Figure 1. Manhattan plot of abdominal aortic aneurysm genome-wide association study. Manhattan plot of discovery study showing genomic control-adjusted −log10 (P values) for additive model from logistic regression incorporating eigenvectors as covariates. High-lighted single-nucleotide polymorphisms are the new confirmed locus (low-density lipoprotein receptor [LDLR]) and 4 previously detected loci (low-density lipoprotein–related protein 1 [LRP1; alpha-2-macroglobulin receptor]; CDKN2B antisense RNA 1 [CDKN2B-AS1]; DAB2 interacting protein [DAB2IP]; and interleukin 6 receptor [IL6R])


Dow

nloaded from



the simplest of a family of LDL-cholesterol cell surface recep-tors24 that also includes the AAA-associated LRP1.13

LDLR rs6511720 is one of the most important LDL-cholesterol–associated variants in genome-wide studies of lipid traits.25–31 Importantly, a relationship between AAA and LDL-cholesterol levels is not presently established. The most complete meta-analysis of the relationship was inconclu-sive, and individual reports are inconsistent.9,32–34 Our report of association between LDLR polymorphism and AAA sus-ceptibility suggests that either LDL-cholesterol is important in AAA pathogenesis or LDLR polymorphism has an effect on another causal pathway. The adjusted results of the New Zealand study suggest that the association is independent of LDL-cholesterol levels.

Despite this uncertainty, our finding is consistent with other knowledge of AAA pathogenesis. LDLR-knockout (Ldl−/−) mice are used as disease models for AAA.35 The same LDLR allele (rs6511720 A) that we report is associated with protec-tion against AAA is also associated with protection against coronary artery25,28,29,36,37 and carotid artery atherosclerotic plaque formation.38 A recent meta-analysis of the use of lipid-lowering therapy in AAA to reduce the rate of expansion has shown some benefit. This finding is consistent with LDL-cholesterol having a direct role in AAA pathogenesis.34

We are aware of the limitations of this study. The strict quality control measures used in this analysis allowed us to identify the LDLR locus by excluding false-positive associa-tions. We accepted the possibility of also excluding some true associations as a consequence. Meta-analysis of genome-wide association studies for AAA may offer a future opportunity to identify further AAA loci that have been excluded. The rs6511720 SNP was associated with AAA at a similar level in the original study analysis, as were many additional likely spurious associations. Although the LDLR locus could have been identified in the original discovery analysis at P<10−4, Bonferroni correction for the attempted replication of many additional putative associations would have vastly reduced the power to replicate true associations. The ORs in the Denmark and New Zealand studies were unaffected by adjustment for age, and the association remained significant after inclusion of

age as a covariate in a logistic regression model. The UK rep-lication study that showed no association with rs6511720 was unaffected by the inclusion of age in the logistic regression.

The use of the standard 3-cm definition of AAA has limita-tions as a phenotype for genetic studies because it is arbitrary, and risk of AAA by this measure is affected by height and sex. Future studies might benefit from using a definition of AAA that incorporates an individual’s stature. Alternatively, longitudinal studies of disease progression have the poten-tial to be useful, although different gene associations might be found to affect progression. Future studies to determine whether LDL-cholesterol levels are causal in the pathogen-esis of AAA could use Mendelian randomization that would require prospective standardized collection of data for serum lipids and the use of lipid-lowering therapies. We do not have sufficient extra information to explore heterogeneity in the rs6511720 association. Possible causes for differences in the effect size between the components of the study include chance and gene–environment interactions because of differ-ences in exposures (tobacco or lipid-lowering therapy) whose prevalence differs over time or between places and case–mix variation (whereby screening-detected AAA, elective surgical cases, and ruptured AAA are differently represented between-study components). Two of the 6 component studies in this investigation did not show a significant association between rs6511720 A and AAA. This is similar to the situation in the 3 other published genome-wide studies of AAA.13–15 Possible reasons for this include low power in individual component studies, type II error, and the winner’s curse.39 Our study made use of the Wellcome Trust Case Control Consortium shared, unscreened controls. The use of population controls offers the advantages of reduced testing for participants, shared costs between projects, and increased power because of the ready availability of participants. A potential drawback is a reduced effect size caused by the presence of a small number of people who would meet the case definition within the control group and the increased risk of type II errors.40

Our finding that LDLR polymorphism affects odds of AAA suggests that study of relationships between LDL-cholesterol

Table. LDLR rs6511720 A Allele

Sample Set Cases (n) Controls (n) Case AF Control AF OR (95% CI) P Value I 2

Discovery study 1755 5314 0.097 0.123 0.76 (0.67–0.86) 3.05×10-5 …

Stage 1 replication

The Netherlands 840 2791 0.093 0.128 0.70 (0.59–0.83) 5.69×10−5 …

Iceland 452 27 712 0.091 0.099 0.91 (0.72–1.15) 0.42 …

Combined 1292 30 503 … … 0.79 (0.69–0.90) 2.34×10−4 68.2

Stage 2 replication

United Kingdom 716 1078 0.111 0.113 0.98 (0.79–1.21) 0.84 …

Denmark 490 172 0.077 0.140 0.52 (0.36–0.77) 8.86×10−4 …

Combined 1206 2456 … … 0.73 (0.57–0.94) 0.014 87.3

Combined stages 1 and 2 replication 2498 32 959 … … 0.78 (0.69–0.87) 1.95×10−5 74.3

New Zealand genome-wide study 885 1000 0.062 0.088 0.68 (0.50–0.93) 0.005 …

All combined 5138 39 273 … … 0.76 (0.70–0.83) 2.08×10−10 61.0

AF indicates allele frequency; CI, confidence interval; and OR, odds ratio.


Dow

nloaded from



homeostasis and AAA pathogenesis may be productive in identifying therapeutic and preventative strategies.

AcknowledgmentsWe thank the participants for their involvement. This study makes use of data generated by the Wellcome Trust Case Control Consortium. A full list of the investigators who contributed to the generation of the data is available from www.wtccc.org.uk. We thank Robert Scharpf and Matthew Ritchie for helpful discussions about using their crlmm software. Drs Bradley and Hughes thank Louis Lau of Belfast Health and Social Care Trust for help recruiting participants.

Sources of FundingFunding for the project was provided by the Wellcome Trust un-der awards 076113 and 085475. This study was supported by the Wellcome Trust (grant number 084695) and the Research and Development Office of the Public Health Agency, Northern Ireland.

DisclosuresDr Bradley was supported by a doctoral fellowship from the Research and Development Office of the Public Health Agency, Northern Ireland. Dr Bown is supported by a Senior Lectureship from the Higher Education Funding Council for England and the Circulation Foundation. J. Golledge receives research funding from the National Health and Medical Research Council and the Office of Health and Medical Research. Dr Clough was funded by the National Institute of Health Research Biomedical Research Center at Guy’s & St. Thomas’ National Health Service Foundation Trust. Additional funding for source cohorts was from the National Health and Medical Research Council (Australia), The Dunhill Medical Trust (Leicester), and The Garfield Weston Trust for Medical Research

into Heart Surgery (Leeds). The replication data sets from Iceland and he Netherlands were funded through a grant in the European Community’s Seventh Framework Program (FP7/2007–2013) and the Fighting Aneurysmal Disease Project grant agreement HEALTH-F2-2008-200647. Dr Child is supported by the St. George’s Hospital NHS Trust Special Trustees and the Peter and Sonia Field Charitable Trust. Dr Harrison is supported by a British Heart Foundation Clinical Training Fellowship (no. FS/11/16/28696). Dr Baas was supported by a grant from The Netherlands Heart Foundation (2009T001). Drs Humphries and Samani hold a chair funded by the British Heart Foundation. Drs Bown, Sayers, and Samani are supported by the National Institute of Health Research Leicester Cardiovascular Biomedical Research Unit. deCODE genetics is a biotechnology company, and authors affiliated with deCODE own stock or stock options in the company. The other authors report no conflicts.

AppendixFrom the Centre for Public Health, School of Medicine, Den-tistry and Biomedical Sciences, Queen’s University Belfast, Bel-fast, United Kingdom (D.T.B., A.E.H., S.A.B.); Department of Surgery, University of Otago, Dunedin, New Zealand (G.T.J.); Center for Cardiovascular Genetics, Institute of Cardiovascular Science, University College London, London, United King-dom (S.C.H., S.E.H.); Department of Cardiovascular Sciences, University of Leicester and The NIHR Leicester Cardiovas-cular Biomedical Research Unit, Glenfield Hospital, Leices-ter, United Kingdom (B.J.W., J.B.W., R.D.S., N.J.S., M.J.B.); Genetics of Complex Traits in Humans Group, Wellcome Trust Sanger Institute, Cambridge, United Kingdom (S.B., S.E., R.G.,

Figure 2. Regional association plot for low-density lipoprotein receptor (LDLR). Regional association plots for LDLR showing genotyped and imputed discovery association genomic control-corrected −log10(P values).


Dow

nloaded from

http://www.wtccc.org.uk



S.E.H., S.P., P.D.); Department of Medical Genetics, University Medical Center Utrecht, Utrecht, The Netherlands (A.F.B.); Population Genomics, deCODE Genetics, Reykjavik, Iceland, & Faculty of Medicine, University of Iceland, Reykjavik, Ice-land (S.G., G.T., U.T., K.S.); Academic Department of Surgery, Kings College London, British Heart Foundation Centre of Research Excellence and NIHR Biomedical Research Centre at Kings Health Partners, St. Thomas’ Hospital, London, United Kingdom (K.B., R.E.C., A.S., M.W.); Departments of Cardiac and Vascular Sciences (A.H.C.) and Vascular Surgery (G.C.), St. George’s, University of London, London, United Kingdom; Department of Vascular Surgery, St. Richard’s Hospital, Chich-ester, United Kingdom (H.H.); Division of Cardiovascular and Diabetes Research, MCRC, Leeds Institute of Genetics, Health and Therapeutics (LIGHT), University of Leeds, Leeds, United Kingdom (D.J.A.S., R.A.S.A., A.J., S.S.); St. George’s Vascu-lar Institute, London, United Kingdom (M.M.T.); Department of Surgery, University of Western Australia, Crawley, Australia (F.M.v.B., P.E.N.); Laekning Medical Clinics, Reykjavik, Ice-land (S.E.M.); Department of Vascular Surgery, VU Medical Center, Amsterdam, The Netherlands (J.D.B.); Department of Vascular Surgery, Catharina Hospital, Eindhoven, The Nether-lands (J.A.W.T.); Department of Genetics, UMC Groningen, Groningen, The Netherlands (C.W.); Departments of Internal Medicine (J.d.G.) and Health Evidence and Urology, Radboud University Nijmegen Medical Centre, Nijmegen, The Nether-lands (L.A.K.); Department of Vascular Surgery, Viborg Hospi-tal, Viborg, Denmark (J.S.L.); Queensland Research Centre for Peripheral Vascular Disease, James Cook University, Towns-ville, Australia (J.G.); Department of Surgery and Cancer, Imperial College London, London, United Kingdom (J.T.P.); Atherosclerosis Research Unit, CMM, Karolinska Institute, Stockholm, Sweden (P.E.); and Department of Health Sciences, University of Leicester, Leicester, United Kingdom (J.R.T.).

References 1. Alcorn HG, Wolfson SK Jr, Sutton-Tyrrell K, Kuller LH, O’Leary D.

Risk factors for abdominal aortic aneurysms in older adults enrolled in The Cardiovascular Health Study. Arterioscler Thromb Vasc Biol. 1996;16:963–970.

2. Duncan JL, Harrild KA, Iversen L, Lee AJ, Godden DJ. Long term out-comes in men screened for abdominal aortic aneurysm: prospective cohort study. BMJ. 2012;344:e2958.

3. Forsdahl SH, Singh K, Solberg S, Jacobsen BK. Risk factors for abdomi-nal aortic aneurysms: a 7-year prospective study: the Tromsø Study, 1994-2001. Circulation. 2009;119:2202–2208.

4. Kent KC, Zwolak RM, Egorova NN, Riles TS, Manganaro A, Mos-kowitz AJ, et al. Analysis of risk factors for abdominal aortic an-eurysm in a cohort of more than 3 million individuals. J Vasc Surg. 2010;52:539–548.

5. Lederle FA, Johnson GR, Wilson SE, Chute EP, Hye RJ, Makaroun MS, et al. The aneurysm detection and management study screening program: validation cohort and final results. Aneurysm Detection and Manage-ment Veterans Affairs Cooperative Study Investigators. Arch Intern Med. 2000;160:1425–1430.

6. Golledge J, Clancy P, Jamrozik K, Norman PE. Obesity, adipokines, and abdominal aortic aneurysm: Health in Men study. Circulation. 2007;116: 2275–2279.

7. Clifton MA. Familial abdominal aortic aneurysms. Br J Surg. 1977;64:765–766.

8. Johnsen SH, Forsdahl SH, Singh K, Jacobsen BK. Atherosclerosis in abdominal aortic aneurysms: a causal event or a process running in parallel? The Tromsø study. Arterioscler Thromb Vasc Biol. 2010;30:1263–1268.

9. Golledge J, van Bockxmeer F, Jamrozik K, McCann M, Norman PE. As-sociation between serum lipoproteins and abdominal aortic aneurysm. Am J Cardiol. 2010;105:1480–1484.

10. Takagi H, Manabe H, Umemoto T. A meta-analysis of association between serum lipoproteins and abdominal aortic aneurysm. Am J Cardiol. 2010;106:753–754.

11. Rizzo M, Krayenbühl PA, Pernice V, Frasheri A, Battista Rini G, Berneis K. LDL size and subclasses in patients with abdominal aortic aneurysm. Int J Cardiol. 2009;134:406–408.

12. Hobbs SD, Claridge MW, Quick CR, Day NE, Bradbury AW, Wilmink AB. LDL cholesterol is associated with small abdominal aortic aneu-rysms. Eur J Vasc Endovasc Surg. 2003;26:618–622.

13. Bown MJ, Jones GT, Harrison SC, Wright BJ, Bumpstead S, Baas AF, et al. Abdominal aortic aneurysm is associated with a variant in low-density lipoprotein receptor-related protein 1. Am J Hum Genet. 2011;89: 619–627.

14. Gretarsdottir S, Baas AF, Thorleifsson G, Holm H, den Heijer M, de Vries JP, et al. Genome-wide association study identifies a sequence variant within the DAB2IP gene conferring susceptibility to abdominal aortic an-eurysm. Nat Genet. 2010;42:692–697.

15. Jones GT, Bown MJ, Gretarsdottir S, Romaine SP, Helgadottir A, Yu G, et al. A sequence variant associated with sortilin-1 (SORT1) on 1p13.3 is independently associated with abdominal aortic aneurysm. Hum Mol Genet. 2013;22:2941–2947.

16. Helgadottir A, Thorleifsson G, Magnusson KP, Grétarsdottir S, Steinthors-dottir V, Manolescu A, et al. The same sequence variant on 9p21 associates with myocardial infarction, abdominal aortic aneurysm and intracranial aneurysm. Nat Genet. 2008;40:217–224.

17. Bown MJ, Braund PS, Thompson J, London NJ, Samani NJ, Sayers RD. Association between the coronary artery disease risk locus on chromo-some 9p21.3 and abdominal aortic aneurysm. Circ Cardiovasc Genet. 2008;1:39–42.

18. Thompson AR, Golledge J, Cooper JA, Hafez H, Norman PE, Humphries SE. Sequence variant on 9p21 is associated with the presence of abdomi-nal aortic aneurysm disease but does not have an impact on aneurysmal expansion. Eur J Hum Genet. 2009;17:391–394.

19. Harrison SC, Smith AJP, Jones GT, Swerdlow DI, Rampuri R, Bown MJ, et al. Interleukin 6 receptor pathways in abdominal aortic aneurysm. Eur Heart J. 2012. doi: 10.1093/eurheartj/ehs354.

20. Scharpf RB, Irizarry RA, Ritchie ME, Carvalho B, Ruczinski I. Using the R Package crlmm for Genotyping and Copy Number Estimation. J Stat Softw. 2011;40:1–32.

21. Guillaume F, Rougemont J. Nemo: an evolutionary and population genet-ics programming framework. Bioinformatics. 2006;22:2556–2557.

22. Han B, Eskin E. Random-effects model aimed at discovering associations in meta-analysis of genome-wide association studies. Am J Hum Genet. 2011;88:586–598.

23. Howie BN, Donnelly P, Marchini J. A flexible and accurate genotype im-putation method for the next generation of genome-wide association stud-ies. PLoS Genet. 2009;5:e1000529.

24. Go GW, Mani A. Low-density lipoprotein receptor (LDLR) family orches-trates cholesterol homeostasis. Yale J Biol Med. 2012;85:19–28.

25. Willer CJ, Sanna S, Jackson AU, Scuteri A, Bonnycastle LL, Clarke R, et al. Newly identified loci that influence lipid concentrations and risk of coronary artery disease. Nat Genet. 2008;40:161–169.

26. Teslovich TM, Musunuru K, Smith AV, Edmondson AC, Stylianou IM, Koseki M, et al. Biological, clinical and population relevance of 95 loci for blood lipids. Nature. 2010;466:707–713.

27. Trompet S, de Craen AJM, Postmus I, Ford I, Sattar N, Caslake M, et al. Replication of LDL GWAs hits in PROSPER/PHASE as validation for future (pharmaco)genetic analyses. BMC Med Genet. 2011;12:131.

28. Aulchenko YS, Ripatti S, Lindqvist I, Boomsma D, Heid IM, Pramstaller PP, et al. Loci influencing lipid levels and coronary heart disease risk in 16 European population cohorts. Nat Genet. 2009;41:47–55.

29. Anand SS, Xie C, Paré G, Montpetit A, Rangarajan S, McQueen MJ, et al. Genetic variants associated with myocardial infarction risk factors in over 8000 individuals from five ethnic groups: the INTERHEART Genet-ics Study. Circ Cardiovasc Genet. 2009;2:16–25.

30. Kathiresan S, Willer CJ, Peloso GM, Demissie S, Musunuru K, Schadt EE, et al. Common variants at 30 loci contribute to polygenic dyslipid-emia. Nat Genet. 2009;41:56–65.

31. Miljkovic I, Yerges-Armstrong LM, Kuller LH, Kuipers AL, Wang X, Kammerer CM, et al. Association analysis of 33 lipoprotein candidate genes in multi-generational families of African ancestry. J Lipid Res. 2010;51:1823–1831.


Dow

nloaded from



32. Rizzo M, Krayenbühl PA, Pernice V, Frasheri A, Battista Rini G, Berneis K. LDL size and subclasses in patients with abdominal aortic aneurysm. Int J Cardiol. 2009;134:406–408.

33. Hobbs SD, Claridge MW, Quick CR, Day NE, Bradbury AW, Wilmink AB. LDL cholesterol is associated with small abdominal aortic aneu-rysms. Eur J Vasc Endovasc Surg. 2003;26:618–622.

34. Takagi H, Yamamoto H, Iwata K, Goto S, Umemoto T. Effects of statin therapy on abdominal aortic aneurysm growth: a meta-analysis and meta-regression of observational comparative studies. Eur J Vasc Endovasc Surg. 2012;44:287–292.

35. Daugherty A, Cassis LA. Mouse models of abdominal aortic aneurysms. Arterioscler Thromb Vasc Biol. 2004;24:429–434.

36. Franceschini N, Muallem H, Rose KM, Boerwinkle E, Maeda N. Low density lipoprotein receptor polymorphisms and the risk of coronary heart

disease: the Atherosclerosis Risk in Communities Study. J Thromb Hae-most. 2009;7:496–498.

37. Martinelli N, Girelli D, Lunghi B, Pinotti M, Marchetti G, Malerba G, et al. Polymorphisms at LDLR locus may be associated with coronary artery disease through modulation of coagulation factor VIII activity and independently from lipid profile. Blood. 2010;116:5688–5697.

38. Bis JC, Kavousi M, Franceschini N, Isaacs A, Abecasis GR, Schminke U, et al. Meta-analysis of genome-wide association studies from the CHARGE consortium identifies common variants associated with carotid intima media thickness and plaque. Nat Genet. 2011;43:940–947.

39. Garner C. Upward bias in odds ratio estimates from genome-wide associa-tion studies. Genet Epidemiol. 2007;31:288–295.

40. Moskvina V, Holmans P, Schmidt KM, Craddock N. Design of case-controls studies with unscreened controls. Ann Hum Genet. 2005;69(Pt 5):566–576.

CLINICAL PERSPECTIVEAbdominal aortic aneurysm (AAA) is a common cardiovascular disease that carries a high risk of mortality in the event of rupture. It can cluster in families, and a family history of the disease is associated with increased risk. The aim of genomic studies of AAA is to identify genetic variants that affect the disease process. This will further the understanding of the pathogenesis of the disease and potentially highlight molecules and biological pathways as possible therapeutic targets, biomarkers, or risk markers. In a comprehensive reanalysis of our genome-wide study, we found that a genetic variant in the low-density lipoprotein receptor (LDLR) gene is associated with risk of AAA. The LDLR gene is important in other cardiovascular diseases and traits. Mutations of this gene cause familial hypercholesterolaemia, and common variations of the gene are associated with coronary artery disease risk. The rs6511720 polymorphism that we found to be associated with AAA is one of the most important genetic variants associated with serum LDL-cholesterol concentrations. However, despite several studies of this subject, the role of LDL-cholesterol in AAA pathogenesis remains controversial and unproven. By showing that an LDLR polymorphism is associated with AAA, our study suggests that function of the LDLR receptor might be important in the pathogenesis of AAA, which may be mediated through LDL-cholesterol levels or other functions of the protein. Future studies will be required to understand the functional basis for the association observed.


Dow

nloaded from


Sayers, Panos Deloukas, Nilesh J. Samani and Matthew J. BownPowell, Per Eriksson, Kári Stefánsson, John R. Thompson, Steve E. Humphries, Robert D.Simon Potter, Jes S. Lindholt, Jonathan Golledge, Paul E. Norman, Andre van Rij, Janet T.

de Graaf, Lambertus A. Kiemeney, John B. Wild, Sarah Edkins, Rhian Gwilliam, Sarah E. Hunt,Unnur Thorsteinsdottir, Jan D. Blankensteijn, Joep A.W. Teijink, Cisca Wijmenga, Jacqueline Frank M. van Bockxmeer, Matthew Waltham, Stefán E. Matthíasson, Gudmar Thorleifsson,

Robert A.S. Ariëns, Anne Johnson, Soroush Sohrabi, Alberto Smith, Matthew M. Thompson, Burnand, Anne H. Child, Rachel E. Clough, Gillian Cockerill, Hany Hafez, D. Julian A. Scott,

Benjamin J. Wright, Suzannah Bumpstead, Annette F. Baas, Sólveig Grétarsdóttir, Kevin Declan T. Bradley, Anne E. Hughes, Stephen A. Badger, Gregory T. Jones, Seamus C. Harrison,

Is Associated With Abdominal Aortic AneurysmLDLRA Variant in

Print ISSN: 1942-325X. Online ISSN: 1942-3268 Copyright © 2013 American Heart Association, Inc. All rights reserved.

Dallas, TX 75231is published by the American Heart Association, 7272 Greenville Avenue,Circulation: Cardiovascular Genetics

doi: 10.1161/CIRCGENETICS.113.0001652013;6:498-504; originally published online September 17, 2013;Circ Cardiovasc Genet.

http://circgenetics.ahajournals.org/content/6/5/498World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://circgenetics.ahajournals.org/content/suppl/2013/09/17/CIRCGENETICS.113.000165.DC1Data Supplement (unedited) at:

http://circgenetics.ahajournals.org//subscriptions/

is online at: Circulation: Cardiovascular Genetics Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer information about this process is available in the

requested is located, click Request Permissions in the middle column of the Web page under Services. FurtherCenter, not the Editorial Office. Once the online version of the published article for which permission is being

can be obtained via RightsLink, a service of the Copyright ClearanceCirculation: Cardiovascular Geneticsin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:


Dow

nloaded from

http://circgenetics.ahajournals.org/content/6/5/498

http://circgenetics.ahajournals.org/content/suppl/2013/09/17/CIRCGENETICS.113.000165.DC1

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://circgenetics.ahajournals.org//subscriptions/


1

SUPPLEMENTAL MATERIAL

Supplemental Methods

Discovery Study Quality Control

The microarray manufacturer provided a list of SNPs that have been reported to show

discordant clustering due to beadpool changes between genotyping platforms used, which

resulted in exclusion of 47 SNPs. One sample with low signal-to-noise ratio (<5) in crlmm

was excluded. Base calls with a probability confidence of <90% were excluded. One hundred

and fifty three individuals were excluded as relatives (>12.5% identical) or duplicates.

Principal components analysis was conducted using Eigensoft v3.0 with reference to the 11

populations in the HapMap phase 3 data release,1 and 428 individuals with outlying ancestry

were excluded (Supplementary Figure 1). Individuals with a genotyping rate < 99% were

excluded. Those with an inbreeding coefficient >0.08 (suggestive of a sample problem) or <-

0.08 (suggesting contamination) were excluded.

Case and control samples were genotyped at different times using different platforms,

which can cause systematic differences in genotyping,2 and is a potential source of false

associations in this study. It was therefore necessary to apply marker-level quality control

measures to the case group as well as the control group. SNPs with minor allele frequency

<1% in case or control groups were excluded. SNPs with overall genotyping rate <99% were

excluded. Assessment for deviation from Hardy Weinberg equilibrium was conducted using

the method of Wigginton et al.3 as implemented in PLINK v1.07. Stringent arbitrary Hardy

Weinberg equilibrium (HWE) thresholds for exclusion were applied (controls, P<0.01,

16,664 SNPs; cases, P<1x10-4, 6,641 SNPs) to avoid detection of false associations. Seven

thousand and sixteen SNPs with P<0.05 for informative missingness were excluded. To

identify aberrant results due to technical problems, case and control genotype counts were

2

each separately compared to HapMap Phase 3 CEU counts using allelic chi-squared tests.

Arbitrary thresholds (P<1x10-3 for controls, 7,160 SNPs; P<1x10-10 for cases, 793 SNPs)

were used to exclude from further analyses SNPs that may have been unreliably genotyped,

with a less stringent threshold for cases to allow for true differences due to the phenotype.

After the association analysis was conducted, SNPs with association P< 1x10-4 were

excluded if there was not a further association with P<1x10-4 within 100kb. This resulted in

exclusion of 138 SNPs. Cluster plots of normalised intensity and genotype were inspected for

the remaining SNPs with P<1x10-4. Correction for genomic inflation (λ=1.03) was conducted

using the median chi-square in PLINK.4 All further analyses were based on corrected P

values. SNPs in or near LRP1, IL6R and CDKN2B-AS (genes known to be associated), and

GPC6 (which we investigated previously11) were not carried through to the replication study.

References: 1. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet. 2006; 38:904-909. 2. Scharpf R, Irizarry R, Ritchie M, Carvalho B, Ruczinsk I. Using the R Package crlmm for Genotyping and Copy Number Estimation. J Stat Softw. 2011; 40:1-32. 3. Wigginton JE, Cutler DJ, Abecasis GR. A note on exact tests of Hardy-Weinberg equilibrium. Am J Hum Genet. 2005; 76:887-893. 4. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Maller J, Sklar P, de Bakker PI, Daly MJ, Sham PC. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007; 81:559-575.

3

Supplementary Table 1. Discovery Genome-wide Association Study Results

CHR SNP Position (base pairs) Effect allele MAF Cases MAF Controls Odds Ratio (95% CI) PGC 1 rs4240872 152702819 G 0.25 0.22 1.21 (1.10 - 1.32) 7.21x10-5 1 rs7514452 152704708 G 0.21 0.18 1.22 (1.11 - 1.34) 6.83x10-5 2 rs4854251 95400974 C 0.19 0.16 1.23 (1.12 - 1.36) 5.53x10-5 2 rs10874471 95401820 G 0.19 0.16 1.23 (1.12 - 1.36) 5.53x10-5 4 rs12507728 11142794 A 0.44 0.49 0.84 (0.78 - 0.91) 1.14x10-5 8 rs2597384 20594816 A 0.47 0.51 0.85 (0.79 - 0.92) 5.98x10-5 8 rs12680047 128828043 G 0.33 0.37 0.84 (0.77 - 0.91) 2.79x10-5 9 rs10116277 22071397 A 0.52 0.48 1.17 (1.09 - 1.27) 5.23x10-5 9 rs4977574 22088574 G 0.53 0.48 1.19 (1.10 - 1.28) 1.76x10-5 11 rs11031122 30504014 G 0.29 0.25 1.21 (1.11 - 1.32) 2.03x10-5 11 rs11031125 30509905 A 0.29 0.25 1.20 (1.11 - 1.31) 2.84x10-5 11 rs495997 100440990 G 0.37 0.41 0.85 (0.79 - 0.92) 8.89x10-5 12 rs4759044 55816937 G 0.42 0.46 0.85 (0.79 - 0.92) 5.40x10-5 12 rs1466535 55820737 A 0.32 0.37 0.84 (0.77 - 0.91) 2.12x10-5 13 rs9301909 93247199 A 0.33 0.29 1.20 (1.10 - 1.30) 2.14x10-5 13 rs2892667 93254940 G 0.33 0.29 1.20 (1.10 - 1.30) 2.56x10-5 13 rs9524249 93281734 C 0.33 0.29 1.20 (1.10 - 1.30) 2.34x10-5 13 rs1411513 93317951 A 0.33 0.29 1.20 (1.10 - 1.30) 2.55x10-5 13 rs9524264 93326419 C 0.36 0.32 1.18 (1.09 - 1.28) 6.59x10-5 15 rs12441037 58889949 G 0.03 0.05 0.64 (0.52 - 0.79) 6.49x10-5 15 rs6494226 58893178 A 0.03 0.05 0.64 (0.52 - 0.79) 5.59x10-5 19 rs6511720 11063306 A 0.10 0.12 0.76 (0.67 - 0.86) 3.05x10-5 19 rs2228671 11071912 A 0.10 0.13 0.77 (0.68 - 0.88) 8.37x10-5 CHR, chromosome; SNP, single nucleotide polymorphism; MAF, minor allele frequency; CI, confidence interval; PGC, P value corrected for genomic inflation.

4

Supplementary Table 2. Results of in silico stage 1 replication study.

SNP, single nucleotide polymorphism; OR, odds ratio; P, P value; *Han Eskin random effects meta-analysis performed in the presence of heterogeneity, otherwise fixed effects meta-analysis.

Netherlands Iceland Meta-analysis

SNP Effect allele OR P OR P OR P I2

rs10874471 G 0.97 (0.83 - 1.13) 0.68 0.96 (0.80 - 1.14) 0.63 0.96 (0.84 - 1.09) 0.54 0 rs12507728 A 0.89 (0.80 - 0.99) 0.039 1.12 (0.98 - 1.28) 0.087 1.00 (1.00 - 1.01) 0.12 85.2* rs2597384 C 0.90 (0.81 - 1.01) 0.068 1.02 (0.89 - 1.17) 0.77 0.95 (0.92 - 1.08) 0.23 45.1* rs12680047 G 1.01 (0.90 - 1.12) 0.93 1.08 (0.94 - 1.25) 0.24 1.04 (0.94 - 1.15) 0.43 0 rs11031125 A 0.97 (0.85 - 1.09) 0.58 0.98 (0.84 - 1.15) 0.84 0.97 (0.87 - 1.08) 0.58 0 rs495997 G 0.88 (0.78 - 0.98) 0.019 0.93 (0.80 - 1.07) 0.28 0.89 (0.82 - 0.98) 0.012 0 rs6494226 A 1.85 (1.18 - 2.89) 7.0x10-3 0.69 (0.41 - 1.17) 0.16 1.14 (1.01 - 1.30) 0.041 87.2* rs6511720 A 0.70 (0.59 - 0.83) 5.7x10-5 0.91 (0.72 - 1.15) 0.41 0.79 (0.69 - 0.89) 2.34x10-4 68.2*

5

Supplementary Table 3. Sequenom iPLEX Oligonucleotides

SNP Primer Sequence

Extension

Direction

rs6511720 Extension CCTAAGACTTCCTTAACATTTT Reverse

rs6511720 Primer-1 ACGTTGGATGGTGTATCTCACTCACCAATC

rs6511720 Primer-2 ACGTTGGATGAACAAGGCCTTGCCTAAGAC

Supplementary Table 4. rs6511720 A and AAA in the UK and Denmark Replication Studies

with Adjustment.

Study

(Adjustment)

Cases Controls Odds Ratio (95%

Confidence Interval)

P

UK (Age) 714 1076 1.03 (0.82 – 1.29) 0.80

Denmark (Age) 490 172 0.52 (0.36 – 0.77) 8.9x10-4

New Zealand

(Age)

885 1,000 0.68 (0.50 – 0.93) 0.016

New Zealand

(LDL-cholesterol)

885 1,000 0.69 (0.52 – 0.92) 0.012

6

Supplementary Figure 1 Principal components analysis of the study population with reference

to HapMap populations

Principal components analysis of the study population with reference to the International

HapMap populations, showing first and second axes, and first and third axes plotted, before (left)

and after (right) removal of population outliers. ASW, African ancestry in Southwest USA;

CEU, Utah residents with Northern and Western European ancestry from the CEPH collection;

CHB, Han Chinese in Beijing, China; CHD Chinese in Metropolitan Denver, Colorado; GIH,

Gujarati Indians in Houston, Texas; JPT, Japanese in Tokyo, Japan; LWK, Luhya in Webuye,

Kenya; MEX, Mexican ancestry in Los Angeles, California; MKK, Maasai in Kinyawa, Kenya;

TSI, Toscans in Italy; YRI, Yoruba in Ibadan, Nigeria.

7

Supplementary Figure 2. Quantile-quantile plot of logistic regression P values

Quantile-quantile plot showing P values before (black) and after (green) correction for genomic

inflation.

Original Article - COnnecting REpositories vascular disease are found at higher rates among people...

Documents

Transcript of Original Article - COnnecting REpositories vascular disease are found at higher rates among people...