Abstract

N-terminal pro-B-type natriuretic peptide (NT-proBNP) is a strong predictor of mortality in coronary artery disease and is widely employed as a prognostic biomarker. However, a causal relationship between NT-proBNP and clinical endpoints has not been established. We have performed a genome-wide association and Mendelian randomization study of NT-proBNP. We used a discovery set of 3740 patients from the PLATelet inhibition and patient Outcomes (PLATO) trial, which enrolled 18 624 patients with acute coronary syndrome (ACS). A further set of 5492 patients, from the same trial, was used for replication. Genetic variants at two novel loci (SLC39A8 and POC1B/GALNT4) were associated with NT-proBNP levels and replicated together with the previously known NPPB locus. The most significant SNP (rs198389, pooled P = 1.07 × 10−15) in NPPB interrupts an E-box consensus motif in the gene promoter. The association in SLC39A8 is driven by a deleterious variant (rs13107325, pooled P = 5.99 × 10−10), whereas the most significant SNP in POC1B/GALNT4 (rs11105306, pooled P = 1.02 × 10−16) is intronic. The SLC39A8 SNP was associated with higher risk of cardiovascular (CV) death (HR = 1.39, 95% CI: 1.08–1.79, P = 0.0095), but the other loci were not associated with clinical endpoints. We have identified two novel loci to be associated with NT-proBNP in patients with ACS. Only the SLC39A8 variant, but not the NPPB variant, was associated with a clinical endpoint. Due to pleotropic effects of SLC39A8, these results do not suggest that NT-proBNP levels have a direct effect on mortality in ACS patients. PLATO Clinical Trial Registration: www.clinicaltrials.gov; NCT00391872.

Introduction

N-terminal pro-B-type natriuretic peptide (NT-proBNP) is widely employed as a prognostic biomarker in clinical settings and predicts mortality in patients with coronary artery disease (CAD) (1,2), acute coronary syndromes (ACS) (3), heart failure (4) and atrial fibrillation (5). The natriuretic peptide precursor B gene (NPPB) encodes the precursor protein pre-proBNP that is secreted by cardiac muscle cells. Pre-proBNP is then processed into proBNP which is cleaved into two parts, the active cardiac natriuretic peptide B-type (BNP) molecule and the inactive NT-proBNP (6). BNP is a peptide hormone that is released as a response to volume overload and increased filling pressure in the heart, e.g. in heart failure. In the kidney, BNP signaling results in discharge of sodium and water through the urine, reducing the blood volume, the systemic blood pressure and afterload. Patients with chronic renal dysfunction have, together with higher BNP levels, a higher atrial pressure, systemic pressure and ventricular mass. It is not clear whether this association is due to poor renal responsiveness to BNP or if BNP is increased as a response to decreased filtration and clearance by the kidney (7). In ACS patients, BNP levels might be raised also in the absence of clinical heart failure but still probably reflecting left ventricular dysfunction (8).

Previous genetic studies have identified variants close to NPPB to be associated with NT-proBNP (9) and BNP (10) levels, and blood pressure (11). The processing of proBNP into BNP and NT-proBNP is not well characterized. One possibility is that corin, a serine peptidase known to be involved in proANP processing (12), is also responsible for proBNP processing. However, this has not been supported by experimental data (13). Another enzyme, for which experimental data supports an involvement in proBNP processing (13), is furin (paired basic amino acid cleaving enzyme). SNPs close to or within the gene encoding furin (FURIN) have been associated with blood pressure (14) and risk of CAD (15). These clinical effects could be mediated by proBNP.

In this study, we performed a genome-wide association study (GWAS) for NT-proBNP levels in patients with ACS. To evaluate the causal effect of NT-proBNP on clinical endpoints, we carried out a Mendelian randomization (MR) study using the identified SNPs.

Results

Baseline characteristics and clinical variables

A total of 9978 patients passed genotype quality control (QC) in the combined discovery and replication data of which 9232 had NT-proBNP measurements available at baseline. Baseline characteristics for patients with NT-proBNP measured at baseline are presented in Supplementary Material, Tables S1 and S2. The median age across cohorts was 62 years. The most commonly self-reported ethnic origin was European (99%) with the remainder reported as African or Asian. In the discovery (N = 3740) and replication (N = 5492) cohort, respectively, the median (interquartile range [IQR]) NT-proBNP at baseline was 459 ng/L (161–1219) and 455 ng/L (144–1322). No major clinical differences were seen between the discovery and replication cohorts (Supplementary Material, Tables S1 and S2). We identified 18 variables (P < 0.10) to be included as covariates in the association analyses (Supplementary Material, Table S3).

Discovery GWAS, replication of initial findings and pooled analyses

A total of 3740 individuals and 1 796 367 SNPs passed the QC in the discovery cohort, and 5492 individuals and 705 132 SNPs in the replication cohort. After imputations and post-imputation QC, 10 962 468 and 10 891 933 SNPs remained in the discovery and replication cohort, respectively (Supplementary Material, Table S4). GWAS was performed separately in the discovery and replication cohorts, as well as in the pooled cohort (discovery + replication). No inflation was observed in any of the GWASes (lambda = 1.018, 1.019 and 1.021 in the discovery, replication and pooled cohort, respectively).

In the discovery GWAS, a total of 148 SNPs met the threshold of P < 1 × 10−5 (Supplementary Material, Table S5). All 148 SNPs from the discovery GWAS had been successfully genotyped or imputed in the replication cohort, and a total of three SNPs (Supplementary Material, Table S5) replicated (P < 0.00034, Bonferroni adjusted P < 0.05). Of these, two SNPs are adjacent on chromosome 12 (rs11105306, P = 6.54 × 10−12 and rs10858848, P = 1.17 × 10−9) and one on chromosome 4 (rs6855246, P = 7.64 × 10−5).

Analyses of the imputed data in the pooled sample sets resulted in 172 genome-wide significant SNPs (Fig. 1A; Supplementary Material, Table S6). These SNPs included the two previous association signals on chromosome 4 (lowest P = 5.99 × 10−10 for rs13107325) and chromosome 12 (lowest P = 1.02 × 10−16 for rs11105306) and one additional signal on chromosome 1 (lowest P = 1.07 × 10−15 for rs198389). Conditional analyses were performed by adjusting for the most significant SNPs (rs198389, rs13107325 and rs11105306) from the three loci identified in the pooled analyses but no additional genome-wide significant signals were seen (Fig. 1B).

Figure 1.

Manhattan plot of the associations with NT-proBNP levels. (A) The primary analyses in the pooled analyses. (B) The conditional analyses, adjusted for the most significant SNP (rs198389, rs13107325 and rs11105306) from each of the three genome-wide significant regions in the primary analyses.

Figure 1.

Manhattan plot of the associations with NT-proBNP levels. (A) The primary analyses in the pooled analyses. (B) The conditional analyses, adjusted for the most significant SNP (rs198389, rs13107325 and rs11105306) from each of the three genome-wide significant regions in the primary analyses.

Functional annotations of the results

The chromosome 1 locus harbors NPPB, encoding pre-proBNP, which has previously been identified in a GWAS for NT-proBNP (10). Our most significant SNP (rs198389) is located in the NPPB promoter (Fig. 2A) and has previously been shown to influence promoter activity (16). We identified that rs198389 is located in an E-box consensus sequence 5′-CANNTG-3′ (on both the plus and minus strand), where it alters the CANNTG motif to CGNNTG on the plus strand and to CANNCG on the minus strand (disrupting the palindromic nature of the binding site). Our most significant SNP in the chromosome 4 region (rs13107325) is located in SLC39A8 (Fig. 2B). rs13107325 is a missense variant, which results in an Ala->Thr amino acid change at position 391 of the protein. This substitution is predicted to be deleterious to the protein (Sorting Intolerant from Tolerant (SIFT) score = 0.01–0.02 and Polymorphism Phenotyping (PolyPhen) score = 0.195–0.839, for the three different protein-coding transcripts available). The association signal on chromosome 12 consists of a large number of SNPs in strong linkage disequilibrium that span both POC1B and GALNT4 (Fig. 2C). A number of SNPs are located in the predicted promoter regions of POC1B and GALNT4, however, the most significant SNP is located in an intronic region with no obvious regulatory function.

Figure 2.

Detailed view of the associated loci. (A) NPPB, (B) SLC39A8 and (C) POC1B/GALNT4 regions. The figures shows the –log(p) from the primary analyses (in black) and from the conditional analyses (in red), adjusted for the top SNPs: rs198389, rs13107325 and rs11105306.

Figure 2.

Detailed view of the associated loci. (A) NPPB, (B) SLC39A8 and (C) POC1B/GALNT4 regions. The figures shows the –log(p) from the primary analyses (in black) and from the conditional analyses (in red), adjusted for the top SNPs: rs198389, rs13107325 and rs11105306.

Multivariable analyses, serial measurements, sensitivity analyses and genotype effect on NT-proBNP levels

Including all three SNPs in a multiple regression model resulted in similar beta estimates as in the GWAS (Table 1). There were no significant interactions between any of the three SNPs (P > 0.05). NT-proBNP was also measured in a subset of the patients at a second visit, approximately 1 month later. There were no large differences in NT-proBNP levels even though there was a change toward higher median and lower mean (Supplementary Material, Table S2). For rs13107325 (SLC39A8), we did not see any significant difference in association between baseline measurements compared with the second visit. However, the association between rs198389 (NPPB) and NT-proBNP level was stronger at the second visit (P = 2.1 × 10−3). On the other hand, association for rs11105306 (at POC1B/GALNT4) tended to be stronger at baseline compared with the second visit (P = 0.052). The additive genetic effects (beta estimates) on the log-transformed baseline NT-proBNP levels are small, ranging from 0.15 to 0.21, which equals 0.10 to 0.14 standard units. In total, the three SNPs explain less than 1% of the variation in NT-proBNP levels. There was an additive effect of all alleles also on untransformed NT-proBNP levels (Table 2). We also performed sensitivity analyses by stratifying on ethnic origin. Because the number of participants with Asian and African origin in this study were very limited (79 and 53 patients respectively), only analyses for the patients with European ancestry (N = 9252) were performed (Table 1). The results of the European ancestry only analyses did not differ from the pooled analyses, confirming that the global selection of samples did not influence the results of from our study.

Table 1.

Multiple regression model including all significant covariates and randomization treatment

Explanatory variables Baseline
All patients, N = 9232
 
Baseline—same patients as measured at the second visit, N = 3105
 
Second visit, N = 3105
 
Biomarker differencea, N = 3105
 
Baseline European descent only, N = 9097
 
Betab SEc Beta P-value Beta Se Beta P-value Beta Se Beta P-value Beta Se Beta P-value Beta Se Beta P 
rs11105306 (POC1B/GALNT4−0.183 0.022 3.3 × 10−16 −0.159 0.037 1.9 × 10−5 −0.088 0.032 6.0 × 10−3 0.071 0.037 5.2 × 10−2 −0.183 0.023 5.6× 10−16 
rs13107325 (SLC39A80.208 0.035 2.0 × 10−9 0.230 0.057 5.0 × 10−5 0.220 0.049 6.6 × 10−6 −0.011 0.056 8.5 × 10−1 0.206 0.035 2.6× 10−9 
rs198389 (NPPB0.152 0.019 1.6 × 10−15 0.087 0.031 4.9 × 10−3 0.181 0.027 1.3 × 10−11 0.094 0.030 2.1 × 10−3 0.150 0.019 4.1 × 10−15 
Explanatory variables Baseline
All patients, N = 9232
 
Baseline—same patients as measured at the second visit, N = 3105
 
Second visit, N = 3105
 
Biomarker differencea, N = 3105
 
Baseline European descent only, N = 9097
 
Betab SEc Beta P-value Beta Se Beta P-value Beta Se Beta P-value Beta Se Beta P-value Beta Se Beta P 
rs11105306 (POC1B/GALNT4−0.183 0.022 3.3 × 10−16 −0.159 0.037 1.9 × 10−5 −0.088 0.032 6.0 × 10−3 0.071 0.037 5.2 × 10−2 −0.183 0.023 5.6× 10−16 
rs13107325 (SLC39A80.208 0.035 2.0 × 10−9 0.230 0.057 5.0 × 10−5 0.220 0.049 6.6 × 10−6 −0.011 0.056 8.5 × 10−1 0.206 0.035 2.6× 10−9 
rs198389 (NPPB0.152 0.019 1.6 × 10−15 0.087 0.031 4.9 × 10−3 0.181 0.027 1.3 × 10−11 0.094 0.030 2.1 × 10−3 0.150 0.019 4.1 × 10−15 

aBiomarker difference is the difference between the NT-proBNP measurement at the second visit (∼1 month after baseline) and the baseline measurement at admission to the hospital.

bBeta is the effect per copy of the minor allele.

cSE, standard error.

Table 2.

NT-proBNP levels (ng/L) for the genotypes at the NPPB, SLC39A8 and POC1B/GALNT4 loci

SNP (gene) Genotype Min First quartile Median Mean Third quartile Maximum N 
rs198389 (NPPBA/A 2.5 126 407 1124 1137 38 410 3138 
A/G 2.5 158 475 1261 1326 31 460 4551 
G/G 176 521 1406 1390 34 830 1709 
rs1310732 (SLC39A8C/C 2.5 148 445 1213 1233 38 410 7983 
C/T 159.2 541 1386 1566 30 730 1346 
T/T 30 229 645 1735 2380 16 790 69 
rs11105306 (POC1B/GALNT4C/C 2.5 167 499 1362 1391 38 410 5285 
C/T 2.5 136 418 1102 1158 34 060 3504 
T/T 110 348 1001 1030 34 830 609 
SNP (gene) Genotype Min First quartile Median Mean Third quartile Maximum N 
rs198389 (NPPBA/A 2.5 126 407 1124 1137 38 410 3138 
A/G 2.5 158 475 1261 1326 31 460 4551 
G/G 176 521 1406 1390 34 830 1709 
rs1310732 (SLC39A8C/C 2.5 148 445 1213 1233 38 410 7983 
C/T 159.2 541 1386 1566 30 730 1346 
T/T 30 229 645 1735 2380 16 790 69 
rs11105306 (POC1B/GALNT4C/C 2.5 167 499 1362 1391 38 410 5285 
C/T 2.5 136 418 1102 1158 34 060 3504 
T/T 110 348 1001 1030 34 830 609 

Associations with other biomarkers and clinical characteristics

Some of the identified SNPs have been reported to be associated with other phenotypes (14,17–20). In our study, the minor allele of rs13107325 was associated with increased body mass index (BMI) (P = 0.000143) and decreased high-density lipoprotein (HDL) (P = 0.036), and we saw a trend of association with lower supine systolic blood pressure at admission (P = 0.070), which agrees with previous studies. However, neither rs198389 nor rs11105306 were associated with any of the above traits (P > 0.1) or with history of diabetes mellitus.

MR and multivariable Cox regression analyses

Hazard ratios (HR) were estimated for rs13107325 (SLC39A8), rs198389 (NPPB) and rs11105306 (POC1B/GALNT4), in relation to the primary composite endpoint (myocardial infarction [MI], stroke or cardiovascular [CV] death) and for CV death alone (Table 3). Neither rs198389 nor rs11105306 were associated with any of the clinical endpoints. For rs13107325, the minor allele (Fig. 3) was associated with an increased risk of CV death (HR = 1.39, 95% CI: 1.08–1.79, P = 0.0095), but not with the primary composite endpoint (Table 3). Because rs13107325 is also associated with HDL and blood pressure, we added clinically relevant baseline variables (log-transformed HDL levels, supine systolic and diastolic blood pressure, treatment at day of randomization with lipid lowering agent, beta -blocker, angiotensin-converting enzyme inhibitor, angiotensin II receptor and calcium channel blocker, as well as previous history of hypertension and dyslipidemia) as covariates in the regression model. This resulted in a slight increase in HR for rs13107325 for CV death (HR = 1.43, 95% CI: 1.11–1.85, P = 0.0071). In contrast, when adding the log-transformed NT-proBNP levels as a covariate, the HR of rs13107325 decreased (HR = 1. 22, 95% CI: 0.93–1.58, P = 0.14).

Table 3.

Cox proportional hazards model adjusted for relevant clinical covariates and the first four principal components

SNP Gene Primary composite endpoint
 
CV death
 
HR (95% CI) Cox regression P HR (95% CI) Cox regression P 
rs198389 NPPB 0.95 (0.87–1.05) 0.31 0.98 (0.84–1.14) 0.77 
rs13107325 SLC39A8 0.94 (0.78–1.12) 0.48 1.39 (1.08–1.79) 0.0095 
rs11105306 POC1B/GALNT4 1.05 (0.94–1.17) 0.37 1.08 (0.9–1.29) 0.41 
SNP Gene Primary composite endpoint
 
CV death
 
HR (95% CI) Cox regression P HR (95% CI) Cox regression P 
rs198389 NPPB 0.95 (0.87–1.05) 0.31 0.98 (0.84–1.14) 0.77 
rs13107325 SLC39A8 0.94 (0.78–1.12) 0.48 1.39 (1.08–1.79) 0.0095 
rs11105306 POC1B/GALNT4 1.05 (0.94–1.17) 0.37 1.08 (0.9–1.29) 0.41 
Figure 3.

The Kaplan–Meier plot. Cumulative incidence of CV deaths for different genetic polymorphisms in rs13107325 (SLC39A8). The heterozygous and homozygous for the rare allele are merged due to the low number of homozygous individuals.

Figure 3.

The Kaplan–Meier plot. Cumulative incidence of CV deaths for different genetic polymorphisms in rs13107325 (SLC39A8). The heterozygous and homozygous for the rare allele are merged due to the low number of homozygous individuals.

Discussion

This GWAS, in a large cohort of ACS patients, identified and replicated two novel genetic loci (SLC39A8 and GALNT4) associated with NT-proBNP levels. NPPB (rs198389), known to be associated with both NT-proBNP (9) and BNP (10) levels, was also replicated. For the first time, we also showed that rs198389 interrupts the consensus sequence of an E-box, which is one of the most well-characterized binding sites located in the promoter of the gene. rs198389 has previously been associated with risk of heart failure (21) and diabetes (22) and was also used in a MR study to show the causative effects of BNP and NT-proBNP on diabetes (23). Another of the most significant SNPs in NPPB identified, rs632793, has been associated with reduced rate of CV readmission in patients with CAD (24). However, in our study, we did not detect any association between rs198389 genotypes and the primary composite endpoint or CV death alone. We also found no association of this SNP with blood pressure or history of diabetes.

Our most significant SNP (rs13107325) in SLC39A8 is a missense variant where the minor allele is associated with increased levels of NT-proBNP and with increased HR of CV death. The minor allele of rs13107325 has previously been associated with decreased levels of HDL (17,18), increased BMI (19), decreased risk of schizophrenia (25) and decreased systolic and diastolic blood pressure (14,20). These associations with HDL, BMI and systolic blood pressure were also replicated in our study. The minor allele of rs13107325 has been suggested to have been subjected to positive selection and is found only in European populations (26). SLC39A8 encodes SLC39A8, also known as ZIP8, a member of the solute carrier family that transports metal ions and is involved in the uptake of zinc, manganese, cadmium and iron from blood in vascular endothelial cells and dispersal to different tissues and organs (27–29). Gene expression is induced by infection and inflammatory stimuli (30) and, recently, SLC39A8 was shown to be a transcriptional target for NF-κB that down-regulate pro-inflammatory responses through zinc-mediated regulation (31). It was therefore suggested that SLC39A8 plays an important role in the regulation of innate immune function by influencing zinc metabolism. rs13107325 is a missense variant that is predicted to be deleterious to the protein. However, no studies have been reported that quantify variation in transporter activity between individuals with polymorphisms in SLC39A8. One possibility is that SLC39A8 interacts with NT-proBNP in controlling blood pressure, by transporting ions or other molecules for urine excretion through the urine. Poor transport activity by SLC39A8 could result in decreased clearance of some molecules, increased BNP release and subsequent lowering of blood pressure. This would be consistent with the deleterious variant being associated with increased NT-proBNP levels and decreased blood pressure. However, it has previously been suggested that the association between the deleterious SLC39A8 variant and decreased HDL levels might be mediated by an inflammatory mechanism (18).

Another of the SNPs identified is located at the POC1B/GALNT4 locus. GALNT4 encodes GalNAcT, a member of the UDP-N-acetyl-alpha-d-galactosamine:polypeptide N-acetylgalactosaminyltransferase (GalNAcT; EC 2.4.1.41) family of enzymes. These enzymes initiate Mucin-type O-glycosylation, which is the most common type of O-glycosylation (32), and GalNAcT4 has been suggested to preferably glycosylate the P-selectin glycoprotein ligand (PSGL-1). It has previously been shown that genetic variants at GALNT4 are associated with risk of CV disease, perhaps via glycosylation of PSGL-1 and P-selectin-mediated cell–cell interactions (33). A genetic variant (rs7136259) located almost 200 kb upstream of GALNT4 has been shown to be associated with risk of CAD in a Chinese population (34), and this variant was also shown to be an expression quantitative trait locus (eQTL) for GALNT4 (35). This variant is nominally associated (P = 0.008) with NT-proBNP levels in our data. However, one of our top SNPs in the region (rs11105298, P = 1.54 × 19−14) has also been suggested to be an eQTL for GALNT4 in monocytes (36). Taken together, these results suggest that GalNAcT4 may play an important role in regulating NT-proBNP production. Even though the mechanisms behind proBNP processing into BNP and NT-proBNP have not been fully elucidated, the cleavage is known to be suppressed by O-glycosylation (37). It is possible that GalNAcT4 also glycosylates proBNP, and thereby influences the rate of proBNP processing into BNP and NT-proBNP.

There are some limitations with this study. First, the PLATO clinical trial was a worldwide multi-center study. This results in a mix of ethnicities that might not be a perfect study design for a GWAS. To reduce these influences, we included principal components in the analyses as well as performed sensitivity analyses that included patients with European ancestry only. Second, the strength of the instruments used in the MR is limited. The NPPB SNP explain less than 1% of the variation in NT-proBNP levels and will therefore serve as quite poor instrument in the MR. It is therefore possible that performing the MR in a much larger sample size (>>10 000 patients) could support the idea of NT-proBNP having causal effects on disease outcomes.

In summary, we have identified two novel genetic loci (SLC39A8 and POC1B/GALNT4) to be associated with NT-proBNP levels and replicated the previously known NPPB association. We also put the rs198389 variant at NPPB in a regulatory context by showing that this variant interrupts a consensus sequence of an E-box. An association between NT-proBNP and mortality in patients with CAD has previously been described (1). Neither the POC1B/GALNT4 nor the NPPB variants showed any association with the clinical outcomes of our study. However, we did identify an increased risk of CV death for the deleterious variant (rs13107325) at SLC39A8. This variant is also associated with BMI, HDL and blood pressure. However, adjusting for these and other clinical variables increased the estimated risk of CV death for the rs13107325 genotype, suggesting that genetically increased NT-proBNP levels might have a causal effect on CV death in patients with ACS.

Materials and Methods

Study population

The PLATelet inhibition and patient Outcomes (PLATO) trial (NCT00391872) was a randomized clinical trial that evaluated the effect of ticagrelor versus clopidogrel in 18 624 patients with ACS (38,39). Baseline venous blood samples for biomarker investigations were collected in ethylenediaminetetraacetic acid (EDTA) tubes within 24 h of admission, prior to the administration of study medication. Additional samples were taken at a second visit at the hospital, approximately 1 month (mean 31.28, standard deviation 7.97 days) after randomization. All participants provided informed consent in accordance with the Declaration of Helsinki. An additional venous blood sample was collected in EDTA tubes for genetic analyses for 10 013 individuals. Participation in the genetic substudy was voluntary and required an additional consent form at the time of enrollment into the genetic substudy.

Biomarker laboratory analysis

Blood samples were obtained by direct venipuncture. Plasma was frozen and stored in aliquots at −70°C until central analysis. NT-proBNP levels were measured using a sandwich immunoassay on the Cobas® Analytics e601 Immunoanalyzer (Roche Diagnostics).

Clinical endpoints

The primary outcome variable in the main PLATO trial was time to the composite endpoint of CV death, MI excluding silent MI and stroke. CV death was defined as death from vascular causes and included CV deaths, cerebrovascular deaths and any other deaths for which there was no clearly documented nonvascular cause. Details regarding endpoint definitions have been published previously (38). Patients were followed up for a median of 9 (range 6–12) months.

GWAS

A two-stage design was used for the GWAS with a discovery cohort and a replication cohort. The discovery consisted of a random set of 3998 individuals genotyped using the Illumina HumanOmni2.5–4v1 (Omni2.5) BeadChip (Illumina, San Diego, CA), and the replication cohort of the remaining 6015 patients, which were genotyped using the Illumina Infinium HumanOmniExpressExome-8v1 BeadChip (Illumina, San Diego, CA). Genotyping was performed according to the manufacturer's instructions. Analysis of the raw data was performed using Illumina GenomeStudio 2011.1 (Illumina, San Diego, CA), the Illumina's Infinium assay and project sample generated cluster files (40,41). QC were performed using the whole genome association analysis toolset PLINK v1.07 (42) (http://pngu.mgh.harvard.edu/purcell/plink) in discovery and replication cohorts separately (Supplementary Material, Table S4).

Classical multidimensional scaling

Pairwise kinship matrices were calculated for the discovery and replication cohorts separately, using genotyped autosomal SNPs (Supplementary Material, Data 1). This was calculated using the ibs function (weighted by the allele frequency) implemented in GenABEL (43), which computes a matrix of average identical-by-state (IBS) values for all individuals. The kinship matrix was used to calculate the pairwise distance matrix between individuals followed by classical multidimensional scaling (MDS) analyses using 10 dimensions. For the pooled (discovery and replication sets) analyses, the principal components were calculated in a similar way for all SNPs that were genotyped and passed QC thresholds in both the discovery and replication cohort (Supplementary Material, Table S4).

Imputation of genotypes at unassayed variants

Imputations were performed in the discovery and replication cohorts separately (Supplementary Material, Table S4), using a pre-phasing approach (44) in SHAPEIT version 2 (45), and IMPUTE2 (version 2.2.2) (46) for imputations. The 1000 Genomes (47) Phase I integrated variant set (NCBI build b37, Mar 2012, updated 24 August 2012) were accessed from the IMPUTE2 Web site and used as reference panel. QC of imputed SNPs (Supplementary Material, Table S4) and merging of the discovery and replication cohorts were performed using QCTOOL version 1.3 (http://www.well.ox.ac.uk/~gav/qctool).

Statistical analysis

The selection of clinical covariates to adjust for in the genetic analyses was performed in a stage-wise approach from a list of demographic and clinical background variables (full list in the Supplementary Material, Table S7). First, univariate estimates for the effects of each of the clinical variables on NT-proBNP were calculated (in the discovery cohort) using linear regression. All variables showing univariate association to NT-proBNP (P < 0.1) were included in a multiple linear regression model. Variables with P < 0.1 in the multiple linear regression model were included as covariates in the GWAS. Our study includes individuals with different genetic backgrounds (Supplementary Material, Fig. S1). To adjust for possible population stratification, all GWAS analyses were adjusted for the four first genetic principal components (48) from the MDS.

GWAS was performed using an additive genetic model. NT-proBNP was analyzed on a natural logarithm scale by linear regression with the palinear function in the ProbABEL package (49). All SNPs with P < 10−5 in the discovery cohort were taken forward for replication. The threshold for significance for a successful replication was set to P = 0.05/number of SNPs for which replication was attempted. In analyses of the pooled sample sets, we used a threshold for significance of P = 0.5 × 10−8. Summary association statistics for all SNPs (discovery, replication and pooled cohort) can be found as Supplementary Material, Table S8. To test for independent genetic effects, the most significant SNPs from the pooled analyses were added as covariates and a new round of GWAS was performed.

Bioinformatic analyses

The significant SNPs from the pooled analyses were imported as custom tracks into the UCSC genome browser (Human Feb. 2009 GRCh37/hg19 Assembly, Data retrieved 27th September 2013). The location of the associated SNPs were compared with the location of known human protein-coding and non-protein-coding genes taken from the NCBI RNA reference sequences collection (RefSeq genes—last updated 25 April 2012). For the missense coding-variants SIFT (50) and PolyPhen (51) scores were retrieved from Ensembl database (http://www.ensembl.org, accessed—27 September 2013).

MR and multivariable Cox regression analyses

Multivariable Cox regression analyses were performed to test the possible relation between SNPs affecting NT-proBNP levels and clinical outcomes. The outcomes analyzed were the primary composite endpoint (MI, stroke or CV death) and CV death. All analyses were adjusted for the following variables: age, sex, BMI, medical history of diabetes mellitus, history of congestive heart failure, history of hypertension, history of chronic renal disease, smoking status, final diagnosis of index event, randomized treatment, history of MI, history of peripheral arterial disease, revascularization history of prior coronary artery bypass grafting surgery, revascularization history of prior percutaneous coronary intervention, history of non-hemorrhagic stroke and the first four principal components.

Supplementary Material

Supplementary Material is available at HMG online.

Conflict of Interest statement. ÅJ.: institutional research grant from AstraZeneca. N.E.: institutional research grant from AstraZeneca. D.L.: institutional research grant and lecture fees from AstraZeneca. C.V.: advisory board member, institutional research grant and speaker fees from AstraZeneca, Boehringer Ingelheim; advisory board member and speaker fees from The Medicines Company; advisory board member for Boehringer Ingelheim. S.J.: institutional research grant, honoraria and consultant/advisory board fees from AstraZeneca; institutional research grant and consultant/advisory board fees from Medtronic; institutional research grants from Terumo Inc., Vascular Solutions; honoraria from The Medicines Company; consultant/advisory board fees from Daiichi Sankyo, Janssen, Sanofi. A.-C.S.: no conflict of interest. T.A.: no conflict of interest. A.S.: institutional research grants from AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, Bristol-Myers Squibb/Pfizer. A.H.: employee of AstraZeneca. B.J.B.: employee of and has stock and stock options in AstraZeneca. R.C.B: scientific advisory board member and grant from AstraZeneca; scientific advisory board member for Bayer, Janssen, Regado Biosciences, Boehringer Ingelheim; safety reviewing committee member for Portola. H.A.K.: grants and personal fees from Roche Diagnostics, AstraZeneca, Bayer Health; grants from St. Jude, Medtronic; personal fees from Daiichi Sankyo. P.G.S.: institution research grants, honoraria and non-financial support from Sanofi, Servier; honoraria and non-financial support from AstraZeneca; honoraria from Amarin, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Eli Lilly, Merck-Sharpe-Dohme, Novartis, Pfizer, Medtronic, Janssen, The Medicines Company, CSL-Behring, GlaxoSmithKline; stockholder in Aterovax. R.F.S.: institutional research grant, consultancy fees, speakers fees and travel support from AstraZeneca, and named by the company as an inventor on patents pending related to discoveries made during the PEGASUS-TIMI 54 study but has no personal financial interest in these; institutional research grant, consultancy fees and speaker fees from Daiichi Sankyo/Eli Lilly; consultancy fees, speaker fees and consumables from Accumetrics; institutional research grant and consultancy fees from Merck; honoraria from Medscape; consultancy fees from Aspen, Correvio, Plaque Tec, Roche, The Medicines Company, Thermo Fisher Scientific, Regeneron, Sanofi-Aventis; travel support from Medtronic. L.W.: institutional research grant, consultancy and lecture fees, travel support and honoraria from GlaxoSmithKline; institutional research grants, consultancy and lecture fees and travel support from AstraZeneca, Bristol-Myers Squibb/Pfizer, Boehringer Ingelheim; institutional research grant from Merck & Co.; institutional research grant from Roche; consultancy fees from Abbott; holds two patents involving GDF-15.

Funding

Genotyping was performed by the SNP & SEQ Technology Platform in Uppsala, which is supported by Uppsala University, Science for Life Laboratory, Sweden (SciLifeLab) and the Swedish Research Council (VR-RFI, Contracts 80576801 and 70374401). The computations were performed on resources provided by SNIC through Uppsala Multidisciplinary Center for Advanced Computational Science (UPPMAX) under projects p2012253 and p2012095. The PLATO study was funded by AstraZeneca. The genetic work was also funded by the Swedish Heart Foundation.

References

1
Jernberg
T.
,
James
S.
,
Lindahl
B.
,
Stridsberg
M.
,
Venge
P.
,
Wallentin
L.
(
2004
)
NT-proBNP in unstable coronary artery disease—experiences from the FAST, GUSTO IV and FRISC II trials
.
Eur. J. Hear. Fail.
 ,
6
,
319
325
.
2
Bibbins-Domingo
K.
,
Gupta
R.
,
Na
B.
,
Wu
A.H.
,
Schiller
N.B.
,
Whooley
M.A.
(
2007
)
N-terminal fragment of the prohormone brain-type natriuretic peptide (NT-proBNP), cardiovascular events, and mortality in patients with stable coronary heart disease
.
JAMA
 ,
297
,
169
176
.
3
Wallentin
L.
,
Lindholm
D.
,
Siegbahn
A.
,
Wernroth
L.
,
Becker
R.C.
,
Cannon
C.P.
,
Cornel
J.H.
,
Himmelmann
A.
,
Giannitsis
E.
,
Harrington
R.A.
et al
. (
2014
)
Biomarkers in relation to the effects of ticagrelor in comparison with clopidogrel in non-ST-elevation acute coronary syndrome patients managed with or without in-hospital revascularization: a substudy from the Prospective Randomized Platelet Inhibition and Patient Outcomes (PLATO) trial
.
Circulation
 ,
129
,
293
303
.
4
Olsson
L.G.
,
Swedberg
K.
,
Cleland
J.G.
,
Spark
P.A.
,
Komajda
M.
,
Metra
M.
,
Torp-Pedersen
C.
,
Remme
W.J.
,
Scherhag
A.
,
Poole-Wilson
P.
(
2007
)
Prognostic importance of plasma NT-pro BNP in chronic heart failure in patients treated with a beta-blocker: results from the Carvedilol or Metoprolol European Trial (COMET) trial
.
Eur. J. Hear. Fail.
 ,
9
,
795
801
.
5
Hijazi
Z.
,
Oldgren
J.
,
Andersson
U.
,
Connolly
S.J.
,
Ezekowitz
M.D.
,
Hohnloser
S.H.
,
Reilly
P.A.
,
Vinereanu
D.
,
Siegbahn
A.
,
Yusuf
S.
et al
. (
2012
)
Cardiac biomarkers are associated with an increased risk of stroke and death in patients with atrial fibrillation: a randomized evaluation of long-term anticoagulation therapy (RE-LY) substudy
.
Circulation
 ,
125
,
1605
1616
.
6
Semenov
A.G.
,
Seferian
K.R.
(
2011
)
Biochemistry of the human B-type natriuretic peptide precursor and molecular aspects of its processing
.
Clin. Chim. Acta
 ,
412
,
850
860
.
7
Daniels
L.B.
,
Maisel
A.S.
(
2007
)
Natriuretic peptides
.
J. Am. Coll. Cardiol.
 ,
50
,
2357
2368
.
8
Morita
E.
,
Yasue
H.
,
Yoshimura
M.
,
Ogawa
H.
,
Jougasaki
M.
,
Matsumura
T.
,
Mukoyama
M.
,
Nakao
K.
(
1993
)
Increased plasma levels of brain natriuretic peptide in patients with acute myocardial infarction
.
Circulation
 ,
88
,
82
91
.
9
Del Greco
M.F.
,
Pattaro
C.
,
Luchner
A.
,
Pichler
I.
,
Winkler
T.
,
Hicks
A.A.
,
Fuchsberger
C.
,
Franke
A.
,
Melville
S.A.
,
Peters
A.
et al
. (
2011
)
Genome-wide association analysis and fine mapping of NT-proBNP level provide novel insight into the role of the MTHFR-CLCN6-NPPA-NPPB gene cluster
.
Hum. Mol. Genet.
 ,
20
,
1660
1671
.
10
Takeishi
Y.
,
Toriyama
S.
,
Takabatake
N.
,
Shibata
Y.
,
Konta
T.
,
Emi
M.
,
Kato
T.
,
Kawata
S.
,
Kubota
I.
(
2007
)
Linkage disequilibrium analyses of natriuretic peptide precursor B locus reveal risk haplotype conferring high plasma BNP levels
.
Biochem. Biophys. Res. Commun.
 ,
362
,
480
484
.
11
Newton-Cheh
C.
,
Larson
M.G.
,
Vasan
R.S.
,
Levy
D.
,
Bloch
K.D.
,
Surti
A.
,
Guiducci
C.
,
Kathiresan
S.
,
Benjamin
E.J.
,
Struck
J.
et al
. (
2009
)
Association of common variants in NPPA and NPPB with circulating natriuretic peptides and blood pressure
.
Nat. Genet.
 ,
41
,
348
353
.
12
Yan
W.
,
Wu
F.
,
Morser
J.
,
Wu
Q.
(
2000
)
Corin, a transmembrane cardiac serine protease, acts as a pro-atrial natriuretic peptide-converting enzyme
.
Proc. Natl. Acad. Sci. USA.
 ,
97
,
8525
8529
.
13
Semenov
A.G.
,
Tamm
N.N.
,
Seferian
K.R.
,
Postnikov
A.B.
,
Karpova
N.S.
,
Serebryanaya
D.V.
,
Koshkina
E.V.
,
Krasnoselsky
M.I.
,
Katrukha
A.G.
(
2010
)
Processing of pro-B-type natriuretic peptide: furin and corin as candidate convertases
.
Clin. Chem.
 ,
56
,
1166
1176
.
14
Ehret
G.B.
,
Munroe
P.B.
,
Rice
K.M.
,
Bochud
M.
,
Johnson
A.D.
,
Chasman
D.I.
,
Smith
A.V.
,
Tobin
M.D.
,
Verwoert
G.C.
,
Hwang
S.J.
et al
. (
2011
)
Genetic variants in novel pathways influence blood pressure and cardiovascular disease risk
.
Nature
 ,
478
,
103
109
.
15
Deloukas
P.
,
Kanoni
S.
,
Willenborg
C.
,
Farrall
M.
,
Assimes
T.L.
,
Thompson
J.R.
,
Ingelsson
E.
,
Saleheen
D.
,
Erdmann
J.
,
Goldstein
B.A.
et al
. (
2013
)
Large-scale association analysis identifies new risk loci for coronary artery disease
.
Nat. Genet.
 ,
45
,
25
33
.
16
Meirhaeghe
A.
,
Sandhu
M.S.
,
McCarthy
M.I.
,
de Groote
P.
,
Cottel
D.
,
Arveiler
D.
,
Ferrieres
J.
,
Groves
C.J.
,
Hattersley
A.T.
,
Hitman
G.A.
et al
. (
2007
)
Association between the T-381C polymorphism of the brain natriuretic peptide gene and risk of type 2 diabetes in human populations
.
Hum. Mol. Genet.
 ,
16
,
1343
1350
.
17
Teslovich
T.M.
,
Musunuru
K.
,
Smith
A.V.
,
Edmondson
A.C.
,
Stylianou
I.M.
,
Koseki
M.
,
Pirruccello
J.P.
,
Ripatti
S.
,
Chasman
D.I.
,
Willer
C.J.
et al
. (
2010
)
Biological, clinical and population relevance of 95 loci for blood lipids
.
Nature
 ,
466
,
707
713
.
18
Waterworth
D.M.
,
Ricketts
S.L.
,
Song
K.
,
Chen
L.
,
Zhao
J.H.
,
Ripatti
S.
,
Aulchenko
Y.S.
,
Zhang
W.
,
Yuan
X.
,
Lim
N.
et al
. (
2010
)
Genetic variants influencing circulating lipid levels and risk of coronary artery disease
.
Arter. Thromb. Vasc. Biol.
 ,
30
,
2264
2276
.
19
Speliotes
E.K.
,
Willer
C.J.
,
Berndt
S.I.
,
Monda
K.L.
,
Thorleifsson
G.
,
Jackson
A.U.
,
Lango Allen
H.
,
Lindgren
C.M.
,
Luan
J.
,
Magi
R.
et al
. (
2010
)
Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index
.
Nat. Genet.
 ,
42
,
937
948
.
20
Wain
L.V.
,
Verwoert
G.C.
,
O'Reilly
P.F.
,
Shi
G.
,
Johnson
T.
,
Johnson
A.D.
,
Bochud
M.
,
Rice
K.M.
,
Henneman
P.
,
Smith
A.V.
et al
. (
2011
)
Genome-wide association study identifies six new loci influencing pulse pressure and mean arterial pressure
.
Nat. Genet.
 ,
43
,
1005
1011
.
21
Pfister
R.
,
Luben
R.N.
,
Khaw
K.T.
,
Wareham
N.J.
(
2013
)
Common genetic variants of the natriuretic peptide gene locus are not associated with heart failure risk in participants in the EPIC-Norfolk study
.
Eur. J. Hear. Fail.
 ,
15
,
624
627
.
22
Choquet
H.
,
Cavalcanti-Proenca
C.
,
Lecoeur
C.
,
Dina
C.
,
Cauchi
S.
,
Vaxillaire
M.
,
Hadjadj
S.
,
Horber
F.
,
Potoczna
N.
,
Charpentier
G.
et al
. (
2009
)
The T-381C SNP in BNP gene may be modestly associated with type 2 diabetes: an updated meta-analysis in 49 279 subjects
.
Hum. Mol. Genet.
 ,
18
,
2495
2501
.
23
Pfister
R.
,
Sharp
S.
,
Luben
R.
,
Welsh
P.
,
Barroso
I.
,
Salomaa
V.
,
Meirhaeghe
A.
,
Khaw
K.T.
,
Sattar
N.
,
Langenberg
C.
et al
. (
2011
)
Mendelian randomization study of B-type natriuretic peptide and type 2 diabetes: evidence of causal association from population studies
.
PLoS Med.
 ,
8
,
e1001112
.
24
Ellis
K.L.
,
Newton-Cheh
C.
,
Wang
T.J.
,
Frampton
C.M.
,
Doughty
R.N.
,
Whalley
G.A.
,
Ellis
C.J.
,
Skelton
L.
,
Davis
N.
,
Yandle
T.G.
et al
. (
2011
)
Association of genetic variation in the natriuretic peptide system with cardiovascular outcomes
.
J. Mol. Cell Cardiol.
 ,
50
,
695
701
.
25
Carrera
N.
,
Arrojo
M.
,
Sanjuan
J.
,
Ramos-Rios
R.
,
Paz
E.
,
Suarez-Rama
J.J.
,
Paramo
M.
,
Agra
S.
,
Brenlla
J.
,
Martinez
S.
et al
. (
2012
)
Association study of nonsynonymous single nucleotide polymorphisms in schizophrenia
.
Biol. Psychiatry
 ,
71
,
169
177
.
26
Price
A.L.
,
Helgason
A.
,
Palsson
S.
,
Stefansson
H.
,
St Clair
D.
,
Andreassen
O.A.
,
Reich
D.
,
Kong
A.
,
Stefansson
K.
(
2009
)
The impact of divergence time on the nature of population structure: an example from Iceland
.
PLoS Genet.
 ,
5
,
e1000505
.
27
He
L.
,
Girijashanker
K.
,
Dalton
T.P.
,
Reed
J.
,
Li
H.
,
Soleimani
M.
,
Nebert
D.W.
(
2006
)
ZIP8, member of the solute-carrier-39 (SLC39) metal-transporter family: characterization of transporter properties
.
Mol. Pharmacol.
 ,
70
,
171
180
.
28
Liu
Z.
,
Li
H.
,
Soleimani
M.
,
Girijashanker
K.
,
Reed
J.M.
,
He
L.
,
Dalton
T.P.
,
Nebert
D.W.
(
2008
)
Cd2+ versus Zn2+ uptake by the ZIP8 HCO3-dependent symporter: kinetics, electrogenicity and trafficking
.
Biochem. Biophys. Res. Commun.
 ,
365
,
814
820
.
29
Wang
C.Y.
,
Jenkitkasemwong
S.
,
Duarte
S.
,
Sparkman
B.K.
,
Shawki
A.
,
Mackenzie
B.
,
Knutson
M.D.
(
2012
)
ZIP8 is an iron and zinc transporter whose cell-surface expression is up-regulated by cellular iron loading
.
J. Biol. Chem.
 ,
287
,
34032
34043
.
30
Begum
N.A.
,
Kobayashi
M.
,
Moriwaki
Y.
,
Matsumoto
M.
,
Toyoshima
K.
,
Seya
T.
(
2002
)
Mycobacterium bovis BCG cell wall and lipopolysaccharide induce a novel gene, BIGM103, encoding a 7-TM protein: identification of a new protein family having Zn-transporter and Zn-metalloprotease signatures
.
Genomics
 ,
80
,
630
645
.
31
Liu
M.J.
,
Bao
S.
,
Galvez-Peralta
M.
,
Pyle
C.J.
,
Rudawsky
A.C.
,
Pavlovicz
R.E.
,
Killilea
D.W.
,
Li
C.
,
Nebert
D.W.
,
Wewers
M.D.
et al
. (
2013
)
ZIP8 regulates host defense through zinc-mediated inhibition of NF-kappaB
.
Cell Rep.
 ,
3
,
386
400
.
32
Bennett
E.P.
,
Hassan
H.
,
Clausen
H.
(
1996
)
cDNA cloning and expression of a novel human UDP-N-acetyl-alpha-d-galactosamine. Polypeptide N-acetylgalactosaminyltransferase, GalNAc-t3
.
J. Biol. Chem.
 ,
271
,
17006
17012
.
33
O'Halloran
A.M.
,
Patterson
C.C.
,
Horan
P.
,
Maree
A.
,
Curtin
R.
,
Stanton
A.
,
McKeown
P.P.
,
Shields
D.C.
(
2009
)
Genetic polymorphisms in platelet-related proteins and coronary artery disease: investigation of candidate genes, including N-acetylgalactosaminyltransferase 4 (GALNT4) and sulphotransferase 1A1/2 (SULT1A1/2)
.
J. Thromb. Thrombolysis
 ,
27
,
175
184
.
34
Lu
X.
,
Wang
L.
,
Chen
S.
,
He
L.
,
Yang
X.
,
Shi
Y.
,
Cheng
J.
,
Zhang
L.
,
Gu
C.C.
,
Huang
J.
et al
. (
2012
)
Genome-wide association study in Han Chinese identifies four new susceptibility loci for coronary artery disease
.
Nat. Genet.
 ,
44
,
890
894
.
35
Erbilgin
A.
,
Civelek
M.
,
Romanoski
C.E.
,
Pan
C.
,
Hagopian
R.
,
Berliner
J.A.
,
Lusis
A.J.
(
2013
)
Identification of CAD candidate genes in GWAS loci and their expression in vascular cells
.
J. Lipid Res.
 ,
54
,
1894
1905
.
36
Zeller
T.
,
Wild
P.
,
Szymczak
S.
,
Rotival
M.
,
Schillert
A.
,
Castagne
R.
,
Maouche
S.
,
Germain
M.
,
Lackner
K.
,
Rossmann
H.
et al
. (
2010
)
Genetics and beyond—the transcriptome of human monocytes and disease susceptibility
.
PLoS One
 ,
5
,
e10693
.
37
Semenov
A.G.
,
Postnikov
A.B.
,
Tamm
N.N.
,
Seferian
K.R.
,
Karpova
N.S.
,
Bloshchitsyna
M.N.
,
Koshkina
E.V.
,
Krasnoselsky
M.I.
,
Serebryanaya
D.V.
,
Katrukha
A.G.
(
2009
)
Processing of pro-brain natriuretic peptide is suppressed by O-glycosylation in the region close to the cleavage site
.
Clin. Chem.
 ,
55
,
489
498
.
38
James
S.
,
Akerblom
A.
,
Cannon
C.P.
,
Emanuelsson
H.
,
Husted
S.
,
Katus
H.
,
Skene
A.
,
Steg
P.G.
,
Storey
R.F.
,
Harrington
R.
et al
. (
2009
)
Comparison of ticagrelor, the first reversible oral P2Y(12) receptor antagonist, with clopidogrel in patients with acute coronary syndromes: rationale, design, and baseline characteristics of the PLATelet inhibition and patient Outcomes (PLATO) trial
.
Am. Heart J.
 ,
157
,
599
605
.
39
Wallentin
L.
,
Becker
R.C.
,
Budaj
A.
,
Cannon
C.P.
,
Emanuelsson
H.
,
Held
C.
,
Horrow
J.
,
Husted
S.
,
James
S.
,
Katus
H.
et al
. (
2009
)
Ticagrelor versus clopidogrel in patients with acute coronary syndromes
.
N. Engl. J. Med.
 ,
361
,
1045
1057
.
40
Gunderson
K.L.
,
Steemers
F.J.
,
Lee
G.
,
Mendoza
L.G.
,
Chee
M.S.
(
2005
)
A genome-wide scalable SNP genotyping assay using microarray technology
.
Nat. Genet.
 ,
37
,
549
554
.
41
Steemers
F.J.
,
Chang
W.
,
Lee
G.
,
Barker
D.L.
,
Shen
R.
,
Gunderson
K.L.
(
2006
)
Whole-genome genotyping with the single-base extension assay
.
Nat. Methods
 ,
3
,
31
33
.
42
Purcell
S.
,
Neale
B.
,
Todd-Brown
K.
,
Thomas
L.
,
Ferreira
M.A.
,
Bender
D.
,
Maller
J.
,
Sklar
P.
,
de Bakker
P.I.
,
Daly
M.J.
et al
. (
2007
)
PLINK: a tool set for whole-genome association and population-based linkage analyses
.
Am. J. Hum. Genet.
 ,
81
,
559
575
.
43
Aulchenko
Y.S.
,
Ripke
S.
,
Isaacs
A.
,
van Duijn
C.M.
(
2007
)
GenABEL: an R library for genome-wide association analysis
.
Bioinformatics
 ,
23
,
1294
1296
.
44
Howie
B.
,
Fuchsberger
C.
,
Stephens
M.
,
Marchini
J.
,
Abecasis
G.R.
(
2012
)
Fast and accurate genotype imputation in genome-wide association studies through pre-phasing
.
Nat. Genet.
 ,
44
,
955
959
.
45
Delaneau
O.
,
Zagury
J.F.
,
Marchini
J.
(
2013
)
Improved whole-chromosome phasing for disease and population genetic studies
.
Nat. Methods
 ,
10
,
5
6
.
46
Howie
B.N.
,
Donnelly
P.
,
Marchini
J.
(
2009
)
A flexible and accurate genotype imputation method for the next generation of genome-wide association studies
.
PLoS Genet.
 ,
5
,
e1000529
.
47
Abecasis
G.R.
,
Auton
A.
,
Brooks
L.D.
,
DePristo
M.A.
,
Durbin
R.M.
,
Handsaker
R.E.
,
Kang
H.M.
,
Marth
G.T.
,
McVean
G.A.
(
2012
)
An integrated map of genetic variation from 1,092 human genomes
.
Nature
 ,
491
,
56
65
.
48
Price
A.L.
,
Patterson
N.J.
,
Plenge
R.M.
,
Weinblatt
M.E.
,
Shadick
N.A.
,
Reich
D.
(
2006
)
Principal components analysis corrects for stratification in genome-wide association studies
.
Nat. Genet.
 ,
38
,
904
909
.
49
Aulchenko
Y.S.
,
Struchalin
M.V.
,
van Duijn
C.M.
(
2010
)
ProbABEL package for genome-wide association analysis of imputed data
.
BMC Bioinformatics
 ,
11
,
134
.
50
Ng
P.C.
,
Henikoff
S.
(
2003
)
SIFT: predicting amino acid changes that affect protein function
.
Nucleic Acids Res.
 ,
31
,
3812
3814
.
51
Adzhubei
I.A.
,
Schmidt
S.
,
Peshkin
L.
,
Ramensky
V.E.
,
Gerasimova
A.
,
Bork
P.
,
Kondrashov
A.S.
,
Sunyaev
S.R.
(
2010
)
A method and server for predicting damaging missense mutations
.
Nat. Methods
 ,
7
,
248
249
.