Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Jun 1.
Published in final edited form as: J Thromb Haemost. 2016 Sep 17;14(10):1960–1970. doi: 10.1111/jth.13431

Single nucleotide polymorphisms in an intergenic chromosome 2q region associated with tissue factor pathway inhibitor plasma levels and venous thromboembolism

J DENNIS *, V TRUONG *, D AÏSSI †,‡,§, A MEDINA-RIVERA ¶,**, S BLANKENBERG †,‡, M GERMAIN †,‡,§, M LEMIRE ‡‡, L ANTOUNIANS **,§§, M CIVELEK ¶¶, R SCHNABEL ††, P WELLS ***, M D WILSON **,§§, P-E MORANGE †††,‡‡‡,§§§, D-A TRÉGOUËT †,‡,§, F GAGNON *
PMCID: PMC6544906  NIHMSID: NIHMS1020235  PMID: 27490645

Summary

Background

Tissue factor pathway inhibitor (TFPI) regulates fibrin clot formation, and low TFPI plasma levels increase the risk of arterial thromboembolism and venous thromboembolism (VTE). TFPI plasma levels are also heritable, and a previous linkage scan implicated the chromosome 2q region, but no specific genes.

Objectives

To replicate the finding of the linkage region in an independent sample, and to identify the causal locus.

Methods

We first performed a linkage analysis of microsatellite markers and TFPI plasma levels in 251 individuals from the F5L Family Study, and replicated the finding of the linkage peak on chromosome 2q (LOD = 3.06). We next defined a follow-up region that included 112 603 single nucleotide polymorphisms (SNPs) under the linkage peak, and meta-analyzed associations between these SNPs and TFPI plasma levels across the F5L Family Study and the Marseille Thrombosis Association (MARTHA) Study, a study of 1033 unrelated VTE patients. SNPs with false discovery rate q-values of < 0.10 were tested for association with TFPI plasma levels in 892 patients with coronary artery disease in the Athero-Gene Study.

Results and Conclusions

One SNP, rs62187992, was associated with TFPI plasma levels in all three samples (β = + 0.14 and P = 4.23 × 10−6 combined; β = + 0.16 and P = 0.02 in the F5L Family Study; β = + 0.13 and P = 6.3 × 10−4 in the MARTHA Study; β = + 0.17 and P = 0.03 in the AtheroGene Study), and contributed to the linkage peak in the F5L Family Study. rs62187992 was also associated with clinical VTE (odds ratio 0.90, P = 0.03) in the INVENT Consortium of > 7000 cases and their controls, and was marginally associated with TFPI expression (β = + 0.19, P = 0.08) in human aortic endothelial cells, a primary site of TFPI synthesis. The biological mechanisms underlying these associations remain to be elucidated.

Keywords: blood coagulation, genetic association studies, genetic linkage, thrombosis, tissue factor pathway inhibitor

Introduction

Tissue factor pathway inhibitor (TFPI) regulates the blood coagulation cascade. It impedes early stages of the extrinsic coagulation pathway by binding the procoagulants tissue factor-activated FVII and activated FX [1], and also limits a growing clot by preventing the incorporation of activated FV into prothrombinase, the enzyme that converts prothrombin into thrombin [2]. Mice lacking the TFPI gene die in utero from disseminated intravascular coagulation [3], and, in humans, prospective studies have shown a threshold effect for low TFPI plasma levels and thrombosis risk. Subjects with baseline TFPI plasma levels in the lowest 5% of the distribution had a nearly two-fold increased risk of incident venous thromboembolism (VTE) in a nested case-control study of 534 VTE cases and 1091 matched controls [4]. Likewise, in a nested case-control study of 296 coronary heart disease patients and 563 matched controls, subjects with baseline TFPI levels in the lowest 10% of the distribution had a greater than two-fold increased risk of incident non-fatal myocardial infarction and coronary death [5].

TFPI plasma levels are associated with both non-genetic and genetic factors. Non-genetic covariates include age, sex, smoking, oral contraceptive use, body mass index, cholesterol levels, and several markers of coagulation and endothelial cell (EC) activation [5–10]. However, whether these correlations are causal or indirect effects of underlying cardiovascular disease is unclear [11].

Between 27% and 52% of the variability in TFPI plasma levels is attributable to genetics [12–14]. Early searches for genetic risk factors focused on TFPI itself, and, although meta-analyses confirmed that rs5940 and rs7586970 in TFPI were associated with TFPI plasma levels [15], much variability in TFPI plasma levels remains unexplained. The Genetic Analysis of Idiopathic Thrombosis (GAIT) Study is the only published genome-wide study of TFPI plasma levels to date, and, through a linkage analysis, identified a single locus on chromosome 2q upstream of TFPI [12]. However, the linkage signal was not explained by rs5940 in TFPI (rs7586970 was not tested). We sought to replicate these linkage results, and to identify the putative variant(s) underlying the linkage signal.

Materials and methods

Overall strategy

Our analytic strategy comprised three steps. First, we conducted a linkage analysis of TFPI plasma levels in 251 individuals from five extended pedigrees from the French-Canadian Family Study on Factor V Leiden Thrombophilia (F5L Family Study) to replicate the GAIT Study linkage findings. Second, we fine-mapped the chromosome 2q region defined by the F5L Family Study and GAIT Study linkage peaks. We meta-analyzed associations between single nucleotide polymorphisms (SNPs) in this region and TFPI plasma levels across the F5L Family Study and the MarseilleThrombosis Association (MARTHA) Study, which included 1033 unrelated cases of VTE, and tested significant SNPs for association with TFPI plasma levels in a third study sample, from the AtheroGene Study, which included 892 subjects with coronary artery disease. We identified one SNP associated with TFPI plasma levels across all three study samples. In the third step, we validated [16] this SNP for its role in thrombotic events in nearly 7000 cases and 53 000 controls from the International Network against Thrombosis (INVENT) collaboration [17].

Study samples

The F5L Family Study included 369 individuals from five French-Canadian families [18,19]. Probands were identified from the Thrombosis Clinic of The Ottawa Hospital, Ottawa, Canada, between 2005 and 2006, and had idiopathic, objectively diagnosed VTE, and carried the FV Leiden (F5L) mutation. Idiopathic VTE was VTE in the absence of immobilization, surgery, fracture with a plaster cast, multisystem trauma, pregnancy, or cancer in the previous 6 months. VTE was diagnosed by venography, compression ultrasound, magnetic resonance imaging, computed tomography, ventilation/perfusion lung scan, spiral computed tomography, and/or autopsy. Probands were free of strong genetic risk factors for VTE: protein S, protein C and antithrombin deficiencies, and homozygosity for the F5L mutation.

Probands’ biological relatives aged ≥ 10 years were eligible to participate in the study. All invited subjects consented (N = 255), completed an interviewer-administered questionnaire on their personal and medical information, and provided blood samples. The research ethics boards of the University of Toronto and the Ottawa Hospital Research Institute approved this study.

The MARTHA Study included 1592 patients with objectively diagnosed VTE seen at the Thrombophilia Center of La Timone Hospital, Marseille, France, between January 1994 and October 2005 [19,20]. Subjects were recruited at least 3 months after the VTE event, and only if none of the following had occurred in the 3 months preceding the VTE event: immobilization for ≥ 7 days, surgery, trauma, oral contraceptive use, or pregnancy. VTE was diagnosed by venography, Doppler ultrasound, spiral computed tomographic scanning angiography, and/or ventilation/perfusion lung scan. None of the participants had strong genetic risk factors for VTE (protein S, protein C or antithrombin deficiency, lupus anticoagulant, or homozygosity for the F5L or F2 G20210A mutations). All subjects were white Europeans, and the majority were of self-reported French descent. A physician interviewed subjects about their personal and medical histories, and participants provided a fasting blood sample. The Ethics Committee at La Timone Hospital approved this study.

The AtheroGene Study included coronary artery disease patients undergoing coronary angiography at the Department of Medicine II of the Johannes Gutenberg- University Mainz or the Bundeswehrzentralkrankenhaus Koblenz between June 1999 and February 2004 [5,21]. Subjects completed a questionnaire and provided a blood sample before the coronary angiography. Subjects were of European origin, of German nationality, and living in the Rhein-Mainz area of Germany. The Ethics Committee of the University of Mainz approved this study.

Genotyping and imputation

F5L Family Study participants were genotyped with a panel of 1079 microsatellite markers from DeCODE [18]. Markers were mapped to cM positions by use of the Marshfield Medical Research Foundation sex-averaged genetic maps [22], and distances were converted to Haldane distances. Markers were spaced, on average, 3.4 cM apart.

Study participants were also genotyped with the Illumina Human660W-Quad beadchip [18,19], and genotype data from all 255 individuals passed quality control. We imputed additional SNP genotypes by using the prephasing approach in shapeit v2.r778 [23,24], the imputation software impute v2.3.1 [25], and the 1000 Genomes Phase 1 Total European Ancestry (EUR) population (23 November 2010 release) [26]. We filtered imputed SNPs with < 20 observations of the minor allele and with low imputation quality (info < 0.3), giving a total of 6 159 088 included SNPs.

MARTHA Study subjects were genotyped with the Illumina Human610-Quad or 660W-Quad beadchips [20], and, after application of quality control filters, 1542 individuals remained in the analysis. We imputed genotypes by using mach version 1.0.18.c [27,28] and haplotypes from the 1000 Genomes Total European Ancestry (EUR) population (August 2010 release), and the 7 804 320 imputed SNPs with a minor allele frequency (MAF) of > 0.01 and acceptable imputation quality (r2 > 0.3) were used in subsequent analyses [29].

AtheroGene Study subjects were genotyped with the Affymetrix Genome-Wide Human SNP 6.0 array. SNPs with significant (P < 10−5) deviation from Hardy-Weinberg equilibrium, with a MAF of < 1% or with a genotyping call rate of < 99% were filtered out. We imputed genotypes in 1762 individuals by using mach version 1.0.18.c [27,28] and haplotypes from the 1000 Genomes Total European Ancestry (EUR) population (November 2010 release), analyzed in minimac (14 March 2012 release) [23,30]. The 9 471 142 imputed SNPs with a MAF of > 0.01 and acceptable imputation quality (r2 > 0.3) were used for replication analysis.

Biological measurements and medication use

We measured free TFPI plasma levels with the Asserachrom Free TFPI enzyme immunoassay from DiagnosticaStago (Asnières, France) in the F5L Family Study, the MARTHA Study, and the AtheroGene Study. The assay’s normal reference interval was 5.2–14.8 ng mL−1. Data were available from 251 individuals from the F5L Family Study, 1170 individuals from the MARTHA Study, and 892 individuals from the AtheroGene Study. Although TFPI was measured from frozen samples in the MARTHA Study, we ensured that values were comparable to those obtained from fresh plasma by calculating the correlation between TFPI and von Willebrand factor (VWF) plasma levels, both of which are markers of EC dysfunction, in the MARTHA Study and the F5L Family Study. TFPI and VWF measured in fresh plasma in the F5L Family study were significantly correlated (r2 = 0.13, P = 0.008), as were TFPI (frozen plasma) and VWF (fresh plasma) in the MARTHA Study (r2 = 0.23, P < 0.0001), suggesting that TFPI in frozen plasma was stable over the storage time. All analyses were of log-transformed TFPI plasma levels to improve model fit, unless noted otherwise. We used the manufacturers’ standard laboratory protocols to measure other lipid and hematologic traits relevant to TFPI.

TFPI plasma levels spike after injection of heparin [6], which is administered in the acute phase of a VTE [31], or prophylactically in patients undergoing coronary angiography [32]. Heparin had been administered to none of the subjects in the F5L Family Study, but to 137 subjects in the MARTHA Study undergoing inherited thrombophilia testing, and to 359 subjects in the AtheroGene Study. Subjects undergoing thrombophilia testing may have differed systematically from the remaining MARTHA Study sample, so we excluded MARTHA Study subjects receiving heparin. In the AtheroGene Study, on the other hand, heparin administration did not vary systematically across patient subgroups, so we included all subjects receiving heparin, and adjusted for heparin use.

Joint segregation and linkage analyses

We used solar [33] to estimate the total additive genetic heritability of TFPI plasma levels in the F5L Family Study, conditioning on age and sex, and to perform variance components multipoint and two-point linkage analyses with microsatellites, as was done in the GAIT Study [12].

We compared the solar results with those from an analysis using loki [34,35], which implements a Bayesian Markov chain Monte Carlo routine and is less sensitive to the trait’s distributional assumptions [36]. Starting values were specified for the number of loci (N = 2), which were assumed to follow a Poisson distribution, and age and sex were included as covariates. Linkage results from loki were reported as Bayes factors, with log10(Bayes factors) between 1 and 10 defining ‘strong evidence’ for linkage [35].

solar showed that the chromosome 2q region was the highest peak genome-wide, and the LOD score of the region exceeded the threshold (LOD > 3) for candidate region significance. In loki, Bayes factors supported ‘strong evidence’ for linkage. We therefore restricted all subsequent linkage analyses to chromosome 2. We intended to fine-map the linkage region by using SNP data, and so next confirmed the microsatellite linkage results by using SNP markers, which were selected from the clean SNPs on the Illumina Human660W-Quad bead-chip. Eligible markers had MAFs of > 0.2 and pairwise r2 < 0.2, and were mapped to cM positions according to Rutger’s sex-averaged map for the Illumina Human660W-Quad beadchip [37]. We selected the first SNP on chromosome 2, followed by all subsequent SNPs separated by at least 0.5 cM, giving a total of 390 markers on chromosome 2, spaced 0.7 cM apart on average. We analyzed the pruned SNP set by using multipoint and two-point linkage analyses in solar, adjusting for age and sex.

We also conducted three additional sensitivity analyses by using the chromosome 2 SNP marker set and solar: we tested an inverse-normal transformation of TFPI plasma levels to account for the slightly elevated residual kurtosis observed in the analysis of log-transformed TFPI plasma levels; we excluded three individuals whose TFPI plasma levels exceeded three times the interquartile range from the third quartile in the F5L Family Study; and we added covariates associated with TFPI plasma levels in the F5L Family Study (HDL cholesterol, lipoprotein(a), VWF, and FVII, identified with a backwards selection model). All SNP linkage analyses included age and sex as covariates.

Candidate region fine-mapping

Ultimately, our linkage analysis led us to a 70-cM region on chromosome 2q defined by the GAIT Study and F5L Family Study (SNP) linkage regions. Linkage cM positions were mapped to bp positions by using Rutger’s sex-averaged map for the Illumina Human660W-Quad beadchip [37], and our region spanned chr2:165,147,999–231,778,585 in the hg19 assembly. This region included 112 603 genotyped and imputed SNPs common to the F5L Family Study and MARTHA Study, each of which was tested for association with TFPI plasma levels in an additive genetic model. In the F5L Family Study, we tested for association with gemma [38], which uses a mixed model to account for family structure, and adjusted for age and sex. In the MARTHA Study, we used mach2qtl [27,28], and adjusted for age, sex, the first four principal components, vitamin K antagonist use, and antiplatelet use. We used an inverse-variance weighted model implemented in metal [39] to meta-analyze results across both studies. Heterogeneity was quantified by use of the I2 statistic and the heterogeneity P.

SNPs with false discovery rate (FDR) q-values of < 0.10 were tested for replication in the AtheroGene Study, as simulation studies have shown that q-value thresholds of > 0.05 may be required in genome-wide asociation studies (GWASs) to discover truly associated SNPs [40]. We modeled associations between SNPs and TFPI plasma levels in the AtheroGene Study under an additive genetic model, and adjusted for age, sex, the first four principal components, heparin use, and antiplatelet use.

As a final step, we attempted to validate [16] any TFPI-associated SNPs for their role in thrombosis by conducting a look-up in a large published meta-GWAS of VTE, the INVENT collaboration, which comprised 7507 VTE cases and 52 632 controls from 12 studies [17].

Results

Our workflow involved looking for TFPI plasma level-associated variants in three distinct study samples (Fig. 1). Individuals in the F5L Family Study were younger than those in the MARTHA Study and the AtheroGene Study, and were equally likely to be male or female, and 6% had a history of VTE (Table 1). TFPI plasma levels ranged from 3.4 ng mL−1 to 28.8 ng mL−1 in the F5L Family Study, from 1.9 ng mL−1 to 59.6 ng mL−1 in the MARTHA Study, and from 1.80 ng mL−1 to 334.0 ng mL−1 in the AtheroGene Study (owing to the higher prevalence of subjects receiving heparin).

Fig. 1.

Fig. 1.

Open in a new tab

Study design and workflow. CAD, coronary artery disease; F5L, factor V Leiden mutation; SNP, single nucleotide polymorphism; TFPI, tissue factor pathway inhibitor; VTE, venous thromboembolism.

Table 1.

Study sample characteristics

Characteristic F5L Family Study (N = 251) MARTHA Study (N = 1033) AtheroGene Study (N = 892)
Female sex, N (%) 126 (50.2) 684 (66.2) 200 (22.4)
Mean age in years (SD) 40.5 (18.0) 47.5 (15.2) 61.1 (9.8)
History of thrombosis*, N (%) 15 (6.0) 1033 (100) 271 (30.4)
Mean TFPI level in ng mL−1 (SD) 8.3 (3.8) 13.5 (6.5) 24.8 (40.4)
F5L mutation carrier, N (%) 63 (25.1) 237 (22.9) ND
Current smoker, N (%) 61 (24.3) 234 (22.7) 360 (40.3)
Female hormone therapy use, N (%) 18 (7.2) 66 (6.4) ND
Mean body mass index in kg m−2 (SD) 26.9 (5.9) 25.4 (4.8) 27.6 (3.7)
Antiplatelet use, N (%) Anticoagulant use, N (%) 17 (6.8) 69 (6.7) 753 (84.4)
 Vitamin K antagonists 8 (3.2) 235 (22.7) ND
 Heparin (low molecular weight or unfractioned), N (%) 0 0 359 (40.2)
Open in a new tab

F5L, factor V Leiden mutation; ND, not determined; SD, standard deviation; TFPI, tissue factor pathway inhibitor.

*

Thrombosis is defined as venous thromboembolism in the F5L Family Study and in the MARTHA Study, and as myocardial infarction in the AtheroGene Study.

The total additive genetic heritability of TFPI plasma levels was 40.9% (P = 1.48 × 10−5) in the F5L Family Study. When the same approach as in the GAIT Study was used, i.e. microsatellite markers and solar, the chromosome 2q peak was the highest peak genome-wide (Fig. 2), with LOD scores of 3.06 in the multipoint analysis and 2.67 in the two-point analysis (peak marker D2S126). The peak was upstream of TFPI. Markov chain Monte Carlo methods in loki confirmed the solar findings, providing ‘strong evidence’ for linkage in the chromosome 2q region (Fig. S1).

Fig. 2.

Fig. 2.

Open in a new tab

Genome-wide multipoint linkage of tissue factor pathway inhibitor plasma levels determined with 1079 microsatellite markers. Analyses were conducted in solar, with adjustment for age and sex. The chromosome 2q region was the highest peak genome-wide, and exceeded the threshold (LOD > 3.0) for candidate region significance.

In the linkage analysis of SNPs on chromosome 2 with solar, the chromosome 2q region had a peak multipoint LOD score of 1.88 (Fig. 3A) and a two-point LOD score of 2.67 at marker rs7564113. The results were similar when the inverse-normal transformation was used and after adjusting for covariates (Fig. S2A,C). When we excluded the three individuals with extreme TFPI plasma levels, however, the peak was reduced (Fig. S2B). All three individuals were from different families, and all were male, but otherwise they had no unifying characteristics.

Fig. 3.

Fig. 3.

Open in a new tab

Fine-mapping of the chromosome 2q linkage region defined by the GAIT Study and F5L Family Study linkage signals. Linkage cM positions were mapped to bp positions by use of Rutger’s sex-averaged map for the Illumina Human660W-Quad beadchip, and our region spanned chr2:165147999–231778585 in the hg19 assembly. Results (log10 [P]) before (A) and after (B) adjustment for rs62187992 were plotted for the 112 603 SNPs whose association with tissue factor pathway inhibitor (TFPI) plasma levels was meta-analyzed in the F5L Family Study and MARTHA Study samples. Single nucleotide polymorphisms (SNPs) in TFPI (± 50 kb) did not explain the linkage signal. F5L, factor V Leiden mutation; GWAS, genome-wide asociation study.

We next fine-mapped the GAIT Study and F5L Family Study chromosome 2q linkage region. Our candidate region spanned 70 cM on chromosome 2, from 170 cM to 240 cM, and from 165 147 999 bp to 231 778 585 bp, hg19 assembly (Fig. 3). We tested the association of 112 603 SNPs in this region with TFPI plasma levels in the F5L Family Study and the MARTHA Study (quantile-quantile plot provided in Fig. S3), and 38 had meta- analyzed q-values below our prespecified 10% FDR threshold. The 38 SNPs corresponded to six independent genomic regions after filtering of SNPs in perfect linkage disequilibrium (r2 = 1 in the 1000 Genomes Phase 1 reference population), and a tag SNP from each region was tested for replication in the AtheroGene Study (Table 2). One region representing three SNPs (rs62187992, rs7602135, and rs73071785) was suggestively associated with TFPI plasma levels in the AtheroGene study (Table S1), and, although the P-values of all three SNPs were not statistically significant after correction for six tests, the direction and magnitude of association of these SNPs with TFPI plasma levels were similar across study samples. We therefore took the lead SNP, rs62187992, forward for further analyses.

Table 2.

Six independent single nucleotide polymorphisms ( SNPs) (r2 < 1 in the 1000 Genomes Phase 1 reference population) were associated with tissue factor pathway inhibitor plasma levels in a meta-analysis of the F5L Family Study and MARTHA Study samples (false discovery rate [FDR] q-value of < 0.10), and were tested for replication in the AtheroGene Study

Meta-analysis
 AtheroGene
SNP Position (hg19) Alleles* Study sample MAF Effect† (SE) P I2 Heterogeneity P Effect† (SE) P FDR q-value MAF Effect† (SE) P
rs58075497 213477859 A/C F5L 0.07 0.12 (0.06) 0.042 0 0.812 0.14 (0.03) 7.61 × 10−6 0.052 0.07 − 0.07 (0.08) 0.364
MARTHA 0.07 0.14 (0.04) 6.64 × 10−5
rs62187961 213503382 C/G F5L 0.09 0.14 (0.06) 0.017 0 0.827 0.12 (0.03) 1.25 × 10−5 0.052 0.08 − 0.12 (0.08) 0.129
MARTHA 0.08 0.12 (0.03) 2.94 × 10−4
rs58997504 213532247 G/T F5L 0.07 0.11 (0.06) 0.061 0 0.718 0.13 (0.03) 3.29 × 10−5 0.099 0.07 − 0.16 (0.08) 0.058
MARTHA 0.06 0.14 (0.04) 1.94 × 10−4
rs62187992 213542695 G/A F5L 0.06 0.16 (0.07) 0.020 0 0.719 0.13 (0.03) 3.38 × 10−5 0.099 0.07 0.17 (0.08) 0.042
MARTHA 0.06 0.13 (0.04) 6.27 × 10−4
rs16851554 215024392 T/G F5L 0.09 − 0.15 (0.06) 0.008 0 0.374 − 0.11 (0.02) 3.50 × 10−6 0.052 0.13 − 0.05 (0.08) 0.581
MARTHA 0.14 − 0.10 (0.02) 8.48 × 10−5
rs17816758 215073685 C/T F5L 0.12 − 0.18 (0.06) 0.001 0 0.321 − 0.14 (0.03) 1.51 × 10−6 0.052 0.07 0.02 (0.08) 0.852
MARTHA 0.12 − 0.12 (0.03) 3.12 × 10−4
Open in a new tab

FDR, false discovery rate; MAF, minor allele frequency; SE, standard error.

*

Allele format is major/minor.

†

Effect is relative to the minor allele.

The rs62187992 ‘A’ allele had a frequency of 0.06–0.07 across all study samples, and had a consistent positive association with TFPI plasma levels (additive allele effect: β = + 0.16 in the F5L Family Study, β = + 0.13 in the MARTHA Study, and β = + 0.17 in the AtheroGene Study). The pooled additive allele effect across all three studies was β = + 0.14 (standard error 0.03), with P = 4.23 × 10−6. The results were unchanged across all sensitivity analyses in the F5L Family Study and MARTHA Study (Table 3).

Table 3.

Sensitivity analyses of the association between rs62187992 and tissue factor pathway inhibitor plasma levels in the F5L Family Study and MARTHA Study

F5L Family Study
 MARTHA Study
N Effect (SE) P N Effect (SE) P
Primary analysis 251 0.16 (0.07) 0.020 1033 0.13 (0.04) 6.27 × 10−4
Covariate adjustment* 227 0.16 (0.06) 0.015 848 0.06 (0.04) 0.088
Males 125 0.26 (0.09) 0.008 349 0.11 (0.05) 0.038
Females 126 0.08 (0.09) 0.363 684 0.11 (0.05) 0.022
Body mass index < 25 kg m−2 108 0.15 (0.09) 0.102 524 0.15 (0.05) 0.006
Body mass index ≥ 25 kg m−2 140 0.21 (0.10) 0.046 470 0.08 (0.05) 0.096
No history of venous thromboembolism 234 0.19 (0.07) 0.006 NA NA NA
History of venous thromboembolism 15 0.05 (0.26) 0.864 1033 0.13 (0.04) 6.27 × 10−4
F5L non-carriers 188 0.11 (0.08) 0.169 793 0.11 (0.04) 0.006
F5L carriers 63 0.23 (0.14) 0.103 237 0.07 (0.09) 0.436
No female hormone therapy use 225 0.17 (0.07) 0.017 966 0.11 (0.04) 0.002
No antithrombotic drug use† 200 0.15 (0.07) 0.046 729 0.12 (0.04) 0.004
Excluding extreme values‡ 248 0.12 (0.06) 0.059 1028 0.11 (0.04) 0.003
Open in a new tab

F5L, factor V Leiden mutation; NA, not applicable; SE, standard error.

*

Covariate adjustment included HDL, lipoprotein(a), von Willebrand factor and FVII in the F5L Family Study, and oral contraceptives, F5L, body mass index, von Willebrand factor and protein S in the MARTHA Study. In both study samples, potential covariates were identified from the literature, and associated covariates were identified by backwards model selection.

†

Antithrombotic medication included vitamin K antagonists and antiplatelet agents.

‡

Extreme values were defined as greater than three times the interquartile range from the third quartile in the study sample from which the individual originated.

Linkage analyses conditional on rs62187992 further implicated the SNP in TFPI plasma level variability. The chromosome 2q signal was reduced when rs62187992 was added as a covariate to the multipoint solar SNP model (maximum LOD of 1.14; Fig. 3B), suggesting that rs62187992 explained at least some of the linkage signal [41].

In an attempt to identify additional SNPs underlying the linkage signal, we also re-ran our candidate region association analyses, conditional on rs62187992. Analyses were run separately in the F5L Family Study and MARTHA Study, and associations were meta-analyzed with metal. The minimum meta-analysis FDR q-value was 0.30 (Fig. 3B; Fig. S3), suggesting no residual allelic heterogeneity across the F5L Family Study and MARTHA Study samples.

In comparison, we also tested two TFPI SNPs, rs5940 and rs7586970, identified in a recent meta-analysis [15]. The chromosome 2q linkage signal was not attenuated when rs5940 or rs7586970 was added to the model (results not shown), and neither SNP was associated with TFPI plasma levels in either the F5L Family Study or the MARTHA Study (Table S2).

Finally, we assessed the association of rs62187992 with the risk of clinical thrombosis in the INVENT meta-GWAS of VTE. The rs62187992 ‘A’ allele was less frequent in 7507 cases than in their 52 632 controls, and increasing copy numbers of the allele were associated with a 10% reduced relative risk of VTE (odds ratio 0.90, 95% confidence interval 0.84–0.98, P = 0.03). A P-value threshold of 0.05 was deemed to be statistically significant, because we tested a single SNP, obviating the need for multiple testing correction. Results were homogeneous across the 12 assessed studies (I2 = 0, P = 0.74), and the inverse association between the rs62187992 ‘A’ allele and VTE risk was consistent with the allele’s positive association with TFPI plasma levels.

rs62187992 is 25.1 Mb upstream of TFPI, in an intergenic region 139 kb upstream of ERBB4 and 322 kb downstream of IKZF2. rs62187992 is also intronic to a predicted long non-coding RNA (ENSG00000273118.1) that overlaps with the 3′-end of IKZF2. To better understand the biological implications of variability at this locus, we tested the association between rs62187992 and the expression of TFPI, IKZF2 and ERBB4 in human aortic ECs, which are primary sites of TFPI synthesis, in 147 heart transplant donors [42]. The rs62187992 ‘A’ allele was marginally associated with increased TFPI (β = + 0.19, P = 0.08) and decreased IKZF2 (β = − 0.05, P = 0.08) expression in this small dataset, but was not associated with the expression of ERBB4 (β = − 0.02, P = 0.57).

Discussion

We replicated the finding of the chromosome 2q linkage to TFPI plasma levels originally reported in the GAIT Study, and identified an SNP that contributed to the linkage peak. The rs62187992 ‘A’ allele was associated with elevated TFPI plasma levels across three European ancestry study samples of patients selected for cardiovascular disease, with a consistent magnitude and direction of effect. The combined effect was highly statistically significant (P = 4.23 × 10−6) in the context of a candidate locus study. Moreover, the rs62187992 ‘A’ allele was associated with a reduced risk of VTE in a recent meta-analysis of nearly 60 000 individuals, and gene expression analysis in a relevant cell type suggested that rs62187992 is an expression quantitative trait locus for TFPI.

Very little is known of the genetics of TFPI, despite high heritability estimates across multiple populations and study designs [12–14]. The GAIT Study is the only genome-wide study of TFPI plasma levels reported to date, and it included 397 individuals from 21 Spanish families, of which 12 families were selected for VTE or arterial thrombosis. The F5L Family Study had a similar study design, investigating families with thrombosis, and found the same linkage signal on chromosome 2 upstream of TFPI. Nonetheless, variants in TFPI (rs5940 and rs7586970) did not explain the linkage peak in the GAIT Study or in the F5L Family Study. rs5940 and rs7586970 were associated with TFPI plasma levels in a previously reported meta-analysis [15], but this finding was not replicated in the present analysis. As an explanation, the meta-analysis result for rs5940 was borderline statistically significant and driven by early, small, candidate gene studies, whose effect sizes may have been upwardly biased, whereas rs7586970 was specifically associated with total as opposed to free TFPI, which was measured in the present analysis. Several small studies have also reported associations between TFPI plasma levels and rare mutations in candidate coagulation genes, including PROS1 [43,44] and F5 [45]. Nonetheless, much remains to be learned about the biology and genetics of TFPI plasma level regulation [15].

rs62187992 (and its perfect proxies rs7602135 and rs73071785) is located in a dynamic regulatory region 25.1 Mb upstream of TFPI. According to haploreg v4 [46], rs7602135 appears to have the most regulatory potential of the three SNPs. It overlaps with an ‘enhancer’ chromatin state in three separate embryonic stem cell lines and a mesenchymal stem cell line (from which ECs and hematopoietic cells differentiate), and its minor allele is predicted to increase the similarity to regulatory motifs for two transcription factors expressed in ECs, BDP1 and COUP-TFII. Further suggesting the dynamic regulatory potential of this region, rs7602135 is 4 bp away from a polymorphic CpG site (rs1320116; MAF of 34.8%) that is differentially methylated in the DNA methylation maps of the Roadmap Epigenomics Consortium [47].

Our gene expression results suggestively associated rs6287992 with TFPI and IKZF2 expression. IKZF2 encodes the Ikaros Family Zinc Finger 2 (Helios), a hematopoietic transcription factor expressed in hematopoietic stem cells [48] and in cells of the lymphoid lineage [49,50] – cells that also express TFPI [51]. One possibility is that Helios affects the epigenetic landscape in hematopoietic stem cells, which impacts on future TFPI expression in differentiated cells. Alternatively, the Genome Tissue Expression Project (GTEx, 38 tissues, v 6.0) reports low IKZF2 expression across non-hematopoietic tissues, and Helios may regulate TFPI expression directly in vascular ECs.

Our study included healthy individuals (most F5L Family Study participants), patients with a history of VTE (MARTHA Study), and patients with coronary artery disease (AtheroGene Study), resulting in differences in the timing of TFPI measurements relative to disease, and in medication use. Cardiovascular disease increases TFPI plasma levels [6,10,11,52,53], and levels tend to stay elevated in coronary artery disease patients [54,55]. In accordance, we observed increasing TFPI plasma levels across the F5L Family Study, MARTHA Study and AtheroGene Study samples. Despite this heterogeneity, we found that the rs62187992 ‘A’ allele was associated with increased TFPI plasma levels in all study samples, even in sensitivity analyses of medication use. Nonetheless, clinical heterogeneity across study samples may have hampered our efforts to identify additional SNPs contributing to the chromosome 2q linkage signal.

Free TFPI, measured in all three study samples, is the most common measure of TFPI in epidemiologic studies [5,6,9,56] but captures only ~ 4% of the total body pool of TFPI [57]. The free TFPI assay measures circulating levels of TFPI-α, one of two common TFPI isoforms in humans [57], but most TFPI-α is either C-terminally truncated and bound to lipoproteins in the circulation, bound to the EC surface and released upon heparin infusion, or sequestered within platelets [57]. Our study could not detect SNPs that affected these pools of TFPI, as no methods exist to measure the fractions of TFPI-α in platelets and monocytes, and nor was our study designed to detect SNPs that affected TFPI-β plasma levels, or TFPI activity. Nonetheless, measures of free TFPI and of TFPI activity are often correlated [58], and the original chromosome 2q signal in the GAIT Study was linked to TFPI activity levels [12].

We replicated the finding of a linkage peak on chromosome 2q originally reported in the only genome-wide study of TFPI plasma levels to date. We confirmed that SNPs in TFPI did not explain the linkage signal, and used multiple complementary study samples to implicate rs62187992 in TFPI plasma level variability and in VTE risk. rs62187992 is intergenic, and the minor allele was marginally associated with IKZF2 and TFPI expression at a primary site of TFPI expression. Our work suggests novel etiologic pathways in thrombosis, and functional experiments are warranted to characterize the biology of our findings.

Supplementary Material

S1

Fig. S1. Microsatellites analyzed with loki provided strong evidence for linkage on chromosome 2q.

S2

Fig. S2. Sensitivity analyses of the chromosome 2 linkage results.

S3

Fig. S3. Quantile-quantile plot of meta-analysis association results for the 112 603 SNPs in the candidate chromosome 2q region before (A) and after (B) adjustment for rs62187992.

TS1

Table S1. SNPs associated with TFPI plasma levels at FDR q < 0.10 in meta-analyses of the F5L Family Study and MARTHA Study and tested for replication in AtheroGene.

TS2

Table S2. Previously reported SNPs in the TFPI gene were not associated with TFPI plasma levels in the F5L Family Study or MARTHA Study.

Essentials.

  • Tissue factor pathway inhibitor (TFPI) regulates the blood coagulation cascade.

  • We replicated previously reported linkage of TFPI plasma levels to the chromosome 2q region.

  • The putative causal locus, rs62187992, was associated with TFPI plasma levels and thrombosis.

  • rs62187992 was marginally associated with TFPI expression in human aortic endothelial cells.

Acknowledgements

This work was supported by the Canadian Institutes of Health Research (Grant MOP 86466) and by the Heart and Stroke Foundation of Canada (Grant T6484). J. Dennis is a Vanier Canada Graduate Scholar and Fellow in the Canadian Institutes of Health Research Strategic Training for Advanced Genetic Epidemiology (CIHR STAGE) program. F. Gaganon and M. D. Wilson hold Canada Research Chairs. M. D. Wilson and A. Medina-Rivera were supported by Heart and Stroke Foundation of Ontario (Bridge Grant 7486). Statistical analyses of the MARTHA and AtheroGene datasets were performed with the C2BIG computing cluster, funded by the Région Ile de France, Pierre and Marie Curie University, and the ICAN Institute for Cardiometabolism and Nutrition (ANR-10-IAHU-05). The authors are grateful to N. L. Smith for his comments during the review process, to V. Codoni for analytic support, and to the individuals who participated in the studies.

Appendix

INVENT Consortium

P. Amouyel, France; M. de Andrade, USA; S. Basu, USA; C. Berr, France; J. A. Brody, USA; D. I. Chasman, USA; J. F. Dartigues, France; A. R. Folsom, USA; M. Germain, France; H. de Haan, the Netherlands; J. Heit, USA; J. Houwing-Duitermaat, the Netherlands; C. Kabrhel, USA; P. Kraft, USA; G. Legal, France; S. Lindström, USA; R. Monajemi, Canada; P. E. Morange, France; B. M. Psaty, USA; P. H. Reitsma, the Netherlands; P. M. Ridker, USA; L. M. Rose, USA; F. R. Rosendaal, the Netherlands; N. Saut, France; E. Slagboom, the Netherlands; D. Smadja, France; N. L. Smith, USA; P. Suchon, France; W. Tang, USA; K. D. Taylor, USA; D. A. Trégouët, France; C. Tzourio, France; M. C. de Visser, the Netherlands; A. van Hylckama Vlieg, the Netherlands; L. C. Weng, USA; K. L. Wiggins, USA.

Footnotes

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.

Addendum

J. Dennis, F. Gagnon, and D. A. Trégouët conceptualized the research. F. Gagnon, P. Wells, D. A. Trégouët, P. E. Morange, S. Blankenberg, and R. Schnabel designed the F5L Family Study, MARTHA Study, and AtheroGene Study, and collected the data. J. Dennis led the data analysis, with support from V. Truong, D. Aïssi, A. Medina-Rivera, M. Germain, M. Lemire, L. Antounians, M. Civelek, and M. D. Wilson. All authors contributed to the revision process and approved the manuscript for publication. A complete list of INVENT collaborators is provided in the Appendix.

Supporting Information

Additional Supporting Information may be found in the online version of this article:

References

  • 1.Novotny WF, Girard TJ, Miletich JP, Broze GJ Jr. Purification and characterization of the lipoprotein-associated coagulation inhibitor from human plasma. J Biol Chem 1989; 264: 18832–7. [PubMed] [Google Scholar]
  • 2.Wood JP, Bunce MW, Maroney SA, Tracy PB, Camire RM, Mast AE. Tissue factor pathway inhibitor-alpha inhibits prothrombinase during the initiation of blood coagulation. Proc Natl Acad Sci USA 2013; 110: 17838–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Huang ZF, Higuchi D, Lasky N, Broze GJ Jr. Tissue factor pathway inhibitor gene disruption produces intrauterine lethality in mice. Blood 1997; 90: 944–51. [PubMed] [Google Scholar]
  • 4.Zakai NA, Lutsey PL, Folsom AR, Heckbert SR, Cushman M. Total tissue factor pathway inhibitor and venous thrombosis. The Longitudinal Investigation of Thromboembolism Etiology. Thromb Haemost 2010; 104: 207–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Morange PE, Simon C, Alessi MC, Luc G, Arveiler D, Ferrieres J, Amouyel P, Evans A, Ducimetiere P, Juhan-Vague I; PRIME Study Group. Endothelial cell markers and the risk of coronary heart disease: the Prospective Epidemiological Study of Myocardial Infarction (PRIME) study. Circulation 2004; 109: 1343–8. [DOI] [PubMed] [Google Scholar]
  • 6.Morange PE, Blankenberg S, Alessi MC, Bickel C, Rupprecht HJ, Schnabel R, Lubos E, Munzel T, Peetz D, Nicaud V, Juhan- Vague I, Tiret L; Atherogene Investigators. Prognostic value of plasma tissue factor and tissue factor pathway inhibitor for cardiovascular death in patients with coronary artery disease: the AtheroGene study. J Thromb Haemost 2007; 5: 475–82. [DOI] [PubMed] [Google Scholar]
  • 7.Morange PE, Renucci JF, Charles MA, Aillaud MF, Giraud F, Grimaux M, Juhan-Vague I. Plasma levels of free and total TFPI, relationship with cardiovascular risk factors and endothelial cell markers. Thromb Haemost 2001; 85: 999–1003. [PubMed] [Google Scholar]
  • 8.Hansen JB, Grimsgaard S, Huseby N, Sandset PM, Bonaa KH. Serum lipids and regulation of tissue factor-induced coagulation in middle-aged men. Thromb Res 2001; 102: 3–13. [DOI] [PubMed] [Google Scholar]
  • 9.Dahm A, Van Hylckama Vlieg A, Bendz B, Rosendaal F, Bertina RM, Sandset PM. Low levels of tissue factor pathway inhibitor (TFPI) increase the risk of venous thrombosis. Blood 2003; 101: 4387–92. [DOI] [PubMed] [Google Scholar]
  • 10.Sakkinen PA, Cushman M, Psaty BM, Kuller LH, Bajaj SP, Sabharwal AK, Boineau R, Macy E, Tracy RP. Correlates of antithrombin, protein C, protein S, and TFPI in a healthy elderly cohort. Thromb Haemost 1998; 80: 134–9. [PubMed] [Google Scholar]
  • 11.Winckers K, ten Cate H, Hackeng TM. The role of tissue factor pathway inhibitor in atherosclerosis and arterial thrombosis. Blood Rev 2013; 27: 119–32. [DOI] [PubMed] [Google Scholar]
  • 12.Almasy L, Soria JM, Souto JC, Warren DM, Buil A, Borrell M, Munoz X, Sala N, Lathrop M, Fontcuberta J, Blangero J. A locus on chromosome 2 influences levels of tissue factor pathway inhibitor: results from the GAIT study. Arterioscler Thromb Vasc Biol 2005; 25: 1489–92. [DOI] [PubMed] [Google Scholar]
  • 13.Warren DM, Soria JM, Souto JC, Comuzzie AG, Fontcuberta J, Blangero J, MacCluer JW, Almasy L. Heritability of hemostasis phenotypes and their correlation with type 2 diabetes status in Mexican Americans. Hum Biol 2005; 77: 1–15. [DOI] [PubMed] [Google Scholar]
  • 14.Bladbjerg EM, de Maat MP, Christensen K, Bathum L, Jespersen J, Hjelmborg J. Genetic influence on thrombotic risk markers in the elderly - a Danish twin study. J Thromb Haemost 2006; 4: 599–607. [DOI] [PubMed] [Google Scholar]
  • 15.Dennis J, Kassam I, Morange PE, Tregouet DA, Gagnon F. Genetic determinants of tissue factor pathway inhibitor plasma levels. Thromb Haemost 2015; 114: 245–57. [DOI] [PubMed] [Google Scholar]
  • 16.Konig IR. Validation in genetic association studies. Brief Bioinform 2011; 12: 253–8. [DOI] [PubMed] [Google Scholar]
  • 17.Germain M, Chasman DI, de Haan H, Tang W, Lindstrom S, Weng LC, de Andrade M, de Visser MC, Wiggins KL, Suchon P, Saut N, Smadja DM, Le Gal G, van Hylckama Vlieg A, Di Narzo A, Hao K, Nelson CP, Rocanin-Arjo A, Folkersen L, Monajemi R, et al. Meta-analysis of 65,734 individuals identifies TSPAN15 and SLC44A2 as two susceptibility loci for venous thromboembolism. Am J Hum Genet 2015; 96: 532–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Antoni G, Morange PE, Luo Y, Saut N, Burgos G, Heath S, Germain M, Biron-Andreani C, Schved JF, Pernod G, Galan P, Zelenika D, Alessi MC, Drouet L, Visvikis-Siest S, Wells PS, Lathrop M, Emmerich J, Tregouet DA, Gagnon F. A multistage multi-design strategy provides strong evidence that the BAI3 locus is associated with early-onset venous thromboembolism. J Thromb Haemost 2010; 8: 2671–9. [DOI] [PubMed] [Google Scholar]
  • 19.Antoni G, Oudot-Mellakh T, Dimitromanolakis A, Germain M, Cohen W, Wells P, Lathrop M, Gagnon F, Morange PE, Tregouet DA. Combined analysis of three genome-wide association studies on vWF and FVIII plasma levels. BMC Med Genet 2011; 12: 102–112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Oudot-Mellakh T, Cohen W, Germain M, Saut N, Kallel C, Zelenika D, Lathrop M, Tregouet DA, Morange PE. Genome wide association study for plasma levels of natural anticoagulant inhibitors and protein C anticoagulant pathway: the MARTHA project. Br J Haematol 2012; 157: 230–9. [DOI] [PubMed] [Google Scholar]
  • 21.Wild PS, Zeller T, Schillert A, Szymczak S, Sinning CR, Deiseroth A, Schnabel RB, Lubos E, Keller T, Eleftheriadis MS, Bickel C, Rupprecht HJ, Wilde S, Rossmann H, Diemert P, Cupples LA, Perret C, Erdmann J, Stark K, Kleber ME, et al. A genome-wide association study identifies LIPA as a susceptibility gene for coronary artery disease. Circ Cardiovasc Genet 2011; 4: 403–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Broman KW, Murray JC, Sheffield VC, White RL, Weber JL. Comprehensive human genetic maps: individual and sex-specific variation in recombination. Am J Hum Genet 1998; 63: 861–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Howie B, Fuchsberger C, Stephens M, Marchini J, Abecasis GR. Fast and accurate genotype imputation in genome-wide association studies through pre-phasing. Nat Genet 2012; 44: 955–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.O’Connell J, Gurdasani D, Delaneau O, Pirastu N, Ulivi S, Cocca M, Traglia M, Huang J, Huffman JE, Rudan I, McQuillan R, Fraser RM, Campbell H, Polasek O, Asiki G, Ekoru K, Hayward C, Wright AF, Vitart V, Navarro P, et al. A general approach for haplotype phasing across the full spectrum of relatedness. PLoS Genet 2014; 10: e1004234. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Howie BN, Donnelly P, Marchini J. A flexible and accurate genotype imputation method for the next generation of genomewide association studies. PLoS Genet 2009; 5: e1000529. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Delaneau O, Marchini J; Genomes Project Consortium. Integrating sequence and array data to create an improved 1000 Genomes Project haplotype reference panel. Nat Commun 2014; 5: 3934. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Li Y, Willer C, Sanna S, Abecasis G. Genotype imputation. Annu Rev Genomics Hum Genet 2009; 10: 387–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Li Y, Willer CJ, Ding J, Scheet P, Abecasis GR. MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes. Genet Epidemiol 2010; 34: 816–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Rocanin-Arjo A, Cohen W, Carcaillon L, Frere C, Saut N, Letenneur L, Alhenc-Gelas M, Dupuy AM, Bertrand M, Alessi MC, Germain M, Wild PS, Zeller T, Cambien F, Goodall AH, Amouyel P, Scarabin PY, Tregouet DA, Morange PE; CardioGenics Consortium. A meta-analysis of genome-wide association studies identifies ORM1 as a novel gene controlling thrombin generation potential. Blood 2014; 123: 777–85. [DOI] [PubMed] [Google Scholar]
  • 30.Fuchsberger C, Abecasis GR, Hinds DA. minimac2: faster genotype imputation. Bioinformatics 2015; 31: 782–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Lip GYH, Hull RD. Rationale and indications for indefinite anticoagulation in patients with venous thromboembolism. UpToDate, Last updated: 19 October 2015 edn: Wolters Kluwer, 2015, Topic 94740 Version 12.0. [Google Scholar]
  • 32.Cutlip D, Levin T. Antithrombotic therapy for elective percutaneous coronary intervention: general use. UpToDate, Last updated: 14 September, 2015 edn: Wolters Kluwer, 2015, Topic 1576 Version 31.0. [Google Scholar]
  • 33.Almasy L, Blangero J. Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet 1998; 62: 1198–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Heath SC. Markov chain Monte Carlo segregation and linkage analysis for oligogenic models. Am J Hum Genet 1997; 61: 748–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Kass RE, Raftery AE. Bayes factors. J Am Stat Assoc 1995; 90: 773–95. [Google Scholar]
  • 36.Wijsman EM, Yu D. Joint oligogenic segregation and linkage analysis using bayesian Markov chain Monte Carlo methods. Mol Biotechnol 2004; 28: 205–26. [DOI] [PubMed] [Google Scholar]
  • 37.Kong X, Murphy K, Raj T, He C, White PS, Matise TC. A combined linkage-physical map of the human genome. Am J Hum Genet 2004; 75: 1143–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Zhou X, Stephens M. Genome-wide efficient mixed-model analysis for association studies. Nat Genet 2012; 44: 821–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Willer CJ, Li Y, Abecasis GR. METAL: fast and efficient metaanalysis of genomewide association scans. Bioinformatics 2010; 26: 2190–1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Craiu RV, Sun L. Choosing the lesser evil: trade-off between false discovery rate and non-discovery rate. Statistica Sinica 2008; 18: 861–9. [Google Scholar]
  • 41.Almasy L, Blangero J. Exploring positional candidate genes: linkage conditional on measured genotype. Behav Genet 2004; 34: 173–7. [DOI] [PubMed] [Google Scholar]
  • 42.Erbilgin A, Civelek M, Romanoski CE, Pan C, Hagopian R, Berliner JA, Lusis AJ. Identification of CAD candidate genes in GWAS loci and their expression in vascular cells. J Lipid Res 2013; 54: 1894–905. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Castoldi E, Simioni P, Tormene D, Rosing J, Hackeng TM. Hereditary and acquired protein S deficiencies are associated with low TFPI levels in plasma. J Thromb Haemost 2010; 8: 294–300. [DOI] [PubMed] [Google Scholar]
  • 44.Mulder R, ten Kate MK, Kluin-Nelemans HC, Mulder AB. PROS1 Heerlen polymorphism is associated with increased free plasma tissue factor pathway inhibitor levels. Thromb Haemost 2012; 107: 594–6. [DOI] [PubMed] [Google Scholar]
  • 45.Duckers C, Simioni P, Spiezia L, Radu C, Gavasso S, Rosing J, Castoldi E. Low plasma levels of tissue factor pathway inhibitor in patients with congenital factor V deficiency. Blood 2008; 112: 3615–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Ward LD, Kellis M. HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res 2012; 40: D930–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Roadmap Epigenomics Consortium, Kundaje A, Meuleman W, Ernst J, Bilenky M, Yen A, Heravi-Moussavi A, Kheradpour P, Zhang Z, Wang J, Ziller MJ, Amin V, Whitaker JW, Schultz MD, Ward LD, Sarkar A, Quon G, Sandstrom RS, Eaton ML, Wu YC, et al. Integrative analysis of 111 reference human epi- genomes. Nature. 2015; 518: 317–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Park SM, Gonen M, Vu L, Minuesa G, Tivnan P, Barlowe TS, Taggart J, Lu Y, Deering RP, Hacohen N, Figueroa ME, Paietta E, Fernandez HF, Tallman MS, Melnick A, Levine R, Leslie C, Lengner CJ, Kharas MG. Musashi2 sustains the mixed-lineage leukemia-driven stem cell regulatory program. J Clin Invest 2015; 125: 1286–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Hahm K, Cobb BS, McCarty AS, Brown KE, Klug CA, Lee R, Akashi K, Weissman IL, Fisher AG, Smale ST. Helios, a T cell-restricted Ikaros family member that quantitatively associates with Ikaros at centromeric heterochromatin. Genes Dev 1998; 12: 782–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Thornton AM, Korty PE, Tran DQ, Wohlfert EA, Murray PE, Belkaid Y, Shevach EM. Expression of Helios, an Ikaros transcription factor family member, differentiates thymic-derived from peripherally induced Foxp3+ T regulatory cells. J Immunol 2010; 184: 3433–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Crawley J, Lupu F, Westmuckett AD, Severs NJ, Kakkar VV, Lupu C. Expression, localization, and activity of tissue factor pathway inhibitor in normal and atherosclerotic human vessels. Arterioscler Thromb Vasc Biol 2000; 20: 1362–73. [DOI] [PubMed] [Google Scholar]
  • 52.Novo G, Caplice N, Tantillo R, Bonura F, Simari R, Novo S. TFPI antigen and activity levels in patients with asymptomatic atherosclerosis and target organ acute and chronic complications. Int Angiol 2005; 24: 366–71. [PubMed] [Google Scholar]
  • 53.Soejima H, Ogawa H, Yasue H, Kaikita K, Nishiyama K, Misumi K, Takazoe K, Miyao Y, Yoshimura M, Kugiyama K, Nakamura S, Tsuji I, Kumeda K. Heightened tissue factor associated with tissue factor pathway inhibitor and prognosis in patients with unstable angina. Circulation 1999; 99: 2908–13. [DOI] [PubMed] [Google Scholar]
  • 54.Winckers K, Siegerink B, Duckers C, Maurissen LF, Tans G, Castoldi E, Spronk HM, Ten Cate H, Algra A, Hackeng TM, Rosendaal FR. Increased tissue factor pathway inhibitor activity is associated with myocardial infarction in young women: results from the RATIO study. J Thromb Haemost 2011; 9: 2243–50. [DOI] [PubMed] [Google Scholar]
  • 55.Brodin E, Borvik T, Sandset PM, Bonaa KH, Nordoy A, Hansen JB. Coagulation activation in young survivors of myocardial infarction (MI) - a population-based case-control study. Thromb Haemost 2004; 92: 178–84. [DOI] [PubMed] [Google Scholar]
  • 56.Winckers K, Ten Cate-Hoek AJ, Beekers KC, Erkens P, Hamulyak K, Ten Cate H, Hackeng TM. Impaired tissue factor pathway inhibitor function is associated with recurrent venous thromboembolism in patients with first unprovoked deep venous thrombosis. J Thromb Haemost 2012; 10: 2208–11. [DOI] [PubMed] [Google Scholar]
  • 57.Wood JP, Ellery PE, Maroney SA, Mast AE. Biology of tissue factor pathway inhibitor. Blood 2014; 123: 2934–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Dahm AE, Andersen TO, Rosendaal F, Sandset PM. A novel anticoagulant activity assay of tissue factor pathway inhibitor I (TFPI). J Thromb Haemost 2005; 3: 651–8. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1

Fig. S1. Microsatellites analyzed with loki provided strong evidence for linkage on chromosome 2q.

NIHMS1020235-supplement-S1.tiff (1.5MB, tiff)
S2

Fig. S2. Sensitivity analyses of the chromosome 2 linkage results.

NIHMS1020235-supplement-S2.tiff (1.5MB, tiff)
S3

Fig. S3. Quantile-quantile plot of meta-analysis association results for the 112 603 SNPs in the candidate chromosome 2q region before (A) and after (B) adjustment for rs62187992.

NIHMS1020235-supplement-S3.tiff (1.5MB, tiff)
TS1

Table S1. SNPs associated with TFPI plasma levels at FDR q < 0.10 in meta-analyses of the F5L Family Study and MARTHA Study and tested for replication in AtheroGene.

NIHMS1020235-supplement-TS1.xlsx (52.4KB, xlsx)
TS2

Table S2. Previously reported SNPs in the TFPI gene were not associated with TFPI plasma levels in the F5L Family Study or MARTHA Study.

NIHMS1020235-supplement-TS2.docx (15KB, docx)

RESOURCES