Common Genetic Variation in Five Thrombosis Genes and Relations to Plasma Hemostatic Protein Level and Cardiovascular Disease Risk
Objective— We undertook a linkage disequilibrium (LD)–based genetic approach to investigate the hypothesis that common sequence variants in 5 thrombosis genes influence plasma hemostatic protein levels or risk of cardiovascular disease (CVD).
Methods and Results— In a reference panel, we characterized LD structure at the fibrinogen gene cluster (fibrinogen-β[FGB], FGA, and FGG), factor VII (F7), and tissue plasminogen activator (PLAT) loci. Forty-one tag single nucleotide polymorphisms (SNPs) were genotyped in 1811 unrelated Framingham Heart Study participants. There were significant associations of 9 FGB SNPs with fibrinogen level (minimum P=0.002) and of 7 F7 SNPs and factor VII level (minimum P<0.0001). SNPs at the PLAT locus were not associated with PLAT level. In stepwise analysis, a single FGB variant explained 1% of the residual variance in fibrinogen level, and 2 F7 SNPs together explained 10% of the residual variance in factor VII level. Two PLAT haplotypes were associated with CVD (multivariable-adjusted global P=0.0004).
Conclusions— A comprehensive survey of common sequence variation demonstrates that cis-regulatory SNPs explain a modest proportion of the residual variance in circulating fibrinogen and factor VII level and PLAT haplotypes increase the risk of CVD. Additional studies are warranted to confirm the association of PLAT sequence variation and risk of CVD.
Plasma levels of blood clotting and fibrinolytic proteins may represent an important intermediate phenotype for cardiovascular disease (CVD) because increased circulating levels of hemostatic proteins have been associated with CVD risk.1–3 Both plasma levels of hemostatic proteins4 and occurrence of CVD5 have a substantial heritable component. Hence, it has been suggested that sequence variation in thrombosis genes may explain interindividual variability in plasma hemostatic protein levels and CVD risk.6
Several sequence variants at the fibrinogen-β (FGB) and factor VII (encoded by the F7 gene) loci have been associated consistently with circulating plasma levels of their respective gene products.7–9 Previous studies have been limited by an incomplete characterization of sequence variation at these loci. The proportion of plasma protein level variation explained by cis-acting sequence variation is unclear.
Recent advances in genomics allow the comprehensive evaluation of common sequence variation at a candidate gene locus for association with phenotype. Neighboring single nucleotide polymorphism (SNP) alleles are often highly correlated (ie, linkage disequilibrium [LD]).10 Genome-wide descriptions of LD structure have been reported recently, including the International Haplotype Map (HapMap) project.11 Within regions of LD, a subset of SNPs that are nonredundant (termed “tag SNPs”) may capture the majority of common variation.12
Thus, to test the hypothesis that common sequence variants in thrombosis genes influence plasma hemostatic protein levels and risk of CVD, we sought to: (1) define the LD pattern for common SNP variants at 5 thrombosis gene loci (FGB, FGA, FGG, F7, and tissue plasminogen activator [PLAT]) and identify a set of tag SNPs that capture underlying common genetic variation; and (2) examine the association of tag SNPs and haplotypes with phenotypes in a large community-based sample.
The Framingham Heart Study (FHS) offspring cohort was recruited in 1971 with the inclusion of 5124 participants who were offspring and spouses of the original cohort established in 1948.13 Participants in the present study are from a subset of individuals (n=2933) who provided blood samples for DNA extraction during the sixth clinical examination (1995 to 1998). Participants were unrelated (ie, 1 person randomly selected from each family or biologically unrelated to any other participants), and selection was designed to include equal numbers of men and women. DNA had been collected without regard to any phenotype feature.
Using these criteria, a panel of DNA samples from 1811 unrelated participants was constructed, and these participants represent the present study sample. The institutional review board at Boston Medical Center approved the study protocols, and all participants gave written informed consent. Clinical characteristics of the study participants are presented in Table 1.
Hemostatic Factor Levels
Fasting collection of samples for hemostatic factor levels and methods of their determination during the fifth cycle examination (1991 to 1995) have been described previously.14 Plasma fibrinogen was determined by the Clauss method.15 Factor VII antigen concentration was measured by ELISA (Diagnostica Stago). Values are expressed as percentage of the standard, which is 100% by definition. tPA antigen was measured by a sandwich ELISA (TintElize TPA; Biopool, AB). The intra-assay coefficients of variation were 2.6% for fibrinogen, 3.0% for factor VII antigen, and 5.5% for tPA antigen.
Assessment of Outcomes
Using prespecified criteria,16 a panel of 3 physicians blinded to genotype status confirmed the occurrence of CVD end points after review of all available hospital records, clinic notes, and pathology reports. For suspected stroke or transient ischemic attacks, a separate review panel including an FHS neurologist adjudicated events.
CVD end points occurring through the end of examination cycle 5 (1991 to 1995) were considered. The composite CVD end point (n=155) was defined by the presence of coronary heart disease (recognized or unrecognized myocardial infarction [MI], coronary insufficiency, and angina pectoris), stroke or transient ischemic attack, intermittent claudication, or congestive heart failure. More restrictive end points included coronary heart disease (n=114) and MI (n=52).
SNP Selection and Genotyping in Reference Panel
From the public National Center for Biotechnology Information dbSNP database and the SeattleSNPs variation discovery resource, we selected markers to encompass the coding region, 10 kb upstream of the transcription initiation site and 5 kb downstream of each of the 5 genes. A SNP map was constructed at each locus meeting the following criteria: (1) 1 common SNP (minor allele frequency ≥5%) approximately every 5 kb, and (2) minimum of 6 common SNPs per LD block (see below for LD block definition). Our effort to characterize LD structure was independent of the HapMap project conducted at the Broad Institute.11
SNPs were genotyped using the Sequenom MassARRAY platform (Sequenom)17 in a reference panel of 12 multigenerational white family pedigrees from Utah/Centre d’Etude du Polymorphisme Humain (CEPH) panel (Coriell Institute for Medical Research, Camden, NJ).17 The reference panel included 93 individuals of European ancestry. Supplemental Table I (available online at http://atvb.ahajournals.org) summarizes the genotyping performed in CEPH reference panel.
LD Structure, Identification of Haplotype Blocks, and Tag SNP Selection in Reference Panel
Haplotype blocks were selected using the “spine of LD” setting in the publicly available Haploview software package v2.03.18 In the reference panel, haplotypes within LD blocks were estimated using the expectation maximization algorithm. The correlation, Rh2, between these haplotypes and the haplotypes predicted using only a subset of SNPs was calculated as described previously (tagSNPs software).19 “Tag SNPs” were defined as SNPs (within a block) that yielded an Rh2 >0.80 for all haplotypes ≥5%.
Genotyping in the FHS
A total of 41 SNPs (25 tag SNPs and 16 additional SNPs) in the 5 genes were genotyped in the FHS sample (supplemental Table I). The 16 additional SNPs were genotyped in FHS to allow for redundancy in the event of genotype failures and to help assess for LD block structure similarity between CEPH and FHS samples. The χ2 test was used to compare observed genotype frequencies with their estimates under Hardy–Weinberg equilibrium. All SNPs genotyped in FHS were in Hardy–Weinberg equilibrium (P≥0.01).
Multivariable regression analyses were conducted using SAS.20 For each hemostatic factor, levels were logarithmically transformed because of skewed distributions. Sex-specific standardized residuals derived from multivariable-adjusted hemostatic factor levels served as the phenotype. Linear regression analysis was performed to test the null hypothesis that the mean level of the standardized residual did not differ by SNP genotype (3 genotype classes per SNP). To identify SNPs with the strongest contribution to overall variance, we further conducted forward stepwise selection that included all SNPs genotyped at each locus. In the individual SNP analysis, we set a threshold for nominal significance at P<0.01. For CVD end point analyses, logistic regression analysis was performed with each of the SNPs to test the null hypothesis that CVD status did not differ by SNP genotype (3 genotype classes per SNP). Haplotype-based association analyses for the hemostatic factor level and CVD phenotypes were conducted using a weighted-regression approach.21 For covariates and a fuller description of haplotype analyses, please see the supplemental Methods, available online at http://atvb.ahajournals.org.
LD Structure and Haplotypes at Candidate Gene Loci in the Reference Panel
In the Figure, we present the genomic organization and LD structure defined in the reference panel for the fibrinogen gene cluster (FGB, FGA, and FGG genes, respectively) on chromosome 4. Forty-four SNPs span a genomic region of 61.8 kb (average spacing ≈1 SNP every 1.4 kb) with 3 blocks of strong LD (extending 12, 13, and 17 kb, respectively). In each block, a small number (4 to 6) of common haplotypes represented >95% chromosomes. Sixteen tag SNPs predicted all haplotypes ≥5% frequency at the fibrinogen gene cluster.
There is a single block of LD at the F7 locus and a single block at the PLAT locus (supplemental Figures I and II, respectively, available online at http://atvb.ahajournals.org). In sum for all 5 gene regions, a total of 25 tag SNPs were identified and genotyped in the FHS cohort along with an additional 16 SNPs (supplemental Table I).
The LD structure defined in the present study agreed well with that seen in the HapMap database as of March 1, 2006 (data not shown).
Haplotype Correlation Between Framingham Participants and Reference Panel
We inferred common haplotypes (frequency ≥5%) within each block in the FHS sample (≈3600 chromosomes), and we assessed the correlation with haplotypes initially defined in the reference panel (96 chromosomes). The haplotype frequencies in the FHS sample were highly correlated with frequencies in the reference sample (Pearson r=0.92; P<0.0001; supplemental Figure III). Of the 22 haplotypes ≥5% frequency in FHS, all were observed in the CEPH reference panel.
Fibrinogen Gene Cluster SNPs, Haplotypes, and Plasma Fibrinogen Level
In association analyses of each of 14 FGB SNPs with plasma fibrinogen levels in FHS participants (Table 2), 9 of 14 FGB SNPs were associated with fibrinogen level (P<0.05). Minor alleles of all 9 SNPs were associated with higher fibrinogen levels. Included among these SNPs was FGB_001437, a polymorphism reported previously to be associated with circulating fibrinogen (also known in previous literature as β-455G/A). FGB_001437 is in strong LD (D′=1) with HindIII β-148C/T (FGB_001744 in present study), another polymorphism related previously to fibrinogen level.
To resolve whether the 14 SNP associations reflected distinct SNP effects or multiple associations attributable to LD, we conducted stepwise analysis. In stepwise analysis, FGB_001437 was the only significant predictor of fibrinogen level (stepwise P=0.026). FGB_001437 explained 1% of the residual variance in multivariable-adjusted fibrinogen level. Higher fibrinogen levels were observed in individuals harboring either 2 copies or 1 copy of the minor A allele of FGB_001437 (314±54 mg/dL, 316±56 mg/dL, and 308±59 mg/dL for the AA, AG, and GG genotypes, respectively).
In a global test for associations of mean plasma fibrinogen level with FGB haplotypes within the single 12-kb block of LD (Figure, block 1), mean plasma fibrinogen level differed across haplotypes (global P=0.03; supplemental Table II). Differences were also noted for individual haplotypes compared with all other haplotypes combined. Hap C and Hap G were each associated with higher fibrinogen level (individual haplotype P=0.02 and P=0.04, respectively; supplemental Table II), whereas Hap A was associated with lower fibrinogen level (individual haplotype P=0.04). Hap C contains the minor alleles of all 9 SNPs associated with fibrinogen level (Figure). Thus, the haplotype findings are consistent with those for individual SNPs.
Although there were haplotype-specific associations with fibrinogen level in block 1, there were no significant associations by the global test in fibrinogen gene cluster blocks 2 and 3.
F7 SNPs, Haplotypes, and Plasma Factor VII Level
In association analyses of plasma levels of factor VII with each of 8 F7 SNPs, 7 SNPs were associated with factor VII level (Table 3). Minor alleles of 5 SNPs were associated with lower factor VII levels, whereas the minor alleles of 2 SNPs were associated with higher factor VII level.
In stepwise selection including all 8 genetic variants, a promoter sequence variant, SNP rs1755685, explained 9% of the residual variance in multivariable-adjusted factor VII level (stepwise P<0.0001). SNP rs964617 explained an additional 1% of the residual variance in factor VII level (stepwise P=0.003). Individuals harboring either 1 or 2 copies of the minor A allele of rs1755685 had lower factor VII level (80%±14%, 95%±16%, and 104%±16% for the AA, AC, and CC genotypes, respectively). Individuals harboring either 1 or 2 copies of the minor T allele of rs964617 had higher factor VII level (105%±17%, 104%±15%, and 99%±17% for the TT, CT, and CC genotypes, respectively). Notably, the previously studied Arg/Gln variant (SNP rs6046) was not retained in the stepwise model, although it was in LD with SNP rs1755685 (D′=0.96; r2=0.79), and it explained a similar proportion of the residual variance in individual SNP analysis.
Mean plasma factor VII levels differed across F7 gene haplotypes (global P<10−6; supplemental Table III). Compared with all haplotypes combined, Hap D was associated with lower levels of factor VII (Hap P<10−6), whereas Hap B and Hap E were each associated with higher levels of factor VII (Hap P<10−6 and Hap P=0.008, respectively). Hap D harbors the minor allele of rs1755685 and haplotypes B and E, the minor allele of rs964617 (supplemental Figure I).
There were no significant associations of the SNPs or haplotypes at the PLAT locus with plasma tPA level (all P>0.05).
Associations of Thrombosis Gene SNPs and Haplotypes with CVD Risk
In analysis of individual variants, a single SNP at the PLAT locus, rs2020918, was associated with increased risk of CVD (Table 4). In age- and sex-adjusted analyses, the CT and TT genotypes were associated with odds ratios (ORs) of 1.5 and 2.7 for increased risks of CVD (95% CI, 1.0 to 2.3 for CT; 95% CI, 1.6 to 4.5 for TT; general model P=0.0006) compared with the referent CC genotype.
As shown in Table 4, presence of the TT compared with CC genotype at rs2020918 conferred a similar to greater magnitude of risk for the more restrictive phenotypes of coronary heart disease (OR, 2.6; 95% CI, 1.5 to 4.8; general model P=0.004) and MI (OR, 4.0; 95% CI, 1.6 to 10.0; general model P=0.006). In the multivariable-adjusted model, the magnitude of the association was modestly attenuated but remained statistically significant (Table 4).
PLAT gene haplotypes were significantly associated with risk of CVD (global P=0.0004; supplemental Table IV). Hap B (at 28% frequency) was associated with increased risk of CVD (Hap P=0.02) and Hap D (at 7% frequency) was similarly associated with an increased risk of CVD (Hap P=0.0006). Conversely, Hap C and Hap F were associated with decreased risk of CVD (Hap P=0.02 for each). The T allele of SNP rs2020918 distinguishes the 2 haplotypes associated with increased risk of CVD (haplotypes B and D) from all others.
There were no associations with CVD for SNPs at FGA, FGB, FGG, and F7 or of other SNPs at the PLAT locus (all P>0.01). Similarly, haplotypes at the fibrinogen gene cluster and F7 were not associated with CVD (global P for each block >0.05).
Using an LD-based approach, we comprehensively surveyed common genetic variation at 5 thrombosis gene loci and assessed for associations with both circulating levels of hemostatic factors and risk of CVD outcomes in 1811 men and women. Several SNPs located in cis-regulatory sequence are associated with plasma protein levels or risk of CVD. After accounting for clinical covariates and correlation among SNPs, a single polymorphic variant at the FGB gene explains 1% of the residual variance in circulating fibrinogen level and 2 F7 gene variants explain 10% of the residual variance in factor VII level. There was no evidence for association of SNPs in FGB or F7 with CVD, but a common SNP in the PLAT gene region was associated with 2.5-fold increased risk of CVD.
Our results provide the following: (1) extend current knowledge about the role of cis-regulatory genetic variation at the fibrinogen gene cluster, F7, and PLAT loci in relation to plasma hemostatic protein levels and CVD outcomes; (2) provide strong support for the ability to use the association approach to identify variants that affect complex phenotypes; and (3) show that this approach may be particularly useful in the search for functional alleles that affect plasma protein levels.
Since an initial report in 1987,9 several individual FGB polymorphisms have been related to plasma fibrinogen level.8 We extend previous work by comprehensively and jointly assessing the role of common variation at the FGB locus. We identified 9 individual FGB variants that are associated with fibrinogen level. These individual SNP results confirm and extend multiple previous reports that the minor A allele of FGB_001437 (β-455G/A22 in previous literature) and other SNPs in LD8 with this variant (including HindIII β-148C/T) are associated with higher fibrinogen level.
When all individually associated variants are considered jointly, FGB_001437 was most strongly associated with fibrinogen level, and this polymorphism explained a modest proportion (≈1%) of the variance in multivariable-adjusted fibrinogen level. This suggests that a single polymorphic site at the FGB locus influences fibrinogen level. Several in vitro studies have shown that the minor allele of FGB_001437 binds less well to a putative repressor protein complex and thus is associated with increased FGB chain transcription.23 Additional functional studies investigating each of the 9 associated variants may help clarify whether FGB_001437 is the true causal site.24
In previous reports, 5 individual F7 SNPs were reported to be related with plasma factor VII levels: 4 in the 5′ sequence and 1 nonsynonymous coding variant.25–29 The minor alleles of the coding variant R353Q (also termed Arg/Gln in the literature) and 3 others have been related to lower factor VII level, whereas the minor allele of the fifth variant has been associated with increased factor VII level. We used stepwise selection to identify the number of independent SNP effects at the F7 locus and found that there were 2 distinct SNP effects at the F7 locus: 1 variant strongly relating to lower factor VII levels and 1 variant relating to higher factor VII levels.
Tissue Plasminogen Activator
We found no association of either PLAT SNPs or PLAT haplotypes with circulating levels of tPA antigen. A recent study reported a similar absence of association with steady-state plasma tPA level.30 In contrast, when examining a dynamic tPA phenotype, vascular release of tPA after stress, Ladenvall et al reported that carriage of the minor T allele at -7351 C/T (or rs2020918) was associated with a 50% decreased vascular release of tPA after stress.31 These findings suggest that decreased local release of tPA, but not altered systemic levels, may increase risk of thrombotic CVD.31
Consistent with this hypothesis, we found a 2.5-fold increased risk of CVD and a 3.1-fold increased risk of MI associated with the T allele of rs2020918. rs2020918 is located 7.2 kb upstream of the transcription initiation site, within an enhancer region. Several in vitro studies have demonstrated this region to contain a binding site for the Sp1 transcription factor.32–34 The C to T substitution at rs2020918 has been shown reduce the affinity of Sp1 to this site, suggesting a mechanism by which rs2020918 is associated with lower tPA transcription.31,34
Our PLAT association results for the combined CVD end point replicate 2 previous studies in small samples of an association of rs2020918 with MI or stroke. In a population-based nested case-control study (61 MI cases and 120 controls), individuals bearing the minor T allele had a significant 2.7-fold increased odds of MI compared with the CC genotype, a risk estimate similar to our finding.35 Separately, in a sample of 182 cases of ischemic stroke and 301 controls, the TT genotype has been associated with increased risk of ischemic stroke (OR, 1.9; 95% CI, 1.01 to 3.6).36
Our study has several limitations. First, our association study sample is entirely white. Our findings may not generalize to other ethnicities/races, in particular those groups such as blacks with much different LD patterns. However, a recent study suggests that the effects of gene variants may be largely consistent across different “racial” or “ethnic” groups.37 Second, population stratification may lead to spurious associations attributable to underlying population substructure with differences in allele frequencies.38 However, this possibility is lessened in our sample given the ethnic homogeneity.
Third, our study examined prevalent CVD at the fifth clinic examination. The apparent association between the PLAT variant and CVD may represent a survival bias if individuals with the risk T allele had a lower case fatality rate than other genotypes. Fourth, because of a limited number of individual event types, our clinical end point investigation used a composite outcome. However, our composite outcome included both MI and stroke, and these phenotypes have been associated previously with PLAT SNP rs2020918.
Fifth, the small number of CVD events may have reduced our statistical power to detect SNP–CVD associations. Sixth, we may have failed to detect SNP-biomarker associations because of insufficient statistical power. We estimate that with a sample size of 1500 unrelated individuals and a significance level 0.001, and under a conservative assumption that nongenetic factors explain 10% of trait variance, we had ≥90% power to detect a quantitative trait locus that accounted for 1.4% of trait variance.
Finally, our association results could represent false-positive findings because of multiple hypothesis testing.39 For the associations between FGB and F7 variants and plasma protein levels, we do not believe this to be the case because our findings replicate multiple previous publications and thus are true positive results. For the CVD association with PLAT SNP rs2020918, there is modest previous supporting literature. Using the criteria set forth by Wacholder,39 we estimate that the false-positive probability rate is 10%. This assumes a previous probability of 1% for the PLAT gene given the established role of tPA in CVD treatment and the 2 previous reported associations.
Implications and Conclusions
Our implementation of an LD-based approach at 5 medically relevant genes comes on the heels of the recent completion of the HapMap project and offers several possible insights. First, studying LD in a reference panel is efficient and generates haplotype frequency estimates similar to those seen in the larger population, at least within a population of European descent.
Second, the strength of associations between genetic variants and circulating blood protein levels supports the general use of circulating biomarker “intermediate phenotypes” to test hypotheses regarding cis-regulatory effects. Sequence variants causally related to circulating protein levels may in turn serve as strong candidate alleles to test for association with disease.
In conclusion, several cis-regulatory sequence variants at the FGB and F7 loci are associated with circulating levels of their respective protein products, and a PLAT cis-regulatory variant increases risk of CVD. Further studies to replicate the PLAT SNP association in diverse populations are warranted.
This work was supported in part through the National Heart, Lung and Blood Institute Framingham Heart Study (contract No. N01-HC-25195) and the Cardiogenomics Programs for Genomic Applications (HL66582). S.K. was supported by the ACC Foundation/Merck Adult Cardiology Research Fellowship Award and the GlaxoSmithKline Research and Education Foundation for Cardiovascular Disease young investigator award.
- Received November 28, 2005.
- Accepted March 17, 2006.
Pradhan AD, LaCroix AZ, Langer RD, Trevisan M, Lewis CE, Hsia JA, Oberman A, Kotchen JM, Ridker PM. Tissue plasminogen activator antigen and d-dimer as markers for atherothrombotic risk among healthy postmenopausal women. Circulation. 2004; 110: 292–300.
Smith A, Patterson C, Yarnell J, Rumley A, Ben-Shlomo Y, Lowe G. Which hemostatic markers add to the predictive value of conventional risk factors for coronary heart disease and ischemic stroke? The Caerphilly Study. Circulation. 2005; 112: 3080–3087.
Feng D, Tofler GH, Larson MG, O’Donnell CJ, Lipinska I, Schmitz C, Sutherland PA, Johnstone MT, Muller JE, D’Agostino RB, Levy D, Lindpaintner K. Factor VII gene polymorphism, factor VII levels, and prevalent cardiovascular disease: the Framingham Heart Study. Arterioscler Thromb Vasc Biol. 2000; 20: 593–600.
Johnson GC, Esposito L, Barratt BJ, Smith AN, Heward J, Di GG, Ueda H, Cordell HJ, Eaves IA, Dudbridge F, Twells RC, Payne F, Hughes W, Nutland S, Stevens H, Carr P, Tuomilehto-Wolf E, Tuomilehto J, Gough SC, Clayton DG, Todd JA. Haplotype tagging for the identification of common disease genes. Nat Genet. 2001; 29: 233–237.
Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families. The Framingham offspring study. Am J Epidemiol. 1979; 110: 281–290.
Kannel WB, Wolf P, Garrison RJ. The Framingham Study: an epidemiological investigation of cardiovascular disease. Section 34. [NIH publication no. 87-2703]. 1987. Bethesda, Md, National Heart, Lung, and Blood Institute. Ref Type: Serial (Book, Monograph).
Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, Higgins J, DeFelice M, Lochner A, Faggart M, Liu-Cordero SN, Rotimi C, Adeyemo A, Cooper R, Ward R, Lander ES, Daly MJ, Altshuler D. The structure of haplotype blocks in the human genome. Science. 2002; 296: 2225–2229.
Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005; 21: 263– 265.
SAS/STAT User’s Guide, Version 8.1 [computer program]. Cary, NC: SAS Institute, Inc; 2000.
Iacoviello L, Vischetti M, Zito F, Benedetta DM. Genes encoding fibrinogen and cardiovascular risk. Hypertension. 2001; 38: 1199– 1203.
Brown ET, Fuller GM. Detection of a complex that associates with the Bbeta fibrinogen G-455-A polymorphism. Blood. 1998; 92: 3286– 3293.
Verschuur M, de JM, Felida L, de Maat MP, Vos HL. A hepatocyte nuclear factor-3 site in the fibrinogen beta promoter is important for interleukin 6-induced expression, and its activity is influenced by the adjacent -148C/T polymorphism. J Biol Chem. 2005; 280: 16763– 16771.
Pollak ES, Hung HL, Godin W, Overton GC, High KA. Functional characterization of the human factor VII 5′-flanking region. J Biol Chem. 1996; 271: 1738–1747.
van’t Hooft FM, Silveira A, Tornvall P, Iliadou A, Ehrenborg E, Eriksson P, Hamsten A. Two common functional polymorphisms in the promoter region of the coagulation factor VII gene determining plasma factor VII activity and mass concentration. Blood. 1999; 93: 3432–3441.
Bulens F, Merchiers P, Ibanez-Tallon I, De VA, Nelles L, Claessens F, Belayew A, Collen D. Identification of a multihormone responsive enhancer far upstream from the human tissue-type plasminogen activator gene. J Biol Chem. 1997; 272: 663–671.
Wolf AT, Medcalf RL, Jern C. The t-PA -7351C>T enhancer polymorphism decreases Sp1 and Sp3 protein binding affinity and transcriptional responsiveness to retinoic acid. Blood. 2005; 105: 1060–1067.
Jannes J, Hamilton-Bruce MA, Pilotto L, Smith BJ, Mullighan CG, Bardy PG, Koblar SA. Tissue plasminogen activator -7351C/T enhancer polymorphism is a risk factor for lacunar stroke. Stroke. 2004; 35: 1090–1094.
Freedman ML, Reich D, Penney KL, McDonald GJ, Mignault AA, Patterson N, Gabriel SB, Topol EJ, Smoller JW, Pato CN, Pato MT, Petryshen TL, Kolonel LN, Lander ES, Sklar P, Henderson B, Hirschhorn JN, Altshuler D. Assessing the impact of population stratification on genetic association studies. Nat Genet. 2004; 36: 388–393.
Wacholder S, Chanock S, Garcia-Closas M, El GL, Rothman N. Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst. 2004; 96: 434–442.