Thrombin Activatable Fibrinolysis Inhibitor Activation Peptide Shows Association With All Major Subtypes of Ischemic Stroke and With TAFI Gene Variation
Objective— Thrombin activatable fibrinolysis inhibitor (TAFI) attenuates fibrinolysis. The aim of the present study was to investigate the possible association between TAFI and overall ischemic stroke and ischemic stroke subtypes.
Methods and Results— The Sahlgrenska Academy Study on Ischemic Stroke (SAHLSIS) comprises 600 cases (18 to 69 years) and 600 matched population controls. Stroke subtype was defined by the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification. TAFI was investigated at the protein level, by analyzing plasma levels of intact TAFI and released activation peptide [AP], and at the genetic level, by genotyping a selection of eleven single nucleotide polymorphisms. After adjustment for traditional risk factors, both TAFI measurements showed association with overall ischemic stroke (AP: odds ratio, 2.22; 95% confidence interval, 1.89 to 2.61; intact TAFI: odds ratio, 1.21; 95% confidence interval, 1.06 to 1.38; for 1-SD increase in AP and intact TAFI, respectively). AP showed associations with all 4 major subtypes of ischemic stroke and intact TAFI to large vessel disease and cryptogenic stroke. TAFI genotypes and haplotypes showed significant associations with both TAFI measurements. In contrast, no association was observed between genetic variants and overall ischemic stroke.
Conclusion— TAFI levels show independent association with overall ischemic stroke. This association is stronger for released AP than for intact TAFI, and for released AP, it is present in all ischemic stroke subtypes.
Thrombin activatable fibrinolysis inhibitor (TAFI), also denoted procarboxypeptidase B1 and procarboxypeptidase U,2 is a zymogen present in human plasma. It can be activated by trypsin-like enzymes such as thrombin, plasmin, or the thrombin/thrombomodulin complex.3–5 On activation of intact TAFI, the activation peptide (AP) (Phe1–Arg92; 20 kDa) is released from the catalytic domain (TAFIa) (Ala93–Val401; 36 kDa).4 Both in vitro and in vivo experiments show that TAFIa retards fibrinolysis.6
TAFIa operates by continuously removing C-terminal lysine residues on plasmin-modified partially degraded fibrin, thus attenuating the rate of plasminogen activation and fibrinolysis.7 Because thrombin and thrombin/thrombomodulin complex can activate TAFI, and because TAFIa suppresses fibrinolysis, this pathway has been proposed to be a molecular link between coagulation and fibrinolysis.
The TAFI gene (named CPB2) maps to chromosome 13q14.11, spans approximately 48 kb and contains 11 exons.8 A number of TAFI gene polymorphisms have been identified, 2 of which result in amino acid substitutions, Thr147Ala and Thr325Ile. The Ile325 variant was shown to have a longer half-life and increased antifibrinolytic properties compared with the 325Thr variant.9 Initial studies reported association between several TAFI gene single nucleotide polymorphisms (SNPs) and circulating levels of TAFI antigen (TAFI Ag).10–12 Furthermore, a combined segregation-linkage analysis suggested the existence of 2 quantitative trait loci that would explain 78% of the variance in plasma TAFI Ag.13 However, it was demonstrated that some ELISAs used for TAFI Ag determination had different assay sensitivity between isoforms,14,15 leading to overestimations of the effects associated with TAFI gene polymorphisms. Results from recent studies using genotype-independent assays have confirmed a genetic effect although of lesser magnitude.16–18
Given the role of TAFI in fibrinolysis, several clinical studies have investigated the possible association between plasma levels of TAFI and thromboembolic events. As shown in a recent review,19 results are inconsistent, and this might be explained by the fact that different methods for TAFI determination have been used. Recent data suggest that it is not the total amount of TAFI protein but the amount of activated TAFI that plays a crucial role in retarding fibrinolysis.20,21 In view of this, it is of interest that there is a recent study reporting increased plasma levels of functional TAFI in acute ischemic stroke.22 However, ischemic stroke is a highly heterogeneous disorder, and as this study included only approximately 120 patients, it was not possible to study the association between TAFI and ischemic stroke subtypes.
Here we aimed to perform a comprehensive investigation of the possible association between ischemic stroke and TAFI among participants in the Sahlgrenska Academy Study on Ischemic Stroke (SAHLSIS), in which ischemic stroke subtype has been determined by Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria.23 To this end, TAFI was analyzed at the protein level by 2 newly developed immunologic assays with distinct reactivities, recognizing either intact TAFI or the released AP,24 and, at the genetic level, by a selection of tagging SNPs over the TAFI locus.
Materials and Methods
SAHLSIS has been described in detail elsewhere.23,25 In short, the study comprises 600 consecutive white patients <70 years of age presenting with acute ischemic stroke and 600 healthy white community controls, matched for age (±1 year), sex, and geographic residence area. All patients underwent neuroimaging and were examined by a physician, both at admission and at 3-month follow-up. Stroke subtype was classified by 2 neurologists according to TOAST and Oxfordshire Community Stroke Project (OCSP) criteria to provide information on etiologic subtype and on the clinical extent of brain damage,26 respectively. Analyses by TOAST subtype were confined to large vessel disease (LVD) (n=73), small vessel disease (SVD) (n=124), cardioembolic (CE) (n=98), and cryptogenic stroke (n=162). Functional outcome after 3 months was assessed using the modified Rankin scale. The study was approved by the Ethics Committee of Göteborg University. All participants gave written informed consent. When patients were unable to communicate, his or her next-of-kin gave informed consent.
Blood Sampling and TAFI Measurements
Among stroke cases, blood sampling was performed within 10 days of the stroke event and at 3-month follow-up. On both occasions, as well as in controls, venous blood samples were collected in tubes containing 10% volume of 0.13 mol/L sodium citrate between 8:30 and 10:30 am after an overnight fast. Plasma was isolated within 2 hours by centrifugation at 4°C and 2000g for 20 minutes. Plasma levels of TAFI were measured by sandwich-type ELISAs using MA-T12D11/MA-T30E5-HRP for intact TAFI24 and MA-T12D11/MA-T18A8-HRP for released AP.24 Values are expressed relative to pooled human plasma. Intra- and interassay coefficients of variation were 6.2% and 8.3% for intact TAFI and 3.1% and 7.3% for AP.24 High-sensitivity C-reactive protein (CRP) was analyzed as described.25
Genotype data from the CEU population in the HapMap project27 was entered into Haploview software package version 3.2 (Jeffrey C. Barrett and Mark J. Daly, http://www.broad.mit.edu/personal/jcbarret/haploview) to select tag SNPs that would capture unmeasured variation over the TAFI gene with an r2>0.80. Using the “Solid spine of LD” setting, 7 SNPs were selected by TAGGER28 to capture all SNPs with MAF>0.05 over a 52.4-kb single-haplotype block. In addition, we genotyped 3 previously studied SNPs (rs2146881, rs3742264, and rs940) and 1 SNP in exon 6 (rs7337140). A schematic presentation of the 11 SNPs is shown in Figure 1. Genotyping was performed by 5′ nuclease (TaqMan) assays (see the online data supplement for details, at http://atvb.ahajournals.org). Genotyping was performed blinded to case/control status.
Differences in characteristics between cases and controls were examined using the χ2 test for proportions and with Student t test for continuous variables. Because of a skewed distribution, both plasma TAFI measurements were analyzed with nonparametric tests or were logarithmically transformed.
Associations between plasma TAFI as well as single polymorphisms and ischemic stroke were investigated using binary logistic regression adjusted for age, sex, hypertension, smoking status, diabetes mellitus, hyperlipidemia, and waist-hip ratio. Associations between each SNP and TAFI levels were investigated using linear regression adjusted for the same covariates as well as CRP level. In all single-locus analyses of TAFI levels, additive and dominant or recessive models were considered. The genetic model showing strongest association with plasma level was used in single-locus analyses on case/control status. Levels of intact TAFI and AP were standardized so that reported odds ratios (ORs) represent an increase by 1 SD (0.234 for logarithmically transformed intact TAFI levels and 0.247 for AP). In ischemic stroke subtype regression models, the entire control population was used, and, when all cases were considered, conditional logistic regression was used. Adjustment for multiple testing was not made. Data were analyzed using SPSS 12.0.1, and statistical analyses were performed in a 2-tailed fashion. P<0.05 was considered significant.
Allele frequencies were derived from genotype data (see the online data supplement), and deviations from the Hardy–Weinberg equilibrium were tested. Haplotype frequencies, geometric means, and pairwise linkage disequilibrium (LD) coefficients, D′, were estimated using the THESIAS software version 3.1.29 Haplotype analyses were also performed with THESIAS. The method uses a Stochastic-EM algorithm for likelihood maximization and allows for simultaneous estimation of haplotype–phenotype association parameters. Covariate-adjusted odds ratios were estimated for each haplotype by comparison with a reference haplotype, represented by the most frequent haplotype.
The number of individuals with missing values has been reported.25 Samples for measurement of plasma TAFI were missing in 2 controls and 25 acute stroke patients. In 46 patients, samples at follow-up were missing because of intervening death,7 technical difficulties,15 or the patient was unwilling to take part in the follow-up examination or provide blood samples.24 Information on missing genotype data are available in the online data supplement. In the regression models, missing values for continuous variables were replaced by the mean value, and, for categorical variables, dummy variables were introduced.
Plasma TAFI in Overall Ischemic Stroke and TOAST Stroke Subtypes
Classic risk factors and CRP levels in SAHLSIS have been described elsewhere.23,25 Compared with controls, overall ischemic stroke showed significantly higher levels of intact TAFI and released AP both in the acute phase and at 3-month follow-up (Figure 2). In patients, released AP levels were slightly higher at follow-up compared with the acute phase (Δln[AP]=−0.024; P=0.03), whereas no significant difference between measurements was detected for intact TAFI. Intact TAFI and AP showed a significant correlation (P<0.001 throughout); Spearman’s ρ was 0.45 in controls and 0.48 and 0.42 in patients in the acute phase and at follow-up, respectively.
In the acute phase, levels of released AP were higher in all TOAST subtypes and intact TAFI in LVD, cryptogenic stroke, and SVD, compared with controls (Figure 2). ANOVA showed a subtype-specific difference for both TAFI measurements. Tukey’s post hoc test revealed a difference between LVD and CE stroke (P=0.01 for both intact TAFI and AP), whereas no differences were detected among other subtypes. There were no significant differences with regard to median time of sampling in relation to stroke onset among TOAST subtypes (data on file). Furthermore, there were no differences in mean intact TAFI or AP level with respect to time of sampling (P>0.20 for both measurements). At 3-month follow-up, there were no significant subtype-specific differences for any TAFI measure. Released AP remained higher in all subtypes, whereas intact TAFI was higher in LVD and cryptogenic stroke compared with controls (Figure 2).
In univariate binary logistic regression analyses, released AP showed significant associations to overall ischemic stroke and to all subtypes at both time points (data not shown). All associations remained after adjustment for traditional risk factors, and adjusted ORs are shown in Figure 3. For intact TAFI, independent associations were observed for overall ischemic stroke and all subtypes apart from CE stroke in the acute phase and for overall ischemic stroke, LVD, and cryptogenic stroke at 3-month follow-up (Figure 3).
Classic Risk Factors and Plasma TAFI
Correlations between plasma TAFI and classic risk factors have been reported elsewhere for a subgroup of the present control group.24 In the present sample, classic risk factors explained 3.3% and 10.2% of the variation in intact TAFI and AP in controls. The corresponding figures were 4.0% and 5.4% for patients in the acute phase and 4.5% and 5.2% at follow-up, respectively. In a subgroup of controls who did not receive any treatment for hypertension, diabetes, or hyperlipidemia (n=485), categorical variables were replaced with corresponding continuous variables. In this group, classic risk factors explained 10.1% and 16.7% of the variation in intact TAFI and AP, respectively.
Plasma TAFI in Relation to the Clinical Extent of Infarction and Outcome After Three Months
There were no significant differences in released AP or in intact TAFI between cases with clinical presentation, indicating an extensive infarct (total anterior circulation infarcts) and other OCSP groups (P>0.4). Furthermore, there were no significant differences in TAFI levels within OCSP groups between the acute phase and 3-month follow-up (P>0.1). Also, the correlations to high-sensitivity CRP were relatively weak (Spearman’s ρ, ≈0.10; P<0.05 for all TAFI-CRP correlations). We have previously shown that functional outcome after 3 months varies by TOAST subtype in this sample.23 TOAST subtype was therefore entered into multivariate models examining the relationship of plasma TAFI variables on outcome. TAFI levels in the acute phase were similar in cases with an unfavorable outcome at 3 months (death or dependency, ie, modified Rankin scale score of 3 to 6) compared with those with a favorable outcome (score 0 to 2). Exclusion of individuals who died before follow-up did not alter these results. At follow-up, however, TAFI levels were significantly higher in cases with an unfavorable outcome (adjusted ORs for death or dependency for intact TAFI and released AP, respectively; intact TAFI: 1.31; 95% confidence interval [CI], 1.02 to 1.68; P=0.04; and released AP: OR, 1.39; 95% CI, 1.06 to 1.83; P=0.02).
LD Structure of the TAFI Locus
An overview of the TAFI locus, typed SNPs, and the observed LD pattern is displayed in Figure 1. SNPs were distributed into 2 haplotype blocks. We identified 4 common (frequency >1%) haplotypes in block 1 (Table), accounting for 96% of the chromosomes. These 4 haplotypes could be identified by 5 tag SNPs, rs2146881 and rs1022953 being in complete LD. Similarly, in block 2, 7 haplotypes accounting for 96% of the chromosomes were tagged by 4 SNPs; rs7337140 and rs940 were in complete LD (Table). In comparison, 12 haplotypes accounted for 93% of the chromosomes when the TAFI locus was analyzed as a single-haplotype block.
TAFI Gene Variation and Plasma TAFI
All genotype distributions were compatible with Hardy–Weinberg equilibrium and are available in the online data supplement. In both cases and controls, TAFI levels differed by single SNP genotype group. Intact TAFI differed for 9 and released AP for 8 of the 11 SNPs (P<0.01 for all) (data not shown). Multivariate analysis showed that the percentage of variance explained by these SNPs varied from 4% to 12% (data not shown). The SNPs that showed strongest associations were rs9526136 and rs17067700. Including both of these variants in the multivariate linear model on data from the control group showed that they explained 16.7% and 13.7% of the variance in released AP and intact TAFI, respectively. Haplotype analysis revealed that the most common haplotype in block 1 (H1A) was associated with a higher expected mean of both released AP and intact TAFI compared with all other H1 haplotypes (P<0.001) (Figure 4). However, no single SNP in our haplotype structure was able to distinguish between H1A and other H1 haplotypes. In block 2, H2B and H2E showed higher levels of both TAFI measurements (P<0.01), suggesting an effect of the intronic rs9526136G allele, or a SNP in LD with that allele (Figure 4). Furthermore, H2D and H2F were associated with lower levels (P<0.05) compared with H2A, suggesting an effect of an allele in LD with the silent rs7337140A allele in exon 6 and with the rs940G allele.
TAFI Gene Variation in Relation to Ischemic Stroke
No significant association was detected between any SNP and overall ischemic stroke in single-locus models. Similarly, no haplotype from either block showed a significant association with overall ischemic stroke (P>0.06). However, subtype analysis revealed an association between H2B and cryptogenic stroke, with an increased risk for H2B carriers (adjusted OR, 1.57; 95% CI, 1.03 to 2.40; P=0.03). Furthermore, the H1B haplotype was associated with a decreased risk of SVD (adjusted OR, 0.55; 95% CI 0.35 to 0.87; P=0.01), and an increased risk of SVD was associated with H2D (adjusted OR, 1.99; 95% CI 1.16 to 3.40, P=0.01) as well as H2E (adjusted OR, 2.48; 95% CI, 1.38 to 4.44; P<0.01). No significant association was observed for any other subtype.
In this large case-control study, we found increased levels of both intact TAFI and released AP, as measured by 2 novel genotype-independent ELISAs, in ischemic stroke patients. The association between plasma TAFI and ischemic stroke was independent of classic risk factors, and it was stronger for released AP than for intact TAFI. Released AP showed independent associations to all ischemic stroke subtypes, whereas intact TAFI showed association with LVD, cryptogenic stroke, and acute-phase SVD. We also found associations between genetic variation at the TAFI locus and both TAFI measurements, but no significant association was detected for overall ischemic stroke.
The finding of increased plasma levels of TAFI in overall ischemic stroke is in agreement with 3 previous studies, despite somewhat different study designs.22,30,31 Montaner et al showed that TAFI Ag levels, measured within 24 hours of onset of the event, were increased in a small sample (n=30) of elderly ischemic stroke patients.31 Santamaría et al investigated 114 young (mean age 56 years) patients with ischemic stroke or transient ischemic attacks.30 TAFI levels, as measured within the first month after the event by an activity assay after full activation of TAFI with thrombin–thrombomodulin, were increased in patients compared with controls. Leebeek et al found increased functional TAFI levels, as determined by a clot lysis assay in 124 patients with a recent first episode of ischemic stroke.22 A subgroup of 36 patients was investigated at 3-month follow-up, and the results indicated that the finding was not attributable to an acute-phase response. Our study included 600 ischemic stroke patients and 600 controls with a mean age of 56 years. It is the first study to measure levels of both intact TAFI and released AP, a marker of the extent of TAFI activation.24 By measuring plasma levels both in the acute phase and at 3-month follow-up, we were able to confirm findings from the work by Leebeek et al that increased TAFI levels in ischemic stroke do not reflect an acute-phase response.22 Indeed, the present study provides further data to support this hypothesis. Firstly, plasma levels of released AP were even higher at 3-month follow-up compared with the acute phase. Secondly, the correlation to high-sensitivity CRP levels was very weak. Finally, patients with clinical signs of extensive infarction did not display higher TAFI levels, which is in line with a smaller study.31
This is also the first large study to explore associations between TAFI levels and TOAST subtypes. In the acute phase, LVD had higher plasma levels of both TAFI measurements than cardioembolic stroke. The only subtype with a significant proportion of patients on anticoagulant therapy was CE stroke. One might speculate that anticoagulant therapy in the CE group contributed to this finding. However, there was no significant difference in acute TAFI levels between CE stroke and either SVD or cryptogenic stroke. Furthermore, a large proportion of patients in the CE group received anticoagulant therapy after the acute-phase blood sample had been drawn. Consequently, the fact that there were no significant subtype-specific differences in plasma TAFI at follow-up speaks against the hypothesis that anticoagulants had any major impact on the results. In line with this, there were no significant differences in any TAFI measurement between patients with or without anticoagulant therapy, either in the CE group or in overall ischemic stroke. Results from multivariate analysis showed an independent association between released AP and all four major ischemic stroke subtypes, indicating that TAFI activation may contribute to the development of ischemic stroke irrespective of the underlying etiology. Indeed the adjusted ORs for follow-up released AP levels ranged from 2 to 3 for all subtypes.
As recently described, released AP is a marker of TAFIa.24 A true measurement of TAFIa requires an approach with regard to both preanalytical conditions and analytic procedures that is not feasible in a clinical study as large as the present one. One might argue that the low cross-reactivity of the AP specific ELISA for intact TAFI (ie, 0.7±0.6%24) can interfere with the interpretation of the data. However, there are several findings speaking against this hypothesis. Firstly and foremost, released AP showed stronger associations to ischemic stroke as compared with intact TAFI. Secondly, the time pattern differed with highest released AP levels at follow-up, whereas the highest intact TAFI levels were seen in the acute phase. Thirdly, released AP showed independent association with all subtypes, whereas intact TAFI only showed association with LVD, cryptogenic stroke, and acute-phase SVD.
Our study also included a comprehensive genetic analysis. To the best of our knowledge, this is the first study of the TAFI gene based on a selection of tagging SNPs from the HapMap project. Classic risk factors explained only 10% to 15% of the variance in both released AP and intact TAFI levels. The combined effect of 2 intronic SNPs, rs9526136 and rs17067700, explained an additional 10% to 15% of the variance in both measurements. This is in line with a recent study by Frère et al showing that genetic variation at the TAFI locus explain approximately 15% of the variability in TAFI Ag levels.18 The authors report that TAFI levels are influenced by 3 quantitative trait nucleotides, 1 in the 3′ untranslated region (1583T>A SNP [rs1087]) and 2 in the 5′ region (−2599C>G and −2345_-2344insG).18 Analyzing differences in TAFI levels across H1 haplotypes in the present study indicated that the association for the intronic rs17067700 was explained by partial LD with the 5′ quantitative trait nucleotides. Furthermore, H2 haplotypes carrying the rs9526136G allele presented highest expected phenotypic means, suggesting that the intronic rs9526136 is a marker of the 3′ untranslated region quantitative trait nucleotides. In line with this, analysis of CEU data from HapMap suggest a strong LD between rs9526136 and rs1087 (r2=0.96).
In accordance with a previous smaller study, we failed to detect any association between TAFI gene variants and overall ischemic stroke.22 However, analysis by TOAST subtype showed an increased risk of cryptogenic stroke in the H2B group, which displays increased TAFI levels. Interestingly, the only difference between the H2A and H2B haplotypes is the rs9526136, which is in strong LD with the putative 3′ untranslated region quantitative trait nucleotide rs1087. The diagnosis of cryptogenic stroke was made when no cause was identified despite an extensive evaluation. It follows that this group represents patients in whom factors other than those directly related to atherosclerosis, lipohyalinosis, and heart disease are relatively more important. For SVD, a decreased risk was observed for H1B and an increased risk for H2D and H2E. In contrast to the finding of cryptogenic stroke, this is difficult to reconcile from a mechanistic point of view because TAFI H2D shows lower and H2E higher plasma levels of both TAFI measurements compared with H2A. However, given the relatively weak associations these may be chance findings, and it follows that genetic associations for subtypes need to be investigated in future studies.
The present design offers some advantages/strengths. These include a large sample size, comprehensive classification of ischemic stroke subtypes, standardized blood sampling at 2 different time points, genotype-independent assays measuring both intact TAFI and released AP, multivariate analysis, and a comprehensive genetic analysis based on HapMap data. There are also some limitations that need to be considered. Firstly, the case-control design is a limitation for interpreting results on plasma levels and other factors that vary over time. Confounding effects of pharmacological therapy in cases cannot be excluded. Consequently multivariate adjustment for classic risk factors was performed using qualitative variables. Secondly, the selection of tagging SNPs were restricted to capture unmeasured variation over a 52.4-kb region with MAF>0.05 according to HapMap data, and we cannot exclude that some rare variants have been missed. However, the frequency spectrum of susceptibility variants for complex phenotypes is likely to include common variants27 and our approach is appropriate to capture these by LD. Finally, case and control ascertainment may influence results via selection bias. However, the stroke admission rate in Sweden is high and the early case fatality rate in ischemic stroke for the age group studied here is low. Furthermore, the control group was recruited by random sampling from the general population in the same geographic areas as patients. This makes the possibility of spurious results caused by population stratification less likely.
In conclusion, plasma TAFI was elevated in ischemic stroke compared with controls, and this difference was more pronounced for released AP than for intact TAFI. With regard to released AP, this association was observed for all ischemic stroke subtypes, indicating that elevated plasma TAFIa increases the risk of cerebral thromboembolic events independent of the underlying etiology. Genetic variation at the TAFI locus explains approximately 15% of the variation in plasma levels of both intact TAFI and released AP. Despite this, no association was found between genetic variants and overall ischemic stroke.
We thank collaborators within the SAHLSIS group. We also express our gratitude to the Göteborg Genomics Unit, Sweden, for excellent genotyping facilities.
Sources of Funding
The present study was supported by the Swedish Research Council (K2005-71X-14605-03A), the Swedish Heart-Lung Foundation 20050603, grants from the Swedish government under the LUA/ALF agreement (ALFGBG2964), the Health and Medical Care Executive Board of the Region Västra Götaland, the Swedish Stroke Association, the Swedish Hypertension Society, the Rune and Ulla Amlövs Foundation for Neurological Research, the John and Brit Wennerström Foundation for Neurological Research, the Per-Olof Ahl Foundation for Neurological Research, the Yngve Land Foundation for Neurological Research, the Katholieke Universiteit Leuven Research Fund (OT/06/66), and the Fund for Scientific Research-Flanders (G.0407.06). A.G. is a postdoctoral fellow of the Fund for Scientific Research-Flanders.
Original received October 10, 2006; final version accepted January 9, 2007.
Eaton DL, Malloy BE, Tsai SP, Henzel W, Drayna D. Isolation, molecular cloning, and partial characterization of a novel carboxypeptidase B from human plasma. J Biol Chem. 1991; 266: 21833–21838.
Bajzar L, Morser J, Nesheim M. TAFI, or plasma procarboxypeptidase B, couples the coagulation and fibrinolytic cascades through the thrombin-thrombomodulin complex. J Biol Chem. 1996; 271: 16603–16608.
Wang W, Boffa MB, Bajzar L, Walker JB, Nesheim ME. A study of the mechanism of inhibition of fibrinolysis by activated thrombin-activatable fibrinolysis inhibitor. J Biol Chem. 1998; 273: 27176–27181.
Schneider M, Boffa M, Stewart R, Rahman M, Koschinsky M, Nesheim M. Two naturally occurring variants of TAFI (Thr-325 and Ile-325) differ substantially with respect to thermal stability and antifibrinolytic activity of the enzyme. J Biol Chem. 2002; 277: 1021–1030.
Henry M, Aubert H, Morange PE, Nanni I, Alessi MC, Tiret L, Juhan-Vague I. Identification of polymorphisms in the promoter and the 3′ region of the TAFI gene: evidence that plasma TAFI antigen levels are strongly genetically controlled. Blood. 2001; 97: 2053–2058.
Franco RF, Fagundes MG, Meijers JC, Reitsma PH, Lourenco D, Morelli V, Maffei FH, Ferrari IC, Piccinato CE, Silva WA Jr, Zago MA. Identification of polymorphisms in the 5′-untranslated region of the TAFI gene: relationship with plasma TAFI levels and risk of venous thrombosis. Haematologica. 2001; 86: 510–517.
Brouwers GJ, Vos HL, Leebeek FW, Bulk S, Schneider M, Boffa M, Koschinsky M, van Tilburg NH, Nesheim ME, Bertina RM, Gomez Garcia EB. A novel, possibly functional, single nucleotide polymorphism in the coding region of the thrombin-activatable fibrinolysis inhibitor (TAFI) gene is also associated with TAFI levels. Blood. 2001; 98: 1992–1993.
Gils A, Alessi MC, Brouwers E, Peeters M, Marx P, Leurs J, Bouma B, Hendriks D, Juhan-Vague I, Declerck PJ. Development of a genotype 325-specific proCPU/TAFI ELISA. Arterioscler Thromb Vasc Biol. 2003; 23: 1122–1127.
Frere C, Morange PE, Saut N, Tregouet DA, Grosley M, Beltran J, Juhan-Vague I, Alessi MC. Quantification of thrombin activatable fibrinolysis inhibitor (TAFI) gene polymorphism effects on plasma levels of TAFI measured with assays insensitive to isoform-dependent artefact. Thromb Haemost. 2005; 94: 373–379.
Frère C, Tregouet DA, Morange PE, Saut N, Kouassi D, Juhan-Vague I, Tiret L, Alessi MC. Fine mapping of quantitative trait nucleotides underlying thrombin-activatable fibrinolysis inhibitor antigen levels by a transethnic study. Blood. 2006; 108: 1562–1568.
Walker JB, Bajzar L. The intrinsic threshold of the fibrinolytic system is modulated by basic carboxypeptidases, but the magnitude of the antifibrinolytic effect of activated thrombin-activatable fibrinolysis inhibitor is masked by its instability. J Biol Chem. 2004; 279: 27896–27904.
Jood K, Ladenvall C, Rosengren A, Blomstrand C, Jern C. Family history in ischemic stroke before 70 years of age: the Sahlgrenska Academy Study on Ischemic Stroke. Stroke. 2005; 36: 1383–1387.
Ceresa E, Brouwers E, Peeters M, Jern C, Declerck PJ, Gils A. Development of ELISAs measuring the extent of TAFI activation. Arterioscler Thromb Vasc Biol. 2006; 26: 423–428.
Ladenvall C, Jood K, Blomstrand C, Nilsson S, Jern C, Ladenvall P. Serum C-reactive protein concentration and genotype in relation to ischemic stroke subtype. Stroke. 2006; 37: 2018–2023.
Lindgren A, Norrving B, Rudling O, Johansson BB. Comparison of clinical and neuroradiological findings in first-ever stroke. A population-based study. Stroke. 1994; 25: 1371–1377.
Santamaría A, Oliver A, Borrell M, Mateo J, Belvis R, Marti-Fabregas J, Ortin R, Tirado I, Souto JC, Fontcuberta J. Risk of ischemic stroke associated with functional thrombin-activatable fibrinolysis inhibitor plasma levels. Stroke. 2003; 34: 2387–2391.
Montaner J, Ribo M, Monasterio J, Molina CA, Alvarez-Sabin J. Thrombin-activatable fibrinolysis inhibitor levels in the acute phase of ischemic stroke. Stroke. 2003; 34: 1038–1040.