The association between protein S levels and anticoagulant activity of tissue factor pathway inhibitor type 1

Protein S (PS) is a vitamin K-dependent coagulation inhibitor of which 40% circulates in a free form and 60% circulates bound to C4b-binding protein (C4BP). PS exerts its anticoagulant activity mainly by being a cofactor for activated protein C (APC) in the inactivation of factor (F) Va and FVIIIa [1Rezende S.M. Simmonds R.E. Lane D.A. Coagulation, inflammation, and apoptosis: different roles for protein S and the protein S-C4b binding protein complex.Blood. 2004; 103: 1192-201Crossref PubMed Scopus (154) Google Scholar]. Tissue factor (TF) pathway inhibitor type 1 (TFPI) is the physiological inhibitor of the TF pathway of blood coagulation. Approximately 80% of circulating TFPI in blood is C-terminal truncated and bound to lipoproteins, while 20% is carrier free full-length TFPI [2Dahm A. Rosendaal F.R. Andersen T.O. Sandset P.M. Tissue factor pathway inhibitor anticoagulant activity: risk for venous thrombosis and effect of hormonal state.Br J Haematol. 2006; 132: 333-8Crossref PubMed Scopus (14) Google Scholar]. The TFPI-mediated inhibition of the TF/FVIIa complex occurs in two steps. Firstly, TFPI binds and inhibits FXa. Secondly, the TFPI/FXa complex binds to the complex of TF and FVIIa to form an inactive quaternary TFPI/FXa/TF/FVIIa complex: 1 The first step of this reaction is rate limiting [3Lindhout T. Franssen J. Willems G. Kinetics of the inhibition of tissue factor-factor VIIa by tissue factor pathway inhibitor.Thromb Haemost. 1995; 74: 910-5Crossref PubMed Scopus (52) Google Scholar], and it is much faster with full-length TFPI than with truncated TFPI [4Wesselschmidt R. Likert K. Girard T. Wun T.C. Broze Jr, G.J. Tissue factor pathway inhibitor: the carboxy-terminus is required for optimal inhibition of factor Xa.Blood. 1992; 79: 2004-10Crossref PubMed Google Scholar, 5Lindhout T. Salemink I. Valentin S. Willems G.M. Tissue factor pathway inhibitor: regulation of its inhibitory activity by phospholipid surfaces.Haemostasis. 1996; 26: 89-97PubMed Google Scholar]. Recently, Hackeng et al. showed in an in vitro model in the absence of APC that PS increases the reaction rate between full-length TFPI and FXa. Moreover, this increase was not observed for truncated TFPI [6Hackeng T.M. Sere K.M. Tans G. Rosing J. Protein S stimulates inhibition of the tissue factor pathway by tissue factor pathway inhibitor.Proc Natl Acad Sci U S A. 2006; 103: 3106-11Crossref PubMed Scopus (241) Google Scholar]. Hence, PS enhances the inhibition of coagulation by full-length TFPI. We have recently reported a TFPI anticoagulant activity assay based on the ability of TFPI to inhibit clotting in a modified diluted prothrombin time assay. This assay measures the activity of full-length TFPI, but not of truncated TFPI [7Dahm A.E. Andersen T.O. Rosendaal F. Sandset P.M. A novel anticoagulant activity assay of tissue factor pathway inhibitor I (TFPI).J Thromb Haemost. 2005; 3: 651-8Crossref PubMed Scopus (33) Google Scholar]. We suggested that the assay discriminated between full-length and truncated TFPI in relation to differences in reaction kinetics with FXa. Based on the findings of Hackeng et al., we hypothesized that the anticoagulant activity of full-length TFPI was associated with the concentration of PS in plasma, and that the concentration of PS would influence the risk for venous thrombosis (VT) associated with low TFPI anticoagulant activity and vice versa. Although the results by Hackeng et al. are primarily applicable to free PS and full-length TFPI, they also found some anticoagulant effect of PS bound to C4BP. We therefore decided to investigate both free and bound PS. To test these hypotheses we used data collected in the Leiden Thrombophilia Study (LETS) that included free and total PS [8Koster T. Rosendaal F.R. Briet E. Van Der Meer F.J. Colly L.P. Trienekens P.H. Poort S.R. Reitsma P.H. Vandenbroucke J.P. Protein C deficiency in a controlled series of unselected outpatients: an infrequent but clear risk factor for venous thrombosis (Leiden Thrombophilia Study).Blood. 1995; 85: 2756-61Crossref PubMed Google Scholar], TFPI free and total antigen, and TFPI anticoagulant activity [7Dahm A.E. Andersen T.O. Rosendaal F. Sandset P.M. A novel anticoagulant activity assay of tissue factor pathway inhibitor I (TFPI).J Thromb Haemost. 2005; 3: 651-8Crossref PubMed Scopus (33) Google Scholar]. LETS is a population-based case–control study of 474 cases with deep vein thrombosis (DVT) and 474 healthy controls matched for sex and age, excluding individuals with cancer. The study has been described in detail previously [9Koster T. Rosendaal F.R. De Ronde H. Briet E. Vandenbroucke J.P. Bertina R.M. Venous thrombosis due to poor anticoagulant response to activated protein C: Leiden Thrombophilia Study.Lancet. 1993; 342: 1503-6Abstract PubMed Scopus (1322) Google Scholar]. The association between free PS and TFPI anticoagulant activity was studied in the 422 controls in whom both free PS and TFPI anticoagulant activity had been assayed. Correspondingly, the association between total PS and TFPI anticoagulant activity was studied in 436 controls. For the calculation of risk for DVT, users of oral contraceptives were excluded due to their confounding effect on TFPI, and users of oral anticoagulation were excluded due to their effect on prothrombin time and PS, as described previously [2Dahm A. Rosendaal F.R. Andersen T.O. Sandset P.M. Tissue factor pathway inhibitor anticoagulant activity: risk for venous thrombosis and effect of hormonal state.Br J Haematol. 2006; 132: 333-8Crossref PubMed Scopus (14) Google Scholar]. Hence, there were 362 cases and 327 controls available for the risk calculations. Single and multiple linear regression with TFPI anticoagulant activity as the dependent variable was used to study how free and total PS was associated with TFPI anticoagulant activity in the controls. The results are presented as regression coefficients with 95% confidence intervals (95% CIs). The risk for DVT is presented as an odds ratio (OR) calculated by logistic regression with 95% CIs derived from the likelihood equation. PS was assayed with conventional ELISAs and the total PS assay included PS bound to C4BP [8Koster T. Rosendaal F.R. Briet E. Van Der Meer F.J. Colly L.P. Trienekens P.H. Poort S.R. Reitsma P.H. Vandenbroucke J.P. Protein C deficiency in a controlled series of unselected outpatients: an infrequent but clear risk factor for venous thrombosis (Leiden Thrombophilia Study).Blood. 1995; 85: 2756-61Crossref PubMed Google Scholar]. TFPI free antigen was assayed with an ELISA specific for full-length TFPI [7Dahm A.E. Andersen T.O. Rosendaal F. Sandset P.M. A novel anticoagulant activity assay of tissue factor pathway inhibitor I (TFPI).J Thromb Haemost. 2005; 3: 651-8Crossref PubMed Scopus (33) Google Scholar]. In the selected controls, mean free PS was 96.8% (range 45–153%), mean total PS was 102.4% (range 51–192%), while mean TFPI anticoagulant activity was 103.3% (range 87.1–122.4%). We found that the level of free PS was associated with the anticoagulant activity of TFPI, with a regression coefficient of 0.14 (95% CI 0.11–0.17). This means that a one-unit increase in free PS was accompanied by an average 0.14-unit increase in TFPI anticoagulant activity (Fig. 1). Similarly, total PS was associated with TFPI anticoagulant activity, with a regression coefficient of 0.15 (95% CI 0.13–0.18). Because TFPI free antigen (which is predominantly full-length TFPI) is the main determinant of TFPI anticoagulant activity [7Dahm A.E. Andersen T.O. Rosendaal F. Sandset P.M. A novel anticoagulant activity assay of tissue factor pathway inhibitor I (TFPI).J Thromb Haemost. 2005; 3: 651-8Crossref PubMed Scopus (33) Google Scholar], we adjusted for the effect of TFPI free antigen by using it as a covariate together with free PS or total PS in two multiple regression models, with TFPI anticoagulant activity as the dependent variable. In the first model, with TFPI free antigen and free PS as covariates, the regression coefficient obtained with simple linear regression for free PS was reduced from 0.14 to 0.059 (95% CI 0.032–0.086). In the second model, with total PS and TFPI free antigen as covariates, the simple regression coefficient for total PS was reduced from 0.15 to 0.064 (95% CI 0.038–0.091). The addition of hormonal state (oral contraceptive users/non-users, sex, pre/postmenopausal) and age to the two regression models resulted in a regression coefficient of 0.054 (95% CI 0.028–0.081) for the model with free PS, and a regression coefficient of 0.051 (95% CI 0.024–0.078) for the model including total PS. Thus, the association between free and total PS and TFPI anticoagulant activity was substantially reduced by adjusting for the levels of full-length TFPI, which means that the levels of free and total PS covary with the levels of TFPI free antigen in plasma. The mean total PS for the 362 cases included in the OR calculations was 108% (range 65–190%), the mean free PS for the cases was 99.5% (range 23–161%), and the mean TFPI anticoagulant activity for the cases was 105% (range 89.2–134%). The OR for DVT was 1.5 (95% CI 0.97–2.5) when TFPI anticoagulant activity was at or below the 10th percentile (97.8%) of the controls included in the OR calculations. When free PS was used as covariate in a logistic regression model, the OR for low TFPI anticoagulant activity was 1.7 (95% CI 1.0–2.8). When total PS was used as a covariate, the OR was 1.8 (95% CI 1.1–2.9). Hence, the levels of free and total PS did not influence the risk for DVT associated with low TFPI anticoagulant activity. Free PS and total PS were associated with TFPI anticoagulant activity to a similar degree. This is primarily a result of covariation between levels of full-length TFPI and free and total PS in plasma. The interaction between PS and full-length TFPI described by Hackeng et al. [6Hackeng T.M. Sere K.M. Tans G. Rosing J. Protein S stimulates inhibition of the tissue factor pathway by tissue factor pathway inhibitor.Proc Natl Acad Sci U S A. 2006; 103: 3106-11Crossref PubMed Scopus (241) Google Scholar] also seems to play a role: there was a clear tendency for persons with higher levels of free and total PS in their plasma to also have higher anticoagulant activity of TFPI, even after adjustment for the covariation with full-length TFPI. Hence, our results may be considered to be an epidemiological confirmation of the in vitro findings of Hackeng et al. The observed covariation between full-length TFPI and both free and total PS is of unknown nature, but one may speculate that TFPI and PS are regulated by similar mechanisms. The influence of PS on TFPI anticoagulant activity did not affect the risk for DVT. This is not surprising, as a low level of PS was not a risk factor for DVT in LETS [8Koster T. Rosendaal F.R. Briet E. Van Der Meer F.J. Colly L.P. Trienekens P.H. Poort S.R. Reitsma P.H. Vandenbroucke J.P. Protein C deficiency in a controlled series of unselected outpatients: an infrequent but clear risk factor for venous thrombosis (Leiden Thrombophilia Study).Blood. 1995; 85: 2756-61Crossref PubMed Google Scholar]. We would expect that in other studies where both low TFPI and low PS are risk factors for DVT these risks would influence each other. In studies of thrombophilic families, PS deficiency is associated with a strikingly increased relative risk for VT [10Vossen C.Y. Conard J. Fontcuberta J. Makris M. Van Der Meer F.J. Pabinger I. Palareti G. Preston F.E. Scharrer I. Souto J.C. Svensson P. Walker I.D. Rosendaal F.R. Familial thrombophilia and lifetime risk of venous thrombosis.J Thromb Haemost. 2004; 2: 1526-32Crossref PubMed Scopus (93) Google Scholar] compared with studies from the general population [8Koster T. Rosendaal F.R. Briet E. Van Der Meer F.J. Colly L.P. Trienekens P.H. Poort S.R. Reitsma P.H. Vandenbroucke J.P. Protein C deficiency in a controlled series of unselected outpatients: an infrequent but clear risk factor for venous thrombosis (Leiden Thrombophilia Study).Blood. 1995; 85: 2756-61Crossref PubMed Google Scholar]. Hence, other factors in the coagulation system might interact with PS in thrombophilic families with PS deficiency. It is likely that full-length TFPI is one such factor. The authors state that they have no conflict of interest.