Portal de Periódicos da CAPES

High titers of autoantibodies to tissue factor pathway inhibitor are associated with the antiphospholipid syndrome

2003; Elsevier BV; Volume: 1; Issue: 4 Linguagem: Inglês

10.1046/j.1538-7836.2003.00102.x

1538-7933

Ricardo Forastiero, Marta Martinuzzo, George Broze,

Monoclonal and Polyclonal Antibodies Research

SummaryAs the activity of the tissue factor pathway inhibitor (TFPI) may be impaired in patients with antiphospholipid antibodies (aPL), 162 aPL patients were evaluated for autoantibodies to recombinant TFPI (anti-TFPI) using an optimized ELISA. Anti-TFPI (>18 U mL−1 for IgG and/or > 15 U mL−1 for IgM) were detected in 54 patients with aPL (33.3%) and in three out of 79 normal controls (3.8%, P < 0.0001). Among aPL patients, the prevalence of positive anti-TFPI was 38.3 and 28.4% in those with or without diagnosis of definite antiphospholipid syndrome (APS). Patients with definite APS had a significantly greater frequency of high titer (>50 U mL−1) anti-TFPI than aPL patients from the no definite APS group (18.5% vs. 6.2%, OR 3.7, P= 0.017). Most aPL recognized full-length TFPI, but not a truncated form of TFPI lacking the C-terminus of the molecule. Isolated IgGs from subjects with anti-TFPI impaired the dose-dependent inhibitory effect of TFPI on factor Xa activity in the presence, but not in the absence of phospholipid vesicles. Thus, aPL with high titer anti-TFPI limit TFPI action and are associated with the APS. As the activity of the tissue factor pathway inhibitor (TFPI) may be impaired in patients with antiphospholipid antibodies (aPL), 162 aPL patients were evaluated for autoantibodies to recombinant TFPI (anti-TFPI) using an optimized ELISA. Anti-TFPI (>18 U mL−1 for IgG and/or > 15 U mL−1 for IgM) were detected in 54 patients with aPL (33.3%) and in three out of 79 normal controls (3.8%, P < 0.0001). Among aPL patients, the prevalence of positive anti-TFPI was 38.3 and 28.4% in those with or without diagnosis of definite antiphospholipid syndrome (APS). Patients with definite APS had a significantly greater frequency of high titer (>50 U mL−1) anti-TFPI than aPL patients from the no definite APS group (18.5% vs. 6.2%, OR 3.7, P= 0.017). Most aPL recognized full-length TFPI, but not a truncated form of TFPI lacking the C-terminus of the molecule. Isolated IgGs from subjects with anti-TFPI impaired the dose-dependent inhibitory effect of TFPI on factor Xa activity in the presence, but not in the absence of phospholipid vesicles. Thus, aPL with high titer anti-TFPI limit TFPI action and are associated with the APS. The antiphospholipid syndrome (APS) is characterized by poor obstetrical outcomes and increased risk of vascular thrombosis involving both venous and arterial territories in the presence of antiphospholipid antibodies (aPL) [1Wilson W. Gharavi A. Koike T. Lockshin M. Branch W. Piette J. Brey R. Derksen R. Harris E. Hughes G. Triplett D. Khamashta M. International consensus statement on preliminary classification criteria for definite antiphospholipid syndrome.Report of and International workshop. Arthritis Rheum. 1999; 42: 1309-11Crossref PubMed Scopus (0) Google Scholar]. It is defined as primary APS in the absence of any underlying disease, and as secondary APS when occurring in the setting of systemic lupus erythematosus (SLE) or SLE-like syndromes [2Love P.E. Santoro S.A. Antiphospholipid antibodies. anticardiolipin and the lupus anticoagulant in systemic lupus erythematosus (SLE) and in non-SLE disorders. Prevalence and clinical significance.Ann Intern Med. 1990; 112: 682-98Crossref PubMed Google Scholar]. Experimental evidence suggests that autoimmune aPL may play a causative role in vascular thrombosis and obstetrics complications [3Pierangeli S.S. Harris E.N. In vivo models of thrombosis for the antiphospholipid syndrome.Lupus. 1996; 5: 451-5Crossref PubMed Google Scholar, 4Holers V.M. Girardi G. Mo L. Guthridge J.M. Molina H. Pierangeli S.S. Espinola R. Xiaowei L.E. Mao D. Vialpando C.G. Salmon J.E. Complement C3 activation is required for antiphospholipid antibody-induced fetal loss.J Exp Med. 2002; 195: 211-20Crossref PubMed Scopus (497) Google Scholar]. The antigen specificity of aPL is highly heterogeneous. It is now widely accepted that autoimmune aPL reacting with β2-glycoprotein I (β2GPI) (anti-β2GPI) are mainly responsible for anti-cardiolipin antibody (aCL) reactivity, whereas those with specificity towards β2GPI or prothrombin (anti-PT) may display lupus anticoagulant (LA) properties. A number of clinical studies have shown that the antigenic specificity of aPL is related to certain clinical features of APS [5Carreras L.O. Forastiero R.R. Martinuzzo M.E. Which are the best biological markers of the antiphospholipid syndrome?.J Autoimmun. 2000; 15: 163-72Crossref PubMed Scopus (0) Google Scholar, 6Forastiero R.R. Martinuzzo M.E. Cerrato G.S. Kordich L.C. Carreras L.O. Relationship of anti β2-glycoprotein I and anti prothrombin antibodies to thrombosis and pregnancy loss in patients with antiphospholipid antibodies.Thromb Haemost. 1997; 78: 1008-14Crossref PubMed Scopus (143) Google Scholar, 7Nojima J. Kuratsune H. Suehisa E. Futsukaichi Y. Yamanishi H. Machii T. Kitani T. Iwatani Y. Kanakura Y. Anti-prothrombin antibodies combined with lupus anticoagulant activity is an essential risk factor for venous thromboembolism in patients with systemic lupus erythematosus.Br J Haematol. 2001; 114: 647-54Crossref PubMed Scopus (0) Google Scholar, 8Cabiedes J. Cabral A.R. Alarcón-Segovia D. Clinical manifestations of the antiphospholipid syndrome in patients with systemic lupus erythematosus associate more strongly with anti-β2-glycoprotein-I than with antiphospholipid antibodies.J Rheumatol. 1995; 22: 1899-906PubMed Google Scholar, 9Guerin J. Feighery C. Sim R.B. Jackson J. Antibodies to β2-glycoprotein I − a specific marker for the antiphospholipid syndrome.Clin Exp Immunol. 1997; 109: 304-9Crossref PubMed Google Scholar]. Several potential mechanisms have been proposed to explain the pathogenic role of aPL [10Carreras L.O. Forastiero R.R. Pathogenic role of antiprotein-phospholipid antibodies.Haemostasis. 1996; 26: 340-57PubMed Google Scholar]. Recent evidence suggests that up-regulation of the tissue factor (TF) pathway and impairment of the tissue factor pathway inhibitor (TFPI) activity in aPL patients may be important in contributing to the hypercoagulable state seen in APS [11Roubey R.A.S. Tissue factor pathway and the antiphospholipid syndrome.J Autoimmun. 2000; 15: 217-20Crossref PubMed Scopus (53) Google Scholar, 12Amengual O. Atsumi T. Khamashta M.A. Hughes G.R.V. The role of the tissue factor pathway in the hypercoagulable state in patients with the antiphospholipid syndrome.Thromb Haemost. 1998; 79: 276-81Crossref PubMed Scopus (263) Google Scholar, 13Adams M.J. Donohoe S. Mackie I.J. Machin S.J. Anti-tissue factor pathway inhibitor activity in patients with primary antiphospholipid syndrome.Br J Haematol. 2001; 114: 375-9Crossref PubMed Scopus (32) Google Scholar]. Furthermore, Salemink et al. [14Salemink I. Blezer R. Willems G.M. Galli M. Bevers E. Lindhout T. Antibodies to β2-glycoprotein I associated with antiphospholipid syndrome suppress the inhibitory activity of tissue factor pathway inhibitor.Thromb Haemost. 2000; 84: 653-6Crossref PubMed Google Scholar] demonstrated that APS-related anti-β2GPI suppress the inhibitory activity of TFPI leading to an increased factor (F)Xa generation. The presence of antibodies directed against TFPI (anti-TFPI) has been recently suggested in a preliminary study including patients with APS or SLE [15Cakir B. Arnett F.C. Roubey R.A.S. Autoantibodies to tissue factor pathway inhibitor (TFPI) are associated with arterial thrombosis/stroke.J Autoimmun. 2000; 15: A11Google Scholar]. TFPI is a trivalent Kunitz-type protease inhibitor that tightly regulates TF-mediated coagulation by forming a quaternary inhibitory complex that contains FVIIa/tissue factor, FXa and TFPI [16Broze Jr, G.J. Tissue factor pathway inhibitor.Thromb Haemost. 1995; 74: 90-3Crossref PubMed Scopus (307) Google Scholar, 17Bajaj M.S. Birktoft J.J. Steer S.A. Bajaj S.P. Structure and biology of tissue factor pathway inhibitor.Thromb Haemost. 2001; 86: 959-72Crossref PubMed Google Scholar]. In the inhibitory complex, the first Kunitz domain (K1) of TFPI binds FVIIa and the second Kunitz domain (K2) binds FXa. TFPI can also directly inhibit FXa in the absence of FVIIa/tissue factor. With physiologic calcium ion concentrations, this process is enhanced by the presence of phospholipids, which presumably provide a surface for the simultaneous binding of FXa and TFPI [18Huang Z.F. Wun T.C. Broze Jr, G.J. Kinetics of factor Xa inhibition by tissue factor pathway inhibitor.J Biol Chem. 1993; 268: 26950-5Abstract Full Text PDF PubMed Google Scholar]. Potential mechanisms contributing to up-regulation of the TF pathway in the APS include increased expression and activity of TF on monocytes and endothelial cells, and decreased TFPI activity due to antibodies directed against TFPI or aPL competing to bind phospholipids on the cellular surface. In this study, an ELISA was designed to detect autoantibodies to TFPI in a large series of patients with aPL. We found that high titers of anti-TFPI are associated with the definite APS. Further studies were performed to determine the domain of TFPI recognized by anti-TFPI and whether these antibodies are able to interfere with TFPI-dependent inhibition of FXa. This study comprised 162 patients with persistent aPL (LA and/or aCL) who were tested because of autoimmune diseases, a history of APS-related clinical features, or abnormal routine clotting tests. The presence of aPL was verified on at least two occasions 3 months apart. Patients were recruited from our group of aPL positive patients consecutively studied between 1998 and 2002. Patients were grouped as having definite APS and not having definite APS (no definite APS group) based on current criteria [1Wilson W. Gharavi A. Koike T. Lockshin M. Branch W. Piette J. Brey R. Derksen R. Harris E. Hughes G. Triplett D. Khamashta M. International consensus statement on preliminary classification criteria for definite antiphospholipid syndrome.Report of and International workshop. Arthritis Rheum. 1999; 42: 1309-11Crossref PubMed Scopus (0) Google Scholar]. Their main clinical characteristics are shown in Table 1. Venous thrombotic events were documented by venography, Doppler ultrasonography, ventilation-perfusion lung scanning or pulmonary angiography. Computerized tomography scan and/or magnetic resonance imaging were used for diagnosis of arterial events. All patients were studied at least 6 months after their last episode of thrombosis. Women with a history of at least three unexplained consecutive abortions before the 10th week of gestation or one or more unexplained fetal deaths at or beyond the 10th week of gestation were considered to have aPL-related obstetric complications for diagnosis of definite APS. In the no definite APS group, there were three women who had had only one or two abortions. Twenty-nine patients were receiving oral anticoagulant therapy at the time of the study. All subjects were interviewed using a standardized questionnaire to collect clinical information about additional risk factors predisposing to thrombosis.Table 1Clinical and laboratory features of 162 aPL patientsDefinite APS nNo definite APS nPatients8181Sex (male/female)35/4624/57Median age in years (range)43 (5–77)37 (5–84)Previous thrombosis68–Venous45–Arterial30–Recurrence of thrombosis31–Previous obstetric complications173Transient cerebral ischemia34Systemic lupus erythematosus1024Thrombocytopenia (<150 × 109 L−1)1510Livedo reticularis12Hemolytic anemia23Severe hypoprothrombinemia–2LA + aCL4839LA alone1719aCL alone1623Anti-β2GPI5236Anti-PT3830 Open table in a new tab Seventy-nine healthy blood donors (32 males, 47 females, median age 37 years, range 19–63) with no history of thrombosis or autoimmune disorders were recruited as a normal control group. We also tested subjects with history of venous or arterial thrombosis (n = 35), poor obstetric history as defined above (n = 31), or SLE (n = 17) as control groups of patients without aPL. Informed consent was obtained from all participants. To test for the presence of the LA, blood samples were drawn by clean venipuncture into plastic tubes containing 0.11 mol L−1 trisodium citrate (ratio 9 : 1). After double centrifugation at 2500 × g for 15 min, platelet-poor plasma was assayed immediately. Sera were obtained from blood collected into glass tubes and allowed to clot at 37 °C. After centrifugation at 1500 × g for 10 min, sera samples were stored at − 80 °C until use. LA activity was identified through screening tests, mixing studies and confirmatory procedures according to the revised criteria of the Subcommittee for Standardization of Lupus Anticoagulants [19Brandt J.T. Triplett DA, Alving B, Scharrer I. Criteria for the diagnosis of lupus anticoagulant: an update.Thromb Haemost. 1995; 74: 1185-90Crossref PubMed Scopus (0) Google Scholar]. aCL of both isotypes were measured by using a standardized ELISA technique and international standards (Louisville APL Diagnostics, Louisville, KY, USA). Results were expressed as standard units (u) for IgG (GPL) or IgM (MPL). Titers greater than 20 u were required for a diagnosis of APS. The home-made ELISAs for anti-β2GPI and anti-PT were performed as previously reported using electron beam- (100 kGy) and γ-irradiated microtiter plates, respectively (Nunc MaxiSorp, Kamstrup, Roskilde, Denmark). The cut-off values (15 arbitrary units for IgG or IgM) were previously determined as the 99th percentiles of 95 normal sera [6Forastiero R.R. Martinuzzo M.E. Cerrato G.S. Kordich L.C. Carreras L.O. Relationship of anti β2-glycoprotein I and anti prothrombin antibodies to thrombosis and pregnancy loss in patients with antiphospholipid antibodies.Thromb Haemost. 1997; 78: 1008-14Crossref PubMed Scopus (143) Google Scholar]. Recombinant full-length and truncated human TFPI (TFPI1-161) were produced in Escherichia coli as previously described [18Huang Z.F. Wun T.C. Broze Jr, G.J. Kinetics of factor Xa inhibition by tissue factor pathway inhibitor.J Biol Chem. 1993; 268: 26950-5Abstract Full Text PDF PubMed Google Scholar]. The mouse monoclonal antibody (mAb) 2H8 (2.7 mg mL−1) is directed against Kunitz domain 1 of TFPI [20Broze Jr, G.J. Lange G.W. Duffin K.L. MacPhail L. Heterogeneity of plasma tissue factor pathway inhibitor.Blood Coagul Fibrinolysis. 1994; 5: 551-9PubMed Google Scholar]. The chromogenic substrate for FXa (S2765) was purchased from Chromogenix (Milano, Italy). Purified FXa was from Diagnostica Stago (Asnières, France). Protein A-Sepharose CL-4B, 1,2-dioleoyl-sn-glycero-3-phosphocholine (PC) and 1,2-dioleoyl-sn-glycero-3-phosphoserine (PS) were obtained from Sigma Chemical Co. (St. Louis, MI, USA). Whole IgG was isolated from serum by affinity chromatography on protein A Sepharose according to the manufacturer's instructions. Purity of the IgG preparations was checked by sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Immunoblot analysis confirmed that β2GPI was not present in any IgG preparation. The antibody concentration was determined using standard methods for proteins. ELISA plates (γ-irradiated, Nunc MaxiSorp) were coated overnight at 4 °C with 50 µL well−1 of recombinant full-length TFPI diluted to 10 µg mL−1 in 0.05 mol L−1 carbonate/bicarbonate buffer pH 9.6. This concentration (0.5 µg well−1) of TFPI was found to be optimal in the anti-TFPI assay. All further incubations were at room temperature. Each well was blocked for 60 min with 75 µL of 0.02 mol L−1 Tris-0.15 mol L−1 NaCl buffer (TBS) pH 7.4 containing 10% of bovine serum albumin (BSA) (Sigma). Immunoblot analysis confirmed that β2GPI was not present in BSA preparation. Thereafter, 50 µL of sample dilution (sera diluted 1/100 in 1% BSA–TBS) was added to each well, and incubated for 90 min. Horseradish peroxidase-conjugated goat antihuman IgG or IgM F(ab′)2 fragment of affinity isolated antibodies (Sigma) diluted 1 : 5000 in 1% BSA-TBS were added (50 µL) to the wells and incubated for 60 min. The plates were washed three times with 100 µL well−1 of TBS after coating, blocking, samples and conjugate incubations. The color was developed by adding 50 µL of o-phenylenediamine (Sigma) in phosphate–citrate buffer pH 5.0 containing hydrogen peroxide for 5 min. The plates were read at 492 nm using a MS2 TitertekPlus microplate reader (ICN-Flow, ICN Biomedicals, Aurra, OH, USA). All sera samples were run in duplicate, and for each serum the optical density (OD) measured on blank wells (coated with carbonate/bicarbonate buffer) was subtracted from the OD obtained in antigen-coated wells. Binding of TFPI to microtiter ELISA plates was confirmed by using an antihuman TFPI mAb diluted 1/10 (0.27 mg mL−1) followed by peroxidase-conjugated affinity-purified goat antimouse IgG. For IgG and IgM isotypes, selected sera from patients with aPL were used in each assay as positive controls in addition to negative controls. Results were expressed as U/mL, referred to internal standards (sera collected from patients with a high titer of anti-TFPI of IgG or IgM isotype) arbitrarily fixed at 100 U mL−1. The 99th percentile in 70 healthy controls was chosen as the cut-off point. The cut-off values were 18 U mL−1 and 15 U mL−1 for IgG and IgM, respectively. The inter- and intra-assay coefficients of variations for the IgG and the IgM assay were < 12 and < 5%, respectively. The ELISA for detecting antibodies to TFPI1-161 was performed as above but using 15 µg mL−1 solution of TFPI in carbonate–bicarbonate buffer pH 9.6. To measure the inhibition of anti-TFPI by fluid-phase recombinant TFPI, serum samples were first diluted in 1% BSA-TBS to twice the concentration exhibiting 50% of the maximum binding to TFPI by ELISA. Diluted samples were then preincubated with an equal volume of buffer or solutions of serially diluted TFPI (400–12.5 µg/mL) for 2 h at 37 °C and overnight at 4 °C. Samples were then retested on the TFPI ELISA. Liposomes containing 25 mol% PS and 75 mol% PC were prepared by vortexing dried phospholipid mixtures in TBS in the presence of recombinant TFPI. Appropriate dilutions of whole human IgG from patients and normal controls were incubated with the liposomes for 2 h at 37 °C. Final concentrations were 1.1 mg mL−1, 400 µg mL−1, 1 mg mL−1 for phospholipids, TFPI and IgG, respectively. After centrifugation at 15 000 × g for 15 min, the supernatants (containing nonadsorbed antibody) were diluted 1 : 10 and assayed in the anti-TFPI ELISA. The effect of various phospholipids and TFPI concentrations in inhibition studies on the antibody binding was investigated by adding TFPI associated with phospholipid vesicles to TFPI-coated wells before purified IgGs (100 µg mL−1) were incubated for 90 min in the anti-TFPI ELISA as detailed above. Phospholipids were used from 20 to 1800 µg mL−1 and TFPI at 200 and 100 µg mL−1. Results were expressed as percentage of binding inhibition. Fresh phospholipid vesicles composed of 25 mol% PS and 75 mol% PC (PSPC) were prepared by standard methods using an assay buffer containing 50 mmol L−1 Tris-HCl, pH 7.4, 150 mmol L−1 NaCl with 5 mmol L−1 CaCl2, and 0.5 mg mL−1 BSA. Recombinant TFPI (150 ng mL−1) was incubated with either PSPC vesicles (20 µmol L−1 ) or assay buffer and either purified IgGs (0.5 mg mL−1 or 0.2 mg mL−1) or assay buffer. After incubation at 37 °C for 30 min, a subsample was drawn from the initial mixture and FXa diluted in assay buffer (100 ng mL−1) was added and further incubated for 10 min at 37 °C. At different time-points, aliquots of 70 µL from the final mixture were transferred to a polystyrene plate and the residual FXa activity was measured by adding 70 µL well−1 of 1.12 mmol L−1 chromogenic substrate S2765. The conversion of the chromogenic substrate was followed at 405 nm, and the residual FXa activity was expressed as a percentage of the control in the absence of TFPI. The χ2 test with Yates correction or Fisher's exact test was used to compare categorical data. Odds ratios (OR) with their 95% confidence intervals (95% CI) were determined as a measure for relative risk. Multivariate logistic regression was performed to adjust for age, sex and the presence of other antibodies. Correlation analysis between antibodies titers were carried out by Spearman test. Values of FXa activity were compared with unpaired t′ test with Welch correction. Significance was defined at P < 0.05. All calculations were made with the SPSS for Windows Release 10.0.1 statistical package (SPSS Inc, Chicago, IL, USA). The specificity of antibody binding in the anti-TFPI ELISA was examined by inhibition experiments using recombinant TFPI in solution. Figure 1(a) illustrates the effect of preincubating an appropriate dilution of two positive serum samples with varying concentrations of fluid-phase TFPI. High concentrations of soluble TFPI produced only partial inhibition of autoantibody binding, suggesting that the autoantibodies preferred surface-bound, rather than fluid-phase TFPI. At the highest concentration (400 µg mL−1) of soluble TFPI tested, only a slight inhibition (<40%) was achieved. On the other hand, adsorption of five anti-TFPI autoantibodies with a TFPI/phospholipid liposome mixture produced on average a 67% (58.8–70.5%) decrease in antibody binding to TFPI in the ELISA. Even low concentrations of soluble TFPI, however, were able to competitively inhibit mAb anti-TFPI binding to TFPI immobilized on the plates. As shown in Fig. 1(b), the inhibition of autoantibody binding varied with the concentration of phospholipids and TFPI. Of 79 normal controls, three (3.8%) were positive for anti-TFPI (one of IgG and two of IgM isotype). Fifty-four of the 162 patients with aPL in this series (33.3%) exhibited anti-TFPI above the cut-off point. There were 24 aPL patients with IgG, 22 with IgM, and eight with both IgG and IgM anti-TFPI. Figure 2 shows the distribution of IgG and IgM anti-TFPI in aPL patients grouped as definite APS and no definite APS, and the normal control group. Anti-TFPI was significantly more prevalent in aPL patients (with or without definite APS) as compared to normal subjects (P < 0.0001). Among the whole aPL group, the prevalence of anti-TFPI of either IgG or IgM isotype in patients belonging to the definite APS or no definite APS group were similar (Table 2). When results were compared regardless of the isotype, the frequency of anti-TFPI appeared to be higher in definite APS than in no definite APS patients although the difference did not reach statistical significance. Approximately 10% of aPL patients had anti-TFPI levels greater than 50 U mL−1. By univariate analysis, the prevalence of these high titer anti-TFPI was significantly greater in the definite APS group compared with the no definite APS group (Tables 2, P= 0.03). After multivariate analysis, this association still remained significant (P = 0.017). As compared to the normal group, only the definite APS but not the no definite APS group showed a significantly greater frequency of high titers of anti-TPFI (P < 0.0001 and P= 0.06, respectively).Table 2Prevalence and statistical analysis of positive results in the ELISA for anti-TFPI in aPL patients with or without definite APS. Data in normal controls are shown for comparisonELISANormal controlsDefinite APSNo definite APSUnivariate analysis OR (95% CI)Multivariate analysis OR (95% CI)n798181Anti-TFPI-IgG1 (1.3)17 (20.9)15 (18.5)1.17 (0.54–2.54)Anti-TFPI-IgM2 (2.5)18 (22.2)12 (14.8)1.64 (0.73–3.68)Anti-TFPI3 (3.8)31 (38.3)23 (28.4)1.56 (0.81–3.02)Anti-TFPI (>50 U mL−1)015 (18.5)5 (6.2)3.45 (1.19–10.0)3.75 (1.27–11.09)Percentages are given in parenthesis. Open table in a new tab Percentages are given in parenthesis. In the whole group of aPL patients, the levels of IgG anti-TFPI correlated poorly with IgG anti-PT titers (rho = 0.17, P= 0.03). Similar correlations were found between IgM anti-TFPI with IgM aCL (Rho = 0.18, p < 0.03), IgM anti-β2GPI (rho = 0.25, P= 0.001), and IgM anti-PT (rho = 0.18, P < 0.03). Anti-β2GPI and aCL of IgG isotype did not correlate with IgG anti-TFPI. All but one of the 17 subjects with SLE who were repeatedly negative for aPL (control group) showed negative results of anti-TFPI. The only patient with a positive result had a level of IgM anti-TFPI of 24 U mL−1. To determine what domain of TFPI is recognized by antihuman TFPI, sera from aPL patients containing autoantibodies of both IgG and IgM isotypes in the ELISA using full-length TFPI (>30 U mL−1) were retested simultaneously on plates coated with full-length TFPI or a truncated form of the molecule, TFPI1-161, lacking Kunitz domain 3 and the carboxy-terminal tail. Most aPL did not bind TFPI1-161. On the other hand, the anti-TFPI mAb showed similar binding with both forms of recombinant TFPI. As shown in Fig. 3, however, three sera from aPL patients did bind to both full-length and truncated human TFPI. Among aPL patients, we examined associations between high levels of anti-TFPI and the main clinical features of the definite APS. As indicated in Table 3, no statistically significant differences in the prevalence of anti-TFPI were observed as regards history of arterial thrombosis or thromboses in specific territories. With respect to venous thrombosis, the frequency of titers >50 U mL−1 of anti-TFPI was significantly higher as compared with patients from the no definite APS group (P = 0.035). However, the association was lost in a multivariate regression model in which all aPL were included. Also the prevalence of anti-TFPI was not statistically different between patients with a single vs. two or more previous episodes of thrombosis (data not shown). In the control group of aPL negative patients with thrombosis, two of 35 exhibited low positive anti-TFPI levels (20 and 22 U mL−1).Table 3Associations between high titers (>50 U mL−1) of anti-TFPI and definite APS-related clinical featuresClinical manifestationnAnti-TFPI (>50 U mL−1)Univariate analysis OR (95% CI)Multivariate analysis OR (95% CI)PositiveNegativeVenous thrombosis459 (20.0)36 (80.0)3.80 (1.19–12.16)nsNo definite APS group815 (6.2)76 (93.8)Arterial thrombosis303 (10.0)27 (90.0)1.69 (0.38–7.55)No definite APS group815 (6.2)76 (93.8)Thrombosis6811 (16.2)57 (83.8)2.93 (0.96–8.92)No definite APS group815 (6.2)76 (93.8)Obstetric complications175 (29.4)12 (70.6)6.25 (1.31–29.94)5.79 (1.13–29.82)No definite APS group*Only women with at least one uncomplicated pregnancy were included. Percentages are given in parenthesis.493 (6.1)46 (93.9)* Only women with at least one uncomplicated pregnancy were included. Percentages are given in parenthesis. Open table in a new tab Within the aPL group of women with previous pregnancies, anti-TFPI were significantly more prevalent in those with obstetric complications than in those without such complications. This was observed not only by univariate (P = 0.02) but also by multivariate (P = 0.03) analysis (Table 3). None of the 31 control women with poor obstetric history who had tested negative for aPL had a positive result in the anti-TFPI assay. Experiments were performed to determine whether antibodies to human TFPI have an effect on the FXa inhibitory activity of TFPI. Initial studies using purified IgG from aPL patients with high titers of anti-TFPI and normal controls in fluid-phase experiments did not demonstrate a significant effect of the antibodies on the dose-dependent inhibition of FXa by TFPI in the absence of anionic phospholipids (data not shown). Because the rate of association of FXa with TFPI is greatly enhanced by the presence of negatively charged phospholipids [18Huang Z.F. Wun T.C. Broze Jr, G.J. Kinetics of factor Xa inhibition by tissue factor pathway inhibitor.J Biol Chem. 1993; 268: 26950-5Abstract Full Text PDF PubMed Google Scholar, 21Willems G.M. Janssen M.P. Salemink I. Wun T.C. Lindhout T. Transient high affinity binding of tissue factor pathway inhibitor-factor Xa complexes to negatively charged phospholipids membranes.Biochemistry. 1998; 37: 3321-8Crossref PubMed Scopus (0) Google Scholar], the experiments were repeated in the presence of PSPC vesicles. Under these conditions (Fig. 4), anti-TFPI(+) IgG clearly reduced TFPI-mediated FXa inhibition. The primary focus of the present study was to investigate whether antibodies directed against TFPI can be found in patients bearing autoimmune aPL. There have been previous and conflicting reports in the literature on TFPI antigen levels in patients with aPL [11Roubey R.A.S. Tissue factor pathway and the antiphospholipid syndrome.J Autoimmun. 2000; 15: 217-20Crossref PubMed Scopus (53) Google Scholar, 12Amengual O. Atsumi T. Khamashta M.A. Hughes G.R.V. The role of the tissue factor pathway in the hypercoagulable state in patients with the antiphospholipid syndrome.Thromb Haemost. 1998; 79: 276-81Crossref PubMed Scopus (263) Google Scholar]. However, most studies demonstrated that TFPI activity is impaired in the APS [13Adams M.J. Donohoe S. Mackie I.J. Machin S.J. Anti-tissue factor pathway inhibitor activity in patients with primary antiphospholipid syndrome.Br J Haematol. 2001; 114: 375-9Crossref PubMed Scopus (32) Google Scholar, 22Adams M.J. Oostryck R. Further investigations of lupus anticoagulant interference in a functional assay for tissue factor pathway inhibitor.Thromb Res. 1997; 87: 245-9Abstract Full Text Full Text PDF PubMed Scopus (7) Google Scholar, 23Jacobsen E.M. Sandset P.M. Wisloff F. Do antiphospholipid antibodies interfere with tissue factor pathway inhibitor?.Thromb Res. 1999; 94: 213-20Abstract Full Text Full Text PDF PubMed Scopus (16) Google Scholar]. Recently, Salemink et al. [14Salemink I. Blezer R. Willems G.M. Galli M. Bevers E. Lindhout T. Antibodies to β2-glycoprotein I associated with antiphospholipid syndrome suppress the inhibitory activity of tissue factor pathway inhibitor.Thromb Haemost. 2000; 84: 653-6Crossref PubMed Google Scholar] described enhanced FXa generation in six patients with aPL (five with and one without APS) and provided further evidence that the anti-TFPI like activity was due to the presence of β2GPI-dependent aPL. True anti-TFPI antibodies in APS have been the topic of a preliminary report [15Cakir B. Arnett F.C. Roubey R.A.S. Autoantibodies to tissue factor pathway inhibitor (TFPI) are associated with arterial thrombosis/stroke.J Autoimmun. 2000; 15: A11Google Scholar], but to our knowledge this work is the first full description of the presence of anti-TFPI in a large series of patients with aPL. We were able to optimize an ELISA for detecting autoantibodies to TFPI. Antibody specificity was ascertained by competition using phospholipid-bound TFPI and inhibition of TFPI functional activity as soluble TFPI competed poorly. Anti-TFPI closely resemble anti-β2GPI and anti-PT in that they preferentially bind their target protein after immobilization on anionic phospholipids or irradiated microtiter plates, in contrast to weak binding to fluid-phase protein [24Forastiero R.R. Martinuzzo M.E. Carreras L.O. Binding properties of antibodies to prothrombin and β2-glycoprotein I assayed by ELISA and Dot blot.Clin Exp Immunol. 1999; 118: 480-6Crossref PubMed Scopus (20) Google Scholar, 25Tincani A. Spatola L. Prati E. Allegri F. Ferremi P. Cattaneo R. Meroni P. Balestrieri G. The anti-β2 glycoprotein I activity in human anti-phospholipid syndrome is due to monoreactive low-affinity autoantibodies directed to epitopes located on native β2 glycoprotein I and preserved during species' evolution.J Immunol. 1996; 157: 5732-8PubMed Google Scholar, 26Arvieux J. Darnige L. Caron C. Reber G. Bensa J.C. Colomb M.G. Development of an ELISA for autoantibodies to prothrombin showing their prevalence in patients with lupus anticoagulants.Thromb Haemost. 1995; 74: 1120-5Crossref PubMed Google Scholar]. Most antibodies to TFPI react with the full-length TFPI molecule but not with the truncated form TFPI1-161 lacking the K3 domain and the positively charged carboxy-terminus region. The epitope could be located in the region involved in the interaction of TFPI with cell surfaces [16Broze Jr, G.J. Tissue factor pathway inhibitor.Thromb Haemost. 1995; 74: 90-3Crossref PubMed Scopus (307) Google Scholar, 17Bajaj M.S. Birktoft J.J. Steer S.A. Bajaj S.P. Structure and biology of tissue factor pathway inhibitor.Thromb Haemost. 2001; 86: 959-72Crossref PubMed Google Scholar, 27Kato H. Regulation of functions of vascular wall cells by tissue factor pathway inhibitor.Arterioscler Thromb Vasc Biol. 2002; 22: 539-48Crossref PubMed Scopus (0) Google Scholar]. However, the recognition of epitopes expressed on K1 or K2 domains cannot be excluded as conformational changes in these domains of TFPI1-161 could account for differences in anti-TFPI binding. Previous reports have shown that anionic phospholipids enhance the rate of association of FXa with TFPI [18Huang Z.F. Wun T.C. Broze Jr, G.J. Kinetics of factor Xa inhibition by tissue factor pathway inhibitor.J Biol Chem. 1993; 268: 26950-5Abstract Full Text PDF PubMed Google Scholar] and are essential for the anticoagulant activity of the FXa/TFPI complex [21Willems G.M. Janssen M.P. Salemink I. Wun T.C. Lindhout T. Transient high affinity binding of tissue factor pathway inhibitor-factor Xa complexes to negatively charged phospholipids membranes.Biochemistry. 1998; 37: 3321-8Crossref PubMed Scopus (0) Google Scholar, 28Kazama Y. The importance of the binding of factor Xa to phospholipids in the inhibitory mechanism of tissue factor pathway inhibitor. the transmembrane and cytoplasmic domains of tissue factor are not essential for the inhibitory reaction of tissue factor pathway inhibitor.Thromb Haemost. 1997; 77: 492-7Crossref PubMed Scopus (0) Google Scholar]. In a system using purified components, anti-TFPI IgGs were found to limit FXa inhibition by TFPI in the presence, but not the absence of anionic phospholipids. These findings suggest that the interaction of the antibodies with phospholipid-associated TFPI impairs the inhibition of FXa. In a recent work, it was demonstrated that anti-β2GPI suppress the inhibitory activity of TFPI because the anti-β2GPI/β2GPI complex competes with the TFPI/FXa complex for the same phospholipid binding sites [14Salemink I. Blezer R. Willems G.M. Galli M. Bevers E. Lindhout T. Antibodies to β2-glycoprotein I associated with antiphospholipid syndrome suppress the inhibitory activity of tissue factor pathway inhibitor.Thromb Haemost. 2000; 84: 653-6Crossref PubMed Google Scholar]. Therefore, both true anti-TFPI and anti-β2GPI may cooperate to interfere with TFPI activity. This hypothesis is consistent with functional studies showing a low TFPI activity in plasma samples from some APS patients [13Adams M.J. Donohoe S. Mackie I.J. Machin S.J. Anti-tissue factor pathway inhibitor activity in patients with primary antiphospholipid syndrome.Br J Haematol. 2001; 114: 375-9Crossref PubMed Scopus (32) Google Scholar]. A high prevalence of anti-TFPI of both IgG and IgM isotypes was found in our series of 162 patients with well-characterized aPL. The frequency of positive results was similar in aPL patients with or without definite APS-related clinical features. Definite APS patients, however, had a significantly greater prevalence of high titers of anti-TFPI than patients with aPL but not fulfilling current criteria for definite APS. Patients with aPL and anti-TFPI titers exceeding 50 U mL−1 had a 3.4- to 3.7-fold increased risk of definite APS-associated clinical complications. These findings suggest the coexistence of anti-TFPI and aPL may contribute to the pathogenesis of the APS. Women with aPL and high titers of anti-TFPI had an approximately sixfold increased risk of obstetric complications compared to aPL women with negative or less than 50 U mL−1 titers of anti-TFPI. In contrast to previous preliminary reports [13Adams M.J. Donohoe S. Mackie I.J. Machin S.J. Anti-tissue factor pathway inhibitor activity in patients with primary antiphospholipid syndrome.Br J Haematol. 2001; 114: 375-9Crossref PubMed Scopus (32) Google Scholar, 15Cakir B. Arnett F.C. Roubey R.A.S. Autoantibodies to tissue factor pathway inhibitor (TFPI) are associated with arterial thrombosis/stroke.J Autoimmun. 2000; 15: A11Google Scholar], an association between high titer anti-TFPI and venous or arterial thrombosis was not detected by multivariate analysis in this study. These subgroups contained relatively few patients, however, and further studies involving larger numbers of patients will be required to establish the clinical relevance of elevated anti-TFPI levels. In conclusion, our data show that autoantibodies to TFPI are commonly found in patients with aPL. Despite the limitations of a retrospective study, the presence of high titers of anti-TFPI seems to identify patients with an increased risk of the clinical complications of definite APS. The majority of anti-TFPI autoantibodies appear to bind to TFPI in the presence of anionic phospholipids and thus decrease the in vitro TFPI-dependent inhibition of FXa. This disturbance in TFPI function could conceivably contribute to the development of the clinical features associated with APS. This work was partly supported by a grant from the National Fund of Science and Technology (PICT 2000–01 N°05–08160), Ministry of Culture and Education, Argentina. The present study is within the Research Project CID-002–99 from the Favaloro University. Additional support was provided by Public Health Service grant R01-HL34462.