Antibodies to prothrombin in antiphospholipid syndrome and inflammatory disorders

Authors


Dr JohnJackson Department of Immunology, St James's Hospital, Dublin 8, Ireland.

Abstract

Antiphospholipid antibodies associated with the antiphospholipid syndrome (APS) have been shown to bind plasma proteins, particularly beta 2-glycoprotein I (β2-GPI). In this study the incidence of antibodies to solid-phase prothrombin was examined in patients with antiphospholipid syndrome and a variety of other inflammatory disorders.

Significantly elevated levels of IgG anti-prothrombin (anti-PT) antibodies were detected in 63% of patients with APS (n = 27, median 22 arbitrary units: AU), 33% with SLE (n = 92, median 14 AU), 45% with rheumatoid factor (n = 22, median 16 AU), 21% with carotid artery stenosis (n = 21, median 15 AU), 32% with stroke (n = 38, median 13 AU), 67% of patients with a false positive serology for syphilis (n = 21, median 24 AU), 37% with HIV (n = 30, median 14 AU), 29% with syphilis (n = 14, median 19 AU) and 3% with infectious mononucleosis (n = 30, median 9 AU). In addition, a group of lupus anticoagulant (LA) positive patients (n = 48) was examined for antibodies to prothrombin, β2-GPI and cardiolipin. 10 (21%) patients had raised levels of IgG anti-PT antibodies, 30 (62%) had significantly elevated levels of anti-β2-GPI antibodies and 15 (31%) had elevated levels of anticardiolipin antibodies (ACA). Of the LA-positive patients, 15 (43%) were identified with definite APS, eight (23%) with probable APS, two (6%) with possible APS and 10 (28%) patients had no clinical evidence of APS.

In conclusion, antibodies to prothrombin were found in a variety of inflammatory disorders and were therefore not specific for the APS. However, identification of the plasma proteins recognized by antibodies from patients with APS may provide insight into the pathogenic mechanisms involved in the heterogenous clinical manifestations of the APS.

The antiphospholipid syndrome (APS), characterized by recurrent venous and/or arterial thrombosis, recurrent fetal loss and thrombocytopenia, is associated with persistently high levels of antiphospholipid antibodies ( Hughes et al, 1986 ; McNeil et al, 1991 ). Antiphospholipid antibodies are identified in serum by their ability to bind phospholipids, most commonly cardiolipin, and in plasma by their ability to interfere with phospholipid-dependent coagulation reactions. However, elevated levels of antiphospholipid antibodies are also found in a variety of infectious conditions, notably human immunodeficiency virus, syphilis, infectious mononucleosis, endocarditis and tuberculosis (reviewed by Loziou et al, 1996 ). In addition, they are found in malignancy and following certain drug therapies and therefore are not specific for APS ( Asherson, 1988).

Recent evidence suggests that these antibodies detected in conventional lupus anticoagulant (LA) and anticardiolipin antibody (ACA) assays are not solely directed at anionic phospholipids but also react with phospholipid-binding plasma proteins, in particular beta 2-glycoprotein I (β2-GPI) and prothrombin, either alone or complexed with phospholipid ( McNeil et al, 1990 ; Galli et al, 1990 ; Bevers et al, 1991 ; Keeling et al, 1993 ; Permpikul et al, 1994 ). Furthermore, antibodies with lupus anticoagulant activity have been reported to interfere with activated protein C, protein S and thrombomodulin anticoagulant function ( Oosting et al, 1993a , b). Measurement of antibodies to some of these plasma proteins may provide a more specific basis for the laboratory diagnosis of APS. Furthermore, identification of the physiologically relevant targets of antiphospholipid antibodies may offer insight into the pathogenic mechanism(s) involved in the thrombotic manifestations of APS.

We and others have shown that the presence of antibodies to β2-GPI is significantly associated with the clinical manifestations of APS, in particular thrombosis ( McNally et al, 1995 ; Balestrieri et al, 1996; Pengo et al, 1996 ; Guerin et al, 1997 ). Additionally, we have shown that elevated levels of antibodies to β2-GPI are highly specific for APS and are not found in infectious conditions, unlike LA and ACA ( Guerin et al, 1997 ).

In this study the incidence of antibodies to solid-phase prothrombin was estimated in patients with a variety of disorders and assessment of their usefulness in the laboratory diagnosis of APS was determined. Additionally, the prevalence of antibodies to prothrombin, along with their sensitivity and specificity for APS, was compared with antibodies to β2-GPI and cardiolipin. Furthermore, patients with LA activity, in the presence or absence of APS, were examined for antibody reactivity to prothrombin, β2-GPI and cardiolipin.

PATIENTS AND METHODS

Subjects

Sera from the following patient groups were examined for antibodies to prothrombin; antiphospholipid syndrome (n= 27), systemic lupus erythematosus (n= 92), rheumatoid factor positive (n= 22), stroke (n= 38), carotid artery stenosis (n= 21), syphilis (n= 14), VDRL false positive serology for syphilis (n= 21), infectious mononucleosis (n= 30), HIV (n= 30) and normal controls (n= 26). Patients with APS and SLE were defined according to established criteria ( Alarcón-Segovia & Sanchez-Guerrero, 1989; Tan et al, 1982 ). In the APS group, 74% of patients had primary APS. From previous studies the prevalence of antibodies to β2-GPI and ACA in these APS patients was found to be 78% and 96%, respectively ( Guerin et al, 1997 ). In the SLE patient group 14% were previously determined to have anti-β2-GPI antibodies and 30% had ACA.

In addition, a separate group of patients with LA activity (n= 48) was also examined for antibodies to prothrombin, β2-GPI and cardiolipin. Lupus anticoagulant activity was defined according to established criteria using the activated partial thromboplastin time (APPT), kaolin clotting time (KCT) and dilute Russell's viper venom time (DRVVT) (Brandt et al, 1995). All patients were DRVVT positive and 27 subjects were positive for both DRVVT and KCT. The medical histories from 35 of these patients were retrieved and independently reviewed for evidence of APS according to the criteria for classification of APS defined by Alarcón-Segovia & Sanchez-Guerrero (1989). The remaining 13 patient charts could not be examined because these individuals attended other hospitals.

Anti-prothrombin (anti-PT) antibody ELISA

Human prothrombin (Sigma Chemical Co., U.S.A.) was coated onto a Nunc maxisorp microtitre plate (Roskilde, Denmark) overnight at 4°C, at a concentration of 5 μg/well in bicarbonate buffer (12 m M disodium carbonate, 35 m M sodium hydrogen carbonate, pH 9.6). The prothrombin preparation contained < 0.01 units of factor X activity and was > 95% pure as judged by SDS-PAGE (data not shown). A minor contaminant was present in the prothrombin preparation which had an apparent molecular weight of 55 kD corresponding to factor X. The plate was washed four times in PBS Tween (100 m M sodium phosphate, 150 m M sodium chloride, pH 7.2, 0.05% Tween 20) and 100 μl of test serum (in triplicate) at a dilution of 1/25 in PBS Tween was added and incubated for 1 h at 37°C. The plate was again washed and 100 μl of peroxidase-conjugated rabbit anti-human IgG (Dakopatts, Denmark) diluted 1/1000 in PBS Tween was added and incubated for 1 h at 37°C. The plate was washed as before and 100 μl of o-phenylenediamine (Dakopatts, Denmark) in 0.1 M citrate phosphate buffer pH 5.0 containing 0.012% v/v hydrogen peroxide (BDH Ltd, Poole) was added. The reaction was developed for 3 min at room temperature and terminated by the addition of 100 μl of 2.5 M sulphuric acid (BDH Ltd, Poole). The absorbance readings were measured at 492 nm on a microtitre plate reader (Titretek Multiscan plus MKII, Flow Laboratories Int., Switzerland). A standard curve was established using serial dilutions from 1/25–1/800 of a known positive serum to which arbitrary units (AU) from 64–2 were assigned. The antibody levels in the test samples were automatically calculated from the standard curve by the MultiCalc Immunoassay Data Management program® (Wallac Oy, Finland). Having determined that the control group (n= 26) was normally distributed using Datadesk® (Odesta Corporation, Illinois, U.S.A.), the reference range was calculated from the mean ±2 standard deviations (SD) of a normal control group. A reagent blank, PBS Tween (mean OD 0.06) and quality control serum (mean value 21 AU) were included on each plate. The intra- and interassay variation determined for the anti-PT assay was 7.3% and 8.5% respectively.

To examine plasma from LA-positive individuals for the incidence of anti-PT antibodies, the anti-PT assay was also standardized and a reference range was established for the level of anti-PT antibodies in plasma.

Inhibition studies

The specificity of antibodies to prothrombin in APS was examined by absorption of antibodies with prothrombin and ovalbumin, an irrelevant control antigen. In addition, to determine whether antibodies recognized a conformational determinant on prothrombin, plasma was absorbed with both solid and fluid phase prothrombin. To measure fluid-phase inhibition, 100 μl of plasma from APS patients (n= 8) diluted 1/25 in PBS Tween was incubated with 100 μl prothrombin or ovalbumin, at a concentration of 40 μg/ml in PBS for 2 h at 37°C and overnight at 4°C. To measure solid-phase inhibition prothrombin and ovalbumin, at a concentration of 40 μg/ml, were coated overnight onto a Nunc microtitre plate as described. The plate was washed in PBS Tween and 100 μl of diluted plasma (n= 8) was added and incubated for 2 h at 37°C and overnight at 4°C. The plasma was recovered from both solid and fluid phase absorption experiments and retested on the anti-PT ELISA as described.

Anti-β2-GPI ELISA

Antibodies to β2-GPI were measured using an anti-β2-GPI antibody ELISA as described ( Guerin et al, 1997 ). This method was slightly modified to facilitate measurement of antibodies to β2-GPI in plasma. A new standard curve was established using a plasma sample from an individual with antibodies to β2-GPI. A reference range was calculated from the mean ±2 SD of the plasma of a normal control group (n= 26).

Anticardiolipin ELISA

Anticardiolipin antibodies were assayed in the LA-positive group using a commercial assay for anticardiolipin antibodies (DIASTAT, Shield Diagnostics Ltd) according to the manufacturer's instructions.

Statistics

The Fisher's exact test was employed to evaluate an association between anti-PT antibodies and thrombosis and the combined clinical features of APS. Assuming a gaussian distribution, Pearson's linear correlation was used to compare the anti-PT, anti-β2-GPI and anticardiolipin antibody assays. All statistical analysis was performed by GraphPad Instat 2.03 (San Diego, Calif., U.S.A.).

RESULTS

Anti-prothrombin antibody levels in the different patient groups

Fig 1 illustrates the level of IgG antibodies to prothrombin in the various patient groups. A level of 21 AU was determined as the positive cut-off value from the mean of 11.3 + 2 SD of the normal control group. Significantly elevated levels of IgG anti-PT antibodies were detected in 63% of patients with APS (median value 22 AU), 33% with SLE (median value 14 AU), 45% with rheumatoid factor (median value 16 AU), 21% with carotid artery stenosis (median value 15 AU), 32% with stroke (median value 13 AU), 3% with infectious mononucleosis (median value 9 AU), 37% with HIV (median value 14 AU), 29% with syphilis (median value 19 AU) and 67% of patients with a false positive serology for syphilis (median value 24 AU).

Figure 1.

AU as the positive cut-off value designated by the horizontal line. These subjects consisted of normal controls, rheumatoid factor positive (RF), systemic lupus erythematosus (SLE), antiphospholipid syndrome (APS), syphilitic (VDRL+, TPHA+), VDRL false positive serology for syphilis (VDRL+, TPHA−), infectious mononucleosis (IM), human immunodeficiency virus infection (HIV), carotid artery stenosis (CAS) and stroke patients.

No association was observed between the presence of elevated levels of anti-PT antibodies and thrombosis in patients with APS according to the Fisher's exact test (P= 0.1320). However, an association between the presence of anti-PT antibodies and all the clinical manifestations of APS (including fetal loss and thrombocytopenia) was observed for this patient group using the Fisher's exact test (P= 0.0043).

Inhibition studies

To ensure that the prothrombin antibodies were specific to prothrombin and not to the trace contaminant factor X, a selection of eight plasmas with anti-PT antibodies were preabsorbed with prothrombin and factor X (Sigma Chemical Co., U.S.A.). None of these plasma samples could be absorbed with factor X and all were significantly absorbed by prothrombin.

The nature of the epitope on prothrombin recognized by anti-PT antibodies was investigated by absorption with prothrombin in both fluid and solid phase (Fig 2). Antibody reactivity in plasma from eight patients with APS was specifically absorbed with solid-phase prothrombin (mean percent inhibition of 42%), whereas antibody reactivity to the control antigen was unchanged. However, no decrease in antibody reactivity was noted in five of eight patients following incubation with fluid-phase prothrombin (Fig 2). In the remaining three patients antibody reactivity was reduced by approximately 35% using fluid-phase prothrombin.

Figure 2.

Fig 2. Antibody binding to solid-phase prothrombin following preincubation of plasma from APS patients (n= 8) with both solid- and fluid-phase prothrombin and ovalbumin (irrelevant control antigen) at a concentration of 40 μg/ml. In the fluid-phase absorption studies anti-PT antibodies in five APS patients were not absorbed (n= 5), whereas the antibody levels in the remaining three patients were significantly reduced (n= 3). The mean absorbance units (AU) + SEM are represented.

The prevalence of IgG anti-PT, anti-β2-GPI and ACA in LA-positive patients

In the group of LA-positive patients examined, 21% had raised levels of IgG anti-PT antibodies, 62% had significantly elevated levels of anti-β2-GPI antibodies and 31% had elevated levels of ACA (Fig 3, Table I). Interestingly, of the 10 patients with anti-PT antibodies, eight had co-existing anti-β2-GPI antibodies. One patient exhibited all three antibody specificities. 11 patients (23%) had both anti-β2-GPI antibodies and ACA, but not anti-PT antibodies. Of the 48 patients, 15 (31%) had neither anti-PT antibodies, anti-β2-GPI antibodies nor ACA.

Figure 3.

0 GPL units/ml and are indicated by horizontal lines. Anti-PT and anti-β2-GPI antibodies are measured in AU and IgG ACA (ACLAG) in GPL units/ml.

Table 1. Table I. The prevalence of IgG antibodies to prothrombin, β2-GPI and cardiolipin in LA-positive patients. Thumbnail image of

No association was observed between the presence of anti-PT and anti-β2-GPI antibodies (r = 0.047, 95% CI −0.32–0.24) nor with anti-PT and ACA (r = 0.025, 95% CI −0.3–0.26) using Pearsons linear correlation.

Lupus anticoagulant positive patients: clinical details

The medical histories were available for 35 of the 48 LA-positive patients examined. The medical charts of these patients were independently reviewed for clinical and serological evidence of APS according to the criteria for classification of APS outlined by Alarcón-Segovia & Sanchez-Guerrero (1989). These were modified to include either LA and/or ACA on two separate occasions. Using these criteria, 15 patients were classified as definite APS (43%), eight as probable (23%), two as possible APS (6%) and 10 patients had no clinical features of the APS (28%). 2 Table II represents the prevalence of antibodies to prothrombin, β2-GPI and cardiolipin in these individuals.

Table 2. Table II. The incidence of anti-prothrombin, β2-GPI and cardiolipin antibodies in APS patients with LA activity. The one patient with anti-β2-GPI antibodies, but without clinical evidence of APS, had an antibody level of 21 AU. Thumbnail image of

The sensitivity and specificity of anti-PT, anti-β2GPI and anticardiolipin antibodies for APS

In all patient groups examined, the specificity of anti-PT, anti-β2-GPI and anticardiolipin antibodies for definite or probable APS was 68%, 99% and 85% and the sensitivity of each assay for definite or probable APS was 61%, 86% and 87% respectively.

DISCUSSION

For a number of years ACA and LA assays have been used for the laboratory diagnosis of APS; however, it is now widely accepted that these assays are not specific for APS. Recently, the antigenic targets of conventional ACA and LA have been more clearly defined as phospholipid-binding plasma proteins, namely β2-GPI and prothrombin, either alone or complexed with phospholipids ( McNeil et al, 1990 ; Galli et al, 1990 ; Bevers et al, 1991 ; Permpikul et al, 1994 ). Furthermore, a number of studies have demonstrated that antibodies to β2-GPI are highly specific for APS and are significantly associated with the clinical manifestations of APS, in particular thrombosis ( McNally et al, 1995 ; Balistrieri et al, 1995 ; Guerin et al, 1997 ). The value of measuring anti-PT antibodies as part of the laboratory diagnosis of APS has not been clearly defined and, although some investigators have reported the presence of anti-PT antibodies in patients with APS, an association between their presence and thrombosis in these patients was not observed ( Pengo et al, 1996 ; Galli et al, 1997 ; Arvieux et al, 1995 ).

In this study and as reported by other investigators, antibody recognition of prothrombin occurred only when prothrombin was absorbed onto a high-affinity binding solid support, notably γ irradiated or Nunc maxisorp but not plain polystyrene plates (data not shown). Interestingly, the behaviour of anti-PT antibodies in ELISA resembles that already observed for anti-β2-GPI antibodies, whereby antibodies can only recognize β2-GPI when it is immobilized onto certain solid supports ( Matsuura et al, 1994 ; Roubey et al, 1995 ; Arvieux et al, 1995 ; Rao et al, 1995 ; Galli et al, 1997 ).

The nature of the epitope recognized by anti-prothrombin antibodies was investigated by inhibition studies. In all the plasmas tested, anti-prothrombin antibodies were specifically absorbed with solid-phase prothrombin at a concentration of 40 μg/ml, whereas no change in antibody reactivity was noted to ovalbumin, an irrelevant control antigen. Furthermore, in five of eight APS patients, antibodies did not recognize prothrombin in fluid phase at a similar concentration. However, three APS patients did bind to fluid-phase prothrombin. These findings suggest that there is a mixture of epitopes recognized by anti-PT antibodies, some of which appear conformationally determined. Other investigators have also proposed that anti-PT antibodies may be directed at separate epitopes on the prothrombin molecule ( Rao et al, 1996 ).

The prevalence of IgG anti-PT antibodies in patients classified as APS according to the classification criteria of Alarcón-Segovia & Sanchez-Guerrero (1989) was 63%. However, an association between the presence anti-PT antibodies and thrombosis in these patients with APS was not observed. Similar to this study, other investigators have also failed to demonstrate an association between the presence of anti-PT antibodies and thrombosis in APS ( Pengo et al, 1996 ; Swadzba et al, 1997 ; Forastiero et al, 1997 ). Pengo et al (1996 ) reported an incidence of 50% for anti-PT antibodies in a group of patients with antiphospholipid antibodies and at least one clinical manifestation of APS. Another group reported a prevalence of 58% for IgG and IgM antibodies to prothrombin in patients with APA; however, only 37% of this patient cohort had clinical manifestations of APS, the remaining patients had separate clinical features ( Galli et al, 1997 ).

In this study, however, when all of the clinical features of APS were examined, an association between the presence of antibodies to prothrombin and APS was observed. This may imply that antibodies to prothrombin in APS are more strongly associated with clinical manifestations such as fetal loss or thrombocytopenia. However, the number of patients in this study is too small to allow this question to be definitely addressed. In contrast to antibodies to prothrombin, anti-β2-GPI antibodies specifically identify all patients with definite APS and are significantly associated with a history of thrombosis in these patients ( Guerin et al, 1997 ).

The prevalence of anti-PT antibodies in SLE was 33% which is similar to the reported incidence of LA and ACA ( Love & Santoro, 1990). Of the SLE patients reviewed (n= 20) for clinical evidence of APS, only one was identified with probable APS. This patient did not have raised levels of anti-PT antibodies. Conversely 25% of this SLE group had elevated levels of antibodies to prothrombin without clinical evidence of APS.

This study also examined the incidence of anti-PT antibodies in a variety of conditions specifically chosen because they result in a high rate of ‘false positives’ in conventional ACA and/or LA. Raised levels of anti-PT antibodies were observed in 67% of patients with a false positive serology for syphilis, 37% in HIV, 29% of syphilitics and 3% of patients with infectious mononucleosis. Elevated levels of anti-PT antibodies were also observed in 21% of patients with carotid artery stenosis and 32% of patients who had experienced stroke and 45% of rheumatoid factor positive patients. Moreover, the level of antibodies to prothrombin was similar in patients with APS and SLE to that in HIV, stroke, carotid artery stenosis and VDRL false positive serology for syphilis. Therefore the presence of anti-PT antibodies was not specific for APS, as high levels of these antibodies were seen in both thrombotic and non-thrombotic conditions. The non-specificity of the anti-PT assay for APS reflects that already seen with the ACA assay.

In addition to the above groups this study also separately examined patients with LA activity for the incidence of antibodies directed at prothrombin, β2-GPI and cardiolipin (Fig 3). 21% of LA-positive patients had elevated levels of IgG anti-PT antibodies. Furthermore, 63% of patients with LA activity had antibodies directed at β2-GPI and 31% had ACA. Antibodies to β2-GPI were more strongly associated with LA activity in these patients. Although 8/10 patients with anti-PT antibodies also had raised levels of anti-β2-GPI antibodies in their plasma, an association between the presence of anti-PT antibodies and anti-β2-GPI antibodies was not observed. Additionally, no association between the presence of anti-PT antibodies and ACA was found. Of a total of 48 patients with LA activity, 15 had neither anti-PT nor anti-β2-GPI antibodies and these patients were also negative for ACA. These findings were supported by Arvieux et al (1995 ) who reported a prevalence of 30% for both IgG and IgM antibodies to prothrombin in a group of LA-positive patients with APS, SLE and SLE-like disease; in addition, 62% of these patients had anti-β2-GPI antibodies. Furthermore, Forastiero et al (1997 ) reported an incidence of 25% for IgG anti-prothrombin antibodies in LA-positive patients. Lupus anticoagulant activity in the remaining patients may be explained by different antigenic targets, for example activated protein C and protein S ( Oosting et al, 1993a ). However, it may be concluded from these results that the plasma proteins, prothrombin and β2-GPI represent the antigenic targets in 67% of LA-positive patients. These findings further support the evidence that patients with LA activity have antibodies not only to phospholipids but also against plasma proteins, in particular prothrombin and β2-GPI.

The medical histories of these LA-positive patients, where available, were reviewed for clinical and serological evidence of APS according to the criteria for classification of APS outlined by Alarcón-Segovia & Sanchez-Guerrero (1989) (modified to include LA and/or ACA). Using these criteria, 15 patients were identified with definite APS, eight with probable APS, two with possible APS, and 10 patients had no clinical evidence of APS. Of the patients categorized as definite and probable APS, 21% had anti-PT antibodies, 83% had antibodies to β2-GPI and 56% had ACA. Interestingly, the majority of patients with APS (definite and probable) had antibodies to β2-GPI, whereas antibodies to prothrombin and cardiolipin occurred at much lower frequencies. In addition, 28% of patients without any clinical features of APS were LA positive, again highlighting the relative non-specificity of LA for the diagnosis of APS.

This study also highlights the problems with the defined classification criteria for APS. Six patients were classified as either definite (n= 1), probable (n= 3) or possible (n= 2) on the basis of these criteria; however, many of these clinical features could be explained on the basis of other prothrombotic pathologies.

In summary, using a wide range of patient groups this study clearly demonstrated that antibodies to prothrombin were not specific for APS and were found in a variety of disorders. In addition anti-β2-GPI antibodies were more strongly associated with features of APS than either LA, anti-PT or anticardiolipin antibodies. Furthermore, anti-PT antibodies do not appear to be associated with thrombosis in APS and may be related to other manifestations such as thrombocytopenia or fetal loss.

Although the presence of antibodies to prothrombin is not specific for APS, identification of the antigenic targets of antiphospholipid antibodies may provide insight into the pathogenic mechanism(s) involved in the spectrum of these clinical manifestations of APS.

Acknowledgements

The authors are grateful to the Health Research Board, Ireland, for funding this work.

Ancillary