Introduction

Lupus anticoagulants (LA) belong to a group of antiphospholipid autoantibodies and are commonly detected through their ability to prolong phospholipid dependent coagulation tests. Antiphospholipid antibodies bind to complexes of negatively charged phospholipids and plasma proteins. LA are common in patients with an antiphospholipid syndrome (APS), and are associated with thromboembolic disease (TE), recurrent spontaneous abortion, and thrombocytopenia [1]. The precise mechanisms for the development of clinical disease are not clear yet. Interestingly, LA are also detected in 1 to 5% of healthy individuals [2]. It is therefore important to differentiate between individuals with LA who are at risk for TE and those who are not. Increased levels of anti-β2-GPI IgG were associated with thrombosis in patients postive for LA [3].

Whether platelets play a role for the development of TE in patients with LA is not clear. There is evidence that LA bind to the phospholipid/plasma protein complex on activated platelet membranes , and thus, influence platelet activation and aggregation [4–11]. Further, monoclonal anti-β2-glycoprotein (GP) I antibodies with LA activity have been shown to bind to platelet membranes in the presence of β2-GP I and lead to platelet activation in the presence of other agonists [12]. Increased levels of anti-β2-GPI IgG were associated with thrombosis in patients positive for LA [3]. Further, findings of increased platelet PAC-1 binding and of CD63 expression [7], strongly suggest that platelets are poised for coagulation in LA patients.

The extent to which platelets respond to various agonists to become activated varies from one individual to the next. It has been shown that some of this variability is associated with inherited polymorphisms of the involved GP complexes Ib/IX/V, Ia/IIa, and IIb/IIIa [13]. Accordingly, some of these inherited polymorphisms have been associated with an increased risk for TE [13, 14].

The platelet GPIbα is the major platelet receptor mediating shear stress induced adhesion and activation. Thus, genetic polymorphisms within this receptor are regarded key candidates to influence its function. These are the variable number of tandem repeats (VNTR), which result from the presence of one to four tandem repeats of a 13-amino acid sequence in the macroglycopeptide region (VNTR A, B, C, or D from largest to smallest) [15], and the Kozak dimorphism, caused by a single nucleotide substitution [(−5)TC] in the Kozak sequence in the GPIbα chain [16]. Whether or not these genetic polymorphisms are associated with an increased risk for TE is controversial [13]. We have shown that the latter polymorphism affects platelet plug formation in healthy individuals under high shear stress rates [17]. We, therefore, were interested to compare allelic frequencies from healthy controls with individuals with LA with a history of TE and those without TE. Assuming that in LA patients, like in healthy individuals, these polymorphisms will influence platelet function, and thus, contribute to thrombotic disease, we evaluated their association with platelet plug formation under high shear stress rates.

Patients

One hundred patients with persistent LA antibodies were included in our study from February 2001 to November 2003. Their clinical characteristics as well as hereditary and acquired risk factors for TE are shown in Table 1. Sixty-seven patients (54 women) had a history of thromboembolic events (group LA/TE+). Their median age at diagnosis of LA was 35.5 years (range, 11.4–77.4 years) and the median age at study entry was 39.7 years (17.1–78.3 years). Thirty-three patients (24 women) without a previous thrombosis (group LA/TE−) were median 53.2 years old (range, 13.4–81.2 years) at diagnosis of LA and median 57.8 years (range, 19.8–86.1 years) at inclusion. In these latter patients, LA was discovered at a routine preoperative coagulation test or with the evaluation of systemic lupus erythematosus (SLE). Fourteen patients (10 LA/TE+ and four LA/TE−) fulfilled the criteria for SLE [18]. The period since the detection of LA until inclusion for this study was similar in both groups (LA/TE− median, 2.4 years; range, 0.8–35.6 years; LA/TE+ median, 3.0 years; range, 0.8–26.3 years; p > 0.05). At the time of the study, 51 LA/TE+ patients and one of the LA/TE− patients received oral anticoagulation by vitamin K antagonists. Four LA/TE+ patients and three LA/TE− patients were under steroid therapy. Two LA/TE+ patients received immunosuppressive therapy.

Table 1 Clinical data of LA/TE+ patients (n = 67) and LA/TE− patients (n = 33)
Full size table

None of the patients had taken any drugs known to affect platelet function for the last 2 weeks before blood sampling. Two hundred thirty-three healthy individuals (99 women; median age 33 years; range, 27–42 years), who were investigated during the same time period as the LA patients served as controls [17]. Informed consent was obtained from all individuals before entry into the study, which had been approved by the Ethics Committee of the Medical University of Vienna.

Materials and methods

LA testing was performed within 3 h of blood sampling. LA was diagnosed according to the criteria proposed by the Subcommittee on Lupus Anticoagulant/Antiphospholipid Antibody of the Scientific and Standardisation Committee of the International Society on Thrombosis and Haemostasis [19], by using two different screening assays, activated partial thromboplastin time (PTT LA-reagent, Diagnostica Stago, Asniere sur Seine, France) and diluted Russel’s viper venom time (DVVT, Gradipore, North Ryde, Australia), mixing tests and confirmatory tests, as described previously [20]. At least two detections of LA antibodies in a period of at least 3 months were required for the definition of persistent LA.

The polymorphisms within the platelet GPIbα were determined by our previously published methodology and results were compared to the published data in 233 healthy controls [17].

High-shear-induced platelet plug formation was tested using the platelet function analyzer PFA-100 (Dade Behring, Marburg, Germany). The PFA-100 measures the time needed for a platelet plug to form after activation of platelets by pathophysiologically relevant stimuli, namely collagen and adenosine diphosphate (CADP) or collagen and epinephrine (CEPI) [21]. We determined both shear-stress induced closure times (CADP-CT, CEPI-CT), but CEPI-CT was used as the primary outcome variable, because it is the more sensitive indicator for platelet function, and the results allowed us comparisons with data that we had obtained in healthy controls [17, 22] at the same time as in LA individuals. Single determinations were performed in one batch of cartridges. Individual day-to-day variability in healthy subjects is 9–12%. Low von Willebrand factor antigen (vWFAg) levels and ristocetin cofactor activity (RiCo) may affect PFA-100 values. Therefore, blood was drawn without venostasis at the same time and under the same conditions as blood for the PFA-100 measurements for the assessment of vWFAg and RiCo. All samples were collected into evacuated tubes containing 129 mmol/l buffered sodium citrate (3.8%, Vacuette tubes, Greiner Bio-One, Kremsmünster, Austria). vWFAg was measured with a fully automated simultaneous thermal analyzer using the STA Liatest vWF (Diagnostica Stago, Paris, France). Plasma activity of RiCo was assayed by turbidometry using a commercial kit (BC von Willebrand reagent; Behring Marburg, Germany), which consists of lyophilized platelets and ristocetin, using the Behring Coagulation System (Behring) [23].

Platelet antibodies against GPIb/IX were determined by MAIPA, as described previously [24].

Statistical analyses

Chi-square test was used for comparison of categorical variables between groups. Mann–Whitney-U test was applied to compare metric variables between two groups, Kruskal–Wallis test to compare more than two groups, respectively. Correlations were calculated in nonparametric Spearman test. Odds ratios (OR) and corresponding 95% confidence intervals (CI) were calculated by binary logistic regression analysis comparing homozygous carriers of the (−5)T-allele of the Kozak polymorphism with carriers of the C-allele (CT and CC). Age was added in bivariate binary logistic regression models. A two-sided p value of less than 0.05 was considered to indicate statistical significance. All analyses were performed with the statistical software package SPSS for Windows.

Results

Allelic frequencies of the GPIbα polymorphisms VNTR and within the Kozak region in the two patients’ groups and healthy controls are shown in Table 2. The latter are similar to data in other Caucasoid populations [13] and were in all groups in Hardy–Weinberg equilibrium (Table 2). While allelic frequencies of the VNTR in patients did not differ from healthy controls (p = 0.45), we observed a significant deviation of the allelic frequencies within the Kozak region (p = 0.047). We, therefore, further compared these frequencies in healthy controls with those from LA/TE+ and LA/TE− individuals separately. There was no difference discernible between healthy controls and the LA/TE− group (p = 0.577). However, there was a significant predominance of the (−5)C allele in the LA/TE+ group compared to controls (p = 0.044).There was no difference between LA/TE+ and LA/TE−. The statistical power to detect a significant difference between LA/TE+ (n = 67) and LA/TE− (n = 33) was 8%. The (−5)C allele was associated with a twofold increased risk for thrombosis compared to controls (OR 2.0; 95% CI, 1.1–3.5; p = 0.02), as it resulted in significantly more (−5)TC in the LA/TE+ than in healthy controls (36% vs 22%, Table 3). No statistically significant odds ratio for thrombosis (LA/TE+ vs LA/TE−) was calculated for (−5)CC+CT genotypes compared to (−5)TT in a bivariate logistic regression analysis adjusted for age at the time of thrombosis (OR 1.3; 95% CI, 0.5–3.4; p = 0.53). As platelets play a major role in arteriosclerosis, we were interested to see if the observed allelic deviation in the LA/TE+ group is a result of an increased frequency of the (−5)C allele in the subgroup of 18 patients with a history of arterial TE disease. However, such a predominance of the (−5)C allele was not discernible (p = 0.77).

Table 2 Allele frequencies of the VNTR and Kozak polymorphisms of GPIbα in patients with lupus anticoagulant and in healthy controls
Full size table
Table 3 Genotypes of the Kozak sequence of GPIbα and platelet function as estimated by CEPI-CT using the PFA-100 in patients with lupus anticoagulant and in healthy controls
Full size table

Applying the platelet function analyzer PFA-100 in healthy individuals, we have observed shorter CEPI-CT in homozygotes with the (−5)TT vs (−5)TC Kozak genotype [17]. Similar to these healthy individuals, we expected shorter CEPI-CT in individuals from the LA/TE− group of the (−5)TT genotype than the (−5)TC genotype. In the LA/TE+ group such an influence of the genotype may be diminished, but we expected overall shortened CEPI-CT in the whole group. There was a significant correlation between CADP-CT and CEPI-CT in both patients’ groups (p = 0.002). CEPI-CT was shorter in LA/TE+ than LA/TE− (LA/TE+ median, 152 s; range, 101–300 s; LA/TE− median, 172 s; range, 69–300 s; p = 0.044). Of note, in both LA groups, we observed prolonged CEPI-CT compared to the respective genotype in healthy controls (Table 3). These were in LA/TE− (−5)TT, and (−5)TC 10 and two individuals, respectively; in LA/TE+ 12 and four patients, respectively, after excluding individuals with platelet counts less than 100 × 109/l, which may already affect the results from the PFA-100 (LA/TE− n = 3, LA/TE+ n = 1). [25] The median CEPI-CT in the two patients’ goups were hardly affected by exclusion of these six patients with platelet counts less than 100 × 109/l. (LA/TE− median, 168 s; range, 60–300 s; LA/TE+ median, 153 s; range, 101–300 s). The prolonged CEPI-CT were not explained by low vWFAg, which was within the normal range (60–180%) in all individuals (LA/TE− median, 130%; range, 64–339%; LA/TE+ median, 156%; range, 70–478%; p = 0.042). However, low RiCo was noted in three LA/TE− individuals with prolonged CEPI-CT, who were therefore excluded from further comparisons. Low RiCo was seen in one LA/TE+ subject with a normal CEPI-CT (not excluded). RiCo activities were in LA/TE− median, 131%, range, 41–354%, and in LA/TE+ median, 143%, range, 53–319%; p = 0.174. There was no difference of the CEPI-CT anymore discernible between the two LA groups after exclusions for low platelet counts and low RiCo (p = 0.21). In contrast to our expectations, CEPI-CT in LA/TE− (−5)TT individuals did not differ from (−5)TC in this group (Table 2, p = 0.42), which is in contrast to healthy controls. The CEPI-CT from patients with arterial disease was also not different from other LA/TE− individuals (Table 3, p = 0.8). Findings were not changed after exclusion of individuals with a CADP-CT > 170 s, which is the upper limit in healthy controls (data not shown).

Platelet antibodies directed against GPIb/IX were detectable in six patients in each patients’ group, and the CEPI-CT was prolonged in four patients from each group (p > 0.05).

Discussion

LA is regarded as a surrogate marker for the risk of TE, which may ensue once its path is paved by a variety of adverse factors coming together. Some of these could be inherited polymorphisms that shift the coagulation cascade towards thrombosis. Thus, platelet GP polymorphisms have been associated with TE [13, 14]. In this report, we show that the (−5)C allel of the Kozak sequence of GPIbα may contribute to the risk for TE in individuals with LA.

Platelets may have distinct roles in the pathomechanism of TE in individuals with LA. First, they may contribute to the generation of LA as a result of the exposure of negatively charged phospholipids, which together with plasma proteins induce LA formation. Alternatively, already activated platelets [7] may have an increased affinity to bind LA. Thus, platelets’ conformation, either in the resting state or activated, may play a significant role in the progress of coagulation, eventually leading to thrombosis. The conformational change of GPIb allows exposure of the binding site for the vWF, binding of thrombin, FXI and FXII, P-selectin and MAC-1 or αMβ2. It was therefore tempted to speculate that GPIbα genes are associated with an increased risk for thrombosis [13, 14].

We observed no deviation of the VNTR gene frequencies of GPIbα in this cohort of LA individuals compared to healthy controls. Our data, however, disclose a predominance of the (−5)C allele of the Kozak sequence in the LA/TE+ population. It has been proposed that this allele is associated with a higher density of the GPIbα than the (−5)T allele [16]. These findings lead to the hypothesis that individuals with the (−5)C allele have platelets of increased function disposing towards an increased risk for coronary disease or stroke [13, 14].

Assuming that platelets in LA/TE− individuals do not differ from healthy individuals, we expected in LA/TE− individuals, like in healthy controls, shorter CEPICT in (−5)TT than in (−5)TC. In LA/TE+ patients, we expected shortened closure times, irrespective the Kozak genotype, indicating increased platelet function. Applying high shear stress rates by the PFA-100, we could not disclose in any of the two LA patients’ groups an increased platelet function, however. Rather, we observed prolonged closure times in individuals from both groups, which were not explained by low vWFAg or RiCo levels. It may be that other factors influence the in vitro response of platelets to high shear stress rates in LA. Platelet autoantibodies, which are common in individuals with LA [26–31], are key candidates to affect platelet function in LA; too few patients had antibodies to GPIb/IX to detect an association between these antibodies and the CEPI-CT, however. Other, so far unknown influences may also play a role in LA individuals’ platelet function, as estimated by the PFA-100. Likewise, Harrison et al. [25] noted prolonged rather than shortened closure times by the PFA-100 in some individuals with the APS.

In a recent report in patients with APS, Levy et al. [32] have noticed an increased platelet deposition on extracellular matrix as determined by the Cone and Plate(let) analyzer in a subgroup of patients with a history of TE. These strong platelet responses were attributed to increased levels of vWFAg in their patients. Likewise, vWFAg levels were high in many individuals from both LA groups in our report, and this elevation was more pronounced in the LA/TE+ group. In accordance with our data on FVIIIAg levels in this patient population [33], a clear differentiation between LA/TE+ from LA/TE− was not possible. As CEPI-CT and CADP-CT were prolonged in some LA individuals from both groups, the PFA-100 may be sensitive to a different malfunction of platelets in LA individuals than the Cone and Plate(let) analyzer or even aggregometry [32].

Our results need cautious interpretation. First, the number of LA individuals in our patient cohorts may be too small to disclose a small or moderate association of the VNTR and the development of LA and/or thrombosis. In particular, larger patient numbers are needed to confirm the association of the (−5)C allele of the Kozak sequence with the risk for TE disease and more rapid plug formation in LA patients, as determined by the PFA-100. Of note, we had to exclude some individuals from both patients’ populations from further evaluations because of variables that prolong the CEPI-CT, like low platelet counts or low RiCo, and therefore may have missed the correlation between a shortened CT and the polymorphism within the Kozak region. Second, the group of individuals with LA who never developed TE is rather small. However, it is difficult to find individuals with a documented persistence of LA without thrombosis. Third, differences between sex and age of the patients and controls may affect our data. The majority of patients was female and of younger age in the LA/TE+ group than in the LA/TE− group. Thus, an estrogen effect may dispose these individuals for TE disease. We hypothesize, however, that LA most often develops already in young individuals, but only pathogenetic significant LA is associated with thrombosis. “Benign” LA that has also been formed in young age is then discovered by chance in elderly subjects who are undergoing medical examination. This hypothesis needs to be evaluated in prospective studies. Fourth, although the observation time was similar in both patients’ groups, we cannot exclude a future thrombotic event in an individual from the LA/TE− group. Longer observation times may be required to address this question. However, only one patient from the LA/TE− patients received oral anticoagulation, which may have prevented future thrombosis. Last, thromboembolism in LA is a possibly fatal disorder, and studying survivors may bias the results, because those who succumb to the disease are lost for evaluation.

In summary, our data suggest an association of the (−5)C allele of the Kozak sequence in LA patients with the risk for TE. However, this genetic disposition may require additional signals for disposing at risk for thromboembolism. The affected platelet function in individuals with LA, as estimated by the PFA-100, is not associated with a particular Kozak genotype or a history of TE, however. These data need confirmation in larger, preferentially multicenter, prospective studies.