Lupus anticoagulant and history of thrombosis are not associated with persistent endothelial cell activation in systemic lupus erythematosus

Abstract

Antiphospholipid antibodies (aPL), especially lupus anticoagulant (LAC), characterize systemic lupus erythematosus (SLE) patients at increased risk for arterial and venous thromboembolic complications. It has been reported that purified human anti-phospholipid antibodies cause endothelial cell activation in in vitro experiments. In order to investigate whether increased endothelial cell activation is associated with thromboembolic events in SLE patients with LAC, we measured plasma levels of thrombomodulin (TM), von Willebrand factor (vWf), sP-selectin, vascular cell adhesion molecule-1 (sVCAM-1) and ED1-fibronectin in a study of 76 patients with SLE. Patients were subdivided on the basis of: no history of thrombosis and LAC-negative (n = 22) or LAC-positive (n = 17); positive history of thrombosis and LAC-negative (n = 16) or LAC-positive (n = 21). The median SLE disease activity index (SLEDAI) was 4. Although concentrations of sTM, vWf, sP-selectin and sVCAM-1 were significantly elevated in SLE compared with values in healthy controls, they did not differ between the four groups, between patients with or without history of thrombosis, and between patients with or without LAC. Presence of anticardiolipin antibodies could not explain these negative findings. Adjustment of the concentrations for significantly associated variables, such as age, hypertension, smoking, immunosuppressive treatment and concentrations of creatinine, cholesterol and homocysteine, did not change the main results of the study. Only sTM was significantly lower in patients with both LAC and thrombosis than in patients without both these features after adjustment for serum creatinine concentrations. In conclusion, we did not find an association between endothelial cell activation and presence of LAC or history of thrombosis in SLE.

Introduction

Systemic lupus erythematosus (SLE) is frequently complicated by arterial and venous thromboembolism [1]. In addition to other disease-associated risk factors and predisposing conditions such as hypertension, corticosteroid treatment, premature atherosclerosis, hyperhomocysteinaemia, vasculitis and endocarditis, presence of antiphospholipid antibodies (aPL), such as lupus anticoagulant (LAC) and anticardiolipin antibodies (aCL), is an important risk factor [2–4]. In recent years, research has been focused on the role of aPL in the pathogenesis of thrombosis. In vitro experiments and animal studies of exposure of endothelial cells to human affinity-purified aPL demonstrated activation of endothelial cells, increased sticking of leucocytes and enhanced thrombus formation. These studies suggested that aPL-induced endothelial cell activation leads to a procoagulant state and is implicated in the pathogenesis of thrombosis [5–8].

In patients with SLE, increased levels of circulating markers of endothelial cell activation have been demonstrated. sVCAM-1 (soluble vascular cell adhesion molecule-1) and von Willebrand factor (vWf) are both synthesized in endothelial cells and were shown to be elevated in plasma of SLE patients in association with disease activity [9–11]. Thrombomodulin, an endothelial anticoagulant, also rises in parallel with disease activity in SLE [12–14]. Increased concentrations of sP-selectin are reported in SLE and other disorders accompanied by endothelial cell activation [15–17]. ED1-fibronectin is a cellular fibronectin variant containing an extra domain (ED1, EDA) produced by alternative mRNA splicing. ED1-fibronectin is exclusively secreted by endothelial cells, fibroblasts and smooth muscle cells [18]. Elevated concentrations were demonstrated in disorders in which endothelial cell activation is present and have recently been reported in rheumatoid vasculitis [15,19].

The aim of the present study was to investigate whether concentrations of circulating endothelial cell activation markers are elevated in SLE patients with a history of thrombosis and/or with LAC, compared with SLE patients without these features.

Patients and methods

Patients

The patients in our study originated from a cohort of 175 unselected consecutive patients with SLE who were treated at the out-patient clinic of the Department of Rheumatology and Clinical Immunology of the University Medical Center Utrecht. After they consented to participate, all study patients were interviewed by means of a standardized questionnaire and underwent a physical examination. All patients fulfilled at least four criteria of the American College of Rheumatology (revised ACR-criteria) for the diagnosis of SLE [20]. Disease activity was measured by the SLE Disease Activity Index (SLEDAI). The maximum score on the SLEDAI is 105, but in clinical practice scores rarely exceed 48 [21]. Hypertension was defined as systolic blood pressure ≥160 mmHg and/or diastolic blood pressure ≥90 mmHg on examination. An extensive analysis of the clinical charts was made. Deep venous thrombosis was diagnosed by venography or ultrasonography, pulmonary embolism by radionuclide scanning, portal vein thrombosis by angiography and thrombophlebitis by history. Ischaemic stroke was documented by brain CT or MRI scan and thrombosis of the abdominal aorta and peripheral arteries by arteriography or surgery. Myocardial infarction was diagnosed if typical electrocardiographic features and an elevated creatine kinase MB fraction were present. Retinal artery thrombosis required relevant abnormalities on funduscopy and fluorescence angiography. TIA (transient ischemic attack) was diagnosed by history.

Since we found that presence of LAC was associated with a six- to seven-fold increase in the risk of thromboembolism, and others also found a strong association of risk of thrombosis with LAC [3,22], we categorized patients according to presence of LAC and history of thrombosis. We calculated that 16 patients per subgroup would be needed to demonstrate one standard deviation difference between subgroups with sufficient precision (α = 0·05 and power = 80%). From the whole cohort of 175 patients, we selected all 18 LAC positive patients without a history of thrombosis and all 18 LAC negative patients with a history of thrombosis. After exclusion of patients with aCL in LAC-negative patients without a history of thrombosis, two complementary groups of 22 individuals were formed in alphabetical order of patients without thrombosis and without LAC, and of patients with both thrombosis and LAC, totalling 80 patients. Afterwards, four patients treated with renal dialysis were excluded: two LAC-positive patients without a history of thrombosis, and one each from both groups of patients with a history of thrombosis.

Blood sampling and laboratory investigations

Blood was sampled by evacuated tube system, collected in citrate anticoagulant (1:10 in 3·1% citrate) and centrifuged immediately at 2000 g for 15 min at 4°C. The supernatant was removed and centrifuged a second time. Plasma samples were stored at −70°C. Commercially available enzyme-linked immunosorbent assay (ELISA) kits were used for measurement of sVCAM-1 (R&D Systems, Abingdon, UK) and soluble thrombomodulin (sTM) (Diagnostica Stago, Paris, France). sP-selectin, vWf and ED1-fibronectin were measured with ELISAs developed at the University Medical Center Utrecht [15,23]. Briefly, microtitre plates were coated overnight with specific monoclonal antibodies (MoAb) at 4°C and then blocked. Each plate contained eight concentrations of a standard. This was either recombinant P-selectin (R&D Systems, UK), pooled human serum or ED1-fn purified from cultured human fetal lung fibroblasts. For the sP-selectin ELISA we used MoAb 2·15 to capture and biotinylated MoAb 1·18 for detection, both recognizing different epitopes [15,24]. For the vWf ELISA we used antihuman vWf (Dako) to capture and peroxidase coupled antihuman vWf (Dako) for detection. ED1-fn was captured with IgM MoAb 3E2 which was raised against fibronectin antigen released by cultures of human breast cancer cell lines (Sigma, St Louis, MO, USA) [25] and detected with biotinylated antihuman fibronectin F(ab)₂-fragments. Biotinylated F(ab)₂ fragments were incubated with streptavidin/horseradish peroxidase. Staining was performed with o-phenylenediamine dihydrochloride. Optical densities were read at 490 nm with the V_max kinetic microplate reader (Molecular Devices Corp, USA). vWf was expressed as percentage of the value in pooled plasma of 40 healthy donors (10 µg/ml).

Reference values from our laboratory (means ±s.d.), derived from 25 healthy controls, are for sTM 29·1 ± 11·5 ng/ml, for sVCAM-1 510·7 ± 135·6 ng/ml, for vWf 115·0 ± 52·4%, for sP-selectin 128·9 ± 41·0 ng/ml and for ED1-fibronectin 1·69 ± 0·9 µg/ml.

Detection of LAC and anticardiolipin antibodies was performed as described previously [4]. For the detection of LAC a dilute prothrombin time was measured, using recombinant human tissue factor (Innovin, Baxter Diagnostics Inc., Deerfield, USA). Positive samples (PT patient:PT control >1·12) were tested for the dilute Russell's viper venom time (dRVVT; IL LAC screen, Instrumentation Laboratory, Milan, Italy) and were retested after 1:1 mixing with normal plasma. If the clotting time of the mixed plasma still was ≥ 20% longer than normal plasma, the dRVVT was repeated with an additional amount of phospholipids (IL test LAC confirm, Instrumentation Laboratory, Milan, Italy). Samples with a ratio of the IL Test Screen/LAC confirm >1·2 were considered positive for LAC. Patients regarded LAC-positive in the present study also had positive tests in previous samples taken during routine follow-up. For the detection of aCL, cardiolipin (48 μg/ml ethanol 70%; Sigma, St Louis, MO, USA) was coated on 96-well polyvinyl assay plates (Costar, Cambridge, MA, USA). Nine calibrators for IgG and IgM anticardiolipin antibodies (Louisville APL Diagnostics Inc., Louisville, KY, USA) were used. Alkaline phosphatase conjugated antibodies were added. P-dinitrophenyl phosphate in diethanolamine buffer was used as substrate and colour development was read in a multiscan photometer (V_max reader; Molecular Devices, Menlo Park, CA, USA).

Data analysis

Data entry and analysis were performed with the SPSS statistical software package. Concentrations are presented as mean values ±s.d. Differences in levels of markers between groups were tested with the Mann–Whitney U-test. The relationship of relevant demographic and clinical variables with the markers was assessed by univariate linear regression. Differences in the concentrations of the markers between the four study groups were adjusted for these variables in a multivariate regression analysis. P-values of <0·05 or 95% confidence limits that did not include 0 were considered statistically significant.

Results

Patients

Characteristics of each group of patients are described in Table 1. Overall, the mean age was 37·6 ± 10·2 years; 89% was female. Mean disease duration was 10·7 years (range 1–38). The median SLEDAI score was 4 (range 0–12). Thirty-nine patients (51%) received immunosuppressive therapy: 38 patients were treated with prednisone and 20, mostly additionally, with azathioprine (n = 16), methotrexate (n = 1) or cyclophosphamide (n = 3). Twenty-nine patients (38%) smoked and hypertension was present in 32 patients (42%). None of the patients had diabetes mellitus. Mean serum cholesterol was 4·8 ± 1·0 mmol/l, mean serum creatinine 95·2 ± 51·6 µmol/l and mean plasma homocysteine 15·9 ± 7·2 µmol/l. Anticardiolipin antibodies were present in 39 patients (51%).

Table 1.

Patient characteristics

	LAC – thrombosis – (n = 22)	LAC + thrombosis – (n = 16)	LAC – thrombosis + (n = 17)	LAC + thrombosis + (n = 21)
Age (years)	35·5 ± 11·6	37·6 ± 10·2	39·4 ± 9·6	38·4 ± 9·6
Female (%)	82	100	88	91
Smoking (%)	36	50	29	38
Hypertension* (%)	59	38	29	38
Cholesterol (mmol/l)	4·4 ± 0·9	5·1 ± 0·8	5·0 ± 0·9	4·8 ± 1·3
SLEDAI†	3 (0–12)	4 (0–12)	2 (0–12)	4 (0–6)
Disease duration (years)	7·5 (2–21)	9·5 (1–18)	12 (2–26)	10 (1–38)
Time since thrombosis (yrs)‡	—	—	5·0 (1·1–18·0)	6·3 (0·4–17·1)
Presence of aCL (%)	0	81	41	90
Creatinine (µmol/l)	89·5 ± 37·9	99·3 ± 69·0	93·3 ± 56·6	99·7 ± 47·8
Homocysteine (µmol/l)	13·5 ± 6·4	14·7 ± 5·9	19·1 ± 9·3	16·6 ± 6·1
Prednisone (%)	55	44	59	43
Other immuno-suppressants (%)§	23	31	35	19
Oral anticoagulants (%)	0	0	24	67
Aspirin (%)	0	0	6	33

Values are expressed as means ± s.d., as numbers (%), or as median (range).

^*Hypertension is defined as systolic blood pressure ≥ 160 mmHg and/or diastolic blood pressure as ≥ 90 mmHg.

^†SLE disease activity index.

^‡Time since last thrombotic event.

^§Azathioprine in four patients in each group, methotrexate in 1 (LAC-T +), and cyclophosphamide in three (one each in LAC-T-, LAC + T- and LAC-T +); T, thrombosis.

Previous arterial thrombotic events had occurred in 22 patients and venous thrombosis in 24 (both arterial and venous events in eight patients). Some patients had more than one complication. Time since the last thrombotic episode ranged from 0·4 to 18·0 years (median 5·8). Venous thrombotic episodes consisted of deep venous thrombosis in 17 cases, pulmonary embolism in nine, superficial thrombophlebitis in six and portal vein thrombosis in one. Arterial thrombotic events included ischemic stroke in 14 patients associated with TIAs in six, isolated TIA in one, myocardial infarction in three, thrombosis of the abdominal aorta in one, peripheral arterial thrombosis in five, renal infarction in one and occlusion of a retinal artery in one.

Concentrations of endothelial cell activation markers

Plasma concentrations of sTM, sVCAM-1, von Willebrand factor, sP-selectin and ED1-fibronectin are shown in Fig. 1. We found no statistically significant differences between the four patient groups. Moreover, there were no significant differences between patients with or without a history of thrombosis, nor between LAC-positive versus LAC-negative patients. Except for ED1-fibronectin, mean levels of all markers were significantly elevated in SLE patients compared with values in healthy controls (see Patients and methods). Mean thrombomodulin in SLE patients was 48·9 ± 36·0 ng/ml (P = 0·004 versus healthy controls); mean sVCAM-1 was 627·1 ± 215·2 ng/ml (P = 0·011), mean vWf was 169·1 ± 65·4% (P < 0·001), mean sP-selectin was 172·1 ± 54·1 ng/ml (P < 0·001) and mean ED1-fibronectin was 1·86 ± 1·08 µg/ml (P = 0·6).

Fig. 1.

Concentrations of plasma thrombomodulin, sVCAM-1, von Willebrand factor, sP-selectin and ED1-fibronectin in SLE patients in the four patient groups: LAC-negative, no previous thrombosis; LAC-positive, no previous thrombosis; LAC-negative and previous thrombosis; LAC-positive and previous thrombosis. Bars represent mean values.

Associations between endothelial cell activation markers and clinical variables: univariate analysis

By means of univariate regression analysis we found statistically significant positive associations of sTM with age, hypertension, immunosuppressive treatment and concentrations of creatinine, cholesterol and homocysteine (Table 2). sVCAM-1 was significantly associated with age, disease duration and creatinine and homocysteine concentrations, and negatively associated with smoking. We found positive associations of vWf with age, immunosuppressive treatment and homocysteine levels, of sP-selectin with smoking and of ED1-fibronectin with none of the variables. None of the markers was significantly associated with sex, with the SLEDAI, or with presence of aCL.

Table 2.

Univariate regression analysis: coefficients β (95% confidence intervals)^*

Variable	Thrombomodulin	sVCAM-1	vWf	sP-selectin
Age	1·1 (0·3, 1·9)	7·9 (3·3, 12·5)	2·0 (0·5, 3·4)	n.s.
Disease duration	n.s.	10·1 (3·6, 16·5)	n.s.	n.s.
Hypertension†	24·3 (8·1, 40·4)	n.s.	n.s.	n.s.
Smoking†	n.s.	−127·7 (− 225·6, −29·7)	n.s.	26·7 (1·4, 51·9)
Immunosuppressive treatment†	18·8 (2·4, 35·1)	n.s.	33·8 (4·5, 63·0)	n.s.
Serum creatinine	0·5 (0·4, 0·6)	1·6 (0·7, 2·5)	n.s.	n.s.
Serum cholesterol	8·8 (0·7, 16·8)	n.s.	n.s.	n.s.
Plasma homocysteine	2·3 (1·2, 3·4)	12·7 (6·3, 19·1)	2·4 (0·3, 4·5)	n.s.

^*The regression coefficients should be interpreted as the increase in concentration for each unit increase of the variable studied, e.g. for 1 year increase of age, thrombomodulin level would increase by 1·1 ng/ml. For the dichotomized variables the coefficient represents the difference in concentration between presence and absence of the variable. The increase is statistically significant if the 95% CI does not include 0.

^†In the case of dichotomized variables, presence of the variable = 1 and absence = 0.

Multivariate regression analysis

The differences in the concentrations of the markers in the four patient groups were adjusted in a multivariate linear regression model. After adjustment for all significantly associated variables from the univariate analysis, no differences in concentrations of sVCAM-1, vWf, sP-selectin and ED1-fibronectin between the four patient groups, between patients with or without a history of thrombosis or between LAC positive and LAC negative patients were found. However, after adjustment for serum creatinine a significantly lower sTM concentration was present in SLE patients with both LAC and thrombosis than in SLE patients lacking both these features (difference −15·1 ng/ml; 95% CI: −25·0, −5·3; P = 0·004). After adjustment for both creatinine and all other significantly associated variables from the univariate regression analysis, this difference in sTM concentrations remained essentially unchanged.

Discussion

Mean plasma concentrations of thrombomodulin, sVCAM-1, vWf and sP-selectin were significantly elevated in the study population as a whole compared with the reference values in healthy controls. However, contrary to our expectation we could not find evidence of increased endothelial cell activation in SLE patients with a history of thrombosis or with LAC, compared with SLE patients without a history of thrombosis or without LAC. This implies that there is endothelial cell activation in clinically quiescent SLE but that this is not associated with LAC or with a history of thrombosis. After adjustment for other relevant factors known to influence endothelial cell activation, such as hypertension, the results remained unchanged.

We investigated the influence of LAC and not of aCL antibodies because LAC has a much higher discriminating potential for risk of thrombosis than aCL [3,22,26]. Horbach et al. found that in comparison with anticardiolipin IgG and IgM, antiβ2GPI and antiprothrombin antibodies, LAC was by far the most important risk factor for thrombosis in patients with SLE. In a stepwise logistic regression model, LAC was the only significant risk factor for arterial thrombosis, while for venous thrombosis both LAC and anticardiolipin IgM titres above 20 MPL antibodies were significant risk factors [22]. However, one could ask if the presence of aCL in the different subgroups could have caused bias. The absence of a statistically significant association between presence of aCL and levels of endothelial cell markers in the univariate regression analysis, as well as raised levels of endothelial cell markers in the absence of aCL in the LAC-negative patients without a history of thrombosis, argue against this suggestion. This also holds for the absence of higher levels of these markers in the two LAC-positive patient groups (with the highest percentages of aCL positive patients) compared with the LAC-negative patient groups.

Our results are in agreement with the results of Kaplanski et al., who did not find differences in elevated sVCAM-1 concentrations between patients with primary antiphospholipid syndrome (APS), SLE patients with both aPL and thrombosis, and SLE patients without aPL or thrombosis [27]. Still, sVCAM-1 seemed to be elevated in subgroups of patients with severe disease, e.g. with more frequent thrombotic events. However, the authors did not adjust levels of sVCAM-1 for the SLEDAI despite scores of up to 32 in the SLE patient groups. We could not confirm the findings of the study of Ames et al., who reported an association of vWf concentrations with history of thrombosis in a mixed population of aPL-positive patients with autoimmune thrombocytopenic purpura, SLE and APS [28]. In another study no differences in plasma vWf concentrations were found between patients with APS and SLE patients without APS [29]. One previous in vitro study showed elevated vWf release after addition of IgG from patients with SLE or APS to endothelial cells compared with IgG from healthy controls. However, the vWf release with patient IgG was not related to the presence of a history of thrombosis [30].

After adjustment for relevant clinical variables, especially serum creatinine [12,14], in a linear regression analysis, a lower sTM concentration was found in patients with both a history of thrombosis and LAC than in patients lacking both these features. In previous reports, contradictory findings are presented. Elevated unadjusted plasma thrombomodulin concentrations are reported in SLE patients with aPL compared with SLE patients without aPL [14,31], whereas others did not find an increase in sTM concentrations in patients with LAC or aCL [12,32,33]. The finding, upon multivariate regression analysis, of a relatively decreased sTM concentration in SLE patients with both LAC and a history of thrombosis needs confirmation. It might indicate a possible role of thrombomodulin in the pathogenesis of thrombosis in a subgroup of LAC-positive SLE patients, since thrombomodulin is able to bind thrombin and as a complex to activate protein C, and therefore downregulates coagulation.

Although in vitro studies showed aPL-induced up-regulation of endothelial cell expression of ICAM-1, VCAM-1 and E-selectin [5–8,34], we could not detect increased concentrations of endothelial cell activation markers in the peripheral blood of SLE patients with LAC. In vitro investigations may not reflect pathophysiological in vivo processes. Differences in the reactivity of human umbilical vein endothelial cells (HUVECs) used in vitro and the human endothelial vascular lining in vivo might be an explanation.

In conclusion, the present study could not confirm an association between endothelial cell activation and presence of LAC or history of thrombosis in patients with SLE.