Vascular Abnormalities, Paraoxonase Activity, and Dysfunctional HDL in Primary Antiphospholipid Syndrome FREE

The antiphospholipid syndrome is an autoimmune disorder characterized by increased levels of circulating antiphospholipid antibodies (aPL), increased thrombotic tendency, and pregnancy morbidity.^1- 2 Although advances in the treatment and follow-up of individuals with positive aPL have markedly reduced the occurrence of thrombotic events, it has become increasingly recognized that accelerated atherosclerosis is an additional problem in patients with aPL, as in other rheumatic diseases.

Preclinical atherosclerotic changes (endothelial dysfunction and increased carotid intima media thickness [CIMT]) have been found in small case-control studies of patients with aPL, but the mechanisms promoting vascular disease remain largely unknown.^3- 5 It has been suggested that aPL antibodies themselves may elicit cellular and humoral responses that are proatherogenic, including enhanced oxidative stress.^5- 6

An inverse relationship between anticardiolipin antibodies and paraoxonase activity, a high-density lipoprotein (HDL)–related antioxidant enzyme that has a key role in defense against lipid peroxidation, has been reported.⁷ Recent studies indicate that HDL, in the setting of low-grade systemic inflammation, may lose its protective properties and acquire a pro-inflammatory and pro-oxidant phenotype promoting atherosclerosis.^7- 9 Furthermore, HDL-targeted antibodies have been observed in patients with aPL.⁷

We compared vascular function and structure with HDL properties in a cross-sectional study of patients with aPL and controls, and determined the relationship between vascular function and structure and paraoxonase activity levels. In addition, we investigated whether vascular abnormalities were related to the occurrence of thrombotic events in aPL.

Between June 2006 and April 2009, we studied 77 women who attended the outpatient clinic of St Thomas' Hospital Lupus Unit and the Charing Cross Hospital Rheumatology Unit, London, England. Inclusion criteria were the presence of medium or high levels of aPL (anticardiolipin IgG >10 GPL [normal values <7.0], IgM >10 MPL [normal values <7.0]) and lupus anticoagulant (dilute Russell viper venom test and activated partial thromboplastin time) antibodies on at least 2 occasions 12 weeks apart. Exclusion criteria were the presence of other vasculopathies, including systemic lupus erythematosus.

All patients with positive aPL who fulfilled the inclusion criteria were identified from our patient database and were invited to participate. Information about cardiovascular risk (CV) factors, current medication, and complications of disease (ie, thrombotic events and fetal losses) in the aPL group was retrieved from patient notes. An equal number of control participants of similar age and sex, matched for CV factors, were recruited from friends of patients with aPL and hospital staff during the same period. The matching process was performed on a frequency and not on an individual basis.

A medical history was taken from all participants followed by anthropometric measurements, vascular assessment, and blood sampling. Participants were asked to refrain from using any medication, such as aspirin, antihypertensive drugs, and antilipidemic drugs, for 48 hours before vascular measures. Institutional review board approval was received and all participants gave written informed consent.

Weight, height, and body mass index (BMI, calculated as weight in kilograms divided by height in meters squared) were measured.² Systolic and diastolic blood pressure (BP) and heart rate were measured (mean of 3 seated measurements) with an automated oscillometric device (Dinamap 9301 vital signs monitor; Critikon, Tampa, Florida).

Blood samples after a 12-hour fast were centrifuged at 3000 × g at 4°C for 10 minutes to obtain serum and frozen at –70°C until assayed. Routine chemical methods were used to determine serum concentrations of total cholesterol, HDL, triglycerides, and glucose (colorimetric enzymatic method; Technicon automatic analyzer RA-1000, Date-Behring Marburg, Marburg, Germany) and high-sensitivity C-reactive protein.

Anti-HDL Antibodies. This method has been previously described.¹⁰ IgG anti-HDL antibodies were measured by enzyme-linked immunosorbent assay using 96 well plates (Polysorp, Nunc, VWR, Portugal) half coated for 1 hour at 37°C with 20 μg/mL human HDL (Sigma-Aldrich, Sintra, Portugal) in 70% ethanol. Blocking was performed by using phosphate buffer solution containing 1% albumin from bovine serum for 1 hour at 37°C. Samples were diluted 1:100 in blocking agent; 100-μL samples and positive control were added to duplicate wells in both halves of the plate for 1 hour at 37°C; 100 μL of alkaline phosphatase–conjugated anti-human IgG (1:1000) was added for 1 hour; and 100-μL p-nitrophenyl phosphate (Sigma-Aldrich) in bicarbonate buffer (pH 9.8) was added and incubated at 37°C, and the absorbance read at 405 nm after 1 hour. Interplate and intraplate coefficients of variation were less than 10%.

Paraoxonase Activity. Serum paraoxonase activity was measured as described by Eckerson et al with some modifications.⁷ Briefly, 1.0-mM paraoxon (Sigma-Aldrich) freshly prepared in 290 μL of 50-mM glycine buffer containing 1-mM calcium chloride (pH 10.5) was incubated at 37°C with 10 μL of serum for 10 minutes in 96 well plates (Polysorp). Formation of p-nitrophenol was monitored at 412 nm and activity was expressed as μmol p-nitrophenol/L/serum/min.

Carotid Intima Media Thickness. The right and left common carotid arteries were scanned with a 5 to 10 MHz linear array transducer (Acuson, Aspen, Mountain View, California). The carotid bulb was identified and longitudinal 2-dimensional ultrasonographic images of the common carotid artery 1 to 2 cm proximal to the carotid bulb were obtained. The optimal longitudinal image was acquired on the R wave of the electrocardiogram and continuously recorded on videotape for 5 seconds. Measurements of the arterial wall were made from stored images, as previously described.¹¹

Flow-Mediated Dilatation. Each woman underwent measurement of endothelium-dependent vascular responses of the right brachial artery by high-resolution ultrasound imaging (Acuson Aspen, 5-10 MHz linear probe). Analysis was performed by using automated software (Brachial Tools; Medical Imaging Application, Coralville, Iowa) as previously described and expressed as a percentage change from baseline diameter.¹² Doppler-derived flow measurements (pulsed wave Doppler signal at a 70° angle) were also obtained continuously. The increase in blood flow after the release of the cuff was expressed as a percentage of the baseline flow. The endothelium-independent response to 25 μg of sublingual nitroglycerine was also calculated as a percentage change from the baseline diameter.

Pulse Wave Velocity. The pressure pulse waveform was recorded transcutaneously from the radial and carotid pulse using a high-fidelity micromanometer (SPC-301; Millar Instruments, Houston, Texas). Pressure pulse recordings were performed consecutively at the 2 points simultaneously with an electrocardiogram signal, which provided an R-timing reference. Pulse wave velocity was calculated by using the mean time difference and arterial path length between the 2 recording points (SphygmoCor version 7.0, Scanmed, England).

High-density lipoprotein from patients with aPL and controls was isolated by sequential ultracentrifugation (d = 1.063-1.21 g/mL) using solid potassium bromide for density adjustment. To ensure that no significant loss of HDL-associated paraoxonase occurred during the procedure, paraoxonase activity and paraoxonase 1 protein were measured in serum (before HDL isolation) and in isolated HDL. There was no significant loss of HDL-associated paraoxonase activity (serum paraoxonase activity: mean [SD], 105 [22] U/mg HDL protein vs HDL-paraoxonase activity: 98 [20] U/mg HDL protein; P = .41) or paraoxonase 1 content (HDL-paraoxonase 1 content: 94% [7%] of serum paraoxonase 1 content, determined by Western blot analysis) after ultracentrifugation.

The effect of HDL (50 μg protein/mL) isolated from women positive for aPL and controls on nitric oxide production in human aortic endothelial cells (HAECs) was assessed by electron spin resonance (ESR) spectroscopy using the spin-probe colloid Fe(DETC)₂ (Noxygen, Elzach, Germany), which was described previously.¹³ Electron spin resonance spectra were recorded on a Bruker e-scan spectrometer (Bruker BioSpin, Billerica, Massachusetts). The coefficients of variation for intra-assay and interassay variability of these measurements were 1.92% and 1.74%, respectively.

The effect of HDL isolated from patients with aPL and controls on endothelial cell superoxide production was assessed in tumor necrosis factor (TNF)–stimulated (5 ng/mL; R&D Systems, Minneapolis, Minnesota) HAECs by ESR spectroscopy using the spin-probe 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine (Noxygen), which was described previously.¹³ Electron spin resonance spectra were recorded after addition of 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine (final concentration, 200 μM) under stable temperature conditions. The coefficients of variation for intra-assay and interassay variability of these measurements were 5.1% and 12.9%, respectively.

Expression of vascular cell adhesion molecule 1 (VCAM-1) in TNF-stimulated (5 ng/mL) HAECs treated with or without HDL (50-μg protein/mL) was determined by Western blot analysis (primary antibody against human VCAM-1; R&D Systems).

Human monocytes were separated by magnetic cell sorting with CD14 antibodies (Miltenyi Biotec, Bergisch-Gladbach, Germany) from peripheral blood mononuclear cells by Ficoll density gradient centrifugation (Vacutainer CPT, Beckton-Dickinson, Basel, Switzerland) and isolated human monocytes were resuspended in RPMI 1640 containing 10% serum (Gibco, Invitrogen, Paisley, Scotland) and labeled with carboxyfluorescein diacetate succinimidyl ester (Molecular Probes, Invitrogen).

The HAECs were cultured to confluency in 24 well plates as described above and stimulated for 3 hours with TNF (1 ng/mL). Afterwards HDL (50 μg/mL) was added to the medium and HAECs were incubated for an additional hour. The adhesion assay was performed after changing the medium by addition of human monocytes (50 000/well) to the HAEC monolayer and incubation for 3 hours in a humidified atmosphere (37°C, 95% air/5% carbon dioxide). Nonadherent cells were removed by washing wells with phosphate buffered saline (Gibco, Invitrogen) and HAECs were counterstained with 4′,6-diamidino-2-phenylindole suspended in mounting medium (Dako, Baar, Switzerland). Adherent monocytes labeled with carboxyfluorescein diacetate succinimidyl ester were counted on 4 randomly selected high-power fields using a fluorescent microscope (DM-IRB; Leica, Heerbrugg, Switzerland) connected to a digital imaging system (Spot-RT; Diagnostic Instruments/Visitron Systems, Puchheim, Germany) and normalized to the number of HAECs present on each field. The coefficients of variation for intra-assay and interassay variability of these measurements were 8.9% and 5.65%, respectively.

The normality of continuous variables was assessed by the Kolmogorov-Smirnov test. All normally distributed variables are expressed as mean (SD), unless otherwise stated. Data not normally distributed are expressed as median (interquartile range [IQR]). For the main vascular outcomes, a sample size of 70 participants in each group was calculated to demonstrate a change by 0.06 mm in CIMT, 0.6 m/s in pulse wave velocity, and 1.2% in flow-mediated dilatation, with 80% power and a 5% significance level assuming an SD of 0.2 mm for CIMT, 2 m/s for pulse wave velocity, and 4% for flow-mediated dilatation.

Descriptive comparisons between patients with aPL and controls were performed using unpaired 2-sample t test. Log transformation was applied for not normally distributed data (ie, triglycerides, C-reactive protein). If transformation was unsuccessful, nonparametric methods (Mann-Whitney U test for 2 groups) were applied.

Multiple comparisons of the changes within the aPL group were performed by 1-way analysis of variance by applying the Bonferroni correction for normally distributed data and the Kruskal-Wallis test analysis of variance for not normally distributed data.

Linear multivariate regression analysis using backward elimination procedure was used to examine relationships between vascular measurements and aPL status controlling for the following potential confounders: age, BMI, BP, smoking, hypercholesterolemia, and paraoxonase activity, which showed an association with the dependent variable in univariate analysis at 10% significance level. Data were expressed as standardized coefficients for significance comparisons. Collinearity was tested between measures using the Spearman correlation coefficient. For parameters such as hypercholesterolaemia and total cholesterol where the correlation coefficient was more than 0.6, only the one that showed the greatest influence on the outcome measure was included in the model.

Within the aPL group, the odds of having a thrombotic event were estimated through multivariate logistic regression analysis using the following as mutually adjusted possible determinants: CIMT as a categorical variable, pulse wave velocity, flow-mediated dilatation, age, systolic BP, and paraoxonase activity. Hosmer and Lemeshow goodness of fit test was applied. Two-sided P<.05 was considered statistically significant. All statistical analyses were performed by using SPSS version 13.0 (SPSS Inc, Chicago, Illinois).

In the aPL group, 38 women had a documented thrombotic event, 23 had a history of at least 3 miscarriages and/or fetal deaths (>10 weeks' gestation), and the remaining 16 had only positive aPL and did not fulfil Sydney criteria for antiphospholipid syndrome. Comparisons of CV factors between women positive for aPL and controls is shown in Table 1.

Classic CV risk factors were similarly distributed between the aPL subgroups, only diabetes was overrepresented in patients with aPL who had experienced a thrombotic event (Table 2). Lipid parameters, including anti-HDL, anti-apolipoprotein A, and paraoxonase activity were unrelated to clinical events.

Carotid Intima Media Thickness . CIMT was significantly greater in patients with aPL vs controls (mean [SD], 0.75 [0.16] vs 0.64 [0.09] mm; 95% confidence interval [CI], –0.14 to –0.06; P < .001). Following multivariate analysis, the only factors that were independently associated with CIMT in aPL were paraoxonase activity (standardized beta coefficient = –0.40, P < .001) and age (beta = 0.26, P = .02). Among patients with aPL, those who had previously experienced a thrombotic event had greater CIMT compared with those without prior thrombosis (mean [SD], 0.8 [0.2] vs 0.7 [0.1] mm; 95% CI, –0.15 to –0.02; P = .01 for those patients with documented thrombosis vs the rest). Following multivariate analysis, CIMT was the only independent determinant of thrombosis in aPL (adjusted odds ratio, 3.6; 95% CI, 1.04-12.2; P = .04; n = 73, Hosmer and Lemeshow test, χ28=4.7; P = .70).

In controls, CIMT was correlated with classic CV risk factors (age, systolic and diastolic BP, total cholesterol, and low-density lipoprotein cholesterol). No association with paraoxonase activity was noted. In multivariate analysis, aPL (beta = 0.33, P = .001), paraoxonase activity (beta = –0.23, P = .001), age (beta = 0.29, P = .001), and systolic BP (beta = 0.29, P = .001) were the only factors independently associated with CIMT.

Flow-Mediated Dilatation . Flow- mediated dilatation was significantly lower in women with aPL compared with controls (6.2% [4.1%] vs 9.6% [4.2%]; 95% CI, 2.02%-4.69%; P < .001). There was no difference in baseline diameter or blood flow changes between aPL and controls (data not shown). In multivariate analysis, aPL (beta = –0.30, P < .001), brachial diameter (beta = –0.32, P < .001), nitroglycerine (beta = 0.15, P = .047), and age (beta = –0.16, P = .03) were significantly associated with flow-mediated dilatation in the whole cohort.

Pulse Wave Velocity . Pulse wave velocity was greater in patients with aPL vs controls (9.2 [1.6] vs 8.5 [1.8] m/s; 95% CI, –1.14 to –0.06; P = .04). There was no relationship between pulse wave velocity and clinical manifestations of aPL. In multivariate analysis, paraoxonase activity (beta = –0.22, P = .008) and systolic BP (beta = 0.25, P = .002) were associated with pulse wave velocity.

HDL and Nitric Oxide Production in HAECs . Incubation of HAECs with HDL from women with aPL and controls did not induce any decrease in HAECs viability. The HAECs in a basal state were found to produce a median 567 (IQR, 522-600) nmol/h/250 000 cells of nitric oxide as assessed by ESR spectroscopy. The HDL from matched controls amplified nitric oxide production from HAECs by 7%. In marked contrast, the nitric oxide production from HAECs was reduced by 21% when cells were treated with HDL from women positive for aPL (median [IQR], 608.8 [529.7-648.0] vs 471.5 [442.6-498.1] nmol/h/250 000 cells for aPL and controls; 95% CI, –202 to –96; P < .001) (Table 3).

Anti-Inflammatory Properties of HDL . The HAECs constitutively expressed a low amount of VCAM-1 as measured by Western blot assay. Stimulated expression of VCAM-1 from HAECs was achieved by 5-ng/mL TNF. Incubation of HAECs with 50 μg/mL of HDL from patients and controls was associated with 27% and 54% reduction in VCAM-1 expression, respectively (median [IQR], 73.2% [63.0%-80.6%] vs 46.2% [36.6%-56.9%] of TNF-treated cells in women with aPL vs controls; 95% CI, 13.6%-37.8%; P < .001) (Table 3).

Monocyte adhesion on HAECs was greater following TNF stimulation. This effect was markedly reduced when cells were pretreated with HDL from both patients positive for aPL and controls. However, the magnitude of this inhibition was substantially greater in controls (median [IQR], 16.2 [12.0-19.5] vs 10.5 [8.0-13.0] of TNF-treated cells in patients with aPL vs controls; 95% CI, 2.3-9.1; P = .009). HDL-induced stimulation of nitric oxide production was inversely associated with monocyte adhesion in HAECs (Pearson correlation coefficient r = –0.7, P = .008) (Table 3).

HDL and Antioxidative Properties . Unstimulated HAECs showed minimal superoxide production, which was greater following TNF stimulation. Addition of HDL from controls abolished the pro-oxidant effect of TNF stimulation and superoxide production was similar to baseline levels (5.5 [5.3-5.6] vs 5.7 [4.9-7.1] nmol/min/250 000 cells, baseline vs controls, respectively; P = .70). However, the antioxidant effect of HDL from aPL was markedly lower compared with HDL isolated from risk factor–matched controls (5.5 [5.3-5.6] vs 8.8 [6.5-10.8] nmol/min/250 000 cells, baseline vs HDL from aPL, respectively; 95% CI, –5.7 to –0.8; P = .008) (Table 3).

We showed that women with positive aPL antibodies, irrespective of prior clinical complications, have functional and structural arterial abnormalities that were associated with reduced activity of paraoxonase, an HDL-related antioxidant enzyme. This implicates HDL and oxidative stress in the causal pathway for atherosclerosis in these patients. We have also shown that HDL in aPL has a “proatherogenic” phenotype. It reduces nitric oxide bioavailability and has impaired anti-inflammatory and antioxidant properties. These findings identify HDL as a potential new target for interventions in this group of patients.

The premature development of atherosclerotic changes in patients with primary antiphospholipid syndrome has been a matter of intense debate for the past decade.^14- 19 The thrombotic tendency of the syndrome, which is its main clinical manifestation, has been attributed mostly to the procoagulant activity that aPL exerts on endothelial cells and to inflammatory/immune mechanisms, leading to auto-antibody–mediated thrombosis rather than to enhanced atherosclerosis.²⁰ Premature atherosclerosis has, however, been found in peripheral arteries but the role of aPL as an independent risk factor for atherosclerosis has been unclear.^3,21 Most studies have been small with confounding by patients with other conditions such as systemic lupus erythematosus and by high level of conventional CV risk factors.^6,22

We used 3 well-established techniques, widely validated as representative of vascular phenotype in early atherosclerosis and linked to the occurrence of later CV events in other conditions.^23- 24 We found impairment in the endothelium-dependent vasodilator response in patients with aPL, pointing to a detrimental effect of aPL antibodies on vascular endothelial function. In addition to these functional changes, we have shown structural arterial abnormalities in aPL, with greater arterial stiffness and CIMT as previously reported.^5,25- 26

Increasing evidence suggests that oxidative stress could be a pathogenic link between aPL and accelerated atherosclerosis. We have observed that impaired vascular function was related to reduced paraoxonase activity. A rapid dose-dependent inhibition of paraoxonase activity after incubation with monoclonal antiphospholipid antibodies has recently been described, suggesting a direct interaction of antiphospholipid antibodies with paraoxonase leading to a reduced paraoxonase activity.²⁷

This concept has been further supported by a recent experimental study demonstrating that injection of aPL antibodies (anti-cardiolipin antibodies) resulted in a reduced paraoxonase activity in an experimental in vivo model.⁷ Dinu et al²⁸ and Delgado Alves et al²⁹ reported the presence of antibodies against apolipoprotein A-1 and HDL in patients positive for aPL. However, it remains unknown whether these antibodies are specifically directed against an antigen present in lipoproteins or simply cross reacting with anticardiolipin antibodies.²⁹

In our study, we were not able to detect differences in these antibodies between patients and controls. However, we were able to confirm decreased activity of paraoxonase in individuals with aPL and demonstrate a strong inverse relationship between paraoxonase and measures of arterial structure. The mechanism of decrease in serum paraoxonase activity in patients with aPL remains speculative. In addition to the above described potential mechanisms, serum paraoxonase activity is generally considered to vary in response to the consumption of paraoxonase for prevention of oxidation.⁹

Experimental studies have indicated that impaired paraoxonase activity leads to dysfunctional HDL, and paraoxonase-1 deficient mice had an accelerated development of atherosclerosis.³⁰ Vice versa, overexpression of human paraoxonase 1 resulted in a reduced atherosclerotic lesion formation in mice, further suggesting that reduced paraoxonase activity may contribute to the development of atherosclerosis.³¹ Moreover, reduced paraoxonase activity has been observed to be a substantially better predictor of CV risk compared with functional paraoxonase genetic variants in patients undergoing coronary angiography, suggesting that reduced paraoxonase activity is associated with a more rapid progression of atherosclerotic vascular disease.³² Therefore, there is evidence to suggest that aPL antibodies may impair paraoxonase activity and that altered paraoxonase activity may lead to dysfunctional HDL and promote development of atherosclerosis.

High-density lipoprotein is thought to exert part of its anti-atherogenic effect by stimulating endothelial nitric oxide production and inhibiting oxidant stress and inflammation.³³ Our study suggests that these atheroprotective effects of HDL are largely impaired in patients with aPL. Furthermore, modification of HDL properties are likely not unique to aPL syndrome but may also occur in coronary disease and acute inflammation.^34- 35

We have demonstrated that HDL has several dysfunctional characteristics in aPL, which could contribute to the vascular abnormalities we have identified. It reduces nitric oxide bioavailability in HAECs and has less antioxidant and anti-inflammatory properties when compared with HDL from controls. Although not tested in our study, a number of different mechanisms could confer pro-inflammatory characteristics to HDL, including increased oxidation in the presence of glycation products, enhanced hydrolysis by secretory phospholipase A2, as well as genetic predisposition.^36- 37

Our study represents the vascular characterization of the largest reported cohort of women with positive aPL. Our findings support a role for aPL in both the etiology of vascular disease and its consequences. We also demonstrated a higher prevalence of cellular events relevant to atherogenesis in the presence of HDL from women positive for aPL compared with that from matched controls, thus confirming the presence of dysfunctional HDL in these patients. We found an association between paraoxonase activity of HDL cholesterol and CIMT and pulse wave velocity in aPL, which suggests a possible interaction between antiphospholipid antibodies and paraoxonase, leading to reduced paraoxonase activity, a concept that was demonstrated in experimental studies.⁸ It is difficult however to establish whether the decreased paraoxonase activity and dysfunctional HDL are the cause or the consequence of the presence of aPL. For example, paraoxonase may modify phospholipids and thus contribute to the development of aPL syndrome itself.³⁸ These links between aPL, paraoxonase, and HDL are intriguing and may be specific to the syndrome. Alternatively, the inflammatory and pro-oxidant milieu, which is present in other chronic inflammatory conditions, may predispose to vascular disease by similar changes in HDL functionality.³⁹

Although the findings of our study provide pathophysiological insights into vascular dysfunction in aPL, the cross-sectional design does not permit causal relationships to be established. We showed an association between CIMT and history of thrombotic events. Nevertheless, we were not powered to examine the effect of HDL function on the clinical manifestations of the aPL syndrome. This would require a much larger study of patients with the full range of presentations rather than of participants selected on the basis of antibody levels selected as in our study.

Strategies to prevent clinical CV complications play an increasingly important role in the treatment of rheumatic diseases, including antiphospholipid syndrome. Our study suggests that targeting the pro-inflammatory and oxidative capacity of HDL merits further consideration in clinical trials, with the potential to delay CV disease evolution and reduce complications in patients who have antiphospholipid syndrome.