Antiphospholipid Antibodies Predict Progression of Abdominal Aortic Aneurysms

Antiphospholipid antibodies (aPLs) frequently occur in autoimmune and cardiovascular diseases and correlate with a worse clinical outcome. In the present study, we evaluated the association between antiphospholipid antibodies (aPLs), markers of inflammation, disease progression and the presence of an intra-aneurysmal thrombus in abdominal aortic aneurysm (AAA) patients. APLs ELISAs were performed in frozen serum samples of 96 consecutive AAA patients and 48 healthy controls yielding positive test results in 13 patients (13.5%) and 3 controls (6.3%; n.s.). Nine of the 13 aPL-positive AAA patients underwent a second antibody testing >12 weeks apart revealing a positive result in 6 cases. APL-positive patients had increased levels of inflammatory markers compared to aPL-negative patients. Disease progression was defined as an increase of the AAA diameter >0.5 cm/year measured by sonography. Follow-up was performed in 69 patients identifying 41 (59.4%) patients with progressive disease. Performing multipredictor logistic regression analysis adjusting for classical AAA risk factors as confounders, the presence of aPLs at baseline revealed an odds ratio of 9.4 (95% CI 1.0–86.8, p = 0.049) to predict AAA progression. Fifty-five patients underwent a computed tomography in addition to ultrasound assessment indicating intra-aneurysmal thrombus formation in 82.3%. Median thrombus volume was 46.7 cm3 (1.9–377.5). AAA diameter correlated with the size of the intra-aneurysmal thrombus (corrcoeff = 0.721, p<0.001), however neither the presence nor the size of the intra-aneurysmal thrombus were related to the presence of aPLs. In conclusion, the presence of aPLs is associated with elevated levels of inflammatory markers and is an independent predictor of progressive disease in AAA patients.


Introduction
Antiphospholipid antibodies (aPL) are a group of heterogenous autoantibodies associated with spontaneous thrombus formation or pregnancy complications in patients with a disease recognized as antiphospholipid syndrome (APS) [1,2]. APLs frequently occur in autoimmune and cardiovascular diseases, correlating with a worse clinical outcome of affected patients [3,4]. The mechanisms leading to the evolvement of aPLs and their functional relevance in vascular diseases are still incompletely understood. Recent experiments indicated that aPLs inhibit the activated protein C pathway, lead to abnormalities in platelet function, up-regulate the tissue factor pathway and cause endothelial dysfunction collectively promoting aberrant thrombus formation and vascular damage [5].
Abdominal aortic aneurysms (AAA) are a common vascular disease with a prevalence of 3% in individuals aged 60 years or older. The pathogenesis of this disease appears to be complex, and immune-mediated mechanisms resulting in the activation of matrix metalloproteinases (MMP) with subsequent disruption of the orderly lamellar structure of the aortic media and tissue degradation play fundamental roles [6]. B-cells occur in the adventitia of AAAs [7] and pro-inflammatory CD4 + and CD8 + Tcells lacking the co-stimulatory molecule CD28 are enriched in peripheral blood and tissue specimens of AAA patients indicating the involvement of adaptive immune-responses in the pathogenesis of the disease [8].
It is well established that intra-aneurysmal laminated thrombi fill the lumina of AAAs to varying extents either covering the entire wall of AAAs or being eccentrically located, leaving part of the aneurysm wall exposed to blood flow. The growth of the intraaneurysmal thrombus is associated with both, aneurysm progression and rupture [9,10]. Intra-aneurysmal thrombi affect the underlying aortic vessel wall leading to chemotaxis of inflammatory cells, adsorption of plasma components and induction of apoptosis in smooth muscle cells [11]. Thus, the intra-aneurysmal thrombus functions as a site of protease release and activation, with subsequent degradation of the extracellular matrix [12].
Previous works reported the development of vascular aneurysms in patients with APS [13]. Given the known association of aPLs with immune-mediated and cardiovascular diseases as well as the evidence for immune-activation in AAA patients, we demonstrate in the Innsbruck AAA study cohort, that aPLs are associated with increased serological markers of inflammation and predict progressive disease.

Statement of the ethics committee
The ethics committee of the Innsbruck Medical University specifically approved this study during their session number 226/ 4.8, 289/5.6 (2618a), study number AM2249 on May 9, 2005 and most recently on May 5, 2010. Informed and written consent was obtained from each subject.

Patients
All AAA patients included in this study are participants of a prospective investigation assessing the role of inflammatory biomarkers and immunocompetent cells in the pathogenesis of AAAs. Out of this AAA study cohort (n = 135) we randomly enrolled 96 AAA patients with an AAA diameter larger than 3 cm (Patients' characteristics and risk factors are summarized in Tables 1 and 2, respectively) as well as 48 healthy volunteers. We excluded neoplasms, chronic inflammatory diseases as well as infections by full medical history, clinical examination and routine blood tests in all subjects. Besides, none of the controls had a history of fetal loss, thrombotic events, thrombocytopenia, livedo reticularis or AAA.

Sonography and serial computed tomography volume measurements
Each AAA patient underwent ultrasound investigations of the abdominal aorta by an experienced vascular surgeon using a 2-5 MHz convex transducer (Toshiba Diagnostic Ultrasound System, Model SSA-660A). Sonographic determination of the aortic diameter was performed as part of clinical routine and disease progression was defined as an increase of the diameter by .0.5 cm per year [14].
A computed tomography (CT)-angiography was routinely performed in case of new diagnosis of AAA, difficulties with reliable measurement of the aortic diameter by sonography, suspected progression of AAA diameter above the threshold for surgical intervention and/or a follow-up of .5 years. Fifty-five (57.3%) AAA patients underwent a CT examination with a multislice CT-scanner (LightSpeed QX/i, GE Healthcare, Milwaukee, USA). After a native scan, depending on the patient's weight, up to 150 ml of a non-ionic contrast agent (Ultravist 370, Bayer HealthCare, Germany) were applied by means of an automatic injector using the SmartPrep software (GE Healthcare) with a flow rate of 4 ml/s. CT-scans were performed during the early arterial and the delayed phase with a slice thickness of 2.5 mm and 5 mm, respectively. Using the axial source images, multiplanar reformations, volume rendering reconstruction and maximum intensity projections as well as volumetric measurements were performed on a postprocessing system (Advantage Windows Ultra Spark 10 Work Station 3.1, Sun Microsystems, Mountain View, CA, USA). Volumes were assessed using the summation area technique, outlining the volume of interest on each CT with the cursor. The postprocessing system automatically determined boundaries around a class of similar voxel intensity values, and the software calculated the volume of all marked areas.
The following 3 volumes of interest were defined semiautomatically: the contrast-filled intra-aneurysm vascular lumen, the total aneurysm volume of the aneurysm sac and the volume of the aneurysm thrombus. The total aneurysm volume included the volume of the aneurysm thrombus and lumen. The volume of the aneurysm thrombus was calculated by subtracting the intraaneurysm vascular channel from the total aneurysm volume [15]. Volume analysis started at the level of the coeliac trunc and ended at the aortic bifurcation.

Inflammatory parameters and three color immunofluorescence flow cytometry
For neopterin measurement the automated version of a commercially available ELISA (ELItest Neopterin, BRAHMS, Berlin, Germany) was used according to the manufacturer's instructions (interassay coefficient of variation: 6.54%, limits of detection without dilution: between 2 and 50 nmol/L). Erythrocyte sedimentation rate (ESR) was determined by the Westergren method.

Statistics
Statistical analyses were performed using the SPSS program, version 21.0 (Chicago, IL, USA). The Kolmogorov-Smirnov test was used to test for normal distribution of metric parameters.
Normally distributed values are expressed as mean 6 standard deviation (SD) and were compared by the Student's t test. In case of non-normally distributed data the median and range are given and the Mann-Whitney-U test was applied. Correlations were assessed by the Spearman rank correlation test. Proportions were calculated by the Chi-square or Fisher's exact tests as appropriate. We performed multivariate logistic regression analysis (maximum likelihood method) to investigate the association between aPLs and AAA disease progression adjusting for possible confounders (age, sex, history of smoking, arterial hypertension, hyperlipidaemia, diabetes mellitus, coronary heart disease, peripheral arterial occlusive disease and AAA diameter). A stepwise backward regression model was conducted excluding parameters failing the pre-specified significance level of 0.1. Goodness of fit was tested using Hosmer-Lemeshow statistics and sensitivity analyses were assessed excluding influential points as well as cases with large Cook and/or DFBETA values. p,0.05 was considered as significant.

Baseline examinations of antiphospholipid antibodies
Results of aPL testing including antibody subtypes are detailed in Table 3. The prevalence of aPLs seemed to be higher in AAA patients (13.5%) than in healthy controls (6.3%), however, the difference did not reach statistical difference (p = 0.189, using the Chi-square test). The presence of aPLs was not related to sex, neither in AAA patients nor in healthy controls.
Medium or high serum titers of antibodies directed against CL or b2GPI were found in 3 (3%) AAA patients and in 1 (2.1%, p. 0.2) healthy control [2]. Nine of the 13 AAA patients with a positive aPL result underwent a second blood test .12 weeks later. Six (66.7%) patients showed persistent aPL antibodies (Table 4).

Antiphospholipid antibodies and AAA disease progression
Clinical and sonographic follow-up examinations were performed in 69 patients (including 10/13 aPL positive AAA patients) over a median of 46.5 months (5-72). Of these 69 patients, 28 (40.6%) had stable and 41 (59.4%) progressive disease as measured by ultrasound and defined as an increase of more than 0.5 cm/ year [9]. Those  We performed step-wise backward logistic regression analysis with AAA progression as the dependent variable, aPLs as the

Antiphospholipid antibodies and intra-aneurysmal thrombus formation
Of 96 AAA patients, 79 (82.3%) presented with an intraaneurysmal thrombus (example given in Figure 2A). The median volume of the thrombus measured by CT was 46.7 cm 3 (1.9-377.5). AAA patients presenting with an intra-aneurysmal thrombus revealed larger AAA diameters [4.5 cm (3-8.7)] compared to those patients without intra-aneurysmal thrombus [3.6 (3-6) (p = 0.003, according to the Mann-Whitney-U test)] as outlined in Figure 2B. The AAA diameter correlated with the size of the intra-aneurysmal thrombus (corr coeff = 0.721, p,0.001, Figure 2C). The presence of an intra-aneurysmal thrombus, however, was not related to AAA progression as determined by Fisher's exact test.

Discussion
This study revealed higher serological signs of inflammation in aPL-positive compared to aPL-negative AAA patients and identified aPLs as a predictor of progressive disease.
All except one aPL-positive patients suffered from progressive disease (90%), whereas only 54.2% of aPL-negative patients showed AAA progression. Multivariate logistic regression analysis revealed aPLs, baseline AAA diameter as well as peripheral arterial occlusive disease as predictors for AAA progression, whereas other traditional risk factors failed the predefined significance level. The most likely explanation for this result is the limited power of our regression analysis to detect weaker  associations between predictors and the dependent variable given the relatively low number of cases and the high number of items included. However, an inclusive model forcing all traditional risk factors into the analysis model (regardless of their significance level) did not alter the association between aPLs and AAA progression thus proving the validity of our step-wise model. An association between aPLs and poor outcome of immunemediated and cardiovascular disease is well established: In systemic lupus erythematosus patients for example, the aPL IgG mediated production of tissue factor and pro-inflammatory cytokines by peripheral blood mononuclear cells contributed to arteriosclerosis [16]. Besides, aPL positive systemic lupus erythematosus patients had an increased mortality rate [17] and patients with giant cell arteritis and aCL antibodies suffered from recurrent flares [18]. Although AAA has been considered as a degenerative disease for many years, recent work suggests an important role of immune-mediation for the pathogenesis of AAAs. In line with this new concept, a case series on 4 APS patients with coexisting AAA reported an early and rapid progression of disease with an AAAgrowth rate of .1.5 cm within 5 months in one of them [13]. Also in patients with cardiovascular diseases, aPLs predicted first myocardial infarction [19] or recurrent stroke [20].
The prevalence of aPLs in 13.5% of AAA patients was consistent with the prevalence of these antibodies in other immune-mediated diseases such as systemic lupus erythematosus (12-30% of patients are positive for CL antibodies), early rheumatoid arthritis (15.7%) and giant cell arteritis (25.5%) [21][22][23][24]. In patients with primary and secondary APS, particularly in cases with underlying vasculitis, an association between aPLs and the development of vascular aneurysms was previously reported [13,[25][26][27][28][29]. In young healthy subjects aPLs were found in 1-5% of cases [21], and with advancing age these autoantibodies are more frequently observed as indicated by a prevalence of 6.3% in our healthy elderly cohort [21,30].
From the pathophysiological perspective, our findings further support the concept of immune-mediated processes contributing to the development and progression of AAA disease. We previously demonstrated that early aged pro-inflammatory T-cells are present in the peripheral blood and tissue specimens of AAA patients [8], and the present study indicates an involvement of B-cells via abnormal production of aPLs in the pathogenesis of the disease as well. The effect of aPLs on AAA progression possibly results from aPL-mediated enhanced MMP-9 activity [31] and/or aPLinduced acceleration of elastin degradation with consecutive remodelling of the arterial wall [32].
As previously reported, we confirmed an association between AAA diameter and the volume of the intra-aneurysmal thrombus [33][34][35]. The role of the intra-aneurysmal thrombus for AAA progression, however, is still controversial: several studies demonstrated that the size of an intra-aneurysmal thrombus is related to AAA progression and/or rupture [12,[36][37][38]. The intra-aneurysmal thrombus may function as a site of protease release and activation [12]. Besides, it may promote oxygen derived free radical development thus contributing to the destruction of the media and leading to AAA progression [38]. In contrast, in our cohort we found no association between thrombus formation and aneurysm growth. This finding is in line with others showing that a thrombus may even reduce shear-stress imposed on the AAA wall [39][40][41]. In summary, the factors determining whether an intraaneurysmal thrombus increases the risk for AAA progression or protects against deterioration of the disease are still unknown [35]. (C) Peripheral levels of CD4 + CD28 2 , but not of (D) CD8 + CD28 2 T-cells as markers of chronic inflammation were increased in aPL-positive AAA patients compared to aPL-negative AAA patients (p = 0.037). Whiskers box plots show 50% of cases within the boxes and all data excluding mavericks between the end-points of the whiskers (lines). P-values ,0.05 were considered as significant (using the Mann-Whitney U test). doi:10.1371/journal.pone.0099302.g001 The limitations of our study are the following: First, we cannot exclude a type II error due to the small sample size of aPL-positive AAA patients (post-hoc power calculation: 28.5%), and the differences between AAA patients and controls concerning age and sex reduced the power to detect a higher aPL prevalence in the former compared to the latter group. The focus of this study, however, was not a prevalence analysis; we rather aimed to examine the possible value of aPLs as a marker of immune activation and as a risk factor for AAA progression. Second, we determined CL-and b2GPI-antibodies by an aPL screen kit, but did not investigate the lupus anticoagulant, which is required for complete evaluation of APS [2]. Third, none of our controls had a history of AAA, however, we did not perform an ultrasound examination of the abdominal aorta to exclude this disease. Fourth, data on current medication of AAA patients were not available and the effects of drugs with pleiotropic effects on the immune system and/or the aortic wall such as statins, angiotensin converting enzyme and angiotensin receptor inhibitors could not be evaluated [42,43].
In conclusion, aPLs occurring in AAA patients are associated with increased serological markers of inflammation and predicted progressive disease. These findings further support the concept of possibly underlying immune-mechanisms in AAA development. No association was observed between aPLs and the presence or size of an intra-aneurysmal thrombus. The presence of an intra-aneurysmal thrombus was associated with larger AAA diameters as assessed by the Mann-Whitney U test. Whiskers box plots show 50% of cases within the boxes and all data excluding mavericks between the end-points of the whiskers (lines). (C) AAA diameters were associated with the volume of the intra-aneurysmal thrombus ranging from 1.9 to 377.5 cm 3 , evaluated by the Spearman-Rho test (corr coeff = 0.721, p,0.001). P-values ,0.05 were considered as significant. doi:10.1371/journal.pone.0099302.g002