Anti-endothelial cell antibodies in rheumatic heart disease

Abstract

To evaluate the anti-endothelial cell antibodies (AECA), anti-cardiolipin antibodies (aCL) and serum mannose-binding lectin (MBL) profiles of a large cohort of Yemeni patients with rheumatic heart disease (RHD) and to correlate these findings with clinical features of the disease. Patients (n = 140) were recruited from Al-Thawra Hospital in Sana'a, Yemen. All had RHD diagnosed according to modified Jones' criteria. We also studied 140 sex- and age-matched healthy blood donors from the same area. Echocardiography was performed according to the recommendations of the American Society of Echocardiography. Solid phase enzyme-linked immunosorbent assays (ELISAs) were used to measure AECA and aCL titres and serum MBL levels. Forty per cent of the patients were AECA-positive, but only 7·8% were positive for aCL antibodies. Serum MBL levels were significantly lower in the RHD group (median 4221 ng/ml versus 5166 ng/ml in healthy controls). AECA titres were correlated positively with patient age, duration of RHD and the severity of aortic stenosis, as determined by echocardiographic findings. In several autoimmune rheumatic diseases, such as systemic lupus erythematosus, vasculitis and scleroderma, AECA have been shown to play pathogenic roles by producing proinflammatory and procoagulant effects (increased expression of adhesion molecules and tissue factors, increased cytokine release) in endothelial cells. In RHD, these autoantibodies might represent a pathological link between activation of the valvular endothelium and valvular damage.

Introduction

Rheumatic fever (RF) is an autoimmune disease triggered in susceptible individuals by untreated group A streptococcal (GAS) pharyngitis. The major clinical manifestations of RF are polyarthritis, carditis, chorea, erythema marginatum and/or subcutaneous nodules [1]. The clinical presentation generally consists of one or more acute episodes. In 30–50% of all cases these lead to progressive/permanent damage to the cardiac valves, known as rheumatic heart disease (RHD) [2].

In industrialized countries the burden of RF declined substantially during the 20th century, mainly as a result of improved living standards and greater access to medical care. However, RF and RHD are still major public health problems in developing countries, where they represent the most common cause of cardiac mortality in children and adults under 40 years of age [3]. In Cambodia and Mozambique, systematic echocardiographic screening has revealed a prevalence of RHD that is roughly 10 times higher than that based on clinical screening with echocardiographic confirmation of suspected cases [4].

RF and RHD are the prototypes of autoimmune disease caused by the presence of epitopes shared by streptococcal and human antigens, the so-called molecular mimicry process [5]. RF may thus be considered one of the most convincing models of human post-infectious autoimmune disease.

Anti-endothelial cell autoantibodies (AECA) have been implicated causally in several autoimmune cardiovascular diseases [6–10], and several lines of evidence suggest strongly that the valve surface endothelium may be an important infiltration site for inflammatory cells in rheumatic carditis [11]. In genetically susceptible individuals, endothelial damage can also be caused by the activation of complement (by autoantibodies or other factors). Mannose-binding lectin (MBL), for example, can activate complement through the lectin pathway, and high serum levels of this protein seem to have proinflammatory effects in chronic diseases [12].

To our knowledge, there are no published data on the roles played in RHD by AECA and conflicting results have emerged from the studies on the possible involvement of MBL in RHD [13–16]. In the present study, we measured AECA and other autoantibody titres and serum MBL levels in large cohort of Yemeni patients with RHD.

Patients and methods

Patients and controls

The study population consisted of 140 patients who were admitted consecutively to Al-Thawrah Hospital in Sana'a, Yemen, for RHD. All had histories of RF defined according to the modified Jones criteria [1]. Patients were classified as having active disease when they met two of the major Jones criteria or one major and two minor criteria at time of enrolment; otherwise the disease was regarded as chronic. We also examined 140 sex- and age-matched controls who were registered blood donors in Yemen. Informed consent was obtained from all patients and controls in accordance with local laws.

Echocardiographic examination

Each patient was examined independently by two cardiologists experienced in the diagnosis and treatment of RHD. M-mode, two-dimensional and colour Doppler studies were performed with a 3–5-MHz sector transducer according to the recommendations of the American Society of Echocardiography. RHD was diagnosed when both examiners found unequivocal colour Doppler evidence of mitral-, aortic- or tricuspid-valve involvement documented in two planes, together with at least two of the following morphological abnormalities: restricted leaflet mobility, focal or generalized valvular thickening and abnormal subvalvular thickening. Mild valvular regurgitation was noted frequently, but in the absence of other findings it was not considered indicative of RHD. The severity of the valve disease was rated (mild, moderate, severe) on the basis of the above morphological features, transvalvular pressure gradients and the area of regurgitant jets (both measured on colour Doppler studies), and the degree of dilation observed at the levels of the mitral valve annulus, the left atrium and the left ventricle.

Routine serological studies

Complete blood counts, erythrocyte sedimentation rates (ESR), C-reactive protein (CRP) levels and anti-streptolysin-O (ASLO) titres were obtained for all patients.

Enzyme-linked immunosorbent assay (ELISA) for anti-endothelial cell antibodies

AECA of the immunoglobulin (Ig)G isotype were detected with a cell-surface ELISA on living endothelial cells (ECs) grown to confluence in microtitre plates. The human umbilical vein endothelial cells were isolated from full-term umbilical cord veins by means of collagenase perfusion, as described previously [17]. AECA titres were expressed as a binding index (BI) calculated with the following formula: 100 × (S-A)/(B-A), where S, A and B are the optical densities (ODs) of the test sample, negative control and a positive reference serum, respectively. AECA positivity was defined as a BI higher than the mean ± 2 standard deviation (s.d.) of values observed in healthy controls. This cut-off corresponds to 50% of the BI observed in a positive reference serum from a systemic lupus erythematosus patient.

ELISA for anti-cardiolipin antibodies

Duplicate serum samples from each participant were subjected to an in-house anti-cardiolipin (aCL) ELISA. Microtitre plates were coated with bovine heart cardiolipin (CL; Sigma-Aldrich, St Louis, MO, USA) in ethanol (50 µg/ml) and incubated overnight at 4°C. With this approach, CL undergoes oxidation during the coating process, as reported previously [18]. After four washes with phosphate-buffered saline (PBS), plates were blocked for 1 h at room temperature with PBS containing 10% fetal calf serum (PBS-F). After four washes with PBS-F, plates were incubated for 90 min at room temperature with serum samples diluted 1:50 in PBS-F. Anti-CL-positive serum was titrated (in triplicate) by serial dilution (1:25 to 1:1000) in PBS-F. After four washes with PBS-F, the plates were incubated for 90 min at room temperature with secondary anti-human IgG/IgM (Sigma-Aldrich) diluted 1:1000 in PBS-F. After multiple washes, paranitrophenyl phosphate in ethanolamine was applied to develop the immunoreactivity, and absorbance was measured at 405 nm in a plate reader. The OD observed for serum in wells containing no antigens was subtracted from the total OD to quantify specific binding. Samples were classified as aCL-positive when the BI exceeded the mean ± 2 s.d. of values observed in healthy controls. All positive results were confirmed by retesting with a commercially available anti-CL ELISA kit obtained from Diamedix (Miami, FL, USA).

Measurement of serum MBL levels

Serum MBL levels were measured in each patient and age-/sex-matched control with the MBL oligomer ELISA kit (Antibody Shop, Copenhagen, Denmark), in accordance with the manufacturer's instructions.

Statistical analysis

The χ² test was used for testing the significance of correlations between autoantibodies tested and clinical/echocardiographic features. Differences between groups were analysed with the Mann–Whitney U-test. The Wilcoxon signed-rank test was used for the analysis of matched pairs. The Spearman test was used for correlation analysis. A probability (P) value of 0·05 was considered significant.

Results

Clinical and echocardiographic findings

We studied 140 patients with RHD (82 females, 58 males; mean age 28 years, range 11–55 years) admitted consecutively to the Cardiology Department of Al-Thawrah Hospital (Sana'a, Yemen) between December 2006 and December 2007. Ten of the patients met the pre-established criteria for active disease. At the time of enrolment, 31 of the 140 patients had fever, five had carditis, three had articular involvement and none had skin involvement or chorea. Seventy-two patients had received secondary prophylaxis, but adherence to the regimen was judged to be satisfactory in only 28. None had received primary prophylaxis.

Eighty patients had elevated ASLO levels (> 200 UI/ml), 78 had elevated CRP levels and ESRs ranged from 1–127 mm/h (median, 31·7 mm/h).

One hundred and fourteen of the 140 patients had evidence of mitral regurgitation (severe in 53, moderate in 61); 85 had moderate (n =60) or severe (n =25) mitral stenosis; 100 had moderate (n =59) or severe (n =41) aortic regurgitation; 28 had moderate (n =14) or severe (n =14) aortic stenosis; 12 had moderate (n =2) or severe (n =10) tricuspid regurgitation; and six had moderate (n =3) or severe (n =3) tricuspid stenosis.

AECA, aCL and MBL measurements

AECA positivity (as defined in Methods) was detected in 56 (40%) of the 140 RHD patients, but only 11 (7·8%) were positive for aCL (IgM in 10 patients, IgG in one). Serum MBL levels in the patients (median: 4221 ng/ml, range 101–5337 ng/ml) were significantly lower than those of controls (median 5166 ng/ml, range 53–6224 ng/ml) (P < 0·001).

No significant differences of AECA positivity and serum MBL levels in patient grouping were observed according to echocardiographic findings, probably because the majority of our patients showed simultaneously regurge and stenosis of more than one valve (Table 1). Interestingly, all patients but one who underwent valve replacement were positive for AECA. Moreover, in the subgroup of patients with aortic stenosis significantly higher AECA titres were found only in those with severe disease (P = 0·0079), although all these patients showed simultaneously involvement of more than one valve.

Table 1.

Echocardiographic picture, anti-endothelial cells antibodies (AECA) positivity and serum mannose-binding lectin (MBL) levels of patients with rheumatic heart disease (RHD).

	n (%)	AECA-positive n (%)	MBL ng/ml median (range)
Pure mitral stenosis	11 (7·85)	5 (45·45)	4329 (5152–431)
Pure mitral regurge	6 (4·28)	1 (16·66)	3869 (4770–2481)
Mitral regurge and stenosis	12 (8·57)	5 (41·66)	4568 (4851–267)
Pure aortic regurge	1 (1·4)	0 (0)	222
Mitral and aortic regurge	22 (15·71)	8 (36·36)	4202 (5159–132)
Mitral and aortic mixed valve disease of whom	67 (47·85)	27 (19·28)	4215 (5337–106)
•Moderate aortic stenosis	11 (7·85)	4 (36·36)	4153 (4650–287)
•Severe aortic stenosis	11 (7·85)	6 (54·54)	4211 (4785–106)
Mitral and tricuspid valve disease	3 (2·14)	0 (0)	4801 (5002–714)
Mitral, aortic and tricuspid valve disease of whom	12 (8·57)	5 (41·66)	3937 (4914–101)
•Moderate aortic stenosis	1 (0·71)	0 (0)	4914
•Severe aortic stenosis	2 (1·42)	1 (0·71)	2194 (4097–290)
Post-valve replacement	6 (4·28)	5 (83·33)	4606 (5073–3192)

AECA titres were correlated positively with patient age (P = 0·001) and with disease duration (P = 0·004) (Fig. 1). Anti-CL titres showed no correlation with disease activity, age, echocardiographic findings or AECA titres.

Fig. 1.

Correlations between anti-endothelial cells antibodies (AECA) titres, age (a) and disease duration (b) in patients with rheumatic heart disease (RHD).

Discussion

This study provides the first documented evidence of the presence of AECA in patients with RHD. These antibodies have been suggested to play pathogenic roles in several diseases characterized by endothelial damage [6–10]. Their proinflammatory and procoagulant effects on ECs include the up-regulated expression of adhesion molecules and tissue factors and increased cytokine release [19]. Their presence might thus represent a link between activation of the valvular endothelial cells and consequent valve damage in RHD.

The involvement of cross-reactive antibodies in the pathogenesis of RHD is still a matter of debate, but several studies point to crucial roles for anti-myosin/anti-GlcNAc antibodies [20–23]. These autoantibodies have been found in serum samples, myocardial tissue and cardiac valves from patients with RF [24,25]. They also share cross-reactive epitopes with other human cardiac antigens, including laminin and vimentin [26]. These proteins are expressed at the level of the basement membrane of the valve surface endothelium, where they would normally be inaccessible to autoantibodies. However, these hidden antigens might conceivably be exposed by AECA-mediated endothelial stress, and their interaction with cross-reactive antibodies might then trigger an inflammatory process within the valvular tissue. AECA have been shown to stimulate the expression of adhesion molecules by human vascular ECs, thereby promoting the infiltration of cross-reactive T clones [27,28].

In this hypothetical chain of events rheumatic valve damage begins at the level of the surface endothelium, and AECA contribute to this damage directly and by promoting autoantibody reaction with basement membrane antigens. The repeated stimulation of valvular ECs also leads to the production of interferon (IFN)-γ, which results in scarring and neovascularization of the normally avascular valves [29]. Therefore, in addition to the inflammatory cells migrating across the surface endothelium, the vascularized valve is also vulnerable to infiltration by cells arriving via the newly formed intravalvular vessels. As a result, subsequent reactivation of the autoimmune cascade by GAS reinfection is likely to produce even more extensive infiltrates. The protective effects of antibiotic prophylaxis in RHD may well lie in its ability to avert such reactivation.

In our RHD patients, AECA titres displayed significant correlation with patient age and disease duration. These findings suggest that the inflammatory response may be amplified over time by the ongoing release of antigens by the activated, damaged ECs. AECA titres were also correlated with the degree of aortic valve stenosis. Flow across this valve under normal circumstances is already much more turbulent than at other sites in the heart. Therefore, the presence of stenosis at this level might be expected to have a particularly strong impact on risk of flow-related endothelial stress with exposure of new autoantigens (e.g. laminin, vimentin).

Cardiac valve lesions similar to those seen in RF have been observed in patients with anti-phospholipid syndrome (APS), suggesting that similar pathogenic mechanisms might be involved in both diseases [30]. Anti-cardiolipin antibodies are known to play key roles in APS, but conflicting data have been reported regarding their possible association with RHD. In 1988, Asherson examined a small number of European patients with RF and found no evidence at all of aCL antibodies [31]. A few years later, however, Figueroa et al. reported anti-CL positivity in 80% of Chileans with acute RF and 40% of those with chronic disease [32]. High positivity rates have also been reported in Russian RF patients, and the titres were correlated with the risk of acute endocarditis [33]. The prevalence of aCL IgG and IgM antibodies among our Yemeni RF patients was quite low, and this finding is consistent with the results of several other studies, which found no significant difference between the aCL titres of RHD patients and healthy controls [34–36]. The discrepancies among these findings may be due partly to the characteristics of the patients studied. In this study, for example, only 10 patients in our cohort had acute RF. Therefore, additional work is needed to determine whether RHD and APS share an anti-CL-related pathogenic pathway.

The relationship between autoimmune disease and infection has been a topic of interest for several decades. It is well known that RF is induced by an autoimmune response triggered by an infectious disease in genetically predisposed individuals. In fact, the development of the disease is related closely to the severity of previous GAS throat infections, which depends not only upon the virulence of the infecting strain but also upon the immune response and genetic background of the host [37,38]. Calcium-dependent lectin MBL plays important roles in innate immunity by promoting complement activation and phagocytosis [39]. Serum levels of MBL are determined genetically, and three different missense point mutations in the MBL2 gene are known to cause MBL deficiency. This is the most common inherited form of immunodeficiency, and it appears to increase the risk, severity and frequency of infections and autoimmune phenomena [40,41].

Thus far, four studies have examined the role of MBL in the development of RF and RHD. The first, reported in 2004, found MBL levels to be elevated significantly in RHD patients and a higher prevalence of MBL deficiency among healthy controls [13]. The same group suggested later that homozygosity for the wild-type MBL2 A allele, which is associated with normal circulating levels of MBL, might increase the risk of RHD, whereas variations involving MBL2 exon 1, which are associated with MBL deficiency, seemed to exert a protective effect [14]. Two years later they also found that other high-production MBL2 genotypes were more frequent in the patients with acute and chronic carditis [15]. In contrast, in the recent study conducted by Ramaswamy et al., homozygosity or compound heterozygosity for exon-1-deficient MBL2 alleles was associated with chronic severe aortic regurgitation in RHD. Because innate immunity serves mainly to limit the replication of infectious agents, these investigators reasoned that MBL deficiency would delay or impair pathogen clearance and that persistence of the infectious agent might trigger the immune response [16]. Our findings support the hypothesis that the breach of self-tolerance leading to RHD is indeed caused by continuous stimulus to the immune system. In fact, circulating MBL levels in our RHD patients were significantly lower than in healthy controls. In the previously mentioned studies the high and low levels of MBL correlated with the presence of the high and low MBL producer alleles, respectively. Similar alleles were found in patients with similar echocardiographic picture of the disease: mitral stenosis has been associated with the ‘A’ allele, which codes for high production of MBL; conversely, aortic regurgitation appeared to be related with the ‘O’ allele, which codes for low production of MBL [42]. In our work we did not find significant differences of serum MBL levels in patient grouping according to echocardiographic findings, probably because the majority of our patients showed simultaneously regurge or stenosis of more than one valve.

In conclusion, our findings strengthen the view that AECA play prominent roles in the cardiac tissue damage associated with RHD. The risk of exaggerated immune response with formation of these antibodies is increased in settings characterized by frequent and severe infections. These conditions would be favoured not only by the socioeconomic conditions in Yemen, but also by the low levels of MBL we found in Yemeni patients with RHD.

Disclosure

The authors declare no interests to disclose.