Viewpoint: The value of non-criteria antiphospholipid antibodies

Abstract

In 2006, at a meeting in Sydney, Australia, consensus was reached by an international group of specialists to establish a number of serological criteria that identify patients with a history of thrombosis or pregnancy complications as having antiphospholipid syndrome (APS). These criteria were originally formulated for research purposes and to compare clinical trials in different centres. However, these same criteria are now generally used and accepted for the diagnosis and treatment of patients. The practice of using these criteria for direct patient care requires that these criteria are based on sound scientific evidence. Indeed, for all the autoantibodies that are officially included in the serological criteria, it has been shown that they induce thrombosis and fetal loss when infused into mice. There are also a number of additional autoantibodies that have been identified in these patients but for these antibodies there was not enough evidence to meet the official APS criteria in 2006. Seventeen years have now passed since the consensus meeting, therefore, this review examines whether additional studies performed with these ‘non-criteria’ autoantibodies have provided sufficient results to suggest the inclusion of these autoantibodies in the official serological criteria of APS.

Introduction

A patient with thrombotic complications or pregnancy morbidity is diagnosed as having antiphospholipid syndrome (APS) if so-called antiphospholipid antibodies are persistently present in the blood [1]. The collective term antiphospholipid antibodies includes immunoglobulin G (IgG) and immunoglobulin M (IgM) antibodies against the phospholipid cardiolipin and the plasma protein β₂-glycoprotein I, as well as antibodies that prolonged clotting assays in a phospholipid-dependent manner, known as lupus anticoagulant. The correlation with these antibodies and the observed clinical manifestations suggests a causal relationship between them. Indeed, animal models of APS have shown that these antibodies are a risk factor for the development of thrombotic complications and fetal losses [2–8]. The presence of antiphospholipid antibodies is often associated with a variety of ‘non-criteria’ clinical manifestations, such as thrombocytopenia, nephropathy, cardiac valve damage and livedo reticularis, which are manifestations not specific enough to be included in the official criteria [9]. A substantial number of patients with APS also have various ‘non-criteria’ autoantibodies in their plasma. There is a wide range of antigens against which these autoantibodies are directed: β₂-glycoprotein-I domain I, annexin a5, vimentin, annexin II, prothrombin, phosphatidylserine/prothrombin, protein S, protein C, thrombomodulin, endothelial cell protein C receptor and probably others [10]. Anti-phosphatidylserine/prothrombin (aPS/PT) antibodies are the most common ones [11]. There are also publications suggesting that immunoglobulin A (IgA) autoantibodies should be included in the criteria [12]. When the criteria for APS were formulated there was not enough evidence to include these ‘non-criteria’ autoantibodies in the definition of APS.

The clinical manifestations of patients with APS are highly diverse: venous or arterial thrombosis, microangiopathy, early and late pregnancy loss. As mentioned, patients can suffer from a variety of additional complaints. It is uncertain whether one specific autoantibody can be responsible for all these different clinical manifestations. It is more likely that the variation in clinical manifestations between patients with APS is due to an individual set of different autoantibodies. There are publications suggesting that anti-β₂-glycoprotein I antibodies and anti-prothrombin antibodies have different clinical consequences [13, 14]; however, these observations have not been confirmed in larger studies. Also, autoantibodies with the same antigen specificity may display different biological complications when present in different individuals [15]. Whether non-criteria antibodies play an additional role in the development of specific (non-criteria) clinical manifestations is unknown. In this review we discuss different ‘non-criteria’ autoantibodies that have been identified in patients with APS and their potential added value for the diagnosis of the syndrome.

Promising non-criteria autoantibodies

Anti-prothrombin antibodies

Prothrombin is an essential component of the blood clotting system. It is proteolytically cleaved by factor Xa to form thrombin, the central enzyme of coagulation. Autoantibodies against prothrombin are frequently found in patients with APS [16]. Interestingly, the presence of anti-prothrombin antibodies is not part of to the official criteria to define APS, but they are part of the criteria through a back door. Anti-prothrombin antibodies can trigger lupus anticoagulant, which is the most predictive assay to identify patients at risk for thrombotic complications and recurrence of fetal loss [17]. Indeed, anti-prothrombin antibodies can induce a more potent lupus anticoagulant than anti-β₂-glycoprotein I antibodies [18] (Fig. 1). The question to be addressed is whether anti-prothrombin antibodies without lupus anticoagulant activity are a risk factor for thrombosis and pregnancy complications. The same question can be raised for antibodies against β₂-glycoprotein I without lupus anticoagulant activity (see below). An assay against prothrombin bound to phosphatidylserine has been developed [19]. This assay correlates better with clinical manifestations than an assay against prothrombin alone, probably because binding of prothrombin to phosphatidylserine induces a conformational change in prothrombin resulting in the exposure of a hidden epitope [20, 21]. There is no sound evidence that anti-prothrombin IgM or IgA is associated with clinical manifestations. Also the value of low titre IgG is questionable and it is unknown whether the upper value of the normal range determined by measuring healthy volunteers coincides with the lower limit of the pathological range.

Prothrombin consists of four domains, an N-terminal Gla domain, two kringle domains and a serine protease domain. At least two distinct subpopulations of PS/PT have been identified that recognize different parts of prothrombin [22], but there are likely more. It is essential to conduct a study correlating the clinical manifestations that characterize APS with the different subpopulations of autoantibodies against prothrombin. Anti-prothrombin antibodies have been tested in a mouse and rat models and an increased risk of thrombosis was found [23–25]. In a more recent study Chayoua et al. demonstrated that only anti-prothrombin antibodies with lupus anticoagulant activity can induce platelet aggregation via the FcγRIIA receptor on platelets [26]. Unfortunately, mice lack this receptor, so the increased risk of thrombosis observed in mice after infusion of anti-prothrombin antibodies cannot be explained by the activation of platelets via this receptor.

There are no major obstacles to including anti-prothrombin IgG antibodies in the official criteria that identify patients with APS. In our opinion, the aPS/PT assay is ready to be included in the official classification criteria as a biomarker for APS. We already have lupus anticoagulant to identify a pathogenic subpopulation of these autoantibodies. It would be helpful if we had a study showing that the aPS/PT enzyme-linked immunosorbent assay (ELISA) preferentially detects plasmas that are lupus anticoagulant positive. Prospective multicentre studies would benefit tremendously if we could use an ELISA as a surrogate for a coagulation test. This would help those laboratories that have problems with lupus anticoauglant testing. Not only is it easier for most laboratories to perform an ELISA, but it also opens the possibility of using serum instead of plasma. Furthermore, the use of an aPS/PT ELISA to detect pathogenic anti-prothrombin antibodies may help to overcome the challenges in lupus anticoagulant testing in patients on long-term vitamin K antagonists.

Anti-domain I antibodies

Plasma protein β₂-glycoprotein I consists of five repeating sushi domains [27]. The fifth domain contains a cluster of lysine residues responsible for binding to negatively charged phospholipids [28]. There is very strong evidence from animal models and patient studies that domain I contains an epitope for autoantibodies responsible for the risk of thrombotic complications [29, 30]. Autoantibodies to each domain of β₂-glycoprotein I have been described [31], but there are no experiments showing the thrombogenicity of antibodies against other domains than domain I [32]. Antibodies against the fifth domain, which are found in patients with leprosy, are not related to thrombosis or pregnancy complications [33]. Antibodies against the combination of the fourth and fifth domains tend to be protective against thrombotic complications caused by domain I antibodies [34].

The question arises whether the assay for anti-β₂-glycoprotein I antibodies should be replaced by an assay for detection of domain I antibodies [30]. Studies with large APS cohorts comparing both assays suggest that the specificity of domain I autoantibodies is much better, but their sensitivity is lower [35–38]. Antibodies against domain I induce thrombosis in a mouse model of APS [39]. In this respect it is interesting to note that isolated domain I can be used as an anti-thrombotic agent in a mouse model of APS [40]. In the absence of studies showing that autoantibodies to other domains of β₂-glycoprotein I are pathogenic, it is recommended to add the anti-domain I antibody ELISA to the official criteria. This would make the diagnosis of APS more specific and identify individuals at high risk for developing thromboembolic events [41, 42]. Nevertheless, the set-up of the ELISA for domain I antibodies should meet strict criteria. Domain I binds to high affinity ELISA trays with its critical epitope towards the surface of the tray, and consequently the critical epitope is badly exposed. Hydrophobic ELISA trays are the plates of choice [30].

Antibodies against oxidized low-density lipoprotein–β₂-glycoprotein I complexes

Autoantibodies against oxidized low-density lipoprotein–β₂-glycoprotein I complexes have been detected in patients with systemic lupus erythematosus (SLE) with or without APS. These antibodies are significantly associated with arterial thrombosis [43]. The levels of these antibodies are higher in SLE patients with APS than in patients without APS. These antibodies are thought to have a pro-atherogenic role in the development of autoimmune vascular complications. The presence of these autoantibodies indicates significant vascular damage. Commercial assays for these autoantibodies are available and may be helpful in predicting progressive atherosclerosis in patients with APS.

Autoantibodies lacking convincing clinical significance

Anti-annexin a5 antibodies

Annexin a5 is a cellular protein with no known function. Mice lacking annexin a5 are fertile and have no significant metabolic or functional defects [44]. Annexin a5 has a strong affinity for negatively charged phospholipids in a Ca²⁺-dependent manner. Annexin a5 forms a protective shield on negatively charged phospholipids thereby preventing clotting factor binding and inhibiting phospholipase activity [45]. Autoantibodies against annexin a5 can be found in some patients with APS [46], and an important role of annexin a5 in the pathogenesis of APS has been proposed [47]. It has been postulated that antiphospholipid antibodies prevent annexin a5 from performing its function as an inhibitor of coagulation. Loss of annexin a5 binding to trophoblasts was also considered to be an important mediator in impaired formation of placentas [48]. An assay has been developed to detect the inhibitory capacity of annexin a5, and it has been shown that patients with APS have lost this inhibitory effect [49]. This loss of sensitivity to annexin a5 correlates well with the presence of a lupus anticoagulant induced by domain I antibodies [50]. However, loss of inhibitory effect of anti-annexin a5 activity has not been associated with the presence of autoantibodies to annexin a5. A literature survey of anti-annexin a5 autoantibodies found no association with thrombosis or miscarriage [51]. Although recent studies suggest that measurement of anti-annexin a5 IgM may be useful for detecting the risk of miscarriage [52], the inclusion of anti-annexin a5 autoantibodies would not improve the diagnosis of APS.

Autoantibodies against annexin II

Annexin II, also known as annexin a2, is a Ca²⁺-dependent phospholipid-binding protein that plays a role in regulation of cell growth and signal transduction pathways. Annexin II has been described as a receptor for the β₂-glycoprotein-I–autoantibody complex on endothelial cells. Annexin II also mediates plasminogen and tissue plasminogen activator formation on endothelial cells, and supports tissue-based fibrinolysis. Blocking of annexin II by the β₂-glycoprotein-I–autoantibody complex is thought to be involved in impaired fibrinolysis observed in patients with APS [53]. Antibodies against annexin II can be found in serum from patients with systemic autoimmune diseases. The clinical significance of anti-annexin II antibodies and their diagnostic value in distinguishing clinical subgroups of patients with APS is unknown [54].

Autoantibodies against the protein C axis

Protein C is a vitamin K-dependent protein that circulates in plasma. It is activated by thrombin to form activated protein C (APC). The activation by thrombin is accelerated >1000 times when thrombin is bound to thrombomodulin, a protein displayed on the surface of the endothelium. APC inactivates factors Va and VIIIa thereby inhibiting coagulation. The inactivation of these cofactors of coagulation by APC is accelerated by the presence of protein S [55]. Partial deficiencies of protein C and S are a risk factor for venous thrombosis. Endothelial cell protein C receptor (EPCR) was identified as a cellular receptor for protein C on endothelial cells [56]. EPCR has an essential role in protein C activation by the thrombin–thrombomodulin complex in larger vessels of the body. Mutations in its gene have been associated with venous thromboembolism and myocardial infarction, as well as with late fetal loss during pregnancy [57]. A polymorphism in the gene for factor V (R506Q) has been described that makes factor V less sensitive to inactivation by APC. The presence of this polymorphism in factor V is known as APC resistance [58]; however, APC resistance can occur without the factor V polymorphism. The presence of APC resistance is a strong risk factor for venous thrombosis.

Patients with APS and lupus anticoagulant do express APC resistance [59] in the absence of factor V polymorphism. Lupus anticoagulants are a heterogenous group of antiphospholipid antibodies that interfere with clotting factors by bind to negatively charged phospholipids thereby inhibiting coagulation [60]. APC also needs negatively charged phospholipids as a cofactor for its activity towards factors Va and VIIIa, so it is not surprising that lupus anticoagulant also inhibits APC activity and induces APC resistance. The presence of APC resistance in APS patients could contribute to their increased risk for venous thrombosis.

The presence of autoantibodies against protein C, protein S and thrombomodulin in patients with APS has been found [61, 62]. The relevance of these autoantibodies is uncertain. Children can develop thrombotic microangiopathy after a varicella infection due to the presence of autoantibodies against protein S [63]. It is conceivable that anti-protein S antibodies can have comparable effects in patient with APS. Whether autoantibodies against protein C or thrombomodulin can also induce thrombotic complication is unknown. A study in which the presence of these antibodies is correlated to thrombotic complications in patients with APS and in particularly with thrombotic microangiopathy is an essential step forward. Studies in mice showing that these autoantibodies can induce thrombosis or fetal loss are necessary before these antibodies can be considered as part of the APS criteria.

Autoantibodies against EPCR were described in 2004 and identified as a risk factor for fetal death [64]. Later on it was found that these autoantibodies are directed against the complex between lysobisphosphatidic acid and EPCR [6]. Originally, these ‘complex’ autoantibodies were described as autoantibodies against cofactor-independent cardiolipin [65]. Apparently, cross-reactivity of autoantibodies against different, closely related antigens makes it difficult to identify the major antigen for the antibodies. In a mouse model of APS these ‘complex’ antibodies induce thrombotic complications [6]. Years ago, lysobisphosphatidic acid was already identified as an important phospholipid involved in APS [66]. Only one article is available on the presence of these anti-EPCR–lysobisphosphatidic autoantibodies in a cohort of APS patients [67]. While animal models suggest the potential value of these antibodies as additional biomarker for APS, it is important that different research groups investigate the presence of these autoantibodies in a large cohort of patients with APS. Positive results in different laboratories are necessary to add this assay to the criteria of APS.

Antiphospholipid antibodies (excluding cardiolipin)

Antibodies directed against phosphatidylethanolamine (PE), a neutral phospholipid, are frequently reported as the sole autoantibodies found in patients with thrombotic manifestations or pregnancy loss [68–70]. Similar to the relevant antibodies against cardiolipin, cofactors for a positive antibody signal towards PE have been suggested, in particular proteins of the contact activation system [71]. A comprehensive review of antibodies against PE in patients with thromboembolic events and pregnancy morbidity has been published [72]. No additional diagnostic value was found for the diagnosis of APS patients. The significance of these antibodies for the diagnosis is doubtful, because the majority of patients with these antibodies lack all other antiphospholipid antibodies [73].

Thin-layer chromatography of a mixture of phospholipids followed by immunostaining revealed autoantibodies against phosphatidic acid, phosphatidylserine and phosphatidylinositol [10]. The value of these antibodies is questionable and the technique to detect them is too difficult to be used in a routine diagnostic environment.

Anti-neutrophil extracellular trap antibodies

The release of neutrophil extracellular traps (NETs) by hyperactive neutrophils has recently been recognized as an alternative to coagulation or platelets as a cause of thrombotic complications. Recently, autoantibodies to NETS have been described in patients with APS. These autoantibodies were able to inhibit the degradation of NETs by patient sera [74]. The presence of these autoantibodies has been associated with venous thrombosis and activation of the complement system. The same antibodies were also found in many patients suffering from COVID-19 infections [75]. Are these anti-NETs antibodies primary antibodies present in patients with APS or secondary antibodies that respond after NETs are formed? Further studies in larger cohorts should answer this question.

Anti-TFPI antibodies

Tissue factor pathway inhibitor (TFPI) is an inhibitor of factors Xa and VIIa. TFPI binds factor Xa and the complex of TFPI–Xa subsequently binds to the VIIa–tissue factor complex, inhibiting thrombin formation [76]. High levels of TFPI correlate with a bleeding tendency [70] whereas low TFPI levels pose a risk of thrombosis [77]. Initially anti-β₂-glycoprotein I antibodies were thought to neutralize the activity of TFPI [78] whereas later autoantibodies against TFPI were found in plasma from patients with APS [79, 80]. The autoantibodies were present in approximately one-third of the APS patients. The autoantibodies were directed against the positively charged C-terminal part of TFPI and inhibit the function of TFPI [79]. Further studies are needed to evaluate the importance of down-regulation of TFPI function in autoantibody-induced thrombosis [81].

Antibodies against contact activation enzymes

The level of factor XII, the first enzyme of the contact activation system of coagulation, is often low in patients with APS [82]. The low levels of factor XII are thought to be the consequence of autoantibodies against factor XII [83]. In a few studies, these autoantibodies have been associated with recurrent embryonic loss [84, 85]. Autoantibodies directed against factor XI in combination with PE have also been described [86]. No further studies have been published in recent years. Whether these antibodies have any additional value for the diagnosis of APS in pregnant women is questionable.

Anti-fibrinolytic antibodies

Low fibrinolytic activity is correlated with an increased risk of thrombosis [87]. In some individual APS patients, autoantibodies against tissue plasminogen activator have been described that are directed against its catalytic domain [88]. It might be possible that in individual cases autoantibodies that interfere with the fibrinolytic system support the higher risk of venous and arterial thrombosis.

Anti-vimentin antibodies

Serum-negative APS refers to patients with clinical symptoms consistent with APS but without autoantibodies in their plasma that belong to the official criteria. A significant number of these patients have antibodies to the complex of vimentin and cardiolipin in their plasma [89]. Vimentin is a structural protein found predominantly in the cytoplasm but it is also secreted or present at the surface of different cells, including endothelial cells and macrophages. As a cytoplasmic protein, vimentin is involved in immune-inflammatory responses [90]. Autoantibodies against vimentin could affect the immune response. A recent study found a higher prevalence of arterial thrombosis in patients testing positive for anti-vimentin/cardiolipin IgA, in both serum-negative and serum-positive APS patients [91]. This study makes the role of these antibodies more complex, because IgA is not able to activate the complement system through the classic pathway, and complement activation seems to be an integrated part of the response to antiphospholipid antibodies [92]. Further studies are needed to understand these interesting antibodies.

IgA

The added value of IgA anti-β₂glycoprotein I antibodies and IgA cardiolipin antibodies for the diagnosis of APS is controversial and a matter of active debate [93]. There are different studies showing that the presence of IgA is strongly correlated with clinical manifestations of APS. However, the presence of IgA is associated with the presence of IgG or IgM, and isolated IgA positivity was rare and not associated with thrombosis or pregnancy morbidity. These data do not support testing for anti-cardiolipin IgA and anti-β₂-glycoprotein I IgA subsequent to the present official serological criteria in identifying patients with thrombosis or pregnancy morbidity [94, 95].

Numerous animal models have shown that IgG antibodies against prothrombin and β₂-glycoprotein I are pathogenic [96]. There are two old publications in which IgM isolated from a patient with APS was infused into mice. One group found a reduced number of fetuses after breeding, and the other found an increased thrombotic risk [97, 98]. No studies were performed in which affinity purified IgA was used in a murine model of APS. Further studies are needed to support a clinically significant role for IgM and IgA in the diagnosis of APS.

Conclusions

Many different autoantibodies have been identified in plasmas of APS patients. Only antibodies against cardiolipin and β₂-glycoprotein I have been accepted as official criteria for the diagnosis of APS. In addition to these two autoantibodies, lupus anticoagulant has been recognized as a laboratory criterion. Lupus anticoagulant is often caused by antibodies towards prothrombin, so there are no arguments to prevent adding anti-prothrombin antibodies to the official criteria for the diagnosis of APS. We do not know whether antibodies against every domain of prothrombin are pathogenic but the same is true for anti-β₂-glycoprotein I antibodies. Epitope characterization is another step to improve the value of the assays. It would be very interesting to identify the pathogenic epitope in prothrombin. There are good arguments to replace anti-β₂-glycoprotein I antibodies with domain I antibodies. This would make the diagnosis more specific. The clinical relevance of all other autoantibodies is unclear. A major problem is that the assays described in this review are poorly standardized, making the comparison between different clinical studies difficult. In addition, the autoantibodies described in the older studies were often incompletely characterized. It would be of interest to isolate specific autoantibodies and test their pathogenicity in a mouse model. This has been done for all autoantibodies accepted for the serological criteria.

An important question is whether the heterogeneity of clinical manifestation between individual patients can be explained by the variation of the pattern of circulating autoantibodies in each individual patient. The various compilations of autoantibodies and the many distinctive clinical manifestations might require different treatment. At present, we have no information to support this hypothesis. The pathogenesis of APS is very complex. Activation of coagulation and complement cascades, monocytes, neutrophils, syncytiotrophoblast endothelial cells and platelets are all involved. Autoantibodies can interfere with all these physiological processes. To correlate the presence of certain autoantibodies with a specific clinical manifestation will be difficult because every patient has a different clinical expression and some of the clinical manifestations are relatively rare.

Combined efforts among basic scientists and clinicians are essential to categorize APS disease manifestation and reliable assays for the different autoantibodies. It should be one of the goals of the Anti-Phospholipid Syndrome Alliance for Clinical Trials and International Networking (APS Action), the international research network established to design and conduct large-scale, multicentre clinical trials in antiphospholipid antibody-positive patient [99]. They are the only consortium that can include enough well-defined patients to answer this complex question.

Data availability

The data underlying this article are available in the article.

Funding

No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: B.dL. and R.G. are employees of Synapse Research Institute, part of the Diagnostica Stago^® group. P.G.dG. declares that this review was written in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References