Classification criteria for antiphospholipid syndrome (APS) require IgG and IgM isotypes of the anticardiolipin antibodies (aCL), anti-β₂-glycoprotein I antibodies (aβ₂GPI), and/or the lupus anticoagulant (LA) to satisfy the laboratory criterion for disease definition [1]. However, over the past 20 years, several other “non-criteria” antiphospholipid antibodies (aPL), directed to other proteins of the coagulation cascade (i.e., prothrombin and/or phosphatidylserine–prothrombin complex), to some domains of β₂GPI, or that interfere with the anticoagulant activity of annexin A5, have been proposed [2]. In some cases, these assays detect specific subsets of pathogenic antibodies or a particular mechanism in APS. The Laboratory Diagnostics Task Force at the 14th International Congress on aPL (Rio de Janeiro, Brazil, 2013) highlighted several non-criteria assays [3]. However, there was consensus that further studies are necessary to obtain high-quality evidence defining their overall roles as risk predictors.

The task force reviewed the literature and conducted new studies between 2013 and 2016; the conclusions were presented at a special session during the 15th International Congress on aPL (www.​apsistanbul2016.​org, North Cyprus, September 2016). This paper updates our recommendations.

Many reports show the clinical utility of phosphatidylserine-dependent antiprothrombin antibodies (aPS/PT) assay in the diagnosis of APS, a conclusion of a task force at the 13th International Congress on aPL (Galveston, TX, 2010) [4] and reviewed, in an evidence-based manner, during the 14th International Congress on aPL (Rio de Janeiro, Brazil, 2013) [3]. The inclusion of aPS/PT antibodies as a laboratory criterion of APS was considered unwarranted then because of poor standardization of its assay and because reproducibility of the strong correlations between aPS/PT and APS manifestations needed confirmation in larger studies.

A recent systematic review suggests that aPS/PT does represent a strong risk factor for arterial and/or venous thrombosis [5]. A group of scientists led by Amengual and Atsumi carried out initial and validation retrospective cross-sectional multicenter studies on aPS/PT [6]. The initial retrospective study acquired data from eight centers from seven countries. Serum/plasma samples were blindly tested for IgG aPS/PT at Inova Diagnostics Inc., United States (USA), using enzyme-linked immunosorbent assay (ELISA) kits provided by two manufacturers: the QUANTA Lite™ aPS/PT IgG ELISA, a Food and Drug Administration (FDA)-approved assay from Inova, and the PS/PT ELISA kit for IgG isotype from Medical and Biological Laboratories Co. Ltd., Nagano, Japan. After completing the initial study, a validation study, using the same methodology for a new cohort of samples from five countries, was carried out.

The initial study comprised 247 subjects. A correlation was obtained with both ELISA kits for the IgG aPS/PT (r = 0.827, p < 0.001). Two hundred and four samples with concordant IgG aPS/PT results in both ELISAs were subsequently analyzed (99 APS, 58 non-APS, and 47 healthy). Immunoglobulin G aPS/PT were more prevalent in APS patients (51%) than in those without (9%), with an OR of 10.8 [95%CI 4.0–29.3], p < 0.0001. For APS diagnosis, sensitivity, specificity, and positive (LR+), and negative likelihood ratio (LR-) were 51, 91, 5.9, and 0.5%, respectively. In the validation study (n = 214), a significant correlation was found for IgG aPS/PT titers (r = 0.803, p < 0.001). Immunoglobulin G aPS/PT concordant samples were again analyzed (n = 182; 76 APS, 57 non-APS, and 49 healthy). Immunoglobulin G aPS/PT were more frequently found in APS patients (47%) than in those without (12%), with an OR of 6.4 [95%CI 2.6–16], p < 0.0001. For APS diagnosis, sensitivity, specificity, and LR+, and LR- were 47%, 88%, 3.9%, and 0.6%, respectively.

Whether to include non-criteria antibodies in the designation aPL-positive is still under discussion. Current APS classification criteria exclude patients with clinical manifestations suggestive of APS who have non-criteria antibodies, sometimes referred to as seronegative APS [7]. The above multicenter study confirms, in both cohorts, high prevalence of IgG aPS/PT in patients with definite APS and in those with APS-associated clinical manifestations in the absence of APS laboratory criteria [6]. Thus, based on the available evidence, the task force suggests the inclusion of IgG aPS/PT in the APS classification criteria.

β₂-Glycoprotein I (β₂GPI) is the main antigenic target for aPL [8]. Anti-β₂GPI may activate the endothelial cells, monocytes, and platelets, triggering the coagulation cascade by recognizing membrane- or receptor-bound β₂GPI [8–10]. Dimerization of β₂GPI or the complexing of β₂GPI with antibodies stabilizes receptor affinity, allowing cell signaling to occur [11].

Experiments in mouse models of APS show that patient-derived autoantibodies against β₂GPI increase the thrombotic risk. Remarkably, epidemiologic studies do not show a strong relation between these antibodies and thrombosis or pregnancy morbidity. Compared to the LA test, the correlation is weak. There are a number of explanations for this incongruity. Lack of standardization of the ELISA results in large differences in results obtained in sample exchange programs. Another possibility is that the ELISAs pick up irrelevant low-affinity antibodies, which lead to many positive results (false-positives) in healthy individuals. A third possibility is that the aβ₂GPI ELISA measures a heterogeneous population of antibodies, and not all autoantibodies directed against β₂GPI are a risk factor for thrombosis or fetal loss.

Immunoglobulin G and IgM aCL were first accepted as valid measures in the 1980s; IgA was not accepted because of high variability among laboratories. When consistent measurement was assured, it became apparent that, rarely, IgA aCL might be the only detectable antibody [26] and that IgA aCL occurs in up to 40% of patients with SLE [27–29]. Recent studies in SLE report a prevalence of 16 to 58% for IgA aβ₂GPI, particularly among those of African-American ethnicity [27–29], and in patients with the primary APS up to 72% [30–34]. In 1995, Pierangeli showed that IgG, IgM, and IgA aCL are pathogenic in a mouse thrombosis model [35]. Later, she showed that aβ₂GPI isolated from four APS patients with only IgA upregulated tissue factor and caused thrombosis in mice [36].

It is not yet clear if measurement of IgA aPL is useful for everyday practice. Some authors emphasize that methodological problems and lack of standardization still exist among commercial preparations and that the addition of IgA to IgG and IgM does not identify increased thrombosis risk in SLE patients.

Tincani et al. used a homemade ELISA for IgA aβ₂GPI to demonstrate positive tests in 28%, 40%, and 3% of 119 APS, 328 SLE, and 78 healthy controls (p < 0.0001 for both patient groups). In SLE patients positive IgA and IgG isotype prevalence was similar, while IgM was lower; in primary, APS IgG and IgM were the most frequent isotypes. Among patients with primary thrombotic APS, 65% of the 31 subjects with recurrent thrombotic episodes had aβ₂GPI IgA compared to 39% of the 46 with one episode (p < 0.05) [37].

The reported experience shows that the routine performance of IgA aβ₂GPI may be useful in SLE patients, as reported by others [38] but also in primary APS, where these antibodies might have a prognostic value. Furthermore, other authors suggested that IgA aβ₂GPI can be an independent risk factor for the development of the first aPL-related event, particularly arterial thrombosis [39].

The task force suggests appropriate prospective studies that will allow the evaluation of IgA antibody as a thrombosis risk factor are still needed.

The APhL assay uses a mixture of negatively charged PL antigens, phosphatidylserine (PS) and phosphatidic acid (PA), with β₂GPI (Louisville APL Diagnostics). It has high sensitivity for identifying APS patients with typical clinical manifestations and has improved specificity when disorders other than APS (e.g., infectious and autoimmune diseases), which often give false-positive aCL results, are studied [40]. The assay derives from older experiments (that antedate discovery of β₂GPI) demonstrating that serum from infectious disease patients and that from autoimmune disease patients differ in their binding to various phospholipids [41]. Sera from syphilis patients had very low affinity for PS and PA despite a high affinity for cardiolipin (CL), while sera from autoimmune patients had high affinity for CL, PS, and PA. Identification of a mixture of two negatively charged phospholipids that enabled the best distinction resulted in the creation the APhL assay in the 1990s [40]. Over 20 years, this test has been proven to identify nearly all aCL-positive APS patients (sensitive) and is usually negative for patients with infectious and other autoimmune diseases (specific) [42–46]. The original assay did not consider the role of β₂GPI; whether it performs the same in the presence and absence of β₂GPI is unknown.

An independent study from Suh-Lailam et al. [47] showed comparable sensitivity for APS between the APhL assay and the aCL assay, when infection-induced antibody was defined by populations of patients with syphilis and parvovirus B19; the specificity of the APhL was greater. The APhL assay also compared favorably with the aβ₂GPI assay in this study, with a sensitivity of 88% and specificity of 98%. (A weakness of this study is that only 16 of 101 aPL-positive patients had known APS; the remainder were drawn from samples submitted to a commercial laboratory and found to be positive.) A review in 2000, which included sera from patients with APS, leishmaniasis, leptospirosis, and syphilis, stated that the aCL assay was positive in 100% of APS sera, the APhL in 98%, and the aβ₂GPI in 74%. Specificities for identifying APS were 73% for aCL, 96% for APhL, and 70% for aβ₂GPI. The aCL and aβ₂GPI tests were more frequently positive in infectious disease sera, making them less specific than the APhL test [48]. In data presented in this study, details of the sources of infectious disease sera are not provided; specificity was calculated using 42 non-APS samples, stated to derive from patients with syphilis, human immunodeficiency virus, Q fever, and other non-APS autoimmune diseases.

These studies suggest that the APhL assay might serve as an alternative to aCL as a first-line test in the APS diagnostic algorithm. Results from a wet workshop at the 13th International Congress on aPL that evaluated the performance of aCL, aβ₂GPI, and APhL assays in the identification of 26 APS and persistent aPL-positive patients versus 21 healthy, infectious disease, and autoimmune controls supports this assertion [49]. The report from the 14th International Congress on aPL called for more extensive testing to confirm this assertion, especially for autoimmune diseases for which data are lacking [3]. Consequently, a comparative analysis was performed in a large number (n: 1178) of well-characterized SLE patients from ethnically diverse SLE cohorts [50]. In this study, IgG aβ₂GPI were highly associated with venous thrombosis in SLE patients, while the APhL and aCL assays were also associated with venous thrombosis, but with smaller OR values. The APhL was the only assay associated with both venous and arterial thrombotic manifestations.

A critical review of the APhL assay was presented at the 15th International Congress on aPL [51]. Six articles met selection criteria; in all the APhL assay correlated with APS, and OR values were greater than those for aCL and aβ₂GPI. The specificity of the APhL assay in diagnosing APS was greater than that for the aCL assay but similar to aβ₂GPI in most studies. Conversely, the APhL assay showed similar sensitivity for APS diagnosis when compared to the aCL assay and improved sensitivity when compared to the aβ₂GPI assays.

The task force concluded that more data on the clinical utility of APHL test are needed before any recommendation can be reached.

Numerous studies show interactions of monoclonal and polyclonal aPL with serine protease (SP) enzymes that regulate hemostasis. Monoclonal human aPL cross-react with SP and bind to thrombin, activated protein C (APC), plasmin, tissue plasminogen activator (tPA), factor (F)IXa, and FXa [52–56], which share amino acid sequence homology at their catalytic sites. Several monoclonal human aPL inhibit inactivation of procoagulant SP and functional activities of anticoagulant/fibrinolytic SP [53, 55, 57, 58], and some aPL may recognize the catalytic domain of SP, leading to dysregulation of hemostasis and vascular thrombosis. Sera from patients with APS (including SLE-associated APS) bind different SP [55, 57].

Factor Xa has a central position in coagulation and mediates cellular inflammatory and anti-inflammatory processes [59]. Given its important position in coagulation and inflammatory pathways, plus recent addition of direct FXa inhibitors as alternative oral anticoagulants, interest has focused upon autoimmune-mediated regulation of FXa as well as other SP.

Subsequent experiments utilizing purified anti-FXa-positive IgG revealed that the avidity of APS-IgG to FXa was higher than that of SLE-IgG. Furthermore, the greatest effects upon prolongation of FXa-activated clotting time (ACT) occurred with APS-IgG and inhibition of FXa enzymatic activity with APS-IgG followed by SLE-IgG when compared to HC-IgG. Antithrombin III inhibition of FXa was reduced by APS-IgG when compared to HC and SLE and did not correlate with binding to ATIII [60].

Inflammation is important in the pathogenesis of the APS through activation of complement and a family of G-protein-coupled receptors, known as protease-activated receptors (PARs) [61] that are present on endothelial cells. Serine protease enzymes, including FXa, activate PARs.

Artim-Esen and Giles hypothesized that polyclonal IgG with anti-FXa positivity may alter PAR-mediated inflammatory as well as coagulant effects in patients with APS and/or SLE. To test this hypothesis, the researchers measured real-time intracellular calcium (Ca²⁺) flux. They found a concentration-dependent induction of Ca²⁺ release by FXa that was significantly potentiated by APS-IgG compared to SLE/APS-IgG and to HC-IgG. Next, they examined the effects of a selective FXa inhibitor, antistasin, hydroxychloroquine, and fluvastatin in the presence or absence of patient IgGs. Treatment with all three drugs reduced FXa-induced and IgG-potentiated Ca²⁺ release.

Anti-FXa IgG isolated from patients with APS enhances both the enzymatic and cellular effects of FXa. Furthermore, FXa-mediated intracellular Ca²⁺ release in human umbilical vein endothelial cells is potentiated by IgG from anti-FXa-positive patients with APS and/or SLE. Further studies are now required to explore the use of IgG anti-FXa positivity as a novel biomarker and its potential to stratify treatment with FXa inhibitors.

The task force concluded that more data on the clinical utility of antibodies to FXa test are needed before any recommendation can be reached.

This novel functional assay is based on the concept that annexin A5 has potent anticoagulant properties that result from its forming two-dimensional crystals over phospholipids, blocking the availability of the phospholipids for critical coagulation enzyme reactions [62–64].

Annexin A5 resistance is specific for APS-derived aPL (compared to aPL induced by syphilis) [65], correlates with risk of thrombosis and pregnancy complications [66, 67] and occurs in children with SLE [68], mainly in the presence of aPL [68, 69]. Resistance to annexin A5 anticoagulant activity correlates with aPL that recognize an epitope on domain I of β₂GPI [70].

The task force concluded that more data on the clinical utility of annexin A5 resistance assay test are needed before any recommendation can be reached.

Thrombin generation (TG) tests , also referred to as thrombography , are methods using fluorogenic substrates that allow measurement of total thrombin activity in vitro in response to low concentrations of tissue factor, summarized as the thrombin generation curve or thrombogram [71]. The thrombogram calculates several parameters: lag time, time to peak, peak height, and the total amount of thrombin activity measured as the area under the curve, which is the endogenous thrombin potential (ETP) [72]. Activated protein C (APC) sensitivity of thrombin generation can be assessed either by adding APC or thrombomodulin. Activated protein C sensitivity can be estimated by an ETP ratio (APC sr) calculated by dividing the ETP measured in the presence of APC at a defined concentration by the ETP without APC. In addition, a dose-response curve of inhibition of ETP with increasing concentrations of APC (IC₅₀-APC) can also be performed. This index corresponds to the APC concentration that produces a 50% inhibition of ETP [73, 74].

Thrombin generation tests can be used for treatment monitoring. In the specific situation of APS, the intensity of anticoagulation can be assessed by means of the ETP. A recent multicenter study, the rivaroxaban in APS (RAPS) [75], included the percentage change in ETP from randomization to day 42 in both treatment arms (warfarin and rivaroxaban) as the primary outcome.

Thrombin generation assays have also been used to evaluate thrombotic risk. Quantitative measurement of LA activity by thrombin generation [76] correlates with thrombotic events [76].