FVIII or FIX inhibitors

Substitutive therapies are frequently used for treating hemophiliacs during bleeding episodes, or as prophylaxis to supply the missing coagulation factor, either blood extracted or recombinant. Patients can detect it as a foreign substance, especially when they totally lack FVIII or FIX synthesis. Autoantibodies develop then and render the substitutive therapy ineffective [48]. When this complication occurs, it is no longer possible to correct hemorrhages by infusing the deficient coagulation factor, and alternative therapeutic strategies must be used [49, 50]. The highest incidence for these autoantibodies concerns patients with severe hemophilia A, unable to synthesize FVIII, even in the inactive form (hemophiliacs A): They can develop high to very high titers of FVIII neutralizing antibodies. It is not rare to observe patients with antibodies able to neutralize various thousand-folds of the normal body’s FVIII concentration. Yet, FVIII concentration is very low, of only about 0.100 μg/mL, and antibodies are able to neutralize several hundreds of μg/mL of FVIII. This situation is especially illustrated by the introduction of prophylaxis with substitutive therapies, which cannot be used for hemophiliacs with inhibitors. In about 20% to 40% of severe hemophiliacs A, a different therapeutic management is required, because of the presence of inhibitors, sometimes transient, but often persistent [50]. Furthermore, many cases of acquired FVIII inhibitors have been reported, and they also require an alternative treatment [51]. When dramatic hemorrhagic events occur, they must be treated using an FVIII bypassing activity (FEIBA) strategy, like that with activated coagulation factors [FEIBA or activated PT complex concentrate (PCC)] or more recently with recombinant activated FVII (rFVIIa), Novoseven^® [51]. Recently, a new engineered drug has been introduced for curative or prophylactic treatment of patients with FVIII inhibitors (neutralizing FVIII antibodies). It is a biphasic antibody, emicizumab (Hemlibra^®), targeted to both FX and FIX/activated FIX (FIXa), and mimics FVIII activity by closing FX and FIX on the phospholipid surface, therefore bypassing FVIII [52]. This drug has a much longer half-life than blood-extracted, recombinant, or even engineered long-lasting FVIII, and it benefits from an extended half-life in blood circulation. Furthermore, it is injected subcutaneously only every 2 to 4 weeks to prevent bleeding, whilst FVIII requires up to 2 injections per week. This drug then offers an improved quality of life to patients. The long-term use of this drug is being followed closely now and it needs a retrospective evaluation for tolerance and safety, as patients with FVIII coagulation inhibitors require a life-long treatment. Autoantibodies to FIX have also been described in severe hemophiliac B, but this complication remains much rarer than that associated with FVIII inhibitors.

Autoantibodies to clotting factors or cofactors

Other autoantibodies to coagulation factors, associated with bleeding diathesis, have been described in the literature. They concern most of the coagulation factors, including fibrinogen [53], PT [54], FX [55], FVII [56], FV [57, 58], vWF [59, 60], and FXIII [61, 62]. When induced by infectious episodes, they are transitory and can provoke various clinical disorders associated with hypocoagulability, like bruises, gum bleeding, intracranial hemorrhages, or more severe bleeding complications. In some cases, they can become life-threatening or fatal. The induced clinical complications result from a double action: i) the direct effect through the inhibition of the targeted coagulation factor, which induces a factor deficiency; ii) the indirect effect resulting from the immune complex binding to body’s cells or organs, focusing the immune response onto them: The deleterious effects include tissue injury, inflammation, thrombo-inflammation, paradoxically disseminated intravascular coagulation (DIC), and thrombosis for some autoantibodies (like those to thrombin), although targeted to clotting factors [63, 64]. Autoantibodies can react with the protein active site and then inhibit directly the coagulation factor or cofactor function, or they react with non-functional epitopes and therefore accelerate the protein clearance, which induces deficiency. They can interfere with the coagulation cascade through their binding to phospholipids, which produces a steric hindrance, preventing the clotting factors from interacting with phospholipids.

Autoantibodies can develop to very specific or particular epitopes, for yet non-identified reasons. This is the case for a reported autoantibody to AT, which mimics the AT binding to heparin and can induce hemorrhages [65]. Other autoantibodies can be targeted to thrombin exosites and are associated with bleeding tendencies by impairing the thrombin clotting activity [66]. These antibodies totally differ from those generated by thrombin itself, like those produced when fibrin glue prepared with bovine topical thrombin was used for healing injuries or during surgery and paradoxically induced thrombosis [63, 64]. They are presented in Table 2, and are more extensively discussed in “Autoimmune thrombotic complications”.

Some contact phase proteins’ deficiencies are induced by autoantibodies to FXII, or Hageman factor, or more frequently to FXI, or Rosenthal factor. Bleeding disorders are mainly observed in some FXI deficiencies, as the result of an autoimmune response [67–69].

Autoantibodies to Plt GPs

Lastly, autoantibodies can be targeted to Plts, especially onto their surface proteins and receptors, and are then associated with thrombocytopenia, which can induce bleeding episodes in severe cases. Some viral infections have been identified as possible causes for the generation of these anti-Plt GPs’ autoantibodies, like those with Herpes, EBV, or cytomegalovirus. Various autoantibody-targeted GPs have been identified, mainly GP1b-GPIX, GP-IIb-IIIa, GPV, and GPVI [70, 71]. These autoantibodies induce thrombocytopenia and are of major concern during pregnancy. They can cross the placenta, and newborns require special caution and follow-up, as they can last some weeks.

Anti-phospholipid syndrome

The most frequent autoimmune-induced thrombotic complications concern the development of APAs, especially those targeted to β2GP1, or to PT, usually reported as APAs, which are responsible for the anti-phospholipid syndrome (APS). These phospholipid-dependent antibodies induce the prolongation of clotting times in coagulation testing and are reported as LA. Although these antibodies are anticoagulant in laboratory tests, they paradoxically provoke a thrombotic tendency in vivo and can be highly thrombogenic [75, 76]. In extreme cases, they can lead to catastrophic APS (CAPS), with disseminated blood activation and coagulation and multi-organ failure, a high-risk clinical complication, with often a fatal issue if not rapidly and appropriately controlled [77]. If β2GP1 and PT are the preeminent targets in APS or CAPS, some other phospholipid-binding proteins, like coagulation PS or A5 have been reported in this disease. In addition, other vitamin K-dependent proteins have been reported to be seldomly associated with APAs, like coagulation PC or PZ, although their incidence remains rare [78, 79]. If APAs generate LA activity, the latter can result from other autoantibodies and is not always associated with thrombotic complications. LA activity can be produced by autoantibodies which induce bleeding complications, like autoantibodies to FV or to PT [57, 80]. Moreover, autoantibodies to PT can induce opposite clinical manifestations, and be hypo-or hyper-coagulant. The mechanisms are yet poorly understood, but they depend on the generating trigger. These anti-PT autoantibodies may be targeted to different protein epitopes or domains, resulting in different consequences [81, 82]. Transitory anti-PT autoantibodies usually produce mild bleeding and are developed following viral infections. In a few cases, they become chronic, and tend to produce thrombotic complications, disseminated or at various body sites, and are then very difficult to manage and control [83, 84]. The inducing primary cause is not always clearly identified. This can result from an abnormal exposure to PT due to a genetic predisposition (HLA system), and the immune system lacks the efficiency to select self-components from the non-self-components, or to correctly recognize altered self-components [85, 86].

PF4 and heparin-dependent antibodies

Another major group of autoantibodies associated with thrombotic complications concerns anti-PF4-dependent immune reactions, responsible for HIT. In a significant number of cases, they are associated with thrombosis and were recently described as the causative factor of the ultra-rare vaccine-induced thrombotic thrombocytopenia (VITT) [87, 88]. PF4 is a chemokine present in Plt α-granules, which is early released upon Plt activation as a high MW complex formed by 8 PF4 tetramers bound to 2 proteoglycans [89]. PF4 half-life in blood circulation is short (a few minutes). When released from Plts, it is displaced from Plt proteoglycans and binds to endothelial cell glycosaminoglycan (GAGs), for which the affinity is higher. When patients are anticoagulated with heparin, that drug displaces PF4 from the endothelial cell storage pool and forms high-affinity HPF4, whose presentation depends on the respective concentrations of heparin and PF4. When stoichiometric concentrations are reached [about 27 international units (IU)/150 μg of heparin per mg of PF4], heparin, a molecule with a high density of electronegative sulfate groups, wraps tightly around the ring of electropositive charges of the PF4 tetramers. This binding alters the PF4 structure and exposes new epitopes, which can become antigenic and generate heparin-dependent antibodies, targeted to complexes of heparin and PF4. These heparin-dependent antibodies provoke deleterious effects when targeted to Plts, blood, or endothelial cells exposing HPF4 complexes on the cell surface [90, 91]. Therefore, HIT pathogenicity results from a “double misfortune”: i) release of PF4 to a concentration which forms stoichiometric complexes with heparin, and which stimulates the immune response; ii) permanence of HPF4 stoichiometric complexes in blood circulation, which bind onto Plt, blood, or endothelial cells and react with the heparin-dependent antibodies, thus provoking the adverse complications. Stoichiometric HPF4 complexes are formed for HPF4 respective concentrations of 0.027 IU/mL of heparin per μg/mL of PF4. In practice, PF4 in blood circulation is present at very low concentrations (< 10 ng/mL), and stoichiometric complexes can only be formed for very low heparin concentrations (0.000027 IU/mL for 10 ng/mL PF4). They are then present at a too-low level for becoming antigenic [87, 90, 91]. However, at the beginning of heparin therapy, PF4 is released from the endothelial cell storage pool, and its concentrations can then be much higher in blood circulation, of various hundreds of ng/mL. PF4 concentration is positively correlated to the hypercoagulability status of heparin-treated patients, and this drug favors PF4 released from Plt α-granules [92, 93]. When high levels of PF4 are in blood circulation (up to 1.0 μg/mL), stoichiometric complexes can be formed with heparin concentrations from 0.0027 IU/mL to 0.027 IU/mL, which still remains very low. But PF4 can accumulate in pathological sites with inflammation or blood activation, where it can reach very high concentrations, of several orders of magnitude higher than in blood circulation. Chances to generate stoichiometric complexes are then much higher, and compliant with heparin therapeutic levels, especially since heparin concentrations vary over time following intra-dermal injections. Immunogenicity can then develop more easily. Heparin blood concentration remains stable in treated patients only when the drug is infused intravenously. This mechanism explains the low patients’ propensity to develop HIT during heparin therapy and the association of incidence with clinical contexts, although any individual can develop an immune reaction when exposed to stoichiometric HPF4. Interestingly, many individuals develop asymptomatic antibodies just after the onset of heparin treatment, whilst only a few of them evolve to HIT, possibly complicated by thrombosis [94]. Conversely, pathological states with high grades of Plt and blood activation, like traumatic surgery or cardiovascular diseases, are associated with a higher frequency of HIT, with an incidence of up to 5% of heparin-treated patients. Thrombotic complications occur at pre-existing pathological sites, especially those associated with inflammation or blood activation, and with chemo-attraction of leukocytes and Plts, like infectious or malignant diseases, metabolic disorders, impaired vessel patency, and atherosclerotic deposits [95]. Disease then develops at sites of former existing disease. In addition, HIT incidence and severity increase with the HPF4 antibody concentration and affinity. These site-targeted and concentration-dependent complications observed in HIT also occur in other autoimmune-associated thrombotic diseases, like those induced by LA and APAs. However, in HIT and also probably in other autoimmune diseases, all autoantibodies are not pathogenic, and many remain asymptomatic. Following the purification of heparin-dependent antibodies in patients with HIT, we demonstrated that only a few of them are able to bind and activate Plts, whilst the vast majority remain asymptomatic [96].

PF4-dependent antibodies

Recently, PF4-dependent autoantibodies have been demonstrated to be responsible for VITT. The pathogenic mechanism and the targeted epitopes on PF4 are totally different from those involved in HIT. Whilst in HIT the immunogen is the HPF4, which expose new PF4 epitopes, in VITT the reactive PF4 epitopes are located on the protein domain which reacts with heparin, i.e., the ring of positive charges [87, 97, 98]. Heparin, even at a very low concentration, is able to displace these antibodies through its strong binding to the involved PF4 epitopes. The causes which induce an autoimmune response to PF4 following the adenovirus vector SARS-CoV-2 vaccine, are unknown and concern only very rare individuals, which suggests that an unusual event has occurred. This is not a general response to the vaccine as in the case of heparin therapy. No genetic predisposing factor has been identified until now. Noteworthily, the vaccine contains an adenovirus vector, but without any replication possibility. The usual route for the vaccine is intradermic, with no contact with blood circulation. A possible cause could be in very rare cases; the vaccine entrance into blood vessels, and adenovirus can then react with Plts and PF4, through the electronegative adenoviral cell-surface hexon moieties. This could then induce a potent immune response to PF4 complexed with adenovirus through an epitope-spreading mechanism [87]. In VITT, the anti-PF4 concentrations are very high, and antibodies are much more reactive to PF4 than to HPF4 complexes. Conversely, in HIT, antibodies are first generated to HPF4 complexes and can be extended to PF4 alone, especially with time of exposure, but the preeminent reactivity to complexes remains. This is a convenient way to differentiate PF4 autoantibodies developed in HIT or VITT by testing their respective reactivity to PF4 or to HPF4 complexes. The autoantibody reactivity optical density (OD) ratio [OD (HPF4):OD (PF4)] is then far greater than 1.00 for HIT and far smaller than 1.00 for VITT. Using a chimeric human-mouse recombinant monoclonal antibody, K070 provided by HYPHEN BioMed (Neuville sur Oise, France), or using affinity-purified antibodies to PF4, we were able to estimate the plasma weight concentrations of those antibodies in HIT and VITT. They are about from 10 μg/mL to > 100 μg/mL in VITT, and from < 5 μg/mL to > 10 μg/mL in pathogenic HIT.

Lastly, anti-PF4 autoantibodies have been described in other pathological complications, favoring the development of thrombotic complications, often intra-cranial, and are paradoxically worsened by heparin therapy, which induces thrombocytopenia and thrombosis. This was observed in some pharyngeal infections, as reported previously in a young child, and confirmed later as “spontaneous HIT” [98, 99]. Antibodies persist for more than 6 months but progressively decrease with time and become asymptomatic. Recently, similar antibodies have been reported in patients with severe COVID-19 evolution or other infectious diseases, and their presence and concentration have been described to be associated with illness severity [17, 100].

Anti-chemokine autoantibodies

Among complications linked with heparin therapy, cases of autoantibodies to IL-8 have been described [101]. Autoantibodies to IL-8 can develop in presence of inflammation, especially chronic, and their presence can contribute to regulating the excess of that inflammatory chemokine. This mechanism was reported for the regulation of other inflammatory cytokines, and it is not expected to complicate the pathological state. However, if any condition targets the cytokine-antibody complex in cells, this can focus the autoimmune response onto these cells, and then autoantibodies become deleterious. For IL-8, this was described in a patient with malignancy and a tendency to develop thrombosis. Hypercoagulability was not reversed by heparin therapy, but by contrast, it worsened and disseminated thrombi were formed [102]. Anti-IL-8 antibodies pre-exist the heparin therapy, but heparin treatment then favors the binding of IL-8 to various blood cells and Plts, which therefore react with autoantibodies: This focuses the autoimmune response onto these cells, provoking thrombocytopenia, although milder than in typical HIT, cell and coagulation activation, and finally thrombosis.

Anti-protamine sulfate antibodies

Another possible at-risk context concerns the use of protamine sulfate to reverse heparin therapy, especially for extra-corporeal circulation (ECC) or extra-corporeal membrane oxygenation (ECMO). For these indications, heparin is used at high concentrations, up to 5 IU/mL in plasma, and is then neutralized by protamine sulfate at the end of the process. In a few observed cases of HIT-like syndrome, but with negative anti-HPF4 antibodies, investigations lead to identifying antibodies to protamine sulfate. They are not autoimmune as protamine sulfate is not a self-component but is usually extracted from salmon sperm [103]. Pathogenicity of some anti-heparin sulfate antibodies results from the binding of heparin protamine sulfate complexes to many blood cells, especially when heparin is slightly in excess. When anti-protamine sulfate antibodies are present, they are then targeted onto these cells, where they focus on the immune response and can generate thrombocytopenia and thrombosis. Nevertheless, many patients exposed to protamine sulfate (to neutralize heparin, or as a component in some insulin preparations used for diabetes treatment) can develop antibodies, but only a very small proportion of them develop pathogenic complications [104, 105]. This tends to occur when patients are exposed various times to protamine sulfate, especially for ECC or ECMO. Yet, antibody concentration is an important factor in the development of pathogenic complications.

Antibodies to coagulation inhibitors

Autoimmune thrombotic complications can be observed in some rare cases of autoantibodies developed to anticoagulant proteins, like PS, PC, or TM, or to thrombin, which impairs the PC activation pathway as discussed in the chapter “Autoimmune thrombotic complications”. Many were already discussed along with the phospholipid-dependent antibodies. Autoimmune thrombotic complications can result from various factors, often combined between i) inhibition of the antithrombotic protein activity by the autoantibody; ii) focus of the autoimmune response onto blood cells, which induces cell activation and destruction, and sometimes thrombosis, especially when cell targets are Plts or endothelial cells; iii) inflammation and blood activation through chemoattraction of neutrophils, and release of NETs, which contribute to development of thrombotic complications. The 3 isotypes, IgG, IgA, or IgM isotypes can be present.

Principles of autoantibodies’ laboratory investigations

The major laboratory techniques used for detecting autoantibodies associated with hemostasis dysfunctions (either bleeding or thrombotic) are presented in Table 4.

Assay methods include: i) functional assays, to evaluate the physiological activity inhibited by autoantibodies; ii) antibody titration methods, to estimate the autoantibody concentration through its capacity to inhibit a biological activity, or to neutralize antigen; iii) immunological methods, which investigate autoantibodies binding to the targeted autoantigens, especially when they can be characterized and available; iv) cell-based methods for autoantibody binding. Antibody titer is measured using serial dilutions of sample containing it, which is mixed with normal plasma, or purified autoantigen at a defined concentration, and evaluating the excess of activity or antigen. The highest dilution which neutralizes a defined activity or antigen amount allows measuring the titer.

Laboratory monitoring of autoantibodies is of special relevance for surveillance of patients’ evolution and adjustment of treatments when possible. These investigations concern autoantibodies’ interference in physiological functions through the direct inhibition of physiologically active sites, accelerated protein clearance by non-inhibitory autoantibodies, or the specific measurement of autoantibodies with capture immune-assays designed autoantigens. When present, autoantibodies induce different consequences, depending on their target autoantigen, their concentration and affinity, and the immune complexes binding onto the various body’s structures and organs.

Functional assays for phospholipid-binding proteins

These assays evaluate the autoantibody impact on biological activity. The most frequently observed concerns prolongation of clotting methods. Antibodies targeted to phospholipid-binding proteins form complexes with electronegative phospholipids, either directly like β2GP1, or indirectly through divalent cations (especially Ca⁺⁺, but also Mg⁺⁺, Zn⁺⁺, and Mn⁺⁺), like vitamin K-dependent proteins or A5. They interfere in all phospholipid-dependent assays used for coagulation testing. Clotting methods are then prolonged or even unclottable. Electronegative phospholipids are exposed on many cell surfaces, like Plts, endothelial cells, and other blood cells, especially when activated, and they bind to phospholipid-dependent proteins through the same mechanisms. This binding targets the immune response onto these cells, focusing its deleterious effects. Antibody affinity and concentration determine the intensity of that harmful activity. There is this laboratory paradox for these autoantibodies, which induce anticoagulant activity in coagulation assays and are thrombogenic in-vivo. Pathogenicity concerns: thrombotic or thrombo-inflammatory complications, thromboembolic diseases, skin or limb extremities necrosis, vasculitis, myocarditis, etc. These autoantibodies are usually detected through their anticoagulant activity, reported as LA: This is a very heterogenous group, reacting with many different phospholipid-binding autoantigens [109–111]. Obviously, autoantigens presenting the highest phospholipid affinity in physiological conditions are the most frequently involved in autoantibodies’ development. Assays must then mimic physiological conditions whenever possible, as testing conditions highly impact autoantigen interaction with phospholipids, and therefore autoantibody detection. For example, whether calcium is present or not can induce opposite effects for the autoantibody interaction with its target autoantigen, as reported for β2GP1 capture assays [112]. Noteworthily, low-affinity autoantigens can express a much higher affinity for phospholipids when dimerized through the divalent autoantibody binding. Specific functional assays designed to detect LA activity in citrated plasma present different specificities and sensitivities [113]. This is why it is recommended to combine 2 different assay principles based on phospholipid dependence for a wide-spectrum diagnosis of LA. In practice, assays are performed at both low and high phospholipid concentrations: The low one allows to detect this anticoagulant activity, whilst the high one confirms LA. The 2 currently used methods are the APTT and the diluted Russel viper venom time (dRVVT). The International Society for Thrombosis and Hemostasis (ISTH) publishes and updates from-time-to-time guidelines for the optimization of LA and APA laboratory testing [114, 115].

Immunoassays

Aside from these large-spectrum functional methods, specific immunoassays can be used when the autoantigen is well-characterized. Capture techniques are designed by using the purified autoantigen, either blood extracted or recombinant or its phospholipid complexes fixed on a solid surface. Usually, the ELISA is used. The coated ELISA plate captures autoantibodies, and then is measured with a tag-antibody directed to human IgGs, or eventually IgMs or IgAs, all isotypes being possibly present in autoimmune diseases. IgGs are often the most pathogenic, but in some cases, IgMs or IgAs can induce clinical complications and must be detected [114, 116, 117]. The autoantigen used for capturing autoantibodies must be in a presentation mimicking the in-vivo conditions, especially those present in pathology. Each assay needs then to be developed and validated for each class of autoantibodies measured. Concerning phospholipid-binding proteins, many have been reported as possible targets for autoantibodies. The highest incidence concerns β2GP1 which binds phospholipids in a calcium-independent way, and whose interaction is highly enhanced through its binding to anti-Β2GP1 autoantibodies. Others concern first PT, then A5, PS, PC, PZ, etc., all proteins, which bind to anionic phospholipids in a calcium-dependent manner.

A similar mechanism is involved in autoantigens’ interaction with electronegative polysaccharides (containing a high density of sulfate groups). The latter can interact with many soluble blood components or structures exposed on cell surfaces. Those with the highest binding affinity for polysaccharides are those involved the most frequently in the development of autoantibodies. Especially, the PF4 chemokine exhibits a strong affinity for heparin and electronegative polysaccharides, and its tetrameric structure exposes a ring of positive charges. In presence of stoichiometric concentrations with heparin, a long chain electronegative polysaccharide with many sulfate groups, and its structure is altered [89, 118]. New epitopes are exposed and can be immunogenic, inducing PF4-dependent autoantibodies. When present, these antibodies bind to HPF4 complexes. This observation allows to design specific assays for identification and estimation of heparin-dependent anti-PF4 autoantibodies, responsible for HIT. Capture autoantigen is then a stoichiometric HPF4, which presents a strong reactivity with autoantibodies involved in HIT. The IgG isotype, alone or combined with IgMs and/or IgAs, is involved in pathogenicity [119]. Many anti-HPF4 antibodies are asymptomatic. Their concentration and affinity for HPF4 are essential factors for pathogenicity. Other heparin-binding chemokines, although with a lower affinity, have been reported to be associated with clinical thrombogenic complications, like IL-8 or neutrophil-activating peptide 2 (NAP-2) [101]. Recently, PF4 has been identified to be the target autoantigen in VITT, but the mechanism differs from that involved in HIT. Conversely, to heparin-dependent antibodies, whose reactivity is the strongest to stoichiometric HPF4 complexes, those developed in VITT are mainly reactive with PF4 alone, and this binding is inhibited by low heparin concentrations [87, 88]. Actually, VITT autoantibodies bind to a defined PF4 domain located on the ring of electropositive charges on the PF4 tetramer, which is the heparin-binding structure [97]. Using HPF4 or PF4 capture immunoassays for the IgG epitope allows differentiating HIT and VITT autoantibodies: Those present in HIT bind preferentially to HPF4 complexes, whilst those developed in VITT exhibit a much higher reactivity with PF4 alone.

Autoantibody concentration and pathology severity

There is a relationship between autoantibody concentration and the development of autoimmune pathology. Using quantitative or semi-quantitative assays could highly improve the immunoassays diagnostic value. The standard immunoassay procedures use a single 1:100 sample dilution (sometimes 1:50). This does not allow differentiating moderate from high or very high autoantibody concentrations. When present at a very large excess in the tested dilution, the assay absorbances correspond to the high concentration range of the ELISA method, and it is not possible to estimate the true concentration. Using successive and higher plasma dilutions allows for measuring better autoantibody concentrations. For HIT, and more especially VITT, we noticed that many of the most severe patients gave responses in the high assay range for plasma or serum dilutions up to 2:2,000 or even 1:5,000. This shows that when these patients are tested at the 1:100 dilution, the response obtained does not even discriminate on a scale from 1 to > 50. To calibrate immunoassays, we now internally use a recombinant monoclonal antibody to PF4 reactive HPF4 or affinity-purified anti-PF4 polyclonal antibodies. Immunoassays allow for detecting autoantibodies, but not their pathogenic effect. Functional assays must be used to confirm their deleterious impact. HIT functional assays are performed by testing patients’ sample capacity to activate/aggregate Plts from healthy donors at a low (0.1 IU/mL to 1.0 IU/mL) and a high heparin concentration (usually 100 IU/mL). The C₁₄-serotonin release assay (SRA) is the most sensitive one and is often considered the gold standard, but PAT, heparin-induced aggregation assay (HIPA), and flow cytometry assay (FCA) have also been successfully used and are part of the panel available for characterization of symptomatic PF4-dependent autoantibodies [89]. When other chemokines are involved, using the purified protein or its recombinant form as capture antigen in the ELISA allows testing the correspondent autoantibody. If isotyping is required, specific tag antibodies reactive with IgG or IgM or IgA are then used.

Challenges of autoantibodies’ laboratory testing

For autoantibody testing, the right autoantigen available in a highly purified form and in a presentation mimicking its in-vivo structure must be used. In addition, the current challenge is to optimize sensitivity while keeping full specificity. Capture solid phase coated autoantigen, like in ELISA, must avoid non-specific binding from the tested specimen, especially when rheumatoid factor or heterophilic antibodies are present, or when “sticky samples” with circulating immune complexes are tested. When inflammation is present, high backgrounds can be generated [120, 121]. To achieve that goal, coated solid surfaces must be saturated with an animal serum (from the same species as that used for the tag antibody) to block possible interactions with antibodies. The same animal serum must be used for the sample diluent at a concentration of ≥ 10%, tenfold higher than that of the tested specimen (measured at the 1:100 dilution, i.e., 1%). Currently, we work with goat serum, which is from the same species as the detection antibody. In these conditions, one cannot react with the solid phase saturation serum or the sample diluent, and non-specific reactions are then suppressed or highly reduced. Another challenge is to define the cut-off value between the normal and pathological ranges and to differentiate the background from the specific antibody reaction [122]. In practice, a panel containing at least 120 normal individuals, male or female, out of any medication or disease, and as many pathological samples with autoantibodies as possible (ideally at least 30 or more), must be tested. The mean (M) value obtained for healthy individuals plus 2 standard deviations (SD) defines the highest value of the normal range, whilst M + 3 SD (or M + 5 SD depending on the local definition and assay specificity) defines the lowest value for the positive range. In between, from M + 2 SD to M + 3 SD (or M + 5 SD), there is a grey zone, which can concern borderline positive results or non-specific reactions. For diagnostic or clinical study purposes to rule out the presence of any autoantibody, the cut-off value for the negative range must be used, and complementary investigations must be performed for any reactivity above that limit.

Limitations and pitfalls of laboratory techniques

A key factor for immunoassay specificity and reactivity remains the assay design and choice of reagents. The target autoantigen used for preparing the capture solid surface can be recombinant or extracted, but its purity grade is essential for avoiding non-specific interactions. Binding of “sticky samples” or containing immune complexes is avoided or reduced by introducing the saturation step, usually performed with animal serum, also used for the sample diluent [120]. We recommend employing a serum from the same species for the tagged antibodies, the saturation step, and the assay diluent to avoid any unwanted reaction. This animal serum can contain proteins homologous to human ones. Animal and human homologous proteins can share some epitopes, but from our experience, this rarely interferes in the assay specificity. When the autoantigen is unknown, running a control plate prepared in the absence of antigen or coated with an irrelevant antigen like bovine serum albumin (BSA) saturated allows for confirming specificity. If tested samples remain reactive with the control plate, presence of a cross-reactivity with the animal homologous protein is possible. The assay must then be performed by using a different plate saturation and sample diluent additive, like BSA, casein, hydrolyzed collagen peptides, or even poly-ethylene-glycol (PEG). The target autoantigen can then be confirmed. Another technical difficulty concerns antigen presentation. In some conditions, autoantibodies only react with a structure modification of the autoantigen inducing the exposure of neo-epitopes. This is the case for β2GP1 and ADAMTS-13 for example. Autoantigen can furthermore be altered when bound to the plate, but this often mimics its binding to in-vivo biological surfaces, exposing modifications responsible for the autoantibody generation. The laboratory challenge is then to reproduce the antigen presentation, react with autoantibodies, and avoid any assay failure due to autoantigen structural change when immobilized. Should this occur, complementary testing could be performed by using other techniques where the autoantigen presentation is preserved or by performing inhibition studies. In addition to ELISAs, automated immune assays become now available for the most frequently tested autoantibodies, like anti-β2GP1, APAs, or anti-HPF4. Assays use autoantigen bound to magnetic latex beads, and captured antibodies are detected by chemiluminescence. For autoantibodies to Plt antigens, the MAIPA technique can be used. Lastly, the global presence of antibodies on cell surface can be estimated using the Coombs method.