Inhibitors in congenital coagulation disorders


Dr Nigel S. Key, Professor of Medicine, Division of Hematology, Oncology, and Transplantation, MMC 480 Mayo Building, 420 Delaware St S.E., Minneapolis, MN 55455, USA.


The development of inhibitory ‘allo’ antibodies to a deficient coagulation factor is arguably now the most severe and important complication of clotting factor concentrate exposure in haemophilia and other congenital coagulation disorders. Furthermore, development of an inhibitor to the factor VIII or factor IX transgene product remains a significant concern in gene therapy protocols for haemophilia. Although the development of an inhibitor does not usually change the rate, initial severity or pattern of bleeding, it does compromise the ability to manage haemorrhage in affected individuals, resulting in a greater rate of complications, cost and disability. The purpose of this review is to summarize current understanding of the epidemiology, immunobiology, laboratory evaluation and management of inhibitors arising in patients with congenital coagulation disorders. An attempt has been made to focus on recent advances in the immunology of inhibitors, and to speculate on their potential clinical application.

During the latter half of the 20th century, the management of patients with congenital coagulation factor deficiency states witnessed dramatic improvements, thanks largely to the development and commercialization of clotting factor concentrates. However, on the heels of the improved quality of life, convenience and independence offered by these plasma-derived concentrates followed the tragic epidemics of blood-borne hepatitis and human immunodeficiency virus (HIV) infections. With the closure of this chapter in history, the development of inhibitory ‘allo’ antibodies (hereafter referred to as ‘inhibitors’) to the missing coagulation factor is arguably now the most severe and important complication of clotting factor concentrate exposure. In practice, inhibitors to factor VIII (FVIII) and to a lesser extent factor IX (FIX), constitute the vast majority of these antibodies, and it is in patients with haemophilia A and B that the most dramatic progress has been made in the understanding of risk factors and optimal management strategies. The bulk of this review will therefore be devoted to inhibitors that occur in these patients.

Definition of inhibitors

An inhibitor is a polyclonal high affinity immunoglobulin G (IgG) antibody that specifically neutralizes the procoagulant activity of the relevant clotting factor. Inhibitors are most commonly defined as either ‘high’ or ‘low’ responder subtype. The FVIII/IX Subcommittee of the Scientific Sub-committees of the International Society of Hemostasis and Thrombosis (ISTH) currently defines high responder inhibitors as those with peak activity >5 Bethesda Units (BU)/ml, that are associated with anamnaesis following replacement of the missing clotting factor (White et al, 2001). In many, but not all high-responder patients, the inhibitor titre will regress to low or even undetectable levels following abstinence from FVIII. However, spontaneous loss of an anamnaestic response to FVIII re-challenge is a very rare event, and should therefore always be anticipated in any patient with such a history.

Although it is inferred that the activity of an inhibitor can be quantified in vitro, non-neutralizing (i.e. non-inhibitory) antibodies that target non-functional epitopes on FVIII may occur in some individuals. These antibodies may be clinically relevant if they result in the accelerated clearance of the transfused clotting factor in vivo. However, the prevalence of non-neutralizing antibodies in the absence of a detectable inhibitor is disputed. While some groups have reported that non-inhibitory anti-FVIII antibodies are quite prevalent (Gilles et al, 1993; Dazzi et al, 1996), others have concluded that they are rather uncommon (Ling et al, 2003). Data on whether the presence of non-inhibitory antibodies correlates with subnormal recovery and half-life of FVIII in vivo are limited, so that the functional relevance of these antibodies currently remains uncertain.

Laboratory characterization of FVIII inhibitors

The Factor VIII/IX Subcommittee of the ISTH has endorsed the recommendation that the Nijmegen-modified Bethesda assay should be adopted to quantify FVIII inhibitors (Giles et al, 1998a). Modifications from the originally described Bethesda assay include (i) the addition of imidazole buffer to the normal pooled plasma substrate; and (ii) the use of FVIII-deficient plasma instead of buffer in the control mixture (Verbruggen et al, 1995). While these modifications undoubtedly improve assay specificity (Giles et al, 1998a), several problems with current testing for FVIII inhibitors remain. First, the assays remain insensitive to low titre inhibitors (<0·4 Nijmegen-BU/ml). In this regard, a novel assay (tentatively called the ‘Nijmegen Low Titre Inhibitor Assay’) currently under development reportedly has a detection limit that is 10–15 times lower than the standard Nijmegen-Bethesda assay (Verbruggen et al, 2003). It will be of interest to determine what proportion of ‘non-inhibitory’ FVIII inhibitors that lead to accelerated clearance of FVIII in vivo can, in fact, be shown to be low titre neutralizing antibodies using this higher sensitivity assay. Secondly, in the absence of a reference antibody standard, the high interlaboratory variability in the quantification of FVIII inhibitors remains a problem. Lastly, it is known that other variables, including the type of FVIII-deficient plasma and specifics of the FVIII assay, can affect assay reliability (Verbruggen et al, 2001).

When FVIII in the test mixture is inactivated in a dose-dependent linear manner, the inhibitor is usually referred to as a ‘type I’ antibody. Most commonly, inhibitors occurring in patients with severe haemophilia A behave in this manner. On the contrary, ‘type II’ inhibitors usually exhibit complex kinetic behaviour, and do not fully inactivate FVIII activity in the test mixture, even after prolonged incubation. These inhibitors are more characteristic of acquired haemophilia, and are also frequently seen in patients with mild haemophilia A that develop inhibitors.

Immunobiology of inhibitors

The synthesis of inhibitors depends upon the activation of CD4+ (helper) T cells specific for FVIII. When the FVIII molecule is endocytosed by a professional antigen-presenting cell (APC), intracellular proteolytic degradation results in the generation of short component peptides. One or more of these peptides may then act as T-cell epitope(s) when presented to cognate T-cell receptors on CD4+ (T helper) cells by major histocompatibility complex (MHC) II molecules located on the APC surface (Fig 1). In order for full and efficient activation of T helper cells to occur, recognition of the MHC II presented peptide by the T-cell receptor must be augmented by additional co-stimulatory signals between the APC and T cell. Several co-stimulatory pathways have been identified (Bretscher, 1999), including the ligation of CD28 by B7-1 (CD80) and B7-2 (CD86) (Coyle & Gutierrez-Ramos, 2001), and the interaction between CD40 and CD40L (CD154). Engagement of the T-cell receptor (signal 1) without subsequent co-stimulation (signal 2) leads to abortive activation of T-helper cells, and possibly even antigen tolerance because of anergy. In the presence of appropriate co-stimulation and cytokine environment however, naïve CD4+ cells (Th0 cells) may be induced to differentiate into T helper 1 (Th1) or Th2 clones. Th1 cells are classically associated with cell-mediated functions and the synthesis of complement-binding IgG subclasses (IgG1 and IgG2), whereas Th2 cells are important in the synthesis of non-complement binding antibodies (IgG4 and IgE), and providing help to B cells (reviewed in Lollar, 2004). Cytokines secreted by the effector Th1 [such as interleukin 2 (IL-2) and interferon-γ] and Th2 (such as IL-4, -5 and -10) clones then direct B-cell synthesis of antibodies, which, in the case of FVIII, may function as inhibitors. Paradoxically, Th2 cells may also downregulate B-cell antibody synthesis under certain circumstances.

Figure 1.

Cooperation of CD4+ T helper cells and B cells is essential for antibody production. Antigen presenting cells (APC; coloured grey) present linear peptide sequences (in red) generated from intracellular proteolytic cleavage of an antigen in a complex with MHC-II to the T-cell receptors of CD4+ T helper cells (in yellow). Full activation and generation of effector T cells requires co-stimulatory signalling between APC and the CD4+ cell. CD28 is constitutively expressed on T cells, whereas B7·1 (CD80) and B7·2 (CD86) must be induced on APCs within an inflammatory environment. T-cell activation by CD28 engagement also upregulates CD40L (CD154), which is essential for the co-stimulatory process. CD40, the receptor for CD40L, is expressed on APCs and B cells (in pale blue). Interaction with APCs via this ligand pairing further upregulates their antigen presenting function, while interaction with B cells promotes their survival and memory function. CTLA4, which is induced on T cells after activation, has a much higher affinity for B7·1 and B7·2 than does CD28, and binding to these molecules inhibits the initiation of, or downregulates established T-cell activation and IL-2 production induced by CD28 (upper right part of panel). In the presence of IL-12 and IL-18 (from dendritic cells), naïve Th0 will differentiate into Th1 clones that secrete IL-2 and IFN-γ, whereas in the presence of IL-4, differentiation to Th2 cells that secrete IL-4, IL-5 and IL-10 is promoted. Th1 cells differentiate into either ‘effector Th1 cells’ or ‘long term memory Th1 cells’. This process may be regulated by inhibitory Th2 clones and Treg cells that produce inhibitory cytokines, including IL-10 and TGF-β.

The FVIII protein consists of domains arranged in the order: A1-A2-B-A3-C1-C2. Small acidic amino acid-rich regions are located between domains A1 and A2, A2 and B, and B and A3. FVIII inhibitors are usually of mixed subclass with a dominant IgG4 fraction, although both Th1 and Th2-driven B-cell responses are probably important in the synthesis of antibodies to FVIII (Gilles et al, 1993; Reding et al, 2002). B-cell epitope mapping studies, initially by immunoblot analysis of inhibitor binding to rFVIII peptides (Scandella et al, 1989), and more recently by functional inhibition of porcine/human FVIII hybrid molecules (Healey et al, 1995), demonstrated several dominant epitopes in the A2, A3 and C2 domains, and in the acidic region between the A1 and A2 domains of the FVIII molecule. These inhibitors typically bind to several critical binding sites in FVIII (Prescott et al, 1997), leading to steric hindrance of the interaction with FIXa, phospholipids and/or von Willebrand factor (VWF). Occasionally, other inhibitory mechanisms have been described, such as binding to neo-epitopes formed by the complex of FVIII with VWF (thereby preventing release of FVIII from VWF), inhibition of FVIII activation by thrombin, or inhibition of binding to substrate FX (reviewed in Ananyeva et al, 2004). Recently, it was demonstrated that in some patients with FVIII inhibitors, FVIII neutralization might be explained by intrinsic proteolytic activity associated with so-called ‘catalytic antibodies’ to FVIII. In 13 of 24 inhibitor patient samples, it was demonstrated that multiple sites of cleavage throughout the FVIII domains were induced by the IgG fraction or affinity-purified antibodies against FVIII (Lacroix-Desmazes et al, 2002). However, because the reaction kinetics of this reaction are very slow, the in vivo significance of this mechanism of FVIII inactivation requires further clarification.

Epidemiology of FVIII inhibitors

Clinical trials with the new recombinant FVIII (rFVIII) concentrates in the late 1980s shed new light on the epidemiology and natural history of inhibitors. As the major concern in that era was the transmission of blood-borne viruses, regulatory authorities demanded that previously untreated patients (PUPs) constitute the target population for prelicensure clinical trials. Participants underwent frequent testing for inhibitors, and the development of an inhibitor was generally defined by a titre >0·6 BU/ml in any sample. The results of several rFVIII PUP studies consistently demonstrated a median of 9–12 exposure days (ED) until inhibitor development, corresponding to a median age of 1·7–3·3 years (Lusher, 2002). About 60% of inhibitors were high titre, and the remainder low titre; more importantly, it was recognized that some patients developed transient inhibitors (usually low titre) that regressed ‘spontaneously’ with time, despite continued intermittent exposure to FVIII. The cumulative incidence of inhibitors, generally defined in severe haemophilia A recipients >5 years of age or following >100 ED, was consistently in the range of 25–32%, although the prevalence eventually fell to approximately 12% as some antibodies disappeared over time. These rates were witnessed in the trials with both first generation (KogenateTM; Bayer Corporation, Elkhart, IN, USA, and RecombinateTM; Baxter, Glendale, CA, USA) and second generation (Kogenate FSTM, Bayer, and RefactoTM; Genetics Institute, Cambridge, MA, USA) rFVIII products (Lusher, 2002). The initial concern with this incidence rate – which was about twice that expected – was that exposure to rFVIII products might induce a higher rate of inhibitor formation compared with a single plasma-derived source of FVIII (pdFVIII). Although this interpretation continues to be expressed (Wight & Paisley, 2003; Aledort, 2004), it is also possible that the apparent discrepancy could be explained by differences in the design, study population and frequency of monitoring strategy used in the more rigorous rFVIII trials compared with older studies with pdFVIII (Hoots & Lusher, 2004; Lusher, 2004; Scharrer & Ehrlich, 2004). This debate is likely to continue in light of recent data demonstrating that, although the rate of inhibitor development decreased during 1977–1990 in the UK, it rebounded during the era of rFVIII in the 1990s (UKHCDO, 2004).

Perhaps more informative in the determination of the true immunogenicity of newly introduced FVIII products are studies in previously treated patient (PTP) populations. As the fear of blood-borne viruses receded, regulatory authorities have discontinued the requirement for PUP studies for the licensure of new rFVIII concentrates, although studies in PTP remain mandatory. The definition of ‘PTP’ requires some clarification. The natural history of time to inhibitor development from the Cooperative Inhibitor Study (McMillan et al, 1988) is shown in Fig 2. Notably, a ‘tail’ of new cases is present well beyond 50 ED, but new events are rare beyond 250 ED. It is apparent that approximately 80% of high titre inhibitors are formed within 50 ED, whereas low titre inhibitors may still arise after many more exposures. Many studies have used as little as 50 ED as a cut-off for defining PTPs, although the FVIII/IX Subcommittee of the ISTH recommends that >150 ED is the most appropriate definition of PTP in these studies (White et al, 1999). Clinical trials to assess the immunogenicity of the most recently licensed third generation rFVIII, AdvateTM (Baxter, Westlake Village, CA, USA), used an eligibility requirement of 150 ED. In this study, one transient low titre inhibitor was seen in one of 108 subjects after 26 ED (Ewenstein et al, 2003), similar to results of this type of study with other rFVIII products (Lusher, 2002, 2004). These data are comparable with several national database analyses showing that the inhibitor incidence rate in PTPs – most of whom are low responders – is in the range of 1–3% over 2 or more years (McMillan et al, 1988; Rosendaal et al, 1993; Giles et al, 1998b; UKHCDO, 2004).

Figure 2.

Incidence of factor VIII inhibitors as a function of cumulative days exposure to Factor VIII for all patients (•—•), patients with inhibitor titres >5 BU (○—○), and patients with inhibitor titres <5 BU (bsl00079bsl00079) among 31 patients who developed inhibitors during the course of the Cooperative Inhibitor Study (McMillan et al, 1988). A FVIII exposure day was defined as a day in which the patient received one or more doses of FVIII as whole blood, plasma, cryoprecipitate or concentrate. Figure and legend reproduced from: White et al (1982) (reprinted by permission from Wiley-Liss, Inc., a subsidiary of John Wiley and Sons, Inc.).

As the development of an inhibitor in a ‘true’ PTP is a sentinel, albeit relatively uncommon event, regulatory authorities are now faced with the challenge of designing the best surveillance strategy in the postmarketing phase of new or re-formulated products (FDA, 2004). The importance of such a system is emphasized by the outbreak of inhibitors that occurred in PTPs exposed to an intermediate purity pasteurized FVIII concentrate (FVIII CPS-P) in the Netherlands and Belgium in 1993 (Peerlinck et al, 1993; Rosendaal et al, 1993). These inhibitors were unusual in that they were specifically directed against the C2 domain of FVIII (Sawamoto et al, 1998). This episode was followed by a second cluster of inhibitors occurring in PTPs with haemophilia A in Belgium and Germany in 1997, following exposure to a double virus-inactivated pdFVIII concentrate (FVIII-SDP) (Peerlinck et al, 1997). This cluster may have been linked to a 40 kDa impurity in some batches of the product (Josic et al, 1999).

Voluntary postlicensure reporting continues to periodically raise the possibility that certain products may be associated with enhanced immunogenicity (Roussel-Robert et al, 2003). However, the notorious under-reporting of adverse events in pharmacovigilance systems suggests that national or even international prospective surveillance databases are more likely to provide a sensitive and accurate early warning system of product neo-antigenicity (Walker et al, 1995; FDA, 2004; UKHCDO, 2004).

Risk factors for inhibitor development

Apart from the issue of product immunogenicity are several recognized host risk factors for inhibitor development. Foremost among these are haemophilia severity, family history of inhibitors and race.

Factor VIII genotype analysis, which determines disease severity, offers some predictive value in the assessment of inhibitor risk. Among patients with severe haemophilia A, more disruptive mutations, including intron 22 inversions, large gene deletions and stop codons are associated with approximately 35% risk of inhibitor formation. In contrast, the risk of inhibitor formation is only approximately 5% in patients with missense mutations or small deletions (reviewed in Goodeve & Peake, 2003). This discrepancy may be explained by a lack of tolerizing FVIII antigen of any kind in the more severe mutations (Oldenburg et al, 2004).

In the Malmo International Brothers’ Study, 15 of 17 twin pairs were concordant for the presence of an inhibitor (Astermark et al, 2001). Data from this study and from a North American Study (Gill, 1999) suggest that approximately 50% of haemophiliac brothers of a proband with an inhibitor will also develop an inhibitor, compared with only approximately 9% of extended haemophiliac family members. Therefore, it is likely that other critical genetic factors apart from the gene defect impact inhibitor expression. As the association of inhibitors with HLA class II determinants is weak at most (Oldenburg et al, 1997), it may be posited that as yet undefined polymorphisms and mutations in other immune response and cytokine expression genes contribute to the risk of inhibitor formation. Furthermore, as discordant inhibitor status has even been noted in monozygotic twins, it should also be presumed that environmental and genetic factors contribute to the risk of inhibitor development. Recent anecdotal observations have focused attention on the possibility that a co-existent ‘inflammatory state’ may increase the likelihood of inhibitor development in the presence of infection, postoperative stress, etc. Matzinger's ‘Danger Model’ proposes that the context in which a potential immunogen (in this case, FVIII) is presented to the immune system is as important as whether it is perceived as a ‘self’ or ‘foreign’ antigen, with respect to determining whether an immune response will be mounted (Matzinger, 2002). The development of appropriate animal models to examine the influence of circulating cytokines and other ‘danger signals’ on inhibitor formation would be a significant scientific contribution.

Two recent retrospective studies have reported a decreased risk of inhibitor formation in children who avoided exposure to FVIII early in life, with very few reported inhibitors occurring in patients receiving their first exposure when >1·5 years of age (Lorenzo et al, 2001; Van der Bom et al, 2003). Notably however, these studies were not controlled for genotype, so that patients with delayed exposure may have had a milder phenotype, perhaps associated with circulating FVIII antigen. The concept that delayed exposure to FVIII has a tolerizing effect is contrary to expectations from animal models. Specifically, in haemophilia B mice and dogs, exposure to a FIX transgene product in the neonatal, but not adult animal, induced immune tolerance (Xu et al, 2003). Despite differences in immune system maturation milestones in humans, it has been speculated that gene transfer in the neonatal period (Ponder, 2003), or even in utero (White, 2003), might be more likely to induce tolerance to the transgene product.

African racial origin is a risk factor for inhibitor development, with studies consistently demonstrating up to a twofold increased risk compared with a white population reference group (Astermark et al, 2001; Lusher, 2002). The precise reason for this discrepancy has not been determined, but may reflect differences in HLA antigen repertoires or other immune response genes.

Inhibitors in moderate and mild haemophilia A

The prevalence of inhibitors in the Canadian Hemophilia Assessment and Resource Management data are 3% and 0·3% in moderate and mild haemophilia A respectively (Walker et al, 1995; Giles et al, 1998b; Food and Drug Administration, 2004). Inhibitors in milder forms of haemophilia more commonly arise under conditions in which the immune system is under intense stimulation and/or exposure to FVIII is unusually high, for example, in the postoperative period (Sharathkumar et al, 2003). It has been suggested that continuous infusion may alter the immunogenicity of the FVIII molecule (Oldenburg, 2002). However, the presence of other confounding factors including infection, tissue damage and exposure to large amounts of antigen renders it difficult to assess the role played by continuous infusion per se. Although often transient and low titre, persistent inhibitors (affecting approximately 40% of mild/moderate patients) tend to respond poorly to immune tolerance induction (ITI). These inhibitors may appear at an older age, with severe bleeding manifestations; both type I and type II inhibitor kinetics have been described.

Mutations that result in a stable abnormal conformation in the FVIII molecule are at particularly high risk for inhibitor formation in mild haemophilia A. Many high-risk mutations are clustered in the A2 and C2 domains of FVIII (Hay, 1998; Lollar, 2004). Some of these mutations provide unique ‘experiments of nature’ to appreciate the immune system's perception of FVIII as an immunogen versus a tolerogen. For example, the Arg593→Cys mutation is representative of the majority of mild haemophilia inhibitors, insofar as tolerance to both exogenous and endogenous FVIII is lost (Thompson et al, 1997). However, the inhibitor from patients with the Arg2150→His mutation neutralizes exogenous, but not endogenous, FVIII (Peerlinck et al, 1999). Correlative studies demonstrate that T-cell clones with a restricted epitope specificity in the C1 domain (corresponding to the mutation site at position 2150) could be generated from peripheral blood lymphocytes of affected individuals, and that these clones recognize wild type but not mutant (self) FVIII (Jacquemin et al, 2003).

Mechanisms of tolerance and opportunities for intervention

The clonal selection theory of self versus non-self discrimination proposes that during ontogeny, a repertoire of self and non-self reactive T-cell clones are generated by the immune system. Clones reactive to self-antigens are subsequently eliminated in generative lymphoid organs, while those specific for non-self antigens are expanded and retained. By birth, the T-cell repertoire is essentially fixed. However, tolerance induction continues throughout life, despite the fact that central tolerance – which is dependent on a functioning thymus gland – is lost early in life. The most important T-cell-dependent mechanisms that participate in peripheral tolerance – which is maintained throughout life – are (i) clonal deletion by apoptosis; (ii) anergy, in which antigen is presented without appropriate co-stimulation, leading to functional unresponsiveness of the T helper cell clone and (3) suppression, induced by the development of regulatory T cells (TReg) (Van Parijs & Abbas, 1998).

Therapeutically, tolerance to FVIII in inhibitor patients may be achieved either by co-stimulator blockade (leading to T-cell anergy) or antigen-specific tolerization of T cells. Antigen-specific tolerance could also theoretically be induced in B cells by anergy or depletion, but ongoing renewal of the B-cell repertoire throughout life makes this less likely to succeed as a long-term strategy. During the co-stimulatory blockade of T cells, the B7 and/or CD40 pathways are targeted for inhibition. In murine models of haemophilia A, where repeated injection of human FVIII induces T-cell-dependent anti-FVIII antibody formation, co-stimulatory blockade has been shown to prevent and eradicate preformed FVIII inhibitors. Intravenous injection of either murine CTLA4-Ig (Qian et al, 2000a) or a monoclonal antibody targeted to CD154 (Qian et al, 2000b) prevented the immune response to FVIII. Furthermore, the blockade of B7-1 or B7-2 and CD154 inhibited the re-stimulation and differentiation of FVIII-specific memory B cells (Hausl et al, 2004). The theoretical disadvantages of co-stimulatory blockade approaches are that the inhibition of T-cell activation will be short lived, necessitating repeated dosing, and there is a risk of undesirable non-specific immunosuppression.

These mechanisms imply that tolerance is an acquired state, which must be learned and maintained. According to this paradigm, a newborn with severe haemophilia would not be assumed to be tolerant to FVIII, unless he had been exposed to maternal blood in utero. Consequently, elucidating why it is that most severe haemophilia A patients, especially those with a higher risk genotype, do not develop inhibitors may contribute more to our understanding of tolerance mechanisms in vivo. Furthermore, even the concept that FVIII is a ‘self’ antigen that does not provoke an immune response in healthy individuals may be oversimplified. CD4+ cell responses to FVIII are detectable in both healthy subjects and haemophiliacs without inhibitors (Reding et al, 2000), and anti-FVIII IgG has been described in both normal subjects (Algiman et al, 1992) and in patients without inhibitors (Gilles et al, 1993). Thus, it may be that, during the development of a ‘normal’ immune system, active tolerization to FVIII is the rule. When this fails in congenital haemophilia, an alloantibody response is generated. Similarly, acquired haemophilia may be the result of inadequate tolerization of a normal immune response to FVIII in a previously healthy individual, resulting in the emergence of a neutralizing auto-antibody to FVIII.

The development of both regulatory T and B cells may be important in ITI. It has been reported that successful ITI may be explained by the development of neutralizing anti-idiotypic antibodies synthesized by regulatory B cells, rather than loss of the FVIII inhibitor (Gilles et al, 1996). CD4+ regulatory T cells may exist as two distinct subsets, ‘natural’ and ‘adaptive’ TReg (Bluestone & Abbas, 2003). Natural TReg cells develop in the thymus, are long lived, and may assist in the maintenance of self-tolerance, whereas adaptive TReg develop from the same population of mature peripheral CD4+ T cells that give rise to the Th1 and Th2 subsets, when antigen exposure or co-stimulation are suboptimal. Adaptive TReg cells produce immunosuppressive cytokines, such as transforming growth factor-β (TGF-β), which inhibits lymphocyte proliferation, and IL-10, which inhibits macrophage activation and expression of co-stimulator molecules. The net effects are inhibition of T-cell activation and formation of effector T cells, and/or inhibition of effector T-cell functions (Fig 1). Interestingly, it may be possible to enhance the generation of adaptive TReg cells in vivo by delivering an antigen via certain routes, such as by intranasal or oral administration. Hypothetically, exposure to a small number of ‘immunodominant’ universal T-cell epitope peptide sequences (i.e. those recognized by the vast majority of patients), particularly if delivered by the intranasal route, could be an attractive and cost-effective means to prevent inhibitor formation and/or induce FVIII-specific immune tolerance (Reding et al, 2001).

Conventional immunosuppressive agents have been used as adjunctive immunomodulatory therapy in haemophiliac patients with inhibitors. While usually indicated in the treatment of acquired haemophilia, drugs such as glucocorticoids and cyclophosphamide, have a less established role in congenital haemophilia complicated by an inhibitor. These drugs were used in the 1970s to suppress anamnaestic responses to FVIII, before falling into disfavour (Nilsson & Hedner, 1976). Recently however, interest has focused on selective B-cell depletion using RituximabTM (IDEC Pharmaceuticals, San Diego, CA, USA), a humanized monoclonal antibody to the CD20 antigen. Here again, data from case series seem to indicate a beneficial effect in acquired haemophilia (Stasi et al, 2004), and possibly in some patients with mild haemophilia A complicated by an inhibitor, where many phenotypic and laboratory features may resemble acquired haemophilia. However, the number of successful case reports in severe haemophilia A or B is extremely limited (Linde et al, 2001; Mathias et al, 2004), so that more information is required before this strategy can be endorsed.

Immune tolerance induction: past, present, and future

Currently, the only proven strategy to achieve antigen-specific tolerance to FVIII or FIX in individuals with inhibitors is the use of one of the described ITI protocols. Since the first description of successful ITI (Brackmann & Gormsen, 1977), several other protocols have been described. All are based on continuous exposure to intravenously administered FVIII/FIX over a period of weeks to months, but the dosing and frequency of administration (reviewed in Mariani et al, 2003) differ considerably. An overview summary of these protocols including daily FVIII dose, median time to success and overall success rate is presented in Table I. Although adjunctive immunomodulation has been used in some protocols – most notably in the Malmo version – there is a growing reluctance to expose affected subjects, who nowadays are usually young children, to the toxicity of agents such as cyclophosphamide or glucocorticoids.

Table I.  ITI success: influence of FVIII dose or immunosuppression.
Protocol (Immunosuppression)FVIII doseMedian Rx time (months)Success rate (%)
  1. Adapted with the permission of Blackwell Publishing from DiMichele (2003).

  2. Comparison of dosing schedule, overall success rate and time to success between the most commonly used protocols for ITI. In column 1, the presence or absence of immunosuppression is shown (+ or −) after the commonly used title of the protocol. (q.o.d., every other day; t.i.w., three times weekly; Rx, treatment). Details of these protocols can be found in Mariani et al (2003).

Bonn (−)200–300 U/kg/d15·073
Malmo (+)≈200 U/kg/d1·380
Dutch (−)25 U/kg q.o.d. t.i.w.11·583
Kasper (+/−)50 U/kg/d3·073

Several existing registries track success rates and identify favourable or adverse risk factors for successful outcome in ITI. These include the International Immune Tolerance Registry (IITR) (Mariani & Kroner, 1999), the North American Immune Tolerance Registry (NAITR) (DiMichele et al, 2002), and the German (Lenk et al, 2000) and Spanish (Haya et al, 2001) registries. The overall success rate of ITI in all these registries is comparable at 70 ± 10%, and there is broad agreement that certain favourable clinical features can be identified. These include a low inhibitor titre (<10 BU/ml) immediately prior to initiation of ITI, and a low historical peak inhibitor titre (<200 BU/ml). There is some disagreement whether (young) age at initiation of ITI or (short interval of) time between inhibitor diagnosis and initiation of ITI predict successful outcome. However, the most controversial prognostic factor remains the daily FVIII dose; whereas the IITR found that a dose ≥200 IU/kg/d was associated with a more favourable outcome, particularly in patients with inhibitor titres >10 BU/ml at the time of ITI initiation, no such association was found in the NAITR data (DiMichele, 2003). The NAITR did however demonstrate that daily dose was inversely related to duration of therapy in cases of successful ITI (DiMichele et al, 2002).

In 2002, an International Immune Tolerance Study was initiated to test the efficacy, morbidity and cost-effectiveness of low versus high dose FVIII in good risk severe haemophilia A patients with inhibitors to FVIII. Worldwide, 150 patients aged ≤7 years with a FVIII inhibitor ≤12 months from diagnosis, and a peak inhibitor titre of 5–200 BU/ml will be randomized to FVIII 50 IU/kg three times a week or 200 IU/kg/d, beginning when the inhibitor titre is <10 BU/ml. As of March 2004, 29 patients have been randomized ( The study hypothesis is that high dose ITI will achieve tolerance more rapidly than low dose ITI, but will have a similar overall success rate. The type and purity of factor concentrate is not controlled, although this might be an important variable, because a preliminary data from the NAITR suggested a better outcome with monoclonal FVIII compared with rFVIII (DiMichele & Kroner, 2000). Furthermore, it has been suggested that switching from high to intermediate purity FVIII may rescue some patients who appear to be failing ITI (Kreuz et al, 1996).

In the International Study, tolerance is defined as a FVIII recovery >66% expected and half-life >6 h. In general, this endpoint is achieved several months after the inhibitor is no longer detectable by the Bethesda or Nijmegen–Bethesda assays (Fig 3). It is also apparent that if an inhibitor remains detectable after 2 years of ITI, the probability of ultimate success is vanishingly small. Although commonly employed, it is unknown whether continued prophylaxis at the conclusion of ITI is required to maintain a state of tolerance (DiMichele et al, 2000). Typically, the relapse rate after successful ITI is <5% (Mariani et al, 2003).

Figure 3.

Time course of disappearance of FVIII inhibitor during ITI. Note that the disappearance of detectable inhibitor (as measured in the laboratory by a Bethesda assay) precedes the time to true tolerance (defined by normal FVIII kinetics) by several months. Data are from the International Immune Tolerance Registry. Reproduced from Mariani et al (2003) (reprinted by permission from Thieme Medical Publishers).

An ancillary study will examine the response to ITI as a function of FVIII genotype. One study reported a lower response rate in patients with a large gene deletion or frameshift mutation compared with those with inversions, nonsense, splice junction or missense mutations (Nakaya et al, 2004). On the contrary, a recent update from the Hemophilia Inhibitor Study reported that large deletions were associated with shorter time to tolerance induction compared with inversion 22 mutations (Ragni et al, 2003).

Haemostatic agents for the management of bleeding and surgery in inhibitor patients

The management of bleeding episodes in patients with inhibitors should take into consideration the current inhibitor titre, the potential for anamnaestic response to FVIII-containing products, and the historical responsiveness of the patient to bypassing therapies. The potential for adverse events, including thrombosis, also needs to be considered when using any FVIII/IX bypassing agent.

Infusion of high dose FVIII is usually preferable in patients with an inhibitor titre <5–10 BU/ml experiencing life or limb-threatening haemorrhage. Extracorporeal immunoadsorption offers the possibility of rapid reduction of plasma inhibitor levels in patients with higher titres, thus facilitating high dose FVIII infusion. The most efficient devices for antibody removal utilize a column containing either staphylococcal protein A (ImmunosorbaTM; Fresenius Hemocare, Redmond, WA, USA) or polyclonal sheep antibodies against human FVIII (Ig-TherasorbTM; Plasma Select, Teterow, Germany) to remove large quantities of inhibitor (Jansen et al, 2001; Freedman et al, 2003). Currently, the Immunosorba system, which utilizes two protein A columns in tandem, is available in the United States under a Humanitarian Device Exemption from the Food and Drug Administration to prepare inhibitor patients for surgery. Theoretically, the use of anti-idiotypic antibodies (Gilles et al, 2004) or ‘decoy’ peptides mimicking B-cell epitopes on FVIII (Villard et al, 2003) could also be used as an alternative short-term means to reduce FVIII inhibitor titres.

Porcine FVIII is a reasonable alternative for selected patients with demonstrated low titre inhibitors to porcine FVIII; overall, anti-human FVIII inhibitors demonstrate a median cross reactivity with porcine FVIII of approximately 15%. Recently however, the availability of porcine FVIII (Hyate CTM; Beaufour-Ipsen, Wrexham, UK) has been severely restricted due to concerns about possible contamination with porcine parvovirus. A phase I study examining the safety and pharmacokinetic profile of a B domain-deleted recombinant porcine FVIII (OBI-1TM; Octagen Corporation, Bala Cynwyd, PA, USA) is currently underway, and a phase II study of bleeding in patients with congenital haemophilia A complicated by a FVIII inhibitor will be initiated shortly.

Bypassing agents induce haemostasis in the absence of FVIII/FIX. (Activated) prothrombin complex concentrates [(a)PCCs] have been used for more than 25 years, despite an incomplete understanding of their mechanism of action. Randomized controlled studies performed in the early 1980s demonstrated a 50–65% efficacy of PCC and aPCC concentrates in the management of joint bleeds, compared with a 25% efficacy rate for placebo (Barthels, 1999). Uncontrolled series demonstrate an efficacy of 80–90% for factor eight inhibitor bypassing activity (FEIBA) or FEIBA-VH in the management of haemorrhagic events (Negrier et al, 1997). Recent data suggest that FEIBA's mechanism of action is probably mediated by a complex of prothrombin and FXa (Turecek et al, 2004). Potential drawbacks of aPCCs include the absence of a monitoring strategy, triggering of an anamnaestic response to FVIII because of the presence of trace amounts of FVIII, and occasional thrombotic events after repetitive high dose administration.

The efficacy, safety, mechanism of action and pharmacokinetics of recombinant activated factor VII (rFVIIa) have been reviewed elsewhere (Abshire & Kenet, 2004). While the recommended dose in haemophilia is 90–120 μg/kg every 2–3 h, it has become apparent that some patients require a higher initial ‘mega-dose’ (200–300 μg/kg) for adequate haemostatic control (Kenet et al, 2003). This impression is supported by a recent retrospective analysis of the US Hemophilia and Thrombosis Research Society's multi-centre registry, which reported a haemostatic efficacy of 97% (n = 119 bleeds) for doses in the range of 200–346 μg/kg, compared with 84% (n = 336 bleeds) for doses <200 μg/kg (P < 0·001) (R. Parameswaran, A.D. Shapiro, J.C. Gill and C.M. Kessler, submitted). Recombinant FVIIa appears to have a low but finite risk of thrombosis (Abshire & Kenet, 2004), and also lacks a suitable monitoring strategy (Key & Nelsestuen, 2004).

The weak enzymatic activity of FVIIa may be enhanced by selectively mutating amino acid residues to enhance membrane binding (Nelsestuen et al, 2001) or intrinsic (tissue factor-independent) activity (Tranholm et al, 2003). Recently, a novel strategy using an adeno-associated viral vector to deliver a modified FVII transgene that can be intracellularly processed and secreted as activated FVII was described in haemophilia B mice (Margaritis et al, 2004). Gene transfer resulted in long-term expression and phenotypic correction of haemophilia, raising the possibility that gene transfer might be feasible in selected patients with inhibitors.

To date, there has been no prospective randomized trial directly comparing the efficacy and safety of aPCC and rFVIIa. However, the International Feiba NovoSeven Comparative (FENOC) Study comparing a single dose of FEIBA (75 U/kg) to two doses of rFVIIa (90 μg/kg) for the treatment of haemarthrosis in a crossover design is underway. Another US study comparing one dose of FEIBA (70 U/kg) to two dosage levels of rFVIIa (270 μg/kg × 1 dose, or 90 μg/kg × 3 doses) in the treatment of haemarthrosis is nearing completion.

Traditionally, elective surgery has been avoided whenever possible in high responder inhibitor patients. However, over the past decade, cumulative experience with FVIIa (usually administered with an anti-fibrinolytic agent) (Hvid & Rodriguez-Merchant, 2002) and a recent re-evaluation of FEIBA's efficacy (Tjonnfjord et al, 2004) during minor and major elective surgery has concluded that haemostasis may be achieved in the majority of instances with an acceptably low rate of adverse events. A longstanding debate focussing on whether FVIIa administered by intermittent bolus is superior to continuous infusion during surgery awaits the result of a recently completed randomized study.

Inhibitors in FIX deficiency (haemophilia B)

Inhibitors in severe haemophilia B occur at a lower frequency than in haemophilia A, although the cumulative incidence may be as high as 4–5%. More than 80% are of the high responder type. Inhibitors in patients with non-severe disease [FIX coagulant activity (FIX:C) >1 U/dl] are very rare. In fact, genotype is a strong determinant of inhibitor risk; patients with gene deletions or rearrangements are at high risk of inhibitor formation (approximately 50%), as are those with frameshift, premature stop, or splice-site mutations (approximately 20%). Missense mutations, which constitute the majority of genotypes in haemophilia B, are at very low risk of inhibitor formation.

Inhibitors in haemophilia B also appear after a median of only 9–11 ED , and the incidence with rFIX is equivalent to that of pdFIX (Poon et al, 2002). However, several clinical features distinguish FIX inhibitors. First, anaphylaxis or anaphylactoid reactions to FIX-containing products may accompany the development of an inhibitor. Data for 88 patients in the ISTH Registry on FIX inhibitors were recently analysed (Warrier, 2003). Of these, 51 (58%) had a history of anaphylaxis to FIX-containing products. Of the 32 subjects who were genotyped, 75% (24/32) had a total gene deletion. The relative lack of success of ITI is another feature that distinguishes FIX inhibitors. In the ISTH Registry, ITI was successful in only five of the 34 patients (15%) in whom it was attempted, two of whom had low responder inhibitors. Furthermore, 13 of these 34 patients (38%) developed nephrotic syndrome as a result of ITI. This complication is not seen in patients undergoing ITI for FVIII inhibitors, which may reflect the much lower amounts of FVIII in plasma (approximately 50 times less on a mg/ml basis) compared with FIX. Nephrotic syndrome has only been reported in subjects who have also experienced allergic reactions to FIX, and has generally responded to withdrawal of FIX-containing products.

Close supervision of patients with a known family history of an inhibitor, and/or a known high-risk inhibitor/anaphylaxis genotype during the first 20 or so FIX exposure days is advisable. Recombinant FVIIa may be a suitable treatment choice for many patients with inhibitors and allergic phenomena to FIX.

Inhibitors in FXI deficiency

Until recently, little was known of the prevalence of inhibitors in patients with congenital deficiency of FXI. In a study of 118 Israeli patients, seven (6%) had an inhibitor, with titres ranging from 3 to 25 BU/ml (Salomon et al, 2003). All seven individuals had received plasma replacement therapy in the past. Genotype analysis demonstrated that all affected patients were homozygous for the so-called type II Jewish mutation (Glu117Stop), which encodes for a baseline FXI:C <1 U/dl. This genotype association was striking, with 33% (seven of 21) inhibitor prevalence in this subgroup, compared with 0% prevalence in other genotypes. Therefore, judicious use of plasma products in type II homozygotes might be advisable, with anti-fibrinolytic agents providing a reasonable alternative in some situations. For the management of bleeding that is unresponsive to anti-fibrinolytics, case reports have described the successful use of rFVIIa (Lawler et al, 2002). There are no good data on the responsiveness to ITI in patients with an established FXI inhibitor.

Inhibitors in other rare bleeding disorders

In the North American Rare Bleeding Disorder Registry, data on 294 individuals with deficiencies of FII, FV, FVII, FX, FXIII, fibrinogen and dysfibrinogenaemia were analysed for the presence of reported inhibitors. Only 3% of patients with FV (n = 1) and FXIII (n = 1) deficiency developed inhibitors following infusion of FFP and FXIII concentrate (Acharya et al, 2004). The literature also contains rare reports of inhibitors in congenital deficiency of FVII and FX.

In type 3 (severe) von Willebrand disease, inhibitors have been described as a complication of treatment, although there are scant data on their true prevalence. Inhibition of VWF binding to FVIII, ristocetin, glycoprotein Ib, glycoprotein IIb/IIIa, and type III collagen have all been described (Batlle et al, 1997; Tout et al, 2000). Bleeding may respond to rFVIII concentrate, administration of platelets, or rFVIIa (Ciavarella et al, 1996).


In the last decade, remarkable progress has been achieved in the management of patients with coagulation factor inhibitors. Illustrative of this statement are data demonstrating that, while the HIV-independent mortality rate in patients with severe haemophilia A and FVIII inhibitors doubled between 1977 and 1992, it subsequently declined during the period 1993–1999, to a level that was identical to patients without inhibitors (UKHCDO, 2004). This startling trend is probably explained by more widespread implementation of ITI, combined with earlier diagnosis of potentially life-threatening bleeding and an improved armamentarium of haemostatic agents. Coincident with improved life expectancy has been an improvement in quality of life, which may now be similar to severe haemophiliacs without inhibitors (Gringeri et al, 2003). The challenge for the next decade will be to improve our understanding of the immunobiology of inhibitor formation, in order to design potent specific and cost-effective therapies to prevent and treat existing inhibitors.


This study was supported in part by grant HL65578 from the National Institutes of Health. The author would like to acknowledge Mark Reding MD, Art Thompson MD and Amy Shapiro MD for helpful discussions and sharing of unpublished data. The technical assistance of Carol Taubert and Omer Aras MD is also gratefully acknowledged.