Immune tolerance in haemophilia: the long journey to the fork in the road


Correspondence: Dr. Donna M. DiMichele, Division of Blood Diseases and Resources, National Heart Lung and Blood Institute, 6701 Rockledge Drive, Bethesda, MD, USA 20892.



Antibody eradication is the ultimate goal of inhibitor management. The only clinically proven strategy for achieving antigen-specific tolerance to factor VIII is immune tolerance induction (ITI). Our knowledge about ITI in haemophilia A and B was, historically, derived from small cohort studies and retrospective national and international ITI registries. Practice is now further influenced by prospective cohort data, and the results of a single prospective randomized international ITI trial. However, due to the low incidence of inhibitors in haemophilia B, there are few comparable data from which to develop a useful evidence-based approach to the prevention and eradication of factor IX inhibitors. The lack of an effective strategy is particularly problematic given the morbidity associated with the unique occurrence of allergic and anaphylactic reactions that often herald factor IX antibody development and preclude effective eradication. This paper will discuss our current understanding of immune tolerance outcome and outcome predictors for haemophilia A and B; review the current consensus practice recommendations for ITI; and summarize the emerging body of immunological science relating to antibody formation and tolerance. It will conclude by suggesting how our knowledge might inform the future investigative priorities impacting the therapeutic and preventative tolerance strategies of tomorrow.

The development of neutralizing antibodies remains a frequent and serious complication of haemophilia replacement therapy. Life threatening allergic reactions may specifically complicate the development of anti-factor IX antibodies in up to 4% of patients with mostly severe haemophilia B (DiMichele, 2007). However, factor VIII (FVIII) inhibiting antibodies (inhibitors) present much more commonly following replacement therapy in haemophilia A, creating a greater burden of clinical disease in this patient population (Chambost, 2010). This is particularly the case with children with severe haemophilia A. A recent analysis of the United Kingdom Haemophilia Centre Director Organization (UKHCDO) database demonstrated that both overall and high titre inhibitor incidence was highest in the 0–4 age group (64·3 and 36·1,respectively, per 1000 at risk person years) (Hay et al, 2011). Furthermore, 74% of all inhibitors (73·7 per 1000 at risk patient -years) first developed before 10 years of age, and 53%of these were high titre antibodies, as defined by a historical peak titre of ≥5 Bethesda units (BU) (Hay et al, 2011). Although less common in non-severe haemophilia A, the development of inhibitors that cross-neutralize endogenous FVIII both negatively impacts the usually mild bleeding phenotype and substantially increases lifelong morbidity due to the general difficulty in eradicating these antibodies (Kempton et al, 2010). Given the inhibitor-associated morbidity resulting from the limited effective treatment options, antibody eradication is the ultimate goal of inhibitor management.

The only clinically proven strategy for achieving antigen-specific tolerance to FVIII or factor IX (FIX) is immune tolerance induction (ITI). First reported over 30 years ago (Brackmann & Gormsen, 1977), much of our current knowledge about ITI in haemophilia A and, to a much lesser extent, in haemophilia B, was derived from small cohort studies (Gruppo et al, 1992; Mauser-Bunschoten et al, 1994; Brackmann et al, 1996; Freiburghaus et al, 1999; Smith et al, 1999; Rocino et al, 2001) and retrospective national and international ITI registries (Mariani et al, 1994; Lenk, 2000; DiMichele & Kroner, 2002). More recently, practice is being influenced by prospective cohort data, and the results of a single prospective randomized ITI trial designed to answer a few outstanding questions about ITI optimization in haemophilia A (Hay & DiMichele, 2012). Due to the relatively low incidence of development (1·5–3%) in haemophilia B patients, there are few comparable data from which to develop a useful evidence-based approach to the prevention and eradication of FIX inhibitors (DiMichele, 2007). The lack of an effective strategy is particularly problematic given the even greater morbidity associated with the almost unique occurrence of allergic, anaphylactoid or frankly anaphylactic reactions that accompany, and often herald, FIX antibody development, complicating attempts to eradicate FIX inhibitors (DiMichele, 2007).Ultimately, successful inhibitor prevention and eradication strategies for both haemophilia A and B will emerge from the clinical translation of our evolving knowledge of the immune mechanisms involved in antibody development and immune tolerance.

This paper, an update of a previously published review of this topic (DiMichele, 2011), will discuss our current understanding of immune tolerance outcome and outcome predictors for haemophilia A and B; review the current consensus practice recommendations for ITI; and summarize the emerging body of immunological science on antibody formation and tolerance. It will conclude by suggesting how current knowledge might inform the future investigative priorities impacting the therapeutic and preventative tolerance strategies of tomorrow.

Haemophilia A: immune tolerance for FVIII inhibitors

The role of host factors in ITI outcome

The historical ITI success rates of >50% in haemophilia A have been defined by variably stringent clinical and laboratory endpoints (Mariani et al, 2003). By international consensus, successful ITI is currently defined as both an undetectable inhibitor titre (≤0·6 Bethesda Units (BU) by Bethesda or Nijmegen assays), and normalized FVIII pharmacokinetics (Consensus Proceedings from the Second International Conference on Immune Tolerance Therapy. Bonn, 1997, unpublished). Among host- related variables known to influence inhibitor development, only race and F8 genotype have been studied in relation to ITI outcome.

The North American Immune Tolerance Registry (NAITR) retrospectively examined race as an outcome predictor. ITI success rate among ethnic Africans (14/21; 67%), as well as Hispanics and Latinos (11/19; 58%) was statistically indistinguishable from that of subjects of other races (99/139, 71%) (DiMichele & Kroner, 2002). A recent single institution retrospective analysis did suggest a significantly lower ITI success rate among African Americans (AAs) (13/23, 58%) when compared with Caucasians (11/12, 92%; P = 0·04); however this difference was largely attributable to higher pre-ITI titres in the AA cohort (Callaghan et al, 2010). Race and ethnicity were examined in a secondary analysis of the International ITI (I-ITI) data and found not to have impacted ITI outcome (Hay & DiMichele, 2012). Notably, Blacks represented only 8% of enrolled subjects.

The impact of F8 genotype on ITI outcome was examined in the Italian PROFIT study (Coppola et al, 2009). A higher ITI success rate of 81% was noted among 16/86 high responder (HR) (historical titre ≥ 5 BU) subjects with F8 mutations predictive of a lower inhibitor risk, compared with that of 47% in 70 subjects with high risk mutations (P = 0·01). Although less significant than pre-ITI and ITI peak titres, F8 genotype remained a predictor of ITI success in a multivariate analysis of this cohort (Odds Ratio 6·2, 95% confidence interval 1·1–36·0, P = 0·04) (Coppola et al, 2009).

The inter-relationship between race and both F8 genotype and haplotype, and its impact on the immunological mechanisms of inhibitor development in severe haemophilia A patients is currently being investigated. These data may ultimately inform the complexity of the host immune response to FVIII, and provide valuable clues to more effective prevention and eradication of inhibitors in populations that are less responsive to our current treatment strategies.

The role of treatment factors in ITI outcome

The International Immune Tolerance Registry (IITR), the German registry and the NAITR all retrospectively identified parameters that influence both success rate and time to successful ITI in severe haemophilia A high titre inhibitor patients (Mariani et al, 1994; Lenk, 2000; DiMichele & Kroner, 2002). Lower pre-ITI, historical peak and on-ITI peak titres have all statistically correlated with ITI success (Mariani et al, 1994; Lenk, 2000; DiMichele & Kroner, 2002). The role of inhibitor titre, including pre-ITI titre, is examined prospectively in the Italian PROFIT registry (Coppola et al, 2009). In the I-ITI study, pre-ITI titre <10 BU was a primary determinant of the ‘good risk’ cohort that defined subject eligibility and, consequently, could not be further studied. However, in a univariate logistical regression model of host- and treatment- related factors relative to ITI outcome, historical peak titre (P = 0·02) and peak titre on ITI (P = 0·002) were the only two variables to correlate significantly with outcome (Hay & DiMichele, 2012). In the multivariate analysis, ITI peak titre retained a significant inverse correlation with ITI success (P = 0·002), thus prospectively validating, in a well-defined subset of ITI subjects, prior observations made by retrospective data analysis (DiMichele & Kroner, 2002).

Additional treatment variables examined in at least one registry and found to be variably predictive of outcome include older age at ITI initiation (Mariani et al, 1994; Lenk, 2000; DiMichele & Kroner, 2002); an interval of >5 years between inhibitor diagnosis and start of ITI (Mariani et al, 1994; Lenk, 2000; DiMichele & Kroner, 2002); and an ITI interruption of more than 2 weeks (Lenk, 2000). The I-ITI study also used these variables to define the ‘good risk’ cohort studied in this trial, and, therefore, did not generate further data on their predictive role in ITI outcome (Hay & DiMichele, 2012).

FVIII dose and dosing regimen

The role of FVIII dosing regimen in the eradication of HR inhibitors has historically provoked the greatest debate due to the conflicting data generated by the IITR and NAITR (Mariani et al, 1994; Lenk, 2000; DiMichele & Kroner, 2002). While the IITR observed better ITI outcomes with a daily FVIII dose of ≥200 u/kg per d, the NAITR reported the more significant impact of FVIII dose (≥50 u/kg per d) on the time to successful tolerance (Mariani & Kroner, 2001; DiMichele & Kroner, 2002). Ultimately, a meta-analysis of both registries determined that, for patients with historical inhibitor titres <200 BU and immediate pre-ITI titres <10 BU, FVIII dose did not impact ITI outcome (Kroner, 1999).

Understanding the comparative cost effectiveness and morbidity associated with non-daily lower dose regimens may be crucial to the broader availability of ITI in both the developed and developing world, thus requiring prospective study. The I-ITI trial randomized 134 good risk severe haemophilia inhibitor subjects from 55 participating centres to receive either 200 u/kg per d or 50 u/kg thrice weekly FVIII for up to 33 months to prospectively examine the effect of FVIII dose on both the overall rate of ITI success and time to success (Fig 1) (Hay & DiMichele, 2012). Among the 66 subjects who reached a study endpoint prior to early termination of the trial, 70% achieved ITI success in either study arm (Hay & DiMichele, 2012). However, due to the limitations of sample size, true equivalence in effectiveness between high and low dose ITI could not be established. Although the study did not observe a FVIII dose effect on the overall time to tolerance, dose was noted to have impacted the time to achievement of both a negative titre and normal FVIII recovery (Table 1) (Hay & DiMichele, 2012).

Figure 1.

Flow-diagram summarizing the protocol for the study, randomization, and monitoring of immune tolerance induction (ITI) and prophylaxis. BU: Bethesda Unit. This research was originally published in Blood. (Hay & DiMichele, 2012). © the American Society of Hematology.

Table 1. Time to achieve ITI milestones by treatment arm in the International Immune Tolerance Study (Hay & DiMichele, 2012)
  N Low dose N High dose P
  1. ITI, immune tolerance induction.

  2. Values are reported as median (interquartile range) months.

  3. This research was originally published in Blood. (Hay & DiMichele, 2012). © the American Society of Haematology.

Phase 1

(Start of ITI to negative titre)








Phase 2

(Negative titre to 1st normal recovery)








Phase 3

(Normal recovery to tolerance)








The I-ITI study did establish an unanticipated but significant impact of dose on ITI-related bleeding morbidity. As shown in Table 2, the intercurrent all rate of all bleeding while on ITI, i.e., haemorrhages/month, was significantly higher in the low dose arm of ITI throughout all phases of ITI, but most significantly so during the interval between that start of ITI and the achievement of a negative inhibitor titre when the bleed rate was over twofold higher on the low dose non-daily ITI regimen (P = 0·00024) (Hay & DiMichele, 2012).

Table 2. All bleed-rate (bleeds per month) by treatment arm and phase of ITI in the International Immune Tolerance Study (Hay & DiMichele, 2012)
 Dose (n)MeanMedianIQRP value
  1. ITI, Immune tolerance induction; IQR, interquartile range.

  2. Phase 1: Start of ITI to negative titre; Phase 2: Negative titre to 1st normal recovery; Phase 3: Normal recovery to tolerance); Phase 4: Prophylaxis following successful tolerance.

  3. This research was originally published in Blood. (Hay & DiMichele, 2012). © the American Society of Haematology.

Phase 1Low (58)0·6230·564(0·093–0·886)0·00024
High (57)0·2820·000(0·000–0·440)
Phase 2Low (27)0·1270·000(0·000–0·073)0·283
High (22)0·0870·000(0·000–0·000)
Phase 3Low (24)0·1500·000(0·000–0·148)0·552
High (22)0·0330·000(0·000–0·000)
Phase 4Low (24)0·1750·150(0·000–0·290)0·112
High (22)0·1020·000(0·000–0·231)

Since one major objective of early tolerance induction is the reduction of haemorrhage-related morbidity, these data add another important dimension to future evaluations of the efficacy and cost-effectiveness. Therefore, I-ITI investigators are performing further analyses of this dataset with respect to both the epidemiology of bleeding on ITI, as well as the impact of bleeding on the pharmacoeconomics of ITI.

FVIII product type

The role of FVIII product type in predicting ITI outcome has been a question raised by a retrospective analysis of the Frankfurt experience suggesting a lower ITI success rate (29%) with high purity FVIII, compared to historical outcomes with the prior use of von Willebrand factor (VWF)- containing product (91%) (Ettingshausen & Kreuz, 2005). Neither the IITR nor the NAITR could corroborate the Frankfurt observation due to the skewed distribution of product use in each registry (Mariani et al,1994; Lenk, 2000; DiMichele & Kroner, 2002). As recently reviewed (Di Minno & Coppola, 2011), tolerance could be successfully achieved in good and poor risk patients with either recombinant or plasma-derived FVIII (Bray et al, 1994; Lusher, 1994; Batlle et al, 1999; Courter & Bedrosian, 2001; Orsini et al, 2005; Barnes et al, 2006; Rocino et al, 2006; Bidlingmaier et al, 2011). Three retrospective analyses of American and Italian poor risk cohorts, further suggested successful ITI with VWF-containing FVIII products may be further influenced by the epitope specificity of the anti-FVIII antibody (Gringeri et al, 2007a; Greninger et al, 2008; Kurth et al, 2008).

International prospective randomized studies, controlled for other variables known to impact ITI outcome, will be required to definitively determine the role of product type in successful tolerance. Due the investigator choice of recombinant FVIII for 90% of ITI courses conducted in this trial, the I-ITI study did not shed further light on this issue (Hay & DiMichele, 2012). The RESIST study plans to examine this question directly by randomizing ITI-naïve poor risk haemophilia inhibitor patients to receive 200 u/kg per d of either recombinant or plasma-derived VWF- containing FVIII (Gringeri et al, 2007ab). Moreover, both the ongoing Italian ITI Registry (Coppola et al, 2009) and multinational Observational ITI (ObsITI) study (Kreuz, 2008) are prospectively examining ITI outcome predictors, including FVIII product type.

The past, current and future role of immune modulation in ITI

Cyclophosphamide and immunoglobulin were historically part of the Malmö ITI regimen (Nilsson et al, 1988). However, later analyses of this protocol demonstrated no outcome advantage relative to standard ITI (Freiburghaus et al, 1999; Berntorp et al, 2000; Rivard et al, 2003). Furthermore, concerns about cyclophosphamide-associated complications and the technical difficulty of performing extracorporeal immunoadsorption, limit the use of this strategy in young children.

Renewed interest in immune modulation has focused on selective B-cell depletion using rituximab (IDEC Pharmaceuticals, San Diego, CA, USA), a humanized monoclonal antibody to B cell CD20 antigen. Compared to acquired haemophilia (Franchini et al, 2006; Stachnik, 2006; Sperr et al, 2007) the published experience with rituximab for ITI in congenital haemophilia has been limited and outcomes difficult to interpret because of small cohort size, protocol variability and potential positive reporting bias. The most comprehensive review of outcomes on rituximab for ITI was the result of a survey conducted by the UK Haemophilia Centre Doctors' Organization (UKHCDO) (Collins et al, 2009). Fifteen patients were found to have been treated with 4–12 infusions of rituximab administered as four consecutive weekly doses of 375 mg/m2. Among these, 6 (40%) had an initial complete response (CR) defined by negative titre alone, and an additional 4 (27%) achieved an initial partial response (PR) defined by a titre <5 BU and effective FVIII prophylaxis. A sustained CR or PR was ultimately achieved in 7 (47%) patients over an observational period of >108 weeks. Interestingly, among the 12 patients who received concomitant FVIII at doses of 100–200 u/kg per d, 10 (83%) individuals achieved an initial CR (6) or PR (4); of these, 7 (58%) were ascertained to have had sustained clinical benefit. Conversely, none of the three patients treated in the absence of concomitant FVIII achieved any response to rituximab therapy (Collins et al, 2009). Overall, the success rate for this cohort was lower when compared to the cumulative results summarized from the previously published literature (Collins, 2006). The frequency of immediate and long-term complications of anti-CD20 therapy in children is still unclear, but continues to be progressively informed through therapeutic intervention studies in disorders other than haemophilia. Infusion reactions including nausea and headache, as well as serum sickness and opportunistic infections have so far occurred infrequently in children (Collins, 2006; Collins et al, 2009). To date, the progressive multifocal leucoencephalopathy, previously described in persons with autoimmune and lymphoproliferative disorders who received prolonged rituximab or combined immunosuppressive therapy, has not been reported in persons with haemophilia (Carson et al, 2009). Nonetheless, caution and careful surveillance must be exercised as more aggressive immunosuppressive therapeutic approaches are applied to antibody eradication in congenital haemophilia (Uprichard et al, 2009). Additional prospective safety and efficacy outcome data is required before the more widespread implementation of these strategies in children with inhibitors. At least two such efforts are underway in the US (C. Leissinger and R. Kruse-Jarres, Tulane University School of Medicine, New Orleans, LA, USA, personal communication). Finally, expansion of earlier efforts to study the immunological mechanisms underlying short and long term tolerance in congenital haemophilia will be required to optimize anti-CD20 therapy (Kessel et al, 2008; Zhang et al, 2010).

Considerable progress in our understanding of the immunology of FVIII antibody development and eradication may eventually enable a safe and more uniformly successful approach to treatment (Waters & Lillicrap, 2009). The mechanisms of central and peripheral tolerance; the antigen-presenting roles of dendritic (+/− cytokine co-stimulation) and memory B cells in primary and secondary anti-FVIII immune responses, respectively; the immune regulatory effect of CD4+, CD8+, CD4+/CD25+ T regulatory cells, specific T-B cell co-stimulatory interactions and apoptotic fibroblasts on these responses; as well as the respective roles of memory B and plasma cells and anti-idiotype antibodies in ongoing anti-FVIII antibody production are all now better understood (Waters & Lillicrap, 2009; Ettinger et al, 2009; Matsui et al, 2009; Kallas et al, 2010; Su et al, 2009; Reipert et al, 2010a; van Helden et al, 2010; Gilles, 2010; Sule et al, 2012; Su et al, 2011; Sakurai et al, 2011; Pautard et al, 2011; Irigoyen et al, 2012).

However, to date, all potentially important immunological mechanisms of tolerance have been studied in animals (Matsui et al, 2009; Su et al, 2009, 2011; Waters & Lillicrap, 2009; Gaitonde et al, 2010; Kallas et al, 2010; Reipert et al, 2010b; Sule et al, 2012); advances in mouse, dog and non-human primate models have recently been reviewed and/or reported (Reipert et al, 2007; van Helden et al, 2011; Qadura et al, 2011). Several such observations have been or are being corroborated through the in vitro study of haemophilic inhibitor plasma (Waters & Lillicrap, 2009; Ettinger et al, 2009; Reipert et al, 2010a; van Helden et al, 2010; Gilles, 2010; Sakurai et al, 2011; Pautard et al, 2011; Irigoyen et al, 2012). In fact, there have been few in vivo corroborative studies. Nonetheless, immunologists appreciate that the immune mechanisms responsible for the down modulation of an established immune response and induction of permanent tolerance now require prospective study in collaboration with clinical ITI trials (Reipert et al, 2007). Several ancillary studies of the immunology of tolerance are currently being conducted in conjunction with the I-ITI Study and more are sure to follow.

Finally, ongoing generation of animal model data continues to suggest that the use of gene transfer technology to modulate immune responses with the goal of FVIII inhibitor prevention and treatment remains an intriguing future therapeutic possibility (Scott & Lozier, 2011). The requirements for adoptive gene transfer of B and T cell-based primary and secondary immunity in mice are currently being studied (Hu et al, 2010; Scott, 2010). Furthermore, the successful induction of tolerance in an adult with a longstanding high titre FVIII inhibitor in the months following a liver transplantation provided possible proof of principle for the potential future efficacy of FVIII gene transfer in eradicating pre-existing antibodies (Ashrani et al, 2004). Further proof of principle has emerged from successful long term inhibitor eradication of high titre FVIII inhibitors in three of four haemophilic dogs who underwent liver-directed gene transfer with a canine FVIII-AAV plasmid constructs (Finn et al, 2010). These and future studies may hold the therapeutic key to effective tolerance protection and re-establishment.

Translation into clinical practice: recommendations for ITI in severe haemophilia A patients with high titre inhibitors

Several years ago, three independent groups developed carefully graded consensus recommendations for the current practice of ITI in patients with severe haemophilia A and high titre (>5 BU) inhibitors, based on a critical review of the existing literature. These included the European Consensus Panel (ECP) (Astermark et al, 2006a), the International Consensus Panel (ICP) (DiMichele et al, 2007), and the UKHCDO (Hay et al, 2006).

All agreed that, despite the importance of beginning therapy soon after inhibitor confirmation, deferring ITI initiation until the titre is <10 BU is preferable (Grade B, Level IIB or III evidence). The UKHCDO and the ICP also recommend avoidance of FVIII exposure and the use of non-amnestic bypass therapy for the prevention and treatment of bleeding during the deferral period (Grade B, Level IIB or III evidence). The ICP added that ITI should be started at >10 BU if the titre does not decline over a period of 1–2 years and/or if inhibitor development or persistence is associated with severe or life-threatening bleeding (Grade C, Level IV evidence) (Astermark et al, 2006a; Hay et al, 2006; DiMichele et al, 2007).

All groups agreed that, prior to the analysis of data from the I-ITI trial, insufficient information existed to support strong recommendations about the initial ITI dosing regimen for good risk inhibitor patients. There was uniform consensus about a critical need for reliable unbiased data, and each group independently recommended that all eligible ITI patients be enrolled in a prospective randomized clinical trial, or, alternatively, in a national or international registry (Grade B or C; Level IIB or III evidence). At this time, several I-ITI investigator panels are conducting critical reviews of the I-ITI study results in order to determine their evidence-based impact on current national ITI practice guidelines for good risk FVIII inhibitor patients.

The ICP also noted that for poor risk ITI patients, defined by a historical titre of >200 BU and/or a pre-ITI titre of >10 BU and/or an interval of >5 years since inhibitor diagnosis, efficacy data is limited to dosing regimens of ≥200 u/kg per d (Grade C, Level III evidence). Participation in prospective comparative clinical trials for such patients was strongly supported (Grade B, Level IIB evidence) (Astermark et al, 2006a; Hay et al, 2006; DiMichele et al, 2007).

All groups also independently concluded that ITI has been successfully performed using recombinant and plasma-derived FVIII replacement therapy (usually the product on which the patient developed the inhibitor), and that none of the published data supported the superiority of any single product type for ITI, pending the results of the RESIST study (Grade C, Level IV evidence). The ECP and ICP added that VWF- containing FVIII concentrates should be considered for patients who fail ITI using high purity FVIII (Grade B, Level IIB, III evidence) (Astermark et al, 2006a; Hay et al, 2006; DiMichele et al, 2007).

The ECP and ICP recommended that there is no current role for immune modulation in primary ITI. However, whereas the ECP concluded that it could not recommend rituximab as an alternate ITI strategy without accrual of additional experience with the drug (Grade C, Level IV evidence), the ICP suggested that rituximab be considered in the case of ITI failure, but in conjunction with participation in an ongoing clinical trial or registry (Grade B, Level IIB evidence). Both advocated for additional safety and efficacy data in haemophilia patients on this and other immune modulatory drugs in advance of any recommendation of the widespread implementation of these treatment modalities (Astermark et al, 2006a; DiMichele et al, 2007).

Although not previously discussed within the context of this paper, supportive care during ITI may be critical to a successful outcome. Both the ICP and ECP included recommendations for safe and reliable venous access, as well as central venous access device (CVAD) use guided by the preferential insertion of Port-a Caths and consensus guidelines for catheter care (Grade B, Level III evidence) (Ewenstein et al, 2004; Astermark et al, 2006a; DiMichele et al, 2007). Ongoing safety and efficacy data continues to be regularly published and will continue to inform ITI practice (Mancuso & Berardinelli, 2010; Hay & DiMichele, 2012). In the I-ITI study, although 86% of subjects received ITI through at least one CVAD, 73% had a catheter placed for routine treatment of bleeding prior to ITI initiation. The frequency of catheter use did not differ significantly between treatment arms. Interestingly, no difference in the rates of CVAD infection between treatment arms was observed. The trial also confirmed that implantable devices were significantly less likely to become infected than external catheters (Hay & DiMichele, 2012). Finally, while the I-ITI study unexpectedly observed no significant impact of CVAD infection on either overall ITI success or the time taken to achieve ITI milestones in this ‘good risk cohort’, a similar evaluation in poor risk inhibitor patients is ongoing in the RESIST study (Gringeri et al, 2007a; Hay & DiMichele, 2012).

Finally, the ICP and ECP recommended that bypass therapy prophylaxis be administered at published doses early in the course of ITI when joint or life-threatening bleeding occurs (Grade C, Level IV evidence). The ICP further recommended close monitoring of inhibitor titres and FVIII recovery during ITI, as well as the prompt cessation of bypass therapy once any FVIII recovery is ascertained (Grade C, Level IV evidence). Both groups advocated for ongoing prospective data collection on the safety and efficacy of prophylaxis in the setting of ITI (Astermark et al, 2006a; DiMichele et al, 2007). Although few I-ITI subjects received bypass therapy prophylaxis, the ENJOIH prospective trial of recombinant activated FVII prophylaxis in children undergoing ITI is expected to yield important additional data (Santagostino et al, 2010).

These practice recommendations have been more recently reviewed by two groups and continue to inform the practice standards pending further analysis of the results of the International Immune Tolerance Study and other evidence based trials (Coppola et al, 2010; Benson et al, 2012).

Special considerations for ITI in severe haemophilia A patients with low titre inhibitors

Substantial data on the practice and outcome of ITI in low titre (LT) (historical peak < 5 BU; no anamnesis) haemophilia inhibitor patients are few. A survey of 21 European haemophilia centres revealed that while the majority (17/21) used ITI regimens of ≥100 u/kg per d for HT inhibitors, 15/21 routinely used <100 u/kg per d, and 10/21 used <50 u/kg per d regimens for LT patients. Outcome data was not reported (Astermark et al, 2006a). The ECP recommended that ITI be initiated in children and adults with persistently low titres for ≥6 months whose bleeding symptoms were not controlled with FVIII replacement (Grade C, Level IV evidence) (Astermark et al, 2006a).

Special considerations for ITI in moderate/mild haemophilia A

Occurring with an incidence of up to 0·84/1000 per year, inhibitors arising in moderate and mild haemophilia patients resemble acquired antibodies in their kinetics, bleeding manifestations and poor response to traditional ITI (Hay et al, 1998). The epidemiology of this problem, suspected genetic and treatment risk factors and current therapeutic options have been reviewed and studied (D'Oiron et al, 2008; Franchini et al, 2009; Kempton et al, 2010).

Little published ITI experience in moderate/mild haemophilia exists, probably due to the relative infrequency of these inhibitors and the misconception of a high ‘spontaneous inhibitor disappearance’ rate (Hay et al, 1998). In the only cohort study to date, Hay reported a comparatively low (25%) success rate with traditional high and low dose ITI initiated in eight poor risk adults because of excessive bleeding (Hay et al, 1998). Desmopressin has been used for both effective prophylaxis and ITI in cases where anti-FVIII antibodies do not cross-react or cross-reacting antibody titres wane (Santagostino et al, 1995; Hay et al, 1998). Traditional immunosuppressive therapy has been reported to be successful in this group (Robbins et al, 2001; Vlot et al, 2002). In a recently published retrospective review of the experience with rituximab therapy for inhibitors in congenital haemophilia, a significantly higher CR rate (12/16, 75%) was observed in moderate/mild patients when compared to those with severe disease (12/28, 43%, P = 0·02) (Franchini et al, 2008).

Based on these few data, the ICP and the UKHCDO both suggested that primary strategy should focus on inhibitor prevention with identification of high risk patients through gene mutation analysis and the preferential use of desmopressin to avoid or minimize particularly high dose FVIII exposure, as well as aggressive surveillance for inhibitor development (Level IV evidence). Both groups agreed on a potential therapeutic role for both traditional ITI and immune modulation in patients whose antibodies fail to disappear and/or who exhibit anamnesis upon FVIII rechallenge. However, the criteria for optimal candidate selection remain unclear (Level IV evidence) (DiMichele et al, 2007). This will hopefully be further elucidated by the ongoing European InSIGHT study. There remains an urgent need for an international registry to which each ITI course in this population would be reported.

Haemophilia B: immune tolerance for factor IX inhibitors

Immunology of factor IX inhibitor development and tolerance

There have been few additional published data on the immunology of FIX inhibitor development since this topic was last reviewed (DiMichele, 2007). In early haemophilia B mouse experiments, single dominant CD4 +  T cell epitopes in mice with both C57BL/6 (H2b) and BALB/C (H2d) backgrounds proliferated in response to subcutaneous injections with human FIX (Lin et al, 1997). However, auto-reactive CD4+ T cells were also noted in normal C57BL/6 mice in the absence of an endogenous immune response to murine FIX, calling into question the specificity of that immune response (Greenwood et al, 2003; Lollar, 2005).

Two human cytokine genes, IL10 (interleukin-10) and TNF (tumour necrosis factor), have recently been linked to FVIII inhibitor development (Astermark et al, 2006b,c). No such data currently exist for FIX inhibitor patients. However, genetic linkage studies in multiple recombinant inbred strains of haemophilia B mice suggested that multiple gene loci could be linked to the inhibitor response in these animals (Lozier et al, 2005). These experiments demonstrated that the major histocompatibility complex (MHC) class II (H-2) and/or K class I-a (Iak) loci were critical to this response (logarithmic odds (LOD) score ~4·8). However, other genes also contributed to FIX antibody development. Noted linkages included polymorphic markers from chromosomes 1 and 10 that approximated the IL10 and IFN1@ (interferon-α) immune regulatory genes (LOD scores ~2·3–2·6) (Lozier et al, 2005).

As is the case with the anti-FVIII antibody response, FIX neutralizing antibodies are thought to be polyclonal in nature. The earliest studies determined that the human anti-FIX antibodies was predominantly IgG4 in nature, based on 10 plasma samples from six haemophilia B inhibitor patients, including five with a history of an allergic phenotype (Sawamoto et al, 1996). Interestingly, transient IgG1 subclass antibodies were also detected in all three allergic phenotype patients whose plasma was procured at the exact time of allergic episode. However, IgG1 subclass antibodies could not be detected in plasma samples obtained more remotely (4 d to >4 weeks) from their allergic event in 2/3 of these patients, as well as an in two additional allergic phenotype inhibitor patients (Sawamoto et al, 1996). These data suggest that the allergic response that occurs in some FIX inhibitor patients may be associated with transient IgG1-subclass antibody production. The polyclonal nature of the FIX antibody response was subsequently confirmed by additional studies of haemophilia B inhibitor patient plasma (Christophe et al, 2001). Furthermore, the FIX epitopes recognized by the predominantly IgG1 and IgG4 subclass antibodies were noted to include the Υ –carboxyglutamic acid (GLA) and serine protease (SP) domains, but not the epidermal growth factor (EGF) and activation peptide (AP) domains (Christophe et al, 2001). Functionally, these antibodies inhibited the activated FIX/activated FVIII intrinsic FX activation complex through at least two mechanisms – interference with FIX binding to phospholipids as well as phospholipid-independent FIX binding to FVIII light chain (Christophe et al, 2001). More in vitro and in vivo studies are required to confirm these data and to further define the immunological and biochemical nature of the FIX inhibitory antibody response.

Similarly, there are fewer data on the immunology of tolerance to FIX when compared with FVIII. However, the relatively recent published experience with AAV hepatic gene transfer in mouse models of haemophilia B suggests that there is a similarly crucial role for CD4 +  CD25 +  FoxP3 +  T regulatory cells in FIX tolerance (Cao et al, 2009; Nayak et al, 2009).

Immune tolerance induction (ITI) for FIX inhibitors

Given the low incidence of FIX inhibitors, the historical experience with immune tolerance in haemophilia B was limited to two small series and a total of seven patients for whom the overall success rate with high dose FIX +/− immune modulation was 71% (Brackmann, 1984; Nilsson et al, 1986). Additional experience with the Malmö protocol suggested that ITI was successful in six of seven haemophilia B patients treated with the Malmö regimen; however tolerance was lost in one patient within 6 months, and a further attempt to re-induce tolerance was unsuccessful (Freiburghaus et al, 1999).

The largest collection of data on ITI in haemophilia B derives from the NAITR and the International Registry for Factor IX Inhibitors. In the NAITR, only 5/16 (31%) completed courses of ITI in haemophilia B were successful on a median dosing regimen of 100 u/kg per d (range 25–200) (DiMichele & Kroner, 2002). Daily dosing regimens and immune modulation were used in 88% and 47% of courses, respectively. Plasmapheresis was used in two courses. High purity or monoclonal FIX concentrates were used in 82% of the reported courses (DiMichele & Kroner, 2002). Given the paucity of these data, no association between ITI outcome and FIX dose or purity can be established at this time.

A demographic analysis of the NAITR FIX inhibitor cohort revealed that subjects with an associated allergic phenotype (10/16) were over-represented. ITI complications specific to this sub-group may have been responsible for the high failure rate in the cohort as a whole. Within the allergic subset of subjects, 8/10 failed ITI (DiMichele & Kroner, 2002). The rate of adverse events (65%) was 10-fold higher than that in haemophilia A inhibitor subjects, and was not dose-related. Allergic reactions accounted for 79% of the adverse events, and all reactions occurred in subjects with a previously identified allergic phenotype (DiMichele & Kroner, 2002). Similarly poor outcomes were reported from the International Registry for Factor IX Inhibitors (Chitlur et al, 2009). Among the 94 registry subjects, 60% had an allergic phenotype. Overall, only 5/39 (13%) of subjects who underwent ITI were tolerized (Chitlur et al, 2009).

Nephrotic syndrome is a complication of ITI in haemophilia B inhibitor patients who presented with an allergic phenotype (Ewenstein et al, 1997). Allergic complications of ITI occurred in NAITR subjects with a previously identified allergic phenotype, and were accompanied by the development of nephrotic syndrome in 3/10 subjects (DiMichele & Kroner, 2002). This complication was subsequently reported from the International Registry for Factor IX Inhibitors (Chitlur et al, 2009). Among the 13 cases compiled from the published international experience, 11 of which were associated with anaphylaxis, this complication presented 8–9 months into the course of high dose ITI (100–325 u/kg per d) with sepsis-like symptoms in conjunction with periorbital oedema, hypoalbuminaemia and proteinuria. FIX products of all types were implicated. Clinical improvement usually followed cessation of the FIX infusions, but the response to steroid therapy was historically poor. So far, the aetiology of this phenomenon remains unclear. Immunohistochemical staining of tissue obtained from a single renal biopsy failed to demonstrate any association with FIX immune complexes (Ewenstein et al, 1997).

Translation into clinical practice: recommendations for ITI in haemophilia B

At the current time, given the overall poor ITI success rate as well as the potential for the development of nephrotic syndrome during ITI, most clinicians either avoid the use of ITI or attempt it with extreme caution in the FIX inhibitor patients with an allergic phenotype. UKHCDO recommendations support this approach (Hay et al, 2006). The ICP additionally recommended early genotyping to assess risk for inhibitor development with/without allergic manifestations and due consideration of desensitization to improve the control of bleeding (Level IV evidence) (DiMichele et al, 2007). Although no specific ITI recommendations were possible, the ICP opined that a plan to proceed with ITI be accompanied by close surveillance for nephrotic syndrome using frequent urinalyses (DiMichele et al, 2007). A recent expert panel review of the literature largely concurred (Benson et al, 2012).

Alternative strategies to traditional immune tolerance

Given the poor success rate with ITI in haemophilia B, there has been anecdotal experience with the use of immune modulatory therapy with mixed success. Successful tolerance was induced in a single FIX inhibitor patient using cyclosporine (Cross & Van Der Berg, 2007). Successful or partially successful tolerance of haemophilia B inhibitor patients with an allergic phenotype, uncomplicated by nephrotic syndrome, was reported with a regimen of mycophenolate-mofetil, dexamethasone, intravenous immunoglobulin and high-dose FIX (Wermes et al, 2000; Klarmann et al, 2008). The experience with rituximab has been mixed. Long-term inhibitor eradication using rituximab was unsuccessful in two haemophilia B inhibitor patients (Mathias et al, 2004; Fox et al, 2006); partially successful in one (Chuansumrit et al, 2008); and completely successful in a single individual with long term remission for a period of 12 months without the development of nephrotic syndrome (Barnes et al, 2010).

Ultimately, FIX inhibitor prevention may prove to be the best strategy. The immunogenicity of future recombinant FIX proteins must be carefully assessed in pre-clinical and pre-licensure human clinical trials (Pipe, 2005). The prospect of a role for gene transfer in achieving permanent tolerance to FIX is intriguing. Immunological tolerance with hepatic gene transfer has been achieved in a haemophilia B mouse model (Mingozzi et al, 2003; Dobrzynski et al, 2006). However, there were qualifying aspects to the success of this preliminary work. These included the need for high expression (>30 ng/ml) of the FIX transgene product; F9 genotype specificity with tolerance more difficult to achieve in complete gene deletion mice; and the requirement for T cell naiveté to FIX in the transgenic animals, an immunological state that could be difficult to replicate in humans. Further experience suggests that the mechanism of tolerance induction in this animal model may be dependent on the mode of gene transfer. While vector-mediated induction of sufficient transgene product-specific regulatory T cells may be critical to tolerance in hepatic gene transfer (Cooper et al, 2009), high FIX expression levels may be crucial to tolerance when using intramuscular gene transfer in the same mouse haemophilia B model (Kelly et al, 2009).


Immune tolerance in haemophilia A and B: where to from here?

Prospective clinical trials (FVIII) and observational studies (FVIII and FIX), that further define their relative safety and efficacy of in risk- stratified subjects will undoubtedly shape the traditional immune tolerance strategies of the future. Progress in establishing an adequate evidence base for clinical practice will necessitate both international collaboration and innovative approaches to trial design.

In prioritizing future research, the investigative community might consider that although present strategies certainly require further refinement and optimization, ITI, as currently practiced, successfully induces tolerance in the majority of FVIII inhibitor patients. However, all success (defined by normal pharmacokinetics), as well as failure in ‘poor risk’ populations is accompanied by an up -front substantial price tag. This may be therapeutically limiting at a time when the expenditure of health care dollars is being scrutinized around the world. Including the absence of musculoskeletal morbidity in the ‘success’ definition for younger patients, or extending proven ITI strategies to larger numbers of ‘ageing’ inhibitor patients in need of expanded interventional health care, could potentially further increase the cost of therapy, making it less accessible globally. The storey for FIX inhibitor eradication is quite different – the traditional strategies used in at-risk allergic phenotype patients are largely ineffective and associated with an unacceptable risk of morbidity.

For haemophilia A and B, the starting points and investigative methods will necessarily be different; however, a common strategy may still apply. Inhibitor prevention must be considered as the investigative priority. Prophylaxis as a mechanism of tolerance will require further study both mechanistically and clinically. Ultimately, our comprehensive understanding of the immunology of inhibitor prevention and eradication may best inform the design of safe and effective immune modulatory therapeutics with the greatest potential to establish durable primary or secondary tolerance to exogenous factor replacement, and eventually, to sustained endogenous FVIII or IX production following successful gene transfer. All told, the future of immune tolerance therapy has never been more threatened, or more promising.