Inhibitors – genetic and environmental factors
It is known that a large number of both genetic and environmental factors contribute to the risk of inhibitor development, but underlying pathogenetic mechanisms are still under investigation. The clinical research on inhibitors towards factor VIII (FVIII) is challenged by the fact that this is an infrequent event occurring in a rare disease. Therefore, it is widely accepted that complementary studies involving animal models can provide important insights into the pathogenesis and treatment of this complication. In this respect, mouse models have been studied for clues to FVIII immunogenicity, natural history of immunity and for different approaches to primary and secondary tolerance induction. In the clinical setting, the type of FVIII product used and the occurrence of product switching are considered important factors which may have an influence on inhibitor development. The evaluation of data currently available in the literature does not prove unequivocally that a difference in the immunogenicity exists between particular FVIII products (e.g. recombinant vs. plasma-derived, full length vs. B-domainless). In addition, national products switches have occurred and, in this context, switching was not associated with an enhanced inhibitor risk. In contrast with severe haemophilia A, patients with moderate and mild haemophilia A receive FVIII treatment infrequently for bleeds or surgery. In this condition the inhibitor risk is low but remains present lifelong, requiring continuous vigilance, particularly after intensive FVIII exposure.
The development of an inhibitory antibody response to factor VIII (FVIII) replacement therapy represents the most serious therapy-related complication of this condition. FVIII inhibitors occur in 25–30% of severe haemophilia A patients with a predominant onset between 10 and 20 exposure days (EDs) to treatment . The inhibitor risk is much lower (approximately 2–3 per 1000 patient-years) in severe patients who have already received hundreds of FVIII infusions [1, 2]. Low inhibitor prevalences (3–13%) are reported in patients with moderate and mild haemophilia A (MHA) [3-6] who, in contrast with severe patients, are less frequently treated and develop inhibitors at an older age, usually after intensive FVIII exposure due to injury or surgery [7-12].
Knowledge of the pathogenetic mechanisms underlying inhibitor development and assessment of therapeutic strategies to mitigate and treat FVIII inhibitors are challenged by the fact that this is an infrequent event occurring in a rare disease. Thus, the conduct of clinical research to elucidate new information concerning inhibitors requires multicentre international studies that often take an extended period of time to organize and report results [13-15]. While, as always, results obtained in animals must be viewed with caution, the conservation of the immune system in vertebrates suggests that lessons from non-human models will often yield knowledge that is highly pertinent to the human condition. .
Factor VIII Inhibitors: Lessons from animal models – David Lillicrap
Animal Models of FVIII Immunogenicity
In the vast majority of cases, the animal model that has been used to evaluate FVIII immune reactivity has been the haemophilic mouse. While haemophilia A dogs can develop inhibitors to their canine FVIII replacement therapy, the number of haemophilic dogs available to perform statistically robust studies is extremely limiting. Interestingly, in the Queen's University haemophilic dog colony , where inhibitor prone dogs have been documented for the past 30 years , a clear genetic predisposition is evident. Nevertheless, while dog studies of FVIII immunity are infrequent, the dog model has been used recently to highlight the potential of FVIII gene transfer for inducing tolerance to FVIII . Finally, it should be noted that all haemophilia A dogs will develop a potent anti-FVIII immune response if infused with human FVIII concentrate and thus any long-term study of FVIII immunity in dogs should use the canine protein or transgene [20, 21].
Mouse models of FVIII immunity
There are now several different mouse models of haemophilia A that have been used to investigate inhibitor development and treatment. The original FVIII knockout mice  have been extensively studied and have been repeatedly been shown to develop a strong immune response to human FVIII infusions. The timing and magnitude of this reaction varies with the FVIII infusion protocol but evidence of anti-FVIII IgM and IgA antibodies develops after a few days and in most animals a potent anti-FVIII IgG response is present after three exposures . There is evidence that the background strain of the mice influences the magnitude of the response, with C57BL/6 mice developing higher titre inhibitors .
As the incidence of the human anti-FVIII antibody response in the original FVIII knockout mice is >95%, recent efforts have been focused on the development of additional mouse models in which the incidence of inhibitors more closely approximates that seen in humans (i.e. ~30%). These efforts have resulted in the generation of at least three alternative mouse models to study FVIII immunogenicity with, in each instance, the application of a different strategy to reduce reactivity to human FVIII exposure.
In the first of these models, the approach that has been taken is to delete the entire mouse MHC II locus and to introduce a single human MHC class II allele (DRB1-1501) into the existing haemophilia A mouse model . This class II allele is associated with an increased likelihood of inhibitor development in humans. Studies of these mice have shown an incidence of anti-human FVIII antibodies of 30–80%, with at least some of this variance likely being the result of the mixed genetic background of the mice. Factor VIII antibody generation in these animals was markedly enhanced by the administration of FVIII by the subcutaneous route and when FVIII was co-administered with lipopolysaccharide. Finally, evidence has been gathered to show that presentation of eight different FVIII-derived peptide regions in this humanized model system results in CD4+ T-cell reactivity. Of note, most of these eight peptide regions contain promiscuous epitopes that can bind several different HLA-DR proteins.
In the second humanized haemophilic mouse model, a human FVIII cDNA transgene, regulated by the liver-specific albumin promoter, has been microinjected into fertilized oocytes, and founder mice were crossed with exon 17 knockout haemophilia A mice . Despite the fact that FVIII mRNA can be found in several tissues in these mice (including liver, brain and gonads), they do not express FVIII in their plasma. Nevertheless, when challenged repeatedly with intravenous human FVIII they do not develop anti-human FVIII antibodies. Only when exposed to FVIII whose immunogenicity has been purposely enhanced is tolerance broken.
The third humanized mouse model that has been generated again involves the insertion of a human FVIII transgene. However, in this instance, a mutant cDNA encoding an Arg593Cys missense change has been utilized . This variant is found as a recurring mutation in humans with mild haemophilia A that are more prone to inhibitor development. Here again there is an absence of circulating FVIII and yet the mice are consistently tolerant to human FVIII unless it is delivered in a manner recognized to be associated with enhanced immunogenicity (e.g. delivered subcutaneously with an adjuvant).
Utility of animal models of FVIII immunogenicity
To date, the mouse models described above have been utilized for a variety of purposes. They have been studied for clues to FVIII immunogenicity [27, 28], for the natural history and details of FVIII immunity  and for the evaluation of many different approaches to primary and secondary tolerance induction [30-34]. With the recent arrival of the various humanized haemophilia mouse models we can expect to see additional studies in which outcomes more pertinent to the human context will be forthcoming.
Influence of product type/switching – Elena Santagostino
Product type as risk factor of inhibitor development
It is well known that the inhibitor risk in previously untreated patients (PUPs) is determined by multiple interactions between genetic and environmental factors. Among the latter, treatment-related determinants including intensity of replacement treatment (FVIII dose and frequency of administration), treatment regimen (i.e. prophylaxis vs. on-demand) and FVIII product type have been reported to influence variably the inhibitor formation [13, 14, 35-38].
A systematic review first highlighted that inhibitor incidence was lower in patients treated with one plasma-derived FVIII (pd-FVIII) brand vs. those who had used multiple brands of pd-FVIII concentrates or a single recombinant FVIII (rFVIII) product . However, the studies included in this analysis were very heterogeneous, particularly in terms of study design and frequency of inhibitor testing. Furthermore, relevant risk factors, such as severity of FVIII defect, F8 genotype, family history of inhibitors and treatment regimen were not taken into account. A more recent systematic review showed that the incidence of inhibitors was nearly twofold higher in patients treated with rFVIII than in those treated with pd-FVIII, nevertheless, the effect of the type of FVIII product on inhibitor incidence was no longer statistically significant after anova because study design and period, inhibitor testing frequency, and duration of follow-up were identified as critical determinants of the differences in inhibitor incidence rather than the type of FVIII product .
Overall, the bulk of data currently available cannot be considered as definite evidence of a different immunogenicity between rFVIII and pd-FVIII. This condition of uncertainty justifies the design of the independent randomized controlled Study on Inhibitors in Plasma-Product Exposed Toddlers (SIPPET; http://www.clinicaltrials.gov/, #NCT 01064284; EUDRACT, #2009-011186-88), currently ongoing and aimed at demonstrating a 50% lower incidence of inhibitors for pd-FVIII . Final results will be analysed cumulatively to compare inhibitor incidence in PUPs treated with the two classes of FVIII products (pd-FVIII and rFVIII).
Although the risk of inhibitor development does not disappear throughout the lifetime, previously treated patients (PTPs) with severe haemophilia and multiple FVIII exposures have a much smaller risk of developing an inhibitor than PUPs. Data of 1257 PTPs from the United States and the United Kingdom confirm a low rate of de novo inhibitors (2.14–3.8 cases for 1000 person-years) [2, 41], although, there is recent evidence of a slight increase in the elderly . Possible reasons for this observation may be related to a delayed inhibitor detection or relapse, intensive replacement treatment for surgical procedures and the decline of natural immune tolerance with ageing. Another factor that may influence inhibitor formation in PTPs is FVIII product switching. Indeed, it is very rare for adults to have used the same product all of their lives , therefore, changing FVIII product type seems to be part of the natural history of haemophilia treatment.
Product switching as risk factor of inhibitor development
Switches from pd-FVIII to rFVIII products and between different rFVIII are quite common in real-world practice. Nevertheless, physicians and patients are often reluctant to change the product in use, mainly because of safety concerns and, especially, of the risk of inhibitor formation.
These concerns first arose in the 1990s when inhibitor outbreaks occurred in PTPs in Belgium and the Netherlands following the introduction of pd-FVIII concentrates that had undergone novel viral inactivation procedures [43, 44]. It is assumed that these inactivation methods resulted in FVIII structural changes that triggered the inhibitor formation.
Reassuring results of a low rate of de novo inhibitors in PTPs who switched from pd-FVIII to rFVIII were shown in prospective premarketing studies carried out with these new products [45-50].
Subsequently, national product switches have provided important pieces of evidence. Two surveillance studies were carried out in Canada during the population switch from pd-FVIII to rFVIII and then from first to second generation rFVIII and neither of these studies showed an increase in inhibitor incidence [51, 52]. A retrospective study performed in Ireland after a national tender with consequent en masse switch to a third generation full-length rFVIII did not detect changes in the rate of de novo inhibitor formation .
In the UK a national tender was floated in 2009–2010 and it required half of patients using rFVIII to change rFVIII brands . Inhibitor testing was performed in all patients prior to the switching date and 6-monthly thereafter. Overall 1217 patients with severe haemophilia A and no inhibitor history were analysed (535 switched and 682 did not). Almost all patients who switched changed to B-domainless rFVIII. The inhibitor incidence was not significantly different from that observed during the previous two decades .
All these studies indicate that switching is not associated with an increased risk of de novo inhibitor formation. However, due to the very low inhibitor incidence in PTPs, all studies were underpowered.
Meta-analyses of PTPs studies were also performed to gain further insight into the available evidence. This methodology was applied to compare the inhibitor risk in PTPs receiving full-length rFVIII with that of patients given B-domainless rFVIII . Unexpectedly, a sevenfold to 10-fold higher inhibitor incidence was found in recipients of B-domainless FVIII . These results were not confirmed in a subsequent systematic review and meta-analysis adopting strict criteria for study selection .
In conclusion, prospective, controlled surveillance programmes on switching and not switching patients are still required to provide robust evidence concerning the inhibitor risk related to product switching. In this respect, inhibitor testing before and after the switch as well as testing of not switching patients is a crucial element to establish the correlation with the new treatment.
Special considerations for product switching
The availability over time of newer therapeutic molecules and the variable market accessibility of different products often entail switching; in this light, patient information on evidences concerning potential risks and benefits associated with product switching is mandatory and should be part of our routine practice. Furthermore, physicians should discuss with patients and their caregivers the different therapeutic approaches and the available product options before the possible need for considering product switch.
In some clinical conditions, that are known to be at high risk of inhibitor formation, specific considerations should be kept in mind.
Since inhibitors usually occur within the first 50 EDs, many physicians prefer not to switch FVIII products during this time. On the other hand, it should be acknowledged that switching between different FVIII products was not associated with an enhanced risk of inhibitor development in the large cohorts of patients with <50 EDs evaluated in the CANAL and RODIN studies [13, 14].
Similarly, intensive treatment and/or surgery are well known determinants of inhibitor development [36, 37]. Therefore, it sounds reasonable to avoid switching FVIII product in the peri-operative period, although, no robust evidence is available to support a specific role of product switch in the inhibitor formation after surgery and intensive treatment [36, 37].
Patients with a family history of inhibitors or a severe F8 gene defect are also known to be at high inhibitor risk, however, no observation has been reported in the literature indicating an increased inhibitor risk when switching FVIII products in these conditions.
Patients with a previous history of inhibitors, including those who achieved success after immune tolerance induction (ITI), are usually considered at risk of inhibitor recurrence. It may be reasonable to surmise that, in this situation, the introduction of a new FVIII product could break the tolerance to FVIII. On the other hand, successful ITI outcome was reported with FVIII products different from that eliciting the inhibitor formation .
Inhibitor development in mild haemophilia A – Karin Fijnvandraat
A substantial lifelong risk of inhibitor development
In patients with MHA inhibitor prevalences between 3% and 13% are reported, but these cross-sectional studies did not take exposure to FVIII concentrate into account [3-6]. As patients with MHA receive factor VIII replacement therapy infrequently for bleeds or surgery, it is especially important to express the inhibitor risk as a function of EDs. This was done in the INSIGHT study that included 2711 persons with MHA (FVIII 0.02–0.40 IU mL−1) from Europe and Australia . The inhibitor risk at 50 EDs was 6.7% (95% CI, 4.5–8.9) and at 100 EDs the risk further increased to 13.3% (95% CI, 9.6–17.0). Furthermore, this study demonstrated that the risk of inhibitor development in patients with MHA remains present after 50 EDs and even after 100 EDs. Thus, in contrast to severe haemophilia A the risk of inhibitor development does not level off after 50EDs and patients with MHA are at lifelong risk of inhibitors, requiring continuous vigilance. This necessitates frequent testing for inhibitors, especially after intensive FVIII replacement for major bleedings or surgery.
Impact of inhibitors in moderate and mild haemophilia
Inhibitors in patients with MHA develop at a median age of 37 years after a median of 29 EDs [7, 8] In about half the patients (57%), the baseline FVIII drops below 0.01 IU mL−1 as the inhibitor cross-reacts with the endogenous FVIII, changing the phenotype into severe haemophilia A . The spontaneous bleeding pattern that ensues frequently resembles the bleeding pattern observed in acquired haemophilia with extensive mucocutaneous bleeding. Severe and uncontrollable bleeding contributes to an increased morbidity and mortality among patients with MHA and inhibitors.
Clinical risk factors of inhibitor development
Inhibitors are frequently provoked by intensive treatment with therapeutic FVIII concentrates for surgery or trauma [8-11]. A cohort study of 41 patients with MHA that received perioperative FVIII replacement reported a 186-fold (95% CI 25–1403) increased risk of inhibitor development for surgery as the reason for first intensive exposure . This extremely high risk arose by the extreme contrasts in the analysis: the time period of 3 months post surgery was compared to all other periods of 3 months during the study. As patients with MHA need therapeutic FVIII concentrates infrequently and months may pass without any exposure to FVIII concentrate, this comparison overestimates the risk that is inflated tremendously. Time post surgery was compared to time periods without any exposure to FVIII concentrates at all! This teaches us that the analysis of clinical risk factors in MHA inhibitor development requires a thoughtful methodological approach. Efforts should be made to compare patient groups that have similar baseline likelihood to develop inhibitors and only differ in the single factor that is under investigation (e.g. FVIII treatment for surgery vs. FVIII treatment for other reasons). Especially the number of previous EDs in both groups should be as similar as possible.
The inhibitor risk of continuous infusion has been the subject of intense debate, as inhibitors were frequently observed following intensive treatment administered by continuous infusion [10, 11]. Other studies could not confirm this association [9, 12]. A large cohort study analysing 1079 continuous infusions in 742 patients with haemophilia A (severe, moderate or mild) established that the absolute inhibitor risk was limited as only nine patients (1.2%) developed an inhibitor .
Genetic risk factors of inhibitor development
There are over 500 reported causative missense mutations for MHA reported in the Haemophilia A database (http://hadb.org.uk/). In patients with missense mutations the presence of circulating endogenous FVIII protein maintains a state of immunological tolerance towards FVIII. Nevertheless, there are certain missense mutations that predispose to inhibitor development in MHA [7, 59, 60] that are clustered in the A2 domain and the C1–C2 domains, e.g. Arg593Cys, Asn618Ser, Asp274Gly, Arg2150His, Arg2159Cys, Trp2229Cys. These missense mutations may contribute to T-cell epitopes that can bind to common HLA-II types. Furthermore, it appears that a class switch in the amino acid substitution (from small/hydrophobic, neutral, acidic or basic to any other of these classes) increases the inhibitor risk, as was recently established in a study of 720 patients with haemophilia and missense mutations .
Treatment of bleeding and inhibitor eradication
As the consequences of inhibitors in MHA are profound, it is extremely important to make strong efforts to prevent this complication. Genotyping can identify which patients with MHA are at increased risk of inhibitor development due to a high risk missense mutation. In these high-risk patients, alternative treatment options such as desmopressin (DDAVP) or bypassing agents should be considered whenever they need surgery or intensive treatment for a large bleed.
Once an inhibitor has emerged, bleeds may be treated by DDAVP when measurable FVIII levels are present. If the response to DDAVP is not adequate or when prolonged hemostatic activity is necessary, FVIII bypassing agents are required, such as recombinant FVIIa (Novoseven®) or activated prothrombin complex (FEIBA®) .
In patients with MHA and inhibitors, avoidance of FVIII concentrate is usually associated with a decline in inhibitor titer. However, re-exposure to FVIII concentrates will often elicit an anamnestic response [8, 9]. Therefore, unless uneventful rechallenge with therapeutic FVIII concentrates has been documented, these patients should not be regarded as “complete remissions” or “successful tolerization”.
In patients with reduced endogenous FVIII levels and a severe bleeding phenotype, it is important to achieve baseline FVIII levels again to mitigate the bleeding phenotype. ITI or immune modulation may be initiated to eradicate the antibodies . The effect of these therapeutic approaches in patients with MHA has not been studied formally. Small ITI case series suggest that traditional ITI regimens used in severe haemophilia A are less efficacious in MHA, with a <30% ‘success’ rate [8, 9, 63]. Case series of Rituximab use have been reported [64-66] and a literature review  suggests a higher than expected success rates (12/16, 75%). This figure should be treated with caution as there may be reporting bias of successful cases and it remains unclear whether cases were re-challenged with FVIII to establish true tolerance, however, it does illustrate an important difference between the eradication of inhibitors in MHA and severe haemophilia A.
ES received speaker fees by Baxter, Biotest, Kedrion and Octapharma, acted as a payed consultant for Bayer, CSL Behring and Grifols and received unrestricted research grants from Novo Nordisk and Pfizer. KF has received unrestricted research funds from CSL Behring, Pfizer, Novo Nordisk and Bayer, and has given lectures at educational symposiums organised by Pfizer, Baxter and Bayer. DL has received research funds from Bayer, Baxter, Biogen-Idec and CSL Behring.