Understanding inhibitor development in haemophilia A: towards clinical prediction and prevention strategies


Antonio Coppola, MD, Regional Reference Centre for Coagulation Disorders, Federico II University, Via S. Pansini, 5 -80131 Naples, Italy.
Tel.: +39 08 174 62317; fax: +39 08 154 66152;
e-mail: antocopp@unina.it


Summary.  Inhibitor development, because of its impact on patients’ morbidity and quality of life, is presently the most serious complication of haemophilia A treatment. The identification of several genetic and non-genetic risk factors may be used for the stratification of inhibitor risk and the definition of prevention strategies, particularly for patients with a high-risk genetic profile. The most extensively studied genetic factor is the type of F8 mutation, i.e. large deletions, nonsense mutations and inversions, which are associated with a higher risk of inhibitor development. This is the basis for the increased risk in patients with inhibitor family history; however, concordance family studies showed that factors other than F8 mutations are involved. An emerging role is investigated for polymorphisms of immune-regulatory genes that may increase (IL-10 and TNF-α) or reduce (CTLA-4) inhibitor risk and whose heterogeneous ethnic distribution may correlate to the higher inhibitor risk in non-caucasian patients. A role for FVIII haplotypes, particularly in black haemophiliacs, has been recently proposed. Recent studies report an increased inhibitor risk for initial intensive treatments (surgery or severe bleeds requiring high-dose and/or prolonged treatment, presence of danger signals), whereas regular prophylaxis (absence of danger signals) exerts a protective effect. A clinical score including the type of F8 mutation, family history of inhibitors and intensive treatment has been recently validated for predicting inhibitor risk. Because of the lack of useful data regarding the role of different types of FVIII concentrates, the stratification of risk in patients starting replacement treatment together with the careful evaluation of indications, doses and duration of treatment at first exposures and further efforts for overcoming barriers to early implementation of prophylaxis are encouraged, particularly for patients with a predictable high inhibitor risk.


Approximately 30% of severe haemophilic patients generates antibodies (inhibitors) against therapeutically administered factor VIII (FVIII), typically during the first 20 exposure days (ED) [1]. Inhibitor development remains the most serious and challenging complication of modern treatment of haemophilia A in developed countries [2], where safe FVIII concentrates are largely available and where prophylaxis is increasingly used to prevent arthropathy. Despite improvements in strategies for the treatment of bleeds with by-passing agents and the recent implementation of prophylactic regimens in an inhibitor setting [3,4], the safe and effective standard of care is precluded in patients with inhibitors, who experience higher levels of morbidity and mortality [5] and a poorer quality of life related to their orthopaedic status than non-inhibitor patients [6,7]. Moreover, the economic burden of such a complication is the highest reported for a chronic disease [6].

Over the past two decades, significant advances in genetics and molecular immunology and multinational efforts in conducting clinical studies enabled us to understand that inhibitors in haemophilia are not simply generated as a result of the immune response which recognizes the transfused FVIII as a foreign protein. There is an interplay of many genetic and non-genetic factors when replacement FVIII infusions are first given, and this may affect the interaction of such an exogenous protein with the patient’s immune system [8,9]. In the light of growing knowledge of these mechanisms and risk factors, inhibitor development is no longer considered a completely unpredictable event and therefore tools to aid in risk stratification, which are useful in clinical practice, have been recently proposed [10]. In keeping with this knowledge, epidemiological data of inhibitor formation may be revisited [1,11] and the identification of non-genetic, potentially modifiable, risk factors may provide a key for defining prevention strategies, particularly for patients having a high-risk genetic profile.

Risk factors for inhibitor development: the genetic background

The first and most extensively studied genetic factor is the causative FVIII gene (F8) mutation. A series of studies showed that the development of inhibitors correlates with the type and location of F8 mutations [9,12–16]. There is general agreement that patients carrying mutations, which cause severe rearrangements of F8 and preclude the synthesis of the gene product, defined as null mutations (large deletions, inversions and nonsense mutations) are more susceptible to developing inhibitors to FVIII. On the other hand, missense mutations, associated with the synthesis of an endogenous but functionally abnormal protein, usually confer a low risk of inhibitor development. Small insertions/deletions and splice site mutations are also considered lower risk genotypes, but this risk is reported more variable with respect to the location of the gene defect and its effects on the gene product. In patients with small deletions/insertions, the risk of inhibitor development is lower for mutations that occur within the A-runs compared with non-A-run abnormalities [13,15]; inhibitors were found from 17% to 44% of patients carrying splice site mutations [13–16]. Therefore, a more detailed stratification of mutation subclasses according to inhibitor risk has recently been proposed [16], but globally most patients carry mutations with a similar risk profile [9,13]. The Malmö International Brother Study (MIBS) clearly showed that for siblings, a family history of inhibitor development is associated with an approximately threefold higher risk to develop an inhibitor [17]. However, this genetic susceptibility could not be predicted by the F8 mutation itself, as the overall concordance between siblings (all or none with a history of inhibitors) was 70% and 63% in all families studied and in those with the intron 22 inversion respectively [17,18]. For the latter, only 38% of all siblings had inhibitors [18]. On the other hand, concordance in families with inhibitors was 42%, and 72% of these inhibitors had the same anamnestic (high-responding) features [18]. Interestingly, concordance was not absolute even in monozygotic twins. The role of genetic determinants other than F8 mutations is also supported by the twofold increase in the risk of inhibitor development in non-caucasian patients [17], whose mutation spectrum is similar to that of causasians. In this respect, the exclusive presence of H3 or H4 FVIII haplotypes in black haemophiliacs, distinct from the H1 and H2 found in all racial groups that match the replacement FVIII products therapeutically used, have been recently proposed as a risk factor for this ethnic group [19]. The search for other determinants of genetic susceptibility to inhibitor formation has been obviously extended to genes involved in the immune response. In spite of a key role in the recognition and presentation of FVIII for initiating the cellular response that results in inhibitor generation [8], conflicting results have been reported regarding a predisposition or a protective role of a variety of leukocyte antigen (HLA) alleles in this setting [9]. Interesting data in the MIBS have been obtained by evaluating polymorphisms of genes encoding immune-regulatory cytokines. An increased inhibitor risk has been shown in patients with a microsatellite polymorphism in the promoter of the interleukin-10 (IL-10) gene compared with non-carriers [20], as well in patients with the homozygous −308A allele of tumour necrosis factor-α (TNF-α) gene compared with those bearing a G allele [21]. On the other hand, a protective effect has been detected in patients with the −318C>T polymorphism of the cytotoxic T-lymphocyte associated protein-4 (CTLA-4) gene [22]. These polymorphisms putatively modulate the cytokine synthesis/release upon antigenic stimuli, thus promoting or inhibiting the expansion of possible inhibitor-producing B-cell clones. The potential role of these polymorphic markers is further supported by their distinct ethnic distribution [9]. On the whole, these and possibly other currently unrecognized immune-regulatory polymorphic markers may contribute to the genetic susceptibility to inhibitor development. From a clinical point of view, the stratification of genetic risk in newly diagnosed patients starting replacement treatment has been collectively referred to F8 mutation type, although the classification into high-risk and low-risk mutations is likely an oversimplification, and to the inhibitor family history, which may include other candidate or hypothesized genetic determinants [10].

Environmental and treatment-related risk factors

Over the past several years, increasing interest has developed in identifying non-genetic (potentially modifiable) factors that predisposes the patient to inhibitor development. The role of these non-genetic factors has been increasingly supported by evolving concepts of the immune response based on the ‘danger model’ [23]: the immune system is activated by alarm signals arising from the injured tissues to a greater extent than by the recognition of non-self. In this respect, the presentation of the exogenous FVIII may not be sufficient for initiating an immune response. In the presence of danger conditions (i.e. severe bleeds, trauma or surgery with major tissue injury), the foreign protein is intensively presented (high-dose and/or prolonged treatment) in association with signals that up-regulate the cellular T and B lymphocyte response. On the other hand, regular exposure to lower doses of antigen, in the absence of danger signals, which occurs in regular prophylaxis, may induce the tolerization of the foreign protein. This hypothesis is supported by recent studies showing a key role of the intensity of treatment at the first FVIII exposures: both in the CANAL cohort, which investigated 366 consecutive previously untreated children (PUPS) born between 1990 and 2000 from 14 centres in Europe and Canada [24], and in the combined analysis of data on 236 patients (FVIII <2%) from the four recombinant FVIII (rFVIII) registration PUPS studies [25], surgical and prolonged (≥5 days) FVIII exposure at first treatment, and high FVIII dose (≥50 IU/kg) during the first 50 ED were associated with a twofold to threefold increase in the risk of inhibitor development. These data are also likely to explain the higher risk of inhibitor development in patients with an early age at first FVIII exposure, detected in previous studies and only at univariate analysis in the CANAL and other studies (Table 1). For this reason, an intensive treatment at initial exposure has been considered the most significant determinant (three points) in the prognostic score determined from the CANAL data [10]. This score includes the genetic factors previously mentioned (high-risk F8 mutation and positive inhibitor family history, each two points) and may help to identify patients at high (>50%, score ≥3), intermediate (about 25%, score 2) or low (about 6%, score 0) risk of inhibitors. In agreement with the ‘danger model’, the CANAL study [24] and a previous case–control Italian study [26] showed by regression models, which take into account other potential inhibitor risk factors, a 60–70% reduction in inhibitor risk in patients on regular prophylaxis compared with those receiving on-demand treatment (Table 1).

Table 1.   Studies reporting multivariate analysis of putative inhibitor risk factors (OR/RR and 95% CI).
Author [reference], (patients, n)End pointF8 mutEthnicityInhibitor family history*Age at 1st exposureType of product§ProphylaxisComments
  1. Inh, inhibitors; n.e., not evaluable; mut, mutations: high-risk mutations vs. non-high-risk.

  2. *Positive vs. negative.

  3. Non-caucasians vs. caucasians.

  4. 12 months or older vs. <6 months.

  5. §rFVIII vs. pdFVIII.

  6. Inhibitor titre >5 BU/mL.

  7. **<1 months vs. >18 months.

  8. ††<6 months vs. >12 months.

  9. ‡‡Two consecutive positive inhibitor detections with FVIII recovery <66%.

  10. §§Intensity of treatment also included in the analyses.

  11. Statistically significant results are indicated in bold values.

Santagostino et al. [26], (108)All inh2.2
0.9–12.0 (<11 months vs. >16 months)
Patients (FVIII:C <2%) treated only with recombinant products. Inhibitor testing every 3 months over the first 100 ED, then every 6 months (at least 150 ED). Subgroup analysis for prophylaxis. Breast feeding included in multivariate analysis
Goudemand et al. [28], (148)All inh2.5
n.e.Patients (FVIII<1%) treated with a single plasma derived (pd) or a full-length recombinant FVIII brand. Longer follow up and FVIII exposure for patients on pdFVIII. Lower but non-statistically different prevalence of non-white patients and frequency of inhibitor testing in the recombinant cohort
High inh + ITI2.2
Chalmers et al. [29], (348)All inh3.3
0.5–2.9 (<1 month)
1.5 0.5–4.1 (<18 months)
n.e.Pateints (FVIII<1%) treated with either a (high-or intermediate purity) pd or a recombinant product. Inhibitor testing at least every 3–6 months, at least 50 ED
Gouw et al. [27], (366)§§Clinically relevant‡‡2.3
Patients (FVIII:C <2%) treated with either a pd (23 brands) or a recombinant (4 brands) product. Frequency of inhibitor testing not clearly stated, at least 50 ED. Intensity of treatment-variables included in the analysis
High titre3.7
Gouw et al. [25], (236)§§All inhn.e.2.7
n.e.n.e.Patients (FVIII:C <2%) from the PUPS rFVIII registration studies with available information up to 50 ED. Intensity of treatment-variables included in the analysis
High inh 3.7

The most controversial issue of treatment-related risk factors remains the type of FVIII concentrate administered [1]. The CANAL study also investigated the risk of inhibitor development with respect to plasma-derived vs. recombinant products, and switching between FVIII products [27]. No significant difference in the risk of inhibitor development was detected between plasma-derived and recombinant FVIII products, even when von Willebrand factor (VWF) content of the former was taken into account. Moreover, the switching between FVIII product brands did not increase the inhibitor risk over the first 50 ED [27]. At variance with the comparison of the retrospective French cohorts that showed lower inhibitor risk in patients treated with a single FVIII plasma-derived vs. a recombinant product [28], the CANAL results were consistent with the findings of another recent English study that failed to detect significant differences in inhibitor risk at multivariate analysis [29], and highlighted that clinically relevant inhibitors develop with substantially comparable figures irrespective of type of product. On the other hand, the prospective long-term rFVIII registration studies clearly showed that approximately one-third of detected inhibitors was transient and that only about half were high-titre (>10 BU/mL) inhibitors (Table 2) [11]. These discrepancies in inhibitor detection may be attributed to the different study designs (retrospective, multicenter and multinational involving many product brands and modality of treatment, different ED) and, particularly to the fact that low-titre and transient inhibitors were probably not detected in the older plasma-derived FVIII studies, resulting in an under-estimation of the overall incidence of inhibitors. Thus, the protective role of VWF in reducing FVIII immunogenicity is supported more by the in vitro data and preclinical experiments in animal models rather than in the clinical setting (Table 1), where there is no conclusive evidence to support the lower risk of inhibitor development of VWF-containing plasma-derived products; indeed, the possible advantages seem confined to some but not to all products and to the development of low-titre inhibitors [27–29]. Moreover, epidemiological data on inhibitor development in the rFVIII PUPs studies [11] may be revisited based on the present knowledge that makes it possible to identify risk profiles for the study populations (Table 2). The most convincing predisposing or protective effects of factors affecting inhibitor development briefly discussed in these paragraphs are represented and summarized in Fig. 1.

Table 2.   Inhibitor prevalence and type and distribution of factors affecting the inhibitor risk profile in the rFVIII PUPS registration studies.
Study [reference], productInhibitors, n Prevalence, % in severe pts (<2%)High-titre (>10 BU/mL), n Prevalence, %Transient, n (LR/HR) % of inhibitorsAfro-American patientsHigh-risk mutationsFamily history of inhibitorsProphylaxis
  1. *Data not available for all patients.

  2. n.a., not available.

Lusher et al. [31], KOGENATE19/65117 (1/6)14n.a.10*n.a.
29.2%16.9%37%22% 15% 
Bray et al. [32]
Rotschild et al. [33]
Goodeve et al. [34], RECOMBINATE
22/7375 (0/5)1931/516*24
Lusher et al. [35], REFACTO32/101129 (1/8)5n.a.n.a.45
31.7%11.9%28%5%  45%
Kreuz et al. [36]9/6052 (0/2)538/571343
Oldenburg et al. [37], KOGENATE BAYER15%8.3%22%8%67%22%72%
Figure 1.

 Protective or predisposing effects of genetic and non-genetic factors that may potentially influence the risk of inhibitor development.

Conclusions and perspectives

The increasing and evolving knowledge of cellular immune response to exogenous FVIII provided new insights into the understanding of inhibitor development in haemophilia A. The environmental conditions at first FVIII exposures interplay with the patient’s genetic background, which influences the recognition of non-self; together with the F8 mutation type, an important role for immune-regulatory genes is emerging, consistent with the up- or down-regulation of cellular response against the foreign antigen in the presence (or absence) of danger signals. Therefore, inhibitor development should be considered a complex process in which multiple genetic and environmental factors interact dynamically, in varied scenarios related to the patient’s clinical conditions (type and severity of bleeding, concomitant diseases) and to treatment characteristics (prophylaxis or on-demand regimens, doses). Among the latter, the relationships between FVIII haplotypes in recipients and in products clinically administered [19] require further investigation in the light of the complexity of the other relevant genetic and non-genetic factors. The interaction of genetic and treatment-related risk factors is also the key for clinical stratification of risk, as reported in the predictive CANAL-derived score [10]. This information may suggest a careful assessment of clinical indications, doses and duration of first replacement treatments and to delay, when possible, elective surgeries [24,25], particularly for patients with high-risk genetic profiles. Early prophylaxis is considered the gold standard of treatment for children with severe haemophilia, but many barriers still hamper its clinical implementation [30]. The protective effects of regular prophylaxis started in the absence of immunological challenges [24,26] further encourage clinical efforts to extend the early start of prophylaxis in all patients, mainly when a high inhibitor risk is predictable. Presently, the potential clinical impact of these prevention strategies may be only speculative. However, two decades of clinical observations provided the pathophysiological background and highlighted the methodological approaches for addressing clinical trials in inhibitor patients, the most challenging issue of haemophilia treatment in the third millennium.


M.F. has received fees for the manuscript. A.C. has received speaker fees from Baxter, Bayer Schering Pharma and CSL Behring. C.S. has acted as a paid consultant for Bayer Schering Pharma. The other authors have declared no conflicts of interest.