http://www.nejm.org/doi/full/10.

1056/NEJMoa1208024#t=article
Factor VIII Products and Inhibitor Development in Severe Hemophilia A
Samantha C. Gouw, M.D., Ph.D., Johanna G. van der Bom, M.D., Ph.D., Rolf Ljung, M.D., Ph.D.,
Carmen Escuriola, M.D., Ana R. Cid, M.D., Sgolne Claeyssens-Donadel, M.D., Christel van Geet,
M.D., Ph.D., Gili Kenet, M.D., Anne Mkipernaa, M.D., Ph.D., Angelo Claudio Molinari, M.D.,
Wolfgang Muntean, M.D., Rainer Kobelt, M.D., George Rivard, M.D., Elena Santagostino, M.D., Ph.D.,
Angela Thomas, M.D., Ph.D., and H. Marijke van den Berg, M.D., Ph.D. for the PedNet and RODIN
Study Group
N Engl J Med 2013; 368:231-239January 17, 2013DOI: 10.1056/NEJMoa1208024

Patients with severe hemophilia A have a deficiency of functional clotting factor VIII (<0.01 IU per milliliter)
and have bleeding in the joints and muscles. To prevent joint destruction, the current standard of care for
children with severe hemophilia A is primary prophylaxis. This includes regular infusions of factor VIII,
which are initiated at the time of the first episode of bleeding in a joint or earlier, aiming at the prevention of
joint damage.1 However, in about 30% of children, inhibitory antibodies to infused factor VIII products
develop, making usual treatment with factor VIII and prophylaxis impossible. There are multiple risk factors
for the development of inhibitory antibodies (inhibitor development).2-18
It has been suggested that recombinant factor VIII products are more immunogenic than plasma-derived
products. However, the outcomes of numerous studies and systematic reviews have been contradictory.19-
23 The studies have been limited by the enrollment of small, heterogeneous study populations and the use
of several factor VIII products, and comparisons among studies have been difficult because of different
study designs.22,24 The inclusion of minimally treated patients and patients who were still at risk for
subsequent development of inhibitory antibodies has led to an underestimation of the incidence of inhibitor
development.25 In addition, prospective postmarketing studies could not include high-risk children who
started bleeding at an early age, which meant that the risk of inhibitor development was underestimated.
Furthermore, small studies with extreme results are more likely to be published than are those with less
extreme findings.22 For these reasons, three systematic reviews of the immunogenicity of factor VIII
products resulted in different conclusions.21-23 Randomized trials comparing the immunogenicity of factor
VIII products have not yet been completed.26
A finding that recombinant and plasma-derived products had a differential risk with respect to inhibitor
development would influence both the decision about which type of product to administer in individual
patients and the availability of the preferred product. Therefore, knowledge of the risk of inhibitor
development associated with recombinant and plasma-derived products is important for both the individual
patient with hemophilia and the hemophilia population as a whole. We assessed whether the type of factor
VIII product and switching among products were associated with inhibitor development in previously
untreated children with severe hemophilia A.
METHODS
Patients
We enrolled consecutive, previously untreated patients with severe hemophilia A (factor VIII activity, <0.01
IU per milliliter) that had been diagnosed in 1 of the 29 participating hemophilia treatment centers. All the
children in the study were born between January 1, 2000, and January 1, 2010. Children who had been
referred to the centers because of the presence of inhibitory antibodies were excluded from the study.
Approval was obtained from the institutional review board at each study center. Parents or guardians of all
children provided written informed consent.
Data Collection
We uniformly collected detailed data on all infusions of factor VIII for up to 75 exposure days or until the
development of inhibitory antibodies, including dates of infusion, doses and brands of factor VIII products,
reasons for treatment, types of bleeding, extravasation of products, and surgery.
Patients were followed until the development of a clinically relevant inhibitory antibody or a cumulative
number of 75 exposure days. (After 75 exposure days, inhibitor development becomes rare [approximately
2 to 5 cases per 1000 patient-years]).27
Outcomes
The primary outcome was the development of clinically relevant inhibitory antibodies, which was defined as
at least two positive inhibitor titers combined with decreased in vivo recovery of factor VIII levels up to the
75th exposure day. The secondary outcome was the development of a high-titer inhibitor, which was
defined as a peak titer of at least 5 Bethesda units per milliliter up to the 75th exposure day.28 A positive
inhibitor titer was defined according to the cutoff level of the inhibitor assay used in the laboratory at each
center. Factor VIII recovery was described as decreased if the level of factor VIII activity was less than 66%
of the expected level 15 minutes after the infusion of factor VIII. The expected level of factor VIII activity
was calculated according to the criteria of Lee et al.29
In the majority of centers (92%), patients were routinely screened for inhibitor development after every 1 to
5 exposure days during the first 20 exposure days and at least every 3 months thereafter. At all centers,
patients were closely monitored for signs of inhibitor development, and investigators performed inhibitor
and recovery testing if there was any suspicion that inhibitory antibodies had developed.
Types of Factor VIII Products
We assessed the incidence of inhibitor development according to the type of product used at subsequent
exposure days (time-varying determinant). We categorized factor VIII products in several ways. First, we
compared the inhibitor risk between plasma-derived factor VIII products and recombinant products.
Second, to investigate whether the content of von Willebrand factor was associated with the risk of inhibitor
development, we categorized factor VIII products into products containing no von Willebrand factor (all
recombinant products), products containing less than 0.01 IU of von Willebrand factor antigen per
international unit of factor VIII antigen (monoclonal antibodypurified plasma-derived products), and
products containing 0.01 IU or more of von Willebrand factor per international unit of factor VIII antigen
(other plasma-derived products).30 Third, we compared inhibitor incidence among the following categories
of factor VIII products: plasma-derived products, first-generation full-length recombinant product (derived
from the full-length complementary DNA sequence of human factor VIII) (Recombinate, Baxter
BioScience), second-generation B-domaindeleted recombinant product, and second- and third-generation
full-length recombinant products.
We did not evaluate Kogenate (Bayer Healthcare), a first-generation full-length recombinant product, or
Refacto AF (Pfizer), a third-generation B-domaindeleted product, because of the small numbers of
patients who received these products (10 patients [7 as first-use product] and 3 patients [3 as first-use
product], respectively). The product type that was used most frequently was selected as the reference
category.
Switching among Products
We evaluated the risk of inhibitor development in children who were receiving a plasma-derived product
who were then switched to a recombinant product, as compared with those who were still receiving a
plasma-derived product. We similarly assessed the association between switching among various types of
factor VIII products and the development of inhibitory antibodies.
Study Conduct
The study was supported by unrestricted research grants from Bayer Healthcare and Baxter BioScience.
The companies did not have a role in the study design, data collection, data analysis, or writing of the
manuscript. Representatives of the companies reviewed the manuscript before it was submitted for
publication. No one who is not an author contributed to the writing of the manuscript.
Three of the authors (including the first author) designed the study, performed statistical analyses,
interpreted the data, and vouch for the integrity of the data, the fidelity of the study to the protocol, and the
accuracy of the data analyses. The first author wrote the first draft of the manuscript. All the other authors
collected the data, critically reviewed the manuscript, and made the decision to submit the manuscript for
publication.
Statistical Analysis
The absolute risk of inhibitor development varies according to the cumulative number of exposure days. To
account for this varying risk, we used pooled logistic regression with the cumulative number of exposure
days as the time variable instead of calendar time. We pooled observations over all exposure days for all
patients into a single sample and then used a logistic-regression model with stratification according to
number of exposure days to relate the risk factors to inhibitor development. This method accounts for
varying risks according to the cumulative number of exposure days and is equivalent to Cox regression
with exposure days as time-variable and time-dependent covariates.31 Relative hazard rates were
interpreted as relative risks.
We calculated both unadjusted and adjusted hazard ratios, with the latter adjusted for race or ethnic group;
age at first exposure to factor VIII; reason for first treatment; interval between exposure days; dose of factor
VIII; F8 genotype; and status with respect to family history of hemophilia and inhibitors, history of switching
among product brands, peak treatment episodes (defined as treatment with factor VIII for bleeding or for
surgery on either 3 consecutive days or 5 consecutive days), a history of major surgery, and regular
prophylaxis. Coding details are provided in theSupplementary Appendix, available with the full text of this
article at NEJM.org.
To assess whether our findings were robust, we also compared the incidence of inhibitor development
according to the product brands used at the first exposure to factor VIII (a time-fixed determinant). We
repeated the analyses in the subgroup of patients who were not included in registration trials (primary
studies of safety and efficacy) for previously untreated patients. An independent statistician who was
unaware of product types repeated all results by means of Cox proportional-hazards regression models.
RESULTS
Patients
A total of 648 patients were eligible for the study. Of these, 17 patients were excluded because of pending
informed consent, and 25 patients were excluded by the investigators for various other reasons (Figure
1FIGURE 1 Enrollment and Outcomes.). Baseline data were available for the remaining 606
patients; the analysis included 574 of these patients (94.7%), for whom detailed exposure data were
available. Their characteristics according to the type of factor VIII product that was first used are presented
in Table 1TABLE 1 Characteristics of the Patients and the Type of Factor VIII Product
Administered during the First Treatment. (see also Table 1S in the Supplementary Appendix).
Primary Outcome
Clinically relevant inhibitory antibodies developed in 177 patients (cumulative incidence, 32.4%; 95%
confidence interval [CI], 28.5 to 36.3). Of these patients, 116 had high-titer inhibitors (cumulative incidence,
22.4%; 95% CI, 18.8 to 26.0). Inhibitory antibodies developed after a median of 15 exposure days
(interquartile range, 10 to 20) at a median age of 15.5 months (interquartile range, 10.7 to 19.6).
Plasma-Derived versus Recombinant Products
Plasma-derived products were used on 4018 exposure days, and recombinant products were used on
25,661 exposure days. Plasma-derived products carried a risk of inhibitor development that was similar to
the risk with recombinant products (adjusted hazard ratio as compared with recombinant products, 0.96;
95% CI, 0.62 to 1.49) (Table 2TABLE 2 Risk of Inhibitor Development, According to the
Type of Factor VIII Product. and Figure 2FIGURE 2 Adjusted Relative Risk of Inhibitor
Development, According to the Type of Factor VIII Product.).
Content of von Willebrand Factor
Seven patients received products with a low von Willebrand factor content on 1 to 11 exposure days (total,
26 exposure days). Inhibitory antibodies developed in two patients after receiving a product with a low von
Willebrand factor content. The risk of inhibitor development with products containing a high amount of von
Willebrand factor was similar to the risk with products containing no von Willebrand factor (adjusted hazard
ratio, 0.90; 95% CI, 0.57 to 1.41).
Specific Product Types
The risk of inhibitor development was similar among plasma-derived products, first-generation full-length
recombinant products, second-generation B-domaindeleted recombinant products, and third-generation
recombinant products. First-generation recombinant products were associated with an unadjusted hazard
ratio of 1.44 (95% CI, 0.71 to 2.90) for high-titer inhibitor development; however, after adjustment, the
hazard ratio was lower. Second-generation full-length recombinant products were associated with a
significantly higher risk of inhibitor development than were third-generation products (adjusted hazard ratio,
1.60; 95% CI, 1.08 to 2.37; P=0.02); for high-titer inhibitor development, the adjusted hazard ratio was 1.79
(95% CI, 1.09 to 2.94; P=0.02) (Table 2).
Switching among Products
Details about the analyses of switching among brands of factor VIII are provided in theSupplementary
Appendix. Switching among products was not associated with the risk of inhibitor development (adjusted
hazard ratio as compared with no switching, 0.99; 95% CI, 0.63 to 1.56) (Table 3S in the Supplementary
Appendix).
Sensitivity Analyses
The results of sensitivity analyses regarding the use of factor VIII products (plasma-derived vs.
recombinant products and specific product types) were similar to those of the primary analysis. Details
regarding the sensitivity analyses are provided in the Supplementary Appendix.
DISCUSSION
In this cohort study involving 574 consecutive, previously untreated children with severe hemophilia A who
were born between 2000 and 2010, recombinant factor VIII products conferred a risk of inhibitor
development that was similar to the risk conferred by plasma-derived products. The von Willebrand factor
content in factor VIII products was not associated with inhibitor development. Second-generation full-length
recombinant products were associated with a higher risk of inhibitor development than were third-
generation full-length products. Switching from a plasma-derived product to a recombinant product or
switching among brands of factor VIII products did not result in an increased risk of inhibitor development.
We directly compared the use of recombinant products and plasma-derived products in one study cohort.
We avoided selection bias by including all consecutive patients who were born between January 1, 2000,
and January 1, 2010. We excluded all patients who were referred from nonparticipating hemophilia centers
because of inhibitor development. We used survival analysis because at the moment of data analysis a
number of patients had not yet reached the study end point and were still at risk for inhibitor development.
This enabled us to include all patients up to the last exposure day and to calculate cumulative incidences.
Collection of detailed information on all 75 exposure days allowed us to adjust the associations for potential
confounding factors. These findings were robust in sensitivity analyses.
Even though we adjusted for potentially confounding factors, we cannot rule out residual confounding. The
observed associations may have been affected by information bias, if frequencies and methods of inhibitor
screening among centers had differed according to the particular factor VIII product. However, we would
not expect that the cumulative incidence of high-titer inhibitor development would be influenced by
variations in inhibitor screening because of a lack of central laboratory testing, since these inhibitory
antibodies will always be detected clinically. Since the associations were similar with respect to both all
clinically relevant inhibitor development and high-titer inhibitor development, information bias would
therefore not have played a major role. Because a relatively small number of patients were treated with
plasma-derived products and because of the variety of plasma-derived products, we may not have been
able to detect potential differences in the risk of inhibitor development among various plasma-derived
products.
Several reports have suggested that plasma-derived factor VIII products, especially those containing
considerable amounts of von Willebrand factor, are less immunogenic than recombinant
products.19,32,33 However, several systematic reviews have yielded inconclusive results.21-23 Our results
are in agreement with the findings of a similarly designed study, the Concerted Action on Neutralizing
Antibodies in Severe Hemophilia A (CANAL) study, in which the risk of inhibitor development was not
clearly lower with plasma-derived products than with recombinant products (relative risk, 0.79; 95% CI,
0.49 to 1.28).20
Unexpectedly, the risk of inhibitor development was 60% higher among children receiving a second-
generation full-length recombinant product than among those receiving a third-generation full-length
product. This association may be a biased finding (through confounding, selection bias, or information
bias), a chance finding, or a causal effect.
We accounted for bias from confounding by adjusting the association for multiple potential confounding
factors. We summarized potentially confounding factors according to the product type used at the first
treatment (Table 1). Children at increased risk for inhibitor development did not receive second-generation
full-length factor VIII products more often than they did third-generation products. Therefore, confounding
does not seem to explain this association.
We avoided selection bias by including all consecutive patients and by excluding all patients who were
referred to the participating center because of a known increased risk for inhibitors. In addition, to further
confirm the absence of selection bias, we performed a sensitivity analysis among patients who were not in
a safety and efficacy trial. Patients who were included in such a trial might have been at reduced risk for
inhibitor development, since they did not have early bleeding.
The observed increase in the risk of inhibitor development with second-generation full-length recombinant
products as compared with third-generation full-length products is not likely to be affected by information
bias, since it is unlikely that patients who were treated with a second-generation full-length product were
more often screened for inhibitors than those treated with a third-generation product. Furthermore, we
observed a similar association in high-titer development. Thus, selection bias and information bias do not
explain the observed increase in risk in second-generation full-length recombinant products, as compared
with third-generation full-length products.
This difference in risk between recombinant products may be due to chance, which seems unlikely, given
the precision of our estimate of effect. But as long as the observation in our study is not confirmed in other
studies, we cannot exclude the possibility. However, since bias is unlikely and the probability of a chance
finding is low, the observed increase in the risk of inhibitor development in patients receiving second-
generation full-length factor VIII products may be real.
Other studies including a systematic review23 and reports of the Kogenate Bayer Study Group25,34
have not shown significant differences in the risk of inhibitor development among various recombinant
factor VIII products. In the registration studies, the incidence of inhibitor development may have been
underestimated because of the inclusion of patients who had already been treated with factor VIII on
several exposure days and a short follow-up period for the subgroup of patients who were still at risk for
inhibitor development. There is no straightforward biologic explanation for a difference in immunogenicity
among recombinant factor VIII products. Further studies are needed to verify these observations and to
identify biologic explanations.
In conclusion, the use of recombinant factor VIII products in children with severe hemophilia A did not have
a significant effect on the risk of inhibitor development, as compared with the use of plasma-derived
products, nor was the von Willebrand factor content of the products or switching among them associated
with the risk of inhibitor development. An unexpected finding was that second-generation full-length
recombinant products were associated with an increased risk of inhibitor development, as compared with
third-generation products.