Exorcizing the devil of factor VIII inhibitors Open Access

“For the Devil is a gentleman, and doesn’t keep his word” – G.K. Chesterton

In this issue of Blood Advances, Butera et al¹ shed some light on the mystery of factor VIII (FVIII) inhibitors, the most diabolical complication of FVIII replacement therapy in patients with hemophilia A. Their work examined the impact of disulfide bonds on the antigenicity of the FVIII molecule, with possible implications for the understanding of immune responses to exogenous FVIII and for drug design.

An immunoglobulin G (IgG) molecule, often remarked to have the shape of the letter Y, might more fancifully be thought to resemble the face of a devil: the constant or crystallizable Fc region making an elongated face with perhaps a pointed beard, the antigen-binding Fab regions forming the horns.

Maybe it is only a hematologist who would make this sinister connection. To most people, IgG is a guardian, an essential agent of the adaptive immune system. But for the hemophilia treater and her patients, IgG can fall from grace. IgG with specificity for functional epitopes on the FVIII molecule, called FVIII inhibitors, can inactivate infused FVIII. And so, treatment with FVIII will work just fine, unless the patient gets an inhibitor and is rendered worse off than before, an uncomfortably Faustian bargain.

There are many unanswered questions about FVIII inhibitors, but perhaps the most essential one is this: Why do they occur in some patients but not in others? The work of Butera et al¹ does not answer this question, which is sulfurously difficult, but does provide an intriguing glimpse through the smoke by considering disulfide bonds.

Disulfide bonds are ubiquitous elements of protein structure. Recent work has demonstrated that disulfide bonds need not all be present in a protein after synthesis, and a protein may exist in a variety of partially disulfide-bonded states. Butera’s team demonstrated that this is the case for both a commercial recombinant FVIII product and a commercial plasma-derived FVIII product; the molecule’s 8 disulfide bonds can be unformed in a large proportion of FVIII molecules. They then set out to examine the effect that this variety of sulfide-bonding states has on FVIII’s antigenicity.

A collection of patient-derived monoclonal anti-FVIII IgG were observed to bind preferentially to FVIII in specific disulfide bonding states, particularly to FVIII molecules in which a specific subset of the 8 disulfide bonds were unformed. The antibodies did not directly reduce the disulfide bonds, and antibody affinity was altered by the disulfide bonding state even if the antibody’s specificity was for an epitope distant from the disulfide bond site. This all suggests that incomplete disulfide bond formation can induce conformational changes in FVIII that affect antibody affinity. In silico simulation supports this conclusion.

The existence of a variety of FVIII molecules, defined by their disulfide bonding states, raises interesting questions. Do these different FVIIIs behave differently in terms of von Willebrand factor binding and, consequently, pharmacokinetics? What role do synthesis and packaging in Weibel-Palade bodies play in determining the disulfide bonding state? Do conformationally distinct FVIIIs behave differently in the intrinsic tenase complex?

But Butera et al’s work has to do with immune responses to FVIII, and this is the area in which their results potentially seem most consequential. Their work is all to do with FVIII’s antigenicity, the molecule’s ability to provide binding targets for anti-FVIII IgG. And so questions about the relationship between disulfide bonding states and FVIII’s immunogenicity, the ability of exogenous FVIII to incite an antigen-specific immune response in the first place, are unanswered.

And this is an area of some mystery, because FVIII is a somewhat peculiar antigen. The incidence of inhibitors in severe hemophilia A is generally observed to be ∼25% to 30% with a much lower incidence in severe hemophilia B.² Why? Some enzyme replacement therapies, such as those used in Fabry disease,³ have a much higher incidence of antidrug antibodies than FVIII. Why? Do variable immune responses to heterogeneous populations of FVIII molecules create a lottery in which some patients develop inhibitors and some do not, in a stochastic fashion, based on randomly determined disulfide bonding states? Do different patterns of disulfide bonding states in other molecules contribute to higher or lower probabilities of developing antidrug antibodies?

If that is the case, then would FVIII that has been engineered to have complete disulfide bonding be less of a target to antibodies in a clinically important way? Existing therapies to bypass FVIII inhibitors, namely recombinant activated factor VII (rVIIa) and factor eight inhibitor bypassing activity (FEIBA), are generally considered to provide hemostasis that is inferior to FVIII and to have substantial limitations, including rVIIa’s well-known very short half-life and a risk for thrombosis and thrombotic microangiopathy when FEIBA is used with emicizumab.⁴ Imagine being able to bypass a FVIII inhibitor with…FVIII! Specifically, a FVIII that has been engineered to be less visible to anti-FVIII IgG. (This is not that hard to imagine, because high-dose FVIII is currently used to overwhelm, rather than bypass, low-titer inhibitors).

Novel nonfactor therapies, such as emicizumab, have surely lessened the burden of FVIII inhibitors for many patients, and it is tempting to regard inhibitors as a problem that has been solved. But, to quote Charles Baudelaire rather than scripture, “The greatest trick the devil ever pulled was convincing the world he didn't exist.” Every patient with an inhibitor remains a calamity that is waiting to happen, particularly when treatment is required to treat or prevent severe bleeding. An improved understanding of FVIII inhibitors of the sort provided by Butera et al is still needed. Their work here provides a fascinating foundation for continued investigation and perhaps even illumination in both basic and clinical science.

Conflict-of-interest disclosure: The author declares no competing financial interests.