Nonhuman sugars cause factor VIII troubles

Immune responses to factor VIII (FVIII), necessary for normal coagulation, remain the most significant complication in treating patients with severe hemophilia A. In this issue of Blood, Arthur et al¹ suggest that posttranslational modifications of recombinant FVIII (rFVIII) products with the non-human carbohydrate αgalactose (α-Gal) promotes the development of anti-FVIII antibodies that prevent FVIII function.

rFVIII products are life-saving interventions for patients with severe hemophilia A. The immune response in the form of neutralizing anti-FVIII antibodies that develop in ∼30% of patients with severe hemophilia A remains the most severe complication in FVIII replacement therapy.² FVIII-neutralizing antibodies, known as inhibitors, prevent FVIII function and directly increase morbidity and mortality. Treatment of patients with an immunogenic response to FVIII currently requires high-intensity factor administration to induce immune tolerance. If this approach fails, long-term use of costly and less effective bypassing therapies becomes necessary.³ Why FVIII-neutralizing antibodies develop in only some patients remains unclear.

Independent cohort studies have described differences among different rFVIII concentrates. A second-generation full-length rFVIII is associated with a 1.6-fold increase in inhibitor risk compared with third-generation products. Another retrospective analysis further reported hazard ratios of 2.81 and 1.64 for inhibitor incidence with second- and third-generation rFVIII, respectively, compared with plasma-derived FVIII.^4-6 The second- and third-generation products refer specifically to Kogenate FS (Bayer) produced in baby hamster kidney cells (BHK; BHK-rFVIII) and Advate (Shire) produced in Chinese hamster ovary cells (CHO; CHO-rFVIII). On the protein level, BHK-rFVIII and CHO-rFVIII differ by a single amino acid at position 1241 in the B domain, aspartic acid, and glutamic acid, respectively. However, the immunologic relevance of this substitution for anti-rFVIII antibody development appears insignificant.

Another hypothesis proposes that carbohydrate modifications of FVIII differ between BHK-rFVIII and CHO-rFVIII and cause differential immunogenicity. FVIII has 25 putative N-linked carbohydrate sites (see figure). Incomplete occupancy of the N-linked carbohydrate sites and incomplete carbohydrate structures contribute to the development of anti-rFVIII antibodies.^4-6 The formation of immunoglobulin M- and immunoglobulin G-FVIII immune complexes contribute to the enhanced clearance of rFVIII in hemophilia A mouse models. N-linked carbohydrates also assume a shielding function, preventing binding of neutralizing anti-rFVIII antibodies.⁷ Non-human cell lines such as BHK and CHO cells can add immunogenic non-human glycan structures, Gal(α1-3)Galβ1-GlcNAc-R (αGal) and 5-glycolylneuraminic acid (Neu5Gc), to rFVIII.^4-6 However, distinct differences in the quantity and localization of αGal between BHK-derived and CHO-derived rFVIII remain unclear.

In this paper, by using mass spectrometry, Arthur et al show that BHK-derived FVIII holds higher non-human αGal levels than CHO-derived FVIII. αGal presence was further demonstrated on immobilized rFVIII and anti-αGal lectin binding. Using αGal knockout mice, which spontaneously generate anti-αGal antibodies, the authors found that BHK-derived FVIII elicits heightened reactivity with serum from αGal knockout mice and increased immunogenicity in αGal knockout recipients. Their data suggest that αGal incorporation onto FVIII carbohydrate structures may result in anti-αGal antibody recognition that could positively influence the development of anti-FVIII antibodies. Consistent with this, BHK-derived FVIII has increased αGal levels, which correspond to increased reactivity with anti-αGal antibodies. Infusion of BHK-derived FVIII, but not CHO-derived FVIII, into αGal knockout mice, which spontaneously generate anti-αGal antibodies, resulted in significantly higher anti-FVIII antibody formation, which suggests that the increased levels of αGal on BHK-derived FVIII can influence immunogenicity in mouse models.

Although this paper convincingly shows the differential presence of αGal on rFVIII between non-human cell line expression systems and the role of αGal presentation and site, there are still multiple gaps in our understanding of the immunogenicity of the 2 products. Specifically, the use of glycoproteomics mass spectrometry approaches to identify structural carbohydrate differences and αGal additions could shed additional light on why different expression systems add more αGal decorations and how this action might affect immune responses. Although the αGal knockout mouse model indicates that more αGal expression increases antibody production, the mechanism by which immune cells recognize different levels and/or presentations of αGal to generate antibodies is unknown. Also, it is unclear if any of the antibodies are neutralizing antibodies. The findings of this study do not explain why only certain patients develop inhibiting antibodies. Mouse models are unlikely to elucidate the differences in immune response between humans. Finally, the authors did not address the role of non-human sialic acid moieties as potential immunogenic structures.

In conclusion, this work furthers our understanding of antibody formation to rFVIII. Much work and insight are needed, however, to resolve the ongoing mystery of interindividual differences in response to non-human carbohydrates in general and on rFVIII specifically.