Buy Article Permissions and Reprints

Due to an absence or deficiency in factor VIII (FVIII), hemophilia A (HA) patients have bleeding episodes that range from bruising and joint bleeds to life-threatening hemorrhages. Accordingly, patients receive FVIII or emicizumab (a FVIII mimetic) to prophylactically prevent bleeding. As emicizumab does not treat bleeds or protect against surgery bleeds, patients on emicizumab may also require FVIII. However, exposure to FVIII can cause inhibitors (anti-FVIII neutralizing antibodies) in 30% of patients with severe disease. These inhibitors negate treatment efficacy,[1] resulting in increased morbidity and mortality, increased cost of care, and decreased quality of life.[2] While a few factors associated with an increased risk of initial inhibitor production have been identified,[3] [4] [5] [6] [7] the exact combination of determinants remain undefined, and thus no strategies exist to prevent inhibitor formation.

To date, all available FVIII products that differ in cell of origin, presence of B domain, and/or various modifications generate clinically significant rates of inhibitors (∼25–40%)[8] [9] in previously untreated patients (PUPs); direct comparison of inhibitor incidence between products is limited, as most studies were single-armed and nonrandomized. The randomized SIPPET trial found that plasma-derived FVIII (PD-FVIII) associates with a slightly lower but still clinically significant incidence of inhibitors than that observed in PUPs infused with standard half-life recombinant FVIII from hamster cells.[8] However, the RODIN study suggests that PD-FVIII, recombinant full-length FVIII (FL-FVIII), and B domain deleted FVIII (BDD-FVIII) products demonstrate a similar risk of inhibitor formation in PUPs, though FVIII products from Baby Hamster Kidney (BHK) cells were found to be more immunogenic than FVIII produced in Chinese Hamster Ovary (CHO) cells.[9]

The underlying mechanisms contributing to both the clinically significant rate of inhibitors observed with all FVIII products and to differences seen between products in some studies remain poorly understood. Understanding the immune response to distinct FVIII products may lead to identification of factors involved in responses to all FVIII products, and thereby approaches to prevent immunity to FVIII.[10] Several studies indicate that inhibitors form through a T cell-dependent process,[11] [12] [13] [14] [15] wherein CD4 T cells are required to form anti-FVIII immunoglobulin G (IgG). However, prior to a B cell receiving help from cognate CD4 T cells, an immunogen must first be recognized by cognate B cell receptors (BCRs) on a B cell. Corroborating this, preclinical studies indicate that shortly following infusion, FVIII is recognized by various immune populations that are posited to present FVIII to cognate BCRs on B cells.[16] [17] [18] Specifically, marginal zone (MZ) B cells have been shown to be required to generate an immune response to a hamster cell origin BDD-FVIII product,[17] and FL-FVIII has been shown to localize with MZ B cells following infusion. However, whether MZ B cells are necessary to generate an immune response to FL-FVIII and modified FVIII products remains to be evaluated. Given the conflicting data in the literature regarding whether distinct biophysical properties of FVIII products (e.g., BDD or containing, cell line of production, pre-complex to von Willebrand factor [VWF]) influence a patient's likelihood to form inhibitors and the strong preclinical evidence that MZ B cells are required to generate inhibitors against BDD-BHK,[17] we hypothesized that the role of MZ B cells in inhibitor formation is not conserved across FVIII products, contributing to the observed clinical differences in inhibitor incidence.

To test this, we first evaluated whether PD-FVIII and FL-FVIII localize with MZ B cells by infusing FVIII-deficient mice with PD-FVIII or FL-FVIII from CHO (CHO-FL) or BHK cells (BHK-FL). Consistent with a previous observation,[16] CHO-FL without or with linkage to polyethylene glycol (CHO-FL.PEG) and BHK-FL similarly localized with MZ B cells ([Fig. 1A, B]). Likewise, PD-FVIII (PD-FL) co-existed with MZ B cells ([Fig. 1A]), though to a lesser extent than FL-FVIII ([Fig. 1B]). As PD-FVIII is human-derived and pre-complexed to VWF, it is possible that glycosylation differences and/or VWF binding impacted PD-FVIII recognition by MZ B cells.

Fig. 1 MZ B cells similarly co-exist with FVIII products and are required to generate an antibody response to BDD and full-length FVIII products of various animal origins and with a protein modification. (A) FVIII-deficient mice were administered saline, full-length plasma derived FVIII (PD-FL; Alphanate, Grifols), recombinant FL-FVIII without (CHO-FL; Advate; Takeda) or with polyethylene glycol expressed by CHO cells (CHO-FL.PEG; Adynovate, Takeda), and recombinant FL-FVIII expressed by BHK cells (BHK-FL; Helixate, Bayer). Fifteen minutes after injection, spleens were harvested for confocal analysis of FVIII localization with MZ B cells that were delineated by CD1d (green), as previously described.[17] [22] IgD (blue) was used to identify follicular B cells. FVIII (red) co-existence with MZ B cells is depicted in yellow and with white arrows. (B) Co-localization of CHO-FL, CHO-FL.PEG, BHK-FL, and PD-FL with MZ B cells was quantified using the Fiji Coloc2 plugin and is demonstrated as a Pearson's correlation coefficient. (C) FVIII-deficient mice were administered saline, BDD-FVIII produced by CHO cells (CHO-BDD; Xyntha, Pfizer), BDD-FVIII expressed by BHK cells (BHK-BDD), BDD-FVIII made by HEK cells (HEK-BDD; Nuwiq, Octapharma), human-porcine FVIII (BDD.H)[24] expressed by BHK cells (BHK-BDD.H), and BDD-FVIII linked to the Fc portion of IgG1 (BDD-IgG1; Eloctate, Sanofi). Fifteen minutes after injection, spleens were harvested for confocal analysis of FVIII localization with MZ B cells that were delineated by CD1d (green), as previously described.[17] [22] IgD (blue) was used to identify follicular B cells. FVIII (red) co-existence with MZ B cells is depicted in yellow and with white arrows. (D) Co-localization of CHO-BDD, BHK-BDD, BHK-BDD.H, HEK-BDD, and BDD-IgG1 with MZ B cells was quantified using the Fiji Coloc2 plugin and is demonstrated as a Pearson's correlation coefficient. Images in panels (A) and (C) were acquired using a UPlanSApo 20x (0.75 NA air) objective and analyzed using Fiji version 2.0.0-rc-69/1.52n. (E) FVIII-deficient mice were administered MZ B cell depleting antibodies (anti-CD11a + anti-CD49d) or isotype control antibodies (Rat IgG2a + Rat IgG2b) on days −4, −2, +10, and +20. On day 0, mice were infused weekly for 4 weeks with 2 μg (molar equivalent to full-length FVIII derived from CHO cells) of recombinant full-length FVIII (CHO-FL and BHK-FL), BDD-FVIII (CHO-BDD and HEK-BDD), or a BDD-FVIII linked to the Fc portion of IgG1 (BDD-IgG1), followed by a 4 μg challenge. Plasma was collected 1 week post challenge (day 42) and evaluated for anti-FVIII IgG by ELISA. Statistics were generated using a (B and D) Kruskal–Wallis test with a post-Dunn's multiple comparison test or a (E) Mann–Whitney test. ****p < 0.0001, ***p < 0.001, **p < 0.01, n.s. = not significant.

While characterization of glycan signatures of PD-FVIII and FL-FVIII revealed that the B domain contains 19/25 potential glycosylation sites,[19] the RODIN study suggests that BDD-FVIII from hamster cells are no less immunogenic than PD-FVIII or FL-FVIII.[9] To examine this, FVIII-deficient mice infused with BDD-FVIII from human and hamster cells were evaluated for FVIII localization with MZ B cells. Removal of the B domain did not alter localization of BDD-FVIII products with MZ B cells ([Fig. 1C, D]). Moreover, addition of the Fc portion of IgG1 to BDD-FVIII from human embryonic kidney (HEK) cells (HEK-BDD) to extend its half-life did not alter the ability of BDD-IgG1 to co-exist with MZ B cells ([Fig. 1C, D]). These results suggest that post-translational modifications and intentional adaptations likely do not contribute to the ability of FVIII to co-exist with MZ B cells.

Although different FVIII products similarly co-existed with MZ B cells, it is possible these products differentially impact MZ B cells, and thereby the ability to generate anti-FVIII antibodies. Given this, we tested whether MZ B cells were required to generate antibodies against a subset of FVIII products. MZ B cell-depleted or isotype-control treated FVIII-deficient mice were administered CHO-FL, BHK-FL, BDD-FVIII from CHO or HEK, or BDD-IgG1 ([Fig. 1E]). Consistent with a previous report,[17] MZ B cell depletion prevented the development of an IgG response to FVIII ([Fig. 1E]) irrespective of the B domain or cell of origin. The requirement of MZ B cells in immunity to BDD-IgG1 was unexpected, as a previous study demonstrates that BDD-IgG1 induces inhibitory signals in B cells compared with recombinant FVIII.[20] Moreover, like patients who develop inhibitors to BDD-IgG1, our preclinical model of HA generated a MZ B cell-dependent anti-FVIII IgG response.

In summary, these results highlight MZ B cells as a conserved initiating player in the immune response to a diverse group of FVIII products. Moreover, these data add to the growing evidence of the role of MZ B cells in the generation of immune responses to intravascularly encountered allogeneic antigens.[17] [21] [22] [23] These findings thereby not only rule out another layer of initiating events that may underlie differential inhibitor formation but also demonstrate the continued need for investigation into how biophysical properties of FVIII products may possess the ability to influence immunity to FVIII.

Ethical Approval Statement

All procedures were performed according to an approved Institutional Animal Care and Use Committee protocol.

*These authors contributed equally.

Publication History

Received: 03 June 2024

Accepted: 30 June 2024

Article published online:
16 July 2024

© 2024. Thieme. All rights reserved.

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany