Hemophilia A Dogs Tolerant to Human Factor VIII Provide a Unique Model to Determine Efficacy and Safety of AAV Delivery of Novel Factor VIII Variants

The hemophilia dog models are valuable for evaluating the efficacy of novel hemophilia therapeutics. The hemophilia B dog was predictive of the therapeutic dose of adeno-associated viral (AAV) vector delivery of human factor IX in clinical trials. In addition, it provides an opportunity to study the long-term efficacy and safety after AAV administration. However, there are several challenges in using the hemophilia A (HA) dog model for gene therapy studies. First, canine factor VIII (cFVIII) has higher specific activity and increased rate of secretion compared to human FVIII (hFVIII). This significant difference between cFVIII and hFVIII prevents the use of a species-specific transgene to predict the efficacy of AAV-hFVIII. Second, the HA dogs are immunocompetent and develop an immune response to the xenoprotein, hFVIII, that precludes the ability to measure transgene expression. Therefore, in order to employ this valuable model for gene therapy studies, we generated a unique cohort of HA dogs that are tolerant to B-domain deleted (BDD) hFVIII. We hypothesized that tolerizing dogs to hFVIII will (1) permit accurate evaluation of hFVIII expression and thus predict the therapeutic vector dose and (2) allow the evaluation of the potential immune response to a novel hFVIII variant. To tolerize the dogs to hFVIII, neonatal HA dogs were treated with a retrovirus (RV-hAAT-hFVIII-BDD-WPRE, 3x10⁹ TU/kg) (n=5) on day 2 of life. The hFVIII expression was between 0.3%-6% at 4 weeks after RV delivery and plateaued after 6 months to 0.8% (S28), 0.3% (S29), 0% (V06), 1.5% (V26) and 1.7% (V27) based on Coatest assay. To determine if the dogs were tolerant to hFVIII-BDD, the dogs were challenged with hFVIII-BDD protein at 5-6 months post-RV administration (Xyntha, 25IU/kg per wk x 6 wks, I.V.). Anti-hFVIII antibodies were monitored closely throughout the challenge and up to 8 weeks after the last challenge. In 4 out of 5 dogs, no anti-hFVIII immune response was observed based on IgG1, IgG2, total IgG or Bethesda titer. In contrast, naïve HA dogs (n=2) developed high level anti-hFVIII IgG2 (1.2-3.2 μg/mL), total IgG (3.4-5.0 μg/mL), and Bethesda titer (4.1-67.8 BU/mL) after the same challenge regimen. Interestingly, the hFVIII activity in one RV-treated dog (V06) was undetectable at 6 months post-RV administration. After the challenge, V06 had anti-hFVIII IgG2 (1.7 μg/mL), total IgG (2.6 μg/mL), and a Bethesda titer (9.5 BU/mL), suggesting that FVIII must be maintained to achieve tolerance. These dogs were used to evaluate the efficacy of AAV serotype 8 (AAV8) delivery of a hFVIII-BDD codon-optimized sequence driven by a hepatocyte promoter, modified transthyretin promoter (TTRm). S29 was delivered AAV8-TTRm-hFVIII-CO (2x10¹² vg/kg). Prior to AAV delivery, the levels of hFVIII activity were 0.5-1% from the tolerization with the RV. After AAV administration the hFVIII activity was 3.8% at d168 and 4.7% at d387, resulting in a 4% increase in hFVIII expression. No anti-hFVIII antibodies were detected. The annual bleeding rate (ABR) for S29 post-RV delivery was 5 and after AAV delivery was 0, showing an improvement in the bleeding phenotype in contrast to untreated HA dogs (ABR=13, n=11). A hFVIII-tolerized littermate, S28, was recently treated with a hFVIII variant, AAV8-TTRm-hFVIII-CO-Δ3-SP/DE (2x10¹² vg/kg). The hFVIII-Δ3-SP/DE variant has a deletion of the furin site (1645-47) and replaces residues SD at 1657-58 with PE. This variant showed higher specific activity (2-fold) in vitro and increased secretion (4-fold) compared to wild type hFVIII-BDD in the setting of AAV delivery in HA mice. Based on the results in S28, we will determine the dosing of V26 and V27. These studies demonstrate that sustained low level hFVIII expression of 0.2-2% up to 4 years post-retroviral delivery were able to induce and maintain tolerance to hFVIII, while allowing for the subsequent assessment of AAV efficacy. A clinically relevant dose of AAV8-TTRm-hFVIII-CO resulted in therapeutic levels of hFVIII expression while ongoing studies will allow investigation of the efficacy of the hFVIII-BDD variant, Δ3-SP/DE, in the setting of AAV administration in a large animal model. Overall, these studies demonstrate that RV-targeting of hFVIII-BDD expression to the liver in neonatal HA dogs leads to tolerance to this xenoprotein and provide a unique large animal model to evaluate both efficacy as well as potential immunogenicity of novel FVIII variants.