52. Therapeutic Factor VIII Expression After AAV Delivery in Non-Human Primates

Abstract

Adeno-associated viral (AAV) vector delivery of factor VIII (FVIII) has been challenging due to its intrinsic properties that result in inefficient expression compared to similarly sized proteins. Early studies of AAV delivery in hemophilia A mice and dogs suggested that the therapeutic vector dose for FVIII will be higher than for factor IX. However, higher vector loads may induce stronger immune responses against capsid antigens, as evidenced in the clinical studies of AAV delivery for hemophilia B. The use of codon-optimization and novel FVIII variants with enhanced biological properties may provide strategies to increase FVIII expression or secretion to support clinical studies for hemophilia A. One published study has reported clinically relevant levels of hFVIII following AAV-hFVIII delivery in non-human primates (NHPs). This study utilized a hFVIII variant that included a 17 amino acid synthetic sequence within the 14 amino acid B-domain region that increased hFVIII expression compared to the parental B-domain deleted FVIII-SQ transgene (McIntosh, 2013). While this and other variants may increase expression after AAV delivery, the use of non-native FVIII sequences may also increase the risk of development of neutralizing antibodies to potential neoantigens. In order to generate an AAV-hFVIII vector capable of expressing therapeutic levels of FVIII at a clinically relevant vector dose without introduction of any neoantigens, 28 hFVIII-SQ sequences were generated and introduced into our optimized expression cassette containing a modified transthyretin (TTRm) promoter. The constructs were initially screened by hydrodynamic delivery of plasmid DNA which identified 11 candidates that expressed FVIII 2-7 fold higher than our first generation codon optimized construct, CO3. AAV vectors (n=9) were generated using a novel AAV capsid, Spark100, with the best performing FVIII constructs. Hem A/CD4 KO mice were administered the vectors alongside CO3 (4×10e12vg/kg). At 8 weeks post vector administration, 2/9 expressed hFVIII similar to CO3, 5/9 were 4-8 fold higher than CO3 while 2/9 (SPK-8003 and SPK-8005) were >10 fold more potent than CO3. SPK-8005 was then evaluated in a dose escalation study in cynomologus macaques (n=3/group) treated with 3 doses: 2×10e12, 5×10e12 and 1×10e13 vg/kg and compared to vehicle controls (n=2). At 2 weeks post AAV administration, average hFVIII levels in the low, mid and high dose cohorts were 12.7 ± 2.1, 22.6 ± 0.8 and 54.1 ± 15.6 percent of normal, respectively. By 3-4 weeks, hFVIII expression started to decline in most of the animals concomitant with generation of antibodies against human FVIII. Of note, this is an expected and well-described observation that occurs in immune competent animal models due to differences between human and endogenous FVIII protein sequences. The 2 macaques that did not develop anti-hFVIII antibodies had sustained FVIII expression through the last time point evaluated. Finally, no vector-related toxicity events were observed. In summary, extensive codon-optimization identified novel AAV-hFVIII constructs capable of achieving therapeutic FVIII levels in macaques at clinically relevant doses. To our knowledge, the hFVIII levels observed in this study are the highest reported in a large animal model after treatment with an AAV vector expressing an unmodified FVIII-SQ protein. These safety and efficacy results in NHPs support the use of SPK-8005 hepatic gene transfer for the potential treatment of hemophilia A.