Abstract 546

While adeno-associated virus (AAV) is a promising gene delivery vector, it has been challenging to deliver FVIII due to the large size of the FVIII cDNA and the high frequency of FVIII antibody formation in hemophilia A (HA) patients. We used two approaches to overcome the size limitation of AAV for FVIII: (1) two-chain delivery in which the canine FVIII (cFVIII) heavy chain (HC) is delivered in one AAV vector and the cFVIII light chain (LC) is delivered in a second AAV vector and (2) single chain delivery in which the B-domain deleted cFVIII cDNA with minimal regulatory elements is within one AAV vector. In the two-chain approach AAV-HC (4.0 Kb) and AAV-LC (3.9 Kb) with a liver specific promoter was co-injected at a dose of 6×1012 vector genomes/vector/kg or 1.25×1013vg/vector/kg using AAV8 or AAV9 via hepatic infusion. Five hemophilia A dogs treated with AAV-HC and AAV-LC expressed 0.5-11% cFVIII in a dose-dependent manner. The mean cFVIII activity based on Coatest assay for the low dose was 1.3% (>1220d)(Linus)(AAV8) and 0.6% (>1770d)(H19)(AAV9), while for the high dose it was 5.2% (800d)(F24)(AAV8) and 2.4% (>1270d)(Woodstock)(AAV9). One dog (J60) had a splenectomy due to a complication at the time of surgery and has maintained high levels of expression (mean 11.0%; >820d). The WBCT consistently remained at a mean of 17.6 min for low dose dogs and 13.7 min for high dose dogs compared to 8-12 min in normal dogs. Using novel reagents that we generated specific to cFVIII, we developed assays to detect cFVIII antigen levels and IgG antibodies. Despite receiving equal doses of each vector, at day 85 the cFVIII-LC antigen levels (71.7 ± 19.2 ng/ml) were >10-fold higher than would be predicted based on activity while the cFVIII-HC antigen levels (14.6 ± 9.2 ng/ml) were >3-fold higher than activity. Since functional FVIII synthesis relies on the co-transduction of AAV-HC and AAV-LC in the same cell, this suggests that only a portion of the vector co-transduces and expresses cFVIII in the same cell and that the light chain is secreted more efficiently than the HC. No IgG antibodies to cFVIII were detected at any time point in these dogs. Three dogs have maintained FVIII expression for >3.5 years and two dogs for >2 years with ongoing observation. No spontaneous bleeding episodes have been observed in these dogs for a cumulative observation of >16 years while >80 bleeding episodes would be expected during this time period. The second approach, the single chain delivery, overcomes the co-transduction requirement of the two-chain approach by ensuring that each transduced cell expresses functional FVIII. However, it is difficult to efficiently package the large 5.2 Kb single chain construct into an AAV vector. Since no significant differences were observed between AAV8 and AAV9 using the two-chain approach, we used AAV8 to deliver the single chain cFVIII by peripheral vein infusion at 2×1013vg/kg or 4×1013vg/kg. The mean cFVIII activity was 0.7% (>430d) for the low dose dog (L51) and 6.8% (>290d) and 2.2% (>110d) for the high dose dogs (M06, M50). cFVIII HC and LC ELISA showed that cFVIII antigen levels correlated with activity. WBCT was a mean of 19.1 min for L51, 15.3 min for M06 and 11.6 min for M50. No spontaneous bleeding episodes have been observed in these dogs. The high dose dogs had no IgG antibodies to FVIII. L51 had transient IgG antibodies to FVIII until d52 in the absence of a Bethesda titer. A rise in FVIII expression in L51 coincided with the disappearance of anti-cFVIII antibodies. Comparison of single chain and two-chain delivery of FVIII reveals that (1) long term therapeutic levels of cFVIII in a dose-dependent manner can be obtained with both delivery approaches; (2) circulating cFVIII antigen levels are >10-fold higher than activity in the two-chain delivery in contrast to single chain delivery in which antigen correlates with activity; and (3) high antigen levels may facilitate tolerance to FVIII in the setting of liver-directed gene transfer, since a transient non-inhibitory antibody was observed in only one dog with very low FVIII levels. Notably, no cellular toxicity due to continuous expression of various forms of FVIII was found in these animals based on long-term sustained FVIII expression levels and normal liver enzymes in all eight HA dogs. Further studies to characterize the immune responses to the transgene will define the optimal vector approach. These data will form the basis for clinical studies in humans with severe HA.

Disclosures:

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution