FVIII activity following FVIII protein infusion or FVIII gene transfer predicts the bleeding risk in hemophilia A rats

Abstract

Background: Prophylactic replacement therapy in hemophilia A (HA) patients does not adequately prevent bleeds and arthropathic complications. A more refined understanding of the relationship between coagulation factor VIII (FVIII) levels and bleeding risk during protein prophylaxis, or with gene therapy, is needed to improve patient care.

Objectives: Investigate this relationship in the HA rat, a model exhibiting spontaneous bleeds and development of arthropathy similar to HA patients.

Methods: Human B domain-deleted FVIII was delivered to HA rats via adeno-associated virus (AAV)-mediated gene transfer or multiple intravenous protein injections.

Results and conclusions: After 12 weeks of observation, both approaches significantly reduced bleeds per animal and increased the proportion of bleed-free animals compared with controls (43% vs 0%, respectively [AAV]; 75% vs 8%, respectively [injection]). Both approaches resulted in an anti-FVIII inhibitory response in 20% to 37% of treated animals, similar to HA patients. Inhibitory antibodies were refractory to clinical improvement (reduction of bleeds) only in the AAV-based prophylaxis. An integrated model-based analysis of data on FVIII exposure and bleeding events was performed. This predicted the bleeding risk at any given circulating FVIII activity. Specifically, 4.8 or 10 IU/dL FVIII (0.048 and 0.1 IU/mL, respectively) were predicted to reduce bleeding risk by 90% or 95%, respectively, compared with untreated controls. Our data establish the utility of the HA rat model in FVIII prophylaxis studies and describe how FVIII activity affects bleeding risk in this setting. These enable further studies on FVIII prophylaxis focusing on disease complications for an optimized treatment of HA patients.

Keywords: animal; genetic therapy; hemophilia A; models; risk; therapeutics.

Fig. 1.. Design to study the effects of hFVIII-BDD prophylaxis.

(A) At study week 0, HA rats were divided into four groups receiving different doses of an AAV-hFVIII-BDD injection. HA control animals received a PBS injection. Blood samples were collected from all rats pre-dosing week 0 and subsequently once weekly until study week 12 when the animals were euthanized following a final blood sample. (B) At study week 0, HA rats were assigned to prophylaxis treatment group and an untreated control group. Rats in the prophylaxis group received recombinant 50 IU/kg hFVIII-BDD, thrice weekly (Mondays, Wednesdays and Fridays) for eight weeks (week 0–7) and blood samples were collected pre-dosing on Mondays for weeks 0–9. Blood samples were also collected on week 0 and 9 in the control group. All animals were euthanized at week 9, following the final blood sampling.

Fig. 2.. hFVIII-BDD expression after AAV-mediated gene transfer.

(A) hFVIII-BDD antigen, (B) activity levels and (C) anti-hFVIII-BDD IgG in HA rats infused with AAV-hFVIII-BDD (N=30). (D) A total of 8 HA animals infused with AAV-hFVIII-BDD were positive for anti-hFVIII-BDD IgG, which was inhibitory in 6 of these animals (measured in Bethesda Units). (E) Kinetics of inhibitor formation in HA rats that were infused with AAV-hFVIII-BDD and developed anti-hFVIII inhibitors (N=6). Mean values of hFVIII antigen (hFVIII:Ag) and activity (hFVIII:C) are shown relative to mean level anti-hFVIII inhibitors (in BUs). (F) Mean hFVIII antigen and activity (hFVIII:Ag and hFVIII:C, respectively) in HA rats infused with AAV-hFVIII-BDD that did not develop anti-hFVIII inhibitors (N=24). For panels A-D, each symbol represents a single animal. Only animals that completed the study are depicted. One IU/dL is equivalent to 0.01 IU/ml.

Fig. 3.. hFVIII-BDD levels in HA rats receiving hFVIII-BDD protein prophylaxis.

(A) In vivo recovery (IVR) within 2–3 min after 1, 2 or 3 doses of hFVIII-BDD in HA rats. Data depicted was obtained from animals during dosing in week 0, shown as mean ± 1 standard deviation (SD). (B) Development of anti-hFVIII-BDD IgG in HA rats receiving protein prophylaxis (N=19). (C) Bethesda Unit titers of antibodies in HA rats receiving protein prophylaxis that were positive for anti-hFVIII-BDD IgG (N=7). In all panels, each symbol represents a single animal. One IU/dl is equivalent to 0.01 IU/ml.

Fig. 4.. Bleeds in HA rats following AAV-hFVIII-BDD gene transfer or hFVIII-BDD protein prophylaxis.

(A) The percentage of rats with no bleeds vs. time is shown for animals that received AAV-hFVIII-BDD (categorized as all animals [AAV All], with or without anti-hFVIII inhibitors [“AAV BU+” or “AAV BU-”, respectively]) or controls (HA). Statistical comparison is shown relative to HA controls. **P<0.01; NS: not statistically different. (B) Number of bleeds in animals that received AAV-hFVIII-BDD (categorized as all animals [“AAV All”], with or without anti-hFVIII inhibitors [“AAV BU+” or “AAV BU-”, respectively]) or controls (HA). Statistical comparison is shown relative to HA controls using a Kruskal-Wallis test. **P<0.01; NS: not statistically different. Data shown as mean ± 1 standard deviation (SD). Each symbol represents a single animal. (C) The proportion of animals with 0, 1, 2, 3 or 4 bleeds in the animals that received AAV-hFVIII-BDD or HA controls is shown as a pie-chart for each cohort. (D) The percentage of rats with no bleeds vs. time is shown for animals that received hFVIII-BDD prophylaxis (categorized as all animals [“Prophy All”], with or without anti-hFVIII inhibitors [“Prophy BU+” or “Prophy BU-”, respectively]) or controls (HA). Statistical comparison is shown relative to HA controls. **P<0.01; ***P<0.001; NS: not statistically different. (E) Number of bleeds in animals that received hFVIII-BDD prophylaxis (categorized as all animals [“Prophy All”], with or without anti-hFVIII inhibitors [“Prophy BU+” or “Prophy BU-”, respectively]) or controls (HA). Statistical comparison is shown relative to HA controls using a Kruskal-Wallis test. *P<0.05; **P<0.01; ***P<0.001; NS: not statistically different. Data shown as mean ± 1 SD. Each symbol represents a single animal. (F) The proportion of animals with 0, 1, 2, 3 or 4 bleeds in the animals that received hFVIII-BDD prophylaxis or HA controls is shown as a pie-chart for each cohort. Cohort size (N) is indicated in all relevant panels.

Fig. 5.. Model assessment based on visual predictive checks (VPCs).

(A) Prediction-corrected VPCs of the plasma concentration-time profiles of hFVIII in HA rats with quantifiable hFVIII expression following AAV-hFVIII-BDD administration week 0 (N=23, including 1 rat euthanized prematurely). The observed plasma concentrations of expressed hFVIII is depicted as dots; the median of the observed expressed hFVIII concentrations is depicted as solid line; the 95% confidence interval of the simulated median is depicted as red shaded area; the observed 2.5th and 97.5th percentiles is depicted as stippled lines; and the 95% confidence interval of the simulated lower 2.5th and upper 97.5th percentiles is depicted as blue shaded areas. (B) VPCs of the repeated time-to-event model comparing the median of the observed Kaplan-Meier curve (solid black line) in both in vivo studies combined (AAV and protein prophylaxis), to the 95% prediction interval of the simulated data (green area) for the 1st (Event 1), 2nd (Event 2), 3rd (Event 3) and 4th (Event 4) bleeding event based on 1,000 simulations. For generation of the Kaplan-Meier VPCs the observed time of bleeding events was set to the middle point of the interval-censored time intervals. One IU/dl is equivalent to 0.01 IU/ml.

Fig. 6.. Relationship between hFVIII levels and reduction in bleeding risk.

(A) Development of the PK-RTTE model allowed us to establish the link between hFVIII levels and bleeding risk. The EC₅₀ (effective hFVIII concentration resulting in a 50% hazard [instantaneous bleeding risk] reduction), was estimated to 0.53 IU/dl. The graph depicts drug effect as a function of hFVIII concentration. Stippled lines indicate the hFVIII concentration (IU/dl) needed (0.53, 2.1, 4.8 or 10) to reduce bleeding risk in treated compared to untreated rats by 50, 80, 90 and 95%, respectively. The reduction in hazard (drug effect) was calculated as shown. One IU/dl is equivalent to 0.01 IU/ml. (B) The developed PK-RTTE model allowed us to visualize how different hFVIII levels modulate bleeding risk. The graph illustrates the predicted mean probability of having one (green solid line), two (red dashed line) or three (purple dotted line) bleeding events within a period of 12 weeks of observation (age 4 to 16 weeks) versus hFVIII plasma concentration in the HA rats, based on 3,000 simulations. At 1 IU/dl hFVIII (0.01 IU/ml, corresponding to 1% in humans) the probability of experiencing one bleed in the study period was 51%, and at 10 IU/dl (0.1 IU/ml, corresponding to 10% in humans) hFVIII the probability of experiencing one bleed was reduced to 9%.