Activity of transgene-produced B-domain-deleted factor VIII in human plasma following AAV5 gene therapy

Abstract

Adeno-associated virus (AAV)-based gene therapies can restore endogenous factor VIII (FVIII) expression in hemophilia A (HA). AAV vectors typically use a B-domain-deleted FVIII transgene, such as human FVIII-SQ in valoctocogene roxaparvovec (AAV5-FVIII-SQ). Surprisingly, the activity of transgene-produced FVIII-SQ was between 1.3 and 2.0 times higher in one-stage clot (OS) assays than in chromogenic-substrate (CS) assays, whereas recombinant FVIII-SQ products had lower OS than CS activity. Transgene-produced and recombinant FVIII-SQ showed comparable specific activity (international units per milligram) in the CS assay, demonstrating that the diverging activities arise in the OS assay. Higher OS activity for transgene-produced FVIII-SQ was observed across various assay kits and clinical laboratories, suggesting that intrinsic molecular features are potential root causes. Further experiments in 2 participants showed that transgene-produced FVIII-SQ accelerated early factor Xa and thrombin formation, which may explain the higher OS activity based on a kinetic bias between OS and CS assay readout times. Despite the faster onset of coagulation, global thrombin levels were unaffected. A correlation with joint bleeds suggested that both OS and CS assay remained clinically meaningful to distinguish hemophilic from nonhemophilic FVIII activity levels. During clinical development, the CS activity was chosen as a surrogate end point to conservatively assess hemostatic efficacy and enable comparison with recombinant FVIII-SQ products. Relevant trials are registered on clinicaltrials.gov as #NCT02576795 and #NCT03370913 and, respectively, on EudraCT (European Union Drug Regulating Authorities Clinical Trials Database; https://eudract.ema.europa.eu) as #2014-003880-38 and #2017-003215-19.

Graphical abstract

Figure 1.

Correlation between FVIII activities measured in OS and CS assays. FVIII activity data from OS and CS assays in BMN 270 clinical trials 270-201 (A), 270-301 (B), and in 20 healthy donors (C) were correlated using linear regression. Each data point represents FVIII activity data from 1 visit; multiple visits for the same participant are represented by the same colored symbol. The slope of the regression curve reflects the constant rate of change between the OS and CS activity measurements.

Figure 2.

Specific activity of recombinant and transgene-produced FVIII-SQ. (A-B) Recombinant FVIII-SQ (Xyntha) was spiked into plasma from individuals with severe HA, to determine the specific activity using the OS or CS assay. (C-D) Mean specific activity for transgene-produced FVIII-SQ in individual participants from BMN 270 clinical trial 270-201 was determined using the OS or CS assay. (E-F) Mean specific activity for transgene-produced FVIII-SQ in individual participants from BMN 270 clinical trial 270-301 was determined using the OS or CS assay. Dashed lines in panels A, C, and E represent the specification range for specific activity reported for Xyntha (5500-9900 IU/mg). Dashed lines in panels B, D, and F represent the specification range for specific activity reported for ReFacto (7 600-13 800 IU/mg). Given the lower activity of recombinant FVIII-SQ in the OS than in the CS assay, the specification range depends on which assay is used for product labeling (OS for Xyntha, CS for ReFacto). Error bars in panels C-F represent intraparticipant standard deviations.

Figure 3.

Kinetics of FIXa, FXa, and FIIa formation in the OS assay. Plasma samples containing transgene-produced FVIII-SQ (gene therapy plasma), collected from 2 different participants treated with BMN 270, and plasma samples containing native FVIII (normal plasma, diluted) with the same CS activity were compared in stop kinetics experiments in the OS assay. FIXa generation (A-B), FXa generation (C-D), and FIIa (thrombin) generation (E-F) were monitored for 60 seconds after addition of CaCl₂. Error bars represent standard deviations between replicate assessments.

Figure 4.

FIXa-triggered thrombin generation assay. Native FVIII in healthy donor samples (normal plasma; n = 40) was compared with transgene-produced FVIII-SQ in trial 270-301 (gene therapy plasma; n = 25), after individual participants (n = 7) had reached nonhemophilic CS activity levels (≥40 IU/dL) between weeks 23 and 26. Box-and-whisker plots were generated for thrombin peak height (A), ETP (B), lag time (C), and time-to-peak (D). The horizontal line in the box represents the median, the box represents the Q1 to Q3 interquartile range (IQR), and the whiskers indicate the minimum and maximum value that fell within Q1 to 1.5 × IQR and Q3 + 1.5 × IQR, respectively. Dashed horizontal lines represent the lower and upper limits of the normal reference range, respectively. P values represent significance levels from Mood’s median test.

Figure 5.

Correlation between frequency of joint bleeds and FVIII activity levels. For each participant, the number of treated joint bleeds per 4-week interval was correlated with the median FVIII activity within each interval, as measured using the OS (blue circles) or CS (red circles) assay in BMN 270 clinical trials 270-201 (A) and 270-301 (B). The predicted bleed frequency per 4-week interval was modeled by negative binomial regression for OS activity (blue solid line) and CS activity (red dashed line). Shaded colored areas represent the corresponding 95% CI. No joint bleeds were reported for FVIII activity >50 IU/dL in either assay; these data were included in the modeling but are not shown here.

Figure 6.

Illustration of kinetic bias leading to higher OS than CS activity of transgene-produced FVIII-SQ. Transgene-produced FVIII-SQ accelerates early FXa formation compared with native FVIII, resulting in a small increase in FXa concentration (picomolar range, pM). This leads to faster thrombin activation and clot formation in the OS assay, which may explain the higher OS measurements, given that this assay uses a kinetic end point (time to visible clot) within the first 1 to 2 minutes of the coagulation reaction. In contrast, the CS assay uses more dilute test samples and determines FXa concentrations after a longer, fixed incubation period of 5 minutes. At this point, FXa generation has been exponentially amplified (nanomolar range, nM) and presumably remains unaffected by the minute differences that occurred earlier in the reaction. The different timing of assay readout therefore leads to a kinetic bias between OS and CS assay, whereby the shorter assay (OS) reports higher FVIII activity values.