Exposure-Response Relationship between Von Willebrand Factor (VWF) Activity and Bleeding Events Following Prophylaxis with Recombinant VWF (rVWF, vonicog alfa) in Patients with Severe Von Willebrand Disease (VWD)

Background: rVWF (vonicog alfa; Vonvendi ^®,Baxalta US Inc, a Takeda company, Lexington, MA, USA) is a purified recombinant VWF concentrate approved for on-demand (OD) treatment of hemorrhage and management of surgical bleeding in adults with VWD. The efficacy and safety of rVWF prophylaxis has also been evaluated in a recent open-label phase 3 trial in adults with severe VWD (NCT02973087).

Aims: To evaluate the exposure-response (ER) relationship between VWF activity (measured by VWF:ristocetin cofactor [RCo]), endogenous factor VIII (FVIII) activity (measured by FVIII:C), and spontaneous bleeding events (sBEs) in patients with severe VWD receiving rVWF prophylaxis for up to 1 year.

Methods: The modeling framework involved developing population pharmacokinetic (PK) and PK/pharmacodynamic (PD) models and conducting simulations to characterize VWF activity/PK and FVIII activity/PD, then developing an ER model for VWF and FVIII activities in association with sBEs. The population PK and PK/PD models were developed using data from 100 unique patients receiving intravenous rVWF in 4 completed clinical studies (NCT00816660; NCT01410227; NCT02283268; NCT02973087). The PK and PK/PD analyses were used to generate model parameters and evaluate predictors of heterogeneity for PK/VWF:RCo and PD/FVIII:C time profiles. The ER relationship was evaluated using sBEs from the phase 3 rVWF prophylaxis trial (NCT02973087) in 23 adults with severe VWD (VWF:RCo <20 IU/dL) requiring VWF therapy to manage BEs during the year before study entry: 13/23 patients were previously treated OD with a VWF (Prior OD group) and 10/23 had received plasma-derived VWF (pdVWF) prophylaxis (Switch group). For this ER evaluation, a repeated time-to-event (RTTE) model was used, including a piecewise exponential additive model, and the covariate effect of previous treatment (OD with a VWF or prophylaxis with pdVWF) was explored. Model selection was performed by comparing the goodness of fit of linear and nonlinear ER models based on the PK and PK/PD models' predicted values for 1) VWF:RCo and FVIII:C trough levels; 2) average VWF:RCo and FVIII:C levels in dosing interval; and 3) average VWF:RCo and FVIII:C levels over 24 h following rVWF treatment, with comparisons of these levels before sBE onset in patients with and without sBEs. The PK and PK/PD models were also used to derive the VWF:RCo and FVIII:C levels for pdVWF, and these were applied to the ER model. The impact of the dosing regimens (twice weekly [BIW] or once weekly [QW]) on the ER for rVWF and pdVWF were investigated based on population simulations. Hazard ratios (HRs) for the probability of bleeding were generated as a function of median VWF activity at steady state for patients with type 3 VWD.

Results: The RTTE model with a linear ER function linking the average levels of VWF:RCo or FVIII:C over 24 h before sBE onset was selected as the best model. A statistically significant ER relationship was observed (p<0.05) for the ER model with VWF:RCo, in which higher exposure to VWF:RCo was associated with a lower risk of sBE occurrence. The covariate effect of previous treatment (OD with a VWF or prophylaxis with pdVWF) was not statistically significant (p=0.6732). Simulations suggested that the HR per 10 IU/dL increment in the average exposure of VWF:RCo 24 h before an sBE was 0.673 (95% CI: 0.454-0.999). The HR per 20 IU/dL increment in the average exposure of VWF:RCo 24 h before an sBE was 0.453 (95% CI: 0.206-0.998). In addition, the predicted risk of a sBE for the 50 IU/kg QW regimen of rVWF and pdVWF was 30% and 43% higher, respectively, compared with the 50 IU/kg BIW regimen of rVWF (ie, reference regimen). The predicted risk of bleeding with the 50 IU/kg BIW regimen of pdVWF was 20% higher compared with the 50 IU/kg BIW regimen of rVWF. A trend was observed for the ER relationship based on FVIII:C (average levels of FVIII over 24 h before the sBE) suggesting a lower risk of sBEs with increased FVIII:C, which was however not statistically significant.

Conclusions: Analysis of exposure to VWF:RCo or FVIII:C vs sBE occurrence indicated a causal association between VWF:RCo and sBEs; higher VWF:RCo was associated with a lower sBE risk. This relationship was independent of the patients' previous treatment (OD with a VWF or prophylaxis with pdVWF). Once further supported with additional data, this ER model could be utilized for individualized dosing strategies to optimize patient outcomes with rVWF prophylaxis.