Pharmacokinetic and Pharmacodynamic (PKPD) Modeling to Determine the Andexanet Alfa Dose to Reverse the Anticoagulant Activity of Direct Fxa Inhibitors

Introduction

Andexanet alfa (AnXa) is a specific antidote under development for direct and indirect factor Xa (fXa) inhibitors.  AnXa is a recombinant, engineered version of human fXa that is catalytically inactive but retains high affinity for direct fXa inhibitors. AnXa binds fXa inhibitors, sequestering them in plasma, thereby reducing the free inhibitor concentration, reversing fXa inhibition, and normalizing thrombin generation.  The decreased unbound concentrations of the direct fXa inhibitors (apixaban, rivaroxaban and edoxaban) cause a portion of the extravascular fXa inhibitors to move into the vasculature.  Reversal of fXa inhibitor-induced anticoagulation requires a molar excess of AnXa relative to the total amount of fXa inhibitor in the blood – i.e., the initial fXa inhibitor plasma concentration plus the redistributed amount. We report a PKPD model that accounts for this extravascular-intravascular redistribution and allows determination of the appropriate AnXa dose to reverse the anti-fXa activity for each approved direct fXa inhibitor.

Methods

A model for rivaroxaban was constructed first by jointly analyzing AnXa concentrations, total and unbound rivaroxaban concentrations, and anti-fXa activity for 5 different doses of AnXa (each administered at the peak plasma level of 20 mg QD rivaroxaban at steady state). Model parameters were estimated by maximum likelihood using nonlinear regression (NONMEM, v. 7.2.0).  The model was used to simulate the potential level of reversal of anti-fXa activity after 2.5-40 mg/day rivaroxaban (QD and BID) for different bolus doses of AnXa and to predict maintenance of reversal by various follow-on infusion rates.  A similar model was developed to characterize the interaction between AnXa and apixaban.  Learnings from the rivaroxaban and apixaban models were used to develop a universal model to predict the level of reversal of anticoagulation for 60 mg edoxaban after a particular dose of AnXa. The model accurately predicted the degree of reversal of edoxaban anticoagulation.  The final model was used to construct a nomogram of AnXa doses that could be used for each fXa inhibitor at each approved dose based on timing of the last dose of the inhibitor.

Results

The final PKPD model was a three-compartment PK model for the fXa inhibitors, including one central and two tissue compartments.  The two tissue compartments included one that rapidly equilibrated with the central compartment and a second that equilibrated slowly.  The combined volume of the central and rapidly equilibrating compartments was similar to the volume of central compartment of the fXa inhibitor PK models in the absence of AnXa.

The AnXa PK was best described by a two-compartment PK model with an additional saturable binding component. The AnXa-fXa inhibitor complex moved to a “sequestration” compartment. Upon release from AnXa, the fXa inhibitor was able to return to the central compartment. The fXa inhibitor did not clear with AnXa.  The binding between AnXa and free fXa inhibitor was characterized by a reversible binding equilibrium. There was a direct PKPD linear relationship between free fXa inhibitor concentrations and anti-fXa activity.

Simulations were used to estimate the best dosing regimen to reverse anticoagulation for each direct fXa inhibitor.  A bolus dose of 800 mg of AnXa followed by an 8 mg/min infusion was sufficient to fully reverse the peak level of anti-fXa activity after the 20 mg QD dose of rivaroxaban.  This represents a >90% decrease in anti-fXa activity compared to pre-AnXa administration. Sustained reversal of peak anti-fXa levels for the 5 mg BID dose of apixaban required a 400 mg bolus AnXa dose with a follow-on infusion of 4 mg/min.  The peak anticoagulant activity of the 60 mg dose of edoxaban could be reversed by an 800 mg bolus dose of AnXa followed by an 8 mg/min infusion.

Conclusions

• A PKPD model was constructed that accurately predicted the AnXa dose necessary to reverse coagulation inhibition of each direct fXa inhibitor. The model incorporated the PK of AnXa, the anti-fXa activity, unbound and total levels of the fXa inhibitor, and the redistribution of each fXa inhibitor between the extravascular and intravascular compartments.
• Simulations were used to predict the AnXa dose necessary to reverse each approved dose of each direct fXa inhibitor and for different times after the last dose taken of the fXa inhibitor.