1. Pharmacokinetic and Pharmacodynamic Modeling of Andexanet Alfa Dose to Reverse the Anticoagulant Activity of FXa Inhibitors in Patients With Acute Major Bleeding
Janet M. Leeds, PhD
Portola Pharmaceuticals, Inc., South San Francisco, CA
Co-authors:
Jaap W. Mandema,1 Genmin Lu,2 John T. Curnutte,2 Truman J. Milling,3 Mark Crowther,4 Stuart J. Connolly,4 Pamela B. Conley2
1Certara, Menlo Park, CA, USA; 2Portola Pharmaceuticals, Inc., South San Francisco, CA, USA; 3Seton Dell Medical School Stroke Institute, TX, USA; 4McMaster University, Hamilton, ON, Canada

2. Disclosures Presenter’s Disclosure
Dr. Leeds is an employee of Portola Pharmaceuticals, Inc.
Other Authors
Drs. Mandema, Crowther, and Connolly have been consultants for Portola Pharmaceuticals, Inc.
Other co-authors are employees of Portola Pharmaceuticals, Inc.
Unlabeled/Unapproved Uses Disclosure
Andexanet alfa is an investigational antidote for factor Xa inhibitors

3. Andexanet: Designed to Reverse Activity of Factor Xa (FXa) Inhibitors
Andexanet acts as a FXa decoy and retains high affinity for direct and indirect FXa inhibitors
It binds FXa inhibitors and counteracts their activity but is no longer capable of assembly into the prothrombinase complex
Andexanet
Native FXa
FXa Inhibitor
High affinity retained
Andexanet has been rationally designed to reverse the activity of FXa inhibitors by acting as a FXa decoy and sequestering the FXa inhibitor away from normal FXa and allowing restoration of normal coagulation
Andexanet is a recombinant engineered derivative of FXa, which is produced in CHO cells.
Two modifications were made to FXa to achieve this:
Have the active site Serine changed to Alanine, to eliminate the catalytic activity and prevent prothrombin cleavage.
Have the GLA (gamma-carboxyglutamic acid) domain removed to prevent competition with endogenous FXa assembly into the prothrombinase complex, eliminating an anticoagulant effect.
There is no known interaction with other coagulation factors except for Tissue Factor Pathway inhibitor (TFPI), which was expected, as the parent molecule, FXa, interacts with TFPI.
Andexanet retains high affinity for Antithrombin III-inhibitor complex and can reverse anticoagulant effects of enoxaparin and fondaparinux.
GLA domain removed to prevent anticoagulant effect
Binding site
A419
S419
GLA
Catalytic Domain
Activity eliminated to prevent thrombin generation S S
N terminal residues retained to reduce immunogenicity
Lu G et al. Nat Med 2013;19(4):446-51

4. Early Andexanet Model Development
Data Phase 2 studies in healthy volunteers were used to develop PK/PD model to predict the andexanet regimen to reverse anticoagulation by FXa inhibitors
FXa inhibitors dosed to steady state. Andexanet administered 3 hours after the last FXa dose
Andexanet, total and unbound fXa concentrations, anti-FXa activity used to build PK/PD model
Evaluation of PK/PD Model Predictions in Bleeding Patients
First Interim cut of data from the ongoing Phase 3b/4 study (ANNEXA-4) used to verify the PK/PD model
Limited information available for patients: anti-FXa activity, total and unbound FXa inhibitor from a small number of time points
Andexanet Clinical Development Programs (Phase 2 and 3 in healthy volunteers) to date have demonstrated significant reversal of PD markers of FXa inhibitors
An ongoing ANNEXA-4 will confirm clinical efficacy and safety in patients with acute major bleeding by assessing both PD markers (anti-FXa activity) and hemostatic efficacy
1Connolly S et al. New Engl J Med. 2016;375(12): Connolly S et al. New Engl J Med. 2016;375(25):

5. Andexanet Pharmacokinetics and Pharmacodynamics in Healthy Subjects
Dose proportional at doses of 90 mg and above
Repeat dose administration or continuous infusion can maintain andexanet levels and reversal of anti-FXa activity

6. Pharmacokinetics of Direct FXa Inhibitors
Equilibrium between:
Plasma
Interstitial and lymphatic fluids
Tissues
Interstitial and lymph fluids
Tissues and
body water
Plasma
AnXa also binds plasma proteins non specifically
Bound
Bound
Bound
Free
Free
Free
Source: Shargel L, Wu-Pong S, Yu ABC: Applied Biopharmaceutics & Pharmacokinetics, 6th Edition:
Copyright © The McGraw-Hill Companies, Inc. All right reserved.

7. Effect of Andexanet on DOAC Pharmacokinetics
Andexanet (AnXa) decreases unbound anticoagulant levels
Tissue
Blood
Tissue
AnXa
AnXa
Unbound apixaban
pharmacologically active
Total apixaban
Unbound FXa inhibitor
Bound FXa inhibitor
Volume of Distribution (Vd, (L))
Rivaroxaban
Apixaban
Edoxaban 50 21
107
FDA-approved product labels: Xarelto®; Eliquis®; Savaysa®

8. A PK/PD Model Was Developed to Describe Mean Trends of Concentrations and Anti-FXa Activity
Structural model shared by three FXa inhibitors
Andexanet PK: 2-compartment with a saturable binding site
FXa Inhibitor PK: 3-compartment with first order absorption
Binding between andexanet and rivaroxaban in central compartment, and a sequestration compartment for eliminated andexanet-rivaroxaban complex
Anti-FXa activity: proportional to free plasma rivaroxaban concentration

9. Andexanet Dose Modeling Provided Dose Recommendations to Reverse Anti-FXa Activity
Model simulations considered the following:
Concentration of FXa inhibitor
Volume of distribution of FXa inhibitor
Half-life of andexanet and FXa inhibitor
Clearance of andexanet and FXa inhibitor
Anti-FXa activity as a function of andexanet bolus dose ending 4 hours after rivaroxaban 20 mg QD
Andexanet dose-dependently reduced anti-FXa activity in the model. Complete reversal of anti-Fxa activity was achieved with a 800 mg bolus of andexanet.

10. Time Course of Andexanet Bolus Dose Predicted by Model to Reverse Anticoagulation
Time after last FXa inhibitor dose (h)
FXa Inhibitor Dose
Apixaban mg BID
Edoxaban mg QD
Rivaroxaban mg QD 3
690
881
753 4
620
736
682 5
560
589 6
510
528
503 8
440
398
377 10
380
312
296 12
350
253
249 14
320
210
225 16
300
180
206

11. PK/PD Model: Andexanet Dosing Recommendations
A single PK/PD model was developed that described the PK/PD interactions between andexanet and each approved direct FXa inhibitor (apixaban, rivaroxaban, and edoxaban)
Model simulations provided recommendations for andexanet dosing
Dose
Initial IV bolus
Follow-on IV infusion
Low dose
400 mg at 30 mg/min
4 mg/min for 120 minutes (480 mg)
High dose
800 mg at 30 mg/min
8 mg/min for 120 minutes (960 mg)
Data we are presenting changed dose at 7 hours. We should explain here that initially we reduced the dose at 7 hours but moved to 8.

12. Validation of the PK/PD Model With Data From the ANNEXA-4 Study
Compare the observed reversal of anti-FXa activity in bleeding patients (ANNEXA-4) and predicted reversal of anti-FXa activity based on the PK/PD model in healthy subjects

13. Bleeding and laboratory assessment
Methodology
In ANNEXA-4, an ongoing prospective, open-label study, bleeding anticoagulated patients received IV andexanet bolus (400 or 800 mg) followed by 120-min infusion (4 or 8 mg/min)
Anti-FXa activity was measured before andexanet administration (baseline), at the end of bolus (EOB), at the end of infusion (EOI), and at 4, 8, and 12 hours after infusion
The relationship between baseline anti-FXa activity and reversal in healthy subjects was derived from the PK/PD model and used to provide a predicted percent reversal for patients with acute major bleeding
Bleeding and laboratory assessment
The assay for assessing anti-FXa levels measures anticoagulant activity and is a well-established assay for FXa inhibitors.
Anti-FXa activity was measured using a modified LMWH chromagenic anti-FXa kit
Andexanet treatment
IV Bolus
2-hour
IV Infusion
After end of infusion
1 h
4 h
8 h
12 h
Day 1

14. Patients
From the first interim analysis of ANNEXA-4, plasma levels from 73 patients were available for model qualification
Drug n
FXa dose
Andexanet dose
Time since last dose
5 BID
2.5 BID
Other
400/4
800/8
7 h
7-18 h
Apixaban 39 19 14 6 10 29
Drug n
FXa dose
Andexanet dose
Time since last dose
20 QD
15 QD
Other
400/4
800/8
7 h
7-18 h
Rivaroxaban 34 22 8 4 30 29

15. Reversal of Anti-FXa Activity: Rivaroxaban
The mean observed percent reversal of anti-FXa activity for rivaroxaban was well predicted by the healthy subject PK/PD model
The point estimates fell within the 95% confidence intervals of predicted values
Time n
Predicted
% reversal
Observed
% reversal [95% CI]
Rivaroxaban
EOB 32
76.4
74.4 [58.3 to 90.4] EI 33
78.3
77.4 [67.6 to 87.2]
4 hours
31.8
32.0 [18.0 to 45.9]
8 hours
40.7
43.6 [35.0 to 52.2]
12 hours
51.0
56.9 [50.2 to 63.7]
EOB: End of Bolus; EOI: End of Infusion

16. Reversal of Anti-FXa Activity: Apixaban
The mean observed percent reversal of anti-FXa activity for apixaban was well predicted by the healthy subject PK/PD model through 4 hours
Time n
Predicted
% reversal
Observed
% reversal [95% CI]
Apixaban
EOB 34
84.1
83.9 [75.3 to 92.5] EI 36
81.4
84.2 [77.3 to 91.1]
4 hours 33
34.4
27.6 [10.2 to 45.0]
8 hours 32
40.2
30.6 [25.1 to 36.1]
12 hours
48.6
32.8 [26.1 to 39.5]
EOB: End of Bolus; EOI: End of Infusion

17. Conclusions
The PK/PD model for andexanet dosing based on data from healthy subjects has been qualified by clinical data from the ANNEXA-4 study in patients with acute major bleeding
The PK/PD model of the interaction of andexanet and FXa inhibitors derived in healthy subjects closely predicted the percent reversal of anti- FXa activity by andexanet in patients receiving apixaban or rivaroxaban who have acute major bleeding
The predictions were within the 95% confidence intervals for rivaroxaban at all time points and through 4 hours after the end of the infusion for apixaban
The PK/PD model was highly predictive of the reversal of anticoagulant activity irrespective of the type of bleed or the level of bleeding in patients

18. Acknowledgements BMS Global Portola Brenda Cinncione Ken Derr BMS KK
Michele Bronson
Janice Castillo
Iwona Bucior
Bill Lis
BMS Global
Brenda Cinncione
BMS KK
Takayo Ueno
Takafumi Ide
Pfizer
Won Byon