Measuring residual anti‐Xa activity of direct factor Xa inhibitors after reversal with andexanet alfa

Abstract Introduction Andexanet alfa (AnXa) was developed for anticoagulant effect reversal of direct factor Xa inhibitors (DXaI) (apixaban, rivaroxaban, edoxaban) in emergency situations. Regular anti‐Xa assays are not suitable to evaluate anti‐Xa activity after AnXa administration because of the high sample dilution resulting in the AnXa‐DXaI dissociation which gives inaccurately high DXaI measured concentrations. This study aimed at developing dedicated STA‐Liquid anti‐Xa test set‐ups for accurately measuring DXaI after reversal with AnXa. Methods Modified anti‐Xa test set‐ups, with reduced sample dilution, were developed to overcome regular assays limitations and to improve measured accuracy with results comparable to Portola microplate reference method used in clinical studies. Both regular and optimized assays were used to measure DXaI concentration in AnXa‐containing samples. Quality controls, normal pooled plasma spiked with five DXaI and three AnXa concentrations, samples from DXaI‐treated patients spiked with AnXa and ex vivo healthy volunteers having received both DXaI and AnXa were used. Results The lower limit of quantitation of optimized anti‐Xa assays was <10 ng/mL with CVs ≤10%. DXaI samples containing 300 ng/mL and 1 µmol/L AnXa resulted in DXaI residual concentrations of 29‐72 ng/mL depending on the DXaI (76%‐90% reversal), compared to 20‐28 ng/mL with reference method (92%‐94% reversal) and 135‐165 ng/mL with regular assays (about 50% reversal). Conclusion Modified test set‐ups are automated alternative to reference method with improved precision and reproducibility. They can be run in all laboratories where regular anti‐Xa assays are performed using commercially available reagents.


| INTRODUC TI ON
Oral direct inhibitors of factor Xa (DXaI such as apixaban, rivaroxaban, edoxaban) are increasingly used in several clinical indications. DXaI are effective therapeutic agents for the prevention and treatment of venous thromboembolism as well as stroke prevention in nonvalvular atrial fibrillation. 1,2 DXaI show more predictable pharmacokinetic and pharmacodynamic profiles than vitamin K antagonists. However, major bleeding events have been reported for patients taking DXaI.
Depending on the drug, the therapeutic indication, and the dosage, the rate of major bleeding in patients can reach 3%-5% patient-year. [3][4][5][6] Because of a large increase in prescription of DXaI, more patients require urgent reversal. A DXaI anticoagulant reversal agent can be useful for the management of serious bleeding or before urgent procedures. 7 Andexanet alfa (AnXa) is a first-in-class recombinant, modified human factor Xa (FXa) molecule developed to reverse anticoagulation in patients taking DXaI or indirect FXa inhibitors. [8][9][10] AnXa is the only specific reversal agent approved for anticoagulation reversal in patients presenting with a life-threatening bleeding episode while treated with rivaroxaban or apixaban. 8,9,[11][12][13] Dedicated anti-Xa chromogenic assays using specific calibrators and controls are used to measure direct and indirect FXa inhibitor levels in plasma. In emergency situations, such as uncontrolled bleeding, DXaI testing is useful to assess the presence of the drug and measure the plasma concentration. 14 As an example, according to the Subcommittee on Control of Anticoagulation of the International Society on Thrombosis and Hemostasis (ISTH), reversal procedure should be considered in patients with serious and life-threatening bleeding if the DXaI concentration measured before AnXa administration exceeds 50 ng/mL. 7 However, the regular anti-Xa assays do not allow an accurate measurement of the residual anti-Xa activity after reversal with AnXa because of the high sample dilution, which can result in erroneously elevated anti-FXa levels. 15 In contrast, the microplate anti-Xa chromogenic assays with specific test set-up using minimal plasma sample dilution can be used to measure the residual anti-Xa activity after reversal. The optimized microplate anti-Xa assay has been used in AnXa clinical studies with inhibitorspecific calibrators. Results are expressed in ng/mL for DXaI. 13 The objectives of our work were to develop automated suggested test set-ups with commercial calibrators and reagents for the measurement of residual DXaI concentration (apixaban, rivaroxaban, and edoxaban) after AnXa administration and then to evaluate the performances of these modified anti-Xa test set-ups in comparison with regular anti-Xa test set-ups and the microplate reference method used by Portola during drug development using patient samples.

| Material
Apixaban, rivaroxaban and edoxaban were provided by the respective manufacturer Bristol-Myers Squibb, Bayer, and Daiichi-Sankyo. Leftover frozen citrated plasma samples were obtained from patients treated with apixaban (10 patients), rivaroxaban (10 patients), or edoxaban (7 patients). Patients did not oppose the use of their plasma for research. Frozen ex vivo samples from healthy volunteers (three treated with apixaban and two treated with rivaroxaban followed by AnXa administration) were provided by Portola. 13,16 Anti-Xa reagents from the Coamatic Heparin kit (Chromogenix) were used for the reference microplate method with inhibitor-specific calibrators. STA-Liquid anti-Xa were used for automated assays with STA-Apixaban, STA-Rivaroxaban, and STA-Edoxaban Calibrators STA-Calibrator-0 is a normal plasma pool free of AnXa and DXaIs. Frozen samples were thawed in a water bath at 37°C for 5 minutes before use. All tests were performed within 4 hours after thawing.

| Microplate reference method
The microplate reference anti-Xa method was performed as described previously. 13 Briefly, the assay was adapted from the commercial Coamatic Heparin kit with modifications to reduce the effect of sample dilution. The reaction mixture consisted of plasma (75 µl), bovine Factor Xa (25 µl, 1x stock solution), S2732 substrate (25 µl, 2x stock solution, supplemented with 25 µl assay buffer). Following incubation at room temperature for 5 minutes, the reaction was stopped by adding quenching buffer (50 µl). Microplate reference method performance study was done in triplicate with five aliquots of samples spiked with 500 ng/mL of apixaban or rivaroxaban and 2.3 µmol/L AnXa. Compared to the regular anti-Xa assays, the modified test setups used neat plasma samples and less reagents. In order to observe the sample dilution effect on DXaI residual activity, 15 a series of conditions with an overall sample dilution from 1:2.6 to 1:12 was automatically realized by the STA-R Evolution with rivaroxaban-spiked samples using STA-Liquid anti-Xa reagent.

| Regular and modified test set-ups
The remaining experiments were conducted using the 1:2.6 sample dilution, which was the lowest working dilution that could be achieved when using liquid reagents. Plasma samples (125 µl) were incubated with STA-Liquid anti-Xa substrate (100 µl) for 4 minutes at 37°C. The reaction was then triggered with prewarmed STA-Liquid anti-Xa Factor Xa (100 µl) at 37°C and Optical Density variation over time (OD/minutes) was measured at 405 nm between 1.33 and 2.50 minutes for rivaroxaban, 0.83 and 1.67 minutes for apixaban, and 0.50 and 3.00 minutes for edoxaban.
Taking into account lower DXaI concentration after reversal, dedicated quality control plasmas were prepared. Either STA-Calibrator-0, Pool Norm, or frozen NPP was used as diluents to dilute STA-Calibrator-1 for apixaban or rivaroxaban and STA-Calibrator-2 for edoxaban and thus prepare extemporaneously ≈ 30 and ≈ 60 ng/mL QC plasmas. Similarly, assay calibrations had to be adapted. STA-Calibrator-2 for apixaban or rivaroxaban and STA-Calibrator-3 for edoxaban were automatically diluted by the instrument. DXaI concentrations in the test samples were interpolated by the instrument from the respective 2nd-order polynomial calibration curves from 0 to 70 ng/mL depending on the initial calibrator titer.
For concentrations between 70 and 130 ng/mL, residual DXaI concentrations were extrapolated from the raw data (OD/minutes) and the respective assay calibration curves allowing a measuring range from 10 (see LoQ determination below) to 130 ng/mL.

| Results and statistical analysis
Reagents lots, replicates, instruments, and samples used in this work are detailed in Table 1

| Sample dilution effect
With the modified test set-up, results obtained with rivaroxabanspiked samples varied according to the dilution as shown in Figure 1.
The measured residual DXaI concentration decreases with decreasing dilutions. A maximum reversal is observed with the 1:2.6 sample dilution.

| Technical performances
Based on the standard deviation analysis, the highest LoD was 6 ng/mL obtained with the modified rivaroxaban test set-up. With a maximum bias of 10 ng/mL compared to STA-Liquid anti-Xa regular assays, the LoQ was estimated to be less than 10 ng/mL for all DXaI. Modified apixaban test set-up bias was found above 10 ng/mL for concentrations higher than 30 ng/mL.
With regular assays, same samples gave a percentage of reversal below 53%.

| Ex vivo samples
Modified anti-Xa set-ups showed dramatic improvement in measuring AnXa reversal activity, but with a mean 18% and median 23% underestimation of percent reversal (ie, higher anti-Xa activity) compared to the microplate reference method (Figure 2). We choose two concentrations of AnXa (1.0 and 2.3 µmol/L) in the spiking experiment to allow reversal of concentrations up to 500 ng/ mL for the three DXaI. However, an incomplete reversal was observed in the excess amount of AnXa (2.3 µmol/L) using the regular anti-Xa assays ( Table 2).

| D ISCUSS I ON
Different test set-ups were evaluated in rivaroxaban-spiked samples in the presence of AnXa. We observed that dilution of the plasma sample had a major impact on results with an improved reversal measurement by decreasing the overall sample dilution in the F I G U R E 1 Sample dilution effect on residual anti-Xa activity based on ng/mL. The results are compared following different dilutions with rivaroxaban (50 to 500 ng/mL)-spiked plasma samples and AnXa concentration of 2.3 µmol/L. The residual concentrations are expressed in ng/mL for each bar which corresponds to a sample and its dilution Residual rivaroxaban concentraƟon (ng/mL) Sample diluƟon effect on residual AnƟ-Xa acƟvity  16,17 Nevertheless, this allowed the first characterization of the performance of the modified anti-Xa methods using commercial reagents on automated analyzers. F I G U R E 2 Ex vivo determination of residual DXaI concentration in five healthy volunteers samples treated by apixaban or rivaroxaban and after AnXa infusion. For apixaban-treated subjects, the samples were take from the 420 mg bolus cohort with no follow-on infusion 16 where rivaroxaban-treated subjects were take from the 720 mg bolus +4 mg/min infusion cohort. 13 Results are displayed for plasma samples drawn immediately after AnXa infusion (1 sample) and 10 (1 sample