Using an old test for new tricks: Measuring direct oral anti-Xa drug levels by conventional heparin-calibrated anti-Xa assay

To the Editor:

The much-anticipated arrival of andexanet alfa, a recombinant factor Xa decoy protein that rapidly reverses the anticoagulant effects of direct oral factor Xa inhibitors (FXai) like apixaban and rivaroxaban,¹ has many hospital systems wondering how to manage patients who may benefit from reversal but fall outside clear indications. For example, patients with delayed drug clearance due to renal failure may experience major bleeding or require emergent surgery. A recent perspective article in Blood provides an excellent overview of FXai testing and highlights both the limited availability and clinical need for drug-specific FXai testing in critical situations.² While development of rapid drug-specific assays is underway, methods to quantitate FXai concentrations quickly, especially given the high cost of antidote, are needed right now. Herein, we provide data suggesting that conventional heparin-calibrated anti-Xa tests, performed using standard coagulometers in many U.S. hospitals, may help guide appropriate and cost-effective use of FXai antidote.

Routine coagulation tests like prothrombin time and activated partial thromboplastin time are of limited and variable utility in measuring FXai, which are best quantified using molecular or drug-specific chromogenic anti-Xa assays. As no FDA-approved drug-specific assays are commercially available, many US hospitals either do not offer drug-specific testing or have limited availability of a “Laboratory Developed Test.” For example, our institution offers FXai testing in a dedicated special coagulation laboratory from Monday through Friday during normal business hours. STAT specimens have a turnaround time of 4 hours, and those received after-hours may not be resulted for up to 3 days. However, the heparin-calibrated anti-Xa assay is available 24/7 in five out of six of our health-system core laboratories, which are not staffed to handle FXai calibration, with a turnaround time of <1 hour. Heparin-calibrated chromogenic anti-Xa assays have been used to screen for the presence or absence of FXai, yet have shown variable effectiveness in quantifying FXai levels and no robust study using patient samples has been reported.^3–5 In light of this, we sought to establish the strength of correlation between our heparin anti-Xa assay and both apixaban- and rivaroxaban-specific assays in our patient population by completing a full (n ≥ 120) validation study, and to determine if FXai drug levels may be predicted using the results of the heparin-calibrated assay.

Institutional guidelines were followed in accordance with the institutional review board determination. Queries of patients actively taking FXai identified the target number of specimens within an 8-week period, resulting in 121 apixaban and 120 rivaroxaban leftover/discarded random-collection-citrated plasma samples. Apixaban and rivaroxaban levels were determined using a chromogenic anti-Xa assay calibrated for apixaban (Biophen Apixaban Calibrator kit, Hyphen BioMed, France) and rivaroxaban (Biophen Rivaroxaban Calibrator kit), respectively, with the HemosIL liquid anti-Xa reagent and performed on the ACL-TOP instrument (Instrumentation Laboratory, Bedford, Massachusetts). Conventional heparin-calibrated anti-Xa testing was performed on all 241 specimens following the manufacturer’s instructions using a HemosIL liquid anti-Xa reagent and a heparin calibrator on the ACL-TOP instrument.

The apixaban level of 121 unique samples from 76 apixaban-treated patients ranged from <13 ng/mL to >600 ng/mL. All levels ≥268 ng/mL (n = 18) had corresponding heparin anti-Xa results over 2.0 IU/mL, the assay’s upper limit of detection, and were thus excluded from linear correlation analysis. Pearson’s correlation of the remaining 103 specimens using GraphPad Prism (Carlsbad, CA) demonstrated significant (P < 0.0001) and strong (R² = 0.9662) positive correlation between the apixaban-specific assay and the heparin-calibrated anti-Xa test (Figure 1A). As correlation is limited to association, we also performed linear regression analysis to determine whether results of the heparin-calibrated assay could be used to predict FXai levels with significance. Such analysis generated the best-fit equation Y = 119.2X + 11, with Y representing the apixaban level (ng/ml) predicted by a given heparin anti-Xa level (X) and P < 0.0001 (Figure 1A).

FIGURE 1.

Correlation and linear regression analysis for conventional heparin anti-Xa assay versus apixaban (A) or rivaroxaban (B) drug-specific assays. N = 103 clinical specimens from patients taking apixaban and n = 99 clinical specimens from patients taking rivaroxaban. R² values generated by Pearson’s correlation analysis demonstrate significance (P < 0.0001) in both A and B. Linear regression analysis generated the best-fit lines as described with significant deviation from zero (P < 0.0001) in both A and B

The 120 samples obtained from 82 rivaroxaban-treated patients revealed rivaroxaban levels ranging from <1 ng/mL to >500 ng/mL. All levels ≥240 ng/mL (n = 21) had corresponding heparin anti-Xa results over 2.0 IU/mL and were excluded from linear correlation analysis. Pearson’s correlation of the remaining 99 specimens likewise demonstrated significant (P < 0.0001) and strong (R² = 0.9755) positive correlation between the rivaroxaban level and heparin anti-Xa result (Figure 1B). Linear regression analysis generated the best-fit equation Y = 125.1X + 4.29, with Y representing the rivaroxaban level (ng/ml) predicted by a certain heparin anti-Xa level (X) and P < 0.0001 (Figure 1B).

Applying threshold concentrations proposed by the International Society on Thrombosis and Haemostasis (ISTH) to the above equations allows for determination of cut-off values. Specifically, ISTH recommends antidote consideration for patients with serious bleeding when FXai drug concentrations exceed 50 ng/mL and in patients requiring urgent intervention with high risk for bleeding at concentrations above 30 ng/mL.⁶ Using our data, this corresponds to heparin anti-Xa levels >0.33 IU/mL for apixaban or > 0.37 IU/mL for rivaroxaban in patients with serious bleeding, and >0.16 IU/mL for apixaban or >0.21 IU/mL for rivaroxaban in patients requiring intervention. This information will be included in the interpretive comment provided to practitioners by our laboratory, although strict cutoff values will not be recommended given the limitations described and importance for clinical correlation. Results for patients with FXai levels exceeding the linear range of heparin anti-Xa testing (>2.0 IU/mL) were not determined. Sample dilution may overcome this limitation; however, assays using drug-specific calibration may be warranted when exact values are necessary.

Our data suggest that conventional heparin anti-Xa testing may have a role in assessing FXai levels when drug-specific testing is not readily available and rapid turnaround time is required. We certainly do not advocate delaying treatment while awaiting anti-Xa results when patients meet clear indications for FXai reversal, nor do we mean to suggest that heparin-calibrated assays be used for nonurgent pharmacokinetic studies when drug-specific assays are more appropriate. In addition, our data are not necessarily generalizable given reagent and coagulometer variability, and each laboratory should establish their own cut-offs and ranges. Nevertheless, we hope our experience helps others investigate correlations between their testing platforms or other alternative methods of FXai testing to aid in optimal utilization of new and expensive FXai antidote.