Chromogenic anti-FXa assay calibrated with low molecular weight heparin in patients treated with rivaroxaban and apixaban: possibilities and limitations

Introduction Clinical application of rivaroxaban and apixaban does not require therapeutic monitoring. Commercial anti-activated factor X (anti-FXa) inhibition methods for all anti-FXa drugs are based on the same principle, so there are attempts to evaluate potential clinical application of heparin-calibrated anti-FXa assay as an alternative method for direct FXa inhibitors. We aimed to evaluate relationship between anti-FXa methods calibrated with low molecular weight heparin (LMWH) and with drug specific calibrators, and to determine whether commercial LMWH anti-FXa assay can be used to exclude the presence of clinically relevant concentrations of rivaroxaban and apixaban. Materials and methods Low molecular weight heparin calibrated reagent (Siemens Healthineers, Marburg, Germany) was used for anti-FXa activity measurement. Innovance heparin (Siemens Healthineers, Marburg, Germany) calibrated with rivaroxaban and apixaban calibrators (Hyphen BioMed, Neuville-sur-Oise, France) was used for quantitative determination of FXa inhibitors. Results Analysis showed good agreement between LMWH calibrated and rivaroxaban calibrated activity (κ = 0.76) and very good agreement with apixaban calibrated anti-Xa activity (κ = 0.82), respectively. Low molecular weight heparin anti-FXa activity cut-off values of 0.05 IU/mL and 0.1 IU/mL are suitable for excluding the presence of clinically relevant concentrations (< 30 ng/mL) of rivaroxaban and apixaban, respectively. Concentrations above 300 ng/mL exceeded upper measurement range for LMWH anti-FXa assay and cannot be determined by this method. Conclusion Low molecular weight heparin anti-FXa assay can be used in emergency clinical conditions for ruling out the presence of clinically relevant concentrations of rivaroxaban and apixaban. However, use of LMWH anti-FXa assay is not appropriate for their quantitative determination as an interchangeable method.


Introduction
Direct oral anticoagulants (DOACs) have been in creasingly used for the prevention and treatment of thromboembolic diseases in recent years. Com pared to vitamin K antagonists (VKAs), clinical appli cation of DOACs does not require routine coagu lation monitoring. However, according to the pre sent expert opinions, there are special clinical situ ations in which laboratory measurement of DOACs in plasma should be performed, including bleed ing or thromboembolic events (acute stroke), emergency surgical or invasive procedures, ex tremes of body weight, renal and/or liver failure resulting with reduced drug elimination and sus pected noncompliance or overdose (14). There are the two main goals of quantitative determina tion of DOACs in abovementioned clinical situa tions. First, to assess the degree of anticoagulation in certain clinical situations, and second to exclude clinically relevant drug concentrations in circula tion.
Introduction of DOACs into clinical practice has faced haemostasis laboratories with new challeng es. On one hand, treatment of patients with DOACs has a significant impact on the results of screening coagulation tests, i.e., prothrombin time (PT) and activated partial thromboplastin time (APTT). Knowing the impact of DOACs on the results of screening coagulation tests is a precondition for the correct interpretation of these assays. Howev er, PT and APTT are not appropriate for direct FXa inhibitors, neither for quantifying drug concentra tion and reliable assessment of their anticoagulant effect, nor to exclude the presence of clinically rel evant drug concentrations in the circulation due to the high differences in sensitivities of individual commercial PT and APTT reagents (59). Therefore, the second major challenge for haemostasis labo ratories refers to implementing methods that al low quantitative determination of DOACs in cer tain clinical situations in order to help clinical deci sion making (1,9,1012). Furthermore, the need for quantitative methods for DOACs measurement is strongly supported by increasingly published fre quent cases of treated patients in which extensive or fatal bleeding is described (13).
Therefore, in parallel with the introduction of DOACs into clinical practice, research has also been focused on the development of specific co agulation methods for quantitative determination of these drugs. In order to achieve this purpose, chromogenic assays with drug specific calibrator materials were recently introduced as promising quantitative methods for particular DOAC drug (1417). The implementation of specific methods for quantifying DOAC concentrations in haemo stasis laboratories has significantly increased re cently. On the other hand, the same chromogenic antiactivated factor X (antiFXa) principle method for both unfractioned heparin (UFH) and low mo lecular weight heparin (LMWH) monitoring is widely used for years. Commercial assays devel oped for all antiFXa inhibitors, including heparins and related drugs, are based on the inhibitory ef fect of the drug present in patient´s plasma on ac tivated factor Xa (FXa) in the test system. However, assays intended to measure particular direct or in direct antiFXa drug need a calibration curve ob tained with a drug specific calibrators. On the oth er hand, it has also been recently shown that as says, even calibrated with drug specific calibrators, are not fully specific, meaning that assay calibrat ed with specific calibrators for one drug can also measure other antiFXa drugs (18). These observa tions resulted with first attempts for potential use of heparin calibrated antiFXa assay as an alterna tive method for direct FXa inhibitors, such as rivar oxaban and apixaban (1923).
Thus, we have hypothesized that LMWH calibrated antiFXa assay could be used in emergency cases for detection of rivaroxaban and apixaban pres ence when specific method is not available. The objective of this study was to assess the relation ship between LMWH calibrated antiFXa chromo genic assay and the same chromogenic method for rivaroxaban and apixaban calibrated with drug specific calibrators. Our particular interest was to evaluate whether commercial method for measur ing LMWH antiFXa activity can be used to exclude the presence of clinically relevant concentrations of rivaroxaban and apixaban in circulation.

Materials
The study was performed as a part of the research project IP2016068208 entitled "New oral anticoagulants: relationship between drug concentration and anticoagulant effect", funded by the Croatian Science Foundation. The institutional ethic com mittee has approved our study protocol as a part of the mentioned research project. All patients signed written informed consent according to the ethical guidelines following the Declaration of Hel sinki.
Samples from patients taking rivaroxaban and apixaban were collected from July to December 2018 at the Department of Neurology and Depart ment of Cardiovascular Diseases, Sestre Milosrdnice University Hospital Center. A total of 61 samples from patients taking rivaroxaban (31 peak and 30 trough) and a total of 53 (30 peak and 23 trough) samples from patients taking apixaban were used in the study. Blood samples were taken from the same patients and on the same day to obtain both, trough (immediately prior the next drug dose) and peak (two hours after drug administration) concen trations of rivaroxaban and apixaban in plasma. All patients were treated with standard and equal drug doses for nonvalvular atrial fibrillation (NVAF) clini cal indication (i.e. rivaroxaban 20 mg once daily and apixaban 5 mg twice daily).
Venous blood samples for determinations of rivar oxaban and apixaban concentrations were collect ed in Vacuette tubes (Greiner BioOne, Kremsmün ster, Austria) containing 3.2%trisodium citrate (volume 3.5 mL). All samples were centrifuged at room temperature for 10 minutes at 1800xg to ob tain platelet poor plasma, aliquoted into labelled tubes and stored at 20°C until analysis. Samples for the conducted study have been chosen in or der to cover as much as possible the whole meas urement range (up to 500 ng/mL) for both rivar oxaban and apixaban concentrations.

Methods
All coagulation assays were performed on Behring Coagulation System XP (BCSXP) analyser (Siemens Healthineers, Marburg, Germany). Low molecular weight heparin antiFXa activity was determined in all samples by chromogenic method using orig inal manufacturers´ reagent kit (Berichrom hepa rin, Siemens Healthineers, Marburg, Germany) and LMWH calibrator (Siemens Healthineers, Marburg, Germany). Results were expressed in antiFXa hep arin equivalent international units (IU/mL). The concentrations of rivaroxaban and apixaban were measured using specific chromogenic antiFXa as say (Innovance heparin, Siemens Healthineers, Marburg, Germany), calibrated with specific cali brators for rivaroxaban and apixaban (Hyphen Bi oMed, NeuvillesurOise, France). Concentrations of rivaroxaban and apixaban were expressed in ng/mL.

Statistical analysis
Data distribution was tested by Komolgorov Smirnov test. Since subgroups with peak and trough concentrations of each drug had 31 sam ples or less, nonparametric statistics was used. Agreement between LMWH calibrated and drug specific calibrated antiFXa activities was tested using kappa statistics. To test the comparability between LMWH calibrated and drug specific cali brated antiFXa activities, results of DOAC concen trations were divided in three categories. The first category included results lower than the lowest of trough concentrations, the second category in cluded results within expected range of concen trations and the third category included results higher than the highest of peak concentrations. Expected ranges for therapeutic drug concentra tions for patients with nonvalvular atrial fibrilla tion are adopted from published guidelines by Gosselin et al. (9). Low molecular weight heparin calibrated FXa activity was categorized in three categories according to measurement range (0.05 -1.26 IU/mL). Results of DOACs concentrations and LMWH antiFXa activity are presented as me dian with 95% confidence interval (95% CI) and in terquartile range (IQR). Differences between peak and trough concentrations of DOACs and LMWH antiFXa activities were tested using the nonpara metric MannWhitney test.
Receiveroperating characteristics (ROC) analysis was done to determine cutoff values of LMWH calibrated antiFXa activity which corresponds to rivaroxaban and apixaban values < 30 ng/mL and < 50 ng/mL. Those cutoff values were used as suggested for the treatment of patients with ex cessive bleeding and perioperative management by Levy et al. (24). P value < 0.05 was considered statistically significant. Statistical analysis was per formed using MedCalc for Windows, version 19.0.3 (MedCalc Software, Ostend, Belgium). Correlation scatter analysis was performed using Microsoft Ex cel version 2010 (Microsoft Corporation, Redmond Washington).

Results
The rivaroxaban concentrations obtained by chro mogenic antiFXa method with drug specific cali brators ranged from 62 to 433 ng/mL for peak and from 4 to 83 ng/mL for trough concentrations. The apixaban peak and trough concentrations ranged from 73 to 415 and from 13 to 98 ng/mL, respec tively.
Results of rivaroxaban and apixaban peak and trough concentrations as well as LMWH antiFXa activites are presented in Table 1. Differences be tween rivaroxaban and apixaban peak concentra tions were not statistically significant (P = 0.745). On the contrary, trough concentrations of apixa ban were significantly higher than rivaroxaban trough concentrations (P < 0.001). Considering ap propriate LMWH antiFXa activities for both peak and trough concentrations, the results demon strated significant difference between LMWH anti FXa activities for rivaroxaban and apixaban peak concentrations (P = 0.011), whereas the difference was not statistically significant for LMWH antiFXa activities measured in samples with trough con centrations of both drugs (P = 0.099).

Discussion
In this study, we evaluated relationship between two chromogenic antiFXa methods, one calibrat ed with LMWH and the other with drug specific calibrators for rivaroxaban and apixaban. The main two questions that we wanted to answer were whether LMWHcalibrated antiFXa activity assay can be used: 1) for excluding clinically relevant concentrations of direct antiFXa drugs in circula tion and 2) for quantifying rivaroxaban or apixa  Concentrations are presented as median (95% confidence interval) and interquartile range. LMWH low molecular weight heparin. P * between apixaban and rivaroxaban peak, and apixaban and rivaroxaban trough concentrations measured using drug specific calibrators. P † between apixaban and rivaroxaban peak and apixaban and rivaroxaban trough concentrations measured using LMWHcalibrated antiFXa activity assay. P < 0.05 was considered statistically significant.  AUC -area under curve. PPV positive predictive value. NPV negative predictive value. NA -not applicable. LMWH -low molecular weight heparin. * Arbitrary defined cutoff value related to perioperative reversal drug administration (24). † Arbitrary defined cutoff value related to reversal drug administration in bleeding patients (24). Table 3. Agreement between categories of apixaban determined with drug specific calibrators and with LMWHcalibrated antiFXa activity assay Table 4. Cutoff values of LMWH calibrated antiFXa activity for rivaroxaban and apixaban arbitrary defined concentrations ban concentrations in plasma as an alternative method to the chromogenic assays calibrated with specific drug.
Our study has shown good agreement between LMWH antiFXa activity and concentrations of ri varoxaban. Furthermore, a very good agreement between LMWH antiFXa activity and apixaban concentrations has been shown. A cutoff value of 0.05 IU/mL corresponded to the apixaban concen tration below 30 ng/mL and could be used in our laboratory for excluding the presence of apixaban with negative predictive value of 100%. In case of rivaroxaban, for concentrations < 30 ng/mL, cut off value at 0.1 IU/mL of LMWH antiFXa activity with negative predictive value of 100% has been obtained. We have chosen cutoff values with the highest sensitivity and negative predictive value to ensure reliable rule out method in our laborato ry. Thus, our results confirmed that if quantitative chromogenic methods calibrated with drugspe cific calibrators are not available, LMWHcalibrated antiFXa activity assay could be used as an alterna tive method for excluding the presence of clinical ly relevant concentration of antiFXa drugs with the negative predictive value of 100%. Namely, an important issue for possible clinical application of LMWH calibrated antiFXa activity assay in exclud ing clinically relevant concentration of antiFXa drugs lies in fact that some emergencies require administration of the reversal agent. The Scientific and Standardization Committee (SSC) of the Inter national Society on Thrombosis and Hemostasis (ISTH) recommended administration of the rever sal agent in the perioperative setting, if plasma concentration of direct FXa inhibitor is above 30 ng/mL, in order to ensure adequate haemostasis (24). In bleeding patients, antidote administration is supported at plasma concentrations above 50 ng/mL (19). Our study also confirmed results of Sa bor et al. who reported that antiFXa response is not the same at the concentrations of 30 and 50 ng/mL of rivaroxaban and apixaban (20).
For both rivaroxaban and apixaban, most of the values fell within the measurement range of the LMWH antiFXa activity. Further, 3/61 and 2/53 samples with high rivaroxaban and apixaban con centrations exceeded the upper level of measure ment range for LMWH calibrated antiFXa method. When considering the upper level of the measure ment range for different commercial LMWHanti FXa activity assays, it is important to note that it is determined by the activity of the highest calibra tor. In case of LMWH antiFXa activity assay used in this study, the upper measurement range of 1.26 IU/mL corresponds to the concentrations of both rivaroxaban and apixaban of approximately 300 ng/mL. Thus, concentrations above 300 ng/mL for both drugs cannot be quantified by this test.
Our results are in accordance with those obtained by Gosselin et al. who have also shown that al though LMWHcalibrated antiFXa activity assay can be used for ruling out the presence of direct antiFXa inhibitors in circulation, it is not suitable for quantitative determinations of antiFXa drugs in plasma (19).
Beside of limited range of quantification of the di rect antiFXa inhibitors using LMWHcalibrated an tiFXa assays, several other issues complicate its use in the accurate assessment of antiFXa inhibi tors. First, the result for LMWH antiFXa assay ac tivity is expressed in international antiFXa units/ mL (antiFXa IU/mL), whereas recommended measurement unit for direct FXa inhibitors is ng/ mL. Further, there is an evidence of substantial variability between different commercial kits and calibrators for LMWHcalibrated antiFXa assays (19,21). This fact additionally complicates compari sons between drug concentrations and antiFXa activity reporting. Both lower and upper limits of quantification were shown to be variable and sig nificantly limited by the calibrator measurement range. Based on these results the authors conclud ed that "on therapy" concentrations of rivaroxa ban cannot be quantitated using UFH, LMWH or hybridcalibrated antiFXa activity assays unless significant modifications are made to extend the current reportable range of the assay. Thus, it seems that results obtained with different com mercial LMWHcalibrated antiFXa assays cannot be extrapolated between different assays. Addi tionally, the authors cautioned that a LMWHcali brated antiFXa activity might not represent the same degree of anticoagulation for apixaban as it does for LMWH (19).
In the light of finding possible solutions to the problems related to need for different drug specif ic calibrators, controls and measurement units (IU/ mL and ng/mL) used in antiFXa assays intended for different antiFXa drugs, a group of authors have recently proposed a new concept that could contribute in solving this problem. These authors suggested a concept for a single antiFXa based laboratory assay for all drugs that directly or indi rectly inhibit FXa. The initiators of the idea of so called "DaXa inhibition assay" suggest a new test that would report inhibitory activity rather than drug concentration or IU/mL (18). The authors as sume that the strategy of universal DaXa inhibi tion assay would eliminate the need for drug spe cific calibrations for antiFXa drugs, heparins and heparin derived drugs (18). However, with assays available at this moment related to the measure ment of antiFXa inhibitors, and based on the re sults of our own study, our opinion is that LMWH calibrated antiFXa activity assay could be applied as an informative method for the purpose of ex cluding or confirming clinically relevant concen trations of antiFXa drugs in circulation. However, as this method is associated with a limited meas urement range, it could not be recommended for quantifying direct antiFXa inhibitors. Instead, a chromogenic antiFXa assay calibrated with drug specific calibrator materials should be used in clini cal situations intended for quantitative deter minations of antiFXa drugs.
The strength of this study relies on the fact that the results are obtained in plasma of patients treated with both direct antiFXa inhibitors, rivar oxaban and apixaban, unlike several previous studies that reported results of drug concentra tions in plasma samples where the certain concen tration of DOACs was added in vitro. Those results are significantly different from the values in the plasma of patients treated with DOACs (25,26). The benefit of our study relies in fact that wide range of concentrations of both direct antiFXa inhibitors were evaluated, thus allowing reliable conclusions based on the study results and related to the is sues that were in our focus.
The limitation of our study is that it was restricted to NVAF as the only clinical indication included.
However, this clinical condition still represents the most common indication for introduction of DOACs in patient management, thus being the real representative patient population for the pur pose of this research. Furthermore, a possible limi tation could be the fact that we compared only one LMWHantiFXa assay with one drugcalibrat ed antiFXa assay for rivaroxaban and apixaban available in the market. However, our intention was to compare the two methods that are in use at our laboratory in order to apply results in manage ment of patients treated at our institution.
In conclusion, the findings of this study will im prove the understanding in terms of both possibil ities and limitations of LMWHcalibrated chromog enic antiFXa assays in patients treated with direct antiFXa inhibitors. Our results strongly suggest that LMWH calibrated antiFXa assay has a poten tial as an alternative first line method for excluding the presence of significant levels of rivaroxaban and apixaban if laboratory has no available specif ic chromogenic antiFXa assay calibrated with par ticular drug. However, in case of positive result suggesting the presence of an oral antiFXa inhibi tor in plasma, specific chromogenic assay for par ticular antiFXa drug should be performed for quantitative determination of these drugs. The use of LMWH calibrated antiFXa assay to quantify ri varoxaban and apixaban concentrations could not be recommended in routine clinical practice as the only method for quantification of antiFXa medi cations. It is of crucial importance that both labo ratory experts and clinicians who treat patients substantially understand the opportunities and limitations of heparincalibrated antiFXa assays in patients treated with antiFXa inhibitors.