Direct oral anticoagulants-Remove versus Taipan snake venom time for detection of a lupus anticoagulant in patients taking oral direct factor Xa inhibitors

1 INTRODUCTION

The use of direct oral anticoagulants (DOACs) has proliferated because of the convenience of fixed dosing and safety profile.¹ There is a need to detect a subset of high-risk patients with thrombosis and antiphospholipid syndrome (APS) who may benefit from a vitamin K antagonist rather than DOACs.^{2, 3} Because of interference on haemostasis assays, testing for lupus anticoagulant (LA) for patients taking DOACs is not generally advised; however, to do so would aid diagnostic accuracy for APS while maintaining the convenience of uninterrupted anticoagulation.^4-6 Two assays of differing analytical principles are required to maximize LA detection rates, which are commonly an LA-sensitive activated partial thromboplastin time (APTT) and the dilute Russell viper venom time (dRVVT).⁷ Several strategies are available to mitigate the effects of DOAC on hemostasis tests and thus potentially detect an LA; temporarily withhold the DOAC to allow “off-therapy testing,” plasma DOAC adsorption methods using activated charcoal (e.g., DOAC Remove, 5-diagnostics, Quadratech, Switzerland, and DOAC Stop, Haematex Research), plasma filter technology, or use of the direct factor Xa inhibitor (DFXaI)-insensitive assays Taipan Snake Venom Time (TSVT) screening test and ecarin time (ET) confirmatory test.^8-10 In 2020, the British Society for Haematology published an addendum to its APS guidance that recommended APTT- or dRVVT-based tests “should not be used to detect LA on samples from patients taking DOAC when there is a detectable drug level.”¹¹ In view of the controversy regarding the optimal laboratory methodology of investigating patients on DOACs, we investigated a strategy of DOAC-Remove treatment before APTT and dRVVT (our first-line standard LA tests¹²) compared with using TSVT/ET on plasma samples from patients taking DFXaI DOACs, where LA investigation had been requested by the clinical team.

2 METHODS

Sample preparation and storage was conducted in accordance with International Society on Thrombosis and Haemostasis (ISTH) guidance.⁶ All LA testing was performed on an ACL TOP 750 (Werfen, Bedford, MA, USA). In this study, only DFXaI DOACs were investigated because the direct prothrombin activation of Taipan and ecarin venoms bypasses FXa inhibition but not thrombin inhibition, rendering dabigatran an interferent in TSVT and ET.¹³

3 RESULTS

4 DISCUSSION

The present study sought to compare two analytical strategies for circumventing DFXaI interference in LA assays and has two important findings. The first is that DOAC Remove has complex effects on plasma in an in vitro plasma spiking model using DFXaIs; the second finding is that there is poor agreement between the TSVT/ET and DOAC Remove strategies for detecting an LA in patient samples. There is significant potential for false-positive or false-negative results in clotting assays used to detect a LA in patients on DFXaIs.^{6, 7}

Spiking LA-negative plasma with DFXaI revealed complex and differential effects for the dRVVT, APTT, and TSVT/ET assays for rivaroxaban, apixaban, and edoxaban (Table 1). Although use of DOAC Remove was generally able to distinguish true- versus false-positive LA, not all assays always returned to the reference range in the LA-negative spiked samples treated with DOAC Remove (e.g., dRVVT screen ratio with edoxaban; TSVT ratio with rivaroxaban, apixaban and edoxaban; TSVT/ET test ratio with rivaroxaban; APTT test ratio for rivaroxaban; described in Results). Incomplete DOAC removal by charcoal adsorbents has been previously described, mainly for high DFXaI levels, and is a limitation of this strategy unless DFXaI assays can be performed to evidence complete removal.^15-17 LA-positive samples remained positive after spiking with all DFXaI, and dRVVT screen ratios were further prolonged, which would be expected. The most striking results were seen with TSVT/ET testing, in which TSVT ratios became more elevated on spiking with rivaroxaban and apixaban, which has not been reported in other studies, and were further prolonged after DOAC Remove treatment with all three DFXaIs. Despite this, spiked and postdrug removal TSVT/ET test ratios were unaffected compared with neat results with rivaroxaban and edoxaban, although they were significantly different with apixaban. There is no immediately obvious explanation for this phenomenon, and further studies would be needed to ascertain any direct interactions between Taipan venom, DFXaIs, and DOAC Remove in the presence of LA. Overall, our data highlight differences in effects (and confounding) of DOAC Remove with different LA assays and DFXaI highlighting the need for further research in this area.

A total of 284 residual samples from consecutive DFXaI anticoagulated patients in routine clinical care were retested by dRVVT and APTT test ratios after anticoagulant removal, and by TSVT/ET test ratio, to assess for effects on routine diagnostic practice where many patients are being tested for LA for the first time. The lower percentages of LA-positive results before DOAC Remove treatment in patients on apixaban and edoxaban compared with rivaroxaban reflect the spiking experiments on LA-negative plasmas. As anticipated from previous reports and mirrored in the spiking experiments, the DOAC Remove procedure markedly reduced the numbers of LA-positive results with dRVVT and APTT, particularly for rivaroxaban. Crucially, frequency of positivity was similar between dRVVT and APTT after DOAC Remove treatment and TSVT/ET test ratio before DOAC Remove treatment, suggesting a degree of diagnostic concordance and accuracy, although the accuracy could only be confirmed with subsequent testing at least 12 weeks later, ideally off anticoagulation. The rise in TSVT/ET test ratio positivity after DOAC Remove treatment reflects the spiking experiments and emphasizes that samples treated for anticoagulant removal with an adsorbent before dRVVT and APTT analysis should not be used for TSVT/ET testing, which should be performed on untreated plasma. Overall, the percentage agreement for TSVT/ET test ratio (on neat plasma) versus LA after DOAC Remove-treated plasma was 90% (Cohen kappa 0.12; poor agreement) in this study of 284 DFXaI samples of which the majority were rivaroxaban.

The TSVT has been previously investigated as a method to detect an LA in patients on DFXaIs.^{10, 13, 18-21} The advantage of a TSVT/ET assay is that no preanalytical inactivation step is necessary for LA detection on DFXaI because prothrombin activation bypasses the effects of factor Xa inhibitors.¹³ The ISTH has previously issued guidance within this area and concluded that there was “no conclusive independent evidence reported on their diagnostic efficacy” and that studies had small numbers of patients mostly on rivaroxaban with challenges in kit standardization and availability likely to limit widespread utility and generalizability.⁶ This has recently been challenged in a multicenter, multiplatform publication by the ISTH SSC Subcommittee on Lupus Anticoagulant/Antiphospholipid Antibodies, which validated the TSVT/ET for patients taking DFXaI.¹³ What remains unclear is whether the discrepancy in LA detection between the techniques we have investigated are because of (i) complex effects of plasma manipulation by DOAC Remove; (ii) differential sensitivity of the TSVT/ET, APTT, and dRVVT for detecting LA (because of the heterogenous nature of LA antibodies, the sensitivity of the TSVT/ET for all LA was 72% and 78% in confirmed APS cases and 87% in triple-positive [i.e., LA, cardiolipin, and beta-2-glycoprotein I positive] APS cases in the recent ISTH SSC publication regarding this¹³); and (iii) confounding of the anticoagulant effects on assays or combinations thereof. In theory, the dRVVT/APTT could be more sensitive to detecting an LA compared with TSVT/ET because there are two assays; however, this was not borne out in the comparison of the two techniques (Table 3). In an ideal validation scenario, a large number of LA-negative samples off anticoagulation would be collected and then subsequent further sample collection would be undertaken after starting DOACs. This would be compared with a large number of people with a positive LA off anticoagulation who had a further sample collected after initiating DOAC therapy. These samples could then be subject to investigation using different LA detection methods. The prospect of such a validation seems remote because of the inherent difficulties in performing such a study. The Clinical and Laboratory Standards Institute advises comparison of methods that this is performed in at least 50 negative and 50 positive samples.²²

This study has limitations. It is a single-center study reflecting our local assays and cannot be generalized beyond our center, and dabigatran was not studied. We have not performed mixing studies as suggested in ISTH guidance because of a lack of available plasma; likewise, factor assays have not been assessed because the Dade Actin FS APTT has been assessed for each patient to ensure no prolongation has been noted.^{7, 23, 24} We have not correlated the assay results with clinical details of the patient samples (e.g., presence of cardiolipin/beta-2-glycoprotein I antibodies and history of obstetric morbidity/thrombosis) and we have no prior knowledge of their LA status off anticoagulation previously. This study did not aim to correlate the assays with patient phenotype but investigated the assay performance for DOAC Remove plus dRVVT/APTT versus TSVT/ET. We also performed spiking experiments with pooled samples (which may introduce confounding) and in vitro results may not relate to in vivo effects. A strength of the work is that we knew the exact DFXaI patients were taking, with the mean DFXaI concentration being 275, 220, and 179 ng/ml for rivaroxaban, apixaban, and edoxaban, respectively, for patient samples. These are within the expected “peak concentrations” on treatment and therefore our results represent testing where there is maximal DFXaI interference with assays.²⁵ Our study did not seek to address how DOAC Remove dealt with different DFXaI concentrations but rather with the evaluation of dRVVT/APTT after DOAC Remove versus the TSVT/ET on neat plasma. Additionally, we did not investigate other charcoal adsorbents (e.g., DOAC Stop), which may not be equivalent to DOAC Remove; a comparative study is required as different adsorbents may have adsorbents have variable capacity, specificity, and kinetics and should not be generalized. Nevertheless, we provide novel data on the comparability and performance of TSVT/ET versus DOAC Remove plus dRVVT/APTT in routine patient care albeit in a specialized hemostasis laboratory.

5 CONCLUSION

In conclusion, we found, first, that in a spiking experiment, DOAC Remove generally distinguished LA-negative versus LA-positive samples though there were some false-positive LA results, and second, LA detection by APTT/dRVVT after DOAC Remove versus testing neat plasma samples for TSVT/ET in patients on DFXaI showed inconsistent results between the two methods. A proposed algorithm could be the testing for an LA on neat plasma by the TSVT/ET (to avoid the confounding effects of charcoal inactivation) and then proceed to DOAC Remove only if the TSVT/ET is negative. This would avoid the initial plasma manipulation step; however, we should recognize that proceeding straight to charcoal absorption is a more straightforward approach.¹³ The controversial issue surrounding the optimal detection of an LA for patients taking DOACs remains and our data highlight discrepancies between methodologies; further studies are required in this area.

RELATIONSHIP DISCLOSURE

Danielle White has nothing to declare. Gary W. Moore reports consultancy fees from Technoclone. Martin Besser reports speakers fee from STAGO and advisory boards for Novartis, Cosmopharma, and Werfen. Stephen MacDonald reports support to attend educational meetings from Sysmex, Werfen, and Stago. Will Thomas reports speakers fees from Bayer, Pfizer, Alexion, and Portola and advisory board for Daiichi Sankyo.

AUTHOR CONTRIBUTIONS

D.W. and W.T. designed the study, wrote the manuscript, and performed data analysis. D.W. performed the laboratory work. M.B., G.W.M., and S.M. had intellectual input and approved the final draft of the manuscript.