Direct oral anticoagulants (DOACs) are a popular and effective treatment for venous thromboembolism (VTE) and nonvalvular atrial fibrillation (1–3). These drugs have several advantages over traditional anticoagulants, such as a more rapid onset of action, as compared with warfarin, and oral dosing and longer half-life, as compared with heparin (3,4). DOAC prescriptions have grown rapidly, and routine monitoring of coagulation in the context of DOAC therapy has not historically been recommended. Adverse events have mounted in recent years as these drugs are prescribed to increasingly complex patient populations, and a lack of U.S. Food and Drug Administration (FDA)-approved diagnostic tests to evaluate patient coagulation status in the context of DOAC therapy may: 1) increase the complexity of the continued adoption of these drugs for complex patient populations (such as those with renal disease, obesity, and polypharmacy) and 2) complicate clinical decision making in emergency and critical care settings (5–15).
Conventional tests for coagulation, such as prothrombin time (PT), activated partial thromboplastin time (aPTT), and activated clotting time are not specific for Factor Xa inhibitor (FXa-I) detection; in addition, these tests lack the sensitivity needed to rule out FXa-I–induced anticoagulation in patients (16). Viscoelastic assays, such as thromboelastoraphy and rotational thromboelastometry, have mixed reports on the sensitivity to all DOACs and appear to be consistently insensitive at lower, although therapeutic, concentrations (17–19). Currently, mass spectrometry and anti-Xa chromogenic assays are the only tests reported to be consistently sensitive and specific to the presence of FXa-I, but these tests are mostly performed in specialized or central laboratories with a 30–120-minute turnaround time, limiting their utility in the emergency setting and in facilities that may not have 24/7 access to this equipment. These tests also require an accurate medical history for drug-specific calibration (20–23). To address these limitations, we developed a microfluidics-based assay, the inhibitor-II-X (i-II-X) test, to detect the presence of FXa-Is in patient samples. We evaluated the i-II-X test for the detection of both apixaban and rivaroxaban in patient blood and whether this test could help clinicians identify patients in which prolonged clotting times (PT/international normalized ratio [INR]) are secondary to the presence of FXa-I.
MATERIALS AND METHODS
Study Design and Sample Collection
We performed a single-center pilot study involving 91 adult patients at the Massachusetts General Hospital (MGH) (Boston, MA, USA) in order to evaluate this new assay approach in emergency department patient samples. All samples and patient information were collected and handled according to MGH and Massachusetts Institute of Technology (Cambridge, MA) Institutional Review Board committee approval (approval numbers: MGH No: 2014P002087; MIT No: 150100681R001); because discarded plasma samples were used, patient consent was not required. Patients admitted to the emergency department were screened and selected for medical histories indicating recent DOAC prescription (Supplementary Table 1, Supplemental Digital Content 1, https://links.lww.com/CCX/A66) and had a 3.2% sodium citrate blood tube drawn for clinician-ordered coagulation testing. On days when DOAC patient samples were collected, additional non–anticoagulated emergency department patient samples were also collected to serve as “negative controls.” Patients were not screened out for any condition or concurrent medication, with the exception of the recent use of other anticoagulants, such as heparin and warfarin. Platelet-poor plasma was collected from the same tube that clinician-ordered coagulation tests were performed, de-identified and stored at –80°C until analysis. Following i-II-X analysis, we retrospectively examined coagulation test results from patient medical records. Coagulation tests ordered in the emergency department included a combination of PT, INR, and aPTT and D-dimer. Due to the design of this study, PT/INR results were not available for every patient (i.e., the attending clinician did not order a PT/INR for 3/43 of the FXa-I patients and 5/48 of the control patients), and although some patients received aPTT or D-dimer, too few patients had these results to allow statistical analysis. It is important to note that no clinicians ordered an anti-Xa assay or mass spectrometry on any sample. The purpose of this experimental design was to compare the i-II-X test results with what the attending clinician ordered to evaluate the patient’s coagulation status. PT/INR was performed on the Destiny Max (Stago Diagostica, Asnieres, France) using the PT HTF reagent (Stago Diagostica). The rivaroxaban anti-Xa chromogenic assay was performed at the MGH Coagulation Laboratory using the STA-Liquid Anti-Xa reagent and Stago rivaroxaban calibrators on the STAR Max Analyzer (Stago, Parsippany, NJ). Apixaban calibrators were not approved for use at MGH at the time of this study; therefore, only rivaroxaban samples were evaluated using the anti-Xa method. Sample volume was insufficient to perform mass spectrometry.
Design and Fabrication of the Microfluidic Devices
The microfluidic devices were manufactured using standard microfabrication techniques. In brief, a single-layer photoresist design (SU-8; MicroChem, Newton, MA), with a 50-μm-thick layer was patterned on one silicon wafer via a photolithography masks and standard processing, according to the manufacturer’s protocols. The resulting patterned wafer was then used as a mold to produce polydimethylsiloxane (Thermo Fisher Scientific, Waltham, MA) devices, which were subsequently, irreversibly bonded to glass slides (1 in. × 3 in.; Thermo Fisher Scientific). The microfluidic design included four channels, each with their own inlet and outlet ports, and one common central imaging area (Supplementary Fig. 1,A and B, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). This configuration allowed for the simultaneous imaging and analysis of multiple conditions (Supplementary Fig. 1C, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). The chips were pretreated with a corona plasma gun (Elveflow, Paris, France) prior to sample loading to eliminate the need for fluid pumps.
Coagulation Curve Generation
The patient samples were run blinded. Coagulation curves were calculated by using our i-II-X microfluidic assay to detect the “time to clot” (TtC) in plasma. Briefly, after thawing plasma at 37°C, we added 20 mM calcium (Boston Bio Products, Boston, MA), 488-conjugated fibrinogen (Thermo Fisher Scientific), and various concentrations of “Agonist A” and “Agonist B” to detect and distinguish the presence of FXa-I and Factor IIa inhibitor (FIIa-I) (Supplementary Fig. 1D, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). Briefly, Agonist A tests for the presence of an inhibitor at or downstream of FXa and Agonist B tests for the presence of an inhibitor at or downstream of FIIa; taken in combination, the test results from Agonists A and B can detect the presence of the FXa-I and FIIa-I. Samples were loaded into the i-II-X microfluidic chip and a fully automated Nikon TiE microscope and NIS Elements Software (Nikon Instruments, Melville, NY) imaged the chip every 15 seconds for up to 10 minutes in order to document the TtC.
Clotting Time Score Generation
Coagulation curves were generated for each sample by plotting TtC for each agonist concentration using GraphPad (GraphPad Software, San Diego, CA) (Supplementary Fig. 1E, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). A predictive model for the detection of FXa-I from the clotting curve was then generated using the R Software using the multivariate logistic regression analysis (24). The resulting model was then used to assign a numerical clotting time score (CTS) to each patient. A cutoff score of 0.5 (functional coagulation level [FCL]/milliliter) was selected for the presence of factor inhibition. Briefly, if the CTS was greater than 0.5 FCL/mL, this indicated that there was inhibition at or downstream of the factor being tested; alternatively, if the CTS was less than or equal to 0.5 FCL/mL, this indicated that there was no inhibition of coagulation at or downstream of the factor being tested (Supplementary Fig. 1D, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). Clotting curves and CTSs were generated using commercially available calibrators for apixaban, dabigatran, rivaroxaban, and edoxaban and in-house generated calibrators for betrixaban made with lyophilized normal control plasma (HYPHEN BioMed, Aniara Diagnostica, LLC, West Chester, OH) (Portola Pharmaceuticals, South San Francisco, CA). Warfarin plasma was purchased from George King Bio-Medical (Overland Parks, KS). Edoxaban calibrators were spike with FEIBA (Shire, Lexington, MA).
Both Excel and GraphPad Prism (GraphPad Software, La Jolla, CA) were used for basic descriptive and comparative statistics. To evaluate conventional hospital coagulation tests, a one-way ANOVA (p < 0.05 for significance) was used to compare PT/INR results for patients without a documented history of anticoagulant use (control) against results from the same tests for DOAC patients. Receiver operating characteristic (ROC) curves were generated, and sensitivity and specificity were calculated to quantitatively assess the utility of PT/INR and the i-II-X test for the detection of FXa-I in plasma samples. Control samples were further subdivided into “normal” and “abnormal” controls using MGH’s reference ranges for conventional coagulation testing: normal PT was defined as greater than 14 seconds and normal INR was defined as greater than 1.2.
Principal component analysis (PCA) was performed to identify patterns in covariance amongst raw clotting curve data for control (C, n = 48), apixaban (A, n = 20), and rivaroxaban (R, n = 23) sample groups across three agonist concentrations. PCA was performed using uncertainty testing to optimize the number of components and the NIPALS algorithm (maximum iterations = 100) of Unscrambler X (CAMO Software, Woodbridge, NJ). The “find outlier” function was used to identify clotting times that caused over-fitting from the control group (n = 1), whereas a second application of this function on remaining measurements identified putative influencers of the model. A PCA triplot was made using centered and standardized data on the same measurement scales.
From January 2017 through August 2017, a total of 91 patient samples (control, n = 48; FXa-I, n = 43) were collected from the emergency department at MGH (Supplementary Table 1, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). The cause for emergency department admittance varied, with many FXa-I patients having exhibited complications of cardiac disease. Other patients were admitted into the emergency department for acute trauma, infection, pain or fever, neoplasia, and chest pain or respiratory distress. Among control patients, one patient was admitted for a bleeding event and two patients were admitted for a clotting event. Among FXa-I patients, one patient was admitted for a bleeding event and one patient was admitted for a clotting event; both patients were on rivaroxaban.
Detection of FXa-I using PT/INR
To evaluate whether clinician-ordered coagulation testing could detect the presence of FXa-I in patient samples, we evaluated PT/INR results. PT and INR in the MGH laboratory are sensitive to the presence of FXa-I in plasma samples with sensitivities 95.12% and 87.80%, respectively (Supplementary Fig. 2,A and B, and Supplementary Table 2, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). As expected, the specificity of PT and INR for FXa-I was low at 54.55% and 75.00%, respectively. Further subdivision of the control samples into “normal” and “abnormal” cohorts confirmed these findings (Supplementary Fig. 2C, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). These results suggest that, although PT/INR may be sensitive to anticoagulant effect DOACs in our laboratory, these tests are not specific for FXa-I–induced anticoagulation.
Comparison of Clotting Times
Clotting curves were produced for each patient sample, and the sensitivity and specificity of the i-II-X assay for the detection of FXa-I were assessed. Control sample i-II-X results had low variability, with higher concentrations of the agonists resulting in decreased TtC (Supplementary Fig. 3A, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). FXa-I patients’ clotting curves showed an increase in TtC (Supplementary Fig. 3B–D, Supplemental Digital Content 1, https://links.lww.com/CCX/A66) with statistically significant differences in TtC relative to control i-II-X results (Fig. 1A). PCA of the TtC results for these three agonist concentrations recapitulated the biphasic behavior between control and FXa-I samples—control i-II-X results (C, blue circle) demonstrate much lower covariance and form a distinct cluster relative to FXa-I results (A and R, pink circle; Fig. 1B). The comparison between the “normal” and “abnormal” control patient TtCs confirmed no significant effect on the i-II-X results in the face of abnormal PT and INR values (Supplementary Fig. 3,E and F, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). In aggregate, these data underscore the sensitivity and specificity of i-II-X for the detection of FXa-I.
Evaluation of CTS
Each patient was assigned a CTS based on their TtC curves. The CTS is indicative of the functional coagulation level (FCL per milliliter) of each specific factor being tested; specifically, it is indicative of the amount of anticoagulation secondary to inhibition from the FXa-I. A CTS of greater than 0.5 FCL/mL indicated the presence of anticoagulation secondary to FXa-I in a patient sample, and a CTS of less than equal to 0.5 FCL/mL indicated the absence of anticoagulation secondary to FXa-I in a patient sample. Plotting CTS data for control and FXa-I patients highlighted distinct differences between the two sample groups (Fig. 2A; Supplementary Fig. 4A, Supplemental Digital Content 1, https://links.lww.com/CCX/A66), and the ROC curve had an area under the curve of 0.984 (Fig. 2B). Collectively, i-II-X test results demonstrated a sensitivity of 93.02% and a specificity of 100.00% for all FXa-I patients (Supplementary Fig. 4,B and C, Supplemental Digital Content 1, https://links.lww.com/CCX/A66) and the CTS algorithm was highly sensitive and specific to both rivaroxaban (91.7% and 100.00%, respectively) and apixaban (95.0% and 100.00%, respectively).
To evaluate whether the CTS was indicative of the FXa-I concentration, we compared CTS-derived rivaroxaban concentrations to the rivaroxaban-calibrated anti-Xa chromogenic assay-derived concentrations (Supplementary Fig. 4,D and E, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). To calculate the drug concentration based on the CTS score, rivaroxaban calibrators were used to generate a best-fit line equation for the relationship between in vitro drug concentration (nanogram per milliliter) and the i-II-X–generated CTS (Supplementary Fig. 4D, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). We then applied this equation to the CTS of each patient and compared the i-II-X drug concentration with the anti-Xa concentration (Supplementary Fig. 4E, Supplemental Digital Content 1, https://links.lww.com/CCX/A66), which yielded an R2 of 0.86. Importantly, only 14 of 23 rivaroxaban samples were included in this comparison—largely owing to insufficient sample volumes for the anti-Xa assay or other reasons, including the presence of hemolysis, lipemia, or icterus, which can affect the anti-Xa assay (25). Based on the anti-Xa test results, we eliminated one rivaroxaban false negative, which turned out to be a true negative due to a rivaroxaban concentration result of 2 ng/mL, which is below the detection limit of the chromogenic anti-Xa assay (25 ng/mL). This increased the i-II-X test’s overall sensitivity and specificity to 95.20% and a specificity of 100.00% for all FXa-I patients and the rivaroxaban sensitivity and specificity to 95.45% and 100.00%, respectively (Supplementary Table S2, Supplemental Digital Content 1, https://links.lww.com/CCX/A66).
DOAC Detection with the i-II-X Test
Samples containing known amounts of DOACs were evaluated using the i-II-X test. Using the described CTS algorithmic approach, best-fit lines were generated using the CTS and drug concentration (Supplementary Fig. 5A–F, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). The CTS curves were generated with both Agonist A and Agonist B, but, importantly, FXa-Isamples tested with the Agonist B test indicated no detectable inhibition at FIIa (CTS < 0, data not shown). Dabigatran, a FIIa-I, demonstrated both Agonist A and Agonist B inhibition, consistent with the logic scheme of the i-II-X test (Supplementary Figs. 1D and 5, E and F, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). Intriguingly, comparison of the various DOAC CTSs highlighted unique curve shapes for each FXa-I (Fig. 3A), suggesting possible differences in each drug’s pharmacokinetics, despite their similar target and mechanism of action. We also used these curves to estimate the current limit of detection (LOD), based on a CTS cutoff of 0.5 FCL/mL, and the expected CTS for the therapeutic ranges of each drug (Fig. 3, B and C).
Although the FDA has historically not recommended routine testing and monitoring of DOACs, the rapid adoption and prescription of these drugs has led to a growing need for an FDA-approved DOAC testing system, especially in the emergency setting (5–7,26). Importantly, the need for DOAC testing will likely continue to mount as DOAC approval extends into vulnerable patient populations, where pharmacodynamics may vary (10,13,27–31). In this single-center pilot trial, patients admitted to the emergency department at MGH were evaluated using the i-II-X microfluidic test for the detection and identification of FXa-Is, rivaroxaban and apixaban. We show that this test is both sensitive and specific to FXa-I, even in the face of variable PT/INR results, suggesting that this test may indicate whether a patient’s prolonged PT/INR may be secondary to a DOAC.
Although other studies evaluating DOAC testing are usually done in controlled patient populations, in this study, we include patients admitted into the emergency department with a variety of diagnoses and comorbidities, including diabetes, neurodegenerative disease, cancer, heart disease, COPD, liver dysfunction, and infections (32). By including patients with complex medical histories, we were able to objectively investigate the robustness of the i-II-X test in real-world patient cohorts. In addition to being more sensitive and specific to the presence of FXa-I than PT/INR, when the PT is prolonged, the i-II-X test can provide valuable information as to whether prolongations may be secondary to an FXa-I. This kind of actionable information is important and timely, especially with the recent approvals of specific DOAC reversal agents (7,21). Additionally, we have preliminary evidence that the i-II-X test can detect and monitor the reversal of the DOAC’s anticoagulant effect in the context of treatments such as anti-inhibitor coagulant complex (FEIBA) (Supplementary Fig. 6, Supplemental Digital Content 1, https://links.lww.com/CCX/A66), suggesting that this test may be useful for the dosing/monitoring of reversal agents, a valuable tool, especially in a setting where rapid decision making and administration of high-cost treatments are necessary (33,34).
There is mounting evidence that there is interpatient variation in the level of anticoagulation secondary to FXa-I administration (35–40). The two i-II-X false negatives had corresponding anti-Xa drug concentrations of 37 and 2 ng/mL. This finding is important, as current surgical guidelines recommend rivaroxaban concentrations less than or equal to 30 ng/mL to avoid adverse bleeding events (7,21). Additionally, rivaroxaban 37 ng/mL is known to be a trough level. The estimated i-II-X–based LODs for these drugs (Fig. 3B) further support current surgical guidelines, including 30 ng/mL cutoff for surgical intervention and 50 ng/mL cutoff for reversal administration in the case of uncontrolled bleeding (7,21). These results suggest that the i-II-X CTS may serve as a physiologically relevant predictor of anticoagulation status secondary to DOAC administration (40,41). Additionally, because the i-II-X test measures a patient’s functional coagulation based on the TtC, it may not suffer from the potential challenges of performing chromogenic-based anti-Xa assays (25,41–46).
Future studies will address some of the inherent limitations of the present clinical study, namely, increasing the number of patients, performing mass spectrometry, having complete coagulation testing data for all patients, and including patients on conventional anticoagulants, that is, warfarin, although preliminary data suggest a lack of interference with warfarin (Supplementary Fig. 7, Supplemental Digital Content 1, https://links.lww.com/CCX/A66). We also plan to examine the effects of non-drug–related coagulopathies on the i-II-X test. One of the rivaroxaban patients exhibited concurrent abnormal PT and INR results (17.4 s and 1.4, respectively), along with an FXa-I CTS with a shift in the clotting curve. This may suggest a concurrent coagulopathy, such as dys- or hypofibrinogenemia, resulting in a clotting curve shift. In this regard, using the i-II-X test in conjunction with coagulation tests, such as viscoelastic testing, could provide a powerful solution for managing patients with complex coagulation profiles and medical histories (Fig. 4).
In this proof of concept pilot study, we have demonstrated that the i-II-X test can sensitively and specifically identify whether patients in an emergency department setting are anticoagulated secondary to the presence of an FXa-I. Although these results are promising, further clinical study should be performed to further establish the robustness of this new technology. In this study, the assay was performed in 10 minutes using less than 10 µL of plasma. We are currently developing an automated system to be used as a point-of-care assay on whole blood, providing results in less than 10 minutes in an emergency department setting. Future trials should include samples from multiple hospitals/institutions, strengthening the comparison between the i-II-X, currently available POC tests, and DOAC plasma levels.
1. Lippi G, Mattiuzzi C, Cervellin G, et al. Direct oral anticoagulants: Analysis of worldwide use and popularity using Google Trends.Ann Transl Med20075332
2. Bielecki S, Lee D, Hamad B. The market for oral anticoagulants.Nat Rev Drug Discov201817617–618
3. Lee LH. Doacs—Advances and limitations in real world.Thromb J20161417
4. DeWald TA, Becker RC. The pharmacology of novel oral anticoagulants.J Thromb Thrombolysis201437217–233
5. Connors JM. Testing and monitoring direct oral anticoagulants.Blood20181322009–2015
6. Douxfils J, Ageno W, Samama CM, et al. Laboratory testing in patients treated with direct oral anticoagulants: A practical guide for clinicians.J Thromb Haemost201816209–219
7. Dubois V, Dincq AS, Douxfils J, et al. Perioperative management of patients on direct oral anticoagulants.Thromb J20171514
8. Dabi A, Koutrouvelis A. Reversal strategies for intracranial hemorrhage related to direct oral anticoagulation medications.Crit Care Res Pract2018490716411
9. Maegele M, Grottke O, Schöchl H, et al. Direct oral anticoagulants in emergency trauma admissions.Dtsch Arztebl Int2016113575–582
10. Martin K, Moll S. Direct oral anticoagulant drug level testing in clinical practice: A single institution experience.Thromb Res201614340–44
11. Touzé E, Gruel Y, Gouin-Thibault I, et al. Intravenous thrombolysis for acute ischaemic stroke in patients on direct oral anticoagulants.Eur J Neurol201825747–e52
12. Hankey GJ, Norrving B, Hacke W, et al. Management of acute stroke in patients taking novel oral anticoagulants.Int J Stroke20149627–632
13. Viktil KK, Lehre I, Ranhoff AH, et al. Serum concentrations and elimination rates of direct-acting oral anticoagulants (DOACs) in older hip fracture patients hospitalized for surgery: A pilot study.Drugs Aging20193665–71
14. Macha K, Marsch A, Siedler G, et al. Cerebral ischemia in patients on direct oral anticoagulants.Stroke201950873–879
15. Tala M, Paoletti O, Dellanoce C, et al. Dabigatran plasma measurement to guide the management of acute bleeding and thrombotic complications.Eur J Case Rep Intern Med20185000947
16. Tripodi A, Ageno W, Ciaccio M, et al. Position paper on laboratory testing for patients on direct oral anticoagulants. A consensus document from the SISET, FCSA, SIBioC and SIPMeL.Blood Transfus201816462–470
17. Seyve L, Richarme C, Polack B, et al. Impact of four direct oral anticoagulants on rotational thromboelastometry (ROTEM).Int J Lab Hematol20184084–93
18. Doubleday MD, Lopex-Espina C, Matthew B, et al. Concentration correlation of direct oral anticoagulants as measured by the TEG 6S oral anticoagulant assay.ISTH2017poster 477-PB
19. Dias JD, Norem K, Doorneweerd DD, et al. Use of thromboelastography (TEG) for detection of new oral anticoagulants.Arch Pathol Lab Med2015139665–673
20. Ebner M, Birschmann I, Peter A, et al. Point-of-care testing for emergency assessment of coagulation in patients treated with direct oral anticoagulants.Crit Care20172132
21. Ten Cate H, Henskens YM, Lancé MD. Practical guidance on the use of laboratory testing in the management of bleeding in patients receiving direct oral anticoagulants.Vasc Health Risk Manag201713457–467
22. Lippi G, Favaloro EJ. Laboratory monitoring of direct oral anticoagulants (DOACs)—The perfect storm?Ann Transl Med201756
23. Siriez R, Evrard J, Dogné JM, et al. Development of new methodologies for the chromogenic estimation of betrixaban concentrations in plasma.Int J Lab Hematol201941250–261
24. R CoreTeamR: A Language and Environment for Statistical Computing. Vienna, Austria, R Foundation for Statistical Computing, 2014. Available at: http://www.R-project.org/
. Accessed July 14, 2019
25. Vera-Aguilera J, Yousef H, Beltran-Melgarejo D, et al. Clinical scenarios for discordant anti-Xa.Adv Hematol201620164054806
26. Food and Drug Administration, HHSIn vitro diagnostic testing for direct oral anticoagulants; public workshop; request for comments.FDAOctober 26, 2015
27. Bellesoeur A, Thomas-Schoemann A, Allard M, et al. Pharmacokinetic variability of anticoagulants in patients with cancer-associated thrombosis: Clinical consequences.Crit Rev Oncol Hematol2018129102–112
28. Graves KK, Edholm K, Johnson SA. Use of oral anticoagulants in obese patients.JSM Atheroscler201721035
29. Malec L, Young G. Treatment of venous thromboembolism in pediatric patients.Front Pediatr2017526
30. Forbes HL, Polasek TM. Potential drug–drug interactions with direct oral anticoagulants in elderly hospitalized patients.Ther Adv Drug Saf20178319–328
31. von Vajna E, Alam R, So TY. Current clinical trials on the use of direct oral anticoagulants in the pediatric population.Cardiol Ther2016519–41
32. Dias JD, Lopez-Espina CG, Ippolito J, et al. Rapid point-of-care detection and classification of direct-acting oral anticoagulants (DOACs) with the TEG® 6s: Implications for trauma and acute care surgery.J Trauma Acute Care Surg2019 Apr 29. [Epub ahead of print]
33. Tomaselli GF, Mahaffey KW, Cuker A, et al. 2017 ACC expert consensus decision pathway on management of bleeding in patients on oral anticoagulants: A report of the American College of Cardiology Task Force on Expert Consensus Decision Pathways.J Am Coll Cardiol2017703042–3067
34. Eikelboom JW, Kozek-Langenecker S, Exadaktylos A, et al. Emergency care of patients receiving non-vitamin K antagonist oral anticoagulants.Br J Anaesth2018120645–656
35. Gulilat M, Tang A, Gryn SE, et al. Interpatient variation in rivaroxaban
and apixaban plasma concentrations in routine care.Can J Cardiol2017331036–1043
36. Testa S, Tripodi A, Legnani C, et al.; START-Laboratory RegisterPlasma levels of direct oral anticoagulants in real life patients with atrial fibrillation: Results observed in four anticoagulation clinics.Thromb Res2016137178–183
37. Testa S, Dellanoce C, Paoletti O, et al. Edoxaban plasma levels in patients with non-valvular atrial fibrillation: Inter and intra-individual variability, correlation with coagulation screening test and renal function.Thromb Res201917561–67
38. Testa S, Paoletti O, Legnani C, et al. Low drug levels and thrombotic complications in high-risk atrial fibrillation patients treated with direct oral anticoagulants.J Thromb Haemost201816842–848
39. Barra ME, Fanikos J, Connors JM, et al. Evaluation of dose-reduced direct oral anticoagulant therapy.Am J Med20161291198–1204
40. Brummel-Ziedins K, Orfeo T, Gissel M, et al. Factor Xa generation by computational modeling: An additional discriminator to thrombin generation evaluation.PLoS One20127e29178
41. Ebner M, Birschmann I, Peter A, et al. Limitations of specific coagulation tests for direct oral anticoagulants: A critical analysis.J Am Heart Assoc20187e009807
42. Ikejiri M, Wada H, Tone S, et al. Comparison of three different anti-Xa assays in major orthopedic surgery patients treated with direct oral anticoagulant.Thromb J20171527
43. Burns ER, Yoshikawa N. Hemolysis in serum samples drawn by emergency department personnel versus laboratory phlebotomists.Lab Med20025378–380
44. Phelan MP, Reineks EZ, Schold JD, et al. Preanalytic factors associated with hemolysis in emergency department blood samples.Arch Pathol Lab Med2018142229–235
45. Giuseppe L, Plebani M, Di Domma S, et al. Hemolyzed specimens: A major challenge for emergency departments and clinical laboratories.Crit Rev Clin Lab Sci201148143–153
46. Al-Sallami HS, Medlicott NJ. Investigation of an anti-activated factor X (anti-Xa) assay for the quantification of enoxaparin in human plasma.J Pharm Pharmacol201567209–214