Laboratory Measurement of the Anticoagulant Activity of the Non–Vitamin K Oral Anticoagulants

Study selection

Our literature search yielded 160 articles. Nine additional references were identified from bibliographies. A total of 152 articles were excluded: 135 did not report an original research study, 14 did not report drug levels measured by LC-MS/MS, and 3 were published as abstracts only. The remaining 17 articles (11–27) met eligibility criteria (Figure 1). Eligible studies were collectively conducted in 9 different countries across a range of dabigatran concentrations from 0 to 1,886 ng/ml. Only 4 studies used ex vivo plasma from dabigatran-treated patients; the remainder involved ex vivo healthy volunteer plasma or normal plasma spiked in vitro with dabigatran as the test material (Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

PRISMA Diagram

This PRISMA (Preferred Reporting Items of Systematic Reviews and Meta-Analyses) flow diagram illustrates dabigatran, rivaroxaban, and apixaban literature searches.

Activated partial thromboplastin time

Twelve eligible studies reported a relationship between the activated partial thromboplastin time (APTT) and dabigatran levels. Three used ex vivo patient plasma, 2 ex vivo healthy volunteer plasma, and 7 normal plasma spiked with dabigatran in vitro (Table 2). Dabigatran prolonged the APTT in a concentration-dependent manner in both ex vivo and in vitro studies. The dose response was linear up to a concentration of 200 to 300 ng/ml and then flattened out at higher drug levels (17,28). This curvilinear relationship did not permit quantitative assessment of dabigatran levels, particularly at higher concentrations.

Commercial APTT reagents differ widely in their sensitivity to dabigatran. The APTT of plasma spiked with dabigatran 120 ng/ml ranged from 26.0 to 91.9 s in a cross-validation study of 9 different APTT methods (27). These findings suggest that coagulation laboratories should perform dose-response studies using calibration standards to determine the sensitivity of their particular APTT method to dabigatran and communicate the results to clinicians. The least sensitive reagents required a dabigatran concentration of approximately 400 ng/ml to produce a 2-fold prolongation in the APTT over control (19). The APTT may not be prolonged in the presence of typical on-therapy trough levels (Table 1), particularly if a relatively insensitive reagent is used. In a study of ex vivo plasma from patients taking dabigatran 150 mg twice daily, 18% of participants had a normal APTT at trough (25). In another study of patients with AF, some samples had an APTT within the normal range despite dabigatran concentrations as high as 60 ng/ml (23).

Prothrombin time/international normalized ratio

Eleven studies reported a relationship between dabigatran levels and the prothrombin time/international normalized ratio (PT/INR). Eight used spiked plasma, 2 ex vivo patient plasma, and 1 ex vivo plasma from healthy volunteers (Table 2). Dabigatran prolonged the PT in a concentration-dependent manner, as defined by an exponential (i.e., nonlinear) relationship (17). The PT/INR was less sensitive to dabigatran than the APTT. In an ex vivo study of patients with AF, an INR of 1.2 or greater was only observed with dabigatran concentrations in excess of 400 ng/ml (23).

As with APTT, commercial PT reagents differ in their sensitivity to dabigatran. In a survey of 71 coagulation laboratories, the PT of plasma spiked to a concentration of 300 ng/ml ranged from 15.7 to 50.2 s, depending on the reagent (27). PT may be measured using 2 assay types. The Quick method is influenced by the entire extrinsic and common pathways of coagulation, whereas the Owren method is affected only by factors II, VII, and X. In studies of spiked plasma, Quick PT reagents were more sensitive to dabigatran than Owren reagents (14,27).

Point-of-care devices are available for measuring the INR (POC-INR) in whole blood in VKA-treated patients. In one study assessing the relationship between dabigatran levels and a single POC-INR system, the POC system yielded INR values 2- to 4-fold higher than those obtained using a laboratory PT/INR method. Dabigatran concentrations >500 ng/ml were beyond the POC system’s limit of detection (22).

Thrombin time

Six studies reported a relationship between thrombin time (TT) and dabigatran levels (Table 2). The TT (in unmodified form) was inordinately sensitive in both ex vivo and in vitro studies. Depending on the reagent, dabigatran concentrations of as little as 25 and up to 150 ng/ml exceeded the limits of detection (15–17,28).

Dilute TT

In the dilute TT assay, the excessive sensitivity of the TT is overcome by diluting the plasma sample (21). Seven studies reported a relationship between the dilute TT and dabigatran concentration (Table 2). Two studies used an in-house modification of the TT (15,20); 5 used the HEMOCLOT thrombin inhibitor assay (HYPHEN BioMed, Neuvillesur-Oise, France), a commercially available dilute TT test (17,21,23–25). Dabigatran prolonged the assay in a concentration-dependent manner. The relationship showed a high degree of linearity, with R² values ranging between 0.92 and 0.99 in both in vitro and ex vivo studies. The lower limit of detection according to the manufacturer is 50 ng/ml (21). Two studies determined the assay to be less accurate and more variable at concentrations below 50 to 100 ng/ml (15,21). The dilute TT is currently not widely available; in a recent survey in Australia and New Zealand, only 9 of 592 laboratories reported using it (29).

Ecarin-based assays

Ecarin is a snake venom that cleaves prothrombin to form meizothrombin, an unstable intermediate of thrombin. Dabigatran inhibits the thrombin-like activity of meizothrombin (24). Two assays use ecarin as an activator: the ecarin clotting time (ECT) and ecarin chromogenic assay (ECA).

Six studies reported a relationship between ECT and dabigatran levels (Table 2). Both in vitro and ex vivo studies demonstrated a high degree of linearity, with R² values ranging between 0.92 and 1.00. Loss of linearity was observed in 2 studies at dabigatran levels in excess of 470 to 500 ng/ml (17,19). A relationship between ECA and drug levels was reported in 4 studies (Table 2). The relationship was linear, with R² values of 0.94 to 0.99 in in vitro and ex vivo samples. One study identified greater variability at dabigatran concentrations <50 ng/ml (23). The ECT and ECA are hampered by lack of standardization, variability in sensitivity to dabigatran among different lots of ecarin, and limited availability (17,20).

Other assays

Relationships between dabigatran level and the dilute PT, prothrombinase-induced clotting time (PiCT), and activated clotting time (ACT) were each reported in a single study. Both the dilute PT and PiCT evinced a complex nonlinear dose-response curve (17). As with heparin, the ACT proved insensitive to lower concentrations of dabigatran. In an ex vivo study, the ACT was normal in 40% of trough samples despite on-therapy dabigatran levels (25).

Study selection

Our literature search yielded 134 unique rivaroxaban articles. Two additional references were identified from bibliographies. We excluded 121 articles: 108 did not report an original research study, 12 did not report a relationship between a coagulation assay and drug levels measured by LC-MS/MS, and 1 was published as an abstract only. The remaining 15 articles (27,37–50) met eligibility criteria (Figure 1). Rivaroxaban concentrations in eligible studies ranged from 0 to >1,000 ng/ml. Four studies used ex vivo plasma from rivaroxaban-treated patients, 1 incorporated ex vivo plasma from healthy controls, and the remainder used normal plasma spiked in vitro with rivaroxaban (Table 3).

Prothrombin time

We found 11 studies evaluating the effect of rivaroxaban on PT (Table 3). In general, rivaroxaban prolonged the PT in a concentration-dependent, linear fashion in plasma spiked with rivaroxaban and in plasma from patients receiving rivaroxaban for approved indications. On-therapy rivaroxaban concentrations showed a modest effect on PT. Typical trough (41 to 60 ng/ml) and peak (219 to 305 ng/ml) concentrations increased PT by 6% to 19% and 50% to 135%, respectively (27,38,42,47). Assay sensitivity varied significantly among thromboplastin reagents. Interassay variability was reduced by use of an international sensitivity index specific for rivaroxaban but not by conversion to an INR used for monitoring VKA therapy (38,42). These observations suggest that coagulation laboratories should perform dose-response studies using calibration standards to determine the sensitivity of their particular PT method for rivaroxaban and communicate the results to clinicians.

Activated partial thromboplastin time

Five studies evaluated the effect of rivaroxaban on APTT (Table 3). Whereas rivaroxaban prolonged APTT in a dose-dependent manner, the overall relationship between rivaroxaban concentration and APTT prolongation was nonlinear, with studies reporting conflicting data regarding the concentration ranges over which nonlinearity was most pronounced (27,41,45). Similar to PT results, there was significant variability among reagents and among individual laboratories in a multicenter study (27,45). Hillarp et al. (41) reported that the APTT assay was insensitive at the lowest drug level studied (25 ng/ml).

Anti-Xa activity

Ten studies assessed the effect of rivaroxaban on anti-Xa activity (Table 3). In general, the studies showed a linear, concentration-dependent relationship between rivaroxaban concentration and anti-Xa activity over a wide range of concentrations (e.g., 20 to 660 ng/ml) when they were measured using a standard curve generated with rivaroxaban calibrators and controls with R² values ranging from 0.95 to 1.00 (39,43,44,49,50). The correlation was less robust at concentrations <100 ng/ml (49). However, Samama et al. (46) demonstrated that low rivaroxaban concentrations could be measured with a modified anti-Xa test using less diluted samples. Investigators found a greater degree of assay imprecision at higher rivaroxaban concentrations (800 ng/ml) in one study (45). In a multicenter study, both intralaboratory and interlaboratory precision were satisfactory except at the lower limit of detection (20 ng/ml); use of a centrally distributed reagent reduced interlaboratory variability (46). Mathematical modeling also decreased interassay variability resulting from different sensitivities of individual reagents to rivaroxaban (40). When commercial anti-Xa assays were used with unfractionated or low-molecular-weight heparin (LMWH) calibrators (rather than rivaroxaban calibrators), the relationship remained linear up to a rivaroxaban concentration of 500 ng/ml (38,41).

Other assays

The relationship between rivaroxaban concentration and the dilute PT, dilute Russell viper venom time (dRVVT), and PiCT was evaluated in a single study (38). Researchers uncovered a linear, dose-dependent relationship between rivaroxaban and the dilute PT. Rivaroxaban increased the dRVVT ratio (expressed as ratio vs. baseline) in a concentration-dependent but nonlinear manner. At low concentrations of rivaroxaban (<200 ng/ml), there was a paradoxical shortening of PiCT, whereas PiCT was prolonged in a concentration-dependent fashion at higher concentrations.

Study selection

Our literature search for apixaban yielded 70 articles; 3 additional references were identified from bibliographies. Sixty-nine articles were excluded: 68 did not report an original research study and 1 was published as an abstract only. The remaining 4 articles (37,54–56) met eligibility criteria (Figure 1). Eligible studies collectively evaluated apixaban across a range of concentrations from 0 to 2,500 ng/ml (Table 4).

Prothrombin time/international normalized ratio

We found 3 studies that reported the relationship between the PT and apixaban levels. One study used both spiked normal plasma and ex vivo plasma from apixaban-treated patients (37), another used spiked plasma as well as ex vivo plasma from healthy volunteers taking apixaban (55), and the third study included only spiked normal plasma samples (56) (Table 4). For the 2 studies that used ex vivo plasma, the relationship was linear in one (37) and curvilinear in the other (55). Correlation was modest, with R² values of 0.36 and 0.41, respectively. Across in vitro and ex vivo samples and for a variety of reagents, the PT was inadequately sensitive to apixaban, not only below, but also above the expected trough concentration of 50 ng/ml.

Activated partial thromboplastin time

Only 1 study compared apixaban concentrations with APTT (Table 3), which was measured using 10 different APTT reagents in normal plasma samples spiked with 10 different concentrations of apixaban. The sensitivity of the APTT was unacceptably low; all assays yielded a ratio of 1.1 times control when the spiked apixaban concentration was 100 ng/ml (i.e., twice the expected trough concentration).

Anti-Xa activity

Three studies compared anti-Xa activity measurements with the plasma concentration of apixaban (Table 4). In 2 of the studies (37,54), ex vivo patient samples were used, whereas the third study (56) included only spiked samples of normal plasma. In general, the relationship was linear at all apixaban concentrations, with R² values ranging from 0.89 to 0.97. Available evidence suggests that an anti-Xa assay calibrated with LMWH standards will also correlate linearly with apixaban concentrations (37,54).