Coagulation Testing in Acute Ischemic Stroke Patients Taking Non–Vitamin K Antagonist Oral Anticoagulants
Background and Purpose—In patients who present with acute ischemic stroke while on treatment with non–vitamin K antagonist oral anticoagulants (NOACs), coagulation testing is necessary to confirm the eligibility for thrombolytic therapy. We evaluated the current use of coagulation testing in routine clinical practice in patients who were on NOAC treatment at the time of acute ischemic stroke.
Methods—Prospective multicenter observational RASUNOA registry (Registry of Acute Stroke Under New Oral Anticoagulants; February 2012–2015). Results of locally performed nonspecific (international normalized ratio, activated partial thromboplastin time, and thrombin time) and specific (antifactor Xa tests, hemoclot assay) coagulation tests were documented. The implications of test results for thrombolysis decision-making were explored.
Results—In the 290 patients enrolled, nonspecific coagulation tests were performed in ≥95% and specific coagulation tests in 26.9% of patients. Normal values of activated partial thromboplastin time and international normalized ratio did not reliably rule out peak drug levels at the time of the diagnostic tests (false-negative rates 11%–44% [95% confidence interval 1%–69%]). Twelve percent of patients apparently failed to take the prescribed NOAC prior to the acute event. Only 5.7% (9/159) of patients in the 4.5-hour time window received thrombolysis, and NOAC treatment was documented as main reason for not administering thrombolysis in 52.7% (79/150) of patients.
Conclusions—NOAC treatment currently poses a significant barrier to thrombolysis in ischemic stroke. Because nonspecific coagulation test results within normal range have a high false-negative rate for detection of relevant drug concentrations, rapid drug-specific tests for thrombolysis decision-making should be established.
Patients on oral anticoagulation treatment at the time of acute ischemic stroke (AIS) pose a frequent challenge as thrombolysis is contraindicated in those who are effectively anticoagulated.1–3 Patients on vitamin K antagonists who present with an international normalized ratio (INR) of ≤1.7 can be thrombolysed without a significant increase in the risk of hemorrhagic complications.4,5 Point-of-care INR testing allows for rapid decision-making with regards to the decision to proceed with thrombolytic therapy.6 In contrast, the decisions around thrombolysis treatment in patients on non–vitamin K antagonist oral anticoagulants (NOAC) are an unresolved issue. Current American Heart Association/American Stroke Association guidelines recommend that thrombolysis can be administered if there is evidence that the patient failed to take the anticoagulant or if there is evidence that excludes significant anticoagulatory activity as measured by sensitive coagulation tests.7,8
Drug-specific coagulation tests are currently not available at the bedside, but are usually performed in central laboratories, which leads to delays in obtaining the necessary results. Furthermore, these tests are frequently not available out of hours.9 Calibrated chromogenic anti-Xa assays allow for the quantitative assessment of drug-specific concentrations in the case of the factor Xa inhibitors rivaroxaban and apixaban.10 For dabigatran, modified versions of the diluted thrombin time (TT) test are available. Mass spectrometry can provide reliable drug-level estimates but are usually unavailable in the emergency setting.11 Importantly, although guidance for the interpretation of coagulation test result allowing safe thrombolysis in patients on NOAC treatment have been proposed,12,13 data supporting their validity is limited.11,14 The impact of time-consuming NOAC-specific coagulation testing in acute stroke management is currently unclear, and the actual use of these tests in clinical practice is unknown.
We report on the current use of standard and specific coagulation tests in assessing NOAC-related anticoagulant activity in routine clinical care among AIS patients enrolled into the prospective multicenter RASUNOA pilot registry (Registry of Acute Stroke Under New Oral Anticoagulants).
Study Design, Setting, and Patients
RASUNOA was an investigator-initiated, multicenter, prospective, observational cohort study without commercial funding (ClinicalTrials.gov, NCT01850797). Thirty-eight Departments of Neurology, with certified stroke units across Germany, participated in the registry. Patients with AIS were prospectively enrolled into the RASUNOA study between February 2012 and February 2015. The inclusion criteria were age ≥18 years and current therapy with a NOAC (ie, apixaban, dabigatran, or rivaroxaban) at the time of stroke onset. Approval was obtained from the ethics committee of the Medical Faculty of Heidelberg, Germany, as well as from the ethics committees of each participating center.
Because of the observational nature of the study, all diagnostic and treatment decisions were left to the discretion of the treating physicians. Participation in RASUNOA had no influence on local standard operating procedures with regards to the use of specific or nonspecific coagulation tests. Routine laboratory results of nonspecific coagulation tests (activated partial thromboplastin time [aPTT], ecarin clotting time [ECT], TT, or INR), drug-specific coagulation tests (anti-factor Xa or hemoclot assay), platelet count, and renal function (creatinine and glomerular filtration rate) obtained at admission were collected using a prespecified case report form. Reference ranges for laboratory values are provided in the expanded methods section in the online-only Data Supplement (including dose-specific concentration ranges; Table I in the online-only Data Supplement). Information about medical history, stroke severity, and clinical course, as well as the time of stroke onset (or last seen well in case of unknown onset) and the time of last drug intake, were documented.
Decision-Making Based on Coagulation Tests
The impact of coagulation test results on the decision to administer thrombolytic treatment to patients with ischemic stroke on NOAC treatment was explored by applying 2 previously published protocols.12,13 The exploratory analysis was limited to patients presenting within a 4.5-hour time window.7 The upper ranges of normal values of nonspecific coagulation tests, as well as peak and trough levels of NOACs, are summarized in the methods section in the online-only Data Supplement. In the protocol by Steiner et al,12 in the absence of a generally accepted upper range of normal of the ECT, we set the upper range of normal to 64 s because this value most closely correlated with the proposed dabigatran concentration threshold of 50 ng/mL (data not shown). We used the 5% of the lowest trough range as the upper range of normal for anti-Xa levels of rivaroxaban (12 ng/mL) and apixaban (34 ng/mL).12,15,16 Following the protocol by Kepplinger et al, we used the INR threshold of <1.4 and performed a sensitivity analysis using an INR threshold of <1.2 to account for different thromboplastin reagents with different sensitivities.13,17
Continuous variables are presented with means and standard deviations. Categorical variables are presented with medians and interquartile ranges, and absolute and relative frequencies are reported. To assess the correlation between specific and nonspecific coagulation test values and the performance of specific tests, the Spearman’s nonparametric correlation is calculated. We calculated test characteristics for the detection of peak range drug concentration levels for the nonspecific tests aPTT, INR, and TT by cross-tabulation of dichotomized nonspecific tests (tests within normal range versus elevated tests values) and drug concentrations (below lower peak range level versus within or above peak range). The false-negative rate was calculated as 1-sensitivity. All statistical tests were 2-sided, and P values of <0.05 were considered statistically significant. If not indicated otherwise, analyses were conducted using IBM SPSS Statistics, version 22.214.171.124 (IBM SPSS, Armonk, NY).
A total of 290 ischemic stroke patients treated with NOACs at the time of stroke were enrolled into the prospective multicenter registry. Baseline characteristics are summarized in Table 1 (mean age 77.1 [SD 9.2] years, 48.3% women). The majority of patients were mildly to moderately affected (National Institute of Health Stroke Scale score at admission, 4 [interquartile range 1–8]).
Availability and Results of Routine Coagulation Testing
Table 2 summarizes the availability of nonspecific and specific coagulation tests. Standard coagulation parameters, such as platelet count, INR, and aPTT, were available in almost all patients. TT was measured in less than half of the patients on dabigatran (44.7%).
On admission, 60.2% of patients anticoagulated with rivaroxaban (100/166), and only 21.3% of those on apixaban (10/47), had an elevated INR (Figure 1C and 1D). Slightly elevated INR levels were also observed in 56.2% of the 73 patients treated with dabigatran (Figure 1A and 1B; Table II in the online-only Data Supplement). In contrast, the TT was above the upper limit of normal (>24 s) in 94% of the dabigatran-treated patients but only in 14% of patients taking factor Xa inhibitors (Table II in the online-only Data Supplement). As expected, the aPTT was also more frequently prolonged in dabigatran (65%) compared with rivaroxaban (32%) and apixaban-treated patients (13%; P<0.001; Table II in the online-only Data Supplement).
Normal values of the aPTT and INR did not reliably rule out peak drug levels as determined by calibrated tests (false-negative rates 11%–44% [95% confidence interval 1%–69%], irrespective of the NOAC used (test characteristics are summarized in the Table III in the online-only Data Supplement). Figure 1 further illustrates the high false-negative rate of the nonspecific parameters, aPTT and INR (Figure 1A, 1C, and 1D), in both factor Xa and direct thrombin inhibitors and TT in factor Xa inhibitors, respectively (Figure 1C and 1D).
Availability and Use of Specific Coagulation Tests
Specific coagulation tests for NOAC treatment were performed in less than half of the patients. Anti-factor Xa activity tests specific for NOACs were more frequently performed in patients on rivaroxaban (42.5%) compared with those on apixaban (17.0%; Table 2). Longer time from symptom onset to admission was associated with a less frequent performance of specific coagulation testing (Spearman’s rho =−0.146 [95% confidence interval −0.28 to −0.1]; P=0.04).
NOAC concentration levels were highly variable at admission (Figure 2A through 2C) even when similar intervals since last intake were compared. In 69 patients with atrial fibrillation taking approved doses of NOACs for the prevention of ischemic stroke, quantitative NOAC concentration measurements were available. Of these, 58% had drug levels within the expected dose-specific trough and peak level ranges. In contrast, 25% had drug levels below the trough level. Interestingly, 17% experienced a stroke, although drug levels exceeded the peak range. Based on specific coagulation tests, or normal TT in the case of dabigatran, 12% (6/50) of patients had apparently failed to take the prescribed NOAC.
Potential Eligibility for Thrombolysis and Anticoagulation Testing
In RASUNOA, only 9 of all AIS patients presenting in the 4.5-hour time window received intravenous thrombolysis (9/159; 5.7%). Suspected or proven NOAC treatment was documented as the main reason for not administering thrombolysis in 52.7% (79/150) of patients.
Different models have been proposed to aid thrombolysis decision-making in patients who are anticoagulated with NOACs. Table 3 summarizes the consequences if the suggested thresholds12,13 had been applied for off-label thrombolysis in patients presenting within a 4.5-hour time window in RASUNOA. The number of patients theoretically eligible for intravenous thrombolysis based on coagulation parameters alone heavily depended on the decision protocol. For example, in rivaroxaban-treated patients, if the decision to thrombolyse had been based on normal anti-Xa levels, only 12% of patients would have been eligible for thrombolysis. In contrast, basing eligibility for thrombolysis on normal aPTT and prothrombin time values, the number of eligible patients (24%) would have doubled. This highlights that the sensitivity of aPTT and prothrombin time/INR for the detection of low or even peak drug concentrations of rivaroxaban is low.
Our study has yielded 5 new findings with regards to the use of coagulation testing in AIS patients treated with NOACs: (1) standard coagulation tests are not reliable in predicting actual NOAC drug levels; (2) specific coagulation tests are performed in less than half of acute stroke patients in the emergency setting; (3) ischemic stroke occurs despite NOAC drug concentrations within the peak range at the time of the stroke; (4) decision-making for off-label thrombolysis in NOAC-anticoagulated patients based on currently proposed protocols yields inconsistent conclusions depending on whether nonspecific or specific coagulation tests are used; and (5) treatment with NOACs is currently a barrier for thrombolysis in AIS.
Although most stroke centers participating in our multicenter study were large and experienced, drug-specific coagulation testing was performed in less than half of the patients. This is suprising given that our observational data demonstrate that nonspecific coagulation tests do not provide reliable information on the current anticoagulation status of NOAC-treated patients. Nonspecific coagulation tests may only be of value if highly sensitive reagents are used and locally determined reagent-specific cutoffs are established for each test and each NOAC.11,18 As a consequence, current available guidance on the use of nonspecific coagulation tests in the decision-making process for thrombolysis administration may have to be revised.7,8
The thrombin inhibitor dabigatran has different effects on standard coagulation parameters when compared with factor Xa inhibitors.19 In our study, the aPTT was more often prolonged in patients on dabigatran, whereas the INR was elevated in the majority of patients receiving rivaroxaban or dabigatran, but not in patients taking apixaban. Use of a less sensitive factor Xa inhibitor recombinant thromboplastin in some centers may explain this finding. Therefore, clinicians should be aware of the thromboplastin time reagent used in their local laboratory. For example, some reagents are sensitive to rivaroxaban (eg, Neoplastin Plus or HemosIL RecombiPlasTin 2G), whereas others barely react (eg, Innovin).17 Importantly, a particularly low sensitivity for apixaban is observed with all current reagents used for prothrombin time testing. Therefore, their use is not recommended with apixaban.10
The occurrence of a stroke in a patient on a NOAC is frequently attributed to failure of anticoagulant intake just prior to the stroke (eg, because of a missed dose).20 This has been of particular concern for NOACs with once-daily dosing regimens where a single missed dose may result in critically low concentrations.21 Notably, results of coagulation tests in our study suggest that failure to take a NOAC immediately before the stroke may not be associated with a substantially increased stroke risk. Only 12% of patients with available specific coagulation tests had no NOAC activity. Instead, concentrations a few hours after symptom onset suggested that drug levels were in the peak range at the time of the stroke or even exceeded established peak range levels. A potential explanation for this finding might be that the patients experienced strokes of other etiologies rather than cardioembolism. Although hypercoagulable states are rare, we cannot rule out the presence of these conditions in individual cases.
Based on the experience with vitamin K antagonists, effective anticoagulation with NOACs is a contraindication to thrombolysis because of a potentially increased risk of symptomatic intracranial hemorrhage.1 A recent study aggregating data from several centers found no increased risk of symptomatic intracranial hemorrhage in patients on NOAC treatment who underwent thrombolysis compared with patients not on anticoagulation treatment.14 However, drug concentrations were not available in the majority of patients, and the median time interval since last NOAC intake was 13 hours.
Two protocols have been proposed, which incorporate laboratory-based nonspecific and specific coagulation test results in decision-making for thrombolysis.12,13 Notably, both protocols lack prospective validation of their safety. Post hoc application of these protocols to our cohort of NOAC-treated AIS patients showed large inter- and intraprotocol differences. In the model proposed by Steiner et al,12 the number of patients treated with rivaroxaban deemed eligible for thrombolysis was nearly doubled when only aPTT and prothrombin time were used (n=12/32 versus n=22/82; Table 3). The number of patients theoretically eligible for thrombolysis was even higher13 when the INR-based threshold alone, as proposed by Kepplinger et al, was used (n=42/83; at INR <1.2). Our data suggest that reliance on aPTT and INR may result in potentially dangerous underestimation of the actual anticoagulatory effect of NOACs. The use of the TT should be limited to patients taking dabigatran, and this can only be used to exclude any dabigatran usage.19 In our series, patients with low dabigatran concentrations still show a markedly prolonged TT (Figure 1).
Preexisting oral anticoagulation treatment in patients presenting with an AIS is a challenge for thrombolysis decision-making. However, the current impact of anticoagulation treatment with NOACs on thrombolysis rates in acute stroke is largely unknown. In our study, only 6% of patients on NOAC treatment at the time of AIS received thrombolytic treatment. In 53% of patients, treatment with an NOAC was the reported reason for withholding thrombolysis. This contrasts with a reported thrombolysis rate of 50% to 64% of ischemic stroke patients presenting within 4.5 hours after symptom onset in a nation-wide quality report.22 This discrepancy highlights that NOAC treatment at the time of a stroke represents a significant barrier for thrombolysis. Establishment and widespread availability of drug-specific point-of-care coagulation devices for NOACs, as well as prospective registry data, may help to refine parameters allowing safe thrombolysis, despite treatment with a NOAC.
Our study has some limitations. We relied on information provided by patients and caregivers with regards to the time of last drug ingestion, and this information may not always have been accurate. Furthermore, delays in blood sampling may have resulted in lower drug concentrations than actually present at the exact time of admission. RASUNOA was performed in a single country in selected centers only, and current practices may have been different in other settings. With regards to the administration of thrombolysis, the local physician’s decision may have been influenced by the relatively mild to moderate deficits of the patients.
In conclusion, our multicenter observational study indicates that drug-specific coagulation testing is not yet part of clinical routine in the majority of major German stroke centers, and standard coagulation tests often do not reflect the anticoagulatory activity of NOACs adequately. More evidence is needed to establish solid reference ranges of coagulation test results, including point-of-care devices, for safe and rapid thrombolysis in acute stroke.
We thank all principal investigators of the RASUNOA study and participating hospitals who enrolled at least one ischemic stroke patient (A–Z). A. Binder (Kiel), M. Dichgans (München), R. Dziewas (Münster), K. Gröschel (Mainz), M. Eicke (Idar-Oberstein), M. Ertl (Regensburg), M. G. Hennerici (Mannheim), C. Hobohm (Leipzig), T. Höhle (Herne), S. Jander (Düsseldorf), E. Jüttler (Ulm), A. Khaw (Greifswald), C. Kleinschnitz (Würzburg), A. Kraft (Halle), M. Köhrmann (Erlangen), F. Meisel (Karlsruhe), T. Neumann-Haefelin (Fulda), C. Opherk (Heilbronn), F. Palm (Ludwigshafen), S. Poli (Tübingen), J. Röther (Hamburg), E. Schmid (Stuttgart), G. Seidel (Hamburg), H. Soda (Bad Neustadt), C. Tanislav (Gießen), G. Thomalla (Hamburg), R. Veltkamp (Heidelberg), K. Wartenberg (Halle-Wittenberg), C. Weimar (Essen).
Sources of Funding
This study is investigator initiated, without commercial funding.
Personal fees, speakers, consulting honoraria, research support were received from Pfizer (Drs Purrucker, Rizos, Poli, Kleinschnitz, Palm, Soda), BMS (Drs Rizos, Poli, Kraft, Kleinschnitz, Palm), Boehringer Ingelheim (Drs Purrucker, Rizos, Poli, Kraft, Kleinschnitz, Palm, Soda), Bayer (Drs Rizos, Poli, Kraft, Kleinschnitz, Veltkamp), Daiichi Sankyo (Rizos, Poli, Kraft, Kleinschnitz, Palm, Soda, Veltkamp), and CSL Behring (Dr Veltkamp) outside of present work. Dr Heuschmann reports grants from German Federal Ministry of Education and Research, European Union, Charité, Berlin Chamber of Physicians, German Parkinson Society, University Hospital Würzburg, Robert-Koch-Institute, Charité–Universitätsmedizin Berlin (within MonDAFIS, supported by unrestricted research grant to Charité from Bayer), University Göttingen (within FIND-AF, supported by an unrestricted research grant from Boehringer Ingelheim), University Hospital Heidelberg (RASUNOA-prime, supported by unrestricted research grant from Bayer, BMS, Boehringer Ingelheim, Daiichi Sankyo) outside of the present work. The other authors report no conflicts.
↵* A list of all RASUNOA investigators is given in the Appendix.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.014963/-/DC1.
- Received July 31, 2016.
- Revision received October 19, 2016.
- Accepted November 8, 2016.
- © 2016 American Heart Association, Inc.
- Hankey GJ,
- Norrving B,
- Hacke W,
- Steiner T
- Veltkamp R,
- Rizos T
- Rizos T,
- Herweh C,
- Jenetzky E,
- Lichy C,
- Ringleb PA,
- Hacke W,
- et al
- Jauch EC,
- Saver JL,
- Adams HP Jr,
- Bruno A,
- Connors JJ,
- Demaerschalk BM,
- et al
- Demaerschalk BM,
- Kleindorfer DO,
- Adeoye OM,
- Demchuk AM,
- Fugate JE,
- Grotta JC,
- et al
- Drouet L,
- Bal Dit Sollier C,
- Steiner T,
- Purrucker J
- Ebner M,
- Peter A,
- Spencer C,
- Härtig F,
- Birschmann I,
- Kuhn J,
- et al
- Steiner T,
- Böhm M,
- Dichgans M,
- Diener HC,
- Ell C,
- Endres M,
- et al
- Seiffge DJ,
- Hooff RJ,
- Nolte CH,
- Béjot Y,
- Turc G,
- Ikenberg B,
- et al
- 16.↵European Medicines Agency. Apixaban. In: Summary of Product Characteristics. 2016. http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Product_Information/human/002148/WC500107728.pdf. Accessed July 31, 2016.
- Lippi G,
- Ardissino D,
- Quintavalla R,
- Cervellin G
- Heidbuchel H,
- Verhamme P,
- Alings M,
- Antz M,
- Diener HC,
- Hacke W,
- et al
- Clemens A,
- Noack H,
- Brueckmann M,
- Lip GY
- Vrijens B,
- Heidbuchel H
- Wiedmann S,
- Heuschmann PU,
- Hillmann S,
- Busse O,
- Wiethölter H,
- Walter GM,
- et al