Introduction Endovascular therapy in acute ischemic stroke is safe and efficient. However, patients receiving oral anticoagulation were excluded in the larger trials.
Objective To analyze the safety of endovascular therapy in patients with acute ischemic stroke and elevated international normalized ratio (INR) values.
Methods Retrospective database review of a tertiary care university hospital for patients with anterior circulation stroke treated with endovascular therapy. Patients with anticoagulation other than vitamin K antagonists were excluded. The primary safety endpoint was defined as symptomatic intracranial hemorrhage (sICH; ECASS II definition). The efficacy endpoint was the modified Rankin scale (mRS) score after 3 months, dichotomized into favorable outcome (mRS 0–2) and unfavorable outcome (mRS 3–6).
Results 435 patients were included. 90% were treated with stent retriever. 27 (6.2%) patients with an INR of 1.2–1.7 and 21 (4.8%) with an INR >1.7. 33 (7.6%) had sICH and 149 patients (34.3%) had a favorable outcome. Patients with an elevated INR did not have an increased risk for sICH or unfavorable outcome in multivariable analysis. The additional use of IV thrombolysis in patients with an INR of 1.2–1.7 did not increase the risk of sICH or unfavorable outcome. These results were replicated in a sensitivity analysis introducing an error of the INR of ±5%. They were also confirmed using other sICH definitions (Safe Implementation of Thrombolysis in Stroke (SITS), National Institute of neurological Disorders and Stroke (NINDS), Heidelberg bleeding classification).
Conclusions Endovascular therapy in patients with an elevated INR is safe and efficient. Patients with an INR of 1.2–1.7 may be treated with combined IV thrombolysis and endovascular therapy.
Statistics from Altmetric.com
Oral anticoagulation increases the risk of intracranial hemorrhage.1 ,2 When IV thrombolysis (IVT) as acute stroke treatment was first introduced, patients with oral anticoagulation were therefore excluded.3 Because many physicians felt that the exclusion of all patients with oral anticoagulation was too strict, they investigated the risk of patients with a subtherapeutic international normalized ratio (INR) <1.7 undergoing IVT. Two larger trials found that an INR <1.7 is not associated with an increased risk for sICH.4 ,5 Consequently, patients with an INR <1.7 are frequently treated with off-label IVT.
Recently, endovascular therapy (EVT) has been proved to be safe and efficient6 but patients with oral anticoagulation were excluded in all studies.7–11 Whether EVT is safe for patients with elevated INR is unclear. Several small trials found no differences in hemorrhagic complications for patients undergoing EVT while taking vitamin K antagonists (VKA).12 ,13 However, these studies did not focus on EVT conducted with a stent retriever, which is the only technique that was proved to be efficient. One study in which the majority of patients were treated with stent retriever showed that there is no significant increase of symptomatic intracranial hemorrhage (sICH) in patients with oral anticoagulation.14 However, patients taking VKA with an INR <1.7 were excluded, unless they received IVT. Moreover, the risk for sICH in subgroups of patients with a subtherapeutic INR was not analyzed separately. Therefore, considerable uncertainty remains about the risk of EVT in patients with elevated INR.
At our hospital, patients eligible for EVT with an INR <1.7 may receive IVT and additional EVT whereas patients with an INR >1.7 receive EVT only. In our study we aimed to analyze the risk for sICH in patients with an elevated INR. In addition, we investigated whether our standard of combined IVT and EVT for patients with an elevated INR of 1.2–1.7 is safe and efficient.
Patients and methods
The study was approved by the ethics committee of the University of Heidelberg (statement S-502/2012). The requirement for informed consent was waived as the analysis was retrospective.
Consecutive patients were retrieved from our prospective stroke database if they were intended to receive EVT for anterior circulation stroke between 2009 and 2015. We excluded patients receiving anticoagulation other than VKA. Symptom onset-to-treatment time (OTT) was defined by the time from symptom onset to the beginning of EVT. OTT for patients with unknown symptom onset but with a defined last-seen-well time was calculated as described previously.15 Patients were selected for EVT based on an interdisciplinary decision of an experienced neurologist and neuroradiologist. All patients had a large vessel occlusion of the internal carotid artery or the proximal middle cerebral artery and early infarct signs in less than one-third of the territory of the middle cerebral artery. Imaging modalities used for selection were CT, including CT angiography with or without CT perfusion, MRI including MR angiography and perfusion-weighted imaging. Reperfusion was graded by experienced neuroradiologists, blinded to this analysis. The modified Thrombolysis In Cerebral Infarction Score (mTICI)16 was used to quantify reperfusion, and relevant reperfusion was defined as an mTICI score of ≥2b. Disagreements were solved by consensus reading.
INR values were taken from the first blood results after symptom onset, before IVT and EVT. Reference values were defined according to our university laboratory (<1.2). To test whether the findings for the safety of IVT when INR values are elevated are also valid for EVT, we categorized the INR value into three groups: (i) <1.2, (ii) 1.2–1.7, and (iii) >1.7.
Safety outcome was defined as sICH according to the European Cooperative Acute Stroke Study II (ECASS II) definition—that is, blood at any site in the brain and clinical deterioration with an increase in the National Institute of Health Stroke Scale (NIHSS) score of at least four points compared with the lowest value within the first 7 days, or any intracerebral hemorrhage (ICH) leading to death. As the optimal sICH definition is a matter of debate,17 we also included an analysis using the SITS (parenchymal hemorrhage type II within the first 36 hours and increase in the NIHSS score of at least four points from baseline or leading to death), NINDS (any new ICH on follow-up imaging with any clinical deterioration within the first 7 days), and Heidelberg bleeding classifications.17 sICH was scored by two authors. Disagreements were solved by consensus reading.
The modified Rankin scale (mRS) score after 90 days served as efficacy outcome. Favorable outcome was defined as a mRS score of 0–2. In contrast to the large randomized trials, patients with a premorbid mRS >1 were also treated. Patients with a mRS of 3 were defined as having a favorable outcome if the discharge mRS was also 3. Ten patients with a premorbid mRS score of 4 and 5 were also treated. All of them had a higher mRS at 90 days and were defined as having an unfavorable outcome. Both outcome parameters were scored by an experienced vascular neurologist who was blinded to this analysis.
Statistics were performed using Microsoft Excel V.2010 and IBM SPSS V.22. The Mann–Whitney U-test or χ2/Fisher exact test were conducted for univariate analysis, depending on the level of measurement. Multivariable analysis was performed using binary logistic regression. The backward elimination method based on likelihood-ratio tests was run for variable selection, and variables were removed if the related p value rose above 0.10. The α level was set to 0.05 and two-sided p values are reported throughout. Interaction analysis between INR and additional therapy with IVT was carried out using SPSS.
As sensitivity analysis, we introduced an error of ±5% to the INR values and repeated the univariate and multivariate analyses.
Four hundred and forty-five patients were retrieved from the database. We excluded two patients for whom an INR value before treatment was not recorded, four patients who received several endovascular therapies within one hospital stay and four patients who did not have a follow-up CT within 24–36 hours because they were moved to the referring hospital (one patient) or because they received palliative care (three patients). Thus, 435 were available for analysis. Median age was 72 (IQR 62–79) and median NIHSS score was 17 (14–20). Ninety per cent of the patients were treated with a stent retriever. Twenty-seven (6.2%) patients had an INR between 1.2 and 1.7 and 21 (4.8%) had an INR >1.7. In total, 45 (10.3%) patients received oral anticoagulation with VKA. A slightly elevated INR was found in eight patients who were not taking oral anticoagulants. Five of them had an INR of 1.20, two patients of 1.21 and one patient of 1.23. Of the 387 patients with a normal INR, 303 (78%) received combined IVT and EVT, whereas this was the case for 17/27 (63%) patients with an INR of 1.2–1.7 and only 1/21 (5%) patients with an INR >1.7 (this patient had denied taking oral anticoagulants, IVT stopped after the bolus). Ten patients with an INR of 1.2–1.7 were not treated with IVT for the following reasons: borderline INR of >1.6 (four patients), diagnosis of erosive gastritis (one patient), prior ICH (one patient), recent surgery (two patients), time window >4.5 hours (one patient), and stroke within the past 24 hours (one patient). Further baseline characteristics are shown in table 1. Thirty-three (7.6%) patients had sICH and 149 (34.3%) had a favorable outcome.
In univariate analysis of patients with and without sICH, those with sICH more often were men (23 (69.7%) vs 202 (50.2%); p=0.032), were more often treated with tirofiban (8 (24.2%) vs 36 (9.0%); p=0.005), and rarely had a favorable outcome (1 (3.0%) vs 148 (36.8%); p<0.001). However, there was no significant difference between the patients with an elevated INR of 1.2–1.7 (3 (9.1%) vs 24 (6.0%)) or >1.7 (1 (3.0%) vs 20 (5.0%); p=0.664, table 2).
To control for relevant predictors for sICH we performed a multivariable logistic regression including age, gender, NIHSS score on admission, additional treatment with IVT, OTT, elevated blood pressure on admission (>185 mm Hg), antiplatelet medication, relevant reperfusion (mTICI=2b/3), treatment with tirofiban and INR (table 3). Again, elevated INR did not increase the risk for sICH (p=0.590 for INR 1.2–1.7 and p=0.644 for INR >1.7 compared with a normal INR of <1.2). Treatment with tirofiban was the only independent risk factor for sICH (OR=3.01, 95% CI 1.23 to 7.79; p=0.016).
We also performed multivariable analysis of predictors for favorable outcome including age, diabetes, OTT, NIHSS score on admission, elevated blood pressure, treatment with IVT, relevant reperfusion, treatment with tirofiban, and INR as independent variables. Elevated INR was not an independent predictor of favorable outcome (p=0.346 for INR 1.2–1.7 and p=0.221 for INR >1.7 compared with a normal INR of <1.2). Independent negative predictors were age (OR=0.95, 95% CI 0.93 to 0.97; p<0.001), diabetes (OR=0.47, 95% CI 0.23 to 0.94; p=0.033), OTT (OR=0.96, 95% CI 0.92 to 0.99; p=0.018), NIHSS score on admission (OR=0.88, 95% CI 0.84 to 0.93; p<0.001), elevated blood pressure (OR=0.32, 95% CI 0.14 to 0.76; p=0.010), and treatment with tirofiban (OR=0.23, 95% CI 0.08 to 0.63) p=0.004), whereas relevant reperfusion was a positive predictor of favorable outcome (OR=5.37, 95% CI 2.90 to 9.93; p<0.001) (table 4). The interaction analysis showed no significant effect of the combination of IVT and elevated INR.
In a sensitivity analysis, there was no relevant difference from the main analysis in the risk and benefit of EVT in patients with elevated INR (see online supplementary tables SI–SIII).
As there is a controversy about which sICH classification is the most accurate, we repeated the univariate analysis for the INR and the multivariate logistic regression analysis with the same parameters as mentioned above for sICH according to the SITS, NINDS, and Heidelberg bleeding classification. Elevated INR was not associated with an increased risk for sICH according to any definition (see online supplementary tables SIV–SIX).
The main finding of our study is that EVT seems to be safe and efficient in patients with elevated INR owing to oral anticoagulation with VKA. Moreover, an algorithm in which patients with an INR of ≤1.7 may receive IVT combined with EVT and patients with an INR >1.7 receive only EVT does not impose an increased risk for sICH or unfavorable outcome for patients.
Our results support the findings of Rebello et al,14 who also found no increased risk for sICH in patients undergoing EVT in the setting of anticoagulation. In their study, they included 29 patients with VKA. The proportion of patients who were additionally treated with IVT, and the INR of these patients, is not specified. The results were obtained for patients with any anticoagulation, including non-vitamin K antagonist oral anticoagulants (NOAC). Moreover, only 67% were treated with stent retrievers. In contrast, our study included only patients with VKA and we were able to present results for 45 patients taking VKA and treated with EVT using mainly stent retrievers. The method used for EVT is important because the bleeding rate is markedly lower using stent retrievers than with older devices like the Merci retriever.18 Other trials have studied the safety of EVT in patients receiving anticoagulation12 ,19 ,20 not primarily treated with stent retriever. In line with our findings, these trials did not find an increased risk for sICH. However, these trials did not focus on patients treated with VKA and therefore had a smaller number of these patient (28 and 7 patients, respectively).12 ,19 One study did not point out how many patients were treated with VKA as they divided them into those with normal and abnormal hemostasis, including elevated partial thromboplastin time and thrombocytopenia,20 thus preventing comparison with our study population.
Another strength of our analysis is the comparability of baseline variables with the large randomized trials proving the safety and efficacy of EVT using stent retriever.6 Baseline NIHSS was 17 (14–20) in the large trials and identical in our population. Our patients also mainly had M1 occlusions. The proportion of patients with favorable outcome corresponds most closely to the MR CLEAN trial,21 which had the smallest number of patients treated outside the trial. It was therefore more likely that our outcome data would be close to this trial than to a more selected population. Moreover, the proportion with relevant reperfusion is also similar to those of the large trials.6 Median age was slightly higher in our population (68 (55–77) in the large trials and 72 (62–79) in our study).
Even though NOAC have substantial advantages over VKA in atrial fibrillation22–24 and are likely to reduce the number of patients taking VKA, they are not beneficial in all cases. For example, patients with severe renal failure and patients with mechanical heart valves25 do not benefit from NOAC. Therefore, patients with VKA will continue to present with acute ischemic stroke and the need for EVT, emphasizing the relevance of our findings. The study of Rebello et al and our results suggest that it seems to be safe to use a risk-based approach in which patients with an INR <1.7 may be treated with IVT and EVT and patients with an INR >1.7 should only be treated with EVT. Evidence with larger statistical power might be obtained from large international stroke registries.
Limitations of our study include the retrospective and monocentric design. We had to exclude three patients who received palliative care as it could not be stated whether they died from sICH. Even though this is a source of bias, all excluded patients comprised only 2% of the sample size. The included patients with a premorbid mRS of 4 and 5 all had an unfavorable outcome, suggesting that other factors influencing outcome might not exert a strong effect on these patients. However, this would not concern our safety endpoint sICH. Patients were recruited over a relatively long period of time and revascularization techniques (eg, proximal balloon occlusion) as well as imaging techniques for patient selection (perfusion imaging techniques) have evolved considerably, which might have led to distortions. Categorization of the INR and the mRS might have led to an oversimplification. Moreover, even though a large number of patients were included in this study, 45 patients taking VKA might not have led to a sufficient power of the multivariable analysis.
In conclusion, patients with oral anticoagulation and elevated INR values may be treated with EVT after a careful risk–benefit evaluation. For the combination of IVT with EVT, our results support the feasibility of a risk-adapted treatment of patients with an elevated INR. It seems to be safe to treat patients with an INR ≤1.7 with IVT and EVT if there are no other contraindications.
Contributors SM planned and conducted the study. He analyzed and interpreted the data, wrote the initial draft of the manuscript, and revised the manuscript for important intellectual content. He is guarantor. AT conducted the study, analyzed the data, and revised the manuscript for important intellectual content. MAM acquired data for the study and revised the manuscript for important intellectual content. MB revised the manuscript for important intellectual content. PAR acquired data for the study, interpreted the results, revised the manuscript for important intellectual content, and supervised the study. He is guarantor.
Competing interests None declared.
Ethics approval Ethics committee of the University of Heidelberg.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.