Symptomatic Intracerebral Hemorrhage Among Eligible Warfarin-Treated Patients Receiving Intravenous Tissue Plasminogen Activator for Acute Ischemic StrokeIntracerebral Hemorrhage, Warfarin, and tPA FREE

Based on pivotal clinical trials and subsequent postmarketing and community studies,^{1 - 4} intravenous (IV) tissue plasminogen activator (tPA) is a proven effective therapy for acute ischemic stroke (AIS). It results in an increased probability of a favorable clinical outcome compared with placebo despite a higher risk of symptomatic intracerebral hemorrhage (sICH). The risk of sICH is increased with advancing age, higher National Institutes of Health Stroke Scale (NIHSS) scores, hyperglycemia, lower platelet counts, higher pretreatment blood pressure, protocol violations, and early infarct changes on brain imaging.^{5 - 8}

The premorbid use of oral antithrombotic medication in patients presenting with AIS has raised further concerns regarding hemorrhagic risk. Despite a modest increase in sICH in patients with prior aspirin use, tPA may still afford clinical benefit in such patients.^{9 - 10} However, to our knowledge, no published data exist assessing safety of tPA among warfarin-treated patients. Excluded in the National Institute of Neurological Disorders and Stroke and European Cooperative Acute Stroke Study tPA trials, patients taking warfarin at baseline represent a growing cohort of patients in whom the safety and efficacy of tPA are uncertain. Nevertheless, current American Heart Association/American Stroke Association guidelines permit IV tPA use in patients taking oral anticoagulants with a baseline international normalized ratio (INR) less than 1.7.¹¹ We aimed to assess the frequency of sICH among warfarin-treated patients who are administered tPA following American Heart Association/American Stroke Association guidelines.

Between October 1, 2002, and February 28, 2009, consecutive patients with AIS who underwent acute revascularization therapies including IV tPA, intra-arterial tPA, and mechanical embolectomy were enrolled in an acute stroke registry. All patients eligible for IV tPA were treated using a standard protocol adopted from American Heart Association/American Stroke Association guidelines. Throughout the study period, it was standard practice at this institution that warfarin-treated patients were eligible for tPA if their INR was less than 1.7. After 2005, all stroke patients were entered into the stroke database of the Specialized Program of Translational Research in Acute Stroke at Columbia University Medical Center, New York, New York. Standardized data collection included baseline demographic characteristics (age, sex, race/ethnicity), medical history (hypertension, diabetes mellitus, hypercholesterolemia, atrial fibrillation, coronary artery disease, prior stroke or transient ischemic attack, medication use including antiplatelet therapy and warfarin), initial NIHSS score, computed tomographic imaging results, laboratory results (glucose level, INR, activated partial thromboplastin time, prothrombin time), and onset-to-treatment time for IV tPA. Warfarin use was defined as self-report or family-report of current use prior to admission. Using the Trial of ORG 10172 in Acute Stroke Treatment criteria based on the available history and clinical and diagnostic evaluation results, subtype classification for each stroke was determined by 2 stroke neurologists (S.P. and J.Y.C.).¹² In the Specialized Program of Translational Research in Acute Stroke registry, sICH is prespecified as parenchymatous hemorrhage within 36 hours and associated with a 4-point increase in the NIHSS score.

We compared the frequency of sICH among warfarin-treated vs non-warfarin-treated patients using Fisher exact test. Because baseline imbalances between those treated with warfarin and those not treated with warfarin may confound this relationship, we also assessed differences between patients with and without prior warfarin therapy using univariable tests (Fisher exact, χ², Mann-Whitney, and t tests) as appropriate. In an exploratory analysis, we used logistic regression to adjust for relevant covariates and potential confounders. P < .05 was considered statistically significant. All statistics were performed using SPSS version 14.0 statistical software (SPSS Inc, Chicago, Illinois). The study was approved by the local institutional review board.

A total of 211 patients with AIS had acute revascularization therapies and were admitted to our institution between October 1, 2002, and February 28, 2009. Ninety-three patients were excluded because they underwent primary or adjunctive endovascular treatments (intra-arterial thrombolysis, embolectomy, angioplasty, and/or stenting due to contraindication to or following IV tPA administration), and 3 additional patients were excluded because they had tPA administration aborted prior to completion. Of the remaining 115 patients, complete data (including baseline warfarin use, onset-to-treatment time, and coagulation profile) were available in 107 patients, who composed the cohort for this analysis.

The mean age of the cohort was 69.2 years, with 43.9% male and 26.1% white. The median initial NIHSS score was 14 (interquartile range, 7-19) and the median time to treatment with tPA was 140 minutes (interquartile range, 115-162 minutes). The median INR was 1.04 (range, 0.82-1.61). Thirteen patients (12.1%) were taking warfarin at baseline; all had INRs less than 1.7 prior to tPA administration (median INR, 1.21; range, 1.04-1.61). Compared with patients not taking warfarin, those taking warfarin were older (mean age, 67.6 vs 80.6 years, respectively; P = .01), more likely to have atrial fibrillation (19.1% vs 69.2%, respectively; P < .001), and more likely to have a higher initial INR (mean, 1.03 vs 1.21, respectively; P < .001). Other baseline characteristics of the cohort are shown in the Table.

There were 7 cases of sICH (6.5%) and 3 cases of asymptomatic intracerebral hemorrhage (2.8%) within 36 hours following IV tPA administration. The computed tomographic scans of patients with sICH are shown in the Figure. The frequency of sICH was higher among warfarin-treated patients than among patients not taking warfarin (30.8% vs 3.2%, respectively; unadjusted odds ratio = 13.5; 95% confidence interval, 2.6-69.9; P = .004). Patients with sICH compared with those without sICH also had trends for higher initial NIHSS scores (median, 18 vs 13, respectively; P = .06) and longer onset-to-treatment times (mean, 159 vs 137 minutes, respectively; P = .12). In an exploratory logistic regression model adjusting for age, NIHSS score, atrial fibrillation, and initial INR, baseline warfarin use remained strongly associated with sICH (P = .004).

In a retrospective study of consecutive patients with AIS who were treated with IV tPA, we found that baseline warfarin use, despite an INR less than 1.7, increased the risk of sICH. Controlling for some potential confounders, warfarin use remained strongly associated with sICH. Our preliminary data raise potential safety concerns regarding the currently accepted guideline of IV tPA use in warfarin-treated patients presenting with AIS in whom the baseline INR is less than 1.7. Larger prospective studies to confirm our preliminary findings are warranted.

With the estimated prevalence of atrial fibrillation in the United States approaching 3 million and expected to double by 2050, the prevalence of anticoagulant use among patients presenting with AIS is not trivial.¹³ It is also well known that maintenance of the INR in the therapeutic range is difficult, with only approximately 60% achieving this goal.¹⁴ Among patients taking warfarin, a subtherapeutic INR is associated with increased stroke risk, more severe strokes, and larger infarct volumes.^{15 - 16} Thus, many patients with atrial fibrillation may present with ischemic stroke and an INR less than 1.7. Although often treated with IV tPA, this subgroup may be at high risk for sICH.

While there are conflicting data showing that premorbid use of antiplatelet medications increases the risk of sICH following IV tPA,^{9 - 10 ,17 - 19} the risk may be even greater in patients taking oral anticoagulants. Given these potential risks, warfarin-treated patients were excluded from the National Institute of Neurological Disorders and Stroke and all 3 European Cooperative Acute Stroke Study trials.^{1 ,20 - 22} Unlike these trials, the Alteplase Thrombolysis for Acute Noninterventional Therapy in Ischemic Stroke study²³ and the phase 4 postmarketing study of tPA (Standard Treatment With Alteplase to Reverse Stroke Study)² included warfarin-treated patients with INRs less than 1.7. However, the risk of hemorrhagic complications in this subset of patients was not reported.

Several mechanisms, besides chance, may explain the observed increased frequency of sICH among warfarin-treated patients in our study. First, the fibrinolytic effects of tPA may be enhanced by the anticoagulant effects of warfarin, even at subtherapeutic levels. Higher recanalization rates with this combination may lead to a greater rate of reperfusion hemorrhage into infarcted tissue. Warfarin use may also be a marker for patients with cardioembolic stroke in whom hemorrhagic transformation is more common and infarct volume is greater. Second, warfarin maintains its anticoagulant effect for 3 days, on average, following the last dose; therefore, the INR may continue to rise to therapeutic levels in the hours following treatment with IV tPA. Extrapolating from the National Institute of Neurological Disorders and Stroke protocol that mandated antithrombotic therapy be withheld for at least 24 hours following tPA owing to concerns about hemorrhage risk, it is possible that the anticoagulation effect of warfarin may increase during this critical period and lead to sICH. Third, overall fibrinolysis and coagulation status may be insufficiently estimated using the INR, which measures the extrinsic pathway only. Lastly, there are rare case reports of coagulopathy caused by tPA itself,²⁴ an unusual pharmacologic effect that may be exacerbated in patients taking vitamin K antagonists.

There are limitations to this study, including its small sample size and tertiary referral bias toward a “sicker” cohort than admitted to community hospitals. We were also unable to assess the effect of early infarct changes on sICH because not all initial computed tomographic scans were available for review at our institution. Serial INR measurements were not performed in every patient in the hours and days following IV tPA. Therefore, we cannot confirm or refute the hypothesis that subsequent INR elevations to greater than 1.7 contributed to sICH formation. We are also unable to assess whether enhanced recanalization mediates this relationship between warfarin and sICH because angiography was not uniformly performed in each patient. Lastly, because we used the European Cooperative Acute Stroke Study definition of sICH, we are unable to report the prevalence of sICH by other definitions. Given these limitations, our study should serve as a hypothesis-generating report that requires confirmation in larger cohorts. Further analyses including more extensive adjustment for confounding variables in larger data sets may prove useful.

Correspondence: Shyam Prabhakaran, MD, MS, Department of Neurological Sciences, Rush University Medical Center, 1725 W Harrison St, Ste 1121, Chicago, IL 60612 (shyam_prabhakaran@rush.edu).

Accepted for Publication: December 30, 2009.

Published Online: March 8, 2010 (doi:10.1001/archneurol.2010.25).

Author Contributions:Study concept and design: Prabhakaran, Vieira, Stillman, Marshall, and Chong. Acquisition of data: Prabhakaran, Rivolta, Stillman, and Chong. Analysis and interpretation of data: Prabhakaran, Vieira, Rincon, and Chong. Drafting of the manuscript: Prabhakaran and Rivolta. Critical revision of the manuscript for important intellectual content: Prabhakaran, Vieira, Rincon, Stillman, Marshall, and Chong. Statistical analysis: Prabhakaran and Rincon. Administrative, technical, and material support: Rivolta, Stillman, and Chong. Study supervision: Vieira, Stillman, Marshall, and Chong.

Financial Disclosure: None reported.

Funding/Support: This work was supported by grant 5P50NS049060 for the Specialized Program of Translational Research in Acute Stroke from the National Institute of Neurological Disorders and Stroke (Dr Marshall).