Risks of Intracranial Hemorrhage Among Patients With Acute Ischemic Stroke Receiving Warfarin and Treated With Intravenous Tissue Plasminogen Activator FREE

Quiz Ref IDIntravenous tissue plasminogen activator (tPA) is currently the only effective treatment to improve outcomes for acute ischemic stroke^1- 2; however, treatment with intravenous tPA carries the risk of symptomatic intracranial hemorrhage (sICH). Of patients who receive intravenous tPA for stroke, 2.4% to 8.8% experience this potentially life-threatening complication.^3- 6 Warfarin-treated patients may be at an increased risk of sICH, but the true absolute risk of sICH in this population remains a matter of significant debate, because warfarin-treated patients were excluded from major trials of tPA.^3- 8 Furthermore, observational studies of bleeding risk among warfarin-treated patients receiving intravenous tPA have been small and inconsistent.^9- 13 Based on limited data, current guidelines of the American Heart Association/American Stroke Association (AHA/ASA) allow use of intravenous tPA in warfarin-treated patients, provided that the patient's international normalized ratio (INR) is 1.7 or lower.^1- 2 Without safety data, however, some investigators have expressed concern regarding administration of intravenous tPA to warfarin-treated patients with stroke.^9,12

We accessed the AHA Get With The Guidelines–Stroke (GWTG-Stroke) Registry to evaluate the association of warfarin treatment and sICH among patients with stroke receiving intravenous tPA in routine clinical practice. Our specific goals were to (1) determine whether warfarin-treated patients were at an increased risk of sICH following administration of intravenous tPA for acute ischemic stroke; (2) to examine the association between INR and sICH in warfarin-treated patients; and (3) to estimate the percentage of warfarin-treated patients in current clinical practice who were otherwise eligible to receive tPA treatment but who did not receive it.

The primary data source was the GWTG-Stroke Registry, an ongoing, voluntary, national stroke registry and performance improvement program sponsored by the AHA/ASA. Details of the design and conduct of the GWTG-Stroke Registry have been previously described.^14- 16 To summarize, the AHA/ASA developed the GWTG-Stroke Registry to improve the quality of care and outcomes for patients hospitalized with stroke. The program includes a set of performance measures to quantify the quality of stroke care and uses the results of those measures to guide the quality improvement efforts at participating centers. As part of this effort, trained hospital personnel are instructed to use an Internet-based patient management tool (Outcome Sciences Inc, Cambridge, Massachusetts) to collect patient-level data on acute stroke care provided to patients enrolled in the GWTG-Stroke Registry. The eligibility of each admission is confirmed through chart review.

Standardized data collection includes patient demographics, medical history, diagnostic testing, brain imaging, in-hospital treatment, and outcomes. The GWTG-Stroke Registry collects data on medications prior to admission, including use of antithrombotics, type of antithrombotic (eg, antiplatelet agent or anticoagulant), and INR values at presentation, using a new version of the data collection form made available in April 2009. The validity and reliability of data collection in the GWTG-Stroke Registry database has been reported.¹⁷ Outcome Sciences Inc serves as the data collection and coordination center for GWTG-Stroke. The Duke Clinical Research Institute serves as the data analysis center and has an agreement to analyze the aggregate deidentified data for research purposes.

Our analyses included patients with acute ischemic stroke receiving intravenous tPA in GWTG-Stroke hospitals between April 1, 2009, and June 30, 2011. We excluded patients who had missing information on warfarin use and INR. We further limited our primary analyses to patients presenting with a baseline INR of 1.7 or lower to conform to the AHA/ASA guideline recommendations. After these exclusions, our primary study population consisted of 23 437 patients from 1203 hospitals (Figure 1).

Warfarin treatment was defined as a patient taking warfarin within 7 days of the index stroke admission. The baseline INR results refer to the first measurement after presentation to the hospital. Quiz Ref IDThe primary outcome measure was sICH, defined as intracerebral hemorrhage within 36 hours, documented by computed tomography or magnetic resonance imaging and by the treating physician's notes indicating clinical deterioration attributable to hemorrhage. This definition is based on the criteria for sICH established in the National Institute of Neurological Disorders and Stroke (NINDS) tPA trials.³ Secondary end points included life-threatening or serious systemic hemorrhage within 36 hours, any tPA complication within 36 hours, in-hospital mortality, and discharge to skilled nursing facility or inpatient rehabilitation facility. Any tPA complication includes sICH within 36 hours, life-threatening or serious systemic hemorrhage within 36 hours, or other serious complications.

Means, medians, and percentages were used to describe the distribution of continuous and categorical variables, respectively. Baseline characteristics were compared between patients with and without warfarin treatment by Pearson χ² test for categorical variables and Wilcoxon rank-sum test for continuous variables. Multivariable logistic regression analyses were performed to investigate the relationships between warfarin use and tPA-related complications of (1) sICH, (2) life-threatening or serious systemic hemorrhage, and (3) any tPA complication. These analyses adjusted for baseline demographic and clinical variables associated with sICH risk in intravenous tPA–treated patients with stroke and included age, sex, race/ethnicity (determined using GWTG-Stroke Registry categories and assessed because race/ethnicity has been identified in previous studies as a risk factor for sICH), baseline National Institutes of Health Stroke Scale (NIHSS) score (a measure of neurologic deficits ranging from 0-42, with higher score for greater stroke severity), systolic blood pressure, and blood glucose levels.¹⁸ Similar logistic regression analyses were performed to evaluate the relationship between warfarin use and in-hospital mortality. The mortality model adjusted for age, sex, arrival mode, medical history of atrial fibrillation, coronary artery disease, prior stroke or transient ischemic attack (TIA), diabetes mellitus, dyslipidemia, and NIHSS score.¹⁹ A multiple imputation method was used to impute missing NIHSS data for 2327 patients (9.9%). Our analyses also accounted for within-hospital clustering using a generalized estimating equations approach.

The robustness of our findings was assessed in several ways. First, we performed a sensitivity analysis using a logistic regression model with all aforementioned clinical factors except for NIHSS score. Second, we restricted the study population to intravenous tPA–treated patients with complete NIHSS data (N = 21 110). Third, we performed subgroup analyses according to age (<75 and ≥75 years), sex, and NIHSS score (≤14 and >14). Fourth, given that the population of greatest interest may be those with higher INR levels, we performed a subgroup analysis focused on patients presenting with INRs between 1.5 and 1.7 (n = 269). In addition, we extended our study cohort to the entire subtherapeutic INR and included patients with an INR of 2.0 or lower (n = 23 510).

Because of the possibility of confounding by tPA treatment selection, we analyzed data from all tPA-eligible warfarin-treated patients (n = 3554) who arrived within 3.5 hours (potentially eligible for the 4.5-hour treatment window) with INRs of 1.7 or lower and who had no documented contraindication or warning signs except for warfarin use. We compared clinical factors between warfarin-treated patients treated with tPA vs those not. We estimated the risk of sICH for 2 groups based on risk factors associated with sICH and determined whether patients with high risk for sICH or those with low risk received tPA. For warfarin-treated patients who did not receive tPA, the model predicts the expected risk of tPA-related sICH had they received tPA.

The association of INR and risk for sICH among warfarin-treated patients was examined graphically and tested for linearity. We then conducted a multivariable analysis including INR and the aforementioned other intracerebral hemorrhage risk factors to determine whether any association between INR and sICH risk persisted after risk adjustment. Because INR is an exponential measure of the ratio of a patient's prothrombin time to normal, we repeated the analysis based on log INR.

In a separate analysis, we determined the percentage of patients with ischemic stroke receiving warfarin therapy in the GWTG-Stroke Registry who were otherwise eligible for intravenous tPA but did not receive it. Based on the AHA/ASA guidelines, we considered a warfarin-treated patient eligible for intravenous tPA if he or she presented within 0 to 2 or 2 to 3.5 hours (potentially eligible for the 0-3 or 3-4.5–hour time window) with an INR of 1.7 or lower and no contraindications or if the only documented issue for intravenous tPA was use of warfarin.^1- 2

All statistical analyses were performed using SAS version 9.3 (SAS Institute Inc). All P values are 2-sided, with P < .05 considered statistically significant. Based on sample size and observed sICH rates, our study had more than 84% statistical power to detect a 1% difference in sICH between warfarin-treated and non–warfarin-treated patients.

The institutional review board of the Duke University Health System approved the study; patient informed consent was not required.

Among 23 437 patients receiving intravenous tPA, 1802 (7.7%) were taking warfarin prior to admission. Table 1 and Table 2 report demographic, clinical, and hospital characteristics. Warfarin-treated patients were older (median age, 77 [interquartile range {IQR}, 68-84] years vs 71 [IQR, 59-82] years); more likely to be women; more often had a medical history of atrial fibrillation or flutter, previous stroke or TIA, coronary artery disease, heart failure, or peripheral vascular disease; and were more likely to present with greater stroke severity as measured by NIHSS score (median, 14 [IQR, 8-20] vs 11 [IQR, 6-17]; P < .001 for all). The baseline INR levels were higher in warfarin-treated patients (median, 1.20 [IQR, 1.07-1.40] vs 1.00 [IQR, 1.00-1.10]; P < .001). Nearly 15% (269/1802) of warfarin-treated patients presented with INRs between 1.5 and 1.7. Time from symptom onset to administration of intravenous tPA was similar in both groups (median, 148 [IQR, 120-174] minutes vs 145 [IQR, 115-175] minutes; P = .28).

We found generally similar demographic, medical history, and clinical characteristics among the tPA-eligible warfarin-treated patients who received tPA vs those who did not (eTable). However, patients who received intravenous tPA were less likely to have a history of previous stroke or TIA and more likely to present with greater stroke severity (median NIHSS score, 14 [IQR, 8-20] vs 9 [IQR, 4-16]; P < .001). Based on the sICH risk prediction model, tPA-treated patients had significantly higher predicted risk of sICH compared with patients not treated with tPA (mean predicted sICH rate, 5.9%; median, 5.0% [IQR, 2.8%-8.4%] vs mean predicted sICH rate, 4.4%; median, 3.1% [IQR, 1.9%-6.2%]; P < .001). Therefore, we found no evidence for preferential treatment of warfarin-treated patients with a lower propensity for tPA-related sICH.

Overall, 1107 patients (4.7%) developed sICH after intravenous tPA administration. Warfarin-treated patients had a higher overall unadjusted rate of sICH than did non–warfarin-treated patients (5.7% vs 4.6%, P < .001; unadjusted odds ratio [OR], 1.22 [95% CI, 0.99-1.51]; P = .06) (Table 3). However, after risk adjustment, warfarin use was not an independent predictor of sICH risk (adjusted OR, 1.01 [95% CI, 0.82-1.25]; P = .94).

These relationships between warfarin use and sICH were consistent in sensitivity analysis (Table 4). When NIHSS score was excluded from our multivariable model, we found that warfarin was still not associated with sICH risk (adjusted OR, 1.11 [95% CI, 0.90-1.37]; P = .32). Our results also remained essentially unchanged in patients with complete NIHSS data (1617 warfarin-treated and 19 493 non–warfarin-treated patients; unadjusted sICH rates, 5.7% vs 4.6%; adjusted OR, 1.00 [95% CI, 0.79-1.25]; P = .99). Further adjustment for use of antiplatelet therapy prior to admission did not substantially alter the association of warfarin and sICH (adjusted OR, 1.04 [95% CI, 0.84-1.29]; P = .71). Stratified analyses by age, sex, and NIHSS score found similar results. Warfarin remained unrelated with the risk for sICH in the subgroup analysis of patients with INRs between 1.5 and 1.7 (adjusted OR, 1.32 [95% CI, 0.85-2.04]; P = .21) and in the exploratory analysis of those with INRs of 2.0 or lower (adjusted OR, 1.00 [95% CI, 0.81-1.23]; P = .97).

The rates of life-threatening or serious systemic hemorrhage were similar in both groups (0.9% vs 0.9%, P = .90) (Table 3), although higher unadjusted rates of any tPA complication (10.6% vs 8.4%, P = .001) and mortality (11.4% vs 7.9%, P < .001) were observed in warfarin-treated patients. However, after multivariable adjustment, warfarin use was not associated with life-threatening or serious systemic hemorrhage (adjusted OR, 0.78 [95% CI, 0.49-1.24]; P = .29), any tPA complication (adjusted OR, 1.09 [95% CI, 0.93-1.29]; P = .30), or in-hospital mortality (adjusted OR, 0.94 [95% CI, 0.79-1.13]; P = .50). Among patients who survived, there was no significant difference in rates of discharge to a rehabilitation facility (adjusted OR, 1.09 [95% CI, 0.97-1.22]; P = .16). However, warfarin-treated patients were more likely to be discharged to a skilled nursing facility (adjusted OR, 1.16 [95% CI, 1.02-1.32]; P = .02).

Graphical trends of the association between unadjusted admission INR and sICH rates following administration of intravenous tPA are shown in Figure 2. International normalized ratio was modeled as a continuous variable because it did not violate linearity assumptions (P = .15). After adjustment for risk factors, INR was not statistically significantly associated with sICH (adjusted OR, 1.10 per 0.1-unit increase in INR [95% CI, 1.00-1.20]; P = .06; adjusted relative risk, 1.09 per 0.1-unit increase [95% CI, 1.00-1.19]; P = .05). These findings were similar when we repeated the analysis based on log INR (adjusted OR, 1.12 [95% CI, 1.00-1.26]; P = .06). We also grouped the INR into 3 categories (0.80-1.19, 1.20-1.49, 1.50-1.70) and found no significant trend of sICH and the degree of anticoagulation (P = .16 by Cochran-Mantel-Haenszel trend test).

Among 443 916 patients with acute ischemic stroke in the GWTG-Stroke Registry during in the study period, 47 358 (10.7%) were classified as having taken warfarin prior to admission. More than half of these warfarin-treated patients (25 762/47 358) presented with a baseline INR of 1.7 or lower. The current AHA/ASA guidelines permit administration of intravenous tPA to eligible patients who can be treated in the period of 0 to 3 and 3 to 4.5 hours after symptom onset.^1- 2 Accordingly, we identified 5884 warfarin-treated patients presenting within 2 hours (potentially eligible for the 0-3–hours window) and 1826 warfarin-treated patients presenting between 2 and 3.5 hours (potentially eligible for the 3-4.5–hours window) with INRs of 1.7 or lower. Overall, 2489 of the patients in the 0-hours to 3-hours time window and 1065 of the patients in the 3-hours to 4.5-hours time window were reported to have no contraindications for intravenous tPA or the only documented issue was warfarin use. Among these patients, 32.1% (799/2489) in the 0-hours to 3-hours time window and 87.3% (930/1065) in the 3-hours to 4.5-hours time window did not receive intravenous tPA treatment. Collectively, 48.6% (1729/3554) of patients receiving warfarin who were otherwise eligible were not treated.

Quiz Ref IDIn this large nationwide contemporary registry of patients with acute ischemic stroke, we found that use of intravenous tPA among warfarin-treated patients with a baseline INR of 1.7 or lower was not associated with increased risk of sICH. These findings were robust across several subgroup analyses and risk-adjustment methods. Warfarin use was also not associated with life-threatening or serious systemic hemorrhage, any tPA complication, or in-hospital mortality. Therefore, although the risk of sICH increases marginally with higher INR levels, intravenous tPA appears to be safe in warfarin-treated patients with a baseline INR of 1.7 or lower. Collectively, these findings provide empirical support for current AHA/ASA guideline recommendations and confirm the safety profile of intravenous tPA in warfarin-treated patients with INRs of 1.7 or lower in routine clinical practice.

Symptomatic intracranial hemorrhage is the most feared complication of thrombolysis for acute ischemic stroke. The NINDS tPA study and the European Cooperative Acute Stroke Study trials (ECASS I-III) excluded patients receiving oral anticoagulant treatment, regardless of INR.^3- 6 The Safe Implementation of Thrombolysis in Stroke–Monitoring Study (SITS-MOST), an observational study to assess the safety profile of intravenous tPA in routine clinical practice, also excluded warfarin-treated patients to conform with the European product license.^7- 8 Although the use of an anticoagulant was not an exclusion criterion in the Standard Treatment With Alteplase to Reverse Stroke (STARS) study (a phase 4 postmarketing study of tPA mandated by the US Food and Drug Administration), the outcome of sICH has not been reported separately in this patient group.²⁰

Despite lack of safety data, intravenous tPA has been used in patients receiving warfarin in clinical practice. To date, only a few studies have investigated the safety of tPA treatment among warfarin-treated patients, with conflicting and inconclusive results.^9- 13,21 Two single-center studies involving a total of 27 warfarin-treated patients reported a 6- to 10-fold increased risk of intracranial hemorrhage in patients with preadmission use of warfarin treated with intravenous tPA.^9,12 In contrast, observational studies from Canada, Finland, Korea, and Switzerland (intra-arterial therapy) did not find significantly increased risk of intracranial hemorrhage associated with preadmission warfarin use.^{10- 11,13,21} However, patterns of stroke care and administration of tPA in the United States may differ from patterns in other nations—especially those with non–United States labeling. Importantly, all of these studies were limited to small samples that included a total of 247 patients receiving oral anticoagulants.

Our study represents the largest clinical experience of the safety of intravenous tPA in warfarin-treated patients who meet clinical guideline eligibility criteria. Among 23 437 patients with ischemic stroke treated with intravenous tPA, 1802 had taken warfarin before stroke onset. The overall sICH rate (4.7%) in this study is slightly lower than pooled results from randomized controlled trials (7.7%) but comparable to rates reported in nonselective patient populations (5.2%).^22- 23 Although the unadjusted incidence of sICH was higher in warfarin-treated patients than in non–warfarin-treated patients, these findings appear to be ascribed to the difference in risk profiles between the 2 groups. Quiz Ref IDAge and stroke severity are strong predictors of sICH.^24- 25 Warfarin-treated patients were significantly older, and stroke severity assessed by the NIHSS score also tended to be greater in these patients compared with those without prior warfarin use. After adjusting for these factors in the multivariate analysis, warfarin treatment was no longer a significant predictor of sICH. Another explanation may be related to the intensity of anticoagulation. Studies have shown that the bleeding risk of warfarin is associated with the intensity of anticoagulation, as reflected by the INR.²⁶ Although INR levels were higher in warfarin-treated patients compared with non–warfarin-treated patients (median, 1.2 vs 1.0), the baseline INR was below the therapeutic range (2.0-3.0 or 2.5-3.5), which may explain the low incidence of sICH in our study population.

Our findings also suggest several new directions for outcomes-based stroke research. The NINDS tPA trial suggested a net benefit of intravenous tPA, even after accounting for increased sICH risk among tPA-treated patients.³ This benefit could be neutralized at some rate of sICH. Although our study supports the safety of intravenous tPA for warfarin-treated patients within the guideline recommendation, it remains unclear how high the INR value could be for this safety to hold. The subgroup analysis of patients with INRs between 1.5 and 1.7 and the exploratory analysis of INRs of 2.0 or lower suggests that intravenous tPA appears to be safe in this population; however, this finding should be interpreted with caution. There were 269 warfarin-treated patients with INRs of 1.5 to 1.7 with 21 sICH events (7.8%) and only 33 patients with subtherapeutic INRs greater than 1.7 with 1 sICH event (3.0%). Further study is warranted to clarify the effectiveness and safety of intravenous tPA for patients beyond the guideline recommended INR range.

In addition, new oral anticoagulants such as dabigatran (a direct thrombin inhibitor) and rivaroxaban (a direct Xa inhibitor) have recently been approved by the US Food and Drug Administration as an alternative to warfarin for the prevention of stroke and thromboembolic disease in patients with nonvalvular atrial fibrillation.^27- 29 Yet there is little experience with and no guidelines for the use of intravenous tPA in this population.^30- 31 Further study is needed to provide guidance on thrombolytic therapy for patients who developed ischemic stroke while taking new oral anticoagulants.

Quiz Ref IDThis study also found that many patients with acute ischemic stroke who were eligible for intravenous tPA did not receive this treatment. Our study found that up to 48.6% of warfarin-treated patients who met AHA/ASA guidelines eligibility criteria did not receive intravenous tPA. In these cases, physician concern about serious adverse effects is one of the main obstacles preventing tPA use.³² Each year, nearly 800 000 new or recurrent strokes occur in the United States.³³ Based on the prevalence of warfarin use in the GWTG-Stroke Registry, we estimate that at least 2400 warfarin-treated patients with ischemic stroke who presented with a baseline INR of 1.7 or lower are otherwise eligible and currently do not receive tPA.

There are several issues to consider in interpreting the results of our study. First, this was a retrospective observational analysis, and treatment selection may bias outcome comparisons. However, among measured potential confounders, we found that patients receiving tPA had greater stroke severity and had higher predicted risk of sICH than tPA-eligible warfarin-treated patients who did not receive tPA. Thus, it could be argued that selection bias is more likely to be against warfarin-treated patients. Second, NIHSS score, a critical determinant of stroke outcomes,³⁴ was missing in 10% of our cohort. Consequently, we performed multiple imputation to fill in missing data with plausible values. Importantly, our findings were consistent for the entire cohort, the subgroup analyses that excluded NIHSS missing data, and the stratified analyses by NIHSS groups.

Third, baseline INR was determined as the first measurement on presentation to the hospital. We were unable to verify whether it was before or after the tPA treatment. Although plausible, the latter seems less likely, because clinical guidelines require an INR result before giving tPA in patients receiving warfarin or in those whose use of anticoagulants is not known.¹ A related issue is the definition of sICH. Computed tomography or magnetic resonance imaging scans were interpreted locally and not centrally adjudicated. The GWTG-Stroke Registry does not have the actual images for review; therefore, we cannot determine the hemorrhage type according to the commonly used ECASS grading system.⁵ Nonetheless, our overall sICH rate (4.7%) is comparable to rates reported in nonselective patient populations outside of randomized controlled trials (5.2%).²³ However, it should be noted that because various definitions are used, it might be difficult to compare our incidence of sICH with incidences from other studies. Fourth, the GWTG-Stroke Registry did not have information on causes of death, so that deaths related to tPA complications could not be determined.

Last, the GWTG-Stroke Registry is a voluntary program. Hospitals participate based on their level of interest in quality improvement in stroke care and their capacity to fulfill the requirements. Therefore, the generalizability of our findings to non–GWTG-Stroke Registry hospitals remains to be established. Regardless, the GWTG-Stroke Registry is the largest stroke registry in the world, representing nearly 25% of all ischemic stroke cases in the United States in 2008.¹⁶ In parallel with increasing rates of hospital participation, this number is expected to increase in later years. Thus, our findings are likely to represent routine clinical practice.

In summary, this study represents the largest clinical experience of the safety of thrombolysis in warfarin-treated patients with acute ischemic stroke who meet clinical guideline eligibility criteria. The use of intravenous tPA among warfarin-treated (INR ≤1.7) patients with ischemic stroke was not associated with increased sICH risk compared with the use of intravenous tPA among non–warfarin-treated patients in routine clinical practice. We found the potential for substantial undertreatment, because up to 50% of warfarin-treated patients who might have been eligible for reperfusion therapy did not receive intravenous tPA. These data provide empirical support of current AHA/ASA guideline recommendations and may help support future stroke quality improvement efforts.