Background and Purpose We sought to identify variables associated with intracerebral hemorrhage in patients with acute ischemic stroke who receive tissue plasminogen activator (t-PA).
Methods We performed subgroup analyses of data from a randomized, double-blind, placebo-controlled trial of intravenous t-PA administered to stroke patients within 3 hours of onset. Using multivariable regression modeling procedures, we assessed the relationship of baseline and after-treatment variables with symptomatic and asymptomatic intracerebral hemorrhage during the first 36 hours after treatment.
Results Overall, t-PA–treated patients had an increase in the absolute risk of symptomatic intracerebral hemorrhage of 6% and a decrease in the absolute risk of 3-month mortality of 4% compared with placebo-treated patients. The only variables independently associated with an increased risk of symptomatic intracerebral hemorrhage in the final multivariable logistic regression model for the 312 t-PA–treated patients were the severity of neurological deficit as measured by the National Institutes of Health Stroke Scale score (five categories; odds ratio [OR], 1.8; 95% confidence interval [CI], 1.2 to 2.9) and brain edema (defined as acute hypodensity) or mass effect by CT before treatment (OR, 7.8; 95% CI, 2.2 to 27.1). This final model correctly predicted those t-PA–treated patients who would or would not have a symptomatic hemorrhage with only 57% efficiency. In the subgroup of patients with a severe neurological deficit, t-PA–treated patients were more likely than placebo-treated patients to have a favorable 3-month outcome (adjusted OR based on multiple outcomes, 4.3; 95% CI, 1.6 to 11.9). These results were similar for the subgroup with edema or mass effect by CT (adjusted OR, 3.4; 95% CI, 0.6 to 20.7). The likelihood of severe disability or death was similar for t-PA–and placebo-treated patients with these two baseline characteristics.
Conclusions Despite a higher rate of intracerebral hemorrhage, patients with severe strokes or edema or mass effect on the baseline CT are reasonable candidates for t-PA, if it is administered within 3 hours of onset.
We recently reported that t-PA administered intravenously within 3 hours of stroke onset increases the likelihood of normal or near normal function at 3 months compared with placebo—an absolute increase of 11%.1 This benefit was seen even after we accounted for a 6% increase in the absolute risk of symptomatic ICH among t-PA–treated patients. However, the question persists whether some subgroups of stroke patients, even among those treated within 3 hours of onset, have an increased or excessively high risk of symptomatic ICH associated with t-PA. For example, the ECASS investigators recently reported that large regions of hypodensity (≥33% of the middle cerebral artery territory) were associated with a higher risk of poor outcome and ICH in patients who received t-PA within 6 hours of onset.2
The purpose of this exploratory analysis was to identify baseline and after-treatment variables associated with symptomatic and asymptomatic ICH in a multicenter randomized study of t-PA for acute ischemic stroke.
Subjects and Methods
Full details of the design and methods have been provided elsewhere.1 In short, the NINDS t-PA Stroke Trial was a sequence of two independent randomized, double-blind treatment studies of t-PA administered within 3 hours of the onset of ischemic stroke compared with placebo. The primary focus of the first study was evidence of clinical activity for t-PA as measured by change in the NIHSSS3 from baseline to 24 hours.1 The primary hypothesis of the second study was that there would be a consistent and persuasive difference between the t-PA and the placebo groups in terms of the proportion of patients who recovered with minimal or no deficit 3 months after treatment.1 However, parts 1 and 2 of the study were conducted in identical fashion with the same study end points and are combined in the present analyses of hemorrhage risk. Both protocols were approved by the Human Research Committee at each site.
Patients had an ischemic stroke with a clearly defined time of onset, a neurological deficit on the NIHSSS (a measure of neurological function),3 and a baseline CT scan of the brain that showed no evidence of intracranial hemorrhage. Exclusions included prior intracranial hemorrhage; prior stroke or serious head trauma within 3 months; major surgery within 14 days; urinary or gastrointestinal hemorrhage within 21 days; an arterial puncture at a noncompressible site within 7 days; seizure at stroke onset; use of oral anticoagulants or heparin within 48 hours with an elevated partial thromboplastin time, elevated partial thromboplastin time, or prothrombin time >15 seconds or platelet count <100 000; or a serum glucose <50 mg/dL (2.7 mmol/L) or >400 mg/dL (22.2 mmol/L). Patients were also excluded if the systolic blood pressure was >185 mm Hg or the diastolic blood pressure was >110 mm Hg at time of treatment or aggressive treatment to reduce blood pressure was needed to reach the specified limits. Informed consent was obtained in all patients.
Patients received placebo or alteplase (Activase, Genentech), a recombinant t-PA, in a dose of 0.9 mg/kg body wt (maximum, 90 mg), 10% of which was given as a bolus in the first minute followed by delivery of the remaining 90% as a constant intravenous infusion over a period of 60 minutes. Of the 624 study patients, 312 were treated with t-PA and 312 were treated with placebo.
The NIHSSS was obtained at baseline, 2 hours after initiation of treatment, and at 24 hours, 7 to 10 days, and 3 months. A Modified Rankin Scale4 and Barthel Index,5 measures of functional outcome, were obtained at 7 to 10 days and 3 months. The Glasgow Outcome Scale6 was obtained at 3 months only. Vital signs were obtained at admission, before treatment, every 15 minutes for the first 2 hours after treatment, every half hour for the next 6 hours, and then every hour up until 24 hours from treatment onset. Blood pressure elevations >180 mm Hg systolic or >105 mm Hg diastolic after initiation of study medication were treated with intravenous antihypertensive medications, most often labetalol, according to a written blood pressure protocol. Baseline laboratory values1 included a prothrombin time, partial thromboplastin time, complete blood count and platelets, and fibrinogen. Fibrinogen level was also determined 2 hours after treatment and at 24 hours. Tests for fibrinogen levels were performed at a control laboratory. The protocol required that no anticoagulants or antiplatelet agents be administered for 24 hours after stroke onset.
CT Imaging and Analyses
All patients underwent a CT scan before treatment to determine eligibility for the trial and to exclude patients with intracranial hemorrhage. All scans were performed on a variety of third- and fourth-generation CT scanners at 45 hospitals from eight centers. In addition to the baseline CT scan before treatment, patients had a CT at 24 hours, 7 to 10 days, 3 months, and at any time when clinical deterioration was observed. All baseline CT scans were obtained with the use of the standard CT imaging protocol of each study hospital (usually slice thickness of 10 mm), but subsequent CT scans at all hospitals were obtained with a slice thickness of 5 mm. At each participating center, the Principal Investigator was responsible for documentation of any neurological deterioration accompanying intracranial hemorrhage.
The study neuroradiologist (Suresh Patel, MD), blinded as to treatment group and to clinical presentation, evaluated each scan for the presence of hemorrhage without benefit of other scans done in the same patient at another time. A CT scan was considered to have an ICH if it met one of the following classifications. Intracerebral hematoma was defined as CT findings of a typical homogeneous, hyperdense lesion with a sharp border with or without edema or mass effect within the brain. This hyperdense lesion could arise at a site remote from the vascular territory of the ischemic stroke or within but not necessarily limited to the territory of the presenting cerebral infarction. Hemorrhage with an intraventricular extension was considered an intracerebral hematoma. Hemorrhagic cerebral infarction was defined as CT findings of acute infarction with punctate or variable hypodensity/hyperdensity with an indistinct border within the vascular territory suggested by the acute neurological signs and symptoms. A confluent, hyperdense appearance could mimic the appearance of an intracerebral hematoma. Intraventricular hemorrhage was defined as blood appearing within the ventricular system. Subarachnoid hemorrhage was defined as blood appearing within the subarachnoid space. Symptomatic ICH was defined as a CT-documented hemorrhage that was temporally related to deterioration in the patient’s clinical condition in the judgment of the clinical investigator. Symptomatic ICH attributable to study medication was defined, before completion of the randomized study, as symptomatic hemorrhage that occurred within 36 hours from treatment onset. Asymptomatic ICH was defined as CT-documented hemorrhage that was not associated with deterioration in the patient’s neurological condition in the judgment of the clinical investigator.
All baseline CT scans were examined by the neuroradiologist at the Coordinating Center for signs of edema and/or mass effect. Edema was defined before completion of the study as a focal or diffuse area of hypodensity that on visual inspection of the CT image was less dense (darker) than white matter but denser (whiter) than cerebrospinal fluid. The definition of edema referred only to the radiographic appearances and not to whether this region of edema represented a developing infarction. The volume of edema was not quantitated prospectively (eg, >33% of middle cerebral artery territory) as for ECASS.2 Mass effect was defined as effacement of the cerebral sulci, sylvian fissure or other basal cisterns, or compression of the ventricular system. The neuroradiologist who made the determinations of mass effect and edema was blinded to all clinical data, treatment assignment, and the results of subsequent CT imaging.
Data Collection and Statistical Methods
Forty-five variables (Appendix 2) constructed from 35 baseline measures were selected to test for an association with symptomatic ICH and all ICH (symptomatic and asymptomatic hemorrhages combined) within 36 hours of treatment onset. Variables were selected because of a biologically plausible relationship to ICH. All variables were assessed for the linearity in the log odds, leading to classification of the NIHSSS into five categories (0 to 5, 6 to 10, 11 to 15, 16 to 20, >20). We included the age-by-NIHSSS interaction and age-by-admission mean blood pressure interaction in the models being tested because these were the predictors of a favorable outcome at 3 months (eg, patients with older ages and higher baseline NIHSSS were less likely to have a favorable outcome). The time-dependent or after-treatment variables included the blood pressure measures, minor external bleeding or oozing, new weakness (defined as an increase of ≥2 points on the motor part of the NIHSS compared with the baseline NIHSSS), level of consciousness (defined as ≥1 point worsening in the level of consciousness part of the NIHSSS compared with baseline NIHSSS) up to 1 hour before the hemorrhagic event, and serum fibrinogen at 2 hours.
All baseline variables were separately tested by logistic regression for symptomatic ICH and all ICH during the first 36 hours after treatment. Those variables significant at the .2 critical levels in the univariate analyses, similar to a screening probability value suggested by Hosmer and Lemeshow,7 were included in the multivariable model. We chose to include variables in our initial stepwise models based on probability values of .2 to increase our power to detect associations with hemorrhage. When hypotheses are tested, type I errors are of concern, and therefore there is often an adjustment to the critical value for multiple comparisons (making α smaller) to protect against falsely rejecting the null hypothesis. In exploratory analyses such as in this report, the goal is to avoid type II errors. We do not want to overlook any potential relationships, even those that might not become evident until other variables are in the model. Thus, we increase the critical value (making α larger) and increase our power to detect effects. This approach is used by Hosmer and Lemshow for screening of variables to be included in a multivariable model.
A shrinkage coefficient8 was obtained for the first step in the multivariable modeling, which included variables chosen after initial screening. Shrinkage measures the extent to which the model may fit well in this set of patients but may not fit well in a new set of patients. If the shrinkage coefficient falls too low (<0.85), model validation in other groups of patients is necessary. A stepwise logistic regression approach was used for modeling. Once the variables were selected, interactions among the selected variables were included and the model was refit. The highly correlated variables (r>.70) that were selected in the univariate analyses were assessed separately in the first step of the multivariable modeling process along with the other variables selected from the univariate analyses. Among those highly correlated variables with a value of P<.05, only the variable with the lowest probability value was included in the final multivariable model in addition to all other variables with a value of P<.05 or any interactions among these variables with a P<.1. Since there were few ICHs in the placebo group (only two symptomatic ICHs), the final models were developed with t-PA–treated patients only. As a secondary analysis, the final multivariable models were evaluated by including both t-PA–and placebo-treated patients and testing for the treatment effects and treatment-by-variable interactions for both symptomatic and all ICH <36 hours, respectively.
To test for associations between ICH and the time-dependent covariates such as external bleeding, a regression analysis based on the Cox proportional hazard model was used. Data for this analysis included any measurements after baseline and up to 1 hour before symptom onset of ICH for both symptomatic and all ICH. The models were developed with t-PA–treated patients only. The highly correlated time-dependent covariates were evaluated separately in the first step of the multivariable model along with other time-dependent covariates selected from the univariate tests. The final models retained the variables with the lowest probability value among those highly correlated variables with a value of P<.05 as well as any other time-dependent covariates with a value of P<.05.
To assess the predictive ability of the final multivariable models, we computed the sensitivities, specificities, and efficiencies of the two models. Efficiency is the percentage of patients correctly classified as having an ICH or not having an ICH. The highest efficiency is desired since the false-positive and false-negative results are essentially equally serious or damaging.9 For example, patients susceptible to an ICH could die if given t-PA. Patients who are not susceptible to an ICH could have increased disability if t-PA was withheld.
To assess treatment benefit within subgroups of patients identified as being at high risk of ICH, we computed the OR of a favorable outcome based on multiple outcome measures at 3 months after stroke.1 The OR was adjusted for covariates that were associated with a favorable outcome in the companion article concerning efficacy of t-PA.10 We used this approach because it is more powerful than choosing one of the four trial outcomes and summarizes treatment benefit across all four outcome measures. A value >1.0 suggests treatment benefit. However, the small sample sizes within the subgroups lead to wide CIs on the ORs. When the CIs encompass 1.0, the results are equivocal, ie, a larger study could show treatment benefit or harm.
Occurrence, Timing, and Presentation of ICH
Symptomatic ICH occurred in 22 patients within 36 hours from initiation of treatment: 20 (6.4%) in the t-PA–treated patients and 2 (0.6%) in the placebo-treated patients (Tables 1⇓ and 2⇓, Fig 1⇓; P<.001). Four symptomatic ICHs occurred among t-PA–treated patients outside of the vascular distribution of the presenting ischemic stroke (20% of all t-PA–related symptomatic ICHs and 1.3% of all t-PA–treated patients). Of the 10 patients who had fatal hemorrhages, 8 (7 t-PA and 1 placebo patient) had onset of symptoms within the first 12 hours, and all had onset of symptoms within the first 24 hours (Fig 2⇓). The presenting signs and symptoms of symptomatic ICH among the 22 patients included deterioration in the level of consciousness in 20, increased weakness in 16, headache in 5, and increased blood pressure or pulse in 11.
An additional 21 patients (13 t-PA and 8 placebo patients) had asymptomatic ICHs during the first 36 hours (Table 1⇑, Fig 3⇓). These asymptomatic ICHs were detected on the safety 24-hour CT scan that was mandatory for all patients. One patient had a large cerebral infarction without ICH on a posttreatment CT. This patient subsequently developed an ICH after craniotomy and surgical decompression for the large edematous cerebral infarction. This ICH is not included in the following analyses. Five patients had symptomatic ICH between 36 hours and 3 months: 3 (1%) in t-PA–treated patients (days 3, 43, and 64) and 2 (0.6%) in placebo-treated patients (days 4 and day 14). These 5 patients had no evidence of ICH on the protocol CT scan performed 24 hours after onset of treatment and were not included in Table 1⇑ or in subsequent analyses. Hemorrhages that occurred after 36 hours were considered unrelated to treatment and could mask treatment/hemorrhage relationships.
Baseline and Time-Dependent (After-Treatment) Covariates and the Risk of Symptomatic ICH Within 36 Hours of Treatment
Baseline NIHSSS (P=.003), edema or mass effect on the baseline CT (P<.001), age (P=.05), history of atrial fibrillation (P=.20), history of other cardiac disease (P=.13), admission diastolic blood pressure >100 mm Hg (P=.15), and admission glucose >300 mg/dL (16.7 mmol/L, P=.07) were associated with an increased risk of symptomatic ICH among t-PA–treated patients in univariate analysis and were included in the multivariable modeling (P<.20, screening criteria of Hosmer and Lemeshow for model building7 ). Current smoking was associated with a decreased risk (P=.09) and was also included. After these baseline covariates were included into the multivariable model, only the NIHSSS and edema or mass effect on the baseline CT remained significantly associated with an increased risk of symptomatic ICH during the first 36 hours after start of treatment (P<.05; Tables 3⇓ and 4⇓). As seen in Tables 2⇑ and 3⇓, only 3% of the 110 t-PA–treated patients with an NIHSSS <10 had a symptomatic ICH compared with 17% of the t-PA–treated patients with an NIHSSS ≥20. Thirty-one percent of patients with edema or infarct on CT developed symptomatic ICH compared with 6% of patients without CT findings. No interaction was detected between these two baseline variables. When both t-PA–and placebo-treated patients were included in the final model, an effect for treatment with t-PA remained (P<.001), after we adjusted for the covariates in the final model. No variables-by-treatment interactions were detected.
Based on univariate tests, three time-dependent covariates were associated with an increased risk of symptomatic ICH and were included in the multivariable modeling (P<.20): minor external bleeding/oozing (P=.10), systolic blood pressure (P=.07), and pulse pressure (P=.06). For example, 9 (45%) of the 20 t-PA–treated patients who developed a symptomatic ICH had minor external bleeding or oozing compared with 31% of the 292 t-PA–treated patients without symptomatic ICH. In the first step of the multivariable modeling, the systolic blood pressure and pulse pressure were assessed separately with minor external bleeding/oozing because they were highly correlated. However, none of the three variables was significantly associated with symptomatic ICH in the final multivariable model (P<.05).
Baseline and Time-Dependent (After-Treatment) Covariates and the Risk of All ICH (Symptomatic and Asymptomatic) Within 36 Hours of Treatment
NIHSSS (P=.002), edema or mass effect on the baseline CT (P=.002), age (P=.02), race (P=.007), admission systolic blood pressure (P=.18), admission systolic blood pressure >190 mm Hg (P=.19), admission diastolic blood pressure >100 mm Hg (P=.06), admission blood glucose >300 mg/dL (16.7 mmol/L; P=.10), and alcohol problems (P=.14) were associated with an increased risk of all ICH among t-PA–treated patients in univariate analysis and were included in the multivariable modeling (P<.20). Current smoking was associated with a decreased risk and was also included (P=.01).
After inclusion of these 10 baseline variables in the final multivariable model of all ICH, only baseline NIHSSS (OR, 1.6; 95% CI, 1.2 to 2.2) and edema or mass effect on baseline CT (OR, 5.3; 95% CI, 1.5 to 18.3) remained significantly associated with an increased risk of all ICH. In contrast, t-PA–treated patients who smoked had a decreased risk of all ICH compared with nonsmokers (OR, 0.25; 95% CI, 0.08 to 0.77). No variable interactions were detected.
When we added the placebo-treated patients to the t-PA–treated patients in the final model and adjusted for the other covariates, there was a smoking-by-treatment interaction (P=.07). The percentage of placebo-treated patients who smoked who had a symptomatic or asymptomatic ICH (5%) was similar to the percentage of placebo-treated patients who did not smoke (3%). However, the percentage of t-PA–treated patients who smoked who had an ICH (4%) was less than the percentage of patients who did not smoke (13%).
Based on univariate tests, six time-dependent covariates were associated with an increased risk of all ICH among t-PA–treated patients and were included in multivariable modeling (P<.20): the 2-hour fibrinogen (P=.17), minor external bleeding/oozing (P=.01), systolic blood pressure (P=.01), mean blood pressure (P=.15), pulse pressure (P=.01), and pulse pressure >50 mm Hg (P=.04). There were high correlations between systolic blood pressure and mean blood pressure (r=.83) and between systolic blood pressure and pulse pressure (r=.78). When we followed the model-fitting procedure described in “Subjects and Methods,” minor external bleeding or oozing (OR, 2.3; 95% CI, 1.2 to 4.6) and pulse pressure (OR, 1.02/mm Hg; 95% CI, 1.004 to 1.035) remained significantly associated with all ICH during the first 36 hours among t-PA–treated patients in the final model (P<.05).
Predictive Abilities of Two Final Multivariable Models
To assess the predictive ability of the final multivariable models for symptomatic ICH and all ICH (symptomatic and asymptomatic) among t-PA–treated patients, we assumed an incidence rate of symptomatic ICH during the first 36 hours of 6.4% and an incidence rate of all ICH during the first 36 hours of 10.5%, as observed in the trial. The final predictive model for symptomatic ICH had 57% efficiency (95% CI, 51% to 62%), 60% sensitivity (95% CI, 39% to 82%), and 56% specificity (95% CI, 50% to 62%). The final predictive model for all ICH had 72% efficiency (95% CI, 67% to 77%), 68% sensitivity (95% CI, 51% to 85%), and 72% specificity (95% CI, 67% to 78%).
The shrinkage coefficients in the first-step multivariable models were 0.76 for the symptomatic ICH model and 0.75 for the all ICH model, indicating the need for validation of the model in other groups of patients.
Treatment and Outcome of ICH
Of the 22 symptomatic hemorrhages during the first 36 hours among t-PA–and placebo-treated patients, 13 received blood products (fresh frozen plasma, cryoprecipitate, or platelets); 12 of the 13 patients were dead at 90 days. Only one patient had operative removal of the ICH, and he subsequently died.
The 3-month mortality of t-PA–treated patients (17%) was lower than that of placebo-treated patients (21%), although the difference was not statistically significant (P=.30). Even during the first week, the mortality rate of t-PA–treated patients was lower (5%) than that of placebo-treated patients (8%),1 although the difference was not significant (P=.11). Of the 20 t-PA–treated patients with a symptomatic ICH, 15 (75%) were dead at 3 months, 3 had a Rankin of 4 or 5 (moderately severe or severe disability), and 2 had a Rankin of 0 or 1 (no symptoms or no disability despite symptoms, can carry out all usual actions and activities). One of the 2 placebo-treated patients with a symptomatic ICH died on the second day, and the other had a Rankin of 4 (moderately severe disability) at 3 months.
When we compare the entire trial cohort of t-PA and placebo patients, irrespective of whether patients had an ICH, 21% of the t-PA–treated patients with a baseline NIHSSS of >20 had a Rankin of 4 or 5 (4=moderately severe disability, unable to walk or attend to bodily needs without assistance; 5=severe disability, bedridden, incontinent, requiring consistent attention), and 48% were dead at 3 months, compared with 38% of placebo-treated patients with a baseline NIHSSS of >20 who had a Rankin of 4 or 5 and 38% who were dead. When we used all four measures of 3-month outcome, t-PA–treated patients in this subgroup had a significantly greater likelihood of a normal or near normal outcome at 3 months than placebo-treated patients (adjusted OR based on multiple outcomes, 4.3; 95% CI, 1.6, 11.9). For example, of those patients with an NIHSSS >20, 10% of t-PA–treated patients had a Rankin of 0 or 1 at 3 months compared with 4% of placebo-treated patients.
Of the 16 t-PA–treated patients with mass effect or edema on baseline CT, 9 (56%) had a Rankin of 4 or 5 or were dead (1 had severe disability and 8 were dead) compared with 19 (52%) of the 19 placebo-treated patients. T-PA–treated patients with edema or mass effect on the baseline CT were more likely to have a normal or near normal 3-month outcome than placebo-treated patients, although the difference was not significant (adjusted OR based on multiple outcomes, 3.4; 95% CI, 0.6 to 20.2). For example, 25% of t-PA–treated patients with edema and mass effect on the baseline CT had a Rankin of 0 or 1 at 3 months compared with 16% of placebo-treated patients.
The absolute increase in the risk of symptomatic ICH was low for all t-PA–treated patients but increased in the presence of a severe neurological deficit (NIHSSS >20 at baseline) and clear signs of brain edema (defined as acute hypodensity) or mass effect on pretreatment CT scan. However, treatment with t-PA was more likely to result in an excellent neurological outcome at 3 months in both subgroups of patients with a higher risk of symptomatic ICH, and the likelihood of severe disability or death at 3 months was similar for t-PA–and placebo-treated patients. Our results suggest that most patients with severe ischemic strokes or signs of brain edema or mass effect by CT are reasonable candidates for intravenous t-PA therapy if given within 3 hours of onset.
Among ischemic stroke patients treated within 6 hours from onset, the ECASS investigators recently reported that treatment with t-PA and advanced age were associated with an increased risk of parenchymal hematoma, while severity of the initial clinical deficit, presence of early ischemic changes on the baseline CT scan, and treatment with t-PA were associated with an increased risk of hemorrhagic infarction.11 12 Hemorrhagic change was not categorized as symptomatic or asymptomatic in the reported ECASS analysis.12 Of the 620 patients in the ECASS cohort, 52 (8%) had a large region of early ischemic change (>33% of middle cerebral artery territory) on the baseline CT scan.2 11 Of the patients with large regions of early ischemic change on the baseline CT, those who were treated with t-PA were more likely to have a poor outcome and to have an ICH than those patients who received placebo.2 11 Patients with smaller regions of early ischemic changes or normal CT scans were more likely to have a good outcome after t-PA treatment than if they received placebo.2
Our own data indicate that most patients with edema (defined as hypodensity) or mass effect on the baseline CT are reasonable candidates for treatment with t-PA, if it is administered within 3 hours of onset. However, because we did not quantitate the volume of hypodensity or mass effect on the baseline CT according to ECASS, we cannot determine from our present analysis whether patients who are seen within the 3-hour time window, but who have a large region of hypodensity or mass effect on the baseline CT, are suitable candidates for t-PA. To address these issues further, a retrospective reevaluation of all our baseline CT scans, including the extent of early ischemic change according to ECASS criteria, is ongoing. However, in the interim, physicians, on the basis of the ECASS data alone, may reasonably decide against using t-PA in patients with large regions of acute hypodensity on the baseline CT, even if the patient could be treated within 3 hours.
In our final models, we could correctly predict symptomatic ICH with only 57% efficiency. Because there were only two symptomatic ICHs in placebo-treated patients, we were unable to determine whether there is an interaction between t-PA therapy and these two baseline variables with respect to symptomatic ICH. In other words, we do not know whether these two baseline characteristics, irrespective of treatment, increase the risk of symptomatic ICH or whether the risk for patients with these characteristics is increased only when t-PA is administered. Finally, our study was designed with sufficient power to test the overall efficacy and safety of t-PA but not within small subgroups.13 Further research is needed on larger numbers of patients to more precisely identify predictors of symptomatic ICH.
The majority of symptomatic ICHs occurred within the first 24 hours after the start of t-PA therapy, underscoring the need to monitor patients closely during this time period. Based on the clinical presentation of patients who had a symptomatic ICH in the present study, nurses and physicians should look for a decrease in the patient’s level of consciousness, increased weakness, increased systolic blood pressure or pulse pressure, or complaints of new headache or vomiting. These signs may indicate a developing ICH.
For all ICH, symptomatic and asymptomatic, the exploratory analysis of time-dependent, posttreatment variables showed a significant association of ICH with elevated arterial blood pressure, confirming findings of the previous NINDS pilot t-PA study14 as well as large studies of t-PA for myocardial infarction.15 16 17 External bleeding or oozing after treatment was also associated with a significantly increased risk of all ICH.
Cigarette smoking was associated with a decreased risk of all ICH (symptomatic or asymptomatic) in t-PA–treated patients. There was a significant interaction between smoking and treatment with t-PA with respect to the risk of all ICH. Previous large myocardial infarction studies of thrombolytic therapy reported that cigarette smoking was associated with a decreased risk of intracranial hemorrhage in univariate analyses,15 17 but this relationship of cigarette smoking and intracranial hemorrhage did not remain significant in their multivariable analyses. Zlokovic and colleagues18 recently noted that nicotine administration in a rat model of focal cerebral ischemia substantially decreases t-PA antigen of the endothelium in brain capillaries and increases the activity of plasminogen activator inhibitor.18 This animal model of decreased endogenous t-PA activity due to nicotine provides one possible explanation for the decreased risk of ICH in smokers who received t-PA.
Differences in the time from stroke onset until treatment with a thrombolytic agent remain a possible explanation for the low rate of ICH among stroke patients treated with t-PA in the present study. One pilot study of t-PA in stroke patients reported a decreased risk of symptomatic ICH with treatment in less than 6 hours compared with treatment beyond 6 hours.19 20 In the four other large randomized thrombolytic treatment studies, rates of ICH were higher than that in the NINDS t-PA Stroke Trial, but the large majority of patients were treated after 3 hours, and only a handful were treated within 90 minutes.2 21 22 23 For example, 301 of our 624 patients had their CT performed and treatment administered within 90 minutes compared with 12 of the 620 ECASS patients (Dr Werner Hacke, written communication, December 18, 1996). Despite the large number of patients who were treated in the NINDS trial within 90 minutes, we did not detect a lower rate of ICH in that group than in those treated at 90 to 180 minutes.
Finally, a trade-off between an early excess in morbidity and mortality and an improved outcome at 3 months was not observed with t-PA therapy when administered within 3 hours of onset. For example, in the North American Symptomatic Carotid Endarterectomy Trial, the surgical patients had an excess incidence of stroke and death of 2.5% during the first 32 days after randomization compared with the medical group. However, surgically treated patients had much better outcomes at 2 years, the crossover occurring at 3 months.24 In contrast, the mortality in t-PA–treated patients in the present study was lower at 3 months, 1 month, and even at 1 week,1 although the difference did not reach statistical significance. Neurological function at 24 hours, as measured by the median NIHSSS, was superior in the t-PA–treated patients compared with placebo-treated patients. Thus, the argument that t-PA within 3 hours of stroke onset offers patients potential long-term improvement but an up-front higher risk of death and disability is fallacious.25
Selected Abbreviations and Acronyms
|ECASS||=||European Cooperative Acute Stroke Study|
|NIH||=||National Institutes of Health|
|NIHSSS||=||National Institutes of Health Stroke Scale score|
|NINDS||=||National Institute of Neurological Disorders and Stroke|
|t-PA||=||tissue plasminogen activator|
The following persons and institutions participated in the NINDS rt-PA Stroke Trial:
Clinical Centers: University of Cincinnati (n=150)—Principal Investigator: T. Brott; Co-investigators: J. Broderick, R. Kothari; M. O’Donoghue, W. Barsan, T. Tomsick; Study Coordinators: J. Spilker, R. Miller, L. Sauerbeck; Affiliated Sites: St. Elizabeth (South), J. Farrell, J. Kelly, T. Perkins, R. Miller; University Hospital, T. McDonald; Bethesda North Hospital, M. Rorick, C. Hickey; St. Luke (East), J. Armitage, C. Perry; Providence, K. Thalinger, R. Rhude; The Christ Hospital, J. Armitage, J. Schill; St. Luke (West), P.S. Becker, R.S. Heath, D. Adams; Good Samaritan Hospital, R. Reed, M. Klei; St. Francis/St. George, A. Hughes, R. Rhude; Bethesda Oak, J. Anthony, D. Baudendistel; St. Elizabeth (North), C. Zadicoff, R. Miller; St. Luke–Kansas City, M. Rymer, I. Bettinger, P. Laubinger; Jewish Hospital, M. Schmerler, G. Meiros.
University of California, San Diego (n=146)—Principal Investigator: P. Lyden; Co-investigators: J. Dunford, J. Zivin; Study Coordinators: K. Rapp, T. Babcock, P. Daum, D. Persona; Affiliated Sites: UCSD, M. Brody, C. Jackson, S. Lewis, J. Liss, Z. Mahdavi, J. Rothrock, T. Tom, R. Zweifler; Sharp Memorial, R. Kobayashi, J. Kunin, J. Licht, R. Rowen, D. Stein; Mercy Hospital, J. Grisolia, F. Martin; Scripps Memorial, E. Chaplin, N. Kaplitz, J. Nelson, A. Neuren, D. Silver; Tri-City Medical Center, T. Chippendale, E. Diamond, M. Lobatz, D. Murphy, D. Rosenberg, T. Ruel, M. Sadoff, J. Schim, J. Schleimer; Mercy General, Sacramento, R. Atkinson, D. Wentworth, R. Cummings, R. Frink, P. Heublein.
University of Texas Medical School, Houston (n=104)—Principal Investigator: J.C. Grotta; Co-investigators: T. DeGraba, M. Fisher, A. Ramirez, S. Hanson, L. Morgenstern, C. Sills, W. Pasteur, F. Yatsu, K. Andrews, C. Villar-Cordova, P. Pepe; Study Coordinators: P. Bratina, L. Greenberg, S. Rozek, K. Simmons; Affiliated Sites: Hermann Hospital, St. Lukes Episcopal Hospital, Lyndon Baines Johnson General Hospital, Memorial Northwest Hospital, Memorial Southwest Hospital, Heights Hospital, Park Plaza Hospital, Twelve Oaks Hospital.
Long Island Jewish Medical Center (n=72)—Principal Investigators: T.G. Kwiatkowski (6/92-), S.H. Horowitz (12/90-5/92); Co-investigators: R. Libman, R. Kanner, R. Silverman, J. LaMantia, C. Mealie, R. Duarte; Study Coordinators: R. Donnarumma, M. Okola, V. Cullin, E. Mitchell.
Henry Ford Hospital (n=62)—Principal Investigator: S.R. Levine; Co-investigators: C.A. Lewandowski, G. Tokarski, N.M. Ramadan, P. Mitsias, M. Gorman, B. Zarowitz, J. Kokkinos, J. Dayno, P. Verro, C. Gymnopoulos, R. Dafer, L. D’Olhaberriague; Study Coordinators: K. Sawaya, S. Daley, M. Mitchell.
Emory University School of Medicine (n=39)—Principal Investigator: M. Frankel (7/92-10/95), B. Mackay (11/90-6/92); Co-investigators: J. Weissman, J. Washington, B. Nguyen, A. Cook, H. Karp, M. Williams, T. Williamson; Study Coordinators: C. Barch, J. Braimah, B. Faherty, J. MacDonald, S. Sailor; Affiliated Sites: Grady Memorial Hospital, Crawford Long Hospital, Emory University Hospital, South Fulton Hospital, M. Kozinn, L. Hellwick.
University of Virginia Health Sciences Center (n=37)—Principal Investigator: E.C. Haley, Jr; Co-investigators: T.P. Bleck, W.S. Cail, G.H. Lindbeck, M.A. Granner, S.S. Wolf, M.W. Gwynn, R.W. Mettetal, Jr, C.W.J. Chang, N.J. Solenski, D.G. Brock, G.F.Ford; Study Coordinators: G.L. Kongable, K.N. Parks, S.S. Wilkinson, M.K. Davis; Affiliated Sites: Winchester Medical Center, G.L. Sheppard, D.W. Zontine, K.H. Gustin, N.M. Crowe, S.L. Massey.
University of Tennessee (n=14)—Principal Investigator: M. Meyer (2/93-), K. Gaines (11/90-1/93); Study Coordinators: A. Payne, C. Bales, J. Malcolm, R. Barlow, M. Wilson; Affiliated Sites: Baptist Memorial Hospital, C. Cape; Methodist Hospital Central, T. Bertorini; Jackson Madison County General Hospital, K. Misulis; University of Tennessee Medical Center, W. Paulsen, D. Shepard.
Coordinating Center: Henry Ford Health Sciences Center—Principal Investigator: B.C. Tilley; Co-investigators: K.M.A. Welch, S.C. Fagan, M. Lu, S. Patel, E. Masha, J. Verter; Study Coordinators: J. Boura, J. Main, L. Gordon; Programmers: N. Maddy, T. Chociemski; CT Reading Centers: Part A: Henry Ford Health Sciences Center, J. Windham, H. Soltanian Zadeh; Part B: University of Virginia Medical Center, W. Alves, M.F. Keller, J.R. Wenzel; Central Laboratory: Henry Ford Hospital, N. Raman, L. Cantwell; Drug Distribution Center: A. Warren, K. Smith, E. Bailey.
NINDS—Project Officer: J.R. Marler.
Data and Safety Monitoring Committee—J.D. Easton, J.F. Hallenbeck, G. Lan, J.D. Marsh, M.D. Walker.
Genentech Participants—Juergen Froelich, MD, Judy Breed, Fong Wang-Chow.
List of 45 Variables Constructed From 35 Baseline Variables
Age, race, sex, current smoking, alcohol problems, alcohol use in 24 hours before stroke, diabetes, history of hypertension, history of atherosclerosis, history of atrial fibrillation, history of other cardiac disease, prior stroke, baseline stroke subtype as determined only by treating physician at baseline (small vessel versus cardioembolic, small vessel versus large vessel, small vessel versus other), baseline NIHSSS (<5, 6-10, 11-15,16-20, >20), edema or mass effect on baseline CT, weight (actual, ranked), percentage of correct dose (ranked), total dose delivered, admission and baseline mean blood pressures, admission and baseline mean blood pressures >130 mm Hg, admission and baseline systolic blood pressures, admission and baseline systolic blood pressures >190 mm Hg, admission and baseline diastolic blood pressures, admission and baseline diastolic blood pressures >100 mm Hg, admission and baseline pulse pressures, admission glucose >300 mg/dL (16.7 mmol/L), history of hepatic disease, history of malignancy, history of hematologic disease, history of hyperlipidemia, history of prosthetic valve, aspirin use before stroke, heparin use before stroke, time from stroke onset to treatment, calcium channel blocker before stroke, age×NIHSSS interaction, age×admission mean blood pressure.
This study was supported by NINDS grants N01-NS-02382, N01-NS-02374, N01-NS-02377, N01-NS-02381, N0-NS-02379, N0-NS-02373, N0-NS-02376, N01-NS-02378, and N01-NS-02380.
Reprint requests to Joseph P. Broderick, MD, Department of Neurology, University of Cincinnati Medical Center, 231 Bethesda Ave, Cincinnati, OH 45267-0525.
The persons and institutions that participated in this trial are listed in Appendix 1.
Almost all of the investigators of the current study are on the speaker panel for Genentech, Inc. Several of the investigators are consultants for an ongoing study of t-PA administered at 3 to 5 hours after onset that is funded by Genentech (ATLANTIS study). Joseph Broderick is a consultant for Genentech in the ATLANTIS study.
- Received June 11, 1997.
- Revision received August 5, 1997.
- Accepted August 5, 1997.
- Copyright © 1997 by American Heart Association
Von Kummer R, Bozzao L, Bastianello S, Manelfe C, for the ECASS Group. Extent of ischemic brain edema and the response to plasminogen activator in acute hemispheric stroke. Stroke. 1997;28:270. Abstract.
Brott TG, Adams HP Jr, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, Holleran R, Eberle R, Hertzberg V, Rorick M, Moomaw CJ, Walker M. Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989;20:864-870.
Van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJA, van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988;19:604-607.
Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J. 1965;14(2):61-65.
Hosmer DW and Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons; 1989.
Copas JB. Regression prediction and shrinkage (with discussion). J R Stat Soc (Series B). 1983;45:311-354.
Galen RS, Gambino SR. Beyond Normality: The Predictive Value and Efficiency of Medical Diagnoses. New York, NY: John Wiley & Sons; 1975.
The NINDS rt-PA Stroke Study Group. Generalized efficacy of t-PA for acute stroke: subgroup analysis of the NINDS t-PA Stroke Trial. Stroke. 1997;28:2119-2125.
Larrue V, von Kummer R, del Zoppo G, Bluhmki E. Hemorrhagic transformation in acute ischemic stroke: potential contributing factors in the European Cooperative Acute Stroke Study. Stroke. 1997;28:957-960.
Fleming TR. Interpretation of subgroup analyses in clinical trials. Drug Information J. 1995;29:1681S-1687S.
Levy DE, Brott TG, Haley EC Jr, Marler JR, Sheppard GL, Barsan W, Broderick JP. Factors related to intracranial hematoma formation inpatients receiving tissue-type plasminogen activator for acute ischemic stroke. Stroke. 1994;25:291-297.
Gore JM, Granger CB, and Simoons ML, Sloan MA, Weaver WD, White HD, Barbash GI, van de Werf F, Aylward PE, Topol EJ. Stroke after thrombolysis: mortality and functional outcomes in the GUSTO-I Trial. Circulation. 1995;92:2811-2818.
Selker HP, Beshansky, JR, Schmid CH, Griffith JL, Longstreth WT Jr, O’Connor CM, Caplan LP, Massey EW, D’Agostino RB, Laks MM. Presenting pulse pressure predicts thrombolytic therapy-related intracranial hemorrhage: thrombolytic predictive instrument (TPI) project results. Circulation. 1994;90:1657-1661.
Zlokovic BV, Wang L, Kittaka M, Berislav V, Schreiber S, Mark J Fisher E, Ning S. Nicotine enhances focal ischemic stroke in rats. Stroke. 1996;27:166. Abstract.
del Zoppo GJ, Poeck K, Pessin MS, Wolpert SM, Furlan AJ, Ferbert A, Alberts MJ, Zivin JA, Wechsler L, Busse O, Greenlee R Jr, Brass L, Mohr JP, Feldman E, Hacke W, Kase CS, Biller J, Gress D, Otis SM. Recombinant tissue plasminogen activator in acute thrombotic and embolic stroke. Ann Neurol. 1992;32:78-85.
Wolpert SM, Bruckmann H, Greenlee R, Wechsler L, Pessin MS, del Zoppo GJ, and the rt-PA Acute Stroke Study Group. Neuroradiologic evaluation of patients with acute stroke treated with recombinant tissue plasminogen activator. AJNR Am J Neuroradiol. 1993;14:3-13.