None of the patients received alteplase prior to 60 minutes after symptom
Saposnik G, Young B, Silver B, Di Legge S, Webster F, Beletsky V, Jain V, Nilanont Y, Hachinski V. Lack of Improvement in Patients With Acute Stroke After Treatment With Thrombolytic TherapyPredictors and Association With Outcome. JAMA. 2004;292(15):1839-1844. doi:10.1001/jama.292.15.1839
Author Affiliations: Stroke Program, Department
of Clinical Neurological Sciences, London Health Sciences Centre, University
of Western Ontario, London (Drs Saposnik, Young, Di Legge, Beletsky, Jain,
Nilanont, and Hachinski and Ms Webster); and Department of Neurology, Henry
Ford Hospital, Detroit, Mich (Dr Silver).
Context The focus of thrombolytic therapy in acute stroke has been on favorable
outcome at 3 months. Few studies have analyzed outcome at 24 hours. An early
and reliable prediction of poor outcome has implications for clinical management
and discharge planning.
Objective To evaluate predictors of lack of improvement at 24 hours after receiving
alteplase and their relationship with poor outcome at 3 months.
Design, Setting, and Participants Prospective cohort of consecutive patients with acute stroke who received
alteplase and were admitted to a university hospital from January 1999 to
March 2003. Participants were recruited from 2 academic centers in a major
city in Ontario and 33 affiliated hospitals from 7 counties.
Main Outcome Measures Lack of improvement defined as a difference between the National Institutes
of Health Stroke Scale score at baseline and at 24 hours of 3 points or less.
Poor outcome at 3 months defined by a modified Rankin Scale score of 3 to
5 or death.
Results Among 216 patients with acute stroke who were treated with alteplase,
111 (51.4%) had a lack of improvement at 24 hours. After adjusting for age,
sex, and stroke severity, baseline glucose level on admission (odds ratio
[OR] 2.89; 95% confidence interval [CI], 1.40-5.99 for a glucose level >144
mg/dL [>8 mmol/L]), cortical involvement (OR, 2.66; 95% CI, 1.36-5.20), and
time to treatment (OR, 1.01; 95% CI, 1.0-1.02 for each 1 minute increase in
time to treatment) were independent predictors of lack of improvement. At
3 months, 43 patients (20.2%) had died; of the 170 survivors, 75 patients
(44%) had poor outcomes. After adjusting for age, sex, and stroke severity,
lack of improvement at 24 hours was an independent predictor of poor outcome
(OR, 12.9; 95%CI, 5.7-29.6) and death (OR, 7.5; 95% CI, 2.9-19.6). Patients
with a lack of improvement had longer lengths of hospitalization (14.5 vs
9.6 days; P = .02).
Conclusions Among patients with acute stroke treated with thrombolytic therapy,
lack of improvement at 24 hours is associated with poor outcome and death
at 3 months. Elevated glucose level, time to thrombolytic therapy, and cortical
involvement were predictors of lack of improvement.
Recombinant tissue plasminogen activator (alteplase) is one of the most
efficacious treatments for ischemic stroke patients. Nine years after its
approval, intravenous alteplase continues to be the only approved thrombolytic
therapy for patients within 3 hours of an acute ischemic stroke.1 However,
several issues remain regarding its use. For example, there was no significant
difference in recovery (measured by >4 points improvement on the National
Institutes of Health Stroke Scale [NIHSS]) at 24 hours between the alteplase
and placebo groups in the National Institute of Neurological Disorders and
Stroke (NINDS) alteplase study.
A greater outstanding issue was the identification of accurate predictors
of outcome.2 Age, sex, mean arterial blood
pressure, NIHSS score, and computed tomographic (CT) findings have been identified
as independent predictors of good clinical outcome at 3 months.3- 6 Other
recent studies analyzed predictors for dramatic recovery or major neurological
improvement at 24 hours after receiving alteplase.7- 10 Nevertheless,
the lack of improvement at 24 hours after receiving alteplase has not been
Many predictors have a univariate re lationship with poor outcome,11- 13 but in multivariate
analysis the relationship is less clear. Identifying predictors of lack of
improvement may improve understanding of the clinical factors that influence
the acute recovery and clinical response to alteplase. This perspective can
help predict poor outcome earlier (24 hours after receiving alteplase) than
at 3 months. An early and reliable prediction of poor outcome has important
implications for clinical management and for discharge planning.
We hypothesized that baseline clinical, imaging, or laboratory factors
are associated with lack of improvement at 24 hours; that these factors are
different from those previously described with poor outcome at 3 months or
major neurological improvement (NIHSS score, ≥8 points) within 24 hours;
and that lack of improvement at 24 hours is an independent predictor of poor
outcome at 3 months.
We analyzed consecutive patients with acute stroke who received alteplase.
All patients were admitted to the university campus of the London Health Sciences
Center (London, Ontario) from January 1999 to March 2003.
London is the largest city in southwestern Ontario with a population
of 336 540 (432 450 in the metropolitan area).14 It
has 2 academic medical centers with 24-hour access to CT and magnetic resonance
imaging. These academic hospitals are a referral center for a large part of
Ontario. In addition to serving the local population, the Health Sciences
Centre receives acute stroke referrals from 33 rural hospitals from 7 counties.
This catchment area covers 7800 square miles and serves a population of 1.5
million.14 Details concerning hospital characteristics
and clinical assessment were outlined in 2 previous articles.15,16
Patients with suspected ischemic stroke were seen within 3 hours of
symptom onset at the Health Sciences Centre. Demographic variables, evaluation
and treatment times, admission, and 24-hour NIHSS scores and outcomes were
entered into a database by stroke fellows. The fellows were trained and certified
in administering the NIHSS. For all patients, time of symptom onset was defined
by the time when they were “last seen to be well.” Time of treatment
with alteplase was obtained from the nursing records as onset to needle time
for the alteplase infusion. Time to alteplase was calculated from these data.
The decision to treat with alteplase was made according to the NINDS
protocol.1 Informed consent was given by the
patient or his/her relatives. The ethics review board at the University of
Western Ontario in London determined that approval was not required because
there was no intervention and the study only observed outcomes as part of
standard care at this institution.
Inclusion and exclusion criteria were applied with one major difference
from NINDS: patients with involvement of more than one third of the middle
cerebral artery territory on the baseline CT scan were excluded. Management
after alteplase infusion followed the published guidelines.1 A
control CT scan was performed at 24 hours to determine the presence of new
infarction, cortical involvement, and extension of the ischemic lesion. The
neuroradiologist who interpreted the CT scans was blinded to the neurological
status of the patient.
A routine evaluation was performed to determine the stroke mechanism
and stroke subtype in all patients, which included routine laboratory tests,
electrocardiogram, transthoracic echocardiogram, and carotid ultrasound. Magnetic
resonance imaging, magnetic resonance angiography, CT angiography, transesophageal
echocardiogram, or conventional angiography were performed when appropriate.
Demographic, clinical, routine laboratory, and hemodynamic variables
were recorded at admission. Data on history of risk factors were obtained
from medical records, the patient, or his/her family. Stroke subtype (lacunar
vs nonlacunar) was based on presenting symptoms, physical examination, and
neuroimaging. The presence of cortical involvement, new infarction, or hemorrhagic
transformation was established according to the neuroradiological report of
the control CT scan.
We analyzed the distribution of all variables by both graphic and analytic
methods (frequency distribution by quartiles or quintiles). If the relationship
between a continuous variable and the primary outcome (lack of improvement,
modified Rankin Scale score, or death) was linear, it was kept as continuous.
If the relationship suggested a cutoff, the variable was categorized. When
there was no clear relationship, we used clinical criteria to analyze the
variable. According to the aforementioned analysis, glucose on admission was
dichotomized as less than, greater than, or equal to 144 mg/dL (8 mmol/L).
In our analysis, age, weight, baseline NIHSS, creatinine levels, hematocrit,
white blood cell count, time from stroke onset to treatment (alteplase), and
total alteplase dose were continuous variables.
Previous studies, including the NINDS alteplase stroke trial, used a
difference of 4 points or more on the NIHSS to reflect a clinically significant
improvement beyond interrater variability.3,17 Other
studies showed a good interrater agreement when examiners did not differ in
the total scores by 3 points or more.18- 20 We
defined lack of improvement as less than or equal to a 3-point difference
between the baseline and 24-hour NIHSS score. Outcome measures were evaluated
using the modified Rankin Scale score at 3 months, which is a commonly used
period in studies of thrombolysis for acute stroke. Poor outcome was defined
as a modified Rankin Scale score greater than or equal to 3 or death.
Two exploratory analyses were performed. First, predictors of lack of
improvement at 24 hours were identified. The association between demographic
characteristics, clinical and hemodynamic variables, and lack of improvement
was examined using univariate logistic regression. We also retrospectively
performed survival analyses to represent the number of patients at risk and
the cumulative proportion of patients with lack of improvement by the time
Second, lack of improvement at 24 hours was evaluated as an independent
predictor of poor outcome (modified Rankin Scale score, 3-5 or death) at 3
months in a multivariate analysis. Step-wise multivariate logistic regression,
allowing for entry at the.15 level of significance based on the score statistic,
was used to determine a subset of these variables independently associated
with lack of improvement. Covariates were checked for collinearity and interaction
effects. Discrimination of the model was assessed by the area under the receiver
operating characteristic curve and calibration was assessed using the goodness
of fit test. No adjustment was made for multiple testing. Analysis was performed
using STATA statistical software (version 7.0, STATA Corp, College Station,
Tex). P<.05 was considered significant.
From January 1999 to March 2003, 1214 patients with acute stroke were
admitted to the Health Sciences Center. During that period, 219 patients (18%)
were treated with intravenous alteplase. Three patients were excluded because
of missing 24-hour outcome data. Of the 216 remaining patients, 111 (51.4%)
had lack of improvement at 24 hours after receiving alteplase (Table 1). Three patients (1.4%) were lost to follow-up.
There were no statistically significant differences in demographic variables,
geographic location of symptom onset (London vs other), risk factors, previous
medication, baseline NIHSS score, and alteplase dose between both groups (lack
of improvement vs improvement; Table 1).
There was no statistically significant difference between the baseline NIHSS
score as a categorical variable (based on clinical categories 0-6, 7-15, ≥16)
and lack of improvement.19,21 Median
change in the NIHSS score was 7 in the group with improvement and 1 in the
group with lack of improvement.
The mean time from symptom onset to arrival in the emergency department
was 83 minutes (median, 70 minutes). The mean time from symptom onset to treatment
was 157 minutes (median, 160 minutes). In the overall sample, only 4.1% of
patients were treated within 90 minutes of stroke onset. The overall asymptomatic
and symptomatic hemorrhage rates at 36 hours were 10.4% and 4.1%, respectively.
Five patients (2.3%) with symptomatic intracranial hemorrhage died. Forty
patients (18%) were treated outside the 3-hour time window. There was no statistically
significant difference between the presence of intracranial hemorrhage and
the time window (within or outside 180 minutes of symptom onset; P = .56).
The presence of hyperglycemia (glucose level >144 mg/dL [>8 mmol/L]),
new infarction, cortical involvement, and time to treatment with alteplase
were independent predictors of lack of improvement in the univariate analysis
(Table 1). In logistic regression analysis,
glucose level (odds ratio [OR], 2.89; 95% confidence interval [CI], 1.40-5.99),
presence of cortical involvement (OR, 2.66; 95% CI, 1.36-5.20), and time to
treatment with alteplase (OR, 1.01; 95% CI, 1.00-1.02 per 1-minute increase)
were independent predictors after adjusting for age, sex, and stroke severity
(Table 2). For each minute increase
in time to treatment, the OR for lack of improvement is 1.01. The longer the
time to treatment, the higher the probability of lack of improvement (Figure).
Lack of improvement was present in all 9 patients with symptomatic intracranial
hemorrhage. In this small group, it is difficult to determine if lack of improvement
was present before or after the development of the intracranial hemorrhage.
Therefore, the clinical deterioration secondary to the intracranial hemorrhage
and concomitant lack of improvement were probably nonindependent events. There
was not a statistically significant difference in the presence of asymptomatic
intracranial hemorrhage or hemorrhagic transformation between patients with
and without lack of improvement (9% vs 12%; P = .50).
Patients with lack of improvement had longer lengths of hospitalization (mean
length, 14.5 vs 9.6 days; P = .02). For
each 5-minute increase, the chance of lack of improvement increased by 5%.
The goodness of fit and Hosmer-Lemeshow tests used to evaluate the calibration
of the model were not significant (goodness of fit P = .27;
Hosmer-Lemeshow P = .13), indicating adequate
fitness. There was no evidence of collinearity in inspection of tolerance
warnings, SEs, and correlation matrix. Interaction was explored between age,
sex, NIHSS score, and risk factors, but none achieved statistical significance.
The model in its entirety was an adequate predictor of lack of improvement
with an area under the receiver operating characteristic curve of 0.73.
Overall in-hospital mortality rate was 12.0% for all stroke admissions,
including 28 deaths among those who received alteplase and 146 deaths among
those not treated with alteplase. Among 213 patients evaluated at 90 days,
43 (20.2%) died. Among the same 43 who died, 5 patients (12%) died within
24 hours after receiving alteplase, while 23 died (53%) at 30 days.
Lack of improvement at 24 hours was significantly more likely in patients
who died at 3 months (35/43 [81%] compared with 79/170 [46%]; P <.001).
Lack of improvement at 24 hours was an independent predictor of death after
adjusting for age, sex, and stroke severity (OR, 7.5; 95% CI, 2.9-19.6).
Of the 170 patients who survived to 90 days, 75 (44%) had poor outcome
(modified Rankin Scale score, 3-5). Lack of improvement was also significantly
more likely in patients who had poor outcome (49/75 [65%] compared with 30/95
[32%]; P<.001). Lack of improvement at 24 hours
was an independent predictor of poor outcome after adjusting for age, sex,
and stroke severity (OR, 12.9; 95% CI, 5.7-29.6).
The benefit of intravenous alteplase has been demonstrated in randomized
clinical trials since 1995.1 It is one of the
most efficacious therapies to date for stroke with a number needed to treat
of only 8.22,23 We examined the
clinical relevance of lack of improvement in a prospective cohort of patients
with acute stroke who received alteplase. We found that the presence of cortical
involvement, hyperglycemia, and time to treatment with alteplase were associated
independently with lack of improvement. Age, baseline NIHSS score, presence
of carotid stenosis, prior medication use, presence of vascular risk factors,
or stroke subtype were not associated with lack of improvement at 24 hours.
Alternately, lack of improvement at 24 hours was independently and strongly
associated with an increased likelihood of poor outcome (modified Rankin Scale
score, 3-5) and death at 3 months after stroke. Lack of improvement at 24
hours increased the risk of poor outcome or death at 3 months to more than
7-fold after adjusting for age, sex, and stroke severity. In addition, patients
with lack of improvement had a longer period of hospitalization, which is
a major factor in determining cost.24
Several studies identified predictors of good or poor outcome at 3 months
after thrombolytic therapy, but only a few analyzed outcome at 24 hours after
studies have shown that baseline NIHSS score, age, mean arterial blood pressure,
no history of diabetes, hyperglycemia, and a normal CT are independent predictors
of good outcome (modified Rankin Scale score, 0-1) at 3 months.4 To
the best of our knowledge, lack of improvement at 24 hours, its predictors
and prognostic value, has not been analyzed previously. This information may
be useful for managing patient and family expectations as well as for organizing
the health care system.
In the NINDS cohort, Brown et al10 reported
major neurological improvement (NIHSS score >8) 24 hours after receiving alteplase.
Age and time to treatment with alteplase were associated with major neurological
improvement in a logistic regression model, which showed a moderate predictive
value of 0.66 under the receiver operating curve area. Grotta et al25 investigated the rate of clinical deterioration following
improvement in the NINDS alteplase trial. The rate of clinical deterioration
following improvement was defined by a 2-point increase in the NIHSS score.
To address the relationship between recanalization/reocclusion and the clinical
course, Grotta et al mainly used single positron emission tomographic scan
of cerebral perfusion. Ten percent of patients in the alteplase group had
a rate of clinical deterioration following improvement at 24 hours with no
statistical difference compared with the placebo group—although neurological
worsening did not suggest reocclusion.25 However,
the cutoff used in the NIHSS score cannot be specific enough for reflecting
the parenchyma damage to the brain, which is probably related to additional
conditions (such as fluctuations in blood pressure or concurrent medications).
More convincingly, an early clinical improvement correlated with recanalization
assessed by transcranial Doppler.8,9 In
these studies, dramatic recovery was defined as a total NIHSS score of 0 to
3 points and early recovery was defined as an improvement of 10 points or
more at 2 hours after treatment with alteplase. Both clinical conditions were
present in 22% of patients—75% of whom had good outcomes at 3 months.
Therefore, detection of early collateral flow or restoration of flow in the
penetrating artery territory assessed by transcranial Doppler seems to be
a predictor of dramatic or early recovery after treatment with alteplase.
Recently, the same group studied the clinical response after early recanalization,
which was assessed by transcranial Doppler in 120 stroke patients following
treatment with alteplase. They found that 37% of patients with early recanalization
did not experience clinical changes or worsening 24 hours following treatment
with alteplase and one third achieved a good outcome at 3 months. The authors
theorized that a “stunned brain syndrome” explained the delayed
recovery.26 These studies differ from ours
in that they analyzed neither clinical predictors of lack of improvement at
24 hours nor the prognostic value of lack of improvement at 3 months.
We observed that the presence of cortical involvement as detected by
imaging of the brain obtained 24 hours after treatment with alteplase is a
predictor of lack of improvement at 24 hours. Few studies identified the prognosis
of cortical involvement in patients with acute stroke.27,28 The
cerebral cortex is usually affected in patients with severe strokes; persistence
of cortical signs is a marker of recanalization in the occluded vessels or
absence of collateral flow.8,9 The
presence of cortical signs, such as aphasia, neglect, and anosognosia, has
been associated with poor long-term functional recovery and outcome.29- 31 Although the present
study did not provide direct evidence on vessel state, the observation that
cortical involvement is an independent contributing factor of lack of improvement
can be explained to some extent by the absence of recanalization or poor collateral
In our study, the benefit of treatment with alteplase was demonstrated
with similar results as the original NINDS trial.4,15,22 Acute
stroke management requires a certain degree of expertise. Only a few institutions
are sophisticated enough to provide highly specialized care for patients with
acute stroke, such as 24-hour availability of transcranial doppler, diffusion-perfusion
imaging, or endovascular therapy.32,33 Many
of the current stroke units and hospitals, especially in developing countries,
have only essential personnel and structure (CT scan of the head, carotid
ultrasound, and basic routine laboratory) required to manage patients with
acute stroke.34- 36 These
centers need to use a straightforward approach by extracting the most useful
clinical information for acute management (eg, hyperglycemia, acute ischemic
changes on CT scan, etc). Some early prediction rules of stroke recovery have
been validated, but they require diffusion in weight imaging.37,38
Our study adds a useful perspective concerning early prediction of outcome
by introducing a clinical variable (lack of improvement) that can be easily
measured. Its recognition can contribute to the management of patients with
stroke after thrombolytic therapy with alteplase in terms of early prediction
of outcome. Our report contains 2 exploratory analyses: predictors of lack
of improvement at 24 hours and whether lack of improvement at 24 hours is
a predictor of poor outcome at 3 months. Exploratory analyses are useful for
generating hypotheses. However, an external validation in a large cohort of
patients may be necessary to create a predictive score. It is possible that
the small number and spurious associations of the sample may limit the generalizability
of our findings.
On the other hand, the present study corroborates previous findings
on the deleterious effect of hyperglycemia in acute stroke patients,3,4 and reinforces the concept that the
timely administration of alteplase is a conditional factor for successful
treatment of acute stroke. These data are not useful for making decisions
about treatment with alteplase or withholding care for those with lack of
improvement at 24 hours. Guidelines from the American Heart Association for
the management of patients with acute ischemic stroke should be followed.39
In summary, the results of our study suggest that lack of improvement
at 24 hours is an independent predictor of poor outcome and death at 3 months.
Time to treatment with alteplase, glucose level on admission, and cortical
involvement were independent predictors of lack of improvement. The identification
of these clinical variables at 24 hours after treatment with alteplase may
improve early prediction of outcome at 3 months.
Corresponding Author: Gustavo Saposnik,
MD, Stroke Service, London Health Sciences Centre, University of Western Ontario,
339 Windermere Rd, Office 7-GE5, London, Ontario, Canada N6A 5A5 (firstname.lastname@example.org).
Author Contributions: Dr Saposnik had full
access to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Saposnik, Young.
Acquisition of data: Saposnik, Silver, Webster, Beletsky,
Analysis and interpretation of data: Saposnik,
Di Legge, Nilanont.
Drafting of the manuscript: Saposnik, Di Legge, Hachinski.
Critical revision of the manuscript for important
intellectual content: Saposnik, Young, Silver, Di Legge, Webster, Beletsky,
Jain, Nilanont, Hachinski.
Statistical analysis: Saposnik.
Obtained funding: Hachinski.
Administrative, technical, or material support:Saposnik,
Silver, Di Legge, Webster, Beletsky, Jain, Nilanont, Hachinski.
Study supervision: Young, Hachinski.
Funding/Support: This research was supported
in part by a grant from the Heart Stroke Foundation of Canada given to Dr
Saposnik. The grant was obtained based on competitive applications following
publication of grant advertisements.
Role of the Sponsor: The investigators acted
as the sponsors of the study. The Heart Stroke Foundation of Canada had no
input on the design, access to the data, analyses, interpretation, or publication
of the study.
Acknowledgment: We thank Bart Demaerschalk,
MD, Blaine Foell, MD, José G. Merino, MD, Fali Poncha, MD, Arturo Tamayo,
MD, and Edward Wong, MD, for their contribution in collecting data. We also
appreciate the administrative and material support of Chris O’Callaghan,
Cheryl Mayer, Connie Frank, Kimberley Hesser, Eva Newhouse, and Maidy Keir
at the London Health Sciences Centre in London, Ontario.