(Stroke. 2000;31:1555.)
© 2000 American Heart Association, Inc.
Original Contributions |
From Service de Pharmacologie Clinique EA 643, Claude Bernard University, Lyon, France (C.C., F.B., J.P.B.); Universita di Milano, Istituto di clinica neurologica (L.C.); Austin and Repatriation Medical Centre, Heidelberg, Victoria, Australia (G.A.D.); Clinique Neurologique, Centre Hospitalier Universitaire, Grenoble, France (M.H., A.J.); and the Acute Stroke Unit, Western Infirmary, Glasgow, UK (K.R.L.).
Correspondence to Catherine Cornu, MD, APRET/EZUS, Faculte RTH Laennec, Rue Guillaume Parradin, BP 8071-69376 Lyon Cedex 08, France. E-mail Catherine.Cornu{at}upcl.univ-lyon1.fr
| Abstract |
|---|
|
|
|---|
MethodsIndividual patient data from the Australian Streptokinase Trial (ASK), Multicentre Acute Stroke Trial-Europe (MAST-E), Multicentre Acute Stroke Trial-Italy (MAST-I), and Glasgow Trial (Glasgow) were pooled. Multivariate modeling determined the interaction between treatment effect and delay from symptom onset to treatment, predicted baseline risk, age, concomitant aspirin or heparin use, and the presence of early CT signs on the outcomes of 10-day death, death and disability, or death alone at 3 or 6 months.
ResultsPatients records were pooled (total 1292 patients; streptokinase, n=653, no streptokinase n=639). The subgroup analysis of treatment effect according to delay from symptoms to inclusion shows only a trend toward a better treatment effect with shorter delay, which is not statistically significant for any outcome. Heavier patients in MAST-E may have had a lower (non significant) risk from the fixed dose of 1.5 million units of streptokinase. Concomitant aspirin increased the excess mortality rates in streptokinase-treated patients (17% without aspirin versus 91% with aspirin, P=0.005). The presence of early CT scan signs did not increase the detrimental effect of streptokinase.
ConclusionsFew factors influenced the response to streptokinase. However, earlier administration, lower doses of streptokinase, and avoidance of concomitant aspirin should be considered if further streptokinase trials in acute stroke are planned.
Key Words: cerebral infarction clinical trials meta-analysis streptokinase stroke, acute thrombolysis
| Introduction |
|---|
|
|
|---|
30% in the treatment group,
with no increase in mortality rates.8 When a 6-hour time
window was used with 1.1 mg/kg IV tPA in the European Cooperative Acute
Stroke Study (ECASS), there was a nonsignificant trend toward better
outcomes but at the cost of an increase in mortality
rates.9 When the dose was reduced to 0.9 mg/kg in the
second European Cooperative Acute Stroke Study (ECASS II), there was no
increase in mortality rate and, again, a trend toward improving
outcomes with therapy.10 A meta-analysis of
published data from these trials shows that overall, there is a clear
early hazard (an increase in fatal intracranial hemorrhage),
but despite this there is a reduction in the number of patients dead or
dependent at the end of follow-up (44 fewer dead or dependent patients
per 1000 treated patients).16 Intravenous tPA
looks slightly more promising than the overall estimate and SK rather
less than the overall estimate. tPA was approved in the United States
for treating stroke patients within 3 hours from symptom onset,
according to the strict NINDS protocol (ie, a selected subgroup of
patients, expert teams). The questions now are whether tPA should be
approved in other countries and how to select the
6% of stroke
patients who could benefit from a thrombolytic
treatment. Finally, should a very large-scale trial be performed to
assess the efficacy of thrombolysis in a large spectrum
of stroke patients, with results generalizable to routine practice? To maximize the information accrued from these trials, meta-analyses of individual patient data are required to control for confounding patient characteristics, to explore the possible sources of heterogeneity between trials, and to attempt to identify subgroups of patients who may benefit, or be at particular risk, from thrombolytic therapy.17 18 19 20 To this end, the Thrombolysis in Acute Stroke Pooling Project (TAS-PP) was initiated. Here we report our initial findings, based on 4 trials of SK.
| Subjects and Methods |
|---|
|
|
|---|
Definition of Variables
The efficacy outcomes were all-cause death and death or severe
disability at the end of scheduled follow-up (3 months or 6 months).
The safety outcomes were early death (within 10 days) and
symptomatic cerebral hemorrhages. Data were
combined after completion of the trials, once each study report was
accepted for publication. Each study provided a data file containing
the variables that were required for the project, with a
description of the file. Data were checked for accuracy and
consistency. The efficacy of the randomization and the
completeness of the follow-up were controlled. After pooling, tabulated
data were edited for each study and sent to the trials
representatives for verification. Definitions of
variables were systematically examined for homogeneity between
trials. A common definition was looked for when
heterogeneous definitions were used in each trial. Because
different neurological scales were used for patient assessment at
baseline, severity at entry was assessed on one hand with the item
"consciousness" of each study as a categorical variable, that
is, "alert" or "drowsy/stupor/coma," and on the other hand with
the definition of "severe" patients used in each study. There was
no standard classification for early CT scan abnormalities. All the
studies had a central CT scan committee. The procedures were similar
between studies: The readers were blind to treatment and looked at the
first CT without looking at the second. However, there was neither a
common rule nor common training of the readers, and there was no common
definition of "early signs" or "hypodensity." Since this
variable might explain part of the outcome, we decided to classify
the patients into 2 categories, that is, "patients with early signs
or hypodensity (any sign related to the present stroke) at
entry," according to each studys committee decision, and patients
without such signs. The primary outcome was a combined outcome of death
and severe disability in all 4 studies; it was evaluated at 3 months in
ASK and at 6 months in MAST-E and MAST-I. Severe disability at the end
of follow-up was defined as a Rankin score
3 in MAST-E and MAST-I
(the scale used was slightly different in the 2 trials, but for both
scales,
3 implies dependency) and a Barthel score <90 in ASK, in
which the Rankin scale was not used. Correspondence between the Barthel
and Rankin scales was possible because both scores were available in
MAST-E. The primary outcome "death or dependency" was available for
all patients. For safety evaluation, death after 10 days of follow-up
was chosen.
Strategy for Statistical Analysis
The statistical analysis was conducted according to the
intention-to-treat principle. Baseline population characteristics were
described for each trial and overall in the 2 treatment groups.
Computation of the treatment effect and its variance across the trials
was performed with the Mantel-Haenszel method (fixed-effect model),
which combined on a log scale the trial-specific relative risks with
weights proportional to the inverse of their variance. The test for
heterogeneity of treatment effect between subgroups was
obtained by logistic regression with the trial entered as a dummy
variable, the treatment group, the variable of interest
(subgroup) entered as an ordinal (ranked) variable, and an
interaction term between this variable and the treatment group.
Assessment of the treatment effect according to baseline risk was
conducted as follows. Patients predicted baseline risk of 3-month
death was computed by means of a prediction function established with a
multivariable logistic model established on the data from the
control group. Five categories of risk then were created with
approximately an equal number of patients in each. All analyses
were repeated with the use of multivariable logistic models with
adjustment on severity, drowsiness, atrial fibrillation,
systolic blood pressure, diastolic blood pressure,
age, sex, concomitant aspirin use, with a trial factor as a fixed
effect. Glasgow data were grouped with MAST-E data, which had the most
similar design, because the data of this study were too limited to be
analyzed separately.
Study Organization
The TAS-PP Steering Committee was constituted before all trials
were completed, with representatives from all of the
eligible trials. This Committee is responsible for the protocol and
plan for analysis, access to the common data file, and
publications. The common data file is held at the Data Handling Unit in
the Department of Clinical Pharmacology, EA643, Claude Bernard
University, Lyon, France.
| Results |
|---|
|
|
|---|
The mean (±SD) delay from onset of stroke to randomization was 3.8±1.08 hours (range 0.4 to 7.0), and 304 patients were included within 3 hours (60% from MAST-I, 23% from ASK, and 17% from MAST-E plus Glasgow).
The patient baseline characteristics differed among the trials
(Table
). Patients from MAST-E had
a more severe neurological deficit (51% drowsy or comatose patients in
MAST-E compared with 9% to 15% in the other studies). Early CT signs
of ischemia were present in 64% of patients in MAST-E and
5% in MAST-I.
|
Outcomes
SK was associated with a significantly greater early death (within
10 days) (relative risk [RR] 1.94, 95% confidence interval [CI]
1.55 to 2.42, P<0.001) and 3-month mortality (RR 1.46, 95%
CI 1.24 to 1.73, P<0.001). There was no evidence that the
combined outcome death or Rankin
3 differed between the 2 groups (RR
0.99, 95% CI 0.92 to 1.06, P=0.72). However, there were
significantly more patients with hemorrhagic transformation (RR 1.85,
95% CI 1.58 to 2.17, P<0.001). After adjustment for
drowsiness, atrial fibrillation, systolic blood pressure,
diastolic blood pressure, age, sex, and aspirin use, the
results remained unchanged.
Delay From Onset of Symptoms to Treatment
The histogram of the patient population by time to treatment
indicates that MAST-I was a major contributor to data of patients
included within 3 hours of treatment (Figure 1
). When dichotomized into 0 to 3 hours
and
3-hour epochs, the subgroup analysis of treatment effect
according to delay from symptoms to inclusion shows only a trend toward
a better treatment effect with shorter delay, which is not
statistically significant for any outcome (Figure 2
).
|
|
Clinical Severity of Neurological Deficit at Baseline and
Age
When patients were categorized according to the severity at
baseline (defined by their predicted baseline risk of 3-month death,
see section "Definition of Variables"), the treatment effect
was not shown to be different on death or severe disability, in the
initial "minimal" (RR 0.89 [0.66 to 1.21]) and "very severe"
(RR 0.97 [0.91 to 1.04]) risk categories, on death at 10 days (RR
2.08 [0.91 to 4.74] and 1.63 [1.16 to 2.28], respectively), and
death at 3 months (RR 1.59 [0.79 to 3.19] and 1.20 [0.97 to 1.49],
respectively). Similar results were obtained when patients were
classified according to their age at inclusion.
Concomitant Antiplatelet Treatment
Concomitant aspirin use had a detrimental effect on the risk of
3-month death (P=0.005) (Figure 3
). However, concomitant aspirin use did
not significantly alter the effect of SK on the risk of "death or
severe disability" (P=0.28). Hence, the negative effect of
the combination of aspirin and SK on mortality rates was compensated by
a positive effect on the functional outcome in surviving patients. The
same results were observed when the analysis was extended to
any antiplatelet drug taken during or up to 48 hours after
thrombolysis.
|
Concomitant Heparin Therapy
One hundred sixty-seven (25.6%) patients in the SK group received
heparin within 48 hours along with 237 (37.1%) patients in the control
group. Patients who received SK without heparin appeared to have
a worse outcome. The risk of 3-month mortality in the SK group was
increased by 56% in these patients as compared with a 4% increase in
patients with heparin (P=0.03). The same trend was observed
for early deaths and death/disability at trial completion.
Dose of SK
In all trials, a dose of 1.5 million units of SK was given to
patients randomized in the active group, regardless of the patients
body weight. An indirect analysis of the relation between the
treatment effect and the dose of SK was performed by classifying the
patients according to their body weight, with heavier patients having
received a proportionally lower dose of treatment. Data were available
only in the MAST-E trial. No differences were observed between the
subgroups with regard to the combined outcome death/severe disability
(Figure 4
). However, there was a marked
trend (although not statistically significant, P=0.12) for a
detrimental treatment effect on the risk of early death in patients
with lower body weight and thus a higher relative dose of SK.
|
Early CT Scan Signs
There was a marginally significant trend (P=0.07) for a
greater deleterious effect of SK on early death in patients with no
early signs at admission (RR 2.27 95% CI 1.65 to 3.13), as compared
with the effect in patients with early signs (RR 1.47 95% CI 1.06 to
2.05). There was no evidence that the presence of early signs at entry
interfered with the effect of SK on the risk of 3-month death and of
death or severe disability.
| Discussion |
|---|
|
|
|---|
Although an interpretation of the data from this meta-analysis clearly indicates that intravenous SK at doses of 1.5 million units is deleterious, it has been shown previously that reperfusion at the tissue level is increased, as are arterial recanalization rates based on data with single-photon emission computed tomography and transcranial Doppler sonography, respectively.27 28 29 However, it does seem likely that these changes are induced too infrequently or too late to result in clinical improvement, as evidenced by increased rates of nonnutritional reperfusion also measured by single-photon emission computed tomography.27
Whereas age and severity of initial deficit appeared to have no impact on the deleterious effect of SK on outcome, the concomitant use of aspirin was significantly associated with death. The reasons for this are unclear and, on the basis of evidence from animal model data, somewhat surprising.30 31 In a rat embolic stroke model, Overgaard et al30 showed that whereas thrombolytic therapy significantly and dose dependently reduced infarct volume, improved clinical score, and increased angiographically verified reperfusion, aspirin coadministration conferred no additional benefit. Thomas et al,31 with the use of a rabbit model of embolic stroke, showed that preadministration of aspirin significantly antagonized the rate and extent of tPA-induced clot lysis by up to 70%.
The effect of concomitant use of anticoagulants within 48 hours of stroke onset cannot be disregarded, given the known risk of hemorrhage into acute cerebral infarcts in anticoagulated patients.3 It should be noted that there was a marked imbalance between therapy and control groups for anticoagulant use; perhaps used less frequently in the SK group because of the early development of cerebral hemorrhages. Because of this, the case for a potential interaction between anticoagulant and thrombolytic agent use remains unproved.
The impact of early ischemic changes, seen on initial CT, on the SK effect on outcome was minimal. Indeed, there was a counterinitiative trend toward increased early deaths for those with a normal CT scan and no effect on 3-month death or death and disability combined. The influence of CT changes on the development of cerebral hemorrhage was similar in the treated and control groups in this data set. It is of interest that CT changes were associated with increased hemorrhage rates after thrombolysis in MAST-E,32 ECASS,33 and the NINDS trial,34 although not with poor clinical outcomes at 3 months in the latter.34 There was no common centralized reading panel for all SK trials, and rates of early CT changes varied considerably between trials. This heterogeneity renders interpretation of these findings difficult, unless standardized assessment becomes possible.
The role of SK should also be considered, now that tPA is "proven" to be effective. A recent meta-analysis of SK and tPA trials demonstrates that for the combined end point of death and disability, the SK trials are internally consistent and produce a neutral result; the tPA trials are internally consistent and produce a positive result; and the results of the SK trials are significantly different from the tPA trials.16 Pharmacological properties of SK may contribute to this difference.
SK has antigenic properties and frequently is associated with hypotension when used after acute myocardial infarction: Ten percent of patients in the ISIS-2 trial developed hypotension compared with 2% of the placebo group.35 Recorded hypotension varied from 20% of the treated group in the ASK trial, through 1.9% in MAST-I, to only 0.6% in the MAST-E study. Hypotension may have been a reason for incomplete infusion in the SK -treated patients in the stroke trials reported here (14% SK versus 2% for placebo). Acute blood pressure reductions could theoretically impair blood flow in the penumbral region, are consistently associated with poorer outcome in preclinical models of ischemic stroke, and have been significantly associated with poorer outcome in recent stroke trials.36 37 Also, the prolonged fibrinogen depletion and anticoagulant effect of SK could be responsible for a greater tendency to hemorrhagic transformation of infarction when compared with tPA; alternatively, this could be construed as an advantage.
An important consideration in the search for reasons for increased mortality rates and lack of efficacy among all trials was that the dose of SK may have been too high. Since doses were not individualized and a standard amount of 1.5 million units was given in all cases, a surrogate of dosage was used by inserting individual patient weights in the outcome model. Although this approach was only possible with the MAST-E data, the observed trend toward worse outcomes in lighter patients would tend to support the view that higher equivalent doses of SK contributed to the increase in mortality rates. However, it would be unwise to select a dose for further study at random: Detailed dose-ranging studies would be required to establish an optimal dose and whether adjustment of dose by weight is preferable.
Further trials with SK could be justified, but the cost-effectiveness and risks of pursuing such trials will need to be carefully weighed against the likelihood of improving on the benefits and costs of rtPA. Also, the exact influence on outcome of the delay from stroke onset, the presence of early CT signs, and severity at baseline could be assessed with the use of meta-analysis on individual data from all thrombolysis stroke trials. This knowledge would help designing any further trial that would also primarily consider the issues of drug dose, timing of treatment, and avoidance of concomitant aspirin.
| Acknowledgments |
|---|
Received December 17, 1999; revision received March 30, 2000; accepted March 30, 2000.
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
G. W. Albers, P. Amarenco, J. D. Easton, R. L. Sacco, and P. Teal Antithrombotic and Thrombolytic Therapy for Ischemic Stroke: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition) Chest, June 1, 2008; 133(6_suppl): 630S - 669S. [Abstract] [Full Text] [PDF] |
||||
![]() |
R.I. Aviv, J. Mandelcorn, S. Chakraborty, D. Gladstone, S. Malham, G. Tomlinson, A.J. Fox, and S. Symons Alberta Stroke Program Early CT Scoring of CT Perfusion in Early Stroke Visualization and Assessment AJNR Am. J. Neuroradiol., November 1, 2007; 28(10): 1975 - 1980. [Abstract] [Full Text] [PDF] |
||||
![]() |
P. Khatri, L. R. Wechsler, and J. P. Broderick Intracranial Hemorrhage Associated With Revascularization Therapies Stroke, February 1, 2007; 38(2): 431 - 440. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. M. Wardlaw and O. Mielke Early Signs of Brain Infarction at CT: Observer Reliability and Outcome after Thrombolytic Treatment--Systematic Review Radiology, May 1, 2005; 235(2): 444 - 453. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. Ergin and N. Ergin Is Thrombolytic Therapy Associated With Increased Mortality?: Meta-analysis of Randomized Controlled Trials Arch Neurol, March 1, 2005; 62(3): 362 - 366. [Abstract] [Full Text] [PDF] |
||||
![]() |
G. W. Albers, P. Amarenco, J. D. Easton, R. L. Sacco, and P. Teal Antithrombotic and Thrombolytic Therapy for Ischemic Stroke: The Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy Chest, September 1, 2004; 126(3_suppl): 483S - 512S. [Abstract] [Full Text] [PDF] |
||||
![]() |
J M Wardlaw, T M West, P A G Sandercock, S C Lewis, and O Mielke Visible infarction on computed tomography is an independent predictor of poor functional outcome after stroke, and not of haemorrhagic transformation J. Neurol. Neurosurg. Psychiatry, April 1, 2003; 74(4): 452 - 458. [Abstract] [Full Text] [PDF] |
||||
![]() |
A.K. Gilligan, R. Markus, S. Read, V. Srikanth, T. Hirano, G. Fitt, M. Arends, B.R. Chambers, S.M. Davis, and G.A. Donnan Baseline Blood Pressure but Not Early Computed Tomography Changes Predicts Major Hemorrhage After Streptokinase in Acute Ischemic Stroke Stroke, September 1, 2002; 33(9): 2236 - 2242. [Abstract] [Full Text] [PDF] |
||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Stroke Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 2000 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |