| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Review |
1 From the Department of Radiology, Duke University Medical Center, Box 3808, Durham, NC 27710-3808 (J.M.P.); Department of Radiology, University of California, Los Angeles (R.J.); Department of Radiology, Mount Sinai Medical Center, New York, NY (T.P.N.); and Department of Medical Imaging, University of Toronto, Sunnybrook & Women’s College Health Sciences Centre, Toronto, Ontario, Canada (A.J.F.). From the 2000 RSNA scientific assembly. Received April 3, 2002; revision requested June 6; revision received October 17; accepted December 10. Address correspondence to J.M.P.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONINTRODUCTIONIMAGING FINDINGS IN HYPERACUTE...VARIABILITY OF VASCULAR...THROMBOLYTIC THERAPYCONCLUSIONREFERENCES |
|---|
© RSNA, 2003
Index terms: Brain, CT, 10.1211 • Brain, MR, 10.1214, 10.12144 • Brain, infarction, 10.78, 17.78 • Cerebral blood vessels, thrombosis, 173.78, 174.78 • Review • Thrombolysis, 10.1265, 17.1265
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONINTRODUCTIONIMAGING FINDINGS IN HYPERACUTE...VARIABILITY OF VASCULAR...THROMBOLYTIC THERAPYCONCLUSIONREFERENCES |
|---|
The U.S. Food and Drug Administration has approved intravenous recombinant tissue plasminogen activator (tPA) for the treatment of acute ischemic stroke within 3 hours of symptom onset. In this review, we will (a) explain the difficulties that are encountered with commonly used neuroimaging techniques for hyperacute stroke assessment, (b) explore the issues involved in determining mechanisms of stroke and the arterial territory involved in an infarct, and (c) critically assess the methods available to reduce the size of cerebral infarctions in the hyperacute setting.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONINTRODUCTIONIMAGING FINDINGS IN HYPERACUTE...VARIABILITY OF VASCULAR...THROMBOLYTIC THERAPYCONCLUSIONREFERENCES |
|---|
For decades, the role of neuroimaging for acute stroke has been one of exclusion of lesions that mimic ischemic stroke (eg, intracerebral hemorrhage, subdural hematoma, cerebritis, hemiplegic or hemisensory migraine, and causes of focal seizures such as tumors and arteriovenous malformations). Before the introduction of the new treatments, imaging could reasonably be performed within a rather wide time window after symptom onset, because imaging findings did not typically alter stroke therapy.
The introduction of intravenous thrombolysis with tPA has radically changed the role of neuroimaging for stroke evaluation. The notable randomized studies of intravenous thrombolysis treatment have been the European Cooperative Acute Stroke Study (ECASS) trial and the American National Institute of Neurological Disorders and Stroke (NINDS) trial (1,2). Neuroimaging played an important (although different) role in each of these studies. The ECASS trial prescribed that patients with stroke symptoms of less than 6 hours in duration and who did not have identifiable infarction of greater than one-third of the middle cerebral artery (MCA) territory on CT images be considered for randomization for possible treatment with intravenous tPA (1). The ECASS results showed that many cases were admitted to the trial despite large infarctions because of nonrecognition of the subtlefindings of hyperacute infarction. These findings, which had been described years earlier, include decrease in gray matter attenuation relative to that of white matter, sulcal compression, and high attenuation in the affected arteries (3). When an expert ECASS panel classified the CT images as showing less or more than one-third of the MCA territory, the large majority of the most serious hemorrhagic complications were associated with infarcts greater than one-third of the MCA territory and had a poorer outcome (4). For those with small or no visible infarction, the outcome was significantly improved and the treatment was validated. The importance of quick and accurate evaluation of acute stroke patients to determine eligibility for thrombolytic therapy was firmly planted.
The NINDS trial established that intravenous tPA treatment is efficacious if administered less than 3 hours after symptom onset (2,5). Because multiple clinical parameters must be verified before treatment, little latitude in time exists for imaging studies to be performed. The clinical catchphrase 
time is brain
is used by stroke associations, stroke teams, and stroke neurologists to emphasize that time-consuming complex imaging studies and analyses decrease the therapeutic options available to patients and the likelihood of a successful intervention.
The field of stroke imaging has greatly advanced in the past few years, and the clinical utility of new techniques is under investigation. However, the time limitations for imaging mandated by the brief therapeutic window has allowed emphasis on easily accessible rapid imaging techniques (such as CT) to be maintained, rather than less accessible longer techniques (such as MR imaging). Unenhanced brain CT remains the most commonly performed neuroimaging technique prior to the decision to deploy intravenous tPA (6). A well-trained radiologist will often be able to discern the extent of early ischemia if appropriate attention is paid to the unenhanced CT study.
|
IMAGING FINDINGS IN HYPERACUTE STROKE |
|---|
|
TOPABSTRACTINTRODUCTIONINTRODUCTIONIMAGING FINDINGS IN HYPERACUTE...VARIABILITY OF VASCULAR...THROMBOLYTIC THERAPYCONCLUSIONREFERENCES |
|---|
CT Findings
CT findings obtained within the first 3–6 hours of cerebral ischemia, when present, are often subtle. Nonetheless, at most institutions CT remains the initial imaging study for evaluation of acute stroke because it is widely accessible, convenient, has a short imaging time, and is sensitive for detection of hemorrhage. However, advances in CT imaging technology and innovations in image assessment have allowed CT findings to be documented earlier and earlier. For instance, the conspicuity of infarcts on unenhanced CT images can be increased by use of variable window width and center level settings at a workstation (as opposed to printed film with standard CT window settings) to accentuate the gray matter–white matter contrast (8). The distinction between normal brain and edematous tissue is thereby increased.
The important CT findings during the early stages of cerebral ischemia can be classified as (a) mass effect, (b) hypoattenuating appearance of gray matter structures, and (c) presence of one or more hyperattenuating arteries. Any combination of these findings may be present, or all may be absent.
The finding of hypoattenuating gray matter structures is typically seen as a gray matter structure becoming isoattenuating to adjacent white matter structures or, in essence, blurring of the gray matter–white matter junction. One example is the so-called insular ribbon sign, in which the insula (which is composed of gray matter) becomes isoattenuating to adjoining white matter (9). Another example is hypoattenuation of the basal ganglia, which then becomes isoattenuating to adjacent white matter structures such as the internal capsule and the external capsule (Fig 1). Examples of early mass effect include narrowing of the sylvian fissure (in MCA infarcts) or loss of cortical sulci (Fig 2). The hyperattenuating artery sign is thought to represent stasis of flow due to arterial thrombus; this sign is most frequently seen in MCA thrombosis but can be seen in any cerebral vessel (3,10). Unfortunately, normal arteries relatively often appear hyperattenuating, and this sign should be carefully interpreted in light of clinical history and other CT findings.
|
|
|
|
|
|
|
After the onset of cell death (cytotoxic edema), mechanisms for maintaining steady-state proportions of intracellular and extracellular water are altered, and the proportion of water in the intracellular space increases. The overall rate of microscopic water motion within such tissue is diminished. Therefore, tissues containing cells undergoing cytotoxic edema appear bright on diffusion-weighted images. High signal intensity on diffusion-weighted images can be seen in the 1st few hours after stroke onset (ie, within the generally accepted time frame for thrombolysis), at a time when T2-weighted images still show a normal appearance (Fig 4) (12). Results of animal studies have shown that abnormal signal intensity on diffusion-weighted images can be seen within minutes after the onset of cerebral ischemia (12). However, because cerebral ischemia does not occur under controlled conditions in humans (such as is the case in animal experiments), the precise time of onset of abnormal signal intensity in humans is not known with certainty. In addition, in many cases of hyperacute stroke in which hyperintense signal is already present on T2-weighted images, diffusion-weighted imaging better defines the size of the ischemic region. Although diffusion-weighted ischemic changes are generally considered to be permanent (and therefore reflect infarction) in clinical studies, recent evidence has shown that in some cases such changes can be reversed with prompt treatment. In one study in which large artery recanalization was achieved by using tPA within 6 hours of symptom onset (13), diffusion-weighted image abnormalities were seen to substantially decrease in a number of patients within 9 hours after thrombolysis. In about half of patients, however, a secondary increase in the size of the diffusion-weighted imaging abnormality was seen within 1 week.
|
|
|
|
|
Another MR perfusion imaging parameter that shows preliminary promise is flow heterogeneity (16). The capacity to alter the heterogeneity of blood transit times (also referred to as flow heterogeneity) is believed to be a major function of the cerebrovascular autoregulatory system. In normal brain tissue, probability density functions of relative flows show a distribution of values that are skewed toward high capillary flow velocities. In animals, decreases in cerebral perfusion pressure are associated with loss of high-flow components (17). Recently, investigators have shown that increases in mean transit time are associated with loss of the high-flow-velocity components and produce a resultant homogenization of capillary flows, which was predictive of final infarct size (16).
It is generally accepted that tissue that is seen to have abnormalities on both perfusion and diffusion-weighted images has already undergone infarction (ie, irreversible ischemia). However, tissue that is seen to have perfusion abnormalities but normal diffusion imaging properties is thought by many investigators to represent reversible ischemia. In particular, tissue showing such a diffusion-perfusion mismatch is thought by many investigators to represent the so-called ischemia penumbra—that is, the region of decreased perfusion that is potentially reversible because it is above the level critical for maintenance of the Na+ K+ -ATPase pump (18). This penumbra has long been sought as a potential therapeutic target; whether this mismatch represents a therapeutic target is a matter of active investigation. However, this information is potentially valuable to the treating physician because it helps determine the risk-benefit ratio in a particular patient. For example, a patient with no (or a very small) degree of mismatch is considered unlikely to benefit from thrombolytic therapy. In such a patient, the ratio of benefit (clinical recovery) to cost (ie, complications of therapy) would generally be considered to be low.
Recently, hemodynamic CT imaging has become available, which may provide the hemodynamic information needed for assessment of hyperacute stroke with the use of CT, rather than MR, imaging. Early results with this technique have shown that it is sensitive for detection of early cerebral ischemia (19). Postprocessing of CT perfusion imaging data, like that for MR perfusion imaging data, is short (ie, can be performed in less than 2 minutes) and does not substantially delay decision making in the hyperacute stroke setting. Results of early investigations of CT perfusion imaging (20) suggest that cerebral blood volume deficits may indicate regions of irreversible hemodynamic deficit (in a manner similar to that provided by diffusion-weighted MR images). Furthermore, cerebral blood volume deficits may predict minimal final infarct size (20). On the other hand, some investigators (21) believe cerebral blood volume and mean transit time deficits on CT perfusion images may represent both infarcted tissue and surrounding tissue that is likely to proceed to infarction if no therapy is provided (ie, the so-called ischemic penumbra).
|
VARIABILITY OF VASCULAR DISTRIBUTIONS AND ETIOLOGY OF BORDER-ZONE INFARCTIONS |
|---|
|
TOPABSTRACTINTRODUCTIONINTRODUCTIONIMAGING FINDINGS IN HYPERACUTE...VARIABILITY OF VASCULAR...THROMBOLYTIC THERAPYCONCLUSIONREFERENCES |
|---|
Difficulties in Determination of Arterial Territories Affected by Infarction
To determine whether an individual has sustained a border-zone infarction, one must be able to determine the territories supplied by the major intracerebral arteries. However, initial indications that one could reliably use the anatomic locus of the infarction to predict the involved vessel are now open to question. It is apparent that identification of specific arterial territories in an individual patient is more difficult than was initially realized. For instance, marked variation has been found in the laterality of the anterior cerebral artery (ACA) supply. One ACA supplies portions of the contralateral cerebral hemisphere in 12%–25% of brains (33–35). The area of contralateral supply is usually small, but in 4%–7% of brains there is a major contralateral supply (33–35). In addition, great variability is seen in the volume of brain supplied by the major cerebral arteries. In one study of vascular distributions of major cerebral arteries, van der Zwan et al (26) showed that no groups of hemispheres exhibited the same combinations of arterial territorial volumes. The ACA could supply 18%–35% of any one hemisphere; the MCA, 34%–64% of the same hemisphere; and the posterior cerebral artery (PCA), 9%–40% of that hemisphere (26). Thus, van der Zwan et al concluded that the location of a cortical infarct in or at the border of the area of variation, visualized with CT or MR imaging, gives no certainty about the pathogenesis of the disease. Although the occurrence of bilateral infarcts may suggest a hemodynamic pathogenesis, the interpretation of the CT or MR images based exclusively on the location of the infarct may lead to false diagnosis (25, p 936).
Identification of Specific Border Zones in Individual Patients
Another problem encountered in determining whether an infarct is a border-zone infarct is that of definition of specific border zones in a particular patient. van der Zwan et al (24–26) studied the vascular territories in 25 unfixed human brains obtained at postmortem examination. After ligation of the posterior medial choroidal and anterior choroidal arteries, these authors individually cannulated the two ACAs distal to the anterior communicating artery, the two PCAs distal to the posterior communicating artery, and the two MCAs. They then simultaneously infused colored medium into all arteries, taking care to maintain the perfusion pressure in each vessel at a constant 93 mm Hg until the distributions of the perfusate stabilized at well-defined borders. They designated these borders the equal pressure boundaries.
van der Zwan et al (24–26) found that the locations of the equal pressure boundaries (EPBs) varied greatly in both the superficial and the deep arterial distributions of each vessel. Not one brain showed a symmetric pattern of intracerebral perfusion. So many variations existed that the authors simply described the results in terms of minimal and maximal regions of distribution for each vessel. Notably, the MCA territory on the convexity extended far upward to reach the interhemispheric fissure, separating the convexity distributions of the ACA from those of the PCA in 26% of cases (25). The EPB for the ACA and PCA lay anywhere along the length of the brain from the superior frontal gyrus to the occipital lobe on both the superior and the medial surfaces of the hemisphere (25). The EPB for the ACA and MCA varied equally widely over the convexity from the superior frontal gyrus to the inferior frontal sulcus, while the EPB for the PCA and MCA varied anywhere from the superior temporal sulcus on the convexity to the occipitotemporal sulcus on the inferior surface of the brain (25). Thus, the superior portion of the precentral gyrus (Brodmann area 4) was supplied by the ACA in 94% of cases, the MCA in 4% of cases, and the PCA in 2% of cases. Furthermore, the superior portion of the postcentral gyrus (Brodmann areas 1–3 and 5) was supplied by the ACA in 78% of cases, the MCA in 6% of cases, and the PCA in 16% of cases (25). Curiously, these authors did not address the problem of bihemispheric ACA supply. They acknowledged that ligation of the choroidal vessels led to distortion of the EPB along the hippocampus but did not elaborate on any possible distortion introduced into the other territories assessed (25).
Because of the great variability outlined above, it is now common to describe cortical infarctions as territorial if they fall completely within the maximum possible van der Zwan territory of a cerebral artery, and as potentially border zone if the infarction falls outside these maxima.
Reassessment of Stroke Mechanism in Border-Zone Infarctions
Another difficulty encountered by the radiologist assessing imaging studies in a patient suspected of having a border-zone infarction is that the mechanism of infarction in border-zone infarcts is less clear than was previously thought. Border-zone (watershed) infarctions were originally conceived of as the common consequence of severe large vessel disease complicated by acute (or acute plus later sustained) hypotension (29–31). However, border-zone infarctions are uncommon in unselected patients with acute stroke and do not occur more often in patients with hemodynamic compromise than in those with cardiac sources of stroke (36). No evidence has been found for a selective increase in cerebral oxygen extraction fraction in patients with carotid artery occlusion (37,38). In one study of 110 patients with carotid artery occlusion (39), no statistically significant difference was found in the incidence of cortical border-zone infarctions between patients with and those without evidence of hemodynamic compromise (measured as increased oxygen extraction fraction at positron emission tomography). Cortical border-zone infarcts have been noted as often in patients with cardiac sources of embolism (3.2%) as in those with severe carotid artery obstruction (>70% stenosis or occlusion) (3.6%) (36). Moreover, border-zone infarctions are rarely (5.2%) the initial manifestation of carotid occlusion, as might be expected for acute low-flow states (40). Instead, border-zone infarction accounts for 72% of delayed strokes in patients with occluded internal carotid arteries (40). Thus, the specific hemodynamic basis of border-zone infarctions is not established.
Considerations of the causes of border-zone infarction now emphasize the known phenomenon of directed embolization (41–44). The majority of cerebral artery bifurcations are of unequal size, with the smaller vessel exiting at an acute angle to the larger parent vessel. As a consequence, particulate matter may be directed with the flow toward the end territories of the major vessels at the border zone (44). It has been shown (41) that experimental embolic infarctions involved the ACA in 8% of trials if only one embolus was released into the circulation but involved the ACA in 50% of trials if several emboli were released (in which case the first embolus never lodged in the ACA). Therefore, an initial occlusion of MCA branches may redirect flow (and flow-directed emboli) into the ACA (41).
These observations have subsequently been confirmed and the concept extended to the entire cerebrovascular tree (42,43). Cerebral watershed infarction as a result of emboli has been documented in patients at postmortem examination, resulting in the conclusion that the infarctions were caused by directed embolization (44). These data indicate the high likelihood that border-zone infarctions result from directed cardiac or arterial emboli. Hennerici et al concluded that the cortical wedge-type of borderzone infarction, said to result from hemodynamic compromise in low-flow perfusion territories, is an ambiguous observation and may be seen in patients with cerebral embolism and hemodynamic compromise due to severe carotid disease (36). Microemboli in relation to border-zone infarctions are now well recognized (29,45). Recently, attempts have been made to synthesize these concepts by suggesting that reduced cerebral perfusion limits the ability of blood flow to clear emboli from the vessels, allowing them to pass to the border zones (46).
Emerging Concepts
The vascular resistance of arteries in the white matter has been shown to be about 3.8 times the vascular resistance in the gray matter (26). Allowing for that difference, the caliber (diameter) of the parent cerebral artery correlates well (r = 0.73) with the volume of the gray and white matter irrigated by each vessel (26). The mean diameters of the postcommunicating ACA (2.24 mm), the postcommunicating PCA (2.04 mm), and the MCA (2.70 mm) lie in direct relative proportion as the supplied territory (26). Measurement of the diameters of these vessels in healthy individuals would provide one measure of the anatomic variations present in each hemisphere and perhaps serve as a guide to later analysis of any strokes.
Deep border-zone infarctions appear to correlate well with hemodynamic compromise (47). It has been shown that in a group of patients with internal carotid occlusion, a rosary pattern of deep white matter infarcts was present only in patients with increased cerebral oxygen extraction fraction in the involved hemisphere (sensitivity, 22%; specificity, 100%) (39). This pattern was described as consisting of three or more lesions 3 mm in diameter or larger arranged in a linear fashion parallel to the lateral ventricle in the centrum semiovale or corona radiata. Results of other studies (48–51) also support the validity of deep watershed infarctions being related to hemodynamic compromise. Thus, at least one imaging sign appears to correlate with hemodynamic compromise.
Analysis of Infarct Frequency Distribution
Another approach to infarct analysis is to determine the frequency with which infarctions affect each of a large number of defined anatomic zones. Specific analysis of the distribution of any infarct may then be compared with the zonal frequency distributions of diverse infarcts as a guide to the nature of the lesion. Application of such analysis to 30 border-zone infarctions studied among 150 supraventricular infarctions showed that parasagittal border-zone cortical infarctions exhibit a bimodal frontal and parietal distribution (52). The anterior cortical-subcortical border zone (ACA/MCA) infarcts show peak frequency at the junction of the superior frontal sulcus and precentral sulcus and involve the adjacent portions of the superior frontal gyrus, middle frontal gyrus, precentral gyrus, and paracentral lobule. The posterior cortical-subcortical (distal ACA/MCA or PCA/MCA) border-zone infarctions most frequently involve the superior parietal lobule and the precuneus. As a consequence, detection of small foci of abnormal signal intensity specifically at these two sites should increase the index of suspicion for a possible source of emboli or other risk factors for stoke. However, until the individual vascular territories can be displayed easily in vivo at the time of stroke assessment, imaging display of infarct topography can provide only inferences with regard to the specific arteries affected and the mechanisms involved.
|
THROMBOLYTIC THERAPY |
|---|
|
TOPABSTRACTINTRODUCTIONINTRODUCTIONIMAGING FINDINGS IN HYPERACUTE...VARIABILITY OF VASCULAR...THROMBOLYTIC THERAPYCONCLUSIONREFERENCES |
|---|
The important role of thrombosis in stroke combined with the success of thrombolysis in acute myocardial infarction has generated great interest in cerebral fibrinolysis. The goal of thrombolytic therapy is rapid restoration of blood flow and preservation of the ischemic penumbra. The U.S. Food and Drug Administration has approved intravenous recombinant tPA for the treatment of acute ischemic stroke within 3 hours of symptom onset (2). In addition, intraarterial thrombolysis has been shown to improve neurologic outcome in patients with acute ischemic stroke (6). The following sections highlight the results of various clinical trials.
Intravenous Thrombolysis
Two thrombolytic agents have been utilized in trials of acute stroke treatment with intravenous medication: streptokinase and tPA (1,2,60–64).
Intravenous streptokinase.—Three trials of intravenous streptokinase for treatment of acute stroke have been reported: the Multicenter Acute Stroke Trial–Europe, the Multicenter Acute Stroke Trial–Italy, and the Australian Streptokinase Trial (60–62). The dose of streptokinase given in these trials was the same as that given in the acute myocardial infarction trials, namely, 1.5 million IU. Treatment was initiated within 4 hours in the Australian trial and within 6 hours in the other two trials (60–62). All of the streptokinase trials were halted prematurely due to poor outcome or an excess rate of mortality in the treated group, making it unlikely that streptokinase will be used in subsequent randomized trials for acute ischemic stroke.
Several reasons can be cited for the failure of streptokinase in these trials. First, the dose of streptokinase was likely high because it was equivalent to the dose used in the coronary thrombolysis trials. A lower dose might have been effective but with lower mortality. For instance, approximately two-thirds of the cardiac dose was used in the successful NINDS intravenous tPA trial (2). Second, in all three trials the patients reporting negative results were treated up to 4 hours and 6 hours after onset of symptoms. The investigators in the Australian trial (62) suggested that streptokinase may be effective if given within 3 hours of symptom onset. However, the number of patients treated within this time was not large enough to show statistical significance. The third reason for failure of streptokinase may have been the use of antiplatelet and antithrombotic agents within the first 24 hours after treatment, resulting in a higher risk of hemorrhage. In the NINDS tPA trial, use of antiplatelet and antithrombotic agents was avoided in the first 24 hours after administration of tPA, perhaps leading to safer use of the thrombolytic drug. On the basis of results from the abovementioned studies, the American Academy of Neurology recommendation regarding streptokinase in treatment of acute stroke (65), published in 1996, states that outside the setting of a clinical trial, streptokinase administration is not indicated for the management of acute ischemic stroke.
Intravenous tPA.—Four phase 3 trials of intravenous tPA for acute ischemic stroke have been published (1,2,63,64). Food and Drug Administration approval of tPA was based on data from the NINDS tPA trial (2). The study was composed of two clinical trials, part I and part II, with both trials conducted in identical fashion by using the same inclusion and exclusion criteria (2). Altogether, 624 patients were treated within 3 hours of symptom onset with a dose of 0.9 mg of tPA per kilogram of body weight and a maximum dose of 90 mg.
In part I of the study, early outcome was evaluated. The primary hypothesis tested in part I was that at 24 hours, a greater proportion of patients treated with tPA would improve by 4 or more points on the National Institutes of Health Stroke Scale (NIHSS), as compared with scores of patients receiving a placebo. Indeed, 47% (67 of 144) of patients treated with tPA improved by 4 or more points on the NIHSS, compared with 39% (57 of 147) in the placebo group (P = .21).
In part II of the study, the long-term functional outcome of patients at 3 months was evaluated. The primary hypothesis tested in part II was that there would be a significant difference between the tPA group and the placebo group in terms of the proportion of patients with minimal or no deficit. Minimal or no deficit was defined according to one of the following four outcome scales: NIHSS score of 0 or 1, Barthel Index greater than 95, modified Rankin Scale score of 0 or 1, and Glasgow Outcome Scale score of 1. For all four outcome measures, the tPA treated group fared better than the placebo group. The tPA patients were 32% (NIHSS), 38% (Barthel Index), 50% (modified Rankin Scale), and 55% (Glasgow Outcome Scale) more likely to have a good outcome as defined above. The absolute percentage difference was 11%–13%, depending on the outcome measure being used—that is, for every 100 patients treated with tPA, there would be an additional 11–13 patients with minimal or no deficit, compared with 100 patients not treated with tPA. The NINDS investigators have also reported follow-up in these patients at 12 months (66,67). The patients treated with tPA were at least 30% more likely to have minimal or no deficit at 1-year follow-up. There was no significant difference in mortality at 12 months between the two groups (24% vs 28%, P = .29). The results indicate a sustained benefit of tPA at 12 months. Overall, 6% of the patients receiving tPA had symptomatic intracranial hemorrhage, compared with 0.6% in the placebo group. Despite the higher rate of symptomatic intracranial hemorrhage in the tPA group, there was no significant difference in mortality between the two groups.
Several explanations have been proposed to account for these apparently contradictory findings (68). On the one hand, tPA may have decreased mortality by reducing the size of infarction and, hence, reducing the likelihood of death. Alternatively, the patients with symptomatic intracranial hemorrhage had large infarction that, even without hemorrhage, would have had high associated mortality. In other words, in the tPA group, patients with large infarctions died with hemorrhagic transformation, while in the placebo group these same patients died but without hemorrhagic transformation. This would result in a higher rate of symptomatic intracranial hemorrhage in the tPA group but no significant difference in mortality between the two groups.
The NINDS investigators sought to identify variables associated with hemorrhage in patients who received tPA (5). The only variables independently associated with a risk of symptomatic intracranial hemorrhage were severity of neurologic deficit as measured with the NIHSS and presence of brain edema or mass effect on CT images obtained prior to treatment. Despite this fact, however, patients with a severe neurologic deficit were more likely to have a favorable outcome if treated with tPA than were those who received the placebo.
The results were similar in the analysis of patients with edema or mass effect seen on pretreatment CT images. The investigators thus concluded that despite a higher rate of symptomatic intracranial hemorrhage, patients with severe stroke or edema or mass effect on pretreatment CT images are reasonable candidates for tPA if it is administered within 3 hours of symptom onset (5). The NINDS investigators also performed a subgroup analysis to identify stroke patients in whom treatment with tPA was particularly hazardous or efficacious (69). They concluded that no pretreatment information significantly affected outcome.
Three additional phase 3 trials with intravenous tPA have been reported: ECASS, ECASS II, and Alteplase Thrombolysis for Acute Non-interventional Therapy in Ischemic Stroke (ATLANTIS) (1,63,64). All had negative results. The ECASS was a prospective, multicenter, double-blind, placebo-controlled study and differed from the NINDS study in several important aspects (1). First, the dose of tPA in ECASS was 1.1 mg/kg with a maximum dose of 100 mg; the NINDS study used a dose of 0.9 mg/kg and a maximum dose of 90 mg. Second, patients were treated up to 6 hours after onset of symptoms. Third, patients with a major early sign of ischemia on CT images (defined as hypoattenuation involving more than one-third of the MCA territory) were excluded. The primary treatment end point included Barthel Index and modified Rankin Scale scoring at 90 days. In the intent-to-treat analysis, no significant difference in either of the two primary end points was seen between the two groups of patients, which was thought to be due to the fact that there were protocol violations in a substantial number of patients included in the analysis. To further address this issue, an analysis of the target population was also performed by excluding 109 patients that were included in the trial but were later found to have been associated with protocol violations. Analysis of the target population revealed no significant difference in Barthel Index scores between the two groups but a significant difference in the modified Rankin Scale score in favor of the tPA treated patients. Sixty-six of the excluded 109 patients were excluded because of abnormalities on CT images (52 had major early infarction signs, two had primary hemorrhage, 12 had unavailable or uninterpretable CT images). The investigators concluded that tPA is beneficial in improving some functional outcomes in a subgroup of patients with moderate to severe neurologic deficits and without extended signs of ischemia on initial CT image. Identification of this subgroup of patients is difficult, however, and largely depends on identification of early signs of infarct on CT images. Furthermore, treatment of ineligible patients is associated with an unacceptable risk of hemorrhage and death. Therefore, the investigators concluded that intravenous tPA administration within 6 hours of symptom onset could not be recommended.
In the ECASS, no significant difference in the frequency of parenchymal hemorrhage in general was found between the tPA and placebo groups in either the intent-to-treat or the target population analyses. However, large parenchymal hemorrhages were more frequent in the tPA-treated group. For that reason, the ECASS II trial was designed with a lower dose of intravenous tPA (0.9 mg/kg, chosen to match NINDS criteria) given within 6 hours of symptom onset. The investigators found no significant difference in the primary end point (modified Rankin Scale score at 90 days) between the tPA and placebo groups. The results did not confirm a significant benefit for tPA.
The approved use of tPA remained restricted to patients presenting within 3 hours of symptom onset. However, this severely restricted the use of the drug, as evidenced by data showing that, since approval of the drug, fewer than 5% of all stroke patients were receiving tPA (64,70,71). Therefore, the ATLANTIS trial set out to assess the safety and efficacy of tPA (0.9 mg/kg, with maximum dose of 90 mg as in NINDS trial) in patients 3–5 hours after symptom onset. This trial was a phase 3, placebo-controlled, double-blind, randomized study. The primary end point was an excellent neurologic recovery at day 90 (NIHSS score 
1). In the target population, 32% of the patients in the placebo group and 34% of the patients in the tPA group had an excellent recovery at 90 days (P = .65). There was a significant increase in the occurrence of symptomatic intracranial hemorrhage in the tPA group (1.1% vs 7.0%; P < .001). Therefore, this study did not show a significant benefit of tPA in patients treated 3–5 hours after ictus.
On the basis of these results, the recommendation of the American Academy of Neurology is that tPA administered intravenously within 3 hours of symptom onset is indicated in patients meeting the inclusion and exclusion criteria as set forth on the basis of the NINDS tPA trial data (65). However, intravenous administration of tPA more than 3 hours after stroke is not recommended.
Despite the approval of tPA for treatment of acute ischemic stroke, outlined above, there has been much trepidation about its use, and several criticisms have been advanced. It has been pointed out that many patients with symptoms of acute ischemic stroke may not have occlusive thromboemboli (72). The evaluation of patients with stroke is typically performed rapidly, with no pathophysiologic assessment to document the presence of an occlusive clot. It has been argued, then, that many patients will receive a potentially dangerous drug although they do not have the problem for which the drug is intended. Additionally, concern has been raised about whether use of tPA would be less efficacious in general community practice, as compared with use in the idealized tertiary-care university hospital conditions under which the trial was carried out.
The Standard Treatment with Alteplase to Reverse Stroke, or STARS, study was designed to address concerns over the use of tPA in the community setting (73). In this prospective multicenter study, the results of intravenous tPA treatment of patients with acute ischemic stroke in 57 medical centers (24 academic, 33 community) in the United States were reported. The results confirmed the beneficial effects of tPA administered within 3 hours of symptom onset, with findings similar to those obtained in the NINDS trial. However, conflicting data were found in another study in which results were reported for stroke patients treated with intravenous tPA in essentially all the hospitals in Cleveland, Ohio (74). The results of that study showed a significantly higher rate of symptomatic intracranial hemorrhage and mortality in patients receiving the drug. However, a large percentage of patients in the study had deviations from national treatment guidelines. As in other studies, evidence of a learning curve for use of tPA, which was reflected in a decrease in the occurrence of guideline deviations and an increase in the rate of intravenous tPA use over time, was seen in the Cleveland experience (74).
Intraarterial Thrombolysis
In order for thrombolytic drugs to induce lysis of acute thromboemboli, a therapeutic dose of the drug must reach the target. However, if major arteries are blocked, intravenous administration may result in insufficient drug delivery. Therefore, much interest has developed in intraarterial delivery of thrombolytic agents. Compared with intravenous therapy, localized intraarterial thrombolysis has the theoretical advantage of achieving faster and more complete recanalization with use of a lower dose. Because the agent is administered during cerebral angiography, clot lysis can be directly assessed, allowing drug infusion to be stopped when clot lysis is achieved (Fig 6). This feature allows an optimal amount to drug to be administered and may diminish the risk of adverse effects. In addition, intraarterial treatment can be initiated up to 6 hours after symptom onset, and the therapeutic window is wider than for that for intravenous therapy.
|
|
|
As with intravenous thrombolysis, only results from randomized trials can be used to answer questions regarding the safety and efficacy of intraarterial therapy. Two such randomized trials have been performed for intraarterial thrombolysis: Prolyse in Acute Cerebral Thromboembolism Trial (PROACT) and PROACT II (6,56). The larger of the two trials, PROACT II, included patients treated within 6 hours of symptom onset who had angiographically demonstrated occlusion of the MCA (M1 or M2 occlusion) (6). The primary outcome of the trial was the ability to live independently at 3 months after stroke. Of the 474 patients who underwent angiography, 180 were enrolled, with 121 receiving intraarterial prourokinase and low-dose intravenous heparin and 59 receiving low-dose intravenous heparin alone. At 2 hours, 67% of patients receiving prourokinase had complete or partial recanalization, compared with 18% in the heparin-only group (P < .001). The primary outcome of the study was attained by 40% of the patients treated with prourokinase, compared with 25% in the heparin-only group (P = .04). However, symptomatic intracranial hemorrhage was seen in 10% of patients undergoing thrombolysis, compared with 2% in the heparin-only group (P = .06). Nonetheless, the authors of the study determined that intraarterial thrombolysis within 6 hours of symptom onset was shown to have a benefit in patients with MCA occlusion. The U.S. Food an Drug Administration did not approve prourokinase after the PROACT II study and instead asked that an additional study be performed, requiring two positive phase 3 trials prior to approval of the drug. Abbott Laboratories is currently considering reopening the PROACT trial.
Intraarterial Thrombolysis Trials in Posterior Circulation
A stroke in the posterior circulation stroke differs in several respects from ischemic stroke in the anterior circulation. First, clinical outcome in patients with vertebrobasilar occlusion is less favorable; death occurs in the majority of patients, and severe deficit occurs in most survivors (83,94,95). In addition, patients with posterior circulation ischemic events often have coexistent severe intracranial large-artery atherosclerotic disease (96). This feature, which presents a unique problem because of the high risk of rethrombosis after treatment, is uncommon in anterior circulation ischemic events (75,83,87). Study results have suggested that thrombolysis beyond 6 hours in the anterior circulation can be associated with high rates of hemorrhagic transformation and poor outcome (102). No conclusive data exist to indicate increased risk beyond 6 hours in posterior circulation strokes; however, to our knowledge, no randomized trials of intraarterial thrombolysis in patients with vertebrobasilar stroke have been performed. In addition, although thrombolysis has been attempted up to 24 hours after symptom onset in patients with posterior circulation ischemic events, the issue of how long a delay after symptom onset can be tolerated before the start of treatment has not been specifically examined (75,77,87).
In one pilot study in which the safety and efficacy of intraarterial urokinase were evaluated in patients with severe brainstem stroke and vertebrobasilar occlusion, 16 patients with vertebrobasilar occlusion were treated within 24 hours of symptom onset (98). Incremental doses of urokinase were administered until clot lysis was achieved or a maximal dose of 1 million U was given. Complete or partial recanalization was initially achieved in 13 of 16 (81%) patients. Of these, reocclusion occurred within 24 hours in two, giving a final recanalization rate of 69%. The 6-month functional status was assessed with the Barthel Index, with a good outcome defined as a score 60 or greater. Eleven (69%) patients survived, nine (56%) with a good outcome and two (12%) with severe deficit. Recanalization correlated with survival (P = .02), but the time between symptom onset and thrombolysis was not predictive of outcome. The authors concluded that intraarterial thrombolysis in the posterior circulation is safe, feasible, and capable of achieving recanalization in the majority of patients.
Combined Intravenous and Intraarterial Treatment
One major disadvantage of intraarterial thrombolysis is treatment delay due to the need to assemble the neurointerventional team and prepare the angiography suite. A combined approach of intravenous and intraarterial thrombolysis allows the advantages of both treatment strategies. Intravenous therapy is initiated without delay but at a lower dose than is used in standard intravenous therapy. This technique allows some tPA to be given intraarterially, which provides a higher recanalization rate than can potentially be achieved than with intravenous treatment alone. The Emergency Management of Stroke study is the only reported trial of combined intravenous and intraarterial thrombolysis (99). This study was a double-bli
