Further Evolution Toward Effective Therapy for Acute Ischemic Stroke

IN THE 5 YEARS since we last reviewed the topic of acute stroke therapy,¹ many advances occurred toward the goal of developing effective therapies to treat acute ischemic stroke (AIS). Most important, the results of the National Institutes of Health (NIH) recombinant tissue-type plasminogen activator (rt-PA) trial showed for the first time that AIS could be successfully treated, although several limitations exist.² The demonstration that rt-PA can significantly improve outcome if initiated within 3 hours of stroke onset is an encouraging and important first step on the road to effective therapy for a disorder that affects more than 400000 new patients per year in the United States and remains the third most common cause of death.³ In addressing the current efforts to develop acute stroke therapy, we will focus on 3 main themes. The first is the use of thrombolytics in stroke, including the differing results for rt-PA and streptokinase. Second, neuroprotective treatments have been explored and, after many failures, there are promising new drugs in clinical and preclinical development. Our discussion of these topics was performed by a MEDLINE search for 1993 through 1997 using the key words cerebral ischemia, thrombolysis, and neuroprotection, which identified 30 articles on thrombolysis and 45 articles on neuroprotective drugs related to cerebral ischemia, focusing primarily on randomized phase 2 and phase 3 clinical trials. Last, we will describe the potential effects of combined thrombolytic and neuroprotective therapy and how new imaging technology could enhance the development and implementation of combination therapies that are likely to have the best chance to markedly improve outcome.

Small controlled trials of thrombolysis were initiated nearly 40 years ago, but the therapy was discarded because of increased risk of death.⁴ With the availability of computed tomography (CT) for acute stroke, allowing for easy identification of brain hemorrhage, interest in thrombolysis reappeared. A meta-analysis reviewed data from 12 heterogeneous clinical trials including approximately 3000 patients and reported a 20% reduction in combined severe disability and death.⁵ However, this was counterbalanced by a 3-fold increase in brain hemorrhage, associated with a mortality rate of nearly 50%. Further evaluation of intravenous (IV) thrombolysis in AIS was also encouraged by the observed benefit with this therapy in acute myocardial infarction.⁶

The thrombolytics under investigation in AIS are rt-PA, streptokinase, urokinase, and prourokinase.^{4 ,7} Clinical interest is presently focused on rt-PA, mainly because of the positive National Institute of Neurological Disorders and Stroke (NINDS) rt-PA Stroke Study,² contrasting with 3 negative trials using streptokinase.^{8 - 10} Recombinant tissue-type plasminogen activator is a relatively fibrin-specific serine protease that has a shorter serum half-life than streptokinase (4-5 minutes vs 18-25 minutes), but both have more prolonged, thrombolytic activity.⁴

A fresh thrombus occluding an artery is likely to respond better to thrombolytic agents than an old, organized thrombus. Cholesterol, calcium, and fibrin-poor thrombi, as well as atheromatous stenosis with only a small superimposed thrombus, are also less likely to respond to thrombolytic drugs. The actual differential response to thrombolysis of these different types of arterial occlusions has not been studied, and the available randomized trials do not provide sufficient data. More information on the response to thrombolytics as a function of thrombus site, type, age, size, and origin may be critical in improving the selection of acute stroke patients who are best suited for this therapy.

One of the problems with IV thrombolysis in AIS is that without thorough investigation of the patient's vascular status, it is not known whether individual patients actually have an occluded artery at the time of treatment. The large randomized clinical trials were nonangiographic and assumed that at least 1 large- or small-artery occlusion was present, which is the case in about 75% of patients with ischemic stroke who undergo angiography within 6 to 8 hours of stroke onset.^{11 - 13} We do not know if there is a differential response to IV thrombolysis in angiographically positive vs negative patients. Even without a visible occlusion, thrombolysis may be effective if distal microvessels (not visible on angiography) are occluded.

As the natural dynamics (including physiological thrombolysis) of cerebrovascular occlusion may vary greatly according to the nature of the occlusion (eg, embolism vs thrombosis in situ, embolism of old vs fresh material), so may the response to thrombolytic therapy. Because ischemic stroke is not a single condition, we need to improve pathophysiological knowledge to better define which patients are the best candidates for thrombolysis.

There have been 5 large randomized trials of IV thrombolysis (Table 1). Only 1, the NINDS rt-PA Stroke Study,² was unequivocally positive. The other trial of rt-PA, the European Cooperative Acute Stroke Study (ECASS),¹⁴ was equivocal, while all 3 trials of streptokinase (Multicenter Acute Stroke Trial–Italy [MAST-I], Multicenter Acute Stroke Trial–Europe [MAST-E], Australian Streptokinase [ASK] Trial)^{8 - 10} were negative. The negative results with streptokinase have many explanations. Streptokinase itself and the relatively high dosage used (vs rt-PA in the NINDS study) may explain the increased risk of hemorrhage. Antithrombotic therapy (eg, heparin, aspirin) was allowed in all 3 streptokinase trials, except in a subgroup from MAST-I, and may have facilitated hemorrhage. Additionally, time to treatment may be a critical factor. While all 624 patients in the NINDS study were treated within 3 hours of stroke onset (including 302 [48%] within 90 minutes), 75% of the patients were treated after 3 hours in MAST-I and ASK. In MAST-E the median delay from onset to treatment was 5 hours. Time to treatment was also one of the main differences between the positive (NINDS) and the equivocal (ECASS) rt-PA trials, since the mean time to treatment in ECASS was 4.3 hours.

In this 2-part randomized (< 3 hours) trial of rt-PA (0.9 mg/kg; maximum, 90 mg; 10% bolus followed by a 60-minute infusion) vs placebo in 624 patients, there was a trend toward neurologic recovery at 24 hours (as assessed by the NIH Stroke Scale). At 3 months 50% of rt-PA–treated patients had no or minimal disability (on the Barthel Index) compared with 38% of the controls. The odds ratio for favorable outcome with rt-PA was 1.7 (95% confidence interval [CI], 1.2-2.6). The symptomatic brain hemorrhage rate was exceptionally small in the placebo group (0.6%), but was increased 10-fold with rt-PA (6.4%) (P< .001). Despite this, mortality at 3 months was still nearly the same with rt-PA (54 of 312, 17%) as with placebo (64 of 312, 21%) (P=.3). The benefit of rt-PA did not vary significantly by stroke subtype diagnosed at baseline. It is somewhat puzzling that patients with presumed lacunar stroke contributed significantly to the positive treatment, because the placebo-treated lacunar stroke patients had a much worse outcome than the previous literature suggests.¹⁵ These issues have engendered considerable controversy concerning the applicability and generalization of rt-PA therapy. In Europe, IV rt-PA for AIS has remained mostly an experimental treatment,¹⁶ justifying the ongoing ECASS-2 trial (rt-PA, 0.9 mg/kg within 6 hours of stroke onset). Recombinant tissue-type plasminogen activator was approved in June 1996 in the United States and promoted under specified conditions by the American Academy of Neurology (AAN) and the American Heart Association (AHA) Stroke Council.^{17 - 18}

In the ECASS,¹⁴ 620 patients with moderate to severe AIS were randomized (< 6 hours) to placebo or rt-PA (1.1 mg/kg; maximum, 100 mg; 10% bolus followed by a 60-minute infusion). A CT exclusion was a major early infarction sign involving more than one third of the middle cerebral artery territory (diffuse sulcal effacement, poor differentiation between gray-white matter, and diffuse hypodensity). At 3 months, the primary end points (a 15-point difference on the Barthel Index and a 1-point difference on the Modified Rankin Scale) showed no significant difference between rt-PA–treated and placebo-treated patients. However, rt-PA was associated with an increased rate of recovery and a shorter hospital stay. Mortality in the rt-PA vs placebo group was higher at 30 days (17.9% vs 12.7%, P=.08) and significantly higher at 3 months (22.4% vs 15.8%, P=.04). Unlike the NINDS study, the development of cerebral hemorrhage was common in the placebo group (36.8%) and not significantly different from the rt-PA group (42.8%). However, large parenchymal hemorrhages were increased 3-fold with rt-PA (19.8% vs 6.5%, P<.001). This led to a significant increase in brain hemorrhage–associated death in the rt-PA group (6.3% vs 2.4%, P=.02). Although overall ECASS was a negative trial, it must be emphasized that 109 of 620 randomized patients were not appropriate (prespecified protocol violators), 54 because of early CT abnormalities. After excluding these 109 patients, outcome analysis in the 511-patient target population showed a statistically significant (P=.04) shift of the Modified Rankin Scale from 3 (moderate disability) to 2 (slight disability) at 3 months. The other primary end point, the Barthel Index, showed no significant difference. That ECASS found the overall negative results of the study were largely driven by inappropriate CT scan interpretation in the haste of emergency management of stroke patients, even in centers known for their expertise in this area, has important implications for applying results of thrombolytics trials to other sites. In its guidelines for implementing rt-PA therapy in AIS, the AHA Stroke Council panel included, in addition to the inclusion-exclusion criteria of the NINDS study,² a specific recommendation to exclude patients in whom baseline CT demonstrates changes suggestive of major infarction.^{17 - 18} In ECASS, even after removing the prespecified protocol violators, the rate of symptomatic brain hemorrhage remained almost 3 times greater in the rt-PA group, suggesting that as in the NINDS study, even in a well-targeted population, IV rt-PA causes a substantial risk of brain hemorrhage.

All 3 trials (MAST-I,⁸ ASK,⁹ and MAST-E¹⁰ ) were halted on recommendation of the Data Monitoring Committees because of unacceptable risks (Table 1). The absence of a difference in outcome at 6 months between the streptokinase and the nonstreptokinase groups in MAST-E and MAST-I may have been caused by the combined effect of excessive early case fatalities and improved disability in survivors. This agrees with the ECASS findings, which suggest that time to treatment and patient selection are critical factors for successful thrombolysis, whereas specific differences between streptokinase and rt-PA may be less important. Another IV streptokinase trial may be indicated, but such a trial should compare streptokinase with rt-PA within 3 hours of stroke onset and should include improved patient selection.

It is difficult to determine which patients are most likely to benefit from IV rt-PA therapy and which patients are at risk, even when using the guidelines for the approved use of IV rt-PA within 3 hours of ischemic stroke onset.¹⁹ Failure to fully understand this issue constitutes a major limitation to the implementation of rt-PA therapy in general practice. Several factors may be critical to selecting patients who will have the best risk-benefit ratio with IV rt-PA.

The differences in the NINDS, ECASS, and ASK findings suggest that time to treatment is critical. However, data from functional imaging (eg, positron emission tomography, diffusion-weighted magnetic resonance imaging [MRI]) suggest that the therapeutic time window may extend well beyond 3 hours in some patients, while in others it may already be closed within 1 hour.²⁰ A simple tool to detect ischemic but viable tissue in individual patients admitted at different times after stroke onset would be useful to select candidates for thrombolysis, especially if this tool also demonstrated occluded arteries that could potentially be reopened. Developing magnetic resonance technology such as angiography, diffusion, and perfusion imaging may be suitable, but this has not been validated.

Factors associated with a more favorable outcome include younger age, absence of cardiac disease or diabetes, lower blood pressure on admission, better neurologic score, absence of early ischemic parenchymal changes, large-artery thrombus visible on baseline brain CT scan, and a well-developed collateral circulation.^{21 - 22} Specific predictors of positive vs negative outcome with thrombolytic therapy still remain to be delineated.

Time to treatment, dose of thrombolytics, blood pressure level, severity of neurologic deficit, and severity of ischemia are risk factors for developing brain hemorrhage, mainly symptomatic bleeding into an evolving infarction.²³ Arterial recanalization may theoretically be a factor for both recovery and bleeding into the ischemic area, but most anecdotal cases of severe intrainfarctional hemorrhage have been reported without recanalization. This finding, which should be evaluated prospectively in larger series, may be of importance, since it suggests that the benefit (ie, vessel reopening) and the risk (ie, hemorrhage) potentials of IV thrombolysis do not run parallel. However, it is also possible that the most severe hemorrhages tend to occur in rapidly worsening and enlarging ischemic areas with early tissue damage, so clinical deterioration may often be wrongly attributed to hemorrhage instead of evolving infarction. Potential complications (in addition to hemorrhage) of thrombolysis include reperfusion injury, arterial reocclusion, and secondary embolization due to thrombus fragmentation.²⁴

Local intra-arterial thrombolysis has the advantage of a higher local drug concentration and a lower systemic concentration. Another advantage is more accurate selection of patients with demonstration of an occluded brain artery, since angiographic documentation is available. More than 10 small uncontrolled series provide anecdotal evidence of benefit of intra-arterial thrombolysis in severe stroke patients, mainly involving the proximal posterior circulation (basilar artery occlusion), usually with urokinase or rt-PA.^{24 - 27} Only 1 preliminary randomized trial is available, the Prolyse in Acute Cerebral Thromboembolism Trial (PROACT),¹³ in which 26 patients with middle cerebral artery territory AIS received 6 mg of prourokinase and 14 patients received placebo within 6 hours of stroke onset, both followed by IV heparin. There was no difference in outcome, but on repeat angiography 1 hour after treatment, partial or complete recanalization was significantly higher in treated (15 of 26; 58%) vs control (2 of 14; 14%) patients (P< .02). PROACT-2 is ongoing (with 9 mg of prourokinase) and other studies are planned, including trials with associated stenting of the thrombolized artery to avoid reocclusion. Local IV thrombolysis has also been performed with anecdotal benefit in patients with cerebral venous sinus thrombosis.^{28 - 29} However, despite the promise and reported spectacular benefits in individual cases, delivery in situ of thrombolytics is likely to remain unavailable to most patients.

Following the NINDS study, IV rt-PA was advocated in AIS patients within 3 hours of onset, provided the inclusion and exclusion criteria applied in the NINDS study are followed. However, this time window and other criteria currently apply to fewer than 5% of stroke patients. It is important that the decision to treat AIS with rt-PA should remain the responsibility of an expert neurologist (and an expert neuroradiologist for baseline CT scan assessment), with appropriate access to optimal acute care facilities. The most important result of the NINDS study may be heightened awareness of acute stroke as a neurologic emergency. The results of ongoing trials such as ECASS-2 and PROACT-2 will be available soon. A potential future development of thrombolytic therapy in stroke treatment may be combination therapy with neuroprotective agents, since this could extend the therapeutic time window. Combination therapy with antithrombotic drugs, in particular platelet antiaggregating agents such as platelet glycoprotein IIb/IIIa complex antagonists, is also attractive, because the thrombolytic process itself has proaggregating effects. New thrombolytic agents (eg, reteplase, a mutant form of endogenous tissue-type plasminogen activator [t-PA]; TNK t-PA, a fibrin-specific variant of t-PA with a longer half-life; E6010, a second-generation t-PA with a longer half-life) are under study and may provide safer and more effective therapy than rt-PA.³⁰

The basic premise of neuroprotective drugs is that the cerebral ischemic region has varying levels of residual blood flow, resulting in evolution of damage from ischemia to cell death, and this residual circulation permits delivery of drugs to the therapeutic target.³¹ At the site, a number of factors involved in cell damage can be the targets of neuroprotective drugs (Figure 1). In animal stroke models, extensive reductions of infarction volume occur with many different types of neuroprotective drugs.³² Unfortunately, none of the therapies that were found to be effective in animals has conclusively proven to be successful in stroke patients, although some benefits are detectable.

Influx of calcium intracellularly enhances cell death, which has led to trials of drugs inhibiting voltage-regulated and receptor-mediated calcium channels. Nimodipine, a dihydropyridine, was extensively evaluated in many clinical trials without benefit.³² However, the time window for the initiation of therapy, 24 to 48 hours from stroke onset, may have been too late to demonstrate a benefit. A meta-analysis of the 9 major nimodipine trials demonstrated that patients whose treatment began within 12 hours of stroke onset had a significant improvement in functional outcome.³³ Currently, a clinical trial with a 6-hour time limit for initiation of treatment with oral nimodipine is being conducted in the Netherlands.

Receptor-mediated calcium channels are fast-conducting calcium channels activated primarily by the excitatory amino acid glutamate.³⁴ The N-methyl-D-aspartate (NMDA) complex, an important and well-characterized receptor-mediated calcium channel, contains glycine and polyamine modulatory sites that are also potential therapeutic targets. Many agents affecting these sites showed promise in animal studies, but this benefit has not been borne out in clinical studies.³² While the noncompetitive NMDA antagonist aptiganel demonstrated possible efficacy and a reasonable safety profile in a phase 2 trial,³⁵ a phase 3 trial was stopped because of concerns about the risk-benefit ratio.

Another approach to inhibiting the NMDA complex is to reduce presynaptic release of glutamate, primarily by affecting sodium-conducting channels. Fosphenytoin, the widely used anticonvulsant, has neuroprotective properties in animal stroke models and a phase 2 trial demonstrated safety.³⁶ Fosphenytoin is currently undergoing a phase 3 study. Lubeluzole³⁷ also has purported effects on presynaptic glutamate release and potential neuroprotective effects, but a European phase 3 trial³⁸ was entirely negative and an American trial³⁹ showed a nonsignificant trend toward reduction of mortality, the primary end point. A small but significant effect on improving functional outcome prompted a third ongoing phase 3 trial to focus on functional outcome.

The a-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) channel is another receptor-mediated channel activated by glutamate that induces primarily sodium, but also intracellular calcium accumulation in ischemic brain tissue.⁴⁰ The AMPA antagonists appear to be generally more toxic than NMDA antagonists and development of the prototype AMPA antagonist, NBQX, was terminated because of nephrotoxic effects.³² Several newer AMPA antagonists are now in early clinical development.

Endogenous growth factors may provide a unique approach to improving outcome after focal brain ischemia. Basic fibroblast growth factor (bFGF) is the most widely studied growth factor. In animal stroke models, bFGF reduces infarction volume when given shortly after the onset of focal ischemia.⁴¹ Interestingly, bFGF given intraventricularly 24 hours after experimental stroke did not reduce infarction size, but did improve functional outcome at 28 days when assessed by sophisticated behavioral measures, presumably related to regenerative effects.⁴² A recently completed phase 2 trial demonstrated that bFGF is well tolerated by stroke patients and phase 3 trials are in progress. Citicoline, a precursor of phosphatidylcholine with antioxidant effects in animals, has been studied in 2 treatment trials in stroke patients.⁴³ In the first trial, functional outcome at 90 days after stroke significantly improved in patients who received 6 weeks of daily oral dosing (500 mg) of citicoline, beginning up to 24 hours after stroke onset.⁴⁴ In the second trial, the benefit of citicoline was apparent only in stroke patients with initially moderate to severe deficits, because the mildly affected patients markedly improved whether they received active therapy or placebo.⁴⁵ The mechanism of citicoline's beneficial effects in stroke patients is uncertain, but is probably more restorative than neuroprotective.

Antioxidants and antileukocyte interventions have been assessed clinically and primarily would appear to be useful following successful reperfusion. In animal models, both of these treatment modalities have consistently reduced infarction size in temporary focal ischemia models, but have little or no documented effects on infarction size in permanent focal ischemia models.³² Tirilazad, a 21-aminosteroid derived from methylprednisolone, was not effective as a primary acute stroke therapy and further development apparently has stopped.⁴⁶ An anti-intercellular adhesion monoclonal antibody directed at leukocytes was tested in a phase 3 trial,⁴⁷ but it worsened outcomes, most likely related to increased fever and infection.

Several other neuroprotective strategies are being explored. γ-Aminobutyric acid (GABA) is an important inhibitory neurotransmitter in the central nervous system that counters the effects of the excitatory neurotransmitters glutamate and glycine.⁴⁸ Spreading waves of depolarizations occur in ischemic regions of experimental animals subjected to focal brain ischemia,⁴⁹ and these depolarizations directly contribute to the expansion of irreversible injury, presumably because of increased metabolic demand in tissue unable to compensate. Enhancing GABA activity could partially block these ischemia-related depolarizations. The GABA agonists muscimol and clomethiazole reduce infarction volume in animal stroke models and a phase 3 clinical trial evaluating clomethiazole was recently completed. The overall results of the trial were negative, but significant benefits on functional outcome were observed in patients with large, cortical strokes when clomethiazole therapy began within 12 hours of stroke onset.⁵⁰ A follow-up phase 3 trial targeting this important stroke subtype is ongoing. Serotonin agonists also promote hyperpolarization and reduce infarction size in animal stroke models. One serotonin agonist, Bay3702, is currently under study in a phase 2 clinical trial.

Nitric oxide is produced in ischemic endothelial cells and neurons and is neurotoxic. Inhibiting neuronal but not endothelial nitric oxide synthase reduces infarction size in experimental stroke models.⁵¹ Specific neuronal nitric oxide synthase inhibitors such as ARL17477 are now available and clinical evaluation of this therapeutic approach will likely begin soon. Adenosine agonists are neuroprotective in experimental stroke models by inhibiting glutamate release, but their deleterious cardiovascular effects limit their usefulness. Recently, the contribution of programmed cell death (apoptosis) to the evolution of ischemic infarction has received increasing attention.⁵² However, the importance of this mechanism of cellular injury in clinical stroke remains uncertain. Apoptotic mechanisms could be important in ischemic regions with initially modest reductions of blood flow or after reperfusion. If the contribution of apoptotic cell death is established, future consideration toward inhibiting this process by therapeutic intervention will be necessary.

The current large number of therapies undergoing preclinical and clinical development for AIS suggest that several new drugs may be available soon for clinical use. However, because of the complexity of focal ischemic brain injury development and evolution, it is highly likely that each drug alone will have only modest beneficial effects. To be approved for clinical use, outcome must be improved by 10% to 15% compared with placebo in randomized clinical trials. Achieving an improvement in outcome of 30% to 40% likely will require the use of multiple targeted therapies.

Future multiple therapies for AIS may be developed and implemented in several ways. Therapy might be used to extend the reperfusion window. The current documented time window for the use of reperfusion with IV rt-PA is within 3 hours, but in animal stroke models, using neuroprotective drugs before reperfusion can extend the time window for the successful use of reperfusion.⁵³ A second option would be to administer rt-PA followed by a therapy with purported ameliorative effects on reperfusion injury such as an antioxidant or a safe antileukocyte intervention. Regardless of the adjunct therapies used, an important consideration with the use of rt-PA is how to maximize benefit and reduce risk for hemorrhagic side effects. Patients with a documented perfusion deficit may be most likely to benefit from early reperfusion; therefore, documenting such a blood flow deficit by perfusion MRI, single-photon emission CT, or ultrasound techniques should be considered in future clinical trial design.

Another important issue for future clinical trials in AIS will be identification of the most appropriate patients for an individual therapeutic modality or combinations. Currently, acute stroke therapies are typically given to groups of stroke patients who fulfill appropriate inclusion criteria within a fixed time window after onset, usually 3 to 6 hours. The time window of 3 to 6 hours is somewhat arbitrarily chosen to include the most patients who are likely to respond to treatment. In most acute stroke treatment trials, a minimum degree of neurological deficit is required for inclusion, with the cutoff equivalent to about 4 on the NIH Stroke Scale. All of the major stroke subtypes are usually included, such that no distinction among large-vessel atherosclerotic stroke, small-vessel lacunar stroke, or cardioembolic stroke is made. In the future, as we evolve toward therapy with multiple agents, our approaches to these and other means of characterizing stroke must change or the goal of maximizing therapeutic response will remain elusive.

The narrow window of initiating treatment within 3 hours of ischemic stroke onset will severely limit the utility of stroke therapy. However, potentially salvageable ischemic tissue may exist in some individuals for much longer than 3 hours.^{54 - 55} Unfortunately, no clinician or imaging technique can determine whether a patient has potentially salvageable ischemic tissue. However, ultrafast diffusion/perfusion MRI has such potential, although existing MRI units must be replaced or upgraded.⁵⁶ If diffusion/perfusion MRI can rapidly provide physiological-based information about the existence or lack of potentially salvageable ischemic tissue for individual stroke patients, then the therapeutic time window could be individualized, increasing the likelihood that a particular patient may benefit from treatment.⁵⁷

Treatment efficacy also may depend on the stroke subtypes. For example, brain white matter does not contain synapses or receptor complexes and lacunar strokes, by classical definition, are purely white matter lesions.^{58 - 59} Therefore, therapy directed at the NMDA receptor complex would not be expected to be effective in this stroke subtype. Additionally, the natural history of untreated lacunar stroke is quite favorable, with 70% or more of such patients achieving a favorable outcome at 180 days without any therapy.⁶⁰ If an anti-NMDA receptor therapy trial included a substantial percentage of patients with lacunar stroke, it is unlikely that any overall treatment effect could be detected. This most likely was the case in the recently reported trial of clomethiazole, a GABA agonist, although the reason for the drug's ineffectiveness in this patient subgroup is unclear. While the overall results of the trial were negative, patients who appeared to have a large cortical lesion benefited from clomethiazole.⁵⁰

Initial stroke severity is expected to correlate with outcome, but its importance has been underscored by the results of recent trials. In the recently reported citicoline and TOAST (Trial of ORG 10172 in Acute Stroke Treatment) trials, patients with initially mild neurological deficits (NIH Stroke Scale scores of 7 or better) had a high rate of recovery with either placebo or active treatment.^{45 ,61} The results of these trials and previous natural history studies strongly imply that most mildly affected ischemic stroke patients recover and do not need the same intensity of therapy as more severely affected patients.⁶² If many mildly affected patients are included in a treatment trial, potential beneficial effects of therapy might be obscured. Conversely, there is emerging evidence from other trials (the lubeluzole study³⁹ and the NINDS study² ) that severely affected patients with NIH Stroke Scale scores of 22 or more may not respond to treatment.

Major gains have been made toward evolving effective, safe, and appropriately targeted therapies for AIS, and the pace of development has accelerated. As therapy for AIS progresses into combination therapy, carefully conceived and hypothesis-driven clinical trial design and implementation will be even more critical. Exploiting potentially synergistic drug effects and developing and applying imaging technology will aid this development. Many important advances should occur over the next 5 years to maximize improvement for this devastating but now treatable disorder.