Early Stroke Risk After Transient Ischemic Attack Among Individuals With Symptomatic Intracranial Artery Stenosis FREE

Several cohort studies have shown that early (≤3 months) risk of stroke following index transient ischemic attack (TIA) is much higher than previously thought.^{1 - 5} Although comparisons across studies are limited by heterogeneity of study populations and clinical settings, they have shown a 90-day stroke risk of 10% to 20% after TIA.⁶ Furthermore, several clinical and imaging models were derived and validated to predict risk of early stroke after TIA.⁷ These models are important considering the unacceptably high proportion of patients with TIA (vs patients with stroke) discharged from the emergency department whose conditions are underinvestigated and undertreated during the period of highest risk for stroke.⁸

Most of these studies included all-cause TIAs; therefore, limited data exist about the effect of specific TIA mechanisms on early prognosis following TIA. Indeed, a powerful predictor of early secondary stroke is large-artery atherosclerosis.⁹ A recent analysis by the North American Symptomatic Carotid Endarterectomy Trial (NASCET) Group¹⁰ revealed that the 90-day risk of ipsilateral stroke was higher after TIA than after hemispheric stroke. However, no data exists about the early risk of stroke following TIA vs stroke in patients with intracranial arterial stenosis, to our knowledge. The Warfarin-Aspirin Symptomatic Intracranial Disease (WASID) study¹¹ provided an opportunity to explore this issue. In this article, we aim to quantify the early risk of ischemic stroke in the territory of a stenotic intracranial artery after TIA and to identify clinical and imaging features associated with increased risk of stroke in the territory among patients with TIA.

The WASID Study Group trial was a prospective multicenter, double-blind, randomized clinical trial conducted at 59 sites with institutional review board approval for the study and was designed to compare warfarin sodium treatment with aspirin for preventing stroke or vascular death in patients with TIA or nondisabling ischemic stroke attributable to angiographically verified 50% to 99% atherosclerotic stenosis of a major intracranial artery and no cardiac source of embolism (eg, atrial fibrillation). Randomization occurred within 90 days of the index event. Details of the WASID Study Group trial design and results of the comparison between warfarin treatment and aspirin have been published previously.^{11 - 12}

In the WASID Study Group trial, a qualifying TIA was defined as a focal neurologic deficit lasting less than 24 hours regardless of the presence or absence of a lesion by imaging. The ischemic stroke end point in the WASID Study Group trial was defined as a new focal neurologic deficit of sudden onset that lasted at least 24 hours and was not associated with bleeding seen on computed tomography or magnetic resonance imaging of the brain. Independent panels of neurologists and cardiologists adjudicated outcomes. Demographics and medical history (before the WASID Study Group trial qualifying event) were self-reported by the patient.^{11 - 12}

Because no difference in efficacy was discerned between the 2 treatment arms in the WASID Study Group trial, subjects in both arms were combined in this analysis. Patients with stroke accompanied by TIA as the qualifying event were analyzed separately to facilitate direct comparison between stroke alone and TIA alone. The time from the qualifying event to randomization was compared between patients with TIA alone vs stroke alone using Wilcoxon rank sum test. We calculated the cumulative risk of ischemic stroke in the territory of the symptomatic artery during the first 90 days after randomization using the Kaplan-Meier method, with events or last follow-up visits occurring after 90 days considered censored at 90 days. We chose the date of randomization as the starting point because, by definition from the WASID Study Group trial protocol, no subject had a stroke end point between the time of the qualifying event and the day of randomization.^{11 - 12} We selected ischemic stroke in the territory of the symptomatic artery as our primary end point for this study to determine the risk specific to symptomatic intracranial stenosis. However, we also present results for any ischemic stroke and for a major ischemic vascular event (ischemic stroke, myocardial infarction, or vascular death) within 90 days of randomization. The time to event for each of these outcomes was compared between patients with TIA alone vs with stroke alone as the qualifying event using log-rank test and Cox proportional hazards regression analysis. Hazard ratios comparing TIA alone and stroke alone are calculated as the risk of event for TIA relative to stroke.

Among patients whose qualifying event was TIA alone, we evaluated the association between various factors and stroke in the territory of the symptomatic artery during the first 90 days after randomization. These factors were selected on the basis of prior WASID Study Group investigations that identified predictors of ischemic stroke after symptomatic intracranial arterial stenosis, including sex, severe stenosis of 70% or greater, serum white blood cell count, and time to randomization after the qualifying event.¹³ Other factors were chosen based on their known clinical association with stroke outcomes and included age, symptom type, blood pressure, diabetes mellitus, location of stenosis, coronary artery disease, presence of infarcts on neuroimaging, and history of stroke (before the WASID Study Group trial qualifying event). These associations were assessed using log-rank test and Cox proportional hazards regression analysis. Similar analyses were performed for patients whose qualifying event was stroke alone. In addition, Cox proportional hazards regression analysis was used to assess the relationship between the qualifying event and stroke in the territory of the symptomatic artery, controlling for baseline characteristics. All reported P values are 2-sided, without adjustment for multiple testing.

Of 569 patients enrolled in the WASID Study Group trial, 222 had TIA alone, 241 had stroke alone, and 106 had stroke combined with TIA as their qualifying events. The median time from the qualifying event to randomization was 17 days, and the mean follow-up was 1.9 years. There was no difference in median times from the qualifying event to randomization between the subjects enrolled with TIA alone (18 days) vs stroke alone (16 days) (P =.99). The 90-day risk of ischemic stroke in the arterial territory was 6.9% after TIA alone compared with 4.7% after stroke alone (hazard ratio, 1.5; 95% confidence interval [CI], 0.7-3.2; P =.32) (Figure). Among patients whose qualifying event was stroke combined with TIA, 13 (12.3%) experienced stroke in the arterial territory within the first 90 days. Rates for other time points among patients whose qualifying events were TIA alone, stroke alone, and stroke combined with TIA are given in Table 1.

Among patients with TIA alone as the qualifying event, 60.0% (15 of 25) of strokes in the arterial territory occurred in the first 90 days compared with 34.4% (11 of 32) among patients with stroke alone as the qualifying event (P =.05). This early disadvantage for patients with TIA was not reflected in the 1-year rate of stroke in the arterial territory (7.9% [95% CI, 5.0%-12.4%] for TIA alone vs 11.7% [95% CI, 8.0%-16.8%] for stroke alone).

Among patients with TIA alone, only the presence of cerebral infarct on baseline neuroimaging was associated with a significantly higher risk of early stroke (Table 2). Age, sex, blood pressure, study treatment (aspirin vs warfarin), location or percentage of arterial stenosis, and ischemic stroke or diabetes mellitus history were not predictors of higher ischemic stroke risk in the arterial territory. None of the symptoms of TIA (including visual, speech, sensory, weakness, vestibular or cerebellar, and others) were predictive of ischemic stroke in the territory of the symptomatic artery during the first 90 days after randomization. Univariate analyses relating time until ischemic stroke in the arterial territory to the same baseline characteristics as in Table 2 for patients with stroke alone as the qualifying event were significant only for time from the qualifying event to randomization (ie, more subjects with stroke as their qualifying event were enrolled within 17 days of their event as opposed to later [8% vs 1%, P =.006]).

The 90-day risk of any ischemic stroke was 6.9% (95% CI, 4.2%-11.2%) after TIA alone compared with 7.6% (95% CI, 4.8%-11.8%) after stroke alone (hazard ratio, 0.9; 95% CI, 0.5-1.8; P = .77). Among patients with TIA alone, all ischemic strokes in the first 90 days were in the territory of the stenotic intracranial artery. The 90-day risk of major vascular event was 7.4% (95% CI, 4.6%-11.7%) after TIA alone compared with 9.6% (95% CI, 6.5%-14.2%) after stroke alone (hazard ratio, 0.8; 95% CI, 0.4-1.4; P =.38).

When we assessed the relationship between the qualifying event and stroke in the arterial territory, controlling for baseline characteristics individually, the qualifying event was not significant in any of these analyses except for the presence of cerebral infarcts on neuroimaging. For stroke in the arterial territory, there was a significant interaction between the qualifying event and the presence of cerebral infarcts on neuroimaging, indicating that the relationship between stroke in the arterial territory and the qualifying event was different depending on whether infarcts were present on brain imaging. This interaction is summarized in Table 3; if infarcts were present on brain imaging, there was a significantly higher risk of stroke in the arterial territory if the qualifying event was TIA.

Although the early incidence of stroke in the symptomatic arterial territory among individuals with intracranial arterial stenosis was numerically higher among those who experienced TIA alone compared with those who experienced ischemic stroke alone, this difference did not reach statistical significance. Nevertheless, the number of outcome events in our study was low and may have compromised our ability to detect a statistically significant difference between the groups. Indeed, if we assume 4.7% to be the true event rate at 90 days for patients with stroke alone as the qualifying event, the true event rate for patients with TIA alone would have to be 10.4% for the study to have had 80% power to detect a difference with the sample sizes we had. However, among subjects with TIA alone as the qualifying event, almost two-thirds of strokes in the arterial territory occurred within the first 90 days compared with one-third of strokes in the arterial territory among subjects with ischemic stroke alone as the qualifying event.

Overall, our results add to the current literature about early risk of stroke following TIA in underscoring the importance of timely management of patients with TIA.^{14 - 16} The current perception among many practitioners is that checking for intracranial stenosis in a given patient with TIA would not alter the timeliness of prescribed stroke prevention management strategies. However, these data suggest that it may be worthwhile to include imaging of intracranial vasculature as part of the workup of patients with TIA so that patients whose TIAs are due to intracranial atherosclerosis can receive early and aggressive medical management to reduce the risk of stroke within 90 days of their index event.

As noted, our risk estimate of stroke after TIA is lower than that found in other studies^{1 - 5} examining early risk of stroke after TIA, but there may be several reasons for this potential underestimation of actual stroke risk. First, we counted the 90-day period from the time of randomization, not from the time of event. Because the median time from the index event to randomization was 17 days, it is conceivable that patients at higher risk were excluded because they were not randomized owing to early recurrence. However, we did this to avoid any bias from incorporating the requisite stroke-free interval per the WASID Study Group trial protocol between the time of the qualifying event and randomization into our recurrent stroke risk assessment. Second, several other studies^{1 - 5} have included patients with all-cause TIA drawn from community hospitals and not the generally more motivated patients exposed to the intense scrutiny and close management associated with a clinical trial such as the WASID Study Group. Third, patients having stroke combined with TIA as their qualifying event had the highest risks of recurrent stroke at 3 months and at 1 year (Table 1), and when these patients were included as having had stroke as the qualifying event in a previous analysis of the WASID Study Group trial data,¹³ patients having stroke had numerically but not significantly higher risk of stroke in the arterial territory at 1 and 2 years compared with patients having TIA alone.

In contrast to other TIA cohorts, we did not find that most previously identified risk factors for recurrent stroke risk after TIA, including age, blood pressure, diabetes mellitus, and clinical features,⁷ were predictive in our study, although not all studies have shown those factors to be predictive of stroke after TIA,^{17 - 18} and it is conceivable that our analysis may have lacked the power to detect these risk factor associations. As in NASCET,¹⁰ we did not find early risk of stroke to be affected by the degree of arterial stenosis, but we found that the presence of cerebral infarcts on neuroimaging was a strong harbinger of stroke risk in the arterial territory after TIA, which is in accord with the results of other studies.^{10 ,19 - 20} Cerebral infarcts on baseline neuroimaging did not confer a higher risk of early stroke in the arterial territory among subjects enrolled in the WASID Study Group trial with stroke alone, unlike among subjects enrolled with TIA alone. This differential role of baseline neuroimaging in predicting short-term recurrent stroke risk among patients with TIA, and not among patients with stroke and intracranial atherosclerosis, suggests that the patients with TIA were experiencing multiple prior cerebrovascular events, perhaps caused by a more unstable source of thromboembolism. This notion of greater atherosclerotic plaque instability in the TIA vs stroke cohorts in the WASID Study Group trial might be further supported by the earlier incidence of subsequent stroke that we noted in the patients with TIA.

This study has important limitations, including the post-hoc analysis with limited statistical power. However, the multicenter design, rigorous methods, and quality control of the WASID Study Group trial point to the study's strengths.

In conclusion, among a cohort of patients with intracranial stenosis, subjects with an index TIA alone had at least as high a risk of stroke in the symptomatic arterial territory within 90 days as patients with an index stroke alone. Furthermore, most subsequent strokes in the arterial territory of a stenotic intracranial artery occurred within 90 days of the index event among patients with TIA alone and among those with stroke alone. These data highlight the importance of prompt diagnosis and management of intracranial stenosis in patients with TIA, particularly those with cerebral infarction on brain imaging.

Correspondence: Bruce Ovbiagele, MD, Stroke Center and Department of Neurology, UCLA Medical Center, University of California at Los Angeles, 710 Westwood Plaza, Los Angeles, CA 90095 (Ovibes@mednet.ucla.edu).

Accepted for Publication: January 3, 2008.

Author Contributions:Study concept and design: Ovbiagele, Cruz-Flores, Lynn, and Chimowitz. Acquisition of data: Ovbiagele, Cruz-Flores, and Chimowitz. Analysis and interpretation of data: Ovbiagele, Cruz-Flores, Lynn, and Chimowitz. Drafting of the manuscript: Ovbiagele, Cruz-Flores, Lynn, and Chimowitz. Critical revision of the manuscript for important intellectual content: Ovbiagele, Cruz-Flores, Lynn, and Chimowitz. Statistical analysis: Lynn. Obtained funding: Chimowitz. Administrative, technical, and material support: Chimowitz. Study supervision: Chimowitz.

Group Information: A list of the WASID Study Group members was published in N Engl J Med. 2005;352(13):1316.

Financial Disclosure: None reported.

Funding/Support: This study was funded by grant 1R01 NS36643 from the National Institute of Neurological Disorders and Stroke (Dr Chimowitz). The following general clinical research centers, funded by the National Institutes of Health, provided local support for the evaluation of patients in the WASID Study Group trial: Emory University (grant M01 RR00039), Case Western University MetroHealth Medical Center (grant 5M01 RR00080), San Francisco General Hospital (grant M01 RR00083-42), The Johns Hopkins University School of Medicine (grant M01 RR000052), Indiana University School of Medicine (grant 5M01 RR000750-32), Cedars-Sinai Hospital (grant M01 RR00425), and the University of Maryland (grant M01 RR165001).