The Warfarin-Aspirin Symptomatic Intracranial Disease Study

Study design and sample size estimate.

This multicenter study was designed to test the hypothesis that warfarin reduces the risk of major vascular events (ischemic stroke in any vascular territory, MI, or sudden death) by 50% compared with aspirin in patients with symptomatic, angiographically proven, high-grade stenosis of a major intracranial artery. A retrospective, nonrandomized cohort study design was employed. Calculation of the required sample size was based on the following estimates: the major vascular event rate in patients treated with aspirin would be 26% during a mean follow-up of 2 years compared with 13% in patients treated with warfarin. This estimated event rate on aspirin was based on the rates of stroke and death in patients with intracranial carotid artery or MCA stenosis who were treated with aspirin in the medical arm of the EC-IC bypass study. ^[5,6] With an alpha of 0.05, the sample size required for 80% power was 163 patients per treatment group. Assuming that each center would enter approximately six to 10 patients per year over the period 1985 to 1991, seven centers were recruited to participate in the study.

Patient recruitment, inclusion and exclusion criteria.

Potential candidates for the study were identified by reviewing the reports of consecutive angiograms performed at each center between 1985 and 1991. An attempt was made to retrieve the angiograms of all patients whose reports indicated a ``moderate,'' ``severe,'' or ``more than equals 50%'' intracranial stenosis. The exact degree of stenosis was measured by the local investigator by comparing the diameter of the vessel at the site of stenosis (D stenosis) with the normal diameter of the vessel just distal to the stenosis (D distal) using the following formula <sup>[13]</sup>: Equation 1 Patients whose angiogram reports indicated a ``severe'' or ``more than equals 50%'' intracranial stenosis but whose angiograms were not available for review (only 6% of patients) were still considered potential candidates for the study. The medical records of patients with a measured intracranial stenosis of more than equals 50% or an angiogram report indicating a ``severe'' or ``more than equals 50%'' stenosis were subsequently reviewed.

Inclusion criteria for the study were (1) 50 to 99% stenosis of one of the following major intracranial arteries: intracranial internal carotid artery (ICA), anterior cerebral artery (ACA), the stem or a major division of the MCA, posterior cerebral artery (PCA), posterior inferior cerebellar artery (PICA), vertebral artery, and basilar artery; (2) a TIA or stroke in the distribution of the stenotic artery; and (3) therapy with aspirin or warfarin.

Exclusion criteria were as follows: occlusion of an intracranial artery; extracranial ICA stenosis more than equals 50% proximal to a stenosis of the intracranial ICA, MCA, or ACA; nonatherosclerotic intracranial vasculopathies such as dissection, moyamoya disease, vasculitis; asymptomatic stenosis of a major intracranial artery; distal branch stenosis of an intracranial artery; coexistent cardioembolic source (ie, atrial fibrillation, mitral stenosis, prosthetic valve, MI within 6 weeks, intracardiac clot, ventricular aneurysm, bacterial endocarditis); a severe neurologic deficit from the qualifying stroke or from a stroke occurring during cerebral angiography; stroke preventive treatment other than aspirin or warfarin (eg, EC-IC bypass, angioplasty, warfarin and aspirin concurrently, ticlopidine, dipyridamole alone); and absence of follow-up data (ie, no notes in the chart following angiography and the patient could not be contacted). Extracranial vertebral artery stenosis or occlusion proximal to an intracranial stenosis in the posterior circulation was not an exclusion criterion for the study.

Treatment, follow-up, and end points.

Treatment with aspirin or warfarin was based on local physician preference. The dose of aspirin and level of anticoagulation were also chosen by the local physician. A poll of the participating centers indicated that the most common dose of aspirin prescribed for stroke prophylaxis was 325 mg per day (see Results), and warfarin therapy was typically adjusted to maintain prothrombin times (PT) in the range of 1.2 to 1.6 times control. None of the centers was using the PT international normalized ratio for measuring levels of anticoagulation during the period 1985 to 1991.

Follow-up of most patients was by chart review and telephone or personal interview. If contact with the patient or family could not be made, follow-up was by chart review alone. If the medical records were not available (eg, a patient was admitted to an outside hospital and the records could not be obtained), follow-up was by telephone or personal interview alone. For the primary analysis, patients were followed until an end point occurred, death from a nonvascular cause, date of last contact, or switch in antithrombotic therapy. Primary end points in this study were stroke in any vascular territory, MI, or sudden death. Patients continued to be followed after a switch in antithrombotic therapy, but vascular events occurring thereafter were not considered primary end points.

Safety of warfarin and aspirin was assessed by recording hemorrhagic complications. Hemorrhages were classified as major (fatal hemorrhage, any intracranial hemorrhage, bleeding requiring hospitalization, bleeding requiring transfusion) or minor (any other bleeding complication).

Statistical analysis.

Vascular risk factors, characteristics of the qualifying events, and mean percent stenosis of the intracranial artery were compared in patients treated with warfarin versus patients treated with aspirin. Chisquare tests or Fisher's exact tests were used to compare rates and proportions of categorical variables; t tests were used to compare the means of continuous variables. Cumulative event-free rates for the time to a major vascular event (stroke, MI, sudden death) were estimated by the Kaplan-Meier product limit method, and the two treatment groups were compared by the log rank statistic. The relative risks for the development of a major vascular event were determined using the Cox proportional hazards model that incorporates the duration of follow-up. Since there were relatively few hemorrhagic complications, comparison of these rates between the two treatment groups was performed by fitting a log-linear model, assuming a Poisson distribution and adjusting for duration of follow-up. All comparisons of primary event rates in the two treatment groups were based on an intention-to-treat analysis. All statistical comparisons were two-tailed.

Patient recruitment, qualifying events, and angiographic lesions.

Over 352 patients were screened for the study, but only 151 patients qualified. Reasons for exclusion in the 201 patients for whom these data were collected were intracranial occlusion (51 patients), treatment other than aspirin or warfarin (33 patients), no follow-up data available (27 patients), asymptomatic intracranial stenosis (20 patients), coexistent cardioembolic source (19 patients), nonatherosclerotic intracranial vasculopathy (18 patients), death or severe deficit from qualifying stroke or angiographic complication (12 patients), tandem extracranial carotid stenosis more than equals 50% (10 patients), intracranial stenosis less than 50% (7 patients), and intracranial branch stenosis only (4 patients).

Among the 151 patients who qualified for the study, stroke was the qualifying event in 89 patients (59%) and TIAs were the qualifying events in 62 patients (41%). All but five patients had their qualifying events within 6 months of angiography (most occurred within 1 month of angiography). The stenotic lesions at angiography involved the anterior circulation in 83 patients (55%) and the posterior circulation in 68 patients (45%). The specific arteries involved were the carotid siphon in 40 patients (26%), MCA stem (35) or division (4) in 39 patients (26%), vertebral artery alone in 31 patients (21%), basilar artery alone in 20 patients (13%), tandem basilar artery and vertebral artery in six patients (4%), tandem basilar artery and PCA in two patients (1%), PCA in six patients (4%), ACA in four patients (3%), and PICA in three patients (2%). The percent stenosis of the symptomatic intracranial artery was 50 to 59% in 29 patients (19%), 60 to 69% in 31 patients (20%), 70 to 79% in 22 patients (15%), 80 to 89% in 19 patients (13%), 90 to 99% in 41 patients (27%), ``severe'' in seven patients (5%), and ``more than 50%'' in two patients (1%).

Risk factors: warfarin group versus aspirin group.

Of the 151 patients in the study, 88 patients (58%) were treated with warfarin and 63 patients (42%) were treated with aspirin. The risk-factor profiles of these two groups are shown in Table 1. There were no significant differences between the two groups in the rates of traditional vascular risk factors, previous stroke, TIA or stroke as the qualifying event, treatment with aspirin at the time of qualifying event, anterior circulation stenosis, or posterior circulation stenosis. Additionally, the mean age and mean percent stenosis of the symptomatic intracranial artery of both groups were similar Table 1. The percentage of patients who had a measured intracranial stenosis of 80 to 99% or an angiogram report indicating a ``severe'' intracranial stenosis was 49% (43 of 88) in the warfarin group and 38% (24 of 63) in the aspirin group.

End points: warfarin group versus aspirin group.

At the end of the study (June 1993), 60 patients (40%) were alive and did not have an end point, 40 patients (26%) had an end point, 38 patients (25%) had their stroke preventive therapy switched, seven patients (5%) could not be contacted but were alive without an end point at the last clinic visit, four patients (3%) had died of nonvascular causes, and two patients (1%) had died from hemorrhages (see Hemorrhagic complications).

Of 63 patients treated with aspirin for a median duration of 19.3 months, 15 patients (24%) had a stroke (12 nonfatal, 3 fatal) and 11 patients (17%) had an MI or sudden death (5 nonfatal MI, 6 fatal MI/sudden death). The overall rate of major vascular events in the aspirin group was 18.1 per 100 patient-years of follow-up. Of the 15 strokes in the aspirin group, nine were in the same vascular territory as the stenotic intracranial artery and six were in a different territory.

In comparison, of 88 patients treated with warfarin for a median duration of 14.7 months, six patients (7%) had an ischemic stroke (5 nonfatal, 1 fatal) and eight patients (9%) had an MI or sudden death (3 nonfatal MI, 5 fatal MI/sudden death). The overall rate of major vascular events in the warfarin group was 8.4 per 100 patient-years of follow-up. Of the six strokes in the warfarin group, five were in the same territory as the stenotic intracranial artery and one was in a different territory. A PT on admission was available in five of 14 patients in the warfarin group who had an end point. These PTs were 10.9 seconds, 15.0 seconds, 15.5 seconds, 16.7 seconds, and 18.0 seconds.

The Kaplan-Meier curves illustrating the proportion of patients in each treatment group who survived free of a major vascular event (stroke, MI, or sudden death) during the study are shown in the Figure 1. These curves show that a substantially higher proportion of patients treated with warfarin survived free of a major vascular event compared with patients treated with aspirin (p less than 0.01, log rank test). Cox proportional hazards analysis showed that the relative risk of stroke, MI, or sudden death in patients treated with warfarin was 0.46 (95% CI, 0.23 to 0.86) compared with patients treated with aspirin.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Graph shows Kaplan-Meier estimates of proportion of patients remaining free of ischemic stroke, myocardial infarction, or sudden death in those treated with warfarin (upper curve) or aspirin (lower curve).

Hemorrhagic complications: warfarin group versus aspirin group.

Hemorrhagic complications occurred in 13 patients on warfarin (3 major, 10 minor) during 166 patient-years of follow-up and in two patients on aspirin (both minor) during 143 patient-years of follow-up (p less than 0.01, log rank test). The three major hemorrhagic complications in patients treated with warfarin were a fatal intracranial parenchymal hematoma (PT 22 seconds on the day of the hemorrhage), a fatal gastrointestinal hemorrhage (PT 12 seconds, partial thromboplastin time 56 seconds--the patient was on heparin after cerebral angiography and warfarin had just been started), and a nonfatal gastrointestinal bleed requiring transfusion (PT 22 seconds on the day of the hemorrhage). Overall, the combined minor and major hemorrhage rates were 7.8/100 patient-years of follow-up in the warfarin group and 1.4/100 patient-years of follow-up in the aspirin group (p less than 0.01, log rank test). The rates of major bleeding complications were 1.8/100 patient-years of follow-up in the warfarin group and 0/100 patient-years in the aspirin group. A Kaplan-Meier analysis that included treatment-related deaths as a study end point (ie, in addition to ischemic stroke, MI, or sudden death) showed that a significantly higher proportion of patients treated with warfarin survived free of these end points compared with patients treated with aspirin (p equals 0.03, log rank test).

Relationship of site and severity of stenosis to outcome.

The relationship of site of intracranial stenosis to outcome is shown in Table 2. Within each treatment group, patients with posterior circulation stenosis had a higher rate of stroke in the same territory as the stenotic vessel than patients with anterior circulation stenosis. Within the anterior circulation, patients treated with aspirin had higher rates of stroke in the same territory as the stenotic vessel and higher rates of MI/sudden death than patients treated with warfarin. In the posterior circulation, patients treated with aspirin had higher rates of stroke in the same territory as the stenotic vessel and higher rates of stroke in a different vascular territory than patients treated with warfarin; however, patients were treated with warfarin for a substantially shorter period. Despite the shorter period of treatment with warfarin, patients with posterior circulation stenosis on warfarin had a slightly higher rate of MI/sudden death than patients with posterior circulation stenosis who were treated with aspirin Table 2.

Of 84 patients with 50 to 79% intracranial stenosis, five patients (6%) had a stroke in the same territory as the stenotic artery during follow-up (4 of 39 [10%] on aspirin, 1 of 45 [2%] on warfarin). In comparison, of 67 patients with 80 to 99% or ``severe'' intracranial stenosis, nine patients (13%) had a stroke in the same territory as the stenotic artery during follow-up (5 of 24 [21%] on aspirin, 4 of 43 [9%] on warfarin).

Aspirin dose and end points.

Of 63 patients treated with aspirin, 54 patients were prescribed 325 mg or less per day (48 patients, 325 mg/d; 5 patients, 81 mg/d; 1 patient, 160 mg/d), six patients were prescribed more than 325 mg per day, and in three patients the dose was unknown. Of the 54 patients treated with less than equals 325 mg of aspirin per day, 15 patients (28%) had a stroke and eight patients (15%) had an MI or sudden death. Of the six patients treated with more than 325 mg of aspirin per day, none had a stroke, but two patients (33%) had an MI or sudden death.

Events after therapy switch.

During follow-up, antithrombotic therapy was switched from warfarin to aspirin in 18 patients and from aspirin to warfarin in seven patients. Of the 18 patients switched from warfarin to aspirin, 12 were switched after they had been asymptomatic for 3 to 6 months on warfarin, three were switched after minor hemorrhagic complications, and three were switched because of newly diagnosed cancer (2 patients) or expense (1 patient). Of the seven patients switched from aspirin to warfarin, four were switched because of recurrent TIAs and three were switched for unspecified reasons.

In 18 patients whose therapy was switched from warfarin to aspirin, six patients (33%) had a stroke (4 patients) or MI/sudden death (2 patients), three patients (17%) had TIAs, and one patient (6%) had a major gastrointestinal hemorrhage during a median follow-up of 16 months after the switch to aspirin. In seven patients whose therapy was switched from aspirin to warfarin, none of the patients had a stroke or MI/sudden death, one patient (14%) had recurrent TIAs (although substantially fewer than when he was on aspirin), and one patient (14%) had a major nose bleed that required surgery during a median follow-up of 14.5 months after the switch to warfarin. Since these events occurred after a switch in therapy, they were not included as primary end points in the study.