Long-Term Mortality and Recurrent Stroke Risk Among Chinese Stroke Patients With Predominant Intracranial Atherosclerosis
Background and Purpose— The goal of this study was to document the long-term outcome of ischemic stroke patients in a population with predominant intracranial atherosclerosis and risk factors for a recurrent event.
Methods— Intracranial and extracranial arteries of consecutive patients with acute ischemic stroke were studied prospectively with transcranial Doppler and duplex ultrasound. All patients were followed up regularly for the development of recurrent stroke, cardiac event, or death.
Results— We included 705 patients with acute ischemic stroke, of whom 345 were documented ultrasonographically as having large-artery lesions. The follow-up period was up to 42 months (mean, 28±5 months). One hundred seventeen patients (17%) died of any cause, and 199 (28%) suffered further cerebrovascular cardiac events. The 3.5-year cumulative mortality rate was 20.8%; for cerebrovascular event, it was 29.5%. The annual recurrent stroke rates during the first year were 10.9% for patients without vascular lesion, 17.1% for intracranial atherosclerosis only, and 24.3% for both intracranial and extracranial atherosclerosis; for the second year, the rates were 7.5%, 8.6%, and 7.7%, respectively. More occurrence of death (log rank, 5.19; P=0.02) or cerebrovascular event (log rank, 9.68; P=0.002) was found among patients with than those without vascular lesions. Patients with both intracranial and extracranial arterial lesions were at highest risk of death (log rank, 9.64; P=0.008) and cerebrovascular event (log rank, 11.56; P=0.003). When death and further vascular event were combined as poor outcomes in a Cox proportional-hazards regression model, number of abnormal arteries, advanced age, diabetes, atrial fibrillation, and previous stroke were significant predictors.
Conclusions— Patients with intracranial atherosclerosis, especially coexisting extracranial carotid disease, are at higher risk of suffering death or further vascular event. Our findings provide important data for planning future randomized clinical trials for this high-risk group of stroke patients.
Stroke is the third-leading cause of death and a major cause of disability in adults, and it is often more disabling than fatal. Primary stroke prevention programs such as traditional risk factor reduction have achieved a reduction in the number of incident strokes. However, with a decline in mortality from initial cerebral infarction and an increase in the life expectancy of the population, the number of patients with recurrent stroke and subsequent cardiovascular events will become greater. Thus, it is important to identify those stroke patients at high risk for stroke recurrence and death.
Carotid atherosclerotic stenosis has been listed as an important risk factor for stroke.1,2 In contrast, intracranial atherosclerosis is not often mentioned as a risk factor associated with recurrent stroke. This is due in part to the lack of long-term outcome data of intracranial atherosclerosis. This study aimed at documenting the annual risk of recurrent stroke and mortality among patients with intracranial atherosclerosis. The short-term outcomes were reported earlier.3
Subjects and Methods
The present study was performed at the Prince of Wales Hospital in Hong Kong, a regional general hospital with accident and emergency services. The protocol was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong.
In this prospective study, we recruited consecutive patients presented with cerebral ischemia within 7 days of symptom onset. The methods were described in detail in an earlier publication, which also described the short-term outcome.3 Briefly, we recruited patients admitted with acute cerebral ischemia, including transient ischemic attack (TIA) and cerebral infarct. All patients underwent a CT scan to rule out intracranial hemorrhage. On admission, baseline data, including age, sex, medical history, and physical examination, were collected. Scoring of prestroke disability was estimated with the modified Rankin scale (mRS), and severity of neurological deficits of index stroke was assessed with the National Institutes of Health Stroke Scale (NIHSS). Vascular risk factors were noted, particularly in patients with any history of smoking, hypertension, diabetes mellitus, ischemic heart disease, atrial fibrillation, and previous TIA or stroke. Blood biochemistry, blood count test, ECG, and chest x-ray were checked routinely.
Transcranial Doppler (TCD) examination was performed on each enrolled patient to document the cerebral arterial condition within 1 week of symptom onset. Using a standardized protocol based on a principle previously described,4 we studied the intracranial large arteries through the temporal, occipital, and orbital windows. Briefly, we examined the following arteries in 4-cm increments: middle cerebral artery, anterior cerebral artery, posterior cerebral artery, siphon internal carotid artery, and vertebrobasilar artery. We also examined the bilateral extracranial carotid arteries. The diagnostic criteria for arterial stenosis or occlusion were given earlier.3 Cerebral arteries that could not be insonated because of poor temporal acoustic windows were regarded as nonocclusive.
All survivors were followed up regularly by telephone or personal interview, in addition to routine medical outpatient clinic attendance. If patients could not be reached at this follow-up, we found their medical records and tried to contact their relatives for more information about the patient’s condition. All telephone and personal interviews were performed by a stroke specialty nurse using a standardized questionnaire.
The specified outcomes were defined as occurrence of a further vascular event (including TIA, stroke, or acute coronary syndrome) or death. The long-term combined poor outcome was regarded as occurrence of either death or a further vascular event. If death and a further vascular event (supposed cause of death) were recorded as occurring simultaneously, 1 event was categorized into the analysis. For example, if a patient died of acute myocardial infarction, we regarded the event as death in the mortality analysis and as a further vascular event in the further-vascular-event analysis. The nature of the vascular event and the cause of death were based on medical records and death certificates when possible. Recurrent events were based of the diagnosis on the discharge summary according to the treating physician. The definite and detailed description of the cause of death by the patient’s relative was used if the cause could not be obtained by other means. Obscure or indefinite information about cause of death was classified as unknown cause.
All data were entered into the SPSS software (version 9.0 for Windows) for storage and analysis. GraphPad InStat software (version 3.02 for Windows) was used as a complementary way to perform a χ2 test of a larger contingency table. Statistical significance was considered at P=0.05 and was 2 sided.
In the present study, age and number of occlusive arteries were used as continuous variables. Sex, vascular risk factors, and end points of long-term outcome were used as categorical variables. We also categorized the scores of mRS and NIHSS as categorical variables. The scores of mRS were dichotomized as 0 to 2 and 3 to 5, and NIHSS scores were grouped as 0 to 1, 2 to 8, and ≥9.
In univariate analyses, independent-sample t test and χ2 test or Fisher’s exact test was used. We also applied the χ2 test for trend to explore the relationship between the number of occlusive arteries and occurrence of a further vascular event or death. In analyses of predictors of the long-term clinical outcome, Cox proportional-hazards regression function was used to estimate impact in terms of risk ratios of possible determinants of survival and further vascular events, taking the time variable into consideration. Hazard ratio (HR), 95% confidence interval (CI), and probability value were observed. In survival analyses, the Kaplan-Meier product-limit method was used to estimate survival condition. The log-rank test was used to compare rate estimates.
A total of 705 patients admitted with acute cerebral ischemia were recruited. Their mean age on admission of index stroke was 67.6±11.9 years, with 67% ≥65 years of age; 58% (409 patients) were men. Vascular risk factors were common among these patients. Hypertension (58%), history of smoking (44%), and previous stroke or TIA (33%) were the 3 most common vascular risk factors. Atrial fibrillation was found in 38 patients (5%). Of 705 patients, 345 patients (49%) were documented ultrasonographically with cerebral arterial lesions in the initial TCD examination; the other 360 (51%) had no vascular lesion identified. Among the 345 patients with vascular lesions, 258 (75%) had intracranial vascular lesions only, 71 (21%) had both intracranial and extracranial lesions, and 16 (4.6%) had extracranial lesions only. The clinical characteristics of patients were reported in detail previously.3
We followed up this cohort for a mean period of 28±5 months (up to 42 months) after stroke onset. Fourteen patients (2%) were lost to follow-up at the end of the study and could not be reached. The duration between the index stroke and last contact date of these 14 patients ranged from 6.3 to 17.8 months. In 8 patients, we had follow-up data of >1 year after the index stroke.
A total of 117 patients (17%) died of any cause (including vascular and nonvascular), and 199 patients (28%) suffered a further vascular event (including fatal and nonfatal). Overall, 691 patients survived the index stroke, with 448 patients alive free of a further vascular event. At the end of follow-up, 85% of patients remained on regular antiplatelet agent (78.5%) or anticoagulant (6.2%). The remaining 15% were not on any antithrombotic treatment mainly because of bleeding contraindications such as gastrointestinal bleeding and intracerebral hemorrhage, which were common in Chinese patients.
Occurrence of Death
Among the 117 patients who died, 59 patients (50.4%) died of vascular causes (including 43 patients of further vascular events and 14 of the index stroke), and 32 patients (27.4%) died of unknown causes. Cerebrovascular events were more commonly the cause of death than cardiac events (27 versus 18, respectively). Table 1 shows the detailed underlying causes of death.
Of the 705 patients, the 3.5-year cumulative mortality rate was 20.8% (95% CI, 16 to 25.6) by the Kaplan-Meier product limit method. The cumulative mortality rate was 5.7% (95% CI, 4 to 7.4) at 6 months, 8.9% (95% CI, 6.8 to 11.0) by 1 year, 14.9% (95% CI, 12.2 to 17.6) by 2 years, and 20.8% (95% CI, 16 to 25.6) by 4 years. The risk of death was greatest in the first year, especially at 6 months, and decreased over the next 2 years (Table 2).
Slightly more deaths were found among patients with vascular lesions (66 of 345 patients, 19%) than those without (51 of 360 patients, 14%). Considering the number of occlusive arteries, the higher occurrence of death was found among patients with more arterial involvement (P=0.02 by χ2 test for trend). The cumulative mortality rate was 7.2% (95% CI, 4.5 to 9.9) by 1 year, 12.4% (95% CI, 9.0 to 15.8) by 2 years, and 15.9% (95% CI, 11.5 to 20.3) by 3 years for patients without vascular lesion and 11% (95% CI, 7.7 to 14.3), 17.8% (95% CI, 13.6 to 22.0), and 32.4% (95% CI, 15.6 to 49.2), respectively, for patients with vascular lesions.
The Kaplan-Meier product limit survival function showed a significantly higher mortality rate for patients with than for those without vascular lesions (log rank, 5.19; P=0.02). In addition, patients with atrial fibrillation also showed a significantly higher mortality rate than those without atrial fibrillation during the long-term follow-up (log rank, 22.4; P<0.0001). Taking the extracranial and intracranial distribution of arterial lesions into consideration, we found that patients with both intracranial and extracranial arterial lesions had the highest mortality rate (log rank, 9.64; P=0.008 by Kaplan-Meier product limit survival function; Figure 1A and Table 2). Because a small number of patients (n=16) had extracranial carotid lesions only, we excluded these patients in this analysis.
Univariate analysis showed that age (P<0.0001), atrial fibrillation (P<0.0001), and number of occlusive arteries (P=0.02) were related to long-term mortality. After adjustment for age, sex, prestroke mRS, NIHSS, hypertension, diabetes, smoking, ischemic heart disease, atrial fibrillation, previous stroke or TIA, and number of occlusive arteries, number of occlusive arteries (HR, 1.16; 95% CI, 1.06 to 1.28; P=0.002) remained the significant predictor of long-term mortality in Cox proportional-hazard regression model. Other independent predictors included advanced age (HR, 1.08; 95% CI, 1.05 to 1.10; P<0.0001) and atrial fibrillation (HR, 2.16; 95% CI, 1.25 to 3.73; P=0.006).
Occurrence of Further Vascular Event
Of the 705 patients, a first major vascular event after index stroke occurred among 199 patients (28%). The major vascular events included 33 TIAs, 120 strokes, and 46 acute coronary syndromes. Of 199 further vascular events, 38 were fatal, including 22 cerebrovascular events and 16 acute coronary syndromes (Table 3). Cerebrovascular events (including TIA and stroke) accounted for 77% of 199 further vascular events.
The estimated 3.5-year cumulative rate of cerebrovascular event was 29.5% (95% CI, 23 to 36) by the Kaplan-Meier product limit method. The cumulative cerebrovascular event rate was 8.8% (95% CI, 6.7 to 10.9) at 6 months, 13.7% (95% CI, 11.1 to 16.3) by 1 year, 21.2% (95% CI, 18.0 to 24.2) by 2 years, and 29.5% (95% CI, 23 to 36) by 3 years. The risk of developing a further cerebrovascular event was greatest in the first year, especially the first 6 months, after a stroke. The annual rate of cerebrovascular event tended to decrease over 2 years (Table 2).
Significantly more patients with than without vascular lesions suffered a further cerebrovascular event (88 versus 65, P=0.02 by χ2 test).
Considering the number of occlusive arteries, a greater occurrence of death was found among patients with more arterial involvement (P=0.004 by χ2 test for trend) The cumulative rate of cerebrovascular event was 10.9% by 1 year, 17.6% by 2 years, and 21.2% by 3 years for patients without vascular lesions and 17.4%, 25.0%, and 46.8%, respectively, for patients with vascular lesions. The Kaplan-Meier product limit survival function showed a significantly higher rate of occurrence of cerebrovascular event in patients with than in those without vascular lesions (log rank, 9.68; P=0.002). Taking the vascular lesion distribution into account, we found patients with both intracranial and extracranial arterial lesions had the highest cerebrovascular event rate (log rank, 11.56; P=0.003 by Kaplan-Meier product limit survival function; Figure 1B and Table 3). Survival function also showed that the occurrence of a further cerebrovascular event during long-term follow-up was significantly high among patients with a history of diabetes (log rank, 4.24; P=0.04), atrial fibrillation (log rank, 4.06; P=0.04), and previous stroke or TIA (log rank, 13.29; P=0.0003). Univariate analysis showed that age (P=0.03), previous stroke or TIA (P<0.0001), and number of occlusive arteries (P=0.004) were related to long-term occurrence of further cerebrovascular event. After adjustment in a Cox proportional-hazards regression model, number of occlusive arteries (HR, 1.15; 95% CI, 1.06 to 1.25; P=0.0006), age (HR, 1.02; 95% CI, 1.00 to 1.03; P=0.02), atrial fibrillation (HR, 1.98; 95% CI, 1.05 to 3.72; P=0.03), and previous stroke or TIA (HR, 1.68; 95% CI, 1.21 to 2.33; P=0.002) were predictive of further cerebrovascular event.
Predictors for Combined Poor Outcome
Both death and subsequent vascular event represent poor outcomes after stroke. When death and subsequent vascular event were combined, a total of 259 patients (37%) of the cohort suffered a poor outcome. The poor outcomes included 26 deaths resulting from nonvascular causes (3.7%), 27 deaths resulting from recurrent stroke (3.8%), 18 deaths resulting from coronary events (2.6%), 14 deaths resulting from index stroke (2%), 32 deaths of unknown causes (4.5%), and 142 subjects alive with subsequent vascular events (20%). The occurrence of combined poor outcome increased notably with an increase in number of occlusive arteries (P=0.005 by χ2 for trend).
Univariate analysis showed that occurrence of combined poor outcome was associated with advanced age (P<0.0001), history of diabetes (P=0.03), ischemic heart disease (P=0.03), atrial fibrillation (P<0.0001), and previous stroke or TIA (P=0.02). No association was found between long-term poor outcome and severity of index stroke, sex, hypertension, and smoking history. After adjustment for a total of 11 variables (age, sex, prestroke mRS, NIHSS, history of hypertension, diabetes, ischemic heart disease, atrial fibrillation, smoking, previous stroke or TIA, and number of occlusive arteries) in a multivariate Cox proportional-hazards regression model, the number of abnormal arteries (HR, 1.13; 95% CI, 1.05 to 1.20; P=0.0004) remained significantly predictive for long-term combined poor outcome, as were advanced age (HR, 1.03; 95% CI, 1.02 to 1.04; P<0.0001), diabetes (HR, 1.34; 95% CI, 1.01 to 1.77; P=0.04), atrial fibrillation (HR, 2.68; 95% CI, 1.77 to 4.07; P<0.0001), and previous stroke or TIA (HR, 1.30; 95% CI, 1.00 to 1.68; P=0.04). Table 4 summarizes the predictors for long-term combined poor outcome.
There has been no large prospective study on mortality rate and recurrence among stroke patients with intracranial atherosclerosis. Intracranial large-artery occlusive disease has been reported to be more predominant in Chinese than white stroke patients.5–8 This study is the largest prospective study to take into consideration the intracranial and extracranial distribution of vascular lesions. We attempted to document long-term prognosis and significance of distribution of large-artery occlusive disease on mortality and recurrence. Patients with vascular lesions had a higher mortality rate and a higher risk of a further vascular event than patients without vascular lesions. Patients with both intracranial and extracranial vascular lesions had a significantly worse outcome than patients with intracranial lesions only. Furthermore, in a multivariate Cox proportional-hazards regression model, the risk of death, subsequent cerebrovascular event, or acute coronary syndrome after stroke increased significantly with 1 more arterial involvement (HR, 1.16; 95% CI, 1.06 to 1.28). Our finding is in agreement with a meta-analysis9 that documented that patients with intracranial carotid stenosis or occlusion had a higher rate of recurrent stroke (rate ratio, 1.09; 95% CI, 1.05 to 1.14) or vascular mortality (rate ratio, 1.10; 95% CI, 1.06 to 1.14). Because most patients (85%) were on antiplatelet medication or an anticoagulant agent, a more effective therapy for stroke patients with cerebrovascular lesions needs to be explored. The high rate of recurrent events could not be accounted for by the differences in the use of antithrombotic treatment. On the contrary, significantly more patients with vascular lesions (314 of 345, 91%) were on antithrombotic treatment than those without vascular lesions (285 of 360, 79.2%; P<0.0001).
The present study of Chinese patients showed that the cumulative mortality rate after stroke is 5.7% at 6 months, 8.9% within the first year, 14.9% by 2 years, and 20.8% by 3 years. Our data showed that patients with intracranial or tandem atherosclerosis are at the greatest risk of stroke or death within the first 6 months of a stroke. Thus, early and aggressive therapy may be needed to prevent these adverse events in this high-risk group of patients.
Besides the number of occlusive arteries, advancing age and atrial fibrillation were powerful independent predictors of death and recurrent events among our patients. This observation was well established, but we found that even when these 2 important factors were taken into account, the number of occlusive vessels remained an independent predictor.
The strengths of our study are the large number of patients and the small number lost to follow-up. The weakness is the lack of angiographic study to confirm the diagnosis of occlusive disease. However, it would be impossible to conduct a large study with intravenous conventional angiography because of the cost and risk associated with this procedure. TCD may not be ideal, but it is accessible for most hospitals (especially important for rural hospitals in developing countries), has a low cost, is safe, and is well accepted by patients. TCD is an accepted method for diagnosing intracranial occlusive disease and has been validated for grading middle cerebral artery stenosis.10,11
Many studies of long-term outcomes of stroke patients reported recurrent events among the survivors of the index stroke. Nowadays, however, many secondary prevention therapies such as aspirin, warfarin, and statin are initiated soon after stroke before discharge from hospital, so we used a pragmatic approach to reporting all deaths and vascular events after the index stroke. Furthermore, some patients died of stroke complications from the index stroke after hospital discharge, and it might be difficult to have a cutoff time for categorizing survival and death.
Although this study was hospital based, our hospital was the one with emergency admission and captured >95% of all hospital admissions in our region. In addition, most patients with stroke required hospital admission for investigations such as CT scan in Hong Kong.
In this study, we used TCD as the only diagnostic tool for intracranial occlusive disease. Regardless of known and accepted TCD criteria for intracranial stenotic disease, TCD is largely operator dependent. Additionally, turbulent flow resulting from anatomic variations in vessel size and course, inability to make angle corrections for the ultrasound beam, and poor acoustic windows are major limitations of TCD. Ideally, angiography such as digital subtraction, MRA, or CTA should be done in those patients with a poor temporal window. Unfortunately, lack of resources in the present study prevented us from performing angiography in our patients. Future studies should consider the use of supplementary angiography to rectify this weakness. Moreover, flow velocities vary significantly with age and presence of vascular risk factors and during the course of the acute phase of stroke. However, even with these limitations, we were able to detect significant differences with the use of TCD only. Further studies using additional diagnostic tests such as angiography may require fewer patients to document the poor prognosis of patients with intracranial occlusive disease. In our laboratory, the TCD criteria used in this study have a sensitivity of 91.4% and a specificity of 82.7% for diagnosing >50% middle cerebral artery stenosis compared with MR angiography.11
The TCD assessments were performed within 7 days of symptom onset. In the acute stroke settings, TCD results may be significantly influenced by the time from stroke onset. The TCD results might have been misinterpreted in terms of hemodynamic changes (stenosis versus recanalization, compensatory flow, collateral circulation) or cause of stroke (embolic versus intracranial stenosis). A significant number of these intracranial atherosclerotic lesions may be of cardioembolic origin. In this study, only 38 patients had known cardioembolism such as atrial fibrillation (5%). Furthermore, we have found that 7% of asymptomatic community subjects >40 years of age have intracranial stenosis, indicating that intracranial disease is common in our population.12 More than 70% of patients with occlusive disease in the middle cerebral artery had persistent disease after 6 months.13 Persistent TCD abnormalities would be more likely to be due to intracranial atherosclerosis.
In summary, our study provided important epidemiological data on poststroke mortality and recurrence rate among our population with a preponderance of intracranial atherosclerosis. Patients with intracranial atherosclerosis are at higher risk than those without intracranial atherosclerosis. Moreover, patients with both intracranial and extracranial occlusive diseases have the worst outcome.
This study was supported by a grant from the Research Grants Council of Hong Kong (grant CUHK 4341/98M).
- Received December 5, 2002.
- Revision received April 22, 2003.
- Accepted May 20, 2003.
Gorelick PB, Sacco RL, Smith DB, Alberts M, Mustone-Alexander L, Rader D, Ross JL, Raps E, Ozer MN, Brass LM, et al. Prevention of a first stroke: a review of guidelines and a multidisciplinary consensus statement from the National Stroke Association. JAMA. 1999; 281: 1112–1120.
Wong KS, Li H, Chan YL, Ahuja A, Lam W, Wong A, Kay R. Use of transcranial Doppler ultrasound to predict prognosis in patients with intracranial large artery occlusive disease. Stroke. 2000; 31: 2641–2647.
Babikian VL, Wechsler LR. Transcranial Doppler Ultrasonography. St Louis, Mo: Mosby-Year Book, Inc; 1993.
Huang YN, Gao S, Li SW, Huang Y, Li JF, Wong KS, Kay R. Vascular lesions in Chinese patients with transient ischemic attacks. Neurology. 1997; 48: 524–525.
Wong KS, Huang YN, Gao S, Lam WW, Chan YL, Kay R. Intracranial stenosis in Chinese patients with acute stroke. Neurology. 1998; 50: 812–813.
Klijn CJM, Kappelle LJ, Algra A, Van Gijn J. Outcome in Patients with symptomatic occlusion of the internal carotid artery or intracranial arterial lesions: a meta-analysis of the role of baseline characteristics and types of antithrombotic treatment. Cerebrovasc Dis. 2001; 12: 228–234.
Wong KS, Huang YN, Yang HB, Li H, Gao S. Prevalence of intracranial large artery stenosis in China: a population-based study. Stroke. 2000; 31: 2793. Abstract.
Wong KS, Li H, Lam WMM, Chan YL, Kay R. Progression of middle cerebral artery occlusive disease and its relationship with further vascular events after stroke. Stroke. 2002; 33: 532–536.