Implication of Earlier Carotid Atherosclerosis for Stroke and Its Subtypes

Yoji Nagai, MD, PhD, Kazuo Kitagawa, MD, PhD, Masayasu Matsumoto, MD, PhD


Prev Cardiol. 2003;6(2) 

In This Article

Carotid Atherosclerosis as a Risk Factor for Specific Stroke Subtypes

Despite the link between carotid atherosclerosis and stroke, the latter is a heterogeneous disease that comprises several subtypes with different etiologies. Because atherosclerosis is a precursor for atherothrombotic infarction, carotid atherosclerosis is a risk factor for this stroke subtype. This concept is reinforced by an association between carotid and major cerebral artery atherosclerosis.[34] In contrast, lacunar infarction (LI) is most often the result of lipohyalinosis, fibrinoid necrosis, or microatheroma in intracerebral small arteries,[35] diluting the implication of carotid atherosclerosis. Moreover, atherosclerosis is not likely to play a direct role in cardioembolic infarction, intracerebral and subarachnoid hemorrhage. To refine the utility of carotid evaluation for stroke risk assessment, we have recently examined the associations between PS and stroke subtypes.[36]

Subjects. The subjects for this investigation comprised 1059 patients (age, 62±11 years) evaluated in our department. Many of them had histories of stroke and/or its risk factors such as hypertension, hyperlipidemia, and diabetes. Also, they included patients with nonspecific neurologic complaints such as dizziness, headache, and memory disturbances. The majority of patients had been referred from other hospitals or departments, for the assessment of cerebral circulation, for the secondary prevention of stroke, or for perioperative risk assessment before surgery. Because of the high prevalence of stroke and its risk factors, carotid ultrasonography was performed for the screening of carotid atherosclerosis and stenosis, or in some cases for the assessment of vertebral artery circulation.

To mitigate direct threats to the cerebral circulation, and to focus on atherosclerosis, patients with the following criteria were excluded: carotid stenosis ≥60% or occlusion by duplex ultrasound; post-carotid endarterectomy; and collagen diseases including Takayasu's arteritis and systemic lupus erythematosus. Patients with TIA were excluded because TIA is highly heterogeneous. Also, patients with stroke within the preceding 2 weeks were excluded, because of the hemodynamic and metabolic modulation often observed after the event.

Carotid Ultrasonography. Duplex carotid ultrasonography was performed as previously described,[24] with the use of linear array 7.5-MHz transducers. In accordance with our previous studies,[24,32,33] severity of carotid atherosclerosis was evaluated by the PS. Briefly, the common and internal carotid arteries were scanned cross-sectionally and longitudinally, whereby distribution of atheromatous plaques was estimated. During the initial scanning, optimal insonation angles were determined for the estimation of plaque heights, and the measurements were performed on the frozen frame, perpendicular to the vascular walls. Bilateral carotid arteries were examined following the same procedures. Thereafter, PS was computed by summing the maximum thickness of all plaques (local increases in IMT ≥1.1 mm) located in bilateral carotid arteries. Carotid stenosis/occlusion was diagnosed by the commonly used criteria.[37]

Diagnoses of Stroke and Its Subtypes. Neurologic signs and symptoms were evaluated by our specialty staffs when patients visited and/or were admitted to the hospital. Histories of neurologic episodes were carefully obtained from the patients and/or their families. On the basis of neurologic signs/symptoms and medical histories, stroke was diagnosed by an acute disturbance of focal neurologic function resulting in either signs or symptoms of presumed vascular origin that persisted for >24 hours. Patients diagnosed to have stroke subsequently underwent brain magnetic resonance imaging scans, which were used as a diagnostic aid for stroke subtypes.

On the basis of neurologic signs/symptoms, medical histories and brain scans, patients were classified into the following six groups: 1) nonstroke; 2) atherothrombotic infarction (AI); 3) LI; 4) cardioembolic infarction (CE); 5) cerebral hemorrhage (CH); and 6) other or unclassified stroke (OU).

Baseline Characteristics. Because the study sample comprised patients of our department, prevalence of stroke and its risk factors was generally high. Stroke was diagnosed in 321 out of 1059 subjects, including 56 AI (17% of total stroke), 117 LI (36%), 65 CE (20%), 26 CH (8%), and 57 OU (18%) patients. The percentages of major stroke subtypes were similar to those reported in Japan,[38] although the proportion of LI was higher than in Western countries.[39,40] Compared with non-stroke patients, cardiovascular risk factors were more prevalent in stroke groups, with the highest prevalence in group AI, supporting associations between stroke and these risk factors.

PS in Patients With Respective Stroke Subtypes. As a measure for carotid atherosclerosis, PS was higher in AI and LI patients than in nonstroke patients, whereas significant differences were not found between other stroke subtype and nonstroke patients (Figure, Table I ). These findings suggest increased severity of atherosclerosis in AI and LI patients. Although average PS also appeared to be higher in CH and OU patients, the differences did not reach statistical significance. However, the number of CH patients was relatively small, limiting our power to detect differences.

Plaque score by stroke subtypes N=nonstroke; AI=atherothrombotic infarction; LI=lacunar infarction; CE=cardioembolic infarction; CH=cerebral hemorrhage; OU=other or unclassified stroke; *p<0.05 vs. group N. (Differences are shown only between group N and each stroke group.) Error bars indicate standard deviation. Reproduced from Nagai et al.[36]

Since PS may be influenced by cardiovascular risk factors in each group, it was further compared by adjusting for age and sex, and subsequently all traditional cardiovascular risk factors. When adjusting for age and sex, PS was little modified, and continued to be higher in AI and LI patients than in nonstroke patients. When adjustment was made for additional risk factors, PS continued to be higher in AI patients than in nonstroke patients. Although such earlier carotid lesions per se would not explain the etiology of AI, the finding further supports involvement of the atherosclerotic process in the evolution of AI. By contrast, the difference between LI and nonstroke patients diminished after risk factor adjustments. Based on this result, the higher PS found for LI patients appears to be induced by cardiovascular risk factors prevalent in such patients.

Prior to the current study, Touboul et al.[41] compared CCA IMT between ischemic stroke and control patients, and demonstrated thicker IMT in all major stroke subtype patients. Their finding is consistent with the higher PS in AI and LI patients found in our study. However, they showed thicker IMT also in CE patients, and the difference persisted after adjustment for cardiovascular risk factors. They measured IMT at plaque-free regions and defined cardiovascular risk factors as binary variables, whereas we quantified plaques and included some risk factors as continuous variables. Also, they obtained the controls primarily from noncardiovascular patients, whereas our nonstroke patients predominantly comprised cardiovascular patients. Such methodologic differences could account for the differences between the two studies.

PS for Stratifying the Likelihood of Specific Stroke Subtypes. Given the higher PS observed for AI and LI patients, PS may aid in the risk assessment of such stroke subtypes. To examine the contribution of PS for stratifying these subtypes, we performed logistic regression analyses, with either AI or LI as an end point ( Table II ). In univariate analysis (model 1), each 1-SD greater PS was associated with 2.5-fold higher risk for AI and 1.4-fold higher risk for LI, compared with nonstroke patients. Adjustments of age and sex little modified these associations (model 2). These findings suggest the potential value of evaluating PS for risk assessment of AI and LI. However, when assessing the risk in clinic, other cardiovascular data may also be taken into account. After further adjustments of cardiovascular risk factors, PS remained a significant associate for AI, which was not the case for LI (model 3). Thus, even when other clinical data are available, carotid evaluation appears to convey additional benefit for risk assessment of AI. Conversely, the benefit for stratifying LI may be limited under such conditions.

To further compare PS to stratify the likelihood for AI and LI, we performed receiver operating characteristic (ROC) curve analyses. The ROC curve is a graph that displays the relationships between sensitivity and specificity of a diagnostic tool across a spectrum of cut points that could be used to classify patients as diseased or nondiseased. Overall accuracy of the diagnosis is expressed in terms of the area under the ROC curve, ranging from 0.5 (no discrimination) to 1.0 (perfect discrimination). Both when using PS alone (model 1) and in combination with other risk factors (models 2 and 3), the area under the ROC curves was greater than 0.5 for AI and LI. Of note, the ROC area found for AI was greater than that for LI in all models, suggesting a greater value of carotid evaluation for risk assessment of AI.

Study Limitations. There are certain limitations of the current study. Given the difficulty of performing cerebral angiography on all stroke patients, the subtypes were diagnosed in a noninvasive manner, leading to possible misdiagnoses. For example, deep small infarction, which was classified in group LI, could be the result of major cerebral artery occlusion or occult cardiogenic emboli. However, we included all equivocal diagnoses in group OU, to reduce the chance for misclassification of major stroke subtypes. Also, due to the high prevalence of cardiovascular risk factors in non-stroke patients, the results of this study cannot be directly transferred to the general population. However, carotid ultrasonography is most often performed for cardiovascular patients in hospitals, supporting the clinical value of our findings.