- Email: [email protected]
Carotid Hemodynamic Parameters Are Useful for Discriminating Between Atherothrombotic Infarction and Lacunar Infarction
Carotid Hemodynamic Parameters Are Useful for Discriminating Between Atherothrombotic Infarction and Lacunar Infarction Yutaka Nishiyama, MD, PhD, Toshiya Katsumata, MD, PhD, Tatsuo Otori, MD, PhD, and Yasuo Katayama, MD, PhD
Using ultrasound, we investigated whether carotid parameters differed among subtypes of ischemic stroke and evaluated the usefulness of these parameters in discriminating among subtypes. Patients with ischemic stroke admitted to Nippon Medical School Hospital were consecutively recruited and grouped into 3 subtypes based on the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification: cardioembolism (group CE), large-artery atherosclerosis (group LAA), and small-vessel occlusion (group SVO). All subjects underwent carotid ultrasonography to determine maximum intima-media thickness (IMT), maximum systolic velocity (Vmax), minimum diastolic velocity (Vmin), mean velocity, and pulsatility index (PI). Carotid parameters that differed among subtypes were statistically identified. A total of 138 patients were enrolled. Intergroup comparisons revealed that the Vmin of the affected side was significantly lower in group LAA than in group SVO (mean 6 SD, 0.12 6 0.05 m/s vs 0.15 6 0.05 m/s; P 5 .02) and the Vmin of the mean of both sides was lower in group LAA than in group SVO (0.12 6 0.04 vs 0.16 6 0.05; P 5 .03). Multivariate analysis showed that the PI of the affected side was a useful adjunct to discriminate between groups SVO and LAA (odds ratio 5 2.94; P 5 .03, group SVO as control). Receiver operating characteristic curve analysis found that the Vmin of the affected side was the most useful parameter for discriminating between group SVO and group LAA. The PI and the Vmin of the affected side were found to differ among stroke subtypes, and thus these may be useful parameters for discriminating among ischemic stroke subtypes. Key Words: Carotid parameter—stroke subtypes—pulsatility index. Ó 2010 by National Stroke Association
Carotid ultrasonography is a noninvasive technique for examining arteriosclerotic changes in the carotid artery that enables a quantitative assessment of intima-media thickness (IMT). The results of ultrasonographic examination have been reported to accurately reflect those of From the Department of Internal Medicine, Division of Neurology, Nephrology, and Rheumatology, Nippon Medical School, Tokyo, Japan. Received May 29, 2009; revision received July 24, 2009; accepted August 15, 2009. Address correspondence to Yutaka Nishiyama, MD, PhD, Department of Internal Medicine, Division of Neurology, Nephrology, and Rheumatology, Nippon Medical School, 1-1-5 Sendagi, Bunkyo-ku Tokyo 113-8603, Japan. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2010 by National Stroke Association doi:10.1016/j.jstrokecerebrovasdis.2009.08.005
pathological examination.1,2 In the Rotterdam Study, Bots et al3 showed that an increase in the IMT of the common carotid artery (CCA) was a risk factor for ischemic stroke. In the ARIC Study, Chambless et al4 found that the mean IMT of the internal carotid artery and the CCA were associated with the incidence of coronary heart disease. These findings indicate that the IMT of the CCA might be a useful indicator of atherosclerosis and could be related to cardiovascular risk.5 The National Institute of Neurological Disorders and Stroke (NINDS) has classified ischemic stroke into the following subtypes: cardioembolic infarction, lacunar infarction, atherothrombotic infarction, and others.6 An effective treatment strategy hinges on an accurate diagnosis of the ischemic stroke subtype based on this classification system.7-9 Comparing plaque scores calculated on the
Journal of Stroke and Cerebrovascular Diseases, Vol. 19, No. 6 (November-December), 2010: pp 441-449
441
Y. NISHIYAMA ET AL.
442 9
basis of the total IMT of both carotid arteries, Nagai et al found that patients with atherothrombotic infarction had higher scores than patients with other ischemic stroke subtypes. Comparing patients with ischemic stroke grouped into lacunar and nonlacunar infarction subtypes, Cupini et al10 found that independent prediction of subtype was possible based on the IMT of the CCA and the presence of atrial fibrillation. Whether measurement of IMT can be used successfully as an adjunct to distinguish the subtype of ischemic stroke remains unclear, however. Recently, Trial of Org 10172 in Acute Stroke Treatment (TOAST) investigators classified ischemic stroke into the following subtypes: cardioembolism, large-artery atherosclerosis, small-vessel occlusion, stroke of other determined etiology, and stroke of undetermined etiology.11 The TOAST classification system is easy to apply to all patients and exhibits good interobserver agreement. The system has been widely used in epidemiologic studies; however, to the best of our knowledge, there have been no reports on echographic parameters of the carotid artery useful for discriminating among the subtypes in the TOAST classification system. Carotid ultrasonography can be used to measure Doppler hemodynamic parameters, including maximum systolic velocity (Vmax), minimum diastolic velocity (Vmin), mean velocity, and pulsatility index (PI). Gosling et al12 proposed that the PI reflects peripheral vascular resistance, and the correlation between PI and peripheral vascular resistance was subsequently confirmed in a study of the brachiocephalic trunks of healthy volunteers.13 To date, no study has used carotid hemodynamic parameters to discriminate among ischemic stroke subtypes, however. The purpose of the present study was to investigate whether carotid hemodynamic parameters differed among ischemic stroke subtypes and whether assessment of the parameters could serve as a useful adjunct in discriminating among the subtypes.
Methods Patients Patients with ischemic stroke admitted to the Nippon Medical School Hospital from January 2000 to September 2002 were consecutively enrolled in this study. Neurologists in our department evaluated the patients’ neurologic signs and symptoms at admission. An ischemic stroke was defined as an acute disturbance of focal neurologic function persisting for 24 hours. Past medical history and risk factors were obtained from the patients or their families. Each patient underwent a computed tomography scan on admission to exclude cerebral or subarachnoid hemorrhage. Informed consent was obtained from all participants. Our institution’s Committee for the Protection of Human Subjects in Research approved the study design.
Evaluation Brain magnetic resonance imaging (MRI) (T1-, T2-, and diffusion-weighted images) and magnetic resonance angiography (MRA) were performed to detect ischemic lesions in the perfused area of the internal carotid artery and to evaluate the intracranial cerebral arteries. Ischemic stroke of an internal carotid artery area was diagnosed when lesions corresponding to a neurologic finding were detected as an anterior circulation area with low signal intensity on T1-weighted images and high signal intensity on T2- and diffusion-weighted images. Carotid ultrasonography was conducted within 2 weeks after onset of stroke. The investigation was performed using a 7.5-MHz linear-array transducer (LOGIQ 500GE, Yokogawa Medical Systems, Tokyo, Japan). The examiner (Y. N.) was blinded to the clinical information regarding ischemic stroke subtype, infarct site, and risk factors. All examinations were performed with the patient laying supine in a dark room with his or her head held in the midline position or slightly tilted to either side. The examination began by scanning the common and internal carotid arteries cross-sectionally and longitudinally, to roughly evaluate the distribution of atheromatous lesions. During this initial scanning, the optimal insonation angles were determined for estimation of the respective IMT sizes. Scanning regions were set within 45 mm proximal and 15 mm distal to the flow divider, according to the method of Handa et al14 (Fig 1). The maximum IMT (max-IMT) was defined as the maximum height of the IMT of the far wall at the point of measurement. Next, the CCA was examined by pulsed Doppler ultrasonography to obtain information on carotid blood flow. On the longitudinal scans, the sample volume was set in the CCA and displayed as linearly as possible. The insonation site was set 15–45 mm proximal to the flow divider, and a range-gate pulsed Doppler sample volume of 5-7 mm was used to measure the blood flow velocity in the CCA. Special care was taken to maintain the incident angle between the CCA and the beam at 30-60 degrees. The Vmax, Vmin, and PI of the CCA were obtained by analyzing the Doppler flow waveform envelope. The PI was calculated as (systolic velocity - diastolic velocity)/mean velocity. Each flow velocity parameter was determined as the median value of 3 consecutive cardiac cycles. No hemodynamic parameter was estimated at any point at which Vmax exceeded 125 cm/second (suggesting an artery stenosis .50%15). Parameters on the affected and contralateral sides were evaluated separately. For statistical analysis, the value obtained on the affected side, the value obtained on the contralateral side, and the mean of these 2 values were evaluated. When the peak systolic velocity measured by ultrasonography exceeded 200 cm/second in the narrow segment (indicating artery stenosis .70%16), or when MRA detected .60% stenosis during endarterectomy using
CAROTID PARAMETERS AND STROKE SUBTYPES
Figure 1. Schematic showing sites of IMTevaluation in the carotid system.
the asymptomatic carotid stenosis method,17 the condition was defined as stenosis of the extracranial or intracranial cerebral arteries. Percutaneous and/or transesophageal echocardiography was performed in all patients to investigate cardioembolic sources.
Classification and Risk Factors Based on neurologic signs and symptoms, medical history, and brain MRI scan findings, the patients were classified into the following 5 groups according to the TOAST classification criteria: (1) group CE (cardioembolism), (2) group LAA (large-artery atherosclerosis), (3) group SVO (small-vessel occlusion), (4) group OE (stroke of other determined etiology), and (5) group UE (stroke of undetermined etiology). When comparing the differences in echographic parameters among the stroke subtypes, patients classified into group OE or group UE were excluded from the study, because the etiology of cerebral infarction in these patients was highly heterogeneous. The following risk factors were evaluated: age, sex, hypertension, diabetes mellitus, hyperlipidemia, atrial fibrillation, smoking, and history of ischemic stroke. These risk factors and patient histories were determined by interviews and physical and laboratory examinations during hospitalization. To evaluate the effects of drugs on echographic parameters of the carotid artery, the presence or absence of argatroban and ozagrel sodium use was noted, and the effect of these drugs on those parameters was studied.
Statistical Analysis All statistical analyses were performed with StatView for Windows version 5.0 (SAS Institute, Cary, NC) or StatMate III for Windows (Atoms, Tokyo, Japan). The c2 test was used to detect intergroup differences of risk factors, and one-way analysis of variance (ANOVA) was used to examine intergroup age differences. The Student t-test
443
was used to compare echographic parameters with and without argatroban and ozagrel sodium use. One-way ANOVA followed by the Tukey-Kramer post hoc test was used to detect intergroup differences in measured carotid parameters. Logistic regression analysis was performed to evaluate risk factors and parameters related to the different ischemic stroke subtypes. Initially, simple logistic regression was used to test the relationships between unadjusted variables and the difference of 2 stroke subtypes (model 1). Then multivariate associations were tested using the backyard multiple logistic regression method (model 2). Finally, a regression model was applied in which variables showing significant relationships in the simple logistic regression were included mandatorily, and other variables were selected by the backyard method (model 3). The outcome variable (dependent variable) was the difference among stroke subtypes (one subtype served as a control). The independent variables were clinically plausible risk factors (age, sex, hypertension, diabetes, hyperlipidemia, smoking, atrial fibrillation, and history of ischemic stroke) and hemodynamic parameters of both the affected and contralateral sides. The original model included all of the foregoing independent parameters. Backyard stepwise regression was performed, and a P value of ..05 was used for variable removal. To evaluate whether ischemic stroke subtypes could be differentiated using these parameters, receiver operating characteristic (ROC) curve analyses were conducted using StatMate III. StatMate III calculates an ROC curve for each parameter of a data set. The area under the ROC curve (AUC) was calculated, and the AUCs of carotid hemodynamic parameters were compared statistically using the method of Hanley et al.18 The best threshold of carotid hemodynamic parameters for stroke subtype discrimination was determined using the Youden index.17 Data are expressed as mean 6 standard deviation (SD) or percentage. A P value ,.05 was considered statistically significant.
Results Out of the patients admitted, a total of 224 patients with ischemic stroke were finally enrolled. Three of the 224 patients (1.3%) were classified into the OD group; 2 of these patients had coagulation and fibrinolytic abnormalities with malignant tumors, and 1 patient had an antiphospholipid antibody syndrome. Eighty-one patients (36.2%) were classified into the UE group; the etiology was unknown in 17 of these patients, and 2 or more etiologies were found in 64 patients. Table 1 presents profiles of the patients’ risk factors. Significant differences in the prevalences of hypertension, hyperlipidemia, atrial fibrillation, and smoking were found among the 5 groups. In comparison with the other groups, hypertension was higher in group OE,
Y. NISHIYAMA ET AL.
444
Table 1. Characteristics of the 5 clinical subtype groups
N Sex, M/F Age, years, mean 6 SD Hypertension, n, (%) Diabetes, n (%) Hyperlipidemia, n (%) Atrial fibrillation, n (%) Smoking, n (%) Old cerebral infarct, n (%)
Group CE
Group SVO
Group LAA
Group OE
Group UE
P value
54 35/19 71 6 11 26 (48) 13 (24) 25 (46) 37 (68) 9 (17) 7 (13)
64 43/21 66 6 11 55 (85) 21 (33) 46 (72) 0 (0) 32 (50) 15 (23)
22 17/5 73 6 11 15 (68) 7 (32) 10 (45) 0 (0) 12 (55) 5 (23)
3 1/2 73 6 10 3 (100) 1 (33) 1 (33) 0 (0) 1 (33) 0 (0)
81 51/30 68 6 11 52 (64) 24 (30) 45 (56) 17 (21) 33 (41) 18 (22)
.531 .069 .003 .881 .037 ,.001 .002 .519
Table 4 compares the carotid ultrasonography parameters of these 3 groups. Comparing the parameters of the affected side by one-way ANOVA revealed significant differences in Vmin among the 3 groups, as did comparing the parameters of the mean of both sides. A similar comparison of the parameters of the contralateral side revealed no statistically significant differences in maxIMT, Vmax, Vmin, or PI among the 3 groups. Post hoc analysis using the Tukey-Kramer test showed that the Vmin of the affected side was significantly lower in group LAA than in group SVO (P ,.05) and that the Vmin of the mean of both sides was lower in group LAA than in group SVO (P , .05); however, max-IMT, Vmax, and PI of the affected side and the mean of both sides were not significantly different. A similar comparison of the parameters of the contralateral side revealed no statistically significant differences in the parameters among the 3 groups. Because in our statistical analysis the carotid parameters differed significantly only between groups SVO and LAA, logistic regression analysis and ROC curve analysis
hyperlipidemia was higher in group SVO, atrial fibrillation was higher in group CE, and smoking was higher in group LAA. There were no significant differences in gender, age, diabetes, or history of ischemic stroke among the 5 groups. Table 2 shows the differences in echographic parameters of the carotid artery on the basis of the presence or absence of argatroban and ozagrel sodium use in all patients. We compared the values on the affected side, on the contralateral side, and the mean of both sides under the presence and absence of these drugs and found no statistically significant differences in any of the parameters (P . .05). Cerebral artery stenosis was observed in 60 of the 224 patients (26.8%). Twenty-two of these patients belonged to group LAA, while the remaining 38 patients belonged to group UE (Table 3). The 84 patients of groups OE and UE were excluded from subsequent analyses. The remaining 140 patients enrolled in the study included 54 patients in group CE, 64 patients in group SVO, and 22 patients in group LAA.
Table 2. Comparisons of the carotid parameters on the basis of the presence or absence of argatroban and ozagrel sodium use Argatroban (1)
Argatroban (2)
P value
Ozagrel sodium (1)
Ozagrel sodium (2)
P value
1.71 6 0.93 0.72 6 0.21 0.16 6 0.06 1.97 6 0.45
2.01 6 1.24 0.69 6 0.21 0.16 6 0.07 1.88 6 0.62
.12 .32 .66 .63
1.96 6 1.29 0.71 6 0.04 0.17 6 0.07 1.95 6 1.09
1.92 6 1.09 0.68 6 0.05 0.15 6 0.07 2.04 6 0.83
.81 .40 .11 .50
1.59 6 0.79 0.74 6 0.23 0.16 6 0.06 1.92 6 0.43
1.83 6 0.93 0.70 6 0.21 0.17 6 0.07 1.88 6 0.62
.11 .24 .50 .64
1.81 6 0.91 0.73 6 0.21 0.17 6 0.08 1.92 6 0.66
1.75 6 0.89 0.70 6 0.22 0.16 6 0.06 1.86 6 0.52
.59 .35 .12 .43
1.65 6 0.77 0.73 6 0.21 0.16 6 0.05 1.95 6 0.40
1.88 6 0.89 0.69 6 0.20 0.16 6 0.06 1.94 6 0.68
.10 .24 .89 .99
1.84 6 0.90 0.72 6 0.19 0.17 6 0.07 1.94 6 0.69
1.82 6 0.85 0.69 6 0.21 0.16 6 0.06 1.95 6 0.58
.82 .34 .07 .89
Affected side Max-IMT Vmax Vmin PI Contralateral side Max-IMT Vmax Vmin PI Mean of both sides Max-IMT Vmax Vmin PI All data are mean 6 SD.
CAROTID PARAMETERS AND STROKE SUBTYPES
445
Table 3. Distribution of cerebral artery stenosis in the patients examined All patients (n 5 224) Single artery stenosis Affected side ICA occlusion Affected side ICA stenosis .50% Affected side MCA stenosis .50% Affected side ACA stenosis .50% Multiple artery stenosis Bilateral ICA stenosis .50% Bilateral MCA stenosis .50% Affected side ICA occlusion and contralateral side MCA stenosis .50% Affected side MCA stenosis .50% and contralateral side ICA stenosis .50% Affected side MCA stenosis .50% and contralateral side ACA stenosis .50% Group LAA (n 5 22) Single artery stenosis Affected side ICA occlusion Affected side ICA stenosis .50% Affected side MCA stenosis .50% Multiple artery stenosis Bilateral ICA stenosis .50% Affected side MCA stenosis .50% and contralateral side ACA stenosis .50%
between these 2 groups were performed. A total of 86 patients in these 2 groups were examined. Table 5 gives the results of the logistic regression analysis. Atrial fibrillation was eliminated as an independent variable, because neither group included any patient with atrial fibrillation. We used the clinical data of group SVO as a control, and compared the findings of this group with those of group LAA. The results of simple regression analysis indicated positive associations with age (P 5 .004) and PI of the affected side (P 5.03) and negative associations with hyperlipidemia (P 5 .03), Vmin of the affected side (P 5 .006), and Vmin of the contralateral side (P 5 0.02) in group LAA (model 1). The results of the multivariate logistic
49 (21.9%) 5 (2.2%) 16 (7.1%) 26 (11.6%) 2 (0.9%) 11 (4.9%) 5 (2.2%) 2 (0.9%) 1 (0.4%) 2 (0.9%) 1 (0.4%) 18 (81.8%) 1 (4.5%) 6 (27.2%) 11 (50.0%) 4 (18.2%) 3 (13.6%) 1 (4.5%)
regression analysis revealed a significant positive association with age and PI on the affected side in group LAA (model 2). The model in which some variables were mandatorily included and other variables were selected by the backyard method revealed a significant negative association with hypertension in group LAA (model 3). The degree applicable to actually measured levels was 77.9% in model 2 and 76.7% in model 3. ROC curve analyses were conducted to evaluate the discrimination ability in groups SVO and LAA. Figure 2 shows ROC curves for the values of the affected side, the contralateral side, and the mean of both sides, with the maximum AUC associated with the Vmin of the
Table 4. Comparisons of the carotid parameters among 3 clinical subtype groups
Affected side Max-IMT Vmax Vmin PI Contralateral side Max-IMT Vmax Vmin PI Mean of both sides Max-IMT Vmax Vmin PI All data are mean 6 SD.
Group CE
Group SVO
Group LAA
P value
1.75 6 0.82 0.68 6 0.23 0.16 6 0.08 1.88 6 0.72
1.74 6 0.79 0.69 6 0.18 0.15 6 0.05 1.88 6 0.48
2.10 6 1.33 0.62 6 0.19 0.12 6 0.05 2.18 6 0.62
.19 .37 .02 .12
1.57 6 0.72 0.67 6 0.24 0.16 6 0.07 1.83 6 0.58
1.74 6 0.82 0.71 6 0.20 0.17 6 0.06 1.89 6 0.52
1.88 6 0.90 0.66 6 0.19 0.13 6 0.05 2.10 6 0.81
.26 .54 .09 .18
1.67 6 0.66 0.68 6 0.22 0.16 6 0.07 1.86 6 0.57
1.72 6 0.70 0.70 6 0.17 0.16 6 0.05 1.88 6 0.42
1.96 6 0.89 0.64 6 0.17 0.12 6 0.04 2.14 6 0.60
.27 .48 .03 .08
Y. NISHIYAMA ET AL.
446
Table 5. Univariate and mulitivariate analysis to test the relationships between cerebrovascular risk factors and hemodynamic parameters and the difference between 2 clinical subgroups Model 1 OR (95% CI) Gender 1.66 (0.54-5.12) Age 1.06 (1.01-1.11) Hypertension 0.35 (0.11-1.11) Diabetes 0.96 (0.34-2.70) Hyperlipidemia 0.33 (0.12-0.89) Smoking 1.20 (0.45-3.17) Old cerebral infarct 0.96 (0.30-3.04) Affected side max-IMT 1.55 (0.91-2.64) Affected side Vmax 0.11 (0.01-1.97) Affected side Vmin 2.05e-7 (3.23e-8-0.11) Affected side PI 3.00 (1.12-8.03) Contralateral side max- IMT 1.22 (0.69-2.15) Contralateral side Vmax 0.28 (0.02-3.87) Contralateral side Vmin 2.55e-6 (4.66e-11-0.14) Contralateral side PI 1.71 (0.79-3.72)
Model 2 P value .38 .004 .07 .93 .03 .71 .95 .1 .13 .006 .03 .5 .34 .02 .17
affected side. A comparison of the Vmin of the mean of both sides and the second-highest AUC level revealed no significant difference (P 5.719); however, a comparison with the PI of the affected side showed a statistically significant difference (P 5 .018). The best threshold of Vmin of the affected side to discriminate between groups SVO and LAA, identified using the Youden index, was 0.14 m/s. At this threshold, the sensitivity was 73% and the specificity was 59%.
Discussion In our study group, certain parameters of carotid ultrasonography differed among distinct ischemic stroke subtypes. One-way ANOVA revealed significant differences in the Vmin of the affected side and the Vmin of the mean of both sides between groups SVO and LAA. These results indicate that carotid parameters might be useful for discriminating between these 2 groups. Examination of the echographic parameters of the carotid artery based on the presence or absence of argatroban and ozagrel sodium use revealed no differences in any of the parameters. Although an increase in cerebral blood flow after treatment with argatroban or ozagrel sodium has been reported previously,19,20 our results suggest that this effect was not strong enough to change the flow velocity in the carotid artery. The most meaningful value for discrimination was the PI of the affected side in the multivariate logistic regression analysis after adjustment for risk factors. The odds ratio shown in Table 5 could be interpreted to mean that a 1-point elevation of PI on the affected side translates to a 2.9-fold increase in the likelihood of suffering a cerebral infarction due to large artery atherosclerosis.
Model 3
OR (95% CI)
P value
1.06 (1.01-1.11)
.02
2.94 (1.10-7.91)
.03
OR (95% CI)
P value
1.03 (0.97-1.09) 0.15 (0.03-0.71)
.4 .02
0.46 (0.14-1.47)
.19
1.82e-6 (1.03e-13-32.25) 1.36 (0.37-4.91)
0.005 (5.27e-9-4068.11)
.12 .64
.44
Although a model in which some variables were mandatorily included was studied, this model’s goodness of fit could not be considered better than that of the model in which variables were selected by the backyard method. The ROC curve analysis revealed that the Vmin of the affected side was the most useful parameter for discriminating between groups SVO and LAA. At the threshold value of 0.14 m/s, the Vmin of the affected side indicated a moderate sensitivity and specificity for discrimination. Several authors have reported that carotid hemodynamic parameters are influenced by sampling position.15,16 But we believe that our careful choice of the sampling point minimized the positional effects of focal stenosis and dilatation of the carotid artery. The PI is a confirmed marker of peripheral vascular resistance12,13 and also may reflect arteriosclerosis in the cerebral arteries distal from the carotid artery. Most of the patients with cerebral artery stenosis in our study were in group LAA, the group with the highest PI values (Tables 4 and 5). This suggests that the PI of the CCA reflects stenosis of a peripheral cerebral artery. Nakatou et al21 reported that the PI is higher in patients with diabetes than in controls and higher in patients with a history of ischemic stroke than in those with no history of ischemic stroke. That study did not discriminate among ischemic stroke subtypes in the analyses, however. If the PI reflects the resistance of blood vessels associated with cerebral infarction, then our result—that PI is positively associated with group LAA—can be considered reasonable. Other studies have reported varying prognoses for atherothrombotic infarction and lacunar infarction, and the diagnosis of ischemic stroke subtype has been found to be essential to establishing a treatment strategy.22,23 Consequently, we undertook this study to determine
CAROTID PARAMETERS AND STROKE SUBTYPES
A
447
1 0.9 0.8
TP RATIO
0.7 0.6 a-Vmin
0.5 0.4
a-PI
0.3
a-Vmin a-PI a-Vmax a-IMT
0.2
AUC
a-Vmax
0.73 0.64 0.59 0.57
a-IMT
0.1 0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
False Positive Ratio
B
1 0.9 0.8
TP RATIO
0.7 0.6 0.5
c-Vmin
0.4
c-PI
0.3
c-Vmin c-PI c-Vmax c-IMT
0.2
AUC 0.69 0.54 0.55 0.55
c-Vmax c-IMT
0.1 0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
False Positive Ratio
C
1 0.9 0.8
TP RATIO
0.7 0.6
m-Vmin
0.5
m-PI
0.4 0.3
m-Vmin m-PI m-Vmax m-IMT
0.2
m-Vmax
AUC 0.71 0.65 0.57 0.55
m-IMT
0.1 0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
False Positive Ratio Figure 2.
ROC curves of the parameters of the affected side (A), the contralateral side (B), and the mean of both sides (C).
whether carotid hemodynamic parameters can serve as useful adjuncts for discriminating between atherothrombotic infarction and lacunar infarction. Several previous studies have identified increased IMT as a risk factor for ischemic stroke,3-5 and higher plaque scores have been found in patients with atherothrombotic infarction compared with patients with other ischemic subtypes.9 Our findings indicate that the max-IMT of the affected side was higher in group LAA than in group
SVO, although our multivariate analysis showed no significant relationship between the group differences. The discrepancy between our results and earlier results may be due to differences in measurement methods. While our study relied on separate determinations of the IMT of the affected and contralateral sides, previous studies used the plaque score9 and mean IMT of the CCAs together.10,24 Previous reports also have noted discrepancies in the severity of carotid arteriosclerosis between
Y. NISHIYAMA ET AL.
448
extracerebral and intracerebral arteries, as well as differences associated with race and gender.25-27 Further studies involving numerous institutions are needed to determine whether IMT is a useful adjunct for discriminating ischemic stroke subtypes. A reduction in Vmin, a parameter of carotid blood flow, indicates stenosis or occlusion of blood vessels peripheral to the measurement site.28,29 We consider our ROC curve analysis results, which indicate that the Vmin of the affected side has superior discrimination ability in group LAA, not to be contradictory if vascular stenosis associated with cerebral infarction is present. The results of our logistic regression and ROC curve analyses indicate that the echographic parameters in group LAA differed with regard to the PI and Vmin of the affected side. This difference is assumed to be due to an insufficient number of subjects and the presence of a correlation between PI and Vmin. Whether PI or Vmin is more useful for the discrimination of stroke subtypes cannot be inferred from our results; larger-scale prospective studies with more subjects are needed. The carotid parameters differed significantly between groups SVO and LAA, while those in group CE did not significantly differ from those in the other groups. It may be that in group CE, infarction was caused by heart disease and had little to do with the ultrasonographic parameters that reflect cervical arteriosclerosis; many patients in group CE also developed concurrent atrial fibrillation. We believe that atrial fibrillation must have changed the carotid blood flow waveform on the ultrasonogram, and also may have affected the parameter measurements. Our results show a positive association with age in group LAA. This suggests that age might be a risk factor relative to atherothrombotic infarction. The relationship between age and stroke has been controversial in previous studies.30-32 Although some studies have found an association between age and lacunar infarction,30,31 Reed et al32 reported a relationship between age and high prevalence of atherosclerosis of the large cerebral arteries. Further studies to investigate this issue are needed. Our study has several limitations. Because the patients classified into groups OE and UE by the TOAST classification criteria were excluded from the investigation of echographic parameters (37.5% of all subjects), the relationship between echographic parameters and stroke subtypes OE and UE cannot be evaluated based on our results. Patients with these subtypes should be investigated within their subtype and placed in more detailed categories. A second limitation relates to the applicability of our data. As a retrospective, single hospital–based registration study, the present study may not have produced results directly applicable to the general population. A multicenter study to confirm our findings is needed. In conclusion, the difference in the PI or the Vmin of the affected side between groups SVO and LAA was
considered to reflect the severity of arteriosclerosis. Thus, we have shown that carotid hemodynamic parameters differ among ischemic stroke subtypes and that these parameters may be useful adjuncts for the discrimination of subtypes.
References 1. Pignoli P, Tremoli E, Poli A, et al. Intimal plus medial thickness of the arterial wall: a direct measurement with ultrasound imaging. Circulation 1986;74:1399-1406. 2. Wong M, Edelstein J, Wollman J, et al. Ultrasonic pathological comparison of human arterial wall: verification of intima-media thickness. Arterioscler Thromb 1993; 13:482-486. 3. Bots ML, Hoes AW, Koudstaal PJ, et al. Common carotid intima-media thickness and risk of stroke and myocardial infarction. Circulation 1997;96:1432-1437. 4. Chambless LE, Heiss G, Folsom AR, et al. Association of coronary heart disease incidence with carotid arterial wall thickness and major risk factors: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Epidemiol 1997;146:483-494. 5. Grobbee DE, Bots ML. Carotid artery intima-media thickness as an indicator of generalized atherosclerosis. J Intern Med 1994;236:567-573. 6. Whisnant JP, Basford JR, Bernstein EF, et al. Special report from the National Institute of Neurological Disorders and Stroke. Classification of cerebrovascular diseases. Stroke 1990;21:637-676. 7. White H, Boden-Albala B, Wang C, et al. Ischemic stroke subtype incidence among whites, blacks, and Hispanics. Circulation 2005;111:1327-1331. 8. Arboix A, Morcillo C, Garcia-Eroles L, et al. Different vascular risk factor profiles in ischemic stroke subtypes: a study from the Sagrat Cor Hospital of Barcelona Stroke Registry. Acta Neurol Scand 2000;102:264-270. 9. Nagai Y, Kitagawa K, Sakaguchi M, et al. Significance of earlier carotid atherosclerosis for stroke subtypes. Stroke 2001;32:1780-1785. 10. Cupini LM, Pasqualetti P, Diomedi M, et al. Carotid artery intima-media thickness and lacunar versus nonlacunar infarcts. Stroke 2002;33:689-694. 11. Adams HP, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke: definition for use in a multicenter clinical trial. Stroke 1993;24:35-41. 12. Gosling RG, King DH. Arterial assessment by Doppler shift ultrasound. Proc R Soc Med 1974;67:447-449. 13. Legarth J, Nolsoe C. Doppler blood velocity waveforms and the relation to peripheral resistance in the brachial artery. J Ultrasound Med 1990;9:449-453. 14. Handa N, Matsumoto M, Maeda H, et al. Ultrasonic evaluation of early carotid atherosclerosis. Stroke 1990; 21:1567-1572. 15. Grant EG, Benson CB, Moneta GL, et al. Carotid artery stenosis: gray-scale and Doppler US diagnosis. Society of Radiologists in Ultrasound Consensus Conference. Radiology 2003;229:340-346. 16. Koga M, Kimura K, Minematsu K, et al. Diagnosis of carotid artery stenosis greater than 70% with power Doppler duplex sonography. AJNR Am J Neuroradiol 2001;22:413-417. 17. Executive Committee for the Asymptomatic Carotid Atherosclerosis Study. Endarterectomy for asymptomatic carotid artery stenosis. JAMA 1995;273:1421-1428.
CAROTID PARAMETERS AND STROKE SUBTYPES 18. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148:839-843. 19. Kawai H, Yuki S, Sugimoto J, et al. Effects of a thrombin inhibitor, argatroban, on ischemic brain damage in the rat distal middle cerebral artery occlusion model. J Pharmacol Exp Ther 1996;278:780-785. 20. Oishi M, Mochizuki Y, Hara H, et al. Effects of sodium ozagrel on hemostatic markers and cerebral blood flow in lacunar infarction. Clin Neuropharmacol 1996;19:526-531. 21. Nakatou T, Nakata K, Nakamura A, et al. Carotid haemodynamic parameters as risk factors for cerebral infarction in type 2 diabetic patients. Diabetic Med 2004;21:223-229. 22. de Jong G, van Raak L, Kessels F, et al. Stroke subtype and mortality: a follow-up study in 998 patients with a first cerebral infarct. J Clin Epidemiol 2003;56:262-268. 23. Grau AJ, Weimar C, Buggle F, et al. Risk factors, outcome, and treatment in subtypes of ischemic stroke. The German Stroke Data Bank. Stroke 2001;32:2559-2566. 24. Touboul PJ, Elbaz A, Koller C, et al. Common carotid artery intima-media thickness and brain infarction. The GENIC case-control study. Circulation 2000;102:313-318. 25. Nijasri CS, Auruma C. Risk factors for atherosclerosis of cervicocerebral arteries: intracranial versus extracranial. Neuroepidemiology 2003;22:37-40.
449 26. Soo JL, Soo JC, Heui-Soo M, et al. Combined extracranial and intracranial atherosclerosis in Korean patients. Arch Neurol 2003;60:1561-1564. 27. Wityk RJ, Lehman D, Klag M, et al. Race and sex differences in the distribution of cerebral atherosclerosis. Stroke 1996;27:1974-1980. 28. Kamouchi M, Kishikawa K, Okada Y, et al. Reappraisal of flow velocity ratio in common carotid artery to predict hemodynamic change in carotid stenosis. AJNR Am J Neuroradiol 2005;26:957-962. 29. Saito K, Kimura K, Nagatsuka K, et al. Vertebral artery occlusion in duplex color-coded ultrasonography. Stroke 2004;35:1068-1072. 30. Rojas JI, Zurru MC, Romano M, et al. Acute ischemic stroke and transient ischemic attack in the very old: risk factor profile and stroke subtype between patients older than 80 years and patients aged less than 80 years. Eur J Neurol 2007;14:895-899. 31. Telman G, Kouperberg E, Sprecher E, et al. Distribution of etiologies in patients above and below age 45 with first-ever ischemic stroke. Acta Neurol Scand 2008;17: 311-316. 32. Reed DM, Resch JA, Hayashi T, et al. A prospective study of cerebral artery atherosclerosis. Stroke 1988;19: 820-825.
Using ultrasound, we investigated whether carotid parameters differed among subtypes of ischemic stroke and evaluated the usefulness of these parameters in discriminating among subtypes. Patients with ischemic stroke admitted to Nippon Medical School Hospital were consecutively recruited and grouped into 3 subtypes based on the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification: cardioembolism (group CE), large-artery atherosclerosis (group LAA), and small-vessel occlusion (group SVO). All subjects underwent carotid ultrasonography to determine maximum intima-media thickness (IMT), maximum systolic velocity (Vmax), minimum diastolic velocity (Vmin), mean velocity, and pulsatility index (PI). Carotid parameters that differed among subtypes were statistically identified. A total of 138 patients were enrolled. Intergroup comparisons revealed that the Vmin of the affected side was significantly lower in group LAA than in group SVO (mean 6 SD, 0.12 6 0.05 m/s vs 0.15 6 0.05 m/s; P 5 .02) and the Vmin of the mean of both sides was lower in group LAA than in group SVO (0.12 6 0.04 vs 0.16 6 0.05; P 5 .03). Multivariate analysis showed that the PI of the affected side was a useful adjunct to discriminate between groups SVO and LAA (odds ratio 5 2.94; P 5 .03, group SVO as control). Receiver operating characteristic curve analysis found that the Vmin of the affected side was the most useful parameter for discriminating between group SVO and group LAA. The PI and the Vmin of the affected side were found to differ among stroke subtypes, and thus these may be useful parameters for discriminating among ischemic stroke subtypes. Key Words: Carotid parameter—stroke subtypes—pulsatility index. Ó 2010 by National Stroke Association
Carotid ultrasonography is a noninvasive technique for examining arteriosclerotic changes in the carotid artery that enables a quantitative assessment of intima-media thickness (IMT). The results of ultrasonographic examination have been reported to accurately reflect those of From the Department of Internal Medicine, Division of Neurology, Nephrology, and Rheumatology, Nippon Medical School, Tokyo, Japan. Received May 29, 2009; revision received July 24, 2009; accepted August 15, 2009. Address correspondence to Yutaka Nishiyama, MD, PhD, Department of Internal Medicine, Division of Neurology, Nephrology, and Rheumatology, Nippon Medical School, 1-1-5 Sendagi, Bunkyo-ku Tokyo 113-8603, Japan. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2010 by National Stroke Association doi:10.1016/j.jstrokecerebrovasdis.2009.08.005
pathological examination.1,2 In the Rotterdam Study, Bots et al3 showed that an increase in the IMT of the common carotid artery (CCA) was a risk factor for ischemic stroke. In the ARIC Study, Chambless et al4 found that the mean IMT of the internal carotid artery and the CCA were associated with the incidence of coronary heart disease. These findings indicate that the IMT of the CCA might be a useful indicator of atherosclerosis and could be related to cardiovascular risk.5 The National Institute of Neurological Disorders and Stroke (NINDS) has classified ischemic stroke into the following subtypes: cardioembolic infarction, lacunar infarction, atherothrombotic infarction, and others.6 An effective treatment strategy hinges on an accurate diagnosis of the ischemic stroke subtype based on this classification system.7-9 Comparing plaque scores calculated on the
Journal of Stroke and Cerebrovascular Diseases, Vol. 19, No. 6 (November-December), 2010: pp 441-449
441
Y. NISHIYAMA ET AL.
442 9
basis of the total IMT of both carotid arteries, Nagai et al found that patients with atherothrombotic infarction had higher scores than patients with other ischemic stroke subtypes. Comparing patients with ischemic stroke grouped into lacunar and nonlacunar infarction subtypes, Cupini et al10 found that independent prediction of subtype was possible based on the IMT of the CCA and the presence of atrial fibrillation. Whether measurement of IMT can be used successfully as an adjunct to distinguish the subtype of ischemic stroke remains unclear, however. Recently, Trial of Org 10172 in Acute Stroke Treatment (TOAST) investigators classified ischemic stroke into the following subtypes: cardioembolism, large-artery atherosclerosis, small-vessel occlusion, stroke of other determined etiology, and stroke of undetermined etiology.11 The TOAST classification system is easy to apply to all patients and exhibits good interobserver agreement. The system has been widely used in epidemiologic studies; however, to the best of our knowledge, there have been no reports on echographic parameters of the carotid artery useful for discriminating among the subtypes in the TOAST classification system. Carotid ultrasonography can be used to measure Doppler hemodynamic parameters, including maximum systolic velocity (Vmax), minimum diastolic velocity (Vmin), mean velocity, and pulsatility index (PI). Gosling et al12 proposed that the PI reflects peripheral vascular resistance, and the correlation between PI and peripheral vascular resistance was subsequently confirmed in a study of the brachiocephalic trunks of healthy volunteers.13 To date, no study has used carotid hemodynamic parameters to discriminate among ischemic stroke subtypes, however. The purpose of the present study was to investigate whether carotid hemodynamic parameters differed among ischemic stroke subtypes and whether assessment of the parameters could serve as a useful adjunct in discriminating among the subtypes.
Methods Patients Patients with ischemic stroke admitted to the Nippon Medical School Hospital from January 2000 to September 2002 were consecutively enrolled in this study. Neurologists in our department evaluated the patients’ neurologic signs and symptoms at admission. An ischemic stroke was defined as an acute disturbance of focal neurologic function persisting for 24 hours. Past medical history and risk factors were obtained from the patients or their families. Each patient underwent a computed tomography scan on admission to exclude cerebral or subarachnoid hemorrhage. Informed consent was obtained from all participants. Our institution’s Committee for the Protection of Human Subjects in Research approved the study design.
Evaluation Brain magnetic resonance imaging (MRI) (T1-, T2-, and diffusion-weighted images) and magnetic resonance angiography (MRA) were performed to detect ischemic lesions in the perfused area of the internal carotid artery and to evaluate the intracranial cerebral arteries. Ischemic stroke of an internal carotid artery area was diagnosed when lesions corresponding to a neurologic finding were detected as an anterior circulation area with low signal intensity on T1-weighted images and high signal intensity on T2- and diffusion-weighted images. Carotid ultrasonography was conducted within 2 weeks after onset of stroke. The investigation was performed using a 7.5-MHz linear-array transducer (LOGIQ 500GE, Yokogawa Medical Systems, Tokyo, Japan). The examiner (Y. N.) was blinded to the clinical information regarding ischemic stroke subtype, infarct site, and risk factors. All examinations were performed with the patient laying supine in a dark room with his or her head held in the midline position or slightly tilted to either side. The examination began by scanning the common and internal carotid arteries cross-sectionally and longitudinally, to roughly evaluate the distribution of atheromatous lesions. During this initial scanning, the optimal insonation angles were determined for estimation of the respective IMT sizes. Scanning regions were set within 45 mm proximal and 15 mm distal to the flow divider, according to the method of Handa et al14 (Fig 1). The maximum IMT (max-IMT) was defined as the maximum height of the IMT of the far wall at the point of measurement. Next, the CCA was examined by pulsed Doppler ultrasonography to obtain information on carotid blood flow. On the longitudinal scans, the sample volume was set in the CCA and displayed as linearly as possible. The insonation site was set 15–45 mm proximal to the flow divider, and a range-gate pulsed Doppler sample volume of 5-7 mm was used to measure the blood flow velocity in the CCA. Special care was taken to maintain the incident angle between the CCA and the beam at 30-60 degrees. The Vmax, Vmin, and PI of the CCA were obtained by analyzing the Doppler flow waveform envelope. The PI was calculated as (systolic velocity - diastolic velocity)/mean velocity. Each flow velocity parameter was determined as the median value of 3 consecutive cardiac cycles. No hemodynamic parameter was estimated at any point at which Vmax exceeded 125 cm/second (suggesting an artery stenosis .50%15). Parameters on the affected and contralateral sides were evaluated separately. For statistical analysis, the value obtained on the affected side, the value obtained on the contralateral side, and the mean of these 2 values were evaluated. When the peak systolic velocity measured by ultrasonography exceeded 200 cm/second in the narrow segment (indicating artery stenosis .70%16), or when MRA detected .60% stenosis during endarterectomy using
CAROTID PARAMETERS AND STROKE SUBTYPES
Figure 1. Schematic showing sites of IMTevaluation in the carotid system.
the asymptomatic carotid stenosis method,17 the condition was defined as stenosis of the extracranial or intracranial cerebral arteries. Percutaneous and/or transesophageal echocardiography was performed in all patients to investigate cardioembolic sources.
Classification and Risk Factors Based on neurologic signs and symptoms, medical history, and brain MRI scan findings, the patients were classified into the following 5 groups according to the TOAST classification criteria: (1) group CE (cardioembolism), (2) group LAA (large-artery atherosclerosis), (3) group SVO (small-vessel occlusion), (4) group OE (stroke of other determined etiology), and (5) group UE (stroke of undetermined etiology). When comparing the differences in echographic parameters among the stroke subtypes, patients classified into group OE or group UE were excluded from the study, because the etiology of cerebral infarction in these patients was highly heterogeneous. The following risk factors were evaluated: age, sex, hypertension, diabetes mellitus, hyperlipidemia, atrial fibrillation, smoking, and history of ischemic stroke. These risk factors and patient histories were determined by interviews and physical and laboratory examinations during hospitalization. To evaluate the effects of drugs on echographic parameters of the carotid artery, the presence or absence of argatroban and ozagrel sodium use was noted, and the effect of these drugs on those parameters was studied.
Statistical Analysis All statistical analyses were performed with StatView for Windows version 5.0 (SAS Institute, Cary, NC) or StatMate III for Windows (Atoms, Tokyo, Japan). The c2 test was used to detect intergroup differences of risk factors, and one-way analysis of variance (ANOVA) was used to examine intergroup age differences. The Student t-test
443
was used to compare echographic parameters with and without argatroban and ozagrel sodium use. One-way ANOVA followed by the Tukey-Kramer post hoc test was used to detect intergroup differences in measured carotid parameters. Logistic regression analysis was performed to evaluate risk factors and parameters related to the different ischemic stroke subtypes. Initially, simple logistic regression was used to test the relationships between unadjusted variables and the difference of 2 stroke subtypes (model 1). Then multivariate associations were tested using the backyard multiple logistic regression method (model 2). Finally, a regression model was applied in which variables showing significant relationships in the simple logistic regression were included mandatorily, and other variables were selected by the backyard method (model 3). The outcome variable (dependent variable) was the difference among stroke subtypes (one subtype served as a control). The independent variables were clinically plausible risk factors (age, sex, hypertension, diabetes, hyperlipidemia, smoking, atrial fibrillation, and history of ischemic stroke) and hemodynamic parameters of both the affected and contralateral sides. The original model included all of the foregoing independent parameters. Backyard stepwise regression was performed, and a P value of ..05 was used for variable removal. To evaluate whether ischemic stroke subtypes could be differentiated using these parameters, receiver operating characteristic (ROC) curve analyses were conducted using StatMate III. StatMate III calculates an ROC curve for each parameter of a data set. The area under the ROC curve (AUC) was calculated, and the AUCs of carotid hemodynamic parameters were compared statistically using the method of Hanley et al.18 The best threshold of carotid hemodynamic parameters for stroke subtype discrimination was determined using the Youden index.17 Data are expressed as mean 6 standard deviation (SD) or percentage. A P value ,.05 was considered statistically significant.
Results Out of the patients admitted, a total of 224 patients with ischemic stroke were finally enrolled. Three of the 224 patients (1.3%) were classified into the OD group; 2 of these patients had coagulation and fibrinolytic abnormalities with malignant tumors, and 1 patient had an antiphospholipid antibody syndrome. Eighty-one patients (36.2%) were classified into the UE group; the etiology was unknown in 17 of these patients, and 2 or more etiologies were found in 64 patients. Table 1 presents profiles of the patients’ risk factors. Significant differences in the prevalences of hypertension, hyperlipidemia, atrial fibrillation, and smoking were found among the 5 groups. In comparison with the other groups, hypertension was higher in group OE,
Y. NISHIYAMA ET AL.
444
Table 1. Characteristics of the 5 clinical subtype groups
N Sex, M/F Age, years, mean 6 SD Hypertension, n, (%) Diabetes, n (%) Hyperlipidemia, n (%) Atrial fibrillation, n (%) Smoking, n (%) Old cerebral infarct, n (%)
Group CE
Group SVO
Group LAA
Group OE
Group UE
P value
54 35/19 71 6 11 26 (48) 13 (24) 25 (46) 37 (68) 9 (17) 7 (13)
64 43/21 66 6 11 55 (85) 21 (33) 46 (72) 0 (0) 32 (50) 15 (23)
22 17/5 73 6 11 15 (68) 7 (32) 10 (45) 0 (0) 12 (55) 5 (23)
3 1/2 73 6 10 3 (100) 1 (33) 1 (33) 0 (0) 1 (33) 0 (0)
81 51/30 68 6 11 52 (64) 24 (30) 45 (56) 17 (21) 33 (41) 18 (22)
.531 .069 .003 .881 .037 ,.001 .002 .519
Table 4 compares the carotid ultrasonography parameters of these 3 groups. Comparing the parameters of the affected side by one-way ANOVA revealed significant differences in Vmin among the 3 groups, as did comparing the parameters of the mean of both sides. A similar comparison of the parameters of the contralateral side revealed no statistically significant differences in maxIMT, Vmax, Vmin, or PI among the 3 groups. Post hoc analysis using the Tukey-Kramer test showed that the Vmin of the affected side was significantly lower in group LAA than in group SVO (P ,.05) and that the Vmin of the mean of both sides was lower in group LAA than in group SVO (P , .05); however, max-IMT, Vmax, and PI of the affected side and the mean of both sides were not significantly different. A similar comparison of the parameters of the contralateral side revealed no statistically significant differences in the parameters among the 3 groups. Because in our statistical analysis the carotid parameters differed significantly only between groups SVO and LAA, logistic regression analysis and ROC curve analysis
hyperlipidemia was higher in group SVO, atrial fibrillation was higher in group CE, and smoking was higher in group LAA. There were no significant differences in gender, age, diabetes, or history of ischemic stroke among the 5 groups. Table 2 shows the differences in echographic parameters of the carotid artery on the basis of the presence or absence of argatroban and ozagrel sodium use in all patients. We compared the values on the affected side, on the contralateral side, and the mean of both sides under the presence and absence of these drugs and found no statistically significant differences in any of the parameters (P . .05). Cerebral artery stenosis was observed in 60 of the 224 patients (26.8%). Twenty-two of these patients belonged to group LAA, while the remaining 38 patients belonged to group UE (Table 3). The 84 patients of groups OE and UE were excluded from subsequent analyses. The remaining 140 patients enrolled in the study included 54 patients in group CE, 64 patients in group SVO, and 22 patients in group LAA.
Table 2. Comparisons of the carotid parameters on the basis of the presence or absence of argatroban and ozagrel sodium use Argatroban (1)
Argatroban (2)
P value
Ozagrel sodium (1)
Ozagrel sodium (2)
P value
1.71 6 0.93 0.72 6 0.21 0.16 6 0.06 1.97 6 0.45
2.01 6 1.24 0.69 6 0.21 0.16 6 0.07 1.88 6 0.62
.12 .32 .66 .63
1.96 6 1.29 0.71 6 0.04 0.17 6 0.07 1.95 6 1.09
1.92 6 1.09 0.68 6 0.05 0.15 6 0.07 2.04 6 0.83
.81 .40 .11 .50
1.59 6 0.79 0.74 6 0.23 0.16 6 0.06 1.92 6 0.43
1.83 6 0.93 0.70 6 0.21 0.17 6 0.07 1.88 6 0.62
.11 .24 .50 .64
1.81 6 0.91 0.73 6 0.21 0.17 6 0.08 1.92 6 0.66
1.75 6 0.89 0.70 6 0.22 0.16 6 0.06 1.86 6 0.52
.59 .35 .12 .43
1.65 6 0.77 0.73 6 0.21 0.16 6 0.05 1.95 6 0.40
1.88 6 0.89 0.69 6 0.20 0.16 6 0.06 1.94 6 0.68
.10 .24 .89 .99
1.84 6 0.90 0.72 6 0.19 0.17 6 0.07 1.94 6 0.69
1.82 6 0.85 0.69 6 0.21 0.16 6 0.06 1.95 6 0.58
.82 .34 .07 .89
Affected side Max-IMT Vmax Vmin PI Contralateral side Max-IMT Vmax Vmin PI Mean of both sides Max-IMT Vmax Vmin PI All data are mean 6 SD.
CAROTID PARAMETERS AND STROKE SUBTYPES
445
Table 3. Distribution of cerebral artery stenosis in the patients examined All patients (n 5 224) Single artery stenosis Affected side ICA occlusion Affected side ICA stenosis .50% Affected side MCA stenosis .50% Affected side ACA stenosis .50% Multiple artery stenosis Bilateral ICA stenosis .50% Bilateral MCA stenosis .50% Affected side ICA occlusion and contralateral side MCA stenosis .50% Affected side MCA stenosis .50% and contralateral side ICA stenosis .50% Affected side MCA stenosis .50% and contralateral side ACA stenosis .50% Group LAA (n 5 22) Single artery stenosis Affected side ICA occlusion Affected side ICA stenosis .50% Affected side MCA stenosis .50% Multiple artery stenosis Bilateral ICA stenosis .50% Affected side MCA stenosis .50% and contralateral side ACA stenosis .50%
between these 2 groups were performed. A total of 86 patients in these 2 groups were examined. Table 5 gives the results of the logistic regression analysis. Atrial fibrillation was eliminated as an independent variable, because neither group included any patient with atrial fibrillation. We used the clinical data of group SVO as a control, and compared the findings of this group with those of group LAA. The results of simple regression analysis indicated positive associations with age (P 5 .004) and PI of the affected side (P 5.03) and negative associations with hyperlipidemia (P 5 .03), Vmin of the affected side (P 5 .006), and Vmin of the contralateral side (P 5 0.02) in group LAA (model 1). The results of the multivariate logistic
49 (21.9%) 5 (2.2%) 16 (7.1%) 26 (11.6%) 2 (0.9%) 11 (4.9%) 5 (2.2%) 2 (0.9%) 1 (0.4%) 2 (0.9%) 1 (0.4%) 18 (81.8%) 1 (4.5%) 6 (27.2%) 11 (50.0%) 4 (18.2%) 3 (13.6%) 1 (4.5%)
regression analysis revealed a significant positive association with age and PI on the affected side in group LAA (model 2). The model in which some variables were mandatorily included and other variables were selected by the backyard method revealed a significant negative association with hypertension in group LAA (model 3). The degree applicable to actually measured levels was 77.9% in model 2 and 76.7% in model 3. ROC curve analyses were conducted to evaluate the discrimination ability in groups SVO and LAA. Figure 2 shows ROC curves for the values of the affected side, the contralateral side, and the mean of both sides, with the maximum AUC associated with the Vmin of the
Table 4. Comparisons of the carotid parameters among 3 clinical subtype groups
Affected side Max-IMT Vmax Vmin PI Contralateral side Max-IMT Vmax Vmin PI Mean of both sides Max-IMT Vmax Vmin PI All data are mean 6 SD.
Group CE
Group SVO
Group LAA
P value
1.75 6 0.82 0.68 6 0.23 0.16 6 0.08 1.88 6 0.72
1.74 6 0.79 0.69 6 0.18 0.15 6 0.05 1.88 6 0.48
2.10 6 1.33 0.62 6 0.19 0.12 6 0.05 2.18 6 0.62
.19 .37 .02 .12
1.57 6 0.72 0.67 6 0.24 0.16 6 0.07 1.83 6 0.58
1.74 6 0.82 0.71 6 0.20 0.17 6 0.06 1.89 6 0.52
1.88 6 0.90 0.66 6 0.19 0.13 6 0.05 2.10 6 0.81
.26 .54 .09 .18
1.67 6 0.66 0.68 6 0.22 0.16 6 0.07 1.86 6 0.57
1.72 6 0.70 0.70 6 0.17 0.16 6 0.05 1.88 6 0.42
1.96 6 0.89 0.64 6 0.17 0.12 6 0.04 2.14 6 0.60
.27 .48 .03 .08
Y. NISHIYAMA ET AL.
446
Table 5. Univariate and mulitivariate analysis to test the relationships between cerebrovascular risk factors and hemodynamic parameters and the difference between 2 clinical subgroups Model 1 OR (95% CI) Gender 1.66 (0.54-5.12) Age 1.06 (1.01-1.11) Hypertension 0.35 (0.11-1.11) Diabetes 0.96 (0.34-2.70) Hyperlipidemia 0.33 (0.12-0.89) Smoking 1.20 (0.45-3.17) Old cerebral infarct 0.96 (0.30-3.04) Affected side max-IMT 1.55 (0.91-2.64) Affected side Vmax 0.11 (0.01-1.97) Affected side Vmin 2.05e-7 (3.23e-8-0.11) Affected side PI 3.00 (1.12-8.03) Contralateral side max- IMT 1.22 (0.69-2.15) Contralateral side Vmax 0.28 (0.02-3.87) Contralateral side Vmin 2.55e-6 (4.66e-11-0.14) Contralateral side PI 1.71 (0.79-3.72)
Model 2 P value .38 .004 .07 .93 .03 .71 .95 .1 .13 .006 .03 .5 .34 .02 .17
affected side. A comparison of the Vmin of the mean of both sides and the second-highest AUC level revealed no significant difference (P 5.719); however, a comparison with the PI of the affected side showed a statistically significant difference (P 5 .018). The best threshold of Vmin of the affected side to discriminate between groups SVO and LAA, identified using the Youden index, was 0.14 m/s. At this threshold, the sensitivity was 73% and the specificity was 59%.
Discussion In our study group, certain parameters of carotid ultrasonography differed among distinct ischemic stroke subtypes. One-way ANOVA revealed significant differences in the Vmin of the affected side and the Vmin of the mean of both sides between groups SVO and LAA. These results indicate that carotid parameters might be useful for discriminating between these 2 groups. Examination of the echographic parameters of the carotid artery based on the presence or absence of argatroban and ozagrel sodium use revealed no differences in any of the parameters. Although an increase in cerebral blood flow after treatment with argatroban or ozagrel sodium has been reported previously,19,20 our results suggest that this effect was not strong enough to change the flow velocity in the carotid artery. The most meaningful value for discrimination was the PI of the affected side in the multivariate logistic regression analysis after adjustment for risk factors. The odds ratio shown in Table 5 could be interpreted to mean that a 1-point elevation of PI on the affected side translates to a 2.9-fold increase in the likelihood of suffering a cerebral infarction due to large artery atherosclerosis.
Model 3
OR (95% CI)
P value
1.06 (1.01-1.11)
.02
2.94 (1.10-7.91)
.03
OR (95% CI)
P value
1.03 (0.97-1.09) 0.15 (0.03-0.71)
.4 .02
0.46 (0.14-1.47)
.19
1.82e-6 (1.03e-13-32.25) 1.36 (0.37-4.91)
0.005 (5.27e-9-4068.11)
.12 .64
.44
Although a model in which some variables were mandatorily included was studied, this model’s goodness of fit could not be considered better than that of the model in which variables were selected by the backyard method. The ROC curve analysis revealed that the Vmin of the affected side was the most useful parameter for discriminating between groups SVO and LAA. At the threshold value of 0.14 m/s, the Vmin of the affected side indicated a moderate sensitivity and specificity for discrimination. Several authors have reported that carotid hemodynamic parameters are influenced by sampling position.15,16 But we believe that our careful choice of the sampling point minimized the positional effects of focal stenosis and dilatation of the carotid artery. The PI is a confirmed marker of peripheral vascular resistance12,13 and also may reflect arteriosclerosis in the cerebral arteries distal from the carotid artery. Most of the patients with cerebral artery stenosis in our study were in group LAA, the group with the highest PI values (Tables 4 and 5). This suggests that the PI of the CCA reflects stenosis of a peripheral cerebral artery. Nakatou et al21 reported that the PI is higher in patients with diabetes than in controls and higher in patients with a history of ischemic stroke than in those with no history of ischemic stroke. That study did not discriminate among ischemic stroke subtypes in the analyses, however. If the PI reflects the resistance of blood vessels associated with cerebral infarction, then our result—that PI is positively associated with group LAA—can be considered reasonable. Other studies have reported varying prognoses for atherothrombotic infarction and lacunar infarction, and the diagnosis of ischemic stroke subtype has been found to be essential to establishing a treatment strategy.22,23 Consequently, we undertook this study to determine
CAROTID PARAMETERS AND STROKE SUBTYPES
A
447
1 0.9 0.8
TP RATIO
0.7 0.6 a-Vmin
0.5 0.4
a-PI
0.3
a-Vmin a-PI a-Vmax a-IMT
0.2
AUC
a-Vmax
0.73 0.64 0.59 0.57
a-IMT
0.1 0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
False Positive Ratio
B
1 0.9 0.8
TP RATIO
0.7 0.6 0.5
c-Vmin
0.4
c-PI
0.3
c-Vmin c-PI c-Vmax c-IMT
0.2
AUC 0.69 0.54 0.55 0.55
c-Vmax c-IMT
0.1 0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
False Positive Ratio
C
1 0.9 0.8
TP RATIO
0.7 0.6
m-Vmin
0.5
m-PI
0.4 0.3
m-Vmin m-PI m-Vmax m-IMT
0.2
m-Vmax
AUC 0.71 0.65 0.57 0.55
m-IMT
0.1 0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
False Positive Ratio Figure 2.
ROC curves of the parameters of the affected side (A), the contralateral side (B), and the mean of both sides (C).
whether carotid hemodynamic parameters can serve as useful adjuncts for discriminating between atherothrombotic infarction and lacunar infarction. Several previous studies have identified increased IMT as a risk factor for ischemic stroke,3-5 and higher plaque scores have been found in patients with atherothrombotic infarction compared with patients with other ischemic subtypes.9 Our findings indicate that the max-IMT of the affected side was higher in group LAA than in group
SVO, although our multivariate analysis showed no significant relationship between the group differences. The discrepancy between our results and earlier results may be due to differences in measurement methods. While our study relied on separate determinations of the IMT of the affected and contralateral sides, previous studies used the plaque score9 and mean IMT of the CCAs together.10,24 Previous reports also have noted discrepancies in the severity of carotid arteriosclerosis between
Y. NISHIYAMA ET AL.
448
extracerebral and intracerebral arteries, as well as differences associated with race and gender.25-27 Further studies involving numerous institutions are needed to determine whether IMT is a useful adjunct for discriminating ischemic stroke subtypes. A reduction in Vmin, a parameter of carotid blood flow, indicates stenosis or occlusion of blood vessels peripheral to the measurement site.28,29 We consider our ROC curve analysis results, which indicate that the Vmin of the affected side has superior discrimination ability in group LAA, not to be contradictory if vascular stenosis associated with cerebral infarction is present. The results of our logistic regression and ROC curve analyses indicate that the echographic parameters in group LAA differed with regard to the PI and Vmin of the affected side. This difference is assumed to be due to an insufficient number of subjects and the presence of a correlation between PI and Vmin. Whether PI or Vmin is more useful for the discrimination of stroke subtypes cannot be inferred from our results; larger-scale prospective studies with more subjects are needed. The carotid parameters differed significantly between groups SVO and LAA, while those in group CE did not significantly differ from those in the other groups. It may be that in group CE, infarction was caused by heart disease and had little to do with the ultrasonographic parameters that reflect cervical arteriosclerosis; many patients in group CE also developed concurrent atrial fibrillation. We believe that atrial fibrillation must have changed the carotid blood flow waveform on the ultrasonogram, and also may have affected the parameter measurements. Our results show a positive association with age in group LAA. This suggests that age might be a risk factor relative to atherothrombotic infarction. The relationship between age and stroke has been controversial in previous studies.30-32 Although some studies have found an association between age and lacunar infarction,30,31 Reed et al32 reported a relationship between age and high prevalence of atherosclerosis of the large cerebral arteries. Further studies to investigate this issue are needed. Our study has several limitations. Because the patients classified into groups OE and UE by the TOAST classification criteria were excluded from the investigation of echographic parameters (37.5% of all subjects), the relationship between echographic parameters and stroke subtypes OE and UE cannot be evaluated based on our results. Patients with these subtypes should be investigated within their subtype and placed in more detailed categories. A second limitation relates to the applicability of our data. As a retrospective, single hospital–based registration study, the present study may not have produced results directly applicable to the general population. A multicenter study to confirm our findings is needed. In conclusion, the difference in the PI or the Vmin of the affected side between groups SVO and LAA was
considered to reflect the severity of arteriosclerosis. Thus, we have shown that carotid hemodynamic parameters differ among ischemic stroke subtypes and that these parameters may be useful adjuncts for the discrimination of subtypes.
References 1. Pignoli P, Tremoli E, Poli A, et al. Intimal plus medial thickness of the arterial wall: a direct measurement with ultrasound imaging. Circulation 1986;74:1399-1406. 2. Wong M, Edelstein J, Wollman J, et al. Ultrasonic pathological comparison of human arterial wall: verification of intima-media thickness. Arterioscler Thromb 1993; 13:482-486. 3. Bots ML, Hoes AW, Koudstaal PJ, et al. Common carotid intima-media thickness and risk of stroke and myocardial infarction. Circulation 1997;96:1432-1437. 4. Chambless LE, Heiss G, Folsom AR, et al. Association of coronary heart disease incidence with carotid arterial wall thickness and major risk factors: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Epidemiol 1997;146:483-494. 5. Grobbee DE, Bots ML. Carotid artery intima-media thickness as an indicator of generalized atherosclerosis. J Intern Med 1994;236:567-573. 6. Whisnant JP, Basford JR, Bernstein EF, et al. Special report from the National Institute of Neurological Disorders and Stroke. Classification of cerebrovascular diseases. Stroke 1990;21:637-676. 7. White H, Boden-Albala B, Wang C, et al. Ischemic stroke subtype incidence among whites, blacks, and Hispanics. Circulation 2005;111:1327-1331. 8. Arboix A, Morcillo C, Garcia-Eroles L, et al. Different vascular risk factor profiles in ischemic stroke subtypes: a study from the Sagrat Cor Hospital of Barcelona Stroke Registry. Acta Neurol Scand 2000;102:264-270. 9. Nagai Y, Kitagawa K, Sakaguchi M, et al. Significance of earlier carotid atherosclerosis for stroke subtypes. Stroke 2001;32:1780-1785. 10. Cupini LM, Pasqualetti P, Diomedi M, et al. Carotid artery intima-media thickness and lacunar versus nonlacunar infarcts. Stroke 2002;33:689-694. 11. Adams HP, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke: definition for use in a multicenter clinical trial. Stroke 1993;24:35-41. 12. Gosling RG, King DH. Arterial assessment by Doppler shift ultrasound. Proc R Soc Med 1974;67:447-449. 13. Legarth J, Nolsoe C. Doppler blood velocity waveforms and the relation to peripheral resistance in the brachial artery. J Ultrasound Med 1990;9:449-453. 14. Handa N, Matsumoto M, Maeda H, et al. Ultrasonic evaluation of early carotid atherosclerosis. Stroke 1990; 21:1567-1572. 15. Grant EG, Benson CB, Moneta GL, et al. Carotid artery stenosis: gray-scale and Doppler US diagnosis. Society of Radiologists in Ultrasound Consensus Conference. Radiology 2003;229:340-346. 16. Koga M, Kimura K, Minematsu K, et al. Diagnosis of carotid artery stenosis greater than 70% with power Doppler duplex sonography. AJNR Am J Neuroradiol 2001;22:413-417. 17. Executive Committee for the Asymptomatic Carotid Atherosclerosis Study. Endarterectomy for asymptomatic carotid artery stenosis. JAMA 1995;273:1421-1428.
CAROTID PARAMETERS AND STROKE SUBTYPES 18. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148:839-843. 19. Kawai H, Yuki S, Sugimoto J, et al. Effects of a thrombin inhibitor, argatroban, on ischemic brain damage in the rat distal middle cerebral artery occlusion model. J Pharmacol Exp Ther 1996;278:780-785. 20. Oishi M, Mochizuki Y, Hara H, et al. Effects of sodium ozagrel on hemostatic markers and cerebral blood flow in lacunar infarction. Clin Neuropharmacol 1996;19:526-531. 21. Nakatou T, Nakata K, Nakamura A, et al. Carotid haemodynamic parameters as risk factors for cerebral infarction in type 2 diabetic patients. Diabetic Med 2004;21:223-229. 22. de Jong G, van Raak L, Kessels F, et al. Stroke subtype and mortality: a follow-up study in 998 patients with a first cerebral infarct. J Clin Epidemiol 2003;56:262-268. 23. Grau AJ, Weimar C, Buggle F, et al. Risk factors, outcome, and treatment in subtypes of ischemic stroke. The German Stroke Data Bank. Stroke 2001;32:2559-2566. 24. Touboul PJ, Elbaz A, Koller C, et al. Common carotid artery intima-media thickness and brain infarction. The GENIC case-control study. Circulation 2000;102:313-318. 25. Nijasri CS, Auruma C. Risk factors for atherosclerosis of cervicocerebral arteries: intracranial versus extracranial. Neuroepidemiology 2003;22:37-40.
449 26. Soo JL, Soo JC, Heui-Soo M, et al. Combined extracranial and intracranial atherosclerosis in Korean patients. Arch Neurol 2003;60:1561-1564. 27. Wityk RJ, Lehman D, Klag M, et al. Race and sex differences in the distribution of cerebral atherosclerosis. Stroke 1996;27:1974-1980. 28. Kamouchi M, Kishikawa K, Okada Y, et al. Reappraisal of flow velocity ratio in common carotid artery to predict hemodynamic change in carotid stenosis. AJNR Am J Neuroradiol 2005;26:957-962. 29. Saito K, Kimura K, Nagatsuka K, et al. Vertebral artery occlusion in duplex color-coded ultrasonography. Stroke 2004;35:1068-1072. 30. Rojas JI, Zurru MC, Romano M, et al. Acute ischemic stroke and transient ischemic attack in the very old: risk factor profile and stroke subtype between patients older than 80 years and patients aged less than 80 years. Eur J Neurol 2007;14:895-899. 31. Telman G, Kouperberg E, Sprecher E, et al. Distribution of etiologies in patients above and below age 45 with first-ever ischemic stroke. Acta Neurol Scand 2008;17: 311-316. 32. Reed DM, Resch JA, Hayashi T, et al. A prospective study of cerebral artery atherosclerosis. Stroke 1988;19: 820-825.