Differential value of intima thickness in ischaemic stroke due to large‐artery atherosclerosis and small‐vessel occlusion

Abstract No study has examined the differential value of arterial intima thickness in the subtypes of acute ischaemic stroke. This study aimed to assess whether intima thickness of carotid artery (CIT), radial artery (RIT) and dorsalis pedis artery (PIT) have an independent and additive value in differentiating ischaemic stroke subtypes due to large‐artery atherosclerosis (LAA) or small‐vessel occlusion (SVO). One hundred and sixty‐one patients with LAA and 79 patients with SVO were recruited. CIT, RIT and PIT were measured with a 24‐MHz ultrasound transducer. Binary logistic regression analysis was used to evaluate the differential values of the different parameters in the two subtypes. ROC curve analyses were plotted to compare the differential performance of different parameters and the combination model. Both RIT and PIT were substantially thicker in LAA than in SVO stroke patients. RIT and carotid intima‐media thickness had similar performances in differentiating stroke subtypes. Introduction of RIT to traditional atherosclerotic associated risk factors had a marginal satisfactory differential performance for LAA and SVO stroke patients (AUC 0.775). RIT is a promising parameter for LAA and SVO subgroup classification. The combination of RIT and traditional risk factors might be a promising tool for differentiating ischaemic stroke subgroups.

occurs in the intima layer of large arteries, including intracranial arteries, as well as extracranial arteries, such as the carotid artery, radial artery and dorsalis pedis artery. In contrast, the pathological changes in SVO include arteriolosclerosis and cerebral amyloid angiopathy. Arteriolosclerosis is mainly characterized by loss of smooth muscle cells from the tunica media, deposition of fibro-hyaline material in the vessel wall, and thickening of the vessel wall. Meanwhile, cerebral amyloid angiopathy is featured by the progressive deposit of congophilic, βA4 immunoreactive amyloid protein in the walls of small-medium sized arteries, as well as arterioles. 5 The differentiation of ischaemic stroke into LAA and SVO subtypes is of great clinical importance in management strategies as well as in prognostication. To date, there are several modalities for differentiation of ischaemic stroke subgroups, including magnetic resonance angiography (MRA), computed tomography angiography (CTA) and transcranial doppler (TCD). MRA and CTA can be used to reveal the angiographical stenosis of cerebral arteries directly; however, both techniques are costly, and allergy to contrast should not be underestimated. TCD can reveal angiographical stenosis indirectly by detecting accelerated blood flow dynamics; however, its precision is limited. In addition, all three techniques are macroscopic in nature, failing to reveal the pathological changes that occur in the arterial intima or media layer microscopically, especially at an early stage of atherosclerosis. There have also been studies involving differentiation of LAA and SVO subtypes with carotid intima-media thickness (CIMT) 6 ; however, CIMT measurement involves combined measurement of both the intima layer and media layer, and its ability to reveal atherosclerosis is limited. Hence, CIMT measurement is no longer recommended to perform in asymptomatic subjects according to the recently published European Society of Cardiology (ESC) guidelines. 7 Recently, we measured the intima and media thickness of carotid and radial arteries in a group of LAA patients and healthy controls by applying a 24-MHz high-resolution ultrasound and found that intima layers were thicker in LAA patients. Such a method could be used to anatomically predict atherosclerosis in the intima layer. 8 In this study, we aimed to differentiate LAA from SVO by applying the same 24-MHz frequency high-resolution ultrasound with the hypothesis that the intima layer is thicker in the LAA subtype than in the SVO subtype, and that it can be used to differentiate the two subtypes of ischaemic stroke.

| Study population
Three hundred and thirty-seven consecutive patients in neurol-

| Arterial ultrasonography and image analysis
The imaging modalities have been described previously. 8

| Statistical analysis
Statistical analysis was performed using SPSS 23.0 (SPSS Inc., Chicago, IL, USA). Continuous data were presented as mean ±SD. Categorical data were presented as numbers (percentages).
Kolmogorov-Smirnov analysis was used to test for normality. An independent t test was performed to compare normally distributed continuous data between groups. Whilst Mann-Whitney U test was used to compare non-normally distributed continuous data. The chi-square test was used to compare categorical data between groups. Binary logistic regression analysis was used to select potential differential parameters for LAA and SVO subtypes. Both forward: LR selection (probability of entry being <0.10) and backward: LR selection (probability of entry and removal being 0.05 and 0.10, respectively) procedures were sequentially performed to select potential differential variables for ischaemic stroke subtypes. Variables significantly contributed to the regression model in both forward: LR selection and backward: LR selection procedures were kept in the combination models and receiver operating characteristic (ROC) curve analysis was performed. A two-tailed p-value <0.05 was considered statistically significant.

| Demographic features
The demographic features of the LAA stroke group and SVO stroke group are listed in Table 1. There were no significant differences between LAA and SVO subgroups in terms of age, sex, SBP, DBP, BMI, the prevalence of coronary artery disease (CAD), hypertension and serum levels of fasting glucose, TC, TG, HDL-C, LDL-C and sdLDL. In patients with LAA stroke, the serum levels of NEFA, ApoA-1, ApoB and Lp(a) were higher than those in patients with SVO stroke. The prevalence of DM and smoking was also higher in the LAA group than in the SVO group (Table 1).

| Ultrasonic measurements in carotid, radial and dorsalis pedis arteries
Ultrasound measurements in patients with LAA and SVO ischaemic stroke subtypes are also listed in Table 1. Measurements of CMT, CIMT, RIT, RMT, RIMT, PIT, PMT and PIMT were substantially higher in the LAA stroke group than in the SVO group (Table 1, Figure 1).
For intracranial LAA subgroup, both CIT and RIT tend to be thinner than those of extracranial and combined groups; however, neither of them was statistically significant amongst groups (Table 2).

| Differentiating value of carotid and radial ultrasound parameters for LAA and SVO subtypes
All parameters, which differed significantly between the LAA and SVO subgroups were introduced into binary logistic regression analysis. After validation with both forward: LR selection and backwards: LR selection, only DM, NEFA, ApoA-1, Lp(a) and RIT remained in the regression model (Table 3).  13 However, the relationship between CIMT and stroke subtypes remains controversial. Some studies found that CIMT was higher in large-vessel than in small-vessel stroke patients, 6,14-16 whilst others found no difference. 17 This paradox indicated that CIMT alone might not be an ideal parameter to reveal pathological changes of different stroke subgroups and for stroke subtype classification. In addition, thickened intima-media thickness (IMT) encompasses not only intimal thickening but also medial hypertrophy, which is in response to long-term hypertension. With the high-resolution ultrasound transducer, we were able to measure intima thickness and media thickness separately and found that not only CIMT but also CMT, RIT, RMT, PIT and PMT were different between LAA and SVO stroke, indicating that these parameters might serve as potential candidates for stroke subtype classification. Surprisingly, there was no difference in CIT between the two groups. This might be due to several reasons. First, the carotid artery is a large elastic vessel, whilst the intracranial arteries are mainly medium-sized muscular vessels, and there are differences in anatomical structures. Secondly, both CIT and RIT were reported to be related to age, SBP and HDL-C 8 ; however, there was no difference between the LAA and SVO subgroups regarding all the above-mentioned parameters in this study. Thirdly, the pathologies in LAA and SVO are complicated, and atherosclerosis might contribute to both subtypes; the carotid artery is more vulnerable to atherosclerosis anatomically in nature and blood flow dynamics, leading to poor differentiating value of carotid parameters for the two groups.
In this study, we found that RIT might help to differentiating between LAA and SVO stroke subtypes with modest differential power (AUC: 0.646). Similar performance of RIT was noticed in our previous study. 8 In addition, RIT was reported as a predictor of adverse cardiovascular events, including stroke. 18 This indicates that RIT could be useful in diagnosis and outcome prediction of atherosclerosis-related diseases. This might be due to the similarity in the anatomical structure of the radial, cerebral and coronary arteries. All three arteries are medium-sized muscular arteries with Note: Data were expressed as mean ±SD or n (%). The highlighted bold value were used to clearly show parameters with a p value <0.05.
In conclusion, RIT is a promising parameter for LAA and SVO subgroup classification. The combination of RIT and traditional risk factors might be a promising tool for differentiating ischaemic subtypes.

ACK N OWLED G EM ENT
This work was supported by grants from the Natural Science

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.