This study was a retrospective analysis of a prospectively collected database of acute ischemic stroke patients (within 7 days from stroke onset) between January 2013 and December 2017. Patients classified by the Trial of Org 10172 in Acute Stroke Treatment (TOAST) as having large artery atherosclerosis with symptomatic proximal internal carotid artery (pICA) disease were enrolled . Significant stenosis of the pICA was defined as > 50% diameter reduction to near-occlusion (artery beyond the stenosis has collapsed, but has remaining patency). Symptomatic pICA disease was considered present when significant stenosis of the pICA was the likely cause of index middle cerebral artery (MCA) or anterior cerebral artery (ACA) infarction [11, 12].
Patients were excluded if they had (1) embolic sources from the heart (e.g., atrial fibrillation or valvular heart disease) or the aorta, (2) other uncommon etiologies such as dissection or moyamoya disease, (3) tandem stenotic lesions at the intracranial ICA and MCA or ACA, or (4) results on contrast enhanced magnetic resonance angiography (MRA) that did not allow assessment of geometric properties (e.g., complete occlusion or poor MRA quality). The study protocol was approved by the institutional review board of Kyung Hee University Hospital, which waived the need for informed consent because of its retrospective nature.
Clinical data and neuroimaging
Demographic characteristics, vascular risk factors, concurrent medications, and baseline laboratory results were obtained from the prospectively acquired stroke database. Hyperlipidemia was defined by patient having past history of hyperlipidemia, using a lipid-lowering agent including statins or patient diagnosed hyperlipidemia at admission for LDL-cholesterol levels .
On the day of admission, all patients underwent magnetic resonance imaging and MRA in the following sequence: DWI, fluid attenuated inversion recovery imaging, gradient echo imaging, T1- and T2-weighted imaging, and intracranial and extracranial contrast enhanced MRA.
The patterns of infarction on DWI were classified according to the number, location, and maximum diameter of lesions [14, 15]. Lesions were classified as single or multiple, and as large (≥15 mm) or small (<15 mm) based on their maximum diameter. Single lesions were further classified as cortical (small), subcortical (small or large), or cortico-subcortical (large). Multiple lesions were classified as small scattered lesions or a large lesion with additional lesions. The topography of ischemic lesions according to vascular territory was determined with reference to published templates . For further analysis, lesion pattern was dichotomized according to the presence or absence of a large lesion. Lesion location was divided in to cortical, subcortical and cortico-subcortical lesions.
Carotid plaque and geometry
The presence and location of plaques and the degree of stenosis in the pICA were assessed on contrast enhanced MRA. Atherosclerotic plaques of the carotid artery were classified as being high-apical or low-body plaques. High-apical plaques were defined as plaques in the transitional zone of the bulb and the proximal cervical ICA segment, with or without the involvement of the body segment. Low-body plaques were defined as those in the transitional zone of the common carotid artery (CCA) and the bulb, and are located mainly in the lower body segment, with or without the involvement of the apical segment (Supplementary Fig 1). To differentiate among type of enlarged plaques, the location of the main plaque component and the level of the most severe stenosis were considered . Two independent researchers evaluated the type of plaque and dichotomized the location to high-apical or low-body type. If any discrepancy exists, a consensus meeting was held to finalize the type of atherosclerosis, with more than three researchers and considering all the imaging modalities available.
The geometry of the carotid artery was quantitatively analyzed by a modification of previous methods . The ICA-external carotid artery (ECA) angle was defined as the angle between the projections of the ICA0-ICA5 and ECA0-ECA5 vectors onto the bifurcation plane (Supplementary Fig 1). The CCA-ICA and CCA-ECA angles were defined similarly. Moreover, ICA planarity was defined as the angle between the out-of-plane components of the CCA and ICA vectors. The ICA-to-CCA diameter ratio was calculated as the ICA5 diameter divided by the CCA5 diameter.
Geometric factors that could affect the risk of embolism or vulnerability of plaque were also evaluated. These factors included 1) the degree of stenosis measured by North American Symptomatic Carotid Endarterectomy Trial (NASCET) criteria; 2) the presence of ulcer, defined as a niche in the plaque surface >2 mm in depth on contrast enhanced extracranial MRA ; 3) carotid webs, defined as thin intraluminal filling defects along the posterior wall of the carotid bulb in oblique sagittal reformatted image on contrast enhanced extracranial MRA ; and 4) ICA kinking, defined as an extreme form of tortuous ICA with angulation of the vessel’s axis ≤ 90° .
Demographic characteristics, vascular risk factors, concurrent medications, laboratory findings, geometrical factors, and ischemic lesion patterns were compared in patients grouped by plaque location. Pearson chi-square tests, independent t-tests, and Mann-Whitney U-tests were used as appropriate. In addition, clinical and imaging characteristics were compared according to different lesion patterns and location on DWI. Multivariable binary logistic regression analysis was performed to investigate the independent association between lesion patterns (small lesions only vs. large lesion) and the type of plaque and the geometric properties of the carotid bifurcation. Odds ratios (ORs) were calculated, along with 95% confidence intervals (95% CIs). A p value <0.05 was considered statistically significant. All statistical analyses were performed using SPSS 22.0 for Windows (IBM Corp. Armonk, NY, USA).