A prospectively maintained registry database at a tertiary care hospital was reviewed to identify all consecutive patients diagnosed with severe symptomatic cerebral atherosclerotic disease between January 2002 and December 2012.

The Asan Medical Center Institutional Review Board approved this study (approved project number: 2014-0772); the requirement for informed consent of patients was waived off. Written informed consent had obtained from all patients for the angiographic procedures.

The inclusion criteria were: 1) patients of ischemic stroke or transient ischemic attack diagnosed on the basis of neurologists’ clinical assessment and findings of magnetic resonance imaging or computed tomography; 2) age range: 30–80 years; and 3) severe (≥70%) stenosis or occlusion of cerebral arteries detected by cerebral angiography [17,18].

Excluded were the patients who had 1) emboligenic heart disease, 2) uncommon vascular disorders such as dissection, vasculitis, or Moyamoya disease, and 3) who had undergone revascularization with thrombolysis, thrombectomy, or endarterectomy and those who exhibited re-stenosis after stenting/angioplasty [17,19,20].

All patients underwent high-resolution, biplane digital subtraction angiography (DSA) (Siemens Axiom Artis Zee biplane angiography system, Siemens AG, Medical Solutions, Erlangen, Germany) of the internal carotid artery (ICA), common carotid artery, vertebral artery, and/or proximal subclavian artery. All lesions were described in terms of location (anterior vs. posterior circulation) and whether they were intracranial (IC) or EC (Fig. 1) [13]. For a more precise localisation of the lesion, the ICA was divided into its embryological vascular segments and the corresponding remnant branch [21]. The ICA segments were then categorized as supraclinoid-terminal, petro-cavernous, or bulb-cervical [13].

Lesions in the petro-cavernous segment confined to the middle of the intradural and EC lesions tended to be closer to the intradural artery as compared to the EC artery. Thus, the petro-cavernous segment, which included the cavernous segment and the bony part of the carotid canal in the petrous bone, was considered as IC artery. The petro-cavernous segment is also known to have deficient adventitia, which renders it similar to the intradural cerebral artery including the supraclinoid-terminal segment beyond the ophthalmic artery, but dissimilar to the bulb-cervical segment of the ICA. The junction between the EC and IC arteries was set as the ICA segment of the lower margin of the carotid canal in the petrous bone for the ICA, and at the level of the foramen magnum for the vertebral artery.

The reason why we only included symptomatic severe stenosis attributed to the practice guideline that catheter angiography was not routinely indicated for patients with IC or EC stenosis especially in asymptomatic ones because sensitivity and specificity with the noninvasive studies such as transcranial Doppler or Duplex sonography, computed tomography angiography, or magnetic resonance imaging (MRI) with MRA was lower compared with standard catheter-based angiography [22]. Therefore, there were no comparable control patients who underwent catheter angiography.

Single vascular lesion was defined as the presence of only one severe lesion. Multiple vascular lesions were defined as presence of at least two severe lesions. In patients with multiple lesions, the lesion that was responsible for the patient’s symptoms at presentation was deemed to be the main lesion. The stenosis on the angiogram was quantified by an on-site neuroradiologist on the basis of the methods used in the Warfarin-Aspirin Symptomatic Intracranial Disease Trial at the time of diagnosis [17,18,23].

Data on the following known risk factors for atherosclerosis were collected: age, sex, HTN, diabetes mellitus (DM), dyslipidemia, coronary heart disease (CHD), CS, alcohol use, previous stroke, family history of stroke, high body mass index (≥30 kg/m2), metabolic syndrome, high erythrocyte sedimentation rate (ESR; ≥9 mm/h), and high C-reactive protein (CRP) level (≥0.6 mg/dL) [24]. Patient age was categorized into 10 year intervals.

HTN was deemed to be present if the patient had a known history of HTN (systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg), or was on antihypertensive treatment. DM was considered to be present if the patient met at least one of the following criteria: past history of known diabetes; glycated hemoglobin ≥6.5%; serum glucose level after an 8 hours fast ≥126 mg/dL; glucose level 2 hours after a 75 g oral glucose tolerance test ≥200 mg/dL; and serum glucose level ≥200 mg/dL on random testing. Dyslipidemia was deemed to be present if the patient qualified at least one of the following criteria: past history of hyperlipidemia or current hyperlipidemia treatment; total cholesterol ≥200 mg/dL; triglycerides ≥200 mg/dL; low-density lipoprotein ≥130 mg/dL; and high-density lipoprotein ≤40 mg/dL. CHD was defined as myocardial infarction, angina, or a history of coronary artery bypass graft or percutaneous coronary intervention. A patient was deemed to use alcohol if they currently drank alcohol or had quit drinking less than 6 months previously. Metabolic syndrome was defined on the basis of the National Cholesterol Education Program criteria, i.e., presence of three or more of the following: 1) abdominal obesity; 2) elevated triglyceride levels (≥150 mg/dL); 3) decreased high-density lipoprotein cholesterol levels (<40 mg/dL for men and <50 mg/dL for women); 4) high blood pressure (systolic blood pressure 140 mmHg, diastolic blood pressure ≥90 mmHg, or use of antihypertensive medication); and 5) fasting glucose levels ≥110 mg/dL Abdominal obesity was defined as waist circumference of ≥90 cm for men and ≥80 cm for women according to the revised Asia-Pacific criteria proposed by the World Health Organization Western Pacific Region [25].

The US Center for Disease Control and Prevention defines non-smokers as those who currently do not smoke cigarettes (i.e., both former and never smokers). Never smokers are defined as those who have never smoked or who have smoked fewer than 100 cigarettes in their entire lifetime. Former smokers are defined as those who have smoked at least 100 cigarettes in their lifetime but have not smoked for at least 6 months. Current smokers are defined as those who have smoked 100 cigarettes in their lifetime and currently smoke cigarettes every day (daily) or some days (non-daily) [26]. Ever smokers consisted of current and former smokers.

Categorical variables are presented as frequencies and percentages. Continuous variables are expressed as means and standard deviations. Continuous variables were analyzed using Student’s t-test and categorical variables were analyzed using the chi-square test. Logistic regression analysis was used to identify independent risk factors for ECAS and ICAS. For the latter analysis, only covariates with a P value less than 0.20 were retained for further analysis. Among the variables which were significant in the multivariate analysis, we focused on CS as a modifiable risk factor of ECAS.

To reduce the effect of selection bias, we performed a propensity score-matching (PSM) method between the smoking group and non-smoking groups. Propensity scores were estimated by using a logistic regression model of the following covariates: sex, DM (only young group), dyslipidemia, alcohol, metabolic syndrome (only young group), body mass index (only old group) which showed significant difference between the smoking and non-smoking group. According to our study assumption in which there is age-dependent difference in the incidence of EC and IC lesions, we analyzed EC vs. IC lesion difference in two different age-subgroups; young (<50) and old (≥50). Histograms showing the density of propensity score distribution in the CS and non-CS groups before and after matching were presented in Fig. 2.

All statistical analyses were performed using R Statistical Software (version 3.1.2.; R Foundation for Statistical Computing, Vienna, Austria). All P-values were two-tailed and a P-value <0.05 was considered indicative of a statistically significant association.

Of 1,709 patients, 794 (46.5%) had EC lesions and the other 915 (53.5%) had IC lesions.

Baseline characteristics of patients in ECAS and ICAS groups are shown in Table 1. There were significantly more smokers (65.1% vs. 44.7%, P<0.001) and male (80.5% vs. 59.3%, P<0.001) in the ECAS group compared to ICAS. Mean age was high in the ECAS group and the age distribution was significantly different between two groups. The young age group (<50 years) had more IC lesion compared to the old age group (≥50 years) (84.5% vs. 47.6%, P<0.001). HTN, DM, CHD, metabolic syndrome, high ESR, and high CRP were observed more frequently in the ECAS group, while dyslipidemia, alcohol consumption, previous history of stroke, familial history of stroke, high BMI were not associated with any group. Anterior circulation was involved more in the ECAS group (88.1% vs. 76.5%), and multiple lesions were also found more in the ECAS group (45.2% vs. 28.2%).

The findings from univariate and multivariate analyses and their association with ECAS are also summarized in Table 1. Of the 794 patients with ECAS, 516 (65.1%) of these were smoker, whereas the remaining 915 patients with ICAS showed lower prevalence of smoker (44.7%). In addition, older patients were more likely to have ECAS. Male gender, HTN, DM, CHD, metabolic syndrome, high ESR, high CRP and anterior circulation were significantly associated with ECAS, while familial history of stroke and single vascular lesion were associated with ICAS. In multivariate analysis, CS (odds ratio [OR], 1.84; 95% confidence interval [CI], 1.26 to 2.68), male (OR, 2.76; 95% CI, 1.80 to 4.23), old age (OR, 1.06; 95% CI, 1.04 to 1.08), CHD (OR, 2.14; 95% CI, 1.39 to 3.28), high ESR (OR, 1.73; 95% CI, 1.20 to 2.49), multiple lesions (OR, 2.17; 95% CI, 1.52 to 3.08), anterior lesion (OR, 2.88; 95% CI, 1.94 to 4.29) were all independently associated with the likelihood of having ECAS after adjusting for retained covariates.

For the analysis, we created a propensity score-matched cohort of 38 (young) and 303 (old) pairs of smokers and non-smokers (Table 2). There were no significant differences in baseline characteristics between smokers and non-smokers after matching. In the cohort, CS group showed significantly more EC lesion in the old age population (65.7% vs. 47.9%, P<0.001). On the other hand, there was no relationship between CS and ECAS in the young group (Figs. 3, 4). Anterior circulation involvement was more frequent in old smokers with borderline significance (83.1% vs. 76.5%, P=0.057).