Silent cerebrovascular disease is the most common incidental finding on brain imaging [1]. One of the best-defined manifestations of silent cerebrovascular disease is white matter hyperintensities (WMHs) of presumed vascular origin [2], which identifies WMHs from others relating to demyelinating diseases, such as infectious, toxic, or metabolic processes. WMHs had been studied over half a century and experts are keeping on investigating the etiopathology as WMHs could lead to increased risk of stroke, dementia, death [3], and impact on the quality of life in stroke survivors [4]. It is recognized that WMHs are associated with age, hypertension, small vascular disease [5], chronic carotid atherosclerosis [6], and carotid artery stenosis [7]. The severity of the internal carotid artery (ICA) stenosis is assumed to be an important factor for WMHs [8, 9]. It is necessary to study protective factors for WMHs further.

ICA stenosis is defined as the maximal luminal narrowing and its severity is rated according to the percentage of stenosis. The 70–99% has been termed severe ICA stenosis that is the main target of the treatment. Studies remain largely constrained to the maximal degree of ICA stenosis while ignoring the collateral blood flow status. The circle of Willis (CoW) is the main collateral system that connects the bilateral carotid systems with the posterior circulation. In patients with severe ICA stenosis, collateral flow provided by the CoW is paramount to maintaining the cerebral perfusion pressure. Three-dimension time-of-flight magnetic resonance angiography (3D TOF MRA) is routinely used for imaging the proximal intracranial vasculature and is available for evaluating cerebral collateral with the advantages of high spatial resolution and lack of an intravenous contrast agent. It has been used for the assessment of the collateral function of the CoW [10]. We investigated whether complete CoW could protect patients with severe ICA stenosis from WMHs.

Materials and methods

Consecutive patients admitted to the Cerebrovascular Disease Center, Department of Neurology, People’s Hospital, China Medical University, between April 2016 and September 2017 were included if they (1) had recordings on age, sex, and risk factors of cerebrovascular diseases including the history of smoking, hypertension, diabetes mellitus, coronary artery disease, atrial fibrillation, and previous stroke, (2) had undergone blood tests, Doppler ultrasound of carotid arteries, transcranial Doppler, electrocardiogram, echocardiography, and brain MR image, (3) had unilateral ICA stenosis ≥ 70% (including occlusion) and contralateral ICA stenosis < 50% or no contralateral stenosis. The patients were excluded if they had (1) ICA stenosis of non-atherosclerotic, (2) infarctions affecting the assessment of WMHs, (3) other possible specific causes of WMHs, (4) possible risks of cardiac embolism according to Adams et al. [11], (5) concomitant diseases like brain tumors, abscess, hydrocephalus, or severe situations such as heart, renal, or respiratory function failure. The Ethics Committee of People’s Hospital, China Medical University, approved the study protocol. All patients or their guardians signed informed consent with regard to participation in our study.

Imaging analysis

The imaging protocols included the series of T1-weighted, T2-weighted, fluid-attenuated inversion recovery (FLAIR), diffusion-weighted image, and 3D-TOF MRA using 3.0T MR imaging (Discovery 750, GE Healthcare, USA) or 1.5T (MR 355, GE BRIVO, USA).

Periventricular WMHs and deep WMHs were judged according to Kim et al. [12] and evaluated separately on the axial section of the FLAIR sequence according to King et al. [13]. Briefly, periventricular WMHs were rated according to the distance perpendicular from the ventricle in the axial plane (thickness of a continuous band or distance of an extension towards the deep white matter) in four categories: (0) < 3 mm, (1) 3 to 10 mm, (2) 10 to 20 mm, and (3) ≥ 20 mm. Deep WMHs were rated according to the largest diameter in four categories: (0) < 3 mm for all lesions, (1) 3 to 10 mm for a single lesion or < 20 mm for grouped lesions, (2) 10 to 20 mm for a single lesion or > 20 mm for grouped lesions, and (3) > 20 mm for a single or confluent lesion. According to Wardlaw et al. [14], lacunar infarction was rated according to the number in three categories: (0) none, (1) 1 to 3, (2) ≥ 4. The perivascular space was evaluated as existent or not.

The circle of Willis (CoW) comprises the A1 segments of the two anterior cerebral arteries, the anterior communicating artery, the P1 segments of the two posterior cerebral arteries and the two posterior communicating arteries. The CoW was graded as complete if all the aforementioned vessels were visible and was graded as incomplete if any of the aforementioned vessels was invisible [15]. According to the completeness of CoW, our included patients were categorized into two groups: those with complete CoW and those with incomplete CoW. Intracranial artery stenosis was evaluated using MRA [16] and extracranial artery stenosis was evaluated using Duplex ultrasound (Siemens Antares, a 6–13 MHz transducer).

All the vascular assessment was conducted with the combination of source imaging with multiplanar reconstruction on 3D-TOF MRA. All the evaluations were performed on each cerebral hemisphere and were performed by two neurologists with the software PACS. When the two neurologists were in different opinions, it was settled by a discussion. If consensus could not be reached, a more senior expertise would be consulted.

Statistical analysis

An unpaired Student t test, or unpaired nonparametric Mann–Whitney U or χ2 test was used to conduct the univariate analysis between patients with complete CoW and incomplete CoW. Logistic regression analysis was used to examine the independent effect of the CoW on WMHs. The covariates included age, sex, and the variables with a P value of < 0.1 in the univariate analysis. A paired nonparametric Mann–Whitney U or χ2 test was used to conduct the univariable analysis between the two sides, ipsilateral and contralateral to stenosed ICA in all included patients, in the patients with complete CoW and in those with incomplete CoW separately. Conditional regression analysis was used to examine the independent effect of the CoW on WMHs. All tests were two-tailed, and P < 0.05 was considered statistically significant. All analyses were performed with SPSS 24.0 software.

Results

WMHs score in patients with complete and incomplete CoW

We included 115 patients with ICA stenosis ≥70%, 60 with complete CoW (52.17%), and 55 with incomplete CoW (47.83%). Compared with the patients with complete CoW, the patients with incomplete CoW were older, had a higher incidence of stroke history and perivascular space, and a higher score of deep WMHs, periventricular WMHs, and lacunar infarction (Table 1). Logistic regression analysis showed that the patients with incomplete CoW had a higher score of deep WMHs (OR = 1.82, 95% CI 1.08–3.06, P = 0.023) and periventricular WMHs (OR = 4.53, 95% CI 2.09–9.81, P = 0.000) in patients with ICA stenosis ≥ 70% (Table 2).

Table 1 Clinical and imaging data on patients with complete and incomplete CoW
Full size table
Table 2 The results of logistic regression analysis in patients with incomplete versus complete CoW
Full size table

WMHs score in all included patients: ipsilateral and contralateral to stenosed ICA

In the included 115 patients of ICA stenosis, the score of the deep and periventricular WMHs was higher in the side ipsilateral to stenosed ICA than that in the contralateral side (Table 3). Regression analysis showed that the score of deep WMHs (OR = 3.96, 95% CI 1.61–9.75, P = 0.003) and periventricular WMHs (OR = 3.37, 95% CI 1.48–7.67, P = 0.004) was higher in the side ipsilateral to stenosed ICA than that in the contralateral side in all patients with ICA stenosis ≥ 70% (Table 4).

Table 3 The comparison of the score of WMHs between the two sides, ipsilateral, and contralateral to stenosed ICA
Full size table
Table 4 The results of conditional regression analysis in the ipsilateral side versus contralateral to stenosed ICA
Full size table

WMHs score in the patients with complete CoW: ipsilateral and contralateral to stenosed ICA

In the 60 patients with complete CoW, the score of periventricular WMHs was higher in the side ipsilateral to stenosed ICA than that in the contralateral side, but there was no significant difference in deep WMHs between the two sides. There was no difference in the score of lacunar infarction and the incidence of perivascular space (Table 3). Regression analysis showed that there was no difference in the score of deep (OR = 2.10, 95% CI 0.37–11.91, P = 0.401) and periventricular WMHs (OR = 2.83, 95% CI 0.99–8.05, P = 0.051) between the two sides ipsilateral and contralateral to stenosed ICA in the patients with ICA stenosis ≥ 70% with complete CoW (Table 4).

WMHs score in the patients with incomplete CoW: ipsilateral and contralateral to stenosed ICA

In the 55 patients with incomplete CoW, the score of deep and periventricular WMHs was higher in the side ipsilateral to stenosed ICA than those in the contralateral side (Table 3). Regression analysis showed that the score of deep WMHs (OR = 4.14, 95% CI 1.33–12.93, P = 0.014) and periventricular WMHs (OR = 5.46, 95% CI 1.16–25.62, P = 0.032) was higher in the side ipsilateral to stenosed ICA than those in the contralateral side in the patients with ICA stenosis ≥ 70% with incomplete CoW (Table 4).

Discussion

In our study, the incidence of complete CoW was 52.17%, similar to the previous reports [17, 18]. In the univariate analysis, the mean age of the patients with incomplete CoW was older than that of complete CoW, which was also observed by prior experts [19]. The incidence of stroke history and the occurrence of lacunar infarction were higher in patients with incomplete CoW than in those with complete CoW, which was consistent with the results of the previous studies reporting that effective collateral circulations reduced the risk of transient ischemic attack and stroke [20], and incompleteness of CoW significantly affected the occurrence of lacunar infarction [21].

In the multivariate analysis, compared with the patients with complete CoW, the patients with incomplete CoW had a higher score of deep and periventricular WMHs, which suggested that complete CoW protected the brain from suffering the change of WMHs in the patients with severe ICA stenosis. This was consistent with the prior results [22, 23]. Another study showed that the WMHs volume and number of lesions increased when two or more segments of the CoW were missing in subjects treated with carotid endarterectomy [24]. It was also demonstrated that different variant models of the CoW have different collateral capacities in the situation of unilateral ICA stenosis [25]. The study by Ryan et al. suggested that incomplete CoW correlates with the incident of white matter disease, but the ICA stenosis of the participants was not assessed [26]. Nevertheless, our results were not consistent with those by Li et al., where the CoW and severe carotid atherosclerosis may not be related to white matter lesions. This might be caused by the different methods of rating WMHs, their exclusion of new and old infarcts which are associated with the CoW and severe carotid atherosclerosis and fewer participants, only 59 participants with unilateral stenosis [27].

Further, we compared the two sides, contralateral and ipsilateral sides to stenosed ICA. In all patients with ICA stenosis, the score of deep and periventricular WMHs was higher in the side ipsilateral to stenosis than that in the contralateral side, this was also true for patients with incomplete CoW. But in the patients with complete CoW, there was no significant difference in the score of deep and periventricular WMHs between the ipsilateral and contralateral sides to the stenosis. This further suggested that complete CoW protected the brain from suffering the change of WMHs in the patients with severe ICA stenosis. Completeness of CoW is regarded as the major source of collateral blood flow and is crucial to maintaining adequate perfusion pressure in patients with severe ICA stenosis. These were supported by the finding that in patients with asymptomatic ICA stenosis, the prevalence of collateral flow via the CoW was significantly increased compared with the symptomatic patients and control subjects. Patients with asymptomatic ICA stenosis demonstrated the largest mean diameter of the arteries of the CoW compared with patients with symptomatic ICA stenosis and control subjects. These studies suggested the importance of adequate hemodynamic compensation via the CoW in patients with ICA stenosis [28, 29]. The study by Men et al. suggested a higher prevalence of leukoaraiosis in southern Han Chinese patients with branch atheromatous disease which is a kind of atherosclerosis [30]. The proposed mechanisms that underlie the pathophysiology of WMHs in chronic atherosclerotic ICA stenosis may include artery-to-artery embolism, branch artery disease, and hypoperfusion [31].

There were some limitations to our study. We only compared CoW completeness versus incompleteness without differentiating the anterior or the posterior part of the CoW. Moreover, we only investigated the image change of WMHs without observing the changes of hemodynamics. Finally, the sample was not large and a larger sample is required for further investigation.

In conclusion, the score of WMHs was higher in patients with incomplete CoW than that in complete CoW in the setting of severe ICA stenosis. The score of WMHs was higher in the side ipsilateral to stenosed ICA than that in the contralateral side in the patients with incomplete CoW, while there was no significant difference in those with complete CoW. This suggested that complete CoW could protect the brain from suffering WMHs even in patients with severe ICA stenosis. Our results suggested that when the ICA stenosis related to brain injury is considered, the maximal degree of ICA stenosis is not the only contributor. CoW supplied collateral blood flow ought to be considered even in severe ICA stenosis. This data reminds us that the completeness of CoW might be included in the risk stratification of ICA stenosis.