Basilar Artery Plaque Distribution, Pontine Infarction and Vertebrobasilar Artery Geometry: A Magnetic Resonance Imaging Study

Background: Basilar artery (BA) atherosclerosis is a common cause of posterior-circulation ischemic stroke. In this study, we investigate the relationship between BA plaque distribution, pontine infarction (PI) and vertebrobasilar artery (VBA) geometries. Methods: 86 patients with BA plaque were enrolled in this study, the VBA geometry was classied into four congurations: walking, tuning fork, lambda, and no conuence. The AP-Mid-BA, Lateral-Mid-BA, and VA-BA angles were measured on three-dimensional time-of-ight magnetic resonance angiography. Patients underwent high-resolution magnetic resonance imaging to evaluate the BA plaque distribution (either anterior, posterior, or lateral wall). Acute and subacute PIs were identied by T2 weighted imaging-uid-attenuated inversion recovery and diffusion-weighted imaging. Results: BA plaques in patients with PI were more frequently located at the posterior wall (50.00%) than at the anterior (10.00%) and lateral (37.50%) walls (P =0.028). Compared with patients without pontine infarction, patients with pontine infarction were more likely to have plaque distributed at the posterior wall (P = 0.009). In the tuning fork group, BA plaques were evenly distributed. BA plaques were more frequently located at the lateral wall than at the anterior and posterior walls in patients with walking, lambda, and no conuence geometry (all P ≤ 0.05). The AP-Mid-BA angle in patients with a tuning fork conguration (14.95 0 ±11.66 0 ) was lower than that in patients with other vascular geometries (P=0.001); there was no signicant difference in the VA-BA angle and lateral-mid-BA angle among the four VBA geometries (P >0.05). Conclusion: VBA conguration strongly inuences BA plaque distribution and BA plaque distribution was associated with PI.

in patients with a tuning fork con guration (14.95 0 ±11.66 0 ) was lower than that in patients with other vascular geometries (P=0.001); there was no signi cant difference in the VA-BA angle and lateral-mid-BA angle among the four VBA geometries (P >0.05).
Conclusion: VBA con guration strongly in uences BA plaque distribution and BA plaque distribution was associated with PI.

Background
Posterior circulation stroke is de ned by infarction occurring within the vascular territory supplied by the vertebrobasilar artery (VBA), and accounts for about 20-30% of all ischemic strokes [1,2]. Klein et al. [3] used HR-MRI to observe 41 patients with pontine infarction (PI), and found that more than 70% of patients had basilar artery (BA) plaque. BA atherosclerosis is a common cause of posterior circulation ischemic strokes, whose underlying mechanisms include artery-to-artery embolism, in situ thromboocclusion, and hemodynamic impairment. In addition, plaques located near the branch artery ori ces may also induce pontine infarction [4]. Previous studies have demonstrated that coronary and middle cerebral artery atherosclerosis plaques tend to be distributed at positions opposite to the ori ces of the perforating artery [5,6]. The BA forms the central core of the posterior circulation and is a rich source of perforating arteries; therefore, it is meaningful to investigate the relationship between BA plaque distribution and pontine infarction.
Usually, the vertebrobasilar system is formed by the basilar and bilateral vertebral arteries. As the diameters of the left and right VAs and their anatomical course are different, a study by Yu J et al [7] classi ed the vertebrobasilar system into four geometric con gurations: Walking, Tuning Fork, Dominant-Lambda, and Hypoplasia-Lambda. Furthermore, they [7] demonstrated that the geometric con gurations of the vertebrobasilar artery strongly in uence BA plaque locations. Wake-buck et al. [8] reported a relationship between vertebrobasilar geometries and differences in hemodynamic distribution. In the study by Zheng J.M. et al [9], VBA geometry was qualitatively classi ed into four basic geometric con gurations: Walking, Tuning Fork, Lambda, and No Con uence and found that the Walking, Lambda, and No Confuence geometry are associated with the presence of BA plaque. In this study, we want to investigate the relationship between BA plaque distribution, pontine infarction and vertebrobasilar artery geometries (Walking, Tuning Fork, Lambda, and No Con uence).

Patients
This study enrolled eighty-seven consecutive patients who presented with posterior circulation ischemic stroke to the Department of Neurology of our hospital from July 2017 to June 2018. Patients with the following characteristics were enrolled: (1) had BA plaque; and (2) displayed image quality su cient for analysis. Patients were excluded if they had any of the following conditions: (1) non-atherosclerotic vasculopathy, such as dissection, arteritis, or Moya-Moya disease; (2) contraindications to MR imaging; (3) poor image quality due to motion artifact; or (4) their vascular geometry could not be classi ed. Based on these criteria, one patient, whose vascular geometry could not be classi ed, was excluded. Ultimately, 86 patients with BA plaque were included in our study.
This study was conducted following the Declaration of Helsinki and was approved by the Ethics Committee of Union Medical College Hospital A liated to Fujian Medical University. All patients signed an informed consent form to participate in the research.

Image analysis
Details of the high -resolution MRI protocol were described elsewhere [9]. On the axial CUBE images, if there was eccentric wall thickening, whereas the thinnest part was estimated to be < 50% of the thickest point by visual inspection, then, a plaque was identi ed [6]. All cross-sections with eccentric plaque were classi ed based on their plaque orientation being centered on the anterior, posterior, or lateral (left or right) sides of the vessel (Fig. 1) [6, 10]. Each cross-section was grouped into one of four quadrants. In cases where the plaque was located at more than one quadrant, the quadrant with the maximal plaque thickness was chosen [6, 10], and we calculated the percentage of plaque distribution at the anterior, posterior, and lateral sides of the basilar wall. Acute and subacute pontine ischemic lesions were identi ed by T2WI-FLAIR and DWI.
The diameter of the VA was measured using TOF-MRA. We measured three consecutive points, 3 mm apart, starting from the vertebrobasilar junction (both VAs and the BA), then, the diameter of each vessel was calculated as the average of the three measurements values [7,11]. The dominant vertebral artery was de ned as the vertebral artery with the widest diameter (difference in diameter ≥ 0.3 mm) [7,11]. Based on the TOF-MRA images, the vertebrobasilar artery geometry was group into four con gurations: walking, tuning fork, lambda, and no con uence. The characteristics of each geometric con guration was de ned elsewhere [9].
The AP-Mid-BA angle was measured using the anteroposterior (AP) view of 3D-reconstructed TOF-MRA, and the VA-BA angle and Lateral-Mid-BA angle were measured using the lateral view of the 3Dreconstructed TOF-MRA (Fig. 2). Imaginary lines were drawn from the mid-BA to the vertebrobasilar junction and the top of the BA in the AP and the lateral views, respectively ( Fig. 2A and 2B) [12]. The maximum angles between these two imaginary lines were considered to be the AP-Mid-BA angle (AP view) and the Lateral-Mid-BA angle (lateral view). Using the same method, imaginary lines were drawn from the vertebrobasilar junction to the BA and the dominant VA; the angle between the two lines was considered the VA-BA angle ( Fig. 2C) [12].

Statistical analysis
Statistical analysis was performed using IBM SPSS statistics software (version 19.0, IBM Corp., Armonk, NY, USA). The Shapiro-Wilk test of normality was used to investigate the distribution of data. Quantitative data are expressed as mean ± SD, and qualitative data are expressed as percentages. Comparisons of BA plaque distribution incidence in the anterior, posterior, and lateral sides of the BA wall were performed using Friedman's M test. Comparisons of plaque distribution incidence between patients with and without pontine infarction were performed using the Mann-Whitney U test, and data comparisons of the four geometric con gurations were performed by variance analysis or the Kruskal-Wallis test. P < 0.05 was considered statistically signi cant.

Results
Of the 86 patients, there were 13 patients (median age, 67 years) with pontine infarction, and 73 patients without pontine infarction (median age, 73 years). Comparisons of clinical risk factors between patients with and without pontine infarction are summarized in Table 1.

Correlation between BA plaque distribution and pontine infarction
In 86 patients, plaques were identi ed in 716 slices. Only four patients had a one-slice plaque; most BA plaques involved multiple slices on high resolution magnetic resonance imaging (HR-MRI). The average length of BA atherosclerosis plaque was 8.00 mm (4.50 mm-11.00 mm) in 73 patients without pontine infarction and 9.00 mm (4.00 mm-11.50 mm) in 13 patients with pontine infarction (P = 0.570).
BA plaques were more frequently located at the lateral walls than at the anterior and posterior walls in all 86 patients as well as in the 73 patients without pontine infarction. However, in the pontine infarction group, plaques were more frequently located at the posterior wall (50.00%) than at the anterior (10.00%) and lateral walls (37.50%, P = 0.028). Compared with patients without pontine infarction, patients with pontine infarction were more likely to have plaque distributed at the posterior wall (P = 0.009). The results of the basilar artery plaque distribution are shown in Table 2.  Correlation between BA plaque distribution and VBA geometry The percentage of plaque distribution at the anterior (13.64%), posterior (37.50%), and lateral walls (37.50%, P = 0.264) was similar in the 11 patients with Tuning Fork geometry. However, in the Walking, Lambda, and No Con uence groups, plaques were more frequently located at the lateral wall than at the anterior and posterior walls (all P < 0.05, Table 3).  The AP-Mid-BA angle in patients with the walking con guration (36.13 0 ±18.27 0 ), Lambda con guration (24.84 0 ±13.02 0 ), and No Con uence con guration (24.17 0 ±10.54 0 ) were higher than that in patients with Tuning Fork con guration (14.95 0 ±11.66 0 ) (P = 0.001); however, there were no signi cant differences in the VA-BA and Lateral-Mid-BA angles among the four vertebrobasilar geometries (P > 0.05, Table 4).

Discussion
In this study, it was observed that BA plaque in patients with pontine infarction was predominantly distributed in the posterior wall. Comparatively, BA plaques in patients without pontine infarction were more frequently distributed in the lateral wall (50.00%) than in the anterior (20.00%) and posterior (22.22%) walls (P ≤ 0.001). Furthermore, we found that vertebrobasilar geometric con gurations in uence plaque distribution.
In a study by Yu et al. [10], BA plaques in the dorsal (posterior) and lateral walls were found to be associated with posterior ischemic stroke, where the branches of BA originate, and that BA plaques in asymptomatic patients were more likely located at the ventral (anterior) wall. In patients with pontine infarction, we found that BA plaque was predominantly distributed in the posterior wall (50.00%); this nding is in line with previous study (9). However, in our study, the percentage of BA plaque distribution in the lateral wall (50.00%) was higher than in the anterior (20.00%) and posterior (22.22%) walls in patients without pontine infarction. BA plaque in the inferior lateral wall or in the upper lateral wall all were de ned as BA plaque of the lateral wall, however, as we all know, the penetrating artery of BA arises from the posterior and inferior lateral wall, thus, BA plaque in the inferior lateral wall more easily induced pontine infraction than in the upper lateral wall. In our study, BA plaque only in the upper lateral wall accounts for about 25% of BA plaque in the lateral wall, which may be explain the result that patients without pontine infarction have high incidence of BA plaque distribution in the lateral wall.
A previous study [7] observed that geometric con gurations strongly in uence BA plaque distribution. In the present study, BA plaques were evenly distributed in the tuning fork geometry, which was consistent with the previous study (7). In the tuning fork geometry, the BA ow is roughly parallel and the velocity pro le peak is rather central in the basilar artery, and the hemodynamic distribution is simple. However, in the vertebrobasilar arteries with walking, lambda, and no con uence con gurations, BA plaques were more frequently occurred at the lateral wall, from where the penetrating artery arose (all P ≤ 0.05). A study by Kim et al. [12] demonstrated that greater mid-BA angulation may enhance lateral plaque formation, and greater BA-VA angulation may enhance posterior plaque formation. In our research results, the AP-Mid-BA angle in walking (36.13 0 ±18.27 0 ), Lambda (24.84 0 ±13.02 0 ), and No Con uence (24.17 0 ±10.54 0 ) geometries were greater than in the Tuning Fork geometry (14.95 0 ±11.66 0 , P = 0.001), and the greater AP-Mid-BA in the Walking, Lambda, and No Con uence con gurations may be induced the high percentage of BA plaque in the lateral wall among the three geometries. In addition, there were no signi cant differences in the VA-BA and Lateral-Mid-BA angles among the four vertebrobasilar geometries, which may be the reason that BA plaque distribution in the posterior wall among the four geometries have no signi cant differences.
Our study has many clinical implications. First, a previous study [10] reported that about two-thirds of BA plaques are located at the lateral and dorsal walls, where the penetrating artery arises, which have a high risk of pontine infarction, this suggests that BA plaque distribution may help us to assess the likelihood of ischemic stroke in the posterior circulation and reduce the risk of complications during stenting. We found about 50% of the total BA plaque located at the posterior wall where the penetrating artery arose in patients with pontine infarction, this is consistent with the previous study (9), which suggested that BA plaque in the posterior wall has a high risk of developing pontine infarction. Besides, we found that the incidence of BA plaque in the lateral wall was higher than in the anterior and posterior walls in patients without pontine infarction, and suggested that BA plaques in the upper lateral wall are unlikely to induce occlusion of the penetrating artery. Second, we demonstrated that geometric con gurations and BA curvature angulation may in uence the BA plaque distribution.
There are several limitations in this study. First, patients with vertebral artery atherosclerosis were not excluded from the study, thus, the hemodynamic effect of VA stenosis to BA atherosclerosis formation is unclear. Second, the sample size was small, only a single center was included. Finally, the hemodynamic distribution in the four geometries was not evaluated. Further studies measuring the hemodynamic distribution to elucidate the mechanism by which the geometry in uences plaque distribution is necessary.

Conclusions
In conclusion, BA plaque in the posterior wall may be related to pontine infarction, besides, there are relationship between vertebrobasilar geometry and BA plaque distribution. Among the Walking, Lambda, and No Con uence geometries, BA plaques were often located at the lateral wall; while in the tuning fork geometry, the BA plaques were evenly distributed. An alignment grid to demonstrate how each cross-section is divided into four quadrants. Figure 2