Anatomical Variations and Dimension of the Intracranial Vertebral Artery: Evaluation With Multidetector Computed Tomography Angiography

The intracranial segment of the vertebral artery (VA) is the unique part of the artery where the two VAs join to form a single vascular channel, viz. the basilar artery. In addition to this typical description, anatomical variations have been described; the presence of anatomical variation has been associated with some pathological processes, neurological complications, and the risk of vascular diseases in the posterior circulatory territory. We evaluated the typical anatomical features and variations of the VA4 component of the VA in a South African population to provide useful data on the prevalence of variation and morphometry of the distal VA. The study is an observational, retrospective chart review of 554 consecutive South African patients (Black, Indian, and White) who had been examined with multidetector computed tomography angiography (MDCTA) from January 2009 to September 2019. We observed various anatomical variations in the VA4 segment of the VA. We report the incidence of VA hypoplasia, hypoplastic terminal VA, and atresia. Fenestration and duplicate posterior inferior cerebellar artery (PICA) origin were also observed. The left intracranial VA was signicantly larger than the right. Our study shows that anatomical variation of the intracranial VA is common in the population studied, with a total prevalence of 36.5%. Imaging of the entire course of the VA from the origin to the point of convergence to form the basilar artery may be necessary to decide a treatment strategy for interventions in the vicinity of the VA.


Introduction
The vertebral artery (VA) emanates from the superior-posterior part of the subclavian artery and proceeds through the foramen transversarium of the cervical vertebra. The left and right VA penetrates the dura mater to enter the intracranial space through the foramen magnum, where they converge to form the basilar trunk at the pontomedullary junction 1 . Anatomically, the VA is divided into four segments. The rst three segments (VA1, VA2, VA3) are the extracranial segments, extending from the origin to where they penetrate the dura mater. The fourth segment of the VA (VA4) is intracranial, extending from the foramen magnum to the point where the left and right VA anastomose to form the basilar trunk 2 . The geometry of the basilar trunk depends on the pattern of the bilateral VAs. When there is an asymmetry of bilateral VAs or other morphological variations, the basilar trunk sometimes bends away from the midline 3 . Previously reported anatomical variations of the intracranial VAs include VA terminating as posterior inferior cerebellar artery (PICA), known as atresia, fenestration, asymmetry, and hypoplasia. Anatomical variations play a signi cant role in the clinical sequelae of an iatrogenic VA injury, which can vary widely 4 . For instance, damage to the dominant VA when the contralateral VA terminates as PICA can result in devastating complications since the dominant VA solely forms the basilar artery.
The presence of variation has been associated with some pathological processes, neurological complications, and the risk of vascular diseases in the posterior circulatory territory. For instance, atresia and hypoplasia have been associated with hypoperfusion of brain tissues and hemodynamic insu ciency, which may predispose to transient ischaemic attacks or acute brainstem ischaemic stroke 5,6 .
Reports on the prevalence of anatomical variations of the intracranial VA are scarce, with few previous reports in the Western population (Caucasians) 1 , the Asian population 5 , and the Turkish population 7 .
Previous studies from the African continent used post mortem and cadaveric histological samples to report on average diameter and incidence of VA hypoplasia 8,9,10 . Advances in modern imaging technology that led to the establishment of multidetector computed tomography angiography (MDCTA) have made endovascular procedures popular. These procedures require a detailed understanding of typical anatomy and the extent of anatomical variations of the VAs. Consequently, a report on the prevalence of possible variant anatomy will help in the interpretation of radiographs, prevention of iatrogenic injuries, and contribute to the advancement of non-invasive surgical intervention.
In the present study, we assessed the typical anatomical features and variations of the V4 segment of the VA using MDCTA. We aimed to determine the dimensional characteristics and prevalence of anatomical variations of the intracranial VA in a South African population. Due to the multiracial composition of the South African population, in addition to the overall incidence of variation, we also report variations based on three racial groups: Black, Indian, and White South African. It is necessary to have correct and detailed information about the typical anatomy and prevalence of anatomical variations. Such information is essential before neurosurgery, endovascular and non-invasive procedures. The detailed information from this study will be useful in neurosurgery, anatomy, endovascular, and non-invasive procedures.

Study population
This study is a retrospective observational review of 554 MDCTA images of South African patients. The patients underwent MDCTA for various reasons between January 2009 and September 2019. Images were obtained from the database of Lenmed Ethekwini Hospital and Heart Center, Durban, South Africa.
The Biomedical Research Ethics Committee of the University of KwaZulu-Natal approved the study (Ethical No: BE 148/19) and waived the need for informed consent as this study utilized retrospective chart reviews. There was no patient contact, and no patient details were released from images. All methods were carried out in accordance with relevant guidelines and regulations. The angiographies were from 307 males (55.4%) and 247 females (44.6%). The average age of the patients is reported as median (IQR): 62 (23) (range: 10-99) years; 61 (23) for male patients and 62 (25) for female patients. The race was de ned according to the guidelines outlined in the modern systems of racial classi cation in the Republic of South Africa 11 . The criteria used in the scheme of racial classi cation include skin colour and ancestry. The South African population is divided into four main racial groups: White, Black, Indian, and Coloured. Three population groups were included in the present study: Black 91 (16.4%), Indian 176 (31.8%), and White 287 (51.8%). According to the modern system of classi cation, a White individual was de ned as a person of European descent. A Black individual was de ned as a person having origins in any of indigenous Africa or Native group. An Indian individual was de ned as a person of Asian descent 11 .
The MDCTA images were analyzed using picture archiving communication system (PACS) tools. The images were examined for vascular variations by a neurosurgeon, a neuroradiologist, and an anatomist using the coronal and sagittal view and the 3D reconstructed images. Exclusion criteria included CTA scans that showed no clarity of the VA's course, and scans with motion artefacts or poor-quality imaging.

CT Angiography Protocol and Imaging Reconstruction
The imaging examination was performed on a 64-detector row computed tomography (CT) scanner

Analysis of Morphological Variation and Dimensions of the V4 Segment
Each MDCTA image was examined for the presence of morphological variations. The following parameters were measured on a coronal view of the MDCTA ( Fig. 1): 1) the diameters were measured along the course of the VA at a distance of 11 mm cranial to the entrance of the VA into the foramen magnum, 2) the length of the VAs was measured from the foramen magnum to the point of union with the contralateral VA, and 3) the angle between the bilateral VA at the vertebrobasilar junction. We were unable to appropriately quantify the frequency of the PICA and the spinal arteries because visualization of branches (such as the PICA and spinal arteries) is usually beyond the limits of the MDCTA. A diameter of ≤ 2 mm was described as hypoplasia; we classi ed the VA as dominant if the diameter was larger than that of the contralateral side by any size difference according to the method provided by Ergun and co-authors 12 . When the bilateral VAs had a similar diameter, we referred to them as "equal" or "codominant." Results were analyzed separately for the left and right sides.

Statistical Analysis
All data were analyzed using SPSS version 27 (SPSS Inc., Chicago, IL, USA). Categorical variables were analyzed using the chi-square test. A Kolmogorov-Smirnov test was used to assess the normal distribution of continuous data. Because the distribution of the data was not normal, nonparametric tests were used. The Kruskal-Wallis test followed by the Wilcoxon Signed-Rank test was used to detect signi cant differences in the obtained values. The interclass correlation coe cient was used to examine the reliability of measurements. All tests were performed at 95% con dence with a p-value of < 0.05.

Results
Continuous and categorical data are presented as the median and interquartile range (IQR) and percentage (N). The interclass correlation coe cient for intra-observer reliability testing was 92% for V4 length, 93% for diameter, and 96% for the angle at the VBJ. For inter-observer reliability testing, the intraclass correlation was 85% for V4 length and diameter; 87% for the angle at the VBJ.

Variation in Morphology
We observed the following variations of the intracranial segment: (1) Table 3.

Page 6/14
The angle at the Vertebrobasilar Junction.

Discussion
Our study shows that MDCTA made it possible to evaluate anatomical variations of the intracranial VA.
We found that variation is common in the population studied, with a total prevalence of 36.5%. The most frequently observed is VA hypoplasia. The incidence of hypoplasia in the present study (8.3% on the right and 6.5% on the left) is similar to the report of Ergun et al. 12 . These authors de ned VA hypoplasia using diameter criteria of ≤ 2 mm and reported an incidence of 7.1% on the right and 9.4% on the left among 254 patients in their angiographic series 12 . By contrast, Songur and co-authors in their autopsy study reported a relatively high incidence of 20.2% on the right, 14.4% on the left, and 4.3% bilaterally using a similar de nition of VA hypoplasia 7 . Sometimes it is challenging to compare data from different populations and research groups due to the differences in study modalities, distribution of data, and average diameter. According to a recent report, an individual's VA diameter may depend on anthropometric parameters such as height 13 . All these factors may contribute to the wide range of differences reported in the literature. VA hypoplasia is a congenital anatomical variation that has been previously described with a cut-off diameter between 2.0 mm and 3.0 mm 14 .
In addition to hypoplasia, we also noticed that the hypoplastic terminal portion of the unilateral intracranial VA is a common anatomical variant of the studied population (Fig. 2b). The VA seems to divide at a spot along its courses to a PICA branch and a tiny branch that joins the contralateral VA. Pekcevik and co-author proposed another terminology for this type of variation; vertebral artery continued as PICA 15 . In our own opinion, this suggested anatomical term can be confused with VA terminating as PICA or VA ending as PICA (atresia). We suggest that this variant anatomy can be simply described as hypoplastic terminal VA.
The reduced diameter of hypoplastic VA has been associated with an increased probability of spontaneous dissection 16 and ipsilateral PICA and lateral medullary infarctions due to suspected atherosclerosis as a result of abnormal hemodynamics 17 . Recently, VA hypoplasia has been associated with an aneurysm of the contralateral dominant VA, most especially at the site of PICA origin 18 .
Knowledge of pathologies associated with VA hypoplasia can provide some clues and help diagnose pathological processes in the posterior circulatory territory.
In our series, the percentage of patients having VA atresia is 6.7% (Fig. 2a). Prevalence of VA atresia has previously been reported as up to 9% 5, 19 . Our results were in accordance with the range of the reported prevalence but most similar to that reported by Liu et al. (6.3%) 5 . Clinically, VA atresia has been previously linked to rotational vertebral artery syndrome (RVAS) 20 and bow hunter's syndrome 2,21 , which may result from compression of this variant vessel.
Another important nding in the present study is the variant origin of atretic VA from the aortic arch ( Fig. 2a). Some authors have previously reported this type of co-existing anatomical variation 19 .
However, the true prevalence has not been established. Out of the 554 patients in our series, 38 left VA directly originated from the aorta arch, and 4 of the 38 VAs were atretic (11%). Knowledge of this variant anatomy is essential when planning for thoracic aortic surgery. Obstruction of this type of VA can increase the risk of posterior circulatory stroke 19 . In addition, VA atresia has been suspected to contribute to ischemic events in the vertebrobasilar system (Liu 2017).
We also observed fenestration at the right intracranial VA in one of the patients (Fig. 3b) and the proximal part of the basilar artery in another patient. Our observation is similar to the report of Dzierzanowski et al., which reported two fenestrations in the Caucasians 1 . Fenestration of the vertebrobasilar artery is a congenital anomaly that involves lumina division of an artery with a single origin into two separate channels that later reunite distally. Embryologically, the VA and the basilar artery develop from different primitive vessels. The VA is formed from the cervical intersegmental arteries, while the basilar artery develops from the longitudinal neural arteries. As a result of these, fenestration at the V4 segment of the VA is due to the absence of obliterations of two intersegmental vessels that fused 22 . Fenestration of the proximal basilar occurs due to partial failure or incomplete fusion of the longitudinal neural arteries and regression of the bridging arteries connecting the longitudinal arteries 23 . Fenestration may predispose to aneurysm around the fenestrated portion of the artery 7,15 , and it has also been previously associated with unexplained subarachnoid hemorrhage 24 . In addition to the associated pathologies, knowledge of this variation is essential in clinical diagnosis as fenestration may be misinterpreted as an aneurysm or a dissection on magnetic resonance imaging 15 .
In the present study, duplicate PICA origin was registered in one of the patients. It is important to note that duplicate PICA origin is different from the duplication of the PICA. In the duplicate origin, the PICA has two separate origins that later converge distally in the course of the artery (Fig. 3a). Whereas in duplication of the PICA, there is no distal arterial convergence 25 : each artery courses separately.
Duplicate PICA origin is a rare congenital morphological variation of the PICA with a prevalence of roughly 1.45% previously reported in the Western population (White and Asian) 25 . Clinically, duplicate PICA origin has been previously reported to highly predispose to intracranial aneurysm formation with an associated incidence between 50% and 71% 25,26 . Embryologically, Lesley and co-authors hypothesized that duplicate PICA origin might be a manifestation of underlying de cient vascular developmental disorganization, which may upraise the tendency toward formation of an intracranial aneurysm 25 .
Considering the unique embryogenesis, adequate perfusion of the regions supplied by the PICA may rely on ow from both origins 26 . Since duplicate PICA origin is an uncommon variation with few previous reports, it should not be overlooked when evaluating the diagnosis and surgical intervention images. Because visualization of branches of the intracranial VA (PICA and spinal arteries) is usually beyond the limits of MDCTA, the frequency of PICA on the left, right, and bilateral PICA in the present study is low (15.7%, 13.9%, and 14.8%, respectively). We also observed bilateral and unilateral double PICA in 5 patients. Therefore, we cannot appropriately quantify the frequency of the PICA in all the VAs.  7 . On the contrary, other authors reported a larger mean angle (85.45 ± 10.76°) 1 . The disparity may have resulted from the con uence of the bilateral VA, which can either be a sharp or blunt edge depending on the pattern and frequency of asymmetry. It is essential to consider the geometry of the vertebrobasilar junction while planning for surgical interventions in this region. This region is of particular interest to neurosurgeons and radiologists due to various interventional neuroradiological procedures conduct in the area to treat vascular diseases such as arterial dissections, aneurysms, arteriovenous malformations, dural stula, or repair of an occlusive disease 29 . In atresia, the VA did not fuse with the contralateral VA but terminated as PICA. The contralateral VA solely proceeds to form the basilar artery. In the case of hypoplastic terminal VA, the contralateral VA predominates the basilar artery with little contribution from the tapering end of the hypoplastic terminal VA. In addition to asymmetry, these two conditions can also cause the basilar artery to bend from the midline (Fig. 2b, also known as bending basilar) 30 . Deviation and prominence of a vessel, such as bending basilar due to dominance of one of the VAs, may cause compression of cranial nerves 15 .

Conclusion
Our study shows that anatomical variation of the intracranial VA is common in the population studied, with a total prevalence of 36.5%. Hypoplasia and hypoplastic terminal VA being the most frequent.
Imaging of the entire course of the VA from the origin to the point of convergence to form the basilar artery may be necessary to select a treatment strategy for interventions in the vicinity of the VA. Understanding the patterns of anatomical variations of the VAs will contribute signi cantly to the interpretation of ischemic areas and diagnosis of various diseases in the posterior circulatory territory.