Risk stratification of vertebral artery vulnerability during surgery for congenital atlanto-axial dislocation with or without an occipitalized atlas
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.158218
Source of Support: None, Conflict of Interest: None
Context: Variability in dimensions and course of vertebral artery (VA) makes it vulnerable to injury during surgery for congenital atlanto-axial dislocation (AAD) with or without an occipitalized atlas.
Keywords: Anatomy; atlanto-axial dislocation; craniovertebral junction; occipitalized atlas; radiology; vertebral artery
Vertebral artery (VA) injury may occur in approximately 4.1% patients during surgery at the craniovertebral junction (CVJ). , This study, conducted in patients with atlanto-axial dislocation (AAD) with or without an occipitalized Atlas More Details, utilizes dynamic, multi-planar reconstruction images of three-dimensional (3D) computed tomographic angiography (CTA), in an attempt to precisely define the course of the third and fourth parts of VA at the CVJ as well after it had entered the intracranial cavity. It also proposes a classification and a prognostic score based on the anatomical variations of VA that determines its relative vulnerability during transoral surgery and posterior stabilization procedures.
A total of 104 patients (65 patients with AAD and 39 without AAD [considered as a control]) were prospectively included in the study conducted from July 2011 to October 2013. Sixty-three of them were male and forty-one female patients, ranging in age from 3 to 76 years. AAD was diagnosed based on a minimal atlanto-dental interval (ADI) >3 mm in adults and 4 mm in those aged ≤9 years on lateral dynamic (in flexion and extension of the neck) radiographs and sagittal reconstructed CT scans of the CVJ.  A magnetic resonance imaging of the CVJ was also performed to assess for co-existing soft tissue abnormalities such as Chiari I malformation or syringomyelia. Patients with inflammatory, tuberculous or traumatic AAD were excluded from the study. The control group constituted of patients presenting with spontaneous subarachnoid hemorrhage in whom a 3D CTA had been performed for detecting the presence of an aneurysm and the CT slices had been extended to include the upper cervical spine. The CTA was performed in all 104 patients using a 64-slice CT scanner with following scanning parameters: 120 kV, 300 mA, 0.75 s gantry rotation time, 0.625 mm slice thickness, pitch of 0.926 and field of view of 250 mm. 50-80 ml of nonionic contrast (Iohexol) was used at a flow rate of 3-5 ml/s. Three dimensional (3D) images were reconstructed and simultaneously analyzed by a radiologist and a neurosurgeon.
The ADI, effective canal diameter, reducibility of AAD and associated bony anomalies such as occipitalization of atlas, Klippel Feil anomaly, facet joint asymmetry, odontoid process anomalies, rotational deformity, and scoliosis were assessed. Basilar invagination was diagnosed on the basis of various craniometric lines including Wackenheim's clival canal line, McGregor's line, McRae's line, and Chamberlain's line [Table 1]. ,
The course of the third and fourth parts of bilateral vertebral arteries was studied using 3D CTA and any variation in size, course, or asymmetry of the VA;  the type of VA groove (VAG) on the lateral one-third of the posterior arch of atlas and the isthmus of axis; , and the point of entry of VA into the foramen transversarium (FT) of atlas (or into the congenital foramen representing the C1 FT in the patients with an occipitalized atlas) were evaluated [Figure 1] and [Figure 2] and [Table 2]. 
Classifying variations in the vertebral artery anatomy
The variations of VA were classified as: Type 1 (based on size): (a) Bilateral symmetrical [Figure 3]a; or, (b) bilateral asymmetrical with unilateral dominance but without hypoplasia/aplasia [Figure 3]b, (c) with unilateral hypoplasia [Figure 3]c or, (d) aplasia [Figure 3]d. Type 2 (based on variations in its course): (a) Having a normal course [Figure 4]; (b) having a cranial entry through a congenital foramen between the occipital bone and occipitalized atlas [Figure 5]a; and when present as (c) a persistent first intersegmental artery [Figure 5]b and c, (d) a fenestrated VA [Figure 5]d or, (e) a low-lying posterior inferior cerebellar artery (PICA) emerging from the VA below the foramen magnum and crossing the C1-2 joint [Figure 6]a. ,, Type 3 (based on anomalous medial deviation of its third segment) showing (a) an absence; or, (b) presence of looping of V3 segment towards the midline [Figure 6]b and c. Type 4 [based on the type of isthmus of the axis above the loop of VA in the C-2 FT; determination of "safe zone" was done by measuring the parameter 'a' that is the vertical distance from the apex of VAG to the upper facet joint surface and parameter 'e' that is the horizontal distance from the entrance of VAG to the lateral margin of the vertebral canal  [Figure 7]a: (a) When the isthmus was wide [Figure 7]b or, (b) when the isthmus was narrow (less than or equal to 4.5 mm) and the VA was high riding [Figure 7]c; resulting in a situation where transarticular and transpedicular screw placement were extremely risky], (c) narrow but with a low-lying VA related to it or, (d) when the isthmus was wide and associated with a highly placed VA (all three variations in the anatomy of axis, therefore, leaving an adequate amount of isthmus of the axis for transarticular and transpedicular screw placement). Type 5 (based on whether or not rotational deformity or tilt at the occipito-C1-C2 joint was associated with AAD): (a) AAD without co-existing rotation or tilt and, (b) AAD associated with rotational deformity or tilt  [Figure 8]. A score of 1 or 2 was assigned to the existing variation in each category. Thus, when there was an anticipated risk of VA injury during a posterior screw and rod or a transarticular screw placement procedure due to the prevalence of a VA anomaly on CTA, a score of '2' was assigned. The latter condition existed in cases where the VA directly crossed the atlanto-axial facet joint from a posterior aspect, in the presence of an exaggerated medial looping of its third segment, when associated with a narrow isthmus with a high riding VA or, when associated with rotational deformity or tilt that caused deviation of VA towards the midline.
In rest of the situations, the score remained '1' for each category. Thus, in the proposed classification, the minimum score 5 was considered as a "safe zone;" and a score between 6 and 9 was considered as presenting an increased vulnerability of VA during surgery (with a higher score representing a progressive greater risk of VA injury) [Table 2] and [Table 3].
Assessing for an abnormal medial displacement of vertebral artery
Forty-two patients with AAD (30 with an occipitalized and 12 with a nonoccipitalized atlas) and 30 control patients were assessed for the existence of an abnormal medial displacement of VA on axial images of CTA using the 3D multiplanar reconstruction method [Table 4].
M1 represented the shortest distance between the VA on either side and the midline on axial CT images in the plane subtended by joining the posterior lip of the hard palate to the tip of the odontoid process. This was obtained by initially drawing a line (representing the axial plane) from the posterior tip of the hard palate to the tip of odontoid process on midsagittal CT reconstruction image of the CVJ. This corresponded to the approximate trajectory of a transoral decompression procedure [Figure 1]a. The corresponding axial image representing this plane was then utilized to measure the shortest distance between the VA on either side and the tip of odontoid (M1) [Figure 1]b.
M2 represented the shortest distance from the midline of the point of entry of the VA into the FT of atlas. This was measured using the following steps. Initially, on coronal 3D reconstructed CTA images, the point of entry of VA into the FT of atlas (or the congenital foramen representing the C1 FT in patients with an occipitalized atlas) was located. At this level, the corresponding axial CT image was obtained, and the distance between the medial edge of FT of C1/medial edge of VA and the midline was measured [Figure 1]c.
M3 represented the shortest distance from the midline of most medial point on the VA on axial CT images in the horizontal plane subtended along the superior margin of the posterior arch of atlas. The following method was adopted to measure this parameter. Initially, on sagittal 3D reconstructed CTA images, the superior margin of C1 was defined and the corresponding axial image was obtained in that plane [Figure 2]a and b. In patients with an occipitalized atlas, the superior margin of C1 corresponded to the natural groove existing between the fused atlas and the occiput. Then, the distance between the most medial point of VA at the superior margin of the posterior arch of C1 and the midline on either side was measured in the representative axial plane.
Comparing the diameter of vertebral artery with its corresponding atlantal foramen transversarium
A comparison of the VA and its corresponding C1 FT diameter was performed. The obliquity of bilateral facet joint angle was also compared on the two sides.  [Table 5] and [Figure 9]a and b.
Confirming the presence of the "stretched loop" sign of vertebral artery in atlanto-axial dislocation
On 3D sagittal and coronal reconstructed CTA images, the redundant loop of the third segment of VA was also assessed by tracing its course after its emergence from the FT of C2 until its exit through the FT of C1. The straightening of this normally redundant loop of VA (termed "the stretched loop" sign  ) was evaluated in the group of patients presenting with AAD and in the control group [Figure 4]c.
The prospectively collected patients' data representing the type of CVJ anomalies and the variation in the course of VA were expressed as a percentage of total cases [n = 65, [Table 1],[Table 2] and [Table 3]. A comparison of the median values of distance of VA from the midline (M1, M2, and M3) at different levels in the three groups, namely, AAD with occipitalization of atlas, AAD without occipitalization of atlas, and the control group was done by the Kruskal-Wallis test [a P < 0.05 was considered significant; [Table 4]]. The mean diameter of VA and its corresponding FT of C1 and the mean facet joint angle in the groups with AAD with or without occipitalized atlas and in control subjects was compared using Mann-Whitney test.
The relation between the diameter of the ipsilateral VA and its C1 FT was also analyzed utilizing Pearson's correlation coefficient. A P < 0.05 was considered significant in both the analysis [Table 5]. The significant difference between the presence or absence of the "stretched loop" sign of VA in the two groups: (a) Patients with AAD and (b) controls were evaluated using Fisher's exact test. A P <.05 was considered as significant.
There were 32 (49%) patients with a reducible and 23 (35%) patients with an irreducible AAD. Forty-seven (72%) of them had an occipitalized atlas. C1-2 facet joint asymmetry and rotational deformity were present in 13 (20%) patients each with AAD, respectively [Table 1].
The proposed classification and scoring system based on variations in the course of VA are given in [Table 2]. An increased predisposition to VA injury was present in 23 (35.4%) patients, (20 [30%] with a persistent first intersegmental artery; 1 [1.53%] with a fenestrated VA, and 2 [3.07%] with a low-lying PICA) in whom the VA or a low-lying PICA was crossing the C1-2 facet joint from the posterior aspect; in 8 (12%) patients, in whom an anomalous deviation of the third segment of VA toward the midline made the latter susceptible to injury during either the transoral approach or posterior fusion; a high riding VA with a narrow isthmus of C2 seen in 12 (18%) patients on the right side and 11 (16.9%) patients on the left side that considerably increased the complexity of screw placement in the isthmus of the axis vertebra; and a rotational deformity or a tilt at the CVJ (seen in 13 [20%] patients) [Table 2]. Thus, 21 patients had the minimum score of 5 that represented a normal situation; and 14 patients had a score of 6, 20 a score of 7, 9 a score of 8, and 1 a score of 9 that heightened their vulnerability to iatrogenic injury during surgery [Table 3].
Medial deviation of the vertebral artery
There was a significantly higher medial deviation of the VA (M1) placing it dangerously close to the trajectory of the transoral approach in the group of patients with AAD with an occipitalized atlas (P = 0.00 on the right and P = 0.001 on the left side) when compared to patients without an occipitalized atlas and controls. Likewise, the shortest distance from the midline of the most medial point on VA in axial CT images in the horizontal plane subtended along the superior margin of the posterior arch of atlas (M3; the trajectory of posterior fusion procedures) showed a significant decrease on the left side (P = 0.04) in patients with an occipitalized atlas when compared with the other two groups. This distance was also less on the right side that, however, did not attain statistical significance. In patients with an occipitalized atlas, this distance was only 10.30 ± 3.1 mm on the right and 10.60 ± 2.2 mm on the left side while it ranged between 11.7 mm and 12.8 mm in the group with AAD without an occipitalized atlas and between 11.8 and 12.35 in the control group. The shortest distance from the midline of the point of entry of VA into the FT of atlas (both in patients with AAD with an occipitalized atlas as well as those with AAD with a nonoccipitalized atlas and in control subjects; M2), however, remained nearly uniform and relatively wide (approximately 22-24 mm) in all the three groups. This signifies that in the presence of an occipitalized atlas, there is a steep medial angulation of VA after it crosses the FT of atlas making it more prone to injury when compared with the other two groups [Table 4].
Dimensions of vertebral artery, its corresponding C1 foramen transversarium and facet joint angle
There was no significant variation in the diameter of VA and its corresponding C1 FT in the three groups (AAD with or without occipitalized atlas and controls [Table 5]). The relationship between the diameter of the ipsilateral VA and its C1 FT, however, showed a positive correlation (P < 0.05 and r = 0.25). The angle subtended by joining the line passing tangentially through the C1-2 facet joint and the vertical plumb-line (the y-axis) on either side was quite variable in the three groups signifying facet joint asymmetry. There was, however, a significant difference in the angle subtended in patients with AAD without occipitalization of atlas as compared to the control group on the right side [Table 5].
The "stretched loop" sign of vertebral artery in patients with atlanto-axial dislocation
There was a significant straightening of the redundant loop of the third segment of VA in 32 (68%) patients with an occipitalized atlas as compared with 2 (4.2%) with a non occipitalized atlas and none in the control group, respectively (P = 0. 001).
Risk factors and their anatomical considerations
The establishment of an objective scoring system helped in the preoperative assessment of the vulnerability of VA to injury during surgery for CVJ anomalies. It also defined the precise bony and VA variations that may be responsible for a heightened intraoperative risk. Thus, a minimum score of 5 represented a normal course of the VA; a higher score proportionately increased the vulnerability of VA to iatrogenic injury. The score also helped in determining the appropriate surgical technique of posterior stabilization based upon the preoperative location of VA relative to the C1-2 facet joints.
The anatomical variations that helped in risk stratification included the presence of a persistent first intersegmental artery, a fenestrated VA, or a low-lying PICA: ,, An anomalous medial deviation of the third segment of VA; a high-riding VA with a narrow C2 isthmus;  a rotational deformity or a tilt at the occipital-C1-2 level that again resulted in a medial deviation of VA;  and, unilateral dominance of VA.  These anatomical variations coexisting with CVJ anomalies are not coincidental but have a definite embryological basis. Persistent first intersegmental artery and fenestration occur in 0.67 and 1% patients without CVJ anomaly  but in 19 and 36.4% patients, respectively, with bony anomalies like occipitalization of atlas, Klippel Feil syndrome or os odontoideum. , If the embryonic first intersegmental artery remains without the persistence of the primary VA, a persistent first intersegmental artery occurs. If the primary VA fails to regress, it is associated with "fenestration." , In our study, the incidence of persistent first intersegmental artery was high (approximately 30%) and mainly occurred in patients having an AAD with the occipitalized atlas. Tokuda identified three types of VA anomalies: (a) VA turned posteromedially after exiting C2 FT and entered the spinal canal between C1-2, not passing through C1 FT (0.67%); (b) VA duplicated after leaving C2 FT with one fenestration continuing its course normally while the second entering the spinal canal between C1-2 and joining the former at the cranial side of C1 (1%); and (c) VA being normal in course but with a "low-lying" PICA emerging from its extradural portion (between atlas and axis) and entering the spinal canal from the caudal side of C1 (0.67%) or from its proximal intradural portion (at the level of foramen magnum).  During development, at the upper cervical level, the lateral spinal longitudinal artery, lying lateral to the spinal cord and ventral to the posterior cervical nerve roots ends in the intradural portion of VA at C1 and/or joins PICA. A low-lying PICA coursing along the posterior aspect of the C1-2 facet joints may occur if a longitudinal continuity is established between the PICA and the lateral spinal longitudinal artery. 
The anatomical variations of the isthmus of the axis also heighten the vulnerability of VA during posterior C1-2 fusion procedures. ,,, The VA makes an acute lateral bend just under the superior articular facet of the axis in approximately 80% patients. If this bending point is too narrow, too medial, too high or too posterior or the height or width of the isthmus of axis narrow, and the VA is high-riding (occurring in about 16.1% patients), there are greater chances of injury. The risk of VA injury from transarticular screw placement when associated with these anatomical variations was 4.1% per patient. , In the presence of a high riding VA, the recommended screw trajectory is through the most posterior and medial part of the narrow isthmus;  or one can bypass the C2 level, and instead fix the rod at C3 level. ,
Vertebral artery variations in the presence of atlanto-axial dislocation associated with an occipitalized atlas
Occipitalization of atlas occurs in 0.1-0.8% of the normal population.  There is a significantly higher incidence of atlantal assimilation in patients with AAD; ,, in our series, this was evident in 47 (72%) patients with AAD particularly the irreducible variety. Irreducible AAD, occipitalized atlas and associated C2-3 fusion are often accompanied by an abnormal fusion and asymmetry of the lateral joints. This is because the bony occipital-C1-2 architecture, the condyles of the occiput as well as the C1-2 facets have a common embryological basis. The CVJ develops from four occipital and two upper cervical sclerotomes. The fourth occipital sclerotome gives rise to proatlas, a transitional bone between the occipital bone and the atlas. The proatlas is the source of origin of the anterior foramen magnum rim, the upper portion of odontoid and the occipital condyles. The inferior aspect of the first cervical sclerotome gives rise to the lateral masses and C1 arches. The failure of development of the intrasclerotomal fissure or the failure of the primitive proatlas to separate from the occiput results in an assimilated atlas. , The associated abnormalities of C1-2 facet joints result in AAD. ,, It is evident from this study that in patients with AAD with an occipitalized atlas, there is a significantly increased incidence of medial deviation of the third and fourth part of VA that makes it vulnerable to injury during the transoral approach (M1) for irreducible AAD (P = 0.00 on the right and P = 0.001 on the left side) when compared to patients without an occipitalized atlas and controls [Table 4]. Moreover, the shortest distance from the midline of the most medial point on the VA in axial CT images in the horizontal plane subtended along the superior margin of the posterior arch of atlas (M3; the trajectory of posterior stabilization procedures) showed a significant decrease on the left side (P = 0.04) in patients with an occipitalized atlas when compared with the other two groups [Table 4]. This is despite the fact that the distance of the point of entry of VA into the FT of atlas (M2) remained nearly the same (and relatively away from the midline) in patients with AAD with or without an occipitalized atlas and in the normal population. In addition, as the occipitalized atlas is often accompanied by a fusion of the occipito-atlanto-axial joints and their asymmetry, the consequent rotation and tilt of the head makes localization of the intracranial VA difficult on 3D CTA images. An effective methodology utilized in the present study that circumvented this problem first plotted the axial plane on sagittal imaging relative to bony landmarks (that would be identifiable later during the course of the surgical exposure). Then, the course of VA was defined in the axial plane relative to the trajectory of the surgical approach using 3D multiplanar CT reconstruction method. Wang et al. classified variations in VA in the presence of an occipitalized atlas into four types. The course of VA in Type I is below the occipitalized C1 lateral mass; in Type II, it travels on the posterior surface or makes a curve on it; in Type III (the most common variety; 61.1%), it enters an osseous foramen created between the fused atlas and occipital bone; and in Type IV, there is absence of VA.  The latter classification, however, neither attempted to focus on the surgical implications of these variations nor did it include a risk stratification of the vulnerability of VA to iatrogenic injury. Moreover, it was directed solely towards patients with an occipitalized atlas; and also did not take into account the medial deviation of VA associated with neck tilt or rotation consequential to facet asymmetry or fusion. These relevant clinical concerns have been adequately addressed in the present study [Table 2].
Relationship of vertebral artery with its foramen transversarium
The FT forms around the developing VA. Its posterior aspect develops from the neural process and the anterior portion from a vestigial costal element that fuses to the vertebral body enclosing the VA. The size of VA has a close developmental relationship with its corresponding FT. In our study, the relationship between the diameter of the ipsilateral VA and its C1 FT showed a positive correlation. Thus, comparing the diameter of C1-2 FT at CVJ on either side may be a useful screening parameter on CTA.  A small VA in a large FT suggests acquired stenosis.  Approximately, 12% patients have unilateral hypoplasia/aplasia of the VA and consequent narrowing of its FT.  Yamazaki et al. reported nearly 16% of patients having a unilateral VA dominance.  The close association between unilateral hypoplasia/aplasia of the VA and segmentation bony anomalies at the CVJ (occipitalized atlas, facet fusion or C2-3 fusion) may be attributed to the common embryonic stage in which resegmentation of embryonic sclerotome and vascular sagittal rearrangement of intersegmental artery occur.  In the presence of a persistent first intersegmental artery or one or both VAs terminating in a low-lying PICA, unilateral FT narrowing may provide a valuable clue regarding the side of VA dominance. An injury to the dominant VA in the presence of its hypoplasia/aplasia on the contralateral side significantly enhances the risk of vertebrobasilar insufficiency and brain stem ischemia. Thus, posterior C1-2 stabilization on the side of the dominant VA should be undertaken with extra precaution. 
Stretching of vertebral artery in the presence of atlanto-axial dislocation
An interesting corollary of the primary study was the detection of straightening of the third part of VA in a significant number of patients with AAD (the stretched loop sign).  The usual redundancy of the two loops normally existing in the third part of VA (that permits unhindered atlanto-axial movements) was often compromised potentially increasing the vulnerability of VA to developing thrombosis or dissection in this most mobile segment of the cervical spine. Normally, at the C1-2 level, compromise of the contralateral VA starts occurring beyond 30° axial rotation of the neck; and at 45° of rotation, the ipsilateral VA also begins to kink.  The acute VA angulations in the presence of an occipitalized atlas, its stretching due the presence of AAD and its medial deviation due to C1-2 facetal asymmetry may lead to an earlier compression of the VA during axial neck movements. This may often precipitate a decreased perfusion in the posterior circulation territory. Compression within the osseous canal may also be an added problem.
The intracranial course of the VA was precisely defined utilizing 3D multiplanar CT reconstruction method relative to the surgical trajectory, focusing on its variations in patients with AAD with or without an occipitalized atlas and in normal patients. A classification is proposed that defines the anatomical variations in the size and course of the VA and its relationship to the bony landmarks at the CVJ. An objective risk stratification of the VA helps in significantly reducing its chances of injury during surgery.
We gratefully acknowledge the help and suggestions given by Professor Uttam Singh, Department of Biostatistics in the preparation of this article.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7], [Figure 8], [Figure 9]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]