Anatomy and Biomechanics

The ligaments that maintain the craniocervical articulation can be divided into two groups (Fig. 313-1). The first set attaches the skull to the atlas and includes the articular capsule ligaments, the anterior and posterior atlanto-occipital ligaments, and two lateral atlanto-occipital ligaments. The anterior atlanto-occipital ligament is a continuation of the anterior longitudinal ligament, and the posterior atlanto-occipital ligament courses between the posterior border of the foramen magnum and the posterior arch of the atlas. The cruciate ligament (which includes the longitudinally oriented extensions of the transverse ligament of the atlas) also contributes to the stability of this articulation.

FIGURE 313-1 Lateral (A) and posterior (B) views of the craniovertebral junction. Note that the cruciate ligament is composed of horizontal fibers (i.e., the transverse ligament) and vertical fibers. The tectorial membrane, the rostral extent of the posterior longitudinal ligament, has been reflected in B to allow visualization of more ventral structures.

The primary source of stability across the CVJ is provided by a second set of ligaments that secure the cranium to the axis. These ligaments include the apical dental ligament, the alar ligaments, the tectorial membrane, and the ligamentum nuchae.^5,6 The alar ligaments are paired structures with two components each: the atlantal alar and the occipital alar. These ligaments connect the tip of the odontoid to the occipital condyles and the lateral masses of the atlas, respectively.⁷ The alar ligaments are the main restraints to axial rotation, which occurs mostly across the C1-2 articulation. The alar ligaments also limit lateral flexion and anteroposterior translation.⁵

The tectorial membrane is the continuation of the posterior longitudinal ligament. It begins at the dorsal surface of the odontoid and inserts on the ventral surface of the foramen magnum.^8,9 The tectorial membrane primarily resists hyperextension, although if incompetent, contact between the posterior arch of the atlas and the occiput will limit hyperextension.^5,10 Flexion is restricted by contact of the odontoid process with the anterior foramen magnum.⁵ The apical dental ligament and the ligamentum nuchae contribute only modestly to stability of the CVJ.

Mechanisms of Injury

Craniocervical dissociation is the consequence of complete or nearly complete disruption of the ligamentous structures between the occiput and upper cervical spine. The extreme forces produced by hyperextension, hyperflexion, lateral flexion, or a combination of these forces lead to ligamentous failure.^11–13 The primary force responsible for producing AOD appears to be hyperextension, which leads to rupture of the tectorial membrane.^14,15 Another theory focuses on damage to the alar ligaments by extreme lateral flexion.^16–18 Incompetence of the alar ligaments and tectorial membrane allows anterior dislocation of the cranium with respect to the upper cervical spine.⁵ Other authors have suggested that hyperflexion forces may also be involved in some cases of AOD based on the observation that the posterior elements of C1 and C2 are commonly separated in the setting of craniocervical dissociation.¹²

The increased incidence of craniocervical dissociation in children^3,12,14 versus adults may be related to the relatively high incidence of automobile-pedestrian accidents, the pediatric anatomy of the CVJ, or both.^5,12 The immature shallow, horizontally oriented surfaces at O-C1 are less biomechanically stable than the deep-seated, vertically oriented articulation in adults. Moreover, a child’s ligaments are not as stiff as an adult’s, and they are subjected to greater stress because of the need to support a proportionately larger head than in an adult for a given body size.

Clinical Findings

High-speed motor vehicle accidents and striking of pedestrians by motor vehicles are the two most common causes of craniocervical dissociation.^3,12 Brainstem or upper cervical spinal cord injury at the time of dislocation probably accounts for the high mortality rate seen with AOD.¹⁹ Patients who survive may be neurologically intact or have dysfunction of the brainstem, cranial nerves, spinal cord, or cervical nerve roots.^18–20 The neurological deficits range from minor to severe. Many patients also suffer a head injury, which can confound the clinical examination.

Patients with craniocervical dissociation may have normal results on motor examination, or they may display deficits, including vegetative posturing responses, cruciate paralysis, or quadriparesis. At the level of C1, the normal spinal canal is capacious. As described by Steel and codified as the Steel rule of thirds, a third of the axial spinal canal diameter is filled by the dens, a third is filled by the spinal cord, and a third is space typically occupied by cerebrospinal fluid.²¹ Cruciate paralysis, described by Bell, refers to weakness in the hands and arms with relative sparing of leg strength.²² This mirrors the cervical injury pattern seen clinically as described by Schneider and colleagues.²³ Weakness often recovers, with improvement seen first proximally and then distally in the hands. Long held to be due to injury to the somatotopically organized crossing lateral corticospinal tract, recent evidence refutes this theory.^24,25

Many patients commonly have hemiparesis that may result from unilateral pyramidal injury, although more rostral structures in the brainstem may also be involved. The cranial nerves injured include VI, IX, X, XI, and XII. Compression, dissection, or stretch injuries may damage the anterior spinal, vertebral, or carotid arteries or any combination of these vessels.

Radiology

Although the radiographic findings in patients with craniocervical dissociation can be quite dramatic, they may be subtle or even absent on initial films.^16–19 When alignment of the bony structures of the CVJ appears normal, one must appreciate indirect signs that craniovertebral dislocation may have occurred. Abnormally prominent prevertebral soft tissues should always increase suspicion for serious spinal injury. In the case of AOD, retropharyngeal hematoma is invariably present. Other signs include retropharyngeal emphysema^14,26 or an increase in the interval between the posterior elements of C1 and C2.¹²

Several measurement protocols using plain lateral cervical radiographs to diagnose craniocervical dissociation have been developed. The basilar line of Wackenheim is drawn as a caudal extension of the posterior surface of the clivus in the midsagittal plane (Fig. 313-2A).²⁷ Normally, Wackenheim’s line lies tangential to the posterior tip of the odontoid process and is not altered by flexion or extension. The line will intersect the dens if the occiput is displaced anterior to the atlantoaxial segment. If the skull is displaced posteriorly, the line will not intersect any portion of the odontoid process.

FIGURE 313-2 Schematic diagrams depicting various criteria developed to assess the craniovertebral junction. Wackenheim’s line (A) is drawn as a caudal extension of the clivus and should tangentially intersect the tip of the dens in a normal individual. A normal dens-basion interval (B) should not exceed 5 mm in an adult and 10 mm in infants. C, The distance between the basion (B) and inner surface of the posterior arch of C1 (C) divided by the distance from the opisthion (O) to the inner surface of the anterior arch of C1 (A) defines the Powers ratio. The mean value in normal subjects is 0.77.

The interval between the odontoid tip and the basion remains constant when the CVJ is in the neutral position (Fig. 313-2B).²⁸ A distance of greater than 5 mm in adults and 10 mm in infants has been suggested as abnormal.^24,25 The sensitivity and specificity of the dens-basion interval, however, have been found to be suboptimal. Powers and coworkers reviewed 100 adult and 50 pediatric cervical spine radiographs and reported that 85% of individuals exceeded the proposed limits. With a reported average distance of 9 mm, as many as 50% of patients with known AOD might fall within the range spanned by the normal population.⁴

Subsequently, Powers and colleagues suggested that the ratio of two measurements could be used to define normal craniovertebral relationships (Fig. 313-2C).⁴ The first distance is measured between the basion (B) and the inner aspect of the posterior atlantal arch (C), and the other represents the distance between the opisthion (O) and the inner aspect of the anterior atlantal arch (A). The mean BC/OA ratio in normal subjects is 0.77, and a value greater than 1.0 may indicate AOD. Precise identification of the anatomic landmarks is necessary to properly determine the Powers ratio, and this factor may be inaccurate when congenital anomalies are present or the atlas is fractured. Additionally, in select situations of posterior or pure longitudinal craniovertebral disassociation, a value of less than 1 may be obtained.^29–31

Lee and associates developed the X-line method to detect craniovertebral dissociation.²⁹ The first line courses the distance between the basion and the midpoint of the C2 spinolaminar line (BC2Sl), and the second runs from the opisthion to the posteroinferior edge of the axis body (C2O). The BC2Sl line should tangentially intersect the posterosuperior aspect of the dens, and the C2O line should tangentially cross the most rostral edge of the C1 spinolaminar line in normal individuals older than 5 years. Although one of these linear associations may be disturbed in normal subjects, alteration of both of these relationships should lead one to be highly suspicious of the presence of AOD. Appropriate use of this method requires a normal atlantoaxial configuration, which is problematic because this segment may be abnormal in more than half of those with craniovertebral dissociation.¹²

Harris and associates described two measurements, the basion–posterior axial line (BAI) and the basion-dental interval (BDI), that may be used to assess the CVJ (Fig. 313-3).³² Their study of 400 adult radiographs showed that the BAI measurement did not exceed 12 mm 98% of the time and that the 95% accuracy range of the BDI was 11.8 mm. Thus, the CVJ can be considered normal when both the BAI and BDI are less than 12 mm. Although the BDI in children younger than 13 years is not reliable because of the variable age until complete ossification of the dens, the BAI measurement held true.

FIGURE 313-3 Schematic diagram depicting the basion–posterior axial line (BAI), and the basion-dental interval (BDI).

In a subsequent publication, Harris and colleagues compared this “rule-of-12 technique” with other methods of CVJ assessment.³³ The BAI/BDI method detected craniocervical dissociation accurately regardless of the magnitude and direction of the dissociation. The Powers ratio and X-line method were unusable 46% of the time because of imaging shortcomings or C1 developmental problems. When present, AOD was correctly determined with the Powers ratio and the X-line method with much lower accuracy rates, 60% and 20%, respectively. Fisher, Deliganis, and their coworkers validated the BAI/BDI technique and found these measurements to be the most useful for diagnosing occipitoatlantal dissociation.^34,35 Although probably superior to other methods of detecting AOD, interobserver and intraobserver reproducibility of the BAI/BDI methods still needs to be addressed.³⁶

The Harris rule-of-12 measurements are effective and universally applicable. They are easily determined on plain lateral cervical radiographs obtained at a 40-inch (1-m) target distance on supine patients. The basion and posterior cortex of the axis are midline structures unaffected by moderate degrees of rotation.

High-resolution computed tomography (CT) is an accurate means of diagnosing AOD^12,29,37,38 and should include sagittal and coronal reconstructed images. A strong case can be made that fine-cut CT should be the study of choice for diagnosing AOD. In blinded tests comparing plain films and CT, with the evaluators being neurosurgeons, orthopedists, radiologists, and emergency department staff, the sensitivity, specificity, and positive and negative predictive values of AOD detection methods were improved, regardless of the measurement techniques used.³⁹ The relevant anatomy was visualized 99.75% of the time with CT versus 39% to 84% of the time with plain films. With the ubiquity of CT scanning and shifting of the standard of care regarding AOD diagnosis, normal anatomic relationships may need to be reevaluated. Rojas and associates found that on CT the normal BDI was less than 8.5 mm 95% of the time as opposed to 12 mm on plain films.⁴⁰ In addition, the BAI was difficult to interpret on CT. The accepted normal adult value for the BDI is significantly different between CT and plain films.

Additional findings on CT that may suggest craniovertebral dissociation include occipitoatlantal joint incongruity, focal hematoma, vertebral artery injury, capsular swelling, cranial nerve canal fracture, and infratentorial hematoma.^35,41

A new criterion for AOD in children was described by Pang and coauthors.⁴² The actual occipital condyle–C1 lateral mass articulation is assessed by CT imaging, and the distance between the two articular surfaces is referred to as the condylar-C1 interval (CCI). The normal value obtained by studying 89 children was 1.28 (standard deviation [SD], 0.26 mm). Left and right normal measurements were essentially symmetrical. The CCI was then measured in a series of 16 patients who had previous radiologic criteria for AOD.⁴³ Its diagnostic sensitivity was 100%, which was much better than that of the other standard tests: Whorley’s, 50%; Powers’, 37.5%; Harris’, 31%. The diagnostic specificity of the CCI was 100%, once again bettering the standard tests. Once the O-C1 joint is injured, it does not return to its preinjury interval. The benefits of using the CCI include essentially no logistic or technical distortions, assessment of dislocation in all three planes, and no concealment by postinjury head and neck relationships.

C1-2 distraction frequently occurs in conjunction with craniocervical dissociation.¹² The lateral mass interval (LMI) between C1 and C2 was ascertained in 92 control patients undergoing CT angiography by Gonzalez and colleagues.⁴⁴ The LMI, typically seen best on coronal CT images, is defined as the widest gap between the C1 and C2 lateral masses when measured perpendicular to the joint space. Control group LMIs averaged 1.77 mm (SD, 0.48; range, 0.7 to 3.3 mm). The 95th percentile range in normal patients was 0.7 to 2.6 mm. Left and right averages were essentially identical. Cup-shaped lateral mass spaces were measured at the narrowest gap interval to avoid false-positive results. When measured in patients with known distraction injuries, the mean LMI was 5.5 mm. The BDI was also calculated, with the normal group averaging 4.7 mm (SD, 1.7 mm). The average BDI in the injured group was significantly increased to 11.9 mm. Magnetic resonance imaging (MRI) was performed in the injured group and showed markedly increased signal bilaterally at the C1-2 lateral mass articulations.

There are generalized MRI findings of injury that can aid in interpreting studies to assess injury at the CVJ. Injury is usually seen as increased T2 signal in the atlantoaxial and atlanto-occipital joints, craniocervical ligaments, prevertebral soft tissues, and spinal cord.³⁵ Injury to the anterior or posterior atlanto-occipital ligaments or the supraspinal ligament may be seen as loss of the black stripe sign and can be reliably diagnosed with MRI.^45,46 In a limited number of cases, craniocervical dissociation has been diagnosed by MRI findings alone.³ Additionally, spinal cord signal abnormalities can be correlated with the results of clinical examination, and vertebral artery injury or thrombosis, or both, may be discovered.

Three specific types of craniocervical dissociation have been identified that can easily be classified on lateral cervical radiographs or sagittal CT images (Fig. 313-4).¹² Type I dislocation is defined as anterior displacement of the occiput with respect to the atlas. Type II is primarily longitudinal distraction with separation of the occiput from the atlas. Type III AOD exists when the occiput is positioned in a posterior direction relative to C1. Rotatory and lateral atlanto-occipital dislocation has also been described.^3,47,48

FIGURE 313-4 Classification of atlanto-occipital dislocation as proposed by Traynelis and associates.¹² The normal craniovertebral junction is represented in A. Dislocations may occur as a result of anterior displacement (B), longitudinal distraction (C), or posterior displacement (D) of the occiput relative to the atlas (arrows).

Treatment Considerations

Maintaining awareness of the possibility of craniocervical dissociation is critical to making the diagnosis. It is important to provide appropriate cardiopulmonary support and spinal immobilization. Surgical hematomas at the CVJ in the setting of craniocervical dissociation are very rare but may require emergency decompressive procedures if associated with neurological deficit.

The use of traction is controversial when treating patients with craniocervical dissociation. The goal of traction, if used, is to decompress the neural elements by realigning the bony structures. Some authors have concluded that traction should not be used because the injury is highly unstable.^3,17,20,31 Others have documented its benefits in achieving a successful outcome.^{4,16,18,19,49}

The type of dislocation (I, II, or III) has been suggested as a means of guiding the decision to use traction.¹² Implementation of traction is best reserved for type I (anterior), type III (posterior), and lateral craniocervical dissociation associated with neurological deficit.^12,48 Traction in the setting of a type II dislocation is contraindicated because the pathology is primarily a result of longitudinal distraction, which would only be worsened by traction. Even with rotatory subluxation, small amounts of traction have resulted in a dangerous amount of distraction.³

A reasonable approach to patients with only minor malalignment or slight neurological deficits is to attempt manual realignment under fluoroscopic guidance. Thus, the use of traction is reserved for patients with grossly abnormal alignment accompanied by significant neurological deficit. The use of fluoroscopy during the initial application of traction is strongly recommended. The amount of weight required should not exceed 5 lb. Frequent clinical and radiographic monitoring is essential. Neurological worsening or an increase in distraction or subluxation should prompt a reduction or discontinuation of the traction. Once the end point of improvement in neurological status or radiographic realignment has been achieved, traction may be reduced to 1 to 2 lb or discontinued and the patient immobilized in a halo vest.

AOD frequently heals poorly with strictly conservative measures because it is primarily a ligamentous injury with severe instability. Although external immobilization in a halo vest may allow the injury to heal, this is unlikely and maintenance of reduction may prove difficult.^3,15,16,50 A more definitive approach is to proceed with posterior fusion once the patient is medically stable.^3,12,20,51 In fact, Dickman and colleagues recommend bypassing traction and performing acute internal reduction and fixation.³ Dorsal fusion with rigid internal fixation that incorporates at least the occiput to C2 is required for an adequate construct.

Whether caused by immediate neurological devastation or trauma to other organ systems, AOD remains a highly lethal injury; however, patients surviving 48 hours after the traumatic event may have a good outcome. Up to a quarter of surviving patients are ultimately intact neurologically, and in another quarter only minor neurological deficits remain.¹²