Occipital-Cervical Spine Injuries
Injuries to the occipital-cervical spine largely fall into two broad categories, (1) atlanto-occipital dissociations (AODs) or craniocervical dissociations (CCDs) and (2) occipital condyle fractures. Historically, the term AOD has been used; however, because these injuries can involve the occipital–C1 articulation, the C1–C2 junction, or a combination of the two, the term CCD is probably better because it is more accurate and encompassing. By pure terminology, AOD is a type of CCD, yet there are other patterns of CCD that really would not fall into the pure definition of AOD.
These injuries involve trauma to the complex articulation, including the occipital bone, the occipitoatlantal articulation, the atlas, the axis, and the ligaments that span from the axis to occiput. The susceptibility of the cervicocranium to injury is related to (1) the large lever arm induced by the mass and immobility of the cranium combined with (2) the relative freedom of movement more caudally with reliance on ligamentous structures rather than on intrinsic bony stability for the maintenance of craniocervical alignment. This functional unit is maintained by highly specialized bony segments connected via a complex ligamentous system whose vulnerability to injury may compromise the structural integrity of the craniocervical junction. Injury to the craniocervical junction is almost always caused by high-energy trauma and is frequently associated with other injuries, including closed head injuries, facial fractures, and either associated atlas or axis fractures or subaxial spine injuries. Craniocervical injuries with associated high cervical spinal cord injuries are thought to account for 10% to 25% of traffic fatalities.
Certainly, within these injuries there is a spectrum of instability, ranging from very stable nonoperative injuries such as isolated, nondisplaced occipital condyle fractures to highly unstable injuries with severe spinal cord injuries such as widely distracted CCDs. Increasing awareness of these injuries and use of routine computed tomography (CT) has resulted in earlier diagnosis and more appropriate, aggressive management, thus allowing an increasing number of these patients to survive. Despite the evolution of learning and improvement in management, cases of catastrophic failure to diagnose and subsequent neurologic deterioration still occur even in experienced trauma centers.
The goal of this chapter is to review the anatomy and methods of diagnosis to identify the wide spectrum of injuries that occur at the occipital cervical junction. Current classifications will be discussed as well as operative and nonoperative management and outcomes of treatment.
The skull base, atlas, and axis comprise the three bony components of the upper cervical spine and form an integrated functional unit. The five unconstrained joints of the upper cervical spine rely primarily on an intact, multilayered ligamentous system for stability. This unique anatomic arrangement allows the upper cervical spine to contribute a substantial portion of neck motion.
The occiput forms the major portion of the foramen magnum ( Fig. 33A-1, A ). From the anterolateral aspect of foramen magnum, the occipital condyles project caudally on each side to form convex bony surfaces that articulate with the matched concave superior articular facet of the atlas with a joint capsule and synovial joint ( Fig. 33A-1, B ). These bony protuberances are semilunar in shape, forming almost a 180-degree arc when viewed sagittally. From the anterior view, they are wedge shaped with increased extension medially tapering off more laterally. The geometry of the occiput as it articulates with C1 allows the condyles to move like a rocking chair within the C1 lateral masses. Posteriorly, on the inside of the skull along the midline, the internal occipital crest extends toward the transverse sulcus and is the key location for occipital fixation. Just lateral and ventral to the occipital condyles is the hypoglossal foramen through which cranial nerve XII descends. Given the close proximity of cranial nerve XII, it is prone to injury with fractures of the occipital condyle and AODs and with C0–C1 transarticular screw fixation. The ventral aspect of foramen magnum is bounded by the basion, which is the caudal extent of the clival plate. The dorsal boundary of foramen magnum is the opisthion.
The atlas is a complicated ring-shaped structure allowing for a critical link between the occiput and the axis. The ring is formed from large lateral masses connected together by thin ventral and dorsal arches ( Fig. 33A-1, C ). The lateral masses viewed anteriorly are trapezoidal in shape to match the occipital condyles, essentially narrow medially and project out to a thickened lateral aspect similar to a bowtie. Lateral to the lateral mass is a foramen through which the vertebral arteries pass as they course rostrally from C2 to wrap posteriorly over the C1 arch and enter the foramen magnum to form the basilar artery. Just anterior to the C1 ring lies the internal carotid artery (ICA) and the hypoglossal nerve. Internally within the ring, the transverse atlantal ligament (TAL) connects the lateral masses together while passing posterior to the dens. The atlas is critical in providing flexion and extension proximally as it articulates with the occipital condyles while providing about 50% of the rotation of the cervical spine as the C1 lateral masses rotate over the lateral masses of C2. Werne and others have shown that the atlanto-occipital articulation has 15, 8, and 0 degrees of motion in flexion–extension, lateral bending, and axial rotation, respectively.
The axis is unique in its anatomy. The dens extends behind the ventral arch of the atlas and is maintained in position by the TAL. The axis is an essential component of the craniocervical articulation because all the major restraining ligaments attach between the axis and occiput and not directly to the atlas. The lateral masses of the C2 are directly under those of C1, thus directly supporting them. The vertebral arteries extend along the inferior lateral border of the C2 body and then exit laterally through the transverse foramina.
The craniocervical junction does not have ligamentum flavum or intervertebral discs (see Fig. 33A-1, B ). The ligamentous structures of the craniocervical junction can primarily be divided into internal ligaments within the canal and external ligaments outside the canal.
The external ligaments include the ligamentum nuchae, the anterior and dorsal occipitoatlantal membrane, and the occipitoatlantal and atlantoaxial facet joint capsules. The anterior and dorsal occipitoatlantal membranes are thin structures that offer minimal protection to the dura. The facet joint capsules are thin and redundant to facilitate wide range of motion yet are key stabilizers to the stability of the occipital cervical junction.
The internal ligaments are also the key stabilizers. The alar ligaments are paired, thick, cordlike structures that project laterally and ventrally from the dens tip to the ventral medial aspect of the occipital condyles. They are the primary restraints to rotation of the upper cervical motion unit. With an average in vitro load to failure of 210 N, however, these vitally important ligaments tolerate less than 50% load to failure than the cruciate ligaments of the knee. These complex ligaments serve a variety of functions. At midposition of the head, they are slack. By turning the head in one direction, the alar ligament contralateral to the direction of rotation tightens, and the ipsilateral ligament slackens. Together with the tectorial membrane, the alar ligaments limit flexion. However, they play no role in limiting extension. The contralateral alar ligament limits lateral bending. The broad tectorial membrane, which constitutes the rostral extension of the posterior longitudinal ligament, effectively limits axial distraction and atlanto-occipital flexion and is considered, along with the alar ligaments, to be one of the major stabilizing ligaments of the craniocervical junction. Anteriorly, the well-developed atlanto-occipital membrane, an extension of the anterior longitudinal ligament (ALL), limits extension with the thinner anterior atlantoaxial membrane contributing to a less significant degree. The apical ligament is a rudimentary structure extending from the dens to the ventral midpoint of foramen magnum.
The cruciate ligament lying behind the dens consists of the transverse ligament and fibers that extend upward to attach to the basion and caudally to the axis. The primary stabilizing ligament of the atlantoaxial motion unit is the cruciate ligament complex, which contains the TAL. By crossing the odontoid at its waist, atlantoaxial flexion, translation, and distraction are minimized, yet rotation is allowed. In flexion, the cruciate ligament is placed under tension, thereby preventing the odontoid from compressing the spinal cord.
Supplemental ligamentous support of the craniocervical junction is provided by a number of smaller ligaments, such as the apical and cruciate ligaments, the obliquely aligned accessory atlantoaxial ligaments, the anterior atlantodental ligament, and the facet joint capsules. Although the facet joint capsules have long been considered secondary stabilizers of the craniocervical junction, a recent biomechanical, cadaveric study has indicated that they share equal importance with the tectorial membrane and alar ligaments in stabilizing the craniocervical junction.
The craniocervical junction accounts for approximately 60% of axial-plane cervical spine rotation, 40% of sagittal-plane flexion–extension motion, and 45% of overall neck motion. The occipital-atlas joint has roughly 15, 8, and 0 degrees of motion in flexion–extension, lateral bending, and axial rotation, respectively. The normal C1–C2 rotational excursion ranges from 80 to 88 degrees and flexion–extension excursion ranges from 20 to 30 degrees. Total left-to-right lateral bending amounts to about 20 degrees at C1–C2.
The potential for upper cervical instability should be considered in all patients who have sustained high-energy injuries. After resuscitation, awake and alert patients are evaluated for spinal tenderness and by neurologic examination. Because cranial trauma and other injuries are common, distracting injury or impaired consciousness precludes adequate evaluation in many patients. In cognitively unimpaired patients, upper cervical spine fractures and dislocations are usually accompanied by neck pain, headache, and nuchal tenderness. Neurologic assessment should be performed according to American Spinal Injury Association guidelines.
Cranial nerve function should be part of any examination of patients with possible head or neck injuries. The abducens and hypoglossal nerves are most commonly affected by craniocervical injuries. A wide variety of neurologic injury patterns are possible in patients with trauma to the upper cervical spine, probably based on the location of neurologic injury relative to the decussation of the pyramidal motor tracts. These range from complete pentaplegia to incomplete injuries, such as cervicomedullary syndromes and disorders affecting brainstem function. The cervicomedullary syndromes, which include cruciate paralysis as described by Bell and hemiplegia cruciata initially described by Wallenberg, represent the more unusual forms of incomplete spinal cord injury and are a result of the specific anatomy of the spinal tracts at the junction of the brainstem and spinal cord. Cruciate paralysis can be similar to a central cord syndrome, although it normally affects proximal more than distal upper extremity function. Hemiplegia cruciata is associated with ipsilateral arm and contralateral leg weakness.
Obtunded patients present increased diagnostic challenges, yet many physical findings can still be assessed. In these patients, physical examination is frequently limited to inspection, palpation, assessment of spontaneous muscle tone, response to painful stimuli, reflexes, and anal sphincter tone. In these patients, early radiographic assessment is critical to help clarify whether there is any instability.
Although now largely supplanted by CT imaging, plain radiographs with or without tomography were historically considered the initial workup for cervical trauma. With regard to the upper cervical spine, lateral radiographs have several shortcomings. Because the typical lateral cervical spine radiograph is centered in the midneck region, interpretation of the craniocervical junction can be impaired by parallax or the obliquity of the C1 superior articular surfaces and the occipital condyles. Even minor malrotation of the head further distorts the occipital condyles and the neural arch of the axis, thus reducing the diagnostic value of such radiographs.
A high degree of emphasis has been placed on interpretation of screening lines on lateral radiographs to warn of the possibility of AOD. Prevertebral soft tissue swelling on the lateral cervical radiograph may be present ( Fig. 33A-2 ). The Powers BC/AO ratio, which compares the distance between the basion and posterior arch of the atlas (BC) with the distance between the anterior arch of atlas and the opisthion (AO), has poor reliability as does the atlanto-odontoid-basion distance, initially described by Wholey and colleagues. Harris and colleagues refined the Basion-Dens interval (BDI) and added the basion-axis interval (BAI). Both the BAI and the BDI should remain 12 mm or less in 95% of adults (“rule of 12s”) ( Fig. 33A-3 ). Although indicative of AOD if positive, 35% of patients may have normal BAIs and BDIs and still have an AOD. Thus, the Harris lines are not completely reliable and seem to have much higher specificity than sensitivity in detecting occipitocervical dissociation.
As implied earlier, radiographic representation of the craniocervical distance also varies with age. Kaufman and colleagues proposed measuring the actual distance between the articular surfaces of the occiput and the superior facet of the atlas on a lateral cervical spine radiograph. These authors held 5 mm as the maximum distance. However, obliquity and rotatory malposition of the head by even a few degrees, as well as mastoid process overlap, can make this measurement attempt challenging, if not nearly impossible. Using a cohort of 16 pediatric patients with AOD and a comparison group of 138 intact patients the authors identified significant false-positive rates for other screening tests, such as Sun’s, Harris’, Wholey’s, and Powers’.
Computed tomography is the imaging modality of choice in high-risk trauma patients. Helical imaging with sagittal and coronal reformats have been shown to be timely, cost effective, and more sensitive and specific in high-risk patients, particularly at the craniocervical junction. Three-dimensional image reformations, obtained from fine-cut CT scans, are rarely clinically useful but may assist with interpretation of more unusual upper cervical injury patterns. The potential for harmful effects of radiation from diagnostic CT, particularly to the thyroid, should be considered.
Occipital condyle fractures are rarely visualized on plain radiographs and are almost always diagnosed from head or cervical spine CT. Reformations of the atlanto-occipital articulation in the coronal and sagittal planes are essential to determine their stability. Displacement may indicate instability caused by rupture of the internal craniocervical ligaments.
Interpretation of the CT scan is critical to the timely diagnose of CCDs. Occipital condyle fractures and other fractures of C1 and C2 may be intrinsically unstable but also may portend a much more serious injury such as CCD.
One of the primary advantages of CT imaging over plain radiographs is that parasagittal and coronal CT images can be used to directly assess the congruency of the occipitocervical junctions. Subluxation or distraction can therefore be directly identified rather than relying on indirect measurement via radiographic lines ( Fig. 33A-4 ). Pang and colleagues reported an average craniocervical interval (CCI) of 1.28 mm in normal children 0 to 18 years of age with a high degree of conformity between left- and right-sided measurements. None of the CCIs exceeded 1.95 mm. In adults, gapping of more than 2 mm between the occipital condyles and C1 lateral masses indicates craniocervical instability. The coronal reformats are also useful to assess widening either between the occiput and C1 ( Fig. 33A-5 ) or between C1 and C2 ( Fig. 33A-6 ). Certain fracture patterns may also be suggestive of distraction such as type I odontoid fractures caused by alar ligament avulsions or type III occipital condyle fractures, also caused by alar ligament avulsions. A horizontal cleavage fracture of the anterior C1 ring has been implicated as having a distractive injury pattern that is associated with AODs ( Fig. 33A-7 ).
Magnetic Resonance Imaging
If magnetic resonance imaging (MRI) is indicated, it is imperative to communicate to the radiologists the goal of the MRI so that it is protocoled appropriately. An MRI examination geared specifically toward the craniocervical junction will be more sensitive than a standard cervical spine MRI. In the setting of trauma, the fat suppression sequences and coronal sections are the most helpful. An MRI is indicated either in the presence of neurologic injury to assess spinal cord injury or to aid in the diagnosis of an AOD. Indicators of a highly unstable injury include significant prevertebral soft tissue swelling, increased joint edema at the occipitocervical joints or C1–C2, tectoral membrane disruption, subarachnoid hemorrhage, and ligamentous injury to the alar ligaments. MRI may be overly sensitive and must be interpreted in light of the mechanism and CT findings.
The traction test is a unique dynamic examination used to help to determine whether a CCD actually exists when CT, MRI, and all other tests are equivocal. There is no other spine trauma situation in which it is used, and even in the realm of CCD, its utility is quite limited because most injuries are delineated with CT imaging either alone or combined with MRI. In the few cases in which CT and MRI are not diagnostic, a radiographic traction test can be performed. Greater than 2 mm of distraction between the occiput and C1 or between C1 and C2 is indicative of an unstable injury ( Fig. 33A-8 ). The amount of weight required for traction testing has not been well defined, although cadaveric studies suggest that the craniocervical traction test reliably demonstrates instability and requires no more than 5 to 10 lb of traction to yield a positive result when the alar ligaments, the tectoral membrane, and the joint capsules are disrupted. In the authors’ experience, the primary role of traction testing has been to confirm that there remains sufficient ligamentous integrity of the craniocervical junction to proceed with nonoperative treatment in situations in which imaging studies have shown some worrisome features for craniocervical instability (e.g., degree of joint subluxation) but other findings (e.g., extent of soft tissue swelling) have been less convincing. Our experience has been that traction testing has served to decrease our diagnosis of occipitocervical dissociation in situations when the diagnosis would otherwise have been made if based on strict interpretation of static imaging parameters, thus saving patients from the morbidity of an unnecessary occipitocervical fusion.
Occipital Condyle Fractures
Mechanism of Injury
Occipital condyle fracture can occur via a variety of mechanisms, which are illustrated in the classification. The three primary mechanisms involve impaction or axial loading, distraction with avulsion, or direct blows to the head with associated skull fractures that may then extend to the occipital condyle.
Classification and Management
The most commonly used classification is the three-part classification described by Anderson and Montesano in 1988 ( Fig. 33A-9 ). Type I fractures are caused by impaction or axial load. Frequently, these are comminuted in nature.
Type II injuries are skull-based fractures that extend into the occipital condyle. Type III injuries are avulsion injuries with the alar ligaments “pulling off” a bony piece of the occipital condyle typically because of distraction. These injuries may potentially be unstable and associated with a CCD.
These injuries are frequently missed on plain radiographs, although, on occasion, an open-mouth odontoid view can demonstrate an occipital condyle injury. With the increased use of CT scans, these injuries are more readily identified. Typically, they are more easily identified with coronal CT reformats ( Fig. 33A-10 ).
The management of occipital condyle fractures has been widely discussed and is largely dependent on whether there is associated CCD. Craniocervical instability is identified when there is displacement of the occipital condyles and C1 lateral masses or by a positive traction test result. Type I injuries can be treated conservatively because they are stable injuries with minimal risk of displacement or neurologic injury. Surgery is not recommended, but even the use of various external orthoses has not been proven to make a difference in outcome.
The management of type II injuries is based on the extent of the skull fracture and underlying head injury rather than on the less relevant occipital condyle injury. In most cases, treatment will consist of hard collar. Rarely, the entire occipital condyle may be sheared off the skull, resulting in the need for occipitocervical stabilization.
If a type III injury is identified or suspected, an MRI of the craniocervical junction can help assess the degree of instability and help ascertain the presence or absence of an associated CCD. If a CCD is not present, then a hard collar can be used, although no studies comparing nonoperative treatment methods are available. Type III fractures with CCD are managed by occipitocervical fusion.
Outcomes and Associated Injuries
Hanson and colleagues retrospectively reviewed 95 patients with 107 occipital condyle fractures. According to the Anderson classification, there were 3, 23, and 65 cases of type I, II, and III, respectively. Unilateral injury was present in 77% of cases. More than one-third of patients had additional cervical spine fractures. Twelve patients had craniocervical instability and were treated surgically. Thus, only 18% with avulsion injuries had associated unstable craniocervical dissociations. Long-term results were related to associated traumatic brain injury (TBI) rather than the occipital condyle fracture itself.
Another retrospective review looked at 100 patients having 106 occipital condyle fractures. TBI was present in 56 percent of patients. Unilateral injuries occurred in 94% of this group. Three patients were treated surgically, all of whom had atlanto-occipital dislocation associated with occipital condyle fracture. At follow-up, no patients treated nonoperatively developed late instability or required other treatment. The treatment algorithm suggested in this study supports that stable injuries without displacement can be treated in a rigid collar for 6 weeks. For cases with mild displacement, a halo orthosis may be beneficial. Occipitocervical fusion was performed in the three patients with associated CCD.
Craniocervical dislocations occur in 0.67% to 1.0% of all acute cervical spine injuries and are present in 8% of victims of fatal motor vehicle accidents. Craniocervical injuries are now recognized as a spectrum of injury patterns with varying degrees of stability.
Traynelis and colleagues identified three craniocervical dissociation patterns based on the direction of displacement ( Fig. 33A-11 ). This system is limited because the extreme instability of AOD injuries renders the position of the occiput relative to the neck arbitrary and more dependent on external positioning forces, and there is an absence of a severity component of the injury.
A useful classification system must quantifiably assess the stability of the craniocervical junction. Signs of instability are translation or distraction of more than 2 mm in any plane, neurologic injury, or concomitant cerebrovascular trauma. The problem lies in segregating patients with minimally displaced ( < 2 mm) craniocervical injuries who can be treated nonoperatively versus those with highly unstable but partially reduced injuries who require operative stabilization in spite of misleading well-aligned static images. The Harborview craniocervical injury classification attempts to identify the severity of the traumatic disruption in a three-tier system analogous to that of basic ligamentous extremity injury ( Table 33A-1 ). Type I injuries are isolated structural injuries and can be treated nonoperatively; these include unilateral type III occipital condyle injuries or isolated alar ligament tears. A type III injury is a complete disruption of all interconnecting ligaments with obviously unacceptable instability; patients are subclassified on the basis of whether they survive for at least 24 hours from the time of their injury.
|Stage||Description of Injury|
|1||MRI evidence of injury to craniocervical osseoligamentous stabilizers|
|Craniocervical alignment within 2 mm of normal|
|Distraction of 2 mm or less on provocative traction radiograph|
|2||MRI evidence of injury to craniocervical osseoligamentous stabilizers|
|Craniocervical alignment within 2 mm of normal|
|Distraction of more than 2 mm on provocative traction radiograph|
|3||Craniocervical malalignment of more than 2 mm on static radiographic studies|
The type II injury, which is a craniocervical disruption with borderline radiographic screening values, is inherently unstable but may be missed on cursory evaluation or even difficult to categorize as unstable based on careful review of the imaging. Clinical evaluation of these patients is often unhelpful because the lesser degree of displacement usually equates to the absence of neurologic deficits. Incomplete, type II stable injuries of the craniocervical junction can have similar degrees of displacement as partially reduced yet highly unstable injuries. The differentiation between the two is a primary challenge to timely recognition of craniocervical dissociation. The authors have found dynamic traction testing to be a useful diagnostic aid in the accurate categorization of these patients with type II injuries (see Fig. 33A-8 ).
Studies have suggested that a delay in diagnosis may result in secondary neurologic deterioration in patients with these potentially life-threatening injuries. Although the advent of a systematic head and neck CT protocol has likely contributed to the reduction of missed craniocervical injuries, improved education and awareness of these injury types among survivors of high-energy trauma has probably played a greater role.
In summary, the diagnosis of CCD is often missed on plain radiographs with sensitivity ranging from 0.57% to 0.76%. The inclusion of the upper cervical spine in routine head CTs obtained for the assessment of the obtunded patient can reveal the presence of a suboccipital hematoma, which may be indicative of CCD. The inclusion of the foramen magnum in routine cranial CT scans has also increased the rate of detection of occipital condyle and type I odontoid fractures, which may be an indicator for a CCD as well. Definitive CT scan of the cervical spine should include reformatted views, including sagittal and coronal views, especially of the transition zones. If there are questionable findings but the diagnosis is still in question, MRI may be warranted. MRI can aid in identifying ligament injuries, intramedullary changes, and hematoma formation in the epidural or paravertebral spaces. In the rare cases of demonstrated cord disruption, MRI may play an important role in the discussion of life-prolonging interventions. Despite the availability of definitive diagnostic testing, patients with CCD continue to be subject to critical delays in timely diagnosis with diagnostic delays as long as 2 years having been reported.
The prognosis of patients with CCD appears to be related to the severity of the initial neurologic findings. If patients survive CCD, there appears to be a trend toward improvement neurologically with some patients even returning to normal but most having long-term residual neurologic deficits.
Neurologic injuries are commonly associated with CCD but can vary widely and can range from quadriplegia with lack of respiratory drive (“pentaplegia”) to a perplexing variety of incomplete cervicomedullary injury syndromes and isolated cranial nerve injuries, such as the cruciate paralysis of Bell and the Wallenberg syndrome.
Vascular injuries are not infrequent with upper cervical spine trauma, although the incidence remains unclear and depends on the diagnostic modalities used. The prospect of concurrent vascular damage commonly dictates CT or MRI-based angiography. Vertebral artery disruption should be considered in any distractive upper cervical spine injury, such as CCD. Lesions include vasospasm, intimal tears, thrombosis, dissection, and pseudoaneurysmal dilatation. In a retrospective study of 29 patients having CCD managed operatively, 15 patients (52%) had 30 blunt cerebrovascular injuries, including 16 vertebral artery and 14 carotid injuries. Three of the 15 had a stroke. Thus, although there is minimal literature with respect to vascular injuries, these are high-mechanism, distractive injuries and appear to have a relatively high rate of vertebral artery and carotid artery injuries. It is therefore ideal to try to obtain either a CT angiogram or other vascular study in these patients.
Associated injuries are common in patients with CCD, including subaxial cervical spine and axis fractures in up to 50% of cases. Neurologic injuries occur in 70% to 100% of survivors, including incomplete and complete spinal cord injury, TBI, Wallenberg syndrome, and cranial nerve injury. Cranial nerve injuries have been reported to include V, VI, VII, IX, X, XI, and XII.
The emphasis on initial management is clearly focused on assuring the best possible chance for patient survival. The Advanced Trauma Life Support (ATLS) principles remain unchallenged in their role of following a principled resuscitation and diagnostic pathway. After establishing vital functions, particularly of the airway, efforts are directed at timely injury recognition and providing protection of the cervical spine. For a patient with a diagnosed CCD, any nonoperative care initially provides temporary stabilization as a bridging measure until definitive care can be rendered. The patient’s head should be immobilized using sandbags, tape, or special head holders while radiographic evaluation is being completed. Skeletal traction is to be avoided. Realignment with a halo vest has been suggested as preferable for acute temporary stabilization but may be inadequate to immobilize severe instability and may have an undesired distractive effect. Accompanying resuscitation efforts include vasopressor support for suspected neurogenic shock and emergent assessment for potential intracranial trauma.
A halo vest can also be considered in adult patients with minimal instability (Stage I CCD) and Stage II CCD who have a negative traction test result. Closed reduction and external immobilization of unstable AODs generally does not lead to a satisfactory outcome because these are typically ligamentous injuries with minimal healing potential. Anatomic alignment is also difficult to maintain over time.
Unlike adults, children seem to have a greater inherent capacity to achieve a stable atlanto-occipital segment through a fibrous ankylosis with nonoperative care. Nonsurgical management in children has the advantage of avoiding surgical injury to the growth centers and disruption of normal craniocervical junction development. More recently, authors have recommended early surgery because of concern for the failure of any nonsurgical measures to provide sufficient stability to this inherently unstable region. van de Pol and colleagues reported on a patient with AOD who sustained a recurrent dislocation while in a halo vest. Halo vest wear can aggravate respiratory compromise in a susceptible patient. There continues to be considerable controversy as to optimal management in the pediatric population, partly driven by the variable degree of instability.
Surgical craniocervical stabilization is indicated for all patients with an atlanto-occipital joint displacement of greater than 2 mm on static imaging studies or with provocative traction testing or in the presence of neurologic injury. In the context of a polytraumatized patient, stabilization is performed as soon as medically possible to prevent further neurologic deterioration.
Manual inline traction, awake fiberoptic intubation, and transnasal intubation are recommended as adjuvant techniques to establish formal airway access for patients with a known CCD while minimizing the potential for secondary injury displacement. An unstable upper cervical spine fracture-dislocation requires atraumatic endotracheal airway access with minimal manipulation. Awake fiberoptic intubation and positioning of a patient allows for clinical neurologic monitoring.
Premature extubation can lead to airway obstruction and a need for emergent reintubation. Assessment of airway swelling before postoperative extubation and a low threshold for delaying extubation until swelling has diminished are important early postoperative management issues. Temporary loss of a patient’s gag reflex should also be taken into consideration in the initial postoperative phase as a means of minimizing the risk of aspiration.
Generally, these patients should have a mean arterial pressure maintained above 85 mm Hg to maintain cord perfusion, particularly in those with known spinal cord injuries.
Electrophysiologic neuromonitoring can be used as an alternative to awake positioning and frequently is the only option because many of these patients arrive intubated, sedated, and in critical condition. Prepositioning signals with motor evoked potentials (MEPs) and somatosensory evoked potentials (SSEPs) are obtained for baseline analysis and followed by repeat studies immediately after positioning. Neuromonitoring is then continued throughout the operative case. These are highly unstable injuries; therefore, any change in signal should prompt an assessment to confirm no change in alignment.
Prone positioning of the intubated patient is performed with Mayfield tongs or a halo on an operating table suitable for spine surgery, allowing full image-intensifier access. The authors’ preferred table is the prone Jackson table. Prone positioning should be performed cautiously, and fluoroscopy should be available to immediately assess alignment. Vertical distraction and subluxation may occur after prone positioning. Reverse Trendelenburg positioning should be avoided because it produces a distraction force between the fixed cranium and the cervical spine.
The head should be positioned in neutral alignment and held rigidly. Sagittal plane alignment should be checked by calculation of the occipitocervical angle. This is the angle formed between the McGregor line (hard palate to occiput) and the inferior end plate of C2. Normal occipitocervical angles range from 10 to 20 degrees. The simpler, as yet scientifically unvalidated, seemingly reliable technique used at the authors’ institution is to align the posterior angle of the mandible with the anterior cortex of C2 ( Fig. 33A-12 ). This has proven to be a successful technique with good outcomes and no cases of significant occipitocervical flexion-extension malalignment. Malalignment of the craniocervical junction can result in airway obstruction, dysphagia, dysphonia, vascular injuries, and difficulty viewing the horizon or the ground.
Approach and Technique
Craniocervical dissociations are most effectively treated through a posterior approach. Following a midline longitudinal incision, the midline intermuscular plane is developed, allowing subperiosteal exposure of the posterior elements. The incision is extended rostrally to the inion to expose the occiput. Care must be taken in working along the occipital–C1 junction as well as the C1–C2 junction to avoid inadvertent durotomy. The atlas should be dissected in a subperiosteal plane, keeping in mind the course of the vertebral artery on the superior aspect of the posterolateral arch. The large, bifid spinous process of the axis is a helpful orientation aid during the early dissection. The C2–C3 interspinous ligament should be preserved if the intended fusion will not extend below C2.
If screw fixation of the axis with pedicle, pars, or transarticular screws is desired, visualization of the superior and medial walls of the C2 pedicles as a reference point is recommended, which requires dissection of the atlantoaxial membrane off the superior lamina of the axis. Exposure of the C1–C2 facet joints may be necessary for a formal arthrodesis of this motion segment, for instance, in the absence of an intact posterior arch of C1. This dissection can result in considerable hemorrhage if meticulous care is not used because of the overlying extensive epidural venous plexus. To facilitate exposure, the C2 nerve root is reflected cranially. When denuding or decorticating the atlantoaxial joint, the vertebral artery’s course immediately lateral to the joint should be taken into account.
Current constructs bridge the occipitocervical junction via contoured rods that connect to an independently placed occipital plate secured with locking screws. An occipital plate can be applied to the midline of the occiput, where the thick midline keel provides the greatest resistance to pullout ( Fig. 33A-13 ). The plate is then attached independently to the atlantoaxial screws with a contoured rod. This stepwise mode of instrumentation increases the ease of occipitocervical instrumentation and provides a powerful tool for reduction and manipulation.
A complete understanding of occipital anatomy is essential to optimize screw safety and fixation. The thickest occipital portion is typically located in the midline at the superior nuchal line, and it has been reported to measure up to 17.5 ± 3 mm. Drilling above the level of the inion should be avoided to avoid hardware prominence and possible injury to the transverse sinus or its confluence with the sagittal sinus with potentially fatal consequences to the patient.
C1 Screw Options
C1 screw fixation is not always necessary in the management of CCDs because the C1 vertebra can often be spanned with instrumentation crossing from the occiput to C2. However, C1 screws can provide additional stability and routine use is the authors’ preference. The C1 lateral mass screw placement as originally described by Goel and colleagues and modified by Harms and Melcher uses a starting point on the posterior aspect of the C1 lateral mass proper, caudal to the prominence where the posterior arch meets the lateral mass ( Fig. 33A-14 ). Access to this starting point requires dissection through the extensive overlying venous plexus, which may lead to problematic bleeding, and requires retraction or transection of the C2 root, either of which may result in occipital numbness and dysesthesias. The authors prefer a C1 screw starting point that is somewhat more rostrally located, at the more readily accessible junction of the posterior arch and lateral mass, which minimizes the need for dissection through the previously mentioned venous plexus and the likelihood of injury to the C2 root ( Fig. 33A-15 ). This modification is not a viable option in all patients, and individual anatomy must be assessed before surgery on CT parasagittal imaging.
Blunt dissection of the C1 posterior arch is carried laterally to the junction of the arch with the lateral mass. This dissection is performed in a strictly subperiosteal manner over the superior aspect of the arch to avoid injury to the vertebral artery. Before placement of instrumentation, one can typically palpate the medial and lateral borders of the lateral mass and the C1–C2 joint, allowing for safe placement of a lateral mass screw. For orientation purposes, the exposure of the pars interarticularis of C2 also helps serve as a general guide to the appropriate starting point for C1 lateral mass screw placement.
A true lateral fluoroscopic view of C1 is then obtained and used to guide a bicortical channel with a drill bit starting approximately 2 mm lateral to the junction of the lateral mass with the posterior arch of C1, generally just medial to where the posterior arch narrows at the vertebral artery sulcus. Creating a bony concavity with a burr is advisable to prevent the drill bit from migrating along this relatively narrow and convex bony ridge. The drill is directed bicortically in the true sagittal plane toward the middle of the anterior margin of C1 on lateral fluoroscopy. This starting point and trajectory help avoid the vertebral artery foramen, the spinal canal, and the atlanto-occipital joint. Rocha and colleagues reported that the width of the C1 lateral mass ranges from 7.7 to 12.8 mm. In general, excellent purchase can be achieved, even in elderly patients with osteoporosis, with screws measuring 22 to 30 mm in length. A bicortical 3.5- or 4.0-mm screw is then placed.
Another potential technique to gain purchase to C1 is with transarticular screws, which allow for fixation to C1 and C2 with the same screw. The nuances of this technique are described in the section on C2 screw options.
C2 Screw Options
Recent advances in anatomic understanding and modern day instrumentation now allow for four primary types of screw fixation to C2: (1) transarticular screws, (2) pedicle screws, (3) pars screws, and (4) translaminar screws. With the advent of many C2 options, the surgeon can tailor the choice based on the patient’s anatomy as interpreted on CT scan to use the safest technique and minimize screw-related complications.
Transarticular C1–C2 screws offer a stiff form of atlantoaxial stabilization complex and were the first screw-based fixation strategy developed for instrumentation of the atlantoaxial complex. This procedure is technically challenging and requires congruous atlantoaxial joint reduction. The presence of anatomic variants, such as the medial vertebral artery coursing across the C2 segment or skeletal dysplasia, can pose significant obstacles to safe completion of this procedure. Other complications include injury to the ICA and the hypoglossal nerve. Because both the hypoglossal nerve and the ICA lie anterior to the lateral portion of the C1–C2 facet joint, they are at risk of penetration with anterior screw insertion. Careful evaluation of the preoperative CT scan is critical to determine the suitability of this technique. Furthermore, the necessary screw trajectory may be blocked by patient body habitus or by the head and neck position required for acceptable fracture alignment. If safe placement of transarticular screws appears doubtful, then other C2 fixation techniques should be considered.
To prevent spinal canal penetration, the medial wall of the isthmus of the axis is visualized and palpated with a neural elevator. Two small paramedial incisions are then made at the cervicothoracic junction to allow the appropriate trajectory of percutaneous drilling and screw placement through a cannulated obturator and drill guide. The starting point for transarticular screws is located in the medial to central third of the inferior articular process of the axis. Drilling with a long Steinmann pin or guide wire for a cannulated screw system is then performed under lateral C-arm guidance, with a 45- to 60-degree vertical inclination trajectory aiming for the mid to upper third of the anterior tubercle of the atlas. Intraarticular passage of the drill or guide wire can be ascertained by direct inspection of the joint. A medial angulation of 0 to 15 degrees is desirable to achieve optimal C1 lateral mass purchase while avoiding an excessively lateral course that may result in vertebral artery and hypoglossal nerve injury ( Figs. 33A-16 and 33A-17 ). Hypoglossal nerve injury can also be avoided by minimizing the extent to which the drill tip penetrates the anterior cortex of the C1 lateral mass or the placement of excessively long screws, the avoidance of which also prevents ICA injury. Cadaveric and radiographic studies have shown that the anterior cortex of the lateral mass becomes engaged when the screw tip lies an average of approximately 6 to 7 mm posterior to the anterior tip of the arch of C1 on lateral radiographs. If a vertebral artery injury is suspected with the placement of the first screw, a contralateral transarticular screw should not be placed.
C2 “pars” screw and “pedicle” screw terminology are frequently used interchangeably, but in reality, the trajectories are quite different, and many screw placements are a combination of pars and pedicle screw with respect to anatomic placement. In general, men have larger C2 pedicle dimensions than women, and men are more likely to safely accommodate C2 pedicle screws. The trajectory of the C2 pedicle screw is much less cephalad and more medially angulated than the transarticular screw ( Fig. 33A-18 ). However, not all C2 pedicles have sufficient bone stock for screw placement with 9% of patients reportedly having inadequate anatomy that precludes the safe placement of C2 pedicle screws. The C2 pedicle screw tends to be less technically demanding than a transarticular screw and does not require anatomic C1–C2 alignment before placement.
Complications can be minimized by meticulously coagulating the venous plexus above the C2 pedicle and inferior to the C2 nerve root and by medially palpating the pedicle wall. The drill or Kirschner wire is aimed in line with the directly visualized pedicle, and the medial cortex is protected to allow safe drilling and screw placement.
C2 pars screws follow a different trajectory than C2 pedicle screws. The trajectory is quite similar to the transarticular screw with a low starting point and minimal medial angulation. The screw tip is intended to end superior to the transverse foramen above the vertebral artery and stop short of the C1–C2 joint ( Fig. 33A-19 ). The C2 pars screw typically averages 24 to 28 mm in length and is the authors’ preferred technique for C2 fixation. The pars screw may be used when the anatomy of the pedicle or the vertebral artery precludes safe placement of transarticular or C2 pedicle screws. For instance, in the case of an unfavorably large and medial vertebral artery foramen, a shorter pars screw can be placed that stops short of the posterior margin of the vertebral foramen.
When placing C2 pedicle and pars screws, the C2 lateral mass is cleared of soft tissue to its lateral border and rostrally up onto the pars. A blunt elevator is used to palpate along the medial border of the pedicle to visualize the trajectory of the screw and to ensure that there is no medial cortical violation. A C-arm is helpful in determining the depth of drilling and in achieving an optimal angle. A more medial trajectory is typically required than that used with transarticular screws—approximately 20 to 25 degrees for C2 pedicle screws versus approximately 10 degrees for pars interarticularis screws. Anticipated screw lengths range between approximately 22 and 30 mm.
The translaminar technique for C2 fixation in which screws are placed in the lamina of C2 starting from the contralateral spinolaminar junction can be used as an alternative to the previously described methods ( Fig. 33A-20 ). The translaminar technique is a viable option for C2 fixation, particularly in patients with anatomy unsuitable for C2 pedicle screws or transarticular screws or in patients in whom there is already a vertebral artery injury. The C2 translaminar construct has proved to be biomechanically sound in stabilizing the atlantoaxial joints. The C2 lamina is the largest in the cervical spine and all elements at risk are visualized directly during insertion, which allows for relatively safe placement.
The first step in C2 translaminar screw fixation is to dissect the lamina. Preoperative imaging is essential to assess dimensions, optimal placement, and anticipated screw lengths. The entry point is located lateral to the spinous process at the spinolaminar junction. A blunt dissector is placed along the medial wall of the lamina in a trajectory aimed toward the lateral mass between the anterior and posterior cortices of the lamina. Although not essential, subperiosteal dissection along the caudal margin of the C2 lamina allows for palpation of the anterior cortex of the lamina to assist with screw trajectory and identify anterior cortical penetration of a drill bit or screw into the spinal canal. Anticipated screw lengths measure 25 to 35 mm. Careful planning is required to avoid screw abutment in the posterior aspect at the junction of the laminae near the spinous process. Ideally, one screw is inserted inferior and is aimed slightly superiorly, and the contralateral screw is begun more superiorly and is aimed slightly inferiorly into the lamina.
Early techniques that used stand-alone onlay bone grafting progressed to include posterior wiring, which provided stability. Pseudarthrosis rates of up to 23% have been reported with onlay grafting and nonrigid fixation (wiring) methods. An 89% fusion rate has been reported using onlay grafting alone, which eliminates hardware-associated complications but requires the use of aggressive postoperative immobilization techniques, including recumbency and skull-tong traction, a Minerva jacket, and a halo. Onlay structural autograft with cerclage wire alone resulted in comparatively improved fusion rates but had the disadvantage of requiring more comprehensive postoperative external immobilization. It was complicated by wire breakage in 78% of patients and late fracture of the graft in up to 15% of patients. The next advance was the development of semirigid fixation using rod-and-wire techniques. In this construct, contoured U-shaped rods, which substituted for the onlay bone graft, were secured to the occiput with threaded wires and linked to sublaminar wires in the suboccipital spine. Wiring techniques have given way to the mechanically superior screw–plate constructs described earlier.
Typically in today’s era, cables are used less frequently because rod–screw constructs are more stable and allow for more reliable fusions. Cables may still be indicated in very young children who are not suitable for screw fixation. Cables can, however, be helpful to aid in securing structural allografts or autografts and to add additional stability.
Bone Graft Options
The authors’ preferred approach is to combine a rigid posterior segmental fixation construct with a structural tricortical iliac crest allograft or autograft that is secured to the occiput and the upper cervical spine as an adjunct to the internal fixation ( Fig. 33A-21 ). The bone graft is attached to the craniocervical junction after decortication and placement of rigid internal fixation devices. This graft is fashioned to cradle the occiput and straddle the C2 spinous process. Allograft extenders or morselized autograft can aid in fusion as well as fill the voids around the structural graft. In a recent review of 48 patients with operatively managed CCDs, there were no cases of pseudoarthrosis using structural grafts in concert with allograft extenders and local autograft.
Reduction and Postoperative Care
After all screws and the occipital plate have been placed, it is imperative to make sure that the occipital–C1 joints and C1–C2 joints are reduced ( Fig. 33A-22 ). This can be difficult to ascertain with C-arm imaging alone, but it is imperative to restore physiologic alignment. One must also ensure appropriate upper cervical flexion-extension alignment as has been previously discussed.
The authors tend to obtain a postoperative CT scan to ensure accurate screw placement and to ensure reduction. Typically, most patients with reasonable bone quality can be managed postoperatively in either a rigid collar or with no immobilization at all. External immobilization should rarely be required beyond 2 to 3 months after surgery. When discontinuing external immobilization, regardless of treatment form, the stability of the injury should be reassessed with flexion–extension and open-mouth odontoid radiographs.
Outcomes and Complications
Most occipitocervical dissociations are fatal. The outcome of survivors depends on (1) the type and severity of associated injuries, particularly closed head injuries; (2) the severity of neurologic injury; and (3) the timeliness with which the diagnosis of AOD is recognized and can be operatively stabilized. In Bellabarba and colleagues’ initial case series, 13 of 17 (76%) patients had at least a 24-hour delay in diagnosis, and five (38%) of these patients deteriorated neurologically before diagnosis compared with no deterioration in the four patients who had been diagnosed without delay. In the follow-up to this series, the delay in diagnosis was decreased to only five of 31 (26%), yet the one patient who had a neurologic deterioration was the only one in whom there had been a delay in diagnosis of over 48 hours. Although the improvement in diagnostic delays (26% vs. 76%) resulted in a much lower missed injury rate (3% vs. 29%), long-term neurologic outcomes were more dependent on the severity of the presenting neurologic injury, which was much worse in the latter group, and suggests that more severely neurologically impaired patients are surviving AODs in greater numbers.
Logic dictates that survivors have less displaced or even spontaneously reduced injuries, and neurologic deficits in survivors are likely to be less severe. Partly because of these reasons and despite substantial advances in neuroimaging, occipitocervical dissociations continue to be frequently missed. Early recognition and timely fixation of these injuries improves outcome by protecting against neurologic deterioration. Delayed diagnosis in these highly unstable injuries has been associated with secondary neurologic deterioration and possibly death in up to 75% of patients. These unacceptably high numbers underscore the importance of improving our current cervical spine trauma screening measures.
Although many of these patients have significant neurologic deficits at admission, the prognosis is better than many other neurologic injuries. In a series of 17 patients, the American Spinal Injury Association (ASIA) motor score improved from 43 to 79, and the number of patients with useful motor function (ASIA grade D or E) increased from 7 (41%) before surgery to 13 (76%) after surgery. With ever increasing improvement in care starting at the scene of the trauma, more patients are surviving injuries that historically certainly would have been fatal. This has led to patients with severe injuries surviving yet with devastating neurologic injuries, adding a whole new level of ethical challenges as well ( Fig. 33A-23 ).
Craniocervical injuries are caused by high-energy trauma. Careful assessment is required to avoid delayed or missed diagnosis. Unfortunately, these remain common and can result in significant morbidity or death. The best diagnostic test is the CT scan with sagittal and coronal reformats. Occipital condyle fractures are mostly stable and can be treated with a cervical orthosis. The avulsion type (type III) injuries may be associated with atlanto-occipital instability requiring surgical treatment. Craniocervical instability or dissociation is classified by the amount of displacement. Patients with small amounts of displacement (<2 mm) that are stable to traction may be treated nonoperatively. Patients with more than 2 mm displacement or atlanto-occipital dislocations should undergo posterior occipitocervical fusion with instrumentation.
Great strides have been made in the reconstruct stability at the occipitocervical junction. Modern rigid craniocervical fusion techniques using screw and plate constructs with structural graft has nearly resolved the issue of pseudarthroses and loss of fixation associated with wire contructs. Potential technical problems include malreduction, which may result in neurologic worsening, and possible penetration of the inner cortex of the skull, which can lead to injury to neural or vascular structures. The bigger challenge in treating survivors of craniocervical dissociation lies in recognizing their often radiographically subtle yet highly unstable injuries and maintaining sufficient stability to protect neurologic function during the initial preoperative treatment phase, which includes the requisite resuscitation and multisystem evaluation in these universally polytraumatized patients.
The level of evidence (LOE) is determined according to the criteria provided in the preface.
Atlas Fractures and Atlantoaxial Injuries
Introduction: Scope and Purpose
The atlas (C1) has a unique anatomy and as a result, its injuries are considered separately from those of other cervical vertebrae. Similarly the relationship between the atlantoaxial articulation differs from the levels caudal to it in the subaxial spine, so the effects of injury to this area will also be considered in this chapter.
Injuries to the C1 ring are frequently associated with injury at the cephalad or caudal levels. Thus, whenever an injury is noted, one must carefully assess the occipitocervical junction, the axis, and subaxial vertebrae. When concomitant injuries occur, it is often the injury to those adjacent levels that dictates treatment. For example, a fracture of the posterior arch of the atlas is often seen with odontoid fracture, but it is the location and displacement of the odontoid fracture that would determine whether operative or nonoperative treatment is warranted. The fracture of the posterior arch of the atlas would only play a role in operative decision making because the posterior arch would not be available as a point of fixation. This chapter will discuss injuries to the C1 ring and atlantoaxial instability.
The atlantoaxial articulation plays an important role in rotational mobility of the cervical spine and is subject to rotary dislocations, which will be covered here. Also the ligamentous support, in particular the transverse ligament, is critical in maintaining stability as the C1 ring rotates about the odontoid process. Injury to the transverse ligament can result in abnormal anterior atlantoaxial translation putting the spinal cord at risk. Injury to the transverse ligament can occur in isolation or in conjunction with atlas fractures, although traumatic rupture of the transverse ligament leading to atlantoaxial instability is rare. These injuries are commonly fatal, but, if the patient survives there is almost always some degree of spinal cord injury. They represent a diagnostic challenge, even postmortem when minimal signs of trauma are evident. These injuries are more frequently seen in older patients with posttraumatic instability developing in the fifth decade of life and beyond.
Mechanism of Injury and Biomechanics
Most upper cervical injuries are the result of automobile accidents or falls. The predominant mechanism of injury is usually forced flexion or extension secondary to unrestrained deceleration forces and from cranial impaction resulting in axial loading to the atlas.
Jefferson initially described the bursting type of atlas fracture, which is an axial loading injury. With an axially directed force, the occipital condyles are driven into superior articular facets of the atlas. Because of the oblique orientation of the facets, as the occipital condyles are driven caudally, the lateral masses of the C1 ring are forced outward, creating anterior and posterior ring fractures and allowing those lateral masses to displace ( Fig. 33B-1 ). Ivancic has confirmed the mechanism of atlas injury in a cadaveric head-first impact model, which demonstrated radial forces producing a burst fracture, with the location of the fracture (anterior or posterior arch) partially being dependent on the relative cross-sectional dimensions of the bone.
Posterior arch fractures of the atlas are often seen in the elderly with concomitant odontoid fractures, occurring with an extension mechanism as the patient falls and strikes the head. Similarly, horizontal fractures of the C1 arch occur from hyperextension injuries and often occur with C2 fractures. Another extension injury is the “plough fracture” where the dens ploughs through the anterior atlantal arch because of forced hyperextension. This unstable fracture usually has an associated posterior atlantoaxial dislocation.
Traumatic rupture of the transverse ligament leading to atlantoaxial instability is caused by high-energy injuries typically resulting from hyperflexion of the neck. This ligament is rigid and tends to experience rupture either centrally or at its lateral insertion as a result of a rapid force application. This is thought to result from a violent injury and is often fatal. It can occur in combination with an atlas fracture. If the transverse ligament is disrupted by the separation of the lateral masses, then the stability of the atlantoaxial relationship is lost. Spence and colleagues defined radiographic parameters as an indirect measure to determine the presence of a disrupted transverse ligament in a cadaveric model. Disruption of the transverse ligament that occurs in isolation, unaccompanied by an atlas ring fracture, is generally the result of a flexion mechanism giving rise to an instability pattern with flexion of the neck, but which reduces in extension.
Injuries to the atlantoaxial joint can be the result of rotational, distraction, or flexion-extension forces, and often also involve the transverse atlantal ligament. Distraction injuries are highly unstable and can occur in isolation or in association with occipito-cervical dissociations. These distraction injuries can be described by their mechanism or direction, similarly to the description of atlanto-occipital injuries, with vertical distraction being the least well described. Vertical distraction injuries have been reported after cervical traction, likely related to undiagnosed ligamentous injuries. Posterior dislocation injuries are hypothesized to be the result of hyperextension and are often accompanied by an odontoid fracture, although several case reports of isolated posterior atlantoaxial dislocation have been reported. The atlas “plough fracture” as described earlier is an anterior arch fracture with posterior atlantoaxial dislocation.
A rotational mechanism can result in either a subluxation or dislocation of the atlantoaxial joints as they pivot around an intact odontoid process. Rotatory injuries of the atlantoaxial joint were first described by Corner in 1907. Rotatory injuries, which usually result from flexion and rotation, can occur with or without a concomitant transverse ligament tear. At approximately 65 degrees of atlantoaxial rotation, ensuing narrowing of the neural canal and subsequent potential damage to the spinal cord usually occur when the transverse ligament is intact. Alternatively, with a deficient transverse ligament, complete unilateral facet dislocation can occur at approximately 45 degrees of rotation with similar consequences. In addition to spinal cord impingement, the vertebral arteries can be compromised by excessive rotation with resultant brainstem or cerebellar infarction and death. Rotatory dislocations at the C1-C2 articulation rarely occur in adults and are significantly different from those in children. Subluxations in children are more frequent, rarely involve severe neurologic manifestations, are related to a viral illness or minor trauma, are almost always self-limited, and generally resolve with conservative treatment. Additionally, the adult form is frequently associated with fracture of a portion of one or both lateral masses and results from a higher-energy trauma (this chapter will concentrate on the adult injury).
Spontaneous atlantoaxial dislocation has been described in patients with preexisting occipito-cervical abnormalities, such as in a male with ankylosing spondylitis, prior iatrogenic destabilization, chronic posttraumatic destabilization, congenital malformations, or rheumatoid arthritis. These have been classified by Xu and colleagues. Further discussion of these nontraumatic lesions, the treatment of which depends highly on their reducibility or lack of reducibility, will not be covered.
These injuries present a diagnostic difficulty because of sometimes subtle neurologic and radiologic findings. Alker and colleagues noted the devastating nature of these injuries in their analysis of 312 victims of fatal traffic accidents, with 24.4% showing evidence of injury to the cervical spine, mostly involving the cervicocranium. Bohlman, in his analysis of 300 patients who sustained acute cervical spine injuries, noted that the correct diagnosis was missed in one-third of patients. Of these injuries, 30% involved the cervicocranium. The presence of a head injury, decreased level of consciousness, alcohol intoxication, multiple injuries, and inadequate radiographs were the extenuating factors that led to delay in diagnosis. With current imaging technology and the use of computed tomography (CT) as the primary screening tool, missed injury should be a less frequent occurrence.
The examination of patients with atlas or atlantoaxial joint injuries is similar to that of any other cervical injuries. Isolated fractures of the ring of the atlas are rarely associated with neurologic deficit; because of wide available space, the tendency of atlas fractures is to radially expand creating even more space. The most common complaint is pain, with neurologic complaints less frequent. The neck pain is usually nonspecific but could localize to the area around the mastoid or referred to the occiput. Specific neurologic complaints unique to these injuries may represent C2 root damage with paresthesia or numbness in the area of the greater occipital nerve. Cranial nerve palsies of the lower six cranial nerves, and injury to the vertebral arteries as they cross the posterior atlantal arch, can also occur. With a rotary atlantoaxial injury, torticollis may be present. Torticollis or “cock robin” appearance could also signify a significant unilateral C1 ring fracture with enough displacement for the occipital condyle to settle caudally onto the superior facet of C2 as the lateral mass of C1 displaces outwardly. This appearance can be present from the initial injury or develop over time as a fracture gradually displaces; thus, the patient should be observed in all subsequent evaluations if nonoperative treatment is chosen. Because of the subtlety of these findings, the diagnosis may be delayed after the initial trauma. Although rare, catastrophic neurologic deterioration as a result of compromise of the neural canal at the cervicomedullary junction has been reported and mandates a high index of suspicion in trauma patients with neck injuries. In traumatic rupture of the transverse ligament, survivors from this usually fatal injury may have a clinical picture ranging from a dense, mixed neurologic deficit to only severe upper neck pain. Neurologic signs may be asymmetric. Additionally, because the vertebral artery is intimately associated with the C2 and C1 vertebrae as it works its way through the foramen transversarium, it is prone to injury. The examination should include an assessment for any evidence of vertebral artery distribution stroke, such as vertigo, dizziness, blurred vision, and nystagmus.
Radiographic analysis can be misleading to physicians not familiar with these injuries because of the complex anatomy and the normal atlantoaxial hypermobility. CT is the mainstay of radiographic evaluation. Because the anatomy of the upper cervical spine is unique, one must be careful to look specifically for the various fracture patterns. The CT axial plane images are usually oriented perpendicular in the midportion of the subaxial spine, which makes them oblique and distorted in the upper cervical spine. Similarly, the coronal reconstructions are usually aligned with the midportion of the subaxial spine making it difficult to visualize the relationship between the occiput, C1, and C2 in the coronal plane. If the axial or coronal cuts are too oblique to fully evaluate the upper cervical region, then one should request that the axials be reconstructed to run in line with the C1 ring and the coronals to run parallel to the odontoid ( Fig. 33B-2 ). Positioning any patient with the head turned will result in physiologic rotational displacement of the atlantoaxial articulation, which may be difficult to distinguish from injury.
Atlantoaxial rotary subluxation or dislocation can be evaluated by several techniques. Pang and Li were able to define normal rotatory C1-C2 movement in children based on CT, although the applicability for adults is uncertain. Wortzman and Dewar suggested a dynamic method of differentiating rotatory fixation from torticollis by using plain radiographs, and Fielding and coworkers suggested cineradiography and CT as additional tools for evaluating these injuries. On an open-mouth odontoid radiograph and with left and right rotation of the atlantoaxial joint, a so-called wink sign may be seen. When plain cervical spine radiographs demonstrate evidence of a rotational anomaly at the atlantoaxial joint, additional radiographic investigation is indicated and should consist of open-mouth odontoid views with the patient’s head rotated 15 degrees to each side to determine whether true atlantoaxial fixation is present. Persistent asymmetry of the odontoid and its relationship to the articular masses of the atlas, with the asymmetry not being correctable by rotation, forms the basic radiologic criteria for the diagnosis of atlantoaxial rotatory fixation. Additionally, cineradiography in the lateral projection may be considered if it is available. Alternatively, a CT through the C1-C2 articulation with the patient’s head rotated to the right and to the left approximately 15 degrees confirms or disproves the presence or absence of rotatory fixation at the atlantoaxial joint. Most commonly, acute or chronic traumatic injuries are in a fixed position. Currently, CT with two- or three-dimensional reconstruction gives the most accurate delineation of the injury. One small pediatric case series advocates a CT under general anesthesia to eliminate muscle spasms or guarding.
If a pure transverse ligamentous injury is suspected, supervised flexion-extension radiographs with constant neurologic monitoring have classically been recommended to assess the atlantodens interval (ADI). However, flexion-extension radiographs in a patient with an acutely unstable spine and neurologic deficit are contraindicated. Normally there is less than 3 mm of anterior displacement of C1 on C2, the ADI, which implies that the transverse ligament is intact. It should be kept in mind that a lack of apparent instability may be caused by protective paraspinous muscle spasm. A displacement of 3 to 5 mm is indicative of transverse ligament rupture, and a displacement greater than 5 mm implies probable rupture and functional incompetence of the transverse and accessory ligaments ( Fig. 33B-3 ). Vertical instability can be seen with traumatic ligamentous C1-C2 injuries, and a high index of suspicion should be maintained when attempting manual reduction in these cases. When examining the C1-C2 facet space and vertical translation, Gonzalez and colleagues noted that 95% of healthy individuals have a C1-C2 lateral mass interval (LMI) of between 0.7 and 2.6 mm. The authors concluded that an LMI greater than 2.6 mm should alert the physician to the possibility of a distraction injury.
Magnetic resonance imaging (MRI) is not required for every atlantoaxial injury but has a definite role, especially in determining injury to the transverse atlantal ligament. It can provide information regarding the extent, location, and compressive pathology in those patients with neurologic injury, especially if the fracture pattern does not match the neurologic examination. Special fine-cut axial images may be necessary to assess details of the transverse ligament rupture so it would be prudent to communicate with the radiologist if this is the intent of the MRI study. Coronal sections are also useful especially to evaluate the alar ligaments. Dickman and colleagues recommended MRI as it is the only way to directly assess the transverse atlantal ligament, as opposed to CT and radiographs, which attempt to indirectly detect injury ( Fig. 33B-4 ). Furthermore, an additional retrospective study found no correlation between bony displacement on CT scan and integrity of the transverse atlantal ligament, recommending MRI for clinical concern of ligamentous injury. Tears of the transverse ligament are represented by interruption in the continuity of the ligament signal pattern. These transverse ligament tears have been classified by both Fielding and Dickman. Fielding’s classification takes into account the alar ligaments, while Dickman’s classification accounts for concomitant C1 lateral mass fractures. Children, patients with Down syndrome, and those with rheumatoid arthritis may all exhibit chronic instability with a widened ADI unrelated to trauma.
If the potential exists for vertebral artery injury, then computed tomography angiography (CTA) would be indicated. This aids in diagnosis of patients with symptoms or signs of vertebral artery stroke and may prove valuable in surgical planning for screw placement. Fractures through the foramen transversarium or involving the posterior arch are at high risk for vertebral artery injury.
Diagnosis and Classification
The diagnosis of C1 ring injuries is generally made from the screening CT scan for trauma patients, then classified based on fracture pattern. Common fracture patterns include isolated posterior or anterior ring, unilateral lateral mass with a fracture anterior and one posterior to the lateral mass, a compression of the lateral mass, and the classic Jefferson fracture with fracture lines anterior and posterior to both lateral masses creating four separate components. With the introduction of CT, Segal and associates in 1987 expanded Jefferson’s original fracture classification to include six subtypes, followed by a seventh subtype added by Levine and Edwards :
Burst fractures (33%), as seen in other areas of the spine, are usually the result of an axial loading force. The force is transmitted through the occipital condyles to the medially oriented superior articulations of C1, which are forced apart, and either three- or four-part fractures are produced. Although most common, they are the least likely to cause neurologic injury ( Fig. 33B-5 ). The degree of displacement dictates treatment, as will be discussed shortly.