Radiographic Features

Basilar impression is difficult to assess radiographically. Many measurement schemes have been proposed; those most commonly referred to are the Chamberlain,¹ McGregor,⁷ and McRae lines in the lateral radiograph (Fig. 30–1) and the Fischgold-Metzger line in the anteroposterior projection. The Chamberlain line¹ (a line drawn from the dorsal marginal hard palate to the posterior lip of the foramen magnum) is seldom used because the posterior lip of the foramen magnum (opisthion) is difficult to define on a standard radiograph (Fig. 30–2A) and is often itself invaginated in basilar impressions.

FIGURE 30–1 Lateral craniometry. Drawing indicates three lines used to determine basilar impressions. Chamberlain line (1939) is drawn from posterior lip of foramen magnum (opisthion) to dorsal margin of hard palate. McGregor line (1948) is drawn from upper surface of posterior edge of hard palate to most caudal point of occipital curve of skull. McRae line (1953) defines opening of foramen magnum. McGregor line is best method for screening because bony landmark can be clearly defined at all ages on routine lateral radiograph.

FIGURE 30–2 A 6-year-old girl presented with a history of unusual gait and recent episode of unconsciousness after mild head trauma. A, Routine lateral radiograph suggests that odontoid is displaced proximally into opening of foramen magnum. McGregor line has been drawn from most caudal portion of occiput to hard palate. Tip of odontoid is more than 5 mm above this line, indicating basilar impression. B, Lateral laminagraph shows subluxation of C1 and C2 and that tip of odontoid is above opening of foramen magnum (McRae line).

The McGregor line (a line drawn from the upper surface of the posterior edge of the hard palate to the most caudal point of the occipital curve of the skull) is easier to identify (see Figs. 30-1 and 30-2) and is preferable. The position of the tip of the odontoid is measured in relation to this baseline, and a distance of 4.5 mm above McGregor line is considered to be on the extreme edge of normality.⁷ The study of normal variations by Hinck and colleagues⁸ showed a wide range of normality, however, and difference between males and females.

The McRae line defines the opening of the foramen magnum and is derived from McRae’s observation that if the tip of the odontoid lies below the opening of the foramen magnum, the patient is likely to be asymptomatic. The McRae line is an accurate guide in the clinical assessment of patients with basilar impression.

A criticism of the lateral lines (McGregor and Chamberlain) is that the hard palate is not actually a part of the skull and may be distorted by an abnormal facial configuration or a highly arched palate, independent of a craniovertebral anomaly. In addition, the patient may have an abnormally long or short odontoid or an abnormality of the axis or occipital facets, which can diminish the value of the measurements. Upper cervical spine measurements on plain radiographs have an unacceptable interobserver reliability and intraobserver reproducibility. Although there are multiple factors for the wide variation between observers, many authors have suggested that the basion and opisthion are difficult to localize on radiographs and may be altered significantly by small changes in patient position.⁹

Since the advent of CT and MRI, the diagnosis of basilar invagination can be made directly rather than based on craniometric lines.¹⁰ These techniques have largely replaced the older tomography techniques, particularly the anteroposterior measurement of Fischgold and Metzger,^10,11 which was previously the most accurate method. The mean and range of normal distances from the odontoid peg to the most frequently used skull baseline employing MRI was 1.2 mm (median 1.5 mm, SD 3 mm) below Chamberlain line; 0.9 mm (median 1.1, SD 3 mm) below McGregor line; and 4.6 mm (median 4.8, SD 2.6) below McRae line.¹⁰ MRI also allows direct assessment of the degree of neural compression and identifies the presence of associated pathologic processes, such as Chiari malformation, hydrocephalus, syringomyelia, or cerebellar herniation (Fig. 30–3).⁴ The lateral reference lines on plain radiographs are still very useful screening tools, however.^12,13

FIGURE 30–3 Secondary basilar invagination with osteochondrodysplasia. A, Sagittal drawing shows acute cervicomedullary angle, ventrally compressed brainstem, herniated hindbrain, and invaginated clival-atlantoaxial complex. B, Sagittal MRI shows these key findings in an 11-year-old girl with Hajdu-Cheney syndrome.

(Images adapted from Sawin PD, Menezes AH: Basilar invagination in OI and related osteochondral dysplasias. J Neurosurg 86:950-960, 1997.)

Clinical Features

Patients with basilar impression frequently have a deformity of the skull or neck (e.g., a short neck in 78%, asymmetry of the face or skull or torticollis in 68%).¹⁴ These physical findings are often noted in patients without basilar impression (Klippel-Feil syndrome, occipitalization) and are not considered pathognomonic.

The symptoms (or lack of them) even in severe basilar impression are difficult to explain.^13,14 The more segments involved in a patient with Klippel-Feil syndrome, the more likely it is that the odontoid will be superior.¹³ Basilar impression is frequently associated with anomalous neurologic conditions, such as Arnold-Chiari malformation¹⁴ and syringomyelia,^15,16 which can cloud the clinical picture further. Symptoms are generally due to crowding of the neural structures at the level of the foramen magnum, particularly the medulla oblongata. The neurologic sequelae of basilar invagination are determined by the degree of cephalad migration of the clivus-atlas-dens complex. In addition to the direct compression of the ventral brainstem, the dens creates a fulcrum for traction on the cervical cord. Cerebellar dysfunction may arise from direct compression or reflect secondary herniation from the posterior fossa.

There is an unusually high incidence of basilar impression in northeast Brazil,¹⁷ and the work of De Barros and colleagues¹⁴ has been helpful in delineating the symptoms and signs. Patients who were symptomatic with pure basilar impression had the dominant complaints of motor and sensory disturbances, and 85% had weakness and paresthesia of the limbs. In contrast, patients who were symptomatic with pure Chiari malformation were more likely to have cerebellar and vestibular disturbances (unsteadiness of gait, dizziness, and nystagmus). In both conditions, there may be impingement of the lower cranial nerves as they emerge from the medulla oblongata, particularly the trigeminal (V), glossopharyngeal (IX), vagus (X), and hypoglossal (XII) nerves.¹⁴ Headache and pain in the nape of the neck, in the distribution of the greater occipital nerve, are common findings.^14,18

Posterior encroachment may cause blockage of the aqueduct of Sylvius, and the presenting symptoms may be from increased intracranial pressure or hydrocephalus.^7,14 Compression of the cerebellum with vestibular involvement or herniation of the cerebellar tonsils (Chiari malformation) is a frequent finding,¹⁴ leading to vertical or lateral nystagmus in 65% of cases. These symptoms may be caused not by direct pressure from the posterior rim of the foramen magnum but rather by a thickened band of dura invisible on plain radiographs; this has prompted several authors to recommend routine MRI evaluation. If this situation is unrecognized, bony decompression alone (without opening the dura) would be unsuccessful in obtaining remission of symptoms or halting progression of the neurologic injury.¹⁴

There is a high incidence of vertebral artery anomalies in basilar impression, atlanto-occipital fusion, and absence of the C1 facet.¹⁹ In addition, the vertebral arteries may be compressed as they pass through the crowded foramen magnum, causing symptoms suggestive of vertebral artery insufficiency, such as dizziness, seizures, mental deterioration, and syncope.^7,19,20 These symptoms may occur alone or in combination with symptoms of spinal cord compression.^7,14,19,21 Michie and Clark²² and Bachs and colleagues²³ theorized that one explanation for the frequent association of syringomyelia or syringobulbia and basilar impression is that the vertebral arteries and the anterior spinal artery are compromised in the region of the foramen magnum, with subsequent degeneration of the spinal cord and medulla. Arteriographic studies are unavailable to confirm this thesis. Children with occipitocervical anomalies may be more susceptible to vertebral artery injury and brainstem ischemia, particularly children who undergo skull traction for correction of scoliosis. Moderate amounts of traction (<15 lb) that normally would be well tolerated may compromise these abnormal vessels. Careful radiographic evaluation of the occipitocervical junction should precede any use of skull traction, even if only minimal traction forces are planned.²⁴

Although this condition is congenital, many patients do not develop symptoms until the 2nd or 3rd decade of life.^13,14 These delayed symptoms may be due to a gradually increasing instability from ligamentous laxity caused by aging, similar to the delayed myelopathies reported after atlantoaxial dislocations or the increasing instability of C1 and C2 in patients with odontoid agenesis.²⁵ These individuals often develop premature cervical osteoarthritis, as found in the family studies of Gunderson and colleagues.²¹ Chamberlain¹ and others theorized that the young developing brain may be more tolerant of compressive effects, which later prove deleterious to older tissues. Similarly, arteriosclerotic changes in the vertebral arteries may make these vessels more susceptible to minor constrictions. Symptoms frequently occur in older patients in whom a congenital anomaly would not ordinarily be considered. Patients with this malformation have been mistakenly diagnosed as having multiple sclerosis, posterior fossa tumors, amyotrophic lateral sclerosis, or traumatic injury. It is important to survey this area whenever such a diagnosis is considered and whenever this malformation is suspected.

Treatment

Treatment depends on the cause of the symptoms and often requires the teamwork of an orthopaedist, neurosurgeon, neurologist, and radiologist. It is possible to have a severe basilar impression without neurologic symptoms, and a search for associated conditions must be conducted. Hydrocephalus from aqueductal stenosis, if present, must be addressed first with ventricular shunting before any other surgical interventions.⁴

Treatment of symptomatic basilar invagination is dictated by the severity of ventral neural compression and reducibility of the deformity. An initial attempt should be made to realign the cervical spine and decompress the neural elements by head positioning and, in some cases, halo traction. Plain radiographs and MRI may be used to determine the adequacy of reduction and neural decompression. Patients who respond to such closed reduction methods may be treated with suboccipital craniectomy and upper cervical laminectomy, followed by occipitocervical arthrodesis. More recent reports suggest that many of the physical changes can be reversed by atlantoaxial distraction fixation surgery.^26,27 Most authors suggest opening the dura to look for a tight posterior dural band.¹⁴ Brainstem compression not relieved by traction demands anterior decompression and resection of the offending anterior atlas, odontoid, or distal clivus. Posterior decompression and fusion are also necessary and may be performed immediately after or as a staged procedure.²⁸

Posterior fossa decompression alone for symptomatic secondary basilar invagination typically provides only transient relief of symptoms, followed by recurrent neurologic deterioration months to years after the index procedure. Studies have shown that secondary basilar invagination in patients with osteochondrodysplasia tends to progress even after occipitocervical fusion with infolding of the skull base around the fusion mass and progressive forward bending.³ Prolonged bracing with an orthotic device and lifelong surveillance may be prudent to halt further recurrence or symptomatic progression.⁴ These recommendations are generalizations regarding treatment, and appropriate references should be consulted in the evaluation of individual patients.

Pathomechanics

The articulation between C1 and C2 is the most mobile part of the vertebral column and normally has the least stability of any of the vertebral articulations. The normal cervical spine permits about 90 degrees of rotatory motion; 50% of this motion occurs in the atlantoaxial joint. Considerable shifting from side to side (lateral slide) also occurs as a component of this rotatory motion. Flexion and extension are permitted to a limited degree (normally about 10 degrees of extension and 5 degrees of forward flexion); more than 10 degrees of flexion indicates subluxation.³⁰ The odontoid acts as a bony buttress to prevent hyperextension, but the remaining normal range of motion is maintained and is solely dependent on the integrity of the surrounding ligaments and capsular structures.

The articulation between the condyles of the skull and the atlas (atlanto-occipital joint) normally allows only a few degrees of flexion-extension—a slight nodding motion of the head. In rotation, the atlas and head turn as a unit. The articulation between the axis and C3 permits some flexion-extension but is similarly restricted in rotation. The atlantoaxial joint is extremely mobile but structurally weak and is located between two relatively fixed points, the atlanto-occipital and C2-3 joints.

Motion of the atlantoaxial articulation is usually accentuated in patients with bony anomalies of the occipitocervical junction. An excellent example is a patient with atlanto-occipital fusion, who is frequently found to have compensatory hypermobility of the atlantoaxial joint. If this same patient has an associated synostosis of C2-3, it is reasonable to expect that this additional stress on the atlantoaxial articulation may eventually lead to significant instability.^28,29,31,32 This assumption has clinical support in that 60% of patients with symptomatic atlanto-occipital fusion have associated fusion of C2-3.^29,32,33

Nonetheless, it is unusual for patients to become symptomatic before the 3rd decade. It must be assumed that at least initially a delicate balance is struck between the hypermobile articulation and the adjacent spinal cord. Motion is maintained without neurologic compromise. This relationship must be altered before symptoms develop. In a symptomatic patient, trauma is immediately suspected, but statistically it is not often associated or is of a minor nature.

More likely, the degenerative changes of aging cause the lower cervical articulations to become more rigid. This gradual restriction of motion below places an increased demand on the ligaments and capsular structures of the atlantoaxial articulation, with the development of instability.³⁴ With aging, the central nervous system itself becomes less tolerant of intermittent compression, and its ability to recover is diminished. There is evidence to suggest that intermittent compression is more harmful or irritating to the spinal cord than static, constant compression.³⁵ Arteriosclerosis and loss of elasticity from aging affect the vertebral arteries, making them more sensitive to compression at the foramen magnum.³⁶

As a consequence, a symptomatic patient often presents with a puzzling clinical picture. Only a few patients present initially with a history of head or neck trauma, neck pain, torticollis, quadriparesis, or signs of high spinal cord compression, any of which would facilitate the diagnosis. A changing, intermittent pattern of symptoms is more typical than the localized pattern suggested by the radiographs alone. Many patients mistakenly receive the diagnosis of a diffuse demyelinating disease such as multiple sclerosis or amyotrophic lateral sclerosis.

In patients with basilar impression or atlanto-occipital fusion, the clinical findings suggest that the major damage is occurring anteriorly from the odontoid. The symptoms and signs of pyramidal tract irritation, muscle weakness and wasting, ataxia, spasticity, hyperreflexia, and pathologic reflexes are commonly found.^33,37 Autopsy findings consistently show that the brainstem is indented by the abnormal odontoid.^37,38 Other, less common complaints include diplegia, tinnitus, earaches, dysphasia, and poor phonation, all of which are due to cranial nerve or bulbar irritation from direct pressure of the odontoid on the medulla.

If the primary area of impingement is posterior from the rim of the foramen magnum, the dural band, or the posterior ring of the atlas (typical of odontoid anomalies), symptoms are referable to the posterior columns with alterations in deep pain and vibratory responses and proprioception. If there is also an associated cerebellar herniation, nystagmus, ataxia, and incoordination may be observed. Symptoms referable to vertebral artery compression—dizziness, seizures, mental deterioration, and syncope—may occur alone or in combination with symptoms of spinal cord compression.³⁴

In children, the presenting symptoms may be quite subtle and nonspecific. Perovic and colleagues³⁹ reported that in most of their patients the only presenting symptom was generalized weakness, manifested as lack of physical endurance, a history of frequent falling, or the child’s asking to be carried. Pyramidal tract signs appeared later, and posterior column signs and sphincter disturbances were less frequently encountered.³⁹ Similarly, in achondroplasia, retardation of motor skills and increasing hydrocephalus may be the only clinical manifestation of impending quadriparesis secondary to basilar stenosis.

Atlantoaxial instability is commonly associated with other anomalies of the spine, many of which lead to scoliosis. Patients with congenital scoliosis, Down syndrome, spondyloepiphyseal dysplasia, osteogenesis imperfecta, and neurofibromatosis all can have significant atlantoaxial instability, which may be unrecognized. Radiologic surveys, particularly flexion-extension views of the C1-2 articulation, should be obtained before administration of general anesthesia or preliminary spinal traction in the management of the scoliosis.

Atlantoaxial Instability and Down Syndrome

Spitzer and colleagues⁴⁰ were the first to describe occipitoatlantal instability and atlantal hypoplasia in Down syndrome. Generalized ligamentous laxity and flat facets predispose to hypermobility and pathologic motion at the craniovertebral and atlantoaxial articulations. Depending on patient age at the time of the study, 10% to 40% of these patients have radiologically detectable atlantoaxial instability.^40,41

The natural history of atlantoaxial instability in patients with Down syndrome is not well known. Burke and colleagues⁴² reported 7 of 32 patients developing atlantoaxial instability over a 13-year period. Morton and colleagues⁴³ found no evidence of de novo atlantoaxial instability in 67 patients with Down syndrome over a 5-year period. In addition, a direct correlation between neurologic deterioration and atlantoaxial instability has not been established. Ferguson and colleagues⁴⁴ identified 17 of 84 patients with atlantoaxial instability on plain radiographs but could not correlate any difference in neurologic symptoms with radiographic findings. Taggard and colleagues⁴⁵ did not find such a benign relationship between C1-2 instability and neurologic function, however, reporting acute neurologic deterioration in six patients after a minor fall or intubation for a general anesthetic.

Radiographic Features

The atlas-dens interval (ADI) is the space seen on the lateral radiograph between the anterior aspect of the dens and the posterior aspect of the anterior ring of the atlas (Fig. 30–4). In children, the ADI should be no greater than 4 mm,³⁰ particularly in flexion, where the greatest distance can be noted. The upper limit of normal in adults is less than 3 mm. Occasionally, the space between the odontoid and the ring of C1 forms a “V” shape in flexion; in this situation, the ADI is measured at the mid-portion of the odontoid. A subtle increase in the ADI in the neutral position may indicate disruption of the transverse atlantal ligament. The ADI is a valuable aid in the evaluation of acute injury, when standard flexion-extension views would be potentially hazardous.^30,46 Fielding and colleagues⁴⁷ noted that the shift of C1-2 does not exceed 3 mm in adults if the transverse ligament is intact. The transverse ligament ruptures within the range of 5 mm.⁴⁸ Similar data for children are unavailable.

FIGURE 30–4 Atlantoaxial joint showing normal atlas-dens interval (ADI), normal space available for spinal cord (SAC), distance between posterior aspect of odontoid, and nearest posterior structure.

The ADI is of limited value in evaluating chronic atlantoaxial instability owing to congenital anomalies, rheumatoid arthritis, or Down syndrome. In these conditions, the odontoid is frequently found to be hypermobile with a widened ADI, particularly in flexion (Fig. 30–5), but not all are symptomatic, and all do not require surgical stabilization.^42,48 In this situation, attention should be directed to the amount of space available for the spinal cord (SAC). Determining the SAC is accomplished by measuring the distance from the posterior aspect of the odontoid or axis to the nearest posterior structure (foramen magnum or posterior ring of the atlas).^29,32,49 This measurement is particularly helpful in evaluating a patient with nonunion of the odontoid or os odontoideum because in both conditions the ADI may be normal, yet in flexion or extension there may be considerable reduction in the space available for the spinal cord (Fig. 30–6).⁵⁰ Lateral flexion-extension views should be conducted voluntarily by the patient, particularly a patient with a neurologic deficit (Fig. 30–7). Most symptomatic patients exhibit significant instability.

FIGURE 30–5 Atlantoaxial instability with intact odontoid. A, Extension—forward sliding of atlas with increased atlas-dens interval (ADI) and decreased space available for spinal cord (SAC). B, Flexion—ADI and SAC return to normal as intact odontoid provides bony block to subluxation in hyperextension.

FIGURE 30–6 Atlantoaxial instability with os odontoideum, absent odontoid, or traumatic nonunion. A, Extension—forward sliding of atlas with reduction of space available for spinal cord (SAC) but no changes in atlas-dens interval (ADI). B, Flexion—posterior subluxation with reduction in SAC and no change in ADI.

FIGURE 30–7 Lateral flexion-extension radiography of os odontoideum. A, Extension. B, Flexion. Odontoid ossicle is fixed to anterior ring of atlas and moves with it in flexion and extension and lateral slide. Space available for spinal cord (SAC) decreases with flexion, and ossicle moves into spinal canal with extension.

Patients with a normal odontoid process and an attenuated or ruptured transverse atlantal ligament are particularly at risk; with anterior shift of the atlas over the axis, the spinal cord is easily damaged by direct impingement against the intact odontoid process,^50,51 such as in atlanto-occipital fusion. The situation is less dangerous if the odontoid process is absent or fractured and is carried forward with the atlas (os odontoideum).⁴⁷ In a large series of patients with os odontoideum reported by Fielding and colleagues,⁴⁷ the average displacement was 1 cm, mostly either anterior or posterior, but some were unstable in all directions.

In patients with multiple anomalies, the usual radiographic views are not always reliable in confirming the presence or absence of an odontoid.³⁹ Similarly, in patients with abnormal bone, such as patients with Morquio syndrome or spondyloepiphyseal dysplasia, the odontoid may be present but dysplastic, blending with the surrounding abnormal bone, and cannot be differentiated. In these situations, adequate visualization can be obtained by using lateral laminagraphic techniques (see Fig. 30–2) or with a CT scan with reconstruction views.^52,53 When the exact cut at the level of the odontoid is determined, it is repeated in flexion-extension to ascertain the stability of the atlantoaxial articulation. Extension views should not be ignored. Many patients have been found to have significant posterior subluxation.^29,47,54 Dynamic (flexion-extension) CT scans have been used to show this area, particularly with lateral reconstruction, and have replaced the laminagraphs.^51–53

Dynamic MRI provides direct vision of the neurologic structures and in many cases the site of bony impingement and cord compression that may not be appreciated on routine views.^52,55–57 CT scans have been extremely helpful in evaluating anomalies of the upper cervical spine. Three-dimensional reconstruction aids in visualizing anatomic relationships. Dynamic CT or MRI flexion and extension views are difficult to perform and time-consuming, but they can provide graphic evidence of looseness.⁵⁸ MRI similarly has been helpful in showing the relationship of the neurologic structures to the bony anatomy, such as the Chiari malformation and syringomyelia.^55,58 Also, the transverse atlantal ligament can be visualized with MRI.⁵⁵

Steel⁵⁹ called attention to the checkrein effect of the alar ligaments and how they form the second line of defense after disruption of the transverse atlantal ligament (Fig. 30–8). This secondary stability no doubt plays an important role in patients with chronic atlantoaxial instability.⁶⁰ Steel’s anatomic studies provided a simple rule that is helpful to physicians evaluating this area. He defined the “rule of thirds”: The area of the vertebral canal at C1 can be divided into one third odontoid; one third spinal cord; and one third “space,” which represents a safe zone in which displacement can occur without neurologic impingement and is roughly equivalent to the transverse diameter of the odontoid (usually 1 cm).⁵⁹

FIGURE 30–8 Atlantoaxial joint as viewed from above. A, Normal. B, Disruption of transverse atlantal ligament (TAL): Odontoid occupies safe zone of Steel. Intact alar ligaments (second line of defense) prevent spinal cord compression.

In chronic atlantoaxial instability, it is of prime importance to recognize when the patient has exceeded the “safe zone” of Steel and enters the area of impending spinal cord compression. At this point, the second line of defense, the alar ligaments, has failed, and there is no longer a margin of safety (see Fig. 30–8).⁴⁸ Experimentally, Fielding and colleagues⁴⁸ found that after rupture of the transverse ligament, the alar ligaments are usually inadequate to prevent further displacement of C1-2 when a force similar to that which ruptured the transverse ligament is applied. Although the alar ligaments appear thick and strong, they stretch with relative ease and permit significant displacement.

McRae³³ was first to call attention to the relationship of neurologic symptoms and the sagittal diameter of the spinal canal (SAC) (see Fig. 30–4). He noted that his patients with atlanto-occipital fusion with less than 10 mm of available space behind the odontoid or atlas were always symptomatic. With the availability of more clinical data, this measurement has been more specifically defined. After an extensive review of the literature, Greenberg³⁴ determined that in adults spinal cord compression always occurs if the sagittal diameter of the cervical canal behind the dens is 14 mm or less. Cord compression is possible between 15 mm and 17 mm and never occurs if the distance is 18 mm or more. Variations in patient size and the presence or absence of soft tissue elements affect the clinical significance of the measurement, however. Jauregui and colleagues⁵⁶ refined these measurements further for children. MRI may be helpful in assessing the integrity of the transverse atlantal ligament.⁵⁵ In some patients, myelography has shown a ventral impingement secondary to a thick wad of radiolucent soft tissue posterior to the dysplastic odontoid and the body of the axis.³⁹

Data on normal variations in sagittal and transverse diameter of the cervical spine have been collected for infants, children, and young adults.^46,61,62 As might be expected, these measurements generally are smaller than in adults and follow a predictable growth curve. Clinical significance is determined via two parameters: (1) comparison of the absolute diameter with the known norms for that vertebral level and age, and (2) comparison of successive vertebral levels within the same individual. Variations in the latter are more sensitive in determining an abnormality involving a single vertebra because there is good correlation between adjacent levels in the same child. These measurements should be readily available when evaluating the growing spine for pathologic narrowing and expansion.⁶³ The transparencies developed by Haworth and Keillor⁶¹ are particularly helpful in this regard and provide an efficient screening test for assessing the transverse diameter of the spinal canal in the growing child. The sagittal diameter of the cervical canal is largest at C1 and gradually narrows in size until C5-7. The appearance is similar to a funnel, and enlargement in the outline suggests an intraspinal mass, even if the absolute measurements do not exceed the upper limits of normal.⁶³

In children, there are several normal variations in cervical spine mobility that can be alarming to physicians who are unaware of them. Cattell and Filtzer⁶⁴ called attention to the frequent (20%) finding of overriding of the anterior arch of the atlas on the odontoid with extension of the neck. This variation is due to normal elasticity of ligaments and diminishes with growth; it is not present after age 7 years. Pseudosubluxation of 3 mm of C2 on C3 can be found in more than one half of children younger than age 8 years (Fig. 30–9).^64–66 Swischuk⁶⁶ suggested a posterior cervical line that is useful in differentiating physiologic from pathologic anterior displacement of C2 on C3. Less frequently, hypermobility occurs at the C3-4 interspace. Recognition becomes particularly important when evaluating a young child who has Klippel-Feil syndrome or has undergone recent trauma.

FIGURE 30–9 A, Pseudosubluxation of C2 on C3 in a child at age 5 years. B, At age 10, flexion view of cervical spine shows normal motion without pseudosubluxation. This normal variation can be found in more than half of children younger than age 8 years.

The normal pattern of ossification of the cervical spine in children can pose problems in radiographic interpretation. At birth, the odontoid is separated from the body of the axis by a wide cartilaginous band, which represents the vestigial disc space, referred to as the neurocentral synchondrosis (see section on anomalies of odontoid). On the lateral radiograph, this lucent line is similar in appearance to an epiphyseal growth plate. This line may be confused with the jointlike articulation between the odontoid and the body of the atlas found in os odontoideum; this is present in nearly all children at age 3 and absent in most by age 6. Consequently, the diagnosis of os odontoideum in children must be confirmed by showing motion between the odontoid and the body of the axis. Atlantoaxial fusion may be difficult to diagnose in a young child because a significant portion of the ring of C1 is unossified at birth. There is usually a 5- to 9-mm gap posteriorly, which ossifies by age 4.³¹ The anterior arch of the atlas is invisible in 80% of infants, and the entire ring may not be completely ossified until age 10 years.^31,53

Myelography can be helpful in defining an area of constriction. In this regard, gas or water-soluble contrast agent (metrizamide) myelography should be used in preference to myelography with oil contrast media (Fig. 30–10).^33,39,67,68 CT in conjunction with metrizamide myelography is particularly helpful in evaluating children with rotational deformities or compromise of the neural canal in the upper cervical spine.^51,67,68

FIGURE 30–10 Normal gas myelography-polytomography. A, A 3-year-old dwarf presented with slightly abnormal odontoid, no atlantoaxial instability, and no neurologic signs or symptoms. B, Atlantoaxial instability without cord compression and myelopathy was noted in a 7-year-old dwarf with congenitally detached odontoid that dislocates forward with anterior arch of atlas. Posterior ring of atlas is absent, and posterior impingement of spinal cord against axis is avoided.

(Courtesy Steven E. Kopits, MD, and Milos Perovic, MD.)

MRI allows a more complete and accurate examination of the soft tissues and brainstem and assists in identifying the area of direct impingement in individuals with occipitocervical anomalies.⁵⁷ Dynamic flexion-extension MRI allows for improved selection of patients for cervical arthrodesis and allows direct determination of the need for concurrent anterior or posterior decompression of neural elements.^52,58 Surgical stabilization may be avoided or delayed in individuals with mild radiographic instability without spinal cord compromise.

Excessive motion at the occiput-C1 may lead to compression of the vertebral arteries.⁵³ Vertebral arteriography is helpful in evaluating patients who exhibit symptoms of transient brainstem ischemia.^24,69,70 If cerebellar herniation (Chiari malformation) is suspected, arteriography combined with myelography can show the anomaly (Fig. 30–11).⁶⁸

FIGURE 30–11 A 12-year-old child presented with nystagmus and cerebellar ataxia. A, Myelography shows Arnold-Chiari malformation with herniation of cerebellar tonsils through foramen magnum. Arrow indicates distal migration of tonsils. B, Vertebral arteriography in same patient shows inferior cerebellar vessels looping down to spinal canal. Arrow A indicates inferior cerebellar vessels; arrow B indicates posterior ring of atlas.

(Courtesy Frank Lee, MD.)

Treatment

Effective treatment generally can be provided only if the exact cause of symptoms has been determined by a careful correlation of the clinical and radiologic findings. Before surgical intervention, reduction of the atlantoaxial articulation should be achieved by either positioning or traction.^28,60,65,71 Operative reduction should be avoided, if possible, because it is associated with increased morbidity and mortality rates.^72,73 The patient should be maintained in the reduced position preoperatively until spinal cord edema and local irritation resolve; this is usually accompanied by improvement of the neurologic status and often remission of symptoms. If neurologic symptoms are unchanged with reduction and rest, the physician should carefully seek other causes.

Various techniques for surgical stabilization of atlantoaxial instability have been used in pediatric patients with good results.^60,74 Sublaminar wiring, C1-2 transarticular screws, transoral decompression with posterior plating, and laminectomy with Steinmann pin occipitocervical fusion all have been reported with good results (Fig. 30–12).^28,45 Choice of surgical procedure for stabilization must be individualized based on clinical presentation and surgeon expertise.^28,71

FIGURE 30–12 An 11-year-old child with Down syndrome presented with gross atlantoaxial instability. The patient’s gait was clumsy. Physical examination revealed poor coordination of the extremities. There was no other evidence of motor or sensory impairment of pathologic reflexes. The patient had no symptoms referable to the cervical spine 2 years after surgical stabilization.

Too little attention has been paid in the past to occult respiratory dysfunction in patients with atlantoaxial instability.^74,75 Many patients have an unrecognized decrease in vital capacity and chronic alveolar hypoventilation as a result of the neurologic injury to the brainstem.^18,76 Similarly, gag and cough reflexes are often depressed,⁷⁴ and the patient may experience a deterioration of pulmonary function in the postoperative period.⁷⁶ Periods of apnea and respiratory distress during surgery or in the immediate postoperative period have frequently resulted in death or have required prolonged respiratory support.⁷⁶ Preoperative pulmonary evaluation can be helpful in limiting the severity of respiratory complications. If pulmonary function is significantly reduced or if gag and cough reflexes are depressed, consideration should be given to preoperative tracheostomy.⁷⁴ Equipment for mechanical respiratory support must be immediately available during the postoperative period.⁷⁶

Laxity of Transverse Atlantal Ligament

Laxity of the transverse atlantal ligament is a diagnosis of exclusion suggested by the clinical occurrence of chronic atlantoaxial dislocation without a predisposing cause.³³ There is no history of trauma, congenital anomaly, infection, or rheumatoid arthritis to account for the radiologic finding. Most patients with this condition have the typical symptoms of atlantoaxial instability and require surgical stabilization.

Laxity of the transverse atlantal ligament is unusually common in patients with Down syndrome, with a reported incidence of 15% (Fig. 30–13).^42,77 The lesion may be found in all age groups, with no age preponderance.⁴² These patients seem to have rupture or attenuation of the transverse atlantal ligament with encroachment of the safe zone of Steel (see Fig. 30–8), but at least initially they are protected by the checkrein action of the alar ligaments from spinal cord compression (see Fig. 30–13). In other words, many patients have excessive motion, but relatively few are symptomatic, and most cases are discovered only by radiologic survey.⁷⁸ If radiographs of the upper cervical spine indicate an ADI of more than 5 mm, instability is considered to be present. Usually, if symptoms are present, instability of greater than 7 to 10 mm is found. MRI can be helpful in assessing the integrity of the transverse atlantal ligament.⁵⁵

FIGURE 30–13 A 16-year-old developmentally delayed child presented with insidious onset of bilateral upper extremity hyperreflexia and gait ataxia. A, Sagittal T1-weighted MRI shows cervicomedullary compression and increased atlantoaxial interval. B, Postoperative lateral radiograph shows reduction of deformity and fixation with transarticular screws augmented with sublaminar wires and bone graft.

(Adapted from Taggard DA, Menezes AH, Ryken TC: Treatment of Down’s syndrome-associated craniovertebral junction anomalies. J Neurosurg Spine 93:205-213, 2000.)

Complete data are unavailable regarding the best way to manage this problem.^42,77,78 Very little is known about the natural history, despite radiographic examination of hundreds of children.⁷⁸ Studies have indicated that some children experience progressive looseness with time; others who manifest small degrees of instability can occasionally become stable.⁴² Currently, radiographic examination on a routine basis is recommended for children with Down syndrome, particularly children who compete in athletics.⁷⁷ Any child with Down syndrome who has a musculoskeletal complaint, such as subluxing patella, dislocating hips, or unsteady gait, should be investigated. In addition, evaluation of the cervical spine should be a part of the preoperative assessment in these cases.

A patient with Down syndrome and no evidence of radiographic instability requires no activity restrictions other than avoidance of collision sports.^42–44 With current knowledge, prophylactic stabilization is not indicated. When a patient develops atlantoaxial instability and is neurologically intact, any sport with the potential to stress the cervical spine should be avoided. The Special Olympics has identified gymnastics, high jump, diving, butterfly and breast stroke swimming, and pentathlon to be at-risk activities.⁴² The use of a cervical collar in these patients is controversial. Progressive instability, neurologic symptoms, or both are indications for surgical stabilization.^34,42

Occipitoatlantal Instability

Occipitoatlantal instability is a rare condition often caused by trauma or, less commonly, congenital abnormalities.^79,80 Previously, most patients did not survive this injury, but with improved resuscitative measures, the problem has been reported more frequently. Clinical and neurologic manifestations include cardiorespiratory arrest, motor weakness, quadriplegia, torticollis, pain in the neck, vertigo, and projectile vomiting.⁷⁹ Surgical stabilization is usually necessary. Anomalies of the upper cervical spine can also lead to this condition, and it has been reported in association with Down syndrome. The curvature of the occipital condyle in healthy children increases by 60% from infancy to adolescence.⁸⁰ Patients who have Down syndrome and occipitoatlantal instability fail to develop this curved architecture in the occipital condyle.⁸⁰

Clinical Features

Most patients have an appearance similar to that in Klippel-Feil syndrome, with a short, broad neck; a low hairline; torticollis; a high scapula; and restricted neck movements.^35,37,83 The skull may be deformed and shaped like a “tower.” Kyphosis and scoliosis are frequent. Other associated anomalies occasionally seen include dwarfism, funnel chest, pes cavus, syndactylies, jaw anomalies, cleft palate, congenital ear deformities, hypospadias, and sometimes genitourinary tract defects.

Neurologic symptoms do not usually occur until the 3rd or 4th decade but can manifest during childhood.^32,84 Symptoms progress in a slow, unrelenting manner and may be initiated by traumatic or inflammatory processes. Symptoms rarely begin dramatically, but they have been reported as a cause of sudden death.⁸⁵ It is difficult to explain why neurologic problems develop so late and progress so slowly in these patients. It may be that the frequently associated atlantoaxial instability progresses with age and the resultant added demands placed on this articulation, producing gradual spinal cord or vertebral artery compromise.

McRae and Barnum⁸³ suggested that the key to development of neurologic manifestations lies with the odontoid and its position, an indication of the degree of actual or relative basilar impression. If the odontoid lies below the foramen magnum, the patient is usually asymptomatic.³³ With the decrease in vertical height of the atlas, the odontoid may project well into the foramen magnum, producing brainstem pressure, a fact well documented by autopsy.^37,38

Anterior compression of the brainstem from the backward-projecting odontoid is most common (see Fig. 30–14). This compression produces various findings, depending on the location and the degree of pressure. Pyramidal tract signs and symptoms (spasticity, hyperreflexia, muscle weakness and wasting, and gait disturbances) are most common, but cranial nerve involvement (diplopia, tinnitus, dysphagia, and auditory disturbances) may be seen less often. Compression from the posterior lip of the foramen magnum or the constricting band of dura may disturb the posterior columns, resulting in loss of proprioception, vibration, and tactile discrimination. Nystagmus, a common occurrence, is probably due to posterior cerebellar compression.

Vascular disturbances from vertebral artery involvement may occasionally result in syncope, seizures, vertigo, and unsteady gait, among other signs and symptoms of brainstem ischemia.^15,53 Disturbed mechanics of the cervical spine may result in a dull aching pain in the posterior occiput and neck with episodic neck stiffness and torticollis.¹⁵ Tenderness noted in the area of the posterior scalp may be due to irritation of the greater occipital nerve. The following most common signs and symptoms occur, in decreasing order of frequency: pain in the occiput and neck, vertigo, unsteady gait, paresis of the limbs, paresthesias, speech disturbances, hoarseness, double vision, syncope, auditory noise or disturbance, and interference with swallowing. Acute respiratory failure secondary to hypoventilation and sleep apnea may occur.¹⁵ All these may be manifestations of underlying atlantoaxial instability, which, as an isolated lesion, may produce neck pain, headaches, and neurologic deficits from cord or root irritation and, rarely, sudden death.^34,73

Radiographic Features

Standard radiographs of this area can be difficult to interpret. Tomography, CT, and MRI may be necessary to clarify the pathologic condition (see Fig. 30–14).^{15,52,53,84,85} Most commonly, the anterior arch of the atlas is assimilated into the occiput, usually in association with a hypoplastic posterior arch. The condition may range from total incorporation of the atlas into the occipital bone to a bony or fibrous band uniting one small area of the atlas to the occiput. CT scans are often necessary to show bony continuity of the anterior arch of the atlas with the occiput. Posterior fusion is not usually evident because this portion of the ring may be represented only by a short bony fringe on the edge of the foramen magnum. Despite its innocuous radiologic appearance, this fringe is frequently directed downward and inward, can compromise the spinal canal posteriorly, and has been found to create a groove in the spinal cord. It is usually assumed that the assimilated atlas is fused symmetrically to the occipital opening, but several autopsy specimens have shown a posterior positioning of the atlas.⁸³ This position, in effect, pushes the odontoid posteriorly, narrowing the spinal canal and the space available for the spinal cord.

Ossification of the atlas proceeds from paired centers—one for each of the lateral masses. These progress posteriorly into the neural arches, which are fully ossified at birth except for a gap of 5 to 9 mm posteriorly that closes by the 4th year.⁵³ The anterior arch of the atlas is invisible in 80% of neonates. This area most commonly ossifies from a single center, which appears during the 1st year of life and fuses to the remainder of the atlas by the 3rd year.^{31,34,53,86,87} A persistent bifid anterior and posterior arch of the atlas beyond age 4 years is observed in skeletal dysplasias, Goldenhar syndrome, Down syndrome, Conradi syndrome, and atlas assimilation.⁸⁸ In an extensive review of CT scans from 1104 patients, Senoglu and colleagues⁸⁹ found a 3.35% incidence. None had neurologic deficits or needed intervention. Rarely, hyperplasia of the posterior arch occurs and has been associated with myelopathy.⁹⁰

There is varying loss of height of the atlas, allowing the odontoid to project upward into the foramen magnum and creating a “relative” basilar impression (see Fig. 30–14). The position of the odontoid relative to the foramen magnum was described earlier (see section on basilar impression). McRae^38,83 measured the distance from the posterior aspect of the odontoid to either the posterior arch of the atlas or the posterior lip of the foramen magnum (SAC), which was closer. He stated that a neurologic deficit would be present if this distance was less than 19 mm. This distance should be determined in flexion because this position most dramatically reduces the space available for the cord.

Flexion-extension stress films (see Fig. 30–14) often show posterior displacement of the odontoid from the anterior arch of the atlas of 12 mm.⁸³ Associated atlantoaxial instability has been reported to develop eventually in 50% of patients^29,31; this is determined by measuring the distance from the anterior border of the odontoid to the posterior aspect of the anterior arch of the atlas. A distance greater than 4 mm in young children who probably have considerable cartilage present and 3 mm in older children and adults is considered pathologic.^86,87,91 The odontoid itself often has an abnormal shape and direction; it frequently is longer, and its angle with the body of the axis is directed more posteriorly.^73,83

McRae³³ was the first to note the frequent occurrence of congenital fusion of C2-3 (70%) and C1-2 instability in patients with atlanto-occipital fusion (Fig. 30–15). This occurrence suggests that greater demands are placed on the atlantoaxial articulation, particularly in flexion and extension when the joints above and below are fused.^37,83 von Torklus and Gehle³¹ noted that approximately 50% of patients develop late onset of atlantoaxial instability and the resultant potential for compromise of the spinal cord.²⁹

FIGURE 30–15 A 23-year-old man presented with Klippel-Feil syndrome, ataxic gait, hyperreflexia, and a history of several episodes of unconsciousness. Lateral laminagraph of cervical spine and base of skull shows C2-3 fusion and fusion of ring at C1 to opening of foramen magnum (occipitalization). Odontoid is hypermobile. Patients with this pattern of fusion are at great risk. With aging, odontoid may become hypermobile, and space available for spinal cord posteriorly may be compromised.

Gholve and colleagues³² used a morphologic classification: zone 1, a fused anterior arch; zone 2, fused lateral masses; zone 3, a fused posterior arch; and a combination of fused zones. In their study, 57% had atlantoaxial instability, with almost half of those having an associated C2-3 fusion. Spinal canal encroachment occurred in 37%, with almost half of these patients having clinical findings of myelopathy. The highest prevalence of spinal canal encroachment (63%) was noted in patients with occipitalization in zone 2.

Another commonly associated abnormality is the presence of a constricting band of dura posteriorly. This band has been found to create a groove in the spinal cord and may be the primary cause of symptoms. The band cannot be visualized on routine radiographs, and it does not correlate with the presence or absence of the posterior bony fringe of the atlas. Consequently, CT and MRI should be an integral part of the evaluation. Water-soluble contrast material (metrizamide) alone or in conjunction with CT can visualize the spinal canal and its contents. A properly performed study yields valuable information regarding the presence of a dural band; tonsillar herniation; and the size, shape, and position of the spinal cord in the spinal canal.^35,92

Treatment

Management of this uncommon problem may be hazardous. In contrast to anomalies of the odontoid, surgical intervention carries a much higher risk of morbidity and mortality.^37,73,93 Nonoperative methods such as cervical collars, braces, plaster, and traction should be attempted initially in some patients. These methods are often helpful in patients with persistent complaints of head and neck pain and are particularly helpful if symptoms follow minor trauma or infection. If neurologic deficits are present, immobilization may achieve only temporary relief. Patients presenting with evidence of a compromised situation in the upper cervical area must take precautions not to expose themselves to undue trauma.

With anterior spinal cord signs and symptoms owing to an unstable atlantoaxial complex, an occiput-to-C2 fusion is suggested, with preliminary traction to attempt reduction, if necessary. If reduction is possible and there are no neurologic signs, surgical intervention carries an improved prognosis.^34,37,73 Operative reduction should be avoided because this has frequently resulted in death.^72,73

Transoral resection of the odontoid should be considered if there is irreducible ventral compression.^28,94 Resection of the odontoid through an anterior approach may destabilize the upper cervical spine. In a large study, Dickman and colleagues⁹⁵ noted that 11 of 19 patients (40%) developed craniovertebral junction instability after transoral removal of the odontoid, 8 of whom did not have clinical or radiographic evidence of instability preoperatively. Instability was more common if there was a congenital bony malformation or if the patient required posterior decompression for an associated condition, such as Chiari malformation or syrinx.⁹⁵

Posterior signs and symptoms and myelographic evidence of bony or dural compression, depending on the degree of neurologic involvement, may be indications for a posterior decompression and stabilization of the occiput to C2.^71,96–99 Satisfactory occipital cervical stabilization and fusion can be obtained by using a halo brace in the postoperative period.⁹⁸ Internal fixation can facilitate postoperative management for certain patients, however. Internal fixation of the occipital cervical junction can pose problems, and several fixation techniques have been found helpful.^27,28,71 The addition of a cervical laminectomy may increase C1-2 instability or lead to the late development of a cervical kyphosis, particularly if the laminectomy involves several levels.^96,100 In these cases, the patient should be managed expectantly with appropriate internal stabilization or a halo vest.^82,96

Anomalies of Ring of C1

Anomalies of the posterior arch can be characterized as median clefts or hypoplasia. Geipel¹⁰¹ found defects in the posterior arch, most of which were median clefts, in 4% of 1613 autopsies. Currarino and colleagues¹⁰² developed a classification scheme (types A to E) for congenital defects of the posterior arch of C1 (Fig. 30–16). Greater than 90% are the type A median cleft variant. These patients are usually asymptomatic, and their anomaly is incidentally identified on radiographic studies.⁸⁹ Types C and D are accompanied by a free-floating posterior tubercle. These patients may experience neck pain and neurologic symptoms before discovery of the anomaly. The free posterior tubercle has been shown to move with neck extension and in some cases may migrate anteriorly and traumatize the spinal cord with neck extension. MRI studies have shown signal abnormality of the posterior spinal cord at the C1-2 level consistent with this mechanism of injury (Fig. 30–17).¹⁰²

FIGURE 30–16 Classification system for posterior C1 ring anomalies. Type A is small midline failure of fusion. Type B defects are unilateral clefts of variable size, ranging from small gap to complete absence of hemiarch. Type C indicates bilateral defects with preservation of most of dorsal arch. Type D is complete absence of posterior arch with retained, midline tubercle. Type E is complete posterior arch and tubercle deficiency.

(Adapted from Currarino G, Rollins N, Diehl JT: Congenital defects of the posterior arch of the atlas. Spine 47:267-271, 2000.)

FIGURE 30–17 Myelopathy secondary to dynamic tubercle movement in patient with type C posterior C1 ring anomaly. A, Sagittal T1-weighted MRI. B, Sagittal T1-weighted MRI with contrast medium enhancement. C, Sagittal T2-weighted MRI of cord contusion at C1-2 level.

(Adapted from Currarino G, Rollins N, Diehl JT: Congenital defects of the posterior arch of the atlas. Spine 47:267-271, 2000.)

Dubousset¹⁰³ called attention to a previously unrecognized problem with the ring of C1, the hemiatlas. Although there had been an occasional case report, Dubousset¹⁰³ presented the first large study to review the problem in depth. He reported 17 patients in whom he was able to document absence of the facet of C1, which led to a severe and progressive torticollis in young children (Fig. 30–18). Initially, the deformity is flexible and can be passively corrected. As the child ages, the torticollis becomes more severe and eventually fixed. Radiographic diagnosis using tomography or CT has been helpful in identifying this deformity. Use of traction to align the head and neck at the time of the study assists in highlighting the problem. If the patient is passively correctable, a single posterior fusion, occiput to C2, is performed. A halo cast is applied postoperatively to maintain the head in a satisfactory alignment until the fusion is complete.

FIGURE 30–18 Absence of C1 facet in a 5-year-old child with severe, progressive torticollis. Laminagraph shows hemiatlas, with complete absence of left portion of C1 ring.

(Courtesy Dr. Luther C. Fisher III, MD.)

Although the time at which this becomes a fixed deformity was not defined, Dubousset¹⁰³ reported good results in teenagers. This deformity can accompany Klippel-Feil syndrome with anomalies of the lower cervical spine as well. Dubousset¹⁰³ found an increased incidence of anomalies of the vertebral vessels in these children and suggested arteriographic evaluation before the use of traction or surgical intervention, which could further compromise a precarious blood supply to the midbrain and spinal cord.

Incidence

The frequency of these anomalies is unknown; similar to many anomalies that may be asymptomatic, they are probably more common than is recognized. They are often incidental findings or are seen in patients sustaining trauma or with symptoms sufficient to require radiographic investigation.

In the authors’ experience, aplasia is extremely rare. Many previous reports have confused aplasia for hypoplasia, and, as previously emphasized, aplasia probably has been a misnomer because it almost never describes an associated absence of the portion below the articular facets that contributes to the body of the axis. Hypoplasia and os odontoideum are infrequently reported and can be considered rare.^{54,104,106–109} With more recent awareness, these lesions are being recognized more commonly, however, than previous literature might indicate—especially os odontoideum.⁴⁷ In a large series reported by Wollin,¹⁰⁹ the average age of diagnosis was 30 years, but in a later series it was 18.9 years, suggesting earlier recognition.⁴⁷ An increasing number of children with these anomalies are being discovered.

In conditions such as Down syndrome, Morquio syndrome, Klippel-Feil syndrome, and some skeletal dysplasias, odontoid anomalies in association with ligamentous laxity producing atlantoaxial instability are much more common than in the general population.^{40,42,50,110–115} Associated regional malformations may occur, but in contrast to many congenital cervical anomalies, they are rare.^{50,104,108,116}

Development of Odontoid

The body of the odontoid is derived from the mesenchyme of the first cervical sclerotome and is actually the centrum of the first cervical vertebra, which becomes separated from the atlas during development to fuse with the remainder of the axis.^82,114,117 The apex of the odontoid process is derived from the mesenchyme of the most caudal occipital sclerotome or proatlas. Ossification of these two segments of the odontoid proceeds along separate lines.

Between the 1st and 5th prenatal months, the dens begins to ossify from two centers, one on each side of the midline. By the time of birth, they have fused into a single mass.^86,87,91 Occasionally, the right and left halves of the odontoid are not fused at birth, and a longitudinal midline cleft may be seen. At birth, the tip of the odontoid has not ossified, is V-shaped, and is known as a dens bicornis (Fig. 30–20). A separate ossification center within the “V,” known as a summit ossification center or ossiculum terminale, usually appears at age 3 years and fuses with the remainder of the dens by age 12.^86,87,118 Cattell and Filtzer⁶⁴ found an ossiculum terminale in 26% of 70 normal children 5 to 11 years old.

FIGURE 30–20 A, Radiograph of a 6-month-old newborn. Odontoid is normally formed and recognizable on routine radiographs at birth but is separated from the body of the axis by a broad cartilaginous band (arrow), similar in appearance to an epiphyseal plate. It represents the vestigial disc space and is referred to as neurocentral synchondrosis. B, Neurocentral synchondrosis is not at anatomic base of dens, at the level of superior articular facets of axis. This open-mouth view shows that embryologic base of odontoid is below articular facets and contributes a substantial portion to body of axis. This radiolucent line is present in nearly all children at age 3 years and in 50% by age 4; it is absent in most individuals by age 5.

An ossiculum terminale may never appear or may occasionally fail to fuse with the dens; it is then called an ossiculum terminale persistens. It is occasionally discernible as either a cyst or an area of increased density. These developmental anomalies are of little clinical significance.^50,119 Sherk and Nicholson¹¹⁴ reported a rare case of quadriplegia and death, however, in a child with Down syndrome that was directly attributable to atlantoaxial instability secondary to an ossiculum terminale. A similar occurrence in an otherwise normal child has been reported.¹²⁰ The ossiculum terminale usually is firmly bound to the main body of the dens by cartilage and consequently is seldom the source of instability.

At birth, the dens is separated from the body of the axis by a cartilaginous band (see Fig. 30–20) that represents the epiphyseal growth plate. This plate does not run across the base of the dens at the level of the superior articular facets of the axis but lies well below this level within the body of the axis (see Fig. 30–20B). The part of the odontoid below the articular facets contributes to the body of the axis. On the open-mouth view, the odontoid fits like a “cork in a bottle,” lying sandwiched between the neural arches (see Fig. 30–20B). This epiphyseal line is present in almost all children by age 3 years and in 50% of children by age 4, but it is absent in most by age 6.^64,91 It rarely persists into adolescence and adult life. If present, the line is not seen at the base of the dens where a fracture would be anticipated but lies well below the level of the superior articular facets within the body of the axis. In a young child, the unossified portions of the odontoid may give the false impression of odontoid hypoplasia. Similarly, one may erroneously conclude that the child has C1-2 instability because the anterior arch of the atlas commonly may slide upward and may protrude beyond the ossified portion of the odontoid⁶⁴ on the lateral extension radiograph (Fig. 30–21).

FIGURE 30–21 A, Lateral radiograph of a 5-year-old boy who at age 2 years fell from a sofa shows normal-appearing odontoid and cervical spine. The child complained of pain in neck and occiput and presented with torticollis. Symptoms and signs gradually resolved over 1 month. B, The child was asymptomatic until age 5, when over a period of 6 months he developed increasing neck pain and stiffness without neurologic findings. Radiographs showed os odontoideum and 7 mm of motion with flexion-extension. The patient subsequently underwent C1-2 stabilization.

The blood supply to the odontoid is from two sources. The vertebral arteries provide an anterior ascending artery and a posterior ascending artery that arise at the level of C3 and pass ventral and dorsal to the body of the axis and the odontoid, anastomosing in an apical arcade in the region of the alar ligaments. These arteries supply small penetrating branches to the body of the axis and the odontoid. Lateral to the apex of the odontoid, the anterior ascending arteries and apical arcade receive anastomotic derivatives from the carotids by way of the base of the skull and the alar ligaments. This curious arrangement of the blood supply is necessary because of the embryologic development and anatomic function of the odontoid. The transient neurocentral synchondrosis between the odontoid and the axis prevents the development of any significant vascular channels between the two structures. The body of the odontoid is surrounded entirely by synovial joint cavities, and its fixed position relative to the rotation of the atlas precludes vascularization by direct branches from the vertebral arteries at the C1 segmental level.

Etiology and Pathogenesis

A congenital etiology for os odontoideum has been assumed, and two theories have been advanced:

There is more often an associated short, stubby projection or hypoplastic odontoid remnant in the area of the odontoid base.^105,109 Hypoplasia and os odontoideum can be acquired secondary to trauma or, rarely, infection.^{33,47,109,118,121–123} Several cases of “os odontoideum” that developed several years after trauma when a normal odontoid was initially present have been reported (see Fig. 30–21).^41,122–124 Most patients have a significant episode of trauma before the diagnosis of os odontoideum.⁴⁷

Fielding and colleagues⁴⁷ suggested that most evidence favors an unrecognized fracture in the region of the base of the odontoid as the most common cause and less often a congenital origin. They postulated that after fracture of the odontoid there may be only slight separation of the fragments, but, with time, contracture of the alar ligaments, which attach to the tip of the odontoid, exerts a distraction force that pulls the fragment away from the base and closer to their origin at the occiput (Fig. 30–22). The blood supply to the odontoid is precarious because it passes up along the sides of the odontoid, is easily traumatized, and contributes to poor fracture healing or callus formation that may retard retraction of the fragment. The position of the odontoid adjacent to the ring of C1 is maintained by the intact transverse atlantal ligament.

FIGURE 30–22 A, Anatomic specimen of os odontoideum from a 17-year-old boy with multiple congenital anomalies who died of renal disease. Previous bony attachment of odontoid to axis was rough and blunted. Fibrocartilage pseudarthrosis was between os odontoideum and C2. B, Occiput and occipital joints, with os odontoideum suspended between facets. Transverse ligament was intact but loosened during preparation of specimen. Alar ligaments remain attached to tip of os odontoideum. Ligaments are shortened and appear to have pulled residual odontoid tip closer to their origin on occiput. Os odontoideum was firmly attached by soft tissue to occiput and ring of C1 and moved freely with these structures on C2. Foramen magnum is incomplete posteriorly. Posterior ring of C1 (not pictured) was intact and otherwise normal in its appearance. Spinal cord was narrowed and attenuated at level of C1.

The blood supply to the fragment is maintained by the proximal arterial arcade, from the carotid to the alar ligaments, and may be sufficient to maintain only a portion of the odontoid. Similarly, the blood supply to the proximal portion of the odontoid can be interrupted by excessive traction on these ligaments.⁴⁷ A familial form¹²⁵ and os odontoideum in identical twins with no history of trauma have been reported.^126,127 It is probable that congenital and post-traumatic forms of hypoplasia and os odontoideum exist.¹¹⁵ Failure of fusion of the apex of the odontoid (derived from the proatlas) to the main body of the atlas (derived from the first cervical sclerotome) is thought to result in the congenital form of os odontoideum.

The free ossicle of the os odontoideum usually appears fixed to the anterior arch of the atlas and moves with it in flexion and extension (see Fig. 30–7).^49,115 Instability can reduce the SAC but not the ADI.³⁰ The instability may be predominantly anterior or posterior or grossly unstable in all directions.^47,49,60 In a group of patients requiring surgery, the average displacement was 11 mm.

There is controversy regarding the presence of associated bony anomalies in the area of the hypoplastic dens or os odontoideum. In the authors’ experience, these changes are occasionally present, but with improved awareness they will undoubtedly be found more often. The posterior arch of C1 may be hypoplastic, whereas the anterior arch is hypertrophied.^34,49,54 Hypertrophy of the C1 anterior arch can be a useful sign of a chronic pathologic condition of C1-2.^53,115,128 If the posterior ring of C1 is narrow and there is abnormal anterior displacement of C1, there is less available space for the cord and an increased danger of neurologic sequelae.

Clinical Features

Patients may present clinically with no symptoms, local neck symptoms, transitory episodes of paresis after trauma, or frank myelopathy secondary to cord compression.^50,73,129 Minor trauma is commonly associated with the onset of symptoms, often of sufficient degree to warrant radiologic evaluation of the cervical spine. Symptoms may be mechanical owing to local irritation of the atlantoaxial articulation, such as neck pain, torticollis, or headache. Neurologic symptoms are due to C1-2 displacement and spinal cord compression. An important factor differentiating os odontoideum from other anomalies of the occipitovertebral junction is that these patients seldom have symptoms referable to cranial nerves¹⁰⁷ because the area of the spinal cord impingement is below the foramen magnum.

If clinical manifestations are limited to neck pain and torticollis (local joint irritation) without neurologic involvement (40% of cases),^49,106 the prognosis is excellent.^{47,107,130,131} Similarly, patients who exhibit only transient weakness of the extremities and dysesthesia after trauma usually have complete return of function. Patients in whom there is an insidious onset and slowly progressive neurologic impairment have a greater potential for permanent deficit.^47,116 Sudden death has been reported.¹⁰⁸ Damage may be mixed, with involvement of the anterior and the posterior spinal cord structures. Weakness and ataxia are more common complaints than sensory loss. Spasticity, increased deep tendon reflexes, clonus, loss of proprioception, and sphincter disturbances in various combinations have all been described, however.⁴⁹

A few patients may have symptoms and signs of cerebral and brainstem ischemia, seizures, mental deterioration, syncope, vertigo, and visual disturbances.^{60,117,131,132} In these patients, there typically is a paucity of cervical spinal cord signs and symptoms, and it is presumed that the patients are experiencing vertebral artery compression at the foramen magnum or just below it.^133–135 The diagnosis can be confusing, and many patients are misdiagnosed or thought to have progressive neurologic illness. Whenever Friedreich ataxia, multiple sclerosis, or other unexplained neurologic complaint is encountered, survey of the occipitocervical junction is suggested.

Radiographic Features

Recommended radiographic views are open-mouth, anteroposterior, lateral, flexion-extension, and CT reconstructions.⁴² Dynamic flexion-extension CT scans are valuable because plain films do not always show the anomaly or the extent of motion (Fig. 30–23; see Fig. 30–7).

FIGURE 30–23 A 7-year-old child was diagnosed with spondyloepiphyseal dysplasia. A, Odontoid was discovered to be absent in this patient. B and C, Lateral laminagraphs show stable atlantoaxial articulation in extension (B) and flexion (C). The patient was neurologically normal.

Cineradiography has been valuable for understanding odontoid anomalies, particularly anomalies that cause atlantoaxial instability,^129,136 but owing to high radiation exposure it has been largely replaced by dynamic CT scans in flexion and extension and CT reconstruction.^51,57 Most children with these lesions have a predominance of either anterior or posterior instability, but these can exist together.⁴⁹

Normal Variations

At birth, the normal odontoid can be visualized in the lateral view with its epiphyseal plate (see Fig. 30–20). A mistaken impression of hypoplasia may be given by a lateral extension radiograph because the anterior arch of the atlas may slide upward and protrude beyond the ossified tip of the dens, especially in very young patients.¹³⁷

Agenesis or Hypoplasia of Odontoid

Agenesis or hypoplasia of odontoid is an extremely rare anomaly that may be recognized from birth onward and is best seen in the open-mouth view, which is sometimes difficult to obtain in infancy (see Fig. 30–19). The diagnostic feature is the absence of the basilar portion of the odontoid, which normally dips down into and contributes to the body of the axis. This basilar portion is well below the level of the superior articular facets of the axis. The lateral view is of little help in distinguishing this anomaly from hypoplasia.

The most common form of hypoplasia manifests as a short, stubby peg of odontoid projecting just above the lateral facet articulations (see Fig. 30–19). Tomography is necessary to confirm whether an os odontoideum is present in addition to the hypoplasia.

In patients with multiple anomalies, the usual radiographic views are not always reliable in confirming the presence or absence of an odontoid. Similarly, in patients with abnormal bone, such as in Morquio syndrome and spondyloepiphyseal dysplasia,⁵⁰ the odontoid may be present but dysplastic, blending with the surrounding abnormal bone, and cannot be differentiated. In these situations, good results have been obtained by using lateral laminagraphic techniques (see Fig. 30–23) or dynamic CT scans in flexion and extension to ascertain the stability of the atlantoaxial articulation. The extension view should not be ignored. Many patients have been found with significant posterior subluxation.^47,54 Giannestras and colleagues⁵⁴ reported a youngster who became quadriparetic from prolonged hyperextension while lying prone watching television.

Os Odontoideum

In os odontoideum, there is a jointlike articulation between the odontoid and the body of the axis that appears radiologically as a wide radiolucent gap (see Fig. 30–19B and C). This gap may be confused with the normal neurocentral synchondrosis (see Fig. 30–19) before age 5 years. In children, the diagnosis of os odontoideum is confirmed by showing motion between the odontoid and the body of the axis. In adults, the diagnosis of os odontoideum is suggested by observing a radiolucent defect between the dens and the body of the axis.

The radiologic appearance of os odontoideum may be similar to a traumatic nonunion, and often they cannot be differentiated.^33,109 In os odontoideum, the gap between the free ossicle and the axis usually extends above the level of the superior facets and is wide with a smooth edge. The ossicle is usually approximately one half the normal size of the odontoid and is round or oval, and the cortex is of uniform thickness. In traumatic nonunion, the gap between the fragments is characteristically narrow and irregular and frequently extends into the body of the axis below the level of the superior facets of the axis. The bone fragments appear to “match,” and there is no marginal cortex at the level of the fracture or the rounded-off appearance found with os odontoideum.¹¹⁸ The odontoid ossicle is fixed firmly to the anterior ring of the atlas and moves with it in flexion, extension, and lateral slide. The anterior portion of the atlas is usually hypertrophied, and the posterior portion of the ring may be hypoplastic or absent.^47,107,128

Recommended radiographic views are open-mouth and lateral flexion-extension views. The literature supports that plain radiographs are sufficient to establish the diagnosis of os odontoideum. Spierings and Braakman¹³⁸ and Watanabe and colleagues¹²⁹ studied flexion-extension radiographs in patients with os odontoideum and attempted to correlate radiographic measurements with myelopathy. The degree of C1-2 instability measured by anteroposterior translation did not correlate with neurologic deficits, but the space available for the cord with neck extension (SAC) was significantly smaller in the myelopathic group. A critical anteroposterior diameter for cervical myelopathy was identified as 13 mm.^129,138

CT reconstructions are indicated when routine views are unsatisfactory in showing the anomaly. Lateral flexion-extension stress views should be conducted voluntarily by patients, particularly patients with a neurologic deficit. The degree of anteroposterior displacement of the atlas on the axis should be documented (Fig. 30–24). The os odontoideum moves with the ring of C1, and consequently measurements of its relationship to C1 are of little value. Measurements can be made using a line projected superiorly from the posterior border of the body of the axis to a line projected inferiorly from the posterior border of the anterior arch of the atlas. Measurements greater than 3 mm should be considered pathologic. Most symptomatic patients exhibit significant instability.⁴⁷ In a large series reported by Fielding and colleagues,⁴⁷ the average measurement was 1 cm; most were either anterior or posterior, but some were unstable in all directions.¹²⁹

FIGURE 30–24 A 7-year-old child had a 1-year history of peculiar posturing and neck stiffness. A, Flexion radiograph showed os odontoideum and subluxation of C1-2. Degree of anteroposterior displacement of atlas on axis should be documented. Os odontoideum moves with ring of C1, and consequently measurements of its relationship to C1 are of little value. Measurements can be made using a line projected inferiorly from posterior border of anterior arch of atlas. B and C, Extension (B) and flexion (C) radiographs after posterior stabilization. Reduction must be accomplished before surgery; if wire stabilization is selected, care must be taken to avoid further flexion of neck during surgery.

When instability has been long-standing, it becomes multidirectional, and the C1-2 articulation is very unstable.¹²⁹ Watanabe and colleagues¹²⁹ noted that patients who have more than 20 degrees of sagittal plane rotation have a higher rate of myelopathy (86%), and patients with myelopathy have a high rate of instability (40%). The greater the sagittal plane rotation and looseness, the more likely is spinal cord compression.

CT scan can help to define the anatomy of the craniocervical junction, including the completeness of the atlas ring and the position of the transverse foramina at C1 and C2. CT scan has also been shown to aid in distinguishing os odontoideum from a post-traumatic odontoid nonunion (Fig. 30–25).¹³⁹

FIGURE 30–25 A and B, Coronal (A) and sagittal (B) CT scans of upper cervical spine clearly show os odontoideum.

MRI is helpful for evaluating the SAC. MRI studies have shown a direct correlation between cord signal abnormality and the degree of myelopathy measured clinically.¹⁴⁰ Dynamic MRI as described by Hughes and colleagues¹⁴¹ allows direct visualization of motion of the os odontoideum, atlas, and axis and surrounding motion throughout the full range of motion of the cervical spine (Fig. 30–26). New open MRI configurations allow analysis of the cervical spine in the physiologic upright position, supporting the weight of the head; this may assist in identifying subtle pathology that is not apparent in the nonphysiologic, supine position.

FIGURE 30–26 Dynamic MRI of cervical spine showing cord compression in a patient with os odontoideum. A, Sagittal MR in neutral position. B, Cervical spine in flexion shows retrolisthesis of os with cord compression. C, Cervical spine in extension shows reduction of os to C2 body and relief of cord compression.

(Adapted from Hughes TB, Richman JD, Rothfus WE: Diagnosis of os odontoideum using kinematic MRI. Spine 24:715-718, 1999.)

Treatment

Patients with congenital anomalies of the odontoid lead a precarious existence. The concern is that a trivial insult superimposed on an already weakened and compromised structure may be catastrophic. In the authors’ experience, patients with these problems either have or develop gross atlantoaxial instability and with it the possibility of progressive myelopathies or even death.⁴⁹

The natural history of os odontoideum is variable, and predictive factors for instability have not been identified.⁴⁹ Indications for surgical stabilization include an os odontoideum in association with occipitocervical pain or os odontoideum with myelopathy. Other factors that may assist in surgical decision making are the severity of C1-2 instability and other associated osseous anomalies. Neural decompression must address anterior bony or soft tissue impingement of the spinal cord or compression posteriorly from the dorsal arch of C1.

Patients with local symptoms or transient myelopathies may expect recovery, at least temporarily.^107,130,131 Cervical traction or immobilization may be helpful in such circumstances. Surgical stabilization is indicated if there is neurologic involvement (even if transient), if there is instability of 10 mm or more in flexion and extension, if there is progressive instability, or if there are persistent neck complaints associated with atlantoaxial instability.

Considerable controversy exists over the role of prophylactic stabilization in asymptomatic patients with instability.^{49,65,107,130,135} Surgical treatment is not required for every patient in whom an os odontoideum is identified. Patients who have no neurologic symptoms or instability at C1-2 can be managed with periodic observation. The lack of C1-2 instability at initial diagnosis does not guarantee that instability will not develop in these patients, however, or that they are not at higher risk for spinal cord injury with trauma. For this reason, longitudinal follow-up with flexion-extension radiographs of these patients is recommended.

Klimo and colleagues⁴⁹ recommended surgical stabilization for all patients with os odontoideum. The safety of stability, and with it the ability to lead a normal active life, must cause one to weigh the possible complications of surgery against the catastrophic dangers of instability with secondary cord pressure. In the pediatric age group, it may be difficult or impossible to curtail activity, even in the presence of marked instability.^47,60,142 If possible, reduction of the atlantoaxial articulation must be accomplished before surgery either by careful positioning of the patient or by skull traction. Manipulative reduction during surgery is discouraged because it has proved extremely hazardous and may result in respiratory distress, apnea, or death.⁶⁵ Ideally, the patient should be maintained in the reduced position several days before surgery to allow recovery of neurologic function and to lessen spinal cord irritation. When fusion is undertaken, regardless of the indication, preoperative halo traction is often required to achieve reduction, which may have to be continued during surgery and postoperatively until transfer to a suitable immobilization device.^107,143

The suggested method of stabilization is posterior cervical fusion of C1-2. Posterior C1-2 arthrodesis in the treatment of os odontoideum provides effective stabilization in most patients. Posterior Gallie and Brook sublaminar wiring techniques with iliac crest bone graft and halo immobilization have been reported with favorable fusion results.^71,144–153 This method is not without risk because slight flexion is often required to pass the wire beneath the posterior ring of the atlas, and it can have tragic results.^107,144 When attempting posterior stabilization using a wire technique in small children, the spinous process of C2 may be small or poorly developed or not yet completely ossified. In this situation, a threaded Kirschner wire through the spinous process, as described by Mah and colleagues,¹⁵⁴ can be very helpful in improving the stability of the wiring technique.

In a patient with a marginally functioning neurologic status, it may be wiser to perform an occiput-to-C2 arthrodesis and plan to maintain immobilization in extension during the postoperative period.¹⁰⁷ In this regard, the halo vest is helpful. Incomplete development of the posterior ring of C1 is uncommon but is reported to occur with increased frequency in patients with os odontoideum.⁴⁷ The completeness of the C1 arch should be evaluated preoperatively because a large gap may preclude wire fixation.¹⁴⁸ If wire fixation is employed, excessive tightening of the wire should be avoided. The articulation is frequently unstable in flexion and extension, and posterior dislocations may occur, owing to overcorrection, with disastrous results.

In patients who have a highly mobile but reducible C1-2 articulation, several authors have reported good results using transarticular screws.^{41,71,144,149,150,152} Atlantoaxial transarticular screws have been shown to be feasible in pediatric patients, and the superior biomechanical stability may eliminate the need for prolonged postoperative halo vest immobilization (see Fig. 30–12).^149,155 A preoperative CT scan with reconstruction views is essential with this technique to define the C1-2 anatomy and course of the vertebral arteries.⁴¹ An aberrant path of the vertebral vessels that are at risk with screw placement is a contraindication for the procedure.⁴¹ Other techniques have been described. Lateral mass screws and transarticular screws have been used with good success being reported.^{152,156–158} Lateral mass screws and fixation to the skull occiput have comparable strength.¹⁵⁶ The carotid artery can be hit with a lateral mass screw.¹⁵⁹

With transarticular fixation of C1-2, the lamina of C1 can be removed without the necessity of extending the fusion to the occiput. The cervical fusion of C1-2 is generally reliable and safe; however, in patients with spinal cord compression, fixed dislocations, or congenital ligamentous laxity, particularly in patients with Down syndrome, extra caution should be employed.^148,160 Smith and colleagues¹⁶⁰ reported significant problems with C1-2 stabilization, particularly in patients who are very unstable or have a myelopathy. Patients with failed fusions or irreducible dislocations were at high risk for perioperative neurologic complications.

Patients in whom the C1-2 dislocation is unreducible after an adequate trial of traction pose a difficult management problem.¹⁶¹ In this situation, posterior decompression by laminectomy has been associated with increased morbidity and mortality.¹⁰⁷ In addition, posterior decompression alone may potentiate C1-2 instability, and if performed it must be accompanied by occiput-to-C2 arthrodesis.^47,161 For patients with no neurologic deficit, a simple in situ posterior fusion is the least hazardous procedure. If reduction of the C1-2 dislocation is considered necessary or if the clinical situation precludes posterior stabilization, an anterior approach should be considered.^28,34,162 The lateral retropharyngeal approach described by Whitesides and McDonald¹⁶² provides anterior exposure of the C1-2 articulation adequate to perform decompression, reduction, and stabilization. This route may be used in place of the transoral or mandibular and tongue-splitting approaches, which are associated with an increased incidence of infection.^28,162 Associated myelopathy can be addressed with reduction of the deformity, dorsal or ventral decompression, and occipitocervical fusion.

A common clinical problem is differentiation between a fractured odontoid and os odontoideum. Discovery is usually made after trauma in both conditions, and an accurate diagnosis may be impossible by radiographic techniques alone. In this situation, a period of immobilization (skull traction or cast) is recommended. Overdistraction, particularly in children, should be avoided because it can lead to os odontoideum.⁴⁷ If the lesion represents an acute fracture, healing usually occurs. If a congenital or traumatic nonunion is present, surgical stabilization is necessary if atlantoaxial instability is shown.

Summary

Anomalous development of the odontoid is uncommon; the clinical significance lies in its potential to produce serious neurologic sequelae owing to atlantoaxial instability. Although there are several recognized variations (aplasia, hypoplasia, and os odontoideum), clinically they share the same signs and symptoms, and treatment is identical. Symptoms are usually due to instability of the atlantoaxial joint, with compression of the spinal cord anteriorly against the axis or posteriorly from the ring of the atlas. Patients may present with no symptoms, persistent neck complaints, or transient or permanent neurologic deficits or may die suddenly. Symptoms from cranial nerve irritation seldom occur, but symptoms of cerebral and brainstem ischemia are occasionally noted owing to compression of the vertebral arteries in the area of the atlas.

If the condition is suspected, the diagnosis can usually be confirmed on lateral flexion-extension radiographs. CT scans with reconstruction views are often very helpful. Flexion-extension stress radiographs are also necessary to determine the presence and degree of atlantoaxial instability.

The role of prophylactic surgical stabilization is not yet established. If marked instability is shown or if there are clinical findings of neurologic compromise, there is general agreement that surgical fusion should be performed. Operative reduction should be avoided, and preoperative correction by traction or positioning is preferred. Posterior surgical stabilization of the C1 and C2 vertebrae is sufficient if this can be accomplished without flexing the head further. In this situation, occiput-to-C2 stabilization is recommended to avoid all possibility of further neurologic trauma during the procedure.

Clinical Features

The classic clinical description of Klippel-Feil syndrome is a triad—low posterior hairline, short neck, and limitation of neck motion—but less than half of patients have all three signs (Figs. 30-30 and 30-31).^112,180 The presence of these signs is directly related to the degree of cervical spine involvement. Clinically, the most consistent finding is limitation of neck motion.¹⁶⁶ If fewer than three vertebrae are fused or if only the lower cervical segments are fused, the patient generally has no detectable limitation.¹⁶⁶ In addition, many patients with marked cervical involvement are able to compensate with hypermobility at the unfused joints and to maintain a deceptively good range of motion.¹¹² Several patients seen by the authors have 90 degrees of flexion-extension, occurring at the only open interspace (Fig. 30–32). Generally, flexion-extension is better preserved than rotation or lateral bending. Rarely, patients have no detectable motion and fixed hyperextension of the neck; this is usually associated with iniencephaly (absence of the posterior cervical laminae and an enlarged foramen magnum) (see Fig. 30–28).¹⁸¹

FIGURE 30–30 A 9-year-old child with Klippel-Feil syndrome. Note short neck with tendency toward webbing, mild torticollis, and asymmetry of eye level. The patient clinically has marked restriction of neck motion, impaired hearing, and mirror motions (synkinesia) of upper extremities.

FIGURE 30–31 Extreme form of webbing of neck—pterygium colli. Note low posterior hairline.

FIGURE 30–32 A-D, Flexion-extension of cervical spine shown clinically (A and C) and radiologically (B and D) in an 18-year-old woman with Klippel-Feil syndrome. Most neck motion is occurring at C3-4 disc space. Clinically, the patient is able to maintain adequate 90-degree range of flexion-extension. She is asymptomatic at present, but with aging this hypermobile articulation may become unstable.

Unless extreme, shortening of the neck is a subtle finding. Similarly, the low posterior hairline is not constant (see Figs. 30-30 and 30-31). Less than 20% of patients with Klippel-Feil syndrome have obvious facial asymmetry, torticollis, or webbing of the neck.^112,166 When extreme, webbing of the neck is called pterygium colli and consists of large skin folds extending from the mastoid to the acromion (see Fig. 30–31).¹⁸² The underlying muscles may be involved, but surgical release generally does not result in improved neck motion.

Sprengel deformity occurs in 15% to 35% of cases unilaterally or bilaterally (Fig. 30–33).^{112,165,166,171,183,184} At the 3rd week of gestation, the scapula develops from mesodermal tissue high in the neck at the level of C4. It descends into the thoracic position by the 8th week, or approximately at the same time that the Klippel-Feil lesion is thought to occur.^165,166 It is logical to expect a significant relationship between these two anomalies. Occasionally, there is a bony bridge between the cervical spine and scapula, an omovertebral bone, or connection to the clavicle.¹⁸⁵ Its removal may permit an increase in neck and shoulder motion.

FIGURE 30–33 A 6-year-old child with Klippel-Feil syndrome and Sprengel deformity on left. A, Frontal view. B, Posterior view. C, Radiographic appearance shows posterior vertebral anomalies of cervical spine and high left scapula. The patient subsequently had left scapuloplasty.

Probably for the same embryologic reasons, other clinical findings are occasionally apparent (see Table 30–1), including ptosis of the eye, Duane contracture (contracture of the lateral rectus muscle),¹⁶⁶ lateral rectus palsy, facial nerve palsy, neurenteric cyst,¹⁸⁶ and cleft or high-arched palate.^187,188 Abnormalities of the upper extremities include syndactyly, hypoplastic thumb, supernumerary digits, and hypoplasia of the upper extremity, which may represent an interruption of the embryonic blood supply.¹⁸⁹ Abnormalities of the lower extremities are infrequent.

With the exception of the anomalies that involve the atlantoaxial joint, no symptoms can be directly attributed to the fused cervical vertebrae. All symptoms commonly associated with Klippel-Feil syndrome originate at the open segments where the remaining free articulations may become compensatorily hypermobile.^180,190,191 Owing to the increased demands placed on these joints or in response to trauma, this hypermobility can lead to frank instability or early degenerative arthritis.^180,183,190

Symptoms may arise from two sources: (1) mechanical symptoms owing to irritation of the joint and (2) neurologic symptoms owing to root irritation or spinal cord compression. Patients with a short segment fusion are less likely to develop symptoms^166,192 because the loss of motion is adequately compensated by the remaining free segments. Patients with synostosis of the lower cervical spine are at less risk because the limitation is minimal and can be adequately compensated by the more normally mobile joints above. Most patients who develop symptoms are in the 2nd or 3rd decade of life,^166,180,190 suggesting that the instability is in part a function of time with increasing ligament laxity.

Neurologic symptoms are generally localized to the head, neck, and upper extremities and result from direct irritation or impingement of the cervical nerve roots with radicular symptoms in the upper extremities.²² The symptoms can usually be localized to the hypermobile joints adjacent to the fused segments.¹⁹³ There may be constriction and narrowing of the nerve root at the foramen from osteophytic spurring.^22,180 If joint instability is progressive or if there is appropriate trauma, the spinal cord may be involved to varying degrees, ranging from mild spasticity, hyperreflexia, and muscular weakness to sudden complete quadriplegia after minor trauma.^{166,183,194,195} Syncope and neurologic compromise may be associated with mechanical compromise of the vertebral arteries secondary to hypermobility, with ischemic episodes or emboli.^196,197

Radiographic Features

In a child with severe involvement, an adequate radiographic evaluation can be difficult. Fixed bony deformities frequently prevent proper positioning, and overlapping shadows from the mandible, occiput, or foramen magnum may obscure the upper vertebrae (Fig. 30–34). In this situation, flexion-extension views may provide the information necessary to assess stability. Another technique that has gained wide acceptance in the diagnosis of cervical instability is CT.^51,180 This technique coupled with flexion-extension of the cervical spine can delineate more precisely the presence or absence of spinal cord compression. CT can be enhanced further by contrast myelography. With MRI, the relationship between the bony elements and neurologic structures can be viewed directly in flexion and extension to show the origin of neural compression.^57,187,198 Use of MRI is particularly helpful for patients who have abnormal bone, such as patients with Morquio disease.¹⁹⁹

FIGURE 30–34 A 6-year-old boy presented with Klippel-Feil syndrome. A, Routine lateral radiograph of cervical spine. Overlapping shadows from shoulder and occiput obscure much of cervical spine. B, Lateral laminagraph in flexion shows anterior hemivertebra, probably C4, and congenital fusion of C2-3, C6-7, and T3-4. C, Lateral laminagraph in extension shows absence of posterior ring of C1 and unstable C1-2 articulation. Flexion-extension laminagraphs are helpful in providing information necessary to evaluate children with severe deformity, particularly if vertebral instability is suspected.

With these techniques, the space available for the spinal cord can be measured directly rather than inferred by use of the ADI.⁵⁷ Knowledge of the normal variations in cervical spine mobility, particularly in children, is important in evaluating patients with Klippel-Feil syndrome.^64,200 Pseudosubluxation of C2 on C3 with flexion can be observed in 45% of normal children younger than 8 years of age (see Fig. 30–9).⁶⁴ Marked angulation at a single interspace during flexion, rather than a uniform arc of vertebral motion, can be observed in normal children (16%)²⁰¹ and may be misinterpreted as vertebral fusion below.

Fusion of cervical vertebrae is the hallmark of Klippel-Feil syndrome. This fusion may be simply synostosis of two bodies (congenital block vertebrae) (Fig. 30–35; see Fig. 30–27) or massive fusion of vertebrae, which was found in Klippel and Feil’s first patient (see Fig. 30–28).¹⁶³

FIGURE 30–35 Postmortem specimen of congenital block of vertebra of C3-4. A, Anterior view. B, Posterior view. Specimen shows complete fusion, but remnants of cartilaginous vertebral bodies can still be seen.

Aside from vertebral fusion, flattening (wasp waist) and widening of the involved vertebral bodies, an increase in the space available for the spinal cord, and absent disc spaces are the most common findings.^202,203 Hypoplasia of the disc space or remnants of it can often be seen (see Fig. 30–35). In a young child, narrowing of the cervical disc space cannot always be appreciated because the ossification of the vertebral body is incomplete, and the unossified endplates may give the false impression of a normal disc space (Fig. 30–36). With continued growth, the ossification of the vertebral bodies is completed, however, and the fusion becomes obvious.¹⁹¹ If fusion is suspected in a child, it may be confirmed by flexion-extension views (Fig. 30–37). Juvenile rheumatoid arthritis (Fig. 30–38), rheumatoid spondylitis, and infection can mimic the radiographic findings, but usually the clinical history and physical examination indicate the correct diagnosis.

FIGURE 30–36 A, Lateral radiograph of an 8-year-old child shows posterior fusion of laminae and spinous process but incomplete fusion of vertebral bodies anteriorly. B, At age 19, there is complete fusion of vertebral bodies at C2-3 and C4-7. Unossified cartilage of vertebral bodies in children can give false appearance of normal disc space.

FIGURE 30–37 A 3-year-old child presented with Klippel-Feil syndrome and congenital scoliosis. A and B, Lateral flexion-extension radiographs of cervical spine show that neck motion occurs predominantly between C4 and C5. Flexion-extension views are helpful in determining type and extent of congenital fusion in young children.

FIGURE 30–38 Juvenile rheumatoid arthritis. A, Radiographic appearance of cervical spine in a patient at age 5, at onset of rheumatoid process. B, Same patient at age 10 with complete fusion of laminae posteriorly and severely restricted neck motion. With further growth, these vertebral bodies would be expected subsequently to fuse.

Hemivertebrae are common (Fig. 30–39); they occurred in 74% of patients in the review of Gray and colleagues,¹⁶⁶ and the incidence increases with the number of segments fused. Fusion of posterior elements usually parallels fusion of the vertebral bodies. In a young child, particular attention should be paid to the laminae because fusion posteriorly is often more apparent than anteriorly in early life (Fig. 30–40).^112,191

FIGURE 30–39 A 14-year-old patient presented with cervical hemivertebrae. Hemivertebrae are common in Klippel-Feil syndrome but usually are found in dorsal or lumbar vertebral segments.

FIGURE 30–40 A 3-month-old infant presented with Klippel-Feil syndrome. Radiograph shows posterior fusion of laminae of C2-3 (arrow). In young children, particular attention should be paid to laminae because fusion posteriorly is often more apparent than anteriorly in early life.

The sagittal and transverse diameters of the spinal canal are usually normal. Narrowing of the spinal canal, if it occurs, is usually seen in adult life and is due to degenerative changes (osteoarthritic spurs) or hypermobility.* Enlargement of the cervical canal is uncommon; if found, it may indicate conditions such as syringomyelia, hydromyelia, or Arnold-Chiari malformation.^62,206 The intervertebral foramina are usually smooth in contour but are frequently smaller than normal and oval rather than circular (Fig. 30–41). Posterior spina bifida is common (45%), but anterior spina bifida is rare. Rarely, there is complete absence of the posterior elements. This condition is usually accompanied by enlargement of the foramen magnum and fixed hyperextension of the neck, referred to as iniencephaly (see Fig. 30–28).¹⁸¹

FIGURE 30–41 Oblique view of cervical spine shows smooth contour of intervertebral foramina, which are frequently smaller than normal and oval rather than circular.

All these defects may extend into the upper thoracic spine, particularly in patients with severe involvement. A disturbance of the upper thoracic spine on a routine chest radiograph may be the first clue to an unrecognized cervical synostosis. With a high thoracic congenital scoliosis, the radiographic evaluation should routinely include lateral views of the cervical spine.

Patterns of Cervical Motion

One can gain insight into the problem of instability by reviewing the lateral flexion-extension films of a patient with Klippel-Feil syndrome. The type or pattern of cervical motion depends on the location and extent of the fused cervical vertebrae. Patients with fusion of the lower cervical vertebrae or with more than two disc spaces between fused segments seem to be at low risk for serious problems.

Pizzutillo and colleagues¹⁹⁰ reviewed patients from the Alfred I. DuPont Institute and patients reported in the literature to determine the long-term problems found in Klippel-Feil syndrome. In their classification, these investigators noted that patients who have upper segment instability were more likely to be younger and to have neurologic problems. Conversely, degenerative changes were more common in older patients with low segment hypermobility. This finding emphasizes the importance of screening the upper cervical spine in a young child for instability. There are three high-risk patterns of cervical spinal motion that potentially have a poor prognosis, from either early instability or late degenerative osteoarthritis.²⁰⁷

Pattern 1 is fusion of C2 and C3 with occipitalization of the atlas. Complications associated with this pattern were first reported by McRae in 1953,³³ and they received substantial support in the literature.^22,208 Flexion-extension is concentrated in the area of C1 and C2. With aging, an odontoid can become hypermobile, narrowing the spinal canal and compromising the spinal cord and brainstem.

Pattern 2 is a long fusion with an abnormal occipitocervical junction (see Fig. 30–28). This is similar to the C2-3 fusion of McRae and could be reviewed as a more elaborate variation. The force of flexion-extension and rotation is concentrated in the area of the abnormal odontoid or poorly developed ring of C1, which cannot withstand the wear and tear of aging.^204,207,208 It is important to differentiate this pattern from the pattern in a patient with a long fusion and a normal C1-2 articulation (Fig. 30–42), which is usually compatible with a normal life expectancy.

FIGURE 30–42 Radiograph of a 45-year-old man with Klippel-Feil syndrome. The patient has complete fusion of C2-7. Flexion-extension occurs only at atlantoaxial articulation. There are no symptoms referable to neck, despite two previous serious falls. This pattern seems to be relatively safe because normal occipitocervical junction serves as protection from late instability.

Pattern 3 is a single open interspace between two fused segments (Fig. 30–43). In this situation, cervical spine motion is concentrated at the single open articulation. In some patients, this hypermobility may lead to frank instability or degenerative osteoarthritis (Fig. 30–44).* This pattern can be easily recognized because the cervical spine appears to angle or hinge at the open segment. Samartzis and colleagues²⁰³ studied cervical fusion patterns in patients with Klippel-Feil syndrome. They developed a classification system in which type I described a single fusion; type II described multiple, noncontiguous fusions; and type III described multiple, contiguous fusions. Samartzis and colleagues²⁰³ found that type I was associated with axial neck symptoms, and types II and III were associated with radicular and myelopathic symptoms.

FIGURE 30–43 Open interspace between two fused segments. This 7-year-old child with Klippel-Feil syndrome has flexion-extension motion of neck occurring primarily at one interspace. Cervical spine appears to angle or hinge at this point. This pattern is worrisome because wear and tear of aging may lead to early degenerative change or instability and narrowing of spinal cord.

FIGURE 30–44 A 54-year-old man presented with a 4-month history of persistent neck pain with radiation into upper extremities. He had no history of neck complaints but a long history of occipital headaches. He recently noted paresthesias in upper and lower extremities. Neurologic examination and electromyography were normal. A, Lateral radiograph of cervical spine shows congenital fusion between C2-3, C4-5, and C6-7. Note marked changes of degenerative osteoarthritis with large osteophyte formation at open interspace of C3-4 and C5-6. B, Myelography during extension of cervical spine shows narrowing of spinal canal owing to large osteoarthritic spurs at C3-4 and C5-6. The patient subsequently underwent spine stabilization with relief of neck complaints.

Associated Conditions

Scoliosis

Scoliosis is the most frequent anomaly found in association with the syndrome.^{112,171,209,210} Of these patients, 60% have a significant degree of scoliosis (>15 degrees by the Cobb method).^112,210 Most of these patients require treatment and should be followed through the growth years. Radiographic examinations should include lateral views of the spine because increasing kyphosis may make the need for treatment of the scoliosis more urgent. If the deformity is recognized early, many children can be successfully managed with standard spinal orthoses. At present, most of these patients have required posterior spinal stabilization, partly owing to late recognition.^{112,171,173,194}

Two types of scoliosis can be identified: congenital scoliosis, owing to vertebral anomalies and differential growth patterns (Fig. 30–45), and compensatory scoliosis, below the area of vertebral involvement. In the authors’ series, congenital scoliosis is more common (55%), and in more than half of the children, the curvature was progressive and required treatment.¹¹² Progressive curves are more likely associated with children who have extensive fusions.²¹⁰ Most children (75%) required posterior spinal fusion to arrest an increasing deformity, and the remainder were controlled with a brace or cast.¹¹² Progressive scoliosis frequently occurred in the normal-appearing vertebrae below the primary congenital curve. If only the congenitally involved segments are examined in follow-up, an increasing compensatory scoliosis in the lower vertebrae may not be recognized, and its significance may not be appreciated until serious deformity results.²⁰⁷

FIGURE 30–45 Radiograph of an 8-year-old child with Klippel-Feil syndrome and deafness. Severe kyphoscoliosis subsequently required spine correction and stabilization.

When surgical intervention is required, the same principles apply as used in congenital scoliosis. When spinal fusion is performed, the orthopaedist should carefully consider the overall alignment of the patient’s spine. The temptation to achieve maximal radiologic correction of the mobile segments must be tempered by careful consideration of the congenitally fixed segments. Failure to observe this principle may result in an unbalanced spine; the patient will have traded one deformity for another that may be even worse than the original.

Documented progression of scoliosis, whether in the congenitally distorted elements or in the compensatory curve below, demands immediate and appropriate treatment to prevent serious additional deformity. Progressive scoliosis in the thoracic spine may seriously compromise pulmonary function.^112,174 More subtle occult abnormalities can lead to respiratory difficulty in some patients with Klippel-Feil syndrome.²¹¹ Abnormal rib spacing, congenital fusion of the ribs, and deformed costovertebral joints may inhibit full expansion of the rib cage during respiration.²¹² Although not causing an angular deformity, fusion of the thoracic vertebrae may decrease the size of the thoracic cage. A dwarf with spondylothoracic deformity may represent a severe form of this problem, leading to early respiratory death.²¹³

Also, Krieger and colleagues⁷⁵ reported on the relationship of occult respiratory dysfunction and craniovertebral anomalies. They noted that in addition to the obvious problems of bony impingement or traction on the brainstem, these patients may have subtle hydrocephalus, which may adversely affect respiratory function. This information has particular application when cervical distraction devices are contemplated in the treatment of scoliosis (halo-femoral or halo-pelvic traction). When considering the use of such devices, the physician should be aware that children with Klippel-Feil syndrome may be more susceptible to neurologic or vascular injury and that the presence of cervical anomalies may preclude the use of cervical distraction.¹¹²

Renal Abnormalities

More than one third of children with Klippel-Feil syndrome can be expected to have a significant urinary tract anomaly. These anomalies are often asymptomatic in young children. Previously, the authors recommended that such anomalies be evaluated with intravenous pyelography (Fig. 30–46).¹¹² It has been found, however, that ultrasonography offers a noninvasive way to screen adequately for the anomalies associated with this syndrome.²⁰¹ The pronephros, the embryologic tissue destined to become the genitourinary tract, develops between the 7th and 14th somites, in the same region and at the same time as the cervical spine,^212,214 quite similar to the scapulae in Sprengel deformity. The most frequent abnormality is unilateral absence of a kidney. Other abnormalities include a double collecting system, renal ectopia, horseshoe kidney, and hydronephrosis from ureteropelvic obstruction. In the authors’ series, 2 of 50 patients developed severe pyelonephritis in the remaining kidney, requiring renal transplantation.¹¹² In Klippel and Feil’s original case report, the patient died of nephritis.¹⁶³

FIGURE 30–46 Radiograph of a 10-year-old child with Klippel-Feil syndrome shows multiple vertebral anomalies, unilateral absence of kidney, and hydroureter. Ureteral reimplantation was required for ureteral reflux and hydronephrosis.

Treatment

A patient with Klippel-Feil syndrome with minimal involvement can be expected to lead a normal active life with no or only minor restrictions or symptoms. Many patients with severe involvement can have the same good prognosis if early and appropriate treatment is instituted when needed; this is particularly applicable in the area of associated scoliosis and renal abnormalities. Prevention of further deformity or complications can be of great benefit. The actual treatment of Klippel-Feil syndrome is confined mostly to the area of associated conditions and is discussed under the respective headings.

At present, treatment choices for the cervical spine anomalies are quite limited. Patients with major areas of cervical synostosis or high-risk patterns of cervical spinal motion should be strongly advised to avoid activities that place stress on the cervical spine. In these patients, the mobile articulations are under greater mechanical demands and are less capable of protecting them against traumatic insults.

As discussed, sudden neurologic compromise or death after minor trauma has been reported in patients with Klippel-Feil syndrome and is usually due to disruption at the hypermobile articulation.^22,166,183 The role of prophylactic surgical stabilization in asymptomatic patients has not yet been defined. There is no satisfactory definition of when the risk of instability warrants further reduction of neck motion.

For symptomatic patients with mechanical problems, the usual treatment measures for degenerative osteoarthritis are applicable and include traction, a cervical collar, and analgesics. Symptoms that suggest neurologic compromise require careful consideration and evaluation by a neurologist, neurosurgeon, and orthopaedist (see Fig. 30–44). The exact area of irritation must be determined before surgical intervention. Attempts should be made preoperatively to obtain reduction of the bony architecture in advance of surgical stabilization. Also, the anesthesiologist may need to plan for airway management and be prepared for difficult intubation.²¹⁷ The physician must be mindful that there are other associated abnormalities in the brainstem and in the spinal cord itself that may be contributing to the symptoms.

Treatment of the cosmetic aspects of this deformity has met with limited success. Occasionally, children with the fixed torticollis posture may be improved with bracing. Bracing requires long-term application, however, and excellent patient cooperation. Correction of the bony deformity by direct means, such as wedge osteotomy or hemivertebrae excision, has been done on a limited basis.²¹⁸ Ruf and colleagues²¹⁸ reported hemivertebrae resection for correction of head tilt. Occasionally, carefully selected patients who have cervical congenital scoliosis may obtain some correction and improvement of appearance by use of the halo vest combined with posterior cervical fusion. Bonola²¹⁹ described a method of rib resection to attain apparent increase in neck length and motion. This procedure is an extensive surgical experience, however, and is a great risk to the patient. No subsequent reports have appeared in the literature. More recently, the vertical expandable prosthetic titanium rib (VEPTR) has been used to correct cervical tilt along with congenital thoracic scoliosis in thoracic insufficiency syndrome.²²⁰ Early results of this treatment have been promising.

Soft tissue procedures, Z-plasty, and muscle resection may achieve cosmetic improvement in properly selected patients.¹⁷¹ These procedures can restore a more natural contour to the shoulders and neck and an apparent increase in neck length. Neck motion is generally not increased, and the scars may be extensive, particularly in a patient with a large skin web. If an omovertebral bone is present or an abnormal connection to the clavicle,¹⁸⁵ its removal may permit an increase in neck and shoulder motion. The risk of brachial plexus injury from traction is higher in patients with Klippel-Feil syndrome because there are likely to be anomalous origins of the cervical nerve roots in these patients. Iniencephaly and absence of the posterior cervical elements (see Fig. 30–28) may be associated with Sprengel deformity and must be identified if surgical correction is considered.¹⁸¹

Summary

Klippel-Feil syndrome is an uncommon condition caused by congenital fusion of two or more cervical vertebrae. Most affected individuals are asymptomatic or have a mild restriction of neck motion. If symptoms referable to the cervical spine occur, it is usually in adult life and is due to degenerative arthritis or instability of the hypermobile articulations adjacent to the area of synostosis. Most patients respond to conservative treatment measures; a small percentage require judicious surgical stabilization. Cosmetic surgery is of limited benefit in treatment of the neck deformity.

The relatively good prognosis of the cervical lesion is overshadowed by the “hidden” or unrecognized associated anomalies. The high incidence of significant scoliosis, renal anomalies, deafness, neurologic malformations, Sprengel deformity, and cardiac anomalies should be of great concern to the physician. Early recognition and treatment of these problems may be of substantial benefit, sparing the patient further deformity or serious illness.

Etiology

At present, congenital muscular torticollis is believed to be the result of local compression of the soft tissues of the neck at the time of delivery.²²⁸ It is usually unilateral, but bilateral torticollis has been reported.²²⁴ Birth records of affected children show a preponderance of breech or difficult deliveries or primiparous births.^162,229 The deformity has occurred after otherwise normal deliveries, however, and has been reported in infants born by cesarean section.^221,229 Davids and colleagues²²⁸ found that MRI studies of the sternocleidomastoid muscle in infants with congenital muscular torticollis have signal changes similar to changes observed in the forearm and leg after compartment syndrome. That finding, coupled with cadaveric dissections, shows that there is a fascial covering of the entire sternocleidomastoid muscle; a review of the children treated by Davids and colleagues²²⁸ strongly suggests that congenital muscular torticollis represents a compartment syndrome.

Microscopic examinations of resected surgical specimens and experimental work with dogs²³⁰ suggest that the lesion is due to occlusion of the venous outflow of the sternocleidomastoid muscle. This occlusion results in edema, degeneration of muscle fibers, and eventual fibrosis of the muscle body. Coventry and Harris²²³ suggested that the clinical deformity is related to the ratio of fibrosis to remaining functional muscle. If sufficient normal muscle is present, the sternocleidomastoid stretches with growth, and the child is not likely to develop the torticollis posture, whereas if there is a predominance of fibrosis, there is very little elastic potential.

Pathologic studies showed that with time the fibrosis of the sternal head may entrap and compromise the branch of the accessory nerve to the clavicular head of the muscle (progressive denervation), leading to a late increase in the deformity.²³¹ Some evidence suggests that the problem may be due to uterine crowding or “packaging syndrome” because in three of four children the lesion is on the right side.^221,228,229 Also, 5% to 20% of children with congenital muscular torticollis have congenital dysplasia of the hip, which is believed to be due to restriction of infant movement in the tight maternal space.^232–234 Boys with developmental dysplasia of the hip were nearly five times more likely to have congenital muscular torticollis.²³³ Radiographs of the cervical spine should be obtained to rule out congenital anomaly of the cervical spine or C1-2 rotatory subluxation.^235,236

Ultrasonography is now widely accepted to be one of the most sensitive tests for the diagnosis of congenital muscular torticollis and developmental hip dysplasia. Tien and colleagues²³⁷ found that a homogeneous, hyperechoic mass within the sternocleidomastoid muscle was diagnostic of congenital muscular torticollis. They found the coexistence of developmental hip dysplasia and torticollis to be 17%; 8.5% of children with identifiable hip dysplasia required some treatment.

Treatment

Surgery

If the condition persists beyond 1 year of age, nonoperative measures are rarely successful.²³⁸ Similarly, established facial asymmetry and limitation of normal motion of more than 30 degrees usually preclude a good result and lead to surgical intervention and poor cosmesis (Fig. 30–50).²³⁸ A good (but not perfect) cosmetic result can be obtained, however, in children 5 to 7 years of age (Fig. 30–51).^{223,236,239,243,244} Asymmetry of the skull and face improves as long as adequate growth potential remains after the deforming pull of the sternocleidomastoid is removed (Fig. 30–52).^223,236,243 Shim and colleagues²⁴⁵ showed that patients older than 8 years, including patients who were skeletally mature, benefit from unipolar or bipolar release of the sternocleidomastoid and postoperative physical therapy. These authors preferred to wait until the child was old enough to comply with the postoperative exercise and bracing.²⁴⁶ All patients had improved functional and cosmetic results postoperatively.

FIGURE 30–50 A, Diagram showing abnormal appearance of sternocleidomastoid (SCM) in congenital muscular torticollis. B, Ultrasonography shows homogeneous, hyperechoic mass on SCM.

(Adapted from Tien YC, Su JY, Lin GT, et al: Ultrasonographic study of the coexistence of muscular torticollis and dysplasia of the hip. J Pediatr Orthop 21:343-347, 2001.)

FIGURE 30–51 A, Clinical appearance of a 6-year-old child with congenital muscular torticollis. Note appearance of two heads of sternocleidomastoid (arrows). B, Operative exposure of the same patient shows complete replacement with fibrous tissue of two heads of sternocleidomastoid.

FIGURE 30–52 An 18-year-old patient after resection of portion of sternocleidomastoid. Incision was placed too near clavicle and consequently spread, becoming cosmetically unacceptable.

Surgery consists of resection of a portion of the distal sternocleidomastoid muscle (see Fig. 30–50). At least a 1-cm segment of the tendon should be removed to guard against recurrence of the deformity. A transverse incision is made low in the neck to coincide with a normal skin fold.²⁴⁷ It is important not to place the incision near or very near the clavicle because scars in this area tend to spread and are cosmetically unacceptable (see Fig. 30–51).¹²¹ Similarly, closure with a subcuticular suture is preferred.¹⁶⁶

The most common postoperative complaint is disfiguring scarring.^221,229,247 The two heads of the sternocleidomastoid are identified, and both are sectioned (see Fig. 30–50). It is important to release the investing fascia around the sternocleidomastoid because this, too, is frequently contracted.²²¹ Rotation of the chin and head at this point generally reveals the adequacy of the surgery, and palpitation of the neck shows any extraneous tight bands that could lead to partial recurrence of incomplete correction (see Fig. 30–52).²²⁹ In an older child, an accessory incision (bipolar) is often required to section the muscle at its origin on the mastoid process. The whole muscle should not be excised because this may lead to reverse torticollis²²⁹ or additional deformity from asymmetry in the contour of the neck.²²¹

The postoperative regimen includes passive stretching exercises performed in the same manner as those done preoperatively. The exercises should begin as soon as the patient can tolerate manipulation of the neck. Occasionally, head traction at night is helpful, particularly in an older child. Bracing or cast correction may be necessary if the deformity has been of long duration or if the torticollis posture has become a strong habit.²⁴⁸ Results of surgery have been uniformly good, with a low incidence of complications or recurrence, and almost all patients are pleased with the results.^{221,223,229,238} Slight restriction of the neck motion and anomalous reattachment occur frequently^229,247 but are generally unnoticed by the patient. If the patient is young, the facial asymmetry can be expected to resolve completely, unless there is persistence of the torticollis, particularly from residual fascial bands (Fig. 30–53; see Fig. 30–52).²²⁹

FIGURE 30–53 A 23-year-old man underwent release of sternocleidomastoid. There is residual fascial band (arrow) and slight restriction of neck motion.

Differential Diagnosis

Torticollis is a common childhood complaint. The etiology is diverse, and identifying the cause can pose a difficult diagnostic problem (Tables 30-2 and 30-3).

TABLE 30–2 Differential Diagnosis of Torticollis

TABLE 30–3 Torticollis Caused by Bony Anomalies

Congenital muscular torticollis is the most common cause of wry neck posture in infants and young children, but there are other problems that lead to this unusual posture. Head tilt and rotatory deformity of the head and neck (torticollis) usually indicate a problem at C1-2, whereas head tilt alone indicates a more generalized problem in the cervical spine. If the posturing of the head and neck is noted at or shortly after birth, congenital anomalies of the cervical spine should be considered. Bony anomalies of the cervical spine, particularly anomalies that involve C1-2, typically manifest as a rigid deformity, and the sternocleidomastoid muscle is not contracted or in spasm.

Gyorgyi²⁴⁹ examined 20 cases of congenital torticollis and found the following coexisting anomalies: congenital cervical fusions, asymmetrical facet joints, basilar impression, atlantoaxial dislocation, assimilation of the atlas, and deformities of the odontoid process. Gyorgyi²⁴⁹ noted that in children with congenital torticollis, 40% had a history of breech presentation. Many occipitocervical malformations manifest as torticollis.^31,33,38,112 De Barros and colleagues¹⁴ noted that 68% had basilar impression, most commonly unilaterally. Approximately 20% of patients with Klippel-Feil syndrome have associated torticollis.^112,250 With asymmetrical development of the occipital condyles or the facets of C1, the head tilt may result in a torticollis unless compensated for by a tilt of the lower cervical spine similar to that which occurs in the milder forms.^14,103,166

If torticollis is noted in the weeks after delivery, the usual cause is congenital muscular torticollis. If the child is younger than 2 months of age, a palpable lump may be found in the sternocleidomastoid. Congenital muscular torticollis is painless, is associated with a contracted or shortened sternocleidomastoid muscle, and is unaccompanied by any bony abnormalities or neurologic deficit. Soft tissue problems are less common and include abnormal skin webs or folds (pterygium colli), which maintain the torticollis posture. Tumors in the region of the sternocleidomastoid, cystic hygroma, branchial cleft cyst, and thyroid teratoma are rare but should be considered.

Inflammatory conditions include local irritation from cervical lymphadenitis, which may lead to the appearance of a wry neck or tilt of the head. Another less frequent cause is a retropharyngeal abscess after inflammation of the posterior pharynx or tonsillitis.²⁵¹ Children with polyarticular juvenile rheumatoid arthritis frequently develop involvement of the cervical joints. Torticollis and limitation of cervical motion may be the only clinical signs. Spontaneous atlantoaxial rotatory subluxation may follow acute pharyngitis.^250,252 Radiographic confirmation is difficult, and the CT methods for evaluation suggested by Phillips and Hensinger²⁵² should be used. Early diagnosis and reduction of the displacement are important.^252,253 If it becomes fixed, it poses a considerable treatment problem.²⁵² A rare inflammatory cause is acute calcification of a cervical disc,^254,255 which can be visualized on routine radiographic study of the neck (Fig. 30–54).

FIGURE 30–54 A 6-year-old girl presented with acute onset of torticollis and neck pain but no history of trauma or recent infection. Lateral radiograph of cervical spine shows acute calcification of disc between C3 and C4 (arrow). The child was treated conservatively with a neck collar, and there was spontaneous resolution of torticollis and symptoms over a 2-week period. Disc calcification was still radiologically visible at 6 months after onset of symptoms but not at 12 months.

Traumatic causes should always be considered and carefully excluded early in the evaluation. If unrecognized, they may have serious neurologic consequences. Generally, torticollis most commonly follows injury to the C1-2 articulation. Minor trauma can lead to spontaneous C1-2 subluxation. Fractures or dislocation of the odontoid may not be apparent in the initial radiographic views (see Fig. 30–18), and consequently a high index of suspicion and careful follow-up are required. Children with bone dysplasia, Morquio syndrome, spondyloepiphyseal dysplasia, and Down syndrome have a high incidence of C1-2 instability and should be evaluated routinely.

Intermittent torticollis can occur in a young child. A seizurelike disorder termed benign paroxysmal torticollis of infancy has many neurologic causes, including drug intoxication. Similarly, Sandifer syndrome, involving gastroesophageal reflux with sudden posturing of the trunk and torticollis, is being recognized more often, particularly in neurologically handicapped children, such as children with cerebral palsy.²⁵⁶

Neurologic disorders, particularly space-occupying lesions of the central nervous system, such as tumors of the posterior fossa or spinal column, chordoma, and syringomyelia, are often accompanied by torticollis. Generally, there are additional neurologic findings, such as long tract signs and weakness in the upper extremities. Uncommon neurologic causes include dystonia musculorum deformans and problems of hearing and vision that can result in head tilt.²⁵⁷ Although uncommon, hysterical and psychogenic causes exist, but these should be diagnosed only after other causes are carefully excluded.

Radiographic Features

All children with torticollis should be evaluated with radiography to exclude a bony abnormality or fracture. Radiographic interpretation of congenital torticollis may be difficult because of the fixed abnormal head position and the restricted motion. In children with a painful wry neck, it may be impossible to position them appropriately for a standard view of the occipitocervical junction. A helpful guide is that the atlas moves with the occiput; if the x-ray beam is directed 90 degrees to the lateral skull, a satisfactory view of the occipitocervical junction usually results. Flexion-extension stress films, dynamic CT scan, or cineradiography may be necessary to confirm atlantoaxial instability. The bony anomalies that may be present in congenital torticollis are documented in the sections on anomalies of odontoid, Klippel-Feil syndrome, and basilar impression.

Rotatory subluxation of C1 and C2 presents a unique problem. Plain radiographs seldom differentiate the position of C1 and C2 during subluxation from that in a normal child whose head is rotated because both give the same picture. Open-mouth views have been difficult to obtain and interpret. Lack of cooperation and decreased neck motion on the part of the child can make it impossible to obtain these special views. Fielding and Hawkins²⁵⁰ recommended cineradiography, but the radiation dosage is relatively high, and patient cooperation may be difficult to obtain because of muscle spasm.

The normal relationship between the occiput and the atlas is thought to be rarely affected in atlantoaxial rotatory subluxation. A lateral radiograph of the skull may show the relative position of the atlas and axis more clearly than a lateral radiograph of the cervical spine, in which tilting of the head also tilts the atlas, and overlapping shadows make interpretation difficult (Fig. 30–55). Similarly, if during CT scans the child is in the torticollis position, the image may be interpreted by the radiologist as showing rotation of C1 and C2. Conversely, in a child with rotatory subluxation, the rotation of C1 and C2 may be within the range of normal, as is usually the case. Early in this condition, the radiologist may attribute the finding to patient positioning.²⁴⁵ This dilemma can be resolved by paying close attention to proper positioning of the patient, obtaining CT cuts at the level of C1, and rotating the head to the right and left and showing the facets to be locked in that position.²⁴⁵

FIGURE 30–55 Obtaining a satisfactory radiograph may be hampered by the patient’s limited ability to cooperate; fixed bony deformity; and overlapping shadows from mandible, occiput, and foramen magnum. A helpful guide is that the atlas moves with the occiput; if x-ray beam is directed 90 degrees to lateral of skull, satisfactory view of occipitocervical junction usually results.