From a perspective of embryology, developmental anatomy, and spinal biomechanics, the upper cervical spine is unique compared with the rest of the spine1. This region is defined as extending from the occiput to the C2-C3 disc space2-4. Congenital osseous anomalies in this region can lead to an increased risk of segmental instability and spinal cord encroachment, which may in turn have neurologic implications. Commonly reported anomalies in this region include Arnold-Chiari malformation, platybasia, basilar invagination, occipitalization of the atlas, a dysmorphic atlas5, a dysmorphic dens5, os odontoideum, vertebral intersegmental fusion, spinal canal stenosis, and segmental instability of the occipitoatlantal and atlantoaxial articulations4,6-10.
Vertebrae develop from the cells induced from somites to surround the notochord and the neural tube. Somites initially appear at 3.5 weeks of gestation, and the full complement of somites is approximately thirty-eight to thirty-nine pairs (four occipital, eight cervical, twelve thoracic, five lumbar, five sacral, and four to five coccygeal). As proposed with the resegmentation theory, during metameric shift, the superior portion of each sclerotome divides and combines with the caudal portion of the sclerotome above. Combined regions develop into a vertebra, whereas the intervertebral disc develops where the sclerotomes divide.
The occiput, atlas, and axis are formed by a separate mechanism from that responsible for the other vertebral bodies. Four sclerotomes are involved in the formation of the occiput, while a part of the first cervical (C1) sclerotome and the cranial portion of the C2 sclerotome contribute cells to the odontoid process and the arch of C1. The centrum of C1 contributes to the odontoid. Also, part of the fourth occipital sclerotome contributes to the atlas. The remaining subaxial cervical vertebrae develop in a manner similar to the rest of the spine4,8.
In describing anomalous development of the upper cervical spine, it is important to recognize variants that can be attributed to normal development. For example, normal closure of the posterior arch of the atlas may not occur until well after birth1,11. Typically, there is a remnant cartilaginous cleft between the osseous posterior arches at birth. An ossification center develops in this cartilaginous cleft around the second year of life, and the posterior arch is usually completely ossified by the age of four years. It is important to keep in mind that complete ossification of the dens, or odontoid process, may not occur until twelve years of age1,11.
Anomalies and variations may occur in the upper cervical spine, either as part of a developmental syndrome or as an isolated congenital anomaly. We previously reported the upper cervical spine anomalies noted in the 22q11.2 deletion syndrome5, the spinal cord dimensions in the Klippel-Feil syndrome12, as well as variations of occipitalization of the atlas13. A large descriptive review of the different osseous anomalies of the upper cervical spine and their frequency in two different cohorts (syndromic compared with nonsyndromic children) has not previously been reported. The purposes of this study were threefold: (1) to determine how children with congenital osseous anomalies of the upper cervical spine present to the orthopaedic surgeon, (2) to evaluate and compare the constellation of upper cervical anomalies present in syndromic and nonsyndromic patient cohorts, and (3) to determine the relationship between these anomalies and the development of instability and/or spinal cord encroachment with possible neurologic implications.
We reviewed the clinical and radiographic records of all children referred to us with anomalies of the upper cervical spine and a clinical problem (either symptoms or signs) between July 1988 and November 2003. Both institutional review board approval and informed consent were obtained. Children with syndromic abnormalities were seen in a multidisciplinary setting. Each patient who had been newly diagnosed as having any syndrome had a comprehensive evaluation by multiple departments.
The patients in the current study underwent an assessment of the occiput and cervical spine in the Division of Orthopaedic Surgery, either as part of the initial evaluation or at a subsequent follow-up appointment. All patients who were less than six months of age were excluded from the current study because occipitocervical findings in this age-group cannot be accurately assessed on radiographs.
Review parameters included (1) demographic data, (2) clinical presentation including a detailed history and physical examination, (3) findings on imaging including radiographs (anteroposterior, lateral, and open mouth views), dynamic lateral radiographs done in maximum cervical flexion and extension, and (4) findings on advanced imaging (computed tomography and/or magnetic resonance imaging). The clinical evaluation was focused on the neurologic symptoms and signs potentially related to the occiput and/or cervical spine. Radiographs were made at the time of the clinical evaluation of each patient. The target-to-film distance was standardized for all radiographs to minimize magnification errors and errors in measurement. The lateral radiographs (neutral, flexion, and extension) were performed in the erect position with the use of a 183-cm target-to-film distance.
Patients were divided into two cohorts: syndromic and nonsyndromic. The most common syndromes included the 22q-deletion syndrome (five patients), spondyloepiphyseal dysplasia (three), Down syndrome (two), and Goldenhar syndrome (two). Radiographs and advanced imaging studies were evaluated to define anomalies in the brain stem; the occipitocervical junction; and the upper cervical osseous canal, including stenosis, instability, and spinal cord anomalies such as canal encroachment. All radiographic evaluations were examined by four of the authors independently, followed by a group review of the findings to resolve any inconsistencies. In cases in which there was a disparity in findings or measurements, agreement was reached by consensus or the values were averaged. The relationship between upper cervical spine instability and spinal cord encroachment and the development of possible myelopathy was evaluated. Data were tabulated, and statistical analysis was performed with use of the Fisher exact test for double cohort analyses.
Radiographic Definitions
The radiographic features that were characterized and assessed included dysmorphic atlas (C1), occipitalization of the atlas, os odontoideum, congenital fusion of C2-C3, basilar invagination, Chiari malformation, segmental spinal stenosis, occipitoatlantal instability (O-C1), and atlantoaxial instability (C1-C2). Canal encroachment was defined and categorized as static, dynamic, and combined. Other rare anomalies were also recorded when observed.
Dysmorphic Atlas5
Because there are no defined criteria for a hypoplastic atlas in the literature, we used our own previous description of dysmorphic atlas5 and included the unusually small and thin atlas (either in part or the total ring) with a variety of other atlas variants such as those with an unusual shape.
Occipitalization of the Atlas
Congenital occipitoatlantal fusion was defined by the evidence of an osseous fusion (anterior, posterior, lateral, or combined) of the atlas and the occiput on lateral cervical spine radiographs and advanced imaging studies14 (Fig. 1) with use of the protocol outlined in our previous study13. Only when the atlas and occiput moved as a functional unit with no evidence of motion between the two on flexion and extension radiographs, with confirming findings on computed tomography and/or magnetic resonance imaging scans, was this diagnosis made.
Os Odontoideum
Os odontoideum refers to an independent osseous structure positioned cephalad to the body of the axis where the normal continuous odontoid process should lie14. The position of this structure can be orthotopic (normal) or dystopic (abnormal). When suspected from the radiographs, os odontoideum was confirmed by a computed tomography and/or magnetic resonance imaging scan (Figs. 2-A and 2-B).
C2-C3 Fusion
Congenital fusion of the C2-C3 vertebrae was defined as either an incomplete fusion involving the posterior vertebral elements only or as a complete block fusion including the vertebral bodies. A fusion of the posterior elements only was observed as a narrowing or elimination of the gap between these elements and was confirmed by the movement of the spinous processes of C2 and C3 as a functional unit on neutral, flexion, and extension lateral radiographs (Fig. 3). Block fusion was confirmed by evidence of fusion of both the posterior elements and the vertebral bodies of C2 and C3 (Fig. 4). Congenital fusion was also confirmed by advanced imaging studies in every case.
Basilar Invagination
Basilar invagination historically has been measured by the radiographic criteria defined by Chamberlain, McGregor, McRae and Barnum, and Fischgold and Metzger14-17. However, reference points near the base of the skull are not always easy to identify on radiographs, so computed tomography and magnetic resonance imaging scans were used to confirm the diagnosis in the present study. A sagittal reformatted computed tomography scan showing the tip of the odontoid at or above the level of the foramen magnum was accepted as confirmation of basilar invagination. On magnetic resonance imaging scans, a midsagittal section demonstrating the tip of the odontoid at or above the level of the foramen magnum or at a short and horizontal clivus with the margin of the foramen magnum above the floor of the posterior fossa was accepted as confirmation of basilar invagination18.
Chiari Malformation19,20
With this anomaly, there is a herniation of the cerebellar tonsils that are displaced caudally through the foramen magnum by =5 mm. When they are displaced <5 mm, the anomaly is described as tonsillar ectopia. The Chiari malformation was diagnosed on a midsagittal magnetic resonance imaging scan (Fig. 5).
Spinal Canal Stenosis
Narrowing or stenosis of the upper cervical spinal canal was diagnosed on magnetic resonance imaging when minimal cerebral spinal fluid was seen on sagittal cuts. Values for the Pavlov-Torg ratio are not available for C2 and cephalad levels in the pediatric population, and therefore this method of determining cervical stenosis was not available. Torg ratios are used for the C3 levels and those more caudal21.
Occipitoatlantal Instability
In the adult literature, occipitoatlantal instability has been defined as >1 mm of translation at the occipitoatlantal articulation with use of the Wiesel-Rothman technique22. Translation is measured from the anterior surface of the occipital condyles to the posterior surface of the anterior arch of the atlas on flexion and extension radiographs22. Given that the cervical spine in children is more mobile than that in adults, we chose the more rigorous cutoff of =2 mm of translation (using the same landmarks) to define increased occipitoatlantal motion5. One millimeter can be difficult to measure accurately on radiographs and may therefore have an element of interobserver variability that was not addressed in this study. Translation is best observed on the extension radiograph1. The Powers ratio was used to confirm the findings in all patients.
Atlantoaxial Instability
Atlantoaxial instability in children is defined by an atlantodens interval of >4 mm as observed on lateral radiographs of the cervical spine in flexion and extension23. The atlantodens interval is the distance between the anterior aspect of the dens and the posterior aspect of the anterior arch of the atlas. This criterion was used to determine atlantoaxial instability in the current study (Fig. 6). For the patients with Down syndrome, the criterion used for instability was an atlantodens interval of >7 mm.
Canal Encroachment
We defined canal encroachment as a reduction (=13 mm) in the space available for the spinal cord or brainstem. Canal encroachment may be due to static or dynamic factors, or both (Fig. 7). Static encroachment includes stenosis that occurs as the result of focal osseous narrowing of the spinal canal, a space-occupying mass such as an intrusion of the cerebellar tonsils with Chiari malformation, or the odontoid protruding through the foramen magnum with basilar invagination. Dynamic encroachment occurs with segmental instability that causes translational encroachment of the space available for the spinal cord.
The space available for the cord or brainstem is measured from the posterior aspect of the odontoid to the anterior aspect of the posterior arch of the atlas1,24. At the level of C2, it is measured from the posterior aspect of the body of C2 to the anterior aspect of the posterior arch of the axis.
Other Anomalies
Although the current study focused on the anomalies listed above, other osseous anomalies were recorded when they were observed. For subaxial (C2-C3) segmental motion, translation of =3 mm was considered as instability.
The embryological explanation for the variations of the upper cervical spine that we noted is unclear, but the abnormalities likely arise during somatogenesis, when errors in cellular proliferation, migration, differentiation, and segmentation can occur25. Failure of segmentation between the fourth occipital and first cervical somites, for example, may cause variable fusion of C1 to the occiput and result in occipitalization of the atlas (Fig. 1). Fusions of C2-C3 also represent segmentation errors between the corresponding cervical somites during the third to the eighth week of fetal life. In younger patients, failure of segmentation may not be evident as an osseous fusion initially because of incomplete ossification, but it may be seen as a reduced disc space or a narrow interval between the spinous processes and posterior elements. Fusions can be confirmed prior to complete ossification by lateral radiographs of the cervical spine in flexion and extension that demonstrate the spinous processes of adjacent vertebrae moving as a functional unit, with no change in the distance between them (Figs. 3 and 4)1. We consider these radiographs essential not only for verification of vertebral fusions (particularly fusions of the posterior vertebral elements) but also for the evaluation of increased segmental motion at adjacent vertebral levels.
In a series consisting primarily of adults, McRae was among the first to report on anomalies of the craniovertebral junction and to find that they have an important association with the development of neurologic sequelae14,26,27. McRae reported neurologic compromise in more than sixty patients with various anomalies including occipitalization of the atlas, basilar invagination, os odontoideum, and chronic atlantoaxial dislocation. Neurologic signs and symptoms, including headache, neck pain, visual and auditory symptoms, weakness and numbness in the extremities, long tract and posterior column signs, ataxia, and nystagmus, were frequently observed. In 1964, Bharucha and Dastur reported similar findings in a series of forty adult and pediatric patients with craniovertebral anomalies28.
Craniovertebral anomalies have been noted to be associated with congenital syndromes, skeletal dysplasias, and metabolic conditions. In our series, twenty-eight of the sixty-eight patients had a syndromic or presumed syndromic diagnosis, with five of the twenty-eight having the 22q11.2 deletion syndrome.
Neck pain and stiffness have been reported to be a presenting complaint in patients with upper cervical anomalies. Bharucha and Dastur28, in their study of forty patients, reported that sixteen patients presented with pain and/or stiffness and ten of these were seen following a minor hyperextension injury of the neck. McRae reported this finding in eleven of twenty-eight patients26. In our series of sixty-eight patients, thirty-seven presented with neck pain and/or stiffness and ten of them noted trauma to be a precipitating factor. It is important to note that neck pain may be absent in many patients with congenital osseous anomalies.
Neurological symptoms in patients with osseous anomalies of the upper cervical spine may be absent initially only to present later in the third and fourth decade of life14,26-30. McRae reported neurological symptoms in nineteen (68%) of twenty-eight patients who had a mean age of thirty-one years (range, seven to forty-five years) at the time of onset26. Fourteen patients presented with motor weakness, eleven had sensory complaints, and thirteen had spasticity. Erbengi and Oge reported neurological symptoms in fourteen of twenty patients at a mean age of onset of 32.5 years (range, sixteen to fifty-seven years)29. In the current study, twenty-one (31%) of sixty-eight pediatric patients had neurological symptoms and the mean age at the time of onset was eight years (range, one month to twenty-four years). On the basis of the earlier reports14,26-30 with cohorts that included mainly adult patients, we expect the incidence of myelopathy in our patients to increase with time. It is imperative to advise families with regard to the natural history of these abnormalities and the risk of myelopathy increasing with time. At the very least, there is the need for a periodic follow-up with both clinical and radiographic examinations. Although the abnormalities in some patients might indicate an option for prophylactic surgical intervention, surgical treatment was not done in the patients in the present study in the absence of clinical signs of myelopathy or instability.
Overall, atlantoaxial instability was noted in thirty-four patients (50%). Twenty-one of them had an associated C2-C3 fusion and thirteen did not. Atlantoaxial instability was also observed by McRae in fifteen patients (53%), and eleven of them had an associated C2-C3 fusion26. Occipitalization with an associated C2-C3 fusion appears to place further strain on the atlantoaxial articulation during flexion and extension26,28,31 and likely predisposes to atlantoaxial instability.
We observed basilar invagination in twenty-five patients and a Chiari type-I or II malformation in twelve patients. Four patients in the present study had an open posterior arch of the atlas, a finding reported by Geipel in 4% of autopsy specimens from adults32. We are unaware of the relevance of this finding, other than the fact that knowledge of its existence is important when surgery with a posterior approach is planned. McRae26 noted posterior angulation of the odontoid in six patients (24%). In our series, five of sixty-eight patients had posterior angulation of the odontoid, although we are unsure of its exact clinical relevance5. Two patients in the current study had an os odontoideum with atlantoaxial instability.
In the current study, a total of 229 anomalies of the upper cervical spine were observed in both cohorts, for an average of 3.4 anomalies per patient, and there was no significant difference between the two cohorts with regard to any specific anomaly or the average number of anomalies observed. The results show that multiple anomalies of the upper cervical spine are common within a single patient, so when a single anomaly is seen in a patient, others should be sought. In an attempt to analyze the effect of underlying genetic variation as a cause for the development of upper cervical spine anomalies, we separated our patients into syndromic and nonsyndromic cohorts. The fact that there was no difference in the frequency of any single anomaly or in the number of anomalies between the two cohorts was unexpected and counterintuitive to us. From a developmental standpoint, this finding suggests a similar developmental pathway for both cohorts, but defining the pathway more fully is beyond the scope of this study.
Biomechanical factors at the craniocervical junction play a large role with regard to injury in children. First, the head in the very young child is relatively large and heavy, and the forces generated with rapid acceleration in flexion and extension are large compared with those occurring in adults1. Additionally, the muscles and ligaments are weaker and less able to resist these acceleration loads1. By adding congenital spinal anomalies with encroachment of the available space for the spinal cord and brainstem, the risk for neural injury may be increased.
Historically, narrowing of the osseous spinal canal has been described as canal stenosis24,30,31,33. The term canal stenosis is inadequate to include all conditions that decrease the space available for the cord, and hence we defined our concept of canal encroachment. With use of this concept, encroachment of the spinal canal may be due to static or dynamic factors, or both (Fig. 7). Static encroachment includes focal osseous narrowing of the spinal canal (stenosis), a space-occupying mass such as a herniation of the cerebellar tonsils with the Chiari malformation, or intrusion of the odontoid through the foramen magnum with basilar invagination. Another cause of canal encroachment occurs in adults because of degenerative changes and is a possible explanation for the delayed onset of the neurological sequelae reported previously26,29. Dynamic encroachment occurs when segmental translational instability encroaches on the space available for the cord. When they occur alone, either of these situations can increase the risk for neurologic injury, but when they occur in combination the risk is greater, particularly when static and dynamic encroachment exist simultaneously. With time and degenerative changes of the spine, a further deterioration of the neurologic situation may result. Fifty-eight (85%) of the sixty-eight patients in the study were deemed at risk for the development of myelopathy or had had compressive myelopathy develop. In the complex pathology of the cervical spine, the encroachment concept has helped us to better evaluate the issues, explain them, and develop a plan for management. We believe that this concept is helpful in surgical planning (including decompression and/or arthrodesis), particularly when static and dynamic encroachments occur simultaneously.
A weakness of our study is that the cohort appears slanted toward surgical treatment, suggesting that most patients are symptomatic early. Likely, this can be explained by selection of the patient pool toward more involved and complex problems seen at a large pediatric tertiary specialty care center. A prospective study would be needed to provide a larger consecutive cohort that includes more asymptomatic patients. Because there was little difference in the incidence of anomalies and associated sequelae between the syndromic and nonsyndromic cohorts in our study, we believe that information could be provided from a prospective study of patients with the 22q11.2 deletion syndrome. Our preliminary retrospective study of patients with this disorder and congenital anomalies of the upper cervical spine suggests that this would be a suitable cohort for such a prospective review5.
Despite this limitation, the study does alert us that the anomalies observed are associated with a risk for neurologic problems. Although the neurologic problems might not be always apparent initially in pediatric patients, the symptoms frequently first occur in the third and fourth decade14,26-30.
On the basis of our experience and observations from this study, we developed an algorithmic approach to the investigation and management of patients with osseus anomalies at the craniocervical junction. Our goal was to identify the risk for neurologic injury by considering the tendency of a given anomaly alone, or in combination, to encroach on the space available for the spinal cord or the brain stem. As outlined in Figure 7, we defined encroachment of the space available for the spinal canal as static or dynamic narrowing, or a combination of both. With asymptomatic patients, we analyze dynamic lateral radiographs of the cervical spine in neutral, flexion, and extension to identify segmental instability. Atlanto-occipital instability is best observed on extension radiographs, and atlantoaxial instability is usually best seen on the flexion radiograph. If the dynamic studies are within normal limits, usually the patient can be managed with observation. If marginal instability is present, then a magnetic resonance imaging scan can help to decide whether the instability is associated with substantial risk for neurologic injury. For example, marginal translation in the presence of a capacious spinal canal may be well tolerated by the more sedentary patients, but the presence of both dynamic and static encroachment might suggest a more important risk worthy of strict activity restriction or even surgical management. To further clarify this finding in borderline situations and to better define the problem, we consider a dynamic magnetic resonance imaging scan, with sagittal cuts, done in flexion and extension, to visualize the translational narrowing of the canal related to the spinal cord or brainstem under dynamic situations. In patients with clinical symptoms and temporary or persistent signs of myelopathy, appropriate surgical treatment is often the course to take. 