33 Rheumatoid Arthritis of the Craniovertebral Junction: Contemporary Surgical Management
Rheumatoid arthritis (RA) is a chronic, relapsing, inflammatory autoimmune disorder characterized by symmetrical erosive synovitis of multiple peripheral joints with a varying degree of systemic involvement. RA is relatively common, affecting 0.8% to 2.0% of the world’s population and 2 million people in the United States alone.1 The cervical spine is the second most frequently affected region (after hands and feet).2 Women are affected more often than men (3:1), although men have a greater risk of advanced cervical spine involvement.3
Garrod first described RA of the cervical spine as a clinical entity.4 The prevalence of cervical spine involvement in RA ranges from 25% to 80%, depending on the diagnostic criteria applied.5 Three types of deformities occur commonly in rheumatoid cervical spine. Atlanto axial sub luxation (AAS) or instability is the most common type, accounting for nearly 65% of the deformities ( Fig. 33.1 ). Rostral migration of the dens (basilar invagination/cranial settling) is the second most common deformity, seen in 20% of patients with RA ( Fig. 33.2 ). Subaxial subluxation is the third most common pathology, occurring in 15% of patients ( Fig. 33.3 ).
Forty to 85% of patients with RA have neck pain. Radiographic evidence of instability is observed in a similar percentage of patients, although only 7% to 13% of them develop neurological deficits. Kauppi and Hakala found that in a Finnish population–based series of 98 patients with RA (mean disease duration ∼17 years), 33% of patients presented with AAS and 27% with atlantoaxial impaction (AAI).6 Pellicci et al. in 1981 reported that, although 80% of their patients with RA and cervical spine involvement demonstrated radiographic progression, 36% had neurological progression.7 Fifty percent of patients with radio-graphic instability were asymptomatic in the series by Collins et al.8 Approximately 10% of patients with RA die of unrecognized spinal cord or brainstem compression.9
The mutilating disease subset of patients with RA, as classified by Olerud et al.,10 usually have global involvement of the cervical spine with associated AAS, subaxial subluxation, and vertical subluxation. The natural course of the disease and the survival rates for this subset are also very discouraging, as shown in Omura’s series, where all patients (n 6) with the mutilating disease type of RA who were managed conservatively died within 3 years.11
Pathomechanics of the Rheumatoid Cervical Spine
No single factor is entirely predictive of progressive cervical spine disease in RA. Nevertheless, disease duration, rapid joint erosiveness, elevated C-reactive protein levels, seropositivity, HLA-DRB1 (human leukocyte antigen–DR beta 1) susceptibility, and severe peripheral joint involvement, including arthritis mutilans, are, in general, suggestive of a more aggressive disease process.12
It is important to understand the various classification systems used in the literature to grade functional capacity, neurological deficit, and pain in patients with craniovertebral junction/upper cervical spine pathology. With these schemes, the outcome from the variety of treatment modalities can be compared in a standardized fashion. The common classification schemes are listed in Tables 33.1, 33.2, 33.3, and 33.4 .
Predictors of Cervical Spine Involvement
The cervical spine, in particular, is affected in RA because of the large number of articulations and their significant mobility. Inflammation of the synovial membrane leads to pannus formation (overgrowth of hyaline cartilage and periarticular inflammation) ( Fig. 33.4 ), which in turn results in bone erosion and synovial cysts. This then leads to joint laxity and subluxation. In the cervical spine, it affects the atlantoaxial joint, periodontoidal ligaments, facet (zygo-apophyseal) joints, uncovertebral joints, retrodental bursa, interspinous ligaments, and ligaments around the atlas. The craniovertebral junction (CVJ) often becomes unstable because of the regional dependence on ligamentous structures for stability. The resulting hypermobility at occiput–C1 and C1–C2 is due to damage to the cruciform-alar ligament complex, the trans-verse ligament, and the joint capsules.15 Degenerative arthritic changes are superimposed on the CVJ, which remains hypermobile. Asymmetric wear results in the superimposition of degenerative arthritis on the region, especially the atlantoaxial articulation. This results in severe AAS, rotatory subluxation, dorsal and lateral AAS, craniocervical scoliosis, and basilar invagination. The atlas often slips over the axis in a ventral-caudal direction and causes a greater upward movement of the dorsal arch of the atlas, resulting in an increased anteverted C1–C2 kyphosis angle. The asymmetrical wear and subluxation ultimately result in the formation of a hypertrophic cicatrix, described as pannus. The pannus itself may be compressive, especially when it is magnified by coronal or sagittal plane malalignment at the CVJ. If these degenerative changes are symmetrical, the patient experiences symmetrical basilar invagination.16 When only one lateral mass is affected by the disease process, fixed rotational tilt of the head to the affected side may occur.
The most common deformity is AAS, representing 60% to 65% of rheumatoid cervical subluxations. The majority are ventral, 20% are lateral, and 10% are posterior.17 Posterior AAS is typically associated with erosion or fracture of the dens, and although this is thought to be more benign than anterior AAS, it actually carries a higher risk of spinal cord compression due to less restricted backward translation of the free-floating atlas.
In contrast, in the subaxial spine, the uncovertebral joints are the predominant focus of the inflammatory process. Disk collapse and autofusion/ankylosis are typical. Subluxation and pannus formation occur later, as the ligaments become involved. Subaxial subluxation usually occurs late in the course of the disease and tends to affect multiple vertebral levels, sometimes causing a classic “stepladder” or “staircase” deformity.
Cranial Settling or Vertical Translocation
Cranial settling is also termed basilar invagination, atlantoaxial impaction, or odontoid vertical migration. Horizontal AAS almost always precedes vertical translocation, usually by 6 years or so. If chronic inflammation affects both atlantoaxial facet joints, their cartilage and bony surfaces may be destroyed. The weight of the skull then forces the atlas downward over the axis, causing cranial settling ( Figs. 33.2 and 33.5 ). Vertical translocation, or ascent of the odontoid peg, is primarily due to C1 lateral mass collapse and may also be partly due to destructive changes in C2 and occipital condyles. Casey et al. reported from Crockard’s series that the physical constraint of the ring of the atlas as it slides down the axis leads to decreased mobility, thus in turn leading to decreased pannus formation and a decreasing atlantodens interval (ADI).16 Diminishing ADI thus should not be considered a sign of amelioration of the patient’s condition. In fact, it suggests increasing vertical translocation, heralding an increased bony compression of the brainstem and spinal cord. This may be associated with a shortened dens because of the erosive process affecting C2.
Histology of Affected Joints
Synovitis with fibrinoid changes of the cervical spine is similar to that of peripheral joints. The transverse ligament becomes insufficient because of inflammatory erosion of the dorsal surface of the dens granulation tissue originating in the synovial joints between the transverse ligament and the dorsal surface of the dens. This loss of tensile strength and stretching of the transverse ligaments initiates AAS. A similar inflammatory response of the synovium may affect the uncovertebral joints, retrodental bursa, facet joints, ligaments around the atlas, and even the intervertebral disk if invaded by synovium. O’Brien et al., in their histological analysis of tissue obtained from transoral resection of the dens, identified two types of synovium: type I, chronic active rheumatoid synovium, and type II, end-stage rheumatoid synovium.18 The latter was associated with decreased spinal cord cross-sectional area, myelopathy, and greater evidence of craniocervical junction osseous destruction. This suggests that ligamentous destruction is followed by replacement of the rheumatoid synovium with fibrous tissue, whereas the osseous structures reveal severe destruction secondary to mechanical stability, rather than to an acute inflammatory process. Therefore, it appears that mechanical degeneration rather than an inflammatory rheumatoid process is responsible for the deteriorating neurological picture in chronic RA at the CVJ.
Imaging in Rheumatoid Arthritis of the Spine
Cervical Spine Radiographs
Radiographs are usually the first line of imaging used to assess the degree of cervical spine involvement from RA. Neutral views, in combination with flexion/extension and rotation, are helpful in identifying and measuring atlantoaxial subluxation ( Fig. 33.6 ), basilar invagination, and subaxial subluxation, as well as for a rough assessment of the extent of osteoporosis. Various measurements, such as ventral ADI, dorsal ADI ( Fig. 33.7 ), and the space available for the spinal cord, can be determined from radiographs, although none are reliable to predict neurological deterioration or to assess the extent of spinal cord compression.
Indications for anteroposterior and lateral radiographs of the cervical spine in patients with RA include19
Prolonged cervical symptoms 6 months
Neurological signs or symptoms
Scheduled procedures requiring endotracheal intubation
Rapidly progressive carpal or tarsal bone destruction
Rapid overall functional deterioration
Evaluation of Basilar Invagination
McGregor Line
The McGregor line is drawn on a lateral plain radiograph from the hard palate to the base of the occiput. Vertical settling of the occiput is defined as migration of the odontoid 4.5 mm above the McGregor line ( Fig. 33.8 ).
Clark Station
The odontoid process is divided into three equal parts in the sagittal plane. If the ventral ring of the atlas is level with the middle third or the caudal third of the odontoid process, a diagnosis of basilar invagination is made ( Fig. 33.9 ).2
Redlund-Johnell Criterion
The distance between the McGregor line and the midpoint of the caudal margin of C2 is referred to as the RedlundJohnell criterion. If it measures 29 mm in women or 34 mm in men, it indicates basilar invagination ( Fig. 33.10 ).20
McRae Line
The McRae line is drawn across the foramen magnum from the basion to the opisthion. A protrusion of the tip of the dens above this line is suggestive of basilar invagi-nation ( Fig. 33.11 ).21
Chamberlain Line
The Chamberlain line is drawn from the dorsal edge of the hard palate to the opisthion. If the tip of the dens lies 3 mm above this line, basilar invagination is suspected ( Fig. 33.11 ).
Wackenheim Line
The Wackenheim line is drawn along the rostral surface of the clivus. A protrusion of the tip of the dens dorsal to this line indicates basilar invagination ( Fig. 33.11 ).
Ranawat Criterion
The Ranawat criterion is measured on lateral plain films. A line is drawn from the pedicles of C2 superiorly along the vertical axis of the odontoid until it intersects a line connecting the anterior and posterior arches of C1. If this measures 15 mm in men or 13 mm in women, basilar invagination is suspected ( Fig. 33.12 ).22
It is often difficult to identify the tip of the dens on plain radiographs. Therefore, the aforementioned radiographic criteria that do not rely on the identification of the tip of the dens for diagnosing basilar invagination are more sensitive. The sensitivity and specificity of these criteria are quite low (70%) if used individually. Riew et al., however, reported that if the Clark station, the Redlund-Johnell criterion, and the Ranawat criterion are positive for basilar invagination, the sensitivity was 94%, with a negative predictive value of 91%, regarding the diagnosis of basilar invagination.23
Computed Tomography
Thin-cut computed tomography (CT) of the CVJ, with reformatted coronal and sagittal views, has become the standard imaging modality for both the initial evaluation of RA of the spine and postoperative follow-up. It provides a three-dimensional appreciation of the extent of subluxation or instability, the extent of basilar invagination, the nature of ventral compression (osseous vs. soft tissue) ( Fig. 33.13 ), the diameter of the spinal canal, and the space available for the spinal cord. It is an extremely important imaging modality for delineating the relation of the vertebral artery to the pars interarticularis, the definition of bony defects, and bone quality.
Magnetic Resonance Imaging
Magnetic resonance imaging (MRI) ( Figs. 33.4 and 33.5 ) gives direct information about the pannus, the nature of ventral compression (whether osseous or soft tissue), the extent of cord compression, subarachnoid space obliteration, spinal cord signal changes, and the extent of basilar invagination. Dynamic MRI of the cervical spine in flexion and extension modes has been suggested by some authors, as MRI in the neutral position alone may be misleading with regard to the amount of spinal cord compression.24,25 T1- and T2-weighted MR images, along with contrast-enhanced T1-weighted spin echo sequences, are generally required. Hypervascular pannus appeared as high signal intensity on T2-weighted graded images with and without contrast, whereas fibrous pannus showed reduced signal intensity on T2-weighted, gradient-recalled echo images and on pre- and postcontrast T1-weighted spin echo images. Such information regarding the differentiation of fibrous versus hypervascular pannus may be useful, as dorsal fusion alone may not cause the ventral mass of fibrous hypovascular pannus to disappear. Conversely, active pannus and synovitis associated with hypervascular tissue may disappear with dorsal stabilization alone. Goel theorized that retroodontoid pannus is a result of laxity and kinking of ligaments secondary to reduction of joint space and lateral mass collapse. He showed that distraction of the facets resulted in immediate postoperative disappearance of the pannus.26