Odontoid process fractures were classified by Anderson and D’Alonzo ³ as types I, II, and III. Type I fractures involve the apical part of the odontoid process, are quite rare, and are usually believed to be stable. However, one report suggested otherwise,⁴ and dynamic imaging should be used to assess stability. Type II fractures involve the neck of the odontoid process and are the most common. Type III fractures extend into the body of C2 and generally heal well with immobilization. However, in a comprehensive review of fractures of the C2 vertebral body, Benzel et al.⁵ noted that the type III fracture described by Anderson and D’Alonzo³ is not an odontoid fracture at all. Benzel et al.⁵ proposed a classification of C2 body fractures that is more comprehensive and more meaningful in regard to mechanisms of injury.

Debate continues regarding the optimal treatment of type II fractures. Reported nonunion rates range from 7%⁶ to 100%.⁷^–¹² A meta-analysis¹³ found that halo vest immobilization produced a fusion rate of 65%. The variable success of immobilization led some authors to try to define parameters that would predict failure with external immobilization. Extent of dislocation (67% nonunion if dislocation is >6 mm,¹⁴ 88% nonunion with dislocation >4 mm¹⁵), patient age (higher failure rate in older patients^15,¹⁶), and direction of subluxation (higher failure rate with dorsal subluxation¹⁶) all have been suggested as predictors, as has a comminuted fragment of bone at the base of the odontoid process (type IIA).¹⁷ Although these studies fail to agree on many points, they do emphasize the nature of the problem.

Degree and direction of offset may be misleading indicators because they have been identified based on single rather than dynamic radiographs (Fig. 142-2). However, age seems to be an important valid indicator of the propensity for nonunion. In a randomized controlled prospective study, Lennarson et al.¹⁸ found the nonunion rate was 21 times greater in patients older than 50 years who were treated with halo immobilization than in younger patients. This study was a key factor leading to a recommendation for surgery in the guidelines for management of acute cervical spine and spinal cord injuries published by the Joint Section of Disorders of the Spine and Peripheral Nerves of the American Association of Neurological Surgeons and Congress of Neurological Surgeons.¹⁹

FIGURE 142-2 A, Patient with unrecognized odontoid process fracture after a fall. B, Flexion film shows severe instability with 100% anterolisthesis, showing the fallacy of relying on a single film to predict the degree of instability.

Because nonoperative treatment of type II odontoid fractures has a high nonunion rate (Fig. 142-3), several alternative methods of surgical fixation have been developed. The traditional operative technique for type II odontoid process fractures has been C1-2 dorsal wiring and arthrodesis. A relatively high fusion rate is associated with this technique; however, rigid postoperative bracing for at least 3 months is necessary, and successful fusion results in a significant reduction in head rotation. The dorsal approach also has an associated traumatic effect on cervical muscles. All these disadvantages can be obviated by using direct ventral odontoid screw fixation techniques.

FIGURE 142-3 Anteroposterior (A) and lateral (B) CT images of a 78-year-old patient with a type II odontoid fracture sustained in a motor vehicle accident. At this juncture, he had been in a halo vest for 6 months and had no evidence of union despite being in good alignment.

Direct Ventral Odontoid Screw Fixation

Direct screw fixation of the odontoid process was first described in 1980 in the Japanese literature by Nakanishi,²⁰ who began using this technique in 1978. This description was followed in 1981 and in 1982 by publications from Böhler,^21,²² who reported his experience dating back to 1968. Although others²³^–²⁶ described their experiences with various approaches to achieve direct odontoid screw fixation, the procedure was not widely accepted. With the development of specialized instrumentation facilitating accurate screw placement and minimal trauma to the patient,²⁷^–²⁹ the procedure has gained in popularity. The technique has the advantages of (1) decreased postoperative pain resulting from lack of extensive muscle dissection, (2) avoidance of bone graft harvest, and (3) maintenance of normal anatomy and rotation at the C1-2 joint.³⁰ Many patients require no postoperative immobilization.

Direct ventral odontoid screw fixation can be used as the primary approach to treat acute type II fractures. Patients with type II dens fractures with concomitant C1 ring fracture may also be candidates for odontoid screw fixation. However, assessment of transverse ligament integrity by MRI preoperatively ³¹ and by flexion fluoroscopy postoperatively is essential. If the latter shows continued C1-2 instability, either a ventral or a dorsal C1-2 fusion is necessary. The direct screw fixation technique may also be used in some patients with chronic nonunion of type II odontoid fractures. Candidates should have a relatively small gap between the odontoid process and the C2 body and a reasonably sized odontoid fragment that has not autofused to C1 and does not have sclerosis of the surface opposing the body of C2. Chronic malunions that do not meet these criteria rarely fuse and ultimately fracture the hardware and become unstable. The chance of successful bony union in one small series of such patients with fractures of more than 18 months of age was only 25%³²; this sharply contrasts with an 88% fusion rate for type II and high type III fractures of less than 6 months of age.³² For this reason, we generally recommend posterior C1-2 fusion for chronically nonunited fractures. Unstable type III odontoid fractures that do not extend too far into the body of C2 are also potential candidates for direct screw fixation.

Contraindications

Absolute contraindications include comminuted fractures of the C2 body and transverse ligament disruption, as defined by MRI or suggested by a C1 lateral mass fracture with extensive lateral displacement (>7 mm total on anteroposterior radiographs)³³; pathologic fractures; and nonunions of fractures that occurred more than 6 to 8 months previously that do not meet the aforementioned criteria. A relative contraindication is severe osteoporosis. In addition, an oblique fracture of the odontoid process, angled caudally and ventrally so that it is parallel to the planned screw trajectory, may not be as suitable for ventral screw fixation because the odontoid process may slide down the fracture plane as the screw is tightened. Although they account for only 16% of the cases in one published series,³² these anterior oblique fractures had a significantly higher failure rate. By starting fixation in a position of slight retrolisthesis and augmenting the construct with a rigid cervical collar to restrict flexion, successful fusion has been achieved in patients with anterior oblique fractures.

A barrel-shaped chest and short neck or a neck that is immobile or kyphotic because of cervical spondylosis can render the surgical approach more difficult, but these are not usually contraindications to the procedure with the instrumentation described subsequently. Finally, using two quality C-arms is preferred, and the procedure should not be attempted without at least one.

Patient Positioning

The patient is placed supine with the neck extended for proper screw trajectory. A folded sheet or blanket is placed under the shoulders. If the neck cannot be initially extended, as judged by careful lateral fluoroscopic monitoring, the head is supported on folded towels in neutral neck alignment. Holter traction with a light weight (5 pounds) hung over the Mayfield U-bar attachment to the operative table is very useful for stabilizing the head.

For odontoid or ventral C1-2 screw placement, biplanar fluoroscopy is necessary. The anteroposterior view is obtained transorally. A wine bottle cork, notched for the teeth or gums, is an ideal radiolucent mouth prop. A single fluoroscope, swung back and forth frequently from the lateral to the anteroposterior position, can be used if necessary. A triangular space for one side of the C-arm can be walled off with drapes and IV poles to facilitate frequent positioning and minimize the need to redrape the C-arm. It is much easier, however, to use a second C-arm fluoroscope if one is available. One C-arm unit is placed laterally with the arc horizontally or up to 45 degrees above the horizon. The other can be brought in at a 45-degree angle from the head of the table and positioned for the transoral view (Fig. 142-4).

FIGURE 142-4 Patient positioned on operating table. Note the folded sheet placed under the shoulders to increase neck extension in this patient whose fracture reduced in extension. Note also the placement of two C-arm fluoroscopic units for anteroposterior (transoral) and lateral fluoroscopic control.

Some adjustments may be needed to optimize the views, but once the optimal view is achieved, the remainder of the procedure is greatly facilitated. The C-arm monitors should be positioned for optimal viewing by the surgeon, who stands on one side of the patient with the assistant on the opposite side. The anesthesiologist remains at the head of the table in the remaining quadrant. This setup provides optimal access to the patient’s head and airway.

Operative Technique

Several screw systems have been used, but all ventral odontoid fixations begin with the same exposure. The initial approach to the spine is the same as for an anterior cervical discectomy. The spine is approached at about the C5 level through a unilateral natural skin crease incision (Fig. 142-5). We use a local injection with epinephrine (1:200,000) to minimize skin bleeding and complete hemostasis with bipolar cautery. The platysma muscle is elevated and divided with monopolar cautery. The sternocleidomastoid muscle fascia is opened along the medial side of the muscle with sharp dissection. Blunt dissection opens the deeper tissue planes medial to the carotid sheath and lateral to the trachea and esophagus to expose the prevertebral space. Dividing the longus colli fascia and the anterior longitudinal ligament in the midline with electrocautery allows the bellies of the longus colli muscle to be elevated bilaterally over approximately 1.5 vertebral segments. Sharp-bladed Caspar retractor blades are set in place below the muscle and attached to the Caspar retractor.

FIGURE 142-5 Skin incision in a natural skin crease at about the C5 level. Inset shows retractor in place.

The loose areolar tissue in the prevertebral space ventral to the longus colli muscles is easily opened with a Kittner or “peanut” dissector held in a curved tonsil clamp. It is swept from side to side while advancing up to the C1-2 level (monitored with lateral fluoroscopy). The Apfelbaum system (Aesculap Instrument Corporation, Center Valley, PA) has an angled retractor blade that reaches into this space under the mandible and holds open the working tunnel. It attaches to one side of the previously placed modified Caspar retractors (see Fig. 142-5). Other systems use different retractors, such as a curved hand-held retractor (Synthes) or small metal hook-shaped hand-held Hohmann retractors that lock over the shoulders of C2 bilaterally alongside the dens, as initially described by Böhler.²¹ The key to the retraction is to create a working tunnel up to the caudal edge of C2, without having any device caudally in the wound that restricts the low trajectory needed for proper screw placement.

At this point, the various instrument systems use different approaches for placing the screws. The Apfelbaum system consists of an outer guide tube with spikes that anchor it to C3 and that can be used to optimize spinal alignment. An inner guide tube, within the outer tube, guides the drilling. After the pilot hole is drilled, the inner guide is removed, the hole is tapped, and the screws are placed through the outer guide tube. First, under biplanar fluoroscopic control, an entry site on the ventral caudal edge of C2 is selected, and a K-wire is impacted into C2 (Fig. 142-6A). If one screw is to be placed, a midline location is chosen. If two are to be placed, a paramedian location is selected 2 to 3 mm from the midline. Care and patience in selecting the entry site and setting the K-wire are rewarded by the remainder of the procedure being expedited. When the K-wire is set, a 7-mm hollow drill is placed over the K-wire and is rotated by hand to create a shallow trough in the face of C3 and in the C2-3 anulus (Fig. 142-6B–D).

FIGURE 142-6 A, A guiding K-wire in place. B–D, Hollow hand drill creates a trough in the face of C3 and C2-3 anulus.

No bone is removed from C2. The two guide tubes are mated together, passed over the K-wire, and walked up the ventral face of the spinal column until the spikes on the outer tube are over the body of C3. The inner guide tube is advanced in the trough to the ventral caudal edge of C2 (Fig. 142-7), and the K-wire is removed. Having the guide tube at the entry site prevents the drill from skipping off the edge of the bone and walking up the ventral face of C2. With the guide tube system firmly engaged in C3, the surgeon can optimize the C2 alignment on the fluoroscopic images by either pushing C2 and C3 dorsally relative to the odontoid-C1 complex or pulling C2 and C3 ventrally. In the case of a retrolisthesed odontoid process, this realignment can be performed while gradually extending the patient’s head and removing the supporting towels beneath it to obtain an ideal working trajectory.

FIGURE 142-7 Drill guide system. Inner and outer guide tubes mate together and are placed over the K-wire. The spikes on the outer guide tube are impacted into C3, and the inner guide tube is advanced into the previously created trough (arrow) to the caudal edge of C2.

A pilot hole is drilled from the ventral caudal edge of C2 to the apex of the odontoid process, advancing the drill slowly under biplanar fluoroscopic control (Fig. 142-8). The dense cortical shell of the odontoid must be pierced to engage the screw properly and avoid splitting. Because the odontoid process is firmly held in position by its periosteum and attached supporting ligaments, it is not displaced as the drill enters from the soft cancellous fracture site. The angle of drilling is such that the drill can penetrate a substantial distance beyond the apex of the odontoid process into the apical ligaments without threatening the dural or neural structures. However, if a more dorsal trajectory is needed, greater care must be taken not to penetrate too far into the spinal canal; this is controlled by visualizing the progress of the drill on the fluoroscope.

FIGURE 142-8 K-wire is removed and replaced with a drill, which is guided fluoroscopically to the apex of the odontoid process after reducing dislocation of the odontoid process. The guide tube is kept in place by steady upward pressure (vertical arrow) to keep the spikes engaged in C3. The alignment of C2 and C3 relative to the odontoid process and C1 is optimized and maintained by lifting up or pushing down on the retractor as appropriate, as indicated by the oblique arrows, before crossing the fracture site with the drill.

A right-angled (dental-type) drill hand piece can be used to avoid interference with the ventral chest wall when drilling; this allows the procedure to be performed even in barrel-chested patients. Once the drill is into the distal odontoid cortex, its depth of penetration is read on the calibrated shaft, and the anteroposterior and lateral fluoroscopic images are saved on the monitor screens. Comparison of future live images with these saved images allows reestablishment of the identical alignment in successive steps.

The drill is withdrawn, and the inner guide tube is removed. A tap is placed through the outer guide tube, and the pilot hole is tapped. Threads are cut in the bone, allowing a more precise bone-screw junction that may reduce bone absorption around the screw caused by pressure necrosis if a self-tapping screw is used. The tap is removed, and a screw is placed through the guide tube (Fig. 142-9). A screw that is a few millimeters shorter than the measured drill depth may be chosen to allow for reduction at the fracture site, but it is important that the screw fully engages the apical cortex. Extending the screw a few millimeters beyond the cortex into the apical ligaments is safe and preferable to having one too short because the latter may (and has in our experience) back out. To achieve some fracture reduction, a partially threaded screw (lag screw) is used to pull the odontoid back toward the body of C2.

FIGURE 142-9 A lag screw is placed through the guide tube (A) and advanced through the tapped pilot hole to its final position (B). In fresh fractures, the gap at the fracture site is reduced (arrows) by the lag effect of the screw pulling the odontoid back toward the body of C2.

At each of these two steps, the odontoid process–C2 alignment that was achieved initially is easily restored by comparing the active fluoroscopic image with the stored image taken at the time of drilling. In practice, we have never had a problem reentering the same hole with the tap or screw, and improved modern fluoroscopes often show the drill track itself. If a second screw is to be placed, the identical series of steps is followed on the contralateral paramedian site except that either a partially threaded lag screw or a fully threaded screw can be used because no further lagging action would be expected to occur.

After removal of the guide tube, bleeding from C3 can be controlled with bone wax. Lateral fluoroscopy in flexion and extension confirms stability. Closure is routine and is performed in layers, closing the sternocleidomastoid fascia, platysma muscle, and subcutaneous tissue with absorbable sutures and the skin with sterile tape strips. No drains are placed. Unless concern exists regarding the patient’s bone quality, external collars are not usually recommended, and patients are allowed to return to work and resume nontraumatic activities promptly.

Several alternative systems have been proposed that are often based on existing long bone screw fixation techniques. These systems use a K-wire to drill the pilot hole and pass a hollow overdrill over this, followed by a cannulated screw.²⁴ Theoretically, after the K-wire is placed, it does not have to be removed so that precise reentry into the same trajectory is assured; however, a drawback of these systems is that they do not appear to have any provision for optimizing alignment with the drill guide, as described earlier, except by trying to do this by repositioning the head or placing additional instruments beside the drill and pushing on C1 or C2. K-wires are suboptimal drills because they lack the torsional rigidity of drill bits and can be deflected by irregular densities within the bone. To redirect them, one must remove the K-wire and select a new starting point.³⁴ In addition, great care must be taken when drilling over the K-wire because the drill can bind to the K-wire and advance it into the spinal canal or cut the K-wire. This can also occur when placing the screw over the K-wire.

Controversies

One-Screw Constructs versus Two-Screw Constructs

Theoretically, with one screw, the odontoid process could rotate on C2, although with fresh fractures the interdigitation of the irregular fracture surfaces may prevent this; this may explain why both techniques have been reported to have similar clinical success.³⁵^–³⁷ Laboratory studies also show no greater resistance to screw fracture from bending with one-screw or two-screw constructs. However, a more recent clinical study by Dailey et al.³⁸ found a significant difference in fusion rates among 57 patients older than 70 years with odontoid fractures: a 56% success rate was achieved if one screw was placed, whereas a 95% fusion rate was achieved when two screws were placed. The different outcomes between this study and the previously mentioned ones are likely due to the fact that the earlier studies had a much younger patient population.

The demographics of patients with odontoid fractures are changing. The previous bimodal age distribution included a large number of younger patients who sustained this fracture when riding as an unrestrained passenger in a motor vehicle involved in an accident. With increased seat belt usage and air bag availability, this is a rare cause of this injury nowadays. Elderly patients who sustain odontoid fractures after relatively minor trauma, often a ground-level fall, constitute most patients with odontoid fractures, highlighting the importance of considering placement of two screws in older and possibly in middle-aged patients.

When two screws are placed, the entry site for each is located paramedially a few millimeters off the midline and the screws are angled toward each other at the odontoid apex. The diameter of the odontoid process should be assessed on the preoperative CT scan to ensure adequate bone volume for a second screw. Some patients may not have a sufficiently wide odontoid process to accommodate two screws side by side,³⁹ but the process may be deep enough that the screws can be placed in such a way that they end up in front of and behind each other to achieve the same fixation.

Screw Size and Type

Biomechanical data suggest that cannulated screws are only about 5% to 10% weaker than solid screws.³⁴ Various screw diameters have been used. These usually range from 3.5 to 4.0 mm and are occasionally larger. The initial experience was with cancellous threaded screws, which have a deeper thread (smaller minor diameter or core) and are better at resisting pull-out. However, pull-out forces on odontoid screws are minimal. An odontoid screw primarily has to resist bending and translational forces. Screw failure, if it occurs, is almost always due to fracture at the level of the bone fracture. Cortically threaded screws (4-mm outer diameter) with a larger minor diameter (2.9 mm vs. the previously used 2 mm) would seem optimal. Such screws have proven substantially stronger in laboratory tests, in which they fail to fracture after 1 million cycles at three times the load at which the older screws fractured at 33,100 cycles. In the experience of the senior author (R.I.A.), these screws are also more effective; no postoperative screw fractures were observed in more than 400 consecutive screw placements compared with a 5% fracture rate in odontoid fixation and 10% fracture rate in C1-2 fixation with the prior screw design. An additional benefit is that the pilot hole is drilled larger (3 mm vs. 2 mm); this makes the drill much more directionally stable, allowing precise correction of pilot hole trajectory to optimize screw placement.

Results

Type II odontoid fractures less than 6 months old treated with this technique have a high rate of fusion. The combined published series ³² of Veres in Hungary and Apfelbaum et al. in Salt Lake City, Utah, encompassed 147 patients whose ages ranged from 15 to 92 years. Successful bony union was achieved in 88% of patients, with an additional 3% achieving stability via fibrous union. These results agree with the results of other published series with fewer patients.^25,^26,^34,^40,⁴¹ Failures usually occur in elderly patients with poor quality bone. In such circumstances, the screw may fail to hold in C2, in particular, if there is an associated fracture in the body of C2. If this complication is recognized early, manipulative realignment and external immobilization have been successful. If not, additional surgery was required. There was one late neurologic complication in which a patient became quadriplegic in the series from Hungary when the construct failed to hold and the fracture subsequently dislocated. There were no other neurologic complications, and other surgical complications are rare.

Other series have shown similar low complication rates; however, there has been a high incidence of dysphagia from the retropharyngeal approach in elderly patients, and some of these patients required temporary feeding tubes.³⁸ Although there are no published results detailing complications when using one of the cannulated systems, numerous serious and even fatal complications are known to result from the K-wire being driven in beyond the odontoid tip.

The high degree of success achieved by using this straightforward, easily mastered technique, usually with minimal complications, warrants its consideration as the primary treatment for many type II odontoid fractures. By allowing the patient to resume normal activity quickly while avoiding the cost and medical and social morbidity of a halo vest, screw fixation seems to be a very cost-effective treatment for this problem.

Ventral C1-2 Transarticular Screw Fixation

Ventral C1-2 transarticular screw fixation may provide an alternative if odontoid screw fixation is impossible or if successful odontoid screw fixation fails to stabilize C2 because of unrecognized concomitant transverse ligament incompetence. Stabilization is accomplished by inserting two screws through the lateral masses of C2 into the lateral mass of C1.⁴² The entry site is just medial to the vertebral artery, which is placed at risk by this approach. The screws angle laterally about 20 degrees and dorsally at a similar angle. The entry site is selected by following the lateral edge of the vertebral body rostrally from the C2-3 interspace to its junction with the lateral mass, staying as medial as possible in that structure. The drill guide system used for odontoid screw fixation can be used for this as well, although the screw length is considerably shorter.

Although it achieves fixation, this technique does not permit placement of substantial bone graft. Rather, long-term stabilization is achieved by promoting arthrodesis of the C1-2 lateral articulation, primarily by immobilization. This stabilization may be enhanced by curetting the joint, but overall, the construct is less likely to succeed than dorsal screw fixation with dorsal grafting. Therefore, we believe it should be reserved for exceptional cases.

Other Ventral Techniques

There are a few reports of using plates and screws transorally.⁴³ Experience with these is limited because they do not appear to have been widely used. The risk of infection and, to a lesser extent, the limited working space seem to have deterred most surgeons from these approaches.

Dorsal Approaches

The dorsal approach is used to stabilize C1 to C2. Rather than providing stabilization and fixation of a fractured bone, it stabilizes what was previously a normal motion segment and provides an optimal environment for bone graft healing. However, bone grafting is almost always required for long-term stabilization because hardware in nonfused segments ultimately fatigues and fails.

Traditional techniques have used various C1-2 wiring strategies with interposed or onlay bone grafting (Gallie,⁴⁴ Brooks,⁴⁵ and interspinous [Sonntag]⁴⁶ fusions). Because these impart limited stability that deteriorates significantly with cyclic loading,⁴⁷ an external rigid orthosis is usually necessary. Fractured or absent dorsal elements can preclude the use of these techniques. Even in optimal situations, nonunion rates of 30% are reported.⁴⁷

Transarticular screw fixation, pioneered by Magerl in 1979, offers immediate stabilization, often without external orthosis.⁴⁸ By optimizing bone graft union with a high chance of success, transarticular screw fixation is a major advance in treating instability in this area. The construct can be extended with various devices to include the occiput and the subaxial spine if needed.

Indications

This procedure is indicated for atlantoaxial instability from almost any cause—traumatic disruption of the transverse ligament, rheumatoid or other degenerative diseases, iatrogenic instability after transoral decompression, congenital or acquired absence of a united odontoid process (os odontoideum), ligamentous incompetence associated with various genetic diseases (Down syndrome, Larsen disease), or chronic nonunited odontoid fractures. Occiput-C1 instability can occur from many of the same causes and, if present, can be treated by extending the stabilization and fusion up to the occiput.

Contraindications

Poor bone quality is always of concern during intraosseous fixation and must be evaluated carefully. It is not an absolute contraindication to surgery but may necessitate using both internal and external devices. Of paramount importance for transarticular screw fixation is an adequate pathway for the screw that traverses the pars interarticularis (isthmus) of C2 (not the pedicle) to the lateral C1-2 articulation before crossing that joint into the lateral mass of C1. Variations in anatomy and secondary effects of vascular elongation coupled with bone softening can result in the vertebral artery looping up into the pars of C2. This looping of the vertebral artery may occur to such an extent that an inadequate pathway remains for safe screw placement. Placing screws in such circumstances has resulted in vertebral artery injury with potentially serious neurologic sequelae. An alternative technique devised by Harms and Melcher ⁴⁹ uses screws placed in the lateral masses of C1 and into the C2 pars or pedicle, which are connected posteriorly with rods. This technique is applicable in cases in which a safe pathway through the C2 pars to C1 does not exist. Another approach using translaminar screws in C2 that are then coupled to C1 screws has been proposed by Wright.⁵⁰

Understanding the patient’s anatomy and the availability of a safe bony pathway for screw placement before proceeding with the surgery is crucial. We obtain thin-section CT scans and build images that are not just in orthogonal planes but are also reconstructed along the planned screw pathways. This construction of images can be done on the consoles of most CT scanners. Alternatively, importing the images into a stereotaxic workstation allows the surgeon to look at various possible pathways and build a three-dimensional model that can be viewed in the operating room. This approach has been invaluable in evaluating whether screws can be passed safely (Fig. 142-10) and in assisting in achieving the desired pathway during surgery.

FIGURE 142-10 Upper views show CT reconstruction through the pars interarticularis of C2 on the right and left sides in trajectory shown on axial scout views in the lower two panels. Note the safe path for the screw on the patient’s right side but abnormally large vertebral artery canal with inadequate bony dimensions for safe screw placement on the left. Only one screw was placed in this patient.

Patient Positioning

A Mayfield three-pin head holder and a cervical collar are placed. The patient is rolled into prone position on bolsters while the surgeon keeps the neck stable with axial traction and maintains it in a neutral position. Before the head clamp is secured to the table, final positioning is performed under lateral fluoroscopic guidance. The best reduction position is often with the neck in extension, but this may preclude C1-2 transarticular screw placement in many cases because lordosis would often dictate a screw trajectory starting within the thoracic cavity. The patient should be positioned with the chin slightly flexed but with the head pulled dorsally (Fig. 142-11). This position has been likened to the posture of military personnel standing at braced attention. This position usually reduces C1-2 dislocations via dorsal translation but leaves the rest of the cervical spine in a flattened or even slightly kyphotic posture so that the needed screw trajectory can be achieved. The head should also be positioned so that it is not rotated; this can be judged by the symmetry of the ear canals relative to the table or floor. Careful monitoring of the fluoroscopic image facilitates safe positioning. An absolutely perfect position is unnecessary; minor residual translational movements are accepted because these can easily be corrected at the time of screw placement.

FIGURE 142-11 Patient positioning for transarticular screw fixation. The head is fixed in a three-pin head holder. Translating the head dorsally with the chin tucked reduces atlantoaxial dislocation and allows the lower spine to be kept in a straight or flexed position. This position facilitates the trajectory for C1-2 screw placement. The patient is positioned while vertebral alignment is monitored via lateral fluoroscopy.

The low angle of screw trajectory needed to traverse the C2 pars interarticularis and enter C1 dictates a starting point at about the T1-3 level. Magerl’s original technique called for an incision and paraspinal muscle retraction down to this level; however, percutaneous tunneling techniques make this unnecessary.

Operative Technique

A dorsal midline incision extending from just below the inion to C3 is usually adequate (Fig. 142-12A). The paraspinal muscles are dissected off the dorsal elements of C1, C2, and the occipital bone and are held back with angled Weitlaner retractors. The exposure may require sharp dissection assisted with electrocautery if the spine is unstable. The full extent of the dorsal elements of C1 and C2 should be exposed, with definition of the lateral aspect of the C2 dorsal elements and the C2-3 facet joint and of the C2 isthmus extending rostrally beneath the C2 nerve root and the associated venous complex (Fig. 142-12B). Bleeding can usually be arrested with small hemostatic gelatin (Gelfoam) pledgets soaked in thrombin or a slurry made with Gelfoam powder and thrombin. It is neither necessary nor desirable to disconnect the inferior midline attachments to the bifid C2 spinous process.

FIGURE 142-12 A, Site of surgical incisions. The level of the stab wound sites inferiorly for placement of the drill guide tube is determined fluoroscopically. B, Details of surgical anatomy. The desired screw placement is just lateral to the lateral edge of the spinal canal. It traverses the isthmus (pars) of C2 and the C1-2 articulation. The screw does not go down the pedicle, which would take it into the body of C2 and not across the C1-2 joint into C1. C, Placement of guide tube (with obturator) through stab wound and into the operative field.

When the anatomic structures are fully exposed, fixation screws are placed. The entry site on each side is just rostral to the C2-3 facet joint and in line with the previously defined pars interarticularis. Standing back from the wound to reduce parallax, a line representing the inferior extension of the trajectory may be drawn on the drapes. A straight instrument such as a K-wire placed along the patient’s neck and superimposed over the desired screw path on the fluoroscopic view provides the intersecting skin coordinates, usually at about T2.

A 1- to 1.5-cm skin transverse incision carried down through the dorsal fascia provides an entry for the drill apparatus (see Fig. 142-12A). As with odontoid fixation, a guide tube system (Aesculap Instrument Corporation) may be used. With this system, a smooth tube fitted with a conical tipped obturator is passed from the skin incision up to the drill entry site at the C2-3 junction by pressing firmly in a rostral direction and gently rotating the instrument back and forth to advance it into the surgical field (Fig. 142-12C). When the guide tube is in place, the obturator is removed, and an awl is used to make a starting hole in the C2 laminar bone. An inner drill guide is placed to support the drill (Fig. 142-13). The guide tube assembly allows precise control of the drilling direction.

FIGURE 142-13 Guide tube in place as seen in sagittal (A) and dorsal (B) views. The obturator is removed (A) and replaced with an inner drill guide (B).

The surgeon must visualize the dorsal lateral and medial borders of the pars interarticularis and direct the drill bit accurately between these limits. A tool, such as a small Penfield dissector held by an assistant, can help visualization of this area. Because it is placed on the dorsum of the pars, it serves as a fluoroscopic marker for that boundary. A low-angled trajectory to carry the drill bit just below the dorsum of the pars and across the C1-2 lateral articulation, as far dorsally as possible, engages the maximum amount of the lateral mass of C1 and keeps the drill above the vertebral artery. On the lateral fluoroscopic image (Fig. 142-14), the projection of the ventral arch of C1 is a helpful target to aim for, especially at its rostral margin. Generally, the screw trajectory should be in a straight paramedian direction or slightly medially as dictated by the bone anatomy and location of the vertebral artery foramen. Aiming too far medially results in a smaller area of C1 lateral mass engagement, whereas aiming laterally can jeopardize the vertebral artery.

FIGURE 142-14 Lateral fluoroscopic view of drilling in process. The instrument, a Penfield dissector, is defining the dorsum of the pars, with the drill passing just ventral to this. The white arrowhead indicates the location of the C1-2 joint. The drill is aimed for the upper portion of the anterior tubercle of C1 (black arrowhead) but is not advanced beyond the anterior cortex of C1 (white arrows). The gray oval indicates the approximate position of the danger zone where the vertebral artery normally resides.

Attempts to adapt image-guided stereotactic systems to this procedure have met with mixed results. It is necessary to track C2 because the vulnerable anatomy is within C2 and because C2 is hypermobile in most of these cases. Tracking arrays placed on the skull clamp or on C1 would not compensate for this motion. However, tracking C2 is difficult because most tracking arrays are bulky and can obstruct the surgeon’s access. They also are difficult to secure solidly to C2. It is also very hard to localize the C2 anatomy accurately to register it with the stereotaxic system. Fluoroscopic devices that obtain multiple images and generate a CT image from these may help in this regard. Despite these difficulties, using image guidance is a good way to evaluate and understand this complex anatomy and obtain the precision necessary to place screws safely. It is recommended to surgeons who lack experience with the technique.

When drilling, increased resistance is felt at the cortical margins of the C2 joint surface, and then at the C1 joint surface and at the ventral cortex of C1. If necessary, C2 can be translated ventrally or dorsally before drilling across the joint space by grasping the spinous process of C2 with a towel clamp, ensuring optimal alignment. When the pilot hole is drilled, the depth is noted on the calibrated drill shaft, and the images are stored. The drill and inner drill guide are replaced with a tap (Fig. 142-15) except in very soft bone. After the hole is tapped, a fully threaded screw is placed because no lag effect is needed. Vertebral alignment is reoptimized just before the C1-2 articulation is crossed by matching the active fluoroscopic image with its stored counterpart. If a polyaxial screw is being used, as in the Harms construct, for example, or to extend the construct to the occiput, a guide tube with an additional inner sleeve is used. Removal of the sleeve increases the inner diameter of the tube so that the polyaxial screw can be placed.

FIGURE 142-15 The inner drill guide tube is removed (A). Through the outer guide, the pilot hole is tapped (B), and the screw is placed (C and D).

Bone bleeding may occur, in particular, in patients with inflammatory disease. However, if brisk arterial bleeding ensues from the drill hole, suggesting a vertebral artery injury, placement of one screw for fixation and tamponade is recommended, but a second screw should not be placed. If this occurs, it would be prudent to obtain postoperative angiographic images to ascertain the status of the vessel and to detect fistula formation.⁵¹

Immediately after the first screw is placed, significant improvement is observed in patients with gross instability. The same sequence of steps is repeated on the contralateral side.

If a safe pathway for transarticular screw fixation is not present, the Harms technique can be used. This technique involves placing a polyaxial screw directly into the lateral mass of C1 and coupling it with a rod to a separate C2 screw. The anatomy of the individual bones must be studied before placing these screws to avoid vascular or neural injury. The C1 lateral mass is accessed by depressing the C2 nerve root where it traverses over the posterior aspect of the lateral mass of C1 below its junction with the posterior arch of C1. The screw usually extends above the bone for 1 cm or more, so a partially threaded screw with a smooth shank proximally is recommended to avoid irritation of the C2 root by the screw threads. Depending on the patient’s anatomy, the C2 screw can be a short pars screw that stops before a prominent vertebral artery foramen, a pars screw aimed anteriorly, a C2 pedicle screw, or a translaminar screw.

The Harms technique widens the options for screw fixation if the C2 pars is unsuitable for placing a transarticular screw, but the surgeon must be aware that the vertebral artery can also be (and has been) injured with the C1 screw, so a thorough understanding of its location and the bone anatomy of C1 is important. The procedure is more complex than transarticular screw placement because four polyaxial screws and the connecting rods must be placed. Using intralaminar screws in C2 also is an option, but this precludes using the lamina of C2 for the inferior end of the graft in the manner discussed in the next section.

Bone Grafting

Screw fixation functions as an internal splint. For long-term stability, bone fusion is required because all hardware ultimately fails without it. Magerl and Seeman ⁴⁸ suggested curetting and then packing bone chips into the C1-2 articulation to encourage arthrodesis. Access to this joint is limited, however, and only a small portion of it can be treated in this manner even in ideal circumstances. We use a dorsal fusion construct if possible.

With intact dorsal elements, a modified construct consisting of a combined interpositional and onlay bicortical iliac crest graft was suggested by Sonntag’s group.⁴⁶ The graft has ventral and dorsal cortices. After the mating surfaces have been denuded with a high-speed bur, the contact sites at the caudal surface of C1 and the laminar surface and rostral edge of the spinous process of C2 are to cancellous bone. The graft and donor surfaces are also contoured for maximum apposition (Fig. 142-16). The graft is secured with a braided titanium cable placed sublaminarly at C1 and around the spinous process of C2, so that the graft is sandwiched between the two layers of cable. This construct provides excellent three-dimensional stability. The screws resist translation and rotation, whereas the graft prevents extension, and the cable prevents flexion. Additional bone chips and curettings may be placed around the construct to enhance fusion.

FIGURE 142-16 Illustration of completed screw fixation. This was the technique originally suggested by Sonntag’s group and shows the cabling technique we prefer to use. In contrast to this illustration, however, we notch the graft to contact both the dorsal and the caudal edge of the C1 laminar arch (see Fig. 142-17) and extend it over and into the lamina of C2 to improve the fusion potential at this site.

(From Marcotte P, Dickman CA, Sonntag VKH, et al: Posterior atlantoaxial facet screw fixation. J Neurosurg 79:234–237, 1993.)

There are significant advantages to being able to use allograft bone as the graft, including elimination of donor site pain and complications, reduced bleeding, and ability to select more optimal bone. Although allograft usually has been unsuccessful in onlay constructs, by modifying the C1-2 dorsal graft placement further so that the graft is notched to fit in close apposition to both the dorsal and the inferior edges of C1 and is mortised into the lamina of C2, we achieve a true interpositional graft (Fig. 142-17). Coupling this graft with internal screw fixation to eliminate all motion has allowed us to use bicortical iliac crest allograft bone with fusion results equal to autograft in this application and significantly reduced patient morbidity (Fig. 142-18).⁵²

FIGURE 142-17 Lateral (A) and posterior (B) views showing the modified grafting technique to increase the graft contact along the inferior aspect of the C1 ring and its dorsal aspect (small black arrows in A). Both of these surfaces of C1 are decorticated and flattened to maximize graft contact. The bicortical graft is also mortised into the decorticated lamina of C2 (large black arrow in A) and contacts the decorticated spinous process of C2 (arrows in B). The artist has shown wiring, but braided titanium cable is used and placed as described by Dickman et al.,⁴⁶ enveloping the graft anteriorly and posteriorly with a cable that passes under the C1 posterior ring and behind the spinous process of C2. Leaving the interspinous ligaments intact on the spinous process of C2 (not shown) helps retain the cable. This graft modification improves grafting success with autograft and allows an equally good result with bicortical iliac crest allograft if desired.⁵²

FIGURE 142-18 A–G, Preoperative and postoperative appearance of a patient with cortically threaded transarticular screws, which have a larger minor diameter. This patient had severe atlantoaxial joint degeneration on the right (A); however, the joint was still mobile. The left C1-2 joint had retrolisthesis (C). The alignment was restored to normal and held with the transarticular screws (B and D). The screw position is close to ideal crossing the C1-2 joints in their middle or slightly medially (E). Note also the interpositional location of the interspinous grafting (F and G). Allograft was used in this patient.

Alternative Systems

Drilling with a K-wire and placing hollow cannulated screws is an alternative technique used by some surgeons.³⁴ The K-wire may be drilled percutaneously or via a similarly placed guide tube. The concerns expressed earlier with regard to lack of precise directional control and the risks of inadvertently advancing the K-wire with the drill and cutting the K-wire pertain here as well.

Screws

Screws measuring 3.5 to 4.0 mm or slightly larger in diameter have been used. As in odontoid screw fixation, there is little pull-out force but significant bending or translational forces. Cortical screws with a larger minor diameter, as described earlier in this chapter with regard to odontoid screws, have replaced the smaller minor diameter cancellous screws previously used. This has been a significant improvement that has eliminated screw breakage. The screw requires a larger drill (3 mm) because of its larger minor diameter. This drill provides more directional control and has resulted in the unexpected benefit of allowing more precise corrections to the trajectory as the drill is advanced.

Postoperative Care

Some surgeons prefer to use a cervical collar, but unless there is a question of bone quality, or if only one screw is placed, this is probably unnecessary. With the immediate elimination of motion, postoperative pain is significantly reduced, and most patients can be discharged from the hospital in about 2 days. Patients are monitored with serial radiographs until fusion is ensured. Nontraumatic activities, including driving, are allowed as soon as the patient is comfortable and no longer receiving narcotic medications.

Results and Complications

Numerous series report excellent stabilization and fusion rates. Magerl and Seeman’s initial series ⁴⁸ reported a 100% fusion rate; Grob et al.⁵³ reported a 99% success rate in 161 patients; Stillerman and Wilson⁵⁴ reported a 95% rate of successful fusion (21 out of 22 patients); and Marcotte et al.⁵⁵ had a 100% rate of fusion in 18 patients. Gluf et al.⁵⁶ reported successful fusion in 98% of 191 adult patients and had similar results in an additional 78 patients since that publication.

This procedure is particularly beneficial to patients with rheumatoid arthritis, who often present with significant pain and who cannot tolerate a halo vest orthosis well. Fusion in these patients also takes substantially longer, and the patients have a higher nonunion rate with conventional techniques. With internal screw fixation, they are stable and protected and have good pain relief and an optimal environment for bone healing.

The procedure is technically demanding. It requires a good knowledge of nuances in anatomy and thorough preoperative evaluation via high-resolution CT scans, preferably with reconstruction along the screw pathway. Neurologic complications from direct injury have not been reported; however, vertebral artery injuries have occurred, and in one case bilateral vertebral artery injuries resulted in a fatal brainstem infarction.⁵⁷ Unilateral injuries have not resulted in neurologic sequelae but have produced arteriovenous fistulae, one of which manifested as a delayed spinal cord compromise from epidural venous engorgement.⁵¹

Fusion of C1 to C2 restricts head rotation by about 50%. In a normal patient, this leaves a residual motion of ±45 degrees. Younger patients can regain some lost motion, often to a surprising degree, presumably by gaining extra motion at each of the subaxial facet joints. Less limber older patients must learn to compensate by torso rotation, and usually they do so without difficulty. Paradoxically, some patients have improved motion almost immediately after surgery because the pain-provoked cervical muscle spasm subsides. Occasionally, patients complain of occipital numbness. Numbness is presumably the result of C2 nerve root trauma during surgery and usually resolves within 3 to 6 months.