Disorders of the cervical spine can lead to compression of neural elements and cause myelopathy, radiculopathy, or a combination of these conditions. Spinal cord or nerve root compression can be caused by a soft herniated nucleus pulposus, osteophyte formation, facet joint hypertrophy, and/or congenital abnormalities of the cervical spine. If nonoperative treatment fails, surgical treatment can often lead to excellent long-term clinical outcomes. The surgical treatment of these disorders depends on patient preference, clinical findings, and the evidence-based literature. In the properly selected patient, posterior cervical decompression for the treatment of radiculopathy provides potential advantages, including preservation of neck motion, avoidance of complications from anterior surgery, and fewer postoperative restrictions.
A posterior fusion is indicated for conditions involving an unstable, kyphotic, or severely spondylotic cervical spine. Advances in the understanding of cervical spine anatomy have greatly enhanced the surgeon's options for osseous fixation to achieve fusion. Cadaveric and imaging studies have demonstrated a high rate of anatomic variability and the importance of preoperative imaging and planning before surgery1-3. These studies have also established reliable guidelines for screw placement to avoid critical vascular, neurologic, and visceral structures during cervical spine instrumentation. Awareness of these principles is essential to avoid complications and provide excellent outcomes.
Foraminotomy
The foramina are bordered superiorly and inferiorly by the respective pedicles, anteriorly by the disc and uncovertebral joint, and posteriorly by the superior articular facet. The average dimensions of a cervical foramen are 9 to 12 mm in height and 4 to 6 mm in width4. These foramina are oriented at an angle of 45° from the midsagittal plane. Consequently, patients are more susceptible to foraminal narrowing in the anterior-posterior plane.
Patients with radiculopathy from nerve root compression in the foramina (Fig. 1) can benefit from a foraminotomy, which decompresses the nerve root by removing a portion of the superior articular facet. This procedure allows the nerve root to be mobilized, increasing the space between it and the disc. In many patients with foraminal disc herniation, a partial pediculectomy and gentle retraction of the nerve root can aid in the discectomy, particularly when there is soft disc pathology rather than a disc-osteophyte complex, which is difficult to remove posteriorly. A preoperative computed tomography (CT) scan is helpful in determining the etiology of the foraminal compression. Posterior foraminotomies are contraindicated for centrally located disc herniations, central spinal canal stenosis, and ossification of the posterior longitudinal ligament.
Axial cross-sectional view of a cadaveric upper cervical spine with facet joint arthrosis (F) and uncovertebral hypertrophy (VB), causing encroachment on the exiting nerve root (N). VA = vertebral artery. (Reprinted, with permission, from: Papadopoulos SM, editor. Manual of cervical spine internal fixation. 1st ed. Philadelphia: Lippincott Williams & Wilkins; 2004.)
A posterior foraminotomy can be performed with the patient in either a sitting or a prone position. The sitting position decreases the incidence of intraoperative bleeding. It carries a theoretical risk of air embolism, but this position has been used in hundreds of cases without that complication5. When the prone position is utilized, Mayfield or Gardner-Wells tongs are used to stabilize the head. Both arthroscopic and microscopic techniques have been described to aid visualization during this procedure. After a preoperative localization radiograph identifies the cervical spine location, an incision is made 1.5 cm off of the midline at the appropriate level. Sharp dissection is then performed through the fascia, with either a self-retaining McCullough retractor or a tube retractor docked onto the lateral mass and the facet joint. A localizing lateral radiograph is obtained to confirm the cervical spine level.
The medial third of the facet joint is identified, and the overlying inferior articular facet is then removed (at the C5 level for a C6 foraminotomy) with a high-speed oscillating burr (Fig. 2). The superior articular facet is well visualized and is removed to unroof the foramen and expose the traversing nerve root. An oscillating burr, micro-Kerrison rongeurs, and angled curets are used to remove up to the medial 50% of the facet joint. To ensure adequate decompression once the nerve root is exposed, a nerve hook can be utilized to carefully palpate the caudad pedicle (at the C6 level for a C6 foraminotomy) (Fig. 2, D). The procedure must be carried out lateral to the pedicle to ensure a complete decompression.
Posterior foraminotomy procedure. After the medial third of the corresponding facet joint is identified, the inferior articular facet is removed with either micro-Kerrison rongeurs (A) or an oscillating burr (B) to expose the overlying superior facet. Decompression of the facet is continued to the level of the nerve root (C). A nerve hook can be used to palpate the pedicle inferiorly and ensure adequate lateral decompression (D). (Reprinted, with permission, from: An HS, Xu R. Posterior cervical spine procedure. In: An HS, Riley LH 3rd, editors. An atlas of surgery of the spine. 1st ed. London: Martin Dunitz; 1998.)
When a soft disc herniation is present, a partial pediculectomy is done to minimize nerve root retraction. Once the medial third of the pedicle is removed with an oscillating burr, the traversing nerve root can be gently mobilized with a nerve hook to expose the disc fragment. Since this space is quite limited, an arthroscopic grabber or micro-pituitary rongeur from a tympanoplasty set can facilitate free disc fragment removal.
Posterior foraminotomies provide excellent clinical outcomes when used for the treatment of cervical radiculopathy5,6. Of 736 patients with a "posterior-lateral foraminotomy," 91.5% were reported to have a good or excellent result at an average of 2.8 years postoperatively5, a finding that has been supported by other reports7,8. Factors that contribute to worsening sagittal alignment after posterior foraminotomy include an age over sixty years and preoperative cervical lordosis of <10°7.
A posterior cervical foraminotomy provides a number of advantages over anterior approaches in properly selected patients. Anterior surgical approaches for radiculopathy may be associated with complications, including recurrent laryngeal nerve palsy, dysphagia, dysphonia, and adjacent-segment degeneration. The incidence of symptomatic adjacent-segment degeneration after anterior cervical discectomy and fusion is 2.9% per year (26% in a group of patients seen at ten years postoperatively)9. Posterior foraminotomy avoids the complications of anterior surgery and may also reduce the risk of adjacent-segment degeneration postoperatively. In a study of 303 patients followed for an average of 7.2 years after a single-level posterior foraminotomy, 4.9% developed symptomatic adjacent-segment degeneration10. The ten-year rate of adjacent-segment degeneration was calculated to be 6.7%. Furthermore, unlike anterior fusion, posterior decompression maintains cervical spine motion. In patients with multilevel pathological involvement, a posterior approach can lead to outcomes that are comparable with those of the anterior approach without limiting neck motion6.
Some authors have suggested that a C5 foraminotomy with a decompressive procedure, such as a laminoplasty, would lead to a decreased incidence of C5 nerve root palsy. Patients with this complication can have severe deltoid weakness, preventing them from performing some activities of daily living. Although the exact cause of C5 nerve root palsy is unclear, many authors have suggested an ischemic etiology from nerve root stretch after decompression. No treatment modalities have been successful in improving the outcome if iatrogenic C5 nerve root palsy occurs, but most patients regain functional deltoid strength within six to nine months after surgery11. Even though a posterior foraminotomy would theoretically decrease nerve root tension after posterior drift of the spinal cord, it is unclear whether this would decrease the incidence of this difficult postoperative complication.
Laminectomy
Posterior cervical laminectomy has primarily been used to treat central cervical spine stenosis caused by spondylosis, neoplastic conditions, or ossification of the posterior longitudinal ligament. Preoperative positioning is similar to that for a foraminotomy, but a more extensive soft-tissue dissection is required to expose the posterior cervical spine.
Bilateral exposure of the lamina-lateral mass junction is required for a full laminectomy. One of several techniques that have been described is use of a high-speed oscillating burr to remove this osseous bridge bilaterally (Fig. 3, A). After removal of the outer osseous cortex, the burr tip can be switched to an extra-rough diamond-tip drill bit to avoid catching soft tissue and dural tears. Excision of the ligamentum flavum is performed both cephalad and caudad to the laminectomy sites. Lamina removal is done with symmetric upward traction with use of a towel clamp at both ends of the decompression (Fig. 3, B). Care must be taken to avoid rotation of the fragments and subsequent impingement on the cervical cord. Remaining ligamentum flavum attachments to the lamina during this part of the procedure can be removed with a Kerrison rongeur.
Standard open posterior cervical laminectomy. After adequate soft-tissue exposure of the lamina-lateral mass junction bilaterally, a high-speed oscillating burr can be used to separate this junction (A). Once the inner cortical bone is adequately separated on each side, symmetric traction is necessary to safely remove the bone from the underlying cervical cord (B). Final inspection of the laminectomy site should demonstrate retention of the facet joint capsules and wide exposure of the cervical dura (C). (Reprinted, with permission, from: Cooper PR, Ratliff JK. Cervical laminectomy. In: Herkowitz HN, editor. The cervical spine surgery atlas. 2nd ed. Philadelphia: Lippincott Williams & Wilkins; 2003.)
Posterior fusion is indicated for cervical spine instability resulting from traumatic, iatrogenic, or degenerative causes; pseudarthrosis after a prior arthrodesis; or multilevel anterior procedures requiring enhanced fixation. Surgeons can choose between wiring and screw fixation techniques. Recent studies have shown screw constructs to have greater rigidity and result in higher fusion rates than spinous process wiring12.
Subaxial Cervical Spine
A number of techniques for subaxial screw fixation of the lateral mass have been described. First described in 197913, the Roy-Camille technique utilizes a starting point at the midpoint of the lateral mass. With use of the neutral alignment of the cervical spine (plumb line), the screw trajectory is perpendicular in the superior-inferior plane with a 10° lateral orientation (Fig. 4). The primary risk with this technique is a breach of the facet joint in the area of both the vertebral artery and the nerve root (Fig. 5). The Magerl (transarticular screw) technique14 utilizes a starting point just medial and superior to that used with the Roy-Camille procedure and orients the screw in a cephalad angle of 30° and a laterally directed angle of 25° (Fig. 4). A broad or prominent spinous process can interfere with the screw trajectory with the Magerl technique, which may increase the risk of lateral screw cutout. There is a risk of nerve root injury with both techniques if they are not performed properly.
Cervical lateral mass fixation with use of either a Magerl or a Roy-Camille technique. While both techniques utilize a starting point at the midpoint of the lateral mass in both the cephalad-caudad and the medial-lateral plane, the Magerl screw is inserted in a 30° superior and 25° lateral orientation. The Roy-Camille screw takes a more perpendicular trajectory (0°) and 10° lateral approach. (Reprinted, with permission, from: Stemper BD, Marawar SV, Yoganandan N, Shender BS, Rao RD. Quantitative anatomy of subaxial cervical lateral mass: an analysis of safe screw lengths for Roy-Camille and Magerl techniques. Spine [Phila Pa 1976]. 2008;33(8):893-7.)
Sagittal cross section of a cadaveric specimen of a lateral mass (LM) in a man operated on with 3.5-mm cancellous screws when he was seventy-two years old. The lateral mass screw (S) is seen violating the inferior articular process of C5 with its tip just short of the vertebral artery (VA) in between the C5 nerve root and C6 ganglion. F = facet joint. (Reprinted, with permission, from: Papadopoulos SM, editor. Manual of cervical spine internal fixation. 1st ed. Philadelphia: Lippincott Williams & Wilkins; 2004.)
Anderson et al. modified the lateral-mass screw technique by using a starting point slightly medial to the midpoint of the lateral mass and a 35° to 40° superior and 10° lateral orientation15. I prefer this method, in which a variable-length drill guide is used to optimize screw lengths at each cervical level. Advancement of drilling in 2-mm increments ensures the strongest bone purchase without violation of the traversing nerve root. This screw trajectory appears to be the most reliable for avoiding critical neurovascular structures.
Stemper et al. demonstrated the variations in the anatomy of the subaxial cervical lateral mass in an in vivo CT study of the cervical spine16. Bicortical screw paths created with use of the Roy-Camille and Magerl techniques were drawn on sagittal CT images of ninety-eight asymptomatic volunteers. Although there was a moderate amount of variation in screw lengths, no correlation was found between screw length and stature, body weight, or neck length. At C3-C6, Magerl screw trajectories were, on average, at least 2 mm longer than Roy-Camille screw trajectories. The screw lengths were the shortest (average, 9.8 mm in males and 8.5 mm in females) at the C7 lateral mass.
Atlantoaxial Junction
Atlantoaxial fusions are indicated for patients showing >5 mm of instability on flexion-extension views; those with severe cervical cord compression; and those with traumatic injuries, such as a Jefferson or unstable hangman fracture. Historically, the use of sublaminar wiring alone to stabilize this junction with the modified Gallie17 or Brooks18 technique has led to acceptable clinical outcomes, but the development of safe and efficient protocols with lateral mass and pedicle screw fixation in the upper cervical spine has increased fusion rates and often obviates the need for halo vest immobilization postoperatively. Patients with concomitant subaxial cervical fracture or osteoporotic bone leading to poor bone purchase may require a halo postoperatively, but the potential for complications with halo use must be carefully considered, especially in elderly patients12.
The Magerl technique utilizes a cancellous screw that crosses the C1-C2 facet joint (Fig. 6). Percutaneous starting points are made as caudad as T1 to obtain the proper angle for adequate fixation. Biplanar fluoroscopy is often required to identify the correct trajectory of 0° in the medial-lateral plane and 45° in the cephalad-caudad plane. With a starting point in the inferomedial quadrant of the C2 lateral mass, the trajectory is aimed toward the anterior tubercle of C1 on a lateral fluoroscopy view. Either cannulated or noncannulated screws can be used, but the placement and control of the initial Kirschner wire can be challenging. One advantage of this technique is that a standard set of fracture repair screws is sufficient for adequate fixation. A preoperative CT scan of the cervical spine is required to track the course of the vascular structures at this level. The presence of an anomalous vertebral artery precludes the use of this screw.
Transarticular screw fixation at C1-C2. The entry point for the transarticular screw is located in the inferomedial quadrant of the lateral mass (B) with a neutral trajectory in the medial-lateral plane (A). Identification of the vertebral artery and the C2 nerve root can aid in the drilling of this screw (B). (Reprinted, with permission of Elsevier, from: Feiz-Erfan I, Klopfenstein JD, Vougioukas VI, Dickman CA. Surgical therapy for fractures and dislocations of the craniocervical junction and upper cervical spine. In: Kim DH, Henn JS, Vaccaro AR, Dickman CA, editors. Surgical anatomy & techniques to the spine. New York: Elsevier; 2006.)
The Harms technique utilizes axial pedicle or pars interarticularis and atlas lateral mass screws19. Pedicle screws have been found to have a greater insertional torque and pullout strength than either laminar or pars interarticularis screws in cadaveric spines20. The C2 pedicle screw is inserted with a starting point that is just superior and medial to the center of the lateral mass (Fig. 7). A small laminotomy and palpation with a nerve hook can often assist with the medial-lateral orientation (10°), while lateral fluoroscopy can guide the cephalad-caudad direction (15°). The C1 entry point is identified at the junction of the lateral mass and the inferior aspect of the posterior arch (Fig. 7). Often, the overhang of the C1 arch should be removed with an oscillating burr at the level of the lateral mass in order to identify the landmarks and provide room for the screw head. A Penfield retractor can be used to palpate the medial aspect of C1 to guide a 10° medial and 20° cephalad trajectory. The most commonly used titanium screw size is 3.5 × 28 mm. With this procedure, the surgeon should try not to violate the C2-C3 interspinous space, and to leave as many muscle attachments as possible to reduce postoperative pain and increase stability. Any bleeding from the venous plexus between C1 and C2 should be controlled with FloSeal Hemostatic Matrix (Baxter Healthcare, Fremont, California); cottonoids; or a combination of Avitene (microfibrillar collagen) (Devol, Warwick, Rhode Island), Gelfoam (Upjohn, Kalamazoo, Michigan), and thrombin.
C1-C2 fixation with use of the Harms technique. The entry point for the C2 pedicle screw is the superomedial quadrant of the lateral mass (A) directed in a 15° cephalad and 10° medial trajectory. The C1 lateral mass screw is started in the midpoint of the lateral mass just below the arch and oriented in a 10° medial direction for bicortical purchase (B). (Reprinted, with permission, from: Harms J, Melcher RP. Posterior C1-C2 fusion with polyaxial screw and rod fixation. Spine [Phila Pa 1976]. 2001;26(22):2467-71.)
Although the C1 lateral mass screw is ideally placed with bicortical purchase, anterior structures such as the internal carotid artery must be avoided. In a study of 149 CT reconstruction images of the cervical spine, Murakami et al. defined the variation of the anatomy of the internal carotid artery anterior to the C1 anterior arch21. In 64% of patients, the internal carotid artery was directly anterior to the middle of the C1 lateral mass, and in 55% of cases, it was over the lateral third. The authors concluded that a 10° medial orientation of the C1 lateral mass screw with the proper entry point can avoid injury to the internal carotid artery. This ideal trajectory was confirmed with an anatomic study of atlas specimens that demonstrated less variability in C1 lateral mass measurements compared with those of other levels22.
When it is not possible to use C2 pedicle or pars interarticularis screws because of anatomic variations, laminar fixation can be considered. With an entry point 5 mm lateral to the midline of the spinous process and 6 mm caudad to the cranial border of the lamina, a 3.5 × 26-mm screw can often be inserted2 (Fig. 8). In a biomechanical analysis of C1 lateral mass-C2 pedicle screw fixation, C1 lateral mass-C2 intralaminar screw fixation, and C1-C2 sublaminar wire fixation techniques, Elgafy et al. demonstrated that all three instrumentation systems were equally stable in flexion-extension and lateral bending23. While both screw constructs were superior to the wiring technique in axial rotation, there were no significant differences between the lateral mass-intralaminar and lateral mass-pedicle constructs in flexion-extension. In an anatomic study of the axis in an Asian population, Ma et al. demonstrated that 83% of patients had anatomic parameters that allowed for a 3.5-mm screw, while only a unilateral screw was suitable in 12% of cases2. In 5% of the cases examined, the laminae were too thin to accommodate screws on either side. This and other studies highlight the importance of preoperative CT planning when either C2 pedicle or C2 laminar screws are considered.
Entry points for C2 laminar screws. (Reprinted, with permission, from: Ma XY, Yin QS, Wu ZH, Xia H, Riew KD, Liu JF. C2 anatomy and dimensions relative to translaminar screw placement in an Asian population. Spine [Phila Pa 1976]. 2010;35(6):704-8.)
Occipital-Cervical Junction
Posterior fixation for occipital-cervical fusion is required for clinical conditions such as atlanto-occipital dissociation/instability or basilar invagination. Depending on the associated condition, a decompression of the opisthion (the hindmost point on the posterior margin of the foramen magnum) may be needed as well.
It is important to identify a number of landmarks prior to formal instrumentation. The external occipital protuberance is a palpable uprising that marks the thickest portion of the bone (Fig. 9, A). The superior and inferior nuchal lines are ridges that extend in the medial-lateral direction from the external occipital protuberance. The path from the external occipital protuberance to the foramen magnum marks a sharp anterior trajectory that can make plate contouring a substantial challenge. All screw fixation should be caudad to the external occipital protuberance to ensure good bone quality and adequate soft-tissue coverage. At the external occipital protuberance, 12 to 18-mm-thick bone is expected. However, the identification of this point is important since a starting point just 1 cm lateral and inferior to it provides only 5 to 7 mm of bone thickness. Although some authors have advocated bicortical screw purchase, unicortical screws at the external occipital protuberance provide adequate pullout strength as the inner table has only 10% of the overall thickness. A wide array of occipital-cervical plate-screw constructs that allow contouring and flexibility in this articulation are available. Regardless of the instrumentation system, the construct should allow adequate soft-tissue coverage, avoid hardware prominence, and provide a stable plate-to-screw interface to avoid postoperative complications (Fig. 9).
Occipital-cervical fixation. The identification of anatomic landmarks (A) is essential to providing the optimum screw fixation in the occiput. Instrumentation systems should provide for screws in the midline of the occiput and a plate-rod interface that is secure and can be contoured to fit the skull (B and C). (Fig. 9, A, reprinted, with permission, from: Herkowitz HN, editor. The cervical spine surgery atlas. 2nd ed. Philadelphia: Lippincott Williams & Wilkins; 2003.)
Bone Graft
The choice of bone graft to be used in conjunction with posterior cervical fusion depends on a number of factors, including fusion rates, the volume required, complications, comorbidities, and cost. Historically, structural and cancellous autologous iliac crest bone-grafting has been successfully utilized in Brooks and Gallie as well as subaxial fusion constructs. Because the posterior cervical spine is a more favorable biologic milieu for bone-healing than is the posterior lumbar spine, iliac crest bone-grafting can often be avoided. Just as important as the bone graft choice is the preparation of the fusion surfaces, which should include a subtotal facetectomy in the subaxial spine and thorough preparation of the lamina, facet joint, and occiput surfaces in the upper cervical spine. Local bone graft from a laminectomy site can provide an osteoinductive stimulus necessary for bone-healing in this environment24. Bone graft extenders such as ceramic scaffolds, used in conjunction with local bone graft, also can increase osseous healing, and this is the method that I prefer. The use of bone morphogenetic protein (BMP) has led to high fusion rates in the posterior cervical spine, but its use may lead to wound complications and local seroma formation25. At the present time, use of BMP in this surgical setting has not been approved by the Food and Drug Administration in the United States. The complications appear to be linked to the dose of growth factor utilized.
In properly selected patients, posterior cervical decompression and fusion can successfully treat cervical myelopathy and radiculopathy. The choice of a posterior approach versus an anterior approach depends on a number of factors, including sagittal alignment, the type and extent of the pathological involvement, and patient preference. Recent advances in spine surgeons’ understanding of anatomy and technique have increased their level of comfort with using instrumentation in the upper and subaxial cervical spine. Furthermore, the use of concomitant minimally invasive techniques with these procedures leads to faster recovery times with equivalent outcomes. Awareness of the strengths and limitations of posterior cervical techniques can greatly enhance the surgeon's ability to achieve excellent clinical outcomes.