Historical Review

Cervical corpectomy including multilevel strut graft techniques has been reported since the early 1980s.¹³ Caspar introduced ventral cervical plates in 1991.¹⁸ Plating systems evolved over time, and the technology was extended to include cervical corpectomy. Cervical corpectomy with the addition of plating, however, revealed a paradoxical increase in the complications of both graft failure and hardware failure, most notably in multilevel corpectomy and reconstruction. Graft failure in these patients often was an early postoperative event. The grafts frequently failed by inferior and ventral displacement of the graft or by pull-out of the screw and plate from the inferior vertebrae of the construct.^1,3,15

Plates evolved from nonlocking plates that used bicortical screws to unicortical locking plate systems and, most recently, dynamic plates. Constrained plates with unicortical locking screws were reported to loosen and pull away from the ventral cervical spine despite successful fusion of the graft.¹ Also, as the graft subsided, the caudal screws of the plate had the potential to cut into the adjacent disc, necessitating reoperation.⁴ Dynamic plates have been promoted due to improved load sharing across the plate and resultant fusion rates.¹⁹ Graft subsidence and loss of lordosis, however, have been shown to be higher with dynamic plating systems.²⁰ Excessive graft subsidence adversely affects the caudal screws of a dynamic plate in the same as it affects a constrained plate.

Biomechanical Studies

DiAngelo and Foley described reversal of the load transfer with plated multilevel strut grafts. With plating, the graft is unloaded in flexion and excessively loaded with extension. After plating, even a small degree of extension (7.5 degrees) caused the strut graft to fail at the caudal vertebra with fracture of the lower vertebra, excessive graft subsidence, graft dislodgement, or plate failure.^21,22 These same patterns of graft failure are found in clinical series.^1,3,4,15 The stand-alone graft in comparison is loaded in flexion and unloaded in extension.^21,22

Brodke et al. showed biomechanical superiority of dynamic plates after simulated subsidence in a single-level corpectomy model due to the load sharing across the graft.²³ Clinical studies have confirmed that dynamic plates increase fusion rates.¹⁹ It is important to remember, however, that graft dislodgement is generally an early postoperative event prior to the anticipated failure from lack of fusion.^1,3 Furthermore, both dynamic plates and constrained plates would be expected to provide a tensile force during extension.

Kirkpatrick et al. studied the biomechanics of three-level corpectomies. Intact specimens, graft alone, ventral plating alone, and dorsal instrumentation alone were compared. Dorsal instrumentation was found to have a greater strength and stiffness in both flexion and extension than a ventral plate. Dorsal fixation provides segmental instrumentation at the level of the corpectomy, as well as above and below, thereby applying three-point bending forces that cannot be achieved with ventral plating.²⁴

Koh et al. compared ventral, dorsal, or ventral-dorsal stabilization of three-column injuries in the cervical spine. Dorsal instrumentation was superior to ventral plating, and the addition of a ventral plate did not significantly increase stability compared with dorsal plating and interbody grafting.¹⁶

Singh et al. studied a two-level corpectomy with ventral plating, dorsal lateral mass screw-rod construct, and combined ventral and dorsal instrumentation. In all pure moments tested, the specimens with dorsal instrumentation outperformed those with ventral plating. No difference was found in any between the dorsal alone and combined ventral-dorsal instrumentation. Furthermore, the authors noted that ventral plating sometimes limited compression of the graft, with the net result being a less biomechanically rigid construct when coupled with dorsal instrumentation.¹⁷

Surgical Techniques

Multilevel anterior cervical corpectomy and strut grafting have a steep learning curve. The incidence of early graft dislodgement has been reported to decrease with experience. One critical test for strut graft security is resistance to displacement under flexion and extension under direct visualization prior to wound closure.²⁵ Placement of a plate compounds rather than corrects poor intraoperative strut graft sizing.²⁶ Optimal graft placement and sizing are especially important in the reconstruction following multilevel corpectomy.

The procedure often is performed with intraoperative neuromonitoring. Somatosensory-evoked potentials, motor-evoked potentials, and free running electromyography of C5-T1 all can be used for intraoperative monitoring. In conjunction with total intravenous anesthesia, the integrity of the descending motor tracts, nerve roots, and peripheral nerves can be ensured using these modalities.²⁷ The patient is positioned supine with Gardner-Wells tongs, and cervical traction of 5 to 10 pounds is used initially.

The approach is typically via a transverse incision in the skin on the right or left based on surgeon preference with undermining of the skin above the platysma muscle. The platysma is then divided in line with the muscle fibers, and placement of stay suture at either end can facilitate exposure and obviate the need for a retractor placed superior to inferior. A vertical incision along the ventral border of the sternocleidomastoid has been described²⁵ but is not a requirement for anterior cervical corpectomy even in the multilevel reconstruction.²⁸ Ventral plating over a long segment is aided by the additional exposure and visualization of the plate at both ends simultaneously. Ventral plating also requires additional soft tissue retraction to triangulate the screws at either end and for plate placement. Cervical corpectomy without plating typically does not require this additional exposure or retraction.

Discetomies at each level are performed initially. The corpectomy trough is made in the central two thirds of the vertebral bodies. Generally the trough is between 14 and 18 mm wide. Some authors have advocated preservation of the posterior longitudinal ligament if no instrumentation is planned.²⁸ With removal of the posterior longitudinal ligament, however, graft displacement rates without plate usage were not found to be increased in one series.²⁹ Careful review of preoperative films for rotational deformity or vertebral artery anomalies should minimize the small risk of a catastrophic intraoperative vertebral artery injury. Small curets can be utilized to safely free osteophytes causing spinal cord compression within a kyphotic deformity. Intentional preservation of the posterior longitudinal ligament, when appropriate from the spinal cord decompression standpoint, will add stability to the strut graft if no supplemental instrumentation is planned.²⁸ Bracing to prevent excessive cervical extension should be utilized if no posterior instrumentation is planned.

Prior to placement of the strut graft, skeletal traction is increased to 30 to 40 pounds, reducing any preoperative kyphosis and facilitating a snug graft placement. Depending on the type of graft to be used, the end plates are prepared to accept the allograft, autograft, or cage. With autogenous fibular or iliac crest autograft, the ends of the graft are rounded, and the end plates are fashioned with seating holes, providing a peg-in-hole design.²⁸ Excessive traction has been associated with oversizing of the strut graft, and similar graft failure rates have been found without the use of intraoperative traction.²⁹ Autograft is associated with additional morbidity from the graft harvest site, reportedly as high as 34%.³⁰

Multiple techniques have been employed to limit the morbidity of autograft with and without plating. Fibular allograft also has been used with good results without plating.^25,26,29 Titanium mesh cages (TMCs) allow lordotic contouring and can be packed with autograft taken from the corpectomy defect. In this case, the sharp footprint off the titanium mesh will subside into the bony end plate, providing a secure fit.¹¹ End caps should be used at the end of the implant to increase the contact area to minimize subsidence.³¹ Unfortunately, the end caps usually do not allow variation of orientation relative to the end plate, and cage subsidence is problematic. Daubs reported a high early failure rate of TMC with ventral plating due to subsidence and distal plate extrusion.¹⁵

Expandable titanium cages that allow for easy insertion also have been designed. Spikes or contouring at either end allow for fixation to the end plates. The broader surface at each end of this implant decreases subsidence.³² Limits of this implant, however, include decreased space for autograft packing, decreased contact of the autograft at the ends of the graft, inability to contour the implant into lordosis, and limitation of postoperative imaging. Titanium prevents accurate assessment of fusion postoperatively with CT and causes significant interaction with MR imaging of the neural elements in the event of neurologic decline. Finally, the force created by an expandable cage can be large, and adjacent-level fracture in the coronal plane has been reported with expandable cages.³³

Lastly, polyetheretherketone (PEEK) now is available in a stackable implant for use in the cervical spine. PEEK grafts are stronger than allograft but softer than titanium and less likely to cause the adjacent bone to collapse. The footprint is significantly wider than TMC, allowing force distribution along the end plate. In addition, the polymer does not interfere with CT scans or MRI, allowing for easier postoperative film interpretation and fusion assessment.³⁴ Clinical studies for efficacy in multilevel cervical corpectomy reconstruction are currently unavailable.