Bony Anatomy

The thoracolumbar spine is divided into five regions: the cervicothoracic junction, the mid-thoracic spine, the thoracolumbar junction, the mid-lumbar spine, and the lumbosacral junction. Each region has distinct anatomic and biomechanical characteristics that must be considered when planning reconstructive and instrumentation surgery. The anatomy typically serves the biomechanics. For example, facet orientation predicts motion segment direction and range of motion. Facet orientation also “couples” motion so that flexion necessitates translation.⁷

Placement of a thoracolumbar implant affects the spinal ligaments directly or indirectly. Direct anterior lumbar interbody approaches require sacrifice of the anterior longitudinal ligament. In disc replacement procedures, to achieve more parallel distraction, the posterior longitudinal ligament may need to be resected. The long-term effect of such ligamentous resection on motion preservation kinematics is unknown.⁸

The thoracic spine can be divided into two subregions: the upper thoracic spine to T8 and lower thoracic spine (T8-thoracolumbar junction). The articulation of the thoracic spine with the rib cage makes the thoracic spine the most rigid portion of the spine. Extending the three-column concept of Denis, an intact rib cage and sternum complex functions as a mechanical fourth column, preventing flexion and extension above T9.⁹

The upper thoracic spine allows significant segmental rotation (10 degrees). Below T8-9, the major plane of motion is flexion and extension.⁹ Before proceeding with a transthoracic approach and spinal reconstruction, the surgeon must be familiar with the articulations between the rib and vertebral body. There are two sets of demifacets: one at the disc level and the other at the transverse process. Radiate ligaments stabilize the articulation further. The rib attaches to the transverse process and the superior aspect of the same-number vertebra (at the level of the pedicle). For example, to reach the T9-10 disc, one can follow the T10 rib to the superior aspect of the T10 body.

The rib–pedicle–transverse process junction is critical in posterior approaches as well. Because the pedicles of the mid-thoracic spine are quite narrow, some authors have recommended an in-out-in approach for pedicle screw insertion. With this technique, the pedicle screw trajectory begins dorsally, but as the pedicle narrows, the screw passes laterally into the space between the rib and the pedicle. In this space, it contains only ligamentous tissues, and penetration does not jeopardize neurologic structures or the lung parenchyma.¹⁰

The thoracolumbar junction represents a straight segment between the lordosis of the lumbar spine and the kyphosis of the thoracic spine. This is a mechanical transition zone from the more rigid thoracic spine and the more mobile lumbar spine. Thoracolumbar fractures most commonly occur in this transition zone of the spine.

The lumbar spine permits significant flexion and extension across all levels. There is a sharp increase in the amount of lateral bending exhibited at L3-4. There is less lateral bending at L2-3 and L4-5. In the lumbar spine, axial rotation is limited by the vertical orientation of the facets.

Thoracolumbar bone and ligaments provide the framework for the development of fixation strategies. To obtain successful spinal fixation, a stable bone-implant interface is required. Given the proximity of neural elements, only certain bone surfaces can be used for spinal instrumentation. In addition, the size and shape of the bony anatomy influence what type of implant can be used and the fixation strength of the implant. Larger pedicles are able to accommodate larger screws for fixation, which increase the stability of fixation and increase holding strength.

There is significant anatomic variation among adults. Using computerized imaging systems, it is possible to use axial and sagittal images for preoperative planning. The maximum length and appropriate screw diameter can be employed.

Neurovascular Anatomy

A primary goal of thoracolumbar instrumentation is to prevent neurologic injury and protect the neural elements. It is crucial to understand the anatomy. The spinal cord typically ends at the L1-2 disc space. Below the conus, the nerve roots pass from the central thecal sac through the neuroforamen into the pelvis. Several cadaveric and imaging studies have described the relationship of the bony elements and their proximity to the neural structures. One cadaveric study measured the average distance from lumbar pedicle to the dural sac medially. From cranial to caudal in the lumbar canal, the range was 1.29 to 1.56 mm; clinically, this means that a medical pedicle breach greater than 1.29 mm has a significant chance of contacting or injuring the dura (Fig. 71–1).The pedicle is farther from the superior nerve root at 4.12 to 5.52 mm but closer to the inferior root, where distances ranged from 1.10 to 1.06 mm. The nerve roots and dura are statistically further from the L5 pedicle than from other pedicles, making the L5 pedicle safer than other lumbar pedicles for screw insertion.

FIGURE 71–1 These images show the importance of good exposure. A, Axial CT scan of a patient with marked facet hypertrophy. In most cases, complete takedown of the soft tissue around the lateral pars and medial transverse process provides enough anatomic clues for appropriate positioning of the pedicle probe. In cases with marked facet hypertrophy such as this one, resection of the facet osteophytes significantly improves identification of normal anatomic landmarks, improving pedicle screw placement accuracy. B, Anteriorly, the soft tissues must be reflected posteriorly enough to palpate the posterior vertebral body margin. Then the surgeon has a markedly improved three-dimensional sense of the position of the spine in space and, it is hoped, can avoid inadvertent penetration of the spinal canal with a screw. Proper positioning of the patient is critical. If the patient is rotated, the surgeon’s orientation to the spine may be confused. That may have contributed to the misplacement of the screws in B and C. Adequate exposure also requires adequate soft tissue release. Particularly in patients with a narrow pelvis, if the thoracodorsal fascia remains tight, the surgeon would have to struggle to achieve proper medialization of the screws. D, Postoperative axial CT scan shows S1 screws. The left-sided screw is in acceptable position, whereas the trajectory of the right-sided screw is not medialized enough. The screw penetrates the anterior sacrum and impinges on the L5 root anteriorly.

It is important to understand the relationship of the pedicle and the exiting nerve root. The cervical spine has eight nerve roots. The corresponding nerve root exits the neuroforamen above the lowest numbered pedicle (i.e., the C6 nerve root exits at the C5-6 level). This numbering changes in the thoracic spine and lumbar spine; for example, at the T11-12 level, the T11 root exits beneath the 11th pedicle. After the L5 nerve root exits the neuroforamen, it travels anterior to the sacral ala. An S1 pedicle screw that is placed too far lateral or is too long places the L5 nerve root at risk for injury. Knowledge of the intimate relationship of the major vessels to the spine is crucial to avoid life-threatening complications during anterior and posterior thoracolumbar instrumentation.

When approaching the upper thoracic spine anteriorly, the approach is typically from the right to avoid the arch of the aorta. Scoliotic curves are typically approached from the convexity of curve, allowing a more complete release. The lumbar spine is typically approached from the left for several reasons. First, the liver is on the right and is more difficult to mobilize. Second, a left-sided approach brings the surgeon into contact with the aorta before the vena cava. The aorta is more easily recognized and is more durable, so the risk of sudden, catastrophic blood loss is less in the event of a vascular injury. With the advent of lumbar disc arthroplasty and anterior lumbar interbody fusion (ALIF), which require an anterior approach, the aorta, inferior vena cava (IVC), and iliac vessels are at risk for potential injury. Magnetic resonance imaging (MRI) is vital for preoperative evaluation of the vascular structures and for detecting potential vascular anomalies. Exposure of the L4-5 disc space requires retraction of the aorta and IVC from left to right. To mobilize these structures, the iliolumbar vein must be identified and ligated. When exposing the L5-S1 disc space, the middle sacral artery and vein must be ligated, and the left common iliac vein is elevated superiorly and laterally. When considering exposure to the lower lumbar spine, vascular calcifications in an older patient may limit retraction of the vessels and may be a contraindication to anterior surgery. In addition, any patient with previous abdominal or hernia surgery must be carefully evaluated because scar tissue may be a contraindication to abdominal surgery (Fig. 71–2).

FIGURE 71–2 Preoperative planning is crucial when selecting the implants or the angle of the approach to the spine. In certain cases, one implant may suit a patient better than another. A and B, Cross-sectional MRI of L4-5 disc space. Note the right iliac vessels are more lateral than expected. This type of anatomy may preclude the use of a transpsoas approach.

When performing posterior thoracolumbar surgery, it is crucial to be cognizant of the anterior vascular anatomy. The heart-shaped thoracic vertebral bodies can allow passage of thoracic pedicle screws out of their respective bodies and place the aorta at risk for injury. One study measured the average distance from the aortic wall to the vertebral body cortex at the apex of the curve by MRI and found it was greater in patients with scoliosis (4.0 mm) than in a normal group (2.5 mm) (P < .05). The distance from the posterior aspect of the aorta to the anterior aspect of the spinal canal was less in the scoliosis group (11.1 mm) than in the normal group (19.2 mm) at T5-12 (P < .05). The aorta was positioned more laterally and posteriorly at the T5-12 vertebral bodies in these patients.¹¹ When placing pedicle screws, preoperative planning is essential to determine the appropriate screw length to minimize the risk of vascular injury. Typically, pedicle screws on the left side place the aorta at more risk of injury, and slightly shorter screws should be placed.

With the development of transforaminal lumbar interbody fusion procedures, there is risk of vascular injury. If there is violation of the anterior anulus with overzealous discectomy, the aorta, IVC, or iliac vessels are at risk. Papadoluas and colleagues reported that the mortality from a vascular injury during lumbar surgery can be 10%. At the L3-4 level, the aorta is more at risk than the IVC and common iliac vein. At L4-5 and L5-S1 levels, the common iliac artery is more at risk than the common iliac vein, IVC, and aorta.

Important Factors in the Selection of Lumbar Implants

The following are important factors to keep in mind when selecting lumbar implants:

Basic Biomechanical Principles

The ability of instrumentation to stabilize the spine adequately is a function of failure load, which represents the mechanical load at which the implant failure occurs, and construct stiffness. Stiffness represents the ability of the spine-instrumentation construct to resist applied forces, such as axial compression. The spine and instrumentation constructs are subjected to linear and circular forces. Linear forces are applied along the line of action. Circular or moment forces occur at any time that force is applied to a point not located along the line of action. Torque represents the magnitude of the applied force times its perpendicular distance from the axis of rotation. Also, forces may be isolated or coupled; coupling refers to two noncolinear forces acting around the same axis.

Stress reflects the cross-sectional area over which the force or load is applied. Host tissue and spinal implants respond to force application by deformation (a change in shape and size). Deformation is predictable, based on the shape, size, and intrinsic material properties of the involved structure. Strain is defined as deformation divided by original length. The stiffness of a structure reflects strain relative to the stress applied. Because strain incorporates the cross-sectional area of the material, cross-sectional area has a marked impact on stiffness. A rod that is half the diameter of another rod made of the same material would have only one quarter the stiffness of the other rod. When a load is removed from an object, elastic deformation occurs if the material recovers its normal shape. Plastic deformation occurs when the load permanently deforms the material; in other words, stress is no longer proportional to strain. The material’s yield point is the load beyond which it can no longer regain its normal shape (i.e., elastic deformation changes to plastic).

A ductile material allows permanent deformation before failure. Ductile materials include metals such as steel and titanium. Brittle materials, such as adult cortical bone and polymethyl methacrylate (PMMA), can fail without deformation. Another mechanical concept that relates to the effect of degenerative change on segmental motion is the neutral zone. Within the neutral zone, the spine offers minimal resistance to motion. Minor changes in load can lead to major shifts in position; for example, a patient with disc damage may exhibit an increase in the neutral zone, allowing motion to occur beyond the pain-free zone under physiologic loads while showing no change in the spine’s overall range of motion. Operative stabilization may decrease pain by decreasing the neutral zone but typically also decreases ultimate segmental range of motion.^12,13

Implant Materials and Corrosion

Most posterior thoracolumbar fixation systems are made of stainless steel, pure titanium, or, most commonly, titanium-aluminum-vanadium alloy. Cages and spacers used anteriorly may also be made of other materials, such as carbon fiber or polyetheretherketone (PEEK). Machined allograft that is shaped like a cage or spacer may be used to facilitate anterior fusion. Surgical stainless steel is very strong, with yield strength of 700 MPa. It is also stiff, having a modulus of elasticity 12 times that of normal bone. Steel implants usually have cobalt-chromium alloy and molybdenum to enhance corrosion resistance. Titanium alloys tend to have greater native biocompatibility and corrosion resistance. Titanium has a modulus of elasticity only six times greater than bone, which makes it easier to bend and insert than steel. These characteristics allow titanium-based devices to transfer load effectively to the vertebral body, shielding the implant from stress. A reduction in stress shielding allows bone grafts to respond to more load appropriately. Use of titanium alloys is increasing because of its improved imaging characteristics, high strength-to-weight ratio, enhanced ductility, and increased fatigue life.

Two types of titanium are typically used. Pure titanium is recommended only when very low strength is needed because it has a low yield strength of only 170 to 485 MPa. More typically, a titanium-aluminum-vanadium alloy, with yield strength of 800 MPa, is employed; its greater strength does not change the favorable modulus of pure titanium (110 GPa).¹⁴ Ultimate tensile strength—the maximum stress a material can sustain without changing shape—may be altered during surgery. Titanium rods are particularly sensitive to notching.¹⁵ If a complex rod contour is required, a template should be employed to minimize the amount of rod bending. In some patients with rigid dual curve deformities, it may be more effective to use separate rods in the thoracic and lumbar spine, employing rod-to-rod connectors (“dominoes”) to complete the construct.

In a biologic environment, fretting and corrosion can occur between the modular components of a spine construct. Kirkpatrick and colleagues subjected 48 spinal implant constructs to surface analysis stereomicroscopy.¹⁶ Titanium alloy implants (n = 25) showed no significant corrosion, but three of the constructs showed fatigue failure of the anchoring screws. The cobalt alloy construct showed no evidence of corrosion. Semirigid stainless steel implants had mild surface alteration, whereas rigid constructs showed moderate to severe corrosion. Based on their findings, the authors recommended avoiding rigid stainless steel implants or constructs with different surface finishes between rods and connectors. The surgeon must use caution when combining implants made of different metals. Mixing stainless steel with titanium would lead to a galvanic response and early corrosion, although titanium has been used with cobalt-chromium alloy without significant corrosion or complications.

Increasingly, nonmetallic implant materials have been used in thoracolumbar implants. Typically, these materials are used as cages, spacers, and graft containment systems rather than as fixation systems. Advantages of radiolucent materials such as PEEK include easier radiographic assessment of graft integration. For some nonmetallic implants, the modulus of elasticity is closer to that of host bone, allowing greater load sharing. Mechanical testing showed acceptable mechanical and fatigue characteristics for PEEK as a load-bearing implant material.¹⁶

The material properties of implants are also affected by manufacturing variables, such as drill holes, structural imperfections, and surface irregularities. Implant fatigue is an important cause of failure. The average spine cycles 3 million times per year.^17–19 Because current implants are overengineered for their designated function, implant failure is more likely to occur from improper selection than from mechanical properties. If bone healing is delayed or incomplete, the implant or construct ultimately fails, so meticulous attention to bone grafting technique is imperative.

Fusion

Two important goals in designing an appropriate spinal construct are to provide spinal stability and to facilitate healing. The trend away from autogenous graft to allograft generally lengthens the duration of bone healing. Numerous animal models have shown that instrumentation increases the rate of fusion maturation.²⁰ A delay in fusion may increase the risk of implant failure or the propensity for collapse of structural grafts.²¹ Use of bone morphogenetic protein generally accelerates healing and may permit use of less fatigue-resistant constructs.

Many postoperative factors affect healing times, including motion, loading, muscle status, medical comorbidities, nutritional status, and tobacco use. The type of fusion performed and the type of graft employed also have significant implications on the likelihood and rate of healing. For fusions expected to heal slowly, more robust forms of instrumentation, possibly with additional anchor points, should be employed. For example, lower fusion rates in smokers may justify instrumentation in settings in which in situ fusion would otherwise be appropriate.

When clinical or mechanical circumstances increase the risk of pseudarthrosis, additional steps, such as addition of L5-S1 interbody fusion below a long posterolateral fusion, should be considered.²² Along the same lines, the anticipation of increased postoperative loading, poor patient compliance, or inadequate postoperative immobilization may warrant more rigid forms of operative stabilization. Examples include patients with neurologic or motion disorders who are subject to increased spinal loads and patients with spinal cord injury or a colostomy for whom brace immobilization is impractical.

Historically, the most common fusion technique was the posterior fusion. The primary advantage of this approach was easy surgical access to the midline posterior elements (spinous processes and lamina). Disadvantages included its limited utility in laminectomy patients. Also, the graft material lies distant from the center of rotation and experiences significant tensile forces with spine flexion. This distance increases tensile stress and motion on the graft that could lead to migration, excessive motion, or graft resorption and ultimate nonunion.²³ The most common contemporary fusion procedure is the intertransverse (posterolateral) fusion in which the facet joints, lateral pars, and transverse processes are decorticated and grafted, leaving the lamina accessible for decompression. In an intertransverse fusion, the graft material is placed closer to the center of vertebral rotation.²⁴ The disadvantage is a poor vascular bed and a decreased area for fusion.

Interbody fusion provides significant mechanical advantages in terms of graft compression and a large, well-vascularized fusion surface. The anterior column fusion spans the neutral zone and, when healed, represents the strongest mechanical block to segmental motion.^23,25 Even a solid intertransverse fusion may fracture or elongate if excessive or repeated load is placed across the motion segment.²⁴

Thoracolumbar implants share applied loads with the spine until a stable fusion occurs. If a construct bears most of the load, stress shielding of the spine results and may lead to device-related osteopenia.²⁶ The clinical sequelae of this shielding include graft resorption and possible implant failure. Increased emphasis has been placed on load-sharing implants in recent years.

Implications of Osteoporosis

Osteoporosis is the most common metabolic bone disorder and results from loss of the crystalline (inorganic) and collagenous (organic) portions of bone. Throughout life, the body constantly remodels bone by removing old bone and creating new bone. Although the pathomechanics are incompletely understood, osteoporosis occurs when the rate of bone resorption exceeds the rate of bone formation. Lower rates of bone formation result in a decline in overall mineral density of bone. Unbalanced osteoclast activity results in disruption of the normal connectivity between bony trabeculae. Bone can be weakened in the material and architectural sense.²⁷ There are an estimated 40 million people at risk for osteoporosis in North America. With the aging of the population, this estimated number is likely to triple over the next 3 decades. Management of spinal disorders in an osteoporotic spine will be a significant problem in the future.²⁸ Thoracolumbar instrumentation options in patients with osteoporosis are limited because implant failure in osteoporotic bone is common. Even if healing occurs uneventfully, patients with osteoporosis are at risk for compression fractures and spondylolisthesis adjacent to rigid constructs.^29,30

For many patients, the diagnosis of osteoporosis and initiation of appropriate management is delayed. Because the consequences of failure to recognize osteoporosis are so high, it is incumbent on the spine surgeon to screen at-risk patients.³¹ There are three main types of osteoporosis: type I (postmenopausal), type II (senile), and type III (secondary). Type I affects the trabecular bone of women, more than men, in their 6th and 7th decades. Type II osteoporosis arises in the 8th and 9th decades and increasingly affects the cortical bone of men and women equally. Although categorizing a thin, elderly, white, or Asian woman at risk for osteoporosis may be relatively straightforward, younger and larger patients are increasingly at risk for secondary (type III) osteoporosis. Excessive endogenous or exogenous cortisol is deleterious to bone mass and is a cause of secondary osteoporosis. Long-term use of thyroid replacement drugs, blood thinners, and various seizure medications may also result in osteoporosis, and patients taking these medications should undergo screening before major spinal reconstruction. At least 30% bone mass loss is needed to identify osteopenia reliably on plain radiographs. Dual-energy x-ray absorptiometry (DEXA) is a much better screening tool than plain radiography.³² Results of DEXA scans are given in T and Z scores. The T score compares the patient’s bone mineral density (BMD) with mean values for healthy, same-gender young adults. For each standard deviation below the norm, fracture risk increases 1.5-fold to 3-fold. A T score of −1 implies a 30% chance of fracture. As the T score decreases from −1 to −2, the risk of instrumentation failure increases significantly. The Z score compares BMD with age-matched controls. A Z score less than −1.5 warrants a more extensive workup for osteomalacia or neoplasm. DEXA values are falsely increased with scoliosis, compression fractures, spondylosis, extraosseous calcification, and vascular disease. In many spine patients, the T score at the hip may be more accurate than the spine value.

Depending on the patient’s activity level, the nature of the intended surgery, and the severity of the osteoporosis, preoperative initiation of antiosteoporotic management and delay of elective spine procedures until follow-up DEXA scores improve may be warranted.³³ Calcitonin, via subcutaneous injection or nasal spray, decreases osteoclastic bone resorption. Over the short-term, calcitonin also enhances bone formation, leading to a slight net bone accretion. Over the long-term, osteoblastic activity slows, however, and bone mass stabilizes. That is, after several years, calcitonin is no longer effective.³⁴ Bisphosphonates dramatically suppress bone resorption and decrease hip and spine fractures. These agents directly stabilize the bone crystal, making it more resistant to osteoclastic bone resorption. They also inhibit osteoclast activity. Bisphosphonates preserve bone architecture and overall density. Weekly and monthly dosing of these agents improves compliance with no increase in toxicity. Most patients with osteoporosis should be receiving a bisphosphonate. For patients who cannot tolerate or have not responded to bisphosphonates and for patients with severe osteoporosis and major spinal instability, more aggressive antiosteoporotic management in the form of pulsed parathyroid hormone administration should be considered. Parathyroid hormone (Fortéo) is anabolic to bone and leads to early, dramatic increases in bone mass.³⁵

When planning a spinal reconstruction procedure in an at-risk patient, it is important to recognize areas of the spine that are vulnerable to the ravages of osteoporosis. Trabecular bone represents 20% of the total bone mass and is found in the metaphyses and epiphyses of long bones and in the cuboid bones (including the vertebrae). Because trabecular bone exhibits eight times greater metabolic activity than cortical bone, the mechanical impact of osteoporosis affects trabecular bone earlier and to a greater degree than cortical bone (Fig. 71–3).^27,36 It was reported in a more recent article that threshold BMD for successful use of anterior spinal instrumentation was 0.22 g/cm (as measured by quantitative computed tomography [CT]).³⁷ Often, combined anterior and posterior surgeries with multiple, additional fixation points are required to achieve adequate fixation.^37,38

FIGURE 71–3 This woman with unrecognized osteoporosis and a fracture of L1 was initially treated with a thoracolumbosacral orthosis. She presented with acute lower extremity weakness and urinary retention. She required urgent surgical intervention to restore her alignment and canal clearance. A, X-ray in brace showing fracture at L1. B, Axial CT scan of osteoporotic burst fracture. C, Sagittal CT scan showing retropulsed burst fracture fragment in canal. D, Postoperative lateral and anteroposterior radiographs of burst fracture reduced with posterior instrumentation.

Other strategies to improve the stability of fixation in an osteoporotic spine include augmentation of screw tracts with PMMA, use of laminar hooks to “protect” inferior pedicle screws, expanding screw designs,³⁹ triangulated screw placement, increased use of transverse connectors, and bicortical vertebral body purchase. BMD is linearly related to screw insertion torque and pullout strength.⁴⁰ The surgeon’s tactile sense of purchase when placing the screw relates to construct strength.⁴⁰ PMMA may be added to screw tracts to increase pullout strength significantly, or bicortical purchase may be sought.^40,41 Overall, BMD has a greater impact than unicortical purchase on screw pullout strength.^42,43 Improved unicortical screw strength can be achieved with triangulated and subchondral placement.^44,45 If bilateral screws are placed in a triangulated pattern, use of a transverse or cross connector further increases fixation strength. Mechanically, stability is improved by the presence of bone between the screws rather than merely by the bone within the threads of each screw individually.⁴⁴ Bicortical screw placement may involve additional surgical risk, although it improves holding power and improves the construct’s resistance to cyclic loading. Bicortical purchase with posterior cortical pedicle screws offers less strength than anteriorly placed vertebral cancellous screws.^46,47 Posteriorly and anteriorly placed screws benefit from purchase into the stronger subchondral bone just below the endplate, but the effect is more pronounced anteriorly.⁴⁷ Bicortical purchase is typically not recommended for thoracolumbar pedicle screws because of risk of injury to anterior vascular structures. For sacral screws, especially in long constructs, bicortical screws improve holding power and sagittal plane correction.^48,49 Some authors have recommended aiming the screw upward into the disc space or through the sacral promontory.^50,51

With particular clinical challenges, such as osteoporosis, changing specific aspects of the implants themselves may improve fixation. In patients with osteoporosis in particular, maximizing pedicle screw diameter improves pullout strength and decreases the risk of fatigue failure of the screw.⁴⁸ Screw length is linearly related to pullout strength. Little difference is seen between self-drilling or self-tapping designs⁴³; if tapping is performed, undertapping by 1 mm leads to greater pullout strength than undertapping by 0.5 mm.⁴⁸ In many cases, the angle of implantation is as important as the implant characteristics themselves. In anterior and posterior constructs, one should seek to avoid parallelogram constructs in which the screws and longitudinal members form a perfect square or rectangle. These constructs resist lateral loads less well than a triangulated pedicle screw construct.⁵² In anterior dual-rod constructs, a trapezoidal short-short/long-long construct is used.

Posterior Approach

The midline posterior approach is the most common approach for placement of thoracolumbar instrumentation. This extensile approach is applicable from the occiput to the sacrum. Below L2-3, the level of the conus medullaris, the dural sac may be safely retracted to afford enough exposure to the posterior disc space for performance of a posterior lumbar interbody fusion. In the upper lumbar spine, the risk of neural injury with dural retraction increases, and more oblique approaches to the disc space, such as transforaminal lumbar interbody fusion, are safer. In the thoracic spine, a midline posterior approach is considered dangerous for decompression of anterior compressive pathology (Fig. 71–4).

FIGURE 71–4 Anterior interbody options are numerous with various types of metal, polyetheretherketone, and allograft cages available. In some cases, an anterior cage alone may be inadequate stabilization for reliable fusion. Numerous devices have been introduced to improve graft or segment stability. A, Example of the many “anti-kickout” implants available. The simplest of these is a screw and a washer. The goal of this instrumentation is to prevent graft extrusion anteriorly. B, More sophisticated and biomechanically rigid technique. This plating system seeks to control extension, a common failure mode for “stand-alone” anterior lumbar interbody fusion. C, This surgically more aggressive means of fixation employs translaminar facet screws to augment dual cylindric cages. Percutaneously inserted pedicle screws to systems have been popular in this application as well. D, Lateral x-ray showing L4-5 degenerative spondylolisthesis. E, Postoperative lateral x-ray after posterior pedicle screw instrumentation. Posterior fusion options are numerous. Numerous devices have been developed to improve segment stability and facilitate fusion.

(A, DePuy Bowt’s plate courtesy of DePuy Spine, Raynham, MA. B, Courtesy of Danek Pyramid Plate, Memphis, TN.)

The powerful retractor systems available for posterior instrumentation procedures create extremely high levels of intramuscular pressure. Over time, this pressure can cause muscle necrosis similar to a compartment syndrome. For longer cases, the retractors should be removed every 2 hours to allow muscle recovery.^33,34 This recovery time is particularly important because preoperative muscle abnormalities may exist in some conditions, which can be accentuated by the trauma of extended muscle retraction. One study found that preoperative paravertebral muscle biopsy specimens in 30 patients with spondylolisthesis were histologically different compared with normal controls.³⁵ Most thoracolumbar instrumentation requires a relatively wide exposure beyond the facet and out into the transverse process.

The lateral extracavitary or costotransversectomy approach allows access to thoracic vertebral bodies from a posterior approach without violation of the pleura or takedown of the diaphragm.³⁶ Visualization is reduced, however, compared with visualization for a corpectomy. Significantly more bone resection (including the rib, costotransverse joint, facets, and pedicle) is required to achieve that visualization.^36,37 Costotransversectomy is associated with increased blood loss, longer operative time, increased paraspinal muscle disruption, and chest wall numbness from intercostal nerve resection. Occasionally, anterolateral cord compression can be addressed with a compromise approach between a standard laminectomy and a formal costotransversectomy. These transpedicular decompressions are known by various names, such as pediculofacetectomy, and are particularly useful in patients with tumors in whom the neoplasm has already destroyed most of the pedicle.³⁸ In this case, the decompression is mainly soft tissue removal, and the anterior compressive elements can be removed indirectly by pulling them away. Although this approach confers limited visualization, it reduces operative time, blood loss, and iatrogenic destabilization.^39,40 Visualization can be improved with a 70-degree endoscope.⁴¹

In the lumbar spine, the Wiltse paramedian approach, although described for resection of far-lateral disc herniations, has been used for interbody and posterolateral fusions. A bilateral paramedian approach is best when no midline decompression is needed. Additionally, this approach could be considered for posterolateral pseudarthrosis revision to avoid midline scar. A Wiltse paramedian type of approach is used in most minimally invasive surgical fusions employing tubular retractors, microscopic dissection, and fluoroscopic guidance. Iatrogenic muscle injury can result from denervation of the primary motor branch of the dorsal primary ramus when the muscle is stripped from the midline beyond the facets.⁴³ Many studies have examined the effects of muscle splitting versus subperiosteal dissection on postoperative muscle health. One MRI study compared muscle enhancement after tubular discectomy versus a standard open discectomy. This study failed to show that microendoscopic discectomy produced less paravertebral muscle disruption than a traditional approach.⁴⁴

Facet-Based Approaches

Numerous methods of facet fixation have been proposed. Historically, small screws were placed directly through the facet joint. Although these screws were able to lock the facets, their short length produced too small a lever arm to counteract the forces to which the lumbar facet is exposed, and many of these implants failed. Techniques at the present time are exploring additional methods of bony purchase. The most common of these is the Magerl technique, in which 4.5-mm, fully threaded cortical screws, usually 50 to 60 mm in length, are passed from the opposite side of the spinous process through the ipsilateral lamina and across the facet.⁵⁵ The Boucher technique is similar but incorporates a greater degree of pedicle penetration.⁵⁶ In both of these techniques, the screws are not placed in lag mode. In addition, they do not provide the strength that the three-column purchase of transpedicular instrumentation affords. These techniques are indicated for one-level or two-level fixation when reduction is not needed.

These translaminar facet screws are often used as posterior column extension supports for anterior lumbar interbody fusion procedures. The advantages of these screws include minimally invasive insertion via a 4-cm percutaneous approach. Translaminar facet screws are much less expensive than transpedicular constructs and offer less impingement on surrounding musculature. Disadvantages include the need for postoperative immobilization, which delays rehabilitation. Contraindications to translaminar facet screws include absence of adequate lamina or facet joints. These screws may be placed after laminotomy procedures but not typically after full laminectomy. Given that these screws are less rigid than transpedicular instrumentation, they are probably best reserved for slim, low-demand, compliant patients with intact laminae and facet joints.⁵⁵

Using contemporary techniques, fusion rates are high, and cases of screw failure are rare. One article examined the outcome of single-level and multilevel fusions for degenerative disc disease and found 91.2% good to excellent outcome in the single-level group and 86.4% good to excellent results in the multilevel group. The only complications in this series of 57 consecutive patients were three iliac crest donor site wound infections.

Laminar Wires and Hooks

Traditionally, posterior thoracolumbar instrumentation has been divided into rigid and nonrigid implants. The earliest constructs incorporated the spinous processes or other posterior elements alone or with slabs of structural iliac crest autograft. Although simple wiring techniques are no longer used in the thoracolumbar spine, many wire-rod techniques continue to be routinely employed. The most common technique—Luque wiring—employs sublaminar wires as anchors. These wires are wrapped around rods to form a segmental, nonrigid spine construct. Such techniques are nonrigid because they allow “pistoning” of the spine in a craniocaudal direction. Luque’s construct was the first segmental system that used multiple wire attachment points.^53,57 In North America today, Luque constructs are preferentially used in the thoracic spine. Common indications include neuromuscular scoliosis, scoliosis with thoracic lordosis, and occasional cases of idiopathic scoliosis. Hybrid constructs using Luque wires with pedicle screws and other bone anchors are more commonly used. Contraindications to sublaminar fixation include absence of laminae. In addition, there is a danger of passing the wire through the spinal canal, particularly in patients with kyphosis or canal stenosis.⁵⁸ Because wiring techniques do not provide axial stability, they are a poor choice for stabilization of pathologic processes with anterior column insufficiency. For example, sublaminar wires should not be used as primary fixation in patients with vertebral fractures or tumors.

In response to the risk of sublaminar wire passage, Drummond proposed a technique, commonly referred to as the Wisconsin method, in which the rod is wired to the spinous processes.⁵⁹ This method has been used with Harrington rods and other rod systems to provide additional stability by segmental fixation.^60,61 A button may be used to decrease wire cut-through of the spinous process. Although this technique is rarely used alone anymore, it may be useful in hybrid constructs, particularly when limitations of bony anatomy or breakout of prior instrumentation renders the pedicles and other posterior elements unusable.

The next evolutionary step after wiring in thoracolumbar instrumentation came with simple, hook-based distraction devices intended to assist in the correction of scoliosis. After Harrington’s initial ground-breaking design, several other modified designs were developed. Harrington rods were used in thoracic curves and for thoracolumbar burst fractures. In trauma cases, the Harrington system provided quick and easy stabilization compared with the earlier devices. This construct could be combined with a rod sleeve to provide a three-point fixation and was particularly advantageous in polytrauma patients. Because Harrington rods produced distraction, they often led to flatback deformities that resulted in persistent pain and fatigue. When the importance of sagittal plane balance was understood and as newer, more powerful systems became available, the Harrington rod fell out of use.

Early hook-based segmental thoracolumbar spine constructs allowed lordosis to be contoured into the rod and held with segmental bone anchorage. These constructs were significantly more rigid than their predecessors. This increased rigidity improved fusion rates and sometimes obviated the need for postoperative bracing or cast immobilization. This was particularly advantageous for children and adolescents undergoing surgical correction of idiopathic scoliosis.

Disadvantages of this construct, as with rigid implants elsewhere, included the effect of fusion on adjacent segments. In contrast to pedicle screw constructs, which employ the concept of three-column spine fixation, hooks anchor to the posterior elements alone and do not have the same power to reduce scoliosis.⁶² Although misplaced pedicle screws can produce catastrophic neurologic or vascular injury, hooks by their very nature are canal intrusive. Mildly misplaced screws produce less canal encroachment than hooks. Polly and colleagues⁶³ have published an analysis of the volumetric effects of hook placement on the spinal canal.

In a comparison of pullout strength, thoracic pedicle screws were found to be significantly stronger than hooks and were recommended for rigid curves.⁶⁴ In terms of operative time, one study showed no difference between the operative time or correction achieved with hook or screw consults.⁶⁵ Finally, the pullout strength of hooks seems less sensitive to osteoporosis than screws, although screws are still stronger than hooks.⁶⁶

Various hooks with different characteristics are available. Each of these is suited to a particular mode of placement and loading. Pedicle hooks, resting on the lamina of the instrumented vertebra and the superior articular process of the next distal vertebra, are the strongest. They are always directed cephalad so that their U-shaped tip captures the pedicle and provides maximal stability for rotation and translation maneuvers. Because of their configuration and their dependence on facet joint anatomy, these implants can be placed from T1 to T10 or T11.⁶⁷ There are subtle variations in hook shape, including short-throated and long-throated configurations and straight or off-set collars. Contemporary “anatomic” hook designs give better bony contact.

Laminar hooks are available in various designs. Variation in blade width and style and the relationship of the blade to the body of the hook allow for optimal hook-bone interface. Between T3-T10 or T11, lamina hooks may also be placed on the superior surface of the transverse process. Here the “transverse process” hook is combined with a pedicle hook to yield a pedicle-transverse grip or “claw.” Although such claws may be constructed over one or two levels, two-level claws are easier to insert and mechanically stronger.⁶⁸ Claws are used mainly at the ends of a construct. Above T3, the transverse process becomes more horizontal, and transverse process hooks do not align with the pedicle hook. The claws must be pedicle–supralaminar hook combinations.

A supralaminar hook enters the spinal canal in a cranial to caudal direction and, while providing the second strongest anchorage, risks iatrogenic cord injury if the hook pistons in and out of the canal during rod manipulation. Thus, careful selection of the hook shape is mandatory. Moreover, one should consider avoiding peri-apical placement or utilization of supralaminar hooks in correction of hyperkyphosis. Finally, infralaminar hooks are placed in a cephalad direction. These anchors are rarely necessary in the thoracic spine because pedicle hooks provide proximally directed forces. Usually, these implants are used inferiorly, along with pedicle screws in hybrid constructs. Occasionally, they are added to transpedicular constructs to protect the screws from pullout.⁶⁹

Pedicle Screws

Pedicle screws are the only three-column fixation devices available at the present time. Because they are inserted into the vertebral body, they allow the surgeon to apply distraction, compression, lordosis, rotation, or translatory forces selectively.⁷⁰ Pedicle screw systems have been studied in Europe since the early 1980s. Modern attempts at pedicle fixation were popularized in North America by Steffee. Their use temporarily declined after a series of television exposés and high-profile lawsuits. Subsequently, many of the available systems received FDA class II status. This approval stopped many of the lawsuits and led to a rapid escalation in the number of systems available and the extent of their use.⁷¹ The generally agreed-on advantages of transpedicular internal fixation of the spine include the following:

Disadvantages and complications of pedicular screw fixation include the following:

Adjacent-segment degeneration issues are of great interest in the recent literature. Park and colleagues⁷² found that pedicle screw placement, more than the actual number of levels fused, was the strongest predictor of adjacent-segment degeneration. This review reported a 12.2% to 18.5% incidence of symptomatic adjacent-segment degeneration during the follow-up period compared with a 5.2% to 5.6% rate in patients fused with other forms of instrumentation or no instrumentation. Damage to the adjacent unfused facet by the screw was thought to be a potentially significant factor producing adjacent-level degeneration. Unintended adjacent-level fusion may also be encountered more frequently in patients undergoing transpedicular instrumentation. One prospective, randomized trial of 130 patients undergoing either instrumented or uninstrumented fusion found that inadvertent fusion had occurred in 19 cases (14%). There was a higher risk of unintended fusion with pedicle screw instrumentation, although functional outcomes were not affected.⁷³

The role of stress shielding in patient outcomes is controversial. The concept was introduced by McAfee and colleagues in 1991.⁷⁴ Subsequent studies have failed to show osteoporotic collapse within a fusion construct. Transpedicular screw fixation does not affect the postoperative reduction of fusion mass.^75–77 The indications for and contraindications to pedicle screw placement remain controversial at the present time. Pedicle screw constructs are widely indicated in cases with clearly defined instability, particularly in the setting of trauma and tumor. In particular, they provide stability where prior laminectomy has been performed.

Many different pedicle screw systems are currently available in North America. These systems may be classified as either degenerative systems or deformity systems. A few systems consider themselves as “universal” either by compromising their design to favor some characteristics of each or by providing modularity that allows for insertion of implants suited for either degenerative or deformity indications. Some systems employ stiffer rods for deformity and trauma application and a less stiff rod for degenerative lumbar conditions. The former maximizes reduction power, whereas the latter facilitates application of graft material and promotes ease of insertion.⁷⁸

Pedicle screw systems may also be divided by the type of longitudinal member or the connection between the anchor and the longitudinal member. Plating systems are one such system and are generally less expensive than modular, polyaxial screw and rod systems. Typically, plate-screw systems require that the screws function as bolts that are rigidly affixed to the plates with nuts and washers. Contemporary polyaxial systems have a U-shaped saddle in the head of the screw into which the rod can be secured. Polyaxial systems fare better under static and cyclic compressive loads. Because higher rod manipulation forces are possible, these connections serve to increase stresses at the end levels of the construct.⁷⁹ Although theoretically their increased freedom of motion should make polyaxial screw constructs less stiff, this has not been borne out in cadaveric testing.⁸⁰

Clinical failure of transpedicular constructs occurs in one of two ways: through loosening with fixation failure or through failure to fuse. Loosening occurs as repetitive loading persists beyond the tolerance of bone. This is usually the result of delayed union or motion at a screw-bone interface, often from excessive activity. The screw is exposed to a combination of cantilever bending and axial pullout loads (Fig. 71–5). Subacute or acute bending failure and implant breakage results from cantilever bending loads in excess of the yield point of the screws resulting in acute failure and breakage. Even severely degenerated and collapsed discs undergo cyclic axial displacement during axial loading. These cyclic displacements may lead to significant cantilever loads and bending moments. These are most pronounced around the screw hub, inside the pedicle.

FIGURE 71–5 When posterior thoracolumbar instrumentation fails, it is usually at the screw-bone interface. A, If a fusion fails to occur, the screws may loosen. Various techniques have been reported to improve screw pullout strength to increase the lift of the construct. B, One technique employs the endplate of the level above for additional purchase. C, In patients with poor bone quality, polymethyl methacrylate may be used in the vertebral body to improve screw pullout strength. Another concern is the appropriate number of fixation points. The more unstable the spine or the poorer the quality of the patient’s bone, the more fixation points are required for a successful construct. D, Failure of a long scoliosis construct presumably because too much force was concentrated at the base of the construct where all the inferior anchors were crowded together.

This effect is multiplied when the screw is forced to bear most or all of the anterior column axial loads, as with burst fractures. If a deformity reduction maneuver is performed during surgery, this subjects the screws to increased load. Typically, the end vertebrae are affected. Various techniques for correction, such as derotation maneuvers, have been described. A strategy that emphasizes correction in the middle segments of the construct with decreased corrective force at the terminal levels is associated with good reduction, while limiting axial tensile forces at the cranial end screw.⁷⁹ After bony healing, fatigue failure of the implants may still theoretically occur but should no longer be of clinical concern. In less unstable situations, such as a minimal degenerative spondylolisthesis, unilateral transpedicular fixation was found to be as good as bilateral fixation.⁸¹

Faraj and Webb⁸² reviewed complications related to transpedicular instrumentation in 648 consecutively inserted screws. Instrumentation was performed for various diagnoses, including scoliosis (34 patients), degenerative lower lumbar spinal disease (25 patients), and lumbosacral spondylolisthesis (3 patients). Intraoperative complications included three cases of screw misplacement, one case of nerve root impingement, two cases of cerebrospinal fluid leak, and two pedicle fractures. Postoperatively, deep wound infections were encountered in four patients, and two loose screws and one rod-screw disconnection occurred. The authors concluded that pedicle screw fixation has an acceptable complication rate and that neurologic injury during this procedure is unlikely.

Another series sought to quantify retrospectively pedicle screw–related complications in 105 consecutive operations. Overall, complications of varying severity were noted in 54%. The rate of deep infections was 4.7%, and all were successfully cured by débridement and antibiotics. Although 6.5% of cases had screw misplacement, there were no permanent neurologic complications. Screw breakage occurred in 12.4% of the patients, inevitably leading to loss of correction. Breakage was especially likely in cases of L5-S1 spondylolisthesis where reduction was performed without anterior support.⁸³

Spondylolisthesis may be stabilized with or without an attempt at slip reduction. Reduction maneuvers include rod contouring with an additional cranial fixation point with screws with an extended head into which the rod can be forcibly delivered, translating the slipped segment posteriorly.^84,85 Some screw-rod systems have been designed solely for slip reduction.⁸⁶ With reduction of an L5-S1 spondylolisthesis, there is increased risk of a postoperative L5 root palsy.⁸⁷ Pedicle screw constructs are also routinely used to increase the likelihood of fusion in the surgical management of junctional stenosis, degenerative spondylolisthesis or scoliosis, prior pseudarthrosis, and more than a three-level fusion (Fig. 71–6).

FIGURE 71–6 A-C, A 65-year-old woman with L5-S1 isthmic spondylolisthesis. She underwent anterior L5-S1 anterior lumbar interbody fusion with instrumentation followed by posterior instrumentation and fusion.

Typically, pedicle screw fixation is avoided in the presence of overt infection, but it may be considered in cases of significant instability. Relative contraindications to pedicle screw fixation include very small pedicles, marked osteoporosis, and inadequate anterior column support. In corpectomy models, 100% cyclic fatigue failure has been reported if anterior reconstruction and grafting has not been performed.^88,89 With partial anterior column load bearing, the posterior screw construct attempts much of the load bearing through cantilever forces.

Ultimately, the surgeon is left with the decision of when to perform pedicle screw fixation. Conflicting clinical outcomes have been reported, with good clinical outcome not correlating well with solid fusion. Although instrumentation enhances the rate of spinal fusion in degenerative conditions of the lumbar spine, such fusion does not guarantee a clinically successful outcome. Conversely, excellent results can be obtained in the setting of radiographic pseudarthrosis.⁹⁰ One prospective randomized controlled trial of 130 patients evaluated supplementary pedicle screw fixation in posterolateral lumbar spinal fusion and found that fusion rates were not significantly different between instrumented and uninstrumented groups. Although functional outcome, as assessed by the Dallas Pain Questionnaire, improved significantly in uninstrumented and instrumented groups, there were no significant differences in outcome between the two groups. A trend toward higher patient satisfaction was noted in the instrumented group (82%) versus the uninstrumented group (74%). The addition of pedicle screws significantly increased operative time, blood loss, and early reoperation rate. The two infections in that series occurred in the implant group, and significant symptoms from screw misplacement were seen in 4.8% of the instrumented patients. The authors concluded that these results did not justify the general use of pedicle screw fixation as a routine adjunct to posterolateral lumbar fusion.⁹¹

A similar prospective randomized controlled trial with 5-year follow-up was performed in 129 patients with severe chronic low back pain from either degenerative instability or isthmic spondylolisthesis. In that series, the reoperation rate was significantly higher in the instrumented group (25%) than the uninstrumented group (14%). There was no difference in work capacity between the two groups and no significant difference between the instrumented and uninstrumented groups in regard to functional outcome, as measured by the Dallas Pain Questionnaire and Low Back Pain Rating Scale. When the subgroups of isthmic and degenerative spondylolisthesis were analyzed separately, patients with isthmic spondylolisthesis had a significantly better outcome after posterolateral fusion without supplemental instrumentation compared with instrumented fusion (P < .03). Patients with primary degenerative spondylolisthesis had more significant improvement with instrumented posterolateral fusion (P < .02).⁷

Another prospective randomized study investigated the role of transpedicular fixation with posterolateral fusion in patients with adult isthmic spondylolisthesis; 37 patients underwent fusion with pedicular fixation, and 40 had uninstrumented fusion. At 2-year follow-up, the level of pain and functional disability were similar in the two groups, and there was no significant difference in fusion rate.⁹² In a separate retrospective review, 57 patients underwent posterior decompression and fusion for L4-5 degenerative spondylolisthesis, with half having transpedicular screw instrumentation. The clinical results and fusion rate were similar in the two groups. The authors concluded that the routine addition of pedicular fixation in these patients was unnecessary.⁹³

Percutaneous transpedicular constructs have been introduced more recently. These may be used to stabilize an ALIF or on the contralateral side of a minimally invasive transforaminal lumbar interbody fusion. Some systems employ a rod-insertion device that links to the screw extension sleeves and allows a precut contoured rod to be placed through the screw saddles via a small stab wound and a muscle-splitting technique. A remote engagement of the screw-locking mechanism is then employed. Long-term outcomes for these techniques are sparse, but short-term success has been reported in small numbers of patients. Proponents of this technique believe that paraspinous tissue trauma is minimized without compromising the quality of spinal fixation.⁹⁴

The role of routine use of pedicle screw instrumentation remains unclear. There are no convincing data to show that pedicle screw fixation is necessary. If in doubt, instrumentation can be safely avoided, especially when optimal purchase and placement are in doubt.

Technical Aspects

In patients with complex deformity, correct rod bending may be quite difficult. The rod should not be contoured in more than one plane at a time. In children and patients with flexible deformities, the rod can be contoured to fit the screws and rotated into proper sagittal balance. Alternatively, a series of force vectors and direct rotational maneuvers may be undertaken. In patients with rigid curves and most adults, accurate rod contouring to fit into the screws can be difficult. It is often helpful to use instead lateral connectors between the rod and screw. Some surgeons, in an effort to maximize correction power, use separate rods in the upper and lower portions of a stiff, dual curve deformity and connect them with a domino on one side and use a neutralization rod on the opposite side.⁶²

When bending rods for hook-based constructs, slight deviation from perfect contour puts a moment on the hook. In the thoracic spine, a kyphotic moment pulls the hook away from the cord. Generally, the aim is for a 30-degree thoracic kyphosis with its apex at T7 or T8. The lumbar lordosis should be 45 degrees with an apex at the L2-3 interspace. The rod should change from kyphosis to lordosis at the T12-L1 interspace. After placement of the hooks, the convex rod is placed first. The more mobile the hooks, the more easily the rod engages. Hook mobility is a function of the amount of bone resection during hook placement. The ease of rod insertion improves with spine mobility, which is a function of facet resection. Torsional stability of anterior and posterior dual rod constructs improves with the use of cross-links between the rods.⁹⁵ Depending on the number of motion segments instrumented, the degree of instability of the spine, and the strength of the bone-implant interface, placement of one to three cross-links has been recommended.^96,97 If two cross-links are used, one should be as proximal as possible and the other as distal as possible. If the instrumentation exceeds 30 cm in length, a third transverse connector should be considered in the middle. In some cases, there is little space for a cross-link between the bulky heads of polyaxial screws. In such instances, the rods could be extended caudally or cranially and a cross-link placed along the extension. Care should be taken to avoid injury to the proximal facet joint by the connector. The addition of a cross-link may mechanically compensate for a missing pedicle screw in a polysegmental construct.⁹⁶

One cadaveric study of transpedicular fixation across the thoracolumbar junction found that rotational and bending stiffness increased significantly with the number of cross-links placed.⁹⁸ Another study compared constructs using no, one, or two cross-links.⁹⁵ This study found that two cross-links were no more effective than one in significantly increasing axial, flexion, and lateral stiffness, although additional cross-links significantly improved the torsional stiffness of the construct. When comparing different types of cross-links, larger cross-sectional area was associated with greater increases in stiffness.

There is some concern that the bulk of some cross-link connectors may reduce the area available for bone grafting and may predispose to pseudarthrosis and construct failure. In addition, too much stiffness may inhibit fusion through stress shielding. Cross-links generally are not indicated in degenerative conditions.

Specifics of Anterior Lumbar Interbody Fusion

In the past 20 years, spine surgeons have become increasingly comfortable with anterior, retroperitoneal approaches to the low lumbar spine. The least controversial indications for anterior cages include reconstruction of tumors or vertebral body fractures. Thoracolumbar cages or strut allografts (see Figs. 71-7 and 71-8) are often employed in the reconstruction of the following:

In addition to numerous interbody cages, an expanding number of allografts have become available (Fig. 71–9). These may be “off-the-shelf” sections of femur, tibia, humerus, or fibula, which the surgeon cuts into the desired size and shape or one of myriad machined “bone products.” Commercial bone implants are typically prefabricated to specific sizes and may have special surface textures to resist extrusion.

FIGURE 71–9 When an anterior corpectomy is performed, various materials are available to reconstruct the defect. A, Example of vertical mesh cage. These devices offer end-caps that engage the host bone, improving rotational stability of the construct. Various allograft materials may be employed as well. B, Use of a single fibular shaft. Graft geometry should match as much as possible that of the host endplate. Particularly in patients with osteoporosis, small corpectomy struts penetrate the endplate, and all axial stability is lost. Disengagement of the hooks posteriorly is noted here. C, Postoperative axial CT myelogram in a case in which a tibial shaft has been used. The graft fills most of the corpectomy. As with all other forms of instrumentation, careful attention to positioning of the implant is crucial with corpectomy devices. D, Grossly malpositioned cage in a patient with postoperative radicular symptoms.

(A, Courtesy Stryker Spine, Mahway, NJ.)

The rationale behind machined allograft and manufactured cages includes the following:¹¹⁹

Although allograft bone spacers may be used alone, they, similar to their metal counterparts, are typically filled with bone graft or bone morphogenetic protein. Interbody spacers come in various shapes and sizes. Cage materials include metal, bone, carbon fiber, PMMA, or PEEK. Numerous absorbable polymers, such as polylactic acid, are also undergoing testing.¹²² Each interbody device has different characteristics, such as modulus of elasticity.

One of the disadvantages of metal cage implantation lies in plain radiographic assessment of fusion. Assessment may be improved when carbon fiber, PEEK, or similar biomaterials are used.¹²³ Even with radiolucent cages, differentiation of bone adherence versus through-growth (true fusion) is difficult, however.

Some differences in bone healing relate to the stiffness of the cage. Stiffer cages stress-shield the bone graft within. The modulus of elasticity with PEEK is closer to host bone and may load the graft more completely.¹²⁴ With resorbable cages, the graft is gradually loaded to a greater and greater degree as the cage is enzymatically digested. One study found that this property was associated with improved fusion rates compared with titanium cages.¹²²

Threaded cages are available in trapezoidal designs for anterior application. This shape is thought to improve segmental lordosis. Other designs have been modified to improve stability. Some newer cage designs include screw holes that allow immediate screw fixation into the vertebral bodies above and below. The addition of screw holes is believed to provide immediate stability and compensate for the extension instability of stand-alone cage constructs.

For posterior interbody cage insertion, various contoured and banana-shaped cages have been designed to facilitate insertion and match the native endplate contour. Anterior cages are broadly divided by purpose into interbody and corpectomy devices. There are two main cage varieties: vertical and horizontal. Vertical devices typically fill corpectomy defects, whereas horizontal (cylindric) cages are used in discectomy procedures. For the latter, the mechanical goal is interspace distraction and restoration of annular tension.^125,126

The metallic tines of vertical cages and the threads or texturing of horizontal cages allow them to resist torsion and displacement better than smooth bone grafts. Horizontal cage designs can be subdivided further by shape into screw-in, box, or mesh. Screw-in cages are exemplified by threaded cylinders such as the Bagby and Kuslich (BAK) device. Box cages are rectangular with flat and textured bearing surfaces to improve axial stiffness and resist extrusion. Numerous manufacturers market machined allograft with surface grooves for a similar effect. The relative merits of each of these designs continue to be debated.¹²⁷

Each of the basic types of interbody devices has modifications designed for particular purposes. For tumor reconstruction, vertical cage modifications include horizontal threads that allow the cage to be vertically expanded in situ.

By providing disc space distraction, interbody devices and procedures restore foraminal height. Interbody techniques can also provide indirect reduction of central canal compression. One cadaveric CT study found that anteriorly or laterally placed interbody devices can reduce anterior listhesis and increase canal and foraminal volume in a degenerative spondylolisthesis model.¹²⁸ One clinical study followed a series of 56 patients with back pain, neuroclaudication, or both from degenerative spondylolisthesis and spinal stenosis who underwent ALIF for reduction and fusion. Outcomes were comparable to the published outcomes of in situ fusion after formal laminectomy, avoiding the risk of epidural fibrosis and “fusion disease” associated with posterior decompression and fusion.¹²⁹

All strut grafts, disc replacements, cages, and other anterior spacers are subject to subsidence and extrusion.^130,131 The rate and degree of this subsidence is related to cage geometry and sizing, endplate coverage, and preparation.^132,133 Several studies have shown excellent early interspace distraction but gradual cage with further follow-up. One clinical study of dual rectangular cages found that 76% of patients developed subsidence, more often into the superior than the inferior endplate, although it did not appear to affect fusion rates or clinical outcomes.¹³⁴ The authors describe this subsidence as typically occurring by 4 months postoperatively. In that study, mean preoperative intervertebral disc height was 11.6 ± 3.1 mm, immediate postoperative height was 16.9 ± 2.0 mm, and final follow-up disc height was 13.2 ± 2.4 mm.

Cylindric implants have significantly more subsidence than plate and graft constructs, rectangular cages, or cages with lateral “wings” for increased axial stability.¹³⁵ Lateral positioning on the endplate is associated with decreased rates of subsidence. Ultimately, cage size and placement in the disc space are more important than implant design.¹³⁶ For similar reasons, subsidence is a major concern for disc replacement procedures. In contrast to fusion procedures, there is no point at which the construct can be said to be healed in position. Implant sizing is crucial. Because an overly large implant is virtually impossible to insert into a degenerated and collapsed disc space, an undersized implant is more commonly inserted owing to difficulty in adequately distracting the collapsed disc space. Because small implants cover very little of the vertebral endplate and offer virtually no end bearing, central positioning of the device risks endplate subsidence.¹³⁷ Finally, if the patient subsequently develops osteoporosis, endplate support itself may decrease.

The transthoracic approach is a modification of the standard approach to aortic aneurysms and offers excellent exposure to the anterior and lateral aspects of the thoracic and upper lumbar spine. Although double-lumen intubation and postoperative chest tube placement are often required, this approach provides a safe avenue for decompression of bony stenosis of the thoracic canal. Approaches to the thoracolumbar junction often require partial diaphragmatic takedown. For patients with marked anterior column disruption and axial instability, the anterior approach allows the most stable reconstruction.

The lateral approach used in the thoracolumbar approach limits subsequent instrumentation placement options. A laterally placed plate is less able to prevent segmental spine extension than a plate on the anterior vertebral surface When approaching the lower lumbar spine anteriorly, the retroperitoneal approach is generally preferred (Fig. 71–10). Transperitoneal approaches enjoyed a short burst of popularity in the mid-1990s when endoscopic techniques were used for threaded cage placement. Although this attempt at less invasive fusion surgery has since given way to mini-open retroperitoneal approaches, it signaled a turning point in the way many surgeons viewed the relative morbidity of anterior and posterior surgical approaches.²⁸

FIGURE 71–10 Thoracolumbar corpectomy operations may be performed with or without anterior or posterior instrumentation. When possible, additional posterior procedures should be avoided. In many burst fracture models, anterior column instrumentation may be biomechanically adequate. There are many options, but the most common are plating and rodding systems. A and B, Plating systems are said to be easier to implant and have a lower profile. C and D, Rod-based systems confer greater mechanical strength in that the longitudinal member (in this case the rods) lies farther from the spine’s instant axis of rotation. This advantage is also a disadvantage, however, in that it increases implant bulk and profile. E, Most rigid anterior instrumentation systems allow distraction of the interspace for safe canal decompression and subsequent compression to improve immediate construct stability. F, For virtually all of these systems, triangulated, bicortical placement of the screws improves holding power in osteoporotic bone.

(E and F, Courtesy of Stryker Spine, Mahway, NJ.)

At the present time, the transperitoneal approach is occasionally used in markedly obese patients and when prior retroperitoneal exposure has been performed. In approaching the sacral promontory, some surgeons recommend infiltration of the tissue with a few milliliters of saline to facilitate the dissection and aid in identification of the presacral parasympathetic fibers. More typically, the anterior aspect of the lumbar spine is approached retroperitoneally. This approach improves access to the upper lumbar spine. For limited exposures to the low lumbar spine, a short transverse or paramedian incision is made. Alternatively, a Pfannenstiel incision may be used. For multilevel exposures, an oblique flank incision is made. In this setting, the patient may be positioned in a semilateral position using blanket or gel rolls.

Minimal Access Surgery

Numerous mini-open modifications of the standard retroperitoneal exposure have been described. These modifications rely chiefly on the experience of the access surgeon and powerful retractor systems to minimize the incision length. In one retrospective 2-year follow-up, 28 patients underwent ALIF via a 6-cm to 10-cm left lower quadrant transverse skin incision.²⁹ A paramedian anterior rectus fascial Z-plasty allowed access to the retroperitoneal space for placement of various implants. No vascular, visceral, or urinary tract injuries occurred, but a mild ileus was noted in three cases. Another report described similar use of a 5-cm left flank incision in 25 patients. No injury to the great vessels or neurologic deterioration was noted.³⁰ This approach is safe, uses a small skin incision, avoids cutting the abdominal wall musculature, allows various interbody fusion techniques, and does not require peritoneal violation or endoscopic instrumentation. Other advantages include fewer x-rays with reduced radiation exposure during surgery and a shortened learning curve because the approach is similar to the anterior open lumbar technique.

Kaiser and colleagues³¹ performed a retrospective comparison of mini-open and endoscopic ALIF in approximately 100 patients. They found that operative times were longer and the risk of retrograde ejaculation was higher with the laparoscopic approach than with the mini-open approach. Length of stay was increased, however, and the immediate postoperative complication rate was greater after mini-open ALIF procedures.

An endoscopically assisted retroperitoneal approach has been termed the balloon-assisted endoscopic retroperitoneal gasless (BERG) approach. In one study of 46 individuals, various devices, including cylindric cages, femoral ring allografts, and vertical cages, were placed. Complications included a left common iliac vein injury not requiring operative repair and a far-lateral cage placement. The average hospital stay was 3 days. An advantage of the BERG approach over traditional endoscopic access is its ability to use standard orthopaedic instruments and implants.³²

Minimally disruptive approaches to the anterior lumbar spine continue to evolve in a quest to reduce approach-related morbidity. One innovation is a lateral retroperitoneal, transpsoas approach to the anterior disc space that allows sufficient access for a complete discectomy, distraction, and interbody fusion without the need for an approach surgeon. Two companies have an FDA-approved device: The XLIF is from Nuvasive (San Diego, CA), and DLIF is from Medtronic (Minneapolis, MN). Advocates of minimal access spinal approaches cite certain advantages over open procedures, including decreased postoperative pain and narcotic requirements, shorter hospital stay, less blood loss, and smaller incisions. The minimally invasive anterolateral approach allows access to the lumbar spine through the retroperitoneal space (Fig. 71–11).

FIGURE 71–11 XLIF device minimal access spinal approaches have certain advantages over open procedures, including decreased postoperative pain and narcotic requirements, shorter hospital stay, less blood loss, and smaller incisions. This patient underwent L2-3, L3-4, and L4-5 XLIF instrumentation followed by posterior stabilization. A, Preoperative MRI showing multilevel stenosis and disc degeneration. B, Preoperative radiographs showing multilevel disc degeneration and lumbar scoliosis. C, Postoperative radiographs.

Complications associated with minimal access surgery include nerve injuries. Although rare, these can be devastating complications, and they are likely to increase as an increasing number of spine surgeons use minimal access retroperitoneal surgery to treat lumbar problems. An anatomic study of the lumbosacral plexus and its relationship to the transpsoas approach of the lumbar spine found that the lumbosacral plexus lay within the substance of the psoas muscle between the junction of the transverse process and vertebral body and exited along the medial edge of the psoas distally. The lumbosacral plexus was most dorsally positioned at the posterior endplate of L1-2. There was a general trend of progressive ventral migration of the plexus on the disc space from L2-3 through L4-5. Average ratios were calculated at each level (distance of the plexus from the dorsal endplate divided by total disc length) and were 0 (L1-2), 0.11 (L2-3), 0.18 (L3-4), and 0.28 (L4-5). This anatomic study suggested that positioning the dilator or retractor, or both, too posteriorly at the disc space may result in nerve injury to the lumbosacral plexus, especially at the L4-5 level. There is also a risk of injuring nerve branches to the psoas muscle and injuring the genitofemoral nerve. Although there have been no published vascular complications, at the time of this writing, it is critical to be aware of the vascular anatomy. A more recent study showed that the safe corridor for performing the DLIF and XLIF procedures becomes more narrow going from L1-2 to L4-5.

Motion-Preserving Implants

Interspinous Spacers

As with the posterior dynamic rod systems, interspinous process distraction devices have been introduced for various indications. Most share a minimally invasive concept in that they can be inserted under local anesthesia with small incisions. Some devices require formal, open exposure of the posterior elements to attach or wrap “artificial ligaments” or other outriggers.

In North America, the most familiar of the less invasive devices is the X-STOP (Medtronic, Minneapolis, MN). As a less invasive alternative to laminectomy, the X-STOP is used to distract the spinous processes in patients with symptomatic spinal stenosis. Subtle segmental flexion or prevention of extension is said to increase canal diameter and relieve neural compression.

In June 2005, a randomized, controlled, prospective multicenter trial of 191 patients with neurogenic claudication compared outcomes of patients treated with X-STOP with outcomes of patients receiving nonoperative care.¹⁴⁷ At 2 years, the symptom severity score of patients treated with X-STOP improved by 45.4% over mean baseline. The control group had only 7.4% improvement. Physical function improved 44.3% in the X-STOP group but declined in the controls. The authors concluded that the X-STOP provided effective treatment for patients with spinal stenosis (Fig. 71–12).

FIGURE 71–12 As a less invasive alternative to laminectomy, an interspinous spacer is used to distract the spinous processes in patients with symptomatic spinal stenosis. Subtle segmental flexion or prevention of extension is said to increase canal diameter and relieve neural compression. A, Preoperative MRI showing multilevel stenosis. B, Preoperative radiographs. C, Postoperative radiographs showing placement of interspinous spacers placed at L4-5 and L3-4.

At least four other interspinous spacer devices are receiving increasing attention at new technology meetings. Most of these devices are intended to treat mechanical low back pain through either distraction or restriction of motion. One example is the Wallis (Zimmer, Warsaw, IN) implant, in which a semielastic blocker seeks to dampen forces passing through the posterior elements while decreasing the neutral zone.¹⁴⁸ The first-generation implant, with a titanium blocker and an “artificial ligament” made of Dacron, was developed in 1986. More than 300 patients were implanted during 1988-1993. Subsequently, the Wallis implant, made of PEEK, was developed. Without permanent fixation into the vertebral bone, the “floating” design of the Wallis and similar implants seeks to avoid the risk of loosening encountered with pedicle screw–based and facet replacement–based dynamic stabilization systems. According to a more recent review of the Wallis implant, significant improvements in back pain were seen “without serious complications.”¹⁴⁸ The Wallis system was recommended for (1) discectomy when massive fragment leads to substantial loss of disc material, (2) first disc herniation recurrence, (3) discectomy for herniation of a transitional disc with sacralization of L5, (4) degenerative disc disease at a level adjacent to a previous fusion, and (5) isolated endplate lesions leading to chronic low back pain.

Facet Replacement Systems

Although there is a North American Facet Replacement trial under way, there is very little information about the rationale and design or even case report material available in the literature. At least three systems are under investigation, but none seeks to treat posterior column mechanical back pain. Rather, these systems are meant to restore stability after laminectomy without necessitating spinal fusion. They fit into a narrow treatment window between clearly unstable spines (e.g., spines with scoliosis or a dynamic spondylolisthesis) that require concomitant fusion and cases in which a narrow decompression can be performed that would not jeopardize postoperative stability.

Creation of a functional facet replacement system is a complex undertaking. Four main components, the superior and inferior right and left facets, have to be sized to the individual. Different designs rely on varying forms of attachment, but generally fixation through the pedicle is required. Various outriggers, cross-links, or other devices hold the system together and balance tension to maintain stability, while maintaining segmental motion. How these devices will function in the long-term, or even in the short-term, is unknown.

Nucleus Replacement

The historical precursor to total disc replacement and nuclear replacement was the stainless steel Fernström ball. This spherical endoprosthesis was developed and implanted in the 1950s and 1960s as a spacer that allowed motion of the adjacent vertebrae.¹⁴⁹ The shape of the device concentrated the loads on a small portion of the implant and vertebral endplate surface. Normal load distribution could not be recovered; because of concerns about device migration and subsidence, the device was subsequently abandoned.¹⁵⁰

More recent efforts at nuclear replacement have sought less stiff materials, typically with viscoelastic behavior. These devices can be divided roughly into preformed implants and injected polymers that cure in situ. Minimally invasive delivery of injectable materials can be accomplished with small annular windows, decreasing the risk of implant extrusion. Preformed implants may have the advantage of improved mechanical strength and fatigue life, but several designs have been associated with high extrusion rates.^151,152

An intermediate approach relies on open implantation of a containment scaffold with subsequent injection or inflation with a hydrogel or other incompressible liquid. An ideal nuclear replacement restores disc mechanics with much less normal tissue ablation than total disc arthroplasty. Some of these devices may offer shock absorption not seen with current total disc arthroplasty designs.¹⁵³

Nuclear replacement devices may be indicated for primary treatment of early, painful lumbar disc degeneration. Alternatively, they may be placed at the time of discectomy in a patient with radicular complaints. In the latter case, the nuclear replacement seeks to maintain or restore disc height and function. In the short-term, this restoration of disc mechanics is thought to decrease postoperative back pain. Ideally, nuclear replacement would prevent further motion segment degeneration, decreasing the likelihood of more invasive and destructive surgeries such as fusion or disc arthroplasty.^153,154

Contraindications for a nucleus prosthesis include disc height less than 5 mm, annular incompetence (in the form of major tears), grade II or greater spondylolisthesis, and Schmorl nodes. Many of the proposed devices remain in preclinical testing for biocompatibility and fatigue life as of this writing.^155,156 The Raymedica PDN (Prosthetic Disc Nucleus, Minneapolis, MN) has undergone several small clinical trials with associated implant redesigns. The device consists of a hydrogel core encased in a polyethylene jacket. The hydrogel absorbs water and expands to fill the nuclear cavity. The jacket prevents the hydrogel from overexpansion. Initially, two devices were implanted at each operated level. After extrusion problems were encountered, a larger, single PDN was designed. Although the device was initially inserted solely through a large posterolateral annulotomy, the consequences of device extrusion into the canal have led to anterolateral insertions as well.¹⁵⁷

Numerous short-term nuclear replacement outcomes studies have been presented, and a subset of these have been published. In one series, 300 patients mainly with radiculopathy were implanted, and a dramatic improvement in ODI and VAS scores was reported for most. The extrusion rate has exceeded 6%, however, even in more recent reports. Occasionally, an intense endplate inflammatory reaction with sclerosis and increased pain occurs.¹⁵⁸ Klara and Ray¹⁵⁹ reviewed the PDN data and found that, since 1996, 10% of the 423 patients undergoing PDN placement have been explanted.

In one more recent article, the authors sought to evaluate the efficacy of PDN for chronic discogenic back pain caused by degenerative disc disease.¹⁶⁰ Of 48 patients undergoing nucleus replacement surgery from January 2001 through May 2002, 46 were followed for 6 months. The mean ODI score was 58.9% preoperatively and improved to 18% at the 1-year follow-up. VAS pain scores improved from a preoperative mean of 8.5 to 3.1 after 1 year. The mean Prolo Scale score also improved from 5.2 preoperatively to 7.2 at 1 year. There were four cases of device migration requiring revision surgery. According to the MacNab criteria, results were excellent in 5 patients (10.9%), good in 31 (67.4%), fair in 3 (6.5%), and poor in 7 (15.1%). The authors concluded that nucleus replacement with the PDN device seemed to be effective in treating patients with chronic discogenic back pain caused by degenerative disc disease. The high failure rate is disturbing, however.

Another emerging nuclear replacement is the spiral Newcleus (Centerpulse, Sulzer SpineTech, Minneapolis, MN). This polycarbonate urethane is shaped in a memory coiling spiral that can be inserted through a small annular window.¹⁶¹ Inside the disc space, it recoils to fill the nucleotomy defect and restore disc height. Theoretically, this design allows for an easily customized amount of the material to be implanted with little risk of extrusion. Korge and colleagues¹⁶² reported implantation into five humans with radicular pain caused by disc herniation. Two years after the implantation, there were no complications, the facet joints were completely functional, and there was clinical improvement indicated by improvement in VAS and ODI scores.

Ahrens and colleagues discussed more recently the DASCOR device that was developed to provide an alternative treatment with a less invasive surgical intervention. The authors looked at 85 patients from 11 European centers who were enrolled in one of two studies between February 2003 and July 31, 2007. Data were collected before surgery and after surgery at 6 weeks and at 3, 6, 12, and 24 months. The clinical outcome measures were obtained from the VAS score for back pain, the ODI, radiographic assessments, and records of analgesic medication use. Mean VAS and ODI scores improved significantly after 6 weeks and throughout the 2 years. Radiographic results showed, at a minimum, maintenance of disc height with no device expulsion and, despite Modic type 1 endplate changes, no subsidence. Serious adverse events were reported in 14 patients, including device explants in 7 patients (7 of 85), in which the main complication was resumed back pain after time. Patients’ rate of analgesic medication decreased dramatically over time, with all patients experiencing significant improvements after 3 months and nearly no analgesic medication or narcotic drug use at 2 years. The authors concluded that the interim outcomes showed significant improvements in mean ODI and VAS scores.

The results of these European studies suggest that the DASCOR device may be a safe and effective, less invasive surgical option for patients with symptomatic degenerative disc disease. Several other companies are working on nucleus replacements that are in varying stages of development and clinical studies.

Lumbar Disc Arthroplasty

After the Fernström ball, the evolution of total disc and nuclear replacement diverged. Lumbar disc replacement, similar to nuclear replacement, is one of an emerging group of motion-preserving technologies whose clinical goals include reduction of pain and avoidance of the morbidity associated with fusion. Lumbar fusion for painful lumbar degeneration has a long, but storied, history and remains controversial. Most importantly, outcomes after these procedures are not consistently satisfactory. Significant numbers of patients remain on narcotic pain medications and remain unable to return to previous levels of employment.¹⁶³

Disc replacement seeks to remove the pain generator while preserving motion. In avoiding fusion, bone graft donor site morbidity is avoided, as is the possibility of pseudarthrosis. Theoretical benefits include decreased rates of adjacent segment degeneration and improved sagittal balance.

In 1982, in what was then Communist East Germany, Büttner-Janz and Schellnack¹⁶⁴ initiated development of their artificial disc at the Charité Hospital in Berlin. They based their design on the “low-friction” principle used in total knee and total hip arthroplasty. In particular, they placed a polyethylene sliding core between two highly polished metal endplates. This sliding core was to mimic the movement of nucleus within its annular containment.

After preliminary mechanical testing, the SB CHARITÉ I disc was first implanted in 1984. In 1985, owing to axial migrations, the artificial disc was modified to SB CHARITÉ II, in which the metal endplates were enlarged with bilateral “wings” to improve support of the implant on the bony endplates of the vertebral bodies. Fractures in Modic II endplates and insufficient instrumentation for implantation led the designer to enlist major orthopaedic device manufacturer Helmut Link to assist in the development of the third generation. The currently available SB CHARITÉ III has basically remained unchanged since Link took over production in 1987. Since 1987, approximately 4000 SB CHARITÉ artificial discs have been implanted (Fig. 71–13).¹⁶⁵

FIGURE 71–13 Newest category of anterior implant is lumbar disc replacement. Numerous systems are being studied. All systems seek to decrease pain while maintaining motion, eliminating some of the morbidity of fusion such as adjacent-segment disease or graft harvest pain. A and B, Flexion-extension radiographs obtained 1 year postoperatively in a pain-free patient.

After the SB CHARITÉ, the next disc arthroplasty system was the Prodisc (Synthes Spine, Paoli, PA). The Prodisc is composed of two titanium endplates with a fixed polyethylene core between them. Motion occurs by articulation of only the upper plate on the convex superior surface of the constrained core. This device was designed and implanted into a group of patients after which the designer, Marnay, refrained from further implantations for 8 years. Subsequently, Marnay reported 8- to 10-year follow-up results on a group of 44 patients who received this prosthesis. He reported 78% good to excellent results.¹⁶⁶

Current indications for lumbar disc replacement are as follows:

Contraindications are more controversial. Although subtle retrolisthesis related to disc collapse is acceptable, disc replacement with greater degrees of translational instability or spinal deformity should be avoided. Disc arthroplasty should not be performed in patients with marked facet osteoarthritis.¹⁶⁷ The ideal arthroplasty patient presents with a single painful level with more than 4 mm of remaining disc height at that level. This ideal patient has intact posterior elements and no facet or adjacent-level degeneration.

In one study of 108 arthroplasty patients, 76% had no back pain at latest follow-up, and 60.9% were completely satisfied. There was a 9% complication rate. Most of the complications were transient, including postoperative L5 radicular pain.

Although disc arthroplasty seems promising, many important issues remain unresolved. First, design considerations include constrained versus semiconstrained implants and differing bearing surfaces, such as metal-on-metal versus metal-on-polyethylene.¹⁶⁸ Proponents of metal-on-metal designs cite the increased rate at which polyethylene wears. Polyethylene particles could lead to osteolytic reactions and loss of bone stock. Thin polyethylene is subject to creep and cracking. Metal-on-metal, although more durable, is approached suspiciously by some spine surgeons because of concerns about metal ions leaching into the bloodstream. Currently available data are inadequate to compare these risks.

Initial wear rate studies seem to suggest excellent prospects for long-term viability.^169,170 When particles are generated in significant numbers, however, they have been shown to migrate along the neuraxis to the brain.¹⁷¹ Serum metal ion levels in patients with long-standing hip prostheses were reported more recently. In this series of patients with implants in situ for more than 30 years, urine and serum cobalt, chromium, titanium, and vanadium were measured. In metal-on-polyethylene and metal-on-metal prosthetic groups, the levels of potentially carcinogenic ions were higher than in the control group. In patients with loose metal-on-metal articulations, the blood and urine cobalt levels were elevated to 50 and 300 times normal. Given that total disc arthroplasty is performed in a largely younger patient population, these metallic ion data should arouse concern and stimulate vigilance.¹⁷²

The issue of device constraint remains unresolved as well. Proponents of unconstrained devices state that these are most capable of restoring the normal, coupled motion of the lower lumbar spine.¹⁷³ Supporters of semiconstrained designs cite occasional polyethylene liner extrusions as evidence that more constraint is needed. The concept of “semiconstrained” is elusive: How much is enough; how much is too much? In particular, do these values hold over various levels of the lumbar spine? Current data suggest that the instantaneous axis of rotation is not constant, and changes depend on the joint position.^174,175

If disc replacement systems restore motion, critics worry that improper motion may be worse than no motion. Disc degeneration rarely occurs in isolation. More typically, at least some facet degeneration coexists. One concern is that maintenance of motion would precipitate further facet degeneration. Given that the facets guide segmental motion, further facet degeneration may disrupt normal kinematics further.¹⁷⁶ In a prospective, clinical study of 64 patients, low-grade facet arthrosis did not influence outcomes after implantation of a semiconstrained metal-on-metal total disc replacement. At 2-year follow-up, the preoperative ODI score of 43.8 had improved to 23.1. VAS back pain score had decreased from 7.6 to 3.2.

More constrained implants require more elaborate fixation methods to the host vertebral body.¹⁷³ Several of the semiconstrained implants in the first generation of devices include keels. Although it is unclear if the keel offers a net advantage to the patient in terms of long-term stability, it would complicate any attempt to remove the implant. Two deep keels at adjacent levels of disc arthroplasty may subject the vertebral body to sagittal split fracture. For all but the tallest degenerated discs, current lumbar disc arthroplasty designs require marked disc space distraction. Too much distraction seems to decrease motion at the adjacent segment. Too little distraction risks implant extravasation.

Finally, critics of the current total disc arthroplasty systems have cited the limited outcomes literature relative to the thousands of discs that have been implanted worldwide. They also point out the suboptimal outcomes reported in the published articles.¹⁷⁷ In particular, the actual amount of motion preserved may be very low in some cases. In a recent retrospective study, 38 patients undergoing one-level and two-level total disc replacement were divided into two groups by the amount of postoperative motion recorded on flexion-extension radiographs at 8.6-year follow-up. Although improvements in ODI and Stauffer-Coventry scores were termed “modest” overall, patients with greater than 5 degrees of motion reported outcomes superior to those with less motion.