CHAPTER 275 Evaluation, Indications, and Techniques of Revision Spine Surgery
Revision spine operations are increasingly being performed and pose a unique set of challenges to the neurosurgeon. Selection of patients for revision spine surgery is more difficult than selection for primary spine operations, and the likelihood of a good clinical outcome declines with each successive operation.1–4 In addition, reoperations are more technically demanding than primary operations because the normal anatomic structures and tissue planes have been disturbed.5 Scar tissue from previous operations further increases the difficulty of achieving adequate exposure and decompression, as well as the risk for postoperative leakage of cerebrospinal fluid and wound infection or dehiscence.5–7 Complication rates, in general, are higher for revision operations, and thus the decision to perform revision surgery must carefully account for both the altered risk-benefit ratios and the decreased chance of achieving a good clinical outcome with each successive procedure.1,3 Most importantly, the goals of surgery must be clearly defined, and the surgically correctable pathology must be accurately identified.
Initial Evaluation
Radiographic evaluation is made more difficult by previous spine operations. Plain anteroposterior (AP) and lateral radiographs provide the clinician with a great deal of anatomic information, as well as data on preexisting spinal instrumentation constructs. Dynamic radiographs such as flexion-extension films are very helpful for investigating the stability of the spinal column and for determining the integrity of instrumentation constructs and bone fusions. Standing 36-inch AP and lateral radiographs taken with both the hips and knees extended are useful in assessing the sagittal and coronal balance of the spinal column, especially in patients with spinal instability and deformity. Evaluation of bone anatomy and surgical instrumentation is best accomplished with computed tomography (CT) and plain radiography. The information provided by high-resolution, multiplanar CT is helpful when revising previous instrumentation constructs because it can serve as a surgical “road map.” This is particularly important during the revision of pedicle or lateral mass screws when planning new screw entry points and trajectories. High-resolution CT also aids in the assessment of spinal fusion, especially when combined with the information from dynamic radiographs; however, the “gold standard” for assessment of fusion remains open surgical exploration.8,9 Magnetic resonance imaging (MRI) is the most useful study for evaluating the neural elements, ligamentous structures, and intervertebral disks. MRI with gadolinium enhancement is often used for the differentiation of recurrent herniated disk material from epidural scar tissue. Frequently, in patients with spinal instrumentation, imaging artifacts significantly limit detailed MRI evaluation of closely adjacent structures. Myelography, often accompanied by postmyelographic CT, is helpful in determining the presence of thecal sac or nerve root compression in cases in which MRI of the neural elements is limited by artifact.
General Surgical Principles
Attention to the integrity of superficial tissues is essential to decrease the risk for postoperative wound complications such as wound infection and dehiscence. Spine reoperations generally take longer than primary operations, and this is directly correlated with the risk for wound infection.7 Several techniques, such as frequent wound irrigation and hourly release of tissue retractors, help ensure adequate blood flow to tissues and lessen the risk for surgical infections. If the wound edges appear hypovascular or otherwise nonviable, they may be excised to allow primary closure of fresh wound edges.
Recurrent Lumbar Disk Herniation
Recurrent lumbar disk herniation is generally defined as ipsilateral or contralateral herniated disk material at the same level as the index operation that causes symptoms of radiculopathy after a postoperative symptom-free interval of at least 6 months.1,10 Reherniation of a lumbar disk at the same level and site of a previous diskectomy causing recurrent radicular symptoms occurs in 3% to 19%, with higher rates being quoted in studies with longer follow-up.11–16 This phenomenon is more common at the L4-5 level, perhaps because of its increased mobility relative to other spinal motion segments.17 Patients typically experience recurrent symptoms of radiculopathy after a postoperative symptom-free period of at least 6 months. Recurrent radiculopathy at the same level may also occur secondary to degenerative foraminal stenosis, a prolapsed annulus, or excessive epidural scar tissue formation.
Recurrent herniation may pose a diagnostic challenge because imaging studies will demonstrate significant postoperative changes in the epidural space at the operated level, such as epidural scar tissue, adhesions, and bony hypertrophy. Gadolinium-enhanced MRI is optimally used for the differentiation of postoperative changes from recurrent herniated disk material (Fig. 275-1).18–20 Scar tissue generally exhibits a homogeneous pattern of enhancement, whereas disk material will usually demonstrate peripheral enhancement. The herniated disk fragment is typically continuous with the intervertebral parent disk unless sequestration of a free fragment has occurred. Neural elements are often displaced away from a herniated disk fragment, but thecal retraction toward epidural scar tissue can also be observed. Flexion-extension radiographs should be obtained to rule out dynamic instability and to evaluate the need for spinal fusion should revision surgery be necessary. If spinal fusion is being considered, standing 36-inch AP and lateral radiographs are used to assess the sagittal and coronal balance of the spinal column.
Epidural scarring, which occurs after any procedure in the epidural space, may also cause recurrent radiculopathy in some patients.21 The degree of scarring has been correlated with the development of recurrent symptoms. Radiculopathy in this setting is thought to be secondary to nerve root tethering. The generally accepted teaching is that in the absence of a herniated disk, patients with epidural scar tissue and concurrent symptoms do not benefit from reoperation on a consistent basis whereas those with a recurrent herniated disk and concordant radiculopathy do benefit from reoperation.22,23 Patients with degenerative foraminal stenosis or annular prolapse with neural compression may also benefit from revision surgery in selected cases.
Postoperative Spinal Instability and Deformity
Iatrogenic disruption of the posterior tension band, paraspinal muscles, and facet joint complexes may result in the development of spinal instability, with or without subsequent deformity. Additionally, abnormalities of the ventral spinal column may lead to or hasten the development of spinal instability after dorsal surgery (Fig. 275-2). Other factors, including disruption of more than half of the medial facets or young age, also increase the risk for iatrogenic spinal destabilization, especially in the cervical spine.24
Neurological injury is one of the most serious complications that may occur during surgery for correction of deformity. Many surgeons elect to perform intraoperative neurophysiologic monitoring (IOM) in the form of somatosensory evoked potentials, motor evoked potentials, and electromyography to minimize this risk.25–29 Although several studies have demonstrated that IOM may be beneficial in predicting and possibly minimizing the occurrence of postoperative neurological deficits, the importance of proper surgical technique and close attention to the integrity of the neural elements can be neither overemphasized nor replaced. Despite the increasing use of IOM in many centers, its routine use has yet to become universally accepted. This may, in part, be due to concerns regarding the time added to the procedure, cost, availability, and the potential for generating misleading information and unnecessary measures taken to address any false-positive alerts generated by IOM.30–32
Postlaminectomy Cervical Kyphosis
Postoperative deformity of the cervical spine most commonly occurs in the form of kyphosis or anterior subluxation. Accordingly, postlaminectomy cervical kyphosis is a well-established complication of cervical laminectomy.33–38 Cervical kyphosis in this regard refers to a loss of the normal lordotic cervical curvature of at least 5 degrees per motion segment (C2-7); however, overt reversal of the natural lordotic curvature to the point of global cervical kyphosis is not uncommonly encountered in this population. Cervical laminectomy involves resection of the spinous processes, laminae, and interspinous and supraspinous ligaments. The muscular attachments of the paraspinal muscles are disrupted and the erector spinae muscles are weakened as a result of partial denervation.39 The disrupted dorsal elements and weakened posterior tension band favor the development of cervical kyphosis because the natural balance between the extension and flexion moments of the cervical spine is shifted toward the latter.40,41 In the pediatric population, this imbalance often leads to the development of anteriorly wedged vertebral bodies because the growth of immature bone is strongly influenced by the asymmetry of the altered flexion and extension moments.42 Resection of more than 50% of the cervical facet joint complexes is also a strong risk factor for the development of cervical kyphosis and anterolisthesis.24,43,44
The true incidence of postlaminectomy cervical kyphosis has not been fully established, in part because of conflicting definitions of deformity, lack of the long-term follow-up required to detect the development of slowly progressive kyphosis in the adult population, and inclusion of only symptomatic patients in some studies. However, the risk for postlaminectomy cervical deformity in the adult population is reported to be between 5% and 50%.35,43,45–47 An incidence as high as 75% has been reported in patients who have undergone extensive (>50%) bilateral facet resection, thus highlighting the significant stabilizing force of the facet joint complexes.40,43 The exact proportion of patients in whom subsequent clinically symptomatic instability develops is not well established, but the rate may be as low as 5%.33,34
Although the precise incidence of postlaminectomy kyphosis is not known, several risk factors have been identified, including young age (<25 years), preoperative evidence of cervical instability or deformity, a neutral cervical spine, laminectomy performed for degenerative spondylosis, coexisting vertebral fracture or ligamentous injury at the levels to be operated, coexistent structural abnormalities of the ventral elements, the extent of laminectomy (partial resection of the facet joint complexes), and the number of segments (more than four) treated.36,40,46,48,49 Given the relative frequency of this complication, many surgeons elect to perform laminaplasty or prophylactically perform a dorsal instrumented fusion in patients with one of the aforementioned risk factors.37,42,48
Techniques for the correction of cervical postlaminectomy kyphosis include a dorsal approach, ventral approach, or a combined dorsal and ventral approach.38–40,50–53 Preoperative traction may be used in patients with a kyphotic deformity that does not fully reduce with simple neck extension but is not clearly a rigid deformity.38,39,54 The choice of surgical approach is dependent on several factors, including the degree of kyphotic angulation, the structural integrity of the anterior and posterior elements, the presence and location of compression of neural elements, and the presence of either a fixed, incompletely reducible or rigid deformity.39 In general, either a ventrally or dorsally performed revision surgery for cervical postlaminectomy kyphosis requires instrumentation and fusion.
Dorsal Approach
A dorsal surgical approach is often used for patients in whom the kyphotic deformity is either completely or satisfactorily reducible with neck extension and the structural integrity of the ventral bony elements is maintained.39 The neural elements should be free of any bone impingement in the reduced position. The basic techniques of revision spine surgery are applied as outlined earlier, such as extending the margins of the previous operative area in the rostral and caudal directions and working from areas of normal anatomy to those of surgically altered anatomy. Use of visual magnification techniques, such as the operating microscope or loupes, greatly facilitates safe dissection of scar tissue from both the bony and neural elements within the previous surgical field. After exposure of the appropriate spinal levels, additional decompression of the neural elements may be performed as necessary. Once decompression is achieved, reduction of the deformity can be performed. Reduction can be accomplished by having an assistant manually reposition the patient’s head into extension. Lateral fluoroscopy is used to confirm adequate cervical lordosis after manual reduction.
Ventral Approach
The ventral surgical approach is chosen when the kyphotic deformity is either rigid or minimally reducible with neck extension and preoperative traction.38,39 In addition, the ventral approach is used when the ventral bony elements are structurally compromised or are of insufficient height to maintain satisfactory alignment and a long-segment interbody fusion is required (see Fig. 275-2). Ventral decompression may be performed in patients with significant neural compression by ventral bony elements, prolapsed intervertebral disks, and osteophytic ridges. In those with high-degree kyphotic deformity, the spinal cord may be draped over the ventral bony elements (Figs. 275-3 and 275-4). In such cases, ventral decompression is required before any procedure for correction of the kyphotic deformity is performed because the application of corrective distraction forces will increase ventral compression of the spinal cord and may lead to spinal cord injury. The ventral approach affords the additional opportunity to perform ventral release of the longus colli muscles, the anterior longitudinal ligament, and the anterior annulus, thereby facilitating further anterior reduction of the kyphotic deformity.
FIGURE 275-4 Flexion (upper left) and extension (upper right) lateral cervical spine radiographs of the patient in Figure 275-3 demonstrating a 48-degree kyphotic deformity measured from C2 to C7 and centered on the C5 vertebral body. The deformity partially reduces to 13 degrees of kyphosis in extension. Lower left, Lateral cervical spine radiograph after 25 lb of Gardner-Wells traction, which provided an additional 4 degrees of correction of the deformity. Lower right, Postoperative lateral cervical spine radiograph 1 year after anterior diskectomies (C3-6), internal reduction, and interbody fusion. A supplemental occipital-T2 posterior fusion with screw-rod stabilization was also performed during the same procedure. An additional 11 degrees of correction of the deformity was achieved and resulted in 2 degrees of lordosis (C2-7).
Gardner-Wells traction is applied and a shoulder roll is placed beneath the scapulae to encourage cervical lordosis. Intraoperative lateral fluoroscopy is valuable in assessing alignment of the cervical spine and centering the incision above the levels to be operated on and during the placement of instrumentation. Either a transverse skin incision or, alternatively, an oblique incision along the anterior border of the sternocleidomastoid muscle (when exposure of three or more spinal segments is required) is performed. The usual method of anterior cervical spinal exposure as described by Cloward is then carried out.55 Caspar distraction posts are placed parallel to the vertebral body end plates, which will lie in a convergent fashion in patients with cervical kyphosis. Distraction is applied along the Caspar posts to place the posts in a parallel orientation. This effectively extends the cervical spine and reduces the kyphotic deformity; however, this maneuver must be used judiciously in patients with suboptimal bone mineral density to avoid vertebral body fracture. Additional distraction and reduction of the deformity are achieved by increasing axial traction via Gardner-Wells tongs by approximately 5 lb per cervical level. Further reduction of the kyphotic deformity is accomplished after performing anterior release of the anterior longitudinal ligament and the outer annulus.
Anterior decompression is performed, as necessary, depending on the degree of kyphotic angulation, the presence of compression of neural elements by ventral structures, and the structural integrity of the ventral bony elements. With regard to correction of deformity, we preferably perform multiple diskectomies rather than corpectomy, when possible, because we are able to obtain improved correction of a kyphotic deformity with the use of multiple lordotic interbody grafts rather than larger strut grafts. Single or multilevel discectomy is performed in the usual manner, depending on the number of segments involved in the kyphotic deformity. Should significant ventral bony compression of the neural elements or evidence of vertebral body collapse exist on preoperative studies, single or multilevel corpectomies are performed at the involved levels. This may be combined with additional discectomies above and below the corpectomy segment or segments as needed, depending on the number of segments with kyphosis. Some authors have advocated preservation of an intermediate point of fixation rather than performing multiple adjacent corpectomies for ventral correction of cervical kyphosis.53 This technique provides three- or four-point bending forces, which enhances the stability of the construct in addition to allowing the surgeon to “deliver” the intermediate point of fixation to a contoured implant for reduction of the deformity. Once complete anterior decompression has been achieved, manual intraoperative neck extension may be used to further increase cervical lordosis.
After decompression, lordotically shaped interbody strut grafts are applied. The interbody grafts should be placed in a ventral position, flush with the anterior margin of the vertebral bodies, to more effectively counteract the flexion moment in the sagittal plane because they will lie in a position as far ventral to the internal axis of rotation (IAR) as possible. This provides the interbody graft with a mechanical advantage to resist the sagittal flexion moment because the force required to resist the flexion moment decreases in a linear manner with increasing distance from the IAR.56 The Caspar distracting pins are then removed and the axial traction is relieved. This maneuver effectively applies an axial load on the interbody strut grafts and thereby enhances fusion and, in effect, offloads the screw-plate system. When performing deformity correction procedures it is preferable to place instrumentation and achieve immediate internal stabilization.50,52,57,58 An appropriately contoured screw-plate construct is then inserted under intraoperative fluoroscopic guidance. If an intermediate point of fixation has been preserved, it may be delivered to the plate system during screw tightening to further enhance lordosis as it is brought in toward the cervical plate. Use of a dynamic anterior screw-plate system, which permits axial subsidence while maintaining lordosis, both lessens the incidence of instrumentation failure and relieves mechanical stress at the screw-bone interface.59–61 The incision is closed in layers and a rigid cervical collar is applied for at least 12 weeks.
If a combined dorsal-ventral procedure has been planned, the patient may be placed in the prone position and dorsal decompression and fusion performed in the same setting or in a staged manner. In general, we prefer to supplement ventral surgery involving corpectomies at two or more levels with a dorsal lateral mass–pedicle screw instrumentation and fusion procedure.40,57,58,62 In cases in which significant posterior spinal canal compromise exists, dorsal decompression can be performed before dorsal fixation, as outlined earlier.
Segmental Instability after Posterior Lumbar Decompression
Dorsal decompression of the lumbar spinal canal is a well-established treatment of degenerative lumbar stenosis.63–67 In some patients who undergo dorsal decompression of the lumbar spine, segmental instability will develop as a consequence of both surgery and, to a lesser degree, structural compromise of the ventral elements. Iatrogenic disruption of the posterior tension band, paraspinal muscles, and facet joint complexes may lead to the development of segmental instability, with or without subsequent spinal deformity.68 Additionally, coexisting abnormalities of the ventral spinal elements, which often arise from degenerative disease, may also contribute to the development of spinal instability after dorsal surgery.
Several risk factors for the development of spinal instability after posterior lumbar decompression have been established, including preoperative radiologic evidence of dynamic instability or anterior spondylolisthesis, preserved disk space height, sagittally oriented facet joints, multilevel decompression, and the extent of facet joint complex or pars resection.44,69–72 It is also important to note that glacial spinal instability may develop in a minority of patients who undergo only partial laminectomy and diskectomy, particularly if more than half to a third of the facet joint complex has been disrupted or if overaggressive resection of the pars interarticularis has been performed.44 This may especially hold true when performing multilevel lumbar laminectomies because the distance between the pars interarticularis progressively decreases at the more cranial vertebral levels. Overaggressive pars resection is more likely to occur at the cranial levels, particularly L3 and above, if the decompression follows a rectangular rather than a trapezoidal shape. Injury to the facet joint from direct capsular cauterization may also contribute to facet joint denervation and glacial destabilization. Nevertheless, the presence one or more of the aforementioned risk factors should encourage the surgeon to consider the inclusion of an instrumented fusion in addition to a dorsal decompressive procedure. Furthermore, inclusion of an instrumented fusion will allow substantially more aggressive decompression of the neural elements because resection of the facet joints and pars interarticularis can be carried out. Such destabilizing maneuvers are frequently required in patients with high-grade canal and foraminal stenosis.
The initial treatment of lumbar spinal instability after lumbar decompression includes a comprehensive trial of nonoperative therapy. If nonoperative therapy fails to control the symptoms or if symptoms of neural compression are present, revision surgery is undertaken. The goals of revision surgery must be carefully and specifically defined. The surgical approach is influenced by the presence of compression of neural elements, the type and severity of the spinal deformity, and the need for grafts or instrumentation to address the deformity. In our experience, patients with lumbar instability and evidence of neural compression will attain maximal benefit from dorsal decompression and instrumented posterior lumbar interbody fusion (Fig. 275-5). The goals of this procedure are to provide decompression of the neural elements, reduction of the deformity, restoration of sagittal alignment, and 360-degree stabilization through one approach. Alternatively, if there is no evidence of neural compression, either an anterior, posterior, or combined stabilization and fusion procedure can be performed. Because many of these patients harbor superimposed progressive multilevel degenerative spinal disease, additional procedures to address the degenerative spine disease at adjacent levels may also be performed as indicated during the same procedure. Accordingly, decompression plus instrumented fusion is often performed at adjacent levels not included in the previous operation.
The surgical technique of posterior lumbar interbody fusion for postlaminectomy lumbar instability consists of exposure of the normal anatomy not included in the previous operation, decompression of the neural elements by further resection of bone and epidural scar tissue, posterior interbody fusion, and instrumented lateral fusion. Under general anesthesia, the patient is placed on a radiolucent Jackson table in the prone position. It is essential that the hips be extended fully to allow the lumbar spine to achieve maximal lordosis. Any hip flexion can be associated with relative kyphotic angulation of the lumbar spine and may predispose to the development of flat back syndrome.73 A radiolucent table is used to aid in intraoperative fluoroscopic imaging in both the coronal and sagittal planes.
After pedicle screw placement, a diskectomy is then performed. Care is taken to preserve the ventral and lateral margins of the annulus to allow the graft material, both structural and morselized, to be contained within the interbody space. Once this is completed, the interbody graft is placed. We prefer the use of two carbon fiber cages packed with morselized graft material, although other types of structural interbody graft can be used as well. The carbon fiber cages that we use have a lordotic sagittal contour with the ventral surface being 2 mm larger than the dorsal surface. In patients with significant loss of lumbar lordosis or a positive sagittal balance, correction of the deformity can be achieved with the use of lordotically shaped interbody grafts.74,75 In patients with spondylolisthesis, a cage of shorter depth is used. For higher grades of spondylolisthesis, the interbody graft is placed after at least partial reduction of the deformity has been accomplished. We perform interbody graft placement via the transforaminal approach. By resecting the medial aspects of the subjacent superior facets, manipulation of the thecal sac and nerve roots is kept to a minimum during placement of the interbody graft.76
Pseudarthrosis
Pseudarthrosis is a commonly encountered complication of both instrumented and noninstrumented spinal fusion surgery.77–86 Pseudarthrosis is defined as the presence of symptomatic bony nonunion more than 1 year after fusion surgery.56 It may lead to motion of the operated spinal motion segment with correlative signs and symptoms of spinal instability. The term nonunion is defined as permanent failure of bone growth across a fracture, regardless of the presence or absence of symptoms and signs of instability. Spinal nonunion can result in a variety of clinical manifestations ranging from an asymptomatic radiographic finding to persistent mechanical or radicular pain or, in its most severe form, to catastrophic construct failure with resultant deformity and neurological injury. Although pseudarthrosis is one of the leading complications of cervical, thoracic, and lumbar arthrodesis surgery, there is a wide range of its reported incidence in the literature.9,78,79,81–83,85–91 This variability is due to diverse surgical and patient factors and the varying methods used to assess bony fusion. The incidence of pseudarthrosis in spinal fusion surgery, in general, has been reported to be approximately 15%91; however, rates approaching 50% have also been reported.85,86,90 This figure varies greatly according to the indications for and the type of surgery, the number of levels fused, the surgical technique, the use of different graft materials or fixation devices, patient demographics, and the presence of systemic factors that could adversely affect bone fusion.
Biology of Bone and Spinal Fusion
Long bone healing occurs in a continuous sequence of acute inflammation, early and late repair, and remodeling.92–94 The acute inflammatory phase occurs 2 to 4 weeks after injury. This phase is initiated by local vascular injury, hemorrhage, and hematoma formation. Local vascular injury leads to an infiltration of acute inflammatory cells and an influx of fibroblasts. Neovascularization and growth of granulation tissue then occur after the acute inflammatory phase. Secretion of a fibrocartilaginous matrix by fibroblasts and formation of soft callus characterize the early repair phase. Callus formation typically occurs approximately 4 to 6 weeks after injury. As the repair process continues, osteoblastic activity progressively replaces the soft fibrocartilaginous callus with woven bone. After a period of 3 to 6 months, the callus is fully replaced by mature cancellous and cortical bone. During the remodeling phase, the fracture callus is remodeled into a new shape that resembles the bone’s original geometry.
Bone graft incorporation plus fusion is a distinct process that entails a phasic response similar to that of fracture healing in which devascularized bone graft is progressively revascularized, resorbed, and incorporated into areas of new bone growth.92,95–97 Decortication and bleeding from the graft recipient site initiate an acute inflammatory phase in which there is local influx of inflammatory cells into the graft recipient site. Revascularization occurs from the host site into the grafted bone. In cortical bone, much of the original bone matrix is replaced during a remodeling phase in which osteoclasts first migrate into the newly vascularized bone graft and resorb the graft matrix. The influx of osteoclasts is followed by activated osteoblasts that produce new bone matrix within the bony voids created by the osteoclasts. The structural integrity of grafted cortical bone is substantially weakened approximately 6 months postoperatively because of osteoclastic resorption until it undergoes remodeling.
In contrast to cortical bone, cancellous bone is first remodeled by an influx of osteoblasts that surround the edges of the grafted cancellous bone trabeculae. Osteoid is deposited by osteoblasts in a lamellar fashion around the dead trabeculae. This process is eventually followed by resorption of the entrapped areas of grafted bone by osteoclastic activity. Thus, in contrast to the early resorption of grafted cortical bone during the remodeling phase, cancellous graft is first strengthened by the formation of new bone before it is resorbed. In this manner, cancellous bone graft attains structural integrity over a period of 3 to 6 months, whereas cortical bone graft attains structural integrity over a period of up to 1 to 2 years. Additionally, revascularization of cancellous bone occurs over a 1- to 2-week period, whereas revascularization of cortical bone takes several months.96,97
A graft is referred to as osteogenic if it can become bone. The process of osteoinduction entails the differentiation of mesenchymal cells into osteogenic cells. Recombinant human bone morphogenetic protein-2 (rhBMP-2) is a well-studied osteoinductive growth factor in spinal fusions.98–100 Osteoconduction involves the growth of capillaries into a graft and the accompanying migration of osteoprogenitor cells along the neovascular channels. These processes are tightly regulated by numerous cytokines, growth factors, and hormonal factors.93,101,102 Autologous graft retains some surviving osteogenic cells, as well as endogenous bone morphogenetic protein (BMP). Autograft is therefore osteogenic, osteoconductive, and osteoinductive.103 Allograft exhibits osteoinductive and osteoconductive properties. Demineralized bone matrix also has osteoinductive and osteoconductive properties.104
Factors Affecting Fusion
Medical Considerations
Bone quality is a well-established factor that affects bone healing and fusion. In addition to age-related osteoporosis, bone quality is also negatively influenced by the presence of systemic diseases such as rheumatoid arthritis, diabetes mellitus, hyperparathyroidism, poor nutritional status, and systemic inflammatory conditions.105–110 Medications used to treat systemic inflammatory diseases, such as nonsteroidal anti-inflammatory drugs (NSAIDs), steroids, and cytotoxic agents, further confound the issue of bone quality in these patients because they are also factors that are known to substantially impede bone healing.102,111–114 Steroids, NSAIDs, and cytotoxic medications interfere with the acute inflammatory phase of bone healing and should be avoided for a minimum of 4 to 6 weeks postoperatively or preferably until fusion has been confirmed.115,116 In patients with the aforementioned risk factors, bone densitometric studies, such as dual-energy x-ray absorptiometry (DEXA), are helpful in quantifying bone quality in an objective manner. DEXA studies are usually performed on the femoral neck, distal radius, and lumbar spine. In patients who have previously undergone lumbar fusion, the femoral neck and radius should be used instead to assess global bone mineral density.117 Quantitative CT is a highly specific and sensitive modality for the detection of osteoporosis that can be used to analyze bone density in both the cortical and trabecular compartments at any skeletal site.118–120 Such use of this modality is currently limited by increased cost and radiation exposure.121 Quantitative evidence of poor bone mineral density provides a scientific basis for the failure of a previous spinal fusion and should compel the clinician to use advanced techniques for enhancement of the chance of successful fusion in subsequent operations. This may be accomplished by improving the construct design, using autograft or biologic modifiers of bone healing, and load sharing with extended constructs or the addition of circumferential stabilization. Modifiable risk factors such as malnutrition and hyperparathyroidism require correction before any elective revision fusion surgery is performed. In patients with rheumatoid disease or other systemic inflammatory conditions, cytotoxic medications, steroids, and NSAIDs should be tapered and discontinued (if possible) preoperatively and avoided for as long as possible during the postoperative period to promote bone healing and fusion.
Osteoporosis is a widely prevalent metabolic disease that mainly affects postmenopausal women. Iatrogenic osteoporosis is also known to occur in patients who have undergone bilateral oophorectomy or long-term corticosteroid therapy. Osteoporosis is characterized by a general loss of bone mineral density and total bone mass. DEXA studies are essential in quantifying the extent of disease. In patients with significantly reduced bone mass, fusion surgery is generally avoided when possible. In the setting of osteoporosis and failed spinal fusion, a multifaceted strategy is required to optimize bone mineral density, as well as the biomechanics of the construct design. Medical therapy, which includes calcium and vitamin D supplementation, should be used to optimize bone mass in these patients.122 Patients who have osteoporosis secondary to oophorectomy are often able to achieve increased bone mineral density after estrogen replacement therapy.123 Bisphosphonate therapy, which has emerged as a mainstay in the treatment of postmenopausal osteoporosis, has been shown to paradoxically decrease spinal fusion rates in several animal models.124–127 Thus, this agent should be avoided in patients undergoing fusion surgery. More recently, recombinant parathyroid hormone (rPTH) has gained increasing acceptance for the treatment of postmenopausal osteoporosis.128 rPTH must be administered systemically on a daily basis and contains amino acids 1 to 34 of the complete PTH molecule, which is the osteogenic component of PTH. rPTH exerts an anabolic affect on bone healing by increasing the rate of new bone formation, as well as by increasing bone mineral density.129,130 rPTH has been shown to increase the rate of bone formation, accelerate spinal fusion, and increase the mineral density of spinal fusion masses in animal models.131,132
Cigarette smoking is a common modifiable risk factor for pseudarthrosis that can drastically affect the rate of spinal fusion. The risk for nonunion in patients who smoke is approximately three times that of nonsmoking patients, with fusion rates being decreased by up to 40% in smokers.133–138 Thus, smoking is generally regarded as the single most deleterious modifiable risk factor for nonunion. Smoking is known to restrict the growth of small vessels into fusion sites, inhibit revascularization of the bone graft, and inhibit osteoblastic function, and it may lead to a generalized loss of bone density.139,140 The nicotine component of tobacco is a well-recognized inhibitor of revascularization of bone grafts and a direct inhibitor of osteoblastic function.139,141 The presence of systemic nicotine itself is an established risk factor for spinal nonunion.141,142 Given the high rate of pseudarthrosis in smokers, many physicians simply refuse to perform elective spinal fusion in patients who continue to smoke. An intermediary approach that involves referral of patients with moderate symptoms and no evidence of structural instability to smoking cessation programs has also been advocated.143 The decision to perform fusion surgery in these patients then depends on a combination of the success of a smoking abstinence program, the presence of additional risk factors for nonunion, and the patient’s symptoms. In patients with symptomatic pseudarthrosis who smoke, however, implementation of a preoperative smoking cessation program may not be feasible. In this setting, the focus should be placed on surgical strategies to improve bone healing, as outlined earlier, in addition to an aggressive postoperative smoking cessation program.135 Although nicotine replacement therapy is useful in preoperative smoking cessation programs, given the deleterious effects of nicotine on spinal fusion, postoperative nicotine replacement therapy should be avoided in patients undergoing fusion surgery.
General Surgical Considerations
In patients with established pseudarthrosis or who harbor additional risk factors for nonunion, it is critical to use all available surgical techniques that will optimize bone healing and maximize the chance of obtaining a successful fusion. Surgical techniques that affect bone healing and improve the chance of achieving successful fusion include proper preparation of the host site, graft placement, and type of graft implanted; the use of biologic agents that enhance bone growth, such as rhBMP-2; and electrical bone stimulation. Meticulous attention should be directed toward placement of a biomechanically sound internal fixation construct to attain immediate and durable stabilization of the operated motion segments. Surgical strategies are often directed toward load sharing by extending the length of the construct over many segments and optimizing the biomechanics of the bone-implant interfaces. Additionally, circumferential stabilization may also be achieved by performing ventral stabilization in addition to dorsal stabilization.4,143–146
Graft Selection
Selection of the graft material depends on the biomechanical environment of the construct, the presence of biologic risk factors for nonunion, and the type of surgery. The gold standard graft material for spinal fusions remains autologous iliac crest.103,147–156 Other options for graft material include locally harvested bone, autologous rib or fibula, morselized nonstructural autograft packed in titanium, polyetheretherketone (PEEK) or carbon fiber cages, allograft (structural or morselized), or bone modification agents such as BMP.103 Each type of graft material has its own advantages and disadvantages.157 In general, use of the patient’s own bone is preferable to the use of allograft or bone modifiers for several reasons. Autograft is genetically identical to the patient, carries no risk for donor-related infection or rejection, and has inherent osteogenic, osteoinductive, and osteoconductive properties.92,103,158 Disadvantages of autograft include harvest site complications such as bleeding, pain, nerve injury, and hernia formation, as well as the chance of harvesting structurally inadequate bone in osteoporotic patients.159–161
Allograft bone is advantageous in that it is readily available in the United States, is structurally sound, and carries no risk for donor site complications. There is a low risk for donor-related infection and immune-mediated rejection, although this risk varies according to the particular methods used for pretreatment of the donor bone.158 Allograft does retain osteoinductive and osteoconductive properties but has no osteogenic properties. Its osteogenic properties may be enhanced, however, by supplementation of the allograft with BMP or autograft. Multiple studies comparing the fusion rates obtained with autograft and allograft continue to substantiate the superiority of autograft in achieving solid bone fusion.148 However, as newer technologies have emerged, the fusion rates seen with allograft continue to improve. Although equivalent fusion rates with autograft and allograft have been observed for certain types of spinal fusion (e.g., single-level anterior cervical fusion),149,162