Chapter 201 Nonunion
Recent advances and refinements in instrumentation have greatly expanded the capacity of surgeons to successfully treat complicated problems over the entire spine (Fig. 201-1). These systems aid the reestablishment of normal or near-normal alignment, may apply complex multidirectional force vectors, and may allow immediate immobilization over multiple spinal segments. As important as the meticulous and thoughtful use of this instrumentation is, careful consideration must be given to the material that will, for a lifetime, bear the stresses that instrumentation supports only temporarily. Solid bony union alone provides enduring spinal stability, and failure to achieve this goal can have consequences ranging from a benign radiographic finding to persistent pain or catastrophic construct failure. The nonunion rate varies with the type of operation performed, being higher when multiple-level fusions are attempted. A pseudarthrosis is defined as a documented failure of continuous bone formation over time that leads to a definitive absence of bone healing through a fracture or new bone formation at an intended arthrodesis site. From a practical standpoint, a nonunion in spine surgery has been defined as the absence of solid fusion 1 year after the operation with concomitant symptoms and signs.1 A large, population-based, prospective review of lumbar spine revisions showed that 23.6% of the indications were for pseudarthrosis.2 Overall, the true incidence of post–spine surgery nonunion is probably underestimated, because many cases are asymptomatic and require no treatment.
Biology of Bone Healing
Bone is a dynamic living tissue that undergoes constant remodeling.3 It is unique in its capacity to repair and regenerate after disruption.4 The amazing qualities of bone are derived from its unique composite structure of organic and inorganic materials. The organic component, chiefly a strongly cross-linked (type I) collagen, gives bone plasticity that allows substantial deformation without fracture and tolerates stress in tension.5 The inorganic component, chiefly in the form of hydroxyapatite, precipitates around the collagen fibers in a process of nucleation and maturation of mineral crystals.6 The inorganic-mineral component of bone gives the tissue tremendous strength in compression and bending.7
The cellular components of bone include osteoblasts, osteocytes, and osteoclasts, connected through an intricate and well-organized system of canals.8
Osteoblasts are derived from mesenchymal stem cells from the bone marrow and periosteum.9–12 They are responsible for bone formation. Osteoblasts fabricate bone in response to many stimuli and under different conditions such as growth, physiologic remodeling, fracture healing, and heterotopic ossification.13–15 Researchers have shown that new bone is formed in response to tumors and infections.16 An investigation has shown that osteoblasts have the ability to form bone during distraction osteogenesis,17 when substituting the void initially filled by autologous or allogeneic bone graft, demineralized bone matrix, or synthetic bone substitutes. When performing anterior cervical discectomy and fusion (ACDF) and plating, 97.5% of fusion with new bone formation has been achieved with either autograft or allograft.18 In a recent study, Jensen showed an 86% union rate after single- and multiple-level ACDF using patella allograft and plating.19 In a study from Japan, Momma has reported complete bone remodeling after 6 to 12 months on CT scan, after the use of B-tricalcium phosphate to fill a partial ventral vertebrectomy defect done for cervical decompression surgery.20
Osteoclasts are derived from hematopoietic stem cells. They exit the circulation close to the site to be remodeled.6,21 They are responsible for the bone resorption.
The mature osteoblast produces proteins such as type I collagen, osteocalcin, and alkaline phosphatase, a key enzyme in bone mineralization. Osteoblasts become entrapped in their own osteoid matrix and develop long cytoplasmic processes to remain in contact with surrounding cells.22 They then begin expressing a whole new set of genes to continue bone turnover and mineral homeostasis. These cells are now considered osteocytes (mature bone cells).6
The process of bone healing after injury is an indistinct continuous sequence of inflammation, repair, and remodeling.10,12 The inflammatory response to injury includes vascular dilatation with exudate and edema, as well as inflammatory cell (polymorphonuclear lymphocytes, macrophages, and lymphocytes) infiltration. A variety of hormones, cytokines, growth factors, and matrix proteins (e.g., bone morphogenetic protein [BMP]) is involved throughout the healing process. Nonsteroidal anti-inflammatory drugs (NSAIDs), steroids, or chemotherapeutic agents given during the first week of healing may blunt the inflammatory response and impair bone healing.10,12,23 As the debris of the inflammatory phase is removed, fibroblasts begin laying down new matrix in the early phases of the repair. Initially, a fracture callus that is composed of fibrous tissue, cartilage, and woven bone may form to bridge the bony defect. This is then replaced by woven bone and, ultimately, by mature cortical or cancellous bone. This process may take 3 to 6 months or longer, depending on age and other factors.10,23 Although the general sequence of inflammation, repair, and remodeling that occurs in long bone fracture healing also occurs with bone graft repair, there are some distinct differences. Autograft bone used in spinal fusion is initially deprived of blood supply, although a robust nonspecific inflammatory response occurs as a result of preparation of the graft recipient bed. The collection of coagulated blood around the graft is somewhat analogous to the hematoma of an acute fracture, with the complex processes of inflammation ongoing within this milieu. Although some of the periosteum, endosteum, mesenchymal cells, and osteocytes within 0.2 to 0.3 mm of the borders survive transplant, most of the transplanted bone cells, separated from their blood supply, die.24 The cancellous portion of the bone graft may be revascularized within 2 weeks, and cortical bone is revascularized within 1 to 2 months. Cancellous bone is more rapidly remodeled and is initially strengthened during the remodeling phase, because osteoblasts are first laid down over the trabeculae.5
Cortical bone is weakened during initial remodeling, and the process is slower than in cancellous bone. Bone graft is gradually replaced with new bone in a process called creeping substitution24 Osteoclasts that act as cutting cones bore into the graft from the margins of host bone, followed by osteoblasts that lay down new bone. This process of healing and remodeling may leave as much as 50% to 90% of the original matrix, even after many years.25,26 The strength of cortical autograft is halved during the first 6 months after fusion but is gradually restored over 1 to 2 years. Autograft bone provides some living bone cells with the ability to make bone (i.e., osteogenic properties). It contains BMP and other substances capable of inducing cellular differentiation (i.e., it has osteoinductive properties), and it provides a scaffolding for bone growth (i.e., osteoconductive properties).27
Harvesting and Handling of the Bone Graft
Surgical exposure of the donor site should be performed to maximize the viability of the graft. Heating of bone with the electric cautery, although frequently unavoidable, has the potential of destroying those surface osteocytes most capable of surviving via diffusion of nutrients. A sharp periosteum elevator can be used to open the subperiosteal plane in a remarkably atraumatic fashion, with minimal blood loss. For tricortical bone grafts placed in compression, harvesting with an oscillating saw provides a graft that is better at resisting compressive loads than a graft harvested with an osteotome.28,29 The clinical significance of this is unknown. For bone used purely for onlay, one should bear in mind that 5 mm is the maximal thickness that can be nourished by diffusion of nutrients.30,31 If possible, the bone graft should be harvested within 30 minutes of planned use. The graft should be kept moist in a saline- or blood-soaked sponge before use. The graft should not be allowed to dry or come into contact with toxic chemicals (e.g., antibiotic solutions).32
Selection of Graft Material and Instrumentation
Ideally, graft material should have the capacity to form bone (i.e., osteogenic properties), induce undifferentiated mesenchymal cells to mature into osteoblasts (i.e., osteoinductive properties), serve as scaffolding for bone healing (i.e., osteoconductive properties), harbor no risk of infection, and be genetically identical to the patient. It should also be mechanically strong, durable, potentially viable, nonreactive to the host tissue, sterile, anatomic, and cost effective.33 Currently, the material that comes closest to these requirements is the patient’s own (autograft) bone. The most common choices for autograft in spine surgery include iliac crest, local bone, or rib. The disadvantages of an autograft include the potential for inadequate bone graft volume or quality, risk of wound hernia, pelvic fracture for iliac crest grafts, blood loss, infection, nerve injury, and, most commonly and bothersome at the iliac crest, chronic graft harvest site pain. The incidence of major complications with autograft harvesting can be as high as 10%33 or even 17.9% when using the same skin incision for iliac crest harvest and the primary spine procedure.34 Chronic persistent pain at the donor site ranges from 2.8% to 70% of patients,35–39 with most series reporting it to be about 20% to 30%.35–37
Allograft bone has the advantage of being readily available in multiple structural forms and without donor site morbidity. Allograft has some osteoinductive and osteoconductive, but no osteogenic, properties. Vascular ingrowth and new bone formation are delayed with allografts.40–42 The mode of preparation of allograft may have an impact on its success as a graft material. Allograft bone may be treated with freezing, freeze drying, or ethylene oxide to reduce its immunogenicity; however, because it is genetically dissimilar to the patient, an inflammatory response similar to graft rejection noted in other tissue transplants may occur.43,44 Fresh-frozen allograft appears to have a superior fusion rate to freeze-dried graft, with ethylene oxide–sterilized grafts demonstrating uniformly poor results.45
Cervical Spine
The reported outcomes of noninstrumented single-level ACDF procedures with the use of autologous ventral iliac crest bone graft (ICBG) include fusion rates between 83% and 100%.46–52 The use of allograft bone for single-level ACDF appears to yield results that are approximately equivalent to use of autograft bone.47,53–60 For multilevel ventral procedures, autograft classically appears to be superior.58,61–63 A recent study, however, shows equal fusion rates of 97.5% using either autograft or allograft and ventral plating for multilevel ACDF.18 For bone struts used over multiple segments, pseudarthrosis rates with allograft are higher (41%) than with autograft (27%).64 When supplemented with dorsal fusion, the pseudarthrosis rate falls to 26%.65
A separate study using notched fibular struts and a halo orthosis demonstrated delayed fusion, but seven of eight patients had good or excellent results.53 In fusions unsupplemented with dorsal instrumentation or a halo orthosis, autograft bone appears to be the favored graft material. For dorsal cervical fusions, autograft bone appears to be superior in some studies.4,66 Ventral plating to add stability to the ACDF construct, especially if multilevel, has clearly increased the union rate of the procedure.48,51,52,67–69 In a recent meta-analysis, the authors concluded that plating significantly increases the fusion rate of ACDF regardless of the number of levels. They also noted that corpectomies had higher fusion rates than multilevel ACDFs and that the use of plates improved the fusion rate of three-level but not two-level corpectomy surgery70 (Table 201-1).
The use of a cylindrical titanium mesh cage packed with bone salvaged from the corpectomy may avoid the need for harvesting a separate graft. Fusion rates using this method of reconstruction have been reported to be greater than 90%.71,72
Lumbar Spine
It is well established that instrumentation in the lumbar spine increases fusion rates. In 1997, Fischgrund et al.73 published a prospective, randomized trial comparing decompression and dorsolateral fusion with and without instrumentation in patients with degenerative spondylolisthesis and spinal stenosis. The average follow-up was 2 years. The fusion rate was significantly better with instrumentation than without (82% vs. 45%, respectively; P = .0015); however, no significant difference was found in clinical outcome. In 2004, Kornblum et al.74 analyzed the patients from the 1997 Fischgrund study (now with a follow-up of 5–14 years) and noted that the patients with a solid fusion did significantly better than patients with a nonunion. Other authors have also conducted prospective, randomized studies looking at the same issue. Whereas Zdeblick’s results also support rigid instrumentation,75 Thomsen et al.76 reported fusion in 68% of instrumented cases and in 85% of noninstrumented cases.
In the only randomized trial comparing circumferential fusion with dorsolateral fusion alone (to our knowledge), a significantly higher fusion rate was found with circumferential fusion (92% vs. 80%; P < .04).77 In a meta-analysis looking at different lumbar fusion procedures, Bono et al. concluded that the highest rate of fusion was obtained with a circumferential technique (91%; P = .06), followed by posterior interbody (89%; P = .05), anterior interbody (86%), and finally posterolateral (85%) techniques.78 The clinical relevance and cost of the circumferential procedure for fusion in the lumbar spine is still open to debate.
In a study of long-segment fusion to the sacrum (mean of 11.9 vertebrae) for adult spinal deformity, the pseudarthrosis rate was 24% (much higher than for short constructs). Half of these pseudarthroses occurred through the thoracolumbar junction and one fourth through the lumbosacral junction. Risk factors were kyphosis of 20 degrees or higher, positive sagittal balance of 5 cm, hip osteoarthritis, a thoracoabdominal approach, age older than 55 years, and incomplete lumbopelvic fixation (complete lumbopelvic fixation defined as L5-S1 interbody fusion and iliac screw fixation). Augmenting the number of fused levels into the upper thoracic spine did not improve the fusion rates.79
In the lumbar spine, allograft bone plays a limited role with dorsolateral fusion. However, its use as a ventral interbody strut (particularly femoral shaft allograft packed with cancellous autograft) when used with dorsal segmental instrumentation has been substantiated.80–82
Pediatric Spine Surgery
The use of allograft versus autograft bone has been well studied for scoliosis.83–86 In uninstrumented cases, autograft performed superiorly. Instrumented dorsal fusions supplemented with allograft bone performed comparably in a pediatric population (although it took a long time to achieve fusion). Ventral allograft struts supplemented with autologous bone (packed into the hollowed marrow space of the allograft), in conjunction with dorsal fusion and segmental instrumentation, yielded better results than allograft fusion without ventral graft supplementation.86 When treating high-grade pediatric isthmic spondylolisthesis, Molinari et al. performed a very interesting study. He divided patients in three groups with the following results:
Bone Graft Substitutes
Demineralized Bone Matrix
Demineralized bone matrix (DBM) facilitates bone fusion through osteoinduction and, to a lesser degree, through osteoconductive properties. During the fabrication of DBM, the demineralization of allograft reduces its antigenicity and may uncover osteoinductive factors, including BMP. However, BMP-2 and BMP-7 exist in nanogram concentrations in DBM, which is one million times less than the concentration of BMP required to produce a lumbar fusion clinically.88–90 DBM may vary in its osteoinductive capabilities, based on the cadaveric bone from which it is derived, by vendor, and even among batches of the same brand. Several animal studies favor DBM when compared with autogenous bone in achieving spinal fusion.91–93 Clinically, the data are more limited. A prospective trial comparing allograft and DBM with autograft in ventral cervical fusion demonstrated only a higher rate of graft collapse and pseudarthrosis in the allograft group.94 In the lumbar spine, the use of DBM plus local bone achieved the same fusion rates as ICBG for a single-level posterolateral fusion.95
Synthetic Bone Substitutes (Ceramics)
Numerous calcium-based synthetic products have emerged as bone graft substitutes and/or extenders in spine fusion. These products serve as scaffolds that support new bone ingrowth. During manufacturing, the porosity of these materials can be optimized for bony ingrowth.96 Calcium sulfate is not sufficient for use as an osteoconductive material, because it absorbs in only a few weeks, much before new bone has formed in a fusion.97 Hydroxyapatite takes several years to be reabsorbed, and its radiopacity makes the radiographic diagnosis of fusion difficult. Beta tricalcium phosphate absorbs in months, thus lasting an adequate period to conduct bone growth during fusion.97 For this reason, most ceramic products used in spine fusion these days are made of beta tricalcium phosphate and/or hydroxyapatite in combination with bovine collagen in varying ratios. Collagen affects the workability and reabsorption rate of the ceramic and may also serve as a carrier for osteoinductive agents, such as BMP.98
Animal studies show excellent fusion rates (superior to the control group with autologous bone) when ceramics are used in combination with bone marrow aspirate (osteoinductive and osteogenic) and very poor results when ceramics are used alone. The authors concluded that the association of bone marrow aspirate is paramount to the success of the procedure.99,100
In prospective case-controlled clinical series, the fusion rates for lumbar posterolateral fusion and transforaminal lumbar interbody fusion using ceramics in conjunction with bone marrow aspirate and/or local autograft yield similar good results as with ICBG (fusion rate from 92% to 100% for a single-level instrumented fusion).101–104 In the cervical spine, Momma has reported complete bone remodeling, after 6 to 12 months, on CT scan, after the use of beta tricalcium phosphate to fill a partial anterior vertebrectomy defect done for cervical decompression.20
Bone Morphogenetic Protein
BMPs are a group of growth factors originally discovered because of their ability to induce the formation of bone and cartilage. Originally, seven proteins from this group were discovered. Of these, six (BMP-2 through BMP-7) belong to the transforming growth factor-beta superfamily of proteins, whereas BMP-1 is a metalloprotease, involved in cartilage development. Since then more BMPs have been discovered, making a total of approximately 20 today.105 BMP bone induction is a sequential cascade. The key steps in this process are chemotaxis, mitosis, and differentiation as shown on early studies by Reddi.106 They are known to stimulate osteoblasts and inhibit osteoclasts.107,108 Currently, the Food and Drug Administration (FDA) has approved two BMPs for use in humans as bone growth inducers—BMP-2 and BMP-7. BMP-2 is the BMP used in almost the totality of current spinal fusion cases because of FDA regulation issues with BMP-7. Multiple preclinical animal studies have shown that the use of BMP results in similar, if not superior, fusion rates with biomechanically stronger fusion masses when compared with autogenous bone graft.109–111
Prospective, randomized clinical studies have shown that BMPs have at least comparable fusion rates and clinical outcomes when compared with ICBG in both interbody and posterolateral lumbar fusions.112–114
One prospective nonrandomized study in the ventral cervical spine reports that fusion rates of allograft and recombinant human bone morphogenetic protein (rhBMP-2; 0.9 mg per level) were slightly better than those of ICBG. However, 50% of the patients receiving the BMP had significant neck swelling.115 Another study reported 27.5% clinically significant neck swelling after ACDF with BMP.116 The safe dose of BMP and the best method for delivery are yet to be determined in the cervical spine. In 2008, the FDA issued a warning concerning its use in cervical surgery.117
Recently discovered bone morphogenetic protein-binding peptide (BBP) is a 19–amino acid peptide that has been shown to bind BMP and potentiate its effect of bone healing in animal studies. BBP may provide for improved fusion rates with a smaller dose of BMP required, potentially reducing cost, as well as potential side effects of BMP such as inflammation and ectopic bone formation.118
Influence of Electromagnetic Stimulation on Bone Healing
Direct current stimulation was first proposed in 1972 as a modality for improving fusion.119,120 Application of pulsed electromagnetic fields for nonunion in long bone fractures appears to have no hazardous side effects. Areas of tension are associated with a net positive charge, and compressive stresses are associated with a net negative charge (10–100 mV) and osteogenesis.120 Electromagnetic stimulation is believed to promote osteogenesis as a result of more rapid angiogenesis and decreased osteoclastic activity.121 The effect of improved osteogenesis may be mediated by growth factors.122 More recent evidence also suggests the activation of a second messenger system involved in bone remodeling. Three broad types of electromagnetic fields are used: implantable direct current, pulsed electromagnetic fields, and capacitively coupled electrical energy. Pulsed electromagnetic fields and capacitively coupled electrical energy are examples of external electromagnetic fields. These are delivered via external electrodes attached to a corset. Implantable direct current requires surgical placement of the electrodes and has been shown to be the most effective.123 Direct current may be more effective than external electrodes secondary to its increased precision in the distribution of current.124 The cost of these devices is not insignificant. Although some investigators have noted no significant benefit of electrical stimulation for canine spinal fusions,125,126
