It is estimated that more than 200,000 spine fusion procedures are performed each year in the United States. Posterolateral lumbar intertransverse process arthrodesis is the most common procedure performed, yet failure to achieve a solid bony union (nonunion) occurs in 10% to 40% of patients with single-level fusions, and more frequently when multiple levels are attempted. 15,19,25,26,30,61 This high rate of nonunions indicates that the physiologic, biologic, and molecular events crucial to this process are not adequately understood. A nonunion frequently leads to unsatisfactory resolution of clinical symptoms, 17,22,68 and usually results in greater medical costs and morbidity, as well as the need for additional surgeries.
The most common clinical approach for preventing nonunions has been the use of internal fixation (e.g., hooks or screws with rods or plates). Although the use of internal fixation has decreased the number of nonunions, it has not eliminated the problem. Nonunions still occur in 10% to 15% of patients. 15,46,65,66 Clearly, other factors must be implicated, yet there is a paucity of knowledge regarding the biologic mechanisms involved. 20,43–45 Such knowledge would be helpful in devising strategies for nonunion prevention and allowing appropriate use of different bone graft substitutes for posterolateral spinal fusion.
The healing of a spine fusion is a multifactorial process, which makes it difficult to study in the clinical setting. The lack of reliable noninvasive techniques for assessing the success or failure of an arthrodesis further limits clinical studies. 16 Thus, an animal model is a practical solution for studying individual factors involved in this complex process. 57 Most models used before 1990 had limited utility for biologic studies for one or more of the following reasons: 1) the successful fusion rate was 100%, which is much higher than that seen clinically; 2) the animals were skeletally immature; 3) the arthrodesis performed was interfacet or interlaminar rather than intertransverse; or 4) the model destabilized the spine and resulted in 0% successful fusions without internal fixation. 35,51,54,55,67 The limitations of these earlier models led to the development of an intertransverse process arthrodesis model that was more clinically applicable to the human situation. 10 The major benefit of this type of model is that nonunions spontaneously occur at a rate comparable with that reported in humans. 10,57
Biology of Spine Fusion
The rabbit lumbar intertransverse process arthrodesis model has been well characterized using autogenous iliac crest as the graft material. Mechanically solid fusions generally occur by the fourth postoperative week with an overall nonunion rate of 30% to 40%. Radiographic analysis has showed progressive remodeling of bone graft material with time, usually by 10 to 12 weeks, but as in humans, radiographs were accurate in assessing success or failure to attain solid fusion only 70% of the time. This model was further validated using control animals with omission of iliac bone graft placement or omission of the decortication step, which resulted in nonunion in all cases. These negative control animals showed that the surgical exposure alone did not automatically result in spinal fusion. This was a limitation of some previous models. Vascular injection studies indicated that the primary blood supply to the fusion mass originated from the decorticated transverse processes, not the surrounding soft tissues. 63 The failure to achieve spine fusion in the absence of decortication emphasized the importance of extensive decortication of the posterolateral spine elements (lateral facet, pars interarticularis, transverse process) in providing bone marrow, vascularization, and osteoprogenitor cells to the fusion mass.
Histologic analysis of the rabbit transverse process fusion has showed three distinct and reproducible temporal phases of spine fusion healing. 12 Maturation of the spine fusion was most advanced at the ends of the fusion mass near the transverse processes (“outer” zone). A similar histologic progression occurred in the “central” zone, but was delayed in time. This central “lag effect,” with a transient cartilaginous area, may explain why many nonunions occur in the central zone of a fusion mass.
A unique temporal and spatial pattern of osteoblast-related gene expression was observed in an RT/PCR analysis of RNA from the different zones of the fusion mass. A lag effect in gene expression that correlated with the previously observed lag effect in the histologic healing sequence was noted in the central zone as compared with the outer zones of the fusion. As with osteocalcin expression, the peak expression of all genes measured was seen in the central zone 1 to 2 weeks later than the peak in the outer zone. This is consistent with the peripheral to central healing pattern observed histologically for fusions using autogenous bone graft.
Expression of the mRNA of several bone morphogenetic proteins (BMPs) also was studied. In the peripheral zones, BMP-2 mRNA expression was increased during weeks 2 through 6, with peak expression in weeks 3 and 4 (40-fold increase). Whereas BMP-6 in the outer zones had a first peak on day 2 (54-fold) and a second peak (100-fold) during week 5, BMP-6 in the central zone showed an initial peak on day 2 (34-fold), but did not demonstrate the later peak. These findings suggest specific time patterns of expression and probably unique roles for each of the various BMPs during spine fusion. It appears that BMP-6 is somewhat unique in that its mRNA levels demonstrated the earliest peak and greatest relative increase of the BMPs studied. The lower level of BMP-6 expression in the central zone of the fusion mass is correlated with the delayed timing and smaller amount of bone formation in the central zone of the fusion. Thus, the predilection for nonunion in the central zone also is apparent at a molecular biologic level. 50
Models of Spine Nonunion
There are several models that can produce a higher rate of nonunion. Implanting only 50% of the normal volume of autograft lowered the fusion rate from 70% to 33% in the rabbit model. 39 Excessive spine motion (produced by lifting the rabbits out of their cage daily after surgery) decreased the fusion success rate from 58% to 14% (P = 0.04). 21 Moderate systemic nicotine exposure (equaling 1 to 2 packs of cigarettes per day) significantly decreased the rate of successful spine fusion from 56% to 0% (P = 0.02) in rabbits. 60 These results highlight the importance of recognizing that each healing environment has unique features, and that bone graft substitutes must be tested in the specific healing environments anticipated to predict efficacy. Ketorolac, commonly used as an intravenous analgesic during the postoperative period, facilitated the fourth nonunion model in the current study. Some nonsteroidal antiinflammatory drugs (NSAIDs) have been shown to inhibit fracture repair. 31,56,62,64 The current author determined that a standard dose (4 mg/kg/day) for 7 days caused a decrease in the successful fusion rate from 75% in the control group to 35% (P < 0.04) in the rabbits treated with ketorolac. 40
In summary, the biology of lumbar spine fusion healing in rabbits has been well characterized with autograft, and important mechanical, vascular, cellular, and molecular influences have been identified. Similar detailed understanding is needed for anterior interbody fusion healing, although this location may be less challenging because of increased bony surface area and compression loading. Bone graft quality, quantity, and host bed vascularity (decortication) are critical and must be assessed carefully by the surgeon. The choice of graft substitutes must be individualized on the basis of the aforementioned factors and the presence of potential local or systemic inhibitory factors.
Augmentation of Bone Healing in the Spine
It is estimated that approximately 500,000 bone graft procedures are performed in the United States each year. Of these bone-grafting procedures, the vast majority (≈50%) are spine fusions, followed by general orthopedic procedures and craniomaxillofacial procedures. This potentially represents a $0.4 to $2 billion per year market for the use of bone repair enhancers or bone graft substitutes (DataMonitor, personal communication).
The most common reason for performing a spinal arthrodesis is either instability (excessive motion) of a spine segment or a deformity that is at risk for progression. The vast majority of fusions are performed to treat degenerative disorders, and the most common site is the lumbar spine. Although spinal fusion is commonly attempted, nonunion is reported to occur in 5% to 45% of patients. 42,52,61,66 Although this may be a clinically disturbing statistic, it makes spine fusion an ideal venue for testing bone augmentation devices. It is far easier to show an improvement in the healing of spine fusion than in the healing of fracture repair, in which the normal healing is quite good. Another rationale for testing bone augmentation in spine fusion is the reality that there is frequently an inadequate supply of autogenous bone graft for performing multilevel spinal arthrodesis. In addition, the morbidity of iliac bone graft harvest is reported to be as high as 30%, with the most frequent complications including chronic donor site pain, infection, fracture, hematoma, and increased operative time and costs. 1,5,23,34,58 The current author estimated that the cost of an iliac crest bone graft harvest is approximately $1500 to $2500.
Currently, there are several competitive strategies either to augment healing or to replace autogenous bone graft in the spine. 9 The most commonly used strategy for healing enhancement is that of rigid internal fixation. The fixation devices range in cost from $3000 to $6000, and are reported to increase the efficacy or the successful fusion rate by approximately 10%. Biophysical stimulation is another strategy that has been pursued in spine fusion for some time, but without good clinical data to support its use. Currently, the costs estimated for electrical stimulation by either direct current or pulsed electromagnetic fields is approximately $4000 to $5000, and although prospective randomized blinded trials have not yet been completed, the maximum benefit in terms of an increased successful fusion rate is estimated to be approximately 10%. 47,48 Low-intensity ultrasound also has been proposed recently for accelerating the healing of long bone fractures, and some experimental evidence in animal models shows that this may also be a viable enhancement strategy for spine fusion. 27,29
Currently, a variety of potential bone graft extenders is available. Demineralized bone matrix (DBM) in a variety of forms, each with variable degrees of osteoinductivity, is available through numerous sources. In general, the DBM cost for augmenting a one-level spinal fusion may be as high as $1100, and the current data support the use of DBM as a bone graft extender only and not as a bone graft enhancer or substitute. 49 Similarly, various ceramics such as natural coral (calcium carbonate), coralline hydroxyapatite, or composites such as hydroxyapatite-tricalcium phosphate have been used on a limited basis. Although clinical studies are lacking, limited animal data suggest that these ceramics may be viable as bone graft extenders when mixed with autogenous bone graft, but that they are unlikely to serve as stand-alone bone graft substitutes. 11 Thus, the need for improved bone repair and augmentation in the area of spine fusion is evident. The remainder of this article focuses on issues pertaining to the selection of an acceptable bone graft substitute for patients undergoing various types of spine fusion. For the purposes of this discussion, a graft extender is considered to be a material that allows the use of less autogenous bone graft with the same end result, or one that allows a given amount of autogenous bone to be stretched over a greater area with the same success rate (Table 1). A bone graft enhancer is a device that when added to autogenous bone graft, increases the successful healing rate of autogenous bone graft, using either the usual amount of graft or a smaller amount of bone graft. Finally, a bone graft substitute is a material that may be used entirely in place of autogenous bone graft to achieve the same or a better fusion success rate.
Because prospective randomized clinical trials that isolate the efficacy of individual bone graft substitutes in the spine do not exist, it can be challenging to garner information from the published animal and clinical studies. Several overriding principles should be kept in mind. The mechanism and timing of bone healing may vary considerably depending on the region of the spine. The anterior or middle column of the spine is primarily cancellous bone and load bearing, whereas the posterior column of the spine has a greater combination of cortical bone and a submuscular healing environment frequently under tension. Thus, there is considerable variation in the type of healing and speed of healing because of these biologic and biomechanical differences in healing environment. The three primary locations are the anterior interbody, the intertransverse process, and the interlaminar–facet joint region. As a result of these regional differences, the dosage and the ideal delivery vehicle for bone graft substitutes may differ by location. Again, it cannot be overemphasized that the results of healing for bone graft substitutes or augmentation devices in one region of the spine cannot necessarily be extrapolated for other regions.
In addition to the region of the spine as a cause for differences in healing, the specific patient diagnosis may give rise to variation in healing success. One example is lower lumbar spinal fusion for degenerative spondylolisthesis. This may be associated with increased motion at the intervertebral disc space to be fused, which decreases the rate of successful fusion. Alternatively, a patient with degenerative lumbar disc disease and disc collapse may actually have decreased disc motion, which may result in an increased likelihood of obtaining a solid fusion. The risks of neurologic complication, which is the most serious complication resulting from spine surgery, also is dependent on location. Cervical and thoracic spine fusions, particularly in the anterior column, carry some risks of spinal cord injury. In contrast, the structure at risk in the lumbar spine generally is the cauda equina, which is a more tolerant neurologic structure.
Another issue unique to spine fusion is the increased difficulty in the actual assessment of intervention efficacy. It is not always obvious when a solid fusion has been obtained, as opposed to a pseudarthrosis or nonunion. Unfortunately, the vast majority of noninvasive evaluation methods are inaccurate. It is reported that plain radiographs carry an accuracy no greater than 70% for predicting whether a spine fusion is solid. 16,18,28,52 Although the current author believes empirically that CT scanning is a more accurate method, there are no good data available to qualify this increased accuracy in the assessment of spine fusion. Another confounding variable involves the frequent dissociation between measurement of an augmentation device’s success (i.e., solid spine fusion) and clinical outcome (e.g., decreased pain or increased function). This dissociation is particularly germane to spine fusion, and is substantiated further by the lack of clinical consensus as to the appropriate indications for spine fusion. 24
Selecting a Bone Graft Substitute: The Burden of Proof
The donor site morbidity, limited supply, and imperfect success rate of autogenous bone graft has led to the commercialization of a variety of bone graft substitutes. These graft alternatives have undergone varying degrees of regulatory scrutiny, so their true safety and effectiveness in patients may not be known before their use by orthopedic surgeons. Moreover, given the different categories of bone graft substitutes and the seemingly similar products within each category, product differentiation has become increasingly varied for the practicing spine surgeon. This section is intended to suggest a framework of principles for considering the burden of proof necessary to allow clinicians to make reasonable choices when deciding on osteoinductive bone graft substitutes.
Osteoconductive substitutes may be used as scaffolds to deliver bone growth factors. These materials facilitators for bone formation, but do not contain growth factors that can attract osteoblast precursor cells and initiate osteoblast differentiation. Osteoconductive materials include collagen, ceramics, polymers, and coral, whereas osteoinductive substitutes contain one or more growth factors such as bone morphogenetic proteins (BMPs) capable of independently attracting precursor cells and inducing new bone formation. Whereas numerous growth factors may be involved in new bone formation including platelet derived growth factor (PDGF), transforming growth factor-beta (TGF-β), and vascular endothelial growth factor (VEGF), evidence suggests that only the BMPs are capable of initiating the entire process of new bone formation. 6,7,9,13,36 Lower doses of the BMPs are contained in demineralized bone matrix. They can be found in higher doses as recombinant (genetically engineered) proteins or purified extracts from bone. These materials are being studied in prospective trials and should soon be available for selected uses.
For understanding the burden of proof, the first principle to consider is that different healing environments (e.g., metaphyseal defect, long bone fracture, interbody spine fusion, and posterolateral spine fusion) have increasing levels of difficulty in forming new bone. For example, a metaphyseal defect permits the successful use of many purely osteoconductive materials. In contrast, a posterolateral spine fusion environment generally does not tolerate the use of purely osteoconductive materials as stand-alone substitutes, and only sometimes permits their use as bone graft extenders. Thus, validation of any bone graft substitute in one clinical anatomic site may not be predictive of its performance in another location. Accordingly, surgeons must be very careful in using graft substitutes that have not been tested definitively in the particular healing environment wherein they are about to use it. Most of the osteoconductive substitutes used currently in the spine have been validated only in nonspine locations.
The second principle to consider is the burden of proof required from preclinical studies to justify the use of an osteoinductive graft material or the choice of one brand over another. An osteoinductive material, by definition, must contain one or more growth factors including BMPs that can induce de novo ectopic bone formation through differentiation of progenitor cells into osteoblasts. Although not commonly recognized, evidence clearly suggests that it is much harder to make bone in humans than in cell culture or rodent models. 8 It is generally accepted that in lower vertebrates, bone heals more readily than in humans. Some osteoinductive materials (e.g., DBM and rhBMP-2) have been successful in lower vertebrates, but have failed to achieve the same result in primates. 2–4 Findings also have shown the rhesus monkey also to be an acceptable model for human bone remodeling and osteoporosis. 32,33,37,38,53,59
Only human trials can determine the efficacy of osteoinductive materials in humans. However, a hierarchy of testing credibility exists, ranging from cell culture assays of bone markers and the rat ectopic bone induction assay, as the least predictable tests, to rabbit ulnar defect, rabbit spine fusion, and large animal long bone–spine fusion, which have intermediate predictability of efficacy in humans (Table 2). Nonhuman primate long bone, interbody spine, and posterolateral spine fusion models offer the highest challenge and therefore the highest predictability of success in humans. 14,41 It is most important to understand that failure of an osteoinductive material to induce bone in any of these progressively challenging models usually means that it will fail at all higher levels. Unfortunately, successful bone induction at any level does not infer success at the next most stringent level. Only testing at the next level can answer that question. Thus, when differentiating between brands of similar osteoinductive products that have not been validated in human clinical trials, the surgeon should strongly consider choosing the product that satisfied the highest burden of proof in preclinical studies within the aforementioned hierarchy.
The third principle involving the burden of proof specifically pertains to products not currently subject to high levels of regulation such as demineralized bone matrix (DBM) and platelet gels containing “autologous growth factors.” Such products are considered to involve minimal manipulation of cells or tissue, and thus are regulated as tissue rather than devices. As a result, no standardized burden of proof level for safety or effectiveness required is before these products are marketed and used in human patients. Although it may be true that such products meet the technical definition of “minimal manipulation,” thereby escaping strict regulation, DBM and “autologous growth factor” platelet concentrates are being marketed and used for their biologic (osteoinductive) effects, not as simple tissue transplants or means of augmentation. Consequently, there is a risk that products or certain brands of products may not produce the expected result in humans when there has been little or no animal testing in the relevant and more challenging models. Many investigators in this field believe that minimal standards demonstrating some level of osteoinductivity (bone induction) should be developed for such products. If surgeons do not demand this soon, or if the manufacturers do not institute this in the near future, these products may risk being more highly regulated by the FDA, making them less accessible to surgeons and patients.
- A basic understanding of the biology of healing in different types of spine fusions and the differences between different categories of bone graft substitutes can help surgeons organize the graft selection process.
- In general, purely osteoconductive substitutes are less effective in adult posterolateral spine fusions, but may be suitable in the anterior spine when it is rigidly immobilized.
- Osteoinductive substitutes are more likely to be successful as extenders, enhancers, or substitutes for posterolateral spine fusion.
1. Arrington ED, Smith WJ, Chambers HG, et al. Complications of iliac crest bone graft harvesting. Clin Orthop 1996; 329: 300–9.
2. Aspenberg P, Lohmander LS, Thorngren K-L. Failure of bone induction by bone matix in adult monkeys. J Bone Joint Surg Br 1988; 70: 625–7.
3. Aspenberg P, Turek T. BMP-2 for intramuscular bone induction: Effect in squirrel monkeys is dependent on implantation site. Acta Orthop Scand 1996; 67: 3–6.
4. Aspenberg P, Wang E, Thorngren K-G. Bone morphogenetic protein induces bone in the squirrel monkey, but bone matrix does not. Acta Orthop Scand 1992; 63: 619–22.
5. Banwart JC, Asher MA, Hassanein RS. Iliac crest bone graft harvest donor site morbidity: A statistical evaluation. Spine 1995; 20: 1055–60.
6. Boden SD. Bioactive factors for bone tissue engineering. Clin Orthop 1999; 367 (Suppl): S84–94.
7. Boden SD. Clinical application of the BMPs. J Bone Joint Surg Am 2001; 83 (Suppl 1): S161.
8. Boden SD, McCuaig K, Hair G, et al. Differential effects and glucocorticoid potentiation of bone morphogenetic protein action during rat osteoblast differentiation in vitro.
Endocrinology 1996; 137: 3401–7.
9. Boden SD, Schimandle JH. Biologic enhancement of spinal fusion. Spine 1995; 20: 113S–23S.
10. Boden SD, Schimandle JH, Hutton WC. An experimental lumbar intertransverse process spinal fusion model: Radiographic, histologic, and biomechanical healing characteristics. Spine 1995; 20: 412–20.
11. Boden SD, Schimandle JH, Hutton WC, et al. In vivo
evaluation of a resorbable osteoinductive composite as a graft substitute for lumbar spinal fusion. J Spinal Disord 1997; 10: 1–11.
12. Boden SD, Schimandle JH, Hutton WC, et al. The use of an osteoinductive growth factor for lumbar spinal fusion: Part I. The biology
of spinal fusion. 1995 Volvo Award in Basic Sciences. Spine 1995; 20: 2626–32.
13. Boden SD, Sumner DR. Biologic factors affecting spinal fusion and bone regeneration. Spine 1995; 20: 102S–12S.
14. Boden SD, Zdeblick TA, Sandhu HS, et al. The use of rhBMP-2 in interbody fusion cages: Definitive evidence of osteoinduction
in humans: A preliminary report. Spine 2000; 25: 376–81.
15. Bridwell KH, Sedgewick TA, O’Brien MF, et al. The role of fusion and instrumentation in the treatment of degenerative spondylolisthesis with spinal stenosis. J Spinal Disord 1993; 6: 461–72.
16. Brodsky AE, Kovalsky ES, Khalil MA. Correlation of radiographic assessment of lumbar spine fusions with surgical exploration. Spine 1991; 16: S261–5.
17. Conaty JP, Mongan ES. Cervical fusion in rheumatoid arthritis. J Bone Joint Surg Am 1981; 63: 1218–27.
18. Dawson EG, Clader TJ, Bassett LW. A comparison of different methods used to diagnose pseudarthrosis following posterior spinal fusion for scoliosis. J Bone Joint Surg Am 1985; 67: 1153–9.
19. DePalma AF, Rothman RH. The nature of pseudarthrosis. Clin Orthop 1968; 59: 113–18.
20. Farey ID, McAfee PC, Gurr KR, et al. Quantitative histologic study of the influence of spinal instrumentation on lumbar fusions: A canine model. J Orthop Res 1989; 7: 709–22.
21. Feiertag MF, Boden SD, Schimandle JH, et al. A rabbit model for nonunion of lumbar intertransverse process spine arthrodesis. Spine 1996; 21: 27–31.
22. Ferlic DC, Clayton ML, Leidholt JD, et al. Surgical treatment of the symptomatic unstable cervical spine in rheumatoid arthritis. J Bone Joint Surg Am 1975; 57: 349–54.
23. Fernyhough JC, Schimandle JH, Weigel MC, et al. Chronic donor site pain complicating bone graft harvesting from the posterior iliac crest for spinal fusion. Spine 1992; 17: 1474–80.
24. Fischgrund JS, Mackay M, Herkowitz HN, et al. Degenerative lumbar spondylolisthesis with spinal stenosis: A prospective, randomized study comparing decompressive laminectomy and arthrodesis with and without spinal instrumentation. Spine 1997; 22: 2807–12.
25. Fischgrund JS, Mackay M, Herkowitz HN, et al. Volvo Award winner in clinical studies: Degenerative lumbar spondylolisthesis with spinal stenosis: A prospective, randomized study comparing decompressive laminectomy and arthrodesis with and without spinal instrumentation. Spine 1997;1997; 22: 2807–12.
26. France JC, Yaszemski MJ, Lauerman WC, et al. A randomized prospective study of posterolateral lumbar fusion: Outcomes with and without pedicle screw instrumentation. Spine 1999; 24: 553–60.
27. Glazer PA, Heilmann MR, Lotz JC, et al. Use of electromagnetic fields in a spinal fusion: A rabbit model. Spine 1997; 22: 2351–6.
28. Hannon KM, Wetta WJ. Failure of technetium bone scanning to detect pseudarthrosis in spinal fusion for scoliosis. Clin Orthop 1977; 123: 42–4.
29. Heckman JD, Ryaby JP, McCabe J, et al. Acceleration of tibial fracture healing by noninvasive, low-intensity pulsed ultrasound. J Bone Joint Surg Am 1994; 76: 26–34.
30. Herkowitz HN, Kurz LT. Degenerative lumbar spondylolisthesis with spinal stenosis: A prospective study comparing decompression with decompression and intertransverse process arthrodesis. J Bone Joint Surg Am 1991; 73: 802–8.
31. Keller J, Bunger C, Andreassen TT, et al. Bone repair inhibited by indomethacin: Effects on bone metabolism and strength of rabbit osteotomies. Acta Orthop Scand 1987; 58: 379–83.
32. King FA, Yarbrough CJ. Medical and behavioral benefits from primate research. Physiologist 1985; 28: 75–87.
33. King FA, Yarbrough CJ, Anderson DC, et al. Primates. Science 1988; 240: 1475–82.
34. Laurie SWS, Kaban LB, Mulliken JB, et al. Donor-site morbidity after harvesting rib and iliac bone. Plast Reconstr Surg 1984; 73: 933–8.
35. Lindholm TS, Ragni P, Lindholm TC. Response of bone marrow stroma cells to demineralized cortical bone matrix in experimental spinal fusion in rabbits. Clin Orthop 1988; 230: 296–302.
36. Ludwig SC, Boden SD. Osteoinductive bone graft substitutes
for spinal fusion: A basic science summary. Orthop Clin North Am 1999; 30: 635–45.
37. Mann DR, Gould KG, Collins DC. A potential primate model for bone loss resulting from medical oophorectomy or menopause. J Clin Endocrinol Metab 1990; 71: 105–10.
38. Mann DR, Rudman GG, Akinbami MA, et al. Preservation of bone mass in hypogonadal female monkeys with recombinant human growth hormone administration. J Clin Endocrinol Metab 1992; 74: 1263–9.
39. Martin G, Boden SD, Morone MA, et al. New formulations of demineralized bone matrix as a more effective graft alternative in experimental posterolateral lumbar spine arthrodesis. Spine 1999; 24: 637–45.
40. Martin G, Boden SD, Titus L. Recombinant human bone morphogenetic protein-2 reverses the inhibitory effect of ketorolac, a nonsteroidal antiinflammatory drug (NSAID) on posterolateral lumbar intertransverse process spine fusion. Spine 1999; 24: 2188–94.
41. Martin GJ, Boden SD, Morone MA, et al. Posterolateral intertransverse process spinal fusion arthrodesis with rhBMP-2 in a nonhuman primate: Important lessons learned regarding dose, carrier, and safety. J Spinal Disord 1999; 12: 179–86.
42. May VR, Mauck WR. Exploration of the spine for pseudarthrosis following spinal fusion in the treatment of scoliosis. Clin Orthop 1967; 53: 115–22.
43. McAfee PC, Farey ID, Sutterlin CE, et al. Device-related osteoporosis with spinal instrumentation. Spine 1989; 14: 919–26.
44. McAfee PC, Farey ID, Sutterlin CE, et al. The effect of spinal implant rigidity on vertebral bone density: A canine model. Spine 1991; 16: S190–7.
45. McAfee PC, Regan JJ, Farey ID, et al. The biomechanical and histomorphometric properties of anterior lumbar fusions: A canine model. J Spinal Disord 1988; 1: 101–10.
46. McGuire RA, Amundson GM. The use of primary internal fixation in spondylolisthesis. Spine 1993; 18: 1662–72.
47. Meril AJ. Direct current stimulation of allograft in anterior and posterior lumbar interbody fusions. Spine 1994; 19: 2393–8.
48. Mooney V. A randomized double-blind prospective study of the efficacy of pulsed electromagnetic fields for interbody lumbar fusions. Spine 1990; 15: 708–12.
49. Morone MA, Boden SD. Experimental posterolateral lumbar spinal fusion with a demineralized bone matrix gel. Spine 1998; 23: 159–67.
50. Morone MA, Boden SD, Martin G, et al. Gene expression
during autograft lumbar spine fusion and the effect of bone morphogenetic protein-2. Clin Orthop 1998; 351: 252–65.
51. Muschler GF, Huber B, Ullman T, et al. Evaluation of bone grafting materials in a new canine segmental spinal fusion model. J Orthop Res 1993; 11: 514–24.
52. O’Beirne J, O’Neill D, Gallagher J, et al. Spinal fusion for back pain: A clinical and radiological review. J Spinal Disord 1992; 5: 32–8.
53. Pope NS, Gould KG, Anderson DC, et al. Effects of age and sex on bone density in the rhesus monkey. Bone 1989; 10: 109–12.
54. Ragni PC, Lindholm TS. Bone formation and static changes in the thoracic spine at uni- or bilateral experimental spondylodesis with demineralized bone matrix (DBM). Ital J Orthop Traumatol 1989; 15: 237–52.
55. Ragni PC, Lindholm TS, Lindholm TC. Vertebral fusion dynamics in the thoracic and lumbar spine induced by allogenic demineralized bone matrix combined with autogenous bone marrow: An experimental study in rabbits. Ital J Orthop Traumatol 1987; 13: 241–51.
56. Ro J, Sudmann E, Marton PF. Effect of indomethacin on fracture healing in rats. Acta Orthop Scand 1976; 47: 588–99.
57. Schimandle JH, Boden SD. The use of animal models
to study spinal fusion. Spine 1994; 19: 1998–2006.
58. Schnee CL, Freese A, Weil RJ, et al. Analysis of harvest morbidity and radiographic outcome using autograft for anterior cervical fusion. Spine 1997; 22: 2222–7.
59. Schnitzler CM, Ripamonti U, Mesquita JM. Histomorphometry of iliac crest trabecular bone in adult male baboons in captivity. Calcif Tissue Int 1993; 52: 447–54.
60. Silcox DH, Daftari T, Boden SD, et al. The effect of nicotine on spinal fusion. Spine 1995; 20: 1549–53.
61. Steinmann JC, Herkowitz HN. Pseudarthrosis of the spine. Clin Orthop 1992; 284: 80–90.
62. Sudmann E, Dregelid E, Bessesen A, et al. Inhibition of fracture healing by indomethacin in rats. Eur J Clin Invest 1979; 9: 333–9.
63. Toribatake Y, Hutton WC, Boden SD, et al. Revascularization of the fusion mass in a posterolateral intertransverse process fusion. Spine 1998; 23: 1149–54.
64. Tornkvist H, Lindholm S. Effect of ibuprofen on mass and composition of fracture callus and bone. Scand J Rheumatol 1980; 9: 167–71.
65. West JL III, Bradford DS, Ogilvie JW. Results of spinal arthrodesis with pedicle screw plate fixation. J Bone Joint Surg Am 1991; 73: 1179–84.
66. Zdeblick TA. A prospective, randomized study of lumbar fusion: Preliminary results. Spine 1993; 18: 983–91.
67. Zerwekh JE, Kourosh S, Scheinberg R, et al. Fibrillar collagen–biphasic calcium phosphate composite as a bone graft substitute for spinal fusion. J Orthop Res 1992; 10: 562–72.
68. Zoma A, Sturrock RD, Fisher WD, et al. Surgical stabilisation of the rheumatoid cervical spine: A review of indications and results. J Bone Joint Surg Br 1987; 49: 8–12.