Autologous Cancellous Bone

Autologous cancellous bone has traditionally been considered to be the “gold standard” of graft materials. Autograft has maintained a track record as the most reliable and effective graft material, particularly in the challenging clinical setting of spinal fusion.^44–47 This finding is reinforced by the recognition that an autogenous cancellous graft provides all three areas of functionality: osteogenic bone and marrow cells; an osteoconductive matrix of collagen, mineral, and matrix proteins; and a spectrum of osteoinductive proteins provided within the transplanted matrix of cells.

However, the limitations and important disadvantages of autogenous cancellous bone have become increasingly evident in recent years as more effective and less morbid grafting options have become available.

The principle disadvantage of autogenous cancellous bone graft relates to the process of graft harvest. Autograft harvest adds operative time, pain, and blood loss, and it carries an increased risk of infection, cutaneous nerve damage, and even local fracture. Autograft harvest leaves the patient with permanent scars and a risk of long-term pain at the graft site. Increased blood loss attributable to the graft harvest results in an increased potential exposure to blood products along with all the associated costs and risks of transfusion. The incidence of major complications associated with the harvest of iliac crest bone graft has been reported to be 5% to 10%.^48–50 The time, effort, and complications associated with autograft harvest are also associated with tangible costs to the health care system, which have been estimated in the range of $700 to $2200, not including the pain and scars. A second disadvantage of autograft is that, in addition to its cost and morbidity, the amount of autogenous bone is limited and may be insufficient in many settings, particularly in children undergoing arthrodesis over multiple segments.

Finally, autogenous bone has a biologic limitation as a cellular graft. Although cancellous bone from the pelvis is the most abundant source of osteoblastic stem cells and progenitors, these cells represent only about 1 in 20,000 cells in normal bone marrow, which is packed with many other cells that do not necessarily contribute to bone healing. When autogenous cancellous bone is harvested, its diverse mixture of highly metabolic cells is dissociated from its blood supply and is then implanted under conditions where it must compete with all of the other cells in bone and marrow for the limited amount of oxygen and other nutrients that are available to diffuse into the graft site. The metabolic demand within the graft site far exceeds the capacity for nutrient diffusion. This results in profound hypoxia as one moves more deeply into the graft. As a result, only cells within 1 to 2 mm of the surface of an autograft are able to survive transplantation. This fact was recognized by Burwell from histologic assessment in the 1960s and can now be defined in more quantitative terms on the basis of chemical engineering principles.⁴¹ Necrosis within the graft then places an additional burden on the site. Necrotic debris must be removed before new bone formation can occur. In addition, cell debris and the cytokines that are released by dying cells escalate the local inflammatory response, bringing in additional cells that further increase local metabolic demand.

Autologous Cortical Bone

Cortical bone grafts are less biologically active than cancellous grafts as a result of several factors. Cortical bone contains fewer osteoblastic stem cells and progenitors than trabecular bone. Furthermore, the cells that are present are less likely to survive because a larger fraction of cells are located within osteons, which are buried in the matrix, where diffusion is insufficient to provide adequate nutrients to support viability. The absence of nonosteogenic marrow cells and endothelial cells further limits the biologic potential of a cortical graft. In addition, in contrast to trabecular bone, cortical grafts have a much lower available surface area per unit volume. This reduces the potential surface for new bone formation (and therefore osteoconductive potential) and may also reduce the bioavailability of osteoinductive factors buried in the matrix. The marked reduction in porosity of cortical bone also represents a barrier to vascular ingrowth and bony remodeling, both of which are critical to bone healing and the development of optimal mechanical strength.

The only advantage of cortical bone versus cancellous bone and other graft materials is its superior mechanical strength and the availability of cortical segments of sufficient size to fill virtually any skeletal defect. The ability to provide immediate mechanical strength at the time of implantation is a critical advantage in many situations, particularly in anterior interbody fusions. However, the mechanical strength of a cortical graft is not constant over time. Allograft bone is remodeled by the process of creeping substitution, resulting in increased porosity and progressive loss of strength during the first 12 to 24 months after implantation before remodeling and new bone formation reconstitute the mechanical properties of the grafted segment.^40,51,52 This is associated with increased risk of graft failure and collapse during the first 24 months after implantation.

Combined grafts consisting of intact cortical and cancellous bone from the iliac crest are common and readily available graft materials with good mechanical properties and biologic properties of incorporation. The mechanical strength of these grafts is variable, however. Grafts from the anterior crest exhibit greater mechanical compressive strength than grafts from the posterior crest.⁵³

Vascularized Autologous Grafts

Vascularized grafts are used extensively in many centers for musculoskeletal reconstructive procedures. High rates of vascular patency can be achieved by experienced microsurgeons. Many studies have shown clear advantages to using vascularized grafts in a number of settings.^54–57 In anterior spinal fusions, donor vessels are available to support the vascularized graft. Suitable grafts with good mechanical strength are available from the anterior iliac crest, posterior iliac crest,⁵⁸ fibula,⁵⁹ or rib.⁶⁰ In addition, an iliac graft pedicle flap on quadratus lumborum has also been described.⁶¹ In intrathoracic procedures, a vascularized rib graft may be mobilized on its intercostal pedicle, with limited additional morbidity and in much less time than a free vascularized graft.^62,63 However, a rib graft provides less mechanical strength when compared with the iliac crest or fibula⁶⁴ and therefore must be mechanically supplemented by additional cortical bone or internal fixation.

Although routine use of vascular grafts is limited by their increased operative time, technical difficulty, and added morbidity, the improved incorporation of these grafts may make them highly desirable in some settings where incorporation of avascular grafts may be compromised, such as in areas of radiation-induced fibrosis or previous infection.

Autologous Bone Marrow

Bone marrow is a valuable and easily accessible source of osteogenic cells that is probably underused in contemporary clinical practice. The osteogenic potential of transplanted bone marrow was first documented by Goujon in 1869,⁶⁵ and later by Senn in 1889.⁶⁶ Studies by Burwell in the 1960s concluded that the formation of new bone following autografting resulted from the differentiation of osteogenic precursor cells contained within the marrow, in addition to osteoblasts on the surface of the graft material itself.⁶⁷ Burwell postulated that following transplantation, these reticular cells free themselves from the sinusoidal walls to become primitive migratory cells; they then differentiate into osteogenic cells when they are exposed to osteoinductive substances released from the necrotic portion of the graft,⁶⁸ or perhaps from osteoinductive materials contained within or secreted by the marrow itself.

Many studies have demonstrated the ability of marrow cells to form bone intramuscularly,^69,70 subcutaneously,⁷¹ interperitoneally,^72–74 in the anterior chamber of eye,⁷⁵ and orthotopically. Using a suspension of marrow cells in diffusion chambers, Friedenstein showed that hematopoietic cells die following transplantation, whereas fibroblasts and other stromal elements are more resistant to hypoxia and may proliferate close to the surface of the graft to produce immature bone, suggesting the presence of an undifferentiated precursor cell in postnatal marrow.^71,74,76,77 It has been well established that the osteogenic cells that contribute to this response are not derived from one homogeneous population of cells, but rather appear to come from two or more compartments of stem cells or progenitor cells that are upstream from the osteoblastic population in bone.^43,71,78,79 A more mature preosteoblastic cell appears to be localizable to the marrow space adjacent to trabecular or osteonal bone. One or more populations of less mature and potentially multipotent cells appear to be present in the more liquid phase of bone marrow, possibly including cells associated with perivascular tissue of vascular sinusoids or other marrow vessels.

Osteoblastic differentiation proceeds in a series of steps, which can be conceptually divided into phases.^80,81 An initial proliferative phase is characterized by expression of H4 histone, c-fos, and c-jun. A matrix synthesis phase is characterized by a reduction in proliferation and upregulation of gene products for type I collagen, osteopontin, osteonectin, and alkaline phosphatase. Finally, a matrix mineralization phase culminates in an osteoblastic phenotype characterized by expression of osteocalcin, bone sialoprotein, and responsiveness to 1,25-dihydroxyvitamin D and parathyroid hormone. A conceptual summary of the large body of literature related to osteoblastic differentiation is presented in Figure 67–1.^43,82

FIGURE 67–1 The osteoblastic pathway. This figure schematically illustrates the stages of osteoblastic differentiation, the predominant activity of the differentiation cell at each stage, some of the characteristic genes expressed at each stage, and the approximate site of principal action for some of the principal osteotropic growth factors and hormones. AP, alkaline phosphatase; bFGF, basic fibroblast growth factor; BMPs, BMP family members; BSP, bone sialoprotein; E2, estradiol; EGF, epithelial growth factor; H4, H4 histone; IGFs, insulin-like growth factors I and II; IL-6, interleukin-6; ON, osteonectin; OP, osteopontin; PDGF, platelet-derived growth factor; TGF-β, transforming growth factor-beta; VDR, vitamin D receptor; Vit D, vitamin D₃.

The value of bone marrow as a bone graft, used alone or as a component in a composite bone graft material, has been supported by numerous studies in rats and rabbits.^{68,71,74–76,83–90} Lane and colleagues⁹¹ demonstrated the efficacy of autogenous bone marrow grafting in a 5-mm rat femoral defect and showed that the efficacy of bone marrow grafts were dependent on transplantation of viable cells. Yasko and colleagues⁹² also showed that bone marrow enhanced the performance of an effective synthetic BMP-2 material in rats, a finding consistent with prior reports on less pure BMP materials.⁸⁸ However, in contrast to clinical practice, almost all of these studies in rodents have used bone marrow obtained by open harvesting of bone and/or irrigation of bone explants, rather than by aspiration.

Increasing evaluation of bone marrow grafting has been carried out in larger nonrodent models. Johnson and colleagues⁹³ found canine bone marrow much less osteogenic than rabbit marrow when transplanted in diffusion chambers. Using a canine tibial model, Tiedeman and colleagues⁸⁹ found that the percutaneous injection of marrow mixed with demineralized bone matrix powder produced overall results comparable with open cancellous grafting.

A few uncontrolled clinical series also imply that aspirated bone marrow can improve bone healing.^68,94–97 Connolly reported successful treatment of 18 of 20 nonunions treated with casting or intramedullary nails plus percutaneous marrow injection. Healy and colleagues⁸⁵ reported healing in five of eight delayed or nonunions of allograft host junction sites using marrow injection alone.

Recognizing the potential biologic values, many surgeons currently use bone marrow as an adjuvant to allograft bone grafts. This practice is currently supported primarily because the risk and morbidity of bone marrow aspiration from the iliac crest is low. The prospective trials needed to document the value or limitations of bone marrow grafting are only now being organized. However, these are informed by a significant volume of clinical information regarding methods and cellular yield of bone marrow aspiration^98–100 and by a robust set of preclinical studies.^101–105

The method of bone marrow aspiration has a significant effect on the concentration and prevalence of bone marrow–derived osteogenic cells. Muschler and colleagues,¹⁰⁶ in a cohort of normal subjects undergoing elective orthopaedic procedures, showed that a mean of approximately 2100 osteoblastic progenitors (colony-forming units or CFU-Os) could be harvested in a 2-mL aspirate of human bone marrow from the iliac crest and that the mean prevalence of CFU-Os among nucleated marrow cells was approximately 1 in 37,000 cells. They further documented that the yield of CFU-Os harvested dropped rapidly as the volume of bone marrow aspirated was increased, owing to dilution with peripheral blood. On the basis of these findings, they recommended that aspiration of marrow be limited to 2 mL from each aspiration site in order to maximize the concentration of CFU-Os in the marrow graft. Further studies have demonstrated that the yield of osteoblastic stem cells and progenitors tends to decrease with age and that the prevalence of these cells may decrease more rapidly in women than in men.^{98,100,107–109} However, these data also show that there is marked variation from individual to individual in the cellularity of marrow and the prevalence of osteogenic cells that is not associated with age or gender.

McLain and colleagues¹¹⁰ obtained transpedicular aspirates from the vertebral bodies of 21 adults undergoing posterior lumbar arthrodesis and pedicle screw instrumentation. Aspirates were obtained from two depths within the vertebral body and were quantified relative to matched, bilateral aspirates from the iliac crest that were obtained from the same patient at the same time and served as a control. Aspirates of vertebral marrow demonstrated comparable or greater concentrations of progenitor cells compared with matched controls from the iliac crest. Progenitor cell concentrations were consistently higher than matched controls from the iliac crest (P = 0.05). The concentration of osteogenic progenitor cells was, on average, 71% higher in the vertebral aspirates than in the paired iliac crest samples (P = 0.05). With the numbers available, there were no significant differences relative to vertebral body level, the side aspirated, the depth of aspiration, or gender. An age-related decline in cellularity was suggested for the iliac crest aspirates. The authors concluded that the vertebral body is a suitable site for aspiration of bone marrow for graft augmentation during spinal arthrodesis.

One clinical grafting product, a collagen ceramic composite called Healos, has been released specifically for use as a delivery system for bone marrow–derived cells and has demonstrated strong clinical performance in terms of spinal fusion rates for both interbody and posterolateral fusion.^111,112

Several authors have addressed the potential value of harvesting bone marrow by aspiration and then processing the cells that are collected to concentrate those that are most likely to be of value, thus eliminating those that are not. Connolly and colleagues⁸³ described concentration of marrow-derived cells using centrifugation techniques and showed that a threefold to fourfold increase in the concentration of nucleated cells increased the amount of bone formation in a diffusion chamber in the rabbit. More recently, methods have been described that allow rapid intraoperative concentration and selection of osteoblastic stem cells and progenitors from bone marrow using an appropriately designed implantable allograft matrix as an affinity column to select osteoblastic cells on the basis of attachment behavior.¹¹³ This strategy has demonstrated increased bone formation, union rate, and mechanical performance in a validated canine posterior spinal fusion model. Performance of these concentrated grafts of marrow-derived cells appears comparable with that of autogenous cancellous bone, when cells are transplanted in an environment that includes a blood or marrow clot. The same strategy has been reported by Kadiyala and colleagues^102,103 demonstrating union of a 5-cm canine femoral defect with both a rate and an outcome equal to autogenous cancellous bone. An early clinical cohort study applying bone marrow concentration strategies to lumbar interbody fusion has also reported a fusion rate of 85%, comparable with historical autograft controls.¹¹⁴

Muschler and colleagues^41–43 have recently published three more detailed reviews of the biologic principles and practical strategies for harvest and use of stem cells and progenitor cells for bone healing applications. Other strategies such as the use of these cells as the delivery system or as the target cells for gene therapy applications are addressed in other chapters in this text.^115–122

Structural Allografts and Cages

Use of allograft bone has been well characterized over the past 30 years.¹²³ There are four principal advantages of allografts. First, they eliminate the morbidity associated with harvesting autologous bone. Second, and in contrast to autograft bone, the volume of available allograft is essentially unlimited. Third, because cortical allografts can be selected from any bone (not just the iliac crest or tibial hemicortex), they provide the surgeon with access to grafts that have mechanical strength and options for shaping that are superior to any autograft site. Fourth, allograft bone can be preprocessed into a wide range of specialized physical forms (e.g., blocks, threaded or nonthreaded dowels, sized rings and wedges, chips, fibers, powder) prepared from cortical and/or cancellous bone. These preprocessed grafts provide opportunity to customize and precertify the physical form and architectural properties (shape, size, mechanical strength, surface area, porosity) of a graft matrix to an individual site.

The method of sterilization and preparation of allograft tissue has a significant impact on osteoconductive, osteoinductive, and mechanical properties, as well as immunogenicity.^124–127 Donor cells and cell fragments are the most immunogenic material in allogenic bone. Processing of allograft bone therefore includes steps that attempt to remove as many cells as possible from the graft. Immunogenicity is further reduced, although not eliminated, by freezing to –20° C.^128–136 Freeze-drying is even more effective at reducing the immunogenicity of allogenic bone, but at the price of reducing mechanical strength by 50%.^130,137

Using contemporary processing and storage techniques, clinical evidence of overt immunologic reaction against the graft is rare. Even so, histologic evidence of a low-grade inflammatory reaction can be found around essentially all allografts. This reaction probably slows the incorporation of many allografts and may contribute to the failure of some, as suggested by several canine studies that have documented improved biologic behavior in antigen-matched allografts.^{125,126,137,138} Antigen matching is not currently considered practical in the clinical setting, however. The relatively high current success rates for allografts makes the large cost of antigen matching not feasible and is probably unwarranted in general practice.¹³⁹

Sterility of frozen allografts is ensured through expedient postmortem harvesting using sterile surgical technique and careful monitoring using surface cultures and recently polymerase chain reaction (PCR) screening for bacterial and viral genome fragments. The current risk of disease transmission via a musculoskeletal allograft is approximately 1 in 1,667,000.^140,141 A variety of secondary sterilization procedures have been designed and may be used, depending on the source of the allograft. Ethylene oxide sterilization was evaluated by Cornell,¹⁴² who found a 70% decrease in bone induction by demineralized bone powder in rats. Other authors have reported variable changes in inductive capacity of ethylene oxide sterilized matrix.^143–145 Heating or autoclaving bone tissue is generally avoided due to their disruption of matrix proteins. Some processing techniques such as high-dose irradiation compromise both the biologic potential of bone matrix, reducing bone formation and union rates,^146,147 and also alter the mechanical properties of the graft.^148–155 For example, irradiation to 2.5 megarads or freeze-dried processing may reduce the torsional strength of the cortical allograft by as much as 50%.¹³⁷

Results from other clinical and experimental studies using allograft bone alone in spinal fusions have been mixed. Some investigators have found allograft to be significantly inferior to autogenous bone grafting when compared with other distinctive preparation methods,^{36,38,149,156–167} whereas others find little or no difference between them.^{36,62,168–182} Allograft bone appears to be particularly valuable in settings that require the graft to serve a significant mechanical function such as struts or ring allografts for anterior lumbar interbody fusions^183–187 or as struts or bone-wire fixation constructs in the upper cervical spine.^188–197 In these settings, allografts have essentially replaced the use of autografts from rib, fibula, tricortical iliac crest, and tibial hemicortex grafts, which are all associated with significant donor site morbidity.

In the past decade options for structural materials that can be incorporated into interbody fusion sites have expanded to include the use of structural cages composed of titanium or carbon fiber.^197–205 These materials are capable of providing the structural function that is necessary to maintain the height and stability of the interbody site. These cages and wedges lack some of the osteoconductive capacity of allograft bone and the capacity of allograft bone to be biologically incorporated into the fusion mass and remodeled, remaining in the site essentially as a permanent foreign body. Metal cages also make radiographic assessment of the fusion site more difficult. However, in comparison with allograft, metal cages do offer the advantage of more consistent material properties, specifically strength and fracture resistance.

Regardless of material used for structural support, there is relatively general consensus that in order to achieve optimal rates of spinal fusion, the environment within and around these structural allografts or cages should be further supplemented with other osteogenic, osteoconductive, and/or osteoinductive graft materials. Autogenous cancellous bone, bone marrow aspirate,²⁰⁶ processed bone marrow–derived cells,¹¹⁴ processed nonstructural allograft materials, and BMPs^3,207,208 are most commonly considered.

Demineralized Allograft Bone Matrix

Demineralization is one means by which the biologic activity of allograft bone can be modified. In settings where mechanical properties are not critical to the graft application, demineralization may enhance the osteoinductive activity of the allograft matrix, by making growth factors embedded in the matrix more available. Several preparations of demineralized bone matrix (DBM) are available; however, these materials may vary in biologic activity, as discussed later.

The history of demineralization as a means to enhance allograft performance is richly linked to many of the recent biologic insights into bone biology and bone healing. It was more than a century ago, in 1889, that Senn reported the repair of long bone and cranial defects in patients with chronic osteomyelitis using hydrochloric acid-treated decalcified heterologous bone implants.⁶⁶ Although his primary motive was to promote antisepsis within the bone cavities, Senn observed rapid substitution of the demineralized tissue with new bone formation invading from the perimeter of the defects. However, several of Senn’s contemporaries obtained equivocal results, and clinical efforts over the next 70 years were minimal.^{188,209–211}

Reddi and Huggins^212,213 revived this concept when they reported on their observation that matrix induced the bone induction phenomenon in rats. Urist went on to demonstrate bone induction using a variety of demineralized matrix preparations in muscular pouches of rabbits, rats, mice, and guinea pigs.²¹⁴ Subsequently, matrix-induced heterotopic bone formation was documented at many soft tissue sites including muscle, tendon, and fascia,^{181,213,315-219} as well as in the thymus⁷¹ and soft connective tissue of visceral organs.²²⁰ Nathanson also observed the differentiation of neonatal embryonic skeletal tissue into cartilage when cultured on demineralized bone matrix substratum and suggested that the tissue transformation of bone induction was analogous to embryonic bone tissue differentiation.^221,222

Reddi subsequently characterized the inductive phenomenon of bone matrix as a cascade of events parallel to those occurring in endochondral ossification and postulated that the process was the result of stimulation by a series of soluble matrix factors that potentiated events along the cascade.^223–225 In this paradigm, bioactive factors in bone matrix stimulate activation and migration of osteogenic stem cells and progenitor cells. Mitogenic factors promote cell proliferation. Angiogenic factors promote local revascularization, and osteoinductive factors promote osteoblastic differentiation. Subsequently, bone matrix has been shown to contain a rich variety of growth factors and other bioactive molecules in concentrations that are bioactive.^226,227

Sato and Urist²²⁸ showed that demineralized bone matrix was both inductive and was synergistic with bone marrow–derived cells in healing of rat femoral defects. He went on to provide a clinical outlet for these discoveries by developing a “chemosterilized, autolyzed, antigen-extracted allogenic bone (AAA),” prepared using chloroform-methanol extraction, 0.6 N hydrochloric acid extraction of soluble proteins with partial demineralization, and neutral phosphate autodigestion.^215–217219 This preparation appeared to reduce the immunogenicity of the allograft matrix without loss of inductive properties. Using this preparation, Urist reported on 40 patients undergoing posterolateral lumbar spinal fusion with an 80% success rate and a pseudoarthrosis rate of 12%.²²⁹

The value of a variety of demineralized bone matrix preparations (chips, fibers, powders) has subsequently been described in a number of settings by a series of authors. Glowacki and colleagues^230–232 were among the first to report successful repair of craniofacial defects. Tiedeman and colleagues⁸⁹ and Wilkins and colleagues²³³ both reported clinical efficacy in long bone defects. Other reports have also shown efficacy for demineralized bone matrix in spine fusion models.^234,235 Some studies have reported a benefit of adding demineralized bone matrix to autograft or ceramic matrices in animal spine fusion models.^87,236–240 There is also recent evidence that some demineralized bone matrix preparations can be used as a substrate for selective attachment and concentration of bone marrow–derived osteogenic stem cells and progenitors.^113,114

Because demineralized bone preparations do not have mechanical properties sufficient to resist external forces, they can easily be dislodged from a graft site. The use of these materials is therefore primarily restricted to grafting of contained defects or graft sites that are protected by rigid internal fixation or to settings in which demineralized matrix is combined with other allograft or synthetic materials that resist displacement or compression.

Of recent clinical concern is evidence that suggests that the biologic efficacy of commercially available preparations of demineralized bone matrix materials can vary significantly, depending on the method of processing, the individual batch of bone that is processed, or the donor of the bone that is used. This variation has raised questions about the possible value of implementing generalized standards for either in vitro or in vivo biologic assays for bone from each batch and donor in order to limit the potential for biologically deficient materials from compromising the performance of clinical grafting procedures.^123,241,242 At present, without evidence that failed graft procedures cluster around individual donors or batches of demineralized bone, and that these failures can be predicted by any of the available assays, the clinical value and cost effectiveness of biologic assays is uncertain. Regardless, some providers of demineralized bone have elected to implement some form of bioassay and use these data to exclude some bone from use and to make claims of superior or more reliable performance in the marketplace.

Deproteinated Heterologous Bone

In contrast to allograft bone, heterologous bone (xenograft) fails to induce osseous repair due to its high level of antigenicity. Partially deproteinated and partially defatted heterologous bone (Kiel bone or Oswestry bone) does exhibit greatly reduced antigenicity and therefore evokes a minimal immune response.²⁴³ The denaturing process, however, also destroys osteoinductive matrix proteins. Accordingly, implantation of such materials in bone defects and muscular compartments has failed to generate bone formation.⁹⁶

The impregnation of this material with cells capable of osteogenic activity, however, has been studied. Salama and colleagues⁹⁶ and Plank and colleagues²⁴⁴ demonstrated that deproteinated xenograft bone supplemented with autologous marrow assisted osteogenesis in both experimental animals and humans. Deproteinated bone, in these experiments, served as an osteoconductive scaffolding, providing a stable mechanical environment for revascularization and proliferation and differentiation of osteogenic cells. Salama and Weissman⁹⁷ reported satisfactory results in clinical attempts to use composite xenograft/autograft (Kiel bone/marrow) in a variety of bone defects. More recently, Rawlinson and colleagues²⁴⁵ reported poor results using bovine-derived Cloward grafts. Due to the wide availability of more effective allograft matrix materials in the United States at similar costs, xenograft materials are not currently used.

Bone Morphogenic Proteins

A major advance occurred in 1978, when Urist and colleagues²⁴⁸ reported the isolation of a hydrophobic, low-molecular-weight protein fraction from insoluble bone matrix gelatin that was responsible for osteoinductive activity.²¹⁸ Further characterization of this inductive factor, BMP, was made possible by quantitative extraction accomplished by differential precipitation in a buffer containing 4M guanidine hydrochloride.²⁴⁹ Lovell and Dawson went on to report the success of a partly purified BMP preparation on polylactic acid strips in a canine segmental spinal fusion model.²⁵⁰

After an extensive search for the protein responsible for the inductive activity of bone matrix extract, Wozney and colleagues,²⁵¹ in 1988, identified and characterized three proteins isolated from a highly purified preparation from bovine bone, each capable of inducing bone formation in a rat subcutaneous bioassay. Human cDNA clones for each peptide were isolated and expressed as recombinant human proteins. Two of the encoded proteins were homologous and described as members of the TGF-β super gene family, whereas the third appeared to be a novel polypeptide (BMP-1). BMP-1 has turned out not to be a growth factor at all. Rather, this molecule has been characterized as a procollagen C-proteinase, which may have a biologic function in the activation of TGF-β-like molecules including the BMPs.^252,253

The BMP story has developed even more rapidly in recent years. At the time of this writing the BMP family of growth factors currently includes 14 members. BMP-2 through BMP-14 are homologous proteins with molecular weight of 12 to 14 kD that are post-transcriptionally modified by glycosylation and are secreted as homodimers or heterodimers of 110 to 140 amino acid peptides linked by one disulfide bond (≈30 kD).^251,254,255 In vivo, these proteins are secreted as soluble factors that have autocrine and paracrine effects. BMP-7 (OP-1) can also be found in systemic circulation and may also have positive hormonal effects on kidney function.^256,257 BMPs are also embedded in bone matrix, both as homodimers and as heterodimers at a concentration of roughly 1 mg/kg of bone, where they are believed to play a role in bone remodeling and the coupling of osteoclastic and osteoblastic activity. Of interest is the finding that as much as 65% of the BMP in bone matrix is BMP-3, which is less active in bone formation and may, in fact, be a negative regulator of bone formation.^258,259

Figure 67–2 summarizes these proteins along with synonyms or alternative names that are now, or have been, used for some of these molecules. Figure 67–3 illustrates the percent RNA sequence homology within and between subgroups of the BMP protein family. Each of these proteins can interact with one or more of a family of cell surface receptors. Cells must express both type I and type II receptors (serine/threonine kinases) in order to be responsive to BMPs because a type I and type II receptor must interact in the presence of a BMP to mediate a cellular response. To date, three type I and three type II BMP receptors have been identified. BMPR-IA binds only BMPs. BMPR-IB binds BMPs and müllerian inhibitory substance (MIS). ALK-2 (also a type I receptor) binds BMPs and activin. BMPR-II binds only BMPs, and the activin type II receptors, ActR-IIA, and ActR-IIB bind both BMPs and activin.²⁶⁰ In vitro, BMPs demonstrate dose-responsive modulation in responsive cells, both primary osteogenic cells and in cell lines, in the range of 1 to 100 ng/mL, but each BMP demonstrates a unique and variable binding pattern for the individual receptors, as illustrated in Figure 67–4. A great deal of functional redundancy and promiscuity between these proteins and receptors appears to be present, though these genes appear to be independently regulated in both space and time in embryonic development and in bone healing settings, and expression of specific inhibitors of BMP function (e.g., noggin, chordin, connective tissue growth factor, follistatin) also plays a role.^{254,261–266}

FIGURE 67–2 The bone morphogenetic protein (BMP) family of proteins. This figure lists the known members of the BMP family of proteins and some of the corresponding names that identify the same protein using other classification terminology. GDF refers to growth and differentiation factor. CDMP refers to cartilage-derived matrix protein.

FIGURE 67–3 Homology between subgroups of the bone morphogenetic protein (BMP) family. This figure illustrates the percent homology in the active protein within and between subgroups of the BMP protein family.

FIGURE 67–4 The bone morphogenetic protein (BMP) receptor family. BMPs act on cells through interaction with a family of membrane-bound cell surface receptors. Cells must express one or more of the type I and the type II receptors in order to be responsive to BMPs. Some of the cross reactivity between BMP-7, BMP-4, and another TGF-β superfamily member, Activin A, are illustrated. BMPs bind first to a type I receptor including Alk-2, Alk-3, Alk-4, and Alk-6. The relative binding affinity of each BMP is reflected by the thickness of the connecting line. The BMP-type I receptor complex is then phosphorylated by a type II receptor protein to activate the signal transduction mechanism in the cell, which results in the BMP-induced response acting through the SMAD pathway (SMAD 1, 4, 5, and 8).

Among the BMP homodimers that are most active in bone induction in vivo are BMP-2, BMP-4, BMP-6, BMP-7 (i.e., Osteogenic Protein-1 or OP-1), and BMP-9. BMP-2 and BMP-7 have both been developed for clinical applications in bone grafting and skeletal reconstruction. BMP-13 (i.e., MP53 or GDF-5) and BMP-14 (i.e., GDF-6) are also under development.^267–269 At present, both BMP-2 and OP-1 (BMP-7) have been fully evaluated and approved by the FDA for use in spinal fusion. This approval is specifically limited to the use of BMP-2 in a collagen carrier (INFUSE, Medtronic Sofamore Danek, Minneapolis) in the setting of anterior interbody fusion using a metallic cage. OP-1 Putty (Stryker, Kalamazoo, Mich.) can be used in patients who have failed a previous spinal fusion surgery and are not able to provide their own bone or bone marrow for grafting because of a condition such as osteoporosis, diabetes, or smoking.

A large number of animal studies have demonstrated the promise and relative safety for these proteins as powerful stimulants of a local bone healing response in rodents, sheep, canines, and nonhuman primates, using various carrier matrices and a dosage range of 100 to 10,000 µg/mL.^{255,270–295} Schimandle and colleagues²⁹³ reported 100% union in an uninstrumented posterolateral intertransverse fusion model in the rabbit using BMP-2 delivered in a collagen carrier, compared with only 42% fusion with autogenous corticocancellous iliac crest bone. Muschler and colleagues,²⁹⁶ using an instrumented posterior canine spinal fusion model, found that BMP-2 delivered in a degradable polymer carrier (PLGA) had comparable efficacy to autogenous cancellous bone. Cook and colleagues²⁹⁷ found similar results in a canine spine model using OP-1 (i.e., BMP-7). The use of rhBMP-6 stOPCs in a carrier of guanidine-extracted demineralized (gDBM) bone matrix significantly enhanced the rate and strength of single-level posterolateral spinal arthrodeses in the New Zealand white rabbit, compared with iliac crest bone graft, gDBM, and decortication alone.²⁹⁸ Using a rabbit model, Fu and colleagues demonstrated enhancement of posterolateral lumbar spine fusion using low-dose rhBMP-2 and cultured marrow stromal cells.

Several prospective clinical trials evaluating BMPs in the setting of spinal fusion have been performed, generally reporting performance that was comparable with autogenous cancellous bone, with fusion rates between 80% and 99%.^299–305 These initial studies strongly support the clinical value of BMPs, particularly BMP-2, which has been most thoroughly studied in the clinical setting to date.

In a prospective, randomized, controlled, multicenter clinical pilot study, Vaccaro and colleagues³⁰⁶ demonstrated that the rates of radiographic fusion, clinical improvement, and overall success associated with the use of OP-1 Putty were at least comparable with that of the autograft controls for at least 48 months after surgery. Singh and colleagues³⁰⁷ performed a prospective, single-institution, clinical case-matched, radiographic cohort study involving 52 patients who underwent posterolateral lumbar arthrodesis with pedicle screw instrumentation. Using thin-slice computed tomography (CT) analysis, the authors demonstrated 97% fusion rates in the rhBMP-2 group compared with 77% in the iliac crest bone graft at 2 years. The authors concluded that the adjunctive use of rhBMP-2 and iliac crest bone graft seems to be safe and results in significantly larger and more consistent posterolateral fusion masses.³⁰⁷ In a human posterolateral lumbar spine trial, OP-1 reliably induced viable amounts of new bone formation, but the fusion success rate evaluated by surgical exploration was only 4 of 7.³⁰⁸ Lewandrowski and colleagues³⁰⁹ reported vertebral osteolysis with the use of rh-BMP-2 in posterior lumbar interbody fusions with 5 out of 68 patients developing osteolysis within 4 months of surgery. Violation of the endplate during decortication was thought to be a contributing factor. This often resolves spontaneously. Similar observations were made in the cervical spine by Vaidya and colleagues.³¹⁰ In a prospective, consecutive patient enrollment with a minimum 24-month follow-up, 30 patients underwent anterior interbody allografts alone and 45 patients underwent anterior interbody allograft filled with rhBMP-2. All cases had posterior pedicle screw instrumentation. A total of 165 surgical levels (62 allograft alone; 103 allograft + BMP) were included. In the allograft rhBMP-2 group, fusion rates were 94%, 100%, and 100% at 6, 12, and 24 months, respectively, after surgery, while in the allograft-only group, fusion rates were 66%, 84%, and 89% at the same time intervals. Clinical outcomes were significantly improved in rhBMP-2 group when compared with the allograft group at 6 months. There were no revisions in the rhBMP-2 group and four revision fusion surgeries (13%) in the allograft group.³¹¹ In a randomized, controlled trial in patients older than 60 years of age, Glassman and colleagues³¹² concluded that RhBMP-2/INFUSE is a viable iliac crest bone graft replacement in older patients in terms of safety, clinical efficacy, and cost-effectiveness.

Recently, several investigators have evaluated the use of BMPs in cervical spine surgery.^310,313,314 Anterior cervical discectomy and fusion performed with rhBMP-2 (0.9 mg BMP per level) allograft was found to be as effective as iliac bone graft in terms of patient outcomes and fusion rates. Safety concerns related to neck swelling and higher initial costs were associated with patients in the bone morphogenic protein group.³¹³ Vaidya and colleagues compared 22 patients treated with rhBMP-2 and PEEK cages with 24 in whom allograft spacers and demineralized bone matrix was used. Radiographic examination following surgery revealed end plate resorption in all patients in whom rhBMP-2 was used. This was followed by a period of new bone formation commencing at 6 weeks. In contrast, allograft patients showed a progressive blurring of endplate-allograft junction. Dysphagia was a common complication and it was significantly more frequent and more severe in patients in whom rhBMP-2 was used. Postoperative swelling anterior to the vertebral body on lateral cervical spine radiograph was significantly larger in the rhBMP-2 group when measured from 1 to 6 weeks after which it was similar. There was no significant difference in the clinical outcome of patients in the two groups at 2 years. The authors concluded that despite providing consistently good fusion rates, they have abandoned using rhBMP-2 and PEEK cages for anterior cervical fusion, due to the side effects, high cost, and the availability of a suitable alternative.³¹⁵ Shields and colleagues³¹⁶ reviewed 151 patients who underwent either an anterior cervical discectomy and fusion (n = 138) or anterior cervical vertebrectomy and fusion (n = 13) augmented with high-dose INFUSE (BMP-2; Medtronic Sofamor Danek). They found a high morbidity rate with a total of 35 (23.2%) patients having complications that the authors thought were attributable to the use of high-dose INFUSE in the cervical spine. Fifteen patients were diagnosed with a hematoma including 11 on postoperative day 4 or 5, of whom 8 were surgically evacuated. Thirteen individuals had either a prolonged hospital stay (>48 hours) or hospital readmission because of swallowing/breathing difficulties or dramatic swelling without hematoma. The authors concluded that putative inflammatory effect that contributes to the effectiveness of INFUSE (Medtronic Sofamor Danek) in inducing fusion may spread to adjacent critical structures and lead to increased postoperative morbidity.³¹⁶ On the basis of their experience with 69 patients who underwent anterior cervical fusion, Smucker and colleagues³¹⁷ concluded that use of rhBMP-2 in the anterior cervical spine is associated with an increased rate of clinically relevant swelling events.

Although BMPs clearly provide valuable tools for use in stimulation of spinal fusion, the next set of challenges is to optimize their use in a broader range of applications. This includes defining those clinical settings where the current formulation of BMPs does or does not provide optimally effective biologic effects. This includes identification of patient groups or settings where clinical performance may be limited even for “on-label” applications (e.g., high-risk patients—diabetes, steroid use, tobacco use, pseudarthrosis). It also includes identification of potential value settings that are currently “off label,” such as posterolateral lumbar fusion and applications in the cervical spine. These must be assessed through prospective trials or carefully selected and documented clinical experience. Until data regarding the use of BMPs in these settings become available, surgeons must remain cautious to avoid overoptimism or wishful thinking that may result in the use of these products in settings where their performance can be expected to be compromised by local biologic factors.

The mechanism of action of BMPs provides many opportunities for further advancement and refinement including modification of dose, formulation, delivery systems, changes in molecular structure, and means of optimizing the environment into which these agents are placed.

One of the principle requirements for optimal BMP activity is the presence of a local population of target cells, most likely osteogenic stem cells and progenitor cells that are responsive to the protein (i.e., they express appropriate receptors). In order for a BMP to be optimally effective, these target cells must be both available and activated in sufficient numbers to produce the desired result. If an optimal number of responsive cells is not present within the tissue volume that is exposed to the protein following implantation, the biologic response to the protein will inevitably be reduced and the implantation of a BMP may be completely ineffective. Variation in the concentration or biologic potential of target cell populations in bone, bone marrow, periosteum, and other tissues (e.g., muscle, fat) may explain much of the apparent variation in the magnitude and type of response seen to BMPs and other growth factors from site to site and individual to individual.^99,100,318

Preclinical evaluation of BMPs in a series of animal models from rats to rabbits to dogs to nonhuman primates demonstrated the need for delivery of dramatically high concentrations of BMP to graft sites in higher animals.²⁷⁴ In fact, the formulations of BMP that are currently available deliver an amount of BMP that is roughly 50 times greater than the total amount of BMP that is present in an entire human skeleton. INFUSE (Medtronic Sofamor Danek) delivers BMP-2 in solution at a concentration of 1.5 mg/mL to be combined with a collagen carrier, resulting in a implanted concentration slightly less than 1 mg/mL. Similarly, the OP-1 device delivers 3.5 mg of OP-1 in a final volume of approximately 4 mL. The reason for this escalation of dose has not been clearly established, though several factors are likely to contribute. Species-specific differences in dose response may exist at the target cell level, though this has not been a consistent finding in in vitro culture of primary osteogenic cells. However, individual species do demonstrate significant differences in the concentration and prevalence of responsive target cells in local tissues and consistently lower numbers in higher animals. The geometry involved in mating active BMP with a responsive target population of cells in larger graft sites may also contribute. For both of these reasons, a huge dose may be necessary to provide a burst of BMP delivery that is sufficiently large that the BMP diffusing away from the graft site will penetrate into regional tissues to a sufficient depth at a sufficiently large concentration to activate enough stem cells and progenitors. As the graft sites become larger, activation of more target cells is required. This results in the need for a greater degree of penetration that can only be achieved with a larger dose. Similarly, if the concentration and prevalence of target cells in regional tissues decrease, as they do in higher animals, activation of a similar number of cells will require even deeper tissue penetration. These factors likely contribute to the seemingly exponential increase in dose that is necessary in larger animals.⁴¹ A second possible explanation for the massive dose relates to the issue of BMP retention at the graft site. A massive initial dose may also ensure that a sufficient, though perhaps small, quantity of BMP will remain at the site long enough to result in activation of target cells that may not enter the graft site until several days after implantation, with the associated inflammatory response or angiogenic response following the surgical trauma.

These issues in BMP delivery and function in a graft site suggest a number of options that could be used to increase the exposure of BMP to an appropriate target population of cells, thereby improving its performance. Direct delivery or supplementation of the target cell population in the graft site is one option. Evidence is provided by a number of studies that addition of a target population (e.g., bone marrow–derived cells) to a site of BMP implantation will significantly improve bone healing.^92,318 Another possible method for enhancing the performance of BMPs is to refine the method and rate of BMP delivery into the graft site. The chemical surface of the matrix may influence the protein binding, its conformation, and stability. In some cases, BMP binding to matrix can actually enhance the biologic performance.^319–321 The four major categories of BMP carrier materials are natural polymers such as collagen, hyaluronans, fibrin, chitosan, silk, alginate, and agarose; inorganic materials such as low- and high-temperature calcium orthophosphates (calcium phosphate cements and sintered ceramics) and calcium sulphate cements; poly(α-hydroxy acid) synthetic polymers such as PLA, polyglycolide (PLG); and their copolymers (poly[D,L-lactide-co-glycolide]) (PLGA).³²² Furthermore, the carrier (collagen, ceramic, polymer) may have its own biologic effects associated with the release of ions or other degradation products.⁴¹ Currently both the OP-1/BMP-7 (Osteogenic Protein-1/BMP-7; Stryker) and INFUSE (Medtronic Sofamor Danek) use type I bovine collagen as the carrier. In the case of OP-1, the protein is lyophilized onto the surface of the collagen. With INFUSE, it is adsorbed onto the surface of the collagen out of an aqueous solution. Other options include release from degradable polymers, liposomes, and collagen-hydroxyapatite microsphere.^323–325

Yet another option for enhancing the performance of BMPs is to prolong the period of time that effective concentrations of the protein are present in the graft site. Residence time of bioactive protein in a graft site is a complex function of the rate of delivery (e.g., release or solubilization); the rate of consumption within the graft site (e.g., degradation, inactivation, binding to inhibitors); and the rate of clearance from the graft site (e.g., diffusion, convection). With BMP preparations currently available, release kinetics are relatively rapid. Pharmacokinetic studies of residence time for OP-1 delivered as a lyophilized protein from a collagen carrier in a rabbit model indicate that the protein is released into the site and retained in measurable concentrations for approximately 7 days, with a maximal release rate in the first 24 hours. The implanted rhOP-1 diffuses out of the immediate graft site at low concentrations, entering systemic circulation, where it is rapidly cleared. Approximately 50% of the implanted dose is excreted in the urine.³²⁶ In the case of BMP-2 in the INFUSE Device, release kinetics are similar, with 50% of the rhBMP-2 cleared from the site in 48 hours and less than 1% remaining at 2 weeks.³²⁷

Longer residence time within the site is associated with improved efficacy at lower protein concentration.³²⁸ Residence time is influenced by the affinity of the protein to the carrier.^329,330 Residence time can also be increased by modifying the BMP protein itself to reduce its solubility and therefore its rate of diffusion out of the implant.³³¹ Prolonged residence time within the graft site may have two effects. It may allow the initial gradient of protein concentration around the site to be maintained for a longer period, which may positively influence chemotactic effects that may be mediated by the protein, drawing more activated progenitors and other cells into the defect. Prolonged residence may also serve to maintain a functional concentration of protein in the graft site for a longer time, providing the opportunity for additional stem cells and progenitors to migrate into the graft site where they may become activated.

Other Growth Factors

A large number of peptide growth factors and hormones are known to have important effects on the recruitment, proliferation, and differentiation of osteoblastic progenitors, which may have potential therapeutic importance. Only some of the examples of candidate growth factors can be listed.³³² Epidermal growth factor (EGF) and platelet-derived growth factor (PDGF) are both capable of inducing colony formation by osteoblastic progenitors in vitro,²⁴ and local injection of PDGF has been shown to result in induction of new bone formation when applied close to a bone surface. Basic fibroblast growth factor (FGF; aka bFGF or FGF-2) will also increase proliferation of human osteoblastic progenitors and reversibly inhibit the expression of alkaline phosphatase and matrix synthesis, in addition to its known potent angiogenic effects. FGF has been evaluated in preclinical studies using a hyaluronic acid delivery system and has been shown to increase local bone formation and union rates.³³³

Vascular endothelial growth factors (VEGFs) are a family of proteins that function as dimers. They have a structure similar to PDGF and interact with transmembrane receptors Flt-1, Flk-1, and Flt-4 to activate a tyrosine kinase signaling cascade. VEGFs play several important roles in angiogenesis, osteoclast migration, and osteoblastic activity.^334,335 VEGFs do not induce bone directly, but in addition to angiogenic effects, VEGF delivery has been shown to upregulate BMP activity in fracture healing and distraction osteogensis.^336,337 When used in combination, VEGFs may enhance the performance of BMPs.³³⁸

Transforming growth factor beta (TGF-β) is another potent osteotropic factor. Of the five known isoforms of TGF-β, two—TGF-β1 and TGF-β2—are synthesized by bone cells. In fact, bone matrix deposited by osteoblasts is the largest source of TGF-β and is mostly present in a latent form that is released during bone remodeling.^339,340 TGF-β has effects on bone formation and remodeling. It will induce new bone formation, but only when implanted or injected in close proximity to bone, suggesting that its bone formation effects are mediated primarily by trabecular or periosteal cells, a different or more limited target cell population than the BMPs.³⁴¹ For example, Joyce and colleagues³⁴² showed that subperiosteal injection of TGF-β can produce a marked periosteal response resulting in rapid formation of a cartilage tissue mass and bone formation via endochondral ossification. IGF-I and IGF-II potentiate a mature osteoblastic phenotype in culture.^40,343–345 Both TGF-β and IGF-1 have been shown to promote spine fusion in a sheep model.³⁴⁶

Collagen

The major components of organic matrix of bone are type I collagen (90% dry weight), a large number of noncollagenous matrix proteins, at least two proteoglycans (biglycan and decorin), and several minor collagens (mostly types III and X). These proteins contribute an osteoconductive substrate for cell attachment and migration and are necessary elements of new bone formation and mineralization. The precise contribution of each element of organic bone matrix is not yet known. Nor is it understood how the structural organization of the various components within the matrix influences the biologic function of these proteins. This makes the engineering of synthetic matrices that mimic specific functional relationships between these proteins difficult.

Type I collagen has been the focus of many efforts to develop optimal tissue engineering scaffolds. Most of the noncollagenous proteins in bone matrix can be solubilized using 4 M guanidine. However, the majority of type I collagen in bone matrix is heavily interconnected by covalent pyridinium crosslinks. This makes native collagen I insoluble and virtually impossible to manipulate as a reagent to create new structures. In addition, demineralized extracted collagen matrix is only modestly effective as osteoconductive material, though it has been used effectively as a delivery vehicle for purified proteins and extracts of bone matrix.

In contrast, fibrillar collagen (uncrosslinked collagen) is soluble and can be extracted from bone and skin. Fibrillar collagen can be engineered to produce a variety of matrices such as gels, sponges, and filaments. These are often secondarily crosslinked to stabilize their structure using a variety of chemical methods. Again, few of these engineered collagen matrices are effective by themselves in strongly promoting bone formation. Some formulations have appeared to actually compromise the efficacy of autograft in a graft site.¹⁰⁵ However, in general collagen matrices have been effective as delivery systems for bone marrow–derived cells⁹¹ and for growth factors. Both of the two BMPs that are now available for clinical use utilize collagen I as their delivery vehicle. In addition, an interconnected porous mesh of purified bovine collagen (pore size ≈ 150 microns) with a thin coating of hydroxyapatite precipitated on its surface (Healos, DePuy Spine, Inc., Raynham, Mass.) has been specifically developed and marketed as a delivery system for bone marrow harvested by aspiration. Clinical assessments have demonstrated high fusion rates in both lumbar interbody fusions and posterolateral fusion sites.¹¹¹

Noncollagenous Matrix Proteins

Bone matrix contains many proteins other than collagens and growth factors.^{98,227,347–349} These proteins may serve a role in organization of the collagenous matrix and other proteins into higher ordered structures. They may provide attachment sites for cells or binding sites for growth factors. They may serve as regulators of mineralization, as in the case of bone and dentin phosphoproteins, bone sialoprotein, osteonectin, and osteocalcin.^350–354 They may also provide a source for release of locally active growth factors and other bioactive molecules during the process of matrix turnover and remodeling, as previously discussed.

Other than growth factors that are embedded in bone matrix, it is difficult to ascribe any one factor with exceptional function in the setting of bone grafting or a high potential for future clinical application. However, some may be relevant and deserve mention. For example, osteoblasts and osteoblastic cell lines appear to express integrins that bind selectively to both fibronectin and vitronectin^355–358 and possibly osteopontin, bone sialoprotein, and laminin. In addition, osteocalcin appears to be chemotactic for osteoclasts and monocytes,³⁵⁹ critical elements of normal bone remodeling. Although these proteins are not likely to be exploited in terms of recombinant manufacturing processes used for BMPs, it is possible that these functions may be localized to specific functional domains of these proteins. This knowledge may be used to design specific low-molecular-weight surrogates that may be applied to tissue engineering constructs.

Ceramics

Calcium phosphate biomaterials fused at their crystal grain boundaries into polycrystalline ceramics by high temperature sintering confer stability to these minerals and reduce bioresorbability.³⁶⁰ A variety of ceramics are currently being evaluated, most of which are composed of either hydroxyapatite (HA) or tricalcium phosphate (TCP). Ceramics may be prepared as porous three-dimensional implants, dense block implants, granular particles (usually 0.5 to 3 mm in size), or thin surface coatings. Almost all calcium phosphate ceramics have a high degree of biocompatibility,^236,361 and some have already been extensively used in dentistry and maxillofacial surgery.^{236,243,244,358–365}

The minimal macropore size in porous ceramics needed for effective ingrowth of bone is approximately 100 µm.³⁶⁶ Most porous ceramics currently being manufactured contain interconnecting macropores ranging from 100 to 400 µm. The various calcium phosphate ceramics generally differ with regard to their bioresorbability characteristics. A number of investigators have reported that ceramic hydroxyapatite does not exhibit extensive bioresorption and is essentially inert.^360,361,367 Conversely, there is unequivocal evidence that ceramic TCP undergoes biodegradation.^{236,360,361,367–369} In addition, implants with a large surface area will tend to exhibit more rapid degradation.²³⁶

Early studies of ceramics suggested that they may be capable of osteogenic stimulation.³⁷⁰ In fact, one can often find new bone formation in an HA ceramic implant placed at heterotopic sites in the absence of other stimuli. This occurs only after several months and would not be likely to contribute to the early success of a bone graft. The role of ceramics, therefore, is primarily that of osteoconduction. One possible mechanism for this apparent late osteoinductive property of HA ceramics is that an implanted HA implant will selectively bind proteins to its surface on the basis of their relative affinity to HA. This may result in the accumulation of some protein growth factors such as BMPs, TGF-βs, and insulin-like binding protein-5 (IGFBP-5), which have strong affinity to HA. Accumulation of these low-abundance proteins and their presentation on a stable surface may secondarily create a local growth factor environment on the ceramic surface that is capable of recruiting local osteoblastic progenitors and inducing bone formation. This affinity of many osteotropic growth factors for the highly charged surface of HA may also make HA ceramics an effective delivery system for growth factors as composite synthetic bone grafting materials are developed.

The stability of a bone-ceramic interface and preparation of local bone are also important. Cameron and colleagues³⁷¹ demonstrated that ceramic implants placed against an unprepared bony cortex do not exhibit bone ingrowth and simply resorb over time. However, when placed subperiosteally and immobilized on a scarified cortex, bone ingrowth readily takes place. The sensitivity of these materials to micromotion likely results from magnification of the mechanical strain within the graft site at the interface between local tissues and the surface of a rigid ceramic block. Similar magnification of strain will occur in the regions of tissue between adjoining ceramic granules, inhibiting bone formation. As a result, the optimal settings for use of these materials may be limited to settings in which mechanical micromotion can be well controlled.

Another drawback of ceramic implants is that they are brittle and have low impact and fracture resistance.³⁷² Furthermore, the limited solubility and remodeling capacity of highly crystalline HA ceramics may retard late stages of bone healing and remodeling, as well as compromise late mechanical properties of the bone formed in a fusion site.³⁶⁰ This concern has been reduced by the work of Ohgushi and colleagues,³⁷³ which showed that ceramic combined with bone marrow exhibited greater biomechanical properties following implantation with marrow cells as a result of new bone formation in the implant. In addition, Muschler and colleagues^104,105 have performed a series of spinal fusion experiments evaluating composites of collagen and ceramic granules (60% hydroxyapatite, 40% TCP). Although these studies found that all composites tested had a significantly higher nonunion rate than autogenous cancellous bone graft, the mechanical properties of successful unions achieved with the collagen ceramic composites were comparable with the mechanical properties of unions resulting from autogenous bone graft, despite the presence of unresorbed granules in the fusion mass. Hing and colleagues³⁷⁴ compared the rate, quality, and extent of osseous healing in a standard rabbit defect model between dense calcium sulfate, ultraporous tricalcium phosphate, and porous silicated calcium phosphate. The authors concluded that in patients in whom bone regeneration may be compromised, the degradation observed with some resorbable bone grafts may contribute to the decoupling of bone regeneration and resorption within the graft site, which may ultimately lead to incomplete bone repair.

Ceramic blocks have been evaluated in a goat anterior cervical fusion model with a reported 50% to 70% fusion rate.^375,376 Several injectable ceramic preparations that crystallize at body temperature have also been described. Resorption rates vary significantly, from weeks to months. These may provide means for improving the initial mechanical fixation for acute fractures, though they do not seem well conceived for achieving long-term fixation. These injectable setting ceramics may also have potential utility in providing extended local delivery for bioactive proteins.³⁷⁷ In a prospective, randomized study with 3-year follow-up, Dai and colleagues³⁷⁸ compared beta-tricalcium phosphate versus autograft in patients undergoing single-level instrumented posterolateral fusion of lumbar spine with beta-tricalcium phosphate versus autograft. The authors reported similar clinical outcomes and fusion rates and suggested that beta-TCP as bone graft substitute may eliminate the need for bone grafting harvesting from the ilium.³⁷⁸ In a different prospective study, Chen and colleagues³⁷⁹ placed autologous iliac crest bone graft in one posterolateral gutter, while on the other side, an equal quantity of autogenous laminectomy bone supplemented with calcium sulfate was placed. In a prospective, matched, and controlled study, Acharya and colleagues³⁸⁰ evaluated hydroxyapatite-bioactive glass ceramic composite as a stand-alone graft substitute for posterolateral fusion of lumbar spine by placing it in the left intertransverse bed. The autograft was placed in the right intertransverse bed. At the end of 1 year, excellent radiologic outcome was seen on the right side (autogenous graft) in all the cases, whereas 95% (21/22) of the cases had poor consolidation on the left side (hydroxyapatite composite). The authors reported fusion rates and fusion size to be similar between the two sides. Epstein³⁸¹ found a 15% pseudarthrosis rate following multilevel laminectomy and one- to two-level noninstrumented posterolateral fusion using lamina autograft/B-TCP. Enriched bone-marrow–derived mesenchymal stem cells were combined with porous beta-tricalcium phosphate in 41 patients undergoing posterior spinal fusion.³⁸² After 34.5 months, 95.1% cases had good spinal fusion results.