Departments of Neurological Surgery (JPS, JMS, JAJ, AHF, CGdP, NES, GAH), Radiology (DFK), Biomedical Engineering (DDM), and Biophysics (JPS), University of Virginia, Charlottesville, Virginia
OBJECTIVE: An optimal method for spinal fusion would induce rapid growth of bone via an osteoconductive and osteoinductive implant. This study examines the spinal fusion enhancement potential of some osteoconductive and osteoinductive biomaterials.
METHODS: Four similar canines received unilateral posterolateral fusions on the left side at T13L1 and L4L5 and on the right side at L2L3 and L6L7. The experiments were grouped as follows: Group A, autogenous bone harvested from the iliac crest; Group B, autogenous bone and collagen; Group C, no implant; and Group D, autogenous bone, collagen, and recombinant human bone morphogenetic protein-2. Radiographic assessment, three-dimensional computed tomographic volumetric analysis, and biomechanical testing were performed at each level.
RESULTS: For Groups A and B, the fusions demonstrated moderate bone formation at 6 and 12 weeks postoperatively. Group D fusions exhibited earlier and more dramatic increases in volume and radiodensity and eventually were comparable in size to the vertebral bodies. Average fusion volumes computed from three-dimensional computed tomographic analysis were: Group A = 1.243 cc, Group B = 0.900 cc, Group C = 0.000 cc, and Group D = 6.668 cc (P = 0.003 compared to Group A). Group D exhibited flexion and extension biomechanical properties much greater than controls. The addition of recombinant human bone morphogenetic protein-2 consistently yielded the strongest fused segments and, on average, enhanced extension stiffness by 626% and flexion stiffness by 1120% over controls.
CONCLUSION: The most advantageous spinal fusion implant matrix consisted of recombinant human bone morphogenetic protein-2, autogenous bone, and collagen. Future investigators, however, need to examine the appropriate quantities of the individual components and clarify the efficacy of the matrix for the various types of spinal fusion approaches. (Neurosurgery 39:548554, 1996)
Key words: Bone morphogenetic protein, Osteoconduction, Osteoinduction, Spinal fusion
Spinal fusion plays a significant role in the treatment of conditions including spinal trauma, tumors, congenital abnormalities, and degenerative diseases (8, 16). Long-term stability in spinal fusion either with or without instrumentation can only be achieved by creating a solid bone arthrodesis. The most significant obstacles in all spinal fusions are poor bone deposition at the fusion site and pseudoarthroses. Moreover, poor bone material properties (e.g., trabecular thickness and mineralized volume and rate) at the fusion site are a particular problem with fusions without instrumentation (12). Autogenous bone grafts are typically used to create a solid bone fusion mass and remain the gold standard (19, 21). Unfortunately, it can be difficult to obtain adequate quantities of bone and the grafts are prone to resorption (10, 21). Moreover, harvesting bone from sites such as the iliac crest or ribs can be painful and increase morbidity (21, 22, 29, 34). Cryopreserved allograft bone has disadvantages of a lower fusion rate and a higher risk of infection compared to autogenous bone grafts (17, 19, 21, 29, 34).
An optimal method for spinal fusion would induce rapid growth of bone at the site via an osteoconductive and osteoinductive implant. One of the first successful attempts at induction of bone growth was achieved using demineralized bone (26). These implants, however, fused inconsistently, tended to be resorbed, and carried the risk of transmission of infectious agents. In another attempt to augment spinal fusion, calcium hydroxyapatite, tricalcium phosphate, and bioglass were implanted into canine spines (18). In those experiments, the calcium hydroxyapatite was the only biomaterial that was incorporated into surrounding trabecular bone; unfortunately, hydroxyapatite alone exhibits poor tensile strength (18) (Costantino PD, Loyola University Medical Center, personal communication, 1995).
More recently, osteogenic growth factors found in demineralized bone and termed bone morphogenetic proteins (BMPs) have been isolated (27). BMPs have been shown to induce transformation and differentiation of mesenchymal cells into osteoblasts (2, 28). Moreover, in situ hybridization studies have illustrated that BMP messenger RNAs are present at the appropriate times and locations during embryonic skeletal development (30). Mammalian serine/threonine kinase receptors (CFK-43A and BRK-1), which have high binding affinities for BMP-2 and -4 but low affinities for transforming growth factor-beta, are expressed in developing limb buds (13, 32). The role of BMPs in bone formation seems indisputable.
Through molecular cloning, nine BMPs have been identified (21). With the exception of BMP-1, BMPs are members of the transforming growth factor-beta superfamily of growth factors and are found in minute quantities in bone. To isolate 1 mg of BMP with current extraction and purification techniques, 1000 kg of bone would be needed (9). Using genetically modified cell lines, recombinant BMP can be produced in vitro in virtually unlimited quantities (31). Recombinant human BMP-2 (rhBMP-2) is free of infectious agents and contaminants that could potentially affect osteoinduction. Another advantage of rhBMP-2 is that the molecule is of human origin and thus does not elicit the same worry of rejection or partial response as xenogeneic agents.
Previous research has shown that the osteoinductive effect that occurs after administration of BMP-2 continues for approximately 6 months (24). In the absence of an appropriate carrier, hemorrhage and edema can act to dilute or wash out the growth factor from the intended location, but the best carrier for rhBMP-2 has yet to be determined. In this study, a potential osteoconductive carrier of bovine Type I collagen was used to bind and deliver the growth factor. Collagen has been shown to provide a framework onto which osteoblasts can freely migrate (6, 7, 23). The goal of this research was to examine radiologically and biomechanically whether the addition of collagen and rhBMP-2 would improve spinous process, lamina, and facet autogenous bone lumbar fusions.
Animals in Group A received autogenous bone alone harvested from both iliac crests. Animals in Group B received fusion grafts of bovine Type I collagen gel and autogenous bone. Animals in Group C received no implant and served as the controls. Animals in Group D received fusion grafts consisting of Type I collagen gel, autogenous bone, and bone morphogenetic protein-2 (rhBMP-2). Each of four canines was fused at the following two vertebral levels: left side of T13L1, right side of L2L3, left side of L4L5, and right side of L6L7. The four graft materials were rotated through the different lumbar sites to minimize any effects that the level might have.
The animals were killed 12 weeks postoperatively with a phenobarbital/pentobarbital mixture. Specimens were delicately excised, cleaned of soft tissue, and subjected to gross examination and biomechanical flexion and extension testing.
Next, incisions were made bilaterally to expose the iliac crests. Bone chips were removed using rongeurs, and the bone was washed and stored in sterile normal saline solution. The iliac crest incisions were then closed in three layers.
In Group A, 2.6 g of autogenous bone chips were added over the fusion level. In Group B, the fusion implant contained 2.6 g of bone chips and 1 g of bovine Type I collagen rehydrated into a cohesive paste material with 2 ml of 0.9 mol/L sterile saline (ReGen Biologics, Franklin Lakes, NJ) (14). Group C had no implants. The Group D implants consisted of 1 g of collagen, 2 ml of sterile saline, 2.6 g of bone chips, and 1.6 mg of rhBMP-2 (Genetics Institute, Cambridge, MA). The fascial layer was then closed in an interrupted fashion using 30 Nurolon sutures (Ethicon, Inc., Somerville, NJ). Subcutaneous tissue was closed in an interrupted fashion using 30 Vicryl sutures (Ethicon, Inc.), and the skin was closed with 30 Dermalon (American Cyanamid Co., Danbury, CT). The canines received postoperative analgesia as needed and postoperative antibiotics for 3 days.
The collagen was tested for cytotoxicity, sterility, hemolysis, pyrogenicity, mutagenicity, and immunogenicity. Cytotoxicity, sterility, hemolysis, pyrogenicity were tested for in accordance with United States Pharmacopeia XXII (25). A modified Ames test was performed to assess mutagenicity (1). Using enzyme-linked immunosorbent assay, the humoral immune response was examined in a rabbit model. Finally, a mouse host was used to examine the immune response to the collagen. In all cases, the results of the safety studies were consistent with the safety requirements for implantable materials.
Volumetric analysis was performed on a Voxel Q workstation (Picker International). A binary image was created by setting the window to zero (0) Hounsfield units. A level of 100 Hounsfield units was chosen, because preliminary data demonstrated that this level most accurately correlated with the volume of cortical bone. Volumetric data were generated based on slice thickness and number of pixels per slice with attenuation values greater than or equal to100 Hounsfield units. Regions of interest were manually created to include only posterior elements. The region of interest was further subdivided by drawing a coronal line bisecting the spinal canal. Using the 12-week postoperative scans, the cortices of intact posterior elements were easily distinguished from adjacent bone grafts. Computer-aided hand segmentation was performed to exclude the intact posterior elements and yield the volume of the new bone graft.
Volumetric analyses of the fusion masses were performed using 3-D CT reconstructions. The results in Figure 3 are the fusion volumes at 12 weeks postoperatively. As expected, Group C (i.e., the control group) exhibited no change in bone volume. Group A fusion volume was 1.243 ± 0.124 cc (± standard error of the mean with a sample size of 4). With the addition of collagen to the bone chips, the fusion volume for Group B was 0.899 ± 0.325 cc, which constituted a 27.7% decrease from that of autogenous bone alone. For Group D, the fusion volume was significantly increased to 6.668 ± 1.139 cc, 536% more than autogenous bone alone. Statistically significant differences exist among Groups A, B, and D and the controls in Group C. Of more interest, statistically significant differences (P = 0.003 in both cases using an unpaired Student's t test) in fusion volume exist between Group D and either Group A or B.
In Figure 4
Biomechanical testing of the explants showed dramatic strength of the rhBMP-2 fusion sites over the controls. Moreover, the consistent biomechanical superiority of the rhBMP-2 implants, regardless of the spinal level, is indicative of a more uniform bone formation process compared to fusions without rhBMP-2. The stiffer fusion results with the autogenous bone (Group A) as compared to the autogenous bone and collagen (Group B) may be caused by a lower density of the fusion mass in the presence of collagen because collagen occupied significant volume in the implant. Furthermore, collagen alone has been shown to delay bone fusion (15).
When the 3-D CT volumetric analysis and the biomechanical testing results were compared, the statistical correlation coefficient (R2) between the fusion volume and the extension stiffness of a specimen was 0.974; the correlation coefficient between the fusion volume and the flexion stiffness of a specimen was 0.976. Thus, a strong correlation exists between the results of the 3-D CT volumetric analysis and the biomechanical testing. In other words, specimens that had larger fusion volumes also were more biomechanically stable. This result may seem obvious, but its significance is substantial. Providing that a solid fusion without pseudoarthrosis has been achieved, future researchers, particularly those conducting human investigations, may gain insight into the flexion and extension strength of a spinal fusion by measuring its fusion volume radiologically.
Several recent animal studies have demonstrated the usefulness of BMPs in the healing of skeletal defects. For instance, Yasko et al. (33) demonstrated that rhBMP-2 stimulated bone formation in a dose-related fashion in rats with 5-mm segmental defects in the femur. Yasko's work was extended by Gerhart et al. (9), who demonstrated that 2-cm femoral defects could be consistently healed with rhBMP-2. In mandibular defects, Toriumi et al. (24) demonstrated that rhBMP-2 effectively healed 3-cm spans. Using a rabbit intertransverse process fusion model, Schimandle et al. (21) showed that rhBMP-2 increased both the strength and stiffness of autograft fusions. Another bone-inducing protein called rhOP-1 has proven effective in healing segmental osteoperiosteal defects in rabbits, dogs, and monkeys (4). This protein also seems to be of potential value in posterior spinal stabilization without the use of autograft (5). The present study has demonstrated that the application of an osteoinductive rhBMP-2, osteoconductive Type I collagen, and autogenous bone matrix increases the volume and biomechanical stability of lumbar spinous process, laminar, and facet fusions.
One limitation of this spinal fusion matrix was the fluidity of the collagen carrier in the implant compartment. Care must be exercised to contain the collagen carrier at the desired fusion site. This fluidity could limit the usefulness of the biomaterial for applications such as cranioplasty. Nevertheless, the fluid quality of collagen is also a beneficial intraoperative property. It can be applied surgically with a cannula or syringe, particularly where the bone defects are irregular or inaccessible by normal surgical means. The collagen carrier and rhBMP-2 implant material could even be administered during an endoscopic procedure. It is also the fluid state of collagen that enables it to establish good osteoconductive contact with the graft bone chips as well as the host bone.
Recent studies at other institutions with hydroxyapatite, which can be injected in a fluid state but hardens to the strength of bone at physiological temperatures within 10 minutes after implantation, have yielded some encouraging results (3, 11). The hydroxyapatite material seems to be an equally good carrier of bone morphogenetic protein with the advantage of easy and rapid containment both of the growth factor and the carrier at the desired site (20). Vascularization of the hydroxyapatite also has been observed within a reasonable timetable (Costantino PD, Loyola University Medical Center, personal communication, 1995). Hydroxyapatite seems to be useful for long bone and cranial defects, but its use in spinal fusion has not been adequately explored.
The ability of rhBMP-2 to enhance bone volume and biomechanical strength is clear, and the potential application of rhBMP-2 and collagen in addition to autogenous bone for spinal fusion looks promising. Moreover, the search for the best in vivo osteoconductive carrier for bone morphogenetic proteins continues with a number of possible candidates (e.g., fibrillar collagen, a collagen sponge, demineralized bone, polymethylmethacrylate, and hydroxyapatite preparations) and the outlook is hopeful. Future research must indicate the best carrier, optimize the appropriate quantities of the individual components, and better clarify the efficacy of the matrix for this and other spinal fusion techniques.
Received, November 9, 1995.
Accepted, March 27, 1996.
Reprint requests: Jason P. Sheehan, M.S., Department of Neurological Surgery, Box 212, University of Virginia Health Sciences Center, Charlottesville, VA 22908.
In conclusion, the authors have documented the benefit of rhBMP-2 in fusion constructs in canines. Hopefully, further work from their laboratory and others will duplicate these results, answer the above questions, and eventually explain the definitive role of rhBMP-2 in a clinical setting.
Volker K.H. Sonntag
Sheehan et al. have performed an interesting and valuable study of methods to perform spinal fusion and primarily to study the effectiveness of using recombinant bone morphogenetic protein (BMP). As expected, the BMP had a beneficial effect on enhancing the fusion and stiffness of the bone graft site.
It will be interesting to see what happens with the use of BMP when it is added without the presence of bone or when it is used with a combination of allograft and/or calcium apatite. If the recombinant BMP can enhance a fusion with any of those substances, less morbidity will occur with spinal fusions.
We are all awaiting the results of clinical trials of BMP. I hope they will be carried out soon and that they will be effective so that fusion surgery can be enhanced.
Stewart B. Dunsker
Biological enhancement of spinal fusion has become a promising area of clinically relevant spinal research. Little doubt exists that such techniques will have widespread clinical applicability in the near future. There are many potential benefits of these methods, including reduced pseudoarthrosis rates, biomechanically stronger fusion bone, and reduced donor site morbidity from diminished reliance on spinal implants. There is also significant potential for application of these methods with concurrent development of minimally invasive spinal surgery techniques. I look forward to further development of these promising methods of spinal stabilization.
Paul C. McCormick
New York, New York
Sheehan et al. have eloquently presented their laboratory experience with rhBMP-2 as an enhancer of lumbar spine fusion. They have statistically demonstrated that the addition of rhBMP-2 volumetrically increases bone growth in spinal fusion, compared to autograft controls. Although an increase in bone volume was observed with the addition of rhBMP-2 to autograft spinal fusion, a statistically significant increase in biomechanical strength was not shown, even when compared to "spine fusions" without graft. The authors stated that a statistically significant strength difference "was approached" between the no graft and autograft plus rhBMP-2 groups. Significance may have been achieved with an increase in the number of animals studied. An increased number of laboratory subjects might also provide additional information regarding the clinically relevant differences between autograft only and autograft plus rhBMP. This is clinically more relevant than comparisons between no graft and graft plus rhBMP.
An increased availability of BMP for laboratory research will increase the statistical power of future investigations by allowing larger studies to be performed. Other variables (such as dosage, efficacy of "carriers," and toxicity and complications) may also be addressed.
Sheehan et al. have clearly demonstrated the potential for BMP application to spinal surgery. They have also established foundation for future research. We, as well as many others, are eagerly anticipating the results of ongoing and future research in this exciting area of medicine. These results almost certainly will alter the manner in which we manage pathological diseases of the spine.
Michael A. Morone
Edward C. Benzel
Albuquerque, New Mexico
Return to Table of Contents
Return to the Neurosurgery Home Page