Isolation of Platelet-Rich Plasma
Platelet-rich plasma was obtained from syngeneic donor mice by double centrifugation (200 g to create platelet/plasma phase, 500 g to pellet platelets). The platelet-rich plasma was adjusted to a concentration of 1.0 × 109 platelets/mL with use of a hemocytometer for cell counts21. A similar concentration of platelets was previously validated for optimal peri-implant osteogenesis in a rabbit model22.
Isolation of Bone Marrow
Bone marrow was obtained from the femora and tibiae of donor mice. The marrow was serially passaged through 23, 25, and 27-gauge needles (five times per needle bore) to separate cell clusters and then was pelleted by means of centrifugation. The number of bone marrow cells was adjusted to a concentration of 1.0 × 108 cells/mL in Dulbecco minimum essential medium, with use of a hemocytometer to determine cell counts.
Murine Fusion Model
All procedures were approved by the Institutional Animal Care and Use Committee and were performed following the National Institutes of Health Guide for the Care and Use of Laboratory Animals23. Adult male C57BL/6 mice (two to three months old) were anesthetized with intraperitoneal pentobarbital (50 mg/kg body weight). The surgical model was similar to that described for rats by Salamon et al.24 and was modified for posterolateral spinal arthrodesis in a mouse13. With use of a sterile technique, a midline incision was made through the dorsal skin and the subcutaneous tissue down to the dorsal lumbar fascia. An operating microscope (Wild OpMi651; Wild-Leica, Gais, Switzerland) at a magnification of as high as 16× was used in all subsequent steps to ensure precision. The fascia was incised 3 mm from the midline, and the paraspinal musculature was split to expose the transverse processes of L4, L5, and L6. This posterolateral approach is similar to the Wiltse approach, which spares the facet joints, posterior elements, and midline soft tissues. The transverse processes were sharply decorticated along their dorsal surfaces until punctate bleeding was seen.
Ten mice were included in each of four study groups. Control animals (Group 1) underwent simple bilateral exposure and decortication of the transverse processes, without placement of a graft. In the three experimental groups, a collagen sponge measuring 10 × 10 × 3.5 mm (INFUSE Kit; Medtronic Sofamor Danek, Memphis, Tennessee) was placed on each side, overlying the decorticated transverse processes. Each collagen sponge was presoaked for fifteen minutes with 31 µg of rhBMP-2 (recombinant human BMP-2; Medtronic Sofamor Danek) in a 100-µL solution containing either saline solution (Group 2), platelet-rich plasma (50,000,000 platelets total) (Group 3), or donor bone marrow (5,000,000 cells total) (Group 4). Thus, a total of 62 µg of rhBMP-2 was used at the arthrodesis site in each mouse in Groups 2, 3, and 4. After placement of the collagen sponge in each experimental group (and after decortication in the control group), the dorsolumbar fascia was approximated with use of a running 5-0 nylon suture and the skin was closed with interrupted 5-0 nylon sutures.
rhBMP-2 Dosage
The concentration of rhBMP-2 used for primate intertransverse arthrodesis varies from 1 to 2.5 mg/mL, with 6 to 12 mg of rhBMP-2 per side9,25 being used in conjunction with a collagen sponge. In the mouse model, we selected an optimal dose of rhBMP-2 that was designed to induce spine fusion on the basis of a review of the literature9,11,12,26-29 and extensive discussions with the manufacturer. An rhBMP-2 dose concentration of 0.31 mg/mL was used, with a total dose per side of the lumbar spine of 31 µg and with the average mouse weighing 25 g. Liao et al.26 and Schimandle et al.27 used 2.7 to 3 mg/side for rabbit posterolateral lumbar spine fusion, with the average rabbit weighing 3 to 4.5 kg.
Postoperative Specimen Processing
Four weeks after surgery, the spine and the surrounding soft tissues were harvested from the mouse and were fixed in 10% formaldehyde in phosphate-buffered saline solution. A Faxitron high-resolution radiography unit (Hewlett-Packard, McMinnville, Oregon) and high-resolution film (EKTASCAN B/RA Film 4153; Eastman Kodak, Rochester, New York) were used to make anteroposterior radiographs of each spine. The radiographs were blinded and were independently graded with regard to fusion by two surgeons (R.D.R., K.G.), with use of a nonparametric scale, as 2 (definitely fused), 1 (probably fused), and 0 (definitely not fused), on the basis of the presence of continuous trabecular bone bridging the intertransverse area.
A video camera (Model DFC280; Leica Microsystems, Cambridge, United Kingdom) was used to make digital images of the radiographs. The radiographs were scanned with use of image-analysis software (Image-Pro Plus Software v 5.0; Media Cybernetics, Bethesda, Maryland) running on a Windows XP workstation. The area and density of fusion were assessed for each mouse spine with use of National Institutes of Health ImageJ software30. The fusion area was measured for each mouse spine from L4 to L6. Radiodensitometric measurement was carried out in the peripheral fusion area to avoid error from the more medially located transverse processes; the density of the imaged region was calibrated with use of a metal step-wedge that spanned the region of maximum bone density and that was evaluated radiographically at the same x-ray exposure parameters as the spine samples. The spine specimens then were evaluated with computed tomography, and representative slices through the fusion mass were measured morphometrically with regard to the area of the fusion, with the volume being reconstructed from serially measured fusion areas.
Samples were decalcified in ethylenediaminetetraacetic acid (EDTA), were embedded in paraffin, were longitudinally sectioned in the dorsal-ventral plane every 5 µm, and were evaluated histologically with hematoxylin and eosin. Two blinded evaluators (V.B.S., U.M.) evaluated each animal with regard to the evidence of fusion on the basis of the presence of bridging trabeculae between the adjacent transverse processes at the two fusion segments. A segment showing abundant osseous trabeculae bridging adjacent transverse processes on a single longitudinal section was graded as "definite fusion" and was assigned a score of 2. Patchy discontinuity in the trabecular architecture between the transverse processes, with some trabecular overlapping of the region of discontinuity in histological sections above or below the representative section, was classified as "probable fusion" and was assigned a score of 1. Obvious gaps between the transverse processes, with complete discontinuity between the trabeculae and no trabecular overlapping in adjacent histological sections, were classified as "definitely not fused" and were assigned a score of 0. Qualitative blinded analysis of two representative sections from each of the forty mice was carried out in conjunction with a trained musculoskeletal pathologist (V.B.S.). These sections were evaluated for the presence or absence of any inflammatory response to the carrier, the amount and quality of induced bone (osteoid), and the interface between the induced bone and native host bone.
Statistical Analysis
Quantitative data were statistically compared among groups with analysis of variance and between groups with the Student-Newman-Keuls post hoc test, with the level of significance set at p < 0.05. Data from the blinded evaluations of the radiographic and histological images were compared with use of the Kruskal-Wallis nonparametric analysis of variance test.
Source of Funding
A grant from Medtronic Sofamor-Danek was used to pay for author (B.C.C.) and technician salaries, fringe expenses, supplies and travel expenses.
Two mice died postoperatively because of poor recovery from anesthesia; these mice were replaced to maintain the number of animals at ten per group. Although hypertrophic bone formation was seen around the decorticated transverse processes, control mice did not show evidence of fusion at any segment (Fig. 1, a). In contrast, fusion occurred in all mice in each of the three experimental groups, with the mice showing a consistent fusion mass between the L4 and L6 transverse processes (Fig. 1, b, c, and d) as assessed on the basis of blinded evaluations of radiographs.
The radiographically measured area of fusion was significantly greater (p < 0.05) for the group treated with rhBMP-2 and bone marrow (Group 4) than for the group treated with rhBMP-2 alone (Group 2) (Fig. 2, a). The control group (Group 1) showed no evident fusion mass, and, thus, the area measurement largely reflected the area of the original spine and decorticated transverse processes. Radiographic fusion density followed a similar pattern; Group 4 had significantly greater fusion density than did Group 2 (p < 0.05) (Fig. 2, b). The control group (Group 1) had no evident fusion mass and therefore there was no fusion density to measure. The volumes of the spines and their fusion masses, as measured on computed tomography images, were greater for the group treated with rhBMP-2 and bone marrow (Group 4) in comparison with the other groups (p < 0.001) (Fig. 2, c). All three experimental groups had greater fusion mass volumes than did the control group (p < 0.001). The group treated with rhBMP-2 and platelet-rich plasma (Group 3) did not show a significant increase in fusion area, density, or volume in comparison with the group treated with rhBMP-2 alone (Group 2).
The histological rate of fusion was significantly greater in each experimental group than in the control group (p < 0.01), but no difference was found among the three experimental groups. A trend toward a qualitatively better fusion was noted in the group treated with rhBMP-2 and bone marrow (Figs. 3-A through 3-D). The fusion sites in all rhBMP-2-treated spines showed the consistent presence of cortical bone in contact with the adjacent paraspinal musculature, with a well-developed inner bone marrow. The control group (treated with decortication only) showed some endochondral ossification near the tip of the decorticated transverse processes, but no bone tissue bridging the transverse processes (Fig 3-A). Fusion sites treated with rhBMP-2 alone had some internal trabeculated bone (Fig 3-B). Trabeculae were more frequently found in the group treated with rhBMP-2 and platelet-rich plasma (Fig 3-C) and the group treated with rhBMP-2 and bone marrow (Fig 3-D); the latter group also demonstrated a tendency toward the development of a thicker cortical perimeter and active bone formation. Osteocytes were in abundance within the calcified osteoid of the cortical and trabecular bone in all experimental groups.
The current United States Food and Drug Administration approval for the use of rhBMP-2 applies to anterior lumbar interbody spinal arthrodesis at a dose concentration of 1.5 mg/mL on a collagen sponge carrier. The total dose of rhBMP-2 used per patient for one lumbar level in this fashion ranges from 3.9 to 8.4 mg28,31. The use of a similar dose of rhBMP-2 and carrier has not been successful in inducing posterolateral spine fusion in Rhesus monkeys9. A higher dose concentration of rhBMP-2 of 2.0 mg/mL, delivered on a hydroxyapatite/tricalcium phosphate carrier, has successfully induced fusion in patients undergoing posterolateral lumbar spine arthrodesis, with reported fusion rates of 89% (thirty-three of thirty-seven)8, 90.6% (forty-eight of fifty-three)7 and 100% (eleven of eleven)2. The total dose of rhBMP-2 used per fusion level in all of those human trials was 40 mg, with unknown long-term side effects from use of this higher dose. In addition, animal studies have shown that increasing the concentration and total dose of rhBMP-2 above a certain threshold has been ineffective for increasing the quality or area of the lumbar spine fusion mass9,25,29. In vitro11 and in vivo studies12 have shown that increasing the dose of BMP beyond an optimum dose has decreased osteoinduction and can result in the formation of acellular cyst-like voids, with decreased biomechanical strength of the developing fusion mass. Sciadini and Johnson hypothesized that this may be a result of increased stimulation of angiogenesis from higher doses of BMP12.
There is limited experimental evidence showing augmentation of osteogenesis in association with the use of multiple growth factors or with the introduction of supplemental target cells. Combining BMP-2 and fibroblast growth factor (FGF) treatment of mesenchymal stem cells has shown greater bone formation following implantation than the use of each factor alone32. Implants that have been constructed in vitro by the addition of BMP-2 to bone marrow have worked as effective anterior lumbar spine fusion implants in rabbits33. Peng et al. reported quantitatively larger and denser bone formation in a critical skull defect in mice when vascular endothelial growth factor (VEGF)-producing stem cells were added to BMP-4-producing stem cells34. Wang et al. reported an equal rate of fusion in rats with both rhBMP-2 and rhBMP-2 in the presence of mesenchymal stem cells but reported more "robust" bone in the presence of the mesenchymal stem cells35. Recently, greater posterolateral fusion in rabbits was found when mesenchymal stem cells were combined with both BMP-2 and FGF for implantation36. The use of bone marrow cells for spine arthrodesis is rational, given their inherent osteogenic capacity attributable to endogenous osteoblastic stem cells and a possible osteoinductive potential from cytokine/growth factors secreted by one or more of the resident cells14-16,37.
The interaction between endogenous BMP-2, locally present osteoprogenitor cells, and growth factors derived from platelets may be a primary mechanism in osteogenesis and fusion. The osteogenic activity of BMP-2 is primarily through osteoprogenitor cells38, whereas platelet-derived growth factor (PDGF) stimulates mitogenesis of mesenchymal cells39 and promotes the formation of extracellular matrix40. Clinical studies have demonstrated the efficacy of bone marrow aspirate or composite bone marrow grafts with osteoconductive agents for the treatment of long-bone nonunion14,41,42 and for spinal fusion43. A 2-mL volume of aspirate taken from the human anterior iliac crest has an average of 2400 alkaline phosphatase-positive (and therefore potentially osteogenic) colony-forming units16. The stand-alone use of platelet-derived growth factors for human posterolateral lumbar fusion has led to mixed results44-47. There is evidence to suggest that the biological effects of platelets may be dose-dependent, with maximum osteogenesis at an optimum concentration and less of an effect at higher concentrations22. The optimum platelet concentration of 109 platelets/mL identified in that study22 was used in the present study.
In the present study, decortication of the mouse transverse processes alone did not result in spine fusion. Similar results were observed by Salamon et al.24 in a rat model, in which none of six animals in the group treated with decortication alone had development of a posterolateral intertransverse fusion mass. In the present study, rhBMP-2, both alone and in combination with bone marrow or platelet-rich plasma, resulted in successful posterolateral lumbar spine fusion. With increases in the area, volume, and density of the fusion mass, the present study quantitatively confirmed a synergistic effect for bone marrow and rhBMP-2 in an in vivo mouse posterolateral lumbar spine fusion model. In the present study, the volume and area of the fusion mass as well as the quality of the fusion mass (as determined by its density) were increased when bone marrow was included. Platelet-rich plasma did not result in increased fusion area, volume, or density in comparison with that obtained with rhBMP-2 alone.
There are several shortcomings with this model system that need to be considered when drawing any conclusions from the present study. First, the mouse model yielded a 100% fusion rate after treatment with BMP-2 alone, thus not providing room for improvement in the primary measure. This high fusion rate may have been due to the smaller animal size for which a thinner fusion mass may not be disrupted as easily with small-scale daily movements following surgery. The high fusion rate also may be due to a high osteogenic response to BMP-2 in the murine system. Using lower BMP-2 concentrations could have yielded a suboptimum fusion rate, as has been elicited with use of autograft alone in this model13, but reducing the BMP-2 amount would have represented another departure from clinical applications of BMP-2. We relied on secondary measures of radiographic density, radiographic area, and computed tomographic volume; with use of these, increased bone mass and density were found with bone marrow supplementation. Although there is no clear evidence that increased bone mass leads to better fusion rates, it is a rational extrapolation given that the goal of spine fusion surgery is new bone growth that achieves a strong mass across the fusion site.
Another possible source of difference between the mouse model and clinical practice is in the quality of bone marrow and the nature and potential of stem cells within the bone marrow preparations. The proportion of stem cells and their pluripotentiality within the mouse bone marrow samples were not evaluated in our system, in which we took a straightforward, clinically simple approach to bone marrow harvest and application. Clearly, this aspect of bone marrow is in need of additional investigation to expand our understanding of stem cell quantity and quality and to determine how bone marrow might be manipulated between harvest and fusion site application to optimize any effect.
The results of the present study should be taken as preliminary but indicative of a potential direction worthy of additional exploration. Synergistic activity between various growth factors and osteoprogenitor cells should be explored as an attempt to increase the success of rhBMP-2-mediated fusion. Understanding the precise mechanisms and synergism involved in osteogenic processes will help to increase the success of spine fusion. Our results show that augmentation of rhBMP-2 with a bone marrow aspirate increases the size and quality of the posterolateral fusion mass in mice. Using the bone marrow aspirate may decrease the need for increasing the dose of rhBMP-2 in patients undergoing posterolateral fusion, minimizing the concerns for any as yet undetermined systemic effects of these higher dosages. Additional studies in higher animals and human clinical trials will be required for implementation of this synergistic activity in clinical practice. 