Experimental Design
This study was divided into three phases. In Phase 1, the proximal part of the femur in seventeen adult male cynomolgus monkeys (Macaca fascicularis, with a mean body weight [and standard deviation] of 7.6 ± 0.2 kg and an age of seven to ten years) contained a 3.5-mm-diameter core defect in the femoral head and neck treated with 360 µg of rhBMP-2/ACS in seven animals, ACS alone in five animals, or no treatment in five animals and was compared with the proximal part of the contralateral, normal femur (Table I). Group sizes were selected in order to detect a significant difference in mean values at a power of 0.8 and a = 0.05 on the basis of the results of a similar study performed in sheep14. Retention of tracer amounts of 125I-rhBMP-2 added to the rhBMP-2 was assessed, with use of a gamma camera, at thirty minutes; at one, three, twenty-four, forty-eight, and seventy-two hours; and at one, two, and three weeks after surgery in the seven animals treated with rhBMP-2/ACS.
All animals were administered intramuscular injections of fluorochrome bone labels. Tetracycline (25 mg/kg) was administered at fourteen and twenty-four days after surgery. Xylenol orange (90 mg/kg) was administered at fourteen weeks after surgery. Calcein (5 mg/kg) was administered at ten and three days before the animals were killed. Ex vivo peripheral quantitative computed tomography and histological evaluations of the proximal part of the harvested femora were performed after the animals were killed at twenty-four weeks.
In Phase 2, the distal end of the femur containing a 3.5-mm-diameter core defect treated with 360 µg of rhBMP-2/ACS in one limb was compared with treatment of a core defect with ACS alone in the distal part of the contralateral limb in the seven rhBMP-2/ACS-treated animals from Phase 1 of this study. Radiographs were made and in vivo peripheral quantitative tomography of the distal femoral core defects in all of the animals was performed prior to surgery; immediately after surgery; and at two, four, eight, sixteen, and twenty-four weeks after surgery. Peripheral quantitative computed tomography and histological evaluations of harvested distal femoral bone segments from all of the animals were performed after they had been killed at twenty-four weeks.
In Phase 3, the histological appearance of the proximal and distal femoral core defects treated with 360 µg of rhBMP-2/ACS in three animals each at one, two, and four weeks after surgery was compared with that of the contralateral proximal femoral core defects treated with ACS alone. Cellular activity in the one-week group was characterized with use of an immunohistochemical evaluation for the presence of tartrate-resistant acid phosphatase (TRAP), osteoprotegerin (OPG), and receptor activator of nuclear factor-?B ligand (RANKL). All procedures were approved by the Institutional Animal Care and Use Committee and were carried out under the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care.
Implant Preparation
The rhBMP-2 protein was produced with use of a CHO cell expression system (Wyeth BioPharma, Andover, Massachusetts)29. The 125I-radiolabeled rhBMP-2 was formulated by New England Nuclear (Boston, Massachusetts) using the IODO-GEN procedure (Pierce, Rockwood, Illinois). The implants were formulated by dripping either 0.24 mL of 1.5 mg/mL of rhBMP-2/buffer solution (a total rhBMP-2 dose of 360 µg) or buffer alone onto collagen sponges that were 15 mm long, 15 mm wide, and 3.5 mm thick. The implants were formulated thirty minutes prior to implantation.
Surgical Procedures
The procedures were performed with use of aseptic technique, with the animals intubated and maintained under general anesthesia with isoflurane. Proximal femoral core defects were made through a stab incision over the lateral aspect of the proximal part of the femur with a 3.5-mm-diameter cannulated drill-bit (Synthes, West Chester, Pennsylvania) inserted over a preplaced 1.25-mm-diameter Kirschner wire with use of fluoroscopic visualization. The core defects were located within the femoral neck and extended to within 3 mm of the femoral articulation. The implants were rolled into a cylindrical shape and were pulled into the defect site with a suture snare insertion device. The defect entry portal was plugged with a dry ACS cylinder. The stab incision was closed with absorbable suture. Distal femoral core defects were created in a similar manner although with a stab incision in the lateral aspect of the distal part of the femur. The defects were centered in the femoral condyles and extended to within 3 mm of the medial cortex. Antibiotics and analgesics (0.01 to 0.03 mg/kg of buprenorphine hydrochloride) were administered intramuscularly twice per day for two days after surgery. Animals were killed by a barbiturate overdose.
Retention of 125I-rhBMP-2 in Proximal Femoral Core-Defect Site
The protocols used to determine 125I-rhBMP-2 retention have been described previously4,30,31. Gamma scintigraphy images were obtained at the designated time points to determine retention of 125I-rhBMP-2 in the proximal part of the femora in Phase 1 of this study. The animals were placed in the appropriate lateral recumbency immediately after treatment, with the limb to be imaged against the gamma camera. The 125I-rhBMP-2 retention profiles were expressed as a percentage of the initial measurement. Noncompartmental pharmacokinetic analysis was performed with use of pharmacokinetic software (WinNonlin; Pharsight, Mountain View, California) to determine the partial area under the percent retention versus time curves.
Peripheral Quantitative Computed Tomography
Bilateral peripheral quantitative computed tomography images (XCT-3000; Stratec, Pforzheim, Germany) of the proximal part of the femora were made from specimens harvested at twenty-four weeks for Phase 1 of this study. Specimens were scanned in conical tubes containing 70% alcohol. Contiguous slices, perpendicular to the femoral neck axis, were obtained, starting at the femoral head and extending to the trochanteric region. The slices were 2.5 mm thick and were placed 2 mm apart, with an in-plane voxel size of 0.2 × 0.2 mm. Core-defect cross-sectional area and the mineral density were determined from a region of interest drawn around a visible area of altered trabecular density created as a result of new bone formation within the core defect compared with the surrounding trabecular bone for the rhBMP-2/ACS-treated animals (Fig. 1). Similar measurements for the ACS alone and untreated animals were obtained from a region of interest drawn manually around the radiolucent core-defect region that remained visible at twenty-four weeks in these animals. The density of the trabecular region surrounding the core defect was determined from a region of interest drawn to exclude the core-defect region and the surrounding cortical bone. Similar measurements were made from corresponding regions of interest from anatomically matched images of the proximal part of the contralateral femur.
In vivo images of the distal part of the femora were acquired at the designated time points from anesthetized animals placed in dorsal recumbency with their hind limbs fixed in a positioning frame. Contiguous slices, perpendicular to the diaphyseal axis, were obtained, starting at the articular surface of the distal part of the femur and extending proximally through the metaphysis. The area of the core defects immediately after surgery and at two and four weeks after surgery were determined by manually drawing a region of interest around the perimeter of the largest visible radiolucent region in the contiguous images. Changes in the density of the trabecular region within the core defects over time were determined by superimposing the region of interest identified as the core defect in the postoperative image on anatomically matched images at subsequent time points. The density of the trabecular region within the core defect and in the surrounding region of the distal part of the femora was determined at twenty-four weeks, with use of the methods described for Phase 1 of this study, from cross-sectional images of specimens harvested after the animals were killed. In all cases, the reported ex vivo values represented the average of measurements obtained from three to four contiguous images within the femoral neck and the distal part of the femur. The coefficient of variation for five repeated measurements of ex vivo core-defect cross-sectional area, core-defect mineral density, and surrounding region mineral density from a single image was 1.9%, 1.8%, and 2.2%, respectively, for the proximal part of a femur and 2.1%, 1.7%, and 2.5%, respectively, for the distal part of a femur treated with rhBMP-2/ACS. Similar values were obtained for measurements of the proximal and distal parts of the femora treated with ACS alone and for those that were untreated.
Histology and Histomorphometry
The methods for processing calcified specimens and preparing thin sections for routine histological evaluation have been described previously32. Qualitative evaluation was performed on both stained and unstained sections with use of a microscope equipped with both transmitted and fluorescent light. Static histomorphometry measurements for Phases 1 and 2 of the study were performed with use of image analysis software (BIOQUANT OSTEO; BIOQUANT Image Analysis, Nashville, Tennessee) on the von Kossa-stained sections with use of black and white digital images displayed on a monitor screen33,34. Dynamic histomorphometry measurements of the surrounding region in Phase 1 of the study were performed with use of the same image analysis software on unstained sections under ultraviolet light fluorescence microscopy33,34. Similar measurements of the core-defect region could not be made because of the diffuse distribution of the calcein fluorochrome labels.
Cross-sectional area and histomorphometric evaluation of the core defects from the rhBMP-2/ACS-treated animals were estimated from a region of interest drawn around a visible area of altered trabecular architecture created as a result of new bone formation within the core defect compared with the surrounding bone. The fluorochrome labeling pattern was used to confirm the proper identification of the core-defect region. Similar measurements for the animals treated with ACS alone and the untreated animals were estimated from a region of interest drawn manually around the radiolucent region created by the core defect that remained visible at twenty-four weeks in these animals. Histomorphometric evaluation of the trabecular region surrounding the core defect in Phase 1 of this study was determined from a region of interest drawn to exclude the core-defect region and the surrounding cortical bone in all of the animals. Similar measurements were made from corresponding regions of interest from anatomically matched images of the proximal part of the contralateral femur. The cross-sectional area of the core defects for Phase 3 of this study was determined from a region of interest drawn around the defect in the preexisting trabecular bone visible at one and two weeks. The reported values represented the average of measurements obtained from three to four contiguous images within the femoral neck and distal part of the femora. The coefficient of variation for five repeated measurements of ex vivo core-defect cross-sectional area, core-defect trabecular volume fraction, and the surrounding region bone volume fraction for a single image was 1.2%, 1.8%, and 1.6%, respectively, for the proximal part of the femur and 2.0%, 1.8%, and 1.4%, respectively, for the distal part of the femur treated with rhBMP-2/ACS. Similar values were obtained for measurements of proximal and distal parts of the femora treated with ACS alone and those that were untreated.
For Phase 3 of the study, decalcified sections from the proximal part of the femora harvested at one week after surgery were also obtained for immunohistochemical characterization of cellular activity. The decalcified sections used for histological analysis were placed in 20% ethylenediaminetetraacetic acid (EDTA) for forty-five days, embedded with paraffin, and sectioned at 5 µm in thickness. After removing the paraffin with xylene, decalcified sections were hydrated with a descending concentration of alcohol and distilled water. Sections were then stained with acid phosphatase in the presence of tartrate to detect the presence of TRAP staining associated with osteoclasts (kit 387; Sigma-Aldrich, St. Louis, Missouri). These sections were counterstained with hematoxylin. Nonspecific antibody binding sites were blocked with normal serum. Additional sections were incubated with antibodies against OPG (1:100 dilution, catalog number sc-11383; Santa Cruz Biotechnology, Santa Cruz, California) and RANKL (1:100 dilution, catalog number sc-7627; Santa Cruz Biotechnology). Bound antibodies were detected with biotinylated secondary antibodies and the biotin-avidin complex (Vector Laboratories, Burlingame, California), with use of diaminobenzidine (DAB) or Vector Red (Vector Laboratories) as the substrate. Additional sections were incubated with nonspecific antibodies, which did not yield positive immunoreactivity. All of the sections were counterstained with hematoxylin stain. The percentage of positive-staining cells of interest (osteoclasts, giant cells, spindle-shaped cells, or osteoblasts, respectively) was determined for each animal by dividing the number of positive-staining cells of interest by the total number of cells of interest counted in each of four fields at 10× magnification for a single slide from each animal. Mean values were obtained by averaging the values for each of the three animals evaluated at one week.
Statistical Analysis
Within-group comparisons of peripheral quantitative computed tomography data and histological measurements were evaluated with use of the paired Student t test. Between-group comparisons for core-defect size and peripheral quantitative computed tomography measurements were evaluated with analysis of variance. Between-group comparisons of histological measurements were evaluated with multivariate analysis of variance. In vivo measurements with use of peripheral quantitative computed tomography of the distal femoral core-defect area and mineral density were evaluated with repeated-measures analysis of variance. When significant group effects were observed, comparisons of all pairs of group means were performed with the Bonferroni-Dunn post hoc test. All tests were two-tailed, and differences were considered significant at p < 0.05.
Source of Funding
Funding for this study was provided by Wyeth Discovery Research.
Phase 1
Retention of 125I-rhBMP-2
Retention of 125I-rhBMP-2 in the proximal part of the femora was a mean (and standard deviation) of 47.1% ± 1.5%, 40.4% ± 7.5%, 39.9% ± 7.6%, 24.2% ± 3.5%, 3.4% ± 1.4%, and 0.5% ± 0.1% of the initial dose at thirty minutes, one hour, and one, seven, fourteen, and twenty-one days, respectively, following implantation in 1.5 mg/mL of rhBMP-2/ACS in core defects (Fig. 2). By comparison, retention of 125I-rhBMP-2 in the proximal part of the femora was a mean of only 6.0% ± 3.9%, 1.46% ± 1.8%, and 0.23% ± 0.01% of the initial dose at thirty minutes, one day, and seven days, respectively, following injection of 4.45 mg/mL of rhBMP-2/buffer solution into core defects in seven nonhuman primates (unpublished data from our laboratory). The partial area under the fractional retention of the initial rhBMP-2/ACS dose versus time curve was a mean of 5.01 ± 2.4 fractional retention days compared with a mean of 0.08 ± 0.04 fractional retention days for rhBMP-2/buffer solution (p < 0.005).
Peripheral Quantitative Computed Tomography
Proximal femoral core defects treated with rhBMP-2/ACS were radiodense at twenty-four weeks (Fig. 1). Core defects treated with ACS alone or left untreated were either partially radiodense or radiolucent and were often surrounded by a thin radiodense rim. The cross-sectional area of new bone within the rhBMP-2/ACS-treated core defects was larger than that of the defects treated with ACS alone (p = 0.0006) and the untreated core defects (p = 0.0018, group effect p < 0.001) at twenty-four weeks (Table II). The mineral density of the core-defect region in the rhBMP-2/ACS-treated defects in the proximal part of the femora, defects treated with ACS alone, and untreated core defects was a mean (and standard deviation) of 80.9% ± 13.9% (p < 0.03), 53.8% ± 2.4% (p < 0.01), and 19.8% ± 0.9% (p < 0.0001), respectively, of the mineral density of the corresponding region in the proximal part of the contralateral femora at twenty-four weeks (see Appendix). This percentage value was greater in the rhBMP-2/ACS-treated group compared with the group treated with ACS alone (p = 0.035) and the untreated group (p < 0.0001, group effect p < 0.0001). The percentage value for the group treated with ACS alone was also greater than that for the untreated group (p = 0.003). The mineral density of the trabecular region surrounding the core defects treated with rhBMP-2/ACS, the core defects treated with ACS alone, and the untreated core defects was a mean of 111.7% ± 9.1% (p < 0.02), 104.9% ± 4.7% (p = 0.21), and 103.9% ± 3.8% (p = 0.28), respectively, of the mineral density of the corresponding region in the proximal part of the contralateral femora at twenty-four weeks (see Appendix). The percentage value was greater in the rhBMP-2/ACS-treated group compared with the group treated with ACS alone (p = 0.037) and the untreated group (p < 0.025, group effect p = 0.045). The percentage value for the group treated with ACS alone was similar to that for the untreated group (p = 0.83).
Histological Analysis
Proximal femoral core defects treated with rhBMP-2/ACS were filled with trabecular bone at twenty-four weeks (Fig. 3). The core defects treated with ACS alone or those left untreated were either partially filled with trabecular bone or remained empty and were often surrounded by a thin rim of dense bone. Cross-sectional area was greater in the rhBMP-2/ACS-treated core defects compared with defects treated with ACS alone (p = 0.005) and untreated core defects (p = 0.003, group effect p < 0.003; Table II). Histological measurements of cross-sectional area were similar to the peripheral quantitative computed tomography measurements (Table II).
The trabecular volume fraction (and standard deviation) within the defects treated with rhBMP-2/ACS, those treated with ACS alone, and untreated core defects was a mean (and standard deviation) of 93.6% ± 18.6% (p = 0.33), 35.9% ± 12.8% (p < 0.0007), and 30.6% ± 9.1% (p < 0.0001), respectively, of the corresponding region in the proximal part of the contralateral femora at twenty-four weeks (see Appendix). The percentage value was greater in the rhBMP-2/ACS-treated group compared with the group treated with ACS alone (p = 0.0001) and the untreated group (p = 0.0001, group effect p = 0.0001). The percentage value of the group treated with ACS alone was similar to that for the untreated group (p = 0.44). The increase in trabecular volume fraction in the rhBMP-2/ACS-treated core defects was due primarily to a mean increase of 42.4% ± 30.0% in trabecular number (p < 0.02) compared with the corresponding region in the proximal part of the contralateral femora. The trabecular thickness in the rhBMP-2/ACS-treated core-defect region was 67.8% ± 3.9% (p < 0.0001) of the value for the corresponding region in the proximal part of the contralateral femora. The trabecular volume fraction of the region surrounding the core defects in the proximal part of femora treated with rhBMP-2/ACS, those treated with ACS alone, and those left untreated was 117.2% ± 10.4% (p < 0.004), 107.7% ± 5.8% (p = 0.19), and 106.8% ± 6.0% (p = 0.19), respectively, of the corresponding region in the proximal part of the contralateral femora at twenty-four weeks (see Appendix). The percentage value was greater in the group treated with rhBMP-2/ACS compared with the group treated with ACS alone (p < 0.018) and the untreated group (p < 0.02, group effect p < 0.027). The percentage value for the group treated with ACS alone was similar to that for the untreated group (p = 0.91). The increase in the trabecular volume of the region surrounding the rhBMP-2/ACS-treated core defects was due primarily to a mean increase of 16.7% ± 8.6% in trabecular thickness (p < 0.002) compared with the corresponding region in the proximal part of the contralateral femora. The trabecular number in the region surrounding the core defects treated with rhBMP-2/ACS was 98.1% ± 7.0% (p = 0.59) of the value for the corresponding region in the proximal part of the contralateral femora.
The mineralized surface of the region surrounding core defects in the group treated with rhBMP-2/ACS, the group treated with ACS alone, and the group of untreated core defects was 209.6% ± 60.8% (p < 0.0001), 127.6% ± 19.6% (p < 0.04), and 140.0% ± 14.9% (p < 0.005), respectively, of the mineralized surface of the corresponding region in the proximal part of the contralateral femora (see Appendix). The percentage value was greater in the group treated with rhBMP-2/ACS compared with the group treated with ACS alone (p < 0.0018) and the group that had no treatment (p < 0.003, group effect p < 0.0025). The percentage value for the group treated with ACS alone was similar to that for the group that had no treatment (p = 0.80). The mineral apposition rate of the region surrounding the core defects treated with rhBMP-2/ACS, those treated with ACS alone, and the untreated core defects was 205.0% ± 91.2% (p < 0.003), 116.0% ± 8.3% (p < 0.01), and 129.1% ± 12.3% (p < 0.01), respectively, of the mineral apposition rate of the corresponding region in the proximal part of the contralateral femora. The percentage value was greater in the group treated with rhBMP-2/ACS compared with the group treated with ACS alone (p < 0.008) and the group that had no treatment (p < 0.02, group effect p < 0.02). The percentage value for the group treated with ACS alone was similar to that for the group that had no treatment (p = 0.69). The bone formation rate in the region surrounding the core defects treated with rhBMP-2/ACS, those treated with ACS alone, and those left untreated was 1075.7% ± 510.3% (p < 0.0009), 134.9% ± 23.5% (p < 0.005), and 127.8% ± 25.3% (p < 0.05), respectively, of the bone formation rate of the corresponding region in the proximal part of the contralateral femora. The percentage value was greater in the group treated with rhBMP-2/ACS compared with the group treated with ACS alone (p < 0.0002) and the group that had no treatment (p < 0.0001, group effect p < 0.0001). The percentage value for the group treated with ACS alone was similar to that for the group that had no treatment (p = 0.96).
The presence of diffuse tetracycline label in the woven bone cores of trabecular bone within the core-defect region indicates mineralization of de novo trabecular bone initiated between fourteen and twenty-four days after surgery in response to treatment with the rhBMP-2/ACS (Fig. 4). The presence of a band of xylenol orange label surrounding the tetracycline-labeled trabecular bone cores indicates mineralization of new lamellar trabecular bone was ongoing at twelve weeks after surgery (Fig. 4, C). The presence of single and double bands of calcein label on the surfaces of trabecular bone within the core defects indicates that mineralization of new bone was ongoing at twenty-four weeks (Fig. 4, C). The presence of sequential distinct tetracycline, xylenol orange, and calcein labels in the trabecular bone surrounding the core defects indicates appositional new bone formation initiated on existing trabeculae in this region at fourteen days and was ongoing during the six-month duration of the study (Fig. 4, E). The presence of distinct xylenol and calcein labels, indicating mineralization of appositional new bone in the dense rim surrounding the core defects in the proximal part of the femora treated with ACS alone (Fig. 4, B and D) and those left untreated, initiated at twelve weeks and was ongoing at twenty-four weeks. Sporadic calcein labeling on the surface of trabecular bone surrounding the core defects in the group treated with ACS alone and the group that had no treatment indicates ongoing normal remodeling activity (Fig. 4, F).
Phase 2
Radiography and Peripheral Quantitative Computed Tomography
Radiographic and peripheral quantitative computed tomographic images of the distal part of the femora demonstrated that core defects treated with rhBMP-2/ACS increased in size by two weeks after surgery compared with images obtained immediately after surgery and compared with the core defects in the contralateral femora treated with ACS alone at the same time points (Fig. 5). A progressive increase in radiodensity was observed within the rhBMP-2/ACS-treated core-defect region during the remainder of the study. The core defects treated with ACS alone were often surrounded by a thin radiodense rim but remained only partially filled with radiodense material during the same time period. Peripheral quantitative computed tomography measurements confirmed an increase in core-defect area in the distal part of femora treated with rhBMP-2/ACS at two weeks after surgery (mean, 65.7 ± 8.0 mm2) compared with the values obtained immediately after surgery (mean, 36.3 ± 3.5 mm2; p <0.001) and at four weeks after surgery (mean, 47.1 ± 8.8 mm2; p < 0.02, p < 0.0001 for group effect of time). No change in core-defect area in the distal part of femora treated with ACS alone was seen at two weeks after surgery (mean, 36.9 ± 9.0 mm2) compared with the values obtained immediately after surgery (mean, 35.5 ± 9.0 mm2; p = 0.78) and at four weeks after surgery (mean, 37.3 ± 9.4 mm2; p = 0.73, p = 0.95 for group effect of time). Ex vivo measurements of the cross-sectional area of new bone within the rhBMP-2/ACS-treated core defects were larger compared with treatment with ACS alone (p = 0.0006) at twenty-four weeks (Table II). The mineral density of the core-defect region in the distal part of femora treated with rhBMP-2/ACS was a mean of 173.9% ± 26.7% greater (p < 0.0001) than the value for the core-defect region in the contralateral femora treated with ACS alone at twenty-four weeks (see Appendix). The mineral density of the trabecular bone region surrounding the core defects in the distal part of femora treated with rhBMP-2/ACS was 116.5% ± 7.0% (p < 0.001) of the corresponding value for the distal part of the contralateral femora treated with ACS alone at twenty-four weeks (see Appendix).
Histological Analysis
As was the case with the proximal femoral defects, distal femoral core defects treated with rhBMP-2/ACS were filled with trabecular bone at twenty-four weeks (see Appendix). Nonmineralized osteoid was still present on the surfaces of new trabecular bone within the rhBMP-2/ACS-treated core defects and on surrounding preexisting trabecular bone. The core defects treated with ACS alone were partially filled with trabecular bone and were surrounded by a thin rim of dense bone at twenty-four weeks. Ex vivo measurement of the cross-sectional area of new bone within the rhBMP-2/ACS-treated core defects was larger than that for the defects treated with ACS alone (p < 0.0004) at twenty-four weeks (Table II). Histological measurements of core-defect cross-sectional area were similar to the peripheral quantitative computed tomography measurements (Table II). The trabecular volume fraction within the core-defect region in the distal part of the femora treated with rhBMP-2/ACS was 192.8% ± 74% (p < 0.006) of the corresponding value in the distal part of the contralateral femora treated with ACS alone at twenty-four weeks (see Appendix). The increase in trabecular volume fraction in the rhBMP-2/ACS-treated core-defect region was due to an increase of 44.5% ± 29.6% in trabecular thickness (p < 0.007) and an increase of 53.6% ± 36.9% in trabecular number (p < 0.008) compared with the corresponding region in the distal part of the contralateral femora. The trabecular volume fraction surrounding the core-defect region in the distal part of the femora treated with rhBMP-2/ACS was 117.7% ± 9.7% (p < 0.007) of the corresponding value in the distal part of the contralateral femora treated with ACS alone at twenty-four weeks (see Appendix). The increase in trabecular volume in the rhBMP-2/ACS-treated core-defect region was due to an increase of 16.3% ± 11.5% in trabecular thickness (p < 0.02). Trabecular number (p = 0.78) was similar in the corresponding region of both groups of distal femora.
Phase 3
Histological evaluation demonstrated no difference in the size of proximal femoral core defects treated with rhBMP-2/ACS and proximal femoral core defects in the contralateral limb treated with ACS alone in the three animals evaluated at one week after surgery (9.5 compared with 9.7 mm2, 9.8 compared with 10.0 mm2, and 9.9 compared with 9.5 mm2; Fig. 6; see Appendix). Multinucleated cells, identified as osteoclasts on the basis of the presence of positive TRAP staining in 95.0% ± 1.5% of the cells, were observed initiating resorption of trabecular bone around the perimeter of the core defects at this time point in the proximal part of the femora treated with rhBMP-2/ACS (Fig. 7, A; see Appendix). Multinucleated cells, identified as giant cells on the basis of the lack of TRAP staining in 80.0% ± 10.5% of the cells, were actively resorbing the perimeter of the ACS (Fig. 7, B; see Appendix). The granulation tissue between the resorbing ACS and the rim of the core defect and the marrow spaces surrounding the trabecular bone at the perimeter of the core defects were highly cellular, containing numerous macrophages, monocytes, and spindle-shaped cells. Approximately 50.0% ± 10.2% of the spindle-shaped cells stained positive for RANKL (see Appendix). Only 12.3% ± 5.1% of these cells exhibited positive staining for OPG (see Appendix). Similar results were observed in the distal part of the femora. Histological evaluation of the proximal femoral core defects in the contralateral limb treated with ACS alone demonstrated the presence of both de novo trabecular bone formation and appositional bone formation on existing trabeculae at the perimeter of the core defect at one week (see Appendix). There was no evidence of bone resorption or ACS resorption at this time point. Only 14.2% ± 7.0% of the spindle-shaped cells exhibited positive staining for RANKL (see Appendix). Approximately 50.5% ± 10.5% of the osteoblasts on preexisting trabecular bone exhibited positive staining for OPG (see Appendix). Similar results were observed in the distal part of the femora.
At two weeks after surgery, the proximal femoral core defects treated with rhBMP-2/ACS were larger and had irregular borders compared with the core defects in the contralateral femora treated with ACS alone (Fig. 6, C; see Appendix). Multinucleated cells were observed resorbing trabecular bone around the perimeter of the core defect. Residual ACS occupied the central third of the core defect. The marrow spaces remained highly cellular and vascularized. Continued de novo trabecular bone formation and appositional bone formation on existing trabeculae were observed in the region immediately surrounding the core defects treated with ACS alone in the proximal part of the contralateral femora at two weeks (Fig. 6, D). This bone formation process resulted in walling off the perimeter of the core defects. Resorption of ACS by multinucleated cells was evident at this time point. Cross-sectional area was greater in the three animals with rhBMP-2/ACS-treated core defects compared with the core defects treated with ACS alone in the contralateral femora (16.7 compared with 9.2 mm2, 14.5 compared with 9.8 mm2, and 15.8 compared with 9.5 mm2) at two weeks. Similar results were observed in the distal part of the femora (see Appendix). Core-defect cross-sectional area was greater in the distal part of the femora treated with rhBMP-2/ACS compared with the distal part of the contralateral femora treated with ACS alone (18.8 compared with 11.6 mm2, 22.0 compared with 11.7 mm2, and 32.6 compared with 12.6 mm2). The larger increase in distal femoral core-defect size was likely due to the decreased trabecular bone volume fraction in the distal part of the femora compared with the proximal part of the femora.
At four weeks after surgery, the enlarged core defects treated with rhBMP-2/ACS were filled with thin, closely packed, nonmineralized and mineralized trabeculae (Fig. 6, E; see Appendix). There was a considerable amount of appositional new-bone formation on the surrounding trabeculae. There were few macrophages, monocytes, or multinucleated cells and no evidence of continued bone resorption. The marrow spaces within the core defect and surrounding trabeculae contained mainly spindle-shaped cells and were highly vascular. There was no evidence of residual ACS at four weeks after surgery. Histological evaluation of the proximal femoral core defects in the contralateral limb treated with ACS alone demonstrated consolidation of the new bone formation around the perimeter of the core defect into a dense rim (Fig. 6, F). De novo trabecular bone formation was also present in the space created by the resorbing ACS, which still occupied the central two-thirds of the core defect.
This study demonstrates that mineral density and trabecular volume were increased within core defects and the surrounding trabecular bone in the proximal femoral defects treated with rhBMP-2/ACS compared with proximal femoral core defects treated with ACS alone and untreated core defects at twenty-four weeks. Increased trabecular volume within the rhBMP-2/ACS-treated core defects was primarily due to de novo bone formation. Increased trabecular volume in the region surrounding the core defect was due primarily to appositional bone formation on existing trabeculae. Bone formation in response to treatment with rhBMP-2/ACS was preceded by transient bone resorption. Bone resorption was initiated as early as one week following surgery and peaked at two weeks following surgery. By four weeks after surgery, bone formation predominated in both the core defects and surrounding trabecular bone.
Transient bone resorption prior to bone formation was reported following implantation of rhBMP-2/ACS in a sheep core-defect model11,14, a sheep spine fusion model15, and a canine tibial bone tunnel model12. Similar results were reported with use of rhBMP-2/buffer dried onto hydroxyapatite-tricalcium phosphate-coated titanium implants placed in the proximal part of the humerus in a canine model22. Bone resorption was also reported in association with use of rhBMP-7 (osteogenic protein-1) delivered in bovine demineralized bone matrix in metaphyseal bone16,18-20.
The role of BMPs in regulating bone resorption compared with bone formation appears to be the result of regulation by BMPs of factors responsible for osteoclast precursor expansion such as macrophage colony-stimulating factor (M-CSF), osteoclast differentiation factors (RANKL), osteoclast inhibitory factors (OPG), and direct effects on osteoclasts35-39. The ratio of RANKL and OPG expression by osteoblast precursors compared with mature osteoblasts has been demonstrated to regulate the balance between bone formation and bone resorption in cell culture models40. The increased RANKL-OPG ratio observed in osteoblast progenitor cells promoted bone resorption. The decreased RANKL-OPG ratio observed in mature osteoblasts promoted bone formation. In this study, the ratio of RANKL to OPG expression in spindle-shaped cells within the rhBMP-2/ACS-treated core defects was high at one week after surgery. Large numbers of macrophages and monocytes, which are the precursor cells for osteoclasts and giant cells, were also present. The juxtaposition of these cell types is likely responsible for the induction of osteoclasts responsible for subsequent bone resorption. In contrast, large numbers of mature OPG-positive stained osteoblasts and fewer RANKL-positive stained spindle-shaped cells were likely responsible for bone formation observed in the core defects treated with ACS alone at one week. Osteoclast apoptosis, or depletion of the osteoclast precursor population, may have been responsible for the observed reversal of bone resorption and subsequent bone formation at four weeks. The transition from bone resorption to bone formation at four weeks could also occur if there was a decrease in the RANKL-OPG ratio in response to the presence of mature osteoblasts observed in the rhBMP-2/ACS-treated core defects at this time point.
Compression of the 0.24-mL volume of rhBMP-2/ACS into a 0.15-mL volume within the 3.5-mm-diameter core defect may also have contributed to the degree of transient bone resorption compared with uncompressed rhBMP-2/ACS placed within interbody cages used for spine fusions11,15. Compression of the collagen sponge may result in a local increase of rhBMP-2 concentration compared with uncompressed collagen sponges because of the >95% retention of the rhBMP-2 on the collagen sponge observed after fifteen minutes of equilibration despite surgical manipulation and compression during placement41.
Transient bone resorption was not observed to precede bone formation following the administration of rhBMP-2 in a calcium phosphate matrix in nonhuman primate metaphyseal core defects42, canine proximal tibial osteotomies43, and canine acetabular components grouted with the same rhBMP-2-carrier combination44 at concentrations similar to rhBMP-2/ACS studies. The carrier-dependent differential response to rhBMP-2 in metaphyseal bone may be related to the release rate of rhBMP-2 from the carrier and the subsequent effect on signaling between osteoblasts and osteoclasts. The more rapid release of rhBMP-2 from the collagen sponge may stimulate a more aggressive response compared with the slower release of rhBMP-2 from the calcium phosphate matrix45. In addition, transient bone resorption was not reported in response to rhBMP-2 or rhBMP-7 administration in animal models of diaphyseal bone repair31,32,45-49 or in patients with open tibial diaphyseal fractures10 or tibial diaphyseal nonunions50. The differential response in metaphyseal compared with diaphyseal bone may simply be due to the availability of appropriate responding cells in the marrow compared with the soft tissue surrounding bone.
Treatment of the core defect with either rhBMP-2/ACS or ACS alone involved de novo bone formation within the defects and appositional bone formation on existing trabeculae surrounding the defects. There was no evidence of endochondral bone formation at any time point. Interestingly, bone formation was initiated as early as one week after surgery in the core defects treated with ACS alone. Bone formation in the rhBMP-2/ACS-treated core defects was not observed until four weeks after surgery. Delayed bone formation in response to rhBMP-2/ACS treatment may have been the result of soluble factors secreted by osteoclasts that inhibit osteoblast differentiation51,52.
One limitation of this study was the estimation of increased core-defect cross-sectional area in the proximal part of the femora treated with rhBMP-2/ACS on the basis of peripheral quantitative computed tomography and histological evaluation at twenty-four weeks. In vivo measurements with use of peripheral quantitative computed tomography confirmed the increase in rhBMP-2/ACS-treated core-defect size observed at twenty-four weeks. In addition, the estimates of core-defect cross-sectional area at twenty-four weeks with use of these two modalities correlated well not only with each other but with direct measurements of core-defect cross-sectional area from histological specimens obtained two weeks after surgery.
This study demonstrates that implanting rhBMP-2/ACS into core defects in the proximal part of the femur of cynomolgus monkeys induces both de novo and appositional bone formation at twenty-four weeks. However, the observed transient bone resorption mandates caution in considering the use of rhBMP-2/ACS for metaphyseal fracture repair when bore resorption may lead to loss of fixation or structural support.