Study Design
Pilot Study
External fixators of three different stiffnesses were constructed as described previously21 and used in a pilot study to determine their influence on the first two weeks of bone-healing in the presence of BMP-2. This study had two aims: (1) to test our hypothesis that the fixator with the least stiffness would promote the most rapid early healing response and (2) to determine a suitable time for increasing fixator stiffness.
Thirty-six rats underwent surgery to create 5-mm femoral defects and were divided into three equal groups of animals that received low, medium, or high-stiffness fixators. All animals received BMP-2. Radiographs of the femoral defects were made after nine and fourteen days. Four rats per group were killed at six, nine, and fourteen days after surgery, and their femora were processed for histological analysis.
The subsequent reverse dynamization protocol was determined by the radiographic and histological data from the pilot study.
Reverse Dynamization
On the basis of the pilot study data, reverse dynamization was implemented by switching from low-stiffness to high-stiffness fixators at day 14. Healing of these animals was compared with that in animals whose low, medium, and high-stiffness fixators were not changed. A small number of control animals that did not receive BMP-2 were included to confirm that the defects did not heal spontaneously. Defects of all animals were monitored with weekly radiographs. At eight weeks, all animals were killed. All specimens were assessed with dual x-ray absorptiometry and microcomputed tomography (micro-CT). Nine specimens from each group were subjected to mechanical testing, and three were processed for histological analysis.
The design of this experiment is summarized in the Appendix.
Methods
Fixators
Defects were stabilized with custom-made external fixators, as described previously21. Their key features are interchangeable connection elements of different stiffnesses. The present work evaluated connection elements with stiffnesses of 114, 185, and 254 N/mm.
Surgery
A 5-mm, critical-sized, midfemoral defect was created in the right hind limb of each rat. We described this model previously, confirming that it does not spontaneously heal but heals when 11 μg of rhBMP-2 is inserted into the defect22,23. Groups of rats were maintained for eight weeks with each of these fixators. An additional group underwent reverse dynamization, whereby a low-stiffness fixator was applied for the first two weeks and then was switched to high-stiffness fixator (see Appendix). The number of control animals was low because of extensive historical data confirming that these defects do not heal spontaneously22,23. The numbers of animals in the treatment groups were based on historical data confirming sufficient statistical power22,23.
Animal care and experimental protocols were followed in accordance with National Institutes of Health guidelines and were approved by our institution’s Institutional Animal Care and Use Committee.
The surgical procedure has been described in detail previously21. Briefly, male Sprague-Dawley rats weighing between 325 and 360 g were anesthetized with isoflurane (2% with 2 L/min O2 by air mask). Before surgery, each rat was given an antibiotic (20 mg/kg of cefazolin) and the analgesic buprenorphine (0.08 mg/kg) intramuscularly in the left leg. An incision of approximately 3.5 to 4 cm was made through the skin, and the shaft of the femur was exposed.
The external fixator bar was used as a positioning guide to permit reproducible positioning of four drill-holes to accommodate the screws used to secure the fixator. After the fixator was in place, a saw guide was used to make the 5-mm segmental defect with use of a 0.22-mm wire Gigli saw. After the defect was created, the saw guide was removed and a collagen sponge impregnated with rhBMP-2 (11 μg in 100 μL of saline solution) was added to the defect area. Control defects received a sponge lacking BMP-2. The wound was closed in layers. On the first three postoperative days, the rat was given analgesic every twelve hours and antibiotic every twenty-four hours.
Radiographic Evaluation
Bone-healing was evaluated with serial radiography with use of a digital dental x-ray unit (Heliodent DS; Sirona, Bensheim, Germany). While under general anesthesia, the rats were placed in a ventral position and the hind limb was laterally rotated so that the external fixator was not in the path of the x-ray source.
Histological Analysis
Femora were fixed for histological analysis in 4% ice-cold paraformaldehyde and were decalcified for six to eight hours in RDO Rapid Decalcifier (Apex Engineering, Aurora, Illinois), testing with a needle as the decalcification proceeded. Fixed and decalcified tissues were dehydrated in graded ethanol up to 100%, transferred to xylene, and embedded in paraffin. Five-micrometer paraffin sections were placed on poly-L-lysine-coated slides, dried overnight, and evaluated immediately or stored at 4°C. Sections were stained with hematoxylin-eosin or safranin orange-fast green.
Microcomputed Tomography
Femora were scanned with use of a desktop microtomographic imaging system (μCT40; Scanco Medical, Bassersdorf, Switzerland) equipped with a 10-mm focal spot microfocus x-ray tube. Femora were scanned with use of a 20-μm isotropic voxel size, at 55 keV of energy, 200-ms integration time, with approximately 720 micro-CT slices per specimen. Evaluation was done only in the 4-mm (200 slices) central defect region to ensure that no preexisting cortical bone was included in the analyses. To evaluate the region of interest, we assessed the following variables: total cross-sectional area or the callus size of the defect (TA in square millimeters), bone area (BA in square millimeters), and bone area over total area (BA/TA in square millimeters). Polar moment of inertia (in millimeters to the fourth power) was also calculated from micro-CT images. Images were thresholded with use of an adaptive-iterative algorithm, and morphometric variables were calculated from the binarized images using direct, three-dimensional techniques that do not rely on any prior assumptions about the underlying structure.
Dual X-Ray Absorptiometry
Bone mineral content (in grams) of the defect region was measured by dual x-ray absorptiometry (Lunar PIXImus II; GE Medical Systems, Madison, Wisconsin). Specimens were placed on a Lucite block during scanning to simulate soft tissue. The scans were acquired with use of the small-animal high-resolution mode.
Ex Vivo Torsion Testing
Specimens were tested to failure in torsion with use of a materials testing system (Synergie 200; MTS Systems, Eden Prairie, Minnesota) to determine the mechanical properties of the healed defect in shear. Before the test, both ends of each specimen were embedded in polymethylmethacrylate to provide a reproducible gripping interface with the testing fixture. All femora were tested to failure under regular deformation control and at the constant deformation rate of 5 rad/min. Angular deformation and applied load data were acquired at 10 Hz. The torque and rotation data were used to calculate the torsional stiffness and strength of the healed defect.
Sample Size and Statistical Analysis
The numbers of animals used were determined by our historical data using this model, in which eight to ten animals per group proved adequate23. A sample-size power analysis showed that the probability of detecting a significant difference with the numbers of animals per group used in our experiments was 95% for all of the imaging (twelve animals) and 85% for mechanical testing data (nine animals). The sample-size study detected a relationship between the independent and dependent variables at a two-sided 5% significance level. The power test confirmed that the animal numbers selected for each modality would be more than sufficient to detect significant differences.
Comparisons of continuous variables between two treatment groups were performed with use of a two-tailed t test, and comparisons between three groups were done with use of one-way analysis of variance. If the difference between the contralateral femur and the treatment groups was significant, a post hoc (Tukey) test was performed. A power analysis after the study was calculated to determine if we had sufficient animals per group for a significant difference. The power levels for all of the data were found to be from 0.8 to 1. Thus, the numbers of animals per group used in these studies is enough to determine a 5% difference between the test groups. All tests were two-tailed, with differences considered significant at p < 0.05. Data are presented as the mean and the standard error of the mean, unless otherwise noted.
Source of Funding
Funding was provided by the AO Foundation (grant S-08-42G), the U.S. Department of Defense (grant W81XWH-10-1-0888), and the National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health (grant R01AR50243).
Pilot Study
To determine a promising time for changing the degree of stiffness and to determine which stiffness fixators to use, we conducted a pilot experiment that focused on the early events in healing. There were three groups representing rats with segmental defects stabilized by low, medium, or high-stiffness fixators. All experimental defects received BMP-2.
Examination of the radiographic images at day 9 (Fig. 1) shows a faint crescent of radiopacity for the low-stiffness fixator group and, to a lesser degree, the medium-stiffness fixator group in the region opposite to the fixator. Little or no defined radiodensity was evident with the high-stiffness fixator group. At two weeks, this ranking was maintained with diffuse radiopacity in defects stabilized with the high-stiffness fixator and obvious, bridging radiopacity with the low-stiffness and medium-stiffness fixator groups. A visual, qualitative assessment of the radiographs suggested that the size of the radiodense callus was greatest for the low-stiffness fixator group and least for the high-stiffness fixator group.
Qualitative histological examination of the defects during this period was consistent with the radiographic findings (Figs. 2-A and 2-B). Staining with hematoxylin and eosin at six days suggested very little intralesional biological activity, and only the collagen sponge was visible in the defect. In the groups with the two lower-stiffness fixators, there was evidence of a periosteal reaction adjacent to the defect gap around the periosteum.
However, by nine days, the defects were filled with new tissue. In the low-stiffness and high-stiffness fixator groups, there was marked thickening of the periosteum, which appeared to migrate across the defect forming a bridge of neoperiosteum. This was more prominent on the side opposite to the fixator. With the low-stiffness fixator, and to a lesser extent the medium-stiffness fixator, there was evidence of new bone formation, often around the collagen sponge, with periosteal new bone formation on the bone adjacent to the defect. This presented as the formation of external callus, with the defect gap completely filled with soft tissue. Defects supported by high-stiffness fixators, in contrast, had no external callus and contained only fibrous soft tissue with little evidence of bone. Although, as in the other groups, there was marked woven bone formation along the periosteum adjacent to the defect, bridging did not occur.
At two weeks, qualitative examination of the histological sections suggested there was robust formation of woven bone with the low-stiffness fixator. Defects supported by the high-stiffness fixator appeared to contain only a little bone.
The defects stabilized with the medium-stiffness fixator had a distinctive gap in the middle of the defect. As described in the next section of the Results, the same feature was also observed on histological analysis and micro-CT images at eight weeks. Sections stained with safranin orange-fast green (Fig. 2-B) showed little evidence of cartilage formation, beyond a few isolated flecks, in defects stabilized with any fixator.
On the basis of these data, we decided, for the reverse dynamization stage of this study, to initiate fixation with the low-stiffness fixator because it gave the most rapid early healing and to switch to the high-stiffness fixator at two weeks to promote the subsequent formation and remodeling of bone.
Reverse Dynamization
Control defects that did not receive BMP-2 (see Appendix) did not heal (data not shown). All other groups received BMP-2 (see Appendix) and mounted healing responses that varied according to the stiffness of the fixator and whether it was dynamized. These responses are described below.
As noted above, two weeks after surgery, the 5-mm defect stabilized with the low-stiffness fixator contained considerable calcified tissue as shown on the radiographs (Figs. 1 and 3). At this point, fixator stiffness was changed from the low-stiffness to the high-stiffness fixator. One week later, the radiographs revealed complete callus bridging with osseous tissue and no evidence of radiolucent lines in the reverse dynamization group. In contrast, soft tissue persisted until four weeks after the surgery in the defects stabilized with the low-stiffness and medium-stiffness fixators and for at least six weeks after surgery in the defects stabilized with the high-stiffness fixator.
After two weeks, the most obvious radiographic change in the reverse dynamization group was an apparent reduction in the width of the bone in the region of the defect by week 4. This phenomenon was not observed in the low-stiffness and medium-stiffness fixator groups until after six weeks. In the groups with the two lower-stiffness fixators, the width of the bone appeared to increase as healing progressed until the end of treatment, whereas in the group with the most rigid fixator, there was no change throughout the entire experiment.
The radiographs further suggest accelerated formation of new cortices in defects subjected to reverse dynamization. This is supported by the histological findings at eight weeks (Fig. 4). Defects subjected to reverse dynamization appeared to be narrower in cross section and had an organized tissue structure, with better architecture; well-formed, evenly distributed neocortices; and only limited trabecular bone, likely because of advanced remodeling. All other defects had persistent callus and contained disorganized woven bone with poor cortication. Defects stabilized for eight weeks with the medium-stiffness fixator maintained the central gap in the defect that was noted earlier (Figs. 2-A and 2-B) and was surrounded by unmineralized soft tissue. Defects stabilized for eight weeks with the high-stiffness fixator contained a prominent band of cartilage (Fig. 4), raising the possibility of development into a nonunion. Cartilage was not seen in any of the other groups at eight weeks.
The main conclusions histologically were confirmed by visual inspection of the high-resolution micro-CT images (Figs. 5-A, 5-B, and 6). The cross-sectional and longitudinal images shown in Figures 5-A and 5-B are consistent with the histological findings in showing more uniform and complete neocortices and less apparent trabecular bone, while appearing smaller in cross section. Reduced cross-sectional area was confirmed quantitatively (Fig. 6). These changes translated into images in which the cortical bone appeared thicker and trabecular bone less abundant compared with the low-stiffness, medium-stiffness, and high-stiffness fixator groups. Furthermore, in the reverse dynamization group, the new cortical bone had an even circumference over the entire length of the healed defect. This was not observed in the groups with constant stiffness.
Quantitative analysis (Fig. 6) confirmed that the bone mineral content of the reversed dynamization group was closer to normal values. The cross-sectional area of bone in the defects healed under reverse dynamization was only 19% higher than normal, whereas the cross-sectional areas in the other groups were considerably higher. The total cross-sectional area of the defects healed under reverse dynamization was also considerably lower. This is consistent with the smaller callus and higher degree of bone formation formed under reverse dynamization. The polar moment of inertia is a quantity used to predict an object’s ability to resist torsion. All experimental groups had polar moment of inertia values greater than those of the control femur.
The femoral defect stiffness following healing under reverse dynamization was significantly higher than that of the intact, contralateral femur and all experimental groups apart from the group stabilized with the low-stiffness fixator (Fig. 7-A). We attribute the high stiffness of the latter to the very large osseous callus (Fig. 4). Although this is disorganized, the sheer mass of bone endows considerable stiffness. Defects healed under conditions of reverse dynamization were considerably stronger than the intact femur and the defects stabilized with low-stiffness, medium-stiffness, and high-stiffness fixators (Fig. 7-B). The combination of high stiffness and unremarkable torque of the defects healed with the low-stiffness fixator means that they are more brittle than normal.
These data support our hypothesis that reverse dynamization should improve the healing of a large segmental bone defect, but indicate a different mode of action from the one we suggested. In particular, initial stabilization of the defect with the least stiff of the tested fixators was predicted to promote chondrogenesis. Instead, little evidence of cartilage formation was seen during the first two weeks of healing. It is possible that the times selected for histological analysis (six, nine, and fourteen days) were not appropriate; however, in that case, endochondral bone formation would have to have occurred very rapidly between six and nine days after surgery.
Our erroneous assumption that healing of a large segmental defect in the rat femur in response to BMP-2 would occur via endochondral ossification stems from a report by Yasko et al.19, who described this process in a similar rat model. Unlike our study, Yasko et al.19 delivered BMP-2 on a demineralized bone matrix that could have supplied additional growth factors favoring chondrogenesis. Alternatively, mechanical factors may be responsible, given the high sensitivity of this system to the mechanical environment noted in the present work. Instead of using external fixation, Yasko et al.19 employed a plate, which was likely quite stiff, although it was not mechanically characterized. In this regard, it is interesting that our stiffest fixator was the only one that led to substantial cartilage deposition.
Histological examination of the early period of healing provided additional valuable information. It was noteworthy that, six days after the insertion of a collagen sponge impregnated with BMP-2, the defects appeared biologically inert and only the sponge was visible. The only obvious biological response was activation of the periosteum adjacent to the cut ends of the defect. It is possible that the sponge prevented the formation of a hematoma and delayed the entry of osteoprogenitor cells. Three days later, the histological findings changed considerably, with defects filled with abundant soft tissue and evidence of osteogenesis in the groups in which low-stiffness and medium-stiffness fixators were used. This is intriguing, given that nearly all of the implanted BMP-2 would have disappeared within the first few days. Further research into the early response of segmental defects to BMP-2 is warranted to provide insight into the mechanism of healing.
The early differences in healing noted with the different fixators were still evident at eight weeks. The low-stiffness fixator produced the largest amount of bone and the high-stiffness fixator, the least. Little or no external callus was visible on radiographs at any time or by micro-CT or histological analysis at eight weeks, and the high-stiffness fixator, unlike the other fixators, generated substantial cartilage. This was evident as a cartilaginous band across the defect at eight weeks and raises concerns about an eventual nonunion. Since this cartilaginous tissue was not seen at the earlier time points (six to fourteen days), its formation appears to be related purely to the mechanical stimuli generated by the fixator and not by the BMP-2, which should have been absent. The medium-stiffness fixator produced the unusual effect of a gap in the center of the defect, which was visible histologically at early time points and was still present at eight weeks, where it could also be seen by micro-CT. This may be related to earlier observations that healing with BMP-2 produces a “shell” of bone lacking internal osseous structure24,25. Of the fixators used continuously without dynamization during the eight-week experiment, the low-stiffness fixator provided consistently the best healing. This runs counter to the present clinical practice of fixing large segmental bone defects as rigidly as possible.
Reverse dynamization in combination with biological cues (e.g., BMP-2) offers numerous advantages in comparison with traditional bone-grafting techniques. There is no need for microvascular surgery, and donor-site morbidity with vascularized bone transfer and harvesting of cancellous autograft is obviated. Reverse dynamization avoids the pain, discomfort, and prolonged healing with distraction osteogenesis. The proposed method also potentiates the potency of a biological signal delivered to the site of the defect.
Reverse dynamization considerably accelerated maturation of the bone within the defect, which was evidenced on qualitative examination of the histological findings and micro-CT images as advanced formation of neocortices, reduced prominence of trabeculae, uniform contour, and accelerated reduction in apparent callus size. Consistent with these observations, quantitative data confirmed that the bone mineral content and bone area of the defects healed by reverse dynamization was closer to normal and had greater mechanical strength. Only one regimen of reverse dynamization was evaluated in this study, and it is possible that different stiffnesses or timing of reverse dynamization would provide even better results.
Although BMP-2 has shown preclinical efficacy in animal models, the clinical effectiveness of BMP-2 in healing of long bones has been disappointing. Our data suggest that an appropriate mechanical environment is necessary for BMP-2 to be effective, and research into the mechanobiology of cellular responses to BMP-2 could be fruitful. This, in turn, has three main components: mechanosensing, signal transduction, and effector cell response. These have been reviewed recently by Morgan et al.26 in the context of fracture-healing.
Although these experiments used a rat model, the data are clear and unequivocal. The results are likely to be relevant to clinical orthopaedics, and it should be feasible to design and construct variable-stiffness fixators for use in humans. To our knowledge, we are the first to show that the healing of critical-sized segmental defects is highly responsive to the ambient mechanical environment and does not necessarily follow the same rules as fracture-healing in this regard. Further study of reverse dynamization could lead to improved clinical management of these difficult cases.
Disclosure: One or more of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of an aspect of this work. In addition, one or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.