Sample Preparation
This study was conducted in two parts. For the first part, demineralized bone matrix processed for clinical use was provided by four different tissue banks: LifeNet Health, Community Tissue Services (CTS), Allosource, and Musculoskeletal Transplant Foundation. All samples provided to the surgical and evaluation team were coded to blind the reviewers to the tissue bank source as well as donor age, sex, and history of bisphosphonate therapy. The code was held in a locked file and was broken only once the results were tabulated. Code numbers were applied randomly so that the bank of origin would not be discernible in the overall reporting of the results. Each tissue bank provided three or four matched batches of demineralized bone matrix: one batch from a donor who had a reported history of using bisphosphonates and one batch from an age and sex-matched donor who had not taken bisphosphonates (see Appendix). It was not possible to control samples for the type of bisphosphonate used or the duration of use because the bone banks do not currently collect this information from donors. Demineralized bone matrix that had been previously shown to have high osteoinductivity was used as a positive control. Demineralized bone matrix from the same lot as the positive control was heated for twenty-four hours at 105°C to inactivate any biologically active protein factors present and served as a negative control. For the second part, mineralized allograft and demineralized bone matrix from the same donor as the positive control above were soaked in 1 mL of phosphate-buffered saline solution with 0, 0.002, 2.0, or 2000 ng/mL of alendronate (Merck, Whitehouse Station, New Jersey)36 and implanted.
Although each tissue bank used a slightly different method for processing demineralized bone matrix, all tissue banks follow a general system that involves cleaning, defatting, and disinfecting cortical bone; grinding the defatted bone to particle sizes between 400 and 1000 μm; demineralizing the bone particles in dilute hydrochloric acid solutions, resulting in a residual calcium content of <8% per the American Association of Tissue Banks standards; and freeze-drying the final product. Some banks also chose to add a terminal sterilization step. Particle size was well controlled by all four tissue banks and was maintained between 400 and 1000 μm.
Demineralized bone matrix (15 mg per sample) was placed in a size-9 gelatin capsule (Torpac, Fairfield, New Jersey) in order to facilitate implantation and to prevent displacement while in situ. We previously demonstrated that these capsules dissolve quickly and play no role in the outcome of the assay33,35. Samples were prepared under aseptic conditions and were kept at room temperature until implantation.
Study Design
The study was conducted under a protocol approved by the Institutional Animal Care and Use Committee at the Georgia Institute of Technology. In the first phase of the study, sixty male athymic nu/nu (nude) mice (Harlan, Indianapolis, Indiana) were divided into fifteen groups of four mice each. This strain of mouse, which has reduced helper T cells, was selected because it reduces the chance that a response is due to immunogenicity rather than to the implants themselves. Demineralized bone matrix samples (with and without bisphosphonates) from the four tissue banks, as well as the positive and negative control demineralized bone matrix samples, were implanted in the gastrocnemius muscle, one implant per limb. Thus, each mouse received two implants, both of the same type to reduce any systemic influences, resulting in eight implants per sample.
In the second part of the study, thirty-two mice of the same strain were divided into eight groups of four each, and mineralized bone allograft or positive control demineralized bone matrix that had been treated with alendronate was implanted bilaterally.
Implantation Protocol
Athymic male nu/nu mice were anesthetized by inhalation of 5% isoflurane in O2. Both hind legs of each mouse were disinfected with use of isopropanol and chlorhexidine. Skin incisions of 0.5 cm were made over each gastrocnemius, and a muscle pouch was prepared by blunt dissection. The implant was inserted into the muscle pouch, and the skin incision closed with wound clips.
The mice were housed in conditions appropriate to their immunocompromised status and given food and water ad libitum for thirty-five days. We have previously shown that this period is sufficient for demonstration of even low levels of osteoinductive activity in this model, despite potential variability of preparations33,35.
Histological Evaluation
The animals were killed at thirty-five days after implantation by asphyxiation with carbon dioxide. The hind limbs were disarticulated and fixed in 10% neutral buffered formalin and examined radiographically (Faxitron, Lincolnshire, Illinois) to localize the implant site. Following fixation, the tissue was decalcified in 5% formic acid, embedded in paraffin, sectioned, and stained with hematoxylin and eosin. Three consecutive cross sections (3 to 4 μm each) of the implant were obtained at three different levels on the longitudinal axis of the sample. We found that this helped to ensure that a section of the implanted region containing a cross section of the tibia and the fibula was obtained to confirm correct orientation, since some of the implants shifted position when the mice returned to weight-bearing. When more than one section with these features was found, the section with the largest cross-sectional area was selected for analysis.
The ability of the samples to induce new bone was qualitatively evaluated by two independent observers, who were blinded to treatment, and was rated according to the following previously published semiquantitative scheme. A score of 1 indicated the presence of particles without any bone; 2, the production of new bone in one site within the section and covering <40% of the surface area examined; 3, the production of new bone in more than one site, covering >40% but <70% of the surface area examined; and 4, the production of new bone in more than one site, covering >70% of the surface area examined33,35.
New bone formation can be found even when heat-inactivated demineralized bone matrix is used, possibly because of particle packing, condensed hematoma, or isolated areas of active factors. If only one or two implants per donor exhibit focal areas of new bone, the mean qualitative score would be <1.5. Therefore, to reduce the likelihood of false positives and to ensure that osteoinduction is a general property of the demineralized bone matrix being tested, demineralized bone matrix implants receiving scores of ≥1.5 were considered osteoinductive.
With use of this method, there was similar induction ability across bone banks, eliminating concerns about demineralized bone matrix quality. Observer scoring was calibrated to an experienced reviewer (Z.S.) prior to beginning the experiment. When the observers’ ratings disagreed, a third observer (B.D.B.) evaluated the disputed sections. The overall grade for each implant was obtained by averaging the scores from all specimens in the group.
On the same histological sections, a histomorphometric analysis was performed with use of a computerized histomorphometric system (version 4.5.1, Image-Pro Plus; Media Cybernetics, Bethesda, Maryland). The section that exhibited the most demineralized bone matrix particles and new bone was selected to represent that sample for area analysis, and the section was scanned with use of a microscope equipped with 1×, 5×, and 10× objectives. The image was captured by video camera, and the data were stored and analyzed with use of the imaging software. Calibration was performed according to the instructions accompanying the software. The areas of new cortical bone and residual demineralized bone matrix particles in each section were measured. Ossicle formation, the sum of new bone and new bone marrow, was also quantified. Representative histological images are presented in Figure 1.
Statistical Analysis
The results of the morphometric analysis are shown as the mean and the standard error of the mean of the specimens examined, with N being the number of bone graft implant sites. For this study, type I error was set as 0.05. For a two-sample t test, assuming alpha = 0.05 and power of 0.8, and with use of a standard deviation of 0.2 from previous experiments, a sample size of eight is adequate to determine effect sizes of 30%. Thus, the sample size we used, eight implants, was appropriate. We previously also calculated our data with use of each animal as an N, averaging the results from the two implants per animal and then assessing the data with an N = 4 and found that the statistical comparisons remained the same33,35. For this reason, we used each implant as an independent N. Significant differences between groups were determined by analysis of variance and the use of the Bonferroni modification of the Student t test. P values of <0.05 were considered significant. The sample size was sufficient to compare samples with respect to processing method: irradiated versus nonirradiated. In addition, it was possible to assess osteoinductivity as a function of donor age, either younger than or older than seventy years. This age was selected on the basis of previous studies showing that demineralized bone matrix from donors older than seventy years is less osteoinductive than bone from younger donors in this model33,34.
Source of Funding
This research was supported by a grant from the Scientific and Technical Affairs Committee of the American Association of Tissue Banks.
Effect of Donor Bisphosphonate Use on Demineralized Bone Matrix Osteoinduction
The average age of donors who had been treated with bisphosphonates was 68.9 ± 2.0 years (see Appendix). Two of these donors were male, and thirteen were female. The average age of donors who had not been treated with bisphosphonates was 69.1 ± 2.5 years, which was not significantly different from the bisphosphonate group. Three of these donors were male and twelve were female.
There was no difference in the semiquantitative assessment of osteoinduction between the two groups. Demineralized bone matrix in the experimental groups had scores similar to the positive control demineralized bone matrix and higher than the negative control demineralized bone matrix (Fig. 2-A). This was the case for demineralized bone matrix from all tissue banks. Ten of fifteen samples from donors not treated with bisphosphonates and nine of fifteen samples from donors treated with bisphosphonates were osteoinductive, on the basis of the mean score (Table I). There were no tissue bank-specific differences in the number of demineralized bone matrix samples that were osteoinductive. New bone formation was significantly higher in all samples compared with the negative control (Fig. 2-B). There was no difference in new bone formation between the experimental groups and the positive control. Only Bank B had significantly increased new bone in the bisphosphonate groups; there was no difference in any of the other groups. The area of newly formed ossicles was increased in demineralized bone matrix samples soaked in bisphosphonate compared with demineralized bone matrix alone (Fig. 2-C). All samples had a larger ossicle area compared with the negative control. Demineralized bone matrix-only groups had a significantly lower ossicle area than the positive control samples.
The areas of residual demineralized bone matrix remaining after thirty-five days were measured (Fig. 3). All samples had significantly more demineralized bone matrix than the negative control, with the exception of Bank C. Only Bank B samples of demineralized bone matrix with bisphosphonates had significantly higher residual demineralized bone matrix than the positive control sample. Bank C had significantly less residual demineralized bone matrix than the positive control samples. There was no difference between demineralized bone matrix only and demineralized bone matrix and bisphosphonate for any of the bone banks.
Because the samples came from four tissue banks with separate preparation methods, we compared the effect of the sterilization methods used, either irradiated or nonirradiated (Table II). We found no difference in qualitative scores or ossicle formation between demineralized bone matrix with or without bisphosphonate in samples with either sterilization method. The age of the demineralized bone matrix donor was also considered (Table III). Ossicle formation and qualitative score were not significantly different between samples of demineralized bone matrix with or without bisphosphonate from donors who were less than seventy years old and from donors who were seventy years and older. Moreover, no significant association was found between donor age and ossicle formation or qualitative score.
Effect of Bisphosphonate Addition on the Osteoinductivity of Demineralized Bone Matrix
To examine the effects of bisphosphonate on demineralized bone matrix osteoinductivity, positive control samples of demineralized bone matrix with the addition of alendronate were examined in our model. Osteoinduction, as measured by qualitative score, was unchanged with increasing doses of alendronate (Fig. 4-A). All samples scored higher than the heat-inactivated demineralized bone matrix samples. The samples also induced a greater area of new bone formation (Fig. 4-B) and ossicle formation (Fig. 4-C) than heat-inactivated samples did, but there was no difference between the treatment doses.
Finally, we examined the effect of bisphosphonate addition on the osteoinductivity of mineralized bone (see Appendix). The qualitative score of osteoinductivity was unchanged for all samples, regardless of bisphosphonate dose. The score of the samples (i.e., 1) indicates that the samples did not induce new bone formation.
This study showed that there was no difference in overall osteoinductivity of demineralized bone matrix prepared from allograft from donors who were known to have been medicated with bisphosphonates and that from donors who were not. This observation was independent of whether the demineralized bone matrix was processed by any specific tissue bank, although different methods of production were used. We used samples that were prepared with use of routine processing methods in each of the tissue banks, and not samples prepared specifically for this study. Thus, which of the several different kinds of bisphosphonates the donors were prescribed, if more than one form of bisphosphonate was used, or the length of time the donors were on bisphosphonate therapy were not known. A comprehensive medical and social history of each donor was used to identify donors who had or had not taken bisphosphonates, but it cannot be entirely ruled out that the donors identified as not having taken bisphosphonates at the time of death had had no prior exposure. It is possible that the failure to detect an effect of bisphosphonate usage on osteoinductivity was because of a failure to definitively screen for bisphosphonate-free donors. Given that each bone bank contributed multiple age-matched pairs of donors, however, the sample population in each treatment group was drawn from a broad spectrum of the population. Therefore, despite the relatively limited sample size, the results strongly support the conclusion that demineralized bone matrix osteoinductivity is not affected by bisphosphonate use.
The fact that we used routinely processed preparations of demineralized bone matrix also allowed us to make important observations. All of the tissue banks produced demineralized bone matrix that was osteoinductive as well as some batches of demineralized bone matrix that were not osteoinductive in the mouse muscle implant model. This suggests that the failures were due to donor-dependent differences and not to a difference in preparation. Although some of the demineralized bone matrix was terminally sterilized by gamma irradiation, there was no difference in its osteoinductivity compared with nonirradiated demineralized bone matrix preparations. The sample size from each bank, and therefore each processing method, was relatively small (six to eight samples, of which one-half were from donors who had used bisphosphonates). Thus, we cannot state in an absolute way that processing is not a concern35.
Bisphosphonates have been shown to modify the response to bone allograft used in bone sites, either enhancing or inhibiting allograft resorption and new bone formation24,45-49. A recent report showed this depended on the concentration of bisphosphonate in an allograft23. When allograft is implanted ectopically, it is resorbed without new osteogenesis34,50, indicating that it is not osteoinductive. We hypothesized that inhibition of osteoclastic activity by alendronate could reduce the resorption of the allograft, potentially allowing it to stimulate bone formation, but this did not occur. Whether or not alendronate was added to the allograft, no bone formation was noted. In contrast, demineralized bone matrix was osteoinductive with or without the addition of alendronate, consistent with another study demonstrating that alendronate did not affect osteogenesis51.
Demineralized bone matrix batches used in the present study were generally from older donors (more than fifty-six years old), as this is the age group most likely to be medicated. We previously showed that demineralized bone matrix from donors more than fifty-six years old had reduced osteoinductivity compared with that from young donors33; however, in the present analysis, all samples exhibited similar new bone formation when implanted in the mouse muscle model with no correlation to age. No young donors were used in the present study; therefore, it is not known if there would have been a higher percent of the samples with osteoinductive properties in demineralized bone matrix processed from a young donor population. Our results support previous observations from our group34 and others52-54, showing that donors who are sixty to seventy years of age continue to possess osteoinductive bone, and suggest that any loss of osteoinductivity occurs later in life.
The results of this study show that demineralized bone matrix from donors treated with or without bisphosphonates has the same ability to induce bone formation. The addition of alendronate to samples of demineralized bone matrix known to be osteoinductive had no effect on osteoinduction. The results suggest that bone formation by means of demineralized bone matrix was not affected by oral bisphosphonates or the addition of bisphosphonates to demineralized bone matrix. However, remodeling of the new bone was not studied, so it is possible that bisphosphonates do not affect bone formation but have longer-term effects on the bone created, possibly during bone remodeling.