Animals, Drug Dosing, and Administration
All methods and animal procedures were reviewed and approved by the University's Institutional Animal Care and Use Committee and met or exceeded all federal guidelines for the humane use of animals in research. Female Sprague-Dawley rats were fed a standard diet and kept caged in pairs in a constant temperature and humidity environment. The rats weighed an average (and standard deviation) of 276 ± 21 g and were approximately three months old at the time of fracture. After fracture, the rats were treated with vehicle (1% methylcellulose, twice a day in ninety-five rats), 4 mg/kg of celecoxib (once a day in fifty-nine rats), or 30 mg/kg of AA-861 (twice a day in eighty-nine rats) by oral gavage. By two weeks after fracture, most rats had begun to gain weight. By five weeks after fracture, vehicle-treated rats had an average 15% increase in body weight, while the AA-861-treated rats had an average 22% increase in body weight. Timed-staged fracture callus specimens were harvested at two, four, seven, and ten days after fracture for cellular proliferation studies; at four, seven, ten, fourteen, seventeen, and twenty-one days after fracture for gene expression studies; and at five weeks after fracture for biomechanical and radiographic analyses. Celecoxib (4-[5-(4-methylphenyl)-3-(trifluoromethyl)-1H-pyrazol-1-yl] benzenesulfonamide) was obtained from Celebrex capsules (Pfizer, New York, NY). AA-861 (2,3,5-trimethyl-6-[12-hydroxy-5,10-dodecadiynyl]-1,4-benzoquinone) was obtained from Wako Chemicals USA (Richmond, Virginia). Celecoxib is a COX-2-selective inhibitor, and AA-861 specifically inhibits 5-LO31,32.
Fracture Model
Closed fractures were generated in the right femora of 243 young adult, female Sprague-Dawley rats by a procedure that has been previously described33,34. Briefly, the rats were anesthetized by intraperitoneal injection of ketamine (50 mg/kg) and xylazine (10 mg/kg). Under aseptic conditions, a medial parapatellar incision (approximately 1 cm) was made in the right hind limb and the patella was dislocated laterally. The medullary canal was entered through the intercondylar notch and reamed with a 21-gauge needle. A 0.71-mm-diameter stainless-steel pin was then inserted into the canal and was secured in the greater trochanter by tamping. The distal portion of the pin was cut flush within the intercondylar notch, and the patella was reduced. The soft tissue and skin were closed in two layers with use of resorbable sutures. The diaphysis of the pinned femur was fractured immediately with use of a three-point bending device. The animals were then caged in pairs and allowed to walk freely after surgery. Animals were killed by CO2 inhalation. Femora were resected and processed for mechanical testing, RNA isolation, and immunohistochemistry.
Measurement of Fracture Callus Leukotriene B4 Levels
Four days after production of the standard closed fracture, the rats were treated with a single oral dose of vehicle (five rats) or with a dose of 1 (five rats), 5 (four rats), 10 (five rats), or 30 mg/kg (four rats) of AA-861. Two hours later, the rats were killed and the fracture calluses were resected and flash frozen in liquid nitrogen. The calluses were stored at -80°C until eicosanoids were extracted and partially purified as described previously11. Leukotriene B4 (LTB4) levels were measured with use of an enzyme-linked immune-assay (Cayman Chemical, Ann Arbor, Michigan) and normalized to callus-extract protein levels35.
Radiographic Analysis
Radiographs were made with use of a Packard Faxitron (McMinnville, Oregon) and Kodak Min-R 2000 mammography film (Eastman Kodak, Rochester, New York), while the rats were under anesthesia or after they were killed. Immediately following fracture, radiographs were made to verify the position and quality of each fracture. Additional radiographs were made weekly until the time when the rats were killed, to determine the degree of healing. Fracture-healing at five weeks was evaluated from the radiographs with use of a 4-point scoring system33. For radiographic scoring, femora were harvested and the soft tissue was carefully removed to avoid damaging the callus. Then radiographs (ventrodorsal view) of the isolated femora were made, and the radiographs were scored in a blinded fashion by three independent investigators familiar with fracture-healing in rats. Each radiograph was scored from 0 to 4 on the basis of apparent bone-bridging across the callus at the left and the right periphery (1 point each) and by apparent bone-bridging between the cortices of the femur on the left and right sides (1 point each). The data were compared between treatment groups with use of analysis of variance on ranks and post hoc Dunn tests.
Biomechanical Analysis
Fractured femora were resected at five weeks, and the intramedullary pin and soft tissue were removed. The ends of the femur were potted in 0.75-in (1.9-cm) hex nuts with use of a low-melt-temperature metal (Wood's metal; Alfa Aesar, Ward Hill, Massachusetts). Torsional testing was conducted with use of a servohydraulic testing machine (MTS Systems, Eden Prairie, Minnesota) with a 20-Nm reaction torque cell (Interface, Scottsdale, Arizona). Testing was performed to failure at an actuator head displacement rate of 2° per second and a data-recording rate of 20 Hz. Peak torque and angle at failure were calculated from the load-deformation curves. Equations used to derive torsional rigidity, maximum shear stress, shear modulus, and the polar moment of inertia have been described10. Data were compared between treatment groups with use of Student t tests.
Histology, Immunohistochemistry, and Cell Proliferation
Rats were injected intraperitoneally with bromodeoxyuridine (BrdU; 30 mg/kg) one hour before they were killed. Femora were resected and fixed in Streck tissue fixative (Streck Laboratories, La Vista, Nebraska). After fixation, samples were decalcified with Morse's solution and embedded in paraffin36. Serial 5-µm longitudinal sections from the midline of the fracture callus were prepared. Sections were dewaxed with xylene and rehydrated through a series of decreasing ethanol concentrations ending in distilled water prior to staining or any immunohistochemistry.
For immunohistochemical detection of bromodeoxyuridine, sections were treated with proteinase K (100 µg/mL) for thirty minutes at 37°C, rinsed three times with phosphate-buffered saline solution (pH 7.4), and fixed with Cell-Prep (American Bio-Safety, Rocklin, California) for forty-five minutes at room temperature. Sections were rinsed with phosphate-buffered saline solution and treated with 3% hydrogen peroxide to inactivate any endogenous peroxidase. The sections were incubated with a biotinylated monoclonal mouse anti-bromodeoxyuridine antibody (Clone Bu20a; DAKO, Carpinteria, California) for fifteen minutes, rinsed with phosphate-buffered saline solution, and treated with peroxidase-conjugated streptavidin for fifteen minutes. Peroxidase activity was detected with use of diaminobenzidine as the chromagen in a ten-minute reaction. Sections were subsequently counterstained with 2% methyl green in citrate buffer (pH 4). A serial section from each sample was stained with hematoxylin and eosin in order to determine the total cell number within the fracture callus area.
Samples were visualized and recorded with use of an Olympus BH2-RFCA microscope (Olympus Optical, Tokyo, Japan) equipped with a Nikon DXM1200F digital camera (Nikon Instruments, Melville, New York). Approximately sixteen to twenty-eight images per sample were captured and analyzed with use of Image-Pro Plus software (version 5.02.9; Media Cybernetics, Bethesda, Maryland). The total number of cells and the number of bromodeoxyuridine-positive cells in the callus were counted. The cell proliferation rate was determined as the percentage of callus bromodeoxyuridine-positive cells to the total number of callus cells. Callus total cell numbers and proliferation rates were compared by two-way analysis of variance and post hoc Holm-Sidak tests with use of drug treatment and time after fracture as the independent variables.
For histomorphometric analysis, the femora and the overlying muscle were resected and fixed in buffered formalin. The specimens were then embedded in polymethylmethacrylate with use of standard procedures37. Longitudinal sections through the femur were cut with use of a diamond saw, glued onto Plexiglas slides, and polished. The slides were stained with van Gieson picrofuchsin and Stevenel blue, which stain mineralized tissue red and cartilage deep blue38. Digital images of each fracture were captured with use of an Olympus BH2 microscope and a Nikon DXM1200F camera. Callus, mineralized tissue, and cartilage areas were measured from the digital images with use of Image-Pro Plus software. Callus area, cartilage area, and mineralized tissue area were compared by two-way analysis of variance and post hoc Holm-Sidak tests with use of drug treatment and time after fracture as the independent variables.
Measurement of Gene Expression by Reverse Transcriptase-Quantitative Polymerase Chain Reaction
After the rats were killed, the femora were resected, the stainless-steel pin was removed, and most soft tissue was removed without disturbing the fracture callus. The callus or mid-diaphysis of the femora was isolated with use of a rongeur and flash frozen in liquid nitrogen. Total RNA was prepared from the samples as described previously39. Briefly, samples were weighed, pulverized with use of a BioPulverizer (Biospec, Bartlesville, Oklahoma), and extracted with TRIzol (10 mL/g of sample; Invitrogen, Carlsbad, California). The aqueous phase of each extract was ethanol-precipitated. The nucleic acid pellet was dissolved in water, and the RNA was further purified by differential binding and elution from glass fiber filters (RNeasy; QIAGEN, Valencia, California). RNA concentration was determined spectrophotometrically, and RNA integrity was confirmed by agarose gel electrophoresis. Only RNAs with acceptable characteristics, including intact 16S and 28S ribosomal RNA, were used for further analysis.
cDNA for each sample was prepared from 1-µg aliquots of total RNA as described previously with use of an oligo(dT)20 primer, RNase inhibitor (SUPERase-In; Ambion, Austin, Texas), and MMLV reverse transcriptase (New England BioLabs, Beverly, Massachusetts)39. Reactions were allowed to proceed for one hour at 42°C and were stopped by incubation at 92°C for ten minutes. cDNA was stored at -20°C until use.
Quantitative polymerase chain reaction (qPCR) was performed with use of the Absolute QPCR SYBR Green mix (ABgene, Rochester, New York) and an Applied Biosystems 7500 Real-Time PCR System (Foster City, California). Reactions contained an aliquot of cDNA corresponding to between 20 and 100 ng of total RNA, depending on the mRNA being quantified, in a 25-µL reaction volume. The increase in SYBR green fluorescence was measured over forty cycles of polymerase chain reaction. RNA was prepared from six fracture calluses for each treatment group at each time point. Because of poor RNA quality or failed polymerase chain reactions, RNA from three to six independent samples were analyzed for each mRNA target, for each treatment group, at each time point. Polymerase chain reactions were performed in triplicate for each mRNA target and for each RNA sample. Mean Ct (cycle threshold) values were calculated with use of the supplied Applied Biosystems software and normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) Ct values. Primer sets and annealing temperatures used for quantitative polymerase chain reaction are summarized in a table in the Appendix. Normalized Ct values were compared by two-way analysis of variance and post hoc Holm-Sidak tests with use of drug treatment and time after fracture as the independent variables. The mean GAPDH normalized elongation factor-1a (EF-1a) Ct value (and standard deviation) for all groups was 0.845 ± 0.024 (range, 0.826 to 0.862; data not shown). Two-way analysis of variance failed to identify any differences between treatment groups or time points for the normalized EF-1a Ct values, which indicates that normalization of target mRNA Ct values to GAPDH Ct values is valid.
Source of Funding
This research was supported by a grant from the New Jersey Commission on Science and Technology.
Dose-Dependent Inhibition of Callus 5-LO with AA-861
The effect of systemic AA-861 treatment on fracture callus 5-lipoxygenase (5-LO) activity was measured indirectly as callus LTB4 levels. AA-861 was administered to rats four days after fracture, and callus LTB4 levels were measured two hours later in order to identify an AA-861 dose that reduced callus LTB4 levels by at least 90%. As shown in Figure 1, all doses of AA-861 significantly reduced callus LTB4 levels relative to the vehicle-treated samples (p < 0.001). The 10-mg/kg dose of AA-861 (five rats) reduced callus LTB4 levels by 88%, while the 30-mg/kg dose (four rats) reduced callus LTB4 levels by 91% relative to the calluses from the five vehicle-treated rats. On the basis of these data, AA-861 was used at 30 mg/kg and administered twice a day to ensure sufficient and continual inhibition of 5-lipoxygenase.
Radiographic Analysis of 5-LO Inhibition on Fracture-Healing
Fracture-healing was initially assessed by radiography in rats treated for twenty-one days after fracture with AA-861 or vehicle. Serial radiographs were made at weekly intervals and showed that fracture-bridging occurred by three weeks after fracture in the AA-861-treated rats compared with four weeks in the vehicle-treated rats (Fig. 2). Fracture callus size also appeared to decline beginning three weeks after fracture in the AA-861-treated rats but not until five weeks after fracture in the vehicle-treated rats. Radiographs were made of all femora resected five weeks after fracture. With use of an established 4-point scoring system, in which 0 indicates no osseous bridging of the fracture and 4 indicates complete osseous bridging of the fracture, the radiographs of the healing femora made five weeks after fracture were scored. The eleven vehicle-treated rats had a mean score (and standard deviation) of 2.33 ± 0.67, while the femora from the eight AA-861-treated rats had a significantly higher mean score of 3.62 ± 0.33 (p < 0.001). These observations suggest that fracture callus-bridging occurs sooner and callus-remodeling is initiated earlier in the AA-861-treated rats, leading to accelerated healing.
Fractures Heal by Endochondral Ossification in the Presence of 5-LO Inhibition
The effects of AA-861 treatment on callus formation were examined histologically. As shown in Figure 3, callus specimens collected at seven or fourteen days after fracture from AA-861-treated rats had morphology consistent with healing occurring by endochondral ossification. The callus from the AA-861-treated rats showed periosteal new-bone formation at the callus periphery and mesenchymal tissue at the center of the callus with cartilage juxtaposed between the new bone tissue and the mesenchymal tissue. Overall callus morphology was similar to the specimens from the vehicle-treated rat callus except there appeared to be more cartilage and new bone in the specimens from the AA-861-treated rats.
Histomorphometric Analysis of the Effect of 5-LO Inhibition on Fracture-Healing
A histomorphometric analysis was performed to determine how AA-861 treatment affected fracture callus cartilage and new-bone formation. Fracture callus, mineralized tissue, and cartilage area were measured at seven, fourteen, and twenty-one days after fracture from five, three, and five vehicle-treated rats, respectively, and from six AA-861-treated rats at each time point. As shown in Figure 4, AA-861 treatment caused a significant, approximately 2.5-fold, increase in callus area seven days after fracture compared with the vehicle-treated samples. Callus area remained approximately the same between seven and twenty-one days after fracture in the AA-861-treated samples, which was consistent with the radiographic observations. The amount of cartilage in the AA-861-treated samples was significantly higher than the values for the vehicle-treated samples. At seven days after fracture, there was approximately 5.6-fold more cartilage in the AA-861-treated samples compared with the vehicle-treated samples (p < 0.001). At fourteen and twenty-one days after fracture, the amount of callus cartilage remained higher in the AA-861-treated samples than in the vehicle-treated samples. Similarly, there was 4.2-fold more mineralized tissue in the AA-861-treated samples compared with the vehicle-treated samples (p = 0.015) after seven days of healing. By twenty-one days after fracture, the amount of callus mineralized tissue in the AA-861-treated samples was approximately half that of the vehicle-treated samples (p < 0.001), which suggests that callus remodeling had already begun in the AA-861-treated rats. This also coincides with the radiographic observations showing apparent earlier callus remodeling in the AA-861-treated rats (Fig. 2).
Biomechanical Measurement of 5-LO Inhibitor Effects on Fracture-Healing
Fracture-healing was quantitatively assessed five weeks after fracture by torsional mechanical testing. The five-week time point was chosen to give the fractures in the vehicle-treatment group sufficient time to heal so that the mechanical testing would compare bridged fractures in both groups rather than bridged and unbridged fractures. As shown in Table I, all torsional mechanical testing values for the healing femora were significantly higher in the AA-861-treated rats than in the vehicle-treated rats. The structural properties of peak torque and maximum rigidity were approximately 40% higher for the eight AA-861-treated samples compared with the fifteen vehicle-treated samples, while the material properties of maximum shear stress and shear modulus were approximately 70% higher for the AA-861-treated samples compared with the controls. These data demonstrate that AA-861 treatment significantly enhanced the mechanical properties of the healing femoral fractures. The mechanical properties of the contralateral, unfractured femur are shown in Table I as a reference only, as all comparisons were made between healing fractured femora of the two treatment groups.
Inhibition of 5-LO Increases Fracture Callus Cell Proliferation
Fracture callus cell proliferation was measured at two, four, seven, and ten days after fracture by bromodeoxyuridine incorporation in rats treated with vehicle (six rats at all time points), celecoxib (six rats at all time points), or AA-861 (four, five, four, and five rats, respectively). Callus specimens were resected, and cells that had incorporated bromodeoxyuridine were detected by immunohistochemistry. Serial sections of the samples were counterstained with hematoxylin to detect cell nuclei. Groups in Figure 5 summarize the total number of callus cells and the proliferation rates at each time point and for each treatment group.
The total number of cells within the fracture callus appeared to reach a maximum at days 4 and 7 after fracture (Fig. 5, A). The data were analyzed by two-way analysis of variance with use of time after fracture and drug treatment as independent variables. On the basis of this analysis, differences in callus cell numbers over time did not appear to be dependent on drug treatment (p = 0.055). Callus cell numbers were the lowest at day 2 (p < 0.001), increased at days 4 and 7, and then declined at day 10 (p = 0.005 for comparison with day 7). Overall, there was no difference in callus cell numbers between the vehicle-treated and celecoxib-treated or the vehicle-treated and AA-861-treated samples. However, there was a significant decrease in the overall number of callus cells in the AA-861-treated rats compared with the celecoxib-treated rats (p = 0.002).
In contrast to callus cell number, cell proliferation rates varied significantly with time after fracture and with drug treatment (Fig. 5, B). Callus cell proliferation rates were compared between treatment groups by two-way analysis of variance with use of time after fracture and drug treatment as independent variables. The data are summarized in a table in the Appendix. Significant effects for time after fracture (p < 0.001) and for drug treatment (p < 0.001) were found. A significant interaction between time after fracture and drug treatment was found (p < 0.001), indicating that drug treatment can alter proliferation rates at specific times after fracture. Significant differences in proliferation rates were noted between all time points in the vehicle-treated rats (p = 0.003) except between the seven and ten-day time points (p = 0.059). Conversely, in the celecoxib-treated rats, the only difference detected was between the proliferation rates at day 4 and day 10 after fracture (p = 0.003). While the proliferation rate peaked at day 4 in the vehicle-treated rats, the proliferation rate was highest at day 2 after fracture in the AA-861-treated rats (p < 0.001), and it declined to a similar rate at days 4 and 7 after fracture, with a still lower value at day 10 (p = 0.011 compared with day 7). Significant differences also were found between treatment groups. Notably, the celecoxib-treated rats had significantly lower proliferation rates at two (1.33% compared with 2.08%; p = 0.025) and four days (1.84% compared with 4.22%; p < 0.001; 95% confidence interval, 2.27 to 4.07) after fracture compared with the vehicle-treated rats. AA-861 caused a significant increase in proliferation rate at two (3.68% compared with 2.08%; p < 0.001; 95% confidence interval, -2.81 to -0.039) and seven days (2.02% compared with 1.05%; p = 0.010) after fracture compared with the vehicle-treated values but was significantly less than the vehicle-treated value at day 4 (2.19% compared with 4.22%; p < 0.001). Similarly, AA-861 treatment led to significantly higher proliferation rates at two and seven days after fracture compared with the celecoxib-treated rats (p < 0.001 and 0.013, respectively).
The data indicate that AA-861 treatment causes an earlier peak in callus cell proliferation compared with the vehicle-treated control rats. Further, the data indicate that celecoxib treatment inhibits callus cell proliferation but does not affect the total number of cells in the callus. Although not shown, the majority of the bromodeoxyuridine-positive cells were located in the periosteum surrounding the fracture site in all treatment groups.
Inhibition of COX-2 and 5-LO Differentially Affect Fracture Callus Gene Expression
The effects of 5-LO and COX-2 inhibition on gene expression during fracture-healing were analyzed. Rats were treated with vehicle, celecoxib, or AA-861 following a standard fracture of the right femur. Animals were killed at four, seven, ten, fourteen, seventeen, and twenty-one days after fracture; the calluses were resected; and total RNA was isolated for reverse transcriptase-quantitative polymerase chain reaction analysis. Between three and six independent RNA samples were analyzed for each target mRNA at each time point for each treatment group.
AA-861 and celecoxib treatment caused significant effects on expression of chondrocyte-related genes. Both drug treatments caused significant increases in aggrecan (Acan) expression during fracture-healing compared with the vehicle-treated samples, although no significant increase in expression was detected at any one time (see Appendix). AA-861 treatment caused a 16.0-fold increase in Type-II collagen (Col2a1) expression and a 6.3-fold increase in Type-X collagen (Col10a1) expression at day 10 after fracture compared with the vehicle-treated or celecoxib-treated samples (p < 0.001 for all comparisons; see Appendix). Celecoxib treatment caused 6.1-fold and 3.3-fold increases in Col2a1 expression at days 10 and 14, respectively, after fracture compared with vehicle-treated samples (p < 0.001 for both comparisons). However, no concomitant increase in Col10a1 expression was detected in the celecoxib-treated samples. These data suggest that callus chondrocytes become hypertrophic sooner or to a greater proportion in the AA-861-treated animals than in the vehicle or celecoxib-treated animals.
Osteoblast gene expression also was altered by the drug treatments. Notably, AA-861 treatment caused a consistent, significant increase in osteocalcin (Bglap) mRNA levels at all times compared with the vehicle or celecoxib-treated samples (p < 0.001 for all comparisons and times except day 14, when p = 0.001 and 0.021 compared with vehicle and celecoxib values, respectively; see Appendix). In contrast, Bglap mRNA levels were similar between the vehicle and celecoxib-treatment samples except at day 17 after fracture, when there was a significant decrease in the celecoxib-treated samples (p = 0.014). Type-I collagen a2-chain (Col1a2) expression appeared to be similar between the AA-861-treated and vehicle-treated samples, although expression was higher at day 4 in the AA-861-treated samples (p < 0.001) and lower at day 21 (p < 0.001). In contrast, Col1a2 expression was significantly lower at days 4, 14, and 21 in the celecoxib-treated samples compared with the vehicle-treated samples (p < 0.001 for all comparisons) or AA-861-treated samples (p = 0.032, <0.001, and 0.003, respectively). These data indicate an inconsistent, positive effect on osteoblast-related gene expression caused by AA-861 treatment and an inconsistent, negative effect on osteoblast-related gene expression caused by celecoxib treatment.
Cathepsin-K (Ctsk) mRNA levels were measured as an indicator of osteoclast abundance in the fracture callus because it is highly enriched in osteoclasts, although not entirely specific for osteoclasts40. AA-861 treatment led to a large, significant increase in the Ctsk mRNA level at day 4 after fracture compared with the vehicle-treated or celecoxib-treated samples (p = 0.014 and 0.035, respectively), which then rapidly declined. In contrast, celecoxib treatment caused a significant peak in Ctsk mRNA levels at day 10 (p = 0.006 compared with AA-861 treatment) and day 14 (p = 0.035 and 0.036 compared with AA-861 and vehicle, respectively) after fracture.
The expression of COX-2 and 5-LO also was monitored during fracture-healing. Celecoxib treatment had little effect on COX-2 expression but did cause an increase in 5-LO expression (p < 0.001 and 0.019 compared with vehicle at days 4 and 14, respectively; see Appendix). In contrast, AA-861 treatment caused a 4.0-fold increase in COX-2 expression at day 10 after fracture. COX-2 expression was higher in the AA-861-treated samples at days 7, 10, and 14 compared with the vehicle-treated samples (p = 0.042, < 0.001, and 0.044, respectively) and higher at days 7 and 10 compared with the celecoxib-treated samples (p = 0.044 and < 0.001, respectively). Further, AA-861 treatment caused a significant and consistent decrease in 5-LO expression at all times after fracture compared with the vehicle or celecoxib-treated samples with a maximum 5.6-fold decrease at day 4 after fracture compared with the vehicle-treated samples (p < 0.001 for all comparisons). Thus, while celecoxib treatment had relatively little effect on COX-2 or 5-LO expression compared with the vehicle-treated animals, AA-861 treatment had diametrically opposite effects on COX-2 and 5-LO expression.
Osteopontin (Spp1) expression during fracture-healing also was affected by AA-861 and celecoxib treatment. In vehicle-treated rats, Spp1 expression increased over the course of healing, reaching the highest levels at day 21 after fracture. Celecoxib treatment caused an early increase in Spp1 expression (day 4, p = 0.005 compared with vehicle treatment) but then a decline, reaching values significantly lower than the vehicle-treatment values at days 17 and 21 after fracture (p = 0.014 and < 0.001, respectively). In contrast, AA-861 treatment caused 17.2-fold and 3.6-fold increases in Spp1 expression compared with the vehicle-treated samples at days 5 (p < 0.001) and 7 (p = 0.024), respectively, and then declined to levels below the vehicle-treated samples at days 17 and 21 after fracture (p = 0.003 and < 0.001, respectively).