Dr. Ma, Dr. Westmore, and Dr. Sato have corporate appointments and own stock with Eli Lilly & Company. All other authors have no conflict of interest.
We investigated the effects of inhibitors of bone resorption (estrogen, raloxifene, and alendronate) on the processes of fracture repair in ovariectomized (OVX) rats. One hundred forty female Sprague-Dawley rats at 3 months of age were either OVX or sham-operated and divided into five groups: sham control, OVX control, estrogen (17α-ethynyl estradiol [EE2], 0.1 mg/kg), raloxifene (Rlx, 1.0 mg/kg), and alendronate (Aln, 0.01 mg/kg) groups. Treatment began immediately after the surgery. Four weeks postovariectomy, prefracture controls were killed and bilateral osteotomies were performed on the femoral midshafts and fixed with intramedullary wires. Treatment was continued and fracture calluses were excised at 6 weeks and 16 weeks postfracture for evaluation by X-ray radiography, quantitative computed tomography (QCT,) biomechanical testing, and histomorphometry. At 6 weeks postfracture, Aln and OVX had larger calluses than other groups. Sham and OVX had higher ultimate load than EE2 and Rlx, with Aln not different from either control. Aln calluses also contained more mineral (bone mineral content [BMC]) than all other groups. By 16 weeks postfracture, OVX calluses were smaller than at 6 weeks and the dimensions for Aln had not changed. Aln had higher BMC and ultimate load than OVX, EE2, and Rlx. EE2 and Rlx had similar biomechanical properties, which were similar to sham. Interestingly, OVX and Aln animals were heavier than other groups at all time points; therefore, ultimate load was normalized by body weight to show no significant differences in strength of the whole callus between groups at either 6 weeks or 16 weeks postfracture. However, Aln strongly suppressed remodeling of the callus, resulting in the highest content of woven bone, persistent visibility of the original fracture line, and lowest content of lamellar bone, compared with other groups. Therefore, the larger Aln callus appeared to be a remarkable, morphological adaptation to secure the fracture with inferior material. In conclusion, OVX-stimulated bone turnover resulted in the fastest progression of fracture repair that was most delayed with Aln treatment, consistent with marked suppression of bone resorption and formation activity. Estrogen and Rlx had similar effects that were generally similar to sham, indicating that mild suppression of bone turnover with these agents has insignificant effects on the progression of fracture repair.
INHIBITORS OF BONE RESORPTION—bisphosphonates, estrogens, and selective estrogen receptor modulators (SERMs)—currently, are used widely in the treatment of osteoporosis in postmenopausal women.(1–5) Alendronate (Aln) may be the most widely prescribed bisphosphonate, which has been shown to prevent loss of bone mass by rapidly suppressing the osteoclastic resorption of bone.(1, 2) However, inhibition of bone resorption has been shown to be followed by a secondary but powerful suppression of bone formation activity in rats and women.(1,2,6,7) The dual suppression of both bone resorption and formation activity results in a substantial reduction in bone turnover, as confirmed by a 90% reduction in activation frequency after 2 years of Aln treatment in women.(7)
Estrogen and the SERM raloxifene (Rlx) have also been shown to prevent loss of bone mass in postmenopausal women with osteoporosis.(3,4,8,9) They also suppress bone resorption activity in ovariectomized (OVX) rats and secondarily suppress formation activity resulting in lower bone remodeling.(10–13) Because patients with osteoporosis are prone to fractures and because Aln, estrogen, and Rlx suppress bone remodeling, we attempted to ascertain what effect they might have on the healing of fractures that might occur during treatment.
Recent studies with another bisphosphonate, incadronate, showed an enlarged callus that was strong, but incadronate delayed callus remodeling in the fractured femora of rats.(14–16) This delay persisted 49 weeks after fracture. Some suggested(17) that strength of the callus is paramount and that the delay in resorption of the callus is of little consequence, because the organism compensated for any negative effect of the drug on composition of the callus by increasing size. However, others have suggested that fracture healing is considered complete when the fracture line is no longer visible radiologically, when the skeletal architecture is restored, and when the mechanical strength is fully restored.(18–20)
Possible effects of estrogens and SERMs in fracture healing are not well understood. Therefore, the effects of estrogen and Rlx on fracture healing were evaluated in OVX rats to ascertain possible effects on fracture repair of two other currently available osteoporosis therapies. These data may help to answer the question as to whether or not patients with osteoporosis undergoing antiresorptive therapy should consider discontinuing treatment after sustaining a nonvertebral fracture.
MATERIALS AND METHODS
Three-month-old, female Sprague-Dawley rats (n = 140; Japan SLC, Inc., Hamamatsu, Japan) were maintained at 20°C on a 12-h light/12-h dark cycle with free access to water and rat food (composition: protein, 24.1%; fat, 5.5%; and mineral, 6.4%; Oriental Yeast Co., Tokyo, Japan). Rats were OVX or sham-operated under general anesthesia with pentobarbital sodium (50 mg/kg intraperitoneally; Abbott Laboratories, North Chicago, IL, USA). Animals were randomized into five groups based on their body weights: sham-OVX controls (sham), OVX controls, OVX treated with estrogen (17-α ethynyl estradiol [EE2]), OVX treated with Rlx, and OVX with Aln. Dosing was initiated on the same day of surgery and continued until necropsy. Sham and OVX groups were given subcutaneous injections of 0.9% saline vehicle. EE2, Rlx, and Aln groups were injected subcutaneously (5 day/week) with 17α-ethynyl estradiol (0.1 mg/kg; Sigma Chemical Co., St. Louis, MO, USA), Rlx (1 mg/kg; Lilly, Indianapolis, IN, USA), or Aln (0.01 mg/kg; Lilly), respectively. Body weights were measured weekly and injection dosages were adjusted accordingly.
After 4 weeks of pretreatment, prefracture groups were killed, and transverse osteotomies in the remaining animals were performed at the midshaft of bilateral femora using a fined-toothed circular microsaw (Kiso Power Co., Osaka, Japan). The fracture was repositioned, fixed tightly by an intramedullary stainless wire (diameter 1.5 mm; Zimmer, Warsaw, IN, USA), and wires were cut at the intercondylar groove to ensure unaffected movement of the knee. Unrestricted ambulation was allowed after recovery from general anesthesia, and dosing was continued postfracture. All rats were double-labeled with a subcutaneous injection of calcein (6 mg/kg; Wako, Ltd., Osaka, Japan) at 9 days and 2 days before death. Rats were killed at 6 weeks or 16 weeks postfracture by exsanguination of the carotid artery under general anesthesia. This experimental protocol (Fig. 1) was approved by the Animal Study Committee of Kagawa Medical University.
Blood was collected before death, stored at −83°C, and osteocalcin levels were assayed using an ELISA (Biomedical Technologies, Inc., Stoughton, MA, USA).
Femora were excised, dissected free of soft tissues, and intramedullary wires were extracted. Anteroposterior soft X-ray radiographs of all femora were taken (30 kVp, 2 mA, for 15 minutes; SRO-40; Sofron, Tokyo, Japan). High-resolution quantitative computed tomography (QCT) was performed using a CT scanner from Enhanced Visions Systems (EVS London, Ontario, Canada). Fracture calluses were imaged with 22.6-μm isotropic voxels to ensure the same resolution in all three orthogonal directions. Analysis was conducted on the fracture plane. The same anatomical region was scanned in the prefractured femora. Within the fracture plane, the image was thresholded(21) into bone and nonbone voxels to measure bone area, bone mineral content (BMC), and moment of inertia.
For mechanical testing, right femora were evaluated by three-point bending using an MTS 1/S testing machine with TestWorks 4 software (MTS Systems Corp., Eden Prairie, MN, USA). Femora were placed, facing anterior down, on two lower support bars (15-mm span). The fracture plane was centered as the loading point, using a strain rate of 10 mm/minute until breakage. Biomechanical properties of the whole callus (ultimate load, stiffness, and energy to break) were determined from the load deformation curve. Intrinsic material properties (ultimate stress, Young's modulus, and toughness) were calculated following the cross-sectional moment of inertia (CSMI).(22) Because of the nonuniformity and variability of the mineral composition of fracture calluses, a weighted moment of inertia was used, taking into account the bone mineral density (BMD) distribution within the callus. Specifically, moment of inertia was normalized by the average BMD within the fracture plane:
where I is the moment of inertia, BMDi is the BMD of the ith voxel, A is the area of each voxel, yi is the perpendicular distance of the ith voxel from the central axis, and BMD is the average BMD value.
After mechanical testing, fractured femora were repositioned, fixed in 70% ethanol, stained in Villanueva bone stain (Polysciences, Inc., Warrington, PA, USA), dehydrated in graded ethanol, defatted in acetone, and embedded in methylmethacrylate. Undecalcified 50-μm-thick cross-sections were cut using a diamond microtome saw (SP1600; Leica Instruments, Nussloch, Germany) within 500 μm from the original fracture line, and contact X-rays were obtained (MX-20; Faxitron X-Ray Corp., Wheeling, IL, USA).
Left femora were fixed in 10% cold neutral-buffered formalin for 3 days and decalcified in 10% EDTA at 4°C for 4 weeks and then embedded in glycol methacrylate (Polysciences, Inc.). Within 100 μm from the original fracture line, 5-μm-thick cross-sections were cut with a microtome (Suppercut 2050; Reichert-Jung, Heidelberg, Germany). TRAP activity of osteoclasts was measured using a leukocyte acid phosphatase kit (Sigma Chemical Co.).
Histomorphometric analysis was performed on 50-μm undecalcified sections with an automatic image analyzer (KSS Stereology; KAA Scientific Consultants, Magana, UT, USA). Polarized light microscopy was used to distinguish lamellar bone from woven bone. The total area (TA), original cortical area, and the medullary area were measured at 100× magnification and the callus area (Ca.Ar) was calculated. Lamellar bone area, woven bone area, single-labeled surface, double-labeled surface, and the interlabeling width were measured in the callus. lamellar/Ca.Ar, mineralizing surface (MS/bone surface [BS]), mineral apposition rate (MAR), and bone formation rate (BFR/bone volume [BV]) were calculated. Osteoclast measurements were performed at 125× magnification in four standardized quarters: anterior, posterior, medial, and lateral aspects as shown previously(14) to include osteoclast number (N.Oc), osteoclast surface (Oc.S), bone-remodeling surface (BS), and derived parameters (Oc.S/BS and N.Oc/BS).
Differences between groups were tested by two-way ANOVA using Statview (SAS Institute, Inc., Cary, NC, USA). If the overall ANOVA was significant, comparisons between pairs of group means were tested by Fisher's protected least significant difference test, where p < 0.05 was considered significant.
Live phase and serum analyses
Among 140 rats at the beginning of the study, 19 were excluded because of death or technical failure of surgeries. All rats gained body weight throughout the experiment. OVX and Aln groups were already heavier than sham before the fracture surgery (Table 1) and were heavier than all other groups at 6 weeks and 16 weeks postfracture. EE2 and Rlx animals weighed less than OVX and Aln at fracture surgery and were lighter than sham after 16 weeks (Table 1).
Table Table 1.. Body Weight and Serum Parameters
At fracture surgery, serum osteocalcin was higher in OVX and Rlx compared with sham, and it was lower in EE2 and Aln compared with OVX (Table 1). At both 6 weeks and 16 weeks postfracture, osteocalcin was higher for OVX compared with all other groups; however, Aln osteocalcin was lower than all other groups at 16 weeks, whereas EE2 and Rlx levels were not different from sham (Table 1). Therefore, osteocalcin levels showed that ovariectomy increased the rate of bone turnover compared with sham, which was suppressed with increasing efficacy by Rlx, EE2, and Aln.
Prefracture groups 4 weeks after ovariectomy
Ovariectomy had no effect on bone area, BMC, or moment of inertia at the midshaft relative to sham controls. Similarly, EE2, Rlx, or Aln had no effect on QCT parameters of the femoral diaphysis relative to sham or OVX controls (Table 2). Biomechanical testing of the intact femoral midshaft at 4 weeks postovariectomy showed no significant differences among groups for biomechanical properties (Table 3) or intrinsic material properties (Table 4).
Table Table 2.. QCT Analysis of the Fracture Plane Before and After Fracture
Table Table 3.. Biomechanical Properties of the Femur
Table Table 4.. Intrinsic Material Properties of the Femur
At 6 weeks postfracture
Images and histomorphometry showed that the whole callus of OVX and Aln generally were enlarged compared with the sham, EE2, and Rlx calluses (Figs. 2 and 3; Tables 2 and 5). The fracture plane was clearly visible for all groups after 6 weeks (Fig. 2). Fracture plane analyses (Table 2) revealed that OVX had bone area, BMC, and moment of inertia that were not different from sham. Treatment with ESTROGEN had no significant effect on bone area, BMC, and moment of inertia relative to sham or OVX controls. Treatment with Rlx had no effect on bone area and BMC, but moment of inertia was reduced relative to OVX. This was primarily because of geometry, as the Rlx calluses tended to be smaller than OVX in TA, although they had a similar amount of mineral and similar bone area. Treatment with Aln increased bone area and BMC, which were both larger than all other groups. Aln moment of inertia was larger than all but OVX controls. Therefore, EE2 and Rlx tended to form smaller calluses that were more similar to sham than OVX, whereas Aln calluses tended to be geometrically more like OVX but with more mineral.
Table Table 5.. Static and Dynamic Histomorphometry of the Fracture
Biomechanical testing of the callus at 6 weeks postfracture showed no differences in biomechanical properties between OVX and sham (Table 3). EE2 and Rlx had lower ultimate load relative to sham and OVX, whereas Aln was not different from OVX. However, normalization of ultimate load by body weight showed no significant differences between groups (Fig. 4).
Intrinsic material properties showed no differences between OVX and sham after 6 weeks postfracture (Table 4). EE2 and Rlx ultimate stress was less than sham, which appeared to be caused by the lower body weight for these two groups. Aln reduced ultimate stress relative to sham but had intrinsic material properties that generally were similar to EE2 and Rlx. Biomechanical and intrinsic material analyses showed that Rlx was not different from EE2 for any parameter.
Histomorphometry (Table 5) revealed that OVX had significant larger TA, Ca.Ar, Ca.Ar/TA, and reduced lamellar/CA compared with sham. EE2 and Rlx had TA and Ca.Ar, which were similar to sham, but the lamellar/Ca.Ar was similar to OVX and less than sham (Fig. 5). Aln TA, Ca.Ar, and Ca.Ar/TA were similar to OVX. At this point, the Aln callus was primarily woven bone and the lamellar/Ca.Ar was reduced compared with all other groups. Rlx, EE2, and Aln reduced the surface and N.Oc'S compared with OVX. OVX increased BFR/BV compared with sham. EE2 and Rlx reduced BFR/BV to between OVX and sham, and lowered MAR relative to OVX. Aln suppressed BFR/BV from OVX to 25–45% of sham levels. These data show that EE2 and Rlx reduced bone formation activity from OVX to sham levels, whereas Aln suppressed bone turnover to below sham levels. After 6 weeks postfracture, Aln calluses were similar to, but not identical with, OVX calluses, while EE2 and Rlx were similar to, but not identical with, sham calluses.
At 16 weeks postfracture
The fracture plane was not always discernible by soft X-ray radiographs, especially for OVX calluses (Fig. 2). Original fracture lines were visible in 0%(0/11) OVX, 56% (5/9) Aln, 27% (3/11) sham, 22% (2/9) EE2, and 10% (1/10) Rlx groups, respectively (Fig. 2). Therefore, fracture healing was most advanced in OVX, and least advanced in Aln, where the original cortical shell remained but became porous. Images showed that all groups but Aln were actively forming a new cortical shell and marrow cavity (Fig. 3). Fracture plane analyses (Table 2) showed that bone area, BMC, and moment of inertia were not different among the groups except for Aln, which had larger bone area, BMC values, and moment of inertia, relative to other groups. Aln calluses were larger than all other groups, and OVX had decreased in size from 6-week levels toward the dimensions of sham, EE2, and Rlx (Fig. 3). These data suggested that callus remodeling had progressed the farthest in OVX, followed by Rlx, EE2, sham, and Aln groups.
Biomechanical testing of the callus (Table 3) revealed that OVX ultimate load was not different from sham. EE2 and Rlx ultimate load were not different from OVX, but Aln ultimate load was greater relative to OVX, EE2, and Rlx. When normalized by body weight, no significant differences were observed among the groups (Fig. 4).
Evaluation of the intrinsic material properties of calluses revealed no significant differences between OVX and sham (Table 4). Ultimate stress was not different among the groups, and EE2 and Rlx Young's modulus was higher than OVX and Aln. These results indicate that EE2 and Rlx calluses had better intrinsic material properties than Aln and OVX.
Histomorphometry revealed that OVX had larger Oc.Ss and N.Oc than sham; but OVX TA and Ca.Ar were reduced from 6-week levels to sham levels after 16 weeks (Table 5). EE2, Rlx, and Aln reduced Oc.S/BS and N.Oc/BS to below OVX. EE2 and Rlx TA and Ca.Ar were now similar but not identical to OVX. Aln TA and Ca.Ar were greater than OVX, EE2, and Rlx. Aln had less lamellar and more woven bone than all other groups (Fig. 6). In fact, 100% of the bone in calluses were lamellar for OVX, EE2, and Rlx, and 87% of the callus was lamellar bone in Aln by this time. Continued elevation of BFR/BV in OVX compared with sham and other groups, and MS (MS/BS) was not different among the groups. EE2, Rlx, and Aln lowered BFR/BV to sham levels. MAR was lower in Rlx and Aln compared with sham, OVX, and EE2. These data indicate a significant delay in callus remodeling with Aln, whereas EE2 and Rlx compared favorably with OVX.
OVX rats have been shown to reproducibly lose bone mass from the axial and appendicular skeleton with declining levels of systemic estrogen, not unlike postmenopausal women.(23) Therefore, the OVX rat has become a useful animal model that mimics the bone loss observed for postmenopausal women. Skeletal pharmacology studies in OVX rats have been predictive of clinical efficacy in postmenopausal women for estrogens,(24, 25) calcitonin,(26) bisphosphonates,(27, 28) tamoxifen,(29–31) and Rlx.(6,11,32)
Similarly, the rat fracture model has been highly useful in elucidating the processes of fracture repair and the possible effects of pharmacologic agents on healing.(33–36) Histomorphometry and biomechanical analyses have shown fracture repair to be a closely regulated process that is orchestrated to restore the mechanical properties, structural geometry, and mobility of the broken bone.(33–35) Early in the healing process, cartilage and woven bone are produced to stabilize the fracture; however, as the callus matures, woven bone is remodeled into lamellar bone. Normally, the callus is formed and remodeled with ever-increasing strength as the mechanical integrity of the callus matures to adopt fully the functional qualities of normal bone.
Currently, Aln, estrogen, and Rlx are available therapies used to treat postmenopausal osteoporosis.(5, 37) Women with osteoporosis have been observed to fracture a variety of bones; therefore, the effects of Aln, estrogen, and Rlx on the fracture repair process were evaluated in femora of OVX rats to model what may happen in women on osteoporosis therapy who sustain nonvertebral fractures. The specific doses of 0.1 mg/kg of estrogen, 0.01 mg/kg of Aln, and 1 mg/kg of Rlx were selected based on previous studies showing maximal skeletal efficacy at these doses in rats.(6, 38) Subcutaneous injection was chosen as the route of administration to eliminate oral bioavailability issues with Aln and to minimize liver metabolism of estrogen and Rlx, resulting in maximum exposure in vivo for the latter two compounds.
In this study, adult rats were confirmed to gain body weight by 4 weeks postovariectomy relative to age-matched sham control, which was mostly caused by a gain in adipose tissue.(32,39,40) Body weights were clearly different between groups before fracture, indicating that calluses were loaded differently from the beginning and through 6 weeks and 16 weeks of fracture repair in our study. Because body weight is proportional to the load applied onto the calluses in vivo, normalization of callus strength (ultimate load) by body weight was an important correction with which to evaluate the functional integrity of the whole callus. Therefore, an important result of our analysis may be that normalization clearly showed no significant differences between groups at both 6 weeks and 16 weeks, indicating that ovariectomy and treatments with inhibitors of bone turnover have only subtle effects on whole callus strength (ultimate load) under our loading conditions and study design. Ovariectomy and treatment with compounds induced significant geometrical, compositional, and biomechanical differences among groups; however, the whole callus appeared to be able to compensate for any negative effects of ovariectomy or treatment by morphological adaptation, as shown previously in fracture studies in intact animals with incadronate.(14, 16) Normalization clearly showed that no group had a substantial advantage over the rest in terms of whole callus strength (ultimate load), indicating that biological adaptation is an important aspect of the fracture healing process.
Ovariectomy was suggested previously to impede fracture repair.(41, 42) Our study showed an enlarged callus at 6 weeks and small reductions in whole callus stiffness of OVX at 16 weeks postfracture. However, no significant effects were observed on any other biomechanical parameter or intrinsic material properties of the fracture after 6 weeks or 16 weeks of healing. Moreover, ovariectomy appeared to be advantageous after 16 weeks in promoting the repair processes because the original fracture plane was not discernible in only the OVX group, and the OVX callus was remodeled completely to lamellar bone. The rapid formation of a new cortical shell and marrow spaces with time in OVX was a likely consequence of the increased rate of bone turnover stimulated by ovariectomy.(10)
Aln increased the size and mineral content of the callus compared with all other groups, while delaying remodeling of the callus woven bone into lamellar bone at both 6 weeks and 16 weeks. Biomechanics showed that the Aln callus was stabilized at the site of fracture, despite the increased woven bone content and delayed remodeling of the Aln callus. Therefore, the larger volume and cross-sectional area (moment of inertia) of Aln appeared to be an adaptation that compensated for the Aln delay of woven bone remodeling into lamellar bone, which is structurally and mechanically superior to woven bone. Essentially, the same data at 6 weeks and 16 weeks postfracture were obtained previously with another bisphosphonate, incadronate.(14) The similarity between the effects of incadronate(14) and Aln indicate that increased size, greater mineral content, and strong inhibition of callus remodeling are effects characteristic of bisphosphonates as a class of antiresorptives. These data show that bisphosphonates do not prevent initiation of fracture healing and formation of callus; however, continued use did significantly impede the processes of callus remodeling. Recent long-term incadronate studies showed long-term retardation of callus remodeling as evaluated 49 weeks postfracture.(16) Other studies with the bisphosphonate clodronate(43) showed delay in the early stages of fracture healing, with no impediment of the repair process as long as treatment was halted after fracture. If clinically relevant, these animal fracture data taken together suggest that cessation of bisphosphonate treatment may be prudent for women on therapy who sustain a nonvertebral fracture.
Estrogen and Rlx had similar effects on fracture healing and were also similar to sham but different from Aln. Estrogen and Rlx calluses were similar in size to sham calluses at 6 weeks; but by 16 weeks, estrogen calluses were smaller than sham, while Rlx calluses were intermediate between sham and estrogen. Estrogen and Rlx appeared to be slightly inhibitory on the processes of fracture healing, but normalization by body weight showed that this was likely caused by estrogen and Rlx animals being significantly lighter than OVX, Aln, and sham animals at 6 weeks and 16 weeks postfracture. Estrogen and Rlx inhibited resorption activity (Oc.S/BS, N.Oc./BS) similar to Aln but mildly suppressed bone formation activity compared with Aln. Therefore, esstrogen and Rlx also reduced remodeling but had insignificant effects on the formation of new cortical shells and repair of the fracture plane, as compared with sham. Normalized biomechanical and intrinsic material properties of estrogen and Rlx calluses compared favorably with sham, indicating that mild suppression of remodeling with these compounds has no significant effect on fracture repair. Therefore, women on estrogen or Rlx who sustain a nonvertebral fracture have no reason to suppose a treatment effect on fracture repair.
In conclusion, Aln induced the formation of large, strong, highly mineralized fracture calluses in OVX rats but delayed callus remodeling by strongly suppressing remodeling of woven bone into lamellar bone. By contrast, estrogen and Rlx mildly suppressed callus remodeling, which had no effect on the progression of fracture repair and had intrinsic material properties that compared favorably with estrogen-replete controls.
We thank Dr. Charles Turner and Dr. David Burr (Indiana University) for recommending a weighted CSMI as a more appropriate correction to derive material properties, based on the non-rodlike geometry of the callus. We thank Miss Kawada and Miss Fukuda for helping to process the bone specimens. Yongping Cao is supported by a Japanese Scholarship from Ministry of Education and Science.