Peri-implant bone remodeling is often studied in relation to how the implant modifies the local mechanical environment1–3 and as a consequence of release of particles from the implant.4, 5 However, surprisingly little is known about the initial bone remodeling response to the placement of the implant. This initial response is probably best thought of in the context of the “regional acceleratory phenomenon” (RAP),6 which has been described as the response of bone and soft tissue to stimuli like surgery.7 The initial response to the injury is thought to be dominated by woven bone formation followed by a remodeling phase.6 For loaded dental implants, there is evidence that elevated bone remodeling in the immediate vicinity of the implant persists indefinitely and may be integral to maintaining osseointegration.8 In an orthopedic model, porous coated implants placed in the distal medial condyle of sheep showed elevated trabecular remodeling with a variable temporal response.9
Knowledge of the RAP can prove useful. For instance, in dentistry increasing the severity of the RAP accelerates the rate of orthodontic tooth movement.10 Even if the host bone response to placement of the implant is only transient, it could still have important consequences by tempering the longer term remodeling related to implant-induced changes in the bone's mechanical environment or to particle-induced changes in the local microenvironment.
Many orthopedic procedures such as joint replacement include disruption of the medullary cavity. We recently provided a thorough histologic and genomic description of changes in the medullary cavity in a rat model following mechanical ablation of the marrow cavity.11 This study showed an early inflammatory response in which woven bone formed within 5–7 days, followed by bone resorption and an apparent reconstitution of the marrow microenvironment within 28 days.11 Although we did not examine bone remodeling dynamics, the expression of most bone remodeling genes such as collagen type I, alkaline phosphatase, osteocalcin, cathepsin K, TRAP5, and RANK had returned to baseline values by 28 days. In a rat osteotomy model, the resorptive phase of the RAP was found to conclude by 21 days.12 These studies suggest that the RAP is transient in rat models following surgically induced injury, subsiding within 3–4 weeks.
Several studies used dynamic histomorphometry to identify the effects of implant treatments on peri-implant bone remodeling in animal models,9, 13–15 the effect of cortical screw placement on osteonal remodeling in the dog,16 and bone remodeling in close proximity to knee replacement implants in humans.17 It is unclear from these studies, however, if the RAP is transient or persistent when an intramedullary implant is placed at the site of skeletal injury.
As the population ages, arthroplasty rates are predicted to increase,18 and postmenopausal osteoporosis will become more prevalent.19 While peri-implant bone remodeling is considered an important determinant of the ability to maintain mechanical fixation of implants, it is unknown if the remodeling response to placement of an implant is transient or if it differs in normal and osteopenic bone. Increased knowledge of the bone remodeling response in normal and osteoporotic bone may give insight into new therapies aimed at increasing the bone-implant construct strength.20, 21 In addition, no information exists on potential differential effects at the trabecular, endocortical, and periosteal surfaces. We hypothesized that intramedullary implant placement in intact and ovariectomized (ovx) rats increases bone remodeling at all three bone envelopes in a time-dependent manner. Additionally, we sought to determine if implant placement and ovariectomy have synergistic effects on bone remodeling.
In an IACUC-approved study, eighty-eight 4.5 months old Sprague–Dawley rats (Harlan, Indianapolis) received ovariectomy (ovx) or sham-ovx (n = 44 in each group). At 11 months of age, 8 sham-ovx and 9 ovx animals were euthanized (baseline groups) while the remaining 71 rats underwent bilateral implant placement and were maintained for periods of 4, 8, or 12 weeks. All animals were euthanized via CO2 inhalation.
Similar surgical procedures were reported by our lab previously.22 Briefly, rats were anesthetized with ketamine and xylazine (100 and 5 mg/kg, respectively) via intraperitoneal injection. Under aseptic conditions, a 1 cm incision was made medial to the patella, and the patella was reflected laterally exposing the distal femoral condyles. A Dremel tool with a 1.5 mm tip was used to penetrate the subchondral bone through the patellar groove, and the drill bit was then used to ream the marrow cavity to the level of the lesser trochanter. The marrow cavity was then irrigated with 10 ml of sterile saline, followed by insertion of a 15 mm-long dual acid-etched titanium rod, 1.5 mm in diameter (Goodfellow, Oakdale, PA).23 The opening in the subchondral plate was plugged with bone wax, the subcutaneous tissue and skin were sutured separately, and the wound was protected with staples. Baseline control rats received no surgical intervention. No complications were observed during implant surgery.
One femur from each animal was designated for scanning and subsequent histomorphometry. The implants remained in situ, and the bones were fixed and scanned in 10% neutral buffered formalin. Scanning was done with a laboratory-based system (model 40, SCANCO USA, Inc., Southeastern, PA). The bones were scanned perpendicular to their long axis at 70 kV, 0.3 s integration time, 1,000 projections/slice, and 2,048 samples/projection, at an isotropic voxel size of 16 µm. A 0.64–1.6 mm thick (40–100 slices) region of interest ∼0.5 mm proximal to the growth plate was chosen for analysis (Fig. 1). The analyzed region included the space between the endocortical surfaces, omitting the 48 µm nearest to the implant to reduce the effect of metal-induced artifacts.24 Comparable regions were scanned and analyzed in the baseline control rats. The threshold chosen to segment the bone was 270 on the Scanco grayscale. The femurs were scanned over a restricted time period, and routine calibrations showed no drift in the scanner's gray scale values. Four trabecular architecture parameters were examined: bone volume per tissue volume (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular spacing (Tb.Sp). These variables were determined using the direct transform method. No evidence of fractures or implant misplacement was noted during scanning. Sample sizes were eight in the sham-ovx baseline group, 9 in the ovx baseline group, 36 in the sham-ovx implant group, and 35 in the ovx implant group (88 total rats).
All rats received subcutaneous calcein injections (10 mg/kg, Sigma-Aldrich, St. Louis, Missouri, USA) at 14 and 4 days prior to euthanasia. Femurs were harvested, fixed in 10% neutral buffered formalin for 3 days, and then placed in 70% ethanol for storage. Sections from the distal metaphysis perpendicular to the long axis of the bone and implant were prepared for dynamic and static histomorphometry at the trabecular and endocortical surfaces; histomorphometry at the periosteal surface was performed at the midshaft superior to the implant (Fig. 1). The implants remained in situ during histological processing. The sections were cut, ground, and polished to ∼100 µm thickness. The fluorochrome labels were used to determine the mineralizing surface per bone surface (MS/BS), mineral apposition rate (MAR), and bone formation rate per bone surface (BFR/BS) following standard methods.25 MS/BS was calculated based on the double label plus one-half the single label method. After analysis of the fluorochrome labels, the sections were stained with basic fuchsin and toluidine blue, and static histomorphometry was performed to determine eroded surface per bone surface (ES/BS). The linear extent of periosteal reaction was measured at the distal site (Fig. 1). Sections were evaluated without knowledge of group membership using commercial software (OsteoMeasure, OsteoMetrics Atlanta, GA). Sample sizes were five in each baseline group and seven–nine at each time point for both the sham-ovx and ovx groups (59 total rats).
Data from one femur in each rat were assessed. One-way ANOVAs were used to determine the effects of implant placement within the sham-ovx and ovx groups, using Bonferroni-corrected post-hoc tests to compare time points (i.e., baseline vs. 4, 8, and 12 weeks). To further explore the effect of ovx on the response to implantation, baseline and 4 week data were evaluated with two-way ANOVAs with presence/absence of the implant and presence/absence of ovx as the between-subjects factors. Means and 95% confidence intervals are reported. Significance was inferred when p < 0.05. Exact p-values are listed. Statistical analyses were performed with commercial software (SPSS version 19, Chicago, IL).
Effect of Implant Placement on Bone Volume and Trabecular Architecture
Neither cortical area nor trabecular BV/TV were affected by implant placement (Fig. 2). Some trabecular architecture parameters did vary as a function of implant placement, including Tb.Th in the sham-ovx group (p = 0.044), Tb.N in the ovx group (p = 0.033), and Tb.Sp in the ovx group (p = 0.045). However, these effects were small, and none of the post-hoc comparisons between time points were significant.
Effect of the Implant on Trabecular Bone Remodeling
In sham ovx rats, MS/BS, MAR, and BFR were elevated at either 4 or 8 weeks post-implantation with a return toward baseline values at 12 weeks (Fig. 3). In the ovx rats, post-implantation increases occurred in MAR and BFR at 4 weeks, and these values, although somewhat variable, tended to remain above baseline values (Fig. 3). Implant placement did not affect trabecular bone eroded surface in either the sham or the ovx rats (Fig. 3).
Effect of the Implant on Endocortical Remodeling
In sham-ovx rats, MS/BS was increased at 4 and 8 weeks compared to baseline while MAR and BFR/BS were increased at 8 weeks compared to baseline with a return towards baseline values at 12 weeks (Fig. 4). In ovx rats MS/BS, MAR, and BFR/BS were increased at 4 weeks compared to baseline (Fig. 4). Placement of the implant did not affect endocortical bone eroded surface in either the sham or the ovx rats (Fig. 4).
Effect of the Implant on Periosteal Remodeling
There was a significant periosteal reaction at the distal site used to evaluate bone remodeling in 11 rats that received implants (Fig. 5). The reaction was composed primarily of woven bone with diffuse, disorganized fluorochrome label present throughout. The periosteal reaction was most notable in ovx rats that received implants and peaked at 4 weeks (Fig. 5, p = 0.005). Although a similar pattern was observed in the sham-ovx rats, the effect was not significant (p = 0.051). Because of the presence of this reaction, periosteal dynamic histomorphometry was performed at the midshaft. In sham-ovx rats MS/BS was increased at 4 weeks compared to 12 weeks while MAR and BFR/BS were increased at 4 weeks compared to baseline (Fig. 6). In ovx rats MS/BS, MAR, and BFR/BS were increased at 4 weeks compared to baseline (Fig. 6). No differences in ES/BS were found in sham-ovx or ovx rats (Fig. 6).
Interaction Between Implant Placement and ovx on Bone Remodeling
To assess interaction between implant placement and ovariectomy on bone remodeling, we performed two-way ANOVA's with presence/absence of ovx and presence/absence of an implant as the between-subjects factors. Because some time-related differences existed among the groups with implants (see above), this analysis was restricted to the baseline and 4 weeks groups. Of the 18 variables examined, only the endocortical MAR and endocortical BFR/BS had significant ovx-by-implant placement interaction terms (p = 0.002 and 0.004, respectively), indicating that the effect of implant placement on bone remodeling was not strongly different in sham-ovx and ovx rats.
While there have been indirect measures of an implant's influence on bone remodeling26, 27 and more direct studies,9, 13–17 to our knowledge this is the first study to measure the effect of an intramedullary implant on peri-implant bone remodeling in a time-dependent manner. Compared to baseline controls, both sham-ovx and ovx rats had increased rates of trabecular, endocortical, and periosteal bone formation 4 or 8 weeks after implant surgery. In general, this initial phase of remodeling had largely subsided by 8–12 weeks. Thus, the hypothesis that intramedullary implant placement alters bone remodeling kinetics in a time-dependent manner was supported. Additionally, only limited evidence was found that the remodeling response to placement of the implant was different in normal and osteopenic bone.
In both sham-ovx and ovx rats, implantation did not affect bone volume compared to non-implant baseline controls as assessed by cortical area and trabecular BV/TV. In addition, none of the other three measures of trabecular architecture varied significantly between the baseline and post-implantation time points. However, placement of the implant led to increased bone formation rates, particularly at 4 and 8 weeks, but no differences were seen for ES/BS, the one measure of bone resorption. These observations are somewhat paradoxical because the lack of change in bone volume implies that the increases in bone formation must have been offset by increases in bone resorption. The period of elevated bone resorption may have occurred before our earliest observation time of 4 weeks, or our method of assessing bone resorption may not have been sufficiently sensitive. Because of the presence of the implant, we used ground sections and assessed bone resorption in terms of eroded surface. While it would be preferable to examine osteoclast number to estimate resorption, osteoclasts are not always easily identified in the thick ground sections necessitated by the presence of the implant.
Bone formation markers tended to be elevated longer at the trabecular and endocortical surfaces compared to the periosteal surface, indicating that the implant's effect was most pronounced in close proximity to the site of implant placement, consistent with a previous prediction6 and direct observations.17, 28 The differences between trabecular and endocortical versus periosteal surfaces could also be due to the prevalence of modeling at the periosteal surface and remodeling at the other two surfaces.
Since increased severity of the RAP is associated with accelerated tooth movement in dentistry,10 it is worth considering if the severity of the RAP in orthopaedics could influence implant migration. This possibility has not been directly tested, but on the one hand the use of bone remodeling inhibitors such as bisphosphonates is associated with depressed migration of cemented tibial components.29 Conversely, a more recent study of cementless tibial components found no difference in implant migration30 in total knee replacement following bisphosphonate treatment.
Although not directly a focus of study, baseline ovx rats had reduced trabecular bone volume and increased bone remodeling compared to the sham-ovx baseline rats. The reduction in trabecular bone volume is consistent with previous work in the rat model, but the higher values for the dynamic histomorphometric formation variables were unexpected because the accepted paradigm is that ovx induces a transient phase of elevated remodeling, leading to bone loss, followed by a new steady state in which the bone remodeling rates return to baseline values.20, 31–34 Our study included a longer period between ovx and observation (6.5 months) than most previous studies (1–4.5 months), perhaps explaining this difference from earlier studies.
A limitation of our study was the lack of age-matched non-operated control rats at the 4, 8, and 12 weeks time points. Other authors found minimal bone was loss in either sham-ovx or ovx rats of the same age as our rats over a 12 months period,35 so it is unlikely that significant bone loss would have occurred over our much shorter follow-up times. Another limitation is that we did not assess implant micromotion, which could have differed between the sham-ovx and ovx rats. Subjectively, implant placement required more manual effort in ovx than sham-ovx rats, presumably due to reduced trabecular bone volume. However, at surgery the implants were deemed to be press fit in both groups, and histology did not show a fibrous membrane suggestive of micromotion at the interface in either group. Finally, our implant, while placed in a weight-bearing bone, did not directly transmit load. An advantage of using a non-weight-bearing implant is that the confounding effects of longer term remodeling associated with the altered mechanical environment or the presence of particles shed from the implant are greatly reduced.
Previously, the injury-induced RAP in a rat osteotomy model resolved in 3 weeks,12 consistent with our recent report on gene expression following marrow ablation in the rat in which no implant was placed.11 The bone remodeling data in the present study indicate that the presence of the implant led to elevated remodeling through ∼8 weeks. Thus, we propose that the presence of the implant prolongs the injury-induced RAP. In the absence of direct data on remodeling rates in the marrow ablation model without subsequent placement of an implant, this is only an inference. Nevertheless, it is impossible to place an implant without first having performed surgery, and the inherent transient in peri-implant remodeling rates may need to be considered in future experimental designs.
The project was supported by NIH Grant T32 AR052272 and a grant from Amgen, Inc., USA. Dr. Thomas Wronski provided guidance on histomorphometry.