Early Changes in Cement-Bone Fixation Using a Novel Rat Knee Replacement Model

Abstract

Trabecular resorption from interdigitated regions between cement and bone has been found in postmortem retrieved knee replacements, but the viability of interdigitated bone, and the mechanism responsible for this bone loss is not known. In this work, a Sprague-Dawley (age 12 weeks) rat knee replacement model with an interdigitated cement-bone interface was developed. Morphological and cellular changes in the interdigitated region of the knee replacement over time (0, 2, 6, or 12 weeks) were determined for ovariectomy (OVX) and Sham OVX treatment groups. Interdigitated bone volume fraction (BV/TV) increased with time for Sham OVX (0.022 BV/TV/wk) and OVX (0.015 BV/TV/wk) group, but the rate of increase was greater for the Sham OVX group (p=0.0064). Tissue mineral density (TMD) followed a similar increase with time in the interdigitated regions. Trabecular resorption, when it did occur, started at the cement border with medullary-adjacent bone in the presence of osteoclasts. There was substantial loss of viable bone (~80% empty osteocyte lacunae) in the interdigitated regions. Pre-surgical fluorochrome labels remained in the interdigitated regions, and did not diminish with time, indicating that the bone was not remodeling. There was also some evidence of continued surface mineralization in the interdigitated region after cementing of the knee, but this diminished over time.

INTRODUCTION:

Initial fixation of cemented joint replacements is achieved through application of doughy cement with pressure to achieve mechanical interlock (interdigitation) between the cement and bone. Clinically, efforts to optimize this initial fixation have included pulsed lavage and brushing to remove marrow from the bone surface, vacuum mixing of the cement to remove porosity, pressurization of the cement in a doughy “does not stick to glove” state, and placement of the implant to potentially further pressurize the cement mantle. Laboratory research has shown that these factors can each contribute to enhanced interdigitation, which in turn results in a stronger cement-bone interface (1). The overall goal of the procedure is to prevent clinical (aseptic) loosening of the implant.

Despite these efforts, the initial implant fixation achieved at surgery is not maintained over time. Trabecular resorption in the interdigitated regions of the cement layer has been found in postmortem retrievals of human knee replacements (2). Further, the amount of trabecular resorption increases with longer time in service and with greater donor age (3). Trabecular resorption in the interdigitated regions results in weaker interfaces (4), and weaker interfaces could contribute to long term aseptic loosening of the implants. Clinically, this can be observed via progressive radiolucencies at the cement-bone interface (5). However, the sequence of biological events responsible for the loss of trabecular interdigitation is not known.

Histological examination of the cement-bone interfaces from postmortem retrievals has focused primarily on cemented femoral components of hip replacements with a focus on the cement-endosteal cortical bone interface. While there is cement-bone apposition, there is often not the mechanical interlock of cement with a trabecular bone bed found in the tibial component of knee replacements. Preclinical models used to investigate cement-bone interfaces to date have typically used a pre-polymerized poly(methyl methacrylate) (PMMA) rod placed in the medullary cavity of long bones (6; 7). There is also a large body of work investigating the effect of cement or polyethylene debris on bone osteolysis (8), alone, and in the presence of an implant (6; 7; 9). These approaches are useful to study the response of host bone to bulk PMMA or debris, but these preclinical models cannot recapitulate the cement-bone interdigitation found in knee replacements. Knee replacement models have been developed for the mouse (10), rat (11; 12), and rabbit (13), but these were not specifically designed to explore cement-bone interdigitation.

The cement-bone interdigitated region is unique in that the that the trabeculae are no longer surrounded by marrow, the trabeculae are stress shielded, and there are small gaps between the cement and trabeculae that can open and close when loaded (14). In this work, we developed a new rat knee replacement model that includes an approach to create an interdigitated cement-bone interface. The objectives of this study were to: 1) determine the morphological and cellular changes in interdigitated and medullary-adjacent bone over time following cementing of a knee replacement using a rat model, and 2) determine if the response is different for a model of osteoporosis via estrogen deficiency-induced bone loss with ovariectomy (OVX) compared to normal (Sham OVX) knee replacement.

METHODS:

All animal procedures were approved by our Institutional Animal Care and Use Committee. Female SAS Sprague Dawley rats were obtained from Charles River Labs (Wilmington, MA) and maintained in community housing (2-3 rats/cage, 22°C) on a 12 hour light/dark cycle with pellet chow (Formulab Diet 5008, LabDiet, St. Louis, MO) and water available ad libitum. Daily welfare observation and biweekly cage and bedding changes were performed in our AAALAC-accredited housing facility (PHS Assurance A3514-01). A total of 43 rats were used in this study with all surgeries performed at 12 weeks of age. Animals were assigned (6/time point) to one of 4 time points (0, 2, 6, 12 weeks) for the sham ovariectomy (Sham OVX) group. A second group (6/time point) received an ovariectomy (OVX) and was assigned to 2, 6, or 12 week time points. Statistical power analysis was conducted a priori, based on literature data from sham and OVX Sprague Dawley rats (bone volume fraction: BV/TV = 0.18±0.04 for sham and 0.06±0.02 for OVX at 10 weeks post-OVX (15)). A difference in BV/TV of 0.06 (one half the 0.12 difference between the sham and OVX groups) between groups could be detected with a power of 90% with a p value of 0.05. Allocation of animals to treatment groups and time points was randomized. A separate 0 week time point for the OVX treatment group was not performed because OVX was performed on the day of surgery. One animal was euthanized early (3 weeks post-op) due to progressive weight loss exceeding 20% body mass. Euthanasia was performed using carbon dioxide asphyxiation followed by cervical dislocation. Animal mass was measured on a biweekly basis throughout the study.

Implant Design

A tibial knee replacement (hemiarthroplasty) was designed to replace the tibial articulating surface and to recreate the general implant shape (tray and keel) found in human knee replacement. Custom polyether ether ketone (PEEK) (Ensinger GmbH, Nufringen, Germany) implants (Fig 1A) were machined with a computer numerical control (CNC) mill to fit the proximal articulating surface of the rat tibia. One implant size was used with a medial-lateral width of 6.0 mm and anterior-posterior depth of 4.8 mm. The articulating surface was polished using 1200 grit silicon carbide grinding paper with water irrigation. Implants were degreased and cleaned using Alconox detergent (White Plains, NY) and sonication in deionized water. PEEK was chosen as an implant material because it is relatively radiolucent and would not obscure cement or bone when μCT imaging. The overall goal of the design and surgical procedure was to achieve cement-bone interdigitation (Fig 1B & C) of metaphyseal bone as occurs in human knee replacement.

Figure 1.

Cemented joint hemiarthroplasty was performed using PEEK implants (A) cemented in the proximal tibial (B) with the goal of achieving regions of interdigitated bone (C). A 1mm thick section in the metaphyseal region was used to create masks and perform morphology analysis of various regions of bone.

Surgical Procedure

A unilateral (left) tibial knee replacement was performed on each animal. Under isofluorane anesthesia, a medial parapatellar arthrotomy was performed (Fig 2A), the patella was dislocated laterally, the joint was exposed after sectioning the ACL and removal of posterior-medial sesamoid bone (Fig 2B). A 1.0 mm diameter surgical burr was used to remove epiphyseal bone above the growth plate, resulting in a flat surface (Fig 2C). A cavity was created to accept the implant keel (Fig 2D), and a central drill hole was made to 4 mm depth. Following test placement of the implant, brush lavage (Fig 2E), saline irrigation were used to prepare the bone surface for cementing. Because a hindlimb tourniquet was not applied, a cellulose absorbent wipe was used to prevent blood flow out from the marrow canal while the cement was prepared. PMMA cement (Simplex, Stryker, Mahwah, NJ) was mixed by hand and applied to the bone surface when the cement reached a ‘does not stick to glove’ viscosity. A 1 mm stainless steel rod was used to push the cement into the distal hole (Fig 2F) and pressurize the cement into the metaphyseal trabecular bone. The implant was then applied and held in place until the cement cured (Fig 2H). Excess cement was removed, the patella was returned to the trochlear groove, and the incision was closed according to anatomic layers (Fig 2I) using interrupted absorbable sutures (5-0 Vicyrl, Ethicon Inc., Cornelia, GA). Interrupted suture and wound clips were used to close the skin incision.

Figure 2.

The surgical procedure (progression from A to I), demonstrated here on a cadaver rat, was performed using an operating microscope. The cement was stained with methylene blue to improve contrast for the images, but was not used in the actual surgical procedure. Implant width (medial-lateral) is 6 mm.

Ovariectomy (OVX) was performed through bilateral dorsal flank incisions. On each side, a single Vicryl ligature was placed around the oviduct followed by excision of the ovary and surrounding fat pad. The sham ovariectomy (Sham OVX) identified the ovary after incision, but left the organ intact. Incisions were closed with interrupted absorbable sutures according to anatomic layers, with wound clips applied to the skin incision. Postoperatively, animals were given 0.03 (day 1), 0.02 (day 2), and 0.01 (day 3) mg/kg buprenorphine at 12 hour intervals for three days. Animals exhibited antalgic gait the evening after surgery and 41 of 42 animals had full weight bearing and function by one week post-implantation. One animal had a poorly aligned implant and did not fully bear weight until week three.

Tissue processing

Fluorochrome labels were administered via subcutaneous injection 1 week prior to surgery (xylenol orange tetrasodium salt, 100 mg/kg, Sigma-Aldrich, St. Louis, MO) and 4 days prior to euthanasia (alizarin complexone, 18 mg/kg, Sigma-Aldrich, St. Louis, MO). Following euthanasia, tibias were harvested with surrounding soft tissue and fixed for 48 hours in 4% formaldehyde in Dulbecco’s PBS (with Ca²⁺, Mg²⁺). Micro-CT scans (55 kV, 145 mA, 200 ms integration time, µCT 40, Scanco, Brüttisellen, Switzwerland) with 8 µm voxel resolution were obtained of the proximal tibia. A hydroxyapatite phantom provided by the manufacturer was used as part of the weekly machine calibration process. The cemented implants were bisected in the sagittal plane using a 0.15 mm thick irrigated diamond blade (Isomet, Buehler Inc, Lake Bluff, IL). The lateral half was processed without decalcification and methyl methacrylate (MMA) embedded, sectioned in 0.15 mm intervals, and ground to 0.015 mm thickness for dynamic histomorphometry. The medial half was decalcified in 10% EDTA, paraffin embedded, and stained with hematoxylin and eosin, and tartrate-resistant acid phosphatase (TRAP) staining (Acid Phosphatase, Leukocyte Kit, Sigma-Aldrich, St. Louis, MO).

Morphological Analysis

CT scan-based morphological analysis was performed on a 1 mm thick axial section in the metaphyseal region just distal of the growth plate. Briefly, an isometric three-dimensional volumetric stack was created in MIMICS (Materialise, Leuven, Belgium) and bone was identified using a lower threshold of 800 mg/cc HA equivalent density. The margins of the cement and cortical bone were identified and masks were created for bone interdigitated with cement, medullary-adjacent trabecular bone, and cortical bone (Fig 1C). The boundary between the cement and medullary-adjacent trabecular bone was defined as the cement border. Trabecular morphologic measures were determined using the BoneJ plugin (16) for Fiji (17) and included bone volume (BV), bone volume fraction (bone volume per total volume (TV), (BV/TV)), trabecular thickness (Tb.Th), spacing (Tb.Sp), and connectivity density (Conn.D). Tissue mineral density (TMD) was determined from the average HA equivalent density for each mask region.

Statistics

Bone volume fraction (BV/TV) and tissue mineral density (TMD) were the primary morphological outcome variables. Additional morphological metrics are included in Supplemental Table 1. Linear regression (JMP, SAS, Cary, NC) was used to determine if the independent variables (BV/TV, TMD) were related to time, as the primary goal was to explore bony changes over time. First, simple linear regression was applied based on region (interdigitated bone, medullary-adjacent bone, and contralateral trabecular bone (not implanted)) and treatment group (Sham OVX and OVX). Following this, analysis of covariance (ANCOVA) was used to determine if the rate of accrual (or loss) was different for interdigitated versus medullary-adjacent or contralateral bone with time as the covariate. Descriptive statistics were determined for each of the histological measures at each time point and treatment group. Overall comparisons were made between groups, where appropriate, via two-sample and paired t-tests.

RESULTS:

Bone Morphology and Mineralization

The interdigitated bone volume fraction (Int BV/TV) increased with time (Fig 3) for the Sham OVX (p<0.0001) and OVX groups (p=0.0005). In contrast, there was not an increase in BV/TV for the medullary-adjacent or contralateral regions for the Sham OVX group. Using ANCOVA, the rate of increase (ΔBV/TV/wk) for the Sham OVX treatment was greater for bone in the interdigitated region compared to medullary-adjacent (p=0.0003) or contralateral (p=0.019) regions. For the OVX treatment, there was a decrease in BV/TV with time in the medullary-adjacent (p=0.0016) and contralateral (p=0.018) regions. Finally, the rate increase in interdigitated BV/TV for the Sham OVX group was greater than for the OVX group (p=0.0064). Along with the increased BV/TV, there was an increase in trabecular thickness in the interdigitated region with time (p<0.0001) (Suppl Table 1). However, trabecular spacing and connectivity density did not change with time of implantation.

Figure 3.

Trabecular bone volume fraction (BV/TV) of interdigitated, medullary-adjacent, and contralateral (not cemented) regions was used as a primary morphological outcome. Regression models with BV/TV as a dependent variable and time as the independent variable was used to determine if there was a significant increase or decrease in BV/TV with time. ANCOVA was used to compare slopes (ΔBV/TV) between interdigitated and medullary-adjacent or contralateral regions. N=6/time point for Sham OVX and OVX groups. Mean and standard error bars are shown.

Tissue mineral density (TMD) also increased with time (Fig 4) in the interdigitated region and medullary-adjacent regions for Sham OVX and OVX treatment groups (p<0.0001). The contralateral trabecular regions did not result in an increase in TMD with time. The rate of increase for interdigitated and medullary-adjacent regions was not different for either Sham OVX or OVX groups. Taken together, the morphology results suggest that there is continued mineralization of interdigitated bone, irrespective of treatment, and there could also be an increase in the quantity of interdigitated bone as documented by an increased BV/TV and trabecular thickness.

Figure 4.

CT-based tissue mineral density (TMD) of interdigitated, medullary-adjacent, and contralateral (not cemented) regions was used as a measure of bone mineralization. Regression models with TMD as a dependent variable and time as the independent variable was used to determine if there was a significant increase or decrease in TMD with time. ANCOVA was used to compare slopes (ΔTMD) between interdigitated and medullary-adjacent or contralateral regions. N=6/time point for Sham OVX and OVX groups. Mean and standard error bars are shown.

Dynamic Histomorphometry

Interdigitated regions (Fig 5a, top) retained the pre-surgery label at all time points and also exhibited the pre-euthanasia label indicating that bone formation continued at the cement interface after cementing. There was also evidence of the second label at the cement border between medullary-adjacent bone and the cement (Fig 5a, bottom). The contralateral limb had significantly less pre-surgical labeled surface fraction (MS/BS) for the Sham OVX (p=0.0024) and OVX (p<0.0001) treatment groups compared to the interdigitated region (Fig 5b). In contrast, the contralateral limb had greater MS/BS compared to the interdigitated region for the pre-euthanasia label for Sham OVX (p=0.0102) and OVX (p=0.034) groups. Together these results suggest that the interdigitated bone is not remodeling after cementing of the bone (retained pre-surgical label) and that the mineralization of interdigitated bone after cementing is substantially diminished compared to normal trabecular bone.

Figure 5.

Undecalcified ground sections (A) of interdigitated regions (top) and cement border (bottom) show presence of pre-surgical (xylenol orange) and pre-euthanasia (alizarin complexone) labels for a 6 week Sham OVX knee. Single label surface fraction (MS/BS) was determined (B) for interdigitated and contralateral (not implanted) regions. N=3/time point for Sham OVX and OVX groups. Mean and standard error bars are shown.

Few Osteoclasts in Interdigitated Regions

The loss of medullary-adjacent bone in the OVX group resulted in interdigitated trabecular bone segments that were not connected to adjacent trabeculae (Fig 6A, top). There was evidence of resorption of some of the original interdigitated bone, starting at the cement border, and this also coincided with the presence of osteoclasts (Fig 6A, bottom). Overall, the fraction of bone surface with TRAP positive cells (Oc.S/BS) for medullary-adjacent and contralateral trabecular bone was much greater than interdigitated regions (p<0.0001) or cement border (p<0.0001). In particular, there were few if any osteoclasts present in the sampled interdigitated regions. The largest proportion of osteoclasts was found in the contralateral limb, at 2 weeks, and in the OVX group. This is consistent with the loss of contralateral trabecular bone in the OVX group.

Figure 6.

TRAP stained sections (A) were used to document presence of osteoclasts in interdigitated, cement border, medullary adjacent, and contralateral regions. Image shown is from a 6 week OVX case and shows local resorption of interdigitated bone starting at the cement border. Osteoclast surface (Oc.S/BS) was measured (B) for N=3/time point for Sham OVX and OVX groups. Mean and standard error bars are shown.

Non-viable Bone in Interdigitated Regions

Hematoxylin and eosin stained sections were used to identify vacant or nucleated osteocyte lacunae, and these were used to create digital masks of non-viable or viable bone (Fig 7A). At implantation (time 0) 98-100% of osteocyte lacunae were nucleated. Subsequently, viable bone had diminished to only 9-23% area fraction in interdigitated bone regions at 2, 6, and 12 weeks (Fig 7B). The medullary-adjacent bone also had a reduction in viable bone (avg 62-67% viable) following implantation, whereas the cortical bone had 72-85% viable bone following implantation at 2 to 12 weeks. For the interdigitated bone, over the 2-12 week time period, there was a weak positive correlation (r²=0.21, p=0.056) between viable bone fraction and time (weeks), suggesting that may be some recovery of viable interdigitated bone with time. The cement border generally served as a demarcation zone with non-viable bone on the interdigitated side, and viable bone on the medullary-adjacent side.

Figure 7.

H&E stained sections (A) were used to identify empty or nucleated osteocyte lacunae in interdigitated, medullary adjacent, and cortical regions. Image shown is from a 12 week Sham OVX case. Viable bone fraction (B) was quantified from masks created in the three regions measured for N=3/time point for Sham OVX and OVX groups. Mean and standard error bars are shown.

DISCUSSION:

In this work we developed a new, cemented arthroplasty model in the rat knee that includes interdigitation between the cement and trabecular bone to mimic the clinical situation found in human knee replacements. Using this model, we found that the interdigitated bone resorption started at the cement border. This pattern of resorption is similar to what has been documented in human postmortem retrievals (18). However, there was not extensive progression of resorption into the interdigitated region in the rat knee over a twelve week period, in contrast to the extensive resorption seen over the long term (decades) in human retrievals (3). It is possible that the resorption process would continue with longer follow-up.

In human postmortem knee replacement retrievals, there was greater loss of cement-bone interdigitation from older donors and with longer time in service (18). An ovariectomy model was used in this study to induce osteoporosis via estrogen deficiency with the goal of simulating osteoporosis in an elderly population. In the current study, while the rate of accrual of interdigitated bone was greater for the Sham OVX group compared to the OVX group, neither group lost bone with time. In this context, the OVX model did not recapitulate conditions of the human clinical population, possibly due to the short time course of the study (12 weeks) or due to the young age of the animals (12 weeks old). There was rapid and substantial loss of medullary-adjacent bone in the OVX model, with near complete loss of medullary-adjacent bone at 12 weeks. This also does not mimic the elderly postmortem retrieval population where some supporting trabecular bone remains (a decrease in BV/TV of 50% from age 55 to 90 years) (18). Overall, the OVX model does not appear to appropriately simulate the morphological changes seen in the elderly clinical population. Use of an older animal at time of implantation, and allowing for longer time in service could produce a model representative of the older clinical population. It should be noted that younger human donors with shorter time in service often maintain interdigitated bone, consistent with what was found in the Sham OVX group.

The most surprising finding from this study was the continued mineralization of the interdigitated bone over time following cementing of the tibia. Hyper-mineralization of bone is known to occur in auditory ossicles and is accompanied by osteocyte death and low bone remodeling (19; 20). Several proteins have been identified as having roles in controlling local matrix mineralization (21), and osteocyte death through apoptosis or necrosis could release these proteins locally, contributing transiently to local matrix mineralization. The hyper-mineralized interdigitated bone found in the current study exhibited extensive osteocyte death and low bone turnover, suggesting that the mechanism of action may be similar to that documented in auditory ossicles. Furthermore, studies by Marsh et al demonstrated that necrotic cell debris can nucleate mineral deposition on devitalized, osteoblast secreted matrix (22), in an alkaline phosphatase dependent manner, perhaps by latent rupture of matrix vesicular bodies (23).

In addition to hyper-mineralization of interdigitated bone tissue, there was also new mineralization (as documented by fluorochrome label) at the bone surface opposing the cement following cementation. Small gaps at the trabeculae-cement border are present (24) in the interdigitated regions due to incomplete filling of marrow spaces with cement, as well as cement shrinkage during polymerization (25). This would leave space for diffusion of mineral ionś and also space for continued matrix mineralization between the trabecular bone surface and cement.

There has been long-standing interest in the biological response and long term viability of PMMA cement since the first applications in hip replacement in the 1960s. Charnley (26) found that there were some areas of cement-bone contact in hip replacement retrievals, but also areas with fibrocartilage formation. A spectrum of tissue reactions from cement-bone contact to fibrous membranes with foreign body giant cells has also been described (27). Transformation of tissues at the interface to synovial-like tissue is known to generate prostaglandin E₂ (28), which in turn stimulates osteoblasts to release factors promoting osteoclastic bone resorption. Along with the changes at the interface leading to progressive loosening, bone necrosis adjacent to the cement has been documented in human and animal studies (29; 30). Necrosis is usually described for cortical bone regions adjacent to the cement, and is often attributed to thermal damage due to exothermic reaction of the curing cement, local cytotoxic effect of the monomer, and interruption of osseous vasculature (29). In the current study, we found non-viable bone in medullary-adjacent regions, and the non-viable bone appeared to be located near the cement border. We believe we can rule out thermal necrosis as a mechanism in our model, because the volume of cement used is very small (~50 mm³). The peak temperature of polymerizing cement increases with larger cement volumes (31). In pilot bench top work, we found a temperature increase of only 4°C over ambient conditions (30°C) for polymerization of a small (100 mm³) bolus of Simplex cement. This is much less than the 50°C threshold described for cell death (32). Therefore, it is likely that local cytotoxic effects of the monomer and disruption of the vasculature contribute to the necrosis within the interdigitated regions and regions adjacent to the cement border.

Studies that explore morphological and cellular changes to interdigitated cement-bone constructs are much more limited when compared to the medullary (cortical-cement) constructs described above. Necrosis has been found in bone adjacent to cement in human vertebroplasty retrievals (33) and in a rabbit vertebroplasty model (34). Retrieval of cemented hip resurfacing constructs (35) have documented 55% empty lacunae in interdigitated regions, suggesting that some of the interdigitated bone may not be viable. In addition, staining via tetracycline label given to patients 5 days prior to revision was not present in the interdigitated regions. This lack of pre-revision tetracycline label is consistent with the limited pre-euthanasia label (particularly at 12 weeks) found in the interdigitated region of the present study. However, the hip resurfacing results could also be confounded by avascular necrosis of the femoral head that may be present at the time of hip resurfacing (36).

There are several limitations to this study. It is difficult to directly relate rat and human ages, but sexual maturity in the rat is reached at about the sixth week. By 12 weeks, the majority of skeletal growth has occurred, but there is still is some tibial lengthening through 18 weeks, weight continues to be accrued (Figure S1), and there is no epiphyseal closure (37). Reproductive senescence does not occur until 15-20 months (38). Therefore, 12-week-old rats may be considered to be relatively young compared to the elderly human total joint population. The mechanism(s) responsible for osteocyte death is not known and could occur through apoptotic or necrotic mechanisms. In addition, it is not known if interdigitated areas that remain viable are associated with vascular supply and if there could be longer term remodeling and increased viability of interdigitated bone. The length scale of the model with cement filling about 3 mm of space is substantially less than found in human joint replacement where cement layers of >10 mm may occur. Greater amounts of cement could alter the damage mechanism to bone through the addition of thermal necrosis.

The use of a rat model without epiphyseal closure of the growth plate resulted in additional challenges in this model system to recapitulate the human tibial tray construct. Ideally, the PEEK implant could be cemented to the epiphyseal bone alone, followed by analysis of the cement-bone interface under the tray. However, the cement-bone interdigitation in the epiphyseal region (~0.5mm) would be very limited. In addition, epiphyseal trabecular bone exhibits very little remodeling when compared to metaphyseal bone in the rat. Once cemented, there is negligible marrow in the epiphyseal region, and no access to metaphyseal marrow due to the presence of the growth plate. We felt it important to have ‘medullary-adjacent’ bone with associated marrow in our analysis zone, as bone marrow is the primary reservoir of bone remodeling cells, and is present in human cement-bone constructs. Surgical removal of the epiphysis was also considered with implant placement directly on the metaphyseal bone. Unfortunately, this compromises the collateral ligaments and the patellar tendon insertion, and was not pursued as a surgical approach.

From a clinical perspective, it is clear that cemented knee replacements function successfully for the majority of patients without clinical loosening. However, late aseptic loosening is a clinical concern with revision rates of ~10% at 20 years (39). If the interdigitated bone is not viable, and is only slowly resorbed (starting at the cement border), then the remaining interdigitated bone would serve little clinical function. The loss of interlock at the cement border also might limit the ability of the viable bone adjacent to the cement border to create sufficient interlock to support mechanical loads across the joint. Efforts to maintain or restore bone viability in the interdigitated region could help long-term function of cemented joint replacements.

While the use of young rats, even with ovariectomy-induced osteoporosis, does not accurately replicate the bone resorption phenotype of elderly human patients, the interdigitated cement-bone preclinical model developed here recapitulates the cement-bone interface of clinical implants. Little is known about the biological fate of interdigitated bone supporting joint replacement implants. The model presented here provides a new platform that could be used in the future to explore how changes in cement formulation or pharmacological interventions could modulate the viability, longevity, and quality of interdigitated bone supporting total joint implants.

Supplementary Material

Supp figS1

Figure S1. Rat body mass was recorded at 2 week intervals. Mean and standard error bars are shown.

Statement of Clinical Significance:

Interdigitated bone with cement provides mechanical stability for success of knee replacements. Improved understanding of the fate of the interdigitated bone over time could lead to a better understanding of the loosening process and interventions to prevent loss of fixation.