Abstract
With in vivo service there is loss of mechanical interlock between trabeculae and PMMA cement in total knee replacements. The mechanisms responsible for the loss of interlock are not known, but loss of interlock results in weaker cement-bone interfaces. The goal of this study was to determine the pattern of resorption of interdigitated bone using a series of 20 postmortem retrieved knee replacements with a wide range of time in service (3 to 22 years). MicroCT scans were obtained of a segment of the cement-bone interface below the tibial tray for each implant. Image processing methods were used to determine interface morphology and to identify supporting, interdigitated, resorbed, and isolated bone as a function of axial position. Overall, the amount of remaining interdigitated bone decreased with time in service (p=0.0114). The distance from the cement border (at the extent of cement penetration into the bone bed) to 50% of the interdigitated volume decreased with time in service (p=0.039). Isolated bone, when present was located deep in the cement layer. Overall, resorption appears to start at the cement border and progresses into the cement layer. Initiation of trabecular resorption near the cement border may be a consequence of proximity to osteoclastic cells in the adjacent marrow space.
Keywords: Implant, loosening, cement, bone, knee replacement
Graphical abstract
With in vivo service there is loss of mechanical interlock between trabeculae and cement in knee replacements. The goal of this study was to determine the pattern of resorption of interdigitated bone using 20 postmortem retrieved knee replacements with a range of time in service. Resorption starts at the distal cement border and progresses into the cement layer. Initiation of trabecular resorption near the cement border may be a consequence of proximity to osteoclastic cells in the adjacent marrow space.
INTRODUCTION
Aseptic loosening continues to be the leading cause of revision in cemented total knee replacements (TKRs). Progressive radiolucencies, component migration and pain are the clinical hallmarks of the loosening process. Recent work with postmortem retrieved knee replacements has shown that the interlock between the trabecular bone bed and poly(methyl methacrylate) (PMMA) cement does not always remain in place with in vivo use (1–3). The loss of interlock results in interfaces with more interface micromotion when loaded (4) and results in weaker joint replacement constructs (5).
Resorption of the trabeculae from within the interlocked cement layer leaves cavities in the shape of the original interlocked trabecular structure (2). The mechanism driving this resorption process is not known, but stress shielding of the trabeculae (6; 7), fluid flow induced osteolysis (8–10), monomer toxicity (11), and thermal necrosis (12) have been proposed. While the amount of fixation has been described in terms of interdigitation depth and cement-bone contact fraction (2; 3) with in vivo service, the pattern of the trabecular resorption process has not been explored.
Using micro-CT image sets of the cement-bone interface from postmortem retrievals, an image processing approach was recently developed to identify locations of trabecular resorption (13) in the interlocked regions. This approach could be used to map the progression of the trabecular resorption process. For example, it is not currently known whether trabecular resorption initiates deep in the cement layer and then progresses towards the border between the cement layer and supporting trabecular bone bed (or vice versa). Alternatively, the resorption pattern could be uniform or random throughout the interlocked region. Using a series of retrieved cement-bone interfaces from knee replacements with a wide range of time in service, the goal of this study was to document the temporal and spatial distribution of trabecular bone resorption. We hypothesized that the amount of trabecular resorption would increase with time in service and donor age, and that resorption would initiate at the cement border with the supporting bone bed and extend into the cement layer with time in service.
METHODS
Cement-bone interface specimens were obtained from the underside of the tibial tray of 20 postmortem retrieved cemented total knee replacements (Table 1). The fresh-frozen knee replacements were obtained from the SUNY Upstate Anatomical Gift Program at the time of death. Donor age ranged from 54 to 90 years old and time in service ranged from 3 to 22 years. There were 15 donors total with 5 donors having bilateral implants. All implants used a metal backed tray design, and 16 of 20 had a central keel. The intact constructs were radiographed in the anterior-posterior and medial-lateral planes. Radiolucencies along the implant-bone interface were scored as no interface radiolucencies (n=13), limited peripheral radiolucencies (n=7), or extensive radiolucencies (n=0). There were no cases of focal osteolytic lesions.
Table 1.
Donor Information for 20 postmortem-retrieved, metal backed tibial components from total knee replacements. There were 15 donors total with 5 donors having bilateral implants. CR/PS: Cruciate Retained/Posterior Stabilized.
| Donor | Age (years) | sex | Weight (kg) | BMI | Time in service (years) | CR/PS | Manufacturer/type | Keel | Radiographic lucencies |
|---|---|---|---|---|---|---|---|---|---|
| A | 78 | M | 65 | 18.5 | 10 | PS | Depuy/AMK | y | none |
| B | 77 | F | 99 | 32.3 | 13 | PS | StrykerScorpio | y | none |
| C | 87 | F | 68 | 23.9 | 3 | PS | Zimmer/Nexgen | y | none |
| D | 69 | F | 62 | 22.5 | 11 | PS | Wright Medical/Advance PS | y | none |
| E | 61 | M | 103 | 32.9 | 5 | CR | Stryker/Triathlon | y | none |
| F | 85 | F | 85 | 36.5 | 22 | PS | Howmedica/Duracon | n | none |
| G | 85 | F | 85 | 36.5 | 10 | CR | Richards/RMC | y | none |
| H | 87 | F | 75 | 29.3 | 5 | CR | Zimmer/Nexgen | n | peripheral |
| I | 87 | F | 75 | 29.3 | 13.5 | CR | Zimmer/Nexgen | n | peripheral |
| J | 83 | F | 69 | 23.7 | 6.5 | PS | Stryker/Scorpio | y | none |
| K | 84 | M | 53 | 23.6 | 8 | PS | Centerpulse/Natural Knee II | y | peripheral |
| L | 76 | F | 91 | 30.8 | 7 | PS | Zimmer/Nexgen LPS | y | none |
| M | 86 | F | 78 | 28.3 | 18 | PS | Zimmer/ Insall Burstein II | y | none |
| N | 86 | F | 78 | 28.3 | 14 | PS | Zimmer/Insall Burstein II | y | peripheral |
| O | 90 | M | 80 | 22.7 | 10 | CR | Zimmer/Nexgen | n | none |
| P | 85 | M | 96 | 32 | 11 | CR | Smith & Nephew/Genesis II | y | none |
| Q | 72 | M | 100 | 29 | 6.5 | PS | Zimmer/Nexgen Porous Coated | y | peripheral |
| R | 72 | M | 100 | 29 | 4 | CR | Zimmer/Nexgen | y | none |
| S | 54 | M | 98 | 28.6 | 10 | PS | Biomet/Maxim | y | peripheral |
| T | 54 | M | 98 | 28.6 | 12 | CR | Biomet/Maxim | y | peripheral |
A water irrigated diamond wafering blade was used to create (nominally) 8mm × 8mm specimens in cross section containing the cement, interlocked cement-bone interface, and supporting trabecular bone. Micro-CT scans (MicroCT 40, Scanco Medical AG, Brüttisellen, Switzerland) were obtained at 16 μm isotropic resolution (55kV, 144 mA, 200s integration time) in air. The scan sets were imported into MIMICS (Materialise, Leuven, Belgium) with the specimen further trimmed to a cross sectional area of 25 mm2 (Figure 1A). MIMICS image processing was used to create three-dimensional masks for four regions of the construct (Figure 1B) and the associated cement layer.
Figure 1.
A CT cross-section of the interdigitated cement-bone interface (A) taken near the cement border. Arrows indicate cement cavities in the general shape of trabeculae. Image processing is used to identify regions of interdigitated bone, resorbed bone, and supporting bone (B). A 2D longitudinal idealized representation (C) of the cement-bone interface is shown to illustrate the four different mask states. The cross sectional mask areas are determined as a function of axial position (D). Integrating the area under curve represents volume of each mask. The 50% position indicates the midpoint of the mask volume as measured from the cement border. A 1mm scale is shown in A.
During cementing of the proximal tibia, doughy PMMA cement is pressed into the trabecular bone bed, where it flows around existing trabecular bone to create a mechanical interlock. The cement cures in place, creating a mold in the shape of the trabeculae. With in vivo service the trabeculae can resorb, leaving cavities in the cement layer, indicating regions where the bone had resorbed. This is analogous to a ‘trace fossil’ where an initial imprint (such as a footprint) is made in a substrate that then solidifies over time, leaving a ‘trace’ of the initial imprint. Examples of the cavities in the cement are shown in Figure 1A. An image processing approach (13) was previously developed and validated to identify the four component regions (Figure 1B); the approach is briefly outlined here.
Bone was identified with a lower threshold of 500 mg/cc HA equivalent. A separate mask was created for the cement and an 8 voxel dilation, followed by an 8 voxel erosion (Mimics ‘close’ operation) was applied to fill the cement cavities with cement. Boolean and region growing operations were then used to identify the components of the interface including the interdigitated bone (inB), resorbed bone (reB), isolated bone (isB), and supporting bone (spB). Interdigitated bone intersected with the ‘closed’ cement space and was contiguous with the supporting bone. The isolated bone was any remaining bone within the ‘closed’ cement space, not contiguous with the supporting or interdigitated bone. The resorbed bone was an estimate of the remaining space in the cement cavities not occupied by bone. Some manual editing was required to identify round cement shrinkage pores in the cement layer; these were excluded from the resorbed bone estimate. The errors associated with predicting the resorbed bone volume fraction (explained below) using this method were previously reported to be 0.02 mm3/mm2 (13).
For each specimen, cross sectional mask areas were determined as a function of axial position. The cement border (CB) was defined as the distal extent of cement into the trabecular bone (Figure 1C). Figure 1C–D illustrates in schematic format how a distribution of component mask areas would map as a function of axial position. The area under each curve (Figure 1D) represents the volume of each mask component. A number of measures were developed to describe the axial distribution of the bone masks. Nomenclature and summary descriptions of each are described in Table 2; nomenclature was chosen to complement existing metrics used for trabecular bone (14). The interdigitated (inBV), resorbed (reBV), and isolated (isBV) bone volume were calculated as the mask volume divided by the cross sectional area of the specimen (25 mm2). The interdigitation depth (ID) was calculated as the axial distance between the 5% and 95% of the inB+reB+isB component masks. The component bone volumes are further presented as bone volume fractions (e.g. inBV fr), which was calculated by dividing the BV parameter by ID. For each mask, the distance from the cement border to the 50% area under the curve (e.g. 50% of inBV) was used to describe the relative position of each mask in the cement layer (e.g. D_inBV, see Figure 1D). This was further normalized to interdigitation depth (ID) to obtain a fractional position of the mask in the cement layer (e.g. D_inBV fr). The supporting bone volume fraction at the cement border (CB spBV/TV) and 2 to 4 mm distal to the cement border (distal spBV/TV) were also determined.
Table 2.
Nomenclature and descriptive statistics for 20 experimental specimens from the cement-bone interface describing parameters for interdigitated, resorbed, isolated, and supporting bone. Specimens had nominal cross sectional areas (CSA) of 25 mm2.
| Parameter | Description | Mean | SD | Median | Min | Max |
|---|---|---|---|---|---|---|
| ID (mm) | Cement-bone Interdigitation Depth | 2.50 | 1.21 | 2.42 | 0.89 | 6.27 |
| inBV (mm3/mm2) | Interdigitated bone volume / CSA | 0.086 | 0.171 | 0.02 | 0.0013 | 0.703 |
| reBV (mm3/mm2) | Resorbed bone volume / CSA | 0.223 | 0.145 | 0.203 | 0.047 | 0.704 |
| isBV (mm3/mm2) | Isolated bone volume / CSA | 0.018 | 0.036 | 0.001 | 0.0 | 0.150 |
| inBV fr (mm3/mm3) | inB Vol / ID | 0.031 | 0.05 | 0.005 | 0.0008 | 0.23 |
| reBV fr (mm3/mm3) | reB Vol / ID | 0.092 | 0.041 | 0.089 | 0.020 | 0.18 |
| isBV fr (mm3/mm3) | isB Vol / ID | 0.0048 | 0.0076 | 0.0006 | 0.0 | 0.024 |
| D_inBV (mm) | Distance to 50% inBV from cement border | 0.80 | 0.83 | 0.49 | 0.0 | 2.88 |
| D_reBV (mm) | Distance to 50% reBV from cement border | 1.13 | 0.54 | 1.02 | 0.56 | 3.06 |
| D_isBV (mm) | Distance to 50% isBV from cement border | 2.41 | 1.28 | 2.24 | 0.50 | 5.60 |
| D_inBV fr (mm/mm) | Relative distance to 50% inBV; D_inBV / ID | 0.30 | 0.25 | 0.22 | 0.0 | 0.78 |
| D_reBV fr (mm/mm) | Relative distance to 50% reBV; D_reBV / ID | 0.49 | 0.14 | 0.53 | 0.18 | 0.69 |
| D_isBV/fr (mm/mm) | Relative distance to 50% isBV; D_isBV / ID | 0.80 | 0.24 | 0.89 | 0.31 | 0.97 |
| CB spBV/TV (mm3/mm3) | Supporting bone BV/TV at cement border (CB) | 0.12 | 0.07 | 0.12 | 0.036 | 0.37 |
| Distal spBV/TV (mm3/mm3) | Supporting bone BV/TV 2–4 mm distal to cement border | 0.070 | 0.036 | 0.067 | 0.017 | 0.18 |
Given the small sample size (n=20) and relative rarity of postmortem retrieved knee replacements, the knee replacements from the 5 bilateral donors were considered to be independent specimens. Descriptive statistics were determined for each outcome measure for the twenty cement-bone specimens. Paired t-tests with correction for multiple comparisons were performed for the bone volume fraction (inBV fr, reBV fr, isBV fr) and relative distance in cement layer (D_inBV fr, D_reBV fr, D_isBV fr) metrics, to determine if the location of the masks was different. To test whether trabecular resorption would increase with time in service and donor age, a linear regression model was performed with inBV, reBV, and isBV as dependent variables and time in service, age, and interdigitation depth as independent variables. Interdigitation depth was included in the regression model to account for the fact that the initial interdigitation depth was variable for the sample population. To test the hypothesis that resorption would initiate at the cement border with the supporting bone bed and extend into the cement layer with time in service, linear regression models were developed with the 50% distance for interdigitated (D_inBV), resorbed (D_reBV) and isolated (D_isBV) bone as dependent variables, and time in service and interdigitation depth as independent variables. Age was not included in this second set of regression models, because it was not a significant variable in the first set of regression models. Finally, to determine if there was condensation of supporting bone at the cement border, a paired t-test was performed to determine if the amount of supporting bone at the cement border was greater than more distal regions.
RESULTS
An example mask area distribution (Figure 2) for a donor with very little trabecular resorption (Donor E) shows that there is more remaining interdigitated bone compared to resorbed bone and that the resorbed bone is located closer to the cement border. There is also more supporting bone near the cement border compared to more distal locations. Note also that in the vicinity of the cement border, there is some overlap between regions of interdigitated, resorbed, and supporting bone. This is because the cement layer does not flow uniformly into the trabecular spaces and is not perfectly orthogonal with the axial direction. The choice of a 5% and 95% of bone in the interdigitated region to define the cement border and interdigitation thickness was chosen to account for these end effects.
Figure 2.
Example of CT mask area fraction as a function of axial position for Donor E. There is more interdigitated bone compared to resorbed bone, and the resorbed bone is located close to the cement border. There was no isolated bone in this sample. There is more supporting bone near the cement border, compared to more distal locations.
Representative examples (Figure 3) of the cement-bone interfaces from postmortem retrievals illustrate that the resorption patterns vary a great deal depending on donor. The specimen from Donor E (61 YO male, 5 years in service) had very little resorbed bone while the specimen from Donor M (86 YO female, 18 years in service) had a substantial amount of resorbed bone with limited interdigitation. Overall, the amount of resorbed bone volume fraction (Table 2) (reBV fr = 0.092 ± 0.041 SD) was greater than the interdigitated bone volume fraction (inBV fr = 0.031 ± 0.05) (p=0.0084). Both reBV fr (p=0.0003) and inBV fr (p=0.0328) were larger than the remaining isolated bone (isBV fr = 0.0048 ±0.0076). The interdigitated bone (D_inBV fr = 0.30 ± 0.25SD) was closer to the cement border than the resorbed bone (D_reBV fr = 0.49 ± 0.14) (p=0.023). The isolated bone was located farther from the cement border (D_isBV fr = 0.80 ± 0.24) when compared to the D_reBV fr (p=0.0016) or D_inBV fr (p=0.0016).
Figure 3.
Representative examples of reconstructed cement-bone interfaces from postmortem retrieved knee replacements. Top row illustrates all mask components. Bottom row shows remaining bone at the time of implant retrieval (without cement or resorbed bone masks). From left to right, there is decreasing interdigitated bone volume and increasing amounts of resorbed bone. The isolated bone, when present, is located deep in the cement layer. The corresponding outcome measures are shown below the figure.
Using a linear regression model (Table 3), the interdigitated bone volume (inBV) decreased with increasing time in service (p=0.0114), was greater for cases with greater interdigitation depth (p=0.0129), and had a non-significant trend of decreasing with increasing age (p=0.071). The amount of resorbed bone (reBV) increased with time in service (p=0.0007) and interdigitation depth (p=0.0012), with a trend for increasing age (p=0.11). The amount of isolated bone (isBV) was also greater with increased time in service (p=0.032) and interdigitation depth (p=0.0006). Overall, the independent variables could explain between 47% and 81% of the variability of the sample population for the dependent variables.
Table 3.
Linear regression models for volume of interdigitated, resorbed, and isolated bone (dependent variables) and the independent variables of time in service (years), initial interdigitation depth of bone into the cement layer (mm), and donor age (years). The overall regression model fit (R2) is shown, as are model estimates and parameter significance levels.
| Independent Variables | |||||
|---|---|---|---|---|---|
|
Dependent Variable |
Overall Regression Model |
Intercept Estimate (p-value) |
Time in Service (TIS) Estimate (p-value) |
Interdigitation Depth (Int Depth) Estimate (p-value) |
Age Estimate (p-value) |
| Interdigitated Bone Volume (inBV) |
R2=0.47 (p=0.015), n=20 | 0.530 (p=0.031) | −0.0207 (p=0.0114) | 0.0813 (p=0.0129) | −0.0056 (p=0.0711) |
| Resorbed Bone Volume (reBV) |
R2=0.81 (p=0.0001), n=20 | −0.271 (p=0.0286) | 0.0154 (p=0.0007) | 0.058 (p=0.0012) | 0.0025 (p = 0.109) |
| Isolated Bone Volume (isBV) | R2=0.73 (p=0.0001), n=20 | −0.0741 (p=0.044) | 0.00257 (p=0.0324) | 0.0188 (p=0.0006) | 0.000246 (p=0.58) |
The relative effects of time in service and donor age can be explored using the regression model parameter estimates. For example, inBV decreased 0.0207 (mm3/mm2) per year in service and 0.0056 (mm3/mm2) per year in age. Assuming an initial interdigitated BV/TV of 0.1, this translates into a loss of interdigitation depth of 0.207mm/yr in service and 0.056mm/yr age of the donor. This shows that the magnitude of the time in service effect is larger than the donor age effect for this set of implants.
The distance to 50% of the interdigitated bone volume (D_inBV) decreased (move closer to cement border) with increasing time in service (p=0.039) and increased (move deeper in cement layer) with greater interdigitation depth (p=0.0018) (Table 4). The distance to 50% of the resorbed bone volume (D_reBV) had a trend to increase with increasing time in service (p=0.093) and increased with greater interdigitation depth (p=0.0007). The distance to 50% of the isolated bone volume (D_isBV) did not depend on either of the independent variables.
Table 4.
Linear regression models for distance from cement border to the median (50%) volume of interdigitated, resorbed, and isolated bone (dependent variables) and the independent variables of time in service (years), initial interdigitation depth of bone into the cement layer (mm). The overall regression model fit (R2) is shown, as are model estimates and parameter significance levels.
| Independent Variables | ||||
|---|---|---|---|---|
|
Dependent Variable |
Overall Regression Model |
Intercept Estimate (p-value) |
Time in Service (TIS) Estimate (p-value) |
Interdigitation Depth (Int Depth) Estimate (p-value) |
| Distance to 50% inBV (D_inBV) |
R2=0.46 (p=0.0056), n=20 | 0.325 (p=0.422) | −0.0769 (p=0.039) | 0.499 (p=0.0018) |
| Distance to 50% reBV (D_reBV) |
R2=0.65 (p=0.0001), n=20 | 0.075 (p=0.732) | 0.032 (p=0.0928) | 0.293 (p=0.0007) |
| Distance to 50% isBV (D_isBV) |
R2=0.74 (p=0.0043), n=11 | −0.208 (p=0.731) | 0.0943 (p=0.251) | 0.536 (p=0.107) |
The amount of supporting bone at the cement border (CB spBV/TV = 0.12 ± 0.07) was greater than the amount of supporting bone distal (2 to 4 mm) to the cement border (distal spBV/TV = 0.07 ± 0.036) (p=0.0005).
DISCUSSION
A series of postmortem retrieved total knee replacements, which were functioning at the time of donor death, were used to explore the pattern of resorption of trabeculae that initially interlock with PMMA cement. Resorption of the trabecular bone below the tibial tray starts at the cement border (at the extent of cement penetration into the bone bed) and progresses into the cement layer. The amount of resorption increases with time in service, and may be more prevalent in older donors, although the age effect was not statistically significant here. Regions of isolated bone are found deep in the cement layer, and these are more likely to be found for cases with large interdigitation depth, and longer time in service. There was more bone at the cement border compared to more distal locations.
This work has several limitations. A single cement-bone specimen was obtained for each donor bone, and specimens for each bone were taken from regions where there was identifiable initial cement-bone interlock. Other regions where there was less initial interlock might have different resorption patterns, but we have noted previously that donor bones with more initial interlock maintain a greater degree of interlock with in vivo service (2). Five donors had bilateral implants and the implants were considered to be independent in the regression models. However, the time in service for each of the bilateral knees was different. A range of donor ages was used in this study (54 to 90 years), and there was a near even distribution between knees from male (n=9) and female (n=11) donors. However, the mean age of male donors was lower (71 years), compared to female donors (83 years). It would be interesting to explore the effect of donor sex on trabecular resorption, particularly given the potential for estrogen deficiency to effect bone loss (15), but this would be best performed if the male and female population spanned the same age ranges, and with a much larger sample size. In the image processing approach used here, there are errors associated with the estimate of resorbed bone volume due to the fact that the cement surface at the completion of surgery does not completely conform to the interlocking trabeculae. Previous work (13; 16) has shown that errors in estimates of resorbed bone volume are 0.02 mm3/mm2; this represents about 10% of the mean predicted resorbed bone volume measured in the current study.
The mechanism (or mechanisms) responsible for the trabecular resorption process in interdigitated cement-bone constructs is not known. Recent work using finite element models of interlocked cement-bone regions that were loaded axially, across the cement-bone interface, indicates that the trabeculae are stress shielded (6; 7; 16) and that the amount of stress shielding increases for bone deeper in the cement layer. The authors (16) also noted that there is more micromotion near the cement border and this micromotion was diminished deeper in the cement layer. If resorption were driven only by stress shielding, one would anticipate that resorption would start deep in the cement layer and move towards the cement border, but this is not consistent with the findings here.
After implantation, very small thin gaps form between the cement and trabeculae, and the cement and bone move relative to each other when loads are applied across the interface. The relative motion causes fluid pumping along the interface, and this could result in shear stresses of sufficient magnitude to result in fluid flow induced lysis (10). Fluid-structure micromechanical models of the trabeculae-cement interface have been used to explore this possibility (8). Fluid shear stresses were found to be highest near the cement border, and decreased further in the cement layer. The shear stresses were estimated to be much higher than what would normally be present on trabecular bone surfaces (17).
Initiation of trabecular resorption near the bone-cement interface may be a consequence of proximity to osteoclastic cells. Trabecular surfaces deep in the cement are initially accessible from the marrow space only through thin interfacial gaps. Osteoclasts (or pre-osteoclasts) would not readily migrate to these deep trabecular surfaces. Interfacial trabecular bone that is either in direct contact with or in close proximity to bone marrow would be most accessible to osteoclasts. Once resorption occurs near the cement border, this would provide a pathway for cell transport deeper in the cement layer. Given the ability of osteocytes to propagate signals over long distances, it remains possible that this interfacial resorption occurs in response to stress shielding of trabecular bone deep in the cement layer. Development of fibrous tissue in the resorbed trabecular spaces could prevent later resorption of trabeculae deep in the cement layer by essentially sealing off access to marrow cells, resulting in residual islands of isolated bone. It is likely a combination of mechanical and biological factors could contribute to the pattern of bone loss seen here.
The amount of supporting bone at the cement border (BV/TV = 0.12±0.07) was of the same magnitude as the sum of interdigitated, resorbed and isolated bone (inBV+reBF+isBV fr = 0.12±0.04) after accounting for the overestimate of resorbed bone (0.008 mm3/mm3). This suggests that for most cases, there is not additional condensation of bone at the cement border. The volume fraction of interdigitated trabecular space (inBV+reBF+isBV fr) in the cement layer at the time of surgery also provides a measure of the initial state of the trabecular bone, prior to in vivo service. The amount of supporting bone distal to the cement border (BV/TV = 0.07±0.04 mm3/mm3) is less than the interdigitated trabecular space volume fraction, suggesting that there is a loss of supporting bone with in vivo service. This is reasonable given the expectation of stress shielding due to placement of the tibial implant (18).
It is important to note that even with substantial trabecular resorption, there can be strut-like support of the cement by the supporting bone (Figure 4A). This is apparently sufficient to maintain stable fixation, even though with loss of trabecular interlock, there is loss of interface strength (5). However, there are cases where the strut-like support is lost, leaving a gap between the cement and bone (Figure 4B). The gap is often filled with organized fibrous tissue, and there is greater condensation of trabeculae adjacent to the interface. This latter scenario would correspond to cases or locations where there are radiolucencies evident on x-ray.
Figure 4.
For a case with extensive loss of interlock, supporting bone can still have points of contact (arrows) that provide a path for load transfer (A). For a case where there is no longer contact between cement and bone, there is fibrous tissue in the space (FT), and condensation of bone (arrows) that no longer resembles trabecular struts (B).
From the results of this study, it is clear that trabecular interlock can be maintained, particularly for younger patients with shorter time in service. If trabecular resorption were driven solely by an acute, short-term process, it would be expected that interlock would be lost within the first few years of service. But this is not the case. It is possible that other systemic factors, such as age related bone loss (19), are needed for the resorption process to occur. Recent clinical studies have shown that patients taking bisphosphonates (20; 21) have a lower revision rate. Bisphosphonates might alter the effects of age related bone loss around the implant, and also possibly mitigate the resorption process at the cement-bone interface. More investigations are needed to understand the role of cell-cell communication, osteoclast access and/or recruitment, mechanobiologic role of osteocytes, and biomechanical signals in the time dependent resorption process. With this additional information, it may be possible to alter or even reverse the resorption process.
Clinical Significance.
Aseptic loosening of joint replacements remains an important clinical problem. This work explores the process and pattern of trabecular bone resorption responsible for loss of interface fixation.
Acknowledgments
Research reported in this publication was supported by the National Institute for Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health under award R01AR42017. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Contributions: Jacklyn Goodheart performed specimen preparation, microCT scanning/image processing, data analysis, and assisted in manuscript preparation. Mark Miller performed the initial specimen preparation, x-ray imaging, data analysis, and assisted in manuscript preparation. Megan Oest contributed to data analysis, interpretation of the results, and writing of manuscript. Kenneth Mann designed experiments, performed statistics, and contributed to manuscript preparation. All authors have read and approved the final submitted manuscript.
References
- 1.Gebert de Uhlenbrock A, Puschel V, Puschel K, et al. Influence of time in-situ and implant type on fixation strength of cemented tibial trays – a post mortem retrieval analysis. Clin Biomech (Bristol, Avon) 2012;27:929–935. doi: 10.1016/j.clinbiomech.2012.06.008. [DOI] [PubMed] [Google Scholar]
- 2.Miller MA, Goodheart JR, Izant TH, et al. Loss of Cement-bone Interlock in Retrieved Tibial Components from Total Knee Arthroplasties. Clin Orthop Relat Res. 2014;472:304–313. doi: 10.1007/s11999-013-3248-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Howard KI, Miller MA, Damron TA, Mann KA. The distribution of implant fixation for femoral components of TKA: a postmortem retrieval study. J Arthroplasty. 2014;29:1863–1870. doi: 10.1016/j.arth.2014.04.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Miller MA, Goodheart JR, Khechen B, et al. Changes in microgaps, micromotion, and trabecular strain from interlocked cement-trabecular bone interfaces in total knee replacements with in vivo service. J Orthop Res. 2016;34:1019–1025. doi: 10.1002/jor.23109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Goodheart JR, Miller MA, Mann KA. In vivo loss of cement-bone interlock reduces fixation strength in total knee arthroplasties. J Orthop Res. 2014;32:1052–1060. doi: 10.1002/jor.22634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Zhang QH, Cossey A, Tong J. Stress shielding in periprosthetic bone following a total knee replacement: Effects of implant material, design and alignment. Med Eng Phys. 2016;38:1481–1488. doi: 10.1016/j.medengphy.2016.09.018. [DOI] [PubMed] [Google Scholar]
- 7.Zhang QH, Cossey A, Tong J. Stress shielding in bone of a bone-cement interface. Med Eng Phys. 2016;38:423–426. doi: 10.1016/j.medengphy.2016.01.009. [DOI] [PubMed] [Google Scholar]
- 8.Mann KA, Miller MA. Fluid-structure interactions in micro-interlocked regions of the cement-bone interface. Comput Methods Biomech Biomed Engin. 2014;17:1809–1820. doi: 10.1080/10255842.2013.767336. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Nam D, Bostrom MP, Fahlgren A. Emerging ideas: Instability-induced periprosthetic osteolysis is not dependent on the fibrous tissue interface. Clin Orthop Relat Res. 2013;471:1758–1762. doi: 10.1007/s11999-013-2896-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Fahlgren A, Bostrom MP, Yang X, et al. Fluid pressure and flow as a cause of bone resorption. Acta Orthop. 2010;81:508–516. doi: 10.3109/17453674.2010.504610. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gough JE, Downes S. Osteoblast cell death on methacrylate polymers involves apoptosis. J Biomed Mater Res. 2001;57:497–505. doi: 10.1002/1097-4636(20011215)57:4<497::aid-jbm1195>3.0.co;2-u. [DOI] [PubMed] [Google Scholar]
- 12.Mjoberg B, Pettersson H, Rosenqvist R, Rydholm A. Bone cement, thermal injury and the radiolucent zone. Acta Orthop Scand. 1984;55:597–600. doi: 10.3109/17453678408992403. [DOI] [PubMed] [Google Scholar]
- 13.Mann KA, Miller MA, Pray CL, et al. A new approach to quantify trabecular resorption adjacent to cemented knee arthroplasty. J Biomech. 2012;45:711–715. doi: 10.1016/j.jbiomech.2011.12.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dempster DW, Compston JE, Drezner MK, et al. Standardized nomenclature, symbols, and units for bone histomorphometry: a 2012 update of the report of the ASBMR Histomorphometry Nomenclature Committee. J Bone Miner Res. 2013;28:2–17. doi: 10.1002/jbmr.1805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Weitzmann MN, Pacifici R. Estrogen deficiency and bone loss: an inflammatory tale. J Clin Invest. 2006;116:1186–1194. doi: 10.1172/JCI28550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Srinivasan P, Miller MA, Verdonschot N, et al. Strain shielding in trabecular bone at the tibial cement-bone interface. J Mech Behav Biomed Mater. 2017;66:181–186. doi: 10.1016/j.jmbbm.2016.11.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Metzger TA, Vaughan TJ, McNamara LM, Niebur GL. Altered architecture and cell populations affect bone marrow mechanobiology in the osteoporotic human femur. Biomech Model Mechanobiol. 2016 doi: 10.1007/s10237-016-0856-4. [DOI] [PubMed] [Google Scholar]
- 18.Chong DY, Hansen UN, van der Venne R, et al. The influence of tibial component fixation techniques on resorption of supporting bone stock after total knee replacement. J Biomech. 2011;44:948–954. doi: 10.1016/j.jbiomech.2010.11.026. [DOI] [PubMed] [Google Scholar]
- 19.Frost HM. On our age-related bone loss: insights from a new paradigm. J Bone Miner Res. 1997;12:1539–1546. doi: 10.1359/jbmr.1997.12.10.1539. [DOI] [PubMed] [Google Scholar]
- 20.Prieto-Alhambra D, Javaid MK, Judge A, et al. Association between bisphosphonate use and implant survival after primary total arthroplasty of the knee or hip: population based retrospective cohort study. BMJ. 2011;343:d7222. doi: 10.1136/bmj.d7222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Prieto-Alhambra D, Lalmohamed A, Abrahamsen B, et al. Oral bisphosphonate use and total knee/hip implant survival: validation of results in an external population-based cohort. Arthritis Rheumatol. 2014;66:3233–3240. doi: 10.1002/art.38789. [DOI] [PubMed] [Google Scholar]