Abstract
INTRODUCTION:
Aseptic loosening continues to be a short and long-term complication for patients with cemented knee replacements. Changes in cemented total knee replacement (TKR) fixation have been limited to assessment of radiographic changes at the implant-bone interface and quantification of component migration. The goal of this study was to determine the interlock morphology between cement and trabecular bone using en bloc postmortem retrieved TKR. Note that these retrievals were not obtained from revision surgery for a loose implant, but rather after death. Thus the implants can be considered successful for the lifetime of the patient. We hypothesized that constructs with greater time in service have less interlock between cement and bone and constructs with more estimated initial interlock sustain more interlock with in-vivo service.
METHODS:
Twelve retrieved tibial components and two lab-prepared constructs with time in service from 0 to 20 years were sectioned in the transverse plane in 10 mm increments, imaged at high resolution, and the current contact fraction (curCF), estimated initial interdigation depth (inID), current interdigitation depth (curID), and loss of interdigitation depth (lossID) were quantified at the cement-bone interface. Contact fraction was determined using a stereology method using random ray projections (Figure 1A). The trabecular shape cast into the cement layer upon curing was used to document the initial penetration of bone into the cement. A line tracing algorithm was used to determine interdigiation depth (Figure 1B).
RESULTS:
There was no biological resorption for the lab prepared constructs (Figure 2A-B). In contrast, a postmortem retrieval (5 years in service) illustrates trabecular resorption with some areas of bone remaining in the cement layer (Figure 2C-D). The curCF (r2 = 0.54, p = 0.0027) and curID (r2 = 0.33, p = 0.033) were lower for constructs with longer time in service (Figure 3). The lossID had a strong positive relationship with time in service (r2 = 0.74, p < 0.0001). Using a two-parameter regression model, constructs with more inID had higher curID (p = 0.011), but constructs with more time in service also had lower curID (p = 0.008).
DISCUSSION:
The cement-trabecular bone interlock that is obtained initially appears to diminish with in-vivo service by resorption of the trabeculae within the cement interlock region. The resulting fixation for longer term constructs (>10 years) was limited, with cement-bone contact fractions of ∼7% and interdigitation depths of ∼0.4 mm. This appears to be sufficient to support load transfer between the metal tray and proximal tibia, as these retrievals were from radiographically well fixed components. The work here supports the surgical concept of obtaining sufficient initial cement interlock (∼3 mm), with the acknowledgement that there will be loss of interlock with in-vivo service. It is possible that loss of interlock contributes to the loosening process of cemented TKR. Further work is needed to understand how loss of interlock affects interface strength.
