UC Berkeley

Total knee replacements (TKR) have become a successful surgical procedure for addressing end-stage osteoarthritis, with ultra-high molecular weight polyethylene and cobalt chrome alloy (UHMWPE/Co-Cr) serving as the bearing materials of choice for decades. However, more than 10% of TKRs fail and require revision surgery. The predominant challenge with UHMWPE is the particulate debris generated through wear-mediated processes; wear debris from the UHMWPE tibial bearing surface leading to loosening is still the main cause for post-fifth-year revisions. UHMWPE wear in hip arthroplasty has been linked to microstructural evolution at the surface from multidirectional sliding in the hip joint but little is known about how the microstructure responds to clinically relevant sliding conditions in the knee. This is likely because wear tests are typically performed under basic motion parameters with simplified geometry (pin-on-disk tests) while the knee has more complex kinematics: it is neither a ball-and-socket joint nor a simple hinge joint, but has 2D sliding, rolling/slip motion, and rotation. There is also disagreement over how to best quantify cross-shear and how to model how much wear it will cause. A custom multidirectional tribo-system was used to investigate the individual and combined effects of the different motions in TKR: 2D sliding, rolling, and rotation, for a total of eight separate kinematic conditions. The trends in wear rates and wear factors for these different motions were compared with many different definitions for magnitudes and ratios of cross-shear. Additionally, the wear surfaces were examined for wear mechanism and the microstructural changes in lamellae orientation for the different motions were analyzed.

To mimic the tribological conditions of a condyle in a TKR, polished Co-Cr spheres were articulated against flat, smooth UHMWPE disks with physiologically relevant loading, speed, and lubrication conditions. The motion parameters were selected based on the lowest and highest reasonable amounts of cross-shear that each motion type would generate during a realistic gait cycle: a reciprocating line or a "figure 8" with a 15° crossing angle, no rolling or a 0.4 slide-to-roll ratio, and no rotation or a 1°/mm of rotation. The amount of wear was measured with optical profilometry of the cross-section at the middle of the wear scar, after allowing for a resting period for the material creep to recover. To calculate the amount of cross-shear at this cross-section, the sliding interface was simulated with a computer Matlab model. Multiple definitions of cross-shear were used, including the traditional, cycle-based approaches and newer, memory-based approaches. The wear surfaces were examined using optical microscopy and scanning electron microscopy (SEM). The lamellar microstructure at the wear surface and below the wear surface was examined using an oxidizing etch to remove the amorphous phase. The remaining lamellae were imaged using SEM and their orientations and alignment was quantified using an image analysis program.

Wear factors were between 0.3 and 8.7 µm^2/(Nm/mm), significantly increasing with motion complexity and cross-shear, with the "figure 8" sliding path having the greatest effect. Volumetric wear rates correlated linearly with the total amount of cross-shear, while wear factors correlated linearly with ratios of cross-shear. The best predictors of the wear factor were the normalized crossing intensity and normalized, memory-based cross-shear ratios, with R2 values of 0.98 and 0.97, although many of the cross-shear ratios also had a good linear fits with wear factor. The kinematic parameters in this experiment did not differentiate the various cross-shear parameters enough to conclusively determine which are most appropriate, although some have a stronger theoretical foundation than others. SEM analysis of the wear scar surface revealed slight scratching and instances of rippling and surface cracking perpendicular to the primary sliding directions. These are consistent with abrasive wear, plastic flow and adhesive wear, and fatigue wear mechanisms reported in other in vitro and in vivo wear studies. The orientations of the lamellae at the wear surfaces were not discernibly different from the lamellae of an unworn section of the disk surface. Similarly, the near-surface regions of the disk cross-section were not discernibly different from the subsurface regions. Previous studies have demonstrated orientation of the microstructure during wear using transmission electron microscopy, X-ray scattering, and Fourier transform infrared spectroscopy techniques, and such methods may be necessary for texture characterization.

These results demonstrate that knee kinematics have a significant effect on the cross-shear and wear of UHMWPE and should not be neglected when designing TKR. A better theoretical understanding of how kinematics contribute to wear can lead to better UHMWPE formulations, improved computer simulations of wear, and optimized TKR designs with longer life-spans.