Clinical biomechanics of wear in total hip arthroplasty

Abstract

Complementary clinical and laboratory studies were performed to identify variables associated with polyethylene wear following total hip replacement, and to elucidate the mechanisms responsible for accelerated wear in the total hip arthroplasty construct. Observational cohort studies were performed using a prospective clinical database of more than 4000 consecutive primary total hip arthroplasties performed by a single surgeon, to identify wear-related variables. These variables included head size, acetabular/femoral component impingement, and third body debris. Novel digital edge detection techniques were developed and employed to accurately measure wear, and to determine the relationships of head size and third body debris to acceleration of wear. A novel sliding-distance-coupled finite element model was formulated and employed to examine the mechanisms responsible for wear. The long-term cohort studies demonstrated smaller head sizes to be associated with less wear. Third body debris generated from cable fretting was associated with an increase in wear, osteolysis, and acetabular loosening, especially with larger head sizes. The sliding-distance-coupled finite element model replicated the wear rates occurring in vitro and in vivo, demonstrating the importance of sliding distance on polyethylene wear following total hip arthroplasty. It also demonstrated substantial increases in wear associated with femoral head scratching from third body debris. Further extension of the finite element formulation demonstrated the potential for acetabular component rim damage from impingement wear, and the enhanced potential for third body ingress to the bearing surface with larger head sizes. Edge detection wear measurement techniques demonstrated that early wear rates were predictive of long-term wear rates. These complementary clinical and laboratory investigations have provided insight into 1) the significance of sliding distance and physiologic loci of motion as contributing factors in minimizing wear, 2) the deleterious effects of third body particulates in accelerating wear, 3) the potential for, and factors related to, impingement wear, and 4) the potential advantages and compromises related to the use of larger head sizes in the bearing surface construct.

Figure 1

Survivorship curve (solid line) and 95% confidence intervals (dashed lines) for Charnley prostheses implanted by the senior author (RCJ) using first generation cementing techniques. The endpoint is aseptic loosening of the acetabular component, confirmed at revision. (Schulte et al., JBJS 1993)

Figure 2

Changes made in prosthesis design or implantation technique over a 26-year period of 4164 hip replacements. Only ten changes were involved, and only two of those changes occurred simultaneously: 22⇒28 mm head size, and Charnley polished flatback⇒Iowa matte finish, in 1981.

A

B

Figure 4

Time-wise variations of contact stress and sliding distance during the articulation cycle in THA. A finite element model (Figure 3a) is used to compute 3-dimensional polyethylene contact stress distributions (contour plots) at serial time points in the gait cycle. Corresponding distributions of bearing surface sliding velocity (vector plots) are determined from recordings of the three-dimensional joint motion patterns. These data are then input to a modified version of the Archard equation. Note that both the contact stress distribution history and the sliding distance history for conventional laboratory wear simulations (left panels) are markedly different from those for human locomotion (right panels), implying very different wear behavior.

Figure 5

Long-term wear behavior was computed using the adaptively-meshed sliding-distance-coupled finite element model. Material removal at late times was based upon extrapolation of per-gait-cycle wear depth distributions (Figure 3b), but with the finite element mesh (and therefore the contact stress distributions) periodically updated to reflect material removal. Patterns of computed long-term wear (left) were consistent with material loss patterns on retrieval cups (middle), and in profile corresponded to unidirectional "test-tube" wear front advance (right).

Figure 6

Site-specific wear coefficients are assigned to model local head roughening effects. A patch of Bezier surface facets, which define the femoral head (left panel), could be assigned an elevated wear coefficient value, associated with roughening. Effects of local roughening (wear coefficient = 1.065 X 10^-7 mm²/N) on computed acetabular wear were simulated for 10⁶ cycles of walking motion (B). Compared to the situation for an undamaged femoral head (A), a 2.13-fold increase in computed volumetric wear was induced. Note also that, for the roughened femoral head, the wear tract becomes less regular than the classic "test tube" pattern.

Figure 7

Finite element analysis of the kinetics of total hip impingement, subluxation and dislocation. The model was driven by triaxial motion sequences recorded from subjects undergoing dislocation-prone activities (e.g., leg crossing, rising from toilet seat). The joint loadings were inferred from a 47-muscle inverse dynamics model of the hip. The model articulates normally until neck impingement on the cup. Computed local contact stresses at the impingement and (later) head egress sites are greatly elevated above those for normal articulation, and substantially exceed the yield stress of UHMWPE (lower left panel), even on the outer cup edge, which corresponds with the outer liner wear damage demonstrated in our dislocation studies and in retrieval studies.

Figure 8

Computational fluid dynamics (CFD) models demonstrate the nonlinear increases in fluid ingress velocities, fluid ingress volume, and concomitant potential third-body debris convection around the subluxating components of increasing head-size total hip constructs.

Figure 9

Application of digital edge detection to measure THA wear radiographically. Search rays are computationally generated at 0.5° increments (here, for clarity, rays are displayed only at 10° increments). The pixel grayscale gradient is calculated at each point along each ray. The points of maximal gradient (denoted by the "ƒ" symbols for the femoral head and the "o" symbols for the cup backing) identify the respective component margins. Ellipses are least-squares best fit to these two sets of points, to determine the apparent penetration of the femoral head into the acetabular component. Doing this at follow-up, and subtracting the corresponding measurement postoperatively, allowed assessment of interval wear between those two time points.,,

Figure 10

Distribution of the separation distances between ellipses on consecutive radiographs (left axis). Point-by-point subtraction of the distribution at the latest follow-up evaluation from the postoperative distribution represents polyethylene wear (right axis). This also allowed precise determination of the direction of maximum linear wear (i.e., peak of the measured wear curve).

Figure 11

Group-average behavior for "bedding-in" (run-in) and long-term penetration into the polyethylene cups, by 22-mm and 28-mm femoral heads.

Figure 12

Effect of the head size on time-dependent accrual of volumetric wear, up to as many as 20 years. Volumetric wear increased in proportion to increases in head size.

Figure 13

Temporal evaluation of radiographically apparent interval wear rates. Data for a single patient cohort are illustrated as a scattergram of individual interval linear-wear rate measurements (197 hips, 1237 archived radiographs taken over a 14-year follow-up period).

Figure 14

The initial clinical "bedding-in" process could be described in terms of a decaying exponential (best-fit), to quantify early-term femoral head behavior. The time point to achieve steady state wear (Wss), at which 95% of the initial transient concluded, was 1.8 years. (This figure has subsequently been rounded to 2 years to simplify description of the "bedding-in" period.39)

Figure 15

Effects of elevated third body debris burden on in vivo wear. Two patient cohorts were compared, both with the same implant: an Iowa femoral component articulating with a cemented 28 mm metal-backed Tibac (Zimmer) cup. One group ("Cable,"197 consecutive patients) had trochanteric fixation with 1.5 mm, 7-strand Co-Cr-W-Ni cable, subsequently found to be fretting prone (top) and its usage discontinued. The other group ("Wire," 157 consecutive patients) had trochanteric fixation with single-strand stainless steel wire. Series wide (bottom) there is a much larger fraction of problem "Cable" patients (10%) with linear wear rates exceeding the clinically problematic rate of 0.2 mm/year.