Bone-on-Bone versus Hardware Impingement in Total Hips: A Biomechanical Study

Abstract

Dislocation remains a serious concern for total hip arthroplasty (THA). Impingement, typically between the implant femoral neck and the acetabular cup, remains the most common dislocation impetus. Wear reductions from recent bearing technology advancements have encouraged introduction of substantially increased femoral head diameters. However, there is some evidence that range of motion with larger head sizes is limited by bone-on-bone, rather than hardware, impingement. While all impingement events are of course undesirable, currently little is known biomechanically if these two impingement modes differ in terms of generation of potentially deleterious stress concentrations or with regard to dislocation resistance. Finite element (FE) analysis was therefore used to parametrically investigate the role of head diameter on the local biomechanics of bone-on-bone versus component-on-component impingement events. Of several dislocation-prone patient motion challenges considered, only squatting consistently resulted in bone-on-bone (as opposed to hardware) impingement. Implant stress concentrations arising from hardware impingement during squatting were greater than those from bony impingement, for all head sizes considered. Additionally, dislocation resistance was substantially greater for instances of bony impingement versus hardware-only impingement. These findings suggest that hardware impingement may still be a/the the predominant mode of impingement even with the use of larger femoral heads, for sub-optimally positioned cups. Additionally, the data indicate that, should impingement occur, impingements between the implant neck and cup are (1) more likely to dislocate, and (2) have a greater propensity for causing damage to the implant compared to impingement events involving bony members.

Introduction

Owing to recent adoption of advanced low-wear bearings, as of 2009, instability had surpassed osteolysis as the leading diagnosis requiring revision surgery in total hip arthroplasty (THA) ¹. The risk factors for dislocation in THA are numerous and varied ^2-10. Many of these factors involve damage to the periarticular soft tissues, or they involve the geometrical design and surgical orientation of the implant. Regardless of the underlying risk factor, impingement is the proximate mechanism for the overwhelming majority of dislocations ¹¹. In this situation, the femur (either native or implant) impinges on bone, on soft tissue, or on the acetabular component of the implant, constituting a fulcrum about which there is tendency for levering-out of the femoral head from the cup. Widely recognized mechanisms to reduce the incidence of impingement include meticulous component positioning and the use of larger diameter femoral heads. Larger femoral head sizes have been statistically associated with improved implant stability in most ^{5, 10, 12, 13} (but not all ¹⁴) clinical studies.

In conventional-sized THAs (22-32mm), impingement typically occurs between the implant femoral neck and acetabular cup. Therefore, the vast majority of experimental studies of THA impingement and range-of-motion have used THA hardware components tested in isolation. Additionally, from the viewpoint of experimental practicality, hardware-only studies are substantially more practical to execute. However, both cadaveric ¹⁵ and mathematical ¹⁶ studies have suggested that bone-on-bone impingement predominates for femoral head diameters greater than about 32mm to 36mm. To date, large-head impingement mechanics studies have been limited by (1) only reporting impingement-free range of motion, and (2) only utilizing simplified joint motions. To clarify the different consequences of bone-on-bone versus hardware impingement as regards instability and/or potential implant damage, a dynamic finite element (FE) model of THA impingement was developed to investigate the role of femoral head size and impingement modality for large-diameter total hips.

Materials & Methods

The study utilized a previously validated ¹⁷ FE model of a widely used contemporary THA implant (Summit stem, 36mm M-spec head, 36mm x 56mm Pinnacle cup, DePuy Orthopaedics, Warsaw, IN) generated from manufacturer-provided engineering CAD files. Implant geometry was pre-processed using TrueGrid (v. 2.3 XYZ Scientific Applications, Inc., Livermore, CA) and Mathcad (v. 14.1, PTC, Needham, MA) software. Seven distinct femoral head diameters were considered (32mm to 44mm, in 2mm increments, Fig. 1). The models were generated by projecting the outer surface of the femoral head mesh onto correspondingly scaled surfaces representing the outer diameter of the implant femoral head. Appropriate mesh densities for each model were determined from prior convergence studies ¹⁸. Bony anatomy of the pelvis and femur was determined by manual segmentation of the Visible Human data set (National Library of Medicine, Bethesda, MD). The bony anatomy was registered to the pelvic reference frame of the FE model. Virtual femoral osteotomy and pelvic reaming were performed using Geomagic Studio (v. 12, Geomagic Inc., Research Triangle Park, NC). The acetabular components were then positioned in a consensus neutral orientation (40° inclination and 15° acetabular anteversion).

Figure 1. The THA FE model consisted of THA hardware and bony anatomy. Seven distinct femoral head diameters were considered. From left to right: 32, 34, 36, 38, 40, 42, and 44mm.

The implants were modeled as linearly elastic metalon- metal, (CoCr, elastic modulus = 210 GPa, Poisson's ratio = 0.3), with 0.029 mm of radial clearance and with a friction coefficient of 0.1 ¹⁹. To economize computational run time, the bony anatomy, cup backing, and distal regions of the femoral component were assigned rigid body definitions. In all cases, the femoral component was positioned in 10° of anteversion.

Two separate FE series were conducted (Fig. 2). The first was undertaken to determine the impingement propensity for various impingement-prone patient motions. The second series investigated differences between hardware and native bony impingement events, for different cup orientations. For the first series, nine candidate impingement challenge motions were considered: pure flexion, as well as six posterior dislocation maneuvers (sit-to-stand from a low chair, sit-to-stand from a normal height chair, seated leg crossing, seated leaning, stooping, and deep squatting) and two anterior dislocation challenges (rolling over in bed, and standing exorotation/pivoting), for a 42mm implant in consensus-neutral cup orientation. These respective impingement challenges were quantified from joint force optimization routines and inverse dynamics analyses conducted for force plate and optoelectronic motion capture trials of THA-aged subjects ⁶. Of these various dislocation challenges, only pure flexion and squatting consistently resulted in bone-on-bone impingement for the neutrally-oriented cup (Fig. 3).

Figure 2. Two separate FE series were conducted. A preliminary series investigated nine distinct impingement-prone maneuvers for a 42mm femoral head with a neutrally positioned cup. Of these candidate challenges, only pure flexion and squatting resulted in bone-on-bone (B-B) impingement. The second series investigated the effect of head size on B-B versus implant-on-implant (I-I) impingement for squatting. Twenty-one distinct squatting FE simulations were performed by investigating seven head diameters times three separate cup orientations and impingement conditions.

Figure 3. Pure flexion (A) resulted in impingement between the bony femoral neck and the rim of the bony acetabulum at approximately 100° of flexion. By contrast, even at 110° of flexion, stooping (B) did not result in impingement. Deep squatting, however, resulted in impingement between the bony femoral neck and the anterior inferior iliac spine after 105° of flexion (C).

After identifying that squatting was the most discriminational motion challenge leading to bone-on-bone impingement, a second (larger) FE series was undertaken (Fig 2). For this second series, the effect of head size on bone impingement was investigated by considering all seven femoral head sizes for the squatting challenge with a neutral (40° abduction) cup orientation, a situation for which for which impingement occurred between the bony members (referred to as “B-B₄₀”). Seven additional FE models were generated with a more horizontal cup orientation (30°), resulting instead in implant neck-on-cup impingement, also for the squatting challenge (referred to as “I-I₃₀”). To then eliminate cup orientation as a confounding variable for stress and resisting moment comparison, seven additional FE models were generated at the same horizontal (30°) cup position, but with contact detection numerically disengaged between the neck and cup. By removing this aspect of contact detection, otherwise present implant neck-on-cup impingement did not occur, as the femoral neck was allowed to artificially penetrate directly into the cup “unscathed”, thus ensuring that any detected contact instead would be only bone-on-bone (“B-B₃₀”).

For these 21 resulting squatting FE simulations, peak implant surface von Mises stress occurring during the impingement-subluxation-edge loading progression was recorded, along with dislocation resisting moment. All simulations were executed using Abaqus/Explicit (v. 6.9, SIMULIA, Providence, RI).

Results

For neutrally-positioned cups, of the nine candidate challenge motions, only pure flexion and deep squatting resulted in impingement, and that impingement was bone-on-bone. None of the other candidate challenges resulted in either form of impingement. Pure flexion resulted in impingement between the bony femoral neck and the anterior acetabular rim, at approximately 100° of flexion. Squatting resulted in bone impingement between the bony femoral neck and a region near the anterior inferior iliac spine, which occurred at about 105° of hip flexion.

For the component-on-component (I-I₃₀) simulations, impingement events during the squatting challenge generated significantly higher impingement-associated edge loading implant stresses than occurred for either of the two bony impingement situations (Fig. 4). Both situations of bony impingement (B-B₄₀ & B-B₃₀) involved generally similar stresses. Edge loading stresses for all three impingement conditions decreased with increased head diameter, the effect being most pronounced for the hardware impingement events. Peak resisting moments developed for the two B-B impingement situations were, on average, 5.5- and 2.6-fold higher than for hardware impingement, and tended to increase with increased head size (Fig 5).

Figure 4. Peak impingement-associated egress site edge loading von Mises (vM) stress for the 21 separate FE simulations. I-I impingement for cups in 30° of inclination (I-I₃₀) demonstrated substantially greater bearing surface stress, compared to B-B impingement for 30° (B-B₃₀) or 40° (B-B₄₀) of cup inclination, for any given value of femoral head diameter. Impingement-associated head edge loading stresses tended to decrease as head size increased.

Figure 5. Peak dislocation resisting moment was greater for instances of B-B impingement than I-I impingement, regardless of femoral head diameter.

Discussion

Impingement of the implant neck on the implant liner (I-I impingement) is commonplace, with corresponding liner rim damage being reported in a majority of explanted conventional-sized THAs ^{13, 14, 20}. However, the recent shift toward advanced low-wear bearing couples has enabled the use of larger diameter femoral components, that theoretically better protect against I-I impingement by allowing for increased angular excursion prior to such contact. It has been previously suggested, however, that the theoretical range of motion in these new larger components actually many not be achieved in practice, due to bone-on-bone impingement first occurring. Using a cadaveric model, Bartz et al. ¹⁵ contended that I-I impingements constituted only 50% of impingements for 28mm femoral heads, and only 30% of impingements for 32mm head diameters, the remainder of impingements being B-B. However, that study was limited in terms of the motion analyzed (only pure hip flexion, ostensibly simulating chair rising), and it involved only a single cup position (neutral). Reports of a similar transition from I-I impingement to B-B impingement with increasing head sizes had been made in an earlier (but even more restrictive) cadaveric study ²¹. Yet an additional cadaveric study ²² maintained that B-B impingement limited hip range of motion for both pure flexion and pure internal rotation, for head sizes greater than 26mm. Cinotti et al. ¹⁶, using a geometric/mathematical model of hip range of motion for 28, 32, 36 and 38mm head sizes, suggested that B-B impingement would be more common than I-I impingement for 36mm and 38mm implants, unlike the case for smaller femoral heads. However, while this latter study involved multiple cup orientations, again, only simplified joint kinematics were considered.

The various purely geometrical issues addressed in those precedent studies unfortunately provide no insight on differences in contact mechanics and construct stability for I-I vs B-B impingement. This knowledge gap motivated the present study, which represents the first investigation of the kinetic (i.e., stress and moment) differences for implant-on-implant vs. bone-on-bone impingement events. While certainly no good can come of any impingement event, some impingements nevertheless are more undesirable than others. The present results show that bone-on-bone impingement events are less prone to result in frank dislocation, and they involve considerably milder impingement-associated edge loading contact stress concentrations, than occurs for implant-implant impingement, other factors being equal. For all but the very largest head diameters considered, the impingement-associated edge loading stresses induced by implant neck-on-liner impingement approached or exceeded the yield strength of cobalt-chrome alloy. Altogether, this suggests that, should impingement occur, events of impingement involving the implant neck and cup are (1) more likely to dislocate, and (2) more prone to result in hardware damage, as compared to bone-on-bone impingements.

It has oftentimes been suggested (based on purely geometrical considerations) that bone-on-bone impingement is the limiting factor for joint range of motion in large-diameter THAs. However, the present investigation provides convincing evidence that hardware-only impingement is physically very possible for various dislocation-prone motions, if the implant components are sub-optimally positioned surgically. Additionally, while B-B impingement was found to occur with pure flexion as well as with deep squatting for neutrally positioned cups, it did not occur for the other motion challenges. This suggests that while B-B impingement may in fact be the predominant limiting factor in certain situations for larger diameter THA, B-B impingement is still a rather unusual occurrence, even for high-risk patient motions.

This investigation has a number of limitations. First, only a single bony geometry was used. While this particular “standard” Visible Human geometry has been very widely used in a host of scientific investigations, it is a reasonable assumption that individual variants in bony geometry could directly influence both the occurrence of bony impingement, and the kinetics of impingement-associated edge loading contact stress generation. However, given the large difference in the intrinsic dislocation resistance and surface stresses between instances of bone-on-bone versus implant-on-implant impingement, it is likely that specific normal-range variants in bony anatomy would have a only a minor/modest effect on impingement dynamics. Secondly, while several impingement-prone kinematic challenges were considered, these obviously still represent only a small subset of possible impingement scenarios. Nevertheless, the selection of motions considered in the present study represents a much more comprehensive sampling than for any other impingement study conducted to date.

In summary, should impingement occur, contact between the bony femur and pelvis is substantially less detrimental than contact between the implant neck and cup. Larger femoral heads, regardless of impingement location, result in less impingement-associated implant edge loading stress and have greater dislocation resistance.