Prosthetic Impingement in Total Hip Arthroplasty—The Trigger for Adverse Wear

Development of total hip arthroplasty (THA) now spans more than 5 decades encompassing combinations of metal-on-metal (MOM), ceramic-on-metal (COM), metal-on-plastic (MOP), ceramic-on-plastic (COM), and ceram-ic-on-ceramic (COC). In every arena of extensive technical development, there exists a data set that when viewed in isolation seemed of little import, but when assembled in-toto may produce a generational shift in perception. Our review focused on two such THA events. Firstly, COC retrieval studies (1999-2001) noted habitual wear patterns on heads and peripheral wear stripes, along with femoral-neck impingement, and ceramic surfaces stained gray by metal debris. These COC data indicated THA risks included, 1) cup edge-loading (E/L) on heads producing “stripe wear”, 2) component impingement releasing metal particles resulting in 3) tissues contaminated by metal debris. A corresponding MOM impingement-debris mechanism was only perceived by Howie (2005) in a McKee-Farrar retrieval study. Our an-ticipation at LLUMC was that MOM retrievals would provide superior wear details to those seen on COC retrievals. We noted stripe wear in the polar zone of CoCr heads and basal stripes in the non-wear areas. The basal-polar stripe combinations were found in all MOM retrievals. Basal-polar neck-E/L, inferior cup-E/L, superior cup-E/l and head-E/L, and ingress of Ti64 particles as a contaminating-roughness effect. Individual MOM cases referred to LLUMC demonstrated dramatic evidence of neck notching. At one end of the debris spectrum, a Ti64-notch model predicting a 6 mm 3 annual wear-rate represented the release of 5700 particles of 126 um-size (ap-proximating daily release of 16 particles). At the other end of the spectrum, if metal particles were crushed between MOM surfaces to the equivalent nanometer size found in tissues, our notch model represented approximately 22-trillion Ti64 particles annually deposited in tissues. The anatomical THA models represented in LPUH videos demonstrated that even 1-degree of head subluxation from a rigid cup created a cup “lift-off” scenario (CLO) that would open a gap of 250 - 400 microns between femoral head and cup. This would void all lubrication potential and focus the total hip-joint force along the beveled cup rim, i.e. stripe wear. It is therefore interesting that MOM im-pingement/debris predictions by Howie et al. have not been confirmed until now or discussed in contemporary literature. Therefore, this review of 50 years of THA data demonstrated that hip impingement was always the trigger for adverse wear and that metal-backed cups represent the potential for release of metal debris at extremes of functional standing and sitting postures.


Introduction
Considerable endeavor has gone into understanding wear-related risks in total hip arthroplasty (THA). Nevertheless, impingement of artificial hip components as a potential failure mechanism has received scant attention. The purpose of this paper is to assemble information that will demonstrate; 1) prosthetic hip impingement is commonplace in THA patients, 2) "stripe" wear is a hallmark indicator for impingement, 3) "prosthetic" impingement risks damage to metallic neck and head, 4) edge-loading during impingement releases metal particles, and 5) retrieval evidence will demonstrate that large metal particles contaminate all arthroplasty types particularly those incorporating metal-backed acetabular cups.
Hip joints may impinge at many locations in functional activities, depending on positioning of spine, pelvis and limbs. We shall assess hip-impingement risks by assembling information from COC, MOM, and MOP retrieval studies. This review begins with ceramic bearings used in THA. Ceramic heads are particularly suited to visualizing contamination by metal debris and also demonstrate a novel form of surface damage termed "stripe" wear [1] [2] [3] [4]. Compilation of evidence will show that stripe wear is a hallmark descriptor for THA im- Figure 2. Autophor COC design, (a) cup-on-neck impingement, (b) head retrieval [12], (c) CIA inclination [58], and SIA [23], (d) ROM algorithm. These ceramic studies confirmed 1) stripe wear on heads (head-E/L), 2) cup-E/L, 3) impingement damage on femoral necks (neck-E/L), and 4) metallic staining as confirmation of circulating metal debris. It was noted that wear stripes on retrievals of contemporary THA appeared identical to the early reports, indicating that stripes were a normal occurrence in COC [16] [17].
Stripe-wear patterns can be readily visualized with reference to THA models and video simulations from La Pitié-Salpêtrière University Hospital (LPUH) that accompany this report (Appendix). Hip range-of-motion (ROM) is blocked when the cup impinges against the femoral neck ( Figure 2  Video-2 depicts "functional-sitting" posture showing the same two segments and pelvic views as video-1. In hip flexion, the neck impinges on the anterosuperior cup rim creating edge-loading on its anterior facet. Note the red stripe representing cup-rim profile corresponding to posteroinferior edge-loading on the head. This stripe corresponds to the red retroverted stripe depicted in the 36mm THA impingement model (Figure 3(d)).

Contemporary Ceramic-on-Ceramic (1990-Present)
Historically, hip impingement was perceived as a risk with ceramic liners, in some cases leading to rim chipping and fracture [18] [19] [20]. As a result, met- anteversion. Revision was planned when radiographs revealed her femoral-neck was notched [21]. Revision surgery at 3 years showed, 1) black staining of periarticular tissues, 2) femoral neck with twin notches, 3) black stained ceramic head, and 4) posteriorly notched Ti64 cup (Figure 4(a)). The neck notches could have been formed in two scenarios, i.e. combined liner and shell impingement creating a "double" notch ( Figure 4(b)) or "twin" notches following 20˚ of head subluxation (Figure 4(c)). Most likely, the posterior rim acted as fulcrum for the neck, enabling the head's anterior subluxation [21]. The revision surgeon trimmed the posterior Ti64 rim with a diamond burr and follow-up at 2-years showed hip noises had been eliminated. It was notable that our patient felt no discomfort or ROM limitation despite this 3-year experience with a squeaking THA. The learning experience was the degree of metallic damage created during this short follow-up.

Ceramic-on-Ceramic Summary
Despite frequent loosening problems in early COC designs, the consensus was that the rims of ceramic cups produced stripe wear on heads (head-E/L). Cups showed rim-E/L anterosuperiorly and posteroinferiorly and sometimes circumferentially ( Figure 1). The large SNHKS study demonstrated stripe wear in 83% of cases, stripes formed in hip flexion being the most common. Overall, there was a consensus that damage commonly found in COC retrievals included head stripes, cup-rim wear, neck impingement, and metallic-stained surfaces [25].
Optimization continues in contemporary THA design but does not eliminate impingement risks. Metal transfer continues to be reported, an unequivocal sign of metallic impingement [26]- [35].
Our learning experience from the 36 mm COC model and LPUH videos was that stripe inclinations were related to femoral-neck widths ( Figure 2, Figure 3).

Pioneering Metal-on-Metal (1965-1975)
A variety of metal-on-metal (MOM) designs emerged during the 1960 era [36].   dimension Z). It helped that the centroidal axis of the MWZ was centered adjacent to component midline and slightly superior to the polar axis ( Figure 6(b)).
The cup MWZ area was larger, corresponding to MWA ratio of 79%. With cup positioned at 30˚ inclination, the MWZ centroidal axis (C) corresponded to  dislocated 7 times over the two years leading to revision [56]. CT-imaging revealed an acetabular cup with steep inclination (65˚) and considerable retroversion (15˚). This case demonstrated main-wear zones, stripe-wear, and large areas of Ti64 contamination. MWZ patterns on head and cup were well polished with average roughness Ra < 25 nm. The differences between this case and our MKF retrieval raised the question, how much bearing surface does the patient habitually use, and how does that vary with THA diameter? Review of the literature and LLUMC data provided six hypotheses: 1) Head wear patterns (MWZ) are circular to mildly elliptical in polar region 2) Head MWA-ratios range up to 55% 3) Narrowest MWZ margin indicates superolateral head position in-vivo 4) Centroidal-axis of head MWZ lies adjacent to stem centerline and superior to polar-axis 5) Inclination of MWZ centroidal axis corresponds to resultant hip-force (R) in-vivo 6) Polar head stripes represent edge-loading by the cup-rim LLUMC received a contemporary THA design with still-fused 50 mm head. This provided the opportunity to validate MWZ methodology on large-diameter MOM. This female patient had a steeply-inclined cup [57]. Her painful left hip emitted creaking and crepitus sensations and was revised at 32 months. The narrow MWZ-margin was identified (Figure 8(a)) and photographed to show the superolateral wear pattern. The prosthesis was then rotated in 90˚ increments to record three more views. Thus, the superior head margin (Z) and inferior margin (M) appeared in two views each. The MWZ centroid was located adjacent to the stem midline and approximately midway between polar (P) and superior (S) axes (Figure 8(a)). Reverse-engineering of MWZ onto patient radiographs illustrated the likely in-vivo position (Figure 8(b)). A satisfactory alignment of MWZ centroidal axis with 15˚ line-of-action of (hypothetical) resultant hip-force (R) [37] [38] was taken as appropriate validation.

Adverse Wear with Metal-on-Metal THA and RA
CoCr retrievals were studied visually under stereo-lens magnification for evidence of stripe wear but these were found difficult to photograph and analyze.
Stripes had to be sketched by hand onto our MWZ-charts using colors to denote basal, equatorial, and polar sites. By definition, polar stripes occurred in the head's main wear zone (MWZ) and basal stripes in the non-wear zone (NWZ).
Basal and polar stripes were found in all MOM retrievals [58]. Basal-polar stripe combinations appeared at simulated prosthetic impingements ( Figure 10). Multiple stripe combinations were observed in some retrievals ( Figure 10, Figure   11(b)) [59] [60]. In contrast to basal-polar combinations, equatorial stripes varied considerably, occasionally following the cup-rim profile for considerable lengths (Figure 11     The "twin" neck-notches observed in some THA retrievals (Figure 4(a)) represented a "subluxation" wear mechanism that does not appear to have been discussed in the literature (Figure 4(c)). The 3D-video LPUH models provide the opportunity to evaluate this "subluxation" hypothesis. Video-3 in functional-standing depicts the femoral head subluxing from the cup during the posterior-impingement maneuver. The appearance of a 2 nd black stripe at a steeper inclination depicts head-E/L during this maneuver. Video-4 in functional-sitting depicts head subluxation following the anterior impingement. The 2 nd red stripe formed at a steeper inclination brings it closer to the head's polar axis. Head stripes crossing within their polar circle (Figure 3(c), Figure 3(d)) were therefore witness to subluxation of the femoral head.
Approximately 15 years ago, Howie et al. [61] published a landmark study identifying 100 μm wide scratches in 20 MKF retrievals. LLUMC termed such scratches "microgrooves" [58] to differentiate them from prior descriptions of "fine CoCr scratches" (0.1 -10 μm quoted range) [62] [63] [64] [65]. LLUMC utilized white-light interferometry (WLI) and scanning electron microscopy (SEM) to characterize microgrooves. The "stripe" damage could be represented by a large microgroove or by arrays of parallel microgrooves particularly when Ti64 contamination was present. Microgrooves varied in width from 40 to 400 μm with 100 μm being typical. Large pits and gouges were found in association with microgrooves, indicative of either impacting debris or sub-surface loss due to fracture [66]. Microgrooves were most conspicuous in the inferior head margins (NWZ). Basal microgrooves varied 3 -20 μm deep with jagged lips of equal height to valleys ( Figure 12). The longitudinal striations in larger microgrooves along with shallow entry and exit termini indicated that these were created by metal particles plowing across CoCr surfaces, i.e. classic 3 rd -body wear mechanism. Basal microgrooves were frequently found contaminated with Ti64 alloy in the non-wear zone. Long smears of Ti64 transfer varied 40 -160 μm in width and quite commonly presented as twin tracks up to 300 μm wide or more ( Figure 12(a)). These revealed Ti64 "islands" with 1 to 5 µm peaks above the CoCr surface ( Figure 12(c), Figure 12(d)). In contrast, polar microgrooves (MWZ) were not so well defined, sometimes resembling sawtooth patterns with 1 -2 μm depth ( Figure 13). With average roughness index 200 -300 nm, these stripes were much rougher than adjacent polished MWZ surfaces (5 -20 nm).
Clearly RSA hips with large femoral necks will impinge more readily than THA [67]. The main difference would be that, in the absence of a prosthetic femoral-neck, there would be no risk of "prosthetic" impingement. This raised the question, would RSA femoral components show similar pits, microgrooves, and stripe wear as THA? We analyzed 12 each THA and RSA retrievals that had adequate clinical information and could be matched by vendor and diameter [68]. RA bearings revealed surface pitting, sometimes singly, sometimes grouped, and frequently in linear formations. The pits were typically 150 -160 µm wide, 5 -15 µm deep, frequently found adjacent to microgrooves, and present in both femoral and acetabular components. In other words, RSA wear damage appeared

Summary of Metal-on-Metal Wear Patterns
The definition of functional wear zones (MWZ) in modular heads represented a critical first step in our analysis. The half-angle subtending the typical MWZ by definition was 60˚ (MWA-ratio = 50%). This was sketched on retrieval photographs to separate main-wear from non-wear regions ( Figure 15). With cups

Impingement Evidence in Metal-on-Metal THA
Basal-polar stripe combinations represented indirect evidence of prosthetic impingement ( Figure 10, Figure 11, Figure 15). Femoral-neck proof was lacking because only 2 femoral stems were received in our study of 45 large-diameter MOM retrievals [58]. We note anecdotally that anodized Ti64 femoral necks frequently show loss of color, an indication of very mild wear by the cup rim ( Figure 16(a), Figure 16(e)). While such rings represent unequivocal evidence of cup impingement, these were too shallow to be called notches so we termed these "circumferential blemishes". LPUH loaned LLUMC a set of ten Metasul and as "blemishes" on 5 others [83]. One Metasul stem had three notches, one superior and two posterior. The latter were the "twin" notches similar to our COC retrieval case (Figure 4(a)). The Metasul model demonstrated cup-inclination CIA-angle of 32˚ on the proximal notch ( Figure 17(a)). The head needed to sublux a further 20˚, enabling the cup rim to impinge more distally creating the 2 nd notch (CIA = −8˚) (Figure 17(b)). It was also apparent in these models that only the cup rim would remain in contact with the femoral neck and head ( Figure 17(b)). We modelled this concept of cup "lift-off" (CLO) using our prior retrieval experience (Figure 4). For simplicity, the beveled cup rim was posi-     years showed periarticular tissues stained black [84]. The revised CoCr head showed Ti64 smears 5 µm thick. The femoral neck had two well-defined notches typical of prosthetic impingement (Figure 18(a)). The notches were not the "twins" that denoted head subluxation (Figure 4(a), Figure 4(c)). In this case, the contours of the Ultamet liner and Pinnacle shell exactly matched the double-notched Ti64 neck. This retrieval became our model for predicting a wear spectrum in neck-notches. Interestingly, except for the shallow rim indent, the Ultima cup showed little damage.
It was notable that polished surfaces in Ti64 notches ( Figure 16(d), Figure   18(a)) resembled "precision machining". There was seldom a suggestion of plastic deformation denoting metal components colliding forcefully as anticipated by McHugh et al. [69]. Notch wear is characterized here (Figure 18 (Figure 17). The actual incidence cannot be predicted but impingement/subluxation may happen regularly, for example while doing yoga, tennis, dancing, golfing, riding horses, power walking, etc.

Summary of MOM Wear Patterns in THA and RA
The COC consensus was that head "stripes" were created by ceramic cup-E/L.  ferentiate such large scratches that to our knowledge were not discussed either before or after this Australian report [58]. The importance of the microgroove was the insight that this provided to abrasive wear mechanisms in MOM bearings.
Note that neck-E/L is typically referred to as a "notch" and head-E/L as a "stripe".
From our point of view, "stripes" and "notches" represent a 2-body wear mechanism that edge-loading by a rigid cup produced over millions of wear cycles.
We therefore proceed with the following observations and hypotheses,

Impingement Evidence in Metal-on-Plastic Retrievals
Our wear hypothesis stated above is that MOM and COC bearings readily crush large metal particles in vivo [86]. Therefore, metal-on-plastic (MOP) retrievals should have retained some evidence showing such metal particles. This we shall demonstrate using an assemblage of MOP reports (Table 1).
MOP designs in 1970 and 1980 era included femoral-head materials such as ceramic, CoCr, stainless steel, and titanium alloy [97] [98] [99] [100]. Following the cemented PE-cups [101], there was a move to non-cemented cups that used metal backings. For brevity, the term CEM-cups will refer to PE liners used with cement (no metal shell) and NC-cups to those using PE-liners with metal-backings. Ti64 femoral heads were also popular in the 1980 era, initially used successfully with CEM-cups. However, when replaced by NC-cups, THA revision rates increased resulting in the Ti64 femoral-heads being abandoned [97] [102]. In a MOP retrieval study of CoCr heads, 3 rd -body wear was visible in 89% of MOP retrievals [103]. SEM imaging described 0.1 -5 µm scratches with jagged lips as typical. The authors concluded that this was 3 rd -body wear by metal particles and was more frequent in cases with NC-cups (Table 1).
MOP impingement denoted by deformed PE-liner rims has an incidence approaching 75% of retrievals [85] [104]. In Ohio State University (MOSU) study sampling of 194 retrievals, 93% of the particles embedded in the polyethylene were found to be metallic (Table 1). MOP retrievals representing impingement and dislocation cases have also shown Ti64 contamination. One described a case with Ti64 layers up to 4 μm thick on the CoCr head [35]. Wear analyses from MOP cases revised at MOSU documented large areas of Ti64 contamination (Figure 19(a)).  (Figure 15) by superimposing MWZ-template showing likely wear zones (Figure 19(b)). In the depicted MOSU example (Figure 20), the metal transfer did not resemble basal stripes seen in MOM (Figures 10-12). Nevertheless, linear smears and Ti64-coated basal microgrooves have also been identified in MOM retrievals.
The roughness on Ti64 "islands" ranged 1 -5 um high (Figure 12(d)). MOP data suggested that during daily activities, circulating Ti64 particles were compressed between the PE-liners and CoCr heads (Figure 4 in MOSU report) [105] and thus (a) coated CoCr heads and (b) likely embedded in PE surfaces [85]. being the greatest risk. It is to be noted that THA with multiple bearings also share impingement risks ( Figure 20).

Assemblage of Impingement Evidence (COC MOM, MOP)
The initial focus of MOM studies (1965)(1966)(1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975) was on polar versus equatorial wear-patterns. Peripheral stripe-wear and neck impingement also received some mention. COC clinical studies had their debut circa 1970-1973 in France and Germany. The consensus in these early studies was that "stripe" wear on ceramic heads represented edge-loading by cup rims. However, these data frequently represented loose components and this somewhat clouded interpretation.
Gray-stained alumina surfaces demonstrated that metal particles had been circulating. Nevertheless, studies of contemporary THA designs later confirmed that stripe wear represented a typical COC wear mechanism. In hindsight, two landmark McKee-Farrar (MKF) studies predicted the future for 2nd generation MOM results. An early MKF report described metallosis and pseudotumors in seven retrievals [107] and this was confirmed recently in long-term studies [43]. A report on 24 MKF retrievals attributed 3 rd -body CoCr wear to large CoCr particles being released at impingement [61] (Table 2). This result was also confirmed recently [58]. Key wear patterns represented polar head wear combined with basal-polar head stripes and pertinent evidence of single, "twin", and "double" notches on femoral necks. Without these key observations, surface pitting, scratching and Ti64 transfer could simply have been written off as 1) surgical damage, 2) dislocation damage, and/or 3) loose beads. We now add that head "stripes" and femoral neck "notches" represent precisely-sited wear mechanisms that could only be replicated by "prosthetic" impingement. The 3D anatomical simulations of impingement in LPUH videos brought awareness of stripe formations and the cup "lift-off" mechanism in functional-standing and sitting postures. This new CLO-concept implies that as the femoral head subluxes from the cup, there will be two dramatic changes, 1) sudden loss of lubrication, and 2) cup rim transmits total hip-joint force onto a narrow strip of head surface. Even one degree of head subluxation from a rigid cup enables 200 microns surface gapping.
The well-polished surfaces of femoral notches represented a wear mechanism functioning over "millions" of load cycles. Estimated metal loss due to neck-notching (Ti64 neck, 3.5-year revision) presented a wear-rate approaching 6 mm 3 /year. Such an annual dose of Ti64 would represent 5700 particles of 126 um-size, a daily release of only 16 micron-size particles. The wear spectrum of neck-notching is unknown but this represents a clinically significant wear mechanism ( Table 2) that has not been discussed in MOM literature (2-body wear).The counterpoint to femoral-neck notching was the formation of stripe wear on femoral heads. Descriptions of abrasive wear in hard CoCr alloy surfaces frequently ascribe such damage to release of surface carbides. As side by side comparisons indicate here, the scale of surface carbides (Figure 21(a): circled < 5 μm) is dwarfed by the typical microgroove of width 100 μm. The jagged lips and longitudinal striations illustrate the power in such 3 rd body wear by metal particles (Figure 21(b)). It can be appreciated that the metal particles traversing this surface had to be at least 100 -200 μm wide.  The obvious circuit-breaker in our foundational impingement hypothesis was that prior revision and simulator studies described CoCr debris as minute, approaching 30 -80 nanometers [63] [87]. There are three pieces of evidence that can explain this enigma. Firstly, most THA patients remain completely unaware of hip impingement and subluxation. This has been termed "repetitive sub-clinical subluxation" (RSS) [59]. We also found quite remarkable that considerable implant damage could materialize with quite short-term follow-ups ( Figure 4, Figure 18, Figure 19). The second confirmation was found in MOP retrieval studies ( Table 2) showing 1) debris embedded in PE liners was mostly metallic of average size 126 µm size [85], and 2) Ti64 transfer onto CoCr heads could be 1 -36 μm thick [35] [105]. The 3 rd piece of evidence was provided by LLUMC simulator studies crushing large CoCr and Ti64 particles in 10-second MOM tests [86]. As follow-up, our second hypothesis introduces the wear mechanism of cup lift-off (Figure 17: CLO). Just 1˚ of lift-off will create 250 to 400 μm surface gapping. Not only does this void all lubrication but it also will trigger adverse stripe wear, i.e. total hip-force is now transferred into the segment of cup-rim in contact with the head.  We also acknowledge financial support to our retrieval studies from 1) Arthroplasty for Arthritis Charity (a4ach.org) 2) Dept of Orthopaedics, La Pitié-Salpêtrière University Hospital, Paris