Background
Modifications in bearing surface modularity and stem designs in total hip replacement (THA) were introduced in the past two decades with the goal of reducing the incidence of aseptic loosening and instability [
1,
2]. One of these modifications included the metal-on-metal (MoM) bearing surface, which was combined with a metallic adapter sleeve for large heads in the early 2000s. The rationale for the revival of this bearing surface included a reduction in volumetric wear and osteolysis compared to conventional metal-on-polyethylene bearings (MoP), decreased impingement throughout range of motion, and decreased rates of dislocation [
1]. A second modification to increase modularity in THA was the introduction of the dual-modular neck. This provided surgeons with increased reconstructive options to potentially match each patient’s anatomy and permit the use of a MoP or ceramic-on-polyethylene (CoP) bearing surface [
2].
The unintended consequence of these implant modifications has been an increasing number of new interacting surfaces of different biomaterials, subject to short-term mechanical and no biologic testing before worldwide marketing and use [
3]. This has resulted in increased MoM implant failures due to a distinct type of cellular/tissue reaction, originally reported as aseptic lymphocyte dominated vasculitis-associated lesions (ALVAL), now collectively referred in the literature as adverse local tissue reactions (ALTR) or adverse reaction to metallic debris (ARMD) [
4‐
12]. Previous histological analyses of retrieved periprosthetic tissue have shown evidence of corrosion products, metallic debris generated by abrasion and/or surface fatigue, extensive soft tissue necrosis, combined macrophagic and lymphocytic infiltrate with variable plasmacytic and eosinophilic components, and vascular wall changes [
5,
13‐
19]. A comprehensive review describing features of periprosthetic inflammation to wear debris has been addressed in a recent review article by Gallo et al. [
20]. The constellation of pathologic findings observed in response to MoM implants was encompassed under the acronym ALVAL by Willert et al. to illustrate the unique lymphocytic component and probable vascular changes not seen in other typical modes of THA failure such as osteolysis or infection [
14]. Failure due to ALTR has predominantly been attributed and described for MoM bearing surfaces, but evidence of head-neck and neck-stem corrosion in modular implants has been reported to result in ALTR [
4,
9‐
12,
21‐
23].
We hypothesized that corrosion products, as previously described in the literature, generated at different contact surfaces of THA implants could be the defining factor of ALTR irrespective of the bearing surface, and these differences would be reflected in the histologic and immunohistochemical profiles of retrieval tissue between different implant types [
5,
7,
11]. In order to investigate this hypothesis, we compared the morphologic and immunohistochemical characteristics from retrieved periprosthetic tissue using two separate classes of implants: 1) MoM bearing surface and a metallic adapter sleeve at the head-neck taper junction and 2) conventional bearing surface (metal-on-polyethylene or metal-on-ceramic) with a dual-modular neck with tapers at the head-neck and neck-stem junctions. The purposes of this study were to analyze implant-based similarities and differences in: 1. Periprosthetic tissue structure, organization, corrosion product morphology, and cellular composition by conventional histology; and 2. Cellular composition by immunohistochemistry.
This is the first study to describe the morphologic and immunohistochemical similarities and differences between ALTR associated with different implant classes. Our results demonstrate that implant design can affect corrosion product morphology and the periprosthetic tissue pathology, and indicate that the interaction between implant design and host biology can have important clinical consequences in surveillance and outcomes of hip orthopedic implants in the future.
Methods
Patients
Between April 2012 and June 2013, all patients with the diagnosis of ALTR based on histological analysis were identified retrospectively from the Osteolysis Tissue Database and Repository at the Hospital for Special Surgery. This prospective database collects demographics, selected clinical data, periprosthetic tissue, and biological fluid (serum, synovial fluid) from all consenting patients undergoing revision THA with suspected ALTR. Ethical committee approval was obtained prior to this study (Institutional Review Board, Hospital for Special Surgery, Protocol Number 26085). Two groups of patients were selected, representing the two major implant classes resulting in ALTR: the Dual-Modular Neck group (DMN) had a MoP or CoP bearing surface with a dual-modular neck (cobalt-chromium-molybdenum, CoCrMo) and TMZF (titanium, molybdenum, zirconia, iron) stem (Stryker, Rejuvenate) [N = 55 hips, 54 patients], and the MoM THA group had a MoM bearing surface (CoCrMo) with a metallic adapter sleeve (CoCrMo) at the head-neck junction and titanium stem (Smith & Nephew, Birmingham THA) [N = 18 hips, 14 patients]. All polyethylene used in the DMN implant was second-generation highly cross-linked polyethylene (X3, Stryker Corporation). Exclusion criteria included previous revision arthroplasty, positive intraoperative cultures, and insufficient tissue retrieval for comparative pathologic examination (less than 5 tissue sections and more than 75% tissue necrosis at light microscopy examination on all slides examined). These two implants were selected because they are examples of recently marketed modular implants with a sufficient number of cases in our institution to allow an in-depth morphologic and immunohistochemical analysis. Preoperative serum cobalt and chromium levels were obtained by quantitative inductively coupled plasma-mass spectrometry at the operating surgeons’ discretion (ARUP Laboratories, Salt Lake City, Utah). Acetabular and stem components were recorded for each implant.
Tissue collection and sampling
All patients suspected of having ALTR at our institution undergo magnetic resonance imaging (MRI) with multi-acquisition variable-resonance image combination (MAVRIC) scan to further reduce susceptibility artifact [
24]. Findings suggestive of ALTR include bulky synovitis, extracapsular disease, tendon/intramuscular edema, and capsular avulsion [
24]. Periprosthetic tissue sampling in revision cases for the implants included in the database has been standardized in our institution since September 2011, when the first cases of the two series of patients described in this report were observed. Areas of inflammation were identified preoperatively on MRI, and used as guidance for tissue sampling by the operating surgeon. Samples were taken from multiple regions around the hip joint including the periprosthetic pseudocapsule, bursal synovium, and adjacent skeletal muscle when necessary and labeled accordingly. The use of cautery was minimized to avoid compromising the tissue for histologic and molecular analysis. Additionally, acetabular and femoral bone samples, core biopsies of osteolytic areas, and/or reamings were sent separately to evaluate possible bone marrow involvement when suitable. Separate tissue samples identified by location were sent to the microbiology laboratory to rule out infection and, if sizable, retrieved after culture preparation for further histological analysis.
The project research coordinator (DM, GW) harvested biological samples with presence of the pathologist (GP) to assure consistency among all the surgeons contributing cases to the database. The tissue was retrieved fresh in labeled tissue cups from the sterile area as soon as possible and kept on ice. One pathologist (GP) performed frozen section by sampling of the fresh tissue in order to assess viability and cell composition and when feasible, a representative tissue sample was processed for RNA isolation for future investigations. Remaining tissue samples were provided between two and six sites surrounding the implant. Extensive sampling was performed at macroscopic examination with care to the orientation of the specimens, including necrotic areas and/or friable, loose material. Acetabular reaming was also collected, osteolytic areas were sampled when present, and cancellous bone was also scraped from the femoral stem and/or the acetabular shell when possible. The number of paraffin blocks containing one or two tissue sections processed per case varied from 7 to 14, to minimize sampling error due to necrosis and to ensure valuable representation of the viable tissue. Photographs of each implant and selected gross tissue specimens were taken.
Histologic analysis
All sections were processed and embedded with standard procedures, stained routinely with hematoxylin-eosin, and examined by an experienced musculoskeletal pathologist (GP) to assess the presence of ALTR. A range of 7 – 14 sections were examined per case depending on tissue availability. Cases were scored by one investigator experienced in examining periprosthetic tissue from revision THA (GP), one experienced surgical pathologist (GM), and a third investigator trained for three months on 100 archival hip revision cases with a full spectrum of adverse reactions (BR). Investigators were blinded from clinical patient characteristics. Discrepancies in scoring were resolved by consensus agreement. The ALVAL scoring system proposed by Campbell et al., which was previously used at our institution as correlative index with MRI imaging analysis, was recorded for each case [
16,
25‐
27].
Histological sections were examined for the presence (Y) or absence (N) of synovial lining loss/hyperplasia, partial or full thickness necrosis of the neo-synovial membrane and subsynovial soft tissue, cell exfoliation, vascular wall changes, high endothelial cell venules (HEV), granulomas (sarcoidosis-like with or without central necrosis), and skeletal muscle inflammatory infiltrate (Table
1). Semi-quantitative evaluation was undertaken for grading of the macrophages [
28]. Macrophages were graded on a 0–3 scale (absent, occasional, clusters, diffuse/sheets). Total lymphocytes were graded as interstitial (band-like) and/or perivascular. Perivascular lymphocytes were graded according to average lymphoctic cuff thickness using a Zeiss Axioskop 40 calibrated reticule and scored as described by Natu et al. on a 0–4 scale with absence or presence of germinal centers [
17]. Neutrophils were graded on a 0–2 scale [(absent, occasional, focally numerous (>5 cells x 10 HPF)]. Plasma cells were graded on a 0–2 scale [(absent, occasional, or numerous (>10 cells per HPF)]. Eosinophils were graded on a 0–1 scale (absent or present), stromal cell cellularity was graded on a 1–3 scale (slight, moderate, marked). Results were expressed as the percentage of samples containing the selected feature.
Table 1
Morphologic comparison of synovial structure, cellularity, macrophage content, and bone marrow involvement between the Dual-Modular Neck and the Metal-on-Metal (MoM) total hip arthroplasty (THA) groups
Synovial structure
|
Cases (%)
|
Cases (%)
| |
Synovial layer loss | 96.4 | 100.0 | |
Synovial layer hyperplasia | 78.2 | 77.8 | |
Cell exfoliation | 87.3 | 94.4 | |
Necrosis | 65.5 | 61.1 | |
Vascular wall changes | 18.2 | 16.7 | |
High endothelial cell venules | 14.5 | 16.7 | |
Granulomas | 18.2 | 11.1 | 0.482 |
Cellularity
| | | |
Macrophages | | | 0.016* |
Grade 1 | 16.4 | 5.6 | |
Grade 2 | 30.9 | 5.6 | |
Grade 3 | 50.9 | 88.9 | |
Lymphocytes | | | 0.066# |
Grade 1 | 1.8 | 16.7 | |
Grade 2 | 9.1 | 5.6 | |
Grade 3 | 52.7 | 61.1 | |
Grade 4 | 34.5 | 16.7 | |
Stromal Cells | | | 0.593 |
Grade 1 | 27.3 | 38.9 | |
Grade 2 | 50.9 | 44.4 | |
Grade 3 | 16.4 | 11.1 | |
Neutrophils | 10.9 | 11.1 | |
Plasma cells sparse | 32.7 | 38.9 | |
Plasma cells numerous | 20.0 | 22.2 | |
Eosinophils | 32.7 | 33.3 | |
Macrophage content
| | | |
Polyethylene particles | 1.8 | 0.0 | |
Metallic particles | 1.8 | 33.3 | <0.005* |
Corrosion products | 100 | 100 | |
Intracellular distribution | Sparse | Diffuse | |
Intracellular morphology | Irregular | Globular + Irregular | |
Extracellular corrosion aggregates | 72.7 | 66.7 | 0.662 |
Bone/bone marrow
| | | |
Necrosis | 47.1 | 28.6 | |
Macrophage infiltration | 47.1 | 100.0 | |
Benign lymphocytic aggregates | 35.3 | 28.6 | |
Germinal centers | 17.6 | 0.0 | |
Macrophage content (polyethylene, metal, and ceramic particles) was graded according to the method used for metallic particles by Natu et al. [
17] (Table
1). Presence of intracellular corrosion products was recorded and extracellular aggregates were graded on a 0–1 scale (absent or present). Presence of hemosiderin deposits and/or suture material was recorded.
Bone marrow sections were evaluated for the presence (Y) or absence (N) of necrosis of bone and marrow cellular elements, macrophage infiltration, and benign lymphocytic aggregates with or without presence of germinal centers. Results were expressed as a percentage of patients displaying each morphologic feature.
Immunohistochemistry
Fifteen cases for each of the DMN and MoM THA groups were analyzed by immunohistochemistry. The cases from the larger DMN group were selected to be representative of the spectrum of histological patterns observed as described in the results section. Conventional immunohistochemistry was performed using standard techniques on consecutive sections (GM). Heat-induced antigen retrieval was performed using a microwave oven and 0.01 mol/L of citrate buffer. All samples were processed using a sensitive ‘Bond polymer Refine’ detection system in an automated Bond immunohistochemistry instrument (Vision-Biosystem, Menarini, Florence, Italy). Antibody dilutions and source are shown in Table
2. Commercially available monoclonal antibodies were used and each batch was tested by titration for optimal dilution on both internal and external controls. Macrophage markers were CD68 (all macrophages) and CD163 (M2 macrophages) [
29,
30]. The lymphocytic response was assessed by expression of CD20 for B cells and CD3, CD4, and CD8 for T cells. Expression of T-bet, GATA3, and FOXP3 was used as marker for transcription factors for Th1, Th2, and Treg cells to sub-classify the T cell distribution [
31‐
33]. High endothelial cell venules were identified as CD123 positive cells. Mast cells were identified as CD117 positive cells [
34].
Table 2
Description of antibodies and dilutions utilized for immunohistochemistry
CD3 | SP7 | THERMO SC. | 1:150 |
CD4 | 4B12 | NOVOCASTRA | 1:150 |
CD8 | C8/144B | DAKO | 1:200 |
CD20 | L26 | NOVOCASTRA | 1:100 |
CD68 | PG-M1 | DAKO | 1:50 |
CD123 | 7G3 | BD Phamingen | 1:100 |
CD163 | 10D6 | NOVOCASTRA | 1:200 |
GATA-3 | L50-823 | BD Phamingen | 1:150 |
FOXP3 | 221D/D3 | SEROTEC | 1:200 |
T-bet | 4B10 | SANTA CRUZ | 1:100 |
Granzyme | GrB-7 | MONOSAN | 1:100 |
CD117 | T595 | NOVOCASTRA | 1:10 |
Semiquantitative analysis was performed for evaluation of macrophage, mast cell, and HEV distributions. Evaluation of CD68 and CD163 stained sections were graded as +, ++, and +++ by three investigators (GP, GM, BR) blinded to the clinical data. CD117 staining was assessed from 0–2 [absent, occasional, numerous (>5 forms per HPF)]. Granzyme immunohistochemistry and the presence of CD123 positive HEVs were assessed by the presence (Y) or absence (N) of positive cells.
A quantitative analysis (Bioquant Osteo, Bioquant Image Analysis Corporation, Nashville, TN) was performed on all sections to evaluate lymphocytic distributions in both perivascular and interstitial regions. Two perivascular and two interstitial areas on each slide were randomly selected and evaluated at high power (×400), and lymphocytes with positive stain were counted manually by two investigators blinded to the clinical characteristics (BR, GP). The results were expressed as percentage of positive cells per mm2. The same areas from consecutive sections were chosen for each stain, ensuring consistency in area of evaluation. The ratios between CD20:CD3, CD4:CD8, and GATA3:T-bet on the same sections were then calculated. The CD20:CD3, CD4:CD8, and GATA3:T-bet were described as a > 2:1, 1:1, or > 1:2 ratio.
A comparison control group of periprosthetic tissue was used for immunohistochemistry. For the control group (N = 17), average age was 63.5 years (standard deviation 14.0) and 71% were females. These included three cases (N = 3) of osteoarthritis with variable amount of lymphoplasmacytic infiltrate without clinical diagnosis of rheumatic disease, three cases of periprosthetic osteolysis from polyethylene/metallic wear debris in standard THA, and three (N = 3) cases of MoM implants not examined in our series (1 resurfacing, 2 MoM THA). Average time of implantation was 30 months in these patients. Additionally, we examined all cases of preoperative native synovial tissue (time zero) available for patients in our series with ALTR and identified five cases (N = 5) with variable perivascular lymphoplasmacytic infiltrate to provide a baseline comparison. These cases underwent the same pathologic and immunohistochemical evaluation as the ALTR cases in this study. Two archival cases of pelvic lymph nodes in patients with history of total hip replacement served as negative and positive immunohistochemistry controls.
Statistics
Categorical variables were reported as frequencies and percentages and compared between the DMN and MoM THA groups by chi-square tests. Continuous variables were summarized as means and standard deviations and compared between groups with independent samples t-tests. In cases where data was not normally distributed, a Mann–Whitney U test was utilized. All statistical tests were two-sided and p-values less than 0.05 were considered statistically significant. Statistical analyses were performed with SAS version 9.3 (SAS Institute, Cary, NC).
Discussion
Failure due to ALTR has previously been described for MoM bearing surfaces and modular junctions at the head-neck and neck-stem [
4,
9‐
14,
21‐
23,
35,
38,
39]. The purposes of this study were to compare implant-based differences in periprosthetic tissue structure, organization, corrosion product morphology, and cellular composition by conventional histology and immunohistochemistry in ALTR resulting from two common implant configurations. Our results demonstrate that similarities between these two implants included spectrum of histologic patterns, composition of the inflammatory infiltrate, and presence of corrosion products. Differences between these implant types included macrophage and lymphocyte distributions, and corrosion product morphology. This is the first study to our knowledge to compare the histologic and immunohistochemical features of ALTR in two different classes of implants.
We have shown convincing histological evidence that similar common morphologic features exist in ALTR with an early phase of cellular activation and proliferation seen in neo-synovial reaction to other particulate implant materials (e.g. polyethylene) followed by a distinctive sequence of cellular and tissue reactions leading to formation of a variable amount of soft tissue necrosis/infarction. Corroborating evidence is provided by the metachronous development of the reaction in various areas of the periprosthetic tissue, contiguous areas of superficial necrosis, preserved neo-synovial architecture, and absence of necrosis in the bursal tissue until dehiscence of the fluid contained within the pseudocapsule. The time to revision in the DMN group was significantly shorter than the MoM THA, and this suggests different progression rates of ALTR with different implant designs. Progression of ALTR may depend on length of device implantation, toxicity/immunogenicity of corrosion particles, implant design and alignment, patient co-morbidities, and host immune reactivity. The modality of failure of the DMN and MoM THA implants analyzed in this study have been attributed in previous publications to the formation of corrosion products at the metallic interacting surfaces and not to technical mistakes or poor design resulting in mechanical failure of the implants [
5,
9,
11]. Gill et al. also found that corrosion at the modular neck-stem junction resulted in early revision relative to the same monoblock stem and bearing components [
9]. Additionally, Cooper et al. have shown a similar time to failure of the DMN implant used in our study, further corroborating our results were not due to technical error [
11]. A possibility of bias in time to revision might exist because the DMN group had a publicized recall of the implant, however, all patients revised in both cohorts were indicated for revision due to elevated metal ion levels, symptomatic hip pain, MRI findings of moderate to severe adverse tissue reaction, and/or positive needle biopsies. Our observations are similar to previous studies that have illustrated the distinct histological aspects of the reaction, predominantly in MoM hip resurfacing implants or in mixed resurfacing and THA implants [
10,
13‐
19]. Previous publications examining ALTR have used the proposed ALVAL score by Campbell et al. as a grading system of the reaction [
16,
18,
25]. If our interpretation of the natural history of the reaction is correct, the score would be an indication of developmental stage of the adverse reaction rather than a grading system of its biological severity, and therefore of limited clinical value in predicting the course or the biological outcome of the reaction for each specific type of implant.
Different histologic subtypes were observed in ALTR in our study. A subset of patients in the MoM THA group had a macrophagic pattern of failure with minimal lymphocytic response and absent or minimal necrosis. These patients may have impingement related failure, suggested by black metallic particles in their soft tissue and/or an immunoprofile that is less responsive to wear debris. A second subgroup of patients had a mixed macrophagic and lymphocytic response with a variable number of plasma cells, eosinophils, and mast cells. This has been described frequently in ALTR from previous studies and represented the most common pattern we observed [
13‐
18]. A third subgroup displayed a granulomatous pattern with or without inflammatory infiltrate or necrosis, and this patient subgroup may have unique immunologic responses to wear debris. We did not observe any cases with an exclusively lymphocytic pattern without presence of particle-laden macrophages, as described by Berstock et al. [
37]. This difference may be due to the extensive sampling performed of periprosthetic tissue in our study. The association of these different histologic patterns and clinical outcomes needs to be investigated in future studies.
We demonstrated an association between the presence of extra-cellular and intra-cellular corrosion products in the periprosthetic tissue with the presence of interstitial and perivascular lymphocytic infiltrate. This association suggests that corrosion particle laden macrophages are instrumental in the formation of the lymphocytic infiltrate, although free particulate material can also significantly contribute to the response. Corroborative evidence of our interpretation was the presence of benign lymphocytic aggregates in the bone marrow associated with particle-laden macrophages as previously reported in hip resurfacing implants [
15,
40]. The appearance of corrosion materials was different between the two implant designs, which also were associated with differing levels of serum cobalt and chromium ions. The MoM THA has two possible sources of corrosion materials or metallic debris: the metal-on-metal articular surface and the head-neck taper junction. The dual modular neck implant also has two possible sources of corrosion: the head-neck taper junction and the neck-stem taper junction, although the predominant one appears to be the latter [
11]. These different surface possibilities likely explain the variable corrosion material appearance and distribution. Xia et al. used electron microscopy and EDX to assess macrophage content in ALVAL due to failure of a MoM bearing surface and their results showed nanometer-sized inclusions within the phagosomes with significant chromium content by EDX [
41]. We hypothesize that the numerous, predominantly globular small intracellular inclusions seen on light microscopy represent corrosion products generated at the bearing surface, which are not present in the dual-modular neck implant. This observation is confirmed by the presence of the same inclusions in resurfacing implants with the same bearing surface (data not shown). In contrast to intracellular corrosion material, both implant types had large extracellular corrosion aggregates of similar morphology. Our data indicate these materials represent corrosion products from the taper junctions at the head-neck and neck-stem, which is consistent with previous studies [
5,
11,
35]. Analysis of material produced from head-neck taper corrosion suggested that chromium orthophosphate was the most common corrosion material produced at modular junctions, and this material could disseminate into the surrounding soft tissue [
5,
11,
35]. These wear products differ in size and shape from the intracellular products that are seen from the bearing surface, possibly explaining biological or clinical differences between different implant types [
42]. Moreover, the stratified appearance of the aggregates at light microscopy possibly suggests mixing of fluid proteins and secondary particles released from the exfoliated macrophages forming products of unknown and untested cytotoxicity. Early involvement of the hematopoietic bone marrow by macrophages and large aggregates of particles can also influence the adverse reaction, and this may have future biological significance.
The ALTR reaction seen in the DMN implant is unlikely to be influenced by polyethylene debris. There have been extensive publications in the literature about wear rates of highly cross-linked polyethylene in vivo, and for the X3, femoral head penetration rates remain low at two years (head penetration <0.06 mm) [
43]. Moreover, between years 1 and 5, wear rates in vivo were less than 0.001 mm/year [
44]. This data suggests that polyethylene wear is unlikely to contribute to the observed reaction to ALTR seen in our study in the DMN group. This is further corroborated by the fact that only 1 of 54 cases examined in our study had polyethylene debris in their periprosthetic tissue at light microscopy, suggesting that polyethylene debris is unlikely to play a major role in ALTR seen in our study.
Immunohistochemistry results showed a predominant T lymphocytic response with a variable B cell component with the formation in some cases of perivascular germinal centers and tall endothelial cell venules as previously reported [
13‐
17,
19]. The analysis of the T cell population pointed towards a mixed pattern with predominant GATA3 positivity (Th2 lymphocytes) but also substantial T-bet and FOXP3 expressing lymphocytes, representing Th1 and Treg subgroups respectively. These findings were associated with the presence of a population of macrophages strongly positive for CD163, a marker of M2 macrophages, a subset frequently correlated with Th2 cytokines [
45]. The frequent finding of a variable number of CD117 positive mast cells is also a new important finding with implications in reaction initiation/progression due to their interactions between T and B cell lymphocytes and eosinophils, and their potential to produce M2 inducing cytokines such as IL-4 [
46]. Reaction initiation and severity may be explained by the release of chemokines from macrophages under oxidative stress and/or direct lymphocyte cytotoxicity [
47‐
50]. Similar lymphocyte distributions were observed in the cases of osteoarthritis at time zero, and the possibility of a non-specific common pathway in different inflammatory conditions of the synovial membrane not representative of the initial response of the adverse reaction must be considered and confirmatory studies with testing of other specific antibodies are needed. It is also possible that the lymphocyte distributions seen in our study reflect an innate immunologic profile of the synovium with subsequent adaptive modulation, and analysis of pathologic gene expression patterns could be helpful to elucidate the role of these lymphocytic sub-populations in initiation and progression of ALTR [
51]. Collectively, the immunohistochemistry studies indicate a complex adaptive immune response potentially involving several cell types. Future molecular analysis will help define the signaling pathways that orchestrate the tissue necrosis and other pathologies underlying ALTR.
The main limitation of this study is the attempt to reconstruct the natural history of the reaction based on one cross sectional observation at the time of implant revision. We compensated for this limitation by extensive topographical sampling of the periprosthetic soft tissue, but we acknowledge that continued longitudinal observation would be needed to confirm our findings.
Competing interests
There are no competing financial or non-financial interests in direct relation to this manuscript for any authors. Author S Jerabek is a consultant for Mako Surgical Corporation.
Authors’ contributions
GP conceived of the study, collected and analyzed synovial tissue, and drafted the manuscript. BR assisted with histologic and immunohistochemistry analysis and drafted the manuscript. SAJ analyzed the clinical data. GM performed immunohistochemistry and assisted with study design. GW and DM collected synovial tissue and assisted with manuscript preparation. SRG participated in study design and interpretation of data. PEP participated in study design and coordination. All authors read and approved the final manuscript.