Inflammatory tissue reactions around aseptically loose cemented hip prostheses: A retrieval study of the Spectron EF stem with Reflection All‐Poly acetabular cup

Abstract The cemented Spectron EF stem in combination with the cemented non‐crosslinked Reflection All‐Poly cup showed a high rate of mid‐term aseptic loosening. However, the failure mechanisms are not fully known. We assessed the inflammatory tissue reactions and wear particles in periprosthetic tissues, implant wear and blood metal ion levels in 28 patients with failed implants. Histological analysis showed a macrophage pre‐dominant pattern with randomly distributed lymphocytes, with various amounts of neutrophils and giant cells. The number of different cell types in the tissue samples from patients in the cup group and in the stem group was similar. Wear particles, mainly ZrO2, CoCrMo, and polyethylene particles of different sizes and shapes, were associated with macrophages/giant cells, and total particle load/mm2 was higher in cases of stem loosening. The Spectron EF stems were heavily worn, abraded, and polished. Stem abrasion correlated with metal ion concentrations in blood. The median polyethylene wear rate of the Reflection cups was 0.23 mm/year. The high proximal roughness of the Spectron EF stem resulted in excessive cement wear during loosening. The resulting inflammatory tissue responses to the degradation products both from the cup and the stem led to massive osteolysis and subsequent implant loosening.

chromium alloy that achieved very good survival rates of 98.4% after 10 years, as reported by the Swedish Total Hip Register 2 and showed superior longevity compared to the cemented Charnley stem. 3 Early reports of the new design raised concerns regarding the stem survival, and this was partly attributed to the roughness of the stem. 4,5 Later, another study also found high levels of osteolysis and stem failure with the Spectron EF. 6 The cemented Charnley, low-friction arthroplasty, pioneered by Sir John Charnley,7 provides the basis of comparison to new designs in modern cemented hip arthroplasty as it has become the gold standard in THA. 8 The Norwegian Arthroplasty Register (NAR) reported a high number of cases of mid-term aseptic loosening of the cobalt-chrome Spectron EF stem in combination with the reflection cemented non-crosslinked All-Poly cup (Smith and Nephew) compared to the Charnley THA. They identified a 3.8-fold T A B L E 1 Patient data, implant duration, and reason for revision. Two patient groups are included in this study: patients with loose cup only and patients with loose stem with cup loosening. A descriptive summary of the age at revision and implant duration is given below the respective group Case no.
Gender, age at revision (year) Implant duration (months) Reason for revision The Mann-Whitney test was used to compare the medians of the cup and stem group (p = .36). b The Mann-Whitney test was used to compare the medians of the cup and stem group (p = .12).
higher risk of revision for this combination. 9 The wear rate of the Reflection All-Poly cup was shown to be very high in a randomized controlled trial. 10 A major cause for revision following THA is aseptic loosening due to osteolysis, that is, bone resorption due to an inflammatory tissue response around the implant. 11 Register (Table 1). 9 We divided the patients in two groups according to the reason for revision: Patients with cup loosening (cup group; n = 10) and patients with stem and cup loosening (stem group; n = 18) ( Table 1).

| Tissue preparation and histological evaluation
Periprosthetic tissue samples from the joint capsule or periprosthetic membranes were collected during revision surgery and fixed in 4% buffered formalin. The specimens were embedded in paraffin, sectioned at 4 μm and stained with hematoxylin and eosin (H&E). Two tissue slides per patient were selected for histological evaluation.
However, in three cases, there was only one sample per patient available. Two pathologists (A. C. J. and K. S.) semi-quantitatively evaluated the tissue slides from all patients in a blinded fashion using a modified Mirra classification 14,15 described by Doorn et al. 16 Three cell-rich highpower fields (HPF, 40Â) per tissue slide were counted by the two pathologists and the number of observations was divided by six and graded for macrophages, acute and chronic inflammatory cells, foreign body giant cells and the amount of metal particles. Hereby, the total number of cells and particle intensity was graded as absent (0), low (1+), moderate (2+), or high (3+), as described in our previous study. 15 Additionally, in the present study, the degree of necrosis in the tissue sections was graded 0-10% (0), 10-50% (1+), 50-100% (2+), or 100% (3+). Tissue samples from three patients (case no. 6, 12, and 26) showed only connective tissue and/or fat tissue, which was not representative for the tissues to be analyzed. The histological results of these samples have been excluded from statistical analysis.  T A B L E 3 Periprosthetic tissue reactions in patients with stem and cup loosening (n = 18); evaluated with the modified Mirra classification Case no.

Neutrophils Macrophages Chronic inflammatory cells Multinucleated giant cells Metal particles/macrophages Necrosis
Frequency distribution (%) and medians of scores 0 to 3+

| Particle characterization
Total wear particle load was assessed from the tissue slides with High-Resolution Optical Darkfield Microscopy (HR-ODM; Auburn, AL) as described before. 15,17 The particles in three particle-rich sections of each tissue slide were counted and measured using image analysis software (NIS-Elements 2.30, Nikon, Japan). Birefringent polyethylene (PE) particles were detected by light microscopy with a polarization filter and counted in three particle-rich areas in 400Â magnification per tissue slide (Olympus CX31, U-TP530 Analyzer). Additionally, PE particles were isolated according to ISO 17853 in eight tissue samples. Following tissue digestion, the PE particles were purified by high-speed centrifugation as described in ISO 17853, which allows all particle sizes F I G U R E 3 Necrosis of periprosthetic tissue. Paler stained, cellfree tissue area, as compared to more cell-rich areas as shown in Figure 1. H&E, 20Â Zeiss AG, Germany) in secondary emission and back scattered modus at 15 kV. Briefly, the paraffin-embedding was removed by soaking the samples in xylene (Merck, Darmstadt, Germany), followed by 100%, 96%, 80%, 70% ethanol (Arcus kjemi, Vestby, Norway) and finally rinsed in distilled water. The chemical composition of the particles in the tissue samples was determined using energy dispersive X-ray (EDX) analysis.

| Stem analysis
The damage of each retrieved stem was assessed using a semiquantitative grading system consisting of three abrasion levels (A, AA, AAA), described by Willert et al. 18 No visible damage was defined as level 0. Each side of the stem was divided into four quadrants using the line between the rough and matt part as a natural division line between the proximal and distal part. The grades of each zone were added to give a total abrasion score of maximum 3 Â 16 = 48 points.

| Cup analysis
Using radiographs, the linear wear of the cup 19 was determined with an X-ray software "Hip X-ray" using pre-defined methods. Additionally, the orthopedic surgeons registered the Paprosky classification

| Blood analysis
Whole blood samples from 22 patients were analyzed as described previously. 15 Briefly, blood samples were drawn just prior to revision surgery from the patient's forearm. An aliquot of 1.5 g whole blood was mixed with 3 ml 60% HNO 3

| Statistical analysis
Statistical analysis was done using GraphPad Prism 9 (GraphPad Software, La Jolla, CA). Data are presented as medians with range and CI, where applicable. Statistical differences between parameters (particle load/mm 2 , metal ion levels) in the independent groups were determined using the Mann-Whitney U test, which compares ranks of two different groups. Relationships between two non-parametric parameters were analyzed using Spearman's rho correlation. Results were considered statistically significant with p < .05.

| Ethics
The project protocol and the retrieval biobank were approved by the  Table 1 lists the patient data, implant duration, and reason for revision for the two groups. Additionally, a descriptive summary of the data (i.e., age of the patient at revision and implant duration) is given in Table 1.  Figure 1A,B). A grade of 3+ was given in 58% and 50% of the cases in the stem group and cup group, respectively (Tables 2 and 3). A few plasma cells, but no lymphoid follicles were observed ( Figure 1A).
Chronic inflammatory cells, that is, lymphocytes, were mainly randomly distributed between the macrophages ( Figure 1B), but not as numerous as these cells. A few scattered neutrophils were found in both groups ( Figure 1C). Multinucleated giant cells, with grade 2+ or 3+, were found in 36% and 19% of the tissue samples in the stem group and the cup group, respectively ( Figure 1D) (Tables 2 and 3). Wear particles were phagocytosed by macrophages and were seen as black and/or dark gray spots in the cytoplasm of these cells (Figure 2). Macrophages with metal particles were more abundant in tissue samples from patients in the stem group (median score 2+) than in patients in the cup group (median score 1+). Moreover, the number of macrophages with metal particles correlated with the total particle load/mm 2 in this group (stem group:  0.036-0.582). In the stem group, 48% of the cases were graded 2+ or 3+, which corresponds to 50 to 100% necrosis, whereas in the cup group, 43% of the tissue samples showed more than 50% necrotic areas.

| Particle analysis
Using high-resolution optical microscopy, a higher total particle load was found in tissue samples from patients in the stem group compared to the samples from the cup group (median stem = 23,893 particles/mm 2 , range: 8127-83,021 particles/mm 2 , vs. median cup = 9244 particles/ mm 2 , range: 232-55,362 particles/mm 2 ; p = .001) (Figure 4). SEM-EDXA showed that zirconium dioxide (ZrO 2 ) particles from the bone cement were frequent in the tissue samples and these particles were usually round and below 2 μm ( Figure 5A). CoCrMo particles were mainly found in the stem-related tissue sections and were normally longer than 1 μm and irregular in shape ( Figure 5B). Using polarization microscopy, PE particles were found in most of the tissue samples from both groups  Figure S1).

| Implant analysis
Stem analysis showed that most of the stems had signs of mechanical wear, seen as abrasion and polishing ( Figure 7A,B). In 44% of the cases, the stems were highly worn (abrasion score > 30). Femoral bone loss was found to be extensive (Paprosky type II and III) in 15 of 18 cases (Table 4). The median linear wear of 10 cups in the stem group was 2.07 mm (range: 0.90-4.05 mm) and the wear rate 0.24 mm/year (range: 0.11-0.36 mm/year). In the stem group, acetabular bone loss was found to be extensive in 83% of cases (15 out of 18 cases; Table 4). Cup analysis showed that most cups were highly worn ( Figure 7C). Third-body wear particles, mostly made of ZrO 2 and some CoCrMo, were found inside the cups, embedded in the plastic surface. In the cup group, the median linear wear of the cups was 1.60 mm (range: 0.82-3.31 mm) and the wear rate 0.21 mm/year (range: 0.11-0.32 mm). Acetabular bone loss was extensive (Paprosky type II and III) in 80% of the cases of the cup group (8 out of 10 cases; Table 4).

| Metal ion analysis
The

| DISCUSSION
In the present study, we aimed to assess inflammatory tissue responses, particle characteristics in the peri-prosthetic tissues, and lesions that are highly lymphocyte-infiltrated do not exhibit a large number of macrophages. 25 There were more macrophages than lymphocytes in the periprosthetic tissue samples from both the cup and the stem group. This is in contrast to a recent study that showed that the macrophage pre-dominant pattern in the periprosthetic tissue is usually associated with Metal-on-Metal (MoM) hip prostheses, whereas a mixed macrophage-lymphocyte pattern was found to be more frequent in tissue samples from Metal-on-Polyethylene (MoP) hip implants. 26 At the femoral side, wear particles of CoCrMo, polymethylmethacrylate (PMMA) and the radiopacifier ZrO 2 were found.
Activated macrophages phagocytize these wear particles, which is evidenced by the particles within these cells (Figure 3) Surprisingly, we detected a slightly higher number of PE particles in the tissue samples from the stem group compared to the cup group.
However, due to detection limitations with polarized light microscopy, we also did SEM analysis of isolated PE particles that showed a high number of submicron PE particles in both groups that were too small to be seen with light microscopy. The median diameter of the PE particles was found to be 0.31 μm (range 0.05-1 μm), which is slightly less than the mean diameter (660 nm) reported by a previous study. 21 However, the small sample size in both groups made statistical analysis unfeasible. Thus, it remains unclear if the tissue samples in the stem group contained more PE particles than samples in the cup group. Implant-derived metal ions have also been shown to play a pivotal role in the osteolytic process. 29 38 In both groups, the blood cobalt ion concentration was higher than that of chromium.
Chromium is known to accumulate to a high degree in the periprosthetic tissue while cobalt ions are rapidly transported to the blood stream and are eliminated via urine. 36 We found that blood metal ion concentrations correlated with the total particle load and stem abrasion, which is not surprising given the roughness of the proximal stem. Similarly, a recent analysis showed elevated Cr ion levels in serum with increasing roughness of CoCrMo femoral head in MoP hip implants. 28 As a result of micromotion at the interface between the stem's rough surface and the cement, metallic and cement particles are generated due to fretting and abrasion. 39,40 These particles, together with the PE particles, originating from the cup, are capable of causing osteolysis and loosening of the stem. 5,6 A potential consequence of the activation by, and phagocytosis of, wear particles is cellular toxicity, including cell death. Previous studies have shown that PE and metal particles can induce apoptosis in macrophages in a size-and dose-dependent manner. 41 However, it has also been demonstrated in cell culture studies, that macrophages stay viable and are able to tolerate PE particles better than metal particles, which led to an increased osteoclast differentiation (favoring osteolysis) of PE-laden macrophages compared to macrophages that phagocytized CoCrMo particles. 21,42 High PE wear, due to the inferior plastic quality, led to osteolysis in the acetabulum and cup loosening. The wear rate is comparable to findings from radiostereometric (RSA) studies with the same cup type of non-crosslinked UHMWPE. 10 Taken together, the design of the Spectron EF stem with its high proximal roughness gives an excessive amount of metallic and cement wear particles during stem loosening. Furthermore, the hard cement particles also increase cup plastic wear through "third-body wear". The resulting inflammatory tissue responses due to the high particle load leads to excessive osteolysis around the implants in patients with the Spectron EF stem and the Reflection All-Poly cup.

ACKNOWLEDGMENTS
The authors thank the patients and the surgeons that contributed with the retrieval samples and Irene Ohlen Moldestad for retrieval handling. Ruiting Zhao, Roohollah Milimonfared and Alex van Tol contributed with retrieval analysis.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon request.