Autopsy
A review of autopsy and surgical records identified seven patients who had received second-generation metal-on-metal total hip replacements and who had died and been autopsied at our hospital between 2002 and 2007. There were four men (five hips) and three women (four hips), with a mean age of seventy-two years (range, fifty-seven to eighty-one years) at the time of death and a mean time between arthroplasty and death of sixty-eight months (range, thirty-six to 120 months). The clinical data are summarized in Table I. The acetabular cup shells (Alloclassic; Zimmer, Winterthur, Switzerland) were of pure titanium. The metallic bearing surface on the acetabular side consisted of CoCrMo alloy and a polyethylene interface layer. The femoral component of the prosthesis (Zweymüller SL; Zimmer) consisted of titanium alloy (Ti6Al7Nb) coupled with CoCrMo-alloy femoral heads, which were mated with the head-neck conical taper of the stems. The Co28Cr6Mo2 carbon alloy (Metasul; Zimmer) complied with ASTM (American Society for Testing and Materials) F 1537 and ISO (International Organization for Standardization) 5832-12 and can be classified as a high-carbon alloy with a fine carbide distribution. According to the manufacturer, the articulating parts of the metal-on-metal bearing system exceed the requirements of ISO 7206-2, with a roughness (Ra) of <0.01 µm, a sphericity of <10 µm, and a clearance between 50 and 120 µm. All femoral heads had a diameter of 28 mm. In compliance with the relevant provisions of Austrian legislation, the implant-bearing femora and acetabula, including the bearing components, were retrieved during routinely performed postmortem examinations. In addition, joint capsule tissue was obtained for histological analysis.
Radiography
All radiographs were evaluated by two independent, experienced orthopaedic surgeons (P.Z. and K.Z.). Postoperative anteroposterior radiographs of six hips (Cases 1 and 3 through 7) were available for comparison with postmortem radiographs. The postoperative cup inclination was measured, and abduction of <40° was considered a varus position and abduction of >50°, a valgus position.
Anteroposterior and lateral radiographs were made of the implant-bearing femora that were retrieved at autopsy. Stability of the femoral component fixation was assessed according to the method described by Engh et al.18, in which the presence of endosteal new-bone formation bridging the gap between bone and implant, the absence of reactive lines around the implant, the occurrence of pedestal formation (a shelf of endosteal new bone at the stem tip partially or completely bridging the medullary canal), and implant migration are evaluated. Osteolytic lesions were defined as a focal area of bone resorption that was =2 mm wide19. Osteolysis adjacent to the stem was recorded according to the zones described by Gruen et al.20. As in previous investigations21, calcar resorption was not classified as osteolysis.
Although ground sections from the retrieved implant-bearing acetabula have not been performed at present, radiographs were made to evaluate the condition of the acetabular bone-implant interface. Osteolytic areas around the cup were classified into the three zones described by DeLee and Charnley22.
Histology
Ground Sections
After radiographs of the implant-bearing femora were made, the specimens were fixed in 7.5% phosphate-buffered formalin. With use of a diamond-coated saw (310 CP; EXAKT, Norderstedt, Germany), four transverse segments of approximately 1 cm in thickness were excised. The level of the sections included the proximal (segment 1) and the distal region (segment 4), as well as the metadiaphysis (segment 2) and the diaphysis (segment 3) (Fig. 1). The segments were dehydrated and embedded in methylmethacrylate without decalcification. With use of a micro-grinding system (400 CS; EXAKT), the embedded specimens were cut into 100-µm-thick transverse sections; 10-µm-thick slides were prepared and stained with toluidine blue. The slides were evaluated with use of conventional and polarized light microscopy (Eclipse 80i; Nikon, Tokyo, Japan) at final magnifications of ×4, ×25, ×100, and ×400.
Joint Capsule Tissue
The obtained joint capsule tissue was fixed in 7.5% buffered formalin, embedded in paraffin, sectioned to a thickness of 2 µm, and stained with hematoxylin and eosin. At least three tissue blocks (range, three to six tissue blocks) from each joint capsule were prepared for investigation. Stained slides were examined with use of conventional and polarized light microscopy (Eclipse 80i; Nikon). The entire tissue sections were studied throughout. From each specimen, representative sites were evaluated and the predominant findings were recorded.
The amount of metal particles and the intensity of diffuse and perivascular lymphocytic infiltrates were graded according to the method of Willert et al.15: the visible metallic particles were defined by their high density, black color, and shape. The amount of metal particles was recorded, according to the grading system, as none (—) indicating no or only isolated phagocytized particles without major macrophage reaction; as few (+), a few particles phagocytized in some spots and/or accumulated perivascularly; as many (++), evident accumulation of particles phagocytized in macrophages; as abundant (+++), tissue is loaded with particles, including foreign-body granulomas; and as excessive (++++), tissue is overstuffed with particles, with foreign-body granulomas dominating the structures everywhere.
Lymphocytic infiltration was assessed by counting the number of diffusely distributed lymphocytes (magnification, ×40) and the number of perivascular lymphocytic agglomerations (magnification, ×4) per field of view, with none (—) indicating =10 diffusely distributed lymphocytes and/or no perivascular agglomeration; few (+), eleven to thirty lymphocytes and/or one to two agglomerations; many (++), thirty-one to fifty lymphocytes and/or three to six agglomerations; abundant (+++), fifty-one to 100 lymphocytes and/or seven to ten agglomerations; and excessive (++++), >100 lymphocytes and/or >10 agglomerations.
To identify inflammatory cells, the specimens were stained immunohistochemically (BenchMark; Ventana Medical Systems, Illkirch, France) with use of the avidin-biotin-complex method. Tissue sections were incubated with antibodies against CD3 (diluted 1:70; Dako, Glostrup, Denmark) to detect T-lymphocytes and CD20 (diluted 1:700; Dako) for B-lymphocytes. For positive controls, lymphoma tissue and tonsillar tissue were stained with use of the same method.
To assess the extent of necrosis, the rating scale described by Doorn et al.23 was used, with + indicating 1 to 2 mm of necrosis per slide; ++ indicating 3 to 9 mm of necrosis per slide; and +++ indicating >1 cm of necrosis per slide.
The joint capsule tissue was evaluated for the presence of tissue changes associated with metal sensitivity after the use of metal-on-metal articulations described as ALVAL (aseptic lymphocyte-dominated vasculitis-associated lesion)15,16. These tissue changes include surface ulceration with fibrinoid necrosis and excessive perivascular or diffuse infiltrates of a mixture of B and T lymphocytes, and they may be found in conjunction with the presence of plasma cells, eosinophilic granulocytes, and swelling of the vascular endothelium.
Wear Analysis
Two-micrometer-thick unstained sections were used to investigate the composition of particles and inclusions within macrophages. For each case, the section containing the highest number of particles was selected for study. In addition, a 5-mm-thick section from segment 1 of the first patient (Case 1), embedded in methylmethacrylate, was used to trace and evaluate metal particles in the periprosthetic osteolytic lesion. Unstained slides were sputtered with gold and inspected with use of a scanning electron microscope (Philips XL40; Philips, Eindhoven, The Netherlands). In order to identify the elements, energy-dispersive x-ray analysis (EDAX; Philips) was used on selected areas.
Tribology
Tribological investigation of the hemispheres of the femoral heads and the inlays was performed by using a coordinate measuring machine (LH 1210 CNC; Wenzel Group, Wiesthal, Germany) and a scanning electron microscope (XL 40; Philips). Measurements were performed as specified in ISO 14242-2. Tribologically relevant parameters, including clearance (joint space), maximal linear wear, and volumetric wear of the articulating surfaces, were determined. In addition to the quantitative analysis, the tribological strained surfaces were investigated with use of scanning electron microscopy to characterize the wear modes involved. In order to identify damage of the metallic matrix, energy-dispersive x-ray analysis was used in a two-dimensional area-mapping mode. As a reference for assessing the quantitative wear analysis, articulating components with an implantation period of twelve days were used.
Source of Funding
There was no external funding source.
Autopsy
Medical records revealed that four patients received six implants in our hospital, and three patients were treated in two other orthopaedic centers. All nine devices were inserted in a primary arthroplasty to treat osteoarthritis. One patient (Case 1) had reported increasing hip pain for the previous year. The other six patients had been free of symptoms concerning the hip replacement at the time of death. None of the patients showed evidence of an implant-associated infection. The cause of death was metastatic lung cancer in two patients; severe atheroscleroses with additional diseases, such as liver cirrhoses, diabetes, and pneumonia, in four; and advanced asthma accompanied by infection in one. None of the deaths were related to the arthroplasty (Table I).
The joint spaces presented various amounts of gray to black-colored debris in seven hips (Cases 1 through 5, 7, and 9). A small rim of debris was adherent to the inner surface of the removed joint capsule tissue in those cases. All hips showed thickened joint capsule tissue with whitish to reddish cut surfaces. There was no gross evidence of impingement or of corrosion at the head-neck junction.
Radiography (Table II)
The postoperative radiographs of six hips (Cases 1 and 3 through 7) showed close contact of the bone-implant interfaces. The angle of cup inclination revealed a standard position, ranging from 46° to 50°, in three hips (Cases 5, 6, and 7) and a valgus position, ranging from 52° to 60°, in three hips (Cases 1, 3, and 4).
The radiographs of one implanted stem (Case 1), in the only symptomatic subject, revealed osteolytic lesions of up to 5 mm in thickness in seven zones (Fig. 2). In addition, there was moderate sclerosis in zones 3 and 5 and signs of slight stem migration. Minimal proximal osteolysis of 3 mm in thickness in zone 14 was observed in two femora (Cases 4 and 7). The remaining stems showed no evidence of osteolysis. Compared with postoperative radiographs, postmortem radiographs of three femora (Cases 1, 5, and 9) demonstrated minimal new-bone formation at the stem tip, but there was no radiolucency between the bone and the implant. All but one of the femora (Case 1) showed intimate contact along the whole or the greater part of the bone-implant interface and were considered to be stable. Signs of atrophy (stress-shielding) in the proximal region were present to a minimal to moderate extent in all but one femur (Case 1).
The radiographic investigation of the retrieved implant-bearing acetabula showed radiolucency in zone 2 in one acetabulum (Case 1), raising suspicion for an osteolytic lesion. However, osseointegration of this cup was evident in zones 1 and 3. One cup (Case 5) revealed linear osteolytic lesions, ranging from 1 to 3 mm in thickness, in all zones, indicating implant loosening. The remaining cups showed complete osseointegration without evidence of linear or focal osteolysis.
Histology (Table II)
Ground Sections
Six femora showed signs of bone loss. One femur (Case 1; Fig. 3) revealed a circular area of osteolysis up to 6 mm in width in segments 1 and 2 and a dorsal area of osteolysis of 5 mm in length and 1 mm in width in segment 3. One femur (Case 7) showed osteolytic lesions, measuring from 3 to 10 mm in length and from 1 to 5 mm in width, in segments 1 and 2. Four femora (Cases 4, 5, 8, and 9) showed focal lesions, measuring from 5 to 7 mm in length and from 1 to 4 mm in width. Within the osteolytic lesions, which were composed mainly of necrotic cell debris demarcated from the adjacent bone by a small rim of connective tissue, some macrophages and a few lymphocytes were diffusely distributed. Sparse dark, spherical, metallic particles were observed freely within necrotic debris and occasionally within the cytoplasm of macrophages. There was no birefringence under polarized light. Except for the above-mentioned focal bone lesions, eight femora (Cases 2 through 9) otherwise showed excellent osseous integration of the implant in all section levels with mainly forceps-like bone ongrowth on the edges of the rectangular implant (Fig. 4). In addition, one femur (Case 1) showed periosteal bone apposition in segments 2 and 3 (Fig. 3, b and c).
Joint Capsule Tissue
The surface of the capsular tissue showed ulceration and fibrinoid necrosis to a various extent in eight hips (Fig. 5, a). Seven hips (Cases 2 through 5, 7, 8, and 9) revealed superficial necrosis of 1 to 2 mm (+) and one hip (Case 1), up to 6 mm (++). One hip (Case 6) showed surface lining cells occasionally containing metallic debris, but showed no evidence of surface ulceration. The necrotic layer was commonly demarcated by metal particle-laden macrophages (Fig. 5, a and b). In addition, two hips (Cases 1 and 8) revealed some multinucleated foreign-body giant cells containing metal particles adjacent to the necrotic layer. Metallic wear debris was present in all hips and was rated as few (+) to many (++). Birefringence was absent in all hips. Within the inner layer of the joint capsule, diffusely distributed lymphocytes rated as few (+) to abundant (+++) were observed in eight hips. One hip (Case 6) had none.
The intermediate vascular layer of joint capsule tissue revealed aggregates of lymphocytes surrounding small capillary vessels rated as none (—) in four hips, as few (+) to many (++) in four hips, and as excessive (++++) in one hip (Case 1). The latter showed some lymphocytes interspersed within the walls of vessels and frequent swelling of the vascular endothelium. In addition, perivascular aggregates of lymphocytes were focally intermixed with plasma cells in two hips (Cases 1 and 7) (Fig. 5, c). Confirmed by immunohistochemical analysis, diffuse and perivascular lymphocytic infiltrates consisted of a mixture of T and B lymphocytes. Diffusely distributed eosinophilic granulocytes were additionally observed in one hip (Case 1). Signs of bleeding characterized by superficial hemosiderin deposits were observed in one hip (Case 2). Granulomas were absent. Fine-grained material and small green to yellow-colored particles, occasionally intermixed with black pigmented, small, metallic particles, were seen within the cytoplasm of mononuclear macrophages in two hips (Cases 1 and 7) (Fig. 6, a). Polymorphonuclear granulocytes were absent in all hips.
Wear Analysis
Energy-dispersive x-ray analysis spectra of aggregates of particles and macrophages identified chromium peaks, but not cobalt. In addition, traces of phosphorus (P) and oxygen (O), indicating the presence of corrosion products (chromium orthophosphate), were evident in three hips (Cases 1, 7, and 8) (Fig. 6, b and c). Titanium and aluminum peaks were absent in all probes. However, free metal particles within necrotic debris at the surface of the joint capsule and within osteolytic periprosthetic lesions could not be detected.
Tribology (see Appendix)
The annual linear wear rate calculated from wear on the ball head and the inlay was 3.6 µm/yr (range, 2.3 to 7.6 µm/yr). Expressed in volumetric terms, 0.73 mm3/yr (range, 0.33 to 2.06 mm3/yr) of metallic wear particles were given off into the joint space each year, corresponding to a mean weight of 6.1 mg/yr (range, 2.8 to 17.3 mg/yr) at a density of 8.38 g/cm3. The joint space (clearance), on the average, was found to be 89 µm (range, 72 to 100 µm). The highest amount of wear (7.6 µm/yr) was produced in Case 1. One hip (Case 9) had to be excluded from evaluation of wear parameters since the metal inlay of the acetabular cup demonstrated severe deformation at explantation. Traces of third-body wear were seen on all surfaces. The surface-embedded particles causing this wear were identified by means of energy-dispersive x-ray analysis as aluminum oxide particles from the grit-blasted implant surfaces. There was no evidence of corrosion of the bearing surfaces.
Second-generation metal-on-metal articulations have shown reduced wear by improved component design and the use of novel manufacturing techniques1. Particulate wear debris is predominantly released by abrasive, adhesive, or third-body wear. Metal particles may accumulate and be stored locally or may be distributed to distant organs, leading to local and systemic effects9,24-26. The cellular response to wear highly depends on the composition, size, and shape of particles27, as well as the concentration and duration of exposure28.
The release of ions from the surfaces of metal implants or from metallic wear particles is the result of electrochemical metal degradation (corrosion), which may continue even intracellularly29,30. Ions may remain locally or may be transported by binding to proteins in the bloodstream and lymph fluid to remote organs31,32. Elevated levels of cobalt and chromium ions have been found in periprosthetic tissue, distant organs, blood, and urine8-10,33. The release of ions and/or reactive oxygen species can activate the immune system34,35. The resulting response by local cells (histiocytes and macrophages) leads to the release of proinflammatory factors, which can induce events resulting in osteolysis and the loss of the implant36,37.
Likely because of their rare occurrence in this study, diffusely distributed metal particles within necrotic debris on the surface of the joint capsule and within periprosthetic osteolytic lesions could not be traced and analyzed by scanning electron microscopy and energy-dispersive x-ray analysis. However, the analysis of particle-containing aggregates of macrophages within the joint capsule tissue revealed small chromium peaks. These findings and the absence of any titanium particles indicate that these particles originated from the CoCrMo alloy. Cobalt peaks were not detected, which can be explained by the dissolution of cobalt into the surrounding environment. Wear particles with reduced cobalt were considered to have been stored in tissue for a longer time30.
The traces of phosphorus (P) and oxygen (O) indicated the presence of chromium orthophosphate corrosion products in three hips that revealed osteolytic lesions to varying extents. Corrosion products probably appeared as small green to yellow-colored particles29,38 within macrophages observed by light microscopy in two hips (Cases 1 and 7). These inclusions strongly resembled previously described drop-shaped unidentified particles found in conjunction with ALVAL-associated tissue reactions15. A possible relationship between corrosion and an implant-related immune response may exist39. Corrosion products are considered to serve as third-body wear and can induce bone resorption in vitro29,38. Since we were not able to detect the source of those particles at the implant surfaces, the present corrosion products, which were observed mainly intermixed with intracellular metallic particles, may have been released from wear particles themselves38,40.
In contrast to titanium alloy, which may be associated with the most successful biocompatible implants41, cobalt and chromium are known to be cytotoxic11-13. They are considered to affect the phagocytic activity of macrophages and may lead to cell death42,43. In addition, the potential carcinogenesis of cobalt-chromium alloy remains a concern. However, studies on large cohorts have revealed rates of most types of cancer similar to those of the general population44,45. Two of our subjects died of lung cancer, which may have been related to smoking.
In the present study, metallic debris was observed to a various extent in all tissue specimens. Furthermore, tissue reactions including diffuse and perivascular lymphocytic infiltrates as well as surface necrosis were found in varying amounts in eight hips. Prominent lymphocytic infiltration has been reported to be a characteristic histological pattern of tissue reaction to metal particles in metal-on-metal articulations, not necessarily resulting in implant loosening15,16,46. The extent and severity of surface ulceration has been associated with the presence of perivascular lymphocytic infiltration in the deeper layer of the joint capsule16. Among our cases, the highest value of lymphocytic infiltrate in conjunction with the greatest extent of surface ulceration was present in one hip (Case 1), which showed signs of implant loosening. However, three hips showing surface necrosis lacked a perivascular lymphocytic infiltration.
The synovial membrane may play an important role in foreign-body reactions. It consists of fibroblast-like type-B cells and macrophage-like type-A cells. The latter are known for their immunoreactivity to several monoclonal antibodies against macrophage-derived substances35,47. Willert et al. suspected that rapidly increasing osteolysis and joint effusion in metal-on-metal articulations may be caused by an immunological response15.
In the present study, two hips (Cases 1 and 7) showing distinct osteolytic lesions revealed simultaneously the most prominent perivascular lymphocytic infiltrates. Although the number of hips we investigated was low, the findings may indicate a relationship between the extent of tissue inflammation and the development of osteolytic lesions in metal-on-metal total hip replacements. However, no correlation between the quantity of metal debris stored in tissue and the extent of osteolysis or the extent of tissue inflammation was evident.
Although one hip (Case 1) revealed the highest value of wear production by tribological investigation (7.6 µm/yr), the storage of wear debris within the capsular tissue was rated as many, likely because of the transportation mechanism. However, in that hip, malposition of the cup (an inclination angle of 60°) was evident and likely the reason for the increased wear and associated osteolysis. However, other hips (Cases 5 and 7) showing a neutral implant position also had osteolytic lesions, which implies that additional factors may be involved in the development of osteolysis.
In comparison with previous investigations, the tribological analysis of eight hips (Cases 2 through 9) showed the amount of wear to be within the normal range (<4 µm/year)48,49. Nevertheless, five of them (Cases 4, 5, 7, 8, and 9) revealed minimal osteolytic lesions by morphological analysis, and two (Cases 4 and 7) did so by radiographic analysis. Thus, the detection of small osteolytic periprosthetic lesions may not be adequate with use of conventional radiography and may require a more effective instrument, such as magnetic resonance imaging50. Except for one hip (Case 1), the implants we investigated were otherwise well fixed. Osteolysis has not been associated with well-fixed metal-on-metal total hip replacements in general3. In a single report, osteolysis was found in a stable, well-fixed metal-bearing implant and was attributed to a multifactorial event51.
In contrast to metal-on-metal total hip replacements, tissue specimens obtained from hips with metal-on-polyethylene articulations have been reported to have less surface ulceration, a prominent foreign-body reaction dominated by macrophages and foreign-body giant cells containing polymer debris, and fibroblasts; metal debris was rare or absent16,23. In some studies, plasma cells were not observed and lymphocytic reactions were thought not to play an important role in polyethylene wear particle-related osteolysis15,16,23,46. However, infiltrates of lymphocytes have been noted previously in several specimens of total hip replacements with polyethylene-on-metal articulations, and the number of polyethylene-laden macrophages had a direct relationship to the extent of bone resorption52.
A previous investigation comparing total hip replacements with metal-on-polyethylene articulations and those with metal-on-metal couples showed similar clinical results3. However, the investigation of postmortem retrieved devices provides the most objective information to evaluate the effect of the implant on the host environment and the effect of the host environment on the implant. Several postmortem studies have examined uncemented femoral components histologically in situ, focusing on the extent of implant coverage by bone53-65. Those studies have included porous-coated and hydroxyapatite-coated stems combined with metal-on-polyethylene couples. A comparison with our findings shows that there may be a similar frequency and extent of osteolysis in well-functioning total hip replacements with metal-on-polyethylene couples. A thorough analysis of particles and the related periprosthetic tissue reaction in well-functioning, osseointegrated devices has been included in a few studies, such as the one by Tonino et al.62. That study showed metal particles with hardly any inflammatory reaction adjacent to the implant. Polyethylene particles with an associated tissue reaction were present in one case, and a small focus of particles without inflammation was present in a second case.
To our knowledge, there is no comparable histological investigation of implant-bearing femora to evaluate osteolytic lesions associated with metal-on-metal total hip replacements. However, to provide a broad base of information regarding the long-term effect of metal-on-metal articulations, further investigation of patients with use of expanded studies, including autopsy analysis, is necessary.