A total of sixteen fourth through seventh-generation Coonrad-Morrey total elbow implants (Zimmer) were retrieved from fourteen patients (see Appendix). In all cases, the retrieved implant was the primary implant for that joint. All elbows had canal preparation with cementing performed under tourniquet control. Cement restrictors were always used in the humerus. Formal cement restrictors were not utilized in the ulna. Low viscosity cement was placed with use of modern cementing techniques including retrograde fill and pressurization.
Fifteen of the implants were obtained during revision surgery for mechanical failure or bushing wear, and one implant that had functioned well was harvested with adjacent bone and soft tissues following the death of a patient for reasons unrelated to the elbow replacement. The implants were retrieved from patients treated at three different medical centers. Revisions for infection were excluded. The study was conducted under a protocol approved by the institutional review board.
All patients had successful pain relief and function initially, but symptoms and/or radiographic evidence of osteolysis subsequently developed or the implant failed in thirteen elbows. There were seven men and seven women, with an average age of fifty-three years (range, thirty-two to eighty-three years) at the time of the initial arthroplasty (see Appendix). Two patients had a bilateral elbow arthroplasty (Cases 1 and 10 and Cases 7 and 9; see Appendix). Eight patients had inflammatory arthritis, four had posttraumatic arthritis, and two had osteoarthritis. The time from the index procedure to implant retrieval averaged fifty-nine months (range, twenty-four to 142 months). Indications for revision surgery included pain, instability, loosening, and/or fracture.
Initial and serial radiographs were analyzed for all elbows, except for the specimen retrieved post mortem. For data collection, anteroposterior and lateral radiographs were divided into zones, analogous to the Gruen classification system for hip replacement and that reported for the Souter-Strathclyde elbow replacement (Fig. 1)22,23. Stems were divided into thirds, past the beaded or precoated periarticular area. Radiolucent changes were recorded at the cement-prosthesis and cement-bone interfaces, respectively. A diffuse pattern of radiolucency was defined as at least three zones with lucent lines on any radiograph. A focal pattern of radiolucency was defined as no more than two zones exhibiting radiolucent lines on any radiograph. Cement mantle quality was graded on the basis of the initial postoperative radiographs, as defined by Morrey et al.24. Alignment of the components relative to the humeral and ulnar axes was determined. The angular alignment of the implant stem was calculated by measuring the included angle between the axis of the implant stem and the long axis of its medullary canal25. As the proximal part of the ulna and the ulnar components are curvilinear, these measurements were made relative to the straight end of the stem tip and the ulnar diaphysis.
The Coonrad-Morrey total elbow prosthesis consisted of Ti-6Al-4V alloy humeral and ulnar stems linked by a two-piece axis pin made of either titanium or cobalt-chromium alloys that passed through two humeral and one ulnar polyethylene bushing. This semiconstrained articulation permitted approximately 7° to 8° of varus or valgus motion as well as some internal and external rotation, theoretically allowing absorption of stress by the implant rather than by the cement mantle. In all cases, the distal aspect of the humeral stems was circumferentially coated with titanium beads to enhance cement-implant bonding. The proximal aspect of the ulnar stems had been precoated circumferentially with polymethylmethacrylate, except for two implants (Cases 1 and 2), which had a circumferential titanium bead coating in this area.
The components available for examination included the ulnar stem alone from nine elbows, both the ulnar and humeral stems from four, and the humeral stem alone from one other elbow. Polyethylene ulnar bushings were retrieved from all sixteen elbow prostheses, and humeral bushings were available from fourteen. Axis pins were retrieved from all but four devices (Cases 11, 12, 13, and 15).
The retrieved humeral and ulnar stems, polyethylene bushings, and axis pins were examined for the presence of wear and deformation with the use of a stereomicroscope (model SMZ-U; Nikon, Tokyo, Japan) at eight to seventy-five times magnification. The modes of wear were categorized as described by McKellop (Table I)26. Damage to the metallic surfaces, such as longitudinal or transverse wear tracks or burnishing, was recorded. The polyethylene bushings were examined for asymmetrical thinning, pitting, delamination, fracture, and deformation. Care was taken to remove the stems without further damaging the polyethylene bushings. Specifically, for the humerus, the bushings were removed before any force was applied to the implant if it was required for removal. Similarly, for the ulna, no hook or device was placed within the open ring of the implant on top of the remaining polyethylene. If the implant stems were well fixed, osteotomies were utilized to facilitate implant removal. Damage to the stems from surgical instruments during removal was minor and easily distinguished under magnification from the finer markings characteristic of in vivo wear processes.
Specimens of periarticular implant-interface tissues were available from fourteen elbow replacements (Cases 1 through 7, 9, and 11 through 16). The nature of wear particles and the histopathological response to their presence were characterized with use of light microscopy of routine paraffin sections stained with hematoxylin and eosin. The histological sections were assessed qualitatively for the degree of cellular infiltrate (histiocytes, giant cells, lymphocytes, plasmacytes, and polymorphonuclear leucocytes) and for the presence of metallic and polyethylene particles. Polyethylene particles were identified with use of a polarizing light microscope in the routine stained sections. Additionally, unstained sections from selected elbows (Cases 1, 6, 11, and 14) were examined with use of a scanning electron microscope (model JSM-6460LV; JEOL, Peabody, Massachusetts). Backscattered electron imaging was used to localize and determine the size of metallic particles. The elemental composition of individual metal and bone-cement particles was determined with use of energy-dispersive x-ray analysis.
Clinical
Nine patients had an insidious onset of pain develop with activity. Eight had evidence of mechanical changes with audible or palpable crepitation during elbow flexion. Four individuals presented with acute pain and dysfunction, two in the setting of a humeral stem fracture, one with acute failure of the axis pin-polyethylene articulation, and one in which both the humeral and ulnar prostheses were grossly loose. The implant retrieved from one patient at the postmortem examination had functioned well up to the time of death.
Radiographic Alignment
Most prostheses exhibited some deviation from perfectly neutral alignment as measured radiographically. Varus alignment of the humeral stem in the coronal plane was observed in six elbows (average, 4°), and valgus alignment was seen in two (4° in both). In the lateral plane, nine humeral components were flexed (average, 6°) and the remaining six were in a neutral position. Four ulnar components were in varus alignment (average, 4°), ten were in neutral, and one component was in 10° of valgus alignment. In the sagittal plane, four ulnar implants were extended (average, 6°), seven were in neutral, and four were in flexion (average, 6°) relative to the ulnar axis. Only two elbows had neutral humeral and ulnar stem alignment on both radiographs.
Radiographic Quality of the Cement Mantle
Initial radiographs revealed twelve humeral components with cement mantles classified as adequate and three as marginal, as defined by Morrey et al.24. Fourteen ulnar mantles were classified as adequate and one as marginal. No initial cement mantles were graded as inadequate.
Radiographic Radiolucent Lines
Immediate postoperative anteroposterior and lateral radiographs revealed humeral cement-bone radiolucent lines in four of fifteen elbows and ulnar cement-bone lines in four elbows. One elbow initially had radiolucent lines between the humeral prosthesis and the cement mantle, and three patients had similar findings in the ulna. There was no pattern to or prevalence of the initial radiolucent lines as defined by radiographic zones.
Analysis of radiographs prior to revision revealed humeral cement-bone interface lines in thirteen of fifteen elbows and ulnar cement-bone lucencies in all elbows. Radiolucent lines between the cement and the prosthesis were observed in the humerus in twelve of fifteen elbows and in the ulna in all specimens.
Several patterns of radiolucencies were noted when serial radiographs were analyzed (Figs. 2-A and 2-B). At the bone-cement interface, six of fifteen humeral stems had a diffuse pattern of lucency, while seven had a more geographic, focal distribution, and two had no radiolucent lines. Of the focal patterns, five stems had lucencies only in the periarticular area (zones 1 and 2) adjacent to the hinge mechanism, one had periarticular (zone-1) and stem tip (zone-4) lucencies, and one had only stem tip (zone-4) lucencies. In the ulna at the bone-cement interface, eight of fifteen stems had diffuse changes, while seven had a more focal pattern. Of the stems with focal patterns, four had isolated periarticular lucency (zones 1 and 2), two had lucencies only at the stem tip (zones 3 and 4), and one had a lucency both in the periarticular region (zone 1) and at the tip (zone 4).
At the cement-prosthesis interface, three of fifteen humeral components had diffuse, progressive lucencies, while eight had more focal changes. Four components had isolated periarticular lucencies (zones 1 and 2), one had changes only at the tip (zone 4), two had changes at both the periarticular region and at the tip, and one had changes centrally (zones 2 and 3). In the ulna at the cement-implant interface, ten of fifteen components had diffuse changes. Three elbows had periarticular lucencies alone (zone 1), while in two elbows, radiolucent changes were only observed in the central stem region (zones 2 and 3). Both of these elbows had an ulnar component with a titanium beaded surface on the proximal end (a fourth-generation prosthesis).
Intraoperative and Postmortem Findings
In nine of the fifteen elbow joints retrieved at revision and in the elbow joint obtained post mortem, the periarticular soft-tissues were grossly gray or had black discoloration, representing the presence of metallic debris. At revision, eleven of the thirteen ulnar components were grossly loose, with one that was fractured but well fixed distally. All humeral components were well fixed except for the distal segment of the three fractured stems.
In the joint harvested post mortem, both the ulnar and humeral components were stable on gross examination. Evaluation of specimen contact radiographs revealed adequate cement mantles in both the humerus and the ulna. No bone-cement or cement-prosthesis radiolucencies were identified.
Implant Analysis
All of the retrieved components exhibited multiple modes of wear. On the humeral and ulnar polyethylene bushings, both mode-1 wear, which represents particle generation from expected motion between two bearing surfaces, and mode-3 wear, from third-body interposition between two bearing surfaces, were observed in nearly all cases. Twenty-seven of twenty-eight humeral polyethylene bushings exhibited asymmetrical thinning. In the ulnar component, fifteen of sixteen bushings had shape deformation from round to elliptical, allowing proximal excursion of the ulnar stem (Fig. 3). In the humeral component, wear was noted along the medial and lateral walls of the polyethylene that articulate with the ulnar component base. Pitting, fracture, embedded metal and/or cement particles, and shape deformation were common findings in the polyethylene bushings.
Mode-2 wear, or generation of particles from contact between a bearing surface and a nonintended articulation, was seen in seven of thirteen implants. All such wear occurred between the metal axis pin and the metallic ulnar base of the prosthesis (Figs. 4-A and 4-B) because of the aforementioned shape change and wear of the ulnar polyethylene.
Degradation of the polyethylene bushings also led to increased generation of metallic particles through substantial unintended contact and metal-on-metal wear of humeral and ulnar component nonbearing surfaces (mode-4 wear) adjacent to the intended articulation. Metal-on-metal wear between the proximal part of the ulnar component and the humeral component was seen in eleven of fourteen prostheses because of asymmetrical polyethylene wear and deformation. Additional mode-4 wear due to proximal excursion of the ulnar component within the deformed polyethylene bushing was observed in three elbows (Cases 6, 14, and 16), where there was articulation of the humeral condyles against cement in the ulna and metal-on-metal wear between the proximal part of the ulnar component and the base of the humeral component (Fig. 5).
Other sites of mode-4 wear due to contact between two nonbearing surfaces included longitudinal wear tracks on the ulnar component stems, consistent with axial pistoning within the cement mantle. These were seen in most ulnar prostheses, as would be expected from a loosened component. Transverse wear tracks, indicating rotational instability, were less frequently seen in the ulnar components and were not observed in the humeral implants. Loss of surface titanium beads was noted in three of five humeral stems, and loss of the polymethylmethacrylate precoating from the ulna was observed in all ten of the loosened components. There was no indication of wear on either the humeral or the ulnar stem of the implants removed from the postmortem specimen, in which neither component had loosened.
Histological Findings in the Periarticular Implant-Interface Tissues
The overall cellular response to particulate wear debris in the periarticular implant-interface tissues and the composition and size of the wear particles were similar in most cases. The synovial tissues and implant interface membranes showed chronic inflammation, fibrosis, and focal necrosis associated with abundant intracellular metallic and polymeric particles. There were a marked number of wear particle-laden histiocytes and a moderate number of multinucleated giant cells. Lymphocytic and lymphoplasmacytic infiltrates were minimal in eleven elbows (Cases 1 through 4, 6, 9, and 11 through 15) and absent in one (Case 5). In the periprosthetic tissues of two elbows (Cases 7 and 16), there was a moderate to marked lymphoplasmacytic infiltrate accompanied in the latter case by a lesser number of eosinophils. Polymorphonuclear leukocytes were rare in the periprosthetic tissues of the fourteen implants for which tissue was available.
Particulate Debris
Titanium alloy wear particles were the most abundant particulate in the periarticular implant-interface tissues. The metallic particles appeared as opaque refractile granules and shards. Most of these particles were submicrometer in size and ranged from 0.5 to 30 µm in greatest dimension by light microscopy. Examination with use of the scanning electron microscope revealed abundant titanium alloy particles (Fig. 6) as small as 0.1 µm, which was the approximate detection limit with use of this technique. The metallic particles were present both in histiocytes and in multinucleated giant cells, often intermixed with particles of polyethylene.
Polyethylene wear particles were nearly equally as prevalent as the metallic particles by light microscopy, and they exhibited a wide range of particle morphology and size. Polyethylene particles included granules =1 µm in size, fibrils measuring 2 to 10 µm, and flakes and shreds ranging from 10 µm to several hundred micrometers in size. The majority of the polyethylene particles were present within histiocytes. The largest particles were within or surrounded by multinucleated giant cells.
Bone-cement particles were the least apparent of the prosthetic wear debris in the periprosthetic tissues, although this was likely an artifact of specimen processing. No polymethylmethacrylate remained in the histological sections following routine processing of the tissues. However, the previous sites of large pieces of cement were evident by remaining empty spaces, either within multinucleated giant cells or rimmed by them. Some of these spaces contained remaining aggregates of barium sulfate particles, the radiographic contrast medium of the cement. Individual particles of barium sulfate (0.2 to 0.8 µm in size) were also detected within histiocytes with use of electron microscopy with energy-dispersive x-ray analysis. In the tissues adjacent to one implant (Case 11), barium sulfate, rather than titanium or polyethylene, was the most prevalent particulate.
To better understand mechanisms of failure of total elbow arthroplasty components and to correlate these with clinical and radiographic findings, we examined the modes of wear of Coonrad-Morrey prostheses retrieved at revision surgery that had failed relatively early, at a mean of five years after implantation. The Coonrad-Morrey total elbow prosthesis is one of the most commonly used semiconstrained implants. The implant has been reported to have a survival rate of 92% to 94% at five to ten years and 68% at up to thirty-one years3,27. The design has undergone multiple changes since its introduction in 1972 in an effort to improve longevity (see Appendix)10,17,28,29. Most (thirteen of sixteen) implants in our study were either fifth or sixth generation, which differed principally in that the axis pin was changed from titanium alloy to cobalt-chromium alloy.
Although our study was limited by its retrospective nature and a relatively small number of devices that had been implanted at three different medical centers, we were, nonetheless, able to identify all four modes of mechanical wear as described by McKellop26. Mode-1 direct bearing wear was observed in both the humeral and ulnar polyethylene bushings, apparently by different mechanisms. The humeral polyethylene showed asymmetrical wear and deformation medially and laterally, most likely due to torsional or varus-valgus loading. While some similar changes were seen in the ulnar polyethylene bushings, fifteen of sixteen bushings revealed distortion into a more elliptical shape, consistent with axial loading. Mode-2 wear between articular and nonarticular surfaces was observed between the ulnar base and the axis pin. Mode-3 third-body wear was evident by the presence of metal and cement debris at the polyethylene bearing surfaces. Mode-4 wear between two nonarticular surfaces was observed between the metal surfaces of the implants and between the cement mantle and the implants. Thus, multiple potential modes of wear that may contribute to particle generation, osteolysis, and mechanical implant failures were identified.
Since surgical technical factors, including the adequacy and symmetry of the cement mantle and the alignment of the prosthetic components, play a large role in the development of osteolysis and aseptic loosening in total hip and total knee replacements, these factors were examined in the present study. Radiographic analysis showed that the initial cement mantle was marginal in only four of thirty stems, and no cement mantle was inadequate as defined by Morrey et al.24. Therefore, within the limitation of our ability to assess the status of the cement mantle radiographically, this factor did not seem to play a role in the failure of the total elbow replacements in this series. Similarly, while some minor radiographic evidence of malalignment from neutral was seen in thirteen elbows, with six of fifteen humeral stems being in varus, this malalignment averaged only 4° and could not be associated with a specific pattern of wear in this small series. Another study of Souter-Strathclyde total elbow arthroplasty has also suggested that implant alignment may play only a minor role in the development of aseptic loosening30.
On the other hand, alignment of total elbow arthroplasty implants within the medullary canals has been shown to potentially affect outcome23,25. For example, it has been shown that if the humeral component of the Coonrad-Morrey prosthesis is inserted in >10° of malrotation, the varus and valgus laxity during the flexion and extension arc follows the structural limits of the hinge31. This removes the theoretical advantage of the sloppy hinge mechanism and may lead to increased stresses on the polyethylene and the bone-cement interface. A varus inclination of this implant may occur on the basis of the humeral stem design31.
We were unable to measure torsional malalignment in this series. Theoretically, torsional malalignment could have contributed to the mechanical wear patterns observed. The components in our study were cemented separately in ten of the fourteen elbows for which this information was available. It may be that cementing these components separately rather than assembled as a unit can contribute to rotational malalignment. We noted evidence of edge-loading and asymmetrical thinning of the polyethylene bushings in fifteen of sixteen ulnar and in twenty-seven of twenty-eight humeral bushings retrieved, suggesting that the implants were subjected to asymmetric forces that could have occurred from subtle stem malalignment. Others have also commented on edge-loading and polyethylene wear with respect to failure of the axis-pin locking mechanism20. In addition, ligamentous instability, which we were not able to assess in this series of retrievals, could also be a factor in the magnitude and patterns of wear.
Periprosthetic osteolysis is generally believed to represent a biologic response to wear particle debris14 and may present either as cystic or linear radiolucencies in focal and diffuse patterns. Progressive radiolucent lines were identified in all elbows that had mechanical failure in this series. While these were seen in the humerus at both the bone-cement and cement-prosthesis interfaces, none of the humeral component stems were found to be loose at revision surgery. Extensive lysis in the periarticular region (zones 1 and 2) may be associated with humeral stem fracture by a cantilever bending mechanism22, as evidenced by the three humeral stem fractures in this series. Two elbows with an ulnar stem with a proximal circumferential beaded surface (fourth generation) exhibited a unique radiographic pattern with progressive radiographic changes only in the central area of the stem (zones 2 and 3). One stem ultimately fractured in this area.
While approximately half of the progressive bone-cement radiolucencies were diffuse in nature, others were more discrete. Five of the thirteen humeri with progressive bone-cement changes had radiolucencies isolated to the periarticular areas (zones 1 and 2). Three of these stems ultimately fractured. Four ulnae had progressive osteolysis develop in the periarticular areas and specifically at the stem tip (zone 4) in three. Osteolysis at the ulnar stem tip was associated with periprosthetic fracture in three elbows. While focal osteolysis at the ulnar stem tip has previously been described with this device, it may indicate a guarded prognosis on the basis of our findings29. In fact, all ulnar components in this series with progressive osteolysis were found to be clinically loose at the time of revision. While the retrospective nature of our study only allows for limited conclusions in this regard, osteolysis around the ulnar component may be a potential harbinger of mechanical implant failure. Wright and Hastings recommended synovectomy and early bushing exchange in any patient with evidence of new onset synovitis, bushing wear, or osteolysis20. We believe that careful, scheduled clinical and radiographic evaluation of patients with a total elbow replacement is mandatory to identify those at risk for sequelae of osteolysis so that timely, appropriate intervention can result in a less extensive surgical revision, longer retention of the implants, and preservation of the cement mantle and bone stock.
We surmise that excessive mode-1 wear and deformation of the polyethylene may be an initiating cause of the other modes of wear identified in this study. Excessive polyethylene mode-1 wear and deformation due to creep eventually allow the axis pin to contact the inner surface of the metal ulnar base abnormally, causing mode-2 wear. The metal particles generated from this mode-2 wear can act as third bodies that can become trapped between the primary bearing surfaces and cause mode-3 wear. Because of a narrow clearance between the distal humeral trochlea and the proximal ulnar base, as the proximal part of the ulna translates proximally from the deformation of its polyethylene, mode-4 metal-on-metal wear can occur, with the generation of substantial metallic debris (Fig. 5). Mechanisms similar to this have been proposed to explain bushing and axis pin failure, emphasizing that this phenomenon is not unique to the Coonrad-Morrey prosthesis19,20,32.
In the present study, examination of the retrieved components clearly documented polyethylene bushing degradation, which can lead to substantial unintended contact of the humeral and ulnar component surfaces and metal-on-metal wear particle generation. However, this study is an observational analysis of wear modes on retrieved implants and cannot definitively determine the primary cause of the observed wear patterns on the implants or the exact source of the particulate debris. The majority of the ulnar implants in this study were of the polymethylmethacrylate precoat design, which appears to have had suboptimal cement-prosthesis bonding as evidenced by the linear longitudinal wear tracks and grossly loose stems at the time of revision surgery. These loose ulnar stems may generate substantial metal and cement particles that could become third bodies that subsequently accelerate polyethylene bushing wear and allow the unintended ulnar base and distal humeral contact. Lee et al.12 recognized that a loose stem can generate metallic and cement particles, leading to osteolysis in the absence of polyethylene wear. Conversely, primary polyethylene bushing wear and deformation due to creep can be the principal cause of metal debris. This was demonstrated in the postmortem elbow replacement specimen. Although neither the humeral nor the ulnar component in that elbow had loosened, the tissue surrounding the prosthetic articulation was black and gray-stained on gross examination, with abundant intracellular titanium-alloy debris seen on histological analysis, as a result of humeral and ulnar component contact and metal-on-metal wear. In one long-term clinical series, total polyethylene bushing wear was seen in 7% of the implants and partial bushing wear in 8% at ten to fifteen years3. A more recent comprehensive review of 919 total elbow replacements revealed a 1.3% rate of bushing exchange for isolated bushing wear12.
Thus, in our study, not only polyethylene particles but also titanium-alloy particles, with nearly equal prevalence, were major constituents of periprosthetic wear debris, inciting the florid histiocytic reaction observed in the histological sections of all of the elbows for which these tissues were available, including several elbows without humeral or ulnar component loosening. Similarly, in conventional metal-on-polyethylene bearings of hip and knee replacements, polyethylene wear particles often predominate26, but metallic wear debris can have an equal or greater presence in the periprosthetic tissues and systemically when unintended wear modes occur33,34.
The titanium-alloy particles generated included 0.1 to 5-µm-sized granules and angular shards of up to 30 µm. Polyethylene particles included granules, fibrils, flakes, and shreds, reflecting the types of wear mechanisms occurring in total elbow bearings. Polyethylene granules and fibrils from <1 to 10 µm in size are related to wear mechanisms and are thought to be most stimulatory to macrophages14,26,35-38. Flakes and shreds measuring from >10 to several hundred micrometers usually result from surface fatigue and abrasive wear mechanisms, respectively26,35,36,38. In this regard, the bearing surface wear mechanisms and the nature of the resulting particles around total elbow prostheses resemble those encountered in total knee replacements, which have higher contact stresses, rather than the polyethylene bearings in hip replacements, in which smaller particulates due to adhesive wear predominate26,35,38. All of these various polymeric and metallic particulates contribute to the formation of particle-induced granulomas and the potential for periprosthetic osteolysis14,39. The importance of titanium alloy particles relative to polyethylene debris should not be underestimated. In vitro studies have shown titanium alloy particles to be even more stimulatory to macrophages in releasing proinflammatory cytokines39, which is believed fundamental to the process of periprosthetic osteolysis.
As is the case with total hip and knee replacement, the complications of osteolysis, aseptic loosening, and fracture (prosthetic and periprosthetic) in total elbow replacement can be mitigated by evolutions in design, materials, and surgical technique. Improved bearing materials, developed by validated laboratory simulations of total elbow replacements (i.e., where laboratory simulations reproduce the wear patterns observed on retrieved implants), will likely diminish the magnitude of debris generation. Particular attention should be directed at the design and geometry of the polyethylene bushings, which are arguably the weak link in the total elbow tribological system. There are newer elbow arthroplasty designs that seek to replicate more closely the flexion-extension axis of the elbow, have larger bearing surfaces, and allow for the option of radial head replacement, which may alter the incidence of loosening, but none have been studied as systematically as the Coonrad-Morrey prosthesis or have been used clinically for a sufficient period of time to know if there are any differences in wear rates, modes of wear, or aseptic loosening rates.