➢ Higher rates of failure have been reported for metal-backed glenoid components, and cemented pegged polyethylene components currently appear to be the component design of choice.
➢ So-called modern cementing techniques have reduced postoperative radiolucent lines and may improve glenoid implant longevity.
➢ Compensating for altered glenoid morphology and bone loss continues to be a challenge.
In the early 1970s, Dr. Charles Neer introduced the first glenoid component. The final design was an all-polyethylene component that was keeled and cemented, with conforming humeral and glenoid radii of curvature. In 1982, Dr. Neer was the first author to report successful pain control following total shoulder replacement surgery1.
Glenoid replacement has improved patient satisfaction, pain control, and clinical outcomes (Fig. 1)2,3. However, to this day, there remains concern regarding the longevity of the glenoid component, which may be considered the weak link of a total shoulder replacement. Glenoid loosening has been reported as the most common reason for revision total shoulder replacement4-8. In one series, forty-four (59%) of seventy-four failed total shoulder arthroplasties were associated with unstable glenoid components9.
The detection of radiolucent lines surrounding the glenoid component has been used as an indication of glenoid component failure. Radiolucent lines surrounding a glenoid component have been reported as early as immediately postoperatively, with the reported prevalence increasing over time to as high as 90%6,8,10,11. However, the correlation of radiolucent lines with clinical loosening is uncertain, and the rate of revision due to loss of fixation has only been reported to be between 0% to 7%4,7,8,11-13. The survival of the original Neer cemented all-polyethylene component has ranged from 83% to 93% at ten years and from 73% to 87% at fifteen years6,7.
Despite being characterized as a non-weight-bearing joint, the glenohumeral joint is subject to forces that may exceed body weight. Plastic deformation occurs in a portion of glenoids that are later revised14. In an investigation involving instrumented joint replacements in four patients, Westerhoff et al. reported that peak glenohumeral contact forces averaged 76% body weight during hair combing and 123% body weight while setting down a coffee pot15. Furthermore, these forces often are transmitted between the humeral head and the glenoid at eccentric locations on the glenoid surface. With use of biplanar fluoroscopy combined with three-dimensional image registration in patients managed with total shoulder arthroplasty, Massimini et al.16 reported that the joint contact center was located an average of 11 mm from the glenoid center for various abduction and rotational arm movements. Forty (62%) of the sixty-five analyzed arm positions resulted in contact points that were located in the superior-posterior quadrant of the glenoid articular surface. These recent clinical data on mechanics have confirmed and elaborated on some earlier cadaveric17,18 and computer modeling estimates19-21 and support the widespread belief that the primary mechanism of glenoid loosening involves repeated, eccentric forces of the humeral head on the glenoid rim, so-called rocking-horse loading (Fig. 2)22-24.
Resistance to cemented glenoid implant loosening relies on a mechanically strong bond between the implant and glenoid bone (Fig. 3). There is a limited amount of glenoid vault bone in which to fasten the component. At 15 mm deep, the vault narrows to an average of only 10 mm wide, with some vaults measuring <6 mm wide25,26. The vault has a retroverted angulation27, with the center of the vault being, on the average, slightly anterior to the center of the glenoid face when looking down the scapular axis26.
This review will focus on evolutionary changes in glenoid component design as they relate to conformity, surface shape, and fixation designs specifically related to cemented, uncemented, and hybrid designs as well as the difficulties in dealing with the morphology of the glenoid. The rationale and impact on operative technique also will be highlighted.
Conformity and Surface Shape
Total shoulder arthroplasty conformity refers to the match or mismatch between the radii of curvature of the humeral and glenoid articulating surfaces. The ideal conformity continues to be an area of debate. A nonconforming design with a relatively larger glenoid radius allows for more normal glenohumeral translations during joint rotations28, and nonconforming designs appear to have more clinical evidence to support their use. The disadvantages include higher contact pressures and potential material yielding as demonstrated by analytical and computational models29,30 as well as instability31. A conforming design, used in many early glenoid components, theoretically results in greater distribution of contact pressure and stable ball-joint-like motions. However, a conforming design is predicted to result in slightly more volumetric wear in association with the much larger contact zone30. In addition, conformity has a complex effect on the loads and stresses imposed on the implant fixation bond32-34. Component surgical positioning is likely to be more critical for conforming glenoid components35, with malpositioned conforming components being vulnerable to rim loading with large resultant side loads and resulting high stresses on the fixation. In a large multicenter clinical study, nonconforming glenoid components with a radial mismatch of >5.5 mm led to fewer glenoid radiolucencies in comparison with their conforming counterparts36. Conforming polyethylene glenoid components that have been retrieved at the time of revision surgery have demonstrated greater abrasion and delamination, whereas nonconforming components have shown more burnishing damage37. Interestingly, as a result of large joint forces, glenoid components can plastically deform and can alter conformity characteristics while in use in vivo14. A hybrid glenoid articular design that includes a conforming central zone and a nonconforming periphery recently has been in use38. In addition to conformity, the wall height of the glenoid influences stability.
There are two main surface shapes for glenoid components: (1) anatomic and (2) oval. There is a paucity of literature to support the use of one design over another. Tammachote et al.39, in a biomechanical model, analyzed the stability ratios for three different glenoid component sizes; small, medium, and large. They found that a modest increase in stability was seen when a larger glenoid component was chosen when the glenoid was between two component sizes. The generally accepted concept is to choose a component (either anatomic or oval) that will fill the glenoid surface but will not allow for overhang.
Glenoid component fixation, in general, can be divided into three main categories: cemented, cementless, and hybrid40. The cemented glenoid component design remains the standard for total shoulder arthroplasty. In general, cementless glenoid component designs initially showed promise, but longer-term follow-up has shown higher failure rates, leading most surgeons to believe that this design is no longer acceptable. As concern regarding glenoid component longevity remains, newer hybrid designs for fixation are evolving.
Cemented designs typically consist of an all-polyethylene component based on the original design described by Neer et al.1. The original backside design of this component was a curved, convex design. There has been some exploration into the concept of an all-polyethylene flat-back design. In a biomechanical study, worse loosening performance was reported in association with a flat-back design41. Szabo et al., at the time of early (two-year) follow-up, reported that the keeled convex-back glenoid component was radiographically better than the keeled flat-back component42. However, in a longer-term prospective trial, there was no significant difference between the two designs in terms of the rates of revision or radiolucent lines43.
Fixation of the glenoid component is most commonly achieved with use of polymethylmethacrylate bone cement. The cement interdigitates into the trabecular bone of the glenoid vault, forming a complex mechanical interlock (Fig. 3) and, theoretically, the more interdigitation, the better the strength of the fixation. In the original technique described by Neer et al.1, the cement was hand-mixed and was manually inserted with finger packing. Today, the so-called modern cementing technique is not standardized, but studies have shown reduced rates of radiolucent lines. Special instrument preparation and pressurization44-47, the weep hole technique (in which a vent is created in the coracoid process to remove fluid so that it is not trapped at the cement interface)48,49, and drying the glenoid with adrenaline or thrombin-soaked gauze45 are a few techniques that have been described.
Edwards et al.50 compared three different techniques of cement preparation: (1) thrombin-soaked gel foam, (2) compressed gas lavage, and (3) saline solution lavage with sponge drying. They showed no differences among the three techniques other than cost. Edwards et al.50 and other groups have highlighted the importance of using specialized instruments and reamers to perfectly match the shape of the back of the glenoid component and compaction of the bone surrounding the peg or keeled design to prevent toggle42,44,46. Earlier glenoid component-preparation techniques involving freehand shaping of the glenoid and the use of curets to create the slots for the component have shown poorer results1,42.
Although we have improved our cementing techniques, there is still concern regarding the progressive development of radiolucent lines around the implant8,10,51. A more recent long-term study of 226 total shoulder replacements52 demonstrated that the survival rate for a flat-back cemented polyethylene component, with revision as the end point, was 99% at five years, 95% at ten years, and 79% at fifteen years. The survival rate with radiographic loosening as the end point was 99% at five years, 80% at ten years, and 34% at fifteen years. In another study, a keeled convex-back glenoid component demonstrated five and ten-year survival rates of 99.7% and 98%, respectively, with revision as the end point and of 99.7% and 52%, respectively, with radiographic loosening as the end point53.
One theory for the development of radiolucent lines is that exothermic bone necrosis occurs in the bone immediately adjacent to the prosthesis, thus weakening the cement-bone interface54,55. Olson et al.56 reported thermal necrosis in association with just glenoid preparation and drilling and recommended the use of frequent irrigation to prevent temperatures from reaching necrosis thresholds. In contrast, Raiss et al.57 reported that the temperatures that occur during glenoid implantation are low and are not high enough to create surrounding osteonecrosis.
Glenoid fixation failure also can occur at the implant-cement interface. In a recent biomechanical experiment in which twenty fixation designs were implanted into synthetic bone and were exposed to cyclic superior loads, components failed at the implant-cement interface, not the cement-bone interface, and failure initiated at the inferior part of the component fixation58. In a study from the same institution, Gregory et al.23 cyclically loaded glenoid components that had been inserted with cement in cadavers and analyzed fixation status with use of computed tomography (CT) scanning. The authors reported that the cement-bone interface remained intact and that the cement fracturing took place at the implant-cement interface, again suggesting that future modifications of all-polyethylene components should focus on strengthening the bond at the implant-cement interface.
Keeled versus Pegged Designs
Vavken et al., in a recent meta-analysis of 1460 patients, showed that pegged glenoid components were associated with less frequent loosening and revision in comparison with keeled components59. However, the authors also reported that the differences were small. Originally, all-polyethylene components had a keeled design, and there has been a trend toward viewing the pegged component as the preferred polyethylene design concept.
“Better cementing” and fewer radiolucent lines have been reported in association with pegged glenoid components as compared with keeled components44,60-62. It also has been reported that incomplete seating is more common in association with keeled components62. Throckmorton et al.63, however, reported no difference in clinical and radiographic outcomes between pegged and keeled components after four years of follow-up.
Lacroix et al.64 performed a finite-element analysis of keeled and pegged anchorage systems and concluded that the pegged system was superior for normal bone, whereas the keeled system was better for rheumatoid bone. In a mechanical test involving synthetic bones, Anglin et al.41 found that a pegged component outperformed a keeled component design in terms of edge displacement following 100,000 cycles of loading.
Nuttall et al.65 used a novel radiostereometric analysis (RSA) technique to measure the position changes of ten keeled and ten four-pegged components. The authors reported that all components moved but noted increased migration, over two years, in association with the keeled components. The components that required corrective reaming because of erosion showed greater change in alignment over two years.
Walch et al.66 highlighted the importance of preserving the subchondral bone in order to improve glenoid component longevity and to minimize component migration. In a multicenter study involving 518 cemented, all-polyethylene keeled glenoid components that were followed for a minimum of five years, the authors found that 32% of the components had evidence of loosening and that 26% of the components had migrated. There were three patterns of migration: superior tilting (10%), component subsidence (8%), and posterior tilting (6%). The authors believed that the migration patterns were related to removal of the subchondral bone due to excessive glenoid preparation in an attempt to obtain a perfect fit to the back side of the glenoid component and to realign the retroverted glenoid. The posterior tilting was correlated with preoperative static posterior subluxation with heightened retroversion and biconcave glenoid wear patterns. Unbalanced soft-tissue tension was suggested as one possible source for the failure. Consideration of posterior bone-grafting was suggested as a potential solution to realign the glenoid, to retension the soft tissues, and to preserve the subchondral bone. There is initial biomechanical evidence that a novel posteriorly augmented glenoid component also may be reasonable in these circumstances67. Current glenoid component designs may need to respect the subchondral bone, particularly superiorly and posteriorly.
A variation of the traditional pegged component is one that has a central fluted uncemented peg surrounded by three shallow peripheral pegs that are cemented to provide initial stability to the component. Wirth et al.68 used a weight-bearing canine model to evaluate the performance of this component and compared it with a standard keeled component design. The authors showed that the cementless fluted peg design was superior to the keeled design for achieving osseous integration and fixation. Clinical follow-up studies at two to five years have shown low rates of loosening69-71. A low rate of radiolucent lines was related to osseous integration within the fins of the central pegs, regardless of whether bone-grafting was used between the fins at the time of implantation. To our knowledge, no long-term studies are yet available for this implant design.
A cementless glenoid implant is, in general, composed of two pieces, including a metal back and a polyethylene inlay. This type of implant is typically fixed to the glenoid vault with use of screws. The use of an ingrowth glenoid component is appealing from the theoretical standpoint that solid integration, if achieved, will allow for increased glenoid fixation longevity. Unfortunately, the survival of cementless metal-backed components has been inferior to that of cemented all-polyethylene glenoid components4,72-77. Wallace et al.72, in what we believe to be the first study comparing cemented and uncemented glenoid components, reported a two-times greater rate of revision and complications in the uncemented group. Soon thereafter, Boileau et al.73 performed a randomized prospective study of thirty-nine patients in which cemented all-polyethylene glenoid components were compared with cementless metal-backed components. At three years of follow-up, the all-polyethylene glenoid components were associated with a significantly higher rate of periprosthetic glenoid radiolucent lines (p < 0.01), but this finding was not correlated with functional scores (p = 0.3). Although radiolucent lines were less common in association with the metal-backed components, when present, they were highly progressive, leading to a 20% rate of glenoid revision in this group. The radiolucent lines around metal-backed glenoid components also were related to severe osteolysis and component shifting, deteriorating functional results, and increasing pain. Martin et al.4 reported on 147 total shoulder replacements, at an average of 7.5 years of follow-up, that were performed with use of an uncemented glenoid component with a plasma-sprayed metal keeled design that featured two screws above and below the keel for fixation. The authors reported an 11% rate of clinical failure, with a ten-year survival rate of 85% and a radiographic survival rate of 76%. Radiographic failure was found to be a sevenfold predictor of clinical failure. Fractured screws and glenoid trays and excessive polyethylene wear were common modes of failure. Fox et al.75, in a study of 1542 shoulders, compared six different implant types, including four all-polyethylene cemented components and two metal-backed components. Survival rates were higher for the all-polyethylene components as compared with the metal-backed components.
A modification of the traditional metal-backed designs—specifically, an uncemented, metal-backed bone-ingrowth design with a central cage screw—was discontinued because of its high rate of glenoid failure at five years78. Montoya et al.78 reported an 11% revision rate after a mean duration of follow-up of sixty-eight months. Five of the six patients who required revision had evidence of breakage of the central cage screw, and four of these five patients had a type-B1 glenoid deformity.
One metal-backed glenoid component design has shown encouraging results. This component is a convex-back, porous-coated, metal-backed component with a large hollow central stabilizing peg and is fixed with two screws above and below the peg. The revised implant design is 7.5 mm thick with a 4.5-mm polyethylene insert. Rosenberg et al.79 reported a four-year implant survival rate of 93% (thirty-two of thirty-four). Castagna et al.80, in a study of thirty-five shoulders, reported no revisions at six years and a 23% rate of periprosthetic radiolucent lines that did not progress after two years. Clement et al.81 reported a ten-year implant survival rate of 86% (thirty-one of thirty-six) with revision as the end point.
Polyethylene components are more compliant and conforming than metal-backed components, leading to a different load transfer from the implant to the periprosthetic cortical and trabecular bone. Several research groups have used finite-element computer analysis to compare all-polyethylene components with metal-backed components64,82-84. There is general consensus that all-polyethylene components result in an overall bone stress pattern that is more similar to that of an intact glenoid as well as in less stress-shielding in the trabecular bone under and around the prosthesis. Local bone resorption surrounding an uncemented metal-backed implant has been predicted with use of bone-modeling simulation85. Higher glenohumeral contact stresses (with presumably increased wear) have been predicted for metal-backed components29, and high-stress regions have been predicted within the polyethylene near the polyethylene metal interface83.
A number of related and additional reasons have been given for the failure of metal-backed glenoid components4,73,75,78, including (1) overstuffing of the joint due to the increased thickness of the overall glenoid component needed to accommodate the metal plate and the polyethylene insert, leading to excessive polyethylene wear and loosening, (2) insufficient thickness of the polyethylene insert (measuring only 4 mm with early designs), potentially predisposing the insert to delamination and eventually increased debris and leading ultimately to the development of metal-on-metal debris, and (3) rigidity of the metal backing due to the large difference in modulus of elasticity between metal and bone, leading to early dissociation of the polyethylene and early accelerated wear73. The quality of initial screw fixation also has been suggested as a potential design flaw and another source of metal debris if the screws loosen or break4.
A recent study demonstrated good short-term clinical results for thirty-four patients who were managed with an uncemented all-polyethylene component with three peripheral small pegs and one central anchor peg86. The central anchor was fluted and packed with bone derived from the reaming. Radiolucent lines were noted in four of the thirty-four shoulders, but this finding was not associated with poorer clinical outcomes.
Hybrid designs are evolving, and there are limited reports on their use. Fucentese et al.87 reported the results associated with a soft-metal-backed glenoid component. The four-pegged all-polyethylene glenoid component was coated with multiple layers of unalloyed titanium mesh that was welded together with an average pore size of 400 µm and had an average porosity volume of 65%. Evaluation of the results for twenty-two patients at a minimum of two years (mean, fifty months) showed a high failure rate (14%) because of implant fractures at the connection between the pegs and component body. The authors reported that loosening appeared to be minimal and that osseointegration was possible. Budge et al.88 performed a biomechanical analysis in which a press-fit, porous tantalum-backed, keeled polyethylene glenoid component was compared with an all-polyethylene cemented component. The authors concluded that the initial fixation of the all-polyethylene cemented glenoid component was superior to that of the porous tantalum-backed component. They suggested that one potential reason for the poorer performance was related to the porous tantalum being raised off the polyethylene back side, which may not allow for a proper “contact profile” on the back of the implant. Budge et al. found that eighteen of nineteen glenoid components had complete ingrowth of the porous tantalum keel at two years; however, four implants (21%) failed as a result of a fracture at the keel-glenoid face junction89. The authors reported that design modifications are required to improve the strength of the polyethylene and porous ingrowth junction.
Addressing Glenoid Bone Deformity
Resurfacing of the glenoid in osteoarthritic patients with severe anteroposterior deformity (Fig. 4) and posterior subluxation continues to be a clinical challenge. In such patients, most surgeons attempt to realign the glenoid toward a neutral version or slight retroversion, although the degree of necessary correction continues to be an area of controversy. Clinical assessments of the effect of component version on outcome have been scarce because of the difficulty of assessing postoperative version90.
Cadaveric and computer-modeling studies have been used to demonstrate the potential effects of excessive retroversion on stability, eccentric glenoid loading, fixation stresses, and predicted glenoid loosening17,18,21,91. A biomechanical study of eight cadaveric shoulders showed that 15° retroverted glenoids produced larger contact pressures than neutrally oriented glenoids during horizontal adduction17 and a finite-element computational analysis of one shoulder showed a threefold increase in cement stresses with the glenoid in 20° retroversion21. A limitation of those studies is the uncertainty regarding soft-tissue constraint forces (especially active muscular forces) that occur in vivo and that vary across individuals and under different conditions. In terms of static posture, some clinical studies have demonstrated that posterior humeral head subluxation was not correlated with glenoid version or its degree of correction following total shoulder arthroplasty92,93, suggesting the need for additional research to better understand the role of glenoid version.
There are technical difficulties associated with measuring glenoid version preoperatively and setting a desired implant version in the operating room. Preoperative measurements of glenoid version on radiographs and axial CT images often have errors exceeding 5° due to malalignment of the imaging plane94,95. In the operating room, limited exposure and lack of anatomical reference points can impede accurate setting of the implant version angle96,97. Patient-specific instrumentation98 and computer-navigated surgery with motion tracking99,100 have been used with some initial success, and future implant designs may likely integrate with the technologies used to align them.
Patients with severely deformed glenoids often have accompanying altered ligamentous, capsular, and muscular environments101. Resetting glenoid version to a more neutral position should be done in conjunction with the intraoperative soft-tissue releases and balancing necessary to attempt to minimize eccentric loading of the new glenoid component. Franta et al., in a review of 136 failed total shoulder arthroplasty procedures, found that 64% of the shoulders demonstrated malalignment (partial subluxation) between the humeral and glenoid components102, indicating the importance of shoulder mechanics and stability in patient outcomes.
The options for correcting excessive retroversion include augmenting the posterior side with bone graft or cement, reaming down the anterior side, and using a posterior-augmented glenoid component. Bone-grafting can be useful in cases of severe deformity, but clinical results have been mixed, with complications including nonunion, shifting, and substantial resorption of the graft103-105. Correction of version by reaming down the anterior side is useful for mildly or moderately retroverted glenoids. In several recent studies, three-dimensional computer modeling has been used to analyze the geometric fit of commercial glenoids within retroverted glenoid vaults106-108. Nowak et al.106 simulated implantation of an in-line three-pegged glenoid component in nineteen CT scans of osteoarthritic patients with a median retroversion of 14°. Retroversion of <12° could be corrected in all cases, whereas retroversion >18° could not.
With a better understanding of retroverted glenoid morphology and difficulties in correcting version, some recently described glenoid designs have included a posterior augmentation to compensate for bone loss. This approach can be used to restore normal glenoid version without reaming away subchondral bone or using a separate graft or cement augmentation. Cofield and colleagues109 reported satisfactory intermediate-term pain relief and improvement in function in thirteen patients who were managed with a glenoid implant that had a sloped backing that was wider on the posterior side. However, the results were mixed in terms of glenohumeral stability. Recently, we and others have reported on the biomechanical characteristics of posterior-stepped glenoid components67,110. Another area of ongoing111 and future research is in the development of personalized custom glenoid components matched to a particular patient.
Increasing the life span of the glenoid component through better fixation and alignment would greatly improve patient satisfaction and would reduce health care costs, but methods of secure, durable fixation remain elusive in a subset of patients, especially those having excessive retroversion with humeral subluxation.
Source of Funding: One of the authors (G.S.L.) was supported by career development award No. 5KL2TR000126-03, funded by the National Institutes of Health and Penn State Clinical and Translational Science Institute.
Investigation performed at the Department of Orthopaedics and Rehabilitation, Penn State Milton S. Hershey Medical Center and College of Medicine, Hershey, Pennsylvania
Disclosure: One or more of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of an aspect of this work. In addition, one or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.
- Copyright © 2013 by The Journal of Bone and Joint Surgery, Incorporated