➢ Primary glenoid bone loss usually occurs in association with osteoarthritis characterized by a posterior wear pattern, whereas secondary glenoid bone loss usually occurs in association with trauma, glenoid loosening, and iatrogenic injury during revision.
➢ Proper preoperative imaging is critical to ascertain glenoid characteristics, including size, version, and vault depth.
➢ Treatment of glenoid bone loss is dependent on the degree of version correction that is required and consists of eccentric reaming, bone or polyethylene augmentation, or the use of reverse shoulder arthroplasty.
➢ Future research should include investigation of the use of different biomaterials, ceramics, glenoid vault endosteal fixation, reverse-polarity implants, and modular glenoid components.
The rate of total shoulder arthroplasty has increased over the last decade. More total shoulder arthroplasties than hemiarthroplasties have been performed annually since 20061. The projected number of total shoulder arthroplasty procedures has been predicted to increase by 192% to 322% from 2007 to 20152. A meta-analysis of patient quality-of-life scores after anatomic total shoulder arthroplasty showed improvement in both physical and pain-related scores, with the greatest improvement in shoulder-specific questionnaires3.
With worsening severity of posterior bone loss, the glenoid typically becomes more retroverted. Posteriorly worn glenoids are associated with posterior instability after total shoulder arthroplasty4. Hopkins et al.5 found that retroverted glenoid components may be more susceptible to loosening after total shoulder arthroplasty. Glenoid component retroversion also has been correlated with diminished shoulder function and decreased range of motion6,7.
Because of the prevalence of glenoid bone loss in patients with arthritic shoulder conditions, various techniques have been developed to address this issue. One such technique is to not implant a glenoid component at all. However, humeral hemiarthroplasty in patients with a degenerative glenohumeral joint has failed to show acceptable results8,9. The so-called “ream-and-run” technique also has been proposed as an alternative to glenoid component implantation but has shown variable results and does not address all cases of bone loss10.
Other methods to correct osseous deformity during total shoulder arthroplasty may allow successful glenoid implantation. These techniques include eccentric reaming, bone-grafting, and the use of augmented implants. In cases of severe bone loss in older patients, reverse shoulder arthroplasty has also been reported, with some success11. This article will review the anatomy and assessment of glenoid morphology, relevant imaging, glenoid-restoration principles, and current surgical treatment options to deal with glenoid bone loss in total shoulder arthroplasty.
Normal and Pathological Glenoid Bone Anatomy and Variability
Normal Glenoid Anatomy
Much research has been done to determine normal glenoid anatomy. Churchill et al.12 evaluated 344 human scapulae and determined that glenoid version—the angulation of the glenoid articular surface to the transverse axis of the scapula (Fig. 1, A)—averaged 1.23° of retroversion (range, 9.5° of anteversion to 10.5° of retroversion). Matsumura et al.13 utilized computed tomography (CT) scanning to compare version in the dominant and nondominant arms of 410 healthy volunteers. The mean retroversion (and standard deviation) was 1° ± 3° on the dominant side, compared with 0° ± 3° on the nondominant side (p < 0.001). In addition, men were found to have increased retroversion compared with women (mean, 1° ± 3° compared with 0° ± 3°; p < 0.001).
Glenoid inclination is a measurement of the slope of the glenoid articular surface in the superoinferior plane. Churchill et al.12 reported a mean of 3.9° to 4.6° of superior angulation and determined that there was no significant difference in mean inclination between men and women or between white and black patients.
The glenoid has been shown to be larger in the superoinferior direction than in the anteroposterior direction. Because the diameter of the lower half of the glenoid is often larger than the diameter of the upper half, the glenoid surface is often referred to as pear-shaped. Checroun et al.14 evaluated 412 scapular specimens and reported that 71% of glenoids were pear-shaped, with the remainder being elliptical. Iannotti et al.15 described the mean size of the glenoid to be 39 ± 3.5 mm in the superoinferior direction and 29 ± 3.2 mm in the anteroposterior direction, with the ratio of the lower half to the upper half being 1:0.80 ± 0.01 in the anteroposterior direction. Churchill et al.12 determined the mean glenoid height to be 37.5 mm for men and 32.6 mm for women, with a mean width of 27.8 mm for men and 23.6 mm for women. No difference was found between specimens from white and black patients.
The glenoid vault is defined as the cancellous bone medial to the glenoid articular surface16 (Fig. 1, B). In an attempt to define the shape of the glenoid vault and find an endosteal source for glenoid fixation, Codsi et al.16 used three-dimensional CT (3D-CT) to evaluate the size and shape of sixty-one cadaveric scapulae. The glenoid vault model that was produced suggested that the glenoid vault is nearly triangular in its entire length and that five sizes of implants could fit all of those modeled. Excessive reaming resulting in decreased vault volume and glenoid surface medialization has been shown to decrease the available area of fixation17.
Pathological Glenoid Changes
Primary Glenoid Bone Loss
Glenoid bone loss is reported as the percentage of the defective area relative to the original intact glenoid. In osteoarthritis, the characteristic wear pattern involves posterior glenoid erosion with posterior subluxation of the humeral head. When this erosion is severe, a second concavity forms and the glenoid is described as biconcave. On the basis of radiographs and CT scans, Walch et al.18 first classified glenoid bone loss in 151 patients with primary glenohumeral arthritis. Glenoid bone loss was classified into three types: type A (concentric; 53.5%), type B (posteriorly subluxated humeral head; 39.5%), and type C (glenoid retroversion of >25° regardless of erosion; 5%) (Fig. 2). Types A and B were further subclassified into type A1 (minor erosion), type A2 (major erosion), type B1 (posterior joint-space narrowing with subchondral sclerosis and osteophytes), and type B2 (biconcave, retroverted glenoid with posterior rim erosion). This classification system has been shown to have moderate interobserver reliability19.
Several studies have analyzed whether variable glenoid dimensions may predispose to a given arthritic wear pattern. Ricchetti et al.20 evaluated glenoid version as a cause of osteoarthritic changes with use of a glenoid vault model. Comparison of pathological shoulders with nonpathological, contralateral shoulders and cadaveric specimens revealed no predisposing premorbid glenoid version or inclination. The authors attributed bone loss in shoulders with osteoarthritis to degenerative changes. Similarly, Habermeyer et al.21 analyzed glenoids with osteoarthritis and determined that eccentric inferior glenoid wear is frequent and independent from retroversion deformity of the glenoid.
Secondary Glenoid Bone Loss
Secondary causes of glenoid bone loss include trauma and glenoid component loosening or revision. Posttraumatic changes to the glenoid often manifest as anterior glenoid rim defects. In the study by Moroder et al., shoulders with recurrent anterior instability and a mean anterior glenoid bone loss of 17.7% had an increased anteversion of 6° compared with the healthy, contralateral side22. Kikuchi et al.23 reported that an increased anteversion of >5° compromised shoulder stability.
Glenoid loosening and bone loss after revision shoulder arthroplasty may be due to poor component placement, mechanical toggling, osteolysis, or infection. Gonzalez et al., in a review of 2657 patients, reported that the overall rate of glenoid loosening after total shoulder arthroplasty was14%24. Antuna et al.25 first classified glenoid destruction in patients undergoing revision arthroplasty as central, peripheral, and combined; the classifications were further subdivided into mild, moderate, and severe. In a clinical series with a minimum five-year follow-up, Walch et al.26 reported that glenoid component migration was associated with low implant positioning, superior tilt, and excessive reaming. In a biomechanical study, Collins et al.27 emphasized the importance of seating the glenoid component completely and congruently to minimize movement, deformity, mechanical loosening, and polyethylene debris with eccentric loading. Ho et al.28 reported that glenoid components that were implanted in >15° of retroversion were associated with increased rates of osteolysis.
Posterior glenoid bone loss is also associated with instability after shoulder arthroplasty. Moeckel et al.29 described three cases of posterior instability after total shoulder arthroplasty that were treated with glenoid component revision and soft-tissue tensioning. Similarly, Steinmann and Cofield30 and Hill and Norris31 reported on patients who had glenoid bone loss and instability after arthroplasty.
Evaluation of glenoid deficiency begins with radiographs. Dedicated three-view shoulder radiographs should include a true anteroposterior (or Grashey) glenohumeral view, a scapular lateral view, and an axillary view. A Grashey radiograph is made at a 30° lateral oblique angulation perpendicular to the scapular plane and the glenohumeral articulation. All shoulder radiographs should be evaluated for glenoid shape, osteophytes, and areas of erosion. An axillary radiograph is useful for evaluating glenoid bone erosion in the anteroposterior plane; however, this view has been shown to have poor intraobserver and interobserver reproducibility and to overestimate glenoid retroversion32.
Advanced imaging modalities can also be useful for glenoid evaluation. Nyffeler et al. reported that CT scanning provided more reproducible measurements of version in many patients33. Recently, magnetic resonance imaging (MRI) has been advocated for preoperative imaging as it provides more accurate results than radiographs, provides similar results to CT in terms of osseous architecture, allows for rotator cuff evaluation, and results in no ionizing radiation34. Three-dimensional CT scanning has gained attention as an aid for determining glenoid vault morphology and for performing dimension analysis16 and should be considered in cases of severe bone loss or retroversion.
Principles of Glenoid Restoration
Restoration of Neutral Version
Ideal glenoid preparation simultaneously achieves anatomic correction of glenoid version, preservation of existing glenoid bone stock, and complete contact between the component and the underlying bone. Malpositioning of the glenoid component produces abnormal stresses, possibly leading to cement mantle failure and component loosening35,36. Retroverted glenoid implants have been shown to increase the rate of osteolysis28.
Neutral glenoid placement is difficult in patients with posterior glenoid wear patterns. An excessive attempt to neutralize the glenoid with eccentric reaming or augmentation may cause component medialization37 or increased interface strain38 and is limited by the narrowing of the triangular dimension of the vault17. In cases of traumatic anterior or posterior body or rim defects, malpositioning may go unrecognized.
Glenoid implantation must also restore anatomic glenoid inclination. Oosterom et al.39 showed that the placement of glenoid components with superior tilt resulted in greater superior rim displacement and decreased shoulder stability.
Achieving glenoid version correction must not be done at the expense of full contact between the glenoid and the prosthesis. Incomplete contact leaves the prosthesis prone to toggling and premature loosening4,27. Recently, with the use of a physiologic model of the scapula, Yongpravat et al.40 demonstrated that, during attempts to restore version, maintenance of cortical bone-to-implant contact is more important than previously thought.
The glenoid component should be centered, with no overhang peripherally. Computer model analyses of glenoid contact stresses have shown that stress from peripheral loading is higher than central loading in multiple component designs41. Reaming should be used for glenoid articular surface preparation as the use of reaming has been shown to be superior to hand-burring or curettage in resisting deformation or edge displacement27.
Retroversion of the Glenoid Component
In some cases of severe posterior bone loss without adequate correction, the glenoid component is implanted in retroversion. Residual retroversion may be left purposefully to avoid excessive correction and to maintain bone stock. Attempts have been made to compensate for this glenoid retroversion with anteversion of the humeral component. Several studies have shown that this technique is not able to counteract the issues seen in association with glenoid implant retroversion33,42.
Unintentionally retroverted glenoid implants have been shown to correlate with preoperative retroversion. Assessment of neutral version can be increasingly difficult in patients with deformity. Gregory et al.43 evaluated CT scans of twenty-nine scapulae preoperatively and postoperatively and reported that component retroversion was reflective of the degree of preoperative retroversion. A clinical study of sixty-eight consecutive total shoulder arthroplasties that had been performed by one surgeon demonstrated a mean retroversion correction of only 2° (from 8° to 6°) after a mean duration of follow-up of thirty-eight months6. Iannotti et al.44 reported that an experienced shoulder surgeon using a 3D surgical simulator placed glenoid components in a mean of 13° of retroversion.
Effects of Poor Implant Position
Improper positioning of the glenoid component has been associated with biomechanical disadvantages. Nyffeler et al.33 tested edge-loading and humeral head displacement with differing angles of glenoid version and found increased loading and humeral head displacement in association with both increased anteversion and retroversion. Farron et al.35 demonstrated that glenoid retroversion of >10° increased stress on the cement mantle by 326% and increased micromotion at the bone-cement interface by 706%. Ho et al.28 reported that postoperative glenoid retroversion of >15° was associated with increased osteolysis around the central peg.
Improper positioning also has been associated with poorer clinical outcomes. Gregory et al.43 evaluated the accuracy of glenoid positioning and correlated it with clinical outcomes. Retroversion and non-neutral rotation were associated with a reduced range of motion. Yian et al.7 reported that patients with glenoid component retroversion had diminished shoulder function and lower Constant scores.
To date, only small studies have been performed to determine the utility of computer-assisted glenoid implantation. Stanley et al.45 reported on an initial series of thirteen patients and reported on the safety and accuracy of this method. Nguyen et al.46 reported improved version in all phases of glenoid reaming and insertion in a study involving sixteen cadaver models. Kircher et al.47, in a prospective randomized trial of ten patients, reported a mean postoperative retroversion of 3.7° ± 6.3° in the navigation group, compared with 10.9° ± 6.8° in the no-navigation group (p = 0.021).
Posterior glenoid wear and retroversion is a difficult problem to treat in shoulder arthroplasty. Currently, operative treatment options include eccentric reaming, bone-grafting, use of augmented glenoid components, and reverse shoulder arthroplasty.
Osteoarthritic wear is characteristically seen in the posterior part of the glenoid. Greater reaming of the anterior aspect of the glenoid (the high side) to correct retroversion is a commonly used technique. This technique has the advantage of reorienting the joint surface. By medializing the center of rotation of the joint, this technique also protects the subscapularis tendon reconstruction and reduces the joint compressive forces.
Eccentric reaming is limited by the decrease in the face diameter of the glenoid, potentially leading to overhang or cortical peg penetration. Gillespie et al.17 simulated varying degrees of retroversion correction in eight cadaveric scapulae and measured the resultant face diameter and ability to implant a glenoid component. A correction of 10° of retroversion decreased the mean anteroposterior glenoid diameter from 26.7 ± 2.5 to 23.8 ± 3.1 mm, whereas a correction of 15° resulted in 50% of specimens having inadequate osseous support. The long-term implication of cortical peg penetration is yet unknown, although two recent studies demonstrated minimal to no increased morbidity48,49.
Nowak et al.50 similarly evaluated the limit to eccentric glenoid reaming in a 3D-CT model of nineteen patients. Less than 18° of preoperative retroversion was required to implant a 46-mm glenoid in neutral without vault penetration.
Glenoid augmentation has been advocated in cases of retroversion in which eccentric reaming may not achieve deformity correction, would result in vault-peg penetration, or would cause excessive medialization. Bone-grafting of glenoid deficiencies may be done in one or two stages. Retroversion defects also may be corrected through single-stage total shoulder arthroplasty with humeral head bone graft or corticocancellous iliac crest autograft (Fig. 3). Neer and Morrison51 reported on twenty shoulders that were treated with internally fixed corticocancellous grafts. At a mean of 4.4 years, sixteen of the twenty shoulders were reported to have excellent clinical results according to the Neer rating system52. Steinmann and Cofield30 reported on twenty-eight shoulders with segmental glenoid wear that were treated with glenoid bone-grafting. After a mean duration of follow-up of 5.3 years, thirteen shoulders were rated as excellent, ten were rated as satisfactory, and five were rated as unsatisfactory according to the Neer rating system52. Radiographically, thirteen shoulders had no radiolucencies, eleven had incomplete radiolucencies, and four had complete radiolucencies, of which three were at least 1.5 mm wide. Hill and Norris31 reported on thirteen shoulders with a preoperative mean retroversion of 33° that were treated with a corticocancellous graft. After a mean duration of follow-up of seventy months, fourteen of seventeen patients maintained improvement to a mean of 4° of glenoid retroversion. Sabesan et al.53 reported on twelve shoulders with a mean preoperative retroversion of 44°. All shoulders received humeral head graft augmentation. After a mean duration of follow-up of fifty-three months, ten of the twelve shoulders had maintenance of corrected version and graft incorporation. Klika et al.54 followed twenty-five shoulders that were treated with structural grafting. After a mean duration of follow-up of 8.7 years, the authors reported a mean of 148° of flexion, 60° of external rotation, and satisfaction with the pain relief. At 7.6 years, radiographic evaluation demonstrated that ten of twenty-four shoulders were at risk of glenoid loosening; the authors concluded that those treated with structural grafting had higher rates of loosening.
Polyethylene component augments are an alternative to bone-grafting (Fig. 4). Metal-backed augments have been shown to have poor clinical outcomes resulting from thin polyethylene inserts and excessive wear52. Kirane et al.55 evaluated ‘‘Poly-step’’ and ‘‘Ti-step’’ glenoids (attached polyethylene or titanium stepblocks) and found that glenoid strain was increased with the Ti-step implant but not the Poly-step implant.
Sabesan et al.37 used a 3D-CT model to compare twenty-nine eccentrically reamed glenoids with twenty-nine glenoids that were fixed with all-polyethylene glenoid augments. In cases of >16° of glenoid retroversion, less medialization was found in association with all-polyethylene augmentation compared with eccentric reaming. The authors concluded that correction of moderate to severe glenoid retroversion by means of eccentric reaming cannot always be done and that the use of augmented components can allow complete correction of retroversion and can minimize the effect of medialization. Iannotti et al.38 evaluated liftoff resistance of the stepped glenoid augment design and found that the stepped design had greater resistance to liftoff and superior fixation compared with spherical, flat angled, and non-augmented glenoids.
Hermida et al.56 evaluated a wedge glenoid component in a scapular model and found that the wedged component decreased stresses and predicted greater bone fatigue life compared with standard glenoid components.
Youderian et al.57, in a study on the preliminary results for twenty-four shoulders that were followed for a minimum of six months (mean, 10.8 months), reported a mean forward flexion of 162° and a mean postoperative Penn58 score of 84. Postoperative CT demonstrated excellent correction (mean, 16.7°) of glenoid version, with minimal loss (mean, 0.45 mm) of the premorbid joint line.
Recent interest in materials designed for osseous ingrowth rather than cemented components could reduce the need for secure peripheral fixation in cases of glenoid bone loss (Fig. 5). With improved central fixation, the periphery may become less influential on overall stability. Churchill et al.59 reported on twenty shoulders that were treated with a partially cemented glenoid prosthesis with an interference-fit centrally fluted peg. At the time of the five-year follow-up, fifteen patients had similar or increased bone presence between the central peg fins, with none having a worsening radiolucency score. Vidil et al.60 reported on twenty-seven patients who were managed with an interference-fit, centrally fluted glenoid peg with three cemented peripheral pegs. At one year of follow-up, twenty-one patients had full osseointegration and four had partial peripheral osseointegration. Wirth et al.61 reported the four-year follow-up for forty-four shoulders that were treated with a similar component. At the time of the latest follow-up, twenty shoulders had perfect seating and no radiolucency and thirty shoulders had increased radiodensity between the flanges of the central peg, indicative of osseous ingrowth.
Cementation may not be necessary at all with improved alternative fixation. A biomechanical study by Sarah et al.62 demonstrated that implantation failure occurred at the cement-implant interface irrespective of the implant design. Groh63, in a subsequent study of eighty-three primary shoulder arthroplasties that had been performed with a cementless fluted glenoid component, found no evidence of radiographic or clinical loosening after minimum of two years of follow-up. De Wilde et al.64, in a study of thirty-four patients who underwent implantation with a cementless fluted glenoid, reported that thirty patients demonstrated no signs of loosening after a mean duration of follow-up of twenty-eight months.
Porous-ingrowth implant designs have also emerged to improve osseous fixation and hopefully to improve longevity (Fig. 6). In the study by Budge et al.65, biomechanical testing demonstrated the stability of porous tantalum with polymethylmethacrylate (PMMA) cementation. A subsequent clinical trial of nineteen patients with limited PMMA cementation demonstrated stable glenoid component ingrowth at a mean of thirty-eight months, but with an unacceptably high rate of clinical complications and failure attributed to implant design66. Based on these findings, porous-ingrowth implants were not shown to be superior to all-polyethylene cemented components.
Glenoid Inset Design
Recent interest in attaining greater fixation on the glenoid surface has led to the development of the inset glenoid technique. This technique leaves a rim of native glenoid bone surrounding the periphery of the component. Glenoid inset fixation potentially improves stability compared with the standard onlay design in cases of severe glenoid structural bone loss. In the study by Gunther et al.67, biomechanical testing demonstrated less distraction of the glenoid rim in association with the use of an inset technique compared with an onlay technique. The inset technique was associated with up to an 87% reduction in displacement. In a separate study, Gunther and Lynch68 evaluated the clinical results at a mean of 4.3 years for seven patients with severe bone deficiencies and reported improvement in terms of range of motion and pain. Radiographic evaluation classified all implants as being at low risk for glenoid loosening.
In cases of anterior or posterior bone loss, the humeral head is often subluxated. The use of an inset glenoid will not improve this issue, and the total shoulder arthroplasty articulation will not be ideal. In cases of bone loss from medialization in which the humeral head remains centrally located, inset implants may be of greater utility.
With the goal of improving the accuracy of glenoid component version, patient-specific instrumentation has been developed. Hendel et al.69 reported the greatest benefit in patients with retroversion of >16°. Iannotti et al.70 found that the use of a patient-specific surgical model with a reusable tool improved guide-pin positioning when compared with standard instrumentation with two-dimensional (2D) and 3D planning. To our knowledge, no cost analysis of the use of patient-specific instrumentation in the shoulder has been reported to date.
Reverse Total Shoulder Arthroplasty
Patients with severe posterior humeral subluxation present a challenge. The use of primary reverse total shoulder arthroplasty should be considered for elderly, low-demand patients with substantial posterior glenoid bone loss. Mizuno et al.11 reported on twenty-seven patients with a mean age of 74.1 years and a mean of 32° of glenoid retroversion and 87% humeral subluxation. Seventeen patients underwent reverse shoulder arthroplasty without bone-grafting, and ten underwent reverse shoulder arthroplasty with bone-grafting to correct >10° of posterior glenoid erosion that was uncorrected by eccentric reaming. After a mean duration of follow-up of fifty-four months, there was substantial improvement in clinical scores and range of motion. This procedure should be considered for elderly patients in order to avoid multiple operations for staged bone-grafting or revision of failed anatomic total shoulder arthroplasty. In younger patients, the other previously described techniques should be considered primarily.
Severe glenoid bone deficiency can prove to be a challenging problem to address during total shoulder arthroplasty. The most common cause of glenoid bone loss is osteoarthritic posterior wear.
In patients with glenoid erosion, it is important to evaluate bone loss with proper imaging. In our practice, we use radiographs with an emphasis on the axillary view. Radiographs sometimes may be difficult to assess, and a CT scan can be performed to improve evaluation. Preoperative planning should emphasize returning the glenoid to neutral version and inclination with proper implant seating and no overhang. Retroversion of glenoid components may or may not be intentional and should be avoided, as the results in these cases are inferior.
Current surgical options to correct glenoid version include eccentric reaming, bone-grafting, glenoid component augmentation, and, in some cases, primary reverse shoulder arthroplasty. The decision about which technique to utilize is dependent on the extent of glenoid bone loss, surgeon familiarity, patient age, and functional demands. With mild posterior glenoid bone loss, up to approximately 15° of retroversion, we prefer eccentric reaming. For patients with larger amounts of retroversion, eccentric reaming combined with bone-grafting or augmented components are options (Table I).
Glenoid wear often creates difficulty with the implantation of traditional glenoid components. Correction of version to neutral often can lead to glenoid vault shortening, medialization, and peripheral peg cortical perforation. The development of augments has helped recreate neutral version without excessive reaming; however, the longevity of these implants is not yet known.
Future research also should evaluate the long-term outcomes of biomaterial-augmented glenoid components; the use of alternative materials, including ceramics; the utility of fixation within the endosteal glenoid vault; and the use of reverse-polarity implants. Increased modularity should be investigated, and metal-backed anatomic glenoid components with conversion to a reverse baseplate should be further elucidated.
Source of Funding: No funding was received for this manuscript.
Investigation performed at William Beaumont Hospital, Royal Oak, Michigan
Disclosure: None 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 any aspect of this work. 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 © 2015 by The Journal of Bone and Joint Surgery, Incorporated