Anatomic vs. reverse shoulder arthroplasty for the treatment of Walch B2 glenoid morphology: a systematic review and meta-analysis

Background Walch B2 glenoid morphology with glenohumeral osteoarthritis is a difficult degenerative pattern to manage for shoulder surgeons. Anatomic total shoulder arthroplasty (TSA) in combination with eccentric reaming or bone grafting are the traditional methods of treatment. Newer approaches such as TSA with posteriorly augmented glenoid components and reverse shoulder arthroplasty (RSA) may offer better stability for the posteriorly subluxated biconcave B2 wear pattern. The aim of this systematic review is to compare mid-term surgical and functional outcomes of Walch B2 glenoids without significant rotator cuff pathology treated with TSA and RSA. Methods The review was performed according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines by searching the MEDLINE (PubMed) and Embase (Elsevier) databases. Inclusion criteria were clinical studies that evaluated the outcomes and complications of TSA or RSA in the setting of B2 glenoid morphology without significant rotator cuff pathology. Data relevant to TSA and RSA surgical outcomes were extracted and compiled, and outcomes were compared. A meta-analysis of proportions of complication and revision rates among TSA and RSA groups was performed. Results Overall, 16 articles were included with 414 TSAs and 78 RSAs. The average follow-up duration was 54.1 ± 14.8 months for patients undergoing TSA and 44.8 ± 10.1 months for patients undergoing RSA. The TSA group was further subdivided based on the use of eccentric reaming (135 TSAs), an augmented glenoid component (84 TSAs), or bone grafting (11 TSAs) or was unspecified (184 TSAs). Overall, patients undergoing TSA and RSA demonstrated mean improvements of 50.1 ± 8.5° and 64.7 ± 5.2° in active flexion, 58.5 ± 10.3° and 68.9 ± not reported° in active abduction, and 31.3 ± 5.7° and 29.0 ± 10.2° in active external rotation, respectively. In regard to functional outcome scores, patients undergoing TSA and RSA showed mean Constant score improvements of 38.8 ± 5.3 and 46.6 ± 3.1 points and American Shoulder and Elbow Surgeons score improvements of 48.2 ± 1.0 and 49.2 ± 25.3 points, respectively. Results of the meta-analysis with mid-term follow-up data demonstrated pooled complication rates of 9% (95% confidence interval [CI], 1%-22%) for TSA and 6% (95% CI, 0%-28%) for RSA and pooled revision rates of 2% (95% CI, 0%-8%) for TSA and 1% (95% CI, 0%-15%) for RSA. Conclusion In the setting of Walch B2 glenoid morphology, TSA with eccentric reaming or an augmented component yields comparable outcomes to RSA. Based on the patient’s age, activity level, and expectations, both TSA and RSA can be considered a reasonable option to treat Walch B2 glenoid morphology.

particularly in patients with significant premorbid glenoid retroversion and/or a flat posterior acromial slope. 17,22,31 In addition, posterior capsular laxity, anterior capsular contracture, and fatty infiltration of the rotator cuff musculature have all been associated with the B2 wear pattern. 7,12 Controversy exists as to which arthroplasty technique best addresses the complexities created by the rotator cuff intact (RCI) B2 glenoid morphology. Traditionally, these deformities have been treated with anatomic TSA performed in conjunction with eccentric reaming (ER) to correct glenoid version. 18,26,30 This technique is limited, however, as correction of >15 of retroversion results in subchondral bone removal and joint line medialization that can cause posterior inferior peg perforation precipitating glenoid component loosening (GCL) and decreased joint stability. 4,26,30 Alternatively, TSA with bone grafting (BG) or the use of a posteriorly augmented glenoid component (PAGC) can permit version and subluxation correction without bone removal. However, the longterm clinical outcomes of these techniques are characterized by either inconsistent results or, in the case of PAGC, a lack of sufficient outcome data. 12,21,26,30,35,45 Overall, the difficulty in addressing the severe biconcave posterior wear and posterior humeral head subluxation with TSA has translated into more frequent GCL, higher incidences of radiolucent lines, increased revision rates, and lower functional outcome scores than other glenohumeral arthritic patterns. 6,9,11,12,17e20,26,30,45 Although previously used only in patients with rotator cuff tear arthropathy, several authors have recently reported good to excellent midterm joint stability, improved functional outcome, and low rates of baseplate loosening using RSA for the B2 glenoid morphology with an intact rotator cuff. 6,17,24,25,29,32 These findings are likely due to the semiconstrained design and robust glenoid baseplate fixation in RSA, eliminating complications created by posterior humeral head subluxation and decreasing the likelihood of GCL. 6,17,30 Despite these results and advantages, little investigation has been performed directly comparing the use of RSA to TSA to treat B2 glenoid deformities, with most studies reporting outcomes of each procedure individually in limited sample sizes. The purpose of this systematic review was to assess range of motion (ROM), functional outcomes, revision rates, and complications of B2 glenoids without significant rotator cuff pathology treated with RSA in comparison to TSA.

Eligibility criteria
All search returns were extracted and examined for relevance, and duplicate search returns were discarded. In phase 2, "screening," titles and abstracts were screened for relevance. Bibliographies of relevant articles were also manually searched for other relevant articles screened out of the database algorithms. Articles were filtered out per the following exclusion criteria: (1) non-English text, (2) revisional cases, (3) only abstract available, (4) hemiarthroplasty only, (5) glenoid morphology other than biconcave (B2) glenoid (6) significant concomitant rotator cuff pathology, (7) review article or meta-analysis, or (8) case reports. Of note, an exception to one of these criteria was made for the included article by Magosch et al 27 which had German text that could be reliably translated using a Google translation plugin. As the article passed all other phases of review, the authors deemed it appropriate for final inclusion. The shoulder osteoarthritis treatments being reviewed included TSA and RSA regarding outcomes specifically pertaining to the Walch classification B2 glenoid. Studies were evaluated only if they were (1) primary TSA or primary RSA; (2) included glenoids with biconcave morphology (Walch B2); (3) included postoperative outcomes such as American Shoulder and Elbow Surgeons (ASES) score, Constant score, ROM, and complications; and (4) reported a minimum of 1 year of follow-up (FU).

Article review
In phase 3, "eligibility," all articles eligible after the screening phase were evaluated for inclusion criteria and relevant data on outcomes after 1 of the surgical treatments of interest. All articles were reviewed, assessed, and data-mined by 3 independent evaluators. All results were then compared to ensure consistency and accuracy. Any conflicts or issues were resolved by review, and in the event of further disagreement, the final decision was made by the senior author (X.L.).

Data extraction and assessment
In phase 4, "included," articles that met inclusion criteria were analyzed for quality, and data were extracted to be used in a metaanalysis. The following items of data were extracted from the included articles: author; publication year; journal title; level of evidence; study design; surgical procedure; number of patients; sex ratio (M:F); mean age at the time of surgery; glenosphere position; concomitant procedures; glenoid correction technique including ER, BG, and PAGC; FU time; active preoperative ROM and postoperative ROM at the final FU in flexion, abduction, and external rotation; preoperative Constant score; postoperative Constant score at the final FU; preoperative ASES score; postoperative ASES score at the final FU; complications; and subsequent procedures.

Quality assessment
To assess the quality of each case series that was included in the analysis, the Risk of Bias in Non-Randomized Studies of Interventions (ROBINS-I) tool is typically applied. 41 However, most studies included in this analysis were of level III or IV evidence, and therefore, applying the ROBINS-I criteria was unnecessary.

Statistical analysis
A meta-analysis of proportions of complication and revision rates among RSA and TSA was performed as this was the most consistent outcome variable that was amenable to this type of analysis reported. The meta-analysis was conducted by using a Freeman-Turkey transformation (arcsine square root transformation) under the random-effects model to calculate pooled estimate rates, whereas 95% confidence intervals (CIs) were estimated with the DerSimonian-Laird estimator. 10 In an effort to account for differences among studies in regard to patient characteristics, surgical characteristics, and study methodology, the random-effects model was chosen. 3 The heterogeneity analysis of the included studies was represented using I 2 , where I 2 represents an estimated percentage of error attributed to interstudy variation. 15 Based on the Cochrane review handbook, values of I 2 between 0 and 40% were deemed to not be important, 30%-60% represented moderate heterogeneity, 50%-90% substantial heterogeneity, and 75%-100% considerable heterogeneity. 14

Results
After a careful search and review of the currently available literature, 1 level II, 13 13 in which the authors describe select patients who received BG in addition to ER but did not uniquely stratify out the results for these patients. Therefore, we could not reliably include them in either the TSA þ ER or TSA þ BG subgroups. In total, there were 4 TSA subgroups analyzed: (1) TSA þ ER, (2) TSA þ PAGC, (3) TSA þ BG, and (4) TSA þ unspecified glenoid approach (UGA). Among the demographic data that could be extracted, the average age of patients undergoing TSA was 63.7 ± 1.8 years vs. 72.8 ± 1.3 years for patients undergoing RSA. In addition, the average FU was 54.1 ± 14.8 months for patients undergoing TSA and 44.8 ± 10.1 months for patients undergoing RSA. Complete demographics provided by each study are detailed in Table I for patients undergoing TSA and in Table II Table VI and compared as weighted averages in Table VII.

Complication and revision rates
Overall pooled complication and revision rates after RSA and TSA were evaluated in a meta-analysis of proportions model. The pooled complication rate among the 2 studies that reported on RSA was 6% (95% CI, 0%-28%) with substantial heterogeneity (I 2 ¼ 71%) compared to the 7 studies that reported on TSA with a rate of 9% (95% CI, 1%-22%) with substantial heterogeneity (I 2 ¼ 80%). The pooled revision rate among the 2 studies that reported on RSA was 1% (95% CI, 0%-15%) with no heterogeneity (I 2 ¼ 0%) compared to TSA that had a reoperation rate of 2% (95% CI ¼ 0%-8%) with substantial heterogeneity (I 2 ¼ 69%). The forest plots of pooled complication and revision rates are presented as Figures 2 and 3, respectively.  For the TSA þ ER subgroup, the complication rate was 5.6% (5/89 shoulders), and the revision rate was 0.9% (1/116 shoulders). Two shoulders had GCL, 3 had failure of the posterior capsule plication procedure (1 had both), and 1 had a rotator cuff tear that underwent conversion to RSA. For the TSA þ PAGC subgroup, the complication rate was 10.5% (4/38 shoulders), and the revision rate was 2.4% (2/84 shoulders). The majority of these complications were glenohumeral subluxation (3 posterior, 1 anterior), of which 1 also demonstrated GCL. The 2 revisions were for GCL. For the TSA þ BG subgroup, the complication and revision rates were both 18.2% (2/11 shoulders). These 2 shoulders were revised for GCL. For the TSA þ UGA subgroup, the complication rate was 20.7% (19/92 shoulders), all from radiographic GCL. The revision rate was 16.3% (15/92 shoulders), of which the majority were for GCL (6 shoulders) and posterior      shoulder dislocations (5 shoulders). Four RSA shoulders were noted for complications. Of these, 3 were nerve palsies (1 axillary, 2 ulnar) and 1 was GCL which was revised to hemiarthroplasty.

Discussion
The purpose of this systematic review is to compare the surgical, radiographic, and functional outcomes of patients with symptomatic glenohumeral joint osteoarthritis and a Walch B2 glenoid without significant rotator cuff pathology managed with either anatomic TSA vs. RSA. We hypothesized that a systematic review of the literature would demonstrate patients with B2 glenoid morphology having superior outcomes and a lower complication and revision rate after RSA than TSA. In regard to ROM and clinical score outcomes, the results of our review suggest largely similar outcomes between RSA, all TSA, and the TSA subgroups in the setting of B2 glenoid morphology. Regarding complications, similar rates were observed between the TSA þ ER (5.6%) and RSA (6%, pooled rate from the meta-analysis) study groups, while more significant rates were observed in the TSA þ PAGC (10.5%), TSA þ BG (18.2%), and TSA þ UGA (20.7%) subgroups. Revision rates of <3% were also demonstrated for all study groups with the exception of the TSA þ BG and TSA þ UGC subgroups which demonstrated much higher rates of 18.2% and 16.3%, respectively. Overall, these results suggest similar outcomes between the traditional TSA þ ER approach and RSA for B2 glenoids, with newer approaches of TSA þ PAGC and TSA þ BG potentially leading to more complications and the need for revision. It is important to note that the 92 shoulders from the review by Walch et al 45 on ER for B2 glenoids were not included in the TSA þ ER subgroup and instead contribute to the TSA þ UGA subgroup; however, the results described by that study, particularly the high complication and revision rates, should also be considered when interpreting the outcomes of the TSA þ ER subgroup from our analysis.
Neer et al 34 recognized that posterior glenoid erosion and static posterior humeral head subluxation was a common wear pattern in symptomatic glenohumeral joint osteoarthritis. Nearly 20 years later, Walch et al 44 described the B2 glenoid as one with static posterior humeral head subluxation and a "posterior cupula," giving the glenoid a biconcave morphology. Since then, the challenge of managing the unique B2 glenoid morphology has been an area of great interest for shoulder surgeons. Multiple techniques including TSA in combination with ER, PAGC, and BG as well as RSA have been described to address this deformity.

TSA with eccentric reaming (TSA þ ER)
In the ER technique, the anterior glenoid is reamed eccentrically to eliminate the biconcavity, correct retroversion, and provide a stable plane for the glenoid component of anatomic TSA. Generally, surgeons aim to correct glenoid retroversion to 5-10 to neutral version. This is in an effort to optimize the glenohumeral contact point, preventing the implant stress that can arise from retroverted implants as has been described in several biomechanical studies. 9,40 Work by Ho et al 16    their series of 66 shoulders, demonstrating increased osteolysis rates around the center glenoid peg and loosening when component retroversion was 15 . It is important to consider, however, that ER in cases of severe deformity involves significant glenoid bone stock removal possibly contributing to GCL and excessive joint line medialization that can disrupt rotator cuff tensioning. 1,4,26 Our review demonstrated superior outcomes in mean flexion (162.7 ± 3.6 ) and abduction (149.6 ± 21.5 ) at the final FU as well as external rotation improvement (40.0 ± 12.3 ) in patients in whom this technique was used. A complication rate of 5.6% and revision rate of 0.9% were reported when the version was corrected to 5-10 of retroversion to neutral. In 2012, Walch et al 45 described the results of 92 B2 glenoids treated with TSA þ ER, most notably reporting a 20.6% GCL rate and a 16.3% revision rate in this cohort. In cases of intermediate glenoid retroversion >30 , the rate of complications reached 62%. This is in contrast to a similar study by Orvets et al 36 who found no instances of GCL in their cohort of 59 B2 glenoids treated with TSA þ ER. In addition, there was no difference in the rate of glenoid radiolucent lines between shoulders with a preoperative glenoid retroversion of 20 (27.8%) vs. >20 (22.7%, P ¼ .670). However, the authors acknowledged that their cohort did not include patients with >30 of glenoid retroversion. Furthermore, the mean radiographic FU was 32 months vs. 77 months in the study by Walch et al. 45 Additional studies by Leschinger et al 23 and Egger et al 8 support TSA þ ER to be an effective option for B2 glenoids regarding functional outcomes, and neither described any revisions in B2 patients. However, neither study indicated the degree of preoperative retroversion in their cohorts. Therefore, overall, the studies by Orvets et al 36 and Walch et al 45 provide the clearest data which may indicate that ER is not appropriate for glenoid retroversion >30 but is a reasonable option for those with retroversion <30 . It is also important to note that as many of the studies do not have long-term FU, it is difficult to assess the true rate of GCL.

TSA with bone grafting (TSA þ BG)
In the glenoid BG technique, asymmetric autograft (typically from the humeral head) is fixated with 2 or 3 independent screws before glenoid component implantation to correct retroversion. In our analysis, the TSA þ BG study group consisted entirely of 11 B2 glenoids reported on by Klika et al. 21  PAGCs were developed to allow for glenoid retroversion correction while avoiding glenoid subchondral bone loss and excessive joint medialization. Early studies demonstrated good results and low revision rates as compared to TSA þ BG and TSA þ ER. 26 Our analysis demonstrated good to excellent outcomes for TSA þ PAGCs in mean Constant score (77.4 ± 1.3 points) and ASES score (90.3 ± 0.9 points) at the final FU. Interestingly, the TSA þ PAGC subgroup also demonstrated the lowest mean improvements in Constant score (36.1 ± 0.6 points) and ASES score (47.6 ± 1.0 points) of any group. The revision rate for this group was only 2.4%; however, the complication rate was higher at 10.5% most commonly because of posterior glenohumeral subluxation.
Rice et al 39    RSA with RCI for B2 glenoid RSA has recently been advocated for as a treatment option for B2 glenoids in RCI shoulders. In theory, the semi-constrained design addresses the problems of recurrent posterior humeral head subluxation and posterior instability while the robust glenoid base plate fixation decreases the likelihood of GCL. However, surgeons may end up encountering a different set of complications unique to RSA. 37 In our analysis, we found B2 glenoids treated with RSA to have superior mean improvements in flexion (64.7 ± 5.2 ) and abduction (68.9 ± NR ) but inferior mean improvement in external rotation (29.0 ± 10.2 ). Mean abduction (128.9 ± 32.2 ) and external rotation (31.5 ± 13.3 ) at the final FU were also inferior to those of all TSA subgroups. Results from our meta-analysis also demonstrated a complication rate of 6% and revision rate of 1% for B2 glenoids treated with RSA.
In a recent study by Collin et al, 5  . Radiographically, however, 9 patients (60%) undergoing TSA experienced some degree of radiolucent lines with 4 patients (26.7%) deemed as radiographic failures. In addition, 2 patients (13.3%) demonstrated GCL, 3 (20%) had failure of the posterior capsule plication, and 2 (13.3%) developed progressive late rotator cuff insufficiency although no revisions were performed. This is in comparison to no radiolucent lines or GCL in the RSA cohort. Overall, this study suggests similar pain relief and improvement in ROM between approaches; however, TSA þ ER may offer superior functional outcomes although with a greater risk of GCL and soft-tissue failure. In a similar study by Magosch et al, 27 a comparable Constant score and active ROM improvement were demonstrated between RCI TSA and RSA study groups, with the only appreciable difference being in external rotation improvement (29.1 for patients undergoing TSA vs. 11.2 for patients undergoing RSA). Unfortunately, revision rates were NR for RCI shoulders specifically; however, the RSA group experienced a higher revision rate of 21.2% (7/33 shoulders) vs. 12.8% (11/86 shoulders) for the TSA group. In comparison to these studies, results from our analysis demonstrated >10 differences in mean flexion and abduction improvements favoring RSA, but conversely >10 differences in mean abduction and external rotation at the final FU favoring TSA. Mean Constant score improvement was also 7.8 points higher in patients undergoing RSA, but the mean ASES score at the final FU was 7.9 points higher in patients undergoing TSA. Results from the meta-analysis demonstrated a higher complication rate of 9% in patients undergoing TSA than the 6% in patients undergoing RSA but similar revision rates of 2% in TSA and 1% in RSA.
Limitations of this study include that this review is dependent on the data reported by each study and its quality. In studies that included only patients with B2 glenoids, data collection was limited to which outcome variables the investigators chose to collect and report. In instances where multiple Walch glenoid types were studied, data collection was further limited to the degree to which the investigators stratified data based on morphology. In the articles included in this review, some form of outcome data specifically pertained to patients with B2 glenoids, although for many of these articles, demographics data were not stratified.
Many articles were excluded from this review for not specifying the patterns of glenoid erosion or, if specified, not stratifying outcomes for the B2 subgroup. In addition, 3 TSA studies included in our review did not specify the glenoid correction technique limiting our subgroup analysis. Most articles also did not specify mean glenoid retroversion or degree of humeral head subluxation, making it difficult to draw conclusions and form recommendations regarding treatment choice for specific ranges of retroversion. This review is also limited by the small sample size in the RSA group, due to a lack of literature on outcomes after RSA for RCI B2 glenoids. All studies except one included in this review were classified as level III or IV evidence. Finally, outcomes after a procedure are likely affected not only by a surgeon's overall experience and skill but also by their experience and skill in a specific procedure.
Strengths of the study are that this is the most comprehensive review of the literature regarding outcomes after TSA and RSA in patients with B2 glenoids without significant rotator cuff pathology. We used well-defined search criteria to identify articles with pertinent, applicable data. We included multiple objective and patient-reported outcome measures to compare the results of various surgical treatments. In addition, we took care to stratify results based on RSA vs. TSA, but we also further stratified the TSA group based on the correction technique used.

Conclusion
This systematic review found comparable outcomes following RSA, TSA þ ER, and TSA þ PAGC for the treatment of patients with Walch B2 glenoids. Similar mid-term outcomes can be expected in regard to ROM improvement, functional score improvement, and revision rate if TSA þ ER, TSA þ PAGC, or RSA is chosen. A similar complication rate between TSA þ ER and RSA can also be expected; however, the use of TSA þ PAGC may increase the risk of glenohumeral subluxation. Overall, the results are limited by a lack of specificity on the severity of retroversion in studies. Treatment choice, therefore, should be made through a shared decisionmaking process and based on surgeon preference, patient age, activity level, and potentially by selecting the best plan with the assistance of newer preoperative planning tools. There remains the need for a randomized controlled trial with a large sample size and long-term FU to determine if any difference in outcome can be expected between these treatment options. Future studies are also indicated to determine the impact of initial glenoid version correction on outcomes in the B2 glenoid, particularly how much is absolutely necessary for proper implant stability as well as more detailed risks associated with correction techniques. Finally, future investigations of patients undergoing RSA and TSA should aim to report preoperative Walch classification using standardized preoperative imaging (CT scans) for measurement, subjective and objective data, and stratify outcomes based on accurate classification of the Walch morphology to allow for more accurate comparison across studies.

Disclaimers
Funding: No funding was disclosed by the authors. Conflicts of interest: The authors, their immediate family, and any research foundation with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.