Medialized versus Lateralized Center of Rotation in Reverse Total Shoulder Arthroplasty: A Systematic Review and Meta-Analysis

One of the original biomechanical principles of reverse total shoulder arthroplasty (RTSA) is medialization of the center of rotation (COR) relative to the native level of the glenoid. Several authors have proposed the lateralized center of rotation, which is characterized by a lateralized (L) glenoid and medialized (M) humeral component. The aim of this review is to compare the clinical and functional outcomes of COR in medialized (M-RTSA) and lateralized (L-RTSA) RTSA in patients with uniform indications and treatment through a meta-analysis. A PRISMA-guided literature search of PubMed, Medline, Embase, Scopus, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews and Cochrane Clinical Answers was conducted from April to May 2021. Twenty-four studies were included in the qualitative synthesis, and 19 studies were included in the meta-analysis. Treatment with RTSA resulted in positive post-operative outcomes and low complication rates for both groups. Statistically relevant differences between L-RTSA group and M-RTSA group were found in post-operative improvement in external rotation with arm-at-side (20.4° and 8.3°, respectively), scapular notching rates (6.6% and 47.7%) and post-operative infection rates (1% and 7.7%). Both lateralized and medialized designs were shown to improve the postoperative outcomes. Nevertheless, a lateralized COR resulted in greater post-operative external rotation.


Introduction
Rotator cuff disorders are the most common cause of disability related to the shoulder [1,2]. Full-thickness rotator cuff tears are present in approximately 25% of individuals over 60 years old and 50% of people older than 80 years [2,3].
Reverse total shoulder arthroplasty (RTSA) is the procedure of choice for treatment of glenohumeral joint disease among patients with severe rotator cuff deficiency [4,5].
To our knowledge, evidence identifying the best location for the COR is lacking, and no recent systematic reviews comparing the two prosthesis designs in a population of patients with standardized indications for RTSA have been reported in the literature.
The aim of this review and meta-analysis is to compare outcome measures, the number of revisions, the number of complications, scapular notching, and the active range of motion (ROM) between M-RTSA and L-RTSA.

Study Selection
The research question was formulated using a PICOS-approach: Patient (P); Intervention (I); Comparison (C); Outcome (O) and Study design (S). The aim of this systematic review is to describe whether patients (P) that underwent RTSA (I) with a lateralized COR reported better clinical and functional results compared to a medialized COR (C). The outcomes (O) assessed were: ROM, American Shoulders and Elbow Surgeons (ASES) score, Simple Shoulder Test (SST), Oxford Shoulder Test (OST), Absolute Constant-Murley score, Visual Analog score (VAS) for Pain, scapular notching, complications, revisions and self-assessed satisfaction.

Inclusion Criteria
Only articles published in English were screened. Peer-reviewed articles of each level of evidence according to Oxford classification were considered. Studies reporting patients undergoing a primary RTSA were included, and the studies were considered eligible if they focused on a medialized or lateralized COR, or on both. The indication for RTSA of the patients enrolled in the trials was limited to cuff tear arthropathy, irreparable cuff tear or cuff tear associated with osteoarthritis.

Exclusion Criteria
Technical notes, letters to editors, instructional courses, or studies, including procedures other than reverse shoulder arthroplasty, were excluded. Articles were discarded if the mean follow-up was less than 12 months. Studies that considered revision RTSA, shoulder hemiarthroplasty and arthroscopic shoulder procedures, and RTSA combined with concurrent tendon transfer, were not included. In addition, articles reporting outcomes of patients with rheumatoid arthritis, acute fracture, post-traumatic fracture sequelae, tumors or active infection were not considered. In vitro, animal, cadaver and biomechanical studies were excluded. Studies that did not specify either the prosthesis design or COR or that were missing data were excluded.
Finally, data regarding patients who underwent procedures such as BIO-RSA were not considered in the statistical analysis, due to lack of standardization of the latter procedure; thus, in articles where the L COR group comprised such procedures, only data from the M COR group were included in the meta-analysis.

Search
A systematic review was performed using the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [32]. Medline, EMBASE, Scopus, CINAHL and CENTRAL bibliographic databases were searched using the following string: ((scapular notching) OR (notching) AND (reverse shoulder arthroplasty) OR (reverse total shoulder) AND (medialized center of rotation) AND (lateralized center of rotation) AND (cuff tear arthropathy) OR (rotator cuff tear) OR (rotator cuff tear arthropathy)). Keywords were used both isolated and combined. Additional studies were searched among reference lists of selected papers and systematic reviews.
The search was performed by two of the authors (B.B. and L.A) from April to May 2021, and articles from the inception of the database to May 2021 were searched.

Data Collection Process
Data extraction was performed by two independent reviewers (B.B. and L.A.), and differences were reconciled by mutual agreement. In case of disagreement on inclusion or exclusion of articles, a third reviewer (S.D.S.) was consulted. The same authors (B.B. and L.A.) performed the review and organization of the titles in order to limit the bias.
The reviewers used the following screening approach: title and abstract were reviewed first, then the full articles. The full text of papers not excluded was evaluated and eventually selected after a discussion between the reviewers. In case of disagreement, the third reviewer (S.D.S.) decided. The number of articles included or excluded were registered and reported in the PRISMA flowchart. Standards reported by Moher et al. were adhered to in designing the PRISMA chart [29].

Data Items
General study characteristics extracted were: primary author, year of publication, type of study, level of evidence (LOE), sample size, mean age, gender totals and number of shoulders treated (Table 1). Moreover, prosthesis design, surgical approach, surgical characteristics (glenosphere size, glenoid tilt and humeral neck shaft angle, all sorted by M and L CORs) and follow-up were considered (in case of multiple time points, only the last follow-up was reported) ( Table 2).

Study Risk of Bias Assessment
Given the designs of the included studies, the Risk of Bias (RoB 2) tool for randomized trials and the Risk of Bias in Non-Randomized Studies of Interventions (ROBINS-I) tool by Cochrane were used to assess the quality of each study [43,44]. Selected articles were independently rated by each reviewer (B.B., A.L.) and verified by a third one (S.D.S.) in case of disagreement.

Statistical Analysis
Categorical data were summarized as frequencies with percentages. Continuous data were summarized as mean values and standard deviations (SD) or ranges (minimum and maximum values). The subgroup meta-analysis was performed using a fixed-effect model or random-effect model, as applicable, and the Der-Simonian and Laird method for the estimation of the between-study variance. To quantify the heterogeneity among the studies, the I 2 statistic was applied, with 50% defined as the threshold for significant heterogeneity [45]. A p-value less than 0.05 was considered statistically significant. All statistical analyses were performed using R software version i368 3.6.1.

Study Selection
The literature search identified 471 articles. No additional studies were found in the grey literature, and no unpublished studies were retrieved. Duplicate removal resulted in the exclusion of 58 studies, leaving 413 articles for screening, and 358 articles were excluded based on the title and abstract. Fifty-five articles were screened by full text, and 31 were excluded (insufficient outcomes, n = 2; unclear data, n = 2; no specified COR, n = 2; potential mix of data, n = 9; indications for RTSA as fractures, infection, inflammatory arthritis, tumors or revision arthroplasty n = 15; and patients with concurrent tendon transfers = 1). At the final screening, 24 articles met the selection criteria and were included in the review. The PRISMA flow-chart of literature search is reported in Figure 1.

Study Characteristics
The level of evidence (LOE) of each of the included was: two level I Randomized Control Trials [25,46]
The 24 studies reviewed included a total of 2199 patients and 2276 shoulders, which were evaluated at a minimum follow-up of 12 months. The sample size for the M-RTSA group was 1138 patients, while the L-RTSA group comprised 1061 patients.
The risk of bias assessments for both RCTs and NRCTs are reported in Figures 2 and 3.
The glenosphere size reported ranged from 32 mm to 46 mm in diameter for both M and L COR groups, while the humeral neck shaft angle reported in the M group ranged from 150 • to 155 • and in the L group from 135 • to 155 • .

Outcome Measures
The data collected about outcome measures were Constant-Murley Score, ASES, SST, OSS and VAS scores. The Constant-Murley Score was collected from eight articles [12,15,26,43,46,49,51,52], the ASES score from five studies [19,43,44,51,53], the SST score from three studies [19,43,44] and the OSS score from two studies [48,50]. No articles on L-RTSA included in the meta-analysis assessed the OSS score; therefore, this questionnaire was not evaluated. No low and moderate risk of bias article measured the VAS; consequently, this score was not assessed. Constant-Murley Score, ASES and SST scores were included in the meta-analysis. No statistically significant differences between M and L in Constant-Murley Score, ASES and SST scores were found (p = 0.40, p = 0.96 and p = 0.76, respectively).

Discussion
The use of RTSA for rotator cuff arthropathy has steadily increased over time due to improved medium-term outcomes [30,54]. The very first RTSA designs were fraught with

Discussion
The use of RTSA for rotator cuff arthropathy has steadily increased over time due to improved medium-term outcomes [30,54]. The very first RTSA designs were fraught with glenoid implant failures. This problem was largely solved by Grammont's design of a medialized COR since it converted shear forces across the glenoid implant-bone interface into compression forces. However, a medialized COR presented a separate set of issues such as scapular notching, limited rotation, instability and poor cosmesis of the shoulder. As glenoid fixation improved with the advent of porous ingrowth posts, central compression screws and locking fixation, RTSA designs began to lateralize the COR to address the issues seen with medialized COR implants. The purpose of this study was to determine which implant design-medial COR or lateral COR-provided better outcomes. We found that both implant designs predictably improve patient's function and reduce their pain. However, our review shows that implants with a lateral COR may provide better rotation and less notching, with no increased incidence of implant failure.
Our findings are consistent with previous studies. Sirveaux et al. [30] reviewed 80 shoulders with a mean follow-up of 44 months. In this study, they reported pain relief in 96% of cases and a significant improvement of the Constant score. However, the study reported scapular notching in 63.6% of cases. Jobin et al. [18] reviewed 37 prostheses, at a mean follow-up of 16 ± 10 months, and noted good post-operative ASES scores. The main limitation of RTSA noted in the study was also scapular notching (reported in 68% of patients). L-RTSA was introduced to overcome this limitation, improving ROM and prosthesis stability [13]. Mollon et al. [40] retrospectively evaluated the results of RTSA with Equinox (Exactech, Inc., Gainesville, FL, USA), reporting a significant increase in abduction and internal rotation.
Several systematic reviews have previously compared L-and M-RTSA. Samitier et al. [5] has reported improvement in external rotation with the arm-at-side for L COR procedures using only the prostheses Reverse Shoulder Prosthesis (DJO Surgical, Austin, Texas), and Arrow Anatomical Shoulder System (Mulhouse, France) compared to the M COR ones in agreeance with our findings. However, their study did not include a meta-analysis and, in addition, the sample of patients in the current systematic review is greater and more homogeneous in terms of indication. The studies by Streit et al. [31] and Helmkamp et al. [23] also reported external rotation improvement in L-RTSA patients. Furthermore, a statistically significant difference was found in terms of scapular notching rates (p < 0.05), with a lower reported incidence for the L COR prostheses ( Figure 6). These results are in line with the study by Alentorn-Geli et al. [55]. However, there was a discrepancy between the studies regarding the indication for RTSA: the current authors considered only cuff tear arthropathy, while Alentorn-Geli et al. did not apply any limitation in terms of inclusion criteria. Consistent with the results of the present systematic review, Heinkamp et al. also found scapular notching rates to be higher in the M group (L = 4.3%, M = 49%) [23].
No statistically relevant differences have been reported in the available literature on complications and revision rates [33,55]. The only statistically noteworthy divergence was found in infection rate, which was higher in the M-RTSA group (p < 0.01). This is difficult to explain and is most likely due to the fact that the first RTSA's performed were with M-RTSA implants. Therefore, the indications were skewed to more difficult cases and the operative time was most likely longer given the learning curve of using RTSA. To our knowledge, no other systematic review reported statistically relevant differences in the infection rates between the two COR groups.
The strength of this review is the homogeneity of patients of the included studies. Only articles on patients who had undergone RTSA for rotator cuff arthropathy or an irreparable cuff tear associated or not with osteoarthritis were selected. Nevertheless, rotator cuff arthropathy is the most reliable indication for RTSA [36,[56][57][58][59].
Additionally, data regarding BIO-RSA procedure were not included in the metaanalysis [12,26,29]: these techniques are not standardized and controlling for lateralization is difficult. These techniques can also be used to simply replace lost bone (i.e., B2 or B3 glenoid) without truly lateralizing the glenoid center of rotation.
Furthermore, all the considered articles presented a minimum follow-up of 12 months and revision surgeries were not considered. The studies were subjectively evaluated by the Cochrane risk of bias tools, RoB 2 and ROBINS-1 [43,44], in order to assess their potential risk of bias, and articles judged as having a serious or critical risk of bias were not included in the meta-analysis.
This study has some limitations. Firstly, the overall level of evidence of the studies included is low due to the limited presence of RCTs comparing M-RTSA and L-RTSA populations [26,51]. Moreover, the NRCTs included ranged from "low" to "critical" risk of bias according to ROBINS-I, thus only permitting the inclusion of low risk of bias and moderate risk of bias articles in the meta-analysis.
Furthermore, this systematic review does not consider the variability in global lateralization of implants [60]. Data regarding glenoid and humeral lateralization were combined into one group. For this reason, additional studies comparing the outcomes of humeral and glenoid lateralization are required, as recently proposed by Nabergoj et al. [61]. However, these functions differ in regard to force needed to generate overhead ROM as well restoration of ER, which is one of the main findings of the authors results. More control for implant type and lateralization would likely result in more meaningful data interpretation. Additionally, BIO RSA procedures were not included by the authors/in the review, nor was the role of the humeral neck shaft angle subject of evaluation.
The small sample size of some included articles downgraded the overall quality of the results. As observational studies constituted the main source for the analysis, selection bias and confounding due to diverse expectations in RTSA patients should be taken in consideration. Moreover, the statistically significant difference in reinfection between lateralized and medialized RSA should be influenced by the different duration (lateralized RSA (2011-2021) and medialized RSA (2007-2012)). In addition, the heterogeneous lengths of follow-up in the examined studies may contribute further inconsistencies. Furthermore, only English studies were included, limiting the available number of articles eligible for this review.

Conclusions
Both L and M designs have been shown to improve postoperative outcomes following RTSA. Nevertheless, it appears that the use of a lateralized COR is more likely to result in greater post-operative external rotation and lower rates of scapular notching. Further high-quality studies are required to compare L-RTSA with M-RTSA.