INTRODUCTION
Reverse shoulder arthroplasty (RSA) is the most reliable surgical treatment for patients with cuff tear arthropathy and irreparable rotator cuff tears with pseudoparalysis [1]. Increased mechanical advantage of the deltoid muscle provided by RSA improves overhead shoulder elevation in posterosuperior rotator cuff tears, but is limited in reconstituting external rotation (ER) deficits [2-6]. Thus, patients with combined loss of active elevation and external rotation (CLEER) as a result of a compromised posterosuperior rotator cuff may not fully recover function with RSA alone [4]. Past studies using the Grammont style prosthesis with a 155° neck-shaft angle showed little to no improvement in ER following RSA in patients with CLEER [7,8].
Latissimus dorsi transfer (LDT) combined with RSA has been utilized to improve functional outcomes in patients with CLEER deficits with reliable results [3,4,9-14]. However, LDT has several downsides including greater surgical complexity, risk of nerve injury, and theoretical loss of internal rotation. While LDT is still frequently performed, some surgeons contend that modern lateralized RSA (relative to the original Grammont design, with a glenoid lateral offset [LO] ≥5 mm and/or a humeral LO ≥5 mm of the Delta III) [15] alone is sufficient to restore ER in patients with preoperative insufficiency through tensioning of the remaining posterior rotator cuff and deltoid [16-20]. However, only a limited number of studies have compared restoration of ER in patients with preoperative insufficiency after RSA with LDT versus lateralized RSA alone.
The goal of this systematic review and meta-analysis was to compare clinical outcomes and complication rates following RSA with LDT versus lateralized RSA alone in patients with preoperative ER deficiency. We hypothesized that greater improvement in postoperative range of motion (ROM) would be observed following LDT with RSA, with similar patient-reported outcome scores to RSA alone.
METHODS
No institutional review board approval was obtained, given that the review did not involve experimentation on human or animal subjects, and the data reviewed are public. This systematic review was performed in accordance with the guidelines for Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [21].
Eligibility Criteria
We included original studies written in English and published prior to January 2023 evaluating the use of RSA with LDT or lateralized RSA alone in patients with ≤0° of active ER with the shoulder adducted to the side [22]. We excluded duplicate studies, non-English text, reviews and meta-analyses, commentaries and editorials, purely radiographic studies, studies that solely evaluated surgical technique, non-human studies, biomechanical studies or those without clinical variables, studies in which greater than 10% of the cases included were post-traumatic, allograft-prosthetic composite reconstruction or tumor cases, and those in which the patient population had preoperative deltoid dysfunction or axillary nerve palsy. To maximize inclusion, functional outcomes and the incidence of postoperative complications were assessed independently (Fig. 1). Thus, studies reporting only functional outcomes or only complications were included.
Search Strategy
We queried the PubMed/Medline, Embase, Web of Science, and Cochrane databases using search terms including: “tendon,” “transfer,” “L’Episcopo,” “teres,” “latissimus,” “reverse,” “inverted,” and “shoulder” (Supplementary Material 1 for database-specific search strategies). Exclusion criteria were applied during title and abstract screening. Full texts of the remaining articles were then reviewed by several authors (CTG, KAH, KMH, TRR, VEB), with expert opinions sought from senior shoulder and elbow surgeons to determine inclusion when questionable.
Study Outcomes
Study variables were selected based on their relevance to CLEER pathology and the frequency with which each variable was reported in the literature. The primary functional outcomes of interest were active ER, active forward elevation (FE), and Constant score. These outcomes were chosen because they are most commonly reported in the body of literature queried in this study. Postoperative complications included were tear in tendon transfer, revision tendon repair, baseplate loosening, peri-prosthetic fracture, neuropraxias and nerve injuries, dislocation, and infection. We also noted radiographic evidence of humeral cortical erosion at the tendon transfer site when reported. Although highly relevant, we anticipated substantial heterogeneity in the reporting of IR, as described previously, and thus it was considered a secondary outcome of this study [23]. IR is variously reported as degree measurements, the most cephalad vertebral level reached by the thumb, and in points according to various scoring systems, with no predominant method in the literature.
Data Extraction
Data extraction was completed by multiple authors (CTG, KMH, TRR, VEB) using a standardized data-collection form. We recorded level of evidence, country of the corresponding author, whether the study was retrospective or prospective, number of patients and shoulders, mean length of follow-up, average age, sex, pre- and postoperative active ROM (ER, FE, and IR), postoperative Constant score, prostheses implanted, whether teres major transfer (TMT) was performed concomitantly with LDT, and relevant complication rates. Prosthesis design was classified as globally medialized versus globally lateralized per the classification system developed by Werthel et al. [15] This classification defines lateralization relative to the original Grammont design (Delta III, DePuy).
Data Analysis and Synthesis
We summarized study characteristics descriptively and calculated weighted means and summary statistics. Our primary analysis compared postoperative values and pre- to postoperative improvement in active ER, active FE, and Constant score between RSA with LDT and Lateralized RSA alone cohorts. A sub-analysis was performed to further stratify the RSA with LDT cohort based on whether the implant used was globally medialized or lateralized in design. The incidence of complications was assessed in a separate analysis. Secondarily, we summarized postoperative IR descriptively and assessed whether it achieved or exceeded the minimum necessary internal rotation (MNIR) needed to perform activities of daily living. The MNIR was previously reported to be the L3 vertebral level [24]. Functional outcomes (ER, FE, and the Constant score) were also compared based on implant design and whether LDT was concomitantly performed. Studies were included for meta-analysis if they reported one or more of the outcomes of interest and provided stratum-specific data. Therefore, studies that utilized multiple implants of varying design or included a mixed cohort without reporting functional outcomes separately for these strata were not included. We anticipated that the designs of the included studies and methodology involved in data collection would result in substantial heterogeneity and thus elected to use a random-effects model a priori [25]. The I2 statistic was used to assess the heterogeneity of results. The true effect size in 95% of the population (95% prediction interval) was calculated using the variance of true effects (T2) and thus the standard deviation of true effects (T). Meta-analysis was performed using the Metafor package [26].
Although mean preoperative active ER <0° was chosen as our inclusion criteria, we anticipated that patients who underwent RSA with LDT might still have lower preoperative ER compared to patients who underwent lateralized RSA alone. Thus, we elected to perform a random-effects multivariable meta-regression in an effort to adjust for preoperative ER and whether the subscapularis was repaired. Meta-regression was performed using the meta package [27]. All statistical analyses were performed using R Software (version 4.2.0, R Core Team) with an α <0.05.
RESULTS
Search Results
Our literature search for RSA with LDT returned 394 publications, of which 199 were unique. We excluded 171 articles during title and abstract screening, leaving 28 articles for full-text review. Following full-text screening, 12 unique articles remained. Separately, the literature search for lateralized RSA without LDT returned 2,237 publications, 946 of which were unique. We additionally excluded 690 articles during title and abstract screening, leaving 256 articles for full-text review. Following full-text screening, four articles remained. In total, 16 studies were identified for inclusion. A bias assessment was completed using the methodological index for non-randomized studies (MINORS) criteria [28]. Supplementary Table 1 shows individual studies and MINORS scores.
Study Characteristics
Of the 16 included articles, most were authored in France (8/16, 50%). The levels of evidence reported were II (1/16, 6%), III (5/16, 31%), and IV (10/16, 63%). Of these, 2/16 (11%) were prospective comparative studies, 9/16 (56%) were retrospective comparative studies, and 7/16 (44%) were retrospective case series. MINORS analysis revealed non-comparative studies had an average score of 12.3 of 16, while comparative studies had an average score of 18 of 24.
Patient and Surgical Characteristics
Outcomes were reported in 12 RSA with LDT articles, encompassing 188 shoulders (63% female; weighted mean age, 70 years; and weighted mean follow-up, 50 months) and four lateralized RSA alone articles including 250 shoulders (63% female; weighted mean age, 72 years; and weighted mean follow-up, 38 months). The “lateralized” group consisted of four very highly lateralized RSAs (VHL-RSA) and five lateralized RSAs (L-RSA). All RSAs in the lateralized group were onlay humerus designs. The “medialized” group (only comprised of RSA with LDT) consisted of five medialized RSAs (M-RSA), as well as three minimally lateralized RSAs (ML-RSA) (Supplementary Table 1), with the total greater than 16 due to several studies using multiple implants [15]. All RSAs in the medialized group were inlay humerus designs. Patients most commonly had a diagnosis of cuff tear arthropathy (58%), followed by osteoarthritis with rotator cuff insufficiency (33%) and massive rotator cuff tear (14%). One study [29] provided multiple diagnoses per patient when applicable.
RSA with LDT vs. Lateralized RSA Alone
In the RSA with LDT cohort, the weighted mean preoperative ER was –14° (range, –42° to 0°) and 22° postoperatively (range, 10°–39°), while FE was 70° preoperatively (range, 59°–116°) and 140° postoperatively (range, 123°–163°). The weighted mean postoperative Constant score was 65 (range, 59–72). There were 148 patients among nine studies with postoperative IR reported (Table 1). Thirty-four percent of included patients reported a mean IR that achieved the MNIR (i.e., L3 vertebral level). In the lateralized RSA cohort, the mean ER was –11° preoperatively (range, –18° to 0°) and 27° postoperatively (range, 25°–32°), FE was 76° preoperatively (range, 53°–83°) and 140° postoperatively (range, 137°–149°). The mean postoperative Constant score was 70 (range, 67–71). There were 250 patients among four cohorts with postoperative IR reported. Four percent of included patients reported a mean IR that achieved the MNIR.
Meta-analysis revealed significantly greater postoperative ER after lateralized RSA alone compared to RSA with LDT (28° [25°–32°] vs. 22° [19°–26°], P=0.010) (Fig. 2A). However, pre- to postoperative improvement in ER was identical after lateralized RSA alone and RSA with LDT (37° [32°–43°] for both, P=0.985) (Fig. 2B). When controlling for mean preoperative ER and percentage of patients with subscapularis repair using multivariable meta-regression, RSA with LDT was associated with poorer postoperative ER (β=–6.2 [–10.8 to –1.6], P=0.014) but not pre- to postoperative improvement in ER (P=0.230) (Table 2). Meta-analysis revealed no difference in postoperative FE or pre- to postoperative improvement in FE (Fig. 3). However, lateralized RSA alone had a more favorable postoperative Constant score compared to RSA with LDT (69 [67–72] vs. 65 [62–67], P=0.014) (Fig. 4).
Sub-analysis: RSA with LDT Based on Implant Lateralization
Sub-analysis revealed that postoperative ER differed between lateralized RSA alone and RSA with LDT using medialized and lateralized implants (28° [25°–32°] vs. 23° [20°–25°] vs. 21° [15°–27°], P=0.014) (Fig. 5), but there was no difference in pre- to postoperative improvement in ER (P=0.525). Multivariable meta-regression demonstrated less favorable postoperative ER for lateralized RSA with LDT compared to lateralized RSA alone (β=–6.9 [–12.6 to –1.2], P=0.023) (Table 3). Sub-analysis revealed no difference in postoperative FE or pre- to postoperative improvement in FE (Fig. 6). Postoperative Constant score was greatest for lateralized RSA alone, followed by medialized and lateralized RSA with LDT (69 [67–72] vs. 67 [63–70] vs. 63 [60–67], P=0.038) (Fig. 7).
Complications
Complications were reported in 9 of 12 RSA with LDT studies (144 shoulders). Complications included tear in the tendon transfer (n=3, 2.1%), revision tendon repair (n=1, 0.7%), nerve-related complications (n=3, 2.1%, [transient neuropraxia, n=3, one axillary, two radial), dislocation (n=4, 2.8%), and infection (n=5, 3.5%). Radiographic evidence of humeral cortical erosion at the transfer site was reported in 19% (n=28) of RSAs with LDT. Separately, complications were reported in 2/4 lateralized RSA alone studies (125 shoulders). Complications included scapular spine fracture (n=2, 1.6%), acromial fracture (n=1, 0.8%), dislocation (n=1, 0.8%), infection (n=4, 3.2%), and unexplained pain (n=1, 0.8%).
DISCUSSION
In this systematic review and meta-analysis, we demonstrated that both lateralized RSA alone and RSA with LDT can restore ROM in ER and FE and yield satisfactory Constant scores in patients with preoperative ER deficits. Patients who underwent lateralized RSA alone had superior postoperative ER and Constant scores compared to RSA with LDT, but there was no difference in postoperative FE. However, there was no difference in pre- to postoperative improvement for ER or FE. The use of a medialized versus lateralized RSA with concomitant LDT did not appear to influence outcomes. In patients who received an LDT, our meta-analysis suggested that the rate of tendon transfer tear was 2.1%, that of nerve-related complications was 2.1% (vs. 0%), and dislocation rate was 2.8% (vs. 0.8%).
Whether patients with preoperative deficits in ER (often in the setting of CLEER) undergoing RSA require concomitant LDT to attain sufficient ER is contentiously debated. Favre et al. [37] demonstrated in vitro that the posterior deltoid moment arm after RSA implantation is biomechanically inadequate to achieve active ER. Consequently, for patients with a deficient posterior rotator cuff, many surgeons believe lateralized RSA alone is insufficient to achieve satisfactory active ER. Boileau et al. [36] initially found that RSA with LDT provided sufficient torque for ER in patients with deficient infraspinatus and teres minor muscles, but this finding was challenged by Berglund et al., [38] who reviewed 33 lateralized RSAs in patients with preoperative active ER <0° and at least two years follow-up, finding no relationship between teres minor atrophy and the restoration of ER.
Prior studies attempted to evaluate whether LDT confers a benefit to RSA alone using modern techniques, but failed to account for the influence of implant design [22,35]. Young et al. [35] performed a randomized controlled trial of 28 patients with cuff tear arthropathy and CLEER pathology who received RSA alone or in combination with a LDT. At 2-year follow-up, no differences were found between RSA with LDT and RSA alone with regard to active ER, IR, FE, and patient-reported outcome scores. However, only 12 of 28 patients were assessed for postoperative ER at 2-year follow-up. Furthermore, Young et al. [35] utilized two different implant systems: the Zimmer Trabecular Metal Reverse Shoulder System and Biomet Comprehensive Reverse Arthroplasty System (Zimmer Biomet). According to the classification system published by Werthel et al. [15], the former is ML and the latter is HL. Loss to follow-up or confounding due to utilization of heterogeneous implant designs may have obscured the influence of the LDT on postoperative ER.
Wiater et al. [22] retrospectively evaluated 31 patients who underwent RSA with LDT and compared them to 33 patients that underwent RSA alone. Both cohorts were similar preoperatively except for marginally poorer active ER in the RSA with LDT group (−8° vs. 0°, P=0.004). Postoperatively, ER was similar between RSA with and without LDT (14°±13° vs. 17°±15°, P=0.43). Regarding implant utilization, patients in the RSA with LDT group received a Zimmer trabecular metal reverse humeral stem (Zimmer) with either a Zimmer reverse glenosphere/baseplate (n=23) or a DJO Encore reverse glenosphere/baseplate. All patients in the RSA alone group received a Zimmer reverse humeral stem and glenosphere/baseplate, which is an ML design per Werthel et al. [15] It is unclear from these data whether use of a lateralized RSA design in the cohort of RSAs without LDT would have resulted in the maintenance of superior ER postoperatively.
Although we attempted to design a fair comparison between RSA with LDT and lateralized RSA alone cohorts by requiring the mean preoperative ER of included studies to be ≤0°, the preoperative mean ER in the RSA with LDT cohorts was lower than the lateralized RSA alone cohort (–14° vs. –11°). This represents selection bias present in the literature and may indicate surgeon preferences to reserve LDT for patients with CLEER pathology and severe ER. After accounting for the mean preoperative ER of included studies in our meta-regression analysis, we found no effect on pre- to postoperative improvement in ER based on whether RSA with LDT or lateralized RSA alone was performed. Although this result suggests that lateralized RSA alone may be sufficient for patient populations with very poor preoperative ER, meta-regression analyses are limited by the range of included data. Therefore, we encourage readers to interpret these findings cautiously, and we advocate for either a case-matched cohort study or prospective trial to definitively determine whether lateralized RSA alone is sufficient to restore ER in deficient patients.
We also found no difference in outcomes when comparing lateralized RSA and LDT with varying prosthesis design. Although prior studies have suggested lateralization alone may restore ER [39,40], these studies were not exclusively comprised of patients with preoperative CLEER or active ER ≤0°. Lateralized designs have been purported to increase ER by improving the length-tension of the rotator cuff, increasing the deltoid wrap, and optimizing the deltoid moment arm [41-44]. Berton et al. [45] in their systematic review of 24 studies and meta-analysis of 19 studies, found significantly greater postoperative ER in lateralized compared to medialized RSA (20.4° and 8.3°, respectively). While lateralization alone may be sufficient in many patients to restore active ER, patients who lack a posterior rotator cuff may receive less benefit and require LDT for functionally successful outcomes. Although Berglund et al. [38] found no association between status of the posterior rotator cuff and restoration of ER using a lateralized design, future studies are needed to corroborate this finding.
Another factor to consider is that TMT is often performed concomitantly with LDT, and our RSA with LDT cohort was comprised of both patients with and without TMT. Many authors have demonstrated that concomitant TMT with LDT plays an important role in ER strength after RSA [8,37]. However, Hones et al. [46] conducted a systematic review and meta-analysis of 19 studies and found no significant differences in ER or Constant score when comparing patients who underwent RSA with LDT alone versus LDT with TMT.
When deciding whether to perform an LDT, surgeons must consider that active ER improvement may come at the cost of higher complication rates. While reported complication rates for primary RSA vary, the overall complication rate is commonly quoted to be approximately 15% [47,48]. Hones et al. [46] reported a complication rate of 14% in their systematic review and meta-analysis of RSA with LDT, and Puskas et al. [5] reported a 22% orthopedic complication rate in their series of 41 RSAs with LDT. There is a higher risk of nerve injury with LDT given the necessary dissection for safe release and transfer of the tendon. Currently available studies of RSA with LDT report the incidence of transient nerve palsy to be 3.1%– 3.4% [46,49], which is markedly higher than that of primary RSA, which has been shown to be about 1.3% [50]. In our analysis, none of the included lateralized RSA alone cohorts reported a postoperative nerve injury.
Another consideration when planning LDT is prosthesis dislocation. We observed a dislocation rate of 2.8% in patients receiving an LDT and 0.8% in lateralized RSA alone. While the higher dislocation rate could be related to the LDT in that cohort, the inclusion of studies that utilized a medialized design may also be responsible. Dislocation rates of primary RSA without tendon transfer have generally ranged from 2% to 5% in the recent literature [51-53]. Subscapularis repair may increase the anterior stability of the RSA and decrease the dislocation rate [54,55]. All RSA with LDT cohorts included in the present study underwent routine subscapularis repair. While operative time and blood loss is not routinely reported in the literature and we therefore did not attempt to evaluate it, longer operative time and greater blood loss are both theoretical downsides of RSA with LDT due to additional tendon harvest. Unique to LDT, we reported a humeral cortical erosion rate of 19%. This may reflect length of follow-up, as Bonnevialle et al. [2] reported an osteolysis rate of 77% after L’Episcopo transfer in 13 patients at a mean of 37 months postoperatively, with osteolysis appearing to worsen over time. However, there was no ultimate impact on clinical outcomes in their series. Furthermore, Valenti et al. [33] evaluated 24 RSAs with LDT at a mean follow-up of 45 months and found that eight (33%) patients demonstrated intact lateral cortex, eight (33%) irregular and eight (33%) lytic lesions, with no differences in clinical outcomes between patients with and without bony lesions.
This systematic review and meta-analysis has several limitations. Limitations include those inherent to all reviews, i.e., the analysis of many retrospective studies with possible individual, compounded reporting bias and the quality of the review being dictated by the quality of the individual studies. Moreover, the included studies had varying follow-up periods. We were unable to statistically analyze IR due to heterogeneity in reporting, which remains an important limitation in comparisons of these procedures, as loss of the latissimus dorsi with or without the teres major as internal rotators is a consideration. While implants were classified according to the Werthel et al. [15] global lateralization classification, modularity of implants may also influence lateralization of the center of rotation [56]. While it is important to attempt to make a fair comparison between groups, the relative rarity of preoperative active ER ≤0° in patients undergoing RSA limited the sample size significantly, and larger populations are necessary to draw stronger conclusions. Furthermore, our cohorts did not have equivalent preoperative ER despite our efforts, which further limits our conclusions. Although we performed a multivariable meta-regression analysis to attempt to control for unequal preoperative ER, our results should be interpreted with caution. We queried highly utilized databases with broad search terms and no time limitation, nevertheless, relevant articles may have been missed. We also did not set a minimum level of evidence as an inclusion criterion given that the majority of studies of RSA with LDT were anticipated to be small retrospective case series.
CONCLUSIONS
Both RSA with LDT and lateralized RSA are reliable options to restore ER in patients with significantly limited preoperative ER. In our meta-analysis and meta-regression, lateralized RS alone was associated with superior postoperative ER and Constant score compared to RSA with LDT, but similar postoperative FE with no differences in pre- to postoperative improvement in ER or FE. Compared with lateralized RSA alone, addition of LDT comes with increased complication risks, namely with regard to nerve injury and dislocation. Overall, addition of an LDT may be necessary in certain patient populations, but our results suggest that lateralized RSA alone can be sufficient to restore ER. Future comparative studies are needed to further ascertain the role of LDT in treating patients with severe ER deficiency indicated for RSA.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Full text via PMC
Download Citation
Supplement
Print
