Lateralization and distalization shoulder angles may not predict clinical outcomes in reverse total shoulder arthroplasty: a systematic review and meta-analysis

Abstract

Background

The purpose of this study was to evaluate the prognostic value of the lateralization shoulder angle (LSA) and distalization shoulder angle (DSA) following reverse total shoulder arthroplasty (rTSA).

Methods

This systematic review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines. PubMed, Cochrane Central Register of Controlled Trials, and Scopus were queried on February 18, 2024. The inclusion criteria encompassed studies reporting correlations between LSA, DSA, and patient-reported outcome scores in patients undergoing rTSA with a minimum follow-up of 2 years. Study quality was evaluated using the Methodological Index for Non-Randomized Studies score. Meta-analysis was performed using a random-effects model with correlation coefficients (r) calculated via Fisher's z-transformation. Heterogeneity was assessed with the Q test and I² values.

Results

After screening, 4 studies met inclusion criteria, representing a total of 974 shoulders with a minimum follow-up of 24 months. The studies included retrospective cohort designs with Methodological Index for Non-Randomized Studies scores ranging from 14 to 15. Meta-analysis revealed no significant correlation between LSA or DSA and functional outcomes (American Shoulder and Elbow Surgeons score, Constant score) or range of motion (ROM) (active anterior elevation [AAE] and active external rotation). The overall correlation coefficient for LSA and DSA with postoperative outcomes was 0.023 (95% confidence interval [CI]: −0.056 to 0.101, P = .572). Similarly, no significant correlations were found between LSA or DSA and AAE or active external rotation, with the random effects model showing an effect size of −0.097 (95% CI: −0.231 to 0.037, P = .156) for AAE and DSA and 0.056 (95% CI: −0.052 to 0.165, P = .309) for AAE and LSA.

Conclusion

DSA and LSA may not predict the postoperative range of motion or clinical outcomes following rTSA. Future studies are warranted to develop and validate measurements of that can be used to help optimize patient outcomes.

The incidence of reverse total shoulder arthroplasty (rTSA) is growing due to expanding surgical indications in an aging population.^4,31,33 As a result, there have been significant changes and advancements in rTSA implant designs over the years.^6,25 The initial Grammont design, characterized by a medialized center of rotation, has evolved towards increased lateralization to address specific limitations associated with the initial design, including scapular notching, diminished range of motion, and suboptimal tension of the remnant rotator cuff and deltoid.^6,7,9,18 Lateralization of the glenosphere shifts the center of rotation, minimizing scapular impingement and reducing notching. rTSA designs also differ in the amount of distalization produced. Distalization increases deltoid tension by lengthening its lever arm, which affects biomechanical efficiency and range of motion.^1,2,21,25 These modifications are designed to compensate for rotator cuff deficiencies, thereby optimizing shoulder function and stability in rTSA.^6,19

The degree of lateralization and distalization of rTSA implants can be quantified on postoperative radiographs.^15,24 Among the various measurement techniques, the lateralization shoulder angle (LSA) and distalization shoulder angle (DSA) have recently been proposed as simple and reproducible methods.⁸ Prior literature has reported that increased values of LSA and DSA significantly correlated with a greater range of motion and functional outcomes after rTSA.⁸ However, conflicting evidence has raised questions about their predictive value for clinical outcome.²³ Recent studies highlight that lateralization and distalization are key variables for improved postoperative stability and motion; however, the correlation between LSA, DSA, and clinical outcomes remains inconclusive.^10,16,20,24

Despite increasing research in this area, there remains a significant gap in identifying reliable, objective measurements that can predict patient-reported outcomes (PROs) following rTSA. As such, the purpose of this systematic review and meta-analysis was to evaluate the prognostic clinical value of the LSA and DSA following rTSA. The hypothesis was neither LSA nor DSA would be reliable measurements for predicting postoperative clinical and functional outcomes after rTSA.

Methods

Study search and identification

This systematic review and meta-analysis were conducted by querying PubMed, the Cochrane Central Register of Controlled Trials, and Scopus on February 18, 2024, utilizing the search strategies detailed in Table I. This review was registered with PROSPERO (ID CRD42024513652). Inclusion criteria encompassed original research articles that examined correlations between LSA or DSA and PROs, with publications limited to the English language. Eligible studies included those reporting on patients who had undergone rTSA with implants with a humeral inclination between 135° and 155° for the treatment of cuff tear arthropathy or primary glenohumeral arthritis, with a minimum follow-up period of 2 years. Exclusion criteria comprised animal studies, biomechanical research, case reports, opinion pieces, review articles, technical notes, and studies that did not report postoperative PROs. All references were uploaded into Covidence (Veritas Health Innovation, Melbourne, Australia, www.covidence.org). Duplicate studies were removed in Covidence. All references underwent a review of titles and abstracts by two independent reviewers who are orthopedic surgeons, (S.S.) and (A.E.O.); conflicts were resolved by the senior author who is a fellowship-trained orthopedic surgeon (A.E.J.). Articles that met inclusion criteria based on title and abstract underwent full-text review to determine whether they would be included in the current review.

Table I.

Search strategies used in each database.

Database	Search strategy
PubMed (234 results)	(reverse shoulder arthroplasty) AND (lateralization OR distalization) AND (angle or radiographic)
CENTRAL (11 results)	1.(Reverse shoulder arthroplasty): ti, ab, kw 2.(lateralization): ti, ab, kw OR (distalization): ti, ab, kw 3.(angle): ti, ab, kw OR (radiographic): ti, ab, kw 4.{And #1-#3}
Scopus (132 results)	(TITLE-ABS-KEY [reverse AND shoulder AND arthroplasty] AND TITLE-ABS-KEY [lateralization] OR TITLE-ABS-KEY [distalization] AND TITLE-ABS-KEY [angle] OR TITLE-ABS-KEY [radiographic])

CENTRAL, Cochrane Central Register of Controlled Trials; ti, title; ab, abstract; ABS, abstract; kw, keyword; KEY, keyword.

Quality assessment

Two orthopedic surgeons (S.S.) and (A.O.) assessed all included articles using the Methodological Index for Non-Randomized Studies (MINORS) scoring system.³⁰ The MINORS score is a scoring tool used to assess the methodological quality of observational studies, consisting of 12 criteria, each rated from 0 to 2, with a maximum total score of 24.³⁰ Any discrepancy in scoring was resolved by the senior author (A.J.). Articles with self-reported levels of evidence and articles without levels of evidence were assigned a value based on the criteria set by Hohmann et al.²²

Data extraction

Two orthopedic surgeons (S.S.) and (A.O.) independently extracted all data using a premade data collection sheet. Discrepancies in the data extraction were resolved by senior author (A.J.). The following variables were recorded from included articles: demographics (age, sex, number of shoulders, follow-up), active ranges of motion (active external rotation [AER], active anterior elevation [AAE]), radiographic measurements (LSA, DSA), and patient-reported outcome scores; American Shoulder and Elbow Surgeons (ASES) score, Constant score (CS).

Statistical analysis

Data were analyzed using Comprehensive Meta-Analysis (V3.3.07; US National Institutes of Health, United States, www.meta-analysis.com). Heterogeneity was assessed using the Q test and I² values.^5,11,13 Q value cutoff for significance is often set at 0.10. I² values of 30%, 50%, and 75% were defined as the upper limits of moderate, substantial, and considerable heterogeneity, respectively.¹³ Forest plots and I² values were generated using RevMan (version 5.4 2020; The Cochrane Collaboration, London, United Kingdom). Both fixed and random effects were examined using Fisher's Z, and the obtained results were transformed into correlation values (r) using Fisher's inverse transformation. Publication bias was assessed using Orwin's fail-safe N, Kendall's tau, Egger's regression, and Funnel Plot.^7,14,26,29 Statistical significance was defined as P < .05.

Results

Study selection and characteristics

The search initially identified the abstracts of 377 articles. After the removal of 106 duplicate records, 271 studies were screened based on title and abstract review. Out of these, 261 studies were excluded for not meeting the predetermined inclusion criteria. Full-text screening of the remaining 10 articles was conducted and ultimately, 4 full-text articles were included in the systematic review and meta-analysis.^3,8,23,27 The Preferred Reporting Items for Systematic Reviews and Meta-analyses study flowchart was shown in Figure 1.

Figure 1.

Flowchart according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.

All studies included were retrospective cohort studies with a level of evidence of III, with MINORS scores ranging from 14^3,23,27 to 15.⁸ The studies involved a total of 974 shoulders, with a minimum of 24-month follow-up periods (ranging from 24^23,27 to 39⁸ months). The mean age of patients ranged from 69³ to 77⁸ years. The details of each study are summarized in Table II.

Table II.

Characteristics of the studies included in this systematic review and meta-analysis.

Study, and year	Study design, LOE	MINORS	No. of shoulders	Follow-up (mo)	Age (yr)
Boutsiadis et al,8 2018	Retrospective Cohort Study, III	15	46	39 ± 18	77 ± 7.5
Imiolczyk et al,23 2023	Retrospective Cohort Study, III	14	630	24 (min)	73.7 (46-90)
Mahendraraj et al,27 2020	Retrospective Cohort Study, III	14	237	24 (min)	71.2 ± 7.2
Berthold et al,3 2021	Retrospective Cohort Study, III	14	61	37.2 ± 8	69.2 ± 8.2

LOE, level of evidence; MINORS, methodological index of non-randomized studies; No., number; min, minimum.

Data reported as mean ± standard deviation (range) or (n).

The study by Boutsiadis et al⁸ retrospectively analyzed DSA and LSA in 46 shoulders with 2 different prosthesis designs (145° and 155° humeral inclinations). Imiolczyk et al²³ conducted a retrospective review of DSA and LSA across 630 shoulders using five different prosthesis designs: S1, 155° humeral inclination in 270 patients; S2, 155° humeral inclination in 44 patients; S3, 150° humeral inclination in 62 patients; S4, 145° humeral inclination in 25 patients; and S5, 135° humeral inclination in 229 patients. Mahendraraj et al²⁷ included 237 shoulders with 2 different prosthesis designs (135° and 155° humeral inclinations) in their retrospective analysis, while Berthold et al³ evaluated 61 shoulders with a 135° humeral inclination.

Meta-analysis

All studies were included in the meta-analysis.^3,8,23,27 The overall Fisher's z-transformed correlation coefficient was 0.023 (95% confidence interval [CI]: −0.056 to 0.101) using a random effects model, indicating no statistically significant correlation (P = .572). The results of the meta-analysis for AAE and AER, as well as for the CS and ASES scores, are illustrated in Figure 2 and 3, respectively.

Figure 2.

Forest plot highlighting the mean difference in AAE and AER in relation to DSA and LSA. AAE, Active anterior elevation; AER, Active external rotation; LSA, Lateralization Shoulder Angle; DSA, Distalization Shoulder Angle.

Figure 3.

Forest plot showing the mean difference in ASES score, and CS in relation to DSA and LSA. ASES, American Shoulder and Elbow Surgeon score; CS, Constant score; LSA, lateralization shoulder angle; DSA, distalization shoulder angle.

Detailed results from the fixed and random effects models between range of motion and DSA/LSA are presented in Table III. The overall effect size across the four studies was minimal, with the fixed effect model estimate of 0.001 (95% CI: −0.03 to 0.033) and the random effect model estimate of 0.002 (95% CI: −0.06 to 0.064). Both models produced nonstatistically significant results. The meta-analysis revealed substantial heterogeneity (I² = 69.99%, Q = 103.31, P < .001), confirming a high degree of variability among the included studies.

Table III.

Relationship between active anterior elevation-active external rotation motions and DSA-LSA.

Outcome	n		Estimated values				Heterogeneity test results
			ES (%95 CI)	SE	Test statistic∗	P value	Q test	P value	I2 (%)
AAE-DSA	4	Fixed	−0.084 (−0.147: −0.02)	0.032	−2.581	.010	24.79	.001	71.76
		Random	−0.097 (−0.231: 0.037)	0.068	−1.420	.156
AAE-LSA	4	Fixed	0.059 (−0.005: 0.122)	0.032	1.806	.071	16.19	.023	56.78
		Random	0.056 (−0.052: 0.165)	0.056	1.016	.309
AER-DSA	4	Fixed	−0.04 (−0.104: 0.023)	0.032	−1.243	.214	20.32	.005	65.56
		Random	−0.076 (−0.198: 0.046)	0.062	−1.222	.222
AER-LSA	4	Fixed	0.071 (0.008: 0.135)	0.032	2.196	.028	25.71	.001	72.77
		Random	0.119 (−0.018: 0.255)	0.070	1.703	.089

ES, effect size; SE, standard error; AAE, active anterior elevation; AER, active external rotation; CI, confidence interval; LSA, lateralization shoulder angle; DSA, distalization shoulder angle; N, number.

^∗Fisher's z.

Correlation between AAE, AER, and DSA-LSA

The correlation between AAE and DSA was significant in the fixed effects model (Effect Size [ES]: −0.084; 95% CI: −0.147 to −0.020; P = .010). However, the random effects model did not support this finding (ES: −0.097; 95% CI: −0.231 to 0.037; P = .156), and had substantial heterogeneity (I² = 71.76%, P = .001) (Table III).

There was a trend toward a positive effect in the fixed effects model between AAE and LSA (ES: 0.059; 95% CI: −0.005 to 0.122; P = .071), although statistical significance was not achieved. The random effects model yielded similar findings (ES: 0.056; 95% CI: −0.052 to 0.165; P = .309), with moderate heterogeneity (I² = 56.78%, P = .023) (Table III).

In the case of AER, neither the correlation with DSA (fixed effects ES: −0.04; 95% CI: −0.104 to 0.023; P = .214, random effects ES: −0.076; 95% CI: −0.198 to 0.046; P = .222) nor the correlation with LSA (fixed effects ES: 0.071; 95% CI: 0.008 to 0.135; P = .028, random effects ES: 0.119; 95% CI: −0.018 to 0.255; P = .089) reached statistical significance. Substantial heterogeneity was observed for both correlations (I² = 65.56%, P = .005) (Table III).

Correlation between patient-reported outcomes and the DSA-LSA

The relationships between patient-reported outcomes, specifically the ASES score and CS, and DSA/LSA were further explored. The pooled estimates for these outcomes demonstrated no significant correlation between DSA and LSA with PROs. The overall effect size for DSA and LSA combined was 0.023 (95% CI: −0.056 to 0.101; P = .572), with substantial heterogeneity (I² = 65.99%, P < .001) (Table IV). Due to substantial heterogeneity across studies, as indicated by the Q-test and I² values, the fixed effect model was not deemed appropriate, and the random effect model was used for a more accurate estimate.

Table IV.

Relationship between patient-reported outcomes (ASES, CS) and DSA-LSA.

Outcome	n		Estimated values∗				Heterogeneity test results
			ES (%95 CI)	SE	Test statistic∗	P value	Q	P value	I2 (%)
Total	4	Fixed	0.039 (−0.005: 0.084)	0.023	1.719	.086	47.05	<.001	65.99
		Random	0.023 (−0.056: 0.101)	0.040	0.565	.572
ASES-DSA	3	Fixed	0.010 (−0.108: 0.128)	0.060	0.171	.865	0.589	.443	0.00
		Random	0.010 (−0.108: 0.128)	0.060	0.171	.865
ASES-LSA	3	Fixed	0.064 (−0.043: 0.171)	0.055	1.178	.239	12.21	.002	83.63
		Random	−0.036 (−0.355: 0.283)	0.163	−0.222	.825
CS-DSA	2	Fixed	0.077 (0.000: 0.153)	0.039	1.971	.049	16.12	.006	68.99
		Random	0.027 (−0.133: 0.186)	0.081	0.328	.743
CS-LSA	2	Fixed	0.001 (−0.076: 0.077)	0.039	0.019	.985	15.76	.008	68.29
		Random	0.055 (−0.102: 0.213)	0.080	0.684	.492

ES, effect size; SE, standard error; CI, confidence interval; ASES, American Shoulder and Elbow Surgeon score; CS, Constant score; LSA, lateralization shoulder angle; DSA, distalization shoulder angle.

^∗Fisher's Z.

No significant association was found between the ASES and DSA (ES: 0.010; 95% CI: −0.108 to 0.128; P = .865), and the heterogeneity was low (I² = 0.00%, P = .443). The ASES-LSA correlation was also nonsignificant (ES: −0.036; 95% CI: −0.355 to 0.283; P = .825), though the heterogeneity was high (I² = 83.63%, P = .002).

Similarly, the analysis of the CS showed no significant correlation with either DSA (ES: 0.027; 95% CI: −0.133 to 0.186; P = .743) or LSA (ES: 0.055; 95% CI: −0.102 to 0.213; P = .492) (Table IV).

Risk of publication bias

The funnel plot for the standard error by Fisher's z showed a relatively symmetrical distribution, indicating no significant publication bias (Fig. 4). Additional tests for publication bias, including Egger's regression coefficient (β = −0.662, P = .263) and Kendall's tau value (−0.037, P = .418), further supported the absence of publication bias. Orwin's fail-safe N was calculated to be 6637, suggesting that an additional 6637 nonsignificant studies would be required to nullify the observed effect, bolstering the robustness of the findings.

Figure 4.

The funnel plot for the standard error by Fisher's z.

Heterogeneity analysis

Significant heterogeneity was observed across the included studies, particularly for the correlations between ASES and LSA (I² = 83.63%), CS and DSA (I² = 68.99%), and CS and LSA (I² = 68.29%) (Tables III and IV).

Discussion

While previous literature has maintained the importance of correcting axial plane deformity (retroversion) in the setting of excess retroversion or posterior glenoid bone loss, the optimization of coronal plane parameters such as lateralization and distalization of the glenosphere remain controversial. In our study, the most important findings was that neither the LSA nor the DSA measured on radiographs after rTSA demonstrated a significant correlation with postoperative functional outcomes or final range of motion at minimum 2 years postoperatively. These results were consistent with the study's hypothesis and suggest that the significance of LSA/DSA and their correlation with outcomes following rTSA may be limited.

The relationship between DSA and LSA measurements and functional outcomes remains inconclusive. While Boutsiadis et al⁸ demonstrated positive correlations between LSA and CS, Imiolczyk et al²³ identified no such correlation. Giovanetti de Sanctis et al¹⁷ reported a weak association between LSA and the CS at 1-year follow-up. Similarly, Berthold et al³ showed a significant correlation between LSA and ASES score, but Mahendraraj et al²⁷ reported that LSA weakly correlated with ASES score. On the other hand, Boutsiadis et al⁸ found a weak correlation between DSA and the CS, while no other studies have reported any significant correlation. The pooled results from our meta-analysis showed that neither LSA nor DSA revealed significant associations with any of the clinical outcomes. These findings indicated that LSA and DSA may not be a reliable postoperative predictor of functional outcomes and PROs after rTSA surgery.

The relationship between ROM and LSA/DSA was also investigated in this meta-analysis, with the findings revealing no significant correlations between AAE and AER. While Boutsiadis et al⁸ reported that optimal AER and AAE are associated with specific ranges of LSA and a DSA, the clinical significance of these angles remains unclear. Berthold et al³ reported a significant correlation between DSA, LSA, and the degree of AAE at the final follow-up (0.31 and −0.28, respectively). In contrast, two studies that were included did not identify any significant association between LSA and AER, nor between DSA and active forward flexion.^23,27 The present review found that neither LSA nor DSA revealed significant associations with any of the ROM, including anterior elevation or external rotation.

These findings call into question the reliability of LSA and DSA measurements in predicting functional outcomes. Current knowledge suggests better clinical outcomes are associated with lateralized rTSA designs^10,16; however, the results of this study demonstrated that lateralization as measured by LSA is not associated with improved ROM or functional outcomes. A possible explanation for this conflicting evidence is that LSA and DSA may not be reliable measurement methods to estimate absolute implant lateralization and distalization. In a recent study, Okutan et al²⁸ evaluated the validity of the LSA and DSA to estimate the exact distance of the lateralization and distalization. The findings identified a weak correlation between LSA and lateralization and no correlation between DSA and distalization. The authors explained this discrepancy is based on geometrical factors that are associated with measurement methods.²⁸ According to the authors, the distance between two intersecting lines, represented by LSA and DSA angles, depends not only on the angle between them but also on the lengths of the line segments, with longer segments resulting in a greater distance even if the angle remains the same. It is also possible that measurements are being done on inadequate radiographic views that could vary throughout studies and impact the measurements reported. Therefore, while both LSA and DSA measurements are simple and reproducible methods, these measurements may not accurately estimate the actual amount of lateralization and distalization. Further, these measurements cannot account for whether the lateralization was achieved from the glenoid or humerus which influences deltoid tensioning, risk of glenoid loosening, and stability.

Although the present study did not assess the intra and interobserver reliability of the LSA and DSA to estimate implant lateralization and distalization, it revealed that the LSA and DSA measurements are unreliable methods for predicting clinical outcomes. Despite this, two studies showed near-perfect interobserver reliability for DSA and LSA.^8,27 This finding may explain the conflicting conclusions reported in the previous studies regarding the impact of LSA and DSA on the outcomes following rTSA. Future studies are needed to develop and validate measurements that are reliable, reproducible, and clinically relevant.

Limitations

This current review is not without important limitations. Only four retrospective studies were included, introducing inherent risks of selection and reporting bias, with a marked discrepancy in sample sizes—most notably the study by Imiolczyk et al,²³ which included a substantially larger patient population than the others and may have disproportionately influenced the pooled results. Considerable heterogeneity in surgical indications, operative techniques, implant types, and postoperative rehabilitation protocols further complicated data synthesis and limited comparability across studies. In addition, important confounding variables such as patient activity level, general physical condition, and psychological status could not be accounted for due to insufficient data.^12,32 Furthermore, while functional outcomes did not appear to be correlated with DSA or LSA, the prevention of postoperative complications such as instability or acromial stress fracture were not assessed in this study and warrant further investigation.

Conclusion

DSA and LSA may not predict the postoperative range of motion or clinical outcomes following rTSA. Future studies are warranted to develop and validate measurements of lateralization and distalization that can be used to help optimize patient outcomes.

Disclaimers:

Funding: No funding was disclosed by the authors.

Conflicts of interest: The authors, their immediate families, and any research foundations 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.