Is Acromial Morphology Different in Patients with Eccentric Glenohumeral Osteoarthritis?

Abstract

Background:

Acromial morphology has been implicated as a potential contributor to eccentric glenohumeral osteoarthritis (GHOA), leading to the development of novel procedures including scapular spine corrective osteotomies. However, there remains a substantial gap in knowledge on the relationship between acromial morphology and eccentric GHOA. This study utilized a comprehensive three-dimensional (3D) semi-automated analysis of acromial morphology to assess its association with eccentric GHOA patterns.

Methods:

A retrospective chart review was conducted to identify symptomatic patients with GHOA undergoing evaluation for total shoulder arthroplasty. Patients were classified as having eccentric versus concentric GHOA based on the Walch criteria. A sample of healthy controls was also included from a previous investigation. Three-dimensional bone models of the scapula were reconstructed from CT scans, and scapular morphology was calculated using custom software to measure: sagittal acromial tilt, coronal acromial tilt, axial acromial tilt, lateral acromial ratio, glenoid inclination, glenoid version, and acromial coverage relative to the scapular body. These morphology variables were compared between groups using ANCOVA adjusting for sex.

Results:

98 patients with GHOA (mean age: 64.7 years; 26.5% female; Walch A: 26, Walch B: 72) and 40 healthy controls (mean age: 55.2 years; 65% female) were included. Patients with Walch B glenoids were 15.0° ± 1.4° more retroverted than patients with Walch A glenoids (p<0.01) and 14.9° ± 1.6° more retroverted than healthy controls (p<0.01). Patients with Walch B (−8.5° ± 0.9°) had more inferiorly inclined glenoids than those with Walch A (−4.1° ± 1.5°), and both groups had more inferior inclination than healthy controls (1.3° ± 1.2°; p ≤0.03). Patients with GHOA exhibited less posterior and overall acromial coverage than healthy controls (p<0.01). Patients with Walch B glenoids had a 6.9° ± 1.6° higher sagittal acromial tilt compared to healthy controls (p<0.01), and no differences were found between groups in coronal tilt or axial tilt. Patients with Walch B glenoids had less lateralized acromions than those with Walch A glenoids and healthy controls (p<0.01).

Conclusions:

There is significant variability in acromial morphology amongst patients with GHOA and compared to healthy controls. While the effect sizes are small, patients with Walch B glenoids had flatter acromion relative to healthy controls and less lateralized acromion relative to both healthy controls and Walch A glenoids. Further research is needed to understand the causative relationship between acromial morphology and the development of eccentric GHOA.

Level of evidence:

III

INTRODUCTION

The etiology of glenohumeral osteoarthritis (GHOA) is multifactorial, and the main drivers of eccentric glenoid erosion remain largely unknown^54,57. Anatomic total shoulder arthroplasty (aTSA) is the current gold-standard treatment, with increasing numbers expected over time^16,36. Patients with more deformed anatomy, including posterior humeral head subluxation and eccentric glenoid erosion (Walch type B)⁵⁵, represent a uniquely challenging treatment group. Type B glenoids are thought to progress from a “pre-osteoarthritic” state at B0 (with static posterior head subluxation) to B3 in a progression of humeral head subluxation and posterior glenoid erosion. This contrasts to type A glenoids, which are postulated to have a separate progression of concentric glenoid wear^4,27,47,57.

Historically, patients with eccentric glenoid wear patterns have experienced higher failure rates and worse postoperative function following aTSA compared to those with concentrically worn glenoids^{19,33,35,50,56}. While the initial Walch classification described only B1 and B2 glenoids⁵⁵, with the bulk of the research focusing on outcomes in those with B2 glenoids³⁵, more recent work has demonstrated minimal clinical differences between subgroups^10,46. Modified techniques have emerged for managing eccentrically worn glenoids, and improved outcomes have been seen with high-side reaming, augmented glenoid components, bone grafting, and increased use of reverse total shoulder arthroplasty (rTSA)^{12–14,18,21,22,24,26,37,42}. Despite these techniques, it remains critical to understand how anatomical variations affect glenoid wear patterns to identify at-risk patients and optimize treatment.

Research on the etiology of eccentric GHOA has primarily focused on glenoid morphology^1,9,15,29,43, coracoid morphology⁵⁸, and biomechanics^39,52,53. While acromial morphology has been associated with instability and rotator cuff tears^{28,34,38,40,48,49,59}, the contributions of acromial morphology to eccentric GHOA are not as thoroughly understood. It has been postulated that a flatter acromion and less posterior coverage of the humeral head decreases posterior bony support and biomechanical efficiency, leading to eccentric GHOA pattern.^2,3,20,23,38 This finding has led to the development of novel acromial osteotomies aimed at altering functional anatomy and native biomechanics²⁰.

To understand the complex scapular anatomy and its effect on GHOA wear patterns, robust three-dimensional (3D) morphological analysis is critical. Two-dimensional measures may be influenced by projection error and the selection of computed tomography (CT) slice on which to perform the measurement⁵¹. In addition, prior work on the association between acromial shape and eccentric glenoid GHOA has described acromial morphologic variations in reference to a glenoid plane which is pathologically deformed. The utilization of the glenoid plane is concerning as it biases the results given its confounding with eccentric OA.^2,3 To minimize this bias, previous literature did not include patients with significant glenoid deformity, which presents other potential biases.^2,3 Fully elucidating morphological relationships may influence arthroplasty options in the eccentric GHOA population. The purpose of this study is to analyze acromial morphology characteristics describing their associations with GHOA wear patterns utilizing comprehensive 3D semi-automated CT analysis referenced to reproducible non-deformed landmarks.

METHODS

A cohort of patients with symptomatic GHOA were identified from four surgeons’ practices at a single institution (n=360). From this cohort, we performed a retrospective chart review (IRB approval #202310211) to identify patients who met the following criteria: previous evaluation for primary total shoulder arthroplasty, had no prior ipsilateral shoulder surgery, and had a pre-operative CT scan that included the entire scapula. The most common reason for exclusion was CT-related (field-of-view did not include the full scapula, or technical issues while obtaining or downloading DICOMs; n=207). Patient demographics were collected from the electronic medical record, and two independent attending surgeons applied the Walch Classification⁵⁵ to each patient based on the pre-operative CT scan. Discrepancies between surgeons in Walch type were addressed by consensus, and patients with glenoid dysplasia (Walch C) were excluded. Additionally, a control cohort of asymptomatic subjects aged 50–60 years were included from a prior investigation during which CTs of the full shoulder girdle were acquired to perform model-based kinematic tracking³². Participants were included in the current study if they were found to have an intact rotator cuff (confirmed via ultrasound) and no evidence of GHOA (confirmed via CT) (n=40).

CT scans were de-identified and downloaded as DICOM files. The scapula was segmented using semi-automatic bone techniques (Mimics, Materalise; Leuven, Belgium) and 3D bone models were subsequently reconstructed. Regions of interest (ROI) for the glenoid and acromion were manually identified using MeshLab¹¹ (Figure 1). Specifically, the acromion ROI was defined as the osseous region of the acromion lateral to the spinoglenoid notch, and the glenoid ROI was defined as the entire glenoid rim and articular surface¹¹.

Figure 1:

Scapular regions of interest (ROIs) and anatomical reference coordinate system. The reference coordinate system was constructed using digitized landmarks at the root of the scapular spine, inferior angle, and spinoglenoid notch. The acromial and glenoid ROI (grey) were manually identified, and the lateral acromial region of interest (red) was automatically identified by our software as the portion of the acromial ROI lateral to a landmark digitized at the base of the coracoid. The axes of the scapular reference coordinate system are colored as follows for a right shoulder: red (+X, anterior), green (+Y, superior), and blue (+Z, lateral).

Acromial and glenoid morphology were then calculated using a custom MATLAB code³⁰. In general, this process involved fitting 3D axes to the glenoid and acromial ROIs (which represent the primary planes of anatomy) and relating these axes to a reference anatomical coordinate system (details below). The selection of this reference coordinate system is an important methodological decision as it represents the reference from which many morphology variables are calculated (e.g., glenoid version, acromial tilt). Unfortunately, there is high variability in the literature regarding which reference coordinate system is used, and thus how morphology variables are calculated. For example, the Friedman line, a two-dimensional parameter which extends from the center of the glenoid to the medial border of the scapula, is frequently used as a reference to describe version^5–7,17,47. Further, the glenoid plane has been used as the reference for acromial morphology². Given the glenoid is known to be deformed in patients with GHOA (especially Walch B glenoids)^9,43,57, which will confound subsequent morphology calculations, we aimed to employ a reference coordinate system that was not expected to be impacted by degenerative changes. Thus, we utilized a reference coordinate system constructed based on the following landmarks, which were digitized on the 3D scapular models by a single user: 1) root of the scapular spine (i.e., the intersection of the scapular spine and medial border of the scapula), inferior angle (i.e., the most inferior point on the scapula), and the spinoglenoid notch (Figure 1; details of landmark digitization are provided in the Appendix). Versions of this reference coordinate system have been used previously^25,32,45.

Once this reference coordinate system was defined, glenoid inclination and version were calculated by relating the axes of a glenoid-based coordinate system to those of the reference coordinate system. The glenoid-based coordinate system was automatically defined in our software using the glenoid ROI (details of calculation are provided in the Appendix)³¹. Given the line of erosion between the neoglenoid and paleoglenoid is not clearly visible in all cases (i.e., especially in B1 and B3 cases), the entire glenoid surface was included in the definition of the glenoid axes. Glenoid inclination was then calculated as the angle between the two superior/inferior (S/I) axes (positive: superiorly-oriented glenoid; negative: inferiorly-oriented glenoid) (Figure 2A), and glenoid version was calculated as the angle between the two anterior/posterior (A/P) axes (positive: anteriorly-oriented glenoid; negative: posteriorly-oriented glenoid) (Figure 2B)³².

Figure 2:

Glenoid and acromial morphology calculations. A) Glenoid inclination was calculated as the angle between the superior axis of the glenoid coordinate system relative to the superior axis of the scapular reference coordinate system. B) Glenoid version was calculated as the angle between the anterior axis of the glenoid coordinate system relative to the anterior axis of the scapular reference coordinate system. C) Acromial sagittal tilt was calculated as the angle between acromial anterior/posterior axis and the scapular superior/inferior axis. D) Coronal tilt was calculated as the angle between acromial medial/lateral axis and the scapular superior/inferior axis. E) Axial tilt angle was calculated as the angle between acromial anterior/posterior axis and the scapular anterior/posterior axis. F) Posterior acromial height was calculated as the perpendicular distance (d) between the posterior acromion and superior/inferior axis of the scapular reference coordinate system. G) Acromial overall, anterior, and posterior coverage was calculated based on vectors between the anterior acromion, glenoid center, and posterior acromion. H) Critical shoulder angle was calculated by reorienting the scapula to align with the glenoid-based coordinate system and calculating the angle between the superior/inferior axis of the glenoid coordinate system and a vector between the inferior glenoid and the lateral-most point on the acromion. I) Lateral acromial ratio was calculated as the ratio of the lateral acromial width (base of coracoid to lateral acromion) to the scapular body width (root of the spine to the base of the coracoid). The axes of the acromial scapular reference coordinate systems are colored as follows for a right shoulder: red (+X, anterior), green (+Y, superior), and blue (+Z, lateral).

Acromial morphology (sagittal, coronal, and axial tilt) was calculated on the 3D models using procedures similar to those described by Beeler et al². First, the 3D scapular model was reoriented such that it was aligned with the scapular reference coordinate system and a single user digitized a landmark at the base of the coracoid (i.e., at the intersection between the most superolateral point of the coracoid base and the superior glenoid). Then, the lateral acromion ROI was automatically identified as any surface of the acromion ROI lateral to the landmark (Figure 1). This approach helped to standardize the definition of the lateral acromion, which was necessary because all acromial tilt measures depend upon it. Then, the axes of the lateral acromion were defined by performing principal components analysis of the surface points defining the lateral acromial ROI. This process is analogous to determining the mean vector of the lateral acromion in each plane (sagittal, coronal, and axial). Finally, acromial sagittal, coronal, and axial tilt were calculated as follows: sagittal tilt was calculated as the angle between the A/P axis of the lateral acromial ROI and the scapular S/I axis in the sagittal plane (Figure 2C), coronal tilt was calculated as the angle between medial/lateral (M/L) axis of the lateral acromial ROI and the scapular S/I axis in the coronal plane (Figure 2D), and axial tilt was calculated as the angle between A/P axis of the lateral acromial ROI and the scapular A/P axis in the transverse plane (Figure 2E). Given the current study employed a novel reference coordinate system which makes comparison between methods challenging, we performed a sensitivity analysis to assess the impact of different scapular reference coordinate systems on the acromial tilt variables. Details and results of this analysis are provided in the Appendix.

Acromial coverage variables were defined by assessing the degree to which the acromion overhangs the glenoid in the sagittal plane using methods similar to those described by Beeler et al² (Figure 2F). Specifically, the anterior-, posterior-, and inferior-most points on the acromial ROI were identified after the 3D scapular model was reoriented such that it was aligned with the scapular reference coordinate system. Overall coverage was then calculated as the angle between the anterior acromion/glenoid center vector and the posterior acromion/glenoid center vector; anterior coverage was calculated as the angle between the anterior acromion/glenoid center vector and the S/I scapular reference axis (a positive value indicates the acromion extended anterior to the scapular S/I axis and a negative value indicates it does not); and posterior coverage was calculated as the angle between the posterior acromion/glenoid center vector and the S/I scapular reference axis. Finally, posterior acromial height was calculated as the perpendicular distance between the inferior most point on the acromion and the scapular plane³⁸.

The critical shoulder angle (CSA) was calculated by reorienting the scapular model to align with the glenoid-based coordinate system (simulating the “true” A/P view⁵¹), then calculating the 3D angle between the glenoid’s S/I axis and a vector between the inferior glenoid and the lateral-most point on the acromion in this view (Figure 2G).³² We also sought to determine acromial lateralization as an independent morphological parameter, notably overcoming the limitations of the CSA by not relying on variable glenoid morphology. We therefore calculated the lateral acromial ratio as the ratio of the lateral acromial width (base of coracoid to lateral acromion) to the scapular body width (root of the spine to the base of the coracoid) in the coronal scapular plane with a larger value suggesting a more lateral acromion (Figure 2H).

To investigate the extent to which manually digitizing anatomical landmarks impacted the calculation of morphological parameters, we performed a reliability analysis on a random subset of our GHOA sample (n=20). Two orthopaedic surgeons independently digitized anatomical landmarks on two separate days approximately 1 week apart. The variability due to repeat landmark digitization was calculated as the average 3D (Euclidean) distance between each individual trial and the centroid across all trials. Further, inter-rater and intra-rater reliability of the morphology variables was assessed using intraclass correlation coefficients (ICC1,1) and standard error of the measurement (SEM). Variability in landmark position due to manual digitization averaged (±SD) 1.8±0.9 mm across all landmarks used to construct the scapular anatomical coordinate system (root of scapular spine: 2.1±1.1 mm; spinoglenonid notch: 1.8±0.7 mm; inferior angle: 1.6±0.8 mm). Inter-rater and intra-rater reliability were good to excellent across all morphology variables (inter-rater: ICC_1,1 0.84–0.99, SEM ≤ 3.2°; intra-rater: ICC_1,1 0.77–0.99, SEM ≤ 3.7°).

Statistical Analysis

Demographics were compared between groups (Walch A glenoids, Walch B glenoids, healthy controls) using ANOVA. Scapular morphology was compared between groups using ANCOVA to adjust for the potential confounding effect of sex given the group differences found (Table 1). To conduct these tests, a three-factor model was performed (independent variable: group, covariate: sex, interaction: group-by-sex) for each morphology variable; if the interaction was not significant, the model was reduced to a parsimonious two-factor model (independent variable: group, covariate: sex). When significant group differences were detected, pairwise comparisons were assessed using Tukey-Kramer adjustments. All data are reported as adjusted mean ± standard error unless otherwise noted. Statistical analyses were performed using SAS OnDemand for Academics (SAS Institute; Cary, NC) with statistical significance set at p<0.05.

Table 1:

Patient demographics presented as mean ± standard deviation.

Variable	Walch A (n=26)	Walch B (n=72)	Controls (n=40)	p-value
Age (years)	64.4 ± 6.1	64.8 ± 9.7	55.2 ± 3.4	<0.01
Sex (% female)	26.9%	26.4%	65.0%	<0.01
BMI (kg/m2)	31.1 ± 6.1	30.8 ± 6.9	26.6 ± 2.9	<0.01
Walch classification	A1: 23.1% A2: 76.9%	B1: 6.9% B2: 72.2% B3: 20.8%	N/A	N/A

RESULTS

The final cohort included 98 patients with symptomatic GHOA and 40 healthy control subjects (Table 1). Patients with GHOA were significantly older than controls (mean ± SD: 64.7 ± 8.9 years vs. 55.2 ± 3.4 years; p<0.01). There were also fewer females in the GHOA group than in the control group (26.5% vs. 65.0%; p<0.01). Among the patients with GHOA, 75% of the cohort had eccentric GHOA (Walch B1: 5/72, 6.9%; B2: 52/72, 72.2%; B3: 15/72, 20.8%) and the remainder had concentric GHOA (Walch A1: 6/26, 23.1%; Walch A2: 20/26, 76.9%). There were no significant differences in age, sex, or BMI between those classified as Walch A (concentric) or Walch B (eccentric).

Patients with Walch B glenoids were on average 15.0° ± 1.5° more retroverted compared to healthy controls (p<0.01) and 14.9° ± 1.6° more retroverted compared to patients with Walch A glenoids (p<0.01) (Table 2). Differences in glenoid version between patients with Walch A glenoids and healthy controls was not statistically significant (p=1.00). Glenoid inclination was superiorly oriented in healthy controls (1.3° ± 1.2°) and was inferiorly oriented in patients with Walch A (−4.2° ± 1.5°) and Walch B (−8.4° ± 0.9°) glenoids, with significant pairwise differences between all groups (p≤0.03).

Table 2:

Summary of the comparison of scapular morphology parameters between patients with Walch A (concentric) glenoids, patients with Walch B (eccentric) glenoids, and healthy controls. Data are presented as adjusted means ± standard error. Negative values for glenoid version and inclination represent a posteriorly- and inferiorly- oriented glenoid, respectively. Groups were compared using ANCOVA to adjust for group differences in sex. Pairwise follow-ups were performed, and letters are used to indicate significant differences between groups within a variable (i.e., groups within a row that share the same letter are not statistically different).

Variable	Walch A (n=26)	Walch B (n=72)	Controls (n=40)	p-value
Glenoid version	−3.2° ± 1.4° (a)	−18.1° ± 0.9° (b)	−3.1° ± 1.1° (a)	<0.01
Glenoid inclination	−4.2° ± 1.5° (a)	−8.4° ± 0.9° (b)	1.3° ± 1.2° (c)	<0.01
Overall coverage	62.9° ± 1.3° (a)	63.1° ± 0.8° (a)	70.1° ± 1.0° (b)	<0.01
Anterior coverage	7.1° ± 1.6°	7.9° ± 1.0°	7.4° ± 1.3°	0.90
Posterior coverage	55.7° ± 1.4° (a)	55.2° ± 0.9° (a)	62.7° ± 1.1° (b)	<0.01
Acromial sagittal tilt	68.2° ± 1.6° (ab)	71.3° ± 1.0° (a)	64.4° ± 1.2° (b)	<0.01
Acromial coronal tilt	75.8° ± 1.9°	79.5° ± 1.2°	78.5° ± 1.5°	0.25
Acromial axial tilt	26.2° ± 1.8°	25.5° ± 1.1°	22.5° ± 1.4°	0.19
Posterior acromial height	13.3 ± 1.1 mm (ab)	15.7 ± 0.7 mm (a)	10.7 ± 0.9 mm (b)	<0.01
Critical shoulder angle	33.7° ± 1.1°	34.1° ± 0.7°	35.0° ± 0.9°	0.65
Lateral acromial ratio	31.5% ± 0.9% (a)	27.8% ± 0.6% (b)	31.1% ± 0.7% (a)	<0.01

Overall acromial coverage differed between healthy controls and patients with GHOA, with controls exhibiting 7.2° ± 1.7° more coverage than patients with Walch A glenoids (p<0.01) and 7.0° ± 1.4° more than patients with Walch B glenoids (p<0.01) (Table 2). There were no statistically significant differences between groups in anterior coverage (p=0.90), but healthy controls exhibited 7.0° ± 1.8° more posterior coverage than patients with Walch A glenoids (p<0.01) and 7.5° ± 1.4° more posterior coverage than patients with Walch B glenoids (p<0.01).

Three-dimensional acromial tilt analysis demonstrated significant differences in the sagittal plane, with patients with Walch B glenoids exhibiting 6.9° ± 1.6° higher sagittal tilt (i.e., flatter acromial roof) than healthy controls (p<0.01) (Table 2). Differences in sagittal tilt between patients with Walch A and B glenoids were not statistically significant (p=0.18). No statistically significant differences were found between any groups in coronal tilt (p=0.25) or axial tilt (p=0.19). Finally, the posterior acromial height in patients with Walch B glenoids was 5.1 ± 0.1 mm greater than that of healthy controls (p<0.01) but not significantly different compared to patients with Walch A glenoids (2.5 ± 1.2 mm, p=0.11).

There were no statistically significant differences in the CSA between groups (p=0.65) (Table 2). However, when the lateral extension of the acromion was calculated as a lateral acromial ratio, patients with Walch B glenoids were found to have a 3.6% ± 1.0% lower lateral acromial ratio (more medialized acromion) compared to patients with Walch A glenoids (p<0.01) and a 3.3% ± 1.0% lower lateral acromial ratio than healthy controls (p<0.01).

DISCUSSION

The relationship between morphology and the development of eccentric GHOA remains unclear, contributing to a gap in knowledge in both pathogenesis and optimal treatment. While newer techniques have shown promising results for managing patients with B2 glenoids^{12–14,18,19,21,24,26,33,56} – the most rigorously studied group of the continuum of eccentric disease – identifying anatomic factors associated with this condition has the potential to shift treatment patterns and further improve patient outcomes. We identified differences in acromial morphology between eccentric and concentric GHOA of unknown clinical significance.

We employed a 3D method with excellent interobserver reliability to comprehensively assess scapular morphology. These methods are similar to those previously utilized by Beeler et al.^2,3 but differs in an important way which may explain the discrepancy in findings between studies. Specifically, the methods employed by Beeler et al. utilize the glenoid as the reference from which to describe acromial morphology. As they acknowledge, the glenoid is often highly deformed within the Walch B GHOA population^9,43,57, and therefore they excluded patients with significant glenoid deformity. Despite this exclusion, descriptions of acromial morphology are likely confounded by glenoid degeneration associated with the study cohort in their analyses. Instead, we utilized a scapular-based reference coordinate system including the entire scapular plane in a representative cohort of glenoid pathology that reflects the scapular anatomic landmarks likely to be unaffected by this pathology. This methodological difference may explain the discrepancy in findings between the current study and Beeler et al. who found more and larger differences in acromial morphology between groups. The comparison of results between a glenoid-referenced calculation and scapular-body reference coordinate system in this analysis demonstrates the sensitivity of acromial morphology to measurement methods and the importance of a non-pathologic reference. This analysis utilizes a healthy shoulder control group in conjunction with a high volume (n=26 [centered] and n=72 [eccentric]) of GHOA that allows for the assessment of acromial morphologic differences in GHOA and specifically, eccentric GHOA.

Our patient cohort generally matched previous studies of primary GHOA patients in demographics. As measured in this study, our average CSA were largely consistent with previous studies; however, unlike prior research, we found no differences between the groups and did not find CSA less than 30° to be associated with GHOA^2,38,52. When lateral acromial ratio was used as a glenoid-independent calculation of lateral overhang, we did find that the ratio was lower for Walch B glenoids, suggesting the acromion extends less laterally in this group. While it is not entirely clear why Walch B glenoids would have less overhang, we postulate that medialization of the acromion alters the deltoid compressive force vector^52,53.

It has been proposed that a flatter acromial roof and undercoverage of the humeral head posteriorly is a potential driver of eccentric GHOA^2,3. Beeler et al. suggested that this anatomy led to humeral head decentralization and a less favorable deltoid vector^2,3. While we found no differences in most orientation parameters between groups, the one acromial orientation we did find to be statistically significant – acromial sagittal tilt in eccentric OA relative to controls – was in agreement with Beeler et al.’s findings. Those authors reported that patients with static posterior subluxation in early osteoarthritis had flatter acromions^2,3. While we found significant differences between controls and osteoarthritic patients in terms of overall and posterior coverage, we did not find differences between Walch A and B, deviating from Beeler et al.’s report of eccentric glenoid posterior undercoverage. Similarly, posterior acromial height (likely a proxy for sagittal tilt) was significantly different between controls and GHOA, but not between eccentric and centered GHOA.

Our results suggest that patients with Walch B glenoids have a flatter acromion (i.e., higher sagittal tilt) than healthy controls and less lateral acromion than both healthy controls and Walch A glenoids. While statistically significant, these differences are small in magnitude. These small effect sizes raise doubts about the clinical significance of the association between acromial morphology and eccentric GHOA. Despite significant efforts, the community’s understanding of the etiology of eccentric GHOA remains in its infancy. To resolve this, continued efforts establishing the complex interplay between kinematics, morphology, activity, intrinsic pathology, and other factors that may contribute to the development of eccentric GHOA are required. This knowledge has the potential to impact decision-making across the spectrum of symptomatic GHOA. Identifying contributing factors will allow for potential interventions to prevent progression of disease or allow for more predictable definitive treatment options (e.g. a well-balanced total shoulder arthroplasty). In the interim, our analysis did not find a substantial association between acromial morphology and the pattern of GHOA. However, further studies should be done to fully elucidate the clinical outcomes and biomechanical effects of distorting a patient’s native anatomy to “normal”.

This analysis is not without limitations. With 72.2% of our Walch B glenoids being B2, the small sample size of B1 and B3 glenoids prevented meaningful subgroup analysis. Additionally, given the study’s retrospective nature, our GHOA patients were older, heavier, and had a lower proportion of females compared to control subjects. These differences resulted from the use of a sample of convenience from a previous investigation to leverage the feasibility of obtaining CT scans on otherwise healthy individuals. Although we employed ANCOVA to control for differences between groups in sex, future studies should seek to employ a more balanced sample to avoid the potentially confounding effect of sex on morphology. Importantly, it is unknown whether any of our control patients could represent a “pre-osteoarthritic” state and eventually develop symptomatic GHOA; this potential early time point in GHOA progression could mitigate differences seen between control patients and our symptomatic GHOA cohort. Additionally, it is possible that acromial morphologic differences between GHOA cohorts and normal cases are age-related. However, literature to date has not identified any changes in acromial morphology as patients age.^8,41,44 Finally, the use of a scapular reference coordinate system required the entire scapula to be within the CT field of view. Consequently, many patient scans were excluded when the field of view did not extend to the inferior angle or root of the scapular spine. To facilitate clinical translation of these methods, future work may explore modified reference coordinate systems that reflect the scapular body without requiring the visualization of distant landmarks.

CONCLUSION

This is a rigorous assessment of the association between acromial morphology and GHOA using semi-automated 3D morphological analysis in healthy control subjects, those with concentric GHOA, and those with eccentric GHOA. These methods can be applied in a variety of patient populations to further understand the complex interplay of anatomic and clinical outcomes. Given the limited association between acromial morphology and eccentric GHOA in this thorough three-dimensional analysis, the clinical benefits of surgically altering acromial morphology to treat posterior humeral subluxation or progression of eccentric GHOA requires further attention. Future work is required to more thoroughly assess scapular morphology’s—specifically acromial morphology—and other biomechanical factors’ contribution to the development of eccentric GHOA.