New Measurement Method for Lateral Acromial Protrusion as an Alternative to Critical Shoulder Angle, Independent of Sagittal Glenoid Tilt

Abstract

Background

Despite the widespread use of the critical shoulder angle (CSA) for assessing lateral acromial protrusion, its reliability and consistency remain debatable. Moreover, significant CSA differences were often observed between shoulders in the same patient, even in the absence of severe bony deformity. Therefore, this study aimed to evaluate the reliability of the CSA across different imaging methods and points of view, and to propose and validate a new measurement method, the acromial protrusion ratio (APR), as a potential alternative for assessing lateral acromial protrusion by complementing the limited reliability of the CSA.

Methods

This study was conducted in 2 phases: an exploratory phase evaluating the reliability of the CSA and APR and a validation phase assessing their clinical applicability. We retrospectively reviewed 25 patients who underwent reverse total shoulder arthroplasty with a CSA difference > 5° between the shoulders in the exploratory phase and prospectively collected data from 38 shoulders in the validation phase. Measurements were independently performed by 2 assessors using different imaging methods (plain radiography and 3-dimensional [3D] reconstructed computed tomography) and points of view at the measurement (orthogonal [3DO] and neutral [3DN] scapular positions). Inter- and intraclass correlations were calculated to assess the reliability of the measurements.

Results

CSA measurements showed significant differences between the shoulders on plain radiographs (p < 0.001) and 3DN views (p < 0.001), but not on 3DO views (p = 0.547). In contrast, the APR showed no significant differences between shoulders or across different measurement methods (all p > 0.05). The APR was calculated using the following formula based on the CSA at the 3DO position: APR (%) = [CSA (°) × 0.007 + 0.053] × 100.

Conclusions

The CSA differed according to the measurement method and scapular tilt, even in an optimized true anteroposterior view. In contrast, the APR was consistent regardless of the measurement method and scapular tilt. Therefore, the APR may serve as a reliable alternative for assessing lateral acromial protrusion to complement the weak reliability of the CSA. However, its clinical relevance should be further validated in future studies.

The critical shoulder angle (CSA) is the most widely used method for evaluating lateral acromion protrusion. It is defined as the angle between the plane of the glenoid fossa and a line drawn from the inferior edge of the glenoid (IEG) to the lateral edge of the acromion (LEA) on a true anteroposterior (AP) view.¹⁾ Despite its wide use, the reliability and consistency of CSA remain debated. Several previous studies presented that the CSA provides a reproducible assessment of anatomical difference regardless of the experience level of examiners and the progression of glenohumeral osteoarthritis (GHOA).²,³⁾ However, significant discrepancies in CSA measurements have been observed between plain radiographs and 3-dimensional (3D) reconstructed images based on computed tomography (CT) or magnetic resonance imaging.⁴,⁵⁾ Furthermore, we often encountered cases with a > 5° difference in the CSA between the shoulders, even in true AP view of plain radiographs of patients without severe scapular deformities (Fig. 1).

Fig. 1. Critical shoulder angle (CSA) difference between the shoulders. Plain radiographs demonstrate a significant CSA difference between the shoulders (right, 30.9°; left, 36.4°) in a true anteroposterior view.

Therefore, this study was designed with 2 main objectives: to evaluate the reliability of CSA by comparing plain radiographs obtained in the true AP view with 3D CT images of the scapula under various sagittal tilting angles, and to propose and validate a new measurement method for assessing lateral acromial protrusion that could complement the weak reliability of the CSA. We hypothesized that the CSA measured at the orthogonal position on 3D CT would not differ between shoulders in the same patient without severe GHOA, and that the CSA could be affected by the sagittal tilt of the glenoid, even in the true AP view. In contrast, we expected that the new method would not be affected by the glenoid sagittal tilt angle.

METHODS

This study was approved by the Institutional Review Board of Seoul National University Bundang Hospital (IRB No. B-2410-931-101). The requirement for informed consent was waived by the Institutional Review Board because of the retrospective nature of the study.

Study Design

This study consisted of 2 primary components: an exploratory phase aimed at evaluating the reliability of the CSA and identifying new measurement techniques and a validation phase designed to assess the reliability of these newly developed methods.

Exploratory Study

We retrospectively reviewed 225 cases of consecutive reverse total shoulder arthroplasty (RSA) performed by the senior author (JHO) of the current study between January 2020 and October 2022. To reduce heterogeneity and measurement errors according to glenoid deformity, cases of revision surgery (n = 14), history of previous surgery (n = 6), fracture (n = 1), or infection (n = 6) on the ipsilateral shoulder, severe GHOA with grade 4 (n = 3) of the modified Samilson-Prieto classification,⁶,⁷,⁸,⁹⁾ glenoid bone defect that required a bone graft (n = 44), or poor image quality (n = 14) were excluded. Therefore, 137 patients who underwent RSA were reviewed as participants for this study.

To evaluate the reliability of CSA, we selected cases with a CSA difference > 5° between the shoulders. This criterion was based on previous studies reporting that the standard deviation of CSA in shoulders without pathology is approximately 3°–4° and that bilateral CSA differences are generally nonsignificant regardless of disease entity.¹⁰,¹¹⁾ Therefore, a difference exceeding 5° was considered to represent a meaningful deviation beyond normal measurement variability. Two shoulder fellowship-trained orthopedic surgeons (YSW and SWL) independently measured the CSA of 137 participants and categorized them for further analyses based on whether the difference between shoulders was > 5°. In instances where the decisions of the 2 assessors did not match, the first and corresponding authors (DHK and HJJ) reevaluated the measurements to confirm the difference in the CSA. Ultimately, 25 of the 137 initial participants were included in further analysis.

Three-dimensional reconstruction and evaluation of the CT scans were independently performed and repeated by the first and corresponding authors, with at least a 3-week interval between analyses. Preoperative CT images covering the entire scapula were obtained using a 64-channel multidetector CT without contrast (Brilliance 64, Philips; detector collimation: 0.612–0.625 mm; table speed: 53–313 mm/sec; slice thickness: 0.67 mm; reconstruction interval: 0.34 mm; effective tube current-time product: 156–338 mAs; tube voltage: 120–140 kVp; and matrix size: 512 × 512).¹²⁾ A commercially available software (Mimics v25.0, Materialise) was used for 3D reconstruction.¹²⁾ This process was automatically performed using a new mask function with a preset Hounsfield unit range of 148–3,071, encompassing both cancellous and cortical bones. However, the coracoid process of the 3D-reconstructed scapula obscured the superior portion of the glenoid fossa, which needed to be exposed for accurate CSA measurements. Therefore, we manually excluded the coracoid process from the 3D reconstruction (Fig. 2).

Fig. 2. Comparison of the orthogonal and neutral positions of the scapula. To define scapular orientation, the vertical (VA, solid black line) and horizontal (dotted black line) axes of the glenoid are drawn at the points where they reach their maximal length in the glenoid en face view. (A) The orthogonal scapular position is defined when the VA is perpendicular to the ground. The lateral end of the acromion (LEA; red dot) and the inferior end (IEG; white dot) and superior end of the glenoid (SEG; green dot) are marked in the true anteroposterior view. (B) The neutral scapular position is based on the X-Y-Z axes, as recorded during imaging. The LEA, IEG, and SEG are also marked in the true anteroposterior view.

To analyze whether the CSA measured in the shoulders was different, we categorized shoulders as shoulders with larger or smaller CSA based on plain radiographs. This categorization was then applied consistently for further analyses. To evaluate the difference in CSA according to glenoid tilt, we compared the CSAs obtained from the 3D-reconstructed scapula in the orthogonal (3DO) and neutral (3DN) views (Fig. 2). To determine the orthogonal position, we identified the superior, inferior, anterior, and posterior ends of the glenoid rim where the lengths of the vertical axis (VA) and horizontal axis (HA) on the en face view of the 3D-reconstructed glenoid were maximized. The center of the glenoid (CG) was defined as the intersection point of the VA and HA (Fig. 2), and the medial pole of the scapular spine (MP) was identified as the most medial point of the line horizontally bisecting the scapular spine (Fig. 3).¹²⁾ The orthogonal position was defined where the VA was perpendicular to the horizon (Fig. 2).¹²⁾ Conversely, the neutral position was based on the X-Y-Z axes as recorded in the CT scans (Fig. 2).¹²⁾ The LEA was defined as the most lateral point of the acromion visible from each point of view used for measurement (Figs. 2, 4, and 5).

Fig. 3. Medial pole of the scapular spine. Medial pole of the scapular spine (blue dot) is defined as the most medial point of the line (dotted line) horizontally bisecting the scapular spine. The lateral end of the acromion (red dot) is indicated solely for reference purposes and does not correspond to the location of the horizontal bisecting line.

Fig. 4. Sequential process of 2-dimensional transformation for radiologic measurement. (A) The lateral end of the acromion (LEA, red dot), inferior end (IEG, white dot) and superior ends of the glenoid (SEG, green dot), and medial pole (MP, blue dot) are marked on the 3-dimensionally reconstructed image. (B) The sketch function transforms the contour of the scapula and marked points into a 2-dimensional representation. (C) The critical shoulder angle is defined as the angle formed by the LEA, IEG, and SEG. (D) The acromial protrusion ratio is defined as the percentage obtained by dividing the length of the line connecting the LEA to the glenoid joint line (dotted double-headed arrow) by the length of the line connecting the MP to the glenoid joint line (solid double-headed arrow).

Fig. 5. Comparison of the marked location for the anatomical landmark between the orthogonal and neutral scapular positions. (A) The lateral end of the acromion (red), center of the glenoid (CG, yellow), and the inferior (white) and superior (green) ends of the glenoid are marked at the orthogonal (solid dot) and neutral (dashed dot) scapular positions. (B) In the glenoid en face view, the scapular contour and marked landmarks are transformed into a 2-dimensional representation using a sketch function. To quantify the location changes between the orthogonal and neutral positions, the angles transformed by the CG and each marked landmark that varied with the glenoid tilt angle are measured.

The CG, MP, LEA, IEG, and superior end of the glenoid rim (SEG) were marked on the 3DO and 3DN views.¹²⁾ Thereafter, the contour of the scapula and marked points were transformed into 2-dimensional (2D) images using the sketch function to mimic a plain radiograph (Fig. 4). CSA was defined as the angle between the glenoid joint line formed by the SEG and IEG and the line connecting the LEA and IEG. Acromial protrusion ratio (APR), the new measurement method, was defined as the percentage obtained by dividing the distance between the LEA and glenoid joint line, by the distance between the MP and glenoid joint line (Fig. 4). Although the APR was originally expressed as a decimal value, it was multiplied by 100 for ease of interpretation and represented as a percentage; APR (%).

The accuracy of the observers was evaluated using the inter- and intraclass correlations of each measurement method on the 3DO and 3DN views. The reliabilities of CSA and APR were compared for each method’s consistency between the different glenoid tilt angles. Additionally, the locations of the anatomical landmarks (LEA, SEG, and IEG) marked on the 3DO and 3DN points of view (POVs) were transformed into 2D images based on the plane parallel to the glenoid en face view to analyze the reasons for the differences in measurements according to the glenoid tilt angle (Fig. 5). To quantify how the marked location of these anatomical landmarks varied with glenoid tilt angles, we measured the angles formed by the CG and each marked landmark at the 3DO and 3DN positions (Fig. 5).

Validation Study

We collected data of 50 RSA cases performed between January 2023 and November 2023 by the senior author (JHO) of the current study. To validate the reliability and clinical applicability of the APR, a clear distinction of the MP of the scapular spine was mandatory (Fig. 6). Therefore, we attached a paper clip to the MP of the scapular spine while obtaining plain radiographs to ensure the clear identification of this anatomical landmark (Fig. 6). Of the 100 shoulders of 50 patients, we excluded those with poor image quality (n = 43), revision arthroplasty (n = 5), severe glenoid erosion (n = 13), or acromial erosion (n = 1). Ultimately, 38 shoulders were included in the validation study.

Fig. 6. Marking the scapular medial pole with a paper clip in the validation test. (A) To measure the extent of lateral acromial protrusion, a clear distinction between the medial scapular poles is mandatory. However, the scapular medial border is often difficult to distinguish from adjacent structures owing to overlapping shadows. (B) A paper clip is attached to the medial pole of the scapular spine to ensure clear identification of the anatomical landmarks.

Since the same 2 assessors (the first and corresponding authors) from the exploratory study performed the measurements in the validation study using identical methods, inter- and intraclass correlations were omitted. Instead, the average value of each measurement was used in the analyses. The CSA and APR were measured in the same manner as in the exploratory study using a plain radiograph with a true AP view and a 3D-reconstructed scapula in both orthogonal and neutral positions. Additionally, we measured the APR in non-3D-reconstructed 2D CT to evaluate the clinical applicability of APR without further processing, including 3D reconstruction, using software other than the picture archiving and communication system (PACS).

To measure the APR on 2D CT, we used the 3D cursor function in the G3 PACS (Infinitt). When the 3D cursor function is located at a specific point in 1 plane, it automatically shows its location in the other 2 planes. The MP of the scapular spine and lateral acromial end were identified in the coronal plane of the 2D CT, with adjustments made based on the sagittal and axial planes. These markings were then copied and pasted to the coronal view, where the glenoid center was located. In this view, a virtual line connecting the superior and inferior ends of the glenoid fossa was drawn to calculate the APR using 2D CT.

Statistical Analyses

All statistical analyses were conducted using R version 4.0.5 (The R Foundation for Statistical Computing) and RStudio version 1.4.1106 (RStudio Inc.). Cronbach’s alpha was used to evaluate the inter- and intraclass correlations between the assessors and agreement between the measurement methods. The Shapiro-Wilk normality test was used to assess data distribution. A paired t-test or Wilcoxon signed-rank test was conducted to identify the CSA differences between the shoulders. A repeated-measures analysis of variance or the Friedman test was used to determine differences according to the measurement methods, followed by Bonferroni post-hoc analysis. A Pearson correlation analysis was used to determine the correlation between CSA and APR in the 3DO views in the validation study, and a linear regression model was established to derive a conversion formula between the 2 parameters. All statistical analyses were performed on both sides, and the significance level was set at p < 0.05.

RESULTS

Exploratory Study

The inter- and intraclass correlations of the CSA and APR at the 3DO and 3DN positions were > 0.8 (all p < 0.001) (Table 1). Therefore, further analyses were conducted using the average values of each measurement. The CSA of the shoulders was significantly different on plain radiographs (larger CSA vs. smaller CSA: 35.2° ± 4.0° vs. 29.0° ± 4.1°, p < 0.001) and 3DN view (30.7° ± 4.1° vs. 27.6° ± 5.4°, p < 0.001). However, the CSA measured on the 3DO view was not different between the shoulders (31.9° ± 4.4° vs. 31.8° ± 6.1°, p = 0.547).

Table 1. Inter- and Intraclass Correlations of the Measurements.

Measurement			Interclass correlation	p-value
Critical shoulder angle
	Orthogonal view		0.903 (0.834–0.943)	< 0.001*
		Assessor 1	0.924 (0.836–0.966)	< 0.001*
		Assessor 2	0.863 (0.713–0.937)	< 0.001*
	Neutral view		0.958 (0.927–0.976)	< 0.001*
		Assessor 1	0.957 (0.906–0.981)	< 0.001*
		Assessor 2	0.960 (0.912–0.982)	< 0.001*
Acromial protrusion ratio
	Orthogonal view		0.907 (0.844–0.946)	< 0.001*
		Assessor 1	0.978 (0.950–0.990)	< 0.001*
		Assessor 2	0.864 (0.715–0.938)	< 0.001*
	Neutral view		0.944 (0.904–0.968)	< 0.001*
		Assessor 1	0.953 (0.898–0.979)	< 0.001*
		Assessor 2	0.964 (0.920–0.984)	< 0.001*

Values are presented as mean (95% CI). Assessors 1 (HJJ) and 2 (DHK) independently measured the critical shoulder angle and acromial protrusion ratio twice at least at 3-week intervals.

^*Statistically significant.

The CSA was measured differently according to the measurement method (p < 0.001). In post-hoc analyses, the CSA measured on 3DN view (29.4° ± 5.0°) was significantly different from that measured on the plain radiograph (32.1° ± 5.1°, p < 0.001) and 3DO view (31.8° ± 5.3°, p < 0.001). The marked locations of the anatomical landmarks varied between the 3DO and 3DN views, and these trends were pronounced at the IEG (Fig. 5). The quantified locational changes of the LEA and SEG according to the different POVs were 3.4° ± 3.8° and 4.3° ± 3.3°, respectively. The IEG showed a larger locational change of 36.9° ± 18.1°. In contrast to the CSA, the APR measured at 3DO (25.8% ± 4.2%) was not different from that measured at 3DN (25.8% ± 4.2%, p = 0.578), and the APR of both shoulders was also not different on both 3DO (larger APR vs. smaller APR: 26.1% ± 3.2% vs. 25.6% ± 5.1%, p = 0.854) and 3DN views (26.1% ± 3.4% vs. 25.6% ± 4.9%, p = 0.985).

Validation Study

Similar to the exploratory study, the CSA was measured differently according to the measurement method (p < 0.001). The post-hoc analyses revealed that the CSA measured on the 3DN view (27.8° ± 4.4°) was significantly different from that measured on the plain radiograph (31.0° ± 4.9°, p < 0.001) and 3DO view (32.7° ± 5.1°, p < 0.001). Meanwhile, the APR was not significantly different according to the measurement methods (plain radiograph: 27.1% ± 4.3%, 3DO: 26.7% ± 3.6%, 3DN: 26.8% ± 3.7%, 2D CT: 27.3% ± 3.4%; p = 0.799).

The CSA and APR on the 3DO view showed a strong correlation (r = 0.844, p < 0.001), therefore, the CSA-APR conversion formula was calculated using regression analysis (APR (%) = [CSA (°) × 0.007 + 0.053] × 100; p < 0.001) (Table 2). The APR on the 3DO view also presented good-to-excellent agreement with other measurement methods (3DO vs. plain radiography: 0.711, vs. 3DN: 0.894, vs. 2D CT: 0.816; all p < 0.001).

Table 2. Estimated Acromial Protrusion Ratio According to the Critical Shoulder Angle at the Orthogonal Position of the Scapula.

Critical shoulder angle (°)	Acromial protrusion ratio (%)
25	22.8
30	26.3
35	29.8
40	33.3

Estimated acromial protrusion ratio is calculated using the formula derived from the regression analysis: acromial protrusion ratio (%) = (critical shoulder angle [°] × 0.007 + 0.053) × 100.

DISCUSSION

The reliabilities of the CSA and APR were evaluated during the exploratory phase of the current study. The CSA of both shoulders were measured differently in plain radiographs and 3DN views. Meanwhile, in the 3DO view, the CSA did not differ between the shoulders, which is consistent with our hypothesis. The CSA varied by measurement method, possibly due to locational changes in anatomical landmark markings. However, APR was consistently measured in both shoulders in the 3DO and 3DN views. The validation phase confirmed the findings of the exploratory phase. In contrast to the CSA, the APR showed consistent measurements not only on 3D CT but also on plain radiography and 2D CT. Additionally, the APR strongly correlated with CSA in the 3DO view, allowing us to estimate the APR at specific CSA values.

Theoretically, lateral acromial protrusion could affect the biomechanics of the shoulder joint,¹³⁾ and the CSA is the most widely used parameter to measure this.¹⁾ An increased CSA, indicating a longer lateral protrusion of the acromion, could increase the superior-inferior shearing forces, which may lead to rotator cuff tears or retears after rotator cuff repairs.¹,¹³,¹⁴⁾ Conversely, a decreased CSA could enhance mediolateral compressive force on the glenohumeral joint, potentially provoking GHOA.¹,¹³⁾ Moreover, CSA might be utilized as a reference to select the medialized or lateralized implant of RSA.¹⁵⁾ Therefore, accurately measuring the extent of lateral acromial protrusion might be important to identify the causes of shoulder pathologies and predict the treatment outcomes. However, the role of the CSA in clinical practice remains controversial. Several previous studies pointed out that the CSA was not correlated with clinical outcomes or could not predict the risk of retear after rotator cuff repair.¹⁶,¹⁷⁾

We believe that the debate regarding the clinical relevance of the CSA may originate from its accuracy and reliability. Obtaining optimal plain radiographs of the true AP view may not always be easy,⁵⁾ and the CSA can be measured differently depending on the imaging method used.⁴,⁵⁾ Furthermore, as demonstrated in this study, significant differences in CSA measurements could occur between the shoulders in the same patient. Although there are some exceptions,¹⁸⁾ the human musculoskeletal system is fundamentally symmetrical. Therefore, the CSA of the shoulders may not have been measured differently. However, we found that the CSA of the shoulders was measured differently on the conventional plain radiograph and the 3DN view. Similar to previous studies,⁴,⁵⁾ CSA measurements varied according to the measurement method.

Although we could not completely explain the exact mechanism of these differences, we hypothesized that the anatomical characteristics of the glenoid may contribute to such variability. In the current study, the marked location of the IEG changed more significantly depending on the measurement position than on other landmarks. The glenoid is a pear-shaped structure, with the lower half being wider than the upper half. Therefore, changes in the marked location between the 3DO and 3DN views due to sagittal tilting may have been exaggerated in the lower half. Furthermore, the scapular posture could be changed by repeated use of the hand.¹⁹,²⁰⁾ Previous studies regarding the asymmetric posture of the scapula in the overhead athlete population showed that the scapula of the hand-dominant side was more anteriorly tilted and internally rotated.¹⁹,²⁰⁾ Considering these factors, CSA differences according to the POV in measurements might occur even in the absence of asymmetry in the anatomical bony structure. We consider that this interpretation is not conclusive but offers a plausible explanation for the CSA variability observed in the present study.

In contrast to the CSA, the APR showed no significant differences between shoulders and demonstrated minimal variation according to different POVs. The glenoid joint line connecting the SEG and IEG is an imaginary anatomical structure that represents the most lateral portion of the glenoid in the true AP view. As the glenoid has a shallow concavity with a relatively constant radius of curvature, similar to that of a dish,²¹,²²⁾ the glenoid joint line might remain largely unaffected, even with significant changes in the marked location of the IEG between the 3DO and 3DN views. Therefore, we hypothesized that this anatomical characteristic of the glenoid might be the cause of consistency in the APR.

Furthermore, the APR was constantly measured across different imaging techniques, including plain radiography and 2D and 3D CT. Considering the clinical relevance of lateral acromial protrusion, measuring the APR in the 3DO view would most accurately reflect the individual anatomical characteristics of the shoulder. However, based on the findings of this study, the clinical applicability of the APR appears to be superior to that of conventional CSA because of its consistency. APR measurement using plain radiography requires clear visualization of the scapular medial border. Although it may not always be well defined, this limitation can be easily addressed by attaching a small metallic marker, such as a paper clip, to the MP, without incurring substantial additional cost. Moreover, the APR can be measured on 2D CT images using a standard PACS viewer without the need for additional software, further enhancing its practicality and feasibility for routine clinical application.

This study had several limitations. We measured the CSA and APR in patients who underwent RSA without severe GHOA. Therefore, the clinical relevance of the APR remains unclear. However, this approach is considered more suitable for evaluating the reliability of the new measurement method itself as it reduces the heterogeneity of patient groups and lowers the risk of measurement errors. In addition, although the anatomical explanation for CSA variability is reasonable, it remains a speculative hypothesis supported by limited evidence. Furthermore, all measurements were performed by only 2 orthopedic surgeons. Although the inter- and intraclass correlations were excellent, we could not completely exclude the possibility of bias arising from a potential compromise in data neutrality. To overcome these limitations, further studies, comparing measurements across various patient groups including healthy controls and involving musculoskeletal radiologists as additional blinded assessors, are warranted. Such an approach would not only enhance clinical relevance but also improve data neutrality and strengthen the external validity of the findings. Additionally, although we cannot completely guarantee the clinical relevance of the APR, we provided the estimated APR values corresponding to the CSA at the orthogonal position of the scapula. These estimates may help interpret APR measurements in clinical settings. The inability to analyze the relationship between treatment outcomes and APR is another limitation of this study. However, because this study was based on the imaging results of patients who underwent RSA, a substantial follow-up period is required for a comprehensive clinical outcome assessment. Given that the primary objective of this study was to evaluate the reliability of CSA and propose an alternative measurement method, we considered the analysis of clinical outcomes beyond the scope of this research. Moreover, we retrospectively reviewed 25 cases in which the CSA difference between the shoulders was > 5° in an exploratory study. The sample size was small, and the criterion of > 5° CSA difference was somewhat arbitrary. We verified the consistency of our findings using a 2-step process of exploratory and validation studies. Therefore, we concluded that the bias related to the sample size was adequately controlled. Although the choice of a 5° CSA difference as an inclusion criterion was determined based on the standard deviation values reported in previous studies,¹⁰,¹¹⁾ no prior research has directly compared CSA differences between bilateral shoulders. Nevertheless, as previous studies have identified CSA values of 30° and 35° as potential risk factors for GHOA and rotator cuff tear, respectively,²³⁾ a 5° difference between shoulders may be regarded as a clinically meaningful variation beyond normal measurement variability. In the absence of more conclusive data, we consider this threshold a reasonable and practical criterion for exploratory analysis. Finally, attaching a small metallic marker for APR measurement is not a routine procedure, and a relatively high exclusion rate in the validation phase may limit the generalizability of the findings. However, although the use of a paper clip is not a standard radiographic practice, most exclusions were attributable not to marker misplacement but rather to difficulty in obtaining an accurate true AP view. This observation is consistent with previous studies highlighting the technical challenges of acquiring a precise true AP view of the shoulder.⁵⁾ Therefore, while the placement of an additional marker may hinder the clinical applicability of the APR, the difficulty in obtaining a true AP view likely has a greater impact on measurement feasibility. Moreover, the APR can be measured on CT scans using PACS software without any additional processing. Considering that CT imaging is routinely performed before shoulder arthroplasty, the clinical applicability of the APR as an alternative to the CSA appears favorable.

In conclusion, CSA could be measured differently according to the measurement method and could be affected by scapular tilt, even in an optimized true AP view. In contrast, the APR was consistent regardless of the measurement method and scapular tilt. Therefore, the APR may serve as a reliable alternative for assessing lateral acromial protrusion to complement the weak reliability of the CSA. However, its clinical relevance should be further validated in future studies.