Preoperative Planning for Reverse Shoulder Arthroplasty: Does the Clinical Range of Motion Match the Planned 3D Humeral Displacement?

Introduction: The functional outcome after reverse shoulder arthroplasty (RSA) is closely linked to how much the humerus shifts because of the implants. While two-dimensional (2D) angle measurements have been used to capture this shift, it can be measured in three dimensions (3D) as the arm change position (ACP). In a previous study, the ACP was measured using 3D preoperative planning software with the passive virtual shoulder range of motion obtained after RSA. The main objective of this study was to evaluate the relationship between the ACP and the actual active shoulder range of motion measured after RSA. The hypothesis was that the ACP and the active clinical range of motion are related such that the ACP is a reliable parameter to guide the preoperative planning of an RSA. The secondary objective was to assess the relationship between 2D and 3D humeral displacement measurements. Materials and methods: This prospective observational study enrolled 12 patients who underwent RSA and had a minimum follow-up of 2 years. The active range of motion in shoulder flexion, abduction, and internal and external rotation was measured. At the same time, ACP measurements were taken from a reconstructed postoperative CT scan, in addition to the radiographic measurements of humeral lateralization and distalization angles on AP views in neutral rotation. Results: The mean humeral distalization induced by RSA was 33.3 mm (±3.8 mm). A non-statistically significant increase in shoulder flexion was observed for humeral distalization beyond 38 mm (R2 = 0.29, p = 0.07). This “threshold” effect of humeral distalization was also observed for the gains in abduction, as well as internal and external rotations, which seemed better with less than 38 mm or even 35 mm distalization. No statistical correlation was found between the 3D ACP measurements and 2D angle measurements. Conclusion: Excessive humeral distalization seems to be detrimental to joint mobility, especially shoulder flexion. Humeral lateralization and humeral anteriorization measured using the ACP seem to promote better shoulder range of motion, with no threshold effect. These findings could be evidence of tension in the soft tissues around the shoulder joint, which should be taken into consideration during preoperative planning.


Introduction
Reverse total shoulder arthroplasty (RSA) has shown excellent functional outcomes and therefore has become widely accepted as the treatment of choice for cuff tear arthropa-

Patients
The research protocol approval was obtained from our institutional review board (IRB: 13B-T-SHOULDER-RM). This prospective observational study enrolled 12 patients (9 women and 3 men), with a mean age of 73 ± 4.3 years (range, 63 to 81 years), who had undergone RSA surgery between 9 January 2015 and 31 August 2016 ( Figure 1). RSA indications were primary osteoarthritis and cuff tear arthropathy (CTA), with six patients in each group. Their average body mass index was 28.5 ± 3 kg/m 2 (19.5-34.4). RSA was performed on the dominant side in nine cases. The shoulder undergoing RSA had never been operated on. The preoperative demographic, clinical, and imaging data of the study population are presented in Table 1.
humeral displacements measured in 3D using the ACP and those in 2D using the LSA DSA.

Patients
The research protocol approval was obtained from our institutional review b (IRB: 13B-T-SHOULDER-RM). This prospective observational study enrolled 12 pati (9 women and 3 men), with a mean age of 73 ± 4.3 years (range, 63 to 81 years), who undergone RSA surgery between 9 January 2015 and 31 August 2016 ( Figure 1). indications were primary osteoarthritis and cuff tear arthropathy (CTA), with six pati in each group. Their average body mass index was 28.5 ± 3 kg/m 2 (19.5-34.4). RSA performed on the dominant side in nine cases. The shoulder undergoing RSA had n been operated on. The preoperative demographic, clinical, and imaging data of the st population are presented in Table 1.

RSA Implants
A reversed shoulder prosthesis (Wright Medical France, Monbonnot Saint Martin, France) was used on all shoulders. The humeral stem was an AEQUALIS ASCEND™ FLEX system (Tornier, Bloomington, MN, USA), which has a 132.5 • fixed inclination Onlay stem mounted with a 1.5 mm to 3.5 mm lateralized humeral plateau with variable indexing and a standard polyethylene insert having a fixed tilt angle of 12.5 • . The final humeral configuration had a 145 • inclination.

Software
The BluePrint ® 3D Planning software (Tornier SAS France, Monbonnot Saint Martin, France) was used to select the type of implants to use in a given patient and to position them in a virtual shoulder joint. Once the implant configuration is finalized, the software simulates the passive RoM and measures humeral displacement in the frontal, sagittal, and axial planes. This is the ACP parameter (expressed in millimeters), which comprises 3 values that correspond to the superior-inferior, mediolateral, and anteroposterior humeral displacements after the virtual implantation of the RSA ( Figure 2). These three values are positive when the displacement is superior, lateral, and anterior. They are negative when the displacement is inferior, medial, and posterior ( Figure 3).

RSA Implants
A reversed shoulder prosthesis (Wright Medical France, Monbonnot Saint Martin, France) was used on all shoulders. The humeral stem was an AEQUALIS ASCEND™ FLEX system (Tornier, Bloomington, MN, USA), which has a 132.5° fixed inclination Onlay stem mounted with a 1.5 mm to 3.5 mm lateralized humeral plateau with variable indexing and a standard polyethylene insert having a fixed tilt angle of 12.5°. The final humeral configuration had a 145° inclination.

Software
The BluePrint ® 3D Planning software (Tornier SAS France, Monbonnot Saint Martin, France) was used to select the type of implants to use in a given patient and to position them in a virtual shoulder joint. Once the implant configuration is finalized, the software simulates the passive RoM and measures humeral displacement in the frontal, sagittal, and axial planes. This is the ACP parameter (expressed in millimeters), which comprises 3 values that correspond to the superior-inferior, mediolateral, and anteroposterior humeral displacements after the virtual implantation of the RSA ( Figure 2). These three values are positive when the displacement is superior, lateral, and anterior. They are negative when the displacement is inferior, medial, and posterior ( Figure 3).   Postoperatively, another software package (PTC Creo ® version 6.0, Parametric Technology Corporation, Needham, MA, USA) was used to capture a postoperative CT scan of the patient's shoulder with the implants in place. These images were uploaded into the Blueprint software to measure the ACP postoperatively, as this corresponds to the actual implant positioning.

Methods
The preoperative imaging assessment included standard AP X-rays in three different rotations plus an axillary view and a CT scan of the shoulder for 3D planning purposes. All the patients were operated on using the deltopectoral approach for the implantation of the prosthesis, in a beach chair position and under general anesthesia combined with an interscalene block. The procedure was performed in a standard manner with no particular technical point during the procedure. The retroversion of the humeral implant was adjusted relative to the forearm axis and ranged from 10° to 30°, depending on the patient. The subscapularis was repaired when it was still present. Biceps tenodesis to the pectoralis major was carried out at the end of the procedure. The patients were discharged 48 h after surgery. Postoperative care included shoulder immobilization with an abduction pillow for 3 to 6 weeks. Physiotherapy was started immediately or at 3 weeks if BIO-RSA had been performed.
The 12 patients were reviewed at a minimum follow-up of 2 years postoperatively. Their active shoulder joint mobility was measured in degrees for flexion (F), abduction (ABD), external rotation (ER1/ER2) with the elbow at the side/with the arm abducted at 90°, and internal rotation (IR1) with the elbow at the side, as it had been during the preoperative consultation. The active internal rotation range of motion measurement was defined as the highest midline vertebral segment of the back that can be reached. This measurement was converted into a 10-point scale according to the Constant-Murley Shoulder Outcome Score guidelines [32].
A complete radiographic assessment identical to that carried out preoperatively (AP view in three rotations and axillary view) was performed; fluoroscopy was used beforehand to ensure the images were reproducible between the patients. The LSA and Postoperatively, another software package (PTC Creo ® version 6.0, Parametric Technology Corporation, Needham, MA, USA) was used to capture a postoperative CT scan of the patient's shoulder with the implants in place. These images were uploaded into the Blueprint software to measure the ACP postoperatively, as this corresponds to the actual implant positioning.

Methods
The preoperative imaging assessment included standard AP X-rays in three different rotations plus an axillary view and a CT scan of the shoulder for 3D planning purposes. All the patients were operated on using the deltopectoral approach for the implantation of the prosthesis, in a beach chair position and under general anesthesia combined with an interscalene block. The procedure was performed in a standard manner with no particular technical point during the procedure. The retroversion of the humeral implant was adjusted relative to the forearm axis and ranged from 10 • to 30 • , depending on the patient. The subscapularis was repaired when it was still present. Biceps tenodesis to the pectoralis major was carried out at the end of the procedure. The patients were discharged 48 h after surgery. Postoperative care included shoulder immobilization with an abduction pillow for 3 to 6 weeks. Physiotherapy was started immediately or at 3 weeks if BIO-RSA had been performed.
The 12 patients were reviewed at a minimum follow-up of 2 years postoperatively. Their active shoulder joint mobility was measured in degrees for flexion (F), abduction (ABD), external rotation (ER1/ER2) with the elbow at the side/with the arm abducted at 90 • , and internal rotation (IR1) with the elbow at the side, as it had been during the preoperative consultation. The active internal rotation range of motion measurement was defined as the highest midline vertebral segment of the back that can be reached. This measurement was converted into a 10-point scale according to the Constant-Murley Shoulder Outcome Score guidelines [32].
A complete radiographic assessment identical to that carried out preoperatively (AP view in three rotations and axillary view) was performed; fluoroscopy was used beforehand to ensure the images were reproducible between the patients. The LSA and DSA in degrees ( • ) were measured on AP views in neutral rotation, as described by Boutsiadis et al. [12] ( Figure 4). DSA in degrees (°) were measured on AP views in neutral rotation, as described by Boutsiadis et al. [12] (Figure 4). Finally, at the same minimum follow-up of 2 years, a CT scan of the operated shoulder was carried out to measure the ACP after RSA using the BluePrint ® planning software. However, the presence of the implants meant that these CT scans were not directly supported by the BluePrint ® software. Several image preparation and processing steps had to be completed before the postoperative ACP could be measured in each patient ( Figure 5). Finally, at the same minimum follow-up of 2 years, a CT scan of the operated shoulder was carried out to measure the ACP after RSA using the BluePrint ® planning software. However, the presence of the implants meant that these CT scans were not directly supported by the BluePrint ® software. Several image preparation and processing steps had to be completed before the postoperative ACP could be measured in each patient ( Figure 5).
These processing steps were as follows: 1.
The extraction of the 3D geometry of the humerus and scapula from the preoperative CT scan; 2.
The manual registration of the preoperative 3D geometry of the humerus and scapula, with RSA implants from the postoperative CT scan, using the PTC Creo ® software; 3.
The creation of planning files integrating the readjusted bone and implant geometries in the BluePrint ® software to measure the postoperative ACP corresponding to the RSA implant configuration specific to each patient.
The data were collated in an Excel ® spreadsheet and analyzed using the JMP ® 11.0.0 software (SAS Institute Inc.©, Cary, NC, USA). A Shapiro-Wilk test was used to evaluate the normal distribution of the continuous quantitative variables. A Student's t-test was performed to compare the means between indication subgroups; linear regression was carried out for the correlations. The significance level was set at p < 0.05. These processing steps were as follows: 1.
The extraction of the 3D geometry of the humerus and scapula from the preoperative CT scan; 2.
The manual registration of the preoperative 3D geometry of the humerus and scapula, with RSA implants from the postoperative CT scan, using the PTC Creo ® software; 3. The creation of planning files integrating the readjusted bone and implant geometries in the BluePrint ® software to measure the postoperative ACP corresponding to the RSA implant configuration specific to each patient. The data were collated in an Excel ® spreadsheet and analyzed using the JMP ® 11.0.0 software (SAS Institute Inc.©, Cary, NC, USA). A Shapiro-Wilk test was used to evaluate the normal distribution of the continuous quantitative variables. A Student's t-test was

Descriptive Data
RSA improved the active shoulder RoM in all patients. These improvements were similar for the two groups (Table 2). Internal rotation was unchanged in the CTA group, whereas it was improved in the primary osteoarthritis group relative to the preoperative measurement. The other RoM (∆F, ∆ABD, ∆ER1, and ∆ER2) had also further increased compared with the preoperative RoM for the primary osteoarthritis group (Table 3).    The mean humeral distalization induced by RSA, assessed using the ACP, was 33.3 mm (±3.8 mm). The mean lateralization was 4.3 mm (±3.7 mm). The mean anterior displacement of the humerus ("anteriorization") was 6 mm (±6 mm). There was no significant difference between the indications for these three parameters ( Table 2). The mean LSA and DSA values were 80.9 • (±11.1 • ) and 48.5 • (±9.7 • ), with no difference between the two indications ( Table 2).

Analytical Data
RoM and Measurement of 3D Humeral Displacement Using the ACP (Figure 6). There were no statistically significant findings; however, there were two inte trends observed: • The first observation was the gains in flexion. Flexion tended to decreas humeral distalization (or lowering). There was a threshold around 38 mm was associated with the worse RoM values; shoulder flexion was better w resulting humerus position was less distal (R 2 = 0.29, p = 0.07). This "thre effect of humeral distalization was also observed for other shoulder motio the correlation was not as strong. The gains in ER1, ABD, ER2, and IR1 s better when the humerus was lowered less than 35 or 38 mm, depending patient. • The second observation was the gains in ER1, which tended to improv humeral lateralization (R2 = 0.29, p = 0.07). Among our other results, a positive linear relationship was found betwe improvement in ABD, F, IR1, and ER2 and the anterior humeral displacement. H lateralization was associated with a slightly worse range of motion in ER2 and IR1 There were no statistically significant findings; however, there were two interesting trends observed:

•
The first observation was the gains in flexion. Flexion tended to decrease with humeral distalization (or lowering). There was a threshold around 38 mm, which was associated with the worse RoM values; shoulder flexion was better when the resulting humerus position was less distal (R 2 = 0.29, p = 0.07). This "threshold" effect of humeral distalization was also observed for other shoulder motions, but the correlation was not as strong. The gains in ER1, ABD, ER2, and IR1 seemed better when the humerus was lowered less than 35 or 38 mm, depending on the patient.

•
The second observation was the gains in ER1, which tended to improve with humeral lateralization (R 2 = 0.29, p = 0.07).
Among our other results, a positive linear relationship was found between the improvement in ABD, F, IR1, and ER2 and the anterior humeral displacement. Humeral lateralization was associated with a slightly worse range of motion in ER2 and IR1.
RoM and Measurement of 2D Humeral Displacement Using the DSA and LSA (Figure 7). A positive linear regression was observed between F, ABD, ER1, ER2, and humer distalization assessed using the DSA. On the other hand, the IR1 decreased when the DS increased. The greater humeral lateralization assessed using the LSA was associated wi better ER1 and IR1 but worse F and ER2. However, none of these findings we statistically significant. A positive linear regression was observed between F, ABD, ER1, ER2, and humeral distalization assessed using the DSA. On the other hand, the IR1 decreased when the DSA increased. The greater humeral lateralization assessed using the LSA was associated with better ER1 and IR1 but worse F and ER2. However, none of these findings were statistically significant.

RoM and
The 2D and 3D Humeral Displacement Measurements (Figure 8).  No correlation was found between the measurements of 3D humeral displacement using the ACP and 2D displacement using the LSA and DSA.

Discussion
Restoring active shoulder flexion is one of the main functional objectives of RSA. In this clinical study, the main finding was that excessive humeral distalization (or lowering) after RSA implantation somewhat reduced flexion amplitude. This observation was made based on the 3D and 2D analyses of humeral displacement. The mean distalization ACP was 33.3 ± 3.8 mm, and the flexion amplitude was better for the smallest ACP values. The mean DSA was 48.5 ± 9.7 • , and the flexion amplitude was also better for the smallest DSA values. Werner et al. [33] reported similar results; they found that arm lengthening ranging between 1 and 2.5 cm was associated with a better constant score. Shoulder flexion increased until humeral lowering reached a value of 25 mm. Beyond that, shoulder flexion decreased. Other previously published studies also supported this finding. Jobin et al. [34] published a prospective cohort study that included 49 patients who underwent RSA for cuff tear arthropathy. They evaluated deltoid lengthening and medialization of the center of rotation radiographically and correlated with RoM. They demonstrated that deltoid lengthening was significantly correlated with superior shoulder flexion. Lädermann et al. [27] compared 183 patients with arm lengthening and those with arm shortening after RSA; they found that postoperative shoulder flexion was significantly greater after arm lengthening, 145 • versus 122 • , with a mean difference of 23 • . However, a lengthening threshold was not found in their study.
In our study, the more uniform distribution of ACP distalization values allowed us to identify a threshold value of around 35 mm. However, the greater dispersion of DSA values made it impossible to identify a humeral lowering threshold value for this criterion. In a recent study by Berhouet et al. [29], it was revealed that the greater the humeral distalization, the better the abduction, with no threshold effect. However, that study was a computer analysis of passive virtual glenohumeral mobility. These new findings leave us wary of the preoperative planning data provided using the dedicated software and their application during the intervention. In other words, even though lowering the humerus is theoretically beneficial for improving the action of the deltoid and therefore joint mobility, too much humeral distalization is harmful in practice, with less RoM in flexion. It becomes nonsensical to plan more than 25 or 30 mm humeral distalization for RSA. Additionally, it would be difficult to physically reduce the implanted prosthesis at the end of the procedure when the planned excessive humeral lowering is carried out.
The counterproductive threshold effect of excessive humeral distalization, measured using the ACP, was also observed for the other RoM values measured in this study: ABD, ER1, ER2, and IR1. On the other hand, no functional limitation was observed relative to humeral lateralization measured using the ACP. This tended to mainly improve the ER1 but was statistically not significant (p = 0.07). In the axial plane, anterior humeral displacement correlates with better RoM in different directions. No threshold effect was observed. These clinical observations are therefore comparable to those reported in a virtual planning study by Berhouet et al. [29]. We propose the following explanation: The humeral displacement induced by RSA in the anteroposterior (6 ± 6 mm) and mediolateral (4.3 ± 3.7 mm) planes is less important than that generated in the frontal plane with humeral distalization. In fact, the soft tissue loading in these different planes is probably less, not exceeding the limits of muscle elongation. We think the rotations are improved by humeral lateralization and anteriorization, as this prevents impingement with the scapular pillar and restores muscular configuration to a more favorable status for the recruitment of the remaining rotator cuff. Thus, virtual preoperative planning may be the most reliable and the least impacted by soft tissues when evaluating shoulder rotation, based on the humeral displacement in the anteroposterior and mediolateral planes [35,36].
The functional consequences of humeral displacement assessed using 2D angle measurements slightly differ from those previously reported with the ACP. Moreover, we found no correlation between the 3D ACP measurements and the 2D angle measurements. The potential deleterious effect on the shoulder flexion of too much distalization, assessed using the DSA, was not observed for other shoulder motions. Lateralization, assessed using the LSA, was favorably related to ER1, ER2, and IR1 but unfavorably to flexion, and had no impact on ABD. The wide dispersion of these different angle values for humeral displacement in our limited study population very likely explains the difficulty in interpreting these results. This may also reflect a limit on these angle measurements being carried out using a 2D image since such measurements are less precise than those in a 3D reference frame. Measuring the 3D humeral displacement via the ACP at the millimeter level allowed for the identification of a threshold value of distalization on the different shoulder motions. This was not the case with the LSA and DSA, which were less accurate. This is further evidence that a 2D parameter is not always suitable for characterizing a 3D movement or position.
The main limitation of this study is the small number of patients included. This contributes to a lack of statistical power and, thus, the inability to identify statistically significant differences. However, this study has several methodological strengths. It was prospective and based on real clinical data. Additionally, it was built around a specific imaging processing protocol for postoperative CT scans, which facilitated the precise measurement of the humeral displacement obtained after RSA implantation and its comparison with that of the preoperative humeral position. In this study, we sought to objectively evaluate the quality of virtual planning information (via the ACP) by comparing it to actual clinical observations. Further investigation with a larger number of patients and long-term follow-up could be carried out to strengthen our findings.

Conclusions
Beyond the 2D angular or 3D millimeter-level evaluation of humeral displacement after RSA, this study reveals how soft tissue tension impacts the functional results with this type of implant. Determining threshold values for the distalization, lateralization, or even "anteriorization" of the humerus is probably one of the first steps in understanding how to restore the humeral position in order to optimize shoulder joint mobility and prevent deleterious effects. Notably, 3D planning software programs are important elements of this process. The ACP measurement generated with the software used in this study is one of the first parameters designed to explore this issue. Funding: No specific grants were received from public, commercial, or non-profit organizations for this study.
Institutional Review Board Statement: Ethics approval for this study was obtained from our institutional review board (IRB: 13B-T-SHOULDER-RM).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data used to support the findings of this study are included within the article. Raw data are available from the corresponding author.