Surgical Outcomes of Weight-Bearing Shoulders: Arthroscopic Rotator Cuff Repair and Reverse Shoulder Arthroplasty

Abstract

Backgroud

This study aimed to report the short- and midterm outcomes of arthroscopic rotator cuff repair (ARCR) and reverse shoulder arthroplasty (RSA) in weight-bearing shoulders.

Methods

This retrospective multicenter study included 19 cases of ARCR and 10 cases of RSA performed in weight-bearing shoulders from 2009 to 2021. In the ARCR group, postoperative 6-month magnetic resonance imaging confirmed the tendon integrity. In the RSA group, scapular notching, acromial fracture, and implant failure were assessed using plain radiographs, and complications were recorded. In both groups, preoperative and postoperative range of motion and functional scores were documented, along with subjective satisfaction and arm use for weight-bearing on the shoulders. For patients followed up for > 5 years, a midterm analysis was performed.

Results

The ARCR group included 8 men and 11 women (average age, 58.8 ± 8.0 years). Initially, Patte types 1, 2, and 3 were noted in 9, 8, and 2 patients, respectively, and 4 patients exhibited full-thickness subscapularis tears. Four patients showed supraspinatus retear, and 2 patients showed subscapularis retear. Retear of any rotator cuff was observed in 5 patients (26.3%). Twelve patients were followed up for > 5 years; 11 (91.7%) used their operated arm for weight-bearing and 9 (75.0%) were satisfied. The RSA group included 5 men and 5 women (average age, 74.3 ± 7.9 years). Procedures included RSAs for cuff tear arthropathy (n = 6), osteoarthritis (n = 3), and fracture nonunion (n = 1). No cases of dislocation, prosthesis loosening, or disassociation were observed throughout the follow-up. However, 1 patient required implant removal due to infection, and 4 patients showed stage 1 scapular notching. Five patients were followed up for > 5 years, all of whom expressed satisfaction and used their operated arms for weight-bearing, despite mean forward flexion (107.5° ± 12.6°) and American Shoulder and Elbow Surgeons score (61.5 ± 5.3) being less than reported patient acceptable symptomatic state (110° and 76, respectively).

Conclusions

Both ARCR and RSA showed promising outcomes in terms of weight-bearing on the operated arm and subjective satisfaction at short- and midterm follow-up. Therefore, neither of these surgeries should be considered contraindicated for patients with weight-bearing shoulder conditions.

Patients with paraplegia or hemiparesis often rely on their arms for weight-bearing. These conditions are usually seen in poliomyelitis, cerebrovascular accident (CVAs), and spinal cord injury (SCI).¹,²⁾ Weight-bearing on the shoulder makes them vulnerable to degenerative shoulder diseases, such as rotator cuff (RC) tears and osteoarthritis.¹,²⁾ The pathophysiological reasons for shoulder degeneration in weight-bearing shoulders have already been described.³⁾ During weight-bearing, the upward force exerted on the humeral head can lead to increased subacromial and glenohumeral joint pressures, resulting in RC tears and arthritic changes.⁴,⁵⁾

However, surgery in patients with weight-bearing shoulders is associated with certain concerns. For patients, shoulder surgery could restrict position changes and mobility and require assistance from other people, which could be a major constraint on the decision for shoulder surgery. For surgical outcomes, weight-bearing on the operated arm can induce pressure and tensile damage to the repaired tendon in arthroscopic rotator cuff repair (ARCR), especially during arm adduction and extension.³⁾ In addition, glenoid implant loosening and humeral implant sinking could occur in reverse shoulder arthroplasty (RSA).⁵,⁶⁾

Previously, some authors have reported the outcomes of shoulder surgery in patients with weight-bearing shoulders.²,⁷⁾ Kerr et al.²⁾ reported 46 cases of ARCR with minimum 24 months of follow-up in wheelchair-dependent patients. They used ultrasound to detect repaired tendon integrity and reported a high retear rate of up to 33% but showed good final clinical outcomes. Fattal et al.⁷⁾ reported 38 cases of arthroscopic or open RCR in weight-bearing shoulders with mean 1.5 years of follow-up. Although tendon integrity was not reported, final shoulder abduction (125°) and pain relief (intensity 0) were reported after surgery. However, few studies have reported surgical outcomes, including satisfaction and arm use for weight-bearing, with extended follow-up durations. In addition, repair integrity analysis using magnetic resonance imaging (MRI) is required to determine the efficacy of ARCR for these patients.

For RSA, Calek et al.⁵⁾ reported on 21 cases in weight-bearing shoulders with a minimum of 2 years of follow-up, demonstrating a revision rate comparable to that of matched normal ambulating patients. However, baseplate dislocation was more frequently observed in the weight-bearing shoulder group. Kemp et al.¹⁾ reported 16 cases of RSA in weight-bearing shoulders with 2 years of follow-up, showing high rate of early complications: 1 case of baseplate failure and 2 cases of early shoulder dislocation. These results appear to discourage RSA in weight-bearing shoulders. However, surgical outcome analysis including early and delayed complications with extended follow-ups have not been widely reported. In addition, there is a lack of study on prevention strategies for early complications, such as glenoid prosthesis failure and dislocation.

Therefore, this study aimed to report the midterm outcomes of ARCR and RSA in weight-bearing shoulders. The hypothesis was that both shoulder surgeries would demonstrate favorable clinical and radiological outcomes, as well as a good level of satisfaction with arm use for weight-bearing after surgery.

METHODS

This multicenter study was performed at 4 hospitals: Samsung Medical Center, Kyung Hee University Hospital, Keimyung University Dongsan Medical Center, and VHS Medical Center. The Institutional Review Board of each hospital approved the study (IRB No. SMC 2022–09–103, KHUH 2024–01–024, KUDH 2023–004–053, and BOHUN 2023–12–019). Informed consent was waived owing to the retrospective design and lack of additional harm to the patients. All surgeries were performed by a single surgeon at each hospital (JCY, YGR, CHC, and HDL).

All patients who underwent ARCR or RSA with sustained wheelchair or walker ambulation between 2009 and 2021 were included in the study. Patients with less than 2 years of follow-up and those with open RCR were excluded. Finally, 19 cases (18 patients) of ARCR and 10 cases (9 patients) of RSA were included. Patients with RC tears but without glenohumeral arthritis underwent ARCR. Single- or double-row transosseous equivalent repair techniques were performed at each hospital. After repair, the footprint coverage was recorded from types 1 to 4 (1, complete coverage; 2, less than half coverage; 3, a small portion of the humeral head exposed; and 4, a large portion of the humeral head exposed).⁸⁾ If there was a concomitant subscapularis (SSC) tear, it was classified using the Yoo and Rhee SSC tear classification.⁹⁾ SSC repair or debridement was performed. SSC repair was specifically indicated for patients classified as Yoo and Rhee type IIB or higher.¹⁰⁾ Concurrent long head of biceps tendon pathology necessitated either tenotomy for patients under 65 years old or tenodesis for those aged 65 years and older.

Patients with massive RC tears, cuff tear arthropathy, osteoarthritis, and sequelae of proximal humerus fracture underwent RSA.¹¹⁾ The reverse shoulder systems utilized in this study included the Aequalis reversed II (Tornier Stryker), Comprehensive (Zimmer Biomet), Equinoxe (Exactech), RSP (DJO Surgical), and SMR (Lima Corporate) systems, chosen based on the preference of the surgeon. The prostheses were categorized based on 4 distinct design configurations: medial glenoid medial humerus (MGMH, found in Aequalis reversed II and SMR), medial glenoid lateral humerus (MGLH, found in Equinoxe), lateral glenoid medial humerus (LGMH, found in RSP), and lateral glenoid lateral humerus (LGLH, found in Comprehensive).

Standard-length humeral stems and press-fit or cementing techniques were used to prevent early humeral prosthesis shrinkage. Low-filling ratio humeral stem with metaphyseal fixation technique to prevent humeral stress shielding was not used for these patients.⁶⁾ Humeral prosthesis was implanted with a retroversion of 20°–40°. Inlay and onlay humeral prostheses were used in this study cohort. The glenosphere was positioned from flush to overhang approximately 3–5 mm inferior to the margin of the glenoid to prevent inferior scapular notching.¹²,¹³⁾ Furthermore, to avoid early failure of the glenoid prosthesis caused by rotational forces during weight-bearing, secure fixation of the baseplate and strict avoidance of upward tilting were performed.¹⁴⁾ Two types of glenoid prostheses with medial and lateral centers of rotation were used in this study. If possible, the SSC was reattached to the lesser tuberosity.¹⁵⁾

Delayed and protected postoperative rehabilitation was performed in these patients compared with patients with non-weight-bearing shoulders. For ARCR, weight-bearing on the shoulder can cause adduction and extension of the arm, leading to stretching of the repaired supraspinatus (SSP) tendon and increased upward pressure on the humeral head against the acromion.¹⁶⁾ This stretching and pressure damage could disturb the healing of the repaired RC tendon.²,¹⁶⁾ Hence, extended immobilization and protection, as well as avoidance of weight-bearing on the operated arm, were performed. An abduction brace was applied for 5–6 weeks or more postoperatively, according to the repaired tendon status or bone quality. Passive ROM exercises were initiated after brace removal. Activeassisted ROM exercises and strengthening exercises using an elastic band were started at 3 months after surgery.

For RSA, weight-bearing on the shoulder can exert significant pressure between the glenosphere and humeral prosthesis, potentially resulting in glenoid prosthesis loosening or dissociation and humeral prosthesis shrinkage, resulting in shoulder dislocation.⁵,⁶,¹⁴⁾ Hence, patients were asked to refrain from weight-bearing on the operated arm until metal-to-bone union, which occurs 3 months after surgery.¹⁷⁾ An abduction brace was applied for 4 weeks after surgery, and passive ROM exercises began 2 weeks after surgery. Active-assisted ROM exercises and strengthening exercises using an elastic band were initiated 3 months postoperatively.

Patients in the ARCR and RSA groups were instructed to be careful about weight-bearing on the operated arm for 6 months after surgery. They were instructed to seek assistance from a third party when changing positions and to use an electric wheelchair for 6 months after surgery. Partial weight-bearing on the shoulder was allowed after 3 months, whereas full weight-bearing and manual wheelchair use were permitted after 6 months. After this period, patients could resume labor and participate in sports activities involving the operated arm.

Radiologic Evaluations

Preoperative T2-weighted MRI was performed for radiological evaluations. The size of the SSP tear was measured using coronal and sagittal oblique views and measured in mm or classified using the Patte classification.¹⁸⁾ The Y-view on the sagittal oblique view was used for evaluating RC muscle atrophy and fatty infiltration by the Goutallier classification.¹⁹,²⁰⁾ The involved tendons were classified as superior (SSP involvement), anterosuperior (AS, SSC, and SSP involvement), posterosuperior (PS, infraspinatus, and SSP involvement), and anterosuperoposterior (ASP, SSC, SSP, and infraspinatus involvement).

In the ARCR group, postoperative tendon integrity was assessed using a 6-month follow-up MRI. The Sugaya classification was used for SSP retears, and types 4 and 5 were considered retears.²¹⁾ SSC retears were evaluated using sagittal, oblique, or axial views on MRI. High signal enhancement with discontinuity on both axial and sagittal oblique images was considered a retear.¹⁰⁾ Plain postoperative radiographs were obtained at each outpatient visit in the RSA group. Inferior scapular notching using the Nerot-Sirveaux classification²²⁾ and implant loosening were evaluated on the final plain radiograph.

Two expert shoulder surgeons at each hospital who were blinded to the study population (SCK and HGK, YGR and SMR, CHC and DHK, and HDL and YSC) performed the radiological evaluations. Reliability was assessed using both intra- and interobserver evaluations. For continuous variables, the surgeons’ mean values were used. For ordinal variables, if a disagreement was observed, agreement was reached between the surgeons.

Clinical Evaluations

The preoperative and 6-month, 1-year, and final (2 years) glenohumeral ROM, functional scores, and shoulder scaption (scapular plane elevation) strength were evaluated. The ROM included forward flexion (FF), external rotation (ER) with the arm at the side, and internal rotation (IR) behind the back. The FF and ER were measured using a goniometer. IR was graded based on the patient's ability to reach the vertebral spinous process with the tip of the thumb on a continuous scale: T1–T12, 1–12 points; L1–L5, 13–17 points; and the buttocks, 18 points. Functional scores included the pain visual analog scale (PVAS), functional visual analog scale (FVAS), American Shoulder and Elbow Surgeons (ASES),²³⁾ and Constant scores.²⁴⁾ The scaption strength was measured using a handheld force gauge that measures pounds-force (lbf).

At the final visit, complications, reoperation, subjective satisfaction (very satisfied, satisfied, same, or poor), and arm use for weight-bearing were evaluated. For patients who were not present at the final visit, a phone survey was conducted to obtain the following information: reoperation, complications, subjective satisfaction, arm use for weight-bearing, and final function (PVAS, FVAS, and ASES scores). In addition, patients with over 5 years of follow-up were identified, and their midterm outcomes were analyzed.

The demographic characteristics of the ARCR and RSA groups are presented in Table 1. One patient from the VHS Medical Center was included in both the ARCR and RSA groups because he underwent subsequent RSA owing to initial ARCR failure. In the ARCR and RSA groups, the final phone surveys were conducted with 3 patients (15.8%) and 1 patient (10%), respectively. Midterm outcome analyses were performed on 12 patients (63.2%) in the ARCR group and 5 patients (50%) in the RSA group, each with > 5 years of follow-up.

Table 1. Demographics.

Variable	ARCR group (n = 19)	RSA group (n = 10)
Age (yr)	58.8 ± 8.0 (44–72)	74.3 ± 7.9 (60–81)
Sex (male : female)	8 : 11	5 : 5
Diagnosis (RCT : CTA : OA : fracture nonunion)	19 : 0 : 0 : 0	0 : 6 : 3 : 1
Cause of weight-bearing shoulder (Polio : CVA : SCI : general weakness)	8 : 8 : 3 : 0	2 : 5 : 2 : 1
Wheelchair : walker	10 : 9	5 : 5
Duration of weight-bearing shoulder (yr)	29.8 ± 21.0 (2–55)	18.1 ± 20.8 (3–55)
Follow-up duration (mo)	73.0 ± 45.8 (24.0–163.6)	67.1 ± 31.1 (36.6–115.8)
Final phone survey	3 (15.8)	1 (10)
Over 5-year follow-up	12 (63.2)	5 (50)

Values are presented as mean ± standard deviation (range) or number (%) unless otherwise indicated.

ARCR: arthroscopic rotator cuff repair, RSA: reverse shoulder arthroplasty, RCT: rotator cuff tear, CTA: cuff tear arthropathy, OA: osteoarthritis, Polio: poliomyelitis, CVA: cerebrovascular accident, SCI: spinal cord injury.

Statistical Analysis

Statistical analyses were performed using R version 4.0.3 (R Foundation for Statistical Computing). The significance level was set at p < 0.05 with a 2-tailed analysis. Continuous variables are presented as means ± standard deviation, and categorical variables are presented as numbers unless otherwise indicated. Preoperative and final clinical outcomes, which were continuous variables, were analyzed using the paired Student t-test. Continuous and categorical variables in independent groups were compared using either Student t-test or Fisher’s exact test. Cohen’s kappa and intraclass correlation coefficients were used to assess the reliability of radiologic measurements; they were rated as excellent (> 0.75), fair to good (0.4–0.75), and poor (< 0.4). The intra- and interobserver reliabilities of the radiologic measurements were excellent (Supplementary Table 1).

RESULTS

Arthroscopic Rotator Cuff Repair

Mean SSP mediolateral tear size was 18.2 ± 10.1 mm on preoperative MRI. SSP medial retraction according to the Patte classification was observed as follows: type 1 in 9 patients, type 2 in 8 patients, and type 3 in 2 patients, indicating that 17 of 19 patients (89.5%) had small-to-medium-sized tears. Partial SSC tears were observed in 7 patients (Yoo and Rhee types I, IIA, and IIB), while full-thickness SSC tears were observed in 4 patients (Yoo and Rhee types III and IV). Follow-up MRI revealed that 4 of 19 patients (21.1%) had SSP retears and 2 of 19 patients (10.5%) had SSC retears. Retear of any RC, whether SSP or SSC, was observed in 5 of 19 patients (26.3%) (Table 2).

Table 2. Radiologic and Surgical Values in the ARCR Group.

Variable			ARCR cases (n = 19)
Preoperative
	SSP tear size (mm)		18.2 ± 10.1 (5–45)
	Patte classification (1 : 2 : 3)
	SSC tear type (no : I : IIA : IIB : III : IV)*		8 : 1 : 4 : 2 : 3 : 1
	Involved tendon (S : AS : PS : ASP)		8 : 9 : 0 : 2
	Goutallier classification (0 : 1 : 2 : 3 : 4)†
		SSC	2 : 7 : 3 : 2 : 3
		SSP	0 : 6 : 5 : 5 : 1
		ISP	4 : 7 : 5 : 0 : 1
Intraoperative
	Type of repair (1 : 2 : 3)‡		13 : 5 : 1
Postoperative
		SSP Sugaya classification (1 : 2 : 3 : 4 : 5)	5 : 7 : 3 : 1 : 3
		SSP retear	4 (21.1)
		SSC retear	2 (10.5)
		Any retear	5 (26.3)

Values are presented as mean ± standard deviation (range), number, or number (%) unless otherwise indicated.

ARCR: arthroscopic rotator cuff repair, SSP: supraspinatus, SSC: subscapularis, S: superior, AS: anterosuperior, PS: posterosuperior, ASP: anterosuperoposterior, ISP: infraspinatus.

^*Yoo and Rhee SSC tear classification.^{29) †}Y-view in the sagittal oblique view was available in 17 patients. ^‡1, complete footprint coverage; 2, less than half footprint coverage; 3, uncovered footprint.

Univariable analysis of SSP retear found no significant variables. Cause of a weight-bearing shoulder, use of a wheelchair or walker, RC tear size, muscle atrophy and fatty infiltration, and type of repair were not associated with SSP retears (all p > 0.05). Among the 4 patients with SSP retear, 2 initially had Patte type 1 SSP tears, while the other 2 initially had Patte type 2 SSP tears. No SSP retear was observed in patients who initially had Patte type 3 SSP tears. One patient with an SSC retear had an initial Yoo and Rhee type IV SSC tear. Another patient who initially had a Patte type 2 SSP tear and a Yoo and Rhee type IIB SSC tear underwent subsequent RSA for both SSP and SSC retears.

Changes in ROM, functional scores, and scaption strength are shown in Fig. 1. All values demonstrated improvement at the final follow-up (2 years) compared with the preoperative values. Significant improvements were noted in PVAS (6.2 to 1.1; mean difference, −5.1; 95% CI, −6.0 to −4.1; p < 0.001), ASES score (46.8 to 79.3; mean difference, 32.5; 95% CI, 25.0 to 40.2; p < 0.001), and Constant score (45.7 to 70.6; mean difference, 24.9; 95% CI, 15.2 to 33.2; p < 0.001). These improvements exceeded previously reported minimal clinically important differences (MCID) for the PVAS (−2.4), ASES score (17.8), and Constant score (10.9).²⁵,²⁶⁾ However, the final ASES score achieved only 91.5% of the reported patient-acceptable symptom state (PASS) of 86.7 points.²⁶⁾

Fig. 1. Range of motion and functional scores in the arthroscopic rotator cuff repair group. (A) Forward flexion (FF). (B) External rotation (ER). (C) Internal rotation (IR). (D) Pain visual analog scale (PVAS). (E) Functional visual analog scale (FVAS). (F) American Shoulder and Elbow Surgeons (ASES) score. (G) Constant score. (H) Scaption strength. ^*Statistically significant by comparing preoperative and final values.

The outcomes of ARCR for large-to-massive tears were further reviewed. Two patients, a 63-year-old man and a 72-year-old woman, both with initial Patte type 3 SSP tears, demonstrated good healing of the repaired tendon. However, they exhibited limited ROM (active FF, 60° and 50°) and weak scaption strength (both 1 lbf). Their satisfaction was moderate, and they experienced restricted arm use for weight-bearing after surgery. Although the number of patients was limited, ARCR for large-to-massive RC tears showed poor clinical outcomes.

Reoperations were performed in 2 patients because of RC retears. As previously mentioned, a 63-year-old male patient, who initially presented with a medium-sized SSP tear (Patte type 2) and a partial SSC tear (Yoo and Rhee type IIB) experienced retears of both the SSP and SSC. Consequently, the patient underwent RSA. Another patient, a 72-year-old woman, underwent revision ARCR and showed good functional recovery after a 24-month follow-up period. Among the other 3 patients with retears, 1 experienced restricted arm use but was satisfied. The remaining 2 patients showed difficulty in using their arms for weight-bearing and poor satisfaction. Finally, 13 of 19 patients (68.4%) were satisfied, and 16 of 19 patients (84.2%) were able to use their operated arms for weight-bearing, either completely or with some restrictions (Table 3).

Table 3. Final Outcomes and Complications in the ARCR Group.

	ARCR group (n = 19)
Reoperation	2 (revision ARCR, 1; RSA, 1)
Final satisfaction (very satisfied : satisfied : same : poor)	8 : 5 : 3 : 3
Final arm usage for weight-bearing (complete : restricted : unable)	12 : 4 : 3

ARCR: arthroscopic rotator cuff repair, RSA: reverse shoulder arthroplasty.

When comparing final outcomes in terms of tendon integrity, patients without retear (n = 14) had higher functional scores compared to those with retear (n = 5; FVAS: 8.2 ± 0.9 vs. 6.5 ± 0.7, p = 0.027; ASES score: 81.8 ± 7.6 vs. 62.5 ± 3.5, p = 0.004). Final satisfaction and use of the operated arm were significantly better in patients without retears (p < 0.01) (Table 4). Considering these findings, we confirmed that healing of the repaired tendon plays a critical role in the recovery of shoulder function in these patients.

Table 4. Final (2 years) Clinical Outcomes in Terms of Repaired RC Tendon Integrity.

	Intact (n = 14)	Retear (n = 5)	p-value
FF (°)	136.5 ± 31.5	123.3 ± 64.3	0.595
ER (°)	45.0 ± 28.8	40.0 ± 17.3	0.780
IR*	9.5 ± 3.5	11.3 ± 3.2	0.435
PVAS	0.9 ± 1.0	1.7 ± 2.9	0.422
FVAS	8.2 ± 0.9	6.5 ± 0.7	0.027†
ASES score	81.8 ± 7.6	62.5 ± 3.5	0.004†
Constant score	71.2 ± 10.6	67.5 ± 0.7	0.644
Scaption strength (lbf)	12.3 ± 4.4	8.0 ± 0	0.212
Final satisfaction (very satisfied : satisfied : same : poor)	8 : 4 : 2 : 0	0 : 1 : 1 : 3	0.007†
Final arm usage for weight-bearing (complete : restricted : unable)	12 : 2 : 0	0 : 2 : 3	< 0.001†

Values are presented as mean ± standard deviation or number.

RC: rotator cuff, FF: forward flexion, ER: external rotation, IR: internal rotation, PVAS: pain visual analog scale, FVAS: functional visual analog scale, ASES: American Shoulder and Elbow Surgeons.

^*T1 to T12 (1–12 points), L1 to L5 (13–17 points), and buttocks (18 points). ^†Statistically significant.

Twelve patients were followed up for > 5 years, and their midterm clinical outcomes are presented in Table 5. The mean FF was 123.0° ± 41.4° and the mean ASES score was 79.9 ± 9.9. Among them, 9 of 12 patients (75.0%) were satisfied, and 11 of 12 patients (91.7%) used their operated arms for weight-bearing with or without restrictions. To assess the potential deterioration of clinical outcomes over time in the ARCR group, functional scores were compared between the 2-year and > 5-year time points. The results indicated no significant deterioration in ROM or functional scores between these time points (Supplementary Table 2).

Table 5. Midterm (> 5 Years) Clinical Outcomes in the ARCR Group (n = 12).

	Preoperative	Midterm	p-value
FF (°)	123.5 ± 40.1	123.0 ± 41.4	0.967
ER (°)	42.0 ± 15.5	41.0 ± 25.9	0.938
IR*	10.4 ± 5.3	9.2 ± 2.9	0.543
PVAS	6.0 ± 1.2	0.9 ± 1.6	< 0.001†
FVAS	3.2 ± 2.6	8.2 ± 1.3	0.002†
ASES score	43.2 ± 9.2	79.9 ± 9.9	< 0.001†
Constant score	50.8 ± 6.4	68.4 ± 11.9	0.021†
Scaption strength (lbf)	5.0 ± 2.0	8.3 ± 1.2	0.077
Final satisfaction (very satisfied : satisfied : same : poor)	-	4 : 5 : 2 : 1	NA
Final arm usage for weight-bearing (complete : restricted : unable)	-	8 : 3 : 1	NA

Values are presented as mean ± standard deviation or number.

ARCR: arthroscopic rotator cuff repair, FF: forward flexion, ER: external rotation, IR: internal rotation, PVAS: pain visual analog scale, FVAS: functional visual analog scale, ASES: American Shoulder and Elbow Surgeons, NA: not available.

^*T1 to T12 (1–12 points), L1 to L5 (13–17 points), and buttocks (18 points). ^†Statistically significant.

Reverse Shoulder Arthroplasty

The radiological and surgical values of the RSA group are shown in Table 6. Five different prostheses were used in this study. Two patients underwent a lateralized glenoid prosthesis design, and 5 patients underwent an onlay humeral prosthesis design. Based on the prosthesis design, the cases were distributed as follows: 4 cases of MGMH, 1 case of LGMH, 4 cases of MGLH, and 1 case of LGLH.

Table 6. Preoperative Radiological and Surgical Values of the RSA Group.

		RSA case (n = 10)
Preoperative*
	SSP tear size (mm)	38.3 ± 5.1 (30–44)
	Patte classification (1 : 2 : 3)	0 : 0 : 6
	Involved tendon (S : AS : PS : ASP)	0 : 0 : 4 : 2
	Goutallier classification (0 : 1 : 2 : 3 : 4)
	SSC	0 : 3 : 5 : 0 : 1
	SSP	0 : 0 : 1 : 4 : 4
	ISP	0 : 2 : 2 : 1 : 4
Intraoperative
	Prosthesis (A : C : D : E : S)	3 : 1 : 1 : 4 : 1
	Lateralized glenoid prosthesis design	2
	Onlay humeral prosthesis design	5
	Prosthesis design (MGMH : LGMH : MGLH : LGLH)	4 : 1 : 4 : 1
	Humeral prosthesis fixation (cemented : press-fit)	4 : 6

Values are presented as mean ± standard deviation (range) or number.

RSA: reverse shoulder arthroplasty, SSP: supraspinatus, S: superior, AS: anterosuperior, PS: posterosuperior, ASP: anterosuperoposterior, SSC: subscapularis, ISP: infraspinatus, A: Aequalis reversed II, C: Comprehensive, D: DJO Encore, E: Equinoxe; S: SMR reverse, MGMH: medial glenoid medial humerus, LGMH: lateral glenoid medial humerus, MGLH: medial glenoid lateral humerus, LGLH: lateral glenoid lateral humerus.

^*Values for 6 cuff tear arthropathy cases.

ROM, functional scores, and scaption strength were improved at 2 years compared with the preoperative values, except for IR (Fig. 2). Significant improvements were noted in FF (65.0°–115.0°; mean difference, 50.0°; 95% CI, 15.0–77.5; p = 0.010), PVAS (5.9–2.4; mean difference, 3.5; 95% CI, 0.3–6.4; p = 0.036), FVAS (3.3 to 6.7; mean difference, 3.4; 95% CI, 0.5–5.9; p = 0.030), and ASES score (18.1–66.0; mean difference, 47.9; 95% CI, 33.2–59.2; p < 0.001). These improvements exceeded the previously reported MCID for FF (12°), PVAS (2), FVAS (1), and ASES scores (10.3) in the RSA.²⁷,²⁸,²⁹⁾ However, the final ASES score did not reach the PASS threshold of 76, achieving 86.8%.²⁷⁾

Fig. 2. Range of motion and functional scores in the reverse shoulder arthroplasty group. (A) Forward flexion (FF). (B) External rotation (ER). (C) Internal rotation (IR). (D) Pain visual analog scale (PVAS). (E) Functional visual analog scale (FVAS). (F) American Shoulder and Elbow Surgeons (ASES) score. (G) Constant score. (H) Scaption strength. ^*Statistically significant by comparing preoperative and final values.

Postoperatively, scapular notching was observed in 4 of 10 patients (40%), all classified as Nerot-Sirveaux type 1 (Table 7). Prosthesis-related complications, acromial insufficiency fractures, and dislocations were not reported. One patient developed a periprosthetic infection and underwent implant removal and cement spacer insertion. A 63-year-old male patient, who had both SSP and SSC retears underwent RSA using an SMR (Lima) prosthesis. The patient had a history of CVA that resulted in left-sided hemiplegia requiring the use of a walker for 36 years. During the follow-up period, no prosthesis-related complications were observed, although type 1 scapular notching was noted. The patient expressed satisfaction and successfully utilized crutches with the operated arm for 5 years. Finally, 9 of 10 patients (90%) were satisfied and able to use their operated arm for weight-bearing, either completely or with some restrictions.

Table 7. Final Radiologic and Clinical Outcomes in the RSA Group.

	RSA group (n = 10)
Scapular notching	4*
Glenoid prosthesis loosening or disassociation	0
Humeral prosthesis loosening or shrinkage	0
Acromial insufficiency fracture	0
Dislocation	0
Infection	1
Revision RSA	0
Final satisfaction (very satisfied : satisfied : same : poor)	6 : 3 : 0 : 1
Final arm usage for weight-bearing (complete : restricted : unable)	7 : 2 : 1

RSA: reverse shoulder arthroplasty.

^*All four cases were Nerot-Sirveaux type 1.

Five patients were available for midterm follow-up, and their clinical outcomes are presented in Table 8. Despite the relatively low values of FF (107.5° ± 12.6°), ASES scores (61.5 ± 5.3), and scaption strength (4.0 ± 0.8 lbf), all patients expressed satisfaction and were able to utilize their operated arms for weight-bearing, with or without restrictions. To assess the potential deterioration of clinical outcomes over time in the RSA group, functional scores were compared between the 2-year and > 5-year time points. The results indicated no significant deterioration in ROM or functional scores between these time points (Supplementary Table 3).

Table 8. Midterm (> 5 Years) Clinical Outcomes in the RSA Group (n = 5).

	Preoperative	Midterm	p-value
FF (°)	62.5 ± 29.9	107.5 ± 12.6	0.013†
ER (°)	25.0 ± 5.7	35.0 ± 5.8	0.092
IR*	12.0 ± 4.7	15.3 ± 1.9	0.381
PVAS	5.0 ± 4.0	2.8 ± 1.0	0.319
FVAS	2.3 ± 2.5	7.9 ± 0.8	0.069
ASES score	15.3 ± 14.6	61.5 ± 5.3	0.004†
Constant score	21.5 ± 16.3	44.0 ± 9.9	0.438
Scaption strength (lbf)	1.0 ± 0.9	4.0 ± 0.8	0.205
Final satisfaction (very : good : same : poor)	-	3 : 2 : 0 : 0	NA
Final arm usage for weight-bearing (complete : restricted : unable)	-	3 : 2 : 0	NA

Values are presented as mean ± standard deviation or number.

RSA: reverse shoulder arthroplasty, FF: forward flexion, ER: external rotation, IR: internal rotation, PVAS: pain visual analog scale, FVAS: functional visual analog scale, ASES: American Shoulder and Elbow Surgeons, NA: not available.

^*T1 to T12 (1–12 points), L1 to L5 (13–17 points), and buttocks (18 points). ^†Statistically significant.

Final Clinical Outcome Comparison between ARCR and RSA Groups

The final (2 years) clinical outcomes were compared between the ARCR and RSA groups. The ARCR group demonstrated superior values across all functional scores, with significant differences in IR (9.9 ± 3.4 vs. 15.4 ± 2.0, p < 0.001), PVAS (1.1 ± 1.4 vs. 2.4 ± 1.0, p = 0.029), FVAS (8.1 ± 1.0 vs. 6.7 ± 2.1, p = 0.044), ASES score (79.3 ± 9.8 vs. 66.0 ± 11.5, p = 0.015), Constant score (70.6 ± 9.7 vs. 44.0 ± 9.9, p = 0.004), and scaption strength (11.5 ± 4.3 lbf vs. 4.0 ± 0.8 lbf, p = 0.035). However, no significant differences were found between the groups in terms of final patient satisfaction and arm use for weight-bearing (all p > 0.7) (Supplementary Table 4).

DISCUSSION

In this study, shoulder surgery, including ARCR and RSA, on weight-bearing shoulders showed promising clinical and radiological outcomes during short- and midterm follow-ups. The retear rate of the repaired SSP or SSC tendon was 26.3% (5 / 19) in the ARCR group, and no complications related to weight-bearing of the operated arm were noted in the RSA group. Most patients in both groups were satisfied and used their operated arms for weight-bearing.

Patients with weight-bearing shoulders face challenges when deciding whether to undergo shoulder surgery. They rely on their arms for various activities such as changing positions and mobility. They usually need assistance from other people after surgery for some period; therefore, favorable surgical outcomes with extended duration should be guaranteed. Therefore, this study investigated the midterm outcomes of shoulder surgery in patients with weight-bearing shoulders, focusing on the use of the operated arm for weight-bearing after surgery.

The healing of tendons in patients with weight-bearing shoulders, compared to those with non-weight-bearing shoulders, is a significant factor in determining the necessity of RCR surgery.³⁾ Some studies have reported the structural outcomes of RCR in patients with weight-bearing shoulders; however, there are some limitations such as missing initial RC tear size and no routine MRI performed for detecting retears.²,³,⁷,³⁰⁾ Additionally, no studies have analyzed SSC healing in these patients. The SSC tendon has been reported as a structure vulnerable to damage in weight-bearing shoulders, which should be addressed during RCR.²⁾

Jung et al.³⁾ reported promising structural and clinical outcomes in open or arthroscopic RCR in 16 wheelchair-bound paraplegic shoulders. Initial RC tear sizes included 2 medium tears (12.5%), 3 large tears (18.8%), and 11 massive tears (68.7%), with a relatively low retear rate of 12%. However, they employed a mixed selection of MRI and ultrasound to detect RC retears. It is generally accepted that ultrasound exhibits lower accuracy than MRI in identifying RC retears.³¹⁾ Additionally, the study's small sample size and short follow-up duration of 13 months were limitations.

Kerr et al.²⁾ performed ARCR in 46 wheelchair-dependent patients, predominantly with SCI (44 / 56), and reported a high retear rate of 33%. However, they did not present the initial SSP tear size, only reporting tendon involvement (87% of SSP, 70% of SSC, and 57% of anteroposterior involvement) and the mean Goutallier classification grade (1.4 for SSC and 1.1 for SSP), making direct comparison of retear rates between studies difficult. Furthermore, ultrasound was used to analyze tendon integrity. Oh et al.³⁰⁾ reported outcomes of open or arthroscopic RCR in patients with lower-extremity disabilities. They included 5 massive tears, 1 large tear, 9 medium tears, and 1 small tear, with retear rates of 25% (4 / 16). However, mixed ultrasound and MRI were used for retear detection with a minimum follow-up of 12 months. Valiquette et al.³²⁾ reported on 15 patients who underwent ARCR with a weight-bearing shoulder and detected 1 patient with a retear using MRI who had sudden weakness and pain during the follow-up period. However, follow-up imaging to detect retears was not routinely conducted, and the retear rate was not documented. Additionally, the initial size of the SSP tears was not reported; only the mean Goutallier grade of 2.64 was reported.

In the current study, routine 6-month postoperative MRI was used for retear detection, and both SSP and SSC retears were analyzed. Previous studies have shown that RC retears are usually observed between 6 and 26 weeks after surgery and infrequently after 3 months, making the timing of follow-up MRI in this study reasonable.³³⁾ SSP retears were observed in 4 patients, specifically 2 of 9 in Patte type 1 and 2 of 8 in Patte type 2 SSP tears. SSC retears were observed in 2 patients. A total of 5 patients showed RC retears, resulting in a retear rate of 26.3%. This overall retear rate is comparable to that reported in previous studies (12%–33%)²,³,³⁰,³²⁾ and falls within the generally accepted range for RC repair retear rates (15%–26%).³⁴⁾

Kerr et al.²⁾ reported a 70% involvement of SSC tears in weight-bearing shoulders. Due to repetitive and forceful wheelchair propulsion, they claimed that RC failure in weight-bearing shoulders occurs primarily anterosuperiorly. However, they did not report detailed SSC healing after surgery. No previous studies have focused on SSC healing despite its increasing importance. In our study, 2 cases of SSC retear led to unfavorable outcomes. One patient required reoperation (RSA), whereas the other showed dissatisfaction, supporting the importance of SSC management in these patients.

The subgroup analysis revealed that patients with retears exhibited significantly poorer functional outcomes. Furthermore, patients with RC retears were dissatisfied and showed difficulty in weight-bearing on the operated arm, which is the primary goal of surgery for these patients. Repaired tendon healing is one of the most important prognostic factors for RC repair surgery.²¹,³⁵⁾

Predisposing factors for RC retears have been well reported in previous studies, including large tear size, advanced fatty infiltration and atrophy of muscle, and incomplete restoration of the tendon to the footprint.⁸,³⁵,³⁶⁾ In weight-bearing shoulders, Kerr et al.²⁾ reported that over 2 tendon tears were associated with RC retears. Goldstein et al.³⁷⁾ reported a case series of 6 patients with SCI and noted retear in 5 of 6 patients, which were associated with smoking, low initial ROM, and advanced SSP and infraspinatus muscle atrophy. Oh et al.³⁰⁾ described using a crutch as a factor for retears. Contrary to previous studies, risk factors for retears were not identified in this study. In addition to previously known factors, various other factors may affect tendon recovery in these patients, such as general health status, contralateral shoulder condition, the degree of shoulder weight-bearing during movement, shoulder protection and rehabilitation after surgery, and socioeconomic status. However, detailed analysis of these factors is challenging.

Arm protection and rehabilitation are important during surgery in these patients. Although some differences were observed in previous studies, an extended period of protection with a sling and a long duration of avoidance of weight-bearing on the shoulder were commonly reported. Kerr et al.²⁾ recommended 6 weeks of electric wheelchair use, with transfer training initiated between 8 weeks and 16 weeks postoperatively. Additionally, rehabilitation was performed under the supervision of a physical therapist 6 weeks after surgery, but showed high retear rate (33%). Oh et al.³⁰⁾ prescribed 8 weeks of electric wheelchair use and 10 weeks of non-weight-bearing assistance. Independent transfer with a manual wheelchair was permitted 10–12 weeks after surgery, resulting in a retear rate of 25%. However, they did not report duration of immobilization. Jung et al.³⁾ implemented the most protective protocol of 8 weeks of immobilization, followed by 6 months of protection from weight-bearing. No wheelchair propulsion was allowed until 6 months postoperatively to protect the repaired cuffs. This approach may have resulted in the lowest retear rate (12%), although 11 of the 17 cases involved massive tears.

In our study, delayed and protected postoperative rehabilitation was also implemented in these patients to protect the repaired RC tendon compared to the non-weight-bearing shoulder. An abduction brace was applied for 6 weeks or more postoperatively, which is longer than the reported immobilization period.²,³⁰⁾ Additionally, weight-bearing on the operated arm was avoided for 6 months. In a previous study, Jung et al.³⁾ performed a longer period of immobilization (8 weeks) than in our study. Consequently, patients were advised to seek assistance for posture changes during this period and recommended the use of an electric wheelchair, similar to previous studies.²⁾ This approach was taken to prevent pressure damage to the RC tendon and ensure secure fixation of the RC to the bone, which would be a key factor for ARCR in weight-bearing shoulders.

Despite these challenges, favorable clinical outcomes of ARCR in weight-bearing shoulders have been reported. Kerr et al.²⁾ reported significant improvements in the ASES (56 to 92) and Constant scores (50 to 80), and Jung et al.³⁾ also reported improved ASES (53 to 85) and Constant scores (48 to 75), which were higher than the MCID, substantial clinical benefit, and PASS threshold.²,³,²⁵,²⁶⁾ Kerr et al.²⁾ showed that approximately 90% of patients were satisfied after surgery, although the retear rate was high (33%). Interestingly, Jung et al.³⁾ reported improved functional scores even in patients with retears. Our study also showed improved pain and function (final ASES score, 79.3; Constant score, 70.6), which were slightly lower than those reported in a previous study.²,³⁾ However, the differences were lower than the MCID of ASES (11.1) and Constant scores (10.4), which supports the use of ARCR even for weight-bearing shoulders.²⁵,²⁶⁾

Different types of complications have been reported after RSA in weight-bearing shoulders.¹,⁵⁾ Calek et al.⁵⁾ reported a 10% revision rate in RSA for weight-bearing shoulders, which was similar to that for non-weight-bearing shoulders. However, glenoid loosening was more frequently observed in weight-bearing shoulders than in non-weight-bearing shoulders.⁵⁾

Premature weight-bearing on the shoulder before bone union with the prosthesis could generate shear and rotational forces on the glenoid prosthesis, potentially leading to failure.¹,⁵⁾ To mitigate this risk, patients should secure metal-on-bone union and implant stabilization with arm protection in the early postoperative period.¹⁷⁾ Previous experimental studies have shown that osteoblast proliferation, differentiation, and mineralization are mostly induced 4 weeks after implantation, with an average of 41.5% bone ingrowth observed.¹⁷,³⁸⁾ Additionally, a solid bond between the bone and metal prosthesis is typically observed 3 months after surgery.¹⁷,³⁸⁾ In this study cohort, a sling was applied for 4 weeks postoperatively, and partial weight-bearing was allowed 3 months after surgery, aligning with previous studies.¹⁷,³⁸⁾ Full weight-bearing and return to normal activities were permitted 6 months after surgery, with assistance from others for movement and position changes during this period.

The design and fixation techniques of the glenoid prosthesis have also been reported to be important. Calek et al.⁵⁾ noted a high rate of glenoid loosening and disassociation, emphasizing the need for initial secure fixation of the baseplate. To reduce rotational force on the glenoid prosthesis, the MG design, which centers the rotation on the glenoid surface, is recommended.³⁹⁾ In this study, most RSA cases (80%) employed the MG design (including SMR, Exactech, and Aequalis prostheses), and no glenoid complications were observed.

In contrast, Alentorn-Geli et al.⁴⁾ reported 16 cases of comprehensive prostheses with the lateral glenoid design, noting no complications related to the glenoid prosthesis. Our study showed similar outcomes, with no complications observed with the LG design. Firm fixation and inferior tilt of baseplate could provide implant stability.³⁹⁾ Comprehensive prosthesis uses a central compression screw and a peripheral compression and locking screw system, providing firm fixation strength.⁴⁰⁾ In addition, preventing superior tilt was performed in our study to reduce the risk of glenoid prosthesis failure.

Shoulder dislocation has been identified as another major complication after RSA in weight-bearing shoulders. Kemp et al.¹⁾ reported 16 cases of RSA for weight-bearing shoulders, and 3 patients showed early postoperative dislocation. They recommended protecting the arm from weight-bearing for 12 weeks after surgery.¹⁾ Unlike these studies, shoulder dislocation was not observed in our study population.

Early weight-bearing can cause subsidence of the humeral component, contributing to tension imbalance following RSA surgery, especially upon the application of a low filling-ratio humeral stem or short humeral stem to reduce stress shielding.⁶,¹⁴⁾ Tross et al.⁴¹⁾ reported a subsidence of 1.4 ± 3.7 mm in uncemented short humeral stems with no implant-related complications after a mean follow-up of 18 months. In our study, all patients used standard-length humeral stems with either press-fit (60%) or cemented (40%) fixation techniques to make stable humeral prosthesis fixation. Also, weight-bearing was avoided for 3–6 months to ensure biological bone ingrowth.

Rehabilitation protocols following RSA for weight-bearing shoulders vary among surgeons, particularly in terms of the duration of immobilization, assistance from third parties, and restrictions on weight-bearing activities. ¹,⁴,⁵⁾ Some studies recommend 3–4 months of non-weight-bearing with assistance for position changes and transfers and showed no prosthesis-related complications and dislocation.⁴,⁴²⁾ Calek et al.⁵⁾ advised 1–2 months of protection from transportation, which was shorter than previous studies, and reported 2 cases of glenoid prosthesis failure among 17 patients.

In our study, education and strict adherence to postoperative rehabilitation protocols was implemented. Patients were recommended to use an electric wheelchair and seek assistance for repositioning for 6 months after surgery, which is longer than the period recommended in previous studies.¹,⁴,⁵⁾ Partial weight-bearing was allowed 3 months after surgery, with full weight-bearing and manual wheelchair propulsion permitted 6 months after surgery. Although the sample size was small, no major complications, such as implant failure or dislocation, were reported, highlighting the importance of postoperative rehabilitation protocols in these patients.

Despite concerns about complications, RSA for weight-bearing shoulders has shown improved clinical outcomes. Kemp et al.¹⁾ reported improvements in FF from 75° to 112° and in ASES scores from 31 to 73. Calek et al.⁵⁾ reported significant improvements in FF from 65° to 95° and in Constant scores from 29.5 to 65.1,⁵⁾ However, not all final outcome scores were satisfactory. The ASES score change in the study of Kemp et al.¹⁾ achieved MCID and SCB, but not PASS (73 of 76, 96% achieved), while the final Constant score in the study by Calek et al.⁵⁾ achieved MCID (12.6), SCB (26.6), and PASS (65 of 59.1, 110% achieved).¹,⁵,²⁷⁾ In addition, the final FF in each study was 112° and 95°, resulting in 102% and 86% achievement of the PASS threshold (110°).²⁷⁾

In our study cohort, ASES scores improved from 18 to 66, achieving both MCID and SCB, but achieving a PASS threshold of 87%.²⁷⁾ However, the final FF was 115°, achieving PASS threshold.²⁷⁾ Despite varied outcome scores, patients generally reported improved shoulder pain and function after surgery. Additionally, 90% of patients were satisfied and utilized their operated arms for weight-bearing, supporting the promising outcomes of RSA in weight-bearing shoulders.

This study has several limitations. First, being a retrospective multicenter study conducted over an extended period, it may have introduced heterogeneity into the cohorts. The cohort consisted of patients with various underlying conditions affecting shoulder weight-bearing, including poliomyelitis, CVA, SCI, and general weakness. Additionally, the severity of lower-extremity disabilities varied, with some participants using wheelchairs while others used walkers. The duration of shoulder weight-bearing also differed among participants. Furthermore, 5 different RSA prostheses were used, which could influence the final surgical outcomes. The study also included different initial diagnoses, such as proximal humerus fracture, osteoarthritis, cuff tear arthropathy, and failed RC repair. Despite this, long durations of shoulder immobilization and protection were consistently implemented across the study cohorts and appeared to be a significant factor in the surgical outcomes for patients with weight-bearing shoulders.

Second, there was insufficient information regarding the contralateral shoulder. The nondominant arm also endures significant loads, which can affect the operated arm both preoperatively and postoperatively. However, comprehensive information on the contralateral shoulder was lacking.

Third, the limited number of cases in our study restricts the generalizability of our results. Although our study cohort showed comparable retear rates and high satisfaction with the outcome of ARCR in weight-bearing shoulders, it primarily included patients with small-to-medium-sized tears.³⁴⁾ Two patients with large-to-massive RC tears did not report high satisfaction despite achieving tendon healing. This may be due to the infrequency of surgeries performed on weight-bearing shoulders. Nonetheless, our study is valuable as a multicenter investigation with midterm follow-up of weight-bearing shoulder surgeries, an area that has not been well-reported previously.

Fourth, there was a high proportion of follow-up loss in both groups at midterm follow-up, and final clinical outcomes for some patients were obtained via phone surveys. The midterm follow-up rates were 63.2% (12 / 19) in the ARCR group and 50% (5 / 10) in the RSA group. Additionally, 2 of the 12 patients in the ARCR group and 1 of the 5 patients in the RSA group were followed up via telephone surveys, but complete information was not obtained. Regular outpatient follow-up was challenging due to mobility restrictions and poor general condition. Nevertheless, critical outcome variables, such as final functional scores, patient satisfaction levels, and weight-bearing capabilities, have been well documented.

Fifth, a long-term outcome analysis is needed for shoulder surgery in weight-bearing shoulders. This study presents a midterm (> 5 years) outcomes analysis. However, outcomes of RSA tend to decline over time; therefore, long-term outcomes could be lower than those reported in this study, potentially altering the implications of our findings. Additionally, a lower IR score at midterm follow-up, specifically at L3, could significantly impact patient comfort, as IR is critical for weight-bearing shoulder patients during positional changes and transfers.

Sixth, our study did not offer guidance on the selection between ARCR and RSA for patients with large-to-massive RC tears. According to our results, ARCR showed better final functional outcomes than RSA, indicating a preference for RC repair. However, for patients with weight-bearing shoulders, ensuring a single surgery with favorable outcomes could support RSA. Our study mainly included patients with small-to-medium RC tears in the ARCR group, with only a few patients undergoing ARCR for large-to-massive RC tears, which is a limitation.

Both ARCR and RSA showed promising outcomes in terms of weight-bearing on the operated arm and subjective satisfaction at short- and midterm follow-up. Therefore, neither of these surgeries should be considered contraindicated for patients with weight-bearing shoulder conditions.

SUPPLEMENTARY MATERIAL

Supplementary material is available in the electronic version of this paper at the CiOS website, www.ecios.org.

Supplementary Table 1

Intra- and Inter-observer Reliabilities of Radiologic Measurement

Supplementary Table 2

Functional Scores between the 2-Year and > 5-Year Time Points in ARCR Group

Supplementary Table 3

Functional Scores between the 2-Year and > 5-Year Time Points in RSA Group

Supplementary Table 4

Final (2 Years) Clinical Outcome Comparison between the ARCR and RSA Groups