2.1 Study design

This is a prospective, clinical, single-centre study with minimum 1-year follow-up. The study received ethics approval from Monash University (27656). All operations were performed by fellowship trained specialist shoulder surgeons with extensive experience in RSA. All shoulders were replaced with a Tornier Perform® Reversed Wedged Augmented Glenoid (Stryker, Michigan, USA) using either a half- or full-wedge augment and cobalt-chromium alloy (CoCr) glenosphere, articulating with an Aequalis™ Reversed II 155° inlay humeral stem with Ultrahigh Molecular Weight Polyethylene (UHMWE) insert (Tornier, Memphis, USA).

Inclusion criteria were patients aged 90-years or less whom required an RSA for the treatment of primary osteoarthritis, cuff tear arthropathy, massive rotator cuff tear, post-traumatic arthritis, having failed a trial of conservative measures. Exclusion criteria included neurological abnormality, inability to comply with postoperative rehabilitation protocols, Walch C glenoid classification, and previous arthroplasty surgery on the same shoulder. Patients with post-traumatic arthritis, instability arthropathy, avascular necrosis or rheumatoid arthritis were also excluded. Patients with an intact rotator cuff were felt to be more appropriate for an RSA rather than TSA due to the degree of glenoid wear, humeral head subluxation, or concerns about the integrity of the rotator cuff in the future. Patients were classified as glenohumeral arthritis with an intact rotator cuff (OA), or Cuff Tear Arthropathy (CTA), including massive cuff tears (MCT), using pre-operative imaging and intra-operative findings.

All patients had a pre-operative CT scan for planning and assessment of glenoid version, inclination and posterior humeral head subluxation was measured using the BluePrint™ software (Blueprint, Stryker, Kalamazoo, MI, USA).

2.2 Surgical technique

The surgical technique is as published previously¹¹ and was carried out under general anaesthetic with the patient supine in a laminar flow theatre using a neurosurgical headrest. A standard deltopectoral approach was utilised with an Alexis® Orthopaedic Protector (Applied Medical, California, USA). The long head of biceps was tenodesed, and the subscapularis reflected from the lesser tuberosity. The axillary nerve was palpated and protected. A thorough capsular release was carried out. The 155° humeral osteotomy was completed in 10° of retroversion using the provided resection guides. The procedures were all planned using BluePrint™ software. The custom 3D-printed patient-specific guide was used to position the glenoid reaming pin in accordance with the preoperative plan, following which glenoid preparation was completed with reaming for either a full- or half-wedge baseplate as previously planned. The baseplate was fixed in position with a central screw followed by two peripheral screws. The choice of baseplate was made during planning to achieve neutral-to-inferior inclination, maximum inferior positioning, and seating ratio of minimum 80 % with the least possible reaming. A glenosphere of either 36 mm or 42 mm was then secured to the baseplate. Proximal humerus preparation was completed via either an uncemented or cemented technique as required to achieve a secure fit. In patients with an intact rotator cuff, if the superior cuff prevented adequate access to the humeral canal and proximal humerus, a window was created in supraspinatus to aid the access. After a trial reduction the definitive liner was inserted and the shoulder reduced. Once stability was confirmed, the subscapularis was repaired with Ethibond™ sutures (Ethicon, Somerville, USA) and the shoulder closed in layers.

Postoperative rehabilitation consisted of sling immobilisation with light exercises for 6 weeks followed by a physiotherapist-supervised home exercise programme, with overhead strengthening beginning no earlier than 3 months.

2.3 Outcome measures

Patients had pre and post-operative assessments including range of movement, for passive external rotation (ER), passive glenohumeral abduction (GHA), active elevation (AE) measured in degrees, and active internal rotation (IR) scored according to the 0–10 point scoring system in the constant-murley shoulder score (CS). American Shoulder and Elbow Surgeons (ASES), Shoulder Pain and Disability Index (SPADI), Disabilities of the Shoulder and Hand (DASH) and Constant-Murley Scores (CS) at 1 year were collected. Radiographs were routinely obtained immediately post operation, and at 3 months and 12 months. These were standardised with views including an anterior-posterior (AP) view of the glenoid with the arm in 20° of external rotation and 20° internal rotation, lateral and axillary views. The radiograph at 12 months post operation was reviewed independently by a shoulder fellow, and assessed for notching according to Sirveaux et al.¹²

Patients were assessed for post operative complications. Stress fractures were defined as symptomatic patients with pain and/or tenderness, with a confirmed fracture on radiograph or CT. Stress reactions were defined as symptomatic patients with pain and/or tenderness, and a corresponding area of increased signal on SPECT bone scan, but no fracture. All patients who were symptomatic for a possible stress fracture or reaction had additional radiographs, and if no fracture was identified they were referred for SPECT. Statistical analysis was performed on IBM SPSS-29. Comparisons between groups with continuous data were performed with independent sample two-sided t-tests, and chi-squared tests performed for categorical data with statistical significance reported at p < 0.05.

3.1 Outcome measures

All patients receiving a reverse shoulder arthroplasty with the Tornier Perform® Reversed Wedged Augmented Glenoid were assessed for inclusion (73 patients). Eleven were excluded for an indication other than glenohumeral arthritis or cuff tear arthropathy. Sixty-two patients were included, with 30 in the OA group, and 32 in the CTA group. The mean age was 76 in the OA, and 78 in the CTA group. There was a higher proportion of females in the OA group, with higher glenoid retroversion and humeral head subluxation, and lower glenoid inclination. (Table 1).

The pre-operative scores showed the OA group had better scores in the DASH (52 vs 62, p = 0.02), and Constant score (27 vs 20, p = 0.02). Post-operative scores were better in all scores for the OA group, with statistical significance for the DASH (13.8 vs 24.7, p = 0.02) and Constant (68.2 vs 58.7, p = 0.02). (ASES 85.7 vs 79.3, p = 0.08, SPADI 11.5 vs 20.3, p = 0.07) (Fig. 1).

Close

Fig. 1 Figure 1. Pre and Post-operative PROMs. ∗ = p<0.05. Scores adjusted so 100 = best score, 0 = worst score).

Export to PPT

Range of movement showed similar ER between the groups pre and post-operatively. GHA was better in the OA group post-operatively (81° vs 71°, p = 0.046), and AE was better in the OA group both pre (86° vs 55°, p=<0.001) and post-operatively (139° vs 122°, p = 0.03). IR was reduced in the OA group pre-operatively (p = 0.041), but the same as the CTA group post operatively (p = 1). (Fig. 2).

Close

Fig. 2 Figure 2. Pre and post operative range of movement. AE, ER and GH abd in degrees, IR in the 0-10 point scale from Constant Score. ∗ = p<0.05.

Export to PPT

3.2 Complications

The OA group had complications in 4 patients (13 %), with 2 temporary, partial axillary nerve palsies, and 2 undisplaced periprosthetic humeral fractures, all managed successfully non operatively. Twelve patients in the CTA group had complications (38 %). There were 4 acromial stress fractures (12.5 %), and 4 acromial stress reactions seen on SPECT. One had a humeral stress reaction on SPECT, one had a dislocation following a fall, one sustained a skin tear and one developed ulnar neuritis in the sling. No patients returned to theatre.

All patients had a radiograph at 12 months post-surgery. Four patients were found to have grade 1 notching according to Sirveaux et al.¹² One patient was in the OA group and 3 were in the CTA group (p = 0.333).

4.1 Indications

Although initially designed for patients with cuff tear arthropathy, RSA is now commonly used for patients with glenohumeral arthritis (GHOA) with an intact rotator cuff. A large study of the United Kingdom National Joint Registry reported that for patients over 60, undergoing a shoulder arthroplasty for glenohumeral osteoarthritis, with an intact rotator cuff, there was no significant differences between RSA and TSA in terms of long-term revision rates, or serious adverse events.¹³ The patients with an intact rotator cuff in our study were thought to be more appropriately managed with an RSA than a TSA, either due to glenoid bone loss, posterior head subluxation, or concerns regarding the integrity or quality of the rotator cuff.

Our study finds good outcomes with a RSA, with both groups achieving significant improvements in outcome scores and movement. We found that patients with an intact rotator cuff have superior outcome scores and range of movement than those patients who have the procedure for cuff tear arthropathy or massive cuff tears (MCT). This is similar to the findings of Testa et al.¹ who reported on 625 patients, and found those with GHOA with an intact cuff had superior ASES (86 vs 77, p < 0.001), Single Evaluation Numeric Assessment (SANE) (86 vs 77, p < 0.001) and Visual analogue scale (VAS) (0 vs 0, p < 0.01) than those with CTA or MCT. They also found improved active elevation (138° VS 127°, P < 0.001), and change in ER (28° VS 15°, P < 0.001), as seen in our study.

Saini et al.¹⁴ found similar results in 311 patients with superior ASES (87 vs 77, p < 0.01), SANE (90 vs 79, p < 0.01) and VAS scores (0.63 vs 1.2, P < 0.01), along with superior active elevation (139° vs 127°, p < 0.01) and external rotation (54° vs 44°, p < 0.01), and change in internal rotation (2.1 points vs 1.2 points, p < 0.01).

Waterman et al.¹⁵ also reviewed the outcomes of 136 patients, and found superior outcomes for GHOA with an intact cuff for external rotation (52° vs 41°, p = 0.006), and Simple Shoulder Test (SST) (11 vs 9, p = 0.048), although the differences for ASES (86 vs 84, p = 0.084), SANE (85 vs 74, p = 0.055), active elevation (143° vs 134°, p = 0.099) were not statistically significant, they did favour GHOA.

Nove-Josserand et al.¹⁶ reported on the outcomes of RSA in cohorts of patients with primary OA and no cuff tear, primary OA with cuff tear, and secondary OA. They found the Constant Scores were significantly higher in those with OA and an intact cuff (74) compared to OA with cuff tear (66) and secondary OA (64, p < 0.001). Active elevation was also higher postoperatively in the cuff intact group.

Kennedy et al.¹⁷ performed a systematic review of the outcomes of RSA by pre-operative diagnosis, including comparison with rheumatoid arthritis (RA), fracture and revision surgery (Rev). They found no difference between post operative ASES scores between OA, CTA and MCT, but the Constant score was significantly higher in the OA group.

Nazzal et al.¹⁸ reviewed 93 RSAs, with 24 with an intact rotator cuff and 69 without an intact rotator cuff, compared with 93 TSAs. They reported no significant difference between those undergoing RSA with cuff intact versus not intact in ASES (75 vs 79, p = 0.49) or VAS (0.9 vs 0.3, p = 0.09), or ROM.

Glenoid wear is a common issue in performing both TSA and RSA, and several techniques have been used to address this. Eccentric reaming can be used, but can be associated with bone loss and only used to address smaller defects. In RSA, bone grafting techniques have reported good outcomes, but are associated with risks of graft failure or resorption. In Waterman's study¹⁵, 42 % of the patients with GHOA required bone grating of the glenoid, compared to 38 % in the CTA group. No patients in Saini's or Testa's studies received bone grafts or augmented baseplates. Augmented baseplates were introduced in the USA in 2011, with Ghanta et al.¹⁹ performing a systematic review of the outcomes, including 810 patients, the majority receiving Exactech Equinoxe (Exactech Inc, Gainesville, FL, USA) augmented baseplates, with 44 patients from Kirsch's study¹⁰ receiving the Stryker Perform augmented baseplate as used in our study. They reported promising outcomes for the use of augmented baseplates in patients with glenoid wear, with low rates of complications.

4.2 Acromial stress fractures and reactions

Kirsch et al.¹⁰ did report a higher than expected number of acromial stress fractures (ASF) in their series with the same implant (11.4 %). We report an overall ASF rate of 6.5 %, (12.5 % in the CTA group, 0 % in GH OA group), with one patient sustaining the fracture in a fall which also resulting in multiple rib fractures. We also noted a high rate of acromial stress reactions seen on SPECT. In our practice, a SPECT is performed if a patient complains of pain around the acromion or scapular spine, which identified a further 4 stress reactions. This may be under reported in other studies where SPECT is not so readily available or performed. It has been reported previously that degenerative joint disease with a rotator cuff tear is an independent risk factor for ASF following RSA²⁰, and this was also found by Testa et al.¹ They reported an ASF rate of 4.9 % in CTA patients, with a rate of 1.0 % in GHOA patients. Saini et al.¹⁴ reported a rate of 5.3 % in CTA patients compared to 1 % in GH OA patients. One possible explanation is that the intact rotator cuff reduces the forces being transmitted through the scapular spine and acromion, reducing the risk of stress reaction or fracture.

Other factors associated with ASFRs include osteoporosis, prosthesis design, screw position and length in the glenoid²¹, rheumatological disease, corticosteroid injections and oral corticosteroid use.²²

Patients in the CTA group who sustained a stress reaction or fracture (ASFR) reported significantly worse outcome scores that those without an ASFR (Table 2). When those who sustained an ASFR were excluded, there were no statistically significant differences between the GHOA and CTA groups (Table 3). Although previous studies have not reported outcomes when patients with ASFRs have been excluded, those studies reporting a difference in ASFR rates^1,14,16, have shown superior outcomes for GHOA over CTA, whereas those not reporting different ASFR rates^17,18, did not report a significant difference. It is possible that some of the difference in reported outcomes is therefore related to the complication profile, particularly ASFR, rather than the indication specifically for the RSA. Further research is required in this area.