Introduction

A reverse shoulder prosthesis is well accepted as treatment for glenohumeral arthritis associated with an irreparable tear of the rotator cuff. The specific design of the reverse prosthesis obviates the need for an intact rotator cuff by medializing the center of rotation of the glenohumeral joint and distalizing the insertion of the deltoid muscle, thereby increasing its lever arm [3, 8]. Since its introduction, many reports describe favorable functional results in patients with cuff tear arthropathy [1, 3, 5, 14, 15, 17–19, 21, 27, 28–31, 33], osteonecrosis of the humeral head, rheumatoid arthritis and irreparable cuff tears [17, 25, 35], tumors of the proximal humerus [10, 12], revision shoulder arthroplasty [4, 8, 9, 11, 17], and proximal humeral fractures [4, 6, 17]. Restoration of function and adequate pain relief seem consistent in all reports.

Constrained shoulder prostheses also have been advocated for treatment of cuff tear arthropathy. These prostheses, however, are associated with a high rate of complications and revisions [1, 20, 23, 24, 34]. In his original study, Post and Jablon [24] reported 21 complications (87.5%) and 13 revisions (54.1%) in 24 prostheses after a mean followup of 103 months. After redesigning the prosthesis, they noted six complications (7.6%) and three revisions (3.8%) in 78 prostheses after a mean followup of 46.5 months. An extended followup of 50 shoulders from the original series showed the modifications of the prosthesis resulted in a decrease of prosthetic failures but in an increase of prosthetic dissociation and aseptic loosening [23].

Medialization of the center of rotation as proposed by Grammont et al. [15, 16] seemingly reduces the risk of loosening of the glenoid component. Loosening of the baseplate from the glenoid remains a concern, but unscrewing of the glenoid components (baseplate-glenosphere) also has been described [3, 8, 13, 29]. Unscrewing of the glenoid components led to modification of the prosthetic design in 1995 using a Morse taper between the glenosphere and the baseplate. Despite these design modifications, two of us (LDW, PD) recently observed complete disengagement of the glenosphere from the baseplate (Fig. 1), a serious complication that also was observed in a French multicenter study and necessitated revision in 75% of the cases [22].

Fig. 1A–B
figure 1

(A) An asymptomatic partial disengagement (arrow) of the glenosphere is shown. The Constant-Murley score at this time was 61. A few months later, this patient (Patient 1) experienced severe pain and loss of function and the Constant-Murley score decreased to 10. (B) A radiograph shows a complete disengagement of the glenosphere with fracture of the central locking screw, necessitating revision of the prosthesis.

Full size image

We therefore determined the incidence of this unusual complication in reverse shoulder arthroplasties and whether disengagement resulted in adverse clinical effects as compared with patients without radiographic evidence of disengagement. We also sought a possible explanation for the differences observed in the two types of reverse prostheses used in the study.

Materials and Methods

We retrospectively reviewed the clinical and radiographic data of 419 patients with 449 reverse shoulder arthroplasties from the French multicenter study (Omarthrose excentrée, Société Française de Chirurgie Orthopédique et Traumatologique, Paris, France, 2006) and 32 from the University Hospital Pellenberg (Pellenberg, Belgium). The French multicenter database includes 627 shoulder prostheses implanted for osteoarthrosis associated with a rotator cuff deficiency in 596 patients. The implants used in this series were 48 hemiarthroplasties (43 patients), 52 bipolar prostheses (47 patients), and 527 reverse shoulder arthroplasties (502 patients). Of these 527 reverse prostheses, 449 were implanted after 1995 in 419 patients (33 bilateral prostheses; 329 females, 120 males). We reviewed only these 419 patients with 449 reverse shoulder prostheses supplemented with all the reverse shoulder prostheses (n = 32; five males, 27 females) implanted at our institution (University Hospital Pellenberg) between 1995 and 2006. Two female patients died from cancer and were excluded from the study. Thus, we were able to analyze the clinical and radiographic data from 446 patients (33 bilateral prostheses; 354 females, 125 males) with 479 reverse shoulder prostheses. The minimum followup was 12 months (mean, 28.6 months; range, 12–72 months). The type of implant used was a Delta III™ reverse prosthesis (DePuy International Ltd, Leeds, UK) in 468 cases and an Aequalis™ prosthesis (Tornier, Grenoble, France) in 11 cases.

To determine whether disengagement was present, we reviewed all the available postoperative radiographic data from the patients until the time of last followup. Radiographic assessment included an anteroposterior (AP) view tangential to the baseplate and an axillary lateral view. We defined disengagement as a visible step between the baseplate and the prosthetic glenosphere on an axillary and/or AP view tangential to the baseplate (Fig. 2). The Constant-Murley score [7] obtained at the last followup was used to assess the clinical outcome.

Fig. 2A–B
figure 2

(A) An AP view or (B) an axillary view can clearly show disengagement of the glenosphere from the baseplate (circle).

Full size image

Results

The incidence of disengagement was 3.2% (16 patients with 16 shoulders among 479 shoulders; Table 1). In 13 patients, the disengagement was partial, ie, the central screw was not broken and the glenosphere remained seated on the baseplate but with a clear step. In three cases, the central screw was broken, leading to total disengagement of the glenosphere from the baseplate. A partial disengagement was present in eight of 468 Delta III™ prostheses (1.7%) and five of 11 Aequalis™ prostheses (45.4%). The three total disengagements occurred in Delta III™ prostheses.

Table 1 Type of prosthesis, degree of disengagement, and length of followup for all patients diagnosed with a disengagement of the glenosphere
Full size table

We observed no difference in clinical outcome at last followup between patients with partial disengagement and those without disengagement (Constant-Murley score of 59 in both groups). However, central screw breakage, leading to complete disengagement of the glenosphere, always led to a dramatic decrease of the Constant-Murley score, with a severe increase in pain and functional disability. There was no difference in Constant-Murley score for partial disengagement for the two types of prostheses and there was no relationship between the score and the amount of partial disengagement. In Patients 8 and 9, both with an Aequalis™ prosthesis in whom no initial disengagement was seen, disengagement could be seen after 2 years’ followup. Reassembly of a glenosphere that was disengaged on previous radiographs was noted in three patients (Patients 5–7; two Aequalis™ prostheses, one Delta III™ prosthesis) after a mean of 16 months. Thus far, only two patients in this had reoperations. Patient 1 had complete disengagement of her glenosphere, with sudden worsening of the Constant-Murley score (Fig. 1). Patient 2 underwent revision because of severe pain and disengagement of her glenosphere. At the time of revision, breakage of the central screw with asymmetric erosion of the baseplate was observed (Fig. 3). Patient 3 had progressive disengagement of his glenosphere but refused revision because of financial reasons.

Fig. 3A–B
figure 3

Disengagement of the glenosphere from the baseplate results in abnormal motion of the glenosphere. This ultimately can result in (A) central screw breakage and (B) asymmetric wear of the baseplate (arrow).

Full size image

Discussion

The reverse shoulder prosthesis has emerged as an excellent solution for patients with arthritis of the glenohumeral joint associated with deficiency of the rotator cuff. The initial good results of the reverse shoulder prosthesis for treatment of rotator cuff arthropathy have resulted in a dramatic increase in the use of this type of implant. Despite a good clinical outcome in most series, the complications associated with this procedure remain frequent [2, 32]. Bohsali et al. [2] reported a mean complication rate of 24.4% (range, 6.25%–50%). Unscrewing of the glenosphere to baseplate primarily occurred in prostheses used before 1995 [3, 8, 13, 29] and resulted in a change in the prosthetic design in 1995 to a Morse taper design with central screw fixation. It was believed this change of design completely resolved the problem of glenoid dissociation. Molé et al. [22], however, reported on one patient with an implant from 2001 who had disassembly of the glenosphere as a result of improper impaction of the Morse cone. Being confronted with three cases of complete disengagement of a glenosphere, we therefore determined the incidence of this complication.

The mean followup in our study is relatively short (28.6 months), clearly a limitation of our study. A longer followup most likely would show whether greater numbers of prostheses develop disengagement and whether asymptomatic disengagements can become symptomatic with time. Longer followup also would allow a better study of the natural history of the disengagement problem in terms of risk of central screw breakage. Moreover, our reported incidence of glenosphere disengagement may be too low because we could not always obtain a true AP view of the shoulder as a result of technical errors. Accurate measurement of the degree of disengagement necessitates radiographic views that are tangential to the baseplate. Variability in the quality of radiographs in this retrospective review may have resulted in underestimation of the incidence as small disengagements potentially may be missed in cases where the beam is not perfectly tangential.

We found 16 cases of disengagement (3.2%): in 13 cases, the disengagement was partial and there was still good contact between the sphere and the baseplate; in three cases, the disengagement had led to breakage of the central screw, leading to a total disengagement of the glenosphere from the baseplate. The presence of partial disengagement of the glenosphere was not associated with a difference in clinical outcome. Nevertheless, because metallic destruction can result from disengagement, the need for early revision is debatable. Roberts et al. [26] recently stated careful radiographic examination is required to reveal early complications with reverse shoulder prostheses. Our study confirms this statement because our examination of the postoperative radiographs of patients who underwent reverse shoulder arthroplasty revealed an incidence of glenosphere disengagement of 3.2%.

Improper seating of the glenosphere and potential instability between the baseplate can occur because of inadequate reaming of bone around the edge of the baseplate or interposition of bone or soft tissue between the glenosphere and the baseplate at the time of impaction. In the case of interpositioning of material in the Morse taper, the central screw will be tightened without locking the Morse taper. This will result in eventual motion between the glenosphere and the baseplate resulting from necrosis of the interposed tissue. The locking screws of the Aequalis™ and the Delta III™ prosthesis are different (Fig. 4): the locking screw of the Delta III™ prosthesis has less than 1-mm slack whereas the Aequalis™ prosthesis has 2-mm slack. Inadequate reaming or soft tissue interpositioning between the glenosphere and the baseplate also will induce disengagement in the Aequalis™ prosthesis, but because this system allows more slack, abnormal motion between the baseplate and the glenosphere will not occur. Because the locking screw of the Delta III™ prosthesis has less than 1-mm slack, spontaneous reassembly is possible only after breakage of the central screw. This screw failure is induced by a seesaw movement of the glenosphere onto the baseplate. In contrast, the Aequalis™ prosthesis central screw does not exhibit this seesaw movement because the slack is greater than 2 mm, explaining why spontaneous reassembly can occur. The two cases of spontaneous disassembly of the glenosphere (both Aequalis™ prostheses) can be explained by the fact that the glenosphere, although initially flush with the baseplate, was not completely locked by the conus.

Fig. 4A–D
figure 4

The differences in the baseplate-glenosphere assembly between (A, B) the Aequalis™ prosthesis and (C, D) the Delta III™ prosthesis are shown. (A, B) The Aequalis™ locking screw is of the central locking type. When inserted, it falls into the thread of the baseplate. Tightening the locking screw brings the baseplate taper closer to the glenosphere taper. Once the screw is fully inserted, the entire thread of the screw is inside the baseplate. There is no more link between the locking screw and the glenosphere. As a consequence, the surgeon will not be able to unscrew the locking screw out of the glenosphere. Inadequate reaming of the glenoid or soft tissue interpositioning between the glenosphere and the baseplate will induce disengagement, but because the Aequalis™ prosthesis allows more slack, abnormal motion between the baseplate and glenosphere will not occur. (C, D) In the Delta III™ prosthesis, the thread of the locking screw is always in contact with the thread of the glenosphere. This makes removal of the glenosphere easier in revision cases, but when there is incomplete reaming or soft tissue interpositioning, the screw will not be seated entirely in the thread of the baseplate. This can result in abnormal motion between the baseplate and the glenosphere and may lead to central screw breakage.

Full size image

To overcome this potentially serious type of impaction default, we advise the following: (1) excellent surgical exposure of the baseplate to prevent soft tissue interpositioning when mounting the glenosphere onto the baseplate; (2) thorough bony reaming around the glenoid with removal of all bony remnants to permit proper impaction of the glenosphere onto the baseplate; and (3) familiarity with the type of locking screws. Locking screws with limited slack, like those of the Delta III™ prosthesis, need more frequent screwing followed by tapering to ensure complete conical locking. When confronted with a patient with a disengaged glenosphere, we suggest (1) regular clinical and radiographic followups at 3- to 6-month intervals and (2) a CT scan, which may be helpful to determine whether there is screw breakage. In elderly patients with an asymptomatic disengagement that remains stable with time, we do not suggest performing revision. In younger patients and in patients with progressive disengagement, we advise revising the reverse prosthesis.