The use of a modular system to convert an anatomical total shoulder arthroplasty to a reverse shoulder arthroplasty

Over the last ten years, the number of shoulder arthroplasties being undertaken has increased.¹ Although the techniques and results of anatomical shoulder arthroplasty (ASA) have improved, the complication rate is still high.² A common cause for failure is secondary insufficiency of the rotator cuff. This can be because of tendon degeneration, trauma or poor surgical technique such as malsizing or malpositioning of the prosthesis. Failure of the rotator cuff after an ASA usually causes pain and a considerable loss of function. In many cases the only solution is to convert the ASA to a reverse shoulder arthroplasty (RSA), because it is not possible to repair the rotator cuff or to centre the prosthesis using the standard components of an ASA.³

When a hemiarthroplasty has been used to treat a fracture of the proximal humerus, problems may also arise from malunion or resorption of the greater tuberosity leading to profound functional limitation.⁴

Some 20 years ago, Grammont and Baulot⁵ described a reverse shoulder prosthesis which moved the centre of rotation medially and caudally, thereby enabling the deltoid muscle to take over active abduction and elevation from the absent rotator cuff. This has given good results as a primary procedure in patients with an irreparable tear of the rotator cuff, although complications such as infection, instability and scapular notching are common.⁶

Until now, revision of an ASA or hemiarthroplasty to a RSA has required the removal of cemented or well-fixed uncemented components before the reverse prosthesis can be implanted. Removal of the humeral stem sometimes requires osteotomy of the shaft: this can in itself cause problems such as fracture or excessive trauma to the soft tissues.^3,7,8 Removal of the glenoid component often leaves a defect in the bone which may cause difficulty in fixation of the glenoid base plate.

With a convertible prosthesis, fixed components such as the humeral stem and the glenoid base plate can be retained. On the humeral side, the humeral head and the neck component are replaced by a reverse body and liner. On the glenoid side, the metal base plate is left in place and the glenosphere can be mounted directly onto it (Fig. 1).

The purpose of this study was to assess the feasibility of using this system to convert an ASA to a RSA and to determine the clinical and radiological results.

Our hypothesis was that the use of a convertible prosthesis facilitates the revision of a failed anatomical arthroplasty thereby reducing patient morbidity: this might be of particular advantage in elderly patients.

Materials and Methods

Between 2007 and 2011, the senior author (JDA) carried out 457 primary shoulder arthroplasties (421 total and 36 hemiarthroplasty). In 411 cases the SMR prosthetic system (Lima Corporate, San Daniele, Italy) was used, in 46 cases implant systems from other manufacturers were used. Of these, 34 had to be revised, 15 for haematoma and four for infection. In total 15 arthroplasties (14 patients, mean age at time of conversion 70 years, 47 to 83) were converted from an ASA to a RSA using the modular, convertible prosthesis (SMR prosthetic system, Lima Corporate, San Daniele, Italy). Of the 14 patients, eight were women and six were men. The right shoulder was operated on in eight cases and the left in seven. In 11 of the 14 patients surgery was carried out on the dominant side.

The original prosthesis was a cementless total shoulder arthroplasty (TSA) in 13 cases and a hemiarthroplasty for a four-part fracture of the proximal humerus (one cemented, one cementless) in two. The conversion was undertaken at a mean of nine months (six to 14) after primary implantation (Table I). All conversions were performed by the senior author (JDA), through a deltopectoral approach with the patient in the deckchair position. Intra-operatively all stems were found to be well-fixed: there were no cases of loosening or infection.

The anatomical head and neck of the prosthesis were disconnected from the stem and removed (Fig. 1). A special pull-out device was used when removing the neck component to avoid loosening of the humeral stem (Fig. 2). No fracture of the humeral shaft or loosening of the stem occurred as a result of this.

The humeral epiphysis was next prepared to allow the body of the reverse prosthesis to be connected to the stem. In the 13 cases in which an ASA was being revised, the polyethylene inlay was undocked from the baseplate of the glenoid: all of these were well-fixed and could be preserved. A glenosphere was then mounted directly onto the retained baseplate. In the two cases in which a hemiarthroplasty was being revised, a standard glenoid component was implanted. A circumferential capsular release was carried out and the appropriate soft-tissue tension assessed using trial reverse inlays. After selecting the correct thickness of inlay, the joint was reduced and tested for range of movement and stability. There was no need to exchange a stem or change the position of the glenoid baseplate in any case because of problems with soft-tissue tension or instability.

There were no complications intra-opeatively or during the hospital stay. The mean duration of the procedure was 64 minutes (45 to 75). Patients were hospitalised for a mean of eight days (five to 11).

All patients were assessed pre- and post-operatively using the American Shoulder and Elbow Surgeons Score (ASES),⁹ a visual analogue scale (VAS), the Oxford shoulder score (OSS),¹⁰ Western Ontario Osteoarthritis of the Shoulder Score (WOOS)¹¹ and the Subjective shoulder value (SSV)¹² (Table II).

They were also evaluated radiologically by an independent musculoskeletal radiologist (DA) who specifically looked for scapular notching, radiolucent lines, loosening, stress shielding, resorption of a tuberosity, heterotopic ossification and fatigue fracture of the scapular spine or acromion, using conventional radiographs in three planes.

Results

No patients were lost to follow-up. The mean follow-up was 43 months (21 to 83) after conversion. The mean age of the patients at follow-up was 74 years (50 to 87).

All clinical scores improved significantly after surgery (Table II). Pain decreased from a mean of 8 pre-operatively on a VAS to 1 at follow-up. One patient still had pain (6 on a VAS) at follow-up: all the others had little or no pain. The mean ASES increased from 12 (0 to 35) to 76 (23 to 95), the mean OSS from 3 (0 to 17) to 39 (9 to 48), the mean WOOS from 1618 (1770 to 815) to 418 (1340 to 10) and the mean SSV from 15 (0 to 50) to 61 (20 to 95).

Radiologically, there was no lucency around the glenoid in any patient. There was one prosthesis with a lucency around the humeral stem and two with evidence of stress shielding. No loosening of the prosthesis or resorption of the tuberosity was seen. In four shoulders there was slight (grade 1) scapular notching.¹³ The other 11 shoulders were normal in this respect. There was heterotopic ossification of < 5 mm in one shoulder at the inferior aspect of the capsule. No fatigue fracture of the acromion or scapular spine was seen.

One patient suffered a dislocation seven months after the conversion and was treated by closed reduction (Tables I and II): there was no further dislocation.

Discussion

The treatment of cuff tear arthropathy by RSA has become a standard procedure over the last decade.^13,14 The results are good, reproducible and with an acceptable rate of complication particularly when undertaken as a primary procedure.^13,15

Wall et al¹⁶ have shown that patients with posttraumatic arthritis and those in whom the RSA is used as a revision procedure for a failed ASA show less improvement and have a higher rate of complications.

Levy et al^17,18 found that pain and function were improved when a hemiarthroplasty was converted to a RSA, while the complication rate was high, mainly because of bone loss from the glenoid and proximal humerus.

We undertook 15 conversions after a mean of nine months (conversion rate 3.28 %) from ASA or hemiarthroplasty. The indications for conversion were varied. In two cases, a hemiarthroplasty was implanted for an acute four-part fracture of the proximal humerus. In both cases, the conversion to a reverse prosthesis was indicated because of resorption of the greater tuberosity with subsequent failure of the rotator cuff, a mode of failure which is common and well described in the literature.^4,19

In nine patients who had undergone ASA for primary osteoarthritis, the indication for the conversion was a failure of the rotator cuff (seven anterosuperior, two posterosuperior).

These patients had a relative indication for conversion and several alternative treatment options were discussed with them. The ultimate aim of intervention was an improvement in active elevation. Thus, in these patients, the indication for conversion was a failure of shoulder function in the older patient (mean 78 years, 61 to 87), not implant failure. For these patients the primary implantation of an anatomical convertible modular prosthesis offered the opportunity of better functional results at the outset with the option of a less complicated conversion if problems occurred with the rotator cuff.

In the other four patients, the reason for conversion was a failure of hardware. A number of defective polyethylene liners were implanted in 2010 which caused irreparable defects in the rotator cuff. Their manufacture subsequently ceased.

Flury et al³ evaluated the outcome of 21 patients after conversion from an ASA to a RSA with a mean follow-up of 46 months. They reported a significant improvement in pain (mean VAS decreased from 8.7 to 3.0), but most patients were not rendered free of pain. Moreover, they reported a rate of complication of 38% compared with the 6.6% in our study. One reason for the substantial difference between the complication rates must be the degree of trauma sustained by the patient in the revision operation. In Flury’s study a complete exchange of all the components of the prosthesis had to be undertaken. They cite as the main intra-operative complication cracks and fractures of the humeral shaft which occurred during explantation of the stem and removal of the cement.

Ortmaier et al⁸ investigated 50 patients who underwent revision arthroplasty with the implantation of a reverse prosthesis. Osteotomy of the humeral shaft had to be carried out in 40%, cementing of the revision stem in 86% and bone grafting for severe glenoid bone loss in 10%. They also reported a rate of complication of 22.7%.

In our study, the mean operating time was significantly shorter (64 minutes, 45 to 75) than in the study by Flury et al,³ who reported a mean operating time of 236 minutes. This was the result of not having to exchange the glenoid baseplate and humeral stem. No additional procedures, such as fenestration or osteotomy of the humerus, cerclage wire or grafting of the humerus or glenoid, were needed. Our short operating time and minimal soft-tissue dissection almost certainly contributed to the absence of post-operative infection. In Flury’s study³ 9.5% of all patients had their prostheses removed because of infection.

Our indications for conversion were similar to those of Flury.³ Most of the patients underwent revision for rotator cuff insufficiency. In the series of Ortmaier et al⁸ nine of 50 patients underwent conversion for infection. This subgroup had worse results than the others.

In Flury’s study,³ only one patient was able to keep his arm in neutral rotation. This was because of poor infraspinatus and teres minor function caused by the loss of bone from the greater tuberosity and proximal humerus while removing the humeral component. In our study, all patients were able to maintain their arm in neutral actively with no external rotation lag sign, as no humeral bone had been lost during the conversion.

Our clinical results were quite favourable. Hattrup et al²⁰ investigated the results of 19 patients after reverse TSA for the treatment of inflammatory arthritis. They found an increase of the mean ASES from 27 to 76, whereas the mean ASES in our study increased from 12 to 76.

Kelly et al²¹ converted an ASA to a RSA in 28 patients because of rotator cuff insufficiency with a minimum follow-up of 24 months. The mean Constant score improved from 24 to 65 and the mean ASES from 55 to 72. This means that their ASES pre-operatively was better, but was worse at follow-up than in the present investigation. A high complication rate of 50% was also reported in this paper.

We found radiological evidence of grade 1 scapular notching in four cases (27%), which is less than that reported in the literature.^3,14,16 This did not significantly influence either our clinical results or shoulder function. Flury³ reported 28% grade 1, 22% grade 2, 17% grade 3 and 6% grade 4 notching.

Sporman et al²² studied cortical bone resorption after 183 TSAs. They found a prevalence of 17%, mostly occurring within the first year of surgery. Nagels et al²³ found stress shielding in 9% (six patients) and suggested rheumatoid arthritis and a large diameter implant as possible risk factors. In our study we found stress shielding in 13%. It had no effect on function or the need for revision as Sporman et al²² and Papadonikolakis et al²⁴ have also reported.

No fatigue fracture of the acromion or scapular spine occurred in our series, despite the fact that it might be argued that tension in the deltoid after conversion could be high and risk fracture of the acromion. In our opinion, this is mitigated by the design of the prosthesis which allows conversion without unnecessary lengthening (Fig. 1).

Castagna et al,²⁵ in a multicentre study with a mean follow-up of 28 months, reported the clinical and radiological results of converting a hemiarthroplasty or ASA to a RSA using the same modular, convertible system as the present study. In 18 cases the patient had sustained a fracture of the proximal humerus and was treated primarily with a hemiarthroplasty: only in eight cases was a TSA converted to a RSA. The mean operating time in their study was very similar to ours. There was no loosening of the primary components and the humeral stem and glenoid baseplate were preserved in each case. There were no complications intra- or post-operatively. The mean VAS for pain decreased from 8 to 2.

There are limitations of this study. The series is small and reviews only 15 converted shoulders. Consequently, detailed conclusions, especially for subgroups, cannot be drawn. A large sample group and a prospective study design including a second study arm would be very informative.

There was, moreover, no long-term follow-up: however, with a mean of 43 months the patients were followed for longer than in comparable studies. We found promising clinical and radiological results, but long-term results would be helpful, as they are now available for series of primary reverse TSA.²⁶

This study shows that a failing hemi- or TSA can be revised to a RSA with excellent pain relief and a significant improvement in shoulder function. In particular, these results show the benefit of using a modular convertible prosthesis which retains the humeral stem and glenoid base plate. This reduces the extent and duration of the revision procedure thereby decreasing the risk of complications and facilitating a rapid recovery.

In summary, our study confirms the results of Castagna et al,²⁵ but we had a longer mean follow-up (43 months compared with 28 months in Castagna’s study).²⁵ In our study most patients underwent revision of a TSA including the conversion of an implanted glenoid and all operations were performed by a single surgeon.

Author contributions:

T. S. Weber-Spickschen: Data collection and analysis, writing the paper.

D. Alfke: Radiological analysis.

J. D. Agneskirchner: Operations, analysis and proof reading.

We thank K. Ali and S. Huffaker for proof reading.

J. D. Agneskirchner has a consultancy contract with Lima Corporate.

The author or one or more of the authors have received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article.

This article was primary edited by A. C. Ross and first proof edited by J. Scott.

Listen live. A podcast by the authors explaining the importance of their paper is available with the online version of this article at www.bjj.boneandjoint.org.uk