Survivorship of a Medialized Glenoid and Lateralized Onlay Humerus Reverse Shoulder Arthroplasty Is High at Midterm Follow-up

Abstract

Background

Reverse shoulder arthroplasty (RSA) is a common treatment of a variety of disabling shoulder conditions. The purpose of this study was to determine revision-free survivorship after RSA using a medialized glenoid and lateralized onlay-type humerus implant and to identify etiologies of revision.

Methods

All RSAs performed using the Comprehensive^® Reverse Shoulder System (Zimmer Biomet, Inc.; Warsaw, IN, USA) at one institution from 2008 to 2014 were identified through a retrospective review. Charts were reviewed to determine whether the RSA was a primary or revision surgery. Patients were contacted by telephone, and survivorship was defined as no subsequent surgery after RSA. Of the 526 RSAs performed, responses were obtained from 403 patients (77%). A Kaplan–Meier analysis was performed to determine survivorship over time. A χ² test was used to determine differences between revision rates after RSA.

Results

Minimum follow-up was 3 years, and average follow-up was 4.83 ± 1.60 years. Survivorship was 96% at 2 years and 93% at 5 years after RSA. Revisions were performed for instability (n = 8), humeral tray-taper junction failure (n = 5), acute fracture (n = 4), infection (n = 3), glenoid loosening (n = 3), osteolysis (n = 1), or notching (n = 1). Fourteen of the 151 patients (9.2%) who had surgery prior to RSA required revision after RSA. Having shoulder surgery prior to RSA was associated with higher rates of subsequent revision after RSA.

Conclusion

Overall, survivorship after RSA using a medialized glenoid and lateralized onlay-type humerus RSA is high, and prostheses implanted in native shoulders have lower rates of revision at midterm follow-up. Instability (1.9%) was the most common reason for revision.

Introduction

Reverse shoulder arthroplasty (RSA) has become a reliable and increasingly used treatment for the management of rotator cuff tear arthropathy, comminuted proximal humerus fractures, glenohumeral osteoarthritis, inflammatory arthritis, and failed humeral hemiarthroplasty or total shoulder arthroplasty. Several studies have examined the midterm survivorship of the Grammont and Delta III implants [1, 4, 7, 8, 11, 13, 15, 17, 18, 20]. By eliminating the neck and reversing the ball-and-socket, the center of rotation (COR) is medialized to the glenoid surface and the deltoid moment arm is increased (Fig. 1). As seen in the figure, the Grammont design has a COR approximately at the face of the glenoid fossa, and humeral offset is approximately 10 mm [9]. They also feature an inlay design of the humeral component. Despite promising early and medium-term outcomes, there have been higher rates of complications than noted with anatomic total shoulder arthroplasty (TSA) [6]. Particular complications associated with the Grammont and Delta III implants were infection, instability, and scapular notching [1].

Fig. 1.

Measured medial-lateral offsets in scapular plane for anatomic shoulder (a) and three different implant designs: Grammont design (b), medialized glenoid, lateralized humerus design (c), and a lateralized glenoid design (d). GH-r center of rotation of glenohumeral joint. Reprinted with permission from Elsevier: Liou et al. [9].

An onlay-type humeral component, in which the humeral tray rests on top of the humeral stem, is an attractive design modification because it facilitates the modular conversion of an anatomic arthroplasty, such as hemiarthroplasty or TSA, to RSA without revising the humeral stem. In one study, humeral fractures occurred in 24% of cases during removal of a well-fixed prosthesis [17]. The added benefit of an onlay-type component is that it lateralizes the humeral position, which increases soft-tissue tension and reduces scapular notching (Fig. 1) [9]. In a biomechanical study, Giles et al. noted that humeral lateralization decreased deltoid forces compared with humeral lengthening and glenoid lateralization [5]. Figure 1 demonstrates an example of the relative COR for a medialized glenoid and lateralized humerus design as being 4 mm lateral to the glenoid fossa, but the offset of the humerus is 21 mm lateral [9]. One potential downside of an onlay-type humeral component is the need to resect more bone from the proximal humerus. Another is the modularity of the component introducing another potential junction for implant failure. There is a paucity of data on the survivorship of implants with a medialized glenoid and lateralized onlay-type humeral component.

Another type of RSA makes use of a lateralized glenoid design (Fig. 1). The COR is noted to be 13 mm lateral to the glenoid fossa and there is 10 mm offset on the humerus. The native shoulder has a COR that is approximately 26 mm lateral to the glenoid fossa; all such RSA designs technically medialize the COR, but the terms medialized glenoid versus lateralized glenoid refer to where the COR is relative to the glenoid fossa [9].

The purpose of this study was to determine the survival of one specific RSA implant that utilizes a medialized glenoid and lateralized onlay-type humeral component, the Comprehensive^® Reverse Shoulder System (Zimmer Biomet, Inc., Warsaw, IN, USA). Our aims were to determine (1) overall survivorship at midterm follow-up, (2) differences in survivorship between RSAs performed in native shoulders compared with the revision setting, and (3) differences in survivorship between RSA performed for cuff tear arthropathy and any other diagnosis. Our hypothesis was that survivorship would be high at midterm follow-up, revision rates would be higher for patients who had ipsilateral shoulder surgery prior to RSA, and that survivorship would be similar between RSA performed for cuff tear arthropathy and for any other indication.

Material and Methods

This study is a retrospective cohort study of RSAs performed using the Zimmer Biomet Comprehensive system at one institution between 2008 and 2014. Patients were identified and included via operating room implant records. Indications for RSA included rotator cuff tear arthropathy (CTA), primary osteoarthritis with either rotator cuff compromise or severe glenoid bone loss, acute proximal humerus fractures with significant comminution, proximal humerus nonunions or malunions, post-traumatic glenohumeral arthritis with rotator cuff compromise, rheumatoid arthritis with rotator cuff compromise, and revision arthroplasty from humeral hemiarthroplasty or anatomic TSA with rotator cuff compromise. In cases of glenohumeral osteoarthritis with posterior glenoid bone loss, the RSA was used when there was concern for jeopardizing the glenoid fixation with either a pegged or keeled component, or posterior instability with subluxation of the humeral head. Rotator cuff compromise was determined by pre-operative imaging with either computed tomographic scanning or magnetic resonance imaging as well as intra-operatively by direct examination.

Medical records were reviewed for patient age, sex, indication for surgery, and whether the RSA was a primary or revision surgery. Survivorship was defined as not having any further surgery on the shoulder. Medical records were reviewed and patients were contacted via telephone to assess whether they had subsequent surgery after the RSA. If a patient had revision surgery after the RSA, the indications and operative notes were reviewed. Risk factors for survivorship were assessed including surgery prior to the RSA and RSA being performed for CTA versus any other diagnosis.

Of the 526 onlay-type RSAs identified, 403 patients (77%) provided responses and 16 patients (3%) had died with the RSA in place. Survivorship analysis was performed on the 403 patients who responded. Minimum follow-up was 3 years and average follow-up was 4.83 years (standard deviation [SD], 1.60; range, 3–8.5 years). Average age at the time of surgery was 72.3 years (SD, 8.84 years), and 64% of patients were female.

Indication for RSA was CTA in 62% (n = 251), osteoarthritis in 9.4% (n = 38), acute fracture in 7.9% (n = 32), post-traumatic arthritis in 7.2% (n = 29), rheumatoid arthritis in 4.2% (n = 17), and failed hemiarthroplasty or TSA in 8.9% (n = 36); 151 patients (37%) had undergone either arthroscopic or open surgery prior to RSA (Table 1). Fourteen surgeons performed the RSAs but only five surgeons did more than ten RSAs.

Table 1.

Baseline patient characteristics (N = 403)

	n	%
Gender
Male	145	36
Female	258	64
Diagnosis
CTA	251	62
Glenohumeral OA	38	9.4
Acute fracture	32	7.9
Post-traumatic arthritis	29	7.2
Rheumatoid arthritis	17	4.2
Failed hemiarthroplasty or TSA	36	8.9
Previous surgery?
Yes	151	37
No	252	63

CTA cuff tear arthropathy, OA osteoarthritis, TSA total shoulder arthroplasty

Operative Technique

The deltopectoral approach was used in all cases. In cases where the subscapularis was intact, it was managed with either subscapularis tenotomy or peel. The subscapularis was routinely repaired unless it was torn and retracted prior to the case and unable to be reduced back to its footprint. In cases of acute trauma, the tuberosities were repaired to the humeral stem. The subscapularis or lesser tuberosity was repaired in 243 cases, not repaired in 120 cases, and not recorded in the operative note in 40 cases. All glenoid implants were coated in hydroxyapatite (HA) and secured with a 6.5-mm compression screw centrally and four peripheral 4.75-mm locking screws. The humeral stems were press fit in 364 cases, with cement used in 39 cases. Of note, the humeral tray/bearing inclination angle of this prosthesis is 147°. Twenty-one of the cases using cement were for acute fractures, ten were for revisions of previous arthroplasty, and eight were for poor bone quality as assessed intra-operatively.

Post-operative Rehabilitation

Post-operatively, all patients were placed in a shoulder immobilizer that held the arm in internal rotation for an average of 4 weeks. The patient was allowed to perform pendulums and use the hand for activities of daily living and hygiene but were limited from lifting activities. Physical therapy began at 2 to 4 weeks post-operatively with passive range of motion, progressing at 8 weeks to activity as tolerated.

Statistical Analysis

Kaplan–Meier survivorship curves with 95% confidence intervals were generated using the end point of any revision shoulder surgery after RSA. A log-rank was used to determine if there was a difference in survival times between RSA performed in native shoulders compared with those with prior surgery, as well as comparing patients undergoing RSA for a diagnosis of CTA versus any other diagnosis. A χ² analysis was used to determine differences between revision rates after RSA. Fisher’s exact test was used to assess the association between subscapularis repair and revision for instability. The level of significance was set at p ≤ 0.05.

Results

Survivorship was 96% (95% CI, 0.94–0.98) at 2 years and 93% (95% CI, 0.90–0.96) at 5 years after RSA (Fig. 2). There was a 6.2% overall revision rate (n = 25) at midterm follow-up after RSA; revisions were performed for instability (n = 8), humeral tray-taper junction failure (n = 5), acute fracture after a fall (n = 4), infection (n = 3), glenoid loosening (n = 3), osteolysis (n = 1), or notching (n = 1). The majority of revisions (92%) were performed at our institution and the two revisions that occurred elsewhere were for acute periprosthetic fractures after a fall. In six of the eight cases of instability, the subscapularis was irreparable at the time of index RSA (p = 0.018). There was no difference in survivorship between the low- and high-volume surgeons (p = 0.58).

Fig. 2.

Kaplan–Meier survivorship curve (and 95% confidence intervals) for overall revision-free survival after lateralized onlay-type reverse shoulder arthroplasty. The number of patients at risk is depicted at the bottom of the graph.

Survivorship for patients who had shoulder surgery prior to RSA was 93% (95% CI, 0.90–0.97) at 2 years and 89% (95% CI, 0.82–0.95) at 5 years. Fourteen of the 151 patients (9.3%) who had surgery prior to RSA required subsequent surgery after RSA. The previous surgeries ranged from arthroscopic rotator cuff repair (n = 9), anatomic total shoulder arthroplasty (n = 4), or proximal humerus nonunion after open reduction and internal fixation (n = 1). Survivorship for patients who had native shoulders prior to RSA was 97% (95% CI, 0.95–0.99) at 2 years and 94% (95% CI, 0.90–0.98) at 5 years. Eleven of the 252 primary RSAs (4.4%) performed in native shoulders required revision after RSA. Having shoulder surgery, whether arthroscopic or open, prior to RSA was associated with higher rates of subsequent revision after RSA (p = 0.049) on χ² testing and lower survival times (p = 0.050) on log-rank testing (Fig. 3).

Fig. 3.

Kaplan–Meier survivorship curve (and 95% confidence intervals) for revision-free survival after lateralized onlay-type reverse shoulder arthroplasty (RSA) segmented based on RSA performed in a native shoulder versus in revision setting. The number of patients at risk is depicted at the bottom of the graph.

Survivorship for patients undergoing RSA for CTA was 95% (95% CI, 0.93–0.98) at 2 years and 93% (95% CI, 0.90–0.97) at 5 years. Survivorship for patients undergoing RSA for any other diagnosis (osteoarthritis, acute fracture, post-traumatic arthritis, rheumatoid arthritis, and failed hemiarthroplasty or TSA) was 96% (95% CI, 0.93–0.99) at 2 years and 92% (95% CI, 0.87–97) at 5 years. There was no difference in rates of revision (p = 0.98) or survival time (p = 0.90) between patients undergoing RSA for CTA and those undergoing RSA for any other diagnosis (Fig. 4).

Fig. 4.

Kaplan–Meier survivorship curve (and 95% confidence intervals) for revision-free survival after lateralized onlay-type reverse shoulder arthroplasty (RSA) segmented based on RSA performed for rotator cuff tear arthropathy versus any other indication. The number of patients at risk is depicted at the bottom of the graph.

Discussion

The purpose of this study was to define midterm survivorship for an RSA utilizing a medialized glenoid and lateralized onlay-type humerus. Defining survivorship with newer designs of RSA implants is important to consider when counseling patients about the expected longevity of their shoulder replacement. Furthermore, identifying potential risk factors for earlier revision also aids in surgical decision-making as well as pre-operative counseling for patients.

This study has several strengths and limitations. The strengths include that our study is one of the largest studies of RSA survivorship to date and is the largest reporting on lateralized onlay-type humerus RSA. Fourteen surgeons contributed patients to this study and the likely differences in surgical technique (method of neck cut, humeral version, placement of the glenoid baseplate, offset of glenosphere, for example) and rehabilitation protocols may contribute to the generalizability of our findings. The limitations include that the minimum follow-up of 3 years is relatively short for examining survivorship of arthroplasty. Additionally, our study was retrospective in design, and although 80% of patients were alive and able to be contacted and included, 20% were lost to follow-up, which could have introduced selection bias. Also, our focus was on revision-free survivorship, and therefore we did not analyze patient-reported outcomes, range of motion, or routine post-operative radiographs. Furthermore, our analysis did not account for potential complications that were managed without revision surgery.

Overall survivorship for RSA using a medialized glenoid and lateralized onlay-type humerus is high at midterm follow-up. There have been limited studies reporting on survivorship of RSA and most report on RSAs with a humeral inlay design. Our findings of overall survivorship of 96% at 2 years and 93% at 5 years are similar to reported midterm survivorship of other implants [1, 3, 4, 7, 8, 15].

RSAs performed in patients who had prior surgery did have a higher rate of revision at midterm follow-up. Complications and revision surgery have been shown to be higher in RSAs performed in a revision setting [14, 17, 21]. This may have implications in pre-operative counseling, particularly in older patients with massive rotator cuff tears where there is consideration for either attempted rotator cuff repair or RSA. Mulieri et al. demonstrated reliable pain relief and function in patients undergoing RSA for massive, irreparable rotator cuff tears and showed worse survivorship when rotator cuff repair failed prior to RSA [12].

There was also no difference in survivorship when comparing RSA performed for CTA and for another diagnosis. This contrasts with the findings of Guery et al., who demonstrated significantly better survivorship in RSA performed for CTA than for other diagnoses (rheumatoid arthritis, fracture, and replacing an anatomic TSA) for the Delta shoulder prosthesis [7]. Our findings are reassuring, given that RSA use has been increasing greatly, including for diagnoses other than CTA.

The most common reason for revision in our cohort was instability (n = 8, 1.9%). Rates of instability in our cohort were less than the 4.7% incidence reported by a systematic review [21]. Time to revision for instability ranged from post-operative day 9 to 3.8 years after RSA. All dislocations occurred anteriorly and five cases were revised within 3 months of RSA. There was a significant increase in revision for instability in cases where the subscapularis was irreparable at the time of RSA (p = 0.018), which is consistent with the findings of Edwards et al., who also demonstrated increased risk of instability after RSA when the subscapularis was irreparable [2]. This contrasts with the findings of another study of this implant, which demonstrated no differences in dislocations with or without subscapularis repair [19]. However, that study only included primary RSA, whereas six of the eight cases of instability in our cohort were on RSAs performed in the revision setting.

The next most common cause of failure in our cohort was at the humeral tray-taper junction (n = 5). This mechanism of failure has previously been reported in modular onlay humerus RSAs [10]. The first-generation Zimmer Biomet Comprehensive RSA implants had a titanium humeral tray and taper. Notably, four of the five taper failures occurred after the patient sustained a fall, and the fifth occurred when the patient was swinging a golf club and felt a pop. The latest humeral tray-taper failure in our cohort occurred on an RSA implanted in April 2011; there have been no reported tray-taper failures in a prosthesis implanted at our institution since then. Zimmer Biomet reports that globally, there were 82 tray-taper failures in approximately 3630 implantations for the first-generation titanium humeral tray (2.26%) (personal communication to Ryan Rauck, from Y. W. Son, November 14, 2018). Time to revision for humeral tray-taper failure ranged from 2.5 to 6.25 years in our cohort. Dissociation can be difficult to discern on plain radiographs, which were initially interpreted as normal in two of the five patients. The key finding on radiograph is radiolucency at the humeral tray-taper junction (Fig. 5). This interface should normally be radiopaque when no separation is present. The second-generation components of the Comprehensive RSA system are cobalt chrome in the humeral tray and taper; these have not been associated with this type of complication, with 43,616 global implantations, according to Zimmer Biomet (personal communication to Ryan Rauck, from Y. W. Son, November 14, 2018). The change from titanium to cobalt chrome occurred in January 2012, and the first-generation (titanium) humeral tray components were recalled by Zimmer Biomet in December 2016 [16].

Fig. 5.

Anteroposterior and axillary radiographs of a left shoulder demonstrating a radiolucency at the humeral tray-taper junction (arrows).

Fracture after RSA led to revision in four cases (1%). These all occurred after the patient sustained a new traumatic event. Infection was a reason for revision in three cases (0.7%). Two of the three infections occurred in patients who had surgery prior to RSA. The two patients underwent revision within 2 weeks of RSA due to the intra-operative cultures being positive for Propionibacterium acnes at post-operative day 10. One infection occurred in a patient being revised from an anatomic TSA with no history of infection, and the other in a patient who had previously had multiple rotator cuff repairs with a history of P. acnes infection. Both patients had undergone pre-operative aspirations with negative cultures, as well as normal inflammatory laboratory markers. The one infection after RSA in a native shoulder occurred 22 months after surgery and was positive for methicillin-sensitive Staphylococcus aureus. The incidence of deep infection after RSA has been reported to be around 4%, with an increased risk in revision than in primary RSA (5.8% versus 2.9%, respectively) [21].

There were three cases of glenoid loosening that required revision in our cohort. In one case, the glenosphere disengaged from the baseplate 6 months post-operatively. Another case of glenoid loosening occurred when the implant was placed too superiorly on the glenoid, leading to the baseplate dislodging and fracturing the glenoid at 4 months post-operatively. The last case of glenoid loosening occurred in a patient who underwent the lateralized onlay-type RSA as a revision of a prior RSA for glenoid loosening. At the time of lateralized onlay-type RSA, the glenoid was reconstructed using femoral head allograft; 2 years post-operatively, recurrent glenoid loosening required revision to a hemiarthroplasty.

The last two revisions were performed for notching and osteolysis. The notching revision was performed for Nerot–Sirveaux grade 2 notching and subjective complaints of instability with no dislocations. The osteolysis revision was performed 5.5 years post-operatively for significant resorption around the proximal humerus and subjective complaints of instability. Intra-operatively, the humeral polyethylene was noted to be extensively worn inferiorly and was exchanged.

In conclusion, in our retrospective study, overall survivorship for RSA using a medialized glenoid and lateralized onlay-type humerus was high, and prostheses implanted in native shoulders had a lower rate of revision at midterm follow-up. Instability was the most common reason for revision and was associated with an inability to repair the subscapularis at the time of RSA. Furthermore, a unique mechanism of failure for a humeral onlay-type RSA was evident: humeral tray-taper junction dissociation. Surgeons must maintain a high index of suspicion for this type of failure with the first generation of this implant, which can be difficult to diagnose on plain radiographs.

Funding Information

The authors received funding from Zimmer Biomet in support of their research and preparation of this work.

Compliance with Ethical Standards

Conflict of Interest

Ryan C. Rauck, MD, Eric P. Eck, BS, Brenda Chang, MS, Edward V. Craig, MD, Joshua S. Dines, MD, David M. Dines, MD, Russell F. Warren, MD, and Lawrence V. Gulotta, MD, all report receiving grant funding from Zimmer Biomet in support of this research and this article. In addition, Edward V. Craig, MD, reports royalties from Zimmer Biomet and from Wolters Kluwer, outside the submitted work. Joshua S. Dines, MD, reports personal fees and grants from Arthrex, editorial board membership from American Journal of Orthopaedics, and a patent with royalties paid from Conmed, outside the submitted work. David M. Dines, MD, reports financial or material support and royalties from Zimmer Biomet, during the conduct of the study, plus personal fees from Wright Medical Technology, Inc., outside the submitted work. Russell F. Warren, MD, reports stock or stock options from Ivy Sports Medicine and Orthonet and royalties from Zimmer Biomet and Smith & Nephew, outside the submitted work. Lawrence V. Gulotta, MD, reports personal fees as a speaker from Zimmer Biomet, during the conduct of the study.

Human/Animal Rights

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2013.

Informed Consent

Informed consent was waived from all patients for being included in this study.

Required Author Forms

Disclosure forms provided by the authors are available with the online version of this article.