The revision of a failed shoulder arthroplasty is one of the most difficult problems facing the orthopedic surgeon today. Reverse shoulder arthroplasty has been proven to be useful in complex shoulder reconstruction. The reverse shoulder prosthesis design allows the surgeon to address soft tissue deficiencies and loss of proximal humeral or glenoid bone stock. The results of reverse shoulder arthroplasty in revisions are inferior to other cases, but it does provide a reliable means of treatment for the most complicated cases.
Reverse shoulder arthroplasty is indicated for failed total shoulder or hemiarthroplasty.
It may be used in cases of component loosening, instability, and cuff deficiency.
The procedure is contraindicated in cases of deltoid dysfunction.
Functional results are inferior to primary reverse shoulder arthroplasty.
Pain relief is typically good.
The complication rate is very high, with dislocation being the most common postoperative complication and fracture of the humerus the most common intraoperative complication.
Deltopectoral approach allows access to the humerus.
Humeral osteotomy facilitates component extraction.
Glenoid bone grafting can be done as a one- or two-stage process.
The prosthesis is implanted in maximal tension to lower the chance of dislocation.
Metallic spacers and variable-thickness polyethylene humeral cups are useful to adjust tension.
The subscapularis should be reattached if possible.
Humeral fracture can occur during forceful component extraction.
Poor visualization of the glenoid can lead to improper component placement.
Inadequate tension across the prosthesis can lead to dislocation.
Rehabilitation may progress slower than normal in order to avoid dislocation.
HISTORY OF REVERSE SHOULDER ARTHROPLASTY
The numbers of total shoulder arthroplasties performed in the United States has more than doubled in the last two decades. Survival rates for total shoulder arthroplasties have generally been reported by experienced surgeons to be approximately 85% at up to 15-year follow-up; however, the procedure is being done more frequently by surgeons with limited experience. Approximately three quarters of all surgeons performing shoulder arthroplasty do only one or two procedures each year. The rate of complications for these low-volume surgeons is higher than that of surgeons who do at least 30 shoulder arthroplasty procedures per year. As the numbers of procedures performed increases, this will undoubtedly lead to a greater number of implant failures in the years to come.
Reconstructive options that allow the surgeon to revise these difficult patients will certainly be in demand; however, shoulder arthroplasty remains one of the most difficult and complex problems facing the shoulder surgeon today. Revision surgery using an anatomic shoulder prosthesis has typically yielded worse results than primary arthroplasty, both in terms of function and longevity.
Before discussing revision shoulder arthroplasty, it is helpful to consider the causes of primary athroplasty failure. In a literature review of 32 reports involving 1615 cases of total shoulder arthroplasty (TSA), Wirth and Rockwood noted the most common postoperative complications to be component loosening, instability, rotator cuff tear, periprosthetic fracture, infection, implant failure, and deltoid dysfunction, in descending order of frequency. In contrast, the most common reason for revision of hemiarthroplasty is the development of painful glenoid arthritis, but infection instability, cuff tear, component malpositioning, and component loosening may all play a role.
The glenoid component is the most common source of loosening in total shoulder arthroplasty ( Fig. 26-1 ). Up to 96% of patients with cemented, all-polyethylene glenoids will have radiolucent lines. It uncertain how many patients with evidence of radiolucent lines progress to actual radiographic and clinical loosening. Mileti et al. reported on 62 of 70 patients with cemented glenoids who demonstrated radiolucent lines at final review. The rate of component loosening was 11%, with a total of 14 % of glenoids being judged “at risk.” Glenoid loosening has been linked to mechanical and biologic causes. Eccentric loading leading to the “rocking-horse” glenoid is an example of a mechanical cause. Polyethylene wear debris may also contribute as a biologic cause.
Glenoid component fixation problems have led several investigators to develop metal-backed, noncemented glenoids. Results have been inconsistent. Boileau et al. reported a 20% loosening rate using a metal-backed implant versus a 0% rate using a cemented, all-polyethylene design in a prospective study of 39 cases with 3-year follow-up. Glenoid loosening was associated with severe glenoid erosion and bone loss. Wallace et al. reported good results using a noncemented, metal-backed component in a retrospective series of 86 patients. The rate of radiolucent lines in the cemented group was three times the rate of the noncemented group. Although the authors did report two cases of early polyethylene dissociation in the metal-backed group, functional results and patient satisfaction were not different at final follow-up.
Although glenoid loosening is the major cause of total shoulder failure, it is the combination of mechanical and biologic factors that result in glenoid bone loss that represents the most difficult reconstructive problem for the surgeon. Surprisingly little has been written on dealing with glenoid defects related to revision surgery. The treatment of a small glenoid defect after component removal can be accomplished by using cement to fill the bony defect. Simple removal of the glenoid implant is an option if bone stock is inadequate to support the new implant. Several authors have advocated bone grafting using both corticocancellous bone and cancellous chips to restore glenoid bone stock. Antuna has described one of the larger series to date to address this issue. Of 48 patients undergoing glenoid revision, 25 were treated with glenoid reimplantation with a noncemented (8 cases) or cemented (17 cases) component. Cancellous allograft was used in conjunction with five of the noncemented implants and only one of the cemented implants. Eighteen additional patients underwent implant removal and glenoid bone grafting using allograft chips. Superior results were seen in those patients in whom a glenoid had been reimplanted. The authors went on to recommend a two-stage approach, grafting of the glenoid defect, and delayed placement of a glenoid component after graft consolidation in those patients who continue to have residual pain.
Glenohumeral instability is the second most common cause of total shoulder arthroplasty failure. In a review of the literature, Cofield reported an overall 5.2% incidence. This mode of failure tends to present relatively early in the postoperative period. Typically, it is related to a soft tissue imbalance of some sort. Anterior instability is often associated with component malrotation and subscapularis rupture. Posterior instability often occurs in patients who have had severe posterior glenoid erosion with static posterior instability.
Revision surgery for instability has shown only modest success due to recurrence of instability. Ahrens et al., in a multicenter study, described 51 cases of anterior instability and 29 cases of posterior instability. They found that 87% of those with anterior instability were related to a rupture of the subscapularis. Closed treatment resulted in recurrent instability in all cases for both anterior and posterior dislocations, although patients with posterior instability showed symptomatic improvement. Open treatment was successful in only 40% of anterior instability cases and 53% of posterior instability cases. Sanchez-Sotelo reported on a series of 32 patients who underwent revision surgery for instability using an anatomic prosthesis. Despite attempts to address the underlying cause of instability, more than 50% of patients remained unstable, with revision surgery restoring shoulder stability to only nine patients. Moeckel described seven cases of anterior instability. All seven patients were found to have failed subscapularis repairs, and although all seven underwent re-repair, instability recurred in three.
The third common cause of total shoulder arthroplasty failure is rotator cuff tear. The rotator cuff plays an important role in function after shoulder arthroplasty, but it is deficient in up to 2% of patients after revision surgery. Larger tears in primary settings are known to be associated with poor outcomes and lead to eventual premature glenoid loosening. Smaller, nonretracted tears seem to be of less clinical consequence but can be significant if associated with severe fatty degeneration of the infraspinatus or subscapularis.
In the absence of properly functioning cuff tissue, revision surgery using a nonconstrained total shoulder offers little functional benefit to the patient and risks progressive glenoid loosening. This has led some authors to suggest the use of hemiarthroplasty in cases in which the rotator cuff is suspect. Although this approach can provide some pain relief, postoperative functional improvement has been minimal when compared with the results of reverse shoulder arthroplasty.
Looking specifically at hemiarthroplasty, several investigators have noted severe glenoid erosion to be the most common cause of failure. In these situations, Sperling et al. have reported some improved motion and pain by revising the patient to a total shoulder arthroplasty, but overall results were disappointing due to relatively modest functional gains. Carroll et al. showed that 47% of hemiarthroplasty patients revised to total shoulder arthroplasty had unsatisfactory overall results secondary to unpredictable pain relief and limited motion. The surgeon must be prepared to address possible subscapulais contracture, posterior capsular laxity, eccentric glenoid bone loss, and humeral component malposition.
Hemiarthroplasty after complex proximal humeral fractures represents a third, distinct class of shoulder arthroplasty. As our population ages, these types of fractures are expected to be seen more frequently, and their incidence may triple in the decades to come. Many authors have recommended hemiarthroplasty in the acute setting after a fracture. Pain relief is normally good, yet functional outcomes are often very disappointing.
Poor functional results can be related to a number of factors. Accurate reconstruction of the normal height and retroversion of the proximal humerus is exceeding difficult, even in experienced hands, but Boileau and Bigliani have separately identified problems with tuberosity osteosynthesis as the most important factor associated with poor results. Tanner and Cofield noted that tuberosity displacement was the most common postoperative complication of proximal humeral arthroplasty after an acute fracture. This problem is magnified in the elderly patient with severely osteoporotic bone. Women over the age of 75 have been shown to fare worse than average. In many cases, tuberosity healing is not possible despite optimal bone grafting and fixation. Attempts to revise the fixation of a failed tuberosity osteosynthesis are fraught with difficulty. Many patients are left with poor motion and function secondary to this disruption of the cuff mechanism.
Current indications for the use of a reverse prosthesis in the revision setting include any patient with preexisting prosthetic instability, glenoid bone stock deficiency, or rotator cuff tear. The prosthesis can be used in patients who are noted intraoperatively to have an attenuated cuff insertion. It is possible to jeopardize the cuff insertion during humeral component extraction, and the prosthesis would be beneficial in these cases as well. Tuberosity nonunion after fracture hemiarthroplasty is another indication for use of the reverse shoulder arthroplasty. The prosthesis will allow for some degree of active elevation and maintains joint stability in the presence of massive tissue deficiencies.
There are two main contraindications to reverse shoulder arthroplasty. The first is deltoid dysfunction. Whether related to an injury to the axillary nerve or dehiscence of the deltoid origin or insertion, deltoid dysfunction remains a contraindication to the procedure. Patients will be left with a poorly functioning extremity and the prosthesis will be prone to dislocation. The second contraindication is insufficient bone stock to support the glenoid component. This is a relative contraindication, as custom designed implants with subacromial plates have been used with limited success in selected cases. An implant with an extended central peg may allow glenoid bone grafting and reconstruction even in cases of severe deficiency.
Careful patient evaluation and selection are of paramount importance. A complete history focuses on prior surgical treatments and determines patient clinical and functional expectations. Previous surgical scars are noted. Active and passive range of motion is documented.
Preoperative radiographic studies consist of anteroposterior (AP) views of the glenohumeral joint in neutral, internal, and external rotation, scapular Y, and axillary positions. In addition, full-length AP comparison views of both the affected and contralateral humerus can be helpful in restoring normal humeral length and deltoid tension. A computed tomography (CT) scan can be used to evaluate glenoid bone stock and version, the status of the rotator cuff tendons and muscles, and humeral head stability.
The patient is placed in the beach-chair position, with the affected shoulder just lateral to the edge of the table. This position should allow hyperextension of the humerus and dislocation of the humeral head, permitting excellent exposure for humeral component extraction and canal preparation. In addition, the arm should be easily supported when placed across the patient’s abdomen. Attention to this detail can be of significant benefit because the surgeon will not have to use one hand or employ an additional assistant to support and position the arm throughout the procedure. The shoulder is then sterilized and draped so that there is access to the lateral half of the clavicle anteriorly and the lateral half of the scapula posteriorly.
We routinely use a deltopectoral approach for this procedure, but a superior approach can be used as well. The deltopectoral approach has several significant benefits over the superior approach in the revision setting. With the deltopectoral approach, there is no disruption of the deltoid fibers. Although there are no reports of significant problems when using the superior approach with deltoid function or dehiscence postoperatively, it is an unnecessary risk when a more extensile exposure can be obtained, which can allow easy access to the anterior humerus. This is especially useful in revision arthroplasty, where humeral osteotomy may be necessary. Moreover, identification of the axillary nerve is desirable in the complex cases, and this is best performed through the deltopectoral approach. Optimal glenoid component placement may be easier through a deltopectoral approach. Difficulties with scapular notching have led some to recommend a slightly inferior placement of the glenoid component with an inferior tilt. This is extremely difficult to obtain through a superior approach, with the soft tissues and humeral head pushing upward on the glenoid reamers. The one significant advantage that might be attributed to the superior approach is a lower rate of dislocation, but the ability to adequately expose the humerus overrides this benefit.
The deltopectoral incision is typically begun at the coracoid process and extended distally and laterally for approximately 15 cm, toward the midpoint of the humerus. The interval is easiest to identify at the most superior and medial portion of the incision. Often a small triangle that is devoid of muscle tissue can be seen and then extended distally, marking the deltopectoral interval. In revision cases, the interval is often difficult to see. If it is still present, the cephalic vein is left laterally with the deltoid muscle. Once the interval is opened adequately, the arm is placed in abduction and external rotation and a Homan retractor is placed over the coracoid process. The arm is returned to full adduction and slight external rotation, and the pectoralis major insertion is identified and then released along the superior 1 to 2 cm. If the underlying circumflex humeral vessels are present at the inferior border of the subscapularis, they are ligated using two absorbable sutures. The ligatures should be placed approximately 1 cm apart and cut long to allow easy identification later in the procedure.
A self-retaining retractor is then placed inferiorly. The coracoacromial ligament is identified and divided just lateral to the coracoid insertion. This improves glenoid exposure. Release of the fascia from the lateral portion of the conjoined tendon and coracobrachialis muscle allows a small blunt retractor to be placed underneath to expose the underlying subscapularis muscle. The axillary nerve is identified by placing the arm in adduction, forward flexion, and neutral rotation, and then by following the anterior surface of the subscapularis medially, underneath the conjoined tendon.
The arm is then placed in abduction and internal rotation in order to locate the biceps, which delineates the lateral border of the subscapularis. It should lay just medial and deep to the pectoralis major insertion tendon if it has not been previously tenodesed. Dissection using a pair of scissors oriented perpendicular to the tendon usually allows one to easily enter the sheath and identify the tendon if it is still intact. Keeping the arm in the same position, the superior border of the subscapularis is found just behind the tip of the coracoid process. Scissors can be used to dissect tangential to the tendon, and as the joint is entered, a rush of articular fluid should be seen.
Once all four borders of the subscapularis have been identified, two stay sutures are placed and the tendon is divided approximately 1.5 cm medial to insertion on the lesser tuberosity, following the anatomic neck of the humerus. The upper portion of the tendon is cut using a scalpel blade. Electrocautery is used to section the muscular portion of the subscapularis and release the inferior capsule at the humeral attachment in order to avoid excessive bleeding.
Humeral component extraction is performed by subluxing the head superiorly out of the wound and using an extraction device to grasp the prosthetic head. If an extraction device is not available, an osteotome or bone tamp can also be used to impact against the inferior surface of the head and loosen the stem. If the implant removal is unsuccessful with the judicious use of force, a humeral osteotomy can be performed by using an oscillating saw to cut a line down the anterior humeral cortex between the pectoralis major and deltoid insertions. The osteotomy should extend beyond the level of the inferior cement plug. Osteotomes can be used to further wedge open the osteotomy site. It is sometimes necessary to complete the osteotomy and cable the proximal humerus at the end of the case.
A humeral head retractor is introduced into the joint in order to sublux the proximal humerus posteriorly, and the subscapularis is then released by performing a juxtaglenoid capsulotomy. This is begun by identifying the superior, or semitubular, portion of the subscapularis tendon. Dissecting scissors can be slid along the superior tendon edge, releasing any subcoracoid adhesions. The deep surface of the muscle is then bluntly dissected free from the underlying capsule and middle glenohumeral ligament. The capsule and middle glenohumeral ligament are then sectioned, working back inferiorly and medially, to the glenoid rim. Next, the previously transected muscle fibers of the inferior subscapularis are found. Lying just posterior to these fibers, which are seen in cross section, is the inferior glenohumeral ligament. The inferior glenohumeral ligament and capsule are dissected free and sectioned superiorly and medially back to the level of the glenoid. The excursion of the muscle is then tested, and, if found to be adequate, the muscle is buried in the subscapularis fossa and protected with a small sponge. If tendon excursion is still inadequate, a blunt instrument can be used to palpate for remaining adhesions on the undersurface of the subscapularis. The muscle is then buried in the subscapularis fossa, and a sharp retractor can be placed in the fossa to retract the medial structures including the subscapularis muscle, axillary nerve, and conjoined tendon. Typically, this is done using a Kolbel retractor, but a curved Homan-type retractor can be substituted if necessary.
In cases of total shoulder arthroplasty, glenoid extraction is not typically difficult. Electrocautery is used to release the inferior glenoid capsular attachments, past the 6-o’clock position, around to the 7-o’clock position in the right shoulder or the 5-o’clock position in the left shoulder. The release is done directly at the level of the bony attachment and extends medial to the glenoid rim for 2 to 3 mm. This may result in a small release of the triceps insertion, but this is of no functional consequence. So long as the dissection is performed at the bony insertion, there is no danger to the axillary nerve, and in our experience we have never injured it using this technique. Moreover, this step is absolutely critical in allowing for proper posterior retraction of the proximal humerus and thereby for proper exposure and preparation of the glenoid.
The humeral head retractor is removed, and the proximal humerus is dislocated anteriorly. A sharp Homan retractor can be used to retract any remaining biceps or cuff tendon. If the biceps tendon is still intact, it is cut at the level of the glenoid insertion. The intramedullary humeral head cutting guide is not normally needed, but it can be used to measure humeral version if desired.
The appropriate retroversion of the cut is still a matter of considerable debate, and recommendations have ranged from 0 to 30 degrees of retroversion. Our current strategy is to place the component such that the metaphyseal portion of the prosthesis is contained within the bone of the proximal humerus to the maximum extent possible. This is done without specific regard for the retroversion of the component. In this way, we tend to use the cutting jig as a retroversion gauge, rather than a retroversion guide. Often this technique yields retroversion of between 0 and 20 degrees.
Epiphyseal reaming is done using a hemispherical power reamer. The back edge of the reamer should be held parallel to the plane of the humeral cut and advanced so that this relationship is maintained. Under normal conditions without significant proximal humeral bone loss, the epiphysis is reamed until the flat edge of the reamer is flush with the surrounding bone.
Diaphyseal preparation begins with the removal of any loose cement fragments. The entire cement mantle is only removed in cases where infection is suspected. The proximal cement mantle may need to be removed to allow proper humeral component fit. A varied assortment of cement extraction tools is helpful in these cases. Reaming is typically done with hand reamers, and trial components are used to ensure proper humeral preparation and fit. The trial components are then removed in order to allow proper glenoid exposure and preparation.
Glenoid preparation begins by reinserting a humeral head retractor into the joint and retracting the head posteriorly. If the proximal metaphysis is intact, this will create an impaction fracture of the thin anterior portion of the humerus, what we have termed a “controlled fracture.” The fracture has no effect on prosthetic implantation or fixation. This step is only possible if the humeral trial has been removed, because the trial will block efforts to retract the head adequately. An additional Homan retractor is placed superiorly, over the top edge of the glenoid.
Glenoid component removal is accomplished using a combination of osteotomes, curettes, and rongeurs. After the component is out, the remaining glenoid bone stock is inspected. Ideal fixation of the component means having a secure press fit for the central peg of the glenoid baseplate and at least three securing screws. A custom baseplate with a longer central peg has been used with some success. Glenoid deficiencies can be addressed with corticocancellous or structural bone grafting in either a one- or two-stage procedure. For central defects, we prefer to use a morcelized corticocancellous graft placed beneath the baseplate. Larger defects require structural iliac crest graft. The baseplate central peg should be anchored in the native scapular bone, with the baseplate screws used to augment fixation. If the quality of the fixation is questionable, the humeral component is not implanted and the prosthesis is completed at a second procedure, after sufficient time to allow for graft incorporation. This normally takes at least 3 months ( Fig. 26-2 ).