This chapter describes one of the most common complications of total shoulder arthroplasty: glenoid loosening. A review of the literature describes the depth of the problem as well as some of the uncertainties regarding the significance of radiolucent lines with respect to loosening. The importance and role of the history and physical examination are detailed, as are the various imaging methodologies, starting with plain film, the gold standard. Surgical indications and techniques are described for arthroscopic evaluation, arthroscopic removal of loose glenoid components, one-stage revision, two-stage revision with bone grafting, and conversion to hemiarthroplasty, and literature review for outcomes associated with each technique is presented.
History and physical examination:
Initial diagnosis, shoulder comorbitidies, and shoulder surgical history may help identify mechanism of failure of primary total shoulder arthroplasty.
Specific complaints (e.g., mechanical symptoms, constitutional symptoms) may similarly help identify failure mechanism.
Active and passive range of motion identifies excessive scarring and mechanical blocks to motion possibly referable to glenoid migration or failure.
Infection must be ruled out both through history and physical examination and through aspiration if necessary.
Serial radiographs remain the gold standard to show loosening or frank migration.
Computed tomography provides accurate assessments of remaining bone stock and glenoid prosthesis version.
Several classification systems are described. In broad terms, all try to provide diagnostic information by evaluating the width and location of radiolucent lines.
Arthroscopic evaluation may be performed for diagnosis of loose glenoid components or for implant removal. Removal can be accomplished either through traditional portals or through a longitudinal incision in the rotator interval.
Open revisions most often utilize the deltopectoral interval from the index operation and can be staged if necessary.
The decision to perform a two-stage revision of a glenoid component depends most heavily on the amount of bone loss—uncontained glenoid bone defects require bone grafting prior to placement of another glenoid component.
Conversion to hemiarthroplasty can also be performed to provide significant pain relief without subsequent reimplantation of a glenoid component.
The deltoid should be left intact if possible. If absolutely necessary, take the deltoid off the humerus, not the clavicle, and perform a thorough repair.
Dividing the subscapularis tendon at its bony insertion instead of midsubstance provides more flexibility for retensioning the tendon when it is repaired.
Moderate to large peripheral bone defects of the glenoid should be addressed with a two-stage revision after bone grafting.
Grafting with cortical bone for moderate and large defects may help prevent medialization of the glenoid as bone is reabsorbed.
Reaming iliac crest cortical grafts in situ may be easier than reaming on a back table or after placement in the shoulder.
Care must be taken to extract a loose humeral component when this is necessary for visualization and removal of the glenoid component. Use of a humeral window can prevent iatrogenic humeral fracture.
Conversion to hemiarthroplasty without bone grafting may result in medialization of the humeral head with decreased myofascial tension and should be avoided. If medialization occurs, increasing the head size of the humeral prosthesis can mitigate the effect on soft tissue tension.
Proceeding with immediate glenoid revision when faced with large peripheral bone defects increases the risk of failure as the revision component will be poorly supported.
Although range of motion is encouraged, postoperative weight-bearing and strengthening exercises may encourage medialization of the humeral head.
Thirteen thousand total shoulder replacements were performed in 2004, the most recent year for which the National Hospital Discharge information is available, and year on year percentage increases in total shoulder arthroplasty rival those of total hip and total knee arthroplasty. Despite the proven effectiveness of total shoulder arthroplasty at alleviating pain in patients with end-stage arthritis and improving the function of a debilitated joint, concern over the stability and continued functionality of total shoulder prostheses is appropriate given the multiple challenges inherent in revision surgery.
Loosening of the glenoid component remains a major concern because it is one of the most common triggers for revision arthroplasty. An early series by Neer et al. reported lucent lines completely surrounding the keeled glenoid component in approximately 12% of patients and incompletely in another 18% but noted the lines were never consistently wider than 1 mm at an average follow-up of 3 years. Taking patients with both complete and incomplete radiolucent lines into consideration, 94% of the lines were seen on immediate postoperative radiographs, suggesting the presence of radiolucent lines to be largely unrelated to component loading or failure of fixation over time. Toward this point, Dr. Neer commented on the unclear relationship between radiolucent lines and functional results and noted that the series included no patients with clinically apparent loosening of the glenoid component over 3 years on average of follow-up ( Fig. 24-1 ).
The subsequent 25 years of experience in total shoulder arthroplasty has not fully elucidated the significance of radiolucent lines around the glenoid, although numerous series have reported on incidence of radiolucent lines as well as glenoid component loosening.
Estimated incidences of radiolucent lines vary widely from series to series, with reported rates ranging from 30% to 84%. Incomplete radiolucent lines tend to be more common than complete radiolucencies. At an average follow-up of 3.3 years, Cofield described complete radiolucent lines around the glenoid component in 37% of patients, compared with incomplete radiolucent lines in 43%. The bias toward incomplete lucent lines was larger in the series of Barrett et al., who found 54% of patients to have an incomplete radiolucent line and only 20% to have a complete radiolucent line at an average follow-up of 3.5 years. Although several series reported little progression in radiolucent lines over the course of clinical follow-up, Cofield’s series demonstrated progression in terms of radiolucent line extent or thickness in 33% of patients. In contrast to Neer’s suggestion that radiolucent lines are largely acquired at the time of component implantation, this suggests a more worrisome relationship between passage of time and radiographic discontinuity of the cement mantle, especially because follow-up in the Cofield series averaged only 3.3 years. Similarly, in a series with a longer average follow-up of 9.6 years, Torchia et al. reported radiolucent line rates of 84% at final follow-up, with development of the radiolucency at greater than 1 year postoperatively in 45% of the patients with radiolucent lines. Another series by Mileti et al. found that 73% of the patients whose glenoids did not have radiolucent lines on initial postsoperative radiographs developed radiolucent lines by final follow-up at an average of 3.9 years later and found that 76% of radiolucent lines seen on initial radiographs progressed during the same period.
Less clear still is the nature of the association between radiolucent lines and shoulder function and pain. Despite multiple series documenting rates of radiolucent lines well above 50% as well as steady progression of the lines seen, the number of patients who required revision surgery for a clinically loose glenoid component is small. Additionally, although multiple authors have equated early radiolucent lines with incomplete cementing, this association is unproven. From Neer’s initial series of 194 shoulders, no revision surgeries for glenoid loosening were required during the 3.3 years of follow-up. Subsequent series have consistently reported low revision rates ranging from 0% to 6% for glenoid loosening. Torchia et al. with the longest average follow-up at 9.7 years, reported that less than 6% of glenoid components were revised for aseptic loosening. Torchia et al. found a strong association between the radiographically loose glenoid components and significantly higher pain levels. This study also found that 93% of patients who eventually developed a radiographically loose glenoid component had radiolucent lines on initial postoperative films, compared with 44% who did not develop loosening. A subsequent series by Mileti et al. however, did not find an association between pain scores and radiographic criteria that put the shoulder “at risk for clinical loosening.” Details of the various radiographic classification systems to evaluate fixation of the glenoid component on plain radiographs are described in detail later.
In response to the occasional need for revision after total shoulder arthroplasty, pegged glenoid components have been developed that were designed with the intention of improving fixation, minimizing bone loss, and encouraging bony ingrowth. Yian et al. described radiolucency at the glenoid component in 11% of patients on immediate postoperative radiographs, which increased to radiolucent lines in 45% at the final follow-up an average of 3.3 years later, similar to radiolucency rates in the pegged components. Three of 47 (6%) patients’ shoulders were thought to be either “possibly loose” or “definitely loose,” and the study reported a statistically significant relationship between pain scores and radiographically loose glenoid components. Several other series have compared radiolucency rates between pegged and keeled components. Although the papers have small differences in the methodology employed to determine degree of lucency, pegged glenoid components were found to consistently have lower radiolucency scores with statistical significance in comparison to keeled components. The series’ from Lazarus et al. and Gartsman et al. evaluated only immediate postoperative radiographs, whereas Klepps et al. investigated films taken up to a year postoperatively. The clinical significance of decreased radiolucent lines in pegged glenoids is uncertain, especially given the short follow-up periods in these studies and the data from the Torchia series demonstrating a large proportion of radiolucent lines developing at greater than 1 year postoperatively.
HISTORY AND PHYSICAL EXAMINATION
As seen earlier, the relationship between radiographic findings and need for revision surgery secondary to aseptic loosening of the glenoid component is not straightforward. It is for this reason that an accurate history and complete physical examination is essential to clinical decision making when evaluating a patient with complaints of pain or instability after total shoulder arthroplasty.
The history begins with clinical information about the patient’s initial diagnosis necessitating total shoulder arthroplasty and the length of time since surgery. Shoulder co-morbidities noted at the time of surgery, such as significant glenoid bone loss or concurrent rotator cuff tear, may suggest modes of total shoulder failure. Franklin et al. described a statistically significant relationship between superior migration of the humeral head as seen in rotator cuff deficiency and radiolucency grade on imaging at an average of 2.5 years after the index operation and invoked the “rocking-horse” mechanism ( Fig. 24-2 ) in explaining this finding—eccentric loading from the migrated humeral head on the superior aspect of the glenoid component promotes tipping of the component and weakens the cement interface. Knowledge of the implant used in the index surgery is essential given situations which the implant contributes to failure of the arthroplasty, such as the high failure rates of early constrained total shoulder replacements as well as metal-backed glenoid components with screw fixation. In addition, the component type may provide information regarding the amount of glenoid bone stock likely available for placement of a glenoid component at the time of revision surgery should this be necessary. Learning about the postoperative physical therapy regimen after the primary arthroplasty and the patient’s adherence to the regimen helps to identify problematic areas that must be addressed before proceeding with a revision arthroplasty that will require similar attention to rehabilitation as the index operation. Shoulder history with respect to secondary surgeries preceding or subsequent to the total shoulder arthroplasty further defines structural limitations of the musculature or bony anatomy and aids in preoperative planning. Occasionally, medical comorbidities may restrict the therapeutic options available to a patient, although this is increasingly rare given more widespread use of regional anesthesia for upper extremity procedures. Neurologic conditions that increase or decrease muscular tone may endanger the stability and function of revision shoulder arthroplasty.
The specific complaints with regard to the shoulder may occasionally offer hints of underlying component loosening. Subjective instability or mechanical symptoms may give evidence of gross loosening of components, although these symptoms are likely nonspecific. Hawkins et al. studied a series of patients who underwent revision total shoulder arthroplasty for glenoid loosening and found that four of the nine patients had a “painful clunk” with forward elevation of the shoulder that began after an initial postoperative pain-free period. The onset of pain and decreased range of motion was gradual in all patients in the series, and all patients experienced an initial symptom-free period before they began to experience slowly progressive pain and disability. Careful attention to symptoms such as recurrent fever and chills and reports of persistent redness or warmth may suggest the possibility of infection, which necessitates a much different treatment algorithm compared with aseptic loosening. Lack of a fever or shoulder erythema must be interpreted cautiously, however, as a series of 33 patients with infected shoulder arthroplasties described by Kozak et al. demonstrated that only 21% and 33% of patients presented with fever and history of shoulder erythema, respectively. Finally, weakness patterns consistent with rotator cuff deficiency may be related to mechanism of failure, as described earlier.
The physical examination should begin with assessment of the active and passive range of motion of the shoulder. Mechanical blocks to passive forward elevation have been linked to gross instability of the glenoid component. Limitation of passive range of motion can suggest excessive soft tissue scarring that may need to be addressed at the time of revision, and, conversely, instability or excessive motion can indicate tendon detachment and deficiency, depending on the direction and degree of laxity. Active elevation and external rotation can be recorded in degrees, and internal rotation is often recorded as the highest vertebral level that patients can reach behind their back. In the revision shoulder arthroplasty series published by Neer and Kirby, significant sources of failure included scarred subscapularis, rotator cuff scarring, deltoid scarring, and detachment of the deltoid. Of particular importance is assessment of rotator cuff integrity in light of the findings of Franklin et al. regarding the association between rotator cuff tears and increased loosening scores of the glenoid component.
Finally, a focused neurologic exam can identify patients who have neuromuscular deficits that will threaten the functionality of their shoulder should they undergo revision surgery.
As a supplement to the information gleaned for the history and physical examination, certain diagnostic laboratory tests are essential to the evaluation of a patient with shoulder pain after total shoulder arthroplasty. First and foremost in the diagnostic workup is the determination of whether a deep infection of the shoulder is present because the management and surgical intervention is quite different from that of the aseptic painful shoulder. Basic tests that are essential to the workup include white blood cell count (WBC), erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) levels. In their published series of six total shoulder replacements with a deep infection involving the prosthesis, Wirth and Rockwood noted the average white blood cell count to be 12.0 × 10 3 and the ESR to average 75 mm/hr. Sperling et al. in their series of 32 patients with infected shoulder arthroplasties, found an average preoperative WBC of 7.4 × 10 3 and preoperative ESR of 47 mm/hr (with 58% of values falling into the elevated range) at an average of 3.5 years postoperatively. Similarly, Coste et al. found an average WBC of 7.9 × 10 3 and an ESR of 55 mm/hr in their series of 49 patients. Finally, from a population of patients with positive intraoperative cultures, Topolski et al. found a similar WBC count but an average ESR of only 12.4 mm/hr. The discrepancy between these sets of laboratory values and the normal values found on many studies highlights the difficulty in discerning the presence of deep infection. Little has been published on CRP values in the setting of deep infection after total shoulder arthroplasty and available published data are discrepant, casting doubt on the similarity of the laboratory assays used.
Joint aspiration when a deep infection is suspected is recommended and often performed even in the absence of symptoms and laboratory values suspicious of infection. Topolski et al. found only one positive culture from preoperative aspiration of 11 patients (9%) who ultimately had positive intraoperative cultures at the time of revision surgery. Coste et al. found four of eight (50%) preoperative aspirations to be positive while Sperling et al. reported that 14 of 18 (78%) aspirate cultures yielded a causative organism. Although there is a large discrepancy in the yield of preoperative aspiration based on these series, preoperative aspiration is recommended because it can direct antibiotic therapy and confirms suspicion about the method of shoulder failure in the event of positive cultures.
Serial imaging using plain films remains the gold standard in diagnosing component loosening or migration ( Fig. 24-3 ). In addition to evaluation of possible looseness of the glenoid component, serial radiographs can be used to gather information about other prior shoulder surgeries, available bone stock, method of component fixation, and presence of concomitant shoulder pathology. Radiographic views should include anteroposterior (AP), scapular lateral, and axillary views of the shoulder to evaluate for gross tilt, shift, or medialization of the glenoid component as well as radiolucent lines as described earlier. Axillary views may provide superior ability to diagnose loosening due to less bony overlap of the glenoid in the line of the beam ( Fig. 24-4 ). Despite adequate radiographs, this method of evaluating loose prostheses may have its limitations ; Boyd et al. described 12 loose glenoid components confirmed at surgery, of which only 4 were suspected to be loose based on preoperative radiographs demonstrating radiolucent lines wider than 2 mm. In the senior author’s experience, an impressive number of symptomatic shoulders have demonstrated gross glenoid motion on arthroscopic examination despite normal preoperative imaging studies.