Data on the Patients
Between 1983 and 1992, we performed a revision total shoulder arthroplasty for the treatment of glenoid arthrosis associated with moderate or severe pain in twenty-two shoulders that had a prosthetic replacement of the humeral head. The mean time from the hemiarthroplasty to the total shoulder replacement was 4.4 years (range, 0.8 to 12.7 years). Of the twenty-two shoulders, eighteen for which complete preoperative and operative records were available and that had been followed for at least two years (mean, 5.5 years; range, 2.3 to 10.0 years) were included in the study (Table II). One shoulder had been followed for between two and three years; one, between three and four years; four, between four and five years; and twelve, for more than five years. Four patients were excluded because they had been followed for less than two years: three of them had died of unrelated causes, and one had been lost to follow-up.
Five of the eighteen hemiarthroplasties were performed at another institution, and thirteen were performed at our institution. A total of 1028 primary hemiarthroplasties were performed at our institution during this same time-period.
The present study included ten women and seven men who ranged in age from twenty-seven to eighty years (mean, fifty-two years). Eleven of the eighteen shoulders were on the dominant side. The indications for the hemiarthroplasty were trauma (ten shoulders), osteoarthrosis (four), rheumatoid arthritis (two), and osteonecrosis secondary to the use of steroids (two). Of the ten patients who had a hemiarthroplasty because of the sequelae of trauma, four had traumatic arthrosis secondary to recurrent dislocation; two, failure of internal fixation of a fracture of the proximal part of the humerus with subsequent avascular necrosis; two, malunion of a fracture of the proximal part of the humerus; one, non-union of a fracture of the proximal part of the humerus; and one, an acute fracture of the proximal part of the humerus.
Seven shoulders had had a previous procedure before the hemiarthroplasty. Four shoulders had had one previous procedure; one, two previous procedures; and two, three previous procedures. The previous operations included open reduction and internal fixation (three shoulders), stabilization (two), exploration (one), and acromioplasty (one).
The patients were identified with the use of a computerized database that contained the files of all patients who had had a joint arthroplasty at our institution since 1969. They were asked to return for a radiographic evaluation, physical examination, and interview with the senior one of us (R. H. C.) at regular follow-up intervals. At the time of the most recent follow-up, five patients were evaluated by the senior one of us and twelve patients who were unable to return to the Mayo Clinic completed a questionnaire for the evaluation of function and satisfaction. In addition, those twelve patients were requested to have a local orthopaedic surgeon send us the results of a clinical examination and recent radiographs.
Clinical Evaluation
At our institution, the results of clinical assessments of all patients who have had an operation on the shoulder are recorded with use of a standard shoulder-analysis form. Pain in the shoulder was graded, according to the scale of Neer et al.28, as 1 point (no pain), 2 points (slight pain), 3 points (pain after unusual activity), 4 points (moderate pain), and 5 points (severe pain). The satisfaction of the patients was assessed by asking them how they felt after the revision operation compared with how they had felt before the revision. One point was given if they felt much better; 2 points, if they felt better; 3 points, if they felt the same; or 4 points, if they felt worse. Active abduction and external rotation were measured in degrees. Internal rotation was measured as the most cephalad posterior vertebral segment that could be reached by the thumb.
We graded the results with use of a modification of the rating system of Neer et al.9,28. The result was considered to be excellent if the patient had no or slight pain, had external rotation to at least 45 degrees, had active abduction to at least 140 degrees, and was satisfied with the result. The result was satisfactory if the patient had no or slight pain or moderate pain only with vigorous activity, had external rotation to at least 20 degrees, had active abduction to at least 90 degrees, and was satisfied with the outcome. If any of the criteria for a satisfactory result were not met, or if the patient had had an additional operative procedure, the result was considered unsatisfactory.
Radiographic Evaluation
An axillary radiograph of the shoulder, a 40-degree posterior oblique radiograph with external rotation of the humerus, and a 40-degree posterior oblique radiograph with internal rotation of the humerus were made. Glenohumeral subluxation was evaluated with regard to the direction and the amount of translation of the center of the prosthetic head relative to the center of the glenoid component. Subluxation was recorded as none, mild if there was less than 25 per cent translation, moderate if there was 25 to 50 per cent translation, and severe if there was more than 50 per cent translation. Before the revision, seven shoulders had posterior subluxation (mild in three and moderate in four), six shoulders had superior subluxation (mild in four and moderate in two), and one shoulder had mild anterior subluxation.
Before the revision, all patients had loss of glenoid cartilage. Fourteen shoulders had erosion of the glenoid, which was graded as mild in nine and moderate in five.
Periprosthetic radiolucency was considered grade 0 if there was no radiolucent line, grade 1 if the line was one millimeter wide and incomplete, grade 2 if the line was one millimeter wide and complete, grade 3 if the line was 1.5 millimeters wide and incomplete, grade 4 if the line was 1.5 millimeters wide and complete, and grade 5 if the line was two millimeters wide and complete. A shift in the position of the component was recorded as either present or absent. Before the revision, a radiolucent line was present around three press-fit humeral components. The line was grade 3 in two of these shoulders and grade 5 in one, and there was a shift in the position of all three components. Postoperative radiographs also were examined for periprosthetic radiolucency, a shift in the position of the component, and glenohumeral subluxation.
Operative Technique
The total shoulder arthroplasty was done through an anterior approach with use of a deltopectoral exposure. Great care was taken to avoid injury to the anterior aspect of the deltoid muscle, the axillary nerve on the undersurface of the muscle, and the rotator cuff tendons. The subdeltoid-subacromial space as well as the space between the conjoined tendons and the subscapularis were freed of scar tissue. In a shoulder with 30 degrees or more of external rotation, the subscapularis tendon was divided over the humeral attachment of the anterior aspect of the capsule. In a shoulder with less than 30 degrees of external rotation, the subscapularis was incised from the humerus. Lengthening of the subscapularis in a z-fashion was necessary in two patients. The inferior aspect of the capsule was incised along the humeral neck with electrocautery, with careful protection of the axillary nerve. The humeral component was then removed from seventeen shoulders because it had not been cemented or it had been cemented in a less-than-optimum position. In the remaining shoulder, the component had been cemented in an excellent position, and it was possible to retract the component posteriorly to expose the glenoid. The anterosuperior aspect of the capsule was incised near the glenoid to increase excursion of the subscapularis muscle and tendon. The glenoid was prepared, the glenoid component was fixed in place, the humeral component was replaced or reduced, and the subscapularis was repaired (Figs. 1-A, 1-B, 1-C and 1-D). A medium-sized tear of the rotator cuff involving the posterior portion of the supraspinatus and the anterior portion of the infraspinatus was repaired in one patient at the time of the revision.
The glenoid component was inserted with cement in ten shoulders, and a porous-coated component was used in eight. Bone-grafting of the glenoid was performed in two shoulders. In one shoulder, a large cystic cavity was filled with cancellous allograft and the glenoid component was inserted with cement. In the other shoulder, a five-millimeter-wide erosion that involved one-half the width of the posterior part of the glenoid was filled with bone graft obtained from the humeral osteotomy and a porous-coated glenoid component was implanted.
Twelve shoulders had had a press-fit humeral-head replacement for the hemiarthroplasty; seven were revised to a porous-coated humeral component, three were revised to a humeral component inserted with cement, and two were revised to a press-fit humeral component. Three shoulders had had a porous-coated replacement, and all three were revised to a porous-coated humeral component. In two shoulders, the humeral component was inserted with cement for both the hemiarthroplasty and the revision. Therefore, after the revision, ten shoulders had a porous-coated humeral component, five had a cemented humeral component, and two had a press-fit humeral component. One cemented humeral component was not revised.
Bone-grafting of the humerus was performed in five shoulders. Allograft bone was placed at the junction of the proximal part of the humerus and the prosthesis (two shoulders), in a defect created during removal of the component (one shoulder), and in the medullary canal to improve bone stock (one). In one shoulder, allograft bone was placed at the sites of non-unions of the humeral tuberosities and at the junction of the proximal part of the humerus and the prosthesis.
At the time of the revision, it was apparent that malposition of the humeral component was a cause of rapid wear of the glenoid in two patients. In one of these patients, the prosthetic humeral head was in 65 degrees of retroversion with associated erosion of the posterior aspect of the glenoid. This shoulder needed revision to a total shoulder arthroplasty eleven months after the hemiarthroplasty. In the second patient, the humeral component was noted to be in an inferior position with associated erosion of the inferior aspect of the glenoid. At the time of the revision, two years after the hemiarthroplasty, the humeral component was inserted with cement 1.5 centimeters superior to its previous position, with bone graft placed in the gap.
All components had a forty-four-millimeter-diameter articular surface. Two types of components were used during this time-period. Cofield humeral and glenoid components (Smith and Nephew Richards, Memphis, Tennessee) were implanted in nine shoulders; Neer humeral and glenoid components (Kirschner Medical, Fairlawn, New Jersey), in six; a Cofield humeral component and a Neer glenoid component, in two; and a Neer humeral component and a Cofield glenoid component, in one. No custom or modular prosthesis was used.
Postoperatively, a shoulder immobilizer was used at night for one month and during the day for the first week. Thereafter, a sling was used intermittently during the day. Rehabilitation was started on the second day with passive range-of-motion exercises. At four to five weeks, active-assisted range-of-motion as well as isometric strengthening exercises were begun. At two to three months, stretching and strengthening exercises with use of an elastic strap were added.
Statistical Methods
For ordinal and continuous measurements, a Wilcoxon signed-rank test was performed to detect differences between preoperative and postoperative values. The Wilcoxon rank-sum test was used to determine whether measurements varied with respect to gender, diagnosis, periprosthetic radiolucency, or subluxation. For categorical variables, a Fisher exact test was used to test for significant relationships between variables.
Complications and Reoperations
Complications occurred in three patients, necessitating a reoperation in two. One patient sustained a traction-induced neuropathy of the brachial plexus involving the upper part of the trunk at the time of the revision. At the most recent follow-up examination, the patient had no pain and had had a full return of muscle function. One patient who had rheumatoid arthritis had removal of the components because of infection nineteen months after the total shoulder arthroplasty. At the most recent follow-up evaluation, six years after removal of the components, the patient had not had an additional operation on the affected shoulder. One patient who had lupus erythematosus had revision of the humeral component with a synovectomy of the shoulder fifty-three months after the total shoulder arthroplasty because of particulate-induced synovitis associated with multidirectional instability of the shoulder.
Pain
Revision to a total shoulder arthroplasty was significantly associated with pain relief. The mean score for pain decreased from 4.3 points before the revision to 2.2 points after it (p = 0.0001). Pain relief did not vary by gender, diagnosis, subluxation, or the presence of a periprosthetic radiolucent line. All patients had had moderate or severe pain before the revision (Table III). No patient had severe pain and three patients had moderate pain after the revision. One patient who had moderate pain had severe superior subluxation. At the time of the revision, she had a medium-sized tear of the rotator cuff. It was repaired, but it either stretched during healing or it did not heal. In another patient, moderate pain was associated with instability. This patient had sustained a neuropathy of the axillary nerve at the time of the hemiarthroplasty, which had been done at another institution. She also had nerve-grafting at the time of the revision total shoulder arthroplasty and had partial improvement in the strength of the deltoid. In the third patient, the moderate pain was associated with synovitis and moderate multidirectional instability. As mentioned previously, the patient had a second revision operation.
Range of Motion
Active abduction and external rotation increased significantly in most patients (Table III). The mean active abduction improved from 94 degrees before the revision to 124 degrees (range, 30 to 180 degrees) after it (p = 0.01). The mean external rotation increased from 32 degrees before the revision to 58 degrees (range, 0 to 90 degrees) after it (p = 0.007). The mean internal rotation (measured as the most cephalad level that the thumb could reach on the spine) increased from the third lumbar vertebra before the revision to the first lumbar vertebra (range, the iliac crest to the sixth thoracic vertebra) after it, but this change was not found to be significant (p = 0.35), with the numbers available.
Radiographic Findings
At the time of the most recent radiographic examination, six shoulders had mild superior subluxation and one had severe superior subluxation (Table III). The shoulder that had severe superior subluxation had had repair of a moderate-sized tear of the rotator cuff at the time of the revision. With the numbers available, we could not detect a significant relationship between the presence of subluxation and the diagnosis.
A radiolucent line was present around six of the ten glenoid components that had been inserted with cement. It was grade 2 in three shoulders, grade 4 in two, and grade 5 in one. No radiolucent line was present around any of the eight porous-coated glenoid components. With the numbers available, we could not detect a significant relationship between the presence of subluxation and a radiolucent line around the glenoid component (p = 0.627).
A radiolucent line was present around four of the ten porous-coated humeral components. It was grade 1 in three shoulders and grade 5, with a shift in the position of the component, in one. No radiolucent line was present around any of the six humeral components that had been inserted with cement or around either of the two press-fit humeral components.
Satisfaction of the Patient
According to the patients' assessments, twelve shoulders were much better, two shoulders were better, and four shoulders were the same or worse than before the revision. Of the four shoulders that did not benefit from the procedure, two had an additional operation and two had active abduction of less than 90 degrees.
Rating of the Results
Eight shoulders had an excellent result; three, a satisfactory result; and seven, an unsatisfactory result. Two shoulders were considered to have an unsatisfactory result because an additional operation had been done. The reasons for the other five unsatisfactory results were limitation of active abduction (three shoulders), a lack of external rotation (one), and limitation of both active abduction and external rotation (one).
The successful management of a patient who has pain after replacement of the humeral head presents the orthopaedic surgeon with two major challenges. First, the cause of the pain in the shoulder must be determined. A systematic assessment of the clinical and radiographic factors is necessary to evaluate fully the potential causes of pain after hemiarthroplasty. The differential diagnosis of pain after replacement of the humeral head includes infection, glenoid arthrosis, instability, impingement, a tear of the rotator cuff, nerve injury, fracture, and malposition or loosening of the component. A thorough examination to evaluate cutaneous changes, deformity, neurovascular status, stability, motion, and strength is essential. Radiographs of the shoulder must be of high quality. In addition to 40-degree posterior oblique radiographs with the shoulder in internal and external rotation and an axillary radiograph, we use spot radiographs with fluoroscopic positioning to assess the implant-bone interface. If a tear of the rotator cuff or glenohumeral synovitis is suspected, an arthrogram may help to confirm the diagnosis. The white blood-cell count with differential and the erythrocyte sedimentation rate should be measured to evaluate for a possible infection. In addition, a bone scan and an indium-labeled white blood-cell scan may allow differentiation between infection and aseptic loosening. If infection is strongly suspected, the shoulder should be aspirated.
Second, the criteria for the selection of patients for a revision procedure must be determined. A revision operation after a previous arthroplasty is a technically challenging type of shoulder arthroplasty because of the presence of scar tissue, muscle-tendon weakness, and bone loss27. The condition of the soft tissues and scarring are important prognostic factors with regard to the range of motion after the operation. Limited motion before the revision is associated with limited motion after it10. The compliance and motivation of the patient are also crucial factors in the decision about whether to perform a revision because a long period of postoperative rehabilitation is necessary.
At the time of the revision, bone-grafting may be necessary to fill defects in the glenoid bone created by erosion. In addition, bone-grafting may be necessary to restore humeral height, to fill defects created during removal of the component, or to restore bone stock. We now insert most all-polyethylene glenoid components with contemporary cementing techniques, which include the use of precise instrumentation, the availability of multiple sizes of components, pulsatile lavage, vacuum-mixed cement, pressurization of the cement, and firm impaction of the component. A porous-coated glenoid component is used in patients in whom bone loss precludes secure fixation with a keel-type implant. A porous-coated humeral component is used unless there are osseous deficiencies that preclude secure fixation without bone cement.
It was clear that certain technical factors were associated with failure of the initial prosthetic replacement of the humeral head. Malposition of the component with regard to proper retroversion and height resulted in irregular wear of the glenoid in two shoulders. As many well fixed prostheses had to be removed and revised, we believe that the use of a modular prosthesis might facilitate revision in some shoulders.
Although the present study included patients who had several different diagnoses and natural histories, all had reached a common end point: painful glenoid arthrosis after a hemiarthroplasty. Two patients had an associated abnormality (a tear of the rotator cuff or an injury of the axillary nerve); however, the clinical and radiographic impression was that glenoid arthrosis was the primary cause of pain, and it was the indication for the revision operation, in all patients. The primary goal of the revision arthroplasty for all patients in our study was relief from moderate or severe pain, and this goal was achieved in fifteen of the eighteen shoulders. Most patients also had significant improvement in the range of motion after the revision to total shoulder arthroplasty, with a mean increase of 30 degrees of active abduction (p = 0.01) and 26 degrees of external rotation (p = 0.007). Complications developed in three of the eighteen shoulders, and two of those had an additional operation.
The development of glenoid arthrosis after replacement of the humeral head can cause severe pain and limitation of motion. The data from the present study indicate that most patients have marked pain relief and improvement in motion after revision to a total shoulder replacement. However, according to a rating system, seven of the seventeen patients who had this procedure had an unsatisfactory result that was due to a limited range of motion or the need for a subsequent operation. Therefore, long-term studies are necessary to evaluate the durability of total shoulder replacement in this group of patients.