Between 1986 and 1988, three women and one man, forty-two, forty-eight, fifty-nine, and seventy-six years old, were managed operatively for a chronic posterior dislocation of the shoulder associated with an anteromedial defect of the humeral head that involved at least 40 per cent of the articular surface. The duration of the dislocation had ranged from one to twelve months. The cause of the dislocation was a grand mal seizure in three patients, one of whom was a known alcohol abuser, and a bicycle accident in one patient. The diagnosis had initially been missed in each patient. One patient had been managed with a physical therapy program and manipulation while under anesthesia.
At the time of presentation, two patients had severe pain, defined as pain every night and regular use of analgesics, and two had moderate pain, defined as occasional nocturnal pain and intermittent use of mild analgesics. Active flexion of the shoulder averaged 35 degrees (30 degrees in three patients and 50 degrees in one), with an average rotation contracture of 40 degrees (15, 40, 45, and 60 degrees) (Table I).
Standard radiographs revealed a locked posterior dislocation in all patients. Three patients had no degenerative changes of the glenohumeral joint, and one (Case 2) had moderate-to-severe osteoarthrotic changes (Figs. 1-A, 1-B, 1-C, 1-D, 1-E, 1-F through 1-G). The impaction fracture of the humeral head was measured on computed tomographic scans made in the horizontal plane at the level of the greatest diameter of the head and was expressed as a percentage of the projected total articular surface. The defect involved between 40 and 55 per cent of the total articular surface in each patient. The architecture of the remainder of the humerus was normal on the computed tomographic scans of two patients (Cases 1 and 4); in the other two patients, who had a more long-standing dislocation, the dislocated part of the humeral head was osteoporotic.
Operative Technique
A deltopectoral approach was used. The conjoint tendon was partially released in order to reach the desired line of incision in the subscapularis tendon in two patients. The axillary nerve was identified and was protected throughout mobilization of the subscapularis and the inferior aspect of the capsule. After a vertical capsulotomy, in which an attempt was made to leave the superior glenohumeral ligament and the coracohumeral ligament intact, adhesions were released between the capsulolabral complex and the dislocated head. There were no relevant fractures involving the posterior aspect of the glenoid rim.
The posterior aspect of the capsule was stretched into a redundant pouch, but it was not detached from the posterior aspect of the scapular neck and it was not operatively addressed. No lesions involving the infraspinatus or supraspinatus, and no relevant injuries of the anterior capsuloligamentous structures, were identified17. Slight internal rotation of the relocated humerus resulted in recurrence of the posterior dislocation in all patients.
The defect of the head was tentatively filled with allograft. This prevented posterior redislocation on internal rotation in all patients and, consequently, the decision was made to reconstruct the humeral head with allogeneic bone. A cryopreserved femoral head was used in three patients, and a fresh autoclaved femoral head was employed in one. The allograft was contoured to fit the segmental defect and to restore the original sphericity of the humeral head (Figs. 1-F, 1-G, and 1-H). In three patients, the graft was fixed with lagged, countersunk 3.5-millimeter cancellous-bone screws (Synthes, Waldenburg, Switzerland). The fourth graft was stable after being press-fit into the defect, and no internal fixation was needed.
The anterior aspect of the capsule was not repaired. The superior glenohumeral ligament, which had not been divided, remained intact. The subscapularis was mobilized from its origin on the anterior surface of the scapula, released from all of its adhesions, and repaired from end to end without lengthening or shortening.
Postoperative Management and Evaluation
The upper limb was kept at the side, with rotation at 0 degrees, in a prefabricated splint for six weeks. The limb was held in the splint by Velcro straps, and the splint was removed for passive external-rotation exercises during this period. Muscle-resistance exercises were started at six weeks. No patient had a problem with wound-healing, neurological injury, or attenuation of the function of the subscapularis muscle.
At the latest follow-up examination, the patients were assessed both clinically and radiographically. Satisfaction, pain, and subjective functional performance were assessed in an interview. All patients were evaluated with use of the functional shoulder-scoring system described by Constant and Murley2. According to this 100-point system, 35 points is assigned for the subjective evaluation of the shoulder and 65 points, for the objective functional evaluation. In the subjective evaluation, a maximum of 15 points is assigned for absence of pain (with 0 points indicating severe pain and 15 points, no pain); a maximum of 4 points each, for the ability to perform activities of daily living and recreational activities; and 2 points, for undisturbed sleep. A maximum of 10 points is assigned for the ability to use the upper limb at graded levels (2 points for use at the waist and 10 points for overhead use). In the objective assessment, a maximum of 40 points is assigned for an active, pain-free, functionally useful range of motion and a maximum of 25 points, for strength. The patients were examined according to the original protocol established by Constant and Murley2 for all parameters except strength, which was evaluated with the Isobex dynamometer (Cursor AG, Bern, Switzerland)6. The patients were required to sustain actively a position of 90 degrees of abduction in the scapular plane, with the elbow extended and the forearm pronated, against a resistance measured in kilograms, derived from the dynamometer, that was applied to the wrist. The average of three readings of an effort of five seconds has been established as the most reproducible measurement6 and was therefore used to calculate the component for strength in the scoring system of Constant and Murley. The scores were expressed as absolute points and as percentage ratios of the values for normal shoulders of age and sex-matched subjects1.
Both plain radiographs and computed tomographic scans were evaluated at the latest follow-up examination. Specific attention was paid to the incorporation and resorption of the allograft, the congruency of the joint, the sphericity of the humeral head and the allograft, and degenerative changes (Figs. 1-I, 1-J, and 1-K). Osteoarthrosis was considered mild if there were osteophytes of less than three millimeters in the humeral head or glenoid rim; moderate if there were osteophytes of three to five millimeters and mild irregularity in the joint line, subchondral sclerosis, or flattening of the head; and severe if there were more extensive changes.
The four patients were evaluated an average of sixty-eight months (range, sixty to seventy-six months) after the operative procedure. Two patients (Cases 3 and 4) reported no pain. One patient (Case 2), who had had advanced osteoarthrosis preoperatively, reported mild pain, as did the fourth patient (Case 1), who had had spontaneous avascular necrosis of the humeral head. With use of the component for pain from the scoring system of Constant and Murley2, three patients had gained 10 points and one, 15 points (Table II). No patient had symptoms consistent with glenohumeral instability. Two patients (Cases 3 and 4) considered the result to be excellent; one (Case 2), good; and one (Case 1), fair.
Flexion of the shoulder was found to have been restored to that of the contralateral shoulder in two patients (Table II). In one patient (Case 1), complete return of function was noted early on; however, after the spontaneous onset of avascular necrosis of the remaining portion of the humeral head in the seventh postoperative year, this patient lost 60 degrees of flexion. The osteoarthrotic shoulder of the fourth patient (Case 2) gained 85 degrees but still had 50 degrees less flexion than the contralateral shoulder. Functional external rotation was restored in all patients. As assessed with the scoring system of Constant and Murley2, function of the shoulder averaged 82 per cent (42, 73, 105, and 109 per cent) that of an age and sex-matched normal population1 at the latest follow-up examination. This value had been 94 per cent (88, 73, 105, and 109 per cent) before the spontaneous collapse of the humeral head at seventy-six months postoperatively in one patient (Case 1).
Radiographs made at the latest follow-up examination revealed no failures of fixation of the allograft, and the humeral head was maintained in an anatomical position in each patient. The humeral head in one patient (Case 1) had undergone spontaneous avascular necrosis and collapse. In the patient who had osteoarthrosis and osteopenia of the dislocated humeral head preoperatively (Case 2), flattening of the contour of the original head with progression of the osteoarthrosis was seen (Figs. 1-D, 1-E, 1-F, through 1-G, 1-I, 1-J, and 1-K). However, the contour and volume of the graft were maintained. In the other two patients (Cases 3 and 4), there was slight flattening of the graft without evidence of avascular necrosis or osteoarthrosis of the remaining portion of the head. These radiographic changes were not associated with any deterioration in the patients' subjective or objective functional status.
Computed tomographic evaluation revealed incorporation of the allograft by three months and consolidation by one year in all four patients.
Chronic posterior dislocations with a large anteromedial defect of the humeral head are, fortunately, rare. Only a few series, each with a limited number of patients, comparable with the number in the current study, have been reported8,16,19.
Hughes and Neer modified McLaughlin's subscapularis transfer and transferred the lesser tuberosity with its attached subscapularis tendon into defects ranging in size between 20 and 40 per cent of the articular surface. For larger defects such as those discussed here, they recommended arthroplasty. Hawkins et al. reported on a series of twenty-five shoulders that were treated for a locked posterior dislocation. Nine shoulders had a subscapularis transfer: four were initially treated by Hawkins et al. and had a satisfactory result, and five had been referred to them for revision because of an unsatisfactory result. The sixteen remaining shoulders had a defect that necessitated an arthroplasty. The over-all results were fair, with an average of 140 degrees of flexion and 30 degrees of external rotation for the shoulders that had had a hemiarthroplasty and 152 degrees of flexion and 40 degrees of external rotation for those that had had a total arthroplasty. Hawkins et al. concluded that transfer of the lesser tuberosity could be recommended if the shoulder had been dislocated for no longer than six months and if the humerus had a defect of between 20 and 45 per cent of the articular surface, whereas replacement arthroplasty seemed preferable for older dislocations or dislocations associated with more important destruction of the head. Walch et al. reported three excellent results, one good result, five fair results, and one poor result after subscapularis transfer in ten patients who had destruction of less than 50 per cent of the head. For defects greater than 50 per cent, those authors attempted rotational osteotomy and reported unsatisfactory results.
Fresh osteochondral allografts have been used to restore joint surfaces, with satisfactory long-term results in weight-bearing joints7,11,18,20. Congruity, mechanical stability, and incorporation of all grafts were achieved where there had been a well vascularized bed, and no allograft failed10,13,14 because of early or late infection, fracture, non-union at the graft-host junction, or instability of the joint. The more that an allograft is loaded, the more probable it is that adaptive structural changes will occur. We therefore do not use this procedure if the humeral head has been dislocated for a sufficient period of time to allow relevant osteopenia with softening of the humeral head to develop. In this setting, the relocated head is at risk for collapse, which would consequently expose the adjacent graft to overload and failure. Progressive collapse of the original head due to avascular necrosis was observed in one of our patients. Although avascular necrosis of the humeral head after operative treatment of locked posterior dislocation of the shoulder has been reported, especially after subcapital humeral rotational osteotomy19, the absence of such changes for six years in this patient refutes a causal relationship between the reconstruction with the allograft and the collapse of the humeral head. Alternatively, the severe, chronic abuse of alcohol by this patient may have been related to the development of the avascular necrosis. This patient had not elected to have an additional operation at the time of writing, but if she has one later it will be a revision to a hemiarthroplasty. It is noteworthy that this procedure could be carried out in a shoulder with normal anatomy of the shaft and tuberosities, an intact rotator cuff, and a stable joint, so that prosthetic reconstruction should be less complicated than after procedures that alter the normal anatomy, such as transfer of the lesser tuberosity or rotational osteotomy.
In summary, after a minimum duration of follow-up of five years, the clinical results of reconstruction of the humeral head with allogeneic bone appeared comparable with those of currently recommended techniques of treatment8,19. For patients who have good bone quality of the residual head and an absence of relevant osteoarthrosis, reconstruction with an allograft represents a viable alternative to subscapularis transfer for smaller defects of the humeral head as well as an alternative to hemiarthroplasty or total joint arthroplasty for carefully selected larger defects.
NOTE: The authors thank Dr. P. M. Ballmer, Thun, Switzerland, and Dr. B. Isler, Winterthur, Switzerland, for their contribution in the beginning of this study.