Between December 1998 and September 2002, sixty procedures with the
Reverse Shoulder Prosthesis were performed by one surgeon in sixty consecutive
patients for the treatment of a rotator cuff deficiency associated with
glenohumeral arthritis. There were forty-one women and nineteen men. The
average age at the time of surgery was seventy-one years (range, thirty-four
to eighty-six years). The average duration of follow-up was thirty-three
months (range, twenty-four to sixty-eight months). All patients in this study
had glenohumeral arthritis and a deficient rotator cuff, but we further
classified them according to the method used by
Walch25, who
recently described patients he had managed with a reversed ball-and-socket
prosthesis. Eleven patients had primary cuff tear arthropathy with collapse of
the humeral head as described by Neer et
al.1. The collapse
of the humeral head was determined on examination of the radiographs.
Seventeen patients had primary cuff tear arthropathy without collapse of the
humeral head, twenty-three patients had failed rotator cuff surgery, seven
patients had a massive rotator cuff tear with chronic pseudoparalysis of the
shoulder (pseudoparalysis with or without anterior-superior dynamic
instability of the humeral head for more than six months), one patient had
post-traumatic arthritis, and one had rheumatoid arthritis. Five patients were
less than sixty years old, and one of those was less than forty years old.
This thirty-four-year-old woman had cuff tear arthropathy with collapse of the
humeral head (Fig. 1). She also
had systemic lupus erythematosus and severe osteonecrosis. Thirty-five
patients had no previous shoulder surgery, while twenty-three had had either
an open or an arthroscopic rotator cuff repair (six of them had multiple
repairs of the rotator cuff), one had had a subacromial decompression, and one
had had a biceps tendon repair.
Preoperatively, all patients were found to have a rotator cuff deficiency
on physical examination and on computerized tomography scans. Most of them had
had a magnetic resonance imaging scan, which also demonstrated the deficiency.
All patients had a complete tear of the supraspinatus with variable
involvement of the subscapularis and the infraspinatus. The computerized
tomography scans were analyzed to verify that there was at least 25 mm of bone
from the glenoid face to the most medial cortex of the scapula to provide
proper seating of the implant. In patients who had insufficient bone stock
because of substantial medial erosion of the glenoid, the Reverse Shoulder
Prosthesis was not recommended. Preoperative radiographs revealed superior
migration of the humeral head, along with erosion of the humeral head,
glenoid, and acromion in all patients. All patients who received a Reverse
Shoulder Prosthesis previously underwent nonoperative treatment, such as
physical therapy and cortisone injections that were unsuccessful, as well as
arthroscopic or open rotator cuff repair. In addition, it was thought that no
other treatment except arthroplasty with the Reverse Shoulder Prosthesis would
provide a reliable method of improving function and diminishing pain.
Exclusion criteria were active infection, axillary nerve palsy, a
nonfunctioning deltoid muscle, insufficient bone to seat the implant
components, or a very high level of physical activity (e.g., participation in
competitive sports or heavy physical labor).
Operative Technique
The senior author (M.F.) performed all operations. Patients were positioned
in the upright beach-chair position with the head firmly secured and the arm
draped free. The operative arm was positioned sufficiently off the side of the
table to allow for unobstructed movement of the shoulder in adduction and
hyperextension. All patients were given general anesthesia in addition to a
scalene block. An extended deltopectoral approach was used, and approximately
two-thirds of the pectoralis major tendon was released. The subdeltoid,
subacromial, and subcoracoid spaces were released. If the subscapularis tendon
was intact, it was released just medial to the long head of the biceps,
allowing atraumatic dislocation of the humeral head with gentle external
rotation and extension of the arm. Osteophytes were resected, a neck cut was
made in 30° of retroversion, and the canal was reamed and broached.
The broach was tapped until it was distal to the neck cut, and this was
followed by reaming of the proximal humeral metaphysis. Next, the glenoid was
exposed with use of thorough capsular releases. A centering hole was drilled,
and the glenoid was prepared with a convex reamer. A fixed-angle glenoid
baseplate was then screwed into place, ensuring at least 60 in-lb (6.8 N-m) of
torque. Additional 3.5-mm peripheral fixation screws were then attached to the
glenoid baseplate. A 29-mm or a 25-mm offset glenosphere was chosen, depending
on the degree of soft-tissue contracture, and was fit onto the baseplate by
means of a Morse taper (Figs. 2-A and
2-B). After reduction with the humeral broach and a trial
polyethylene component, the appropriate size of humeral implant that would
allow a 2-mm circumferential cement interface around the component was
selected and routinely cemented in place. The reduction was checked for
stability, especially in abduction, extension, and internal rotation, and
achievement of full passive elevation was confirmed. Finally, the
subscapularis was repaired through drill-holes followed by routine closure
with use of number-2 braided polyester sutures.
Postoperative Rehabilitation
A shoulder immobilizer was worn for four weeks, and passive range-of-motion
exercises were performed. The passive range-of-motion exercises, performed
with the patient supine, were started the day after surgery with a physical
therapist. The standard limits of 90° of elevation and 0° of external
rotation were used. After the first four to six weeks, the patient was managed
with a sling and active-assisted activities were initiated. Active range of
motion was not started until eight to ten weeks postoperatively. Resistive
exercises were delayed until the subscapularis tendon was healed, usually at
twelve weeks.
Preoperative and Postoperative Clinical Assessment
Several validated scoring instruments were used to assess the patients
clinically26.
Patients completed forms for the American Shoulder and Elbow Surgeons
assessment for pain and function, a visual analog scale for pain and function,
and a rating of overall satisfaction. Patients were asked to indicate their
personal satisfaction with the outcome of the surgery as dissatisfied,
satisfied, good, or excellent. An orthopaedic surgeon who was not involved in
the treatment of any of the patients measured the range of motion
preoperatively and postoperatively using a goniometer while digital clinical
videos were played back on a computer. Preoperative and postoperative flexion
and abduction data were available for all patients. Postoperative external
rotation data were available for all patients, but preoperative values were
available for only sixteen patients.
Radiographic Analysis
An orthopaedic surgeon who was not involved in the treatment of any of the
patients performed the radiographic evaluation. The initial preoperative and
the most recent (at least two-year) postoperative radiographs of the shoulder,
which included anteroposterior, axillary, scapular Y lateral, and
Grashey27
radiographs with the shoulder in both internal and external rotation, were
evaluated. Complete radiographic data were available for all patients.
Preoperative radiographs demonstrated subluxation of the humeral head as
well as erosion of the glenoid and humeral head in all patients. Erosion of
the acromion was present in all but two patients. The subluxation and erosion
were usually directed anteriorly and superiorly. The preoperative
acromiohumeral distance measured from the Grashey radiographs was an average
(and standard deviation) of 2.6 ± 1.68 mm (range, 0 to 6 mm). An
acromial fracture was seen preoperatively in five patients; four were distal
and one was at the acromial base.
Statistical Analysis
Pain and function scores and range of motion were compared preoperatively
and postoperatively by an independent statistician using a difference-of-means
test (StatView; SAS Institute, Cary, North Carolina). The influence of a
previous operation on the results was also analyzed with use of a
difference-of-means test. The subset of thirty-five patients who had no
previous shoulder surgery and the subset of twenty-five patients who had
previous shoulder surgery were compared with use of the Levene test for
equality of variances and a t test for equality of means.
The clinical results including the scores on the American Shoulder
and Elbow Surgeons assessment and the visual analog scale, as well as the
range-of-motion measurements, are summarized in
Table I. Data for external
rotation at 0° of abduction were not complete. Postoperative data on
external rotation were available for all patients, but only sixteen had
preoperative measurements. The mean external rotation for the sixteen patients
for whom we had complete data improved from 12° preoperatively to 41°
postoperatively; the difference was significant (p < 0.0001). For all sixty
patients, the mean postoperative external rotation measured 35.9° (range,
5° to 60°). Forty-one (68%) of the sixty patients rated the outcome as
good or excellent, sixteen (27%) were satisfied, and three (5%) were
dissatisfied.
The patients were also divided into two groups with respect to those who
had had previous shoulder surgery and those who had not. No difference between
the groups was detected in terms of demographic data, preoperative scores,
postoperative visual analog scale scores, or range of motion. However, the
postoperative mean total score and mean pain and function scores according to
the American Shoulder and Elbow Surgeons system were significantly higher for
the group that had not had previous shoulder surgery than for the group that
had had previous shoulder surgery (p = 0.015, 0.047, 0.044, respectively).
There were thirteen complications in ten patients (17%) in this study. One
patient had a scapular fracture at three months and an acromial fracture at
twelve months, neither of which were believed to be related to the surgery.
The scapular fracture occurred during therapy while the patient was undergoing
passive flexion in the supine position. The acromial fracture occurred as the
result of a fall. Both fractures were treated nonoperatively and healed
uneventfully. Another patient had an acromial fracture at thirteen months
postoperatively. It was treated with open reduction and internal fixation and
healed uneventfully. Both of these patients rated the outcome as excellent.
One patient had a hardware failure from a preexisting acromial fracture with
an infection five months postoperatively. The patient underwent a revision
open reduction and internal fixation along with irrigation and
débridement. The patient rated the overall outcome as satisfactory.
There were no nerve or vascular injuries.
Postoperatively, progression of subluxation or erosion was not seen
radiographically. Glenoid radiolucency was seen in three of the sixty
patients, with a radiolucency of >2 mm in a patient with glenoid baseplate
failure (Fig. 3) and a
radiolucency of 1 to 2 mm in two patients. Scapular notching was absent in all
patients. No humeral component showed any evidence of loosening or failure on
postoperative radiographs. However, radiolucent lines that did not progress
were seen around the humeral component and were most often located along the
inferior-lateral margin of the prosthesis. These radiolucencies were 1 to 2 mm
in width, and none was >2 mm.
Revision Operations
Seven patients in whom the device failed required eight revisions at an
average of 21.4 months (range, eleven to thirty-three months) after the index
procedure. In all eight revisions, inspection of the porous surface of the
glenoid baseplate revealed no evidence of osseous ingrowth. Two shoulders were
converted to a hemiarthroplasty because of insufficient bone stock and a deep
infection. One of these patients had a humeral dissociation with glenoid
loosening, and the other patient had glenoid baseplate failure. Both were
treated with six weeks of intravenous antibiotics and implantation of an
antibiotic spacer followed by a hemiarthroplasty. The other five patients had
baseplate failure and were managed with a revision to another Reverse Shoulder
Prosthesis. One of these five patients had a second baseplate failure and had
revision to a third Reverse Shoulder Prosthesis. Of the five patients who had
a revision to another Reverse Shoulder Prosthesis, four rated the outcome as
excellent and one rated it as good at the time of the most recent follow-up.
These five patients were followed for an average of thirty-six months (range,
twenty-eight to forty-seven months) after the initial surgery and 13.6 months
(range, eight to twenty-one months) after the revision surgery. The two
patients whose shoulders were converted to a hemiarthroplasty rated the
outcome as good and satisfactory at the most recent follow-up examination.
The treatment of glenohumeral arthritis in the setting of advanced
rotator cuff disease has long been a perplexing problem for the
upper-extremity surgeon. Early attempts to manage this problem with a
constrained prosthesis proved
disappointing17,28,29.
Hemiarthroplasty, the "nonconstrained" option, has long been
the standard of care for cuff tear arthropathy. However, careful examination
of the literature revealed that the results have not been uniform
(Table II). Even though our
patients had a clinical complex of glenohumeral arthritis and massive rotator
cuff tears not classically described as rotator cuff arthropathy, surgical
management of these patients would historically entail hemiarthroplasty as
seen in the series described by Williams and
Rockwood30. Other
authors have shown less successful outcomes, and relief of pain has been
inconsistent10,11,31-33.
Early reversed shoulder arthroplasties, in effect a
"semiconstrained" design, were plagued with difficulties related
to glenoid failure and soon fell out of
favor17. Our study
demonstrates that, in the short term at least, the Reverse Shoulder Prosthesis
may be a viable alternative for the treatment of glenohumeral arthritis and
rotator cuff deficiency.
The authors of several studies conducted in Europe have reported promising
results in the short and medium term with use of a reversed or inverted
shoulder
implant22,24,34,35.
The most recent
investigation34, a
multicenter study in Europe in which seventy-seven patients (eighty shoulders)
with glenohumeral osteoarthritis and a massive rupture of the cuff were
treated with the Delta-III prosthesis, described an improvement in the mean
constant score of 42 points, an increase of 65° in forward elevation, and
minimal or no pain in 96% of the patients. However, forty-nine patients
(63.6%) were noted to have medial component encroachment and scapular notching
without evidence of loosening. Progressive loosening was noted in five glenoid
components, two of which had been revised at the time of publication. In seven
patients, the glenosphere and baseplate dissociated. This uncoupling was
progressive in three patients, with one requiring revision.
In our study, we found similar improvements in active elevation, which
increased from 55° preoperatively to 105° postoperatively, with use of
the Reverse Shoulder Prosthesis. Even though the preoperative external
rotation data were incomplete, the average postoperative measurement of
35.9° in our study is encouraging, especially compared with the 11.2°
of external rotation reported in the study by Sirveaux et
al.34. No patient
in the present study had medial encroachment or progressive erosion of the
glenoid.
Eight shoulders in seven patients needed revision because of failure of the
glenoid baseplate at a mean of 21.4 months postoperatively. In these failures,
it was observed intraoperatively that there was no osseous ingrowth into the
baseplate. The lack of osseous ingrowth made the revision of these devices to
another Reverse Shoulder Prosthesis possible because of minimal loss of
glenoid bone. In our experience, patients with noninfectious etiologies of
symptomatic glenoid loosening and adequate bone stock should have a revision
to another Reverse Shoulder Prosthesis.
A critical time-period for these devices is the first two years because, if
ingrowth does not occur, mechanical failure may result from metal fatigue. The
mode of failure of the baseplates in our study appeared to be metal fatigue in
the screws. However, another cause of failure in the longer term may be the
generation of wear debris particles, which could induce osteolysis. At this
point, there has been no evidence of osteolysis in the form of scapular
notching; however, long-term follow-up will be necessary to determine the
prevalence of this problem.
The device in this study differs from the Delta III in several ways that we
believe are clinically important. The Reverse Shoulder Prosthesis has a more
lateral center of rotation, approximately matching that found in the normal
glenohumeral joint. It also more closely reproduces normal lateral offset and
preoperative shoulder contour (Figs. 4-A,
4-B, and 4-C) In
addition, this lateral center of rotation diminishes medial scapular notching,
which is observed commonly with the Delta-III
prosthesis19,23,24,34,36.
Valenti et al.24,
in a study of thirty-nine patients managed with the Delta-III prosthesis and
followed for at least five years, reported that twenty-two had a scapular
notch on the inferior part of the glenoid and eight of those had erosion of
the inferior screw. Another study noted scapular notching in 65% of the
patients and found that notching substantially affected the Constant score
when the notch was over the screw or was
extensive23. Favard
et al.19 found that
fifty of the eighty shoulders with the Delta-III implant had scapular
notching, which reached the inferior screw in thirteen of the fifty shoulders.
An analysis of a Delta-III prosthesis retrieved post mortem at eight months
after implantation revealed a large notch at the inferior pole of the scapular
neck, which extended beyond the inferior fixation
screw36. Scapular
notching, when extensive, can negatively affect the Constant
score37. The
lateralized center of rotation produces an increased moment at the glenoid
prosthesis interface. Design features, such as a central compressive 6.5-mm
screw, options for multiple peripheral locking and non-locking screws, and a
curved contoured interface, provide stability comparable with that of the
Delta III with its more medial center of rotation.
In summary, the results of arthroplasty with the Reverse Shoulder
Prosthesis in sixty patients with glenohumeral arthritis and severe rotator
cuff deficiency at greater than two years of follow-up were good or excellent
for forty-one patients and were satisfactory for sixteen. While it is too
early to reliably predict the longevity for this "semiconstrained"
implant, it is important to report on early complications and address
theoretical concerns about this prosthesis with its nonanatomic design.
?