Study Group
Between May 1995 and June 2003, 240 consecutive reverse total shoulder
arthroplasties were performed at the Clinique Sainte Anne Lumière. All
procedures were performed by one of two surgeons (G.W. or L.N.-J.). The
prosthesis that was implanted was either the Delta III (DePuy France, Saint
Priest, France) or the Aequalis system (Tornier, Montbonnot, France), both of
which are based on the Grammont design with a medialized center of rotation.
The Delta III implant was used in 209 shoulders, and the Aequalis was used in
thirty-one shoulders.
The indications for reverse total shoulder arthroplasty were classic
rotator cuff tear arthropathy as described by Neer et
al.1; a massive
irreparable rotator cuff tear with chronic loss of elevation (of more than six
months' duration) that had failed to respond to treatment with physiotherapy;
posttraumatic glenohumeral arthritis with rotator cuff compromise; primary
osteoarthritis with rotator cuff compromise; primary osteoarthritis with
severe glenoid bone loss and static posterior instability that prevented
insertion of an unconstrained glenoid component; rheumatoid arthritis with
rotator cuff compromise; an acute comminuted displaced proximal humeral
fracture in an elderly patient; a shoulder girdle tumor requiring resection of
all or a portion of the rotator cuff, creating rotator cuff compromise; and
revision arthroplasty with rotator cuff compromise. In the present series,
rotator cuff compromise was defined as an irreparable tear of at least two
tendons or grade-3 or 4 fatty infiltration of the infraspinatus or
subscapularis on preoperative computed tomography images according to the
classification system described by Goutallier et
al.17. Although no
specific age limits were imposed, the prosthesis generally was reserved for
older patients or for younger patients in whom it would have been impossible
to reconstruct the rotator cuff. Severe posterior or superior glenoid bone
loss was used as a relative indication for reverse total shoulder
arthroplasty. The reverse total shoulder arthroplasty was used in cases in
which the glenoid defect jeopardized glenoid keel fixation or prevented the
implantation of the glenoid component at an acceptable angle to prevent
posterior subluxation. The use of the press-fit component allowed for easier
structural grafting of the glenoid, and the constrained nature of the reverse
total shoulder arthroplasty prevented recurrent posterior subluxation.
Two hundred and forty prostheses were implanted into 232 patients, with
eight patients having a bilateral procedure. The average age of the patients
was 72.7 years (range, twenty-three to eighty-six years). One hundred and
eighty-four procedures were performed in female patients, and fifty-six were
performed in male patients. One hundred seventy-three prostheses were placed
in the right shoulder, and sixty-seven were placed in the left shoulder. The
dominant shoulder was involved in 169 cases, and the nondominant shoulder was
involved in sixty-seven cases; in the remaining four cases, the patient was
ambidextrous.
The patients were grouped according to the primary etiology of the shoulder
disorder (Table I). The
radiographic classification system described by Hamada et al. was used to
grade the preoperative radiographs as needed to differentiate rotator cuff
tear arthropathy from a massive rotator cuff tear without arthritis
(Fig.
1)18.
According to this system, stage 1 is associated with minimal radiographic
changes, stage 2 is characterized by narrowing of the subacromial space to
=5 mm, stage 3 is defined as erosion and so-called acetabulization of the
acromion secondary to superior migration of the humeral head, stage 4 is
associated with glenohumeral arthritis and is subdivided into stage 4a
(without acetabulization) and stage 4b (with acetabulization), and stage 5 is
characterized by the presence of humeral head osteonecrosis. Patients who met
the criteria for Hamada stage-4a, 4b, or 5 radiographic changes, which
provided evidence of the glenohumeral changes according to the definition
described by Neer et
al.1, were
considered to have cuff tear arthropathy (Figs.
2-A and
2-B). Patients with Hamada
stage-1, 2, or 3 changes were considered to have a massive rotator cuff tear
without arthritis. The distribution of patients according to Hamada
classification is shown in Table
II.
The patients who were classified as having glenohumeral osteoarthritis with
a rotator cuff tear had no proximal migration of the humeral head. The
patients who were classified as having posttraumatic arthritis included those
who had glenohumeral arthritis and a history of a fracture involving the
proximal part of the humerus. The patients who were managed with revision
arthroplasty included those who had had a previous hemiarthroplasty (thirty
shoulders) and those who had had a previous total shoulder arthroplasty
(twenty-four shoulders).
Data Collection
The present study was approved by our center's bioethics committee, and all
subjects provided informed consent to allow their information to be used in
the study. All patients were evaluated preoperatively and postoperatively by
an examiner (B.W.) other than the operating surgeon. Preoperative and
postoperative ranges of motion were documented, and Constant scores were
collected19.
Subjective results were graded by asking the patients to rate their overall
experience with the procedure as very satisfied, satisfied, uncertain, or
disappointed.
Preoperatively, computed tomography and computed tomography-arthrography
were used to evaluate the quality of the rotator cuff and glenoid bone stock.
Postoperative standardized radiographs, including anteroposterior views of the
glenohumeral joint in neutral rotation, internal rotation, and external
rotation as well as outlet and axillary views, were made under fluoroscopic
control.
For seventeen patients who were unable to return to the clinic for
follow-up, a telephone interview was conducted and radiographs were made at an
outlying facility and were sent in for review. For an additional five
patients, a physical examination (including measurement of range of motion,
computation of the Constant score, and assessment of the subjective result)
was conducted by a local physician.
Operative Technique
All but three of the prostheses in this series were placed with use of a
deltopectoral approach. From 1995 to 2002, the Delta III implant was used (209
shoulders). In 2002 and 2003, the Aequalis Reversed implant was used
(thirty-one shoulders). Both of these implants are based on the Grammont
design with a medialized center of rotation. The humeral stem was cemented in
all cases but one. The number of screws that were used to fix the glenoid
baseplate was dictated by the available glenoid bone stock and surgeon
preference. All four screws were used in 203 shoulders, and three screws were
used in thirty-one shoulders; the remaining six shoulders required a custom
glenoid implant because of glenoid bone loss. In cases in which the
subscapularis was still intact, it was divided through the tendinous portion,
approximately 1.5 cm medial to the insertion, in line with the anatomic neck
of the humerus. Subscapularis tendon repair with use of transosseous
nonresorbable sutures was possible in 137 cases. No attempt was made to
reattach the tendon in the remaining 103 cases.
Postoperative Rehabilitation
Postoperatively, the shoulder was immobilized with use of a simple sling
that held the arm in internal rotation for one month. The sling was removed
for patient hygiene, and the patient was allowed to use the hand on the
involved side for simple activities of daily living but was instructed to
avoid lifting activities. Passive range of motion was begun immediately. After
one month, use of the sling was discontinued and the patient was allowed
activity as tolerated.
Data Analysis
Analysis of variance was used to compare the functional scores and ranges
of motion in the various etiologic groups. Improvements in functional scores
and range of motion from the preoperative period to the time of the most
recent follow-up within each etiologic group were analyzed with
repeated-measures analyses. Satisfaction and complication rates were compared
between groups with use of chi-square analyses. All post hoc tests were
adjusted for multiple comparisons with use of the Sidak method. The level of
significance was set at p = 0.05.
Of the original 232 patients, five had a reverse shoulder arthroplasty
following tumor resection (two patients) or for the treatment of acute
fracture (two patients) or rheumatoid arthritis (one patient). These patients
were excluded from the analysis because the small number of subjects with
these etiologies would not yield meaningful group comparisons.
Of the remaining 227 patients, twenty-two (9.7%) died before the minimum
two-year follow-up, eight (3.5%) had removal or revision of the prosthesis,
and eleven (4.8%) were lost to follow-up. This left 191 shoulders in 186
patients (forty-one men and 145 women) who were available for study after an
average duration of follow-up of 39.9 months (range, twenty-four to 118
months).
All 186 patients completed a questionnaire and were evaluated
radiographically; five patients were unable to return for a clinical
examination. The mean age at the time of follow-up was 75.3 years (range,
twenty-six to eighty-nine years). The dominant side was affected in 133
patients (71.5%), and the nondominant side was involved in fifty-two; the
remaining patient was ambidextrous. Four patients had undergone bilateral
reverse shoulder arthroplasty.
Overall Functional and Clinical Outcomes
The average Constant score improved from 22.8 points before surgery to 59.7
points at the time of follow-up. All components of the Constant score improved
significantly after reverse total shoulder arthroplasty
(Table III). Across all
etiologic groups, the mean active elevation improved from 86° to 137°
(p < 0.001) and the mean internal rotation improved from L5 to L4
(Table IV). No significant
improvement was seen in terms of external rotation with the arm at the side or
with the arm at 90° of abduction. Patients with repair of the
subscapularis had greater improvement in the amount of internal rotation (from
L5 to L4) than did those without repair (from the sacrum to L5).
When asked to grade the subjective result of the procedure, 111 (59.7%) of
the 186 patients reported being very satisfied, sixty-two (33.3%) reported
being satisfied, eleven (5.9%) reported being uncertain, and two (1.1%)
reported being disappointed.
Functional and Clinical Outcomes According to Etiology
Patients with primary rotator cuff tear arthropathy, primary osteoarthritis
with a rotator cuff tear, and a massive rotator cuff tear without arthritis
had the best final outcomes (Tables
III and
IV). These three groups did not
differ significantly from one another with respect to postoperative Constant
scores (p = 0.540), range of motion (p = 0.350), or the subjective rating of
the outcome (p = 0.127).
In contrast, the patients in the posttraumatic arthritis and revision
arthroplasty groups had significantly worse postoperative Constant scores (53
and 52, respectively) in comparison with the other three groups (p = 0.006)
(Table III). Patients in the
posttraumatic arthritis and revision arthroplasty groups also had
significantly worse postoperative ranges of elevation (115° and 118°,
respectively) in comparison with the other three groups (p = 0.001)
(Table IV). In addition, the
percentage of patients who stated that they were very satisfied or satisfied
with the outcome was lower in the posttraumatic arthritis and revision
arthroplasty groups (89%) than in the other three groups (96%), although this
difference did not achieve significance (p = 0.083).
The postoperative Constant scores were significantly related to the
patients' subjective ratings (p < 0.01) in all of the etiology groups
except the massive rotator cuff tear group (p = 0.648). The postoperative
Constant scores were significantly related to the postoperative active range
of elevation in all of the etiology groups (p < 0.001). The postoperative
Constant scores were not related to the preoperative Hamada score (p = 0.503)
in any of the etiology groups.
Patients undergoing reverse total shoulder arthroplasty for revision of a
hemiarthroplasty had a significantly worse mean preoperative Constant score
(16.4 points) (p = 0.038) and worse preoperative strength (0.6 kg) (p = 0.025)
in comparison with those undergoing reverse total shoulder arthroplasty for
revision of a total shoulder arthroplasty (25.6 points and 2.5 kg,
respectively). However, following reverse total shoulder arthroplasty, both
groups had approximately the same degree of improvement from their respective
baseline values for the Constant score, strength, and range of motion. Thus,
previous surgery had no apparent effect on the relative degree of improvement
following reverse total shoulder arthroplasty when performed as a revision
procedure, but patients who had had failure of a previous hemiarthroplasty had
lower preoperative ratings and thereby had a worse final outcome in comparison
with those who had had failure of a previous total shoulder arthroplasty.
Radiographic Outcomes
Two patients had signs of glenoid loosening. In both cases, the loosening
was thought to be attributable to surgical error. In the first patient, the
superior approach had been used and the glenoid component had been placed with
a substantial superior tilt. The end result was superior cutout and loosening
of the implant. The second patient had had severe posterior glenoid wear at
the time of the reverse arthroplasty, and a structural iliac crest graft had
been used to reconstruct the glenoid bone stock. The central peg of the
baseplate, however, had not been anchored in the native host bone. This
patient had rapid loosening of both the graft and implant (Figs.
3-A and 3-B,
3-C,
3-D). In both cases the
arthroplasty was converted to hemiarthroplasty, and both patients reported
good pain relief but little functional improvement.
Two patients had signs of loosening of the humeral component. One patient
had undergone revision arthroplasty with an uncemented stem and had required a
longitudinal osteotomy of the humerus. The second patient had been managed
with a cemented stem for the treatment of cuff tear arthropathy.
Improper seating of the glenosphere on the metaglene baseplate was seen in
five patients. In all cases this appeared to be a stable problem, and none of
the patients required a revision. There was no deleterious effect on either
the functional scores or the range of motion. At the time of review, these
five patients had a mean Constant score of 57.2 points and a mean elevation of
141°.
Of the 186 patients with two-year follow-up radiographs, 152 had adequate
radiographs to evaluate potential scapular notching. Seventy-seven (50.7%) of
the 152 patients had evidence of scapular notching. Notching could be seen on
only the anteroposterior view in sixty-two cases, on only the axillary view in
twelve cases, and on both views in three cases. The mean postoperative
Constant score was 60.6 points for patients with notching and 58.7 points for
those without notching. The mean elevation was 132° for patients with
notching and 131° for those without notching. There were no cases of
glenoid loosening in the group of patients with notching.
Complications
The eight patients who had removal or revision of the reverse prosthesis
before completing the two-year follow-up period were included in this portion
of the analysis, bringing the total number of cases reviewed to 199.
Thirty-six patients had a total of thirty-eight complications (prevalence,
19.1%). The most common complications were dislocation (fifteen cases;
prevalence, 7.5%) and infection (eight cases; prevalence, 4.0%). Glenoid
fractures, postoperative humeral fractures, symptomatic hardware,
musculocutaneous nerve palsy, radial nerve palsy, glenoid sphere loosening,
and glenoid base loosening also occurred in five or fewer cases each. The risk
of complication associated with revision surgery (36.7%; eighteen of
forty-nine) was significantly higher than the risk of complication associated
with primary surgery (13.3%; twenty of 150) (p < 0.001). With the numbers
studied, repair of the subscapularis was not related to the occurrence of
postoperative complications (p = 0.123) or dislocations (p = 0.115).
During thirteen (24.1%) of the fifty-four revision procedures, a humeral
fracture occurred during removal of the primary prosthesis or cement mantle.
These events were not counted as complications because they were thought to be
related to the revision surgery and not to the reverse prosthesis itself.
The short-term functional results of reverse total shoulder arthroplasty
appear to be excellent, and the level of patient satisfaction also appears to
be
high12,20-22.
Overall, patients in the current series had 37 points of improvement in terms
of the Constant score and 51° of improvement in terms of active elevation,
which are gains that are comparable with the findings described in previous
reports of reverse total shoulder arthroplasty for the treatment of rotator
cuff tear
arthropathy12,20-22.
As has been the case in previous
reports22,23,
the patients in the current study had no gains in active external rotation
with the arm at 0° of abduction; this finding can be attributed to the
prosthesis design (which produces limited lateral offset of the glenospher e
and medialization of the center of rotation) and, possibly, to the status of
the teres minor. We believe that the present report describes the first study
in which functional and clinical outcomes of reverse total shoulder
arthroplasty have been compared according to etiology.
Our results showed that patients who were managed with a reverse total
shoulder arthroplasty for the treatment of posttraumatic arthritis or for a
revision arthroplasty fared worse than patients with cuff tear arthropathy,
primary osteoarthritis associated with a massive rotator cuff tear, or a
massive rotator cuff tear alone. The patients in the posttraumatic arthritis
and revision arthroplasty groups had worse preoperative Constant scores and
worse active elevation as compared with those in the other three groups. Thus,
although patients in the posttraumatic arthritis and revision arthroplasty
groups had improvements of similar magnitudes in terms of shoulder motion and
function, they did not achieve the level of performance observed among
patients in the other groups at the time of the most recent follow-up.
Previous studies have indicated that the results of reverse total shoulder
arthroplasty as a revision procedure are less predictable than those of
reverse total shoulder arthroplasty as a primary
procedure22,23.
Our results showed that the improvement in function associated with reverse
total shoulder arthroplasty appears to be stable for at least three years,
although previous investigations have indicated that improvement can be
maintained for much longer. For example, Valenti and coauthors reported on a
group of sixty patients who underwent reverse shoulder arthroplasty for the
treatment of cuff tear
arthropathy13.
After an average duration of follow-up of eighty-four months, they showed no
increase in the rate of loosening or deterioration in function, with a
survival rate of 93% at five years. Similarly, Guery et al. reported a
survival rate of 84% at ten years in a group of seventy-seven patients who had
undergone reverse shoulder
arthroplasty24.
Several previous studies have indicated that hemiarthroplasty for the
treatment of cuff tear arthropathy decreases but does not eliminate pain and
that it increases active elevation to between 85° and
120°25-29.
In comparison, eighty-four of our 186 patients reported having no pain in the
involved shoulder at the time of the most recent follow-up; an additional
seventy-eight of our 186 patients reported having only mild pain. In addition,
our patients with cuff tear arthropathy were able to achieve an average active
elevation of 142° (range, 95° to 180°).
Latissimus dorsi transfer is commonly performed for the treatment of
massive, irreparable tears of the rotator cuff that have failed to respond to
nonoperative treatment. Latissimus dorsi transfer has been shown to increase
active elevation to between 135° and 170° and to improve functional
ability30-33.
Some studies, however, have indicated that a substantial proportion
(23%33 to
42%30) of patients
may have development of osteoarthritic changes within three years after the
procedure. In the present study, reverse total shoulder arthroplasty was
successful for restoring active elevation and function among patients with
irreparable rotator cuff tears while addressing detected or undetected
arthritic changes.
The complication rate in the current study (19%; thirty-eight of 199) was
lower than has been previously reported in association with reverse total
shoulder arthroplasty. For example, one study demonstrated a complication rate
of 50%22. That
study, however, included relatively minor complications, such as wound
hematomas, that did not affect the final outcome. In addition, 70% of the
arthroplasties in the previous series were revision procedures. The
complication rate associated with revision procedures in the present study
(37%; eighteen of forty-nine) was similar to the overall complication rate
reported in that
study22, but the
complication rate associated with our primary procedures (13%; twenty of 150)
was considerably lower. Thus, patients undergoing reverse total shoulder
arthroplasty as a revision procedure or for the treatment of trauma are at the
greatest risk for postoperative complications.
The present study had several limitations. The retrospective design did not
allow for a direct comparison between reverse total shoulder arthroplasty and
other treatments for the included etiologies. The inclusion of patients who
had been managed by two experienced surgeons who were working at the same
specialty clinic with a high referral caseload also introduces the possibility
of selection bias; less experienced surgeons may not obtain the same
results34. In
addition, the minimum duration of follow-up of twenty-four months is
relatively short when dealing with the results of arthroplasty.
We believe that the present study represents the largest reported series of
patients who have been managed with reverse total shoulder arthroplasty to
date. The present study shows that the reverse total shoulder arthroplasty
prosthesis may be used not only for patients with cuff tear arthropathy but
also for those with a number of other complex shoulder problems in whom the
soft tissues or glenoid bone stock may be deficient. The advanced age of the
patients in this series and the relatively short duration of follow-up suggest
that the prosthesis should continue to be used judiciously, particularly
because the use of this prosthesis in difficult cases, for example, for
revision arthroplasty or for the treatment of posttraumatic arthritis, results
in less improvement and higher complication rates. ?