Patients
Fifty-eight consecutive patients with painful pseudoparesis (defined as
active shoulder elevation of <90° in the presence of free passive
anterior elevation) caused by a massive irreparable rotator cuff tear were
treated with the Delta III reverse-ball-and-socket total shoulder replacement.
All patients gave their written consent prior to the procedure. Indications
for surgery included severe pain, pseudoparesis, absence of overt septic
arthritis, and an irreparable rupture of at least the supraspinatus tendon and
either the subscapularis or the infraspinatus tendon. The rotator cuff was
considered to be irreparable if the pseudoparesis was chronic, the
acromiohumeral distance was <7 mm on a plain anteroposterior radiograph
made with the shoulder in neutral rotation
(Fig. 2-A), and fatty
infiltration of the supraspinatus and infraspinatus muscles was greater than
stage two according to the Goutallier
classification25
(either on computed tomography scans or as adapted for magnetic resonance
imaging by Fuchs et
al.26).
There were fifteen men and forty-three women. Their mean age at the time of
surgery was sixty-eight years (range, forty-four to eighty-four years). The
dominant shoulder was involved in forty-four patients. Seventeen patients with
cuff tear arthropathy had no previous operation (Group A), whereas, in
forty-one patients, the implantation of the Delta III prosthesis was a
revision procedure (Group B). Of the forty-one revisions, twenty-one were
revisions of another prosthesis (Group B1): fifteen were revisions of a failed
hemiprosthesis that had been used to treat a humeral head fracture; five,
revisions of a failed total shoulder replacement that had been used to treat a
humeral head fracture; and one, a revision of a failed hemiprosthesis that had
been used to treat osteoarthritis. Twenty patients in the revision group had
undergone a previous shoulder operation other than implantation of a
prosthesis (Group B2): seventeen had had a previous reconstruction to treat a
rotator cuff tear; two, arthroscopic débridement for a cuff tear; and
one, open reduction and internal fixation as well as cuff reconstruction for a
proximal humeral head fracture associated with a cuff tear.
Surgical Technique and Postoperative Rehabilitation
All patients were operated on in a beach-chair position with use of general
anesthesia, regional anesthesia, or a combination of both. A deltopectoral
approach without detachment of the anterior aspect of the deltoid was used in
all patients. In patients with a preserved subscapularis tendon, the superior
part of the tendon was released from the lesser tuberosity. The axillary nerve
was exposed and protected. The humeral resection guide was used to achieve
<20° of humeral retroversion. After exposure of the glenoid, all soft
tissues, including the long head of the biceps if it was present, and
peripheral osteophytes were removed. The inferior aspect of the capsule and
the long head of the triceps were released to provide space for the humeral
component. The axis of the glenoid was determined, and the insertion point for
the guide pin was chosen approximately 2 mm inferior to the middle of the
glenoid and approximately centered
anteroposteriorly18.
Our ongoing clinical observations and a biomechanical
investigation27
suggested that the optimal position for the convex glenoid component is more
inferior than that suggested by the manufacturer and that this position can be
obtained if the inferior border of the glenoid baseplate is flush with the
inferior rim of the glenoid surface. The glenoid baseplate was therefore
progressively shifted more distally, to this 2-mm inferior position, over the
course of this study. Glenoid version was measured on preoperative computed
tomography scans and corrected if necessary to 5° to 10° of
retroversion with use of an adjustable guide pin. The humeral component was
implanted with gentamicin-impregnated bone cement (Palacos; Essex Chemie,
Lucerne, Switzerland) in forty-five patients and without cement in thirteen.
The glenoid was replaced with a 36-mm glénosphère in fifty-six
patients and with a 42-mm glénosphère in two. A standard
lateralized humeral polyethylene cup was used in all patients. Suction drains
were also used in all cases.
Physical therapy, consisting of passive and active-assisted exercises, was
initiated immediately after the surgery. Passive and active mobilization in a
pool was started immediately after removal of the suction drain, with the
wound protected by an impermeable dressing. Between physical therapy sessions,
the arm was kept in a sling. An abduction brace was never used. After two
weeks, a free active range of motion was initiated as far as tolerated.
Patients were encouraged to continue physical therapy sessions until six weeks
after the index operation. This protocol is in contrast to the postoperative
immobilization in an abduction splint for six weeks described by Grammont and
Baulot18.
Clinical assessment was performed at a minimum of twenty-four months
(average, thirty-eight months) postoperatively by two of us (C.M.L.W. and
P.A.S.) who had not been involved in any of the surgical procedures. Like the
prospectively performed preoperative examinations, the follow-up assessment
consisted of a structured interview, clinical examination, photographic
documentation, and radiographic evaluation. Also, the shoulders were assigned
scores according to the system of Constant and
Murley28, with use
of the protocol previously described by one of us
(C.G.)29, and the
patients were asked to estimate the value of their shoulder as a percentage of
an entirely normal shoulder (the subjective shoulder
value30).
True anteroposterior, axillary lateral, and Neer scapular lateral
radiographs were made under fluoroscopic control to assess component position,
radiolucent lines, osteolysis, heterotopic ossification, and infrascapular
notching. Inferior scapular notching was classified according to
Nérot's system, as described by Valenti et
al.31 and Sirveaux
et
al.32,33.
Statistical analysis of the results was performed with use of the Wilcoxon
signed rank test (for comparison of specific preoperative and postoperative
values), Spearman rank correlation (for correlation of outcome values with
differences in the preoperative state of the shoulder and according to
subgroups), and the Mann-Whitney U test (for analysis of differences between
subgroups). As multiple comparisons between the groups were not performed, the
Bonferroni-Dunn correction was not necessary to reduce the ß error. The
significance level was set at p = 0.05.
Losses to Follow-up
At the time of follow-up, four patients had died from unrelated
causes, three patients could not be reexamined because of their poor general
health status, and one patient refused to be reexamined because of a poor
clinical outcome (painful dysfunction with subluxation of the prosthesis).
Thus, fifty patients were available for clinical follow-up.
Complications and Failures
(Table I)
The complication and failure rates were known for all fiftyeight patients,
even though we were able to personally examine only fifty of them.
Complications are therefore reported for the entire study group. A
complication occurred in twentynine (50%) of the fifty-eight patients. The
complication had an effect on the final outcome in six patients, as it led to
removal of the prosthesis in two of them and to conversion to a
hemiarthroplasty in four. The complication did not affect the final outcome in
twenty-three patients. The majority of the complications occurred within a few
weeks after the index operation. Nine complications occurred in eight of the
seventeen patients who had undergone primary intervention (group A), thirteen
complications occurred in twelve of the twenty-one patients who had had a
revision of a prosthesis (group B1), and eleven complications occurred in nine
of the twenty patients who had had a revision of a non-prosthetic procedure
(group B2).
Reoperations
Nineteen patients (33%) underwent a total of thirty-one (range, one to
three) reoperations after implantation of the Delta III prosthesis. Eighteen
percent (three) of the seventeen patients in group A, 38% (eight) of the
twenty-one patients in group B1, and 40% (eight) of the twenty patients in
group B2 had a reoperation.
Failures
The six patients who had removal or conversion of the prosthesis were
considered to have had a failure. Two of these patients were in group B1, and
four were in group B2. No patient in group A had a complication necessitating
removal or conversion of the prosthesis. Details of these six cases are
summarized in Table I. Their
functional results are not included with the reported results for the
forty-four patients in whom the Delta III prosthesis was in situ at the time
of the final follow-up.
In two patients, the prosthesis had to be removed because of an infection
that had been unexpectedly identified by positive results of bacteriological
cultures of intraoperative specimens. In one of these two patients, the
prosthesis was replaced by a cement spacer one week after the index operation.
Because of persistent fungal infection, however, the spacer was later removed
and not replaced. In the other patient, the prosthesis was not reimplanted
after removal of the prosthesis and implantation of a cement spacer because
the patient had pain-free, subjectively satisfactory shoulder function. The
patients both had a poor clinical outcome with relative Constant scores of 14%
and 7% (10 and 5 points) at the time of follow-up; these scores corresponded
to almost no shoulder function. However, the patients were pain-free and did
not request additional surgical treatment.
In four patients, the prosthesis had to be converted to a hemiarthroplasty,
as a result of loosening of the glenoid component in three and because of
repeated anterosuperior dislocation of the prosthesis in one. The relative
Constant scores for these patients were 23%, 57%, 75%, and 20% (17, 39, 48,
and 14 points). Three of the patients had moderate pain, and one had severe
persistent pain.
The most common complication, which did not affect the ultimate outcome,
was postoperative hematoma, recorded for twelve patients. Aspiration of the
hematoma alone was performed in five of these patients, and the other seven
patients underwent open revision (partly because of concomitant complications
requiring revision).
In addition to the patient who had the conversion to the hemiarthroplasty,
as described above, four patients had anterosuperior dislocation of the
prosthesis, which was attributed to insufficient soft-tissue tension and was
recurrent in one of them. Closed reduction was the definitive treatment for
one patient. The three remaining patients were treated successfully with
surgical lengthening of the neck of the prosthesis by placement of a
réhausseur, which is a 9-mm screw-in piece that can be added to the
epiphysis of the prosthesis.
The polyethylene inlay dislocated and had to be revised in one patient. The
polyethylene cup was replaced, following which a primary infection developed.
It was necessary to temporarily remove the prosthesis and implant an
antibiotic-impregnated cement spacer before the prosthesis was reimplanted.
Three other patients (in addition to the two treated with prosthetic removal,
as described above) had positive cultures and biopsies of specimens taken at
the time of the index operation. The bacteriological examinations revealed
coagulase-negative Staphylococcus in one case and Propionibacterium
acnes in two. All three infections were successfully treated with
intravenous antibiotics.
One axillary nerve lesion, which resolved completely, was recorded
postoperatively. Another patient had a full loss of shoulder function due to
persistent axillary nerve palsy following the resection of an intraspinal
cervical ependymoma one year after the shoulder surgery. In another patient,
progressive telescoping of an uncemented humeral component into the humeral
shaft led to instability of the prosthesis. This patient underwent revision of
the humeral component and implantation of a larger-diameter shaft with
cement.
A fracture of the acromion or the scapular spine was recorded in four
cases. These injuries were considered to be fatigue fractures of an eroded
acromion or scapula and, in two patients, open reduction and internal fixation
(tension band wiring) was necessary. The other two patients were successfully
treated conservatively.
Clinical Outcome (Table
II)
Apart from the six patients who had a hemiarthroplasty or removal of the
prosthesis, as reported above, and the one patient who refused to return for
follow-up because of a poor clinical outcome, the patients had a high degree
of satisfaction.
Constant Score
The Constant score had at least doubled in all groups of patients in whom
the prosthesis was still in place (forty-four patients), and was similar in
the primary and revision groups. The mean relative Constant score increased
from 29% preoperatively to 64% postoperatively (a gain of 35%, p < 0.0001)
in the overall group, from 35% to 72% in group A, from 25% to 55% in group B1,
and from 27% to 61% in group B2.
Subjective Shoulder Value
The average subjective shoulder score approximately tripled in all three
groups. It increased from 18% to 56% (a gain of 38%, p < 0.0001) in the
overall group, from 19% to 59% (a gain of 40%) in group A, from 12% to 53% (a
gain of 41%) in group B1, and from 23% to 52% (a gain of 29%) in group B2.
Pain
The average Constant pain score (with 0 points indicating the most severe
pain imaginable and 15 points indicating no pain) improved from 5.2 to 10.5
points (a gain of 5.3 points, p < 0.0001) overall. It improved from 6.2 to
12.3 points (a gain of 6.1 points) in group A, from 4.1 to 9.6 points (a gain
of 5.5 points) in group B1, and from 5.6 to 8.9 points (a gain of 3.3 points)
in group B2.
Range of Active Motion
The average active anterior elevation increased from 42° to 100° (a
gain of 58°, p < 0.0001) in the forty-four patients with the prosthesis
in place. It increased from 43° to 103° in group A, from 39° to
96° in group B1, and from 44° to 97° in group B2. The average
abduction increased from 43° to 90° (a gain of 47°, p < 0.0001)
overall, from 39° to 84° in group A, from 43° to 95° in group
B1, and from 47° to 91° in group B2. The average active external
rotation with the arm at the side decreased from 17° to 12° (a loss of
5°, p = 0.04).
Differences Between Subgroups
With the numbers available, no significant difference in the occurrence of
complications, the subjective shoulder value, or patient satisfaction could be
found between group A and group B or between group B1 and group B2. However,
the rate of reoperations in group A (18%) was significantly higher than that
in group B (39%) (p = 0.005). Also, there was a significant negative
correlation between previous operations and the final Constant score, which
averaged 72% in group A and 58% in group B (p = 0.03), and the postoperative
pain score, which averaged 12.3 points in group A and 9.2 points in Group B (p
= 0.01). The number of complications encountered did not appear to influence
the overall satisfaction of the patients in whom the Delta III prosthesis had
been preserved (p = 0.64).
Radiographic Outcome
Radiolucencies were detected around parts of the humeral stem in eight
patients (one in group A, three in group B1, and four in group B2). Seven of
the stems were cemented, and one was uncemented. Only one stem (uncemented)
was also clinically loose and had to be replaced. Radiolucent lines were found
around the baseplate in five patients (Fig.
2-B); three baseplates were considered to be loose (with lucencies
around the screws as well; Fig.
3). Notching of the inferior part of the scapular neck was found
on 96% of the forty-eight follow-up radiographs that were available for
interpretation (Figs. 2-B and
4). The notching did not extend
to the inferior fixation screw of the implant (grades 1 and 2 according to the
Nérot system reported by Sirveaux et
al.32,33)
in twenty-five cases (54%), and it reached or extended beyond the inferior
screw (grades 3 and 4) in twenty-one cases (46%). The notching could not be
correlated with any objective or subjective clinical result or complication.
In thirty-eight (79%) of the forty-eight cases in which notching progression
was analyzed on subsequent radiographs
(Fig. 4), there was no
enlargement of the osseous defect after one year. In half of those cases, the
bone defect reached its final extent within a few months after implantation.
In the ten patients with progression of the osseous defect, the notching had
increased by a maximum of one grade according to the Nérot
system32,33
at the time of the most recent follow-up.
The Delta III reverse-ball-and-socket shoulder prosthesis was
developed by
Grammont17,18
to restore shoulder function in the presence of an irreparable rotator cuff
tear associated with glenohumeral osteoarthritis. As a result of the design of
the implant, the center of rotation of the glenohumeral joint is medialized
and the insertion of the deltoid muscle is moved distally. This increases the
lever arm and the pretension of the deltoid muscle. The Delta III prosthesis
can therefore address both the degenerative changes of the glenohumeral joint
and the missing rotator
cuff17,18,24,31-34.
All patients in our series had an irreparable rotator cuff tear and severe
functional impairment, with active anterior elevation of the shoulder not
exceeding 90° (so-called pseudoparesis). This functional impairment was
associated with either degenerative changes of the glenohumeral joint (group
A) or with previous shoulder surgery (group B), including total joint
replacement. Both the subjective and the objective outcome values at a minimum
of two years after the operation showed that the procedure had improved the
patient's function substantially, with similar improvements in those treated
with a primary arthroplasty and those treated with a revision. The operation
resulted in a high degree of patient satisfaction when no complications
requiring removal of the prosthesis had occurred. Whereas primary and revision
arthroplasties resulted in similar improvements in the range of motion, the
pain and relative Constant scores were significantly higher following the
primary procedures. This finding is in contrast to that in the studies by
Sirveaux et
al.32,33,
who reported that the Constant score was not influenced by previous surgery.
The subjective shoulder value after implantation of the Delta III prosthesis
was similar for all treatment groups in our study and was generally higher
than the results reported following
hemiarthroplasty12,14,35-40.
However, patient satisfaction was low when the prosthesis had to be converted
to a hemiarthroplasty or had to be removed.
Active external rotation was, as expected, not improved but actually
decreased by a mean of 5° (p = 0.04). This finding reflects the inability
of the deltoid to control external rotation, particularly in elevation and
abduction41. The
loss of active external rotation was also observed by Baulot et
al.42, and patients
should be informed about this outcome before the surgery. For patients with a
severe external rotation lag and a torn teres minor, an additional tendon
transfer29 might
improve this function and is the subject of a current investigation.
The complication rate in our series is higher than previously reported
rates18,24,31-34.
One or more complications occurred in twenty-nine (50%) of the fifty-eight
patients. In six patients, the complication required removal of the prosthesis
or conversion to a hemiarthroplasty. The higher complication rate in our
series might have been due to the fact that we considered hematoma formation,
which occurred twelve times, to be a complication. This high rate of hematoma
formation might be explained by the fact that (1) compared with conventional
prosthetic designs, the reverse Delta-III design leaves a larger space free of
soft-tissue after implantation, and (2) the initiation of postoperative
physiotherapy before the wound has healed completely might favor hematoma
formation. Another reason why the complication rate was high is that five
infections that had not been apparent before the operation and were diagnosed
with bacteriological examination of culture specimens obtained at the index
operation were reported as complications. Finally, some complications were
recorded separately even though they represented a single clinical event
(e.g., loosening of the humeral component leading to dislocation of the
prosthesis with a concomitant hematoma). Interestingly, in this series, in
which the percentage of revisions was high (33%; nineteen patients), group-B
patients did not have more complications than did group-A patients, but they
did have a much higher rate of subsequent revisions, indicating that the
complications were more serious in that group.
Subsidence of the humeral component was the reason for one revision.
Because of this complication, we stopped implanting the humeral stem without
cement. The manufacturer does not make any recommendations regarding the use
of cement. Radiolucencies around the glenoid component were rare compared with
the rates of glenoid radiolucency following conventional total shoulder
arthroplasty8,43-47.
This finding could be due to the fact that the glénosphère
overlaps the
baseplate32 by
about 1 mm in situ and, therefore, small radiolucent lines could remain
undetected even on perfectly oriented true anteroposterior radiographs. On the
other hand, biomechanical
analyses17 have
suggested that shear forces at the interface between the baseplate and the
scapula depend on the force developed by the deltoid in the absence of the
rotator cuff. This force and the shear force on the implant both decrease as
the lever arm of the deltoid increases. This is an advantage of the Delta III
prosthesis and may explain the low rate of early glenoid failures compared
with the rates associated with fixed-fulcrum designs.
Asymptomatic osteolysis at the distal part of the scapular neck, also
referred to as notching, was observed in forty-six of the forty-eight patients
with satisfactory follow-up radiographs (true anteroposterior views allowing
correct grading with the Nérot
system31). The
lesion appeared rapidly after implantation and was stable after one year in
thirty-eight (79%) of the patients. In contrast to the findings reported by
Sirveaux et al.32,
higher-grade (grade-3 or 4) notching did not correlate with lower Constant
scores or any other clinical entity.
Nérot31
proposed that grade-4 notching corresponds to glenoid loosening. In our
patients, however, grade-4 notching was not associated with glenoid loosening.
This finding might be due to the fact that the screws can still ensure
stability even in the absence of contact between the baseplate and the glenoid
bone (Fig. 4). We therefore
think that additional radiolucencies around the screws are necessary to make a
diagnosis of glenoid loosening with this prosthesis. Also, in contrast to the
observation by Sirveaux et
al.33, the
radiographic presentation of grade-4 notching did not have the same congruent
appearance as the lower-grade notching. Rather, it seemed to creep along the
backside of the baseplate, as is seen in cases of osteolysis due to
polyethylene wear. This finding was also reported by Nyffeler et
al.48.
Lateralization of the humeral component with use of a réhausseur (the
additional epiphyseal part of the prosthesis) does not prevent the development
of inferior scapular notching. It is, however, particularly useful when
soft-tissue tension is lacking after implantation of the component. Also, the
réhausseur can be used to lateralize the arm, allowing somewhat more
adduction to occur, in shortbodied obese patients who have a very medialized
scapula.
Although the etiology of inferior scapular notching remains unknown, the
morphology of the lesion suggests a mechanical impingement between the
polyethylene cup of the humeral component and the inferior bone of the
scapular
neck27,48.
Implantation of the glenoid component with an inferiorly tilted inclination
might be considered in order to reduce impingement at the inferior aspect of
the scapular neck. However, this would require reaming of the inferior aspect
of the glenoid toward the lateral pillar of the scapular blade. The lateral
border of the scapula would thereby be brought into direct contact with the
adducted humeral component, and mechanical impingement would ensue.
Furthermore, reaming in this fashion decreases the cortical substance left for
implantation of the glenoid component and might be a factor predisposing to
glenoid
loosening27. The
main factor associated with notching is positioning of the
métaglène too superiorly on the glenoid, as was reported in
detail by Nyffeler et
al.27. It is
important to recognize that the position recommended by the manufacturer
(DePuy) does not correspond with the optimal position shown by testing and
should not be used.
On the basis of the findings in this series, we concluded that the
implantation of a Delta III prosthesis is a valuable procedure for the
treatment of severe dysfunction of the shoulder due to an irreparable rotator
cuff tear associated with glenohumeral pathology (degenerative osteoarthritis
or prior surgery, including total shoulder replacement). However, because of
the high complication rate and the fact that there may be long-term
complications that are not yet known, arthroplasty with this implant should be
reserved as a salvage procedure for situations in which an acceptable clinical
outcome cannot be expected with another treatment modality.