FAILED SHOULDER ARTHROPLASTY: REVISION AND ARTHRODESIS

Successful prosthetic reconstruction of the shoulder
joint requires accurate alignment of the prosthetic joint surfaces and
careful balancing of the surrounding soft tissues. Instability
following arthroplasty can result from failure to achieve one or both
of these objectives and is best considered according to the direction
of instability (Table 103.3).

Rates of postoperative subluxation or dislocation
following total shoulder replacement have been reported to be as high
as 18.2% (18). There are some conflicting reports

about the most common instability pattern seen after arthroplasty (5,18,72,73).
A collective review of clinical series published before 1988 found that
anterior instability occurred most frequently, followed by posterior
and inferior instability (18). In contrast, Boyd et al. (5)
reported a series of 127 shoulders and found that 23% of shoulders
demonstrated superior migration. When approaching prosthetic
instability, categorization according to the direction of subluxation
is useful for developing a rational approach to treatment.

Rotator cuff rupture occurs in 2% of patients who have undergone total shoulder arthroplasty (18). Neer et al. (56) reported five rotator cuff tears in their 1982 review of 261 total shoulders. Two patients underwent surgical

repair, two had no pain but were unable to raise their arms, and one
patient had intermittent pain but refused further surgical
intervention. Cofield (14)
reported five rotator cuff ruptures in his review of 77 unconstrained
Neer prostheses. One patient suffered severe pain, had limited range of
motion, and was considering further surgery. The remaining four
patients had no pain but had functional weakness.

Rotator cuff tears that occur after total shoulder
replacement for osteoarthritis with an initially intact cuff may
present with pain and increasing weakness. Base the decision to repair
the cuff on the degree of symptoms and functional compromise. If the
humeral component was placed too proud, revise the humerus at the time
of cuff repair, lest the compression against the acromion and the
increased stress in the cuff tendon as it winds around the overly
superior head lead to a retear. In such a situation, the cuff is most
easily repaired while the prosthetic head is out.

Infection complicates shoulder arthroplasty in less than 0.5% of cases (18).
This rate is lower than those seen in replacement of the hip and knee,
which range from 0% to 3%. This may be in part a result of the
excellent soft-tissue coverage and abundant vascularity present in the
shoulder. The hip and knee literature have identified rheumatoid
arthritis, male sex, skin ulcers, diabetes, postoperative urinary tract
infections, trauma, and revision arthroplasty as potential risk factors
for the development of sepsis (23,69).
Prosthetic replacement for shoulder trauma has higher rates of
infection than that for arthritis, with rates ranging from 3.1% to 4.8%
(13,18,72). This increase is likely because of the additional soft-tissue trauma, bone damage, and hematoma.

The presentations of aseptic loosening and late infection after total shoulder arthroplasty can be similar. Codd et al. (13)
found no preoperative investigation to be a reliable indicator for
active sepsis in an arthroplasty. They found a 20% false-negative rate
for joint aspiration, a 40% false-negative rate for triple-phase bone
scans, and a 13% false-negative rate for intraoperative cultures.
Analysis of preoperative blood work showed that white blood cell counts
were usually within normal limits, and in 13% the sedimentation rate
was normal. (These problems occur in the hip and knee as well.) When
infection is suspected in a patient with shoulder arthroplasty, a
thorough evaluation is required. When in doubt, an open biopsy sent for
pathologic evaluation and culture is the most reliable assessment tool.

Early reports recommended resection arthroplasty for infected shoulder replacements (15,53).
In contrast, treatment options for infected total hips and knees in-
clude irrigation and debridement, immediate exchange arthroplasty or
staged reimplantation, and resection arthroplasty. The application of
these principles to infected shoulder arthroplasties has been reported
recently (13).

Because shoulder arthroplasty infection is uncommon, a
treatment algorithm has yet to be defined. Given the poor functional
results for activities performed at or above shoulder level with
resection arthroplasty (15,52), reimplantation is the preferred treatment (Fig. 103.3).

Codd et al. (13), in a series of
16 patients with septic shoulder arthroplasties, showed that
reimplantation of another prosthesis after treatment of the deep
infection is a reasonable option. These patients had no recurrence of
infection at an average follow-up of 47 months. Base decisions about
timing and whether to do another arthroplasty on the relative virulence
of the organism, the chronicity of the infection, and the overall
health of the patient. In rare cases of early infection (less than 6
weeks after surgery) by antibiotic-susceptible organisms, irrigation
and debridement without component removal may be considered by
inference from the hip literature; however, data for the shoulder are
limited.

The increased frequency of arthroplasty in the treatment
of arthritic shoulder conditions, and the increased functional demands
of patients with shoulder implants have led to concern over implant
loosening. Aseptic loosening

in
hip and knee prostheses is well documented and is thought to be induced
by wear particles that stimulate giant cell–mediated bone resorption (31,35,36,38,57,67).
A recent study characterized the particulate debris generated in the
total shoulder arthroplasty when radiographic evidence of osteolysis
was present. The particles were larger and more varied in shape than
those in total hip arthroplasties that required revision (70). No information is currently available regarding the immune response generated by particles of this size.

Clinical series of shoulder implants have demonstrated
few revisions for symptomatic component loosening, particularly on the
humeral side. Indeed, aseptic loosening of cemented humeral components
is extremely rare. In their 1982 series, Neer et al. (56)
reported only two patients with such loosening among 94 total shoulders
with radiographic evidence of humeral loosening. Although both were
symptomatic, neither underwent revision arthroplasty. Information
regarding press-fit humeral stems is more varied, but the rate of
loosening would appear to be, if anything, higher than that of cemented
stems. Cofield (16) reported that less than 1%
of cemented humeral stems demonstrated radiographic evidence of
loosening, compared to 12% of uncemented stems. Several stems with
textured surfaces have been introduced to facilitate bone adherence to
the prosthesis, but long-term follow-up is lacking (16,22,47).
A new problem introduced by these designs is the difficulty of removing
a malaligned or infected ingrowth stem that is not loose.

Most concerns for long-term loosening have centered on
the glenoid component. Lucent lines about the glenoid component have
been reported in 30% to 83% of cases (18,53),
with progression occurring in 12% to 16% of cases. The clinical
significance of these radiolucent zones is unclear. Several authors
have attributed glenoid lucencies to technical errors occurring at the
time of glenoid placement (7,12,63,64). Neer (53)
believed that radiolucencies about a glenoid component could be caused
by a number of factors, including inconsistent radiographic technique,
variable density and strength of glenoid bone stock, stress shielding
caused by the glenoid component, and disuse osteoporosis. Both Neer (53) and Brems (7)
reported higher incidences of radiolucent lines in cases performed
earlier in their series, and that it might be related to inexperience
with the technique in earlier cases. Clear radiographic evidence of
humeral loosening is pres- ent when a completely circumferential
radiolucent line of greater than 1 mm width is seen to progress at the
bone–cement/prosthesis interface, or when the prosthesis changes
position on serial plain radiographs.

Revision for symptomatic glenoid loosening (Fig. 103.5) remains uncommon (2.8%) (7).
In a series of 70 total shoulder replacements followed for 5–11 years,
complete lucent lines of 1.5 mm or more were found in one third of the
glenoid components, but only three components required revision (32).
However, unlike the situation with a lower extremity, the use of one
arm may be decreased by compensating with the other arm; thus, the rate
of revision may underestimate the true rate of component loosening.

Neurologic injury after total shoulder replacement is,
fortunately, a rare occurrence. The rates of injury are much lower than
the 0% to 3% reported for total hips and knees. The cause of nerve
injury is most often traction or compression, but accidental
transection is possible. Injuries to the axillary, musculocutaneous,
ulnar, and radial nerves have been reported (18,42,72,73). Most injuries are neuropraxias that resolved with time. Lynch et al. (42)
reviewed 417 patients undergoing total shoulder ar- throplasty and
found that 18% of shoulders had a neu- rologic deficit after surgery.
Most of those deficits involved the upper and middle trunks of the
brachial plexus. Of the 18 shoulders affected, four had lesions that
interfered with their rehabilitation. This series identified the long
deltopectoral approach, use of methylmethacrylate, and shorter
operative times as risk factors for injury.

Patients undergoing total shoulder arthroplasty should
have a careful neurologic evaluation postoperatively. If a deficit is
identified, institute physical and occupational therapy to maintain
joint mobility. When indicated, provide support with splints. If there
is no clinical evidence of neurologic recovery in 4–6 weeks, obtain
electromyography and nerve conduction studies to document the extent
and type of injury and to provide a baseline for monitoring improvement
in nerve function. If no improvement is seen in 3 months, consider
exploration of the nerve for neurolysis or nerve grafting.

Fracture following prosthetic shoulder replacement is rare, occurring in 1.6% to 1.8% of patients (6,74).
Boyd et al. reported seven patients with humeral shaft fracture
following shoulder arthroplasty. Two were treated nonoperatively: One
developed a painful nonunion but refused surgical intervention, and the
other attained union but required revision for a loose glenoid
component. Five patients underwent operative intervention when
conservative treatment failed after an average of 8 weeks of treatment.
Two of those patients developed radial nerve palsies following the
fracture, which prompted early surgical exploration. All fractures
united.

Wright and Cofield (74) classified periprosthetic fractures according to their morphology:

Fractures that fell into category C behaved like closed
humerus fractures, and those in categories A and B had a greater risk
of nonunion. The authors recommended surgical treatment for type A
fractures, operative treatment for type B fractures when satisfactory
alignment could not be achieved with closed means or if no evidence of
union was present by 3 months, and nonoperative management for type C
fractures. Groh et al. (30) reported on four
postoperative fractures that occurred an average of 14 months after
arthroplasty. They were treated with a fracture brace and early
range-of-motion exercises. At an average of 9 weeks, all fractures had
united.

Given the paucity of data, it is difficult to establish
reliable rules for the treatment of these fractures. We prefer
conservative treatment with immobilization in a functional brace if the
fracture is well aligned and has a long oblique pattern with a stable
stem. However, if the fracture is unstable or the humeral stem loose,
we recommend open reduction and internal fixation.

In an age of progressive improvements in shoulder ar- throplasty and other reconstructive procedures, the indications

for shoulder arthrodesis have become limited. However, glenohumeral
arthrodesis remains an important and valuable treatment option for
restoring useful function in situations where arthroplasty either has
failed or is not indicated. Whereas the optimal results seen with
glenohumeral arthrodesis do not compare with those seen with
conventional arthroplasty, it should be noted that when arthrodesis is
indicated, significant and meaningful improvement in quality of life
can be obtained. Although arthrodesis is performed uncommonly, numerous
reports show that a functional fusion can be obtained reliably with an
acceptable complication rate (2,17,33,40,51,59,60,61 and 62,65).

Arthrodesis is generally the treatment of choice for a
flail shoulder. Paralytic shoulder secondary to anterior poliomyelitis,
upper-trunk brachial plexus lesions, and axillary nerve palsy in the
presence of a nonfunctional rotator cuff are indications for
arthrodesis. The flail shoulder is a particularly strong indication for
an arthrodesis when good elbow and hand function and strong
periscapular musculature are present (44,45,59,61,65).
In addition to limited function, patients with flail shoulders may have
significant pain associated with inferior luxation of the humeral head.

A flail elbow or hand ipsilateral to a flail shoulder is
not necessarily a contraindication to glenohumeral arthrodesis. A
concurrent procedure to restore elbow flexion can be performed, either
staged or simultaneously. Indeed, an arthrodesis can significantly
improve the results of an elbow flexorplasty by stabilizing the arm in
reference to the axial skeleton and by placing the arm in a more flexed
position. For the completely flail upper extremity, a shoulder
arthrodesis can be performed with an elbow flexorplasty, to improve
quality of life, even with a dysfunctional hand (59,61,75) (see Chapter 59, Chapter 60 and Chapter 178).

Axillary nerve palsy as an indication for shoulder
arthrodesis is controversial. Most agree that deltoid dysfunction alone
is not a sufficient indication—rotator cuff dysfunction is generally
required in addition. The restoration of rotator cuff and/or deltoid
function with muscle transfers is difficult and generally requires
multiple surgeries.

Before considering an arthrodesis, consult a hand and
upper extremity surgeon; for example, in the case of an axillary nerve
laceration, cable grafting can give surprisingly reliable functional
return. Other nerve grafting and muscle transfer procedures are
available that may make shoulder fusion unnecessary (see Chapter 59, Chapter 60 and Chapter 178).

Septic arthritis was historically the primary indication for glenohumeral arthrodesis (58,68).
While it remains an important indication, arthroplasty has become more
accepted. The surrounding musculature is often preserved and functional
despite prior sepsis, making it amenable to prosthetic reconstruction (13). Thus, if the septic arthritis can be eradicated, an arthroplasty is preferred.

Failed prosthetic arthroplasty is a rare but accepted indication for arthrodesis when seen in conjunction with loss

of both deltoid and rotator cuff functions. When function of the
deltoid and/or rotator cuff muscles is preserved, a revision
arthroplasty is preferred. Anterior superior instability, however, may
best be treated with arthrodesis. In these people, the rotator cuff as
well as the anterior deltoid is usually dysfunctional. Incompetence of
the coracoacromial arch can result in migration of the prosthesis
anteriorly and superiorly into a subdeltoid, almost subcutaneous,
location. In addition to significant dysfunction, these patients often
have severe pain. Because periscapular musculature is still intact, a
successful arthrodesis could alleviate the problem and provide
significant functional gains. However, arthrodesis after failed
shoulder arthroplasty is difficult because of the loss of proximal
humeral bone, in particular the head of the humerus.

Failed reconstructive procedures for instability have
been an indication for glenohumeral fusion. More recently, arthrodesis
does not appear to be a good alternative for failed shoulder
instability reconstructions. Outcome data suggest that many people who
obtain arthrodesis in this situation continue to be unhappy,
persistently perceiving instability in the glenohumeral joint despite a
radiographic fusion (59). Additionally, pain
from the glenohumeral joint is often persistent despite any obvious
objective reasons. Before exploring arthrodesis as a surgical
alternative, exhaust every other treatment option. Assess the patient’s
psychological makeup carefully.

For shoulder dysfunction and pain secondary to rotator
cuff tear, osteoarthritis, or rheumatoid arthritis, advances in
conventional reconstructive procedures have made glenohumeral
arthrodesis relatively contraindicated. The deciding factor in each of
these three clinical scenarios is the function of the rotator cuff and
deltoid. Loss of these two important muscle groups makes arthrodesis
the best choice in most patients.

Glenohumeral arthrodesis is often the procedure of
choice following removal of periarticular shoulder tumors when the
proximal humerus can be preserved. Because the rotator cuff and the
deltoid are often removed as part of the resection, stability is
difficult to achieve without arthrodesis. After aggressive bone
resection, vascularized bone graft or allograft reconstruction may be
necessary to achieve an arthrodesis (11,41,43).

Unless a rigid external orthosis is necessary because of
inadequate internal fixation, use a soft, removable abduction pillow.
It will unload the weight of the arm and is easily removed for hygiene;
thus it is better tolerated than the orthosis for the 8- to 12-week
period that is often necessary to obtain a solid fusion. Once clinical
or radiographic fusion has been obtained, place the patient on a
rehabilitation program for periscapular strengthening. Give particular
attention to strengthening the trapezius, rhomboids, and serratus
anterior. Throughout the postoperative period, employ elbow
range-of-motion and hand exercises in conjunction with shoulder
rehabilitation.

Glenohumeral fusions are often performed in osteopenic
bone, where a stress riser distal to the plate is dangerous. It is not
surprising that a high fracture rate can occur in this area. Cofield
and Briggs (17) reported a fracture rate

of 15%. Because this complication often occurs with only minor trauma,
closed treatment is often successful. If a glenohumeral arthrodesis is
established and internal fixation of a distal humerus fracture becomes
necessary, then removal of the proximal plate is recommended for the
future to prevent a stress riser.

Infection has been a relatively rare complication. Treat it as you would infected fractures or implants elsewhere (see Chapter 132 and Chapter 135).
In general, avoid removal of in- ternal fixation devices until you
achieve an arthrodesis. Following this, remove all internal fixation
and perform repeat irrigation and debridements and treat with
intravenous antibiotics as necessary to eradicate the infection. If the
internal fixation has loosened in the presence of infection, remove it
and convert to external fixation augmented by an external orthosis.

Clinical union of the arthrodesis site has been a
reliable measure of outcome for glenohumeral arthrodesis. Nonunion is
rare. In Cofield’s (17) series of 71 shoulders, only 4% went to nonunion. Richards et al. (59) reported similarly high fusion rates.

In general, patients with arthrodesis can tolerate a
large range of flexion and rotation positions. Glenohumeral abduction,
in contrast, is poorly tolerated in adults at angles greater than 40°.
This is because of the significant scapular winging and trapezial
tension that occur when the arm is resting at the side. When
malposition causes significant pain or dysfunction, an osteotomy may be
required. It is generally performed in the humerus distal to the
glenohumeral fusion.