Between June 1975 and March 1991, fifty-seven second-generation
Swanson trapeziometacarpal joint prostheses were implanted in the
hands of forty-five patients by three experienced surgeons (one
of us [R.D.] and two others who have retired). Eight
patients died, one could not be traced, and one refused to participate
in the study. The remaining thirty-five patients (forty-five implants)
were available for clinical follow-up. Of the forty-five prostheses,
thirty-five were implanted in women and twenty-two were placed in
the dominant hand. The mean age at the time of the operation was sixty-one
years (range, 42.5 to 85.0 years). All of the joints were affected
by primary osteoarthritis. The indication for surgical intervention
was long-standing pain that was not responsive to conservative treatment.
The original surgical technique described by Swanson et al. was
used12,13. In six patients, an
additional ligament reconstruction with a strip of the flexor carpi
radialis tendon combined with Kirschner-wire stabilization from the
first to the second metacarpal was performed. Postoperatively, the
hands were immobilized in a cast for six weeks. After the cast and
the Kirschner wire were removed, the patients began functional exercises,
guided by a physiotherapist if necessary. The mean duration of clinical
follow-up was 13.8 years (range, 7.5 to 23.5 years).
Clinical Evaluation
The cases of all thirty-five patients (forty-five implants) were
reviewed by one of us (H.G.J.v.C.), who had not been involved in
the surgical procedures. The same clinical scoring system that was described
in detail in our previous study of the de la Caffinière
prosthesis was used14. In short,
this clinical assessment consisted of four items, pain, function,
movement, and power, which were assigned a maximum score of 10 points
each (Table I).
A total score of 36 to 40 points (the maximum score) was rated as
an excellent result, and 30 to 35 points was considered a good result.
As in the assessment of pain, the subjective result was measured
by the patient on a visual analog scale, with a score of 8, 9, or
10 points representing a good-to-excellent result.
Radiographic Evaluation
Joint degeneration in the twenty-eight thumbs for which preoperative
radiographs were available was staged according to the criteria
of Eaton et al.15. Thirteen hands
(46%) had stage-II joint degeneration; eleven (39%),
stage-III; and four (14%), stage-IV. Standard anteroposterior
and oblique radiographs of the wrist were made at the time of the
clinical review, and they were evaluated for the position, wear,
and deformation of the prosthesis as well as for evidence of osteolysis
around the stem and degenerative bone cysts, indicating silicone-induced
synovitis.
In addition to the clinical and radiographic evaluation, all
complications as well as the indications for and the mode of any
revision operations were obtained from the files.
Complications and Revisions
After the primary operation, three patients had loss of sensory
function of the superficial radial nerve. Two of them had resolution
of the sensory loss within eight weeks, but the third had permanent loss.
Postoperatively, one patient had a deep infection at the site of
a Kirschner wire. The prosthesis was removed, and a ligament reconstruction
with tendon interposition was performed. One patient had mild reflex
sympathetic dystrophy that resolved quickly but ultimately necessitated
resection of the implant followed by tendon interposition because
of persistent pain.
In one patient, symptomatic silicone-induced synovitis developed
more than five years after the primary surgery. Radiographically,
subluxation of the implant had been noted since six months after implantation.
Before removal of the implant, radiographs showed osteolysis around
a deformed and broken prosthesis (Fig. 1). After removal, histological examination
showed synovitis around silicone particles and giant cells with
intracellular silicone debris.
Eighteen prostheses (40%) dislocated. Seven of the dislocations
were diagnosed within three months after surgery; four, within six
months; three, within one year; three, within two years; and one,
within four years. Eight dislocations were painful, and a repeat
operation was performed within a few weeks to one year after the
diagnosis was made. At the time of removal, one of the prostheses
was broken at the junction of the body and the stem. In one exceptional
case, the dislocation was tolerated for thirteen years until a painful
adduction contracture developed and the patient requested a revision.
Of the remaining nine thumbs with a dislocated prosthesis that was
not revised, six had development of an adduction deformity with
secondary hyperextension of the metacarpophalangeal joint at the
time of follow-up.
After revision surgery, two patients had permanent loss of sensory
function of the radial nerve and one had full recovery of sensory
function. One of the two patients who had revision of a Swanson
prosthesis had redislocation within six months. The indications
for and outcome of all revisions are listed in Table II.
Clinical Results
The clinical and subjective results are shown in Table III. Twelve of
the original forty-five prostheses had been revised. Resection of
the Swanson prosthesis had been followed by reimplantation of a
new one in two thumbs and by tendon interposition with or without
ligament reconstruction in five. Five thumbs had had a simple resection
of the implant (Table II).
The results in the thirty-three thumbs that had not had a revision
depended strongly on the position of the prosthesis. Of the nine
thumbs that had a dislocation, only two had a good score on the
clinical examination and only three had a good score on the subjective
evaluation; the less satisfactory scores were mainly due to poor
function and power. Pain relief was reasonable. Of the twenty-four
thumbs that had not had a dislocation, twenty-one (88%) had
a good or excellent score (50% had an excellent score),
both clinically and subjectively. Of these twenty-four thumbs, seven
had radiographic evidence of subluxation but all seven had a good
or excellent result. The mean abduction of the twenty-four thumbs
was 32°, and all but five were able to oppose the tip of the little
finger. In contrast, the mean abduction of the nine dislocated thumbs
was 25°, and only five were able to oppose the tip of the little
finger.
Of the twelve thumbs that had had a revision, four had a good
result clinically and three had a good result subjectively. Again,
the main reasons for the poor results were loss of power and function.
Pain relief was acceptable. The mean abduction of the thumbs was
30°, and eight thumbs could oppose the tip of the little finger.
The results were independent of the type of revision. Furthermore,
the final results of the revisions performed because of dislocation
were no better than those of the thumbs with an unrevised dislocated
implant, but the thumbs were less painful than they had been before
the revision.
Radiographic Results
Follow-up radiographs were available for forty-two thumbs. Of
thirty thumbs with an unrevised implant for which radiographs were
available, eight (27%) demonstrated a dislocation and seven
(23%) had a subluxation. Five prostheses (17%),
including one that was dislocated, were deformed, which was manifested
mainly as loss of height on the ulnar side of the body. Four of
these five thumbs also had bone cysts in the base of the metacarpal
and the scaphoid, indicating synovitis. Of the three patients who
declined radiographic evaluation at the time of the latest follow-up,
one had a dislocation that had been radiographically demonstrated
six months postoperatively. The other two had no clinical evidence
of dislocation or synovitis. Radiographically identified dislocations
were always clinically apparent.
Follow-up radiographs were made after all twelve revisions. One
of the two revision prostheses had redislocated. The body of the
second revision prosthesis was completely disintegrated, and large osseous
defects were seen in the metacarpal base, indicating synovitis.
All of the thumbs that had had a resection of the prosthesis showed
subsidence of the first metacarpal compared with that seen on preoperative
radiographs. The maximum subsidence was 10 mm, in a thumb that had
not had tendon interposition.
In our experience, the most serious problem with the Swanson
silicone trapezium implant has been dislocation. Most of the eighteen
dislocations occurred in the first six months after surgery, when the
soft tissues were still healing. Painful dislocation was the indication
for nine of the twelve revisions. The remaining nine thumbs that
had a dislocation received no additional treatment because the patient
did not want it, although six of them thought that the outcome was
poor. Furthermore, the patient who had redislocation of a revision
prosthesis also did not want additional surgery although the result
was poor. In addition to the dislocations, radiographs revealed
subluxation in seven thumbs. All of these patients had a good result.
The subluxated implants, which were still functioning as interposition
arthroplasties, were considered to be stable after a minimum duration
of follow-up of 7.5 years.
Problems with the stability of the Swanson implant, leading to
dislocation rates as high as 55%, have been frequently
reported in the literature7,12,16-19.
In an effort to improve stability, Swanson and others suggested
various procedures for capsular reinforcement with use of different
tendon slips12,19,20. In 1981,
Swanson et al. described the temporary use of a Kirschner wire placed
through the bone or the prosthesis itself to enhance stabilization;
they later recommended use of an absorbable suture instead of a
Kirschner wire13. Allieu et al.
suggested that the osteotomy of the base of the metacarpal should
be in 10° of varus for a more stable seating of the implant16. Niebauer designed a silicone prosthesis
with Dacron mesh and wires for ingrowth and tying of the implant21. Eaton developed a prosthesis with
a hole through the body, making a firm tenodesis possible22. Stability improved with these methods,
but only the series with the best results13,22 had
a rate of instability of <10%.
In our opinion, two important factors lead to instability of
the Swanson prosthesis. The first is the sacrifice of all ligaments
and capsules attached to the trapezium. The prosthesis functions
as an interpositional arthroplasty and, after suturing, it becomes
surrounded by one newly formed tight capsule augmented with ligament
reconstructions if necessary. However, the capsule is not attached
to this "new trapezium" and therefore never offers
the same stability. The second factor is that the prosthesis makes
the lever arm of the first metacarpal longer. In our series, measures
to improve stability, according to the advice of Swanson et al.13, were thought to be necessary in
six thumbs. Despite these measures, four implants dislocated. Furthermore,
a deep infection developed around a fifth implant, which had to
be removed.
Wear and deformation of the prosthesis and silicone-induced synovitis
were less problematic than stability in our series. Movement of
the implant is not always restricted to its articulation with the scaphoid
because bending often takes place at the junction between the body
and the stem9,16. The latter,
however, can lead to breakage of the prosthesis, which we encountered
twice. One dislocated prosthesis appeared to be broken at the time of
revision, and silicone-induced synovitis was the indication for
one revision of a deformed, broken, and subluxated prosthesis. Of
the thirty-two prostheses (thirty that had not been revised and
two that had been revised) for which radiographs were available,
six (19%) showed some degree of wear and deformation. Five
of the six also had radiographic signs of silicone-induced synovitis,
but there were no clinical symptoms. Pellegrini and Burton suggested
that the two problems of instability and wear form an unresolved
conflict4,7. They found that the
increased stability of the Eaton prosthesis, compared with that
of the Swanson design, led to increased wear, deformation, and breakage.
Silicone-induced synovitis as an inflammatory reaction to small
wear particles rather than to gross deformation has often been discussed
in the literature6-8,17,23-25.
Swanson and de Groot Swanson argued that synovitis is related to
instability9. Creighton et al.
disagreed, as they found high rates of cystic changes with or without
the use of a Kirschner wire10.
Although clinical symptoms of synovitis do not occur very often,
the swelling and inflammation can be very painful and revision can
be difficult because of the soft-tissue damage and the sometimes
severe osteolysis.
In contrast to the good results after various revision procedures
reported by Weilby and Søndorf19 and
by Burton4, our results after
revision were poor. Conolly and Rath also reported unpredictable
results after revision26. As we
have pointed out, the main reason for our poor overall results was
dislocation, which occurred in eighteen thumbs (40%). Half
of these thumbs were painful, which led to revision. Although pain
relief was better after revision, power and function, in terms of
daily activities and movement, were not. The degree of pain was acceptable
in the other nine thumbs, and a revision was not requested. However,
the outcome was poor in most of them because of the loss of power
and function due to the instability of the thumb. A z-deformity
developed in six of these thumbs. In sharp contrast to these poor
results were the good clinical and subjective results seen in twenty-one (88%)
of the twenty-four thumbs that had had no dislocation or revision.
However, because of the multiple dislocations, the overall results,
both clinically and subjectively, were satisfactory for only 60% of
the thumbs.
We are aware of only a few other reports on Swanson arthroplasties
for primary treatment of osteoarthritis of the basal joint of the
thumb with a comparable long-term follow-up. In 1987, Hofammann
et al. reported on a series of eighteen patients with twenty silicone
prostheses, sixteen of which were Swanson prostheses, who were clinically assessed
at a mean of 8.1 years5. Despite
a rate of dislocation of 15%, a rate of wear of 65%,
and a rate of destructive bone changes of 50%, the rate
of patient satisfaction was 90%, mainly because of pain
reduction. In 1988, Sollerman et al. reported on a series of thirty-three patients
(thirty-nine prostheses) who had been followed for a mean of twelve
years11. Ninety percent of the
patients had a good result in terms of pain relief and 62%,
in terms of grip strength. Eighteen percent of the thumbs had a
dislocation, which was positively correlated with pinch-grip weakness
but not with pain. The rate of radiographic wear and cystic changes
was >60%, but these findings did not correlate
with the clinical results. It should be noted, however, that these thirty-nine
cases were those remaining after a positive selection of sixty-six
thumbs that had not had a dislocation in the original series of
100 cases with three years of follow-up as described by Haffajee18 in 1977. In 1991, Creighton et al.
described cystic bone changes but did not mention dislocations at
a mean of 4.3 years after 151 arthroplasties10.
Cystic changes were noted in 56% of the scaphoids and in
74% of the metacarpals. Again, these rates did not correlate
with the 84% rate of patient satisfaction.
It is difficult to compare our results with those in the other
reports because of the differences in assessment. Patient satisfaction
may appear to be somewhat better in the other series than it was
in ours, but in the other series it was based mainly on pain relief.
The poor results in our study were mainly due to the loss of power
and function, but pain relief was quite reasonable even in the thumbs with
a dislocation. However, we think that the objective results in our
series and in the other series are equally distressing, and they
have led us to advise against the use of the Swanson prosthesis.