Forty-eight patients were admitted to our institution for the treatment of
an infection at the site of a total knee arthroplasty from 1999 to 2003.
Twelve of these patients eventually underwent a knee arthrodesis. In all
twelve cases, a long intramedullary nail was used for fixation. All procedures
were performed by the senior author (N.G.S.). Eleven patients were followed
for a minimum of two years (range, two to six years; mean, 4.1 years), and one
patient died eighteen months after the arthrodesis from unrelated causes.
Hospital records and serial radiographs of all patients were reviewed to
evaluate patient status and the results of the operation.
The patients included seven women and five men who had a mean age of
sixty-eight years (range, sixty-one to seventy-seven years) at the time of the
arthrodesis. The number of previous procedures per patient, comorbidities, and
bacteria causing the infections are shown in
Table I.
The treatment goals were the eradication of infection and the achievement
of a solid knee fusion. All patients were carefully assessed monthly until
fusion occurred, both for healing of the arthrodesis site and for signs of
infection.
The C-reactive protein level, erythrocyte sedimentation rate, and white
blood-cell count were determined at each visit, followed by a thorough
clinical evaluation.
Clinical and radiographic union was defined as the ability to walk without
pain with no tenderness at the arthrodesis site, combined with the appearance
of circumferential bridging callus on both anteroposterior and lateral
radiographs14,15.
In order to assess functional outcome, the Western Ontario and McMaster
Universities Osteoarthritis Index
(WOMAC)16 was
determined before the removal of implants and again at the time of the final
office visit. At the same time, the limb-length discrepancy was recorded.
Surgical Technique
In all patients, the procedures were carried out through previously
utilized anterior approaches. All patients underwent the same staged
procedure. After the removal of implants, thorough soft-tissue
débridement, and removal of infected necrotic bone, antibiotic-loaded
polymethylmethacrylate spacers were prepared and inserted. We used 6 g of
tobramycin and 6 g of vancomycin per 40 g of polymethylmethacrylate.
In three patients, débridement and exchange of spacers had to be
carried out twice because of evidence of recurrence of infection. On the basis
of cultures and the results of sensitivity testing, appropriate intravenous
antibiotics were administered for a period of four to six weeks, followed by
an observation period of four to eight weeks.
Knee arthrodesis was not performed until clinical and laboratory data
suggested eradication of infection. These data included a normal C-reactive
protein level and erythrocyte sedimentation rate after the discontinuation of
intravenous antibiotics as well as evidence that the wound had healed and had
remained quiescent.
At the time of the proposed arthrodesis, the spacers were removed.
Débridement of the soft tissue and the medullary canals was performed
and the bone end cuts were freshened, followed by thorough irrigation with
pulsed lavage. Tissue samples were obtained from multiple sites (including the
medullary canal of both the tibia and the femur) for frozen-section analysis.
If the polymorphonuclear leukocyte count was less than ten per high-power
field17, knee
arthrodesis was then carried out.
In the first five cases, parallel tibial and femoral bone cuts were made in
order to achieve maximum contact of viable bleeding cancellous bone. An effort
was made to make the bone cuts parallel, with no or minimal gaps. Some knees
required substantial shortening to achieve maximum bone contact, viability,
and stability.
Because we used a curved nail with an 18° radius in the sagittal plane,
achievement of maximum contact was technically difficult at the arthrodesis
site. Therefore, in the next seven cases, we used a different technique. A
pair of matched convex-to-concave reamers were used to prepare the tibial and
femoral surfaces. This technique created a ball-and-socket configuration that
provided large contact areas of viable bleeding bone (Figs.
1-A and 1-B), improved
stability, and maintained congruency after the insertion of the bowed nail
(Figs. 2-A and 2-B). The tibia
was reamed to its maximum diameter, and then an intramedullary nail (Biomet,
Warsaw, Indiana) with a diameter of 1.0 to 1.5 mm less than the diameter of
the last reamer was chosen. The femur was then prepared to the same size as
the last reamer used on the tibia, and the long intramedullary nail was
inserted and locked distally with two cross-locking screws. Finally, to ensure
apposition and compression, the nail was hammered in reverse and then was
statically locked proximally.
Two patients with excessive scarring due to multiple skin incisions or
previous sinuses had a local flap transfer for wound coverage at the time of
arthrodesis. In the remaining ten patients, wound closure over a drain was
carried out in a routine fashion.
All patients were allowed to walk with an aid on the second postoperative
day, with partial weight-bearing as tolerated.
Mean values and standard deviations for the time to fusion, the WOMAC
score, and limb-length discrepancy were calculated with use of a standard
computer spreadsheet program (Excel; Microsoft, Redmond, Washington).
Institutional review board approval was obtained, and all patients were
informed that data concerning their cases would be submitted for
publication.
Ten of the twelve patients had achievement of a solid radiographic and
clinical fusion, were able to walk with no pain, and had no evidence of
instability or recurrent infection (Figs.
2-A and 2-B). One patient
underwent an above-the-knee amputation. Despite débridement and
exchange of the intramedullary nail, the infection persisted. Two months after
the index arthrodesis, the patient was diagnosed with a lymphoma and was
managed with chemotherapy and radiation therapy. This patient died eighteen
months after the arthrodesis from causes that were not related to the index
procedure.
In one patient, who was morbidly obese, the knee was assessed as being
clinically and radiographically fused at 5.5 months. The patient was able to
walk without pain at the time of the sixteen-month follow-up visit. Three
years after the index procedure, the patient was readmitted because of the
sudden onset of knee pain and breakage of a nail at the arthrodesis site.
There was no evidence of recurrent infection. The patient refused additional
surgical intervention and was walking with a walker and an extension brace at
the time of the latest follow-up.
In one patient, the nail was removed because of a suspected low-grade
infection at one year. Reaming and irrigation was carried out. Reinsertion of
the nail was not necessary because a solid fusion had already been achieved.
There were no other complications or reoperations in the present series.
The average time to union was 5.5 months (range, 3.5 to 12.4 months). The
average time to union for the seven patients who underwent preparation of the
bone ends with convex-to-concave reamers was shorter than that for the
remaining five patients (4.3 compared with 7.4 months).
The mean WOMAC score improved from 41 points (range, 34 to 47 points)
before removal of the prostheses to 64 points (range, 40 to 72 points) two
years postoperatively.
The average postoperative limb-length discrepancy was 5.5 cm (range, 3.1 to
8.0 cm). Two patients walked with an insole, whereas the rest of the patients
required external shoe-lifts. Of the ten patients with a successful
arthrodesis, one used a walker, seven used a cane (five on a regular basis and
two on an occasional basis), and two walked without any aid at the time of the
latest follow-up. The times to union, WOMAC scores, and limb-length
discrepancies are shown in Table
II.
The indications for the conversion of a total knee arthroplasty to an
arthrodesis because of infection are not clear, and the decision-making
process is difficult. The surgeon must keep in mind that the functional
results after multiple total knee arthroplasty revisions are often
poor4,5.
Patients with compromised general health or comorbidities such as diabetes
mellitus, smoking, obesity, poor local wound conditions, failure of the
extensor mechanism, and/or infection with virulent, antibiotic-resistant
bacteria are at a higher risk for the recurrence of infection and failure
after reimplantation of a total knee
replacement2,18,19.
For instance, successful treatment of an infection at the site of a total knee
arthroplasty in the presence of methicillin-resistant Staphylococcus
aureus is known to be very
difficult20. Both
of the treatment failures in the present series occurred in patients who had a
history of either immunosuppression or diabetes and obesity and the presence
of methicillin-resistant Staphylococcus aureus.
In patients with a failed total knee arthroplasty, external fixation with a
variety of devices and intramedullary nailing have both been
employed21-23.
Damron and McBeath, in a review of fifty-six knees, reported a fusion rate of
94% in patients managed with intramedullary rods compared with a 64% rate in
patients managed with external
fixators24. In
general, fusion rates in studies in which external fixators were used were
inferior to rates in studies in which intramedullary nails were
used3,7,11,21,24-30.
In a review of the literature,
Wiedel10 reported
that external fixation devices were associated with fusion rates ranging from
43% to 71% in five series ranging in size from seven to seventy-one
patients9,13,31-33
and that intramedullary nailing was associated with success rates of 83% to
100% in nine series ranging in size from nine to twenty-one
3,11,28-30,33-37.
Intramedullary nails provide greater
stability38, avoid
pin-track infection, allow faster weight-bearing, and generally are better
tolerated than external fixators
are3,4,39,40.
Several authors have reported good results in association with the use of
intramedullary nailing after staged eradication of
infection7,35.
One-stage procedures have been used for the treatment of infection in knees
with
prostheses29,30,41,
but the results of two-stage procedures have been reported to be better in
terms of both union rate and healing
time11,42.
In the present series, union was achieved in ten of twelve patients. We
believe that two factors contributed substantially to these favorable results.
First, every effort was made to ensure that the infection was eradicated by
the time of the arthrodesis. Second, maximum contact and compression of viable
cancellous bone ends was the ultimate intraoperative goal. Rigid fixation,
compression, and an adequate biologic environment are known to be key elements
for
bone-healing12,13,21,32,40.
Charnley believed that the high success rate that he achieved in association
with knee arthrodeses (99%; 169 of 171) was due to two well-coapted surfaces
of cancellous bone with intact
circulation12.
In seven of our patients, a pair of matched convex-to-concave reamers were
used to provide bone-contact areas with a ball-and-socket configuration.
Large, bleeding surfaces were prepared in this manner (Figs.
1-A and 1-B). The
convex-to-concave configuration allowed maximum contact even in knees in which
nail insertion necessitated alterations of the knee alignment. Although our
numbers are too small to allow us to draw solid conclusions, none of these
seven patients had a complication and the mean time to fusion was shorter than
that for the first five patients in the present series, who did not have this
treatment (4.3 compared with 7.4 months).
We believe that a staged procedure provides a better environment for
bone-healing because eradication of infection allows the bone and soft tissue
to revascularize. We recommend the use of a curved nail with an 18° radius
of curvature in the sagittal plane and aggressive bone resection in order to
achieve maximum contact and stability of fixation. Substantial shortening of
the affected extremity following knee fusion has been reported in the
literature8,25,30,43,44.
In our series, the average limb-length discrepancy was 5.5 cm (range, 3.1 to
8.0 cm).
Most of our patients were able to walk with an aid and were using
shoe-lifts at the time of the last follow-up. The mean postoperative WOMAC
score was 64 points (range, 40 to 72 points), and the patients with a solid
fusion reported a stable, painless extremity with difficulty climbing stairs
and when sitting in cars and airplanes.
We believe that arthrodesis can provide a painless, functional limb in
patients who are physically and psychologically compromised following multiple
failed surgical interventions. We agree with Conway et al., who stated that
the overall outcome of a successful arthrodesis is better than the outcome of
an amputation or a poor
revision2. In the
present series, the eradication of infection before nail insertion and the
preparation of large congruent contact surfaces of viable bleeding cancellous
bone resulted in a high success rate and favorable functional results. We
believe that conversion of a total knee arthroplasty to an arthrodesis should
be considered for the treatment of infection, particularly in patients who
have had multiple operations and in those who are medically compromised.
?