Two-stage revision arthroplasty1-3,9,12,17,20,26,28 appears
to be the most successful treatment for late hematogenous, deep
infection. The first stage consists of irrigation and d衲idement
of the infected joint with removal of the prosthesis. Often, an
antibiotic-loaded cement spacer is used to increase available levels
of antibiotics and to prevent contracture of the joint space. The
patient is subsequently treated with a course of intravenous antibiotics
(usually six weeks) and then is returned to the operating room for
implantation of the prosthesis. This strategy has proven reliable
in the treatment of infections at the sites of total knee arthroplasties,
with successful eradication of 80 to 100 percent of the infections
in series of twenty to sixty patients15,17,26,28.
Despite this success, between 3 and 20 percent of patients managed
with two-stage revision have a recurrence of the infection. In most
series reported in the literature, the prosthesis was reimplanted without
the performance of additional cultures of specimens from the joint
before the discontinuation of antibiotic therapy26,28.
In our study, we examined the role of preoperative cultures performed
at least four weeks after antibiotic therapy had been discontinued.
Patients with positive cultures were treated with more irrigation and
d衲idement and a second course of antibiotics; revision was performed
only when preoperative cultures were negative. The goal of our study was
to see whether this approach reduced the morbidity and increased
the successful outcome of treatment of infection at the site of
a total knee arthroplasty when compared with a standard method of
two-stage reimplantation without culture before the reimplantation.
Sixty-nine consecutive patients with a deep infection following
total knee arthroplasty that was treated with two-stage revision
arthroplasty between January 1, 1989, and December 31, 1993, were
prospectively followed. The deep infections developed from one to
ninety-seven months after the index knee arthroplasty. Demographic
information such as preoperative history, operative findings, and
postoperative course were obtained from hospital and outpatient
records and recorded. In addition, preoperative and postoperative
radiographs, nuclear medicine scans, laboratory findings, and culture
results were reviewed. Anteroposterior and lateral radiographs were made
annually to check for progressive radiolucencies or migration of
the prosthesis. The patients were divided into two groups depending
on the treatment protocol, which was selected solely on the basis
of the preference of one of the two senior surgeons (M. A. M. or
D. S. H.) and not on the basis of the severity of the infection
or any other factors. The patients in Group I were managed by one surgeon,
and the patients in Group II were managed by the other surgeon.
The objective clinical rating system of the Knee Society14 was used
to follow the patient's clinical course, and the radiographic scoring
system of the Knee Society6 was used to assess postoperative radiographs.
Study Groups
Group I consisted of thirty-five knees treated with removal of
the components, irrigation and d衲idement of the joint, six weeks
of intravenous antibiotics, and reimplantation of a total knee prosthesis.
Group II was composed of thirty-four knees treated with removal
of the components, irrigation and d衲idement, six weeks of intravenous
antibiotics, and cultures of material obtained from one knee-joint
aspiration performed four weeks after completion of the antibiotic
therapy. If the culture was negative after two weeks, the patient
underwent reimplantation. If the culture was positive, the patient
was managed with repeat irrigation and d衲idement of the joint
and then was reentered into the protocol. The operative procedure
was similar for all of the patients. The joint was extensively debrided
of all grossly infected tissue; the d衲idement included a complete
synovectomy. The components were removed, and the bone surfaces
were aggressively curetted. A cement spacer impregnated with two
grams of tobramycin (one gram per pack of cement) was placed loosely
into the joint, and the wound was closed over two suction drains.
For all of the patients, intravenous antibiotics were chosen after consultation
with the infectious disease service. Antibiotic levels were carefully
monitored, and a bactericidal titer of at least 1:8 was maintained
for the six-week course. As mentioned, the treatment protocol was
determined preoperatively solely by the preference of the attending
surgeon.
Demographics
Group I consisted of seventeen men and eighteen women. The mean
age of the patients at the time of infection was sixty-four years
(range, forty-six to eighty years). The mean time from the index
arthroplasty until the diagnosis of infection was thirty-three months
(range, two to ninety-seven months). The mean duration of symptoms
until diagnosis was forty-eight days (range, one to 138 days). There
were thirty-one primary and four revision arthroplasties. All knees
had a late deep infection proven on culture. The mean duration of follow-up
was sixty-eight months (range, thirty-six to 114 months).
Group II consisted of sixteen men and eighteen women. The mean
age of the patients at the time of infection was sixty-nine years
(range, fifty-six to eighty-two years). The mean time from the index arthroplasty
until the diagnosis of infection was thirty-five months (range,
one to eighty months). The mean duration of symptoms until the diagnosis
was fifty-six days (range, four to 322 days). There were thirty
primary and four revision arthroplasties. All knees had a late deep
infection proven on culture. The mean duration of follow-up was fifty-eight
months (range, thirty-six to ninety-one months).
There was no significant difference between the two groups in
terms of the male:female ratio, age, preoperative Knee Society score,
time from the index arthroplasty until the diagnosis of infection, types
of infectious organisms, duration of symptoms before diagnosis,
number of previous revisions, or duration of follow-up (p > 0.05).
Treatment
At the time of reimplantation, all patients had careful d衲idement
of the affected joint, removal of the cement spacer, and reinsertion
of a knee prosthesis. All prostheses were fixed with cement impregnated
with antibiotics (tobramycin, one gram per pack). If a patient was
found to be free of infection at the time of reimplantation (with
negative gram stains and specimens that were completely free of
polymorphonuclear leukocytes), no antibiotics were given. If a knee
was found to have a persistent infection (polymorphonuclear leukocyte counts
of ten or more cells per high-power field), it was debrided again,
another course of antibiotics was given, and the patient was reentered
into the protocol. The patients were followed with yearly physical
and radiographic examinations, and data from the most recent follow-up
evaluation were recorded as the final results in this report. The
reimplantation was considered a success when the objective Knee
Society score was 75 points or more and the patient was free of
infection at the time of the most recent follow-up.
Data Analysis
The primary issue was whether performing preoperative aspiration
and culture had a significant effect on the rates of reinfection
and successful clinical outcomes following two-stage revision arthroplasty.
The null hypothesis suggested that performing preoperative aspiration
and culture would have no effect on the rate of reinfection or on
the final clinical outcome. Differences in frequencies of infection
between Groups I and II were analyzed with the chi-square test.
Differences in Knee Society scores were analyzed with the two-tailed Student
t test.
Group I (Tables I and II)
Of the thirty-five patients in Group I, thirty (86 percent) had
no recurrence of infection. These patients all had a well functioning
prosthesis and were free of infection at the most recent follow-up evaluation.
Of the five patients (14 percent) who had a recurrence of infection,
one was managed with repeat delayed exchange arthroplasty. At the time
of the last (fifty-one-month) follow-up, this patient had retained
the second revision prosthesis, was free of apparent infection,
and had a Knee Society score of 82 points. Three of the other patients who
had recurrent infection had successful arthrodesis of the knee24. One patient continued with chronic
antibiotic suppression and did not undergo removal of the revision
prosthesis. The mean time to reinfection was four months (range,
two to six months). All reinfections were from the organism that
had caused the initial infection, although in three of the five
patients the sensitivities of the organism to antibiotics had changed.
The mean postoperative Knee Society score associated with successful clinical
outcomes was 86 points (range, 80 to 95 points). The mean knee flexion
at the time of the latest follow-up in the patients with initial
eradication of the infection was 95 degrees (range, 30 to 122 degrees),
and the mean extension was -5 degrees (range, 0 to -10 degrees).
Group II (Tables I and III)
Of the thirty-four patients in Group II, thirty-one (91 percent)
had a negative culture after the course of antibiotics had ended.
These patients were managed with second-stage total knee arthroplasty
after the negative culture was obtained. Thirty of the thirty-one
patients were free of infection and had a functioning prosthesis
at the time of the last follow-up. One patient had a recurrent infection
after the second-stage revision; the infection was caused by Staphylococcus
aureus; this was different from the initial infecting organism,
which was a streptococcus species. This patient eventually had a
knee arthrodesis24. The three
patients (9 percent) who had a positive culture were managed with
six weeks of intravenous antibiotics after more operative d衲idement. All
three had the same infecting organisms as they had had preoperatively,
although the sensitivities to antibiotics had changed. The three
patients subsequently had a negative culture and a successful revision
arthroplasty.
The mean postoperative Knee Society score for all of the knees
in Group II was 87 points (range, 64 to 98 points). At the time
of the final follow-up, the mean knee flexion in the patients with
initial eradication of the infection was 98 degrees (range, 63 to
130 degrees), and the mean extension was -4 degrees (range, 0 to
-12 degrees).
There were no signs of loosening, on the most recent postoperative
radiographs, of any of the surviving prostheses in either group.
No prosthesis had migrated compared with the position seen on postoperative
radiographs. Three revision femoral components showed partial, nonprogressive
radiolucent lines anteriorly (in zone 1) on the lateral radiograph.
Two prostheses had partial, nonprogressive radiolucent lines under
the medial tibial tray (in zones 1 and 2) on the anteroposterior
radiograph.
The optimum treatment for an infection complicating total knee
arthroplasty continues to be controversial. An ideal treatment would
entail a minimum of surgical procedures, an acceptable level of
joint function during the treatment, and a high rate of success.
Each method represents a tradeoff between the success rate and the
morbidity of the treatment itself. Protocols that include retention
of the affected components seek to reduce the need for additional
surgical treatment and the period of disability. These procedures
include arthroscopic irrigation and d衲idement8,25,
open irrigation and d衲idement5,18,22,23,
as well as one-stage revision arthroplasty7,10,11.
Unfortunately, the high rates of reinfection seen with these methods
have been disappointing.
Two-stage revision arthroplasty is the most commonly reported
alternative. This method involves removal of the prosthesis followed
by at least six weeks of antibiotics. Studies of two-stage revision have
demonstrated the most consistent and successful results. Windsor
et al.28 reported a success rate
of 97 percent (thirty-two) of the thirty-three knees treated. Borden
and Gearen4 reported success in
ten of eleven knees treated with two-stage revision. Even with these
uniformly successful results, reinfections do occur. The most likely
source of reinfection is bacteria retained in the joint following
the first-stage d衲idement. Bacterial resistance, inadequate d衲idement,
or a new infection may account for persistence of infection in some
cases.
Rationale for Prerevision Cultures
Despite the superior clinical results obtained with two-stage
arthroplasty in a number of centers, there may be modifications
that can improve this method. Preoperative cultures may be a sensitive way
of detecting persistent infection. A number of recommendations have
been made regarding these cultures16.
Insall et al.13 recommended that
antibiotic therapy be stopped and then cultures be performed before
revision arthroplasty. Other authors have recommended that cultures
be performed while the patient is receiving antibiotic therapy or
have believed that there should be no aspiration at all before a
revision procedure26-29. Any joint
that becomes reinfected following revision may have harbored bacteria
that were suppressed by the antibiotic course. These joints show few
or no clinical signs of continued infection, and intraoperative
sampling may not be helpful because of low bacterial counts. Allowing
four weeks off antibiotics before culture of specimens from the
joint gives any bacteria present an opportunity to proliferate and
potentially to be detected by culture of the aspirate. This technique
also allows a period for clinical symptoms from bacterial proliferation,
if present, to manifest themselves. The disadvantage of repeat cultures
is a delay in definitive operative treatment as well as the prolonged
morbidity associated with a minimally functioning knee joint. Many
patients with an infection at the site of a prosthesis are often
those who are least able to tolerate a long period of relative immobility19,21. In the current study, there
were no significant differences between groups with regard to the
most recent follow-up scores of the knees with successful reimplantation;
this finding indicates that the delay in treatment had no long-term
effects.
One weakness in our study is the lack of randomization of the
patients. The patients were managed according to the protocol of
the two senior surgeons, and differences in their approaches may
account for the reported results. However, all of the patients were
managed with similar operative techniques, operating-room teams,
equipment, and assistant surgeons. Since all six recurrences happened
after less than one year of follow-up, the minimum three-year follow-up
used in this study should have been adequate to detect all persistently
infected joints.
Two-stage revision arthroplasty was found to be an effective
method for treating infection after total knee arthroplasty. This
study suggests that prerevision cultures, after the discontinuation
of antibiotic treatment, should be performed in all patients treated
with this procedure. With identification of continuing infection,
the morbidity associated with recurrent infection may be avoided and
the number of surgical procedures may be reduced.