Once the diagnosis has been established, the variables that must be
considered before treatment is initiated include (1) whether the infection is
superficial or deep, (2) the time that has elapsed between the arthroplasty
and the diagnosis of infection, (3) host factors that may adversely affect the
treatment of the infection, (4) the condition of the soft-tissue envelope
surrounding the knee (specifically, the integrity of the extensor mechanism),
(5) whether the implant is loose or well fixed, (6) the pathogen(s)
responsible for the infection, (7) the physician's ability to provide the
appropriate level of care, and (8) the patient's expectations and functional
requirements.
Goals of treatment of an infection at the site of a total knee arthroplasty
include eradication of the infection, alleviation of pain, and maintenance of
a functional extremity. The six basic treatment options include (1) antibiotic
suppression, (2) open débridement, (3) resection arthroplasty, (4)
arthrodesis, (5) amputation, and (6) implantation of another prosthesis.
Discerning the most appropriate treatment for a given clinical situation is
paramount, as this step in the management process will likely define the
ultimate outcome. Secondary attempts at treating infection are often adversely
affected by progressive scarring, devitalization of the soft-tissue envelope,
development of antibiotic-resistant organisms, and continued bone loss
associated with previous failures.
Antibiotic Suppression
Antibiotic treatment alone should be considered only in extreme situations,
as this strategy will not eliminate deep periprosthetic infection and is
generally associated with a very poor
prognosis25. In a
multicenter study, antibiotic suppression was successful in only forty (18%)
of 225 knees26.
Analysis of several series together revealed that antibiotic suppression was
successful in sixty-two (24%) of 261
knees18. This
treatment should be considered only when all of the following criteria are
met: (1) it is not feasible to remove the prosthesis (usually because of a
medical condition that precludes an operative procedure), (2) the
microorganism is of low virulence, (3) the microorganism is susceptible to an
oral antibiotic, (4) the antibiotic can be tolerated without serious toxicity,
and (5) the prosthesis is not
loose27. The
presence of other joint prostheses should be considered a contraindication for
this
treatment28.
Débridement with Retention of the Prosthesis
Although both arthroscopic and open débridement techniques have been
described, the inability to address the glycocalyx "slime" layer
on the underside of polyethylene components of modular total knee prostheses
without formal removal of the liner, coupled with an increased difficulty in
adequately débriding the back of the knee, has steered most surgeons
away from arthroscopic management. Open débridement may be indicated
for the occasional acute infection in the early postoperative period (Type II)
or for an acute hematogenous infection (Type III) at the site of a securely
fixed and functional prosthesis. Suggested criteria for this treatment include
(1) a short duration of symptoms of infection (less than two weeks), (2)
susceptible gram-positive organisms, (3) absence of prolonged postoperative
drainage or a draining sinus tract, and (4) absence of prosthetic loosening or
radiographic evidence of
infection29.
The results of débridement are difficult to assess as the authors of
existing studies have reported on a plethora of conditions, resulting in a
wide range of findings and mitigating against meaningful comparisons. A
multicenter study demonstrated success with open débridement in thirty
(19%) of 154
knees26. In an
analysis of multiple series, success was found in 140 (31.5%) of 445
knees18. Many of
the reports on débridement included patients with chronic
periprosthetic infection.
Perhaps the most critical factor in achieving success with open
débridement is expeditious treatment. This is particularly true for
Staphylococcus aureus infections, as delaying treatment beyond
forty-eight hours after the onset of symptoms resulted in a significant
decrease in the success rate (relative risk, 4.2; 95% confidence interval, 1.6
to
10.3)29,30.
Our experience has been that the distinction between gram-positive and
gram-negative organisms is inconsequential when one is deciding whether to
attempt a débridement procedure. Others also have challenged the
traditional belief that treating an infection at the site of a total joint
arthroplasty yields less satisfactory results when the organisms are
gram-negative31.
Other factors that have been correlated with better outcomes of
débridement for the treatment of infection at the site of a knee
arthroplasty include the presence of a cemented rather than an uncemented
implant and an unconstrained knee prosthesis rather than a constrained
prosthesis5,32,33.
We currently administer intravenous antibiotics, chosen on the basis of the
results of cultures, for four weeks following débridement and then
decide whether additional chronic oral suppressive antibiotics are also
needed. Implantable pump delivery systems have been developed for local
antibiotic delivery in combination with surgical
débridement34.
At present, this technique has not been rigorously evaluated, and the treating
physician should exercise caution if he or she is planning to use it on a
routine basis.
Resection Arthroplasty
Definitive resection arthroplasty is primarily reserved for patients with
polyarticular rheumatoid arthritis with limited ambulatory
demands35. Although
this treatment option has been shown to effectively eradicate
infection32,36,
the primary disadvantage of resection arthroplasty is the frequent occurrence
of knee instability associated with pain during transfer or walking. The three
fundamentals of the operative technique include (1) initial débridement
and removal of all infected tissue and foreign material, (2) temporary
fixation with pins or sutures to maintain alignment and apposition of the
tibia and femur, and (3) cast immobilization, with weight-bearing permitted,
for at least six months.
A previous study of twenty-six patients, eleven of whom had rheumatoid
arthritis, showed that 89% were free of infection at an average of five years
following resection
arthroplasty37.
Functional results were suboptimal, as only fifteen patients were able to walk
independently and all required walking aids. Only five patients had sufficient
knee stability to walk without external support; eight required a
knee-ankle-foot orthosis, and two used a splint. Although resection
arthroplasty usually results in satisfactory resolution of the infection, most
patients have some pain, knee instability, and a limited ability to walk.
Arthrodesis
Arthrodesis has traditionally been considered the gold-standard treatment
for infection at the site of a total knee arthroplasty; it has an excellent
potential for eradicating infection, alleviating pain, and providing stable
knee function. Although elimination of knee motion can make sitting and other
activities cumbersome, a recent study of thirteen patients with deep infection
demonstrated comparable Oxford scores between patients who had undergone an
arthrodesis and those treated with a two-stage
revision38.
Indications for knee arthrodesis in patients with a failed total knee
arthroplasty include (1) high functional demand, (2) single-joint disease, (3)
young age, (4) disruption of the extensor mechanism, (5) a poor soft-tissue
envelope requiring extensive soft-tissue reconstruction, (6) systemic
immunocompromise, and (7) microorganisms requiring highly toxic antibiotic
therapy or those resistant to conventional antibiotics. Relative
contraindications include (1) bilateral knee disease, (2) ipsilateral ankle or
hip disease, (3) severe segmental bone loss, and (4) amputation in the
contralateral
extremity35.
Inherent difficulties with a knee arthrodesis following an infection at the
site of a total knee arthroplasty include an inability to eradicate the
infection, bone loss, and extremity shortening. Fusion can be achieved in the
face of infection, but infection markedly decreases the success of this
procedure. In one study, the union rate was 62% in patients with controlled
infection and 19% in those with persistent
infection39.
Adequate apposition of vascularized cancellous bone is the most important
factor influencing the success of an arthrodesis. Implants with a hinged
design or with intramedullary stems often induce additional bone loss and are
associated with a lower rate of successful
arthrodesis35,39.
Arthrodesis in patients with severe segmental bone loss can be addressed with
the use of intercalary bone grafts or with adjunctive distraction histogenesis
to correct limb-length
inequality40.
The techniques of knee arthrodesis include external fixation with use of
compression with external pin-frame devices and internal fixation with
intramedullary nailing or with plates. In the presence of infection, many
knees can be treated with one thorough débridement (with excision of
all necrotic tissue, scar tissue, infected tissue, and foreign material)
followed by definitive fixation. In cases with extensive tissue necrosis, it
is advisable to perform serial débridements, to better discern tissue
viability, prior to definite fixation.
The use of an external fixation device in the presence of active infection
has the advantage of leaving no residual foreign implants. Additional
advantages include the potential for adjusting the fixator position and
accessing the soft tissues. Disadvantages include nonrigid fixation, the
potential for neurovascular injury during pin insertion, pin-site
complications, and the need for a second procedure for fixator removal.
Bone-grafting is performed in knees with <50% surface bone contact between
the tibia and femur. In the presence of severe bone loss, the main obstacle
involves the difficulty of maintaining rigid fixation until
union41. Although
the conversion from a single to a double frame and the insertion of sagittal
pins have increased the rigidity of these devices, surgeons should be
cautioned about using single and biplane external fixators in the presence of
severe bone loss as low fusion rates have been documented in this
setting25,35,42,43.
Improved fusion rates (93% to 100%) have been reported with the Ilizarov
device44,45,
but complication rates can be as high as
80%44.
Intramedullary nailing may offer the best method for rigid fixation in the
presence of severe bone loss and the best chance of knee fusion (80% to
100%)41. However,
arthrodesis with an intramedullary nail is not recommended in the presence of
active infection because of the potential for extension of the infection into
the femoral or tibial medullary
canal46. This mode
of treatment has been reported to have good success following the eradication
of
infection47-49,
but the primary disadvantage of using an intramedullary nail in this setting
is that, should infection recur, the infectious process will extend into the
diaphyseal portions of the femur and tibia.
Arthrodesis with plate fixation is another option for obtaining fusion
after resolution of an infection at the site of a total knee arthroplasty.
This method typically employs a two-staged approach. First, the knee is
adequately débrided, with removal of all foreign material, and is
subsequently treated with parenteral antibiotics. On resolution of the
infection, the patient is taken back to the operating room for definitive
fixation. The use of two staggered dynamic compression twelve-hole plates has
been previously
described50
(Figs. 1-A and 1-B).
Occasionally, unilateral plate fixation may be necessary to augment
intramedullary fixation when there is severe bone
loss51. Although
plates provide excellent mechanical fixation, soft-tissue coverage over dual
plates can be difficult and plate-contouring can be arduous.
Amputation
Fortunately, amputation is rarely indicated except for life-threatening
systemic sepsis or persistent local infection associated with massive bone
loss. Amputation is performed in <5% of patients who are treated for an
infection at the site of a total knee
arthroplasty18. The
most common factors leading to amputation include multiple revision attempts
for treatment of chronic infection, severe bone loss, and intractable
pain52. It is
advisable to consider arthrodesis early in the treatment of persistent
infection as repeated attempts at revision surgery may lead to a situation
that requires amputation. Functional ability, especially in elderly patients,
is often limited following amputation. In a study of twenty-three patients
treated with above-the-knee amputation following a failed total knee
arthroplasty, more than half were ultimately confined to a
wheelchair53.
Reimplantation
Although implantation of another prosthesis remains the treatment option
preferred by patients, the potential for improved function must be weighed
against a higher risk of reinfection with the potential for additional bone
loss, the development of multiresistant organisms, and psychosocial
implications. Generally accepted contraindications for insertion of another
prosthesis include persistent or recalcitrant infection, medical conditions
that prevent multiple reconstructive procedures, disruption of the extensor
mechanism, and a poor soft-tissue envelope about the knee
joint18.
Reimplantation following infection is done either as a one-stage or a
two-stage procedure. The success of a direct exchange appears to depend on a
gram-positive status for the infection, use of antibiotic-loaded cement for
fixation of the new prosthesis, and prolonged use of antibiotics after the
revision surgery54.
Use of antibiotic-loaded bone cement appears to be particularly important, as
direct-exchange procedures performed without the use of such cement were
successful in only eleven of nineteen knees in one
study18, whereas,
in an analysis of multiple series combined, the success rate for one-stage
exchanges performed with antibiotic-impregnated bone cement was 131 (74%) of
176 knees18. The
direct-exchange technique may be best reserved for highly selected patients
and performed by surgeons with adequate experience treating infections at the
site of a total knee
arthroplasty54,55.
Delayed reimplantation after administration of intravenous antibiotics
appears to offer better success rates than direct-exchange techniques
(Fig. 2), and delayed
reconstruction is the most commonly accepted approach in North America for
treatment of infections following knee
arthroplasty18.
This treatment entails removal of the prosthesis and all cement with thorough
débridement of bone and soft tissues, followed by four to six weeks of
parenteral antibiotics and then by implantation of a new
prosthesis56. A 97%
success rate was reported following use of this protocol, for treatment of the
original microorganism, for sixty-four knees with an infection at the site of
a replacement57.
When patients who had a reinfection with a different organism were also
considered, the ultimate rate of knees free of infection was 90.6% at an
average of 7.5 years
postoperatively57.
In a recent study in which both the mid-term and long-term outcomes of this
protocol were evaluated in a series of ninety-four patients (ninety-six knees)
who had undergone a two-stage reimplantation because of an infection at the
site of a total knee arthroplasty, the early success was well maintained
throughout the long-term follow-up
period40. The
survivorship with implant removal for any reason as the end point was 90% at
five years and 77.3% at ten
years58. The
survivorship with implant removal because of reinfection as the end point was
93.5% at five years and 85% at ten
years58.
Although a six-week course of intravenous antibiotics administered prior to
reimplantation has had excellent success rates and is currently the most
commonly accepted clinical
standard57,59,
most studies in which the results supported this approach did not include use
of adjunctive treatments such as antibiotic-impregnated cement spacers or
antibiotic-impregnated cement for definitive reimplantation of the
prosthesis57,59.
Excellent success with a shorter duration of intravenous antibiotics (three or
four weeks) has been reported when antibiotic-impregnated cement was used at
the time of
reimplantation24,58,60,61.
Ideally, the duration of antibiotic therapy should be individualized for each
patient on the basis of the virulence of the microorganism, the comorbidities,
and whether antibiotic-impregnated spacers or beads are also being used to
deliver adjunctive antibiotics.
The major disadvantage of staged revision is the interval of time between
the primary resection and the later reimplantation, which is often associated
with pain, difficult mobility, and knee
instability54,55,62.
Other reported shortcomings include the development of scar tissue, shortening
of the extensor mechanism, ligament shortening, and capsular retraction.
Temporary joint spacers have been introduced in an effort to prevent these
problems. The different types of block spacers include a simple tibiofemoral
block63, a molded
arthrodesis
block64, and a
mobile articulating
spacer65,66.
The primary functions of block spacers are delivery of local antimicrobial
agents and maintenance of collateral ligament length. Potential disadvantages
include the presence of a foreign body and bone loss incurred while the
patient is awaiting reimplantation.
The original spacer block was the simple tibiofemoral block, which was
preformed in the shape of a hockey puck or an L-shaped block and then was
inserted into the tibiofemoral space after the cement had polymerized
(Figs. 3-A and 3-B).
Difficulties with this type of block spacer include the inability to match the
shape of the distal part of the femur with that of the proximal part of the
tibia, subluxation of the osseous surfaces off of the spacer surface,
instances of extensor mechanism necrosis, wound breakdown, and progressive
bone loss67.
Some of the difficulties encountered with preformed spacer blocks can be
avoided with use of the molded arthrodesis block. This spacer is fabricated so
that the cement is placed within the knee in a doughy state and is polymerized
within the knee so that the cement can conform to the irregular contour of the
femur and tibia (Figs. 4-A and
4-B). This macrointerdigitation of the cement into bone defects
and the intercondylar notch and extension into the medullary canals and
suprapatellar pouch creates a cement arthrodesis of the knee
joint64. This
stability improves patient comfort and prevents spacer migration and
progressive bone erosion. It is recommended that the knee be positioned in
slight flexion during this procedure to help avoid migration of the cast
during the period of
immobilization64.
The mobile articulating spacer allows the patient to move the knee through
a range of motion during the time period between removal and reinsertion of
the prosthesis. Present designs include a system of molds that incorporate
small metal runners and polyethylene tibial trays so that cement surfaces are
not articulating against each
other65. This
avoids the problem of cement wear debris, which was noted in higher amounts in
knees with this temporary articulating spacer than in knees in which no spacer
was used68. The
theoretical advantages of mobile articulating spacers include the potential
for an improved functional outcome and a better range of motion. To date,
these claims remain unproven, with studies demonstrating conflicting
findings69,70.
Articulating spacers do simplify the surgical exposure at the time of the
reimplantation, and they are particularly helpful for patients who require
simultaneous, bilateral removal of knee replacements because of
infection71.
The use of antibiotic-impregnated cement for fixation at the time of
reimplantation of a prosthesis has also been found to exert a beneficial
effect. In a study of eighty-nine knees treated with reimplantation because of
an infection at the site of a knee arthroplasty, reinfection developed in
seven (28%) of twenty-five knees in which antibiotic-impregnated cement had
not been used for prosthetic fixation compared with only three (5%) of
sixty-four in which it had been
used24.
Analysis of the contributory benefits of antibiotic-loaded cement and the
time delay before the insertion of a new prosthesis revealed several
interesting trends (Fig. 2).
Direct exchange (early reimplantation at less than three weeks after removal)
without antibiotic-loaded cement was successful in only eleven of nineteen
knees, whereas direct exchange with antibiotic-loaded cement was successful in
131 (74%) of 176
knees18. In reports
detailing a two-stage protocol, reimplantation without any antibiotic-loaded
cement was successful in sixty-five (88%) of seventy-four
knees18, whereas
the use of antibiotic-loaded bone cement was successful in 254 (92%) of 277
knees18. These
findings suggest that a time delay, even in the absence of antibiotic-loaded
cement, yields higher success rates than direct-exchange techniques in
conjunction with antibiotic-loaded bone cement.
Thus, the most effective protocol for eradicating infection appears to be a
two-stage reconstruction with use of antibiotic-loaded bone cement for the
final fixation of the component. The dose and combination of antibiotics that
should be used in the cement for both temporary spacers and the final,
reimplanted component, as recommended by the senior author (A.D.H.), are
outlined in Table II. The
systemic safety of such high doses of antibiotics in the temporary spacer was
established in a study in which average total doses of 10.5 g of vancomycin
and 12.5 g of gentamicin mixed in static spacers consisting of an average of
3.4 40-g batches of bone cement (in conjunction with a full six-week course of
intravenous antibiotics) were not associated with an increase in serum
creatinine levels or renal
dysfunction72.
One of the most important issues for both the patient and the surgeon is
when it is safe and appropriate to proceed with reimplantation. Erythrocyte
sedimentation rates would not be expected to normalize in just four or six
weeks; however, the values obtained just prior to reimplantation should have
improved8.
C-reactive protein levels typically normalize by the twenty-first day after
surgery, and levels that remain elevated may suggest persistent
infection8. It has
been suggested that open biopsy or aspiration to obtain specimens for culture
and sensitivity testing should be performed before proceeding with
reimplantation73.
These procedures have not been beneficial in our practice, and we prefer to
use an intraoperative decision-making process based on the appearance of the
knee joint supplemented by analysis of frozen sections. It is important to
recognize that this method requires considerable experience on the part of the
surgeon and the pathologist, who must keep in mind that the presence of
spacers or beads may alter the appearance of the tissues.
The surgical approach for reimplantation is typically more difficult than
that for implant removal, particularly when there has been a delay of more
than six to eight weeks. There are often associated bone deficiencies and
attendant ligamentous insufficiencies, which may require the use of more
constrained prosthetic designs to achieve knee
stability74.
Currently, posterior stabilized prostheses fixed with antibiotic-loaded cement
are used for the majority of the reimplantation procedures at our institution
(Figs. 5-A, 5-B,
5-C, 5-D). Hinged constrained
prostheses are not used if at all possible, but occasionally they are
required. Bone graft soaked in antibiotic solution (ideally in conjunction
with uncemented components) has been used to address substantial bone
deficiency, with good
outcomes75. With
cemented components, the use of bone graft is avoided by utilizing
alternatives like modular wedges or by filling bone defects with
antibiotic-loaded bone cement.
Although reimplantation has become a commonly accepted treatment for
patients with an infection at the site of a knee prosthesis, the poor outcomes
in those in whom reinfection develops after reimplantation have not been fully
appreciated. Of twenty-four patients treated for reinfection after
reimplantation of a knee prosthesis, ten ultimately had a knee arthrodesis,
five retained the prosthesis without resolution of the infection and were
maintained on suppressive oral antibiotics, four (three of whom had a failed
hinge prosthesis) had an above-the-knee amputation, three had a persistent
pseudarthrosis, one had a resection arthroplasty, and one retained the
prosthesis with resolution of the
infection76. In
three patients, a third attempt at reimplantation was made, but only one of
these third attempts was successful.
Overall, the difficulties encountered in treating infections following
total knee arthroplasty can be considerable, and the treatment must be
carefully planned after appropriate clinical and diagnostic work-up. The use
of algorithms based on scientific data from good studies with long-term
follow-up and evaluation of clearly defined outcomes will enable both the
treating surgeon and the patient to make informed decisions, with the ultimate
goals of eradicating the infection, alleviating pain, and maintaining function
(Fig. 6).