Orthopaedic Devices
Titanium-alloy (Ti6Al4V) pins (Biomet, Warsaw,
Indiana), 15 mm long and 2.8 mm in diameter, were used. The lower one-quarter
of the pin is threaded, whereas the remaining segment of the pin is blasted
with silica carbide (30 grit). Prior to surgery, rabbits were randomized to
receive either uncoated pins or pins that had been coated with a combination
of minocycline and rifampin. The antibiotic coating was applied evenly and
directly (without use of a carrier) by spraying the blasted portion of the
metal implant with a methanol solution that contained 17 mg/mL of both
minocycline and rifampin. The coated device was air-dried to allow evaporation
of the solvent, leaving a coat of pure antibiotic powder on the device
surface. The device was packaged in Tyvek (high-density polyethylene) pouches
and was terminally sterilized with gamma radiation, with minimum and maximum
delivered doses of 26.4 and 31.4 kGy. The amounts of minocycline and rifampin
that are noncovalently bound (by means of van der Waals forces or hydrogen
bonding) to the devices used in this animal study are 35 µg each per square
centimeter of surface area. Both minocycline and rifampin detach relatively
rapidly from the device surface on hydration to provide local antimicrobial
concentrations that are sufficiently high to effectively kill bacteria. In
vitro elution analysis showed that 94% of minocycline and 95% of rifampin
elute by six hours after immersion of the coated device in buffered saline
solution.
Infecting Pathogen
The methicillin-sensitive Staphylococcus aureus P1 strain is a
variant of the ATCC 25923 strain that was originally used in a rabbit model of
catheter-related
infection12 and was
subsequently utilized in animal models to cause infections in association with
a variety of surgical implants, including prosthetic heart
valves13, penile
implants14, tissue
expanders15, and
vascular grafts16.
The Staphylococcus aureus P1 strain is susceptible to the two coating
antibiotics, minocycline (minimum inhibitory concentration, 0.5 µg/mL) and
rifampin (minimum inhibitory concentration, 0.01 µg/mL), as well as to
antibiotics that are commonly used in patients with orthopaedic devices, such
as cefazolin (minimum inhibitory concentration, 2 µg/mL) and vancomycin
(minimum inhibitory concentration, 2 µg/mL). The combination of minocycline
(a bacteriostatic drug) and rifampin (a bactericidal agent) provides only
additive activity (because of rifampin's rapid and powerful bactericidal
activity) against most strains of Staphylococcus aureus, including
the P1 strain, and is synergistic against almost 30% of Staphylococcus
aureus strains, particularly methicillin-resistant
isolates17-19.
In Vitro Antimicrobial Activity
Using a modified Kirby-Bauer
technique20, we
assessed the in vitro antimicrobial activity of gamma-irradiated pins coated
with minocycline and rifampin by determining the zone of inhibition against
the same P1 strain of Staphylococcus aureus that was tested in the
animal study. The zones of inhibition were assessed at baseline as well as at
one and three days after suspension of the antimicrobial-coated pins in bovine
calf serum (HyClone, Logan, Utah) at 37°C and daily changes of the serum.
Bacteria were grown at 37°C for eighteen hours in trypticase soy broth to
a concentration of 108 CFU (colony-forming units)/mL. A cotton swab
was dipped in the bacterial suspension and then rubbed in a streaking fashion
across the whole surface of a Mueller-Hinton agar plate. The
antimicrobial-coated pin was then tightly embedded in a groove created by
removing some agar from the center of the agar plate. After incubation of the
agar plate at 37°C for twenty-four hours, the zone of inhibition was
determined by measuring the diameter of the clear zone perpendicular to the
long axis of the pin.
Surgical Implantation
This animal study was approved by our institutional review board. A
previously described rabbit model of infection in association with a femoral
intramedullary screw was used with some
modifications21.
Twenty-eight New Zealand White, specific-pathogen-free rabbits (body mass, 3
to 4 kg) were anesthetized with an intramuscular injection of ketamine (37.5
mg/kg) and acetopromazine (0.52 mg/kg) in a solution that contained ketamine
(44 mg/mL) and acetopromazine (0.61 mg/mL). The right hindleg of each rabbit
was shaved from the knee to the ankle, cleaned with triclosan solution and
then with povidone-iodine, and aseptically draped. After a 2-cm incision was
made over the right knee joint, the patella was dislocated and the knee joint
and the intercondylar notch were exposed. A hole was drilled with use of a
2.8-mm drill bit through the intercondylar notch into the medullary canal of
the femur (Fig. 1). A 25-µL
aliquot of a 2 × 104 CFU/mL suspension (equivalent to an
absolute bacterial inoculum of 5 × 102 CFU) of
Staphylococcus aureus P1 strain was introduced through the drilled
hole. In a pilot trial that was done to compare the ability of different
inocula of the Staphylococcus aureus to cause infection, we concluded
that a bacterial inoculum of 5 × 102 CFU was optimal as it
resulted in colonization of most uncoated pins. Either an antimicrobial-coated
pin (fourteen rabbits) or an uncoated pin (fourteen rabbits) was screwed into
the drilled hole. After relocation of the patella, the wound was closed with
use of a monofilament nylon suture material. To ensure adequate pain control,
an intramuscular injection of the anti-inflammatory agent ketoprofen (2 mg/kg)
was given immediately after the surgery and then daily for three days. The
animals were monitored daily to evaluate the appearance of the surgical wound,
mobility, signs of sepsis, and the ability to thrive.
Collection of Samples for Cultures
The rabbits were killed at one week after the surgery with an intracardiac
injection of phenobarbital. The knee of the right hindleg was cleaned, and a
5-cm incision was made lateral to the surgical wound that had been created at
the time of pin implantation. The patella was dislocated, and the knee joint
and the intercondylar notch were exposed. The titanium-alloy pin was removed
in a sterile fashion. The track where the implant had been placed was swabbed
to obtain a specimen for culture. Bone surrounding the device track and blood
samples from the heart were also collected in a sterile fashion from each
rabbit after death.
Device Cultures
Each explanted device was placed in a sterile tube that contained 2 mL of
normal saline solution, following which the tube was sonicated with use of a
Bransonic ultrasonic cleaner (Danbury, Connecticut) in a water bath for five
minutes at room temperature with use of an output of 100 W. After vortexing
for thirty seconds, 100-µL aliquots of the original suspension and of
subsequent dilutions in normal saline solution (one in ten, one in 100, one in
1000, and one in 10,000) were inoculated onto sheep blood trypticase soy agar
plates. After the plates were incubated at 37°C for forty-eight hours, the
numbers of bacterial colonies were determined.
Bone Cultures
The removed femoral bone fragments were weighed (mean weight, 8.2 g) and
then were homogenized in 10 mL of normal saline solution. The tube containing
the homogenized bone was sonicated in a water bath for five minutes with use
of an output of 100 W. After vortexing for thirty seconds, 100-µL aliquots
of the original suspension and of subsequent dilutions in normal saline
solution (one in ten, one in 100, one in 1000, and one in 10,000) were
inoculated onto sheep blood trypticase soy agar plates. After the plates were
incubated at 37°C for forty-eight hours, the numbers of bacterial colonies
were determined.
Cultures of Swabs Used to Obtain a Specimen from the Pin Track
Each swab that had been used to obtain a specimen from the pin track was
immersed in a tube that contained 100 µL of normal saline solution, and the
tube was sonicated in a water bath for five minutes with use of an output of
100 W. The tube was then removed and was vortexed for thirty seconds. A
100-µL aliquot of the original suspension was inoculated onto sheep blood
trypticase soy agar plates. Bacterial growth was assessed after the plates
were incubated at 37°C for forty-eight hours.
Blood Cultures
A 200-µL aliquot of whole blood was inoculated onto sheep blood
trypticase soy agar plates. Bacterial growth was assessed after the plates
were incubated at 37°C for forty-eight hours.
Assessed Outcomes
The three outcomes of the animal study were (1) device colonization,
defined as growth of the same infecting strain (confirmed by antibiotic
susceptibility assay) of Staphylococcus aureus on culture of the
explanted device (detectability limit, 20 CFU); (2) osteomyelitis, defined as
growth of the same infecting strain of Staphylococcus aureus in the
sonication culture of bone (detectability limit, 100 CFU); and (3)
device-related osteomyelitis, diagnosed by growth of the same infecting strain
of Staphylococcus aureus in cultures of both the device and the bone.
A two-tailed Fisher exact test was used to compare the rates of device
colonization, osteomyelitis, and device-related osteomyelitis between the two
groups of devices.
In Vivo Outcomes
Of the twenty-eight rabbits that had undergone surgery, three (two in the
uncoated group and one in the coated group) died within three to five days
after the surgery as a result of a failure to thrive but without evidence of
sepsis (i.e., they did not have a fever or positive blood cultures). The
remaining twenty-five rabbits (thirteen with an antimicrobial-coated device
and twelve with an uncoated device) were killed at one week after the surgery,
and the results of the microbiologic evaluations of all of those animals were
included in the analysis. The antimicrobial-coated devices had a significantly
lower rate of colonization than the uncoated devices (five of thirteen
compared with twelve of twelve, p = 0.0016; relative risk = 2.60 [95%
confidence limits, 1.31 to 5.17]). Additionally, the antimicrobial-coated
devices were associated with significantly lower rates of osteomyelitis (six
of thirteen compared with twelve of twelve, p = 0.005; relative risk = 2.17
[95% confidence limits, 1.20 to 3.90]) and device-related osteomyelitis (five
of thirteen compared with twelve of twelve, p = 0.0016; relative risk = 2.60
[95% confidence limits, 1.31 to 5.17]). Except for one rabbit that had
colonization of an antimicrobial-coated device but a negative culture of the
swab used in the device track, all rabbits that had device colonization
(twelve in the uncoated group and four in the coated group) also had growth of
Staphylococcus aureus on culture of the swab used in the device
track. No rabbit had a positive culture of the swab used in the device track
along with a negative device culture. All blood cultures (detectability limit,
5 CFU/mL) performed in this study were sterile.
In Vitro Zones of Inhibition
The in vitro zones of inhibition provided by the antimicrobial-coated pins
against Staphylococcus aureus P1 strain were 36 mm at baseline
(Fig. 2), 12 mm at day 1, and
undetected at day 3.
Antimicrobial coating of medical devices has recently emerged as a
potentially effective method to prevent device-related infections. The results
of this animal study indicate that titanium-alloy orthopaedic devices coated
with a combination of minocycline and rifampin can provide protection in this
rabbit model against colonization and infection by Staphylococcus
aureus, the most common cause of such
infections22. The
protective clinical efficacy of incorporating minocycline and rifampin onto
the surfaces of vascular catheters was accurately predicted by the findings of
a rabbit study of the anti-infective efficacy of such coated catheters against
Staphylococcus
aureus23.
Although rabbits usually resist infection better than humans, it is unclear
whether that was a factor in the animal model used in the present study, in
which an infection developed in all animals treated with the uncoated device.
Furthermore, since the study had a control arm of uncoated devices that were
placed in the same type of rabbits as those that received the
antimicrobial-coated devices, the impact on the results of any inherent
resistance to infection by the studied rabbits would be minimal.
Notwithstanding the potential differences between animals and humans with
regard to host characteristics and mode of bacterial inoculation, the findings
of this rabbit study suggest that orthopaedic devices coated with minocycline
and rifampin may be clinically protective.
Staphylococcus aureus adheres to a variety of host-derived
components of the biofilm layer, including fibronectin, fibrinogen, collagen,
and, to a lesser extent,
laminin5. The
anti-infective capacity of these antimicrobial-coated orthopaedic devices may
be related, at least in part, to the production of an effective zone of
inhibition that inhibits the adherence of organisms not only to the surface of
the device but also to the biofilm layer around the device. This concept is
supported by the results of an in vitro dynamic adherence study of vascular
catheters (modified Robbins
devices)24 and a
scanning electron microscopic examination of vascular catheters removed from
patients25. An in
vitro zone of inhibition that exceeds 10 to 15 mm is commonly regarded as
being effective, as it has been shown to accurately predict the likelihood of
antimicrobial-coated devices protecting against Staphylococcus aureus
colonization and infection in both
animal5,12
and clinical
studies6-9.
By providing zones of inhibition of 36 mm at baseline and 12 mm a day later,
the antimicrobial-coated pins in our animal study gave adequate protection
against infection by the Staphylococcus aureus organisms with which
the animals had been inoculated at the time of pin implantation.
For antimicrobial-coated orthopaedic devices to provide optimal protection,
it is essential that they provide activity not only against the most common
causative organisms (i.e., Staphylococcus aureus and
Staphylococcus epidermidis) but also against other potential
pathogens so that superinfection does not ensue. Combining minocycline with
rifampin broadens the spectrum of antimicrobial activity against gram-negative
bacteria. Although the spectrum of the antimicrobial activity of pins coated
with minocycline and rifampin was not assessed in this study, nonmetallic
devices coated with the same antimicrobial combination have been demonstrated
in vitro to provide activity against a wide variety of potential pathogens,
including gram-positive cocci and gram-negative
bacilli23. Another
advantage of combining minocycline with rifampin is that it lessens the risk
of the development of resistance to rifampin because of differences in the
mechanisms of activity of minocycline (inhibition of protein synthesis) and
rifampin (inhibition of DNA-dependent RNA
polymerase)23,26,27.
Despite the expanded clinical use of non-orthopaedic devices coated with the
combination of minocycline and rifampin, there has been no evidence of
emergence of antibiotic
resistance6,7.
Although similar findings would be expected with the clinical use of
orthopaedic implants coated with minocycline and rifampin, it would be
important to monitor antimicrobial susceptibilities. Anti-infective
orthopaedic devices should be used only for patients for whom the risk of
infection is high or the medical/economic consequences of infection would be
very serious. When used in that manner, the clinical protection against
infection that is afforded by vascular catheters coated with minocycline and
rifampin has been associated with large
cost-savings6,9.
Our approach of coating orthopaedic devices with a combination of
minocycline and rifampin is unique for three reasons. First, to our knowledge,
it is the only antimicrobial coating approach that has been demonstrated in
clinical trials to significantly (p < 0.05) reduce the rate of clinical
infection (as compared with that associated with uncoated devices) in
association with a variety of non-orthopaedic devices, including vascular
catheters6-9,
ventricular
catheters10, and
penile implants11.
Second, studies of animals that have demonstrated that coating of vascular
catheters with the combination of minocycline and rifampin significantly (p
< 0.05) reduces the rate of infection by Staphylococcus
aureus23 have
accurately predicted the likelihood of clinical protection against
Staphylococcus aureus in subsequent clinical trials
6-9,
whereas other types of antimicrobial delivery systems for orthopaedic devices
have not, to our knowledge, been evaluated in a similar fashion. Third, the
direct application of minocycline and rifampin to the surface of the metallic
implant without the use of a carrier is unique and is intended to prevent
infection. In contrast, most other antimicrobial delivery systems that have
been studied in animal models of osteomyelitis have involved the use of
carriers (such as resorbable polymers and permanent polymers like
polymethylmethacrylate, hydroxyapatite, and collagen), with a focus on the
controlled release of antimicrobial agents to treat an existing infection.
As the device used in this animal study contained only 35 µg of
minocycline and rifampin each per square centimeter of surface area, the
maximal amounts of minocycline and rifampin incorporated onto the largest hip
prosthesis (including both the femoral and the acetabular component) used in
patients would be 15 mg each. These antibiotic amounts constitute a very small
fraction of the daily systemic dose of minocycline (7.5% of 200 mg) and
rifampin (2.5% of 600 mg). Although we did not assess the serum and urinary
levels of minocycline and rifampin in this animal study, it is very unlikely
that such levels would be detectable. This scenario is similar to that of
central venous catheters coated with minocycline and rifampin, which are not
associated with detectable serum antibiotic levels and are used to prevent
infection25, but is
different from that of polymethylmethacrylate spacers impregnated with large
amounts (comparable with systemic doses) of other antibiotics (such as
vancomycin, aztreonam, and gentamicin) that are intended to treat an existing
infection28,29.
If almost all antibiotics bound to the surface of the orthopaedic device
dissolve into the space between the implant surface and the adjacent bone
tissue by means of a diffusion path of 1 mm, temporary local concentrations of
minocycline and rifampin in excess of 100 µg/mL would be achieved. Such
temporary local concentrations would greatly exceed the minimal inhibitory
concentrations of minocycline for 90% of methicillin-susceptible (0.12
µg/mL) and methicillin-resistant (8 µg/mL) Staphylococcus
aureus strains
30 and the minimal
inhibitory concentration of rifampin (0.015 mg/µL) for 90% of both
methicillin-susceptible and methicillin-resistant Staphylococcus
aureus strains
31. The short-lived
activity of an antimicrobial-coated device is intended to effectively combat
bacteria that are introduced perioperatively into the implantation site as a
result of contamination of the implant, bone, or soft tissue. Although most
pathogens responsible for infections associated with joint prostheses are
thought to adhere to the device or the surrounding layer of biofilm during the
perioperative
period21, infection
with low-virulence organisms such as Staphylococcus epidermidis may
not manifest clinically until months or years after insertion of a
prosthesis27.
Although the results of this animal study demonstrate the protective
efficacy of these antimicrobial-coated orthopaedic devices in this rabbit
model, the potential clinical benefit of such coated devices can be confirmed
only in prohibitively large human trials. ?