Preparation of Antibiotics and Calcium Sulfate Flakes
Twenty grams of calcium sulfate (Sigma, St. Louis, Missouri) was sterilized
in ethylene oxide gas and then stirred with 10 mL of sterile water until a
uniform consistency was reached. It was then injected, with a syringe, into
6-mm-diameter by 8-mm-deep molds, where it was allowed to dry for twenty-four
hours. The resulting pellets were crushed in a mortar and pestle to form
flakes. The same procedure was used to form the gentamicin-loaded calcium
sulfate flakes, but with gentamicin present in the sterile water used for the
mixture. The gentamicin solution was mixed with the calcium sulfate in a ratio
that resulted in 128 mg of gentamicin per 1 g of loaded calcium sulfate.
Study Design
After approval was granted by our institutional animal care and use
committee, female Sprague-Dawley retired breeder rats with weights ranging
from 200 to 400 g were obtained to carry out a prospective, randomized,
blinded study. This study was piloted with three groups of five rats to ensure
that enough infections occurred to effectively differentiate among the results
of the different treatments. In a previous, unpublished study in which a lower
dose of bacteria and no foreign body were used, the infections readily
resolved with all forms of antibiotic treatment, and the relative efficacy of
the treatments could not be differentiated. After the pilot study, the formal
study included seven different treatment groups composed of ten rats each
(described below). The group treated with systemic gentamicin and the one
treated with local gentamicin were then extended to include twenty-five and
twenty-seven rats, respectively, because the results in those two groups had
suggested that there may be a significant difference between them if larger
numbers were studied.
Surgical Procedure
Isoflurane anesthesia (2.0% to 3.0%, to effect) was administered, and the
preoperative weight of each rat was determined. The lateral aspect of the
right thigh was then depilated with electric clippers. The area was prepared
with iodine solution and was draped in a sterile fashion. The skin was opened
with a 10-mm longitudinal incision in the lateral aspect of the thigh with use
of scissors. A 2 × 2-cm pocket was formed, with blunt dissection, in the
quadriceps muscle down to the femur. A 32-gauge stainless-steel suture was
placed around the femur inside the wound to act as a surgical implant or
foreign body to promote infection. Approximately 8.0 × 105
colony-forming units (CFUs) of gentamicin-sensitive Staphylococcus
aureus (from American Type Culture Collection [ATCC] 29523) was pipetted
into the pocket and allowed to remain for 2.5 minutes. The rats were then
randomly allocated to one of seven treatment groups. The wound was irrigated
with 0.5 mL of sterile saline solution (except in the rats in which the wound
was irrigated with bacitracin solution instead), and the irrigant was allowed
to run out of the wound. The bacterial concentration and the small irrigant
volume were based on the amounts that had resulted in consistent infections in
the pilot studies. The rat then received the appropriate treatment for its
group.
The seven types of treatment were:
Control—no treatment was given, and the incision was simply
closed.Plain calcium sulfate flakes—62.5 mg of flakes only were placed deep
in the wound pocket, and the wound was closed.Bacitracin irrigation—the wound was irrigated with 0.5 mL of
bacitracin (50 U/mL) instead of saline solution and then was closed.Systemic gentamicin—the wound was closed, and then 0.16 mL of aqueous
gentamicin (50 mg/mL) was administered subcutaneously (a dose equal to 20
mg/kg).Local gentamicin—the wound was closed, and 0.16 mL of 50 mg/mL
aqueous gentamicin solution (a dose equal to 20 mg/kg) was injected into the
depths of the closed surgical wound.Calcium sulfate loaded with gentamicin—flakes loaded with 62.5 mg of
gentamicin (a dose equal to 20 mg/kg) were placed deep in the surgical pocket,
and the wound was closed.Combination treatment—systemic gentamicin (10 mg/kg) was delivered
subcutaneously and gentamicin-loaded calcium sulfate flakes (10 mg/kg) were
placed in the wound; irrigation was performed with 0.5 mL of 50 U/mL
bacitracin instead of with saline solution.
Control—no treatment was given, and the incision was simply
closed.
Plain calcium sulfate flakes—62.5 mg of flakes only were placed deep
in the wound pocket, and the wound was closed.
Bacitracin irrigation—the wound was irrigated with 0.5 mL of
bacitracin (50 U/mL) instead of saline solution and then was closed.
Systemic gentamicin—the wound was closed, and then 0.16 mL of aqueous
gentamicin (50 mg/mL) was administered subcutaneously (a dose equal to 20
mg/kg).
Local gentamicin—the wound was closed, and 0.16 mL of 50 mg/mL
aqueous gentamicin solution (a dose equal to 20 mg/kg) was injected into the
depths of the closed surgical wound.
Calcium sulfate loaded with gentamicin—flakes loaded with 62.5 mg of
gentamicin (a dose equal to 20 mg/kg) were placed deep in the surgical pocket,
and the wound was closed.
Combination treatment—systemic gentamicin (10 mg/kg) was delivered
subcutaneously and gentamicin-loaded calcium sulfate flakes (10 mg/kg) were
placed in the wound; irrigation was performed with 0.5 mL of 50 U/mL
bacitracin instead of with saline solution.
The incision was closed superficially with skin clips. Post-operatively,
the rats received acetaminophen elixir diluted in water (100 mg/100 mL) for
pain control until they were killed.
Outcome Measurement
After two days, all rats were killed with a CO2 overdose.
Whenever a rat died before the scheduled time, the outcome was measured at the
time that the death was discovered. The original surgical site was opened by
swabbing the skin with alcohol and then excising the overlying skin to prevent
contamination of the underlying wound with skin flora, after which the
surgical pocket was opened with spreading dissection with a hemostat. A
cotton-tipped applicator was then swabbed through the wound, with an attempt
to uniformly contact all parts of the surgical pocket, to obtain a sample to
be used for quantitative culture. The culture swab was placed in 1 mL of Luria
broth and was vortexed. Serial 1:10 dilutions were performed, and the specimen
broth was inoculated onto Luria-Bertani agar and spread with use of a bent
Pasteur pipette. The plates were assessed for the number of CFUs at
forty-eight hours. Concentrations of the bacteria (in CFU/mL) were calculated
from the plates that had countable numbers of colonies. A plate was considered
satisfactory for counting if it had between ten and 200 colonies. If more than
one plate had a countable number, the average was calculated. If no plate for
a given specimen had a countable number of colonies, the most dilute specimen
with any number of colonies present was used to calculate the concentration.
When an animal had either no difference or a greater than two orders of
magnitude difference between one serial dilution and the next, the data were
discarded as changes of one order of magnitude would be expected. Failure to
meet these criteria was interpreted as indicating methodological error.
The Kruskal-Wallis nonparametric test was used for the statistical
evaluations of colony counts because the assumptions of the t test were not
met in the data produced (i.e., the data were not normally distributed). A p
value of <0.05 was considered to be significant.
Mortality was high in the groups that did not receive gentamicin: six of
nine rats in the control group (the data for one rat in that group was
discarded because it did not meet criteria for "countable" as
described under "Outcome Measurement"), seven of the ten in the
plain-calcium-sulfate group, and five of the ten in the bacitracin group died
during the two days after the surgery. No rats in the gentamicin-treated
groups died before they were killed at forty-eight hours after the initial
surgery.
The weights of the rats were recorded preoperatively and at the time that
they were killed, but no significant differences or trends were found among
the groups.
The mean numbers of CFUs in the seven groups are shown in
Figure 1. Analysis-of-variance
testing on ranks showed significant improvements compared with the control
values in all gentamicin-treated groups (p < 0.05 for all groups). It
should be noted that 50,000 or 100,000 CFUs are typically recognized as the
cutoff for the number of bacteria signifying local infection.
There was no significant difference in the bacterial counts between the
control group and the plain-calcium-sulfate group or the bacitracin-treated
group (p > 0.05). Similarly, there were no significant differences among
the systemic gentamicin, calcium sulfate with gentamicin, or
combination-treatment groups (p > 0.05).
Kruskal-Wallis testing showed a significant decrease (p = 0.0003) in the
mean bacterial count in the group treated with locally administered aqueous
gentamicin as compared with the mean count in the group treated with systemic
gentamicin. The median counts in the two groups were 0 and 14 CFU/mL,
respectively.
Contrary to our hypothesis, calcium sulfate loaded with gentamicin did not
appear to be more effective than systemic delivery of gentamicin for
prophylaxis against wound infection. However, simple local injection of
aqueous gentamicin into the wound cavity after wound closure was significantly
more effective than systemic administration. This may have been due to the
tendency of the calcium sulfate to act not only as a drug delivery system, but
also as a foreign body to which the bacteria could adhere. Another possible
explanation for this result may stem from the fact that the peak
concentrations of gentamicin would not have been as high when it slowly eluted
from the calcium sulfate flakes as when it was injected in an aqueous
form.
Neither plain calcium sulfate nor bacitracin appears to be significantly
more effective than no treatment. While we did not expect plain calcium
sulfate to lower the bacterial load, it should be noted that the common
clinical practice of bacitracin irrigation also failed to lower the bacterial
load compared with that in the control group. This would suggest that
bacitracin irrigation alone is insufficient to prevent surgical wound
infections. We suspect that antibiotic irrigation is less effective than one
might expect because the irrigation fluid is usually aspirated from the wound
prior to wound closure and therefore the bacteria are not exposed to high
concentrations of antibiotic for a long enough time for it to be effective.
There is extensive evidence that wound irrigation, with or without
antibiotics, lowers infection rates, presumably as a result of dilution and
removal of bacteria. All of the wounds in this study were irrigated with
saline solution except for those irrigated with
bacitracin9. We
believe that irrigation should remain part of the armamentarium for
prophylaxis against surgical wound infection. However, the results of this
study support the practice of placing any antibiotic locally in the wound by
injection after closure rather than in the irrigation fluid, where it will
mostly be washed away before closure. As bacitracin is a relatively toxic
antibiotic, it probably is not suitable for injection after closure of the
wound, where much of it may be absorbed.
The bacterial counts seemed remarkably low in the group in which plain
aqueous gentamicin was injected into the wound after closure. We expected that
such an injection would be less effective because we believed that the local
antibiotic levels would drop very rapidly as a result of escape from the wound
and systemic absorption. Initial analysis-of-variance testing on ranks with
only ten rats in each group indicated a promising trend for locally
administered gentamicin to be more effective than systemic gentamicin, but
that trend was not significant. Therefore, we extended those two groups to
twenty-five and twenty-seven animals and were then able to determine that this
difference was significant (p = 0.0003).
We used gentamicin in this study because there is a long history of local
administration of this antibiotic and because it provides broad-spectrum
coverage for the bacteria that commonly cause surgical wound infection.
However, we have no reason to believe that other antibiotics might not be used
locally in the same way.
The dose of gentamicin that we chose to use in this study was 20 mg/kg.
This dose is higher than the dose commonly suggested for rats, for which the
upper limit is typically 12
mg/kg/day10.
However, studies have shown that no side effects occur when gentamicin is
given to rats at a dose of 20 mg/kg/day for twenty-four
days11. Thus, no
adverse effects would be expected with a one-time dose at this level. In this
study, we injected 8 mg into an approximately 2-cc wound. Guidelines suggest
that 5 to 7 mg/kg of gentamicin can be administered to humans every
twenty-four hours, so it should be safe to administer 350 mg of gentamicin as
a one-time dose to a 70-kg man. A wound size of 87.5 cc (which seems to be a
reasonable size to expect in many orthopaedic procedures) containing 350 mg of
gentamicin would contain the same concentration of antibiotic as the wounds in
these study rats.
Because both systemic and local antibiotics appear to be effective
separately, one might expect that administration of both local and systemic
antibiotics together might be even more effective. However, our group in which
gentamicin was administered both systemically and locally did not have a lower
mean bacterial count. It must be recognized that, because this study involved
a single antibiotic (except for irrigation with the topical antibiotic
bacitracin) and the gentamicin has substantial toxicity, we cut the local and
systemic doses in half in our combination group to maintain a consistent total
dose. We suspect that systemic administration of cephalosporin and local
administration of gentamicin might prove to be more effective than either drug
alone, but that possibility awaits future study.
One potential problem with the local administration of high doses of
gentamicin is that the concentration inside the wound may be high enough to
cause local tissue toxicity, including cell death or inhibition of healing.
Miclau et al. found that sustained in vitro tissue-culture levels of
gentamicin of 400 µg/mL resulted in decreased cell replication, and levels
of 10,000 µg/mL resulted in cell
death12. We
recognize that the concentration injected in this study (50,000 µg/mL) is
higher than this level, but once 0.12 mL is injected into a wound containing 1
mL of solution the level is diluted almost immediately to approximately 8000
µg/mL. Additionally, antibiotic levels in tissue cultures are sustained,
whereas the levels found in wounds rapidly decrease as a result of diffusion
into the wound hematoma and systemic absorption of the antibiotic. Humphrey et
al. showed that local wound levels of gentamicin eluted from a bovine collagen
sponge dropped rapidly in a rabbit
model13. They
administered 3 mg/kg (which would be 12 mg in a 4-kg rabbit) into an
approximately 2 × 2 cm wound and found the gentamicin level in the
hematoma to be 600 µg/mL at four hours and <7 µg/mL at twenty-four
hours. This wound size is approximately the same as the surgical pocket in our
study, but we delivered less antibiotic (8 mg). Because we had no drug
delivery system in place, the antibiotic level would be expected to have
dropped even faster than that in the study by Humphrey et al. Furthermore,
Miclau et al. showed that bone graft and demineralized bone matrix eluted 70%
and 45%, respectively, of their antibiotic load by twenty-four
hours14. Negligible
amounts were detected in both of these media at one
week14. Again, we
would expect that the antibiotic levels following local delivery with our
aqueous injection technique would have fallen even faster because the
antibiotic was delivered in a solution instead of with the sustained-release
collagen delivery system.
On the basis of this study, we concluded that both local and systemic
antibiotics are effective for prophylaxis against surgical wound infections.
However, gentamicin solution injected directly into the surgical wound after
wound closure kills bacteria more effectively than does systemically
administered gentamicin. Local administration produces high concentrations of
antibiotic inside the wound while minimizing systemic levels that can lead to
toxicity. We recommend that the use of local antibiotic prophylaxis against
infection be evaluated in larger animals and in controlled human studies
before clinical application of the findings of the present appraisal.
Furthermore, it seems likely that a combination of locally and systemically
administered antibiotics of different classes and side-effect profiles may be
even more effective, but this option was incompletely evaluated in this study
and should be more carefully assessed in future work. ?
Note: The authors thank Dr. Chris Elkins, PhD, and Kevin Smith
for consultations regarding bacterial culture and quantification; Arun Aneja
for contributions to the study design and methods; and Jeanine Matuszewski for
contributions to the statistical analysis.