Patients
From January 2002 until February 2006, a total of twenty-one patients
underwent reimplantation after hip resection arthroplasty for the treatment of
infection. All patients were operatively managed at the Hospital Clinic of
Barcelona. The patients included eleven men and ten women with a mean age (and
standard deviation) of 65 ± 11.65 years. The reason for the primary
arthroplasty was osteoarthritis in all cases. None of the patients had
rheumatoid arthritis or inflammatory disease. All of the patients had clinical
symptoms of infection (pain, inflammatory signs, or a fistula) as well as
analytical signs of infection (a C-reactive protein level of >1 mg/dL, or
an erythrocyte sedimentation rate of >30 mm/hr) and/or a positive finding
on technetium-99m hexamethylpropylene-amine oxime-labeled leukocyte
scintigraphy. A two-stage exchange was performed for all patients. Samples for
histological and microbiological studies were always obtained during the first
and second stages of the procedure before the administration of antibiotics.
After implant removal (the first stage), the standard antibiotic regimen
consisted of cefepime and vancomycin until the results of the microbiological
studies were obtained. All patients received a minimum of six weeks of
antibiotic therapy on the basis of the known sensitivities of the infecting
organism. Antibiotic bone cement spacers containing 1 g of powdered gentamicin
per 40 g of polymethylmethacrylate were used in all patients. In nineteen
cases, the infection at the site of the primary arthroplasty was confirmed on
the basis of the finding that at least two cultures of deep samples
demonstrated growth of the same microorganism. In the remaining two cases,
cultures of deep samples were negative and the infection was confirmed on the
basis of the presence of pus around the
prosthesis17.
Reimplantation (the second stage) was considered only when the patient had
shown persistent normalization of the erythrocyte sedimentation rate and the
C-reactive protein level and had demonstrated no clinical symptoms of
infection for at least two weeks after discontinuation of antimicrobial
therapy. The mean duration (and standard deviation) between resection and
reimplantation was 212 ± 202 days. Bone scintigraphy or
pre-reimplantation aspiration of the hip to support the resolution of
infection was not performed. Reimplantation of the new prosthesis was
performed in all patients regardless of the results of frozen-section
analysis. Broad-spectrum antibiotics were initiated after samples were
obtained for histological and microbiological studies. The antibiotic therapy
was maintained or stopped depending on whether the results of the
microbiological studies were positive or negative. The same criteria that were
used to define infection during the first stage (that is, the finding that at
least two cultures of deep samples demonstrated growth of the same
microorganism or the presence of pus around the
prosthesis17) were
applied during the second stage. If the same organism was isolated from
cultures during both the first and second stages, the infection was
interpreted as persistent. If different organisms were isolated, then the
infection at the time of the second stage was interpreted as new.
Histological Analysis
At the times of both the first and second-stage operations, tissue samples
for frozen-section analysis and permanent histological analysis were obtained
from the pseudocapsule, the membrane around the prosthesis or spacer, or
tissue that was suspected to be infected. Each of two samples from each
patient were divided into two parts, one for frozen-section analysis and one
for permanent paraffin-embedded section analysis. The samples that were used
for frozen-section analysis were snap-frozen in carbon dioxide; 4-µm
sections were then cut and stained with hematoxylin and eosin. The samples
used for histological analysis of paraffin-embedded sections were fixed in
formalin and embedded in paraffin prior to staining with hematoxylin and
eosin.
For the present study, the pathologist (C.M.) reviewed the frozen and
paraffin-embedded sections that had been obtained during the second stage. The
revision was blinded because the pathologist did not know the results of the
cultures. The most cellular areas in the tissue sample were chosen for
evaluation, and the number of neutrophils (in the frozen and paraffin-embedded
sections), lymphocytes, and plasma cells (in the paraffin-embedded sections)
per high-power field (×400) in at least ten separate microscopic fields
were counted. Two histological criteria were used to diagnose infection at the
site of a revision arthroplasty: (1) Criterion A (the Feldman
criterion11,12,14),
defined as the presence of at least five neutrophils per high-power field
(×400) in at least five separate microscopic fields and (2) Criterion B
(the Athanasou
criterion10,14,18,19),
defined as the presence of at least one neutrophil per high-power field
(×400), on average, after examination of ten microscopic fields. In
addition, the number of lymphocytes and plasma cells was quantified according
to Criterion B. Fibrosis was considered to be present when mature collagen
fibers and ordered fibroblasts were found. Fibrosis was stratified as 0
(absent), 1 (moderate), or 2 (severe).
Microbiological Analysis
At the times of both the first and second-stage operations, at least three
periprosthetic samples from different sites were submitted to the laboratory
for culture. Liquid samples that were aspirated from the operative site with
use of a sterile syringe were immediately inoculated into Bactec 9000 Blood
Culture Systems (Becton Dickinson Diagnostic Instruments, Sparks, Maryland)
and were incubated for five days. Positive flasks were subcultured in aerobic
and anaerobic agar media. Swab samples were obtained by passing a sterile swab
(Deltalab invasive sterile eurotube collection swab with Stuart transport
medium; Rubí, Catalonia, Spain) over the area of tissue, bone, or fluid
that was suspected of being infected. Solid tissue samples from pseudocapsule,
the membrane around the tissue or spacer, or tissue that was suspected to be
infected were immediately placed into a separate sterile universal bottle.
Solid tissue samples and swab samples were cultured in aerobic and anaerobic
agar media and in thioglycolate broth enriched with vitamin K and hemin and
were incubated for ten days. Positive cultures were sent for organism
identification and sensitivity testing.
Statistical Analysis
The specificity (true negatives/[false positives + true negatives]),
sensitivity (true positives/[false negatives + true positives]), positive
predictive value (true positives/[true positives + false positives]), and
negative predictive value (true negatives/[true negatives + false negatives])
of Criterion A and Criterion B at the time of reimplantation were determined,
with growth of the same microorganism on at least two cultures of deep samples
of periprosthetic tissue being used as the gold standard for defining
infection. For comparison of proportions, we applied the Fisher exact test.
The level of significance was set at p < 0.05.
The C-reactive protein level, the histological and microbiological results,
and the interval between the first and second stages of each revision
arthroplasty are shown in a table in the Appendix. During the second stage of
the revision arthroplasty, there were no macroscopic signs of infection. In
the cases of eleven (52.4%) of the twenty-one patients, the cultures were
negative. In those cases, the antibiotic therapy was stopped after the
microbiological results had been obtained and none of the patients showed
clinical evidence of infection after at least six months of follow-up.
In the cases of three patients (14.3%) (Case 2, 11, and 19), only one
culture was positive and the microorganisms were considered to be
contaminants. The microorganisms that were isolated were coagulase-negative
staphylococci in two cases (both from solid tissue samples) and
Streptococcus viridans in one case (from a swab specimen). These
three patients were not considered to have an infection, and none of them had
clinical signs of infection after at least ten months of follow-up.
In the cases of seven patients (33.3%), cultures of at least two samples
were positive for the same pathogen. Six of seven patients were considered to
have a persistent infection, and one was considered to have a new infection.
The total number of positive cultures was nineteen. Seventeen cultures
(representing eight liquid samples, six solid samples, and three swab samples)
were positive for coagulase-negative staphylococci, and two (both representing
liquid samples) were positive for Candida species (see Appendix). All of these
patients received at least six weeks of specific antibiotic therapy guided by
antimicrobial testing. Six of these patients (Cases 3, 5, 6, 7, 9, and 10)
showed no clinical signs of infection after at least nine months of follow-up.
One patient (Case 15), who had an infection with Candida species, had
development of an acute infection twelve days after the reimplantation.
Débridement was performed, and cultures were again positive for Candida
species. Antifungal agents were administered for three months, and the patient
remained without clinical evidence of infection after six months of
follow-up.
Tissue at the site of a reimplantation arthroplasty is composed of
reparative fibrous and granulation tissue, which commonly contains a chronic
inflammatory cell infiltrate. The results of analysis of frozen and
paraffin-embedded sections are shown in a table in the Appendix. According to
Criterion A, the histological result was positive for two patients (Cases 6
and 15) and, in both cases, at least two cultures were positive for the same
pathogen. In the remaining nineteen cases, the histological result was
negative. In five of these nineteen cases the results of culture were
positive, and in fourteen the results of culture were negative. The results of
analysis of frozen sections coincided with the results of analysis of
paraffin-embedded sections in all cases. The sensitivity, specificity,
positive predictive value, and negative predictive value of frozen-section
analysis for the detection of the presence of microorganisms according to
Criterion A were 28.5%, 100%, 100%, and 73.6%, respectively. When Criterion B
was used, there were differences between the results of analysis of frozen
sections and paraffin-embedded sections. For this reason, the results were
analyzed with respect to the histological technique that was used.
Histological analysis of frozen sections revealed a positive result for ten
patients and a negative result for eleven patients. Of the ten patients who
had a positive result on frozen-section analysis, five had a positive result
on culture and five had a negative result on culture. Of the eleven patients
who had a negative result on frozen-section analysis, two had a positive
result on culture and nine had a negative result on culture. The sensitivity,
specificity, positive predictive value, and negative predictive value of
frozen-section analysis for the detection of the presence of microorganisms
according to Criterion B were 71.4%, 64.2%, 50%, and 81.8%, respectively.
Histological analysis of paraffin-embedded sections revealed a positive result
for seven patients and a negative result for fourteen patients. Of the seven
patients who had a positive result on analysis of paraffin-embedded sections,
five had a positive result on culture and two had a negative result on
culture. Of the fourteen patients who had a negative result on histological
analysis of paraffin-embedded sections, two had a positive result on culture
and twelve had a negative result on culture. The sensitivity, specificity,
positive predictive value, and negative predictive value of analysis of
paraffin-embedded sections for the detection of the presence of microorganisms
according to Criterion B were 71.4%, 85.7%, 71.4%, and 85.7%, respectively.
The percentage of patients who had at least one neutrophil, on average, in ten
microscopic fields on analysis of paraffin-embedded sections was significantly
higher among those who had a positive result on culture than among those who
had a negative result on culture (71.4% compared with 14.3%, p = 0.01). A
neutrophilic inflammatory infiltrate was sometimes accompanied by the presence
of other inflammatory cells. The number of lymphocytes and plasma cells on
analysis of paraffin-embedded sections was described with use of Criterion B.
In fourteen (67%) of the twenty-one patients, at least one lymphocyte in ten
microscopic fields was found. Among the seven patients who had a positive
result on culture, six (85.7%) had an average of at least one lymphocyte in
ten microscopic fields. Among the fourteen patients who had a negative result
on culture, eight (57.1%) had an average of at least one lymphocyte in ten
microscopic fields. In ten (48%) of the twenty-one patients, at least one
plasma cell in ten microscopic fields was found. Among the seven patients who
had a positive result on culture, five (71.4%) had an average of at least one
plasma cell in ten microscopic fields. Among the fourteen patients who had a
negative result on culture, five (35.7%) had an average of at least one plasma
cell in ten microscopic fields. Therefore, patients with a positive result on
culture had lymphocytes and plasma cells more frequently than did those who
had a negative result on culture, but these differences were not significant
(p > 0.05). Finally, there was a trend toward more severe fibrosis (a score
of 2) in samples from patients with a negative result on culture than in
samples from patients with a positive result on culture (85.7% compared with
42.8%; p = 0.06).
Intraoperative frozen-section histological analysis is often done to
evaluate the presence or absence of infection at the time of reimplantation
after hip resection arthroplasty that is being performed for the treatment of
an
infection2,5,7,8.
However, to our knowledge, the usefulness of this technique has been evaluated
in only one
study20. In that
study, the authors found a sensitivity of 25% and a specificity of 98% with
use of at least ten neutrophils per high-power field (×400) as a cutoff
point. Our results confirm the low sensitivity (28.5%) and high specificity
(100%) of frozen-section analysis with use of the Feldman
criterion11,12,14
(at least five neutrophils per high-power field [×400] in at least five
fields). In order to improve the detection of infection, we analyzed the
samples on the basis of the Athanasou
criterion10,14,18,19,
which considers infection to be present when an average of at least one
neutrophil per high-power field (×400) is found after examination of ten
microscopic fields. According to this criterion, the sensitivity of
frozen-section analysis increased (to >70%) but the specificity decreased
(to 64.2%)14.
A possible explanation for the low sensitivity of the Feldman
criterion11,12,14
could be the low inflammatory response induced by coagulase-negative
staphylococci. The Feldman
criterion11 was
first described in 1976 by Mirra et
al.9 in patients
infected with Pseudomonas species (seven), Escherichia coli (three),
Proteus species (four), Streptococcus species (two), and Staphylococcus
aureus (one). It is of note that only one case was due to
coagulase-negative staphylococci. Therefore, the presence of fewer than five
neutrophils per high-power field can not exclude the presence of low-virulence
microorganisms such as Propionibacterium acnes, Corynebacterium
species, or coagulase-negative staphylococci. However, it has not been well
established when these microorganisms should be considered to be true
pathogens. The criterion applied in the present study and in
others17 was the
finding that at least two samples were positive for the same microorganism,
but other authors21
have considered true pathogens to be present when at least three samples are
positive. In the present study, the percentage of patients who had at least
one neutrophil per field was significantly higher among those who had a
positive result on culture than among those who had a negative result on
culture (71.4% compared with 14.3%, p = 0.01). This finding supports the
usefulness of the cutoff point applied in the present study. In addition,
according to Levine and
Evans22, liquid
samples have a high specificity (100%) and sensitivity (92%), and it is of
note that in our study five of seven patients with an infection had positive
liquid samples.
The successful treatment of infection after reimplantation in patients who
had a positive result on culture could have been due to the excision of all
infected tissue and antibiotic treatment for at least six weeks. Indeed,
one-stage exchange has been associated with good results when used for the
treatment of infection due to low-virulence microorganisms at the site of an
arthroplasty23-25.
On the other hand, even when the Athanasou
criterion10,14,18,19
was applied, two patients had fewer than one neutrophil per high-power field
and a positive result on culture, and we cannot exclude the possibility that
these microorganisms were contaminants. However, Tunney et
al.26 studied
paraffin-embedded sections of periprosthetic tissue from eighteen patients who
had a documented infection with low-virulence microorganisms and reported that
they did not find any neutrophils in eight cases, which suggests that
infection due to this type of microorganism cannot be excluded even in the
absence of neutrophils. The percentages of cases with at least one lymphocyte
or plasma cell per high-power field at the time of reimplantation were higher
(67% and 48%, respectively) than the percentages described by Pandey et
al.18 at the time
of resection arthroplasty (35% and 6%, respectively). However, the number of
lymphocytes or plasma cells observed in tissue was not significantly
associated with positive or negative cultures; therefore, the analysis of
these cells did not help in the diagnosis of infection. It is of note that
there was a trend toward an association between extensive fibrosis and a
negative culture. This finding suggests that the presence of active infection
may not allow the formation of fibrous tissue. However, the score that was
used to measure the degree of fibrosis was subjective and therefore less
reproducible.
The main drawback of the present study was the low number of patients
included. However, all of the patients were managed by the same team of
surgeons and infectious-disease specialists following a strict protocol for
the treatment of prosthetic joint infection as well as for the number of
samples taken for pathological and microbiological studies.
In conclusion, frozen-section analysis with use of the Feldman
criterion11,12,14
had a low sensitivity but high specificity for predicting the presence of
low-virulence microorganisms in samples of periprosthetic tissue at the time
of reimplantation after hip resection arthroplasty. With use of the Athanasou
criterion10,14,18,19,
the sensitivity increased but the specificity decreased. Therefore, in the
presence of at least five neutrophils per high-power field the probability of
infection is high, but in the presence of fewer than five neutrophils per
high-power field it is not possible to rule out infection. In these cases, we
recommend obtaining culture specimens of periprosthetic tissue and maintaining
antibiotic treatment until definitive results are determined.
Tables showing clinical, culture, and histologic results for all patients
in the study are available with the electronic versions of this article, on
our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?